solar energy and pv systems

International Journal of Photoenergy

Solar Energy and PV SystemsGuest Editors Ismail H Altas and Adel M Sharaf

Solar Energy and PV Systems



Guest Editors Ismail H Altas and Adel M Sharaf

Copyright copy 2014 Hindawi Publishing Corporation All rights reserved

This is a special issue published in ldquoInternational Journal of Photoenergyrdquo All articles are open access articles distributed under theCreative Commons Attribution License which permits unrestricted use distribution and reproduction in any medium provided theoriginal work is properly cited

Editorial Board

M S A Abdel-Mottaleb EgyptXavier Allonas FranceNicolas Alonso-Vante FranceWayne A Anderson USAYanhui Ao ChinaRaja S Ashraf UKV Augugliaro ItalyDetlef W Bahnemann GermanyI R Bellobono ItalyRaghu N Bhattacharya USAPramod H Borse IndiaAlessio Bosio ItalyStephan Buecheler SwitzerlandGion Calzaferri SwitzerlandC Chen ChinaSung Oh Cho Republic of KoreaV Cimrova Czech RepublicJuan M Coronado SpainYing Dai ChinaD D Dionysiou USAPingwu Du ChinaM M El-Nahass EgyptPolycarpos Falaras GreeceChris Ferekides USAPaolo Fornasiero ItalyHermenegildo Garcıa SpainGerma Garcia-Belmonte SpainE I Garcia-Lopez ItalyBeverley Glass AustraliaM A Gondal Saudi ArabiaJr-Hau He TaiwanShinya Higashimoto JapanCheuk-Lam Ho Hong KongWing-Kei Ho Hong KongFuqiang Huang ChinaAdel A Ismail EgyptChun-Sheng Jiang USAMisook Kang Republic of KoreaShahed Khan USASun-Jae Kim Republic of Korea

Jong Hak Kim Republic of KoreaSungjee Kim Republic of KoreaCooper H Langford CanadaTae-Woo Lee Republic of KoreaLecheng Lei ChinaXinjun Li ChinaZhaosheng Li ChinaYuexiang Li ChinaStefan Lis PolandVittorio Loddo ItalyGongxuan Lu ChinaDongge Ma ChinaN M Mahmoodi IranNai Ki Mak Hong KongRajaram S Mane IndiaD Mantzavinos GreeceUgo Mazzucato ItalySheng Meng ChinaJacek Miller PolandClaudio Minero ItalyAntoni Morawski PolandFranca Morazzoni ItalyF Morlet-Savary FranceM Muneer IndiaKun Na Republic of KoreaEbinazar B Namdas AustraliaMaria Neves PortugalTebello Nyokong South AfricaKei Ohkubo JapanHaridas Pal IndiaLeonidas Palilis GreeceLeonardo Palmisano ItalyRavindra K Pandey USAH Park Republic of KoreaPierre Pichat FranceGianluca Li Puma UKTijana Rajh USAPeter Robertson UKAvigdor Scherz IsraelElena Selli Italy

Ganesh D Sharma IndiaJinn Kong Sheu TaiwanPanagiotis Smirniotis USAZofia Stasicka PolandElias Stathatos GreeceJ Subbiah AustraliaM Swaminathan IndiaKazuhiro Takanabe Saudi ArabiaMohamad-Ali Tehfe CanadaK R Justin Thomas IndiaYang Tian ChinaNikolai V Tkachenko FinlandAhmad Umar Saudi ArabiaThomas Unold GermanyVeronica Vaida USARoel van De Krol GermanyMark van Der Auweraer BelgiumRienk Van Grondelle The NetherlandsWilfried Van Sark The NetherlandsSheng Wang ChinaXuxu Wang ChinaMingkui Wang ChinaEzequiel Wolcan ArgentinaMan Shing Wong Hong KongDavid Worrall UKJeffrey C S Wu TaiwanYanfa Yan USAJiannian Yao ChinaMinjoong Yoon Republic of KoreaJiangbo Yu USAHongtao Yu USAYing Yu ChinaKlaas Zachariasse GermanyJuan Antonio Zapien Hong KongTianyou Zhai ChinaLizhi Zhang ChinaGuijiang Zhou ChinaYong Zhou ChinaRui Zhu China

Contents

Solar Energy and PV Systems Ismail H Altas and Adel M SharafVolume 2014 Article ID 408285 2 pages

Performance Analysis of Hybrid PVDiesel Energy System inWestern Region of Saudi ArabiaMakbul A M Ramli Ayong Hiendro and H R E H BouchekaraVolume 2014 Article ID 626251 10 pages

NewMultiphase Hybrid Boost Converter with Wide Conversion Ratio for PV SystemIoana-Monica Pop-Calimanu and Folker RenkenVolume 2014 Article ID 637468 17 pages

Reconfigurable Charge Pump Circuit with Variable Pumping Frequency Scheme for Harvesting SolarEnergy under Various Sunlight Intensities Jeong Heon Kim Sang Don Byeon Hyun-Sun Moand Kyeong-Sik MinVolume 2014 Article ID 437641 9 pages

Optimization of p-GaNInGaNn-GaN Double Heterojunction p-i-n Solar Cell for High EfficiencySimulation Approach Aniruddha Singh Kushwaha Pramila Mahala and Chenna DhanavantriVolume 2014 Article ID 819637 6 pages

Buck-BoostForward Hybrid Converter for PV Energy Conversion Applications Sheng-Yu TsengChien-Chih Chen and Hung-Yuan WangVolume 2014 Article ID 392394 14 pages

ANovel Parabolic Trough Concentrating Solar Heating for Cut Tobacco Drying System Jiang Tao LiuMing Li Qiong Fen Yu and De Li LingVolume 2014 Article ID 209028 10 pages

Very Fast and Accurate Procedure for the Characterization of Photovoltaic Panels from DatasheetInformation Antonino Laudani Francesco Riganti Fulginei Alessandro Salvini Gabriele Maria Lozitoand Salvatore CocoVolume 2014 Article ID 946360 10 pages

An Improved Mathematical Model for Computing Power Output of Solar Photovoltaic ModulesAbdul Qayoom Jakhrani Saleem Raza Samo Shakeel Ahmed Kamboh Jane Labadinand Andrew Ragai Henry RigitVolume 2014 Article ID 346704 9 pages

A Virtual PV Systems Lab for Engineering Undergraduate Curriculum Emre Ozkop and Ismail H AltasVolume 2014 Article ID 895271 17 pages

Effect of Ambient Temperature on Performance of Grid-Connected Inverter Installed inThailandKamonpan Chumpolrat Vichit Sangsuwan Nuttakarn Udomdachanut Songkiate KittisontirakSasiwimon Songtrai Perawut Chinnavornrungsee Amornrat Limmanee Jaran Sritharathikhunand Kobsak SripraphaVolume 2014 Article ID 502628 6 pages

Grid Connected Solar PV System with SEPIC Converter Compared with Parallel Boost ConverterBased MPPT T Ajith Bosco Raj R Ramesh J R Maglin M Vaigundamoorthi I William ChristopherC Gopinath and C YaashuwanthVolume 2014 Article ID 385720 12 pages

ADifferentThree-Port DCDC Converter for Standalone PV System Nimrod VazquezCarlos Manuel Sanchez Claudia Hernandez Eslı Vazquez Luz del Carmen Garcıa and Jaime ArauVolume 2014 Article ID 692934 13 pages

Improved Fractional Order VSS Inc-Cond MPPT Algorithm for Photovoltaic SchemeR Arulmurugan and N SuthanthiravanithaVolume 2014 Article ID 128327 10 pages

A Simple and Efficient MPPTMethod for Low-Power PV Cells Maria Teresa Penella and Manel GasullaVolume 2014 Article ID 153428 7 pages

BICOMPPT A Faster Maximum Power Point Tracker and Its Application for Photovoltaic PanelsHadi Malek and YangQuan ChenVolume 2014 Article ID 586503 9 pages

Simulation Analysis of the Four Configurations of Solar Desiccant Cooling System Using EvaporativeCooling in Tropical Weather in Malaysia M M S Dezfouli S Mat G Pirasteh K S M Sahari K Sopianand M H RuslanVolume 2014 Article ID 843617 14 pages

Solution-Processed Bulk Heterojunction Solar Cells with Silyl End-Capped SexithiopheneJung Hei Choi Mohamed E El-Khouly Taehee Kim Youn-Su Kim Ung Chan Yoon Shunichi Fukuzumiand Kyungkon KimVolume 2013 Article ID 843615 9 pages

Maximum Power Point Tracking Method Based on Modified Particle Swarm Optimization forPhotovoltaic Systems Kuei-Hsiang Chao Long-Yi Chang and Hsueh-Chien LiuVolume 2013 Article ID 583163 6 pages

Effect of Subgrains on the Performance of Mono-Like Crystalline Silicon Solar Cells Su ZhouChunlan Zhou Wenjing Wang Yehua Tang Jingwei Chen Baojun Yan and Yan ZhaoVolume 2013 Article ID 713791 8 pages

EditorialSolar Energy and PV Systems

Ismail H Altas1 and Adel M Sharaf2

1Karadeniz Technical University 61080 Trabzon Turkey2Sharaf Energy Systems Inc Fredericton NB Canada E3C 2P2

Correspondence should be addressed to Ismail H Altas ihaltasaltasorg

Received 20 November 2014 Accepted 20 November 2014 Published 22 December 2014

Copyright copy 2014 I H Altas and A M Sharaf This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

The utilization of solar photovoltaic (PV) systems has gaineda tremendous momentum due to decreasing costs of PVarrays and interface systems by asmuch as 50during the lastfive years The advancements on electric utility grid interfacesystems andutilization of PVarrays in standalone local powergeneration and smart buildings with storage battery andback-up hybrid systems are increasing the PV system utiliza-tion as the emerging form of renewablealternative energysource In many countries the government has institutedspecial incentives and tax credits as well as feed-in tariffand energy purchase back legislation programs in order topromote and encourage manufacturers and consumers andboost new investments in solar PV energy use in differentsectors

As the solar PV systems emerge as viable and eco-nomic source of green energy with increasing installationsites every year attempts are made to find economic andtechnological solutions to the problems arising from variousaspects of the PV utilization schemes The state of the artresearch is continuing in all areas from material sciences tomanufacturing and interfacing in order to ensure efficientutilization and commercial viability in terms of cost securityand durability of PV and hybrid PV-wind-storage systemsSpecific areas focus on PV array topologies dynamic suntracking maximum power point control storage devicesand efficient decoupled interface with smart grid and smartbuilding to ensure dynamic matching of energy to loadrequirements with minimal impact on the host utility gridBesides energy management studies in smart grids anddistributed generation have become other additional areas of

demand sidemanagement and energy efficient hybrid utility-renewable energy

We invited investigators to contribute original researcharticles as well as review articles that will stimulate the contin-uing efforts and promote new research directions to addressthe undergoing challenges and technological requirements inPV systems utilization in order to ensure commercial viabilityand improve usability security reliability and integration ofsustainability of converting sun power to electricity

Hybrid PV-wind-fuel cell-microgas turbines with storageLi-ion batteries and super capacitors are promising tomodifythe way smart grid manages efficient electrical energy andensure demand-side management and peak shifting as wellas shaving of peak demand during summer months due tomassive air-conditioning loads

The inherent problems of PV interfacing include theeffects of solar insulation and temperature changes affectingthe PV array powerenergy as well as interface power qualityand required dc-ac isolation and grid supply security andreliability

The effects of mismatching conditions and partial shad-ingclouding problems require novel control and powertracking algorithms new architecture usingmulti convertersand sittinglocation dynamic exchanges of PV arrays usingseries-parallel (SP) topologies

The special edition is a collection of accepted papersfocused on photovoltaic systems emerging technology andcurrent applications including interfacing energy efficientutilization emerging technologies fabrication and new con-trol strategies for maximum power point tracking under

Hindawi Publishing CorporationInternational Journal of PhotoenergyVolume 2014 Article ID 408285 2 pageshttpdxdoiorg1011552014408285

2 International Journal of Photoenergy

contingenciesmismatching and cloudypartial shading con-ditions with PV farmpark utilization and field studies Tech-nologies using solar energy in heating and cooling systemsadvancements in manufacturing processes developmentsin power electronic devices for utility interfacing issuesshading effects maximum power point tracking algorithmsand efficient energy management for higher efficiency in PVsystems are some other topics presented in this special issue

Ismail H AltasAdel M Sharaf

Research ArticlePerformance Analysis of Hybrid PVDiesel Energy System inWestern Region of Saudi Arabia

Makbul A M Ramli1 Ayong Hiendro2 and H R E H Bouchekara3

1 Department of Electrical and Computer Engineering King Abdulaziz University Jeddah 21589 Saudi Arabia2Department of Electrical Engineering Universitas Tanjungpura Pontianak 78124 Indonesia3 Constantine Electrical Engineering Laboratory (LEC) Department of Electrical Engineering University of Constantine 125000 Constantine Algeria

Correspondence should be addressed to Makbul A M Ramli makbulanwarigmailcom

Received 29 October 2013 Accepted 24 March 2014 Published 13 May 2014

Academic Editor Adel M Sharaf

Copyright copy 2014 Makbul A M Ramli et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

The potential implementation of hybrid photovoltaic (PV)diesel energy system inwestern region of Saudi Arabia is analyzed in thispaper The solar radiation intensity considered in this study is in the range of 415ndash717 kWhm2day The HOMER software is usedto perform the technical and economical analysis of the system Three different system configurations namely stand-alone dieselsystem and hybrid PVdiesel systemwith and without battery storage element will be evaluated and discussedThe analysis will beaddressed to the impact of PV penetration and battery storage on energy production cost of energy number of operational hoursof diesel generators fuel savings and reduction of carbon emission for the given configurations The simulation results indicatethat the energy cost of the hybrid PVdieselbattery system with 15 PV penetration battery storage of 18696MWh and energydemand of 32962MWhday is $0117kWh

1 Introduction

The Kingdom of Saudi Arabia is blessed with abundantenergy resources It has the worldrsquos largest oil reserves andthe worldrsquos fourth largest proven gas reserves In addition theKingdom also has abundant wind and solar renewable energyresources However in this country the use of its renewableenergy resources to generate electricity is negligibly small andalmost all its electricity is produced from the combustion offossil fuels [1] During the last two decades electrical energyconsumption in Saudi Arabia increased significantly due torapid economic development and the absence of energy con-servation measures It is expected that peak loads will reach60GW in 2023which causes total investmentmay exceed $90billion Therefore there is an urgent need to develop energyconservation policies for sustainable development [2]

Remarkable efforts to diversify energy sources and tointensify the deployment of renewable energy options have

been increasing around the world In recent years a set ofrenewable energy scenarios for Saudi Arabia has been pro-posed to examine the prospects of renewable sources fromthe perspective of major oil producers The drive towardsrenewable energy in Saudi Arabia should not be regarded asbeing a luxury but rather amust as a sign of good governanceconcern for the environment and prudence in oil-productionpolicy [3 4]

Thefirst priority in intensifying renewable energy deploy-ment in the 21st century is the combined effects of the deple-tion of fossil fuels and the awareness of environmental deg-radation [5] Therefore policy makers and researchers arepaying more attention to research in this field For instanceAlnatheer has conducted researches on environmental im-pacts of electric energy system expansion in Saudi ArabiaIt has been concluded that the use of renewable energyand energy efficiency resources gives significant environ-mental benefits [6 7] Apart from local conservation efforts



the country has an option to reduce domestic diesel con-sumption and increase its oil exports By reducing domesticdiesel consumption subsidies can be used to promote the useof renewable energyThis in turn contributes to reducing airpollution and greenhouse gas emissions [3 8]

As one of renewable energy sources solar energy is a site-dependent inexhaustible benign (does not produce emis-sions that contribute to the greenhouse effect) and potentialsource of renewable energy that is being developed by anumber of countries with high solar radiation as an effort toreduce their dependence on fossil-based nonrenewable fuels[9] Saudi Arabia located in the heart of one of the worldrsquosmost productive solar regions receives the most potent kindof sunlight [10] With the average annual solar radiation of2200 kWhm2 in the Arabian Peninsula applications of solarenergy have been growing since 1960 [9 10] Now and inthe future exploitation of this important energy resourcebecomes more imperative for Saudi Arabia [11]

Makkah is the most populous province of Saudi ArabiaIt is located in western region of Saudi Arabia and has annualsolar radiation of 2475Wm2 There are many factors affect-ing the electricity demand in this area such as weatherchanges social life activities (work school and prayer times)and special events (Ramadan and Hajj) [12] With the highelectricity demand during both day- and nighttime replacingdiesel generators with PVbattery system is not a wisesolution Therefore very large sizes of PV and battery areneeded to meet the electricity demand otherwise electricityshortages will occur

Many researchers have reported that hybrid PVdieselbattery system is more economically viable than stand-alonediesel system [13ndash16] It is not happening in Makkah at thepresent time Operation cost for the stand-alone dieselgenerators is relatively cheap in Makkah because of the lowdiesel fuel price However diesel generators are not environ-mentally friendly Although hybrid PVdieselbattery systemis more expensive than the stand-alone diesel the hybrid sys-tem gives other various advantages such as improved relia-bility and reduced pollution and emission

In this paper a hybrid PVdiesel system is designed toreach its optimum performance to meet load demand inMakkahDiesel generators are used as a backup for the hybridsystem Minimum sizes of the hybrid system componentsrequired to achieve zero unmet electric loads are determinedusing hybrid optimization model for electric renewable(HOMER) software [17]

2 Solar Irradiance Data

Saudi Arabia is one of the driest and hottest countries in theworld The global solar irradiation in Saudi Arabia is shownin Figure 1 Either the clearness index or the solar irradiationdata can be used to represent the solar resource Basedon data from NASA surface meteorology and solar energy(httpeosweblarcnasagov) the solar irradiation inMakkah(21∘261015840 North 39∘491015840 East) is between 415 kWhm2day and717 kWhm2dayThe scaled annual average of the solar radi-ation is estimated to be 594 kWhm2day Figure 2 shows the

2150 2300 2450

(kWhm2)

0 200

(km)

Global horizontal irradiation Saudi Arabia

Average annual sum period 1999ndash2011

lt2000

Figure 1 Solar irradiation map in Saudi Arabia

8

6

4

2

0

10

08

06

04

02

00

Global horizontal radiation

Dai

ly ra

diat

ion

(kW

hm

2d

)

Clea

rnes

s ind

ex

Daily radiationClearness index

Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

rFigure 2 Solar irradiation data

solar irradiation data on the right axis is the clearness indexof the solar irradiation It is clearly shown that solar irradianceis high (above the average) in MarchndashSeptember with a peakin June while solar irradiance is low in January FebruaryOctober November and December as shown in Figure 3

3 Design and System Specifications

31 Primary Load The load demand in Makkah variesmonthly Three different reasons for increases in the loaddemand in Makkah are due to (1) special occasions (Eid al-Fitr National Day) (2) religious occasions (Hajj Ramadanand Umra) and (3) climate conditions The maximum peakload occurs in the summer season Sometimes there is anoverlapping between the summer season and the Hajj orRamadanmonth resulting in a much higher load demand forthat period

International Journal of Photoenergy 3

08

04

00

May

00 00

0 12 24 0 12 24 0 12 24

0 24 0 12 24 0 12 24

0 12 24 0 12 24 0 12 24

0 12 24 0 12 24 0 12 24

08

04

00

08

04

00

08

04

00

08

04

08

04

00

08

04

00

08

04

08

04

08

04

00

08

04

00

08

04

January February March

April June

July August September

October November December

Figure 3 Monthly solar irradiation data

25

2

15

1

05

0

times106

Load

(kW

)

Seasonal profile

MaxDaily highMeanDaily lowMin

Ann

ual

Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r

Figure 4 Monthly load profile of Makkah

Load profile ofMakkah is presented in Figure 4 From theload profile it is shown that peak load inMakkah is 2213MWwith energy consumption of 32962MWhdayThe peak loadis about 0023 or 2 hours during the year

32 Design Specification In this design the hybrid PVdieselbattery system consists of four main system components

(1) PV modules (2) storage batteries (3) diesel generatorsand (4) inverters The configuration of the hybrid PVdieselbattery system is shown in Figure 5

321 Diesel Generator (DG) Adiesel generator (DG) is char-acterized by its fuel consumption and efficiency The fuelcharacteristic describes the amount of fuel the generator


Load

Converter

AC DC

Battery

PV

AC generatorset 1

AC generatorset 2

AC generatorset 3

AC generatorset 4

AC generatorset 5

AC generatorset 6

AC generatorset 7

Figure 5 Configuration of hybrid PVdieselbattery system

Efficiency curve40

30

20

10

0

0 20 40 60 80 100

Effici

ency

()

Output ()

Figure 6 Efficiency curve

Table 1 Generator groups

Group Number of units Total capacity (MW)Generator 1 4 320Generator 2 4 320Generator 3 4 320Generator 4 4 320Generator 5 4 320Generator 6 4 320Generator 7 2 160

consumes to produce electricity The efficiency curve defineselectrical energy coming out divided by the chemical energyof fuel going in

In this design the DGs have the fuel intercept coefficientof 001609 LkWh and the fuel slope of 02486 LkWh Theefficiency curve of the DGs is shown in Figure 6 The DGs

Table 2 DG data

DGSize 80MWLifetime 15000 hrMin load ratio 40Capital cost $400kWReplacement cost $350kWOperating and maintenance cost $005hr

Table 3 PV data

PV systemSize 11ndash22 GWLifetime 20 yrDerating factor 90Capital cost $2500kWReplacement cost $2000kWOperating and maintenance cost $3yr

are used as a backup during peak demand periods whichcannot be fulfilled by PV and battery The DGs also supportthe battery at nighttime when the PV has stopped producingelectricity In order to cover the peak load of 2213MW80MWunit DG is used in the simulation There are 26DGsemployed in this design to meet the load demand Theyare distributed into 7 groups of generators as illustrated inFigure 5 Table 1 presents amount of DGs in each group TheDG cost and technical data are provided in Table 2

322 Photovoltaic (PV) Solar energy is used as the base-loadpower source PV array size is dependent on the load profilesolar radiation and renewable fraction The renewable frac-tion is the fraction of the energy delivered to the load thatoriginated from renewable power sources and in this case therenewable fraction is related to the PV production

With the peak load of 22 GW the initial PV size of22 GW is fair enough for the PVdieselbattery hybrid sys-temThePV size can be either increased or decreased accord-ing to the amounts of unmet electric load and renewablefraction set in the design This PV size will be used to caterfor the variety of load demand in a year PV array will onlygenerate electricity at daytime from 6 am to 6 pm Theexcess generated power will be used to charge the batteryThePV cost and technical data are provided in Table 3

323 Inverter The PV arrays produce direct current (DC) ata voltage that depends on the design and the solar radiationTheDC power then runs to an inverter which converts it intostandard AC voltage The inverter size is rated based on theselected PV size in order to maximize the quantity of energywhich is harvested from the PV arrays For 22 GW ratedoutput PV the inverter is rated at 22 GW to fully supply thepower from the PV However it is frequently sized below thePV rated output because the PV does not always produce itsfull rated power Smaller size inverter will minimize inverter


6

5

4

3

2

1

0

times103

times103

0 20 40 60 80 100

14

12

10

8

6

4

2

0

Cycle

s to

failu

re

Depth of discharge ()

CyclesThroughput

Life

time t

hrou

ghpu

t (kW

h)Figure 7 Lifetime curve

Table 4 Inverter data

InverterSize lt22GWLifetime 10 yrEfficiency 90Capital cost $400kWReplacement cost $375kWOperating and maintenance cost $20yr

Table 5 Battery data

BatteryType Surrette 4KS25PLifetime 12 yrBatteries per string 12Nominal voltage 4V (48V)Nominal capacity 1900AhNominal energy capacity of each battery 760 kWhCapital cost $1200quantityReplacement cost $1200quantityOperating and maintenance cost $60yr

cost but does not reduce the system performance A briefsummary on the data for inverter is provided in Table 4

324 Battery Battery is used as a storage device which hastwo operation modes charging and discharging Excess elec-tricity from PV or other sources can be stored in the batteryThe purpose of the battery is to alleviate the mismatchbetween the load demand and electricity generation

State of charge (SOC) indicates the level of battery chargeWhen the battery is fully charged the SOC level is 100Battery has its specificminimumSOCallowed to operate andit is usually recommended by the battery manufacturers

The battery chosen is Surrette 4KS25P It is a 4-volt deepcycle battery rated at 1900Ah at 100 hour rate The batteryrsquos

safe operating SOC is between 40 and 100 Lifetime ofthe battery is 12 years for operating within the safe regionIt will shorten the batteryrsquos lifetime if operated below theSOC of 40 or over DOD of 60 as shown in Figure 7 Thebattery lifetime throughput is 10569 kWh when operatedwith minimum SOC of 40 or maximum DOD of 60 Thedata for battery is provided in Table 5

4 Cost of Carbon Emissions

Carbon emissions cause economic costs of damage and re-sulting climate changeThe cost of carbon emissions is calcu-lated bymultiplying tons ofCO

2emitted for each type of plant

system by an assumed cost per ton for carbon emission Thecost per ton for carbon emissions is not set in Saudi Arabiasince there is currently no CO

2marketmechanismHowever

emission penalties can be added to analyze the total annualcost of the power system on the assumption that the penaltiesare $50t for CO

2 $900t for SO

2 $2600t for NO

119909 and

$2800t for PM [18 19]

5 Simulation Results and Discussions

Performance of the stand-alone diesel system hybrid PVdiesel system without battery and hybrid PVdiesel systemwith battery is discussed in this section Simulations for var-ious configurations are performed by considering the totalbattery storage sizes of 18696MWh for 5minautonomy(equivalent to 5min of average load) while the hourly averageload is 1373434MWhhr

51 Stand-AloneDiesel System From the simulation results itcan be found that stand-alone diesel system without renew-able penetration gives total net present cost (NPC) of$17335490560 and CO

2emission of 8460421632 kgyr

This system offers 0 for both the unmet load and excesselectricity This is according to the diesel price of $0067LThe cost of energy (COE) for this stand-alone diesel systemis $0102kWh

Monthly average electric production and cash flow sum-mary are shown in Figures 8 and 9 respectively

52 Hybrid PVDiesel System without Battery To determinethe feasibility of hybrid PVdiesel installation four configu-ration options will be analyzed

(1) option 1 PV (11 GW) with DGs(2) option 2 PV (22GW) with DGs(3) option 3 PV (33 GW) with DGs(4) option 4 PV (44GW) with DGs

521 Option 1 PV (11 GW) with DGs From the simulationresults it can be noticed that this system gives total NPC of$20139882496 and CO

2emission of 7198296576 kgyrThe

COE for this system is $0119kWh with PV penetration of15

Monthly average electric production and cash flow sum-mary are illustrated in Figures 10 and 11 respectively


2

15

1

05

0

times106

Pow

er (k

W)

Monthly average electric production

Generator 1Generator 2Generator 3Generator 4

Generator 5Generator 6Generator 7

Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r

Figure 8 DGs monthly average electric production



4

3

2

1

0

times109

Net

pre

sent

cost

($)

Cash flow summary

Gen 1 Gen 2 Gen 3 Gen 4 Gen 5 Gen 6 Gen 7

Figure 9 DGs cash flow summary

522 Option 2 PV (22 GW) with DGs From the simulationresults it can found that this system gives total NPC of$20995399680 and CO


The COE for this system is $0124kWh with PV penetrationof 26


523 Option 3 PV (33 GW) with DGs From the simulationresults it can be seen that this system gives total NPC of$22260590592 and CO




524 Option 4 PV (44 GW) with DGs From the simulationresults it can be noticed that this system gives total NPCof $23976376320 and CO





2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r

Figure 10 Monthly average electric production

35

3

25

2

15

1

05

0

times109

Net

pre

sent

cost

($)



Cash flow summary

Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

PV

PV

Converter

Con

vert

er

Figure 11 Cash flow summary



2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r



3

2

1

0

times109

Net

pre

sent

cost

($)



Cash flow summary

Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

PV

PV

Converter

Con

vert

er

6

5

4




25

2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r


2

0

times109

Net

pre

sent

cost

($)

Cash flow summary



Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

PV

PV

Converter

Con

vert

er

8

6

4




25

2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r


2

0

times109

Net

pre

sent

cost

($)

Cash flow summary



Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

PV

PV

Converter

Con

vert

er

12

8

10

6

4



From the simulation results it can be found that option 1is the cheapest and theminimum system requirement tomeetall demands All of them offer the unmet load of 0 as sum-marized in Table 6

High PV penetrationmight result in difficulties in controlwhile maintaining stable voltage and frequency The level ofrenewable energy penetration in real application is generallyin the range of 11ndash25 The utilization of the bigger PV arraysize will result in a higher value of the total NPC as well asthe COE On the other hand reducing the PV size will resultin higher dependence of DGs and give more CO

2emission

Therefore the use of PV array size between 11 and 22GW isjustified

53 Hybrid PVDiesel System with Battery From the simula-tion results it can be seen that this system gives total NPCof $19849900032 and CO



Table 6 Hybrid PVdiesel system without battery performance

Config Unmet load () Excess elect () NPC ($) COE ($kWh) PV penetn () CO2 emissions (kgyr)Option 1 0 094 20139882496 0119 15 7198296576Option 2 0 609 20995399680 0124 26 6276211200Option 3 0 1460 22260590592 0131 32 5742476288Option 4 0 2320 23976376320 0141 36 5408787456



2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r


2

15

05

1

0

times109

Net

pre

sent

cost

($)

Cash flow summary



Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

S4KS

25P

PV

PV

Surrette 4KS25PConverter

Con

vert

er

35

3

25



From the previous results it is shown that PV array sizebetween 11 and 22GW offers satisfying options To deter-mine the feasibility of hybrid PVdiesel with battery instal-lation two options of configurations are considered

(1) option 1 PV (11 GW) with battery and DGs(2) option 2 PV (22GW) with battery and DGs



2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r


531 Option 1 PV (11 GW) with Battery and DGs Monthlyaverage electric production and cash flow summary areillustrated in Figures 18 and 19 respectively

532 Option 2 PV (22 GW) with Battery and DGs Fromthe simulation results it can be found that this systemgives total NPC of $20690055168 and CO

2emission of

6247786496 kgyr The COE for this system is $0122kWhwith the PV penetration of 26


The summaries of hybrid PVdiesel system with batteryare presented in Table 7

54 Comparing Designs The hybrid PVdiesel system usingPV array size of 11 GWgives 15 renewable penetrationThispenetration value makes sense for real-world applicationFurther the utilization of PV array size more than 11 GW isout of consideration since it would result in higher values ofthe total NPC as well as the COE In addition higher con-tribution of renewable energy penetration might give prob-lems related to system instability

The summaries of the stand-alone diesel system hybridPVdiesel system without battery and hybrid PVdiesel sys-tem with battery are presented in Table 8 By using the pro-posed hybrid PVdiesel systemwithout battery the total NPCis $20139882496 This system is the most expensive systemconfiguration as can be seen in Table 8 One of the main rea-sons is that the power generated by PV is not being fully util-ized If there are no storage devices the excess solar electricity


Table 7 Performance of hybrid PVdiesel system with battery

Config Unmet load () Excess elect () NPC ($) COE ($kWh) PV penetn () CO2 emissions (kgyr)Option 1 0 084 19849900032 0117 15 7176592896Option 2 0 593 20690055168 0122 26 6247786496

Table 8 Stand-alone diesel and hybrid PVdiesel with and without battery

Config Unmet load () Excess elect () NPC ($) COE ($kWh) PV penetn () CO2 emissions (kgyr)Diesel 0 000 17335490560 0102 0 8460421632PVdiesel 0 094 20139882496 0119 15 7198296576PVdieselbattery 0 084 19849900032 0117 15 7176592896

3

2

1

0

times109

Net

pre

sent

cost

($)

Cash flow summary



Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

S4KS

25P

PV

PV


Con

vert

er5

4

6


cannot be stored and is considered as losses When the PVcannot meet the load demand the DGs will be then operatedto cope for the demand The yearly load has to be providedby DGs if the PVdiesel system does not use battery storages[20]

Although storage devices are typically very expensivethey are very important to ensure that the excess electricityproduced fromPV can be stored for later use It would greatlyoptimize the systemand as a result the PVdiesel systemwithbattery is less expensive than the PVdiesel system withoutbattery

The COE of hybrid PVdieselbattery system (15 PVpenetration) with 5-minute battery autonomy is $0117kWh(diesel fuel price of $0067L) This value is lower than theCOE of hybrid PVdiesel system under similar conditionwhich is $0119kWh As reported in some pieces of literaturethe COE of PVdiesel system in some countries is in the rangeof $022ndash096kWh [14 21ndash27]TheCOEvaries depending ondiesel fuel prices PV penetrations and interest rates

The present cost of electricity production by diesel powerplant in Makkah is about $010kWh This cost of electricityproduction matches the simulation result for the stand-alonediesel as shown in Table 8 The average electricity price for

residence in Makkah is about $00133ndash00693kWh (1 $ =375 SAR) Residential buildings consume most of electricityin Saudi Arabia and estimated 45ndash47 of the total electricalenergy generated in the country [20 28] The difference be-tween cost and price is paid from the government resourceswhich subsidize fuel and electricity prices

As shown in Table 8 the stand-alone diesel system ischeaper than the hybrid PVdiesel system either with orwithout battery for application inMakkah It is because of thecheap subsidized diesel fuel price in Makkah

By renewable energy penetration of 15 (as in hybridPVdiesel with battery) the use of diesel fuel can be reducedfrom 3212823296 Lyr to 2725292544 Lyr In this case thecountry can save 487530752 L of diesel fuel per year Fromenvironmental viewpoint the use of hybrid PVdieselsystem will significantly reduce CO

2emission from

8460421632 kgyr to 7176592896 kgyr

6 Conclusion

The HOMER software has simulated three different systemconfigurations namely stand-alone diesel system hybridPVdiesel system and hybrid PVdieselbattery system fortwo options of PV array size that is 11 GW and 22GWThehybrid PVdiesel system using PV array size of 11 GW gives15 renewable penetration This penetration value makessense for the real-world application From the simulationit has been clearly demonstrated that the stand-alone dieselsystem has the lowest COE but the highest CO

2gas emission

The use of hybrid PVdiesel system will significantly reduceCO2gas emission from environmental point of view On

the other hand the configuration of hybrid PVdiesel systemwithout battery is the most expensive system One of themain reasons is that the power generated by PV is not beingfully utilized Since the storage devices are very important toensure that the excess electricity produced by PV array canbe stored for later use it would greatly optimize the systemAs a conclusion the PVdiesel system with battery is moreeconomical than the PVdiesel system without battery

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper


Acknowledgment

This paper was funded by the Deanship of Scientific Research(DSR) King Abdulaziz University JeddahTherefore the au-thors acknowledge with thanks the DSR financial support

References

[1] Y Alyousef and P Stevens ldquoThe cost of domestic energy pricesto Saudi Arabiardquo Energy Policy vol 39 no 11 pp 6900ndash69052011

[2] S A Al-Ajlan A M Al-Ibrahim M Abdulkhaleq and F Al-ghamdi ldquoDeveloping sustainable energy policies for electricalenergy conservation in Saudi Arabiardquo Energy Policy vol 34 no13 pp 1556ndash1565 2006

[3] Y Al-Saleh ldquoRenewable energy scenarios for major oil-produc-ing nations the case of Saudi Arabiardquo Futures vol 41 no 9 pp650ndash662 2009

[4] H A Saleh ldquoRenewable energy research development and ap-plications in Saudi Arabiardquo inWorld Renewable Energy CongressVI A A M Sayigh Ed chapter 345 pp 1665ndash1668 PergamonOxford UK 2000

[5] H H Chen H-Y Kang and A H I Lee ldquoStrategic selection ofsuitable projects for hybrid solar-wind power generation sys-temsrdquo Renewable and Sustainable Energy Reviews vol 14 no 1pp 413ndash421 2010

[6] O Alnatheer ldquoThe potential contribution of renewable energyto electricity supply in Saudi Arabiardquo Energy Policy vol 33 no18 pp 2298ndash2312 2005

[7] O Alnatheer ldquoEnvironmental benefits of energy efficiency andrenewable energy in Saudi Arabiarsquos electric sectorrdquo Energy Pol-icy vol 34 no 1 pp 2ndash10 2006

[8] S M Rahman and A N Khondaker ldquoMitigation measures toreduce greenhouse gas emissions and enhance carbon captureand storage in Saudi Arabiardquo Renewable and Sustainable EnergyReviews vol 16 no 5 pp 2446ndash2460 2012

[9] S M Shaahid and I El-Amin ldquoTechno-economic evaluation ofoff-grid hybrid photovoltaic-diesel-battery power systems forrural electrification in Saudi Arabia-A way forward for sustain-able developmentrdquo Renewable and Sustainable Energy Reviewsvol 13 no 3 pp 625ndash633 2009

[10] J Dargin ldquoSaudi Arabia UAE promote energy from sun andwindrdquo Oil amp Gas Journal vol 107 no 12 pp 18ndash22 2009

[11] S H Alawaji ldquoEvaluation of solar energy research and its appli-cations in Saudi Arabiamdash20 years of experiencerdquo Renewable ampsustainable energy reviews vol 5 no 1 pp 59ndash77 2001

[12] A J Al-Shareef andM F Abbod ldquoThe application of committeemachinemodel in power load forecasting for thewestern regionof Saudi Arabiardquo Engineering Sciences Journal vol 22 no 1 pp19ndash38 2011

[13] SM Shaahid andMA Elhadidy ldquoEconomic analysis of hybridphotovoltaic-diesel-battery power systems for residential loadsin hot regions-A step to clean futurerdquoRenewable and SustainableEnergy Reviews vol 12 no 2 pp 488ndash503 2008

[14] E S Hrayshat ldquoTechno-economic analysis of autonomous hy-brid photovoltaic-diesel-battery systemrdquo Energy for SustainableDevelopment vol 13 no 3 pp 143ndash150 2009

[15] D P Papadopoulos and E Z Maltas ldquoDesign operation andeconomic analysis of autonomous hybrid PV-diesel power sys-tems including battery storagerdquo Journal of Electrical Engineer-ing vol 61 no 1 pp 3ndash10 2010

[16] J Dekker M Nthontho S Chowdhury and S P ChowdhuryldquoEconomic analysis of PVdiesel hybrid power systems in dif-ferent climatic zones of South Africardquo International Journal ofElectrical Power amp Energy Systems vol 40 no 1 pp 104ndash1122012

[17] HOMER Energy httpwwwhomerenergycom[18] N Kulichenko and E EreiraCarbon Capture and Storage in De-

veloping Countries A Perspective on Barriers to DeploymentWorld Bank Washington DC USA 2012

[19] J Bergerson and L Lave ldquoThe long-term life cycle private andexternal costs of high coal usage in the USrdquo Energy Policy vol35 no 12 pp 6225ndash6234 2007

[20] SM Shaahid andMA Elhadidy ldquoOpportunities for utilizationof stand-alone hybrid (photovoltaic + diesel + battery) powersystems in hot climatesrdquo Renewable Energy vol 28 no 11 pp1741ndash1753 2003

[21] S Rehman and LM Al-Hadhrami ldquoStudy of a solar PV-diesel-battery hybrid power system for a remotely located populationnear Rafha SaudiArabiardquoEnergy vol 35 no 12 pp 4986ndash49952010

[22] K Karakoulidis K Mavridis D V Bandekas P AdoniadisC Potolias and N Vordos ldquoTechno-economic analysis of astand-alone hybrid photovoltaic-diesel-battery-fuel cell powersystemrdquo Renewable Energy vol 36 no 8 pp 2238ndash2244 2011

[23] G Rohani andM Nour ldquoTechno-economical analysis of stand-alone hybrid renewable power system for Ras Musherib inUnited Arab Emiratesrdquo Energy vol 64 pp 828ndash841 2014

[24] K Y Lau M F M Yousof S N M Arshad M Anwari andA H M Yatim ldquoPerformance analysis of hybrid photovol-taicdiesel energy system under Malaysian conditionsrdquo Energyvol 35 no 8 pp 3245ndash3255 2010

[25] M S Ismail M Moghavvemi and T M I Mahlia ldquoTechno-economic analysis of an optimized photovoltaic and diesel gen-erator hybrid power system for remote houses in a tropical cli-materdquo Energy Conversion andManagement vol 69 pp 163ndash1732013

[26] M S Adaramola S S Paul and OM Oyewola ldquoAssessment ofdecentralized hybrid PV solar-diesel power system for appli-cations in Northern part of Nigeriardquo Energy for SustainableDevelopment vol 19 pp 72ndash82 2014

[27] N E Mbaka N J Mucho and K Godpromesse ldquoEconomicevaluation of small-scale photovoltaic hybrid systems for mini-grid applications in far north Cameroonrdquo Renewable Energyvol 35 no 10 pp 2391ndash2398 2010

[28] S M Shaahid and M A Elhadidy ldquoProspects of autonomousstand-alone hybrid (photo-voltaic + diesel + battery) power sys-tems in commercial applications in hot regionsrdquo RenewableEnergy vol 29 no 2 pp 165ndash177 2004

Research ArticleNew Multiphase Hybrid Boost Converter withWide Conversion Ratio for PV System

Ioana-Monica Pop-Calimanu1 and Folker Renken2

1 Applied Electronics Department ldquoPolitehnicardquo University of Timisoara 300223 Timisoara Romania2 Power ElectronicsDepartment of Engineering ldquoJade Universityrdquo University of Applied Sciences26389 Wilhelmshaven Germany

Correspondence should be addressed to Ioana-Monica Pop-Calimanu ioanamonicapopgmailcom

Received 1 December 2013 Accepted 2 March 2014 Published 30 April 2014

Academic Editor Ismail H Altas

Copyright copy 2014 I-M Pop-Calimanu and F Renken This is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited

A new multiphase hybrid boost converter with wide conversion ratio as a solution for photovoltaic energy system is presented inthis paper To ensure that all the phases of the converter operate at the same switching frequency we use interleaving topology Theproposed converter can be used as an interface between the PV system and the DC loadinverter This multiphase converter hasthe advantage of reduced value and physical size of the input and output capacitor as well as the effort for the inductors To validatethe operation of the converter we provide the analyses and the simulation results of the converter

1 Introduction

Photovoltaic (PV) energy has attracted interest as an energysource capable of solving the problems of the energy crisisSolar PV energy is becoming an increasingly important partof the renewable energy resources It is considered one of themost promising energy resources due to its infinite powerdelivered directly for free and many other advantages Theseinclude reliability availability zero pollution or destructionof the land reasonable installation and production costlong life-span and the capability of supporting microgridsystems and connecting to electrical grids [1ndash5] One ofthe challenges in the case of grid connection applicationsor high voltage DC applications requirements is the lowvoltage of the PVmodule For this reason many PVmodulesshould be either connected in series tomeet these applicationrequirements or to use a step-up DCDC converter Thistype of converter is widely used in PV systems Theoreticallya traditional step-up converter can achieve a high step-up voltage gain with an extremely high duty ratio near to100 [6] The step-up voltage gain in practice is limited

due to the effect of power switches rectifier diodes theequivalent series resistance of inductors and capacitors andthe saturation effects of the inductors and capacitors [7] Dueto the fact that these traditional converters are not able tooperate efficient with a high duty ratio near to 100 scientificliterature presents many topologies [8ndash11] that provide a highstep-up voltage gain without an extremely high duty ratioA very good review of nonisolated high step-up DCDCconverters used in renewable energy applications is presentedin [12 13] where the advantages and disadvantages of theseconverters and the major challenges are summarized In [14]the architecture of a high step-up converter which containsseven parts including a PV module input circuit a primary-side circuit a secondary-side circuit a passive regenerativesnubber circuit a filter circuit a DC output circuit anda feedback control mechanism is presented The authorsfrom [14] for raising the voltage gain are using a coupledinductor with a low-voltage rated switch In [15] three types ofstep-up converters with high-efficiency by using PV systemwith reduced diode stresses sharing are proposed Throughthe employ of coupled inductor and switched capacitor the



UOutUIn

DO

S

D1

D2

D3

L1

L2

Figure 1 Single-phase hybrid boost DCDC converter

proposed converters attain high step-up conversion ratiowithout operating at extreme duty ratio Very often in theliterature is presented high step-up DCDC converter thatutilizes at least one coupled inductor One of these structuresis presented in [16] where the operating principle and steadystate analyses are discussed in detail and a prototype ofthe circuit is implemented Another configuration of step-up converter with coupled inductor is presented in [17] Thisconverter achieves a high step-up voltage conversion ratiowithout extreme duty ratios and the numerous turnrsquos ratiosof a coupled inductor The leakage inductor energy of thecoupled inductor is efficiently recycled to the load Otherhigh step-up DCDC converters with coupled inductor arepresented in [18ndash21]

In this paper is presented and analysed the structures ofthe step-up hybrid boost converter [7 22] or as named insome papers switched inductor boost type

One of the main disadvantages of this circuit is that theeffort for the input and output capacitor in the case of asingle-phase DCDC converter is very high This is the samewith the effort of the inductors Based on the structure ofthe hybrid boost converter built in multiphase design wepresent a method for reducing this disadvantage through anew multiphase hybrid boost converter

2 Single-Phase Hybrid Boost Converter

Figure 1 shows the step-up hybrid boost converter whichwas introduced in [22] The hybrid boost DCDC converterconsists of a classical boost converter in which is inserted anL-switching structure The L-switching structure consists oftwo inductors and three diodes We can simply say that theinput inductor from a classical boost converter was replacedby the two inductors in the new hybrid converterThis type ofconverter provides high gain and high efficiency and is usedfor many applications such as solar cell energy conversionsystems [7] fuel cell energy conversion systems battery back-up systems for uninterruptible power supplies and highintensity discharge lamp ballast for automobile head lamps[23]

The gain of the hybrid boost converter is higher than thetraditional boost converter by a factor of (119889 + 1) Figure 2shows a comparison between the gain of the hybrid boostconverter and the traditional boost converter

Hybrid boost converter

Traditional boost converter

50

40

30

20

10

02 04 06 08 10

d

UO

utU

In

(UOutUIn) = (d + 1)(d minus 1)

(UOutUIn) = 1(d minus 1)

Figure 2 Conversion ratio of a hybrid boost and traditional boostDCDC converter

UOutuSUIn COCI

IIn iID iOP IOut

iOCiIC iS

DO

S

D1

D2

D3

L1

L2


UOutUInSwitched-on

state

Switched-offstate

UOutUIn

S

S

Figure 4 Current flow in switched-on and switched-off state of thehybrid boost DCDC converter

To ensure a better understanding this section continuesby presenting and discussing the power part of the hybridconverter which is presented in Figure 3

The power part consists of an input network with twoinductors and three diodes a step-up circuit with the powerswitch S and diode DO as well as a DC-link capacitor atthe input and output side of the converter Based only onthe circuit structure we can observe that the energy can betransferred only in one direction from input to output


i

t

IIn iIP

ton toff

i

t

i

t

iIC

TP 2TP 3TP

iL = iL1= iL2

IL AV ΔiL

Figure 5 Current waveforms at the input of a single-phase hybridboost DCDC converter (119889 = 03)

02 04 06 08 10

d

12

10

08

06

04

02

PInP

InR

(UOutUIn) = 15 (UOutUIn) = 3 (UOutUIn) = 9

Figure 6 Maximum power transfer in case of fixed output voltage

07

06

05

04

03

02

01

02

d

04 06 08 10

ΔiL max = 05 middot IL AVΔiL max = 0

IICIL

AV

Figure 7 RMS current in the input capacitors of a single-phasehybrid boost DCDC converter

i

i

i

t

t

t

iOP IOut

ton toff iOC

1

TP 2TP 3TP

iL = iL1= iL2

IL AV ΔiL

Figure 8 Current waveforms at the output of a single-phase hybridboost DCDC converter (119889 = 03)

07

06

05

04

03

02

01

02

d

04 06 08 10


IO

CIL

AV

Figure 9 RMS current in the output capacitors of a single-phasehybrid boost DCDC converter

For operation of the converter we use a pulse controlmethod Figure 4 shows the current flow in the circuit in theswitched-on and the switched-off state

In the switched-on state the inductors are connected inparallel and the current of both inductors flows in the inputphase and in the switch element For this reason during theswitched-on state the current in the output phase 119894OP is zero(Figure 8) In the switched-off state the inductance currentflows in series through the inductors and the output diode Inthis moment the inductors are connected in series while theinput phase current 119894IP and output phase current 119894OP are thesame as the inductor currentThe expectation being that onlythe average value of the input phase current would flow in theconverter input current 119868In while theAC-current component


would flow in to the capacitor 119862I and the average value of theoutput phase current would flow in the converter output 119868Out

For calculations we assumed that all elements of theconverterwouldworkwithout losses the voltage and currentsat the input and output were ideally DC-values and theconverter would be controlled with pulse width modulation[24] With these requirements in the converter switchingstates the voltage at the inductors can be determined byapplying the voltage-second balance on the inductor asfollows

119905on 119906119871= 119880In

119905off 119906119871= minus

119880Out minus 119880In2

(1)

where 119906119871is the inductor voltage119880In is the input voltage and

119880Out is the output voltageAs we can see from the above formulas the voltage at the

inductors is positive in the switched-on state and negative inthe switched-off state In circuit operation the positive andnegative voltage-time-area at the inductors must always bethe same With this condition the conversion ratio of thehybrid boost DCDC converter can be calculated Consider

119880Out =1 + 119889

1 minus 119889

sdot 119880In with 119889 =119905on119879P 1 minus 119889 =

119905off119879P (2)

21 DCDC Converter Input Circuit In Figure 5 the currentwaveforms at the input of the hybrid boost converter ispresented The first waveform shows the currents in bothinductors andwe can see that they are the sameThe inductorcurrent rises in switched-on state and fall in switched-offstate so that a triangular shape current is generated Themiddle waveform presents the input phase current 119894IP andthe lower waveform presents the input capacitor current 119894IC

The average current 119868119871AV depends on the duty cycle 119889

Consider

119868119871AV =

119868In1 + 119889

(3)

In order to calculate the required inductance and capaci-tance in the circuit we needed to fix conditions for the circuitoperation We first needed to know the rated output voltageas well as the minimal input voltage at rated power whichshould be transferred With these conditions the conversionratio of the circuit for rated input power 119875In119877 transfer is fixed

In this operation point the DC-input current and theaverage current 119868

119871AV in the inductors can be calculated Itis assumed that this average current 119868

119871AV is the maximumDC-value in the inductances Figure 6 shows the possiblepower transfer dependent on the duty cycle for three differentconverter designs The inductivity must be calculated for themaximum inductor voltage-time-area within the pulse peri-ods and the maximum acceptable current variation duringthis time In the switched-on state 119905on the input voltage 119880Inis connected at the inductors The voltage 119880In is described in

formula (5) as a function of the output voltage and the dutycycle Consider

1198711= 1198712=

119880In sdot 119905onΔ119894119871

(4)

1198711= 1198712=

119880Out sdot 119879PΔ119894119871

sdot

119889 sdot (1 minus 119889)

1 + 119889

(5)

Considering the full duty cycle range the voltage-time-area within the pulse periods reaches its maximum for a dutycycle of approximately 41 For this duty cycle in generalthe converter inductance is specified The maximum accept-able current variation Δ119894

119871 max is chosen normally between10 and 30 of the rated average inductance current 119868

119871AV119877These inductances can be realized as individual or mutualcoupled inductances

1198711= 1198712=

119880Out sdot 119879PΔ119894119871 max

sdot (3 minus 2 sdot radic2)

with Δ119894119871 max = (01ndash03) sdot 119868119871AV119877

(6)

We continue with the calculation of the capacity of theinput sideWe beginwith theworst case scenariowhich couldhappenwhen the converter input current is an ideally averagevalue and the overall AC-current of the input phase flows inthe capacitor (see Figure 4)This AC-current in the capacitorproduces an AC-voltage at the input that is overlaid with theDC-input voltage For this reason the maximum acceptablevoltage variation at the capacitor must be chosen during thecapacity dimension Consider

119862I =119894C sdot 119905onΔ119906C

119862I =119868119871AV sdot 119879PΔ119906Inmax

sdot 119889 sdot (1 minus 119889)

(7)

The current-time-area depends on the duty cycle havingits maximum at 50 and at the rated inductance current119868119871AV In practice the acceptable static voltage variation at theconverter input is chosen smaller than 1 of the rated inputvoltage 119880In119877

119862I =119868119871AV119877 sdot 119879P4 sdot Δ119906Inmax

with Δ119906Inmax le 001 sdot 119880In119877 (8)

In the input of the DCDC converters electrolytic capac-itors are often used [25] The main design criteria for thesecapacitors are the RMS current loads The RMS current inthe input capacitor is calculated as a function of the dutycycle for an average inductance current 119868

119871AV and amaximumcurrent variation Δ119894

119871max Consider

119868IC

= 119868119871AV

sdot radic119889 sdot (1 minus 119889) + (

Δ119894119871max119868119871AV

)

2

sdot

1198892sdot (1 minus 119889)

2sdot (1 + 3 sdot 119889)

12 sdot (1 + 119889)2sdot (3 minus 2 sdot radic2)

(9)


The result can be split into two One is dependent onthe average current 119868

119871AV and the other is dependent on thetriangle current variation in the inductances In Figure 7the current in the RMS capacitor is shown The maximumvalue of the capacitor current is higher than the half averageinductance current 119868

119871AV The current variation has howeveronly a small influence on the total RMS current in thecapacitor

22 DCDC Converter Output Circuit Figure 8 shows thecurrent and voltage waveforms at the output of the hybridboost DCDC converter

The first waveform in Figure 8 shows the current inboth inductors The middle waveform presents the outputphase current 119894OPThe lower waveform shows the AC-currentin the output capacitor This current can be calculated bysubtraction of output phase current 119894OP and output current119868Out

We continue with the calculation for the capacity of theoutput side We begin with the worst case scenario whichwould happenwhen the converter output current is an ideallyaverage value and the overall AC component of the outputphase current flows in the capacitor (see Figure 8) This AC-current in the capacitor produces an AC-voltage at the outputthat is overlaid with the DC-output voltage For this reasonthe maximum acceptable voltage variation at the capacitormust be chosen during the capacity dimension The current-time-area in the capacitor within the pulse periods can bedescribed as a function of the output current and the dutycycle Consider

119862O =119868Out sdot 119879PΔ119906Out max

sdot 119889

119862O =119868119871AV sdot 119879PΔ119906Out max

sdot 119889 sdot (1 minus 119889)

(10)

The current-time-area depends on the duty cycle withits maximum at 50 and at rated inductance current 119868

119871AVIn practice the acceptable static voltage variation is chosensmaller than 1 of the nominal output voltage119880Out Consider

119862O =119868119871AV119877 sdot 119879P

4 sdot Δ119906Out maxwith Δ119906Out max le 001 sdot 119880Out119877 (11)

Also on the output side of the hybrid boost converterselectrolytic capacitors are often used For design of thecapacitors the RMS current in the output must be calculatedIn the next formula the RMS current is determined as afunction of the duty cycle for an average inductance current119868119871AV and a maximum inductance current variation Δ119894

119871 maxConsider119868OC

= 119868119871AV


Δ119894119871max119868119871AV

)

2

sdot

1198892sdot (1 minus 119889)

3


(12)

The result can also be split into two one is dependent onthe average inductor current 119868

119871AV and the other is dependenton the triangle current variation in the inductances InFigure 9 the current in the RMS capacitor is shown Themaximum value is more than the half average inductancecurrent 119868

119871AV The current variation also has only a smallinfluence on the total RMS current in the capacitor

The effort for the input and output capacitor in the caseof a single-phase DCDC converter is very high In thenext section a method for reducing the capacitor currents ispresented With this method the effort for the inductors canalso be reduced

3 Multiphase DCDC Converter

Several solutions and their control were presented in theliterature for interconnection of the converters [26ndash32]Multiphase configuration is one of them [33ndash35] Althoughthe physical connection of the multiphase looks exactly thesame like parallel connection the main difference betweenthem is the method of how they timecontrol their mainswitches The parallel configuration is operated by havingthe switching signals for main switches coincide with eachother (eg when switch from first power converter turns onso does the switch from the second power converter andvice versa) The main advantage of parallel configuration isthat the control circuit must only provide a single switchingsignal

Inmultiphase converter each switch operates at a differenttime with a phase shift between the switch gate drivers [36]but with common frequency The result of the phase delayallows multiphase configurations to exhibit higher overallefficiency due to ripple cancelation smaller output filterrequirements and smaller output voltage ripple [37ndash40]The output voltage ripple is reduced because the outputfrequency is increased by the number of phase times ofthe individual switch frequency Higher output frequencymakes the output ripple easier to filter which allows smallercomponents and further increase in efficiency [36] In [41]is presented another two-phase boost converter topologywhich due to the fact that the multiphase configurationat the output the circuit will have just a single capacitorand the output voltage will have ripple component twicethe operating switching frequency of each individual boostconverter The frequency multiplication effect also occurs atthe input side of the converter and will reduce the inputfilters and will improve the quality of the input current[41]

The hybrid boost DCDC converter presented in thispaper can also be built in multiphase design Therefore thedifferent phases are connected at a common input and outputcapacitor The total current is subdivided in the differentphases By interleaving switching topology the AC-currentsin the input 119862I and output capacitor 119862O can be reducedclearly In addition the frequency of the capacitor currentsis increased With suitable circuit design the effort of theinductances can be decreased Figure 10 shows a DCDCconverter in two-phase design


uS2

iIP2

iOP2

iS1

UOutUIn

iIP iOP1iOP IOut

iOCiIC

IIn

Phase 1

Phase 2

iIP1

D11

D31

D21 uS1S1

DO1

CI CO

DO2

D12

D32iS2

S2D22

L11

L12

L22

L21

Figure 10 Two-phase hybrid boost DCDC converter

i

i

i

i

i

i

i

t

t

t

t

t

t

ton toff

ton toff iIP1

iIP

iIP2

IIn

iIC

TP2

TP 2TP 3TP

iL11= iL21

IL119899 AV ΔiL119899

IL119899 AV ΔiL119899iL12

= iL22

Figure 11 Current waveforms at the input of a two-phase hybridboost DCDC converter (119899 = 2 119889 = 03)

07

05

03

02

01

d

1 phase

2 phase

3 phase

02 04

04

06

06

08 10

ΔiL119899 max = 05 middot IL119899 AVΔiL119899 max = 0

IICIL1

AV

Figure 12 RMS current in the input capacitors of a one- two- andthree-phase hybrid boost DCDC converter

i

i

i

i

i

i

t

t

t

t

t

ton toff

ton toff iOP1

iOP2

iOP

iOC

IOut

TP2

TP 2TP 3TP

iL11= iL21

IL119899 AV

IL119899 AV

ΔiL119899

ΔiL119899iL12

= iL22

Figure 13 Current waveforms at the output of the two-phase hybridboost DCDC converter (119899 = 2 119889 = 03)


31 DCDC Converter Input Circuit In Figure 11 the currentwaveforms at the input of a two-phase hybrid boost converterare shown First the currents in both inductances andthe input current of phase one are shown Representedunderneath in green are the same currents of phase twoThe triangle shaped inductance currents of the two phasesare shifted with about a half pulse period At switched-ontime the input current of each phase is twice as small as theinductance current from a single-phase converter But in thiscircuit the overall input phase current 119868IP consists of the sumof all the input phase currentsTheoverall input phase current119868IP with theDC-component 119868In is shown in Figure 11 Also forthis circuit it is assumed that only the DC-current is flowingin the circuit input With these conditions for multiphaseconverters the function between the input current 119868In and theaverage inductance current of the individual phases 119868

119871119899AV can

be calculated Consider

119868119871119899AV =

119868In119899 sdot (1 + 119889)

(13)

The total AC-component of the overall input phasecurrent 119868IP flows in the input capacitor (Figure 11) Thiscurrent is clearly smaller in comparison to a single-phasehybrid boost converter In addition the frequency of thecapacitor current is doubled For these reasons the filter effortis reduced

We continue with the calculation of the inductances andcapacitors of a multiphase hybrid boost converter The sametechnical conditions for a single-phase converter are appliedand the calculation of inductances for an 119899-phase design canbe accomplished in the same way as for single-phase circuitsTaking into account the input current which is subdividedinto the different phases the maximum current variationin the inductances must be based on the maximum DC-current at rated power in the individual phases For designthe maximum current variation is selected between 10 and30 of the phase DC-current in the inductances at ratedpower Consider

1198711119899= 1198712119899=

119880Out sdot 119879PΔ119894119871119899max


with Δ119894119871119899max = (01ndash03) sdot 119868119871

119899AV119877

(14)

The amplitude of the triangle current variation in theinductors is dependent on the duty cycle of the converterBecause the duty cycle in all phases has the same value thecurrent variations in all inductors are equivalent Consider

Δ119894119871119899

=(119889) sdot (1 minus 119889)

(1 + 119889) sdot (3 minus 2 sdot radic2)

sdot Δ119894119871119899max (15)

The necessary input capacity of the multiphase hybridboost converter will now be calculated Compared to a single-phase design the capacitor current is reduced and the currentfrequency is increased by the number of phases As a con-sequence the current in the capacitor has a voltage variation

at the input For the capacitor design the permissible staticvoltage variation in general is selected smaller than 1 of therated input voltage Consider

119862I =119879P sdot 119868119871

119899AV

4 sdot 119899 sdot Δ119906Inmaxwith Δ119906Inmax le 001 sdot 119906In119877 (16)

In these DCDC converters electrolytic capacitors areused often A main design criterion of these capacitors is theRMS current load For this reason for multiphase DCDCconverters the RMS current in the input capacitor 119862I will becalculated Therefore the switching processes in the phasesare assumed as ideal Moreover it is accepted that the currentin the input is an ideal DC-current The worst case scenariowill happen when the capacitor is loaded with the total AC-current component

The input capacitor current for two-phase converters isrepresented in Figure 12 In the case of closer inspection thecapacitor current can be split into two different componentsa rectangle part and a triangle partThe RMS current of thesetwo components has been calculated The RMS result of therectangle portion for multiphase converters is shown in theformula below

119868ICΠ

=

1003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816

radic1198682

119871119899AV sdot 1198992sdot (119889 minus

0

119899

) sdot (

1

119899

minus 119889) if 0119899

le 119889

le

1

119899

radic1198682


1

119899

) sdot (

2

119899

minus 119889) if 1119899

le 119889

le

2

119899

radic1198682


2

119899

) sdot (

3

119899

minus 119889) if 2119899

le 119889

le

3

119899

radic1198682


119899 minus 1

119899

) sdot (

119899

119899

minus 119889) if 119899 minus 1119899

le 119889

le

119899

119899

(17)

In this formula it is necessary to consider that the averageinductance current 119868

119871119899AV becomes smaller with an increasing

number of phases For example the average inductancecurrent 119868

119871119899AV in a two-phase converter is only half as big as in

a single-phase converter With this fact in mind the rectanglecomponent of the input capacitor current can clearly bereduced with a multiphase design

In addition the RMS current of the triangle has beencalculated The next formula shows the RMS results for amultiphase circuit design Consider


07

05

03

02

01

04

06

d

02 04 06 08 10

1 phase

2 phase

3 phase


IO

CIL1

AV

Figure 14 RMS current in the output capacitors of a one- two- and three-phase hybrid boost DCDC converter

70

60

65

50

55

40

45

30

35

20

25

10

15

5

0

0 5 10 15 20 25 30 35 40 45 50

Time (ms)Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Ripple fac Harmonic Form fact

0 50m 0 60 60 60 60 60 84853m 3 1 1 11414my MinX MaxX MinY MaxY

28

Figure 15 Input voltage

Signal 1 Signal 2

Time (ms)

10

9

8

7

6

5

4

3

2

1

0

40 40020 40040 40060 40080 40100 40120

40007m40007m 0

0

0

0

0

50m50m

3320

3320

5762

5762

10 10

10

10

10

3320

3320

4709

4709

165982m165980m

1736

1736

3012

3012

817335m817335m

1736

1736

t y MinX MaxX MinY MaxY Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Harmonic Form fact Ripple fac

Figure 16 PWM pulse applied to switch one (pink signal) respective switch two (red signal)


Time (ms)

10

9

8

7

6

5

4

3

2

1

0

40 40020 40040 40060 40080 40100 40120

y MinX MaxX MinY MaxY Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Harmonic Form fact 0 050m 6973 2494 2653 6973 2494 124694m 1064 2796 340942m 1064

20

Ripple fac 904460m

Figure 17 Input current of phase one 119894IP1

Time (ms)

10

9

8

7

6

5

4

3

2

1

0

40 40020 40040 40060 40080 40100 40120

y MinX MaxX MinY MaxY Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Harmonic Form fact

00 050m 6992 2496 2655 6992 2496 905309m 124809m 1064 2801 340948m 1064

15

Ripple fac

Figure 18 Input current of phase two 119894IP2

40 40020 40040 40060 0080 40100 40120

Time (ms)

10

9

8

7

6

5

4

3

2

1

0


0 050m 10296 4990 5080 10296 4990 949092m 249503m 1018 2063 186847m 1018

3

Ripple fac

Figure 19 Overall input phase current 119894IP


3

2500

2

1500

1

500

0

40 40020 40040 40060 40080 40100 40120

Time (ms)t y MinXTrace MaxX MinY MaxY Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Harmonic Form fact

3487

3487

1871

1871

1875

1875

3487

3487

1871

1871

135711m135711m

93529m93529m

1003

1003

1864

1864

72360m72360m

1003

1003

0

0

0

0

0

0

0

0 50m50m

iL11iL21

Ripple fac (1)

(2)

iL11

iL21

Figure 20 Current through inductors 11987111and 119871

21 11989411987111

11989411987121

40 40020 40040 40060 40080 40100 40120

Time (ms)

3

2500

2

1500

1

500

0

iL12

t y MinX MaxX MinY MaxY Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Harmonic Form fact

40024m 1879 0 050m 3496 1872 1877 3496 1872 135792m 93608m 1003 1867 72343m 1003

Ripple fac

Figure 21 Current through inductor 11987112 11989411987112

40 40020 40040 40060 40080 40100 40120

Time (ms)

10

9

8

7

6

5

4

3

2

1

0


00 50m minus157172120583 5081 1251 1535 5081 1251 890262m 62531m 1227 4063 579928m 1227

6

Ripple fac

Figure 22 Output current of phase one 119894OP1


10

9

8

7

6

5

4

3

2

1

0

40 40020 40040 40060 40080 40100 40120

Time (ms)

7

y MinX MaxX MinY MaxY Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Harmonic Form fact 1828 0 50m minus157115 120583 4851 1252 1537 4851 1252 891238m 62581m 1228 3876 580038m 1228

Ripple fac

Figure 23 Output current of phase two 119894OP2

6

5500

5

4500

4

3500

3

2500

2

1500

1

500

0

40 40020 40040 40060 40080 40100 40120

Time (ms)yt MinX MaxX MinY MaxY Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Harmonic Form fact

40007m 1828 0 50m 0 9931 2502 2661 9931 2502 905742m 125112m 1063 3969 340360m 1063

8

Ripple fac

Figure 24 Overall output phase current 119894OP

3

2500

2

1500

1

500

0

40 40020 40040 40060 40080 40100 40120


40007m 2501 0 50m 0 3275 2490 2492 3275 2490 112534m 124483m 1 1315 45154m 1

iR1

Ripple fac

Figure 25 Output current 119868Out


40 40020 40040 40060 40080 40100 40120

Time (ms)

2540

2530

2520

2510

2500

2490

2480

2470

2460

2450

2440

yt MinX MaxX MinY MaxY Mean RMS Peak-peak Abs mean RMS ac Integral Ripple Harmonic Form fact 40007m 2501 0 50m 0 3275 2490 2492 3275 2490 112534m 124483m 1 1315 45154m 1

iR1

Ripple fac

Figure 26 Zoom on output current

40 41 42 43 44 45 46 47 48 49

130

120

110

100

90

80

70

60

50

40

30

20

10

0


40007m 0 50m 0 1 1315 45154m 1120029 157198 119504 119626 157198 119504 5402 5975

Ripple fac

V[ 9]

Figure 27 Output voltage 119881Out

40 40020 40040 40060 40080 40100 40120

122

121500

121

120500

120

119500

119

118500

118


40007m 0 50m 0 1 1315 45154m 1120029 157198 119504 119626 157198 119504 5402 5975

V[ 9]

Ripple fac

Figure 28 Zoom on output voltage 119881Out


119868ICΔ

=

10038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816

radicΔ1198942

119871119899max sdot119899 sdot [[2 minus (119899 + 1) sdot 119889]

2sdot (119889 minus 0119899)

3+ [0 minus (119899 + 0) sdot 119889]

2sdot (1119899 minus 119889)

3]

12 sdot (3 minus 2 sdot radic2)2

sdot (1 + 119889)2

if 0119899le 119889 le1

119899

radicΔ1198942


2sdot (119889 minus 1119899)

3+ [2 minus (119899 + 1) sdot 119889]

2sdot (2119899 minus 119889)

3]


sdot (1 + 119889)2

if 1119899le 119889 le2

119899

radicΔ1198942


2sdot (119889 minus 2119899)

3+ [4 minus (119899 + 2) sdot 119889]

2sdot (3119899 minus 119889)

3]


sdot (1 + 119889)2

if 2119899le 119889 le3

119899

radicΔ1198942

119871119899max sdot119899 sdot [[2 sdot 119899 minus (119899 + 119899) sdot 119889]

2sdot (119889 minus (119899 minus 1) 119899)

3+ [(2 sdot 119899 minus 2) minus (119899 + 119899 minus 1) sdot 119889]

2sdot (119899119899 minus 119889)

3]


sdot (1 + 119889)2

if 119899 minus 1119899le 119889 le119899

119899

(18)

The current variation Δ119894119871119899max in this formula is selected

in relation to the rated average inductance current 119868119871119899AV

during the circuit dimension This will mean that with thesame circuit dimension and increasing number of phases thecurrent variation is reduced

The geometrical addition of rectangle and triangle RMScomponents results in the total capacitor current for an 119899-phase hybrid boost DCDC converter Consider

119868IC = radic1198682

ICΠ + 1198682

ICΔ (19)

Figure 12 shows the RMS current in the input capacitorsof multiphase DCDC converters The current load of thecapacitor decreases with the increasing number of phasesThe influence of the rectangle RMS component is clearlydominant (dotted lines 119894

119871119899max = 0) The maximum capacitor

current component produced from the rectangle portion issmaller by a factor of 1119899 for 119899-phase convertersThe trianglecapacitor current is not dependent on the converter powerand has for all loads the same value In case of rated powerthe additional load from the triangle current in the capacitorsis small

32 DCDC Converter Output Circuit Figure 13 shows thecurrent waveforms at the output of a two-phase hybrid boostDCDC converter First the currents in both inductors andthe output current of phase 1 are represented In the middlethe current of phase 2 is shown The triangle inductancecurrent of the two phases is shifted within about a half pulseperiod of each other At switched-off times the inductors areconnected in series and the current flows to the output phasejust as in a single-phase converter However in this circuitthe overall output phase current 119868OP consists of the sum of allthe output phase currents The overall output phase current119868OP with the DC-component 119868Out is shown Figure 13 It isassumed that only the DC-current is flowing in the circuitoutput the total AC-component of the overall output phasecurrent 119868OP flows in the output capacitor In comparison toa single-phase hybrid boost converter this current is clearly

smaller In addition the frequency of the capacitor currenthas been doubled

The necessary output capacity of the multiphase hybridboost converter will next be calculated Compared to asingle-phase design the output capacitor current is reducedand the current frequency is increased by the number ofphases As a consequence the current in the capacitor hasa voltage variation at the output For the capacitor design thepermissible static voltage variation in general is selected atless than 1 of the rated output voltage Consider

119862O =119879P sdot 119868119871

119899AV

4 sdot 119899 sdot Δ119906Out maxwith Δ119906Out max le 001 sdot 119906Out119877

(20)

The RMS current in the output capacitor 119862O will next becalculated for multiphase hybrid boost DCDC convertersThe output capacitor current for a two-phase converter isrepresented in Figure 13 Like the input capacitor the outputcapacitor current can also be split into a rectangle anda triangle component The RMS current of the rectangleportion for multiphase converters is shown in the formulabelow

119868OCΠ

=

1003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816

radic1198682


0

119899

) sdot (

1

119899

minus 119889) if 0119899

le 119889 le

1

119899

radic1198682


1

119899

) sdot (

2

119899

minus 119889) if 1119899

le 119889 le

2

119899

radic1198682


2

119899

) sdot (

3

119899

minus 119889) if 2119899

le 119889 le

3

119899

radic1198682


119899 minus 1

119899

) sdot (

119899

119899

minus 119889) if 119899 minus 1119899

le 119889 le

119899

119899

(21)

The output capacitor RMS current that is producedby the rectangle component is exactly the same as that in


the input capacitor It is necessary to also consider that theaverage inductance current 119868

119871119899AV becomes smaller with the

increasing number of phases The next formula shows theoutput capacitor RMS component produced by the trianglepart of multiphase converters Consider

119868OCΔ =

1003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816

radicΔ1198942

119871119899max sdot

119899 sdot 1198892sdot [(119899 minus 1)

2sdot (119889 minus 0119899)

3+ (119899 minus 0)

2sdot (1119899 minus 119889)

3]

12 sdot (3 minus 2 sdot radic2)

2

sdot (1 + 119889)2

if 0119899

le 119889 le

1

119899

radicΔ1198942

119871119899max sdot

119899 sdot 1198892sdot [(119899 minus 2)

2sdot (119889 minus 1119899)

3+ (119899 minus 1)

2sdot (2119899 minus 119889)

3]


2

sdot (1 + 119889)2

if 1119899

le 119889 le

2

119899

radicΔ1198942

119871119899max sdot

119899 sdot 1198892sdot [(119899 minus 3)

2sdot (119889 minus 2119899)

3+ (119899 minus 2)

2sdot (3119899 minus 119889)

3]


2

sdot (1 + 119889)2

if 2119899

le 119889 le

3

119899

radicΔ1198942

119871119899max sdot

119899 sdot 1198892sdot [(119899 minus 119899)

2sdot (119889 minus (119899 minus 1) 119899)

3+ [119899 minus (119899 minus 1)]

2sdot (119899119899 minus 119889)

3]


2

sdot (1 + 119889)2

if 119899 minus 1119899

le 119889 le

119899

119899

(22)

The geometrical addition of rectangle and triangle RMScomponents results in the total capacitor current for 119899-phaseconverters Consider

119868OC = radic1198682

OCΠ + 1198682

OC Δ (23)

In Figure 14 the RMS current in the output capacitors ofmultiphase DCDC converters is shown The current load ofthe capacitor decreases with the increasing number of phasesIf neglecting the current variation (Δ119894

119871119899max = 0 dotted

lines) the maximum current in the output capacitor 119862O issmaller for 119899-phase converters by a factor of 1119899 comparedto a single-phase design The load in the output capacitorsproduced by the triangle current is fixed by the inductanceconverter design This capacitor current is not dependent onthe converted power If the converter is operated with ratedoutput power this current has only a small influence on theoverall capacitor current In the figure the output capacitorcurrent is represented with a current variation of Δ119894

119871119899Nmax =

05119868119871119899AV The influence of the triangle current at the output

capacitors is smaller than at the input side of the converterThe calculations indicate that the triangle current load

in the input and output capacitor for multiphase convertersis clearly reduced Putting this knowledge into practice thetriangle current in the inductances can be selected morelargely [42 43] The dynamics of the hybrid boost DCDCconverter can be improved and beyond that the inductanceand capacitor effort of the circuit can be reduced substantially

Besides the calculated current in the capacitors also isflowing harmonics current produced by switching processesof the phases [44]This current has been examined in [45ndash47]for different converters In consequence these results can alsobe used for a hybrid boost converter The additional currentcan contribute substantially to the output capacitors heatingup the capacitors during the small load of the converterHowever with higher power the calculated capacitor current

dominates A statement for all semiconductor types cannotbe made Beyond that this additional current is dependenton the pulse frequency of the converter

33 Simulation of the Two-Phase Hybrid Boost DCDC Con-verter We made this simulation of the two-phase hybridboost DCDC converter in CASPOC simulation programwith a power source not with a PV module The followingparameters are used for simulation 119875In = 300W 119880In = 60V119868In = 5A 119871

11= 11987121

= 11987112

= 11987122

= 82355 120583H 119862I =15625 120583F 119862O = 5208 120583F and 119891119904 = 50 kHz

In previous Figures 15 16 17 18 19 20 21 22 23 2425 26 27 and 28 are presented the results of simulationFigure 16 presents the PWM pulses applied to switch 1corresponding to first phase of the converter respective thePWMpulses applied to switch 2 corresponding to the secondphase of the converter Figure 17 show the input current ofphase 1 Figure 18 show the input current of phase 2 and theoverall input phase current which consists of the sum of allthe input phase currents in our case phase one and phase twoit is presented in Figure 19 It is obvious from Figure 20 thatthe currents of phase one through inductors 119871

11and 119871

21(11989411987111

11989411987121

) are equal The currents of phase two through inductors11987112and 119871

22(11989411987112

11989411987122

) are equal but here is represented justone of them in Figure 21 Output current of phase 1 respective2 and the overall output phase current which consists of thesum of all the output phase currents in our case phase oneand phase two are presented in Figures 22 23 and 24 In thelast figures are presented the output current and voltage andthe ripple of them

4 Conclusion

Beginning from a step-up hybrid boost converter which wasintroduced in [22] after analyses this circuit and seeing if it


is suitable for a PV system we realize that one of the maindisadvantages of this circuit is that the effort for the inputand output capacitor in the case of a single-phase DCDCconverter is very high This is the same with the effort ofthe inductors A method for reducing this disadvantage wasbased on the same structure of the hybrid boost converter butbuilt in a multiphase design To ensure that all the phases ofthe converter operate at the same switching frequency andwith phase-shift between them we used interleaving switch-ing strategy After we analyze this new multiphase hybridboost converter made the theoretical calculation and sketchthewaveform our presumption became trueThe effort of theinductors of the input and output capacitor is decreased andat the same time their size is reduced The frequency of thecapacitor currents is increased with the number of phase Tovalidate and confirm our theoretical calculation we simulatethis circuit in CASPOC [48] simulation program

Therefore how it is obvious from theoretical calculationandwaveforms associated to it comparedwith the simulationwaveforms we can see a very good accordance Based on thisnext step will be to put this circuit into practice



Acknowledgments

This work was partially supported by the strategic GrantPOSDRU 10715S77265 inside POSDRU Romania 2007ndash2013 cofinanced by the European Social FundmdashInvesting inPeople

References

[1] A Campoccia L Dusonchet E Telaretti and G ZizzoldquoComparative analysis of different supporting measures for theproduction of electrical energy by solar PV and Wind systemsfour representative European casesrdquo Solar Energy vol 83 no 3pp 287ndash297 2009

[2] T Andrejasic M Jankovec and M Topic ldquoComparison ofdirect maximum power point tracking algorithms using en50530 dynamic test procedurerdquo IET Renewable Power Genera-tion vol 5 no 4 pp 281ndash286 2011

[3] S SWWalkerNK SooriyaarachchiND B Liyanage P AGS Abeynayake and S G Abeyratne ldquoComparative analysis ofspeed of convergence ofMPPT techniquesrdquo inProceedings of the2011 6th International Conference on Industrial and InformationSystems (ICIIS rsquo11) pp 522ndash526 Kandy Sri Lanka August 2011

[4] M Liserre T Sauter and J Y Hung ldquoFuture energy systemsintegrating renewable energy sources into the smart powergrid through industrial electronicsrdquo IEEE Industrial ElectronicsMagazine vol 4 no 1 pp 18ndash37 2010

[5] J M Carrasco L G Franquelo J T Bialasiewicz et al ldquoPower-electronic systems for the grid integration of renewable energysources a surveyrdquo IEEE Transactions on Industrial Electronicsvol 53 no 4 pp 1002ndash1016 2006

[6] A Das V K Rajput A Chakraborty M Dhar S Ray andR Dutta ldquoA new transformerless DC-DC converter with high

voltage gainrdquo in Proceedings of the 2010 International Conferenceon Industrial Electronics Control andRobotics (IECR rsquo10) pp 91ndash94 December 2010

[7] O Abdel-Rahim M Orabi E Abdelkarim M Ahmed andM Z Youssef ldquoSwitched inductor boost converter for PVapplicationsrdquo in Proceedings of the 27th Annual IEEE AppliedPower Electronics Conference and Exposition (APEC rsquo12) pp2100ndash2106 February 2012

[8] R J Wai C Y Lin R Y Duan and Y R Chang ldquoHigh-efficiency dc-dc converter with high voltage gain and reducedswitch stressrdquo IEEE Transactions on Industrial Electronics vol54 no 1 pp 354ndash364 2007

[9] B R Lin and F Y Hsieh ldquoSoft-switching zeta-flyback converterwith a buck-boost type of active clamprdquo IEEE Transactions onIndustrial Electronics vol 54 no 5 pp 2813ndash2822 2007

[10] K C Tseng and T J Liang ldquoNovel high-efficiency step-upconverterrdquo IEE Proceedings Electric Power Applications vol 151no 2 pp 182ndash190 2004

[11] T F Wu Y S Lai J C Hung and Y M Chen ldquoBoost converterwith coupled inductors and buck-boost type of active clamprdquoIEEE Transactions on Industrial Electronics vol 55 no 1 pp154ndash162 2008

[12] W Li X Lv Y Deng J Liu and X He ldquoA review of non-isolated high step-up DCDC converters in renewable energyapplicationsrdquo in Proceedings of the 24th Annual IEEE AppliedPower Electronics Conference and Exposition (APEC rsquo09) pp364ndash369 IEEE February 2009

[13] A Tomaszuk and A Krupa ldquoHigh efficiency high step-upDCDC convertersmdasha reviewrdquo Bulletin of the Polish Academyof Sciences Technical Sciences vol 59 no 4 pp 475ndash483 2011

[14] R JWaiW HWang and C Y Lin ldquoHigh-performance stand-alone photovoltaic generation systemrdquo IEEE Transactions onIndustrial Electronics vol 55 no 1 pp 240ndash250 2008

[15] V M E Thirumurugan and S B E Sugumar ldquoHigh-efficiencystep-up converter by using PV system with reduced diodestresses sharingrdquo International Journal of Engineering andAdvanced Technology vol 4 no 2 pp 319ndash324 2014

[16] Y P Hsieh J F Chen T J Liang and L S Yang ldquoNovelhigh step-up dc-dc converter for distributed generation systemrdquoIEEE Transactions on Industrial Electronics vol 60 no 4 pp1473ndash1482 2013

[17] S M Chen T J Liang L S Yang and J F Chen ldquoA safetyenhanced high step-up DC-DC converter for AC photovoltaicmodule applicationrdquo IEEE Transactions on Power Electronicsvol 27 no 4 pp 1809ndash1817 2012

[18] S Y Tseng andH YWang ldquoA photovoltaic power system usinga high step-up converter forDC load applicationsrdquoEnergies vol6 no 2 pp 1068ndash1100 2013

[19] S M Chen T J Liang L S Yang and J F Chen ldquoAcascaded high step-up DC-DC converter with single switchfor microsource applicationsrdquo IEEE Transactions on PowerElectronics vol 26 no 4 pp 1146ndash1153 2011

[20] Y P Hsieh J F Chen T J Liang and L S Yang ldquoA novelhigh step-up DC-DC converter for a microgrid systemrdquo IEEETransactions on Power Electronics vol 26 no 4 pp 1127ndash11362011

[21] K Varuna and S Kumar ldquoModeling amp simulation of gridconnected photovoltaic system using high step up DCDCconverterrdquo International Journal of Latest Advances in Electronicand Electric Engineering vol 2 no 2 pp 153ndash158 2013


[22] B Axelrod Y Berkovich and A Ioinovici ldquoSwitched-capacitorswitched-inductor structures for gettingtransformerless hybrid DC-DC PWM convertersrdquo IEEETransactions on Circuits and Systems I Regular Papers vol 55no 2 pp 687ndash696 2008

[23] L Huber and M M Jovanovic ldquoDesign approach for serverpower supplies for networking applicationsrdquo in Proceedings ofthe 15th Annual IEEE Applied Power Electronics Conference andExposition (APEC rsquo00) pp 1163ndash1169 February 2000

[24] F Renken and V Karrer ldquoDCDC converters for automotiveboard-net structuresrdquo in Proceedings of the International PowerElectronics and Motion Control Conference (EPE-PEMC rsquo04)Riga Latvia September 2004

[25] F Renken ldquoHigh temperature electronics for future hybriddrive systemsrdquo inProceedings of the 13th EuropeanConference onPower Electronics and Applications (EPE rsquo09) pp 1ndash7 BarcelonaSpain September 2009

[26] L Huber and M M Jovanovic ldquoDesign approach for serverpower supplies for networking applicationsrdquo in Proceedings ofthe 15th Annual IEEE Applied Power Electronics Conference andExposition (APEC rsquo00) pp 1163ndash1169 New Orleans La USAFebruary 2000

[27] K P Yalamanchili M Ferdowsi and K Corzine ldquoNew dou-ble input DC-DC converters for automotive applicationsrdquo inProceedings of the 2006 IEEE Vehicle Power and PropulsionConference (VPPC rsquo06) Windsor UK September 2006

[28] M Gavris O Cornea and N Muntean ldquoMultiple input DC-DC topologies in renewable energy systemsmdasha general reviewrdquoin Proceedings of the 3rd IEEE International Symposium onExploitation of Renewable Energy Sources (EXPRES rsquo11) pp 123ndash128 Subotica Serbia March 2011

[29] M J V Vazquez J M A Marquez and F S Manzano ldquoAmethodology for optimizing stand-alone PV-system size usingparallel-connected DCDC convertersrdquo IEEE Transactions onIndustrial Electronics vol 55 no 7 pp 2664ndash2673 2008

[30] T Hosi ldquoControl circuit for parallel operation of self-commutating invertersrdquo Japanese Patent S56-13101 1975

[31] R Gules J D P Pacheco H L Hey and J Imhoff ldquoAmaximumpower point tracking system with parallel connection forPV stand-alone applicationsrdquo IEEE Transactions on IndustrialElectronics vol 55 no 7 pp 2674ndash2683 2008

[32] S V G Oliveira and I Barbi ldquoA three-phase step-up DC-DCconverter with a three-phase high frequency transformerrdquo inProceedings of the IEEE International Symposium on IndustrialElectronics 2005 (ISIE rsquo05) vol 2 pp 571ndash576 June 2005

[33] PThounthong and B Davat ldquoStudy of a multiphase interleavedstep-up converter for fuel cell high power applicationsrdquo EnergyConversion and Management vol 51 no 4 pp 826ndash832 2010

[34] G Y Choe J S Kim H S Kang and B K Lee ldquoAn optimaldesign methodology of an interleaved boost converter forfuel cell applicationsrdquo Journal of Electrical Engineering andTechnology vol 5 no 2 pp 319ndash328 2010

[35] H B Shin J G Park S K Chung H W Lee and T A LipoldquoGeneralised steady-state analysis of multiphase interleavedboost converter with coupled inductorsrdquo IEE ProceedingsElectric Power Applications vol 152 no 3 pp 584ndash594 2005

[36] T Soepraptob S Sidopeksoc and M Taufik ldquoA Comparisonof multiple boost configurations for small renewable energybattery storage systemsrdquo in Proceedings of the InternationalConference on Sustainable Energy Engineering and Applicationpp 93ndash96 Yogyakarta Indonesia 2012

[37] W Xiao B Zhang and D Qiu ldquoAnalysis and design of anautomatic-current-sharing control based on average-currentmode for parallel boost convertersrdquo in Proceedings of theCESIEEE 5th International Power Electronics and Motion Con-trol Conference (IPEMC rsquo06) pp 1293ndash1297 August 2006

[38] T Taufik R Prasetyo D Dolan and D Garinto ldquoA newmultiphase multi-interleaving buck converter with bypass LCrdquoin Proceedings of the 36th Annual Conference of the IEEEIndustrial Electronics Society (IECON rsquo10) pp 291ndash295 PhoenixAriz USA November 2010

[39] Z Lukı S M Ahsanuzzaman A Prodı and Z Zhao ldquoSelf-tuning sensorless digital current-mode controller with accuratecurrent sharing formulti-phaseDC-DCconvertersrdquo inProceed-ings of the 24th Annual IEEE Applied Power Electronics Confer-ence and Exposition (APEC rsquo09) pp 264ndash268Washington DCUSA February 2009

[40] P L Wong P Xu B Yang and F C Lee ldquoPerformanceimprovements of interleaving VRMs with coupling inductorsrdquoIEEE Transactions on Power Electronics vol 16 no 4 pp 499ndash507 2001

[41] T Taufik T Gunawan D Dolan and M Anwari ldquoDesignand analysis of two-phase boost DC-DC converterrdquo in WorldAcademy of Science Engineering and Technology pp 912ndash9162010

[42] F Renken V Karrer and P Skotzek P ldquoThe starter generatormdashsystems functions and componentsrdquo in Proceedings of the 30thFISITA World Automotive Congress Barcelona Spain May2004

[43] V Karrer and F Renken ldquoPower electronics for the integratedstarter generatorrdquo in Proceedings of the Optimization of thePower Train in Vehicles by Using the Integrated Starter Generator(ISG rsquo02) pp 222ndash247 Haus der Technik e V MunichGermany 2002

[44] F Renken ldquoAnalytic calculation of the DC-link capacitorcurrent for pulsed single-phase H-bridge invertersrdquo EuropeanPower Electronics and Drives Journal vol 13 no 4 pp 13ndash192003


[46] F Renken ldquoAnalyses of the DC-link current in discontinuousmodulated three-phase invertersrdquo in Proceedings of the 12thInternational Power Electronics and Motion Control Conference(EPE-PEMC rsquo06) Portoroz Slovenia August-September 2006

[47] F Renken ldquoDC-link current in pulsed H-bridge invertersrdquo inProceedings of the International Exhibition amp Conference forPower Electronics Intelligent Motion and Power Quality (PCIMrsquo09) Nurnberg Germany May 2009

[48] ldquoCASPOC Simulation programrdquo httpwwwcaspoccom

Research ArticleReconfigurable Charge Pump Circuit with Variable PumpingFrequency Scheme for Harvesting Solar Energy under VariousSunlight Intensities

Jeong Heon Kim Sang Don Byeon Hyun-Sun Mo and Kyeong-Sik Min

School of Electrical Engineering Kookmin University Seoul 136-702 Republic of Korea

Correspondence should be addressed to Kyeong-Sik Min mkskookminackr

Received 13 November 2013 Revised 1 March 2014 Accepted 1 March 2014 Published 15 April 2014


Copyright copy 2014 Jeong Heon Kim et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

We propose variable pumping frequency (VPF) scheme which is merged with the previous reconfigurable charge pump (RCP)circuit that can change its architecture according to a given sunlight condition Here merging the VPF schemewith the architecturereconfiguration can improve percentage output currents better by 214 and 224 than RCP circuit with the fixed pumpingfrequencies of 7MHz and 15MHz respectively Comparing the VPF scheme with real maximum power points (MPP) the VPFcan deliver 919 of the maximum amount of output current to the load on average In terms of the power and area overheads theVPF scheme proposed in this paper consumes the power by 04 of the total power consumption and occupies the layout area by161 of the total layout area

1 Introduction

Solar energy is one of the most popular energy sources thatcan be harvested from environment because of its ubiquitousavailability and high power density and that DC voltage andcurrent can be directly obtained from sunlight [1ndash3] Due tothese advantages that are mentioned just earlier harvestingsolar energy becomes more popular and useful in variousapplications such as wireless sensor networks (WSN) [4 5]Among various solar energy harvesting systems in this paperwe consider a simple capacitor-based charge pump circuitfor harvesting a small amount of solar energy [6 7] Thecharge pump circuit with small-scale energy harvesting canbe implemented with low cost and simple architecture thusit can be suitable to tiny low-power and low-cost systems suchas WSN systems

One problem in developing solar energy harvesting sys-tems is that sunlight is not static but changes dynamicallyAs a result of this dynamic variation of sunlight an amountof solar energy that can be converted to electrical voltageand current by energy harvesting systems can be changedWhen sunlight is very strong the amount of solar energy

that can be converted to electrical voltage and current islarge On the contrary for weak sunlight the amount ofsolar energy is small thereby the converted amounts ofvoltage and current become small too Figure 1(a) shows thecurrent-voltage relationship of solar cell for various sunlightconditions For the sunlight intensity as strong as 38mAmaximum power point (MPP) that means a bias point thatcan deliver maximum amount of current-voltage product ofsolar cell is given at119881SC = 297Vand 119868SC = 327mAHere119881SCand 119868SC are the solar cellrsquos voltage and current respectivelyIf the sunlight intensity becomes as small as 2mA in weaksunlight condition the MPP moves to 119881SC = 181V and119868SC = 161mA as shown in Figure 1(a) Figure 1(b) showsthe relationship of solar cellrsquos voltage and power where thesolar cellrsquos power is calculated with the product of 119881SC and119868SC One more thing to note here is that the sunlight intensityshould be expressed by lumen or candela unit In this paperhowever we expressed the sunlight intensity by amount ofsolar cellrsquos photocurrent because the solar cellrsquos photocurrentis directly depending on the sunlight intensity [8] Insteadof using lumen or candela the solar cellrsquos photocurrent



60

40

20

0

0 1 2 3 4

Maximum power point

Intensity = 9mA

Intensity = 2mA

Intensity = 38mA

ISC

(mA

)

VSC (V)

(a)

0 1 2 3 4

150

100

50

0

Maximum power point

Intensity = 38mA

Intensity = 9mA

Intensity = 2mA

VSC (V)

PSC

(mW

)(b)

Figure 1 (a) Relationship of solar cellrsquos current (119868SC) and voltage (119881SC) according to sunlight intensity variation (b) Relationship of solar cellrsquospower (119875SC) and voltage (119881SC) according to sunlight intensity variation

for representing sunlight intensity can make us easier inunderstanding the energy harvesting system

Lee et al proposed a reconfigurable charge pump (RCP)circuit that can adjust the architecture among the serial theparallel-serial and the parallel modes according to sunlightvariation to deliver the largest amount of power to the outputat a given sunlight condition [9] The conceptual diagrams ofoperation of RCP circuit are shown in Figures 2(a) 2(b) and2(c) where weak moderate and strong sunlight conditionsare illustrated respectively In Figure 1(a) we can know that119881SC and 119868SC in weak sunlight are smaller than moderateand strong sunlight conditions In Figure 2(a) 119881SC in weaksunlight is as low as 181 V which is about 713 of the open-circuit voltage (119881OC) in Figure 1(a) On the contrary theoutput voltage 119881BAT should be the same with the chargingvoltage of lithium ion batteries that is as high as 42 V Thusthe number of stages of charge pumprsquos architecture should belarge enough to generate 42 V from 181 V

One more thing to consider here is that an amount ofpumping current should be small in weak sunlight becausethe solar cell can deliver only small amounts of 119881SC and 119868SCto the load Here we can think that the serial architecture ofcharge pump in Figure 2(a) is more suitable than the parallel-serial and parallel architectures that are shown in Figures 2(b)and 2(c) respectively In the serial architecture 4 unit pumps(UPs) are connected in series as shown in Figure 2(a)

In Figure 2(b) we assume that sunlight intensity ismoderate where 119881SC becomes 240V that is higher than119881SC = 181V in weak sunlight For the amount of 119868SC 119868SCin moderate sunlight is larger than weak sunlight as shownin Figure 1(a) Owing to these higher 119881SC and larger 119868SC inthe moderate sunlight condition than the weak conditionwe can consider the parallel-serial architecture more suitable

than the serial in Figure 2(a) in delivering larger outputcurrent to the load In the parallel-serial architecture twoUPs are in parallel and two parallel-connected pumps arein series as shown in Figure 2(b) Figure 2(c) shows theparallel architecture that has fourUPs in parallelThis parallelarchitecture can be thought to be suitable to deliver thelargest amount of output current to the load than the serialand parallel-serial architectures in Figures 2(a) and 2(b)respectively

As explained just earlier RCP circuit can maximize theamount of output current by choosing the most suitablearchitecture among the serial parallel-serial and parallelarchitectures at a given sunlight condition [9] Howeverperformance of RCP circuit can be improved more byadding variable pumping frequency (VPF) scheme to thereconfigurable architecture Bymerging theVPF schemewiththe reconfigurable architecture the charge pump which isproposed in this paper can track MPPs better than chargepumps only with the reconfigurable architecture

2 Variable Pumping Frequency Scheme withReconfigurable Charge Pumprsquos Architecture

Figures 3(a) 3(b) and 3(c) show the amounts of outputcurrent of RCP circuit for weak moderate and strongsunlight respectively with varying the pumping frequencyIn Figure 3(a) with weak sunlight intensity from 15mA to25mA the pumping frequency is varied from 1MHz to30MHz to find the best pumping frequency at which thecharge pump can deliver the largest amount of output current[10] From Figure 3(a) it is indicated that the MPP is foundaround 119891PUMP = 25MHz for the sunlight intensity as lowas 15mA If the sunlight intensity becomes 2mA the best


Weak sunlight

UP

UP

UP

UP

Solarcell

VBAT = 42V

VSC = 181V

VBAT

VSC

(a)

Moderate sunlight

UP

UP

UP

UP

Solarcell

VSC = 240VVBAT = 42V

VBAT

0V

VSC

(b)

UP

UP

UP

UP

Solarcell

Strong sunlight

VSC

0V

VBAT = 42V

VSC = 30V

VBAT

(c)

Figure 2 (a) Conceptual diagram of operation of RCP circuit and the serial architecture in weak sunlight condition (b) conceptual diagramof operation of RCP circuit and the parallel-serial architecture in moderate sunlight condition and (c) conceptual diagram of operation ofRCP circuit and the parallel architecture in strong sunlight condition

pumping frequency of delivering the maximum power to theload moves to around 275MHz For the light intensity asmuch as 25mA in Figure 3(a) the charge pump can be thebest in harvesting solar energy around 119891PUMP = 3MHzHere we can know that pumping frequency forMPP becomeshigher with increasing the sunlight intensity

Similarly withmoderate sunlight we varied the pumpingfrequency to find the MPPs where the largest amount ofpower can be harvested from environmental solar energyFor the moderate sunlight intensities of 3mA 9mA and

15mA respectively in Figure 3(b) the pumping frequenciesofMPPs are found at 3MHz 7MHz and 9MHz respectivelyFigure 3(c) shows the amounts of output current with strongsunlight intensities such as 18mA 38mA and 58mA Thecircuit simulation indicates that the pumping frequenciesof 9MHz 15MHz and 23MHz can deliver the maximumamounts of output current to the load for the sunlightintensities of 18mA 38mA and 58mA respectively

From Figures 3(a) 3(b) and 3(c) we can know thatthe VPF scheme that is merged with the reconfigurable


0 10 20 30

00

02

04

06

08

IO

UT(m

A)

Frequency (MHz)

MPP

Weakintensity

Intensity = 15mAIntensity = 2mA

Intensity = 25mA

(a)


Intensity = 15mA

5

4

3

2

1

0

Moderateintensity

0 10 20 30

Frequency (MHz)

MPP

IO

UT(m

A)(b)

0

5

10

15

20

25

Strong intensity


Intensity = 58mA

0 10 20 30

Frequency (MHz)

MPP

IO

UT(m

A)

(c)

0 20 40 600

10

20

30

Sunlight intensity (mA)

fPU

MP

atM

PPs(

MH

z)

(d)

Figure 3 (a) Output current of RCP circuit in weak sunlight intensities with varying pumping frequency (b) Output current in moderatesunlight with varying pumping frequency (c) Output current in strong sunlight with varying pumping frequency (d) The pumpingfrequencies at MPPs with varying sunlight intensity from 15mA to 58mA

architecture can harvest more power from environmentthan the charge pump circuit with only the fixed pumpingfrequency Figure 3(d) shows the pumping frequencies atMPPs with varying sunlight intensity from 15mA to 58mAthat includes weak moderate and strong sunlight intensitiesFrom Figure 3(d) it is indicated that the pumping frequencyas low as 25MHz for the sunlight intensity = 15mA canharvest the largest amount of power from solar energy If thesunlight intensity becomes 28mA the frequency at MPP isincreased to 12MHz For the largest sunlight intensity as large

as 58mA the best frequency for delivering the maximumamount of output current can be found around 23MHz

Figure 4 compares output currents of RCP circuit with thefixed pumping frequencies of 119891PUMP = 7MHz and 119891PUMP =15MHz and the found MPPs that are shown in Figures 3(a)3(b) and 3(c) Here the fixed pumping frequency of 119891PUMP =7MHz is decided by the MPP of moderate sunlight intensityas low as 9mA If we fix pumping frequency of RCP circuit by7MHz the amounts of output current at 119891PUMP = 7MHz arevery similar with the maximum output currents in moderate


0 20 40 60

01

100

10

1


fPUMP = 7MHzfPUMP = 15 MHzMPP

IO

UT(m

A)

Figure 4 Comparison of output currents of RCP circuit with thefixed pumping frequencies (119891PUMP = 7MHz and 119891PUMP = 15MHz)and MPPs that are shown in Figures 3(a) 3(b) and 3(c)

sunlight intensities from 25mA to 155mA On the contraryif the intensity is increased stronger than 155mA the outputcurrent of RCP circuit with 119891PUMP = 7MHz begins todelivermuch smaller output current thanMPPs For the fixedpumping frequency of 15MHz the amounts of output currentwith 119891PUMP = 15MHz are very comparable to the MPPs instrong sunlight intensities from 155mA to 58mA Howeverfor lighter intensities the fixed pumping frequency of 15MHzcannot give the output current as much as MPPs

From Figure 4 we can know that no fixed frequency canhave the amounts of output current that are similar with theMPPs in the entire range of sunlight intensity Thus we needto propose the VPF scheme that can be added to the previousRCP circuit where the pumping frequency can be changedaccording to a given sunlight condition to deliver the largestamount of output current to the load at this given intensity

To give more analytical explanation on relationship ofMPP and 119891PUMP we can start from the following equation ofthe input current of RCP circuit 119868IN [10]

119868IN =1

120578

119868OUT + 1205721119891PUMP (1)

Here 120578 is the ideal current efficiency of RCP circuit and1205721is the proportional coefficient between the charge pumprsquos

input current and the pumping frequency that accountsfor switching loss and reverse current of RCP circuit thatdepend on the pumping frequency [10] In this paper we canthink that the solar cell delivers the input current to RCPcircuit with the additional control circuit which is needed toadjust the architecture and pumping frequency according to

a sunlight variation The solar cellrsquos current can be expressedwith [10]

119868SC = 119868IN + 119868DC + 1205722119891PUMP (2)

Here 119868DC means the static current which is consumed inRCP circuit and the controller circuit 120572

2means the switching

loss in the additional control circuit By combining (1) and (2)we can calculate 119868OUT with

119868OUT = 120578 (119868SC minus 119868DC minus (1205721 + 1205722) 119891PUMP) (3)

As we increase 119891PUMP 119868SC becomes larger thus we canhave more 119868OUT at the load However once 119868SC starts tosaturate 119868OUT should be lowered because of minus(1205721 + 1205722)119891PUMPin spite of increasing 119891PUMP Thus we can have an optimumpumping frequency for a given sunlight condition at whichthe charge pump circuit can deliver the largest amount ofoutput current to the load

From (3) we can extract three parameters of 120578 119868DC and1205721+ 1205722using the least-square fitting method [11] For the

strong sunlight condition with the parallel architecture theextracted values of 120578 119868DC and 1205721 + 1205722 are 0607 389 120583A and0744mAHz respectively

If we know the value of 120578 we can rewrite (3) as follows

120578119868SC minus 119868OUT = 120578 (1205721 + 1205722) 119891PUMP + 120578119868DC (4)

Equation (4) is shown in Figure 5 where the symbolsrepresent the simulated values of 120578119868SC minus 119868OUT and the line iscalculated with 120578(120572

1+ 1205722)119891PUMP + 120578119868DC using the values of 120578

119868DC and 1205721 + 1205722 that are stated just earlier as we increase thepumping frequency from 1MHz to 30MHz From Figure 5we can see that the simulated values of 120578119868SCminus119868OUT are in goodagreement with the calculated values of 120578(120572

1+ 1205722)119891PUMP +

120578119868DC The good agreement between the simulation and themodel in (4) indicates that (3) can describe the charge pumprsquosoutput current well

3 New Reconfigurable Charge Pump CircuitMerged with Variable Pumping FrequencyScheme

In this session a new RCP circuit is proposed where thereconfigurable architecture is merged with the VPF schemeas shown in Figure 6(a) Here the solar cell is connected to thevoltage divider with the division ratio as much as 05 Thuswe can have a voltage as high as 05 sdot 119881SC using the dividerBy applying the sampling frequency 119891SAMPLE which is muchslower than119891PUMP for the open-circuit condition the dividedvoltage 05 sdot 119881OC can be sampled by the sample and holdcircuit (SampH) In this paper119891SAMPLE used in the simulation isas slow as 1 KHz [9] Comparing119891SAMPLE with119891PUMP119891SAMPLEis sim1000x slower than 119891PUMP thus the timing overhead dueto the sampling of the open-circuit voltage can be ignoredHere it should be noted that 05 sdot 119881OC is half the sampledsolar cellrsquos voltage with the open-circuit conditionThe SampHrsquosoutput 05 sdot 119881OC goes into the architecture reconfigurationblock (ARB) where the charge pumprsquos architecture can be


15

10

5

0

20 25 300 5 10 15

Strong sunlight

120578(1205721 + 1205722)

120578IDC

fPUMP (MHz)

120578(1205721 + 1205722)fPUMP + 120578IDC

120578I

SCminusI

OU

T(m

A)

120578ISC minus IOUT

Figure 5 120578119868SC minus 119868OUT versus 120578(1205721 +1205722)119891PUMP +120578119868DC with increasing119891PUMP The good agreement between 120578119868SC minus 119868OUT and 120578(120572

1+

1205722)119891PUMP + 120578119868DC indicates that (3) can describe the charge pumprsquos

output current driven by the solar cell well

reconfigured by controlling the reconfiguration switches [9]As well explained in [9] three modes of RCP circuit areavailable in Figure 6(a) They are the serial parallel-serialand parallel architecture respectivelyThe sampled 05sdot119881OC iscompared with the predetermined reference voltages such as119881REF PM and 119881REF SM thereby the charge pump architecturecan be reconfigured C1 and C2 are the comparators for thearchitecture reconfiguration If 05 sdot 119881OC lt 119881REF SM thearchitecture is changed to the serial mode [9] If 05 sdot 119881OC isbetween 119881REF SM and 119881REF PM the circuit is reconfigured bythe parallel-serialmodeWhen 05sdot119881OC is larger than119881REF PMthe mode is fixed by the parallel architecture

The 05 sdot 119881OC that is a signal before the SampH cir-cuit in Figure 6(a) goes into the variable frequency block(VFB) where the solar cellrsquos voltage 119881SC is compared with05 sdot 119896 sdot 119881OC minus Δ and 05 sdot 119896 sdot 119881OC + Δ to decide the optimumpumping frequency at which the maximum solar powercan be harvested from environment Here 119896 is a fractionalconstant of 119881OC and the value used in this simulation is 08It is well known that the solar cell can deliver the maximumpower to the load when 119881SC is close to a fractional open-circuit voltage 119896 sdot 119881OC even though 119896 can be different forvarious kinds of solar cells [12] If the optimum pumpingfrequency is found the voltage-controlled oscillator (VCO)applies the obtained 119891PUMP of MPP to RCP circuit Theschematic of UP circuit is shown in the inset of Figure 6(a)where119872

1and119872

3are the precharging NMOSFETs and119872

2

and 1198724are the transferring PMOSFETs The width and

length of 1198721and 119872

3are 120120583m and 035 120583m respectively

The width and length of1198722and119872

4are 240 120583m and 035 120583m

respectively The pumping capacitors are CAP1and CAP

2in

Figure 6(a) The capacitance used in the simulation is 50 pF

CLKB and CLK are the two-phase clocking signals that enterCAP1and CAP

2 respectively Figure 6(b) shows the detailed

schematic of VFB with VCO circuit For VCO circuit thegeneral scheme which is based on simple ring oscillator isused in this paper [13]

Figure 6(c) shows the voltage and current waveforms inFigure 6(a) If 05 sdot 119881SC is between 05 sdot 119896 sdot 119881OC minus Δ and05 sdot119896 sdot119881OC+Δ the charge pump (CP) in VFB keeps its outputvoltage 119881

119862 which controls the VCO If 119881

119862is not changed

the VCOrsquos output frequency 119891PUMP is not changed too Nowlet us explain the closed-loop operation of VPF scheme inFigure 6(a) more in detail as shown in Figure 6(c) First wecan assume that 05 sdot 119881SC is larger than 05 sdot 119896 sdot 119881OC + ΔAt this time the two comparators of 119862

3and 119862

4can give UP

signal to CP and VCO in Figure 6(a) to increase the pumpingfrequency higher As the pumping frequency becomes higher05 sdot 119881SC becomes lower If 05 sdot 119881SC is between 05 sdot 119896 sdot 119881OC minusΔand 05 sdot 119896 sdot 119881OC + Δ the two comparators of 119862

3and 119862

4give

STOP signal to CP andVCO thereby the pumping frequencyis stabilized If 05 sdot 119881SC is lower than 05 sdot 119896 sdot 119881OC minus Δthe comparators 119862

3and 119862

4generate DOWN signal to CP

and VCO thus the pumping frequency becomes lower and05 sdot 119881SC can be raised little by little until 05 sdot 119881SC is between05 sdot 119896 sdot 119881OC minus Δ and 05 sdot 119896 sdot 119881OC + Δ

4 Simulation Results and Layout

The proposed circuit is verified by SPECTRE circuit sim-ulation which is provided by Cadence Inc The SPECTREsimulation model was obtained from Magna 035 120583m CMOSprocess technology Figure 7(a) compares the percentageamounts of output current among 3 schemes of the recon-figurable charge pump circuit with respect to MPPs Hereldquo119891PUMP = 7MHzrdquo in Figure 7(a) means that the output cur-rent of RCP circuit is obtained when the pumping frequencyis fixed at 7MHz For ldquo119891PUMP = 7MHzrdquo even though thesunlight condition is changed the pumping frequency is fixedat 7MHz for all the sunlight conditions In Figure 7(a) thefixed pumping frequency as low as 7MHz is decided by thefact that the charge pump circuit can deliver the maximumoutput current at 119891PUMP = 7MHz for the moderate sunlightcondition of the intensity = 9mA as shown in Figure 3(b)In Figure 7(a) 119891PUMP = 15MHz is the output current ofRCP circuit when the pumping frequency of Figure 6(a) isfixed at 15MHz Similarly with ldquo119891PUMP = 7MHzrdquo the fixedpumping frequency as high as 15MHz is decided becausethe charge pump can have the largest output current for thestrong sunlight condition of intensity = 38mA as shown inFigure 3(c) In Figure 7(a) ldquoVPFrdquo means that the pumpingfrequency of the reconfigurable charge pump can be variableaccording to the sunlight conditions Here the pumpingfrequency is decided to meet the condition where the solarcellrsquos voltage 05119881SC should be between 05 sdot 119896 sdot 119881OC minus Δ and05 sdot 119896 sdot 119881OC + Δ This variable frequency for each sunlightcondition can be tracked by the circuit shown in Figures 6(a)and 6(b) In Figure 7(a) ldquoMPPrdquomeans the amounts of outputcurrent of RCP circuit at the maximum power points forvarious sunlight conditions The maximum amounts of 119868OUT


ARB (architecture reconfguration block )

RCP (reconfgurable charge pump)

UP

UP

UP

UP

UP

UP

UP

UP

UP

UP

UP

UP

UPSerial

architectureParallel

architectureParallel-serial

architecture

VBATIOUT

BAT

Solar cell

VCO

k

C3

C1

C2

C4

05

+

+

minus

minus

+

minus

+minus

CP

VFB (variable frequency block)

VREF SM

VREF PM

M1 M2

M3M4

S and H

CLK

IN OUT

CLKBCAP1

CAP2

fSAMPLE

ISC IIN

fPUMPSWSM

SWSPM

SWPM

05 middot VSC

05 middot VSC

05 middot VOC

VSC

05 middot k middot VOC + Δ

05 middot k middot VOC minus Δ

(a)

VCO

M1

M2

M3

M4

G1

G2

COMP2

COMP1

Vbp

Vbn

C1

Vc

VSC+

+

05 middot VOC

05 middot VSC

k

fPUMP

minus

minus



(b)

05VSC

05kVOC + Δ

05kVOC minus Δ

VCfPUMP for MPP

fPUMP

IOUT

(c)

Figure 6 (a) Block diagram of the new RCP circuit that is proposed in this paper (b) Circuit schematic of the VFB (c) Voltage and currentwaveforms of the VFB

for various sunlight conditions are obtained by the circuitsimulation of the charge pump with varying the pumpingfrequency from 1MHz to 30MHz

From Figure 7(a) ldquo119891PUMP = 7MHzrdquo shows 996 ofthe maximum output current at the moderate sunlight con-dition However ldquo119891PUMP = 7MHzrdquo shows only 631 and489 at the weak and strong sunlight conditions respec-tively This fact tells us that 119891PUMP as low as 7MHz candeliver the largest amount of 119868OUT in only moderate sun-light intensity compared to weak and strong sunlight

intensities If sunlight intensity becomes different fromthe moderate condition percentage of 119868OUT with respect tothe MPP becomes much smaller ldquo119891PUMP = 15MHzrdquo candeliver 999 of the maximum output current in strong sun-light In spite of this large output current in strong intensityRCP circuit at the fixed pumping frequency of 15MHz cangenerate as low as 338 and 747 of the maximum 119868OUTin the weak and moderate sunlight condition respectivelyUnlike the cases of the fixed pumping frequencies of 119891PUMP =7MHz and 15MHz the VPF scheme which is proposed in


100

50

0

Sunlight intensity

Average

VPF MPPfPUMP = 7MHz fPUMP = 15MHz

Weak(15mA)

Moderate(9mA)

Strong(38mA)

IO

UT()

(a)

10

1

01

0 20 40 60


MPPVPF

IO

UT(m

A)(b)

Figure 7 (a) The percentage 119868OUT for various sunlight conditions with the fixed pumping frequencies of 7MHz and 15MHz and the VPFscheme Here MPP means the simulated maximum output currents (b) The output current comparison between the proposed VPF schemeand the simulated maximum output currents (MPP)

Reconfgurable charge pump

2080 120583m

290120583m

VFB ARB

Figure 8 Layout of RCP circuit with the added VPF scheme

this paper shows 919 of the maximum amounts of currentfor all the sunlight conditions from weak to strong intensityon average The large amounts of output current in the VPFscheme can be delivered to the load in awide range of sunlightintensity from weak to strong Figure 7(b) compares the VPFschemewith respect toMPPs from the sunlight intensity from15mA to 58mA For the entire range of sunlight intensity theVPF scheme shows that amounts of the output current aresimilar with MPPs that are obtained from Figures 3(a) 3(b)and 3(c)

Figure 8 shows the layout of the proposed RCP circuitwith VPF scheme The VPF scheme which is added in thispaper occupies an area as small as 161 of the total layoutarea The power overhead which is caused by the added VPFscheme is estimated only as small as 04 of the total powerconsumption for the sunlight condition as strong as 38mAIf the sunlight conditions become as weak as 9mA the poweroverhead is increased to 17 This is due to the fact thatthe solar cellrsquos current becomes smaller with decreasing the

sunlight intensity while the amount of power consumptionof the VPF scheme is changed very littleThus the percentagepower overhead becomes worse with decreasing sunlightintensity

5 Conclusion

In this paper the VPF scheme was proposed to be mergedwith the previous RCP circuit that can change its architectureaccording to a given sunlight condition Merging the VPFschemewith the architecture reconfiguration can improve thepercentage 119868OUT better by 214 and 224 than the fixedpumping frequencies of 7MHz and 15MHz respectivelyComparing the VPF scheme with the available maximumoutput currents the VPF can deliver 919 of the maximum119868OUT to the load on average In terms of the power and areaoverheads the VPF scheme proposed in this paper consumesthe power by 04 of the total power consumption andoccupies the layout area by 161 of the total layout area




Acknowledgments

The work was financially supported by NRF-2011-220-D00089 NRF-2011-0030228 NRF-2013K1A3A1A25038533NRF-2013R1A1A2A10064812 and BK Plus with the Educa-tional Research Team for Creative Engineers on Material-Device-Circuit Co-Design (Grant no 22A20130000042)funded by the National Research Foundation of Korea (NRF)and the Industrial Strategic Technology Development Pro-gram funded by the Ministry of Trade Industry and Energy(MOTIE Korea) (10039239) The CAD tools were supportedby IC Design Education Center (IDEC) Daejeon Korea

References

[1] J Kim J Kim and C Kim ldquoA regulated charge pump with alow-power integrated optimumpower point tracking algorithmfor indoor solar energy harvestingrdquo IEEE Transactions onCircuits and Systems II Express Briefs vol 58 no 12 pp 802ndash806 2011

[2] C Lu S P Park V Raghunathan and K Roy ldquoLow-overheadmaximum power point tracking for micro-scale solar energyharvesting systemsrdquo in Proceedings of the 25th InternationalConference on VLSI Design (VLSID rsquo12) pp 215ndash220 January2012

[3] J Schmid T Gaedeke T Scheibe and W Stork ldquoImprovingenergy efficiency for small-scale solar energy harvestingrdquo inProceedings of the European Conference on Smart ObjectsSystems and Technologies pp 1ndash9 June 2012

[4] C Alippi and C Galperti ldquoAn adaptive system for opimalsolar energy harvesting in wireless sensor network nodesrdquo IEEETransactions on Circuits and Systems I Regular Papers vol 55no 6 pp 1742ndash1750 2008

[5] S Bader and B Oelmann ldquoShort-term energy storage forwireless sensor networks using solar energy harvestingrdquo inProceedings of the IEEE International Conference on NetworkingSensing and Control pp 71ndash76 April 2013

[6] P Favrat P Deval andM J Declercq ldquoA high-efficiency CMOSvoltage doublerrdquo IEEE Journal of Solid-State Circuits vol 33 no3 pp 410ndash416 1998

[7] F Pen and T Samaddar Charge Pump Circuit Design McGraw-Hill New York NY USA 2006

[8] J Nelson The Physics of Solar Cells Imperial College PressLondon UK 2003

[9] E S Lee J M Choi Y S Kim et al ldquoSunlight-variation-adaptive charge pump circuit with self-reconfiguration forsmall-scale solar energy harvestingrdquo IEICE Electronics Expressvol 9 no 17 pp 1423ndash1433 2012

[10] H Shao C-Y Tsui andW-H Ki ldquoAn inductor-less micro solarpower management system design for energy harvesting appli-cationsrdquo in Proceedings of the IEEE International Symposium onCircuits and Systems (ISCAS rsquo07) pp 1353ndash1356 May 2007

[11] T Kariya and H Kurata Generalized Least Squares JohnWileyamp Sons New York NY USA 2004

[12] J Ahmad ldquoA fractional open circuit voltage based maximumpower point tracker for photovoltaic arraysrdquo in Proceedings of

the 2nd International Conference on Software Technology andEngineering (ICSTE rsquo10) pp 247ndash250 October 2010

[13] P E Allen and D R Holberg CMOS Analog Circuit DesignOxford University Oxford UK 2002

Research ArticleOptimization of p-GaNInGaNn-GaN Double Heterojunctionp-i-n Solar Cell for High Efficiency Simulation Approach

Aniruddha Singh Kushwaha123 Pramila Mahala4 and Chenna Dhanavantri23

1 Indian Institute of Technology Bombay (IIT-B) Powai Maharashtra 400076 India2 Academy of Scientific and Innovative Research New Delhi 110001 India3 Council of Scientific and Industrial Research-Central Electronics Engineering Research Institute (CSIR-CEERI) PilaniRajasthan 333031 India

4 School of Solar Energy Pandit Deendayal Petroleum University (PDPU) Gandhinagar Gujarat 382007 India

Correspondence should be addressed to Pramila Mahala pramilamahala98gmailcom

Received 29 May 2013 Accepted 13 March 2014 Published 10 April 2014


Copyright copy 2014 Aniruddha Singh Kushwaha et al This is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited

We have conducted numerical simulation of p-GaNIn012

Ga088

Nn-GaN p-i-n double heterojunction solar cell The dopingdensity individual layer thickness and contact pattern of the device are investigated under solar irradiance of AM15 for optimizedperformance of solar cell The optimized solar cell characteristic parameters for cell area of 1times 1mm2 are open circuit voltage of226 V short circuit current density of 331mAcm2 fill factor of 846 and efficiency of 643 with interdigitated grid pattern

1 Introduction

The direct and tunable band gap of InGaN semiconductoroffers a unique opportunity to develop high efficiency solarcell The direct band gap of the InGaN semiconductor canvary from 07 to 34 eV [1] which covers a broad solarspectrum from near-infrared to near-ultraviolet wavelengthregion InGaN alloys show the unique properties such astunable band gap high carrier mobility and high radiationresistance which offers a great advantage in design and fabri-cation of high efficiency devices for photovoltaic applications[2ndash4]High absorption coefficient of InGaNalloys alsomakesit suitable to use in solar cells [5ndash7] Earlier theoreticalcalculations have shown that it is possible to achieve efficiencyup to 50 with InGaN alloys [8] In spite of high efficiencyprediction fabricated solar cells could not demonstrate highefficiency [9ndash13] The epitaxial growth of InGaN layers facesmany issues such as lack of defect-free substrate indiumsegregation at higher composition and large latticemismatchbetween InN and GaN which leads to phase separation [14]In addition there are significant polarization charges at theInGaNGaN interface which alter the electric field in InGaN

layer [15] As a consequence the fabricated InGaN solar cellshows very low conversion efficiency compared to theoreticalcalculation

In this paper a simulation work is carried out tooptimize p-GaNInGaNn-GaNsapphire device structureof solar cells with 12 indium composition under AM15illumination Simulation is carried out by optimizing dopingconcentration and thickness of p-GaN InGaN and n-GaNlayer respectively by changing only one material parameterat a time and keeping other parameters constant Simulationverifies that device efficiency is strongly dependent on theintrinsic-layer (InGaN) thickness as most of the spectrumis absorbed in this layer We simulate p-GaNInGaNn-GaNSapphire solar cell to investigate the effect of incor-porating grid contact pattern on the device characteristicparameter It is observed that optimization of grid contactspacing helps greatly in device efficiency enhancement

TCAD SILVACO Version ATLAS 5163R is used forsimulation The simulator works on mathematical modelswhich consist of fundamental equations such as Poissonrsquosequation continuity equation and transport equations Inour simulation we have used the models such as AUGER for



Table 1 Material parameter used in simulation

Parameter GaN In012Ga088N InNBand gap 119864

119892(ev) [10] 342 293 07

Lattice constant (A∘) [11] 318 323 36Minority carrier life time (ns) [21] 1 1 1Spontaneous polarization (sheet charge per cm2) [22] minus21211989013 minus21811989013 minus26311989013

Piezoelectric polarization (sheet charge per cm2) [22] 0 minus88711989012 99811989013

Auger coefficient n-type [11] 1119890 minus 34 1119890 minus 34 1119890 minus 34

Auger coefficient p-type [11] 1119890 minus 34 1119890 minus 34 1119890 minus 34

Auger recombination SRH for Shockley Read Hall recom-bination OPTR for optical recombination KP model foreffective masses and band edge energies for drift diffusionsimulation Mathematical models are used in simulationwhich consists of fundamental equations such as Poissonrsquosequation continuity equation and transport equations New-tonrsquosmethod is used as the solutionmethod in simulation Allthe above models are used from standard TCAD library

This paper is organized as follows in Section 2 materialparameters are described which are used in simulationmodels Section 3 discusses simulation and results Finallyconclusion is presented in Section 4

2 Material Parameters

Theabsorption coefficient 120572(119864) of In119909Ga1minus119909

N semiconductoras a function of energy [16] energy band gap and electronaffinity as a function of band gap energy 119864

119892(119909) [17 18] and

electron and hole mobility as a function of doping [19] can beexpressed as

120572 [119864 (120582)] = 1205720radic

119864 (120582) minus 119864119892(119909)

119864119892(119909)

119864119892(119909) = 07119909 + 34 (1 minus 119909) minus 143119909 (1 minus 119909) eV

120594 = 41 + 07 times (34 minus 119864119892(119909))

120583119894(119873) = 120583min119894 +

120583max119894 minus 120583min119894

1 + (119873119873119892119894)

120574119894

(1)

where 119864(120582) is the energy of photon corresponding to respec-tive wavelength and 119864

119892(119909) is the band gap of In

119909Ga1minus119909

Nsemiconductor It is assumed that 120572

0for In

119909Ga1minus119909

N is thesame as that of GaN 119894 denotes electrons (e) or holes (h)119873 is doping concentration and 120583min 120583max 120574 and 119873119892 arethe parameter given for a specific semiconductor [19] Thematerial parameters for InGaN such as absorption coefficientlattice constant and polarization charges are derived andextracted by interpolation of known material parameters ofInN and GaN Some of important material parameters usedduring simulation are listed in Table 1

p-Contact

p-GaN

n-GaN

i-InGaN

Sapphire

n-Contact

Figure 1 p-i-n InGaNGaN double heterojunction solar cell struc-ture

3 Simulation Results and Discussion

31 Optimization of p-i-n Structure The schematic diagramof InGaNGaN p-i-n solar cell is shown in Figure 1 The p-GaN layer thickness is varied from 70 to 120 nm The InGaNand n-GaN layer thicknesses 100 nm and 15 120583m and theirdoping densities 1times 1016 cmminus3 and 6times 1018 cmminus3 respectivelyare kept constantwhile optimizing p-GaN layer thickness anddoping density

The effect of p-GaN layer thickness and doping concen-tration on solar cells characteristic parameters such as shortcircuit current density open circuit voltage fill factor andefficiency is shown in Figure 2 The short circuit currentdensity 119869sc increases with p-GaN layer thickness As thethickness of p-GaN layer increases more photon absorptionoccurs in the p-region resulting in more electron and holecarrier generation that contributes to enhancement of currentdensity On the contrary thicker p-GaN layer increasessurface recombination rate after a certain thickness currentdensity starts decreasing The open circuit voltage 119881oc andfill factor FF do not show significant change with respectto p-GaN layer thickness as these parameters largely dependon the bulk property of semiconductor rather than surfaceproperty

The effect of doping density is analyzed by simulatingthe structure for different doping concentrations of 1 times1016 cmminus3 1 times 1017 cmminus3 and 1 times 1018 cmminus3 respectively Thesimulation result shows that the short circuit current densityis higher for doping concentration of 1 times 1016 cmminus3 However


1875

1850

1825

1800

65 70 75 80 85 90 95 100 105 110 115 120 125

P-GaN thickness (nm)

Jsc(m

Acm

2 )

(a)

2286

2284

2282

2280

2278

65 70 75 80 85 90 95 100 105 110 115 120 125


Voc

(V)

(b)

0902

0901

0900

0899

0898

65 70 75 80 85 90 95 100 105 110 115 120 125


FF

1 times 1016

1 times 1017

1 times 1018

(c)

1 times 1016

1 times 1017

1 times 1018

256

252

248

244

65 70 75 80 85 90 95 100 105 110 115 120 125


Effici

ency

(d)

Figure 2 Effect of p-GaN layer thickness and doping concentrations on GaNInGaN solar cell characteristic parameters (a) short circuitcurrent density (b) open circuit voltage (c) fill factor and (d) efficiency

Jsc(m

Acm

2 ) 35

30

25

20

15

10

0 200 400 600

InGaN thickness (nm)

(a)

Voc

(V)

222

220

218

216

0 200 400 600


(b)

090

088

086

084

082

080

0 200 400 600


FF

(c)

0 50

15

20

25

30

35

40

45

100 150 200 250 300 350 400 450 500 550 600


Effici

ency

(d)

Figure 3 Effect of InGaN layer thickness on characteristic parameters of solar cells (a) Short circuit current density (b) Open circuit voltage(c) Fill factor (d) Efficiency

it is found that 119881oc is higher for doping concentration of 1 times1017 cmminus3 The combined effect of 119869sc and 119881oc determinesthe efficiency curve which follows curve pattern of 119869sc andshows higher efficiency for the doping concentration of 1 times1017 cmminus3 where 119881oc is high

The effect of intrinsic InGaN layer thickness on solarcell characteristic parameters of p-GaNInGaNn-GaN p-i-n solar cell is shown in Figure 3 The InGaN layer thicknessis varied from 50 to 550 nm The p-GaN and n-GaN layerthicknesses 100 nm and 15 120583m and their doping densities 5


030027

50

40

30

20

10

0

033 036 039 042 045 048

Qua

ntum

effici

ency

()

Wavelength (120583m)

Figure 4 Quantum efficiency of optimized p-i-n GaNInGaN double heterojunction solar cell

160

326

328

330

332

334

336

338

340

180 200 220 240 300280260 320

Shor

t circ

uit c

urre

nt d

ensit

yJ

sc(m

Acm

2 )

Grid spacing (120583m)

5grid fingers

4grid fingers

3grid fingers

(a)

160 180 200 220 240 300280260 320


59

60

64

63

62

61

5grid fingers

Effici

ency

()

4grid fingers

3grid fingers

(b)

p-Contact

n-Contact

Sapphire substrate

i-InGaN

n-GaN

p-GaN

6 times 1018 cmminus3


(c)

Figure 5 Effect of (a) short circuit current density (b) efficiency on grid spacing for different number of grids and (c) p-i-n InGaNGaNdouble heterojunction solar cell with grid type contact


Table 2 Characteristic parameters of simulated InGaN p-i-n solar cell

Contact type Effective device area (mm2) Grid spacing (120583m) Indium () 119869sc (mAcm2) 119881oc (V) FF () 120578 ()Square pad 096 mdash 12 564 227 82 416Grid pattern

5 grid fingers 090 180 12 326 217 836 5924 grid fingers 092 210 12 331 226 846 6343 grid fingers 094 260 12 339 217 832 612

times 1017 cmminus3 and 6 times 1018 cmminus3 respectively are kept constantwhile optimizing InGaN layer thickness

It is found that 119869sc increases with InGaN layer thicknesssince photons get longer path for absorption and also unab-sorbed lower energy photons from p-GaN layer get absorbedin this region Photon absorption is also supported by higherabsorption coefficient of InGaN material [20] On the otherhand the open circuit voltage is found to be decreasedand remain nearly constant after layer thickness of 250 nmHigher thickness generates more carriers and recombinationresults in decreased open circuit voltage

The fill factor also starts decreasing with respect toincreased InGaN layer thickness The series resistanceincreases with increasing InGaN layer thickness The effi-ciency curve resembles that of current density which rep-resents the combined effect of all parameters 119878

119862 119881ic and

FF After the p-GaN and InGaN-layers the n-GaN layerthickness and doping density concentration are optimizedThe n-GaN thickness is varied from 04 to 24 120583mThe p-GaNand InGaN layers thicknesses are 100 nm and 450 nm anddoping densities 5 times 1017 cmminus3 and 1 times 1016 cmminus3 respectivelyare kept constant while optimizing n-GaN layer thicknessand doping density The n-GaN layer thickness and dopingdensity do not play an important role with respect to solar cellparameters It may be because most of the available photonsare already absorbed in the p-GaN and InGaN layer and veryfew carriers are generated in n-GaN layer

The final p-i-n structure consists of n-type GaN layer15 120583m thickness and doping density of 6 times 1018 cmminus3 unin-tentionally doped InGaN layer 450 nm thickness and dopingdensity of 1 times 1016 cmminus3 with 12 percent indium compositionand top p-type GaN layer 100 nm thickness with acceptordensity of 5 times 1017 cmminus3The quantum efficiency of optimizedstructure is calculated and shown in Figure 4 and the max-imum quantum efficiency is found to be in the wavelengthrange of 04 nm to 045 nm wavelength which correspondsto GaNInGaN material band gap region

32 Effect of using Interdigitated Grid Pattern We simulatedthe p-i-n structure with interdigitated grid pattern in orderto further enhance the efficiency by use of grid type contactwhich helps in increasing the carrier collection Simulationresults of incorporating grid patterns on times and efficiencywith different numbers of grid fingers such as 3 4 and 5and different grid spacing from 175 to 375 120583m are shownin Figure 5 The size of the cell considered here is 1 times1mm2 The number of grids will be reduced with increasingfinger-to-finger grid spacingThe short circuit current density

decreases with increasing grid numbers due to decreasein effective area available for photon absorption Howeverthere is an enhancement in efficiency due to more carriercollectionThe characteristic parameters for different contactpatterns along with square pad are compared in Table 2

4 Conclusion

The optimization of p-GaNInGaNn-GaN double hetero-junction p-i-n solar cell with square contact and grid patternis studied The photovoltaic parameter of solar cell stronglydepends on p-GaN and InGaN layers thickness and dopingdensity Photovoltaic parameters such as 119881oc 227V 119869sc564mAcm2 FF 82 and 120578 416 are obtained for squarecontact type under 1 sun AM15 illumination with optimizedp-GaN (100 nm 5e17 am-3) and InGaN (In = 012 450 nm)layers and high quantum efficiency of sim50 is also achievedin wavelength range of 04 nm to 045 nm which correspondsto In012

Ga088

N absorption region In addition to optimizingstructure use of grid contact pattern with finger spacing of210 nm improves conversion efficiency to 634



Acknowledgments

The authors gratefully acknowledge the Director of CSIR-CEERI Pilani for his encouragement in this work They alsothank all ODG members for their help and cooperation

References

[1] J Wu WWalukiewicz K M Yu et al ldquoSmall band gap bowingin In1minus119909

Ga119909N alloysrdquo Applied Physics Letters vol 80 no 25 pp

4741ndash4743 2002[2] X M Cai S W Zeng and B P Zhang ldquoFabrication and char-

acterization of InGaN p-i-n homojunction solar cellrdquo AppliedPhysics Letters vol 95 Article ID 173504 2009

[3] Y Nanishi Y Saito and T Yamaguchi ldquoRF-molecular beamepitaxy growth andproperties of InNand related alloysrdquo Journalof Applied Physics vol 42 part 1 pp 2549ndash2559 2003

[4] X Zheng R H Horng D S Wuu et al ldquoHigh-qualityInGaNGaN heterojunctions and their photovoltaic effectsrdquoApplied Physics Letters vol 93 Article ID 261108 2008


[5] C J Neufeld N G Toledo S C Cruz et al ldquoHigh quantumefficiency InGaNGaN solar cells with 295 eV band gaprdquoApplied Physics Letters vol 93 Article ID 143502 2008

[6] R Singh D Doppalapudi T D Moustakas and L T RomanoldquoPhase separation in InGaN thick films and formation ofInGaNGaN double heterostructures in the entire alloy com-positionrdquo Applied Physics Letters vol 70 no 9 pp 1089ndash10911997

[7] J F Muth J H Lee I K Shmagin et al ldquoAbsorption coefficientenergy gap exciton binding energy and recombination lifetimeof GaN obtained from transmission measurementsrdquo AppliedPhysics Letters vol 71 no 18 pp 2572ndash2574 1997

[8] A De Vos Endoreversible Thermodynamics of Solar EnergyConversion Oxford University Press Oxford UK 1992

[9] X Zhang X L Wang H L Xiao et al ldquoSimulation ofIn065

Ga035

N single-junction solar cellrdquo Journal of Physics DApplied Physics vol 40 no 23 p 7335 2007

[10] H Hamzaoui A S Bouazzi and B Rezig ldquoTheoretical pos-sibilities of In

119909Ga1minus119909

N tandem PV structuresrdquo Solar EnergyMaterials and Solar Cells vol 87 no 1ndash4 pp 595ndash603 2005

[11] Md R Islam A N M E Kabir and A G Bhuiyan ldquoPro-jected Performance of In

119909Ga1minus119909

N-based multi-junction solarcellsrdquo Istanbul University-Journal of Electrical and ElectronicsEngineering vol 6 no 2 pp 251ndash255 2006

[12] L Hsu and W Walukiewicz ldquoModeling of InGaNSi tandemsolar cellsrdquo Journal of Applied Physics vol 104 Article ID024507 2008

[13] B W Liou ldquoIn119909Ga1minus119909

N-GaN-based solar cells with a multiple-quantum-well structure on SiCNndashSi(111) substratesrdquo IEEE Pho-tonics Technology Letters vol 22 no 4 2010

[14] Y-S Lin K-J Ma C Hsu et al ldquoDependence of compositionfluctuation on indium content in InGaNGaN multiple quan-tumwellsrdquoApplied Physics Letters vol 77 no 19 pp 2988ndash29902000

[15] J Wierer A Fischer and D Koleske ldquoThe impact of piezo-electric polarization and nonradiative recombination on theperformance of (0001) face GaNInGaN photovoltaic devicesrdquoApplied Physics Letters vol 96 Article ID 051107 2010

[16] J Hubin and A V Shah ldquoEffect of the recombination functionon the collection in a p-i-n solar cellrdquo Philosophical Magazine Bvol 72 no 6 pp 589ndash599 1995

[17] M E Levinshtein S L Rumyantsev and M S Shur Propertiesof Advanced Semiconductor Materials Wiley Chichester UK2001

[18] N Li Simulation and analysis of GaN-based photoelectronicsdevices [PhD dissertation] Institute of Semiconductors Chi-nese Academy of Sciences Beijing China 2005 (Chinese)

[19] T T Mnatsakanov M E Levinshtein L I Pomortseva S NYurkov G S Simin and M A Khan ldquoCarrier mobility modelfor GaNrdquo Solid-State Electronics vol 47 no 1 pp 111ndash115 2003

[20] K Kumakura T Makimoto N Kobayashi T Hashizume TFukui and H Hasegawa ldquoMinority carrier diffusion lengthin GaN dislocation density and doping concentration depen-dencerdquo Applied Physics Letters vol 86 Article ID 052105 2005

[21] G F Brown J W Ager III W Walukiewicz and J Wu ldquoFiniteelement simulations of compositionally graded InGaN solarcellsrdquo Solar Energy Materials and Solar Cells vol 94 no 3 pp478ndash483 2010

[22] ATLAS Userrsquos Manual Device Simulation Software 061108Ed Silvaco Data Systems Inc Santa Clara Calif USA 2008

Research ArticleBuck-BoostForward Hybrid Converter for PV EnergyConversion Applications

Sheng-Yu Tseng Chien-Chih Chen and Hung-Yuan Wang

Department of Electrical Engineering Chang-Gung University 259 Wen-Hwa 1st Road Tao-Yuan 333 Taiwan

Correspondence should be addressed to Sheng-Yu Tseng sytsengmailcguedutw

Received 16 August 2013 Revised 28 February 2014 Accepted 28 February 2014 Published 7 April 2014


Copyright copy 2014 Sheng-Yu Tseng et alThis is an open access article distributed under theCreative CommonsAttribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

This paper presents a charger and LED lighting (discharger) hybrid system with a PV array as its power source for electronicsign indicator applications The charger adopts buck-boost converter which is operated in constant current mode to charge lead-acid battery and with the perturb and observe method to extract maximum power of PV arrays Their control algorithms areimplemented by microcontroller Moreover forward converter with active clamp circuit is operated in voltage regulation conditionto drive LED for electronic sign applications To simplify the circuit structure of the proposed hybrid converter switches of twoconverters are integrated with the switch integration technique With this approach the proposed hybrid converter has severalmerits which are less component counts lighter weight smaller size and higher conversion efficiency Finally a prototype of LEDdriving system under output voltage of 10V and output power of 20W has been implemented to verify its feasibility It is suitablefor the electronic sign indicator applications

1 Introduction

In recent years light emitting diodes (LEDs) are becomingmore prevalent in a wide application Due to materialadvances over the past few decades efficiencies of LEDshave increased many times [1] and their applications haverapidly grown for automotive taillights LCD back lightstraffic signals and electronic signs [2 3] Moreover seriousgreenhouse effect and environmental pollution caused byoverusing fossil fuels have disturbed the balance of globalclimate In order to reduce emission of exhausted gases zero-emission renewable energy sources have been rapidly devel-oped One of these sources is photovoltaic (PV) arrays whichis clean and quiet and an efficient method for generatingelectricity As mentioned above this paper proposes an LEDdriving systemwhich adopts the PV arrays for electronic signapplications

In electronic sign applications using PV arrays the powersystemwill inevitably need batteries for storing energy duringthe day and for releasing energy to LED lighting during thenightTherefore it needs a charger and discharger (LED driv-ing circuit) as shown in Figure 1 Since the proposed powersystem belongs to the low power level applications buck

boost buck-boost flyback or forward converter is moreapplied to the proposed one [4ndash12] In these circuit structuresaccording to the relationships among PV output voltage 119881PVbattery voltage 119881

119861 and output voltage 119881

119874 the proposed

hybrid converter can choose functions which are of step-upand -down simultaneously as the charger or discharger for awider application Due to the previously described reasonsthe charger of the proposed one adopts buck-boost converterand the discharger uses forward converter Moreover sinceforward converter exits two problems which are the energiestrapped in leakage inductor and magnetizing inductor oftransformer 119879

119891 it needs a snubber or circuit to recover

these energiesTherefore forward converter can use an activeclamp circuit to solve these problems In order to simplifythe proposed hybrid converter and increase its conversionefficiency a bidirectional buck-boost converter and activeclamp forward converter are used as shown in Figure 2Since charger and discharger (LED driving circuit) of theproposed hybrid converter are operated in complement andthey use switch 119878

1to control their operational states inductor

1198711of buck-boost converter and magnetizing inductor 119871

119898of

transformer 119879119891can be merged Therefore switches of the

bidirectional buck-boost converter and active clamp forward



Charger

Power management

unit

Battery management

unitMPPT

unit

LED driving circuit

(discharger)LED

lighting

PWM ICunit

PWM ICMicrochipController

PV arrays VPV

VPV

IPV

VB

M1 M2

M1 M2IB

VB

VO

VO

minus

+

minus

+

minus

+

Figure 1 Block diagram of the proposed hybrid converter for electronic sign applications

1 N

Buck-boost converter Active clamp forward converter

VPV

M1

minus

+ minus

+

minus

+

M2 M3

M4C1

C2

L1

L2D1 D2

D3

Lk

Lm

Tf

S1

CO VORL

VB

Figure 2 Schematic diagram of the hybrid converter for electronic sign applications

converter are integrated with synchronous switch technique[13] to reduce their component counts as shown in Figure 3With this circuit structure the proposed one can yield higherefficiency reduce weight size and volume and increase thedischarging time of battery under the same storing energysignificantly

The proposed hybrid converter using PV arrays suppliespower to LED lighting for electronic sign applications Theproposed one includes charger and discharger Since theproposed one uses PV arrays as its power source it mustbe operated at the maximum power point (MPP) of PVarrays to extract its maximumpowerManymaximumpowerpoint tracking (MPPT) methods of PV arrays have beenproposed [14ndash21] They are respectively power matching[14 15] curve-fitting [16 17] perturb-and-observe [18 19]and incremental conductance [20 21] methods Since powermatching method requires a specific insolation condition orload it will limit its applications MPPT using curve-fittingtechnique needs prior establishment characteristic curve ofPV arrays It cannot predict the characteristics includingother factors such as aging temperature and a possible

breakdown of individual cell The incremental conductancetechnique requires an accurate mathematical operation Itscontroller is more complex and higher cost Due to a simplercontrol and lower cost of perturb and observe method theproposed hybrid converter adopts the perturb and observemethod to implement MPPT

For electronic sign applications using LED the powersystem needs battery to store energy during day and todischarge energy for driving LED during night In orderto generate better performances of battery charging manybattery charging methods have been proposedThey are con-stant trickle current (CTC) constant current (CC) and CCand constant-voltage (CC-CV) hybrid charge methods [22]Among these methods the CTC charging method needs alarger charging time Battery charging using CC-CVmethodrequires to sense battery current and voltage resulting in amore complex operation and higher cost Due to a simplercontroller of battery charger using CC charging method itis adopted in the proposed hybrid converter According todescription above the proposed hybrid converter uses theperturb and observe method to track MPP of PV arrays and


1 NVPV

M1

minus

+ minus

+

minus

+

M2

L2D1 D2

D3

Lk

Lm

Tf

S1

CO

CC

VORL

VB

IN2

IN1IDS1 IDS2ILm

ILk

IOIPV

IL2

IB

Figure 3 Schematic diagram of the proposed hybrid converter for electronic sign applications

adopts the CC charging method to simplify battery chargingAll overall power system can achieve battery charging andLED driving

2 Circuit Structure Derivation ofthe Proposed Hybrid Converter

The hybrid converter consists of a bidirectional buck-boostand active clamp forward converter as shown in Figure 2Due to complementary operation between two converterstwo switch pairs of (119872

11198724) and (119872

21198723) can be operated

in synchronous It will do not affect the operation of theproposed original converter Since switch pairs of (119872

21198723)

has a common node they meet the requirements of switchintegration technique [11] According to principle of switchintegration technique switches 119872

2and 119872

3can be merged

as shown in Figure 4(a) Since charger and LED drivingcircuit (discharger) are operated in complementary switch1198722and 119872

3are also regarded as an independent operation

Therefore voltages across switches1198722and119872

3are the same

value in each operation state Diode 119863119865231

and 119863119865232

canbe removed while diode 119863

119861231and 119863

119861232can be shorted

as shown in Figure 4(b) In Figure 4(b) since 119871119870≪ 119871119898

119871119870can be neglected The inductor 119871

1and magnetizing 119871

119898

are connected in parallel Although features of inductors 1198711

and 119871119898are different their design rules are to avoid them

to operate in saturation condition Therefore they can bemerged as inductor 119871

1119898 as shown in Figure 4(c)

From Figure 4(c) it can be seen that switch 1198721and

1198724have a common node They can use switch integration

technique to combine them as shown in Figure 4(d) Sincevoltages across119872

1and119872

4are the same values diodes119863

119861141

and 119863119861142

are shorted and diodes 119863119865141

and 119863119865142

can beremoved as shown in Figure 4(e) From Figure 4(e) it canbe found that capacitors 119862

1and 119862

2are connected in parallel

They can be integrated as capacitor 119862119862 as illustrated in

Figure 4(f) To simplify symbols of components illustrated inFigure 4(f) component symbols will be renamed as shownin Figure 3 Note that switch 119878

1can be operated by manual

or automatic method to control the operational states of theproposed hybrid converter

Buck-boost and forward converters are combined to formthe proposed hybrid converter Since operation of buck-boost converter is the same as the conventional buck-boot

converter its operational principle is described in [8] Itwill not be described in this paper The forward converterwith the active clamp circuit recovers the energies stored inmagnetizing and leakage inductors of transformer 119879

119891and

achieves zero-voltage switch (ZVS) at turn-on transition forswitches119872

1and119872

2 It operational mode can be divided into

9 modes and their Key waveforms are illustrated in Figure 5since their operational modes are similar to those modes ofthe conventional converter illustrated in [23] It is also notdescribed in this paper

3 Design of the Proposed Hybrid Converter

The proposed hybrid converter consists of buck-boost con-verter and active clamp forward converter Since switchesand inductors in two converters are integrated with thesynchronous switch technique design of the proposed onemust satisfy requirements of each converter Since design ofthe active clamp forward converter is illustrated in [23] buck-boost converter is only analyzed briefly in the following

31 Buck-Boost Converter Since buck-boost converter isregarded as the battery charger under constant currentcharging Its design consideration is to avoid a completelysaturation of inductorTherefore duty ratio119863

11and inductor

119871119898are analyzed in the following

311 Duty Ratio 11986311 Within charging mode since battery

voltage119881119861is regarded as a constant voltage during a switching

cycle of the proposed hybrid converter maximum duty ratio11986311(max) of the proposed one can be determined by volt-

second balance of inductor 119871119898 Its relationship is expressed

as

119881PV(min)11986311(max)119879119904 + (minus119881119861(max)) (1 minus 11986311(max)) 119879119904 = 0(1)

where 119881PV(min) is the minimum output voltage of PV arrays119881119861(max) is the maximum voltage across battery and 119879

119904repre-

sents the period of the proposed hybrid converter From (1)it can be found that119863

11(max) can be illustrated by

11986311(max) =

119881119861(max)

119881PV(min) + 119881119861(max) (2)


1 NVPV

M1 M23

M4

minus

+ minus

+

minus

+

L2

L1

D1

DF232

DB232

DF231

DB231 D2

D3

Lk

Lm

Tf

S1

CO

C2

C1

VORL

VB

(a)

1 N

M23

M4

L2D2

D3

Lk

Lm

Tf

S1

CO

C2

VPV

M1

minus

+

L1

D1

C1

minus

+

VORL

minus

+

VB

(b)

1 N

M23D3Lm

S1

CO

M4

C2

M1

VPVminus

+

L2D2TfD1

C1

minus

+

VORL

minus

+

VB

(c)

1 N

M23

DB141

DB142

DF142

DF141

D3Lm

S1

CO

C2

M14

VPVminus

+ minus

+

L2D2TfD1

C1

RL

minus

+

VB

VO

(d)

1 N

M23

D3Lm

S1

CO

C2

M14

VPVminus

+

L2D2TfD1

C1

RL

minus

+

VB

minus

+

VO

(e)

1 N

D3Lm

S1

CO

CC

M23

M14

VPVminus

+

L2D2TfD1

RL

minus

+

VB

minus

+

VO

(f)

Figure 4 Derivation of the proposed hybrid converter for battery charger


0

0

0

0

0

0

0

0

0

0

0

0

0

t0 t1t2 t3t4t5 t7t8t9t6

VO

IO

VCC

IN2

IN1

ILm

ILk

VDS2

IDS2

VDS1

IDS1

M2

M1

Figure 5 Key waveforms of forward converter with active clampcircuit over one switching cycle

Moreover transfer ratio11987211(max) can be also determined as

follows

11987211(max) =

11986311(max)

1 minus 11986311(max) (3)

When type of battery is chosen its maximum chargingcurrent 119868

119861(max) is also determined The charging current 119868119861

can be changed from its maximum charging current 119868119861(max)

to 0 by variation duty ratio11986311of switch119872

1

312 Inductor 119871119898 Since the proposed hybrid converter is

operated in CCM to obtain the maximum charging current119868119861(max) its conceptual waveforms of inductor current 119868

119871119898and

charging current 119868119861are illustrated in Figure 6 If the proposed

one is operated in the boundary of discontinuous conductionmode (DCM) and CCM the charging current 119868

119861is expressed

by

119868119861(av) =(1 minus 119863

11)2

119881119861119879119904

2119871119898119861

(4)

where 119871119898119861

is the inductance 119871119898at the boundary condition

According to (4) variation of duty ratio 11986311

can obtain

t0

t0

t

t

0

0M1

IB

ILm

IB(a)

TS

D11TS (1 minus D11)TS

ΔILm(max)ILm(0)

Figure 6 Conceptual waveforms of inductor current 119868119871119898

andcharging current 119868

119861in buck-boost converter

a different charging current 119868119861 In general the maximum

charge current 119868119861(av)max occurs at the maximum battery

voltage 119881119861(max) and the maximum output voltage 119881PV(max)

of PV arrays Therefore the boundary inductor 119871119898119861

can bedetermined by

119868119861(av)max =

(1 minus 11986311119861)2

119881119861(max)119879119904

2119871119898119861

(5)

where 11986311119861

is duty ratio of switch 1198721under 119881

119861(max) and119881PV(max) From (5) it can be found that the maximumboundary inductor 119871

119898119861(max) can be expressed as

119871119898119861(max) =

(1 minus 11986311119861)2

119881119861(max)119879119904

119868119861(av)max

(6)

Since the proposed hybrid converter is operated in CCMinductor 119871

119898must be greater than 119871

119898119861(max) Therefore when119881PV(max)119881119861(max) 119868119861(av)max and119879119904 are specified theminimuminductance 119871

119898(min) (=119871119898119861(max)) can be determinedIn order to avoid the core of transformer 119879

119891operated

in saturation state the working flux density 119861max must beless than the saturation flux density 119861sat of core Since 119861maxis proportional to the maximum inductor current 119868

119871119898(peak)119868119871119898(peak) must be first determined In Figure 6 119868

119871119898(peak) canbe expressed as

119868119871119898(peak) = 119868119871119898(0) + Δ119868119871119898(max) (7)

where 119868119871119898(0)

is the initial value of inductor current 119868119871119898

operated in CCM and Δ119868119871119898(max) represents its maximum

variation value In general its maximum value Δ119868119871119898(max) can

be determined by

Δ119868119871119898(max) =

119881PV(max)11986311119877119879119904

119871119898

(8)


where 11986311119877

represents the duty ratio of switch 1198721under

119881PV(max) and 119881119861(max) Furthermore the maximum chargingcurrent 119868

119861(av)max can be expressed as

119868119861(av)max =

(1 minus 11986311119877)

2

(119868119871119898(peak) + 119868119871119898(0))

=

(1 minus 11986311119877)

2

(2119868119871119898(0)+

119881PV(max)11986311119877119879119904

119871119898

)

(9)

From (9) the initial value 119868119871119898(0)

can be determined as follows

119868119871119898(0)=

119868119861(av)max

1 minus 11986311119877

minus

119881PV(max)11986311119877119879119904

2119871119898

(10)

From (7) (8) and (10) 119868119871119898(peak) can be denoted as

119868119871119898(peak) =

119868119861(av)max

1 minus 11986311119877

+

119881PV(max)11986311119877119879119904

2119871119898

(11)

According to datasheet of core which is supplied by coremanufacturer the number of turns 119873

1on the primary side

of transformer 119879119891can be obtained by

1198731= radic119871119898

119860119871

(12)

where 119860119871represents nH per turns2 That is 119871

119898= 1198732

1119860119871 By

applying Faradayrsquos law 119861max can be determined as

119861max =119871119898119868119871119898(peak) times 10

4

1198731119860119888

(13)

where119860119888is the effective cross-section area of the transformer

core In order to avoid saturation condition of core119861maxmustbe less than saturation flux density 119861sat of core

4 Configuration of the ProposedPV Hybrid Converter

Since the proposed PV power system includes charger anddischarger and adopts PV arrays as its power source itscircuit structure and control algorithm are described in thefollowing

41 Circuit Structure of the Proposed PV Power System Theproposed PV power system consists of battery charger LEDdriving circuit (discharger) and controller as shown inFigure 1 The battery charger and LED driving circuit usingbuck-boost and active clamp forward hybrid converter areshown in Figure 3 In addition controller adopts microchipand PWMIC formanaging battery charging and LEDdrivingcircuit The microchip is divided into 3 units (MPPT powermanagement and battery management units) to implementMPPT of PV arrays and battery charging The PWM IC isused to regulate output voltage of LED driving circuit Inthe microchip of the controller the MPPT unit senses PVvoltage 119881PV and current 119868PV to achieve MPPT which adopts

Table 1 Parameter definitions of control signals shown in Figure 4

Symbol Definition119881PV Output voltage of PV arrays119868PV Output current of PV arrays119875119875 Maximum power of PV arrays119875119861 Charging power of battery (119875

119861= 119881119861119868119861)

119875119861(max)

Maximum charging power of battery(119875119861(max) = 119881119861119868119861(max))

119868119861 Charging current of battery119881119861 Battery voltage119881119874 Output voltage of forward converter119868119874 Output current of forward converter119881119861(max) Maximum voltage of battery119881119861(min) Minimum voltage of battery

119881119874(max)

Maximum output voltage of forwardconverter

119868119874(max)

Maximum output current of forwardconverter

119868119861(max) Maximum current of battery

119881refReference voltage for obtaining thedesired output voltage 119881

119874

119868119862

Current command for obtaining thedesired charging current 119868

119861

119878119872

Control signal of operational mode (119878119872=

0 battery charging 119878119872= 1 LED driving)

119878119899

Insolation level judgment (119878119899= 1 low

insolation level 119878119899= 0 high insolation

level)119881119891 Feedback signal of PWM IC119881119890 Error value1198661 1198662 pwM signals

11987211198722 Gate signals of switches119872

1and119872

2

perturb and observe method The battery management unitacquires battery voltage 119881

119861and current 119868

119861for implementing

CC charging of battery Since the proposed hybrid converteris required to match MPPT of PV arrays and CC chargingmode the power management unit can manage the powerflow between PV arrays and battery depending on the rela-tionship of the generated power of PV arrays and the requiredpower of battery charging All of protections are imple-mented by microchip The protections include overcurrentand overvoltage protections of the proposed hybrid converterand undercharge and overcharge of battery Therefore theproposed one can achieve the optimal utility rate of PV arraysand a better performance of battery charging

42 Control Algorithm of the Proposed Hybrid Converter InFigure 1 the controller of the proposed hybrid converterincludes microchip and PWM IC to achieve battery chargingand LED driving In order to implement battery charging andLED driving block diagram of the hybrid converter is shownin Figure 7 In Figure 7 control signals are defined in Table 1


Photosensor

Operationalmode

judgment

Switch selector PWMgenerator

Erroramplifer

Mode selector

Te proposed hybrid converter

A

B

(PWM IC)

Perturband

observemethod(MPPT

unit)

Batterymanagement

unit

Powermanagement

unit

Microchip

Controller

LOD2

D3

Lk

Lm

Tf

S1

Sn

SM

SM

SM

SP

CO

CCVPV

VPV

M1

M2

minus

+

D1

minus

+

VO

VO

VB(max)

VO(max)

VB(mix)

VB

VeVf

Vref

Vref

RL

IC

IPV

IB

IO

IO(max)

IB(max)

G1

G2

PB(max)

PB

PP

1 N

Figure 7 Block diagram of the proposed hybrid converter

In the following control algorithms for battery charging andLED driving are briefly described

421 Battery Charging Since the proposed hybrid convertersupplies power to load from PV arrays the proposed onemust perform MPPT for PV arrays and battery chargingfor battery The MPPT control method and battery chargingmethod are described as follows

MPPT Algorithm Since solar cell has a lower output voltageand current a number of solar cells are connected in seriesand parallel to form PV arrays for attaining the desired PVvoltage and current Their output characteristic variationsdepend on ambient temperature and insolation of sun Figure8 illustrates P-V curve of PV arrays at different insolation ofsun from which it can be seen that each insolation level hasa maximum power 119875max where 119875max 1 is the most insolation

of sun while 119875max 3 is the one at the least insolation Threemaximum power point 119875max 1 sim 119875max 3 can be connect by astraight lineThe operational area is divided into two areas Aarea and B area When operational point of PV arrays locatesin A area output current 119868PV of PV arrays is decreased tomake the operational point close to itsmaximumpower point(MPP) If operational point is set in B area current 119868PV willbe increased to operate PV arrays at its MPP

The proposed power system adopts perturb and observemethod to implement MPPT Its flowchart is shown inFigure 9 In Figure 9119881

119899and 119875119901separately represent their old

voltage and power and119875119899(=119881119899119868119899) is its newpower According

to flowchart procedures of MPPT using perturb and observemethod first step is to read new voltage 119881


119899of

PV arrays and then to calculate new PV power 119875119899 Next step

is to judge relationship of 119875119899and 119875119901 Since the relationship of

119875119899and119875119901has three different relationships they are separately


0

P

V

A area B area

Increase loadPmax1

Pmax2Pmax3

Figure 8 Plot of P-V curve for PV arrays at different isolation ofsun

119875119899gt 119875119901 119875119899= 119875119901 and 119875

119899lt 119875119901 Each relationship can

be corresponded to the different relationship of 119881119899and 119881

119901

Therefore when the relationship of 119875119899and 119875

119901is decided

next step is to find the relationship of 119881119899and 119881

119901 According

to the relationship of P-V curve of PV arrays when therelationships of 119875

119899and 119875119901and 119881

119899an 119881119901are decided working

point of PV arrays can be specified When working point ofPV arrays locates in A area power system connected in PVarrays to supply load power must decrease output power toclose the distance between working point and MPP of PVarrays On the other hand when working point sets in B areapower system must increase output power to make workingpoint to approachMPP of PV arrays Finally119881

119901is replaced by

119881119899and119875119901is also substituted by119875

119899The procedure of flowchart

returns first step to judge next working point of PV arraysMoreover when 119875

119899= 119875119875and 119881

119899= 119881119875 working point of

PV arrays set in the MPP of PV arrays The maximum power119875119875of PV arrays is transferred to power management unit for

regulating power of battery chargingBattery Charging Method The proposed hybrid converteruses CC charging method to charge battery Accordingto battery specifications charging voltage and current arelimited for extending its life cycle Therefore the powerlimitation curve for battery charger will be limited Figure 10depicts conceptual waveforms of charging current voltageand power for battery charger with CC charging methodThe battery charging time is from 119879

0to 119879119888 When 119905 = 119879

0

the proposed power system begins to charge battery andbattery voltage 119881

119861is at the minimum value 119881

119861(min) When119905 = 119879

119888 battery is charged to its maximum voltage 119881

119861(max)According to limitation of the maximum battery chargingcurrent 119868

119861(max) power limitation curve of battery chargingcan be determined from 119879

0to 119879119888 The charging power of

battery follows the power limitation curve for extending itslife cycle

Since power limitation curve of battery has upper andlower values they are respectively 119875

119861(min) (=119881119861(min)119868119861(max))and 119875

119861(max) (=119881119861(max)119868119861(max)) According to relationship

among 119875PV(max) 119875119861(min) and 119875119861(max) they can be dividedinto three operational states 119875PV(max) lt 119875119861(min) 119875119861(min) le119875PV(max) lt 119875119861(max) and 119875PV(max) gt 119875119861(max) as shown inFigure 11 When 119875PV(max) lt 119875119861(min) power curve of batterycharging follows 119875PV(max) When 119875

119861(min) le 119875PV(max) lt 119875119861(max)power limitation curve and 119875PV(max) intersects at A point

where its intersecting time is 119879119860 Power curve tracks power

limitation curve before 119905 = 119905119860 while it traces119875PV(max) after 119905 =

119905119860 as shown in Figure 11(b) If operational state of 119875PV(max) gt119875119861(max) power curve is regulated by power limitation curve as

shown in Figure 11(c) As mentioned above battery chargingcan be operated in a better charging mode

In order to implement a better battery charging powermanagement and battery management units are adopted andthey are implemented by microchip In the following powermanagement and battery management are briefly described

(1) Power Management In Figure 7 the controller includesmicrochip and PWM IC When the microchip is usedto execute power management its control procedures aredepicted in Figure 12 First step is to set 119878

119875= 0 and

then is to read control signals The control signals include119881119874(max) 119881119861(max) 119881119861(min) 119881119861 119868119861 119868119861(max) 119868119874 119868119874(max) 119881ref and119875119875 When control signals are obtained by microchip next

step is to calculate 119875119861(max) (=119881119861119868119861(max)) and 119875

119861(=119881119861119868119861)

Since 119875119875 which is attained by MPPT control method is the

maximum output power of PV arrays when 119875119875ge 119875119861(max)

is confirmed 119875set is set to equal 119875119861(max) If 119875119875 ge 119875119861(max)

is denied 119875set = 119875119875 The 119875119875is the power command of

battery charging Therefore power error value ΔP can bedetermined It is equal to (119875set minus119875119861) When ΔP is determinedcurrent command 119868

119862can be obtained It is equal to (ΔP119881

119861)

The current command 119868119862is sent to PWM IC to determine

gate signals 1198661and 119866

2 Next step is to judge next current

command

(2) Battery Management In Figure 12 the right hand side offlowchart shows procedures of battery management Whenthe microchip reads control signals the procedure of batterymanagement is to judge overcurrent condition When 119868

119874ge

119868119874(max) is confirmed overcurrent condition of the proposedhybrid converter occurred When overcurrent conditionoccurred signal 119878

119875is set to 1 The 119878

119875is sent to PWM

IC to shutdown PWM generator and the proposed hybridconverter is also shutdown Next step is to judge next currentcommand Moreover when 119881

119874ge 119881119874(max) (overcharge

condition) 119881119861le 119881119861(min) (undercharge condition) and 119881119861 ge

119881119861(max) (overcharge condition) the control procedure enters

to set 119878119875= 1 and to shutdown the proposed hybrid

converter According to previously describing proceduresbattery can be properly controlled to complete a bettercharging condition

(3) PWM IC In the battery charging mode PWM IC isused to control charging current with CC method Firstphotosensor is used to detect insolation level of sun Wheninsolation is a high level 119878

119899= 0 If insolation is a low

level 119878119899= 1 The signal 119878

119899is sent to operational mode

judgment to obtain mode control signal 119878119872 When 119878

119872= 0

the hybrid converter enters battery charging mode That isthe insolation of sun is at a high level and 119878

119899= 0 If 119878

119872= 1 the

one is in LED driving modeThe signal 119878119899= 1 and insolation

is at a low level The mode control signal 119878119872is sent to mode

selector switch selector and switch 1198781 When mode selector


Start

Calculate

Judge

Judge

Decreaseload

Increaseload

To powermanagement

unit

Replace

A area A area A areaB areaB areaB area

Read Vn In

Pn = VnlowastIn

Pn = PP Pn gt PP

Pn lt PP

Pn and PP

Vn and PP

JudgeVn and PP

JudgeVn and PP

Vn = PP

PP

Vn gt PP Vn lt PP

Vn lt PP

Vn lt PP Vn ge PPVn ge PP

(ΔI)

Decreaseload(ΔI)

Decreaseload(ΔI)(ΔI)

Increaseload(ΔI)

Increaseload(ΔI)

VP = Vn

PP = Pn

= PP = Pn

Set PPV(max)

Figure 9 Flowchart of MPPT using perturb and observe method for PV arrays system

receives 119878119872= 0 the feedback signal 119881

119891is set to equal IC

The feedback signal 119881119891and reference single 119881ref are sent to

error amplifier to obtain error value 119881119890 When error value 119881

119890

is attained PWM generator can depend on 119881119890to determine

duty ratios of PWM signals 1198661and 119866

2 When 119866

1and 119866

2are

specified and 119878119872= 0 switch selector can set that119872

1= 1198661

and 1198722= 1198662to control the charging current 119868

119861of battery

As mentioned above the proposed hybrid converter can usemicrochip and PWM IC to achieve battery charging

422 LED Driving The LED driving mode is regarded asbattery discharging mode When operational mode entersLED driving mode 119878

119872= 1 The mode selector can be

operated to set 119881119891= 119881119874 The 119881

119891and 119881ref are sent to error

amplifier to attain 119881119890 The 119881

119890is through PWM generator to

generate signals 1198661and 119866

2 Since 119878

119872= 1 switch selector is

controlled by 119878119872

to set 1198721= 1198662and 119872

2= 1198661 Therefore

the proposed hybrid converter can depend on duty ratios of

gate signals1198721and119872

2to supply power to LED until battery

voltage119881119861is equal to or less than119881

119861(min) When119881119861le 119881119861(min)

the proposed hybrid converter is shutdown

5 Experimental Results

In order to verify the circuit analysis and component design ofthe proposed power system a prototype which is composedof a charger and LED driving circuit (discharger) with thefollowing specifications was implemented

51 Buck-Boost Converter (Charger)

(i) Input voltage 119881PV DC 17Vsim21 V (PV arrays)(ii) Switching frequency 119891

1199041 250KHz

(iii) Output voltage 119881119861 DC 5sim7V (lead-acid battery

6 V23 Ah)(iv) Maximum output current 119868

119861(max) 23A


(power limitation curve)

tCharging time0

PmaxVB(min)

VB(max)

IB(max)

TOTC

Figure 10 Conceptual waveforms of charging current voltage and power for battery charger with CC charging method

0

Power limitation curve

Power curve

PB(max)

PB(min)

PPV(max)

tCharging timeTO TC

(a) 119875PV(max) lt 119875119861(min)

0


Power curve

A

PB(max)

PB(min)PPV(max)

tCharging time

TO TA TC

(b) 119875119861(min) le 119875PV(max) lt 119875119861(max)

0


Power curve

tCharging time

PB(max)

PB(min)

PPV(max)

TO TC

(c) 119875PV(max) gt 119875119861(max)

Figure 11 Conceptual waveforms of maximum output power of PV arrays and power limitation and power curves of battery from 119879119874to 119879119862

(a) 119875PV(max) lt 119875119861(min) (b) 119875119861(min) le 119875PV(max) lt 119875119861(max) and (c) 119875PV(max) gt 119875119861(max)

52 Active Clamp Forward Converter (LED Driving Circuit)

(i) Input voltage 119881119861 DC 5Vsim7V

(ii) Switching frequency 1198911199042 250KHz

(iii) Output voltage 119881119874 DC 10V

(iv) Maximum output current 119868119874(max) 2A

According to previously specifications and design of thehybrid converter inductors 119871

2and 119871

119898and capacitor 119862

119862can

be determined In (17) illustrated in [23] since inductor 119871119898

must be greater than 709 uH under 119881119874= 10V 119881

119861= 7V

and119881PV = 17V 1198712 is chosen by 40 uH According to (6) and(22) illustrated in [23] the magnetizing inductor 119871

119898must

be greater than 68 uH under 119873 = 5 119881119861= 7V and 119871

2=

40 uHTherefore magnetizing inductor 119871119898is determined by

40 uH while its leakage inductor 119871119870is obtained by 02 uH

Moreover capacitor 119862119862can be attained by (29) illustrated in

[23] Its capacitance119862119862is 022 nF under119873 = 5 and119881

119861= 7V

Therefore119862119862is chosen by 024 uFThe components of power

stage in the proposed hybrid converter was determined asfollows

(i) switches11987211198722 PSMN005-75B

(ii) diodes11986311198632 UF601

(iii) transformer 119879119891 EE-25 core

(iv) inductor L2 EE-22 core


Start

Read

Calculate

Shutdownpowersystem

Judge

Judgeovervoltage

overcharge

overvoltage

Judgeundervoltage

Judge

ToPWM

IC

No Yes

Yes

Yes

No

Yes

No

Battery management(battery management)

Power management(power management)

No

Shutdownpowersystem

Judge

Judgeovervoltage

overcharge

overvoltage

Judgeundervoltage

Judge

ToPWM

IC

No Yes

Yes

Yes

No

Yes

No

No

Judge

Set

No Yes

Calculate

ToPWM IC

Set SP = 0

VO(max) VB(man) VB(min)PP IB VB IO Vref

IB(max) IO(max)

PB(max) = VBIB(max)PB = VBIB

SetSP = 1

PP ge PB(max)

Pset = PP

SetPset = PB(max)

CalculateΔP = Pset minus PB

IC = ΔPVB

IO ge IO(max)

VO ge VO(max)

VB ge VB(max)

VB le VB(min)

Figure 12 Flowchart of power and battery managements of the proposed hybrid converter

(v) capacitor 119862119874 47 120583F25V and

(vi) switch 1198781 IRFP540

Since the charging current 119868119861of battery can be varied by

duty ratio of switch 1198721in buck-boost converter its value

is proportional to duty ratio 11986311 Figures 13(a) and 13(b)

respectively depict the measured voltage 119881119863119878

waveform ofswitch 119872

1and current 119868

119861waveform under duty ratio of

031 and 035 illustrating that the charging current 119868119861can

be increased by duty ratios increase Measured waveformsof PV arrays current 119868PV and voltage 119881PV using the perturband observe method are illustrated in Figure 14 Figure 14(a)illustrates the MPP of PV arrays at 10W while Figure 14(b)depicts that at 20W Figure 15 shows the measured batteryvoltage 119881


119861under MPP of PV arrays at 10W

from which it can be found that the maximum charging


VDS

IB

LeCroy

(VDS 25Vdiv IDS 500mAdiv 1 120583sdiv)

(a)

VDS

IB

LeCroy


(b)

Figure 13 Measured voltage 119881119863119878

waveform of switch1198721and the charged current 119868

119861waveform operated in duty ratio of (a) 031 and (b) 035

for working in the charging state

VPV

IPV

PPV

LeCroy

(VPV 20Vdiv IPV 2Adiv PPV 50Wdiv time 100msdiv)

(a)

VPV

IPV

PPV

LeCroy


(b)

Figure 14 Measured voltage 119881PV current 119868PV and power 119875PVwaveforms of PV arrays using the perturb and observe method totrack MPPT of arrays (a) under 119875PV(max) = 10W and (b) under119875PV(max) = 20W

current 119868119861is limited at 15 A under battery voltage119881

119861of 65 V

due to control of power managementWhen the proposed hybrid converter is operated in the

discharging state (LED driving state) active clamp forward isin workingMeasured voltage119881

119863119878and current 119868

119863119878waveforms

of switched1198721and119872

2are respectively illustrated in Figures

16 and 17 Figures 16(a) and 16(b) show those waveformsunder 20 of full load while Figures 17(a) and 17(b) depictthose waveforms under full load From Figures 16 and 17 itcan be seen that switches119872

1and119872

2are operated with ZVS

VB

IB

LeCroy

(VB 5Vdiv IB 1Adiv time 10msdiv)

Figure 15 Measured battery voltage VB and current IB waveformsunder 119875PV(max) = 10W

at turn-on transition Comparison of conversion efficiencybetween forward converter with hard-switching circuit andwith the proposed active clamp circuit from light load toheavy load is depicted in Figure 18 from which it can befound that the efficiency of the proposed converter is higherthan that of hard-switching one Its maximum efficiency is90 under 80 of full load and its efficiency is 83 under fullload Figure 19 illustrates step-load change between 20 offull load and full load illustrating that the voltage regulation119881119874has been limited within plusmn2 From experimental results

it can be found that the proposed hybrid converter is suitablefor electronic sign applications

6 Conclusion

In this paper the buck-boost converter combined withactive clamp forward converter to form the proposed hybridconverter is used to implement battery charger and driv-ing LED Circuit derivation of the hybrid converter withswitch integration technique is presented in this paper toreduce component counts Operational principle steady-state analysis and design of the proposed hybrid converterhave been described in detail From efficiency comparisonbetween forward converter with hard-switching circuit andwith the proposed active clamp circuit the proposed activeclamp converter can yield higher efficiency An experimental


VDS1

IDS1

ZVS

(V 10Vdiv IDS1DS1 4Adiv 1120583sdiv)

(a)

VDS2

IDS2

ZVS


(b)


and current 119868119863119878

waveforms of (a) switch1198721and (b) switch119872

2for working under 20 of full load

ZVS


VDS1

IDS1

(a)

ZVS

VDS2

IDS2


(b)


and current 119868119863119878


2for working under full load

6065707580859095

Effici

ent (

)

10 20 30 40 50 60 70 80 90 100Load ()

With propose oneWith hard switching circuit

Figure 18 Comparison conversion efficiency between the conven-tional hard-switching forward converter and the proposed one fromlight load to heavy load for working in the discharging state

prototype for a battery charger for lead-acid battery of6V23 Ah anddischarger for LEDdriving under 10V2Ahasbeen built and evaluated achieving the maximum efficiencyof 90 under 80 of full load and verifying the feasibil-ity of the proposed hybrid converter Moreover constantcurrent charging method MPPT with perturb and observemethod and power management have been implemented bymicrochip and PWM IC

VO

IO

(VO 5Vdiv IO 1Adiv 500120583sdiv)

Figure 19 Output voltage119881119874and output current 119868

119874under step-load

charges between 30 and 100 of full load of the proposed forwardconverter operated in the discharging state



References

[1] Y Jiang A W Leung J Xiang and C Xu ldquoLED light-activatedhypocrellin B induces mitochondrial damage of ovarian cancercellsrdquo International Journal of Photoenergy vol 2012 Article ID186752 5 pages 2012


[2] C-H Liu C-Y Hsieh Y-C Hsieh T-J Tai and K-H ChenldquoSAR-controlled adaptive off-time technique without sensingresistor for achieving high efficiency and accuracy LED lightingsystemrdquo IEEE Transactions on Circuits and Systems I vol 57 no6 pp 1384ndash1394 2010

[3] C C Chen C-Y Wu and T-F Wu ldquoLED back-light drivingsystem for LCD panelsrdquo in Proceedings of The Applied PowerElectronics Conference and Exposition pp 19ndash23 2006

[4] S-Y Tseng and C-T Tsai ldquoPhotovoltaic power system with aninterleaving boost converter for battery charger applicationsrdquoInternational Journal of Photoenergy vol 2012 Article ID936843 15 pages 2012

[5] J-H Park and B-H Cho ldquoNonisolation soft-switching buckconverter with tapped-inductor for wide-input extreme step-down applicationsrdquo IEEE Transactions on Circuits and SystemsI vol 54 no 8 pp 1809ndash1818 2007

[6] K Mehran D Giaouris and B Zahawi ldquoStability analysis andcontrol of nonlinear phenomena in boost converters usingmodel-based Takagi-Sugeno fuzzy approachrdquo IEEE Transac-tions on Circuits and Systems I vol 57 no 1 pp 200ndash212 2010

[7] B Bryant and M K Kazimierczuk ldquoOpen-loop power-stagetransfer functions relevant to current-mode control of boostPWM converter operating in CCMrdquo IEEE Transactions onCircuits and Systems I vol 52 no 10 pp 2158ndash2164 2005

[8] M R Yazdani H Farzanehfard and J Faiz ldquoEMI analysisand evaluation of an improved ZCT flyback converterrdquo IEEETransactions on Power Electronics vol 26 no 8 pp 2326ndash23342011

[9] K I Hwu and T J Peng ldquoA novel buck-boost convertercombining KY and buck convertersrdquo IEEE Transactions onPower Electronics vol 27 no 5 pp 2236ndash2241 2012

[10] F Xie R Yang and B Zhang ldquoBifurcation and border collisionanalysis of voltage-mode-controlled flyback converter based ontotal ampere-turnsrdquo IEEE Transactions on Circuits and SystemsI vol 58 no 9 pp 2269ndash2280 2011

[11] B-R Lin and H-Y Shih ldquoImplementation of a parallel zero-voltage switching forward converter with less power switchesrdquoIET Power Electronics vol 4 no 2 pp 248ndash256 2011

[12] D Wang X He and J Shi ldquoDesign and analysis of aninterleaved flybackforward boost converter with the currentautobalance characteristicrdquo IEEE Transactions on Power Elec-tronics vol 25 no 2 pp 489ndash498 2010

[13] T-F Wu and T-H Yu ldquoUnified approach to developing single-stage power convertersrdquo IEEE Transactions on Aerospace andElectronic Systems vol 34 no 1 pp 211ndash223 1998

[14] K I Hwu W C Tu and C R Wang ldquoPhotovoltaic energyconversion system constructed by high step-up converter withhybrid maximum power point trackingrdquo International Journalof Photoenergy vol 2013 Article ID 275210 9 pages 2013

[15] A Braunstein and Z Zinger ldquoOn the dynamic optimal couplingof a solar cell array to a load and storage batteriesrdquo IEEEtransactions on Power Apparatus and Systems vol 100 no 3 pp1183ndash1188 1981

[16] A SWeddell G VMerrett andM A-H Bashir ldquoPhotovoltaicsample-and-hold circuit enablingMPPT indoors for low-powersystemsrdquo IEEE Transaction on Circuits and Systems vol 59 no6 pp 1196ndash1204 2012

[17] S M M Wolf and J H R Enslin ldquoEconomical PV maximumpower point tracking regulator with simplistic controllerrdquo inProceedings of the EEE 24th Annual Power Electronics SpecialistConference pp 581ndash587 June 1993

[18] C-L Shen and S-H Yang ldquoMulti-input converter with MPPTfeature for wind-PV power generation systemrdquo InternationalJournal of Photoenergy vol 2013 Article ID 129254 13 pages2013

[19] R Leyva C Olalla H Zazo et al ldquoMPPT based on sinusoidalextremum-seeking control in PV generationrdquo InternationalJournal of Photoenergy vol 2012 Article ID 672765 7 pages2012

[20] N Onat ldquoRecent developments in maximum power pointtracking technologies for photovoltaic systemsrdquo InternationalJournal of Photoenergy vol 2010 Article ID 245316 11 pages2010

[21] K H Hussein I Muta T Hoshino and M Osakada ldquoMax-imum photovoltaic power tracking an algorithm for rapidlychanging atmospheric conditionsrdquo IEE Proceedings GenerationTransmission and Distribution vol 142 no 1 pp 59ndash64 1995

[22] S R Rajwade K Auluck J B Phelps K G Lyon J T Shawand E C Kan ldquoA ferroelectric and charge hybrid nonvolatilememorymdashPart II experimental validation and analysisrdquo IEEETransactions on Electron Devices vol 59 no 2 pp 450ndash4582012

[23] S Y Tseng H Y Wang and C C Chen ldquoPV power systemusing hybrid converter for LED indictor applicationsrdquo EnergyConversion and Management vol 75 pp 761ndash772 2013

Research ArticleA Novel Parabolic Trough Concentrating Solar Heating forCut Tobacco Drying System

Jiang Tao Liu Ming Li Qiong Fen Yu and De Li Ling

Solar Energy Research Institute Yunnan Normal University Kunming 650092 China

Correspondence should be addressed to Ming Li lmllldy126com

Received 31 October 2013 Accepted 4 February 2014 Published 26 March 2014


Copyright copy 2014 Jiang Tao Liu et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

A novel parabolic trough concentrating solar heating for cut tobacco drying system was established The opening width effect ofV type metal cavity absorber was investigated A cut tobacco drying mathematical model calculated by fourth-order Runge-Kuttanumerical solution method was used to simulate the cut tobacco drying process And finally the orthogonal test method was usedto optimize the parameters of cut tobacco drying process The result shows that the heating rate acquisition factor and collectorsystem efficiency increase with increasing the opening width of the absorber The simulation results are in good agreement withexperimental data for cut tobacco drying process The relative errors between simulated and experimental values are less than 8indicating that thismathematical model is accurate for the cut tobacco airflow drying processThe optimumpreparation conditionsare an inlet airflow velocity of 15ms an initial cut tobacco moisture content of 26 and an inlet airflow temperature of 200∘CThe thermal efficiency of the dryer and the final cut tobacco moisture content are 6632 and 1415 respectivelyThe result showsthat this parabolic trough concentrating solar heating will be one of the heat recourse candidates for cut tobacco drying system

1 Introduction

Solar energy currently represents the most abundant inex-haustible nonpolluting and free energy resources that couldhave a positive meaning in alleviating the global energyshortage and environmental pollution There are abundantsolar energy resources and good economic tobacco industryin Yunnan Province located in the southwest of ChinaHowever drying cut tobacco of the production process incigarette factory needs a large amount of heat energy (150∘Csim300∘C) If abundant solar energy resources are used for dryingcut tobacco of the production process in cigarette factory wecan not only save a large amount of fossil energy but alsowiden the field of utilization of solar energy In this papersolar energy drying cut tobacco process will be investigated

However the solar energy is intermittent in nature andthat received on earth is of small flux density due toatmospheric scattering and abortion making it necessaryto use large surfaces to collect solar energy for drying cuttobacco process utilization The major technologies usedfor solar energy conversion to heat are thermal processes

comprising of solar collectors There are two basic types ofsolar collectors the flat-plate and the concentrating solarcollectors The flat-plate collector has advantage of absorbingboth beam and diffused radiation and therefore still func-tions when beam radiation is cut off by the cloud The areaabsorbing solar radiation is the same as the area interceptingsolar radiation However flat-plate collectors are designedfor applications requiring energy delivered at temperaturesquite lower than 100∘C indicating that they are not suitablefor drying cut tobacco process The concentrating collectorutilizes optical systems like reflectors refractors and soforth to increase the intensity of solar radiation incidentson energy-absorbing surfaces It has the main advantage ofgenerating high temperatures andmay be used for drying cuttobacco process

Parabolic trough solar collector (PTC) is one type ofthe intermediatehigh concentrating collectors being widelyused in the field of electricity generation air-conditioningheating desalination and so on Conventional receiver usedin PTC is the vacuum tube Because of different expandingcoefficients between the metallic tube and the glass tube the



cost of vacuum tube receiver is relatively high and its usefullife is shortThereforemany researches focus on investigationof cavity absorbers Boyd et al [1] firstly developed a receiverthat consists of an annular cylindrical tube with an apertureparallel to the axis of the cylinderThe aperture is illuminatedby means of a focusing concentrator such as a lens and theradiation energy is scattered inside the cavity absorbed by thewalls and transmitted as heat to the working fluid flowingaxially inside the annulus To reduce radiative losses the tubeis surrounded by a layer of a thermal insulator Principaladvantages of this design lie in exclusive use of commonlyavailable materials and current fabrication technology Inorder to minimize the heat loss this annular cylindricaltube was improved by Barra and Franceschi [2] Eight smallcylindrical tubes were scattered inside the big cylindrical tubeand used for fluid transferring Qiaoli et al [3] proposedthe scheme of tightly connecting cluster and the inner wallof pipe Moreover Zhang et al [4] established a series ofabsorbers with triangular circular hemispherical and squareand simulated the optical performance and thermal prop-erty using the software programs TracePro and Fluent Theresults showed that the triangular cavity absorber structureowned the best optical performance and thermal propertywhich could gain collecting efficiency above 40 under thecondition of collecting temperature 150∘C Based on theseabove studies a novel cavity absorber was investigated bythis paper PTC system with V type metal cavity absorbersimilar to triangle structure would be used for drying cuttobacco process and the thermal loss is constrained bythe cavity radiation In order to lower the absorber costaluminum alloy was selected as the material of absorb-er

Because of the coupling effect among the heat transfermass transfer and energy air drying process became com-plex In order to analyze air drying cut tobacco process sim-ulation model was discussed Fukuchi et al [5] regarded thecut tobacco as equal volume sphere and simulated themotioncharacteristics indicating the simulation results agreed wellwith the experimental results The model equivalent spherewas established to simulate drying process of superheatedsteam by Pakowski et al [6] The result showed the mois-ture content of shredded tobacco simulated and the outlettemperature simulated agreed well with the experimentalvalues Based on these simulation results simulation modelof air drying cut tobacco process was studied and the processparameters of process of drying cut tobacco were optimizedin this research

In this paper parabolic trough concentrating systemwith V type metal cavity absorber was used for providingheat for cut tobacco drying system In order to reduce theincident light energy loss and prevent cavity absorber fromdeformation the width of V type metal cavity absorberwas calculated ensuring that stable heat energy providedby the PTC system is enough for cut tobacco dryingprocess Moreover a mathematical model of cut tobaccodrying process was calculated by using fourth-order Runge-Kutta numerical solution method and the accuracy of themodel compared with the experimental results was ana-lyzed

Table 1 Main parameters of cut tobacco drying system

Parabolic trough concentrator mirrorReflectivity 09Opening width 32mLighting area 30m2

Length 10mMetallic V cavity absorber

Opening width 4 cm 7 cmInternal surface coating Aluminum anodic oxide filmAbsorptivity 085sim09Emissivity 01sim02

Variable diameter drying pipePart 1 Height 12m diameter 200mmPart 2 Height 2m diameter 300mmTotal height 325m

2 Materials and Methods

21 Materials Before solar energy drying process the cuttobacco samples were pretreated by humidifier and equili-brated water content for 24 h in conditions of temperature20plusmn2∘Cand relative humidity 60plusmn2Moisture content of cut

tobacco samples was determined by YCT31-1996 standardtest methods [7]

22 Experimental Equipment

221 Parabolic Trough Concentrating Solar Heating for CutTobacco Drying System Diagram of parabolic trough con-centrating solar heating for cut tobacco drying system wasshown in Figure 1(a) The main parameters of the design aregiven in Table 1 The parabolic trough concentrating solarheating for cut tobacco drying process was illustrated asfollows Firstly when sunlight arrived at the parabolic troughconcentrator mirror the solar tracker was adjusted to makethemetal V cavity body coincidewith focal length of the PTCBecause of photothermal conversion heat transfer oil wasused to exchange heat with the cavity body and then heatedto the predetermined temperature Secondly the circulatingfan was opened and the hot air exchanged with heat transferoil was transferred into the drying pipe When temperatureof the hot air reached target temperature in the drying pipecut tobacco pretreated was put into the drying pipe for dryingprocess Thirdly when finishing the drying process productof cut tobacco dried was followed with airflow into thewhirlwind separator and separated from the airflow Fourthlypart of airflow was recycled into next drying process and theleft airflow was supplied by the valve S2 The electric heaterwas used to keep the output temperature of heat conductionoil stable when solar irradiance changed

222 Test Schematic of Energy Flux Density DistributionIn order to analyze the effect of cavity opening width onenergy flux density distribution the test schematicwas shownin Figure 2 As shown in Figure 2 the width and the focallength of parabolic trough concentrator mirror were 32m


M

P

Vent line

Fresh steam

Product in

Product out

1

2

3

4

5

6

78

9

S1

S2

Sunlight

(1) V-shaped metal cavity(2) Cyclone(3) Circulating fan(4) Air flow meter(5) Variable diameter drying pipe

(6) Heat exchanger(7) Electric heater(8) Oil pump(9) Collecting mirror

(a) Diagram of parabolic trough concentrating solar heating for cut tobacco dryingsystem

(b) Photograph of parabolic trough concen-trating solar heating for cut tobacco dryingsystem

Figure 1 Parabolic trough concentrating solar heating for cut tobacco drying system

and 117 cm respectivelyThere were a width of 10 cm Lamberttarget neutral density attenuator (ND-filter) CCD industrialcameras computers and solar tracking control system andother parts Test method was detailed by our previous work[8]

3 Cut Tobacco Drying Mathematical Model

31 Assumption In this work a relatively simple one-dimensional model [9 10] was adopted assuming no radialor axial dispersion of flow The shape of cut tobacco particleswas considered as equal volume sphere and had a uniformsize Diameter and density variations of cut tobacco particleswere ignored in the drying process The wall of drying pipehas good insulation and there was no heat exchange with theatmospheric environment

32 Mathematical Model of the Drying Process Themomen-tum equation of cut tobacco particle was as follows

119889V119901

119889119911

=

3

4

120585

120588119892(V119892minus V119901)

2

V119901119889119901120588119901

minus (1 minus

120588119892

120588119901

)

119892

V119901

minus

01

119863

(1)

The relationship between 120585 and Re in (1) was shown inTable 2 [11]

PTC

ComputerData line

ND-filter

Sunlight

x

Lamberttarget

z

CCD camera

O

Sunlight

Figure 2 The measurement schematic of focal line energy fluxdensity distribution

The movement equation of air stream was as follows

119889V119892

119889119911

= minus

119889119875

120588119892V119892119889119911

minus

119892

V119892

minus

3119873120585(V119892minus V119901)

2

4V119892119889119901

minus

120582V119892

2119863

(2)

The parameter 119873 of (2) was the number of particles perunit volume in the drying pipe and identified according to theexperimental conditions


Table 2 The relationship between 120585 and Re

Parameter Numerical valueRe 0-1 1ndash500 500ndash15000120585 24Re 24Re05 044

The differential equations of cut tobacco particles andairflow humidity distribution were as follows

119866119901119889119883 + 119866

119892119889119884 = 0 (3)

119866119875119889119883 +119882

119889119889119860 = 0 (4)

The equation of dry mass transfer rate was as follows

119882119889= 119870119910(119884eq minus 119884) (5)

The differential equation of cut tobacco particles humiditywas as follows

119889119883

119889119911

= minus

6119870119910(119884eq minus 119884) (1 + 119883)

120588119901119889119901V119901

(6)

The mass transfer coefficient 119870119910 the Schmidt number

Sc and the Prandtl number Pr in (6) were associated andcalculated according to the correlation proposed by literature[12]

ℎ

119870119910

= 119862119892(1 + 119884) (

ScPr

)

056

Pr =(119862119892]119892120588119892)

120582119892

Sc =]119892

119863V119886

(7)

Combining (3) and (6) the differential equation of airflowand humidity could be obtained as

119889119884

119889119911

= minus

6119866119901119870119910(119884eq minus 119884) (1 + 119883)

119866119892120588119901119889119901V119901

(8)

The heat balance equation of cut tobacco particles was

119889119905119901

119889119911

= minus

119862119908119905119901

119862119904+ 119862119908119883

119889119883

119889119911

+

6 (1 + 119883) [ℎ (119905119892minus 119905119901) minus 119870119910(119884eq minus 119884) (119903

0+ 119862119908119905119901) ]

120588119901V119901119889119901(119862119904+ 119862119908119883)

(9)

The convective heat transfer coefficient ℎ between the airstream and cut tobacco particles in (9) was associated withRussell number Nu and also calculated by literature [12]

The equation of gas heat balance was as follows

119889119905119892

119889119911

= minus

1199030

119862119892

119889119884

119889119911

minus

6119866119901(1 + 119883) [ℎ (119905

119892minus 119905119901) minus 119870119910(119884eq minus 119884) (119903

0+ 119862119908119905119901)]

119866119892V119901119889119901119862119892

(10)

The differential equation of pipe diameter was as follows

119889119863

119889119911

= 2 cot 120579 (11)

There 120579 was the expansion angle of the adjustable pipeAccording to the equations from (1) to (2) and (6) to

(11) the tapered tubular air drying cut tobacco mathematicalmodel was formed Then this mathematical model was cal-culated using fourth-order Runge-Kutta method and Matlabsoftware program

4 Results and Discussion

41 The heating Experiments of Parabolic Trough Concentrat-ing System Experiments were studied to analyze the effect ofcavity opening width on collector temperatures as shown inFigure 3

As can be seen from Figure 3 when the solar irradiationintensity is low the cavity with opening width of 7 cm hasfaster heating rate because the cavity with opening widthof 4 cm has relatively larger heat loss When the run timeincreases to about 1 month the cavity opening width of 4 cmhas larger deformation As for the cavity with opening widthof 7 cm it was almost not deformed except in enclosure seamsof insulated enclosureThese changes before and after heatingabout the cavity shape are shown in Figure 4

In order to analyze the effect of cavity opening width onenergy flux density distribution the test results were shownin Figure 5 During test process the direct solar radiationduring measurement was 568Wm2 As can be seen fromFigure 5 the 95 of energy is concentrated in the focal lineof concentration at the 0ndash75 cm When the cavity absorberwith opening width of 4 cm is used there is part of theenergy outside the cavity focused on the insulation enclosureThis may be because the insulation housing material and theabsorber material were different The cavity absorber withopening width of 4 cm has severe deformation when it isstressed The cavity absorber with opening width of 7 cm canabsorb most of the energy The insulation enclosure absorbsless energy and only causes deformation of shell commissure

When the impact of tracking accuracy is ignored theoptical efficiency of this collector system is equal to theproduct of mirror reflectivity acquisition factor and thecavity absorptivity The acquisition factors of the cavityabsorber with the opening width of 4 cm and 7 cm are


1320 1325 1330 1335 1340 1345 1350 135540

60

80

100

120

140

160

180

200

220

240

Collector outlet temperatureCollector inlet temperatureDirect solar radiation

Time

Tem

pera

ture

(∘C)

400

500

600

700

800

900

1000

Dire

ct so

lar r

adia

tion

(Wm

2)

(a) 7 cm


1150 1210 1230 1250 1310 133080

100

120

140

160

180

200

220

240

400

500

600

700

800

900

1000

1100

1200

1300

Time

Tem

pera

ture

(∘C)

Dire

ct so

lar r

adia

tion

(Wm

2)

(b) 4 cm

Figure 3 Effect of cavity opening width on collector temperature(experimental conditions mass of heat conduction oil = 4265 kgand flow rate = 015 kgs)

081 and 096 respectively The collector system efficiency iscalculated using the following formula

120578sys =119898119862oil (119905out minus 119905in)

119860119886119867119887

(12)

When the heat conduction oil temperature is 230∘Cthe collector system efficiency of the cavity absorber withthe opening width of 4 cm and 7 cm is 109 and 252respectively The result shows that the heating rate acqui-sition factor and collector system efficiency increase withincreasing the opening width of the absorber and can alsoprevent the cavity from deformation It naturally follows thatstability of heat conduction oil output temperature can beimproved

(a) 4 cm

(b) 7 cm

Figure 4 The changes about the cavity shape before and afterheating

00 10 20 30 40 50 60 70

0

7

14

21

28

35

42

49

56

times103

Flux

(Wm

2)

x0 (cm)

x1 = 7 cm

x2 = 4 cm

Energy flux density distribution

Figure 5 The test results of focal line energy flux density distribu-tion

42 Design and Construction of the VariableDiameter Drying Pipe

421 Approximation of Variable Diameter Drying Pipe Thispaper uses the Fedorov method on the drying tube diameterand height of approximate calculation

The number119870119894is calculated by the following equation

119870119894= 119889119901times3

radic

4119892 (120588119901minus 120588119892)

31205832

119892120588119892

(13)

Through the Figure 6 found the relationship betweenRe119901and 119870

119894and through the Figure 7 found the relationship

between Re119901and Nu the heat transfer coefficient of 119870

between air and material particles is as follows

119870 =

Nu sdot 120582119892

119889119901

(14)


1 10 102

103

01

1

10

102

103

104

Rep

Ki

2345

2345

2345

23

23

45

02

03

04

2 3 4 5 6789 2 3 4 5 2 3 4 5

Figure 6 The relationship between Re119901and 119870

119894

0 50 100

2

5

8

Nu

Rep

Figure 7 The relationship between Re119901and Nu

Suspension velocity of particles is as follows

120592119886=

Re119901120583119892

119889119901

(15)

The total surface area of the particle is as follows

119860119904=

6119866119901

119889119901120588119901

(16)

The mean temperature difference between the material andthe air is

Δ119905 =

1199051+ 1199052

2

minus 119905119904 (17)

The time of the drying process is

120591 =

119876

119870119860119904Δ119905

(18)

The length 119911 of the drying pipe is as follows

119911 = (V119892minus V119886) 120591 (19)

(a) 120579 = 50∘

(b) 120579 = 52∘

Figure 8The velocity of flow field in variable diameter drying pipe

The diameter 119863 of the drying pipe is

119863 = radic

4119866119892120588119904

3600120587120592119892

(20)

The design parameters are substituted into the aboveformula of cut tobacco drying The parameters are shown inTable 3

422 The Simulation Speed Flow of Variable Diameter DryingPipe Reducing type dry pipe consists of two different diam-eters of straight pipe and an expansion pipe connection Inthe actual environment the improper value of 120579 will lead tosediment in the drying pipe on drying process According tothe actualmodel usingANSYS Fluent for the fluid flowmodelthe sediment will make the flow channel blockage and have agreat influence on the drying effect even an accident

Figure 8 shows the entrance of hot air temperature is200∘C velocity of 15ms the variable angle 120579 is 50

∘ and52∘ respectively We get the velocity of flow field in variable

diameter drying pipe through ANSYS FluentFigure 8 shows that when the 120579 is 50

∘ there is refluxarea in variable diameter of variable diameter drying pipeThe reflux area will make sediment in the drying pipe ondrying process When the 120579 is 52∘ there is no reflux area invariable diameter drying pipe In practical design the heightand heat loss of variable diameter drying pipe are considered


1200mm

250mm

300mm

2000mm

150mm

82∘

360mm

938mm

23∘

30∘

200

mm

45∘

60∘

50mm

200mm

(a) The parameters of the drying pipe (b) The photo of variable diameter drying pipe

Figure 9 The diagram of variable diameter drying pipe

Table 3 The design parameters of cut tobacco drying

Parameters Density(kgm3)

Specific heat(kJkgsdot∘C)

Initial velocity(ms)

Initial moisturecontent (kgkg)

Initial temperature(∘C)

The water contentafter drying (kgkg)

Diameter(m)

Cut tobacco 800 1467 0 025 20 0115sim014 12 times 10minus3

Air 075 1025 15 0025 190 mdash mdash

the variable angle 120579 of 60∘ can well meet the requirements ofcut tobacco drying process

423 Production of the Variable Diameter Drying PipeAccording to the above calculation and analysis the variablediameter drying pipe is designed and ismade of stainless steel304 with the thickness of 2mm It is fixed through the bracketto ensure the drying tube vertically The parameters of thedrying pipe and the photo are shown in Figure 9

43 Cut Tobacco Drying Process

431 Simulation of Cut Tobacco Drying Process The distri-bution simulation of airflow and cut tobacco parameters isshown in Table 4 The velocities shown in Figure 10 varywith the height of drying process In the accelerating section(range from 0m to 12m) the airflow velocity decreasesslightly and that of cut tobacco increases quickly withincreasing the height of drying pipe In the transition section(from 12m to 125m) the airflow velocity decreases sharplyand that of cut tobacco increases slowly with increasing the

height of drying pipe But in the constant section (from 125mto 325m) the velocities of the airflow and the cut tobaccodecrease gradually with the height increasing

As shown in Figure 10(b) the temperature of air flowdecreases with the increases of drying pipe height The tem-perature of cut tobacco increases rapidly in the acceleratingsection and then increases slowly with the increases of dryingpipe height This is the cut tobacco is put into the dryingpipe and separated quickly by high-speed airflow Then thecontact area between the airflow and the cut tobacco isincreased indicating that both heat and mass transfer ratesare improved In the accelerating section because the relativespeed of cut tobacco particles is large both heat transfer areasfor gas phase together with solid phase and the volumetricheat transfer coefficient are improved It naturally follows thatcut tobacco temperature increases rapidly in the acceleratingsection But in the constant section the relative speed ofairflow and cut tobacco particles basically keeps unchangedso the speed of cut tobacco particles no longer increasesand the time of cut tobacco in drying pipe is prolong-ed


Table 4 Parameters of cut tobacco drying (cut tobacco)

Intensitykgsdotmminus3 SpecificheatJsdot(kgsdot∘C)minus1

Initial moisturecontent Diameterm Solids-gas

ratioAirflow

temperature∘CAirflow

speedmsminus1Initial temperature

∘C255 1467 2505 12times10-3 01 200 15 20

00 04 08 12 16 20 24 28 32

0

2

4

6

8

10

12

14

16

Cut tobacco velocityAir velocityCut tobacco moisture content

Drying pipe height z (m)

Moi

sture

cont

ent (

db

)

Velo

city

(ms

)12

14

16

18

20

22

24

26

(a)

Cut tobacco temperatureAir temperatureCut tobacco moisture content

00 04 08 12 16 20 24 28 32

12

14

16

18

20

22

24

26

30

60

90

120

150

180

210

Tem

pera

ture

(∘C)

Drying pipe height z (m)M

oistu

re co

nten

t (

db

)

(b)

Figure 10 Distribution simulation of airflow and cut tobacco parameters (simulation experimental conditions inlet air temperature = 200∘Cflow rate = 15ms and initial moisture content of cut tobacco = 25)

A1 B1 C1 C1998400 C2

112

116

120

124

128

132

136

140

144

148

Moi

sture

cont

ent (

)

Run number

MeasuredSimulated

Figure 11 Final cut tobacco moisture contents of simulation andexperiment results

432 Validation of Mathematical Model The mathematicalmodel of the cut tobacco particles drying process was cal-culated using fourth-order Runge-Kutta numerical solutionmethod Experimental conditions [12] and the results of both

simulation and experiment are shown in Table 5 and Figures11 12 and 13 respectively

Figures 11 12 and 13 show the final cut tobacco moisturecontents temperatures and the final airflow temperaturesalong with the simulation results for the same It can be seenfrom it that the simulation results are in good agreement withexperimental data for cut tobacco drying processThe relativeerrors between simulated and experimental values are lessthan 8 indicating that this mathematical model is accuratefor the cut tobacco airflow drying process

44 Optimization Parameters of Cut Tobacco Drying ProcessThe effects of initial cut tobacco moisture content inlet air-flow velocity and temperature on cut tobacco drying processare studied by the orthogonal experiment in three factors andthree levels Criteria for determining the optimum parameterconditions are the final cut tobacco moisture content andthermal efficiency of the dryer The optimization results areshown in Table 6

According to the orthogonal test result it can be seen thatthe thermal efficiency increases with increasing the airflowtemperature and the inlet cut tobacco moisture content butdecreases with the decrease of airflow velocity The resultshows the optimum preparation conditions are an inletairflow velocity of 15ms an initial cut tobacco moisturecontent of 26 and an inlet airflow temperature of 200∘CThe thermal efficiency of the dryer and the final cut tobaccomoisture content are 6632 and 1415 respectively The


Table 5 Summaries of experimental conditions

Run number 119905119892in V

119892in V119892out 119905

119901out 119883in 119883out

A1 1995 15 1634 532 2403 1237B1 1998 15 1569 587 2602 1415C1 2004 15 1577 593 2505 1339C11015840 2004 20 1526 612 2505 1123C2 1903 15 1485 528 2505 1433

Table 6 Result of orthogonal experiment

Run number V119892in (ms) 119883in () 119905

119892in (∘C) 120578 ()

1 15 24 180 62462 15 25 190 63543 15 26 200 66324 18 24 190 59465 18 25 200 61346 18 26 180 60647 20 24 200 54568 20 25 180 52359 20 26 190 5749K1 64107 58827 58483 mdashK2 60480 59077 60163 mdashK3 54800 61483 60740 mdashRange 9307 2656 2257 mdash

MeasuredSimulated

A1 B1 C1 C2Run number

20

25

30

35

40

45

50

55

60

65

70

Tem

pera

ture

(∘C)

C1998400

Figure 12 Final cut tobacco temperatures of simulation andexperiment results

final cut tobacco moisture content meets the requirementsof cut tobacco drying process The result shows that thisparabolic trough concentrating solar heating will be a poten-tial heat recourse for cut tobacco drying system

5 Conclusion

A novel parabolic trough concentrating solar heating forcut tobacco drying system was established The result shows

100

110

120

130

140

150

160

170

180

MeasuredSimulated


Tem

pera

ture

(∘C)

C1998400

Figure 13 Final airflow temperatures of simulation and experimentresults

that the heating rate acquisition factor and collector systemefficiency increase with increasing the opening width of theabsorber and can also prevent the cavity from deformationThe simulation results using the cut tobacco drying math-ematical model were in good agreement with experimentaldata for cut tobacco drying process The optimum prepara-tion conditions were an inlet airflow velocity of 15ms aninitial cut tobacco moisture content of 26 and an inlet


airflow temperature of 200∘C The thermal efficiency of thedryer and the final cut tobaccomoisture content were 6632and 1415 respectivelyThe result showed that this parabolictrough concentrating solar heating would be one of the heatrecourse candidates for cut tobacco drying system

Nomenclature

119860 Drying pipe cross-sectional area m2119862119904 Specific heat capacity of dry material

kJ(kgsdot∘C)119862119908 Specific heat capacity of water kJ(kgsdot∘C)

119862119892 Specific heat capacity of moist airkJ(kgsdot∘C)

119862oil Specific heat capacity of heat transfer oilkJ(kgsdot∘C)

119863 Drying pipe diameter m119889119901 Particle size m

119866119892 Oven dry air flow kgs

119866119901 Oven dry particle size flow kgs

ℎ Convective heat transfer coefficientsW(m2 sdot ∘C)

119870119910 Mass transfer coefficient kg(m2sdots)

119873 The number of particles per unit volume1m3

Nu Nusselt number119875sat Saturated vapor pressure PaPr Prandtl number1199030 0∘C latent heat of vaporization of water

kJ(kgsdot∘C)Re Reynolds number119905119892 Air temperature ∘C

119905119901 Particle temperature ∘C

V119892 Steam velocity ms

V119901 Particle velocity ms

119882119889 Mass transfer rate W(m2 sdot ∘Csdots)

119883 The moisture content of the material db

119884 Absolute humidity of air db119884eq Moisture content of air saturation119911 Drying pipe length m

Greek

120585 Drag coefficient120582119892 Thermal conductivity of air J(msdot

∘C sdots)120588119901 Wet density of the material kgm3

120588119892 Wet density of the air kgm3

120588119904 Oven dry density of the material kgm3

120579 Variable diameter angle120578sys Collector system efficiency120578 Drying pipe thermal efficiency

Subscript

In inletOut outlet



Acknowledgments

The present study was supported by the National NaturalScience Foundation China (Grant no U1137605) and theApplied Basic Research Programs of Science and TechnologyDepartment Foundation of Yunnan Province China (Grantno 2011DFA60460)

References

[1] D A Boyd R Gajewski and R Swift ldquoA cylindrical blackbodysolar energy receiverrdquo Solar Energy vol 18 no 5 pp 395ndash4011976

[2] O A Barra and L Franceschi ldquoThe parabolic trough plantsusing black body receivers experimental and theoretical analy-sesrdquo Solar Energy vol 28 no 2 pp 163ndash171 1982

[3] C Qiaoli G Xinshi C Shuxin et al ldquoNumerical analysis ofthermal characteristics of solar cavity receiver with absorber ofa bundle of pipesrdquo Acta Energiae Solaris Sinica vol 16 no 1 pp21ndash28 1995

[4] L Zhang H Zhai Y Dai and R-Z Wang ldquoThermal analysisof cavity receiver for solar energy heat collectorrdquo Journal ofEngineering Thermophysics vol 29 no 9 pp 1453ndash1457 2008

[5] J Fukuchi Y Ohtaka and C Hanaoka ldquoA numerical fluidsimulation for a pneumatics conveying dryer of cut tobaccordquoin Proceedings of the CORESTA Congress pp 201ndash212 LisbonPortugal 2000

[6] Z Pakowski A Druzdzel and J Drwiega ldquoValidation of amodel of an expanding superheated steam flash dryer for cuttobacco based on processing datardquo Drying Technology vol 22no 1-2 pp 45ndash57 2004

[7] L Huimin and M Ming YCT31-1996 Tobacco and TobaccoProductsmdashPreparation of Test Specimens and Determinationof Moisture ContentmdashOven Method China Standard PressBeijing China 1996

[8] X Chengmu L Ming J Xu et al ldquoFrequency statistics analysisfor energy-flux-density distribution on focal plane of parabolictrough solar concentratorsrdquo Acta Optica Sinica vol 33 no 4Article ID 040800 pp 1ndash7 2013

[9] R Legros C A Millington and R Clift ldquoDrying of tobaccoparticles in a mobilized bedrdquo Drying Technology vol 12 no 3pp 517ndash543 1994

[10] R Blasco R Vega and P I Alvarez ldquoPneumatic drying withsuperheated steam bi-dimensional model for high solid con-centrationrdquo Drying Technology vol 19 no 8 pp 2047ndash20612001

[11] P Yongkang andW ZhongModern Drying Technology Chem-ical Industry Press Beijing China 1998

[12] W E Ranz andWRMarshall ldquoEvaporation fromdrops part IrdquoChemical Engineering Progress vol 48 no 3 pp 141ndash146 1952

Research ArticleVery Fast and Accurate Procedure for the Characterization ofPhotovoltaic Panels from Datasheet Information

Antonino Laudani1 Francesco Riganti Fulginei1 Alessandro Salvini1

Gabriele Maria Lozito1 and Salvatore Coco2

1 Department of Engineering Universita di Roma Tre Via Vito Volterra 62 00146 Roma Italy2 DIEEI Universita di Catania Viale A Doria 6 95125 Catania Italy

Correspondence should be addressed to Francesco Riganti Fulginei rigantiuniroma3it

Received 1 November 2013 Accepted 12 February 2014 Published 24 March 2014


Copyright copy 2014 Antonino Laudani et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

In recent years several numerical methods have been proposed to identify the five-parameter model of photovoltaic panels frommanufacturer datasheets also by introducing simplification or approximation techniques In this paper we present a fast andaccurate procedure for obtaining the parameters of the five-parameter model by starting from its reduced form The procedureallows characterizing in few seconds thousands of photovoltaic panels present on the standard databases It introduces andtakes advantage of further important mathematical considerations without any model simplifications or data approximations Inparticular the five parameters are divided in two groups independent and dependent parameters in order to reduce the dimensionsof the search space The partitioning of the parameters provides a strong advantage in terms of convergence computational costsand execution time of the present approach Validations on thousands of photovoltaic panels are presented that show how it ispossible to make easy and efficient the extraction process of the five parameters without taking care of choosing a specific solveralgorithm but simply by using any deterministic optimizationminimization technique

1 Introduction

The one-diode model for the photovoltaic (PV) panel char-acterization has been widely used within both specific soft-ware toolboxes for the estimation and the prediction of theelectrical power produced by PV plants [1ndash4] and algorithmsfor the Maximum Power Point Tracking [5ndash7] or irradiancemeasurements [8 9] Indeed it guarantees a good trade-off between accuracy and complexity for its setup [10 11]On the other hand the extraction of the five-parametermodel at standard reference conditions (SRC) (ie an inverseproblem) has been widely faced in the literature Althoughtwo approaches are generally the most adopted (the onethat uses the datasheet information and the other one thatexploits the experimental data on I-V curves) the use of onlydata provided by manufacturer on datasheet appears moreinteresting because it does not require a specific experimentalstudy on PV module Nevertheless the approaches proposedthe in literature differ between them and often it is difficult to

understand what is the best one to be used Indeed on onehand several works proposed different equationsapproachesfor the extraction of the five parameters on the otherhand almost any kinds of optimization techniques havebeen presented to solve the inverse problem of the extrac-tion of the five parameters This is essentially due to thenature of the involved equations which are transcendentaland hard to manage Just to give some references withinthe wide literature regarding this issue hereafter some ofthe more recent works are briefly reported Regarding thetechniques for finding the inverse problem solutions in [12]an improved differential evolution algorithm is presentedfor the extraction of five parameters from both syntheticdata and experimental 119868119881 data in [13] penalty differentialevolution is used in a similar way in [14 15] pattern searchand Bacterial Foraging Algorithm are used respectively andso on (see the referencewithin these works for further journalarticles) Regarding the alternative analytical approaches in[16] an explicit 119868-119881 model of a solar cell which uses Pade



G

Iirr

I

V

RS

RSH

minus

+

Figure 1 One-diode equivalent model for a PV module

approximation is presented that is the exponential functionis approximated by means of a rational function in [17] theTaylor series is instead used in [18ndash20] the 119868-119881 relationsare more explicitly written by means of the 119882 Lambertfunction [21] and then the extraction of the five parametersis performed by numerical techniques Although the aim ofthese works is to effectively solve the problem they sufferfrom unsuitable mathematical approximations (which leadto errors in the results) or complicated implementations andhigh computational costs As a consequence these approachesare not so easy to be applied

In this paper we present a fast and accurate procedurefor obtaining the parameters of the five-parameter modelby starting from its reduced form [22] which allows theidentification of thousands of PV panels available on thestandard databases (such as Californian Energy Commissiondatabase [23]) Indeed by using suitable initial guesses it ispossible to fully characterize thousands of PV panel in fewseconds without any model simplifications or data approx-imations simply by introducing further important mathe-matical considerations about the five-parameter model Thepaper is structured as follows in Section 2 the traditional one-diode model and the problem related to its characterizationare presented in Section 3 the reduced form of the five-parameter model is described the validation results obtainedon thousands of PV panels are shown in Section 4 finallySection 5 is for the conclusions

2 The One-Diode Model andIts Reduced Form

The equivalent circuit of the one-diode model is shownin Figure 1 The relation between current 119868 and voltage 119881for a PV arraypanel of arbitrary dimension (119873

119875parallel

connected strings of 119873119878series-connected PV cells) at the

equivalent port is [10]

119868 = 119873119875119868irr minus 1198731198751198680 [119890

119902(119881+119868(119873119878119873119875)119877119878)119873119878119899119896119879

minus 1]

minus

119881 + 119868 (119873119878119873119875) 119877119878

(119873119878119873119875) 119877SH

(1)

where 119868irr is the irradiance current (photocurrent) 1198680 is thecell reverse saturation current (diode saturation current) 119902is the electron charge (119902 = 1602 times 10

minus19 C) 119899 is the cellideality factor 119896 is the Boltzmann constant (119896 = 13806503 times10minus23 JK)119879 is the cell temperature and119877

119878and119877SH represent

the cell series and shunt resistance respectively In (1) thegoverning variables 119899 119877

119878 119868irr 1198680 and 119877SH can be assumed as

dependent or not on the irradiance (119866) and the temperature(119879) but in any case they are in function of certain reference(ref) parameters at SRC (119866ref = 1000Wm2 119879ref = 25

∘C)(hereafter we present and utilize the relations proposed byDe Soto et al in [11] other slightly different relations arepresented in several works such as [24 25] but their usedoes not affect the effectiveness and validity of the presentedprocedure)

119899 = 119899ref (2)

119877119878= 119877119878ref (3)

119868irr =119866

119866ref[119868irrref + 120572119879 (119879 minus 119879ref)] (4)

1198680= 1198680ref[

119879

119879ref]

3

119890[119864119892ref119896119879refminus119864119892119896119879]

(5)

119877SH =

119866

119866ref119877SHref (6)

In (5) 119864119892

= 117 minus 473 times 10minus4

times (1198792(119879 + 636)) is

the bandgap energy for silicon in eV Thus there are fiveunknown parameters at SRC 119899ref 119877119878ref 119868irrref 1198680ref and119877SHref to be foundwithin (2)ndash(6)Then by starting from theirvalues and by using the above relations it is possible to writethe 119868-119881 curves for every temperature and irradiance valuesIn order to determine these five unknowns we need five inde-pendent equations based on datasheet information Usuallythe PV panel manufacturers provide several information ondatasheet at standard reference conditions (SRC) that is forthe irradiance 119866ref and the temperature 119879ref the values ofthe short-circuit current (119868SCref) and the open-circuit voltage(119881OCref) the current and voltage values at the maximumpower point (119868mpref and 119881mpref) In addition the datasheetsreport the temperature coefficients (or percentage) of boththe short-circuit current (120572

119879or 120572119879) and the open-circuit

voltage (120573119879or 120573119879) On the basis of the three characteristic

points open-circuit short-circuit and maximum powerpoints at SRC it is possible to write the first four equationsof the five-parameter model [10 11 15 22 26] indeed thefirst equation arises by writing (1) for the open-circuit (OC)condition the second equation arises by using (1) for theshort-circuit (SC) condition the third equation arises byexploiting the current and voltage values at the maximumpower point (MPP) conditionThe fourth equation is writtenby imposing the slope of the 119875-119881 curve (power versusvoltage) over the MPP equal to zero 119889(119881 sdot 119868)119889119881 = 0that can be also expressed in terms of 119868mpref119881mpref ratioThe last fifth equation used to complete the five-parametermodel is written by exploiting the temperature dependenceof (1) at the open-circuit condition and irradiance 119866 =

119866ref by using the previously stated temperature coefficients(120572119879

and 120573119879) [10ndash12] Before showing the five equations

it is useful to briefly recall the constants specified in (7)adopted in order to simplify the writing of the mathematicalexpressions Furthermore the temperature-dependent factor


0

0002

0004

0006

0008

001

1

2

0

1

minus1

minus2

minus3

minus4

minus5

minus6

05

15nref

R Sre

f

log10[f1(nref RSref )

2+ f2(nref RSref )

2]

(a)0

0001

0002

0003

0004

0005

0006

0007

0008

0009

001

1

2

05

15

nre

fRSref

Portion of the feasible domain and solution found in it

(b)

Figure 2 Graph of the functional (25) and the feasible domain for the BP 3 235 T PV panel (multi-Si technology) A unique physical solutionexists

119870119879= [119879119879ref]

3119890[119864119892ref119896119879refminus119864119892119896119879] is used in (5) and the shunt

conductance 119866SHref = 119877minus1

SHref is adopted as unknown in (6)instead of 119877SHref

1198621=

119896119879ref119902

1198622=

119881OCref

1198731198781198621

1198623=

119881mpref

1198731198781198621

1198624=

119868mpref

1198731198751198621

1198625=

119868SCref

1198731198751198621

(7)

Thus the five equations are the following

0 = 119868irrref minus 1198680ref (1198901198622119899ref

minus 1) minus 119866SHref11986211198622 (8)

11986211198625= 119868irrref minus 1198680ref (119890

1198625119877119878ref119899ref

minus 1) minus 11986211198625119866SHref119877119878ref

(9)

11986211198624= 119868irrref minus 1198680ref (119890

(1198623+1198624119877119878ref)119899ref

minus 1)

minus 119866SHref1198621 (1198623 + 1198624119877119878ref) (10)

1198624

1198623

=

(1198680ref119899ref1198621) 119890

(1198623+1198624119877119878ref)119899119903119890119891

+ 119866SHref

1 + (1198680ref119877119878ref119899ref1198621) 119890

(1198623+1198624119877119878ref)119899ref + 119866SHref119877119878ref

(11)

0 = 119868irrref + 120572119879 (119879 minus 119879ref)

minus 1198680ref119870119879 (119890

119902(11987311987811986211198622+120573119879(119879minus119879ref))119873S119899ref119896119879

minus 1)

minus 119866SHref11987311987811986211198622+ 120573119879(119879 minus 119879ref)

119873119878

(12)

In (12) a value of119879 = 119879ref plusmn10K is used even if variationsof temperature Δ119879 belonging to the range [minus10 +10] withrespect to 119879ref return very similar solutions [10 11]

The five-parameter model is thus defined by a system offive equations (8)ndash(12) with the five unknowns (parameters)119899ref 119877119878ref 119868irrref 1198680ref and 119866SHref Due to the presence oftranscendental equations this problem is not so simple tomanage and it can be only solved by means of numericalmethods Since it is practically an inverse problem manyminimization algorithms can be used and almost any kinds ofcomputing techniques have been tested in the literature forexample in [27] the comparison between several techniquesto extract the five parameters is presented and comparedby using the criteria of applicability convergence stabilitycalculation speed and error on various types of 119868119881 dataIn addition due to its nonlinear nature the system returnssolutions that are very sensitive to the choice of the initialguesses [22 26 27] As the following section shows thisproblem can be easily overwhelmed by using a reducedform of the model employing only a set of two equations intwo unknowns For the readerrsquos convenience the list of thetechnical parameters used in this work is reported at the endof the paper

3 Reduction to a Two-Parameter Model

In [22] it has been proven that the five-parameter modelcan be reduced to a two-parameter model improving theefficiency of the algorithm finding the solution By usingthis reduced form of the system instead of the original oneit is also possible to demonstrate (i) the uniqueness of


0

0005

001

0015

002

1

2

0

1

minus1

minus2

minus3

minus4

minus5

05

15nref

R Sre

f


2+ f2(nref RSref )

2]

(a)0

0002

0004

0006

0008

001

0012

0014

0016

0018

1

2

05

15

nre

fRSref


(b)

Figure 3 Graph of the functional (25) and feasible domain for the Sharp NT175 panel (mono-Si technology) A unique physical solutionexists

the solution for the problem (ii) the existence of a uniquesolution without physical meaning for some PV panels (iii)the matter of the optimal choice of the initial guesses thatmake easy and effective the solution of the inverse problemAs first thing let us show theway to reduce the five-parametermodel to a two-parameter model This is obtained by simplealgebraicmanipulations of three of the five equations (8)ndash(11)Indeed from the first equation (8) it is possible to obtain 119868irrrefas a function of 119899ref 119877119878ref and 119877SHref as follows

119868irrref = 1198680 (1198901198622119899ref

minus 1) + 119866SHref11986211198622 (13)

By substituting (13) in (10) it is possible to write

11986211198624= 1198680ref (119890

1198622119899ref

minus 1) minus 1198680ref (119890

(1198623+1198624119877119878ref)119899ref

minus 1)

minus 119866SHref1198621 (1198623 minus 1198622 + 1198624119877119878ref)(14)

that can be written also as

1198680ref =

11986211198624+ 119866SHref1198621 (1198623 minus 1198622 + 1198624119877119878ref)

(1198901198622119899ref minus 119890(1198623+1198624119877119878ref)119899ref)

(15)

On the other hand from (11) we can also obtain 1198680ref

1198680ref =

1198623119866SHref minus 1198624 minus 1198624119866SHref119877119878ref

119890(1198623+1198624119877119878ref)119899ref ((119862

4119877119878ref minus 1198623) 119899ref1198621)

(16)

Furthermore by posing the expression (15) equal to the(16) we can write

11986211198624+ 119866SHref1198621 (1198623 minus 1198622 + 1198624119877119878ref)

(1198901198622119899ref minus 119890(1198623+1198624119877119878ref)119899ref)

=


119890(1198623+1198624119877119878ref)119899ref ((119862

4119877119878ref minus 1198623) 119899ref1198621)

(17)

from which it is possible to obtain 119866SHref as a function of 119899refand 119877

119878ref

119866SHref

=

1198624

1198624119877119878ref minus 1198623

sdot

(1 + (1198623minus 1198624119877119878ref) 119899ref) 119890

(1198623+1198624119877119878refminus1198622)119899ref

minus 1

1 + ((1198624119877119878ref + 1198623 minus 1198622) 119899ref minus 1) 119890

(1198623+1198624119877119878refminus1198622)119899ref

(18)

Now substituting (18) in (15) or (16) also allows express-ing 1198680ref as a function of 119899ref and 119877119878ref

1198680ref

=

11986211198624

1198623minus 1198624119877119878ref

sdot

(21198623minus 1198622)

1198901198622119899ref + ((119862

4119877119878ref + 1198623 minus 1198622) 119899ref minus 1) 119890

(1198623+1198624119877119878ref)119899ref

(19)


0

002

004

006

008

1

23

4

501

0

0 04

0

1

minus1

minus2

minus3

minus4

minus5

nref

R Sref


2+ f2(nref RSref )

2]

(a)

0 001 002 003 004 005 006 007 008 009

1

2

3

4

5

15

25

35

45

nre

f

RSref


(b)

Figure 4 Graph of the functional (25) and feasible domain for the PV panel Xunlight XR12ndash88 panel (thin film technology) A uniquephysical solution exists

4

5

6

7

8

9

10

0951

105

115

12512

11

0

1

minus1

minus2

minus3

minus4

minus5

minus6

nref

R Sref


2+ f2(nref RSref )

2]

times10minus3

(a)

12

11

4 45 55 655 6 7

095

1

105

115

125

nre

f

RSref

Close-up of the feasible domain and solution found outside it

times10minus3

(b)

Figure 5 Graph of the functional (25) and feasible domain for the PV panel BP Q 230 panel (multi-Si technology) A unique unphysicalsolution exists (it lies outside the feasible domain) that is this panel cannot be modelled by using the five-parameter model


Finally (19) and (18) are utilized together with (13) so that119868irrref can be written as a function of 119899ref and 119877119878ref

119868irrref =119862111986241198622

(1198623minus 1198624119877119878ref)

+

(11986211198624(21198623minus 1198622)) (119862

3minus 1198624119877119878ref) (119890

1198622119899ref

minus 1 minus (1198622119899ref) 119890

(1198623+1198624119877119878ref)119899ref

)

1198901198622119899ref + ((119862

4119877119878ref + 1198623 minus 1198622119899ref) minus 1) 119890

(1198623+1198624119877119878ref)119899ref

(20)

In (18) (19) and (20) we have written 119866SHref 1198680ref and 119868irrrefas functions of 119899ref and 119877

119878ref respectively This means thatnow there are only two independent unknowns called 119899refand 119877

119878ref to be found by using the two equations coming outfrom the other two conditions not yet utilized in the previoussteps They are (9) and (12) obtained from the evaluation of(1) at short-circuit (119881 = 0 119868 = 119868SCref) condition and for120573119879asymp (119881OC minus 119881OCref)(119879 minus 119879ref) condition respectively Thus

the reduced form of the original five-equation system is thefollowing

1198623(21198624minus 1198625)

1198623minus 119877119878ref1198624

+

1198624(1198622minus 21198623)

1198623minus 119877119878ref1198624

119890((1198625119877119878refminus1198622)119899ref)

+ [

11986251198623minus 11986241198622

1198623minus 119877119878ref1198624

minus

11986221198624+ 11986251198623minus 11986251198622

119899ref]

times 119890((1198623+119877119878ref1198624minus1198622)119899ref)

= 0

[120572119879(119879 minus 119879ref) minus

120573119879(119879 minus 119879ref) 1198624

119873119878(1198623minus 119877119878ref1198624)

+

11986211198624(21198623minus 1198622)

(1198623minus 119877119878ref1198624)

]

+ [120572119879(119879 minus 119879ref) (

1198623+ 119877119878ref1198624 minus 1198622

119899refminus 1)]


+ [

1198624120573119879(119879 minus 119879ref)

119873119878

(119899ref minus 119877119878ref1198624 + 1198623)

119899ref (1198623 minus 119877119878ref1198624)]


+

11986211198624119870119879(1198622minus 21198623)

(1198623minus 119877119878ref1198624)

times 119890(119902120573119879(119879minus119879ref)119873119878119896119879119899ref+(1198622119899ref)(119879ref119879minus1))

+

11986211198624(119870119879minus 1) (2119862

3minus 1198622)

(1198623minus 119877119878ref1198624)

119890(minus1198622119899ref)

= 0

(21)

31 Solution of the Reduced form of the Five-Parameter ModelAlthough the two equations (21) of the reduced form of thefive-parameter model are transcendental equations they arequite affordable that simple and fast numerical methods canbe utilized to find the solutions instead of more complex andexpensive algorithms in terms of computational costs [28ndash31] On the other hand the effectiveness of the reduced formwith respect to the original system based on five equations

is evident since it returns the same unique solution alsoby using different numerical methods Moreover (18)ndash(21)allows making several important considerations about thesolutions of the five-parameter model

(i) Since the values of 119868irrref 1198680ref and 119877SHref must bepositive in order to obtain a physical meaning forthese three parameters it is possible to find the condi-tions for the range of the independent unknowns 119899refand 119877

119878ref from (18) (19) and (20) In particular themaximum admissible value for 119877

119878ref is a function of119899ref according to the following relations [22]

119877119878ref =

(1198622minus 1198623)

1198624

0 lt 119877119878ref lt 119877

max119878ref (119899ref)

(22)

with 119877max119878ref as a function of 119899ref

119877max119878ref (119899ref) =

119899ref1198624

[1 +119882minus1(minus119890(1198622minus119899refminus21198623)119899ref

)] +

1198623

1198624

(23)

whereas the Lambert 119882 function [8] in (23) hasbeen used this special mathematical function allowsobtaining a closed form representation for the 119868-119881curves and it is often successfully used for the analysisof PV modules [14 15]

(ii) Thus it is possible to individuate the feasible domainfor the two remaining independent unknowns byassuming

05 ⩽ 119899ref ⩽ 25 (24)

(iii) It is also possible to graph the 2D functional used forsolving the system (21)

119865 (119899ref 119877119878ref) = 1198911(119899ref 119877119878ref)2

+ 1198912(119899ref 119877119878ref)

2

(25)

where 1198911(119899ref 119877119878ref) = 0 and 119891

2(119899ref 119877119878ref) = 0

represent the first and the second equation of thesystem (21) respectively The graph of the functional119865(119899ref 119877119878ref) is smooth and free from local minima(some examples are shown in the next section)The presence of only one minimum (ie one globalminimum)makes the problem of finding the solutionof the system (21) a convex problem allowing theuse of simple initial guesses without choosing specificoptimization algorithms Indeed as also discussedin [26] where an empirical approach is adopted


0 5 10 15 20 25 30 35 40

0

1

2

3

4

5

6

7

8

9

Voltage

Curr

ent

3 par model4 par model5 par model

Current-voltage curves at SRC

Figure 6 Current-Voltage curves at SRC of BP 3 235T traced byusing 345 parameter models

the choice of the initial guesses is one of the morecritical aspects regarding the identification of the five-parameter model [22]

(iv) As a consequence it is also possible to state thatthe system (21) that is the reduced form of theoriginal five-equation system has a unique solutionthat corresponds to the one with physical meaning

(v) By observing the graph of the functional 119865(119899ref 119877Sref)it is also possible to verify that some PV panelshave the minimum (ie the solution of the problem)lying outside the feasible domainThis means that thesolution still exists but it is not physical (ie at leastone among the five parameters is negative)

32 Some Examples and Graphs In order to prove the aboveconsiderations in Figures 2 3 4 and 5 the results of fourdifferent panel modules are shown a mono-Si PV panel(Sharp NT-175UC1) a multi-Si PV panel (BP 3235 T) a thinfilm PV panel (Xunlight XR12ndash88) and another multi-Si PVpanel (BP Q Series 230 W) Each figure shows the graph ofthe functional 119865(119899ref 119877Sref) together with its feasible domain(ie the set of points of the two independent parameters 119899refand119877

119878ref for which the dependent parameters 1198680ref 119868irrref and

119866SHref have physical meaning) and the solution (minimumof the functional) of the reduced system (21) Figures 2ndash5 clearly prove the uniqueness of the solutions of the fourpanels Furthermore the quasi-monotonic behaviours of thefunctionals allow for an easy search of the solution (convexoptimization) The initial guesses chosen for the searchprocedure of the solutions were the following

119877guess119878

= 09 sdot 119877max119878ref (119899

guess) (26)

18 20 22 24 26 28 30 32 34 36

6

65

55

75

85

7

8

Voltage

Curr

ent

MPP3 par model4 par model

5 par model


Figure 7 Close-up around maximum power point of Current-Voltage curves at SRC of BP 3 235T traced by using 345 parametermodels

0 5 10 15 20 25 30 35 400

50

100

150

200

250

Pow

er

Voltage

Power-voltage curves at SRC


Figure 8 Power-Voltage curves at SRC of BP 3 235T traced by using345 parameter models

with 119899guess

= 10 for multi-Si and mono-Si PV panels and119899guess

= 20 for thin film PV panels It can be noted by observ-ing Figure 5 that the PV panel BP Q Series 230 W does notprovide for physical solutions of the five parameters model(ie the solution exists but it lies outside the feasible domainand then at least one among the dependent parameters islower than zero) It is worth noting that all the PV panels ofBP Q series cannot be modelled by using the five-parametermodel The issue about the existence or not of the solution


24 26 28 30 32 34

190

200

210

220

230

240

Voltage

Pow

er



5 par model

Figure 9 Close-up aroundmaximumpower point of Power-Voltagecurves at SRC of BP 3 235T traced by using 345 parameter models

is really complex and nothing can be said a priori by simplyobserving the datasheets of the PV panels

4 Tests on California EnergyCommission Database

In order to prove the effectiveness of the proposed procedureaimed at the identification of the five-parameter model sim-ply by starting from PV datasheets in this section a statisticalvalidation is presented In particular the tests have involvedaround 11000 PV panels belonging to the California EnergyCommission (CEC) database [23] (updatedmonthly) Table 1shows the number of tested PV panels ( PV panels) fromCEC database grouped by type of technology With theaim to demonstrate both the robustness and the fastnessof the proposed approach several initial guesses have beenchosen and three different numerical algorithmsfunctionshave been used in Matlab fsolve (suitable for systems withnonlinear equations) fminsearch (generic unconstrainedminimization function) and lsqnonlin (function aimed tosolve least squares problems) The simulations showed veryaccurate results (ie with functional values less than 1119864minus20)which are independent of the adopted algorithm and theunique solutions have been found at first launch for almostall panels The execution time for the extraction of the fiveparameters of all the 11764 PV shown in Table 1 performedon an Intel i5 core 25 GHz based notebook with 4 GB ofRAMwas around 90 seconds for the most efficient algorithm(fsolve with trust-region-dogleg) and 400 seconds for theslowest algorithm (fsolve with trust-region-reflective) Thismeans that also for the worst cases the herein proposedextraction procedure of the five parameters spent less than30msec for each panel

Table 1 Number of tested PV panels from CEC database groupedby type of technology

Technology PV panelsMono-Si 4801Multi-Si 6435Amorphous and thin films 253Other (CIS CIGS CdTe etc) 275Total 11764

Table 2 Results obtained with Matlab fsolve for 119877guess119878

=

05119877max119878ref(119899

guess)

Algorithm Trust-region-dogleg

Trust-region-reflective

Levenberg-marquardt

Mean steps 1556 1755 1752Std steps 948 591 989Mean FEs 4116 5565 8084Std FEs 2638 1785 3457


=

09119877max119878ref(119899

guess)



Levenberg-marquardt


Table 4 Results obtained with different Matlab functions with119877guess119878

= 09119877max119878ref(119899

guess)

Matlabfunction fminsearch lsqnonlin lsqnonlin

Algorithm Nelder-Meadsimplex method


Levenberg-marquardt


The comparisons of the performance achievable by usingvarious algorithms and initial guesses are reported in Tables2 3 and 4 in particular the results are expressed in terms ofaverage number (mean steps) and standard deviation (std steps) of iterations steps average number (mean FEs) andstandard deviation (std FEs) of function evaluations (FEs)(in Table 2 the initial guesses are the ones proposed in [22])

It is worth noting that the proposed initial guess (26)allows obtaining effective results also by using one of themost generic solvers for minimization problems fminsearchwhich employs the Nelder-Mead simplex method discussedin Lagarias et al [32] By using instead more effective Matlabfunctions (such as fsolve or lsqnonlin) and algorithms (such


as Levenberg-Marquardt [33] or trust-region-dogleg [34] algo-rithms) the number of iterations and FEs becomes extremelylow (around 7 for the average number of iterations and 45for the one of FEs) Consequently the computational costsof the proposed procedure is quite negligible as the varioussoft computing based approaches like the ones in [12ndash15]typically require thousands of FEs In addition it is worthnoting that the obtained results have physical meaning formore than 97 of the total number of panels (11488 on11764 PV panels) The unphysical solutions could be dueto the impossibility of identifying the five-parameter modelas was for the previous case of BP Q series PV panelsNevertheless since we do not have any information abouthow the datasheets are loaded into the CEC database weassumed they were correct and no check was made about theexactness of data Thus some unphysical solutions could bealso due to this lastmatter and caused by the presence of someerrors within the CEC database Finally with the aim to showthe importance of adopting an accurate 5-parameter modelrather than approximated ones (such as for example the 3-parameter and 4-parameter models [35]) the 119868-119881 and 119875-119881curves at SRC for BP 3 235T module have been consideredas last test The 3-parameter and 4-parameter models seemto be very similar to the 5-parameter one Nevertheless withthe aim to simplify the characterization problem 119877

119878ref = 0

and 119877SHref = inf conditions are used for the 3-parametermodel and 119877SHref = inf condition is used for the 4-parametermodel The 119868-119881 and 119875-119881 curves for the three models areshown in Figures 6 and 8 whereas the close-up around themaximum power point (MPP) is shown in Figures 7 and 9It is evident by observing the 119875-119881 curves that (1) the 5-parameter and 4-parameter models are both accurate in theevaluation of MPP whereas the 3-parameter model is not(2) the 4-parameter model overestimates the power aroundthe MPP causing possible errors in the prediction of electricpower produced by a PV plant

5 Conclusions

In this paper a fast and accurate procedure has been presentedfor the characterization of thousands of photovoltaicmodulesin few seconds by starting from themanufacturer datasheetsThe proposed procedure utilizes the five-parameter modeland takes advantage from its reduced form [22] in orderto decrease the dimensions of the search space Indeedthe reduced system provides a strong advantage in termsof convergence computational costs and execution time ofthe present approach (less than 30 msec for each panel wasspent on a simple Intel i5 core 25 GHz based notebook)In particular it allows (1) choosing suitable initial guesseswithin a well-defined feasible domain (2) using very simpleand standard numerical algorithms for finding parameters(3) proving the existence or not of the unique physicalsolution of the five-parameter model for each PV panel(4) proving the existence of only unphysical solutions forthe cases in which the five-parameter model cannot beidentified The results of the tests performed on around11000 photovoltaic modules belonging to the CEC database

demonstrated both the fastness and the effectiveness of theproposed method

Technical Parameters

119902 1602 times 10minus19 (C)

119896 13806503 times 10minus23 (JK)

119864119892 Bandgap energy

119866 Irradiance119879 Cell temperature1198680 Reverse saturation current

119868irr Photocurrent119899 Ideality factor119877119878 Series resistance

119877SH Shunt resistance119873119904 Number of series modulescells

119873119901 Number of parallel connected strings

119879ref 25∘C at SRC119866ref 1000 Wm2 at SRC119899ref Ideality factor at SRC119877119878ref Series resistance at SRC

119868irrref Photocurrent at SRC1198680ref Reverse saturation current at SRC119866SHref = 119877

minus1

119878119867ref Shunt resistance at SRC119881OC Open circuit voltage119868SC Short circuit current119881mp Maximum power voltage119868mp Maximum power current119881OCref Open circuit voltage at SRC119868SCref Short circuit current at SRC119881mpref Maximum power voltage at SRC119868mpref Maximum power current at SRC120572119879 Temperature coeff for 119868SC

120572119879 Percentage temperature coeff for 119868SC

120573119879 Temperature coeff for 119881OC

120573119879 Percentage temperature coeff for 119881OC

1198621 119896119879ref119902

1198622 119881OCref1198731198781198621

1198623 119881mpref1198731198781198621

1198624 119868mpref1198731198751198621

1198625 119868SCref1198731198751198621



References

[1] A D Rajapakse and D Muthumuni ldquoSimulation tools forphotovoltaic system grid integration studiesrdquo in Proceedings ofthe IEEE Electrical Power and Energy Conferenc (EPEC rsquo09) pp1ndash5 October 2009

[2] D Menicucci and J Fernandez Userrsquos Manual for PVFORMSAND85-0376 A Photovoltaic System Simulation Program ForStand-Alone and Grid-Interactive Applications Sandia NationalLaboratories Albuquerque NM USA 1988

[3] ldquoPVWATTSrdquo 2011 httpwwwnrelgovrredcpvwatts


[4] N J Blair A P Dobos and P Gilman ldquoComparison of photo-voltaic models in the system advisor modelrdquo in Proceedings ofthe Solar Baltimore Md USA April 2013

[5] J Ma K Man T Ting N Zhang S U Guan and P W WongldquoDem direct estimation method for photovoltaic maximumpower point trackingrdquo Procedia Computer Science vol 17 pp537ndash544 2013 1st International Conference on InformationTechnology and Quantitative Management

[6] I T Papaioannou and A Purvins ldquoMathematical and graph-ical approach for maximum power point modellingrdquo AppliedEnergy vol 91 no 1 pp 59ndash66 2012

[7] M Carrasco F Mancilla-David F R Fulginei A Laudaniand A Salvini ldquoA neural networks-based maximum powerpoint tracker with improved dynamics for variable dc-link grid-connected photovoltaic power plantsrdquo International Journal ofApplied Electromagnetics and Mechanics vol 43 no 1 pp 127ndash135 2013

[8] B Schulz T Glotzbach C Vodermayer GWotruba MMayerand S Grunsteidl ldquoEvaluation of calibrated solar cells andpyranometers regarding the effective irradiance detected bypv modulesrdquo in Proceedings of the 25th European PhotovoltaicSolar Energy Conference and Exhibition5th World Conferenceon Photovoltaic Energy Conversion pp 4797ndash4800 October2012

[9] F Mancilla-David F Riganti-Fulginei A Laudani and ASalvini ldquoA neural network-based low-cost solar irradiance sen-sorrdquo IEEE Transactions on Instrumentation and Measurementvol 63 no 3 pp 583ndash591 2014

[10] H Tian F Mancilla-David K Ellis E Muljadi and P JenkinsldquoA cell-to-module-to-array detailed model for photovoltaicpanelsrdquo Solar Energy vol 86 no 9 pp 2695ndash2706 2012

[11] W de Soto S A Klein andW A Beckman ldquoImprovement andvalidation of amodel for photovoltaic array performancerdquo SolarEnergy vol 80 no 1 pp 78ndash88 2006

[12] L L Jiang D L Maskell and J C Patra ldquoParameter estimationof solar cells and modules using an improved adaptive differen-tial evolution algorithmrdquo Applied Energy vol 112 pp 185ndash1932013

[13] K Ishaque Z Salam S Mekhilef and A Shamsudin ldquoParam-eter extraction of solar photovoltaic modules using penalty-based differential evolutionrdquo Applied Energy vol 99 pp 297ndash308 2012

[14] M F AlHajri K M El-Naggar M R AlRashidi and A KAl-Othman ldquoOptimal extraction of solar cell parameters usingpattern searchrdquo Renewable Energy vol 44 pp 238ndash245 2012

[15] N Rajasekar N K Kumar and R Venugopalan ldquoBacterialforaging algorithm based solar PV parameter estimationrdquoSolar Energy vol 97 pp 255ndash265 2013

[16] S Xian Lun C Jiao Du G Hong Yang et al ldquoAn explicitapproximate iv characteristic model of a solar cell based on padapproximantsrdquo Solar Energy vol 92 pp 147ndash159 2013

[17] S Xian Lun C Jiao Du T Ting Guo S Wang J Shu Sang andJ Pei Li ldquoA new explicit iv model of a solar cell based on taylorsseries expansionrdquo Solar Energy vol 94 pp 221ndash232 2013

[18] A Jain and A Kapoor ldquoExact analytical solutions of theparameters of real solar cells using Lambert W-functionrdquo SolarEnergy Materials and Solar Cells vol 81 no 2 pp 269ndash2772004

[19] A Jain and A Kapoor ldquoA new method to determine the diodeideality factor of real solar cell using Lambert W-functionrdquoSolar Energy Materials and Solar Cells vol 85 no 3 pp 391ndash396 2005

[20] Y Chen X Wang D Li R Hong and H Shen ldquoParametersextraction from commercial solar cells I-V characteristics andshunt analysisrdquo Applied Energy vol 88 no 6 pp 2239ndash22442011

[21] R M Corless G H Gonnet D E G Hare D J Jeffreyand D E Knuth ldquoOn the Lambert W functionrdquo Advances inComputational Mathematics vol 5 no 4 pp 329ndash359 1996

[22] A Laudani F Mancilla-David F Riganti-Fulginei and ASalvini ldquoReduced-form of the photovoltaic five-parametermodel for efficient computation of parametersrdquo Solar Energyvol 97 pp 122ndash127 2013

[23] California Energy Commission ldquoCECPV calculator version40rdquo 2013 httpwwwgosolarcaliforniaorgtoolsnshpcalcula-torindexphp

[24] V lo Brano A Orioli G Ciulla and A di Gangi ldquoAn improvedfive-parameter model for photovoltaic modulesrdquo Solar EnergyMaterials and Solar Cells vol 94 no 8 pp 1358ndash1370 2010

[25] A Mermoud and T Lejeune ldquoPerformance assessment of asimulation model for PV modules of any available technologyrdquoin Proceedings of the 25th European PV Solar Energy Conferencepp 4786ndash4791 2010

[26] A P Dobos ldquoAn improved coefficient calculator for the cal-ifornia energy commission 6 parameter photovoltaic modulemodelrdquo Journal of Solar Energy Engineering Transactions of theASME vol 134 no 2 Article ID 021011 pp 1ndash6 2012

[27] Y Li W Huang H Huang et al ldquoEvaluation of methods toextract parameters from currentvoltage characteristics of solarcellsrdquo Solar Energy vol 90 pp 51ndash57 2013

[28] J Fourie R Green and Z W Geem ldquoGeneralised adaptiveharmony search a comparative analysis of modern harmonysearchrdquo Journal of Applied Mathematics vol 2013 Article ID380985 13 pages 2013

[29] A Laudani F Riganti-Fulginei and A Salvini ldquoClosed formsfor the fully-connected continuous flock-of-starlings optimiza-tion algorithmrdquo in Proceedings of the 15th International Confer-ence on Computer Modeling and Simulation (UKSim rsquo13) April2013

[30] A Laudani F Riganti-Fulginei A Salvini M Schmid and SConforto ldquoCFSO3 a new supervised swarm-based optimiza-tion algorithmrdquo Mathematical Problems in Engineering vol2013 Article ID 560614 13 pages 2013

[31] F R Fulginei A Salvini and G Pulcini ldquoMetric-topological-evolutionary optimizationrdquo Inverse Problems in Science andEngineering vol 20 no 1 pp 41ndash58 2012

[32] J C Lagarias J A Reeds M H Wright and P E WrightldquoConvergence properties of the Nelder-Mead simplex methodin low dimensionsrdquo SIAM Journal on Optimization vol 9 no 1pp 112ndash147 1998

[33] J J More ldquoThe levenberg-marquardt algorithm implementa-tion and theoryrdquo in Numerical Analysis pp 105ndash116 SpringerNew York NY USA 1978

[34] T F Coleman and Y Li ldquoAn interior trust region approach fornonlinear minimization subject to boundsrdquo SIAM Journal onOptimization vol 6 no 2 pp 418ndash445 1996

[35] A Luque and S HegedusHandbook of Photovoltaic Science andEngineering John Wiley amp Sons New York NY USA 2011

Research ArticleAn Improved Mathematical Model for ComputingPower Output of Solar Photovoltaic Modules

Abdul Qayoom Jakhrani1 Saleem Raza Samo1 Shakeel Ahmed Kamboh2

Jane Labadin3 and Andrew Ragai Henry Rigit4

1 Energy and Environment Engineering Department Quaid-e-Awam University of EngineeringScience and Technology (QUEST) Nawabshah Sindh 67480 Pakistan

2Department of Mathematics and Computational Science Faculty of Computer Science and Information TechnologyUniversiti Malaysia Sarawak Kota Samarahan 94300 Sarawak Malaysia

3 Faculty of Computer Science and Information Technology Universiti Malaysia Sarawak Kota Samarahan94300 Sarawak Malaysia

4 Faculty of Engineering Universiti Malaysia Sarawak Kota Samarahan 94300 Sarawak Malaysia

Correspondence should be addressed to Abdul Qayoom Jakhrani aqunimashotmailcom

Received 31 May 2013 Revised 26 December 2013 Accepted 9 January 2014 Published 17 March 2014


Copyright copy 2014 Abdul Qayoom Jakhrani et alThis is an open access article distributed under theCreative CommonsAttributionLicense which permits unrestricted use distribution and reproduction in anymedium provided the originalwork is properly cited

It is difficult to determine the input parameters values for equivalent circuit models of photovoltaic modules through analyticalmethods Thus the previous researchers preferred to use numerical methods Since the numerical methods are time consumingand need long term time series data which is not available in most developing countries an improved mathematical model wasformulated by combination of analytical and numerical methods to overcome the limitations of existing methods The valuesof required model input parameters were computed analytically The expression for output current of photovoltaic module wasdetermined explicitly by Lambert W function and voltage was determined numerically by Newton-Raphson method Moreoverthe algebraic equations were derived for the shape factor which involves the ideality factor and the series resistance of a single diodephotovoltaic module power output model The formulated model results were validated with rated power output of a photovoltaicmodule provided by manufacturers using local meteorological data which gave plusmn2 error It was found that the proposed modelis more practical in terms of precise estimations of photovoltaic module power output for any required location and number ofvariables used

1 Introduction

The photovoltaic (PV) modules are generally rated understandard test conditions (STC) with the solar radiation of1000Wm2 cell temperature of 25∘C and solar spectrumof 15 by the manufacturers The parameters required forthe input of the PV modules are relying on the meteoro-logical conditions of the area The climatic conditions areunpredictable due to the random nature of their occurrenceThese uncertainties lead to either over- or underestimationof energy yield from PV modules An overestimation up to40 was reported as compared to the rated power outputof PV modules [1 2] The growing demand of photovoltaicstechnologies led to research in the various aspects of itscomponents from cell technology to the modeling size

optimization and system performance [3ndash5] Modeling ofPV modules is one of the major components responsiblefor proper functioning of PV systems Modeling providesthe ways to understand the current voltage and powerrelationships of PV modules [6ndash8] However the estimationofmodels is affected by various intrinsic and extrinsic factorswhich ultimately influence the behavior of current andvoltage Therefore perfect modeling is essential to estimatethe performance of PV modules in different environmentalconditions Hernanz et al [9] compared the performance ofsolar cells with different models and pointed out that themanufacturers did not provide the values of the resistance inseries and parallel of the manufactured cell Andrews et al[10] proposed an improved methodology for fine resolutionmodeling of PV systems using module short circuit current



(119868sc) at 5min time scales Their work was a modified versionof the Sandia array performance model by incorporatingnew factors for the calculation of short circuit current (119868sc)to justify errors (including instrumentation alignment andspectral andmodule power tolerance errors) Chakrasali et al[11] investigated the performance of Nortonrsquos circuit model ofsolar PV module with the existing models using Matlab andreported that it is a well-suited way to predict the behavior ofPV modules operated for longer periods of time Chouder etal [12]modeled a PVmodule by a single diode lumped circuitand evaluated its main parameters by considering the powerconversion efficiency Chouder et al [13] presented a detailedcharacterization of the performance and dynamic behaviorof PV systems by using the LabVIEW platformThe LambertW function was applied for the solution of equations by Jainand Kapoor [14] Jain et al [15] Ortiz-Conde et al [16] andothers [17ndash19] Picault et al [17] presented a novel methodto forecast existing PV array production in diverse environ-mental conditions and concluded that Lambert W functionfacilitates a direct relationship between current and voltage ofmodules as it significantly reduces calculation time Chen etal [18] proposed an optimized method based on polynomialcurve fitting and Lambert W function for extraction ofparameters from the current-voltage (I-V) characteristics ofcommercial silicon solar cells The Lambert W function wasused for translation of transcendental equation into explicitanalytical solution Fathabadi [19] presented a novel methodfor characterization of silicon solar cells modules and plasticsolar cells Artificial neural network together with LambertW function was employed for determination of I-V and P-Vcurves of silicon and plastic solar cells and modules [20]

Moreover Krismadinata et al [21] used a single diodeelectrical equivalent circuit model for determination of PVcell characteristics and found that output of PV moduleswere strongly affected by the intensity of solar irradiationand ambient temperature Lu et al [22] investigated variousPV module layouts using full size as well as halved solarcells The performance of module layouts was investigated bypartially shading the PV cells using a solar cell equivalentcircuit model with SPICE software They found that theseries-parallel hybrid connection of cells within a modulehas a significant improvement on the power output of thePV module under partial shading conditions Mellit et al[23] employed a methodology for estimation power profileof a 50Wp Si-polycrystalline PV module by developing twoartificial neural networks (ANNs) for cloudy and sunny daysand found that the ANN-models performed better than theexisting models and also did not need more parametersunlike implicit models Singh [24] reviewed various modelsof PV cells and concluded that the accuracy of modelscan be improved by including series and shunt resistanceinto the model In addition the author also discoveredthat the estimation of models can further be improved byeither introducing two parallel diodes with independent setsaturation current or considering the diode quality factoras a variable parameter instead of fixed value like 1 or 2Thevenard and Pelland [25] reported that the uncertainties ofmodel predictions can be reduced by increasing the reliabilityand spatial coverage of solar radiation estimates appropriate

familiarity of losses due to dirt soiling and snow anddevelopment of better tools for PV system modeling Tian etal [26] presented a modified I-V relationship for the singlediode model The alteration in the model was made in theparallel and series connections of an array The derivation ofthe adapted I-V relationship was begunwith a single solar celland extended up to a PV module and finally an array Themodified correlation was investigated with a five-parametermodel based on the data provided by the manufacturers Theperformance of themodel was examined with a wide range ofirradiation levels and cell temperatures for prediction of I-Vand P-V curves maximum power point values short circuitcurrent and open circuit voltage Vincenzo and Infield [27]developed a detailed PV array model to deal explicitly withnonuniform irradiance and other nonuniformities across thearray and it was validated against data from an outdoortest system However the authors reduced the complexity ofthe simulations by assuming that the cell temperatures arehomogeneous for eachmodule Yordanov et al [28] presenteda new algorithm for determination of the series resistanceof crystalline-Si PV modules from individual illuminatedI-V curves The ideality factor and the reverse saturationcurrent were extracted in the typical way They found thatthe ideality factor at open circuit is increased by about 5It was established from the review that Lambert W functionis a simple technique to give the analytical explicit solutionof solar photovoltaic module characteristics as compared tothe other methods However some problems still exist for thederivation of the required model equations

Equivalent electrical circuit model is one of the keymodels under study since the last few decades It is configuredwith either single or double diode for investigation of current-voltage relationships The single diode models usually havefive four or three unknown parameters with only oneexponential term The five unknown parameters of a singlediode model are light-generated current (119868

119871) diode reverse

saturation current (119868119900) series resistance (119877

119904) shunt resistance

(119877sh) and diode ideality factor (119860) [29 30] The four-parameter model infers the shunt resistance as infinite andit is ignored [31] The three-parameter model assumes thatthe series resistance is zero and shunt resistance is infiniteand thus both of these parameters are ignored whereas thedouble diode models have six unknown parameters with twoexponential terms [32 33]

In fact both single and double diode models require theknowledge of all unknown parameters which is usually notprovided bymanufacturers Nevertheless the current-voltageequation is a transcendental expression It has no explicitanalytical solution It is also time consuming to discover itsexact analytical solution due to the limitation of availabledata for the extraction of required parameters [34ndash36] Forthat reason the researchers gradually focused on searchingout the approximate methods for the calculation of unknownparameters The analytical methods give exact solutions bymeans of algebraic equations However due to implicit natureand nonlinearity of PV cell or module characteristics itis hard to find out the analytical solution of all unknownparameters Analytical methods have also some limitationsand could not give exact solutions when the functions


are not given Thus numerical methods such as Newton-Raphson method or Levenberg-Marquardt algorithm werepreferred It is because of the fact that numerical methodsgive approximate solution of the nonlinear problems withoutsearching for exact solutions However numerical methodsare time consuming and need long term time series datawhich is not available in developing countries

It was revealed from the review that a wide variety ofmodels exist for estimation of power output of PV modulesHowever these were either complicated or gave approximatesolutions To overcome the limitations of both numericaland analytical methods an improved mathematical modelusing combination of numerical and analytical methods ispresented It makes the model simple as well as compre-hensive to provide acceptable estimations for PV modulepower outputs The values of required unknown parametersof I-V curve namely light-generated current diode reversesaturation current shape parameter and the series resistanceare computed analytically The expression for output currentof PV module is determined explicitly by Lambert W func-tion and voltage output is computed numerically by Newton-Raphson method

2 Formulated Model for Computing PowerOutput of PV Modules

The power produced by a PV module depends on intrinsicelectrical characteristics (current and voltage) and extrinsicatmospheric conditions The researchers generally incorpo-rate the most important electrical characteristics and influ-ential meteorological parameters in the models for the sakeof simplicity It is almost unfeasible to obtain a model thataccounts each parameter which influences the performanceof PV modules The models generally include those parame-ters which are commonly provided by manufacturers suchas the electrical properties of modules at standard ratingconditions [37] The standard equivalent electrical circuitmodel of PV cell denoted by a single diode is expressed as[38]

119868 = 119868119871minus 119868119863minus 119868sh (1)

where 119868119871is light-generated current 119868

119863is diode current and

119868sh is shunt currentThe diode current (119868119863) is expressed by the

Shockley equation as [39 40]

119868119863= 119868119900[119890120585(119881+119868119877

119904)minus 1] (2)

The shunt current (119868sh) is defined by Petreus et al [41] as

119868sh =119881 + 119868119877

119904

119877sh (3)

Therefore the final structure of five-parameter one diodeelectrical equivalent circuit model is graphically shown inFigure 1 It is also algebraically expressed as [40 42ndash44]

119868 = 119868119871minus 119868119900[119890120585(119881+ 119868119877

119904)minus 1] minus

119881 + 119868119877119904

119877sh (4)

V

I

IoIL

Rs

Rsh

Figure 1 Equivalent electrical circuit model of a PV module

where 119868119900is the reverse saturation current and 120585 is a term

incorporated for the simplicity of (4) which is expressed as

120585 =

119902

120582119896119879119888

(5)

where 119902 is electronic charge 119896 is Boltzmannrsquos constant 119879119888is

the cell temperature and 120582 is the shape factor which is givenas

120582 = 119860 times 119873CS (6)

where 119860 is the ideality factor and 119873CS is the number ofseries cells in a module The ideality factor (119860) does notdepend on the temperature as per definition of shape factor(120582) from semiconductor theory [45] The five unknownparameters namely 119868

119871 119868119900 119877119904 119877sh and 119860 can only be found

through complicated numerical methods from a nonlinearsolar cell equation It requires a close approximation of initialparameter values to attain convergence Otherwise the resultmay deviate from the real values [32 43] It is impracticalto find a method which can properly extract all requiredparameters to date [40 46] Thus for simplicity the shuntresistance (119877sh) is assumed to be infinity Hence the last termin (4) is ignored [47] Therefore the simplified form of four-parameter single diode equivalent circuit model which isused for this study is defined as [48]

119868 = 119868119871minus 119868119900[119890120585(119881+119868119877

119904)minus 1] (7)

From (7) a continuous relationship of current as functionof voltage for a given solar irradiance cell temperature andother cell parameters can be obtained

21 Determination of Unknown Parameters of FormulatedModel The expressions for unknown parameters such aslight-generated current (119868

119871) and reverse saturation current

(119868119900) were adopted from previous available models The

expressions for the shape factor (120582)which involve the idealityfactor (119860) and the series resistance (119877

119904) were algebraically

derived from existing equations The PV module used forthis study was NT-175 (E1) manufactured by Sharp EnergySolution Europe a division of Sharp Electronics (Europe)GmbH Sonninstraszlige 3 20097 Hamburg Germany

211 Light-Generated Current (119868119871) Light-generated current

(119868119871) is a function of solar radiation and module temperature


if the series resistance (119877119904) and the shape factor (120582) are

taken as constants For any operating condition 119868119871is related

to the light-generated current measured at some referenceconditions as [14 29]

119868119871= (

119878119879

119878119879119903

) [119868119871119903

+ 120583119868sc(119879119888minus 119879119888119903)] (8)

where 119878119879and 119878

119879119903are the absorbed solar radiation and 119879

119888

and 119879119888119903

are the cell temperatures at outdoor conditionsand reference conditions respectively 119868

119871119903is light-generated

current at reference conditions and 120583119868sc

is coefficient oftemperature at short circuit currentThe cell temperature (119879

119888)

can be computed from the ambient temperature and otherdata tested at nominal operating cell temperature (NOCT)conditions provided by the manufacturers

212 Reverse Saturation Current (119868119900) The reverse saturation

current (119868119900) is a function of temperature only [49] It is given

as

119868119900= 119863119879

3

119888119890120585120576119892119860 (9)

The reverse saturation current (119868119900) is a diminutive number

but its value increases by a factor of two with a temperatureincrease of 10∘C [50] It is actually computed by taking theratio of (9) at two different temperatures thereby eliminatingthe diode diffusion factor (119863) It is related to the temperatureonly Thus it is estimated at some reference conditions as thesame technique used for the determination of light current[49] Consider

119868119900= (

119879119888

119879119888119903

)

3

119890120585120582120576119892(119879119888119903minus119879119888)119860

(10)

213 Shape Factor (120582) Mostly manufacturers provide infor-mation of I-V characteristic curve at three different pointsusing reference conditions at open circuit voltage (119881oc) atshort circuit current (119868sc) and at optimum power point forboth current and voltageThe correlation for the given pointsare 119868 = 0 and 119881 = 119881oc at open circuit conditions 119868 = 119868scand 119881 = 0 at short circuit conditions and 119868 = 119868mp and119881 = 119881mp at maximum power point [48] By substituting theseexpressions in (7) it yields

119868sc119903 = 119868119871119903

minus 119868119900119903

[119890(120585119903119868sc119903119877119904)

minus 1] (11)

where

120585119903=

119902

120582119896119879119888119903

(12)

119868119871119903

minus 119868119900119903

[119890(120585119903119881oc119903)

minus 1] = 0 (13)

119868mp119903 = 119868119871119903

minus 119868119900119903

[119890120585119903(119881mp119903+119868mp119903119877119904)

minus 1] (14)

The reverse saturation current (119868119900) is a very small quantity

on the order of 10minus5 to 10minus6 A [47] It lessens the influence

of the exponential term in (11) Hence it is assumed to be

equivalent to 119868sc [32] One more generalization can be maderegarding the first term in (13) and (14) which could beignored Regardless of the system size the exponential term ismuch greater than the first termThus the equations become

119868119871119903

cong 119868sc119903 (15)

119868sc119903 minus 119868119900119903

[119890(120585119903119868sc119903119877119904)

] cong 0 (16)

119868mp119903 cong 119868119871119903

minus 119868119900119903

[119890120585119903(119881mp119903+119868mp119903119877119904)

] (17)

Solving (16) for reverse saturation current at reference condi-tions (119868

119900119903) is obtained as

119868119900119903

= 119868sc119903 [119890minus(120585119903119881oc119903)

] (18)

By substituting the value of reverse saturation current atreference conditions (119868

119900119903) from (18) into (17) it yields

119868mp119903 cong 119868sc119903 minus 119868sc119903 [119890120585119903(119881mp119903minus119881oc119903+119868mp119903119877119904)

] (19)

The Equation (17) can also be solved for 120585119903 which is given as

120585119903=

ln (1 minus 119868mp119903119868sc119903)

119881mp119903 minus 119881oc119903 + 119868mp119903119877119904 (20)

Finally the value of the shape factor (120582) can be obtained bycomparing (12) and (20) as

120582 =

119902 (119881mp119903 minus 119881oc119903 + 119868mp119903119877119904)

119896119879119888119903ln (1 minus 119868mp119903119868sc119903)

(21)

214 Series Resistance (119877119904) The series resistance (119877

119904) is an

essential parameter when the module is not operating nearthe reference conditionsThis characterizes the internal lossesdue to current flow inside the each cell and in linkagesbetween cells It alters the shape of I-V curve near optimumpower point and open circuit voltage however its effect issmall [29 36] I-V curve without considering 119877

119904would be

somewhat dissimilar than the curves outlined including itsvalue On the basis of annual simulation the predicted poweroutput from PV systems will be 5 to 8 lower when correctseries resistance is not used [32 51] It can be determined as

119889119881

119889119868

10038161003816100381610038161003816100381610038161003816119881oc119903

minus 119877119904= 0 (22)

To obtain differential coefficient for (22) first the current (119868)can be extracted explicitly as a function of voltage (119881) byusing Lambert W function from (7) and is expressed as

119868 = (119868119871+ 119868119900) minus

119882(120585119877119904119890120585(119881+119868

119871119877119904+119868119900119877119904))

120585119877119904

(23)

By differentiating (23) with respect to 119881 and taking itsreciprocal at 119881 = 119881oc119903 and 119868

119871= 119868119871119903

and substituting into(22) it gives

119882(120585119877119904119868119900119890120585(119881oc119903+119868119871119903119877119904+119868119900119877119904)

)

[1 +119882(120585119877119904119868119900119890120585(119881oc119903+119868119871119903119877119904+119868119900119877119904))] 119877

119904

minus 119877119904= 0 (24)


A number of simplifications have beenmade in order to solve(24) analytically for 119877

119904 For example 119868

119900is usually taken in the

order of 10minus5 to 10minus6 [47] Its value for this study was taken

as the order of 10minus6 Similarly the values of 119868119871119903

and119881oc119903 weretaken from the manufacturers data The expression for 119877

119904is

obtained based on the above simplifications and by puttingthe value of 120585 in (24) as

119877119904

= 18Re(119879119888119882(minus15 times 10

71198900022(47+(17times10

5

119879119888)))

119879119888119882(minus15times 10

71198900022(47+(17times10

5119879119888)))+4400

)

(25)

where Re represents the real part because the negativeexpression inside the Lambert W function results in acomplex number However in practical problems only realvalues are to be considered

22 Determination of Optimum Power Output Parameters ofProposed Model The optimum power output parameters ofmodel were determined by deriving the equations for current(119868) and voltage (119881) by putting the values of unknown parame-ters namely 119868

119871 119868119900119877119904 and 120582 in the respective equationsThe

power (119875) is the product of current (119868) and voltage (119881) [48]therefore it can be expressed as

119875 = 119868119881 (26)

By substituting 119868 from (23) into (26) the value of 119875 iscomputed as [52]

119875 = (119868119871+ 119868119900) minus

119882[120585119877119904119890120585(119881+119868

119871119877119904+119868119900119877119904)]

120585119877119904

119881 (27)

Mathematically the optimum power occurs at the point119881maxof P-V curve where the slope of tangent line is equal to zeroas follows

119889119875

119889119881

10038161003816100381610038161003816100381610038161003816119881max

= 0 (28)

By differentiating (27) with respect to voltage (119881) and takingthe RHS equal to zero

119882[120585119877119904119868119900119890120585(119881+119868

119871119877119904+119868119900119877119904)]119881

1 +119882[120585119877119904119868119900119890120585(119881+119868

119871119877119904+119868119900119877119904)] 119877119904

+

1

120585119877119904

times 119882[120585119877119904119868119900119890120585(119881+119868

119871119877119904+119868119900119877119904)] minus 120585119877

119904(119868119871+ 119868119900) = 0

(29)

Newton-Raphson method is applied to (29) in order tofind the critical value of 119881 for 119881max The value of 119881max issubstituted into (27) in order to solve the maximum power(119875max) Consider

119875max = (119868119871+ 119868119900) minus

119882[120585119877119904119890120585(119881max+119868119871119877119904+119868119900119877119904)

]

120585119877119904

119881max (30)

0 5 10 15 20 25 30 35 40

25

2

15

1

05

0

Curr

ent (

A)

Voltage (V)

Vmax

Imax

Isc

Voc

Simulated dataManufactured data

Optimum powerpoint

Figure 2 Typical I-V characteristic curve of a PV module

Consequently the desired maximum current (119868max) can beobtained from (30) by dividing the maximum power (119875max)with maximum voltage (119881max) Consider

119868max =119875max119881max

(31)

3 Simulation of I-V and P-V CharacteristicCurves of a Selected PV Module

The familiarity of current and voltage relationship of photo-voltaic modules under real operating conditions is essentialfor the determination of their power output Normally thecells are mounted inmodules andmultiple modules are usedin arrays to get desired power output Individual modulesmay have cells connected in series and parallel combinationsto obtain the required current and voltage Similarly thearray of modules may be arranged in series and parallelconnections When the cells or modules are connected inseries the voltage is additive and when they are attached inparallel the currents are additive [52ndash56] The power outputof PV modules could be predicted from the behavior ofcurrent-voltage I-V and power-voltage P-V characteristiccurves The current-voltage and power-voltage characteristiccurves are graphically shown in Figures 2 to 9

The current-voltage I-V characteristic of a typical PVmodule is shown in Figure 2When the output voltage119881 = 0the current is the short circuit current (119868sc) and when thecurrent 119868 = 0 the output voltage is the open circuit voltage(119881oc) Mostly the current decreases slowly at a certain pointand then decreases rapidly to the open circuit conditionsThe power as a function of voltage is given in Figure 3 Themaximum power that can be obtained corresponds to therectangle of maximum area under I-V curve At the optimumpower point the power is 119875mp the current is 119868mp and thevoltage is 119881mp Ideally the cells would always operate at theoptimum power point that matches the I-V characteristic ofthe load Hence the load matching is essential for extractingthe maximum power from the solar photovoltaic modules


0

0

5 10

10

15 20

20

25 30

30

35 40

40

50

60

70

80

Voltage (V)


Vmax

Optimum power

Pow

er (W

)

Pmax

Figure 3 Typical P-V characteristic curve of a PV module

0

0

1

2

3

4

5

5 10 15 20 25 30 35 4540

Voltage (V)


Curr

ent (

A)

GT = 1000 (Wm2)

GT = 800 (Wm2)

GT = 600 (Wm2)

GT = 400 (Wm2)

GT = 200 (Wm2)

Optimum powerpoint

Figure 4 I-V characteristic curves at various solar radiation levels

Therefore the maximum power point tractors are preferredto optimize the output power from solar PV systems

I-V characteristic curves at various solar irradiation levelsand temperatures are shown in Figures 4 and 5 respectivelyThe locus of maximum power point is indicated on thecurves The short circuit current increases in proportion tothe solar radiation while the open circuit voltage increaseslogarithmically with solar radiation As long as the curvedportion of the I-V characteristic does not intersect theshort circuit current is nearly proportional to the incidentsolar radiation If the incident solar radiation is assumedto be a fixed spectral distribution the short circuit currentcan be used as a measure of incident solar radiation I-Vcharacteristics curves for the combination of irradiance andtemperatures are illustrated in Figure 6 It was observedthat the temperature linearly decreases the output voltage ascompared to current Consequently the decrease of voltagelowers the power output of PV module at constant solarirradiation level However the effect of temperature is smallon short circuit current but increases with the increase ofincident solar radiation

0

0

1

2

3

4

5

5 10 15 20 25 30 35 40

Voltage (V)


Curr

ent (

A)

Optimum powerpoint

Ta = 0∘C

Ta = 25∘C

Ta = 50∘C

Ta = 75∘C

Figure 5 I-V characteristic curves at various temperatures

0

0

1

2

3

4

5

5 10 15 20 25 30 35 40

Voltage (V)


Curr

ent (

A)

Optimum powerpoint

GT = 500Wm2

Ta = 20∘C

Ta = 60∘C

GT = 1000Wm2

Figure 6 I-V characteristic curves for various set of solar radiationand temperature

The P-V characteristics curves for various solar irradi-ation levels at constant temperature of 25∘C and at severaltemperatures with constant solar irradiance of 1000Wm2is illustrated in Figures 7 and 8 respectively Increasingtemperature leads to decreasing the open circuit voltageand slightly increasing the short circuit current Operatingof cell temperature at that region of the curve leads to asignificant power reduction at high temperatures The P-Vcharacteristics curves for the combination of irradiance andtemperatures are shown in Figure 9

The power output of photovoltaic module by formulatedmodel gave aplusmn2 error when comparedwith the rated powerof PV module provided by manufacturers on average basisHowever at higher solar radiation and temperature valuesthemodel simulated results were somehow deviated from therated power of PV module Since the shunt resistance (119877sh)was assumed to be infinity in the proposed model It wasfound from the analysis that the increase of temperature anddecrease of incident solar radiation levels lead to lower poweroutput and vice versa The power output from PV modules


0 5 10 15 20 25 30 35 4540

Voltage (V)


Optimum power

GT = 1000 (Wm2)

GT = 800 (Wm2)GT = 600 (Wm2)GT = 400 (Wm2)GT = 200 (Wm2)

180

160

140

120

100

80

60

40

20

0

Pow

er (W

)

Figure 7 P-V characteristic curve at constant temperature of 25∘C

0 5 10 15 20 25 30 35 4540

Voltage (V)


Optimum power180

160

140

120

100

80

60

40

20

0

Pow

er (W

) Ta = 0∘C

Ta = 25∘C

Ta = 50∘C

Ta = 75∘C

Figure 8 P-V characteristic curve at constant solar radiation of1000Wm2

0 5 10 15 20 25 30 35 40

Voltage (V)


Optimum power160

140

120

100

80

60

40

20

0

Pow

er (W

)

Ta = 60∘C

Ta = 20∘C

GT = 1000Wm2

GT = 500Wm2

Figure 9 P-V characteristic curve at various set of solar radiationand temperature

approaches zero if the amount of solar radiation tends todecrease and the temperature goes up

4 Conclusions

The proposed mathematical model is formulated by inte-gration of both analytical and numerical methods Therequired parameters of current-voltage (I-V) curve such asthe light-generated current diode reverse saturation currentshape parameter and the series resistance are computedanalytically The expression for output current from PVmodule is determined explicitly by Lambert W function andvoltage output is computed numerically by Newton-Raphsonmethod The main contribution of this study is algebraicderivation of equations for the shape factor (120582)which involvethe ideality factor (119860) and the series resistance (119877

119904) of single

diode model of PV module power output These equationswill help to find out the predicted power output of PVmodules in precise and convenient manner

The current-voltage (I-V) and the power-voltage (P-V)characteristic curves obtained from the proposed modelwere matching with the curves drawn from the PV moduleat standard test conditions The variation of incident solarradiation and temperature were found to be themain cause ofmodifications in the amount of PV module power output Alinear relationship between the power output of PV moduleand the amount of incident solar radiation were observed ifother factors were kept constant

The estimated results of the proposedmodel are validatedby PVmodule rated power output provided bymanufacturerwhich gave a plusmn2 error The model is found to be morepractical in terms of the number of variables used andpredicted satisfactory performance of PV modules

Nomenclature

119878119879 absorbed solar radiation Wm2

119896 Boltzmannrsquos constant 1381 times 10minus23 JK

119879119888 Cell temperature at actual conditions K

119868 Current output of cell A119868119863 Diode current or dark current A

119863 Diode diffusion factor -119868119900 Diode reverse saturation current A

119860 Ideality factor 1 for ideal diodes andbetween 1 and 2 for real diodes

119882 Lambert W function -119868119871 Light-generated current A

120576119892 Material band gap energy eV 112 eV for

silicon and 135 eV for gallium arsenide119868mp Maximum current of PV module A119875mp Maximum power of PV module W119881mp Maximum voltage of PV module V119873cs Number of cells in series -119881oc Open circuit voltage of PV module V120585 Parameter 119902119896119879

119888120582

119877119904 Series resistance Ω

120582 Shape factor of I-V curve -


119868sc Short circuit current of PV module A119877sh Shunt resistanceΩ120583119868sc Temperature coefficient of short circuitcurrent -

120583119881oc Temperature coefficient of voltage VK

119881 Voltage output of cell V119866119879 Solar radiation Wm2

119902 Electron charge 1602 times 10minus19C

119879119886 Ambient temperature ∘C

The Subscript

119903 In any notation the corresponding value ofparameter at reference conditions



References

[1] W Durisch D Tille A Worz and W Plapp ldquoCharacterisationof photovoltaic generatorsrdquo Applied Energy vol 65 no 1ndash4 pp273ndash284 2000

[2] A Q Jakhrani A K Othman A R H Rigit S R Samo andS R Kamboh ldquoA novel analytical model for optimal sizing ofstandalone photovoltaic systemsrdquoEnergy vol 46 no 1 pp 675ndash682 2012

[3] A N Celik ldquoA simplified model based on clearness index forestimating yearly performance of hybrid pv energy systemsrdquoProgress in Photovoltaics Research and Applications vol 10 no8 pp 545ndash554 2002

[4] A H Fanney and B P Dougherty ldquoBuilding integrated photo-voltaic test facilityrdquo Journal of Solar Energy Engineering vol 123no 3 pp 194ndash199 2001

[5] M A Green ldquoShort communication priceefficiency correla-tions for 2004 photovoltaic modulesrdquo Progress in PhotovoltaicsResearch and Applications vol 13 pp 85ndash87 2005

[6] A Q Jakhrani A K Othman A R H Rigit and S R SamoldquoModel for estimation of global solar radiation in SarawakMalaysiardquo World Applied Sciences Journal vol 14 pp 83ndash902011

[7] A Q Jakhrani A K Othman A R H Rigit and S R SamoldquoAssessment of solar and wind energy resources at five typicallocations in Sarawakrdquo Journal of Energy amp Environment vol 3pp 8ndash13 2012

[8] R Ramaprabha andB LMathur ldquoDevelopment of an improvedmodel of SPV cell for partially shaded solar photovoltaic arraysrdquoEuropean Journal of Scientific Research vol 47 no 1 pp 122ndash1342010

[9] J A R Hernanz J J C ampayoMartin I Z Belver J L LesakaE Z Guerrero and E P Perez ldquoModelling of photovoltaicmodulerdquo in Proceedings of the International Conference onRenewable Energies and Power Quality (ICREPQ rsquo10) GranadaSpain March 2010

[10] R B Andrews A Pollard and J M Pearce ldquoImproved para-metric empirical determination of module short circuit currentfor modelling and optimization of solar photovoltaic systemsrdquoSolar Energy vol 86 pp 2240ndash2254 2012

[11] R L Chakrasali V R Sheelavant andHNNagaraja ldquoNetworkapproach tomodeling and simulation of solar photovoltaic cellrdquoRenewable and Sustainable Energy Reviews vol 21 pp 84ndash882013

[12] A Chouder S Silvestre N Sadaoui and L Rahmani ldquoMod-eling and simulation of a grid connected PV system basedon the evaluation of main PV module parametersrdquo SimulationModelling Practice andTheory vol 20 no 1 pp 46ndash58 2012

[13] A Chouder S Silvestre B Taghezouit and E KaratepeldquoMonitoring modeling and simulation of PV systems usingLabVIEWrdquo Solar Energy vol 91 pp 337ndash349 2013

[14] A Jain and A Kapoor ldquoA new approach to study organic solarcell using Lambert W-functionrdquo Solar Energy Materials andSolar Cells vol 86 no 2 pp 197ndash205 2005

[15] A Jain S Sharma and A Kapoor ldquoSolar cell array parametersusing Lambert W-functionrdquo Solar Energy Materials and SolarCells vol 90 no 1 pp 25ndash31 2006

[16] A Ortiz-Conde F J Garcıa Sanchez and JMuci ldquoNewmethodto extract the model parameters of solar cells from the explicitanalytic solutions of their illuminated I-V characteristicsrdquo SolarEnergy Materials and Solar Cells vol 90 no 3 pp 352ndash3612006

[17] D Picault B Raison S Bacha J de la Casa and J AguileraldquoForecasting photovoltaic array power production subject tomismatch lossesrdquo Solar Energy vol 84 no 7 pp 1301ndash1309 2010


[19] H Fathabadi ldquoNovel neural-analytical method for determiningsiliconplastic solar cells and modules characteristicsrdquo EnergyConversion and Management vol 76 pp 253ndash259 2013


[21] Krismadinata N A Rahim H W Ping and J Selvaraj ldquoPho-tovoltaic module modeling using simulinkmatlabrdquo ProcediaEnvironmental Sciences vol 17 pp 537ndash546 2013

[22] F Lu S Guo T M Walsh and A G Aberle ldquoImproved PVmodule performance under partial shading conditionsrdquo EnergyProcedia vol 33 pp 248ndash255 2013

[23] A Mellit S Saglam and S A Kalogirou ldquoArtificial neuralnetwork-based model for estimating the produced power ofa photovoltaic modulerdquo Renewable Energy vol 60 pp 71ndash782013

[24] G K Singh ldquoSolar power generation by PV (photovoltaic)technology a reviewrdquo Energy vol 3 pp 1ndash13 2013

[25] D Thevenard and S Pelland ldquoEstimating the uncertainty inlong-term photovoltaic yield predictionsrdquo Solar Energy vol 91pp 432ndash445 2013

[26] H Tian F Mancilla-David K Ellis E Muljadi and P JenkinsldquoA cell-to-module-to-array detailed model for photovoltaicpanelsrdquo Solar Energy vol 86 pp 2695ndash2706 2012

[27] M C D Vincenzo and D Infield ldquoDetailed PV array model fornon-uniform irradiance and its validation against experimentaldatardquo Solar Energy vol 97 pp 314ndash331 2013

[28] G H Yordanov O-M Midtgard and T O Saetre ldquoSeriesresistance determination and further characterization of c-SiPV modulesrdquo Renewable Energy vol 46 pp 72ndash80 2012


[29] W De Soto S A Klein andW A Beckman ldquoImprovement andvalidation of amodel for photovoltaic array performancerdquo SolarEnergy vol 80 no 1 pp 78ndash88 2006

[30] E Karatepe M Boztepe and M Colak ldquoNeural network basedsolar cell modelrdquo Energy Conversion and Management vol 47no 9-10 pp 1159ndash1178 2006

[31] Y-C Kuo T-J Liang and J-F Chen ldquoNovel maximum-power-point-tracking controller for photovoltaic energy conversionsystemrdquo IEEE Transactions on Industrial Electronics vol 48 no3 pp 594ndash601 2001

[32] R Chenni M Makhlouf T Kerbache and A Bouzid ldquoAdetailed modeling method for photovoltaic cellsrdquo Energy vol32 no 9 pp 1724ndash1730 2007

[33] A Wagner Peak-Power and Internal Series Resistance Mea-surement under Natural Ambient Conditions EuroSun Copen-hagen Denmark 2000

[34] M Chegaar G Azzouzi and P Mialhe ldquoSimple parameterextraction method for illuminated solar cellsrdquo Solid-State Elec-tronics vol 50 no 7-8 pp 1234ndash1237 2006

[35] K Bouzidi M Chegaar and A Bouhemadou ldquoSolar cellsparameters evaluation considering the series and shunt resis-tancerdquo Solar Energy Materials and Solar Cells vol 91 no 18 pp1647ndash1651 2007

[36] J C Ranuarez A Ortiz-Conde and F J Garcıa Sanchez ldquoA newmethod to extract diode parameters under the presence of para-sitic series and shunt resistancerdquoMicroelectronics Reliability vol40 no 2 pp 355ndash358 2000

[37] M A De Blas J L Torres E Prieto and A Garcıa ldquoSelecting asuitable model for characterizing photovoltaic devicesrdquo Renew-able Energy vol 25 no 3 pp 371ndash380 2002

[38] D Petreus C Frcas and I Ciocan ldquoModelling and simulationof photovoltaic cellsrdquo Acta Technica Napocensis Electronica-Telecomunicatii vol 49 pp 42ndash47 2008

[39] V D Dio D La Cascia R Miceli and C Rando ldquoA mathe-matical model to determine the electrical energy production inphotovoltaic fields under mismatch effectrdquo in Proceedings of theInternational Conference on Clean Electrical Power (ICCEP rsquo09)pp 46ndash51 Capri Italy June 2009

[40] W Kim andW Choi ldquoA novel parameter extractionmethod forthe one-diode solar cell modelrdquo Solar Energy vol 84 no 6 pp1008ndash1019 2010

[41] D Petreus I Ciocan and C Farcas An Improvement onEmpirical Modelling of Photovoltaic Cells ISSE Madrid Spain2008

[42] M S Benghanem and N A Saleh ldquoModeling of photovoltaicmodule and experimental determination of serial resistancerdquoJournal of Taibah University for Science vol 1 pp 94ndash105 2009

[43] A N Celik ldquoEffect of different load profiles on the loss-of-loadprobability of stand-alone photovoltaic systemsrdquo RenewableEnergy vol 32 no 12 pp 2096ndash2115 2007

[44] Q Kou S A Klein and W A Beckman ldquoA method forestimating the long-term performance of direct-coupled PVpumping systemsrdquo Solar Energy vol 64 no 1ndash3 pp 33ndash40 1998

[45] J A Duffie and W A Beckman Solar Engineering of ThermalProcesses John Wiley amp Sons 3rd edition 2006

[46] A H Fanney B P Dougherty and M W Davis ldquoEvaluat-ing building integrated photovoltaic performance modelsrdquo inProceedings of the 29th IEEE Photovoltaic Specialists Conference(PVSC rsquo02) New Orleans La USA May 2002

[47] R L Boylestad L Nashelsky and L Li Electronic Devices andCircuit Theory Prentice-Hall 10th edition 2009

[48] E Saloux A Teyssedou and M Sorin ldquoExplicit model ofphotovoltaic panels to determine voltages and currents at themaximum power pointrdquo Solar Energy vol 85 no 5 pp 713ndash722 2011

[49] J Crispim M Carreira and C Rui ldquoValidation of photovoltaicelectrical models against manufacturers data and experimentalresultsrdquo in Proceedings of the International Conference on PowerEngineering Energy and Electrical Drives (POWERENG rsquo07) pp556ndash561 Setubal Portugal April 2007

[50] F M Gonzalez-Longatt Model of Photovoltaic Module inMatlab II CIBELEC Puerto La Cruz Venezuela 2006

[51] D Sera and R Teodorescu ldquoRobust series resistance estimationfor diagnostics of photovoltaic modulesrdquo in Proceedings of the35thAnnual Conference of the IEEE Industrial Electronics Society(IECON rsquo09) pp 800ndash805 Porto Portugal November 2009

[52] A Q Jakhrani A K Othman A R H Rigit R Baini SR Samo and L P Ling ldquoInvestigation of solar photovoltaicmodule power output by various modelsrdquo NED UniversityJournal of Research pp 25ndash34 2012

[53] A Q Jakhrani A K Othman A R H Rigit and S RSamo ldquoComparison of solar photovoltaic module temperaturemodelsrdquoWorld Applied Sciences Journal vol 14 pp 1ndash8 2011

[54] A Q Jakhrani A K Othman A R H Rigit and S R SamoldquoDetermination and comparison of different photovoltaicmod-ule temperaturemodels for Kuching Sarawakrdquo inProceedings ofthe IEEE 1st Conference on Clean Energy and Technology (CETrsquo11) pp 231ndash236 Kuala Lumpur Malaysia June 2011

[55] A Q Jakhrani S R Samo A R H Rigit and S A KambohldquoSelection of models for calculation of incident solar radiationon tilted surfacesrdquoWorld Applied Sciences Journal vol 22 no 9pp 1334ndash1341 2013

[56] A Q Jakhrani A K Othman A R H Rigit S R Samo andS A Kamboh ldquoSensitivity analysis of a standalone photovoltaicsystem model parametersrdquo Journal of Applied Sciences vol 13no 2 pp 220ndash231 2013

Research ArticleA Virtual PV Systems Lab for EngineeringUndergraduate Curriculum

Emre Ozkop and Ismail H Altas

Department of Electrical and Electronics Engineering Karadeniz Technical University 61080 Trabzon Turkey

Correspondence should be addressed to Emre Ozkop emreozkophotmailcom



Copyright copy 2014 E Ozkop and I H Altas This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

Design and utilization of a Virtual Photovoltaic Systems Laboratory for undergraduate curriculum are introduced in this paperThelaboratory introduced in this study is developed to teach students the basics and design steps of photovoltaic solar energy systems ina virtual environment before entering the fieldThe users of the proposed virtual lab will be able to determine the sizing by selectingrelated parameters of the photovoltaic system to meet DC and AC loading conditions Besides the user will be able to analyze theeffect of changing solar irradiation and temperature levels on the operating characteristics of the photovoltaic systems CommonDC bus concept and AC loading conditions are also included in the system by utilizing a permanent magnet DC motor and anRLC load as DC and AC loading examples respectively The proposed Virtual Photovoltaic Systems Laboratory is developed inMatlabSimulink GUI environment The proposed virtual lab has been used in Power Systems Lab in the Department of Electricaland Electronics Engineering at Karadeniz Technical University as a part of undergraduate curriculum A survey on the studentswho took the lab has been carried out and responses are included in this paper

1 Introduction

As the utilization of photovoltaic (PV) solar energy systemsgain importance day by day training seminar sessions shortcourses and certificate programs are offered by companiesin order to close the gap of the experienced professionals indesigning and utilizing the PV systems As a part of encourag-ing the use of renewable energy sources government inmanycountries recommend schools and universities to includethe renewable energy sources such as solar and wind intheir curriculum Modeling analyzing and designing of PVsolar energy systems are taught in a course called Designof Low Voltage Power Systems and a set of PV experimentsare included in Power Systems Lab in the Department ofElectrical and Electronics Engineering at Karadeniz Techni-cal University Turkey In order to teach the modeling andcharacteristic behaviors of the PVarrays a virtual PVSystemsLab is developed and offered to be used by the students as apart of undergraduate curriculum

The changes and developments such as globalization eco-nomic crisis and technological innovations involve recon-sidering education system curriculum [1] In conventional

education system courses have been based on the princi-pal of printed word and blackboard [2] Studying mainlymathematics and basic concepts to understand the dynamicsystem responses causes students to be less interested inessential subject If the students are provided with variousways to reach learning sources of information which isone of the flexible education aims classical education basedon textbooks and chalkboards becomes time consumingand insufficient in completion realize the dynamic systemresponses nowadays [3 4]

While students attain theoretical knowledge in the class-room practical knowledge and experiences are given inlaboratories [2] Laboratory is an integral part of under-graduate studies particularly in engineering education [1]Laboratory works give students the ability to design andconduct experiments analyze and interpret data design asystem and component and develop social and team workskills and also increase learning efficiency in engineeringeducation [2 5 6] There are many studies to improveengineering studentsrsquo skills [7]

Even though students get theoretical background verywell in detail in classroom and they are provided labmanuals



to study and become familiar with the experiments they aregoing to do in laboratory they always face difficulties inpractice when it comes to setting up and performing theactual experimental tests It has been always a task to havethe students gain the ability to turn the theoretical knowledgeinto practical reality In particular the engineering studentsshould be able to combine the theory with the practiceCourse laboratories in schools and universities are the firstand important places for the students to put their knowledgeinto practice However every student does not have thechance to use the labs for practicing before the lab hour forthe course starts Every student does not have the income toestablish hisher own high cost test laboratory for checkingthe experiment designed to be performed

Experimental setup has a high cost and also lacks doingexperiments again and again in nonlinear range [8 9]There are generally limited laboratory equipments spacetime period and danger for laboratory experiments suchthat it is not possible for the students to do an experimentalone in laboratory Besides the increasing number of thestudents every year becomes a heavy burden to departmentswith laboratories [1 5] Consequently either the numberof experiment groups or the number of students in theexperiment groups must be increased In any case moreteaching assistants and lab equipments are required todeliver the experiments efficiently Otherwise the studentswill not be able to learn and interpret effectively in theexperiments [1 9ndash11] If the required conditions for a highquality or a normal laboratory are not provided then thepractical experiment part of the education will be insufficientresulting in inexperienced graduates deprived of scientificknowledge and abilities [3] Insufficient laboratory conditionsand experiments decrease the education quality and result inthe presence of incompetent graduates with less ability [1]Therefore new alternatives are always the challenge to keepthe laboratory experiments and the practical abilities of thegraduates highThat is where the virtual laboratories becomeimportant Virtual laboratories are software based simulatorsof the actual systems and require only the computers to beperformed

Since computer prices decrease and it is possible tofind easily variable versatile software programs computer-aided education has become a part of both classroomand laboratory experiments [12] With new technologicalinnovations computers have been benefited as auxiliaryequipments and teaching tools in many universities [13] Theuses of computers make a positive impact in lectures andlaboratory work and also contribute economically towardsmitigating the education cost [1] A lack of concentrationlessons and studentsrsquo performances are positively affectedwith computer applications as compared to hands-on labsalone [1] Computer based virtual lab software has beenused as a part of education tools to develop the studentsrsquocomprehension in both theoretical and experimental topics[14] When the comparison is made between the hands-on experiments and the virtual learning experiments it isobvious that the virtual one provides a flexible time invariantand location free learning environment to perform manydifferent and less costly experiments Moreover each student

can realize more complex and extensive experiments aloneso that self-learning potential of the students is brought toforward [14] The virtual laboratory enables easy interactionfor students and has flexible parameter variations for ongo-ing experiments and gives prominent understanding intoexperimental dynamics [6] Virtual learning environmentsgive students opportunity to establish a connection fromtheoretical concepts to practical applications [14] The vir-tual learning environments corroborate students to designexperiments based on the concept they are studying andalso provide immediate graphical feedback It is obviousthat the virtual learning has a comprehensive and interactiveenvironment [1] Nowadays many universities and institu-tions have applied virtual laboratories A virtual laboratorycan be utilized by the means of combining both physicalexperiments and numerical simulations [14]

Commercially available software packages have beenprogressively used for educational purposes as the highperformance and cost efficient computers developed [1516] There are many different software packages developedusing independent language platforms These virtual labsare usually developed for certain purposes and are notexpandable for more detailed experiments On the otherhand there are some software platforms that are suitablefor developing virtual laboratories MATLAB is one of thesesoftware platforms used by a great number of universitiesthroughout the world and it might be the better choice sinceengineering students are generally acquaintedwithMATLAB[2 12 16ndash19]The user can design develop expand simulateanalyze and visualize data with MATLAB and its toolboxSimulink software [16 20]

Graphical user interface (GUI) ability ofMatlabSimulinksoftware environment is used to develop the Virtual PVSystems lab introduced in this paper Mathematical modelof the PV cell and PV array given in [21] is modified andused in this study The Virtual PV Systems lab providesthe measurement and analysis of PV arrays I-V and P-Vcharacteristics at constant and variable solar irradiation andtemperature levels The user is able to modify the numberof panels in series and in parallel so that a desired arraycombination can be designed tomatch the load requirementsThe virtual lab also includes a power electronics converterto give the opportunity of controlling the output voltage ofthe PV arrays A permanent magnet DC motor is also usedas a load to analyze the operational characteristics of thePV system under variable load The proposed virtual lab hasbeen used in Power Systems laboratory in the Department ofElectrical and Electronics Engineering at Karadeniz technicalUniversity Turkey and the responses from the students areevaluated

2 System Architecture

The Virtual PV Systems Lab (VPVSL) is intended to be usedas an auxiliary supporting virtual environment for the actualPV System labThe idea behind the combination of the virtualand actual labs is based on student outcome requirementsdescribed by ABET as A-K engineering education criteriarsquos


Figure 1 The main menu of VPVSL

[22] Two of the A-K criteriarsquos highlight the importanceof the ldquoability to design and conduct experiments as wellas to analyze and interpret datardquo and ldquoability to design asystem component or process to meet desired needs withinrealistic constraints rdquo [22] In order to have the studentsgain the ability of designing and conducting experiments onPV systems proposed VPVSL is developed

The lab is conducted in the following order

(i) Define the size and operational specifications of thesystem

(ii) Determine the values of required parameters andenter them into the VPVSL software using the userdata interfacing windows shown in Figures 1 2 3 45 and 6

(iii) Select the configuration (DC or AC load) type to beused

(iv) Obtain I-V and P-V characteristics of the PV array forvarious solar irradiation and temperature levels andanalyze them

(v) Obtain the optimum operating power points forvarious operating conditions

(vi) Run the VPVSL record the results and analyze them

Students write their lab reports based upon the data theyobtained from the VPVSL and keep them to compare withthose obtained using actual PV system lab The use of theVPVSL before the actual lab is that students get to knowthe PV system and its operational characteristics so that theybecome familiar with the designing steps and characteristicsTherefore they have no difficulties when performing the realtime PV system experiments as well as designing real time PVsystem applications The VPVSL software is developed usingMatlabSimulink GUI environment Figure 1 shows the firstuser interfacing screen which is also called themain windowThis window includes the following five modules which areexplained in the next sections

(1) PV characteristics(2) PV solar irradiation

(3) PV temperature(4) PV powered PMDCmotor(5) PV powered AC loads

21 Module 1 PV Array Characteristics Thismodule uses thePV cell model described by

119881119862=

119860119896119879119862

119890

ln(119868ph + 1198680 minus 119868119862

1198680

) minus 119877119878119868119862 (1)

where 119860 is curve fitting constant used in solar cell I-Vcharacteristics 119890 is electron charge 119896 is Boltzmann constant119868119862is the cell output current 119868ph is photocurrent 119868

0is the

reverse saturation current of diode 119877119878is series resistance of

cell 119879119862is reference cell operating temperature and 119881

119862is the

cell output voltage The cell parameters values used in thisstudy are given in the Appendix All these cell parametersand modelling details are given in [21] The module runsthe characteristic equation in the back and offers the user toenter the simulation parameters that are needed to obtain thePV array current-voltage and power-voltage characteristicsunder certain solar irradiance and ambient temperature Theuser interface window of Module 1 is shown in Figure 2Number of cells in series (119873

119878) in a branch and number of

parallel connected series branches (119873119875) are the main inputs

for sizing the PV array to get the required power Solarirradiation and ambient temperature are the other two inputsin order to obtain the PV array characteristics for these lightand temperature levels

Students are able to resize the PV array to meet the loadrequirements and obtain the I-V and P-V characteristicsAfter entering the panel configuration and external condi-tion the module returns the characteristics along with themaximum power point operating quantities This moduleteaches the student how the resizing affect the power outputof a PV array under constant solar irradiation and ambienttemperature

22 Module 2 Solar Irradiation Effects on PV Array Char-acteristics This module shows how the changes in solarirradiation affect the output current and voltage and thepower of a PV arrayThe inclusion of the effects of the changesin solar irradiation level in PV arraymodeling is described bythe following equations which are discussed in [23 24]

119862119879119881= 1 + 120573

119879(119879119886minus 119879119909) (2)

119862119879119868= 1 +

120574119879

119878119862

(119879119909minus 119879119886) (3)

119862119878119881= 1 + 120573

119879120572119878(119878119909minus 119878119862) = 1 + 120573

119878(119878119909minus 119878119862) (4)

119862119878119868= 1 +

1

119878119862

(119878119909minus 119878119862) = 1 + 120574

119878(119878119909minus 119878119862) (5)

Δ119879119862= 120572119878(119878119909minus 119878119862) (6)

119881119862119883= 119862119879119881119862119878119881119881119862 (7)

119868ph119909 = 119862119879119868119862119878119868119868ph (8)


Figure 2 The photovoltaic array characteristics interface

where 119879119886is the reference ambient temperature 119878

119862is the

reference solar irradiation and 119879119909and 119878

119909are the current

values of the ambient temperature and solar irradiationrespectively 120573

119879and 120574119879are constants representing the effect

of temperature on PV cell voltage and current respectivelyThe constant 120572

119878represents the slope of the change in the

cell operating temperature due to a change in the solarirradiation Constants 120573

119878and 120574119878represent effect of the solar

irradiation on cell voltage and photocurrent respectivelyEquations (2) and (3) yield the temperature gain parameters119862119879119868

and 119862119879119881

for current and voltage respectively while (4)and (5) give the solar irradiation gain parameters119862

119878119868and119862

119878119881

for current and voltageThese gain parameters are used in (7)and (8) in order to update the values of PV cell voltage andcurrent to the new cell voltage and new photocurrent undernew cell temperature119879

119909and solar irradiation 119878

119909 Temperature

constants (120573119879and 120574119879) and solar irradiation constants (120572

119878 120573119878

and 120574119878) are obtained experimentally for the PV cell type used

in both simulation and test studies Since the PV cell typesand the properties are assumed to be the same in a PV panelthese constants may also be obtained by running the similartests with the PV panels used After recording the referencevalues at known temperature and solar irradiation levels thesame quantities are measured at another known temperatureand solar irradiation level and the differences are comparedThe same test may be repeated for a few times at differenttemperature and solar irradiation levels for better correlationThen (2) to (8) are used backwards to obtain the constantparameters used in these formulas After this process theobtained constant can only be used for the models of the PVcells that are used in the testing

As shown in Figure 3 panel configuration and cell tem-perature are given as the constant inputs besides the fourdifferent solar irradiation values After running the modulethe students get four different normalized I-V andP-V curvesone for each solar irradiation The module also returns themaximum power point operating power voltage and currentvalues 119875max 119868mpp and119881mpp for all four solar irradiation levels

Figure 3 The photovoltaic array solar irradiation interface

It is clear that this module gives a good understanding andinterfacing to the student to analyze the effects of variablesolar irradiation levels on PV array characteristics

23 Module 3 Temperature Effects on PV Array Character-istics This module uses the same equation set (2) to (8)given in previous section about Module 2 The studentsare able to configure the array and analyze the effects ofvariable temperature on PV array characteristics while thesolar irradiation is kept constant Figure 4 shows the userinterface screen for Module 3 After running the module thestudents get four different normalized I-V and P-V curvesone for each temperature level The module also returns themaximum power point operating power voltage and currentvalues 119875max 119868mpp and 119881mpp for all four temperature levelsIt is clear that this module gives a good understanding andinterfacing to the students to analyze the effects of variabletemperature on PV array characteristics

24 Module 4 PMDC Motor Load Fed from PV Array Oncethe students learn and understand the PV array character-istics under different environmental conditions they cancontinue the simulation based experiments by starting tofeed the loads The first load in VPVSL is a permanentmagnet DC (PMDC) motor This scheme includes a PVarray with the required voltage and power configurationa filter circuit to eliminate the discontinuities in currenta PID controlled DCDC chopper a PID controller and aPMDC motor Figure 5 shows the user interfacing windowsand the connection diagramof the PVpoweredPMDCmotorscheme The user interface scheme has PV array data motordata and PID controller data as three input areas and fourgraphic windows as the output areas Time response of thePMDCmotor speed PMDCmotor voltage and PV array I-Vand P-V characteristics under loading conditions are plottedin the output section Since a maximum power point tracker(MPPT) is not added to the scheme the operating powermay not be at its maximum value and may be changed as thereference voltage is changed by the user An MPPT may be


Figure 4 The photovoltaic array temperature interface

considered to be added to the scheme for a more completesimulation model

This module gives the students an opportunity of under-standing DC load operation and control A PID controller isemployed to operate the PMDC motor at a constant voltagewhile the temperature and solar irradiation levels of the PVarray changeThemodule can be simulated without and withthe controller to see the effects of the changes in sunlightlevel In order to have a constant motor speed the voltageto the motor must be kept constant Therefore the constantspeed means constant DC voltage which may be assumed asa DC bus used to collect the DC power from different sourcesin distributed generation networks Therefore this modulealso teaches the common DC bus concept with the controlin practice

25Module 5 AC Loads Fed from PVArray TheDC chopperin previous module is replaced by a three-phase inverterin order to obtain a three-phase AC voltage with nominalvoltage and constant frequency As shown in user interfacescreen ofModule 5 in Figure 6 thismodule includes six inputareas which are used for panel configuration RLC AC loadfilter parameters RL DC load controller parameters and athree-phase isolation transformer

There are six output plots in the user window Thestudents are able to plot and observe the voltage at the outputterminals of the inverter the voltage of the PV array underload I-V characteristics of the array under loading the ACvoltage at load terminals PV array current under load andP-V characteristics of the panel under loading conditions ADC load is connected before the inverter in order to initiatethe operation of the PV array

3 Simulation Results

This section gives the results from the virtual experimentsdone to test the proposed VPVSL Five modules describedabove are tested in given order First the PV array charac-teristics then the effects of the solar irradiation the effects

Figure 5 The PV powered PMDCmotor interface

Figure 6 The PV powered AC loads interface

of temperature DC load operation and finally the AC loadoperation are tested

31 PV Array Characteristics The GUI screen given inFigure 2 forModule 1 is used to perform the tests for PV arraycharacteristicsThe system is simulated for eight different PVarray configurations under reference temperature and solarirradiation levels which are 20∘C and 100mWcm2The arrayconfigurations and corresponding maximum power pointoperating values are given in Table 1 for a sample case studyThese sets of experiments give the students an insight look atthe design of PV systems in terms of sizing

Results from the sample case study are depicted inFigure 7 The size of the PV array is changed by differentcombinations of series and parallel connected PV panelsand the resultant operating parameters are recorded fromeach case The sizing effect on open circuit voltage (119881oc)short circuit current (119868sc)maximumpower (119875max)maximumpower point current (119868mpp) and maximum power point


120ndash140100ndash12080ndash10060ndash80

40ndash6020ndash400ndash20

1

2

34

020406080

100120140

12

34

Ope

n ci

rcui

t vol

tage

(V)

NP

NS

(a) Open circuit voltage surface


5ndash100ndash5

1

2

34

0

5

10

15

20

25

12

34

Shor

t circ

uit c

urre

nt (A

)

NP

NS

(b) Short circuit current surface

1400ndash16000012 ndash1400

1000ndash1200800ndash1000


1

2

34

0200400600800

1000120014001600

12

34

Max

imum

pow

er (W

)

NP

NS

(c) Maximum power surface

16ndash1814ndash1612ndash1410ndash128ndash10


1

2

34

02468

1012141618

12

34

Max

imum

pow

erpo

int c

urre

nt (A

)

NP

NS

(d) Maximum power point current surface



1

2

34

0102030405060708090

100

12

34

Max

imum

pow

erpo

int v

olta

ge (V

)

NP

NS

(e) Maximum power point voltage surface

Figure 7The sizing effects on the array characteristics at reference temperature and solar irradiation levels (a) Open circuit voltage surface(b) short circuit current surface (c) maximum power surface (d) maximum power point current surface and (e) maximum power pointvoltage surface


Table 1 The PV array characteristics experiment parameters and outputs

Case I II III IV V VI VII VIII119873119878

1000 2000 3000 4000 1000 1000 1000 2000119873119875

1000 1000 1000 1000 2000 3000 4000 2000119875max (W) 90147 18090 27138 36100 18080 27152 36248 36169119868mpp (A) 4210 4043 4121 4186 8077 12037 16533 8134119881mpp (V) 21416 44741 65803 86234 22384 22558 21925 44465119868sc (A) 4998 5000 5000 5000 9996 14992 19985 9999119881oc (V) 30561 61123 91664 12225 30561 30561 30561 61123Case IX X XI XII XIII XIV XV XVI119873119878

2000 2000 3000 3000 3000 4000 4000 4000119873119875

3000 4000 2000 3000 4000 2000 3000 4000119875max (W) 54205 72339 54161 81381 108550 72372 10851 14467119868mpp (A) 11914 16262 8364 12233 16550 8118 12042 16332119881mpp (V) 45496 44483 64751 66524 65592 89146 90109 88582119868sc (A) 14998 19998 10002 15002 20001 10001 15001 20003119881oc (V) 61123 61123 91684 91684 91684 12225 12225 12225

Table 2 Maximum power point surface 119875max (W) at a fixedconfiguration

119873119904= 2 119879

119909

119873119901= 2 minus20 0 20 40

119878119909

80 316 298 280 261100 410 386 362 337120 512 481 451 418140 618 581 543 505160 733 688 642 595180 850 797 745 691200 977 915 853 790

voltage (119881mpp) are shown in Figure 7 by surface plots wheresolar irradiation and temperature levels are kept constantThesurfaces in Figure 7 are called the Design Surfaces and usedas the base load power matching in design process Thesedesign surfaces can be extended to include the effects of thechanges in solar irradiation and temperature levelsThereforethese surfaces can be converted into dynamic design surfacesso that both the maximum power operating point values canbe tracked when the environmental conditions are changedand the configuration switching for load power matching canbe done

32 Solar Irradiation and Temperature Effects Simulationresults of Modules 2 and 3 of VPVSL are discussed in thissection The students get the I-V and P-V curves of the PVarray at different solar irradiation and temperature levels fora fixed configuration by keeping 119873

119878and 119873

119875unchanged For

the sample case study the array configuration is set to119873119878= 2

and119873119875= 2

First operating temperature is assumed to be constant inModule 2 and only the solar irradiation is changedThereforetests from Module 2 give the characteristics for changing

sunlight levels at constant temperature as given in Figure 8Solar irradiation is assumed to be constant in Module 3and only the temperature is changed to see the effects oftemperature on the PV array characteristicsThe results fromModule 3 are given in Figure 9

Both Module 2 and Module 3 show that the generatedpower from the PV array changes when temperature andsolar irradiation levels changeTherefore the recordings fromthese two modules can be combined to yield the maximumpower generated by the PV array at different pairs of solarirradiation and temperature as given in Table 2

A similar table can also be obtained easily for maximumpower point current and voltages However for the sake ofusing optimum number of pages these tables are not givenhere The plots of the maximum power points maximumpower point operating currents and maximum power pointoperating voltages are given in Figure 10 instead

The results from Module 2 and Module 3 are evaluatedtogether in order to see the solar irradiation and tem-perature effects at the same time The combination of theresults from these two modules enables the user to obtainmaximum power points maximum power point operatingcurrents and maximum power point operating voltages atdifferent solar irradiation and temperature levels Three-dimensional surface plots of these maximum power pointquantities are shown in Figure 10 Actually these surfaces areused as dynamic design surfaces as expressed before Thesesurfaces are called the dynamic design surfaces because themaximum power operating points move on these surfaceswhen the solar irradiation and temperature levels change asunpredictable and uncontrolled input variables The usersof the VPVSL or the students can use these surfaces asdesign surfaces within an interval of solar irradiation andtemperature universes The array configuration is used toexpand the power output range while the maximum powerpoint surfaces are used to track and operate the PV array atits maximum power output [25]


Table 3 The PV powered PMDC motor GUI interface parameters

PV Array and PID controller PMDCMotorNumber of solar cells connected in series 119873

11990410 Resistance 119877

11989815Ohm

Number of solar cells connected in parallel 119873119901

10 Inductance 119871119898

000805HAmbient temperature 119879 20∘C Back emf constant 119870

119898008Vsdotsrad

Ambient solar irradiation () 119878 100 Viscous friction constant 119861119898

7 times 10minus4Nmsrad

Proportional constant 119870119901

06 Rotor moment of inertia constant 119869119898

5 times 10minus4 kgsdotm2

Integral constant 119870119894

0 Actual rated speed 119908119899

1265 radsDerivative constant 119870

1198890 Voltage source 119881

119898300V

Table 4 The PV powered AC load experiment parameters

PV array AC loadNumber of solar cells connected in series 119873

1199048 Frequency 119891

11989960Hz


1 Nominal phase-phase voltage 119881119899

208VAmbient temperature 119879 16∘C Active power 119875 500WAmbient solar irradiation () 119878 103 Inductive reactive power 119876

119871200VAR

Capacitive reactive power 119876119862

500VARController DC Load


1 Resistance 119877 40OhmIntegral constant 119870

1198940 Inductance 119871 00002H

Filter Three-phase transformerResistance 119877

119891005Ohm Nominal power 119878 1000VA

Inductance 119871119891

005H Primary winding phase-phase voltage (rms) 1198811

208VCapacitance 119862

11989120 times 10

minus6 F Secondary winding phase-phase voltage (rms) 1198812

208V

33 PV Array with a PMDC Motor Load The VPVSL istested under two loading conditions First a PMDC motorload is fed from the PV system A PMDC motor is selectedas the DC load in order to give an example of stand-aloneDC load feeding A controlled DC-DC chopper is used tocontrol the voltage to DC loads DC chopper is controlledfor two different purposes which are constant DC voltageoperation and constant motor speed operation The first oneis to obtain a constant DC voltage at the output terminals ofthe DC-DC chopper in order to have a common DC busThis operating condition gives the student the importanceof common DC bus in distributed generation systems Thesecond purpose of controlling DC-DC chopper is for testingthe system control tomeet the load operating conditionsThemotor is operated at a constant speed without considering aconstantDCvoltageThepurpose here is to operate themotorat constant speed whatever the voltage from PV array to DC-DC chopper is Therefore the second operating case showsthe user how to eliminate the effects of the changes in solarirradiation and temperature levels on the load performancewhile the first case shows how the same effects are eliminatedfor the common DC bus voltage

TheGUI screen shown in Figure 5 is usedwith the samplecase parameters given in Table 3 The connection diagramin Figure 5 includes a reverse current blocking diode (D)a current smoothing input filter (119877

119891and 119871

119891) and a storage

capacitor (119862119891) to overcome voltage discontinuities A PID

controlled type-D DC-DC chopper is used to adjust thevoltage at load terminals

The PID controller used in VPVSL generates the controlsignal which is sent to a pulse width modulator (PWM) togenerate the required switching pulses For the two-quadrantDC-DC chopper four pulses are generated

The system outputs which are observed are shown inFigures 11 12 13 14 15 and 16 Figure 11 shows the controlledoutput voltage of DC-DC chopper This voltage is keptconstant considering that it is a voltage of common DC buswhich acts as a station to connect different power generatingunits together in a distributed power system The speedvariation of the PMDC motor load is given in Figure 12 Forthis operating case the DC chopper is controlled to keep thespeed constant I-V and P-V characteristics of the PV systemduring the constant DC bus voltage operating condition areshown in Figures 13 and 14 respectively Since there is noMPP tracking controller in the system the operating pointon I-V and P-V curves swings along the characteristics untilit settles down at operating values that match the load sidevalues The operating voltage and power values can be seenclearly in Figures 15 and 16 in which the time responses ofthe PV array voltage and power are shown

34The PV Powered AC Load The second loading conditionof the VPVSL is a three-phase AC load The connectiondiagram given in Figure 6 is used for the AC load operationThe connection scheme includes reverse current blockingdiode (D) and input filter (119877

119891 119871119891 and 119862

119891) for input current

smoothing and input voltage continuity a DC R-L load forthe system initiation a PI controlled three-phase inverter an


Table 5 Assessment data student evaluations (Section-A)

(1) When I compare this experiment with other experiments belonging to this laboratory this experiment isVery easy (1) Easy (2) Reasonable (3) Difficult (4) Very difficult (5)

(2) When I compare this experiment with other experiments belonging to this laboratory workload of this experiment isToo much (1) Much (2) Reasonable (3) Few (4) Too few (5)

(3) Experiment realization speed isVery fast (1) Fast (2) Reasonable (3) Slow (4) Very slow (5)

(4) This experiment is completelyVery good (1) Good (2) Reasonable (3) Inadequate (4) Highly inadequate (5)

(5) This experiment instructor is completelyVery good (1) Good (2) Reasonable (3) Inadequate (4) Highly inadequate (5)

Table 6 Assessment data student evaluations using a Likert-scale (Section-B)

6 I understood the main idea7 Experiment coordination was very poor8 The experiment was attractive9 Experiment sheet (laboratory handout) was well arranged10 I learned something that I will remember future11 Experiment sheet helped me to comprehend orders effectively12 Using MatlabSimulink GUI environment was good13 Using MatlabSimulink GUI environment helped me to interpret the experiment results14 Using the model development of the system for MatlabSimulink GUI environment enhanced my knowledge and skills

15 I find the model development of the system for MatlabSimulink GUI environment useful for learning related issue incorresponding course

16 Using the model development of the system for MatlabSimulink GUI environment make me be more interested in this subject17 Experiment goal was clear and understandable

18 I was able to fully use the model development of the system for MatlabSimulink GUI environment by following the instructionsprovided

19 It was difficult to gather from this experiment20 Instructor showed effective control and guidance during the experiment21 How experiment will be done was explicitly explained in experiment sheet22 It was easy to attend the order since experiment sheet was well organized23 I will need more information to prepared experiment reportNote strongly agree 1 agree 2 neither agree or disagree 3 disagree 4 strongly disagree 5

isolation transformer for filtering and isolation purposes anda three-phase AC RLC load The input parameters used inFigure 6 for the sample study case are given in Table 4

As mentioned earlier the students or the users are ableto plot and observe the voltage at the output terminals ofthe inverter the voltage of the PV array under load I-Vcharacteristics of the array under loading the AC voltageat load terminals PV array current under load and P-Vcharacteristics of the panel under loading conditions Thecontroller gains (119870

119875 119870119868) can be adjusted to provide the

AC load demands (required voltage and frequency withminimum disturbances) PV array characteristics can also beobserved from this operating processThe virtual experimentoutputs are shown in Figures 17 18 19 21 and 22 Figures 17and 18 show the time variations of the PV array voltage andcurrent respectively These two figures show the steady-state

operating voltage and current of the array The array voltagegiven in Figure 17 is the DC voltage converted to three-phaseAC voltage by the inverter

The output voltage of the inverter is given in Figure 19Since the voltage at the output terminals of the inverter isnot pure AC an isolation transformer is used as a filteringdevice to minimize the voltage harmonics so that the ACvoltage waveform shown in Figure 20 is obtained with lessharmonics

I-V and P-V characteristics of the PV array are shown inFigures 21 and 22 After the initial swinging of the operatingpoint on the curves it settles down to steady-state operatingvalues as depicted in Figures 17 and 18 The operating pointsof the PV array on I-V and P-V characteristics in Figures 21and 22 are marked using the information from Figures 17 and18


0 5 10 15 20 25 30 350

1

2

3

4

5

6

Voltage (V)

Curr

ent (

A)

Sa

Sb

Sc

Sd

(a) I-V characteristics

0 5 10 15 20 25 30 350

20

40

60

80

100

120

Voltage (V)

Pow

er (W

)

Sa

Sb

Sc

Sd

(b) P-V characteristics

Figure 8 The effect of solar irradiation on I-V and P-V characteristics of the PV array

0 5 10 15 20 25 30 350

1

2

3

4

5

Voltage (V)

Curr

ent (

A)

Ta

Tb

Tc

Td


0 5 10 15 20 25 30 350

20

40

60

80

100

Voltage (V)

Pow

er (W

)

Ta

Tb

Tc

Td


Figure 9 The effect of temperature on I-V and P-V characteristics of the PV array

4 Student Assessment and Evaluation

The effectiveness of learning based on the GUI environmentis determined by experiment report and questionnaire Theexperiment report is based on the GUI interface explainedthoroughly in the experiment sheet (laboratory handout) Aquestionnaire is an important source to observe the studentsrsquoreactions Individual students share their perceptions of theexperiment Thus the students were asked to fill in a ques-tionnaire about the experiment characteristics of the envi-ronment and opinion about instructor during experimentThe students can specify the pros and cons of the environment

and experiment The questionnaire consisted of thirty-threequestions as shown in Tables 5 6 and 7 The questionnaireconsists of three sections (Section-A Section-B and Section-C) We used Likert scales most commonly applied ratingscales in spite of some limitations [26] General opinionabout experiment and instructor are roughly researched inSection-A The studentsrsquo observations and comments abouteffectiveness of the experiment and instructor are deeplytaken on in Section-B and Section-C respectively

The students gave a grade between 5 (strongly disagree)and 1 (strongly agree) based on Likert Scale The studentsrsquoobservations and comments given in Tables 5ndash7 are evaluated


80

100120

140160

180200

0100200300400500600700800900

1000

020

40

Max

imum

pow

er (W

)

minus20

900ndash1000800ndash900700ndash800600ndash

ndash700

500 600


Tx

S x

(a) Maximum power point surface

80

100120

140160

180200

02468

101214161820

020

40

Max

imum

pow

er p

oint

curr

ent (

A)

minus20


ndash14

10 12


Tx

S x

(b) Maximum power point operating current surface

80

100120

140160

180200

0

10

20

30

40

50

60

020

40

Max

imum

pow

er p

oint

vol

tage

(V)

minus20



Tx

S x

(c) Maximum power point operating voltage surface

Figure 10 Maximum power point surfaces under variable solar irradiation and temperature levels with fixed sizing (a) Maximum powerpoint surface (b) maximum power point operating current surface and (c) maximum power point operating voltage surface

and their success based on their laboratory reports due aweek after the completion of the laboratory is describedin percentages for each group shown in Figure 23 Thereports were marked according to experiment sheet (hand-out) explained above The global result gotten from thequestionnaire (average scores of each question) is shown inFigures 24 25 and 26 and Table 8

5 Conclusion

Design and utilization of a Virtual Photovoltaic SystemsLaboratory for engineering undergraduate curriculum areintroduced in this paper The Virtual Photovoltaic SystemLaboratory described in this study is developed to teachstudents the basics and design steps of photovoltaic solar


Table 7 Assessment data student evaluations using a Likert-scale(Section-C)

24 Instructor was powerful communicator25 Instructor was eager to teach26 Instructor performance interacted with me27 Instructor explanation was explicit and fluent28 Instructor explanation complicated taking notes29 Instructor enabled me to concentrate more30 Instructor behavior was friendly and pleasant31 Instructor was well prepared32 Instructor was self-confident33 I do not ask for instructor help for other thingsNote strongly agree 1 agree 2 neither agree nor disagree 3 disagree 4strongly disagree 5

0 1 2 3 4 50

50

DC

mot

or lo

ad v

olta

ge (V

)

100

150

200

250

300

Time (s)

VrefV

Figure 11 DC motor load voltage waveform

energy systems in a virtual environment before entering thefield Proposed VPVSL offers five modules to the users Thefirst module can be used to analyze the I-V and P-V charac-teristics of PV systems based on sizing solar irradiation andtemperature conditions Module 1 provides a GUI windowto the user to change solar irradiation level temperatureand the sizing parameters so that both design and theeffects of the changing environmental conditions are set andanalyzed Module 2 can be used to analyze the effect of solarirradiation level under constant temperature and fixed sizingconditions The effects of the changes in temperature can beanalyzed usingModule 3 under constant solar irradiation andfixed sizing conditions The proposed system is tested undertwo loading conditions Module 4 provides the testing GUIwindow for a PMDCmotor load APID controller is includedin this module along with a DC-DC chopper in order to havea constant DC voltage to be treated as a common DC busat which a PMDC motor is connected Module 5 is used totest the VPVSL for AC loading A three-phase inverter and

0 1 2 3 4 50

500

1000

1500

2000

2500

3000

Time (s)

Spee

d (r

ads

)

Figure 12 Speed-time waveform

0 50 100 150 200 250 300 3500

10

20

30

40

50

Voltage (V)

Curr

ent (

A)

Figure 13 PV current-voltage waveform

0 50 100 150 200 250 300 3500

2

4

6

8

10

Voltage (V)

Pow

er (W

)

times103

Figure 14 PV power-voltage waveform


0 1 2 3 4 50

50

100

150

200

250

300

350

Time (s)

Volta

ge (V

)

Figure 15 Variation of PV array voltage in time domain

0 1 2 3 4 5

0

2

4

6

8

10

Time (s)

Pow

er (W

)

times103

minus2

Figure 16 Variation of PV array power in time domain

0 002 004 006 008 01

140

160

180

200

220

240

Time (s)

PV v

olta

ge (V

)

Figure 17 Load PV voltage waveform

0 002 004 006 008 010

1

2

3

4

5

Time (s)

PV cu

rren

t (A

)

Figure 18 PV current waveform

0 002 004 006 008 01

0

50

100

150

200

Time (s)

Vab

inve

rter

(V)

minus50

minus100

minus150

minus200

Figure 19 Inverter voltage waveform

0 002 004 006 008 01

0

50

100

150

Time (s)

Vab

load

(V)

minus50

minus100

minus150

Figure 20 AC load voltage waveform


120 140 160 180 200 220 240 2600

1

2

3

4

5

PV voltage (V)

PV cu

rren

t (A

)


120 140 160 180 200 220 240 2600

100

200

300

400

500

600

700

800

PV voltage (V)

PV p

ower

(W)


83 8393

7583 83

90 88

60

88

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10

Succ

ess (

)

Group

83 83757

83 8390 88

60

88

Figure 23 The group success in experiment in percentages

254

380

290230

160

0

1

2

3

4

5

1 2 3 4 5

Aver

age

Question

254290

2301601 60

Figure 24 Summary of responses to student opinion survey(Section-A)

Table 8 Student evaluation results

Question SD () D () N () A () SA ()6 2 2 6 46 447 28 34 22 12 48 2 4 18 44 329 0 10 14 33 4310 0 6 4 56 3411 0 6 24 38 3212 2 4 6 32 5613 0 6 28 38 2814 0 12 28 32 2815 2 12 32 36 1816 0 8 24 34 3417 2 2 12 32 5218 8 12 14 40 2619 12 42 16 20 1020 0 6 8 34 5221 2 8 6 37 4722 2 4 10 40 4423 4 28 26 22 2024 0 0 2 37 6025 2 0 12 30 5626 2 0 14 42 4227 0 0 6 28 6628 32 38 22 2 629 0 6 16 42 3630 2 0 0 18 8031 0 0 6 28 6632 2 0 4 20 7433 78 16 2 2 2SD strongly disagree D disagree N neutral A agree SA strongly agree

a three-phase transformer are included in this module tosupply power to an RLC load

Virtual test results for sample cases are obtained anddiscussed in the paper It has been shown that the proposedVPVSL can be used in designing and analyzing purposesThe students can analyze and understand the operationalcharacteristics of the PV systems as well as being able to stepin to the design stage of the PV systemsTheproposedVPVSL


172

370

200 192 182 204164

212 224 244206

170236

326

168 182 180

274

0

1

2

3

4

5

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Aver

age

Question

7272 200 192 182182 2041641 64

212 224 244206

17070236

326

1686 182182 18080

272

Figure 25 Summary of responses to student opinion survey (Section-B)

142 162 178140

388

192126 140 136

466

0

1

2

3

4

5

24 25 26 27 28 29 30 31 32 33

Aver

age

Question

142 1626 1781 78140

388

192126 140 136

Figure 26 Summary of responses to student opinion survey(Section-C)

has been used in Power Systems Lab in the Department ofElectrical and Electronics Engineering at Karadeniz Tech-nical University as a part of undergraduate curriculum Asurvey on the students who took the lab has been carried outand responses are included in this paperThe survey indicatesthat the students get benefit of using the lab They found itusable easy and understandable

A maximum power point tracker a battery chargingunit and power conditioning filter circuit options can beadded to this study to expand the scope of the virtual labA wind energy conversion module a fuel cell module anda small hydro module can also be added to this systemto develop a general virtual renewable energy laboratoryBesides the proposed VPVSL can be expanded to handlereal time experiments remotely However if the VPVSL iscombined with a real time measurement and experimentalsystem some additional components and software will beneeded In order to perform the real time experimentsremotely a communication link and a device onoff switchingcontrol software are required In this case the connectionand data transferring speed will be important as well as thebandwidth of the communication channels to handle thedensity of coinciding user demands Due to large size andadditional research in a different area such as communicationand networking the idea of establishing a remote VPVSL isleft for the future work and is not studied in this paper

Appendix

The Photovoltaic Cell Parameters

119879119862(∘C) = 20119878119862() = 100120573119879= 0004

120574119879= 006

119879119886(∘C) = 20120572119878= 02

119860 = 62

119896 (JKminus1) = 13806488 times 10minus23

119890 (coulombs) = 1603 times 10minus19

1198680(A) = 001119877119878(Ω) = 002

Nomenclature

PV PhotovoltaicGUI Graphical user interfaceVPVSL Virtual photovoltaic systems labPMDC Permanent magnet DCMPPT Maximum power point tracker119881119862 Solar cell output voltage119860 Curve fitting constant used in solar cell I-V

characteristics119896 Boltzmann constant119879119862 Reference solar cell operating temperature119890 Electron charge119868ph Solar cell photocurrent1198680 Diode reverse saturation current119868119862 Solar cell output current119877119878 Series resistance of solar cell equivalent

circuit119873119878 Number of cells in series119873119875 Number of cells in parallel119862119879119881

Temperature dependent scaling factor ofsolar cell voltage

120573119879 Temperature constant affecting solar cell

voltage


119879119886 Reference ambient temperature119879119909 Current values of ambient temperature119862119879119868 Temperature dependent scaling factor of

solar cell current120574119879 Temperature constant affecting solar cell

current119878119862 Reference solar irradiation120572119878 Solar irradiation constant119878119909 Current values of solar irradiation119878119862 Reference solar irradiation120573119878 Effect of the solar irradiation on cell voltage119862119878119881 Solar irradiation dependent scaling factor of

solar cell voltage119862119878119868 Solar irradiation dependent scaling factor of

solar cell current120574119878 Effect of the solar irradiation on

photocurrentΔ119879119862 Temperature change due to changing solarirradiation level

119881119862119883

Solar cell operating voltage119868ph 119909 Solar cell operating photocurrent119875max Maximum power photovoltaic array119868mpp Maximum power point current photovoltaic

array119881mpp Maximum power point voltage photovoltaic

array119868sc Short circuit current photovoltaic array119881oc Open circuit voltage photovoltaic array



References

[1] M Hashemipour H F Manesh andM Bal ldquoAmodular virtualreality system for engineering laboratory educationrdquo ComputerApplications in Engineering Education vol 19 no 2 pp 305ndash3142011

[2] B Balamuralithara and P C Woods ldquoVirtual laboratories inengineering education the simulation lab and remote labrdquoComputer Applications in Engineering Education vol 17 no 1pp 108ndash118 2009

[3] J A Kypuros and T J Connolly ldquoStudent-configurable web-accessible virtual systems for system dynamics and controlscoursesrdquo Computer Applications in Engineering Education vol16 no 2 pp 92ndash104 2008

[4] R Dormido H Vargas N Duro et al ldquoDevelopment of a web-based control laboratory for automation technicians the three-tank systemrdquo IEEE Transactions on Education vol 51 no 1 pp35ndash44 2008

[5] E Tanyildizi and A Orhan ldquoA virtual electric machine lab-oratory for effect of saturation of the asynchronous machineapplicationrdquo Computer Applications in Engineering Educationvol 17 no 4 pp 422ndash428 2009

[6] E Lindsay and M Good ldquoThe impact of audiovisual feedbackon the learning outcomes of a remote and virtual laboratoryclassrdquo IEEE Transactions on Education vol 52 no 4 pp 491ndash502 2009

[7] H Rehman R A Said andYAl-Assaf ldquoAn integrated approachfor strategic development of engineering curricula focus onstudentsrsquo design skillsrdquo IEEE Transactions on Education vol 52no 4 pp 470ndash481 2009

[8] G C Goodwin A M Medioli W Sher L B Vlacic andJ S Welsh ldquoEmulation-based virtual laboratories a low-costalternative to physical experiments in control engineeringeducationrdquo IEEE Transactions on Education vol 54 no 1 pp48ndash55 2011

[9] S Azaklar and H Korkmaz ldquoA remotely accessible andconfigurable electronics laboratory implementation by usingLabVIEWrdquo Computer Applications in Engineering Educationvol 18 no 4 pp 709ndash720 2010

[10] C Elmas and Y Sonmez ldquoAn educational tool for powerelectronics circuitsrdquoComputer Applications in Engineering Edu-cation vol 18 no 1 pp 157ndash165 2010

[11] A See and A J Leong ldquoAn advanced-automated system forequipotential field lines mapping utilizing motion controlrdquoComputer Applications in Engineering Education vol 19 no 1pp 171ndash182 2011

[12] H Y Yamin I A Altawil A F Al-Ajlouni and A S Al-Fahoum ldquoA new developed educational approach to improveconventional teaching methodology of the power electronicslaboratoryrdquo Computer Applications in Engineering Educationvol 19 no 1 pp 193ndash200 2011

[13] A F Al-Ajlouni H Y Yamin and B Harb ldquoEffect of instruc-tional software on achievement of undergraduate students inDigital Signal Processingrdquo Computer Applications in Engineer-ing Education vol 18 no 4 pp 703ndash708 2010

[14] E S Aziz S K Esche and C Chassapis ldquoContent-richinteractive online laboratory systemsrdquoComputer Applications inEngineering Education vol 17 no 1 pp 61ndash79 2009

[15] T Yigit and C Elmas ldquoAn educational tool for controlling ofSRMrdquo Computer Applications in Engineering Education vol 16no 4 pp 268ndash279 2008

[16] J R R Ruiz A G Espinosa and J A Ortega ldquoValidation of theparametric model of a DC contactor using Matlab-SimulinkrdquoComputer Applications in Engineering Education vol 19 no 2pp 337ndash346 2011

[17] J A Calvo M J Boada V Dıaz and E Olmeda ldquoSIM-PERF SIMULINK based educational software for vehiclersquosperformance estimationrdquoComputer Applications in EngineeringEducation vol 17 no 2 pp 139ndash147 2009

[18] I Colak S Demirbas S Sagiroglu and E Irmak ldquoA novelweb-based laboratory for DC motor experimentsrdquo ComputerApplications in Engineering Education vol 19 no 1 pp 125ndash1352011

[19] J R R Ruiz A G Espinosa and X A Morera ldquoElectric fieldeffects of bundle and stranded conductors in overhead powerlinesrdquo Computer Applications in Engineering Education vol 19no 1 pp 107ndash114 2011

[20] E Aziz ldquoTeaching and learning enhancement in undergradu-ate machine dynamicsrdquo Computer Applications in EngineeringEducation vol 19 no 2 pp 244ndash255 2011

[21] I H Altas and A M Sharaf ldquoA photovoltaic array simulationmodel formatlab-simulinkGUI environmentrdquo inProceedings ofthe International Conference on Clean Electrical Power (ICCEPrsquo07) pp 341ndash345 Capri Italy May 2007

[22] Criteria for Accrediting Engineering Programs Effective forReviews during the 2012ndash2013 Accreditation Cycle ABETEngineering Accreditation Commission 2011


[23] M Buresch Photovoltaic Energy Systems Design and Installa-tion McGraw-Hill New York NY USA 1983

[24] I H Altas and A M Sharaf ldquoA fuzzy logic power tracking con-troller for a photovoltaic energy conversion schemerdquo ElectricPower Systems Research vol 25 no 3 pp 227ndash238 1992

[25] I H Altas and A M Sharaf ldquoA novel on-line MPP search algo-rithm for PV arraysrdquo IEEE Transactions on Energy Conversionvol 11 no 4 pp 748ndash754 1996

[26] D R Hodge and D F Gillespie ldquoPhrase completion scales abetter measurement approach than Likert scalesrdquo Journal ofSocial Service Research vol 33 no 4 pp 1ndash12 2007

Research ArticleEffect of Ambient Temperature on Performance ofGrid-Connected Inverter Installed in Thailand

Kamonpan Chumpolrat Vichit Sangsuwan Nuttakarn UdomdachanutSongkiate Kittisontirak Sasiwimon Songtrai Perawut ChinnavornrungseeAmornrat Limmanee Jaran Sritharathikhun and Kobsak Sriprapha

National Electronics and Computer Technology Center 112 Thailand Science Park Phahonyotin Road Klong 1 Klong LuangPathumthani 12120 Thailand

Correspondence should be addressed to Kamonpan Chumpolrat kamonpanchumpolratnectecorth

Received 28 October 2013 Revised 7 January 2014 Accepted 4 February 2014 Published 6 March 2014


Copyright copy 2014 Kamonpan Chumpolrat et alThis is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in anymedium provided the originalwork is properly cited

The effects of temperature on performance of a grid-connected inverter and also on a photovoltaic (PV) system installed inThailandhave been investigated It was found that the maximum efficiency of the inverter showed 25 drop when ambient temperature wasabove 37∘C The inverter performed efficiently in November and December the months of high irradiance and monthly averageambient temperature of lower than 35∘C allowing relatively high system performance ratio in this period Our results show thathigh temperature provides negative impacts not only on the PV modules but also on the performance of the inverter Thus theeffect of temperature on the inverter efficiency should be taken into account when predicting energy yield or analyzing losses ofthe PV systemsmdashespecially in high temperature regions

1 Introduction

Thailand receives an annual average solar irradiation of182MJm2mdashday which is relatively high compared to othertropical and mid-latitude counties [1] Owing to abundantsolar energy and government support scheme for megawattssolar farms Thailand is expected to be an emerging photo-voltaic (PV) market of Southeast Asia Nonetheless severeclimatic conditions of tropical countriesmdashhigh temperatureand high humiditymdashhave negative impacts on performanceand reliability of PV systems Since high temperature causesa reduction in output power of PV modules the temperatureeffects are one of main concerns when forecasting energyproduction or analyzing losses of the PV systems There arethus many reports regarding the effects of temperature onthe performance of various types of PV modules operatingin tropical countries including Thailand [2ndash5] Besides theperformance of the PV modules inverter efficiency is alsoa critical factor which greatly influences the system per-formance therefore its actual behavior needs to be evalu-ated Furthermore temperature-dependent performance ofinverter is also worth investigating because the efficiency

of electronic devices including inverter also depends onthe operating temperature [6ndash9] Since the temperature-dependent behavior of the inverter for PV systems has not yetbeen reported in this studywehave investigated performanceof a high-efficient grid-connected inverter installed in Thai-land in particular with respect to the temperature effect Thefindings of our work are expected to be useful information forenergy yield prediction and loss analysis of the PV systemsespecially in high temperature regions

2 System Installation and Monitoring

The 224 kWp grid-connected PV system has been installedat National Science and Technology Development Agency(NSTDA) Pathumthani province Thailandmdashlatitude14∘ 41015840 4610158401015840N longitude 100∘ 361015840 410158401015840 Emdashin March 2010supplying electricity for load of the office building This PVsystem consists of tandem amorphous silicon (a-SiHa-SiH) prototype modules manufactured by NSTDA Thecharacteristics of the PV modules inverter and the PVsystem in this study are described in Table 1The PVmoduleshave been installed in an open rack at a tilt of 14∘ facing



Table 1 PV module and system characteristics

Item Details

Module

Type Thin film a-SiPeak power output 40WPeak power voltage 475 VPeak power current 084A

Temperature coefficient for power minus020∘CArea 12m times 065m (078m2)

InverterType Single-phase string transformer based grid-connected

Nominal output power 5 kWMaximum efficiency 96

System

Number of modules 56 (8 modulesstring 7 strings)Nominal power output 224 kWPeak power voltage 380VPeak power current 589A

Figure 1 PV test site at NSTDA Thailand

South as shown in Figure 1 An automatedmeasuring systemwas constructed to collect data of the system that is systemoutput module temperature (119879

119898) ambient temperature (119879

119886)

and in-plane solar irradiance at 1min interval The collecteddata from 5 AM to 7 PM during the period of 7 months(from October 2010 to April 2011) was used for the systemperformance analysis in this study


The DC output power as a function of the solar intensityand a variation of the module temperature are shown inFigure 2 It can be seen that the module temperature wasabout 30∘C under the irradiance of lower than 250Wm2 andthen gradually increased with increasing irradiance risingto 60∘C at the irradiance of 800ndash1000Wm2 It should benoted that in this test site the average module temperatureranged from 42∘C to 47∘C rarely being 25∘C or lower duringthe operating hours Figure 3 indicates a linear relationshipbetween the DC output from the PVmodules (inverter inputpower) and the AC output from the inverter The inverterstarted to produce the AC output at the DC output powerof about 58W suggesting energy consumption of 58W at itsoperating mode

0 200 400 600 800 1000 1200 14000

500

1000

1500

2000

2500

PV o

utpu

t pow

er (W

)

Solar intensity (Wm2)30

∘C40

∘C50

∘C60

∘C

Figure 2 Relation of solar irradiance to PV output power andvariation in module temperature

0 500 1000 1500 2000 25000

500

1000

1500

2000

2500

Inve

rter

out

put p

ower

(W)

Inverter input power (W)

R2= 1

Intercept = minus5870

Slope = 09942

Figure 3 Relation between input and output power of inverter

Interestingly it was found that the actual maximumefficiency of the inverter strongly depended on the ambienttemperature As shown in Figure 4 the inverter efficiency


25 30 35 40 4590

92

94

96

98

Inve

rter

effici

ency

()

Ambient temperature (∘C)

Figure 4 Relation between inverter efficiency and ambient temper-ature

00 01 02 03 04 0530

40

50

60

70

80

90

100

Inverter output power (W)

Inve

rter

effici

ency

()

Ta lt 37∘C

Ta gt 37∘C

Figure 5 Average inverter efficiency of two different operatingconditions 119879

119886lt 37∘C and 119879

119886gt 37∘C

(120578ivt) reaches its maximum value of 96ndash965 when theambient temperature is below 37∘C and shows 25 dropwhen the temperature increases above 37∘Cmdashunder the samesolar irradiance of 1000ndash1100Wm2 An obvious difference inthe inverter efficiency of two different operating conditionsof which 37∘C is set as a borderline is displayed in Figure 5 Inboth cases the inverter efficiency similarly tended to remainconstant when the AC output power exceeded 1500Wor 27of the inverterrsquos rated capacity however the difference in theinverter efficiency of these two different temperatures wasobvious

It is well known that the characteristics of the inverterare dependent on the temperature of the electronic circuitmdashtemperature inside the inverter case Although we did notcollect continuous data of the temperature inside the inverterwe occasionally checked it and found its relation to theambient temperature The inverter temperature is alwayshigher than the ambient temperature During the day time

a temperature difference of about 10ndash14∘C is found whenthe ambient temperature rises higher than 32∘CThis impliesthat the ambient temperature of 37∘C corresponds to theinverter temperature of about 47ndash51∘C The grid-connectedinverter in this study is a single-phase string inverter withtransformer It contains 3 major parts Maximum PowerPoint (MPP) tracking bridge and transformer Among thesethree parts bridge is the part that is the most sensitiveto the operating temperature because it contains switchingdevices The temperature effects possibly can be mitigatedby optimizing inverter topology and its internal design Aseparation of different types of components into differenttemperature zones within the inverterrsquos overall enclosure isconcerned to be an effective way This approach is generallyutilized in low frequency transformer-based inverter design[10] Since transformer causes heating and also induces powerloss development of a transformerless inverter is likely to bepreferable in respect to the temperature effect Additionallyan efficient cooling system is essential to maintain perfor-mance and extend lifetime of the inverter

The experimental results have revealed the temperature-dependent behavior of the inverter and moreover clearlypointed out negative impact of high temperature on theinverter performance [7] In high temperature regions theoperating temperature of the inverter thus is a critical factorwhich should be concerned when analyzing the losses in thePV systems

Normalized frequency distributions of four differentinverter operating conditions are shown in Table 2 Herethe input power of 27 of the inverterrsquos rated capacity andthe ambient temperature of 37∘C are used as borderlineconditions It can be seen that the conditionswhose119879

119886gt 37∘C

(numbers 2 and 4) occupied 33 of the total frequency dis-tribution indicating relatively high possibility of undesirableoperating conditions for the inverter Furthermore the idealcondition normalized input power gt27 and 119879

119886lt 37∘C

occupied only 11 In this test site the normalized frequencydistribution of the low input conditions (numbers 3 and 4)was found to be as high as 80This was because the installedPV module capacity was 224 kWp only 45 of the inverterrsquosrated capacity

The monthly cumulative solar irradiancemdashwhich is heredenoted by solar energy received (119864

119903)mdashand the monthly

average ambient temperature are shown in Figure 6The aver-age ambient temperature from October 2010 to December2010 was found to be lower than 35∘C while from January2011 to April 2011 the temperature increased above 35∘C Inthis test site the 119864

119903ranged from 5 to 65MWh especially

being high in January The monthly average inverter effi-ciency performance ratio (PR) of the PV modules and PRof the PV system are indicated in Figure 7 It is obviousthat the inverter efficiency was strongly affected by theambient temperature that is high efficiency during low-temperature period and less efficient performance duringhigh-temperature months The high average inverter effi-ciency was observed in November and December whenthe average ambient temperature was lower than 35∘C andtheir monthly cumulative irradiance was relatively highSince the inverter efficiency is directly proportional to the


Table 2 Normalized frequency distributions of four different inverterrsquos operating conditions

Number Inverterrsquos operating conditions Normalized frequency ()Normalized input power () 119879

119886(∘C)

1 gt27 lt37 112 gt27 gt37 93 lt27 lt37 564 lt27 gt37 24

Oct

10

Nov

10

Dec

10

Jan

11

Feb

11

Mar

11

Apr

11

0

1

2

3

4

5

6

7

8

Month

Sola

r ene

rgy

rece

ieve

d (M

Wh)

0

5

10

15

20

25

30

35

40

Aver

age a

mbi

ent t

empe

ratu

re (∘

C)

Ta

Er

Figure 6 Trends of monthly solar energy received (119864119903) and average

ambient temperature (119879119886)

solar intensity higher efficiency should be observed in themonth of higher irradiance however in January the monthof the highest monthly cumulative irradiance the inverterefficiency was found to be relatively low This was perhapsdue to high temperature in January demonstrating thetemperature effects on the inverter performance Althoughthe highest inverter efficiency was obtained in December thePR of the thin film a-SiH modules significantly droppedwhich was likely to be caused by the red-shift spectrumdistribution during this period [11ndash14] The PR of the PVmodules tended to recover in January however it was stilllow due to high temperature during January 2011ndashApril 2011which eventually resulted in the low system PR Accordingto the results the PR of the PV system depended on bothPV module and inverter performance both of which werestrongly influenced by the operating temperature

The output energy of the PV system can be expressed bythe following equation which is the modified equation of amethod previously used for field-test analysis of PV systemoutput [15]

119875out = 119870119879 sdot 119870ivt sdot 119870119890 sdot 119875in sdot 120578119904 (1)

where 119875out is AC output energy 119875in is in-plane irradi-ance and 120578

119904is conversion efficiency of the PV modules

under the standard testing conditions (STC)mdashirradiance of

60

65

70

75

80

85

90

95

100

Perfo

rman

ce ra

tio o

r effi

cien

cy (

)

Inverter efficiencyPR of PV

PR of system

Oct

10

Nov

10

Dec

10

Jan

11

Feb

11

Mar

11

Apr

11

Month

Figure 7 Trends of monthly inverter efficiency PR of PV modulesand PR of PV system

1000Wm2 module temperature of 25∘C and air mass 15global spectrum 119870

119879denotes a correction coefficient for

module temperature which can be given by

119870119879= 1 + 120572 sdot (119879

119898minus 25) (2)

where 120572 is a temperature coefficient for power of the PVmodule which is minus020∘C in this case 119870ivt is a correc-tion coefficient for inverter performance Here the 119870ivt isexpressed by the following equation

119870ivt = 120578∘

ivt sdot 120572ivt = 120578lowast

ivt (3)

where 120578∘ivt is the nominal maximum efficiency of the inverter120572ivt is a correction factor for inverter efficiency and 120578lowastivt isthe actual efficiency of the inverter It is well known that theinverter efficiency depends mainly on the input DC powerwhich corresponds directly to the solar irradiance thus thevariation of the inverter efficiency can be presented in termsof the solar irradiance However the inverter efficiency underhigh-irradiance conditionmdashirradiance gt500Wm2mdashtendedto be saturated and remained constant the correction factorfor the inverter performance is therefore presented as theapproximate equation (3) A correction coefficient for fac-tors except module temperature and inverter performancewhich includes effects of shading reflection soiling and


Table 3 The monthly basis of irradiance average ambient temperature average module temperature and correction coefficients

Month Irradiance (MWh) Ambient temp (∘C) Irradiance weightedmodule temp (∘C) 119870

119879119870ivt 119870

119890

Oct 10 5398 306 475 0955 0898 0807Nov 10 5654 320 498 0950 0926 0778Dec 10 5926 334 497 0951 0935 0761Jan 11 6648 363 496 0951 0897 0779Feb 11 5398 376 492 0952 0873 0788Mar 11 5071 360 465 0957 0862 0784Apr 11 5795 374 492 0952 0870 0783

degradation is denoted by 119870119890 The 119875in 119875out 119879119898 and 120578

lowast

ivtwere obtained from the field-test data while the 120578

119904and

120578∘

ivt were known values therefore the 119870119879 119870ivt and 119870119890 can

be derived to evaluate the effects of module temperatureinverter efficiency and other factors on the PV systemperformance

We have derived the correction coefficients119870119879119870ivt and

119870119890and summarized in Table 3 These correction coefficients

whose value must be gt0 and lt1 can be used as primaryindicators of energy loss thus we can simply compare thelosses due tomodule temperature inverter and other factorsIt can be seen that the 119870

119879did not show obvious variation

while the 119870ivt tended to depend on the monthly averageambient temperature During high-temperature period the119870ivt was found to be small suggesting a large amount ofenergy loss of the inverter Interestingly the 119870

119890in December

was obviously smaller than that of the othermonthsThis wasperhaps due to increased spectrummismatch loss during thatperiod which resulted in the PR drop of the thin film a-SiHmodules

The efficiency of the inverter needless to say certainlyinfluences the total performance of the PV systems Thetemperature effects on the inverter is thus a meaningfulfinding and the quantitative analysis of this kind of lossis useful for forecasting energy yield of the PV systemsmdashespecially in high temperature regions Although at presentwe have evaluated and analyzed the results of only one high-efficient inverter it is certain that high operating temperaturehas negative effects on all inverters However the amountof loss perhaps depends on inverterrsquos type and manufacturewhich is worth investigating further

4 Conclusion

High temperature has negative impacts also on the perfor-mance of the inverter not only on the PVmodules Accordingto our experimental results the ambient temperature ofhigher than 37∘C caused 25 drop in the inverterrsquos maxi-mum efficiency During high-temperature period when themonthly average ambient temperature was gt35∘C the PRof the PV system was found to be low which was likelyto be due to a large amount of loss arising from highoperating temperature Consequently in high temperatureregions likeThailand the effect of temperature on the inverterperformance cannot be neglected and it must be taken into

account in the energy yield prediction or the loss analysis ofthe PV systems



References

[1] S Janjai Solar Radiation Maps from Satellite Data for ThailandDepartment of Alternative EnergyDevelopment and EfficiencyMinistry of Energy Bangkok Thailand 2010

[2] W Suponthana N Ketjoya W Rakwichian and P InthanonldquoPerformance evaluation AC solar home systems in Thailandsystem using multi crystalline silicon PVmodule versus systemusing thin film amorphous silicon PV modulerdquo InternationalJournal of Renewable Energy vol 2 no 2 pp 35ndash52 2007

[3] K Phaobkaew N KetjoyW Rakwichian and S Yammen ldquoPer-formance of a-Si p-Si and HIT PV technological comparisonunder tropical wet climate conditionrdquo International Journal ofRenewable Energy vol 2 no 2 pp 23ndash34 2007

[4] A Sasitharanuwat W Rakwichian N Ketjoy and S YammenldquoPerformance evaluation of a 10 kWp PV power system proto-type for isolated building in Thailandrdquo Renewable Energy vol32 no 8 pp 1288ndash1300 2007

[5] P Kamkird N Ketjoy W Rakwichian and S Sukchai ldquoInvesti-gation on temperature coefficients of three types photovoltaicmodule technologies under Thailand operating conditionrdquoProcedia Engineering vol 32 pp 376ndash383 2012

[6] D Wolpert and P Ampadu Managing Temperature Effects inNanoscale Adaptive Systems Springer New York NY USA2012

[7] S Islam A Woyte R Belmans P J M Heskes and P M RooijldquoInvestigating performance reliability and safety parameters ofphotovoltaic module inverter test results and compliances withthe standardsrdquo Renewable Energy vol 31 no 8 pp 1157ndash11812006

[8] B Burger andR Ruther ldquoInverter sizing of grid-connected pho-tovoltaic systems in the light of local solar resource distributioncharacteristics and temperaturerdquo Solar Energy vol 80 no 1 pp32ndash45 2006

[9] M Eltawil and Z Zhao ldquoGrid-connected photovoltaic powersystems technical and potential problemsmdasha reviewrdquo Renew-able and Sustainable Energy Reviews vol 14 no 1 pp 112ndash1292010


[10] J Worden and M Martinson ldquoHow inverters work what goeson inside the magic boxrdquo Solarpro no 23 pp 68ndash85 2009

[11] Y Nakada H Takahashi K Ichida T Minemoto and HTakakura ldquoInfluence of clearness index and airmass on sunlightand outdoor performance of photovoltaic modulesrdquo CurrentApplied Physics vol 10 no 2 supplement pp S261ndashS264 2010

[12] T Minemoto S Nagae and H Takakura ldquoImpact of spectralirradiance distribution and temperature on the outdoor per-formance of amorphous Si photovoltaic modulesrdquo Solar EnergyMaterials and Solar Cells vol 91 no 10 pp 919ndash923 2007

[13] A Limmanee and K Chumpolrat ldquoChanges in spectral irradi-ance and their effects on PV module performancerdquo in Proceed-ing of the 6thConference on EnergyNetwork ofThailand (ENETTrsquo10) 2010

[14] R Ruther G Kleiss and K Reiche ldquoSpectral effects on amor-phous silicon solar module fill factorsrdquo Solar Energy Materialsand Solar Cells vol 71 no 3 pp 375ndash385 2002

[15] K Nishioka T Hatayama Y Uraoka T Fuyuki R Hagiharaand M Watanabe ldquoField-test analysis of PV system outputcharacteristics focusing on module temperaturerdquo Solar EnergyMaterials and Solar Cells vol 75 no 3-4 pp 665ndash671 2003

Research ArticleGrid Connected Solar PV System with SEPIC ConverterCompared with Parallel Boost Converter Based MPPT

T Ajith Bosco Raj1 R Ramesh1 J R Maglin1 M Vaigundamoorthi1

I William Christopher1 C Gopinath1 and C Yaashuwanth2

1 Department of Electrical and Electronics Engineering Anna University Chennai 600 025 India2Department of Computer Science Engineering SRM University Tamil Nadu 603 203 India

Correspondence should be addressed to T Ajith Bosco Raj ajithboscorajgmailcom and R Ramesh rrameshannaunivedu

Received 22 August 2013 Revised 13 January 2014 Accepted 14 January 2014 Published 3 March 2014


Copyright copy 2014 T Ajith Bosco Raj et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

The main objective of this work is to study the behaviour of the solar PV systems and model the efficient Grid-connected solarpower systemThe DC-DCMPPT circuit using chaotic pulse width modulation has been designed to track maximum power fromsolar PV module The conversion efficiency of the proposed MPPT system is increased when CPWM is used as a control schemeThis paper also proposes a simplifiedmultilevel (seven level) inverter for a grid-connected photovoltaic systemThe primary goal ofthese systems is to increase the energy injected to the grid by keeping track of the maximum power point of the panel by reducingthe switching frequency and by providing high reliability The maximum power has been tracked experimentally It is comparedwith parallel boost converter Also this model is based on mathematical equations and is described through an equivalent circuitincluding a PV source with MPPT a diode a series resistor a shunt resistor and dual boost converter with active snubber circuitThis model can extract PV power and boost by using dual boost converter with active snubber By using this method the overallsystem efficiency is improved thereby reducing the switching losses and cost

1 Introduction

Because of constantly growing energy demand grid-con-nected photovoltaic (PV) systems are becoming more andmore popular and many countries have permitted encour-aged and even funded distributed-power-generation sys-tems Currently solar panels are not very efficient with onlyabout 12ndash20 efficiency in their ability to convert sunlight toelectrical power The efficiency can drop further due to otherfactors such as solar panel temperature and load conditionsIn order to maximize the power derived from the solar panelit is important to operate the panel at its optimal powerpoint To achieve this a maximum power point tracker willbe designed and implemented

TheMATLABPSPICE model of the PVmodule is devel-oped [1ndash4] to study the effect of temperature and insolationon the performance of the PVmoduleThe power electronicsinterface connected between a solar panel and a load orbattery bus is a pulse width modulated (PWM) DC-DCconverter or their derived circuits used to extract maximum

power from solar PV panel 119868-119881 characteristic curve ofphotovoltaic generators based on various DC-DC converters[5ndash8] was proposed and concluded that SEPIC converter isthe best alternative to track maximum power from PV panelThe various types of nonisolated DC-DC converters for thephoto voltaic system is reviewed [9]

The maximum power tracking for PV panel using DC-DC converter is developed [10] without using microcontrol-ler This approach ensures maximum power transfer underall atmospheric conditions The analogue chaotic PWM isused to reduce the EMI in boost converter The conversionefficiency is increased when CPWM is used as a controltechnique [11ndash13] To increase conversion efficiency an activeclamp circuit is introduced into the proposed one to providesoft switching features to reduce switching losses Moreoverswitches in the converter and active clamp circuit are inte-grated with a synchronous switching technique to reduce cir-cuit complexity and component counts resulting in a lowercost and smaller volume [14]



Multilevel inverter consists of an array of power semi-conductor switches capacitor voltage sources and clampingdiodes The multilevel inverter produces the stepped voltagewaveforms with less distortion less switching frequencyhigher efficiency lower voltage devices and better electro-magnetic compatibility [15] The commutation (process ofturn off) of the switches permits the addition of the capacitorvoltages which reach high voltages at the output [16]

A modular grid-connected PV generation system pre-sents an actual behavioural model of a grid tied PV systemsuitable for system level investigations Simplified means formodelling the PV array and investigating a gradient basedMPPT into a very simple averaged model of the power con-verter was developed and themodel has been experimentallyvetted [17 18] A single-phase grid-connected inverter whichis usually used for residential or low-power applications ofpower ranges that are less than 10 kW [15] Types of single-phase grid-connected inverters have been investigated [19]A common topology of this inverter is full-bridge three-levelThe three-level inverter can satisfy specifications through itsvery high switching but it could also unfortunately increaseswitching losses acoustic noise and level of interference toother equipment Improving its output waveform reduces itsharmonic content and hence also the size of the filter usedand the level of electromagnetic interference (EMI) generatedby the inverterrsquos switching operation [20]

MATLAB-based modelling and simulation schemewhich is suitable for studying the 119868-119881 and 119875-119881 characteristicsof a PV array under a nonuniform insolation due to partialshading [21] was proposed The mathematical model of solarPV module is useful for the computer simulation The powerelectronics interface connected between a solar panel anda load or battery bus is a pulse width modulated (PWM)DC-DC converter or their derived circuits used to extractmaximum power from solar PV panel [22] The main draw-back of PV systems is that the output voltage of PV panels ishighly dependent on solar irradiance and ambient temper-ature Therefore PV panels outputs cannot connect directlyto the load To improve this a DC-DC boost converter isrequired to interface between PV panels and loads [23] Theboost converter is fixing the output voltage of the PV systemConverter receives the variable input voltage which is theoutput of PV panels and gives up constant output voltageacross its output capacitors where the loads can be connectedIn general a DC-DC boost converter operates at a certainduty cycle In this case the output voltage depends on thatduty cycle If the input voltage is changed while the duty cycleis kept constant the output voltage will vary Duty cycle isvaried by using a pulse width modulation (PWM) technique[24]

Silicon carbide (SiC) represents an advance in silicontechnology because it allows a larger energy gap SiC is classi-fied as a wide-band-gap (WBG) material and it is the main-stream material for power semiconductors [25 26] Amongthe different types of power semiconductors the power diodewas the best device to adopt SiC technologyThemain advan-tage of SiC is high-breakdown voltage and reverse-recoverycurrent is small [27ndash29] As a result higher efficiency andhigher power density can be brought to power electronic

Rp

Rs

ID

DIsc

IPV

VPV

Figure 1 Equivalent circuit of solar PV module

systems in different applications [30 31] In this researcha new active snubber circuit is proposed to contrive a newfamily of PWM converters This proposed circuit providesperfectly ZVT turn on andZCT turn off together for themainswitch of a converter by using only one quasiresonant circuitwithout an important increase in the cost and complexityof the converter This paper proposes to implement ChaoticPWM as a control method to improve the steady state perfor-mance of the DC-DC SEPIC converter based MPPT systemfor solar PV module The nominal duty cycle of the mainswitch of DC-DC SEPIC converter is adjusted so that thesolar panel output impedance is equal to the input resistanceof the DC-DC converter which results in better spectralperformance in the tracked voltages when compared toconventional PWM control The conversion efficiency of theproposed MPPT system is increased when CPWM is usedthiswill be comparedwith parallel boost converterMultilevelinverters are promising as they have nearly sinusoidal output-voltage waveforms output current with better harmonicprofile less stressing of electronic components owing todecreased voltages switching losses that are lower than thoseof conventional two-level inverters a smaller filter size andlower EMI all of whichmake them cheaper lighter andmorecompact [29]

2 MATLAB Model of L1235-37WSolar PV Module

The output characteristics of the solar PVmodule depend onthe irradiance and the operating temperature of the cell Theequivalent circuit of PV module is shown in Figure 1

From Figure 1 the current and voltage equation is givenby

119868sc = 119868119863

+ 119868PV + (

119881119863

119877119901

)

119881PV = 119881119863

minus (119868PV lowast 119877119904)

(1)

where diode current is 119868119863

= 119868119900

+ (119890(119881119863119881119879)minus 1)

Based on the electrical equation (1) and the solar PVmod-ule are modelled in MATLAB as shown in Figure 2 whichis used to enhance the understanding and predict the 119881-119868characteristics and to analyze the effect of temperature andirradiation variation If irradiance increases the fluctuationof the open-circuit voltage is very small But the short circuit


1000

Insolation

InsolationProduct

Voltage-current characteristics

Power-current characteristics

PV1

PV module (I)

IPV

IPV VPV

VPV

IPV ramp

PPV

PPV

PPV

Figure 2 MATLAB model for PV module

Figure 3 L1235-37W solar module under test

current has sharp fluctuations with respect to irradianceHowever for a rising operating temperature the open-circuitvoltage is decreased in a nonlinear fashion [4]

The 119881-119868 characteristics are validated experimentally inthe L1235-37Wp solar module as shown in Figure 3 Thetechnical specifications of L1235-37Wp solar module undertest are given inTable 1 Figure 4 shows the119881-119868 characteristicsof L1235-37Wp which is based on the experimental resultsunder irradiation (119866) = 1000Wm2 and temperature = 25∘C

21 Space Modelling of SEPIC Converter Input at MPP Therelation between input and output currents and voltage aregiven by

119881OUT119881IN

=

119863

(1 minus 119863)

119868IN119868OUT

=

119863

(1 minus 119863)

(2)

The duty cycle of the SEPIC converter under continuousconduction mode is given by

119863 =

119881OUT + 119881119863

119881IN + 119881OUT + 119881119863

(3)

Table 1 Specifications of L1235-37W solar PV panel

Short circuit current (119868sc) 25 AVoltage at MPP (119881

119898) 164

Current at MPP (119868119898) 225

Open circuit voltage (119881oc) 21 VLength 645mmWidth 530mmDepth 34mmWeight 4 kgMaximum power (119875max) 37W

3

25

2

15

1

05

0

0 5 10 15 20 25

Am

ps (A

)

Voltage (V)

MPP

Figure 4 119881-119868 characteristics of L 1235-37W solar panel

119881119863is the forward voltage drop across the diode (119863) The

maximum duty cycle is

119863max =

119881OUT + 119881119863

119881IN(MIN) + 119881OUT + 119881119863

(4)

The value of the inductor is selected based on the belowequations

1198711

= 1198712

= 119871 =

119881IN(MIN) lowast 119863max

Δ119868119871

lowast 119891119878

(5)


Δ119868119871is the peak-to-peak ripple current at the minimum input

voltage and 119891119878is the switching frequency The value of 119862

1

depends on RMS current which is given by

1198681198621(RMS) = 119868OUT lowast radic

119881OUT + 119881119863

119881IN(MIN) (6)

The voltage rating of capacitor 1198621must be greater than the

input voltage The ripple voltage on 1198621is given by

Δ1198811198621

=

119868(OUT) lowast 119863max

1198621

lowast 119891119878

(7)

The parameters governing the selection of the MOSFET arethe minimum threshold voltage 119881th(min) the on-resistance119877DS(ON) gate-drain charge 119876GD and the maximum drain tosource voltage 119881DS(max) The peak switch voltage is equal to119881IN + 119881OUT The peak switch current is given by

1198681198761(Peak) = 119868

1198711(PEAK) + 119868

1198712(PEAK) (8)

The RMS current is given by

1198681198761(RMS) = 119868OUTradic(119881OUT + 119881IN(MIN)) lowast

119881OUT119881IN(MIN)2

(9)

The total power dissipation for MOSFETs includes conduc-tion loss (as shown in the first termof the above equation) andswitching loss (as shown in the second term) 119868

119866is the gate

drive current The 119877DS(ON) value should be selected at max-imum operating junction temperature and is typically givenin the MOSFET datasheet

119875switch = (1198681198761(RMS) lowast 119877DS(ON) lowast 119863MAX)

+ (119881IN(MIN) + 119881OUT) lowast 1198681198761(Peak) lowast

(119876GD lowast 119891119878)

119868119866

(10)

The output diode must be selected to handle the peak currentand the reverse voltage In a SEPIC converter the diode peakcurrent is the same as the switch peak current 119868

1198761(Peak) The

minimum peak reverse voltage the diode must withstand is

119881RD = 119881IN(MAX) + 119881OUT(MAX) (11)

22 Dynamic Input Characteristics of a SEPIC Converter atMPP The input voltage and the equivalent input resistanceof the converter are 119881

119878and 119877

119894 respectively As the input

power 120588119894to the converter is equal to the output power 120588

119900of

the solar PV module

120588119894= 120588119900

=

1198812

119878

119877119894

(12)

The rate of change 120588119894with respect to 119881

119878and 119877

119894can be

shown below

120597120588119894=

2119881119878

119877119894

120597119881119878

minus

1198812

119878

1198772

119894

120597119877119894 (13)

At the MPP the rate of change of 120588119894equals zero and 119877

119894=

119903119892

120597120588119894= 0 hence

120597119881119878

120597119877119894

=

119881119878

2119877119894

(14)

The equation gives the required dynamic resistance char-acteristics of the tracker at MPP

23 Generation of Chaotic PWM In order to improve thesteady state performance of solar powered system direct con-trol Chaotic Pulse width modulated (CPWM) SEPIC con-verter is proposed to track maximum power from solar PVmodule Therefore in order to get chaotic frequency 119891

Δor

chaotic amplitude 119860Δ chaos-based PWM (CPWM) is ana-

lyzed to generate chaotic PWM The MATLAB simulation iscarried out as shown in Figure 5The analogue chaotic PWMhas its advantages over the digital in its low costs and easy-to-design making it suitable for high-frequency operation andsituations when design flexibility high converter conversionefficiency and low cost In order to generate chaotic pulsewidth modulation Chuarsquos diode is used to trigger the mainswitch of SEPIC converter and to be used for reducingspectral peaks in tracked converter voltage

The CPWM adopts sawtooth to modulate but its carrierperiod 119879

1015840

Δchanges according to

1198791015840

Δ=

119883119894

Mean (119909)

lowast 119879Δ (15)

where 119879Δis invariant period 119883

119894 119894 = 1 2 119873 a chaotic

sequence 119909 = (1199091 1199092

119909119873

) and Mean(119909) average of thesequence defined as

Mean (119909) = Lim119873

sum

119894=1

1003816100381610038161003816119883119894

1003816100381610038161003816

1

119873

119873 997888rarr infin (16)

Similarly the CPWM also adopts sawtooth to modulate butits carrier amplitude 119860

1015840


1198601015840

Δ= 1 + 119870

119883119894

Mean (119909)

119860Δ (17)

where 119860Δis the invariant amplitude 119883

119894 119894 = 1 2 119873 a

chaotic sequence 119909 = (1199091 1199092

119909119873

) andMean(119909) average ofthe sequence and119870 is themodulation factor of the amplitudewhich can be set required in practice The value of 119870 isselected as low so that the ripple in the output voltage of theSEPIC converter is low Also the ripple in the output voltagecontrolled by chaotic PWM is lowThe analog chaotic carrieris generated based on the circuit the resistances (119877

1198891sdot sdot sdot 1198771198896

)

are used to realise linear resistor called Chua diodeThe para-meters for Chuarsquos diode are designed and chosen as 119877

1198891=

24 kΩ 1198771198892

= 33 kΩ 1198771198893

= 1198771198894

= 220 Ω and 1198771198895

= 1198771198896

=

20 kΩ The other parameters of Chuarsquos oscillator used in theexperiment are 119871

1= 22mH 119862

1= 47 nF 119862

2= 500 pF and

119877 = 175KΩ


Reset

Relational

Randomnumber

Discrete-time

Chaotic PWM

Chaos carrier

integrator1

A1

1z

operator1

Unit delay1

03

ge

ge

Constant2

K Ts

z minus 1

Figure 5 Chaotic PWM pulse generation

Figure 6 Hardware setup

Figure 7 Chaotic PWM

24 Experimental Setup Standalone PV System Figure 6shows the experimental setup of the proposed SEPICconverter-based MPPT for solar PV module which is con-stituted by a power stage and a control circuit The powerstage includes an inductor 119871

1 1198712 capacitor 119862

1 1198622 a switch

119878 a load resistance and a solar PV module (L1235-37Wp)The analog chaotic carrier is generated based on the hardwareoutput of CPWM is in Figure 7

3 Mathematical Model for Parallel BoostConverter with Active Snubber Circuit

Figure 12 represents the circuit diagram of the parallel boostconverter with active snubber It consists of five inductors119871119891119894 1198711198912 1198711198771 1198711198772 119871119899and three capacitors 119862

119904 119862119903 119862119900 119881119892

and 119881119900represents supply and output voltage respectively

119878 (1198781 1198782) is an active primary switch 119863 (119863

1198911 1198631198912

) is a free-wheeling diode 119863

119904(1198631 1198632 1198633) is a Snubber diode and 119877

119871

is the load resistance 119878 (1198781 1198782 1198783) operates at a switching fre-

quency 119891119878with duty ratio 119889

Choose the switching frequency of switches 1198781

= 1198782

=

100KHz and 1198783

= 200KHzWhen 119878

1= 1198782

= 0 and 1198783

= 1 as in Figure 8

119889119894119871119865

119889119905

=

1

119871119865

[119881119892

= 119881119900]

119889119881119900

119889119905

=

1

119862119900

[119894119871119865

minus

119881119900

119877119871

minus 119894119871119878]

(18)

Also the switches 1198781

= 1198782

= 1198783

= 1 as in Figure 9

119889119894119871119865

119889119905

=

1

119871119865

[119881119892

minus 119881119900]

119889119881119900

119889119905

=

1

119862119900

[119894119871119865

minus

119881119900

119877119871

]

(19)

Similarly the switches 1198781

= 1198782

= 1 and 1198783

= 1 or 0 as inFigure 10

119889119894119871119865

119889119905

=

119881119892

119871119865

119889119881119900

119889119905

=

1

119862119900

[minus

119881119900

119877119871

]

(20)


Vg

minus

+

Lf1

Lf2

Co RLLsVo

iLs

iLf

Figure 8 When 1198781

= 1198782

= 0 and 1198783

= 1

Vg

minus

+

Lf1

Lf2

CoRL Vo

iLf


= 1198782

= 1198783

= 1

By using state-space averagingmethod the state equationsduring switch-on and switch-off conditions are

1199091

=

minus (1 minus 1198891)

119871119865

1199092

minus

(1 minus 1198892)

119871119865

1199092

+

119881119892

119871119865

1199092

=

minus1

119877119871119862119900

1199092

+

(1 minus 1198891) 1198892

119862119900

1199091

+

(1 minus 1198891) (1 minus 119889

2)

119862119900

1199091

(21)

where 1199091and 119909

2are the moving averages of 119894

119871119865and 119881

119900

respectively

4 Proposed Parallel Boost Converter forPV Application

Figure 11 shows the BlockDiagram of PV based parallel boostconverter with active snubber It is the combination of newactive snubber circuit with parallel boost converter Threeswitches 119878

1 1198782 and 119878

3are used 119878

1and 1198782act as main switch

and 1198783acts as an auxiliary switch 119878

1and 1198782are controlled by

ZVT and ZCT respectively also 1198783is controlled by ZCSThis

circuit operates with the input of solar powerAssume both the main switches (119878

1and 119878

2) operate in

the same frequency The features of proposed parallel boostconverter are as follows

(i) All the semiconductors work with soft switching inthe proposed converter

(ii) The main switches 1198781and 119878

2turn on with ZVT and

turn off with ZCT(iii) The secondary switch is turned on with ZCS and

turned off with ZCS

Vg

minus

+

Lf1

Lf2

CoRL Vo

iLf


= 1198782

= 1 and 1198783

= 1 or 0

(iv) All other components of the parallel boost converterfunctions based on this soft switching

(v) There is no additional current or voltage force on themain switches 119878

1and 1198782

(vi) There is no additional current or voltage force on thesecondary switch 119878

3

(vii) Also there is no additional current or voltage force onthe main diodes 119863

1198911and 119863

1198912

(viii) According to the ratio of the transformer a part of theresonant current is transferred to the output loadwiththe coupling inductance So there is less current stresson the secondary switch with satisfied points

(ix) At resistive load condition in the ZVT process themain switches voltage falls to zero earlier due todecreased interval time and that does not make aproblem in the ZVT process for the main switch

(x) At resistive load condition in the ZCT process themain switches body diode on state time is increasedwhen the input current is decreased However thereis no effect on the main switch turn off process withZCT

(xi) This parallel boost converter operates in high-switch-ing frequency

(xii) This converter easily controls because the main andthe auxiliary switches are connected with commonground

(xiii) Themost attractive feature of this proposed converteris using ZVT and ZCT technique

(xiv) The proposed new active snubber circuit is easilyadopted with other basic PWM converters and alsoswitching converters

(xv) Additional passive snubber circuits are not necessaryfor this proposed converter

(xvi) SIC (silicon carbide) is used in the main and auxiliarydiodes so reverse recovery problem does not arise

(xvii) Theproposed active snubber circuit is also suitable forother DC-DC converters

41 Procedure for Constructing a Proposed Converter Steps toobtain a system level modeling and simulation of proposedpower electronic converter are listed below


Active snubber

Main diodesMain inductors MLI

Grid

Main capacitorMPPTSolar PV

Main switches

Snubber circuit

Auxiliary

Input source

(Lf1 Lf2)

Parallel boost converter

(S1 S2)

(L1235)

switch (S3)

Figure 11 Block diagram of PV based parallel boost converter with active snubber

Table 2 Specification of parallel boost converter with activeSnubber

Main inductor 1198711198911

750 120583HMain inductor 119871

1198912750 120583H

Upper Snubber inductor 1198711198771

5 120583HLower Snubber inductor 119871

1198772(119871119898

+ 119871119889) 2 120583H

Magnetization inductor 119871119872(119871119899

+ 1198710119897) 3 120583H

Parasitic capacitor 119862119904

1 120583FSnubber capacitor 119862

11987747 nF

Output capacitor 119862119900

330 120583F450VOutput load resistance 119877 = 119877

119871530Ω

(i) Determine the state variables of the proposed powercircuit in order to write its switched state-spacemodel for example inductance current and capaci-tance voltage

(ii) Assign integer variables (ON-1 and OFF-0 state) tothe proposed power semiconductor to each switchingcircuit

(iii) Determine the conditions controlling the states ofthe proposed power semiconductors or the switchingcircuit

(iv) Assume the main operating modes apply Kirchhoff rsquosCurrent law and Kirchhoff rsquos Voltage law and combineall the required stages into a switched state-spacemodel which is the desired system-level of the pro-posed model

(v) Implement the derived equations with MATLABSimulink

(vi) Use the obtained switched space-state model todesign linear or nonlinear controllers for the pro-posed power converter

The algorithm for solving the differential equations andthe step size should be chosen before running any simulationThis step is only suitable in closed-loop simulations [21]

42 Operation of Proposed Boost Converter with Snubber Cir-cuit Theproposed PV based converter is shown in Figure 12and it is based on a dual boost circuit where the first one(switch 119878

1and choke 119871

1198911) is used asmain chock of boost con-

verter circuit and where the second one (switch 1198782and choke

1198711198912) is used to perform an active filtering The proposed

converter applies active snubber circuit for soft switchingThis snubber circuit is built on the ZVT turn on andZCT turnoff processes of the main switches Specification of proposedparallel boost converter with active snubber is in Table 2

The power from the solar flows through the two parallelpaths High efficiency was obtained by this method So asto reach soft switching (SS) for the main and the auxiliaryswitches main switches turn on with ZVT and turn off withZCT The proposed converter utilizes active snubber circuitfor SS This snubber circuit is mostly based on the ZVT turnon and ZCT turn off processes of the main switch 119871

1198772value

is limited with (119881out1198711198772

)119905rise1198782 le 119868119894max to conduct maximum

input current at the end of the auxiliary switch rise time(119905rise1198782) and119871

1198771ge 21198711198772 To turn off 119878

1with ZCT the duration

of 119905ZCT is at least longer than fall time of 1198781(119905fall1198781)119905ZCT ge

119905fall1198781 Though the main switches are in off state the controlsignal is functional to the auxiliary switch The parasiticcapacitor of the main switch should be discharged absolutelyand themain switches antiparallel diode should be turned onThe on state time of the antiparallel diode is named 119905ZVT andin this time period the gate signal of the main switch wouldbe applied So the main switch is turned on below ZVS andZCS with ZVT

Whereas the main switches are in on state and ways inputcurrent the control signal of the auxiliary switch is appliedAfter the resonant starts the resonant current should behigher than the input current to turn on the antiparallel diodeof the main switchThe on state time of the antiparallel diode(119905ZCT) has to be longer than the main switches fall time (119905

1198911198781

)After all these terms are completed while antiparallel diodeis in on state the gate signal of the main switch should becutoff to provide ZCT for the main switch Auxiliary switchturn on with ZCS and turn off with ZCSThe auxiliary switchis turned on with ZCS for the coupling inductance limits thecurrent rise speed


Lf1

Lf2

S1 S2

S3

D2

Ld

Lm

LR1

D1

Ln

RLoad

Co

Lo1 D4

Df1

Df2

D3

CR

PVArray

Figure 12 Circuit diagram of PV based parallel boost converter with active snubber with resistive load

The current passing through the coupling inductancemust be partial to conductmaximum input current at the endof the auxiliary switch rise time (119905

1199031198783) So the turn on process

of the auxiliary switch with ZCS is offered To turn off theauxiliary switch with ZCS though the auxiliary switch is inon state the current passing through the switch should fallto zero with a new resonant Then the control signal can becutoff If 119862

119878is ignored 119871

1198771value should be two times added

with 1198711198772

to make the auxiliary switch current fall to zero Asthe current cannot stay at zero as long as the auxiliary switchfall time (119905

1198911198783

) the auxiliary switch is turned off nearly withZCS

The proposed Simulink topology is shown in Figure 13The inductors 119871

1198911and 119871

1198912have the similar values the diodes

1198631198911-1198631198912

are at the same type and the same guess was for theswitches (119878

1amp 1198782) All the inductors have individual switches

and they resemble paralleling of classic converters

5 Design of MLI Module

A multilevel converter is a power electronic system thatsynthesizes a desired output voltage levels from theDC inputssupply Compared with the traditional two-level voltageconverter the primary advantage of multilevel converters istheir smaller output voltage step which results in high powerquality lower harmonic components better electromagneticcompatibility and lower switching losses The functionalityverification of the simplified seven-level inverter is doneusing MATLAB simulation which is shown in Figure 14

This single-phase simplified seven-level inverter wasdeveloped using a single-phase full bridge (H-bridge)inverter two bidirectional auxiliary switches and a capac-itor voltage divider formed by 119862

1 1198622 and 119862

3 as shown

in Figure 14 The simplified multilevel inverter topology is

Table 3 Switching pattern for the single-phase seven-level inverter

1198810

1198781

1198782

1198783

1198784

1198785

1198786

119881dc 1 0 0 1 0 02119881dc3 0 0 0 1 1 0119881dc3 0 0 0 1 0 10 0 0 1 1 0 00lowast 1 1 0 0 0 0minus119881dc3 0 1 0 0 1 0minus2119881dc3 0 1 0 0 0 1minus119881dc 0 1 1 0 0 0

significantly advantageous over other topologies The advan-tages of simplified topology are requirement of less powerswitch power diodes and less capacitors for this inverterPhotovoltaic arrays were connected to the inverter via a DC-DC SEPIC converter The power generated by the inverter isto be delivered to the power network so the utility grid ratherthan a load was used The DC-DC SEPIC converter wasrequired because the PV arrays had a voltage that was lowerthan the grid voltage High DC bus voltages are necessary toensure that power flows from the PV arrays to the grid Afiltering inductance 119871

119891was used to filter the current injected

into the grid Proper switching of the inverter can produceseven levels of output-voltage (119881dc 2119881dc3 119881dc3 0 0

lowastminus119881dc3 minus2119881dc3 minus119881dc) from the DC supply voltage Table 3shows the switching pattern for the single-phase simplifiedseven-level inverter

6 Grid-Connected Solar Power System

The modelling and simulation of PV MPPT CPWMSEPICconverter simplified seven-level MLI and controller had


Discrete

Powergui

0

Volt

Power

Current

Duty

Boost

vs1 is4

Mean

Mean

Mean value1

solar

Mean value2

Output1Output2

Current measurement

Output4

i

minus

+

minus

+

+

+

+

+ +

+

+

minus+

s

Display

Lf1

Lf

IGBTdiode

g

g

c

c

E

E

L

Pulse generator2

Mosfet

Display1

Transfer Fcn1

isolation

cycle

pwm

panel

Pulse

Pulse

generator3

generator1

IGBTdiode 1R

minus+

minus

+

Multilevel inverter

Subsystem1

s = 24e minus 05s

C1

C0

LO1

sg D

1

den(s)

PV current3

Figure 13 Simulink model of proposed PV based parallel boost converter with active snubber circuit with MLI

+

1

2

Out1

Out1

Out1 Out1

Step1Gain

1

1

V

I P Q

Mag V I

i

Total powerf(u)s

Dg

sDg

sDg

sDg

sDg

2

minus

minus

+

+

+

+

+

+

+ +

+

sDg

g

minus

iminus+

C1

C2

C3

P5

P6

D1

D3

D5

D7

M6

M5

D2

D4

D6

D8

M1

M3

P3

P1

P4

P2

M2

M4

+

Figure 14 Simulated model for seven-level inverter

been carried out in MATLAB Simulink environment Thebasic block diagram of reliable high efficient grid-connectedsolar power system has been shown in Figure 15

The grid-connected PV system consists ofMPPT trackingusing SEPIC converter which is used to track the maximumvoltage The tracked voltage is boosted in to 325V A simpli-fied seven-levelMLI is designed to convert into anAC voltagewith seven levels which should connect to gridThe simulatedresults for the MLI output are in Figures 16 and 17

7 Results and Discussions

A modular solar PV based DC-DC converter using parallelboost converter with active filter of the proposed systemis simulated using MATLAB Simulink program The wave-forms of parallel boost converter voltage and MLI filteredoutput voltage is shown in Figures 18 and 19 The controlsignals of the switches are shown in Figures 20 and 21respectively The simulation results show the proposed PVbased soft switched parallel boost DC-DC converter hasthe proper response The detailed comparison of SEPIC andparallel boost converter is in Table 4

Table 4 Comparison of SEPIC and parallel boost converter

Parameters SEPIC converter Parallel boost converterDuty cycle 45 47No of switches 7 9Input 148 148Output 325V 448VEfficiency 9215 987

8 Conclusion

The behaviour of solar module (L1235-37Wp) is studied Themaximum power is extracted from solar PV module usingCPWM and PWM for different converters The spectrumperformance is improved when CPWM control is used forMPPT purposes The performance of MLI is studied and theproto type model of MLI is carried out The main objectiveof this research was to improve efficiency of the solar PVbased parallel boost converter and reduce the switchinglosses Simulations were initially done for conventional boostconverter with snubber circuit The changes in the inputcurrent waveform were obtained A parallel boost converter


MPPT technique

CPWM SEPIC converter

Controller

multilevel inverter

Utility gridSun7 levels

Figure 15 Block diagram of reliable high efficient grid-connected solar power system

50

40

30

20

10

0

minus10

minus20

minus30

minus40

Curr

ent

Time0 05 1 15

Time offset 0

Figure 16 MLI output current

300

200

100

0

Volta

ge

minus100

minus200

minus300

Time0 05 1 15

Time offset 0

Figure 17 MLI output voltage

475

470

465

460

455

450

445

440

0 002 004 006 008 01 012 014 016 018 02

Time offset 0

Figure 18 Parallel boost converter output voltage

800

600

400

200

0

minus200

minus400

minus600

minus800

0 01 102 03 04 05 06 07 08 09

Time offset 0

Figure 19 MLI filtered output voltage (119881119900)

550

500

450

400

350

300

250

200

150

100

50

0

0 02 04 06 08 1 12 14 16

times10minus3

Time offset 0

Figure 20 Control signals of switches 1198781and 119878

2

600

500

400

300

200

100

0

0 02 04 06 08 1 12 14 16

times10minus3

Time offset 0

Figure 21 Control signals of switch 1198783


was designed with soft switching which is provided by theactive snubber circuit The main switches and all the othersemiconductors were switched by ZVT and ZCT techniquesThe active snubber circuit was applied to the parallel boostconverter which is fed by solar input line This latest con-verter was achievedwith 148V input Due to themain and theauxiliary switches have a common ground the converter wascontrolled easilyTheproposednewactive snubber circuit canbe simply functional to the further basic PWM convertersand to all switching converters thereby increasing efficiencyand improving output voltage



References

[1] J A Gow and C D Manning ldquoDevelopment of a photovoltaicarray model for use in power-electronics simulation studiesrdquoIEE Proceedings vol 146 no 2 pp 193ndash200 1999

[2] H Patel and V Agarwal ldquoMATLAB-based modeling to studythe effects of partial shading on PV array characteristicsrdquo IEEETransactions on Energy Conversion vol 23 no 1 pp 302ndash3102008

[3] M G Villalva J R Gazoli and E R Filho ldquoComprehensiveapproach to modeling and simulation of photovoltaic arraysrdquoIEEE Transactions on Power Electronics vol 24 no 5 pp 1198ndash1208 2009

[4] GWalker ldquoEvaluatingMPPT converter topologies using amat-lab PVmodelrdquo Journal of Electrical and Electronics Engineeringvol 21 no 1 pp 49ndash56 2001

[5] H S-H Chung K K Tse S Y Ron Hui C M Mok and MT Ho ldquoA novel maximum power point tracking technique forsolar panels using a SEPIC or Cuk converterrdquo IEEE Transactionson Power Electronics vol 18 no 3 pp 717ndash724 2003

[6] M Veerachary ldquoPower tracking for nonlinear PV sourceswith coupled inductor SEPIC converterrdquo IEEE Transactions onAerospace and Electronic Systems vol 41 no 3 pp 1019ndash10292005

[7] K K Tse BM T Ho H S-H Chung and S Y R Hui ldquoA com-parative study of maximum-power-point trackers for photo-voltaic panels using switching-frequency modulation schemerdquoIEEE Transactions on Industrial Electronics vol 51 no 2 pp410ndash418 2004

[8] K K Tse M T Ho H S-H Chung and S Y R Hui ldquoA novelmaximum power point tracker for PV panels using switchingfrequencymodulationrdquo IEEETransactions on Power Electronicsvol 17 no 6 pp 980ndash989 2002

[9] M H Taghvaee M A M Radzi S M Moosavain H Hizamand M Hamiruce Marhaban ldquoA current and future study onnon-isolated DC-DC converters for photovoltaic applicationsrdquoRenewable and Sustainable Energy Reviews vol 17 pp 216ndash2272013

[10] A Safari and SMekhilef ldquoSimulation and hardware implemen-tation of incremental conductance MPPT with direct controlmethod using cuk converterrdquo IEEE Transactions on IndustrialElectronics vol 58 no 4 pp 1154ndash1161 2011

[11] H Li Z Li B Zhang F Wang N Tan and W A HalangldquoDesign of analogue chaotic PWM for EMI suppressionrdquo IEEE

Transactions on Electromagnetic Compatibility vol 52 no 4 pp1001ndash1007 2010

[12] H Li W K S Tang Z Li and W A Halang ldquoA chaotic peakcurrent-mode boost converter for EMI reduction and ripplesuppressionrdquo IEEE Transactions on Circuits and Systems II vol55 no 8 pp 763ndash767 2008

[13] Z Wang K T Chau and C Liu ldquoImprovement of electromag-netic compatibility of motor drives using chaotic PWMrdquo IEEETransactions on Magnetics vol 43 no 6 pp 2612ndash2614 2007

[14] S-Y Tseng andH-YWang ldquoAphotovoltaic power systemusinga high step-up converter forDC load applicationsrdquoEnergies vol6 pp 1068ndash1100 2013

[15] V G Agelidis and M Calais ldquoApplication specific harmonicperformance evaluation of multicarrier PWM techniquesrdquo inProceedings of the 29thAnnual IEEEPower Electronics SpecialistsConference (PESC rsquo98) pp 172ndash178 1998

[16] Y Cheng C Qian M L Crow S Pekarek and S Atcitty ldquoAcomparison of diode-clamped and cascadedmultilevel convert-ers for a STATCOMwith energy storagerdquo IEEE Transactions onIndustrial Electronics vol 53 no 5 pp 1512ndash1521 2006

[17] L Zhang K Sun Y Xing L Feng and H Ge ldquoAmodular grid-connected photovoltaic generation system based on DC busrdquoIEEE Transactions on Power Electronics vol 26 no 2 pp 523ndash531 2011

[18] M E Ropp and S Gonzalez ldquoDevelopment of a MATLABsimulink model of a single-phase grid-connected photovoltaicsystemrdquo IEEE Transactions on Energy Conversion vol 24 no 1pp 195ndash202 2009

[19] N A Rahim K Chaniago and J Selvaraj ldquoSingle-phase seven-level grid-connected inverter for photovoltaic systemrdquo IEEETransactions on Industrial Electronics vol 58 no 6 pp 2435ndash2443 2011

[20] H S Patel and R G Hoft ldquoGeneralized techniques of harmonicelimination and voltage control in thyristor invertersmdash1 har-monic eliminationrdquo IEEETransactions on Industry Applicationsvol IA-9 no 3 pp 310ndash317 1973

[21] T Esram andP L Chapman ldquoComparison of photovoltaic arraymaximum power point tracking techniquesrdquo IEEE Transactionson Energy Conversion vol 22 no 2 pp 439ndash449 2007

[22] E J Duran M Galan Sidrach-de-Cardona and J M AndujarldquoMeasuring the I-V curve of photovoltaic generators-analyzingdifferent DC-DC converter topologiesrdquo IEEE Industrial Elec-tronics Magazine pp 4ndash14 2009

[23] J L Santos F Antunes A Chehab and C Cruz ldquoA maximumpower point tracker for PV systems using a high performanceboost converterrdquo Solar Energy vol 80 no 7 pp 772ndash778 2006

[24] W Jiang and B Fahimi ldquoActive current sharing and sourcemanagement in fuel cellbattery hybrid power systemrdquo IEEETransactions on Industrial Electronics vol 57 no 2 pp 752ndash7612010

[25] M Bhatnagar and B J Baliga ldquoComparison of 6H-SiC 3C-SiC and Si for power devicesrdquo IEEE Transactions on ElectronDevices vol 40 no 3 pp 645ndash655 1993

[26] Q Zhang R Callanan M K Das S-H Ryu A K Agarwaland J W Palmour ldquoSiC power devices for microgridsrdquo IEEETransactions on Power Electronics vol 25 no 12 pp 2889ndash28962010

[27] A Elasser M H Kheraluwala M Ghezzo et al ldquoA comparativeevaluation of new silicon carbide diodes and state-of-the-artsilicon diodes for power electronic applicationsrdquo IEEE Transac-tions on Industry Applications vol 39 no 4 pp 915ndash921 2003


[28] M M Hernando A Fernandez J Garcıa D G Lamar and MRascon ldquoComparing Si and SiC diode performance in commer-cial AC-to-DC rectifiers with power-factor correctionrdquo IEEETransactions on Industrial Electronics vol 53 no 2 pp 705ndash7072006

[29] B Ozpineci and L M Tolbert ldquoCharacterization of SiC Schot-tky diodes at different temperaturesrdquo IEEE Power ElectronicsLetters vol 1 no 2 pp 54ndash57 2003

[30] G Spiazzi S Buso M Citron M Corradin and R PierobonldquoPerformance evaluation of a Schottky SiC power diode in aboost PFC applicationrdquo IEEE Transactions on Power Electronicsvol 18 no 6 pp 1249ndash1253 2003

[31] A M Abou-Alfotouh A V Radun H-R Chang and C Win-terhalter ldquoA 1-MHz hard-switched silicon carbide DC-DC con-verterrdquo IEEE Transactions on Power Electronics vol 21 no 4 pp880ndash889 2006

Research ArticleA Different Three-Port DCDC Converterfor Standalone PV System

Nimrod Vaacutezquez12 Carlos Manuel Sanchez3 Claudia Hernaacutendez1 Esliacute Vaacutezquez4

Luz del Carmen Garciacutea1 and Jaime Arau5

1 Electronics Engineering Department Technological Institute of Celaya 38010 Celaya GTO Mexico2 Applied Mathematics Division Potosino Institute of Scientific and Technological Research 78216 San Luis Potosi SLP Mexico3 Electrical Engineering Department Michoacana University 58030 Morelia MICH Mexico4 Engineering Faculty Veracruz University 94294 Boca del Rio VER Mexico5 Electronics Engineering Department Cenidet 62490 Cuernavaca MOR Mexico

Correspondence should be addressed to Nimrod Vazquez nvazquezieeeorg

Received 29 November 2013 Revised 17 January 2014 Accepted 17 January 2014 Published 2 March 2014


Copyright copy 2014 Nimrod Vazquez et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

A three-port converter suitable for standalone applications is proposed in this paper Each port is used for specific input or outputand its functions depend on the port they are the renewable source the battery set and the output portThis proposed converter isconsidered for standalone operation but it is not limited to it Not only the system is able to deliver energy independently from eachinput source or in a mixed way for the output but also the battery system may be charged from the renewable source just in caseit is required The battery port is only used when it is required this allows increasing battery lifetime Another important featureis that in case of a renewable source failure the energy is automatically demanded from the battery set like an uninterruptiblepower supply The system is able to track the maximum power of the renewable source when it is required Operation analysissimulation and experimental results are described in detail

1 Introduction

Since fossil fuels are depleting it is important to studysome different choices in order to be able to supply therequired worldwide energy Renewable energy is takeninto account more seriously nowadays There exist appli-cations of renewable energy which employ hundreds ofMW (high power) and there are also those which usehundreds of W (low power) Applications may be distin-guished depending on if they are connected to the gridor not also known as cogeneration or standalone systemsrespectively the last one is especially employed in remoteplaces where utility is not available The proposed convertercan be used in remote places where the utility is notavailable but also as satellite power supply electric vehiclesand other applications that consider a battery set as aninput

Photovoltaic panel arrays should be the main sourceof energy in standalone systems Since there is no utilityline which implies that an effective use of energy is veryimportant a battery set is needed in order to be able to supplyenergy in case the renewable source is an irregular wayleaving aside the power management a power converter stageis needed in order to assure output power DCDC converterssuitable for standalone applications can be classified intomultistage or integrated-stage topologies (Figure 1)

Multistage converters [1ndash12] are able to control eachinput independently but share a single output with theaid of a common DC bus voltage as shown in Figure 1(a)for standalone operation this configuration implies a bidi-rectional converter for the battery port as a consequenceoperation becomes complex because both sources should beable not only to deliver energy but also to manage the batterychargingdischarging In these schemes the energy to charge



DCDC

DCDC

Vrs

Vbat

Vo

+

ndash

(a)

DCDC

Vrs

Vbat

Vo

+

ndash

(b)

Figure 1 Standalone systems (a) multistage (b) integrated stage

Vrs

VbatVo

+

ndash

Figure 2 Three-port converter proposed in [8]

Vrs

Vbat

Vo

+

ndash


the battery is processed twice the first one to deliver energyto the common bus and then to charge the battery

Integrated-stage converters [1ndash11] should be able todemand current from both inputs in the easiest way(Figure 1(b)) Some of them do not include bidirectionalconverters and the battery charging is not possible unlessanother converter is considered [5 6] There are three-portconverters suggested in the literature that consider a batteryset [7ndash12]

Proposal in [8] is shown in Figure 2 unfortunately thisconfiguration uses the battery set during all the differentmodes of operation which reduces its lifetime Figure 3shows a converter based on full-bridge inverter and a trans-former [9] mainly the disadvantage of this topology is thesemiconductors count There are other schemes for three-port converter [10ndash12] but these are very similar to those twomentioned previously

A different three-port converter topology is proposed inthis paper not only semiconductors count is reduced and thebattery set is used when it is only required to do so but also

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2

Figure 4 Proposed converter

the power flow is controlled easily The system accepts twoinput voltages and the energy may be supplied from bothinputs simultaneously or independently depending on theavailability of the input voltage source One of these inputsis the battery set which may be charged from the renewablesource if it is necessary Battery set is involved only when thesystem requires it so that its lifetime increases

Information is organized in this paper as follows three-port converter proposal is discussed in Section 2 whichincludes systemoperation and analysis Section 3 is addressedfor control operation of the proposed system Experimentalresults are discussed in Section 4 and some final conclusionsare given

2 Proposed Three-Port Converter

Three-port converter proposal shown in Figure 4 is suitablefor renewable energy application especially for standaloneapplications where a battery set and power flowmanagementare required This last feature can be carried out easily in theproposed converter

The system is composed of three switches three diodestwo inductors and one capacitor as shown in Figure 4 Eachswitch determines the operation of the system this allows aneasy and good power management

For standalone systems energy is mainly provided fromrenewable source which depends on weather conditionsThen the battery set is considered in order to guarantee


S2

S1

S3

(Vrs minus Vo)L1

t0 t1

VrsL1

IL1

IL2

(a) Waveforms

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2

(b) Subcircuit for 1199050

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2

(c) Subcircuit for 1199051

Figure 5 Waveforms and subcircuits when the energy is obtained from the renewable source (mode 1 no charging)

energy supply Since energy consumption is restricted to amaximum limit this has to be optimized

When considering a photovoltaic system and a specificload connected in the standalone operation there exist manydifferent possibilities these scenarios must be taken intoaccount in the power converter in order to provide a constantand regulated output voltage no matter weather conditionsEach scenario considers different stages depending on therequired functions next are described the operating modesof the system

21Mode 1 Delivered Energy from the Renewable Source Thismode happens when there is enough energy to feed the loadhowever the battery set may be charged or not dependingon its conditions Particularly for this operation mode theswitch 119878

3is turned on all the time (Figures 5 and 6) Then

1198782is used to regulate the output voltage and 119878

1is used to

charge the battery set The two switches are easily controlledindependently because these are uncoupled

In this mode the renewable source voltage is higher thanthe battery set voltage and then the diode in antiparallel of 119878

1

is not conducing even if the battery is being charged

211 When No Battery Charging Waveforms for this modeare shown in Figure 5(a) In these conditions the converteroperates as described next

(1) During 1199050 1198781is off and 119878

2and 1198783are on the equivalent

subcircuit is shown in Figure 5(b) and then theinductor 119871

1is being charged with the renewable

source the load is fed only by the output capacitor

(2) During 1199051 1198781remains off and 119878

3remains on but 119878

2

is turned off the equivalent subcircuit is shown inFigure 5(c) and then energy stored at the inductor 119871

1

is delivered to the output capacitor

212 When Battery Charging Waveforms for this modeare shown in Figure 6(a) In these conditions the converteroperates as described next

(1) During 1199050 1198783is on but also 119878

1and 119878

3are turned

on the equivalent subcircuit is shown in Figure 6(b)so that inductors 119871

1and 119871

2are charged with the

renewable source Load is only fed from the outputcapacitor


S2

S1

S3

(Vrs minus Vo)L1

t0 t1 t2

VrsL1

minusVbat L2

(Vrs minus Vbat )L2

IL1

IL2

(a) Waveforms

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2

(d) Subcircuit for 1199052

Figure 6 Waveforms and subcircuits when the battery set is being charged (mode 1 charging)

(2) During 1199051 1198783remains on and 119878

1may be turned off

but 1198782is still on the equivalent subcircuit is shown in

Figure 6(c) so that inductor 1198711is still charged while

inductor 1198712is discharged into the battery set and by

the diode1198631

(3) During 1199052 1198783remains on but now 119878

1and 119878

2are

off the equivalent subcircuit is shown in Figure 6(d)so that stored energy at inductors 119871

1and 119871

2is

delivered to the output capacitor and the battery setrespectively

In this mode of operation the switch 1198782is always used

to control the output voltage And the switch 1198781is used to

control the charge of the battery set a MPPT tracker is usedfor this purpose except when the renewable energy availableis too high then the current for charging the battery is limitedto a maximum valueThe energy is taken from the renewablesource

22 Mode 2 Delivered Power from the Battery Set Thismodeoccurs if the photovoltaic panel has no energy (point A inFigure 7) or if the maximum energy which may be obtainedfrom the panel (point B in Figure 7) is lower than the output

Renewablesource

Battery

A

B

CPrs (W)

Vrs (V)

Figure 7 Characteristic waveforms of renewable sources anddifferent power outputs

power (point C in Figure 7) then it is necessary to use abattery in order to deliver the required amount of energy tothe load


S2

S1

S3

(Vrs minus Vo)L1

t0

t1

t2

VrsL1

(Vrs minus (Vo + Vbat ))L2

(Vrs minus Vbat )L2

IL1

IL2

(a) Waveforms

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Figure 8 Waveforms and subcircuits when the energy is obtained from both sources (mode 2 energy available from renewable source)

By assuming that the converter is operated in mode 1(switch 119878

3is on) and suddenly the renewable source suffers

a power variation that reduces its power capacity for anycircumstance the battery set is automatically connected tothe inductor 119871

1through the diode in antiparallel of 119878

1 acting

as uninterruptible power supply therefore the power to theload is guaranteed no matter weather conditions

The above implies that the operation mode 2 occurswithout any change to the switch control signals of mode 1However the energy demanded to the renewable source couldnot be themaximum available from it therefore the switch 119878

3

is used to track the maximum power point during mode 2

221 When Energy Available in Renewable Source Wave-forms for this mode are shown in Figure 8(a) In theseconditions the converter operates as described next

(1) During 1199050 1198781and 1198783are off but 119878

2is on the equivalent

subcircuit is shown in Figure 8(b) the inductors 1198711

and 1198712are discharged

(2) During 1199051 1198781remains off but 119878

2and 1198783are turned on

the equivalent subcircuit is shown in Figure 8(c) and

inductors 1198711and 119871

2are charged from the renewable

source Load is only fed from the output capacitor

(3) During 1199052 1198781is still off and 119878


2

is turned off the equivalent subcircuit is shown inFigure 8(d) Stored energy at inductor 119871

1is delivered

to the output while 1198712is charged

During all these stages the battery set and the renewablesource are delivering energy It should be noticed that thecurrent in 119871

2is negative because the battery is delivering

energy therefore the diode 1198631 in antiparallel of 119878

1 is

conducting even if control signal of 1198781is off

222When Energy Available Only in Battery Set Waveformsfor this mode are shown in Figure 9(a) In these conditionsthe converter operates as described next

(1) During 1199050 1198781is off but 119878


subcircuit is shown in Figure 9(b) so that bothinductors119871

1and119871

2are chargedwith energy provided

from the battery set Load is only fed from the outputcapacitor


Vbat (L1 + L2)

(Vbat minus Vo)(L1 + L2)

S2

S1

S3

t0 t1

IL1

IL2

IL1= minusIL2

(a) Waveforms

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Figure 9 Waveforms and subcircuits when energy is supplied from the battery set (mode 2 only battery)


3is still on but 119878

2

is turned off the equivalent subcircuit is shown inFigure 9(c) Stored energy at inductors 119871

1and 119871

2is

delivered to the output capacitor

Similar to mode 1 the switch 1198782is used to control

the output voltage however energy can be taken from therenewable source and the battery set or only just from thebattery setThe switch 119878

3is used to track themaximumpower

point of the photovoltaic panel If there is no energy in therenewable source then the battery will deliver all the powerto the load

23 Mode 3 No Power Available Thismode occurs when therenewable source and the battery cannot supply the energyrequirements to the load because it should not be forgottenthat the amount of energy is finite and it depends on thebattery set and weather conditions When this occurs thesystem is in idle condition and only the battery set could becharged (once the renewable source is available)

When the renewable source and the battery set do nothave enough energy the output voltage cannot be kept intoits acceptable limits of operationThismode will remain until

the battery set is completely charged To guarantee no energyis delivered to the load a relay is used to disconnect the load

Table 1 summarizes different switching states whichestablish the power flow of the system The switching fre-quency of the different states is determined by a single carriertherefore all the switches work under the same frequencywhen they are operating the switching period for the systemwas selected at 20120583s

24 Analysis of the Converter Since the converter may beoperated in several modes and each one has a differentbehavior analysis was made for each specific condition

In mode 1 switch 1198783is turned on but 119878

1and 119878

2are

commuting and then only appear at the equations where theduty cycle is associated with these switches State variablerepresentation for the system is then as follows

1198891198941198711

119889119905

=

119881rs1198711

minus

119881119900

1198711

(1 minus 1198891198782

)

1198891198941198712

119889119905

=

119881rs1198712

1198891198781

minus

119881bat1198712


Table 1 Switching states of the converter

Operation mode Switches Functions1198781

1198782

1198783

Mode 1 0 X 1 Feeding the load with 119881rsX X 1 Charging battery and feeding load with 119881rs

Mode 2 0 X X Feeding the load from 119881rs and 119881bat0 X 1 or 0 Feeding the load only with 119881bat

Mode 3 X 0 1 Charging battery0 0 1 System off

ldquo1rdquo means ldquoonrdquo ldquo0rdquo means ldquooffrdquo and ldquoXrdquo means ldquocommutatingrdquo

Table 2 Equilibrium point of the operation mode

Operation mode Equilibrium point Functions

Mode 1119881119900=

1

1 minus 1198891198782

119881rs Feeding the load with 119881rs

119881bat = 1198891198781

119881rs Charging battery if required

Mode 2

119881119900=

1

1 minus 1198891198782

minus 1198891198783

119881rsFeeding the load from 119881rs and 119881bat

119881bat = 119881rs minus (1 minus 1198891198783

)119881119900

119881119900=

1

1 minus 1198891198782

119881bat Feeding the load only with 119881bat

Mode 3 119881bat = 1198891198781

119881rs Charging battery

119889119881119900

119889119905

=

1198941198711

119862

(1 minus 1198891198782

) minus

119881119900

119877119862

(1)

where 119881rs is the renewable voltage source 119881bat is the batteryset voltage 119881

119900is the output voltage 119894

1198711

is the inductor1198711current 119894

1198712

is the inductor 1198712current 119862 is the output

capacitance 1198711and 119871

2are the inductances 119889

1198781

is the dutycycle of the switch 119878

1 and 119889

1198782

is the duty cycle of the switch1198782In mode 2 119878

1is turned off 119878

2is always commuting and

1198783may be switching depending on the availability of the

renewable source State variable representation for the systemwhere the renewable source is available is then as follows

1198891198941198711

119889119905

=

119881rs1198711

minus

119881119900

1198711

(2 minus (1198891198782

+ 1198891198783

))

1198891198941198712

119889119905

=

119881rs minus 119881bat1198712

minus

119881119900

1198712

(1 minus 1198891198783

)

119889119881119900

119889119905

=

1198941198711

119862

(1 minus 1198891198782

1198891198783

) 1198891198782

(1 minus 1198891198783

) minus

119881119900

119877119862

(2)

Where 1198891198783

is the duty cycle of the switch 1198783

In this mode when the renewable source cannot provideenergy the valid equations for the system are then as follows

1198891198941198711

119889119905

=

119881bat1198711+ 1198712

minus

119881119900

1198711+ 1198712

(1 minus 1198891198782

)

1198891198941198712

119889119905

= minus

1198891198941198711

119889119905

119889119881119900

119889119905

=

1198941198711

119862

(1 minus 1198891198782

) minus

119881119900

119877119862

(3)

Particularly for this case 1198941198711

= minus1198941198712

Converter in mode 3 operates only in charging mode

therefore the system equations are as follows

1198891198941198712

119889119905

=

119881rs1198712

1198891198781

minus

119881bat1198712

(4)

The equilibrium points of the different operating modesare obtained by making the set of differential equations equalto zero these are shown in Table 2 As it is easily seen theoutput voltage always depends on the duty cycle of 119878

2 hence

it is the switch employed to do itParticularly for mode 2 when energy is delivered by both

sources switch 1198783 which is related to the battery set is used

for having control of the energy delivered by this source itshould be noticed that 119878

3also affects the output voltage and

in order to maintain a regulated output voltage the 1198782must

be adjusted by the controllerIt should be noted that 119878

1affects only the battery hence

it is used only when battery requires to be charged


Table 3 Comparison between converters

Converter Reported in [8] Figure 2 Reported in [9] Figure 3 Proposed schemeActive

Diode 3 0 3MOSFET 3 12 3

PassiveInductor 1 3 2Transformer 1 1 0

Battery lifetime Bad Good GoodOperation Simple Complex Simple

25 Converter Design Issues The converter was designed tobe operated in continuous conduction mode (CCM) Notonly inductors were designed based on the desired ripple butalso the capacitor

Inductors 1198711 1198712and the capacitor may be obtained by

considering the operation in mode 1 and using the followingequations

1198711=

119881rs1198891198782

119879119904

Δ1198681198711

1198712=

(119881rs minus 119881bat) 1198891198781

119879119904

Δ1198681198712

119862 =

119881119900(1 minus 119889

1198782

) 119879119904

Δ119881119862119877

(5)

where 119879119878is the switching periodΔ119868

1198711

is the ripple of 1198711Δ1198681198712

is the ripple of 1198712 and Δ119881

119862is the ripple of 119862

26 Comparison of Proposal with Previous Schemes Table 3shows a brief comparison of this proposed converter it con-siders the number of semiconductors the passive elementsthe operation of the battery set and the general form ofoperation Schemes appear illustrated in Figure 2 [8] andFigure 3 [9] both are themost representative in the literatureIt is easily observed that converter in this proposal and theone reported in [8] have not only less components than thosereported in [9] but also an easier form of operation Althoughthis proposal has similar features than those offered in [8] itis not the same with the battery use this proposal will onlyemploy the battery when it is required to do so however theother schemewill use it during all the operatingmodes whichdiminish its lifetime

The other two schemes consider isolation which maybe a disadvantage in this proposal since it is not taken intoaccount however if more components are added it maybecome suitable

3 Controlling the Converter

There are three switches and each one is used for onepurpose as mentioned before in Section 24 Output voltagefor this converter is controlled using a traditional controller

PI compensator

MPPT

Modeselection

block

Carrier

S2S1 S3

Vor

Vo

+

ndash+ndash+ndash+ndash

Vrs

Vbat Min PPTdS2

dS1

1 minus dS3

Figure 10 Block diagram of the controller

but an integrator must be in it the employed switch forthis purpose is 119878

2 This switch controls the output voltage

independently if the converter is operated inmode 1 or 2 evenfromwhere the energy is taken which permits to operate eas-ily the system It is important to notice that even if the renew-able source is suddenly not available the system operatesproperly because in a naturalway the energy is taken from thebattery set due to the diode 119863rs and the antiparallel diode of1198781For controlling the charge of battery switch 119878

1is used

and this occurs in modes 1 and 3Particularly when operating in mode 2 when switches 119878

2

and 1198783are commuting 119878

3establishes the amount of energy

taken from each input source it comes either from the renew-able source or from the battery set A maximum power pointtracker (MPPT) algorithm is used in order to optimize theuse of the renewable source although a traditional maximumpower point tracker is reported in the literature a minimumpower point tracker (MinPPT) focused on the battery issuggested in this paper and is only necessary to sense thebattery

In Figure 10 the controller employed is shown as itmay be observed the voltage compensator (PI) the MinPPTtracker the battery charger (MPPT) the pulse width mod-ulation (PWM) section and the mode selection block areused Battery voltage is considered to determine if the batteryrequires to be charged or not while the battery current is usedto switch betweenmodes and enable the control signal of eachsemiconductor


31 Voltage Regulator A traditional controller which is a PIcompensator is used in order to regulate the output voltageThe output capacitor voltage and reference are introducedinto the compensator in order to obtain the duty cycle andthen 119889

1198782

is considered in a typical PWM controller based ona carrier signal which determines the switching frequency asshown in Figure 10 The switch used is 119878

2

The equation considered is then as follows

1199061198782

= 119896119901119890119900+ 119896119894int 119890119900119889119905 = 0 (6)

where 119890119900is the output voltage error (119890

119900= 119881119900minus 119881or) and 119881or is

the voltage referenceIt should be noticed for this controller that although

perturbations may occur the integral part reduces the steadystate error Actually output voltage depends on both switchesin mode 2 as seen in Table 2 and for reducing the couplingissue the compensator includes an integral term

32 119872119894119899119875119875119879 Algorithm Maximum power point of the

renewable source is obtained by centering the algorithm onthe battery set and then instead of aMPPTadifferentmethodis used to obtain the maximum energy from the renewablesource This algorithm is used when the battery set providesenergy to the load and there is still energy from the renewablesource (mode 2)

The system is focused on obtaining the minimum powerfrom the battery set in order to make sure that the maximumpower is delivered from the renewable source where thiscondition is achieved easily Therefore a minimum powerpoint tracker (MinPPT) for the battery set is considered thebattery power is not required since the battery set voltage isalmost constant and then only the current of the battery setmay be used

Flowchart of the proposed MinPPT algorithm is shownin Figure 11 Since current is the only variable which ismeasured the proposed technique becomes simpler thanother methods Once it is detected that the battery set isin use and delivering energy to the load the algorithm isperformed then the duty cycle of 119878

3is changed At the

beginning the duty cycle is equal to one and then decreasesuntil the minimum power is obtained the system remainsoscillating at the MinPP until the current of the battery setis zero this means that the renewable source is able to deliverall the energy to the load and the battery set is not required

After obtaining the duty cycle it is introduced into aPWM it should be noticed that the comparator inputs arechanged and the compliment of the 119889

1198782

is used in order toobtain the signals as the proposed operation in mode 2

33 Charging the Battery A voltage-current mode methodwhich is similar tomethods in uninterruptible power supplies(UPS) was implemented for charging the battery set Voltageis employed to determine when the battery set either is fullycharged or requires to be charged the current is used tocontrol the battery set charging

This stage may occur in mode 1 or 3 As the energy istaken from the renewable source a maximum power point

Start

NoYes

No

Yes No

Yes

Yes

Yes

No

NoIL2

(k) lt 0

IL2(k) lt 0

IL2(k) lt 0

IL2(k) gt IL2

(k minus 1)

IL2(k) gt IL2

(k minus 1)

Sense IL2

dS3larr 1

dS3(k) larr 1

Decrease dS3

Increase dS3

Figure 11 Flowchart of the tracking algorithm

is considered It is assumed that the battery set voltage is con-stant and then only the current is required in the algorithmsimilar to that implemented in the previous section but herethe traditional maximum power tracker is used

Additionally to theMPPT the battery current is boundedto a maximum value which allows avoiding the increase ofthe power to the battery that may cause damage in it

34 Mode Selection Block For establishing the mode ofoperation a mode selection block is considered it enablesthe respective control signal of each mode and this is madedepending on conditions for the system Basically the batteryvoltage and current are used


System functionality was experimentally evaluated with abuilt prototype so that the proposed idea was validated Thebattery set is of 24V the renewable source is a photovoltaicsolar panel emulator of Agilent (6131) the output voltage is100V and the converter was designed for an output powerof 100W the switching frequency is 50KHz 119871

1and 119871

2are

700 120583H and 119862119900is 100120583F


S1

S2

S3

1 2

2

100A 500mA100 120583s 100 MSs

10k points

Figure 12 Operation when the energy is obtained from therenewable source From top to bottom 119878

1 1198782 1198783 1198681198711

and 1198681198712

Thecurrent reference of both inductors is the same

S1

S2

S3

1 2

2

100A 500mA100 120583s 100 MSs

10k points

Figure 13 Operation when the battery set is being charged fromtop to bottom 119878

1 1198782 1198783 1198681198711

and 1198681198712

The current reference of bothinductors is the same

The experimental prototype was tested under differentoperating conditions First the performance of the converterin the differentmodes in steady state is addressed second theperformance under power variation of the renewable sourceis addressed third some tests were applied to the systemunder transitions due to the battery set conditions and finallya test under load variations is made

41 Steady State Operation Figures 12 through 15 showexperimental results for different operationmodes they wereobtained with a mixed signals oscilloscope For illustrationpurpose results were not graphed at the designed conditionswhich allow comparing with the theoretical waveformsthis is explained because the ripple values cannot be wellappreciated at the full output power and designed conditionsThen the output power in this case is 20W

Figure 12 shows operation inmode 1 when the renewablesource is available but no charging is required For this casethe available energy from the photovoltaic panel is higher thatthe load power so that no battery needs to be used As itis easily seen only inductor 119871

1provides energy to the load

since inductor 1198712has no current the battery is not in useThe

efficiency in this condition is around 93Figure 13 shows system operation in mode 1 when the

battery set is charged from the renewable source For this caseit is considered that required energy to the renewable sourceis able to satisfy the load necessities it is clearly seen how bothinductors have current 119871

1provides energy to the load and 119871

2

S1

S2

S3

1 2

2

100A 500mA100 120583s 100 MSs

10k points

Figure 14 Operation when the energy is obtained from the batteryset from top to bottom 119878

1 1198782 1198783 1198681198711

and 1198681198712

The current referenceof both inductors is the same

S1

S2

S3

1 2

2

100A 500mA100 120583s 100 MSs

10k points

Figure 15 Operation when the energy is obtained from the batteryset and the renewable source from top to bottom 119878

1 1198782 1198783 1198681198711

and1198681198712

The current reference of both inductors is the same

provides to the battery set The efficiency in this condition isaround 93

Figure 14 shows operation in mode 2 when the availableenergy from the renewable source is null then all the energyis obtained from the battery set It is seen how both inductors1198711and 119871

2 have the same current in magnitude but opposite

sign for this case the battery is discharged through bothinductors The efficiency in this condition was around 89

Figure 15 shows operation in mode 2 when the availableenergy from the photovoltaic panel is less than that requiredby the output then the battery set is used as a compliment itis easily seen how both sources deliver energy to the load andhow current in both inductors is different however they aredelivering energy to the load The efficiency in this conditionwas around 91

It is important to notice that the efficiency depends on thedevices in use and also the operating conditions this shouldbe improved if other devices are considered Different inputsandor outputs were considered because the battery port isbidirectional the efficiency was calculated as

120578 =

sum119899119875onsum119898119875im (7)

where 119875on is the ldquonrdquoth output power and 119875im is the ldquomrdquothinput power

42 Renewable Power Source Variation Theproposed systemwas tested under a renewable source variation in order to


2

3

4Ch1 Ch2Ch3 Ch4

500V200V 200V

500mA P 200 s

Figure 16 Operation when suddenly the maximum power energyof the renewable source becomes lower than the output power fromtop to bottom 119881

119900 minus1198681198712

1198782 and 119878

1

carry consistently validationHenceforth the proposed powerpoint tracker algorithm was tested

Figure 16 shows the systemoperationwhen the renewablesource suffers a variation The initial MPP of the renewablesource is 119881mpp = 48V and 119868mpp = 25A (119881oc = 60V 119868sc =3A) which means a higher output power than the initiallyconsidered in the test (119875

119900= 100W) then suddenly the MPP

is changed to119881mpp = 35V and 119868mpp = 25A (119881oc = 43V 119868sc =3A) this represents a lower power demanded by the load Itshould be noticed that the systemautomatically demands partof the energy to the battery (see current of inductor 119871

2) the

system is properly maintained in operationIt is important to notice that after the transition the MPP

of the renewable source is not assured but thenMinPPT startsto operate As it is seen in the figure the current of the batteryset is changed until the minimum battery current is obtainedby reaching the MPP of the renewable source this was madeby changing the duty cycle of the switch 119878

3

Figure 17 shows the system operationwhen the renewablesource returns to the initial condition then the energyavailable from the renewable source is enough to satisfy theload requirements As it may be noticed after the transitionthe battery does not deliver energy

43 Operation under Different Battery Set Conditions Toverify the system operation some tests were made undervariation of battery set conditions when charging starts andwhen stop is considered and when charging starts and whenstop is considered

Figure 18 shows the operation of the systems when thebattery set is being chargedTheMPP of the renewable sourceis 119881mpp = 48V and 119868mpp = 25A (119881oc = 60V 119868sc =3A) this means that initial power is higher than the outputpower considered in the test (119875

119900= 100W) the battery is

required to be charged and then the system starts to charge

2

3

4Ch1 Ch2Ch3 Ch4

500V200V 200V

500mA P 200 s

Figure 17 Operation when suddenly the maximum power energyof the renewable source becomes higher than the output power fromtop to bottom 119881

119900 minus1198681198712

1198782 and 119878

1

2

3

4Ch1 Ch2Ch3 Ch4

500V200V 200V

500mA P 200 s

Figure 18 Operation when battery set charging starts from top tobottom 119881

119900 minus1198681198712

1198782 and 119878

1

it but demanding the maximum power available from therenewable source

Figure 19 shows the operation of the systems when thebattery set is stopped to be charged The MPP of therenewable source is 119881mpp = 48V and 119868mpp = 25A (119881oc =60V 119868sc = 3A) this means that initial power is higher thanthe output power considered in the test (119875

119900= 100W) the

battery initially is being charged and when it is detected tobe fully charged then the charging is stopped

44 Load Variation A load variation was made in order tocomplete the tests Figure 20 shows the performance of thesystem when the load is changed from 50 to 100 of thepower the systemwas operated undermode 1The triggering


2

3

4Ch1 Ch2Ch3 Ch4

500V200V 200V

500mA P 200 s

Figure 19 Operation when battery set charging stops from top tobottom 119881

119900 minus1198681198712

1198782 and 119878

1

2

3

4

Ch2Ch3 Ch4

P

200V 100Vsim

250V A Ch2 150V200ms

Figure 20 Operation under a load variation from top to bottomtriggering signal 119878

2 and 119881

119900(AC mode)

signal of the event the control signal of 1198782 and the output

voltage put in AC mode are also illustrated As it is seen thesystem operates in a satisfactory manner the settling time isaround 5ms and the undervoltage is around 2V

5 Conclusion

Three-port DCDC converter in renewable applicationsshould be able to handle both the renewable source and thebattery set this is because the energy from these sourcesdepends on their availability and especially for the battery setits utility period of life should be taken into account

A different three-port DCDC converter is suggested inthis paper one port for the renewable source the secondfor the battery set and finally the output port Power may

be demanded from both input voltages simultaneously oreach one independently The battery set may also be chargedfrom the renewable source This proposal has the featurethat battery port is used only when it is required this offersan increase of lifetime Also when the renewable sourcesuddenly is not available the system automatically takesenergy from the battery set without any change only after anMPPT is used to optimize the use of the renewable source

The operation and analysis of the converter were givenExperimental results were also shown



References

[1] Y-M Chen Y-C Liu and F-Y Wu ldquoMulti-input DCDCconverter based on themultiwinding transformer for renewableenergy applicationsrdquo IEEE Transactions on Industry Applica-tions vol 38 no 4 pp 1096ndash1104 2002

[2] Y-M Chen Y-C Liu and S-H Lin ldquoDouble-input PWMDCDC converter for high-low-voltage sourcesrdquo IEEE Trans-actions on Industrial Electronics vol 53 no 5 pp 1538ndash15452006

[3] B G Dobbs and P L Chapman ldquoA multiple-input DC-DCconverter topologyrdquo IEEE Power Electronics Letters vol 1 no1 pp 6ndash9 2003

[4] Y-M Chen Y-C Liu S-C Hung and C-S Cheng ldquoMulti-input inverter for grid-connected hybrid PVwind power sys-temrdquo IEEE Transactions on Power Electronics vol 22 no 3 pp1070ndash1077 2007

[5] N Vazquez A Hernandez C Hernandez E Rodrıguez ROrozco and J Arau ldquoA double input DCDC converter forphotovoltaicwind Systemsrdquo in Proceedings of the IEEE PowerElectronics Specialists Conference (PESC rsquo08) pp 2460ndash2464Rodes Greece 2008

[6] C L Shen and S H Yang ldquoMulti-input converter with MPPTfeature for wind-PV power generation systemrdquo InternationalJournal of Photoenergy vol 2013 Article ID 129254 13 pages2013

[7] V M Pacheco L C Freitas J B Vieira Jr E A A Coelhoand V J Farias ldquoA DC-DC converter adequate for alternativesupply system applicationsrdquo in Proceedings of the 17th AnnualIEEE Applied Power Electronics Conference and Expositions pp1074ndash1080 Dallas Tex USA March 2002

[8] H Al-Atrash F Tian and I Batarseh ldquoTri-modal half-bridgeconverter topology for three-port interfacerdquo IEEE Transactionson Power Electronics vol 22 no 1 pp 341ndash345 2007

[9] J L Duarte M Hendrix and M G Simoes ldquoThree-portbidirectional converter for hybrid fuel cell systemsrdquo IEEETransactions on Power Electronics vol 22 no 2 pp 480ndash4872007

[10] H Tao J L Duarte and M A M Hendrix ldquoThree-port triple-half-bridge bidirectional converter with zero-voltage switch-ingrdquo IEEE Transactions on Power Electronics vol 23 no 2 pp782ndash792 2008


[11] H Krishnaswami and N Mohan ldquoThree-port series-resonantDC-DC converter to interface renewable energy sources withbidirectional load and energy storage portsrdquo IEEE Transactionson Power Electronics vol 24 no 10 pp 2289ndash2297 2009

[12] Z Qian O Abdel-Rahman H Al-Atrash and I BatarsehldquoModeling and control of three-port DCDC converter inter-face for satellite applicationsrdquo IEEE Transactions on PowerElectronics vol 25 no 3 pp 637ndash649 2010

Research ArticleImproved Fractional Order VSS Inc-Cond MPPT Algorithm forPhotovoltaic Scheme

R Arulmurugan12 and N Suthanthiravanitha2

1 Department of Electrical and Electronics Engineering Anna University Regional Zone Coimbatore India2Department of Electrical and Electronics Engineering Knowledge Institute of Technology KIOT CampusNH-47 Salem to Coimbatore Road Kakapalayam Salem 637 504 India

Correspondence should be addressed to R Arulmurugan arullectyahoocom



Copyright copy 2014 R Arulmurugan and N Suthanthiravanitha This is an open access article distributed under the CreativeCommons Attribution License which permits unrestricted use distribution and reproduction in any medium provided theoriginal work is properly cited

Nowadays a hot topic among the research community is the harnessing energy from the free sunlight which is abundant andpollution-free The availability of cheap solar photovoltaic (PV) modules has to harvest solar energy with better efficiency Thenature of solar modules is nonlinear and therefore the proper impedance matching is essential The proper impedance matchingensures the extraction of the maximum power from solar PVmodule Maximum power point tracking (MPPT) algorithm is actingas a significant part in solar power generating system because it varies in the output power from a PV generating set for variousclimatic conditions This paper suggested a new improved work for MPPT of PV energy system by using the optimized novelimproved fractional order variable step size (FOVSS) incremental conductance (Inc-Cond) algorithmThe new proposed controllercombines the merits of both improved fractional order (FO) and variable step size (VSS) Inc-Cond which is well suitable for designcontrol and executionThe suggested controller results in attaining the desired transient reaction under changing operating pointsMATLAB simulation effort showsMPPT controller and aDC toDCLuo converter feeding a battery load is achievedThe laboratoryexperimental results demonstrate that the new proposed MPPT controller in the photovoltaic generating system is valid

1 Introduction

Renewable energy sources are considered as an importantsource of energy in the 21st century that is in use to fulfillour needs and growing demands of electricity Among allrenewable energy sources solar energy is readily availablefree of cost The production cost of solar photovoltaic basedsystem is decreased considerably The advancement in PVtechnology also causes less cost per unit and thus PVtechnology do not contribute to global warming [1] Theextraordinary diffusion of solar PV system in electricitygeneration is evident from the fact that the PV scheme isanticipated to be the largest source of electricity generationamong all the accessible nonconventional energy sourcesThey are considered feasible in residential applications andare suitable for roof top installations [2]The PVmodules areprimarily a current source device and the current is producedwhen light falls on the surface of solar device The character-istics curve of the PV module shows its nonlinear behavior

The nonlinear 119881-119868 curve of PV module has only one point ofmaximumpower extractionTherefore the energy harvestingat maximum efficiency is not simple enough The survival ofonly one unique point of maximum power requires specialtechniques to function the scheme at the point of maximumpower These operating techniques are named as MPPT[3] MPPT techniques control the power electronic interfacesuch that the source impedance is matched with the loadimpedance and hence maximum power is transferred Incontrast with the nonlinear characteristicsMPPT techniquesare vital for any solar PV system

Different methods have been reported in literature fortracking the maximum power point (MPP) Among the20 distinct methods reported by [4] the methods such asperturb and observe (PampO) incremental conductance (Inc-Cond) fractional open circuit voltage (FOCV) fractionalshort circuit current (FSCC) fuzzy logic and neural networkalgorithm are widely used by the researchers Among these



methods the FOCV and FSCC are considered as offlineMPPT techniques because they isolate the PV array whenthey track the MPP and calculate the operating point forMPPT [5 6] These techniques adopt both analog as well asdigital implementations [7] However the periodic isolationof the PV array is power loss and the change in operatingpoint depends on irradiance (119866) therefore the periodicpower loss is to be avoidedwe need irradiance sensor that canmeasure the 119866 and hence PV array needs not to be isolated[8] The fuzzy logic andor neural network based MPPTtechnique have good performance under fast changing envi-ronmental circumstances and display improved performancethan the PampO method [9] However the main drawback ofthis technique is that its efficiency is extremely reliant onthe technical information of the engineer in calculating theerror and approaching up with the fuzzy rule based tableIt is importantly reliant on how a designer assembles thesystem based on his experience and skill Perturb and observealgorithm can be failure under fast varying environmentalcircumstancesThe Inc-Cond technique is constructed on theslope of the solar photovoltaic panel power curve This tech-nique has partly solved divergence of perturb and observemodel [10]

In this paperwe suggested a novel technique that will tunethe onlineMPPT techniques based on changingweather con-ditions The proposed algorithm modifies the existing con-ventional Inc-Cond controller based on improved fractionalorder variable step size which differs from the existing Thedifference is based on the datasheet of the panel on the novelcontroller and is constant for any particular PV array Theproposed algorithm is implemented intoMATLABSimulinkenvironment and it is tested and validated

The structure of the system is organized as followsSection 2 discuss the modelling of PV modules ImprovedFOVSS Inc-Cond controller and analysis of DC to DC Luoconverter Section 3 provides the simulation and experimen-tal setup hence results validate the controller performanceFinally Section 4 concludes remarks

2 Proposed System Description

The schematic circuit diagram for the suggested system isshown in Figure 1 It contains PV panel designed novelFOVSS Inc-Cond control algorithm synchronous DC to DCLuo converter and battery load The power switches of thedesigned DC to DC Luo converter are controlled by the gatedrivers programmed via a controller module The designedconverter delivers required levels of the output power to thestand alone battery load The impedance of the battery loadshould be assumed as a suitable one for subsequent analysisThe DC to DC converters are responsible for MPPT andvoltage regulations Simulation and experimental models areestablished in MATLABSimulink and controller processorenvironment

21 Modeling of PV Modules PV systems convert sunlightinto electrical energy without causing any environmentalissues Various equivalent models are available in the litera-ture for better understanding of concept of PV array Among

PV array

or panel

DC to DCboost-buck Luo

converterLoad

Improved FO VSS Inc-CondMPPT algorithm

Switch control signal

Vin

Iin

+

minus

+

minus

Io

Vo

Figure 1 The proposed optimized novel FOVSS Inc-Cond MPPTsystem

D

G

Radiation

ID

IR

Rsh

Rs

IoVo

Figure 2 Equivalent circuit model of solar cell

the models Figure 2 is considered as good which supportsaccuracy and user friendliness [11] For the constant weatherconditions the curve has only one unique point of maximumpower (MP) and the 119881-119868 characteristic of an irradiatedcell is nonlinear It depends on several factors includingthe temperature and irradiance With a varying irradiancethe short circuit current varies however the open circuitvoltage changes significantly with changes in temperatureThe varying atmospheric conditions make the MPP keepshifting around the PV curve In the PV simulation resultsshow the cumulative effect of the nonhomogenous weatherconditions on MPP The analytical expression based on thetemperature (119879) and irradiance (119866) variation can be writtenas follows

119868PV = 119896 sdot 119866 sdot 119878 (1)

where 119868PV is the photovoltaic current source119868119889is the single exponential junction current and is given

by

119868119889= 119868119900sdot (119890119860119881119889

minus 1) (2)

119868 is the output current and is given by 119868 = 119868PV minus 119868119889minus 119881119889119877sh

119881 is the output voltage and is given by 119881 = 119881119889minus 119877119904sdot 119868

119868sc (119866 119879) = 119868sc (STC) sdot

119866

1000

sdot (1 + 120572119868scΔ119879) (3)

119881oc (GT) = 119881oc (STC) sdot (1 + 120573119881ocΔ119879) (4)

119875119898(119866 119879) = 119875

119898(STC) sdot

119866

1000

sdot (1 + 120574120588Δ119879) (5)

120578 =

119875119898

119866119860

= (119875119898(STC) sdot

(1 + 120574120588Δ119879)

119860

) (6)

where Δ119879 = 119879119888 minus 25∘C


22 A New Design of Improved Fractional OrderVSS Inc-Cond Controller

221 FractionalOrderDifferentiator AFOsystemcomprisedby a fractional differential or an integral equation and sys-tems covering few equations has been deliberate in engineer-ing andphysical appliances for example active control signalprocessing and linear and nonlinear response controllerThe generally utilized approaches have been anticipated fornumerical assessment of fraction derivatives by Riemann-Lioville and Grunwald-Letnikov definition [12] It reflects acontinuous function 119891(119905) where its 120572th order derivative canbe conveyed as follows [13]

119889120572119891 (119905)

119889119905120572

= limℎrarr0

1

ℎ120572

120572

sum

119903=0

(minus1)119903(

120572

119903

)119891 (119905 minus 119903ℎ)

120573 = (

120572

119903

) =

120572

119903 (120572 minus 119903)

(7)

where 120573 is the coefficient binomial and 120572 is an integerpositive order We use the guesstimate approach arising theGrunwald Letnikov definition as

119863120572

119905119891 (119905) asymp ℎ

minus120572

[119905ℎ]

sum

119903=0

(minus1)119903120573119891 (119905 minus 119903ℎ) (8)

For generalization it is suitable to adopt 119905 = 119899ℎ where ldquo119905rdquois the opinion at which the derivative is appraised and ℎ isthe discretization step We can rewrite the estimate of the 120572thderivative as follows

119863120572

119905119891 (119905) =

119889120574

119889119905120574[119863minus(120574minus120572)

119905]

119863120572

119905119891 (119905) asymp (

119905

119899

)

minus120572 119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (minus120572) Γ (119903 + 1)

119891 (119905 minus 119903

119905

119899

)

(9)

where 120574 is an integer satisfying 120574 minus 1 lt 120572 le 120574 Clearly theFO calculus leads to an immeasurable dimension while theintegral calculus is a finite dimension Reflect119891

119898(119905) = 119905

119898119898 =

1 2 3 4 and the 120572th derivative is

119863120572

119905119905119898

asymp

119905119898minus120572

Γ (minus120572)

119899120572

119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (119903 + 1)

(1 minus

119903

119899

)

119898

(10)

If we expand [1 minus (119903119899)]119898 by the binominal theorem [3 6]

(10) becomes

119863120572

119905119905119898

asymp

119905119898minus120572

Γ (minus120572)

119898

sum

119896=0

(minus1)119896(

119898

119896

) 119899120572minus119896

119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (119903 + 1)

119903119896 (11)

119870 equiv

119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (119903 + 1)

119903119896 (12)

If y is an unstipulated and if 119895 is an integer positive then 119910 119895fractional is defined as

119910(119895)

= 119910 (119910 minus 1) (119910 minus 2) sdot sdot sdot (119910 minus 119895 minus 1)

Γ (119910 + 1) = 119910(119895)

Γ (119910 minus 119895 + 1)

(13)

So an integral power of 119910 can be expressed as a factorialpolynomial as

119910119896=

119896

sum

119895=1

120585119896

119895119910(119895)

=

119896

sum

119895=1

120585119896

119895

Γ (119910 minus 119895 + 1)

Γ (119910 + 1)

(14)

where the 120585 is the sterling values Let 119910 = 119903 in (14) besubstituted in (12) and replace 119899 by 119899 minus 119895 and 120572 by 120572 minus 119895 then

119870 =

119896

sum

119895=1

120585119896

119895(

119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (119903 + 1 minus 119895)

) =

119896

sum

119895=1

120585119896

119895

Γ (119899 minus 120572)

Γ (119899 minus 119895)

(

1

119895 minus 120572

)

(15)

Equation (11) becomes

119863120572

119905119905119898

asymp

119905119898minus120572

Γ (minus120572)

119898

sum

119896=0

(minus1)119896(

119898

119896

)

119896

sum

119895=0

120585119896

119895119899120572minus119896

Γ (119899 minus 120572)

(119895 minus 120572) Γ (119899 minus 119895)

lim119899rarrinfin

119899120572minus119896 Γ (119899 minus 120572)

Γ (119899 minus 119895)

=

1 if 119895 = 119896

0 if 119895 lt 119896

(16)

where119898

sum

119896=0

(minus1)119896(

119898

119896

)

1

(119896 minus 120572)

= 119861 (minus120572119898 + 1) (17)

A general fractional order differentiator can be expressed asfollows

119863120572

119905119905119898

asymp

Γ (119898 + 1)

Γ (119898 + 1 minus 120572)

119905119898minus120572

(18)

For all 120572 positive negative andor zero 119898 = 0 1 2 3 4 Note the select of 120572 can be seen as selecting the spectaclesthat will be modeled By selecting 0 lt 120572 lt 1 anomalous phe-nomena such as heat conduction diffusion viscoelasticityand electrode-electrolyte polarization can be described [1]

222 Design of New Improved VSS Inc-Cond ControllerGenerally step size is fixed for the Inc-CondMPPT techniqueThe produced power from the PV panel with a higher stepsize plays to quicker dynamics but results in extreme steadystate fluctuations and subsequent poor efficiency [14] Thiscondition is inverted through the MPPT by operating witha lesser step size Thus the tracking with constant step sizemakes a suitable trade-off among the fluctuation and dynam-ics Thus the problem can be resolved with VSS restatement[15 16] Even though all the conventional methods are simpleperturb and observe method produce oscillations occurringat maximum power point and hence output power is notachieved at desired level and results in poor efficiency TheInc-Cond method is envisioned to resolve the difficulty ofthe conventional perturb and observe method under quickvarying environment circumstances [17] Hence in this paperthe performance of the FOVSS Inc-Cond method in quicklyvarying environment conditions by using voltage versus cur-rent graph [18] Condition 1 the curve power versus voltage ispositive and the indication of the altering voltage and current


Sample

Obtain dV120572= ΔV

120572= (Vz minus

d120572I = ΔI = Iz

dP120572

dP120572

=

S = Mtimes

d120572I = 0

dV120572gt 0

V(z) = V(zminus1)

V(z) = V(zminus1)

Eq (25) = Eq (26)

Eq (25) gt Eq (26) V(z) = V(zminus1) minus S

V(z) = V(zminus1) + S

V(z) = V(zminus1) minus S1V(z) = V(zminus1) + S2

dV120572gt 0

andd120572I gt 0

dV120572lt 0 and

d120572I lt 0

End

Iz rarr Izminus1

Vz rarr Vzminus1

V(z) = V(zminus1) minus S

YY

Y

Y

Y

YY

N

N

N

N

N

minus 120572Izminus1

magnitude ( d120572I)

dV120572= 0

Vzminus1)120572

V(z) I(z)

ΔV120572times ΔI

Po lt P

Figure 3 Novel improved FOVSS Inc-Cond MPPT algorithm

is the same simultaneously the algorithm recognizes that 119866is in quickly accumulative environmental circumstances andreduces the voltage Condition 2 on the other side if theslope of the power versus voltage graph is positive alteringcurrent and voltage are opposite concurrently the algorithmrecognizes that it is quickly reducing environment situationsand rises the voltage Condition 3 lately if altering 119868 and119881 are in conflicting directions the algorithm for tracingsupreme power upsurges the119881 as the Inc-Cond conventionalalgorithmThus this algorithm eludes difference from the realMPP in quickly varying environmental circumstances

In this report a VSS procedure is suggested for theimproved Inc-Cond tracking technique and is dedicated tosearch an easier and active way to increase tracking dynamicas well as correctness In every tracking application thepossible power follower is attained by joining a DC to DCconverter among the PV panel and load system [19] Thepower output of the PV is utilized for energetic control

of the DC to DC converter pulse width modulation (119863)to diminish well the complication of the structure [20]The flowchart of the FOVSS improved Inc-Cond trackingalgorithm is illustrated in Figure 3 where the power DCto DC converter PWM (119863) recapitulation step size tunedautomatically The power output of PV panel is involved toregulate the power DC to DC converter PWM (119863) donatingto a shortened control scheme where the outputs 119868 and 119881 ofthe PV array represent119881

(119911)and 119868(119911)

at time 119911 respectivelyTheVSS implemented to diminish the problem represented aboveis written in the equation as follows

119863 (119911) = 119863 (119911 minus 1) plusmn 119872 times

10038161003816100381610038161003816100381610038161003816

119889119875

119889119881

10038161003816100381610038161003816100381610038161003816

(19)

In the above equation 119872 denotes the scaling factor which isadjusted at the period to regulate the step size The VSS canalso be recognized from the incline of the power versus dutycycle graph in [16] for perturb and observe tracking writtenas follows


1003816100381610038161003816100381610038161003816

Δ119875

Δ119881

1003816100381610038161003816100381610038161003816

(20)

In the above written equation Δ119863 represents the change instage 119863 at earlier sample period As illustrated in the powerversus voltage the derivative of (119889119875119889119881) of a PV panel canbe seen to be changing efficiently and is suggested in [15]as an appropriate constraint for determining the VSS of theperturb and observe method So the derivative (119889119875119889119881) isalso working herein to control the VSS for the Inc-Condtracking method The modern rule for PWM (119863) can beacquired as the following equation

119863(119911) = 119863 (119911 minus 1) plusmn 119872 times

10038161003816100381610038161003816100381610038161003816

119875 (119911) minus 119875 (119911 minus 1)

119881 (119911) minus 119881 (119911 minus 1)

10038161003816100381610038161003816100381610038161003816

(21)

The 119872 is necessarily determined by the effectiveness of thetracking structure Physical fine-tuning of this constraintis boring and resultant output may be effective only for agiven structure and operating circumstance [15] A modesttechnique is used to determine whether the 119872 is suggestedhere Initially higher step size of the maximum duty cycle(119863max) for constant step size tracking scheme was selectedBy such results the active development is best adequatebut gives poor steady state performance The stable stateassessment instead of dynamic assessment in the start-updevelopment of themagnitude119875 divided by119881 of the PVpaneloutput can be estimated under the constantVSSworkingwithmaximum duty cycle which will be selected as the superiorcontroller as VSS Inc-Cond tracking technique To confirmthe conjunction of the tracking superior rule the variable step(VS) rule should observe the following

119872 times

10038161003816100381610038161003816100381610038161003816

119889119875

119889119881

10038161003816100381610038161003816100381610038161003816fized step=Δ119863max

lt Δ119863max (22)

In the above equation |119889119875119889119881|fized step=Δ119863maxis the |119889119875119889119881|

at FSS operation of maximum duty cycle The 119872 can beobtained as follows

119872 lt

Δ119863max|119889119875119889119881|fized step=Δ119863max

(23)


In the equation above the VSS improved Inc-Cond trackingwill be operating with FSS of the early set superior controllerΔ119863max The above equation delivers an easier supervision todetermine the 119872 of the VSS Inc-Cond tracking techniqueWith the fulfillment of above calculation superior scalingfactor shows a relatively quick reaction than a minor scalingfactor The SW will become minute as derivative power tovoltage becomes very slight nearby the maximum power [21]

223 The Control Process of Improved FOVSS Inc-CondAlgorithm The 119881-119868 characteristics of a single module areresolute and enlarge to control the performance of a PV arrayas illustrated in Figure 3 It seems 119889119868119889119881 lt 0 with rising 119881

as 119868 is diminishing Based on (1)ndash(3) current and voltage arecontingent on environment and electricity transmission Theirregular singularities can be designated as FOD Thus the119889119868119889119881 can be altered as follows

119889120572119881 (119868)

119889120572119868

= limΔ119881rarr0

119881120572(119868) minus 119881

120572

119900(119868 minus Δ119868)

Δ119868

(24)

119889119881120572

119889120572119868

asymp

(119881 minus 119881119900)120572

119868 minus 120572119868119900

(25)

The efficiency of the weighing Δ119868 is altered as 120572 gt 0 and 120572 isan even number If 120572 = 1 then it yields to the rate of changequickness For 120572 = 2 outside the range it yields accelerationTherefore for 0 lt 120572 lt 1 the appearance can be called as thefractional rate of the alteration of operation Equation (25) isutilized to direct the FO incremental variations of the 119868 and119881

of the PV array The VSS incremental conductance load canbe modified as follows

119889120572

119889119881120572(minus

119881119900

119868119900

)

= (minus

1

119868119900

)

119889120572119881120572

119900

119889120572119868

+ (minus119881119900)

119889120572119868minus1

119874

119889120572119868

= (minus

1

119868119900

)(

Γ (2)

Γ (2 minus 120572)

)119881119900

1minus120572+ (minus119881

119900)

Γ (0)

Γ (minus120572)

1198681minus120572

119874

(26)

where Res(Γ minus119911) = ((minus1)119911119911)119885 = 0 minus1 minus2 minus3 minus4 with

remainder Γ(0) = Res(Γ minus 0) = 1 Thus the procedureof improved FOVSS Inc-Cond method examines the 119881 as avariable at which the MPP has an increasing or diminishingduty cycle

Figure 3 shows the flowchart of the improved FOVSSInc-Cond control algorithm By using the radiation meterthis control technique can modify the working mode in theprogram Based on the power output of the PV module MPPvaries hence the suggested control technique increases ordiminishes the voltage output of the PV module as a similarpath and it can be traced to the MPP It regulates the 119863

by the immediate values 119868119911and 119881

119911at existent iteration step

and their consistent values of 119868119911minus1

and 119881119911minus1

deposited at theend of the foregoing repetition step The VSS incrementalchanges in 119868 and 119881 are approached as 119889120572119868 asymp (119868

119911minus 120572119868119911minus1

) =

Δ119868 and 119889119881120572

asymp (119881119911minus 119881119911minus1

)120572

= Δ119881120572 correspondingly To

evade underestimating the employed state under numerous

+

+minus

+

minus

minus

R

Io

VL2+

minus

+

minus

L2

S2

S1

Vd

Vd L1

L c

VC2

VC1

Figure 4 DC to DC Luo converter

conditions the first voltage 119881119911can be set to 0119881 or default

values rendering to the 119879 differences Rendering to the fourconclusions the control process of improved FOVSS Inc-Cond method algorithm can be expressed as follows

Situation one if (Δ119881120572

= 0 and Δ119868 = 0) not anycontroller accomplishment is requiredSituation two if (Δ119868 = 0 and Δ119881

120572gt 0) a controller

action is required to enhance the Δ119881120572 to present

voltage 119881 with a cumulative119863 step sizeSituation three if (Δ119868 = 0 and Δ119881

120572lt 0) a controller

action is required to decrease the Δ119881120572 to present

voltage 119881 with a diminishing119863 step sizeSituation four calculated power output is equal tomultiplication of voltage and current output 119875 = 119881119868If 119875119900lt 119875 modernize the 119881 119881

119911minus1= 119881119911and 119868119911minus1

= 119868119911

and then dismiss the controller process

23 Analysis of Synchronous DC to DC Luo Converter Whenrecommending a MPP tracker the most important processis to choose and analyze a highly suitable converter whichis invented to function as the foremost fragment of thetracker (MPPT) Therefore switching mode power suppliesare suitable to operate with high efficiency Among allthe complete topologies existing the series of buck-boostconverters provide the opportunity to have either higheror lower output voltage compared with the input voltageThe conventional buck-boost formation is cheaper than theLuo one even though some drawbacks occur such as lessefficient weak transient reaction high peak current in powerapparatuses and discontinuous current input On the otherside the Luo converter has the highest efficiency with lowswitching losses amongst nonisolated DC to DC convertersand no negative polarity regulated output voltage comparedto the input voltage It can deliver an improved current outputcharacteristic due to the output stage inductor Thus theLuo configuration is an appropriate converter to be active indeceiving the MPPT [21]

The DC to DC Luo converter provides a positive polarityregulated output voltage with respect to the input voltagewhich is shown in Figure 4 The process of the synchronousLuo converter with ZVS and ZCS technique is for droppingthe switching loss of the primary switch In addition thefreewheeling diode is replaced by power switch to reduce


+

minus

+minus

R

Vd

VoC

C1

Mode-1

L1

+

minus

VL2

L2

(a)

+

minus

+minus

Vd

VoC

C1

Mode-2

L1

+

minus

VL2

L2

(b)

Figure 5 Equivalent modes of converter (a) main switch on (b) main switch off

Continuous

Pv

+minus

+minus

+

+

minus

minus

+

minus

im gS D

mg

SD

IMean

times

NOT

Mosfet1

Mosfet2

Batt lSOC

Battery

Ramp2

Relationaloperator2

Transportdelay

Transportdelay1

Clock

D initSwitch

EmbeddedMATLAB INC-Cond

fcn

i

d

inew

newdnew

Powergui

0

2

C4C2

C5 L1

L2

C6

C1

Transport delay2

Figure 6 Simulation layout of the proposed FOVSS Inc-Cond system

conduction losses too The designed circuit two powersMOSFET switches are utilized to reduce switching andconduction lossesThe energy storage elements are capacitors1198621and 119862

2and inductors 119871

1and 119871

2 119877 is the load resistance

To analyze the process of the DC to DC Luo converter thecircuit can be divided into two equivalent modes [22]

231 Modes of Operation In mode one operation when thepower switch 119878

1is turned on the inductor 119871

1is charged by

the input supply voltage 119881in At similar time the inductor1198712absorbs the energy from input source and the primary

capacitor 1198621 The load is delivered by the capacitor 119862

2 The

equivalent method of DC to DC Luo converter operatingmode 1 is shown in Figure 5(a)

In the mode 2 process when the switch is in turnedoff state the input current drawn from the source becomeszero as shown in Figure 5(b) The inductor current 119868

1198711flows

through the power 1198782to charge the capacitor119862

1The inductor

second current 1198681198712flows through119862

2to load resistance circuit

and the second switch 1198782to keep it continuous


31 Simulation Setup The PV array is modeled and coupledwith the DC to DC Luo converter and is controlled bysuggested tracking algorithm To examine the performanceand effectiveness of suggested FOVSS Inc-Cond controller itis tested on the experimental prototype of the photovoltaicMPPT controller and the complete simulation structure of aproposed system is illustrated in Figure 6 [23] It is made upof multi and mono crystalline silicon materials of 40 watt PVarray The Table 1 shows the specifications for single 10 wattPV module [10]

32 Analysis of PV Results To confirm the enactment of thesuggested system the119881-119868 and119881-119875 characteristics of single PV


Table 1 Electrical parameters of PV module

Designation Peak maximum power Peak maximum voltage Peak maximum current Open circuit voltage Short circuit currentValue [units] 10Wp 164V 0610A 21V 0700A

08

07

06

05

04

03

02

01

0

Curr

ent (

A)

Voltage (V)Voltage (V)0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16

Voltage versus current curve Voltage versus power curve10

9

8

7

6

5

4

3

2

1

0

Pow

er (W

)

Model 02KWm225∘CModel 04KWm230∘CModel 06KWm235∘C

Model 08KWm240∘CModel 1KWm245∘C



Figure 7 Simulated 119881-119868 and 119881-119875 characteristics of single PV module with variation of solar GampT which are installed on the floor of thelaboratory at GCE Salem (sponsored by IIT Bombay)

module of proposed panel are plotted for different values ofsolar insulation and cells temperature as shown in Figure 7Simulation uses the standard design method which showsthat an increased number of modules can deliver a nominallevel of operating charging current for normal range of 119866From this PV curves it was discovered that the decreasein the maximum power causes increase in temperature Thefollowing operating conditions are observed from this study(1) when increasing the load current causes drops in the PVvoltage (2) when increase in temperature causes reductionin power output due to rises of internal resistance across thecell (3) when increasing the insolation the power outputPV increases as more photons hit out electronics and furthercurrent flow causing higher recombination The variation ofpower output acts as a function of module voltage and isaffected by altered working conditions Also the output 119881

versus 119868 characteristics of the single PV module is observedunder various conditions of 119879 and 119866 [23]

33 Results for Proposed System under Dynamic WeatherConditions To distinguish the enactment of the designedimproved FOVSS Inc-Cond MPPT control algorithm whichcan automatically regulate the step size with the traditionalincremental conductance algorithm the MATLAB simu-lations are constructed under similar circumstances Thesampling period carried out for the conventional Inc-Condalgorithm was selected as 002 second Consequently thePWM duty cycle (119863) of the DC to DC Luo converteris modernized for each 002 seconds The performance ofoutput power of conventional Inc-Cond maximum trackingcontrol with a fixed size step is 002 under an irradiance stepvarious from 200Wm2 at temperature 25∘C to 800Wm2at temperature 27∘C at 05 seconds which are shown in

Figure 8(a) To differentiate the consistent photovoltaicpower output response of the designed improved FOVSS Inc-Cond maximum tracking control algorithm with allowablepossible duty size Δ119863 is 010 and is illustrated in Figure 8(b)It is observed that the fluctuations happening at steady statein conventional Inc-Cond algorithm are nearly eliminated bythe design of improved FOVSS Inc-Cond tracking algorithmAlso the dynamic enactment of the designed method isnoticeably quicker than the conventional technique by fixedsize step of 002The outcomes point out that the fluctuationsat steady state conditions are significantly reduced by usingthe designed FOVSS Inc-Cond maximum tracking controlalgorithm

The performance is compared between conventional Inc-Cond and proposed FOVSS Inc-Cond tracking algorithmand is obtained in Table 2 Compared with the conventionalincremental conductance fixed step size of Δ119863 is 010which shows good performance but results in greater steadystate fluctuation The proposed FOVSS Inc-Cond techniquesolves this problem The fluctuation at the steady state isnearly exterminated by the use of very small magnitude of(119889119875120572119889120572119868) and the resultant output power of PV array is

395W Furthermore the dynamic performance of proposedFOVSS Inc-Cond technique is quicker than conventional Inc-Cond technique which is shown in Figure 8

34 Experimental Setup and Results Theprocess of improvedFOVSS Inc-Cond maximum tracking algorithm has beenassessed by experiment The experimental test was carriedout on the laboratory test bench of the standalone PV systeminstalled on the floor of the Electrical and Electronics Engi-neering at Government College of Engineering Salem Indiasponsored by IIT Bombay A model of the suggested scheme


Table 2 Comparison of conventional and proposed tracking algorithm performance

Technique Parameter

Irradiance-200Wm2

and temperature is minus25∘CIrradiance-800Wm2

and temperature is minus27∘C Under steady stateconditionsOutput power Sampling period

in seconds Output power Sampling periodin seconds

ConventionalInc-Cond Δ119863 = 010 119875

119900 129W 002 seconds 119875

119900 387W 05 seconds More fluctuation

takes placeProposed FOVSSInc-Condalgorithm

119872 = 0056 119875119900 135W 002 seconds 119875

119900 395W 05 seconds Eliminate the

fluctuation

10

20

30

40

030 040 050 060

Times (s)

Out

put p

ower

(W)

850Wm2 27∘C

200Wm2 25∘C

850Wm2 27∘C

200Wm2 25∘C

(a)

10

20

30

40

030 040 050 060

Times (s)

Out

put p

ower

(W)

850Wm2 27∘C

200Wm2 25∘C

850Wm2 27∘C

200Wm2 25∘C

(b)

Figure 8 Simulated photovoltaic power output response under sudden change in GampT (a) conventional Inc-Cond algorithm (b) designedimproved FOVSS Inc-Cond tracking technique

(a) (b)

Figure 9 Photos of prototype setup (a) PV array (b) DC to DC Luo converter with improved FOVSS Inc-Cond MPPT algorithm

depicted in Figure 9 is composed of (a) photovoltaic paneland (b) DC to DC Luo converter with suggested controllingtechnique The DC to DC Luo converter specifications areselected as follows The input voltage is 21 V capacitance 119862

1

and capacitance 1198622are 220120583F inductances 119871

1and 119871

2are

15mH and 2mH respectively switching frequency is 10 Khz

and 12V battery Note that these passive components aredesignated to fill design criteria distilled based on equationsIn the test there are four PV modules mounted side byside and connected in series and parallel manner Atmega8 microcontroller was used to deliver the control pulsesfor the DC to DC Luo converter The 119862 language code of


Figure 10 Initial waveforms ofMPPTwith PV array (channel-1 PVvoltage channel-2 PV current channel-3 gate pulse)

the improved FOVSS Inc-Cond controller and PWM gen-erator system is constructed debugged and executed withthe assistance of the Arr studio development tool and Proispsoftware [16 17]

The initial graph with improved FOVSS Inc-Cond peaktracking control algorithm is illustrated in Figure 10 Whenthe scheme attains close to the peak power the size of the stepbecomes very tiny outcoming in an excellent power graphThe power and current of the PV rises to a length due to greatstep size change at the starting An adjustable resistive loadwas straight joined with the PV panel as well to investigatethe peak power The peak power distinguishing between thePV panel could be fashioned and the modules outputs withthe suggested FOVSS Inc-Cond peak tracking technique arewithin numerous watts Thus the peak tracking efficiency ofthe suggested technique under the present situation is about9892The peak tracking efficiency variance is not clear dueto theminor step size selected for the fixed step size Inc-CondalgorithmThe reason of this paper is to advance the dynamicreaction and investigate the change in irradiance further [18ndash20] A dual switch is familiarized to series with one set ofseries assembled PV module to simulate the consequence ofthe irradiance on the PV scheme When the SW is off oron both the voltage and power output of the PV panel willhit a step variation simulating a poor operational conditionfor the maximum tracking control When the SW is off themodules of the PV altered from three to four The equivalentPV schemepower output graphswith the suggested improvedFOVSS Inc-Cond peak tracking algorithm controller areillustrated in Figure 11 while Figure 12 demonstrates individ-uals graph for the modules of the PV that suddenly variedfrom four to three The sampling periods of the improvedFOVSS Inc-Cond peak tracking algorithm are selected toachieve almost steady state accuracy From the outcome of thefigures it can be illustrated that the PV schemewith improvedVSS gets the peak power within 13 seconds to trace the peakpower when the power output of the PV is instantly variedFrom the result it is concluded that the improved FOVSS Inc-Cond peak tracking control algorithm has the best dynamicenactment

Tek Stop M Pos 2040ms

M 100msMATH 100 vv

M

Figure 11 Change in power when the number of PV modules isincreased from three to four

Tek Stop

MATH 100 vv

M

M Pos

M 100 s

minus3440 s

Figure 12 Change in power when the number of PV modules isdecreased from four to three

4 Conclusion

In this paper a novel improved fractional order variable stepsize (FOVSS) incremental conductance (Inc-Cond) trackingalgorithm is designed and verified with MATLAB simula-tion and experimental environment The major differencebetween the suggested technique and existing tracking tech-nique includes elimination of the additional PI control loopand investigates the effect of novel Improved FOVSS Inc-Cond control technique This paper includes huge contribu-tions such as how improved VSS Inc-Cond is derived basedon fractional order derivative method how DC to DC softswitching Luo converter is designed and how comparisonbetween the proposed scheme and existing system is donewith the help of simulation and experimental arrangementThe experimental and simulation results demonstrate thatthe suggested controller tracks the peak power of the pho-tovoltaic scheme in variable insulation with quick transientresponse Since current and voltage of the solar photovoltaicare utilized as input elements it has controller characteristicswith variable step size Thus fluctuations around peak powerare significantly eliminated Thus the suggested FOVSS Inc-Cond based peak tracking algorithm increase the poweroutput 475 times the conventional power output for lowload conditions Accordingly it is seen that the suggestedtechnique is favorable for quick varying climatic situation


Nomenclature

119879 Temperature119866 IrradianceMPPT Maximum power point trackingMPP Maximum power pointPV PhotovoltaicInc-Cond Incremental conductanceADC Analog to digital converterFSS Fixed step sizeFOVSS Fractional order variable step size119863 Duty cycle119860 AppendixSW SwitchVSS Variable step size119868 Current119881 VoltageMP Maximum powerFO Fractional orderFOD Fractional order derivativeZVS Zero voltage switchingZCS Zero current switching



References

[1] C-H Lin C-H Huang Y-C Du and J-L Chen ldquoMax-imum photovoltaic power tracking for the PV array usingthe fractional-order incremental conductancemethodrdquoAppliedEnergy vol 88 no 12 pp 4840ndash4847 2011

[2] A Al Nabulsi and R Dhaouadi ldquoEfficiency optimization of aDSP-based standalone PV system using fuzzy logic and Dual-MPPT controlrdquo IEEE Transactions on Industrial Informaticsvol 8 no 3 pp 573ndash584 2012

[3] S Subiyanto A Mohamed and M A Hannan ldquoIntelligentmaximum power point tracking for PV system using Hopfieldneural network optimized fuzzy logic controllerrdquo Energy andBuildings vol 51 pp 29ndash38 2012

[4] N Patcharaprakiti S Premrudeepreechacharn and Y Sri-uthaisiriwong ldquoMaximum power point tracking using adaptivefuzzy logic control for grid-connected photovoltaic systemrdquoRenewable Energy vol 30 no 11 pp 1771ndash1788 2005

[5] T Tafticht K Agbossou M L Doumbia and A CheritildquoAn improved maximum power point tracking method forphotovoltaic systemsrdquoRenewable Energy vol 33 no 7 pp 1508ndash1516 2008

[6] A A Ghassami S M Sadeghzadeh and A Soleimani ldquoA highperformance maximum power point tracker for PV systemsrdquoElectrical Power and Energy Systems vol 53 pp 237ndash243

[7] T K Soon S Mekhilef and A Safari ldquoSimple and lowcost incremental conductance maximum power point trackingusing buck-boost converterrdquo Journal of Renewable and Sustain-able Energy vol 5 pp 023106ndash023110 2013

[8] L Guo J Y Hung and R M Nelms ldquoComparative evaluationof sliding mode fuzzy controller and PID controller for a boostconverterrdquo Electric Power Systems Research vol 81 no 1 pp 99ndash106 2011

[9] D Rekioua A Y Achour and T Rekioua ldquoTracking powerphotovoltaic system with sliding mode control strategyrdquo EnergyProcedia vol 36 pp 219ndash230 2013

[10] K Punithaa D Devaraj and S Sakthivel ldquoDevelopment andanalysis of adaptive fuzzy controllers for photovoltaic systemunder varying atmospheric and partial shading conditionrdquoApplied Soft Computing vol 13 pp 4320ndash4332 2013

[11] A I Dounis P Kofinas C Alafodimos andD Tseles ldquoAdaptivefuzzy gain scheduling PID controller formaximumpower pointtracking of photovoltaic systemrdquo Renewable Energy vol 60 pp202ndash214 2013

[12] S Lalouni andD Rekioua ldquoOptimal control of a grid connectedphotovoltaic systemwith constant switching frequencyrdquo EnergyProcedia vol 36 pp 189ndash199 2013


[14] F Liu S Duan F Liu B Liu and Y Kang ldquoA variable stepsize INCMPPT method for PV systemsrdquo IEEE Transactions onIndustrial Electronics vol 55 no 7 pp 2622ndash2628 2008

[15] Q Mei M Shan L Liu and J M Guerrero ldquoA novel improvedvariable step-size incremental-resistanceMPPTmethod for PVsystemsrdquo IEEETransactions on Industrial Electronics vol 58 no6 pp 2427ndash2434 2011

[16] D A R Wati W B Pramono and R D G WibowoldquoDesign and implementation of fuzzy logic controller basedon incremental conductance algorithms for photovoltaic poweroptimizationrdquo in Proceeding of the International Conference onSustainable Energy Engineering andApplication (ICSEEArsquo12) pp6ndash8 Yogyakarta Indonesia November 2012

[17] M H Taghvaee M A M Radzi S M Moosavain HHizam and M H Marhaban ldquoA current and future study onnon-isolated DC-DC converters for photovoltaic applicationsrdquoRenewable and Sustainable Energy Reviews vol 17 pp 216ndash2272013

[18] M A Al-Saffar E H Ismail and A J Sabzali ldquoFamily of ZC-ZVS converters with wide voltage range for renewable energysystemsrdquo Renewable Energy vol 56 pp 32ndash43 2013


[20] K Ishaque Z SalamM Amjad and S Mekhilef ldquoAn improvedparticle swarm optimization (PSO)-based MPPT for PV withreduced steady-state oscillationrdquo IEEE Transactions on PowerElectronics vol 27 no 8 pp 3627ndash3638 2012

[21] D P Hohm and M E Ropp ldquoComparative study of maximumpower point tracking algorithmsrdquo Progress in PhotovoltaicsResearch and Applications vol 11 no 1 pp 47ndash62 2003

[22] R Arulmurugan and N V Suthanthira ldquoIntelligent fuzzyMPPT controller using analysis of DC to DC novel Buckconverter for photovoltaic energy system applicationsrdquo in Pro-ceedings of the International Conference on Pattern RecognitionInformatics andMobile Engineering pp 225ndash231 February 2013

[23] R Rahmani R Yusof and M Seyedmahmoudian ldquoHybridtechnique of ant colony and particle swarm optimization forshort term wind energy forecastingrdquo Journal of Wind Engineer-ing and Industrial Aerodynamics vol 123 part A pp 163ndash1702013

Research ArticleA Simple and Efficient MPPT Method for Low-Power PV Cells

Maria Teresa Penella1 and Manel Gasulla2

1 Urbiotica SL Parc UPC Edifici Nexus II CJordi Girona 29 08034 Barcelona Spain2Universitat Politecnica de Catalunya CEsteve Terradas 7 08860 Castelldefels Spain

Correspondence should be addressed to Manel Gasulla manelgasullaupcedu

Received 31 October 2013 Accepted 28 December 2013 Published 27 February 2014


Copyright copy 2014 M T Penella and M GasullaThis is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in anymedium provided the originalwork is properly cited

Small-size PV cells have been used to power sensor nodesThese devices present limited computing resources and so low complexitymethods have been used in order to extract themaximumpower from the PV cells Among them the fractional open circuit voltage(FOCV)method has been widely proposed where the maximum power point of the PV cell is estimated from a fraction of its opencircuit voltage Here we show a generalization of the FOCV method that keeps its inherent simplicity and improves the trackingefficiency First a single-diode model for PV cells was used to compute the tracking efficiency versus irradiance Computationswere carried out for different values of the parameters involved in the PV cell model The proposed approach clearly outperformedthe FOCVmethod specially at low irradiance which is significant for powering sensor nodes Experimental tests performed witha 500mW PV panel agreed with these results

1 Introduction

The advances in electronics and communication protocolshave led to a widespread use of wireless sensor networks(WSN) In most applications the sensor nodes of the WSNare required to be wireless both for communication andpowering As for the power supply the use of small-sizePV cells or modules has been proposed The power-voltage(P-V) curve of a photovoltaic (PV) cell or panel presentsa maximum power point (MPP) that changes with tem-perature and irradiance To extract the maximum powerunder varying conditions an MPP tracker can be usedSeveral MPP tracking (MPPT) methods have been proposedin the literature [1ndash4] Since sensor nodes present limitedcomputing resources low complexity MPPT methods arepreferred for this particular application Because the locationof the sensor nodes is mostly determined by the applicationa wide range of irradiance can be expected at the sensorplacement Thus a high efficiency is desirable over a widerange of irradiance and specially for low irradiance as thepower source is scarce

One of the simplest and most popular MPPT methods isthe fractional open circuit voltage (FOCV) technique which

estimates theMPP voltage (119881MPP) from a fraction of the opencircuit voltage (119881OC) that is

119881MPPest = 119896119881OC (1)

where 119881MPPest is the estimated value of the actual 119881MPP and 119896is an empirical constant whose value should be set followinga thorough characterization of the PV panel under varyingmeteorological conditions (irradiance and cell temperature)119881OC is eithermeasured periodically (bymomentarily openingthe output of the PV panel) or by using a pilot cell (iean additional solar cell of the same type configured in opencircuit voltage configuration) Typical reported values for 119896range from 073 to 08 depending on the PV panel type andcharacteristics [2 3] Because of its simplicity the FOCVmethod has been recently applied to small-size PV cells inorder to power autonomous sensors [5ndash9]

In this work we propose to generalize (1) in order toestimate 119881MPP by using a linear fit that is

119881MPPest = 119886119881OC + 119887 (2)

where 119886 and 119887 are empirical coefficients The use of (2) willbe referred to as the linear open circuit voltage (LOCV)



method In fact the FOCV method can be considered aparticular case of the proposed LOCV method with 119887 = 0Both computed and experimental results of the proposedapproach will be presented and compared with the FOCVmethod As will be shown the LOCV method significantlyimproves the performance of the FOCV method especiallyat low irradiance while maintaining its inherent simplicityThe work presented here builds upon [10] where we firstpresented (2) and some initial results

2 Solar Cell Model

Different equivalent circuits have been used in the literaturein order tomodel the currentvoltage (I-V) characteristic of asilicon PV cell [11ndash15] Among them the single-diode modelshown in Figure 1 offers a good compromise between sim-plicity and accuracy [13] whereby 119868PH is the photogeneratedcurrent 119868 is the cell current 119881 is the cell voltage and 119877sand 119877p are respectively the series and shunt resistancesThismodel will be used here in order to generate computed dataof the I-V curve of a PV cell

The corresponding expression of the I-V characteristic isgiven by [16]

119868 = 119868SC minus 119868O [119890((119902(119881+119877s119868))119899d119870119879cell)

minus 1] minus

119881 + 119877s119868

119877p (3)

where 119868PH has been approximated by 119868SC the short circuitcurrent of the cell 119868O is the saturation current of the diode119902 is the electron charge 119899d is the ideality factor of the diodewhich for silicon is usually between 1 and 2 [5 7] 119870 isthe Boltzmann constant and 119879cell is the cell temperature inKelvin

By considering open circuit conditions (119868 = 0 and 119881 =119881OC) in (3) we can write the parameter 119868O as

119868O =119868SC minus (119881OC119877p)

[119890(119902119881OC119899d119870119879cell) minus 1]

(4)

The parameters 119868SC and 119881OC in (3) and (4) change with theirradiance and temperature as

119868SC (119879cell 119866) =119866

1000

[119868SCr+ 120572 (119879cell minus 119879r)] (5)

119881OC (119879cell 119866) = [119881OCr+ 120573 (119879cell minus 119879r)]

times [1 + 120588OC ln(119866

119866OC) ln( 119866

119866r)]

(6)

where119879cell is the cell temperature119866 is the incident irradiance(in Wm2) 120572 and 120573 are the current and voltage temperaturecoefficients respectively 119868SCr

and119881OCrare given at a reference

irradiance (119866r) and reference cell temperature (119879r) and120588OC and 119866OC are two empirical constants used to modelthe significant variation of 119881OC at low 119866 Typically 119866r =1000Wm2 (=100mWcm2) and 119879r = 25∘C Values of 120588OC =

minus004 and 119866OC = 1000Wm2 are adequate for many siliconPV cells [17] When directly illuminated solar cells heat up

minus

+

I

V

Rs

RpIPH

Figure 1 Single-diode model of a solar cell with series (119877s) andshunt (119877p) resistances

above the ambient temperature (119879a) which is known as theself-heating effect and 119879cell can be obtained from

119879cell = 119879a +119879cellNOCT minus 20

800Wm2119866 (7)

where 119879cellNOCT known as the nominal operating cell temper-ature (NOCT) is the temperature of the cell when exposedto 800Wm2 at 119879a = 20

∘C and wind speed of 1ms Itis empirically determined and for silicon solar cells rangebetween 42∘C and 48∘C

3 Computed Results

We employed (3) to (7) with the following parameter valuestypical of monocrystalline solar cells [17] 119869SCr

= 35mAcm2119881OCr

= 06V 120572119860 = 125 120583Acm2∘C 120573 = minus2mV∘C and119879cellNOCT = 42∘CThe parameters 119868SCr

and 120572 in (5) can be re-spectively obtained by multiplying 119869SCr

and 120572119860 by the areaof the cell For the computations a single solar cell with anarea of 1 cm2 was used Nevertheless as will be justified at theend of this section the results and the derived conclusionsare equally valid to cells of any area and PV panels composedof an arbitrary number of solar cells disposed in parallel andseries Figure 2 shows the computed I-V and P-V curves atthree different levels of irradiance for the particular case of119877s = 0 119877p = infin 119899d = 15 and 119879a = 25

∘C To obtain the datawe simulated the PV cell model in SPICE The power valueswere obtained bymultiplying 119868 by119881 at each data point As canbe seen both 119881OC and 119881MPP slightly decrease at the highestirradiance which is due to the self-heating effect of the PVcell

From the data of the I-V and P-V curves severalparameters can be obtained such as 119881OC 119881MPP and 119875MPP(power at the MPP) Table 1 shows numerical values of thoseparameters for an irradiance range from 119888119886 20Wm2 to1000Wm2 Fourteen points of irradiance logarithmicallyequally spaced were selected to provide a dynamic rangearound 100 in 119875MPP Other cases that will be discussedthroughout this section are also shown in Table 1 Figure 3


Table 1 Computed VOC VMPP and PMPP data at fourteen points of irradiance logarithmically equally spaced and for different values of theparameters of the PV cell model

G(Wm2)

119899d = 15 (Figure 1) 119899d = 1 119899d = 2 119903s = 01 (Figure 6) 119903p = 10 (Figure 8)VOC(V)

VMPP(V)

PMPP(mW)

VOC(V)

VMPP(V)

PMPP(mW)

VOC(V)

VMPP(V)

PMPP(mW)

VOC(V)

VMPP(V)

PMPP(mW)

VOC(V)

VMPP(V)

PMPP(mW)

237 0262 0194 0135 0262 0207 0153 0262 0185 0121 0262 0193 0134316 0311 0237 0227 0311 0251 0254 0311 0226 0205 0311 0235 0225422 0357 0277 0362 0357 0293 0401 0357 0264 0330 0357 0275 0359 0241 0126 0095562 0398 0313 0553 0398 0331 0609 0398 0300 0507 0398 0311 0548 0306 0168 016875 0434 0347 0826 0434 0365 0904 0434 0332 0761 0434 0343 0816 0369 0221 0299100 0467 0376 12 0467 0396 1312 0467 0361 1113 0467 0372 1186 042 0282 05241334 0495 0402 1725 0495 0422 1874 0495 0385 1600 0495 0395 1693 0462 0337 08991778 0518 0423 2431 0518 0444 2635 0518 0406 2259 0518 0415 2374 0494 0379 14772371 0537 0440 338 0537 0461 3657 0537 0422 3146 0537 0429 3279 0519 0409 23213162 055 0452 464 055 0474 5013 055 0434 4321 055 0437 4458 0537 043 35004217 0558 0458 628 0558 0481 6786 0558 0441 5853 0558 0439 5961 0548 0443 50945623 056 0459 8386 056 0482 9067 056 0441 7812 056 0433 7821 0553 0448 71857499 0555 0454 11031 0555 0477 11943 0555 0436 10264 0555 042 10035 055 0446 98501000 0543 0442 14254 0543 0464 15472 0543 0424 13234 0543 0398 12511 054 0436 1313

0 01 02 03 04 05 060

125

25

375

50

0

4

8

12

16PMPP versus VMPP

1000Wm2

500Wm2

100Wm2

I(m

A)

V (V)

P(m

W)

Figure 2 Generic I-V and P-V plots at several values of 119866 and at119879a = 25

∘C for a single solar cell with an area of 1 cm2 A curve joiningthe MPPs is also plotted

represents the fourteen computed points (diamonds) of119881MPPversus 119881OC and two least-squares regression lines fitted tothe computed data corresponding respectively to the FOCVmethod that is (1) with 119896 = 0809 and the LOCV methodthat is (2) with 119886 = 0894 and 119887 = minus0041 As can be seen theregression line corresponding to the LOCVmethod better fitsthe computed dataThe inferred parameters of the regressionlines (119896 119886 and 119887) were used to obtain 119881MPPest at the fourteenirradiance points for each of the two methods by using (1)and (2) respectively

The corresponding power values at 119881MPPest 119875MPPest wereinferred from the computed P-V curves in order to obtain thetracking efficiency which is given by

120578MPP =119875MPPest

119875MPP (8)

015

020

025

030

035

040

045

050

02 025 03 035 04 045 05 055 06

nd = 15 Ta = 25∘C Rs = 0 and Rp = infin

VMPP versus VOC

VM

PP(V

)

VOC (V)

LOCV VMPPest = 0894VOC minus 0041

FOCV VMPPest = 0809VOC

Figure 3 Computed 119881MPP versus 119881OC for a single PV cell Twoleast-square regression lines are also represented the grey linecorresponds to the FOCV method with 119896 = 0809 whereas theblack line corresponds to the LOCV method with 119886 = 0894 and119887 = minus0041 The parameters of the PV cell model are 119879a = 25

∘C119899d = 15 119877s = 0 and 119877p = infin

This parameter is used in the literature to comparethe performance among different algorithms Obviously avalue of 1 (100) is the ultimate goal Figure 4 shows thecomputed values of 120578MPPT versus 119866 at 119879a = 25

∘C forthe fourteen irradiance points We added the results at twomore temperatures 0∘C and 50∘C For these temperaturesthe P-V curves were recalculated but we still used thesame regression lines of Figure 3 This makes sense as aPV panel can be characterized at a single temperature forexample 25∘C and the calculated regression lines used for


94

95

96

97

98

99

100

10 100 1000

nd = 15 Rs = 0 and Rp = infin

120578M

PPT

()

G (Wm2)

LOCV Ta = 25∘C

LOCV Ta = 0∘C

LOCV Ta = 50∘C

FOCV Ta = 25∘C

FOCV Ta = 0∘C

FOCV Ta = 50∘C

Figure 4 Computed 120578MPPT versus119866 by using the FOCV and LOCVmethods for a single PV cell at different values of 119879a and with 119899d =15 119877s = 0 and 119877p = infin

the full working temperature range As can be seen at lowirradiance the LOCVmethod clearly outperforms the FOCVmethod At higher irradiance a rather high value of 120578MPPT(gt99) is achieved by both methods although the FOCVmethod presents the lowest efficiency from 119888119886 100Wm2 to1000Wm2 at 119879a = 0

∘C The value of 120578MPPT for the LOCVmethod was always higher than 998 at the three computedtemperatures

More computations were carried out at 119879a = 25∘C for

119899d = 1 and 119899d = 2 (see Table 1) Again better linearfits were obtained with the LOCV method (not shown)Figure 5 shows the corresponding computed values of 120578MPPTFor each of the cases the parameters of the correspondingregression lines are provided Again the LOCV methodclearly outperforms the FOCVmethod at low irradiance andslightly at medium irradiance

Finally computations were performed for 119899d = 15 119879a =25∘C nonzero values of 119877s and finite values of 119877p In [13]

normalized values for 119877s and 119877p were defined as

119903s =119877s

[119881OC119868SC]STC

119903p =119877p

[119881OC119868SC]STC

(9)

This normalization allows for an immediate comparisonamong different PVmodules (of different area and character-istics) Based on [13] in ourworkwe considered the followingvalues from 001 to 01 for 119903s and from 100 to 10 for 119903p Theperformance for both 119903s = 001 and 119903p = 100 was almostidentical to that shown in Figure 2 (for 119879a = 25

∘C) with theLOCV method outperforming the FOCV method So these

120578M

PPT

()

G (Wm2)

96

965

97

975

98

985

99

995

100

10 100 1000

Ta = 25∘C Rs = 0 and Rp = infin

VMPP est = 0926VOC minus 0037

LOCV nd = 1


LOCV nd = 2

VMPP est = 0850VOCFOCV nd = 1


Figure 5 Computed 120578MPPT versus119866 by using the FOCV and LOCVmethods for a single PV cell at different values of 119899d and with 119879a =25∘C 119877s = 0 and 119877p = infin

results are not shown here Table 1 shows the data for thelimiting cases 119903s = 01 (with 119903p = infin) and 119903p = 10 (with119903s = 0) For the case of 119903p = 10 the data for the lowest twoirradiance levels were not used as they provided negligiblevalues of 119875MPP (well below of 1 of the resulting 119875MPP at1000Wm2) As for 119903s = 01 Figure 6 shows the computedvalues of 119881MPP versus 119881OC and the fitted regression linesDue to the high value of 119903s (highest limit) the data valuespresent a folded form at the highest irradiance levels So theregression lines of the FOCV and LOCV methods cannot fitthe data corresponding to the high irradiance levels as well asthat in Figure 3 Otherwise both lines are very similar in thiscase Consequently the computed values of 120578MPPT shown inFigure 7 (119903s = 01) are quite similar (and indeed relativelyhigh) for both methods

As for 119903p = 10 Figure 8 again shows the computed valuesof 119881MPP versus 119881OC and the fitted regression lines Due tothe low relative value of 119903p (lowest limit) the regression lineof the LOCV method cannot fit the data corresponding tothe low irradiance levels as well as that in Figure 3 Even sothe computed values of 120578MPPT also shown in Figure 7 stillpresent a high efficiency outperforming the FOCVmethod atall the irradiance levels but specially at the low ones Finallywe computed 120578MPPT for 119903s = 01 and 119903p = 10 (not shown) Inthat case the LOCV method also outperformed the FOCVmethod at low and medium irradiance levels

Increasing the PV cell area or adding identical PV cells inparallel will scale up the values of currents and thus of powersbut the values of 119881OC and 119881MPP will remain the same and sothe derived tracking efficiencies Tracking efficency will also


015

02

025

03

035

04

045

05

02 03 04 05 06

rs = 01 nd = 15 Ta = 25∘C and rp = infin

VM

PP(V

)

VOC (V)



VMPP versus VOC

Figure 6 Computed 119881MPP versus 119881OC for a single PV cell with119879a = 25

∘C 119899d = 15 119903p = infin and 119903s = 01 Two regressionlines corresponding to the FOCV and LOCV methods are alsorepresented

75

80

85

90

95

100

10 100 1000

nd = 15 and Ta = 25∘C

120578M

PPT

()

G (Wm2)

LOCV rs = 01 LOCV rp = 10

FOCV rs = 01 FOCV rp = 10

Figure 7 Computed 120578MPPT versus 119866 by using the FOCV and theLOCV methods for a single PV cell with 119879a = 25

∘C n119889= 15 and

for both 119903s = 01 (and 119903p = infin) and 119903p = 10 (and 119903s = 0)

remain constant by adding PV cells in series both 119881MPP and119881OC will scale up by the number of serial cells but their ratiowill remain constant and so the derived tracking efficiencies


The LOCVmethod was tested with a 500mW (119868SC = 160mA119881OC = 46V) PV panel (MSX-005 Solarex) and comparedwith the FOCV method These low-power panels are usedfor example to power autonomous sensors [5ndash10] In order

0005

01015

02025

03035

04045

05

02 03 04 05 06

rp = 10 nd = 15 Ta = 25∘C and rs = 0

VM

PP(V

)

VOC (V)



VMPP versus VOC


∘C 119899d = 15 119903s = 0 and 119903p = 10 Two regressionlines corresponding to the FOCV and LOCV methods are alsorepresented

to achieve reproducible results we implemented a PV panelsimulator by connecting a current source (GS610 Yokogawa)in parallel with the PV panel which was coated with anopaque cover (Figure 9) In this way the short circuit current(119868SC) of the PV panel was adjusted by the current sourceemulating different levels of irradiance Since the panel wasnot illuminated 119879celLNOCT = 20∘C (ie the panel is not over-heated) The current source was configured to cover the fullrange of the PV panel varying from 5mA to 158mA in 9mAsteps The PV panel simulator was characterized by usingthe GS610rsquos measurement unit to measure the panelrsquos voltagea 2001 multimeter (Keithley) to measure the current of thepanel and a programmable voltage source (Agilent E3631A)in parallel with a 10Ω1W resistor acting as a load Figure 9shows the experimental setup

All the instruments were controlled via the GPIB buswith a dedicated program using the graphical developmentenvironment LabVIEW For each current value (119868SC) thevoltage of the E3631A was increased from 0V to 5V in 01 Vsteps PV output voltages and currents were measured andthe corresponding power values were calculated in order toobtain the I-V and P-V curves From each P-V curve the val-ues of119881OC 119881MPP and119875MPP were obtainedThe limit values for119875MPP were respectively 82mW (119868SC = 5mA) and 5459mW(119868SC = 158mA)

Figure 10 represents the experimental data of119881MPP versus119881OC and two fitted least-squares regression lines correspond-ing to the FOCV and LOCV methods As can be seen theregression line corresponding to the LOCV method betterfits the experimental data From the two regression lines thevalues of 119881MPPest corresponding to the FOCV and LOCVmethods were derived Then from the P-V curves thevalues of 119875MPPest and 120578MPPT were obtained Figure 11 shows120578MPPT versus 119875MPP In agreement with the computed resultsof Section 3 the LOCV method clearly outperformed the


minus

minus

+

=

ISC iPV

I-V current source I-V diode I-V PV array simulator

Covered PV panel

GS610 Power source

PV array simulator

Multimeter 2001 (keithley)

E3631APower source

(Agilent)IDiode

10Ω

PV

Figure 9 The PV array simulator and the setup used for its characterization

3 32 34 36 38 4 42 44 46 48 52

25

3

35

4PV array simulator

VMPP versus VOCLOCV VMPPest = 1067VOC minus 1182


VM

PP(V

)

VOC (V)

Figure 10 119881MPP versus 119881OC for a 500mW panel simulator at 119879a =25∘C and two fitted least-square regression lines corresponding to

the FOCV and LOCV methods

0 100 200 300 400 500 60092

93

94

95

96

97

98

99

100


120578M

PPT

()


PMPP (mW)

Figure 11 Experimental 120578MPPT versus 119875MPP by using the FOCVmethod (grey line) with 119896 = 0798 and the LOCV method (blackline) with 119886 = 10674 and 119887 = minus1182 for a 500mW PV panel

FOCV method at low irradiance with 120578MPPT always beinghigher than 9996The efficiency increase at low irradianceis of significant importance in order to power sensor nodes

5 Conclusion

PV cells have been proposed in the literature in order topower the sensor nodes of WSN Because of the limitedcomputing capabilities of the sensor nodes simple MPPTmethods have to be used Among them the FOCV methodhas been widely proposed and used Tracking efficienciesthough are lower than that achieved with more complexmethods In this work we have proposed the LOCVmethodwhich outperforms the FOCV method while maintainingits inherent simplicity Computations show that the LOCVmethod achieves a high efficiency for all the irradiance rangewhereas the FOCVmethod fails in achieving a high efficiencyat low irradiance levels for most of the cases The presenceof extremely low values of shunt resistance of the PV cellnegatively impacts the achieved efficiency on both methodsbut specially that of the FOCVmethod Finally experimentaldata from a low-power 500mW PV panel confirmed thegood performance of the LOCV method for a wide range ofirradiance which is of significant value for powering sensornodes



Acknowledgment

Thisworkwas supported by the SpanishMinistry of Economyand Competitiveness under Contract TEC2011-27397

References




[3] D P Hohm and M E Ropp ldquoComparative study of maximumpower point tracking algorithmsrdquo Progress in Photovoltaics Re-search and Applications vol 11 no 1 pp 47ndash62 2003

[4] V Salas E Olıas A Barrado and A Lazaro ldquoReview of themaximum power point tracking algorithms for stand-alonephotovoltaic systemsrdquo Solar Energy Materials and Solar Cellsvol 90 no 11 pp 1555ndash1578 2006

[5] D Dondi A Bertacchini D Brunelli L Larcher and L BeninildquoModeling and optimization of a solar energy harvester systemfor self-powered wireless sensor networksrdquo IEEE Transactionson Industrial Electronics vol 55 no 7 pp 2759ndash2766 2008

[6] A S Weddell G V Merrett and B M Al-Hashimi ldquoPho-tovoltaic sample-and-hold circuit enabling MPPT indoors forlow-power systemsrdquo IEEE Transactions on Circuits and SystemsI vol 59 no 6 pp 1196ndash1204 2012

[7] A Chini and F Soci ldquoBoost-converter-based solar harvester forlow power applicationsrdquo Electronics Letters vol 46 no 4 pp296ndash298 2010

[8] F I Simjee and P H Chou ldquoEfficient charging of supercapaci-tors for extended lifetime of wireless sensor nodesrdquo IEEE Trans-actions on Power Electronics vol 23 no 3 pp 1526ndash1536 2008

[9] O Lopez-Lapena andM T Penella ldquoLow-power FOCVMPPTcontroller with automatic adjustment of the sampleampholdrdquoElectronics Letters vol 48 no 20 pp 1301ndash1303 2012

[10] M T Penella and M Gasulla ldquoOptical energy harvestingrdquo inPowering Autonomous Sensors chapter 5 section 55 pp 101ndash107 Springer Dordrecht The Netherlands 2011

[11] J A Gow and C D Manning ldquoDevelopment of a photovoltaicarray model for use in power-electronics simulation studiesrdquoIEE Proceedings Electric Power Applications vol 146 no 2 pp193ndash200 1999


[13] C Carrero J Amador and S Arnaltes ldquoA single procedure forhelping PV designers to select silicon PVmodules and evaluatethe loss resistancesrdquo Renewable Energy vol 32 no 15 pp 2579ndash2589 2007


[15] F Adamo F Attivissimo A Di Nisio and M SpadavecchialdquoCharacterization and testing of a tool for photovoltaic panelmodelingrdquo IEEE Transactions on Instrumentation andMeasure-ment vol 60 no 5 pp 1613ndash1622 2011

[16] F Lasnier and T G Ang Photovoltaic Engineering HandbookAdam Hilger New York NY USA 1990

[17] A Luque and S HegedusHandbook of Photovoltaic Science andEngineering 2003

Research ArticleBICO MPPT A Faster Maximum Power Point Tracker andIts Application for Photovoltaic Panels

Hadi Malek1 and YangQuan Chen2

1 Electrical amp Computer Engineering Department Utah State University Logan UT 84321 USA2MESA Lab School of Engineering University of CA Merced California 95343 USA

Correspondence should be addressed to Hadi Malek hadimalekaggiemailusuedu

Received 29 May 2013 Accepted 3 January 2014 Published 27 February 2014


Copyright copy 2014 H Malek and Y Chen This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

This paper develops a maximum power point tracking (MPPT) algorithm to optimize photovoltaic (PV) array performance andmake it more compatible to rapidly varying weather conditions In particular a novel extremum seeking controller (ESC) whichuses a Bode ideal cutoff (BICO) filter in its structure is designed and tested on a simulated PV arrayThe new algorithm is comparedagainst the commonly used ESCMPPT algorithmwith first-order filtersThe BICO extremum seeking controller achieves transientrise to the MPP faster than the common extremum seeking MPPT which is the faster and more robust method among all othermethodsThis claim has been discussed and proved mathematically in this paper in addition to simulation illustrationsThis fasterextremum seeking algorithm enables PV systems to detect rapid variations in the environmental conditions like irradiation andtemperature changes

1 Introduction

The PV cells exhibit a nonlinear119881-119868 characteristics as shownin Figure 1 and their power output mainly depends on thenature of the connected load Since connecting the loaddirectly to the PV system results in a poor overall efficiency tominimize the life cycle cost of any PV system increasing theefficiency by changing the operating point of the systemusingan intermediate maximum power point tracker (MPPT) isdesirable

MPPT controls the output current and voltage andconsequently output power of the PV panel adaptively tomaintain maximum efficiency and better performance inthe presence of environmental variations Typically MPPTalgorithms are implemented on a solar array using a switchingpower converter for instance in a grid-tied inverter the solararray charges a capacitor and then the current is switched outof the capacitor at an optimal varying duty cycle in order toextract maximum power from the PV array

A number of solar power converter architectures withMPPT are discussed in the literature [1 2] As discussed inthese works convergence speed is one of the most important

features among all different MPPT algorithms Brunton et alhave pointed out in their paper that ldquoas irradiance decreasesrapidly the IV curve shrinks and theMPV andMPI decreaseIf theMPPT algorithmdoes not track fast enough the controlcurrent or voltage will fall off the IV curverdquo [3] Consequentlyany improvement in the rise time of MPPT improves thereliability of the system and increases the power extractionand efficiency of the whole system

11 Maximum Power Point Tracking Algorithms There aremany different maximum power point tracking techniquesfor photovoltaic systems which are well established in theliterature [4ndash11] These techniques vary in many aspects assimplicity convergence speed digital or analogical imple-mentation sensors required cost range of effectiveness andother aspects In analog world short current (SC) openvoltage (OV) and CV are good options for MPPT otherwisewith digital circuits that require the use of microcontrollerPerturb and Observe (PampO) Incremental Conductance (IC)and temperature methods are easy to implement [12] Table 1and Figure 2 present the comparison among different MPPT



Maximumpowerpoint

Irra

dian

ce

Arr

ay cu

rren

tI

Terminal voltage V

Figure 1 119868-119881 curves at various solar irradiances [3]

Ener

gy g

ener

ated

()

100

90

80

Low Medium High

P and Ob

P and Oa ICa

ICb

OV

TP

TGSC

P and Oc

CV

Cost

Figure 2 Comparison of the MPPT methods [12]

methods considering the costs of sensors microcontrollerand additional power components and also their efficienciesIn this table A means absence L means low M meansmedium and H means high

Currently the most popular and workhorse MPPT algo-rithm is PampO because of its balance between performanceand simplicity However it suffers from the lack of speedand adaptability which is necessary for tracking the fast tran-sients under varying environmental conditions [3] Recentlyanother adaptive algorithm called extremum seeking con-trol has been developed [13] to overcome these weaknesses

12 Extremum Seeking Control A promising new robustMPPT algorithm is the method of extremum seeking (ES)control which carries all PampOrsquos benefits like simplicity andperformance and in addition improves its weaknesses [3]Figures 3 and 4 present the comparison between ESC andPampO where the inverter controls the current and voltage

6000

5000

4000

3000

2000

0 500 1000 1500

Pow

er (W

)

Time (s)

6000

5000

4000

3000

2000

0 500 1000 1500

Pow

er (W

)

Time (s)

(a)

Curr

ent (

A)

15

10

5

0 500 1000 1500

Time (s)

(b)

MPPES

PO

400

350

300

Volta

ge (V

)

0 500 1000 1500

Time (s)

(c)

Figure 3 Comparison of current controlled PampO and ESC MPPTcontroller [3]

6000

5000

4000

3000

2000

0 500 1000 1500

Pow

er (W

)

Time (s)

(a)

Curr

ent (

A)

15

10

5

0 500 1000 1500

Time (s)

(b)

MPPES

PO

400

350

300

Volta

ge (V

)

0 500 1000 1500

Time (s)

(c)

Figure 4 Comparison of voltage controlled PampO and ESC MPPTcontroller [3]


Table 1 MPPT methods comparison [12]

MPPT algorithm Additional components Sensors Micro-Controller TotalCV A L AL LSC H M AL MOV H LM AL LMP and Oa A M L LMP and Ob A M L LMP and Oc A M M MIC A M M MTG A MH M MHTP A H MH H

according to the set-points which are provided by these twoMPPT algorithms

The power current and voltage are plotted versus timefor the ESC and PampO algorithms as well as the truemaximumpower of the system

PampO and ESC methods oscillate closely around the realmaximum power voltage as seen in the power versus timeplot Obviously the ESC method rises to the MPP orders ofmagnitude more rapidly than the PampO

ESC MPPT has some advantages from hardware imple-mentation point of view Brunton et al have mentioned intheir paper that ldquothe ripple-based ES algorithm has goodMPPT performance over a range of inverter capacitor sizesTypically the choice of capacitor is expensive because it mustbe well characterized and large enough to maintain a smallripple However because the ES control signal exploits thenatural inverter ripple a smaller capacitor allows the trackingof rapid irradiance changesrdquo

ESC for peak power point tracking method has beensuccessfully applied to biochemical reactors [14 15] ABScontrol in automotive brakes [16] variable cam timing engineoperation [17] electromechanical valves [18] axial compres-sors [19] mobile robots [20] mobile sensor networks [21 22]optical fibre amplifiers [23] and so on A good survey of theliterature on this topic prior to 1980 can be found in [24]and a more recent overview can be found in [25] Astromand Wittenmark rated extremum seeking as one of the mostpromising adaptive control methods [26]

Since extremum seeking control has better features andperformance compare to PampO which is the best knownMPPT algorithm in this paper the improvement of betterthan the best MPPT algorithm has been investigated

2 Basic Regular Extremum Seeking Algorithm

As shown in Figure 5 ESC algorithm employs a slow periodicperturbation sin(120596119905) which is added to the estimated signal120579 If the perturbation is slow enough the plant appears as astatic map 119910 = 119891(120579) and its dynamics do not interfere withthe peak seeking scheme [13] If 120579 is on either side of 120579lowast whichis the optimal point the perturbation signal 119886 sin(120596119905) willcreate a periodic response of 119910 which is either in phase orout of phase with 119886 sin(120596119905)The high-pass filter eliminates theldquoDC componentrdquo of 119910Thus 119886 sin(120596119905) and high-pass filter will

Non linear map

Lawpassfilter

Highpassfilter

120574

s

120594

120579

sin(120596t)120596t)

J

y = f(120579)

asin(

120592

Figure 5 Extremum seeking algorithm scheme

be approximately two sinusoids which are in phase if 120579 lt 120579lowast

or out of phase if 120579 gt 120579lowast

The integrator 120579 = (120574119904)120594 approximates the gradient

update law 120579 = 119896(119886

22)(119891)

1015840(120579) which tunes

120579 to 120579lowast [13]

System of Figure 5 can be summarized in mathematicalequations as follows

119910 = 119891 (120579 + 119886 sin (120596119905))

120579 = minus120574120594

120594 = 120592 lowast Lminus1HLPF (119904)

120592 = [119910 lowast Lminus1HHPF (119904)] sin (120596119905)

(1)

where lowast is the convolution operator and Lminus1 is the inverseLaplace transform operator The transfer functions for HHPFand HLPF in the regular SISO ESC scheme are 119904(119904 + 120596

ℎ) and

120596119897(119904 + 120596

119897) respectively where 120596

119897lt 120596 lt 120596

ℎ[25] This model

will be used for stability analysis in this paperIn the following sections after introducing BICO filter

[27 28] the advantages of using this filter in the ESCalgorithm from the stability and robustness point of view willbe discussed

3 BICO Extremum Seeking Control MPPT

In this section we are going to introduce a filter which wasstrongly favored by Bode [29] called Bodersquos ideal cutoff


minus80

minus70

minus60

minus50

minus40

minus30

minus20

minus10

10

Gai

n (d

B)

Frequency (rads)

HP BICO filter gain HP filter gain

0

10minus1

100

101

102

Figure 6 High-pass BICO and regular high-pass filters

characteristic (BICO) filter [27 28] The general transferfunction of low-pass BICO filter is

119866LP-BICO (119904) =119896

(119904120596119888+ radic1 + (119904120596

119888)2

)

119903 119903 isin R

+ (2)

Its corresponding time response in the special case (120596119888=

1 rads) was derived byOberhettinger and Baddi [30] but thetime response of (2) is presented as

119892BICOLP(119905) = 119896

119903119869119903(120596119888119905)

120596119888119905

(3)

where 119869119903is the 119903th order Bessel function

By replacing 119904 by 1119904 high-pass BICO filter is obtainedFigure 6 compares the frequency response of high-pass BICOfilter and regular high-pass filter (119904(119904 + 120596

119888)) with the same

cutoff frequency of 120596119888= 10 rads

As seen in this figure BICO filter has a sharp edgein its cutoff frequency This great feature causes almostno attenuation for frequencies higher than 120596

119888and a large

attenuation in the lower frequencies Therefore the behaviorof this filter is close to an ldquoidealrdquo filter This sharp edgepresents in the low-pass BICO filter By combining high-passand low-pass BICO filters the band-pass BICO filter withsharp edges in both sides can be obtained as well

4 Stability Analysis of ExtremumSeeking Control

41 Mathematical Modeling of ESC Scheme The stabilityanalysis of ESC algorithmhas been investigated in [13 31ndash34]In these literatures traditional extremum seeking control withregular first-order filters has been considered In this paperwe have considered the similar stability analysis approach as[34] to compare the stability and robustness of BICO-ESC

and regular SISO ESC The difference between this work and[34] is that in this paper the stability analysis has been doneby considering both low-pass and high-pass filters in the ESCstructure but in [34] authors have investigated the stability ofsimplified ESC with only low-pass filter in its structure

According to the previous discussion the nonlinear mapin the ESC scheme is considered to be concave and is assumedto have only one extremum point Since in the PV systemapplicationsMPPT is employed to extractmaximumamountof power from PV panels therefore extremum point in thiscase is maximum point (120579lowast) where at this point 120597119891(120579lowast)120597120579 =0 and also 1205972119891(120579lowast)1205971205792 lt 0 In ESC algorithm the output ofthe nonlinear map is

119910 = 119891 (120579 + 119886 sin (120596119905)) (4)

where 119886 and 120596 are the amplitude and angular frequency ofperturbation signal Since the perturbation signal assumes tobe small therefore the Taylor expansion of (4) is

119910 = 119891 (120579) +

119889119891 (120579)

119889120579

119886 sin (120596119905) +HOT (5)

where HOT stands for higher order terms and 120579 is the

approximation of 120579lowast The high-pass filtered signal will be

119869 ≃ Lminus1

119904

119904 + 120596ℎ

lowast 119891 (120579)

+ Lminus1

119904

119904 + 120596ℎ

lowast

119889119891 (120579)

119889120579

119886 sin (120596119905)

(6)

High-pass filter acts as a derivative opereator in serieswith a low-pass filter (119904(1(119904 + 120596

ℎ))) By multiplying the

modulation signal to the resulted signal from the high-passfilter and passing the modulated signal through the low-passfilter and since 120596

119897lt 120596ℎ the output signal of integrator will be

120579 ≃

119886120574

2

Lminus1

1

119904

lowast Lminus1

120596119897

119904 + 120596119897

lowast

119889

119889119905

(

119889119891 (120579)

119889120579

)

(7)

Under the assumptions that the amplitude of the sinu-soidal perturbation is small and the harmonics of high-passfilter are attenuated by low-pass filter output of low-pass filteris proportional to the gradient of the nonlinear map withrespect to its input and time Therefore in the neighborhoodof the extremum point the amplitude of the estimated signalis small since the gradient is small It can be seen that thisamplitude depends on 120574 and 119886

42 Stability Analysis of Averaged ESC Scheme Theaveragingmethod is typically used to analyze the periodic steady statesolutions of weakly nonlinear systems Since the amplitude ofperturbation in the ESC scheme is small this system can be


evaluated by its averaged model The averaged form of signal119909(119905) is

119909 (119905) =

1

119879

int

119879

0

119909 (119905) 119889119905 (8)

where 119879 = 2120587120596 Therefore the averaged model of ESCscheme is

120579 = 120579

119910 = 119891(120579) = 119891(

120579) = 119891 (120579)

120579 =

120574119886

2

119889

119889119905

(

119889119891 (120579)

119889120579

) lowast Lminus1HLPF lowast L

minus1

1

119904

(9)

On the other hand in the neighborhood of the extremumpoint 119910(119905) can be approximated as

119910 ≃ 119891 (120579lowast) +

119889119891 (120579)

119889120579

10038161003816100381610038161003816100381610038161003816120579=120579lowast

(120579 minus 120579lowast)

+

1

2

1198892119891 (120579)

119889120579

100381610038161003816100381610038161003816100381610038161003816120579=120579lowast

(120579 minus 120579lowast)2

(10)

If the difference between the extremum point and aver-aged point is defined by 120579 = 120579minus120579

lowast and since 119889119891(120579)119889120579|120579=120579lowast =

0 thus

119910 ≃

1

2

1198892119891(120579)

1198891205792

100381610038161003816100381610038161003816100381610038161003816120579=120579lowast

1205792 (11)

By defining (12)(1198892119891(120579)119889120579

2)|120579=120579lowast = 119870 then

(119889119891(120579)119889120579)|120579=120579lowast = 2119870120579 Therefore (119889119889119905)(119889119891(120579)119889120579)|

120579=120579lowast =

2(119889119870119889119905)120579lowast Substituting this relationship in (9) gives

120579 = (120574119886120579

lowast 119889119870

119889119905


minus1

1

119904

(12)

Without loss of generality and by assuming that 119889119870119889119905is the output signal from a high-pass filter H

119867119875119865 which its

cutoff frequency is higher than cutoff frequency of the low-pass filter H

119871119875119865 then (12) can be rewritten as

120579 = (120574119886120579

lowast119870)Lminus1HHPF lowast L

minus1HLPF lowast L

minus1

1

119904

(13)

This system can be considered as a feedback systemas shown in Figure 7 which 119870 can be considered as aperturbation The loop gain of this system depends on thedemodulation gain 119886 the integral gain 120574 and the curvatureof nonlinear map119870 This result completely matches with theresults obtained from other analysis method in the simplifiedcase [13 31ndash33]

K120574a

1

sHLP (s) HHP(s)

Figure 7 Averaged BICO-ESC scheme

5 Analysis of ESC MPPT on PV Panels

51 PV Cell Model The PV cell 119881-119868 curve V = 119891(119894 119866) ismodeled using the light emitting diode equations [3]

119868 = 119868119871minus 119868OS (exp(

119902

119860119896119861119879

) (119881 + 119877119868) minus 1)

119868OS = 119868OR (119879

119879119877

) exp(119902119864119866

119860119896119861

(

1

119879119877

minus

1

119877

))

119868119871=

119866

1000

(119868SC + 119870119879119868 (119879 minus 119879119877))

119881 =

119860119896119861119879

119902

ln(119868119871 minus 119868119868OS

+ 1) minus 119877119868

(14)

Values and definitions for further simulations are shownin Table 2

52 BICO ESC vsersus Regular SISO ESC In order tocompare the qualitative behavior of BICO MPPT with theregular SISO ESC MPPT the Root-Locus (RL) analysis hasbeen chosen Clearly other stability analysis methods areapplicable at this point

Since Root-Locus (RL) method is serving for linearsystems it is important to point out that the curvatureconstant 119870 is an uncertain parameter which depends ondifferent factors like the PV panel manufacturer and weatherconditions According to [33] 119870 isin [minus05 minus5]

As seen in Figure 7 the characteristic polynomial ofaveraged ESC system is 1 + 119886120574119870HHPFHLPF and to analyzethe behavior of this system Root-Locus (RL) analysismethodcan be employed Since there is no command in MATLABto plot Root-Locus (RL) for BICO transfer function thisequation has been solved for different 119886120574 and the roots havebeen plotted To compare the behavior of regular SISO ESCwith BICO ESC the roots for this system has been plottedwith the same method instead of using ldquorlocusrdquo command

Figure 8 shows the comparison between the root locusof averaged ESC using BICO and regular first-order filtersIn this root locus plot the constant gain is assumed to be119886120574120579lowast= 1 and cutoff frequencies in both filters are assumed

to be 120596119888= 10 rads Clearly by using first-order filters in

the averaged ESC scheme for some values of 119870 system hascomplex poles which cause an oscillatory behavior in theresponse of the system On the other side by using BICOfilters roots of the characteristic polynomial (system poles)are always real numbers for any value of119870


Table 2 Values of the considered PV model [3]

119879119877

298 Reference temperature119868OR 225119890 minus 6 Reverse saturation curent at 119879 = 119879

119877

119868SC 32 Short circuit current119864119866

138119890 minus 19 Silicon band gap119860 16 Ideality factor119896119861

138119890 minus 23 Boltzmanrsquos constant119902 16119890 minus 19 Electron charge119877 001 Resistance119870119879119868

08 119868SC Temperature coefficient

0

0

1

2

3

4

Real

Imag

e 0

0

2

BICO ESCo Regular ESC

minus1

minus2

minus2

minus3

minus4

minus4

minus350 minus300 minus250 minus200 minus150 minus100 minus50

minus10minus20

Figure 8 Root locus of the averaged BICO and the regular first-order low-pass filters

Besides the location of the poles as can be seen in thisfigure for the same value of 119870 BICO filter has farther poleswith respect to the origin compared to the first-order filtercase Therefore the bandwidth of BICO filter is higher thanthe first-order filter with similar cutoff frequency Higherbandwidth means faster response for BICO ESCThis featureis justifiable by looking at the frequency response of bothfilters Since the bandwidth of BICO is close to the bandwidthof an ideal filter BICO ESC becomes the fastest achievableMPPT algorithm

6 Simulation Illustrations

Figures 9 and 10 show the maximum power point of aPV panel with the defined parameters in Table 2 As canbe seen in these figures environmental conditions andespecially variations in the sun irradiation will change thenonlinear behavior of PV panels Shadows cloudy or dustyweather and temperature variations cause moving of optimaloperating point in PV systems When temperature increasesthe maximum output power of PV panels decreases and vice

0 5 10 15 20 250

5

10

15

20

25

30

35

40

45

50

55

Voltage (V)

Pow

er (W

)

= 30∘C

= 25∘C

= 20∘CTT

T

Figure 9 119875-119881 chart of considered PV model for different tempera-tures (irradiation = 1000Wm2)

versa In Figure 9 the variations of optimal operating point bychanging the environmental temperature from 20∘C to 30∘Care illustrated As Figure 10 presents variation of peak outputpower happens in the wider range when sun irradiationchanges Generally in PV panels when sun irradiationincreases the output power increases but when temperatureincreases the maximum extractable power reduces

To compare the proposed MPPT method which is calledBICO MPPT with the ESC MPPT the working conditionsof both algorithms have been defined to be same Forall simulations the ambient temperature is 25∘C and theirradiation is assumed to be 1000Wm2 Also the cutofffrequency of high-pass filters is 120596

ℎ= 100 rads and for

low-pass filter this frequency is 120596119897

= 50 rads Underthese conditions from Figure 9 the maximum amount ofpower which can be extracted from the simulated PV panelis 48 Watts and this maximum happens around 17 VoltsTo implement the BICO MPPT the discrete approxima-tion of this filter in MathWorks Incrsquos website has beenused

Figure 11 shows the outputs of extremum seeking algo-rithms resulted from two different MPPTmethods Figure 12


0 5 10 15 20 250

5

10

15

20

25

30

35

40

45

50

55

Voltage (V)

Pow

er (W

)

G = 1100Wm2

G = 1000Wm2

G = 900Wm2

Figure 10 119875-119881 chart of considered PV model for different irradia-tions (temperature = 25∘C)

0 2 4 6 8 100

5

10

15

20

25

30

35

40

45

50

Time (s)

Pow

er (W

)

ESC MPPTBICO MPPT

Figure 11 Comparison of power tracking in BICOMPPT and ESCMPPT

represents the maximum voltage tracking in BICO MPPTalgorithm and ESC with first-order filters As expected andproved before BICO MPPT converges to the peak powerpoint two times faster than the regular ESCMPPT algorithmFaster convergence speed is due to the higher bandwidth ofthe averaged BICO system

Figures 13 and 14 show the performance of the proposedMPPT approach compare to the ESC in the presence of whitenoise As can be seen in these figures in the presences ofa white noise (noise power = 004) which is consideredas the variations in the nonlinear map behavior 119870 BICO

0 2 4 6 8 100

2

4

6

8

10

12

14

16

18

Time (s)

Volta

ge (V

) and

curr

ent (

A)

ESC voltageBICO voltage

Figure 12 Comparison of voltage tracking in BICOMPPT and ESCMPPT

0 2 4 6 8 100

5

10

15

20

25

30

35

40

45

50

Time (s)

Pow

er (W

)

ESC MPPTBICO MPPT

Figure 13 Performance of BICO MPPT and ESC MPPT in thepresence of a white noise with noise power 004

MPPTpreforms better than ESCMPPT from attenuation andtracking point of view

Another parameter which is considered in these simula-tions is the robustness of these two methods to the variationsof the system gain Figures 15(a) and 15(b) illustrate theperformance of BICO MPPT and regular ESC MPPT to thegain variations respectively From these figures it can beconcluded that BICOMPPT tolerates higher gain variationsbut by increasing the gain of the integrator in ESC MPPT itstarts oscillating


0 2 4 6 8 100

2

4

6

8

10

12

14

16

18

20

Time (s)

Volta

ge (V

) and

curr

ent (

A)


Figure 14 Voltage tracking of BICO MPPT and ESC MPPT in thepresence of a white noise with noise power 004

7 Conclusions

The ubiquity of usingMPPT in the renewable energy systemshas stimulated the persistent development of various MPPTalgorithms According to the comparison in this paper oneof the most popular methods of maximum power trackingmethod is the PampO method However this method fails inthose situations when there is a need to track the MPPsrapidly

ESC is a new robust MPPT algorithm which carries allPampOrsquos benefits and improves its weaknesses In this paper anovel ES algorithm has been presented which is called BICOMPPTThis algorithm employs BICO filter in its structure toimprove the ESC rise time response even further BICO filteris a forgotten part of Bodersquos research that for the first time inthis paper was used for PV applications

BICO filter is the closest filter to the ideal filter andthis feature of BICO improves the performance of ESCalgorithm Also the advantages of BICO has been discussedin this paper As discussed in this paper using BICO in theESC structure increases the bandwidth of the system whichimproves the system response

In addition applying BICO filter to the ESC algorithmmoves all the roots of characteristic polynomial of averagedBICO ESC system to the real axis Consequently this systemhas no pole with imaginary part and therefore theoreticallythe system has no oscillatory response by increasing itsintegrator gain These poles however can have imaginarypart by increasing the gain of the system in the regular ESCThis feature increases the robustness of the BICO MPPTcompared to regular ESC

As can be seen in the results BICO MPPT not only canfollow themaximumpower point faster than regular ESC butalso it shows more robustness in the presence of disturbancein the system or gain variations

0 2 4 6 8 10

0

10

20

30

40

50

Time (s)

Pow

er (W

)

minus10

Gain = 30

Gain = 55

Gain = 75

Gain = 95

(a)

0 2 4 6 8 10

0

10

20

30

40

50

Time (s)

Pow

er (W

)

minus10

Gain = 30

Gain = 55

Gain = 75

(b)

Figure 15 Sensitivity of BICO MPPT and ESC MPPT to the gainvariation of integrator



References

[1] B Bose P Szczesny and R Steigerwald ldquoMicrocomputer con-trol of a residential photovoltaic power conditioning systemrdquoIEEE Transactions on Industry Applications vol 21 no 5 pp1182ndash11191 1985

[2] R A Mastromauro M Liserre T Kerekes and A DellrsquoAquilaldquoA single-phase voltage-controlled grid-connected photovoltaicsystem with power quality conditioner functionalityrdquo IEEE


Transactions on Industrial Electronics vol 56 no 11 pp 4436ndash4444 2009

[3] S L Brunton C W Rowley S R Kulkarni and C ClarksonldquoMaximum power point tracking for photovoltaic optimizationusing ripple-based extremum seeking controlrdquo IEEE Transac-tions on Power Electronics vol 25 no 10 pp 2531ndash2540 2010


[5] J JNedumgatt K B Jayakrishnan SUmashankarDVijayaku-mar and D P Kothari ldquoPerturb and observe MPPT algorithmfor solar PV systems-modeling and simulationrdquo in Proceedingsof the Annual IEEE India Conference Engineering SustainableSolutions (INDICON rsquo11) pp 1ndash6 December 2011

[6] R Alonso P Ibanez V Martınez E Roman and A Sanz ldquoAninnovative perturb observe and check algorithm for partiallyshaded PV systemsrdquo in Proceedings of the 13th European Con-ference on Power Electronics and Applications (EPE rsquo09) pp 1ndash8September 2009

[7] N Femia G Petrone G Spagnuolo and M Vitelli ldquoOptimiza-tion of perturb and observe maximum power point trackingmethodrdquo IEEE Transactions on Power Electronics vol 20 no4 pp 963ndash973 2005

[8] G J Kish J J Lee and P W Lehn ldquoModelling and controlof photovoltaic panels utilising the incremental conductancemethod for maximum power point trackingrdquo IET RenewablePower Generation vol 6 no 4 pp 259ndash266


[10] H Malek S Dadras Y Q Chen R Burt and J Cook ldquoMaxi-mum power point tracking techniques for efficient photovoltaicmicrosatellite power supply systemrdquo in Proceeding of the SmallSatellite Conference Logan Utah USA 2012

[11] H Malek S Dadras and Y Q Chen ldquoA fractional order max-imum power point tracker Stability analysis and experimentsrdquoin Proceeding of the IEEE 51st Annual Conference on DecisionandControl (CDC rsquo12) pp 6861ndash6866MauiHawaii USA 2012

[12] R Faranda and S Leva ldquoEnergy comparison of MPPT tech-niques for PV systemsrdquoWSEAS Transactions on Power Systemsvol 3 no 6 pp 1ndash6 2008

[13] M Krstic and H-H Wang ldquoStability of extremum seekingfeedback for general nonlinear dynamic systemsrdquo Automaticavol 36 no 4 pp 595ndash601 2000

[14] M Guay D Dochain and M Perrier ldquoAdaptive extremumseeking control of continuous stirred tank bioreactors withunknown growth kineticsrdquo Automatica vol 40 no 5 pp 881ndash888 2004

[15] G Bastin D Nesic Y Tan and I Mareels ldquoOn extremumseeking in bioprocesses with multivalued cost functionsrdquoBiotechnology Progress vol 25 no 3 pp 683ndash689 2009

[16] S Drakunov U Ozguner P Dix and B Ashrafi ldquoABS controlusing optimum search via sliding modesrdquo IEEE Transactions onControl Systems Technology vol 3 no 1 pp 79ndash85 1995

[17] D Popovic M Jankovic S Magner and A Teel ldquoExtremumseekingmethods for optimization of variable cam timing engineoperationrdquo in Proceeding of the American Control Conferencepp 3136ndash3141 Denver Colo USA June 2003

[18] K S Peterson and A G Stefanopoulou ldquoExtremum seekingcontrol for soft landing of an electromechanical valve actuatorrdquoAutomatica vol 40 no 6 pp 1063ndash1069 2004

[19] H-H Wang S Yeung and M Krstic ldquoExperimental applica-tion of extremum seeking on an axial-flow compressorrdquo IEEETransactions on Control Systems Technology vol 8 no 2 pp300ndash309 2000

[20] C G Mayhew R G Sanfelice and A R Teel ldquoRobustsource-seeking hybrid controllers for autonomous vehiclesrdquo inProceedings of the American Control Conference (ACC rsquo07) pp1185ndash1190 New York NY USA July 2007

[21] E Biyik and M Arcak ldquoGradient climbing in formation viaextremum seeking and passivity-based coordination rulesrdquo inProceedings of the 46th IEEEConference onDecision and Control(CDC rsquo07) pp 3133ndash3138 New Orleans La USA December2007

[22] P Ogren E Fiorelli and N E Leonard ldquoCooperative controlof mobile sensor networks adaptive gradient climbing ina distributed environmentrdquo IEEE Transactions on AutomaticControl vol 49 no 8 pp 1292ndash1302 2004

[23] P M Dower P M Farrell and D Nesic ldquoExtremum seekingcontrol of cascaded Raman optical amplifiersrdquo IEEE Transac-tions on Control Systems Technology vol 16 no 3 pp 396ndash4072008

[24] J Sternby ldquoExtremum control systems an area for adaptivecontrolrdquo in Proceedings of the International Symposium onAdaptive Systems vol 24 pp 151ndash160 San Francisco Calif USA1980

[25] K B Ariyur and M Krstic Real-Time Optimization byExtremum-Seeking Control Wiley-Interscience 2003

[26] K J Astrom and B Wittenmark Adaptive Control Addison-Wesley 2nd edition 1995

[27] Y Luo T Zhang B Lee C Kang and Y Q Chen ldquoDisturbanceobserver designwith bodersquos ideal cut-offfilter in hard-disc-driveservo systemrdquoMechatronics vol 23 no 7 pp 856ndash862 2013

[28] L E Olivier I K Craig and Y Q Chen ldquoFractional orderand BICO disturbance observers for a run-of-mine ore millingcircuitrdquo Journal of Process Control vol 22 no 1 pp 3ndash10 2012

[29] T T Hartley and C F Lorenzo ldquoOptimal fractional-orderdampingrdquo in Proceedings of the ASME Design EngineeringTechnical Conferences and Computers and Information in Engi-neering Conference pp 737ndash743 Chicago Ill USA September2003

[30] Oberhettinger and Baddi Tables of Laplace TransformsSpringer Berlin Germany 1973

[31] Y Tan D Nesic and I Mareels ldquoOn stability properties of asimple extremum seeking schemerdquo in Proceedings of the 45thIEEE Conference on Decision and Control (CDC rsquo06) pp 2807ndash2812 December 2006

[32] Y Tan D Nesic and I Mareels ldquoOn non-local stabilityproperties of extremum seeking controlrdquo Automatica vol 42no 6 pp 889ndash903 2006

[33] R Leyva C Olalla H Zazo et al ldquoMppt based on sinusoidalextremum-seeking control in pv generationrdquo International Jour-nal of Photoenergy vol 98 no 4 pp 529ndash542 2011

[34] R Leyva P Artillan C Cabal B Estibals and C AlonsoldquoDynamic performance of maximum power point trackingcircuits using sinusoidal extremum seeking control for photo-voltaic generationrdquo International Journal of Electronics vol 98no 4 pp 529ndash542 2011

Research ArticleSimulation Analysis of the Four Configurations of SolarDesiccant Cooling System Using Evaporative Cooling in TropicalWeather in Malaysia

M M S Dezfouli1 S Mat1 G Pirasteh2 K S M Sahari3 K Sopian1 and M H Ruslan1

1 Solar Energy Research Institute (SERI) Universiti Kebangsaan Malaysia 43600 Bangi Selangor Malaysia2 Department of Mechanical Engineering Faculty of Engineering University of Malaya 50603 Kuala Lumpur Malaysia3 Department of Mechanical Engineering Universiti Tenaga Nasional Jalan IKRAM-UNITEN 43000 Kajang Selangor Malaysia

Correspondence should be addressed to M M S Dezfouli salehisolargmailcom

Received 29 May 2013 Revised 13 November 2013 Accepted 29 November 2013 Published 2 February 2014


Copyright copy 2014 M M S Dezfouli et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

A high demand for air conditioning systems exists in hot and humid regions because of the warm climate during the yearThe highenergy consumption of conventional air conditioning system is the reason for our investigation of the solar desiccant cooling systemas an energy-efficient cooling system Four model configurations were considered to determine the best configuration of a solardesiccant cooling system one-stage ventilation one-stage recirculation two-stage ventilation and two-stage recirculation Thesemodels were stimulated for 8760 hr of operation under hot and humid weather in Malaysia Several parameters (ie coefficientof performance or COP room temperature and humidity ratio and the solar fraction of each system) were evaluated by detectingthe temperature and humidity ratio of the different points of each configuration by TRNSYS simulation The latent and sensibleloads of the test room were 0875 kW and 2625 kW respectively By investigating the simulation results of the four systems theventilation modes were found to be higher than the recirculation modes in the one- and two-stage solar desiccant cooling systemsThe isothermal dehumidification COP of the two-stage ventilation was higher than that of the two-stage recirculation Hence thetwo-stage ventilation mode desiccant cooling system in a hot and humid area has higher efficiency than the other configurations

1 Introduction

Because of the high electrical consumption of conventionalvapor compression systems the solar desiccant cooling sys-tem has been considered as one of the promising alternativesto cooling air where the sensible and latent heats of air areremoved separately [1]The first desiccant cooling systemwaspresented by Pennington in 1955 [2] Generally depending onthe dehumidification material used two kinds of desiccantcooling system exist solid and liquid Many scientists andresearchers have studied both kinds of desiccant coolingusing renewable energy [3] The common materials usedin the solid and liquid desiccant cooling systemwere silicagel and liquid water-lithium chloride The desiccant cool-ing system includes three main units namely desiccantheat-source and cooling units [4] The elements of each

unit are designed depending on the weather conditionsThe components of the basic cycle consist of the soliddesiccant wheel (DW) as dehumidifier evaporative cooleras the cooling unit and the heat recovery wheel as theheater and cooler in the regeneration and process air sidesrespectively In recent years according to the Penningtoncycle different types of solid desiccant cooling cycles havebeen designed in the world by different researchers Thecoefficient of performance (COP) of the desiccant coolingsystem regeneration temperature mass flow rate fresh airand relative humidity of the supply air are the importantparameters considered by the researchers Haddad et al [5]studied the simulation of a desiccant-evaporative coolingsystem for residential buildings They found that the useof solar energy for regeneration of the DW can provide asignificant portion of the auxiliary thermal energy needed



Fong et al [6] designed a simulation model (TRNSYS) ofan integrated radiant cooling by absorption refrigerationand desiccant dehumidification Salehi Dezfouli et al [7]investigated the solar hybrid desiccant cooling system in hotand humid weather in Malaysia They found that a solarhybrid solid desiccant cooling system provided considerableenergy savings in comparison with the conventional vaporcompression in hot and humid area

Researchers have studied and compared the COP ofventilation and recirculation cycles Jain et al [8] evaluatedventilation and recirculation cycles based on weather data inIndia to determine the effectiveness of evaporative coolerson the COP of a cooling system Boundoukan et al [9]evaluated a comparison between ventilation and recircu-lation procedure modes critical values were achieved forcycle component performance specifics required to yield agiven supply air temperature condition Mehmet et al [10]evaluated First low and second Low for COPs of ventilationand recirculation cycles Panaras et al [11] investigated thebehavior of the COP of the ventilation and recirculationdesiccant cooling systems in terms of the air flow rateThey found that the ventilation cycle presents a higher COPand the required air flow rate is similar for both cyclesSphaier et al [12] studied the simulation of ventilation andrecirculation desiccant cooling systems to analyze the impactof cycle component characteristics on the overall systemperformance They concluded that although all componentscan influence the COP the COP values of the recirculationcycle are lower than the ventilation cycle

With regard to the increase in temperature of the processair after dehumidification and the limitation capacity ofthe dehumidification by the DW the required regenerationtemperature is high especially in hot and humid regions[13ndash16] Many researchers have attempted to determine thebest solution for the high required regeneration temperatureproblem in one-stage desiccant cooling systems The isother-mal dehumidification process is a great solution to solve thisproblem The required cool process air after dehumidifica-tion and reduced regeneration temperature can be obtainedusing isothermal dehumidification technology or multistagedehumidification process Meckler [17] proposed a two-stagesolid desiccant air conditioning system integrated with anHVAC system A two-stage system was also introduced byHenning [18] A novel rotary desiccant cooling cycle withisothermal dehumidification and regenerative evaporativecooling was proposed by La et al [19] They found that theisothermal dehumidification was relatively lower tempera-ture requirement for the heat source because the regenerationtemperature is reduced from 80∘C to approximately 60∘CGe et al [20] evaluated the performance of a two-stagerotary desiccant cooling (TSRDC) system The requiredregeneration temperature of the TSRDC system is low andthe COP of the system is high La et al [21] performed atheoretical analysis of a solar-driven TSRDC system assistedby vapor compression air conditioning They concludedthat the solar-driven two-stage desiccant cooling system isreliable and energy efficient Li et al [22] investigated theTSRDCheating system driven by evacuated glass tube solarair collectors Their experimental results indicated that the

average thermal COP in the cooling cycle was 097 and thecooling capacity was in the range from 163 kW to 256 kWunder hot and humid ambient conditions Li et al [23]carried out experimental investigation on a one-rotor two-stage desiccant coolingheating system driven by solar aircollectors The average thermal COP in the cooling cycle is095 in hot and humid climate conditions

Given the use of ambient air in solar desiccant cooling sys-tems the ambient conditions of each region have a significantimpact on system performance Although various studiesand publications on the one-stage and two-stage desiccantcooling systems have been based on differentweather data nosuch study compares one-stage and two-stage solar coolingtechnologies that have been undertaken in Malaysia Thehigh demands of air conditioning systems because of hot andhumid weather during the year as well as the high energyconsumption of conventional AC systems have resulted inthe investigation of the solar desiccant cooling system as anefficient air conditioning system in Malaysia The presentpaper presents a comparison simulation study among thefour configurations of solar desiccant cooling system in hothumid weather in Malaysia

2 Methodology

The comparison performance between the one-stage andtwo-stage solar desiccant cooling systems under the venti-lation and recirculation modes is used as the methodologyin this study The one-stage and two-stage desiccant coolingsystems are explained in the following sections

21 Case Study Description Four kinds of solar desiccantcooling systems were considered to supply air for a testroom in a technology park (UKM) in Malaysia The one-stage desiccant cooling system under two modes of ventila-tion and recirculation and the two-stage desiccant coolingsystem under two modes of ventilation and recirculationwere considered to evaluate the four types of desiccantcooling systemThe latent and sensible loads of the test roomwere 0875 and 2625 kW respectively The cooling capacitywas 1 ton According to the American Society of HeatingRefrigerating and Air Conditioning Engineers (ASHRAE)comfort condition indoor condition designs should consistof temperature of 25∘C relative humidity of 50 and humid-ity ratio of 00098 kgkg [24] According to the Malaysianweather in outdoor condition design the temperature shouldbe 30∘C and the humidity ratio should be 00200 kgkgThe cooling system includes four main parts namely DWas dehumidifier heat recovery wheel evaporative cooler ashumidifier and solar-evacuated tube collector as heat sourceSimulation design using the TRNSYS software was carriedout to investigate the solar desiccant cooling in the differentconfigurations to find the best operation of the componentsand the cooling system based on the hot and humid weatherin Malaysia In this study one stage (ventilation and recircu-lation modes) and two stages (ventilation and recirculationmodes) were simulated under the same effectiveness of thecomponents


Table 1 Specific assumptions used in simulation

Item FactorDW Type 683 119865

1= 01 119865

2= 007 power consumption 02 kW and speed of rotor rotation 8 rh

DEC Type 506c saturation efficiency = 08 and power consumption 01 kW

HRW Type 760b sensible efficiency = 756 latent efficiency = 0 and power consumption017 kW

HX Type 91 effectiveness = 07 specific heat of hot side 419 kjkgsdotK and specific heat of hotside 105 kjkgsdotK

Pump Type 3b maximum power consumption 60W and water flow rate 180 kghBlower Type 112a motor efficiency = 085 and maximum air flow rate 8704 kghRoom (load) Type 690 sensible load 2625 latent load 0875 SHR = 075Weather data Type 109-TMY2 all of weather data is based on Kuala Lumpur (Malaysia)Solar collector Type 71 evacuated tube area of collector 20m2 and the efficiency of collector 120578 060Hot water tank Type 4c the loss coefficient 055Wm2 KBackup heater Type 6 the maximum heating rate 10 kW119866 The one year average solar radiation of Malaysia (2012) 500Wm2

In the simulation process the regeneration temperaturecan be detected by adjusting and setting the required humid-ity ratio after the dehumidification process in DW as thehumidity ratio set pointThree set points (ie 00100 00080and 00050 kgkg) for the humidity ratio were considered forthe one-stage ventilation and recirculation modes to evaluatethe behavior of regeneration temperatures and COPs Ina comparative evaluation between one-stage and two-stagesolar desiccant cooling systems the humidity ratio set pointsfor the first and second DW in the two-stage ventilationand recirculationmodes were 00100 kgkg and 00050 kgkgrespectively The properties of the simulation componentsused in the four configurations are shown in Table 1The solardesiccant cooling system was divided into two parts namelythe one-stage and the two-stage that are explained in Sections22 and 23 respectively The TRNSYS simulation was vali-dated by the measurement data from the experimental setupof the one-stage ventilation solar desiccant cooling systemThis system had been installed in the technology park at theUniversiti Kebangsaan Malaysia [7] After the validation ofTRNSYS it was used to simulate the four configurations ofthe solar desiccant cooling system

22 One-Stage Solar Desiccant Cooling System The one-stagesolar desiccant cooling includes the ventilation and recircu-lation modes The components of the one-stage desiccantcooling are oneDW one heat recoverywheel two evaporativecoolers one heat exchanger one auxiliary heater and anevacuated tube collector

The air flow rate of the one-stage ventilation and recircu-lation in the process and regeneration sides of the system was8704 kghr

221 One-Stage Ventilation Mode Figure 1 shows that theventilation mode of the solar desiccant cooling is an opencycle that provides supply air to the room from ambient air

Backup heaterSolar

collec

torWater tank Pump

Pump

HX

HRWDW DEC 1Room

DEC 2

1 2 3 4

56789

SA

Exhaust air RA

Ambient air

Figure 1 Solar desiccant cooling in ventilation mode schematic

Backup heater

Water tank Pump

Pump

HX

HRWDW DEC 1 Room

DEC 21 2 3 4

56789

SARA

Exhaust air

Ambient air

Solar co

llector

Figure 2 Solar desiccant cooling in recirculation mode schematic

The returning air from the room after few processes isconverted to ambient temperature as exhaust air Thereforetwo kinds of air are present the process side that produces


7

DW 1 DW 2 HRW 2HRW 1DEC 1

DEC 2

Room

HX 1 HX 2

Ambient air

Exhaust air Exhaust airExhaust airEx

haus

t air

1 2 3 4 5 6SA

RA

89101112131415

Am

bien

t air

Figure 3 Schematic of two-stage solar desiccant cooling system-ventilation mode


DEC 2

Room

HX 1 HX 2

Ambient air

Exhaust air Exhaust airExhaust airExhaust air

1 2 3 4 5 6SA

RA

7891011121314

Figure 4 Schematic of two-stage solar desiccant cooling system-recirculation mode

the supply air and the regeneration side that releases thereturning air from room to ambient temperature In theprocess side first the ambient air becomes dry because ofthe DW Then the heat recovery wheel cools the air in theprocess side In the next step the air becomes cold because ofthe evaporative cooler then the air moves into the room assupply air In the regeneration side the returning air whoselatent and sensible loads were left in the room becomes coldbecause of the evaporative cooler The heat recovery wheelacts as an energy conservation wheel in the regeneration airside Then the heat from the heat exchanger is transferred tothe air In the last step in the regeneration side the hot air thusabsorbed the humidity of the DW and released it outside

222 One-Stage Recirculation Mode Figure 2 shows therecirculation mode of the solar desiccant cooling systemGenerally this system is not an open cycle The process airside in the recirculation mode is a closed loop whereas theregeneration air side is an open cycle

In the process air side the room air that includes thesensible and latent loads moves to the DW to remove thelatent load The heat recovery wheel acts as a cooler in theprocess air side In the next step the air grows cold becauseof the evaporative cooler and it moves into the room assupply air In the regeneration side the ambient air grows coldbecause of the evaporative cooler In the next step the heatfrom the solar collectors and the backup heater is transferredto the air by means of a heat exchanger Thus in the last step

in the regeneration side the hot air absorbs the humidity ofthe DW and is released as exhaust air outdoors

23 Two-Stage Solar Desiccant Cooling System The two-stagesolar desiccant cooling includes the ventilation and recircu-lation modes The components of the two-stage desiccantcooling system are two DWs two heat recovery wheelstwo evaporative coolers with different capacities two heatexchangers two auxiliary heaters and an evacuated tubecollector According to the high dehumidification capacity ofthe two-stage DW and the humid weather in Malaysia theset point of dehumidification was adjusted at a 00050 kgkghumidity ratio The air flow rate of the two-stage ventilationand recirculation in the process side was 8704 kghr and4352 kghr in the regeneration side

231 Two-Stage Ventilation Mode Figure 3 shows the two-step dehumidification of the process air in the solar desiccantcooling system using the evaporative cooling units in venti-lation mode This mode is an open cycle The air propertiesof each of the 15 points are characterized by the simulationmodel The air properties are explained in Section 3

The supply air in the process side is produced from theambient air by undergoing a few stages such as the two-step dehumidification in the DWs (2 and 4) two-step coolingin the heat recovery wheels (3 and 5) and one-step coolingin the direct evaporative cooler (6) In the regeneration airside the returning air (7) is mixed with the ambient air (8)


0

10

20

30

40

50

60

70

80

90

100

Time (hr)

AmbientRegenerationRoom (load)

Supply airProcess air after DW

Supply air

Ambient

Room (load)

Process air after DW

Regeneration

Tem

pera

ture

(∘C)

85008000750070006500600055005000450040003500300025002000150010005000

Figure 5 Temperatures (∘C) of different components versus time (hr) in ventilation desiccant cooling system

00000

00050

00100

00150

00200

00250

Hum

idity

ratio

(kg

kg)

Time (hr)AmbientSupply air

Room (load)Process air after DW

Ambient

Supply air

Room (load)


85008000750070006500600055005000450040003500300025002000150010005000

Figure 6 Humidity ratio (kgkg) of different components versus time (hr) in ventilation desiccant cooling system

before being cooled in one direct evaporative cooler Thenthe output air from the evaporative cooler (9) is divided intotwo stages with the same air flow rate In each stage air isheated in the heat recovery wheel (10 and 13) heated in thesolar heaters (11 and 14) and humidified in the DWs (12 and15)

232 Two-Stage Recirculation Mode Figure 4 shows theschematic of the two-stage solar desiccant cooling system

under the recirculation mode The process air side (1ndash6) isa closed loop and the regeneration side (7ndash14) is an opencycle Supply air is produced from the returning air in theprocess air side by undergoing a few stages such as the two-step dehumidification in the DWs (2 and 4) two step coolingin the heat recovery wheels (3 and 5) and one-step coolingin the direct evaporative cooler (6) In the regeneration sideambient air (7) is mixed from the air in the cooler (8) andthen divided into two stages for the heating and humidifying


process (8ndash11 and 8ndash14)The process air under the ventilationmode is an open loop whereas it is a closed loop under therecirculation mode

24 Determination of the COP of theDesiccant Cooling SystemThe COP of the solar desiccant cooling system can becalculated by the rate of heat extracted against the rate of thecooling system driving energy The rate of heat extracted isthe cooling capacity of the system The rate of the coolingsystemdriving energy is the total thermal energy requirementfor the cooling system generated from solar collectors and thebackup heater Therefore the COP of the system is obtainedby the following relationship [4]

COP =119876cool119876119879

(1)

241 One Stage The COP of the solar desiccant coolingsystem under the ventilation mode is defined as the ratio ofthe enthalpy change from ambient air to supply airmultipliedby the mass air flow and external heat delivered to theregeneration heat exchanger It can be written as follows

COP =119898119904(ℎ1minus ℎ4)

119898119903(ℎ8minus ℎ7)

(2)

The COP of the solar desiccant cooling under recirculationmode can be written as

COP =119898119904(ℎ1minus ℎ4)

119898119903(ℎ8minus ℎ7)

(3)

where 119898119904(kghr) is the mass flow rate of the supply air 119898

119903

(kghr) is the mass flow rate of the regeneration air and ℎ(kJkg) is the enthalpy of air

242 Two Stages The COP of the two-stage solar desiccantcooling under the ventilation mode can be written as [20]

COP =1198981199041(ℎ1minus ℎ6)

11989811990313(ℎ14minus ℎ13) + 11989811990310(ℎ11minus ℎ10)

(4)

The COP of the two-stage solar desiccant cooling under therecirculation mode can be written as

COP =1198981199041(ℎ1minus ℎ6)

11989811990312(ℎ13minus ℎ12) + 1198981199039(ℎ10minus ℎ9)

(5)


119903

(kghr) is the mass flow rate of the regeneration air and ℎ(KJkg) is the enthalpy of air

25 Solar Fraction Generally solar collectors convert solarradiation into thermal energy in solar cooling systems Con-sidering the insufficient thermal energy from solar collectorsin cloudy weather the backup heater is used to achieve therequired thermal energy for the total cooling system drivingenergy The percentage ratio of the thermal energy producedby the solar collectors to the total cooling system driving

Table 2 Air properties of one-stage solar desiccant cooling systemin ventilation mode

Points number Temperature (∘C) Humidity ratio (kgkg)1 30 002002 495 001003 318 001004 22 001375 308 001526 2495 001767 456 001768 816 001769 50 00262

energy is known as the solar fraction which can be expressedas follows [25]

SF =119876119880

119876119879

(6)

where119876119879is the total cooling system driving energy produced

by the solar collectors and the backup heater 119876119880

is thethermal energy produced by the solar collectors that can beexpressed as [26]

119876119880= 119860 times 120578 times 119866 (7)


The results of the four different systems were explained andthen compared in the next sections The temperature andhumidity ratio against time of the specified points in the foursystems were analyzed

31 Results of the One-Stage Solar Desiccant Cooling System

311 Ventilation Mode The temperature of ambient airprocess air after the DW supply air room (load) and regen-eration temperature versus time (hr) are shown in Figure 5The temperatures at different points were specified for 8760 hafter running the solar desiccant cooling system The regen-eration temperature is one of the important parameters thatplay a main role in the changes in the COP of the desiccantcooling system The regeneration temperature under theventilation mode for 00100 kgkg dehumidification (humid-ity ratio reduction from 00200 kgkg to 00100 kgkg) wasalmost 816∘C (average) Figure 6 shows the humidity ratio ofthe ambient air supply air load and process air after the DWversus time under the ventilation mode

The humidity ratio of the supply air is one of the impor-tant parameters that play a main role in the amount latentload removed from the roomby the desiccant cooling systemespecially in hot and humid weather inMalaysiaThe averagetemperature and humidity ratio of the specified points in theone-stage ventilation system for 8760 h operation are shownin Table 2 The temperature and humidity ratio of the supplyair (point 4) are 22∘C and 00137 kgkg respectively whereasthose in the room (point 5) are 3080∘C and 00152 kgkgrespectively


0030

0028

0026

0024

0022

0020

0018

0016

0014

0012

0010

0008

0006

0004

0002

0000

848076726864605652484440363228242016

Hum

idity

ratio

(kg

kg)

Temperature (∘C)

1

2

3

4

6 7

8

9

5

100 R

elativ

e hum

idity

90

8070

60

50

40

30

20

10

Figure 7 Psychometric chart of the one-stage solar desiccant cooling system under the ventilation mode

0

10

20

30

40

50

60

70

80

90

100

Time (hr)AmbientRegenerationRoom (load)


Supply air

Ambient

Room (load)


Regeneration

Tem

pera

ture

(∘C)

8500

8000

7500

7000

6500

6000

5500

5000

4500

4000

3500

3000

2500

2000

1500

1000

5000

Figure 8 Temperatures (∘C) of different components versus time (hr) in recirculation desiccant cooling system

Figure 7 shows the psychometric chart of the one-stagesolar desiccant cooling system under the ventilation modein which the air property and process air on all pointsare specified The process-air and regeneration-air sides areindicated by blue and red lines respectively Because of thedehumidification process between points 1 and 2 the airtemperature increases whereas the humidity ratio decreasesBy the cooling process between points 2 and 3 the airtemperature decreases whereas the humidity ratio of air inthe heat recovery wheel remains constant The temperaturedecreases and the humidity ratio increases because of the

evaporative cooling process between points 3 and 4 Theline between points 4 and 5 shows that the air propertychanges from supply air to the returning air inside the roomThe regeneration side includes the four stages namely theevaporative cooling process between points 5 and 6 heatingbetween points 6 and 7 heating process between points 7 and8 and process of exhausted air to outdoor between points 8and 9

312 Recirculation Mode Figure 8 shows the temperatures(∘C) of the ambient air process air after the DW supply


00000

00050

00100

00150

00200

00250

AmbientProcess air after DW

Ambient

Supply air

Hum

idity

ratio

(kg

kg)

Room (load)


Time (hr)

8500

8000

7500

7000

6500

6000

5500

5000

4500

4000

3500

3000

2500

2000

1500

1000

5000

Room (load)Supply air

Figure 9 Humidity ratio (kgkg) of different components versus time (hr) in recirculation mode

Table 3 Air properties of one-stage solar desiccant cooling systemin recirculation mode


air the room and the regeneration versus time (hr) underthe recirculation mode The regeneration temperature underthe recirculation mode for 00100 kgkg dehumidification(humidity ratio reduced from00200 kgkg to 00100 kgkg) isalmost 702∘C (average) The humidity ratios of the ambientair and process air after DW in the room and the supplyair versus time (hr) in the one-stage recirculation mode areshown in Figure 9

The average temperature and humidity ratio of thespecified points in the one-stage recirculation system for8760 h operation are shown in Table 3 The temperatureand humidity ratio of the supply air (point 4) are 214∘Cand 00136 kgkg respectively whereas those for the room(point 1) are 31∘C and 00150 kgkg respectively Figure 10shows the psychometric chart of the one-stage solar desiccantcooling system under the recirculation mode where the airproperty changes in the different parts of the solar desiccantcooling system are described In the process side the airreturning from the room to the DW for dehumidification is

shown between points 4 and 1 The ambient air temperatureis decreased by the evaporative cooling process (5 and 6)The heating process (6 and 7) heating process (7 and 8) andprocess of exhausted air to outdoor (8 and 9) are performedin the regeneration side

32 Comparison between the One-Stage Ventilation andRecirculation Modes Table 4 shows the simulation resultsof the one-stage ventilation and recirculation modes inthree different humidity ratio set points (ie 00050 00080and 00100 kgkg) The results show that by increasing thehumidity ratio set point in both modes the trend of theregeneration temperature decreased whereas the trend ofsupply air temperature return air temperature and COPincreased By increasing the humidity ratio set point becauseof the decreasing regeneration temperature the COP ofthe system increased The temperature and humidity of theroom also increased In the evolution of air conditioningsystem parameters although the COP is the most importantparameter reaching the comfortable condition of the roomis another important parameter that must be consideredFor example in ventilation mode when the humidity ratiodecreased from 000100 kgkg to 00050 kgkg the roomtemperature and humidity ratio also decreased from 3080∘Cto 2822∘Cand from00152 kgkg to 00115 kgkg respectivelyThe COP decreased from 065 to 057 The set point of00050 kgkg humidity ratio in comparison with another setpoint thus leads the system to reach near thermal comfortconditions

By comparing the temperature results of the one-stageventilation and recirculation in each humidity ratio set pointthe regeneration temperature under the ventilation modeis found to be considerably higher than that under therecirculation modeThe temperatures of the supply air under


848076726864605652484440363228242016

Temperature (∘C)

1

2

3

4

100 R

elativ

e hum

idity

90

8070

60

50

40

30

20

10

5

6 7 8

9

0030

0028

0026

0024

0022

0020

0018

0016

0014

0012

0010

0008

0006

0004

0002

0000

Hum

idity

ratio

(kg

kg)

Figure 10 Psychometric chart of one-stage solar desiccant cooling system in recirculation mode

Table 4 Air properties results of ventilation and recirculation modes based on different humidity ratio set points

Humidity ratioset point(kgkg)

Ventilation mode Recirculation mode

Reg 119879 (∘C) SA RA COP Reg 119879 (∘C) SA RA COP119879 (∘C) HR (kgkg) 119879 (∘C) HR (kgkg) 119879 (∘C) HR (kgkg) 119879 (∘C) HR (kgkg)

00050 12200 1948 00100 2822 00115 057 9690 1850 00100 2850 00115 02900080 9840 2095 00121 3024 00136 061 8030 2080 00121 3046 00135 04000100 8160 2200 00137 3080 00152 065 7020 2140 00136 3100 00150 050

0

10

20

30

40

50

60

70

80

90

100

Time (hr)AmbientRegeneration 1Room (load)

Supply airRegeneration 2

Supply air

Room (load)

Ambient

Regeneration 1

Regeneration 2

85008000750070006500600055005000450040003500300025002000150010005000

Tem

pera

ture

(∘C)

Figure 11 Temperatures (∘C) of different components versus time (hr) in two-stage ventilation desiccant cooling system


00000

00050

00100

00150

00200

00250H

umid

ity ra

tio (k

gkg

)

Time (hr)Ambient

Supply airRoom (load)Process air after DW 1Process air after DW 2

Process air after DW 2


Supply air

Room (load)

Ambient

85008000750070006500600055005000450040003500300025002000150010005000

Figure 12 Humidity ratio (kgkg) of different components versus time (hr) in two-stage ventilation desiccant cooling system

both modes are almost the same In addition the roomtemperatures under both modes are almost the same

By comparison the humidity ratio results of the ven-tilation and recirculation show that the humidity ratio ofthe room and the supply air under both modes are almostthe same The COPs under the ventilation and recirculationmodes are calculated using (1) and (2) Therefore the COPsof the desiccant cooling system under the ventilation modeare higher than those under the recirculation mode

33 Result of the Two-Stage Solar Desiccant Cooling SystemThe simulation results of the two-stage desiccant coolingsystem under the two modes of ventilation and recirculationare explained in the following sections

331 Two-StageVentilation Figure 11 shows the temperaturetrend in the different points of the two-stage solar desiccantsystem such as the temperatures of the regeneration ofthe first and second DWs room ambient air and supplyair versus time (hr) The first and second regenerationtemperatures are approximately 80∘C The humidity ratio ofthe ambient air process air after DW room and supply airversus time (hr) under the two-stage ventilation mode areshown in Figure 12

The humidity ratios of the air after the first and seconddehumidification are 00100 and 00050 kgkg respectivelyduring the 8760 hr system operation According to thetemperature and humidity ratio results under the ventilationmode the important points of the air property are detectedand shown in Table 5

Figure 13 shows the psychometric chart of the two-stagesolar desiccant cooling system under the ventilation modeThe process air side of this system includes the two-stepdehumidification process (1 and 2 and 3 and 4) by the twoDWs the two-step cooling process (2 and 3 and 4 and 5)

Table 5 Air properties of two-stage solar desiccant cooling systemin ventilation mode

Points number Temperature (∘C) Humidity ratio (kgkg)1 30 002002 537 001003 302 001004 50 000505 283 000506 1762 000967 273 001098 2971 001459 234 0016810 451 0016811 821 0016812 415 0027013 482 0016814 80 0016815 34 00280

by the two heat recovery wheels and the one-stage coolingprocess (5 and 6) by the evaporative cooler The processfrom points 1 to 5 is defined as isothermal dehumidificationThe regeneration air side includes a few processes such asthe mixing of ambient air and returning air process (7 and8) evaporative cooling process (8 and 9) two-step heatingprocess (9 and 10 and 9ndash13) by the heat recovery wheels two-step heating process (10 and 11 and 13 and 14) by the solarheater and two-step process of exhausted air outdoors (11 and12 and 14 and 15)

332 Two-Stage Recirculation The temperature trend invarious points in the two-stage recirculation solar desiccant


848076726864605652484440363228242016

Hum

idity

ratio

(kg

kg)

Temperature (∘C)

1

2

3

4

100 R

elativ

e hum

idity

9080

70

60

50

40

30

20

10

5

6

7

8

9 10 11

12

13

14

15

0030

0028

0026

0024

0022

0020

0018

0016

0014

0012

0010

0008

0006

0004

0002

0000

Figure 13 Psychometric chart of two-stage solar desiccant cooling system in ventilation mode

0

10

20

30

40

50

60

70

80

90

100

Time (hr)

AmbientRegeneration 1Room (load)


Ambient

Room (load)

Supply air

Regeneration 1

Regeneration 2

Tem

pera

ture

(∘C)

85008000750070006500600055005000450040003500300025002000150010005000

Figure 14 Temperatures (∘C) of different components versus time (hr) in two-stage recirculation desiccant cooling system

cooling system versus time (hr) is shown in Figure 14 Thehumidity ratio trend of the various points in the two-stagerecirculation system versus time (hr) is shown in Figure 15According to the temperature and humidity ratio results ofthe simulation model of the two-stage system under therecirculationmode the important points of air properties aredetected and shown in Table 6

Figure 16 shows the psychometric chart of the two-stagesolar desiccant cooling system under the recirculation modeThe process air side of this system includes several processessuch as the returning air process from the room to the firstDW (6ndash1) two-step dehumidification process (1 and 2 and

3 and 4) by the two DWs two-step cooling process (2 and3 and 4 and 5) by the two heat recovery wheels and one-stage cooling process (5 and 6) by the evaporative coolerThe regeneration air side includes several processes such asthe evaporative cooling process (7 and 8) two-step heatingprocess (8 and 9 and 8ndash12) by the heat recovery wheel two-step heating process (9 and 10 and 12 and 13) by the solarheater and two-step process of exhausted air to outdoor (10and 11 and 13 and 14)

34 Comparison Results between the Two-Stage Ventilationand Recirculation Modes The humidity ratios of the process


00000

00050

00100

00150

00200

00250H

umid

ity ra

tio (k

gkg

)

Time (hr) h amh supply airh room (load)

h process air after DW 1h process air after DW 2



Supply air

Room (load)

Ambient

85008000750070006500600055005000450040003500300025002000150010005000

Figure 15 Humidity ratio (kgkg) of different components versus time (hr) in two-stage recirculation mode

Table 6 Air properties of two-stage solar desiccant cooling systemin recirculation mode

Points number Temperature (∘C) Humidity ratio (kgkg)1 27 001092 3482 001003 273 001004 4933 000505 3074 000506 18 000967 30 002008 261 002119 413 0021110 80 0021111 587 0025312 329 0021113 50 0021114 426 00231

air after the first and second DWs under both modes are00100 and 00050 kgkg respectively By comparing thetemperature results of the two-stage ventilation and recir-culation modes the regeneration temperatures under theventilationmode are found to be higher than those under therecirculationmodeThe supply air temperature and humidityratio under the ventilationmode are 176∘C and 00096 kgkgrespectively whereas those under the recirculation modeare 18∘C and 00096 kgkg respectively The room air tem-perature and humidity ratio are 273∘C and 00109 kgkgrespectively whereas those under the recirculation modeare 27∘C and 00109 kgkg respectively The COPs underthe ventilation and recirculation modes are calculated by(3) and (4) The COPs of the ventilation and recirculation

are 106 and 043 respectively Although the cooling systemdriving energy in ventilation mode was higher than therecirculationmode the rate of cooling capacity in ventilationmode was higher than the recirculation mode The COPsof the one-stage and two-stage desiccant cooling systemsunder the ventilationmode were higher than those under therecirculation mode

35 Comparison Results of the Four Configurations of the SolarDesiccant Cooling Systems The regeneration temperaturesupply air temperature supply air relative humidity supplyair humidity ratio room air temperature room air relativehumidity room air humidity ratio COP and solar fractionof the four configurations of the solar desiccant cooling areshown inTable 7The values are average for one yearThefinalhumidity ratio set point before the evaporative cooler for thefour configurations is 0005 kgkg humidity ratio Whereasthe one-stage desiccant cooling dehumidification occurs inone step the two-stage desiccant cooling dehumidificationoccurs in two steps

The supply and return air temperatures and humidityratios of the two-stage system are less than those of theone-stage system Regarding the thermal comfort conditionparameters the results of the two-stage desiccant coolingsystem were better than the one-stage desiccant coolingsystem Given the isothermal dehumidificationin process inthe two stage desiccant cooling systems the regenerationtemperature of the two-stage ventilation and recirculationwas lower than the one stage ventilation and recirculation Bycomparing the best configuration of the one-stage and two-stage ventilation modes the two-stage ventilation solar cool-ing system was selected as the best configuration among thefour configurations because of its higher COP as well as lowerroom temperature and humidity ratio By comparing the


848076726864605652484440363228242016

Hum

idity

ratio

(kg

kg)

Temperature (∘C)

1

2

3

4

100 R

elativ

e hum

idity

9080

70

60

50

40

30

20

10

5

6

7

8 910

11

12 13

14

0030

0028

0026

0024

0022

0020

0018

0016

0014

0012

0010

0008

0006

0004

0002

0000

Figure 16 Psychometric chart of two-stage solar desiccant cooling system in recirculation mode

Table 7 Performance comparison of four desiccant cooling configurations under 0005 kgkg humidity ratio set point in hot and humidclimate (30∘C 20HR)

Type of configurations 119879 reg (∘C) SA RA COP SF ()119879 (∘C) HR (kgkg) RH () 119879 (∘C) HR (kgkg) RH ()

One-stage ventilation 12200 1948 00100 704 2822 00115 479 057 39One-stage recirculation 9690 1850 00100 750 2850 00115 471 029 50Two-stage ventilation 8210ndash8000 1760 00096 762 2730 00109 480 106 69Two-stage recirculation 8000ndash5000 1800 00096 743 2700 00109 487 043 85

solar fractions of the four configurations the solar fractionpercentage for the two-stage ventilation desiccant coolingsystemwas 69Themaximumandminimum solar fractionswere for the two-stage recirculation and one-stage ventilationsystems

4 Conclusions

This paper has presented a comparison study among fourconfigurations of a solar desiccant cooling system in the hotand humid weather in Malaysia The four systems namelythe one-stage ventilation one-stage recirculation two-stageventilation and two-stage recirculation were simulated dur-ing an 8760 hr operation by TRNSYS The COP of the two-stage ventilation mode was higher than those of the otherconfigurations The two-stage ventilation produced supplyair at 176∘C and 00096 kgkg to remove the sensible load ofthe room whereas the supply air temperatures of the otherconfigurationswere higherTherefore based on the conditionof ambient air (30∘C and 00200 kgkg) the two-stage solardesiccant cooling system under the ventilation mode is moresuitable than the other configurations

Nomenclature

119860 Collector area [m2]AC Air conditioning system [-]ASHRAE American Society of Heating Refrigerating

and Air Conditioning Engineers [-]COP Thermal coefficient of performance [-]DEC Direct evaporative cooler [-]DW Desiccant wheel [-]F1 F2 Effectiveness of desiccant wheel119866 Solar radiation [Wm2]ℎ Enthalpy [kjkg]hr Time [hour]HR Humidity ratio [kgkg]HRW Heat recovery wheel [-]HVAC Heating Ventilation and Air-Conditioning

[-]HX Heat exchanger [-]119876cool Cooling capacity of the system [kW]119876119879 The cooling system driving energy119876119880 Solar thermal power [kW]

reg RegenerationRA Return air [-]SA Supply air [-]


SF Solar fraction []119879 Air temperature [∘C]TSRDC Two-stage rotary desiccant cooling [-]119898119904 Mass flow rate of the supply air [kghr]119898119903 Mass flow rate of regeneration air [kghr]120578 Solar collector efficiency [-]


The authors of the paper do not have any financial relationwith the TRNSYS software company They confirm that thepaper is just a research workwhich has no financial benefitfrom TRNSYS Company or other companies

Acknowledgment

The authors would like to thank the Solar Energy ResearchInstitute (SERI) Universiti KebangsaanMalaysia for provid-ing the laboratory facilities and technical support

References

[1] D La Y J Dai Y Li R Z Wang and T S Ge ldquoTechnicaldevelopment of rotary desiccant dehumidification and air con-ditioning a reviewrdquo Renewable and Sustainable Energy Reviewsvol 14 no 1 pp 130ndash147 2010

[2] E Hurdogan O Buyukalaca T Yılmaz and A HepbaslıldquoExperimental investigation of a novel desiccant cooling sys-temrdquo Energy and Buildings vol 42 no 11 pp 2049ndash2060 2010

[3] A Y T Al-Zubaydi ldquoSolar air conditioning and refrigerationwith absorption chillers technology in australiamdashan overviewon researches and applicationsrdquo Journal of Advanced Science andEngineering Research vol 1 no 1 pp 23ndash41 2011

[4] K Daou R Z Wang and Z Z Xia ldquoDesiccant coolingair conditioning a reviewrdquo Renewable and Sustainable EnergyReviews vol 10 no 2 pp 55ndash77 2006

[5] K Haddad B Ouazia and H Barhoun ldquoSimulation of adesiccant-evaporative cooling system for residential buildingsrdquoin Proceedings of the 3rd Canadian Solar Building Conferencepp 1ndash8 Fredericton Canada August 2008

[6] K F Fong V I Hanby and T T Chow ldquoHVAC systemoptimization for energymanagement by evolutionary program-mingrdquo Energy and Buildings vol 38 no 3 pp 220ndash231 2006

[7] M M Salehi Dezfouli Z Hashim M H Ruslan B BakhtyarK Sopian andA Zaharim ldquoExperimental investigation of solarhybrid desiccant cooling system in hot and humid weather ofmalaysiardquo in Proceedings of the 10th International Conference onEnvironment Ecosystems and Development (WSEAS rsquo12) 2012

[8] S Jain P L Dhar and S C Kaushik ldquoEvaluation of solid-desiccant-based evaporative cooling cycles for typical hot andhumid climatesrdquo International Journal of Refrigeration vol 18no 5 pp 287ndash296 1995

[9] P Bourdoukan EWurtz and P Joubert ldquoComparison betweenthe conventional and recirculation modes in desiccant coolingcycles and deriving critical efficiencies of componentsrdquo Energyvol 35 no 2 pp 1057ndash1067 2010

[10] K Mehmet M Ozdinc Carpınlıoglu andM Yıldırım ldquoEnergyand exergy analyses of an experimental open-cycle desiccantcooling systemrdquo Applied Thermal Engineering vol 24 no 5-6pp 919ndash932 2004

[11] G Panaras E Mathioulakis and V Belessiotis ldquoAchievableworking range for solid all-desiccant air-conditioning systemsunder specific space comfort requirementsrdquo Energy and Build-ings vol 39 no 9 pp 1055ndash1060 2007

[12] L A Sphaier and C E L Nobrega ldquoParametric analysis ofcomponents effectiveness on desiccant cooling system perfor-mancerdquo Energy vol 38 no 1 pp 157ndash166 2012

[13] Z Huan and N Jianlei ldquoTwo-stage desiccant cooling systemusing low-temperature heatrdquo Building Services EngineeringResearch and Technology vol 20 no 2 pp 51ndash55 1999

[14] Z Q Xiong Y J Dai and R Z Wang ldquoDevelopment of a noveltwo-stage liquid desiccant dehumidification system assisted byCaCl2 solution using exergy analysis methodrdquo Applied Energyvol 87 no 5 pp 1495ndash1504 2010

[15] S Techajunta S Chirarattananon and R H B Exell ldquoExperi-ments in a solar simulator on solid desiccant regeneration andair dehumidification for air conditioning in a tropical humidclimaterdquo Renewable Energy vol 17 no 4 pp 549ndash568 1999

[16] K-H Yang and S-H Wang ldquoThe analysis of a solar-assisteddesiccant evaporative cooling system and its applicationrdquo Jour-nal of the Chinese Institute of Engineers vol 11 no 3 pp 227ndash234 1988

[17] GMeckler ldquoTwo-stage desiccant dehumidification in commer-cial building HVAC systemsrdquo Ashrae Transactions vol 95 no2 pp 1116ndash1123 1989

[18] H-M Henning ldquoSolar assisted air conditioning of buildingsmdashan overviewrdquo Applied Thermal Engineering vol 27 no 10 pp1734ndash1749 2007

[19] D La Y Dai Y Li T Ge and RWang ldquoCase study and theoret-ical analysis of a solar driven two-stage rotary desiccant coolingsystem assisted by vapor compression air-conditioningrdquo SolarEnergy vol 85 no 11 pp 2997ndash3009 2011

[20] T S Ge Y Li R Z Wang and Y J Dai ldquoExperimental studyon a two-stage rotary desiccant cooling systemrdquo InternationalJournal of Refrigeration vol 32 no 3 pp 498ndash508 2009

[21] D La Y Li Y J Dai T S Ge and R Z Wang ldquoDevelopmentof a novel rotary desiccant cooling cycle with isothermaldehumidification and regenerative evaporative cooling usingthermodynamic analysis methodrdquo Energy vol 44 no 1 pp778ndash791 2012

[22] H Li Y J Dai Y Li D La and R Z Wang ldquoCase study ofa two-stage rotary desiccant coolingheating system driven byevacuated glass tube solar air collectorsrdquo Energy and Buildingsvol 47 pp 107ndash112 2012

[23] H Li Y J Dai Y Li D La andR ZWang ldquoExperimental inves-tigation on a one-rotor two-stage desiccant coolingheating sys-tem driven by solar air collectorsrdquoAppliedThermal Engineeringvol 31 no 17-18 pp 3677ndash3683 2011

[24] ASHRAE HandbookmdashHVAC Systems and Equipment Ameri-can Society of Heating Refrigerating and Air ConditioningEngineers Atlanta Ga USA 1996

[25] A M Baniyounes G Liu M G Rasul and M M K KhanldquoAnalysis of solar desiccant cooling system for an institutionalbuilding in subtropical Queensland Australiardquo Renewable andSustainable Energy Reviews vol 16 no 8 pp 6423ndash6431 2012

[26] Solar-Assisted Air-Conditioning in Buildings A Handbook ForPlanners Springer New York NY USA 2004


Research ArticleSolution-Processed Bulk Heterojunction Solar Cells withSilyl End-Capped Sexithiophene

Jung Hei Choi1 Mohamed E El-Khouly2 Taehee Kim1 Youn-Su Kim1

Ung Chan Yoon3 Shunichi Fukuzumi4 and Kyungkon Kim15

1 Photo-Electronic Hybrids Research Center Korea Institute of Science and Technology Hwarangno 14-gil 5 Seongbuk-guSeoul 136-791 Republic of Korea

2Department of Chemistry Faculty of Science Kafrelsheikh University Kafr ElSheikh 33516 Egypt3 Department of Chemistry Pusan National University Jangjeon-dong Kumjeong-gu Busan 609-735 Republic of Korea4Department of Material and Life Science Graduate School of Engineering Osaka University ALCAJapan Science and Technology Agency (JST) Suita Osaka 565-0871 Japan

5Department of Chemistry and Nanoscience Ewha Womans University 52 Ewhayeodae-gil Seodaemun-guSeoul 120-750 Republic of Korea

Correspondence should be addressed to Shunichi Fukuzumi fukuzumichemengosaka-uacjp

Received 15 July 2013 Accepted 25 November 2013


Copyright copy 2013 Jung Hei Choi et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

We fabricated solution-processed organic photovoltaic cells (OPVs) using substituted two sexithiophenes aw-bis(dimethyl-n-octylsilyl)sexithiophene (DSi-6T) and aw-dihexylsexithiophene (DH-6T) as electron donors and [66]-phenyl-C

61-butyric acid

methyl ester (PCBM) as an electron acceptor Solution-processed OPVs using DH-6T and DSi-6T showed good photovoltaicproperties in spite of their poor solubility The best performance was observed on DSi-6T PCBM 1 5 (ww) blend cell with anopen circuit voltage (119881oc) of 063V short circuit current density (119869sc) of 134mAcm2 fill factor (FF) of 55 and power conversionefficiency of 044 under AM 15G illumination Although DH-6T has higher hole mobility than DSi-6T the DSi-6T PCBMblend cell showed higher hole mobility than DH-6T PCBM cell Therefore DSi-6T cell showed higher device performance thanDH-6T cell due to its silyl substitutions which lead to the increase of the solubility The incorporation of solution-processed TiO

2

interfacial layer in the DSi-6T PCBM devices significantly enhances FF due to the reduced charge recombination near activelayerAl interface

1 Introduction

Organic solar cells have received strong attention due tothe possibility to realize flexible and low-cost large photo-voltaic cells [1ndash5] Polymer based organic photovoltaic cells(OPVs) demonstrated power conversion efficiencies (PCEs)in excess of 7 by the results of synthesis of low bandgappolymers [6 7] However the polymers still have highbatch-to-batch variations and difficulty in purification andthere have been many efforts to discover solution-processedsmall molecules Recently oligothiophenes have receivedmuch attention as donor materials in OPVs because of theirhigh hole mobility easy multigram synthesis high purityand simple chemical modification [8ndash10] Some studies on

oligothiophene based solar cells have already been reportedusing bilayer heterojunction solar cells [11ndash15] By controllingthe film morphology using coevaporation of excess fullerene(C60) bulk heterojunction organic photovoltaic cells con-

sisting of sexithiophene (6T) as a donor and C60

as anacceptor showed good photovoltaic properties [16] Howeverthe small molecules based on sexithiophene have not beeninvestigated solution-processed OPVs The low performanceof sexithiophene-based solar cells was mainly attributed tothe difficulties encountered in fabricating solution-processedbulk heterojunction cells

We choose 120572120596-dihexylsexithiophene (DH-6T) as the ol-igothiophene material because it is well known to exhibit


high field-effect mobility as high as 10 cm2Vs and a repre-sentative oligothiophene material [17] Recently we report-ed 120572120596-bis(dimethyl-n-octylsilyl)sexithiophene (DSi-6T) forvacuum-deposited or solution processable organic thin filmtransistor application [18] The silyl end-capped sexithio-phene (DSi-6T 22 gL in CHCl

3) was more soluble than

hexyl substituted sexithiophene (DH-6T 1 gL in CHCl3)

[19] Thus the silyl end-capped sexithiophene may be a goodcandidate for solution processable solar cell application

In regard to our continuing interest for sexithiophene-based organic solar cells we report herein analysis of the per-formance of solution-processed organic solar cells based on120572120596-functionalized linear sexithiophene PCBM systems Wefabricated the solution-processed OPVs using two linear sex-ithiophenes 120572120596-dihexylsexithiophene (DH-6T) and 120572120596-bis(dimethyl-n-octylsilyl)sexithiophene (DSi-6T) as donorsand [6 6]-phenyl-C

61-butyric acid methyl ester (PCBM) as

an acceptor (Figure 1) Solution-processed bulk heterojunc-tion (BHJ) solar cells using DH-6T and DSi-6T with PCBMshowed good photovoltaic properties in spite of their poorsolubility Although DH-6T has higher hole mobility thanDSi-6T the DSi-6T PCBM blend cell showed higher holemobility than DH-6T PCBM cell Therefore DSi-6T cellshowed higher device performance than DH-6T cell due totheir silyl substitutions which lead to the increase of thesolubility The higher photovoltaic performance of DSi-6Tcell can be explained by the filmmorphology dependence onthe solubility of donor In addition utilizing the nanosecondtransient absorption spectroscopy in polar media in thevisible and near-IR region proved electron-transfer reactionsin DH-6T PCBM and DSi-6T PCBMmixtures

2 Experimental Section

21 Instruments The absorption spectra were measuredusing a Perkin Elmer Lambda 35 UV-vis spectrometer inspin-coated films at room temperature The morphologyof the DSi-6T PCBM and DH6T PCBM was determinedusing atomic force microscopy (AFM) (Park systems) oper-ating in a noncontact mode under ambient conditionsSteady-state fluorescence measurements were carried out ona Shimadzu spectrofluorophotometer (RF-5300PC) Phos-phorescence spectra were obtained by a SPEX Fluorolog 120591

3

spectrophotometer Emission spectra in the visible regionwere detected by using a Hamamatsu Photonics R5509-72photomultiplier A deaerated 2-MeTHF solution containingDH-6T and DSi-6T at 77K was excited at indicated wave-lengths Cyclic voltammograms (CV) and differential pulsevoltammograms (DPV) techniqueswere carried on aBASCV50W Voltammetric Analyzer A platinum disk electrode wasused as working electrode while a platinum wire served asa counter electrode SCE electrode was used as a referenceelectrode All measurements were carried out in deaeratedbenzonitrile containing tetra-n-butylammonium hexafluo-rophosphate (TBAPF

6 010M) as a supporting electrolyte

The scan rate = 50mVsDSi-6T andDH-6Twere excited by a PantherOPOpum-

ped by NdYAG laser (Continuum SLII-10 4ndash6 ns fwhm) at

120582 = 440 nm with the powers of 15 and 30mJ per pulseThe transient absorption measurements were performedusing a continuous xenon lamp (150W) and an InGaAs-PIN photodiode (Hamamatsu 2949) as a probe light and adetector respectively The output from the photodiodes anda photomultiplier tube was recorded with a digitizing oscil-loscope (Tektronix TDS3032 300MHz) All measurementswere conducted at 298KThe transient spectra were recordedusing fresh solutions in each laser excitation

22 Fabrication of BHJ Cells BHJ solar cells were fabri-cated on patterned ITO (indium tin oxide) coated glasssubstrates as the anode ITO on glass substrates was sequen-tially cleaned with isopropyl alcohol acetone and isopropylalcohol in ultrasonic baths Afterwards it was dried inoven at 80∘C for 10min and then treated with UV-ozonefor 20min The surface of the ITO substrate was modi-fied by spin-coating conducting poly(34-ethylenedioxythi-ophene) poly(styrenesulfonate) (PEDOT PSS H C StarckBaytron P VP Al 4083) with a thickness of around 40 nmfollowed by baking at 120∘C for 10min in ovenThemolecularstructures of the donor (DSi-6T and DH-6T) and acceptormaterials (PCBM) and their energy band diagram are shownin Figure 1 DSi-6T was synthesized according to literatureprocedures [18]DH-6Twas purchased from Aldrich Chem-ical Co and was used without further purification DSi-6T PCBM (1 4 1 5 1 6 ww) and DH-6T PCBM (1 41 6 ww) were blended together and dissolved in chloroformat a total concentration of 20mgmLminus1 And then DSi-6T PCBM blended solution was no filtration but DH-6T PCBM solution was filtered out through a 045 120583m filterThe active layer was spin-cast at 4000 rpm from the solutionof DSi-6T or DH-6T and PCBM (Nano-C) in chloroformThe thickness of the active layers was around 80 nm asmeasured with an Alpha-step IQ A solution containing05 wt TiO

2nanoparticles in ethanol was spin-coat at a

rate of 4000 rpm on top of the active layer The synthesisof TiO

2nanoparticles is described elsewhere [20 21] The

cathode consists of 100 nm of aluminum and was thermallyevaporated on top of the film through the use of a shadowmask to define an active area of 012 cm2 under a base pressureof 1times 10minus6 Torr (1 Torr = 13332 Pa)Thedevices were annealedat 60∘C for 10min in the thermal evaporator The currentdensity-voltage curves were obtained using a Keithley 2400source-measure unit The photocurrent was measured underillumination using a Newport class-A 100mWcmminus2 solarsimulator (AM 15G) and the light intensity was calibratedwith an NREL-calibrated Si solar cell with KG-1 filter forapproximating one sun light intensity External quantumefficiency (EQE) was measured in range from 300 to 800 nmusing a specially designed EQE system (PV MeasurementsInc) A 75W xenon lamp was used as a light source forthe generating monochromatic beam A calibration wasperformed using a silicon photodiode which was calibratedusing the NIST-calibrated photodiode G425 as a standardand IPCE values were collected under bias light at a lowchopping speed of 10Hz


SSiS

SS

SS Si

SS

SS

SS

DSi-6T

DH-6T

DH-6T

O

O

PCBM

C8H17C8H17

C6H13 C6H13

DSi-6T

PEDOTPSS

48 eV51 eV

51 eV

43 eV44 eV

27 eV

43 eV

62 eV

81 eV

TiO2

ITO

AI

PCBM

or

Figure 1 Chemical structures and energy band diagram of DSi-6TDH-6T and PCBM

300 350 400 450 500 550 60000

02

04

06

08

10

Abso

rban

ce

Wavelength (nm)

PCBMDH-6TDSi-6T

DH-6T PCBM (1 6)DSi-6T PCBM (1 5)

Figure 2 UV-vis absorption spectra of the DSi-6T PCBM (1 5ww) andDH-6T PCBM (1 6 ww) blend films


31 Electron-Transfer Reaction of DSi-6TPCBM and DH-6TPCBM The UV-vis absorption spectra of spin-coatedfilms of DSi-6T and DH-6T blended with PCBM are shownin Figure 2 Both DH-6T and DSi-6T exhibited similar

absorption spectra in a dilute chloroform solution withabsorption maxima observed at 440 nm and 442 nm respec-tively The absorption spectra of sexithiophene films exhib-ited a noteworthy blue shiftwith respect to that in solution forboth compounds but this shift is more significant forDH-6Tthan forDSi-6T While theDH-6T film showed the stronglyblue-shifted absorption spectrum with maxima at 362 nmthe DSi-6T film showed an absorption peak at 415 nmSuch feature is attributed to molecular excitons formed as aconsequence of the intermolecular interactions in the solidstate [22] and is found to be blue-shifted by about 06 eVwith respect to the characteristic absorption peaks of theisolated molecules The PCBM film has an absorption bandof about 340 nm and significantly greater absorption in thevisible region (350ndash750 nm) compared to the PCBMsolutionconsistent with the literature [23] The UV-vis absorptionspectra of DSi-6T PCBM (1 5 ww) and DH-6T PCBM(1 6 ww) blend films showed similar shape due to their samebackbone of donor and strong absorption of PCBM

Fluorescence spectra of the singlet-excited state of DH-6T and DSi-6T in benzonitrile exhibited emission bandsat 514 and 520 nm respectively from which the energy ofthe singlet state of DH-6T and DSi-6T were estimatedas 241 and 239 eV respectively (See Supporting Informa-tion Figure S1 of Supplementary Material available onlineat httpdxdoiorg1011552013843615) Fluorescence quan-tum yields of DH-6T and DSi-6T was determined as 013and 036 respectively Fluorescence lifetimes of the singletstates of DH-6T and DSi-6T were determined as 10 and


06

05

04

03

02

01

00

1600140012001000800600400Wavelength (nm)

06050403020100

20151050Time (120583s)

800nm

700nm

2120583s8120583s

ΔAb

sorb

ance

ΔAb

sorb

ance

(a)

Time (120583s)

12

10

08

06

04

02

00

0 10 20 30 40

300

250

200

150

100

0 20 40 60 80 100

000

001

005

010

ΔAb

sorb

ance

[PCPM] (120583M)

[PCPM] (mM)

k1

sttimes103

(b)

Figure 3 (a) Nanosecond transient spectra of DSi-6T (005mM) in the presence of PCBM (005mM) in deaerated benzonitrile 120582ex =440 nm Inset Decay time profile of 3DSi-6Tlowast (700 nm) and rise and decay profiles of DSi-6T∙+(800 nm) (b) Dependence of rate constantof the formation of DSi-6T∙+ at 800 nm on concentration of PCBM in deaerated benzonitrile Inset first-order plot

18 ns respectively In the case of the DH-6TPCBM andDSi-6TPCBM mixtures it was found that the fluorescenceintensities and the lifetimes are quite close to those of theDH-6T and DSi-6T respectively This observation suggeststhe intermolecular electron transfer between theDH-6T andDSi-6T with PCBM but not the intramolecular electrontransfer

The redox potentials of the examined DH-6T DSi-6Tand PCBM have been examined by using cyclic voltammetry(CV) anddifferential pulse voltammetry (DPV) techniques toevaluate the driving forces for the electron transfer (minusΔ119866et

119879)The first reduction potential (119864red) of the PCBM was locatedat ndash856mV versus AgAgNO

3 while the oxidation potentials

(119864ox) of DH-6T andDSi-6T were located at 732 and 600mVversus AgAgNO

3(see Supporting Information Figure S2ndash

S5)The redox potential measurements of theDH-6TPCBMand DSi-6TPCBM mixtures did not show significant inter-action suggesting no interaction in the ground state Theredox potentials of the examined DH-6T DSi-6T andPCBM suggest their potentials as promising materials inthe photovoltaic cells The feasibility of the electron-transferprocess via the triplet-excited states is controlled by the freeenergy change (Δ119866et

119879) which can be expressed by the Rehm-Weller relation [24] (1)

Δ119866et119879= 119864ox minus 119864red minus 119864119879 + 119864119888 (1)

where 119864ox is the first oxidation potential of the DH-6Tand DSi-6T 119864red is the first reduction potential of PCBM119864T is the triplet energy of DH-6T (178 eV) and DSi-6T

(179 eV) (we found from phosphorescence measurements indeaerated benzonitrile that energy level of the triplet DH-6T and DSi-6T at 178 and 179 eV resp) and 119864c is theCoulomb energy term (approximately 006 eV in the polarbenzonitrile) [25] The free energy change (Δ119866et

119879) valuesvia the triplet DH-6T and DSi-6T were estimated as minus005and minus017 eV respectively The negative Δ119866et

119879 values suggestthat the quenching process should be close to the diffusion-controlled limit (119896diff) [26]

By photoexcitation of DSi-6T in deaerated benzoni-trile using 440 nm laser photolysis the transient absorp-tion spectrum immediately after the laser pulse exhibitedonly an absorption band at 700 nm which assigned to thetriplet-excited state of DH-6T (3DH-6Tlowast) (See SupportingInformation Figure S6) By fitting the decay profile of3DSi-6Tlowast the decay rate constant was found to be 660 times104 sminus1 from which the lifetime of 3DSi-6Tlowastwas estimatedas 152 120583s By photoexcitation of DSi-6T in the presenceof PCBM [001ndash010mM] in Ar-saturated benzonitrile using440 nm laser photolysis the transient spectra exhibit thecharacteristic band of 3DSi-6Tlowast at 700 nm With its decaythe concomitant rises of the DSi-6T radical cation (DSi-6T∙+) at 800 and 1500 nm and the PCBM radical anion(PCBM∙minus) at 1000 nm were observed (Figure 3(a) and FigureS7) These observations show clear evidence of occurrenceof intermolecular electron transfer from the triplet-excitedstate of DSi-6T to PCBM In oxygen-saturated solutionsan intermolecular energy transfer from 3DSi-6Tlowast to oxygenemerges suppressing the electron-transfer process Similar


00 02 04 06 08

00

Voltage (V)

Curr

ent d

ensit

y (m

Ac

m2)

minus08

minus04

minus12

minus16

DSi-6T PCBM (1 4)DSi-6T PCBM (1 4) with TiO2



Figure 4 J-V characteristics of ITOPEDOT PSSDSi-6T PCBMAl (active layer DSi-6T PCBM 1 4 ww (I) 1 5 ww () 1 6 ww(◻)) and ITOPEDOT PSSDSi-6T PCBMTiO

2Al (active layer

DSi-6T PCBM 1 4 ww (∙) 1 5 ww (998771) 1 6 ww (◼)) underillumination of AM 15 100mWcmminus2

electron-transfer features were recorded in the case of DH-6TPCBMmixture in deaerated benzonitrile (See supportinginformation Figure S8)

A more detailed picture of the kinetic is shown inFigure 3(b) where the rate constant of the electron-transferprocess (119896et) was evaluated by monitoring the formation ofthe DSi-6T∙+ as function of the concentrations of PCBMThe formation of DSi-6T∙+ was fitted with clean first-orderkinetics each rate constant is referred to (119896

1st)The linear con-centration dependence of the observed 119896

1st values gives the119896et which is calculated as 224 times 109Mminus1 sminus1 which is nearthe diffusion-controlled limit (119896diff = 56 times 109Mminus1 sminus1) inbenzonitrile [27] Similarly the 119896et value of

3DH-6TlowastPCBMmixture was found to be 265 times 109Mminus1 sminus1 which is slightlylarger than that of 3DSi-6TlowastPCBMmixture (See supportinginformation Figure S9)

32 Photovoltaic Properties of DSi-6TPCBM and DH-6TPCBM For investigating the photovoltaic properties of sex-ithiophenes OPVs were fabricated with the configuration ofITOPEDOT PSSDonor (DSi-6T or DH-6T) PCBMAlDSi-6T PCBM and DH-6T PCBM were blended togetherand dissolved in chloroform at a total concentration of20mgmL After Al deposition the devices were thermallyannealed at 60∘C for 10min Figures 4 and 5 show thecurrent density-voltage (J-V) curves of the OPV devices withDSi-6T PCBM and DH-6T PCBM using different blendedratios under AM 15 illumination 100mWcmminus2 Table 1 sum-marizes the photovoltaic performance of the functionalized

00 01 02 03 04 05 06 07 08Voltage (V)

00

Curr

ent d

ensit

y (m

Ac

m2)

minus08

minus06

minus04

minus02

minus10

DH-6T PCBM (1 4)DH-6T PCBM (1 4) with TiO2


Figure 5 J-V characteristics of ITOPEDOT PSSDH-6T PCBMAl (active layer DH-6T PCBM 1 4 ww (X) 1 6 ww (998771))and ITOPEDOT PSSDH-6T PCBMTiO

2Al (active layer DH-

6T PCBM 1 4 ww (∙) 1 6 ww (◼)) under illumination of AM15 100mWcmminus2

sexithiophene based solar cells The sexithiophene DSi-6Twas mixed with PCBM to form bulk heterojunction activelayers for use as a donor with PCBM as an acceptor Theweight ratios of the donor to acceptor were 1 4 1 5 and 1 6For comparison the OPV devices based on DH-6T PCBMwere also fabricated with commercialDH-6T as a donor andthe weight ratio of 1 4 and 1 6 The short circuit currentdensity (119869sc) values of the DSi-6T PCBM BHJ solar cellsincrease as the weight ratio of donor to acceptor increasesHowever the weight ratio of donor to accepter could not belarger than 1 4 due to low solubility of sexithiophenes Thephotovoltaic performance was observed on DSi-6T PCBM1 5 (ww) blend cell with an open circuit voltage (119881oc)of 060V a 119869sc of 129mAcm2 a fill factor (FF) of 42and power conversion efficiency of 032 Under the sameconditions DH-6T PCBM 1 6 (ww) blend cell showeda 119881oc of 063V a 119869sc of 060mAcm2 a FF of 44 andPCE of 017 The 119869sc values of the DH-6T PCBM cellswere constant with 060mAcm2 regardless of the weightratio of DH-6T to acceptor This result could be ascribedto the saturated solution of DH-6T PCBM (1 6 ww) Infact a factor strongly affecting the 119869sc of DSi-6T PCBMcells is the filtration of the active solution The 119869sc of solarcell with the filtration of DSi-6T PCBM solutions is 20lower than that of solar cell without filtration The mostDSi-6T donors remained in the filter due to their strong 120587-120587interaction However the DH-6T cells without filtration ofactive solutions showed the poor photovoltaic performance

For the development of OPVs with increased PCE andlifetime OPVs with the TiO

2interfacial layer between the

active layer and the Al electrode were fabricated togetherwith typical OPVs without TiO

2 According to the literature


Table 1 Photovoltaic characteristics of DSi-6T PCBM andDH-6T PCBM blended cells

Device name 119881oc (V) 119869sc (mAcm2) FF PCE ()DSi-6T PCBM

1 4 061 126 039 0301 5 060 129 042 0321 6 057 088 037 018

DSi-6T PCBM (with TiO2)1 4 064 134 050 0421 5 060 134 055 0441 6 062 107 049 032

DH-6T PCBM1 4 052 062 033 0111 6 063 060 044 017

DH-6T PCBM (with TiO2)1 4 051 061 042 0131 6 065 057 056 021

aOpen circuit voltage (119881oc) short circuit current density (119869sc) fill factor (FF) and PCE at DSi-6T and DH-6T as electron donors PCBM at weight ratio Thedevice performance was consistent and reproducible The active area is 012 cm2

the TiO2interfacial layer improved the PCEs and reduced

the sensitivity of such devices to oxygen and water vapour[21 28] As shown in Figure 4 the FF of DSi-6T PCBMcells with TiO

2layer is enhanced up to 055 which is 30

higher than that of the solar cell The best performance wasobserved onDSi-6T PCBM 1 5 (ww) blend cell with119881oc of063V 119869sc of 134mAcm2 FF of 55 and power conversionefficiency of 044 Also the FF of DH-6T cells is exactly 27higher than that of solar cells TheDH-6T PCBM 1 6 (ww)blend cell with TiO

2layer exhibited a 119881oc of 065 V 119869sc of

057mAcm2 FF of 56 and PCE of 021 (Figure 5) Theincorporation of solution processed TiO

2interfacial layer in

the sexithiophene PCBMBHJ devices significantly enhancesFF mainly due to the reduced charge recombination nearactive layerAl interface

Figure 6 shows the external quantum efficiency (EQE)plot for the DSi-6T PCBM (1 5 ww) and DH-6T PCBM(1 6 ww) cells The sexithiophene-based cells absorb thesolar light in narrow visible range and show low EQEs dueto a small amount of donor by limited solubility For theDH-6T PCBM solar cell the maximum EQE is 113 at 340 nmthat is caused by strong absorption of PCBMWhen theDSi-6T was used the DSi-6T PCBM solar cell shows a higherEQE in the all wavelength with a maximum of 161 at410 nmThe absorption of theDSi-6T donorwas increased inthe EQE spectrum to be different from the UV-vis spectrumIt is expected that horizontal intermolecular packing due tothe bulky silyl side chains of DSi-6Tmay increase the chargetransfer at the interfaces so the EQE increases

To investigate the charge transporting property of theOPVdevices wemeasured holemobility of the donor PCBMblend layers with the space-charge limited-current (SCLC)model Hole-only devices were fabricated with a device con-figuration of ITOPEDOT PSSDonor PCBMMoO

3Al

because high work function of molybdenum oxide (MoO3)

blocks the injection of electrons from the Al cathode Theblend ratios of the Donor PCBM layers were 1 5 (ww)

300 400 500 600 700 8000

2

4

6

8

10

12

14

16

18

20EQ

E (

)

Wavelength (nm)

DSi-6T PCBM (1 5) with TiO2

DH-6T PCBM (1 6) with TiO2

Figure 6 EQE curves of DSi-6T PCBM (1 5) andDH-6T PCBM(1 6) blend cells

and 1 6 (ww) for DSi-6T PCBM and DH-6T PCBMrespectively

The hole transport through the active layers is limited bythe space charge and the SCLC is described by

119869 =

9

8

1205761199031205760120583ℎ

1198812

1198713 (2)

where 119869 is the current density 1205760is the permittivity of vacuum

120576119903is the dielectric constant of the material (assumed to be 3)120583ℎis the mobility 119881 is the applied voltage (119881applied) corrected

from built-in voltage (119881bi) arising from difference in the workfunction of the contacts and voltage drop (119881

119903) due to the

series resistance of the electrodes and 119871 is the thickness ofthe blend layer [29] Equation (1) is valid when the mobility


1000 2000 3000 4000 5000 6000

minus41

minus42

minus43

minus44

minus45

minus46

Ln(JL3V2)

E12 (V05

m05)

DH-6T PCBMDSi-6T PCBM

(a)

00 05 10 15 200

10

20

30

40

50

ExpCal (1)Cal (2)

ExpCal (1)Cal (2)

J05

(A05m

)

Vapp minus Vbi minus Vr (V)


(b)

Figure 7 (a) ln (11986911987131198812) versus11986412 curves according to the SCLCequation (2) (b) 11986912 versus119881 curves showing the electric-field-dependenceof hole mobilities at highly biased region Dotted and solid lines are calculated curves with (1) and (2) respectively

is field independent It has been found that the behaviour ofelectric field dependence of the mobility can be described byan empirical rule 120583

ℎ= 120583ℎ0exp(12057411986412) where 120583

ℎ0is the hole

mobility at zero field 120574 is field dependence prefactor and 119864 isthe electric fieldTherefore the electric field dependent SCLCmodel can be expressed by

119869 =

9

8

1205761199031205760120583ℎ0exp (120574radic119864) 119881

2

1198713 (3)

The zero-field hole mobility 120583ℎ0

and the prefatory 120574of DSi-6T PCBM (1 5 ww) and DH-6T PCBM (1 6 ww)blend films were obtained from (2) as shown in Figure 7(a)(120583ℎ0

= 77 times 10minus6 cm2Vs 120574 = 33 times 10minus4m05 Vminus05 forDSi-6T PCBM and 120583

ℎ0= 16 times 10minus6 cm2Vs 120574 = 45

times 10minus4m05 Vminus05 for DH-6T PCBM) The low 120574 of DSi-6T PCBM indicates low energetic disorder related to theinteraction of each hopping charge in randomly orienteddipoles [30 31] Figure 7(b) shows the SCLC fitting of DSi-6T PCBM (1 5 ww) and DH-6T PCBM (1 6 ww) blendfilms according to (1) and (2) It is observed that the SCLCcalculated from the field independent model deviates fromthe experimental data at the high voltage region In actualoperating devices the active materials are in the effectivebias of sim1 V (corresponding to the energy level difference ofHOMO and LUMO of an electron donor and acceptor) atthe short circuit conditions Taking this into considerationwe estimated the field dependent hole mobility of the activeblend layers at 1 V120583

ℎ= 27times 10minus5 cm2Vs forDSi-6T PCBM

and 120583ℎ= 88 times 10minus6 cm2Vs for DH-6T PCBM The higher

hole mobility of the DSi-6T PCBM blend film than that ofthe DH-6T PCBM film is a key property of the enhancedphotovoltaic performance It seems that the higher hole

mobility of DSi-6T PCBM blend layer is inconsistent withthe pure donorrsquos field-effect mobility However it shouldbe noted that the mobility in organic thin-film transistorsstrongly depends on the molecular orientation due to thehighly anisotropic charge carrier mobility In the OPV deviceconfiguration the DSi-6T PCBM BHJ layer yielded thebetter vertical charge transport than theDH-6T PCBM dueto the improved solubility of DSi-6T

To study the thermal annealing effect of this system themorphologies of sexithiophene PCBM films before and afterannealing at 60∘C were studied by atomic force microscopy(AFM) (Figure 8) The morphological studies indicate themore amorphous nature of DSi-6T PCBM compared toDH-6T PCBM films The root-mean-square (rms) rough-ness is 47 38 and 23 nm for DSi-6T PCBM 1 4 1 5 and1 6 (ww) respectively and 71 nm for DH-6T PCBM 1 6(ww) In DSi-6T PCBM films the weight ratio donor toacceptor increased as the roughness of films decreased Afterthermal annealing the roughness was 37 29 and 14 nmfor DSi-6T PCBM 1 4 1 5 1 6 (ww) respectively and76 nm for DH-6T PCBM 1 6 (ww) For the solar cellbased on theDSi-6T PCBM cell the rms roughness of filmsafter thermal annealing at 60∘C was significantly reduced ascompared to that of films In the case of DH-6T PCBMcell the morphological change after annealing was slightlyobserved Unfortunately all films were quite rough com-pared to the polymer solar cells The inferior morphologiesof sexithiophene based BHJ solar cells may be related tothe solubility of donors which is expected to influencethe photovoltaic performance For a comparison the DSi-6T PCBM cell was thermally annealed at 135∘C which wasthe temperature employed for the P3HT PCBM cell Whena film was annealed at a high temperature the decrease in


200

(nm

) 100

0

(a)

(nm

)

80

40

0

minus40

minus80

(b)

(nm

)

75

50

25

0

minus25

minus50

(c)

(nm

)

20

10

0

minus10

minus20

(d)

(nm

)

120

80

40

0

minus40

(e)

(nm

)80

40

0

minus40

(f)

(nm

)

60

40

20

0

minus20

(g)

(nm

)

20

10

0

minus10

minus20

(h)

Figure 8 AFM images of the blended films under various ratios (a) DSi-6T PCBM 1 4ww nonannealed (e) annealed and (b) DSi-6T PCBM 1 5 ww non-annealed (f) annealed and (c)DSi-6T PCBM 1 6ww non-annealed (g) annealed and (d)DH-6T PCBM 1 6wwnon-annealed (h) annealed The image scan sizes is 5 120583m times 5 120583m

PCE (0034) becamemore significantThe future workmayconcentrate on optimization of the annealing temperature oftheDSi-6T PCBM devices to allow better efficiency

4 Conclusion

In conclusion we have demonstrated solution processedorganic photovoltaic cells that incorporate sexithiophenederivatives having silyl side chain or hexyl chain in theend of thiophene ring as donor materials and PCBM as anacceptor Owing to the particular electronic properties ofDSi-6T DH-6T and PCBM such combinations seem to beperfectly suited for the study of electron-transfer processesin the polar media via the triplet states The electron-transferprocess from the triplet states of DSi-6T DH-6T to PCBMwas confirmed in this study by utilizing the nanosecond laserphotolysis technique in the visible and NIR regions DSi-6T showed higher photovoltaic performance than DH-6Tdespite its poor quality film morphology The optimal DSi-6T PCBM blend ratio was found to be 1 5 and its powerconversion efficiency was 044 under AM 15G illumina-tion Although the power conversion efficiency remains to beimproved the present study provides valuable insight into thefurther development of the solution-processed efficient OPVdevices


The authors declare that they have no conflict of interests

Acknowledgments

This researchwas supported byGrants-in-Aid (No 20108010)from the Ministry of Education Culture Sports Science andTechnology Japan and NRFMEST of Korea through WCU(R31-2008-000-10010-0) and GRL (2010-00353) Programs

References

[1] G Yu J Gao J C Hummelen F Wudl and A J Heeger ldquoPoly-mer photovoltaic cells enhanced efficiencies via a network ofinternal donor-acceptor heterojunctionsrdquo Science vol 270 no5243 pp 1789ndash1791 1995

[2] N S Sariciftci L Smilowitz A J Heeger and F Wudl ldquoPhot-oinduced electron transfer from a conducting polymer tobuckminsterfullerenerdquo Science vol 258 no 5087 pp 1474ndash14761992

[3] S Gunes H Neugebauer and N S Sariciftci ldquoConjugatedpolymer-based organic solar cellsrdquo Chemical Reviews vol 107no 4 pp 1324ndash1338 2007

[4] P W M Blom V D Mihailetchi L J A Koster and D EMarkov ldquoDevice physics of polymer fullerene bulk heterojunc-tion solar cellsrdquo Advanced Materials vol 19 no 12 pp 1551ndash1566 2007

[5] W Tang J Hai Y Dai et al ldquoRecent development of conjugatedoligomers for high-efficiency bulk-heterojunction solar cellsrdquoSolar Energy Materials and Solar Cells vol 94 no 12 pp 1963ndash1979 2010

[6] H-Y Chen J Hou S Zhang et al ldquoPolymer solar cells withenhanced open-circuit voltage and efficiencyrdquoNature Photonicsvol 3 no 11 pp 649ndash653 2009

[7] Y Liang Z Xu J Xia et al ldquoFor the bright future-bulk hetero-junction polymer solar cells with power conversion efficiency of74rdquo Advanced Materials vol 22 no 20 pp E135ndashE138 2010

[8] Electronic Materials The Oligomer Approach edited by KMullen andGWegnerWiley-VCHWeinheimGermany 1998

[9] Handbook of Oligo- and Polythiophenes edited by D FichouWiley-VCH Weinheim Germany 1999

[10] A Mishra C-Q Ma and P Bauerle ldquoFunctional oligothio-phenes molecular design for multidimensional nanoarchitec-tures and their applicationsrdquo Chemical Reviews vol 109 no 3pp 1141ndash1176 2009

[11] N Noma T Tsuzuki and Y Shirota ldquo120572-thiophene octamer asa new class of photoactive material for photoelectrical conver-sionrdquo Advanced Materials vol 7 no 7 pp 647ndash648 1995


[12] C Videlot A El Kassmi and D Fichou ldquoPhotovoltaic prop-erties of octithiophene-based Schottky and pn junction cellsinfluence of molecular orientationrdquo Solar Energy Materials andSolar Cells vol 63 no 1 pp 69ndash82 2000

[13] P LiuQ LiMHuangWPan andWDeng ldquoHigh open circuitvoltage organic photovoltaic cells based on oligothiophene de-rivativesrdquo Applied Physics Letters vol 89 no 21 pp 213501ndash213503 2006

[14] K Schulze C Uhrich R Schuppel et al ldquoEfficient vacuum-deposited organic solar cells based on a new low-bandgap oli-gothiophene and fullerene C

60rdquoAdvancedMaterials vol 18 no

21 pp 2872ndash2875 2006[15] J Ah Kong E Lim K K Lee S Lee and S Hyun Kim ldquoA

benzothiadiazole-based oligothiophene for vacuum-depositedorganic photovoltaic cellsrdquo Solar Energy Materials and SolarCells vol 94 no 12 pp 2057ndash2063 2010

[16] J Sakai T Taima and K Saito ldquoEfficient oligothiophe-nefullerene bulk heterojunction organic photovoltaic cellsrdquoOrganic Electronics vol 9 no 5 pp 582ndash590 2008

[17] M Halik H Klauk U Zschieschang et al ldquoRelationshipbetweenmolecular structure and electrical performance of olig-othiophene organic thin film transistorsrdquo Advanced Materialsvol 15 no 11 pp 917ndash922 2003

[18] J H Choi DW Cho H J Park et al ldquoSynthesis and character-ization of a series of bis(dimethyl-n-octylsilyl)oligothiophenesfor organic thin film transistor applicationsrdquo Synthetic Metalsvol 159 no 15-16 pp 1589ndash1596 2009

[19] F Garnier A Yassar R Hajlaoui et al ldquoMolecular engineeringof organic semiconductors design of self-assembly propertiesin conjugated thiophene oligomersrdquo Journal of the AmericanChemical Society vol 115 no 19 pp 8716ndash8721 1993

[20] E Scolan and C Sanchez ldquoSynthesis and characterization ofsurface-protected nanocrystalline titania particlesrdquo Chemistryof Materials vol 10 no 10 pp 3217ndash3223 1998

[21] W-S Chung H LeeW Lee et al ldquoSolution processed polymertandem cell utilizing organic layer coated nano-crystalline TiO

2

as interlayerrdquoOrganic Electronics vol 11 no 4 pp 521ndash528 2010[22] A Sassella R Tubino A Borghesi et al ldquoOptical properties

of oriented thin films of oligothiophenesrdquo Synthetic Metals vol101 no 1 pp 538ndash541 1999

[23] S Cook H Ohkita Y Kim J J Benson-Smith D D C Bradleyand J R Durrant ldquoA photophysical study of PCBM thin filmsrdquoChemical Physics Letters vol 445 no 4-6 pp 276ndash280 2007

[24] D Rehm and A Weller ldquoKinetik und mechanismus der elek-tronubertragung bei der fluoreszenzloschung in acetonitrilrdquoBerichte der Bunsengesellschaft fur physikalische Chemie vol 73pp 834ndash839 1969

[25] R J Sension A Z Szarka G R Smith and R M HochstrasserldquoUltrafast photoinduced electron transfer to C

60rdquo Chemical

Physics Letters vol 185 no 3-4 pp 179ndash183 1991[26] Handbook of Photochemistry edited by S I Murov Marcel

Dekker New York NY USA 1985[27] M E El-Khouly O Ito P M Smith and F DrsquoSouza ldquoInter-

molecular and supramolecular photoinduced electron trans-fer processes of fullerene-porphyrinphthalocyanine systemsrdquoJournal of Photochemistry and Photobiology C vol 5 no 1 pp79ndash104 2004

[28] K Lee J Y Kim S H Park S H Kim S Cho and A J HeegerldquoAir-stable polymer electronic devicesrdquo Advanced Materialsvol 19 no 18 pp 2445ndash2449 2007

[29] P W M Blom M J M de Jong and M G van Munster ldquoElec-tric-field and temperature dependence of the hole mobility inpoly(p-phenylene vinylene)rdquo Physical Review B vol 55 no 2pp R656ndashR659 1997

[30] D H Dunlap P E Parris and V M Kenkre ldquoCharge-dipolemodel for the universal field dependence of mobilities in mo-lecularly doped polymersrdquo Physical Review Letters vol 77 no3 pp 542ndash545 1996

[31] G G Malliaras J R Salem P J Brock and C Scott ldquoElectricalcharacteristics and efficiency of single-layer organic light-emitting diodesrdquo Physical Review B vol 58 no 20 pp R13411ndashR13414 1998


Research ArticleMaximum Power Point Tracking Method Based on ModifiedParticle Swarm Optimization for Photovoltaic Systems

Kuei-Hsiang Chao Long-Yi Chang and Hsueh-Chien Liu

Department of Electrical Engineering National Chin-Yi University of Technology No 57 Section 2 Zhongshan RoadTaiping Distatrict Taichung 41170 Taiwan

Correspondence should be addressed to Kuei-Hsiang Chao chaokhncutedutw

Received 9 June 2013 Accepted 20 September 2013


Copyright copy 2013 Kuei-Hsiang Chao et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

This study investigated the output characteristics of photovoltaic module arrays with partial module shading Accordingly wepresented a maximum power point tracking (MPPT) method that can effectively track the global optimum of multipeak curvesThis method was based on particle swarm optimization (PSO)The concept of linear decreases in weighting was added to improvethe tracking performance of the maximum power point tracker Simulation results were used to verify that this method couldsuccessfully track maximum power points in the output characteristic curves of photovoltaic modules with multipeak values Theresults also established that the performance of the modified PSO-based MPPT method was superior to that of conventional PSOmethods

1 Introduction

The output power of photovoltaic (PV) systems is heavilyinfluenced by irradiation and temperature Thus the outputpower of these systems presents nonlinear changes Maxi-mum power point tracking (MPPT) must be incorporated tostabilize the output power at maximum power points Thisresearch topic is critical for photovoltaic power generationsystems

The voltage feedback method [1] is the simplest of conve-ntionalMPPTmethods However this method requires priortesting of maximum power point (MPP) voltage In add-ition MPPs cannot be tracked when the photovoltaic mod-ules deteriorate and cause the MPPs to change unless re-measurement is performed The framework of the constantvoltage tracking method [2 3] is simple and does notrequire complex formulasThis method employs the extremesimilarity of MPP voltages under varying amounts of irra-diation using MPP voltage under standard test conditionsas reference points to set the operation of the photovoltaicmodule arrays at these points However when the amount ofirradiation is low or when photovoltaic module temperatureschange the difference in values between theMPP voltage andthe reference voltage becomes substantial reducing tracking

accuracyTheperturb and observemethod [4] uses periodicalincreases or decreases in voltage to perturb the system Ifthese perturbations increase the output power the same trendis used to change (ie increase or decrease) voltage thefollowing time If output power decreases the reverse trendis used to change voltage However this method is unableto track MPPs accurately In addition the method osci-llates near the MPP and increases tracking losses therebyaffecting power generation efficiency These conventionalmethods only perform MPPT on module arrays with single-peak characteristic curves When multipeak values appear inthe characteristic curves thesemethods frequently track localMPPs and miss global MPPs

Recently numerous scholars have presented intelligentMPPT methods [5ndash11] for photovoltaic module arrays bothto track MPPs accurately and to improve the dynamic andsteady-state tracking performance However these methodsare applicable only to MPPT in photovoltaic module arrayswithout shading Nevertheless the appearance of multi-peak output curves because of partial module shading inphotovoltaic module arrays is commonTherefore the devel-opment of an algorithm for accurately tracking the trueMPPs of complex and nonlinear output curves is crucialReference [12] presented a MPP tracker based on particle


2

16

12

08

04

00 20 40 60 80 100

02

06

1

18

14

One module at 30 shadow ratio

With no shadowTwo modules at 30 shadow ratio

Three modules at 30 shadow ratio

Four modules at 30 shadow ratio

VPV (V)

IPV

(A)

Figure 1 I-V output characteristic curves in the four-series one-parallel module array with varying numbers of modules and shaderatios of 30 as simulated by Solar Pro software

swarm optimization (PSO) for photovoltaic module arraysAlthough this tracker was capable of tracking global MPPsof multipeak characteristic curves because fixed values wereadopted for weighing within the algorithm the trackingperformance lacked robustness causing low success rateswhen tracking global MPPs Even though the MPPs weretracked successfully the dynamic response speed was low

Therefore this study used PSO and added improvementspreventing it from being trapped in local MPPs (ie search-ing only local MPPs) and enabling it to track global MPPsquickly and consistently on the multipeak characteristiccurves of photovoltaic module arrays

2 PV Module Array Characteristic Analysis

Within photovoltaic power generation systems photovoltaicmodules are typically connected to formmodule arrays usingseries and parallel methodsWhen certain modules malfunc-tion or are shaded their electric current flow is impeded andhot spots occur damaging the photovoltaic modules Thusbypass diodes through which electric current circulatesare generally installed This enables the other photovoltaicmodules to operate normally However when some modulesmalfunction or are shaded the output voltage and current ofthe module arrays decrease producing multipeak values inthe P-V and I-V output characteristic curves

To find out the output characteristics of series and parallelmodule arrays when some modules are shaded a four-seriesone-parallel photovoltaic module array consisting of HIP2717 modules produced by Sanyo was investigated Figures1 and 2 show the I-V and P-V output characteristic curves ofone two three and four modules with 30 shading in thefour-series one-parallel module array simulated using SolarPro software [13] The figures show that shading on somemodules within the array causes the emergence of multipeakvalues in the output characteristic curves In addition themaximum output power points showed growing declines asthe number of shaded modules increased

120

20

00 20 40 60 80 100

40

60

100

80

VPV (V)





PPV

(A)

Figure 2 P-V output characteristic curves in the four-series one-parallel module array with varying numbers of modules and shaderatios of 30 as simulated by Solar Pro software

3 Particle Swarm Optimization

PSO is an optimization theory presented by Kennedy andEberhart in 1995 [14] The theory is an algorithm that usescollective intelligence it belongs to a branch of evolutionarycomputation The two scholars were inspired by the foragingbehavior of birds and applied this phenomenon to resolveproblems related to search and optimization [15 16] Eachbird flying in a space is called a particle All particles movingin the space have fitness values mapped by an objective func-tion and individual velocities which are used to determinethe direction and distance of their movement Two memoryvalues influence themovement of the particles119875best and119866bestEach particle stores the best position it is currently seekingin an individual best memory position (119875best) Furthermorememory intercommunication is present among the particlesthe particle swarm compares individual positions to locatethe best position and stores this as the swarm best position(119866best) The particle swarm uses this method to revise thedirection and velocity of its movement continuously therebyrapidly converging toward a global optimum

The PSO calculation process is as follows

(1) Set the number of particles the maximum number ofiterations the weight and the learning factors

(2) Initialize the particle swarm and randomly assignpositions and velocities for each particle

(3) Substitute the initial positions into the objective fun-ction to assess the fitness values for each particle

(4) Compare the fitness values and the individual bestmemory positions (119875best) of each particle to selectbetter positions and update 119875best

(5) Compare 119875best and the swarm best memory value119866best If 119875best is superior to 119866best update 119866best


Photovoltaic array

MOSFETdriver

Modified PSO-based MPPT

controller

DCACinverter Load

PWM control signal

Cin

R1

R2

Cout

V

L

D

PV IPV

Figure 3 System architecture for modified PSO-based MPPT control

(6) Use the core PSO formulas to update particle veloc-ities and positions These formulas are shown asfollows

119881119895+1

119894= 119882 times 119881

119895

119894+ 1198621times rand1 (∙) times (119875best minus 119875

119895

119894)

+ 1198622times rand2 (∙) times (119866best minus 119875

119895

119894)

(1)

119875119895+1

119894= 119881119895+1

119894+ 119875119895

119894 (2)

In (1) and (2) 119881119895119894and 119875119895

119894are the velocity and position of

particle 119894 at iteration j respectively Variables rand1(∙) andrand2(∙) are random number generators that generate realnumbers between 0 and 1 randomly these numbers are usedto strengthen the variability of the particle swarm 119882 isthe weight 119862

1and 119862

2are the learning factors 119875best

119894

is theindividual optimum of particle 119894 and 119866best is the swarm orglobal optimum

(7) Stop tracking if the stop conditions are met Other-wise rerun Steps 4 through 6The stop conditions areeither locating the global optimum or reaching themaximum number of iterations

The search efficiency and success rate of PSO are deter-mined primarily by the values assigned for the weights andthe learning factors [17] When the weight is too high theparticle search might lack accuracy because the movementstep sizes are too large However if the weight is low par-ticle movement becomes slow and the local optimum trapmight be unavoidable when facing multipeak values Thusweighting is typically based on the objective function

4 PSO-Based MPPT for PV Systems

Conventional PSO is fast and accurate when searching forthe output characteristic curves of PV module arrays withsingle peak values However when somemodules are shadedweights in conventional PSO must be readjusted appropri-ately based on various multipeak curve characteristics If thisis not performed excessively high or low weights result in

tracking failure Thus conventional PSO-based MPPT mustbe modified when some of the modules in a photovoltaicmodule array are shaded

To solve these problems linear decreases in line withincreasing iteration numbers were adopted in this study forthe weighting of the PSO kernel formulas The modified wei-ghting formula is as follows

119882 = (119882max minus119882min) times(119899 minus 119895)

119899

+119882min (3)

where119882max is the maximum weight 119882min is the minimumweight 119899 is the maximum number of iterations and 119895 is thecurrent iteration number

The physical meaning of this modified weighting formulais that greater step sizes are used to increase the particlesearch velocity during the initial search because the distanceto the global optimum is relatively large This prevents anexcessively small step size from making local optimum trapsunavoidable However 119882 decreases gradually as the num-ber of iterations increases Because the particles are nowapproaching the MPP these decreases in 119882cause the stepsin the particle movements to shrink enabling the particles totrack the MPP more accurately

In addition the output curve of the photovoltaic modulearray appears only in the first quadrant Therefore regionswith output power below zero cannot be optimum positionsand the lower limit for particle tracking is set to zero thatis particles automatically return to zero when tracking regi-ons with values of less than zero This greatly reduces thetime wasted by particles tracking in erroneous regions Thepredicate for these conditions is as shown in (4)

119875best119894

=

119875best119894

119875best119894

gt 0

0 119875best119894

le 0

(4)

The 119875best119894

value obtained in (4) is the solution of (2)

119875best119894

= 119875119895+1

119894 (5)

Figure 3 shows the proposed system architecture formodified PSO-based MPPT control This system contains


0 10 20 30 40 50 60 70 80 900

20

40

60

PPV

(W)

VPV (V)

(a)

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

Iterations

Modified PSO

Conventional PSO

Gbe

st(W

)

(b)

Figure 4 Simulation results for a four-series one-parallel modulearray with one module shaded 30 and one module shaded 55and(a) P-V characteristic curve (b) tracking comparison resultsbetween the conventional PSO (W = 04) and modified PSO-basedMPPT methods

two major subsystems (1) a boost converter and (2) a MPPTcontroller The MPPT controller is used to control the dutycycle of the boost converter [18] allowing the photovoltaicmodule array to output maximum power despite partialshading of some modules


MATLAB software [19] was used to simulate and comparethe application of the conventional and modified PSO-basedMPPT control methods to track photovoltaic module arraysin four shading situationsThe first test situation involved onemodule with 30 shading and one module with 55 shadingin a four-series one-parallel module array Figure 4(a) showsthis P-V characteristic curve and Figure 4(b) shows theresults of a comparison between the conventional (with aweight of 04) and modified PSO-based MPPTmethodsThefigures indicate that the P-V characteristic curve exhibitedthree peak values when the two modules in the sameseries had differing amounts of shade In this situation theconventional PSO-based MPPT method could track onlylocal maxima whereas the modified PSO method couldtrack global MPPs The modified PSO method also had afaster response speed than the conventional PSO method Inthe second test situation one module was shaded 25 andthe other (in another series) was shaded 30 in the four-series one-parallel module array Figure 5(a) shows the P-V characteristic curve and Figure 5(b) shows the results ofa comparison between the conventional (with a weight of04) and modified PSO-based MPPT methods Under theseworking conditions the P-V characteristic curve had twopeaks The MPPT results in Figure 5(b) indicate that theconventional PSO method tracked the local optimum for

0 10 20 30 40 50 60 70 80 900

20

40

60

80

PPV

(W)

VPV (V)

(a)

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

Iterations

Modified PSO

Conventional PSO

Gbe

st(W

)

(b)

Figure 5 Simulation results for a four-series one-parallel modulearray with one module shaded 25 and one module shaded 30and (a) P-V characteristic curve (b) tracking comparison resultsbetween the conventional PSO (W = 04) and modified PSO-basedMPPT methods

0 5 10 15 20 25 30 35 40 450

20

40

60

80

PPV

(W)

VPV (V)

(a)

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

Iterations

Modified PSO

Conventional PSO

Gbe

st(W

)

(b)

Figure 6 Simulation results for a two-series two-parallel modulearray with onemodule shaded 25 and onemodule shaded 30 (a)P-V characteristic curve (b) tracking comparison results betweenthe conventional PSO (W = 04) and modified PSO-based MPPTmethods

the first peak and was unable to avoid the local maximumtrap In contrast the modified PSO method successfullytracked the globalMPP In the third test situation onemodulewas shaded 25 and the other was shaded 30 in a two-series two-parallel module array Figure 6(a) shows the P-V characteristic curve and Figure 6(b) shows the results ofa comparison between the conventional (with a weight of


Table 1 Parameter settings for the two PSO methods

Methods Parameter valuesWeight (119882) Learning factors (119888

1 1198882) Maximum number of iterations

Conventional PSO method (i) 04 with module shading (2 2) 100(ii) 07 without module shading

Modified PSO method Linearly decreasing (2 2) 100119882max 09 119882min 04

0 10 20 30 40 50 60 70 80 900

50

100

150

PPV

(W)

VPV (V)

(a)

0 10 20 30 40 50 60 70 80 90 1000

50

100

150

Iterations

Conventional PSO

Modified PSO

Gbe

st(W

)

(b)

0 10 20 30 40 50 60 70 80 90 1000

50

100

150

Iterations

Modified PSO

Conventional PSO

Gbe

st(W

)

(c)

Figure 7 Simulation results for a four-series one-parallel modulearray without any shading (a) P-V characteristic curve (b) trackingcomparison results between the conventional PSO (W = 04) andmodified PSO-based MPPT methods (c) tracking comparisonresults between the conventional PSO (W = 07) andmodified PSO-based MPPT methods

04) and modified PSO-based MPPT methods Figure 6(b)indicates that although the conventional PSO method wasalso able to track the true MPP successfully its responsespeed was substantially slower than that of the modifiedPSO method The fourth test situation involved a four-seriesone-parallel module array without any shading Figure 7(a)shows the P-V characteristic curve and Figure 7(b) showsthe results of a comparison between the conventional (witha weight of 04) and modified PSO-based MPPT methodsThe figures indicate that the modified PSO method wasalso able to track the true MPP accurately without the

occurrence of multipeak characteristic curves In contrastbecause weighing was maintained at 04 the conventionalPSO method was unable to track the true MPP Figure 7(c)shows that both the conventional and the modified PSOmethods could track the true MPP when the weighting ofconventional PSO method was adjusted to 07 However themodified PSO method still had better response performancein dynamic tracking

Theweight of the conventional PSOmethod tested in thisstudy was set at 04 with which its tracking success rate wasthe highest when themodules were shadedWithout shadingbecause output power increased substantially the weight ofthe conventional PSOmethodwas reset to 07 to increase stepsize during tracking thereby allowing the tracking successrate to reach 100 Linear decreases from 09 to 04 were usedfor weighting the modified PSO method in both situationsFor both tracking methods the learning factors 119862

1and 119862

2

were set to a fixed value of 2 and the maximum number ofiterations was set to 100 Table 1 shows the detailed parametersettings and Table 2 shows the comparison results for thesuccess rates of the two methods after 100 tracking attemptsTable 2 indicates that the success rate of the modified PSOmethod for tracking true MPPs was significantly greatercompared to the conventional PSO method regardless ofwhether modules were shaded This is because the modifiedPSO method employed linearly-adjusted weighting whichenabled it to track true MPPs at a 100 success rate undervarious shading conditions In contrast for the conventionalPSO method to reach a 100 success rate regarding single-peak curves and no shading appropriate weights based onoutput power were required Furthermore tracking using theconventional PSOmethod frequently became trapped in localMPPs when multipeak characteristic curves appeared

6 Conclusion

The modified PSO-based MPPT method presented in thisstudy is a novel algorithm based on conventional PSO Theprimary feature of this method is the linear decreases usedto adjust weighting in contrast to the fixed weights adoptedby conventional PSOThemodified PSOmethod was appliedto maximum power tracking in photovoltaic module arrayssuccessfully solving the inability to track true MPPs becauseof partial module shading in photovoltaic module arraysThetracking success rate reached 100 for the modified methodand the tracking speed was superior to that of conventionalPSO This could reduce energy loss during the MPPT pro-cess thereby substantially enhancing the power generationefficiency of photovoltaic power generation systems


Table 2 Comparison of the tracking success rates for the conventional and modified PSO methods

MPPT methods

Module array configuration and shade conditions

Four-series one-parallelone module with 30

shading and one modulewith 55 shading



Two-series two-parallelone module with 25


Four-series one-parallel nomodule shading

Conventional PSOmethod 75100 attemptslowast 72100 attempts 71100 attempts

(i) With weight set to 0449100 attempts

(ii) With weight set to 07100100 attempts

Modified PSOmethod 100100 attempts 100100 attempts 100100 attempts 100100 attemptslowastNote Number of successestotal number of attempts


The authors of the paper declare that there is no conflict ofinterest with any of the commercial identities mentioned inthe paper

Acknowledgment

This work was supported by the National Science CouncilTaiwan Republic of China under the Grant NSC 102-ET-E-167-001-ET

References

[1] A Barchowsky ldquoA comparative study of MPPT methods fordistributed photovoltaic generationrdquo in Proceedings of the IEEEInnovative Smart Grid Technologies Conference pp 1ndash7 January2012

[2] M A S Masoum H Dehbonei and E F Fuchs ldquoTheoret-ical and experimental analyses of photovoltaic systems withvoltage- and current-based maximum power-point trackingrdquoIEEE Transactions on Energy Conversion vol 17 no 4 pp 514ndash522 2002

[3] Y S Xiong S Qian and JM Xu ldquoResearch on constant voltagewith incremental conductance MPPT methodrdquo in Proceedingsof the Power and Energy Engineering Conference March 2012

[4] N FemiaDGranozio G Petrone G Spagnuolo andMVitellildquoPredictive and adaptive MPPT perturb and observe methodrdquoIEEE Transactions on Aerospace and Electronic Systems vol 43no 3 pp 934ndash950 2007

[5] K H Chao and Y H Lee ldquoA maximum power point trackerwith automatic step size tuning scheme for photovoltaic sys-temsrdquo International Journal of Photoenergy vol 2012 Article ID176341 10 pages 2012

[6] R Ramaprabha V Gothandaraman K Kanimozhi R Divyaand B L Mathur ldquoMaximum power point tracking using GA-optimized artificial neural network for solar PV systemrdquo inProceedings of the 1st International Conference on ElectricalEnergy Systems (ICEES rsquo11) pp 264ndash268 Newport Beach CalifUSA January 2011

[7] A H Besheer ldquoAnt colony system based PI maximum powerpoint tracking for stand alone photovoltaic systemrdquo in Proceed-ings of the IEEE International Industrial Technology Conferencepp 693ndash698 March 2012

[8] MAdly ldquoAnoptimized fuzzymaximumpower point tracker forstand alone photovoltaic systems ant colony approachrdquo in Pro-ceedings of the 7th IEEE Industrial Electronics and ApplicationsConference pp 113ndash119 July 2012

[9] K H Chao and C L Chiu ldquoDesign and implementation of anintelligentmaximumpower point tracking controller for photo-voltaic systemsrdquo International Review of Electrical Engineeringvol 7 no 2 pp 3759ndash3768 2012

[10] H T Yau and C U Wu ldquoComparison of extremum-seekingcontrol techniques for maximum power point tracking inphotovoltaic systemsrdquo Energies vol 4 no 12 pp 2180ndash21952011

[11] H Zazo E del Castillo J F Reynaud and R Leyva ldquoMPPTfor photovoltaic modules via newton-like extremum seekingcontrolrdquo Energies vol 5 pp 2653ndash2666 2012

[12] L R Chen C H Tsai Y L Lin and Y S Lai ldquoA biologicalswarm chasing algorithm for tracking the PVmaximum powerpointrdquo IEEE Transactions on Energy Conversion vol 25 no 2pp 484ndash493 2010

[13] K H Tang K H Chao Y W Chao and J P Chen ldquoDesignand implementation of a simulator for photovoltaic modulesrdquoInternational Journal of Photoenergy vol 2012 Article ID368931 6 pages 2012

[14] J Kennedy and R C Eberhart ldquoParticle swarm optimizationrdquoin Proceedings of the IEEE International Conference on NeuralNetworks vol 4 pp 1942ndash1948 December 1995

[15] R C Eberhart and J Kennedy ldquoA new optimizer using particleswarm theoryrdquo in Proceedings of the 6th International Sympo-sium onMicroMachine and Human Science pp 39ndash43 October1995

[16] D Srinivasan W H Loo and R L Cheu ldquoTraffic incidentdetection using particle swarm optimizationrdquo in Proceedings ofthe IEEE Swarm Intelligence Symposium pp 144ndash151 April 2003

[17] W H Han P Yang H Ren and J Sun ldquoComparison study ofseveral kinds of inertia weights for PSOrdquo in Proceedings of the1st IEEE International Conference on Progress in Informatics andComputing (PIC rsquo10) pp 280ndash284 Shanghai China December2010

[18] M Elshaer A Mohamed and O Mohammed ldquoSmart optimalcontrol of DC-DC boost converter in PV systemsrdquo in Proceed-ings of the IEEEPES Transmission and Distribution Conferenceand Exposition Latin America (T and D-LA rsquo10) pp 403ndash410Sao Paulo Brazil November 2010

[19] MATLAB Official Website httpwwwmathworkscompro-ductsmatlab


Research ArticleEffect of Subgrains on the Performance of Mono-LikeCrystalline Silicon Solar Cells

Su Zhou Chunlan Zhou Wenjing Wang Yehua Tang Jingwei ChenBaojun Yan and Yan Zhao

The Key Laboratory of Solar Thermal Energy and Photovoltaic System Institute of Electrical EngineeringChinese Academy of Sciences Beijing 100190 China

Correspondence should be addressed to Wenjing Wang wangwjmailieeaccn



Copyright copy 2013 Su Zhou et alThis is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

The application of Czochralski (Cz) monocrystalline silicon material in solar cells is limited by its high cost and serious light-induced degradation The use of cast multicrystalline silicon is also hindered by its high dislocation densities and high surfacereflectance after texturing Mono-like crystalline silicon is a promising material because it has the advantages of both mono- andmulticrystalline silicon However when mono-like wafers are made into cells the efficiencies of a batch of wafers often fluctuatewithin a wide range of gt1 (absolute) In this work mono-like wafers are classified by a simple process and fabricated into laserdoping selective emitter cellsThe effect andmechanism of subgrains on the performance of mono-like crystalline silicon solar cellsare studiedThe results show that the efficiency of mono-like crystalline silicon solar cells significantly depends on material defectsthat appear as subgrains on an alkaline textured surfaceThese subgrains have an almost negligible effect on the optical performanceshunt resistance and junction recombination but significantly affect the minority carrier diffusion length and quantum efficiencywithin a long wavelength range Finally an average efficiency of 182 is achieved on wafers with hardly any subgrain but with asmall-grain band

1 Introduction

Most current industrial solar cells are made of Czochralski(Cz)-grown monocrystalline silicon material and cast mul-ticrystalline silicon substrates [1] However Cz monocrys-talline silicon material has a high cost and undergoes seriouslight-induced degradation (LID) of efficiency under sunlightDefects with very high dislocation densities and high sur-face reflectance after texturing can limit the application ofmulticrystalline silicon [2ndash4] Mono-like crystalline siliconwhich has the advantages of both Cz and multicrystalline sil-icon is fabricated using a Czmonoseed layer or by optimizedgrowth nucleation during ingot casting [5 6] Recently thesesquaremono-like crystalline wafers have gained considerableattention because of their low structural defect density lowfabrication cost and weak LID [7 8] Several approaches toimprove the quality ofmono-like crystalline silicon have beenpresented [9ndash11] However an important remaining issue isthat when mono-like wafers are made into cells the range of

efficiencies of a batch of wafers fluctuates within a wide rangeof gt1 (absolute) [12]

This paper aims to explain the abnormal efficiency fluctu-ation of mono-like wafers Mono-like wafers were classifiedby the subgrain content after alkaline texturing and thenfabricated into laser doping selective emitter (LDSE) cellsThe optical and electrical performances of these cells wereanalyzed and several measurements were performed toestimate the effects of subgrains on solar cells Finally semi-mono-like crystalline silicon wafers which have almost nosub-grain but have a small-grain band are chosen to prepareLDSE cells to demonstrate the effect of subgrains on cellperformance

2 Experimental

All wafers used in this study were commercial grade 1Ωsdotcm119901-type mono-like crystalline Si wafers with an area of 6square inches (2434 cm2) and thickness of sim200120583m Wafers


(a)

Subgrains

Monocrystalline area

(b)

Figure 1 Optical image of a part of the alkaline textured wafer with sub-grain regions (a) vertical view and (b) with a tilt angle of 45∘

are typically optically perfect and appear like square Czmonowafers After texturing in a mixture of 1 NaOHand 4 isopropyl alcohol some subgrains appeared onsome parts of the mono-like wafers as shown in Figure 1Figure 1(a) shows an alkaline textured surface like monocrys-talline wafer However when the image was taken at a tiltangle of 45∘ subgrains with an estimated size of 3ndash7mm canbe observed as shown in Figure 1(b)

According to the content of these subgrains the waferswere divided into three classes Wafers with ratios of sub-grains area to total wafer area of lt10 sim50 and gt90 ontheir surface were defined as grades A B and C respectivelyAll wafers after alkaline texturing were phosphorus diffusedto a sheet resistance of 80 Ω◻ with POCl

3liquid source An

industrial wet chemical etching process was then performedto achieve edge junction isolation A SiN

119909antireflection

coating (ARC) was deposited onto the front surface of thewafers using an industrial remote plasma-enhanced chemicalvapor deposition system Aluminum (Al) paste was thenscreen printed on the rear surface of the wafers and fired in abelt furnace at 900∘C to form the back surface field of the cellsDiluted phosphoric acid was spin coated on the SiN

119909film

A 532 nm Q-switched NdYAG laser was used to remove thedielectric layer and pattern laser-doping n+ finger patterns onthe 119899-type surface simultaneously Nickel (Ni) and silver (Ag)were then plated onto the patterned fingers by light-inducedplating and sintered to form Ni silicate which provided low-resistance contact

The surface reflectance of textured and passivated sam-ples and internal quantum efficiency (IQE) of cells within therange of 300 nm to 1200 nm were measured using a solar cellspectral responsequantum efficiency measurement system(QEX7 PV Measurement) The electroluminescence (EL)images light beam-induced current (LBIC) and diffusionlength of fabricated solar cells were characterized using aninfrared defect inspection tool (ELT C02 ASIC) and tabletopPV measurement system (WT-2000 Semilab) A current-voltage (IndashV) tester was used to obtain both dark and illu-minated IndashV curves and to assess the electrical performances

of the laser-doped 119901-type mono-like crystalline Si solar cellswith different sub-grain amounts


31 Optical Performances and IQE Figure 2 shows an exper-imental comparison of the percentage reflectance of texturedand ARC-coated wafers with the IQE of cells with sub-grain and monocrystalline areas within a wavelength rangeof 300 nm to 1200 nm The overall reflectance of sub-grainareas is similar to that of monocrystalline areas exceptfor a slight increase within the short wavelength range of300 nm to 400 nm The weighted reflectances of sub-grainand monocrystalline areas after ARC coating are 403 and404 respectively The similar weighted reflectance meansthat the subgrains have minimal effect on the light trappingof the pyramid texture However as shown in Figure 2 theIQE of cells with sub-grain areas is significantly decreasedwithin the range of 600 nm to 1100 nm wavelength comparedwith the result of monocrystalline areas According to thedeep penetration of long wavelength light in silicon thereduction of IQE may be attributed to the recombinationin silicon substrate Thus the recombination rate is likely tobe higher in sub-grain areas than in monocrystalline areaswhich decreases the IQE of fabricated cells within the longwavelength range under the same light condition

32 Analysis on Solar Cell Parameters Table 1 comparesvarious cell parameters fabricated by wafers with differentsub-grain contents For each type of silicon wafer 10 solarcells were fabricated andmeasuredWith increased sub-graincontent from lt10 to gt90 the cell efficiency decreasesfrom 176 to 160 With increased sub-grain content theopen-circuit voltage (119881oc) of the cell decreases from 6289mVto 6159mV and the short-circuit current density (119869sc) ofthe cell decreases from 3687mAcm2 to 3435mAcm2 Thedecrease in the efficiency can be attributed to the significantdecrements in 119881oc and 119869sc


IQE of wafer with monocrystallineIQE of wafer with subgrainsTextured wafer with monocrystallineTextured wafer with subgrainsARC-coated wafer with monocrystallineARC-coated wafer with subgrains

Wavelength (nm)

50

40

30

20

10

0

100

80

60

40

20

01200400 600 800 1000

IQE

()

Refle

ctiv

ity (

)

Figure 2 Reflectance of textured and ARC-coated wafers and IQEof cells with subgrains and monocrystalline areas

0 01 02 03 04 050

1

2

3

Grade AGrade BGrade C

Loca

l ide

ality

fact

or

Voltage (V)

Figure 3 Local ideality factor curves derived from the dark 119868-119881 curves of monocrystalline LD solar cells fabricated on differentsubstrates

The local ideality factor 119898 of a solar cell in the dark isgiven by

119898 =

1

119881119879

(

119889119881

119889 ln (119868)) (1)

where 119881 and 119868 are the measured dark voltage and currentrespectively and 119881

119879= 0026 eV The local ideality factors in

different voltage regions indicate differentmechanismswhich

Table 1 Average electrical parameters of 10 solar cells fabricated ondifferent substrates

Grade 119881oc (mV) 119869sc (mAcm2) FF () Eta ()A 6289 plusmn 12 369 plusmn 01 758 plusmn 06 176 plusmn 02

B 6216 plusmn 23 359 plusmn 02 752 plusmn 04 168 plusmn 02

C 6159 plusmn 30 344 plusmn 02 757 plusmn 06 160 plusmn 03

may have an effect on cell performance [13 14] Figure 3shows the local ideality factor curves derived from the dark119868-119881 curves of mono-like crystalline laser-doping (LD) solarcells fabricated on different substrates The local idealityfactors of different-grade wafers are similar (gt2) around thelow-voltage region (lt04V) indicating that shunting mayhave occurred in all these cells This shunting problem maybe caused by insufficient edge isolation The local idealityfactors of different-grade wafers around the medium-voltageregion (near the maximum power point) are also similarindicating that the subgrains hardly affect the formationof localized Schottky contacts The result shows that theshunting property andmetal contact of mono-like crystallineLD solar cells are independent of the subgrains

33 EL Image Figure 4 shows the EL images of cells fabri-cated with different sub-grain contents With increased sub-grain content dark line clusters spread from a small part toalmost the entire surface These dark line clusters indicatea low EL intensity which can be caused by series resis-tance variations or locally enhanced recombination Thesedark line clusters are also closely correlated with subgrainsobserved on the wafer surface This finding indicates thatthese subgrains which can represent material defects suchas grain boundaries and dislocations play an important rolein enhancing series resistance or recombination of minoritycarriers Grain boundaries and dislocations are known to beeasily generated during casting and the dislocation densityincreases from bottom to top of the ingots [15 16] Thedislocation density especially dislocation clusters correlatedwith subgrains has been found to affect the recombinationof light-generatedminority carriers significantly and thus thesolar cell efficiencies of cast multicrystalline silicon [17 18]Therefore dislocations caused by sub-grain formation can beinferred to result in the recombination of minority carriersand the dark line clusters found in the EL images

34 Minority Carrier Diffusion Length Figure 5 shows var-ious spatial distributions of cells with different sub-graincontents Red represents the regions with a minority carrierdiffusion length (MCDL) of sim130 120583m and black representsthose with an MCDL of sim420120583m Strong local variationsin MCDL can be clearly seen in different regions fromFigures 5(a) to 5(c) With increased sub-grain content thediffusion length significantly decreases and the diffusionlength distribution changes from uniform to nonuniformMoreover these regions of low diffusion length in Figures5(b) and 5(c) also correspond with subgrains observed onthe wafer surface These reduced diffusion length regionswhich indicate locally enhanced recombination may lead to


(a) (b)

(c)

Figure 4 EL images of solar cells fabricated on different substrates ((a)ndash(c)) grades AndashC respectively

decreased119881oc Those longitudinal lines in figures are resultedfrom fingers of solar cells by blocking the testing laser spotlight during the test

35 LBIC Measurement Further characterizations were per-formed to determine the influence of subgrains on recom-bination The IQE distribution which was measured by thetwo-dimensional LBICmethod can represent recombinationactivity in different regions at different wavelengths of light[19ndash21] The IQE distributions measured at 405 and 979 nmare shown in Figures 6 and 7 The uniform and similar IQEdistributions of cells with different sub-grain contents inFigure 5 indicate that the locally enhanced recombinationcaused by subgrains has hardly any effects on the light-induced current at 405 nmThe penetration depth of 405 nmwavelength light is about 500 nm in crystalline siliconTherefore the IQE measured at this wavelength indicatesthe information of emitter and PN junction The strongfield passivation effect caused by the PN junction and thedominating auger recombination caused by the highly doped

119899-type emitter may result in uniform light-induced currentand IQE at 405 nm

However the result measured at 979 nm shows a sig-nificant difference With increased sub-grain content theIQE significantly decreases and the IQE distribution changesfrom uniform to non-uniform These non-uniform areasin Figure 7 correspond with sub-grain regions The pene-tration depth of 979 nm wavelength light is about 99120583min crystalline silicon Therefore the IQE measured at thiswavelength indicates information on the bulk material Thelocally enhanced recombination caused by defects and grainboundaries shown as subgrains significantly affects the light-induced current in the bulk silicon

36 Solar Cell Performance of Wafers without Subgrains Toillustrate the effect of subgrains on cell performance waferswith hardly any sub-grain but with a small-grain band whichmeans an area with many small grains were chosen toprepare LDSE cells Figures 8(a) and 8(b) show optical andEL images of fabricated solar cells respectively Hardly any


(a) (b)

(c)

Figure 5 Spatial distribution of minority carrier diffusion length of solar cells fabricated on different substrates ((a)ndash(c)) grades AndashCrespectively

Table 2 Average electrical parameters of 10 solar cells fabricated onwafers with hardly any sub-grain but with a small-grain band

119881oc (mV) 119869sc (mAcm2) FF () Eta ()6341 plusmn 14 372 plusmn 01 769 plusmn 05 182 plusmn 01

dark line clusters can be observed in the EL image The smallstraight line dark regions are caused by the slight peelingof grid lines The average electrical performance results areshown in Table 2 As shown in Figure 8(a) the small-grainband becomes a high-reflectivity area after alkaline texturingleading to more optical loss Despite the high reflectivity ofthe small-grain band 119881oc and 119869sc of cells with a small-grainband are still higher than that of the grade A mono-likecells which have a ratio of subgrains area to total wafer areaof lt10 Average 119881oc increases from 6289mV to 6341mVand average 119869sc increases from 369mAcm2 to 372mAcm2respectively These increments in119881oc and 119869sc indicate that thegrain boundaries and defects represented by subgrains havemore effects on the electrical performance of mono-like solarcells than high-reflectivity small-grain band

4 Conclusions

The application of optically perfect mono-like wafers canbe challenging because of abnormal efficiency fluctuations

This study showed that cell efficiency decreases withincreased sub-grain content The reflectance of monocrys-talline and sub-grain areas after texturing and ARC coatingis the same However the IQE of cells with sub-grain areassignificantly decreases within 600 nm to 1100 nm comparedwith monocrystalline areas The local ideality factor resultshows that the shunting property andmetal contact of mono-like crystalline LD solar cells are independent of subgrainsHowever subgrains shown as dark line clusters in the ELimage significantly decrease the diffusion length and the IQEat long wavelengths in these regions This finding indicatesthat defects caused by subgrains act as the recombinationcenter for minority carriers degrading the IQE in middle-and long-wavelength range and further affecting the solarcell performance Finally an average efficiency of 182 wasachieved on wafers with hardly any sub-grain but with asmall-grain band to indicate that the negative effect of sub-grain on cell performance would be more serious than thatof small-grain band This average efficiency also illustratesthe promising application potential of mono-like wafers Itcan be inferred that higher efficiency would be obtained onthe mono-like wafers without any sub-grain areas and small-grain bandsThus defects appeared as subgrains significantlyaffect cell performance and dislocations must be eliminatedby optimizing the casting process for the industrial applica-tion of mono-like crystalline silicon


(a) (b)

(c)

Figure 6 IQE distribution of solar cells fabricated on different substrates at 405 nm wavelength ((a)ndash(c)) grades AndashC respectively

(a) (b)

(c)



(a) (b)

Figure 8 Optical (a) and EL (b) images of solar cells fabricated on wafers with hardly any sub-grain but with a small-grain band

Acknowledgments

This work was supported by the National High TechnologyResearch and Development Program of China (Grant no2011AA050515) and National Basic Research Program ofChina (Grant no 2012CB934204)

References

[1] N Asim K Sopian S Ahmadi et al ldquoA review on the roleof materials science in solar cellsrdquo Renewable and SustainableEnergy Reviews vol 16 no 8 pp 5834ndash5847 2012

[2] D Cunningham A Parr J Posbic and B Poulin ldquoPerfor-mance comparison between BP solar mono2 and traditionalmulticrystalline modulesrdquo in Proceedings of 23rd EuropeanPhotovoltaic Solar Energy Conference and Exhibition(EUPVECrsquo08) pp 2829ndash2833 Valencia Spain 2008

[3] A Jouini D Ponthenier H Lignier et al ldquoImprovedmulticrys-talline silicon ingot crystal quality through seed growth for highefficiency solar cellsrdquo Progress in Photovoltaics Research andApplications vol 20 no 6 pp 735ndash746 2012

[4] K Nakajima K Morishita R Murai and K KutsukakeldquoGrowth of high-quality multicrystalline Si ingots using non-contact crucible methodrdquo Journal of Crystal Growth vol 355no 1 pp 38ndash45 2012

[5] N Stoddard B Wu I Witting et al ldquoCasting single crystalsilicon novel defect profiles from BP solarrsquos mono2 wafersrdquoSolid State Phenomena vol 131-133 pp 1ndash8 2007

[6] V Prajapati E Cornagliotti R Russell et al ldquoHigh efficiencyindustrial silicon solar cells on silicon Mono2 cast materialusing dielectric passivation and local BSFrdquo in Proceeding of 24thEuropean Photovoltaic Solar Energy Conference and Exhibition(EUPVSEC rsquo09) pp 1171ndash1174 Hamburg Germany 2009

[7] X Gu X Yu K Guo L Chen D Wang and D Yang ldquoSeed-assisted cast quasi-single crystalline silicon for photovoltaicapplication towards high efficiency and low cost silicon solarcellsrdquo Solar EnergyMaterials and Solar Cells vol 101 pp 95ndash1012012

[8] Y Tsuchiya H Kusunoki N Miyazaki et al ldquoCorrelationbetween carbon incorporation and defect formation in quasi-single crystalline siliconrdquo in Proceedings of the 38th IEEE

Photovoltaic Specialists Conference (PVSC rsquo12) pp 000297ndash000301 2012

[9] Q Yu L Liu W Ma G Zhong and X Huang ldquoLocal design ofthe hot-zone in an industrial seeded directional solidificationfurnace for quasi-single crystalline silicon ingotsrdquo Journal ofCrystal Growth vol 358 pp 5ndash11 2012

[10] W Ma G Zhong L Sun Q Yu X Huang and L LiuldquoInfluence of an insulation partition on a seeded directionalsolidification process for quasi-single crystalline silicon ingotfor high-efficiency solar cellsrdquo Solar Energy Materials and SolarCells vol 100 pp 231ndash238 2012

[11] A Black J Medina A Pineiro and E Dieguez ldquoOptimizingseeded casting of mono-like silicon crystals through numericalsimulationrdquo Journal of Crystal Growth vol 353 no 1 pp 12ndash162012

[12] D Hu T Zhang L He et al ldquoThe characteristics of subgrainsin the mono-like silicon crystals grown with directional solid-ification methodrdquo in Proceedings of the 38th IEEE PhotovoltaicSpecialists Conference (PVSC rsquo12) pp 002735ndash002738 AustinTex USA 2012

[13] B Tjahjono S Wang A Sugianto et al ldquoApplication of laserdoped contact structure on multicrystalline solar cellsrdquo in Pro-cessding of 23rd European Photovoltaic Solar Energy Conferencepp 1995ndash2000 Valencia Spain 2008

[14] K R McIntosh Lumps humps and bumps Three detrimentaleffects in the current-voltage curve of silicon solar cells [PhDthesis] Center of Excellence for Advanced Silicon Photovoltaicsand PhotonicsUniversity of New South Wales Sydney Aus-tralia 2001

[15] T Y Wang S L Hsu C C Fei K M Yei W C Hsu and CW Lan ldquoGrain control using spot cooling in multi-crystallinesilicon crystal growthrdquo Journal of Crystal Growth vol 311 no 2pp 263ndash267 2009

[16] CW LanWC Lan T F Lee et al ldquoGrain control in directionalsolidification of photovoltaic siliconrdquo Journal of Crystal Growthvol 360 pp 68ndash75 2012

[17] T F Li K M Yeh W C Hsu and C W Lan ldquoHigh-qualitymulti-crystalline silicon (mc-Si) grown by directional solidifi-cation using notched cruciblesrdquo Journal of Crystal Growth vol318 no 1 pp 219ndash223 2011


[18] H Sugimoto K Araki M Tajima et al ldquoPhotoluminescenceanalysis of intragrain defects in multicrystalline silicon wafersfor solar cellsrdquo Journal of Applied Physics vol 102 no 5 ArticleID 054506 2007

[19] W Dimassi L Derbali M Bouaıcha B Bessaıs and HEzzaouia ldquoTwo-dimensional LBIC and Internal-Quantum-Efficiency investigations of grooved grain boundaries in mul-ticrystalline silicon solar cellsrdquo Solar Energy vol 85 no 2 pp350ndash355 2011

[20] W S M Brooks S J C Irvine V Barrioz and A JClayton ldquoLaser beam induced current measurements ofCd1minus119909

Zn119909SCdTe solar cellsrdquo Solar Energy Materials and Solar

Cells vol 101 pp 26ndash31 2012[21] S Litvinenko L Ilchenko A Kaminski et al ldquoInvestigation of

the solar cell emitter quality by LBIC-like image techniquesrdquoMaterials Science and Engineering B vol 71 no 1-3 pp 238ndash243 2000







Editorial Board




Contents















































1 Introduction














2150 2300 2450

(kWhm2)

0 200

(km)



lt2000


8

6

4

2

0

10

08

06

04

02

00


Dai

ly ra

diat

ion

(kW

hm

2d

)

Clea

rnes

s ind

ex


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe






08

04

00

May

00 00

0 12 24 0 12 24 0 12 24

0 24 0 12 24 0 12 24

0 12 24 0 12 24 0 12 24

0 12 24 0 12 24 0 12 24

08

04

00

08

04

00

08

04

00

08

04

08

04

00

08

04

00

08

04

08

04

08

04

00

08

04

00

08

04


April June




25

2

15

1

05

0

times106

Load

(kW

)

Seasonal profile


Ann

ual

Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r







Load

Converter

AC DC

Battery

PV

AC generatorset 1

AC generatorset 2

AC generatorset 3

AC generatorset 4

AC generatorset 5

AC generatorset 6

AC generatorset 7


Efficiency curve40

30

20

10

0

0 20 40 60 80 100

Effici

ency

()

Output ()






Table 2 DG data


Table 3 PV data







6

5

4

3

2

1

0

times103

times103

0 20 40 60 80 100

14

12

10

8

6

4

2

0

Cycle

s to

failu

re


CyclesThroughput

Life

time t

hrou

ghpu

t (kW

















2 $900t for SO

2 $2600t for NO

119909 and

$2800t for PM [18 19]














2

15

1

05

0

times106

Pow

er (k

W)




Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r




4

3

2

1

0

times109

Net

pre

sent

cost

($)

Cash flow summary
















2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r


35

3

25

2

15

1

05

0

times109

Net

pre

sent

cost

($)



Cash flow summary

Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

PV

PV

Converter

Con

vert

er




2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r



3

2

1

0

times109

Net

pre

sent

cost

($)



Cash flow summary

Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

PV

PV

Converter

Con

vert

er

6

5

4




25

2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r


2

0

times109

Net

pre

sent

cost

($)

Cash flow summary



Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

PV

PV

Converter

Con

vert

er

8

6

4




25

2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r


2

0

times109

Net

pre

sent

cost

($)

Cash flow summary



Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

PV

PV

Converter

Con

vert

er

12

8

10

6

4





2emission









2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r


2

15

05

1

0

times109

Net

pre

sent

cost

($)

Cash flow summary



Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

S4KS

25P

PV

PV


Con

vert

er

35

3

25







2

15

1

05

0

times106

Pow

er (k

W)

PV


Janu

ary

Febr

uary

Mar

ch

April

May

June July

Augu

st

Sept

embe

r

Oct

ober

Nov

embe

r

Dec

embe

r




2emission of











3

2

1

0

times109

Net

pre

sent

cost

($)

Cash flow summary



Gen

1

Gen

2

Gen

3

Gen

4

Gen

5

Gen

6

Gen

7

S4KS

25P

PV

PV


Con

vert

er5

4

6









2emission from

8460421632 kgyr to 7176592896 kgyr

6 Conclusion


2gas emission






Acknowledgment


References





































1 Introduction





UOutUIn

DO

S

D1

D2

D3

L1

L2










50

40

30

20

10

02 04 06 08 10

d

UO

utU

In




UOutuSUIn COCI

IIn iID iOP IOut

iOCiIC iS

DO

S

D1

D2

D3

L1

L2


UOutUInSwitched-on

state

Switched-offstate

UOutUIn

S

S





i

t

IIn iIP

ton toff

i

t

i

t

iIC

TP 2TP 3TP

iL = iL1= iL2

IL AV ΔiL


02 04 06 08 10

d

12

10

08

06

04

02

PInP

InR



07

06

05

04

03

02

01

02

d

04 06 08 10


IICIL

AV


i

i

i

t

t

t

iOP IOut

ton toff iOC

1

TP 2TP 3TP

iL = iL1= iL2

IL AV ΔiL


07

06

05

04

03

02

01

02

d

04 06 08 10


IO

CIL

AV







119905on 119906119871= 119880In

119905off 119906119871= minus


(1)




119880Out =1 + 119889

1 minus 119889


119905off119879P (2)



Consider

119868119871AV =

119868In1 + 119889

(3)






1198711= 1198712=

119880In sdot 119905onΔ119894119871

(4)

1198711= 1198712=

119880Out sdot 119879PΔ119894119871

sdot

119889 sdot (1 minus 119889)

1 + 119889

(5)




1198711= 1198712=

119880Out sdot 119879PΔ119894119871 max



(6)


119862I =119894C sdot 119905onΔ119906C

119862I =119868119871AV sdot 119879PΔ119906Inmax

sdot 119889 sdot (1 minus 119889)

(7)






119871max Consider

119868IC

= 119868119871AV


Δ119894119871max119868119871AV

)

2

sdot

1198892sdot (1 minus 119889)

2sdot (1 + 3 sdot 119889)


(9)









sdot 119889


sdot 119889 sdot (1 minus 119889)

(10)



119862O =119868119871AV119877 sdot 119879P




= 119868119871AV


Δ119894119871max119868119871AV

)

2

sdot

1198892sdot (1 minus 119889)

3


(12)










uS2

iIP2

iOP2

iS1

UOutUIn

iIP iOP1iOP IOut

iOCiIC

IIn

Phase 1

Phase 2

iIP1

D11

D31

D21 uS1S1

DO1

CI CO

DO2

D12

D32iS2

S2D22

L11

L12

L22

L21


i

i

i

i

i

i

i

t

t

t

t

t

t

ton toff

ton toff iIP1

iIP

iIP2

IIn

iIC

TP2

TP 2TP 3TP

iL11= iL21

IL119899 AV ΔiL119899

IL119899 AV ΔiL119899iL12

= iL22


07

05

03

02

01

d

1 phase

2 phase

3 phase

02 04

04

06

06

08 10


IICIL1

AV


i

i

i

i

i

i

t

t

t

t

t

ton toff

ton toff iOP1

iOP2

iOP

iOC

IOut

TP2

TP 2TP 3TP

iL11= iL21

IL119899 AV

IL119899 AV

ΔiL119899

ΔiL119899iL12

= iL22




119871119899AV can


119868119871119899AV =

119868In119899 sdot (1 + 119889)

(13)



1198711119899= 1198712119899=

119880Out sdot 119879PΔ119894119871119899max



119899AV119877

(14)


Δ119894119871119899

=(119889) sdot (1 minus 119889)


sdot Δ119894119871119899max (15)



119862I =119879P sdot 119868119871

119899AV




119868ICΠ

=

1003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816

radic1198682


0

119899

) sdot (

1

119899

minus 119889) if 0119899

le 119889

le

1

119899

radic1198682


1

119899

) sdot (

2

119899

minus 119889) if 1119899

le 119889

le

2

119899

radic1198682


2

119899

) sdot (

3

119899

minus 119889) if 2119899

le 119889

le

3

119899

radic1198682


119899 minus 1

119899

) sdot (

119899

119899

minus 119889) if 119899 minus 1119899

le 119889

le

119899

119899

(17)








07

05

03

02

01

04

06

d

02 04 06 08 10

1 phase

2 phase

3 phase


IO

CIL1

AV


70

60

65

50

55

40

45

30

35

20

25

10

15

5

0

0 5 10 15 20 25 30 35 40 45 50



28


Signal 1 Signal 2

Time (ms)

10

9

8

7

6

5

4

3

2

1

0

40 40020 40040 40060 40080 40100 40120

40007m40007m 0

0

0

0

0

50m50m

3320

3320

5762

5762

10 10

10

10

10

3320

3320

4709

4709

165982m165980m

1736

1736

3012

3012

817335m817335m

1736

1736




Time (ms)

10

9

8

7

6

5

4

3

2

1

0

40 40020 40040 40060 40080 40100 40120


20

Ripple fac 904460m


Time (ms)

10

9

8

7

6

5

4

3

2

1

0

40 40020 40040 40060 40080 40100 40120


00 050m 6992 2496 2655 6992 2496 905309m 124809m 1064 2801 340948m 1064

15

Ripple fac


40 40020 40040 40060 0080 40100 40120

Time (ms)

10

9

8

7

6

5

4

3

2

1

0


0 050m 10296 4990 5080 10296 4990 949092m 249503m 1018 2063 186847m 1018

3

Ripple fac



3

2500

2

1500

1

500

0

40 40020 40040 40060 40080 40100 40120


3487

3487

1871

1871

1875

1875

3487

3487

1871

1871

135711m135711m

93529m93529m

1003

1003

1864

1864

72360m72360m

1003

1003

0

0

0

0

0

0

0

0 50m50m

iL11iL21

Ripple fac (1)

(2)

iL11

iL21


21 11989411987111

11989411987121

40 40020 40040 40060 40080 40100 40120

Time (ms)

3

2500

2

1500

1

500

0

iL12


40024m 1879 0 050m 3496 1872 1877 3496 1872 135792m 93608m 1003 1867 72343m 1003

Ripple fac


40 40020 40040 40060 40080 40100 40120

Time (ms)

10

9

8

7

6

5

4

3

2

1

0


00 50m minus157172120583 5081 1251 1535 5081 1251 890262m 62531m 1227 4063 579928m 1227

6

Ripple fac



10

9

8

7

6

5

4

3

2

1

0

40 40020 40040 40060 40080 40100 40120

Time (ms)

7


Ripple fac


6

5500

5

4500

4

3500

3

2500

2

1500

1

500

0

40 40020 40040 40060 40080 40100 40120


40007m 1828 0 50m 0 9931 2502 2661 9931 2502 905742m 125112m 1063 3969 340360m 1063

8

Ripple fac


3

2500

2

1500

1

500

0

40 40020 40040 40060 40080 40100 40120


40007m 2501 0 50m 0 3275 2490 2492 3275 2490 112534m 124483m 1 1315 45154m 1

iR1

Ripple fac



40 40020 40040 40060 40080 40100 40120

Time (ms)

2540

2530

2520

2510

2500

2490

2480

2470

2460

2450

2440


iR1

Ripple fac


40 41 42 43 44 45 46 47 48 49

130

120

110

100

90

80

70

60

50

40

30

20

10

0


40007m 0 50m 0 1 1315 45154m 1120029 157198 119504 119626 157198 119504 5402 5975

Ripple fac

V[ 9]


40 40020 40040 40060 40080 40100 40120

122

121500

121

120500

120

119500

119

118500

118


40007m 0 50m 0 1 1315 45154m 1120029 157198 119504 119626 157198 119504 5402 5975

V[ 9]

Ripple fac



119868ICΔ

=

10038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816

radicΔ1198942


2sdot (119889 minus 0119899)

3+ [0 minus (119899 + 0) sdot 119889]

2sdot (1119899 minus 119889)

3]


sdot (1 + 119889)2

if 0119899le 119889 le1

119899

radicΔ1198942


2sdot (119889 minus 1119899)

3+ [2 minus (119899 + 1) sdot 119889]

2sdot (2119899 minus 119889)

3]


sdot (1 + 119889)2

if 1119899le 119889 le2

119899

radicΔ1198942


2sdot (119889 minus 2119899)

3+ [4 minus (119899 + 2) sdot 119889]

2sdot (3119899 minus 119889)

3]


sdot (1 + 119889)2

if 2119899le 119889 le3

119899

radicΔ1198942


2sdot (119889 minus (119899 minus 1) 119899)


2sdot (119899119899 minus 119889)

3]


sdot (1 + 119889)2

if 119899 minus 1119899le 119889 le119899

119899

(18)





119868IC = radic1198682

ICΠ + 1198682

ICΔ (19)







119862O =119879P sdot 119868119871

119899AV


(20)


119868OCΠ

=

1003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816

radic1198682


0

119899

) sdot (

1

119899

minus 119889) if 0119899

le 119889 le

1

119899

radic1198682


1

119899

) sdot (

2

119899

minus 119889) if 1119899

le 119889 le

2

119899

radic1198682


2

119899

) sdot (

3

119899

minus 119889) if 2119899

le 119889 le

3

119899

radic1198682


119899 minus 1

119899

) sdot (

119899

119899

minus 119889) if 119899 minus 1119899

le 119889 le

119899

119899

(21)






119868OCΔ =

1003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816100381610038161003816

radicΔ1198942

119871119899max sdot

119899 sdot 1198892sdot [(119899 minus 1)

2sdot (119889 minus 0119899)

3+ (119899 minus 0)

2sdot (1119899 minus 119889)

3]


2

sdot (1 + 119889)2

if 0119899

le 119889 le

1

119899

radicΔ1198942

119871119899max sdot

119899 sdot 1198892sdot [(119899 minus 2)

2sdot (119889 minus 1119899)

3+ (119899 minus 1)

2sdot (2119899 minus 119889)

3]


2

sdot (1 + 119889)2

if 1119899

le 119889 le

2

119899

radicΔ1198942

119871119899max sdot

119899 sdot 1198892sdot [(119899 minus 3)

2sdot (119889 minus 2119899)

3+ (119899 minus 2)

2sdot (3119899 minus 119889)

3]


2

sdot (1 + 119889)2

if 2119899

le 119889 le

3

119899

radicΔ1198942

119871119899max sdot

119899 sdot 1198892sdot [(119899 minus 119899)

2sdot (119889 minus (119899 minus 1) 119899)

3+ [119899 minus (119899 minus 1)]

2sdot (119899119899 minus 119889)

3]


2

sdot (1 + 119889)2

if 119899 minus 1119899

le 119889 le

119899

119899

(22)


119868OC = radic1198682

OCΠ + 1198682

OC Δ (23)


119871119899max = 0 dotted


119871119899Nmax =







11= 11987121

= 11987112

= 11987122

= 82355 120583H 119862I =15625 120583F 119862O = 5208 120583F and 119891119904 = 50 kHz


11and 119871

21(11989411987111

11989411987121


22(11989411987112

11989411987122


4 Conclusion







Acknowledgments


References



























































1 Introduction






60

40

20

0

0 1 2 3 4

Maximum power point

Intensity = 9mA

Intensity = 2mA

Intensity = 38mA

ISC

(mA

)

VSC (V)

(a)

0 1 2 3 4

150

100

50

0

Maximum power point

Intensity = 38mA

Intensity = 9mA

Intensity = 2mA

VSC (V)

PSC

(mW

)(b)











Weak sunlight

UP

UP

UP

UP

Solarcell

VBAT = 42V

VSC = 181V

VBAT

VSC

(a)

Moderate sunlight

UP

UP

UP

UP

Solarcell


VBAT

0V

VSC

(b)

UP

UP

UP

UP

Solarcell

Strong sunlight

VSC

0V

VBAT = 42V

VSC = 30V

VBAT

(c)







0 10 20 30

00

02

04

06

08

IO

UT(m

A)

Frequency (MHz)

MPP

Weakintensity


Intensity = 25mA

(a)


Intensity = 15mA

5

4

3

2

1

0

Moderateintensity

0 10 20 30

Frequency (MHz)

MPP

IO

UT(m

A)(b)

0

5

10

15

20

25

Strong intensity


Intensity = 58mA

0 10 20 30

Frequency (MHz)

MPP

IO

UT(m

A)

(c)

0 20 40 600

10

20

30


fPU

MP

atM

PPs(

MH

z)

(d)






0 20 40 60

01

100

10

1



IO

UT(m

A)





119868IN =1

120578

119868OUT + 1205721119891PUMP (1)




119868SC = 119868IN + 119868DC + 1205722119891PUMP (2)













1+ 1205722)119891PUMP +





15

10

5

0

20 25 300 5 10 15

Strong sunlight

120578(1205721 + 1205722)

120578IDC

fPUMP (MHz)

120578(1205721 + 1205722)fPUMP + 120578IDC

120578I

SCminusI

OU

T(m

A)



1+





1and119872


2





4are 240 120583m and 035 120583m


2in









3and 119862

4can give UP


3and 119862

4give


3and 119862








UP

UP

UP

UP

UP

UP

UP

UP

UP

UP

UP

UP

UPSerial



architecture

VBATIOUT

BAT

Solar cell

VCO

k

C3

C1

C2

C4

05

+

+

minus

minus

+

minus

+minus

CP


VREF SM

VREF PM

M1 M2

M3M4

S and H

CLK

IN OUT

CLKBCAP1

CAP2

fSAMPLE

ISC IIN

fPUMPSWSM

SWSPM

SWPM

05 middot VSC

05 middot VSC

05 middot VOC

VSC



(a)

VCO

M1

M2

M3

M4

G1

G2

COMP2

COMP1

Vbp

Vbn

C1

Vc

VSC+

+

05 middot VOC

05 middot VSC

k

fPUMP

minus

minus



(b)

05VSC

05kVOC + Δ

05kVOC minus Δ

VCfPUMP for MPP

fPUMP

IOUT

(c)






100

50

0

Sunlight intensity

Average


Weak(15mA)

Moderate(9mA)

Strong(38mA)

IO

UT()

(a)

10

1

01

0 20 40 60


MPPVPF

IO

UT(m

A)(b)



2080 120583m

290120583m

VFB ARB





5 Conclusion





Acknowledgments


References
























Ga088


1 Introduction









119892(ev) [10] 342 293 07










119892(119909) [17 18] and


120572 [119864 (120582)] = 1205720radic

119864 (120582) minus 119864119892(119909)

119864119892(119909)


120594 = 41 + 07 times (34 minus 119864119892(119909))

120583119894(119873) = 120583min119894 +

120583max119894 minus 120583min119894

1 + (119873119873119892119894)

120574119894

(1)



119909Ga1minus119909


0for In

119909Ga1minus119909


p-Contact

p-GaN

n-GaN

i-InGaN

Sapphire

n-Contact







1875

1850

1825

1800

65 70 75 80 85 90 95 100 105 110 115 120 125


Jsc(m

Acm

2 )

(a)

2286

2284

2282

2280

2278

65 70 75 80 85 90 95 100 105 110 115 120 125


Voc

(V)

(b)

0902

0901

0900

0899

0898

65 70 75 80 85 90 95 100 105 110 115 120 125


FF

1 times 1016

1 times 1017

1 times 1018

(c)

1 times 1016

1 times 1017

1 times 1018

256

252

248

244

65 70 75 80 85 90 95 100 105 110 115 120 125


Effici

ency

(d)


Jsc(m

Acm

2 ) 35

30

25

20

15

10

0 200 400 600


(a)

Voc

(V)

222

220

218

216

0 200 400 600


(b)

090

088

086

084

082

080

0 200 400 600


FF

(c)

0 50

15

20

25

30

35

40

45

100 150 200 250 300 350 400 450 500 550 600


Effici

ency

(d)





030027

50

40

30

20

10

0

033 036 039 042 045 048

Qua

ntum

effici

ency

()



160

326

328

330

332

334

336

338

340

180 200 220 240 300280260 320

Shor

t circ

uit c

urre

nt d

ensit

yJ

sc(m

Acm

2 )


5grid fingers

4grid fingers

3grid fingers

(a)

160 180 200 220 240 300280260 320


59

60

64

63

62

61

5grid fingers

Effici

ency

()

4grid fingers

3grid fingers

(b)

p-Contact

n-Contact

Sapphire substrate

i-InGaN

n-GaN

p-GaN



(c)









119862 119881ic and





4 Conclusion


Ga088




Acknowledgments


References













Ga035



119909Ga1minus119909



119909Ga1minus119909






















1 Introduction





119874 the proposed






119898of





Charger

Power management

unit

Battery management

unitMPPT

unit

LED driving circuit

(discharger)LED

lighting

PWM ICunit


PV arrays VPV

VPV

IPV

VB

M1 M2

M1 M2IB

VB

VO

VO

minus

+

minus

+

minus

+


1 N


VPV

M1

minus

+ minus

+

minus

+

M2 M3

M4C1

C2

L1

L2D1 D2

D3

Lk

Lm

Tf

S1

CO VORL

VB







1 NVPV

M1

minus

+ minus

+

minus

+

M2

L2D1 D2

D3

Lk

Lm

Tf

S1

CO

CC

VORL

VB

IN2

IN1IDS1 IDS2ILm

ILk

IOIPV

IL2

IB





11198724) and (119872



21198723)


2and 119872

3can be merged




3are the same


and 119863119865232


119861231and 119863





119898








1and119872


119861141

and 119863119861142


and 119863119865142


1and 119862








119891and


1and119872






11and inductor






as



119904repre-



11986311(max) =

119881119861(max)

119881PV(min) + 119881119861(max) (2)


1 NVPV

M1 M23

M4

minus

+ minus

+

minus

+

L2

L1

D1

DF232

DB232

DF231

DB231 D2

D3

Lk

Lm

Tf

S1

CO

C2

C1

VORL

VB

(a)

1 N

M23

M4

L2D2

D3

Lk

Lm

Tf

S1

CO

C2

VPV

M1

minus

+

L1

D1

C1

minus

+

VORL

minus

+

VB

(b)

1 N

M23D3Lm

S1

CO

M4

C2

M1

VPVminus

+

L2D2TfD1

C1

minus

+

VORL

minus

+

VB

(c)

1 N

M23

DB141

DB142

DF142

DF141

D3Lm

S1

CO

C2

M14

VPVminus

+ minus

+

L2D2TfD1

C1

RL

minus

+

VB

VO

(d)

1 N

M23

D3Lm

S1

CO

C2

M14

VPVminus

+

L2D2TfD1

C1

RL

minus

+

VB

minus

+

VO

(e)

1 N

D3Lm

S1

CO

CC

M23

M14

VPVminus

+

L2D2TfD1

RL

minus

+

VB

minus

+

VO

(f)



0

0

0

0

0

0

0

0

0

0

0

0

0


VO

IO

VCC

IN2

IN1

ILm

ILk

VDS2

IDS2

VDS1

IDS1

M2

M1



follows

11987211(max) =

11986311(max)

1 minus 11986311(max) (3)





1



119871119898and



119861is expressed

by

119868119861(av) =(1 minus 119863

11)2

119881119861119879119904

2119871119898119861

(4)

where 119871119898119861



can obtain

t0

t0

t

t

0

0M1

IB

ILm

IB(a)

TS


ΔILm(max)ILm(0)








can bedetermined by

119868119861(av)max =

(1 minus 11986311119861)2

119881119861(max)119879119904

2119871119898119861

(5)

where 11986311119861




119871119898119861(max) =

(1 minus 11986311119861)2

119881119861(max)119879119904

119868119861(av)max

(6)





119891operated




119868119871119898(peak) = 119868119871119898(0) + Δ119868119871119898(max) (7)

where 119868119871119898(0)




be determined by

Δ119868119871119898(max) =

119881PV(max)11986311119877119879119904

119871119898

(8)


where 11986311119877




119868119861(av)max =

(1 minus 11986311119877)

2

(119868119871119898(peak) + 119868119871119898(0))

=

(1 minus 11986311119877)

2

(2119868119871119898(0)+

119881PV(max)11986311119877119879119904

119871119898

)

(9)



119868119871119898(0)=

119868119861(av)max

1 minus 11986311119877

minus

119881PV(max)11986311119877119879119904

2119871119898

(10)


119868119871119898(peak) =

119868119861(av)max

1 minus 11986311119877

+

119881PV(max)11986311119877119879119904

2119871119898

(11)




1198731= radic119871119898

119860119871

(12)


119898= 1198732

1119860119871 By


119861max =119871119898119868119871119898(peak) times 10

4

1198731119860119888

(13)








119861= 119881119861119868119861)

119875119861(max)



119881119874(max)


119868119874(max)




119874

119868119862


119861

119878119872



119878119899





1and119872

2







Photosensor

Operationalmode

judgment


Erroramplifer

Mode selector


A

B

(PWM IC)

Perturband

observemethod(MPPT

unit)

Batterymanagement

unit

Powermanagement

unit

Microchip

Controller

LOD2

D3

Lk

Lm

Tf

S1

Sn

SM

SM

SM

SP

CO

CCVPV

VPV

M1

M2

minus

+

D1

minus

+

VO

VO

VB(max)

VO(max)

VB(mix)

VB

VeVf

Vref

Vref

RL

IC

IPV

IB

IO

IO(max)

IB(max)

G1

G2

PB(max)

PB

PP

1 N











119899of





0

P

V

A area B area

Increase loadPmax1

Pmax2Pmax3


119875119899gt 119875119901 119875119899= 119875119901 and 119875



119901


119901is decided


119901 According


119899and 119875119901and 119881







119899= 119875119875and 119881

119899= 119881119875 working point of



0to 119879119888 When 119905 = 119879

0



119861(min) When119905 = 119879







119861(min) (=119881119861(min)119868119861(max))and 119875










119875= 0 and



119861(=119881119861119868119861)







119861)




command


119874ge


119875is set to 1 The 119878










level 119878119899= 1 The signal 119878



119872= 0


119899= 0 If 119878

119872= 1 the





Start

Calculate

Judge

Judge

Decreaseload

Increaseload

To powermanagement

unit

Replace


Read Vn In

Pn = VnlowastIn

Pn = PP Pn gt PP

Pn lt PP

Pn and PP

Vn and PP

JudgeVn and PP

JudgeVn and PP

Vn = PP

PP

Vn gt PP Vn lt PP

Vn lt PP


(ΔI)

Decreaseload(ΔI)


Increaseload(ΔI)

Increaseload(ΔI)

VP = Vn

PP = Pn

= PP = Pn

Set PPV(max)






119890



2 When 119866

1and 119866

2are


1= 1198661


119861of battery









2 Since 119878



to set 1198721= 1198662and 119872






119861(min) When119881119861le 119881119861(min)






1199041 250KHz



119861(max) 23A



tCharging time0

PmaxVB(min)

VB(max)

IB(max)

TOTC


0


Power curve

PB(max)

PB(min)

PPV(max)

tCharging timeTO TC

(a) 119875PV(max) lt 119875119861(min)

0


Power curve

A

PB(max)

PB(min)PPV(max)

tCharging time

TO TA TC


0


Power curve

tCharging time

PB(max)

PB(min)

PPV(max)

TO TC

(c) 119875PV(max) gt 119875119861(max)









2and 119871


119862can



119861= 7V


119898must


2=





119861= 7V



(i) switches11987211198722 PSMN005-75B

(ii) diodes11986311198632 UF601




Start

Read

Calculate

Shutdownpowersystem

Judge

Judgeovervoltage

overcharge

overvoltage

Judgeundervoltage

Judge

ToPWM

IC

No Yes

Yes

Yes

No

Yes

No



No

Shutdownpowersystem

Judge

Judgeovervoltage

overcharge

overvoltage

Judgeundervoltage

Judge

ToPWM

IC

No Yes

Yes

Yes

No

Yes

No

No

Judge

Set

No Yes

Calculate

ToPWM IC

Set SP = 0


IB(max) IO(max)


SetSP = 1

PP ge PB(max)

Pset = PP

SetPset = PB(max)


IC = ΔPVB

IO ge IO(max)

VO ge VO(max)

VB ge VB(max)

VB le VB(min)









1and current 119868








VDS

IB

LeCroy


(a)

VDS

IB

LeCroy


(b)





VPV

IPV

PPV

LeCroy


(a)

VPV

IPV

PPV

LeCroy


(b)



119861of 65 V



119863119878and current 119868





1and119872


VB

IB

LeCroy





6 Conclusion



VDS1

IDS1

ZVS


(a)

VDS2

IDS2

ZVS


(b)


and current 119868119863119878



ZVS


VDS1

IDS1

(a)

ZVS

VDS2

IDS2


(b)


and current 119868119863119878



6065707580859095

Effici

ent (

)

10 20 30 40 50 60 70 80 90 100Load ()




VO

IO







References

































1 Introduction






















M

P

Vent line

Fresh steam

Product in

Product out

1

2

3

4

5

6

78

9

S1

S2

Sunlight










119889V119901

119889119911

=

3

4

120585

120588119892(V119892minus V119901)

2

V119901119889119901120588119901

minus (1 minus

120588119892

120588119901

)

119892

V119901

minus

01

119863

(1)


PTC

ComputerData line

ND-filter

Sunlight

x

Lamberttarget

z

CCD camera

O

Sunlight



119889V119892

119889119911

= minus

119889119875

120588119892V119892119889119911

minus

119892

V119892

minus

3119873120585(V119892minus V119901)

2

4V119892119889119901

minus

120582V119892

2119863

(2)






119866119901119889119883 + 119866

119892119889119884 = 0 (3)

119866119875119889119883 +119882

119889119889119860 = 0 (4)


119882119889= 119870119910(119884eq minus 119884) (5)


119889119883

119889119911

= minus

6119870119910(119884eq minus 119884) (1 + 119883)

120588119901119889119901V119901

(6)



ℎ

119870119910

= 119862119892(1 + 119884) (

ScPr

)

056

Pr =(119862119892]119892120588119892)

120582119892

Sc =]119892

119863V119886

(7)


119889119884

119889119911

= minus

6119866119901119870119910(119884eq minus 119884) (1 + 119883)

119866119892120588119901119889119901V119901

(8)


119889119905119901

119889119911

= minus

119862119908119905119901

119862119904+ 119862119908119883

119889119883

119889119911

+


0+ 119862119908119905119901) ]

120588119901V119901119889119901(119862119904+ 119862119908119883)

(9)



119889119905119892

119889119911

= minus

1199030

119862119892

119889119884

119889119911

minus

6119866119901(1 + 119883) [ℎ (119905


0+ 119862119908119905119901)]

119866119892V119901119889119901119862119892

(10)


119889119863

119889119911

= 2 cot 120579 (11)









1320 1325 1330 1335 1340 1345 1350 135540

60

80

100

120

140

160

180

200

220

240


Time

Tem

pera

ture

(∘C)

400

500

600

700

800

900

1000

Dire

ct so

lar r

adia

tion

(Wm

2)

(a) 7 cm


1150 1210 1230 1250 1310 133080

100

120

140

160

180

200

220

240

400

500

600

700

800

900

1000

1100

1200

1300

Time

Tem

pera

ture

(∘C)

Dire

ct so

lar r

adia

tion

(Wm

2)

(b) 4 cm




119860119886119867119887

(12)


(a) 4 cm

(b) 7 cm


00 10 20 30 40 50 60 70

0

7

14

21

28

35

42

49

56

times103

Flux

(Wm

2)

x0 (cm)

x1 = 7 cm

x2 = 4 cm






119870119894= 119889119901times3

radic

4119892 (120588119901minus 120588119892)

31205832

119892120588119892

(13)





119870 =

Nu sdot 120582119892

119889119901

(14)


1 10 102

103

01

1

10

102

103

104

Rep

Ki

2345

2345

2345

23

23

45

02

03

04

2 3 4 5 6789 2 3 4 5 2 3 4 5


119894

0 50 100

2

5

8

Nu

Rep



120592119886=

Re119901120583119892

119889119901

(15)


119860119904=

6119866119901

119889119901120588119901

(16)


Δ119905 =

1199051+ 1199052

2

minus 119905119904 (17)


120591 =

119876

119870119860119904Δ119905

(18)


119911 = (V119892minus V119886) 120591 (19)

(a) 120579 = 50∘

(b) 120579 = 52∘



119863 = radic

4119866119892120588119904

3600120587120592119892

(20)








1200mm

250mm

300mm

2000mm

150mm

82∘

360mm

938mm

23∘

30∘

200

mm

45∘

60∘

50mm

200mm










Diameter(m)


Air 075 1025 15 0025 190 mdash mdash











ratioAirflow



∘C255 1467 2505 12times10-3 01 200 15 20

00 04 08 12 16 20 24 28 32

0

2

4

6

8

10

12

14

16



Moi

sture

cont

ent (

db

)

Velo

city

(ms

)12

14

16

18

20

22

24

26

(a)


00 04 08 12 16 20 24 28 32

12

14

16

18

20

22

24

26

30

60

90

120

150

180

210

Tem

pera

ture

(∘C)


oistu

re co

nten

t (

db

)

(b)


A1 B1 C1 C1998400 C2

112

116

120

124

128

132

136

140

144

148

Moi

sture

cont

ent (

)

Run number

MeasuredSimulated










119892in V119892out 119905

119901out 119883in 119883out

A1 1995 15 1634 532 2403 1237B1 1998 15 1569 587 2602 1415C1 2004 15 1577 593 2505 1339C11015840 2004 20 1526 612 2505 1123C2 1903 15 1485 528 2505 1433



119892in (∘C) 120578 ()


MeasuredSimulated


20

25

30

35

40

45

50

55

60

65

70

Tem

pera

ture

(∘C)

C1998400



5 Conclusion


100

110

120

130

140

150

160

170

180

MeasuredSimulated


Tem

pera

ture

(∘C)

C1998400





Nomenclature



















Greek






Subscript

In inletOut outlet



Acknowledgments


References






















1 Introduction





G

Iirr

I

V

RS

RSH

minus

+






119875parallel



119868 = 119873119875119868irr minus 1198731198751198680 [119890

119902(119881+119868(119873119878119873119875)119877119878)119873119878119899119896119879

minus 1]

minus

119881 + 119868 (119873119878119873119875) 119877119878

(119873119878119873119875) 119877SH

(1)








119899 = 119899ref (2)

119877119878= 119877119878ref (3)

119868irr =119866


1198680= 1198680ref[

119879

119879ref]

3

119890[119864119892ref119896119879refminus119864119892119896119879]

(5)

119877SH =

119866

119866ref119877SHref (6)

In (5) 119864119892


times (1198792(119879 + 636)) is









0

0002

0004

0006

0008

001

1

2

0

1

minus1

minus2

minus3

minus4

minus5

minus6

05

15nref

R Sre

f


2+ f2(nref RSref )

2]

(a)0

0001

0002

0003

0004

0005

0006

0007

0008

0009

001

1

2

05

15

nre

fRSref


(b)


119870119879= [119879119879ref]




1198621=

119896119879ref119902

1198622=

119881OCref

1198731198781198621

1198623=

119881mpref

1198731198781198621

1198624=

119868mpref

1198731198751198621

1198625=

119868SCref

1198731198751198621

(7)




11986211198625= 119868irrref minus 1198680ref (119890

1198625119877119878ref119899ref


(9)

11986211198624= 119868irrref minus 1198680ref (119890

(1198623+1198624119877119878ref)119899ref

minus 1)

minus 119866SHref1198621 (1198623 + 1198624119877119878ref) (10)

1198624

1198623

=

(1198680ref119899ref1198621) 119890

(1198623+1198624119877119878ref)119899119903119890119891

+ 119866SHref

1 + (1198680ref119877119878ref119899ref1198621) 119890

(1198623+1198624119877119878ref)119899ref + 119866SHref119877119878ref

(11)

0 = 119868irrref + 120572119879 (119879 minus 119879ref)

minus 1198680ref119870119879 (119890

119902(11987311987811986211198622+120573119879(119879minus119879ref))119873S119899ref119896119879

minus 1)


119873119878

(12)






0

0005

001

0015

002

1

2

0

1

minus1

minus2

minus3

minus4

minus5

05

15nref

R Sre

f


2+ f2(nref RSref )

2]

(a)0

0002

0004

0006

0008

001

0012

0014

0016

0018

1

2

05

15

nre

fRSref


(b)



119868irrref = 1198680 (1198901198622119899ref

minus 1) + 119866SHref11986211198622 (13)


11986211198624= 1198680ref (119890

1198622119899ref

minus 1) minus 1198680ref (119890

(1198623+1198624119877119878ref)119899ref

minus 1)



1198680ref =

11986211198624+ 119866SHref1198621 (1198623 minus 1198622 + 1198624119877119878ref)

(1198901198622119899ref minus 119890(1198623+1198624119877119878ref)119899ref)

(15)


1198680ref =


119890(1198623+1198624119877119878ref)119899ref ((119862

4119877119878ref minus 1198623) 119899ref1198621)

(16)


11986211198624+ 119866SHref1198621 (1198623 minus 1198622 + 1198624119877119878ref)

(1198901198622119899ref minus 119890(1198623+1198624119877119878ref)119899ref)

=


119890(1198623+1198624119877119878ref)119899ref ((119862

4119877119878ref minus 1198623) 119899ref1198621)

(17)


119878ref

119866SHref

=

1198624

1198624119877119878ref minus 1198623

sdot

(1 + (1198623minus 1198624119877119878ref) 119899ref) 119890

(1198623+1198624119877119878refminus1198622)119899ref

minus 1


(1198623+1198624119877119878refminus1198622)119899ref

(18)


1198680ref

=

11986211198624

1198623minus 1198624119877119878ref

sdot

(21198623minus 1198622)

1198901198622119899ref + ((119862


(1198623+1198624119877119878ref)119899ref

(19)


0

002

004

006

008

1

23

4

501

0

0 04

0

1

minus1

minus2

minus3

minus4

minus5

nref

R Sref


2+ f2(nref RSref )

2]

(a)

0 001 002 003 004 005 006 007 008 009

1

2

3

4

5

15

25

35

45

nre

f

RSref


(b)


4

5

6

7

8

9

10

0951

105

115

12512

11

0

1

minus1

minus2

minus3

minus4

minus5

minus6

nref

R Sref


2+ f2(nref RSref )

2]

times10minus3

(a)

12

11

4 45 55 655 6 7

095

1

105

115

125

nre

f

RSref


times10minus3

(b)




119868irrref =119862111986241198622

(1198623minus 1198624119877119878ref)

+

(11986211198624(21198623minus 1198622)) (119862

3minus 1198624119877119878ref) (119890

1198622119899ref

minus 1 minus (1198622119899ref) 119890

(1198623+1198624119877119878ref)119899ref

)

1198901198622119899ref + ((119862


(1198623+1198624119877119878ref)119899ref

(20)





1198623(21198624minus 1198625)

1198623minus 119877119878ref1198624

+

1198624(1198622minus 21198623)

1198623minus 119877119878ref1198624

119890((1198625119877119878refminus1198622)119899ref)

+ [

11986251198623minus 11986241198622

1198623minus 119877119878ref1198624

minus

11986221198624+ 11986251198623minus 11986251198622

119899ref]


= 0

[120572119879(119879 minus 119879ref) minus

120573119879(119879 minus 119879ref) 1198624

119873119878(1198623minus 119877119878ref1198624)

+

11986211198624(21198623minus 1198622)

(1198623minus 119877119878ref1198624)

]

+ [120572119879(119879 minus 119879ref) (

1198623+ 119877119878ref1198624 minus 1198622

119899refminus 1)]


+ [

1198624120573119879(119879 minus 119879ref)

119873119878

(119899ref minus 119877119878ref1198624 + 1198623)

119899ref (1198623 minus 119877119878ref1198624)]


+

11986211198624119870119879(1198622minus 21198623)

(1198623minus 119877119878ref1198624)


+

11986211198624(119870119879minus 1) (2119862

3minus 1198622)

(1198623minus 119877119878ref1198624)

119890(minus1198622119899ref)

= 0

(21)






119877119878ref =

(1198622minus 1198623)

1198624

0 lt 119877119878ref lt 119877

max119878ref (119899ref)

(22)


119877max119878ref (119899ref) =

119899ref1198624


)] +

1198623

1198624

(23)



05 ⩽ 119899ref ⩽ 25 (24)


119865 (119899ref 119877119878ref) = 1198911(119899ref 119877119878ref)2

+ 1198912(119899ref 119877119878ref)

2

(25)

where 1198911(119899ref 119877119878ref) = 0 and 119891

2(119899ref 119877119878ref) = 0



0 5 10 15 20 25 30 35 40

0

1

2

3

4

5

6

7

8

9

Voltage

Curr

ent










119877guess119878

= 09 sdot 119877max119878ref (119899

guess) (26)

18 20 22 24 26 28 30 32 34 36

6

65

55

75

85

7

8

Voltage

Curr

ent


5 par model



0 5 10 15 20 25 30 35 400

50

100

150

200

250

Pow

er

Voltage




with 119899guess




24 26 28 30 32 34

190

200

210

220

230

240

Voltage

Pow

er



5 par model








=

05119877max119878ref(119899

guess)



Levenberg-marquardt



=

09119877max119878ref(119899

guess)



Levenberg-marquardt



= 09119877max119878ref(119899

guess)




Levenberg-marquardt






119878ref = 0


5 Conclusions




119902 1602 times 10minus19 (C)

119896 13806503 times 10minus23 (JK)








minus1





1198621 119896119879ref119902

1198622 119881OCref1198731198781198621

1198623 119881mpref1198731198781198621

1198624 119868mpref1198731198751198621

1198625 119868SCref1198731198751198621



References

















































1 Introduction



















119868 = 119868119871minus 119868119863minus 119868sh (1)





119868119863= 119868119900[119890120585(119881+119868119877

119904)minus 1] (2)


119868sh =119881 + 119868119877

119904

119877sh (3)


119868 = 119868119871minus 119868119900[119890120585(119881+ 119868119877


119881 + 119868119877119904

119877sh (4)

V

I

IoIL

Rs

Rsh




120585 =

119902

120582119896119879119888

(5)



120582 = 119860 times 119873CS (6)




119868 = 119868119871minus 119868119900[119890120585(119881+119868119877

119904)minus 1] (7)














119868119871= (

119878119879

119878119879119903

) [119868119871119903

+ 120583119868sc(119879119888minus 119879119888119903)] (8)

where 119878119879and 119878


119888

and 119879119888119903





119888)




as

119868119900= 119863119879

3

119888119890120585120576119892119860 (9)



119868119900= (

119879119888

119879119888119903

)

3

119890120585120582120576119892(119879119888119903minus119879119888)119860

(10)


119868sc119903 = 119868119871119903

minus 119868119900119903

[119890(120585119903119868sc119903119877119904)

minus 1] (11)

where

120585119903=

119902

120582119896119879119888119903

(12)

119868119871119903

minus 119868119900119903

[119890(120585119903119881oc119903)

minus 1] = 0 (13)

119868mp119903 = 119868119871119903

minus 119868119900119903

[119890120585119903(119881mp119903+119868mp119903119877119904)

minus 1] (14)





119868119871119903

cong 119868sc119903 (15)

119868sc119903 minus 119868119900119903

[119890(120585119903119868sc119903119877119904)

] cong 0 (16)

119868mp119903 cong 119868119871119903

minus 119868119900119903

[119890120585119903(119881mp119903+119868mp119903119877119904)

] (17)



119868119900119903

= 119868sc119903 [119890minus(120585119903119881oc119903)

] (18)




] (19)


120585119903=

ln (1 minus 119868mp119903119868sc119903)

119881mp119903 minus 119881oc119903 + 119868mp119903119877119904 (20)


120582 =

119902 (119881mp119903 minus 119881oc119903 + 119868mp119903119877119904)

119896119879119888119903ln (1 minus 119868mp119903119868sc119903)

(21)


119904) is an


119904would be


119889119881

119889119868

10038161003816100381610038161003816100381610038161003816119881oc119903

minus 119877119904= 0 (22)


119868 = (119868119871+ 119868119900) minus

119882(120585119877119904119890120585(119881+119868

119871119877119904+119868119900119877119904))

120585119877119904

(23)


119871= 119868119871119903


119882(120585119877119904119868119900119890120585(119881oc119903+119868119871119903119877119904+119868119900119877119904)

)

[1 +119882(120585119877119904119868119900119890120585(119881oc119903+119868119871119903119877119904+119868119900119877119904))] 119877

119904

minus 119877119904= 0 (24)








119904is


119877119904

= 18Re(119879119888119882(minus15 times 10

71198900022(47+(17times10

5

119879119888)))

119879119888119882(minus15times 10

71198900022(47+(17times10

5119879119888)))+4400

)

(25)





119875 = 119868119881 (26)


119875 = (119868119871+ 119868119900) minus

119882[120585119877119904119890120585(119881+119868

119871119877119904+119868119900119877119904)]

120585119877119904

119881 (27)


119889119875

119889119881

10038161003816100381610038161003816100381610038161003816119881max

= 0 (28)


119882[120585119877119904119868119900119890120585(119881+119868

119871119877119904+119868119900119877119904)]119881

1 +119882[120585119877119904119868119900119890120585(119881+119868

119871119877119904+119868119900119877119904)] 119877119904

+

1

120585119877119904

times 119882[120585119877119904119868119900119890120585(119881+119868

119871119877119904+119868119900119877119904)] minus 120585119877

119904(119868119871+ 119868119900) = 0

(29)


119875max = (119868119871+ 119868119900) minus

119882[120585119877119904119890120585(119881max+119868119871119877119904+119868119900119877119904)

]

120585119877119904

119881max (30)

0 5 10 15 20 25 30 35 40

25

2

15

1

05

0

Curr

ent (

A)

Voltage (V)

Vmax

Imax

Isc

Voc


Optimum powerpoint



119868max =119875max119881max

(31)





0

0

5 10

10

15 20

20

25 30

30

35 40

40

50

60

70

80

Voltage (V)


Vmax

Optimum power

Pow

er (W

)

Pmax


0

0

1

2

3

4

5

5 10 15 20 25 30 35 4540

Voltage (V)


Curr

ent (

A)

GT = 1000 (Wm2)

GT = 800 (Wm2)

GT = 600 (Wm2)

GT = 400 (Wm2)

GT = 200 (Wm2)

Optimum powerpoint




0

0

1

2

3

4

5

5 10 15 20 25 30 35 40

Voltage (V)


Curr

ent (

A)

Optimum powerpoint

Ta = 0∘C

Ta = 25∘C

Ta = 50∘C

Ta = 75∘C


0

0

1

2

3

4

5

5 10 15 20 25 30 35 40

Voltage (V)


Curr

ent (

A)

Optimum powerpoint

GT = 500Wm2

Ta = 20∘C

Ta = 60∘C

GT = 1000Wm2





0 5 10 15 20 25 30 35 4540

Voltage (V)


Optimum power

GT = 1000 (Wm2)


180

160

140

120

100

80

60

40

20

0

Pow

er (W

)


0 5 10 15 20 25 30 35 4540

Voltage (V)


Optimum power180

160

140

120

100

80

60

40

20

0

Pow

er (W

) Ta = 0∘C

Ta = 25∘C

Ta = 50∘C

Ta = 75∘C


0 5 10 15 20 25 30 35 40

Voltage (V)


Optimum power160

140

120

100

80

60

40

20

0

Pow

er (W

)

Ta = 60∘C

Ta = 20∘C

GT = 1000Wm2

GT = 500Wm2



4 Conclusions


119904) of single




Nomenclature










119888120582









The Subscript




References


































































1 Introduction






























119881119862=

119860119896119879119862

119890

ln(119868ph + 1198680 minus 119868119862

1198680

) minus 119877119878119868119862 (1)


0is the



119862is the







119862119879119881= 1 + 120573

119879(119879119886minus 119879119909) (2)

119862119879119868= 1 +

120574119879

119878119862

(119879119909minus 119879119886) (3)

119862119878119881= 1 + 120573

119879120572119878(119878119909minus 119878119862) = 1 + 120573

119878(119878119909minus 119878119862) (4)

119862119878119868= 1 +

1

119878119862

(119878119909minus 119878119862) = 1 + 120574

119878(119878119909minus 119878119862) (5)

Δ119879119862= 120572119878(119878119909minus 119878119862) (6)

119881119862119883= 119862119879119881119862119878119881119881119862 (7)

119868ph119909 = 119862119879119868119862119878119868119868ph (8)




119862is the










and 119862119879119881


119878119868and119862

119878119881



119909 Temperature


119878 120573119878
























1

2

34

020406080

100120140

12

34

Ope

n ci

rcui

t vol

tage

(V)

NP

NS



5ndash100ndash5

1

2

34

0

5

10

15

20

25

12

34

Shor

t circ

uit c

urre

nt (A

)

NP

NS


1400ndash16000012 ndash1400



1

2

34

0200400600800

1000120014001600

12

34

Max

imum

pow

er (W

)

NP

NS




1

2

34

02468

1012141618

12

34

Max

imum

pow

erpo

int c

urre

nt (A

)

NP

NS




1

2

34

0102030405060708090

100

12

34

Max

imum

pow

erpo

int v

olta

ge (V

)

NP

NS






1000 2000 3000 4000 1000 1000 1000 2000119873119875


2000 2000 3000 3000 3000 4000 4000 4000119873119875



119873119904= 2 119879

119909

119873119901= 2 minus20 0 20 40

119878119909

80 316 298 280 261100 410 386 362 337120 512 481 451 418140 618 581 543 505160 733 688 642 595180 850 797 745 691200 977 915 853 790



119878and 119873

119875unchanged For


and119873119875= 2









11990410 Resistance 119877

11989815Ohm


10 Inductance 119871119898


119898008Vsdotsrad










119898300V



1199048 Frequency 119891

11989960Hz




119871200VAR





1198940 Inductance 119871 00002H






11989120 times 10


208V



119891and 119871







119891 119871119891 and 119862
























0 5 10 15 20 25 30 350

1

2

3

4

5

6

Voltage (V)

Curr

ent (

A)

Sa

Sb

Sc

Sd


0 5 10 15 20 25 30 350

20

40

60

80

100

120

Voltage (V)

Pow

er (W

)

Sa

Sb

Sc

Sd



0 5 10 15 20 25 30 350

1

2

3

4

5

Voltage (V)

Curr

ent (

A)

Ta

Tb

Tc

Td


0 5 10 15 20 25 30 350

20

40

60

80

100

Voltage (V)

Pow

er (W

)

Ta

Tb

Tc

Td








80

100120

140160

180200

0100200300400500600700800900

1000

020

40

Max

imum

pow

er (W

)

minus20


ndash700

500 600


Tx

S x


80

100120

140160

180200

02468

101214161820

020

40

Max

imum

pow

er p

oint

curr

ent (

A)

minus20


ndash14

10 12


Tx

S x


80

100120

140160

180200

0

10

20

30

40

50

60

020

40

Max

imum

pow

er p

oint

vol

tage

(V)

minus20



Tx

S x




5 Conclusion





0 1 2 3 4 50

50

DC

mot

or lo

ad v

olta

ge (V

)

100

150

200

250

300

Time (s)

VrefV



0 1 2 3 4 50

500

1000

1500

2000

2500

3000

Time (s)

Spee

d (r

ads

)


0 50 100 150 200 250 300 3500

10

20

30

40

50

Voltage (V)

Curr

ent (

A)


0 50 100 150 200 250 300 3500

2

4

6

8

10

Voltage (V)

Pow

er (W

)

times103



0 1 2 3 4 50

50

100

150

200

250

300

350

Time (s)

Volta

ge (V

)


0 1 2 3 4 5

0

2

4

6

8

10

Time (s)

Pow

er (W

)

times103

minus2


0 002 004 006 008 01

140

160

180

200

220

240

Time (s)

PV v

olta

ge (V

)


0 002 004 006 008 010

1

2

3

4

5

Time (s)

PV cu

rren

t (A

)


0 002 004 006 008 01

0

50

100

150

200

Time (s)

Vab

inve

rter

(V)

minus50

minus100

minus150

minus200


0 002 004 006 008 01

0

50

100

150

Time (s)

Vab

load

(V)

minus50

minus100

minus150



120 140 160 180 200 220 240 2600

1

2

3

4

5

PV voltage (V)

PV cu

rren

t (A

)


120 140 160 180 200 220 240 2600

100

200

300

400

500

600

700

800

PV voltage (V)

PV p

ower

(W)


83 8393

7583 83

90 88

60

88

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10

Succ

ess (

)

Group

83 83757

83 8390 88

60

88


254

380

290230

160

0

1

2

3

4

5

1 2 3 4 5

Aver

age

Question

254290

2301601 60







172

370

200 192 182 204164

212 224 244206

170236

326

168 182 180

274

0

1

2

3

4

5

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Aver

age

Question

7272 200 192 182182 2041641 64

212 224 244206

17070236

326

1686 182182 18080

272


142 162 178140

388

192126 140 136

466

0

1

2

3

4

5

24 25 26 27 28 29 30 31 32 33

Aver

age

Question

142 1626 1781 78140

388

192126 140 136




Appendix


119879119862(∘C) = 20119878119862() = 100120573119879= 0004

120574119879= 006

119879119886(∘C) = 20120572119878= 02

119860 = 62



1198680(A) = 001119877119878(Ω) = 002

Nomenclature






voltage








119881119862119883






References




































1 Introduction








Item Details

Module





System





119886)




0 200 400 600 800 1000 1200 14000

500

1000

1500

2000

2500

PV o

utpu

t pow

er (W

)


∘C40

∘C50

∘C60

∘C


0 500 1000 1500 2000 25000

500

1000

1500

2000

2500

Inve

rter

out

put p

ower

(W)


R2= 1


Slope = 09942




25 30 35 40 4590

92

94

96

98

Inve

rter

effici

ency

()



00 01 02 03 04 0530

40

50

60

70

80

90

100


Inve

rter

effici

ency

()

Ta lt 37∘C

Ta gt 37∘C


119886lt 37∘C and 119879

119886gt 37∘C






119886gt 37∘C


119886lt 37∘C










119886(∘C)


Oct

10

Nov

10

Dec

10

Jan

11

Feb

11

Mar

11

Apr

11

0

1

2

3

4

5

6

7

8

Month

Sola

r ene

rgy

rece

ieve

d (M

Wh)

0

5

10

15

20

25

30

35

40

Aver

age a

mbi

ent t

empe

ratu

re (∘

C)

Ta

Er









60

65

70

75

80

85

90

95

100

Perfo

rman

ce ra

tio o

r effi

cien

cy (

)


PR of system

Oct

10

Nov

10

Dec

10

Jan

11

Feb

11

Mar

11

Apr

11

Month





119870119879= 1 + 120572 sdot (119879

119898minus 25) (2)


119870ivt = 120578∘


ivt (3)





119879119870ivt 119870

119890



lowast


119904and

120578∘








119890in December



4 Conclusion





References


























1 Introduction











Rp

Rs

ID

DIsc

IPV

VPV






119868sc = 119868119863

+ 119868PV + (

119881119863

119877119901

)

119881PV = 119881119863

minus (119868PV lowast 119877119904)

(1)


= 119868119900

+ (119890(119881119863119881119879)minus 1)



1000

Insolation

InsolationProduct



PV1

PV module (I)

IPV

IPV VPV

VPV

IPV ramp

PPV

PPV

PPV






119881OUT119881IN

=

119863

(1 minus 119863)

119868IN119868OUT

=

119863

(1 minus 119863)

(2)


119863 =

119881OUT + 119881119863

119881IN + 119881OUT + 119881119863

(3)



119898) 164

Current at MPP (119868119898) 225


3

25

2

15

1

05

0

0 5 10 15 20 25

Am

ps (A

)

Voltage (V)

MPP




119863max =

119881OUT + 119881119863

119881IN(MIN) + 119881OUT + 119881119863

(4)


1198711

= 1198712

= 119871 =


Δ119868119871

lowast 119891119878

(5)




1



119881OUT + 119881119863

119881IN(MIN) (6)



Δ1198811198621

=


1198621

lowast 119891119878

(7)


1198681198761(Peak) = 119868

1198711(PEAK) + 119868

1198712(PEAK) (8)



119881OUT119881IN(MIN)2

(9)


119866is the gate




(119876GD lowast 119891119878)

119868119866

(10)


1198761(Peak) The


119881RD = 119881IN(MAX) + 119881OUT(MAX) (11)


119878and 119877



119900of

the solar PV module

120588119894= 120588119900

=

1198812

119878

119877119894

(12)


119878and 119877

119894can be

shown below

120597120588119894=

2119881119878

119877119894

120597119881119878

minus

1198812

119878

1198772

119894

120597119877119894 (13)


119894=

119903119892

120597120588119894= 0 hence

120597119881119878

120597119877119894

=

119881119878

2119877119894

(14)



Δor




1015840


1198791015840

Δ=

119883119894

Mean (119909)

lowast 119879Δ (15)


119894 119894 = 1 2 119873 a chaotic

sequence 119909 = (1199091 1199092

119909119873


Mean (119909) = Lim119873

sum

119894=1

1003816100381610038161003816119883119894

1003816100381610038161003816

1

119873

119873 997888rarr infin (16)


1015840


1198601015840

Δ= 1 + 119870

119883119894

Mean (119909)

119860Δ (17)


119894 119894 = 1 2 119873 a


119909119873


1198891sdot sdot sdot 1198771198896

)


1198891=

24 kΩ 1198771198892

= 33 kΩ 1198771198893

= 1198771198894

= 220 Ω and 1198771198895

= 1198771198896

=


1= 22mH 119862

1= 47 nF 119862

2= 500 pF and

119877 = 175KΩ


Reset

Relational

Randomnumber

Discrete-time

Chaotic PWM

Chaos carrier

integrator1

A1

1z

operator1

Unit delay1

03

ge

ge

Constant2

K Ts

z minus 1





1 1198712 capacitor 119862

1 1198622 a switch




119904 119862119903 119862119900 119881119892



1198911 1198631198912



119871




= 1198782

=

100KHz and 1198783

= 200KHzWhen 119878

1= 1198782

= 0 and 1198783

= 1 as in Figure 8

119889119894119871119865

119889119905

=

1

119871119865

[119881119892

= 119881119900]

119889119881119900

119889119905

=

1

119862119900

[119894119871119865

minus

119881119900

119877119871

minus 119894119871119878]

(18)


= 1198782

= 1198783

= 1 as in Figure 9

119889119894119871119865

119889119905

=

1

119871119865

[119881119892

minus 119881119900]

119889119881119900

119889119905

=

1

119862119900

[119894119871119865

minus

119881119900

119877119871

]

(19)


= 1198782

= 1 and 1198783


119889119894119871119865

119889119905

=

119881119892

119871119865

119889119881119900

119889119905

=

1

119862119900

[minus

119881119900

119877119871

]

(20)


Vg

minus

+

Lf1

Lf2

Co RLLsVo

iLs

iLf


= 1198782

= 0 and 1198783

= 1

Vg

minus

+

Lf1

Lf2

CoRL Vo

iLf


= 1198782

= 1198783

= 1


1199091

=


119871119865

1199092

minus

(1 minus 1198892)

119871119865

1199092

+

119881119892

119871119865

1199092

=

minus1

119877119871119862119900

1199092

+

(1 minus 1198891) 1198892

119862119900

1199091

+

(1 minus 1198891) (1 minus 119889

2)

119862119900

1199091

(21)

where 1199091and 119909


119871119865and 119881

119900

respectively



1 1198782 and 119878

3are used 119878






1and 119878

2) operate in






turned off with ZCS

Vg

minus

+

Lf1

Lf2

CoRL Vo

iLf


= 1198782

= 1 and 1198783

= 1 or 0



1and 1198782


3


1198911and 119863

1198912













Active snubber


Grid


Main switches

Snubber circuit

Auxiliary

Input source

(Lf1 Lf2)


(S1 S2)

(L1235)

switch (S3)





1198912750 120583H



1198772(119871119898

+ 119871119889) 2 120583H


+ 1198710119897) 3 120583H



11987747 nF



119871530Ω









1and choke 119871






1198772value




1198771ge 21198711198772 To turn off 119878





1198911198781



Lf1

Lf2

S1 S2

S3

D2

Ld

Lm

LR1

D1

Ln

RLoad

Co

Lo1 D4

Df1

Df2

D3

CR

PVArray





119878is ignored 119871


with 1198711198772


1198911198783



1198911and 119871


1198631198911-1198631198912







1 1198622 and 119862

3 as shown



1198810

1198781

1198782

1198783

1198784

1198785

1198786









Discrete

Powergui

0

Volt

Power

Current

Duty

Boost

vs1 is4

Mean

Mean

Mean value1

solar

Mean value2

Output1Output2

Current measurement

Output4

i

minus

+

minus

+

+

+

+

+ +

+

+

minus+

s

Display

Lf1

Lf

IGBTdiode

g

g

c

c

E

E

L

Pulse generator2

Mosfet

Display1

Transfer Fcn1

isolation

cycle

pwm

panel

Pulse

Pulse

generator3

generator1

IGBTdiode 1R

minus+

minus

+

Multilevel inverter

Subsystem1

s = 24e minus 05s

C1

C0

LO1

sg D

1

den(s)

PV current3


+

1

2

Out1

Out1

Out1 Out1

Step1Gain

1

1

V

I P Q

Mag V I

i

Total powerf(u)s

Dg

sDg

sDg

sDg

sDg

2

minus

minus

+

+

+

+

+

+

+ +

+

sDg

g

minus

iminus+

C1

C2

C3

P5

P6

D1

D3

D5

D7

M6

M5

D2

D4

D6

D8

M1

M3

P3

P1

P4

P2

M2

M4

+








8 Conclusion



MPPT technique


Controller

multilevel inverter



50

40

30

20

10

0

minus10

minus20

minus30

minus40

Curr

ent

Time0 05 1 15

Time offset 0


300

200

100

0

Volta

ge

minus100

minus200

minus300

Time0 05 1 15

Time offset 0


475

470

465

460

455

450

445

440

0 002 004 006 008 01 012 014 016 018 02

Time offset 0


800

600

400

200

0

minus200

minus400

minus600

minus800

0 01 102 03 04 05 06 07 08 09

Time offset 0


550

500

450

400

350

300

250

200

150

100

50

0

0 02 04 06 08 1 12 14 16

times10minus3

Time offset 0


2

600

500

400

300

200

100

0

0 02 04 06 08 1 12 14 16

times10minus3

Time offset 0






References











































1 Introduction






DCDC

DCDC

Vrs

Vbat

Vo

+

ndash

(a)

DCDC

Vrs

Vbat

Vo

+

ndash

(b)


Vrs

VbatVo

+

ndash


Vrs

Vbat

Vo

+

ndash






Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2









S2

S1

S3

(Vrs minus Vo)L1

t0 t1

VrsL1

IL1

IL2

(a) Waveforms

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2








1is used to



1



(1) During 1199050 1198781is off and 119878







2


1




1and 119878

3are turned


1and 119871




S2

S1

S3

(Vrs minus Vo)L1

t0 t1 t2

VrsL1

minusVbat L2

(Vrs minus Vbat )L2

IL1

IL2

(a) Waveforms

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2




1may be turned off




the diode1198631


1and 119878

2are


1and 119871

2is






Renewablesource

Battery

A

B

CPrs (W)

Vrs (V)




S2

S1

S3

(Vrs minus Vo)L1

t0

t1

t2

VrsL1


(Vrs minus Vbat )L2

IL1

IL2

(a) Waveforms

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2







1 acting



3















2


1is delivered





1 is



(1) During 1199050 1198781is off but 119878



1and119871




Vbat (L1 + L2)


S2

S1

S3

t0 t1

IL1

IL2

IL1= minusIL2

(a) Waveforms

Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2


Vrs Vbat

Vo

+

ndashD1

D0

S2

S1

S3

L1

L2 C

Drs

IL1

IL2





2


1and 119871

2is












1and 119878

2are


1198891198941198711

119889119905

=

119881rs1198711

minus

119881119900

1198711

(1 minus 1198891198782

)

1198891198941198712

119889119905

=

119881rs1198712

1198891198781

minus

119881bat1198712




1198782

1198783







Mode 1119881119900=

1

1 minus 1198891198782


119881bat = 1198891198781


Mode 2

119881119900=

1

1 minus 1198891198782

minus 1198891198783



)119881119900

119881119900=

1

1 minus 1198891198782


Mode 3 119881bat = 1198891198781


119889119881119900

119889119905

=

1198941198711

119862

(1 minus 1198891198782

) minus

119881119900

119877119862

(1)



1198711


1198712




1198781


1 and 119889

1198782






1198891198941198711

119889119905

=

119881rs1198711

minus

119881119900

1198711

(2 minus (1198891198782

+ 1198891198783

))

1198891198941198712

119889119905

=

119881rs minus 119881bat1198712

minus

119881119900

1198712

(1 minus 1198891198783

)

119889119881119900

119889119905

=

1198941198711

119862

(1 minus 1198891198782

1198891198783

) 1198891198782

(1 minus 1198891198783

) minus

119881119900

119877119862

(2)

Where 1198891198783



1198891198941198711

119889119905

=

119881bat1198711+ 1198712

minus

119881119900

1198711+ 1198712

(1 minus 1198891198782

)

1198891198941198712

119889119905

= minus

1198891198941198711

119889119905

119889119881119900

119889119905

=

1198941198711

119862

(1 minus 1198891198782

) minus

119881119900

119877119862

(3)


= minus1198941198712



1198891198941198712

119889119905

=

119881rs1198712

1198891198781

minus

119881bat1198712

(4)


2 hence


















1198711=

119881rs1198891198782

119879119904

Δ1198681198711

1198712=

(119881rs minus 119881bat) 1198891198781

119879119904

Δ1198681198712

119862 =

119881119900(1 minus 119889

1198782

) 119879119904

Δ119881119862119877

(5)


1198711

is the ripple of 1198711Δ1198681198712







PI compensator

MPPT

Modeselection

block

Carrier

S2S1 S3

Vor

Vo

+


Vrs

Vbat Min PPTdS2

dS1

1 minus dS3





1is used


2







1198782


2


1199061198782

= 119896119901119890119900+ 119896119894int 119890119900119889119905 = 0 (6)


119900= 119881119900minus 119881or) and 119881or is







3is changed At the



1198782




Start

NoYes

No

Yes No

Yes

Yes

Yes

No

NoIL2

(k) lt 0

IL2(k) lt 0

IL2(k) lt 0

IL2(k) gt IL2

(k minus 1)

IL2(k) gt IL2

(k minus 1)

Sense IL2

dS3larr 1

dS3(k) larr 1

Decrease dS3

Increase dS3







1and 119871

2are

700 120583H and 119862119900is 100120583F


S1

S2

S3

1 2

2

100A 500mA100 120583s 100 MSs

10k points


1 1198782 1198783 1198681198711

and 1198681198712


S1

S2

S3

1 2

2

100A 500mA100 120583s 100 MSs

10k points


1 1198782 1198783 1198681198711

and 1198681198712










2

S1

S2

S3

1 2

2

100A 500mA100 120583s 100 MSs

10k points


1 1198782 1198783 1198681198711

and 1198681198712


S1

S2

S3

1 2

2

100A 500mA100 120583s 100 MSs

10k points


1 1198782 1198783 1198681198711

and1198681198712








120578 =

sum119899119875onsum119898119875im (7)




2

3

4Ch1 Ch2Ch3 Ch4

500V200V 200V

500mA P 200 s


119900 minus1198681198712

1198782 and 119878

1





2) the



3






2

3

4Ch1 Ch2Ch3 Ch4

500V200V 200V

500mA P 200 s


119900 minus1198681198712

1198782 and 119878

1

2

3

4Ch1 Ch2Ch3 Ch4

500V200V 200V

500mA P 200 s


119900 minus1198681198712

1198782 and 119878

1



119900= 100W) the




2

3

4Ch1 Ch2Ch3 Ch4

500V200V 200V

500mA P 200 s


119900 minus1198681198712

1198782 and 119878

1

2

3

4

Ch2Ch3 Ch4

P

200V 100Vsim

250V A Ch2 150V200ms


2 and 119881

119900(AC mode)



5 Conclusion







References






















1 Introduction












PV array

or panel


converterLoad



Vin

Iin

+

minus

+

minus

Io

Vo


D

G

Radiation

ID

IR

Rsh

Rs

IoVo



119868PV = 119896 sdot 119866 sdot 119878 (1)


by

119868119889= 119868119900sdot (119890119860119881119889

minus 1) (2)



119868sc (119866 119879) = 119868sc (STC) sdot

119866

1000

sdot (1 + 120572119868scΔ119879) (3)


119875119898(119866 119879) = 119875

119898(STC) sdot

119866

1000

sdot (1 + 120574120588Δ119879) (5)

120578 =

119875119898

119866119860

= (119875119898(STC) sdot

(1 + 120574120588Δ119879)

119860

) (6)

where Δ119879 = 119879119888 minus 25∘C




119889120572119891 (119905)

119889119905120572

= limℎrarr0

1

ℎ120572

120572

sum

119903=0

(minus1)119903(

120572

119903

)119891 (119905 minus 119903ℎ)

120573 = (

120572

119903

) =

120572

119903 (120572 minus 119903)

(7)


119863120572

119905119891 (119905) asymp ℎ

minus120572

[119905ℎ]

sum

119903=0

(minus1)119903120573119891 (119905 minus 119903ℎ) (8)


119863120572

119905119891 (119905) =

119889120574

119889119905120574[119863minus(120574minus120572)

119905]

119863120572

119905119891 (119905) asymp (

119905

119899

)


sum

119903=0

Γ (119903 minus 120572)

Γ (minus120572) Γ (119903 + 1)

119891 (119905 minus 119903

119905

119899

)

(9)


119898(119905) = 119905

119898119898 =


119863120572

119905119905119898

asymp

119905119898minus120572

Γ (minus120572)

119899120572

119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (119903 + 1)

(1 minus

119903

119899

)

119898

(10)


(10) becomes

119863120572

119905119905119898

asymp

119905119898minus120572

Γ (minus120572)

119898

sum

119896=0

(minus1)119896(

119898

119896

) 119899120572minus119896

119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (119903 + 1)

119903119896 (11)

119870 equiv

119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (119903 + 1)

119903119896 (12)


119910(119895)


Γ (119910 + 1) = 119910(119895)

Γ (119910 minus 119895 + 1)

(13)


119910119896=

119896

sum

119895=1

120585119896

119895119910(119895)

=

119896

sum

119895=1

120585119896

119895

Γ (119910 minus 119895 + 1)

Γ (119910 + 1)

(14)


119870 =

119896

sum

119895=1

120585119896

119895(

119899minus1

sum

119903=0

Γ (119903 minus 120572)

Γ (119903 + 1 minus 119895)

) =

119896

sum

119895=1

120585119896

119895

Γ (119899 minus 120572)

Γ (119899 minus 119895)

(

1

119895 minus 120572

)

(15)


119863120572

119905119905119898

asymp

119905119898minus120572

Γ (minus120572)

119898

sum

119896=0

(minus1)119896(

119898

119896

)

119896

sum

119895=0

120585119896

119895119899120572minus119896

Γ (119899 minus 120572)

(119895 minus 120572) Γ (119899 minus 119895)

lim119899rarrinfin

119899120572minus119896 Γ (119899 minus 120572)

Γ (119899 minus 119895)

=

1 if 119895 = 119896

0 if 119895 lt 119896

(16)

where119898

sum

119896=0

(minus1)119896(

119898

119896

)

1

(119896 minus 120572)

= 119861 (minus120572119898 + 1) (17)


119863120572

119905119905119898

asymp

Γ (119898 + 1)

Γ (119898 + 1 minus 120572)

119905119898minus120572

(18)




Sample


120572= (Vz minus

d120572I = ΔI = Iz

dP120572

dP120572

=

S = Mtimes

d120572I = 0

dV120572gt 0

V(z) = V(zminus1)

V(z) = V(zminus1)

Eq (25) = Eq (26)




dV120572gt 0

andd120572I gt 0

dV120572lt 0 and

d120572I lt 0

End

Iz rarr Izminus1

Vz rarr Vzminus1


YY

Y

Y

Y

YY

N

N

N

N

N



dV120572= 0

Vzminus1)120572

V(z) I(z)

ΔV120572times ΔI

Po lt P





(119911)and 119868(119911)



10038161003816100381610038161003816100381610038161003816

119889119875

119889119881

10038161003816100381610038161003816100381610038161003816

(19)



1003816100381610038161003816100381610038161003816

Δ119875

Δ119881

1003816100381610038161003816100381610038161003816

(20)



10038161003816100381610038161003816100381610038161003816

119875 (119911) minus 119875 (119911 minus 1)

119881 (119911) minus 119881 (119911 minus 1)

10038161003816100381610038161003816100381610038161003816

(21)


119872 times

10038161003816100381610038161003816100381610038161003816

119889119875

119889119881

10038161003816100381610038161003816100381610038161003816fized step=Δ119863max

lt Δ119863max (22)



119872 lt


(23)





119889120572119881 (119868)

119889120572119868

= limΔ119881rarr0

119881120572(119868) minus 119881

120572

119900(119868 minus Δ119868)

Δ119868

(24)

119889119881120572

119889120572119868

asymp

(119881 minus 119881119900)120572

119868 minus 120572119868119900

(25)



119889120572

119889119881120572(minus

119881119900

119868119900

)

= (minus

1

119868119900

)

119889120572119881120572

119900

119889120572119868

+ (minus119881119900)

119889120572119868minus1

119874

119889120572119868

= (minus

1

119868119900

)(

Γ (2)

Γ (2 minus 120572)

)119881119900

1minus120572+ (minus119881

119900)

Γ (0)

Γ (minus120572)

1198681minus120572

119874

(26)







and 119881119911minus1


119911minus 120572119868119911minus1

) =

Δ119868 and 119889119881120572


)120572



+

+minus

+

minus

minus

R

Io

VL2+

minus

+

minus

L2

S2

S1

Vd

Vd L1

L c

VC2

VC1












119911minus1= 119881119911and 119868119911minus1

= 119868119911





+

minus

+minus

R

Vd

VoC

C1

Mode-1

L1

+

minus

VL2

L2

(a)

+

minus

+minus

Vd

VoC

C1

Mode-2

L1

+

minus

VL2

L2

(b)


Continuous

Pv

+minus

+minus

+

+

minus

minus

+

minus

im gS D

mg

SD

IMean

times

NOT

Mosfet1

Mosfet2

Batt lSOC

Battery

Ramp2

Relationaloperator2

Transportdelay

Transportdelay1

Clock

D initSwitch


fcn

i

d

inew

newdnew

Powergui

0

2

C4C2

C5 L1

L2

C6

C1

Transport delay2




1and 119871





1is charged by



2 The



1198711flows


1The inductor










08

07

06

05

04

03

02

01

0

Curr

ent (

A)



9

8

7

6

5

4

3

2

1

0

Pow

er (W

)















Technique Parameter

Irradiance-200Wm2





119900 129W 002 seconds 119875



119872 = 0056 119875119900 135W 002 seconds 119875


fluctuation

10

20

30

40

030 040 050 060

Times (s)

Out

put p

ower

(W)

850Wm2 27∘C

200Wm2 25∘C

850Wm2 27∘C

200Wm2 25∘C

(a)

10

20

30

40

030 040 050 060

Times (s)

Out

put p

ower

(W)

850Wm2 27∘C

200Wm2 25∘C

850Wm2 27∘C

200Wm2 25∘C

(b)


(a) (b)



1


1and 119871

2are








M 100msMATH 100 vv

M


Tek Stop

MATH 100 vv

M

M Pos

M 100 s

minus3440 s


4 Conclusion



Nomenclature




References
































1 Introduction




119881MPPest = 119896119881OC (1)



119881MPPest = 119886119881OC + 119887 (2)





2 Solar Cell Model



119868 = 119868SC minus 119868O [119890((119902(119881+119877s119868))119899d119870119879cell)

minus 1] minus

119881 + 119877s119868

119877p (3)



119868O =119868SC minus (119881OC119877p)

[119890(119902119881OC119899d119870119879cell) minus 1]

(4)


119868SC (119879cell 119866) =119866

1000

[119868SCr+ 120572 (119879cell minus 119879r)] (5)


times [1 + 120588OC ln(119866

119866OC) ln( 119866

119866r)]

(6)





minus

+

I

V

Rs

RpIPH




800Wm2119866 (7)



3 Computed Results


= 35mAcm2119881OCr








G(Wm2)


VMPP(V)

PMPP(mW)

VOC(V)

VMPP(V)

PMPP(mW)

VOC(V)

VMPP(V)

PMPP(mW)

VOC(V)

VMPP(V)

PMPP(mW)

VOC(V)

VMPP(V)

PMPP(mW)

237 0262 0194 0135 0262 0207 0153 0262 0185 0121 0262 0193 0134316 0311 0237 0227 0311 0251 0254 0311 0226 0205 0311 0235 0225422 0357 0277 0362 0357 0293 0401 0357 0264 0330 0357 0275 0359 0241 0126 0095562 0398 0313 0553 0398 0331 0609 0398 0300 0507 0398 0311 0548 0306 0168 016875 0434 0347 0826 0434 0365 0904 0434 0332 0761 0434 0343 0816 0369 0221 0299100 0467 0376 12 0467 0396 1312 0467 0361 1113 0467 0372 1186 042 0282 05241334 0495 0402 1725 0495 0422 1874 0495 0385 1600 0495 0395 1693 0462 0337 08991778 0518 0423 2431 0518 0444 2635 0518 0406 2259 0518 0415 2374 0494 0379 14772371 0537 0440 338 0537 0461 3657 0537 0422 3146 0537 0429 3279 0519 0409 23213162 055 0452 464 055 0474 5013 055 0434 4321 055 0437 4458 0537 043 35004217 0558 0458 628 0558 0481 6786 0558 0441 5853 0558 0439 5961 0548 0443 50945623 056 0459 8386 056 0482 9067 056 0441 7812 056 0433 7821 0553 0448 71857499 0555 0454 11031 0555 0477 11943 0555 0436 10264 0555 042 10035 055 0446 98501000 0543 0442 14254 0543 0464 15472 0543 0424 13234 0543 0398 12511 054 0436 1313

0 01 02 03 04 05 060

125

25

375

50

0

4

8

12

16PMPP versus VMPP

1000Wm2

500Wm2

100Wm2

I(m

A)

V (V)

P(m

W)





120578MPP =119875MPPest

119875MPP (8)

015

020

025

030

035

040

045

050

02 025 03 035 04 045 05 055 06


VMPP versus VOC

VM

PP(V

)

VOC (V)




∘C119899d = 15 119877s = 0 and 119877p = infin




94

95

96

97

98

99

100

10 100 1000


120578M

PPT

()

G (Wm2)

LOCV Ta = 25∘C

LOCV Ta = 0∘C

LOCV Ta = 50∘C

FOCV Ta = 25∘C

FOCV Ta = 0∘C

FOCV Ta = 50∘C








119903s =119877s

[119881OC119868SC]STC

119903p =119877p

[119881OC119868SC]STC

(9)



120578M

PPT

()

G (Wm2)

96

965

97

975

98

985

99

995

100

10 100 1000



LOCV nd = 1


LOCV nd = 2








015

02

025

03

035

04

045

05

02 03 04 05 06


VM

PP(V

)

VOC (V)



VMPP versus VOC



75

80

85

90

95

100

10 100 1000

nd = 15 and Ta = 25∘C

120578M

PPT

()

G (Wm2)




∘C n119889= 15 and





0005

01015

02025

03035

04045

05

02 03 04 05 06

rp = 10 nd = 15 Ta = 25∘C and rs = 0

VM

PP(V

)

VOC (V)



VMPP versus VOC







minus

minus

+

=

ISC iPV


Covered PV panel

GS610 Power source

PV array simulator


E3631APower source

(Agilent)IDiode

10Ω

PV


3 32 34 36 38 4 42 44 46 48 52

25

3

35

4PV array simulator



VM

PP(V

)

VOC (V)



0 100 200 300 400 500 60092

93

94

95

96

97

98

99

100


120578M

PPT

()


PMPP (mW)



5 Conclusion




Acknowledgment


References



























1 Introduction








Maximumpowerpoint

Irra

dian

ce

Arr

ay cu

rren

tI

Terminal voltage V


Ener

gy g

ener

ated

()

100

90

80

Low Medium High

P and Ob

P and Oa ICa

ICb

OV

TP

TGSC

P and Oc

CV

Cost





6000

5000

4000

3000

2000

0 500 1000 1500

Pow

er (W

)

Time (s)

6000

5000

4000

3000

2000

0 500 1000 1500

Pow

er (W

)

Time (s)

(a)

Curr

ent (

A)

15

10

5

0 500 1000 1500

Time (s)

(b)

MPPES

PO

400

350

300

Volta

ge (V

)

0 500 1000 1500

Time (s)

(c)


6000

5000

4000

3000

2000

0 500 1000 1500

Pow

er (W

)

Time (s)

(a)

Curr

ent (

A)

15

10

5

0 500 1000 1500

Time (s)

(b)

MPPES

PO

400

350

300

Volta

ge (V

)

0 500 1000 1500

Time (s)

(c)













Non linear map

Lawpassfilter

Highpassfilter

120574

s

120594

120579

sin(120596t)120596t)

J

y = f(120579)

asin(

120592





update law 120579 = 119896(119886

22)(119891)

1015840(120579) which tunes

120579 to 120579lowast [13]


119910 = 119891 (120579 + 119886 sin (120596119905))

120579 = minus120574120594



(1)


ℎ) and

120596119897(119904 + 120596


119897lt 120596 lt 120596

ℎ[25] This model






minus80

minus70

minus60

minus50

minus40

minus30

minus20

minus10

10

Gai

n (d

B)

Frequency (rads)


0

10minus1

100

101

102



119866LP-BICO (119904) =119896

(119904120596119888+ radic1 + (119904120596

119888)2

)

119903 119903 isin R

+ (2)



119892BICOLP(119905) = 119896

119903119869119903(120596119888119905)

120596119888119905

(3)






119888and a large






119910 = 119891 (120579 + 119886 sin (120596119905)) (4)


119910 = 119891 (120579) +

119889119891 (120579)

119889120579

119886 sin (120596119905) +HOT (5)



119869 ≃ Lminus1

119904

119904 + 120596ℎ

lowast 119891 (120579)

+ Lminus1

119904

119904 + 120596ℎ

lowast

119889119891 (120579)

119889120579

119886 sin (120596119905)

(6)





120579 ≃

119886120574

2

Lminus1

1

119904

lowast Lminus1

120596119897

119904 + 120596119897

lowast

119889

119889119905

(

119889119891 (120579)

119889120579

)

(7)





119909 (119905) =

1

119879

int

119879

0

119909 (119905) 119889119905 (8)


120579 = 120579

119910 = 119891(120579) = 119891(

120579) = 119891 (120579)

120579 =

120574119886

2

119889

119889119905

(

119889119891 (120579)

119889120579


minus1

1

119904

(9)


119910 ≃ 119891 (120579lowast) +

119889119891 (120579)

119889120579

10038161003816100381610038161003816100381610038161003816120579=120579lowast


+

1

2

1198892119891 (120579)

119889120579

100381610038161003816100381610038161003816100381610038161003816120579=120579lowast

(120579 minus 120579lowast)2

(10)



0 thus

119910 ≃

1

2

1198892119891(120579)

1198891205792

100381610038161003816100381610038161003816100381610038161003816120579=120579lowast

1205792 (11)

By defining (12)(1198892119891(120579)119889120579

2)|120579=120579lowast = 119870 then


120579=120579lowast =


120579 = (120574119886120579

lowast 119889119870

119889119905


minus1

1

119904

(12)


119867119875119865 which its



120579 = (120574119886120579


minus1HLPF lowast L

minus1

1

119904

(13)


K120574a

1

sHLP (s) HHP(s)




119868 = 119868119871minus 119868OS (exp(

119902

119860119896119861119879

) (119881 + 119877119868) minus 1)

119868OS = 119868OR (119879

119879119877

) exp(119902119864119866

119860119896119861

(

1

119879119877

minus

1

119877

))

119868119871=

119866

1000

(119868SC + 119870119879119868 (119879 minus 119879119877))

119881 =

119860119896119861119879

119902

ln(119868119871 minus 119868119868OS

+ 1) minus 119877119868

(14)










119879119877


119877





0

0

1

2

3

4

Real

Imag

e 0

0

2


minus1

minus2

minus2

minus3

minus4

minus4


minus10minus20





0 5 10 15 20 250

5

10

15

20

25

30

35

40

45

50

55

Voltage (V)

Pow

er (W

)

= 30∘C

= 25∘C

= 20∘CTT

T









0 5 10 15 20 250

5

10

15

20

25

30

35

40

45

50

55

Voltage (V)

Pow

er (W

)

G = 1100Wm2

G = 1000Wm2

G = 900Wm2


0 2 4 6 8 100

5

10

15

20

25

30

35

40

45

50

Time (s)

Pow

er (W

)

ESC MPPTBICO MPPT




0 2 4 6 8 100

2

4

6

8

10

12

14

16

18

Time (s)

Volta

ge (V

) and

curr

ent (

A)



0 2 4 6 8 100

5

10

15

20

25

30

35

40

45

50

Time (s)

Pow

er (W

)

ESC MPPTBICO MPPT





0 2 4 6 8 100

2

4

6

8

10

12

14

16

18

20

Time (s)

Volta

ge (V

) and

curr

ent (

A)



7 Conclusions






0 2 4 6 8 10

0

10

20

30

40

50

Time (s)

Pow

er (W

)

minus10

Gain = 30

Gain = 55

Gain = 75

Gain = 95

(a)

0 2 4 6 8 10

0

10

20

30

40

50

Time (s)

Pow

er (W

)

minus10

Gain = 30

Gain = 55

Gain = 75

(b)




References













































1 Introduction










2 Methodology






1= 01 119865










Backup heaterSolar

collec

torWater tank Pump

Pump

HX

HRWDW DEC 1Room

DEC 2

1 2 3 4

56789

SA

Exhaust air RA

Ambient air


Backup heater

Water tank Pump

Pump

HX

HRWDW DEC 1 Room

DEC 21 2 3 4

56789

SARA

Exhaust air

Ambient air

Solar co

llector




7


DEC 2

Room

HX 1 HX 2

Ambient air


haus

t air

1 2 3 4 5 6SA

RA

89101112131415

Am

bien

t air



DEC 2

Room

HX 1 HX 2

Ambient air


1 2 3 4 5 6SA

RA

7891011121314










0

10

20

30

40

50

60

70

80

90

100

Time (hr)



Supply air

Ambient

Room (load)


Regeneration

Tem

pera

ture

(∘C)

85008000750070006500600055005000450040003500300025002000150010005000


00000

00050

00100

00150

00200

00250

Hum

idity

ratio

(kg

kg)



Ambient

Supply air

Room (load)


85008000750070006500600055005000450040003500300025002000150010005000








COP =119876cool119876119879

(1)


COP =119898119904(ℎ1minus ℎ4)

119898119903(ℎ8minus ℎ7)

(2)


COP =119898119904(ℎ1minus ℎ4)

119898119903(ℎ8minus ℎ7)

(3)


119903



COP =1198981199041(ℎ1minus ℎ6)

11989811990313(ℎ14minus ℎ13) + 11989811990310(ℎ11minus ℎ10)

(4)


COP =1198981199041(ℎ1minus ℎ6)

11989811990312(ℎ13minus ℎ12) + 1198981199039(ℎ10minus ℎ9)

(5)


119903






SF =119876119880

119876119879

(6)




119876119880= 119860 times 120578 times 119866 (7)







0030

0028

0026

0024

0022

0020

0018

0016

0014

0012

0010

0008

0006

0004

0002

0000

848076726864605652484440363228242016

Hum

idity

ratio

(kg

kg)

Temperature (∘C)

1

2

3

4

6 7

8

9

5

100 R

elativ

e hum

idity

90

8070

60

50

40

30

20

10


0

10

20

30

40

50

60

70

80

90

100



Supply air

Ambient

Room (load)


Regeneration

Tem

pera

ture

(∘C)

8500

8000

7500

7000

6500

6000

5500

5000

4500

4000

3500

3000

2500

2000

1500

1000

5000






00000

00050

00100

00150

00200

00250


Ambient

Supply air

Hum

idity

ratio

(kg

kg)

Room (load)


Time (hr)

8500

8000

7500

7000

6500

6000

5500

5000

4500

4000

3500

3000

2500

2000

1500

1000

5000











848076726864605652484440363228242016

Temperature (∘C)

1

2

3

4

100 R

elativ

e hum

idity

90

8070

60

50

40

30

20

10

5

6 7 8

9

0030

0028

0026

0024

0022

0020

0018

0016

0014

0012

0010

0008

0006

0004

0002

0000

Hum

idity

ratio

(kg

kg)






00050 12200 1948 00100 2822 00115 057 9690 1850 00100 2850 00115 02900080 9840 2095 00121 3024 00136 061 8030 2080 00121 3046 00135 04000100 8160 2200 00137 3080 00152 065 7020 2140 00136 3100 00150 050

0

10

20

30

40

50

60

70

80

90

100



Supply air

Room (load)

Ambient

Regeneration 1

Regeneration 2

85008000750070006500600055005000450040003500300025002000150010005000

Tem

pera

ture

(∘C)



00000

00050

00100

00150

00200

00250H

umid

ity ra

tio (k

gkg

)

Time (hr)Ambient




Supply air

Room (load)

Ambient

85008000750070006500600055005000450040003500300025002000150010005000













848076726864605652484440363228242016

Hum

idity

ratio

(kg

kg)

Temperature (∘C)

1

2

3

4

100 R

elativ

e hum

idity

9080

70

60

50

40

30

20

10

5

6

7

8

9 10 11

12

13

14

15

0030

0028

0026

0024

0022

0020

0018

0016

0014

0012

0010

0008

0006

0004

0002

0000


0

10

20

30

40

50

60

70

80

90

100

Time (hr)



Ambient

Room (load)

Supply air

Regeneration 1

Regeneration 2

Tem

pera

ture

(∘C)

85008000750070006500600055005000450040003500300025002000150010005000







00000

00050

00100

00150

00200

00250H

umid

ity ra

tio (k

gkg

)





Supply air

Room (load)

Ambient

85008000750070006500600055005000450040003500300025002000150010005000









848076726864605652484440363228242016

Hum

idity

ratio

(kg

kg)

Temperature (∘C)

1

2

3

4

100 R

elativ

e hum

idity

9080

70

60

50

40

30

20

10

5

6

7

8 910

11

12 13

14

0030

0028

0026

0024

0022

0020

0018

0016

0014

0012

0010

0008

0006

0004

0002

0000






4 Conclusions


Nomenclature









Acknowledgment


References







































61-butyric acid


2


1 Introduction
















3












SSiS

SS

SS Si

SS

SS

SS

DSi-6T

DH-6T

DH-6T

O

O

PCBM

C8H17C8H17

C6H13 C6H13

DSi-6T

PEDOTPSS

48 eV51 eV

51 eV

43 eV44 eV

27 eV

43 eV

62 eV

81 eV

TiO2

ITO

AI

PCBM

or


300 350 400 450 500 550 60000

02

04

06

08

10

Abso

rban

ce

Wavelength (nm)

PCBMDH-6TDSi-6T








06

05

04

03

02

01

00

1600140012001000800600400Wavelength (nm)

06050403020100

20151050Time (120583s)

800nm

700nm

2120583s8120583s

ΔAb

sorb

ance

ΔAb

sorb

ance

(a)

Time (120583s)

12

10

08

06

04

02

00

0 10 20 30 40

300

250

200

150

100

0 20 40 60 80 100

000

001

005

010

ΔAb

sorb

ance

[PCPM] (120583M)

[PCPM] (mM)

k1

sttimes103

(b)

















00 02 04 06 08

00

Voltage (V)

Curr

ent d

ensit

y (m

Ac

m2)

minus08

minus04

minus12

minus16





2Al (active layer








00 01 02 03 04 05 06 07 08Voltage (V)

00

Curr

ent d

ensit

y (m

Ac

m2)

minus08

minus06

minus04

minus02

minus10














1 4 061 126 039 0301 5 060 129 042 0321 6 057 088 037 018


DH-6T PCBM1 4 052 062 033 0111 6 063 060 044 017

DH-6T PCBM (with TiO2)1 4 051 061 042 0131 6 065 057 056 021












3Al



300 400 500 600 700 8000

2

4

6

8

10

12

14

16

18

20EQ

E (

)

Wavelength (nm)






119869 =

9

8

1205761199031205760120583ℎ

1198812

1198713 (2)




119903) due to the



1000 2000 3000 4000 5000 6000

minus41

minus42

minus43

minus44

minus45

minus46

Ln(JL3V2)

E12 (V05

m05)


(a)

00 05 10 15 200

10

20

30

40

50

ExpCal (1)Cal (2)

ExpCal (1)Cal (2)

J05

(A05m

)



(b)



ℎ= 120583ℎ0exp(12057411986412) where 120583

ℎ0is the hole


119869 =

9

8

1205761199031205760120583ℎ0exp (120574radic119864) 119881

2

1198713 (3)












200

(nm

) 100

0

(a)

(nm

)

80

40

0

minus40

minus80

(b)

(nm

)

75

50

25

0

minus25

minus50

(c)

(nm

)

20

10

0

minus10

minus20

(d)

(nm

)

120

80

40

0

minus40

(e)

(nm

)80

40

0

minus40

(f)

(nm

)

60

40

20

0

minus20

(g)

(nm

)

20

10

0

minus10

minus20

(h)



4 Conclusion




Acknowledgments


References

























2






60rdquo Chemical

















1 Introduction






2

16

12

08

04

00 20 40 60 80 100

02

06

1

18

14





VPV (V)

IPV

(A)







120

20

00 20 40 60 80 100

40

60

100

80

VPV (V)





PPV

(A)











Photovoltaic array

MOSFETdriver


controller

DCACinverter Load

PWM control signal

Cin

R1

R2

Cout

V

L

D

PV IPV



119881119895+1

119894= 119882 times 119881

119895


119895

119894)


119895

119894)

(1)

119875119895+1

119894= 119881119895+1

119894+ 119875119895

119894 (2)

In (1) and (2) 119881119895119894and 119875119895



1and 119862


119894









119899

+119882min (3)




119875best119894

=

119875best119894

119875best119894

gt 0

0 119875best119894

le 0

(4)

The 119875best119894


119875best119894

= 119875119895+1

119894 (5)



0 10 20 30 40 50 60 70 80 900

20

40

60

PPV

(W)

VPV (V)

(a)

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

Iterations

Modified PSO

Conventional PSO

Gbe

st(W

)

(b)





0 10 20 30 40 50 60 70 80 900

20

40

60

80

PPV

(W)

VPV (V)

(a)

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

Iterations

Modified PSO

Conventional PSO

Gbe

st(W

)

(b)


0 5 10 15 20 25 30 35 40 450

20

40

60

80

PPV

(W)

VPV (V)

(a)

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

Iterations

Modified PSO

Conventional PSO

Gbe

st(W

)

(b)









0 10 20 30 40 50 60 70 80 900

50

100

150

PPV

(W)

VPV (V)

(a)

0 10 20 30 40 50 60 70 80 90 1000

50

100

150

Iterations

Conventional PSO

Modified PSO

Gbe

st(W

)

(b)

0 10 20 30 40 50 60 70 80 90 1000

50

100

150

Iterations

Modified PSO

Conventional PSO

Gbe

st(W

)

(c)





1and 119862

2


6 Conclusion




MPPT methods















Acknowledgment


References





























1 Introduction




2 Experimental



(a)

Subgrains


(b)




3liquid source An




119909film









Wavelength (nm)

50

40

30

20

10

0

100

80

60

40

20

01200400 600 800 1000

IQE

()

Refle

ctiv

ity (

)


0 01 02 03 04 050

1

2

3


Loca

l ide

ality

fact

or

Voltage (V)



119898 =

1

119881119879

(

119889119881

119889 ln (119868)) (1)












(a) (b)

(c)








(a) (b)

(c)





4 Conclusions




(a) (b)

(c)


(a) (b)

(c)



(a) (b)


Acknowledgments


References


























solar energy and pv systems

Documents