Better Than Oil-Free:

Why Screw Chillers with Pure Speed Capacity Control Save More Energy Than Magnetic Bearing Chillers

Carrier Corporation Syracuse, New York May 2015

2

TABLE OF CONTENTS

INTRODUCTION .................................................... 2 ACHIEVING HIGHER EFFICIENCY ................. 2, 3 TO BUILD A BETTER CHILLER, BUILD A BETTER COMPRESSOR ............................................................ 3, 4 Bearings ........................................................................ 4 OIL-FREE CHILLERS AND THE IMPACT OF OIL ON HEAT TRANSFER................................................ 4, 5 ELIMINATING MECHANICAL UNLOADERS IMPROVES PERFORMANCE .................................... 5, 6 IDEAL FAN LAWS .................................................... 6-10 Warm Condenser Water Operation .............................. 8 Cold Condenser Water Operation ................................. 8

IDEAL FAN LAWS (cont) The Ideal Centrifugal Speed and Less Than Ideal Operating Conditions .................................................... 9 THE MISSING FACTOR IN COMMON INDUSTRY METRICS .................................................................. 10-13 The BIN Method’s Unintended Impact .................... 11 The Counterintuitive Energy Retrofit Case .............. 12 Warm Climates ......................................................... 13 CHILLER PLANT OPERATION: IDEAL vs ACTUAL OPERATION ................................ 13 Performance Degradation Over Time ...................... 13 SUMMARY .................................................................... 14 REFERENCES ................................................................ 15

____________________________________________________________________________________

INTRODUCTION

The proposed benefits of oil-free, magnetic bearing chillers have been widely marketed, and many current trade journals include multiple references to these chillers. Interestingly, the benefits of chillers that offer variable speed screw technology with pure speed capacity control1 have not received the same attention. This paper will compare and contrast the two technologies and illuminate a simple, logical approach to higher energy savings. See Table 1 for a comparison summary.

ACHIEVING HIGHER EFFICIENCY

In the case of oil-free magnetic bearing chillers, eliminating energy losses due to friction and

1 The use of mechanical unloaders of any kind is not required

under normal operating conditions.

reducing resistance to heat transfer are marketed as contributing to higher efficiency. In the case of variable speed screw chillers with pure speed capacity control, the elimination of mechanical unloaders and surge are, in contrast, marketed primarily as reliability and operational flexibility benefits. Equipped with these explanations, and provided only with the chillers’ efficiencies as shown in Figure 1, can you tell which of the chillers shown is oil free?

The answer may surprise you. The chiller on the left, with the higher kW/ton value, is the oil-free chiller; the more efficient chiller on the right is the variable speed screw chiller. This result is consistent across a wide range of capacities in which the two technologies are offered, as shown in Table 2.

Centrifugal Chillers with Oil-Free Magnetic Bearings Screw Chillers with Pure Speed Capacity Control

Frictionless bearings* Improved heat transfer† Less service and maintenance** Low Sound

No losses due to mechanical unloaders No use of hot gas bypass to prevent surge Less service and maintenance Low sound

* Main bearings require power for the levitation system, operating at higher motor speeds may increase in windage losses. Touchdown (back up) bearings are anti-friction bearings. † The impact of the studied benefit is still less than 0.1 – 1% depending on load, ignoring the benefit of oil “wetting” on the upper rows of evaporator tubes. ** Magnetic bearings include multiple sensors and coils as well as the electronics to control and power them.

Table 1 – Comparison of Claims

C(22334

*O 2

T

Capacity Tons)

250 290 360 390 450

Oil-free chiller eff23XRV efficiency

Table 2 – Chil

Oil-Free ChilleClaim (kW/ton0.633 / 0.357 0.634 / 0.328 0.576 / 0.327 0.604 / 0.330 0.590 / 0.346

W

ficiency values ay values are per

ller Efficienc

er n)*

23X(kW0.590.570.570.570.57

Which

FOil-Free M

are presented in pAHRI certified se

cy Values

Figu

XRV Chiller* W/ton)92 / 0.327 72 / 0.313 74 / 0.303 75 / 0.315 78 / 0.307

h chill

Figure 1 – AchMagnetic Beari

public domain, elections.

ure 2 – Chiller

3

TO A B

Givethe bthe moiledundeefficthe chillCondcondloadincre14%

ler is o

hieving Higher ings vs Pure Sp

Losses by Com

BUILD A BETTER CO

en the well-pbenefits of oilmost efficientd chillers. erstand how ciency. The pcontribution

ler at 100%ditioning, Heditions. Thes except the eased while h

% at full load t

oil fre

Efficiency – peed Capacity

mponent

BETTER OMPRESSO

publicized exl-free chillerst chillers in thTo understaneach chiller

pie chart showof losses w

% load, opereating and Re results wercompressor a

heat exchangeto less than 3%

ee?

Control

CHILLER,OR

xplanation su, it is worth n

he industry arnd why, wecomponent i

wn in Figure 2within a watating at AH

Refrigeration re similar atand VFD coner losses drop% at 25% loa

, BUILD

urrounding noting that re actually e need to influences 2 displays ter-cooled

HRI (Air-Institute)

t reduced ntributions pped from d.

Tblt

Aeuiemim

B

Wbc

Ibrmbcprep

Mictbdp

2

3

The outcome better chiller,osses within

the losses in th

A screw chileliminates alunloaders sucmprovement

efficiency momarketed benmpact of oil

minor in mod

Bearings

While markebearings in oconsume pow

In screw cobearings offerrolling elememechanical lobearing lossecompressor published kWrolling elemenelectronic bopotential mod

Many oil-freenstead of m

compressors othe selection bearings are mdo consume positioning s

2 The “other” in t

miscellaneous bypass in the sinefficiency alo

3 When operatedbearing in a 23

suggests tha start with athe compres

he chiller.2

ller with purl losses ass

ch as guide vato the comp

ost. In contrnefit surroundl on heat traern productio

eted as frictioil-free chillwer.

ompressor dr long bearingent bearing oss in a low es further deunloads and

W/Ton of tnt bearings dooards to opdes of failure.

e compressormechanical boperate at ove

of magneticmarketed as “

power. Powensors and

the compressor asmall refrigerant

system piping. Tone was over 50%d at AHRI condit3XRV compresso

NOT IMP

at if you wana better compsor represent

re speed capociated withanes, providinponent that crast, most of

ds the bearingansfer. Theson chillers.

ionless, the mlers actually

designs, rollg life and simlosses represspeed compre

ecline with sd are incluthe chiller. o not require

perate, elimi

rs use magnbearings. (Mer 30,000 rpmc bearings.) frictionless,” wer is requcontrol. Elec

and other categot side pressure d

The compressor m% of the chiller’s ions, the lowest

or is over 50 yea

PACTED BY O

Figure 3 –

nt to build apressor. Thet over half of

pacity controlh mechanicalng significantcontributes tof the oil freeg friction andse losses are

magnetic y do

ling elementmplicity.3 Thesent a smallessor. Thesespeed as theuded in the

Furthermore,capacitors ornating these

netic bearingsMany oil-freem, influencing

While thesethey actuallyired for thectrical losses

ory represented drops and mechanical total kW losses.L10 life of any rs.

OIL

– Impact of Oi

4

a e f

l l t o e d e

t e l e e e , r e

s e g e y e s

assocheat,a fluIn ahighelectwiththe mwill

Evenmechdesigthat or cwithbeari

In slosse

OILOF

The fracequibene

ASHRefrReseconcfromchillfractequivload

4 Me

is gmalow0.5

l on Heat Tran

ciated with th, which does

uid and wouldaddition, the h speed may tric motor. S

h the movememotor as it sincrease wind

n in comprehanical bearign. To ensurthere is an in

component fah a mechaniing.

summary, thes. No additi

L-FREE CHOIL ON HE

e net result ofction of a fraivalent to 0.1efit of oil).

HRAE (Amrigerating, aearch Projeccentration an

m RP 751 to ler, the net retion of a frvalent to 0..4

easurements in agreatest near theay improve heat twer parts of the b5%, too low to im

nsfer is Minima

his function anot advance

d therefore beoperation ofincrease the ince windageent of the cospins, in gendage losses.

ssors that usings are notre the shaft is nability to coailure) the coical bearing

he AHRI ratonal benefit s

HILLERS AEAT TRAN

of oil on chilaction of a fr1% to 1% (ig

merican Soand Air-Concts 601 andd heat transfdetermine t

esult of oil onfraction of a1% to 1%

actual full bundle e liquid level. At transfer by wettin

bundle the oil conmpact heat transfe

al

are generally ethe purpose oe categorized f compressorwindage los

e losses are aooling mediuneral, increasi

se magnetict eliminated supported inntrol levitatio

ompressor is called a to

tings includeshould be assu

AND THE IMNSFER

ller efficiencfraction, esseignoring the

ociety of nditioning Ed 751 pertaifer. Using the impact onn chiller effica fraction, eimpact depe

tests show oil cothe top of the bu

ng the tube surfancentration is leser.

emitted as of chilling as a loss.

rs at such ses in the associated

um around ing speed

bearings, from the

n the event on (power equipped

ouchdown

e bearing umed.

MPACT

cy is a entially e wetting

Heating, Engineers) in to oil the value n the full

ciency is a essentially ending on

oncentration undle, oil ace. In the ss than

TtprhTrdttfwpcn

AbfcterisaT5

ARcacmumpucacThcmrRd

5

These works that referencproperly. Freferences a heat transfer This should nreduction in cdeals only withe evaporatoto create the lfeed and drawere coveredpossible in ancomparisons not translate r

As there was be a water filfouling and mchiller has oilthe overall evaporator tubrefrigerant filmmpact evapo

shown in the accounts for The impact o5% of the loss

Another comResearch Proconcentrationa college camcurrent desigmanufacturedused R-11 management, pressure in understood tcommonly haare not issucentrifugal anThe inherent higher differechillers is onmodern desigresults from R-134a positidecades of im

5 The total kW lo

comparing a mchilled water teof a chiller withtemperature eqa perfect theor

are quite tecces to theseFor example“22-25% redcoefficient dunot be misintchiller efficienith the refriger. Electric caoad. The te

ain system thd in liquid rn actual chilleof oil to oil-

reliably from

no water in lm resistance,

metal resistancl in it or not.resistance t

be. The remm and is the rator heat trapie chart in

only 5% of tf oil is therefses in the chil

mmonly citedoject 601. s in samples

mpus. These wgn. Half od prior to 197

refrigerant. oil retards

low pressurthat old nead oil recoveues with pond screw chil

improvemenential pressuree of the signgn. It woul this study

ive pressure mprovements.

osses in an evap

model of a chiller emperature less h a zero degree aqual to saturatedretical evaporato

chnical and ite publication, Research

duction in refue to averageterpreted as ncy. Researcerant side heartridge heateest bundle righereby assurirefrigerant, a

er. In general-free in real studies using

the tubes, the, or fouling. ce are the sam. These repreto heat tran

maining 27% ronly opportunansfer. (See Fn Figure 2, ththe losses in fore 22 to 25ller.

d reference This paptaken from 1were older ch

of the ten c70 and all o Without the refrigera

re chillers. egative pressery challengesositive pressllers in produ

nt in oil recoes associated nificant advanld be incorreare applicab

chillers built

porator were detewith a normal apsaturated suctionapproach (leavind suction temperar).

t is importantns are cited

Project 751frigerant sidee oil effects.”a 22 to 25%h Project 751

eat transfer iners were used used an overing all tubesa set up notl, quantitativebundles maybundle rigs.

ere could not Water film,

me whether aesent 73% ofnsfer in anresides in thenity for oil toFigure 3.) Ashe evaporator

the system.5

% of 27% of

is ASHRAEer cites oil10 chillers onhillers, not ofchillers weref the chillerseffective oilant-oil vapor It is wellsure chillerss. But thesesure R-134auction today.overy due towith R-134a

ntages of theect to imply

ble to modernt after 3 to 4

ermined by pproach (leaving n) with a model g chilled water ature, signifying

5

t d 1 e ”

% 1 n d r s t e y

t , a f n e o s r 5 f

E l n f e s l r l s e a .

o a e y n 4

Givefree overdata.smalchillelimperfothe d

ELIUNLPER

To bspeethe aA diprovmech

RefrcompAlonturneto rerecircondcond

Imagthroureduunit usedguidcreat

en the well-pdesign impro

restimate the . As outlinedll. In the clers, it may b

minating meormance. It dramatic impa

IMINATINGLOADERS RFORMAN

better illustrated capacity coadvantages ofiscussion of svide ample ophanical unloa

rigerant flowspressor and ng this path, ted and acceleestrict, recircurculation aredenser) or hdenser, evapo

gine a pipe wugh it. Now

uce flow and oof throughpu

d to reduce flode vanes in ate a pressure

Figure 4Losses

publicized expoves efficienc

impact in td above, the icase of the be obvious thechanical may also be act this has on

G MECHAIMPROVE

NCE

te the compreontrol, it mayf eliminating speed and refpportunity to aders used to r

s from the eis dischargethe refrigeran

erated. Mechulate or bypa

e hot gas bhot gas rec

orator, and firs

with a large vow place a resobserve what

ut. In a chilleow. While re

a centrifugal drop of 1% to

4 – Inlet Guide s Increase as th

planations ofcy, it would bthe absence impact is actuvariable spe

hat reducing unloaders easy to unde

n performanc

ANICAL ES

ession benefiy be helpfulmechanical u

frigerant gas observe the

restrict flow.

evaporator thrd into the cnt may be comhanical unloadass flow. Exaypass (bypa

circulation (bst stage impel

olume of fluistriction in tht happens to pr, inlet guide educing flowcompressor (o 13%.

Vane Mechanihe Vanes Close

f how oil- be easy to of actual

ually quite eed screw

losses by improves

erestimate ce.

its of pure to realize

unloaders. flow will impact of

rough the condenser. mpressed, ders serve amples of

assing the bypassing ller).

id moving he pipe to power per vanes are

w, the inlet (Figure 4)

ical e

Tcitt

I

Tauti

6

This increascompressor thmpact observ

to 50%.6 Notthe <1% from

IDEAL FAN

This section wapproach (deuses only speethe ways in wmpact efficie

6 In the paper “A

Centrifugal Coof inlet guide vcompressor, isaccounts for. Taerodynamic eon the compre

1s

2n

3r

*It

es the heahereby increaved in comprte, the impact

m the oil free e

N LAWS

will comparescribed aboved to control

which eliminaency, we will

Aerodynamics of

mpressors” 1996vanes on efficiencs far higher than tThis observation

efficiency due to cssor map are als

st Law: Mass

nd Law: Lift (a

rd Law: Power

t may be less tha

d requiremeasing power.ressor performt is significanteffect discusse

e the mechanve) to a comflow. By coting mechaniunderstand th

Rotatable Inlet G6, JJ Brasz notescy, 4% to 50% inthe increase in h suggests that chchanges in the oso a factor.

flow (capacit

ability to gene

r is the cube o

an cubic due to th

ents on the The actualmance is 4%tly more thaned earlier.

nical unloadermpressor thatnceptualizingcal unloadershe magnitude

Guide Vanes for s that the impact n a centrifugal head alone hanges in

operating point

ty) is linear w

rate head) is

of speed*

he proximity of g

Figure 5 – Id

6

e l

% n

r t g s e

of themplof scentrrefrigadhe

Signinduand pa cruabiliomit

ith speed

the square of

uide vanes to the

deal Fan Laws

he savings. loy limited spspeed controrifugal comprgerant type,

ere to the idea

nificant markeustry on the 3r

power. Unfoucial role in eity for the chitted.

f speed

e impeller inlet

s

Oil-free cepeed control,ol where poressors (regarmotor type, o

al fan laws, ill

eting has beenrd law, relation

ortunately, theequipment effller to stay on

entrifugal chi leveraging th

ossible. Howrdless of beaoiled or oil flustrated in F

n done to edunship betweene 2nd law, whificiency (and in line), is com

illers also he benefit

wever, all aring type, free) must igure 5.

ucate the n speed ich plays in the

mmonly

Ics8od

Telmibhafss

Tc

7

In simple tercentrifugal cospeed. At 8080% x 80% operating condesign head, t

To avoid sureither speed uoad balance

magnetic bearnclined to be

based on capahead requiredat 50% load afirst law wouspeed, but thspeed. (See F

This raises ancompressor m

7 Centrifugal com

between the locondenser. If treversal will oc

rms, the abiliompressor is r0% speed, a = 64% of

ndition requthe centrifuga

rge, the centup or engage h

valves or rering machinelieve that comacity, it is ac

d. For exampand 75 F enteuld suggest whe 2nd law reFigure 6.)

n interesting qmust operate

mpressors moveow pressure evapthe pressure diffeccur (surge).

Figure 6 –

ity to generarelated to the

centrifugal wits design hires more th

al compressor

trifugal comphot gas bypascirculation fl

es). So whilempressor speectually often ple, if a chilleering condenswe could opeequires us to

question: If that 84% spee

gas through an

porator to the higerence (head) is

– Determining

ate head in asquare of thewill generate

head. If thehan 64% ofr will surge.7

pressor mustss (also calledlow on manye we may beed is typicallybased on the

er is operatingser water, theerate at 50% run at 84%

he centrifugaled to develop

open path gher pressure

too large, flow

Compressor Sp

7

a e e e f

t d y e y e g e

% %

l p

suffi50%inletbalanof dour inletconsor vabypagas bto mmeet(hot first The morewhiccooli

Speed Required

icient head, %? This requt guide vanesnce valves (h

diverting or rsimple pipe

t guide vasumption andariable geomass or recircubypass or rec

move throught the load. Tgas bypass) stage of thenet result i

e refrigerant ch representsing at the ope

d to Develop Su

how do we uires an unlos, discharge

hot gas bypassrestricting floexample, it

anes will id reduce effic

metry diffusersulation flow wcirculation flo

h the compresThis flow by or the conde compressois the comprthan is requ

s an increase erating point.

ufficient Head

limit the caoading deviceflow restrictis) or some othow. Thinkingbecomes obvincrease theiency. If gus are removedwill be requiow allows hissor than is nypasses the edenser, evapor (recirculatiressor is com

uired to meet in power pe

apacity to e such as ions, load her means g back to vious that e energy uide vanes d, hot gas ired. Hot gher flow needed to evaporator orator and ion flow). mpressing the load,

er unit of

W

NchlcepcsrFcwTcewccsmdromwca

8

Warm Cond

Now considerchiller. Screwhead at any spaws. So if th

capacity at 7essentially rupoints of spechiller. To uscrew comprreduction is gFigure 7 illucentrifugal anwith constant The variable capacity contreven with conwater temperchiller is less chiller enjoysspeed reductiomust employ decrease in erepresents mooutput). In mechanical unwarmer climachillers will advantage.

Any speed

8 Maintaining hig

multiple chiller expected to opbuilding loads chillers needed3-chiller plant).

Figure 7Cond

FiguC

denser Wate

r this same opw chillers propeed and are nhe screw chil5 F entering

un at 50% speed reductionunderstand horessor just rood speed redustrates this nd screw perf

entering conspeed screw

rol reduces spnstant 85 F (2rature. Effic

than half loas the same reon is limited mechanical

efficiency is ore power conthe graphic nloaders are ates pure spe

provide a

reduction isand more

gh efficiency aboplants since only

perate below 40%below 40% x 1/Nd to meet building.

7 – Chiller Effidenser Entering

ure 7 – Chiller Condenser Ente

er Operatio

perating pointovide the samnot bound byller needs to

g condenser wpeed. This rn relative to ow the physiremember: duction and m

principle byformance at vndenser waterw chiller withpeed as the lo29.4 C) entericiency increaad.8 While theduction in lby the 2nd idemeans to unlobserved (hi

nsumed for a gbelow it is required anded capacity can enormou

s good speede is better.

ove 40% is highlyy the last chiller o

% load and this oN, where N is theg full load (for ex

ciency at Consg Water Tempe

Efficiency at Cering Water Te

on

t with a screwme amount ofy the ideal fan

provide 50%water, it willrepresents 34a centrifugalcs favors the

Any speedmore is better.y comparingvarying loadsr temperature.h pure speed

oad is reduceding condenserases until thehe centrifugalload, becauseeal fan law, itload, and theigher kW/tongiven coolingobvious that

d therefore incontrol screw

us efficiency

d reduction

y valued in on line is

only occurs at a e number of xample 13% in a

stant 85 F erature

Constant 85 F emperature

8

w f n

% l 4 l e d .

g s . d d r e l e t e n g t n w y

Cold

Is it opernot capplicontiguidbypaFiguovertempmay varyprofitons wate

A vacapaefficentemorcent

9 Un

equ

d Condense

possible to crating range wcome into plaication, with

tinuously, wode vanes, varass and flow rure 8 depicts r a wide varperatures. No

be fixed, they throughout file; significan

with signifier temperature

ariable speedacity controlciencies overering condenre than constrifugal chil

less there is conuator for example

Figure 8 – ChCondens

er Water Op

compare the twhere the mecay? It may sh chillers ruould provideriable geomerecirculation chillers operariety of enteote, that while outdoor amthe data ce

nt hours willicantly reduce.9

d screw chill performs arall, leveragnser water testant or variallers.

nstant entering coe.

hiller Full Loaser Entering W

peration

two technolochanical unloseem that a dunning at 10e a case whetry diffuserswould not beating at 100%ering condenle the data ce

mbient tempernter’s 24x7 l be spent at ed entering c

ller with purat higher opeging the reduemperature able speed

ondenser water,

ad Efficiency atWater Temperat

gies in an oaders will data center 00% load here inlet s, hot gas e required. % capacity nser water enter load rature will operating 100% of

condenser

re speed erating uction in even

on the

t Varying ture

Acitevrwaco2earfnbarudcc

BtwwbamccecorbsFftmhcr

As the condencooler in the cmproving ref

through the evaporator prvapor. Threfrigeration will not conalready changcompressor mof 120% (10020% due flaentering condand sub coolrefrigerant liqflashing acronow be moviboiling in thacross the orremains constunloading frodesign head compressor mchiller capacit

But even as tthe speed reqwider marginwater temperabecomes irreleat play, furmechanical comparing tcentrifugal cexclusively bcentrifugal opobserved. Threfrigerant floblade, from cspins, the morFigure 9 belofixed and the the 2nd ideal more energy head at the opcompressing required.

F

nser water temchiller becomfrigerant qualorifice from

ressure, somehis vapor w

cycle, includntribute to evged state. A

may be movin0% due to boishing acrossenser water telers in chillequid to the oss the orificeing a volume

he evaporatorrifice). Wtant at 100%, om 120% to

to a reducmass flow rety has been re

the speed neequired to gen. Therefore, ature is reducevant. Since rther insight unloaders

the constancurves. Sby mass flowperating withhe impeller wow throughoucenter to tipre energy is trow. Since tspeed is greafan law, the on the gas

perating poinand perfor

Figure 9 – Impe

mperature decmes more effeity. As refri

m condensinge refrigerant will move ding the comvaporator tonAt design cong a volume ling in the ev the orifice)emperatures ters provide corifice. The e. The come of 110% (r, and 10%

While the chithe compress110% of ma

ced head coequired to aceduced.

eded for massnerate head d

as the enterinced, the 2nd ithe 2nd ideal

as to thebecomes vi

nt and varSpeed is nw and the

h excess tip swill impart eneut the entire . The faster ransferred, asthe impeller

ater than requiimpeller bladflow than is

nt. The centrirming more

eller Blade

clines, the subective therebygerant moves

g pressure towill flash tothrough the

mpressor, butns as it has

onditions, theof refrigerant

vaporator, and). At lowerthe condensercolder denserresult is less

mpressor may100% due todue flashing

iller capacitysor is actuallyss flow fromndition. Thechieve 100%

s flow drops,declines by ang condenserideal fan lawfan law is not

e impact ofisible whenriable speedow dictatedimpact of aspeed can beergy onto thelength of thethe impeller

s illustrated ingeometry is

ired to satisfyde will impart

required forifugal is over

work than

9

b y s o o e t s e t d r r r s y o g y y

m e

%

, a r

w t f n d d a e e e r n s y t r r n

Screheadcomppoinspeeperfolevertempspeethe pthe mcentrexplsimpthe sefficflow

TheIdea

The ideabe.

The requrequcentrwhenrequunlolowespeecenteflowheadcompcomp

Comthat occuexacAbovcentrfor emplflowcompchillan afromwill

ew compressod at any speedpressing at th

nts. What maed screw wiorms at highraging the redperature evened centrifugal performance mechanical unrifugal compranation is bey

ple terms, thescrew compreciency of the

w, low head co

e Ideal Cental Operatin

e further the al speed curv

ideal fan uirement for fuirement for rifugal compn the speed r

uirements, meoad a centrifuer efficiency ed control aloer, we disco

w often exceed. In this cpress, but presses with l

mbining thesethe ideal sp

urs when thctly matches tve this sperifugal to op

capacity aloyed. Belo

w exceeds the pression. Inlers with pureadvantage. T

m the ideal spbe.

ors develop d, and so by dhe high capaciay be unexpecith pure spher operatingduction in enn more thanchillers. In tadvantage c

nloaders alonressors are oyond the scope resulting isessor is higher

centrifugal conditions.

trifugal Speg Condition

e operating pve, the great

laws defineflow and a sep

the generapressor. Earlrequired for hechanical meugal compres

than a scrone. Next, iovered that thds the speed case, both c

the screwless penalty.

e observationspeed for a ce speed reqthe speed requeed, the hierate at highand mechanow this speedhead requirem

n either case e speed capaThe further tpeed curve, t

the same adefinition areity, low head cted is that thpeed capacityg efficiencietering conden

n constant orthis operating

cannot be attrne. Both the sver compresspe of this papsentropic effir than the aercompressor at

ed and Lessns

point is fromter the impa

e a minimuparate minimation of helier, we obse

head exceededeans were ressor, which reew compresin the case ohe speed reqnecessary to

ompressor tyw compress

s, it becomescentrifugal coquired for muired to genegh head fo

her speeds thanical unloadd, the speed ments resultinvariable spe

acity control othe operatingthe greater th

amount of also over operating

he variable y control s overall, nser water r variable g scenario, ributed to screw and sing. The per, but in iciency of rodynamic t the high

s Than

m the ct will

um speed mum speed ead in a erved that d the flow equired to esulted in sor using f the data

quired for o generate ypes over sor over

s apparent ompressor

mass flow erate head. orces the an needed ders are for mass

ng in over eed screw operate at g point is he impact

Cc1ctTtcrratTa

OcwlbcIwttpsc

1

1

Consider ovecommon prac15%, but thischillers wouldthe entering The net effecthe left fromcapacity has remains veryreduction, furtand mechanicthe chiller is oThis effect wallows a chille

Once a chillchillers will rwould therefoow delta T an

building is acondenser wIronically, thewhere the spethe speed reqthe use of mpoint where speed requirecompression).

0 If chillers are

full load efficie1 Technically th

delta T across100% load.

Figu

er sizing foctice to overs may come ad be operatincondenser w

ct is that the m A to A1 (

been reducey close to dther speed red

cal unloaders oversized, the

will be observer to be taken

ler is taken ramp up to hore be runningnd other operat part load,water tempe chiller wenteed required tquired to genemechanical un

speed required for head .

commonly oversency seems illoghe head droppeds the condenser

ure 10 – Ideal

or example. size the chilat a price. A

ng at 85 to 90water would b

operating po(Figure 10). ed, but the hesign.11 Wduction is sevmust be use

e bigger the imved until the bn off line.

off line, thhigh load. Tg at 100% loarating realitie it is expec

perature hat from operatito generate heerate flow (thnloaders) to red for flow

(thus result

sized, emphasizigical. d by 1.5 F due to

water at 85% loa

Centrifugal Sp

It may beller by 10 toAt peak load,0% load10 butbe at design.oint moves to

The chillerhead requiredWithout headverely limitedd. The more

mpact will be.building load

he remainingThese chillersad (subject to

es). Since thected that theas dropped.ing at a pointead exceededhus requiringan operatingexceeds the

ting in over

ng or prioritizing

the reduction in ad versus of

peed

10

e o , t .

o r d d d e .

d

g s o e e . t d g g e r

THEIND

The elegpoinor o

The examcompbetwtwo kW/tchilldoesHowdiffeseencons

Theswateidealallowdecliunloat thapprshow12%operwill tempclimtempside operadvireprea sinemplmann

12 AHcon

13 AHcom

Lo10755025

E MISSINGDUSTRY M

e reality of acgant than IPnts where thover compres

practice of omple of how pare chillers

ween centrifumost commoton and IPLVlers are overss not accuwever, this reerences betwen in the subsiders four dis

se four poiner temperaturl fan curve ws speed reqine together

oaders or overhese four discropriate, but twn in the abo

% and 45% rating at 65 F

not supportperatures for

mates will experatures even

economizerrating hours ses that IPLesentation of ngle chiller oloy more thanner in which

HRI 550/590 reconducting a life cy

HRI 550/590 sectmprehensive too

Enterioad Water00% 5% 0% 5%

G FACTORMETRICS

ctual operatPLV, resultin

e use of mecssion are mo

oversizing chthe most com

s may not rugal and screonly referenceV (integrated psized, the sub

urately depiceality is not aeen compressbmittal data.screte operatin

nts assume thre is droppingperspective, quirements freducing ther compressio

crete operatinthis is rarely ove chart theof the time

F. The weatht 65 F ente

57% of thexperience con at high loars reduce th

at low loadLV “was dthe average p

only.”13 Mosn one chiller

h chillers are

ommends local wycle analysis. tion D.2 further e

ol should be used

ing Condensr Temperatur

85F 75F 65F 65F

R IN COMM

tion is muchng in operatichanical unore likely to

hillers providemmon metricreflect the dew compressoed metrics arepart load valubmittal full lct actual oacknowledgesion types w. The IPLng points:

he entering cg with load. this operatin

for flow ande need for mn. If chillers

ng points this the case (if e

e assumption e, the chillerher in warmerering condene ton hours.12

lder condensads while air he number ods. AHRIderived to ppart load efficst chilled wato meet the lsequenced in

weather to be use

explains that a md for life cycle an

ser re Weigh 1% 42%

45% 12%

MON

h less ing loaders occur.

es a good cs used to differences ors. The e full load ue). If the load point operation. d and the

will not be V metric

condenser From an

ng profile d head to mechanical s ran only would be

ever). As is that at

r will be r climates ser water 2 Cooler ser water and water of chiller I 550/590 provide a ciency for ater plants load. The n multiple

ed when

ore alysis.

ht

chrmtAmpobtAcpog

Cory

W6mbcWiwaFat1n

1

chiller plants hours occurrinrarely depicts most chiller pthe local weatAHRI profilemuch less elepoints where over compresbecause the athe ideal speeAHRI IPLV chillers with possess induobserved in agreater.

Consider for operating prorepresented oyellow and bl

We observe t67% (two thimet by operatblue chiller chillers will lWhile the chtself is just 6

water temperaas listed as thFor this examagain when ththus requires 100% load ponot all 100% l

4 Subject to low

Figure 1

will typicallng in the 75%an actual chi

plants have mther likely doe. The realiegant than IPL

the use of sion are mor

actual operatind curve than load profile. pure speed

ustry leadingactual operatio

example a ofile of ch

on the chart iack lines.

hat when theirds load) theting just 2 ofis turned ofload up to 1

hillers are at 67% loaded, ature would nhe 100% loa

mple we will ahe building o

only one coints to the AHload points ar

delta T issues a

11 – Chiller Op

ly result in 2% load rangeiller’s load prmore than onoes not matchity of actual LV, resultingmechanical

re likely to ocng points arethe operating While thecapacity con

g IPLVs, on is expecte

three-chiller hillers 1, 2 in Figure 11

e building hase needed capf the 3 chillerff, the yellow100% to mee

100% load, so the enterin

not be expectead point on tassume 75 F.operates at 3

chiller. ComHRI submittare the same.

and other operatin

perating Profil

/3 of the tone. The IPLVrofile becausene chiller andh the standard

operation isg in operatingunloaders orccur. This is further from

g points in thee VFD screwntrol alreadythe benefits

ed to be even

plant. Theand 3 areby the blue,

s unloaded topacity can bers. After thew and blacket set point.14

the buildingng condensered to be 85 Fthe submittal. This occurs3% load and

mparing theseal reveals that

ng realities.

les

11

n V e d d s g r s

m e

w y s n

e e ,

o e e k 4 g r F . s d e t

OperCond

AHRCondActuaCond(#3)ActuaCond(#5) The centbut tsubm

The speeoperflowoverefficCentchillopereitheshouat opThismuchdata hourwe overof ocond

The

The enouanalymodreflemanuwill load to remetr

rating dition

Lo

RI dition

10

al dition

10

al dition

10

e screw chilletrifugal chilthis benefit mitted full lo

AHRI pointed curve for crating point d

w and head arr compressiciency as trifugal Speeler outperformrating points, er the submittuld be noted, tperating points observation h emphasis iwhen a ma

rs occur at ofaccept the

rsized, most coperation at denser water t

e BIN Metho

impacts of lugh that it isysis cannot b

dified such tect local weufacturers haprovide thesprofile, use o

ecognize that ric, and BIN

oad EnteringCondenWater Temper

00% 85 F

00% 75 F

00% 65 F

er further oller at operawill not be soad point or

t will fall vecentrifugal codoes not, as re now quite ion occurs,seen previod graph in Fms the centrbut this bene

ted full load that far more ts 3 and 5, tha

n raises the qis placed on ajority of theff peak tempeidea that chchillers will 100% load

temperature.

od’s Uninte

local weathers recommend

be performed,that temperaather and ch

ave simple mse values baof economizet IPLV or mometrics use a

g nser

rature

DesigFlow

~ 100

~ 100

~ 100

utperforms ting points 3seen in either the IPLV.

ery close to mpressors. Tthe speed ndifferent. In, impactingously on tFigure 10. Trifugal chillerefit will not bpoint or the Iton hours are

an at operatinquestion as tothe full load e high load eratures. In adhillers are cexperience zwith design

ended Impac

r and staging ded that whe, the IPLV eqatures and whiller staging

modeling progased on yourers, etc. It is odified IPLVaverage value

n Design Head

% ~ 100%

% ~ 80%

% ~ 60%

the 3 and 5, er the

the ideal The actual needed for n this case g chiller the Ideal The screw r at these be seen in IPLV. It e expected ng point 1. o why so submittal operation

ddition, if commonly zero hours n entering

ct

are large en hourly quation be weightings g. Most

grams that r building important

V is a BIN es. Using

12

average values can be convenient but it can also oversimplify. Consider a spring day with a cool morning and a warm afternoon. In the morning the building requires some heat and in the afternoon some cooling. On average, the building requires neither heating nor cooling but you should not expect your utility bill to reflect this average. The same is true for BIN data.

Within each load BIN, chillers operate over a wide range of conditions. On average, the chiller may be operating at 75% load with 75 F entering condenser water temperature. In actual operation, the loads and condenser water temperatures vary above and below the average. Figure 12 compares chillers of similar full load kW/ton to evaluate the variance between the 75% point published and the surrounding operating points. We find that IPLV uses the average operating point, not the average efficiency for the range of operating points.

For the variable speed centrifugal chiller evaluated, the average of operating points A,B,C,D is 5.6% less efficient than the rating point. For the variable speed screw chiller evaluated, this variance is just 1.5%. These results are consistent with the compressor behavior presented earlier. For centrifugal chillers operating at warmer temperatures, the head limits speed reduction and mechanical unloaders must be used. At colder temperatures, the centrifugal compressor is over compressing. The average efficiency resulting from the use of mechanical unloaders and over compression is not equal to the IPLV rating point, but is actally 5.6% lower!

Earlier we presented AHRI certified data showing that oiled screw chillers have better IPLVs than centrifugal chillers. Here we see that in actual operation, the advantages are likely larger than the full load and IPLV metrics would indicate. As we

move from IPLV to actual operation, the impact of averaging alone accounts for a large variance in operation. This observation leads to the following question: What other common issues experienced in actual practice impact chiller performance, and do they impact screw chillers differently than centrifugal chillers?

The Counterintuitive Energy Retrofit Case

A facility manager once boasted that his load had been significantly reduced by installing a new roof, new lighting and other changes. Based on his new lower cooling demands, he intended to upgrade his chillers to new units that specifically featured great efficiencies at part load. The solution was counter intuitive. The chiller that had the higher kW/ton (worse) values listed at the 50% and 25% load points at AHRI operating conditions was actually better. Reviewing in more detail we find that while he lowered the cooling demand, he could not lower the outdoor wet bulb. The AHRI IPLV efficiencies listed at 50% and 25% load are based on 65 F entering condenser water. His new chillers would be operating at these new lower loads, but with the warmer condenser water temperatures normally associated with chillers operating at higher loads. In essence, he had shifted his operating profile, creating oversized chillers. Only after accounting for both load and condenser water temperature did the answer become clear. Centrifugal chillers operated at high speed due to head requirements and used mechanical unloaders to get to 50% and 25% load. Screw chillers could use speed control alone, even with high entering condenser water temperatures, yielding true energy savings.

A B A B62.5% 87.5% 62.5% 87.5%

80F 0.525 0.5 80F 0.448 0.483

70F 0.376 0.398 70F 0.346 0.381D Average of Four Corners = 0.45 kW/Ton C D Average of Four Corners = 0.415 kW/Ton C

Variance =1.056 Variance =1.015

Centrifugal Chiller Screw Chiller

0.426

Entering Cond Water Temp.

Load Load

Entering Cond Water Temp.

0.409

IPLV reflects the average operating point, but the numerical average of operating at points ±5ºF and ±12.5% capacity above and below the IPLV point can be 5.6% worse for centrifugal chillers.

Figure 12 – Comparison of Actual Operating Loads to Average Loads

13

Warm Climates

In warmer climates, the speed of a centrifugal chiller is more likely to be dictated by the head required. This increases the use of mechanical unloaders and increases the expected relative operating savings of using variable speed screw compressors with pure speed capacity control.

CHILLER PLANT OPERATION:

IDEAL vs ACTUAL OPERATION

A typical life cycle analysis often assumes that a chilled water plant will not only achieve ideal operating conditions, but sustain them over a 25-year period. Ignored are the less than stellar maintenance practices, wear and tear, imperfect controls, low delta T syndrome, the fact that the equipment is oversized by 10 to 15% and, of course, unusual override requirements. What impact will these factors have on the chiller operating condition? The operating point is moved further from the ideal speed curve of a centrifugal chiller. As we move away from this point, the difference in compression matters more.

An interesting exercise to demonstrate this impact would be to perform a sensitivity analysis on a life-cycle evaluation. Change the simulation assumptions to reflect an actual load of 85% of the nominal chiller nameplate capacity. To simulate poor maintenance, use a tower approach of 10 F instead of 7 F. The results will show the effects of actual operating conditions on chiller performance.

Performance Degradation over Time

The concept of performance degradation over time occasionally sparks some debate. Some have suggested oiled chillers may worsen over many years of operation due to an assumed increase in oil concentration. There is no evidence to support this occurrence on modern positive pressure chillers. With respect to oil concentrations in the field, an ASHRAE subcommittee noted that, “the data is extremely sparse and statistically insignificant.”15 Any nominal impact of oil is already included in the AHRI submittal data and reflected in the approach temperature both on the submittal and in test results.

15 RTAR from Subcommittee 8.2 “Oil Concentration of Field-

Installed Chillers with Flooded Type Evaporators.”

Despite the seemingly large numbers often referenced in claims of oil impact on heat transfer, the actual impact of oil is a fraction of a fraction of a fraction, totaling just 0.1 % to 1% of the total losses in a chiller, as shown earlier in Figure 3. Oil-related losses are included in the AHRI certified performance. Comparisons of AHRI submittals reveal oiled chillers may have lower approach temperatures than oil-free chillers. Much is made of the impact of oil, but it is machine design and heat exchanger surface area that drive performance.

The approach temperatures of chillers and the instances requiring corrective action are commonly logged. Common causes of increasing approach temperatures are tube fouling, low refrigerant charge or issues with the metering device between the condenser and evaporator. Low refrigerant levels may contribute to oil accumulation in the evaporator and therefore, oil is often blamed; however, the actual cause of the change in performance is the low refrigerant charge. If the refrigerant charge level is adjusted to the appropriate level, the approach temperature will decrease. If all of the oil is removed from the evaporator and the refrigerant level is not corrected, the approach may actually get worse because the refrigerant level is still not covering the tubes and without the wetting benefit of oil, further degradation may occur.

Rather than focus on any one hypothetical cause of performance degradation for marketing gain, we should instead recognize that the maintenance of high performance variable speed chillers is more valuable than ever. Poor maintenance practices of many kinds have a common symptom. The head required of the compressor will rise. Poor tower maintenance, poor water treatment, skipping tube brushing, low refrigerant charge and many other issues all result in higher approach temperatures which increase the head on the compressor. In the case of centrifugal compressors, raising the head will raise the speed, and now the poor maintenance practice has elevated the power by the cube. A variable speed screw chiller speed is largely unaffected as lift increases. So while power will also increase, it is less sensitive to such issues.

S

Sfcwgaaac

NerHtaelnDwt0b

IewCtrbr

C

SUMMARY

Screw comprfan laws andcan be preciswithout the ugas bypass oa significantadvantage, imaccounts forchiller.

Not all compoefficiency equrefrigerant filHowever, evatotal losses accounting fevaporator loosses (water

not change wDespite the swith respect transfer coeff0.1 and 1% debenefit of oil)

In contrast, vaefficiency bywhere over hCentrifugal coto operate arequiring inletbypass or flrefrigerant tha

COMPARISO

Y

ressors are nd are not subsely matcheuse of mech

or flow recirct compressiomproving th

r over 50% o

onents and suually. Oil-frelm resistanceaporator losse

in a chillefor just rousses. The vafilm, fouling

whether the cseemingly larto the impac

ficient, the reepending on .

ariable speed y focusing ohalf of the lompressors (oat higher spt guide vaneslow recirculaan needed) t

ON CHART

not bound bybject to surg

ed to the opeanical unloaculation. Thon efficiencyhe componenof the total lo

ubsystems inflee chillers se

e associated es represent jur, with refr

ughly a quast majority og and metal rchiller is oilerge numbers ct to refrigerasulting benefload (ignorin

screw chilleron the composses in a coiled or oil-frpeeds to prs (flow restrication (compro reduce cap

by the ideal ge, so speed rating pointaders, hot his providesy nt that osses in a

fluence chillereek to reducewith boiling.ust 5% of therigerant filmarter of allof evaporatorresistance) dod or oil-free.often quotedant side heatfit is betweenng the wetting

rs garner highpressor itself,chiller reside.ee) may need

revent surge,ction), hot gasressing morepacity for the

14

t

s

r e . e

m l r o .

d t n g

h f, .

d , s e e

operby thspeepoingas signiimpr50%

Whilat colikelreduneedflowthe melegaand in mthe IstagiredulocatWarmcondwithCoolThe mechwithpoinconseveryenergchanless becochillwith

We counundemechin towate

16 IP

avbececoth

rating point. he ideal fan laed can be prnt without the

bypass or fificant comroving the c

% of the total l

le screw chilommon indusly much higheuces head andd for mechan

w recirculationmetric. The ant. Chillerscapacity con

multiple chilleIPLV load ping (on/off)

uced. Furthertions will nmer weathe

ditions that reh mechanical uler weather maverage of

hanical unlohout either.16

nts, the moresider the imyday occurrgy upgrades

nges in occupthan ideal

omes a great lers with purh such consiste

place great vnt on, not juser pressure. hanical equip

otality and in er plants.

PLV uses averagverage of operatelow the 75% IPentrifugal chillersonsistently, with he IPLV point.

Screw compaws and are nrecisely matce use of mecflow recirculmpression ecomponent thlosses in a chi

llers often prostry metrics lier in reality. d capacity innical unloaden at the four reality of acs operate at anditions. Moser plants that profile (and wof chillers armore, the we

not match ther results iequire centrifuunloaders and

may contributover compre

oaders is no The further

e significant mpact of acturences such such as lig

pancy or use,maintenanc

comfort thatre speed capency.

value on peost on their be

The sapment. Chillerecognition o

e operating pointing ±5ºF and ±12LV load point ca

s. In contrast, scthe numerical av

pressors are nnot subject toched to the chanical unloaation. This

efficiency ahat accounts iller.

ovide better ke IPLV, the The IPLV lo

n tandem, limers, hot gas br points that ctual operatioa wide varietst chillers areare unlikely

weighting) ds the buildineather profile

he IPLV assin the high

fugal chillers td their resultite to over comession and thot equal to r off of the IP

the impact. ual operationover sizing

ghting and i, varying weace and opet variable spepacity contro

ople and toolest day, but

ame can be ers should be of our imperfe

ts, but the nume2.5% capacity abn be 5.6% worsecrew chillers opeverage falling wit

not bound o surge, so

operating aders, hot

provides advantage,

for over

efficiency benefit is

oad profile miting the bypass or comprise

on is less ty of head e installed to follow

due to the ng load is es in most umptions. her head to operate ing losses. mpression. he use of operating PLV load

As we n due to

g chillers, insulation, ather, and ration, it eed screw ol operate

ls we can when put

said of evaluated

ect chilled

erical bove and e for erate thin 1.5% of

15

REFERENCES

1. AHRI Standard 550/590 or Performance Rating of

Water-Chilling and Heat Pump Water-Heating Packages Using the Vapor Compression Cycle (2011).

1. ASHRAE Research Project 751 (RP-751),

“Experimental Determination of the Effect of Oil on Heat Transfer with Refrigerants HCFC and HFC-123 and HFC-134a,” July 1999.

2. ASHRAE Research Project 601 (601-TRP), “Chemical

Analysis and Recycling of Used Refrigerants From Field Systems,” April 1990.

3. Brasz, J. J., "Aerodynamics of Rotatable Inlet Guide

Vanes for Centrifugal Compressors" (1996), International Compressor Engineering Conference. Paper 1196.

16

This paper is provided for informational and marketing purposes only and shall not be deemed to create any implied or express warranties or covenants with respect to the products of Carrier Corporation or those of any third party.

© Carrier Corporation 2015 www.carrier.com/commercial 04-581080-01 Printed in U.S.A. 5-15 Replaces: New