smart grid and renewable energy · the smart grid and renewable energy (online at scientific...

Smart Grid and Renewable Energy, 2010, 1, 63-107 Published Online August 2010 in SciRes (http://www.SciRP.org/journal/sgre/)

Copyright © 2010 SciRes. SGRE

TABLE OF CONTENTS

Volume 1 Number 2 August 2010

A Prepaid Smart Metering Scheme Based on WiMAX Prepaid Accounting Model

R. H. Khan, T. F. Aditi, V. Sreeram, H. H. C. Iu…………………………………………………………………………………63

Renewable Diesel Fuel from Processing of Vegetable Oil in Hydrotreatment Units:

Theoretical Compliance with European Directive 2009/28/EC and Ongoing Projects in Spain

D. Garraín, I. Herrera, C. Lago, Y. Lechón, R. Sáez…………………………………………………………………………………70

Transient Stability Analysis of Hu-Liao HVDC and AC Parallel Transmission System

P. Ye, Y. Q. Sui, Y. H. Yuan, X. M. Li, J. Q. Tao…………………………………………………………………………………74

The Initial Parameters Design of the Voltage Source Converter Fed SMES

X. D. Song, Z. Xu, T. Y. Xiang……………………………………………………………………………………………………81

Composite Cost Function Based Solution to the Unit Commitment Problem

S. Subramanian, R. Anandhakumar………………………………………………………………………………………………88

Separation of Biomass Pyrolysis Oil by Supercritical CO2 Extraction

J. H. Wang, H. Y. Cui, S. Q. Wei, S. P. Zhuo, L. H. Wang, Z. H. Li, W. M. Yi…………………………………………………98

Smart Grid and Renewable Energy (SGRE)

Journal Information

SUBSCRIPTIONS

The Smart Grid and Renewable Energy (Online at Scientific Research Publishing, www.SciRP.org) is published quarterly by

Scientific Research Publishing, Inc., USA.

Subscription rates: Print: $50 per issue.

To subscribe, please contact Journals Subscriptions Department, E-mail: [email protected]

SERVICES

Advertisements

Advertisement Sales Department, E-mail: [email protected]

Reprints (minimum quantity 100 copies)

Reprints Co-ordinator, Scientific Research Publishing, Inc., USA.

E-mail: [email protected]

COPYRIGHT

Copyright©2010 Scientific Research Publishing, Inc.

All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by

any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as described below, without the

permission in writing of the Publisher.

Copying of articles is not permitted except for personal and internal use, to the extent permitted by national copyright law, or under

the terms of a license issued by the national Reproduction Rights Organization.

Requests for permission for other kinds of copying, such as copying for general distribution, for advertising or promotional purposes,

for creating new collective works or for resale, and other enquiries should be addressed to the Publisher.

Statements and opinions expressed in the articles and communications are those of the individual contributors and not the statements

and opinion of Scientific Research Publishing, Inc. We assumes no responsibility or liability for any damage or injury to persons or

property arising out of the use of any materials, instructions, methods or ideas contained herein. We expressly disclaim any implied

warranties of merchantability or fitness for a particular purpose. If expert assistance is required, the services of a competent

professional person should be sought.

PRODUCTION INFORMATION

For manuscripts that have been accepted for publication, please contact:

E-mail: [email protected]

Smart Grid and Renewable Energy, 2010, 1, 63-69 doi:10.4236/sgre.2010.12010 Published Online August 2010 (http://www.SciRP.org/journal/sgre)


63

A Prepaid Smart Metering Scheme Based on WiMAX Prepaid Accounting Model Reduan H. Khan1, Tanzim F. Aditi1, Victor Sreeram2, Herbert H. C. Iu2

1Bangladesh University of Engineering & Technology (BUET), Dhaka, Bangladesh; 2University of Western Australia (UWA), Crawley, Australia. Email: [email protected] Received August 8th 2010; revised August 10th 2010; accepted August 12th 2010. ABSTRACT Prepaid energy meters have been widely adopted by utilities in different countries across the world as an innovative solution to the problem of affordability and consumption management. However, the present smart card based systems have some inherent problems like added cost, low availability and lack of security. In the future Smart Grid paradigm, use of smart meters can completely overhaul these prepaid systems by introducing centralized accounting, monitoring and credit-control functions using state-of-the-art telecommunication technologies like WiMAX. In this paper we pro-pose a prepaid smart metering scheme for Smart Grid application based on centralized authentication and charging using the WiMAX prepaid accounting model. We then discuss its specific application to Demand Response and Roam-ing of Electrical Vehicles. Keywords: Smart Metering, WiMAX, Smart Grid, Demand Response, Electric Vehicles

1. Introduction Over the last decade, prepaid energy metering has been considered as an effective tool to facilitate affordability and reduce the cost of electricity. Smart card based sol-id-state energy meters have already been introduced in many parts of the world-especially in the developing countries [1]. Almost all the prepaid energy meters available today are based on smart cards. However, the use of smart cards for utility payment systems has some major drawbacks like added cost, low availability, and risk of complete disconnection due to system failures that outweigh their benefits. In addition, Data storage in a remote meter for a long period poses a number of poten-tial threats like inception, tampering and other fraudulent activities [2].

With the advent of new Smart Grid applications like Demand Response and Plug-in Electric Vehicles, the utility operators need to profoundly change the way pre-paid energy systems operate by bringing telecommunica-tions to the core of their activities. Using a smart meter-ing infrastructure with a centralized authentication and charging system, a utility operator can eliminate most of the above mentioned drawbacks and increase the adop-tion of prepaid energy meters to a large extent. In addi-tion, prepaid energy metering through smart meters pro-vides both the customers and utility operators a wide

range of benefits: 1) Enables utility operators to optimize cash flow by

charging the user’s energy consumption in real-time and giving dynamic offers like discounts and bonuses.

2) Allows customers to purchase energy in convenient ways such as by using credit card online or just sending a SMS from their prepaid phone.

3) Allows the customers to switch from prepaid to postpaid mode seamlessly just by provisioning an unlim-ited prepaid quota.

4) Enables utility operators to buy electricity back (net metering) easily from renewable and distributed energy sources such as solar, wind, hydroelectric, biomass [3].

5) Allows customers to charge their Plug-in Electric Vehicles (PEVs) while roaming in another utility net-work.

6) Allows utility operators to execute various load control programs with the aid of Non-Intrusive Load Monitoring (NILM) techniques.

Being the first and only 4G broadband technologies available today, WiMAX has been emerged as the best candidate to build the next generation Smart Grid Com-munication network [4]. The open-standard technology platform of WiMAX supports continuous innovation and provides multi-vendor interoperability. In addition, the state-of-the-art authentication, authorization and ac-counting (AAA) capabilities of WiMAX makes it suit-



64

able for real-time smart metering. In this paper we pro-pose a prepaid smart metering scheme for Smart Grid based on centralized authentication and charging using the WiMAX prepaid accounting model [5]. We also pre-sent specific applications of the proposed scheme to De-mand Response and Roaming of Electrical Vehicles.

The rest of the paper is organized as follows. Section 2 describes the architecture, operation and key features of the WiMAX prepaid accounting model. Section 3 pre-sents a similar architecture for prepaid smart metering, provides a sample usage scenario, and then describes two additional applications—Demand Response and Roam-ing of PEVs-that the model can support. Finally, Section 4 provides conclusions and directions for future work.

2. The WiMAX Prepaid Accounting Model WiMAX accounting framework is based on RADIUS (Remote Authentication Dial In User Server) protocol [6]. RADIUS protocol generally supports offline (post-paid) accounting. To enable online (prepaid) accounting, a prepaid extension of RADIUS protocol has been released by WiMAX Forum in [7]. Although the model can sup-port a number of pricing schemes, for simplicity, we shall discuss only volume based charging scheme which is relevant to pricing of energy.

2.1 Architecture The architecture of the WiMAX prepaid accounting model is shown in Figure 1. Here, only the network enti-ties relevant to the prepaid charging model will be dis-cussed. A more detail description on the other WiMAX network entities can be found in [8].

A brief description of the WiMAX prepaid accounting entities are given below:

2.1.1 Network Access Server (NAS) The NAS is a generic RADIUS entity. It is the first AAA Client where AAA messages are originated and delivered to. In the prepaid accounting model, the Prepaid Client (PPC) functionality is implemented in the NAS. The PPC interacts with the Authentication, Authorization & Acco- unting Server (AAA-S) for dynamic charging information

Figure 1. WiMAX prepaid accounting model

using RADIUS protocol. The NAS along with PPC is physically implemented in the ASN (Access Service Network) Gateway of WiMAX ASN.

2.1.2 AAA Server/Proxy The AAA-S is typically responsible for authentication of the WiMAX Customer Premises Equipment (CPE). In the prepaid accounting model, the AAA-S communicates with the Prepaid Server (PPS) using RADIUS protocol in order to authorize services to the CPE and perform pre-paid accounting and charging.

The AAA proxy transparently routes AAA messages to the destination servers. In roaming scenario, the AAA proxy performs additional interworking functionalities between the home and visited network.

2.1.3 Prepaid Server (PPS) The PPS is located in the WiMAX Connectivity Service Network (CSN). It communicates with the AAA-S & NAS (PPC) for the purposes of authentication and au-thorization. In addition, The PPS also performs the fol-lowing functions:

1) Keeps the subscriber’s account balance (Balance manager);

2) Rates access service requests in real-time (Rating Engine);

3) Manages quota for a particular prepaid session (Quota Server).

The PPS checks the subscriber account before au-thorization. The rating entity in the PPS converts the credit into energy units, called the “quota”. This quota is then returned to the requesting PPC.

2.2 Operations The volume based prepaid model is built on an iterative authorization paradigm as the network does not have any prior knowledge about the consumption profile of the customer. In that case, the PPS will reserve a fraction of subscriber’s balance into a quota, each time an authoriza-tion request is made. The operational procedures of the system are discussed in the subsequent paragraphs.

2.2.1 Session Initialization When a WiMAX CPE attaches to the network, the NAS initiates authentication and authorization procedure by sending a RADIUS Access-Request message to the AAA-S as shown in Figure 2. The message contains standard RADIUS attributes, i.e., the base AVPs (Attrib-ute Value Pairs) along with a special attribute called Prepaid Accounting Capability (PPAC) that indicates the prepaid capabilities of the PPC. The PPAC is a Vendor Specific Attribute (VSA) which is specifically introduced for WiMAX Prepaid Accounting Model.

The PPS checks the subscriber account and authorizes the subscriber. During this procedure, the PPS takes into consideration the capabilities of PPC. Upon successful



65

Access Request

Access Request

Access Accept

Access Accept

Authentication

(Base AVPs, PPAC)

(Base AVPs, PPAC)

(Base AVPs, PPAQQID,VQ,VTH)

(Base AVPs, PPAQQID,VQ,VTH)

PPC AAA-S PPS

Accounting Request (Start)Accounting Request (Start)

MeterTurns ON

Metering Start

Quota Assigned

Figure 2. Message flow for prepaid session initialization

authorization, PPS generates an Access-Accept message containing another VSA called PPAQ (Pre-Paid Ac-counting Quota). The PPAQ attribute includes the fol-lowing information:

1) Quota ID (QID), which is set by the PPS to a unique value that is used to correlate subsequent quota requests.

2) Volume Quota (VQ), which is set to a value repre-senting a portion of the subscriber’s credit.

3) Volume Threshold (VTH) that indicates when the PPC should request additional quota.

4) RADIUS AVP Service Type is set to 6 which means “Authorize-Only”.

Upon receiving Access-Accept message from the PPS, the AAA-S appends the usual service attributes and for-wards the packets to the PPC. The PPC authorizes the service to the user and starts the metering session. The PPC sends an Accounting-Request (Start) packet to the PPS to indicate the start of the service. Once the Author-ize-Only Access-Request message is validated, the AAA-S forwards the Authorize-Only Access-Request to the ap-propriate PPS.

2.2.2 Mid Session Negotiation When the allocated volume threshold has been reached, the PPC sends an Access-Request message to the AAA-S with PPAQ attributes as shown in Figure 3.

The AAA-S validates the message after verifying the Message-Authenticator field. The PPS locates the prepaid

Figure 3. Message flow for mid-session negotiation

session state using the QID contained within the PPAQ. The PPS takes the most recently allocated quota and sub-tracts it from the user balance. If sufficient balance re-mains, the PPS allocates an additional quota (VQ) and calculates a new threshold value (VTH).

Upon successful re-authorization, the PPS generates an Access-Accept containing the PPAQ attribute. Upon re-ception of message, the PPC continues to provide service until the new threshold is reached. If the request for addi-tional quota cannot be fulfilled then the PPC lets the subscriber use the remaining quota and terminates the session.

2.2.3 Session Termination The termination phase is initiated when 1) the service is terminated, 2) the subscriber’s balance is exhausted, or 3) when the PPC receives an unsolicited Disconnect Mes-sage from the PPS. In these cases, the PPC sends an Au-thorize-Only Access-Request message with a PPAQ and Update-Reason attribute set to either “Client Service Termination” or “Remote Forced Disconnect”. Figure 4 shows the RADIUS message flow for session termination operation.

2.3 Additional Features 2.3.1 Tariff Switching The model supports tariff switching mechanism based on different time of a day. For example, as shown in Figure 5, peak rate is applied after 6 pm. The mechanism re-quires the PPC to report usage before and after the switch has occurred.

During authentication, the PPC indicates support for tariff switching by setting the appropriate bit in the PPAC. If the PPS needs to inform a tariff switch time, it will send the PTS (Prepaid Tariff Switching) VSA which indicates the point in time when the switch will occur.

Figure 4. Message flow for session termination

Figure 5. Example of tariff switching



66

PTS attribute contains sub-type Tariff Switch Interval (TSI) which indicates number of seconds from current time.

After the interval, the PPC sends another Access-Request message, as a result of either the session being turned off or the volume threshold being reached. The PPC reports how much volume was used in total (in a PPAQ attribute) and how much Volume was Used After the Tariff Switch (VUATS) using the VUATS subtype-attribute in PTS AVP.

2.3.2 Multiple Concurrent Services The model allows differentiation of services within a single session using separate quota for each service and allows for those quotas to be consumed at different rates.

As shown in the Figure 6, a session could be associ-ated with multiple (N) services. Each service is identified by a service identifier (Service-ID). The format of the Service-ID could be expressed as an IP flow using a ser-vice context containing Source-IP and Port, Destina-tion-IP and Port and protocol type. Port number can be allocated based on service type. PPS will provide a ser-vice ID for each connection inside a session and will allow a separate quota for each service.

2.3.3 Rating Group The model allows a wide range of pricing schemes using “Rating Group” concept that supports complex rating functions efficiently. Rating groups are typically config-ured at the PPC. As shown in Figure 7, Rating Groups are formed by a group of services from a particular user. However, unlike Figure 6, a single quota is allocated for the entire group.

During the usage of a service that is associated with a Rating Group, the PPC sends the ID of the Rating Group to the PPS. The PPS authorizes the Rating Group by al-locating a quota to it. When an additional service that belongs to an already authorized Rating Group is initi-ated, the PPC meters the service such that it draws from the already allocated quota. Therefore, no RADIUS messages need to be exchanged in this case. This limits the amount of traffic between the PPC and the PPS.

2.3.4 Support for Roaming The model supports roaming scenarios for prepaid charging. In roaming scenarios, the user is always authenticated

Figure 6. Multiple services within a single session

Figure 7. Multiple services with a single quota

by its home network. Authorization for the prepaid ses-sion and quota replenishing occurs at the home network as well.

3. The Proposed Smart Metering Scheme A similar model, based on the volume based charging scheme of WiMAX prepaid accounting model, can be used for smart-metering. Although there are a lot of dif-ferences between a telecom device and a smart-meter, the fundamentals of real-time charging are same. The same iterative charging model can be used, except the fact that, here the subscriber is getting energy service instead of multimedia services.

3.1 Architectural Model Figure 8 shows the architectural model of the proposed scheme. Unlike WiMAX CPE, the metering session is invoked by a WiMAX enabled smart meter in the pro-posed scheme. Hence, the proposed model is composed of the three logical entities–Smart Meter, ASN and CSN- in accordance with the WiMAX prepaid accounting ar-chitecture described in Section 2. A brief description of each entity is given below.

3.1.1 Smart Meter The smart meter calculates instantaneous power con-sumption of the connected load at regular intervals and transmits the data to the nearest WiMAX Base Station (BS). The block diagram in Figure 9 provides a simpli-fied re-presentation of a smart metering circuitry.

As shown in the diagram, the smart meter calculates instantaneous voltage (V) and current (I) of the load us-ing a voltage and current transformer. From V & I, real power consumption is calculated by a wave-form analy-sis algorithm using a microcontroller. Then the analogue information is converted into digital values by an ADC

Figure 8. Prepaid charging model for smart metering

Figure 9. Block diagram of a simple smart metering circuit



67

(Analogue to Digital Converter). The digital value is embedded in a metering protocol over the IEEE 802.16 PHY & MAC PDUs (Protocol Data Unit) by a WiMAX chip and transmitted through its built-in micro-antenna.

To ensure network security, the smart-meter can be periodically authenticated by the AAA server in WiMAX CSN. In addition, the information exchange between Smart Meter and the Core Network can be encrypted by the shared encryption keys received during authentica-tion.

3.1.2 Access Service Network (ASN) The WiMAX BS in the ASN extracts the metering pro-tocol from the IEEE 802.16 PDUs and tunnels them to the ASN-GW over the IP interface using a standard relay protocol. The ASN-GW extracts the power consumption information from the metering protocol and provides it to the PPC.

The PPC in the ASN-GW acts as the charging broker between the Smart Meter & the Prepaid Server. The PPC must have the capability to analyze the metering infor-mation sent by the Smart Meters periodically and convert them into energy units in KWH. The PPC tracks the en-ergy consumption of the meter and requests for a new quota to the PPS once the previous one has been con-sumed.

3.1.3 Connectivity Service Network (CSN) The main entity in CSN for prepaid smart-metering is the PPS. The PPS updates the account balance based on en-ergy consumption and tariff profile. In addition, it per-forms additional pricing and load control activities in conjunction with the backend infrastructures. All the communication between PPC and PPS is done via one or more AAA servers/Proxies.

The PPS is connected with several backend infra-structures such as, voucher server for account recharging, billing server for customer invoicing, provisioning sys-tem and other application servers for energy management and consumption profiling. A firewall is used to separate the network domains. Figure 10 illustrates the backend connectivity of PPS.

3.2 Sample Usage Scenario To illustrate the functionality of the proposed scheme, we describe a sample usage scenario with the message flows shown in Figure 11:

1) After turning ON, the meter sends a metering re-quest to the PPC using the WiMAX radio network.

2) The PPC opens a session for the meter and sends a RADIUS Access-Request message to the PPS via the AAA.

3) The AAA server authenticates the user, authorizes the service and forwards the Access-Request to the PPS.

4) The PPS verifies the account balance of the user and reserves money equivalent to the initial quota of 5 KWH.

HTTP

TCP/IP

Figure 10. Connectivity of PPS with the backend infra-structures

Figure 11. Message flow for the proposed charging model

Then it sends an Access-Accept message to the PPS with a Volume-Quota (VQ) of 5 KWH and Volume-Threshold (VTH) indication of 4.5 KWH. It further associates the Quota Identifier (QID) = YYYY to this quota reserva-tion.

5) Upon reception of message, the PPC continues the user session and meters the energy consumption periodi-cally.

6) When the allowed 4.5 KWH energy has been con-sumed by the user, the PPC generates an Access-Request that contains the amount of consumed quota, and the request for replenishment.

7) Upon reception of message, the PPS identifies the user from the QID and PPS subtracts the reserved money from the account. Then it sends an Access Accept mes-sage to the PPS with a new VQ of 5 KWH, VTH of 4.5



68

KWH and a new QID as ZZZZ. 8) Upon reception of message, the PPC updates its re-

cords and continues provisioning energy access to the user. The procedure runs iteratively until the prepaid ac-count balance of user has been used up or the energy consumption has been stopped.

9) In case, the meter is turned off or a re-authentication procedure is initiated, the PPC sends an Access-Request message with the reason “Client Service Termination” and specifies the amount of energy consumed from the allocated quota. The PPS subtracts the portion of the re-served money equivalent to consumed energy and reim-burses the balance.

3.3 Applications In addition to the real-time prepaid charging, the pro-posed scheme can support the following two major smart grid applications using the additional features of Wi-MAX prepaid accounting model described in Section 2.

3.3.1 Demand Response Demand response enables the utility operator to opti-mally balance between power generation and consump-tion either by offering dynamic pricing or by implement-ing various load control programs [9]. Using the pro-posed scheme that supports tariff switching, charging of multiple services in a single session, rating groups, utility operators can perform various demand response pro-grams effectively and efficiently with minimum addi-tional costs.

For example, during peak hours, the utility company can charge a premium rate for the non-essential appli-ances such as electric heater, air-conditioner and washing machine using the tariff switching functionality. Fur-thermore, the rating group feature of the proposed scheme allows the utility operator to offer flexible pricing schemes to its customers and thus maximize their profit. For example, a utility operator may wish to rate energy such that the first N KWHs are free, then the next M KWHs are rated at $5 per KWH and volume above (N + M) KWH be rated at $6 per KWH.

Demand Response using automatic load control can also be performed using an “Unsolicited Session Termi-nation” operation on a particular group of loads and thus reduce the load from the grid. The differentiation of loads can be performed by using non-intrusive load monitoring techniques as described in [10]. In this case, as shown in Figure 12, the PPS should dynamically in-teract with a SCADA server located in the WiMAX CSN which will manage and implement the load-control pro-grams via the WiMAX ASN.

3.3.2 Roaming of PEVs PEVs are regular hybrid vehicles that have a large high- capacity battery bank. The batteries on these vehicles can take power from the grid during off-peak hours to charge

the batteries and provide power back to the grid during peak hours. Thus they can play a key role in balancing power generation and consumption of the grid [11]. Since the PEVs are highly mobile, integrating them into the Smart Grid can be a major challenge.

The proposed scheme enables utility operators to allow flexible charging and de-charging of PEVs in any utility network using just an on-board smart meter. Although PEVs are mobile, the charging stations are static. There-fore, no mobile communication network is needed. Fig-ure 13 shows the roaming scenario of a PEV with an on-board WiMAX smart meter.

The charging station is located at the visited WiMAX network. The smart meter in the PEV authenticates itself and sends metering information to the AAA and Prepaid Server of its home network. The communication between home and visited network is done via the AAA proxy servers located at the edge of each network. Here, the visited network merely provides network access to the PEV.

4. Conclusions Prepaid charging of energy has the potential to provide both customers and utility operators a wide range of ben-efits including consumption regulation, load management and integration of distributed energy resources. In this paper, we have provided a detail study of WiMAX pre-paid accounting model along with its key additional fea-tures. This study is then used to present a similar prepaid charging scheme for smart metering to be used in Smart Grid. Specific applications of the proposed scheme

Figure 12. Automatic load control using the prepaid model

Figure 13. Prepaid Roaming of Electric Vehicles



69

to Demand Response and Roaming of Electrical Vehicles are also discussed.

The proposed scheme makes use of existing WiMAX core infrastructure as much as possible while providing secure charging using the state-of-the-art security capa-bilities of WiMAX network. Since the scheme is based on a telecom framework, it requires a separate protocol to transfer the metering information from smart meters to the WiMAX core network. In addition, the scheme needs specific guidelines for data storage, session maintenance, authentication, quota allocation, and handling of outage and quality related events.

The continuation of the research will include devel-opment of necessary protocols and guidelines for support-ing various metering, control, and pricing applications on the proposed scheme. An experimental examination on the performance of the proposed scheme is also worth further investigation.

REFERENCES [1] A. A. Casarin and L. Nicollier, “Prepaid Meters in Elec-

tricity. A Cost-Benefit Analysis,” IAE Business School, Austral University, March 2008. http://www.sandiego.edu

[2] P. McDaniel and S. W. Smith, “Security and Privacy Challenges in the Smart Grid,” IEEE Computer and Re-liability Societies, Los Alamitos, 2009

[3] Whitepaper, “Electricity Smart Metering: Business Driv-ers,” Atos Origin, November 2009.

http://www.fr.atosorigin.com/ [4] Whitepaper, “WiMAX Applications for Utilities,” Wi-

MAX Forum, October 2008. http://www.senza-fili.com/ [5] “WiMAX End-to-End Network Systems Architecture-

Stage 2: Architecture Tenets, Reference Model and Ref-erence Points,” WiMAX Forum, November 2005.

[6] C. Rigney, A. Rubens, W. Simpson and S. Willens, “Re-mote Authentication Dial In User Server (RADIUS),” RFC 2856, Internet Engineering Task Force, June 2000.

[7] “WiMAX End-to-End Network Systems Architecture- Stage 3: Detailed Protocols & Procedures, Annex: Pre-paid Accounting,” WiMAX Forum, February 2009.

[8] Whitepaper, “Mobile WiMAX–Part I: A Technical Over-view and Performance Evaluation,” WiMAX Forum, March, 2006. http://www.wintegra.com/

[9] S. Braithwait and K. Eakin, “The Role of Demand Re-sponse in Electric Power Market,” Edison Electric Insti-tute, October 2002. http://www.smartgridnews.com

[10] C. Laughman, K. Lee and R. Cox, “Power Signature Analysis,” IEEE Power and Energy Magazine, Vol. 1, No. 2, 2003, pp. 56-63.

[11] A. Brooks, T. Gage and A. C. Propulsion, “Integration of Electric Drive Vehicles with the Electric Power Grid—A New Value Stream,” 18th International Electric Vehicle Symposium and Exhibition, Berlin, October 2001. http://www.smartgridnews.com



Renewable Diesel Fuel from Processing of Vegetable Oil in Hydrotreatment Units: Theoretical Compliance with European Directive 2009/28/EC and Ongoing Projects in Spain

Daniel Garraín, Israel Herrera, Carmen Lago, Yolanda Lechón, Rosa Sáez

Spanish Ministry of Science and Innovation, CIEMAT (Public Research Centre on Energy, Environment and Technologies), Energy Department, Energy Systems Analysis Unit, Madrid, Spain. Email: [email protected] Received August 2nd 2010; revised August 5th 2010; accepted August 10th 2010.

ABSTRACT

Oil hydrotreating units in refineries are aimed at reducing the sulfur content of fuels to accomplish standard particular specifications. However, this process is currently one of the best available technologies to produce biofuels from vege-table oil in a refinery. Vegetable oils can be processed or co-processed in these units if several adaptations are per-formed, so some properties could be improved in comparison with conventional fuel such as density and cetane number. This study highlights the theoretical greenhouse gases (GHG) emissions (using a life cycle assessment–LCA-approach) of a hydrotreated vegetable oil (HVO) from bibliographical data. Results were compared with other biofuel production processes, such as those obtained by transesterification of vegetable oil (FAME, fatty acid methyl ester). It has also been included the comparison with conventional fossil diesel as a benchmark in order to assess the theoretical compli-ance with GHG savings proposed in European Directive 2009/28/EC. Finally, ongoing projects and future perspectives in Spain are mentioned. Keywords: Hydrotreated Vegetable Oil, Fatty Acid Methyl Ester, GHG Emissions Savings, Directive 2009/28/EC,

Renewable Diesel Fuel

1. Introduction

Oil hydrotreating units in refineries are aimed at reducing the sulphur content of fuels to accomplish standard par-ticular specifications. However, hydrotreating of vegeta-ble oils or animal fats is an alternative process to esteri-fication for producing biobased diesel fuels. This practice is a modern way to produce very high-quality biobased diesel fuels without compromising fuel logistics, engines, exhaust aftertreatment devices, or exhaust emissions. These fuels (hydrotreated vegetable oils–HVO-) are now also referred to as ‘renewable diesel fuels’ instead of esterificated ‘biodiesel’ which is reserved for the fatty acid methyl esters (FAME) [1].

Selected FAME and HVO properties have been com-pared in Table 1 since they represent two different ap-proaches for making diesel fuel from vegetable oil. HVO has excellent diesel fuel properties including an ex-tremely high cetane number (measure of a diesel fuel’s

ignition delay). FAME has lower heating value (LHV) because of its oxygen content and also has other unde-sirable properties such as high density, and high NOx emissions. Overall, HVO appears to be a superior prod-uct [2].

This paper focuses on the greenhouse gas (GHG) emissions study of a theoretical HVO process, from bib-liographical data, to identify problematic stages in the production chain in order to reduce environmental im-pacts. For this evaluation, life cycle assessment (LCA) methodology was the approach chosen to calculate the GHG emissions profile associated with the production of this new renewable diesel fuel.

ISO 14040:2006 and ISO 14044:2006 standards [4,5] define LCA as a methodology for the comprehensive assessment of the impact that a product or process has on the environment throughout its life span (from extraction of raw materials through manufacturing, logistics and use to scrapping and recycling, if any), which is known as a



71

Table 1. Comparison of FAME and HVO properties [1-3]

Properties FAME HVO

% Oxygen 11 0

Density (15ºC) (g/ml) 0.883 to 0.885 0.775 to 0.780

Viscosity (40º) (mm2/s) 4.5 2.5 to 3.5

Cloud point (ºC) –5 to 0 –5 to –30

Sulphur content < 10 ppm < 10 ppm

LHV (MJ/kg) 37.5 to 38 44

Storage stability Very challenging Good

Cetane number 50 to 65 80 to 99

% change in NOx emission +10 0 to –10

“from cradle-to-grave” analysis.

2. Goal and Scope

The first step in a LCA is to define the scope and goals of the study. This work aims to assess the theoretical pro-duction of HVO from soybean under an LCA framework focused on the global warming impact category in order to calculate GHG emissions. Counterbalanced biblio-graphical data and previous own revisions were selected to study the process in detail. Results will be compared with other biofuel production processes, such as those obtained by transesterification of vegetable oil (FAME, fatty acid methyl ester). It will also be included the com-parison with conventional fossil diesel as a benchmark in order to assess the theoretical compliance with GHG savings proposed in European Directive 2009/28/EC.

The scope of this assessment is cradle-to-grave, from

acquisition of the raw materials in agricultural labours through the production of HVO in refinery to the final combustion. Stages taken into account are detailed in Table 2 of the following Section 3.

According to ISO 14040:2006 standard, the functional unit provides a reference to which inputs and outputs are normalised. In this case study, the amount of fuel ex-pressed in energy units (MJ, Megajoules) was consid-ered.

3. Life Cycle Inventory

3.1 Data Collection

Life cycle inventory (LCI) phase involves data collection and modeling of the product system, as well as descrip-tion and verification of data. This encompasses all data related to environmental and technical quantities for all relevant unit processes within the study boundaries that compose the product system. The procedure for LCI of the HVO and FAME productions are summarised in Ta-ble 2.

Inventory data for those energy and material inputs were obtained from eco-profiles within SimaPro7.1 software [11], representing average production in a European context.

3.2 Allocation Procedures

According to ISO 14044:2006 standard [5], allocation process is defining as the partitioning of the input/outputs flows of a process to the system product under study. The allocation procedure in a multi-product process is the most critical issue in LCA, so it is recommended avoiding allocation whenever possible either through subdivi- sion of certain processes or by expanding the system limits to include the additional functions related to them. Where allocation cannot be avoided, the environmental loads could be assigned allocated into two or more sub

Table 2. Procedure of LCI of HVO and FAME production and data sources

Phase (product) Main input data Main output data Data source

Seed farming (HVO and FAME) Pesticides, fertilizers, fuel – [6]

Oil extraction (HVO and FAME) Hexane, electricity, natural gas Soybean meal [6]

Oil refining (HVO and FAME) Caustic soda, aluminium sulphate, ammonia nitrate, bentonite, electricity, natural gas

Soap pulp (waste) [6]

Oil transesterification (FAME) Chloride acid, catalyser, methanol, electricity, natural gas Glycerol Adapt.from [7]

Oil hydrotreating (HVO) Hydrogen, electricity, steam Naphta, ateam, electricity [8]

Transports (HVO and FAME) Distances, type of transport – [6,9,10]

Combustion (HVO and FAME) – – [10]



72

processes or expanding the product system to include the additional functions related to co products. Oil extraction, transesterification and hydrotreating phases yield more than one product, therefore, allocation is necessary. En-ergy allocation was considered in the case because of the purpose of the main product.

4. Results and Discussion

4.1 FAME vs HVO

LCA was conducted by means of SimaPro v.7.1 software [11], using characterization factors from CML 2 baseline 2000 methodology [12]. Low heating values used has been 44.0 MJ/kg and 37.2 MJ/kg, for HVO and FAME respectively. Table 3 shows the GHG emissions of both products by stage of the whole life cycle.

GHG emissions in the combustion phase are consid-ered null due to the biogenic origin of the carbon in the plant. Total values of the table show that theoretical en-vironmental benefits are achieved from the processing of vegetable oil with hydrogen against transesterification process to obtain biofuels.

4.2 Compliance with European Directive 2009/28/EC

European Directive 2009/28/EC establishes a common framework for the promotion of energy from renewable sources. It sets mandatory national targets for the overall share of energy from renewable sources in gross final consumption of energy and for the share of energy from renewable sources in transport. Article 17 refers to the sustainability criteria for biofuels and bioliquids, high-lighting that the GHG emission saving from the use of biofuels and bioliquids shall be at least 35%. With effect from January 1st 2017, that saving shall be at least 50%, and from January 1st 2018, shall be at least 60% for bio-fuels and bioliquids produced in installations in which production started on or after January 1st 2017. For bio-fuels, for the purposes of the calculation referred to GHG savings, the fossil fuel comparator emissions shall be the latest available actual average emissions from the fossil part of petrol and diesel consumed in the Community as reported under Directive 98/70/EC. If no such data are available, the value used shall be 83.8 g CO2 eq/MJ [13]. Table 4 illustrates the GHG savings for both biofuels using the previous default value for conventional diesel.

These values show that policy objectives can be achieved when theoretical data of the processes are taken into account. This gives an idea of the real possibility of compliance with the Directive. Nevertheless, in order to certify a new biofuel in terms of reducing GHG emis-sions and, therefore, compliance with the Directive, original data of the whole products and processes in-volved must be submitted. In case that actual data source

Table 3. GHG emissions by stage for HVO and FAME (g CO2 eq/MJ)

Phase FAME HVO

Seed farming 5.87 4.96

Oil extraction 7.73 6.54

Oil refining 0.40 0.34

Oil transesterification 5.05 –

Oil hydrotreating – 4.48

Transports 4.18 3.98

Combustion 0 0

TOTAL 23.23 20.30

Table 4. Percentage of GHG savings for FAME and HVO

FAME HVO

% GHG savings 72.30 75.78

is not available, the Directive states typical and default values for agricultural and processing systems.

5. Conclusions

This work focused on the theoretical environmental per-formance comparison of two different processes to obtain biofuels from vegetable oil: hydrotreatment versus trans-esterification. The products obtained in each case are called HVO and FAME, respectively. It has been shown that the hydrotreating process could achieve good envi-ronmental performance in terms of GHG emissions, from theoretical data. Furthermore, the possibility of produc-ing HVO is feasible and it does not require the construc-tion of any new infrastructure, because it could be manufactured in oil refineries. However, optimization operations of the hydrotreatment unit should be carried out, since the chemical reaction forms CO and CO2 gases that could damage the catalyser. Moreover, a thermal balance should be done in the furnace due to the exo-thermic performance of the hydrogenation.

Finally, a theoretical compliance with European Di-rective 2009/28/EC has been shown due to the percent-age of GHG savings in the entire life cycle of the biofu-els. Nonetheless, original data of the whole products and processes involved are necessary in order to obtain the final policy certification.

6. Perspectives and Future Works

The first commercial scale HVO plant with a capacity of



73

170 Mtons per year (3800 bbl per day) was started up in 2007 at Neste Oil’s Porvoo oil refinery in Finland. This technology is based on a separate unit at an oil refinery site while at the same time using existing logistics, quality-control laboratories, and energy plant. A separate unit like this can be optimized and run without risking the refinery units, which may be a problem if bio-oils are fed into existing refinery units as blended with fossil feeds [2].

Currently, Spanish oil company Repsol is developing some projects to demonstrate the technical feasibility of the production of biofuels at an industrial scale, through co-processing of vegetable oils with fossil fuel in hydrotreatment units. A theoretical study from Garraín et al. [14] has been shown that co-processing would be feasible from the environmental point of view against other biofuel production processes, such as transesterification or separately hydrotreatment.

Two first industrial-scale demonstrations have been completed. The first, as indicated in the oil company, was positive, since the product obtained has suitable properties, including high-cetane, low density and ultra- low sulphur. Due to these sterling properties, the ob-tained product could be fitted on the production of diesel, complementing the addition of seven percent of biodiesel that is currently allowed to use in conventional diesel. Second test was better because of optimizing operational factors and energy consumption associated with the pro-duction. The new product had identical properties as the first renewable diesel.

At present (July 2010), Energy System Analysis Unit from CIEMAT is working on calculating the environ-mental impacts of these new products, in order to reveal the ‘green’ benefit over other biofuels and their corre-sponding production processes.

REFERENCES [1] H. Aatola, M. Larmi, T. Sarjovaara and S. Mikkonen,

“Hydrotreated Vegetable oil (HVO) as a Renewable Die-sel fuel: Trade-Off between NOx, Particulate Emission, and Fuel Consumption of Heavy Duty Engine,” Helsinki University of Technology & Neste Oil, Finland, 2008.

[2] R. Marinangeli, T. Marker, J. Petri, T. Kalnes, M. McCall, D. Mackowiak, B. Jerosky, B. Reagan, L. Nemeth, M. Krawczyk, S. Czernik, D. Elliott and D. Shonnard, “Op-portunities for Biorenewables in Oil Refineries,” Final Technical Report DE-FG36-05GO15085 from UOP LLC to U.S. Department of Energy, Des Plaines, 2005.

[3] T. Kalnes, T. Marker and D. R. Shonnard, “Green Diesel: A Second Generation Biofuel,” International Journal of

Chemical Reactor Engineering, Vol. 5, A48, 2007.

[4] ISO 14040:2006 Environmental management-Life cycle assessment - Principles and framework.

[5] ISO 14044:2006 Environmental management-Life cycle assessment-Requirements and guidelines.

[6] J. A. Hilbert, L. B. Donato, J. Muzio and I. Huerga, “Comparative Analysis of Energetic Consumption and Greenhouse Gas Emissions from the Production of Bio-diesel from Soy under Conventional and no Till Farming Systems,” Boletín nº 6, Doc N° IIR-BC-INF-06-09, INTA, Argentina, 2009.

[7] Y. Lechón, H. Cabal, C. de la Rúa, C. Lago, L. Izquierdo, R. Sáez and M. Fernández, “Análisis del ciclo de vida de combustibles alternativos para el transporte. Fase II: Análisis del ciclo de vida comparativo del biodiésel y del diésel,” Centro de Publicaciones–Secretaría Gral. Técnica– Ministerio de Medio Ambiente, Madrid, 2006.

[8] G. Reinhardt, S. O. Gärtner, H. Helms and N. Retten- maier, “An Assessment of Energy and Greenhouse Gases of NExBTL,” Final Report from Institute for Energy and Environmental Research Heidelberg GmbH (ifeu) by or-der of the Neste Oil Corporation (Porvoo, Finland), Hei-delberg, 2006.

[9] R. Dones, C. Bauer, R. Bollinger, B. Burger, M. Faist Emmenegger, R. Frischknecht, T. Heck, N. Jungbluth, A. Röder and M. Tuschsmid, “Life Cycle Inventories of Energy Systems: Results for Current Systems in Switzer-land and other UCTE Countries,” Ecoinvent report No. 5, Paul Scherrer Institut Villigen, Swiss Centre for Life Cycle Inventories, Dübendorf, Switzerland, 2007.

[10] JEC, “Well-to-Wheels Analysis of Future Automotive Fuels and Powertrains in the European Context, Well-to- Wheels Report, Version 2c,” CONCAWE/EUCAR/EC- JRC-IES Report, 2007.

[11] PRé “SimaPro ® 7.1”, PRé Consultants, 2007, Amers- foort, The Netherlands. http://www.pre.nl

[12] J. B. Guinée, “Handbook of Life Cycle Assessment– Operational Guide to the ISO Standards,” Kluwer Academic Publishers, Dordrecht, 2002.

[13] Directive 2009/28/EC of the European Parliament and the Council of 23 April 2009 on the promotion of the use of energy from renewable sources and amending and subse-quently repealing Directives 2001/77/EC and 2003/ 30/EC, 05/06/2009, Official Journal of the European Union.

[14] D. Garraín, I. Herrera, C. Lago, Y. Lechón and R. Sáez, “Viabilidad Medioambiental del Co-Procesamiento de Aceites Vegetales en Unidades de Hidrotratamiento Para Obtener Biocarburantes en el Marco del ACV,” XIV In- ternational Conference on Project Engineering, Madrid, 2010.



Transient Stability Analysis of Hu-Liao HVDC and AC Parallel Transmission System*

Peng Ye1, Yuqiu Sui1, Yinghua Yuan1, Xiaoming Li1, Jiaqi Tao2

1Northeast Electric Power Research Institute Co. Ltd Shenyang, China; 2Northeast China Grid Company Limited Shenyang, China. Email: [email protected] Received June 17th 2010; revised July 10th 2010; accepted July 17th 2010.

ABSTRACT

In this paper, transient stability analysis was focused on Hu-Liao HVDC and AC parallel transmission system. The Hu-Liao HVDC project was introduced; Simulation method and mathematic models of AC and DC systems were stud-ied as well as corresponding regulators and controllers. The dynamic performance and the interaction between AC and DC systems during serious disturbance were researched by detail time-domain simulation. Comparison was also made under different operation schemes. The research will bring important and significant reference for further operation and stability control of Hu-Liao HVDC and AC system. Keywords: HVDC, Transient Stability, Control Strategy, Time-Domain Simulation

1. Introduction

Since the first High Voltage Direct Current transmission project was commissioned into commercial operation in 1954, HVDC has been developed so rapidly that it has been widely applied in such fields as large power trans-mission over long distance, interconnecting two asyn-chronous systems, power transmission through subma-rine cables for supplying power to islands and so on. Compared with three-phase AC transmission systems, conventional HVDC is superior in the following aspects [1]: Firstly, HVDC need less cost in constructing and operating; Secondly, it needs not keep operating syn-chronously between the two AC systems; Thirdly, it is easy to control and adjust power flow, etc.

Among the many HVDC long transmission schemes around the world, very few operate in parallel to AC transmission of comparable capacity. Problems for par-allel AC/DC operation is primarily related with the coor-dination between AC and DC power flows and how each system reacts to any disturbance [2]. It is well known that AC transmission systems have the inherent means to reschedule their power flows and to provide timely and sufficient synchronizing torque to secure such flows fol-lowing disturbances such as AC faults, load rejection or generator tripping, etc. How a HVDC in parallel to AC system reacts in those situations has always been a cen-

tral question, particularly for planning and daily opera-tion of such a complex scheme. In reference [3-6], the interaction action between AC and DC parallel transmis-sion system were studied, the theory and operation rules of such power system were demonstrated with simulation examples. Results show that HVDC schemes in parallel operation with AC transmission are prone to both tran-sient swing angle and voltage instabilities. And the risks of instability will increase during disturbances. In refer-ence [7-9], research on advance control strategy for a HVDC scheme in parallel operation with AC systems was discussed. By these unconventional control strate-gies, the HVDC scheme can actively participate in the instantaneous rescheduling of power and improve the dynamic performance of power network. In reference [10,11], a real AC and DC parallel transmission system in South China was studied from operation and control aspects.

The HuLunbeier-Liaoning (abbreviated as Hu-Liao hereinafter) HVDC project was the first AC/DC parallel transmission system in North China, which was em-ployed to transfer electricity from Hulunbeier energy center to Liaoning province. In particular, in the sending side of HVDC, Hulunbeier has a very weak network. As a result, the stability problem in operation is very critical. It is a challenge work for the operator to keep such a special AC/DC parallel transmission system operating in an economic and secure state.

In this paper, transient stability analysis was focused

*This paper is supported by key project of State Grid Cooperation of China: “Dynamic behavior and coordination control of AC and DC transmission system in northeast electric network”



75

on the Hu-Liao HVDC and AC parallel transmission system. The Hu-Liao HVDC project was introduced; Simulation method and mathematic models of AC and DC systems were studied as well as corresponding regu-lators and controllers. The dynamic performance and the interaction between AC and DC systems during serious disturbance were researched by detail simulation. Com-parison was also made under different operation schemes. The research will bring important and significant refer-ence for further operation and stability control of Hu- Liao HVDC and AC system.

2. Hu-Liao HVDC and AC Parallel Transmission System

Northeast China Electric Power Network System consists of four provincial networks, Liaoning, Jilin, Heilongjiang and East of Inner Mongolia. These networks have been linked with 500 kV transmission lines. And Liaoning electric network is connected with Huabei network by back-to-back HVDC links with a rated capacity of 1500 MW. Figure 1 shows the schematic of Hu-Liao HVDC and AC parallel transmission system.

Hulunbeier is rich in coal, large scale power plants are under constructing. Liaoning province is a rapid devel-oping province in economic but much lack of energy. So based on the mutual interest and the idea of optimizing resources, Hu-Liao HVDC project is put into operation.

Liaoning network

Heilongjiang network

Fengtun TCSC

Mujia

Gaoling

Huabei network

Jiangjiaying

Hu-Liao HVDC

Bayantuohai

Yimin

Hulunbeier Energy(2 × 60)

Ewenke (2 × 60)

Yimin 1, 2 (2 × 50, 2 × 60)

Yimin 3 (2 × 60)

Jilin network

Figure 1. Schematic of Hu-Liao HVDC and AC parallel transmission system

Hu-Liao HVDC transmission is a bipolar 12-pulse HVDC transmission system with rated DC voltage ± 500 kV, rated power 3000 MW, rated current 2500 A. Over-head lines have a length of 908 km long. Yimin converter station locates at east of Inner Mongolia, which is 10 km away from Yimin power plants. It acts normally as a rec-tifier and its AC side rated voltage is 500 kV. Mujia con-verter station is in the center of Liaoning province and connected with Anshan 500 kV station with two 21 km lines. It acts principally as an inverter and its AC side rated voltage is also 500 kV. The Hu-Liao HVDC trans-mission system operation modes include bipolar mode, monopolar ground return mode, monopolar metallic re-turn model and monopolar parallel line ground return model. The HVDC system can be operated under rated voltage and lower voltage. The Hu-Liao HVDC trans-mission normally operates in P mode (constant power control mode). I mode (constant current control mode) can be used as a back-up mode.

As mentioned above, large scale power plants are un-der construction in Hulunbeier. Before Hu-Liao HVDC implemented, there are two power plants, namely: Yimin 1 and 2, with altogether capacity of 2200 MW. Electric power is transferred to the west of Heilongjiang province through two 500 kV lines. At Fengtun station, Thyristor Controlled Series Compensation (TCSC) is installed to improve the transfer capability. Up to now, three new power plants, Yimin 3, Hulunber Energy and Ewenke, are set up. The electric power capacity is 3600 MW and they are mainly transferred to Liaoning province by Hu-Liao HVDC system.

There are disconnecting switches between the buses of Yimin 1, 2 and Yimin 3 power plants. When they are closed, Hu-Liao HVDC and AC system are operating in parallel. There is power exchanging through AC and DC system. Generally, when AC and DC system is operated in parallel, the fault occurred in either AC or DC system would lead to instantaneous or permanent power imbal-ance in power system, and quantity of power will shift through AC and DC system, which would be a great im-pact to transient stability. It is a challenge for system operation, especially due to such a weak delivering side.

3. Mathematic Models and Criterian in Simulation

The electromechanical transient simulation of this AC and DC hybrid system was made through PSASP. Northeast electric power network data include about 1500 buses, 330 generators and two groups of DC lines. Only 500 kV and 220 kV voltage network are considered in calculation. As PSASP offered adequate mathematics models for each type of electric elements, a majority of them are defined by PSASP according to the requirement of the simulation. Individuals are developed through user program interface such as DC and TCSC models and



76

controllers. The details are as follows:

3.1 Models for AC System

For most of generators, five-order model is adopted, in which the variations of Eq˝, Ed˝ and Eq΄ are considered. They are fit for detail simulation of salient pole syn-chronous generators. As for the individual small hydroe-lectric equipment, two-order model is adopted, in which it is approximately hold that Eq΄ could keep constant. Most of wind turbines are modeled as doubly-fed di-rect-drive wind power generators. The models of the corresponding regulators such as excitation system, speed control system and PSS, are selected and defined in the software according to the practical case.

For system load, it is described as combination of fifty percent constant impedance and fifty percent induction motor during dynamic simulation. Three-order model of induction motor is used here to simulate the dynamic features of loads. The parameter of stator leakage reac-tance is selected as 0.18 pu.

The electric distance is relatively near between TCSC and DC lines. The Yi-Feng TCSC is composed of two parts: fixed part and variable part. The fixed part occu-pies thirty percent of the total transmission line capacity and the variable part occupies fifteen percent. In the dy-namic process, the TCSC acts as follows: The fixed part is sure not to be bypassed when fault occurred in trans-mission lines; The variable part is to be bypassed when three or two phase fault occurred in the line; while when single phase fault occurred, the fault phase is bypassed and the forced compensation will take action in normal phase; The forced compensation will also take action when fault occurred in neighbor line. The bypass time of TCSC is 0.05 s after fault occurring. The control logic of the variable part is shown in Figure 2. Where, P and V are measured power and voltage, which is used for oscil-lation control; ttrigger is the signal of beginning time and Tforced is the continuous time of the action. The maximum compensation capacity is forty-five percent.

Oscillation controller

∑

PL0

PL

xTCSC

+

_

max

ttrigger

Tfored

∑

∑

v0

v

+ _

∑

protectioncontrollor

xTCSC

Inertia and

limter

xTCSC

ttrigger

Tfored Bypass

controllorTransient stability

controllor

Figure 2. Control scheme of TCSC

3.2 Models for DC System

The DC model used in steady state calculation is shown as the following equation, in which approximation was made in reactive power calculation and in this way the equation form was much simplified.

0

tan

cos

ac d d

ac ac

d d

P U I

Q P

U U

(1)

where, Ud0 is the converter transformer no-load DC voltage, Pac and Qac is the active and reactive power from AC to DC. Id and Ud is the current and voltage of DC line.

In normal operation, HVDC links required to transmit a scheduled power. In such an application, the master control layer receives the power schedule, modifies by auxiliary power control and then converts the power sig-nal into the coordinated bi-pole current order commen-surate with the DC voltage. Pole control is the core of HVDC control and activates the appropriate controller of the rectifier and inverter station according to the state of AC/DC systems. Then it produces the firing angle for both rectifier and inverter stations. The control scheme is shown in Figure 3.

Pole control at the rectifier side has a current controller, which takes the maximum and minimum current con-strains and the VDCOL into consideration. The mini-mum firing angle control is embedded implicitly in the

Figure 3. Control scheme of DC system

Current controller

∑

cosα cosβ

∑

Idorder

Imod

ΔI +

_

_

+

Id

Idorder

P/VVd

Pmod

Pdorder

∑

Idorder-Im

Id ΔI +

_

∑ Vdorder ΔV

+

_

Rdc

+

∑ γ Δγ

_

γ0

Current limiter

Controllogic

Id Vdr Vdi

Vdi

Current controller

voltage controller

γ controller

Id



77

current controller by angle limits. Pole control at the in-verter station includes a voltage controller, a constant current controller and a constant extinction angle con-troller. Current error Im provides a transition between the current control and voltage control to facilitate control stabilization. The shift logic of these controllers is im-plemented by:

0 0cos cosd r d id

xr d xi

U UI

d R d

(2)

where: dxr, dxi are the equivalent resistance of the rectifier and inverter. Rd is the resistance of DC line. α and γ cor-respond to the rectifier ignition angle and inverter extinc-tion angle.

When situation needed, additional control will be joined through Pmod or Imod to fully exert the DC features of fast power control and improve the dynamic perform-ance of AC system.

3.3 Criterian for Transient Stability

According to power system stability guideline of China, to keep transient stability, the following conditions must be satisfied at the same time:

Angle stability: after disturbance, any rotor angle be-tween two generators in the same AC system takes on a damping oscillation.

Voltage stability: the continuous time of low voltage under 0.75 pu is within 1 s. The voltage of pivot buses is above 0.8 pu when the fault is clear.

Frequency stability: the frequency collapse will not happen with secure measures such as loads shedding and generator tripping. The frequency can restore to the nor-mal level and the large unit operation will not be af-fected.

4. Transient Stability Analysis

Transient stability criterion for the studied system re-quires the system to be stable after clearing of any single fault or successful reclosing. Amount of simulations were done on this system. For AC system, the worst con-dition occurs when three phase permanent fault happens near Yimin power plants. For DC system, the worst con-dition occurs when bipolar blocking happens. Stability measures are necessary for most cases.

4.1 Dynamic Behavior when AC Faults

The typical case was studied when a permanent three phase fault occurred in the exit of 500 kV Yi-Feng line. The steady state condition is 6 units for DC power send-ing and DC capacity 3000 MW. Stability measure is that three unit tripping in Yimin plants and AC/DC separated within 150 m. Because the close electric distance with Yimin bus, AC bus voltage of Yimin converter reduced a lot and is close to zero in the fault instant. As a result, DC voltage brought down along with AC voltage till it

can not work its way and quitted temporarily. The mis-balance between energy and power of AC system in Yimin and Hulunbeier areas was further increased and the instability of the system became even worse. After one second the fault line is cleared, the AC voltage is recovered. When condition is permitted, DC system re-started and DC power is restored, which is much helpful for AC system stability. Voltage, rotor angle and DC power curves during the disturbance were shown in Fig-ure 4, Figure 5 and Figure 6.

To study the influence of DC power to AC stability, simulation was done under 4 units and different DC power. Results were shown in Figure 7 without measure and in Table 1 with necessary measure. Results showed that power exchanging from AC to DC system is advan-tageous to improve AC system stability.

4.2 Dynamic Behavior when DC Faults

When fault occurred in DC line, the bipolar blocking is the most serious one. In this case, mass power shifts to Yi-feng lines and make a great impact on AC system. If

Figure 4. Voltage variation of Yimin and Fengtun bus

Figure 5. Rotor angle variation of Yimin unit

Figure 6. DC power and current variation



78

Figure 7. Rotor angle variation of Yimin unit under differ-ent DC power

Table 1. Stability measures under different DC power

DC capacity Measures To Keep stability

3000 MW No measures needed

2600 MW two units tripping in Yimin plants within 150 ms

2200 MW two units tripping in Yimin plants and AC/DC separated within 150 ms

1800 MW and AC/DC separated within 150 ms

DC can restart successfully, the oscillation can be ap-peased and AC system keeps stable; Else, AC system will lose stability without secure control measures.

In the above simulation, The DC system experiences a restart failure for the first time and a success start over a lowering voltage for the second time. Along with DC restore, the AC system got a smooth resumption. To keep stability, the control measures under different DC powers are listed in the following table in case of bipolar block-ing and failure restart. Results showed that the severity of the fault is closely related with the exchanging power between AC and DC system and the unit boot mode of Yimin plants.

Figure 8. Rotor angle variation of Yimin unit

Figure 9. DC power and current variation

Table 2. Stability measures under different DC power in bipolar blocking

DC capacity Measures to keep stability

6 unit, 3000 MW 5 units tripping




4.3 Additional Controls for AC/DC Hybrid System

The following simulation was done in such an operation: 4 units for DC power sending and DC power 1800 MW. First, to research the features of additional controls for AC/DC hybrid system, simulations were calculated by three cases: that is no additional control, with TCSC forced compensation control and with DC emergency power control. Results were shown in Figure 10. When the additional control of TCSC and DC is activated, the transient stability can be improved but in a limit com-pared with unit tripping under such serious faults.

Figure 10. Additional control effect for AC and DC hybrid system

Figure 11. Ewenke frequency when AC and DC is separated



79

Secondly, frequency restore will meet problem when AC/DC is separated during dynamic process. It is mainly due to the AC simple structure in the rectifier side of DC system and the mismatch power between the DC system and AC islands system. The frequency simulation was done in two cases: without additional DC control and improving DC power by 15%. Results show that fre-quency can be recovered in an acceptable range with DC control.

The HVDC system with island generators in sending side will meet a frequency stability problem when the AC/DC parallel transmission system changed from con-nection into separation. Frequency stability depends on the generation capacity in the sending side and the HVDC power. Improving the generation capacity in the sending side and decreasing the mismatch power be-tween the generation capacity and DC power in the sending side will be helpful to the improvement of fre-quency stability. DC emergency power control is neces-sary in the frequency control process. Amount of simula-tion has done to explore the frequency stability rules of such a special HVDC system. The results and stability terms are listed in the following table, which shows the relationship between the mismatch power and frequency stability. Advanced DC additional power controller is still needed for delicate frequency control.

5. Conclusions

In this paper, transient stability analysis was carried out focus on the Hu-Liao HVDC and AC parallel transmis-sion system. The dynamic performance and the interac-tion between AC and DC systems during serious distur-bance were researched by detail simulation. Only several examples are showed and discussed here as space limited. Results showed that:

1) As for the weak network of Hu-Liao rectifier side and the limit of AC transfer capacity, transient stability problem is rather serious. The dynamic interaction be-tween AC and DC system during disturbance is intense.

2) The exchanging power between AC and DC system have a corresponding influence on dynamic performance and control strategies for keeping transient stability. Power exchanging from AC to DC system is advanta-geous to improve AC system stability.

3) By proper control, TCSC forced compensation and DC emergent power transfer can improve system stabil-ity. Compared with generator tripping and AC/DC sepa-rating, they are only a subsidiary control method for the researched system.

4) Frequency stability by additional DC controls is necessary when AC/DC parallel transmission system changed from connection into separation.

The research will bring important and significant ref-erence for further operation and stability control of Hu-Liao HVDC and AC system. Further research still

Table 3. The maximum line transmission of Yimin-Yimin converter to keep frequency stability

DC island generation

Unit number Unit generation

Line transmission limit of Yimin-Yimin converter (MW)

4 No limit –650 ~ 850

> 1550 –350 ~ 850 3

< 1550 –350 ~ 550

> 650 –350 ~ 550 2

< 650 –50 ~ 550

remained on the topics such as the optimal operation of AC and DC system, DC separated operation and control, strategies design for stability control and so on.

REFERENCES [1] IEEE Committee Report, “AC-DC Economics and Alte-

matives-1987 Panel Session Report,” IEEE Transaction on Power Delivery, Vol. 5, No. 4, October 1990, pp. 1956-1976.

[2] H. Ritva, “Torsional Interaction between an HVDC Link and Large Turbine-Generators,” Saehkoe Electricity and Electronics, Vol. 62, No. 6, June 1989, pp. 38-41.

[3] IEEE Committee Report, “HVDC Controls for System Dynamic Performance,” IEEE Transaction on Power system, Vol. 6, No. 2, May 1991, pp. 743-752.

[4] R. John and U. Edvina “Study of Power Transfer Capa-bility of DC Systems Incorporating AC Loads and a Par-allel AC Line,” IEEE Transactions on Power Delivery, Vol. 12, No. 1, January 1997, pp. 426-434.

[5] K. W. V To, A. K. David and A. E. Hammad, “A Robust Coordinated Control Scheme for HVDC Transmission with Parallel AC System,” IEEE Transactions on Power Delivery, Vol. 5, No. 4, July 1994, pp. 1710-1716.

[6] A. E. Hammad, “Stability and Control Strategy for Paral-lel Operation of AC and DC Transmission Systems,” Proceedings of 6th International Conference on AC and DC Power Transmission, London, September 1996, pp. 163-171.

[7] A. E. Hammad, “Stability and Control of HVDC and AC Transmission in Parallel,” IEEE Transactions on Power Delivery, Vol. 14, No. 4, October 1999, pp. 1545-1551.

[8] H. Z. Cai, Z. H. Qu and D. Q. Gan, “A Nonlinear Robust HVDC Control for a Parallel AC/DC Power System,” Computers and Electrical Engineering, Vol. 29, No. 1, 2002, pp. 135-150.

[9] B. L. Quan, “Stability of Tian-Guang HVDC and HVAC Transmission System,” ICPST, Beijing, 1994.

[10] Y. Jing, Z. Ren, B. Q. Li and S. R. Ma, “Research on Transmission Capability of Tian-Guang AC and DC Hy-brid System,” Power System Technology, Vol. 26, No. 8,



80

2002, pp. 52-55.

[11] Y. Jing, L. C. Li, Z. Ren, “Stability Control of Tian-guang AC and DC Parallel Transmission System,”

Automation of Electric Power Systems, Vol. 26, No. 1, 2002, pp. 49-52.



81


Xudong Song1, Zhi Xu2, Tieyuan Xiang1

1School of Electrical Engineering, Wuhan University, Wuhan, China; 2Jiangsu Huai’an Power Supply Company, Huai’an, China. Email: [email protected], [email protected], [email protected] Received May 24th 2010; revised July 6th 2010; accepted July 10th 2010.

ABSTRACT

The initial parameters of the superconducting magnetic energy storage (SMES) fed by a voltage source converter (VSMES) are studied and the setting rules are designed in this paper. A time-domain simulation model is established by using the software PSCAD/EMTDC. Based on this model, the application of the VSMES in the power system is used to test the designed rules. The simulation results are valuable for the further research of the initial parameters design of the VSMES. Keywords: Superconducting Magnetic Energy Storage (SMES), VSMES, Initial Parameters, Voltage Sag, Time-Domain

Simulation

1. Introduction

Superconducting magnetic energy storage system (SMES), a new electric power regulating device used in the power system which combines the superconducting magnet with the power electronic converter, is to store up the electric energy in the form of magnetic energy. If neces-sary, SMES can compensate the power system for the energy shortage as soon as possible. SMES can not only realize the adjustment of input and output power of SMES in four quadrants, but also control the bi-directional flow of power between the SMES and the power system with a rapid energy exchange rate. Therefore, the SMES is one of important aspects in the application field of superconducting technology [1].

According to the main circuit topology of power elec-tronic converter, the SMES can be classified into the current source and the voltage source. Though current characteristic is the inherent feature of SMES, the volt-age source converter, as the mainstream development technology currently, has a lower cost and mature tech-nology. Hence, the SMES system adopts voltage source converter more. Lately, a great deal of literatures had made relative analysis both on the control strategy of SMES and the stability improvement of power system, however, researches concerned on the time-domain mod-eling and simulation of SMES connected to the power system are rare. This paper establishes the time-domain simulation model of VSMES by the use of PSCAD/ EMTDC, analyses the cases that VSMES compensates

the system instantaneous voltage sag and enhances the transient stability of the generator, which have provided references for the practical applications of VSMES in power system [2].

2. Basic Configuration and Work Principle of VSMES

As Figure 1 shows, the VSMES mainly consists of the voltage source converter, the support capacitance, the chopper, and the superconducting magnet (SM) and is connected to power system through the transformer. The superconducting magnet is equivalent to a pure induc-tance without resistance and cooperates with the chopper, making the support capacitance a direct current and con-stant voltage source. VSC is the core of the whole VSMES. Controlling the phase angle and width of trigger pulses of gate turn off thyristor (GTO) in VSC and changing the phase and the amplitude of voltage at the joint of VSMES and the system, which can indirectly control the current between the VSMES and the power system and ultimately realize the control the bidirectional power flow [3].

3. The Time-Domain Simulation Model of VSMES

This paper adopts the user-defined components contained in PSCAD/EMTDC software to build a time-domain digital model that possesses four functions-magnetizing, freewheeling, tracing, demagnetization of VSMES.



82

Power System Transformer

Line impedance

VSC

Support Capacitance

Chop

SMPower system

VSC Chop

Transformer Line impedance

Support capacitance

SM

Figure 1. Basic construction plan of VSMES

In the model shown as Figure 2, the 0.2 Ω resistance and the 0.005 Ω inductance are used to simulate the line loss and the 10 H inductance is used to simulate the su-perconducting magnet. Es is the phase voltage at the joint of VSMES and the transformer. The voltage source con-verter employs the three-phase six-impulse bridge in-verter. The control pulse of GTO is G1~G6 in turn. Ac-cording to the control pulse signal produced by the target power, changing the amplitude and the phase of Es can realize the control of the power.

The power control system (PCS) of VSMES adopts double closed-loop serial control, including a voltage outer loop and a current inner loop. Since the direct cur-rent control strategy of current inner loop is introduced, the dynamic response of VSMES is improved to a large extent and also the anti-disturbance ability. Through the d-q transform of three-phase voltage and based on the target tracing power, the current value in the d-q coordi-nate is obtained, moreover, the target current value needed to be regulated can be derived from the decoup-ling transform of the current above, which achieves the purpose of direct current control [4].

The chopper comprises two-quadrant chopping circuit which is reversible in voltage. It adopts bang-bang con-trol strategy and maintains the voltage of the support capacitor constant. The control of the converter aims at power controlling, while the control of the chopper aims at voltage invariableness which supports the capacitance.

In this way, it decreases the difficulty of dynamic match- ing between the convertor and the chop to a large extent.

4. The Initial Parameters Design of the VSMES

When the disturbance occurs, the VSME can compensate the power system with the storing energy. And the stored energy, in other words the initial parameters of the VSMES decides its compensating capability. So, to de-sign the initial parameters of the VSMES is very impor-tant. The initial parameters include L, Ism_ref, Udc and ES. Where L represents the pure inductance of the supercon-ducting magnet; Ism_ref is the value of initial magnet cur-rent; Udc is the real time voltage the supports the capaci-tance; ES is the voltage magnitude on AC side.

4.1 The Rule for Selecting L

Superconducting magnet in the superconductive state is equivalent to a pure inductance without resistance in the VSMES. The relationship between the current I, L and energy W is following:

21

2W LI (1)

When L is constant, I is in direct proportion to W in (1). But the current carrying capacity of the SM limits the magnitude of I. Meanwhile, Equation (2) shows that the DC voltage on the SM and the SM current all limits the selecting of L.

smessmes

diU L

dt (2)

where Usmes is the DC voltage on the SM; ismes is the SM current. Considering the above factors and the real SMES made by HUST, the L is setting for 10 H in this paper.

Figure 2. The circuit model of VSMES



83

4.2 The Rule for Selecting Ism_ref

Considering that the VSMES can compensate the three unbalanced power of the power system during the oscil-lating period which has the biggest oscillatory power, the reference value _sm refI of the VSMES can be fixed

according to the unbalanced power which must be com-pensated and the compensating time.

2_

1

2 sm refE L I (3)

where E is the energy which the VSMES stores and just can meet the need of the power compensating.

Supposing the max compensating power by the VSMES during the oscillating period which has the biggest oscil-latory power is maxP ; and the keeping compensating

time is t , it must have:

maxE P t (4)

It must be pointed out that, when the unbalanced power of the power system likes the low frequency os-cillation showed in the Figure 3, E must be

max

2

2E P t (5)

Equations (3) and (4) both suppose that the VSMES keeps to supply the energy to the power system during all the compensating time. So, when the fault ends, the SM current is lowest. And the lowest current can be set as Ism_min. Considering the energy loss during the compen-sating and the loss energy is corresponding to the current

_sm lossI varied in the SM, the wanted Ism_ref is:

_ _ _ min _sm ref sm ref sm sm lossI I I I (6)

In summary, the rule for selecting Ism_ref is (7).

max

_

_ _ min

1

2 sm ref

sm ref ref sm loss

E P t

E L I

I I I I

(7)

t

maxPP

t

P

Figure 3. The curve of low frequency oscillation

where, Ism_min and _sm lossI need artificially setting and

regulating according to the actual simulating result. And Ism_min is very critical.

4.3 The Rule for Selecting Udc

In normal condition, the instantaneous adjustable output power by the VSMES meets (8).

2 2sm sm dc smesP Q U I (8)

where, smP is the output active power by VSMES; smQ

is the output reactive power by VSMES. Along with the VSMES’s work, its adjustable instan-

taneous power range gradually is also changing. When the VSMES absorbs energy from the power system, Ismes up, Udc Ismes up, and the adjustable power range is bigger; otherwise, when the VSMES supplies energy to the power system, Ismes down, Udc Ismes down, and the adjust-able power range is smaller. Meanwhile, the suitable allowance must be considered when selecting Ism_ref. If the allowance is not enough, the VSMES cannot provide the needed power to the power system. In another word, the needed power exceeds the adjustable power range of the VSMES. This will cause the whole VSMES system crash. And Ism_min represents the allowance of Ism_ref actu-ally. So, it is the reason why Ism_min is very critical. It must meet (9).

2 2_ minsm sm dc smP Q U I (9)

According to the analysis above, supposing the system instantaneous power shortage which is needed to com-pensate is S, the rule for selecting Udc is (10).

_ mindc

sm

SU

I (10)

4.4 The Rule for Selecting ES

In order to ensure the power regulation of the VSMES in the controlled linear range, the signal modulation ratio M in the SPWM modulation must belong to 0 to 1. The relevant calculation of M is (11). And according to the Udc and the P and Q which are the power needed to out-put by the VSMES, ES can be got.

Maintaining M at 0.5, the corresponding ES can be got by the method showed in Figure 4. It is difficult to de-duct ES directly from Udc according to (11). So the in-verse method showed in Figure 4 is adopted. Presetting the ES, to calculate U'dc, and then compare it with the setted Udc, judging its suitability by the setting ES. If U'dc and Udc are more or less the same, then select the value of ES; if U'dc and Udc are very different, then continue to adjust the ES.



84

Figure 4. The method of selecting ES

2 2

1 2 3

1 2 3

1 2 3

1 2 3

2 2

2 2

22 /

3

3

20

rd rq dc

rd rd rd rd

rq rq rq rq

drd sd rd q rd d

qrq sq rq d rq q

sd sqd

sd sq

sq sdq

sd sq

sd s

sq

M U U U

U U U U

U U U U

diU U U Li U Ri L

dtdi

U U U Li U Ri Ldt

PU QUI

U U

PU QUI

U U

U E

U

0,1M

(11)

4.5 Amendment of Ism_ref, Udc and ES

The above rules of the VSMES initial parameters are the basis general principles, which are based on the VSMES compensating the power system with equivalent three unbalanced power. But in the real applications of VSMES in power system, due to the existence of outer controller that is designed for achieving the set goals, under the control strategy of the outer controller, the compensating by the VSMES is not equivalent three un-balanced power compensating. Based on these circum-stances, it must adjust the values of Ism_ref and Udc, ac-

cording to the actual situations. In the permitted extension of Ism_ref and Udc, it should

increase Udc comparatively and reduce Ism_ref. Lower Ism_ref on one hand can help to reduce the internal power loss, on the other hand, help to reduce the design index of current carrying capacity for the SM and the power con-verter device.

In addition, Es is designed according to (11), under the situation that M = 0.5. While in the applications of the VSMES in power system, if Es can be lower and still be able to keep the power conditioning of VSMES in the SPWM linear regulate region , it can set a lower Es. This helps reduce the design index of the withstand voltage for the SM and the power converter device.

5. VSMES Used for the Voltage Instantaneous Sag

Figure 5 is the circuit diagram of VSMES compensating the voltage instantaneous sag, in which the rating voltage of the generatrix is 115 kV and the system capability is 500 MVA. When the simulation runs to the 21st s, the system fails for 0.75 s because of the three-phase ground- ing short circuit. During the 0.75 s, the bus voltage falls from 0.78 (p.u.) to 0.61, then recovers to 0.78.

If the system active power is abundant, VSMES merely need to provide enough reactive power for the system to compensate the voltage instantaneous sag. As Figure 5 shows, VSMES is connected to the bus through the transformer. The outer loop power control of the VSMES system adopts the bus voltage as the feedback control quantity. Comparing the real-time value of the bus voltage with the rating value, we can get the error signal U which will be regulated by PI controller to



85

Figure 5. The circuit diagram of VSMES for compensating the voltage instantaneous sag generate the reactive power control signal of the VSMES while the active power control signal remains zero [5-7].

According to the aforementioned design rules for the initial parameters of VSMES, the four initial parameters are setting as below:

The L is setting for 10 H. The support capacitance voltage of the VSMES is setting for 100 kV and the ini-tial current of the superconducting magnet is setting for 5 kA. The VSMES is connected to the system in parallel through the step-down transformer, and the secondary voltage of the transformer is 6 kV.

Figure 6 is simulation results of the bus voltage. Fig-ure 6(a) depicts real-time comparison curves of the bus voltage during the whole simulation. Figure 6(b) is tran-sient state comparison curves when the VSMES com-pensates for the voltage sag. Where Vpu_0 represents the system bus voltage without the availability of the VSMES, and Vpu represents the system bus voltage with the VSMES.

It’s obvious that the VSMES compensates the instan-taneous voltage sag of the system voltage favorably.

Figure 7 presents each monitoring variable of VSMES during the simulation. Figure 7(a) provides the real-time power changing curve of VSMES, in which the upper curve represents the active power Psm, and the lower curve represents the reactive power Qsm. Figure 7(b) describes the superconducting magnet current. Figure 7(c) displays the voltage of the support capacitance.

When the VSMES compensates the system voltage sag, it is found that the system needs about 125 MVar reac-tive power. Despite the fact that the VSMES is only modulating reactive power, the magnet current will de-cline due to losses of some components such as the con-vertor. In Figure 7(b), the magnet current declines from

(a)

(b)

Figure 6. The generatrix voltage. (a) Comparison graph of the generatrix voltage (p.u.); (b) Transient state comparison graph of the generatrix voltage(p.u.) the original 5 kA to 1.8 kA during compensating. While Figure 7(c) shows that the voltage of the support capaci-tance maintains invariable after magnetizing.

It’s shown that the VSMES can compensate the in-stantaneous sag of system voltage rapidly and the setting of initial parameters is logical. As shown in this simulation,



86

(a)

(b)

(c)

Figure 7. The monitoring variables of VSMES. (a) The real- time power of VSMES; (b) The superconducting magnet current; (c) The voltage of the support capacitance in the end step when the VSMES compensates the volt-age decline and outputs 125 MVar reactive power, the magnet current declines to 1.8 kA, however, the magnet current and the support capacitance voltage of 100 kV can still enable the VSMES adjust the reactive power of 125 MVar, and ensure that the compensating on the voltage sag keeps normal, avoiding the voltage collapse which supports the capacitance. Put it another way, the rules of the initial parameters design of the VSMES is feasible.

However, if the initial parameters are changed not ac-cording to the designed rules, for example: the initial current of the superconducting magnet is lowered to 4.5 kA, the support capacitance voltage will decrease when VSMES is compensating the voltage sag as Figure 8

(a)

(b)

Figure 8. The variables of VSMES after initial parameters changed. (a) The superconducting magnet current after initial parameters changed; (b) The voltage of the support capacitance after initial parameters changed shows.

It’s obvious that the energy stored in the superconducting magnet is not enough for the compensating. So it needs additional energy which supported by the support capacitance at the end of the compensating. Then the voltage of the support capacitance decreases. This will have negative effect on the control of VSMES. Therefore, the changed initial parameter is not reasonable. Mean-while it can be approved that the designed rules in this article are feasible and practical.

6. Conclusions

Based on the principle analysis of the VSMES, this paper establishes a time-domain simulation model of the VSMES with six-pulse and studies on the application of the VSMES in the power system.

To sum up: 1) On the basis of the detailed research on the opera-

tional characteristics of the VSMES, the time-domain simulation model of the VSMES is established.

2) The initial parameters setting of the VSMES di-rectly determines the effects on the power system and involves the problem of the power range. This paper proposes a basic rule for parameters setting and uses an example to testify its feasibility.

3) The VSMES can adjust power in four quadrants and



87

the dynamic response is rapid, which is desirable for compensating the voltage sag of the system.

REFERENCES

[1] Y. Yang, “Development of Superconducting Technique and Its Application in Power System,” Power System Technology, Vol. 25, No. 9, 2001, pp. 48-60.

[2] Y. Shirai and T. Nitta, “On-line Evaluation of Power System Stability by Use of SMES,” IEEE Power Engi-neering Society Winter Meeting, New York, Vol. 2, 2002, pp. 900-905.

[3] Y. Ohsawa, Y. Maruoka, H. Takeno, et al., “Determina-tion of Installation Location of SMES for Power System Stabilization,” Proceedings of the 2000 IEEE Interna-tional Symposium on Circuits and Systems, Geneva, 2000, pp. 233-236.

[4] Y. Li, S. J. Cheng, Y. Pan and Y. J. Tang, “Time Domain Simulation of the Characteristics for a Voltage Source Converter Fed SMES,” Automation of Electric Power Systems, Vol. 26, No. 18, 2002, pp. 60-63.”

[5] X. Liu, X. G. Zhu, X. Chu and X. H. Jiang, Voltage Sag Compensation by SMES,” Automation of Electric Power Systems, Vol. 28, No. 3, 2004, pp. 40-44.

[6] H. Zhang, P. C. Zhu, Y. Kang and J. Chen, “Improving Damping of Generator using Superconducting Magnetic Energy Storage Systems,” Power Electronics, Vol. 37, No. 1, 2003, pp. 50-53.

[7] S. X. Zhou, W. Wu, J. L. Wu and N. An, “Application of Superconducting Magnetic Energy Storate to Improve Transient Voltage Stability,” Power System Technology, Vol. 28, No. 4, 2004, pp. 1-5.




Srikrishna Subramanian, Radhakrishnan Anandhakumar

Department of Electrical Engineering, Annamalai University, Annamalainagar, India. Email: [email protected] Received May 29th 2010; revised June 29th 2010; accepted July 3rd 2010.

ABSTRACT

This paper presents a new approach via composite cost function to solve the unit commitment problem. The unit com-mitment problem involves determining the start-up and shut-down schedules for generating units to meet the forecasted demand at the minimum cost. The commitment schedule must satisfy the other constraints such as the generating limits, spinning reserve, minimum up and down time, ramp level and individual units. The proposed algorithm gives the com-mitted units and economic load dispatch for each specific hour of operation. Numerical simulations were carried out using three cases: four-generator, seven-generator, and ten-generator thermal unit power systems over a 24 h period. The produced schedule was compared with several other methods, such as Dynamic programming, Branch and bound, Ant colony system, and traditional Tabu search. The result demonstrated the accuracy of the proposed method. Keywords: Composite Cost Function, Generation Scheduling, Unit Commitment

1. Introduction

Unit commitment is to determine the commitment and generation levels of generating units over a period of time to minimize the total operation cost [1]. In solving the unit commitment problem, generally two basic deci-sions are involved, namely, the “unit commitment” deci-sion and the “economic dispatch” decision. The “unit commitment” decision involves the determination of the generating units to be running during each hour of the planning horizon, considering the system capacity re-quirements, including the system constraints. The “eco-nomic dispatch” decision involves the allocation of the system demand and spinning reserve capacity among the operating units during each specific hour of operation. The unit commitment problem has commonly been for-mulated as a non-linear, large scale, mixed-integer com-binatorial optimization problem with constraints. Re-search endeavours, therefore, have been focused on effi-cient, near-optimal solutions. A survey of literature on unit commitment methods reveals that various numerical optimization techniques have been employed to address the unit commitment problems.

An interior-point/cutting-plane method for non differ-entiable optimization is used to solve unit commitment problem [2]. This method has two advantages of sub- gradient and bundle methods that have better conver-gence characteristics and does not suffer from the pa-

rameter-tunning drawback. A parallel repair genetic al-gorithm has been proposed to solve unit commitment problem [3]. This algorithm provides a modeling frame-work less restrictive than previous approaches such as dynamic programming or lagrangian relaxation.

A new cooperative coevolutionary algorithm has been described for unit commitment problem which combines the basic ideas of LR and GA to form a novel two-level approach [4]. A successive sub problem solving method is developed and applied to solve the unit commitment problems with identical units [5]. The commitments of the identical units can be differentiated and the homoge-nous oscillations are avoided. The unit commitment problem has been solved with dual variable constraints [6].

An evolutionary programming based Tabu search has been applied to solve unit commitment problem [7]. The security constrained unit commitment problem is de-composed into two coordinated problems, based on benders decomposition, which include a master problem for optimizing UC and a sub problem for minimizing network violations [8]. The price-based unit commitment problem has been solved based on mixed integer pro-gramming [9].

The fuzzy logic based UC scheduling using absolutely stochastic simulated annealing have been proposed in which they introduce the sign bit vector to reduce eco-



89

nomic load dispatch calculations [10]. An ant colony system approach for unit commitment problem has been proposed [11]. The algorithm implements the movement of ants in the search space and also discusses the accu-racy of the solution with respect to the solution time.

The mixed-integer linear formulation for the unit commitment of thermal units have been applied to solve this problem, this method requires fewer binary variables and constraints than previously reported models, yielding a significant computational saving [12]. The unit com-mitment problem has been solved by Simulated Anneal-ing and they maintained spinning reserve capacity, re-sulting in near-optimal UC solutions [13].

A Tabu search based hybrid optimization approach for a fuzzy modeled unit commitment problem has been pro-posed [14]. A unit commitment problem with probabilis-tic spinning reserve and interruptible load has been for-mulated [15]. Recently Bacterial foraging technique has been applied to solve unit commitment problem [16].

The artificial intelligence approaches Genetic Algo-rithm (GA), Evolutionary Programming (EP), Simulated Annealing (SA), Tabu Search (TS) and Expert Systems (ES) have been proposed to solve the UC problem and these methods require high computational time for large scale systems. This article presents a composite cost function based solution algorithm for solving unit com-mitment problem and it directly gives the units to be committed and the economic dispatch of the committed units while satisfying the equality and inequality con-straints.

2. Problem Formulation

2.1 Nomenclature

ai , bi , ci Fuel cost coefficients unit i Shi – cost Hot start cost in Rs Sci – cost Cold start cost in Rs c–s–hour Cold start hour in hours MUi Minimum up time in hours MDi Minimum down time in hours N Number of generating units Pi max Maximum output power of unit i in MW Pi min Minimum output power of unit i in MW Pit Power produced by unit i in time t PGi Power generation of the plant i in MW PDt Power demand at hour t in MW PRt Spinning reserve requirement at hour t in MW Rs Rupees ini state Initial status of the unit in hours SDi–cost Shut down cost in Rs STi Start up cost in Rs CCF Composite cost function h hour MW Mega Watt

i Index of generating units ( i = 1 ,2,….,N) Xi

on (t) Duration of continuously ON of unit i in hour t Xi

off (t) Duration of continuously OFF of unit i in hour t λ Incremental production cost

2.2 Unit Commitment Problem

Suppose there are N thermal units and the time horizon is T. The unit commitment problem is to determine the commitment and generation levels of all units over the period T so that the total generation cost is minimized. It is formulated as a mixed-integer optimization problem

1 1

2

min ,

,

cos

cos

N T

i i i ii t

i i i i i i i

i

i

C with C C P t ST t SD t

Where C P t a P t b P t c

ST startup t

SD shutdown t

(1)

Subject to the following constraints.

2.3 Constraints

2.3.1 System Power Balance Constraint The generated power from all the committed units must satisfy the load demand, which is defined as

1

N

it ti

P PD

(2)

2.3.2 Generation Limit Constraints Each unit has a valid generation range, which is repre-sented as

min max , 1,2,...i it iP P P i N (3)

2.3.3 Spinning Reserve Constraints

1

N

it t ti

P PD PR

(4)

The total amount of power available at each hour must be greater than the load demanded. The reserve power available, denoted by PRt , is used when a unit fails or an unexpected increase in load occurs.

2.3.4 Initial Status Constraints At the beginning of the schedule, the unit initial status must be taken into account.

2.3.5 Minimum Up and Down Time Constraints Once a unit is committed/decommitted, there is a prede-fined minimum time after which it can be decommitted/ committed again.

oni iMU X t (5)

offi iMD X t (6)



90

2.3.6 Startup Cost

offci i

i

hi

S if X c s hourST

S

(7)

2.3.7 Ramp Level From the beginning to end of the schedule the ramp level that is the output of the unit is maintained with in pre-scribed limits.

3. Composite Cost Function

The composite cost coefficients are derived as follows. The total fuel cost of the ‘N’ unit system can be writ-

ten as

1 2 3 .......T NF F F F F (8)

For most economical generation,

1 1 1 1 1 12 ; 2a P b P b a

2 2 2 2 2 22 ; 2a P b P b a

3 3 3 3 3 32 ; 2a P b P b a

2 ; 2N N N N N Na P b P b a

2 ; 2G GA P B P B A (9)

where, is the incremental production cost of the plant in MW.

The total generation of the plant can be written as,

1 2 3 ....G NP P P P P

1 2 3

1 1 2 2 3 3

1 2 3

1 1 2 2 3 3

1 2 3

2 1 1 1 ........ 1

1 2 .......

2 1 1 1 1 ....... 1 ]

.......

1 1 1 1 ....... 1

G N

n N

N G

N N

N

P a a a a

b a b a b a b a

a a a a P

b a b a b a b a

a a a a

(10)

by comparing (9) & (10)

1 2 31 1 1 1 ....... 1 NA a a a a (11)

1 1 2 2 3 3 ....... N NB b a b a b a b a A (12)

The fuel cost can be rewritten as 2 2

1 1 1 1 14 4F a b a c ;

2 22 2 2 2 24 4F a b a c ;

2 24 4N N N N NF a b a c ;

2 24 4TF A B A C (13)

From (13)

1 2 3

2 2 2 21 1 2 2 3 3

2

.........

4 4 4 ... 4

4

N

N N

C c c c C

b a b a b a b a

B A

(14)

4. Description of CCF Based Unit Commitment and Economic Dispatch

The detailed computational flow of the proposed meth-odology is presented as a flow chart in Figure 1 and the algorithmic steps of CCF based solution for unit com-mitment problem are presented as follows.

Step 1: Read the unit characteristics, cost coefficients and load.

Step 2: Generate the possible binary combinations us-ing 2n, where n is the number of generating units.

Step 3: Choose the possible combinations to satisfy the power demand and spinning reserve constraints for first hour.

Step 4: Compute CCF coefficients and λ for each com-bination.

Step 5: Evaluate the generation of units for each com-bination.

Step 6: Compute the total operating cost for each com-bination.

Step 7: Choose a combination with minimum total op-erating cost.

Step 8: If it is last hour go to step 9 else increase the hour and go to step 3.

Step 9: Print the unit commitment schedule and dis-patches.


The proposed technique has been implemented in MATLAB on a 2.10 GHz core 2 Duo processor PC. The performance of the algorithm has been evaluated through simulation. Simulation studies have been carried out on four-generator [11], seven-generator [14] and ten-generator [11] sample systems. Before proceeding to the dispatch of demand to generators, careful selection of units is im-portant. In the proposed method first generate the possi-ble binary combinations using 2n, where n is the number of generating units. For the four unit system, the possible binary combinations are 0000 to 1111. The first combi-nation is not necessary because all the units are off. The maximum generation value is substituted in the combina-tions where the unit was on and the total generation of the each combination is obtained.

The first hour load demand is 410 MW. The possible binary combinations are five, such as 0110, 0111, 1011, 1110, 1111 having generations of 630, 610, 690, 440,



91

550 MW. For each combination CCF coefficients for committed units, lambda, power generation and total operating cost are obtained. Then the combination 1111 is chosen for first hour because its operating cost is low (Rs.862/-). Hence for the first hour the unit combination is 1111 and the economic schedule for the committed unit is P1 = 72 MW, P2 = 138 MW, P3 = 140 MW and P4 = 60 MW. In case if there is any violation of power limit, the power is fixed at its maximum limit or mini-mum limit. Then the CCF coefficients, lambda, power generation and total operating cost are obtained. The same procedure is continued for next hour.

5.1 IEEE 4-Unit Test System

The proposed approach has been tested on a sample sys-tem consisting of four generating units. Table 1 summa-rizes the committed units, operating cost and start up cost. The committed unit shows that there was no need to add startup and shut down cost because all the units were ON

Start

Read system data

h = 1

Choose the possible combinations satisfy power demand and spinning reserve constraints

Generate 2n possible combinations

Compute CCF coefficients and λ for each combination

Evaluate the generations of units for each combination

Compute the total operating cost for each combination

Choose a combination with minimum total operating cost

Print the unit commitment schedule and dispatches

If h = T

Stop

NO

YES

h = h+1

Figure 1. Flow chart of the proposed method

Table 1. Committed units and cost for 4-unit system

Hour U1 U2 U3 U4 Cost ST.Cost

1 1 1 1 1 862 0

2 1 1 1 1 1057 0

3 1 1 1 1 1237 0

4 1 1 1 1 1433 0

5 1 1 1 1 1188 0

6 1 1 1 1 946 0

7 1 1 1 1 841 0

8 1 1 1 1 935 0

9 1 1 1 1 1140 0

10 1 1 1 1 1301 0

11 1 1 1 1 1152 0

12 1 1 1 1 1046 0

13 1 1 1 1 946 0

14 1 1 1 1 1095 0

15 1 1 1 1 1263 0

16 1 1 1 1 1366 0

17 1 1 1 1 1128 0

18 1 1 1 1 979 0

19 1 1 1 1 851 0

20 1 1 1 1 1039 0

21 1 1 1 1 1220 0

22 1 1 1 1 1327 0

23 1 1 1 1 1176 0

24 1 1 1 1 1019 0

Total cost (Rs) 26547

during entire 24 hour. Table 2 gives the optimum gen-eration schedule for various system demands. The ramp level should be maintained for entire 24 hour. The results do not show any violation of reserve and power demand constraints. The proposed method in comparison with Dynamic Programming (DP), Branch and bound and Ant colony system as shown in the Table 3. It is clear from



92

Table 2. Economic Schedule of committed units for 4-unit system

Hour P1 P2 P3 P4 Load (MW)

1 72 138 140 60 410

2 80 181 179 60 500

3 80 221 214 60 575

4 80 250 260 60 650

5 80 210 205 60 555

6 80 155 155 60 450

7 70 134 136 60 400

8 80 152 153 60 445

9 80 200 195 60 535

10 80 234 226 60 600

11 80 202 198 60 540

12 80 179 176 60 495

13 80 155 155 60 450

14 80 190 186 60 516

15 80 226 219 60 585

16 80 247 238 60 625

17 80 197 193 60 530

18 80 163 162 60 465

19 71 136 138 60 405

20 80 177 175 60 492

21 80 217 211 60 568

22 80 239 231 60 610

23 80 208 202 60 550

24 80 172 171 60 483

Table 3. Comparison results of cost of generation for 4-unit system

Methods Production cost of generation (Rs)

Dynamic programming [11] 26986.40

Branch and bound [11] 26921.94

Ant colony system[11] 26921.94

CCF (PM) 26547.00

the comparison, that the proposed approach gives the better optimum solution for all load demands when compared to the other methods reported in Reference [11].


The proposed approach has been tested on a sample sys-tem the optimum generation schedule for various system de consisting of seven generating units. Table 4 summa-rizes the committed units, operating cost and start up cost. The committed units show that it is necessary to add the start up cost at Hour 10 and Hour 24. Table 5 gives mands. The results do not show any violation of reserve and power demand constraints. In this article [14] there is no shutdown cost and ramp level for any generating units. The proposed method is compared with Tabu Search (TS) and the comparison is shown in Table 6. It is clear from the comparison, that the pro-posed approach gives the better optimum solution for all load demands when compared to the other method re-ported in Reference [14].


The proposed approach has been tested on a sample sys-tem consisting of ten generating units. Table 7 summa-rizes the committed units, operating cost and start up cost. The committed units shows that it is necessary to add startup cost Hour 4 and shutdown cost at Hour 1, Hour 20 and Hour 24. Table 8 gives the optimum generation schedule for various system demands. The ramp level should be maintained for entire 24 hour. The results do not show any violation of reserve and power demand constraints. The proposed method in comparison with Dynamic Programming (DP), Branch and bound and Ant colony system as shown in the Table 9. It is clear from the comparison, that the proposed approach gives the better optimum solution for all load demands when compared to the other methods reported in Reference [11].

The data are obtained for four generator [11], seven- generator [14] and ten-generator [11] as reported in ref-erences. From the results, it was clear that the proposed approach provides solution with close agreement with conventional method. The composite cost function-based solution method has the following distinctive features.

1) The conventional methods gives the committed units and total production cost only but the proposed method gives the committed units, cost and economic schedule for committed units.

2) The proposed approach directly express the opti-mum generation of each unit using CCF coefficients. Hence the computation time for unit commitment and economic schedule becomes an easy task.

3) The iterative steps are completely eliminated.



93


Hour U1 U2 U3 U4 U5 U6 U7 Cost ST.Cost

1 1 1 1 1 1 1 1 86238 0

2 1 1 1 1 1 1 1 77574.66 0

3 1 0 1 1 1 1 1 79388 0

4 1 0 1 1 1 1 1 79161 0

5 1 0 1 1 1 1 1 78823 0

6 1 0 1 1 1 1 1 79642 0

7 1 0 1 1 1 1 1 77394.7 0

8 1 0 1 1 1 1 1 79642 0

9 1 0 1 1 1 1 1 78823 0

10 1 1 1 1 1 1 1 78283.22 2525

11 1 1 0 1 1 0 1 59918 0

12 1 1 0 1 1 0 1 59586 0

13 1 1 0 1 1 0 1 54320 0

14 1 1 0 1 1 0 1 53604 0

15 1 1 0 1 1 0 1 53299 0

16 1 0 0 1 1 0 1 45932 0

17 1 0 0 1 1 0 1 44228 0

18 1 0 0 1 1 0 1 43240 0

19 1 0 0 1 1 0 1 45932 0

20 1 0 0 1 1 0 1 45583 0

21 1 0 0 1 1 0 1 45327 0

22 1 0 0 1 1 0 1 42750 0

23 1 0 0 1 1 0 1 52763 0

24 1 1 1 1 1 1 1 86238 12425

Total cost (Rs) 1542640



94


Hour P1 P2 P3 P4 P5 P6 P7 Load (MW)

1 60 80 98 102 150 150 200 840

2 60 65.52 78.13 80.99 122.36 150 200 757

3 60 0 100 115 150 150 200 775

4 60 0 100 113 150 150 200 773

5 60 0 100 110 150 150 200 770

6 60 0 100 104 164 150 200 778

7 60 0 96.73 100.27 150 150 200 757

8 60 0 100 104 164 150 200 778

9 60 0 100 110 150 150 200 770

10 60 66.93 79.56 82.48 125.03 150 200 764

11 60 80 0 108 150 0 200 598

12 60 80 0 105 150 0 200 595

13 60 70 0 85 130 0 200 545

14 60 68 0 83 127 0 200 538

15 60 67 0 83 125 0 200 535

16 60 0 0 82 124 0 200 466

17 57 0 0 77 115 0 200 449

18 55 0 0 74 110 0 200 439

19 60 0 0 82 124 0 200 466

20 60 0 0 81 122 0 200 463

21 60 0 0 80 120 0 200 460

22 54 0 0 73 107 0 200 434

23 60 0 0 120 150 0 200 530

24 60 80 98 102 150 150 200 840

Table 6. Comparison results of cost of generation for 7–unit system

Method Production cost of generation (Rs)

Tabu Search [14] 1543179

CCF (PM) 1542640



95


Hour U1 U2 U3 U4 U5 U6 U7 U8 U9 U10 Cost ST.cost

1 1 1 1 1 0 1 1 1 1 1 2422.47 0

2 1 1 1 1 0 1 1 1 1 1 2589.33 0

3 1 1 1 1 0 1 1 1 1 1 2873.41 0

4 1 1 1 1 1 1 1 1 1 1 3296 180

5 1 1 1 1 1 1 1 1 1 1 3579 0

6 1 1 1 1 1 1 1 1 1 1 3906 0

7 1 1 1 1 1 1 1 1 1 1 4146 0

8 1 1 1 1 1 1 1 1 1 1 4230 0

9 1 1 1 1 1 1 1 1 1 1 4378 0

10 1 1 1 1 1 1 1 1 1 1 4378 0

11 1 1 1 1 1 1 1 1 1 1 4317 0

12 1 1 1 1 1 1 1 1 1 1 4230 0

13 1 1 1 1 1 1 1 1 1 1 4146 0

14 1 1 1 1 1 1 1 1 1 1 3932 0

15 1 1 1 1 1 1 1 1 1 1 4038 0

16 1 1 1 1 1 1 1 1 1 1 3932 0

17 1 1 1 1 1 1 1 1 1 1 3579 0

18 1 1 1 1 1 1 1 1 1 1 3161 0

19 1 1 1 1 1 1 1 1 1 1 2965 0

20 1 1 1 1 0 1 1 1 1 1 2718 0

21 1 1 1 1 0 1 1 1 1 1 2611 0

22 1 1 1 1 0 1 1 1 1 1 2505.23 0

23 1 1 1 1 0 1 1 1 1 1 2589.33 0

24 1 1 0 1 0 1 1 1 1 1 2345 0

Total cost (Rs) 83180.77



96


Hour P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 Load (MW)

1 200 136.31 150 258.54 0 120.78 85.34 89.32 69.71 60 1170

2 200 147.96 150 285.08 0 136.67 94.17 97.90 78.22 60 1250

3 200 173.45 150 343.09 0 150 113.46 110 80 60 1380

4 200 176 150 349 179 150 116 110 80 60 1570

5 200 203 150 409 208 150 120 110 80 60 1690

6 200 232 150 477 241 150 120 110 80 60 1820

7 200 255 150 520 265 150 120 110 80 60 1910

8 200 270 150 520 280 150 120 110 80 60 1940

9 200 320 150 520 280 150 120 110 80 60 1990

10 200 320 150 520 280 150 120 110 80 60 1990

11 200 300 150 520 280 150 120 110 80 60 1970

12 200 270 150 520 280 150 120 110 80 60 1940

13 200 255 150 520 265 150 120 110 80 60 1910

14 200 235 150 482 243 150 120 110 80 60 1830

15 200 244 150 503 253 150 120 110 80 60 1870

16 200 235 150 482 243 150 120 110 80 60 1830

17 200 203 150 409 208 150 120 110 80 60 1690

18 200 165 150 323 165 150 107 110 80 60 1510

19 200 150 150 292 150 141 97 100 80 60 1420

20 200 157 150 307 0 150 101 105 80 60 1310

21 200 150 150 288 0 139 95 99 79 60 1260

22 200 142.14 150 271.81 0 128.72 89.76 93.61 73.96 60 1210

23 200 147.96 150 285.08 0 136.67 94.17 97.90 78.22 60 1250

24 200 153 0 300 0 145 100 102 80 60 1140

Table 9. Comparison results of cost of generation for 10-unit system

Methods Production cost of generation (Rs)

Dynamic Programming [11] 83652.40

Branch and bound [11] 83475.25

Ant colony system [11] 83491.42

CCF (PM) 83180.77



97

6. Conclusions

An effective, robust unit commitment solution is a nec-essary contribution to the operating on/off plans of the generating units. In this paper, a new composite cost function based solution algorithm for solving the unit commitment problem is presented. The proposed algo-rithm uses the composite cost coefficients to select the committed units and give the economic schedule for each specific hour. This new algorithm produces better results than the Branch and bound, Dynamic programming, Ant colony, and Tabu search methods in addition to satisfac-tion of the system constraints. The proposed algorithm is most suitable for system having smaller number of units. From the results, it is clear that the proposed method provides the quality solution with low cost and has a po-tential for on-line implementation.

7. Acknowledgements

The authors gratefully acknowledge the authorities of Annamalai University, Annamalainagar, Tamilnadu, In-dia, for their continued support, encouragement, and the extensive facilities provided to carry out this research work.

REFERENCES [1] A. J. Wood and B. F. Wollenberg, “Power Generation,

Operation and Control,” Wiley, New York, 1984.

[2] M. Madrigal and V. H. Quintana, “An Inte-rior–Point/Cutting-Plane Method to Solve Unit Commit-ment Problems,” IEEE Transactions on Power Systems, Vol. 15, No. 3, August 2000, pp. 1022-1027.

[3] J. M. Arroyo and A. J. Conejo, “A Parallel Repair Ge-netic Algorithm to Solve the Unit Commitment Problem,” IEEE Transactions on Power Systems, Vol. 17, No. 4, November 2002, pp. 1216-1224.

[4] H. Y. Chen and X. F. Wang, “Cooperative Coevolution-ary Algorithm for Unit Commitment,” IEEE Transactions on Power Systems, Vol. 17, No. 1, February 2002, pp. 128-133.

[5] Q. Z. Zhai, X. H. Guan and J. Cui, “Unit Commitment with Identical Units: Successive Subproblem Solving Method Based on Lagrangian Relaxation,” IEEE Trans-actions on Power Systems, Vol. 17, No. 4, November 2002, pp. 1250-1257.

[6] A. L. Motto and F. D. Galiana, “Unit Commitment with Dual Variable Constraints,” IEEE Transactions on Power Systems, Vol. 19, No. 1, February 2004, pp. 330-338.

[7] C. C. A. Rajan and M. R. Mohan, “An Evolutionary Pro-gramming-Based Tabu Search Method for Solving the Unit Commitment Problem,” IEEE Transactions on Power Systems, Vol. 19, No. 1, February 2004, pp. 577- 585.

[8] B. Lu and M. Shahidehpour, “Unit commitment with Flexible Generating Units,” IEEE Transactions on Power Systems, Vol. 20, No. 2, May 2005, pp. 1022-1034.

[9] T. Li and M. Shahidehpour, “Price-Based Unit Commit-ment: A Case of Lagrangian Relaxation Versus Mixed Integer Programming,” IEEE Transactions on Power Sys-tems, Vol. 20, No. 4, November 2005, pp. 2015-2025.

[10] A. Y. Saber, T. Senjyu, T. Miyagi, N. Urasaki and T. Funabashi, “Fuzzy Unit Commitment Scheduling Using Absolutely Stochastic Simulated Annealing,” IEEE Transactions on Power Systems, Vol. 21, No. 2, May 2006, pp. 955-964.

[11] S. P. Simon, N. P. Padhy and R. S. Anand, “An Ant Col-ony System Approach for Unit Commitment Problem,” Electrical Power and Energy Systems, Vol. 28, No. 5, 2006, pp. 315-323.

[12] M. Carrion and J. M. Arroyo, “A Computationally Effi-cient Mixed-Integer Linear Formulation for the Thermal Unit Commitment Problem,” IEEE Transactions on Power Systems, Vol. 21, No. 3, August 2006, pp. 1371- 1378.

[13] D. N. Simopoulos, S. D. Stavroula and C. D. Vournas, “Reliability Constraints Unit Commitment Using Simu-lated Annealing,” IEEE Transactions on Power Systems, Vol. 21, No. 4, November 2006, pp. 1699-1706.

[14] A. A. Victoire and A. E. Jeyakumar, “A Tabu Search Based Hybrid Optimization Approach for a Fuzzy Mod-elled Unit Commitment Problem,” Electric Power Sys-tems Research, Vol. 76, No. 6-7, 2006, pp. 413-425.

[15] F. Aminifar, M. Fotuhi-Firuzabad and M. Shahidehpour, “Unit Commitment with Probabilistic Spinning Reserve and Interruptible Load Considerations,” IEEE Transac-tions on Power Systems, Vol. 24, No. 1, February 2009, pp. 388-397.

[16] M. Eslamian, S. H. Hosseinian and B. Vahidi, “Bacterial Foraging Solution to the Unit-Commitment Problem,” IEEE Transactions on Power Systems, Vol. 24, No. 3, August 2009, pp. 1478-1488.




Jinghua Wang1, Hongyou Cui1*, Shuqin Wei1, Shuping Zhuo1, Lihong Wang2, Zhihe Li2, Weiming Yi2

1School of Chemical Engineering, Shandong University of Technology, Zibo, China; 2Shandong Research Center of Engineering and Technology for Clean Energy, Shandong University of Technology, Zibo, China. Email: [email protected] Received May 17th 2010; revised June 18th 2010; accepted June 23rd 2010.

ABSTRACT

Supercritical CO2 extraction was employed to separate simulated and real bio-oils. Effects of extraction pressure, tem-perature and adsorbents on distribution coefficient (or enrichment coefficient) of five representative compounds were investigated using a simulated bio-oil, which was composed of acetic acid (AC), propanoic acid (PA), furfural (FR), acetylacetone (AA) and 2-methoxyphenol (MP). The distribution coefficients of AA, FR and MP between supercritical CO2 phase and liquid phase were bigger than 1.5, while those of AC and PA characteristic of relatively strong polarity were less than 1. Temperature and pressure also had impacts on the distribution coefficients of AA, FR and MP, espe-cially remarkable for AA. The extraction of simulated bio-oil spiked on three adsorbents shows that adsorbents influ-ence extraction efficiency and selectivity by changing intermolecular forces. High extraction pressure and relative low temperature are beneficial to reduce the water content in the extract. In addition, the feasibility of supercritical CO2 extraction of real bio-oil was examined. After extraction in the extraction fraction total ketones increased from 14.1% to 21.15~25.40%, phenols from 10.74% to 31.32~41.25%, and aldehydes from 1.92% to 3.95~8.46%, while the acids significantly dropped from 28.15% to 6.92~12.32%, and water from 35.90% to 6.64~4.90%. In view of extraction effi-ciency, the optimal extraction temperature was determined to be 55. Extraction efficiency of the real bio-oil in-creased with rising pressure. The maximal extraction efficiency of real bio-oil on water-free basis could reach to 88.6%. After scCO2 extraction, the calorific value and stability of the extract fraction evidently increased and the acidity slight decreased with nearly 100% volatility below 140, suggesting potentially applicable as substitute for engine fuel. Keywords: Supercritical, Extraction, Simulated Bio-Oil, Adsorbent, Carbon Dioxide

1. Introduction

Facing upcoming depletion of fossil fuels and increasing environmental concerns, great effort has been devoted in exploration of biomass energy in the past few decades all over the world, not only because it is recognized as one of the most attractive alternative energy resources in the current century but also it is available in abundance, re-newable and environmentally friendly [1]. Biomass as an energy source is considered sustainable since it is CO2 neutral in the life cycle, causing almost zero net emis-sions of CO2. Moreover, it contains negligible contents of sulfur, nitrogen, and ash, and gives much lower emis-sion of SO2, NOx, and soot, by combustion, than the conventional fossil fuels [2]. Among the various biomass utilization technologies, conversion of biomass into bio-oil by fast pyrolysis has been shown promising for internal combustion engine fuels and high value added chemicals from the viewpoint of efficiency and econom-

ics [3-5]. Bio-oil is a dark brown, crude oil-like liquid mixture,

usually obtained from thermal biomass decomposition under moderate pyrolysis temperature with very high heating rate and short vapor residence time [6]. Being a non-thermodynamic equilibrium mixture, it is highly viscous, non-volatile, poor in heat value, and corrosive [7]. Moreover, it is such a complicated mixture, contain-ing almost all kinds of oxygenated organic compounds, including alcohols, aldehydes, carboxylic acids, ketones, esters, saccharides, phenols, guaiacols, syringols, furans and multifunctional compounds [8], that severely re-strains its direct application in vehicle engines as fuels. To meet the demands for substituting fossil fuels, there-fore, considerable endeavors have been contributed to upgrading bio-oils, such as catalytic hydrotreatment [9,10], catalytic cracking [11], emulsification [12] and catalytic esterification [13]. Unfortunately, all of these



99

technologies have some shortcomings due to the com-plexity and thermal instability of bio-oil.

Separation and refining of bio-oil possess advantages in keeping original components, simple operation, low investment, and probably obtaining valuable chemicals. Several conventional separation techniques, like distilla-tion, molecular distillation, solvent extraction and ad-sorption, have been attempted, but they seemed not to be very satisfying [14,15]. Recently, supercritical technol-ogy was also introduced into this field, but mainly fo-cused on the bio-oil upgrading by esterification and py-rolysis in supercritical alcohols [16], as well as hydrode-oxygenation of phenols in supercritical n-hexane [17]. Supercritical CO2 extraction has been demonstrated a powerful tool in separation of thermal sensitive chemi-cals in the past few decades because of the relatively low critical pressure (73.8 atm) and critical temperature (31.1) of CO2. Compared with organic solvents, carbon dioxide is nontoxic, nonflammable, noncorrosive, cheap, and readily available in abundance with high purity. Su-percritical CO2 extraction has been extensively used in extracting natural products. Very recently, scCO2 extrac-tion of bio-oil was attempted by Rout et al. [18]. Their research results clearly showed that most of the valuable compounds such as furanoids, pyranoids and benzenoids could be enriched into the extract fraction, while water could be removed effectively after extraction. As a result, the calorific value of the bio-oil was improved.

In this work, supercritical CO2 extraction of bio-oil was systemically studied. Based on the experimental estimation of two-phase distribution coefficient (or en-richment coefficient) of a simulated bio-oil, effects of extraction pressure, temperature and adsorbents on su-percritical CO2 extraction of real bio-oil were investi-gated and the quality of the bio-oil after extraction was evaluated in comparison with the original oil.

2. Experimental

2.1 Experimental Materials

The bio-oil used in the extraction experiment was made by flash pyrolysis of pulverized corn stalk with size of 0.1 mm~0.25 mm in the temperature range of 477~480. 5A molecular sieve, activated carbon, silica and Karl- Fischer reagent were purchased from Kermel chemical Reagent Co. (Tianjin, China). 1-methoxy-2-propanol was obtained from Acros Organics Co. with purity no less than 98.5%. All other chemicals were analytic reagents with purities more than 98% and used as received. For preparation of a simulated bio-oil, five compounds were chosen to be representative compounds and directly mixed in approximate connection with the composition in the real bio-oil. Namely, it was composed of 14.24% acetylacetone, 24.87% acetic acid, 5.87% 2-methoxyphenol, 6.39% propanoic acid, 12.37% furfural and 36.25% de-

ionized water respectively.

2.2 Supercritical Fluid Extraction Setup

All supercritical extraction experiments were conducted on a supercritical CO2 extraction setup as shown in Fig-ure 1, which mainly consists of a CO2 reservior, a CO2

delivering pump, an autoclave extractor (150 mL, 316 stainless steel) equipped with electromagnetic agitation and water-bath heating system, a separator (100 mL, 316 stainless steel) and a wet gas flow meter. The tempera-tures both for the extractor and the separator were meas-ured by thermal couples and the pressures were gauged by manometers. When performing extraction experiment, bio-oil or adsorbed bio-oil was firstly charged to the ex-tractor, followed by introduction of CO2 using the CO2 delivering pump, and then heated up to the desired ex-traction temperature and pressure. The temperature and the pressure were controlled in accuracies of 1 and 0.1 MPa respectively.

In the case of adsorbed bio-oil equilibration experi-ment, 12 g bio-oil was spiked on 30 g absorbent (silica, molecular sieve or active carbon) in advance and allowed to store in a closed vessel for 24 h at room temperature. Before extraction, they were wrapped up with a piece of nylon filtration fiber and then put into the extractor. For the phase behavior experiments without absorbents, elec-tromagnetic agitator was used to enhance transportation between scCO2 phase and liquid phase kept at constant temperature and pressure for at least 1.0 h to approaching equilibrium, while in the experiments for absorbed bio-oil where agitator couldn’t be used, the equilibration time extended to at least 3.0 h.

The sampling method is similar to that in the literature [19]. Briefly, In phase equilibration experiments when sampling from liquid phase (or scCO2 phase), as shown in Figure 1, valve 2 (or valve 4) was first turned off, and then valve 1 (or valve 3) was opened until a pressure balance between the sampling tube and the extractor was reached. Afterward, valve 1 (or valve 3) was turned off and valve 2 (or valve 4) slowly opened to depressurize and collect the sample in a flask. The sampling tubes are made of a stainless steel tube with an inner diameter of 4 mm and a length of 50 mm.

The scCO2 extraction experiments were carried in an intermittent mode. After static extraction at the required temperature and pressure for 30 min or 1.0 h, depending on the cases with or without the absorbents, CO2 effluent was depressurized into the separator. Then the releasing valve was closed and CO2 was recharged up to the origi-nal pressure. In this way, the operation ran several times. The CO2 volume was measured by a wet gas meter and calculated into volume at standard state (0 and 1 atm) according to the ideal gas state equation, in a unit of normal liter, NL.



100

PExtractor

CO2 reservior CO2 delivering pump Electromagnetic stirrer

PT

Airing

wet gas meter

PT

M

Separator

Solvent

Water-bath

Sampling tube

Valve 1 Valve 2

Valve 3 Valve 4

Figure 1. Schematic diagram of the experimental setup for supercritical CO2 extraction of simulated/real bio-oil 2.3 Methods of Analysis

The quantitative analysis of simulated bio-oil was ac-quired on a gas chromatography (Varian CP-3800), which was equipped with a 50 m × 0.25 mm × 0.33 μm PEG capillary column. Both the inlet temperature and the detector temperature were fixed at 250. Column tem-perature was controlled by a temperature program, which is: starting at 80 for 1 min, heating to 130 at a ramp rate of 10/min, followed by heating to 200 at a ramp rate of 30/min, and finally holding at 200 for 8 min. 1-Methoxy-2-propanol was used as an internal standard compound.

For the real bio-oil analysis, both qualitative and quan-titative analyses were carried on a GC-MS system. (GC6890-MS5973N, Agilent Co.)equipped with 60 m × 0.25 mm × 0.25 μm Innowax 19091N-136 capillary col-umn. He was used as carrier gas with a flow rate of 1mL/min and split ratio was 80. Inlet temperature was fixed at 250. A ramp temperature program was adopted, which started at 10/min from 60 to 120, then heated at 5/min to 200, and held at 200 for 8 min. All the acquired component compositions were based on area normalization. Water content is determined by Karl-Fischer method [20]. The calorific value was de-termined in a static bomb calorimeter [21].

Distribution coefficient (Di), enrichment coefficient (Ri) and percentage extraction (Ei) were calculated ac-cording to the following equation.

ii

i

YD

X

100%ii

i

mR

M

100%i

mE

M

where Yi and Xi depict the mass percentage of compo-nent i in the extract and in the liquid faffinate after re-moval of CO2, %; mi and Mi represent the mass of com-ponent i in the extract and the total mass of component i in the simulated bio-oil respectively, g; i indicates acetic acid (AC), propanoic acid (PA), furfural (FR), acety-lacetone (AA) or 2-methoxyphenol (MP); m and M de-scribe the mass of extract and the total mass of bio-oil, g.


3.1 Two-Phase Distribution of Simulated Bio-Oil Extraction

The effects of CO2 pressure on the components distribu-tion coefficients between supercritical CO2 phase and liquid phase was investigated at 45 in a pressure range of 7~17 MPa (Figure 2). 24.0 g simulated bio-oil was used in each experiment.

6 8 10 12 14 16 180

1

2

3

4

5

6

Di

P, MPa

Acetylacetone Guaiacol Furfural Acetic acid Propanoic acid

Figure 2. Distribution coefficients of various components in simulated bio-oil between supercritical CO2 phase and liquid phase as a function of pressure Extraction conditions: 45, pressure range from7 to17 MPa, 24.0 g simulated bio-oil



101

Figure 2 shows that the distribution coefficients of various components were very close to each other at low pressure (low CO2 density). Such a phenomenon might be caused by the combined effects of vaporization and scCO2 solvating power. Except for MP (b. p. 205), the boiling points at atmosphere of AC (118), PA (140.7), FR (162) and AA (140.5) are relatively close, suggesting they have relatively approximative partial vapor pressures. MP is weak in polarity and readily ex-tracted by nonpolar CO2, where as AC, PA, FR and AA relatively difficult to be extracted as they are polar com-pounds. As the pressure rising, DAA increased remarkably from 1.5 to 4, DFR and DMP increased gradually, while DAC and DPA varied very little. Strong intermolecular interactions between homogeneous species and hetero-geneous species of acetic, propanoic acid and water, ca-pable of forming hydrogen bonds, might be a good ex-planation. Even under high pressure the polarity of CO2 is still very weak and hard to break up these hydrogen bonds, resulting in difficulty in extraction of AC and PA into supercritical CO2 phase. In contrast, in spite of polar compounds and capable of forming intermolecular hy-drogen bonds with water, AA and FR cannot form ho-mogeneous intermolecular hydrogen bonds. Conse-quently, they exhibited higher distribution coefficient under high CO2 density than under low CO2 density. Considering the fact bio-oil typically contains 20~40% water, therefore, hydrogen bonding effect might be bene-ficial to selective isolation or enrichment of some chemicals in supercritical CO2 extraction.

Figure 3 depicts the effect of temperature on the dis-tribution coefficients in simulated bio-oil at constant pressure of 15.0 MPa. Likewise, 24.0 g simulated bio-oil was used in each experiment.

35 40 45 50 55 600.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Di

T, 0C

Acetylacetone Guaiacol Furfural Propanoic acid Acetic acid

Figure 3. Effect of temperature on distribution coefficients of various components in simulated bio-oil Experimental conditions: 15.0 MPa, 24.0 g simulated bio-oil

In the experimental temperature range of 35~60, temperature had very weak effect on DAC and DPA, around 0.6 and 1.0 respectively. DPA is always bigger than DAC, indicating polarity played a much more impor-tant role than partial vapor pressure as the partial vapor pressure of AC is bigger than that of PA at same tem-perature. Low distribution coefficients both for AC and PA are attributed to the formation of hydrogen bond with water. DAA, DFR and DMP increased at the beginning and then decreased with elevating the temperature. Such an interesting tendency might be related to the CO2 density because at constant pressure CO2 density decreased and resulted in a drop in its dissolving powder when raising the temperature. On the other hand, the vapor pressures increased with temperature. As a result, the contents of low boiling point compounds in the extract tend to rise. Acetylacetone has a relative low boiling point and cannot form hydrogen bond with water. For this reason, it is not surprise that DAA showed the most evident temperature dependence.

Adding absorbents might have also effect on scCO2 extraction, in theory, because it can influence the inter-molecular forces of various species. 5A molecular sieve, activated carbon and silica were chosen as the absorbents. In each experiment, 5.0 g simulated bio-oil was spiked on 20.0 g absorbent. Due to the difficulties in knowing the exact amount of the various components in the ab-sorbents after extraction, enrichment coefficient was used instead of distribution coefficient. Figure 4 shows the enrichment coefficients at 45 and 18.3 0.1 MPa in the presence of various absorbents.

In the presence of absorbent, the interaction force be-tween the component and the absorbent depends on not only the component but also the type and internal texture of the absorbent [22,23]. Competitive adsorption be-tween species occurred on the surface of the absorbent after spiking. For this reason, the intermolecular forces between different components could be changed and thus influence their extractability by supercritical CO2. In contrast to the case in absence of absorbent, the enrich-ment coefficients of FR, AA and PA declined, whereas those for MP and AC rose. Among all of the five com-pounds, AC is the strongest and MP is the weakest in polarity. All of the other three compounds whose en-richment coefficients declined in the presence of absor-bents are moderate in polarity. Therefore, extraction se-lectivity might be linked to the surface polarity of the absorbents. AA, FR and MP are high value chemicals. Thus it is reasonable to expect much more of these com-pounds could be preferentially extracted into the scCO2 phase. In this context, except for MP, all three adsorbents seemed not to be very satisfying. Therefore, functional-ized adsorbents should be taken into consideration in the future research work.

Water content in the bio-oil is of significance since it



102

AA AC FR PA MP0.0

0.4

0.8

1.2

1.6

2.0

2.4

2.8

3.2

Ri

without adsorbent 5A molecular sieve Silica Active carbon

Figure 4. Enrichment coefficients of various components in simulated bio-oil in the presence of adsorbents Experimen-tal conditions: 45 and 18.3 0.1 MPa, 5.0 g simulated bio-oil, 20.0 g absorbents influences both the heat value and combustibility in the internal combustion engines. Effects of extraction pres-sure and temperature on the water content in the extract were also surveyed (Figure 5 and Figure 6).

It can be seen in Figure 5, water content in the extract decreased with raising the pressure. For example it was 13.28% at 9.4 MPa and dropped drastically to only 5.52% at 18.4 MPa. This was in accordance with the ex-planation in the section of pressure effect on distribution coefficients. The volatility of a compound at low pres-sure is a dominant factor while the dissolving capacity of CO2 becomes dominant at high pressure. High water content in the extract is attributed to the lowest boiling point and strongest polarity of water among all of the components in the simulated bio-oil.

According to the above explanation, it is expectable to see in Figure 6 that water content in the extract gradually

8 10 12 14 16 18 204

6

8

10

12

14

Wat

er,

wt%

P, MPa

Simulated bio-oil

Figure 5. Effect of extraction pressure on the water content in the extract at 45 Experimental conditions: 45, 5.0 g simulated bio-oil

30 40 50 60 70 80

5

10

15

20

25

30

Wat

er,

wt%

T, oC

Figure 6. Variation of water content in the extract as a function of extraction temperature under 12.0 MPa pres-sure Experimental conditions: 12.0 MPa, 5.0 g simulated bio-oil increased with elevating the temperature at 12.0 MPa. At 75, the water content could go up to as high as 28.3%. Elevating temperature creates two effects on the extrac-tion: 1) reducing the density of CO2 and thus its dissolv-ing power, resulting in poorer extraction efficiency when rising the temperature; 2) increasing the saturated vapor pressure of extractable compounds, as a result, leading to higher extraction efficiency. As a compromise, dissolv-ing power of CO2 plays a dominant role at low tempera-ture, while vapor pressure turns to be a dominant factor at high temperature. Given a fact that the polarity of wa-ter is very strong, its solubility is still very low in CO2 phase even though at low temperature. Therefore, the tendency that the water content increased with tempera-ture is understandable.

3.2 Supercritical CO2 Extraction of a Real Bio-Oil

For the extraction of real bio-oil, pyrolysis bio-oil of corn stalk powder was selected and subjected to extraction by supercritical CO2, whose composition acquired by GC- MS was summarized in Table 1. It could be clearly seen that large amount of water (35.9%) and organic acids (24.51% acetic acid, 2.07% propanoic acid, 1.57% acryl acid) were contained in the original bio-oil, which were dominantly responsible for the corrosiveness of bio-oils. Phenols which resulted from the pyrolysis of the lignin in corn stalk occupied about 10% in bio-oil, mainly existing in the form of phenol, methyl phenol, methoxy phenol and ethyl phenol. Except for the compounds listed in Table 1, other kinds of compounds, like saccharides and charcoals, might also be contained in the bio-oil, but could not be measured by GC-MS and weren’t taken into account in this study.



103

Table 1. Composition of the pyrolysis bio-oil of corn stalk

Compound Content, area%

Compound Content, area%

ethanol 4.19 2-furanmethanol 1.81

water 35.90 3-methyl-2-hydroxy-2-cyclopentene ketone

2.39

1-hydroxy pro-panone

7.56 2-methoxy phenol 1.21

2-cyclopenten ketone

1.69 3-methyl phenol 1.22

1-hydroxy-2-butanone

2.46 Phenol 4.66

acetic acid 24.51 4- methyl phenol 1.73

ethylene glycol 1.92 4-ethyl phenol 1.92

furfural 1.92 2-acryl acid 1.57

propanoic acid 2.07 Others 1.41

3.2.1 Changes of Various Categories of Compounds in the Extracts after scCO2 Extraction

For scCO2 extraction of the real bio-oil, 12.0 g bio-oil was spiked on 30.0 g adsorbents (5A molecular sieve or silica) and then subjected to extraction at 45 and 26.0 MPa. Typical profiles for various compound categories were given in Table 2 when the CO2 volume was 50NL and extraction efficiency was 20%.

Given a fact that the composition of bio-oil varies with the biomass resources, pyrolysis processes, and the op-eration conditions, and most of the compounds in the bio-oil are in very low concentration, our experiments focused on the extractability of various categories of compounds rather than individual compound. Similar to the extraction of simulated bio-oil, intermittent operation mode was used for the extraction of the real bio-oil. The extraction pressure swing range was between 26.0 MPa and 10.0 MPa. All of compounds identified by GC-MS were classified into 8 groups. They are: 1) alcohols in-cluding ethanol, methanol, ethanediol, isobutyl alcohol

furfuryl alcohol, etc; 2) ketones including 2-propanone- 1-hydroxy, 2-cyclopentanone, 1-hydroxy-2-butanone, 2- hydroxy-cyclopentanone, 3-hydroxy-2-butanone, 4-methyl-cycloheptanone, acetyl-acetone, furan acetophenone, 2,2-dimethyl-1-propeneketo-1-cyclohexenyl, etc; 3) acids including acetic acid, propanoic acid, butyric acid, vinyl acid, crotonic acid, 9,10-diene stearic acid, etc; 4) phe-nols including 4-ethyl-2-methoxyphenol, 2,5-dimethyl- phenol, 4-methyl phenol, 3-methoxy-phenol, 3-methyl- phenol, 3-methyl-4-ethyl phenol, 4-ethyl phenol, 2,5-dimethyl-phenol, 2,4-dimethyl-3-dimethyl-etherbase-4-methoxy-phenol, 2,6-dimethyloxy-phenol, 2-me-thoxy-phe-nol, 2-ethoxy-phenol, phenol, etc; 5) aldehydes including furfural, 2-furyl glyoxal, 4-ethoxy-benzaldehyde, 3-hydroxy-4-methoxybenzaldehyde, etc; 6) esters including methyl acetate, 4-methyl-amyl methyl ester, 1-propylene glycol acetate, 2-methyl ethylene, etc; 7) water, and 8) the others which refer to all compounds which were de-tectable by GC but unidentified by MS.

It has been reported that hundreds of compounds might be contained in bio-oils. However, for our original bio-oil only 17 compounds could be identified by GC- MS analysis. Besides the nonvolatile compounds, there were still a variety of compounds whose contents were so low that they couldn’t be detected out by the GC-MS. After scCO2 extraction the compound species detected in the extract rose up to 80, suggesting that scCO2 extrac-tion can be used as a useful pretreatment tool for qualita-tive and quantitative analysis of bio-oils. Table 2 shows the enrichment of all categories of compounds when the extraction efficiencies on water-free basis were about 20%.

Compared with the original bio-oil, there was only a little change of alcohols in the extract fraction after ex-traction which might be attributed to the low content al-cohols in original bio-oil. Water content decreased sig-nificantly, about one fifth of that in the original bio-oil (35.90%) suggesting that scCO2 extraction was very ef-fective in reducing the water content and thus boosting up the heat value of bio-oil. The total ketones in bio-oil increased from 14.1% to 21.15~25.40%, phenols from 10.74% to 31.32~41.25%, aldehydes from 1.92% to 3.95~8.46%, while the acids were significantly reduced from 28.15% to 6.92~12.32%. These tendencies are in

Table 2. Typical profiles for various categories of compounds in the extracts after scCO2 extraction

Treatment condition Alcohols Water Ketones Acids Phenols Aldehydes Esters Others

Original bio-oil 7.92 35.90 14.1 28.15 10.74 1.92 0 1.41

Without adsorbent 2.55 6.64 21.15 11.98 35.22 3.95 1.84 18.41

5A molecular 7.03 5.28 25.40 12.32 31.32 3.99 1.76 12.91

Silica gel 4.81 4.91 21.99 6.95 41.25 8.46 1.35 8.55



104

accordance with the experimental results of the simulated bio-oil.

In summary, although the selectivity differed from one compound to another, depending on the types of absor-bents, water and acids can be greatly reduced after scCO2 extraction, indicating significant heat value increment and corrosiveness reduction (lower pH) from the view-point of upgrading bio-oil. In addition, selective extrac-tion or enrichment of high value-added chemicals might be probable by selecting appropriate adsorbent.

3.2.2 Effect of Pressure on Extraction Efficiency Effect of pressure on extraction of real bio-oil was stud-ied at fixed temperature of 45. Figure 7 shows the pressure dependence as a function of CO2 consumption in standard volume, NL. In each experiment, 30.0 g original bio-oil was used.

As shown in Figure 7, extraction pressure had strongly positive influence on the extraction efficiency of real bio-oil. The higher the extraction pressure at same CO2 volume, the higher extraction efficiency was. When the CO2 volume was 560 NL, for example, extraction efficiencies of the bio-oil were respectively 8.00% at 7.0 MPa, 24.20% at 10.0 MPa, 36.53% at 15.0 MPa, 47.59% at 20.0 MPa, 58.85% at 25.0 MPa and 77.58% at 30.0 MPa. Since the dissolving power of CO2 mainly de-pended on by its density and was tunable by modifying the pressure, it is easy to understand the pressure effect on extraction efficiency. In addition, pressure also affects the polarity of CO2. The polarity of scCO2 has been con-firmed to be increased with elevating the pressure at con-stant temperature. Therefore, the composition of various compounds in the extraction fraction varied with the pressure. It is reasonable to expect that the compounds with weak polarity are preferentially extracted at low pressure since CO2 is a nonpolar compound, while the compounds with strong polarity can be extractable at

0 100 200 300 400 500 600

10

20

30

40

50

60

70

80

Ei,

%

Volume of CO2, NL

7.0 MPa 10.0 MPa 15.0 MPa 20.0 MPa 25.0 MPa 30MPa

Figure 7. Extraction efficiency of bio-oil as a function of CO2 volume under different pressure Experimental condi-tions: 45, 30.0 g bio-oil

high pressure although their solubility in scCO2 might still be very low. According to this, the solubility of wa-ter in scCO2 phase should also be increased. However, the experimental results showed that the water content in the extract decreased with elevating the pressure at con-stant temperature. To explain this, the competitive ex-traction between various compounds and the phase be-havior in such a complicated system need to be taken into account.

3.2.3. Effect of Temperature The effect of extraction temperature on extraction effi-ciency of bio-oil was surveyed at 30.0 MPa using 30.0 g bio-oil. Figure 8 shows the variation of extraction effi-ciency at different temperature as a function of CO2

volume. Extraction temperature didn’t show remarkable impact

on the extraction efficiency in the experimental tempera-ture range of 45~65. As discussed in Subsection 3.1, temperature in scCO2 extraction has two adverse effects. Therefore, it is not surprised to see that the extraction conducted at 55 afforded the highest extraction effi-ciency, which might be a compromise of these two ef-fects.

It is well known that modified CO2 using low molecu-lar weight organic compounds like methanol or acetic acid, has been shown to be more powerful in extracting polar substances than pure CO2. Considering that ethanol and methanol are available from conversion of biomass and can be used directly as engine fuels, they were em-ployed in the investigation of the impacts of modifiers on the extraction. When performing the experiment, certain amount of methanol or ethanol was directly poured into 30.0 g bio-oil. The experimental results at constant con-sumption of CO2, 156NL, showed that extraction effi-ciency could be increased about 20% and 10% respec tively compared to the case without modifiers when 3.0 g

0 100 200 300 400 500 60010

20

30

40

50

60

70

80

90

Ei,%

Volume of CO2,NL

45oC

55oC

65oC

Figure 8. Temperature dependence of bio-oil extraction efficiency as a function of CO2 volume at 30MPa Experi-mental conditions: 30.0 MPa, 30.0 g bio-oil



105

modifier was used in 30.0 g bio-oil. Evidently, methanol is more eligible as a modifier than ethanol in enhancing the dissolving power of scCO2. It should be cautious that, however, the water content in the extract was significant higher when excess methanol was used than that in the case of pure CO2 extraction.

3.2.4 Property Comparison between Original Bio-Oil and Extracted Oil

One can see that in Table 3, pH value increased from 2.10 to 4.1~4.5 and the calorific value from 13.95 to 18.59~25.41 kJ/kg in different treatment cases, while the density decreased from 1.15 to 0.92~0.98 and the mois-ture from 35.9% to 4.29~6.64%. The stability of the bio-oils was judged by after 6 month storage under room temperature to observe whether phase splitting or no-ticeable change in viscosity occurred. The original bio- oil was dark brown opaque liquid with high viscosity, increased noticeably in viscosity, and phase splitting happened after storage, while the extracts from scCO2 extraction were light brown transparent liquids with rela-tively low viscosities and hadn’t noticeable changes in viscosity and phase splitting, implying relatively high stability.

3.2.5 Bio-oil Volatility Volatility is vital in evaluating transport fuels since it relates to the atomization and combustion performance. To estimate the volatility of bio-oil, 2.0 g bio-oil was accurately weighed and allowed to temperature pro-grammed evaporating in an open dish in air drying oven. The mass variation was determined by analytical balance. The extract and the faffinate used were obtained from the scCO2 extraction under 35 and 30.0 MPa with extrac-tion efficiency of 88.6% (water-free basis). Figure 9 shows temperature dependence of bio-oil volatility as a function of evaporation time.

As shown in Figure 9, the extract fraction showed evi-dently higher in volatility than the original bio-oil, while

the faffinate showed significantly lower. For example, at 60 for 15 min, the volatility percentages of original bio-oil, extract and faffinate were 50.32%, 68.98% and 45.34%, respectively. When elevating temperature to 140, more than 97.69% of extract became volatile, whereas only 89.56% and 83.92% for original bio-oil and faffinate. If taking the fact that most of water was left in the faffinate into account, the volatility experiments strongly suggested that the quality upgrading of the ex-tract faction in view of potential application in combus-tion engine since the nonvolatile compounds were mainly left in the faffinate after extraction.

4. Conclusions

Supercritical CO2 extraction was an effectively powerful tool in separation and upgrading of bio-oil. By adjust-ment of operation temperature, pressure, and using ab-sorbents, weakly polar compounds are capable of selec-tive separation and enrichment into scCO2 phase. The

0.0 0.5 1.0 1.5 2.0 2.5 3.0

0

20

40

60

80

100

Vo

latil

ity p

erc

ent

age

, %

Time, h

Original bio-oil Extract Faffinate

140oC120oC100oC80oC60oC

Figure 9. Variation of volatile as a function of time at dif-ferent temperatures

Table 3. Properties of bio-oil and extracts at different conditions

Properties Original bio-oil Without adsorbent 5A molecular Silica gel

pH 2.10 4.30 4.11 4.53

Density g.cm-1 1.1500 0.9521 0.9288 0.9837

H2O content, % 35.90 6.64 5.28 4.29

Calorific value, kJ/kg 13.95 18.59 23.55 25.41

dark brown light brown light brown light brown Appearance

opaque Transparent transparent transparent

Stability unstable Stable stable stable



106

adsorbent can affects the intermolecular forces between different components. Therefore, valuable chemicals might be expected to be isolated from bio-oil by selecting appropriate adsorbent. High extraction pressure and rela-tive low extraction temperature favor in effectively re-ducing the water content in extracted bio-oil. Extraction efficiency of real bio-oil increased with raising the pres-sure. The maximum extraction efficiency of supercritical CO2 extract bio-oil reached to 88.6% on water-free basis at 30.0 MPa and 35. After scCO2 extraction, the quality of bio-oil can be improved significantly from the view-point of water content reduction, acidity, calorific value, stability and appearance. Water content could be reduced to one fifth of that of original bio-oils. Calorific value could increase to be doubled. pH value went up from 2.1 to above 4.1. The extract fraction showed nearly 100% volatility below 140 since most of the nonvolatile compounds was left in the faffinate.

5. Acknowledgements

This research is financial supported by Natural Science Foundation of Shandong Province (Grant No. ZR2009 BL023) and National High Technology Plan (863) Pro-ject No. 2009AA05Z401 Granted by the Ministry of Science and Technology of the People’s Republic of China.

REFERENCES [1] M. Baratieri, P. Baggio, L. Fiori and M. Grigiante, “Bio-

mass as an Energy Source: Thermodynamic Constraints on the Performance of the Conversion Process,” Biore-source Technology, Vol. 99, No. 15, 2008, pp. 7063- 7073.

[2] J. L. Zheng and W. M. Yi, N. N. Wang, “Bio-Oil Produc-tion From Cotton Stalk,” Energy Conversion and Man-agement, Vol. 49, No. 6, 2008, pp. 1724-1730.

[3] A. V. Bridgwater and G. V. C. Peacocke, “Fast Pyrolysis Processes for Biomass,” Renewable and Sustainable En-ergy Reviews, Vol. 4, No. 1, 2000, pp. 1-73.

[4] C. David, O. Anja and S. Yrjö, “Power Generation Using Fast Pyrolysis Liquids from Biomass,” Renewable and Sustainable Energy Reviews, Vol. 11, No. 6, 2007, pp. 1056-1086.

[5] A. V. Bridgwater, D. Meier and D. Radlein, “An Over-view of Fast Pyrolysis of Biomass,” Organic Geochemis-try, Vol. 30, No. 12, 1999, pp. 1479-1493.

[6] M. Peter, “Energy Production from Biomass (Part 2): Conversion Technologies,” Bioresource Technology, Vol. 83, No. 1, 2002, pp. 47-54.

[7] Q. Lu, W. Z. Li and X. F. Zhu, “Overview of Fuel Prop-erties of Biomass Fast Pyrolysis Oils,” Energy Conver-sion and Management, Vol. 50, No. 5, 2009, pp. 1376- 1383.

[8] T. A. Milne, F. Agblevor, M. Davis, et al., “A Review of

Chemical Composition of Fast Pyrolysis Oils,” In: A. V. Bridgwater, Ed., Developments in Thermochemical Bio-mass Conversion, London, Blackie Academic & Profes-sional, 1997, pp. 409-424.

[9] V. A. Yakovlev, S. A. Khromova, O. V. Sherstyuk, et al., “Development of New Catalytic Systems for Upgraded Bio-Fuels Production from Bio-Crude-Oil and Biodiesel,” Catalysis Today, Vol. 144, No. 3-4, 2009, pp. 362-366.

[10] Y. Xu, T. J. Wang, L. L. Ma, et al., “Upgrading of Liquid Fuel from the Vacuum Pyrolysis Of Biomass over the Mo–Ni/γ-Al2O3 Catalysts,” Biomass and Bioenergy, Vol. 33, No. 8, 2009, pp. 1030-1036.

[11] L. Garcia, R. French, S. Czernik, et al., “Catalytic Steam Reforming of Bio-Oils for the Production of Hydrogen: Effects of Catalyst Composition,” Applied Catalysis A: General, Vol. 201, No. 2, 2000, pp. 225-239.

[12] I. Michio, S. Maria and H. Ed, “Emulsification of Pyroly-sis Derived Bio-Oil in Diesel Fuel,” Biomass and Bio-energy, 2003, Vol. 24, No. 3, pp. 221-232.

[13] Y. Xu, J. Chang, Q. Zhang, et al., “Upgrading of Bio-Oil by Esterification over Solid Base Catalyst,” Petrochemi-cal Technology, Vol. 35, No. 1, 2006, pp. 615-618.

[14] F. Mahfud, H. Melia′n-Cabrera, I. R. Manurung and H. J. Heeres, “Upgrading of Flash Pyrolysis Oil by Reactive Distillation Using a High Boiling Alcohol and Acid,” Trans IChemE, Part B, Process Safety and Environmental Protection, Vol. 85, No. B5, 2007, pp. 466-472.

[15] B. Tuya, M. Chaala, C. R. Garcia-Perez, et al., “Colloidal Properties of Bio-Oils Obtained by Vacuum Pyrolysis of Softwood Bar Characterization of Water-Soluble and Water-Insoluble Fraction,” Energy Fuels, Vol. 18, No. 3, 2004, pp. 704-712.

[16] J. Peng, P. Chen, H. Lou and X. M. Zheng, “Catalytic Upgrading of Bio-Oil by HZSM-5 in Sub- and Su-per-Critical Ethanol,” Bioresource Technology, Vol. 100, No. 13, 2009, pp. 3415-3418.

[17] H. H. A. Hussain, “Supercritical Hexane Extraction: a New Solvent System for Extracting Bio-Oil,” Division of Fuel Chemistry, Vol. 53, No. 2, 2008, pp. 862-864.

[18] P. K. Rout, M. K. Naik, S. N. Naik, et al., “Supercritical CO2 Fractionation of Bio-oil Produced from Mixed Bio-mass of Wheat and Wood Sawdust,” Energy & Fuels, Vol. 23, No. 12, 2009, pp. 6181-6188.

[19] H. Y. Cui, T. Wang, F. J. Wang, et al., “Kinetic Study on One-pot Synthesis of Dimethyl Carbonate in Supercritical CO2 Conditions,” Industrial & Engineering Chemistry Research, Vol. 43, No. 24, 2004, pp. 7732-7739.

[20] H. Wang, R. H. Liu, C. M. Zhang and J. D. Luan, “An Experimental Study on Determination of the Water Con-tent in Bio-Oil by Karl-Fischer Titration,” Renewable Energy, Vol. 3, No. 1, 2005, pp. 17-20.

[21] C. M. Jiao and F. He, “Heating Value Determination of Liquid Product from Corn Stalk Pyrolysis,” Journal of Shandong University of Technology, Vol. 20, No. 2, 2006, pp. 11-13.

[22] Y. K. Matsui, Y. S. Fukuda and T. K. Inoue, “Effect of



107

Natural Organic Matter on Powdered Activated Carbon Adsorption of Trace Contaminants: Characteristics and Mechanism of Competitive Adsorption,” Water Research, Vol. 37, No. 18, 2003, pp. 4413-4424.

[23] C. Pelekani and V. L. Snoeyink, “Competitive Adsorption between Atrazine and Methylene Blue on Activated Car-bon: The Importance of Pore Size Distribution,” Carbon, Vol. 38, No. 10, 2000, pp. 1423-1436.

smart grid and renewable energy · the smart grid and renewable energy (online at scientific...

Documents