A Novel Coordination Scheme of Transaction Scheduling in Multi-level Power Market Rui Bo, Yang Li, Fubin Liu, and Guoqing Tang

Abstract-With power system deregulation undergoing worldwide, it’s urgent for China to establish multi-level power market in large interconnected power system to better configure energy resources and gain greater economic benefit of power exchanges. But it’s not clear how to maintain network security in such environment. This paper therefore suggests a coordination information based security coordination scheme under which the upper-level and lower-level power market can cooperate smoothly for the system security and potential power exchanges among lower-level markets can be greatly boosted. The method of obtaining the coordination information for lower-level market and the transaction scheduling model for upper-level market are also discussed. The suggested scheme is tested on IEEE 30 bus system and proved to be feasible. Numeric results show that the scheme and related algorithms proposed in this paper can make great economic benefit which is near to the ideal results. Index Terms-Broker System, coordination information, multi-level power market, power exchange, pre-scheduling, security coordination, transaction scheduling.

I. INTRODUCTION

S the push for deregulation of power industry continues, the economics and efficiency of energy resource configuration are under great concerns. In China, up to now, more than six provincial power grids are on the way of restructuring. However, it’s just the first step. The energy development level and resource structure of each province may be quite different from one another. For example, the generation costs in some provinces are relatively much higher than those in the neighboring provinces due to the scarcity of energy resources, higher fuel price and labor price. Considering of these facts and others like the weak structure of power grid and its vast territory, China therefore has constituted policies that emphasize on the development of regional power market on large interconnected power grid which is usually combined by several provincial grids. The regional power market is designed to facilitate and boost

A

This work was supported in part by the Nari-Southeast University Excellent Thesis Fund. Rui Bo is with the Department of Electrical Engineering, Southeast University, Nanjing, 210096 China(e-mai1: [email protected]). Yang Li is with the Department of Electrical Engineering, Southeast University, Nanjing, 210096 China(e-mail: lijang@seu,edu.cn). Fubm Liu is with the Depamnent of Electrical Engineering, Southeast University, Nanjing, 210096 China(e-mail: [email protected] ). Guoqing Tang is with the Department of Electrical Engineering, Southeast University, Nanjing, 210096 China(e-mail: [email protected]).

0-7803-8110-6/03/$17.00 02003 IEEE

I

potential power exchanges among provinces and hence make better resource configuration and economic benefit. The first attempt in China is East China Regional Electricity Market (ECREM) which consists of five provincial grids in geography. This is a two-level power market and organized in a decentralized dispatch mode. The lower-level market refers to any of the five provincial power market. In general, the provincial market is organized in POOL mode and its market members include generators and customers within the provincial grid. The upper-level market is called regional market which retains control of the tie-lines connecting provincial grids. The five provincial markets are market members of the regional market. Every provincial market will be able to exchange power with each other on an economic basis through regional market. The IS0 of each provincial power market, which is usually undertaken .by the provincial power company in China, acts as two roles: on one hand, it is responsible for operation of the provincial market as an ISO; on the other hand, it is a market member of the regional market and will participate in the regional market as the deputy to the provincial power market. In ECREM, East China Power Company acts as the IS0 of the regional market. In the next step of planning for ECREM, generators and bulk customers can directly take part in the regional market although currently the provincial market is their only choice. ECREM is established on a power system that covers a vast temtory since it is the combination of five provincial grids. So it is reasonable for China to select multi-level market rather than a unified single-level market as the organization mode although the latter is more advantageous in economics in theory. This decision can be viewed as the compromise between economics and technical complications. In the multi-level market, energy resources will be deployed in an order of priority: the generation resources, for example, will be dispatched fiistly in the provincial market to satisfy the energy demand in the same market; Afterward, the extra generation resources can be dispatched in the regional market to meet the demand in other provinces. Broker trading mode will be adopted as market model of the regional market according to the existing experience, expertise and advise of consulting companies. The regional spot market of ECREM will be arranged through a broker system. Besides, market participants can exchange energy with one another through bilateral contracts. Both the two types of transactions in ECREM are inter-province

85

transactions. The features of ECREM can be summed up as follows: (1) It’s a multi-level market; (2) The lower-level market, namely, provincial market, acts independently and has its own dispatching control; (3) The upper-level market-regional market, should be contrived to encourage the members to fully exchange power so as to gain economic benefit as much as possible. These features are the foothold of the following discussion in this paper. Multi-level power market is a new research field and hasn’t attracted much attention up to now since single-level power market like POOL is the main research focus in recent years. Reference [l] discusses the economic power exchanges in power brokerage system. Network flow algorithm and dynamic programming are presented to achieve greater savings. This paper points out that the High-Low matching algorithm does not necessarily maximize savings subject to the network constraints. The optimal energy exchanges within a large system of interconnected electric utilities is discussed in [2] and wheeling penalties are introduced to consider the influence of wheeling power. Reference [3] and [4] presents a brokerage system that uses Linear Programming (LP) to maximize the savings or profits to each interconnected utility. Similarly, reference [6] represents a LP formulation of an energy brokerage system with emission trading. Furthermore, a new method of pricing mechanism for reserve margins and transmission losses is proposed in [7]. The theoretical bases for benefit optimization in centralized and decentralized electric power systems are carefully examined in [ 5 ] . Although these researches focus on the approaches for optimal power exchange in decentralized power system, there still lies two important issues: (1) In the interconnected power system (or multi-utility environment in the US), each subregion (or utility) is always modeled as an equivalent generator simplifying the detailed internal grid in order to reduce the computation complication. However, it may lead to conspicuous computing errors and even unfeasible solution; (2) In multi-level power market, the upper and lower-market are tightly related rather than independent of each other. So when technical problems such as transaction scheduling in upper-level market are examined, their consequential influence on lower-level market must be investigated simultaneously. Considering the above issues, this paper proposes a novel coordination scheme of transaction scheduling for multi-level market through which great economic savings can be obtained and the security of the whole grid is maintained. The remainder of this paper is organized as follows: firstly, a new problem encountered in multi-level power market will be introduced in the following section. Afterward, in section 111, this paper will propose a novel coordination information based security coordination scheme of transaction scheduling to solve the problem. The next section will discuss the corresponding method for obtaining bidding data and coordination information for the lower-level market under the proposed scheme. A mathematic model of transaction

scheduling for regional market will be proposed in section V. Then, a case study will be illustrated in the next section. Finally, conclusion will be drawn in section VII. 11. NEWPROBLEM-SECURITY COORDINATION PROBLEM In multi-level power market described above, provincial power market is a special type of market member for regional market: (1) It is not destined to be a seller or buyer. It depends on the practical circumstances. For instance, a provincial market will buy power through regional market when it is short of power energy or the additional generation cost is much higher than other provincial markets; in the opposite case, the provincial market will be a seller when it has extra cheaper generation capacity; (2) This type of member is a market itself and retains control of its ovtrl.1 grid. So, unlike other members as generators and bulk customers, it is based on network rather than bus or buses. Although each provincial grid is only controlled by corresponding provincial market-precisely, its ISO, power flow on the provincial grid can be envisaged as two components: one is caused by intra-province transactions; the other by inter-province transactions. These two types of transactions are scheduled in provincial market and regional market individually. So, the maintenance of network security needs the cooperation between upper and lower markets. The markets should coordinate to ensure the network security and thus boost market transactions at the same time.

III. S O L U T I O N ~ O O R D I N AINFORMATION TION BASED SECURITY COORDINATION SCHEME

A novel security coordination scheme for transaction scheduling is proposed here and the processes in provincial market and regional market are shown in Figure 1 and Figure 2 respectively. The flowchart in Figure 2 represents the detailed process of the process block labeled with an asterisk (‘*’) in Figure 1. It should be pointed out that the open time of provincial market is designed^ to be earlier than that of regional market while the clear time of provincial market is later.

Transaction scheduling in regional market(*) Get results of scheduled transactions from regional market

,

c

I Transaction scheduling for its own market I

e3 Market clear

86

-

Fig. 1. Transaction scheduling process in provincial market under the proposed security coordination scheme

(

M

*

X

1

2

d

Accept bids and coordination information submitted by market members

.c

Transaction scheduling (using coordination information)

+

Send results to corresponding market members

( Market clear ) Fig. 2. Transaction scheduling process in regional market under the proposed security coordination scheme

provincial grids

I

Gird of province A

Fig. 3. Some interconnected power system

Firstly, after provincial market performs pre-scheduling with zero inter-province transactions, it's found that line 1 4 2 still have 40 percent available capacity. Also, it's obvious that bus 1 will undertake all the possible inter-province transactions since there is no generator at bus 2. Based on these facts, provincial market A will submit the following data to regional market: LCPP of line 1+2 is LCPP12=40%; BPPF at bus 1 is al=l,BPPF at bus 2 is' a2=0. With these coordination information instead of the detailed prescheduling result of each provincial market, the regional market can deduce as follows the maximum power sale amount that the network of provincial market A can undertake: firstly, aI=l and az=O represent that the inter-province transactions will be totally provided at bus I and the power will flow through line 1-2 as can be seen from Figure 3; then, LCPPIz=40% indicates that line 1 4 2 have 40 percent capacity available. So, it can be concluded that the maximum sale amount of provincial market A should be no more than

Under this scheme, each provincial market submits two type of coordination information. With these information, the regional market can evaluate the network security of provincial grids when performing transaction scheduling. The first type of coordination information is called Line Capacity Participation Percent (LCPP) which is defined as the percent of provincial transmission lines capacity available for carrying inter-province transactions. The other type of coordination information is called Bus Power Participation Factor (BPPF) which represents the proportion of contributed looMW* 40% = 40MW ;otherwise, transmission constraints power of each bus in undertaking the inter-province 1 transactions. BPPF can further be classified into two kinds: will be violated. This conclusion is identical to the truth. one kind related with sold power is called Generation From the above example, we can see that the proposed Contribution Factor (GCF); the other kind of BPPF which is security coordination scheme can achieve accurate results and related with purchased power is called Load Extraction Factor be used for transaction scheduling as long as the submitted (LEF). Elaborate description and definition can be referenced coordination information is accurate enough. The method for in [8]. It should be noted here that it's proper to define both obtaining the coordination information for provincial market GCF and LEF at generation buses because whether the and the way to use these information in transaction scheduling provincial market sells or buys power, the influenced buses of regional market will be discussed in the following two are always generation buses. The calculation method of LCPP sections. and BPPF will be discussed in section IV. LCPP reflects the remained available capacity of provincial DATAAND I v . METHOD FOR OBTAINING BIDDING transmission lines after the pre-scheduling of the provincial COORDINATION INFORMATION market with zero inter-province transactions. BPPF reflects how the pre-scheduling results of the provincial market will A. Bidding data change with different level of inter-province transactions. So, For a provincial market, the sell bid and buy bid can be with these two types of coordination information, the regional quantified respectively based on additional and avoidable market is able to know how power flow in a provincial grid costs. In conventional methods, for example, the bidding will change with inter-province transactions and furthermore curve can be directly drawn from the system marginal cost estimate the network security of the provincial grid. It should curve. be pointed out that this paper mainly discusses transmission B. LCPP constraints of network security. According to the definition of LCPP, it can be computed Here give a simple example to illustrate the principle of the proposed scheme. Suppose the two-bus system in Figure 3 by subtracting the occupied percent of the transmission line represents a provincial grid. The network and load parameters capacity that is determine by pre-scheduling of provincial market with zero inter-province transactions. are labeled in the figure. For instance, if a% of the capacity of line i+j in downstream direction is occupied after the pre-scheduling, as 87

1

-

P,""

ps-,

j

The above method for calculating LCPP is based on the assumption that DC power flow is used for computation. So Superposition principle is applicable for line power flow, that is, the ultimate power flow can be seen as the algebraic addition of two power flow components: one is caused by intra-province transactions; the other is caused by interprovince transactions.

vector of participant's maximum and minimum sale amount respectively; vector of participant's maximum and minimum Pb-, pbmin purchase amount respectively; Pgm, Pgmin vector of maximum and minimum bus generation amount respectively; Pd-,Pdmin vector of maximum and minimum bus load amount respectively; -max Pline

available transmission line capacity vector in downstream direction;

P-

available transmission line capacity vector in upstream direction.

-line

B. Mathematic model The mathematic model is just briefly reviewed here since the detailed derivation of the model has been introduced in 181.

C. BPPF According to the definition of BPPF, it can be calculated as follows, @Gi BPPF. = 1 A P where BPPFi denotes the BPPF at bus i; AP

Gi

Ps- I T - u l P S m

p6"" I T T * ~ I P b m a X represents the

power change amount at bus i when the provincial market exchanges additional At' power with other provinces through the regional market. Equation (1) means that BPPF can be calculated by twice pre-scheduling computation: one with the additional AP power and the other without AP . More detailed description of the method for obtaining bidding data and coordination information can be referenced in [9].

v. TRANSACTIONSCHEDULING METHOD FOR REGIONAL POWER MARKET In order to get optimal social benefit, a new method is proposed for transaction scheduling in broker system that takes the network constraints into account directly. A mathematical programming model is established to implement the 'method. A. Nomenclature We will use the following notations. m total number of market participant in regional market; psi sell bid of participant i; pbj buy bid of participant j; fij transaction amount sold'by participant i to participant j;

t<

existing bilateral transaction amount between participant i and participant j ; T transactions matrix representing transactions among m participants; U vector of ones of dimension m; a Generation Contribution Factor matrix; B Load Extraction Factor matrix; PTDF Power Transfer Distribution Factor matrix;

Pg- l a T T - u l P g m a X s.t.

Pd- I pTTTuI Pdm

Equation (2) is the mathematic model of transaction scheduling for regional power market. The objective is to maximize the social welfare realized by the transactions. The first and second inequalities respectively represent the restrictions of power sale amount and purchase amount. The third and fourth inequalities respectively represent the restrictions of bus generation amount and load amount. The fifth inequality is the line power flow restrictions. The sixth inequality represents the transaction amount restrictions. Transaction amount should not be less than the amount of existing bilateral transactions. It should be noted that the implication of

-max Pline

and

varies in different cases: for the tie-lines connecting provincial grids, they respectively equal the maximum transmission line capacity in downstream and upstream direction; for the intraprovince lines, they respectively equal the product of maximum transmission line capacity in downstream and upstream direction multiplying corresponding LCPP. Network transmission constraints are incorporated into the optimization model directly. Extension of the model such as power loss, wheeling power, multi-segment bidding and ramp rate constraints is carefully discussed in [9].

VI. CASESTUDY The algorithm of 'proposed security coordination scheme is implemented by MATLAl3 and tested on IEEE 30 bus system. 88

’

The system data is obtained from Cornel1 University’s MATPOWER package which can are available in [ 101. The network is shown in figure 5, the whole grid is grouped to 3 control areas, each with 2 generators. Each control area here can be looked as a provincial power market and the whole grid as a regional market. So, the transactions between control areas can be viewed as inter-province transactions.

4.8

4.6 4.4

= F 5;

h_.

F-

4.2 4.0

e,

E

3.8

M

3.6 3A

L

k

,

,

l

L

I 4

0

12

8

-

16 20 Power Amount (MW)

24

28

32

36

Fig. 6. Supply-demand curve TABLE I BIDDING DATAOF EACHAREA

Member Fig. 5. IEEE 30 bus system

Area 1 To make the result easier to understand, it’s assumed that no bilateral contracts have been made initially and each provincial market will calculate bidding data on cost based method like’ that introduced in section IV, rather than some tactical or game theory method. The maximum transmission capacity of line 2-4 is revised to 17 MVA to make a transmission bottleneck and thus will reduce the power exchange capacity of area 1. The case study is designed to include the following 4 steps: Firstly, each provincial market calculates its bidding data and coordination information, i.e., LCPP and BPPF, using the method introduced in section IV. The results are shown in Table I and II. The supply-demand curve is depicted in Figure 6. (Note: The digits labeled near each segment of the curve indicate the number of control area by whom the bid is submitted.) Then, with these data, transaction scheduling in regional market is performed, and, Each provincial market performs its own transaction scheduling considering the inter-province transactions scheduled in the previous step. Finally, the total cost will be calculated and compared with two other cases. Results are shown in Table III.

Area 2

I I

I

4 4

3.8300 4.5556

NIA NIA NIA NIA

NIA NIA NIA NIA

1

4 4 4 4 4

I

I

4 4 4 4 4 4 4 4

1.

3.8077 3.8921 3.9910 4.0927 4.1972

I

I

3.3784 4.4404 4.3285 4.2173 4.1066 3.9965 3.7017 3.6091

I

N/A NIA NIA

1

I

NIA NIA NIA

I I

TABLElI GENERATION CONTRIBUTION FACTORAND LOAD EXTRACTION F A ~ OFR . EACHAREA

I

BusNumber Area 1 Area 2

1

Area3

I

I

GCF

1 2 13 23 22 27

LEF

0.46 0.54 0.64 0.36 0.15 0.85

0.46 0.54 0.53 0.47 0.15 0.85

TABLE III SCHEDULING RESULTS AND TOTALCOST

l”SACrr0N

Net export amount of each area (MW)

I 89

CaseA CaseB CaseC

I 1 1

5.8 6.3 0

1

-16.1 -17.9 0

I

10.3 11.6 0

I

I

1

576.0 574.7 583.4

I

Three cases are compared in Table III. The results of Case A represent the outcome computed under the coordination scheme and corresponding algorithms proposed in this paper. Case C represents the case with zero inter-area transactions. In Case B, the results are obtained by supposing the whole grid is controlled and dispatched under centralized mode. As shown in Table I to Table HI, the security coordination scheme proposed in this paper is a feasible solution. The transactions scheduled in regional market are proved to be feasible and favorable for the provincial markets, that is, every provincial market is able to implement the scheduled inter-province transactions and willing to get profit from the power exchanges. From the results of Case C in contrast with those of Case A and Case B, we can see that the total costs can be remarkably reduced with reasonable power exchanges. As we know, if all the energy resources are organized in a single-level market and dispatched in centralized mode, optimal economic benefit or minimum total cost will be achieved. So Case B is the optimal case. Since Case A is related to multi-level market and decentralized mode, Case A is inferior to Case B in theory from the perspective of economics. Even now, from the results shown in Table 111, we can see that the scheduling results and consequential total costs of Case A are quite near to those of Case B, which is a good illustration of the challenging performance in economics of the proposed scheme. Furthermore, each provincial market can dispatch independently under this scheme except for just one requirement to submit coordination information to regional market. It's consistent with the rationale of this paper: multilevel power market and decentralized organization mode.

great economic benefit that is near to the ideal optimal results obtained under centralized dispatch mode. It indicates that under the proposed scheme, power exchanges are fully boosted, which rightly meets the original intention of the establishment of multi-level power market. W I . REFERENCES K. W. Doty, P. L. McEntire, " An analysis of electric power brokerage system," IEEE Trans. Power Apparatus and Systems, vol. PAS-101, no. 2, pp. 389-396, Feb. 1982. Mark J. Huggins, Michael S. Mirsky, " Optimal energy transactions in interconnected electric systems," IEEE Trans. Power Apparatus and Systems, vol. PAS-104, no. 11, pp. 2994-3003, Nov. 1985. George Fah4 Daniel A. Richards, Gerald B. Sbeble, " The Implementation of an energy brokerage system using linear programming," IEEE Trans. Power Systems, vol. 7, no. 1, pp. 90-96, Feb. 1992. George Fahd, Gerald B. Sheble, " Optimal power flow emulation of interchange brokerage systems using linear programming," IEEE Trans. Power Systems, vol. 7, no. 2, pp. 497-504, May. 1992. Fuyuhiko Nishimura, Richard D. Tabors, Marija D. Ilic, Joaquin R. Lacalle-Melero, " Benefit optimization of centralized and decentralized power systems in a multi-utility environment," IEEE Trans. Power Systems, vol. 8, no. 3, pp. 1180-1186, Aug. 1993. D. Chattopadhyay, " An energy brokerage system with emission trading and allocation of cost savings," IEEE Trans. Power Systems, vol. 10, no. 4, pp. 1939-1945, NOV.1995. Jayant Kumar, Gerald Sheble, " Framework for energy brokerage system with reserve margin and transmission losses," IEEE Trans. Power Systems, vol. 11, no. 4, pp. 1763-1769, Nov. 1996. Bo Rui, Liu Fubin, Li Can, Wei Ping, Li Yang, Tang Guoqing, "Transaction scheduling considering transmission network constraints in regional electricity market," Automation of Electric Power Systems, vol. 26, no. 22, pp. 10-15, Dec. 2002. Bo Rui, " Research on transaction scheduling and congestion management in regional electricity market," M.S. dissertation, Dept. Elec. Eng., Southeast University, China, 2002. [lo] Ray Zimmerman, D'qiang Gan. (1997, Dec.). MATPOWER Package (version 2.0). Available: http://www.pserc.comelI.edu/matpwer/

IX. BIOGRAPHLES

W. CONCLUSION Multi-level power market is now becoming a cynosure recently in power industry, especially in countries of vast territory such as China. Members in interconnected power system can achieve great economic benefit by power exchange with neighbors through multi-level power market. This paper proposes a novel security coordination scheme of transaction scheduling which can coordinate the transactions scheduled in upper-level (or regional) market and lower-level (or provincial) market to maintain the security of the whole system. Some coordination information is defined and used in the scheme. With this information, the upper-level market can estimate the influence of power exchange among lower-level markets (or inter-province transactions) on the security of the grids of lower-level markets. The method for obtaining the coordination information as well as bidding data for lower-level market is also discussed. Furthermore, the transaction scheduling method for upperlevel market is introduced. Finally, a case study is conducted on IEEE 30 bus system. The results show that the proposed scheme and related algorithms are feasible and can achieve

Rui Bo (St. M. IEEE) was bom in Nanjing, China, on November 11, 1978. He received B.S. degree from Southeast University in July 2000. He is now pursuing for M.S. degree in Southeast University and has finished his M.S. dissertation. His current interesting area is power system operation and planning and power system restructuring.

Yang Li (M. IEEE) was born on August 7,1961. He received his B.E. and M.E. respectively in July 1982 and April 1992 from Southeast University, China. From June 1998 to June 1999, he was a researcher in Aichi Institute of Technology in Japan. His current interesting is electricity market and Demandside Management. Fubm Liu (St. M. IEEE) was bom in Yangzhou, China, on June 22, 1975. He graduated from Southeast University in July 1997 and April 2000, and respectively received B.E. and M.E. His current interesting area is power system operation, electricity market, and operations research applications in power system

Guoqing Tang (M. IEEE)was born in Shanghai, China, on Oct.IO 1937. Now he is a professor worked in department of electrical engineering of Southeast University. His current interesting field is power system operation and control, artificial intelligence applied to power system and electricity market.

90