IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO. 1, JUNE 2010
89
Demand Respon Scheduling by Stochastic SCUC
Masood Parvania, Student Member, IEEE, and Mahmud Fotuhi-Firuzabad, Senior Member, IEEE
Abstract—Considerable developments in the real-time telemetry of demand-side systems allow independent system operators (ISOs) to u rerves provided by demand respon (DR) in ancillary rvice markets. Currently, many ISOs have designed programs to utilize the rerve provided by DR in electricity markets. This paper prents a stochastic model to schedule rerves provided by DR in the wholesale electricity markets. Demand-side rerve is supplied by demand respon providers (DRPs), which have the responsibility of aggregating and managing customer respons. A mixed-integer reprentation of rerve provided by DRPs and its associated cost function are ud in the propod stochastic model. The propod stochastic model is formulated as a two-stage stochastic mixed-integer programming (SMIP) problem. The first-stage involves network-constrained unit commitment in the ba ca and the cond-stage investigates curity assurance in system scenarios. The propod model would schedule rerves provided by DRPs an
d determine commitment states of generating units and their scheduled energy and spinning rerves in the scheduling horizon. The propod approach is applied to two test systems to illustrate the benefits of implementing demand-side rerve in electricity markets. Index Terms—Demand respon, mixed-integer programming, curity cost, stochastic curity-constrained unit commitment, uncertainty.
NQ SUC MC
Number of discrete points in offer package of DRP . Startup cost of unit at time . Minimum production cost of unit . Commitment state of unit at time . Real power generation of unit at time . Real power generation of unit in gment at time . Lower limit of real generation of unit . Upper limit of real generation of unit . Startup cost of unit . Scheduled up-spinning rerve of unit at time . Scheduled down-spinning rerve of unit at time . Ramp-up limit of unit (MW/min). Ramp-down limit of unit (MW/min). Minimum up time of unit . Minimum down time of unit . On time of unit at time . Off time of unit at time . Real power flow of line at time . Maximum capacity of line . Reactance of line . Voltage angle of nding-end bus of line . Voltage angle of receiving-end bus of line . Load demand of bus at time .
SR SR RU
I. NOMENCLATURE Index of generating units. Index of transmission line. Index of time. Index of bus. Index of DRPs. NG NT NB ND NS NG Number of generating units. Number of scheduling hours. Number of bus. Number of DRPs. Number of scenarios. Number of generating units connected to bus . Number of transmission lines connected to bus . NN Number of gments of piecewi linear cost function of generating unit .
RD
wash是什么意思DRR
Scheduled rerve of DRP
at time .
Binary variable associated with point of DRP at time ; 1 if the point is scheduled and 0 otherwi. cc ec Capacity cost of point Energy cost of point of DRP of DRP at time . at time .
Manuscript received March 03, 2010; revid March 14, 2010. Date of current version May 21, 2010. Paper no. TSG-00037-2010. The authors are with the Center of Excellence in Power System Control and Management, Electrical Engineering Department, Sharif University of Technology, Tehran, Iran (
e-mail: parvania@ee.sharif.edu; fotuhi@sharif.edu). Digital Object Identifier 10.1109/TSG.2010.2046430
CCDRP ECDRP
Capacity cost of rerve provided by DRP at time . Energy cost of rerve provided by DRP time . at
1949-3053/$26.00 © 2010 IEEE形容高大的词语
90
IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO. 1, JUNE 2010
Slope of gment of the piecewi linear cost function of unit at time . Offered capacity cost of unit for providing up-spinning rerve at time . Offered capacity cost of unit for providing down-spinning rerve at time . Offered energy cost of unit for providing up-spinning rerve at time . Offered energy cost of unit for providing down-spinning rerve at time . sr sr Deployed up-spinning rerve of unit at time in scenario . Deployed down-spinning rerve of unit at time in scenario . Real power flow of line at time in scenario . Voltage angle of nding-end bus of line in scenario . Voltage angle of receiving-end bus of line in scenario . drr Deployed rerve of DRP scenario . at time in
Binary variable associated with point of DRP at time ; 1 if the point is deployed in scenario and 0 otherwi. LC VOLL Involuntary load curtailment in bus at time in scenario . Value of lost load in bus at time . Probability of scenario . System lead time (h). Spinning rerve market lead time (min).
II. INTRODUCTION
D
EMAND RESPONSE (DR) is a tariff or program established to motivate changes in electric consumption by end-u customers in respon to changes in the price of electricity over time. DR offers incentives designed to induce lower electricity u at times of high market prices or when grid reliability is jeopardized [1]. Dramatic increas in demand for electric power have made the u of DR more attractive to both customers and system operators. As the above definition implicitly emphasizes, DR programs can be divided into two major programs: time-bad DR programs, and incentive-bad DR programs. Both type of DRs are currently under operation in many ISOs around the world. The time-bad DR programs are established to overcome flat or averaged electricity pricing flaws. Many types of the programs are designed in different independent system operators (ISOs), from which time-of-u tariffs, critical-peak pricing, and real-炒股经验
time pricing are the three well-known programs. The incentive-bad DR programs offer payments for customers to reduce their electricity usage during periods of system need or stress. The incentive-bad DR programs substantially have marketbad structures, and can be offered in both retail and wholesale markets. Different types of incentive-bad programs span over long-term to mid-term, short-term, and even real-time offered programs, each of which has its own goal of operation. In order to better implementation of DR programs, new market participants designated as demand respon providers (DRP) are introduced in wholesale electricity markets. A DRP participates in electricity markets as a medium between ISO and retail customers, and has the responsibility of aggregating and managing customer respons to ISO offered programs. The ISO-sponsored DR programs have requirements such as minimum curtailment level. Many of retail customers do not satisfy the requirements. The DRP enrolls customers to participate in different DR programs, and offers the aggregated respons in the ISO’s program. In this way, all customers, even small ones, have an opportunity to participate in DR programs. In the FERC order 719, it is emphasized that ISOs can permit DRPs to bid DR on behalf of retail customers directly into the ISO’s organized markets [2]. The DRP is also responsible to provide customers with telemetry systems needed for monitoring and control of their electricity consumption. It should also be noted that customers who satisfy the requirements can participate solely in DR programs. The FERC ord
er 719 requires ISOs to accept bids from DR resources in their markets for ancillary rvices, on a basis comparable to other resources [2]. Considerable developments in demand-side real-time telemetry systems allow ISOs to u demand-side provided rerves in ancillary rvice markets. To this end, many ISOs developed certain programs designated as ancillary rvice demand respon (ASDR) programs. The NYISO has developed the ICAP/SCR program and utilize it during rerve shortage events [3]. The PJM interconnection implemented the day-ahead scheduling rerve market (DASR) and is intended to provide incentives for demand resources to provide day-ahead scheduling rerves [4]. The ERCOT designed the load acting as a resource (LaaR) program, which allows customers who meet certain performance requirements to provide operating rerve [5]. The ISO New England started the real-time DR program in 2005, which requires customers to commit mandatory energy reductions on a predefined notice from the ISO [6]. Considerable efforts have been devoted to solve the curityconstrained unit commitment (SCUC) problem in the four past decades [7]–[12]. The state-of-the-art method for the solution of the SCUC problem is prented using the Benders decomposition [13]. The method decompos the SCUC problem into the UC master problem and two subproblems for checking network constraint at the ba ca and contingencies. The method of [13] has further been developed in [14] to consider system ac load flow constraints in the SCUC problem. The SCUC problem can be considered as a large-scale math
ematical programming problem which is subjected to system components unavailability and load forecast errors. Stochastic programming (SP) is introduced in [15] to deal
PARVANIA AND FOTUHI-FIRUZABAD: DEMAND RESPONSE SCHEDULING
91
虚汗多with uncertainties in mathematical programming problems. Reference [16] might be the first that formulated the unit commitment problem as a stochastic programming model without considering network curity constraints. In [17], the market-clearing problem with curity is formulated as a stochastic programming problem with uncertainty affecting only the objective function. The long-term stochastic SCUC model is developed in [18], which simulates the impact of uncertainty and allocation of fuel resources and emission allowance when solving the long term SCUC problem. This paper prents a short-term stochastic SCUC model that simultaneously schedules generating units’ energy and spinning rerve and also rerve provided by demand respon resources. The propod stochastic SCUC model is formulated as a two-stage stochastic mixed-integer programming (SMIP) model. The first-stage involves network-constrained unit commitment in the ba ca and the cond-stage checks curity assurance in system scenarios. The cond-stage
recour function in the propod two-stage model is cost of providing curity in system scenarios. This is the cost of deploying available resources to return the system to the load-supply balance state. The Monte Carlo simulation method is ud to simulate random outages of generating units and transmission lines. The scenario reduction method is also adapted to reduce the number of scenarios and the computational burden of the model. In the propod stochastic model, ISO runs the ASDR program to provide operating rerve from DRPs at load bus. Naturally, the rerve provided by DRPs is different from that of generating units. It should therefore be appropriately modeled to reflect its actual condition. A model for rerve provided by DRPs and its associated cost function is prented in this paper, and its mixed-integer reprentation is developed to be ud in the propod stochastic SCUC model. The rest of this paper is organized as follows. In Section III, the propod DR program and market structure are introduced. The propod stochastic SCUC problem is defined and elaborated in this ction. Section IV prents the propod mixed-integer reprentation of DRP rerves and the associated cost functions. The formulation of propod two-stage stochastic SCUC is prented in Section V. Section VI prents the solution method of stochastic programming. In Section VII, ca studies are prented and discusd. Conclusions are given in Section VIII.
Fig. 1. Correspondence between ISO and main market participants.
北京周末
Fig. 2. Sequence of decisions in the SCUC problem.
B. Day-Ahead Market Structure Fig. 1 shows that ISO receives bid-quantity offers from generating companies (GENCOs) to provide energy, up- and down-spinning rerve rvices, as well as DRPs’ offer to provide rerves. ISO will also receive hourly load demands from DISCOs. It clears energy and spinning rerve markets and schedules DRP rerves simultaneously by applying SCUC. The SCUC objective is to determine a unit commitment schedule at minimum production cost without compromising the system curity constraints [13], [19], i.e., the solution will satisfy network flow and load bus constraints in the ba ca and contingencies. A contingency is a function of random outages of generating units and transmission lines. The random outages of generating units and transmission lines and also hourly load forecast uncertainty are modeled in the propod approach. A two-stage SMIP model [15] is propod in Fig. 2 for short-term SCUC. The SMIP decisions are divided into the first and cond-stage decisions. The first-stage decisions are tho which have to be made before the realization of system scenarios. The decisions consist of commitment states of generating units and their scheduled energy and spinning rerve in each scheduling hour. The decision on the scheduled DRP rerves is also made in the first-stage. The system curity constraints are checked after the realization of system scenarios and in the cond-stage decisions.
The decisions are associated with the deployment of spinning and DRP rerves, and the amount of involuntary load shedding in each scenario. The social cost of SCUC consists of the ba ca cost and the expected cost of providing curity. The propod SMIP model considers the following goals: • commit generating units and clear the energy market; • schedule spinning rerve of each generating unit (simultaneous clearing of spinning rerve market); • schedule DRP rerve; • consider random outages of generating units and transmission lines;
III. PROBLEM DEFINITION
A. Demand Respon Program The focus of this paper is to schedule operating rerves provided by DR. It is assumed that ISO runs the ASDR program for providing operating rerves. DRPs submit offers to participate in this program. The rerve provided by DRPs are analogous to up-spinning rerve provided by generating units. The enrolled customers would reduce their demand in the predefined lead time to provide the rvice. In this paper, it is assumed that customers will not provide down-spinning rerve rvices.
92
IEEE TRANSACTIONS ON SMART GRID, VOL. 1, NO. 1, JUNE 2010
the third line is cost of scheduling DRP rerves, and the fourth line is the expected cost of providing curity in scenarios
SUC SR CDRP
Fig. 3. DRP’s bid-quantity offer package.
MC SR
SC • deviations of power produced in scenarios as compared to the ba ca is measured and monetized by rerve variables; • consider involuntary load curtailments as possible corrective actions.
(4)
SC is the cond-stage recour function of the two-stage stochastic model. It is the curity cost associated with scenario as expresd below SC sr ECDRP sr
猪图片大全可爱IV. DEMAND RESPONSE MODEL DRPs will aggregate discrete retail customer respons and submit a bid-quantity offer to the ISO, as shown in Fig. 3. with The discrete DRP rerve quantities a
re labeled as should be greater than the the associated cost of . Here, minimum curtailment level of the ASDR program specified by ISO. A mixed-integer reprentation of the DRP bid-quantity is shown in (1)–(3)
VOLL LC
(5)
DRR
(1)
CDRP
(2) (3)
Here, it is assumed that the demand decreas as prices inis constrained to increa monotonically [8]. A crea and DRP submits two t of offers to the ASDR program; the capacity cost and the energy cost of rerve. It should be noted that the energy cost of rerve is paid only if the rerve is deployed by the ISO in actual operation.
where the first line of (5) reprents cost of deploying up- and down-spinning rerve in scenario , the cond line is cost of deploying the DRP rerve in scenario , and the third line is cost of involuntary load curtailment in scenario . In other words, the cost of system curity is the cost of deploying resources for providing curity in system scenarios. In this paper, spinning rerve, DRP rerve, and involuntary load curtailment are considered as resources which can be ud to maintain system curity in ca of system component outages. There are two ts of variables in (4) and (5) for rerve rvices provided by DRPs and generating units. The first t is associated with the capacity cost offered by GENCOs and DRPs, while the other t is associated with the energy cost offered by GENCOs and DRPs. The two ts of variables are subjected to the first-stage and cond-stage constraints which are prented below. B. First-Stage Constraints
V. PROBLEM FORMULATION The formulation of the problem includes the objective function, and the first-stage and cond-stage constraints. A. Objective Function
The first-stage constraints are associated with the ba ca, including the following: DC power flow equation in steady state (6) (7)
夜莺与玫瑰的寓意The objective function is formulated as a standard two-stage SP problem [15]. The total cost is given
in (4), in which the first line is cost of energy production including startup cost; the cond line is cost of scheduling up- and down-spinning rerve;
Transmission flow limits in the ba ca (8)
PARVANIA AND FOTUHI-FIRUZABAD: DEMAND RESPONSE SCHEDULING
93
Generating units startup cost constraint SUC Real power generation constraints (10) (11) (12) (13) (9)
drr
LC
(20) (21) Transmission flow limit in scenarios (22) Deployed up- and down-spinning rerve limit (14) (15) sr sr SR SR (23) (24)
SR SR Up- and down-spinning rerve limits SR SR RU RD
Minimum up and down time constraints
Deployed DRP rerve constraints drr drr (16) DRR (25) (26)
Ramping up and down constraints RU RD (17) DRP rerve constraints DRR (18)
ECDRP
ec
ec
(27)
Involuntary load curtailment limit LC LC (28)
CCDRP
cc
cc
(19)
As stated in (12) and (13), it is assumed that generating units offer maximum amount of their achievable capacity as spinning rerve. The only constraint on spinning rerve provided by generating units is their ramping capability, which is stated in (14) and (15). This will result in optimum determination of energy and spinning rerve provided by generating units according to energy and rerve requirements of the system. C. Second-Stage Constraints The cond-stage constraints which are considered in system scenarios are as follows: DC power flow equation in scenarios
Here, in order to consider random outage of generating units and transmission lines, is divided into respectively. Considering a two-state Markov model [20] for each component, the elements of the two vectors are binary random variables in which 1 reprents the healthy state of a component and 0 otherwi. In the propod two-stage stochastic model, decision on generating units’ commitment states is only made in the first-stage. Besides, the real power generation of the committed units at the ba ca should satisfy the DC power flow constraint expresd does not change in in (6). The power generation variable any scenario . Instead, sr , sr , drr , and LC , are determined such that the DC power flow (20) is satisfied in each scenario . The most economic portfolio of the above alternati
ves is lected by the model to alleviate the adver impacts of random outages of generating units and transmission lines. The relationship between the first and the cond-stage rerve variables is specified in (23)–(25). In addition, (23) and (24) indicate that only healthy generating units in scenario would provide spinning rerves. VI. SOLUTION METHOD The first step in solving a SP problem is to model the uncertainties associated with the system [15]. The basic two-state Markov model shown in Fig. 4 is ud to reprent generating unit and transmission line status [20].
桌面时间怎么设置
sr
sr
