Sci.Int(Lahore),26(5),1931-1940,2014
ISSN 1013-5316; CODEN: SINTE 8
1931
CO-OPTIMIZATION OF ENERGY AND RESERVE CONSIDERING DEMAND RESPONSE PROGRAM Atena Darvishi1, Hossein Akhavan Hejazi1, Behrooz Vahidi1*, Seyed Hossein Hosseinian1, Mehrdad Abedi1 1-Department of Electrical Engineering, Amirkabir University of Technology, Tehran 1591634311, Iran * Corresponding author email:
[email protected]
ABSTRACT: Co-allocation of energy and reserve is an efficient approach in market clearance. Many different factors contribute to this problem while accounting for system security and credible contingencies may increase the allocated spinning reserve and the total costs, load flexibility as a result of Demand Side Programs may contribute with a positive effect on the total amount and cost of spinning reserve as well as the dispatched energy. This paper presents the integration of Time of Use Program in the securityconstrained energy and reserve Co-allocation market and the effect that TOUP has on the total costs and amount of reserve services are studied in an actual unit commitment market clearing problem. Differential Evolution Algorithm is utilized for an efficient 24-hour unit commitment approach which accounts for both system security and load elasticity. The effectiveness of proposed method is evaluated by application of the algorithm on IEEE 24 bus reliability test system. The results of the algorithm are compared in various cases with other approaches for the unit commitment problem and it is shown that considering TOU program has a noticeable positive effect on the obtained optimum solution. Keywords: Energy and reserve market, Demand side management, Time of use program, Differential evolution NOMENCLATURE E Elasticity of the demand Q The demand value (MWh) Electricity energy price ($/MWh) 0 Initial electricity energy price ($/MWh) q0 Initial demand value (MWh) d(t) Customer demand in tth hour (MWh). (t) Spot electricity price in tth hour ($/MWh). 0(t) Nominal electricity price when the demand is nominal f Total cost of energy and reserve procurement. CP,i,t, CSR,i,t Energy and spinning reserve cost of ith generation unit. in period "t" Pi,t, , Qi,t Active and reactive generation of ith unit in period "t" Voltage of ith unit in period "t" V i ,t
Pi max , Pi min Upper and lower limits of active power
Ng, Npq Nc Gij ,Bij xn,k , vn,k , un,k
Pij ,t , Qij ,t
SRi,t
SR
max i
i ,t imin ,imax N, Nl
Upper and lower limits of reactive power generation in ith unit in period "t" Spinning reserve of ith generating unit in period "t" Maximum amount of reserve that ith unit is offering.
Sijmax
ij ,t
and
reactive
power
of
period "t" max PQi
V
,
min PQi
V
max min SRslack SRslack
,
max Pmin Pslack , slack
Pds,t, Pds,t(c)
Phase angle of ith generating unit in period "t" Upper and lower limits of Phase angle of ith generating unit in period "t" Numbers of system buses and transmission lines.
Active
transmission line between bus i and j in period "t" Maximum allowable apparent power of transmission line between bus i and j. Phase angle between bus i and j in
generation in ith unit in period "t"
Qimax , Q imin
Numbers of generating units and PQ buses. Number of contingencies. Conductance and susceptances of transmission line between bus i and j. Vectors representing nth member, mutant and trial vectors of kth generation.
PL,t , Pl(c)
1. INTRODUCTION November-December, 2014
Upper and lower voltage limits of ith PQ bus. Upper and lower slack bus limits for spinning reserve. Upper and lower slack bus limits for active power generation. Sum of dispatched power to all generators excluding slack unit, in steady state and cth contingency. Total amount of active load in steady state and cth contingency in period "t" .
1932
ISSN 1013-5316; CODEN: SINTE 8
ISO is responsible for provision of ancillary services, load balancing, congestion management, and security of the new power markets. Among all the ancillary services, reserve services including spinning and supplemental are among the services that are procured through the market. ISO has to provide sufficient power reserve as one of the ancillary services for maintaining security [1]. Reserve services are allocated based on two main market auction structures. In the first structure, the energy and reserve markets are unbounded and reserve market auctions are performed independently after clearing the power market [2-4]. In the second reserve and energy are integrated and the auctions are performed simultaneously [5-7]. Both energy and reserve are affected in this structure when maintaining the system security and thereby there is more flexibility in allocating the resources. One important factor in maintaining the system security is load flexibility that surely will have an effect on the energy and reserve market. Because of moving to the new smart grid which sends online prices to the customers, load may change according to the market prices or electricity tariff, or even participate in spinning reserve market. Many works have been done for procurement of energy and reserve. In these works, the required levels of reserve services for maintaining security of the system is set in three methods which are Deterministic, probabilistic, and stochastic methods. In deterministic method it is common to schedule enough reserve in order to compensate the loss of the largest generator or other prescribed contingencies [811]. [12] Suggests evaluating the “credibility” of failures and their “expected” consequences by means of probabilistic methods in power system security analysis methods. Chattopadhyay and Baldick [13] adopt the loss of load probability (LOLP) and an approximation of this quantity using an exponential function with system-dependent parameters in the UC optimization problem formulation. Loss of load probability is as an extra linear constraint in order to account for the associated risk. In another probabilistic method [14], it is stated that local reserves, being critical elements of the security-constrained marketclearing problem, must be treated as decision variables of the optimization process and not pre specified parameters. Also Bouffard and Galiana [15] propose a pool market clearing process, considering a probabilistic reserve determination. In some works in the area of power market clearance, stochastic formulation, accounting for the likelihood of the contingencies and allowing the load shedding are applied. The security constraint is applied by considering the expected Energy not served (EENS) and its costs due to random line and generator outages as well as load-side shedding [16-18]. For instance [19] and [20] by using EENS as the index of customers' reliability and by considering and by considering spinning reserve allocation, presented an approach for clearing power market. In all the mentioned works, in order to reduce the costs, maintaining security is obtained by allowing the load shedding during the contingencies.
Sci.Int.(Lahore),26(5),1931-1940 ,2014
In this work a security-constrained energy and reserve allocation is performed in such a way that accounts for all the generating units outages and allows for load elasticity; reducing the total costs is obtained by considering the demand response program as a replacement for load shedding. The demand is shifting from peak period to medium and low periods and thereby will be lowered in the peak times where the cost is high, but the load will not be curtailed during the contingencies. Furthermore, it has been shown that in this way the total cost will be lower than the case in which the load shedding is allowed and the load is considered to be inelastic. Differential Evolution Algorithm is used for a 24-hour unit commitment problem to determine the system optimal operating points that will be affected by different load levels as well as the previous operating points and unit ramping rates. Each hourly optimal power flow considers credible contingencies and system constraints including voltage and transmission limits in system’s steady state and after probable contingencies. The effect of applying Time of Use Program, as a subgroup of Time-based Demand Response programs, on daily demand curve and thereby the total cost reduction of energy and reserve have been studied. The comparison of the results of 24 hours of running the UC problem is done in two cases, In the First case, the results of security-constrained unit commitment with and without applying TOUP are compared. It is shown that applying DRP is effective on reducing the total cost of energy and reserve compared to the same scenario when the DRP is not applied. In the second case the result of security-constrained UC when considering the DRP is compared with the case in which the load considered inelastic and the load shedding is applied. The structure of the paper is as follows: Section 2 presents Integration of TOUP into UC problem. The case studies and the comparison of the results are given in Section 3 and Section 4 concludes the paper. 2. INTEGRATION OF DSM INTO SECURITY CONSTRAINED ENERGY AND RESERVE COOPTIMIZATION The purpose of this paper is to consider the effect of load participation in the form of demand response programs in total energy procurement costs in the simultaneous clearance of energy and reserve markets. Participation of consumers to gain maximum benefit may also cause a decrease of demand in peak times and hence lower the total generation costs. In order to account for the impact of demand response programs (TOUP) to market auctions, the settlements within an interval of 24 hours UC problem are considered. 2.1. The Optimization Objective The objective of the optimization formulation is to minimize the sum of energy costs and spinning reserve costs within an interval of 24 hours. The total cost of energy and reserve procurement within an interval of 24 hours can be shown as: t 24 Ng
f
[C t 1 i 1
November-December, 2014
P ,i , t
CSR ,i ,t ]
(1)
Sci.Int(Lahore),26(5),1931-1940,2014
ISSN 1013-5316; CODEN: SINTE 8
where CP,i,t and CSR,i,t denotes the cost of energy and reserve in every period of time respectively. Prices related to each generating unit are assigned here through units' price curves. The energy cost curve can be either based on market-based bids submitted by the resources or cost-based bids calculated from their operational costs. The combined cost of offers made by generating units should be minimized complying with all market and operation constraints. 2.2. The Supply Constraints The active and reactive power supplies are limited by available capacity of each generating unit in each period of time. It should not exceed its maximum operating limit, that is:
Pi min Pi ,t Pi max min i
Q
i 1,..., N g
(2)
t 1,..., 24 i 1,..., N g
Qi ,t Q
max i
t 1,..., 24
……
(3)
Each unit has a limited capability for spinning reserve which depends on its ramping rate and margin time. The spinning reserve associated with ith generating unit should meet:
0 SRi ,t min( SRimax ,t , MT * Ramp)
Pi ,t SRi ,t Pi max
i 1,..., N g
(4)
t 1,..., 24
i 1,..., N g
(5)
t 1,..., 24
2.3. Power Balance Constraints The power balance constraints are described by the AC power flow equations. In a practical implementation there may be multiple resources connected to each node, each resource injection in this work is modeled as a separate variable and the bus injection is considered to be the sum of all the resource injections. The generation and load balance for real and reactive power at a bus can be described as follows: N
Pg , i Pd , i V i
V
j
(G ij cos ij B ij sin ij )
j 1
i 1,..., N g t 1,..., 24 (6)
N
Qg ,i,t Qd ,i ,t Vi,t
V
j ,t
(Gij sin ij ,t Bij cosij ,t )
j 1
i 1,..., N g
(7)
t 1,..., 24
It is remarkable that the amount of Pd,i,t is changed in each hour through TOUP in such a way that the benefit of customer is maximized too, it is discussed in section 2.6. 2.4. Network Constraints Constraints in this paper include the following types; the voltage magnitude and phase angle limits and the branch flow limits; The voltages of NPV PV buses are kept as specified.
V i , t V i spec
i 1, ..., N P V t 1, ..., 24
(8)
The rest of bus voltages should set within their limits:
Vi min Vi ,t Vi max
1933
i 1, ..., N P Q
(9)
t 1, ..., 24
The voltage phase angle of all buses except the slack bus should be within the predefined value. i 1,..., N PQ PV imin i ,t imax (10) t 1,..., 24 The apparent power of each transmission line should not violate its transmission capacity:
Pij2,t Qij2,t Sijmax
(11)
Pij2,t [Gij (Vi,2t Vi,tVj,t cosij,t ) Bij,tVi,tVj,t sin cosij,t ]2
(12)
Qij2,t [Bij (Vi,tVj,t cosij,t Vi.2t ) GijVi,tVj,t sin cosij,t ]2
(13)
2.5. Contingency Limits The contingencies considered in this work are the outage of all generation units. Each contingency consists of the outage of a single generating unit. Errors in the load forecast and the outage of transmission lines are not taken into account. All system constraints mentioned above including voltage and line flow limits shouldn't also be violated in operating point reached after contingencies; therefore all the probable contingencies that are all the single outages are actually simulated and the power supply constraints, the power balance constraints and the system constraints are all applied. 2.6. Load Economic in Demand Response Program In the first years of the deregulation, Consumers usually had not effective participation in the power markets. They had not enough knowledge and information hardware to effectively participate in the markets. But in the new power market environment, specially, in the new smart grids due to online pricing and the management system, consumers have more potential and facility to take part and benefit through Demand Response. Program (DRP) is divided into two basic categories and several subgroups. The first group is incentive-based program and the second group is time based program which is described in detail in [21-23]. Because of moving to the new smart grid by sending online prices to customers, time based programs have more potential for widely being used in the demand side, |therefore, in this paper TOUP is applied as one of the best choices of demand side program. In TOUP, the electricity prices are determined based on the production costs in the same period [24-26]. Thus, usually the price in the low load period will be cheap, in the off-peak period will be moderate, and will be high in the peak period. By running this program, the consumers, especially those able to move their consumption, will adjust themselves with the prices. So, the peak demand will be reduced and loads will be transferred from the peak period to off-peak and low price periods. Due to the changes of demand versus price, Elasticity is defined as the demand sensitivity with respect to the price
November-December, 2014
ISSN 1013-5316; CODEN: SINTE 8
E
q 0 dq . q0 dp
(14) In this paper, TOUP formulation is used to model the effects of it on the electricity demands and costs, and how the maximum benefit of customers could be achieved due to these programs. For the single period modeling suppose that the customer demand change is as follows [28]:
d (t ) d o (t ) d (t )
( MWh)
(15)
Therefore, the customer’s benefit, S ($), for ith hour will be as follow: S (d (t )) B(d (t )) d (t ). (t ) $ (16) Where (t) is the spot electricity price in tth hour ($/MWh).and B(d(t)) is the Customer's income in tth hour To maximize the customer’s benefit, customer's consumption which is discussed in detail in [28] in the multi period will be acquired as follows: 24
d (t ) d 0 (t ) E0 (t , t1 ). t1 1
d 0 (t ) .[ (t1 ) 0 (t1 )] 0 (t1 )
t 1,..., 24 (17) Above equation shows how much should be the customer's consumption to achieve maximum benefit in a 24 hours interval. 2.7. Differential Evolution Algorithm Application of the Optimization Problem Differential evolution (DE) developed by Storn and Price [28] has gained more and more attention recently due to its simplicity, strong ability in searching global optimal solution and the efficiency and robustness of the approach. Like nearly all evolutionary algorithms DE is a population based optimizer. The DE provides a population-base search procedure in which the individuals or vectors advance in successive iterations based on the principles of the natural evolution [29] a randomly initialized population composed of NP individuals evolves over several generations to reach an optimal solution. DE includes several steps that is generating the individuals, initialization, load flow and fitness evaluation, defining the global best individual, mutation, cross over, selection and have been explained in detail in [28]. 3. CASE STUDY 3.1. Test case description The capability of the 24 hour unit commitment problem is tested on the IEEE 24 bus, one area Reliability Test System (RTS) which consists of 12 generator buses, 13 load buses, 33 transmission lines and 5 transformers. This system has 6 hydro units, 24 thermal units and 2 nuclear units ranging from 12 MW to 400 MW [30]. The total installed capacity is 3405 MW and the peak base load is 2850 MW. The generation data including ramp –up and ramp down limit were taken from [30]. The generation units cost function coefficients and their capacity limits are given in Table 1 and can be obtained based on [30]. The bidding price of
Sci.Int.(Lahore),26(5),1931-1940 ,2014
spinning reserve associated with each period of unit’s output is given in table 2 and can be obtained from [31]. In this paper the data of hourly load of 23rd week in summer season is considered for studies [32]. The average electricity energy price in this system is assumed to be 20 $/ MWh. the energy tariff for summer weekdays is considered to be 7 $/ MWh in low load period (12 p. m. to 8 a. m. ) , 20 $/MWh in offpeak period ( 8 a.m. to 6p.m. ) and 48 $/MWh for peak period (6 p.m. to 11 p. m. ). The self and cross elasticity data are given in Table 3 and are extracted from [33]. The diagonal elements of the above table indicate the self elasticity of demand in the specified period while the offdiagonal value represent the cross elasticities. However the depicted 3 by 3 matrix expands to a 24 by 24 matrix due to the performed 24 unit commitment. The results of unit commitment problem is tested on 2 cases .In the first case , the result of 24-hour unit commitment by with and without using TOUP is compared with each other, in this case no load shedding is allowed. In the second case, the result of 24-hour unit commitment by applying TOUP is compared with the case that no Demand Response program is applied but allows for load shedding through stochastic model. 3.2. Case 1 In this case the load shedding is allowed neither for original load nor for changed load using TOUP. Fig. 1. Shows the initial load curve and the changed load curve due to running of TOUP. It can be seen that based on the differences of the prices and elasticities in different periods, loads are transferred from expensive periods to cheap periods. Fig. 2. Shows the load and reserve level due to running TOUP. The total cost of energy and reserve for the case no Demand Respond Programs applied that is the case with original load curve is 661293 $, compared to the TOUP which is 640438 $. The hourly costs of running unit 2600 2400 Load Level
[27]:
2200 2000 1800 1600 1400
Original Load Curve Load Curve After TOUP
2
4
6
8
10
12 14 Time(hour)
16
18
20
22
24
Figure 1. Original load curve and the changed load curve due to running of TOUP.
2000 Capacity(MW)
1934
1500 1000 500 0
Load Reserve 5
10
15
20
Time(hour)
Figure 2. Load level and reserve level due to running of TOUP program.
November-December, 2014
Sci.Int(Lahore),26(5),1931-1940,2014
ISSN 1013-5316; CODEN: SINTE 8
commitment formulation, that is the cost of energy and reserve in every hour for each program, with and without considering TOUP program with their related load level is shown in. table 4and Fig. 3. It is notable that all the constraints which are mentioned before in steady state and all contingencies are in their limit.
Figure 3. Cost of original load curve and changed load after using TOUP without load shedding. By comparison of the value of total cost and the hourly costs in different load levels, the following results are observed: Applying the TOUP leads to smoothing of the load curve so that the load level in peak hours decreases in contrast to the increase of demand in low periods. This is further demonstrated in Fig. 1. The total cost in TOUP is considerably lower than that in the original load case. This difference of 20855 $ is due to the fact that TOUP keeps the operating points in medium levels and the generation of units will be more economical considering the non-linear generation cost curve of units; further more there is no need to use the more expensive units in peak hours. The sudden increase in load level of successive hours leads to a considerable additional increase in the total cost; this increase in total cost apart from the rise of load level is due to ramp up limit of generating units especially the units with lower generating cost, since the generation of cheaper units are restricted by previous operating point and the ramping rate thereby they cannot deliver as much power as they should. For instance in hour 9 and hour 23, there is the same load level of 2231 MW, however the total cost of hour 9 is 29989 $ which is much higher 28780 $ of hour 23, the rise of load level from hour 8 to 9 is 282 MW compared to the rise of load level from hour 22 to 23 that is 154 MW. From the above discussion, it can be inferred that the smoothness of load curve as a result of TOUP can cause an additional saving in total cost apart from reducing demand in peak hours and the related savings. It can be seen in table 4 that the load level of hour 9 in no program is very close to that of hour 14 in TOUP; however the cost of hour 14 is lower than that in hour 9, Even though the load level of hour 14 is a bit higher. Also the cost in hour 8 of no program is higher than hour 8 of TOUP with nearly same load levels.
1935
Response program is applied, and the load is inelastic, but it allows for load shedding if necessary in order to reduce the costs. The specifications of the test case for these two approach is the same, except that additional data for the probability of each unit outage are obtained based on unit reliability indices [32] and the loss of curtailed load added to the total cost in the second approach [31]. This value is obtained from the sum of lost load value in each contingency that is the product of the Expected Energy Not Served and the Value Of Lost Load, i.e. EENSi*VOLL, the EENSi is obtained by product of the amount of energy that demand loose in occurrence of an outage and the probability of each outage. The results of running UC for the mentioned approaches are depicted in Table (5). It can be seen that the total cost of procuring energy and reserve in 24 hours by applying TOU program is lower than that, when load is considered inelastic, despite that in our approach the security is maintained whit no load shedding in occurrence of any unit outage. The total cost of energy with TOU program is (640438$) while the total cost in the second approach is (678851$). It is depicted in Fig. 4 for better comparison. It is observed from this figure that the amount of decrease of cost in high price period is greater than the amount of increase of cost in low price period and therefore total amount of cost due to TOUP decreases. 4. CONCLUSION Because of moving to new smart grids this paper proposes a new formulation of the UC problem that take into account directly demand response program namely TOUP .Differential Evolution algorithm is used as an optimization tool for deriving the global optima of the problem.
Figure 4. Cost of original load curve with load shedding through stochastic method and changed load after using TOUP.
3.3. Case 2 In the second case, the result of 24-hour unit commitment by applying TOUP is compared with the case that no Demand November-December, 2014
ISSN 1013-5316; CODEN: SINTE 8
1936
Sci.Int.(Lahore),26(5),1931-1940 ,2014
TABLE 1. THE GENERATION UNITS COST FUNCTION COEFFICIENTS AND THEIR CAPACITY LIMITS PiMin
PiMax
ai
bi
ci
Unit
(MW)
(MW)
K$/(MW2)
K$/(MW)
K$
1
2.4
12
0.0253
25.5472
24.3891
2
2.4
12
0.02649
25.6753
24.4110
3
2.4
12
0.02801
25.8027
24.6382
4
2.4
12
0.02842
25.9318
24.7605
5
2.4
12
0.02855
26.0611
24.8882
6
4
20
0.01199
37.5510
117.7551
7
4
20
0.01261
37.6637
118.1083
8
4
20
0.01359
37.7770
118.4576
9
4
20
0.01433
37.8896
118.8206
10
15.2
76
0.00876
13.3272
81.1364
11
15.2
76
0.00895
13.3538
81.2980
12
15.2
76
0.00910
13.3805
81.4641
13
15.2
76
0.00932
13.4073
81.6259
14
25
100
0.0623
18.0000
217.8952
15
25
100
0.00612
18.1000
218.3350
16
25
100
0.00598
18.2000
218.7752
17
54.25
155
0.00463
10.6940
142.7348
18
54.25
155
0.00473
10.7154
143.0288
19
54.25
155
0.00482
10.7367
143.3179
20
54.25
155
0.00487
10.7583
143.5972
21
68.95
197
0.00259
23.0000
259.1310
22
68.95
197
0.00260
23.1000
259.6490
23
68.95
197
0.00263
23.2000
260.1760
24
140
350
0.00153
10.8616
177.0575
25
100
400
0.00194
7.4921
310.0021
26
100
400
0.00195
7.5031
311.9102
November-December, 2014
Sci.Int(Lahore),26(5),1931-1940,2014
ISSN 1013-5316; CODEN: SINTE 8
1937
TABLE 2. BIDDING PRICE OF SPINNING RESERVE ASSOCIATED WITH EACH PERIOD OF UNIT’S OUTPUT Unit number
Perid 1 (MW)
bidding price SR($/MW)
Perid 2 (MW)
bidding price SR($/MW)
Perid 3 (MW)
bidding price SR($/MW)
1-2
15.8-16
2.50
16-19.8
3.098
19.8-20
3.6
3-4
15.2-38
2.41
38-60.8
2.6679
60.8-76
3.1
5-6
15.8-16
2.50
16-19.8
3.098
19.8-20
3.6
7-8
15.2-38
2.41
38-60.8
2.6679
60.8-76
3.1
9-11
25-50
1.400
50-80
1.5109
80-100
1.608
12-14
68.9-118
1.43
118-57.6
1.5044
157.6-197
1.5707
15-19
2.4-6
1.01
6-9.6
1.0900
9.6-12
1.24
20
54.25-93
4.19
93-124
4.3565
124-155
4.565
21
54.25-93
4.44
93-124
4.3565
124-155
4.566
22
100-200
4.44
200-320
4.5392
320-400
4.666
23
100-200
4.4495
200-320
4.5392
320-400
4.660
24-25
54.25-93
2.06
93-124
2.1782
124-155
2.2820
26
140-227.5
4.32
227.5-280
4.5334
280-350
4.7500
TABLE 3. SELF AND CROSS ELASTICITIES PEAK
OFF-PEAK
LOW
PEAK
-0.10
0.016
0.012
OFF-PEAK
0.016
-0.10
0.01
LOW
0.012
0.01
-0.10
TABLE 4. THE HOURLY COSTS OF RUNNING UNIT COMMITMENT FORMULATION FOR CASE 1 Load MW
Load MW Cost ($)
Hours
Original
Cost ($) TOUP
NOP
TOUP
1
1641
23867
1854
25413
2
1539
19252
1738
20857
3
1487
18143
1680
20747
4
1436
18027
1622
20025
5
1436
17096
1622
20003
6
1487
18590
1680
20157
7
1641
20379
1854
23120
8
1949
24661
1945
23989
9
2231
29989
2181
27853
10
2436
33718
2133
27938
November-December, 2014
ISSN 1013-5316; CODEN: SINTE 8
1938
Sci.Int.(Lahore),26(5),1931-1940 ,2014
11
2539
33189
2223
29482
12
2565
33180
2245
29698
13
2539
32240
2223
29511
14
2565
32750
2245
29274
15
2565
33015
2245
29716
16
2488
32102
2178
29716
17
2462
31810
2406
31081
18
2462
31839
2406
31129
19
2385
31317
2331
29431
20
2359
29964
2306
29365
21
2359
30061
2306
29526
22
2385
30372
2331
29758
23
2231
28780
2181
27154
24
2077
26952
2030
25495
TABLE 5. THE HOURLY COSTS OF RUNNING UNIT COMMITMENT FORMULATIONFOR CASE 2 Load MW
Cost ($)
Load MW
Cost ($)
Original
Through stochastic
TOUP
Through TOUP
Hours
method 1
1641
24721
1854
25413
2
1539
19252
1738
20857
3
1487
18543
1680
20747
4
1436
18427
1622
20025
5
1436
17906
1622
20003
6
1487
18790
1680
20157
7
1641
20779
1854
23120
8
1949
25661
1945
23989
9
2231
30980
2181
27853
10
2436
34717
2133
27938
11
2539
34185
2223
29482
12
2565
34185
2245
29698
13
2539
33240
2223
29511
14
2565
33753
2245
29274
15
2565
34015
2245
29716
16
2488
33102
2178
29716
17
2462
32810
2406
31081
18
2462
32839
2406
31129
November-December, 2014
Sci.Int(Lahore),26(5),1931-1940,2014
ISSN 1013-5316; CODEN: SINTE 8
1939
19
2385
31317
2331
29431
20
2359
30864
2306
29365
21
2359
30961
2306
29526
22
2385
31372
2331
29758
23
2231
29080
2181
27154
24
2077
27352
2030
25495
In this model, the required reserve was determined with regards to system credible contingencies and demand response programs. The results of unit commitment problem are tested on the IEEE 24 bus on 2 cases with and without allowing load shedding. The result bus demonstrated that applying TOUP can lead to smoother load level, more optimum value of spinning reserve and consequently the total cost is minimized effectively. REFERENCES [1] Verbic, G. Gubina, F. “Cost-based models for the powerreserve pricing of frequency control”, IEEE Trans. Power Syst, 19(4): 1853-1958 (2004). [2] Zhong, J. Bhattacharya, K. “Design of competitive markets for spinning reserve service”, Power Engineering Society Summer Meeting, pp. 1627-1632 (2002). [3] Zhao, H. Bhattacharya, K. Zhong, J. “A spinning reserve market considering security and biddable reserve,” Power India Conference, (2006). [4] Liu, Y. Alaywan, Z. Liu, Sh. Assadian, M. "A rational Buyer’s algorithm used for ancillary service procurement", IEEE Power Engineering Society Winter Meeting, vol. 2, pp. 855-860 (2000). [5] Wu, T. Rothleder, M. Alaywan, Z. Papalexopoulos, A. D. “Pricing energy and ancillary services in integrated market systems by an optimal power flow”, IEEE Transactions on Power Systems, 19(1): 339-347 (2004). [6] Zheng, T. Litvinov, E. “Contingency-based zonal reserve modeling and pricing in a co-optimized energy and reserve market”, IEEE Transaction on Power Systems, 23(2): 277-286 (2008). [7] Chen, J. Thorp, J. S. Thomas, R. J. Mount, T. D. “Locational pricing and scheduling for an integrated energy-reserve market”, in Proc. 36th Hawaii Int. Conf. System Sciences, pp. 54-63 (2003). [8] Baldick, R. Helman, U. Hobbs, B. F. O’Neill, R. P. “Design of efficient generation markets”, Proc. IEEE, 93(11): 1998-2012 (2005). [9] Liu, Y. Alaywan, Z. Rothleder, M. Liu, S. Assadiam, M. “A rational buyer’s algorithm used for ancillary service procurement”, in Proc. IEEE PES Summer Meeting, Singapore, (2000). [10] Galiana, F. D. Bouffard, F. Arroyo, J. M. Restrepo, J. F. “Scheduling and pricing of coupled energy and primary, secondary, and tertiary reserves”, Proc. IEEE, 93(11): 1970-1983 (2005).
[11] Rau, N. S. “Optimal dispatch of a system based on offers and bids—a mixed integer LP formulation”, IEEE Trans. Power Syst, 14(1): 274-279 (1999). [12] Kirschen, D. S. “Power system security”, Power Eng. J., 16(5): 241–248 (2002). [13] Chattopadhyay, D. Baldick, R. “Unit Commitment with probabilistic reserve”, Power Engineering Society Winter Meeting, pp. 280-285 (2002). [14] Arroyo, J. M. Galiana, F. D. “Energy and reserve pricing in security and network-constrained electricity markets”, IEEE Trans. On Power Systems. 20(2): 634643 (2005). [15] Bouffard, F. Galiana, F. D “An electricity market with a probabilistic spinning reserve criterion”, IEEE Trans. On Power Systems. 19(1): 300-307 (2004). [16] Dhillon, J. S. Parti, S. C. Kothari, D. P. “Stochastic economic emission load dispatch”, Elect. Power Syst. Res, 26(3): 179-186 (1993). [17] Chang, C. S. Fu, W. “Stochastic multiobjective generation dispatch of combined heat and power systems” Proc. Inst. Elect. Eng., Gen. Transm. Distrib., 145(5): 583-591 (1998). [18] Ozturk, U. A. Mazumdar, M. Norman, B. A. “A solution to the stochastic unit commitment problem using chance constrained programming,” IEEE Trans. Power Syst. , 19(3): 1589-1598 (2004). [19] Hajian-Hoseinabadi, H. Fotuhi-Firuzabad, M. Hajian, M. “Optimal allocation of spinning reserve in a restructured power system using particle swarm optimization”, Universities Power Engineering Conference, pp 1-4 (2008). [20] Porkar, S. Fotuhi-Firuzabad, M. Abbaspour-Tehrani fard, A. A. T. Porkar, B. “An approach to determine spinning reserve requirements in a deregulated electricity market”, Power Systems Conference and Exposition, PSCE '06, pp1341-1344 (2006). [21] USAID/IIEC, “DSM Best Practices Guidebook 2005. [22] Canadian Energy Efficiency Alliance, “Demand side management framework for Ontario”, February 2004. [23] FERC, "Regulatory commission survey on demand response and time based rate programs/ tariffs", www. FERC. Gov, August 2006 [24] Li, Y. Peter, F. "Deregulated power prices: changes over time", IEEE Transaction on Power Systems, 20(2): 565-572 (2005). [25] Tang, Y. Song, H. Ho, F. "Investigation on TOU pricing principles", IEEE TBD conference, pp. 1-9 (2005).
November-December, 2014
1940
ISSN 1013-5316; CODEN: SINTE 8
[26] Kirschen, D. S. Strbac, G. "Fundamentals of power system economics", John Wiley & Sons, N.Y, 2004. [27] Aalami, H. A. Yousefi, G. R. Parsa Moghadam, M. "Demand Response Model Considering EDRP and TOU Programs", PES 2008, Chicago, USA, 2008. [28] Price, K. V. Storn, R. M. Lampinen, J. A. “Differential evolution-a practical approach to global optimization,” Natural Computing Series, SpringerVerlag 2005. [29] Vesterstrom J. Thomsen, R. “A comparative study of differential evolution, particle swarm optimization, and evolutionary algorithms on numerical benchmark problems,” in Proc. Congr. Evolutionary Computation, Portland, OR, pp. 1980–1987 (2004).
Sci.Int.(Lahore),26(5),1931-1940 ,2014
[30] Grigg, C. Wong, P. Albrecht, P. Allan, R. Bhavaraju, M Billinton, R. Chen, Q. Fong, C. Haddad, S. Kuruganty, S. Li, W. Mukerji, R. Patton, D. Rau, N. Reppen, D. Schneider, A. Shahidehpour, M. Singh, C. “The IEEE reliability test system - 1996,” IEEE Trans. Power Syst, 14(3): 1010-1018 (1999). [31] Hajian-Hoseinabadi, H. Fotuhi-Firuzabad, M. Hajian, M. “Optimal allocation of spinning reserve in a restructured power system using particle swarm optimization,” Universities Power Engineering Conference, pp 1-4 (2008). [32] Billinton R. Allan, R. N. Reliability Evaluation of Power Systems. New York: Plenum, 1996. [33] Kirschen, D. S. Strbac, G. Cumperayot, P. Mendes, D. "Factoring the elasticity of demand in electricity prices", IEEE Transaction on Power Systems, 15(2): 612-617 (2000)
November-December, 2014