Research Journal of Recent Sciences _________________________________________________ ISSN 2277-2502 Vol. 3(8), 74-79, August (2014) Res.J.Recent Sci. International Science Congress Association 74 Estimating Optimal Size of GenCos in a Restructured Power System, in order to Improve Reliability Based on Monte Carlo and Sensitivity Analysis MethodsMohammad Reza Behmaneshfar, Shahrokh Shojaeian and Ali Behmaneshfar2*Department of Electrical Engineering, Khomeinishahr Branch, Islamic Azad University, Isfahan, IRAN Young Researchers and Elite Club, Najafabad Branch, Islamic Azad University, Najafabad, Isfahan, IRANAvailable online at: www.isca.in , www.isca.me Received 7th February 2014, revised 5th April 2014, accepted 4th June 2014Abstract In this paper, an algorithm is proposed in order to optimal sizing of the GenCos in a restructured power system based on well-known reliability indices. The proposed method uses Monte Carlo simulation to find indices after connecting each GenCo in an appropriate bus which has been found by sensitivity analysis. To show the validation and effectiveness of the proposed method simulation on IEEE reliability test system (IEEE-RTS) are presented. Keywords: Monte Carlo simulation, sensitivity analysis, power system reliability, GenCosizing. Introduction The main objective of a power system is to provide reliable and continuous electricity for its customers. In a restructured power system, it is possible to plan a suitable generation program for individual GenCos which can reduce the effects of contingencies. Because of the probabilistic nature of the contingencies, it is clean that such a plan has to be done with a probabilistic point of view. Well-known reliability indices are used for taking into account direct and indirect of the contingencies1, 2, 3. Wang and Billinton present a technique for evaluating the customer load point reliability in a regulated power system considering customer choice on reliability. In this technique, a generation company (GenCo) is represented by an equivalent multistate generation provider (EMGP) and an equivalent multistate reserve provider (EMRP) based on the market function of a GenCo. A GenCo, which has reserve agreements with other GenCos, is represented by an equivalent multistate generation provider with reserve agreement (EMGPWR). The transmission system between a GenCo and its customer considering reserve agreements is represented by an equivalent multistate transmission provider with reserve agreement (EMTPWR). Reliability network equivalent techniques have been extended and combined with the equivalent assisting unit approach to determine the reliability model of the EMGPWR. Pilo and Gelli improve an algorithm for the optimal allocation of automatic sectionalizing switching devices for the maximum exploitation of intentional islanding. Line faults and overloads have been considered as causes of interruptions. Stochastic models have been adopted to assess the probability of overloads and of properly functioning intentional islands. The application to real world case studies has highlighted the benefits achievable with intentional islanding as well as the inability of common reliability indexes (e.g. SAIFI, SAIDI) to properly perceive advantages that are inherently local. Billinton and Bai propose a methodology for capacity adequacy evaluation of power system including wind energy. The results and discussions on two representative systems containing both conventional generation units and wind energy conversion systems (WECS) are presented. A Monte Carlo simulation approach is used to conduct the analysis. The study shows that the contribution of a WECS to the reliability performance of a generation system can be quantified and is highly dependent on the wind site conditions. In this paper, the most economical size of the GenCos which have been installed in a restructured power system is investigated using combination of Monte Carlo method and sensitivity analysis. Simulations are made on IEEE-RTS to show the effectiveness of the proposed method. Material and Methods In a restructured power system each power generation scheduling that can reduce loss of load expectation (LOLE) of the system will improve its reliability. Then size of GenCos in the IEEE-RTS is obtained by sensitivity analysis so that minimum LOLE to be obtained. There are 32 generation units in this power system, which of them can be considered in up or down state. The Down State probability is known as Forced Outage Rate (FOR). For all units, it is possible to consider 232 states that each can result in its individual total Capacity. 232 is about 4.3 billion and Research Journal of Recent Sciences ______________________________________________________________ ISSN 2277-2502Vol. 3(8), 74-79, August (2014) Res. J. Recent Sci. International Science Congress Association 75 can cause a heavy computation load even for modern digital computers. In all 4.3 billion cases, total capacity has to be calculated and compared with the total load of the system. If the load demand can be supply, this state will be a success state, and else it will be a fail state. In practical power systems, number of generation units is considerably higher. It can complicate the system simulation intensively. To overcome this problem, Monte Carlo method has been proposed. Monte Carlo Simulation: In recent decades, Monte Carlo method is used in different fields of sciences and engineering7-10. As a simple example to show how this method works, a simple one unit constant load power system can be considered. A random number between 0 and 1 is generated. If the generated number is greater than or equal to the unit FOR, this state will be a success state else it is a fail. Repeating this process more and more, results in a converged loss of load probability (LOLP) which is equal to the ratio of the number of fail states in the total number of iterations. Results and Discussion IEEE-RTS is shown in figure 1. Network parameters are mentioned in IEEE-reliability test system11. To simulate the system generation states, a 1*32 matrix with random components between 0 and 1 is generated. If each component is greater than or equal to the FOR of the corresponding unit, this unit is considered on, else it is off. Consequently, total generation capacity can be calculated and compared with the total load demand. If total capacity can meet the total demand this state is a success state else it is fail. This process is repeated as needed to convergence. The LOLE is calculated as follow: 365* numberoffailstates LOLE numberoftotaliterations (1) Simulating IEEE-RTS with this method is converged after 7*10 iterations as shown in figure 2. Figure-1 IEEE-RTS network7 Research Journal of Recent Sciences ______________________________________________________________ ISSN 2277-2502Vol. 3(8), 74-79, August (2014) Res. J. Recent Sci. International Science Congress Association 76 Here LOLE is about 43.84 day/year. To determine optimal size of GenCos, it is needed to find unserved demand for each of 7*10 states which mentioned above. All these states are saved. Then, a typical capacity for each GenCo is considered and number of the fail states which after coming this GenCo to the system can be changed into the success state are obtained. By dividing the number of the changed state to the number of all states (7*10), probability of improving the system LOLP can be calculated. This calculation is repeated for different GenCo sizes to find the sensitivity. Table 1 and figure 3 show the results.As the results show, improving in system LOLE has a saturated form. Thus, increasing the GenCo size from 600 to 1300 MW and even more, has a negligible effect on the system LOLP improvement and then it is not economical. Optimal GenCo size is one which can get the maximum reduction in LOLP as well as the minimum cost. Figure 4 shows sensitivity in the function of GenCo size. It is clear that the maximum sensitivity is for the sizes lower than 150 MW. Then, the figure is zoomed for this section in figure 5 and figure 6. The sizes of 25, 30 and 50 MW have the higher sensitivity respectively. On the other hand, from (1): 365 LOLE LOLP D D= (2) Then, if it is aimed to reduce LOLE, for example for 7 days, it is needed to decrease the LOLP by 1.9178%. Therefore, as table 1 shows, it is needed to use a GenCo by the Capacity of 40 MW or more. Also, from figure 6 GenCo with 50 MW Capacity will be the final choice, which can improve the LOLP with economical cost. With this choice LOLE is become 33.142 days which is 10.7 days lower with respect to the original network (with no GenCo).Figure-2 Convergence of system success probability in Monte Carlo method Figure-3 Probability of reliability improvement based in function of GenCo value in HLI level 0.020.040.060.080.10.120.140200400600800100012001400 Research Journal of Recent Sciences ______________________________________________________________ ISSN 2277-2502Vol. 3(8), 74-79, August (2014) Res. J. Recent Sci. International Science Congress Association 77 Table-1 Calculation probability of production lack in HLI Level Generation curtailment Number of curtailed generation, less than X Variation of probabilityin ideal conditionsSensitivity 10 119010.0017 0.000170 20 746790.010668 0.000533 25 1072150.015316 0.000613 30 1228990.017557 0.000585 40 1392930.019899 0.000497 50 2051870.029312 0.000586 60 2126760.030382 0.000506 70 2484940.035499 0.000507 75 2651350.037876 0.000505 80 3008770.042982 0.000537 90 3150800.045011 0.000500 100 3558190.050831 0.000508 125 4163970.059485 0.000476 150 4581350.065448 0.000436 175 4967160.070959 0.000405 10 119010.0017 0.000170 20 746790.010668 0.000533 200 5483030.078329 0.000392 225 5792320.082747 0.000368 250 6123380.087477 0.000350 275 6544430.093492 0.000340 300 6784700.096924 0.000323 325 7043790.100626 0.000310 350 7302900.104327 0.000298 375 7456520.106522 0.000284 400 7625580.108937 0.000272 425 7741480.110593 0.000260 450 7873670.112481 0.000250 475 7955110.113644 0.000239 500 8045190.114931 0.000230 525 8112860.115898 0.000221 550 8162730.11661 0.000212 575 8200830.117155 0.000204 600 8245430.117792 0.000196 625 8276950.118242 0.000189 650 8296440.118521 0.000182 675 8324370.11892 0.000176 700 8342390.119177 0.000170 725 8352690.119324 0.000165 750 8365220.119503 0.000159 775 8375340.119648 0.000154 800 8382150.119745 0.000150 825 8386830.119812 0.000145 850 8393180.119903 0.000141 875 8397040.119958 0.000137 900 8399310.11999 0.000133 925 8402030.120029 0.000130 950 8404260.120061 0.000126 975 8405430.120078 0.000123 1000 8406310.12009 0.000120 Research Journal of Recent Sciences ______________________________________________________________ ISSN 2277-2502Vol. 3(8), 74-79, August (2014) Res. J. Recent Sci. International Science Congress Association 78 1025 8407270.120104 0.000117 1050 8407830.120112 0.000114 1075 8408190.120117 0.000112 1100 8408600.120123 0.000109 1125 8408840.120126 0.000107 1150 8409020.120129 0.000104 1175 8409130.12013 0.000102 1200 8409210.120132 0.000100 1225 8409300.120133 0.000098 1250 8409330.120133 0.000096 1275 8409360.120134 0.000094 1300 8409380.120134 0.000092 Figure-4 Sensitivity analysis based on GenCo value Figure-5 Sensitivity analysis based on GenCo value for less than 150 M.W. Research Journal of Recent Sciences ______________________________________________________________ ISSN 2277-2502Vol. 3(8), 74-79, August (2014) Res. J. Recent Sci. International Science Congress Association 79 Conclusion In this paper, a simple and relatively fast method is presented to find optimal size of GenCos in a restructured power system. Sensitivity analysis and Monte Carlo method are used to find reliability indices and they are selected as the essential measures for determining the system performance with individual sizes of GenCos. Simulation of the proposed method for IEEE-RTS shows its capability to improve system reliability.References 1.Tollefson G., Billinton R. and Wacker G., Comprehensive bibliography on reliability worth and electric service consumer interruption costs 1980–1990, IEEE Trans. Power Syst., , 1508–1514 (1991)2.Billinton R., Wacker G. and Wojczynski E., Customer damage resulting from electric service interruptions, Canadian Electrical Assoc., Montreal, P.Q., Canada, R and D Rep. 907 U131, 1 and 2, (1982)3.Chowdhury A. A. and Koval D.O., Value-Based Power System Reliability Planning, IEEE Transactions on Industry Applications, 35(2), (1999)4.Wang P. and Billinton R., Reliability assessment of a restructured power system considering the reserve agreements, Power systems, IEEE Transactions on, 19(2), 972-978 (2004)5.Pilo F. and Gelli G., Improvement of reliability in active network with intentional islanding, IEEE International Conference on Electeric Utility Deregulation, Restructuring , and Power Technologies, 474-479 (2004)6.Billinton R. and Bai G., Generating capacity adequacy associated with wind energy, IEEE Trans. Energy Convers.19(3), 641–646(2004) 7.Ali Amjad, Salahuddin, Alamgir, Testing Goodness-of-Fit in Autoregressive Fractionally Integrated Moving Average Models with Conditional Hetroscedastic Errors of Unknown form, Research Journal of Recent Sciences, 2(5), 39-43 (2013) 8.Mohseni Roozbahani K. and Shahzadi A., Simulation of Doppler Frequency Estimation in Satellite Communication Using MIMO-OFDM Technique, Research Journal of Recent Sciences, 3(1), 78-82 (2014) 9.Izadi E., Sobhi H.R., Sajadi S.M. and Behmaneshfar A., Comparing Net Present Value of the Installation of Carbon-Active and Nano-Tube-Carbon Filters Using Monte Carlo Simulation, International Journal of Economy, Management and Social Sciences, 2(6), 318-321 (2013)10.Nasiripour A. Afshar Kazemi M. and Izadi A., Effect of Different HRM Policies on Potential of employee Productivity, Research Journal of Recent Sciences, 1(6), 45-54 (2012) 11.Reliability test system task force of the IEEE subcommittee on the application of probability methods, IEEE reliability test system, IEEE Trans.,98(6), 2047–54 (1979)