Assessment of Marginal and Long-term Surplus Power in Orissa A Case Study

Transcription

1 1 Chandra 16th NATIONAL POWER SYSTEMS CONFERENCE, 15th-17th DECEMBER, Assessment of Marginal and Long-term in Orissa A Case Study Chandra Shekhar Reddy Atla, A.C. Mallik, Dr. Balaraman K and Dr. Nagaraja R. 1Abstract- A power system serves to supply customers with electrical energy as economical as possible and with an acceptable degree of reliability and quality. Probabilistic methods get importance in order to meet the customer demands with economical investment and acceptable reliability of supply as deterministic methods used in past are not respond to nor reflect the probabilistic or stochastic nature of system behavior of customer demands or of component failures. In the paper two types of probabilistic methods namely, analytical and Monte Carlo methods are developed and are validated with IEEE RTS. A case study has been carried out for a surplus state, Orissa, in the country to analyze the quantum of surplus power or deficiency of power during 11 th and 12 th plan during planning phase. I. INTRODUCTION A power system serves one function only and that is to supply customers, both large and small, with electrical energy as economical as possible and with an acceptable degree of reliability and quality. Present day society has come to expect that the supply should available on demand. This is not physically possible due to random system failures which is generally outside the control of power system engineers, although the probability of customers being disconnected can be reduced by increasing investment during either the planning phase, operating phase or both. It is evident therefore that the economic and reliability constraints can conflict, and this can lead to difficult managerial decisions at both the planning and operational phases [1]. The criteria and techniques generally used in practical applications were all deterministically based (like percentage reserve method based on largest unit size or fixed percentage of demand) during planning and operating phases. The essential weakness of deterministic criteria is that they do not respond to nor reflect the probabilistic or stochastic nature of Shekhar Reddy Atla, System Engineer, Member IEEE, is with M/s Research and Development Consultants Pvt. Ltd., Bangalore, India. ( sekhar.atla@gmail.com) A.C. Mallik, Director, M/s. GRIDCO Ltd, Bhubaneswar, Orissa, India. Dr. Balaraman K, CGM Systems, Member IEEE, is with M/s Research and Development Consultants Pvt. Ltd., Bangalore, India. ( balaraman@prdcinfotech.com) Dr. Nagaraja R, Managing Director, Member IEEE, is with M/s Research and Development Consultants Pvt. Ltd., Bangalore, India. ( nagaraja@prdcinfotech.com) system behavior, of customer demands or of component failures. A functional problem in system planning is the correct determination of reserve margin. Too low a value means excessive interruption; while too high a value results in excessive investment. Forced outage rates (FOR) of generating units are known to be a function of size and type and therefore a fixed percentage reserve cannot ensure a consistent risk. The need for probabilistic evaluation of system behavior has been recognized since at least the 1930 s, but it was not widely used. The main reasons were lack of data (like FOR of generating units), limitations of computational resources, lack of realistic techniques, aversion to use of probabilistic and a misunderstanding of the significance and mean of probabilistic criteria and indices. None of these reasons need to valid today as most utilities have reliability databases, computing facilities are greatly enhanced, reliability evaluation techniques are highly developed and most engineers have a working understanding of probabilistic techniques. In the report probabilistic methods namely analytical and Monte-Carlo simulation methods are developed in VC++ environment and both models are validated with IEEE reliability test system (RTS). The developed models are applied to determine the quantum of surplus power or deficiency power during 11 th plan ( to ) and 12 th plan ( to ) in Orissa, India. II. METHDOLOGY The basic aspects of power system reliability are system adequacy and system security [1]. System adequacy involves the existence of sufficient facilities in the system to satisfy the customer demand. These facilities include the generating capacity required to generate enough energy and transmission and distribution elements needed to transfer the generated energy to the customer load points. System security, however, concerns the ability of the system to respond to disturbances. systems have to maintain certain levels of static and spinning reserves in order to achieve a required level of adequacy and security. A power system consists of three functional zones of generation, transmission and distribution shown in Fig. 1. These functional zones can be combined to form hierarchical levels. Hierarchical Level I (HL-I) is concerned with only the generation facilities, while Hierarchical Level II (HL-II) includes both the generation and transmission facilities,

2 16th NATIONAL POWER SYSTEMS CONFERENCE, 15th-17th DECEMBER, Hierarchical Level III (HL-III) includes all the three functional zones to provide a complete system. Adequacy evaluation at HL-I involves the determination of the total system generation required to satisfy the total load requirement and the model for adequacy evaluation at HL-I is shown in Fig. 2. The basic approach to perform adequacy evaluation at HL-I consists of three segments shown in Fig. 3. The generation model and the load model are combined to form the risk model. The risk indices obtained are overall system adequacy indices and do not include transmission constraints and transmission reliabilities. Fig. 1: Hierarchical level Fig. 2: Model for adequacy evaluation at HL-I load characteristics, and the potential assistance available from neighboring systems. The LOLP or LOLE approach is by far the most popular and can be used for both single systems and interconnected systems. Expectation indices are most often used to express the adequacy/deficiency/surplus of the generation systems while meeting the demand with sufficient capacity to perform corrective and preventive maintenance on the generating facilities. A wide range of methods has been developed to perform generating capacity reliability evaluation. These techniques can be categorized into two types, analytical methods and Monte Carlo simulation method. Analytical methods represent the system by mathematical model and evaluate the reliability indices from this model using mathematical solutions. Mont Carlo simulation methods, however, estimate the reliability indices by simulating the actual process and random behavior of the system. The method therefore treats the problem as a series of real experiments. There are merits and demerits in both methods. Generally Monte Carlo simulation requires a large amount of computing time as compared to analytical methods, but with present day availability of high speed computers this drawback is not considerable. A. Analytical Method The generation model is formed in analytical method by creating a capacity outage probability table (COPT). This table represents capacity outage state of the generating system together with the probability of each state. A.1. Generating unit Models: The two state model for base load generating unit is shown in Fig. 4. The base load unit can also be modeled with derated or partial output state as shown in Fig. 5. This model can be expanded to include more derated states. Fig. 4: Two state model for base load unit [1] Fig. 3: Conceptual model in adequacy assessment at HL-I. Criteria such as loss of load probability (LOLP, the probability of occurrence of loss of load), loss of load expectation (LOLE, expected number of days (hours) where the loss of load occurs), and loss of energy expectation (LOEE) are now widely used by electric power utilities. There is considerable confusion within and outside the power industry on the specific meaning of these expected indices and the use that can be made of them. A single expected value is not a deterministic parameter: it is mathematical expectation of a probability distribution, i.e. it is therefore the long run average value. These expected indices provide valid adequacy indicators which reflect factors such as generating unit size, availability, maintenance requirements, Fig. 5: Three state model for base load unit [1]. where, ST is service time, FOT is forced outage time, P up is probability of generating unit in UP state, P down is probability

3 16th NATIONAL POWER SYSTEMS CONFERENCE, 15th-17th DECEMBER, of generating unit in DOWN state, P derated is probability of generating unit in DERATED state. Peak load generating unit can be modeled by using four state model [2] and is presented in Fig. 6. Fig. 6: IEEE four-state model for a peak load generating unit [2] Conditional forced outage rate, P = f *( FOT ) ST f *( FOT ) 1 T and, demand factor, f 1 1 D T where, T is average reserve shutdown time between periods of need, exclusive of periods for maintenance (hrs), D is average in service time per occasion of demand, P s is probability of starting failure. In practice the utilities will not have the above full data for peak load generating units, under this condition; these energy limited units can be modeled based on their energy availability considering the unit past history [3]. A.2. Recursive Algorithm: Capacity outage probability table (COPT) used in loss of load method can be formed using a recursive algorithm [1]. The cumulative probability of a particular capacity outage state of X MW after a unit capacity C MW and forced outage rate U is added is given by 1 1 P( X ) (1 U ) P ( X ) ( U ) P ( X C)...for 2-sate model n 1 ( ) i ( )...for derated units i 1 P X PP X C where, P (X) and P(X) is the cumulative probability of the capacity outage state of X MW before and after the unit is added, n is number of unit states, C i is capacity outage state i for the unit being added, p i is probability of existence of the unit state i. The COPT is complete after all the generating units in the system are added. A.3. Loss of Load Method: The generation system model illustrated in the previous section can be convolved with the system load model to produce a system risk model as shown in Fig. 3. The load model can be either be the daily peak load variation curve (DPLVC) which only includes the peak loads of each day, or the daily peak load duration curve (LDC) which represents the hourly variation of the load. In the loss of load method, the daily peak loads (or hourly values) are combined with the COPT to obtain the expected number of days (or hours) in the given period in which the daily peak load (or hourly load) exceeds the available capacity. The index in this case is designated as the loss of load expectation (LOLE). The LOLE is given by: n LOLE P( C L ) days or hours/period i 1 i i i where, n is number of days or hours of period under scope, C i is available capacity on the day or hour i, L i is forecasted peak load on day or hour I, P i (C i -L i ) is cumulative probability of loss of load on day or hour i, this value is obtained directly from the capacity outage cumulative probability table. The loss load probability (LOLP) is given by: LOLE LOLP (Study period in years* x) where, x is 365 days when the daily peak load is considered or 8760 when hourly load data is considered. After the calculation of expected indices like LOLP or LOLE, these indices will be compared with standard index value specified by utilities and the sensitivity study is required to determine the adequacy or surplus power or deficiency power while meeting the demand with available generation considering the limitation of reliability index. B. Monte Carlo Simulation Method Monte-Carlo Simulation methods estimate the reliability indices by simulating the actual process and random behavior of the system. Probabilistic or stochastic simulation itself can be used in one of two ways: sequential or random. The sequential approach simulates the basic intervals in chronological order. The random approach simulates the basic intervals of the system lifetime by choosing intervals randomly. In the paper the model is developed based on random approach and is presented below [1]. Modeling unit states in Monte Carlo Simulation method for two state unit is achieved by generation of uniform random number, U, in the range (0, 1). This value of U is compared with the FOR. If U < FOR, then the unit is deemed to be in down state; otherwise the unit is deemed to be available. This principle can be extended to any number of states. For three state unit as shown in Fig. 5, a uniform random number U is again generated. If U < P(down), the unit is deemed to be in the totally down state; if P(down) < U < [P(down)+ P(derated)], the unit is deemed to be in derated state; otherwise the unit is deemed to be available. After the operating histories of all the generating units are generated, the system available capacity during one simulation time span is calculated. The super imposition of the available capacity on the load is shown in Fig. 7. The loss of load occurrence is the total number of occurrences of loss of load during the study period. From this analysis LOLE can find in days or hours/period of study based on the load profile considered.

4 16th NATIONAL POWER SYSTEMS CONFERENCE, 15th-17th DECEMBER, Fig. 7: Superimposition of the available capacity on the load III. VALIDATION Probabilistic methods namely analytical and Monte-Carlo simulation methods have been developed in VC++ environment based on the methodology presented in previous section. Both programs are validated with IEEE Reliability Test System (RTS) [4] and presented below. The total installed capacity of the RTS is 3405 MW with peak load of 2850 MW. Table 1 presents a set of representative results for validation purposes at selected capacity levels. From Table 1, it is seen that the probability values are matching up to 5 to 8 decimal points. State Cap. Out TABLE 1 REPRESENTATIVE GENERATION MODEL DATA Analytical Method by IEEE RTS [1] Individual Cumulative Individual Analytical Method Developed program Cumulative probability probability probability probability In practice, COPTs are truncated and rounded, which, when convolved with load models, which may also be approximated, can give results with varying degrees of inaccuracy based on the computer precision [5]. Table 2 presents the LOLE calculation results for IEEE RTS by developed model and results published in [5] using analytical method. The difference in degree of accuracy is based on the precision of computers used [5]. LOLE (days/year) TABLE 2 VALIDATION OF LOLE CALCULATION Analytical Methodpublished [5] ENS: Energy Not Supplied Analytical Method Developed program LOLE hrs/year) Monte Carlo Simulation Method IV. SYSTEM STUDIES The generation reliability evaluation method, Monte-Carlo Simulation Method, mentioned in section II is applied to Orissa (in India) power system to evaluation the adequacy/surplus/deficiency in generation system while meeting the system demand with reliability index, LOLP, limit of 2% according to OERC Standard [6]. The study has been performed for the 11 th plan ( to ) and 12 th plan ( to ) durations. The study results are presented by annually and monthly during 11 th and 12 th plans. The captive power plant (CPP s) support is not considered in the analysis. The present generation in Orissa without CPP s is around 4894 MW, with hydro generation of 2085 MW, Thermal generation in Orissa of 880 MW and the central sector share of 863 MW. KAHALGAON generation (Orissa share) is around 144 MW is considered from onwards. The future units, which are going to commence in future, have been considered in the study based on the information provided by M/s Gridco Ltd. According to this, the year wise capacity additions (Orissa Share) during 11th and 12th plans are presented in Table 3. Capacity addition by month wise has been considered for the monthly studies. TABLE 3 YEAR WISE CAPACITY ADDITION DURING 11 TH AND 12 TH PLANS S.No. Year Capacity Addition, Orissa share, in MW The Forced Outage rates (FOR) and auxiliary consumption is considered from the OPTCL monthly reports for existing units and for future units it is considered from the OERC Standard [6] and same is presented in Tables 4 and 5. TABLE 4 GENERATING UNIT PLANNED OUTAGE RATES AND FOR [6] Unit Type Planned Outage Forced Outage (days/year) (%) (days/year) (%) Hydro Electric to to 10 Steam Thermal Gas turbine TABLE 5 GENERATING UNIT AUXILIARY CONSUMPTION [6] Size/Type Auxiliary Consumption 1 Coal based Thermal Station i) 200 MW ii) 500 MW 9.5% 8.0% 2 Gas Based Thermal Station i) Combined cycle ii) Open cycle 3.0% 1.0% 3 Hydro Station 0.5% The forecasted demand including transmission loss by OPTCL is used for the analysis and the same is presented in

5 16th NATIONAL POWER SYSTEMS CONFERENCE, 15th-17th DECEMBER, Table 6. The hourly data of is considered as the base data. According to Table 6, the base hourly data is forecasted for the future years during 11 th and 12 th plans. TABLE 6 FORECASTED DEMAND BY OPTCL S.No. Year Forecasted demand including transmission loss in MW Based on the data presented above, the Monte-Carlo simulation method was developed as mentioned in section II. Base load units are modeled by considering the FOR and outage rates. Hydro units or energy limited units are modeled by their specific capacity and available energy [4], and it is available yearly energy for annual studies, available monthly energy for monthly studies. By convolving the operating histories of all the generating units and load duration curve the reliability index, LOLP, is evaluated for different cases presented below. A. Year ahead Studies The following points are considered to model the system for annual studies and the results are presented in Table 7. The maximum demand that can meet with available generation with LOLP of 2% is also presented in Table 7. From Table 7, it is seen the annual surplus power is available only from onwards. 1. Generation unit availability is considering base on the unit and forced outages [6]. 2. All available generation up to the study year is considered. (Ex: for study, the generation available up to 1 st, April, 2012 is considered). 3. LOLP of 2% is considered. 4. The following information is used for the calculation of surplus power from Base Load Units. a. Hydro unit capacity is not considered. b. Unit capacity is considered after auxiliary consumption. c. Both forced outage rate and outage rate is considered for unit availability. B. Month ahead Studies The studies have been formed for all the months during 11 th and 12 th plans and the results are presenting only for and where the surplus power is available. The following points are considered to model the system for monthly studies and study results are presented in Tables 8 and Generation unit availability is considering based on the forced outages [6]. S.No. Study year TABLE 7 ANNUAL GENERATION EVALUATION FOR 11 TH AND 12 TH PLAN Total available capacity (Hydro Capacity) Peak Demand Capacity Max. peak demand required to that can meet with meet peak available generation demand with with LOLP of 2% LOLP of 2% (from thermal, central sector & future units) TABLE 8 MONTHLY GENERATION EVALUATION FOR THE YEAR Total Generation S. No Month available Forecast ed Peak Demand Max. peak Capacity demand that can meet with available generation required (after aux.) to meet peak demand with LOLP of 2% 1 April [without outage rate] No surplus 1340 [with outage rate] No surplus 2 May June July Aug Sep Oct Nov Dec Jan Feb Mar All available generation up to the study month is considered. (Ex: for April, 2012 medium - term study, the generation available up to 1 st, April, 2012). 3. LOLP of 2% is considered. 4. The following information is used for the calculation of monthly surplus power from Base Load Units: a. Hydro unit capacity is not considered. b. is presented in with and without outage rate (scheduled maintenance is considered based on the demand).

6 16th NATIONAL POWER SYSTEMS CONFERENCE, 15th-17th DECEMBER, c. Capacity is considered after auxiliary consumption. d. The presented available monthly surplus power is after the allocation of yearly surplus power. For example, for the year , according to annual studies, surplus power allocated is 1340 MW. After allocating this surplus power, the available surplus power month of August, 2016 is 3909 without considering the scheduled maintenance (by subtracting yearly surplus of 1340 MW from total monthly surplus of 5249 MW). Total S. No Month Generation available TABLE 9 MONTHLY GENERATION EVALUATION FOR YEAR Forecasted Peak Demand Capacity required (after aux.) to meet peak demand with LOLP of 2% [without outage rate] [with outage rate] 1340 MW is available for the year and the monthly surplus will be available from the month May, 2015 onwards. REFERENCES [1] Roy Billinton and R. N. Allan, Reliability Evaluation of Systems, second edition, Plenum Press, New York, [2] Report of the IEEE Task Group on Models for peaking Service, A four State Model for Estimation of Outage Risk for Units in Peaking Service, IEEE Transaction on Apparatus and Systems, Vol PAS-91, no.2, Mar/Apr 1972, pp [3] IESO 2009 Comprehensive Review of Resource Adequacy, Covering the Ontario Area for the period 2010 to 2014, Aug-2009, Approved by NPCCRCC. [4] A Report prepared by the Reliability Test System Task Force of the Application of Probability Methods Subcommittee, The IEEE Reliability test System 1996 IEEE Transactions on power Systems, vol. 14, no.3, August 1999, pp [5] R.N. Allan, Roy Billinton and N.M.K. Abdel-Gawad, The IEEE Reliability Test System Extensions to and Evaluation of the Generating System, IEEE Transactions on Systems, vol.pwrs- 1, no. 4, Nov. 1986, pp.1-7. [6] Transmission and Bulk Supply Standards, OERC Standard, India. 1 April May June July Aug Sep Oct Nov Dec Jan Feb Mar From the Table 8, it is seen that the surplus power is available from May, 2015 onwards to sale outside Orissa and From Table 9, it is seen that the surplus power is available in all the months of the year V. CONCLUSIONS The probabilistic methods, analytical and Monte Carlo simulation, are used in the analysis to determine the adequacy/surplus in generation system. The developed programs, analytical and Monte Carlo simulation, are validated with results published in [1] and [5]. The Monte- Carlo Simulation method has been applied to Orissa power system to determine the quantum of surplus power available during 11 th and 12 th plan with available generation while meeting the demand with LOLP of 2% [6]. The generation model and Load model have been developed with the input data provided by M/s Gridco Ltd and is presented in section IV. From the results it is seen that the annual surplus power of