On Sizing of Standalone Hybrid Wind/Solar/Battery Micro-grid System

Transcription

1 International Conference on Renewable Energies and Power Qality (ICREPQ 7) Malaga (Spain), 4 th to 6 th April, 27 Renewable Energy and Power Qality Jornal (RE&PQJ) ISSN X, No.5 April 27 On Sizing of Standalone Hybrid Wind/Solar/Battery Micro-grid System U. Akram, M. Khalid and S. Shafiq Department of Electrical Engineering King Fahd University of Petrolem and Minerals (KFUPM) Dhahran 326, Kingdom of Sadi Arabia s: g25293@kfpm.ed.sa, mkhalid@kfpm.ed.sa and saifllah@kfpm.ed.sa Abstract. Capacity optimization of renewable energy sorces sch as solar and wind and the optimal sizing of associated battery energy storage (BES) is essential for economic and reliable operation of a standalone micro-grid (MG). In this work a constraint based iterative search algorithm is proposed to find the optimal sizes of solar, wind and battery capacity in a standalone MG. The proposed method is based on two basic principles, i.e., maximm reliability and imm cost. The proposed techniqe is robst and garantees to avoid the over and nder sizing. In addition it considers the forced otage rates (FORs) of solar and wind and tilization factor of battery storage which makes it more practical. Renewables and demand data of Dammam region of KSA is sed to test the proposed method. A reliability and economic analysis is also carried ot to validate the proposed techniqe. Key words Micro-grid, wind power, solar power, battery energy storage, optimization. Introdction Most recently, tilizing fossil fels sch as natral gas, oil and coal prodces major proportion of electricity. Brning of fossil fels adds massive amont of carbon in the atmosphere. Moreover, fossil fels reserves are depleting rapidly and their prices are flctating. On the other hand, demand is increasing and new environmental policies are also being made to mitigate carbon emission problems. So a decrease in the se of fossil fels and search for other alternatives are essential for reliable and eco-friendly power generation. Renewable energy sorces (RES) can be sed as a sbstitte of fossil fels as they provide green energy and are freely available. Remote areas sch as Islands, which are far away from the tility grid, can also tilize local available green energy sorces to meet their demand [], [2]. Wind and solar are most widely sed renewable sorces for the generation of electricity [3]. Otpt of wind trbine (WT) and solar photovoltaic (PV) are highly variable becase they harvest energy from natral resorces which have low controllability [4], [5]. The intermittent otpt and reliability problems associated with RES can potentially be addressed by sing energy storage systems (ESS) [6]-[9]. In particlar, battery energy storage (BES) can be employed for energy storage in a system tilizing renewables []-[2]. A hybrid wind solar generation system tilizes both wind and solar to extract the maximm available energy. It is more reliable and efficient than solar or wind alone. In addition a hybrid wind-solar generation system reqires less storage capacity making it more economical [3]. In addition to these benefits, the most important is to design a micro-grid (MG) system with optimal sizing of WT, PV and BES for efficient, reliable and economical operation. Several methods have been developed for optimal sizing of WT, PV and BES. In [4], an iterative techniqe is proposed based on imizing the difference between the generation and load. This proposed techniqe may lead to oversizing of PV, WT and BES. In [5], optimal size is detered considering mlti-objectives, i.e., high spply reliability and imization of cost. The major problem in this method is that it sets the imm limit for PV and WT to spply the average demand dring day and night times respectively. This assmption may lead to incorrect optimal soltion. Moreover complementary characteristics of wind and solar are considered which is not the case for all the geographical locations is. This limits the proposed method to the areas, which have complementary characteristics. In [6], optimal sizing is done considering the reliability and cost objectives. Particle swarm optimization (PSO) is sed to find the optimal sizes of grid connected PV, WT and BES considering reliability cost and emissions objectives [7]. In this paper, an innovative optimization methodology based on advanced constraint-based iterative search algorithm is proposed for the capacity optimization of solar PV, WT, and BES in a standalone MG system. The proposed algorithm generates a search space consisting of different combinations of PV, WT and BES. The optimal soltion is then obtained from the given search space on RE&PQJ, Vol., No.5, April 27

2 the basis of reliability and cost ratio. The proposed method is robst as so there is no over-sizing and ndersizing problem. Moreover the techniqe is general and can be applied to other types of generations and storages. The remainder of the paper is organized as follows. Section 2 discsses abot the strctre and mathematical model of standalone MG. Section 3 describes the proposed methodology. Section 4 presents the reslts while conclsion is given in Section Micro-grid A MG is a small scale power station that tilizes locally available resorces to flfil the reqired load which can be operated along with main grid or in standalone mode. In this stdy standalone MG is considered which is shown in Fig.. PVs and WTs are employed for hybrid power generation while for storage battery technology is tilized. PV and WT and BES are connected to DC bs via DC/DC and AC/DC and DC/DC converters respectively. The DC bs is connected to load bs throgh DC/AC converter and a transformer. and its efficiency. In this stdy a linear relation between the solar irradiation and otpt power of PV is considered. C. Battery Energy Storage System The BES system is composed of series and parallel connected battery banks. The normalized energy of BES can be calclated as E B (t) = E B (t ) + η B {E G (t) E L (t)} (2) ( DOD) E B (t) (3) where η B is the charging/discharging efficiency of BES, DOD is the depth of discharge, and E G, E B and E L are energy spplied generating sorces, energy spplied by BES and energy demand respectively. 3. Proposed Methodology The power generated by a MG system tilizing PVs and WTs can calclated as: i N PV j N WT P (i,j) G (t) = i P PV (t)pv stats (t) + j P WT (t)wt stats (t) (4) N i PV N PV N PV max N j max WT N WT N WT A. Wind System Fig.. A typical stand-alone MG system. A WT transforms the kinetic energy of moving wind to electrical energy. The power otpt of a WT can be calclated sing the following eqation [8]. v < v ci P rated v v ci v P WT (v) = v r v ci v < v r ci () P rated v r v < v co { v v co where v r, v ci, and v co are rated, ct-in and ct-ot wind speeds respectively. The WT generates no power below the ct-in speed, and between ct-in and ct-ot speeds power otpt increases with the speed. The WT generates P rated between rated and ct-ot speeds, and no power beyond ct-ot speed. B. Solar PV System A solar PV system converts solar energy to electrical energy. The power otpt of PV system depends on solar irradiation, area of the PV array, atmospheric temperatre, where N WT and N PV are nmber of WTs and PVs respectively, P WT is the otpt power of one WT and P PV is the otpt power of one PV, WT stats and PV stats are the stats of WT and PV which are calclated based on FORs and N WT, N max max PV, N WT and N PV are imm and maximm nmber of WTs and PVs. The instantaneos error p(t) is the difference between the generation and demand at any instant of time can be calclated sing the following eqation p (i,j) (t) = P L (t) P G (i,j) (t) (5) where P L is the system demand. The total error P can be calclated by smg the absolte vales p. The cmlative error matrix P contains P for all possible combinations of N WT and N PV and N WT and N PV are (i ) and ( j) vectors that contain all vales nmber of WTs and PVs that are considered in this work. A search space can be formed sing P and N PV and N WT. This search space can be redced by applying the constraint, P P. The redced search space is given as RS space = [ P N PV N T WT ] (6) j where RS space is the redced search space, P, is a ( j) vector that contain the cmlative errors that satisfy the constraint and N WT and N PV contain the corresponding vales of N WT and N PV. The power RE&PQJ, Vol., No.5, April 27

3 generated by case chosen from RS space is P G (t) which can be calclated as P G (t) = N PV P PV (t) + N WT P WT (t) (7) The gap in load and generation at any instant of time is given as p gap (t) = P L (t) P G (t) (8) The P gap is a vector whose entries are non-negative nmbers and each entry is the smmation of all the previos entries. It is calclated sing (8) P gap P gap r (r) = p gap (t) t= (r) = { P gap (r) P gap (r) > else (9) () The maximm BES capacity that can be installed to store the maximm of srpls energy can be calclated by sing the following expression B max = max(p gap ) () where B max is the maximm BES capacity for the case. The reqired BES capacity is less than or eqal to B max is calclated as B cap = { B max x = CBS x = (2) where B cap is the reqired BES capacity and it is eqal to B max if the battery discharges flly to its imm allowable point after its complete charging, this condition is indicated by x =. Otherwise B cap will be eqal to corrected battery size (CBS). The CBS is detered sing the region redction algorithm. Finally the optimal capacity of BES is detered based on the tilization factor. B optimal = { B cap UF BU limit BCS (3) else where B optimal is the optimal BES size. If UF is less than BU limit, B optimal will be eqal to battery corrected size (BCS ). The BCS is calclated sing an iterative algorithm. Reliability and cost can be sed to assess the performance of any system. In this stdy a ratio of energy served and cost is sed to find the optimal soltion. The total cost C t can be detered as C t = C c + C m (4) where C m and C c are maintenance and capital costs respectively. The cost of generation can be calclated by sing the following eqation C g = N i= N i= NPV i NDES i (5) where C g and N is the generation cost and the total nmber of renewable sorces respectively. The optimal decision variable (ODV) can be calclated as ODV = E s (6) C nit where C nit and E s are cost per nit and the energy served respectively. The soltion vector (SV) is a row vector given as SV = [ODV ODV j ODV j ] (7) The optimal WT, PV and BES combination corresponds to the index of maximm vale of SV. 4. Reslts and Discssions The proposed methodology is tested sing residential demand and renewables data of Dammam city which is sitated in the Eastern province of Kingdom of Sadi Arabia. Time series data of both renewables and demand for a length one year and with a resoltion of one hor is sed. The demand data is normalized to a peak of MW. The normalized average daily demand is shown in Fig. 2. The imm daily load demand appears arond 6 am to 8 am which increases after that and daily peak demand occrs at arond 8 pm. The normalized daily average wind power and solar are shown in Fig. 3 and Fig. 4 respectively. It can be observed from that solar gives power dring the day time while wind power keeps on flctating throghot the day. Normalized average demand Normalized average wind power Fig. 2. Normalized daily average demand Fig. 3. Normalized average wind power profile RE&PQJ, Vol., No.5, April 27

4 Normalized average solar power Case nmber Storage capacity (MWh) Fig. 4. Normalized average solar PV power profile PV capacity (MW) Fig. 5. capacities combinations wind and solar X: 8.3 Y: 63 Y: Case nmber Fig. 6. Optimal storage size. In this stdy the RS space consists of 8 cases. The capacities for WT and PV corresponding to each case are shown in Fig. 5. The optimal BES capacity for each combination of WT and PV is presented in Fig. 6. It is clear from Fig. 5 and Fig. 6 that the cases for which overall installed capacity is small and very large have small size of energy storage. The performance of any system can be analyzed sing its cost and reliability. In this stdy an economic and reliability analysis of the cases selected for the RS space is carried ot in order to find the optimal soltion. The optimal soltion is selected on the basis of energy served to cost ratio. A high vale of ratio means higher reliability at relatively lower cost. The case which has the maximm ratio will be the optimal soltion provided that nserved energy mst be approximately zero. The energy served to cost ratio is shown in Fig. 7 and energy served and energy nserved are shown in Fig. 8. It can be observed from the Fig. 7 and Fig. 8 that the local maximm appears for the case 3 bt it cannot be selected as an optimal soltion becase the corresponding energy nserved is very large. The global maximm appears for the case nmber 63 and Wind capacity (MW) the nserved energy for this case is almost zero which is depicted in the Fig. 7 and Fig. 8. The demanded energy for the whole year is sccessflly served for the optimal case. The WT, PV and BES capacities corresponding to the optimal case can be obtained from Fig. 6 and Fig. 7. Energy served to cost (/$) Energy (GWh) Cost ($/kwh) X: 3 Y: 82.2 Y: Case nmber Fig. 7. Energy served to cost ratio. Energy served Energy nserved Y: 47.8 Y: Case nmber Fig.8. Energy served and energy nserved comparison LOLP Y: Case nmber Fig.9. System estimated Cost per nit Y: Case nmber Fig.. LOLP for different cases RE&PQJ, Vol., No.5, April 27

5 The cost per nit and LOLP for different cases are shown in Fig. 9 and Fig.. It is evident from these figres that the cost is reasonable and LOLP is almost zero for the optimal soltion. It can be conclded from the Fig. 6 and Fig. that cost is highly dependent pon the storage size. The tilization factor of BES for the cases is shown in Fig.. It can be seen that tilization factor is greater than 8% for all the cases as per constraints. Utilization factor Case nmber Fig.. Utilization factor of BES for different cases. 5. Conclsion In this paper, a methodology is proposed for the optimal sizing of solar, wind and BES in a stand-alone MG system. The proposed techniqe avoids nder and over sizing as it searches every possible soltion. Moreover, to make the techniqe more realistic, FORs of PV and WT and tilization factor of BES is also considered. The proposed techniqe is validated sing solar, wind and load data of Dammam. Reliability and cost analysis are performed and compared for different cases. The optimal soltion is detered on the basis of reliability to cost ratio. LOLP and energy nserved are almost zero and the cost per nit is reasonable for optimal soltion. Acknowledgement We thank the spport provided by Research Institte at King Fahd University of Petrolem and Minerals and Sadi Electricity Company for providing time series data. References Y:.9943 [] C. Li, X. Li, Y. Cao, P. Zhang, H. Shi, L. Ren, and Y. Kang, A time-scale adaptive dispatch method for renewable energy power spply systems on Islands, IEEE Transactions on Smart Grid, vol. 7, no. 2, pp , 26. [2] T. Ma, H. Yang, and L. L, A feasibility stdy of a standalone hybrid solar wind battery system for a remote Island, Applied Energy, vol. 2, pp , 24. [3] Y. Zh, B. Li, and X. Sn, Freqency-based power management for PV/battery/fel cell stand-alone microgrid, in 2nd InternationalFtre Energy Electronics Conference (IFEEC). IEEE, 25, pp. 6. [4] B. Lynch, A. Kiger, and K. Hatipogl, An overview of hydrogen electrical energy storage meeting energy demand with intermittent renewable resorces, in IEEE SotheastConf, 26. IEEE, 26, pp. 6. [5] B. Zho and T. Littler, Local storage meets local demand: a technical soltion to ftre power distribtion system, IET Generation, Transmission & Distribtion, vol., no. 3, pp. 74 7, 26. [6] P. Denholm and R. M. Margolis, Evalating the limits of solar photovoltaics (pv) in electric power systems tilizing energy storage and other enabling technologies, Energy Policy, vol. 35, no. 9, pp , 27. [7] B. Yang, Y. Makarov, J. Desteese, V. Viswanathan, P. Nyeng, B. McMans, and J. Pease, On the se of energy storage technologies for reglation services in electric power systems with significant penetration of wind energy, in 5th International Conference on the Eropean Electricity Market. IEEE, 28, pp. 6. [8] A. Khatamianfar, M. Khalid, A. V. Savkin, and V. G. Agelidis, Improving wind farm dispatch in the Astralian electricity market with battery energy storage sing model predictive control, IEEE Transactions on Sstainable Energy, vol. 4, no. 3, pp , 23. [9] M. Khalid and A. V. Savkin, A model predictive control approach to the problem of wind power smoothing with controlled battery storage, Renewable Energy, vol. 35, no. 7, pp , 2. [] S. Shafiq and A. T. Al-Awami, Reliability and economic assessment of renewable micro-grid with V2G electric vehicles coordination, in Jordan Conference on Applied Electrical Engineering and Compting Technologies (AEECT). IEEE, 25, pp. 6. [] A. V. Savkin, M. Khalid, V. G. Agelidis, A Constrained monotonic charging/discharging strategy for optimal capacity of battery energy storage spporting wind farms, IEEE Transactions on Sstainable Energy, vol. 7, no. 3, pp , 26. [2] M. Khalid, A. Ahmadi, A. V. Savkin, and V. G. Agelidis, Minimizing the energy cost for microgrids integrated with renewable energy resorces and conventional generation sing controlled battery energy storage', Renewable Energy, vol. 97, pp , 26. [3] Z. Benhachani, B. Azoi, R. Abdessemed, and M. Chabane, Optimal sizing of a solar-wind hybrid system spplying a farm in a semi-arid region of algeria, in 47th International Universities Power Engineering Conference (UPEC). IEEE, 22, pp. 6. [4] W. Kellogg, M. Nehrir, G. Venkataramanan, and V. Gerez, Generation nit sizing and cost analysis for stand-alone wind, photovoltaic, and hybrid wind/pv systems, IEEE Transactions on Energy Conversion, vol. 3, no., pp. 7 75, 998. [5] L. X, X. Ran, C. Mao, B. Zhang, and Y. Lo, An improved optimal sizing method for wind-solar-battery hybrid power system, IEEE Transactions on Sstainable Energy, vol. 4, no. 3, pp , 23. [6] H. Yang, L. L, and W. Zho, A novel optimization sizing model for hybrid solar-wind power generation system, Solar Energy, vol. 8, no., pp , 27. [7] L. Wang and C. Singh, Pso-based mlti-criteria optimm design of a grid-connected hybrid power system with mltiple renewable sorces of energy, in IEEE Swarm Intelligence Symposim. IEEE, 27, pp [8] S. Roy, Market constrained optimal planning for wind energy conversion systems over mltiple installation site, IEEE Transactions on Energy Conversion, vol. 7, no., pp , RE&PQJ, Vol., No.5, April 27