2011 3rd International Conference on Computer Modeling and Simulation (ICCMS 2011)

Novel Simulation Based Evolutionary Optimization Algorithm to Design Stand-alone Hybrid Energy Systems A.T.D.Perera D.M.I.J.Wickremasinghe D.V.S. Mahindarathna R.A.Attalage K.K.C.K.Perera E.M.Bartholameuz Abstract—This paper introduces a simplified novel method to optimize Hybrid Energy Systems (HES). In order to carry out this work, mathematical modeling of system components was performed in combination with system simulation. Based on the simulation Levelized Energy Cost (LEC) was optimized using Genetic Algorithm (GA). Further, impacts of GA parameters to optimized results were investigated. Finally, comparison of the novel method with existing enumerative-based software (Homer) was carried out, which illustrates that novel method is having a good capability to reduce fuel consumption in system operation.

I. INTRODUCTION Stand alone Hybrid Energy Systems (HESs) have become an area of interest to study due to its versatile applicability in producing electricity where extending existing electricity grid is not viable due to economic expenses. At the same time HESs produce less Green House Gasses (GHGs) [1], consume lesser amount of fuel and cost effective compared to Internal Combustion Generators (ICGs). Due to the above mentioned advantages a number of researches have focused on different configurations of HESs [2],[3]. In the theoretical investigation on HESs, mathematical modeling of renewable energy sources is the first step. The present states of HES modeling is reviewed by Deshmukh et al [4]. Mathematical modeling is followed by HES simulation, which comprise of HES control strategy, which depends upon the system components and other auxiliary information such as meteorological data and Electricity Load Demand (ELD). In most of the studies, system simulation had been carried out considering a year of operation. Several methods such as graphical [5],[6] numerical [7], enumerative and heuristic [8],[9] had been proposed for the optimization due to complexity of decision space variables and objective functions. Several software were also been implemented in order to handle the difficulties of optimization [10].

Among those existing methods, algorithm proposed HOMER [11] and HOGA [8] becomes significant due to its capability to optimize both system configuration and operation strategy. In HOMER, analytical method was used based on the dispatch strategy introduced by Barley et al [12], which has the limitation of considering opportunity cost of battery storage with the chaotic variation of renewable energy sources. In order to take this into account HOGA introduces a heuristic based method to optimize system operation strategy, considering the limits obtained analytically for critical load and dispatch load, producing a complex combined algorithm. The main objective of this paper is to present a simplified optimizing scheme to optimize system configuration and operation strategy using evolutionary techniques based on system simulation II. MODELING SYSTEM COMPONENTS Mathematical modeling was carried out in this study for a HES consisting of wind turbines, Solar PV (SPV) panels, ICG, battery bank, AC-DC converters, DC-AC converters and a battery charger (Fig. 1). Hourly global irradiation on a horizontal plane was used to calculate hourly tilted global irradiation using Climed-2 model [13] and Klucher [14] model for anisotropic effects. Semi empirical model proposed by Durisch et al. [15] was used to calculate the efficiency of SPV modules. Hourly wind speed throughout the year at anemometer height (10m) was taken to calculate wind speed at hub level using power law approximation. Wind turbine power was calculated using wind turbine model proposed by Chedid et al [16]. Rain Flow algorithm was used to calculate battery life and inverter efficiencies were taken as constant value (95%) assuming variation of the load factor as negligible.

A.T.D. Perera is with Department of Mechanical Engineering, University of Moratuwa, Sri Lanka. (e-mail: [email protected] ) D.M.I.J. Wickremasinghe was with Department of Mechanical Engineering, University of Moratuwa, Sri Lanka. (e-mail: [email protected] ) D.V.S. Mahindarathna is with Department of Mechanical Engineering, University of Moratuwa, Sri Lanka. (e-mail: [email protected] ) R.A.Attalage is a professor, Director Post Graduate Division - Faculty of Engineering and senior staff member of Department of Mechanical Engineering, University of Moratuwa, Sri Lanka. (e-mail: [email protected] Tel: 94-112650621) K.K.C.K. Perera is a professor and a senior staff member of Mechanical Engineering Department, University of Moratuwa, Sri Lanka. (e-mail: [email protected] ) E. M. Bartholameuz was with Department of Mechanical Engineering, University of Moratuwa, Sri Lanka. (e-mail: [email protected] ).

978-1-4244-9243-5/11/$26.00

C

2011 IEEE

V1-272

Fig. 1 System Configuration

2011 3rd International Conference on Computer Modeling and Simulation (ICCMS 2011)

III. SIMULATION HES simulation combines hourly varying meteorological data and ELD according to the operation strategy decided by the system designer known as dispatch strategy. Hourly average wind speed and solar irradiation data values of 1995, 1997 and 1998 at Hambanthota, southeast location of Sri Lanka (06°07′ N 81°07′ E) was taken in order to simulate the system. Hambanthota is an area having ample renewable energy potential, both wind and solar [17],[18]. It was assumed that the ELD varies according to summer - weekly load curve given by [19] which was scaled to 7.5 kW. Dispatch strategy of the system consists of five states (as shown in Fig 2) which includes frugal discharge strategy, State of Charge (SOC) strategy, set point strategy and load following strategy [12] based on load difference P, between renewable energy sources (PRE) and load (PL) according to (1). P = PL - PRE (1) In order to increase the battery life and ICG efficiency, minimum SOC (SOCmin) value and minimum ICG power (Pmin) values were also optimized along with critical load (Pcri), dispatch load (Pd) and SOCset point values in the optimization algorithm [8]. Further Unmet load was calculated using (2) as measures of consistency of supplying ELD which was set as a constraint in the optimization

fraction. LEC of the system consists of acquisition cost, installation cast, operation and maintenance cost. Acquisition cost consists of capital required to purchase system components such as wind turbines, SPV panels, ICG etc. Installation cost considers the capital required to install the system components. Finally, operation and maintenance cost was calculated based on the system simulation, which includes maintenance cost of system components such as wind turbines, SPV panels and ICG, components replacement cost for battery bank and ICG and fuel cost. Integer vector was used to insert variables into the optimization algorithm as shown in Fig. 3. Operating range of the ICG from nominal load (Pngen) to Pmin was divided into Nd equal parts and the appropriate dispatch load Xd (Fig. 4) was optimized using GA instead of using analytical equations [12]. In order to calculate the critical load, range between dispatch load and Pngen is divided in to Nc equal parts and critical load Xc (Fig 4) was also optimized in a similar way. All the parameters in the integer vector including SOCmin, Pmin and set point values were optimized within the range given in Table 1. Finally, optimization algorithm was combined with system modeling and simulation as shown in Fig. 5. Wind SOC SPV SPV Wind No of Turbine SOCmin Pmin set Type Panels Turbines batteries Power Value

Xd Xc

Fig. 3 Integer vector used to insert variables

Unmet Load Fraction = Σ Unmet ELD/ Σ Hourly ELD (2)

Fig 4: selection of Xd and Xc

Fig. 2 Dispatch strategy

IV. OPTIMIZATION Constrained optimization was carried out using Genetic Algorithm (GA) with real coded crossover and mutation to optimize Levelized Energy Cost (LEC) without unmet load

Fig. 5 Optimization algorithm

In order to fine tune parameters relevant to optimization such as crossover rate, mutation rate and population size,

V1-273

2011 3rd International Conference on Computer Modeling and Simulation (ICCMS 2011)

system optimization was carried out by varying those parameters according to the Table 2. Obtained results from the optimization (Fig. 6-8) indicate that except with one case novel algorithm is capable to approach the optimum results within around 200 generations.

V. RESULTS AND DISCUSSION Obtained configurations (Table 3) from the novel method are slightly different from the ones obtained using Homer, even though their LEC values are almost same. Variation in the operation strategy, system component modeling (battery

TABLE 1 LIMITS OF VARIABLES Variable

Minimum

Maximum

Interval

SPV Type

0

3

1

SPV Panels

0

220

1

Wind Turbine Power 0

2

1

Wind Turbines

0

20

1

No of batteries

0

100

4

SOC min

30%

50%

2.5%

Pmin

30%

50%

2.5%

SOC set Value Nd

70% 0

100% 7

3.75% 1

Nc

0

7

1

TABLE 2 PARAMETER SETTINGS FOR GA Varying Set Variable 1 Set variable 2 Variable Population (40) Mutation (0.3)

Crossover

Range 0.7~0.95

Cross over (0.9) Population

Mutation

0.1~0.4

Cross over

Population

20~100

Mutation

Fig. 7 Impact of mutation for results

Fig 8: Impact of mutation for results

bank, SPV performance) and concern on inverter efficiency in the novel method are some of the reason for this variation. Most importantly, the optimized operation strategy of the novel method consumes lesser amount of fuel while increasing the renewable energy component in the total load supplied which was achieved through evolutionary learning obtained from system simulation. VI. CONCLUSION In this study a simplified novel method was introduced to model and optimize HESs based on simulation using evolutionary method. It was shown that this method is having a good capability to approach towards the optimum solution within the range selected for GA parameters in this study. Finally obtained results from the optimization shows that novel method is capable of reducing ICG fuel consumption while managing renewable energy sources in an optimum manner producing a much eco friendly sustainable HES. ACKNOWLEDGMENT

Fig 6: Impact of mutation for results

Authors would like to acknowledge Prof K. Deb IIT, Kanpur for his help on multi objective optimization, Prof Ajith de Alwis University of Moratuwa, Ms Anusha

V1-274

2011 3rd International Conference on Computer Modeling and Simulation (ICCMS 2011)

Wijewardana University Loughborough and Mr Indika Perera IUPUI for their help in various other ways. TABLE 3 COMPARISON OF HOMER AND NOVEL METHOD

Homer

ICG capacity (kVA) 2.5

LEC ($)

SPV panels

SPV energy (kWh/yr)

Wind turbines

Wind energy (kWh/yr)

Fuel con. (l/yr)

Battery Bank size

0.56

84

27799

1

17571

4390

92 72

Novel

2.5

0.57

101

39180

1

17469

4047

Homer

5.5

0.33

1

331

1

17571

13254

16

Novel

5.5

0.32

0

0

2

34938

6947

44

REFERENCES [1]

[2] [3] [4] [5] [6]

[7] [8] [9] [10]

[11]

[12] [13] [14] [15]

[16]

[17]

G. Giatrakos, T. Tsoutsos, P. Mouchtaropoulos, G. Naxakis, and G. Stavrakakis, “Sustainable energy planning based on a stand-alone hybrid renewable energy/hydrogen power system: Application in Karpathos island, Greece,” Renewable Energy, 2009, pp. 1-9. J.L.B. Agustín and R.D. Lo´ pez, “Simulation and optimization of stand-alone hybrid renewable energy systems,” Renewable and Sustainable Energy Reviews, vol. 13, 2009, pp. 2111-2118. W. Zhou, C. Lou, Z. Li, L. Lu, and H. Yang, “Current status of research on optimum sizing of stand-alone hybrid solar–wind power generation systems,” Applied Energy, vol. 87, 2010, pp. 380-389. M. Deshmukh and S. Deshmukh, “Modeling of hybrid renewable energy systems,” Renewable and Sustainable Energy Reviews, vol. 12, 2008, pp. 235-249. S. Diafa, D. Diafb, M. Belhamelb, M. Haddadic, and A. Louchea, “A methodology for optimal sizing of autonomous hybrid PV/wind system,” Energy Policy, vol. 35, Apr. 2010, pp. 5708-5718. M. Muselli, G. Notton, P. Poggi, and A. Louche, “PV-hybrid power systems sizing incorporating Battery storage: an analysis via simulation Calculations,” Renewable Energy, vol. 20, Apr. 2010, pp. 1-7. W. Kellogg, M. Nehrir, G. Venkataramanan, and V. Gerez, “Optimal unit sizing for a hybrid wind/photovoltaic generating system,” Electric Power Systems Research, vol. 39, Apr. 2010, pp. 35-38. R.D. Lo´pez and J.L.B. Agustı´n, “Design and control strategies of PV-Diesel systems Using genetic algorithms,” Solar Energy, vol. 79, 2005, pp. 33-46. G.C.G.C.S. Hochmuth, “A combined optimisation concept for the design and Operation strategy of hybrid-pv energy systems+,” Solar Energy, vol. 61, 1997, pp. 77-87. D. Connolly, H. Lund, B. Mathiesen, and M. Leahy, “A review of computer tools for analysing the integration of renewable energy into various energy systems,” Applied Energy, vol. 87, 2010, pp. 1059-1082. T. Givler and P. Lilienthal, Using HOMER® software, NREL’s micropower optimization model, to explore the role of gen-sets in small solar power systems case study: Sri Lanka, National Renewable Energy Laboratory, 2005. C. Barley and C. Winn, “Optimal dispatch strategy in remote hybrid power systems,” Solar Energy, vol. 58, 1996, pp. 165-179. A.D. Miguel, J. Bilbao, H. .Kambezidis, and E. Negro, “Diffuse solar irradiation model in the north Mediterranean Belt area,” Solar Energy, vol. 70, 2001, pp. 143-153. T. Klucher, “Evaluation of models to predict insolation on tilted surfaces,” Solar Energy, vol. 23, 1979, pp. 111-114. W. Durisch, B. Bitnar, J. Mayer, H. Kiess, K. Lam, and J. Close, “Efficiency model for photovoltaic modules and demonstration of its application to energy yield estimation,” Solar Energy Materials & Solar Cells, vol. 91, 2007, pp. 79-84. R. Chedid, H. Akiki, and R. Saifer, “A decision support technique for the design of hybrid solar-wind power systems,” IEEE Transactions

[18] [19]

V1-275

on Energy Conversion, vol. 13, 1998, pp. 76-83. M. Young and R. Vilhauer, Sri Lanka wind farm analysis and site selection assistance, National Renewable Energy Laboratory, 2003. D. Renné, R. George, B. Marion, and D. Heimiller, Solar resource assessment for Sri Lanka and the Maldives, National Renewable Energy Laboratory, 2003. “IEEE reliability test system: A report prepared by the reliability test system task force of the application of probability methods subcommittee,” IEEE Transactions on Power Apparatus and Systems, vol. PAS-98, 1979, pp. 2047 - 2054.