Optimal Allocation and Sizing of Distributed ...

12 downloads 7985 Views 211KB Size Report
Department of Electrical Engineering, Setif1 University, Email: ... placement and sizing problem in the radial distribution system. Firstly a brief review about.
Revue des Sciences et de la Technologie - RST-

Volume 6 N°1 / janvier 2015

Optimal Allocation and Sizing of Distributed Generation with Particle Swarm Optimization Algorithm for Loss Reduction K. Ras Guerriche * and T. Bouktir** *Department of Electrical Engineering, Setif1 University, Email: [email protected] ** Department of Electrical Engineering, Setif1 University, Email: [email protected] Abstract— an optimal placement and sizing of distributed generation units in distribution system is usually done for the purpose of loss reduction and voltage profile improvement. Research work included by this paper focuses on using an optimization methodology for identifying proper location and size of DG units. Solution’s particle swarm optimization algorithm for DG units placement and sizing has been developed in terms of parameter selection, to obtain the maximum loss reduction and voltage profile improvement. The performance of this proposed methodology is tested on IEEE 33-Bus distribution system. Results show the efficiency and robustness of the proposed PSO algorithm to finding the very near optimal location and sizing of the DG units for enhancing the loadability of radial distribution system. Keywords – Particle swarm optimization (PSO), Optimal size, Optimal location, Distributed generation (DG), Active power losses.

I.

INTRODUCTION

The need to provide acceptable power quality and reliability will create a very favorable climate for the penetration of renewable and nonrenewable distributed generation resources around the world. As a result of this penetration of DG resources at the distribution level, distribution system are no longer passive supplying loads, but are active with power flows and voltages determined by the generation and loads. A number of steps should be followed concerning, one of this steps is the best use of existing distribution network through the optimal allocation and sizing of the DG resources. In order to achieve the desired performance in DG resources and minimizing power loss, improve the voltage profile, increase reliability and improving the power quality parameters of the electric grid, suitable placement and size need to provide for this DG units. There are two methods for sitting and sizing of DG in the distribution network. The first method is traditional based such as optimal power flow (OPF), sensitive factor and repetitive load flows (reload flow). In the second method, the artificial intelligent (AI) is used to apply with DG placement and sizing like Ant Colony Algorithm (ACO), Genetic Algorithm (GA), Tabu Search (TS), Differential Evolution (DE) and Particle Swarm Optimization (PSO). A lot of papers and studies have been carried out in the recent years to present methodologies in the general topic of DG units placement and sizing [1- 6]. A particle swarm optimization (PSO) algorithm was introduced to determine the optimum size and location of a single DG unit to minimize the real power losses of the system in [7]. In [8], a novel optimization approach that employs an Artificial Bee Colony (ABC) algorithm to find the optimum DG size, power factor, and location in order to minimize the total system real power loss. A genetic algorithm (GA)-based technique along with optimal power ow (OPF) calculations were used to determine the optimum size and location of DG units installed to the system in order to minimize the cost of active and reactive power generation is used in [9]. In [10], an analytical method to determine the optimum location–size pair of a DG unit was proposed in order to minimize only the line losses of the power system. In [11], a combined PSO and GA algorithm was used to nd the optimal location of a xed number of DG units with speci c

59

K. Ras Guerriche , T. Bouktir

total capacity such that the real power loss of the system is minimized and the operational constraints of the system are satis ed. [12] Present a genetic algorithm (GA) approach for the location and sizing of DG in three different loading conditions which were peak load, medium load and low load. In [13], A novel optimization approach that employs an Artificial Bee Colony (ABC) algorithm to find the optimum DG size, power factor, and location in order to minimize the total system real power loss. In [14], a sensitivity analysis of power losses in terms of DG size, location and operating point has been performed to nd the optimal size and location of DG units. [15], a PSO algorithm was used to place multiple DG units with nonunity power factor but the objective was to minimize only the real power loss of the system. Pavlos, S in [16]. Present a review serves as a guide to aid researchers and power system engineers on the available DG placement models and methodologies. In this paper a particle swarm optimization algorithm is presented, to solve DG units optimal placement and sizing problem in the radial distribution system. Firstly a brief review about distributed generation technology. Section II explains the proposed approach. A problem formulation in Section III. The results and discussion are presented in Section IV. And finally the conclusion is given in Section V. Definition of Distributed Generation (DG): As known, distributed generation signify the electric power generation within distributed network to meet the rapid energy demand of consumers. However, There is many terms and definitions used for explain DG and that’s create a various perspectives: The Electric Power Research Institute (EPRI) de nes distributed generation as generation from ‘a few kilo-watts up to 50 MW’ [17]. International Energy Agency (IEA) de nes distributed generation as generating plant serving a customer on-sit or providing support to a distribution network, connected to the grid at distributed level voltages [18]. The International Conference on large High Voltage Electric Systems (CIGRE) de nes DG as ‘smaller than 50-100 MW’ [17]. Although there are variations in de nitions, however, the concept is almost same. DG can be treated as small scale power generation to mitigate the consumer energy demand. Distributed Generation can come from a variety of sources and technology. Here, we will consider the Distributed Generation as an Electric power source connected directly to the distribution system.

II.

PARTICLE SWARM OPTIMIZATION:

One of the most recent metaheuristic algorithms is the Particle Swarm Optimization (PSO) is a population based stochastic optimization technology [19,20] by Dr. Eberhart and Dr. Kennedy in 1995, inspired by social behavior of bird flocking and fish schooling. It is used for optimization of continuous nonlinear functions [21, 22]. PSO is a swarm intelligence algorithm, inspired by the social dynamics and an emergent behavior that arises in socially organized colonies. PSO algorithm exploits a population of individuals to probe promising regions of the search space. In this context, the population is called swarm and the individuals are called particles or agents. In conventional PSO, particles change their positions (states) with time. Let 'u' and 'v' denote a particle coordinates (position) and its corresponding flight speed (velocity) in a search space respectively. The position vector ui and the velocity vector vi of the i th particle in the ndimensional search space can be represented as

Optimal Allocation and Sizing of Distributed Generation with Particle Swarm Optimization Algorithm for Loss Reduction

60

K. Ras Guerriche, T. Bouktir

ui

(ui1 , ui 2 ,.....,uin )

And

vi

(vi1 , vi 2 ,......,vin )

respectively. The best previous position of the i the particle is recorded and represented as

Pbest

(ui1pbest , uipbest ,....., uinpbest ) . 2

The index of the best particle among all the particles in the group is represented by :

(u1Gbest , u2Gbest ,.....,unGbest ) .

Gbest

The modified velocity and position of each particle can be calculated as per following formulas [23]:

vi( k

1)

vi( k )

c1r1 ( Pbesti( k ) ui( k ) )

Pr evious Velocity

Cognitive Component

c2 r2 (Gbest

(k )

ui(k ) )

(01)

Social Component

ui( k

1)

ui( k ) vi( k

1)

(02)

Where , c1, c2 0, k is the iteration number. : is the inertia weight factor. c1 and c2 : are the acceleration coefficients. rl and r2:are two random numbers within the range [0,1].

vi(k ) , ui(k ) : are the velocity and the current position of particle i in the search space at iteration k, respectively. In general, the inertia weight provides a balance between global and local explorations (control the influence of the previous history of the velocities on the current one). It is set according to the following equation:

(

max

max

min,

max

min

)

kmax

k

(03)

: initial and final inertia factor weights.

kmax : maximum iteration number. k: current iteration number.

K. Ras Guerriche , T. Bouktir

61

K. Ras Guerriche , T. Bouktir

The constants c1 and c2 pull each particle toward Pbest positions (cognitive component of velocity) and Gbest positions (social component of velocity). The position is updated with respect to (02). Time Varying Acceleration Coefficients The time-varying inertia weight (TVIW) can locate a good solution at a significantly faster rate but its ability to fine tune the optimum solution is weak, due to the lack of diversity at the end of the search. It has been observed by most researchers that in PSO, problem-based tuning of parameters is a key factor to find the optimal solution accurately and efficiently [24, 25]. In TVAC, this is achieved by changing the acceleration coefficients and with time in such a manner that the cognitive component is reduced while the social component is increased as the search proceeds. TVAC-PSO is successfully implemented for economic load dispatch (ELD) problem in [26].

vi( k

1)

(k ) (k ) i

v

c1( k ) r1 ( Pbesti( k ) ui( k ) )

c2( k ) r2 ( Pgbesti( k ) ui( k ) ) c1( k ) (k ) 2

c

(c1i c1 f ) (c2 f

c2i )

k kmax k kmax

(04)

c1i (05)

c2i

Where:

ci(k ) is the ith acceleration coefficient at iteration k.

c ii and c if are initial and final values of the ith acceleration coefficient respectively. (k )

is the same value of inertia weight in (03).

III.

PROBLEM FORMULATION

The objective of this work is to minimize the active power loss in the radial distribution system as well as to improve the voltage profile of the system by solving the distributed generator placement and sizing problem. The complexity of this problem lead the researchers to introduce simple assumptions such as [27]: The section load is balanced and uniformly distributed, and the load current is constant. The power factor of the section load is 1.0. The candidates of DG installation position for every feeder section are given. The capacity of distributed generator must be selected from given capacity candidates (discrete values). One DG can be allocated for one candidate position. The maximum number of installable DGs is given. The total installation capacity of DGs is given.

Optimal Allocation and Sizing of Distributed Generation with Particle Swarm Optimization Algorithm for Loss Reduction

62

K. Ras Guerriche, T. Bouktir

Two main constraints are: The upper and lower limits of node voltages. The capacities of conductors. The objective function of our problem is to minimize the total active power loss and according to the previous assumptions the formulation of problem can be listed as fellow [28]:

Ri Ii2

fitness function Minimize

(06)

i

Subject to:

0.95 p.u. V j 1.05 p.u.

(07)

Ii

(08)

Ii max

In which:

Ri : is the ith branch resistance. Ii : is the current of the ith branch. I i max : is the maximum current of the ith branch. IV.

RESULTS AND DISCUSSION

To check the validity of the proposed PSO algorithm, the IEEE 33-Bus (Fig.1) radial distribution feeder system was considered in different cases. This network has a voltage of 12.66kV, load size of 3.715MW and 2.300 MVar and consists of 32 line and 33 buses. The size of the distributed generation used will vary automatically between the ranges of 0% to 30% of the total load until it reaches the optimal values; the DG voltage is 12.66 kV. That will be able to show the effects of the optimal placement and sizing of DGs units on the network parameters (voltage profile, active power losses). Therefore, with the application of the proposed approach (PSO) we calculate and compare the result with those obtained via other methods. Furthermore, we analyzed two scenarios, Scenario I and scenario II. For the first scenario we have three cases to studying. Case I (the reference case) is when the system without distributed generation unit, Case II is to determine the optimal size and location for a single DG unit and the Case III for two DG units simultaneously. Scenario II on the other hand, represents the situation where two load levels are applied the light load (50% from the full load) and heavy load (150% from the full load). The results of the feeder system due to above mentioned scenarios are shown below: The voltage profile before and after DG installation is shown in Fig.2. The lowest voltage occurred in the Bus 18 in the case I and II with the amount of 0.91 p.u and 0.97 p.u, for the case III the lowest voltage is in Bus 25 with the value of 0.98 p.u. Noticed an important improvement of voltage profile in case II and III compared to the reference case.

K. Ras Guerriche , T. Bouktir

63

K. Ras Guerriche , T. Bouktir

Fig. 1. 33 Bus Under study Radial Distributed System Table.1. shows the most appropriate location and size and corresponding real power for the scenario I, The results show that a significant real power loss reduction and voltage profile improvement after the installation of DG units in the system. Noticed that the installation of DG units can improve the system loadability.

TABLE I. SUMMARY OF RESULTS FOR THE SCENARIO I Initial Power losses (kW) Optimal DG Location Optimal DG size (MW)

Case I 202.66

Loss Reduction (%) Minimum Voltage (p.u)

0.91 (Bus 18)

Case II 202.66 06 2.43

Case III 202.66 13 , 30 0.82 , 1.16

69.52%

85.77%

0.97 (Bus 18)

0.98 (Bus 25)

Optimal Allocation and Sizing of Distributed Generation with Particle Swarm Optimization Algorithm for Loss Reduction

64

K. Ras Guerriche, T. Bouktir

Fig. 2

Voltage profile of the scenario I at the different cases

To evaluate and prove the effectiveness of the proposed PSO Algorithm we compare the results obtained by this technique using the IEEE 33-BUS system with those obtained in [29],[30]. The table.2 shows the DG units optimal location, sizing and the total power losses of the system. TABLE II.RESULTS AND COMPARISON Case II

Case III

Approach

Proposed PSO

[29]

Proposed PSO

[30]

Optimal DG location

6

6

13 , 30

6,8

Optimal DG Size (MW)

2.43

2.28

0.82 , 1.16

1.72, 0.84

Total power loss (kW)

61.77

63.26

28.83

55.67

The table.2 reveals that the proposed PSO algorithm could gain better results compared to [29],[30] approaches in real power loss reduction so the voltage profile improvement too, by reducing the real power losses from 202.6 kW to 61.77 kW and 28.83kW. TABLE III SUMMARY OF SCENARIO II RESULTS FOR SINGLE DG UNIT Light Load

Normal Load

Heavy Load

Initial power Losses (kW)

46.8

202.66

497.12

Optimal DG location

6

6

6

Optimal DG size (MW)

1.22

2.43

3.69

Total power losses (kW)

14.96

61.77

142.78

Further study to the proposed PSO when the system is under different loads levels. The light load (50% from the full load) and heavy load (150% from the full load) in order to see how the DGs Units can react with the load variation. Moreover the optimum size and location of DGs units for the voltage profile improvement and active power reduction using the proposed approach in the different load levels are illustrated below:

K. Ras Guerriche , T. Bouktir

65

K. Ras Guerriche , T. Bouktir

TABLE IV SUMMARY OF SCENARIO II

RESULTS FOR TWO DGS UNITS

Light Load

Normal Load

Heavy Load

Initial power Losses (kW)

46.8

202.66

497.12

Optimal DG location

13 , 30

13 , 30

13 , 30

Optimal DG size (MW)

0.41 , 0.56

0.82 , 1.16

1.25 , 1.69

Total power losses (kW)

7.03

28.83

65.36

From Tables.3 and 4, for the different levels loads the DGs size is change simultaneously with the load levels (preserve the same place of the DG units). Moreover the voltage profile of feeder system due to different load levels are varying in matching to the load levels as illustrated in Figure.3 and 4.

Fig. 3 Voltage profile of the scenario II for a single DG unit

Fig. 4 Voltage profile of the scenario II for tow DG units

Optimal Allocation and Sizing of Distributed Generation with Particle Swarm Optimization Algorithm for Loss Reduction

66

K. Ras Guerriche, T. Bouktir

V.

CONCLUSION

A particle swarm optimization approach is proposed in this paper to solve the DGs units allocation and sizing problem. Comparing the proposed approach to other optimization algorithms, showed an improved performance and better results. The proposed PSO was applied to the IEEE 33-Bus system radial distribution system. Based on the results the proposed algorithm has the capability to provide the optimal placements and sizes for the DGs units. Moreover, the results illustrate the efficiency of this approach for the voltage profile improvement and power losses reduction. In order to better evaluate the robustness of this approach we should be tested in larger distribution systems. ACKNOWLEDGMENT The author thanks Prof. Tarek Bouktir and all members of his SMART GRID team. REFERENCES [1] H. Hedayati, S. A. Nabaviniaki and A. Akbarimajd, “A Method for Placement of DG Units in Distribution Networks,” IEEE Transactions on Power Delivery, Vol. 23, No. 3, July 2008, pp.1620-1628. [2] Lalitha, M.P., Reddy, V.C.V., Usha, V.: “Optimal DG placement for minimum real power loss in radial distribution systems using PSO”, J. Theor. Appl. Inf. Technol., 2010, 13, (2), pp. 107–116. [3] Gandomkar, M., Vakilian, M., Ehsan, M.: “A combination of genetic algorithm and simulated annealing for optimal distributed DG allocation in distributed networks”. Proc. IEEE Electrical and Computer Engineering Canadian Conf., 2005, pp. 645–648. [4] H. Yoshida, K. Kawata and Y. Fukuyama, “A Particle Swarm Optimization for Reactive Power and Voltage Control Considering Voltage Security Assessment,” IEEE Transactions on Power System, Vol. 15, No. 4, 2000, pp.1232-1239. [5] Zhu, D., Broadwater, R.P., Tam, K., Seguin, R., Asgeirsson, H.: “Impact of DG placement on reliability and ef ciency with time-varying loads”, IEEE Trans. Power Syst., 2006, 21, (1), pp. 419–427. [6] Keane, A., O’Malley, M.: “Optimal distributed generation plant mix with novel loss adjustment factors”. IEEE Power Engineering Society General Meeting, 2006. [7] Y. A. Katsigiannis and P. S. Georgilakis, “Optimal Sizing of Small Isolated Hybrid Power Systems Using Tabu Search,” Journal of Optoelectronics and Advanced Materials, Vol. 10, No. 5, May 2008, pp. 1241-1245. [8] T. Niknam, S. I. Taheri, J. Aghaei, S. Tabatabaei, M. Nayeripour, “A modified honey bee mating optimization algorithm for multiobjective placement of renewable energy resources”, Applied Energy, vol. 88, pp. 4817–4830, 2011. [9] Mardaneh, M., Gharehpetian, G.B.: “Siting and sizing of DG units using GA and OPF based technique”. IEEE Region 10 Conf., 2004, vol. 3, pp. 331–334. [10] Gozel, T., Hocaoglu, M.H.: “An analytical method for the sizing and siting of distributed generators in radial systems”, Int. J. Electr. Power Syst. Res., 2009, 79, pp. 912–918. [11] Safari, A., Jahani, R., Shayanfar, H.A., Olamaei, J.: “Optimal DG allocation in distribution network”, Int. J. Electr. Electron. Eng., 2010, 4, (8), pp. 550–553

K. Ras Guerriche , T. Bouktir

67

K. Ras Guerriche , T. Bouktir

[12] D. Singh, D. Singh, and K. S. Verma, “GA based energy loss minimization approach for optimal sizing and placement of distributed generation”, International Journal of Knowledge-based and Intelligent Engineering Systems vol. 12, pp. 147-156 2008. [13] F. S. Abu-Mouti, M. E. El-Hawary, “Optimal Distributed Generation Allocation and Sizing in Distribution Systems via Artificial Bee Colony Algorithm”, IEEE Trans. on Power Delivery, vol. 26, 2011. [14] Kashem, M.A., Le, D.T., Negnevitsky, M., Ledwich, G.: “Distributed generation for minimization of power losses in distribution systems”. IEEE Power Engineering Society General Meeting, 2006. [15] AlHajri, M.: “Sizing and placement of distributed generation in electrical distribution systems using conventional and heuristic optimization methods”. Ph.D. thesis, Dalhousie University, Halifax, Nova Scotia, 2009. [16] P. S. Georgilakis, S. Member, and N. D. Hatziargyriou, “Optimal Distributed Generation Placement in Power Distribution Networks : Models, Methods, and Future Research,” IEEE Transactions, vol. 28, no. 3, pp. 3420-3428, August 2013. [17] T. Ackermann, G. Andersson, and L. Sder, “Distributed generation: a de nition,” Electric Power Systems Research, vol. 57, no. 3, pp. 195-204, 2001. [18] “IEA publication, distributed generation in liberalized electricity market, 2002. (page19).” [Online]. Available: http://www.iea.org/textbase/nppdf/free/2000/distributed2002.pdf [19] M. A. Abido, “Optimal power flow using particle swarm optimization” Electrical power and energy systems” 24(2002)563- 571- 1985. [20] J. Kennedy and R. Eberhart, “A Particle Swarm Optimization” Proceedings of IEEE International conference on Neural Networks, vol.IV, pp.1942- 1948, Perth, Australia, 1995. [21] Ioan Cristian TRELEA, « L’essaim de particule vu comme un système dynamique : convergence et choix des paramètres », Séminaire « L’optimisation par essaim de particules», Paris, 2003. [22] Maurice Clerc et Patrick Siarry, « Une nouvelle métaheuristique pour l’optimisation difficile : la méthode des essaims particulaires », Séminaire « L’optimisation par essaim de particules», Paris, 2003. [23] Salhi A, Bouktir T, Naimi D. “Economic dispatch resolution using adaptive acceleration coefficients based PSO considering generator constraints”. Proceedings of the first IEEE International Conference on Control, Decision and Information Technologies CoDIT’13, Hammamet, Tunisia, 2013. [24] Niknam T, Mojarrad H D, Meymand H Z. “Non-smooth economic dispatch computation by fuzzy and self-adaptive particle swarm optimization”. Applied Soft Computing, 2011, 11(2): 2805-2817. [25] Safari A, Shayeghi H. “Iteration particle swarm optimization procedure for economic load dispatch with generator constraints”. Expert Systems with Applications, 2011, 38(5): 6043–6048. [26] Kumarasamy K, Raghavan R. “Particle swarm optimization algorithm for voltage stability improvement using multiple STATCOM”. Proceedings of International Conference on Emerging Trends in Electrical Engineering and Energy Management (ICETEEEM), Chennai, India, 2012, 235–242. Optimal Allocation and Sizing of Distributed Generation with Particle Swarm Optimization Algorithm for Loss Reduction

68

K. Ras Guerriche, T. Bouktir

[27] Nara, K, Hayashi, Y, Ikeda, K and Ashizawa, T, “Application of tabu search to optimal placement of distributed generators”, IEEE Power Engineering Society Winter Meeting, 2001, Volume: 2, PP.918 –923. [28] M. R. Aghaebrahimi, M. Amiri “An Immune-Based Optimization Method for Distributed Generation Placement in Order to Optimize Voltage Profile”, IEEE International Conference on Sustainable Power Generation and Supply, Nanjing, 978-14244-4934-7, 2009. [29] B. Hanumantha Rao and S. Sivanagaraju,“Optimum allocation and sizing of distributed generations based on Clonal Selection Algorithm for loss reduction and technical benefit of energy savings,” International conference on Advances in Power Conversion and Energy Technologies (APCET), pp. 1-5, 2-4 August 2012. [30] T. N. Shukla, S. P. Singh, V. Srinivasarao and K. B. Naik, “Optimal sizing of distributed generation placed on radial distribution systems,” Electric Power Components and Systems, vol. 38, pp. 260-274, 2010.

K. Ras Guerriche , T. Bouktir

69