[ieee 2013 ieee 7th international power engineering and optimization conference (peoco) - langkawi...

Effect of Population Size for DG Installation using EMEFA

1S. R. A. Rahima, 2I. Musirinb, 2M. M. Othmanc, 1M. H. Hussaind , 3M. H. Sulaimane, and 1A. Azmif

1 School of Electrical System Engineering, Universiti Malaysia Perlis (UniMAP),

Kampus Pauh Putra, 02000, Arau, Perlis, Malaysia. 2Faculty of Electrical Engineering, Universiti Teknologi MARA,

40450 Shah Alam, Malaysia 3Faculty of Electrical & Electronic, Universiti Malaysia Pahang (UMP),

Pahang, Malaysia.

[email protected], [email protected], [email protected], [email protected], [email protected]

Abstract- This paper presents a new Embedded Meta Evolutionary-Firefly Algorithm (EMEFA) for DG installation which considers the effect of population size on loss and cost minimization while improving the performance of the system. The proposed EMEFA technique is to alleviate the setback experienced in the Meta-EP and firefly in terms slow convergence and less accurate. Implementation of the proposed technique in minimizing both the distribution losses and fuel cost separately has indicated promising results, while maintaining the voltage at acceptable levels. Assessment on its performance with respect to other optimization techniques revealed that the proposed technique is superior in terms fast convergence and achieving more accurate solution, validated on a chosen IEEE Reliability Test System.

I. INTRODUCTION Recent trend changes in the electric utility infrastructure

have created the opportunities for the large number of distributed generation (DG) in the distribution networks environment. DG is defined as small scale generator connected directly to the distribution network on the customer site of the meter. DGs are modular and can be combined with management and storage energy systems in order to improve the operation of distribution systems [1]. The DG unit can be categorized as several forms of sources. Among others are the renewable and non-renewable energy resources. Technologies that utilize non-renewable or conventional energy resources included fuel cells, gas turbines and microturbines. For renewable energy resources, it comprises wind power, biomass, solar or photovoltaic (PV) generation. Indeed, renewable energy can significantly reduce the greenhouse emission as compared to the conventional power plants [2].

Along with the rapid development of DG technology into the distribution grids, it also require new electricity grid structures and strategies for their operation, control and management to ensure efficiency, sustainability, security, reliability and high quality of power supply. However DG can have both negative and positive impacts depending on their

size and location. It is very important and critical for utilities to minimize the negative impact and maximize the positive impact of DG. This is to determine the suitable location and sizing of DG to ensure for greatest benefits that can be obtained from the installation of DG [3, 4].

Numerous papers have reported studies on DG technologies which consider sizing and allocation using different computational intelligence (CI) techniques. This has also proven that AI is superior which translate the behavior of insects, fish and other living habitats into mathematical equations and eventually managed to solve power system optimization problems. Conventional analytical methods such as loss sensitivity, Linear Programming and Lagrangian have been applied to solve the DG allocation and sizing problems. Nevertheless, these methods are very time-consuming and less reliable, especially when dealing with complex power system problems. Other approach which is categorized under the CI have also been employed to solve different DG problems such as Stochastic Algorithms [5], Immune Algorithms, Evolutionary Algorithm [6, 7], Physical Algorithms [8] and Swarm Intelligence [9] are promising and still evolving in this field. With the advancement in CI technologies especially in Nature Inspired Algorithm (NIA), many researchers have turned their attention to meta-heuristic approach. Most studies are only limited to DG issues in terms of allocation and sizing, without taking into account the insight of the algorithm and properties. This is significant as the performance of the optimization techniques can be influenced by the effect of population size, search steps, probability (if any) and even to the level of mutation or breeding techniques. Thus, it is timely that this issue needs to be addressed accordingly.

This paper presents EMEFA technique for DG installation in Distribution System which aims to investigate the effect of population size on the optimization performance. The results obtained from the study revealed that the proposed EMEFA is capable to achieve optimal solutions within the significant minimal fuel cost by considering the effect of population size.

978-1-4673-5074-7/13/$31.00 ©2013 IEEE

2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June 2013

746

II. PROBLEM FORMULATION For several years, various techniques and procedures had

been developed by many researchers for loss minimization and allocation of DG. In order to improve the overall efficiency of power delivery in the distribution system, the reduction of I2R loss is very important parameter to be considered. The I2R loss can be separated into two parts which are based on the active and reactive components of branch currents. The total I2R loss in the distribution system with n number of branches is given in (1). The loss associated with the active and reactive power components of branch currents is given in (2) and (3) respectively.

j

n

jjloss RIP ∑

==

1

2 (1)

j

n

jajaL RIP ∑

==

1

2 (2)

1

2j

n

jrjrL RIP ∑

== (3)

Where; Ij = current magnitude Rj = resistance at branch I Iaj = active current component at branch I Irj = reactive current component at branch I

In order to study the effectiveness of the proposed technique, two different cases are considered in the investigation of the effect of population size for DG installation in the system. The installation of DG was reported as a method of minimizing the loss and could affect both the active and reactive power losses in the system. In this study, the DGs are modeled as PQ nodes. Total real power loss for the base case of the system are calculated, where no DG is installed in the system. To study the impact of DG installation on the system, the following two cases tabulated in Table 1 are considered:

TABLE 1 CASE STUDY

Case 1 The objective function (Of1) is to minimize loss in the presence of DG. Of1 is proposed to minimize active power loss as shown in (4).

Case 2 The objective function (Of2) is to calculate both the active power loss and fuel cost in the presence of DG. Of2 is proposed to minimize both active power loss and the fuel cost as shown in (5).

)(1 lossPMinf = (4)

)(2 CostlossPMinf += (5)

∑=

=n

jjj .RI

lossP

1

2 (6)

)1

2(∑=

++=NG

jGjP

ic

GjP

jb

jaCost (7)

Where aj, bj, and cj are three cost coefficient of the ith

generator. The value of these coefficients depends on different power system and different generators. Other cost such as purchase and installation cost for DG is neglected [10]. In this study, IEEE 14-bus test system is used as the test specimen and the corresponding coefficients are given in TABLE 1I.

TABLE II

COST COEFFICIENTS OF GENERATOR G1 G2 G3 G6 G8 DGi

a 0 0 0 0 0 0

b 20 20 40 40 40 40

c 0.043 0.25 0.01 0.01 0.01 0.01

The conceptual configuration designed for effect of of the

study involving the population size for DG installation is shown in Fig 1. The outputs of this study are loss and cost minimization in order to produce the best performance of DG installation while the population size generator varies from 5 to 50 numbers of population.

Fig 1: Conceptual configuration for effect of population size for DG

installation.


747

III. PROPOSED TECHNIQUE Generally, the idea is to apply the new EMEFA technique

and to investigate the performance of DG installation in the distribution system. The detailed description of the proposed technique is clearly presented in this section. The proposed technique is tested on the IEEE 14-bus distribution system, along with the discussion of results described in the next section.

A. Proposed Embedded Meta Evolutionary - Firefly

Algorithm (EMEFA) The proposed Embedded Meta Evolutionary - Firefly

Algorithm (EMEFA) technique was developed with an objective to minimize the distribution losses and fuel cost while satisfying the voltage constraint in the system. The EMEFA is developed based on embedded FA properties into the EP. Based on the literature study, it is found that hybridizing them together with other algorithms could improve it to be faster, efficient, and more robust [11]. The complete algorithm of the EMEFA can be summarized in Fig.2.

The proposed technique started by generating control variable of Xi,α which depends on the number of population size and control variables. A new population is bred by mutating the initial existing population by implementing the mutation operator. Mutation is the only variation operator used for generating the offspring from each parent [12]. The fitness of the offspring was calculated by calling the load flow

program. The mutation formula for EP is shown in (8) and (9).

1010exp )),(iτN),'N(((j)iη(j)iη' += τ (8)

) ,(i(j)NiηjBjB ii 10')()(' += (9)

Where: Bi (j), Bi ′(j), ηi(j) and ηi′(j) is the jth component of the respective vector. N(0,1) with mean 0 and 1. Nj(0,1) indicates that the random number will be new for each value of j.

In this method, the combined parents and offspring are ranked according to the fitness value. Based on the sorted fitness value, the current best value is selected from the first half value and set as the initial locations of fireflies. Consequently, the FA operation is performed by comparing the initial location of ith solution with its jth neighboring solution [13]. The firefly attractiveness, β can be defined by (10). The distance between any two fireflies, i and j are calculated using the Cartesian distance in (11). The movement of firefly i is attracted to another more attractive (brighter) firefly j, determined by (12).

2r-e0γββ = (10)

d

k)j,kxi,k(x jxixijr ∑

=−=−=

12

(11)

No Yes

Yes

No

Yes

No No

Yes

Start

variablecontrol of no

size population k :where

k x :sizematrix ,

:X variablecontrol Generate

=

=

α

ααkX

Constraint violation test

Violation test fail?

Pool full?

Data base collection

Fitness calculation

variablecontrol of no

size population k :where

k x :sizematrix ,,

:)(offspringm

XMutation

=

=

α

ααkmX

Fitness calculation

Combine parents and offspring & Sort population based on fitness value

Set initial location of ‘n’ fireflies

Converge?

End

A

A

Calculate distance and attractiveness

f(Xi >f(Xj) ?

Update position

Calculate fitness value for new solution

Sort population based on fitness value

Embedded Fireflies

Set no of population size (k)

Fig 2: Complete Algorithm for proposed method


748

)()ix-j(x2r-e0i xix 0.5-randαγβ ++=

(12)

Where β0, γ and r are the predefined attractiveness, light absorption coefficient, and distance respectively.

xi,k is the kth component of the spatial coordinate xi of ith fireflies.

Implementation on several case studies, EMEFA characteristics are employed to give better performance. Its characteristics are tabulated in TABLE III with the number of population size varied from 5 to 50.

IV. RESULTS AND DISCUSSION The proposed EMEFA technique was simulated and tested

on the IEEE 14-bus test system [10]. The effect of population size to loss and cost minimization with the DG installation in terms of different objective functions was observed by installing the DG at bus 14. In this case, Bus 14 was chosen due to its ability avoid under and over-compensated in DG

problems. The results are compared with those obtained using MEP and FA techniques. In the simulations, two cases are addressed which are loss minimization and cost minimization. The value of total loss without DG is given in Table IV and it includes the total fuel cost for the system.

TABLE IV

RESULT OF THE SYSTEM WITHOUT DG FOR IEEE 14-BUS

Total Loss (MW) Cost ($/h) F

13.55 8175.5 8189.1

A. Case 1: Objective Function, Of1

The effectiveness of the proposed EMEFA is compared with both MEP and FA for determining the effect of population size in minimizing the distribution losses, while satisfying the voltage constraint in the system. Of1 in case 1 is defined as in (4). Table V shows the results with different population size using the three techniques. With the increase of population size from 5 to 50 in stages, EMEFA and MEP displayed similarity and consistency compared to FA. The results for these three methods shows the small population sizes (5-10 population size) indicate inconsistencies compared to large population size (20-50 population size). The suitable population size for this case is 20 population size based on the consistency for every number of simulation. DG sizing can be summarized as in Table V. It is found that the DG sizing is considered to be a large scale with a value of greater than 50 MW.

TABLE III EMEFA CHARACTERISTIC

τ= 1 Number of DG unit: 1

α (scaling parameter): 0.25 Minimum value of β(attractiveness): 1

γ (absorption coefficient): 1

TABLE V RESULT OF OBJECTIVE FUNCTION, OF1 WITH INSTALLATION OF DG FOR IEEE 14-BUS

No of simulation

No of Population size

5 10 20 50

DG Sizing OF1 DG Sizing OF1 DG Sizing OF1 DG Sizing OF1

EMEFA Technique

1st 65.84384 9.28726 65.74409 9.28653 65.74404 9.28653 65.74404 9.28653

2nd 65.78647 9.28684 65.74933 9.28657 65.74404 9.28653 65.74404 9.28653

3rd 65.88425 9.28756 65.75157 9.28659 65.74404 9.28653 65.74404 9.28653

4rd 65.76923 9.28672 65.74446 9.28654 65.74404 9.28653 65.74404 9.28653

5th 65.79162 9.28688 65.74456 9.28654 65.74404 9.28653 65.74404 9.28653

MEP Technique

1st 65.74515 9.28654 65.74544 9.28654 65.74471 9.28654 65.74708 9.28655

2nd 65.98693 9.28834 65.74470 9.28654 65.74539 9.28654 65.74406 9.28653

3rd 65.74661 9.28655 65.74786 9.28656 65.74482 9.28654 65.74638 9.28655

4rd 65.78094 9.28680 65.74475 9.28654 65.74622 9.28655 65.74455 9.28654

5th 65.74918 9.28657 65.74615 9.28655 65.74408 9.28653 65.74417 9.28653

FA Technique

1st 68.00760 9.30788 66.95017 9.29666 68.00950 9.30790 68.49165 9.31374

2nd 68.28465 9.31118 66.31742 9.29099 69.60906 9.32901 67.68515 9.30422

3rd 63.06163 9.37732 68.41081 9.31273 65.45505 9.38908 67.52738 9.30251

4rd 68.28465 9.31118 65.89850 9.28767 68.76458 9.31725 69.90663 9.33349

5th 65.76894 9.28671 67.04871 9.29761 69.62138 9.32920 68.05472 9.30843


749

Fig 3 displayed the result of Of1 without and with installation of DG using three different techniques. Percentages of loss reduction without and with DG using EMEFA and MEP are 31.46%. However FA technique is less than proposed technique which is 31.3%.

.

Fig 3: Result of Objective Function, Of1 with installation of DG

B. Case 2: Objective Function, Of2 The analysis was done to determine the effect of population

size in cost minimization with the presence of DG. The Of2 is proposed to minimize both active power loss and the fuel cost as shown in (5). Table VI shows the results with different population size using three techniques. With the increase of population size from 5 to 50 in stages, EMEFA exhibits similarity and consistency compared to MEP and FA techniques. The results for these three methods are suitable for 20 population size. The results for small population sizes (5-10) indicate inconsistencies. It is found that the DG sizing is considered to be a medium scale with a value between 5 MW to 50 MW. Fig 4 illustrates the result of Of2 without and with the installation of DG using three different techniques. Percentages of cost reduction without and with DG using EMEFA, MEP and FA are 0.882%, 0.88 and 0.875% respectively.

TABLE VI RESULT OF OBJECTIVE FUNCTION, OF2 WITH INSTALLATION OF DG FOR IEEE 14-BUS

No of simulation

No of Population size

5 10 20 50 DG Sizing OF2 DG Sizing OF2 DG Sizing OF2 DG Sizing OF2

EMEFA Technique

1st 25.739355 8116.814447 25.751572 8116.814431 25.752879 8116.814431 25.751835 8116.814431

2nd 25.726897 8116.814496 25.752062 8116.814431 25.751773 8116.814431 25.75172 8116.814431

3rd 25.814859 8116.814848 25.767192 8116.814456 25.751675 8116.814431 25.751795 8116.814431

4rd 25.752096 8116.814431 25.749847 8116.814432 25.751867 8116.814431 25.751752 8116.814431

5th 25.849524 8116.815432 25.753871 8116.814432 25.751835 8116.814431 25.751773 8116.814431

MEP Technique 1st 28.976076 8117.897593 26.816338 8116.93296 25.38071 8116.828869 25.380679 8116.828871 2nd 25.579367 8116.817547 25.380695 8116.82887 25.380708 8116.828869 25.380782 8116.828863 3rd 27.87808 8117.286401 25.380805 8116.828862 25.380668 8116.828872 25.38085 8116.828858 4rd 29.650233 8118.396019 25.381265 8116.828826 25.38073 8116.828867 25.380732 8116.828867 5th 29.234833 8118.077845 25.380768 8116.828865 25.380735 8116.828867 25.380707 8116.828869

FA Technique 1st 28.163876 8117.421496 28.874206 8117.830412 28.189105 8117.434234 29.877952 8118.585477 2nd 27.981037 8117.333119 29.897552 8118.602281 29.27602 8118.107808 28.845485 8117.811858 3rd 26.663422 8116.901378 28.813055 8117.791112 29.501316 8118.277879 28.934957 8117.870219 4rd 29.49901 8118.276086 28.423464 8117.558849 28.899912 8117.847163 29.790354 8118.511338 5th 26.382001 8116.856004 28.37403 8117.531619 29.21531 8118.063765 28.198713 8117.439119


750

Fig 4: Result of Objective Function, Of2 with installation of DG

V. CONCLUSION In conclusion, the assessment of the performance of DG in

terms of loss and cost minimization at the identified location with optimal sizing was successfully tested on the test systems. The results show that the proposed technique was capable to reduce the system losses with the appropriate population size. The results for these three methods shows the small population sizes (5-10 population size) indicate inconsistencies compared to large population size (20-50 population size). The suitable population size for this case is 20 population size based on the consistency for every number of simulations.

ACKNOWLEDGMENT The authors gratefully acknowledge the School of Electrical Systems Engineering, Universiti Malaysia Perlis and Faculty of Electrical Engineering, Universiti Teknologi Mara for their continuous support, encouragement and the facilities provided to carry out this research work.

REFERENCES

[1] B.-R. Angel A, "Future development of the electricity systems with distributed generation," Energy, vol. 34, pp. 377-383, 2009.

[2] W. El-Khattam and M. M. A. Salama, "Distributed generation technologies, definitions and benefits," Electric Power Systems Research, vol. 71, pp. 119-128, 2004.

[3] D. Jing, Z. Ren-jun, Z. Si, and R. Yu-lin, "Multi-objective Allocation of Distributed Generation Considering Low-Carbon Effect," Energy Procedia, vol. 11, pp. 2629-2637, 2011.

[4] M. H. Sulaiman, O. Aliman, and S. R. A. Rahim, "Optimal embedded generation allocation in distribution system employing real coded genetic algorithm method," Proceedings of World Academy of Science, Engineering and Technology, vol. 62, pp. 591-596, 2010.

[5] K. Nara, Y. Hayashi, K. Ikeda, and T. Ashizawa, "Application of tabu search to optimal placement of distributed generators," in Power Engineering Society Winter Meeting, 2001. IEEE, 2001, pp. 918-923 vol.2.

[6] Z. M. Yasin, T. K. A. Rahman, I. Musirin, and S. R. A. Rahim, "Optimal sizing of distributed generation by using quantum-inspired evolutionary programming," in Power Engineering and Optimization Conference (PEOCO), 2010 4th International, 2010, pp. 468-473.

[7] M. H. Sulaiman, O. Aliman, and S. R. A. Rahim, "Optimal allocation and contingency analysis studies of embedded generation in distribution systems," International Energy Journal, vol. 12, pp. 67-76, 2011.

[8] A. Parizad, A. H. Khazali, and M. Kalantar, "Sitting and sizing of distributed generation through Harmony Search Algorithm for improve voltage profile and reducuction of THD and losses," in Electrical and Computer Engineering (CCECE), 2010 23rd Canadian Conference on, 2010, pp. 1-7.

[9] L. Y. Wong, S. R. A. Rahim, M. H. Sulaiman, and O. Aliman, "Distributed generation installation using particle swarm optimization," in PEOCO 2010 - 4th International Power Engineering and Optimization Conference, Shah alam, Selangor, 2010, pp. 159-163.

[10] Q. Kang, T. Lan, Y. Yan, L. Wang, and Q. Wu, "Group search optimizer based optimal location and capacity of distributed generations," Neurocomputing, vol. 78, pp. 55-63, 2012.

[11] H. Zang, S. Zhang, and K. Hapeshi, "A Review of Nature-Inspired Algorithms," Journal of Bionic Engineering, vol. 7, Supplement, pp. S232-S237, 2010.

[12] T. K. A. Rahman, S. R. A. Rahim, and I. Musirin, "Optimal allocation and sizing of embedded generators," National Power and Energy Conference, 2004. PECon 2004. Proceeding, 2004, pp. 288-294.

[13] X.-S. Yang, S. S. Sadat Hosseini, and A. H. Gandomi, "Firefly Algorithm for solving non-convex economic dispatch problems with valve loading effect," Applied Soft Computing, vol. 12, pp. 1180-1186, 2012.


751

[ieee 2013 ieee 7th international power engineering and optimization conference (peoco) - langkawi...

Documents