10
Maximization of Loadability of Distributed System by Optimal placement and sizing of DG using Parallel Particle Swarm Optimization (PPSO) A.Marimuthu a* ,K.Gnanambal b ,S.Divya c , T.Jaipoorani d ,P.Lakshmi e a Associate Professor, Department of Electrical and Elecronics Engg, K.L.N.College of Engineering, Pottapalayam-630612, India b Professor, Department of Electrical and Elecronics Engg, K.L.N.College of Engineering, Pottapalayam-630612, India c,d,e UG Students, Department of Electrical and Elecronics Engg, K.L.N.College of Engineering, Pottapalayam-630612, India Abstract Recently, Distributed Generation (DG) is increasing adopted in the distribution systems because of its high penetration levels and role on the voltage stability. In this paper, PSO with parallel mutation namely Parallel Particle Swarm Optimization (PPSO) is proposed to maximize system loadability by optimally placing and sizing of DG units in a radial distribution system The proposed method is to overcome the drawback of Particle swarm Optimization (PSO), which works with less number of iterations, but as the iterations increases, there is a possibility of getting trapped in local optima. The efficiency of proposed method is evaluated using IEEE 33 bus and IEEE 69 bus radial distribution test systems. The results show the integrity and effectiveness of the proposed method. Keywords: DG; Voltage stability; PSO; PPSO; RDS; 1. Introduction Distributed Generation, which is placed in distribution systems, has the ability to meet out the demand of growing energy markets. DG is mainly preferred because of its capability to improve maximum loadability, voltage profile and stability [1-5]. To achieve maximum technical and economic benefits, optimum location must be determined. Optimal location has been carried out by using both conventional as well as evolutionary algorithm. Conventional algorithms [6] including numerical method [7], analytical methods [8] and evolutionary algorithms such as bacterial foraging optimization [9], Particle Swarm Optimization [10], Imperialist competition algorithm [11] have been applied to the optimization problem of DG units. The hybrid genetic algorithm have developed for multi-objective location and sizing of DG [12]. To reduce cost, power loss and emission, a study has developed based on hybrid modified shuffled frog algorithm and differential evolutionary algorithm [13]. The result shows that the combined method has a higher probability of finding the optimal solution. However different strategies for installing multiple DGs are not considered. To optimally solve the allocation problem of DG, Artificial Bee Colony algorithm has also been presented [14, 15]. Multiple wind turbines are connected to distribution network using genetic algorithm [16, 17]. Different types of renewable DGs are investigated using modified honey bee mating algorithm [18]. Investigation on impact of loadability in distribution system is proposed [19, 20]. Weakest bus is identified as an optimum location for DG placement based on Continuation Power Flow (CPF) method [21]. Maximum loading up to the voltage constraint for optimal placement of DG is also considered [22]. The study [23] has concluded that the maximum loadability of distribution system has been limited by voltage limit rather than the thermal limit. Overall voltage profile is thus improved by improving maximum loadability. Due to increased loading power systems are heavily stressed which results in voltage stability problem. Several methods have been proposed to calculate maximum loadability limit. To maximize savings, PSO approach has been used to optimally determine the location of capacitors [24]. The optimal size and sites of DGs and capacitor combination are determined by Artificial Bee Colony algorithm [25]. _____________ * E-mail address:[email protected]. This paper presents parallel mutation in Particle Swarm Optimization technique to locate optimally DG JASC: Journal of Applied Science and Computations Volume 5, Issue 6, June /2018 ISSN NO: 0076-5131 Page No:258

Maximization of Loadability of Distributed System by Optimal ...Maximization of Loadability of Distributed System by Optimal placement and sizing of DG using Parallel Particle Swarm

  • Upload
    others

  • View
    6

  • Download
    0

Embed Size (px)

Citation preview

  • Maximization of Loadability of Distributed System by Optimal placement and sizing of DG using Parallel Particle Swarm

    Optimization (PPSO)

    A.Marimuthua*,K.Gnanambalb,S.Divyac, T.Jaipooranid,P.Lakshmie a Associate Professor, Department of Electrical and Elecronics Engg, K.L.N.College of Engineering, Pottapalayam-630612, India

    b Professor, Department of Electrical and Elecronics Engg, K.L.N.College of Engineering, Pottapalayam-630612, India c,d,e UG Students, Department of Electrical and Elecronics Engg, K.L.N.College of Engineering, Pottapalayam-630612, India

    Abstract

    Recently, Distributed Generation (DG) is increasing adopted in the distribution systems because of its high penetration levels and role on the voltage stability. In this paper, PSO with parallel mutation namely Parallel Particle Swarm Optimization (PPSO) is proposed to maximize system loadability by optimally placing and sizing of DG units in a radial distribution system The proposed method is to overcome the drawback of Particle swarm Optimization (PSO), which works with less number of iterations, but as the iterations increases, there is a possibility of getting trapped in local optima. The efficiency of proposed method is evaluated using IEEE 33 bus and IEEE 69 bus radial distribution test systems. The results show the integrity and effectiveness of the proposed method.

    Keywords: DG; Voltage stability; PSO; PPSO; RDS;

    1. Introduction

    Distributed Generation, which is placed in distribution systems, has the ability to meet out the demand of growing energy markets. DG is mainly preferred because of its capability to improve maximum loadability, voltage profile and stability [1-5]. To achieve maximum technical and economic benefits, optimum location must be determined. Optimal location has been carried out by using both conventional as well as evolutionary algorithm. Conventional algorithms [6] including numerical method [7], analytical methods [8] and evolutionary algorithms such as bacterial foraging optimization [9], Particle Swarm Optimization [10], Imperialist competition algorithm [11] have been applied to the optimization problem of DG units. The hybrid genetic algorithm have developed for multi-objective location and sizing of DG [12]. To reduce cost, power loss and emission, a study has developed based on hybrid modified shuffled frog algorithm and differential evolutionary algorithm [13]. The result shows that the combined method has a higher probability of finding the optimal solution. However different strategies for installing multiple DGs are not considered. To optimally solve the allocation problem of DG, Artificial Bee Colony algorithm has also been presented [14, 15]. Multiple wind turbines are connected to distribution network using genetic algorithm [16, 17]. Different types of renewable DGs are investigated using modified honey bee mating algorithm [18]. Investigation on impact of loadability in distribution system is proposed [19, 20]. Weakest bus is identified as an optimum location for DG placement based on Continuation Power Flow (CPF) method [21]. Maximum loading up to the voltage constraint for optimal placement of DG is also considered [22]. The study [23] has concluded that the maximum loadability of distribution system has been limited by voltage limit rather than the thermal limit. Overall voltage profile is thus improved by improving maximum loadability. Due to increased loading power systems are heavily stressed which results in voltage stability problem. Several methods have been proposed to calculate maximum loadability limit. To maximize savings, PSO approach has been used to optimally determine the location of capacitors [24]. The optimal size and sites of DGs and capacitor combination are determined by Artificial Bee Colony algorithm [25]. _____________

    * E-mail address:[email protected].

    This paper presents parallel mutation in Particle Swarm Optimization technique to locate optimally DG

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:258

  • considering maximum loadability and minimum power losses. In addition, Parallel Particle Swarm Optimization takes the advantage of reduced computation time compared to PSO technique. The proposed algorithm will allocate different part of the task to multiple processors and each processor will execute different part of the program. Furthermore, the proposed algorithm is tested in IEEE 33-bus and 69-bus radial distribution systems. The results obtained by proposed algorithm show that the proposed algorithm is outperformed by other evolutionary optimization methods. This paper is organised as follows: In section 2, the need for DG on voltage stability is discussed. Section 3, will briefly discuss about the Particle Swarm Optimization (PSO) algorithm. In section 4, the proposed PPSO algorithm is presented. Section 5, presents the problem formulation. Discussion about PSO algorithm on 33-bus and 69-bus system is in section 6. In section 7, proposed PPSO algorithm is used to test the system.

    2. Role of Distributed Generation on voltage stability To analyse the potential effect of DG on distribution system following definitions are used: Active line loss reduction:

    ���� = {����}������� �� �{����}���� ��

    {����}������� �� × 100 (1)

    Reactive line loss reduction:

    ���� = {����}������� ���{����}���� ��

    {����}������� �� × 100 (2)

    DG penetration level:

    �� ����������� ����� = ������� �������� �����

    �������� ����� ���� �� ���� × 100 (3)

    3. Overview of Particle Swarm Optimization PSO is an optimization technique based on the movement of swarms. It was developed by James Kennedy and Russell Eberhart. Each particle keeps track of its position in the solution space which is associated with best solution that is achieved so far by that particle. The value is called as personal best, pbest. Another best value is tracked by PSO is the best value obtained so far by any particle in the neighbourhood of that particle. This value is called gbest. Velocity of particle i at time k+1 is given by Eqn.4

    ����� = ���

    � + ��������� − ��� � + �� ������� − ��

    � � (4)

    Fig. 1 Depiction of velocity and position updates in PSO

    Where

    ��� : Current position;

    ����� : Modified new position

    �� : Current velocity ���� : Modified velocity ��

    : Personal best g� : Global best 3.1 Parameter selection

    Vki Xk

    i+1

    gb

    pb

    Vk+1

    Xki

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:259

  • The performance of the PSO is greatly affected by its parameter values. In this case, the selection of the PSO parameters follows the strategy of considering different values for each particular parameter and evaluating its effect on the PSO performance. Each particle fitness value has to be evaluated using a power flow solution at each iteration; the number of particles must be small to reduce the particles are chosen as an appropriate population sizes. 3.2 Inertia weight The inertia weight is linearly decreased. The purpose is to improve the speed of convergence of the results by reducing the inertia weight from an initial value of 0.9 to 0.1 in even steps over the maximum number of iterations.

    1max

    18.09.0

    iteration

    iterationWk

    Where �� : Inertia weight at iteration k��������� : Iteration number ������������ : Maximum number of iterations

    3.3 Acceleration constant A set of three values for the individual acceleration constants are evimportance to the individual’s best or the swarm’s best: C1 = {1.5, 2 and 2.5}. The value for the social acceleration constant is defined as: C2 = 4.5

    4. Overview of Parallel Particle Swarm Optimization Parallel processing is concerned with producing the same results using multiple processors with the goal of reducing the running time. The two basic parallel procesparallelism. The principle of pipeline processing is to separate the problem into a cascade of tasks where each of the tasks is executed by an individual processor. Data are transmitted through each processor wdifferent part of the process on each of the data elements. Since the program is distributed over the processors in the pipeline and the data moves from one processor to the next, no processor can proceed until the previous processor has finished its task. Data parallelism is an alternative approach which involves distributing the data to be processed amongst all processors which then executes the same procedure on each subset of the data. Data parallelism has been applied fairly widely incl(communication), each subpopulation will receive some new and promising chromosomes to replace the worst chromosomes in a subpopulation. This helps to avoid premature convergence. The parallel genetic algoribeen applied to vector quantization based communication via noisy channels successfully (Pan, McInnes & Jack 1996). Fig. 3 shows the flow chart for PPSO algorithm.

    The performance of the PSO is greatly affected by its parameter values. In this case, the selection of the PSO considering different values for each particular parameter and evaluating its

    effect on the PSO performance. Each particle fitness value has to be evaluated using a power flow solution at each iteration; the number of particles must be small to reduce the computational effort. Swarms of 5 and 20 particles are chosen as an appropriate population sizes.

    The inertia weight is linearly decreased. The purpose is to improve the speed of convergence of the results by eight from an initial value of 0.9 to 0.1 in even steps over the maximum number of

    : Inertia weight at iteration k

    : Maximum number of iterations

    A set of three values for the individual acceleration constants are evaluated to study the effect of giving more importance to the individual’s best or the swarm’s best: C1 = {1.5, 2 and 2.5}. The value for the social

    C2 = 4.5 – C1. Flow chart of PSO algorithm is as shown in Fig. 2

    Fig. 2 Flow chart for PSO algorithm

    4. Overview of Parallel Particle Swarm Optimization

    Parallel processing is concerned with producing the same results using multiple processors with the goal of reducing the running time. The two basic parallel processing methods are pipeline processing and data parallelism. The principle of pipeline processing is to separate the problem into a cascade of tasks where each of the tasks is executed by an individual processor. Data are transmitted through each processor wdifferent part of the process on each of the data elements. Since the program is distributed over the processors in the pipeline and the data moves from one processor to the next, no processor can proceed until the previous

    nished its task. Data parallelism is an alternative approach which involves distributing the data to be processed amongst all processors which then executes the same procedure on each subset of the data. Data parallelism has been applied fairly widely including to genetic algorithms. (communication), each subpopulation will receive some new and promising chromosomes to replace the worst chromosomes in a subpopulation. This helps to avoid premature convergence. The parallel genetic algoribeen applied to vector quantization based communication via noisy channels successfully (Pan, McInnes & Jack 1996). Fig. 3 shows the flow chart for PPSO algorithm.

    The performance of the PSO is greatly affected by its parameter values. In this case, the selection of the PSO considering different values for each particular parameter and evaluating its

    effect on the PSO performance. Each particle fitness value has to be evaluated using a power flow solution at computational effort. Swarms of 5 and 20

    The inertia weight is linearly decreased. The purpose is to improve the speed of convergence of the results by eight from an initial value of 0.9 to 0.1 in even steps over the maximum number of

    (5)

    aluated to study the effect of giving more importance to the individual’s best or the swarm’s best: C1 = {1.5, 2 and 2.5}. The value for the social

    . Flow chart of PSO algorithm is as shown in Fig. 2

    Parallel processing is concerned with producing the same results using multiple processors with the goal of sing methods are pipeline processing and data

    parallelism. The principle of pipeline processing is to separate the problem into a cascade of tasks where each of the tasks is executed by an individual processor. Data are transmitted through each processor which executes a different part of the process on each of the data elements. Since the program is distributed over the processors in the pipeline and the data moves from one processor to the next, no processor can proceed until the previous

    nished its task. Data parallelism is an alternative approach which involves distributing the data to be processed amongst all processors which then executes the same procedure on each subset of the data. Data

    With this migration (communication), each subpopulation will receive some new and promising chromosomes to replace the worst chromosomes in a subpopulation. This helps to avoid premature convergence. The parallel genetic algorithm has been applied to vector quantization based communication via noisy channels successfully (Pan, McInnes & Jack

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:260

  • The performance of the PPSO is also highly dependent on the level of correlation between parameters and the nature of the communication strategy. Three communication strategies have been developed for parallel particle swarm optimization algorithm. The firobservation that if the parameters are independent or are only loosely correlated, then the better particles may get good results quite quickly. Thus multiple copies of the best particles imutated particles migrate and replace the worst particles in the other groups for every R1 iterations.

    However, if the parameters of the solution are loosely correlated the better particles in each group may not get optimum results particularly quickly. In this case, a second communication strategy may be applied as depicted in Fig 5. This strategy is based on selfits neighbour groups to replace some of the more poorly performing particles for R2 iterations.an associated position vector xi (t) and velocity vector vupdate of each particle should be done by taking into account coordinates personal best and global best. This is done by calculating the signed distance between these coordinates and the current posIteration ‘t’ and integrating these computations into the calculation of the updated velocity vector. The velocity of particle ‘i’ at iteration‘t’ is determined by taking into account the previous velocity of the particle ‘i’, vi (t − 1) weighed by a inertia factor w (t). The current position of the particle, weighed by R1 , a random value within the range (0, 1) generated by a uniform distribution at each iteration, and by C1 , a positive valued constant named the cognitive acceleration constant, with a suggested value of 2.0. This second component represents the linear attraction towards the best position ever found by particles personal best and therefore is named “memory” or “self-knowledge”best value namely global best, and the current position of the particle, weighed by R2, a random value within the range (0, 1) generated by a uniform distribution at each iteration, and by C2 , a positive valuethe social acceleration constant, with a suggested value of 2.0. This last component is the linear attraction towards the best position ever found by any particle in the group, thus receiving the name

    Fig. 3 Flow chart of proposed PPSO algorithm

    The performance of the PPSO is also highly dependent on the level of correlation between parameters and the

    nature of the communication strategy. Three communication strategies have been developed for parallel particle swarm optimization algorithm. The first communication strategy shown in Fig. 4 is designed based on the observation that if the parameters are independent or are only loosely correlated, then the better particles may get good results quite quickly. Thus multiple copies of the best particles in each group are mutated and those mutated particles migrate and replace the worst particles in the other groups for every R1 iterations.

    Fig. 4 Loosely correlated parameter

    However, if the parameters of the solution are loosely correlated the better particles in each group may not get optimum results particularly quickly. In this case, a second communication strategy may be applied as depicted

    on self-adjustment in each group. The best particle in each group is migrated to groups to replace some of the more poorly performing particles for R2 iterations.

    (t) and velocity vector vi (t) in the hyperspace. As referred earlier, the position update of each particle should be done by taking into account coordinates personal best and global best. This is done by calculating the signed distance between these coordinates and the current position of each particle at Iteration ‘t’ and integrating these computations into the calculation of the updated velocity vector. The velocity

    is determined by taking into account the previous velocity of the particle ‘i’, − 1) weighed by a inertia factor w (t). The current position of the particle, weighed by R1 , a random value

    within the range (0, 1) generated by a uniform distribution at each iteration, and by C1 , a positive valued tive acceleration constant, with a suggested value of 2.0. This second component

    represents the linear attraction towards the best position ever found by particles personal best and therefore is knowledge”. The signed distance between the coordinates associated with the global

    best value namely global best, and the current position of the particle, weighed by R2, a random value within the range (0, 1) generated by a uniform distribution at each iteration, and by C2 , a positive valuethe social acceleration constant, with a suggested value of 2.0. This last component is the linear attraction towards the best position ever found by any particle in the group, thus receiving the name

    t=R1+1

    MUTATE AND UPDATE

    J=00 J=01 J=10

    t=1

    t=2

    t=R1

    The performance of the PPSO is also highly dependent on the level of correlation between parameters and the nature of the communication strategy. Three communication strategies have been developed for parallel particle

    st communication strategy shown in Fig. 4 is designed based on the observation that if the parameters are independent or are only loosely correlated, then the better particles may

    n each group are mutated and those mutated particles migrate and replace the worst particles in the other groups for every R1 iterations.

    However, if the parameters of the solution are loosely correlated the better particles in each group may not get optimum results particularly quickly. In this case, a second communication strategy may be applied as depicted

    adjustment in each group. The best particle in each group is migrated to groups to replace some of the more poorly performing particles for R2 iterations. Each particle has

    (t) in the hyperspace. As referred earlier, the position update of each particle should be done by taking into account coordinates personal best and global best. This is

    ition of each particle at Iteration ‘t’ and integrating these computations into the calculation of the updated velocity vector. The velocity

    is determined by taking into account the previous velocity of the particle ‘i’, − 1) weighed by a inertia factor w (t). The current position of the particle, weighed by R1 , a random value

    within the range (0, 1) generated by a uniform distribution at each iteration, and by C1 , a positive valued tive acceleration constant, with a suggested value of 2.0. This second component

    represents the linear attraction towards the best position ever found by particles personal best and therefore is en the coordinates associated with the global

    best value namely global best, and the current position of the particle, weighed by R2, a random value within the range (0, 1) generated by a uniform distribution at each iteration, and by C2 , a positive valued constant named the social acceleration constant, with a suggested value of 2.0. This last component is the linear attraction towards the best position ever found by any particle in the group, thus receiving the name “cooperation” or

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:261

  • “social knowledge”. When the correlation property of the solution space is known the first and second communication strategies work well. However, they can perform poorly if applied in the wrong situation. In the cases where the correlation property is unknown, a hybrid communication strategy can be applied. The hybrid communication strategy separates the groups into two subgroups with the first subgroup applying the first communication strategy for R1 iterations and all groups applying the second communication strategy for R2 iterations.

    Fig. 5 Strongly correlated parameter

    Parallel mutation uses multiple processors to yield the same results but with reduced run time. Pipeline processing and data parallelism are the two basic methods used for parallel processing. Pipeline processing is aimed to segregate the problem into a series of tasks each task is evaluated by an individual processor. The main disadvantage of this method is that a processor can proceed only if the previous processor has completed its work. Data parallelism overcomes the disadvantage of pipeline processing. In this work, data parallelism is deployed for creating a parallel mutation PSO (PPSO) algorithm. 5. Problem Formulation PSO and PPSO techniques are used to find the maximum loadability of the system, real and reactive power losses and also to observe the increase in active and reactive load using Eqn. (6, 7) till divergence is attained. P����� = P������� × � (6) Q����� = Q������� × � (7) Where � : Loading factor P������� : Initial active power Q������� : Initial reactive power P����� : Final active power Q����� : Final reactive power Size of DG: 0 ≤ ∑ ���

    ���� ≤ ����� (8)

    Position of DG: 2 ≤ �� �������� ≤ ����� ��� (9) The amount of active and reactive power are calculated using Eqn. (7, 8, 9)

    J=00 J=01 J=10 J=11

    t=1

    t=2

    t=R2

    t=R2+1

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:262

  • ����������� = ��������������� × ����� ������ (10)

    ��������������� = ����������������� − �����������

    � (11)

    ������������ = �∑ �������

    ���+ ∑ ��

    ��������

    (12)

    Table 1 shows the system details of 33-bus and 69-bus radial distribution system

    Table 1. Test details of 33-bus and 69-bus radial distribution system

    Test system Active Power in KW

    Reactive Power in KW

    Apparent Power in KVA

    Power Factor

    33-bus system 3715 2300 4369.35 0.8502

    69-bus system 3801.9 2694.1 4659.67 0.8159

    6. Implementation PSO in radial distribution system

    In this section, optimum multi – DG unit placement with PSO technique will be applied on 3 phase, 12.66 KV standard 33-bus and 69-bus test system. DG will provide active and reactive power hence DG is considered as a PQ load. 6.1. IEEE 33-bus radial distribution system The base load of 33-bus radial distribution system is 4369.35 KVA. For DG placement, when PSO algorithm is applied on 33-bus radial distribution system, the obtained results are presented in Table 2.

    Table 2. Application of PSO algorithm for optimum single and multi-DG unit’s placement on 33-bus radial distribution system

    Parameters Without DG PSO technique for DG allocation in 33 - bus system

    No. of DG units

    --

    Single DG unit

    Two DG units

    Three DG units

    DG size ( Position)

    --

    1258.2500 (30)

    12.0000 (22)

    463.7316 (30)

    24.0000 (10)

    402.8337 (30) 1112.2106 (33)

    Active power losses (KW)

    210.9991

    151.3733

    141.8485

    138.8979

    Reactive power losses (KW)

    143.1775

    103.8981

    96.5329

    94.4287

    Maximum loadability

    3.41

    3.9827

    4.4320

    4.6821

    From Table 2, It can be found that for base case the active and reactive power losses were found to be 210.9991 KW and 143.1775 KW. For single DG unit the losses are found to be reduced. When multi-DGs are included the losses are reduced further than the previous cases. Maximum loadability is improved with addition of DG. The variation of real power losses with the number of DG unit is as shown in Fig. 6.

    Fig. 6 Real power loss for different DG units in 33-bus system with PSO technique

    With PSO algorithm the reactive power losses were found to be reduced with optimal DG placement as shown

    0 10 20 30 40 50 60 70 80 90 100135

    140

    145

    150

    155

    160

    165

    Iteration

    Plo

    ss (

    KW

    )

    Single DG (30)

    Two DG (22 & 30)

    Three DG (10,30 &33 )

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:263

  • in Fig. 7

    Fig. 7 Reactive power loss for different DG units in 33-bus system with PSO technique

    6.2. IEEE 69-bus radial distribution system The base load of 69-bus radial distribution system is 4659.67 KVA. For DG placement, when PSO algorithm is applied on 69-bus radial distribution system, the obtained results are presented in Table 3.

    Table 3. Application of PSO algorithm for optimum single and multi-DG unit’s placement on 69-bus radial distribution system

    Parameters Without DG PSO technique for DG allocation in 69 - bus system

    No. of DG units

    --

    Single DG unit

    Two DG units

    Three DG units

    DG size ( Position)

    --

    1329.8991 (61)

    16.0000 (61)

    334.6632 (26)

    20.0000 (20)

    252.1558 (61) 1142.4090 (26)

    Active power losses (KW)

    224.9521

    152.0051

    146.4946

    145.6454

    Reactive power losses (KW)

    102.3115

    70.6369

    68.3470

    66.8185

    Maximum loadability

    3.32

    4.0179

    4.3578

    4.6801

    From Table 3, It can be found that for base case of 69-bus system the active and reactive power losses were found to be 224.9521 KW and 102.3115 KW. The losses are reduced with single DG unit. When multi-DGs are included the losses are reduced further than the previous cases. Maximum loadability can also be improved with number of DG unit’s placement. The variation of real power losses with the number of DG unit is as shown in Fig. 8.

    Fig. 8 Real power loss for different DG units in 69-bus system with PSO technique

    The reactive power losses were found to be reduced with optimal placement of DG units as shown in Fig. 9. 7. Implementation Proposed PPSO in radial distribution system Optimum multi-DG unit placement with Proposed PPSO technique will be applied on 3 phase, 12.66 KV standard 33-bus and 69-bus test system. As the DG unit produces active and reactive power, it is considered as PQ load. The power factor of 33-bus system is 0.85 and for 69-bus system is 0.8159.

    0 10 20 30 40 50 60 70 80 90 10094

    96

    98

    100

    102

    104

    106

    108

    110

    112

    Iteration

    Qlo

    ss (

    KW

    )

    Single DG (30)

    Two DG (22 & 30)

    Three DG (10,30 &33)

    0 10 20 30 40 50 60 70 80 90 100145

    150

    155

    160

    165

    170

    175

    180

    185

    190

    195

    Iteration

    Plo

    ss (

    KW

    )

    Single DG (61)

    Two DG (26 & 61)

    Three DG (20,26 & 61)

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:264

  • Fig. 9 Reactive power loss for different DG units in 69-bus system with PSO technique

    7.1. 33-bus radial distribution system The base load of 33-bus radial distribution system is 4369.35 KVA. For DG placement, when PPSO algorithm is applied on 69-bus radial distribution system, the obtained results are presented in Table 4. Table 4. Application of proposed PPSO algorithm for optimum single and multi-DG unit’s placement on 33-bus radial distribution system

    Parameters

    Without DG

    PPSO technique for DG allocation in 33 - bus system

    No. of DG units

    --

    Single DG unit

    Two DG units

    Three DG units

    DG size ( Position)

    --

    1137.9077 (32)

    13.0000 (30)

    813.2147 (24)

    24.0000 (13)

    895.3480 (30) 911.1455 (21)

    Active power losses (KW) 210.9991 111.0285 87.2528 73.5895 Reactive power losses (KW)

    143.1775

    81.7274

    59.7651

    50.9988

    Maximum loadability

    3.41

    4.1241

    4.5107

    4.8107

    From Table 4, it can be found that for base case the active and reactive power losses were found to be 210.9991 KW and 143.1775 KW. The losses are reduced and maximum loadability is improved with addition of DG unit. The variation of real power losses with the number of DG unit is as shown in Fig. 11.

    Fig. 10 Real power loss for different DG units in 33-bus system with proposed PPSO technique

    Fig.11 Reactive power loss for different DG units in 33-bus system with Proposed PPSO technique

    0 10 20 30 40 50 60 70 80 90 10066

    68

    70

    72

    74

    76

    78

    80

    82

    Iteartion

    Qlo

    ss (

    KW

    )

    Single DG (61)

    Two DG (26 & 61)

    Three DG (20,26 & 61)

    0 10 20 30 40 50 60 70 80 90 10070

    80

    90

    100

    110

    120

    130

    140

    150

    160

    170

    Iteration

    Plo

    ss (

    KW

    )

    Single DG (32)

    Two DG (24 &30)

    Three DG (13,21 & 30)

    0 10 20 30 40 50 60 70 80 90 10050

    60

    70

    80

    90

    100

    110

    120

    130

    140

    150

    Iteration

    Qlo

    ss (

    KW

    )

    Single DG (32)

    Double DG (24 & 30)

    Three DG (13,21,30)

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:265

  • The reactive power loss of the system can be reduced with multi-DG unit rather than a system with no DG placement as shown in Fig.11. 7.2 69-bus radial distribution system The base load of 69-bus radial distribution system is 4659.67 KVA. When PPSO algorithm is applied on 69-bus radial distribution system, the results obtained are as shown in Table 5. Table 5. Application of proposed PPSO algorithm for optimum single and multi-DG unit’s placement on 69-bus radial distribution system

    Parameters Without DG PPSO technique for DG allocation in 69-bus system

    No. of DG units

    --

    Single DG unit

    Two DG units

    Three DG units

    DG size ( Position)

    --

    1872.6293 (61)

    17.0000 (61)

    568.5644 (36)

    49.0000 (61)

    589.4083 (22) 1761.6589 (18)

    Active power losses (KW)

    224.9521

    83.1894

    71.7659

    70.5379

    Reactive power losses (KW)

    102.3115

    40.6482

    36.0753

    33.3004

    Maximum loadability

    3.32

    4.1107

    4.3591

    4.7347

    From Table 5, the following results are observed for base case of 69-bus system, the active and reactive power losses were found to be 224.9521 KW and 102.3115 KW. When DG units are included both active and reactive power losses are reduced. Maximum loadability can also be improved with number of DG unit’s placement. The variation of real power losses with optimal DG unit placement is as shown in Fig. 12.

    Fig. 12 Real power loss for different DG units in 69-bus system with Proposed PPSO technique

    Fig. 13 shows the reduction in reactive power losses in 69-bus system with DG placement using PPSO technique.

    Fig. 13 Reactive power loss for different DG units in 69-bus system with proposed PPSO technique

    8. Conclusion This paper has presented an effective new approach for optimal placement of DG with Particle Swarm Optimization involving Parallel mutation. The DG placement and sizing are based on maximization of loadability and to minimize active and reactive power losses. PPSO will help in getting the best result considering the objective function. The proposed technique is applied on 33-bus and 69-bus radial distribution system and it was found that PPSO technique works better than simple PSO technique.

    0 10 20 30 40 50 60 70 80 90 10065

    70

    75

    80

    85

    90

    95

    100

    Iteration

    Plo

    ss

    (K

    W)

    Single DG (61)

    Two DG (36 & 61)

    Three DG (18,22 & 61)

    0 10 20 30 40 50 60 70 80 90 10032

    34

    36

    38

    40

    42

    44

    46

    48

    50

    Iteration

    Qlo

    ss (

    KW

    )

    Single DG (61)

    Two DG (36 & 61)

    Three DG (18,22 & 61)

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:266

  • References

    [1] Chiradeja P, Ramakumar R. An approach to quantify the technical benefits of distributed generation. IEEE Trans Energy Convers

    2004; 19(4):764-73. [2] Bayod-Rújula AA. Future development of the electricity systems with distributed generation. Energy 2009; 34(3):377-83. [3] Zangiabadi M, Feuillet R, Lesani H, Hadj-Said N, Kvaløy JT. Assessing the performance and benefits of customer distributed

    generation developers under uncertainties. Energy 2011; 36(3):1703-12. [4] Soroudi A, Ehsan M. A distribution network expansion planning model considering distributed generation options and techno-

    economical issues. Energy 2010; 35(8):3364-74. [5] Bracco S, Dentici G, Siri S. Economic and environmental optimization model for the design and the operation of a combined heat and

    power distributed generation system in an urban area. Energy 2013; 55(0):1014-24. [6] Tan W-S, Hassan MY, Majid MS, Abdul Rahman H. Optimal distributed renewable generation planning: a review of different

    approaches. Renew Sustain Energy Rev 2013; 18:626–45. [7] Al Abri RS, El-Saadany EF, Atwa YM. Optimal placement and sizing method to improve the voltage stability margin in a distribution

    system using distributed generation. IEEE Trans Power Syst 2013; 28:326–34. [8] Gozel T, Hocaoglu MH. An analytical method for the sizing and siting of distributed generators in radial systems. Elect Power Syst

    Res 2009; 79:912–8. [9] Devi S, Geethanjali M. Application of modified bacterial foraging optimization algorithm for optimal placement and sizing of

    distributed generation. Expert Syst Appl 2014; 41:2772–81. [10] Mistry KD, Roy R. Enhancement of loading capacity of distribution system through distributed generator placement considering

    techno-economic benefits with load growth. Int J Elect Power Energy Syst 2014; 54:505–15. [11] Soroudi A, Ehsan M. Imperialist competition algorithm for distributed generation connections. IET Generation Transm Distrib 2012;

    6:21–9. [12] Moradi MH, Abedini M. A combination of genetic algorithm and particle swarm optimization for optimal DG location and sizing in

    distribution systems. Int J Elect Power Energy Syst 2012; 34:66–74. [13] Doagou-Mojarrad H, Gharehpetian GB, Rastegar H, Olamaei J. Optimal placement and sizing of DG (distributed generation) units in

    distribution networks by novel hybrid evolutionary algorithm. Energy 2013;54:129–38. [14] El-Zonkoly A. Optimal placement and schedule of multiple grid connected hybrid energy systems. Int J Elect Power Energy Syst

    2014;61:239–47. [15] Lalitha MP, Reddy NS, Reddy VV. Optimal DG placement for maximum loss reduction in radial distribution system using ABC

    algorithm. Int J Rev Comput 2010;3:44–52. [16] Mokryani G, Siano P. Optimal wind turbines placement within a distribution market environment. Appl Soft Comput 2013;13:4038–

    46. [17] Ugranlı F, Karatepe E. Optimal wind turbine sizing to minimize energy loss. Int J Elect Power Energy Syst 2013;53:656–63. [18] Niknam T, Taheri SI, Aghaei J, Tabatabaei S, Nayeripour M. A modified honey bee mating optimization algorithm for multiobjective

    placement of renewable energy resources. Appl Energy 2011;88:4817–30. [19] Alonso M, Amaris H. Voltage stability in distribution networks with DG. In: IEEE PowerTech conference 2009. pp. 1-6. [20] Hemdan NGA, Kurrat M. Distributed generation location and capacity effect on voltage stability of distribution networks. In:

    Conference distributed generation location and capacity effect on voltage stability of distribution networks. p. 1-5. [21] Hedayati H, Nabaviniaki S, Akbarimajd A. A method for placement of DG units in distribution networks. IEEE Trans Power Deliv

    2008;23(3):1620-28. [22] Alonso M, Amaris H. Voltage stability in distribution networks with DG. In: IEEE PowerTech conference 2009. pp. 1-6. [23] Prada RB, Souza LJ. Voltage stability and thermal limit: constraints on the maximum loading of electrical energy distribution feeders.

    IEE Proc Gener Transm Distrib 1998;145(5):573-7. [24] Singh SP, Rao AR. Optimal allocation of capacitors in distribution systems using particle swarm optimization. Int J Electr Power

    Energy Syst 2012;43:1267–75. [25] Abu-Mouti FS, El-Hawary ME. Optimal distributed generation allocation and sizing in distribution systems via artificial bee colony

    algorithm. IEEE Trans Power Deliv 2011;26(4):2090–101.

    JASC: Journal of Applied Science and Computations

    Volume 5, Issue 6, June /2018

    ISSN NO: 0076-5131

    Page No:267