+Model JESIT 46 1–12
ARTICLE IN PRESS Available online at www.sciencedirect.com
ScienceDirect Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
Impact of DG different types on the grid performance
1
2
A.M. Abd-rabou a , A.M. Soliman b,∗ , A.S. Mokhtar a
Q1
a
3
b
4
Faculty of Electrical Engineering, University Technology, Malaysia Electronics Research Institute, Power Electronics Department, Egypt
Received 15 December 2014; received in revised form 15 March 2015; accepted 24 April 2015
5
Abstract
6
21
Distributed generation (DG) is playing an important role in power system to improve the grid performance. Optimizing the locations of DGs is necessary to enhance the Grid performance and to avoid the degradation of the power system networks. The type of DG directly influences the penetration level and the placement of DG. In this study, genetic algorithm (GA) technique is proposed to find the optimum location for three different types of distributed generation, these types are synchronous generator, wind turbine and photovoltaic. The obtained results are also validated using Particle Swarm Optimization (PSO). Three different types of DGs will be penetrated individually to find their optimum location and investigate their impact on the grid performance. This approach will be applied on IEEE 13 bus system which is simulated using Power System Computer Aided Design (PSCAD) and Matlab software. GA and PSO are used to find the optimum solution of multi-objective function; the objective function combines the overall number of buses experience voltage sag, the number of buses experience voltage drop, the number of buses experience voltage less than 10%, the overall number of buses experience voltage swell and network power loss. Finally, results are analyzed, discussed and validated which show that the optimum locations of each DG will vary according to the type of DG but all of them will be around the load center. In the mean time, results showed that the highest grid overall performance was for synchronous generator DG and the lowest for photovoltaic DG. © 2015 Electronics Research Institute (ERI). Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
22
Keywords: IEEE 13 bus; Voltage sag mitigation; Grid performance; Optimal DG locations; Genetic algorithm
7 8 9 10 11 12 13 14 15 16 17 18 19 20
23
1. Introduction
24
25 26 27 Q3
Q2
Utility all over the world have for decades worked on investment of what is known as power quality. Voltage sag is considered as one of the most serious hazard of power quality problems and can lead to significant damage in sensitive devices (Jahromi et al., 2007; Dhas and Prakash, 2011). It is defined as a short reduction in RMS (Root Mean ∗
Corresponding author. Tel.: +20 1003438162; fax: +20 233310551. E-mail addresses:
[email protected] (A.M. Abd-rabou),
[email protected] (A.M. Soliman),
[email protected] (A.S. Mokhtar). Peer review under the responsibility of Electronics Research Institute (ERI).
http://dx.doi.org/10.1016/j.jesit.2015.04.001 2314-7172/© 2015 Electronics Research Institute (ERI). Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12 2
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
65
Square) voltage magnitude and can be produced due to short circuit, wind contamination on electrical insulator and starting of large motors. The sag phenomena ranges from 2 cycles up to 10 cycles and its magnitude ranges from 0.1 to 0.9 (Jahromi et al., 2007; Dhas and Prakash, 2011). Many solutions are proposed to mitigate voltage sag such as DVR (Dynamic Voltage Restorer) series and shunt configuration to inject active and reactive power to compensate the voltage (Dhas and Prakash, 2011; Martinez-Velasco, 2007; Song-cen et al., 2008) in which authors focused only on the control procedure and used battery banks with limited energy storage. Other solutions were the placement of DGs (distributed generations) in the electrical network due to a lot of their benefits such as improving protection reliability and the voltage profile (Collins and Jiang, 2008), reducing losses (Dhas and Prakash, 2011; Martinez-Velasco, 2007; Gozel et al., 2005; Kamalinia et al., 2007; Sedighzadeh and Rezzadeh, 2007) but not to mitigate the voltage sag. Another author used genetic algorithm optimization technique to mitigate voltage sag (Jahromi et al., 2007) but with disadvantages of using combination of single phase DG and three phase DG which is not realistic to propose single phase DG with approximately 500 kW. In-addition the researcher in (Jahromi et al., 2007) used only general type of DG, rather than he applied three phase short circuit to simulate voltage sag while single phase short circuit is frequently occurs, almost 80% (Bolen, 2000). Different types of distributed generations are widely used nowadays due to lot of their benefits but each of them has its own characteristics and response in the electrical network (Jahromi et al., 2007; Gozel et al., 2005; Sedighzadeh and Rezzadeh, 2007). Hence, it is highly important to place each type at the correct location to avoid the bad performance of the electrical network grid. For example, if the number of buses experience voltage sag increased when DGs are inserted in the electrical network grid than the case without DGs or there is no optimum DG location then, more and more devices and equipments connected to these buses will exposed to serious damage and lot of control systems may breakdown rather than many operations controlled by the sensitive devices may stopped or blocked. In this paper, three individuals DGs resources are used to improve the performance of unbalanced radial system used for medium voltage distribution. These resources are photovoltaic, wind turbine and diesel generator. IEEE 13 bus approach is used to represent the unbalanced redial network. The optimum locations for DGs are presented for each resource separately using GA technique in which results had been validated using PSO technique. In the mean time, the load flow was studied in which power losses in the network were calculated and discussed for each DG type. The optimum locations will be implemented by constructing an objective function; this function will combine four factors. The first factor is the overall number of 1 ph buses experience voltage sag all over the whole process, the sag will be produced by executing short circuit (1 ph and 3 ph) at all the buses for a specific period of time 0.3–0.5 s for each DG location. 2nd factor is the overall number of 1 ph buses experience voltage drop 0.9 over the whole process. The 3rd factor is the overall number of 1 ph buses experiences zero voltage. The 4th one is the overall number of 1 ph buses experience voltage swell 1.1. To minimize the search space, some assumptions are considered such as; the inserted DGs are two identical types and size each three-phase 500 kW. An important assumption is that the load is considered fixed with time to avoid the necessity of increasing or decreasing the DG sizes. This paper is organized as follows; Section 2 presents the research methodology that used to formulate the flow chart of the optimization procedures using GA. Results and discussions for the three types of distributed generations including their impacts on voltage mitigation and grid performance are discussed in Section 3 in which also results were validated using PSO optimization technique. Finally, conclusion and future work are presented in Section 4.
66
2. Research methodology
67
2.1. Test system
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64
68 69 70 71 72 73 74
Since most of distribution networks are radial systems and include different types of balanced and unbalanced loads and different types of short transmission lines with different configurations, then it is highly important to propose a network example that contains all these features. One of these networks is IEEE 13 bus system which is a radial network which contains balanced loads, unbalanced loads, single phase loads, three phase loads, short lines of different configurations and two types of transformers (Farag et al., 2012; Martine et al., 2011). IEEE 13 bus is modeled using PSCAD software to start modeling this network, all the transmission lines included in IEEE 13 bus system are estimated as short transmission lines. The IEEE 13 bus system single line diagram is shown in Fig. 1. Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
3
Fig. 1. Single line diagram of IEEE 13 bus system. 75
76 77 78 79 80 81 82 83 84 85 86 87 88
89 90 91
2.2. Problem formulation and optimization technique The main target of this study is to find the optimum location for each DG type individually to improve the power quality. To implement this purpose, IEEE 13 bus grid is simulated using PSCAD for two cases “with and without DG insertion” at normal operating condition. The bus voltages must be within the limit and the power flow is performed using Newton Raphson method. As starting step to formulate genetic algorithm (GA) optimization technique, we begin by modeling the IEEE 13 bus in PSCAD without DGs connection to the grid during normal operating condition and apply two types of faults, single phase and three phase faults. This case is taken as reference (threshold) for all DG locations. All the bus voltages are recorded at normal and faults conditions for 0.1 s duration. The buses experienced voltage sag, voltage drop, voltage below 0.1 and voltage swell are recorded and counted to determine the objective function value for this case as a reference. Voltage sag, voltage drop, voltage below 10% and voltage swell levels are selected according to IEEE standard (Jahromi et al., 2007; Bolen, 2000). At the same time GA starts encoding the IEEE 13 bus system by setting the bus 1 when DG is connected to the bus while setting the bus 0 when there is no DG connected to the bus. After that, GA selects two random locations for two similar and identical DG types and as follows: 1. Utilizing the same model of IEEE 13 bus built by PSCAD with the random locations selected by genetic algorithm, the two similar DG models are inserted in PSCAD. The model is running without any fault application and then the buses voltages are recorded to be taken into account and added to total number of buses during the fault conditions. Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12 4
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
Table 1 IEEE 13 bus encoding.
92 93 94 95 96 97 98 99
Bus
632
633
634
671
692
675
680
DG random location 1 DG random location 2
0 0
1 1
0 0
1 0
0 0
0 1
0 0
For accurate results each single phase bus is considered as one bus because single phase fault and three phase fault are applied. For example, the three phase bus 671 is considered as 3 buses. 2. Four variables are used to combine objective function, these variables are the overall number of buses experience voltage sag, the overall number of buses experience voltage drop, the overall number of buses experience zero voltage (voltage less than 0.1 p.u.), and finally the overall number of buses experience voltage swell. This objective function is applied on the case without DG to be a reference. In order to construct the objective function of different indices, weighting factors are needed to combine all the respective indices as proposed in this study and shown in Eq. (1). Fobj = a × Nsag + b × Ndrop + c × Nzero + d × Nswell
100
101 102 103 104 105 106 107
3.
108 109 110 111
4.
112 113 114
5.
115 116 117 118
6.
119 120 121 122 123 124 125 126
7.
127 128 129 130 131 132
8.
(1)
where Nsag is the overall number of single phase buses experience voltage sag; Ndrop is the overall number of 1 ph buses experience voltage drop; Ndrop is the overall number of buses experience zero voltage and finally, Nswell is the overall number of buses experience voltage swell. The weighting factors are a = 0.7, b = 0.1, c = 0.1 and d = 0.1. These factors are set according to the application and the power quality phenomena and can be changed correspondingly. These values are obtained from heuristic search. Different values are tried until reaching the suitable values. Calculate the four variables for the randomly selected DG locations and from these variables the objective function is calculated, then this value is compared to the reference value (the objective function of the case without DG inserted). If the objective function value is better than no DG case (lower than the case without DG) then this location is considered as a potential solution. An assumption is proposed, all the DG types are constant in size. Furthermore, that DG types are three phase because their size is 25% of the total load, such as 0.5 MW at voltage 4.16 kV. The loads are considered to be constant during the whole simulation process. Any solution violated constrains and boundaries will be ignored. In the proposed optimization technique, the number of solutions is dependent on the total number of generations, in which the number of generation depends on the total number of selected locations. Also, each bus is considered as location at which DG can be connected to this bus. For each solution, two identical DG models are connected to two buses simultaneously according to the genetic algorithm selection. In IEEE 13 bus system the number of three phase buses are 7 buses then the total number of solutions for two different DG sizes penetration are 2187 if two different DG sizes is selected because each DG size has state and when no DG has another state, the three states of DGs are 0 where no DG, 1 for the 1st DG and 2 for the 2nd DG. Since the possible solutions are 2187 then it is clearly concluded that the number of possible solutions is very large, this leads to very long time to find the optimum solution. To reduce the number of possible solutions and the solution time, the proposed limitations are considered. The selection of two identical DG size and type lead to only two states only, 0 where no DG and 1 when DG is inserted. The number of solutions in this assumption will be 128. The three phase buses as shown in Table 1 are, 632, 671, 692, 675, 680, 633 and 634. For more reduction for the number of possible solution in the search space, the bus 634 is ignored because it is low voltage not medium voltage bus. Then the number of possible solutions becomes 64. The actual randomly selected locations are at the locations 671 and 633, and the other locations are at 675 and 633 as shown in Table 1. The simulation is conducted without the application of any fault and all the buses voltage measurements are recorded to be accounted and added to the other buses voltage measurements during fault conditions. This is to combine the steady state operation and fault conditions during calculating the fitness value.
Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
5
Table 2 Objective function and the four variables.
133 134 135 136 137 138
139
Objective function
Nsag
Ndrop
Nzero
Nswell
Fobj
Franking
fitness noDG
310
40
31
77
231.8
0.829119
9. Single phase short circuit must be considered because it is approximately estimated 80% occurrence of all fault types (Bolen, 2000) as long as three phase short circuit too which is the most severe short circuit. Both fault types are applied to all possible buses as described in the case without DG previously in PSCAD. 10. The ranking value is calculated from Eq. (2) which created form the objective function. This value is used for ranking the solution for selection and reproduction the new offspring (Jahromi et al., 2007; Sedighzadeh and Rezzadeh, 2007; Alibabadi et al., 2008). Franking = 1 − (Fobj /Fobj-av )
(2)
162
In Eq. (2) Fobj-av is the sum of all the objective function for all solutions. The fitness function sets a fitness number to each solution and according to this number, the probability of selection of each solution for reproduction increases when this number increases. In this study, the calculated fitness value in the case without DG is used as threshold and is compared to the fitness of all solutions. The solution is considered a potential solution if its fitness value is higher than the threshold value. The fitted solutions are selected and preceded by the three genetic algorithm operators, crossover, mutation and selection. Each two solutions are exposed to the crossover operator at a percentage of 80% then the mutation operator is applied too to reproduce an offspring in the next generation (Jahromi et al., 2007; Kwang and El-Sharqawi, 2008). The offspring behaves as a new solution, the new solution is then added to the IEEE 13 bus, and this solution is implemented by inserting the two identical generators as two DGs. Simulating the new solution with PSCAD and repeating the short circuit step, then the objective function and the fitness function are applied to evaluate if the new solution is fit or not. These steps are repeated until reaching the optimum solution. The flow chart shown in Fig. 2 represents the optimization technique proposed by genetic algorithm. After finding the optimum location solution of synchronous generator as the first DG type, the second DG type is penetrated which is wind turbine in this study and the third DG type is penetrated also as PV distributed generations. The same genetic algorithm procedures are applied to the other types to find the optimum location solution of these DG types. It is highly important to investigate if the wind turbine or PV has an optimum location similar to synchronous generator or not and it provides better performance than the synchronous generator or not. The optimization technique utilized in this study to find the optimum location for three DG types and generate an output results is needed to be verified and to assure from these results are correct or not. One of the methods to verify or validate the results is to use another optimization technique such as PSO (Particle Swarm Optimization) technique. PSO is one of the evolutionary computation techniques (Alinejad et al., 2008; El-Zonokly, 2011). Matlab file is designed to implement PSO technique for finding the optimum DG location of the objective function in Equation 1. The number of iterations and number of swarm are defined for starting the PSO optimization technique.
163
3. Results and discussion
140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161
164 165 166 167 168 169 170 171 172 173
The simulated PSCAD file for IEEE 13 bus system is shown in Fig. 3. In this figure, the IEEE 13 bus was constructed from main generator 5 MVA 115 kV, main transformer (T1) 5 MVA 115/4.16 kV, second transformer 0.5 MVA 4.16/0.4 kV, different transmission lines configuration of different lengths, and different types of loads. Moreover, two identical synchronous generator models 0.5 MW, two identical wind turbines models 0.5 MW and finally two identical photovoltaic generator models 0.5 MW are modeled and inserted as DGs. These generators are considered as the distributed generation, each two identical distributed generation is selected to be 25% of the total load (Jahromi et al., 2007). The distributed generations are considered to be constant over the whole simulation period. All variables are calculated for all the SC (short circuit) cases at the normal operation condition without DG as shown in Table 2. From these variables the objective function is determined from Eq. (1) and the ranking value is calculated from Eq. (2) to be used as a reference. Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12 6
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
Fig. 2. Proposed optimization technique flow chart.
174
175 176 177 178 179 180 181 182 183 184
3.1. Optimization of synchronous DG 3.1.1. Using genetic algorithm Table 3 provides useful information for all the solutions, specially the offspring solution 671–680 which has overall number of buses experience voltage sag 77 is the same as offspring solution 671–675. On the contrary, the objective function of 671–680 is 71.6 while the solution 671–675 is 70.2. These offspring solutions can reproduce because they have ranking better than the reference. Looking carefully to Table 3 which shows that DGs penetration are highly essential due to the clear improvement of the performance compared to the reference case but attention must be taken into account because some location could lead to degrade the performance as shown in the location 632–645b. These results prove that the solution 633–692 is the best solution which provides the optimum solution and the best location. Then the synchronous generators provide better impact on the grid by improving the performance and mitigating the voltage sag. Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
7
Fig. 3. Proposed PSCAD for model IEEE 13 bus system. Table 3 The overall variables, objective function and fitness value of attained solutions.
185 186 187 188 189
DG locations
Nsag
Ndrop
Nzero
Nswell
Fobj
Franking
fitness noDG fitness DG633 671 fitness DG633 675 fitness DG633 680 fitness DG633 692 fitness DG632 671 fitness DG632 692 fitness DG632 645b fittness DG671 675 fittness DG671 680
310 52 69 73 51 313 305 360 77 77
40 20 38 21 20 47 47 8 30 32
31 12 43 34 12 8 14 35 5 5
77 158 292 278 159 10 12 0 128 140
231.8 55.4 85.6 84.4 54.8 225.6 220.8 256.3 70.2 71.6
0.829119 0.95916 0.936896 0.937781 0.959602 0.83369 0.837228 0.811058 0.948249 0.947217
3.1.2. Using PSO to validate the obtained results The objective function is used to be optimized and inserted in the PSO Matlab file taking into account the boundaries which are the case without DG threshold as an initial value. The number of generations is selected to be 20. Fig. 4 shows that there is convergence to the value 54.8 but still the average is 60.78. If the number of generations is increased, the PSO became more convergence to the best value 54.8. Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12 8
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
Fig. 4. PSO objective function when 20 generations.
Table 4 Voltage and losses of the heavy loaded buses without DG. Sending-receiving bus
Sending bus voltage (p.u)
Receiving bus voltage (p.u)
Active losses (MW)
Reactive losses (Mvar)
650–632 632–671 633–634 671–684
1.00 1.041 1.039 1.02
1.041 1.02 1.02 1.019
0.02 0.03 0.01 0.01
0.38 0.11 0.01 0.00
Table 5 Voltage and losses of the heavy loaded buses with DG.
190 191 192 193 194 195 196 197
Sending-receiving bus
Sending bus voltage (p.u)
Receiving bus voltage (p.u)
Active losses (MW)
Reactive losses (Mvar)
650–632 632–671 633–634 671–684
1.00 1.05 1.049 1.034
1.05 1.034 1.03 1.034
0.01 0.03 0.01 0.00
0.21 0.08 0.00 0.00
3.1.3. Impact on grid power losses For verifying the effect of DG on the network grid power losses, the proposed IEEE 13 bus network grid during normal operating conditions is modeled using Digsilent software. Both cases are simulated, the case when there is no DGs connected to the grid and when two DGs are connected to the solution 633–692. The results are shown in Table 4 for the case without DG while Table 5 shows the results when two identical synchronous generators as two identical DGs are connected to the grid. Both tables show the power losses and bus voltages for the heavy loaded buses. Comparing the obtained results in Tables 4 and 5, the results show improvement and reduction of the losses in case of DG usage. Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
9
Table 6 The overall variables, objective function and fitness value of attained solutions. DG locations
Nsag
Ndrop
Nzero
Nswell
Fobj
Franking
noDG DG633-671 DG633-675 DG633-680 DG633-692 DG632-671 DG632-692 DG671-675 DG671
310 302 308 302 305 300 309 309 331
40 8 3 6 16 60 3 3 21
31 32 35 34 29 29 32 32 1
77 131 91 161 83 63 92 92 81
231.8 228.5 228.5 231.5 226.3 225.2 229 229 242
0.888117 0.889709 0.889709 0.888261 0.890771 0.891302 0.889468 0.889468 0.883193
Fig. 5. PSO Algorithm score for wind turbine.
198
199 200 201 202 203 204 205 206 207 208
209 210 211 212
3.2. Optimization of wind turbine DG 3.2.1. Using genetic algorithm The results shown in Table 6 provides 9 DG locations, o671 degrades the network performance because it has Nsag 331 while the reference is 310. Only one DG is inserted in this solution to ensure that two penetrated DG are better than only one. The other DG locations improve the network performance because they have overall number of bus experience sag lower than the reference. The lowest overall number of buses experience voltage sag value is 302 for the location 632–671. It can be concluded that this location is different than the location for the synchronous generator. It can be observed that the best location is 632–671 and the objective function value is 225. If this location is compared the best location when synchronous generator is inserted as DG, it is found that the best location of synchronous generator is 633–692. Then, the DG type is highly effects the voltage sag mitigation and the best location that mitigate the voltage sag is dependent on the type of distributed generation. 3.2.2. Using PSO to validate the obtained results As mentioned in synchronous generator as DG, wind turbine is optimized using Particle Swarm Optimization technique. The number of generations is selected 100. Observing Fig. 5, it is found that the best score or objective value is 225.2 and the mean score value reached to 225.2 when the swarm go toward convergence. Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12 10
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
Table 7 The overall variables, objective function and fitness value of attained solutions. DG locations
Nsag
Ndrop
Nzero
Nswell
Fobj
Franking
DG633 noDG DG633 675 DG633 680 DG633 692 DG633 671 DG632 671 DG632 692 DG632 675 DG632-633 DG633
310 339 336 331 336 336 336 336 330 341
40 18 12 12 12 12 12 12 23 32
31 3 2 2 2 2 2 2 2 3
77 122 81 84 84 79 79 79 80 77
231.8 251.6 244.7 241.5 245 244.5 244.5 244.5 241.5 249.9
0.904981 0.896864 0.899693 0.901004 0.89957 0.899775 0.899775 0.899775 0.901004 0.897561
215
Comparing the proposed genetic algorithm technique in this study with the swarm convergence, the swarm converged to 225.2 while the minimum objective function is 225. Hence, it can be concluded that both techniques approximately provide the same results, but the PSO requires more generations for better convergence.
216
3.3. Optimization of photovoltaic DG
213 214
217 218 219 220 221 222 223 224 225
3.3.1. Using genetic algorithm Looking carefully to Table 7, it is found that the reference case has an objective function 231.8 and ranking 0.90498 which is better than all solutions. It means, when the PV as a DG is inserted to the grid, it degrades the network performance and increases the overall number of buses experience voltage sag, the reference Nsag is 310. But anyway, it can be found the best location that improve the performance than the other solutions to get the minimum objective function value and the minimum overall number of buses experience voltage sags. It is clearly obvious that both the locations 633–692 and 632–633 are the locations which have the minimum objective function value but the location 632–633 have overall number of buses experience voltage sag less than the location 633–692. It means that the best location is 632–633.
Fig. 6. PSO Algorithm score for Photovoltaic.
Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
11
231
3.3.2. Using PSO to validate the obtained results Optimizing PV using particle swarm optimization is implemented similar to the procedures applied to both synchronous generator and wind turbine. The number of generations and the number of iterations are defined. Fig. 6 shows the best score of the objective function. Comparing the proposed genetic algorithm technique above with the swarm convergence, it can be concluded that both techniques approximately provide the same results.
232
3.4. Comparison between the impact of the three DG types on the grid
226 227 228 229 230
237
It is highly important looking to the synchronous generator as DG and the wind turbine compared to the PV as DG to ensure proof that the best location is dependent on the type of DG. It is noted that synchronous generator as DG provides better impact on the grid performance than wind turbine and PV at totally different location. At the same time, the best objective function value of photovoltaic when selected as DG is 241.5 provides the poorer performance on the grid than synchronous DG and wind turbine DG.
238
4. Conclusion and future work
233 234 235 236
251
DGs locations are needed to be optimized efficiently to improve the electrical network performance and to avoid the degradation of the network function using optimization technique like genetic algorithm and Particle Swarm Optimization. In This study genetic algorithm is used to optimize the location of three types of DG. The three proposed DG models which are simulated using PSCAD software are synchronous generator, wind turbine and photovoltaic. Each model of these types were inserted separately to investigate its behaviors on the grid precisely because it was found that each model provide different impact on the grid. The results showed that synchronous generator is mitigating the sag and enhance the grid performance better than wind turbine and photovoltaic while wind turbine is better than photovoltaic. PV as discussed degrade the voltage sag mitigation but finding the best location which make the performance better than the others is helpful because all the countries all over the world looking for more utilization of clean and sustainable energy specially the power generated from photovoltaic sources. In the mean time, PSO approach shows approximately similar results which validate that the GA proposed approach generates correct results. It is suggested in the future to combine two or three different DG types and monitor their response on the grid and optimize their location to recognize the best location and impact on the grid when they are integrated in the same grid.
252
References
239 240 241 242 243 244 245 246 247 248 249 250
253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
Alibabadi, M., Behbahani, B., Jalilvand, A., 2008. Combination of GA and OPF for allocation and active and reactive power optimization in distributed generation units. In: International Conference on Power and Energy (PECon), 1–3 Dec 2008, Johor Bahru, Malaysia, pp. 1541–1544. Alinejad, Y., Sedighizadeh, M., Sedighi, M., 2008. A Particle swarm optimization for sitting and sizing of distributed generation in distributed network to improve voltage profile and reduce THD and losses. In: 43rd International Universities Power Engineering Conference UPEC2008, 1–4 September 2008, Padova, Italy, pp. 1–5. Bolen, M., 2000. Understanding Power Quality Problems Voltage Sag and Interruptions. IEEE Press series on power engineering 2000, 2nd ed. Wiley IEEE Press. Collins, E., Jiang, J., 2008. Voltage sags and the response of a synchronous distributed generator: a case study (2008). IEEE Trans. Power Deliv. 23, 442–448. Dhas, G., Prakash, T., 2011. A novel approach for voltage sag mitigation using FACTS device interline dynamic voltage restorer. In: 2nd International Conference on Electronics Computer Technology ICECT, 8–10 April 2011, Kanyakumari, India, pp. 37–41. El-Zonokly, A., 2011. Optimal placement of multi-distributed generation units including different load models using particle swarm optimization. Swarm Evol. Comput. 1, 50–59. Farag, H.E.Z., Elsaadany, E.F., Seethapathy, R., 2012. A two ways communication-based distributed control for voltage regulation in smart distribution feeders, smart grid. IEEE Trans. 3, 271–281. Gozel, T., Hocaoglu, M., Eminoglu, U., Balikkci, A., 2005. Optimal placement and sizing of distributed generation on radial feeder with different static load models. In: 2005 International Conference on Future Power System, 18–18 November 2005, Amsterdam, Netherlands, pp. 2–6. Jahromi, M., Farajah, E., Zolghardi, M., 2007. Mitigation voltage sag by optimal allocation of distributed generation using genetic algorithm. In: 9th International Conference on Electrical Power Quality and Utilization, 9–11 October 2007, Barcelona, Spain, pp. 1–7. Kamalinia, S., Afsharnia, S., Khodayar, M., Rahimikian, A., Sharbafi, M., 2007. A combination of MADM and Genetic Algorithm for optimal DG Allocation of Distribution System. In: 42nd International Conference Universities Power Engineering Conference (UPEC), 4–6 September 2007, Brighton, UK, pp. 1031–1035.
Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001
+Model JESIT 46 1–12 12 275 276 277 278 279 280 281 282 283 284 285
ARTICLE IN PRESS A.M. Abd-rabou et al. / Journal of Electrical Systems and Information Technology xxx (2015) xxx–xxx
KwangF L., El-Sharqawi, M., 2008. Modern heuristic optimization techniques. IEEE Press series on power engineering 2008, 2nd ed. Wiley IEEE Press. Martine, C.R., Renato, G., Arturo, S.B., Roberto, C.L., 2011. System unbalance and fault impedance effect on faulted distribution network. In: 3rd Global Conference on Power Control Optimization, Computer Mathematics with Applications, Vol. 60, Elsevier, August 2010, pp. 1105–1114. Martinez-Velasco, J., 2007. Distributed generation impact on voltage sags in distribution networks. In: 9th International Conference on Electrical Power Quality and Utilization EPQU, 9–11 October 2007, Barcelona, Spain, pp. 1–6. Sedighzadeh, M., Rezzadeh, A., 2007. Using genetic algorithm for distributed generation allocation to reduce losses and improve voltage profile. In: IEEE, UPEC2007, Sustainable Power Generation and Supply, 6–7 April 2009, Nanjing, China, pp. 1–6. Song-cen, W., Kun-shan, Y., Guang-fu, T., 2008. Mitigation of voltage sags by grid-connected distributed generation system in series and shunt configuration. In: Power System Technology and IEEE Power Indian conference Joined with International Conference POWERCON, 12–15 October 2008, New Delhi, India, pp. 1–8.
Please cite this article in press as: Abd-rabou, A.M., et al., Impact of DG different types on the grid performance. J. Electr. Syst. Inform. Technol. (2015), http://dx.doi.org/10.1016/j.jesit.2015.04.001