Minimization of Network Losses by Optimal Placement and Sizing of DG using Flower Pollination Algorithm

Transcription

1 International Research Journal of Engineering and Technology (IRJET) e-issn: Volue: 5 Issue: 12 Dec 21 p-issn: Miniization of Network Losses by Optial Placeent and Sizing of DG using Flower Pollination K. Mareesan Lecturer (Sr.Grade)/EEE, VSVN Polytechnic College, Virudhunagar, Tailnadu, India *** Abstract - Reliable and Quality power supply to the consuer is the ajor objectives of power syste planner. Recent trends in the power syste infrastructure enhanceent, increases the efficiency in the syste operation and control. One aong the is the advanceent in the usage of Distributed (DG) which helps to provide ore flexible power transaction within the syste utilities. The proposed schee in this paper explored the features of Distributed in achieving the objective of network loss iniization by optially sizing and placing the DG within the security liits. Nature based Flower pollination algorith () has been applied to extract the best exploration and exploitation balanced solution. The proposal has been tested with the two radial distribution syste networks such as 33 and 9 bus systes with Unity and Variable Power Factor(VPF) DG. Solution based coparison of with the standard Genetic () and Particle Swar Optiization() algorith depicts the syste stability by reducing the network loss at the greater extent. Keywords: Distributed (DG), Flower Pollination (),,, Variable Power factor (VPF) 1. INTRODUCTION Effective production and aintenance in ters of supplying sustainable power supply to the custoer utility is the ain focus of a power systes planner. With the advent of sart technologies in the power syste planning and control, the structure of power syste has been reodelled and invited ore nuber of arket participants in electricity arket. The copetiveness in the electricity arket akes the syste utilization at the higher level which akes the syste security at the ost vulnerable position. Hence there is a great challenge in aintaining power balance during the load variations. The optial usage of distributed generations (DG) helps to provide solution in achieving greater power balance in ters of generation, transission and distribution of electricity in the power syste network. Thanks to Griffin et.al and Ackerann et.al [1-2] for adopting the distribution generation (DGs) in a power syste which will resolve any power shortage proble by directly connecting the DGs to distribution network or to custoer side (Load Center). It also iproves the following paraeters of power syste network such as power quality, continuous power supply and security, constant bus voltage profile, voltage stability during load variations. Recent years there are larger possibilities of renewable and non- renewable DGs of different capacities such as solar, wind, diesel, gas and icro turbines are available for power integration however the benefit depends on the location and sizing of the DGs. Wang et al. (24) have shown that iproper allocation not only costs heavier power loss but also collapsed the syste operation [3]. It is andatory to choose the optial location as well as sizing for the reliable operation. DGs placeent and sizing is considered as non-linear prograing for iniizing the power losses. There are ultitudes of classical literatures [4-5] available for analytical based solution for this proble Das et al., Willis et al. and Acharya et al. are noted aong the. Introduction of eta-heuristic algoriths for coputing coplex power syste proble have invaded in the region of optial location and sizing proble. Acharya et al. [] and Zhu et al. [7] have approached this proble using popular nature inspired algorith. Evolutionary prograing (EP) has been exploited by the De sauza et al. [], Cano have used Fuzzy systes[9], Ant Colony optiization (ACO) has been proposed by Favuzza et al. [1], Srinivasa rao et al. have ade use of the plant growth siulation [11], Bee Colony optiization algorith (BCO) as a tool for solving optial DG allocation and sizing proble has been exploited by Abu-outi et al. [12]. A hybrid algorith of with has been utilized for optial placeent of DG Moradi et al. [13]. Miniizing the technical aspects of power losses [14] while iproving the voltage profile is achieved independently by Abou et.al, Khalesi et al. [15]. In this paper, it is proposed to iniize the network losses by optial placeent and sizing of DG using a nature inspired Flower Pollination () in distributed syste. The effectiveness of the algorith is tested using standard IEEE 33 and 9 bus radial distribution syste networks. To attain the real tie syste results, DGs with various power factors are also considered and the results are recorded. The effectiveness of this algorith is copared using the test results taken fro standard and algoriths. 2. MATHEMATICAL FORMULATION 2.1 Objective Function: Best solutions of iniizing the real power loss of the distribution syste are extracted by eans of forulating the proble as per detail given below. Miniization of (F) is the Objective function. 21, IRJET Ipact Factor value: ISO 91:2 Certified Journal Page 599

2 International Research Journal of Engineering and Technology (IRJET) e-issn: Volue: 5 Issue: 12 Dec 21 p-issn: The network loss of the syste is defined as follows Where is the i th generator bus real power, is the i th bus real power load deand value, and are the i th bus connected branch s sending and receiving end voltages, and are the adittance agnitude and phase angle of the i th bus connected branch, are the receiving and sending end voltage angle of the i th bus connected branch. NB is the total nuber of bus in radial distribution syste (). Conventional real power loss is obtained using Newton Raphson s power flow ethodology Constraints The above single objective proble needs to be solved by satisfying the following static operational liit values. Constraint I: Voltage Liit Constraint Constraint II: DG Capacity Constraint (2) (3) (4) Reactive power forulation of the distribution generation fro the real Power is given as below Constraint III: Power balancing Constraints (5) (7) Where and are the iniu and axiu perissible voltage at bus, and are the iniu and axiu perissible real power value of each DG capacity, and are the real and reactive power fro DG in p.u, and are the iniu and axiu perissible power factor value of each DG capacity, and are the real and reactive power load deand in p.u, and are the voltage agnitude at the and bus in p.u, and are the voltage angle at the and bus in p.u, N is the total nuber of bus in distributed syste, is the agnitude of adittance atrix and is the angle of adittance atrix between buses in p.u. 2.2 Optiization structure The objective of achieving the iniised network loss in the Distribution Syste (DS) with the optial allocation of pre-deterined nuber of DG s without violating the syste operation liits as given in the previous section, Stochastic algoriths such as Genetic (), Particle Swar Optiization () and Flower Pollination () are used to solve this optiization proble. The structure of the optiization algorith has been explained as follows Flower Pollination structure Flower Pollination () is a nature inspired stochastic algorith recently cae to lie light for solving hard nonlinear optiization probles. It is proposed by Xin-She Yang in 212, is based on flower pollination behaviour. Pollination ay occur fro pollen of sae flower or of different flowers in sae plant or between flowers of different genus, the forer is known as self-pollination, and the latter is known as cross pollination. Biotic pollination known as global pollination initiated by the bats, bees, birds, insects and flies which can fly a long distance. So it has characteristics of Levy flight jups as Levy flight distribution, one of the control paraeter. The pollination process is dictated by four andatory rules such as 1) Global pollination is biotic and pollen carrying agents do the Levy flights 2) Selfpollination is abiotic and tered as local pollination 3) Reproduction probability is attributed to flower constancy and depends on the siilarity of the two flowers involved 4) p [, 1], is assued as switch probability which controls both local and global pollination. The pollination and reproduction of the fittest is given as below () (s s o ) (9) Where v i t is the pollen i or solution vector v i at iteration t, and d * is the current best solution found aong all solutions at the current generation/iteration. L is the strength of the pollination, which is a step size. Pollinators can ove over a long distance with various distance steps known as Levy flight distribution. Γ(µ) is the standard gaa function, and this distribution is valid for large steps s >. In all our siulations, we have used µ = 1.5 and s [, 1]. L > is assued for Levy distribution. The local pollination and flower constancy can be represented as ( ) (1) Where, v j t and v k t are pollens fro the different flowers of the sae plant species. This essentially iics the flower constancy in a liited neighbourhood. Matheatically, if v j t and v k t coes fro the sae species or selected fro the sae population, this becoe a local rando walk if 21, IRJET Ipact Factor value: ISO 91:2 Certified Journal Page

3 33 BUS No per International Research Journal of Engineering and Technology (IRJET) e-issn: Volue: 5 Issue: 12 Dec 21 p-issn: we draw ε fro a unifor distribution in [, 1]. Pollination activities in algorith can occur at both local and global scale. Flowers at closer proxiity are ore likely to pollinate copared to flowers at distance, so switch probability or proxiity probability p to vary between p =.5(initially) to..the value of p =. works better for ost applications. The flow chart of is given in fig.1 radial distribution test syste has 32 branches with the total syste load of 3.72 MW and 2.3 MVAR. The initial syste power loss of the 33 bus syste is 21.9 and the reactive power loss is 143 kvar. The syste voltage base is 12. kv and the base apparent power is 1 MVA. The second test syste is the 9 bus radial distribution test syste with the total syste load of 3. MW and 2.9 MVAR. The initial syste power loss of the 9 bus syste is and the reactive power loss is 12.1 kvar. Both 33 bus and 9 bus radial distribution syste data are taken fro the literature by Moradi et al. (212)., and based optiization algoriths are used to extract the best optiized solution by eans of finding best distributed generation site and sizing. Three best locations for DG placeent have to be identified and the best sizing for each DG should be less than 2 KW has been assued. In this proposed work, DG power generation with both y and variable leading power factors are optiized to achieve the reduced network loss. In DGs with variable power factor optiization, the power factor is also taken as control paraeter along with location and DG sizing. MATLAB siulation of, and algoriths for the 33 and 9 bus test syste have been conducted with the DG s y and variable operational power factors. The siulation results have been recorded and the details are provided below with the explanation. 3.1 Siulation results of single objective proble forulation with y power factor DGs The ost excellent optiu location and sizing to extract the best real power loss is identified by eans of ipleenting the three algoriths such as, and. The siulation results of the three algoriths are recorded and it is provided in the Table 1 and Table 2 for the 33 and 9 bus test syste respectively. Table 1, and based optiization results for 33 bus syste Fig1. Flower Pollination Flow chart 3. RESULT AND DISCUSSION The proposed single objective forulation has been ipleented with the help of two standard radial distribution test systes 33bus and 9bus. The 33 bus , IRJET Ipact Factor value: ISO 91:2 Certified Journal Page 1

4 Syste Miniu Mean Maxiu Standard Deviation No of Iteration for convergenc e p.u p.u 9 BUS No per International Research Journal of Engineering and Technology (IRJET) e-issn: Volue: 5 Issue: 12 Dec 21 p-issn: Table2., and based optiization results for 9 bus syste It is inferred fro the Table 1 and Table 2 that the provides best iniized network loss (F) in 33 bus and 9 bus test radial distributed systes. The provides better solution copared to the and Optiization algorith s perforance easures The statistical easures such as ean, best, standard deviation of the three algoriths for the two test systes are recorded in the Table 3 by conducting 2 different trials. It is inferred fro the statistical perforance Table 3 that the optial solutions of for the two test syste is better copared to and and also the nuber of iterations for extracting best solution using is better than copared to and for all the test syste. The above is obvious fro the three algorith s convergence graph in Figure 2 and Figure 3 for the 33 bus and 9bus respectively. The solution of the has been copared with that of the cobined / based ethodology proposed in the literature by Moradi et al. (212). It is inferred fro the result that the based solution provides better extraction of iniized network loss copared to the cobined / based ethodology. The above percentage coparison of real power loss fro the base case for the based proposed proble forulation and the cobined / based proble forulations are depicted in the Figure 4. Table 3. Statistical easures of, and Fig.2., and based optiization convergence graph for 33 bus Fig3., and based optiization convergence graph for 9 bus , IRJET Ipact Factor value: ISO 91:2 Certified Journal Page 2

5 3 3 B No per Fitness Fitness 9 BUS % of Loss reduction fro the base case No per International Research Journal of Engineering and Technology (IRJET) e-issn: Volue: 5 Issue: 12 Dec 21 p-issn: Real Power loss coparison Fig.4. Cobined / and based real power loss(reduction) coparison 3.2 Siulation results of single objective proble forulation with variable power factor DGs The proposed single objective proble is solved by finding the optial sites and their respective sizing of DG with the help of finding optial power factor within the specified liit. DGs with the variable power factor based optial sizing and location to achieve the best optial solutions are extracted and it is recorded in the Table 4 and Table 5 for the 33 and 9 bus syste respectively. The convergence graph for the three optiization algoriths siulations are depicted in the figure 5 and. It is inferred fro the graphs that the based optial solution extraction solution converges faster copared to that of and. It is inferred fro the table 4 & 5 that based siulation provides the iproved solutions copared to that of and siulations. The statistical easures such as ean, best, standard deviation of the three algoriths for the two test systes are recorded in the Table by conducting 2 different trials. It is inferred fro the Table that the optial solutions of for the two test syste is better copared to and and also the nuber of iterations for extracting best solution using is better than copared to and for all the test syste. Table 4, and based optiization results for 33 bus syste 9 syte / Table5., and based optiization results for 9 bus syste Fig. 5 Variable power DG based optiization Convergence graph for 33 bus syste Fig. Variable power DG based optiization Convergence graph for 9 bus 21, IRJET Ipact Factor value: ISO 91:2 Certified Journal Page 3

6 % of loss reduction fro the base case network loss Syste Miniu Mean Maxiu Standard Deviation No of Iteration for convergenc e % of loss reduction fro the basecase network loss International Research Journal of Engineering and Technology (IRJET) e-issn: Volue: 5 Issue: 12 Dec 21 p-issn: Table. Variable power factor DG based optiization algorith s statistical easures Variable and y power factor DGs optiization solution coparison It is obvious fro siulation results that the real power loss solution extracted fro variable power factor DG optiization gets largely iproved when coparing with the y power factor DG optiization. The percentage iproveent of reduction in power loss fro the base case for the 33 and bus test systes are depicted in the fig.7 and fig.. DG with Unity operational Power factor Real Power loss coparison graph DG with variable operational Power factor Fig.7 Variable power factor DG and y power factor DG based real power loss coparison for 33 DG with Unity operational Power factor DG with variable operational Power factor Fig. Variable power factor DG and y power factor DG DG based real power loss coparison for 9 bus 4. CONCLUSION This paper projects the application of DG in extracting the network loss iniization by optially deterining the placeent and sizing of DG with y and variable power factor by considering the syste security liits. Fro the siulation results, it is evident that the proposed optiization proble extracts the better solutions. Also it is concluded that the application of in solving the proposed optiization provides the ost definite solution by aintaining the good balance in exploring the global inia and exploiting the local inia. REFERENCES Real Power loss coparison graph [1] Griffin, Tosovic, et al. "Placeent of dispersed generation systes for reduced losses." Syste Sciences, 2. Proceedings of the 33rd Annual Hawaii International Conference on. IEEE, 2 [2] Ackerann, T, Andersson, G & Soder, TF 21, Distributed generation: a definition, Electric Power Systes Research, Elsevier. [3] Wang, Caisheng, and M. Hashe Nehrir. "Analytical approaches for optial placeent of distributed generation sources in power systes." IEEE Transactions on Power systes 19.4 (24): [4] Das, Do, D. P. Kothari, and A. Kala. "Siple and efficient ethod for load flow solution of radial distribution networks." International Journal of Electrical Power & Energy Systes 17.5 (1995): [5] Willis, H. Lee, ed. Distributed power generation: planning and evaluation. CRC, Press, 2. 21, IRJET Ipact Factor value: ISO 91:2 Certified Journal Page 4

7 International Research Journal of Engineering and Technology (IRJET) e-issn: Volue: 5 Issue: 12 Dec 21 p-issn: [] Acharya, Naresh, Pukar Mahat, and Nadarajah Mithulananthan. "An analytical approach for DG allocation in priary distribution network." International Journal of Electrical Power & Energy Systes 2.1 (2): 9-7. [7] Zhu, Ji Zhong. "Optial reconfiguration of electrical distribution network using the refined genetic algorith." Electric Power Systes Research 2.1 (22): [] De Souza, Benear A., and Joao MC De Albuquerque. "Optial placeent of distributed generators networks using evolutionary prograing." Transission & Distribution Conference and Exposition: Latin Aerica, 2. TDC'. IEEE/PES. IEEE, 2. [9] Cano, Edwin B. "Utilizing fuzzy optiization for distributed generation allocation." TENCON IEEE Region 1 Conference. IEEE, 27. [1] Favuzza, Salvatore, et al. "Optial electrical distribution systes reinforceent planning using gas icro turbines by dynaic ant colony search algorith." IEEE Transactions on Power Systes 22.2 (27): [11] Rao, R. Srinivasas, S. V. L. Narasiha, and M. Raalingaraju. "Optial capacitor placeent in a radial distribution syste using plant growth siulation algorith." International journal of electrical power & energy systes 33.5 (211): [12] Abu-Mouti, Fahad S., and M. E. El-Hawary. "Optial distributed generation allocation and sizing in distribution systes via artificial bee colony algorith." IEEE transactions on power delivery 2.4 (211): [13] Moradi, Mohaad Hasan, and M. Abedini. "A cobination of genetic algorith and particle swar optiization for optial DG location and sizing in distribution systes." International Journal of Electrical Power & Energy Systes 34.1 (212): -74. [14] El-Ela, AA Abou, S M. Alla, and M. M. Shatla. "Maxial optial benefits of distributed generation using genetic algoriths." Electric Power Systes Research.7 (21): [15] Khalesi, N., N. Rezaei, and M-R. Haghifa. "DG allocation with application of dynaic prograing for loss reduction and reliability iproveent." International Journal of Electrical Power & Energy Systes 33.2 (211): BIOGRAPHY K.MAREESAN (Author ) is currently working with VSVN Polytechnic College, Virudhunagar, Tailnadu, India. His area of interest are Power Syste in deregulation environent and Electrical Machines. 21, IRJET Ipact Factor value: ISO 91:2 Certified Journal Page 5