4th International onference on Mechatronics, Materials, hemistry and omputer Engineering (IMME 05) A practical Shunt apacitor Placement Algorithm in Distriution Netork Zhilai Lv, a, Wei Wang,, Hai Li, c, Dejin Wang, d, Zifa Liu, e and Yunpeng Wang, f Being XUJI Electric o.,ltd., Being 00085, hina; School of Machine and Electric Engineering, Shandong Science and Technology University, Taian 709, hina; School of Electrical and Electronics Engineering, rth hina Electric Poer University, hangping District, Being 006 a helen_lr@6.com, angei-dongfang@6.com clihai@jxj-xjgc.com, d angdejin@sina.cn, e764896@qq.com, fhrang@sina.com Keyords: distriution netork, capacitor shunt placement, matching flo, sensitivity, cycle compensation method. Astract. For capacitor shunt placement the proposed methods do not consider the folloing conditions; there are fe measures and the first compensation points are easily over compensation in distriution netorks. So this paper proposes a practical method that can full of utilize the availale measures and avoid over compensation for the first compensation point. First, the non-measured measures are evaluated y the improved matching flo. Secondly, the step-y-step re programming method ased on sensitivity is used to find compensation location and capacity. Finally, the cycle compensation method is used to eliminate over compensation and etter compensation results are get. The practical 85 nodes feeder verifies the accuracy and practical for the proposed algorithm.. Introduction Research on capacitor shunt placement in distriution netorks is of great realistic and theoretical significance. It can reduce the line loss effectively, improve the economic enefit. The prolem is a nonlinear integer programming troule in mathematics. If compensation point numer, location, and compensation capacity are simultaneously optimized, the search space is huge and easy to fall into the curse of dimensionality in distriution netorks. Therefore, the method of the present study is mainly to find location firstly, and then determine the compensation capacity. That is uses the method of sensitivity [-], re secondary precise moment [], the load impedance [4] and hierarchical clustering method [5] to select compensation point. Then the numer of compensation groups is calculated y using the method of mathematical programming or intelligent optimization algorithm. The disadvantage of sensitivity method is easily lead to over compensation, and it is difficulty to deal ith the mutual influence eteen the compensation locations. Reference [6] considers the influence eteen the compensation points y second order netork loss re sensitivity matrix to a certain extent. But the compensation capacity and test method for choosing node size ill affect the study results. This is difficult to avoid in a separate source and capacity optimization. Re secondary precise moment, the load impedance or methods of hierarchical clustering method can avoid the influence of virtual high sensitive nodes to a certain extent. But they still rely on the re compensation efore netork tide. Dynamic impact of netork tide and sensitivity after the installation of re compensation capacity are not considered. The numer of nodes and numer of partitions are not easy to determine. Finally, the proposed methods are usually assumed that the load is measured at the maximum, average and minimum load conditions. But the aove data is difficult to e given y the distriution netork company that affects the practicaility of the algorithm. 05. The authors - Pulished y Atlantis Press 005
For dealing ith the aove prolems, this paper eliminates the over compensation phenomenon and proposes a practical parallel compensation method for distriution netork.. Improved matching flo According to Pseudo measurement flo calculation ased on non measurement load of matched flo calculation, such as the ranch, current and node voltage value can e calculated. The smaller the error eteen the calculated values and the measured values, the more likely that the estimate is more close to the actual value. The ojective function of the pseudo measurement estimation of the non measurement nodes ith the minimum error is given as elo: min ε = + np nv i= V, i P V i, P V + p,, =,, V, i I, P, Q, i + ni nq = = Q, I, I Q,, I Q here, are the eight of the real-time measurement variales of the node voltage amplitude, ranch current amplitude, ranch and re respectively; n, n, n, n are the numers of nodes ith real-time measurement of node voltage amplitude, ranch V I P Q current amplitude, and re of the ranch respectively; V, V are the i, i voltage amplitude flo calculation and real-time measurement of the node i respectively; I, I are, the ranch current amplitude flo calculation value and real-time measurement value of node numer i, j respectively; P, P 和 Q, Q are the ranch, re flo,, calculation and real time measurement value of the node numer i, j respectively. The equation aove also satisfies the flo, ranch current and node voltage constraints. The flo algorithm of improved matching algorithm is presented: Initialization, the admittance matrix is formed according to the netork topology structure; alculate the load pseudo measurement y using the first - end coefficient method, and calculate the distriution coefficient of the load mismatch α, β ; The root node of the average voltage as the node voltage, the amount of mismatch P, Q = 0 ; the numer of iterations k = ; Σ Σ m m 4 Regard P + α P and Q + α QP as the ne load pseudo measurement data; Di i Σ Di i Σ 5 The voltage of each node is calculated y a flo calculation; 6 The loss of the root node is P, Q according to the node voltage; Σ Σ 7 Determine hether the value of the target function type () is less than the given value, if it is less than the given value, go to step 8, otherise, go to step 4 8 alculate the pseudo measurement value of the non real time measurement load. Through the aove algorithm, e can make full use of all kinds of real time measurement. The algorithm can improve the accuracy of the estimation of the non measurement load.. Ojective function for shunt capacitor placement This paper takes the maximum ratio of investment and enefit as the target. max normal min ( DP t + DP t + DP t ) K max normal min D Max F = () Nc n γ ( + γ ) ( Q K ) + N K n c op = j ( + γ ) max normal min Where P, P, P are saving the loss in the largest, the general, the minimum load condition respectively, t, t, t are the year running time in the largest, the general, the minimum max normal min load condition respectively; K is the electricity price; N is the group numer of compensation D c capacitor; Q is the capacity of capacitor; is the single group capacity cost for compensating 006 ()
capacitor; capacitor; K is the compensation group numer; γ is the discount rate; n is the operation time of is the operation cost of single group capacitor. op The equation aove also satisfies the flo, ranch current and node voltage constraints. 4. Sensitivity method The folloing method of sensitivity only considers the relationship eteen the re component of the current and the loss of the netork. P iopt = ( t I R ) j j j = t I R () j j j = t R j j = Where I is the re component of the current efore is incorporated into the ranch j;if the j ranch j is on the path of the node i to the root node, t = ± ; if the ranch j is not on the path of the node i to the root node, t = 0 ; If the direction of the road j is the same as the road, t =,else t = ; R is the resistance of the ranch j; is the ranch numer. j According to the formula (), the changes of the loss after the parallel units re can e calculated quickly. Then it can fast calculate the largest sensitivity of nodes. 5. Loop compensation method for shunt capacitor placement In order to get a greater investment returns ratio, the specific algorithm flo is shon in Figure. Make the compensation node i=, the ratio of maximum investment enefit Fmax= According to the second section method, estimate the non measurement load pseudo measurement According to the. section method, calculate target function F and the numer of compensation node N F>FmMx compensation point, end Record the location and capacity of the compensation node Fmax=F Reduce a set of compensation capacity of i compensation node The compensation capacity of less than i is constant, and calculate the target function and the numer of compensation nodes according to the. section. F>FmMx Record the location and capacity of the compensation node, Fmax=F i=i+ i N Output the location and capacity of the compensation node hen Fmax, end Fig. Flo char of algorithm 007
6. Examples and analysis As is shon in Figure, the feeder has 85 nodes, root node voltage is 0kv. Active supply of a year is 0700000kh. Maximum load of a year is 5800k. The current of the feeder load 7 is.6ka. The current of the feed line 5 is.9ka under the maximum load condition. Feeder line model is LGJ-0. The planning period is 5 years. The price is 0.5 / degree. The discount rate is 0.. Parameters of the system can e find in reference[7]. 7. 50 6 49 7 5 5 54 5 8 4 77 4 4 44 45 46 47 48 4 40 76 6 4 4 5 8 75 85 40 7 60 8 9 0 6 4 5 69 74 6 7 8 9 0 4 5 6 7 4 5 6 55 9 56 57 58 6 0 64 5 6 65 66 67 68 70 77 7 5 78 80 8 8 79 6 4 5 84 9 0 59 7 8 7 9 6 7 8 9 0 8 5 4 6 7 8 9 Fig. 85 nodes practical feeder The improved matching flo results are shon in Tale. Ta. n-measured measures estimated values( unit k, re unit kvar) numer re numer re numer re 60. 08. 6 79.7 55. 9. 8. 79.6 55.8 7 6.7 0.8 5. 5.6 79.8 55.4 8 65.7 7. 7.. 4 6.9 0. 9 80.8 55.4 4 8. 55.7 5 80. 56.5 0 6.4 9 5 9.7 88.4 6 4.5 7.9 65.8 44. 6 6.8 4.6 7 65.4 4.4 4.7 8. 7 64. 08.8 8 9.7 8. 4. 6. 8 58.9 05.6 9 4. 8. 4 8.6 88.8 9 5.9 0.6 0 60.9 09.8 5 6.4 4.9 40.9 86.5 60.4 6. 6 6.. 4 9.9 95.4 8.5 54.8 7 4.7 6.5 4 6.6 06. 60.4 74.6 8 8. 56.4 4 58.6.5 4 69.0 05.9 9 77.4 58.7 44 65. 4 5 09 69.8 0.7 89.6 45 8. 59 According to Tale, the pseudo measurement are shon in Tale. Ta. Results of var optimal programming ompensation method location ompensation capacity (kvar) The method of this paper de 7,79 600,600 Sensitivity method de 7,80 000,50 From tale, it can e concluded that the sensitivity method is easy to cause first compensation node over compensation. 8 9 0 008
Tale gives the comparison of the main economic and technical indexes of the method of this paper and the method of sensitivity. Ta. Index of economic and technology Investment enefit ratio Line loss rate the loest node voltage (kv) The method of this paper.9% 9.64 Sensitivity method.9.95% 9.57 From tale, it can see that this method can e improved oth in economy and technology compared ith the sensitivity method in re compensation. 7. Summary Parallel compensation method for capacitor in distriution netork is presented in this paper. Example results sho that the proposed algorithm improves the economic and technical indicators of re compensation. The algorithm has strong practicaility and accuracy. The algorithm is entirely easy to develop practical re optimization of distriution netork planning softare. The method can increase more the utility's economic enefit than the exciting sensitivity method. References []. Baran M E, Wu F F. Optimal capacitor placement on radial distriution systems[j]. IEEE Trans. Poer Delivery, 989, 4(): 75-74. []. Baran M E, Wu F F. Optimal sizing of capacitors placed on radial distriution systems[j]. IEEE Trans. Poer Delivery, 989, 4(): 74 74. []. Zhang Tingchang, Geng Guangfei. Re optimization for medium voltage distriution netork ased on improved particle sarm optimization. Poer System Technology 0, 6(): 58-6(in hinese). [4]. Jiao Zhihong-ai, Zhong qin, Guo Zhizhong. Accurate moment method for optimization of capacitors in radial distriution systems[j]. Relay, 00, 0(9): -4(in hinese). [5]. Yan Wei, Xu Zheng. impedance moment method for optimal location of re compensation on 0 kv Feeder[J]. Proceedings of the hinese Society of Universities for Electric PoerSystem andautomation, 005, 7(5):9-(in hinese). [6]. Wang Shao, Zhou Xin. Optimization of re in distriution netork ith hierarchical clustering and ant colony algorithm. Poer System Technology,0, 5(8):6-68(in hinese). [7]. Zhu Jingmin. Study on distriution netork re optimization allocation. Master. Shandong Science and Technology University. hian, 05.4-45. (in hinese). 009