National Power & Energy Conrerelice (PECon) 2004 Proceedings, KuaIa Lumpur, Malaysia 41

Static Var Compensator Planning Using Artificial Immune System For LOSS’ Minimisation And Voltage

Improvement S. Jshak, A. F. Abidiii and T. K. A. Rahinan

Abstract--Loss minimization in power system is an important colisideration research issue. Transmission losses in a power system call bc minimized by incans of reactive power compensation. Installing Static Var Compensation (SVC) in a power system has known to be able to improve voltage level i n the system and hence minimizing the system losses. This paper presents a technique to determine the location of the SVC in order to minimize loss in the system. The performance of this technique is tested using 14 buses IEEE Reliability Test System. A load flow programmed writtcn in M A T L A B by using Artificial Immune System (AIS) tech’nique was used to compute power tlow. The test result shows that the location and sizing of the SVC identified by the proposed technique Inns been able to improve voltage level of the system and also mlnlnilze the losses.

Keyivords--Static Var Compensator (SVC), Artificial Immune System (AIS), loss minimization and voltage improvenicnt.

I. INTRODUCTION

Most large power system blackouts, which occurred worldwide over the last twenty years, are caused by heavily stressed system with large amount of real and reactive power demand and low voltage condition. Low limit may be exceeded due to the voltage draining off VAR rcserves [I] . When the voltages at system buses are low, the losses will also be increased. This study is devoted to develop a technique for improving the voltage and minimizing the losses and hence eliminate voltage instability in a power system. Many techniques of conipensation were used by power system industries to minimize the losses and improve the voltage, such as On-load tap changing transformer (OLTC) and Static Var Compensator. Similarly, many technique of optimization have been in used for the past few years. These tecllniques are Artificial Neural Network (ANN), Generate Algorithm (GA), Evolutionary Programming (EP) and AIS. The purpose of this paper is to present a new technique for SVC planning using AIS technique in order to minimize the losses and improves the voltage profile in a power system.

S. Ishak, A. F. Abidin and T. K. A . Rnhman are with Faculty of Eleclricat Enginccting, Univcrsiti Tckrrologi MARA, 40450 Slid1 hlani, Scloiignr, Malaysia.

0-7803-8724-4/04/$20.00 02004 IEEE.

11. STATIC VAR COMPENSATOR (SVC)

Over the last decade, SVC has become popular nieans of providing fast-acting reactive support in power systems. These devices are used for voltage support, minimization of reactive power and system losses and improving stability linuts [Z]. Another advantages of installing SVC, are an increasing operation efficiency, ensuring certain security level, improving service quality and can be forniulated into an optimization problem [ 3 ] . Traditionally, SVC is installed in heavy load areas and at the weakest voltage buses to alleviate stressed systenls.

111, ARTIFICIAL IMMUNE SYSTEM (AIS)

Over the last Tew years, there has been an increase of interest in the area of AIS and their applications. Among the many works in this new field of research, are of Ishida (1996), Hunt and Cook (I996), Dasgupta (1999) and Hofmeyr and Forrest (1999) [4]. The AIS uses an idea gleaned from immunology in order to develop systems capable of performing direrent facts in varies areas of research.

In this paper, the clonal selection concept will be reviewed together with the affinity maturation process and demonstrate that these biological principles can lead to the developnient of useful computational tools, The algoritlini to be presented focuses on a systematic view of the immune system and docs not take into account cell-cell interactions. It is not our concern to model exactly any biological phenomenon, but to show that some basic immune principles can help us to not only better understand the immune system ilself, but also to solve complex engineering tasks [4][5]. In this course of work, AIS technique was implemented to show how the suitable SVC could be determined before being installed at the bus in load flow.

IV. METHODOLOGY

This study is aimed to investigate the effect of SVC placement at the weakest voltage buses in power system for loss minimization. Firstly, when the 14 buses load flow program was run, the systcni identified 3 load buscs vollages with lowest for the purpose of installing the SVC. The AIS optimization technique was then used to deternune the suitable

The AIS optimization technique was implemented in the following procedures. First, 3 initial values were generated

valuc o r the SVC.

randomly. Then, these values will be the size of SVC to be installed and transferred into the load flow for the purpose of evaluating the total losses. This technique was repeated until IO values of total losses subject to voltage range were obtaided at each load bus. Second, the size of SVC and total loss were cloned. Then, the value of clone, was mutated followed by running again the load flow to get new total loss. This process again was repeated until the minimum new total loss was obtained. Based on the yield above, the value o f SVC will be defined and installed in the load flow. Fig. 1 shows the flow chart of AIS technique.

'7, WPULATIIN

Fig. I . Flow chart for AIS technique

A. Test System

The technique was tested on the IEEE 14-bus reliability test systems as shown in Figure 2. This system consists of 5 generators (PV bus) and 9 load buses (PQ bus) with 20 interconnected lines. In this study, SVC can be installed at any of weakest voltage at load buses from 6-14. For practical and economic considerations, the number of SVC units is limited to not exceeding three [2]. After running the system, buses 8, 12 and 13 were selected to perform the test.

Ftg. 2. 14-bus IEEE Reliability Test System

B. Artificial Immune System Algorithm

42

Initialiy, the series of random number, x i is generated by uniform distribution number, where:

where: i = 1,2 and 3

a = number of row b = number of column

off set = max. range number that can be achieve min set = plus the max. range number

The random number generated using the uniform distribution number will be assigned as the transformer tap changer setting in the system. Since the test system is the IEEE 14-bus Reliability Test System, thus three variables namely x , , x2 and xj were generated to represent the sized of SVC. During the initialization process, some constrains need to be considered, i.e.:-

Total Loss I Total Los met (3)

During the initialization the constraints imposed are the bus voltage range must be greater than 0.95 Y and less I . 05 V . Total Loss must be less that Total Loss set.

The test was repeated until 10 values of SVC and Totnl Loss subject to voltage range were obtained at each load bus. Then, the sue of SVC and Total Loss were cloned.

Clone = repmat ( A , [ i I j ] ) (4)

where; A = subject to clone i = clone the row j = clone the column

Maximum fitness, minimum fitness, sum of fimess and average fitness as follows:-

'I'hese values will be used for the i&~tation process. The random number, xi was pcrrormed on iiiutalioii LO produce ofkpring. The equation for mutation process is:-

19) ' $

~ / t m n j = ~ a j + ~ ( O , ~ ( ~ ~ " , ~ ~ ~ - ~ , , , i i , , ) ( ~ ) ) f nnx

where:

43 V. RESULTS AND DISCUSSIONS

TABLE 1 STUDY SYSTEM BEFORE INCREASING TlIE l.OAl>

Table 1 shows the data bus for 14-bus system at stability conditions. For making the system to be instable conditions, h e load of 20% and 56" were increased. Following the initial condition, the voltage at each bus is decreasing and the total loss will be increased. This part shows the test result before and after iiistalliiig the SVC at both conditions.

x i+,uJ = nititatecl parent (oKspring) x ,J =parents

N = Gaussian random variable with mean p and variance y

p = mutalion scale, O<P <1 A. 211% Loud Iticrease x jm.r = maximum random number for every

x, = minimum random iiuniber for every variable

variable f; = fitness for the i'" random number

f,,, = maximum fitness

In selection, the offsprings produced by mutation process will be sort and choose the best 10 from 100 the offsprings subject to T i , d Lvss.

Stopping criteria or convergence ' criterion is specified when the value of difference between maximum Total Los3 and minimum Total Loss is I less than 0.0001. The mathematical equation is given as :-

Toral Loss mr'l - rofoz loss m,,l I 0.000 1 (10)

Total loss: 54.987 M w

44

TABLE 3 STUDY SYSTEM WITH OPTIMAL SVC REINFORCEMENT

Gen I

iterations the load flow program was converges as shown in Table 5 . - B. 5UK Loud Increase

TABLE 6 STUDY SYSTEM WITHOUT SVC

Mvar3 I Total Loss 1

Total loss: 54.1 136 Mw Total loss: 12 1.994Mw

TABLE 4 FIRST GENERATION SVC

I I Mvarl I MvarZ I Mvar3 1 Total LosslMw) I

TABLE 5 CONVERGES SVC

Mvar2 1 Mvar3 1 Tolal I

Based on Table 2, when 20% load was increased, the system will be instable and optimal SVC sites are at bus 8, 12 and 13. With such SVC placement, voltage bus at bus 8, bus 12 and 13 was increased from 0.915p.u to 0.956p.u,0.935p.u to 1.01Xp.u and 0.929p.u to 1.003p.u. Resulting in total loss to reduce from 54.987 MW to 54.1 136MW as shown Table 3.

It is found that the installation of the SVC in the system can help to minimize the losses and improve the voltage. Table 4 shows the first generation sized of SVC and Total Loss after run the load flow program at the 20% load increased. After 26

TABLE 7 STUDY SYSTEM WITH SVC REINFORCEMENT

Total loss: 9 1.9GGOMw

TAULE E PlKST GENERATION SVC

1 Mvarl I Mvar2 I Mvar3 1 Tom1 h s s ( M w ) 1

45 TAULE 9

CONVERGES SVC 2. Reactive power flow control by tap-cliaugiiig trans rormcrs

or by simply improving the program itself in such a way of reducing computational speed (lesser itcration) and computcr memory. * r

VITI. REFERENCES

11 j A . h l!l-Kcil), XMB, “Applicaliori oI‘ Arli l icial Nc~:r l NC[WII~ it1 Vollage Snbili ly Assessmcnl”, IEEE Traiis. On I’owcr System, Vol. 10. No.4, Nov.1995. A A . El-Einary,” Formula For Tlic ElTccl 0 On Synchionizing ’Ibrquc Cocllicicti!”. I Dislrib., Vol. 143. Noh, Nov. 1996. C.S.Clioiig and .f.S Iluang, “Op~imal MulliutJ~lcclivc SVC t’la~rning Iiv Vollogc Slabilily Enhnnccnicnl”. [EBE 1’roc.-Gcocr. Transm. DislIih., Vol.145, No.2, Mar. 1908. I-candro Nunea de Caslro and Pemandn J. Vnii Zubcn,”Thc Clonal Sclclioii Algorihin will1 liiigiiiccriiig Aplilicalio~ls”, 111 Workshop Proceeding or Gccco’OO, pp. 36-37, Workshop on Artificial lniniutie Syslcm and ’flleir Applications, Las Vegas. USA, July 2000. Leandro N. de Castroand J . Von Zuben, “Leaming and Opl imimlio~i Using tlic Cional Selection Principle”, IEEE ~ransniission on Evoluiinii, Vo1.4. No.3, Junc 2002.

[2]

[3]

141

Same wilh tlle above condition, the SVC placement of

0.82Op.u to 0.971p.u, 0.885p.u to 0.985p.u and.0.880p.u to 0.913p.u as shown 1 Table 8 and Table 9, respectively. As a result, the total loss is also reduced from 121.994p.u to 9 I .9GGOp.u.

flere, it can be concludcd that tlie iilci-ement of 50% load is better than the increment of 20% load. For 50% load of increment, total loss reduction obtained is 30.0280Mw compared with 20% load of incremcnt oiily 0.8734Mw.

Table 8 shows the first generation sized of SVC and Total Loss after run the load flow program at the 50% load increased. After 17 iterations the load flow program was converges as shown in Table 9.

[SI voltage bus at ‘9 12 and 13 were increased u~iit voltage rroln

.

. .

vr. CONCLUSION

A new technique to identify the best location for SVC and its sizing is proposed. This technique IS used to minimize the total loss and improve the voltage in power system. The correct placement o f SVC is determined by looking a1 [lie weakest bus voltage in power flow. In this project, MATLAB is used to simulate the artificial immune system in order to determine the optimal size of the SVC for loss minimization i n power system. The proposed technique was tested on the 14- bus IEEE reliability test system at 20% and 50% load increased. The result indicated that it is possible to be implemented practicaIly.

VII. FUTURE DEVELOPMENT

This paper only presents SVC planning using AIS technique. It is hoped that with the presentation of this paper the understanding o f SVC, AIS, load flow and i t associated methods of solution can be comprehended for future studies and’modifications. Such modifications may be i n the foi-in of including the following active devices:

1 .Local or remote voltage control by tap-changing transformers.