optimal placement of custom power devices

International Journal of Electrical Engineering and Technology (IJEET), ISSN 0976 – 6545(Print), ISSN

0976 – 6553(Online) Volume 3, Issue 3, October – December (2012), © IAEME

187

OPTIMAL PLACEMENT OF CUSTOM POWER DEVICES IN POWER

SYSTEM NETWORK FOR LOAD AND VOLTAGE BALANCING

D.K. Tanti

1, M.K. Verma

2, Brijesh Singh

3, O.N. Mehrotra

4

1,4Department of Electrical Engineering, Bihar Institute of Technology, Sindri (INDIA)

E-mail: [email protected] ,

4 [email protected]

2,3Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi

(INDIA)

E-mail: [email protected] ,

[email protected]

ABSTRACT

In this paper, a criterion based on Artificial Neural Network (ANN) has been developed

for optimal placement of Distribution Static Compensator (DSTATCOM), Dynamic Voltage

Restorer (DVR) and Unified Power Quality Conditioner (UPQC) in a power system network for

balancing of load voltage and current against switching of unbalanced load across it, and to

balance voltage at all other buses which get affected due to connection of unbalanced load in the

system. A feed forward neural network with back propagation algorithm has been trained with

unbalanced bus voltages with targets defined as balanced bus voltages prior to connection of

unbalanced load in the system. The optimal bus has been taken as the bus having maximum

squared deviation of three phase unbalanced bus voltage from its target value. The DSTATCOM,

DVR and UPQC have been placed at the optimal bus or in the line connecting optimal bus. Case

studies have been performed on IEEE 14-bus system. Simulations have been carried out in

standard MATLAB environment using SIMULINK and power system block-set toolboxes. The

effectiveness of proposed approach of placement of custom power devices in load and voltage

balancing has been established on the test system considered.

KEYWORDS: Load balancing, Voltage balancing, DSTATCOM, DVR, UPQC, ANN

INTERNATIONAL JOURNAL OF ELECTRICAL ENGINEERING

& TECHNOLOGY (IJEET)

ISSN 0976 – 6545(Print)

ISSN 0976 – 6553(Online)

Volume 3, Issue 3, October - December (2012), pp. 187-199

© IAEME: www.iaeme.com/ijeet.asp

Journal Impact Factor (2012): 3.2031 (Calculated by GISI)

www.jifactor.com

IJEET

© I A E M E



188

1. INTRODUCTION

The present distribution systems are facing severe power quality problems such as poor

voltage regulation, high reactive power demand, harmonics in supply voltage and current, and

load unbalancing [1]. Therefore, maintenance of power quality is becoming of increasing

importance in worldwide distribution systems. Industrial consumers with more automated

processes require high quality power supply else equipments such as microcontrollers, computers

and motor drives may get damaged. High quality power delivery includes balanced voltage

supply to consumers. Connection of unbalanced load at a bus may cause unbalanced voltage and

current drawn by other loads connected at that bus. Switching of unbalanced load at a bus may

also result in unbalanced voltage at some other buses. Unbalanced voltages contain negative and

zero sequence components which may cause additional losses in motors and generators,

oscillating torques in Alternating Current (AC) machines, increased ripples in rectifiers,

saturation of transformers, excessive neutral currents and malfunctioning of several type of

equipments.

With the advancement in power electronics, new controllers known as Flexible AC

Transmission System (FACTS) have been developed [2]. These controllers have been proved to

be quite effective in power flow control, reactive power compensation and enhancement of

stability margin in AC networks [3]. Power electronics based controllers used in distribution

systems are called custom power devices. Custom power devices have been proved to be quite

effective in power quality enhancement [1]. The custom power devices may be series, shunt, and

series-shunt or series-series type depending upon their connection in the circuit. Most prominent

custom power devices include Distribution Static Compensator (DSTATCOM), Dynamic

Voltage Restorer (DVR) and Unified Power Quality Conditioner (UPQC) [1]. There are several

papers reported in literature on placement of custom power devices in balancing of unbalanced

load in radial distribution systems. Load voltage balancing using DVR against unbalanced

supply voltage in radial distribution system has been considered [4], [5]. Placement of

DSTATCOM in weak AC radial distribution system for load voltage and current balancing has

been considered in [6]. Balancing of source currents using DSTATCOM in radial distribution

system has been considered in [7]. In [7], unbalancing has been caused by connection of

unbalanced and non-linear load. Load compensation using DSTATCOM against unbalancing

caused by opening of one of the phase of the load in radial distribution system has been

considered in [8]. Balancing of supply across an unbalanced 4-phase load along with power

factor improvement using DSTATCOM has been suggested in [9]. A Voltage Source Converter

(VSC) based controller has been proposed in [10] to balance terminal voltage of an isolated

standalone asynchronous generator driven by constant speed prime mover. A non-linear and

unbalanced load has been connected at the generator terminals in [10] to create unbalance in

supply voltages. A DVR/APF (Active Power Filter) based on Proportional Resonant (PR)

controller has been proposed in [11] to protect sensitive industrial loads at the point of common

coupling, against voltage harmonics, imbalances and sags. The Artificial Neural Network (ANN)

based methodologies have been successfully applied in several areas of the Electrical

Engineering, including detection of voltage disturbances, voltage and reactive power control,

fault detections [12]-[14]. An AI based UPQC has been modeled using MATLAB toolbox to

improve power quality [15 ].To generate switching signals for the series compensator of the

UPQC system NNC algorithm such as MRC and NARMA-12 has been used. The paper [16 ] has



189

proposed a model for UPQC to compensate input voltage harmonics and current harmonics

caused by non-linear load. The control strategies are based on PI and ANN controller. Again, in

paper [17 ] the ANN based controller has been designed and trained off-line using data from the

conventional proportional, integral controller. The performance of ANN and PI controller has

been studied and compared for UPQC using MATLAB simulation. An ANN based approach for

optimal placement of Custom Power Devices to mitigate voltage sag in a meshed interconnected

power system, has been suggested in [18].

Unbalanced load connected at a particular bus may cause voltage unbalances at several

other buses in an interconnected power system network. No effort seems to be made in optimal

placement of custom power devices in an interconnected power system network in balancing bus

voltages at all the buses caused by unbalanced load connected at a particular bus. In this paper,

an Artificial Neural Network (ANN) based approach has been proposed for optimal placement of

custom power devices to balance unbalanced voltages in the whole power system network. The

ANN has been trained with Levenberg Marquardth back-propagation algorithm (trainlm ). Case

studies have been performed on IEEE 14-bus system. [19].

2 CUSTOM POWER DEVICES MODEL

2.1 DSTATCOM model

In the present work, the DSTATCOM has been represented as three independently

controllable single phase current sources injecting reactive current in the three phases at the point

of coupling. The proposed DSTATCOM model has been shown in Figure-1. The control scheme

consists of three control switches which can be set on/off as per compensation requirement. The

maximum and minimum reactive power injection limit of DSTATCOM has been taken as +50

MVAR and -50 MVAR, respectively.

Figure-1. Proposed DSTATCOM model



190

2.2 DVR model

In the present work, the DVR has been represented as three independently controllable

single phase voltage sources injecting complex voltages in series with the line in the three

phases. The magnitude and angle of injected voltages may be controlled to balance load voltage

at different buses. The proposed DVR model has been shown in Figure-2 . The control scheme

consists of six control switches which can be set on/off as per compensation requirement. During

off condition, the three control switches connected in series with the controllable single phase

voltage sources are open and the other three control switches in parallel with controllable voltage

sources, are closed. When compensation is required , the three switches connected in series with

independently controllable voltage sources are closed, and the remaining three switches are made

open. This permits injection of controllable complex voltages in the three phases of the line

which causes load balancing and voltage balancing of different buses.

Figure-2. Proposed DVR model

2.3 UPQC model

In the present work, UPQC has been considered as combination of DSTATCOM and

DVR models suggested in sections 2.1 and 2.2, respectively.

3. METHODOLOGY

In this work, feed forward Artificial Neural Network with back propagation algorithm

has been used to find optimal location for DSTATCOM placement. The architecture of this

network has been shown in figure-3.

In figure-3, the input data p(1), p(2), ……….p(R) flow through the synapses weights wi,j.

These weights amplify or attenuate the input signals before being added at the node represented

by a circle. The summed data flows to the output through an activation function f. The neurons

are interconnected creating different layers. An elementary neuron with R inputs has been shown

in figure-3. Each input is weighted with an appropriate weight w. The sum of the weighted inputs

and the bias, b forms the input to the transfer function f.

International Journal of Electrical Engineering and Technology (IJEET), ISSN 0976

0976 – 6553(Online) Volume 3, Issue 3, October

Once the network weights and biases are initialized, the network is ready for training.

The training process requires a set of examples of proper network behavior

and target outputs, t. During training the weights and biases of the network are iteratively

adjusted to minimize the network performance function. The default performance function for

feed forward networks is Mean Square Error (MSE)

network outputs and the target outputs. The gradient is determined using a technique called back

propagation, which involves performing computations backward through the network.

In the proposed neural network archite

layers. This network can be trained to give a desired pattern at the output, when the

corresponding input data set is applied. The training process is carried out with a large number of

input and output target data. The system has been made unbalanced by connection of highly

unbalanced load at different load buses. The three phase balanced per unit (p.u.) voltages of

buses prior to connection of unbalanced load, have been taken as output target data. The three

phase p.u. voltages of buses under unbalanced loading conditions have been considered as input

data to train the neural network. Once the network is trained some data are used to test the

network. The testing results provide information about the optimal l

DSTATCOM controller. Mean Square Error has been computed for all the buses. The load bus

corresponding to highest mean Mean Square Error value has been selected as the optimal bus for

the placement of DSTATCOM controller.

lines connected to the optimal bus. The line where placement of DVR results in the maximum

balancing of voltage and load is considered as the optimal line for the placement of DVR. The

UPFC placement is considered in optimal line towards optimal bus.

Figure-

Electrical Engineering and Technology (IJEET), ISSN 0976 –

6553(Online) Volume 3, Issue 3, October – December (2012), © IAEME

191


The training process requires a set of examples of proper network behavior—network inputs, p



feed forward networks is Mean Square Error (MSE) — the average squared error between the

network outputs and the target outputs. The gradient is determined using a technique called back


In the proposed neural network architecture there are 20 hidden layers and 14 output



data. The system has been made unbalanced by connection of highly





network. The testing results provide information about the optimal location for the placement of



the placement of DSTATCOM controller. The placement of DVR is considered in each of the



red in optimal line towards optimal bus.

(a)

(b)

-3. Artificial Neural Network architecture

– 6545(Print), ISSN


network inputs, p



he average squared error between the

network outputs and the target outputs. The gradient is determined using a technique called back-


hidden layers and 14 output



data. The system has been made unbalanced by connection of highly





ocation for the placement of



lacement of DVR is considered in each of the





192

4. CASE STUDY

Case studies were performed on IEEE 14-bus system [15] having 14 buses and 20 lines.

The system consists of 5 synchronous machines three of which are synchronous condensers.

There are 11 loads in the system having a net real and reactive power demand of 259 MW and

81.3 MVAR, respectively. The single-line-diagram of the system has been shown in figure-4.

Simulation model of IEEE 14-bus system was developed using software package

MATLAB/SIMULINK [16]. The simulation block diagram of the system has been shown in

figure-5. The developed plant model shown in figure-5 was used to find three phase balanced bus

voltages prior to switching of unbalanced load, unbalanced three phase voltage and current at the

bus where unbalanced load is switched on, and unbalanced three phase voltages at other buses in

the system. In order to create unbalance loading condition, an additional Y- connected highly

unbalanced load ; Phase A [P=1MW, Q=100MVAR] , Phase B [ P=25KW,

Q=200KVAR] , Phase C [ P=1KW, Q=0.1KVAR] was connected at each bus considered at a

time, with all other buses having balanced base case loadings. A feed forward neural network

was trained with three phase unbalanced bus voltages. The balanced three phase voltages of

different buses prior to connection of unbalanced load at a bus were considered as target data for

the neural network. The Mean Square Errors (MSE) were calculated for all the buses using

training data and target data. The MSE of all the buses have been shown in figure-6. It is

observed from figure-6 that bus-5 has maximum MSE value. Therefore, bus-5 was selected as

the optimal location for the placement of DSTATCOM controller. Placement of DVR was

considered in each of the lines connected to bus-5 viz. line 5-1, line 5-2, line 5-4 and line 5-6,

respectively, and the three phase voltages of different buses were observed. It was found that

placement of DVR in line 5-4 was more effective in voltage load and voltage balancing

compared to DVR placement in line 5-1, line 5-2 and line 5-6, respectively. Therefore, line 5-4

was selected as the optimal line for the placement of DVR controller. UPFC placement was

considered in optimal line 5-4 towards optimal bus-5.

Three phase voltage at all the buses and three phase current at the bus with unbalanced

load were plotted versus time for the four cases – (i) without any controller (ii) with placement

of DSTATCOM at the optimal bus (iii) with the placement of DVR in the optimal line and (iv)

with the placement of UPQC in optimal line towards optimal bus. The relative performance of

DVR, DSTATCOM and UPQC in load and voltage balancing is studied to decide most suitable

controller out of the three controllers considered. The variation of three phase voltage with

respect to time for all the buses and variation of three phase current with respect to time at the

bus with unbalanced load were plotted using MATLAB software [16]. Three phase voltage and

current at bus-2 with unbalanced load connected at bus-2 have been shown in figure-7. Three

phase voltage at bus-5 and at bus-10 with unbalanced load connected at bus-2 have been shown

in figure-8. Three phase voltage and current at bus-10 with unbalanced load connected at bus-10

have been shown in figure-9. Three phase voltage at bus-4 and at bus-5 with unbalanced load

connected at bus-10 have been shown in figure-10. Three phase voltage and current at bus-12

with unbalanced load connected at bus-12 have been shown in figure-11. Three phase voltage at

bus-4 and at bus-7 with unbalanced load connected at bus-12 have been shown in figure-12. It is

observed from figures 7, 9 and 11 that placement of custom power devices in the network results

in considerable balancing of load voltage and current at the bus with unbalanced load. It is

observed from figures 8, 10 and 12 that placement of custom power devices in the network is

also able to produce considerable voltage balancing at other buses.



193

Figure-4. Single-line-diagram of IEEE 14-bus system

Figure-5. IEEE-14 Bus system (MATLAB/SIMULINK) model



194

Figure-6. Mean Square Error for different buses (IEEE 14-bus system)

Unbalance load connected at Bus 2

Bus No. 2 (Voltage waveform) 2 (Current waveform)

Without

Controller

With DVR

in Line 5-4

With

DSTATCO

M at Bus 5

With

UPQC at

Line 5-4

Toward

Bus 5

Figure-7. Three phase voltage and current at bus-2 with unbalanced load connected at bus-2



195


Bus No. 5 (Voltage waveform) 10 (Voltage waveform)

Without

Controller

With DVR in

Line 5-4

With

DSTATCOM

at Bus 5

With UPQC

at Line 5-4

Toward Bus 5

Figure-8. Three phase voltage at bus-5 and at bus-10 with unbalanced load connected at bus-2



Without

Controller

With DVR in

Line 5-4

With

DSTATCOM

at Bus 5

With UPQC

at Line 5-4

Toward Bus 5




196



Without

Controller

With DVR

in Line 5-4

With

DSTATCO

M at Bus 5

With UPQC

at Line 5-4

Toward Bus

5

Figure-10. Three phase voltage at bus-4 and at bus-5 with unbalanced load connected at bus-10



Without

Controller

With DVR

in Line 5-4

With

DSTATCO

M at Bus 5

With UPQC

at Line 5-4

Toward Bus

5




197



Without

Controller

With DVR

in Line 5-4

With

DSTATCO

M at Bus 5

With UPQC

at Line 5-4

Toward Bus

5

Figure-12. Three phase voltage at bus-4 and at bus-7 with unbalanced load at connected at bus-12

5. CONCLUSION

In this work, an Artificial Neural Network based approach has been suggested for the

placement of Custom Power Devices in power system to balance three phase voltage and

current at a bus where a highly unbalanced load is switched on, and to balance three phase

voltage at all other buses which become unbalanced due to connection of an highly unbalanced

load at a particular bus. Case studies were performed on IEEE 14-bus system using

MATLAB/SIMULINK. Simulation results on the test system validate the effectiveness of the

proposed approach of placement of custom power devices in load and voltage balancing. The

placement of UPQC seems to be more effective in load and voltage balancing compared to

placement of DSTATCOM and DVR controllers. The proposed approach of optimal placement

of custom power devices is quite simple and easy to adopt.

REFERENCES

[1] A. Ghosh and G. Ledwich, “Power quality enhancement using custom power devices”,Kluwer

Academic Publishers (London) 2002.

[2] N. G. Hingorani and L. Gyugyi, “Understanding FACTS: Concepts and technology of Flexible AC

Transmission System”, IEEE publication, 2000.

[3] J. Dixon, Luis Moran, Jose Rodriguez, “Reactive power compensation technologies: State of art

review”, Proceedings of the IEEE, Vol.93, No. 12, pp. 2144-2164, December 2005.



198

[4] ArindamGhosh and Gerald Ledwich, “Compensation of distribution system voltage using DVR”,

IEEE Transactions on Power Delivery, Vol. 17, No. 4, pp. 1030-1036, October 2002.

[5] PendroRoncero-Sanchez, Enrique Acha, Jose Enrique Ortega-Calderon, Vicente Feliu, and Aurelio

Gareia-Cerrada, “A versatile control scheme for a Dynamic Voltage Restorer

for power quality improvement”, IEEE Transactions on Power Delivery, Vol. 24, No. 1, pp. 277-284,

January 2009.

[6] ArindamGhosh and Gerald Ledwich, “Load compensating DSTATCOM in weak AC systems”, IEEE

Transactions on Power Delivery, Vol. 18, No. 1, pp. 1302-1309, October 2003.

[7] C. N. Bhende, Dr. M. K. Mishra, and Dr. H. M. Suryawanshi, “ A D-STATCOM modeling, analysis

and performance for unbalanced and non-linear loads”, Institutions of Engineers(India) Journal – EL ,

Vol. 86, pp. 297-304, March 2006.

[8] Wei-Neng Chang and Kuan-DihYeh, “Design and implementation of D-STATCOM for fast load

compensation of unbalanced loads, “Journal of Marine Science and Technology” , Vol. 17, No. 4, pp.

257-263, 2009.

[9] Zakir Husain, Ravinder Kumar Singh and ShriNiwasTiwari, “ Balancing of unbalanced load and

power factor correction in multiphase ( 4 phase ) load circuits using D-STATCOM”, Proceedings of the

World Congress on Engineering 2010, Vol. II WCE-2010, June 30-July 2, 2010, London (U.K).

[10] Gaurav Kumar Kansal and Bhim Singh, “Harmonic elimination, voltage control and load balancing

in an isolated power generation”, European Transactions on Electrical Power, Vol. 20, Issue 6, pp.

771-784, September 2010.

[11] Pablo Fernandez-Comesana, Francisco D. Freijedo, Jesus Doval-Gandoy, Oscar Lopez, Alejandro G.

Yepes, JanoMalvar, “Mitigation of voltage sags, imbalances and harmonics in sensitive industrial loads

by means of a series power line conditioners”, Electric Power Systems Research, Vol. 84, Issue 1, pp. 20-

30, March 2012.

[12] E.A.Mohamed , N.D. Rao, “Artificial Neural Network based fault diagnostic system for electric

power distribution feeders,” Electric Power Systems Research, Vol. 35, No. 1, pp. 1-10, 35 October 1995.

[13] Ernesto Vazquez, Hector J. Altuve, Oscar L. Chacon, “Neural network approach to fault detection in

electric power systems”, IEEE International conference on Neural Networks, Vol. 4, pp. 2090-2095, June

3-6, 1996, Washington, DC, USA.

[14] F.J. Alcantare, J. R. Vazquez, P. Salmeron, S.P. Litran, M.I. Arteaga Orozco, “On line detection of

voltage transient disturbances using ANNs,” International Conference on Renewable Energies and Power

Quality (ICREPQ 09) 15th

to 17th April 2009, Valencia, Spain.

[15] Moleykutty George, “ Artificial Intelligence based three phase Unified Power Quality Conditioner”,

Journal of Computer Science (3) 7 : pp. 465-477, 2007.

[16] R.V.D. Rama Rao, Dr. SubhransuSekhar Dash, “ Power Quality Enhancement by Unified Power

Quality Conditioner using ANN with Hysteresis control”, International

Journal of Computer Applications (0975-8887), Vol. 6-No.-1, pp. 9-15, Sept.2010.

[17] N. Ramchandra, M. Kalyanchakravarthi, “ Neural Network Based Unified Power Quality

Conditioner”, International Journal of Modern Engineering Research. (IJMER), Vol.2, Issue 1, pp. 359-

365, Jan.-Feb.2012.

[18] D. K. Tanti, M. K. Verma, Brijesh Singh and O. N. Mehrotra, “ Optimal Placement of Custom Power

Devices in Power System Network to Mitigate Voltage Sag under Faults”, International Journal of Power

Electronics and Drive System (IJPEDS), Vol. 2, No. 3, pp. 267-276, September 2012

[19] “Power systems test case archieve” available at

http://www.ee.washington.edu/research/pstca/pf14/pg_tca14bus.htm

[20] MATLAB 7 User’s Guides for SIMPOWER SYSTEMS and Neural Network Tool-box.



199

BIOGRAPHIES

D. K. Tanti received B.Sc. (Eng.) degree in Electrical Engineering from Muzaffarpur Institute of

Technology (India) in 1990 and M.Sc. (Eng.) degree in Electrical Engineering from Bihar

Institute of Technology, Sindri (India) in 2000. Presently, he is Associate Professor in the

Department of Electrical Engineering, Bihar Institute of Technology, Sindri (India), and pursuing

for his Ph.D degree at Vinoba Bhave University, Hazaribag (India). His research interests

include application of FACTS controllers, power quality and power systems.

M. K. Verma received B.Sc. (Eng.) degree in Electrical Engineering from Regional Engineering

College, (presently National Institute of Technology), Rourkela (India) in 1989, M.Sc. (Eng.)

degree from Bihar Institute of Technology , Sindri (India) in 1994 and Ph.D. degree from Indian

Institute of Technology, Kanpur (India) in 2005. Presently, he is Associate Professor in the

Department of Electrical Engineering, Indian Institute of Technology (BHU), Varanasi (India).

His research interests include voltage stability studies, application of FACTS controllers,

operation and control of modern power systems, power quality and smart grid.

Brijesh Singh received B.Tech. degree in Electrical Engineering from Faculty of Engineering

and Technology, Purvanchal University, Jaunpur (India) in 2003 and M.Tech. degree from

Kamla Nehru Institute of Technology, Sultanpur (India) in 2008. Presently, he is persuing for his

Ph.D. degree at Indian Institute of Technology (BHU), Varanasi (India). His research interests

include modeling and analysis of power systems, application of FACTS controllers and power

quality.

O. N. Mehrotra received B.Sc. (Eng.) degree in Electrical Engineering from Muzaffarpur

Institute of Technology (India) in 1971, M.E. (Hons.) degree in Electrical Engineering from

University of Roorkee, (presently Indian Institute of Technology, Roorkee, India) in 1982 and

Ph.D. degree from Ranchi University (India) in 2002. Presently, he is Professor (retired),

Department of Electrical Engineering, Bihar Institute of Technology, Sindri (India). His research

interests include control and utilization of renewable energies, power quality and power systems.

optimal placement of custom power devices

Documents

balancing of load voltage

load voltage balancing

power system network

power electronics

load balancing

power system block

bus system

power quality enhancement