Computational Intelligence Based Tehnique for Load Shedding Scheme

COMPUTATIONAL INTELLIGENCE BASED TEHNIQUE FOR LOAD SHEDDING SCHEME

LUKMAN HAKIM BIN HAMRON

i

FACULTY OF ELECTRICAL ENGINEERING

UNIVERSITI TEKNOLOGI MARA

MALAYSIA

COMPUTATIONAL INTELLIGENCE BASED TECHNIQUE FOR LOAD SHEDDING SCHEME

Project report is presented in partial fulfillment for the award of the

Bachelor of Electrical Engineering (Hons)

Universiti Teknologi MARA (UiTM)

ii

LUKMAN HAKIM BIN HAMRON

Faculty of Electrical Engineering

UNIVERSITI TEKNOLOGI MARA

40450 SHAH ALAM, SELANGOR

A report submitted to Faculty of Electrical Engineering, Universiti Teknologi MARA in partial fulfillment of the requirement for Bachelor of Electrical Engineering (Hons).

This thesis is approved by:

……………………………….

Associate Professor Dr. Ismail Musirin

Project supervisor

Faculty of Electrical Engineering

Universiti Teknologi MARA (UiTM)

40450 Shah Alam

Selangor.

Date:………………..

iii

DECLARATION

It is hereby declared that all materials in this thesis are the result of my own work and

all the materials that are not the result of my own work have been clearly

acknowledged. Although, certain result on this thesis is effort from other dispute.

iv

ACKNOWLEGMENT

First and foremost, all praise to Allah S.W.T. The Most Gracious and Most Merciful

who had given me the strength, ability and patient upon completing this final year

project.

I wish to conveymy deepest gratitude and appreciation to my supervisor, Assoc. Prof.

Dr. Ismail Musirin for his guidance, concern, valuable time, effort, constant

encouragement and patience in supervising this project from the start until the

completion if this thesis.

I also wish to take this opportunity to express my gratitude to my family especially to

my mother and my father for supporting me along way my journey in this field. They

have encourage me throughout my education , and I will always be grateful for their

sacrifice, generosity and love. May Allah S.W.T. bless them all.

Not forget to my friends and anyone who directly or indirectly giving their support

and contribution to finished this project. May Almighty Allah bless and reward them

for their generosity. Thank you very much.

v

ABSTRACT

Losses in generation and overloading effect are two phenomena that may

occur due to progressing demand at the load side. This may lead to system instability

in forms of voltage and frequency. In order to avoid this problem, the under voltage

load shedding scheme can be performed to shed some amount of load before the

disturbance occur. This paper presents computational intelligence technique for load

shedding. The study involves the development of fuzzy rules in order to make

decision on load shedding. This method functions will determine the amount of load

that needs to be shed depending on the measured minimum voltage of the system.

The result of this paper will show the performance of under voltage load shedding

scheme in determining power system stability by shedding some amount of the load

demand. The technique has been validated on the IEEE 30-bus system.

Index Terms—voltage collapse, system stability, fuzzy logic, under voltage load shedding

vi

TABLE OF CONTENTS

CHAPTER DESCRIPTION PAGE

DECLARATION i

ACKNOWLEDGEMENT ii

ABSTRACT iii

TABLE OF CONTENTS iv

LIST OF FIGURES vi

LIST OF TABLES vii

LIST OF ABBREVIATIONS viii

1.0 INTRODUCTION 1

1.1 INTRODUCTION 1

1.2 PROBLEM STATEMENT 3

1.3 OBJECTIVE 3

1.4 SCOPE OF THE PROJECT

1.5 RESEARCH FRAMEWORK

1.6 OVERVIEW OF THE REPORT

4

5

6

2.0 LITERATURE REVIEW 5

2.1 ECONOMIC DISPATCH (ED) 5

2.2 DYNAMIC ECONOMIC DISPATCH(DED) 6

2.2.1 Ramp Rate Constraint 8

2.3 PARTICLE SWARM OPTIMIZATION (PSO)

2.4 METHOD TO SOLVE DED

9

10

vii

3.0 METHODOLOGY 15

3.1 INTRODUCTION 15

3.2 DYNAMIC ECONOMIC DISPATCH (DED)

FORMULATION 15

3.2.1 Objective Function 15

3.2.2 Equality Constraint 16

3.2.3 Inequality Constraint 17

3.2.4 Dynamic Constraint 17

3.2.5 Fitness Function 18

3.3 PARTICLE SWARM OPTIMIZATION (PSO) 19

3.3.1 Basic PSO Algorithm 19

3.3.2 Particle’s Velocity Update 20

3.3.3 Constriction Factor Approach (CFA) 20

3.3.4 Particle’s Position Update 21

3.3.5 Representation of Particle’s Position 21

3.4 DED BASED ON PSO TECHNIQUE 23

4.0 RESULTS AND DISCUSSION 26

4.1 DATA FOR IEEE 26-BUS TEST SYSTEM 26

4.2 PSO PARAMETERS SETTING 28

4.3 SIMULATION RESULTS FOR SOLUTION

OF DED BASED ON PSO 29

4.4 ANALYSIS OF PSO METHOD ON DED

SOLUTION 32

5.0 CONCLUSION 36

6.0 RECOMMENDATIONS FOR FUTURE WORKS 37

REFERENCES 38

APPENDICES 43

viii

LIST OF FIGURES

FIGURE TITLE PAGE

2.1 Single line diagram of transmission line 6

3.1 Matrix representation of particle’s position 22

3.2 Modification of gBest according to generator constraints 22

3.3 Flow chart for DED based on PSO process 25

4.1Variation of Cost with Power Demand Curve for 6 units

system30

4.2Variation of Power loss with the Load Demand for 6

units system30

4.3Graph of Fuel Cost against Load Demand for

comparison between PSO and Newton Raphson method33

4.4Convergence Characteristics of PSO Method for 6 units

system34

ix

LIST OF TABLES

TABLE TITLE PAGE

4.1 Generating Unit Capacity and Coefficients for IEEE 26-

bus system

26

4.2Initial output power and Ramp Rate limits for IEEE 26-

bus system27

4.3 Load Demand for IEEE 26-bus system of 24 hours 27

4.4 Transmission loss Coefficients for IEEE 26-bus system 28

4.5 PSO parameters 28

4.6Optimal MW Generation for each unit, Transmission loss

and Fuel Cost of 24 hours29

4.7 Comparison of PSO and Newton Raphson Result 32

4.8 Result for the Variation Number of Particles 35

x

LIST OF ABBREVIATIONS

ED - Economic Dispatch

DED - Dynamic Economic Dispatch

PSO - Particle Swarm Optimization

SED - Static Economic Dispatch

AI - Artificial Intelligent

FACTS - Flexible Alternative Current Transmission Systems

DP - Dynamic programming

GA - Genetic Algorithm

SA - Simulated Annealing

EP - Evolutionary Programming

LP - Linear Programming

NLP - Non-Linear Programming

QP - Quadratic Programming

DE - Differential Evolution

ANN - Artificial Neural Network

HNN - Hopfield Neural Network

CFA - Constriction Factor Approach

IEEE - Institute of Electrical and Electronics Engineers

xi

CHAPTER 1.0

INTRODUCTION

1.1 BACKGROUND OF THE STUDY

In power system operation, the balance between load demand and the available

generation is important to make sure the stability of the system is in good condition

[1]. Nowadays, there are many situation occur where the demand load have reached

the limit of an available generation in certain place. When this condition occurs, there

will be same situation as in 2005 where there was power outage in Malaysia where

many states of Malaysia’s northern peninsular, including Perlis, Perak, Penang and

Kedah due to the occurred fault. This situation happened due to the load demand used

by the user has exceed the limit that the available generation can support. From this

situation a load shedding scheme is initiated to avoid the system from collapsed [2].

There are many factories have take improvement step to prevent this phenomena

happening again by developing a new alternative extensively to ensure the power

system network operates in the normal steady state condition conveniently [3].

A system enters a state of voltage instability when a disturbance, increase in

load demand, or change in system condition causes a progressive and uncontrollable

drop in voltage [4]. The main factor for instability is the inability of the power system

to meet the demand of increased reactive power. Literally, it will cause the system

collapse.

There are several studies that indicate about voltage stability of the power

system. One of these studies is about estimating the voltage stability of power system

[5]. This study is based on the fast calculation of indicators of risk of voltage

1

instability has been developed. These indicators can detect on-line voltage instability

and signal the tendency towards a critical situation.

Several methods have been developed to prevent the voltage from collapse. In this

paper load shedding is applied to the selected bus so the voltage minimum will

increase and the system become stable. This technique is proposed to make sure the

system in a balanced condition. In [6], there are several methods to perform the load

shedding technique such as under-voltage load shedding and under-frequency load

shedding. The best way to perform load shedding scheme in a system is by

minimizing the amount of load to be shed [7] for voltage collapse prevention. In [7],

the paper study about the practical approach to perform the load shedding scheme.

In order to perform the developed technique, a fuzzy logic algorithm was

proposed. This algorithm provides solution as decision making to determine which

load bus that need to be shed and how much load will be shed to make sure the system

recover to the normal operation. Fuzzy logic was a useful algorithm where it can be

used in wide area of study. In [8], fuzzy logic was used to solve the unit commitment

problem. While in [9], fuzzy load shedding based algorithm is performed by using

voltage stability indicator for averting voltage collapse. In this paper, fuzzy logic is

performed by monitoring the minimum voltage by running the load flow. Then under

voltage load shedding will be perform to get the system back to normal operation. The

variable is selected from the load flow results.

This paper presents computational intelligence based technique for load

shedding scheme. The study involves the development of fuzzy rules in order to make

decision on load shedding. Results from the experiment indicated that the proposed

technique is successful to solve the load shedding problems. The load levels increase

are divided into several different loading factors. The fuzzy technique is applied to

each case to select load bus to be shed and to calculate the amount load to be shed to

prevent voltage instability.

2

1.2 PROBLEM STATEMENT

Everyday people are using equipment continuously and the load demand for

each distribution network is increasing with the increasing of electric usage among the

user. Each generation that was established in Malaysia is enough to support the load

demand in certain area depending on the load usage. There are some cases where the

load demand is higher than the generation level. This will cause voltage collapse in

the area. For example in 1995, blackout situations happen in Malaysia due to high

load usage. The reason why this situation happened is because of the hot weather at

that time. The same situation occurred in 2005 where the biggest blackout happened in

Malaysia where there is no electricity due to the fault of the main cable transmission

line grid.

As the usage of equipment is increasing, the load demand will also increase.

This condition will burden the generation to support the load demand. A generation

has their limit to support the load demand in each area. When the consumer load

demand has gone beyond the limit of available generation, it may lead to blackout.

When this situation happens, it will cause problem to all consumer. This reason

becomes the why a new method is needed to overcome this problem.

1.3 OBJECTIVE

i. To develop load shedding scheme in power systemii. To identify the Selected bus for load shedding and amount of load demand

that should be shed for stable power system operation

iii. To improve the power balance in power system operation by using

computational intelligence

3

1.4 SCOPE OF THE PROJECT

The scope of this project is to analyze the balance between the load demand

and the available generation. The data will be taken from legal resource as the first

step of this project. Later, it will be analyzed to match with the load shedding

technique. This technique is used to develop an algorithm as the solution for solving

the load shedding problem. A selected load bus will be chosen for shedding based on

the output of the develop algorithm.

The under voltage load shedding scheme is employed in order to determine

which load and amount of load that need to be shed. The voltage magnitude, active

power and reactive power at load bus will be assigned as the input variable to fuzzy

logic system. This fuzzy system will be implemented using MATLAB software. The

proposed method gives satisfactory results in term of blackouts prevention and

minimum voltage improvement.

Moreover, this project will show the performance of fuzzy logic algorithm to

be effective and useful in problem concerning the load shedding. The results of this

method will be used to decide which of the load is the most suitable to be removed for

maintaining the stability of the system.

Flow chart in Figure 1 below summarizes the involved process:

4

Preparing the system data

Initialize the load shedding scheme

Develop the fuzzy logic algorithm in matlab

Determine the shedding load to balance the system

Figure 1: Scope of project

1.5 RESEARCH FRAMEWORK

5

START

KNOWLEDGE ACQUISITION

MATLAB PROGRAMMING

FUZZY LOGIC ALGORITHM

LOADSHEDDING

DEVELOPMENT OF SIMPLELOAD SHEDDING TECHNIQUE

AND UVLS

DEVELOPMENT OF CONCEPTUAL MODEL OF

LOAD SHEDDING

DEVELOP THE PROGRAMMING CODES IN MATLAB

IMPLEMENTATION OF FUZZY LOGIC ALGORITHM FOR LOAD

SHEDDING

Figure 2: Research framework

1.6 OVERVIEW OF THE REPORT

This thesis consist of five chapters explain about solving under voltage load shedding

(UVLS) schemes implemented by using fuzzy logic system. Chapter 1 describes an

introduction of the project which includes the objective of this research and also scope

of work to complete this project.

In Chapter 2, the theory and basic of voltage stability, under voltage load shedding

and fuzzy logic systems are reviewed and explained properly. The summary are

include the full details of problem in power system, theory of UVLS scheme, theory

of fuzzy logic and its application and some literature review on method to solve the

load shedding problems.

This project thesis was followed by the design methodology that explained clearly in

Chapter 3. This chapter explains the DED formulation algorithm including all the

constraints and the PSO techniques algorithm and lastly implementation of PSO

techniques to DED problems. This chapter also indicates the flow chart of DED based

on PSO techniques.

Next is Chapter 4 that illustrated all the results obtained together with the discussion

of the results. All the tables and graph plotted are discussed clearly in this chapter

including the analysis of PSO techniques on DED solution.

On the Chapter 5, a conclusion that has been made upon the result obtains and the last

chapter is Chapter 6 which discusses the recommendations for future works in order to

improve the solution for DED problems. The last parts of this thesis are the references

and appendix.

6

CHAPTER 2.0

LITERATURE REVIEW

2.1 POWER SYSTEM VOLTAGE STABILITY

In a power system, the operation condition should be in a stable condition

where the voltage and the frequency is in equilibrium state. It means that the criteria

of the system operation should be meet the various operational, and it should also be

secure in the event of any credible contingency. That is why it is important to

maintaining the system in stable condition and secures the power system operation.

Nowadays, the problem and the challenges keep coming causing the power system

that being operated closer to their stability limits and the voltage in the system

dropping where it become unstable. Voltage instability and voltage collapse have been

considered as a major threat to present power system networks due to their stressed

operation. The disturbance that occur cause the voltage decreasing continuously and

lead to voltage collapse where the value of the voltage below its normal value. To

prevent the system collapse, lots of mechanism has been develop such as VAR

compensators, undervoltage load shedding and underfrequency load shedding [10].

Voltage collapse is the process by which the voltage falls to a low, unacceptable value

as a result of an avalanche of events accompanying voltage instability [11]. Once

associated with weak systems and long lines, voltage problems are now also a source

of concern in highly developed networks as a result of heavier loading.

In this chapter, the concept of voltage stability and the conventional method of

voltage stability analysis which is undervoltage load shedding is presented. Simulation

results on test power systems are presented to illustarate the problem of voltage

stability and the under voltage load shedding scheme to analyze the problem. The

undervoltage load shedding scheme the being implemented by using one of the

artificial intelligence technique which is called fuzzy logic.

7

2.2 VOLTAGE STABILITY

According to the IEEE Power System Engineering Committee, voltage

stability is being defined as “Voltage stability is the ability of a system to maintain

voltage so that when load admittance is increased, load power will increase, and so

that both power and voltage are controllable.” [12]. If there is disturbance occur and

the voltage in the system dropping, it will become voltage instability and lastly it will

cause voltage collapse. And nowadays, this voltage collapse is one of the major

problems which electric power networks might face [13]. The voltage stability

practically can be classified into two subcategories which are Long term and Short

term.

2.2.1 LONG TERM VOLTAGE STABILITY

In power system, the long-term voltage stability involves rather slow acting

equipment such as tap changing transformers, thermostatically controlled loads, and

generator current limiters. To analyze system dynamic performance, the long term

simulation is required. In these studies, stability is usually determined by the resulting

outage of equipment, rather than the severity of the initial disturbance.

2.2.2 SHORT TERM VOLTAGE STABILITY

The short-term voltage stability in power system involves dynamics of fast

acting load components such as induction motors, electronically controlled loads, and

HVDC converters. The study period of interest is in the order of at most several

seconds, and analysis requires solution of appropriate system differential equations

which is similar to analysis of rotor angle stability. Dynamic modeling of loads is

often essential. In contrast to angle stability, short circuits near loads are important.

This kind of voltage instability could easily happen in the result of a serious fault

occurrence in the power system network. So, it would have close relations with

electric power system protection methods [4].

8

2.3 VOLTAGE INSTABILITY

Figure 2.0-1: Power System with Remote Generation

Fig. 2.1 illustrates a simplified power system with a remote generator

supplying a substantial portion of the load at the load center through six transmission

lines. Es is the voltage at the remote generator buses 4 and Eg is the voltage at the

load center buses. As lines between the remote generators and the load center trip, the

MW power flows over fewer lines resulting in increased Var losses.

Figure 2.0-2: Real Power (MW) vs. Voltage (P-V) Curve

9

Figure 2.2 illustrates how voltage decays as lines trip. In power system, the

utilities system planners use this type of P-V curves analysis as an analysis tool to

determine the real power transfer capability across a transmission interface to supply

local load. Generally, all the planning engineers call this type of curves as nose

curves. The reason why this type of P-V curves is called as nose curves is because

when there is condition starting from a base-case system (all lines in-service),

computer-generated load flow cases are run with increasing power transfers while

monitoring voltages at critical buses. When power transfers reach a high enough level,

a stable voltage cannot be sustained and the system voltage collapses. As illustrates in

Figure 2.2, the shape of the nose of the curve depends on the nature of the load at the

load center. High levels of motor load combined with capacitor bank support of load

center voltage tend to make the voltage drop very rapidly for a small increase of

power at the nose of the curve.

The set of P-V curves illustrates that for baseline conditions shown in curve A,

the voltage remains relatively steady (changing along the vertical axis) aslocal load

increases. System conditions are secure and stable to the left of point A1. After a

contingency occurs, such as a transmission circuit tripping, the new condition is

represented by curve B, with lower voltages (relative to curve A). This is because the

power being transmitted from the remote generators now follows through five, rather

than six, transmission lines. The system must be operated to stay well inside the load

level for the nose of curve B. If the B contingency occurs, then the next worst

contingency must be considered. The system operators must increase local generation

(Eg) to reduce the power being transmitted for the remote generators to reduce losses,

as well as increase voltage at the load center to within the safe zone, to avoid going

over the nose of curve C.

2.4 DISTURBANCE IN POWER SYSTEM

Usually all the electrical energy that has been provided by the electrical utility is

safe and reliable. The problem comes when there are disturbance, disruptions,

irregularities and nature of electricity occurred. For an example there are lightning,

equipment failures, and high winds can cause power line disturbance and also will

affect the voltage in the system. If electrical equipment is being used and the is power

disturbance occur, it will cause data or memory losses, altered data and other

10

functional errors, as well as equipment damage. And if there is no preventive method

towards this problem, then it may cause scheduling problems, downtime and

expensive troubleshooting. Before proceeding to the preventive method, first thing to

do is by understanding the causes of the problems.

2.5 CAUSES OF DISTURBANCE

There are several types of irregularities that affect electrical power which are

surges, sags. transients, noises and power outage.

2.4.1 SAGS

Sags is the condition when the voltage in a system is lower than the stable

range which caused by power failures, down lines, utility recloser operations and

storms. In power system, sags are the most common problem compared to others and

it can be assume that voltage below 0.95p.u. is considered sags.. This problem can be

corrected by using backup power sources such as UPSs, generators or voltage

restoration technologies.

2.4.2 SURGES

The different between surges and sags is the surges is the condition when the

voltage in a system is above from the stable range. The voltage that considered as

stable range is between 0.95p.u. to 1.05p.u.. The affect of surges in a system is it will

damage the equipment used. They may be seen more frequently in facilities with

rapidly varying electrical loads, often caused by the switching on / off of electric

motors (inductive load switching). Air conditioners, electrical power tools, furnace

igniters or ignition systems, arc welders, electrostatic copy machines and elevators are

most likely to create surges.

2.4.3 TRANSIENTS

Transient can be define as a change happen in voltage causes by the short

duration and sharp impulse. In power system and if there is disturbance occur, a

transient voltage may exceed the normal voltage level by five or ten times. Present,

the transients normally caused by a lightning strike and the normal operation of

11

electrical equipment such as switching on/ off electrical motors. Normally, the

presence of transient voltage can only be detected with special monitoring equipment.

2.4.4 NOISES

Noises can be classified as interferences that can be generated by any electrical

equipment. Usually the noise comes from equipment that not being installed correctly

and properly. This equipment may include: radio transmitters, fluorescent lights,

computers, business machines and even simple devices such as light sockets, wall

receptacles, plugs and loose electrical connections. These types of disturbances can

result in computer errors.

2.4.5 POWER OUTAGE

In power system, power outage can be defined as total losses of power. This

condition can be momentarily or last for extended periods of time. Generally, the

power in generation system must be equal to the load demand by customer plus the

losses. So if the load demand increase higher than the generation can support, it may

lead to the power outage. Besides that, the power outage also can be caused by

electrical load switching in utility power stations. Even a momentary outage, of only a

fraction of a second, will affect a computer and can result in data loss and the need for

data re-entry or reprogramming.

2.6 UNDERVOLTAGE LOAD SHEDDING SCHEME

Theoretically, the philosophy of UVLS is that when there is a system

disturbance and the voltage drops to a pre-selected level for a pre-determined time,

then selected loads are shed. It means that when there is voltage instability occurs due

to a disturbance and the load shedding is performed, the voltage will recover to

acceptable level thereby avoiding a more widespread system voltage collapse.

Practically, combination between protection engineers and system planners, who

together can determine the amount of load and time in the shedding program is

required to develop the undervoltage load shedding program. The system planning

engineers will analyze numerous studies using P-V curves as well as other analytical

methods to determine the amount of loads that to be shed to retain voltage stability

under credible contingencies. Voltage collapse is most probable under heavy load

12

conditions where large amounts of power are to be transported from remote generation

sites and the bulk of the system load consists of motors.

In under voltage load shedding scheme, there are two types that being applied

in the system which are centralized and decentralized (distributed). A centralized

scheme is a method where it has undervoltage relay installed at key system buses

within the area and trip information is transmitted to shed load at various locations. As

the security is added to the system, sometimes the additional logic is applied. While a

decentralized scheme is where it has relays installed at the loads to be shed. The relays

will start to shed the load at the selected location when the voltage condition at the

locations begins to collapse. Moreover, this type of scheme is similar to the under

frequency load shedding schemes. Many of these schemes are categorized as “special

protection“or “wide area” protection schemes. These schemes require high-speed and

reliable communication to properly operate.

In [11, 14], it is been shown that load shedding is an effective counter measure

against voltage collapse. As been mention before, generally undervoltage load

shedding scheme is designed to shed a specific amount of load from one or more

locations within a power system after finite amount of time upon detecting the onset

of voltage collapse. There are three main areas for consideration in under voltage load

shedding which are the amount of load to shed, and the location where load is to be

shed.

2.6.1 THE AMOUNT OF LOAD TO BE SHED

Theoretically, there are many research indicate that as certain the amount of

load that is appropriate to shed under given conditions. If there is less load that been

trip than necessary, it is obvious that it would be not effective in arresting voltage

collapse. But if tripping too much load may result in transitioning the system from an

under-voltage to an over-frequency condition as the resulting system will have more

generation than load.

Load characteristic in power system play an important role in determining the

ability of the system to regain a stable equilibrium after a disturbance. The incorrect

presumption of load characteristics in load-flow and dynamic studies may render a

UVLS scheme ineffective and perhaps even inadvertently impose an over-frequency

13

condition upon the power system. In [14], the paper discuss about procedure to

calculate the amount of load to be shed where the amount of load to be shed is

calculated based upon the difference between the pre-contingency (steady-state)

power drawn by load and the instantaneous power drawn at the instant of system

disturbance. In this case, dynamic load model parameters are estimated on-line using a

non-linear least squares method in order to calculate the load shed amount.

In an actual power system, the granularity with which load can be shed is

limited due to pragmatic considerations. In general, the smallest block of load that can

be shed is equal to the load served through one substation-class distribution breaker

since it is this breaker that is employed to interrupt the load. Furthermore, the

distribution feeders served out of a particular substation in most cases have different

aggregate load characteristics and demand profiles making the predetermination of the

amount of load available for shedding challenging. This means that the design of a

UVLS should incorporate the impact of errors as a result of the differences between

the load that is presumed to be shed and the load that is actually shed.

The design should also take into account the impact of intentional load

shedding on distribution feeders serving, for example, police and fire stations,

hospitals, schools, power plant or bulk transmission system control centers, prisons

and army bases.

2.6.2 THE LOCATION WHERE LOAD IS TO BE SHED

An important factor to consider within a UVLS design is the location where

load is shed. Small disturbance analysis coupled with dynamic simulation and in some

cases optimal power flow methodology is some tools employed in the determination

of the location of load shed [15]. In this case, the load buses are ranked in the order of

the weakest to the strongest. The weakest bus tends to have the highest component

and tends to be most susceptible to voltage collapse given the relatively large reactive

power consumption for a small reduction in bus voltage. Therefore, often it is this bus

that is the most appropriate candidate for load shedding initially.

14

In [14], the proposed UVLS scheme detects voltage collapse at every bus in

the ten bus system considered. Rather than shedding load at the weakest one through

ranking buses, each bus is monitored for voltage collapse and upon detection of this,

the UVLS is triggered at that bus. A major drawback to this approach, as noted by the

authors of the paper, is that the optimum amount of load will not be shed given that

the power-voltage characteristics of the lines would change upon load shed at one bus.

Furthermore, this approach does not distinguish between the bus at which the reactive

power demand is increased and the adjacent buses whose voltages follow suit. This

means that the load at adjacent buses may be shed in the case where load rejection at

the weakest bus alone would have arrested voltage collapse.

In [15], the system overcomes this approach by pre-determining the weakest

buses in the system under various contingencies (N-1, N-2 and N-3). In this instance,

training scenarios consisting of eight different system configurations were subjected to

these contingencies. The resulting unstable scenarios were identified and the weakest

buses were noted for each unstable scenario. This was followed by a common ranking

of the load buses as it was postulated that the optimal load shedding locations will be

nearly the same for all unstable scenarios of the set. The preceding approach identifies

the common weakest buses for all conceivable contingencies and optimizes the

location of the load to be shed.

2.7 CONCEPT OF UNDER VOLTAGE LOAD SHEDDING

15

When a transmission system becomes stressful due to the overload, the

voltage instability or voltage collapse could be experience by the system [16]. The

philosophy of UVLS is that whenever the system is perturbed and voltage drops to

a certain pre-selected level for a certain pre-determined time period, then selected

loads may be cut off [17]. In some research, by shedding some of the loads in a

system the voltage magnitude will recover to its normal level. In practical, load

shedding schemes requires coordination between protection engineers and system

planners to set up the amount to be shed without affecting its security.

Data below shows the acceptable range value to be the reference during

generation of data. Some function has been created to devide the results to

determine their range of stable voltage.

Min(Vm)<0.95 = unstable

0.95<min(Vm)<1.05 = stable

2.8 FUZZY LOGIC

Seminal paper on fuzzy logic was introduced by Prof. Lofti A. Zadeh

in 1965 [19].Since then, many developments have taken place in different

parts of the world. Since the 1970s Japanese researchers have been the primary

force in the implementation of fuzzy theory and now have thousands of patents

in the area.

The world response to fuzzy logic has been varied. On the one hand,

western cultures are mired with the yes or no, guilty or not guilty, of the binary

Aristotelian logic world and their interpretation of the fuzziness causes a

conflict because they are given a negative connotation. On the other hand,

Eastern cultures easily accommodate the concept of fuzziness because it does

not imply disorganization and imprecision in their languages as it does in

English.

Practically, fuzzy logic is a powerful problem-solving methodology with a

myriad of applications in embedded control and information processing. Fuzzy

provides a simple way to draw definite conclusions from vogue, ambiguous or

imprecise information [20]. Moreover in computational intelligence, fuzzy logic

16

resembles human decision making with its ability to work from approximate data and

find precise solutions

Classical set theory is based on the fundamental concept of a set, in

which individuals are either a member or not a member. A sharp, crisp, and

ambiguous distinction exists between a member and a non-member for any

well-defined set of entities in this theory, and there is a very precise and clear

boundary to indicate if an entity belongs to a set. Thus, in classical set theory

an element is not allowed to be in a set (1) or not in a set (0) at the same time.

This means that many real-world problems cannot be handled by classical set

theory. On the contrary, the fuzzy set theory accepts partial membership values

μ ƒ ϵ [0, +1], and therefore, in a sense generalizes the classical set theory to

some extent.

As Prof. Lotfi A. Zadeh suggests by his principle of incompatibility:

“The closer one looks at a real-world problem, the fuzzier becomes the

solution,” and thus, imprecision and complexity are correlated [21].

Complexity is inversely related to the understanding we can have of a problem

or system. When little complexity is presented, closed-loop forms are enough

to describe the systems. More complex systems need methods such as neural

networks that can reduce some uncertainty. When systems are complex

enough that only few numerical data exist and the majority of this information

is vague, fuzzy reasoning can be used for manipulating this information.

2.8.1 Concept of Fuzzy Logic

A simple way to define fuzzy logic is logical system which is the

extension of mutivalued logic. In a specific way, fuzzy logic is almost

synonymous with the theory of fuzzy sets, a theory which relates to classes of

object with unsharp boundaries in which membership is a matter of degree

[25]. Fuzzy inference is the process of formulating the mapping from a given

input to an output using fuzzy logic. The mapping then provides a basis from

which decisions can be made.

In MATLAB, fuzzy logic can be implementing by using fuzzy logic

toolbox as the decision making. The fuzzy logic system consists of three parts

which are fuzzification, fuzzy inference and defuzzification [22, 23]. In

17

fuzzification, it will involve the process of transforming input variable into a

membership for linguistic terms of fuzzy sets. While fuzzy inference system is

used as a drawing conclusion from the set of fuzzy rules. The fuzzy rule is a

set of if-then linguistic term [15]. For the defuzzification, it converts

the fuzzy output values back into output actions. In this paper,

fuzzy logic algorithm is allowed to determine the suitability of

each bus and the one with the highest suitability chosen for

load shedding. The FLS Editor displays general information

about the fuzzy inference system.

2.8.2 Membership Function

The purposed of membership function is to determine or find the input and

output of the system. It is a curve that defines how each point in the input space is

mapped to a membership value between 0 and 1.It is the first step of the fuzzy logic

control process where a fuzzy algorithm categorises the information entering a system

and assigns values that represent the degree of membership in those categories.

In fuzzy logic, the membership function is a graphical representation of the

magnitude of participation of each input. It associates a weighting with each of the

inputs that are processed, defined functional overlap between inputs, and determines

and output response. The rules use the input membership values as weighting factors

to determine their influence on the fuzzy outputs sets of the final output conclusion.

Once the the functions are inferred, scaled, and combined, they are defuzzified into

crisp output which drives the system.

Input membership functions themselves can take any form the designer of the

system requires triangles, trapezoids, bell curves or any other shape as long as those

shapes accurately represent the distribution of information within the system, and as

long as a region of transition exists between adjacent membership functions.

18

Figure 1: Membership function

Due to their simple formulas and computational efficiency, both

triangular membership functions and trapezoidal membership functions have

been used extensively, especially in real-time implementation. However since

the membership functions are composed of straight-line segments, they are not

smooth at the switching points specified by the parameters.

2.8.3 Fuzzy Inference System

Fuzzy inference is the process of formulating the mapping from a given

input to an output using fuzzy logic. The mapping then provides a basis from

which decision can be made, or pattern discerned. The process of fuzzy

inference involves all of the pieces that are described which are the

membership function or fuzzification, fuzzy logic operators, and the fuzzy

rules. In fuzzy logic toolbox, there are two type of fuzzy logic system which is

Mamdani and Sugeno.

Mamdani’s fuzzy inference method is the most commonly used in

fuzzy methodology. Mamdani’s method is among the first control systems

built using fuzzy set theory. It was proposed in 197 5 by Ebrahim Mamdani as

an attempt to control a steam engine and boiler combination by synthesizing a

set of linguistic control rules obtained from experienced human operators.

Mamdani’s effort was based on Lofti Zadeh’s 1973 paper on fuzzy algorithm

for complex systems and decision processes. Although the inference process

19

describe differs from the methods described I the original paper, the basic idea

is much the same.

Mamdani type expects the output membership functions to be fuzzy

sets. After the aggregation process, there is fuzzy set for each fuzzy output

variable that needs fdefuzzification. It is possible and in many cases much

more efficient to use a single spike as the output membership functions rather

than a distributed fuzzy set. This is sometimes known as a singleton output

membership function and it can be thought of as a pre-defuzzified fuzzy set. It

enhance the efficiency of the defuzzification process because it greatly

simplifies the computation required by the more general Mamdani method,

which finds the centroid of a two dimensional function. Rather than integrating

across the two-dimensional function to find the centroid, that used the weight

average of a few data points. Sugeno type systems support this type of model.

Generally, Sugeno type systems can be used to model any inference system in

which the output membership functions areeither linear or constant.

2.8.4 Defuzzification

A defuzzifiation process is use to obtain the crisp output. This result is

obtained from fuzzy inference system where it maps an input vector to a crisp output.

The input to the defuzzification process is a fuzzy set (the aggregated output fuzzy

set), and the output of the defuzzification process is a single number. Many

defuzzification techniques have been proposed in the literature. The most commonly

used method is the centroid. Other methods include the maximum, the means of

maxima, height, and modified height method.

20

CHAPTER 3.0

METHODOLOGY

3.1 INTRODUCTION

The objective of a UVLS scheme is to restore reactive power balance in

the power system, to prevent voltage collapse and to keep a voltage problem

within a local area rather than allowing it to spread out by shedding some loads

[18]. In power system, the power generated by the generation system must be

equal to the load demand and the total losses. If the load demand is higher than

the generation can support, the it may lead to voltage collapse. To solve the

problems, fuzzy logic has been implemented to the UVLS schemes by testing

it o IEEE 30 bus system that has six generators, four under load tap changing

transformers, two shunt capacitor and thirty seven lines. The program for the

implementation of fuzzy logic to the UVLS schemes was done using

MATLAB programming.

3.2 UNDERVOLTAGE LOAD SHEDDING SCHEME

3.2.1 Preparing System Data

The test system used in this study is the IEEE-30 RTS. The system has

six generators, four under load tap changing transformers, two shunt capacitor

and thirty seven lines. In the base case the total system load is 2.834 pu, the

swing bus (bus number 1) generates real power of 2.5687 pu, while the other

21

generators generate 0.4 pu real power. Figure 1 illustrates the single line

diagram of IEEE 30-bus system.

Figure 0-3: IEEE 30-BUS TEST SYSTEM

(Source: ljs.academicdirect.org)

Fuzzy logic load shedding need data to perform the rules. As IEEE 30

bus system, it has five generator buses, one slack bus, 6

intermediate buses and eighteen load buses. The load

shedding technique is performed by creating several

conditions. Load factor is increased in order to indicate load

variation in the system. By increasing the load bus, the system

stability will change and may cause voltage instability due to

the load increase in the system. Load shedding then perform

by selecting the weakest bus.

22

http://ljs.academicdirect.org/A14/204_220.htm

3.2.2 Power Flow Analysis

Power flow analysis, coomonly referred as load flow is an important

tool of power system analysis and design. It is use for planning, operation and

economic scheduling. In this project, the transmission system is modelled by a

set of buses or nodes interconnected y transmissionlink. Generators and loads,

connector to various nodes of the system inject and remove power from the

transmission system. However in power system, power are known rather than

current. So the resulting equation is in terms of power, known as the power

flow equation become non-linear and must be solve by iterative solution. The

most common technique used for the iterative solution of non-linear algebraic

equation are Newton-Raphson, Gauss-Seidel and Quasi-Newton methods.

Newton-Raphson load flow is implement in this project to get the desired

output.

3.2.2.1 Newton-Raphson load flow

In power system, this method is widely used to solve the simultaneous algebraic

equations. Newton-Rahpson method is a successive approximation procedure based

on an initial estimate of the unknown and the use of Taylor’s series expension.

Because of its quadratic convergence, Newton-Raphson method is mathematically

superior to the Gauss-Seidel method and is less prone to divergence with ill-

conditioned problems. For large power systems, the Newton-Raphson method is found

to be more efficient and practical. The number of iterations required to obtain a

solution is independent of the system size, but more functional evaluations are

required at each iteration. Since the power flow problem, real power and voltage

magnitude are specified for the voltage-controlled buses, the power flow equation is

formulated in polar form.

3.2.3 Load Shedding Algorithm

Load shedding algorithm involves several steps before it complete.

First thing to do is by creating the cases. Generally, it is a fact that in base

case, the system is already stable. UVLS scheme is perform when the system

23

in unstable. To makes system unstable, the values of the load at the load bus

need to be increased. The load bus consists of active power (Pd) and reactive

power (Qd).

After increasing the value of voltage in the load bus, the Newton-Raphson load

flow will execute. Upon execution, the programme will read the bus data of IEEE 30

RTS and run the power flow solution. The programme also will call necessary

routines to display the desired results. Upon completion of the load flow, the

programme will display minimum voltage in the system, the new value of the load,

and the minimum voltage at each bus. From the results, it can be determine whether

the system become unstable or not. If the system becomes unstable, then the UVLS

scheme can be perform. If the system still stable, then the programme will increase

again the load until the system become unstable. For this project, several cases is

developed to perform analysis of the system.

As been discussed, UVLS scheme is perform by selecting the appropriate

location for load shedding before it can shed the load. In IEEE 30 bus RTS, the

programme will determine the weakest bus in the system to perform the load

shedding. When the load at the weakest bus has been shed, then the Newton-Raphson

will run again the power flow solution to determine whether the system become stable

or not. If yes, it means that the programme is success but if no the programme need to

shed again the load in the system so that the system comeback to the stable state. The

step to perform the UVLS scheme can be found in Figure 4.

The used of fuzzy logic system in this project is it will implement the method

of undervoltage load shedding by determining the suitable location of load shedding

and how much the value of the load that need to be shed. The fuzzy logic will use

IF… THEN rules to control the input and output variable.

24

Figure 0-4: UVLS flowchart

25

3.3 FUZZY LOGIC SYSTEM

When there is disturbance or an unexpected event occurs in a large network of

power system, in some cases the probability and uncertainties of the incidence

represented. However, it is made clear that some of the uncertain functions are

intrinsically fuzzy in nature and difficult to handle to handle effectively by

probability. By using fuzzy logic, it provide a good solution that is not easily solve by

other methods and are readily applicable to power system problems [24]. The function

of fuzzy logic in this project as been mention before is to determines the suitability of

each bus for load shedding where the bus with the lowest value of the voltage is

chosen as the most appropriate bus for load shedding. The same system is develop in

fuzzy logic to determine the amount of the load that need to be shed where the fuzzy

will decide how much of the amount in the particular bus that need to be shed to

restore back the system into stable condition. The input variable and the output has

been develop and the rules has been created.

3.3.1 Bus Selection for load shedding

The fuzzy FIS Editor for selected bus load shedding is illustrated in Figure 3, 4

and 5. The input variable of the FIS Editor are voltage magnitude (Vm) which divides

into four categories as shown in Figure 4 and loading factor in Figure 3. The output is

the percentage of selected bus to be shed also divides into 4 categories as shown in

Figure 5.

Inputs: Lf (Loading factor) Trapezoidal membership function as shown in Figure 3

VM (Voltage Magnitude) Triangle membership function as shown in Figure 4.

Output: PSBLS (Percentage Selected Bus Load Shedding) Triangle membership

function as shown in Figure 5.

26

Figure 0-5: loading factor

Figure 0-6: Voltage Magnitude (Vm)

Figure 0-7: Percentage selected bus

In fuzzy logic system, an IF-THEN basic rule-based system is used. IF

statement is refer as antecendent while THEN statement is as consequent. In this

section, fuzzy rule system is determined to form decision on the fuzzy input derived

from the voltage magnitude and loading factor. For this fuzzy to find the selected bus

for load shedding, 4 rules were developed which are:

Rule 1: IF loading factor is 1.1 AND Voltage magnitude is low THEN percent

selected bus is high

Rule 2: IF loading factor is 1.1 AND Voltage magnitude is medium THEN percent

selected bus is high-medium

Rule 3: IF loading factor is 1.1 AND Voltage magnitude is high-medium THEN

percent selected bus is medium

Rules 4: IF loading factor is 1.1 AND Voltage magnitude is high THEN percent

selected bus is low

27

As illustrated in Figure 10, the rows of plots represent the rules while the

columns represent the variables. The first two columns of plots (yellow) show the

membership functions of the input variables, while the fourth column of plots (blue)

shows the membership functions of the output.

3.3.2 Algorithm for Amount of Load to be Shed

The fuzzy FIS Editor for selected bus load shedding is illustrated in

Figure 6, 7, 8 and 9. The input variable of the FIS Editor are voltage

magnitude (Vm), active power (Pd) and reactive power (Qd) while the output

is the percentage amount of load to be shed.

Inputs: Pd (Active Power) Triangle membership function as shown in Figure 6

Qd (Reactive Power) Triangle membership function as shown in Figure 7

VM (Voltage Magnitude) Trapezoidal membership functions as shown in

Figure 8.

Output: PAL (Percentage amount of load to be shed) Triangle membership function

as shown in Figure 9.

28

Figure 0-8: Active Power (Pd)

Figure 0-9: Reactive Power (Qd)

Figure 0-10: Voltage Magnitude (Vm)

Figure 0-11: Percentage Amount Load to be shed

The fuzzy analysis of this method was developed using the same technique

as described in the previous method. The fuzzy rules to find the amount load to

be shed are as in Table 1. It is necessary to establish a meaningful system for

representing the linguistic variables in the matrix. For this case the following

will be used:

29

Table 1: Fuzzy decision matrix

AND

Voltage Magnitude

L LM M HM H

Pd,

Qd

L M M LM LM L

LM HM M M LM LM

M HM HM M M LM

HM H HM HM M M

H H H HM HM M

“L”: “low”

“LM”: “low-medium”

“M”: “medium”

“HM”: “high-medium”

“H”: “high”

25 fuzzy rules are derived and here are some examples of the fuzzy rules

listed in Table 1:

Rule 1: IF Pd is low AND Qd is low AND Voltage magnitude is low THEN

percent load to be shed is medium

Rule 2: IF Pd is low-medium AND Qd is low-medium AND Voltage

magnitude is low THEN percent load to be shed is high-medium

When fuzzy rule has multiple antecedents or input variable, the fuzzy operator

AND for minimization operator is used to obtain a single number that represents the

result of the antecedent evaluation. Fuzzy rules involve the operations between input

fuzzy sets, as illustrated graphically in Figure 10. It is based on fuzzy inference

described previously.

30

Figure 0-12: Fuzzy rules analysis

As illustrated in Figure 10, the rows of plots represent the rules while the

columns represent the variables. The first three columns of plots (yellow) show the

membership functions of the input variables, while the fourth column of plots (blue)

shows the membership functions of the output.

31

CHAPTER 4.0

RESULTS AND DISCUSSIONS

4.1 INTRODUCTION

Based on the developed methodology, all the results and discussion of

the Under-voltage Load Shedding scheme are presented in this chapter. The

results include output of fuzzy logic system to determine the bus Selection for

load shedding and the amount that need to be shed to stabilize the

system.Several cases has been selected to represent the output of the program.

The cases includes of loading factor from base case until loading factor = 2.

This study is conduct to show how reliable of computational intelligence

system to perform the load shedding in a system.

1.1.1. loading factor = 1.4

Determination of selected bus for load shedding using the proposed fuzzy

system is shown in Figure 11. This fuzzy system performed by referring the data from

the load flow results as shown in Table 2. Loading factor = 1.4 is selected as the test

case conditions.

Table 2: Fuzzy output for selected bus of load shedding at loading factor 1.4

B

us

N

o.

Minim

um

Voltag

e (Vm)

Percent

age

selected

bus (%)

B

us

N

o.

Minim

um

Voltag

e (Vm)

Percent

age

selected

bus (%)

1 1.0600 27.2 16 1.0098 33.1

2 1.0230 32.4 17 1.0010 33.3

3 0.9983 34.2 18 0.9848 39.7

4 0.9859 39.3 19 0.9806 41.2

32

5 0.9800 41.4 20 0.9864 39.1

6 0.9850 39.6 21 0.9899 37.8

7 0.9710 44.2 22 0.9907 37.5

8 0.9900 37.8 23 0.9829 40.4

9 1.0256 32.2 24 0.9730 43.6

10 1.0082 33.2 25 0.9706 44.3

11 1.0820 20.7 26 0.9444 51.5

12 1.0291 31.9 27 0.9818 40.8

13 1.0610 27 28 0.9820 40.7

14 1.0068 33.2 29 0.9520 49.5

15 0.9996 33.5 30 0.9348 54.1

Based on the newton-raphson load flow results, it is shown that the minimum

voltage of the system has dropped below the stable condition which at 0.9348 p.u..

Fuzzy logic system operates by using the data from newton-raphson load flow results

to determine the suitable bus for load shedding.

Figure 11: FIS for Selected Bus for Load Shedding at Bus (30) with 1.4 loading factor

As illustrated in Table 2, it shows that bus 30 is the weakest in the system.

Therefore bus 30 is selected as the appropriate bus to perform the load shedding.

Figure 12 shows the fuzzy based load shedding system to determine the amount load

to be shed at bus 30.

33

At bus 30 with 1.4 loading factor, the value of active power and reactive power

are 14.84MW and 2.66MVAR respectively.

Figure 13: FIS for percentage amount to be shed at Bus (30) with loading factor = 1.4

Based on the Figure 12, it is shown that fuzzy based load shedding system

decided to shed the amount of load up to 65.1%. As the results, the amount of load to

be shed is 9.6608MW and 1.7317MVAR. The minimum voltage at the system

increases from 0.9348 p.u. to 0.9631 p.u. at bus 26. It is shown that the system is

improved to a stable condition.

As the result, the total amount load to be shed in loading factor = 1.4 is 9.6608MW

and 1.7317MVAR. At loading factor = 1.4, the program need to perform 1 stage of

load shedding to get back the system to a stable condition.

1.1.2.loading factor = 1.5


system is shown in Figure 12. This fuzzy system is performed by referring to the data

from the load flow results as shown in Table 3. Loading factor = 1.5 is selected as the

test case conditions.


Bus

No.

Minimum

Voltage (Vm)

Percentage

selected bus (%)

Bus

No.

Minimum

Voltage (Vm)

Percentage

selected bus (%)

34

1 1.060 27.2 16 0.9969 34.8

2 1.0230 32.4 17 0.9880 38.5

3 0.9916 37.1 18 0.9697 44.6

4 0.9782 42 19 0.9655 45.8

5 0.9800 41.4 20 0.9719 43.9

6 0.9765 42.5 21 0.9761 42.6

7 0.9649 45.9 22 0.9768 42.4

8 0.9800 41.4 23 0.9674 45.2

9 1.0173 32.8 24 0.9569 48.1

10 0.9961 35.2 25 0.9546 48.8

11 1.0820 20.7 26 0.9260 56.7

12 1.0171 32.8 27 0.9671 45.3

13 1.0510 28.9 28 0.9721 43.9

14 0.9930 36.5 29 0.9344 54.2

15 0.9853 39.5 30 0.9155 60.1


voltage of the system has dropped below the stable condition which is 0.9155 p.u.

Fuzzy logic system operated by using the data from newton-raphson load flow results

to determine the selected bus for load shedding.

Figure 14: FIS for Selected Bus for Load Shedding at Bus (30) with loading factor =

1.5

35

From the result above, it shows that bus 30 is the weakest in the system.




At bus 30 loading factor = 1.5, the value of active power and reactive power





be shed is 11.3526MW and 2.0349MVAR. The minimum voltage in the system

increase from 0.9155 p.u. to 0.9469 p.u. which occur at bus 26. It shows that the

system is still in unstable condition.

Therefore the program decides to shed load at bus 26 as the next stages to

make the system in stable condition. At bus 26 with loading factor= 1.5, the value of

active power and reactive power are 5.25MW and 3.45MVAR respectively.

36




be shed is 3.108MW and 2.0424MVAR. The minimum voltage in the system increase

from 0.9469 p.u.to 0.9700p.u. which occur at bus 19. It is shown that the system is


As the result, the total amount load to be shed in loading factor = 1.5 is

14.4606MW and 4.0773MVAR. At loading factor = 1.5, the program need to perform

2 stages of load shedding to get back the system to a stable condition.







Bus

No.

Minimum

Voltage (Vm)

Percentage

selected bus (%)

Bus

No.

Minimum

Voltage (Vm)

Percentage

selected bus (%)

1 1.06 27.2 16 0.9845 39.8

2 1.013 33 17 0.9758 42.7

3 0.9848 39.7 18 0.9552 48.6

4 0.9704 44.4 19 0.951 49.7

37

5 0.97 44.5 20 0.9581 47.8

6 0.9704 44.4 21 0.9633 46.4

7 0.9561 48.4 22 0.964 46.2

8 0.98 41.4 23 0.9527 49.3

9 1.01 33.1 24 0.9421 52.1

10 0.985 39.7 25 0.9408 52.5

11 1.082 20.7 26 0.9097 62.4

12 1.0052 33.3 27 0.9552 48.6

13 1.041 30.5 28 0.9662 45.6

14 0.9794 41.6 29 0.9196 58.8

15 0.9715 44 30 0.899 66.7





Figure 175: FIS for Selected Bus for Load Shedding at Bus (30) with loading factor =

1.6





38





decided to shed the amount of load up to 75%. As the results, the amount of load to be

shed is 12.72MW and 2.28MVAR. The minimum voltage increased in the system

increase from 0.899 p.u. to 0.9183 p.u. which occur at bus 26. It shows that the

system is still in unstable condition.




39





from 0.9381 to 0.9644p.u.. which occur at bus 24. It is shown that the system is



16.276MW and 4.6168MVAR. At loading factor = 1.6, the program need to perform 2

stages of load shedding to get back the system to a stable condition.







Bus

No.

Minimum

Voltage (Vm)

Percentage

selected bus (%)

Bus

No.

Minimum

Voltage (Vm)

Percentage

selected bus (%)

1 1.06 27.2 16 0.9708 44.3

2 1.013 33 17 0.962 46.8

3 0.9778 42.1 18 0.9391 52.9

4 0.9623 46.7 19 0.9348 54.1

5 0.97 44.5 20 0.9426 52

6 0.9617 46.8 21 0.9484 50.4

7 0.9498 50.1 22 0.9492 50.2

8 0.97 44.5 23 0.9361 53.8

9 1.0011 33.3 24 0.9248 57.1

10 0.9721 43.9 25 0.9235 57.5

11 1.082 20.7 26 0.8897 66.9

12 0.9926 36.7 27 0.9392 52.9

13 1.031 31.7 28 0.956 48.4

14 0.9648 46 29 0.9003 66.5

15 0.9563 48.3 30 0.8778 67.6

40





Figure 208: FIS for Selected Bus for Load Shedding at Bus (30) with loading factor = 1.7







41




be shed is 13.533MW and 2.4257MVAR. The minimum voltage increased in the

system increase from 0.8778 p.u. to 0.9183 p.u. which occur at bus 26. It shows that

the system is still in unstable condition.





42



shed is 4.4625MW and 2.9325MVAR. The minimum voltage in the system increase

from 0.9183 to 0.9497p.u.. which occur at bus 19. It is shown that the system is still

not improved to the stable condition.








increase from 0.9497 to 0.9597p.u.. which occur at bus 24. It is shown that the system

is improved to a stable condition.


27.2979MW and 8.6875MVAR. At loading factor = 1.7, the program need to perform

3 stages of load shedding to get back the system to a stable condition.




43




Bus

No.

Minimum

Voltage

(Vm)

Percentage

selected bus

(%)

Bus

No.

Minimum

Voltage

(Vm)

Percentage

selected bus

(%)

1 1.06 27.2 16 0.9589 47.6

2 1.013 33 17 0.9503 49.9

3 0.9731 43.6 18 0.925 57

4 0.9571 48.1 19 0.9207 58.4

5 0.96 47.3 20 0.9293 55.7

6 0.9573 48 21 0.9361 53.8

7 0.942 52.1 22 0.9369 53.5

8 0.97 44.5 23 0.9217 58.1

9 0.9946 35.9 24 0.9103 62.1

10 0.9616 46.9 25 0.91 62.2

11 1.082 20.7 26 0.8735 67.9

12 0.9815 40.9 27 0.9277 56.2

13 1.021 32.5 28 0.9513 49.7

14 0.9518 49.5 29 0.8855 67.1

15 0.943 51.9 30 0.8612 69.2





44


From the Table 3, it shows that bus 30 is the weakest in the system. Therefore

bus 30 is selected as the appropriate bus to perform the load shedding. Figure 23

shows the fuzzy based load shedding system to determine the amount load to be shed

at bus 30.







45















46






not improved to a stable condition.





47




from 0.9485 to 0.9659p.u.. which occur at bus 7. It is shown that the system is still not



40.0991MW and 17.1099MVAR. At loading factor = 1.8, the program need to

perform 4 stages of load shedding to get back the system to a stable condition.

1.1.6. loading factor = 1.9






Bus

No.

Minimum

Voltage (Vm)

Percentage

selected bus (%)

Bus

No.

Minimum

Voltage (Vm)

Percentage

selected bus (%)

1 1.06 27.2 16 0.9474 50.7

2 1.003 33.3 17 0.9371 53.5

3 0.9638 46.3 18 0.9105 62

4 0.9464 51 19 0.9055 64.1

5 0.96 47.3 20 0.9145 60.5

6 0.9466 50.9 21 0.9212 58.2

7 0.9346 54.2 22 0.922 58

8 0.96 47.3 23 0.9067 63.6

9 0.9851 39.6 24 0.893 66.8

10 0.9487 50.3 25 0.8915 66.8

11 1.082 20.7 26 0.852 70.6

12 0.9726 43.7 27 0.9099 62.3

48

13 1.021 32.5 28 0.9396 52.8

14 0.9404 52.6 29 0.8639 68.9

15 0.9305 55.4 30 0.8374 73.4









at bus 30.



49








make the system in stable condition. At bus 26 with loading factor = 1.9, the value of



50


















51





increase from 0.9365 to 0.9641p.u.. It is shown that the system is improved to a stable

condition.




1.1.7. loading factor = 2






Bus

No.

Minimum

Voltage

Percentage

selected bus

Bus

No.

Minimum

Voltage

Percentage

selected bus

52

(Vm) (%) (Vm) (%)

1 1.06 27.2 16 0.9397 52.8

2 1.003 33.3 17 0.9283 56

3 0.96 47.3 18 0.9002 66.6

4 0.9424 52 19 0.8947 66.7

5 0.96 47.3 20 0.9043 64.6

6 0.9436 51.7 21 0.9112 61.8

7 0.9317 55 22 0.912 61.5

8 0.96 47.3 23 0.8961 66.7

9 0.9799 41.4 24 0.881 67.3

10 0.9405 52.6 25 0.8796 67.4

11 1.082 20.7 26 0.8372 73.5

12 0.967 45.4 27 0.8994 66.7

13 1.021 32.5 28 0.9359 53.8

14 0.9325 54.8 29 0.8498 70.9

15 0.9218 58 30 0.8213 78.1





53





at bus 30.











active power and reactive power are 7MW and 4.6MVAR respectively.

54









active power and reactive power are 19MW and 6.8MVAR respectively.


55













increase from 0.9247 to 0.9524 p.u.. It is shown that the system is improved to a stable

condition.




2.1.

56

CHAPTER 5.0

CONCLUSIONS

In this paper, particle swarm optimization (PSO) technique was use to solve the unit

commitment problem with several constraint as stated before. The result shows that

the proposed method was capable of obtaining optimum operating cost for UC

problem for 24 hour period interval of load demand.. In addition, the wind turbine

generator was attached to improve an operating cost. The wind generator was

successfully shown the effectiveness in minimizing the operating cost. Thus, the

purpose of unit commitment to meet a demand with minimum cost has been achieved.

For recommendation, PSO can be combined with another algorithm such as

Evolutional Programming (EP), Ann Colony and Bee Colony to improve a

performance of the technique to solve unit commitment problem. Moreover, the others

green energy such as solar and nuclear can be implement to study their effect toward

the UC problem.

57

CHAPTER 6.0

RECOMMENDATIONS FOR FUTURE WORKS

There are several addition and development that can be done on DED problems in

order to have high quality and accurate solutions. The improvement that can be done

is such as taking into account other generator constraints such as spinning reserve

requirement and emission constraint. All the constraints will give more accurate result

to the solution of DED problems. In addition, accurate modeling of DED problem will

be improved when the valve point loadings effects in the generating units are taken

into account. Valve point effect are are usually modelled in two form which is i)

consider the prohibited zones as the inequality constraint and ii) implement the effect

as the non-smooth cost function for the fuel cost function[10]

58

REFERENCE

[1] Atputharajah, Arulampalam, and Tapan K. Saha. "Power system blackouts-Literature review." Industrial and Information Systems (ICIIS), 2009 International Conference on. IEEE, 2009.

[2] Taylor, Carson W. "Concepts of undervoltage load shedding for voltage stability." Power Delivery, IEEE Transactions on 7.2 (1992): 480-488.

[3] Calderaro, V., Galdi, V., Lattarulo, V., & Siano, P. (2010). A new algorithm for steady state load-shedding strategy. Optimization of Electrical and Electronic Equipment (OPTIM), 2010 12th International Conference on, 48-53.

[4] P. Kundur, Power System Stability and Control, vol. IV. New York: McGraw Hill, 1994, pp. 959- 1024

[5] Kessel, P., and H. Glavitsch. "Estimating the voltage stability of a power system." Power Delivery, IEEE Transactions on 1.3 (1986): 346-354.

[6] Saffarian, Alireza, and Majid Sanaye-Pasand. "Enhancement of power system stability using adaptive combinational load shedding methods." Power Systems, IEEE Transactions on 26.3 (2011): 1010-1020.

[7] Wang, Y., Pordanjani, I. R., Li, W., Xu, W., & Vaahedi, E. (2011). Strategy to minimise the load shedding amount for voltage collapse prevention. Generation, Transmission & Distribution, IET, 5(3), 307-313.

[8] Kadam, D. P., et al. "Fuzzy Logic Algorithm for Unit Commitment Problem."Control, Automation, Communication and Energy Conservation, 2009. INCACEC 2009. 2009 International Conference on. IEEE, 2009.

[9] Abdelaziz, A. Y., et al. "Fuzzy based load shedding approach against voltage instability." International Journal of Engineering, Science and Technology 4.3 (2013): 15-44.

[10]Terzija, Vladimir V. "Adaptive underfrequency load shedding based on the magnitude of the disturbance estimation." Power Systems, IEEE Transactions on 21.3 (2006): 1260-1266.

[11]C. W. Taylor, Power System Voltage Stability, McGraw-Hill, 1994.[12]A. Wiszniewski, “New criteria of voltage stability margin for the purpose of load

shedding,” IEEE trans.Power del., vol. 22, no. 3, July 2007, pp. 1367-1371.[13] A. Guzmán, D. Tziouvaras, E. O. Schweitzer and Ken E. Martin, “Local and wide-

area network protectionsystems improve power system reliability,” Schweitzer Engineering Laboratories technical papers, 2004.

[14] R. Balanathan, N. Pahalawaththa, and U. Annakkage, “A strategy for undervoltage load shedding in power systems,” International Conference on Power System Technology, vol. 2, pp. 1494–1498, Aug. 1998.

[15] C. Moors, D. Lefebvre, and T. V. Custem, “Design of load shedding schemes against

voltage instability,” ser. 23-27, vol. 2, Power Engineering Society Winter Meeting, 2000. IEEE, Jan 2000, pp. 1495–1500.

[16] Verayiah, R., Ramasamy, A., Abidin, H. Z., & Musirin, I. (2009, December). Under Voltage Load Shedding (UVLS) study for 746 test bus system. In Energy and Environment, 2009. ICEE 2009. 3rd International Conference on (pp. 98-103). IEEE.

59

[17] M. Begovic, D. Fulton, M. R. Gonzalez, J. Goossens, E. A. Guro, R. W. Haas, C. F. Henville, G. Manchur, G. L. Michel, R. C. Pastore, J. Postforoosh, G. L. Schmitt, J. B. Williams, K. Zimmerman, and A. A. Burzese, "Summary of "System Protection and Voltage Stability"," IEEE Transactions on Power Delivery, vol. 10, pp. 631-638, 1995.

[18] Mozina, Charles. "Undervoltage load shedding." Power Systems Conference: Advanced Metering, Protection, Control, Communication, and Distributed Resources, 2007. PSC 2007. IEEE, 2007.

[19] Zadeh LA (1965) Fuzzy sets. Info Control 8(3):338–353[20] E. Cox. “ Fuzzy fundamentals” (IEEE Spectrum, October 1992, pp. 58-61).[21] Zadeh LA (1973) Outline of a new approach to the analysis of complex systems and

decision processes, IEEE Trans Syst Man Cyber SMC 3:28–44[22] Grewal, G.S.; Konowalec, J.W.; Hakim, M. “Optimization of a load shedding scheme” ,

Industry Applications Magazine, IEEE, vol 4, pp 25-30, July/August 1998

[23] Afiqah, R. N., Musirin, I., Johari, D., Othman, M. M., Rahman, T. K. A., & Othman, Z. (2009). Fuzzy logic application in DGA methods to classify fault type in power transformer. SELECTED TOPICS in POWER SYSTEMS and REMOTE SENSING, 83-88

[24] Momoh J. and Tomsovic K., 1995. Overview and literature survey of fuzzy set theory in power systems, IEEE Transactions on Power Systems, Vol. 10, No. 3, pp. 1676-1690.

[25] Naaz, Sameena, et al. "Effect of different defuzzification methods in a fuzzy based load balancing application." IJCSI International Journal of Computer Science Issues 8.5: 261-267.

APPENDICES

MATLAB PROGRAMMING

Main Program

60

http://ieeexplore.ieee.org.ezaccess.library.uitm.edu.my/xpl/RecentIssue.jsp?punumber=2943

Momoh J. and Tomsovic K., 1995. Overview and literature survey of fuzzy set theory in power systems, IEEE Transactions on Power Systems, Vol. 10, No. 3, pp. 1676-1690.clear, close allclc

pso.psoMethod = 'constriction'; pso.saveResults = 'true'; pso.maxIter = 10 pso.noParticles =30;M. Begovic, D. Fulton, M. R. Gonzalez, J. Goossens, E. A. Guro, R. W. Haas, C. F. Henville, G. Manchur, G. L. Michel, R. C. Pastore, J. Postforoosh, G. L. Schmitt, J. B. Williams, K. Zimmerman, and A. A. Burzese, "Summary of "System Protection and Voltage Stability"," IEEE Transactions on Power Delivery, vol. 10, pp. 631-638, 1995.

Mozina, Charles. "Undervoltage load shedding." Power Systems Conference: Advanced Metering, Protection, Control, Communication, and Distributed Resources, 2007. PSC 2007. IEEE, 2007.

Grewal, G.S.; Konowalec, J.W.; Hakim, M. “Optimization of a load shedding scheme” , Industry Applications Magazine, IEEE, vol 4, pp 25-30, July/August 1998

Afiqah, R. N., Musirin, I., Johari, D., Othman, M. M., Rahman, T. K. A., & Othman, Z. (2009). Fuzzy logic application in DGA methods to classify fault type in power transformer. SELECTED TOPICS in POWER SYSTEMS and REMOTE SENSING, 83-88.

pso.noVars = 6; pso.c1 = 2.05; pso.c2 = 2.05; pso.xMin = 0; pso.xMax = 1; pso.vMax = 1;pso.vMin = -1; pso.pMin = [100,50,80,50,50,50];pso.pMax = [500,200,300,150,200,120]; pso.consFactor = getConstrictionFactor(pso.c1,pso.c2);

saveStringInit = 'F:\Final Year Project\FYP azuwam\Matlab Programming\PSO editted azuwam.mat';

saveString = 'F:\Final Year Project\FYP azuwam\Matlab Programming\PSO editted azuwam.mat'; gBest = PSO(pso, seed, saveStringInit, saveString);

61

http://ieeexplore.ieee.org.ezaccess.library.uitm.edu.my/xpl/RecentIssue.jsp?punumber=2943

FinalResult;

PSO Main Program

function gBest =PSO(pso, seed, saveString1, saveString2)

gBest.fitness = 0;gBest.xVal = zeros(1, pso.noVars);

if strcmp(pso.saveResults,'true') gBest.hist = zeros(pso.maxIter, 2 + pso.noVars);end for i = 1 : 1 : pso.noParticles for j = 1 : 1 : pso.noVars particles(i).velocity(j) = rand; particles(i).xVal(j) = rand; particles(i).bestXVal(j) = particles(i).xVal(j); end % particles(i).fitness = fitnessFcn(outputPower(particles(i).xVal, pso));

particles(i).pBest = particles(i).fitness; end gBest.xVal = particles(1).xVal; gBest.fitness = particles(1).fitness; diff = 1000; for i = 2 : 1 : pso.noParticles if (abs(gBest.fitness - particles(i).fitness) < diff) gBest.fitness = particles(i).fitness; gBest.xVal = particles(i).xVal; diff = abs(gBest.fitness - particles(i).fitness); end if strcmp(pso.saveResults,'true') gBest.hist(1,:) = [0, gBest.fitness, gBest.xVal]; end

62

end gBest.iter = 0;

if strcmp(pso.saveResults,'true') save(saveString1); end

iter = 1; t = cputime; while ((iter ~= pso.maxIter) && (gBest.fitness < pso.objective)) for i = 1 : 1 : pso.noParticles for j = 1 : 1 : pso.noVars vid = particles(i).velocity(j); pid = particles(i).bestXVal(j); xid = particles(i).xVal(j); pgd = gBest.xVal(j); if strcmp(pso.psoMethod,'constriction') vid = pso.consFactor * (vid + pso.c1 * rand * (pid - xid) + pso.c2 * rand * (pgd - xid)); end if (vid > pso.vMax) vid = pso.vMax; elseif vid < pso.vMin vid = pso.vMin; end xid = xid + vid; if xid > pso.xMax xid = pso.xMax; vid = 0; elseif xid < pso.xMin xid = pso.xMin; vid = 0; end particles(i).velocity(j) = vid; particles(i).xVal(j) = xid;

63

end end for i = 1:1:pso.noParticles particles(i).fitness = fitnessFcn(outputPower(particles(i).xVal, pso)); if (particles(i).fitness > particles(i).pBest) particles(i).pBest = particles(i).fitness; particles(i).bestXVal = particles(i).xVal; end if (particles(i).fitness > gBest.fitness) gBest.fitness = particles(i).fitness; gBest.xVal = particles(i).xVal; gBest.iter = iter if strcmp(pso.saveResults,'true') gBest.hist(iter + 1,:) = [iter, gBest.fitness, gBest.xVal]; end end end iter = iter + 1; end gBest.optTime = cputime - t; if strcmp(pso.saveResults,'true') save(saveString2); end

Fitness Ramp Rate Program

function [PD,PL,Fcost] = fitnessRamprate(Pi) pMin = [100,50,80,50,50,50];pMax = [500,200,300,150,200,120]; UR = [80,50,65,50,50,50]; DR = [120,90,100,90,90,90]; Po = [440,170,200,150,190,110];

64

Prmin = [320,80,100,60,100,20]; Prmax = [520,220,265,200,240,160]; for i = 1:1:6 if Pi(i) < pMin(i) fitness = 0; return; elseif Pi(i) > pMax(i) fitness = 0; return; endend for i = 1:1:6 if Pi(i) < Prmin(i) Pi(i)= Prmin; Po(i)= Pi(i); Prmin(i)= Po(i)-DR(i); elseif Pi(i) > Prmax(i) Pi(i)= Prmax; Po(i)= Pi(i); Prmax(i)= Po(i)+UR(i); endend B = [0.0017, 0.0012, 0.0007, -0.0001, -0.0005, -0.0002;... 0.0012, 0.0014, 0.0009, 0.0001, -0.0006, -0.0001;... 0.0007, 0.0009, 0.0031, 0.0000, -0.0010, -0.0006;... -0.0001, 0.0001, 0.0000, 0.0024, -0.0006, -0.0008;... -0.0005, -0.0006, -0.0010, -0.0006, 0.0129, -0.0002;... -0.0002, -0.0001, -0.0006, -0.0008, -0.0002, 0.0150]; Bo = (1.0e-03)*[-0.3908, -0.1297, 0.7047, 0.0591, 0.2161, -0.6635]; Boo = 0.0056; basemva = 100; PL = Pi*(B/basemva)*Pi'+Bo*Pi'+Boo*basemva; P = sum(Pi) - PL;PD = round(sum(P)); loaddemand = 955; if PD < loaddemand fitness = 0;

65

return;end cost = [240,7,0.007;... 200,10,0.0095;... 220,8.5,0.009;... 200,11,0.009;... 220,10.5,0.008;... 190,12,0.0075]; alpha = cost(:,1);beta = cost(:,2);gamma = cost(:,3); for i = 1:1:6 F(i) = alpha(i) + beta(i)*Pi(i) + gamma(i) * (Pi(i)^2);endFcost = sum(F);Ppbc = sum(Pi) - loaddemand - PL; fitness = 1 / (Fcost + Ppbc);

Constriction Factor Program

function consFactor = getConstrictionFactor(c1,c2) theta = c1 + c2; if theta <= 4 error('Theta must be more than 4.')end consFactor = 2/abs(2-theta-sqrt(theta^2-4*theta));

Output Power Program

function powerOut = outputPower(xVal, pso) for i = 1:1:pso.noVars powerOut(i) = (pso.pMax(i) - pso.pMin(i)) * xVal(i) + pso.pMin(i);end powerOut

Final Result Program

clear, close all

66

clc

load('F:\Final Year Project\FYP azuwam\Matlab Programming\PSO editted azuwam.mat'); psogBest = gBest.xVal;ccc = outputPower(psogBest,pso);psogBest = gBest.hist; disp(strcat('Optimization time:',num2str(gBest.optTime))); [PD, PL,Fcost] = fitnessRamprate(ccc);PDPLFcost

67

Documents

Computational Intelligence Based Tehnique for Load Shedding Scheme