OPTIMIZATION OF POWER DISTRIBUTION SYSTEM WITH DISTRIBUTED

GENERATION USING HARISH HAWKS OPTIMIZATION TECHNIQUES

(CASE STUDY: DEBRE BERHAN POWER DISTRIBUTION SYSTEM)

MASTER OF SCIENCE IN POWER SYSTEM AND ENERGY ENGINEERING

BY

SEID AHMED MUHYE

HAWASSA UNIVERSITY, HAWASSA, ETHIOPIA

FEBRUARY, 2021
OPTIMIZATION OF POWER DISTRIBUTION SYSTEM WITH DISTRIBUTED
GENERATION USING HARISH HAWKS OPTIMIZATION TECHNIQUES
(CASE STUDY: DEBRE BERHAN POWER DISTRIBUTION SYSTEM)

BY

SEID AHMED MUHYE

A THESIS SUBMITTED TO DEPARTEMENT OF ELECTRICAL AND

COMPUTER ENGINEERING, HAWASSA INSTITUTE OF TECHNOLOGY,
SCHOOL OF GRADUATE STUDIES.

HAWASSA UNIVERSITY
HAWASSA, ETHIOPIA

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE

OF MASTER OF SCIENCE IN POWER SYSTEM AND ENERGY
ENGINEERING

FEBRUARY, 2021
 HAWASSA UNIVERSITY

INSTITUTE OF TECHNOLOGY

DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING

SCHOOL OF GRADUATE STUDIES

ADVISOR’S APPROVAL SHEET

This is to certify that the thesis entitled ”OPTIMIZATION OF POWER DISTRIBU-

TION SYSTEM WITH DISTRIBUTED GENERATION USING HARRIS HAWKS
OPTIMIZATION TECHNIQUES (CASE STUDY: DEBRE BERHAN POWER DIS-
TRIBUTION SYSTEM) ” submitted in partial fulfillment of the requirements for the
degree of Masters of Science in Electrical Engineering with specialization in Power
system and Energy Engineering, The Graduate Program of the Department of Elec-
trical and Computer Engineering, and has been carried out by Seid Ahmed Muhye
ID No- PGPSER/012/10, under my supervision. Therefore, I recommend that the
student has fulfilled the requirements and hence hereby can submit the thesis to the
department.

Dr. Baseem Khan ........................ ........................

Advisor Signature Date

Mr. Essayas Gdey ........................ ........................

Co-advisor Signature Date

 SCHOOL OF GRADUATE STUDIES

HAWASSA UNIVERSITY

EXAMINER’S APPROVAL SHEET

We the under signed Board of examiners of the final open defense by Seid Ahmed have
read and evaluated his thesis entitled ”Optimization of Power Distribution system
with Distribution Generation using Harris Hawks Optimization Techniques (Case
Study: Debre Berhan Power Distribution System) and examined the candidate. This
is therefore, to certify that the thesis has been accepted in partial fulfilment of the
requirements for the degree.

Name Signature Date

————————
———————— ———————
Chair Person
Dr. Baseem Khan
———————— ———————
Advisor
—————————
———————— ———————
External Eximiner
—————————–
———————— ———————
Internal Examiner
——————————–
———————— ———————
SGS

Final approval and acceptance of the thesis is contingent upon the submission of the
final copy of the thesis to the school of graduate studies (SGS) through the depart-
ment Graduate Committee (DGC) of Electrical and Computer Engineering.

Stamp of SGS ...........

Declaration

I, the undersigned, declare that this thesis is my original work, has not been presented
for fulfillment of a degree or otherwise in this or any other university, and all sources and
materials used for the thesis have been acknowledged. All Advisor’s comments are duly
incorporated.

Seid Ahmed .............................. ........................

Name Signature Submission Date

Place: Hawassa, Ethiopia

This thesis has been submitted for examination with my approval as a university
advisor.
Dr. Baseem Khan ........................ ........................

Advisor Signature Date

Acknowledgment

Firstly, it is my pleasure to give special thanks to Allah, the most Gracious, and the most
Merciful who is the wonderful source of my strength. He gave me the grace, wisdom, good
health and guided me in all the way of my life. Secondly, I would like to acknowledge and
extend my deepest gratitude to Dr. Baseem Khan my advisor who offered his support,
and encouragement, his valuable and academically essential guidance, hints, motivation,
ideas, support and constructive comments were so much beneficial. Throughout this thesis
work, I received so much help from him, and completing this thesis work would have been
all the most difficult without his support. In addition, he encourages and advising me to
engage in research for the future of my life. I would like to acknowledge and extend my
gratitude to Mr. Essayas Gdey my co-advisor for his essential guidance and motivation
and members of electrical staff for their polite, continuous and unmeasurable cooperation
by providing me the necessary comments during the thesis progress presentation. I would
like to thank also Debre Berhan district utility officials for their sincere cooperation in
providing me the necessary data and information. Finally, I would like to thank my
families and friends whom I respect and love dearly for their cooperation.

I
Abstract

Growing concerns over environmental impacts, conditions for improvement of the whole
distribution network, shortage and expensiveness of fossil fuel, the deficiency in generating
capacities, and ever-increasing demand for electricity have set the way towards distributed
generation (DG) units in commercial and domestic electrical power systems. The major
problems of distribution systems, such as load growth, power outage, overloaded lines,
quality of supply, and reliability can be solved by optimally placing distributed gener-
ation near to the customer side. However, the non-optimal placement and size of DG
units may lead to high power losses and bad voltage profiles on the power network. This
thesis paper aims to minimize system real and reactive power losses, cost and improve the
voltage profile of the system by determining the optimal size and penetration of wind-
based distributed generation using Harris hawks optimization. The results of base case
load flow analysis showed that the case study distribution network feeder has a base case
active and reactive power loss of 1629.04KW and 609.513KVAr respectively. The feeder
minimum bus voltage and total voltage drop index at the base case is 0.8497V and 0.4407
respectively. The propossed HHO determines the optimal location and size of DG based
on minimum loss reduction index, voltage deviation index, and the cost of DG at the
same time to use the existing distribution network in an optimal manner. The optimal
location of DG is determined to be 33 and 38 with a size of 2.4997MW and 2.4896MW
respectively. The active and reactive power loss reduced to 290.097KW and 254KVAr
after the DG integration. The method implemented and tested on the Sheno feeder of
is
the Debre Berhan power distribution system.

Key words: Distributed Generation, Optimal location, Optimal size, Objective func-
tion, technical constraints, loss reduction

II
Contents

Acknowledgment I

Abstract II

List of Tables VIII

List of Figures IX

List of Abberivations XI

List of Symbols XIV

1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Statement of Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Objectives of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.1 General Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.2 Specific Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Scope of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 Significance of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 Motivation for DG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.7 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 LITERATURE REVIEW 8
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Distributed Generation Technologies . . . . . . . . . . . . . . . . . . . . . 9
2.2.1 Geothermal Power . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

III
 2.2.2 Fuel Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.3 Biomass Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.4 Solar Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.5 Wind Energy System . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.5.1 Wind Speed Data of the Site . . . . . . . . . . . . . . . . 14
2.2.5.2 Power and Energy Estimation by a Wind Turbine Based
on Rayleigh Approach . . . . . . . . . . . . . . . . . . . . 15
2.3 Impact of Distributed Generation . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.1 Impact of DG on Loss . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.2 Impact of DG on Voltage Regulation . . . . . . . . . . . . . . . . . 18
2.3.3 Impact of DG on Voltage stability . . . . . . . . . . . . . . . . . . . 18
2.3.4 Impact of DG on Power Reliability . . . . . . . . . . . . . . . . . . 19
2.3.5 Impact of DG on Harmonics . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Distributed Generation Benefits . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4.1 Economic Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4.2 Technical Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4.3 Environmental Benefits . . . . . . . . . . . . . . . . . . . . . . . . . 22

3 MODELING AND PROBLEM FORMULATION 23

3.1 Mathematical Modeling of Load Flow Analysis using Newton Raphson
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2 System Power Flow Sensitivity Factors . . . . . . . . . . . . . . . . . . . . 24
3.2.1 Analysis of Change in Active Power Flow . . . . . . . . . . . . . . . 25
3.2.2 Analysis of Change in Reactive Power Flow . . . . . . . . . . . . . 26
3.2.3 Formulating the Sensitivity Factors for Power Flow . . . . . . . . . 27
3.2.4 Analysis of Change in Real Power Loss . . . . . . . . . . . . . . . . 28
3.2.5 Analysis of change in Reactive Power Loss . . . . . . . . . . . . . . 29
3.2.6 The Power Loss Sensitivity Factors Formulation . . . . . . . . . . . 30
3.3 Objective Function Formulation . . . . . . . . . . . . . . . . . . . . . . . . 31
3.3.1 Minimizing Real Power Losses . . . . . . . . . . . . . . . . . . . . . 31
3.3.2 Minimizing Reactive Power Losses . . . . . . . . . . . . . . . . . . . 32

IV
 3.3.3 Minimizing Voltage Drop . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.4 Minimizing Annual Economic Loss . . . . . . . . . . . . . . . . . . 32
3.4 Objective Function Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.4.1 Real Power Loss Reduction Index . . . . . . . . . . . . . . . . . . . 34
3.4.2 Reactive Power Loss Reduction Index . . . . . . . . . . . . . . . . . 34
3.4.3 Voltage Deviation Index . . . . . . . . . . . . . . . . . . . . . . . . 34
3.4.4 Annual Cost Reduction Index . . . . . . . . . . . . . . . . . . . . . 35
3.5 Multi-objective Function Formulation . . . . . . . . . . . . . . . . . . . . . 35
3.6 Constraints Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.6.1 Real and Reactive Power balance constraints . . . . . . . . . . . . . 36
3.6.2 Bus Voltage Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.6.3 DG Capacities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.7 Optimization Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.7.1 Genetic Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.7.1.1 Choice of GA Parameters . . . . . . . . . . . . . . . . . . 38
3.7.2 Particle Swarm Optimization . . . . . . . . . . . . . . . . . . . . . 40
3.7.3 Proposed Harris Hawks Optimization (HHO) Algorithm . . . . . . 41
3.7.3.1 Exploration phase . . . . . . . . . . . . . . . . . . . . . . 42
3.7.3.2 Transition from exploration to exploitation . . . . . . . . . 43
3.7.3.3 Exploitation phase . . . . . . . . . . . . . . . . . . . . . . 43
3.7.3.4 Soft besiege . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.7.3.5 Hard besiege . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.7.3.6 Soft besiege with progressive rapid dives . . . . . . . . . . 44
3.7.3.7 Hard besiege with progressive rapid dives . . . . . . . . . 46
3.7.4 Application of HHO in DG allocation . . . . . . . . . . . . . . . . . 46

4 MODELING OF DFIG BASED WIND ENERGY GENERATION SYS-

TEM 48
4.1 Overview of Wind Energy Generation System . . . . . . . . . . . . . . . . 48
4.2 Modeling of Wind Turbine system . . . . . . . . . . . . . . . . . . . . . . 50
4.2.1 Aerodynamic Modeling . . . . . . . . . . . . . . . . . . . . . . . . . 50

V
 4.2.2 Mechanical Shaft and Gearbox Modeling . . . . . . . . . . . . . . . 51
4.2.3 Controller for Wind Turbine . . . . . . . . . . . . . . . . . . . . . . 52
4.3 Doubly-Fed Induction Generator (DFIG) Modeling . . . . . . . . . . . . . 54
4.3.1 Modeling in αβ Stationary Reference Frame . . . . . . . . . . . . . 54
4.3.2 Modeling in dq Synchronously Rotating Reference Frame . . . . . . 56
4.3.3 Control Systems for DFIG . . . . . . . . . . . . . . . . . . . . . . . 58
4.3.3.1 Rotor side control . . . . . . . . . . . . . . . . . . . . . . 59
4.3.3.2 Grid side controller . . . . . . . . . . . . . . . . . . . . . . 63

5 SIMULATION STUDIES AND RESULT ANALYSIS 66

5.1 System under Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2 Cause of Interruption in the Feeder . . . . . . . . . . . . . . . . . . . . . . 68
5.3 Simulation result for different cases . . . . . . . . . . . . . . . . . . . . . . 70
5.3.1 Results during Base case load flow (without DG) . . . . . . . . . . 70
5.3.2 Results during Single DG placement with single objective . . . . . . 71
5.3.3 Results during two DG placement with single objective . . . . . . . 74
5.3.4 Results during Two DG placement with multi objective . . . . . . . 77
5.4 Cost analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.5 Simulation results of Generating optimal DG size from DFIG based Wind
Energy System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

6 Conclusion and Future Work 86

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.3 Beneficiaries of this work . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.5 Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

A Appendix 1 95
A.1 Line and Load data of Sheno feeder . . . . . . . . . . . . . . . . . . . . . . 95
A.2 Wind turbine and DFIG Rating Parameter for the test system . . . . . . . 97

VI
A.3 Proposed Harrish Hawks Optimization Code for optimal size and location
of DG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

VII
List of Tables

2.1 Monthly Average Wind Speed of the Site . . . . . . . . . . . . . . . . . . . 14

2.2 Wind Energy Potential Based on Rayleigh Analysis at 10 m height . . . . 16

5.1 Data for main transformer in Debre Berhan distribution substation . . . . 67

5.2 Distribution transformer and load data for Sheno feeders . . . . . . . . . . 68
5.3 Annually interruption duration and Cost of ENS for Sheno feeder . . . . . 68
5.4 Percentage duration and frequency of each causes of fault . . . . . . . . . . 68
5.5 Percentage duration and frequency of each types of fault . . . . . . . . . . 69
5.6 Bus Voltage profile without installation of DG . . . . . . . . . . . . . . . . 71
5.7 Power loss at different power factor . . . . . . . . . . . . . . . . . . . . . . 73
5.8 Simulation results of the proposed and PSO algorithms for single DG . . . 74
5.9 Simulation results of the proposed and PSO algorithms for Two DG Single
objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.10 Multi objective Weight factors . . . . . . . . . . . . . . . . . . . . . . . . 78
5.11 Simulation results of the proposed and PSO algorithms for Two DG multi
objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.12 The economic evaluation of the optimal distributed generation . . . . . . . 82

A.2 Parameter of Nordex N80/2500 wind turbine . . . . . . . . . . . . . . . . . 97

A.3 DFIG Rating Parameter for the test system . . . . . . . . . . . . . . . . . 97

VIII
List of Figures

2.1 Distributed generation definition . . . . . . . . . . . . . . . . . . . . . . . . 9

2.2 Schematic diagram of geothermal energy [23] . . . . . . . . . . . . . . . . . 10
2.3 Schematic diagram of a fuel cell [24] . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Schematic diagram of a biomass energy [25] . . . . . . . . . . . . . . . . . 12
2.5 Schematic diagram of solar power [26] . . . . . . . . . . . . . . . . . . . . . 13
2.6 Schematic diagram of wind energy system [28] . . . . . . . . . . . . . . . . 14
2.7 Monthly Average Wind Speed of the Site at 10m . . . . . . . . . . . . . . 15

3.1 Single line diagram of a two bus RDS . . . . . . . . . . . . . . . . . . . . . 24

4.1 Mechanical arrangement of wind turbine drive train . . . . . . . . . . . . . 52

4.2 Schematic diagram of abc to αβ stationary reference frame transformation 56
4.3 Schematic diagram of αβ to dq synchronously rotating reference frame
transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4 Synchronous rotating dq reference frame aligned with the stator flux space
vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.5 Closed loop PI current controller diagram of idr and iqr . . . . . . . . . . . 61
4.6 Reduced second order closed loop system (a) for idr and (b) for iqr . . . . . 62
4.7 Reduced second order closed loop system for mechanical speed . . . . . . . 63

5.1 Single line diagram of Debre Berhan substation out going feeders . . . . . 67
5.2 Percentage duration of each causes for power interruption . . . . . . . . . . 69
5.3 Percentage interruption frequency of each faults for power interruption . . 70
5.4 Bus voltage profile without installation of DG . . . . . . . . . . . . . . . . 71
5.5 Bus voltage profile comparison with single DG installation . . . . . . . . . 72

IX
5.6 Bus voltage profile comparison between single DG of Type1 and Type2
with single objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.7 Bus voltage profile comparison of two DG single objective . . . . . . . . . 75
5.8 Bus voltage profile comparison between two DG of Type one and Type two
with single objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.9 Bus voltage profile comparison with Two DG multi objective . . . . . . . . 79
5.10 Bus voltage profile comparison between Two DG of Type one and Type
two with Multi objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.11 Simulated rotor speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.12 Simulated rotor id currents . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.13 Simulated rotor iq currents . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.14 Simulated rotor three phase currents . . . . . . . . . . . . . . . . . . . . . 84
5.15 Simulated rotor vd voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.16 Simulated stator three phase voltage . . . . . . . . . . . . . . . . . . . . . 85
5.17 Simulated stator three phase current . . . . . . . . . . . . . . . . . . . . . 85
5.18 Simulated power delivered to the grid . . . . . . . . . . . . . . . . . . . . 85

X
List of Abberivations

%P LR Percent of active power loss reduction

%QLR Percent of ractive power Loss reduction

AECW DG Annual Economic Loss with DG

AECW ODG Annual Economic Loss without any DG

CDG Cost of DG generated power per kW

LT Total DG life in years

PDG Active Power from Distributed Generation

P LW DG Active Power Loss with Distributed Generation

P LW DG Total active power loss after DG installation

P LW ODG Active Power Loss without Distributed Generation

P LW ODG Total active power loss before DG installation

QDG Reactive Power from Distributed Generation

QLW DG Total reactive power loss after DG installation

QLW ODG Total reactive power loss before DG installation

T &D Transmission and Distribution Systems

T SR Tip speed ratio

V DW DG Total voltage drop with DG

XI
V DW ODG Total voltage drop without DG

AC Alternate Current

ACRI Annual Cost Reduction Index

CHP Combined Heat and Power

DC Direct Current

DFIG Doubly fed Induction Generator

DG Distributed generation

GA Genetic Algorithm

Gbest Best Global Positio

GHG Greenhouse House Gas

GSC Grid side control

HHO Harris Hawks Optimizer

KVAr Kilo volt amper

KW Kilo Watt

LF Levy flight

MOF Multi objective Function

MW Mega Watt

NASA National Aeronautics and Space Administration

O&M Operation and Maintenance

Pbest Best Personal Positio

Pf power Factor

PL Active Power Loss

XII
PLRI Active power loss reduction index

PMSG Permanent Magnet Synchronous Generator

PSO Particle Swarm Optimization

PV Photovoltaic

QL Reactive Power Loss

QLRI Reactive power loss reduction index

RDS Radial Distribution System

RSC Rotor side control

SLD Single Line Diagram

VD Voltage Drop

VDII Voltage deviation index

WECS Wind Energy Conversion System

XIII
List of Symbols

cp Cofficient of power

ct Cofficient of torque

E indicates the escaping energy of the prey

Ed wind energy density at the site

Eo the initial state of energy

Et Energy available at the site

Jg the generator moment of inertia

Jtotal The total moment of inertia

Jturb Wind turbine moment of inertia

N Gear box ratio

Tm Generator Mechanical Torque

Tturb Turbine torque

Tem GeneratorElectromagnetic Torque

Vm Average wind speed at reference height

Vw Available Wind speed at reference height

Zo Surface roughness length of the site

λ Tip speed ratio

ρ the air density

ωm the generator mechanical speed

XIV
Chapter 1

Introduction

1.1 Background

In current scenario, the electricity demand grew rapidly due to industrial and domestic
needs and conventional energy sources are decreasing rapidly. In the conventional power
system, the power transmission carrying capacity has been falling due to increased elec-
tricity demand that result in an over-utilized transmission and distribution system with
an increased level of losses and higher probability of disturbances. Because of increased
demand in the distribution network, the radial distribution lines would be carrying cur-
rent beyond their maximum capacity, in some cases, at the limit, resulting in increased
losses.
There are some conventional methods like Load balancing, bifurcation of overloaded
lines, installation of shunt capacitors to reducing the loss in power distribution system.
In these types of distribution networks, addition of new loads and supplying the load
during interruption becomes very difficult and practically impossible. The thermal loading
conditions of the initial sections of the conductors would violate even if there exist a
marginal increment in the connected load. Currently in our country there is centralized
power system in which there is no any distributed generation near to the load center. With
the concern on shortage and expensiveness of resources like fossil fuel and environmental
concerns, renewable distributed generation (DG) units such as biomass, wind and solar
are emerging as an alternative energy solution. The author in [1] defined distributed
generation as the use of any modular electric generation that is sited near the point of
use (interconnected to the distribution system) to lower the cost of service.

1
In general, a distributed generation is nothing but a small-scale power generations that is
connected near load center instead of expanding central generation station to minimizing
the losses and enhancing voltage stability in power systems [2, 3]. The installation of
distributed generation can have either positive or negative impacts [4].Thus, the generators
connected in distribution systems in such a manner that it reduce system power loss
and operating cost, improve voltage profile, and avoids degradation of power quality
and reliability. Along with these assistances, inadequate allocation of DG in terms of its
location and capacity may lead to voltage deviation, reduce stability, reliability and power
quality, increase line losses, increase system capital and operating costs. The location and
capacity of distributed generation units is a complicated optimization issue and should
be carefully addressed [5].
The selection of optimal location and size of DG is essential to attain potential ben-
efits with less investment. Recently, many different algorithms have been implemented
and proposed to optimally locate and sizing of DG unit. They differ from one another
based on the formulation of problem, methodology, constraints and assumptions being
made. Network losses are the only objective function considered for determining the op-
timal allocation of multiple distributed generations using a novel approach method in [6].
The author in [7] have optimally determined the location and the size of PV distributed
generation systems particularly for improving voltage support in distribution networks
based on particle swarm optimization (PSO). The paper in [8] summerized recent trends
of energy usage from renewable sources and reliable operation of hybrid renewable energy
system. Maximizing power stability index and minimizing of real power losses considered
at the same time to find the global optimal solution for DGs allocation in [9]. A modern
interior point method and increasing and replacement strategies are used to obtain the
optimal configuration location and sizing of dispersed wind generation to reduce the dis-
tribution network loss and improve voltage quality effectively in [10]. The author in [11]
studied the detailed analysis of power system reliability with the optimal employment of
distributed generation using digital simulation and electrical network calculation program
(DIgSILENT). A novel model proposed in [12] for DG planning to analyze the influence
of DG location and capacity on the probability of protection mal−operation in distribu-
tion networks. This paper also considers the minimization of the total cost including DG
investment cost, the cost of energy loss, the cost of protection coordination, and also the

2
mal-operation cost. The author in [13, 14] use GA and (PSO) techniques considering short
circuit level in the distribution system to find the best location of distributed generations.
Moradi and Abedini proposed a hybrid method based on GA–PSO algorithm to find the
optimal location and sizing of DG units considering power loss reduction as an objective
function within the distribution system [15]. The improved Gravitational Search Algo-
rithm is applied to optimally install and size the renewable distributed generation in a
distribution system to minimize the total power losses and average total voltage harmonic
distortion [16]. The author in [17] considers the voltage sag as a power quality index to
find the optimal allocation of DG units using a genetic algorithm. The paper in [18] also
studied the influence of types of DG namely synchronous generation, Wind Turbine and
Photovoltaic system on the penetration level and the placement of DG.
Most of the paper above consider only single-objective problems. In this paper, multi-
objective Harris hawks optimization technique based on integrating distributed generation
is considered and an attempt made to determine the optimal size(s) of DG to minimize
real power losses, operating cost of DG, and improve voltage profile in radial distribution
systems. The result in this paper have been obtained considering the DGs as Type one
(injecting only real power into the system) and as Type two DG providing real and reactive
power injection into the system. The annual cost of loss saving due to minimize the power
loss is also calculated and a comparison has been provided.

1.2 Statement of Problem

The electric power consumption in Ethiopia is increasing from time to time due to popula-
tion growth, industrial growth, social and economic developments whereas supply growth
is constrained by scarce resources, which results in increased current drawn from the
source, voltage drop & power loss in the distribution systems. Because of these, power
plants are not sufficient to satisfy the people’s fundamental services of the country and
both utility and connected customers suffer from these loss and voltage drop problems.
The people in the Debre Berhan city, which consists of many industry loads, residential
and commercial customers, are victims of such problem. Especially for factories and in-
dustries, it is really challenging to tolerate such problems since it causes much revenue
losses.

3
Therefore, there must be new system facilities, and improvements in production and use
of electricity to tackle such problems in the power distribution system.
So far different classical methods have been applied to satisfy the load demand, minimize
power loss and improve the voltage profile by capacitor placement, conductor upgrading
and feeder reconfiguration, tapping cables to the transformers irregularly, and hence over-
loading and heating is resulted which in turn causes unnecessary power losses. However,
it is not a promising approach to solve these problems.
Distributed generation and its integration with the power distribution network is one
way of addressing such problems in power systems. However, non-optimal location and
sizing of DG units may lead to losses increase together with bad effect on voltage profile of
the distribution systems. Therefore, this paper determined the amount of DG penetration
and optimum DG placement to minimize the negative impacts and enhance the positive
influences of DGs using Harris hawks optimization.

1.3 Objectives of the thesis

1.3.1 General Objective

• To determine the optimal placement and sizing of DG for optimal operation of the
system using Harris hawks optimizatio for optimizing operation of the distribution
system by reducing power loss, cost of DG, and improving voltage profile.

1.3.2 Specific Objectives

• To assess the wind resource availability in the study area

• To identify main cause of power interruption in the feeder

• To determine the maximum possible DG integration with respect to the given con-
straints

• To compare the results of the system with DG implementations with the present
grid

• To draw appropriate conclusions and recommendations from the result of the inves-
tigation

4
1.4 Scope of the thesis

The paper mainly limited on voltage profile improvement and power loss reduction of
distribution feeders by optimally locating and sizing DG using optimization techniques.
This thesis work mainly includes:

• Steady state analysis of a power distribution system using load flow analysis

• Comparision of system operating parameters with and without distributed genera-

tion

• Develop a program for optimal placement and size of the DG in matlab based on
Harris hawks optimization techniques for power loss minimization and voltage profile
improvements

• Test and compare the program on Debre Berhan radial distribution feeder to confirm
its eligibility

1.5 Significance of the thesis

Some of the significance of this thesis paper includes:

• Presence of DGs in the distribution network can extend equipment maintenance

intervals, reduce electrical line losses, and improve distribution system reliability
and voltage profile, all with cost savings to utilities

• DGs also contribute to the application of competitive energy policies, diversification

of energy resources, and reduction of on-peak operating cost, lower losses, lower
transmission, and distribution costs

• Useful tool for the utilities for practical implementation of DG

1.6 Motivation for DG

There are mainly three units of electric power system namely production, transmission,
and distribution unit and devices that make use of the electricity. The main requirement
of the electric power system is to enable power transfer between the generators and the

5
distributing system maintaining an acceptable reliability and voltage quality for all cus-
tomers (producers and consumers). However, the far distance between the production
unit and end-users increase the transmission and distribution loss, interruption duration
that reduces optimal operation of the systems. Power engineers find DG motivation and
energy producers as both the utility and customers benefits in DG applications in terms of
cost and quality supply. The utility invest much cost for the expansion and maintenance
of transmission and distribution (T &D) systems, which is a tough and expensive task
for utilities. Improvements are regularly necessary to keep up with energy consumption
growth.
By employing DG at the distribution level where there is significant growth, cost
related to improvements of the equipment size can be avoided while the energy losses
related to T &D can be reduced. Studies can reveal that T &D costs attributed 30% of
those costs that customers pay for electricity [19]. The on-site integration of DG reduce
such extra cost on customer due to T &D costs.
Regarding environmental matters, several of the conventional types of production
result in the emission of carbon dioxide with the much-discussed global warming as a
very likely consequence. Changing from conventional production based on fossil fuels,
such as coal, gas, and oil, to renewable sources, such as sun and wind, would reduce the
emission.

1.7 Thesis Outline

This thesis work is structured into six chapters:

Chapter 1 presents the background, problem statements, objectives, and motivation for
DG integration. The need to shift the power system to a decentralized power system is
explained well in this chapter. It also explains that the location and size of DG have an
impact on distribution system operation.
Chapter 2 explain the concept and types of distributed generation (DG), impact of inte-
grating distributed generation into the distribution system. The wind resource assessment
also explained in this chapter. are

In chapter 3, different factors and indexes namely the power loss and voltage deviation
indexes are formulated to help in solving the optimal location and sizing problem.

6
The weight factors are used in the formulation of the multi-objective function and the
operational constraints are done in this chapter.
In chapter 4, the modeling of wind turbine and DFIG with their respective controller is
presented. The MPPT controller for the wind turbine and the vector controlled mecha-
nism for DFIG are implemented.
In chapter 5, the results obtained are presented and analyzed in detail in terms of tables
and graphs.
Chapter 6 presents conclusions, recommendations, and future works.

7
Chapter 2

LITERATURE REVIEW

2.1 Introduction

This chapter presents the definition of distributed generation, types of distributed genera-
tion technology, impact of distributed generation, and benefits to the distribution network.
There are different definitions of DG in different literature. Distributed generation named
as decentralized generation, site generation that generates electrical energy through alter-
nate energy-producing resources close to the customer side to reduce the expansion of the
present expensive electric transmission system [20]. In general, the standard definition of
distributed generation (DG) is installing small-scale generation units (typically 1KW –
50MW) to produce electricity at different locations tied directly to the distribution sub-
station or near to the load for power flow reduction, loss minimization, voltage profile
enhancement and voltage stability improvement [21]. Some of the distributed generators
include combustion gas turbines, fuel cells, solar photovoltaic, geothermal, and wind tur-
bines. Usually, distributed generation integration to the distribution system is via power
electronic converter or other power electronic devices [22]. When distributed generation
integrated into a distribution system, it can have a positive or negative impact on power
system operating parameters. Different researchers apply different techniques to get its
optimal location and size so that to reduce its negative impact on the distribution sys-
tem. DG includes both renewable and non-renewable sources which mainly support the
increased energy-demand, peak shaving, and to use as a backup generator.

8
 Figure 2.1: Distributed generation definition

2.2 Distributed Generation Technologies

There are two major classifications of distributed generation in the world today applying
in power systems. Based on the type of fuel they use for their operation, DGs classified
into two groups namely; renewable and non-renewable distributed generation resources.
Both of them help the power distribution system to increase the efficiency and reduce the
cost of operations, installations, and maintenances.
Non-renewable DG resource: They incorporate technology such as natural gas,
coal, gas turbine, microturbine, and petroleum to produce energy for different operations.
Mostly used as distributed generations in the areas where the available renewable en-
ergy resource is not sufficient for reliable operation. However, the burning of fossil fuels
has contributed the largest percentage to global warming, greenhouse gases, acid rain,
petrochemical pollution, and eventually run out of use because of the high rate of energy
demand and not sustainable and not replaced by a natural means.
Renewable DG technologies: Renewable DG technologies are the electric power
generation resources that directly connected to the consumer’s load on the distribution
systems at the medium voltage or low voltage levels. The renewable DG technologies
include geothermal power, biomass, solar power, hydro power, and wind turbines. They
provide a technical and economical improvement in power system networks and at the
same time to reduce the need to support the increasing demand and reduce CO2 emission.

9
Commonly used DG technologies are explained below and wind energy system are selected
in this paper to implement as distributed generation source in the distribution systems
considering the availability around Debre Berhan power distribution systems.

2.2.1 Geothermal Power

Geothermal energy extract heat from the sub-surface of the earth, which originates from
radioactive decay of materials or inject water in the case of dry rock technology. The
heat generated requires conversion by mechanical means or used directly. The geother-
mal heat is used faster than it can be recovered by the heat in the center of the earth that
makes the production of some geothermal plant in order to dropped siginificantly over
time.Production of power from geothermal energy requires wells to access the system to
derive turbine connected to electrical generator. The extracted steam pumped through
steam pipelines and the kinetic energy of the steam is transformed into rotational me-
chanical power. The alternator connected to the turbine axis rotates and transforms
mechanical rotational energy into electrical energy, which transferred to the distribution
network. Not all the steam pumped to the turbine and the steam exits the turbine re-
turned to the liquid state in a condenser. A cooling tower used to cool the water produced
from the steam condensation process. The gases that cannot be condensed are dispersed
in the atmosphere and the condensed water is then re-injecting it into the deep rock layers
[23].

Heat
Production well Turbine Alternator Grid
exchanger
for steam water
extraction

Heat from subsoil Condensor

Cooling tower

Figure 2.2: Schematic diagram of geothermal energy [23]

10
2.2.2 Fuel Cell

A fuel cell is an electrochemical cell that generate a fuel through the chemical reaction
of hydrogen ions with oxygen without combustion. The electrochemical devices convert
the chemical energy of a fuel directly to electricity and heat. The waste heat from fuel
cells used for heating applications in low-pressure steam tied into hot water distribution
systems. In addition to electrical power generation technology, fuel cells used in CHP
applications.
There are several types of fuel cells, but commonly all of them consists of a cathode
and an anode separated by an electrolyte that allows ions to move between the two sides of
the fuel cells. The fuel (for example, hydrogen or natural gas) supplied to the anode. The
second electrode (the cathode) supplied with oxygen by simply pumping air in. Oxidation
reactions that generate ions (often positively charged hydrogen ions) and electrons take
place at the anode and these ions move through the electrolyte from the anode to the
cathode. At the same time, electrons flow through an external circuit from the anode to
the cathode, producing direct current electricity. The ions, electrons, and oxygen react
with each other at the cathode due to the catalyst in it, forming water and possibly
other products [24]. Now the most commercially available type of fuel cell is the proton
exchange membrane due to its highest energy density per volume rate.

Electric Current
Fuel in e-

e-
H+
e- O2 and
O2 Other gases in
e-
Excess Un used
H2O Gases out
Fuel H2 H+

Cathode
Anode Electrolyte

Figure 2.3: Schematic diagram of a fuel cell [24]

11
2.2.3 Biomass Energy

Biomass is a renewable source of fuel developed from organic materials and a sustainable
source of energy like scrap limber, forest debris, certain crops, manure, forest waste,
industrial waste, etc. The waste burned to produce steam that runs the turbine to make
electricity or that provides heat to industry and home. Biomass converted to electricity
directly by combustion to produce heat, or indirectly after converting it to discrete forms
of biofuel. In some biomass industries, it used as combined heat and power (CHP) systems
in which the extracted or spent steam from the power plant is also used for manufacturing
processes or to heat buildings, which greatly increase the overall energy efficiency of the
system [25].
Conversion of biomass to biofuel achieved by electrochemical conversion methods,
thermal, chemical, and biochemical conversion. On the other hand, it is possible to
convert the solid biomass into a fuel gas. Biomass has a valuable contribution to the
overall energy supply system mainly because of its very low cost and the fact that it is
renewable. There is still considerable scope for making better use of the existing biomass
energy supplies and well positioned to play a major role to reduce greenhouse emissions.
Biomass has the advantage of controllability and availability compared to other renewable
energy when needed. The fuel needs to be delivered, stored, procured, and paid for which
is the disadvantage of biomass energy.

Exhaust
Furnace boiler

Turbine

Preparation Generator Grid

Storage and
Processing

Ash
Biomass
materials
Condenser

Figure 2.4: Schematic diagram of a biomass energy [25]

12
2.2.4 Solar Power

The conversion of energy from sunlight into electrical energy with the help of directly using
photovoltaic (PV) cell or indirectly using concentrated solar energy called solar power.
A solar PV cell is a large-area semiconductor diode consisting of a p-n junction created
by doping into the semiconductor crystal. The crystal consists of covalent bonds to the
neighboring atoms for the most commonly used silicon solar cells. The location of the sun
in the sky and the amount of cloud cover, determine the amount of power produced by a
solar power installation. The location of the sun in the sky shows a predictable daily and
seasonal variation caused by the rotation of the earth on its axis and around the sun.

Sunlight

Electron flow

Load

P-Type
N-Type Junction

Figure 2.5: Schematic diagram of solar power [26]

The cell generates less amount of DC power, so the PV cells connected in series and
parallel to form a module, panel, and array to increase the output of current and voltage
in DC. Solar panels is the most widely used, often installed on the roof of a building that
only transfers part of the solar energy into electrical energy [27].

2.2.5 Wind Energy System

The wind energy is also a renewable energy that converts kinetic energy from the hori-
zontal displacement of air into kinetic energy of the rotation of a turbine by means of a
number of blades to turn the electric generator and traditionally to do other works, like
pumping and mining.

13
The electrical generator then transforms this mechanical rotational energy into electrical
energy. The conversion of wind energy to an electrical energy system consists of wind
turbines having an aerodynamic rotor, a mechanical transmission system, an electric gen-
erator, and a rotor control system. The wind system offers a good power solution because
of being cost-effective, low operation and maintenance (O&M) costs, readily available,
plentiful, renewable, free, clean, emit no gas such as Co2 that makes it growing at a rapid
rate, and it is one of the dramatic growing energy source.

Wind
Gear system
and Generator Power grid
coupling

Controller

Figure 2.6: Schematic diagram of wind energy system [28]

2.2.5.1 Wind Speed Data of the Site

The available wind speed measured at 10 meters height for Debre Berhan from NASA
indicates that wind with minimum speed occurs during August (3.7m/s), while the wind
with high velocity occurs during December, November, February, and January; 5.31, 5.28,
5.17, and 5.15m/s respectively. The wind resource data collected for five years at 10m
height and the monthly average wind Speed of the Site is given in table 2.1 bellow.

Table 2.1: Monthly Average Wind Speed of the Site

Month Jan Feb Mar April May Jun July Aug Sept Oct Nov Dec Average
Speed
5.15 5.17 5.03 5.09 4.90 4.6 4.07 3.70 4.04 5.24 5.28 5.31 4.81
(m/s)

The table above indicates that the mean wind speed of the month at 10 m height is
4.81m/s.

14
 Monthly average wind speed for 2015-2019
6

Wind Speed (m/s)

4

0
Jan Feb Mar April May Jun July Aug sept Oct Nov Dec
Month

Figure 2.7: Monthly Average Wind Speed of the Site at 10m

The annual average wind speed at 10m for the Debre Berhan site is already known
before which is 4.81m/s and the average wind speed at the desired height( h) with the
wind shear exponent of Zo is calculated as follows:
" #
ln( zho )
Vw = Vm (2.1)
ln( hzor )

Where;
Vw = wind velocity at desired height (h)
Vm = average wind speed at reference height (hr)
Zo = surface roughness length of the site
The annual average wind speed at 10m for the Debre Berhan site is already known before
which is 4.81m/s and using equation (2.1) the average wind speed at the desired height
of 80m with the wind shear exponent of 0.12 is calculated to be 7.07m/s.‘

2.2.5.2 Power and Energy Estimation by a Wind Turbine Based on Rayleigh

Approach

Each wind turbine has a wind speed power curve that describes its own characteristic.
The shape of the wind speed power curve is influenced by the wind speed, the blades area,
the choice of airfoil, the number of blades, the blade shape, the optimum tip speed ratio,

15
the speed of rotation, the cut-in wind speed, the shutdown speed, the rated speed, and
gearing and generator efficiencies.
Energy produced at any wind speed can be obtained by multiplying the number of
hours of its duration by the corresponding turbine power at this wind speed obtained by
the turbine’s power curve. The total energy produced is calculated by summing the energy
produced at all the wind speeds within the operating range of the turbine. Considering
Rayleigh distribution, wind energy density at the site is given by the expression:

 
3
Ed = ρVm3 (2.2)
π

Energy available at the site for the unit area of the rotor, over a time can be estimated
using the expression:
 
3
Et = T ∗ Ed = T ∗ ρVm3 (2.3)
π

Where;
Ed is wind energy density at the site
Et is Energy available at the site
T is no of hour/month, ρ is the air density

Table 2.2: Wind Energy Potential Based on Rayleigh Analysis at 10 m height

Month Vm(m/s) Ed (KW/m2) Et (KWh/m2)

Jan 5.15 0.157 117
Feb 5.17 0.159 107.7
Mar 5.03 0.146 108.88
April 5.09 0.151 109.2
May 4.9 0.135 100.66
Jun 4.6 0.111 80.594
July 4.07 0.078 57.683
Aug 3.7 0.058 43.338
Sep 4.04 0.075 54.597
Oct 5.24 0.165 123.101
Nov 5.28 0.169 121.879
Dec 5.31 0.172 128.101

From the table 2.2 above, it can be concluded that there is potential wind energy from
the site to supply parts of the distribution feeders.

16
 2.3 Impact of Distributed Generation

With an increased number of distributed generation integrated to distributed networks,

it leads to some challenges to the network planning. Traditionally, the distribution sys-
are
tems designed to deliver the electrical power in one direction only from the source to the
load direction. When DG installed, it changes the normal characteristics of the distri-
bution system and maximizes green energy penetration in the utility network, increases
the concern of voltage and frequency stability [32]. The effect of DG on enhancing power
system quality by improving the voltage profile and power loss reduction also investigated.
The integration of Distributed generation in radial distribution systems introduces a new
source in the system and it causes to change the direction of power flow, which consider-
ably affect the power flow and voltage conditions at both, customers and utility side. The
major impacts of integrating distributed generation to power distribution system iden-
tified as voltage variations and unbalance, current and voltage harmonics, short circuit
level, and other power quality issues, such as flicker and stress on distribution transformer
[33]. The distributed generation can affect the system both positively and negatively de-
pends on the operating characteristics of the distribution system and the DG as well. It
can also affect the system technically or economically.

2.3.1 Impact of DG on Loss

Normally the real power loss reduction draws more attention for the utility, as it reduces
the efficiency of transmitting energy to consumers. The placement of DG along the
network feeders can help both the customers and the utility to reduce system loss as it
provide both real and reactive power to the system. The integration of DG may increase
or decrease the loss in the distribution systems depending on the capacity and location of
DG on each part of the distribution network [34]. At the lower penetration of DG, the loss
is reduced, however as the penetration increases, the power loss begins to increases. This
means there is a maximum penetration level beyond which the DG will increase loss in the
network. Locating and sizing the DG units should be as optimum as possible to achieve a
better reliability of the system with reduced losses and to reach an optimal performance
of the power network. The studies presented the positive effects of the small capacity of
DGs on loss reduction when connected to feeders having high losses in the system, and

17
it is a great benefit for the utilities. Different penetration levels used to determine the
impact of increased DG penetration on grid losses. Grid losses decreased at lower DG
penetrations and showed an increase in grid losses with increased penetration.

2.3.2 Impact of DG on Voltage Regulation

Voltage regulation is one of the main problems in radial distribution networks. The most
common device and techniques used for voltage regulation in the distribution systems are
load tap changer transformers, voltage regulators, and shunt capacitors. The penetra-
tion of DG in the system changes the power flow of real and reactive power, as well as
voltage magnitude, is also changed. These make the Distribution network active and non-
unidirectional. Nevertheless, the presence of DG can provide either a positive or negative
impact on voltage regulation depending on the characteristic, location, and size of DG
[35]. It causes to occur voltage fluctuation and more loss, which are the negative effects of
DG but the voltage profile improved well if DG is placed at the optimal location. When
the DG unit is beyond 10MW, the impact on the primary feeder is high and then voltage
regulation analysis is required to make sure the feeder voltage within suitable limits [36].

2.3.3 Impact of DG on Voltage stability

Always power system required to operate near to its stability limit due to the rapid
growth of power demand. However, the entire system operated on the border of the
voltage instability region while the voltage profiles of the system are still within the
normal operating margins. DG benefits the customers by providing electricity within
a short distance between the source and the customer that reduce the measurement of
the voltage stability margin (VSM). The study in [37] uses different optimal capacities
distributed generation to check the effect on the voltage stability of the system in model
of medium voltage meshed distribution system. The result showed that penetration of
DG increases the active power injected into the bus and will improve the overall voltage
profile and system voltage stability has increased. Distributed generation is a way of
improving voltage stability and power loss and maintain steady and acceptable voltage
at all buses in the system after subjected to a disturbance from a given initial operating
condition. The integration of DG at a feeder provides active power locally and has the
potential to provide added reactive power support.

18
2.3.4 Impact of DG on Power Reliability

Electric power system should continuously supply power to satisfy the customers. This
named as reliability. The reliability of the system affected by the placement and control
of DG and protection devices with respect to DG penetration [38]. If the penetration
is not optimal and the control is not good, the system reliability levels may decline to
meet the user demand for power supply, otherwise you can improve the reliability of the
distribution network. After DG and power grid, when DG interrupts the power supply
unit, the system can compensate for the load in the shortest time difference. The optimal
penetration of distributed generation in the distribution system has a positive impact
on distribution system reliability. This is due to interruption frequency and interruption
duration reduce as the distance between of load point and DG reduces. DG used as
a back-up generator when the main power supply in case interrupted. In addition, the
distribution network’s restoration capability can be improved and give additional capacity
to feeders or substations. DG can eliminate network constraints (voltage drop or feeder
loading) during the restoration process. The author in [39] showed that reliability of
distribution system increases with integrating single DG unit into the distribution system
and injecting multiple DG at different locations and near to load points in distribution
network further increases reliability of distribution system, while introducing multiple DG
at single location improves reliability of distribution system.

2.3.5 Impact of DG on Harmonics

Harmonics are always present in power systems to some extent caused by, for instance,
non-linearity in transformer exciting impedance or loads such as fluorescent lights, AC
to DC conversion equipment, variable-speed drives, switch-mode power equipment, arc
furnaces, and other equipment. The integration of DG produce harmonics in the network
either from the generation unit itself (generator) or from the power electronics equipment
such as inverters used to transfer the DC power generated from renewable energy to AC
to supply the grid. The paper in [40] considers different types of distributed generation
(DG) units like a mini-hydro synchronous generator, photovoltaic system, and double fed
induction generator wind turbine to analyze harmonic impact in low voltage distribution
systems.

19
Therefore, care should be given to the impact of DG on harmonics and when there exists
a worst case in the operation, the equipment at the DG may need to be disconnected to
remove the extra heating caused by the harmonics.

2.4 Distributed Generation Benefits

In the centralized way of transmission and distribution system, a large number of residen-
tial, commercial, and industrial loads supplied from long distances by large size of sources,
which is costly and wastes the electricity because of aging transmission equipment, sys-
tem outage, and growing congestion. Unlike large central power plants, DG required less
investment cost for connecting the source to the load as it installed near the load center
by decreasing the distance it travels and the amount that lost with ratings range from
5KW up to 100MW and [41].
Some of the majior benefit drawn through the interconnection of DG units into electric
power networks are economic, environmental and technical benefit.

2.4.1 Economic Benefits

The act of integrating DG units closer to the load center leads to avoidance of the need
for building new and costly transmission and distribution lines, upgrading the existing
power supply system, and reducing transmission and distribution capacity during network
planning. Distributed generation sources provide good flexibility and adaptation to the
changes in the economic environment due to their short construction lead-time and small
size. Moreover, distributed generation technologies can help suppliers to fill in the slots
in the liberalized market and provide electricity services as required by the customers.
Thus, DG can be a least-cost planning alternative and cause a considerable reduction in
required investment cost for supplying increased load in future years [42].
Some economic benefits of DG are as follows:

• DGs can be installed in a very short period at any location as modules and its
capacity can be increased or decreased by adding or removing more modules, re-
spectively.

• Lowering cost by avoiding long-distance high voltage transmission and promote

competitive markets.

20
 • Distributed generation supply power during peak demand periods (Peak Shaving)
which contributes to avoiding electricity price fluctuations and provides a flexible
reaction in evaluating electricity prices.

• Provide ancillary services, cost saving by reducing peak demand for central power
station and remote power applications.

• DGs satisfy an increase in load growth, by sizing it in small increments.

• Integration of DG reduces power demand from central plant, which reduce the
wholesale power price by supplying power to the grid.

• In addition, the planning of DG is a short-term investment approach due to low

capital cost and reducing time for installation and hence minimizing investment
risk.

• It improve economic benefit by minimizing the number of voltage regulator and

capacitor reduce the feeder, power loss and minimizing the cost of loss and by
decreasing the loading on electric equipment, minimizing their maintenance cost.

• Running costs more or less constant over the period with the use of renewable
sources.

2.4.2 Technical Benefits

Both the utility company and the consumer are benefit from the integration of distributed
generation into the utility’s conventional resources. DG has an important role in power
distribution systems to improve service reliability by reducing the number of system out-
ages and by avoiding line extensions to remote areas. Such systems can also relieve
thermal overloads in selected utility distribution systems. Other important benefits also
exist, such as loss reduction on both distribution and transmission lines and voltage sup-
port. DGs can provide different technical benefits based on different factors, for example,
location and technology. The following are some of the technical benefits [43].

• Improving availability and reliability of the power supply network by having a back-
up generation.

• Voltage regulation support and Power-loss reduction

21
 • Give benefit to utility system such as system security by diversifying the energy
sources and solving power quality problems, such as voltage sags, as the installation
of a DG increase the voltage level in the network.

• Reducing power flow inside the transmission network to fit certain constraints and
improve its voltage profile

• DGs can help in peak load shaving avoiding problems that might be associated with
high demands such as load shedding and transmission line congestion.

• Other benefits of DG include providing ancillary service, and adding self-generation

to customer options

2.4.3 Environmental Benefits

The generation of Electric power from large and centralized power plants emit a significant
amount of carbon dioxide, sulfur oxide hydrocarbons, and nitrogen oxide, which is a major
contributor to climate change. The need to reduce carbon emission into the environment
is also to reduce noise pollution and protect other creatures made possible by DG power
source. Recent DG technologies offer an environmentally friendly source of electrical
energy by limiting the Greenhouse house Gas (GHG) emissions. Distributed generation
can also play a role, as it allows optimizing the energy consumption of firms that have a
large demand for both heat and electricity [44]. Another benefit of DG is that it reduces
the tendency of construction of new transmission circuits and large networks that takes
land resource available in the customer’s hand.

22
Chapter 3

MODELING AND PROBLEM

FORMULATION

These thesis works have mainly two modeling and problem formulation. The first one
is mathematical modeling of the load flow analysis of the power system to calculate the
optimal size and location of distributed generation. The second one is modeling and
formulation of the energy conversion system from the wind (WECS) to generate the
determined optimal size of DG from the wind available in the area, which is explained in
the next chapter.

3.1 Mathematical Modeling of Load Flow Analysis

using Newton Raphson Method

This chapter deals with the problem formulation including the general form of optimiza-
tion procedure and mathematical formulation of sensitivity factor for power loss, real and
reactive power flow using newton raphson method. The multi-objective function is for-
mulated taking into consideration some key factors like real power loss reduction, reactive
power loss reduction, and voltage profile improvement index. Both equality and inequality
constraints to which the multi-objective function is subjected to are also defined here. For
the analysis, considering the single line diagram(SLD) of two buses connected between
buses p and q as shown in figure 3.1.

23
 Figure 3.1: Single line diagram of a two bus RDS

Where;
PP and Qp is real and reactive power flow at pth bus respectively
Pq and Qq is real and reactive power flow at q th bus respectively
Vp and Vq is voltage at pth and q th bus respectively
Rp and Xq is resistance and reactance of the line connecting pth bus to q th bus
δp and δq is power angle at pth bus and q th bus respectively
Ip is current flow in the branch connecting pth bus to q th bus

3.2 System Power Flow Sensitivity Factors

Power flow sensitivity of the power system is the change in power flow in a transmission
or distribution line connected between two buses say bus p and bus q due to unit change
in the power injected at any buses in the system. Considering the single line diagram
above, the complex power injected by a source into a bus of a power system is given by:

Sp = Pp + jQp = Vp Ip∗ p = 1, 2, 3, . . . , n (3.1)

Where;
Vp is the voltage at the bus with respect to ground
Ip is the source current injected into the bus
In order to handle load flow analysis more conveniently, the use of Ip rather than Ip∗ is
encouraged. As a result the complex conjugate of the above equation is considered, that
is:

Sp∗ = PP − jQP = Vp∗ Ip (3.2)

24
The source current is given by:

X
Ip = Ypq Ipq p = q = 1, 2, 3, ...n (3.3)

Thus substituting equation (3.3) in to equation (3.2), we have:

X
Pp − jQp = Vp∗ Ypq Vq (3.4)

The active and reactive power of bus p calculated by equating real and imaginary parts
of equation (3.4) above:
X
PP = Re(Vp∗ Ypq Vq ) (3.5)
X
QP = Re(Vp∗ Ypq Vq ) (3.6)

The voltage and bus admittance of the power system can be expressed in polar form as:

VP = |VP |ejδp (3.7)

Ypq = |Ypq |ejθpq (3.8)

where;
θpq is the angle of the pq th element of the Ybus matrix
Substituting the polar representations given in the equation (3.7) and (3.8) above into
equation (3.5) and (3.6), the real and reactive powers expressed in general as shown:

X
PP = |VP | |Vq ||Ypq |cos(θpq + δpq ) p = q = 1, 2, . . . , n (3.9)
X
QP = −|VP | |Vq ||Ypq |sin(θpq + δpq ) p = q = 1, 2, . . . ., n (3.10)

3.2.1 Analysis of Change in Active Power Flow

The active power flow in a line k connecting two buses, bus p and bus q can be expressed
as:
Ppq = VP Vq Ypq cos(θpq + δpq ) − Vp2 Ypq cosθpq (3.11)

25
Where;
VP and Vq are the voltage magnitudes at buses p and q respectively
δp and δq are the voltage angles at buses p and q respectively
Ypq is magnitude of the pq th element of the Ybus matrix
θpq is the angle of the pq th element of the Ybus matrix
Mathematically, the active power flow sensitivity can be expressed as:
 
∆Ppq
∆Pn
(3.12)
 
 
∆Ppq
∆Qn

Ignoring second and higher order terms the change in real line flow using Taylor series
approximation can be expressed as:

∂Ppq ∂Ppq ∂Ppq ∂Ppq

∆Ppq = ∆δp + ∆δq + ∆Vp + ∆Vq (3.13)
∂δp ∂δq ∂Vp ∂Vq

The partial derivatives of real power flow with respect to variables δ and V gives the
coefficients appearing in the equation (3.13) above as shown below:

∂Ppq
= VP Vq Ypq sin(θpq + δpq ) (3.14)
∂δp
∂Ppq
= −VP Vq Ypq sin(θpq + δpq ) (3.15)
∂δq
∂Ppq
= Vq Ypq cos(θpq + δpq ) − 2Vp Ypq cos θpq (3.16)
∂Vp
∂Ppq
= Vp Ypq cos θpq (3.17)
∂Vq

3.2.2 Analysis of Change in Reactive Power Flow

The reactive power flow in a line k connecting two buses, bus p and bus q can be expressed
as:
Qpq = −VP Vq Ypq sin(θpq + δpq ) + Vp2 Ypq sinθpq (3.18)

Mathematically, the reactive power flow sensitivity can be written as:

 
∆Qpq
∆Pn
(3.19)
 
 
∆Qpq
∆Qn

26
Ignoring second and higher order terms the change in reactive line flow using Taylor series
approximation can be expressed as:

∂Qpq ∂Qpq ∂Qpq ∂Qpq

∆Qpq = ∆δp + ∆δq + ∆Vp + ∆Vq (3.20)
∂δp ∂δq ∂Vp ∂Vq

The coefficients appearing in the equation (3.20) can be obtained using the partial deriva-
tives of real power flow with respect to variables δ and V as shown below:

∂Qpq
= VP Vq Ypq cos(θpq + δpq ) (3.21)
∂δp
∂Qpq
= −VP Vq Ypq cos(θpq + δpq ) (3.22)
∂δq
∂Qpq
= −Vq Ypq sin(θpq + δpq ) + Vp Ypq cos θpq (3.23)
∂Vp
∂Qpq
= −Vp Ypq sin(θpq + δpq ) (3.24)
∂V q

3.2.3 Formulating the Sensitivity Factors for Power Flow

The change in the real power flow over a transmission or distribution line connected
between bus-p and bus-q due to the change in active power injected at any other bus
represents sensitivity factors for the real power flow. Similarly, the change in the reactive
power flow over a transmission or distribution line connected between bus p and bus q
due to the change in reactive power injected at any other bus represents sensitivity factors
for the reactive power flow. The equations for the changes in the line flows arranged in
matrix form and expressed as:
    
∂Ppq ∂Ppq
 ∆Ppq   ∂δ ∂V   ∆δ 
 =   (3.25)
∂Qpq ∂Qpq
∆Qpq ∂δ ∂V
∆V

The variables ∆δ and ∆V can be obtained from the full Newton Raphson load flow
Jacobian matrix as follows:
      
 ∆P   ∆δ   J1 J2   ∆δ 
  = [J]  =   (3.26)
∆Q ∆V J3 J4 ∆V

27
Then the variables ∆δ and ∆V can be obtained from equation (3.26)above as:

   −1  
 ∆δ   J1 J2   ∆P 
 =    (3.27)
∆V J3 J4 ∆Q

Now substituting equation (3.27) for ∆δ and ∆V into equation (3.25) for the change in
line flows we have:
    −1  
∂Ppq ∂Ppq
 ∆Ppq   ∂δ ∂V   J1 J2   ∆P 
 =     (3.28)
∂Qpq ∂Qpq
∆Qpq ∂δ ∂V
J3 J4 ∆Q

The sensitivity factors for the real power flows are represented as:
     
∂Ppq ∂Ppq
∂Pn   LP P   T −1  ∂δ
= = J (3.29)
 
  
∂Ppq ∂Ppq
∂Qn
LP Q ∂V

The sensitivity factors for the reactive power flows are represented as;
     
∂Qpq ∂Qpq
∂Pn   LQP   T −1  ∂δ
= = J (3.30)
 
  
∂Qpq ∂Qpq
∂Qn
LQQ ∂V

Where;
LP P is the real power flow sensitivity related to the real power injection
LP Q is the active flow sensitivity related to the reactive power injection
LQP is the reactive power flow sensitivity related to the active power injection
LQQ is the reactive power flow sensitivity related to the reactive power injection
J is the Jacobian matrix of power flow, and the superscript T indicates the transpose
Here the four sensitivities are column vectors with dimension of the number of the system
busses.

3.2.4 Analysis of Change in Real Power Loss

The real power loss in a line connecting two buses, bus p and bus q can be expressed as:

PL(pq) = αpq Vp 2 + Vq 2 − 2Vp Vq cos(δpq )

(3.31)

28
Thus, the total active power loss in the circuit can be expressed as:

Xnb
[αpq Vp 2 + Vq 2 − 2Vp Vq cos(δpq ) ]


PL(total) = (3.32)
i=1

Where;
nb is the number of lines of the network
αpq is the conductance of the line p − q
Vp is the nodal voltage of bus p
Vq is the nodal voltage of bus q
δpq is the phase angle difference between the busses p and q
Mathematically, loss sensitivity for the real power can be written as:
 
∆PL(pq)
∆Pn
(3.33)
 
 ∆PL(pq)

∆Qn

Ignoring second and higher order terms the change in real power loss using Taylor series
approximation can be expressed as:

∂PL(pq) ∂PL(pq) ∂PL(pq) ∂PL(pq)

∆PL(pq) = ∆δp + ∆δq + ∆Vp + ∆Vq (3.34)
∂δp ∂δq ∂Vp ∂Vq

The partial derivatives of real power loss with respect to variables δ and V gives the
coefficients appearing in the equation (3.34) above as shown below:

∆PL(pq)
= 2αpq Vp Vq sin (δpq ) (3.35)
∂δp
∆PL(pq)
= −2αpq Vp Vq sin (δpq ) (3.36)
∂δq
∆PL(pq)
= 2αpq (Vp − Vq sin δpq ) (3.37)
∂Vp
∆PL(pq)
= 2αpq (Vq − Vp sin δpq ) (3.38)
∂Vq

3.2.5 Analysis of change in Reactive Power Loss

The reactive power loss in a line connecting two buses, bus p and bus q can be expressed
as:
Xnb
QL(total) = [−bpq sh (Vp 2 + Vq 2 ) − bpq (Vp 2 + Vq 2 − 2Vp Vq cos δpq )] (3.39)
i=1

29
Where;
bsh
pq is the shunt susceptance of the line p-q

bpq is the susceptance of the line p-q

Mathematically, loss sensitivity for the reactive power can be written as:
 
∆QL(pq)
∆Pn
(3.40)
 
 ∆QL(pq)

∆Qn

Ignoring second and higher order terms the change in reactive power loss using Taylor
series approximation can be expressed as:

∂QL(pq) ∂QL(pq) ∂QL(pq) ∂QL(pq)

∆QL(pq) = ∆δp + ∆δq + ∆Vp + ∆Vq (3.41)
∂δp ∂δq ∂Vp ∂Vq

The partial derivatives of reactive power loss with respect to variables δ and V gives the
coefficients appearing in the equation (3.41) above as shown below:

∆QL(pq)
= 2bpq Vp Vq sin (δpq ) (3.42)
∂δp
∆QL(pq)
= −2bpq Vp Vq sin (δpq ) (3.43)
∂δq
∆QL(pq)
= −2[bpq sh Vp + bpq (Vp − Vq cos(δpq )) (3.44)
∂Vp
∆QL(pq)
= −2[bpq sh Vq + bpq (Vq − Vp cos(δpq )) (3.45)
∂Vq

3.2.6 The Power Loss Sensitivity Factors Formulation

The change in the real power loss over a transmission or distribution line connected
between bus-p and bus-q due to the change in active power injected at any other bus
represents the sensitivity factors for real power loss. Similarly, the change in the reactive
power loss over a transmission or distribution line connected between bus-p and bus-q due
to the change in reactive power injected at any other bus represents sensitivity factors
for reactive power loss. To calculate the changes in the line flows the equations can be
arranged in matrix form as:
    
∂PL(pq) ∂PL(pq)
 ∆PL(pq)   ∂δ ∂V   ∆δ 
 = ∂QL(pq) ∂QL(pq)
  (3.46)
∆QL(pq) ∂δ ∂V
∆V

30
Applying the same procedure similar to the sensitivity factors for the real and reactive
power flows, the sensitivity factors for the real and reactive power loss represented by:
     
∂PL(pq) ∂PL(pq)
∂Pn   MP P   T −1  ∂δ
= = J (3.47)
 
 ∂PL(pq)
 ∂PL(pq)

∂Qn
MP Q ∂V

     
∂QL(pq) ∂QL(pq)
∂Pn   MQP   T −1  ∂δ
= = J (3.48)
 
 ∂QL(pq)
 ∂QL(pq)

∂Qn
MQQ ∂V

Where;
MP P is the real power loss sensitivity related to the real power injection
MP Q is the real power loss sensitivity related to the reactive power injection
MQP is the reactive power loss sensitivity related to the active power injection
MQQ is the reactive power loss sensitivity related to the reactive power injection
Here the four sensitivities are column vectors with dimension of number of the system
busses.

3.3 Objective Function Formulation

Integrating DG in the distribution system is mainly to minimize power losses, voltage drop
and to maximize the stability and reliability of the power distribution system. In this
paper, the proposed multi-objective Harish hawks optimization method aims to minimize
the power loss, voltage deviation, and total operating cost of the distribution system.
The performance index is computed for each objective function to optimally locate and
sizing of DG units and to enhance the voltage stability of the radial distribution system.
Mathematically, the expression for objective functions and constraints are formulated in
the following subsection.

3.3.1 Minimizing Real Power Losses

The radial structure of the power system increase active power losses in the distribution
system hence, it is important to reduce the active power losses (PL).

Xnb
M in, P L = (Ii 2 × Ri ) (3.49)
i=1

31
Where;
i is the branch number
Nb is the total number of branches
Ri is branch resistance

3.3.2 Minimizing Reactive Power Losses

Most papers do not consider reactive power loss minimization as the objective function.In
this thesis,the reactive power loss (QL) also included along with real power loss for cal-
culating optimal DG sizes and location.

Xnb
M in, QL = (Ii 2 × X i ) (3.50)
i=1

Where;
Xi is branch reactance

3.3.3 Minimizing Voltage Drop

By installing the DG, the proposed method will try to minimize the voltage drop(VD) by
minimizing the gap between the rated voltage that is usually one and the actual voltage
of the bus nearer to zero. Therefore, it improves the voltage stability and the network
performance.
XN n
M in, V D = (Vni − V rated )2 (3.51)
i=1

Where;
N n is total number of bus
Ni is receiving bus number
Vni is voltage of bus Ni
Vrated is the rated voltage

3.3.4 Minimizing Annual Economic Loss

The problem is finding the best combination of some DG units among available categories
and placing them on suitable buses not only to minimize the power loss but also to

32
minimize the annual cost of losses and installed units. It is not advisable to install a
very high-capacity DG in the network as it increase the cost of real power supplied by
the DG. Here the system has two operational costs one from the real power supplied
from the substation and the other from real power drawn from the DG. The cost from
the substation can be minimized by reducing the total loss of the system and the cost
of the DG can be minimized by reducing the size of the DG. After integrating DGs in
to the network, the total active power loss with DG (P LW DG ) is reduced compared to
total active power loss without DG (P LW ODG ). Therefore, total Annual Economic Loss
without any DG (AECW ODG ) given by the equation below:

AECWODG = PLWODG × Ce × 8760 (3.52)

Where; Ce = Energy loss cost per kWh in birr

Total Annually Economic Loss with DGs (AECW DG ) will be:

PNDG
CDG i=1 PDG
AECWDG = PLWDG × Ce × 8760 + (3.53)
LT

Where;
NDG is Number of DG installed in the system
CDG is Cost of DG generated power per kW
LT is Total DG life in years
So, Annual Savings(AS) is calculated as:

AS = AECWODG − AECWDG (3.54)

3.4 Objective Function Index

To evaluate the performance of the system at each objective function the performance
index namely real power loss reduction index, reactive power loss reduction index, voltage
profile improvement, and annual economic loss reduction index are takes into consideration
in the objective function.

33
3.4.1 Real Power Loss Reduction Index

A common strategy for sizing and placement of DG is to minimize system power loss of
the power system. The total active power loss reduction index (PLRI) is the ratio of
percentage reduction in real power loss from the base case when a DG is installed which
is expressed in the equation as below:

P LWDG
PLRI = (3.55)
P LWODG

Where;
P LW ODG is the total active power loss before DG installation
P LW DG is the total active power loss in the system after installation of DG
By minimizing (PLRI), the reduction in real power losses in the presence of DGs unit will
be a maximum value.

3.4.2 Reactive Power Loss Reduction Index

To determine the effect of DG on reactive power losses, the Reactive Power Loss Reduction
Index is incorporated as an objective function. This refers to the ratio of percentage
reduction in reactive power loss from base case when a DG is installed at bus i.
Reactive Power Loss Reduction Index (QLRI) is expressed as:

QLWDG
QLRI = (3.56)
QLWODG

Where;
QLW ODG is the total reactive power loss before DG installation
QLW DG is the total reactive power loss in the system after installation of DG

3.4.3 Voltage Deviation Index

In a power system, the voltage at each bus should be within the acceptable range and
the line flow within the limits. These limits are important so that the integration of
distributed generation into the system does not increase the cost for voltage control or
replacement of existing lines.

34
The Voltage Deviation Index (VDI) help to identify the size-location pair, which gives
higher voltage improvement from the base voltage and is defined as:

V DWDG
V DI = (3.57)
V DWODG

Where;
V DW ODG is total voltage drop without DG
V DW DG is total voltage drop with DG

3.4.4 Annual Cost Reduction Index

This refers to the ratio of percentage reduction in annual operating cost from base case
when a DG is installed at a selected bus. The Annual cost of reduction Index (ACRI)
defined as follow:
AECWDG
ACRI = (3.58)
AECWODG

3.5 Multi-objective Function Formulation

The separate objective functions index listed above combined together to form the multi-
objective function of the distribution networks using the weighting factor. The MOF is
a multiple index function proposed to be optimal location and sizing of DG units and
to enhance the optimal operation of the radial distribution system. The lowest values of
PLRI, QLRI, and VDI are to reduce the real, reactive power losses and voltage drop of
the distribution system respectively. The multi-objective function (MOF) formulated in
the equation below as:
M OF = W1 + W2 + W3 + W4 (3.59)

The values of weighting factors play an important role in optimizing the multi-objective
problem, and are decided according to the network designer at which it optimize the sys-
tem. The values of the weighting factors will be larger weight for the individual objective
function that outperforms the others higher in terms of importance and benefits and vice
versa. The sum of the weight factors assigned to all individual objective function should
add up to one. That is:
W1 + W2 + W3 + W4 = 1 (3.60)

35
3.6 Constraints Formulation

The aim is to minimize the multi-objective function formulated above subjected to various
inequality and equality operational constraints so as satisfy to the electrical requirements
for the distribution network.

3.6.1 Real and Reactive Power balance constraints

Power balance is similar to the non-linear power flow equations, which state that the
sum of complex power flows at each bus in the distribution system injected into a bus
minus the power flows extracted from the bus including line losses should equal zero. The
constraints for power balance formulated as follows:

Pi = PDGi − (PDi − PLi ) (3.61)

Qi = QDGi − (QDi − QLi ) (3.62)

Where;
Pi is real power flow at bus i in KW
Qi is reactive power flow at bus i in KVAr
PDGi is distributed real power generations at bus i in KW
QDGi is distributed real power generations at bus i in KW
PDi is real power demand at bus i in KW
QDi is reactive power demand at bus i in KVAr
PLi is real power loss at bus i in KW
QLi is reactive power loss at bus i in KVAr

3.6.2 Bus Voltage Limits

This includes the upper and lower voltage magnitude limit, Vmin and Vmax at each node.
Bus Voltage magnitude is to be within acceptable operating limits throughout the opti-
mization process to maintain power quality and failure of electrical device.

36
The bus voltage limit of the system given by:

Vmin ≤ Vi ≤ Vmax (3.63)

Where;
Vmin and Vmax are lower and upper bound of bus voltage limits in p. u

3.6.3 DG Capacities

The DG capacity of each unit should be defined around its nominal value to operate
the system within the acceptable limit. So that each DG unit is maintained within an
acceptable limit. This includes the upper and lower real and reactive power generation
limits of distributed generators (DGs) connected at each bus. The capacity of DG is given
as follows:
P DGmin ≤ P DGi ≤ P DGmax (3.64)

QDGmin ≤ QDGi ≤ QDGmax (3.65)

Where;
P DGmin is the minimum real power generation from DG capacity in KW
P DGmax is the maximum real power generation from DG capacity in KW
QDGmin is the minimum reactive power generation from DG capacity in KVAr
QDGmax is the maximum reactive power generation from DG capacity in KVAr

3.7 Optimization Techniques

Up to now, many intelligent search techniques have been proposed and developed to solve
the optimization problems that are complex problems instead of using traditional methods
due to their accuracy and robustness. In this paper, three different algorithms explained
below used to solve the optimal DG placement problem considering different objectives.

are

37
3.7.1 Genetic Algorithm

There are many optimization algorithm techniques for solving optimization problems.
Due to the limitation of classical optimization methods in finding global minimum value,
the heuristic optimization methods are widely used because of their reliability, flexibility,
and robustness in seeking optimum value in recent years [45]. GA is a heuristic search
method to find the global optimal solution in a complex multi-dimensional search space.
The Genetic Algorithm used to solve an optimization problem with objective function
is
based on the mechanism of the natural selection and evolution that is survival of the fittest.
To find the best fitness, GA needs the initial value and randomly generates individual
solutions. The objective function used to calculate the fitness of each randomly generated
individual population to select the most fitted individuals and let them generate more fit
individuals, just like evolution in nature.
More generations are produced by applying GA operators. This eventually leads to
a generation of high performing individuals. A typical genetic algorithm mainly finds
the optimal point usually based on three operators [46]: production operator (elitism),
recombination (also known as the crossover) operator, and mutation operator. GA needs
to combine to a method that has good convergence such as PSO to obtain a robust solution
for the problem at hand.

3.7.1.1 Choice of GA Parameters

GA parameters are crucial for its faster convergence, depend on its selection. In the
absence of any guiding rule to choose these parameters, some mechanism has to be devel-
oped. The GA parameters include:
Population size
Initially, the genetic algorithm creates many chromosomes randomly under the constraint
conditions. The number of chromosomes named as a possible solution to the problems
is the population size. Enlarging the population size leads to more accurate solutions
increasing the search scope. However, it is not always necessary to have so large popu-
lation. Therefore, the researchers should find a suitable size of population. This initial
population is a set of many possible solutions evaluated on the main goal of the problem
to see if it is a good or bad answer. The genetic algorithm will start from this generation,
and select and inherit the high fitness individuals for the next generation.

38
The best solutions compared to each other until we reach the stopping criteria to find the
best answer.
Selection
The chromosomes of the population sorted based on their fitness value and selecting indi-
viduals with high fitness and grouping these individuals into a new population according
to the fitness value to keep the better individuals for the next generation, thereby making
the individuals in the population come close to the optimal solution gradually. The chro-
mosomes mixed to other chromosomes with less fitness to generate the next population
(or change the current population intelligently). The chromosomes with a higher amount
of fitness copied than others even if each chromosome has its own probability. Generally,
the selection process determines the number of solution to be selected from the current
population to create the new generation, and which of the current solutions should be
erased to make room for new solutions. The most commonly used scheme for the process
of selection is Roulette Wheel Selection. The roulette wheel operates based on the fitness
of each of the solution candidates.
Crossover
In genetic algorithms, the purpose of the crossover operator is to randomize recombi-
nation operations and to change the programming of chromosomes by exchange genetic
information between two selected solutions from the current population (parents) that
produces unique new solutions to be included in the next generation. It is also a recom-
bination process of taking more than one parent solution and producing a child solution
from them. Generally, the types of crossover are one-point, two-point, and N-point, and
random multipoint crossover and the choice of mutation depends on the user. The under-
lying idea in the crossover operation is that “good” solutions may produce better ones by
interchanging their genetic material. A crossover operator is typically applied to a per-
centage of selected solutions determined by the parameter called the crossover probability
(Pc). Typical values for Pc are in the range of 0.6-0.9 [47].
Mutation
Mutation removes the overcome in the crossover by creating new genetic material in the
population to maintain diversity transformation of gene from one generation of a popu-
lation of genetic algorithm chromosomes to the next. It used to changes the new solution
completely from the old result by changing one or more gene values in a chromosome from

39
its initial state. The rate of its occurrence determined by the parameter called the prob-
ability of mutation, Pm, with typical values in the range of 0-1 [48]. GA follows different
types of mutation like bit string mutation, flip bit mutation, uniform and non-uniform
mutation to implement the process.
Elitism
Elitism is the genetic operator implemented to put the most fit individuals in the next
generation and to replace the worst chromosome in the newly generated population with
the best chromosome in the old population. This is operated if the best number in the
newly generated population is worse than that in the old population and to ensure the
algorithm’s convergence. It used to preserve the elite parent.

3.7.2 Particle Swarm Optimization

Particle swarm optimization (PSO) is a metaheuristic algorithm based on the concept of

swarm intelligence. PSO proposed by Eberhart and Kennedy in 1995, considering social
behavior in the flocks of birds that flew together [49]. It takes its operation based on the
movement of organisms and social behavior of animals such as birds, fishes, insects, their
communities, and the members of the entire population maintained through the search
procedure. PSO focused on how they manage as a group, rather than as individuals,
recreating themselves and adapting in accordance with the changes in the surrounding
environment, to search for food or to migrate.
It was imitated the behavior of individual swarms which are a flock of birds, school of
fish, and other insect groups and optimizes a problem by improving, in an iterative way,
candidate solutions called particles regarding a given measure of quality.
In a PSO algorithm, the particle is an individual member & each particle flies around
the multi-dimensional search space with a velocity and position. The particle updates its
velocity and position continuously comparing itself with its own experience and experience
of the neighbor’s particles or the experience of the whole swarm. Then each particle moved
through the information related to best personal position (Pbest) and best global position
(Gbest). The mathematical expression of the algorithm also expressed in the following

40
equations.

(k+1)
Vi = Vik + c1 ∗ r1 (X Pbest − Xki ) + c2 ∗ r2 (X Gbest − Xki ) (3.66)
(k+1) (k+1)
Xi = Xik + Vi (3.67)

Where;
Xik is the current individual position of particle i at iteration k
Xik+1 is modified position of particle i
Vik is the velocity of particle i of the previous vector at iteration k
Vik+1 is modified velocity of particle i
c1 , c2 are random number between [0 1]
XP best is the personal best position of a particle
XGbest is the global best position of the particle

3.7.3 Proposed Harris Hawks Optimization (HHO) Algorithm

Harris Hawks Optimizer (HHO) is a new novel population-based, nature-inspired opti-

mization based on the cooperative hunting behavior of the most intelligent and distin-
guished predator birds in nature. In this intelligent strategy, several hawks from different
directions cooperatively pounce the potential animal (rabbit) in a group for its food.
Initially, a group of hawks is considered as the initial population that tries to chase the
targeted prey rabbit from different directions, and initially, the leader hawk tries to attack
the prey [50]. If the leader hawks fail to grab the prey because of the dynamic nature and
escaping behavior of the prey, the switching tactics are followed, so that the other party
members (hawks in the group) will hit the escaped prey until seized.
HHO is applied for determining the optimal size and location of DG at different
operating power factors (pf) with the aim of minimizing the total active power loss,
reducing the voltage deviation (VD), and increasing the voltage stability (VS) considering
the operational constraints of the distribution system. It is effective for any optimization
problem subject to a proper formulation using exploration and exploitation phases and its
performance is improved using the rabbit location instead of the random location. The
mathematical formulation of each phase of the proposed HHO inspired by exploring a
prey, surprise pounce, and different attacking strategies of Harris hawks presented in the

41
following subsections:

3.7.3.1 Exploration phase

The exploration phase based on the concept of a group of hawks try to perch randomly
on some locations and wait to detect a prey. The group of hawks can track and detect
the prey usually a rabbit by their powerful eyes. The leader hawks perch based on the
position of family members and prey to detect a prey maybe after several hours since the
position of prey cannot see easily.
The exploration process in HHO can be expressed using two strategies. Firstly, when the
hawks and family member locations close enough for the condition of q < 0.5 and the
second strategy is when the hawks located at random trees for condition of q ≥ 0.5. The
mathematical implementation of these strategies is modeled as:

 X rand (t) − r1 |X rand (t) − 2r2 X(t)|

q ≥ 0.5
X(t + 1) = (3.68)
 (X rabbit (t) − Xm (t)) − r3 (LB + r4 (U B − LB))
 q < 0.5

Where;
t is the current iteration
X(t + 1) is the position vector of hawks in the next iteration
Xrand (t) is a randomly selected hawk from the current population
Xrabbit (t) is the position of rabbit
X(t) is the current position vector of hawks
r1 , r2 , r3 , r4 , and q are random numbers between (0, 1), which are updated in each iteration
LB and UB show the upper and lower bounds of variables
Xm is the average position of the current population of hawks
The average position of the current population of hawks mathematically expressed as:

1 XN
Xm (t) = Xi (t) (3.69)
N i=1

Where;
Xi (t) indicates the location of each hawk at iteration t
N denotes the total number of hawks

42
3.7.3.2 Transition from exploration to exploitation

The value of the escaping energy (E) of the prey determines the transition from exploration
to exploitation phase and different exploitative behaviors. The energy of a prey usually
the rabbit decreases considerably during the chasing or escaping behavior. The energy of
a prey mathematically modeled as:

t
E = 2EO (1 − ) (3.70)
T

Where;
E indicates the escaping energy of the prey
T is the maximum number of iterations
Eo is the initial state of its energy changes randomly between -1 and 1
When the value of Eo decreases from 0 to -1, it indicates that the rabbit is physically
flagging and when the value of Eo increases from 0 to 1, the rabbit is strengthening.
During the iterations, the initial state of energy Eo have a decreasing or increasing trend
while the dynamic escaping energy E has a decreasing trend. The HHO performs the
exploration phase when the escaping energy |E| ≥ 1, hence, the hawks search different
regions to explore a rabbit location, and HHO starts exploiting near to the rabbit place
when |E| < 1.

3.7.3.3 Exploitation phase

In the HHO, the chance of the rabbit to escape (r) and escaping energy (E) determine
whether the exploitation phase implemented or not. In this phase, the Harris’ hawks
perform the switching tactics to attack the intended prey based on the escaping behaviors
of the prey and the chasing strategies of the Harris’ hawks. The exploitation phase
performed in four possible strategies to model the attacking stage. When r < 0.5, the
chance of a prey named as successfully escaping and not when r ≥ 0.5 before surprise
pounce.
The hawks increase the probability of killing the rabbit by attacking the prey from
different directions named as a hard or soft besiege. After several minutes, the escaping
prey will lose its energy and the hawks get closer to the intended prey to intensify the
besiege process and effortlessly catch the exhausted prey.

43
The HHO optimization utilizes the escaping energy E to switch between soft and hard
besiege processes. In this regard, when r ≥ 0.5 and |E| ≥ 0.5, the soft besiege happens,
and when r ≥ 0.5 and |E| < 0.5, the hard besiege occurs.

3.7.3.4 Soft besiege

During soft besiege, the rabbit has enough energy and tries to escape with the aid of
random jumps notwithstanding the hawks surround it softly. The soft besiege performed
when both of the chance of the rabbit to escape (r) and absolute value of the escaping
energy (E) ≥ 0.5. During these attempts, the Harris’ hawks encircle it softly to make
the rabbit more exhausted and then perform the surprise pounce. Mathematically this
besiege modeled as follow:

X(t + 1) = ∆X(t) − E|JX rabbit (t) − X(t)| (3.71)

J = 2(1 − r5 ) (3.72)

∆X(t) = X rabbit (t) − X(t) (3.73)

Where;
∆X(t) is the difference between the position vector of the rabbit and the current location
in iteration t
r5 is a random number inside (0, 1)
J represents the random jump strength of the rabbit throughout the escaping procedure

3.7.3.5 Hard besiege

The hard besiege performed when r ≥ 0.5 and |E| < 0.5, the prey is so exhausted and it
has a low escaping energy while the hawks hardly surrounded the intended prey to finally
perform the surprise pounce. In this situation, the action mathematically formulated as:

X(t + 1) = X rabbit (t) − E|∆X(t)| (3.74)

3.7.3.6 Soft besiege with progressive rapid dives

This besiege performed when r < 0.5 and |E| ≥ 0.5 in which the rabbit has enough
energy to successfully escape and still a soft besiege is constructed before the surprise

44
pounce. The main besiege that distinguishes the HHO over the other swarm methods
is soft besiege with progressive rapid dives, which is an intelligence strategy than the
previous case. This besiege is mathematically formulated based on the concept of the
levy flight (LF), the escaping patterns of the prey, and leapfrog movements as follow:

Y = X rabbit (t) − E|JX rabbit (t) − X(t)| (3.75)

Where;
Y denotes the soft besiege position
The hawks dive based on the LF-based patterns using the following equation:

Z = Y + S × LF (D) (3.76)

Where;
D indicates the problem dimension
S is a vector of random values with dimension 1 × D
The levy flight (LF) mathematically expressed as:

w×δ
LF (x) = 0.01 × 1 (3.77)
uβ
Γ(1 + β × sin( πβ
2
)) 1
δ=( ) (3.78)
( β−1
Γ( (1+β)
2
) ×β×2 2
) β

Where;
w and u are random values inside (0,1)
β is a default constant set to 1.5
Hence, the final strategy for updating the positions of hawks in the soft besiege phase can
be performed by equation (3.79):

 Y if F (Y ) < F (X(t))

X(t + 1) = (3.79)
 Z if F (Z) < F (X(t))


Where;
Y and Z are obtained using equation (3.75) and (3.76)

45
3.7.3.7 Hard besiege with progressive rapid dives

During this besiege the rabbit has not enough energy to escape and it is performed when
|E| < 0.5 and r < 0.5. The hard besiege is constructed before the surprise pounce to
catch and kill the prey. The situation of this step in the prey side is similar to that in
the soft besiege, but this time, the hawks try to decrease the distance of their average
location with the escaping prey. Therefore, the following rule is performed in hard besiege
condition:

 Y if F (Y ) < F (X(t))

X(t + 1) = (3.80)
 Z if F (Z) < F (X(t))


Where; Y and Z are obtained using equation bellow as:

Y = X rabbit (t) − E|JX rabbit (t) − Xm (t)| (3.81)

Z = Y + S × LF (D) (3.82)

In the proposed HHO algorithm, if the hawks becomes out of the position limits, the
following equation is used to restrict the hawks within the minimum and maximum limit
of the problem variables:

X(t + 1) Xmin ≤ X(t + 1) ≤ Xmax





X(t + 1) = Xmax X(t + 1) > Xmax (3.83)



 Xmin

X(t + 1) < Xmax

3.7.4 Application of HHO in DG allocation

The application of the HHO into DG allocation can be summarized in the following steps:
Where;
Step 1: Read the system data ( line and load data) and define the objective function.
Step 2: Randomly initialize a set of hawks’ searches within the upper and lower limits of
the DG sizes and locations, HHO parameters, and maximum number of iterations Kmax .
Step 3: Rub the power flow and calculate the objective function for each search hawk.
Step 4: Store the best solution Xrabbit.

46
Step 5: Update the parameters of HHO (E, Eo , and J).
Step 6: Update the sizes and locations of the best solutions based on the exploration and
exploitation phases’ strategies.
Step 7: Check the sizes and locations limits and update the position
Step 8: Check if (k < Kmax ) Step 2.
Step 9: Return the final best solution stored ( DG location and sizes).
Step 10: Run the power flow and obtain the voltage profile.
In this paper, a novel population-based, multi-objective Harris Hawks Optimization al-
gorithms mainly based on the intelligence behavior of the Harris hawks in chasing prey
is proposed and applied for determining the optimal placement of distribution generation
(DG) in the radial distribution systems.

47
Chapter 4

MODELING OF DFIG BASED

WIND ENERGY GENERATION
SYSTEM

4.1 Overview of Wind Energy Generation System

The generation and integration of wind power to the grid consists of mechanical and elec-
trical parts like wind turbines, gear, and electrical generators. The size and type of the
wind turbine to generate a large amount of energy determined by the wind speed sources,
is
cut-in and cut-out wind speed mainly to contribute significant renewable fraction and this
can be performed using single large wind turbine or number of smaller turbines. The elec-
trical generators mostly used today to convert the wind energy system are either doubly
fed induction generator (DFIG) or permanent magnet synchronous generator (PMSG)
type. WECS with a Permanent magnet synchronous generator use the synchronous gen-
erator to convert the mechanical power from the wind available in the area to electrical
power to the grid. PMSG has fewer mechanical stress issues and high efficiency with less
maintenance due to the absence of a gearbox [51]. The PMSG is usually a multi-polar
generator, which is relatively makes the system large and heavy and causes inconvenience
for the installation and naturally needs permanent magnets, which will increase the cost
for this wind turbine concept considering the current market. The power converters used
in PMSG based WTGS is a full-scale power rating, which will cause high losses, generate
high harmonic components, and have high cost [52].

48
 WECS with DFIG uses a wound rotor induction electric generator where both windings
transfer significant power between shaft and grid. The rotor side windings of the doubly
are
fed induction generator connected to a back-to-back converter via slip rings, while the
stator side windings directly connected to the grid. The power converters could control
the rotor frequency and thus the rotor speed. The power rating of the power converters is
typically about 25-30% total generated power since the rotor of the doubly fed induction
generator would only deal with slip power and the rest fed to the grid directly from the
stator [53]. The smaller rating of the power converters leads to many advantages, such as
reduced converter cost, reduced filter volume and cost, less switching losses, less harmonic
injections into the connected grid, improved overall efficiency, and reduce overall cost of
the system. Besides, this type of wind turbine can also achieve the desired reactive power
compensation. The doubly fed induction generator has additional features that allow
them to run at a speed slightly above or below their natural synchronous speed which is
useful for variable speed wind turbine because wind speed can change suddenly.
Nowadays, the most frequently used wind energy conversion system is using DFIG-
based wind turbine-generator system in industry for large wind turbines and a variable
speed generator is the preferred option in newer wind turbine installations because of the
fact that machine with this technology exhibits more power capture and less mechanical
stress [54]. The paper also explains the technology of integrating generator, gearbox, main
shaft, and shaft bearing within a common housing, variable speed operation over a large,
but restricted, range makes WECS with doubly fed induction generator interesting. This
concept makes the system less weight, less size of generators, become cheaper and more
reliable system than that of the standard one, but it loses its efficiency.
Considering the merits of the DFIG systems, this thesis will only focus on the modeling
and control schemes of DFIG based wind turbine generator systems. Nordex N80/2500
wind turbine taken for this analysis because among the wind turbine available in the
market, it start generating power at the lowest cut-in wind speed of 3 m/s. This property
is beneficial to low speed regions like those considered in this thesis.The generation of
wind energy system requires modeling of wind turbine and generator to integrate the
power to the grid.

49
4.2 Modeling of Wind Turbine system

The modeling of wind turbine systems involves the aerodynamic wind turbine modeling,
the drive train system modeling, and the modeling of the wind turbine controller.

4.2.1 Aerodynamic Modeling

There are two arrangement in WECS that converts kinetic energy of air in motion into
electrical energy. The first arrangement converts the kinetic energy of the wind flow into
the mechanical power to turn the wind turbine rotor. The second one is the generator that
transforms the rotational mechanical power to electrical power. The wind speed passing
is
through a turbine rotor plane and the available input power in the wind (Pw ) related in
the equation below as:
1
pw = ρAVw3 (4.1)
2

Where;
ρ is the destiny of air in kg/m3
A is the swept area measured in m2
Vw is the available wind speed in m/s
However, the power that can be converted for useful work is limited by a factor called
“Betz Limit” which is 0.5925. That means the maximum efficiency that the wind turbine
converts the available power into useful power is 59.25%.
Maximum turbine output mechanical power available given as follow:

1
pt = ρAVw3 Cp (4.2)
2
Pt
Cp = (4.3)
Pw

Where;
Pt is the output power of wind turbine
Pw is the available or input power of wind turbine
Cp is the wind power coefficient
For a given wind turbine, if we fix the pitch angle beta at a constant value then the tip
speed ratio (TSR) only determines the wind power coefficient.

50
The tip-speed radio (TSR) defined as:

Rωturb
λ= (4.4)
Vw

The turbine rotor torque and the power output of the wind turbine related in the equation
below as:

Pt ρAVw3 Cp ρπR3 Vw3 Ct

Tturb = = = (4.5)
ωturb 2ωturb 2

Where;
Ct is the coefficient of torque
The coefficients of power Cp and torque Ct are related by the equation below as:

CP (λ) = λCt (λ) (4.6)

4.2.2 Mechanical Shaft and Gearbox Modeling

The DFIG based wind energy conversion system includes a gearbox, which connects the
slow shaft of the turbine to the fast shaft to increase the speed of the input shaft to
that required by the generator. An increase in speed is needed because wind turbine
rotors, and hence main shafts, turn at a much slower speed than required by electrical
are
generators. The gearbox model categorized as a single mass, two mass, and three mass
models based on the speed–up ratio of a gearbox and number of stages. The two mass
model is developed by neglecting the moment of inertia for the shafts and the gearbox
wheels because they are small compared with the moment of inertia of the wind turbine or
generator. Then, the one-mass model which is obtained in this case can be described by
further neglected the stiffness and the damping factor for both low speed and high-speed
shafts. The equivalent physical model considered for the representation of the mechanical
transmission system dynamics shown in figure 4.1.
is
Then, the model of the drive system in this case described by the following equation:

dωm
Ttotal = Jtoal (4.7)
dt

51
The equivalent torque, the equivalent moment of inertia and the mechanical speed given
by:

Ttotal = Tm − Tem (4.8)

Jturb
Jtoal = Jg + (4.9)
N2
ωm = N ∗ ωturb (4.10)
1
Tm = Tturb (4.11)
N

Where;
N is the gear box ratio
Tm is the torque applied to the shaft of the generator
Tem is the generator electromagnetic torque
Ttotal is the total torque
ωm is the generator mechanical speed
Jtotal is the total moment of inertia
Jturb is wind turbine moment of inertia
Jg is the generator moment of inertia

Figure 4.1: Mechanical arrangement of wind turbine drive train

4.2.3 Controller for Wind Turbine

The controller is based on the most successful control strategy named as MPPT and its
objective is to rotate the turbine over a certain wind speed range in order to maintain

52
the tip speed ratio (λ) at its optimal value λopt which makes the turbine to operate at
optimum power coefficient (Cpopt ).
The optimum tip−speed radio (TSR) calculated as:

opt
Rωturb
λopt = (4.12)
Vw

For a wind turbine, from the mechanical modelling of the systems:

dωturb
Jturb = Tturb − Tm (4.13)
dt

The controller of the wind turbine is achieved through controlling ωturb by adjusting the
electromagnetic torque of the generator (Tem ). Here we adopt maximum power point
∗
tracking controller to generate the torque reference signal Tem .
When the turbine is working on the maximum power point, the maximum turbine torque
extracted by the turbine is then given by:

ρπR3 Vw3 Ctopt ρπR5 ωturb

2
Cpopt
Tturb (max) = = (4.14)
2 2(λopt )3

Moreover, from the mechanical modeling of the wind turbine;

Tturb
Tm = (4.15)
N

Substituting equation (4.14) into equation (4.15), we can get the maximum mechanical
torque given by:

ρπR5 ωturb
2
Cpopt ρπR5 Cpopt ωm
2
2
Tm (max) = = = Kopt ωm (4.16)
2(λopt )3 N 2(λopt )3 N 3

Where;

ρπR5 Cpopt
Kopt = (4.17)
2(λopt )3 N 3

∗
Now set the Tem value at Tm (max) such that the controller operate to extract maximum
power from each variable wind speed.

53
4.3 Doubly-Fed Induction Generator (DFIG) Mod-

eling

It is essential to know the DFIG dynamic modelling for the purposes of better under-
standing and designing control schemes for DFIG based wind turbine-generator system.
The space vector model of the DFIM can be represented into both αβ and dq reference
frame. When the DFIG is connected to a three phase balanced supply, the relationships
between voltages, currents and flux linkages of each phase of the machine on the stator
and rotor side given by:

dΨs
Vs = Rs Is + (4.18)
dt
dΨr
Vr = Rr ir + (4.19)
dt
ψs = Ls is + Lm ir (4.20)

ψr = Lr ir + Lm is (4.21)

Where;
The subscripts s denotes that one space vector is referred to the stator
The subscripts r denotes that one space vector is reference to the stato
The symbols v,i and Ψ represents voltages, current and flux linkages respectively
Rs and Rr represents the stator and rotor winding resistances respectively
Ls represents the stator leakage inductance
Lm represents mutual leakage inductance
Lr represents the rotor leakage inductance
Modelling and analysis of voltage, current and mutual inductance in (abc) stationary
reference frame depends on time. Such system is difficult to manage and needs to convert
time-variant parameters in to time-invariant by transforming them first into two-phase
stationary reference frame and then transforming to an appropriate two-phase rotating
reference frame.

4.3.1 Modeling in αβ Stationary Reference Frame

Modeling the DFIG in αβ reference frame model requires space vector theory to the basic
electric equations of the machine and it resembles the dq modeling but it is stationary.

54
Typically, three different rotating reference frames utilized to develop space vector-based
models of the DFIM. The stator reference frame (αβ) is a stationary reference frame,
the rotor reference frame (DQ) rotates at ωm and the synchronous reference frame (dq)
rotates at ωs .
The voltage equations in equation (4.18) and (4.19) represented in to two-phase stationary
reference frame (αβ) using the following set of equations:

dΨαs
vαs = Rs iαs + (4.22)
dt
dΨβs
vβs = Rs iαs + (4.23)
dt
dΨαr
vαr = Rr iαr + ωm ψβr + (4.24)
dt
dΨβr
vβr = Rr iβr − ωm ψαr + (4.25)
dt

Similarly, the stator and the rotor flux expressions derived in αβ stationary reference
frame expressed as:

ψαs = Ls iαs + Lm iαr (4.26)

ψβs = Ls iβs + Lm iβs (4.27)

ψαr = Lr iαr + Lm iαr (4.28)

ψβr = Lr iβr + Lm iβr (4.29)

Where;
ωm is the rotational mechanical speed in (rad/s)
vαs , vβs , vαr , vβr , iαs , iβs , iαr , iβr are voltages and currents of the stator and rotor in α and
β axis
ψαs , ψβs , ψαr , ψβr are flux linkages of the stator and rotor in α and β axis
Here we use a transformation matrix to calculate voltage and current in αβ reference
frame given by:
 
 Xa 
   
 Xα   cos θs sin θs   
 =  Xb 
 (4.30)
Xβ − sin θ cos θs  
Xc

55
Where;
Xα and Xβ represents voltage and current in αβ reference frame separated by 90◦
Xa , Xb and Xc represents voltage and current in abc reference frame separated by 120◦

Figure 4.2: Schematic diagram of abc to αβ stationary reference frame transformation

4.3.2 Modeling in dq Synchronously Rotating Reference Frame

The space vector model of the DFIM represented by defining the dq reference frame,
which will be rotating at the synchronous angular speed of the system. The two-phase
synchronously rotating dq reference frame voltage equations formulated as below:

dΨds
vds = Rs ids − ωs ψqs + (4.31)
dt
dΨqs
vqs = Rs iqs + ωs ψds + (4.32)
dt
dΨdr
vdr = Rr idr − ωr ψqr + (4.33)
dt
dΨqr
vqr = Rr iqr + ωr ψdr + (4.34)
dt

56
Similarly, the expression for fluxes given by:

ψds = Ls ids + Lm idr (4.35)

ψqs = Ls iqs + Lm iqr (4.36)

ψdr = Lr idr + Lm ids (4.37)

ψqr = Lr iqr + Lm iqs (4.38)

Where;
ωs is the angular slip frequency (rad/s)
omegar is the generator angular electrical speed in (rad/s)
vds , vqs , vdr , vqr , ids , iqs , idr , iqr are voltages and currents of the stator and rotor in d and q
axis
ψds , ψqs , ψdr , ψdr are flux linkages of the stator and rotor in d and q axis
Here we use a transformation matrix to calculate voltage and current in d and q reference
frame given by:
    
 Xd   cos θs sin θs   Xα 
 =   (4.39)
Xq − sin θ cos θs Xβ

Figure 4.3: Schematic diagram of αβ to dq synchronously rotating reference frame trans-

formation

In order to align the dq reference frame with the stator flux, the angle of the stator can
be calculated by representing the stator flux in the (abc) reference frame to αβ stationary

57
reference frame as:

ψs = ψαs + jψβs = |ψs |hθs

ψβs
θs = tan−1 ψαs (4.40)
q
|ψs | = ψαs 2 + ψβs 2

The angle θs represents the location of the stator’s rotating magnetic field at an instant
of time.
There is a production of magnetic energy due to the mutual flux between rotor and
stator, which is stored in the magnetic field. This stored energy responsible to produces
an electromagnetic torque in the stator. The electromagnetic torque in the stator, the
stator and rotor active and reactive powers for the DFIG represented in dq reference frame
as:

3
Ps = (vds ids + vqs iqs ) (4.41)
2
3
Qs = (vqs ids − vds iqs ) (4.42)
2
3
Pr = (vdr idr + vqr iqr ) (4.43)
2
3
Qr = (vqr idr − vdr iqr ) (4.44)
2
3
Tem = (idr iqs − iqr ids ) (4.45)
2

4.3.3 Control Systems for DFIG

The control of doubly fed induction generator (DFIG) completed by control of the variable
frequency converter, which mainly related to the grid side and the rotor side controllers
as well. Mainly the controller controls the active and reactive power in DFIG based wind
energy conversion system. Among the different alternative control methods developed for
the DFIG, only the vector control in synchronous reference frame strategy is chosen for
implementing the control strategies for the rotor and grid side converter. The operation of
vector control of the DFIG performed in a synchronously rotating dq frame, in which the
d-axis is oriented with the stator flux so that the active and reactive power of the stator
controlled independently by means of the quadrature and the direct current, respectively.
The control of DFIG required mainly two sub-control levels namely Rotor side con-
verter controller and Grid side converter controller.

58
The objective of the RSC is to allow for decoupled controlling the active and reactive
power. This facilitates high flexibility which enables the turbine to capture maximum
energy from wind and to provide reactive power support to the grid. The objective of the
GSC is to maintain constant voltage across the capacitor irrespective of the magnitude
and direction of the rotor power.

4.3.3.1 Rotor side control

The rotor currents in abc stationary reference frame can be resolved into the quadrature
(idr ) and the direct current (iqr ) components. The direct current component produces
a flux in the air gap, which is aligned with the rotating flux vector linking the stator,
whereas the quadrature component produces flux at right angles to this vector. Hence
ψds = ψs and ψqs = zero.

Figure 4.4: Synchronous rotating dq reference frame aligned with the stator flux space
vector

Because the stator is connected directly to the grid at constant AC voltage, the stator
flux is constant, and assuming that the stator resistance is small, then, the term (dψds )/dt
and Rs ids in equation (4.31) are zero. Consequently, equation (4.31) and (4.32) becomes
as follows:

dΨds
vds = Rs ids − ωs ψqs + =0 (4.46)
dt
dΨqs
vqs = Rs iqs + ωs ψds + = ωs ψs (4.47)
dt

59
Given that ψds = Ls ids + Lm idr and ψqs = Ls iqs + Lm iqr , we have:

ψds − Lm
ids = idr (4.48)
Ls
−Lm
ids = iqr (4.49)
Ls

The electromagnetic torque, active and reactive powers of the stator calculated above
reduced to the equation below:

3 ψs ωsLm
Ps = ( iqr ) (4.50)
2 Ls
3 L ψs ψs
Qs = ψs ωs m (idr − ) = KQ (idr − ) (4.51)
2 Ls Lm Lm
3 L
Tem = − p m ψs iqr = KT iqr (4.52)
2 Ls

The above last three equations reveal that the direct rotor current is proportional to the
stator reactive power and that the quadrature rotor current is proportional to the torque
and stator active power. The component idr then controls the reactive power and that
of iqr controls the torque or active stator power. The reference signal can be obtained
from the desired electromagnetic torque (or the desired stator real power) and the desired
stator reactive power, as shown below:

2 Ls 2 Ls
i∗qr = ( ∗
)Tem = 2
Kopt ωm (4.53)
3 pψs Lm 3 pψs Lm
ψs 2 Ls
i∗dr = + Q∗ (4.54)
Lm 3 ψs ωs Lm s

The controller PI parameters are determined by comparing with the Butterworth poly-
nomial that is described in the below section.
If we substitute equation (4.35 - 4.38) into equation (4.33) and (4.34), we have get:

didr Lm d|Ψs |
vdr = Rr idr + σLr − ωr σLr iqr + (4.55)
dt Ls dt
diqr Lm
vqr = Rr iqr + σLr + ωr σLr idr + ωr |Ψs | (4.56)
dt Ls

Where;
L2m
σ =1− Ls ∗Lr
Lm d|Ψs |
The stator flux is constant and ψqs = 0 consequently, the term Ls dt
is zero.

60
If we take Laplace transform on both side of last two above equation, we can get the
following equation in frequency domain as:

vdr + ωr σLr iqr

idr = (4.57)
Rr + σLr s
vqr − σLr idr ωr LLms
iqr = |ψs | (4.58)
Rr + σLr s

Let us draw the closed-loop with PI current controller of idr and iqr with their respective
reference value of i∗dr and i∗qr respectively as in the figure below:

Figure 4.5: Closed loop PI current controller diagram of idr and iqr

The block diagram in figure 4.5 can be reduced to a second-order closed-loop system
for both currents with appropriate gain of the PI regulator as below:

61
 Figure 4.6: Reduced second order closed loop system (a) for idr and (b) for iqr

The gain function is given by:

(sKp +Ki )
idr(s) iqr(s) sKp + Ki M
∗
= ∗
= 2
= (Kp +Rr )
(4.59)
idr(s) iqr(s) σLr s + (Kp + Rr )s + Ki s2 + +Ki
M M

Where; M = σLr
The transfer function of a Butterworth polynomial of second order denominator is given
as:

C(s) ωn2
= 2 (4.60)
R(s) s + 2ωn + ωn2

Where;
C(s) is out put of the transfer function
R(s) is input put of the transfer function
The controller PI parameters determined by comparing the denominators coefficient in
equation (4.59) with the denominators coefficient of the Butterworth polynomial and then
choosing appropriate ωn .

62
Therefore;

(Kp + Rr )
2ωn = (4.61)
M
Kp = 2ωn M + Rr (4.62)
Ki
= ωn2 (4.63)
M
Ki = ωn2 M (4.64)

Where;
ωn is the bandwidth of the current controller, which depends upon the design value
Similarly, following the procedure above, we can get a second-order closed-loop system
for mechanical speed with appropriate gain of the PI regulator as below:

𝑆𝐾𝑝𝜔 + 𝐾𝑖𝜔
𝜔𝑚 ∗ 𝐽 2 ωm
𝑝 𝑆 + 𝐾𝑝𝜔 𝑆 + 𝐾𝑖𝜔

Figure 4.7: Reduced second order closed loop system for mechanical speed

Then, the PI parameters of the mechanical speed is calculated to be:

J
KP ω = 2ωn (4.65)
P
J
Kiω = ωn (4.66)
P

4.3.3.2 Grid side controller

To design the grid side controller similar to the rotor side controller the system configured
by the grid side converter, filter, and grid voltage can ideally represented by the equation
bellow as:

didg
vdf = Rg idg − ωs Lg iqg + Lg + vdg (4.67)
dt
diqg
vqf = Rg iqg + ωs Lg idg + Lg + vqg (4.68)
dt

63
Where;
Vdf and Vqf are the voltage imposed by the grid side converter in α and β reference frame
Vdg and Vqg are the grid side voltage in dq reference frame
Lf is inductance of the grid side filter
Rf is resistive part of the grid side filter
In order to perform the vector control technique, the voltage along d axis of the rotating
frame is aligned with the grid voltage space vector vg as shown in Figure ?? giving the
resulting dq components of the grid voltage as:

vdg = |vg | (4.69)

vqg = 0 (4.70)

Thus, the total active and reactive powers exchanged with the grid are calculated as:

3
Pg = (vdg idg + vqg iqg ) (4.71)
2
3
Qg = (vqg idg − vdg iqg ) (4.72)
2

Substituting the relations in equation (4.69) and (4.70) into the above last two equation;
the power calculation simplified as:

3 3
Pg = vdg idg = |vg |idg (4.73)
2 2
3 3
Qg = − vdg iqg = − |vdg |iqg (4.74)
2 2

The above last two equation indicate that idg control implies Pg control, while iqg control
implies Qg control. The current references i∗dg and i∗qg are calculated from the grid side
active and reactive powers as:

Pg
i∗dg = 3 KP G Pg (4.75)
v
2 dg
Qg
i∗qg = = KQG Qg (4.76)
− 23 vdg

Where;
1
KP G = 3
v
2 dg

64
 1
KQG = − 32 vdg

Similar to the rotor side controller, the current idg and iqg can be expressed in the frequency
domain as:

vdf + ωs Lg iqg
idg = (4.77)
Rg + sLg
vqf − Lg idg
iqg = (4.78)
Rg + sLg

Then parameters of the PI regulators determined by the equation below as:

Kpg = 2ωng Lg − Rg (4.79)

2
Kig = ωng Rg (4.80)

Where;
ωng is is the bandwidth frequency of the reactive power controller
The GSC mainly controls the voltage across the capacitor and its PI parameter can be
calculated by expressing the DC voltage dynamics in DC-link as:

CVdc = Kvdc (Vdc∗ − Vdc ) (4.81)

Where;
Kiv
 
Kvdc is PI controller for the DC voltage given as: Kvdc = Kpv + s

Then, The equation (4.81) written as:

   
Kiv ∗ Kiv
CVdc = Kpv + Vdc − Kpv + Vdc (4.82)
s s

The gain function becomes:

1
Vdc C
(sKpv + Kiv )
∗
= (4.83)
Vdc s2 + s KCpv + KCiv

Then the PI parameter calculated to be:

Kpv = 2Cωdc (4.84)

2
Kiv = Cωdc (4.85)

65
Chapter 5

SIMULATION STUDIES AND

RESULT ANALYSIS

5.1 System under Study

This thesis work performed on the Sheno feeder of the Debre Berhan power distribution
substation and the single line schematic diagram of the substation is shown in figure 5.1
below. This chapter begins by describing the performance of the Sheno feeder without
the installation of distributed generators. The total real and reactive power loading of the
system is 8.0284MW and 4.7223kVAr respectively. It also presents the result obtained
using HHO Simulation Algorithm. The implementation steps of the proposed algorithm
is programmed using Matlab software. The obtained results subdivided into different
sections depending on the number of DG being optimally place and sized. In other words,
one or more DGs have been installed in the test systems for more economic and technical
benefits.
The maximum and minimum ratings of DGs are taken as 80% and 10% of total connected
loads respectively [55]. All the buses are considered to be possible candidate buses for DG
placement except the slack bus. The IEEE standard minimum voltage used is 0.95 and
the maximum voltage is 1.05, which is ±5% compared to the rated value.. The optimum
DGs power factor is determined for maximum capacity utilization and hence, optimum
benefit can be extracted from DGs.
The substation supplied from 132KV grid transmission line and there is two main
power transformer in the substation supplied the city with a primary voltage of 15 and

66
33KV. There are eight (8) radially configured feeders to distribute primary voltage level
power to the distribution transformer and industrial loads. The 33KV voltage supplies
four outgoing feeders, these are: Sheno, Enwary, AliuAmba, Dashin, and 15KV side
similarly have four outgoing feeders such as blanket factory, Ankober, Mendida, and
Sheno2.

Sendafa

132 KV

T2
T1
20 MVA
20MVA

15 KV 33 KV

Blanket Mendida Ankober Sheno Sheno2 Enwary Aliuamba Dashin

2.5MVA 4.35MVA 4MVA 9.3MVA 1.115 MVA 5.515 MVA 3.65 MVA 3.5 MVA

Figure 5.1: Single line diagram of Debre Berhan substation out going feeders

All required data including line resistance and load data, power interruption frequency,
and duration are collected from the Debre Berhan district utility. To reduce the complex-
ity of the analysis in the system the Sheno (15kV) feeder selected due to:
is
• They fed power to all kinds of loads such as industrial, commercial, and domestic

• High numbers of customers are connected and rate of load increment is higher.

• Served more Debre Berhan town customers than the other feeders

• There is a higher imbalance between the load and supply even if the substation try
to supply the load with two feeders

Table 5.1: Data for main transformer in Debre Berhan distribution substation

Transformer type Transformer quantity Voltage level (KV) Transformer capacity(MVA)

Main trnasformer1 1 33 20
Main trnasformer2 1 15 20

67
 Table 5.2: Distribution transformer and load data for Sheno feeders

Transformer Total number Active Reactive

quantity
KVA rating of transformer power (MW) power (MVAr)
25 4
50 2
100 8
200 4 31 8.0674 4.7223
315 6
630 1
800 3

5.2 Cause of Interruption in the Feeder

Among the different types of causes for power interruption, frequently occurring at Sheno
feeder include component failure(CF), component maintenance(CM), line shedding, bro-
ken line, animal, wind and tree failure(W,A&T), and unkowun cause(UC) causing per-
manent and transient earth fault(EF), permanent and transient short circuit(SC), open
circuit fault(OC) and longer interruption duration for operation and maintenance. The
interruption duration is taken for one year to identify the main cause of power interruption
in the feeder, the duration of the cause of faults, and the amount of energy not supplied
due to interruption.

Table 5.3: Annually interruption duration and Cost of ENS for Sheno feeder

Types of fault occur due to the cause

Cause ENS
EF SC OC OI
Dur Freq Dur Freq Dur Freq Dur Freq
of fault (MWh)
(Hr.) (Int/yr.) (Hr.) (Int/yr.) (Hr.) (Int/yr.) (Hr.) (Int/yr.)
CF 95.46 10 75.55 13 69.64 9 8.55 2 168.21
W, A
50.70 12 52.5 15 49.70 10 14.5 7 112.995
&T
CM 134.87 27 115.30 11 85.80 14 20 4 355.97
UC 106.56 19 95.65 20 96.75 7 17.5 3 213.61
Total 387.6 68 339 59 301.9 40 60.5 19 850.78

The peak load of the feeder, which is 0.675 MW is used to calculate energy not supplied
due to interruption in table 5.3 above.

Table 5.4: Percentage duration and frequency of each causes of fault

Cause CF W, A & T CM Unknown cause

%DR 19.77 13.28 41.84 25.11
%FR 18.78 24.31 29.83 27.07

68
 As we observe from table 5.4, the feeder experiencing longer interruption duration
and frequency of interruption due to component operation and maintenance occurred in
the feeder. The total contribution of each cause for power interruption in the feeder is
presented in figure 5.2 below.

% duration

19.77%
25.11%

13.28%

41.84%

CF W,A&T CM UC

Figure 5.2: Percentage duration of each causes for power interruption

As can be seen from the chart above, 41.8 % of the interruption duration is due the
component operation and maintenance, 19.77% is due to component failure, 13.28% is
due to wind, animal, and tree fall. The remaining 25.11% account for unknown causes.
The main cause of power interruption is component operation and maintenance. The
most frequently occurred fault in the feeder is the earth fault, which accounts for 37.15%
of the total interruption frequency.

Table 5.5: Percentage duration and frequency of each types of fault

Types of fault EF SC OC OI
%DR 35.6 31.13 27.72 5.6
%FR 37.15 32.24 21.8 8.7

The total percentage of frequency interruption of each fault type in the feeder is
presented in figure 5.3 below.

69
 % interruption frequency

8.70%

37.15%
21.80%

32.24%

EF SC OC OI

Figure 5.3: Percentage interruption frequency of each faults for power interruption

The most frequently occurred fault in the feeder is earth fault, which accounts for
37.15% of the total interruption frequency.

5.3 Simulation result for different cases

To analysis the power loss and voltage drop, the following case studies are undertaken to
the Sheno feeder.

1. Base case load flow (without DG)

2. Single DG placement

3. Two DG placement

5.3.1 Results during Base case load flow (without DG)

Initially, load flow was run on the selected Sheno feeder to get the voltage at each bus, and
the total power loss of the system. The system active and reactive power loss during base
case load flow analysis is 1629.04KW and 609.513KVAr respectively. The total voltage
drop index of the existing system is 0.4407. The bus voltage profile obtained without DG
shown in table 5.6 below.

70
 Table 5.6: Bus Voltage profile without installation of DG

Bus number Voltage (pu) Bus number Voltage (pu)

1 1.0000 21 0.8581
2 0.9673 22 0.8560
3 0.9452 23 0.9295
4 0.9424 24 0.9159
5 0.9393 25 0.9068
6 0.9377 26 0.8988
7 0.9363 27 0.8907
8 0.9353 28 0.8839
9 0.9348 29 0.8781
10 0.9346 30 0.8752
11 0.9344 31 0.8738
12 0.9188 32 0.8715
13 0.9074 33 0.8708
14 0.9013 34 0.9281
15 0.8874 35 0.9270
16 0.8816 36 0.9267
17 0.8735 37 0.8529
18 0.8682 38 0.8507
19 0.8641 39 0.8500
20 0.8603 40 0.8497

Voltage profile before DG instalation

1

0.98

0.96
Voltage profile (P.U)

0.94

0.92

0.9

0.88

0.86

0.84
0 5 10 15 20 25 30 35 40
Number of Buses

Figure 5.4: Bus voltage profile without installation of DG

5.3.2 Results during Single DG placement with single objective

The objective function here is only to minimize the power loss. The optimal DG size and
location obtained with the proposed HHO during single DG placement with the aim of
only minimizing power loss were 3.9652MW and bus 18 respectively.

71
The total real power loss after a single DG allocation reduced to 402.988KW. The per-
centage real power loss reduction is 75.26% with respect to base case loss. While that of
PSO reduce the power loss to cause a 53.86%, reduction, which is an indication of HHO
better performance than PSO. The base case and improved voltage magnitudes using sin-
gle DG allocation with PSO and the proposed HHO were plotted against their respective
bus numbers in order to see the improvement in voltage profile after single DG location
and sizing.

Voltage profile of Single DG placement with single objective

1
HHO
0.98 PSO
Base case

0.96
Voltage profile (P.U)

0.94

0.92

0.9

0.88

0.86

0.84
0 5 10 15 20 25 30 35 40
Number of Buses

Figure 5.5: Bus voltage profile comparison with single DG installation

Figure 5.5 shows a comparison of the voltage profiles for the base case, PSO, and the
proposed HHO case after a single DG allocation. The optimal allocation of single DG
with HHO reduces the voltage deviation index from 0.4407 to 0.1397 causing a 68.3%
improvement in the overall voltage profile of the system as compared to the base case
voltage profile. While that of PSO cause a 61.288% improvement, which is an indication
of the positive effect of the DG. HHO improves the overall voltage profile of the system by
18.11% as compared to the PSO voltage profile. These all are when the single DG supplies
active power only (Type one). This means the DG operated at unit power factor. When
the DG supplies both active and reactive power, it operates at a power factor between zero

72
and one. The optimal power factor is the one that gives minimum power loss at optimal
DG size and location. Mostly the optimal power factor of DG lies between 0.8 and 0.9 to
reduce the system loss optimally [56]. Table 5.7 shows the power loss at different power
factors.

Table 5.7: Power loss at different power factor

Pf 0.8 0.81 0.82 0.83 0.84 0.85 0.86 0.87 0.88 0.89 0.9 0.91 0.92 1
loss 331 325 314 317 321.3 327 336 343.5 351 357 364 379 390 402

From table 5.7, we can determine the optimal power factor of the DG to be 0.82 as the
loss start to increases below these power factor.At these power factor, the DG supplies
both active and reactive power of size 3.9652MW and 2.7677KVAr respectively. In order
to see the improvement in voltage profile after the optimal single DG supplies both active
and reactive power (Type two DG), the bus voltage magnitudes along these two types
were plotted against their respective bus numbers.

voltage profile of Type1 and Type2 Single DG placement with single objective
1.02
HHO 1DG Type2
1 HHO 1DG Type1
Base case
0.98
Voltage profile (P.U)

0.96

0.94

0.92

0.9

0.88

0.86

0.84
0 5 10 15 20 25 30 35 40
Number of Buses

Figure 5.6: Bus voltage profile comparison between single DG of Type1 and Type2 with
single objective

Figure 5.6 shows a comparison of the voltage profiles of single DG allocation for Type1,
Type2, and the base case using HHO optimization. The overall voltage deviation index for

73
the base case, Type1 and Type2 with single DG allocation using HHO is 0.4407, 0.1397,
and 0.0851 respectively. These indicate single DG location of Type2 minimizes the voltage
drop by 80.68% and 39.08% compared to the base case and Type1 DG respectively.

Table 5.8: Simulation results of the proposed and PSO algorithms for single DG

Type DG capacity Power loss %PLR %QLR VDI

DGL
Methods of DG MW MVAr KW KVAr
0.4407
No
1629.04 609.513
DG
Type
17 3.4378 487.00 320.449 70.11 47.43 0.1706
PSO one
Type
17 3.4378 2.3995 381.991 218.713 76.55 64.12 0.1059
two
Type
18 3.9652 402.988 314.098 75.26 48.47 0.1397
HHO one
Type
18 3.9652 2.7677 314.984 213.758 80.66 64.85 0.0851
two

When the single DG supplies both active and reactive power, the power loss reduced by
21.83% with HHO compared to the DG supplies only active power. From the table 5.8, the
percentage reduction in real power losses obtained using Type2 DG is much interest and
quite important and Harris hawks optimization (HHO) shows better performance than
particle swarm optimization (PSO) in reducing the total active power loss and voltage
drop in the system.

5.3.3 Results during two DG placement with single objective

Results for this case are tabulated in table 5.9. For this setting, the Optimal placement and
sizing of two units of DG with HHO that minimizes the power loss located at buses 17 and
25, with sizes of the DG 2.4929MW and 2.5MW respectively. Here, total real power losses
reduced to 261.827KW from 1629.04KW, and consequently the reactive power reduced
to 191.178KVAR from 609.513KVAr. This indicates an active power loss reduction of
about 83.92% and reactive power loss reduction of 68.64 % with respect to base case
loss. Similarly, PSO reduces the active and reactive power loss by 76.9% and 60.79%
respectively. The bus voltage magnitude at each bus after two DG allocations with HHO,
PSO, and base case system is drawn in figure 5.7 below to show the improved voltage
profile compared to the base case.

74
 Voltage profile of Two DG placement with single objective
1
HHO
0.98 PSO
Base ase
c

0.96
Voltage profile (P.U)

0.94

0.92

0.9

0.88

0.86

0.84
0 5 10 15 20 25 30 35 40
Number of Buses

Figure 5.7: Bus voltage profile comparison of two DG single objective

Figure 5.7 above shows a comparison of the voltage profiles for the base case, PSO, and
the proposed HHO case after two DG allocation considering only minimizing the power
loss. Here, total VDI reduced to 0.0937 and 0.1526 with HHO and PSO respectively.
These indicate that optimal allocation of two DG with PSO caused a 65.37% improvement
in the overall voltage profile of the system as compared to the voltage profile of the base
case while the proposed HHO cause a 78.74% and 38.59% improvement in voltage profile
compared to the base case and PSO. Which is an indication of the positive effect of the
DG. In order to see the improvement in voltage profile when the DG supplies both active
and reactive power (Type two DG ) compared to the DG supplies only active power (
Type one DG) the bus voltage magnitudes along these two types were plotted against
their respective bus numbers.

75
 voltage profile of Type1 and Type2 Two DG placement with single objective
1.02
HHO 2DG Type2
1 HHO 2DG Type1
Base ase
c

0.98
Voltage profile (P.U)

0.96

0.94

0.92

0.9

0.88

0.86

0.84
0 5 10 15 20 25 30 35 40
Number of Buses

Figure 5.8: Bus voltage profile comparison between two DG of Type one and Type two
with single objective

Figure 5.8 shows a comparison of the voltage profiles of two DG allocation for Type1,
Type2, and the base case using HHO optimization. The overall voltage profile index for
single objective two DG allocation of Type2 using HHO is 0.0263. These indicate that
Type2 DG minimizes the voltage drop by 94% and 71.93% compared to the base case and
Type one DG respectively.

76
Table 5.9: Simulation results of the proposed and PSO algorithms for Two DG Single
objective

Type DG capacity Power loss %PLR %QLR VPI

DGL
Methods of DG MW MVAr KW KVAr
0.4407
No
1629.04 609.513
DG
14 2.2437
Type
376.02 238.972 76.9 60.79 0.1526
PSO one
28 2.1895
14 2.2437 1.566
Type
98.2359 44.833 83.96 86.93 0.0189
two
28 2.1895 1.5283
17 2.4927
Type
261.827 191.178 83.92 68.64 0.0937
HHO one
28 2.500
17 2.4927 1.7398
Type
113.406 76.8894 93.04 89.92 0.0263
two
28 2.500 1.7450

When the optimal two DG supplies both active and reactive power, the power loss
reduced by 93.04% and 56.68% with HHO compared to the system without DG, and
the DG supplies only active power respectively. The proposed HHO reduced the power
loss by 30.37% compared to PSO during two DG locations in the system. This indicates
that Harris hawks optimization (HHO) shows better performance than particle swarm
optimization (PSO) in reducing the total active power loss and voltage drop in the system
with two DG system similar to a single DG location.

5.3.4 Results during Two DG placement with multi objective

For this scenario, objectives PL ,QL ,CDG , and VDI are considered for minimization by
optimal placement of two DGs. Therefore, the system will have minimum active and
reactive power loss with lesser DG cost and voltage drop simultaneously after penetrations.
The optimum result in these case is tabulated in table 5.10. As mentioned in chapter three,
the allocation of the various weights in a given multi-objective function varies according
to the engineer’s concern. In this thesis work, more emphasis is given to real power
loss reduction since this results in a considerable decrease in total cost of operation and
this is not to mean that the other three factors are not important. Thus taking this
into consideration a study of the effect of the weights on the fitness was done so as to
determine the best weights combination to adopt in coming up with the multi-objective
function. During this study, the values of the weights were assumed positive and restricted

77
as follows:

Table 5.10: Multi objective Weight factors

K1 K2 K3 K4 Fitness
0.6 0.2 0.1 0.1 0.1443
0.6 0.1 0.2 0.1 0.1371
0.6 0.1 0.1 0.2 0.1515
0.7 0.1 0.1 0.1 0.1521

From the table 5.10 above, the combination of weights chosen is the one that gave the
minimum best fitness. Thus, the weights chosen were; K1=0.6, K2=0.1, K3=0.2, K4=0.1
and the multi-objective function given by equation becomes:

M OF = 0.6 ∗ P LRI + 0.1 ∗ P LRI + 0.2 ∗ V P II + 0.1 ∗ ACRI (5.1)

During the MOF case, the minimum power loss obtained is 290.097 KW that is slightly
greater than the single-objective case. This is due to the way of searching the optimal
location and size not only to minimize the active power loss but also reactive power loss,
voltage drop, and annual cost of operation. In order to see the performance of HHO
compared to PSO and the improvement in voltage profile after two DG locations and
sizing the base case and bus voltage magnitudes with two DG in multi-objective were
plotted against their respective bus numbers.

78
 Voltage profile of Two DG placement with multi objective
1
HHO
PSO
0.98 Base ase
c

0.96
Voltage profile (P.U)

0.94

0.92

0.9

0.88

0.86

0.84
0 5 10 15 20 25 30 35 40
Number of Buses

Figure 5.9: Bus voltage profile comparison with Two DG multi objective

Figure 5.9 shows the voltage profiles with the proposed HHO and PSO after two DG
allocation with multi-objective compared to the base case. The optimal allocation of two
DG with HHO reduce the voltage drop index to 0.0664 from 0.4407 causing an 84.93%
improvement in the overall voltage profile of the system as compared to the voltage profile
of the base case and 29.13%, improvement compare to the optimal allocation of trwo DG
with HHO with single objective. This is an indication of the positive effect of considering
voltage profile improvement in the objective function.

79
 voltage profile of Type1 and Type2 Two DG placement with Multi objective
1.04
HHO 2DG Type2
1.02 HHO 2DG Type1
Base ase
c
1

0.98
Voltage profile (P.U)

0.96

0.94

0.92

0.9

0.88

0.86

0.84
0 5 10 15 20 25 30 35 40
Number of Buses

Figure 5.10: Bus voltage profile comparison between Two DG of Type one and Type two
with Multi objective

The overall voltage profile with HHO during two DG Type one multi-objective is
0.0664 while during two DG Type two multi-objective case the overall voltage profile is
0.0169, which indicates an improvement of 74.54 %. Two DG multi-objective of Type two
also reduce the power loss to 106.39 KW which is 63.3% reduction compared to two DG
Type one multi objective.

80
Table 5.11: Simulation results of the proposed and PSO algorithms for Two DG multi
objective

Type DG capacity Power loss %PLR %QLR VDI

DGL
Methods of DG MW MVAr KW KVAr
0.4407
No
1629.04 609.513
DG
33 2.1416
Type
385.143 251.899 76.35 58.67 0.085
PSO one
39 2.2960
33 2.1416 1.6008
Type
188.147 152.802 88.45 74.91 0.0233
two
39 2.2960 1.6665
33 2.4997
Type
290.097 254.925 82.19 58.18 0.0664
HHO one
38 2.4896
33 2.4997 1.7447
Type
106.39 161.97 93.46 73.56 0.0169
two
38 2.4896 1.7377

5.4 Cost analysis

This paper focuses on minimizing the economic operation of the distribution network
due to distributed generation. Network simulation studies have been carried out on dis-
tribution systems to which single and two DGs have been connected. It is true that
transmission network losses, maintenance, and operational cost have reduced with in-
creased penetration of DGs to the system while distribution network losses are dependent
on the amount of DG capacities connected to the distribution network. In general, there is
a reduction of network losses, power outage duration, maintenance and operational costs
when the overall network is considered thereby giving financial benefit from DGs added to
the system. However, in this cost analysis, the economic loss due to transmission power
loss, transmission and distribution maintenance, and operational cost are not considered
due to there is no relevant data. Considering these the cost saved due to minimizing the
loss for the whole case above is described in table 5.12 bellow. The revised current tariff
of Ethiopian Electric Utility is used to calculate the power loss cost of the system [57].
The average of the general tariff is calculated from the revised tariff to be 1.75 Birr/kwh.
The American Wind Energy Association (AWEA) explained the cost of power injected by
DGs is taken as $ 1200 per KW (42000 Birr per KW) that includes capital investment for
DG units with installation, operation, and maintenance cost and the average wind DG

81
life is taken as 25 years.

Table 5.12: The economic evaluation of the optimal distributed generation

Total
Total
DG DG annual
PL annual
annual
Case Methods Size Investment
(KW) economic
economic
(KW) cost
loss (Birr)
saving (Birr)
Base
1629.04 - - 24,973,183 -
case
Single PSO 381.991 3437.8 5,775,504 11,630,031 13,343,152
DG HHO 314.981 3965.2 6,661,536 11,475,156 13,498,027
2141.6
Two PSO 188.14 7,455,168 10,339,354 14,633,829
2296.0
DG 2465.8
HHO 106.39 8,322,720 9,953,678 15,019,504
2489.0

From table 5.12 the annual cost saving using integration of two DG is better than
integrating a single DG in the feeder. The average annual energy not supplied due to
outage and interruption of the line is 850.8MWh, which corresponds to an annual economic
loss of 1,488,900 Birr. The DG supply the feeder during power outage and minimize such
economic loss due to power interruption.

5.5 Simulation results of Generating optimal DG size

from DFIG based Wind Energy System

The simulation is conducted based on a doubly fed induction generator based wind en-
ergy system in MATLAB/SIMULINK to generate the optimal DG size and to evaluate
the performance of the vector-controlled system. The simulation does not consider the
start-up of the DFIG system.
The DFIG required to generate generate nearly 2.5MW that is the optimal size of DG
determined using the proposed optimization techniques. The synchronous speed of the
DFIG is 1500rpm. The generator rotates nearly at 900rpm when the available wind speed
is between 5m/s and 7m/s and it operates beyond the synchronous speed when the avail-
able wind speed is between 12m/s and 15m/s.

82
All the results figure out here are after the system reaches a steady-state in super-
synchronous modes of operation. We know that DFIG supplies the grid through both
the stator and rotor. Therefore, 2MW DFIG means it deliver this power to the grid
through the stator only. The total rated power (pmec ) generated by the wind turbine is
the sum of the power through the stator and rotor.
The optimal DG size determined is nearly 2.5MW; therefore, the rating of the wind turbine
selected is 2.5MW. Then the power delivered through the rotor (pr ) is 0.5MW assuming
loss in the system zero.The simulation is performed at a unit power factor supplying zero
reactive power (Q = 0) to the grid. The simulation results in the super-synchronous
mode obtained by applying a wind speed of 12.5m/s which fed a speed of 1800rpm into
the speed control loop of the rotor doubly fed induction generator is presented in the
figure below:.

Figure 5.11: Simulated rotor speed

Figure 5.11 above demonstrates the ability of the speed control of the DFIG to main-
tain the specified speed at the reference signal

Figure 5.12: Simulated rotor id currents

The rotor id current in figure (5.12) maintained at zero value all the time since the
simulation is performed at unit power factor and the corrosponding rotor vd voltage will

83
be zero.

Figure 5.13: Simulated rotor iq currents

Figure 5.14: Simulated rotor three phase currents

Figure 5.14 above is the three phase current of the rotor and its frequecy is the slip
time the frequency of the stator current that is 50 Herth in this analysis. Its magnuted
is nearly equal to the rotor iq current as the rotor id current is zero.

Figure 5.15: Simulated rotor vd voltage

84
 Figure 5.16: Simulated stator three phase voltage

Figure 5.17: Simulated stator three phase current

Figure 5.17 and 5.16 are the three phase current and voltage respectively supplied to
the stator of the DFIG from the grid.

Figure 5.18: Simulated power delivered to the grid

figure 5.18 shows the total power which is nearly the optimal size of DG deliverd to
the grid through the doubly fed induction generator.

85
Chapter 6

Conclusion and Future Work

6.1 Introduction

This chapter presents a summary of this thesis work. Mainly the chapter highlighted the
general conclusion, recommendation, and suggestions for further works.

6.2 Conclusion

This research work showed the formulation of main equations of electrical power systems
for power flow analysis and the formulation and implementation of the Harris hawks opti-
mization algorithm. The algorithm helps in reducing system power losses and improving
the voltage profile by optimizing the location and size of multi-type DGs. The power
loss, voltage deviation, and economic cost reduction index used effectively in reducing the
search space for the algorithm.
The results obtained were compared with those obtained using the particle swarm
optimization algorithm. For the selected feeder, the particle swarm optimization found
the optimal DG location to be bus 33 and 39 with a size of 2.1416 MW and 2.296 MW
respectively. The power loss reduced to 188.147 KW considerind the DG supplies both
active and reactive power to the grid. This is 88.45% total loss reduction in the feeder. The
proposed HHO found the optimal DG location to be 33 and 39 with a size of 2.4997 MW
and 2.4896 MW respectively.The power loss reduced to 106.39 KW which is a 93.46% total
loss reduction in the feeder. Similarly, the voltage profile improved as well. Furthermore,
the proposed HHO had a higher performance than the PSO method in order to calculate
the optimal size and location of DG.

86
This thesis also studied firstly energy sources in the area and deals with modeling a
generation system based on wind renewable energy. Finally, the paper explained the
generation of the determined DG size from wind distributed generation.

6.3 Beneficiaries of this work

Different parties benefit from this research work both directly and indirectly. Distribution
utility and customers are the direct beneficiaries of this research work. Some of the direct
benefits from this work includes:

• This thesis paper will help the distribution companies in reducing both real and
reactive power losses in their networks that enable them to avoid some of the penal-
ties and compensations they incur and hence result in an improvement in annual
profit.

• This paper guide the distribution company to incorporate renewable distributed

generation into the distrbution network.

• Both the utility and the customer save theire materials from damege caused by
voltage deviation.

• The electric tarif will be reduced due to the reduction of much investment cost in
long transmission and distribution sytem.

6.4 Recommendations

Considering the cause that makes difficulties in the analysis and the result of the studies
carried out in this thesis, the following points are recommended:

• We strongly recommend that EEP must consider the option of using distributed gen-
eration such that the utility company can save cost, increase transmission efficiency
and reduce socio-economic problem.

• It was difficult to identify each load point and the corresponding data of that load
point. This made the analysis in this thesis to be difficult. So, we recommend to
EEU and power distribution substation to distinguish each load point and their
corresponding failure data.

87
6.5 Future Works

The following items provide a shortlist of potential for future work:

• The hypred optimization with the proposed HHO algorithm can be developed for
optimal DG allocation using other optimization methods such as Particle Swarm
Optimization technique, Greedy Search Algorithm, and Genetic algorithm.

• Further analysis can also be done by using the developed model to simultaneously
allocate more than two DG in radial distribution networks with minimum economic
cost.

88
References

[1] Thomas Ackermann, Goran Andersson, Lennart Soder “Distributed Generation: a

definition,” Electric Power Systems Research, Vol.57, pp.195-204, 2001.
[2] O. A. Zongo and A. Oonsivilai, “Optimal placement of distributed generator for power
loss minimization and voltage stability improvement”, Energy Procedia, Vol.138, pp.134-
139, 2017.
[3] U. Sultana, A. B. Khairuddin, M. M. Aman, A. S. Mokhtar, and N. Zareen, “A review
of optimum DG placement based on minimization of power losses and voltage stability
enhancement of distribution system,” Renewable and Sustainable Energy Reviews, Vol.
63, pp. 363-378, 2016.
[4] K. Balamurugana , Dipti Srinivasana, Thomas Reindlb, “Impact of Distributed Gen-
eration on Power Distribution Systems,” Energy Procedia, Vol. 25, pp. 93 – 100, 2012.
[5] H. Doagou-Mojarrad, G. B. Gharehpetian, H. Rastegar, and J. Olamaei, “Optimal
placement and sizing of DG (distributed generation) units in distribution networks by
novel hybrid evolutionary algorithm,” Energy, Vol.54, pp. 129-138, 2013.
[6] Sarfaraz Nawaz, Sonali Singh, Supriya Awasthi, “ Power Loss Minimization in Radial
Distribution System using Multiple DG Units and Network Reconfiguration,” European
Journal of Scientific Research, Vol. 148, no 4, pp. 474-483, March, 2018.
[7] Krischonme Bhumkittipich and Weerachai Phuangpornpitak,” Optimal Placement and
Sizing of Distributed Generation for Power Loss Reduction using Particle Swarm Opti-
mization,” Energy Procedia, Vol. 34, pp. 307 – 317, 2013.
[8] Binayak Bhandar et al, “Optimization of Hybrid Renewable Energy Power Systems”
International Journal of Precision Engineering and Manufacturing-Green Technology Vol.
2, no. 1, pp. 99-112, Jan 2015.
[9] Ruhaizad Ishak et al, “Optimal DG Placement and Sizing for Voltage Stability Im-
provement Using Backtracking Search Algorithm,” Conference on Artificial Intelligence,

89
Energy and Manufacturing Engineering, June 9-10, 2014.
[10] Zhang Xiaozhao, Duan Jiandong, Han Yuhui, “Locating and Sizing of Dispersed
Wind Generation in Active Distributed System,” IEEE PES Asia-Pacific Power and En-
ergy Conference, pp. 1872 –76, 2016.
[11] Atthapol Ngaopitakkul et al, “A Reliability Impact and Assessment of Distributed
Generation Integration to Distribution System ,” Energy and Power Engineering, Vol. 5,
pp. 1043-1047, 2013.
[12] M. Esmaeilzadeh, I. Ahmadi and N. Ramezani, “ Optimal Distributed Generation
Planning in Radial Distribution Networks Considering Protection Coordination Limits,”
Iranian Journal of Electrical & Electronic Engineering, Vol. 14, no. 2, Jun 2018.
[13] Nafar M, “Optimal placement of DGs in distribution systems considering voltage
stability and short circuit level improvement using GA,” J Basic Appl Sci Res, Vol. 2,
no. 1, pp. 368–75, 2012.
[14] Nafar M, “PSO-based optimal placement of DGs in distribution systems considering
voltage stability and short circuit level improvement” J Basic Applied Science Res, Vol.
2, no. 1, pp. 703–9, 2012.
[15] Moradi MH, Abedini M, “A combination of genetic algorithm and particle swarm
optimization for optimal DG location and sizing in distribution systems,” Electric Power
Energy System, Vol. 34, pp. 66–74, 2012.
[16] A. F. A. Kadir et al, “An improved gravitational search algorithm for optimal place-
ment and sizing of renewable distributed generation units in a distribution system for
power quality enhancement,” Journal of Renewable and Sustainable Energy, Vol. 6, no.
3, 2014.
[17] Biswas S, Goswami SK, Chatterjee A. Optimum distributed generation placement
with voltage sag effect minimization. Energy Convers Manage, Vol.53, no.1, pp.163–74,
2012.
[18] A. M, Abdrabou Ahmed “Voltage sag Mitigation by Optimizing the location of Dis-
tributed Generation using Genetic Algorithm for three distributed generation types,”
Master of Engineering (Electrical), May 2014.
[19] G. Pepermans et ai, “Distributed generation: definition, benefits and issues,” Energy
Policy, Vol. 33, pp.787–798, 2005.
[20] Ali M. ELTAMALY et al, “ Impact of Distributed Generation (DG) on the Distribu-

90
tion System Network,” International Journal of Engineering Science, April 2019.
[21] Dr. Mounir Bouzguenda, Abdullah Samadi, S. RajaMohamed, ”Optimal Placement
of Distributed Generation in Electric Distribution Networks,” Ieee International Confer-
ence on Intelligent Techniques in Control, Optimization and Signal Processing, 2017.
[22] Onwunta Emea Kalu Onwunta, ”Modelling and Simulation of the Impacts of Dis-
tributed Generation Integration into the Smart Grid”, Ph.D Dessertation, Cape Peninsula
University of Technology, Nov 2014.
[23] Diagram of geothermal power plant, MULTIMEDIA, Retrieved from http://www.eniscuola.net/en/m
of-a-geothermal-power-plant.
[24] Fuel cell, WIKIPEDIA, Retrieved from, https://en.wikipedia.org/wiki/Fuel cell, Nov
2020.
[25] Biomass For Electricity Generation, WBDG, Retrieved from, https://www.wbdg.org/resources/biom
electricity-generation, 09-15-2016. [26] L. Priyadarshanee, ”modeling and control of hy-
brid ac/dc micro grid,” Rourkela, 2012.
[27] Solar (PV) cell module, Samlex solar, retrieved from, https://www.samlexsolar.com/learning-
center/solar-cell-module-array.aspx.
[28] Flow-Diagram-of-a-Wind-Turbine-System, retrieved from, https://www.researchgate.net/figure/Flo
Diagram-of-a-Wind-Turbine-System.
[29] Gaber El Saady et al, “Analysis of Wind Turbine Driven Permanent Magnet Syn-
chronous Generator under Different Loading Conditions,” International Journal of Tech-
nology in Education and Science, Vol.4, no.14, pp.97-111, Dec, 2013.
[30] A. PRAVEEN VARMA and K. BALA CHAKRI, “STUDY OF GRID CONNECTED
INDUCTION GENERATOR FOR WIND POWER APPLICATIONS”, Department of
Electrical Engineering National Institute of Technology, Rourkela, 2012.
[31] Vineet Dahiya and Leena G, “Comparative Study of Doubly Fed Induction Genera-
tor and Permanent Magnet Synchronous Generator in Wind Energy Conversion System,”
International Journal of Electrical Engineering and Technology, Vol.10, no.3, pp. 73-79,
2019.
[32] A. M. Eltamaly, Y. S Mohamed, A. H. M. el-sayed, and A. N. A. elghaffar, “impact
of distributed generation (dg) on the distribution system network,” International Journal
of Engineering, April, 2019.
[33] Jhansi S, “Effects of Distributed Generation on Electrical Power Network and Pro-

91
tection,” Journal of Electrical & Electronic Systems, Vol. 8, no 1, pp.1-3, 2019.
[34] A Hasibuan, et al, “Effect of distributed generation installation on power loss us-
ing genetic algorithm method,” IOP Conf. Series: Materials Science and Engineering,
Vol.308, pp.1-11, 2018.
[35] Angel Fernández Sarabia, “Impact of distributed generation on distribution system,”
MSC dissertation, Faculty of Engineering Science and Medicine, Aalborg University, June
2011.
[36] P. P. Barker and R. W. de Mello,, “Determining the impact of distributed generation
on power systems. Part l radial distribution systems,” Power Engineering Society Sum-
mer Meeting, 2000.
[37] Kumaraswamy, B. Venkataprasanth and S.Tarakalyan, “Role of Distributed Gener-
ation in Voltage Stability Enhancement,” International Journal of Current Engineering
and Technology, Vol.4, No.1, Feb 2014.
[38] Jinsong Liu , Junyang Zhang , Da Zhang, “Effect of Distributed Generation on Power
Supply Reliability of Distribution Network,” International Conference on Grid and Dis-
tributed Computing, pp. 32-35, 2015.
[39] Sanaullah Ahmad, Sana Sardar,” Impact of Distributed Generation on the Reliability
of Local Distribution System”, (IJACSA) International Journal of Advanced Computer
Science and Applications, Vol. 8, no. 6, July 2017.
[40] A.F. Abdul Kadir, A. Mohamed , H.Shareef,“Harmonic Impact of Different Dis-
tributed Generation Units on Low Voltage Distribution System,” IEEE International
Electric Machines & Drives Conference (IEMDC), pp.1201-1206, 2011.
[41] N. Rugthaicharoencheep, S. Auchariyamet , “Technical and Economic Impacts of Dis-
tributed Generation on Distribution System,” International Journal of Electrical, Com-
puter, Energetic, Electronic and Communication Engineering, Vol.6, no.4, 2012.
[42] M. Al-Muhaini and G. T. Heydt, “Evaluating future power distribution system reli-
ability including distributed generation,” IEEE Transactions on Power Delivery, vol. 28,
no. 4, pp. 2264–2272, 2013.
[43] Pathomthat Chiradeja, “An Approach to Quantify the Technical Benefits of Dis-
tributed Generation,” IEEE TRANSACTIONS ON ENERGY CONVERSION, Vol. 19,
no. 4, Dec 2004.
[44] G. Pepermans, J. Driesen, D. Haeseldonckx, W. D’haeseleer and R. Belmans, “DIS-

92
TRIBUTED GENERATION: DEFINITION, BENEFITS AND ISSUES,” Article in En-
ergy Policy, April 2005.
[45] Ozge Akkus and Emre Demis, “Comparison of some classical and meta–heuristic
optimization techniques in the estimation of the logit model parameters,” International
Journal of Advanced Research, Vol.4, No.1, Oct 2016.
[46] John McCall, “Genetic algorithms for modelling and optimisation,” Journal of Com-
putational and Applied Mathematics, Vol.184, pp.205-222 2005.
[47] T. Monish kumar1, Dr V. Ganesh,“ Optimal allocation of DG unit for the radial
distribution system using genetic algorithm,” INTERNATIONAL JOURNAL OF IN-
NOVATIVE RESEARCH IN ELECTRICAL, ELECTRONICS, INSTRUMENTATION
AND CONTROL ENGINEERING, Vol. 3, no 1, pp. 87-90, Jan, 2015.
[48] Ahmad Hassanat et al,“Choosing Mutation and Crossover Ratios for Genetic Algo-
rithms—A Review with a New Dynamic Approach,” Jornal of information, Vol.10, No.12,
Nov 2019.
[49] D.B. Prakash, C. Lakshminarayana, “Multiple DG Placements in Distribution Sys-
tem for Power Loss Reduction Using PSO Algorithm,” Procedia Technology, Vol. 25,
pp.785 – 792, 2016.
[50] Heidari et al, “Harris hawks optimization: Algorithm and applications,” Article in
Future Generation Computer Systems, Vol. 97, pp. 849-872, March 2019.
[51] Omessaad Elbeji, Mouna Ben Hamed, Lassaad Sbita, “PMSG Wind Energy Conver-
sion System: Modeling and Control,” International Journal of Modern Nonlinear Theory
and Application, Vol.3, pp. 88-97, July 2014.
[52] Rashmi.R, Ramanujan.V, Purushotham.M, “Permanent Magnet Synchronous Gener-
ator Configuration in Wind Turbines,” International Journal of Scientific & Engineering
Research, Vol. 6, no. 2, pp.23-31, February-2015.
[53] S. Soued1, H. S. Ramadan, M. Becherif, “Effect of Doubly Fed Induction Genera-
tor on Transient Stability Analysis under Fault Conditions,” Energy Procedia, Vol. 162,
pp.315–324, 2019.
[54] A. Dekhane et al, “DFIG Modeling and Control in a Wind Energy Conversion Sys-
tem,” First International Conference on Renewable Energies and Vehicular Technology,
pp.287-292, March 2012.
[55] Johan MORREN, Sjoerd W.H. de HAAN, “Maximum penetration level of distributed

93
generation without violating voltage limits,” Conference Paper, July 2008.
[56] Bindeshwar Singh, Bindu Jee Gyanis, “Impact assessment of DG in distribution sys-
tems from minimization of total real power loss view point by using optimal power flow
algorithms, Energy Report, Vol. 4, pp. 407-417, 2018.
[57] Impacts and drivers of policies for electricity access Micro- and macro-economic analy-
sis of Ethiopia’s tariff reform,retrieved from, https://energyeconomicgrowth.org/node/236.

94
Appendix A

Appendix 1

A.1 Line and Load data of Sheno feeder

[Table A.1: Line and Load data of sheno feeder]

Real Reactive
Line From Resistance Reactance
To bus power power
no bus (ohm) (ohm)
(MW) (MVAr)

1. 1 2 0.3471 0.1936 0 0

2. 2 3 0.1446 0.2099 0.6758 0.5671

3. 3 4 0.1446 0.0695 0.2635 0.1224

4. 4 5 0.1157 0.0606 0.0450 0.0104

5. 5 6 0.0867 0.1035 0.0765 0.0311

6. 6 7 0.2314 0.1242 0.1700 0.1091

7. 7 8 0.2314 0.2390 0.1649 0.0982

8. 8 9 0.0578 0.0177 0.0607 0.0250

9. 9 10 0.1157 0.0311 0.0413 0.0240

10. 3 11 0.1157 0.1276 0.0675 0.0370

11. 11 12 0.0867 0.0644 0.0807 0.0384

12. 12 13 0.2314 0.1813 0.2000 0.1854

13. 13 14 0.0867 0.0783 0.03110 0.2590

14. 14 15 0.2603 0.1788 0.0842 0.0465

15. 15 16 0.1735 0.1478 0.0400 0.0127

16. 16 17 0.0289 0.0626 0.2669 0.1822

95
17. 17 18 0.0289 0.1641 0.6332 0.4218

18. 18 19 0.2578 0.1641 0.2603 0.1600

19. 19 20 0.1578 0.0714 0.1649 0.1186

20. 20 21 0.1778 0.0774 0.0453 0.0243

21. 21 22 0.2867 0.1774 0.5447 0.4132

22. 3 23 0.0867 0.0399 0.2644 0.1022

23. 23 24 0.3471 0.8277 0.2669 0.1246

24. 24 25 0.2314 0.1774 0.5525 0.2888

25. 25 26 0.0489 0.0227 0.1900 0.1340

26. 26 27 0.1735 0.2217 0.1755 0.0933

27. 27 28 0.1735 0.1299 0.0503 0.0132

28. 28 29 0.1446 0.1695 0.2100 0.1260

29. 29 30 0.1446 0.0340 0.6758 0.2494

30. 30 31 0.0867 0.1094 0.0713 0.0455

31. 31 32 0.0867 0.04035 0.0503 0.0241

32. 32 33 0.3471 0.2114 0.2669 0.1046

33. 23 34 0.1735 0.0576 0.2530 0.0838

34. 34 35 0.2314 0.1242 0.1159 0.0346

35. 35 36 0.1246 0.1091 0.0842 0.0476

36. 22 37 0.0578 0.0325 0.0913 0.0300

37. 37 38 0.1735 0.9607 0.1950 0.1580

38. 38 39 0.1446 0.0961 0.0605 0.0476

39. 39 40 0.2144 0.1537 0.1700 0.0976

40. 0.0882 0.0310

96
A.2 Wind turbine and DFIG Rating Parameter for

the test system

Table A.2: Parameter of Nordex N80/2500 wind turbine

Parameter Value(unit)
Rotor diameter 80 m
Hub height 80 m
Cut in speed 3 m/s
Cut out speed 25 m/s
Nominal wind speed 12.5 m/s
Speed range 9-16 rpm
Optimum TSR 7.2
Maximum power coefficient 0.44

Table A.3: DFIG Rating Parameter for the test system

Parameter Value (unit)

Nominal stator active power 2 MW
Nominal torque 12.732 K.NM
Stator voltage 690 V
Nominal speed 1500 rpm
Speed range 900-200 rpm
Pole pair 2
Lm 2.5 *10ˆ-3H
Lr 87 *10ˆ-6
Ls 87 *10ˆ-6
Rr 0.026 ohm
Rs 0.029 ohm

A.3 Proposed Harrish Hawks Optimization Code for

optimal size and location of DG

% H a r r i s ’ s hawk o p t i m i z a t i o n code f o r DG placement and s i z i n g .

% T: maximum i t e r a t i o n s , N: p o p u l a t o i n s i z e
clear all
clc
close all
warning ( ’ o f f ’ )
N=30;
T=100;

97
UB=40; LB=2;
l b =1; ub=4;
dim=2;
d i s p ( ’HHO i s now t a c k l i n g your problem ’ )
tic
% i n i t i a l i z e th e l o c a t i o n and Energy o f t he r a b b i t
R a b b i t L o c a t i o n=z e r o s ( 1 , dim ) ;
Rabbit Energy=i n f ;
d i =1;
%I n i t i a l i z e t he l o c a t i o n s o f H a r r i s ’ hawks
f o r run=2
f o r i =1: d i
i f s i z e (UB,1)==1
x ( : , 1 ) = c e i l ( rand (N, d i ) . ∗ ( UB−LB)+LB ) ;
x ( : , 2 ) = round ( rand (N, d i ) . ∗ ( ub−l b )+ lb , 4 ) ;
end
i f s i z e (UB,1) >1
f o r i =1: d i
high=UB( i ) ; low=LB( i ) ;
HIGH=ub ( i ) ;LOW=l b ( i ) ;
x ( 1 , i )= c e i l ( rand ( 1 ,N) . ∗ ( high−low)+low ) ;
x ( 2 , i )=round ( rand ( 1 ,N) . ∗ ( HIGH−LOW)+LOW, 4 ) ;
end
end
end
[ f ]= G e t F u n c t i o n s d e t a i l s ( x ,N ) ;
B=f ;
CNVG=z e r o s ( 1 ,T ) ;
t =0; % Loop c o u n t e r
w h i l e t<T
f o r i =1: s i z e ( x , 1 )
% Check b o u n d r i e s

98
 FU=x ( i ,1) >UB; FL=x ( i ,1) <LB ; x ( i , 1 ) = ( x ( i , 1 ) . ∗ ( ˜ ( FU+FL)))+UB. ∗FU+LB. ∗ FL ;
FU=x ( i ,2) > ub ; FL=x ( i ,2) < l b ; x ( i , 2 ) = ( x ( i , 2 ) . ∗ ( ˜ ( FU+FL)))+ ub . ∗FU+l b . ∗ FL ;
% f i t n e s s of locations
f i t n e s s=B( i , : ) ;
% Update th e l o c a t i o n o f Rabbit
i f f i t n e s s <Rabbit Energy
Rabbit Energy=f i t n e s s ;
R a b b i t L o c a t i o n=x ( i , : ) ;
end
end
E1=2∗(1−( t /T ) ) ; % f a c t o r t o show th e d e c r e a i n g e n e r g y o f r a b b i t
% Update th e l o c a t i o n o f H a r r i s ’ hawks

E0=2∗rand () −1; %−1<E0<1

Escaping Energy=E1 ∗ ( E0 ) ; % escaping energy of rabbit
% Escapi ng Energy = 0 . 1 ;
i f abs ( Es caping Energy)>=1
%% E x p l o r a t i o n :
% H a r r i s ’ hawks perch randomly based on 2 s t r a t e g y :
f o r i =1:N
q=rand ( ) ;
rand Hawk index = f l o o r (N∗ rand ( ) + 1 ) ;
X rand = x ( rand Hawk index , : ) ;
i f q <0.5
% perch based on o t h e r f a m i l y members
X( i , : ) = X rand−rand ( ) ∗ abs ( X rand −2∗rand ( ) ∗ x ( i , : ) ) ;
e l s e i f q>=0.5
% p erch on a random t a l l t r e e ( random s i t e i n s i d e group ’ s home range )
X( i , : ) = ( R a b b i t L o c a t i o n ( 1 , : ) − mean ( x))− rand ( ) ∗ ( ( ub−l b ) ∗ rand+l b ) ;
end
end
[ f ]= G e t F u n c t i o n s d e t a i l s (X,N) ;

99
 H=f ;
f o r i =1:N
i f H( i , : ) <B( i , : ) % improved move?
x ( i , : ) =X( i , : ) ;
end
end

e l s e i f abs ( Escaping Energy )<1

%% E x p l o i t a t i o n :
r=rand ( ) ; % p r o b a b l i t y o f each even
i f r >=0.5 && abs ( Escaping Energy ) <0.5 % Hard b e s i e g e
f o r i =1:N
X( i , : ) = ( R a b b i t L o c a t i o n )− Escaping Energy ∗ abs ( R a b b i t L o c a t i o n −x ( i , : ) ) ;
end
end
[ f ]= G e t F u n c t i o n s d e t a i l s (X,N) ;
H=f ;
f o r i =1:N
i f H( i , : ) <B( i , : ) % improved move?
x ( i , : ) =X( i , : ) ;

end

end

i f r >=0.5 && abs ( Escaping Energy )>=0.5 % Soft besiege

f o r i =1:N
Jump strength =2∗(1−rand ( ) ) ; % random jump s t r e n g t h o f t h e r a b b i t
X( i , : ) = ( R a b b i t L o c a t i o n −x ( i , : ) ) − Escaping Energy ∗ abs . . .
( Jump strength ∗ R a b b i t L o c a t i o n −x ( i , : ) ) ;
end
[ f ]= G e t F u n c t i o n s d e t a i l s (X,N) ;

100
 H=f ;
f o r i =1:N
i f H( i , : ) <B( i , : ) % improved move?
x ( i , : ) =X( i , : ) ;

end

end
[ f ]= G e t F u n c t i o n s d e t a i l s ( x ,N ) ;
s=f ;
f o r i =1: s i z e ( x , 1 )
% f i t n e s s of locations
f i t n e s s=s ( i , : ) ;
% Update th e l o c a t i o n o f Rabbit
i f f i t n e s s <Rabbit Energy
Rabbit Energy=f i t n e s s ;
R a b b i t L o c a t i o n=x ( i , : ) ;
end
end

end
%% phase 2 : p e r f o r m i n g team r a p i d d i v e s ( l e a p f r o g movements )
i f r <0.5 && abs ( Escaping Energy ) >=0.5 ,

Jump strength =2∗(1−rand ( ) ) ;

f o r i=N
X1=R a b b i t L o c a t i o n −Escaping Energy ∗ abs . . .
( Jump strength ∗ R a b b i t L o c a t i o n −x ( i , : ) ) ;
[ f ]= G e t F u n c t i o n s d e t a i l s (X1 ,N) ;
H=f ;
i f H<B( i , : ) % improved move?
x ( i , : ) = X1 ;

101
 else
[ o]= l e v y ( dim ) ;
X2=R a b b i t L o c a t i o n −Escaping Energy ∗ abs . . .
( Jump strength ∗ R a b b i t L o c a t i o n −x ( i , : ) ) + rand ( 1 , dim ) . ∗ o ;
[ f ]= G e t F u n c t i o n s d e t a i l s (X2 ,N) ;
R=f ;
i f R<B( i , : ) % improved move?
x ( i , : ) = X2 ;
end
end
end
end
i f r <0.5 && abs ( Escaping Energy ) <0.5 ,
Jump strength =2∗(1−rand ( ) ) ;
f o r i=N

X1=R a b b i t L o c a t i o n −Escaping Energy ∗ abs . . .

( Jump strength ∗ R a b b i t L o c a t i o n −mean ( x ) ) ;
[ f ]= G e t F u n c t i o n s d e t a i l s (X1 ,N) ;
H=f ;
i f H<B( i , : ) % improved move?
x ( i , : ) = X1 ;
else
[ o]= l e v y ( dim ) ;
X2=R a b b i t L o c a t i o n −Escaping Energy ∗ abs . . .
( Jump strength ∗ R a b b i t L o c a t i o n −mean ( x))+ rand ( 1 , dim ) . ∗ o ;
i f R<B( i , : ) % improved move?
x ( i , : ) = X2 ;
end
end
end
end

102
 [ f ]= G e t F u n c t i o n s d e t a i l s ( x ,N ) ;
s=f ;
f o r i =1:N
% f i t n e s s of locations
f i t n e s s=s ( i , : ) ;
% Update th e l o c a t i o n o f Rabbit
i f f i t n e s s <Rabbit Energy
Rabbit Energy=f i t n e s s ;
R a b b i t L o c a t i o n=x ( i , : ) ;
end
end
end
t=t +1;
CNVG( t )= Rabbit Energy ;
end
r=Rabbit Energy ;
R=R a b b i t L o c a t i o n ;
i f Rabbit Energy<r
r=Rabbit Energy ;
R=R a b b i t L o c a t i o n ;
end
end
toc
figure ,
h o l d on
s e m i l o g y (CNVG, ’ Color ’ , ’ b ’ , ’ LineWidth ’ , 4 ) ;
t i t l e ( ’ Convergence curve ’ )
xlabel ( ’ Iteration ’ ) ;
y l a b e l ( ’ Best f i t n e s s o b t a i n e d so f a r ’ ) ;
axis tight
grid off
box on

103
l e g e n d ( ’HHO’ )
d i s p l a y ( [ ’ The b e s t l o c a t i o n o f HHO i s : ’ , num2str (R ) ] ) ;
d i s p l a y ( [ ’ The b e s t f i t n e s s o f HHO i s : ’ , num2str ( r ) ] ) ;

104