Ain Shams University
Faculty of Engineering
Electrical power & Machines Department
Adaptive Load Shedding for Stand-Alone Power
Systems
M.Sc. Thesis
By
Eng. Taghreed Ibrahim Abdel Hady Shaarawy
Submitted in partial fulfillment of the requirements for the M.Sc. degree
in Electrical Engineering
Supervised By
Prof. Dr. Hossam El Din Abd Allah Talaat
Dr. Rania Abdel Wahid Abdel Halim
Electrical Power and Machines Department,
Faculty of Engineering, Ain Shams University
Cairo, 2017
Examiners Committee
For the Thesis
Adaptive Load Shedding for Stand-Alone Power Systems
By
Eng. Taghreed Ibrahim Abdel hady Shaarawy
A thesis Submitted to the Faculty of Engineering Ain Shams University
in partial fulfillment of the requirements for the M.Sc. Degree in
Electrical Power and Machine Engineering
Name, title and affiliation Signature
1. Prof. Dr. Hanafy Mahmoud Ismail
Electrical Power and Machines Eng. Department,
Faculty of Engineering, Ain Shams University.
2. Prof. Dr. Rania Metwally Awad Elsharkawy
Head of Electrical and control Eng. Department,
Arab Academy for Science and Technology
and Maritime Transport, Cairo.
3. Prof. Dr. Hossam El Din Abdallah Talaat
Electrical Power and Machines Eng. Department,
Faculty of Engineering, Ain Shams University
Supervisors Committee
For the Thesis
Adaptive Load Shedding for Stand-Alone Power Systems
By
Eng. Taghreed Ibrahim Abdel hady Shaarawy
A thesis Submitted to the Faculty of Engineering Ain Shams University
in partial fulfillment of the requirements for the M.Sc. Degree in
Electrical Power and Machine Engineering
Approved by:
Name, title and affiliation Signature
1. Prof. Dr. Hossam El Din Abdallah Talaat
Electrical Power and Machines Department,
Faculty of Engineering, Ain Shams University
2. Dr. Rania Abdel Wahid Abdel Halim
Electrical Power and Machines Department,
Faculty of Engineering, Ain Shams University
Statement
ii
STATEMENT
This Thesis is submitted to Ain Shams University in partial fulfillment
of the requirements of M.Sc. degree in Electrical Engineering.
The included work in this thesis has been carried out by the author at the
department of electrical power and machines, Ain Shams University. No
part of this thesis has been submitted for a degree or a qualification at
any other university or institution.
Name: Taghreed Ibrahim Abdel Hady Shaarawy
Signature:……………………………………….
Date: / /
ACKNOWLEDGEMENT
iii
ACKNOWLEDGEMENT
I would like to express my sincere gratitude to Professor. Dr. Hossam
El Din Abd Allah Talaat for giving me the opportunity to work under
his supervision, for the continuous support of my study and research, for
his patience, motivation, immense knowledge, and teaching me how to
think. His guidance helped me in all the time of research and writing of
this thesis.
I wish to express my deep thanks to Dr. Rania Abdel Wahid Abdel
Halim for her supervision, supporting, helpful advice, continuous
guidance, theoretical advice and assistances during the stages of
preparation of this work.
Finally, I am grateful for my parents and my husband, who helped me
throughout all of this work. Thank you for supporting me in every way.
Taghreed Ibrahim Abdel Hady Shaarawy
Cairo, 2017
Abstract
iv
ABSTRACT
Load frequency control is commonly used to maintain power
system frequency very close to its nominal value. A small
deviation from the rated frequency may have harmful impact on
the components of the system. Under-Frequency Load Shedding
(UFLS) is to be used when the regular frequency control loops fail
to restore the frequency to its rated value. In such conditions, the
frequency continues to fall until adequate loads are disconnected
from the network. Usually, renewable energy generation in
microgrids represents a considerable percentage of the total
generation. Therefore, application of UFLS in microgrids is crucial
since they are exposed to rapid deficiency in generation due to
weather conditions.
There are two main categories of UFLS: conventional and adaptive
load shedding. Conventional UFLS has a fixed number of stages
and fixed time delays while the adaptive UFLS applies shedding,
in most cases, in one stage in the shortest possible delay.
This work introduces two adaptive UFLS schemes for an islanded
microgrid both are based on artificial intelligence (AI). These
schemes rely on estimating the value of load to be shed in one
stage. As a base of comparison, a heuristic methodology for load
shedding has been used. This heuristic load shedding is applied by
creating a look up table using the loads history and willingness to
pay.
The under-frequency load shedding schemes developed in this
work has utilized two well-known AI techniques, namely fuzzy
logic and artificial neural network. Both schemes have two inputs
and one output. The inputs are the maximum deviation in
frequency and the absolute value of the rate of change of frequency
Abstract
v
at the instant of islanding. The output of the proposed scheme is
the amount of load to be shed to maintain the frequency of the
power system within the acceptable limits. Training patterns
obtained from the simulated system are used to tune both the fuzzy
rule base and the neural network so as to give accurate results.
Fifteen simulation case studies have been used for this purpose.
Testing stage have been implemented using additional 15
simulation case studies.
The obtained results prove the effectiveness of the developed
scheme in estimating the amount of the load to be shed to restore
the system frequency back near to its nominal value. The main
advantage of the proposed scheme is its capability in performing
the load shedding in one stage with minimum load disconnection
which enhances the system reliability and improves the transient
response of the frequency deviation of the system.
Keywords: Distributed generation, Renewable sources, Speed
governor, Underfrequency load shedding, Stand-alone power
system, Fuzzy logic, Artificial Neural Network.
Table of Contents
vi
TABLE OF CONTENTS
ABSTRACT iv-v
TABLE OF CONTENTS vi-viii
LIST OF TABLES ix
LIST OF FIGURES x-xii
LIST OF ABBREVIATIONS xiii
1. INTRODUCTION
1.1. General …………………………………………………………..
1.2. Objectives of the thesis ………………………………………….
1.3. Thesis Outlines ………………………………………………….
1-3
1-2
2
2-3
2. THEORETICAL BACKGROUND AND LITERATURE
REVIEW
2.1. General …………………………………………………………..
2.2. Distributed Generation (DG) …………………………………….
2.2.1. DG Applications ……………………………………………...
2.2.2. Types of Distributed Generators ……………………………...
2.2.2.1. Traditional Combustion Generators: Micro-turbine
(MT) …………………………………………………...
2.2.2.2. Non-Traditional Combustion Generators ……………...
A. Electromechanical Devices: Fuel cell
B. Storage Devices …………………………………….
C. Renewable Devices …………………………………
2.2.3. Types of Distributed Generators ……………………………...
2.2.4. Advantages of distributed generation ………………………...
2.2.5. Disadvantages of distributed generations …………………….
2.3. Review of Methods for Islanding Detection Techniques ………..
2.4. Effect of Increasing Renewable Sources in Power System ……..
2.5. Frequency Regulation in Power Systems ………………………..
2.5.1. The need for frequency regulation ………………………..
A. Impact of frequency variation on steam turbines ……...
B. Impact of under frequency variations on Generators ….
C. Impact on other network components …………………
2.5.2. Role of load frequency control ……………………………
4-27
4
4-9
4-5
5-9
5
6
6
6
6
6-7
8
8-9
9-10
10-13
14-18
14-15
14-15
15
15
16
Table of Contents
vii
2.5.3. Evaluation of frequency performance in typical power
networks …………………………………………………..
2.6. Underfrequency Load Shedding Schemes ………………………
2.6.1. Industrial Underfrequency Load shedding schemes………..
2.6.2. Review of Underfrequency load shedding techniques……..
16-18
19-27
19-21
21-27
3. SIMULATION OF HEURISTIC UFLS SCHEME
3.1. General …………………………………………………………
3.2. Problem Description …………………………………………...
3.3. Role of speed governor and its limits ………………………….
3.3.1. Role of speed governor …………………………………….
3.3.2. Limits of speed governor control …………………………..
3.4. System Understudy Single Line Diagram ……………………..
3.5. Simulink Model ………………………………………………..
3.6. Application of Heuristic Scheme ………………………………
3.6.1. Procedure for creating the look up table ………………….
3.6.2. Algorithm of Underfrequency load shedding …………….
3.7. Simulation Results ……………………………………………..
3.7.1. Case study (I) ……………………………………………..
3.7.2. Case study (II) ……………………………………………
28-49
28
28-31
32-34
32
32-34
34
34-36
37-40
37-38
38-40
40-49
40-43
43-49
4. ARTIFICIAL INTELLIGENT-BASED UFLS SCHEMES
4.1. General …………………………………………………………..
4.2. Fuzzy Logic Fundamentals ……………………………………...
4.3. Proposed Methodology ………………………………………….
4.4. Generation of case studies ………………………………………
4.5. Design steps …………………………………………………….
4.5.1. Range of variables ……………………………………..
4.5.2. Membership functions …………………………………
4.5.3. Rule-Base ……………………………………………...
4.6. Considerations of On-Line Application of FL-UFLS Scheme ….
4.7. Results of Fuzzy UFLS scheme …………………………………
4.7.1. Case study (7) ………………………………………...
4.7.2. Case study (10) ……………………………………….
4.7.3. Summary of results …………………………………..
4.8. ANN Fundamentals ……………………………………………..
4.9. Methodology of ANN- UFLS …………………………………..
50-84
50
50
50-51
51-52
52-55
53
53-54
54-55
55-56
56-60
57
58
59-60
60-61
61-62
Table of Contents
viii
4.10. Design of ANN-UFLS Scheme ……………………………
4.11. Results of ANN-UFLS Scheme ……………………………
4.12. Comparison between Developed UFLS Schemes …………
4.12.1. Comparison between Heuristic and FL-UFLS ……..
4.12.1.1. Effect of the location of the load to be shed …….
4.12.1.2. Comparison between Heuristic and FL-UFLS ….
A. Performance of Heuristic and FL-UFLS for
PDrated= 40 MW …………………………………
B. Performance of Heuristic and FL-UFLS for
PDrated= 43 MW ………………………………….
4.12.2. Comparison between fuzzy and ANN techniques …..
4.12.2.1. Simulation results ……………………………...
4.12.2.2. Summary of results …………………………….
62-64
64-73
74-84
74-79
74-75
75-79
76-77
77-79
79-84
80-83
83-84
5. CONCLUSIONS 85-86
REFERENCES 87-89
LIST OF PUBLICATION 90
LIST OF TABLES
ix
LIST OF TABLES
Table 2.1 Types of DGs and their capacity 7
Table 2.2 WECC Load shedding stages 20
Table 2.3 Additional Load shedding stages 21
Table 2.4 Seven load shedding steps in the Egyptian power
system 22
Table 3. 1 Example of the look up table 38
Table 3. 2 Look up table for case study I 42
Table 3. 3 Look up table for case study II scenario 1 45
Table 3. 4 Look up table for case study II scenario 2 48-49
Table 4.1 Summary of simulation of 15 case studies 52
Table 4.2 Developed Output Fuzzy Rule Base 55
Table 4.3 Summary of results 59-60
Table 4.4 Testing case studies 64
Table 4.5 Summary of Testing Results 65
Table 4.6 Performance Evaluation of the 4 UFLS Schemes 70
Table 4.7 Summary of Testing Results for ANN31, ANN32,
ANN33 and ANN34 72
Table 4.8 Performance Evaluation of the 4 UFLS Schemes 73
Table 4.9 Effect of location on maximum overshoot and steady
state frequency for fuzzy scheme 74
Table 4.10 Effect of location on maximum overshoot and steady
state frequency for heuristic scheme 75
Table 4.11 Comparison between heuristic and FL-UFLS by
shedding loads from same location 75
Table 4.12 Estimated load shed (p.u.) for the best ANN scheme
and the FL scheme 80
Table 4.13 Estimated Pshed for FL-UFLS and ANN3-UFLS
versus target shedding load 83
Table 4.14 Performance Evaluation of the 2 UFLS Schemes 84
LIST OF FIGURES
x
LIST OF FIGURES
Figure 2.1 DG Technologies 5
Figure 2.2 Islanding detection techniques 9
Figure 2.3 Global Installed wind power capacity (MW) 11
Figure 2.4 Global Wind Power Cumulative Installed Capacity 12
Figure 2.5 IEEE C37-106 Worst case frequency withstand 15
Figure 2.6 Average frequency values in Continental Europe, June
2003 and June 2010 for Swiss grid
17
Figure 2.7 Frequency quality behaviour in Continental Europe
during the last ten years in Swiss grid 18
Figure 2.8 NERC Areas 20
Figure 3.1 Micro grid connected to main grid 28
Figure 3.2 Frequency dip within the European power system 29
Figure 3.3 Frequency variation with different power deficiency 31
Figure 3.4 Frequency deviation with speed governor control 33
Figure 3.5 Frequency deviation for islanding condition 34
Figure 3.6 Single line diagram 35
Figure 3.7 Simulink model 36
Figure 3.8 Algorithm for heuristic underfrequency load shedding 39
Figure 3.9 loads configurations for case study I 41
Figure 3.10 Frequency variation with time for case study (I) 43
Figure 3.11 loads configurations for case study II scenario 1 44
Figure 3.12 Frequency variation with time for case study (II)
scenario 1 46
Figure 3.13 loads configurations for case study II scenario 2 47
Figure 3.14 Frequency variation with time for case study (II)
scenario 49
LIST OF FIGURES
xi
Figure 4.1 Proposed FL-UFLS Scheme 51
Figure 4.2 Membership function of input 1 53
Figure 4.3 Membership function of input 2 54
Figure 4.4 Membership function of the output 54
Figure 4.5 algorithm of FL-UFLS scheme 56
Figure 4.6 Frequency variation with time for case study (7) 57
Figure 4.7 Frequency variation with time for case study (10) 58
Figure 4.8 General Architecture of the ANN scheme 61
Figure 4.9 Architecture of the ANN scheme 61
Figure 4.10 Methodology of ANN-based UFLS scheme 62
Figure 4.11 Internal Architecture of the ANN 63
Figure 4.12 ANN1-UFLS Frequency Variation for PDrated = 38.5
MW 66
Figure 4.13 ANN2-UFLS Frequency Variation for PDrated = 38.5
MW 66
Figure 4.14 ANN3-UFLS Frequency Variation for PDrated = 38.5
MW 67
Figure 4.15 ANN4-UFLS Frequency Variation for PDrated = 38.5
MW 67
Figure 4.16 ANN1-UFLS Frequency Variation for PDrated = 43.5
MW 68
Figure 4.17 ANN2-UFLS Frequency Variation for PDrated = 43.5
MW 68
Figure 4.18 ANN3-UFLS Frequency Variation for PDrated = 43.5
MW 69
Figure 4.19 ANN4-UFLS Frequency Variation for PDrated = 43.5
MW 69
Figure 4.20 Bar chart for ANN1, ANN2, ANN3 and ANN4
comparing mean error and rms 71
LIST OF FIGURES
xii
Figure 4.21 Bar chart for ANN31, ANN32, ANN33 and ANN34
comparing mean error and rms 73
Figure 4.22 FL-UFLS Frequency Variation for PDrated = 40 MW 76
Figure 4.23 Heuristic Frequency Variation for PDrated = 40 MW 77
Figure 4.24 FL-UFLS Frequency Variation for PDrated = 43 MW 78
Figure 4.25 Heuristic scheme – Scenario 1, Frequency Variation
for PDrated = 43MW 78
Figure 4.26 Heuristic scheme – Scenario 2, Frequency Variation
for PDrated = 43MW 79
Figure 4.27 FL-UFLS Scheme Frequency Variation for
PDrated=38.5 MW 81
Figure 4.28 ANN34-UFLS Scheme Frequency Variation for
PDrated=38.5 MW 81
Figure 4.29 FL-UFLS Scheme Frequency Variation for PDrated =
47.5 MW 82
Figure 4.30 ANN34-UFLS Scheme Frequency Variation for
PDrated = 47.5 MW 82
Figure 4.31 Bar chart for ANN34 and fuzzy comparing mean error
and rms 84
LIST OF ABBREVIATIONS
xiii
LIST OF ABBREVIATIONS
DG Distributed Generators
WECC Western Electricity Coordinated Council
AI Artificial Intelligent
UFLS Under-Frequency Load Shedding
ANN Artificial Neural Network
NERC North American Electric Reliability Council
PLL Phase Locked Loop
FLLSC Fuzzy Logic Load Shedding Controller
LSCM Load Shedding Controller Module
ROCOF Rate Of Change Of Frequency
ROCOFLi Rate Of Change Of Frequency for load i
FL Fuzzy Logic