13 5. 5.10. 3. Sensitivity analysis for different redundancy of Inverter and .... The General system configuration of remote area power supplies 51. Fig. 3.8(a) ...... Clint (Jito) Coleman has reported in /23/ (1985) that, as a result of the ... installation of 25 stand-alone systems ranging from nine watts to over five kilowatts. (Twenty ...
OPTIMAL SIZING AND RELIABLE CONTROLLER OF HYBRID PHOTOVOLTAIC POWER SYSTEMS
by Eng. Mohamed Bayoumy Abdel-Kader Zahran Assistant Researcher - Photovoltaic Cells Dept. Electronics Research Institute A Thesis submitted to the Faculty of Engineering at Cairo University in Partial Fulfillment of the Requirements of the Degree of DOCTOR OF PHILOSOPHY
Ill
ELECTRICAL POWER AND MACHINES DEPARTMENT
FACULTY OF ENGINEERING, CAIRO UNIVERSITY GIZA, EGYPT December, 1998
OPTIMAL SIZING AND RELIABLE CONTROLLER OF HYBRID PHOTOVOLTAIC POWER SYSTEMS
by Eng. Mohamed Bayoumy Abdel-Kader Zahran Assistant Researcher - Photovoltaic Cells Dept. Electronics Research Institute A Thesis submitted to the Faculty of Engineering at Cairo University in Partial Fulfillment of the Requirements of the Degree of DOCTOR OF PHILOSOPHY
in ELECTRICAL POWER AND MACHINES DEPARTMENT
FACULTY OF ENGINEERING, CAIRO UNIVERSITY GIZA, EGYPT December, 1998
OPTIMAL SIZING AND RELIABLE CONTROLLER OF HYBRID PHOTOVOLTAIC POWER SYSTEMS
by Eng. Mohamed Bayoumy Abdei-Kader Zahran . Assistant Researcher - Photovoltaic Cells Dept. Electronics Research Institute A Thesis submitted to the Faculty of Engineering at Cairo University in Partial Fulfillment of the Requirements of the Degree of DOCTOR OF PHILOSOPHY in ELECTRICAL POWER AND MACHINES DEPARTMENT
Under the Supervision of Prof.Dr. Said Helmi El-Hefnawi
Prof.Dr. Adel Abdel-RaoofHanafy Faculty of Engineering, Cairo University
Electronics Research Institute Photovoltaic Cells dept.
Assoc. Prof. Osama Altmed Maltgoub Faculty ofEngineering, Cairo University.
Dr. Maltmoud lbraltim Kamel Benha Higher Institute of Tech., Benha, Egypt
FACULTY OF ENGINEERING, CAIRO UNIVERSITY GIZA, EGYPT December, 1998
11
OPTIMAL SIZING AND RELIABLE CONTROLLER OF HYBRID PHOTOVOLTAIC POWER SYSTEMS
by
Eng. Mohamed Bayoumy Abdel-Kader Zahran Assistant Researcher - Photovoltaic Cells Dept. Electronics Research Institute A Thesis submitted to the Faculty of Engineering at Cairo University in Partial Fulfillment of the Requirements of the Degree of DOCTOR OF PHILOSOPHY
.
Ill
ELECTRICAL POWER AND MACHINES DEPARTMENT
Approved by the Examining Committee
Prof.Dr. Adel Abdel-Raoof Hanafy
Thesis Main Advisor
Prof.Dr. Hassan Taher Doraa
Member
Prof.Dr. Mohamed Mostafa Al-Saied
Member
FACULTY OF ENGINEERING, CAIRO UNIVERSITY GIZA, EGYPT December, 1998 111
Acknowledgment
The author would like to express his deep thanks to Prof Dr. Adel AbdelRaoof Hanafy for his kind supervision, useful discussion and comments, general helping during the preparation of this thesis. Also I would like to thank Prof Dr. Said El-he:fuawi for his supervision and encouragement. I would like to express my sincere thanks to Assoc. Prof Osama Mahgoub and Dr. Mahmoud Kamel for their help during the implementation phase and final production of this work. I pray to God to forgive Prof Dr. Ahmed El-Tobshy asking God to keep him in the high degree in the Paradise (Elferdaws El-Aala). I also express my deep thanks to all my colleagues in ERI, specially the PV Cells Dept. group. I would like also to express my appreciation for DAAD (German Academic Exchange Service) for giving me the chance to do a part of the simulation and implementation of this work in Germany (1995-1996). So, I would like to express my deep
thanks for Prof Dr.-Ing. Franz N. Fett (Head of the Institute for Energy
Technology) and Prof Dr.-Ing. J. Mario Pacas (Head ofPower Electronics Institute), Siegen University and their working group. I also like to thank the Physics Dept., Oldenburg University.
IV
Contents Contents
v
List of Tables
xii
List of Figures
xiii
List of Symbols
xix
Abstract
xxi
Chapter 1 Introduction to the PV Remote Area Systems 1. 1. Introduction
1
1.2. Conventional PV Hybrid Systems
5
Chapter 2 Survey on Photovoltaic Remote Area Systems 8
2.1. Introduction 2.1.1. Photovoltaics
9
2.1.2. Solar Cells
11
2.1.3. Working Principles of Solar Cells
12
2.1.4. Solar Cell Modules
12
2.2 Availability of Solar Energy
14
2.3. Types ofPV Systems and their applications
14
2.4. Survey of PV Pilot Plants
18
2.5. PV Systems in Egypt
21
2.6. Economical Analysis ofPV Systems
23
2. 7. PV Systems Design
25
2.8. PV Systems Management and Control
26
2.9. PV Systems Reliability
27
2.10. Economics and Reliability ofPV Hybrid Systems
28
Chapter 3 Photovoltaic Remote Area Systems Sizing and Systems Performance Study 30
3. 1. Introduction
v
3 .2. Photovoltaic system sizing-present methods
34
3 .2.1. Energy balance on a daily basis method
34
3.2.2. Loss ofload probability system sizing methods
36
3.2.3. Minimization oftotal-life-cycle-cost (TLCC) method
40
3.3. Proposed sizing method
48
3.4. Sizing ofPV remote area systems
49
3 .4 .1. Case study
49
3.4.2. General system configuration
51
3 .4. 3. Inverter size
51
3.4.4. Meteorological data-base (MOB) module
53
3.5. PV battery stand-alone (PVB) system Sizing
55
3.6. PV battery Diesel generator (PVBD) hybrid system sizing
60
3.7. Diesel generator battery (DB) hybrid system sizing
62
3.8. Economical analysis of the PV remote area systems
65
3.9. General PV remote area systems sizing flow chart and sizing procedure 67 3.10. Results of the PV remote area systems sizing
70
3.1 0.1. Results ofPVB stand-alone system sizing
70
3.10.2. Results ofPVBD hybrid system sizing
71
3.10.3. Results ofDB hybrid system sizing
72
3 .11. PVBD hybrid system configuration
74
3.12. Conclusion
76
Chapter 4 Sizing Sensitivity Analysis of Photovoltaic Remote Area Systems 4.1.
Introduction
78
4.2. Sensitivity analysis on sizing of the PVB system
78
4 .2.1. Effect of the load peak variation on optimum system components 78
SIZe
4.2.2. Effect of the load peak variation on system LOLP and kWh cost
80
4.2.3. Effect ofthe dynamic load duration variation on optimum components size
82
4.2.4. Effect of the dynamic load duration variation on system LOLP
Vl
and kWh cost
83
4 .2. 5. Effect of the site of application distance on the optimum generated kWh cost
85
4.3. Sensitivity analysis on sizing ofPVBD hybrid system
85
4.3.1. Effect ofthe load peak variation on optimum system components 85
SiZe
4.3.2. Effect ofthe load peak variation on system LOLP and kWh cost
87
4.3.3. Effect ofthe dynamic load duration variation on optimum 88
components size 4.3.4. Effect of the dynamic load duration variation on system LOLP and kWh cost
90
4. 3. 5. Effect of the site of application distance on the optimum generated 91
kWh cost 4.4. Sensitivity analysis on sizing of the DB hybrid System
92
4. 4 .1. Effect of the load peak variation on optimum system components 92
SiZe
4.4.2. Effect of the load peak variation on system LOLP and kWh cost
94
4.4.3. Effect of the dynamic load duration variation of optimum components size
95
4.4.4. Effect of the dynamic load duration variation on system LOLP and kWh cost
97
4. 4. 5. Effect of the site of application distance on the optimum generated 98
kWh cost 4. 5. Comparison between different remote area systems performance
99
4.5.1. Effect of dynamic load duration variation on the optimum kWh cost
99
4.5.2. Effect ofload peak changes on the optimum kWh cost
101
4.5.3. Effect of dynamic load duration changes on system LOLP and kWh cost
102
4.5.4. Effect ofload peak changes on system LOLP and kWh cost 4.6. General Conclusions
104 106
Vll
Chapter 5 Remote Area Systems Operation Reliability 5 .1. Introduction
108
5 .2. Reliability Definition and Relationships
109
5 .2.1. Reliability of systems with n-items in series
110
5 .2.2. Reliability of systems with m-items in parallel
111
5 .2.3. Reliability of series-parallel systems
112
5. 3. Reliability Design and Optimization
109
5.3.1. Under budgetary constraint.
113
5.3.2. Under safety constraints.
113
5.4. Application ofFuzzy Logic to Systems Reliability
114
5. 5. Reliability consideration for remote area systems
116
5.6. Remote Area Systems Fault-Tree Analysis
120
5.6.1. PVBD Hybrid System Fault Tree Analysis
121
5.6.2. PVB system Fault Tree Analysis
123
5.6.3. DB hybrid system Fault Tree Analysis
125
5. 7. Results of fault tree analysis of remote area systems
126
5.8. Sensitivity analysis ofthe factor cr on the remote area systems reliability 127 5.9. Reliability Improvement ofRemote Area Systems
129
5. 9 .1. System with nonrepairable components
130
5. 9 .2. Systems with repairable components
130
5.9.3. Remote area system availability
131
5.1 0. Sensitivity analysis of the critical components MTBF on the system reliability
13 2
5.10. 1. Reliability Sensitivity analysis for critical system components 133
MTBF 5.10.2. Reliability improvement including the effectiveness of preventive maintenance.
13 5
5.10. 3. Sensitivity analysis for different redundancy of Inverter and System Controller
136
5. 10. 4. Effectiveness of preventive maintenance and redundancy on the system reliability.
141
5.11. Conclusions
143
Vlll
Chapter 6 Review of Present Hybrid PV System Management 6.1. Introduction
146
6.2. Goals of control strategies in PV systems
147
6.3. Present Control Strategies
147
6.3.1. PVB with Shunt Regulator Configuration "A"
148
6.3.2. PVB with Trickle Charging Technique Configuration "B"
149
6.3.3. PVB with de-de Converter Configuration "C"
150
6.3.4. PVB with Array Strings on/off Control Configuration "D"
151
6.3.5. PVB with Partial Shunt Switching Capability Configuration "E"
152
6.3.6. PVB with Diesel generator Backup Hybrid System Config. "F"
153
6.4. PV Pilot Plants
154
6.4.1. NASA (USA) PV Pilot Plant
154
6.4.2. Vulcano (Italy) PV Pilot Plant
158
6.4.3. Pellworm (Germany) PV Pilot Plant
159
6.4.4. Nice (France) PV Pilot Plant
162
6.5. Conclusion
163
Chapter 7 Development of Reliable Controller and its Implementation. 7 .1. Introduction
166
7.2. Fault-Tolerant System Controllers
168
7.3. Fault -Tolerant Systems Controllers Description
169
7. 3 .1. Dual redundancy architecture
17 0
7.3.2. TMR redundancy architecture
171
7.3.3. N- Modular redundancy
172
7.3.4. Quad-modular architecture
174
7.4. Proposed reliable controller subsystem
175
7. 4 .1. Proposed Reliable Controller Subsystem Functional Diagram
17 6
7.4.2. Operation Principles of the Proposed Controller Subsystem
177
7.4.2.1. Reconfigurable Switch (Voter) Operation
177
7.4.2.2. Watch-dog hardware checker circuit operation
178
7.5. The proposed system controller block diagram
182
7.6. Conclusion
183
IX
Chapter 8 Laboratory Testing of the Developed Reliable Controller 8.1. Introduction
185
8.2. The System I-V and P-V Curves Characteristics
187
8.3. PV array Voltage regulation
188
8.3.1. PV Array Voltage SelfRegulation
188
8.3.2. PV Array Voltage Forced Regulation
189
8.3 .2.1. Parallel Regulator
189
8.3.2.2. Series Regulator
190
8.3.2.3. PV Array Strings Switching Regulator
191
8.4. PVBD hybrid system simulation
192
8. 4 .1. The PV Array Model
192
8.4.2. Storage Battery Bank Model
194
8.4.3. Load current and battery parameters determinations
194
8.4.4. System performance simulation based on INSEL.500 usage
195
8.4.5. System Performance Simulation Based on MatLab Program Usage 197 8.5. System control signals
200
8. 5.1. System controller input signals
201
8. 5. 1.1. PV array current sensor
201
8.5.1.2. Load current sensor
201
8.5.1.3. Battery SOC signal sensor
201
8. 5. 1.4. Diesel generator status signal
202 203
8.5.2. Controller output signals 8.5.2.1. PV array strings control circuit
203
8.5.2.2. Diesel start/stop control
204
8.6. Sampling rate of system controller
205
8.6.1. PV Current Sampling
205
8.6.2. Battery SOC sampling
206
8.6.3. Load current Sampling
206
8. 7. Proposed Control Strategies
207
8.8. Battery SOC estimation
208
8. 9. Experiments test results
212
8. 10. Conclusion
214
X
Chapter 9 Fuzzy Logic Based PVBD Hybrid System Control 9. 1. Choosing of the appropriate control algorithm
216
9.2. Specification of the control system
216
9.3. Proposed system control block diagram
217
9.4. What is Fuzzy Logic?
218
9. 5. Fuzzy Logic Controller of the proposed PVBD hybrid System
220
9. 5.1. System Operation Algorithm
220
9.5.2. Fuzzy logic controller algorithm
222
9.6. Simulation of the FLC ofFOPD process using fuzzy toolbox under MatLab
226
9.7. Implementation ofMicrocontroller- Based Fuzzy Logic Controller
230
9.7.1. Signal Conditioning Circuits (Adaptation ofunipolarto be bipolar one & vice-versa)
230
9.7.2. Normalization ofthe conversion ofthe sensed signals
231
9.7.3. Calculation ofthe two fuzzy logic controller inputs
231
9.7.4. Implementation ofthe fuzzy logic variables using microcontroller subsystem
231
9.7.5. Rules Identifications
232
9.7.6. Retrieve the amount offuzzy output L1I
232
9. 7. 7. Calculating the crisp value of L1I
233
9.8. Problems faced us during the implementation
234
9. 9. Conclusion
234
Chapter 10 Conclusion and Future work 10 .I. Conclusion
23 6
10.2. Future work
240
References
241
Publications
247
Appendix A, The Developed Reliable Controller Wiring Diagram
248
Appendix B, The System Modes of Operation and Control Strategy Flow Chart
251
Xl
List of Tables Table 1.1,
PV system versus Diesel generator.
1
Table 2.1,
Solar cell efficiency values of different solar cell types.
11
Table 2.2,
PV pilot plants.
20
Table 3.1,
The case study load parameters.
49
Table 3.2,
The optimum tilt angle.
53
Table 5.1,
The linguistic function of item reliability.
116
Table 5.2,
The system components MTBF.
118
Table 5.3,
The PVBD system components MTBF
119
Table 5.4,
The system components MTTR.
132
Table 6.1,
The relation between the system voltage and the controller action
156
Table 6.2,
Comparison between the present pilot plants and proposed system 164
Table 8.1,
The load current values.
194
Table 8.2,
The battery parameters at different SOC.
195
Table 8.3,
Relation between SOC and specific gravity.
210
Table 8.4,
Relation between SOC and pH probe output.
211
Table 9.1,
The fuzzy rules of the PVBD system
223
Table 9.2,
The values of the membership grades
232
Table 9.3,
The Rule Base for the Fuzzy Logic Controller Output consequence ~I
233
Xll
List of Figures Fig. 1.1,
Comparison between the difference remote area systems.
3
Fig. 2.1,
Silicon solar cell construction
12
Fig. 2.2,
Solar cell characteristics (I - V) curves.
13
Fig. 2.3,
Three basic types ofPV system configuration.
16
Fig. 2.4,
History and Forecast for PV module Shipments, 1975-2000.
16
Fig. 2.5,
depicts the different applications classifications for PV system
17
Fig. 2.6,
Trends in Flat-plate PV Module Cost.
23
Fig. 3.1,
Power flow diagram for sizing ofPV stand-alone system
35
Fig. 3.2,
Flow Chart of the Simulation Program
39
Fig. 3.3,
The Sizing procedure of /21/ (1988).
40
Fig. 3.4,
Battery storage requirements for 1% LOLP /20/ (1987).
42
Fig. 3.5,
Life-Cycle-Cost calculation algorithm /20/ (1987).
47
Fig. 3.6,
The case study daily Load Profile
50
Fig. 3.7,
The General system configuration of remote area power supplies
51
Fig. 3.8(a),
One Inverter with a separate softstarter
52
Fig. 3.8(b),
Two separate Inverters for the two load types
53
Fig. 3.9,
The block diagram of the Meteorological data-base module
54
Fig. 3.10,
The insolation curve for site of application, (El-Giza City)
55
Fig. 3.6',
The Load daily Profile
61
Fig. 3.11,
Diesel-generator battery hybrid system configuration
62
Fig. 3.12,
The sizing routine flow chart
69
Fig. 3.13(a),
The PVB system sizing results
70
Fig. 3.13(b),
The kwh cost in L.E ofPVB system
71
Fig. 3.14(a),
The PVBD hybrid system sizing results
71
Fig. 3.14(b),
The kwh cost in L.E ofPVBD hybrid system
72
Fig. 3.15(a),
The DB hybrid system sizing results
73
Fig. 3.15(b),
The kWh cost in L.E ofDiesel battery system.
73
Fig. 3.16,
The PVBD hybrid system configuration with its control signals
76
Fig. 4.1,
PVB, load peak sensitivity on components size and kWh cost
80
Fig. 4.2,
PVB, load peak sensitivity on system LOLP and kWh cost
81
Xlll
Fig. 4.3,
PVB, load duration sensitivity on components size and kWh cost
83
Fig. 4.4,
PVB, load duration sensitivity on system LOLP and kWh cost
84
Fig. 4.5,
The kWh cost as a function of distance in PVB system
85
Fig. 4.6,
PVBD, load peak sensitivity on components size and kWh cost
87
Fig. 4.7,
PVBD, Load peak sensitivity on system LOLP and kWh cost
88
Fig. 4.8,
PVBD, load duration sensitivity on components size and kWh cost 89
Fig. 4.9,
PVBD, load duration sensitivity on system LOLP and kWh cost
91
Fig. 4.10,
The kWh cost as a function of distance in PVBD system
92
Fig. 4.11,
DB, load peak sensitivity on components size and kWh cost
94
Fig. 4.12,
DB, load peak sensitivity on system LOLP and kWh cost
95
Fig. 4.13,
DB, load duration sensitivity on system LOLP and kWh cost
96
Fig. 4.14,
DB, load duration sensitivity on system LOLP and kWh cost
98
Fig. 4.15,
The kWh cost as a function of a distance in km of DB system
98
Fig. 4.16(a),
kWh cost ofPVRAS at 100% Rel. & load duration variations
99
Fig. 4.16(b),
kWh cost ofPVRAS at 95% Rel. & load duration variations
100
Fig. 4 .16( c),
kWh cost ofPVRAS at 90% Rel. & load duration variations
100
Fig. 4.17(a),
kWh cost ofPVRAS at 100% Rel. & load peaks variations
101
Fig. 4.17(b),
kWh cost ofPVRAS at 95% Rel. & load peaks variations
101
Fig. 4.17(c),
kWh cost ofPVRAS at 90% Rel. & load peaks variations
102
Fig. 4.18(a),
kWh cost and LOLP ofPVRAS at 100% Rel. & load duration variations (fixed components size)
Fig. 4.18(b),
kWh cost and LOLP ofPVRAS at 95% Rel. & load duration variations (fixed components size)
Fig. 4.18(c),
Fig. 5.1,
105
kWh cost and LOLP ofPVRAS at 95% Rel. & load peaks variations (fixed components size)
Fig. 4.19(c),
104
kWh cost and LOLP ofPVRAS at 100% Rel. & load peaks variations (fixed components size)
Fig. 4.19(b),
103
kWh cost and LOLP ofPVRAS at 90% Rei. & load duration variations (fixed components size)
Fig. 4.19(a),
103
105
kWh cost & LOLP ofPVRAS at 90% Rel. & load peaks variation (fixed components size)
106
A system made up of two subsystems or parts in series.
110
XIV
111
Fig. 5.2,
A system made up of two redundant paths in parallel
Fig. 5.3,
Series-Parallel system configuration
Fig. 5.4,
The functional block diagram of conventional remote area systems 118
Fig. 5.5,
The fault tree analysis ofPVBD Hybrid System
121
Fig. 5.6,
The fault tree analysis of the PVB system
124
Fig. 5.7,
The fault tree analysis of the DB hybrid system
126
Fig. 5.8,
The remote area systems reliability at a
127
Fig. 5.9,
The remote area systems reliability at different a
129
Fig. 5.10,
The remote area systems availability
13 2
Fig. 5.11,
PVBD system reliability, sensitivity analysis on critical components
112
=
50%
MTBF (10,000-50,000) operating hours Fig. 5.12,
PVB system reliability, sensitivity analysis on critical components MTBF (10,000-50,000) operating hours.
Fig. 5.13,
133
134
DB system reliability, sensitivity analysis on critical components MTBF (10,000-50,000) operating hours.
134
Fig. 5.14,
Effectiveness of preventive maintenance on system reliability.
136
Fig. 5.15,
The functional block diagram with redundant components.
13 7
Fig. 5.16,
The Fault-tree analysis of systems with redundant components
13 8
Fig. 5.17,
PVBD system, results of sensitivity analysis on system with redundant components at different degrees of redundancy
Fig. 5.18,
PVB system, results of sensitivity analysis on system with redundant components at different degrees of redundancy
Fig. 5.19,
13 9
13 9
DB hybrid system, results of sensitivity analysis on system with redundant components at different degrees of redundancy
140
Fig. 5.20,
The systems reliability including double redundancy degree
142
Fig. 5.21,
The systems reliability including quad redundancy degree
143
Fig. 6.1(a),
PVB with Shunt Regulator system block diagram.
148
Fig. 6.1(b),
PVB with Shunt Regulator system control strategy.
148
Fig. 6.1 (c),
PVB with Trickle Charging system block diagram.
149
Fig. 6.1 (d),
PVB with Trickle Charging system control strategy.
149
Fig. 6.1 (e),
PVB with de-de Converter system block diagram.
150
Fig. 6.1(±),
PVB with de-de Converter system control strategy.
150
XV
Fig. 6.1 (g),
PVB with Array Strings on/off Control system block diagram.
151
Fig. 6.1(h),
PVB with Array Strings on/off Control system control strategy
151
Fig. 6.1(i),
PVB with Partial Shunt Switching system block diagram.
152
Fig. 6.1Q),
PVB with Partial Shunt Switching system control strategy.
152
Fig. 6.1(k),
PVB with Diesel generator Backup Hybrid System block diagram
153
Fig. 6.1(1),
PVB with Diesel generator Backup Hybrid System control strategy 153
Fig. 6.2,
Simplified System Block Diagram of the NASA (USA) 155
PV Pilot Plant Fig. 6.3,
Simplified System Block Diagram of the Vulcano (Italy) PV Pilot Plant
Fig. 6.4,
158
Simplified System Block Diagram of the Pellworm (Germany) PV Pilot Plant
Fig. 6.5,
160
Simplified System Block Diagram of the Nice (France) PV Pilot Plant
162
Fig. 7.1,
Dual architecture with redundant input modules .
170
Fig. 7.2,
TMR architecture with redundant input modules.
171
Fig. 7.3(a),
Basic tri-modular redundancy (TMR) configuration.
173
Fig. 7.3(b),
The use of TMR voters to remove single points of failure from a network
173
Fig. 7.4,
Basic quad-modular redundancy configuration.
174
Fig. 7.5,
The Functional Diagram of the Proposed Controller Subsystem
176
Fig. 7.6(a),
Proposed reliable controller subsystem operation flow chart 179
Fig. 7. 6(b ),
Switching Voter subroutine
180
Fig. 7.6(c),
Watch-dog Checker Subroutine
181
Fig. 7.7,
Proposed reliable controller block diagram
183
Fig. 7.8,
A photo of the proposed controller subsystem
184
Fig. 8.1,
The implemented lab testing PVBD hybrid system configuration with its control signals.
185
Fig. 8.2,
The PV array of the system under test.
186
Fig. 8.3,
The Diesel generator unit with its accessories.
186
Fig. 8.4,
The load bank.
187
Fig. 8.5,
The I-V curves of the PV array at different insolation levels
XVl
188
& 25 °C.
Fig. 8.6,
The I-V curves of the PV array at different temperature levels & 1000W/m2 .
188
Fig. 8.7,
Shunt (parallel) regulator.
189
Fig. 8.8,
Series regulator.
190
Fig. 8.9,
The system performance at minimum SOC and 28 °C.
196
Fig. 8.10,
The system performance at medium SOC and 28 °C.
196
Fig. 8.11,
The system performance at maximum SOC and 28 °C.
197
Fig. 8.12,
The system block diagram of the Matlab simulation program.
198
Fig. 8.13,
The system variables at 30% sun & 28 °C
(battery discharging 199
mode). Fig. 8.14,
The system variables at 60% sun & 28 °C
(battery charging 199
mode). Fig. 8.15,
The system performance at actual insolation curve, daily load pattern & 28 °C.
200
Fig. 8.16,
The photo of the two current transducers.
202
Fig. 8.17,
The storage battery with the pH probe.
202
Fig. 8.18,
The generator status circuit block diagram.
203
Fig. 8.19,
The PV arrays driving circuit block diagram.
203
Fig. 8.20,
PV strings control and Diesel control Driving circuits, Filters circuit and AID signal conditioning circuit.
204
Fig. 8.21,
The PV array blocking diodes and the relays control bank.
204
Fig. 8.22,
PV insolation curve (output power) under different skies.
205
Fig. 8.23,
Charge/discharge with different rates curves.
209
Fig. 8.24,
Proposed reliable controller block diagram.
212
Fig. 8.25,
System performance during battery discharge mode Ch1 -load current and Ch2 - the load voltage.
213
Fig. 8.26,
The PV current and battery charging current at different loading
213
Fig. 8.27,
System performance during battery discharge mode Ch1 - PV current and Ch2 - the load voltage.
214
Fig. 9.1,
The classifications of the control systems
216
Fig. 9.2,
Proposed PVBD hybrid system control block diagram
217
XVll
Fig. 9.3,
Basic configuration offuzzy system
219
Fig. 9.4,
The relation between the SOC and SP
221
Fig. 9.5,
The meaningful of the +ve and -ve error corresponding to the system control signal locus
222
Fig. 9.6(a),
Error membership function
223
Fig. 9.6(b),
Change ofError membership function
223
Fig. 9.6(c),
Change of Current membership function
224
Fig. 9.7,
The fuzzy logic controller algorithm
225
Fig. 9.8,
Block diagram ofFLC under MatLab
226
Fig. 9.9,
The plots of Error membership function
227
Fig. 9.10,
The view surface ofthe fuzzy rules of the PVBD system
227
Fig. 9.11,
The FLC of the PVBD system control block diagram
228
Fig. 9.12,
The output response using step reference with 1Os dead time
228
Fig. 9.13,
The output response step change in reference input
229
Fig. 9.14,
The output response sine-wave variation in reference input
229
Fig. 9.15,
The AID converter signal conditional circuit
230
XVlll
List of symbols ANOTSM
actual number of total system modules
ANOSM
actual number of series modules
ANOPS
actual number of parallel strings
BSF
basic fault tolerant systems
Cll
local signal of system 1
C12
remote signal of controller 1
C22
local signal of system 2
C21
remote signal of system 2
D-G
Diesel generator
DB
Diesel battery hybrid system
DM
Deutsch mark
DGC
Diesel generator counter
DSCS
system components size and kWh cost at load duration variation
DSLU
system loss of load probability and kWh cost at load duration variation
DGR
Diesel generator rating
DOUT1&2
output data from system 1&2
DACT
actuator data
DM1&2
medium point data of system 1 & 2
DOD
battery depth of discharge
IARS
industrial applications and remote systems
INSEL.5
INteractive Simulation of renewable Electrical energy supply systems
LOLP
loss ofload probability
LOLC
loss ofload constraint
'A
failure rate
LED
light emitting diode
MTBF
mean time between failure
MDB
meteorological data base module
XIX
MTTR
mean time to repair
MPPT
maximum power point tracking
MC 1&2
microcontroller system 1&2
NAD
number of autonomous days (battery as alone source)
NOSM
number of series modules
NOPS
number of parallel strings
NPS
Northern Power Systems
NOSB
number of series batteries
PV
photovoltaic
PVB
photovoltaic battery system
PVBD
photovoltaic battery Diesel hybrid system
PVBWR
photovoltaic battery with redundancy
PVRAS
photovoltaic remote area systems
soc
state of charge
TVA
Tennessee valley authority
TLCC
total life cycle cost
TMR
three modular redundancy
T1 & T2
watchdog 1 & 2 output signal
UE
unserved energy
WD1&WD2
watchdog 1 & 2
XX
Abstract
The combination ofphotovoltaic (PV) power supplies, and/or potentially other renewable energy sources, with a diesel generator (D-G) in a hybrid electric power system configuration permits solution to many remote sites power problems. But, the sizing of such systems, depending on short-term data, does not give an exact estimate or an appropriate size for the system components, where the number of sunless days is assumed by the designer to assure a reliable power supply. The main problem for longterm systems dynamic sizing and simulation is the site meteorological data base (MDB). In this thesis, the MDB module for different sites in Egypt, as example, is created by using the modern renewable energy systems
simulation program,
INSEL.500, at horizontal and tilted PV arrays plane. The global, direct and diffUse insolation curves during one complete year in El-Giza city, Egypt (site of application) are created and the available solar energy are estimated also. A computer program for systems dynamic sizing and simulation is developed in FORTRAN language to be compatible with INSEL output data format. In this program, the load profile is divided into two load types, dynamic loads and shallow loads to optimize the
operation of different system components, D-G and PV
subsystem. The D-G rating is estimated in the 3kW power-range to handle the load peaks and dynamic loads, and the PV battery subsystem is designed to handle the shallow loads. This program is applied on a case study for an actual data of isolated farm, 300km from El-Giza city, Egypt. The minimum, and hence economical optimum, size of storage battery determined.
and PV array area of a different degrees of reliability is
The loss of load probability (LOLP) and unserved energy (UE) are
evaluated and the number ofD-G operating hours is determined at different degrees of reliability from 90% to 100%. The system total life cycle cost (TLCC), the kWh cost is determined and fitted for different degrees of reliability. The MDB module and sizing program have the ability to be applied for any PV system any where. Many systems consist of subsystems, which in turn consist of components. Failure of any one of these components may cause failure of the entire system. The performance of a system for operation under various environmental conditions is critical in many military and industrial remote area applications. Due to the increasing
XXl
demand for highly reliable, safer, and cheaper systems, two techniques for achieving high system reliability have been identified by either using redundant components or using components of higher reliability. It is often more economical to increase the number of redundant components than to improve component reliability's because component cost may increase exponentially with increased reliability. Digital computers (controllers) systems have been used for control and safety application in industrial applications and remote systems for many years. In most cases, these computer systems must be very reliable because they are critical to the economical and/or safe operation of the process. For increased reliability, many of the more recent digital computer designs provide hardware redundancy in a multi-channel arrangement, relying on redundant processors and fault diagnostic routines to allow continued control function given failures in a single processor. Redundancy is often extended to other portions ofthe system as well (e.g. input modules, input sensors, and system clocks (watchdog timers)). Because of their capability to detect single faults and to isolate the failed equipment (i.e. to ensure that the equipment that is operable takes over the control function), redundant digital control systems that use fault diagnostic routines are called fault-tolerant digital control systems (FTDCSs). The advantage of FTDCSs over those that are not fault tolerant is that the former will continue to function after the occurrence of most single hardware faults. This capability greatly improves the reliability of the system. The operation of a stand-alone photovoltaic system depends not only on the quality of the individual system components but also on their interaction within the total system. The demands on the operation control, to coordinate the interactions, increase with the system complexity. The significance of the operation control is due to its influence on the reliability of the energy supply and the component lifetimes (primarily that of the battery and the auxiliary energy generator).
XXll
Chapter 1
Introduction to the PV Remote Area Systems 1.1. Introduction
The remote area photovoltaic systems can (with respect to the uses) be classified into two types; the photovoltaic battery
stand-alone (PVB) and
photovoltaic battery Diesel generator hybrid power systems (PVBD). A lot of reports are made to study the performance of different remote area photovoltaic systems. For example, the following two references introduce the advantages and disadvantages of such systems and make a comparison between such remote area photovoltaic systems. Doug Danley and Craig A. Shoots have reported in 11/(1989), that the combination of photovoltaic PV power supplies and/or potentially other renewable energy source, with a Diesel-generator (D-G) in a hybrid electric power system configuration is providing highly reliable and cost effective solutions to many remotesite power problems. The presence of the D-G eliminates the need to oversize the PV system to account for "worst-case" solar radiation conditions, resulting in capital cost savings. The renewable energy source, PV, permits a limited and controlled operation of the D-G with a corresponding reduction in operating costs. The characteristics of the PV and D-G are highly complementary, Table 1.1 shows the PVB system versus Diesel generator alone system. Table 1.1, PV system versus Diesel generator. Diesel generator alone systems
Photovoltaic battery systems dependent on the solar sources
independent of natural cycles
charge the battery relatively slowly
typically charge the batteries quickly
difficulty supplying load peaks
handle the load peaks rather easily
offer excellent inherent reliability
reliability mode,
lS
dependent on
its operating
fuel availability and maintenance
which are required on more frequent basis high initial cost and operating cost
very low low initial cost and relativily high operating cost.
So, a combination of the photovoltaic battery with the Diesel generator as a hybrid power supply for remote applications can offer many advantages over either
1
technology taken alone. The PVBD hybrid can provide improved energy availability at a lower life cycle cost than a conventional Diesel or PVB stand alone system. Specific advantages of a PV BD hybrid system include: 1. Higher reliability through two independent power sources, one of which has no moving parts or fossil fuel requirements, and low maintenance needs,
2. Diesel fuel consumption reduced 70% or more, 3. Diesel maintenance reduced to 1/3 or less of normal requirements, 4. Reduced capital costs through reduced PV module and battery storage capacities,
5. Fuel delivery and maintenance will be reduced,
6. Diesel component lifetimes extended to 4 times,
7. Diesel overhaul extended to 10 years.
M. S Imamura has reported in /2/, that a hybrid system needs a didicated system controller (Microcontroller and/or computer)
typically a process logic
controller (PLC) for real-time monitoring and control functions, and it also requires many sensors because of the need to optimize the overall operation of the system, considering; available energy from the PV array, battery operation constraints, and operation constraints of the auxiliary power source D-G to optimize its efficiency and lifetime. V.A.P. von Dijk has reported in /3/, that by using the SOMES simulation model, the technical and economical performance of PVB, DB, and PVBD systems was compared. The hourly energy flows in the system are determined using the hourly load, insolation data, the characteristics of the system components and a control strategy. Because the gasoline generators have a moderate part load efficiency, the calculations in the system control strategy is used in which the generator is always operated at full load. The SOMES results show, that under the given assumptions on load, climate, life-cycle-costs and components specifications, the addition of a gasoline generator to a PVB stand system can be a very cost effective measure to reduce the energy shortage to a level below 5% where: • In the first option, DB, the energy shortage is 14% at the cost about 1.4 ECU/kWh, • In the second option, PVB, the energy shortage is 10% for the equal DB cost. The PVB system can supply electricity with a lower shortage than a DB system, where; 2
by doubling the installed PV capacity (PVBWR), the energy shortage can be reduced to 4% but the kWh cost will increase by 70% compared with the first, • In the third option, PVBD, the energy shortage is 1% whereas the production of the electricity increases about 17% compared with the first.
The summary of the remote area systems comparison are shown in Figure 1.1.
Comparison between PVRAS 2.5
-8
2
1,/)
.c
1.5
~
.., ~
0.5 0 FVB
DB System type
FVBWR
FVBD
Fig.1.1, Comparison between the difference remote area systems.
The prev1ous study presents the advantages of the PV Battery Dieselgenerator hybrid system where the system availability of generated energy is improved to be 99% instead of 83% with only 17% cost increase. The hybrid system concept can be divided into three modules; sizing and optimization module, control algorithm module and reliability module. The following paragraphs will introduce a survey on each module alone. Richard Derlombard has reported in /4/ that the sizes of the PV array and the battery for a stand-alone/ photovoltaic system are normally based on the amount of the
sun light available (insolation) at the planned installation site, the electrical
load requirements (power level and daily profile) and the available funds. The actual daily energy output of the PV
array will depend
characteristics of the installation site. If there energy
on
the actual insolation
is a short term deficit in the array
relative to the load, the battery will supply the required energy. For a long
term deficit, the engine generator will supply the required energy.
3
Stephan J Phillips has affirmed in /5/ the result of V.A.P. Van Dijk /3/, since he reported that; in remote areas of Australia and in many other parts of the world, Diesel-generators are typically used to provide electrical power. Such systems are often
characterised by either poor efficiency and high maintenance costs because of
prolonged operation at low load levels, or intermittent power because the unit is only run during periods of significant load. By addition of a battery bank and inverter to the Diesel-generator, to produce a series hybrid system, this system has identified
been
as producing a number of benifits. These include an improvement in
Diesel-generator efficiency, reduce operating times and a reduced cost per unit of energy. However such series systems also have disadvantages, including the need for large battery banks, the significant loss of efficiency by having all the power pass through the batteries and large inverter capacity. The modified system is to make the PV inverter operate in hybrid
parallel
with
array and storage battery with
Diesel-generator.
The unique features of this
system have significant implications for the sizing ofDiesel-generator sets
and battery banks. The peak loads can now be met by the Diesel-generator and the inverter together. Overall, Diesel-generators can now be utilised and considerably better fuel efficiency and engine life expectancy are achieved. Load sharing between the Diesel-generator and the inverter is effected by power electronic control of the two sources. A key objective in developing the control algorithm was that of maintaning Diesel-generator operation in the range of 60% to 90% of its rated capacity, thus ensuring fuel efficient operation of the unit. Sizing of a remote area power supply is achieved usmg the load demands of the commmunity and
available
meteorological data. With a hybrid systems, it is
still important to size the system properly to obtain the most economical solution, so each component must be sized so as to optimise the overall system performance and the performance of each individual component. Gary J. Jones has reported in /6/ that, the first issue in the design of any photovoltaic system is the considerations in potential
the
selection
selection
will
of be
the appropriate plant rating. The major the power or energy desired by the
owner, and the available capital. These three factors: power, energy, and
capital cost determine the maximum possible size of the system.
4
The second factor in determining the plant size is the proposed site for the plant, and its available area. The only considerations with regard to the selection of
the photovoltaic module have to do with reliability and fault tolerance. Field
experiance has shown that commercially available
offers a 10 year warranty (garantee) on
reliable. At least one manufacturer now module
performance.
configuration
which
configuration results in
The
photovoltaic
emphasizes
module hardware is extremely
module should offer an internal circuit
fault tolerance. For flate plate modules, this
solar cells series connected in the horizontal dimension
and parallel connected in the vertical dimension. The control algorithm of /5/ is developed based on one of three modes: •
Inverter only operation (PV stand-alone).
•
Diesel-generator and inverter in parallel (battery discharge mode).
•
Diesel-generator and inverter in parallel with inverter acting in reverse as a battery charger (battery charge mode). The controller element is a microprocessor which supervices the operation
of the system, selecting the most appropriate mode of operation for the system load being supplied without interrupting the supply of power. The availability of power has been enhanced through the incorporation of the
Diesel-generator in the system design. No interruption in power will occur
when
the
battery if the system
Diesel-generator is automatically started and stopped to charge the battery SOC becomes low. In the unlikely event that the PV power
is unable to provide power to the loads, the Diesel-generator alone is
capable of supplying the required power to all the loads. In the event that both the inverter
(PV
subsystem) and the Diesel-generator are failed, the standby inverter is
capable of providing power to the earth station and the audio converter (critical loads). This part shows that the PVBD is more reliable than the other remote area systems in energy generation mode, but it has some problems in energy transmission mode. 1.2. Conventional PV hybrid systems
The problems of conventional PV hybrid systems can be summarized in two points; sizing methods and the energy transmission reliability. PH. Taslides & A. Thanailakis have reported in 17I, that the sizing depending on the short term data, 5
doesn't give an exact estimation or an appropriate size for the system components, because the number of sunless days (battery capacity) is selected by the designer to assure an absolutely reliable power supply. For the loss of load probability tends to zero, the system at this moment is not economical. Also, E.S. Gavanidau, and A. G. Bakirtiz, have reported in /8/, that the conventional design methods separate the sizing of generators and the sizing of the storage battery, this gives a conservative design with oversized components. The second problem is the energy transmission reliability. M.S. Imamura has reported in /2/, that the high cost of the equipment, its lifetime, and the parts required to maintain it are other justifications for emphasis on high reliability. It would be extremely unfortunate if the failure of a 1OODM part in the inverter caused a complete stoppage of a multi-million DM plant operation, or a failure of a small 50DM bias power supply to the data acquisition equipment wiped out data recording for several months. The design and performance ofPV Diesel-generator Hybrid power system are presented in this thesis as follows:
Chapter 1 gives an introduction to the PV hybrid system, Chapter 2 includes bachground and survey on different PV remote area systems, Chapter 3 intrtoduces the conventional sizing methods and the advantage and disadvantage of each one. The proposed sizing technique is explained and the main problem of the availability of the meteorological data base for system sizing is solved by developing an improved meteorological data-base module based on the INSEL.5 program (INteractive Simulation of renewable Electrical energy supply systems). The proposed sizing method, the developed sizing program and their applications on the design of different PV remote area systems are introduced also in this chapter. In this chapter, the optimal sizing based on the desired degree of sizing reliability, economical analysis are presented. The block diagram of the PV battery Diesel generator hybrid system configuration is also given in this chapter.
Chapter 4 introduce the sizing sensitivity analysis of the three proposals; PV battery stand alone system, PV battery Diesel hybrid system and Diesel battery hybrid system subject to changes in load peaks, load durations and distance of the site of application. The desiegn charts of different PV remote area systems to estimate the relevant system for such application are also presented. 6
Chapter 5 introduces the system reliability studies and estimation methods. In this chapter the system performance is determined using the fault tree analysis technique. The results show that the three critical items in hybrid system are system controller, inverter
and
system distribution subsystem. The distribution subsystem has
approximately no effect on system reliability, but the inverter has the strongest effect on the system reliability since the duplication of the inverter subsystem improve the system reliability by at least 10%. The reliability of different remote area system is studied and the results are presented in this chapter. Also, a sensitivity analysis are made on the system reliability subject to the components mean time between failure (MTBF) and the subsystems degree of redundancy.
Chapter 6 gives a preview of the proposed hybrid PV system management. In this chapter, the present control strategies are introduced and discussed. Also, some ofthe famous PV pilot plants are studied to develop the proposed control strategy.
Chapter 7 will introduce the development of the reliable controller and its implementation. The fault tolerant systems are studied to develop the proposed reliable controller, and their operation strategy. The complete and detailed block diegram and real photo ofthe proposed reliable controller subsystem are given.
Chapter 8
reports on
the Lab testing of the PV battery Diesel generator hybrid
system management, control hardware and software implementation. In this chapter, the system performance is studied and simulated using the INSEL.S and the Matlab simulation programs. A real photo and block diagrams of individual system components and overall system control board are presented. The system control signals and their sampling rates are also introduced.
Chapter 9
introduces the fuzzy logic based PV battery Diesel generator hybrid
system control. The classification of the control systems and the reason for using fuzzy are demonstrated in this chapter. The proposed system control block diagram with the feedback signals are developed. The basics of the fuzzy and its application on the PVBD system is studied and simulated using the fuzzy logic toolbox under Matlab. The implementation of microcontroller based fuzzy logic controller of the PVBD hybrid system is successfully completed and tested on the Lab PVBD subsystem. The FLC implementation algorithm is presented in this chapter.
Chapter 10 introduces the conclusion, results and futur work.
7
Chapter 2
Survey on Pltotovoltaic Remote Area Systems
2.1. Introduction GTZ has reported in /9/ (1993) that, approximately 80 percent of the world's 5.3 billion people live in rural areas of developing countries, most of them with no access to electricity. Sunrise and sunset still mark the beginning and end of the working day.
There is no doubt that electricity spurs the social and economic
development of rural areas: Often the availability of electric power is decisive for the supply of good drinking water, the conservation of food, the storage of medical supplies, telecommunication, radio, TV, etc. It is obvious that along the anticipated path
of development, many developing countries will increase their energy
consumption. A large part of it will be covered by conventional sources like oil and coal. This will contribute to a steady increase in the world's carbon-dioxide (C02) production. Solar panels are one of the very few C02 -free energy convertors. Today, for a range of applications, they are a technically feasible and economical by viable alternative based on a physical process that requires no moving parts. This results in a relatively long services life of solar generators. Solar radiation provides us at zero cost with 10,000 times more energy than is actually used. Most developing countries receive as much as 50 to 100 percent more insolation than countries in temperate zones. Nevertheless, solar or photovoltaic (PV) systems do not come for free. The introduction of such a new technology takes time and effort. The financial barrier (especially regarding the initial investment) is too high for many enterprises and families.
Stuart R. Wenham et.al have reported in /10/ (1994) that, the high cost of extending the electricity grid to customers has meant that many communities, properties and households around the world rely on Diesel, petrol or renewable energy based power supply systems. This applies particularly to countries such as Australia where
there are large remote areas and difficult terrain. Photovoltaic
systems offer an attractive option or supplement to the older technologies. They are widely used in small systems, and are being used increasingly in larger systems. The
8
cost effective region for using a stand-alone system, versus connecting to the grid, varies with load, distance from the grid
and the stand-alone system chosen. For
instance, in Australia, for a grid connection cost of Au$40, 000 the annual load would need to be at least 6000 kWh for the grid to be cost effective than a stand-alone system. For smaller loads, say less than 3000 kWh/annum, a stand-alone system would be more cost effective once grid connection costs exceeded about $A20,000. In the US, a PV system can be cost effective if the grid extension is more than 5 miles and the load is less than 1500 kWh/month., typical of2 houses. Smaller than 2 kW systems can be cost effective at grid extensions of only 1/3 of a mile.
2.1.1. Photovoltaics
Erik H. Lysen, have reported in Ill/ (1994) that, two billion people are without an electrical connection, but the cost of hooking up their homes to a conventional grid system will be too high for most developing countries. Photovoltaic solar systems are a cheaper alternative and can reach virtually any site on earth. For some decades now, photovoltaics (PV) has been on the energy science in the developing world, particularly in established markets such as telecommunications, marine, railway signaling, and cathodic protection of pipelines. The challenge will be to apply PV to the provision of energy to rural homes for lighting, refrigeration, and TV. The most successful to date is the individually owned solar home system: typically a 50W solar panel that charges a battery by day and powers loads after dark. Owners are proud of having their own power system and are not susceptible to grid failure or inconsiderate neighbors, while the equipment itself can be expanded when required.
Richard Griswold, Frank Mancini and Ray Williamson have reported in /12/ (1991) that, the cost of extending utility lines to remote areas is extremely expensive to both utilities and its customers, costing up to $30000 per mile. This high cost discourages the development of remote areas for home sites and other applications. Many people who would like to develop remote sites believe that expensive engine generators using non-renewable fuels are the only option to provide power to remote sites. Photocomm, Inc. has
developed a new hybrid power trailer that combines
9
photovoltaic panels, a battery bank, and a back-up propane generators. It can be utilized for various remote power uses, including remote homes, as well as for emergency applications. In such systems, the design of the electronic controller for array regulation and automatic generator start is a key operation of the system.
The federal Minister for Research and Technology in Germany (BMFT) has reported in /13/ (1992) that, photovoltaics (PV) is the technology of converting light directly to electrical energy. Solar cells, large-area diodes of semiconductor material, are used as energy converters. When light is absorbed by the semiconductor material, an electric potential is built up at the metallic contacts of the diode. Current flows once a load is connected. The absorbed energy has been converted to electrical energy. The ratio of the electrical energy produced to the incident radiant energy is called the efficiency value of the solar cell. All semiconductor materials display a photovoltaic effect, but solar cells have been made from only a few of them in large quantities to date. These are primarily the materials which are also important for electronic components, specially silicon (Si) and gallium arsenide (GaAa), but recently also the somewhat less common semiconductor materials, cadmium telluride (CdTe) and copper indium diselenide (CuinSe2, "CIS"). The market today is dominated by silicon, which also plays the leading role in other areas of semiconductor technology. 1839
Becquerel discovers the photovoltaic effect.
1954
First silicon solar cell in the Bell laboratories.
1958
First satellite with photovoltaic power supply.
1974
First amorphous Si cells.
1983
First photovoltaic power station with a capacity exceeding 1 MW.
1985
First silicon solar cell with an efficiency value higher than 20%.
1989
First tandem cell with an efficiency value of more than 30% under concentrated light.
Silicon is the second most common chemical element in the earth's crust and can be obtained, for example, from quartz sand. Nevertheless, a number of process steps are necessary to gain silicon of high purity which is needed both for microelectronics and for solar cell technology. 10
2.1.2. Solar Cells BMFT has reported in /13 I (1992) that, as in other areas of semiconductor technology, the basis for monocrystalline silicon solar cells are wavers sawn from single massive crystals. Solar cells with efficiency values between 15 and 18 % are produced industrially from this material. At present, these cells command the highest share of the market. It is less expensive to make solar cells out of silicon in other forms, e.g. from multicrystalline or amorphous silicon. Multicrystalline silicon forms when molten silicon is left to cool in a large mould. To obtain amorphous silicon, thin layers of silicon are deposited onto a substrate (e.g. glass) when a gaseous compound such as silane (SiH4) is decomposed. The efficiency of multicrystalline cells is somewhat lower than the monocrystalline cells. The reduction in efficiency is still greater for cells of amorphous silicon, which also have the disadvantage of further degradation in the efficiency value over time caused by solar radiation. Today, crystalline silicon solar cells are 0.3-0.5 mm thick. This thickness guarantees sufficient mechanical stability as well as complete absorption of the solar radiation, for which a thickness
of at least 0.2 mm is needed. Amorphous silicon has a much higher
absorption coefficient, so that layers of only a few micrometers are adequate for solar cells. Silicon materials for solar cells can be defined as: monocrystalline
Wafers sawn from massive single crystals
multicrystalline
wafers sawn from a coarsegrained multicrystalline block
amorphous
thin film deposited onto a substrate from the gas phase
Table 2.1 shows Solar cell efficiency values of different solar cell types: Type
Laboratory cm2
'11%
Production cm2
'11%
Si. -monocrystalline
4
23.3%
100
15-18
Si. -multicrystalline
4
17.8
100
12-14
amorphous
1
11.5
1000
5-8
GaS a
0.25
25.7
4
17
CdTe
1
10.9
-
-
CuinSe2
3.5
14.1
-
-
GaAs/GaSb tandem
0.005
34
-
-
11
2.1.3. Working Principles of a Solar Cell BMFT has reported in /13/ (1992) that, the absorbed light creates electronhole pairs in the semiconductor. The electrons and holes are separated from each other by certain semiconductor structures, as shown in figure 2.1, e.g.
solar radiation (photons)
Fig. 2.1, Silicon solar cell construction /13/ (1992). 1. by a p-n junction between two differently doped layers; 2.
by a metal-insulator double layer on the surface of the semiconductor (MIS structure). A voltage potential is created at the metal contacts on the surfaces, which
causes a current to flow when the external circuit is connected. With a fixed resistance in the external circuit: •
the voltage depends on the choice of the semiconductor material
•
the current depends on the radiation intensity.
2.1.4. Solar Cell Modules In order to get more power, single solar cells are combined to from modules. When solar cells made of crystalline wafers are used, contacts are made on the individual cells and these are soldered together into strings. As in battery construction, a suitable choice of parallel and series connections is made to give the required voltages and currents. The strings are then embedded in a glass/plastic laminate to protect them from the environment, and provided with a frame for mounting /13/ (1992). J.C. Schaefer has reported in 114/ (1990) that, photovoltaic modules, whether single crystal, amorphous silicon, or concentrators, are built up electrically from individual cells. Cells are small, outputs of0.6 volts and one or two amperes. They are 12
wired in series to provide greater voltage and in parallel to provide greater current. An entire plant might contain millions of cells. PV cells, modules, and plants are
typically rated at an insolation (incident solar radiation) level of 1000W/m2, about what occurs at noon on a clear day. Electrical output is a linear function of insolation. PV cell power output depends upon cell temperature also, declining for crystalline modules by about 0. 5 per cent per degree celsius. Therefore, ratings for cells, modules and plants must be based on a specific temperature. Researchers have usually used cell temperature 25 °C; for modules it is more logical to use module temperatures, which will be slightly below cell temperatures. For arrays or plants it is more logical to use ambient air temperatures. Figure 2.2 show the I-V curves characteristics.
1-V Curves with temperature variation (0-60)degree. 2.5
2
4
0 E-
220 V de bus,
0.99
at
120 V de bus,
0.979
L.
1
U(E. -L .)=
1
1
1
1 3.55
if 0
E..
60
-+-90% Rei. --11-91% Rei. ---.6.--92% Rei. ~93%Rel.
3: .loO: ~ (,)
--*-94% Rei. -+-95% Rei. ---+-96% Rei. -97%Rel. -98%Rel.
~99%Rel
0
Cll
= co "'
-e-100% Rei.
40 20 0 7.2
8
8.8
9.6
10.4
11.2
12
Array Area m2
Fig. 3 .14 (a), The PVBD hybrid system sizing results.
71
12.8
PV Diesel-G. Hybrid System kWh Cost 1.4
-+-90% Rei. __..__ 92"k Rei. ---.-94% Rei. -;-96% Rei. -98%Rel. -s-100% Rei.
1.2
u.i _j
.5 en 0
(.)
0.8
--91% Rei. ---*--93% Rei. -+-95% Rei. -97%Rel. -&--99% Rei
0.6
.c
~
0.4 0.2 0 7.2
8
9.6
8.8
10.4
11.2
12
12.8
Array Area m2
Fig. 3 .14 (b), The kwh cost in L.E of PVBD hybrid system. Figure 3.14(a) shows the relation between the PV array area in m2 and storage battery capacity in kWh at different degrees of reliability. In these curves, it can be seen that the optimal sizing range is present between (8 -12)m2 PV array and (520)kWh battery capacity. For a reliability change from 90% to 100%, of course the reduction in PV array area and battery capacity is due to the major contribution of Diesel unit. Figure 3.14(b) shows the kWh cost ofPV Diesel generator Hybrid system at different degrees of reliability. The results shows that the minimum optimal kWh cost is app. 0.5 L.E at 90% reliability and the maximum optimal kWh cost is app. 0.66 L.E at 100% reliability. Therefore a reduction from 1.7357 L.E to 0.66 L.E has resulted when a Diesel unit has been introduced.
3.10.3. Results of the DB hybrid system sizing The results of the Diesel-generator battery system sizing is present in Figure 3.15(a) and (b). The relation between the Diesel-generator power in W and the battery capacity in kWh with the corresponding yearly
Diesel operating hours is
present in Fig. 3.15(a) at different degrees of reliability. The Diesel operating hours versus Diesel rating is present also in this figure. It can be seen that with 100% reliability, the need for battery capacity is too high compared with other operating states.
72
300
4500
250
4000
:2
~
t7J
200
c
>-
3500:;:;
::a
Qj
~
~
:!:::
u
CQ
c. CQ
(.)
~ Qj
3000
=
Qj
c. en 0 ..
150
-
3: _g::1 Qj
100
i5
CQ
lXI
2500
50 0
2000
2634
2355.8
2125.5
1935.1
1775
llesel -G. Power W
Fig. 3.15(a), The DB hybrid system sizing results.
For reliability less than 100%, battery size is constant at very low value up to certain Diesel size. Beyond this size (2125.5) a continuous increase in battery size is needed with increasing Diesel unit size, but on the other hand, the number ofDiesel operating hours is decreased. For 100% reliability a dramatic increase in battery size is needed specially for Diesel unit size below 2125.5W. It seems 100% reliability case is not the best choice for the DB system especially if small size ofDiesel unit is used. The same analysis goes for change in kWh cost with values at 100% reliability higher than either the PVB- or PVBD-systems. 3 2.5
ui _j
2
.5
-rn0
1.5
(.)
.c
~
0.5
0 2634
2355.8
2125.5
1935.1
llesel-G. Power W
Fig. 3.15(b), The kWh cost in L.E ofDB hybrid battery system.
73
1775
Figure 3 .15(b) shows the optimal operating states of Diesel battery system at different Diesel generating ratings and different degrees of reliability. With minimum Diesel rating, the system is not economical at required high degree of reliability. However decreasing reliability only by 1% (99%) would dramativally decrease the kWh cost from 2.5L.E to less than O.SL.E for Diesel size below 2125W. Selecting Diesel size of2125 W would reduce battery size to 150 kWh at 99%.
3.11. PVBD hybrid system configuration The proposed system configuration depends strongly on the results of system sizing, based on the desired degree of sizing reliability. If the 100% degree of reliability (0% LOLP) is taken, then, as the results of sizing program, the system components have the following sizing: • the PV array area is:
11.2 m2,
• the battery capacity is:
10230 Wh,
• The Diesel generator rating is 3kW, automatic and manual starting technique, self speed (voltage and frequency) regulation using governor, which provides the Diesel engine with appropriate fuel rate at different loads.
The number of system modules and hence the system configuration depends strongly on the module parameters. The available PV module have the following parameters: • Monocrystalline PV modules with
12% conversion efficiency,
• the number of cells/module are
20 cell, all in series connection,
• the cell area is
0.01m2,
• The nominal power is PM
23.2W,
• The nominal voltage (VM) is
9.6V.
By matching the results of system sizing and the present module parameters, the system configuration can be determined as follows: The total number of system modules (NOTSM) is : NOTSM
=
PV array area I module area
=
11.2 I (0.01
74
* 20)
=
56 modules
The number of modules in series (NOSM) depends on the system voltage and the site parameters also. The system voltage (VS) is estimated to be 24.0V, and then the number of series modules can be determined as follows:
* (1 -13 * (MAXT- 25.0)))
ANOSM
=
VSI(VM
NOSM
=
int(ANOSM) + 1.0,
The number of parallel strings (NOPS) can be determined as also follows:
where;
NOTSM I NOSM
ANOPS
=
NOPS
= nint(ANOPS)
13
is the temperature coefficient, and
MAXT
is the maximum cell temperature.
The total
number of system batteries depends also on the used battery
capacity, the present battery capacity is 110 Ah at 12V, or 1320Wh, then the number of system batteries will be: NOSB =Nearest integer part of (total battery capacity I battery capacity) =
nint (1 0230 I 1320)
=
8 batteries.
The number of batteries in senes (NOSB) will be, the system voltage by battery voltage (VB); NOSB = VS I VB, = 2 batteries.
The final results of system configuration, in this case (100%) reliability, can be listed as follows: The number of string modules is
4
modules,
The number of parallel strings is
16
string,
The number ofbatteries in series is
2
batteries,
The number ofbattery strings is
4
string.
75
The wiring diagram (block diagram) of system configuration and the system control signals are shown in Figure 3 .16.
PV-Strings Control signals ::~::::~::::::::·~
-- .. .. .. .. .. .. ... ... ... ... ... ... ... ...
- ...., .... - - l .. - - .. - - . .
,
'
•
•
.. .•.......
- --
----:- .. :
-_ -_ -_ -_ .. -_ .. .. -_ .._-_ .. ..
To ac load
.
.
I ....................... ·I .................... :
D-G. rating =3 kW
j
!
·----------------------------~
Fig. 3 .16, The PVBD hybrid system configuration with its control signals.
3.12.Conclusion The present stzmg methods are studied and introduced; energy balance method, LOLP method, and minimization ofthe TLCC method. The advantage and disadvantage of each method are discussed and the proposed sizing method is introduced. The optimal sizing of three PV remote area systems; PVB, PVBD, and DB are estimated and the minimum kWh cost are determined. The results show that under the same operation conditions, the same load profile and the LOLP condition value, the 76
PVBD hybrid system is the most economical proposal compared to the other two proposals; PVB and DB. The PVBD hybrid system configuration is determined based on the 0% LOLP design parameter, and the system block diagram is shown. In the next chapter, a sensitivity analysis is made to study the different remote area systems performance under the change of the load parameters either in the load peaks or in the dynamic load durations. Also, the effectiveness of distances between the site of application and the nearest city on the kWh cost are studied also.
77
Chapter 4
Sizing Sensitivity Analysis ofPlwtovoltaic Remote Area Systems 4.1. Introduction To qualitatively measure the performance of the remote area systems, the expected load variation should be studied. The system sensitivity analysis will study the system performance to determine the relevant system for such load profile and their expected parameters change. The sensitivity analysis is made in terms of the following five cases: 1. sensitivity of optimum system components size and the generated kWh cost with load peak variation (PSCS), 2. sensitivity of the system LOLP with load peak variation at the predesigned (estimated) optimal system components size (PSLU), 3. sensitivity of optimum system components size and the generated kWh cost with dynamic load duration variation (DSCS), 4. sensitivity of the system LOLP with dynamic load duration variation at the predesigned (estimated) optimal system components size (DSLU), 5. sensitivity on the generated kWh cost as a function of the distance between the site of application and the nearest city in km at optimum system components size.
4.2. Sensitivity analysis on sizing of the PVB system In the following, the effect of load parameters changes (load peak and dynamic load duration) on the system components size, kWh cost and system LOLP are studied according to the previous five cases.
4.2.1. Load peak variation on optimum system components size. In this case, the optimum system components size (PV array area and battery capacity) are calculated to meet the variation in load peaks and to meet the desired degree of reliability. In this case, may be small difference in kWh cost is present compared to the similar state due to the change ifthe incrementation value in components
78
s1ze. Figures 4.1(a), (b) & (c) show the variation ofPV optimum array area in m2 and kWh cost in L.E versus the load peak variations in range (-20% to 20%). The results show that the optimum PV array area is sensitive to the load peak variation, since the PV array varies in a range (40.3-60)m2 at 100% reliability and (31. 7-46. 1)m2 at 90% for peak load change from -20% to +20%.
2.5
65 60 >. oiS ~ u N
E Ill
(.)
>.
Cll
...... = lXI Ill
1.5
:.45
Ill
0
(.) .J:
~
40
Ill
.
. 40
.:::
IV
1.25
c(
20 0 -4
-3
-2 -1 0 1 2 3 Variation in Dynamic Loads Duration (hr)
Fig. 4.3(b).
82
4
~
100
2
80
1.75 u.i ...i .E 1.5 Ul
N'
§. 60 Ill
-
.. .. Ill
c(
0
>. 40
()
..c
Ill
c(
1.25
20
~
0 -4
-3
-2
-1
0
2
3
4
Variation in Dynamic Loads Duration (hr)
Fig. 4.3(c). Fig.4.3, PVB, load duration sensitivity on components size and kWh cost.
The Figures 4.3 (a), (b) & (c) show that the variation in PV array area versus load duration variation. The results show that the PV stand-alone system is too sensitive to the load duration variation, since the PV array area changes app. in a linear function with the load duration variation.
4.2.4. Dynamic load duration variation on system LOLP and kWh cost. In this case, the effects ofvariation ofthe dynamic load duration on the system LOLP at the fixed precalculated system components size is studied and the results are shown in Figure 4.4. 60
4
50 3
:.le 0 ~ 40
. w ...i .5
0 ...130 E
2 1ii
Ill
()
~20
..c
-
0
(/)
1
10 0
0
-4
-3
-2
0 2 3 Variation in Dynamic Loads Duration (hr) -1
Fig. 4.4(a).
83
4
:: ~
4
60
50
3 ui
~
~40
.....i .5 2 ti
...1
0
...130
E
0
(.)
Cll
.c
~20
en
1
~
10 0 -4
-3
-2 -1 0 1 2 Variation in Dynamic Loads Duration (hr)
3
Fig. 4.4(b).
4
60
50 ~ 0
3
ui
~ 40 0
.....i .5
E 30
2 ti
...1
0
Cll
(.)
~ 20 en
.c
1 ~
10 0
0 -4
-3
-2
-1
0
2
3
4
Variation in Dynamic Loads Duration (hr)
Fig. 4.4(c). Fig.4.4, PVB, load duration sensitivity on system LOLP and kWh cost.
The Figures 4.4(a), (b) & (c) show the variation in system LOLP% versus the load duration variation. The results show that the system LOLP is sensitive to load duration variation since app. 30% of load is lost if the load duration increases with 4 hours. The results show that the kWh cost is sensitive in both negative variation and positive variation, and the minimum kWh cost is present at 0% variation in dynamic loads duration. The figures show increase in LOLP as peak dynamic load duration is increased while a decrease in kWh cost would result. A minimum value for kWh is detected at duration higher than the nominal design value.
84
4.2.5. Site of application distance on the optimum generated kWh cost. Figure 4.5 shows the kWh cost in L.E as a function of distance in k:m between the site of application and the nearest city. From this results, it can be seen that the distance strongly affects the cost of kWh due to the system components and accessories transportation cost and maintenance cost. As the reliability level is increased the kWh cost will also increase for the same distance.
kWh cost as a function of llstance in PV Stand-alone System
4.5 4 3.5 ui _j 3 ~ 2.5 u; 0 () 2 ..c: ~ 1.5 1 0.5 0 0
0 0 T""
0 0
N
0 0
C'")
0 0
0
0 ...,.
L()
0
0
0 0
0>
0 0
0 T""
Distance in km
Fig. 4.5, The kWh cost as a function of distance in PVB system.
4.3. Sensitivity analysis on sizing of the PVBD hybrid system In this system, a similar study for the PVB is made to study the PVBD hybrid system performance subject to changes in the different load parameters. In this system, may be the number of the Diesel generator operating hours change to meet the optimum kWh cost.
4.3.1. Load peak variation on optimum system components size. In this case, the optimum system components size are calculated to meet the variation in load peaks and to evaluate the desired degree of reliability. Fig. 4.6(a),(b)&(c) show that the PV battery Diesel-generator components size is more sensitive to the load peaks variation as compared with the PV battery stand-alone system,
85
smce the PV array area changes from 3.2m2 at -20%
to 20m2 at 20% load peaks
variations for 100% reliability. To a less extent, the kWh cost is sensitive to the load peak variation. But, the kWh cost in PV-battery-Diesel generator is usually lower compared with the PV battery stand-alone system due to the major contribution ofDiesel unit.
0.8
u..i
_j
-
0.6 .5 I ll
0.4
0
(.)
.t::
s:
-+-Array A.
-.a.- Batt.-Cap.
...:
0.2
--kWh C.
Variation In Load Peaks
Fig.4.6(a).
25 20
0.8
u..i
C\1
E
_j
-
IU
15
0.6.5
>.
10
0.48
5
s: 0.2...:
0
0
~ .
en 10
8..c
0.2 .¥ ==
'#. l{)
'#.
'#.
'
'
Variation in Load Peaks
.....
0 .....
0
Fig.4.7(a).
87
'#.
0 .....
'#. l{)
.....
50
~ 0
40
0.8
u.i
..J3Q
..i 0.6 .5
~20
ti 0 0.4 0
(/) 10
0.2
a.
g
.c
i
~
'# 0
....-
'
Variation In Load Peaks
Fig.4.7(b).
50 ~ 0
a.
..J
0
40
0.8
30
..i 0.6 .5
20
ti 0 0.4 0
10
0.2.:.:
..J
E rn
Cll
~
>-
(/)
u.i
0
=""r-----,r----+
'#
~
'#
'#
....-
....-
10
'
0
'
'#
U{
'# 0
'#
10
0
'# ~
Variation In Load Peaks
Fig.4.7(c). Fig.4.7, PVBD, Load peak sensitivity on system LOLP and kWh cost.
However the rate of increase ofLOLP with increasing load peak is much higher than that corresponding to PVB case. Also, as reliability level is decreased, this increase in LOLP assumes higher values unlike the case ofPVB system.
4.3.3. Dynamic load duration variation on optimum system components size. In this case, the new system components sizes are estimated to meet the variation m the dynamic load duration and to evaluate the desired degree of reliability. The results are shown in Figure 4.8.
88
>. GIS :t::
15
2
12
1.6
u.i ..i 1.2 .5
(J
N
E [,..9 IIJ
-
~
J
!
tn
0
~ .¥
c:C
0.8 0
=
>.cu""t5
IIJ ..
. 6
.c
c:C
0.4
3
~
0 -4
-3
0 2 -2 -1 Variation In Dynamic loads Duration (hr)
3
Fig.4.8(c). Fig.4.8, PVBD, load duration sensitivity on components size and kWh cost.
89
The results show that the PV battery Diesel-generator is insensitive to the variation in the dynamic load duration. Unlike PVB where area of PV increase as duration increases, the area is slightly decreasing as load duration increases. Also unlike PVB system, as load duration increases the kWh cost decreases steadily, while for PVB system a minimum value for kWh cost exists.
4.3.4. Dynamic load duration variation on system LOLP and kWh cost. In this case, the effects ofvariation of the dynamic load duration on the system LOLP at the fixed precalculated optimum system components size are studied and the results are shown in Figure 4. 9.
12
2
10
1.6
:::!! 0
u.i
...J
1.2 .5
c.. 8
_j
0
...J
E
-
6
f/1
0
0.8
Gl
1ii 4 >.
(..)
.c
C/)
0.4
2
0
0 -4
-3
-2 -1 0 2 Variation in Dynamic Loads Duration (hr)
4
3
Fig.4.9(a). 12
2
10
1.6
:::!! 0
u.i
...J
1.2 .5 1ii 0 0.8 (..)
c.. 8
0
...J
_j
6
E Gl 1ii 4 >.
.c
C/)
0.4
2
0
0 -4
-3
0 3 -2 -1 2 Variation in Daynamic Loads duration (hr)
Fig.4. 9(b).
90
4
3: -"
~
12
2
10
1.6
~ 0
Cl. ..J
8
..J
E 6
-
Ul
Ql
0.8
~ 4
en
0.4 0
w
....i 1.2 .5
0
-a-k\11/hC. -4
-3
0
()
.c
~
0 -2
0
-1
2
3
4
Variation In Daynamlc Loads IAiration (hr)
Fig.4.9(c).
Fig.4.9, PVBD, load duration sensitivity on system LOLP and kWh cost.
The results show that the PV battery Diesel-generator is sensitive to the dynamic load duration variation, since both the system LOLP and the kWh decreases with increasing the dynamic loads duration. The minimum system LOLP is present at the minimum dynamic load (-4 hrs from the normal duration (5-7) hrs). It can be seen that the system LOLP is less than the desired value in a range (-3 to 4)hrs of dynamic load duration. Also here unlike PVB system as dynamic load duration increases the LOLP decreases. This ensures the result that PVBD is in favor of increasing dynamic load duration unlike PVB system. While PVB system has a better LOLP as duration decreases but the higher cost of kWh. Also kWh cost is monotinically decreasing with load duration increase, unlike PVB system where a minimum value of kWh cost exists.
4.3.5. Site of application distance on the optimum generated kWh cost.
Figure 4.10 shows the kWh cost in L.E as a function of distance in km between the site of application and the nearest city. From this results, it can be seen that the distance has a strongly affect on the cost of kWh due to the system components and accessories transportation cost and
91
maintenance cost. However, the kWh cost is still much less than that ofPVB system for the same distance. From this results, it can be seen that the distance has a strongly affect on the cost of kWh due to the system components and accessories transportation cost and maintenance cost. However, the kWh cost is still much less than that ofPVB system for the same distance.
kWh cost as a function of []stance in PV Hybrid System
1.6 1.4 . 1.2 w _j
.£
1
~ 0.8
{)
~ 0.6 ~
0.4
~90%Rel
0.2
-a-95% Rei -t.r- 100% Rei
0 0
0
0 .....
0
0 C\1
0 0
("')
0 0 "
0 0 0
.....
Cistance in km
Fig. 4.1 0, The kWh cost as a function of distance in PVBD system.
4.4. Sensitivity analysis on sizing of the DB hybrid system. In this system, the optimum system components size (Diesel rating and the battery capacity) as well as the system LOLP and kWh cost are determined subject to the load parameters changes.
4.4.1. Load peak variation on optimum system components size. In this case, the new Diesel-generator and storage battery sizes are estimated to satisfY the desired degree of reliability when the load peak varies in a range ( -20% 20%). Figure 4.11 shows that the Diesel-battery hybrid system size is sensitive to the
92
load peaks variations, since the Diesel rating increases app. in a linear function with load peaks increase. But, on the other hand the kWh cost decreases with a increase in load peak. This result leads us to a conclusion that, the high ratings ofDiesel generator power is more economical than the small one if the Diesel is still operated in the optimal range (highly loaded).
3
2.5
u..i
.....i .5
-
2
I ll
0
(.)
..c
1.5
1
"#.
"#.
0 N
lO
Variation in Load Peaks
Fig.4.ll(a).
0.
0
ci Qj Ill
Cll
0
3000
0.4
2500
0.39
u..i
.....i 0.38 .5
[2000
u;
Qj 1500 ~
0.37
0
0.1000
0.36
500
0.35
0
"#.
~
"#. lO
'7
"#.
"#.
"#. 0
"#. lO
0 lt{ ..-' Variation In Load Peaks
Fig.4.11(b).
93
"#. 0
"#. lO
"#. ~
0
(.)
..c
:: ..10:
~
3000
0.4
2500
0.39
~ ~2000 (!)
ui
...i 0.38.5
... 1500
1ii
~ ~ 1000
0
0.37 (.)
..r:::
0
3:
0.36.:.::
500
0.35
0
*"10..--.
*"..--0.
*"~ Variation in Load Peaks
Fig.4.11(c). Fig. 4.11, DB, load peak sensitivity on components size and kWh cost.
4.4.2. Load peak variation on system LOLP and kWh cost. In this case the system performance index LOLP is studied when the load peaks varies in a range (-20% - 20%), but the system components sizes are still fixed at the predetermined optimal values. The results present in Figure 4.12 show that the LOLP of Diesel-generator battery hybrid system is sensitive only in the positive part of load peak variation, but the kWh cost is insensitive due to the increasing in the actual generated energy from the system because the Diesel-generator is over loaded. But the increase in LOLP with increasing load peak is effectively less than that for both PVB and PVBD systems.
~ 0
a.
...J
0
25
3
20
2.5 2
15
...J
1.5 1ii 0
-
E (II 10
(.)
1
Ul
>.
r.n
ui
...i .5
5
0.5
0
0
*"~
*"10..--.
*"10..-Variation in Load Peaks
Fig. 4.12(a).
94
*~
..r::: 3: .:.::
~ 0
c.. 0
..J
25
3
20
2.5 2
15
.5
..J
E ell
1ii
1.5 ~ 10
(.)
1
>.
en
u.i
...i
5
.c
~
0.5
0
0
'~*
'*. 10
......
'......*. 0
*
'*
0
It(
'*
10
'......* 0
'......*
10
'~*
Variation in Load Peaks
Fig. 4.12(b). 25
3
20
2.5
~
•c..
..J
0
2 15
.5
..J
E
ien
u.i
...i
1.5 ~ 10
(.)
1 5
.c
~
0.5
0
0
'~*
'......*. 10
'......*. 0
* It(
* 0
'*
10
Variation in Load Peaks
'......* 0
'......*
10
'~*
Fig. 4.12(c). Fig.4.12, DB, load peak sensitivity on system LOLP and kWh cost.
4.4.3. Dynamic load duration variation on optimum system components size. In this case the relevant system components sizes are estimated according to the variation in the dynamic load duration variation in hrs to evaluate the desired system degree of reliability. The results are shown in Figure 4.13. The results show that the Diesel-generator battery hybrid system size is too sensitive to the dynamic load duration variation in hours, since the Diesel power changes from (372 W- 2716 W), while the corresponding change in the battery capacity is from (11 - 370) kWh's.
95
4000
10
3500 o6
[
..
8
3000
>.
u.i ...i 6 .5
;!:
~ -2500
Q. 0
~~~2000 >. 3:
0
=-
a. Gi Gi
Ill
Cll
i:3
Ill
co
..:0:
I ll
4
1500
0
(.)
.s:::
3:
1000
2
..:0:
500 0
0 -4
-2 -1 0 2 3 -3 Variation in Dynamic Loads Duration (hr)
4
Fig. 4.13(a). 0.6
4000
..
3500
a. a. [2000
2500
0.5w ...i .5 0.4 ~
Gi
1500
.s:::
Cll
1000
~
0
3000
(.)
Ill
i:3
0.3
3:
..:0:
500 0
0.2 -4
-3
0 3 -2 -1 2 Variation in Dynamic Loads Duration (hr)
4
Fig. 4.13(b). 4000
.. ~
0.6
3500 0.5 u.i ...i .5 0.4 Ill
3000
2500 a. a. [2000
0
Gi
Ill
Cll
i:3
0
(.)
1500
.s:::
0.3
1000
~
500 0
0.2 -4
-3
-2
-1
0
2
3
4
Variation In Dynamic Loads Duration (hr)
Fig. 4.13(c). Fig. 4.13, DB, load duration sensitivity on system LOLP and kWh cost.
96
But the generated kWh cost is not sensitive because the increasing in the diesel operating hours meets decreasing in the required storage battery, and then eliminate the increase in the generated kWh's cost.
4.4.4. Dynamic load duration variation on system LOLP and kWh cost. In this case the system performance factor LOLP is studied when the dynamic load duration varies in a range (-4- 4) hours, but the system components sizes are still fixed at the predetermined optimal values. The results are shown in Figure 4. 14.
-
en
0
(.)
..c
~
20 0
0 -4
-3
-2
-1
0
2
Variation in Dynamic Loads Duration (hr)
Fig. 4.14(b).
97
3
4
100
5
80
4
60
w ...i 3 .5
~ 0
0.
...I
0
-
...I
Ill
0
E
2 (.) .c
40
Ill
::.:.::
~ Ill
20 0
0 -4
-3
0 -2 -1 2 Variation in Dynamic Loads duration (hr)
3
4
Fig. 4.14(c). Fig. 4.14, DB, load duration sensitivity on system LOLP and kWh cost.
4.4.5. Site of application distance on the optimum generated kWh cost. Figure 4.15 shows that for reliability less than 100% DB-system is approximately insensitive to change in distance and a high increase in kWh cost with increase in distance at 100% reliability where the least cost corresponding to the PVBD system. However for reliability less than 100%, DB system results in least kWh cost for long distances.
kWh Cost as a function of Distance in Diesel-Battery Hybrid System 4 3.5
Ill
0
(.)
.c
~
3 2.5 2 1.5 0.5 0
Fig. 4.15, The kWh cost as a function of a distance in km in DB battery system.
98
4.5. Comparison between different remote area systems performance The
results of the sensitivity analysis of the system performance subject to
changes in the load parameters are summarized to make a comparison between the different PV remote area system at different load profiles.
4.5.1. Effect of the dynamic load duration variation on optimum kWh cost Figure 4.16(a), (b) and (c) show the comparison between the generated kWh cost from the different remote area systems if the dynamic load duration varies in a range (-4 to 4) hours.
Remote Area Systems Performance (100% Rei.- DSCS)
8 -+-FVB -11-FVBD
u.i
6
--.t.-DB
_j
.5 +"'
Ill
4
0
0
.c
~
2
0 -4
-3
-2 0 3 -1 1 2 Variation in Dynamic Loads Duration in (hr)
4
Fig. 4.16(a), kWh cost ofPVRAS at 100% Rei. & load duration variations.
The results show that, for 100% reliability, the optimum kWh cost for the PVBD system has the lowest value for very wide change in dynamic load duration ( 4hr) from the nominal value. However, the DB-system can result in comparable cost as load duration increases from the nominal value.
99
Remote Area Systems Performance (95% Rei.-DSCS)
2
ui 1.5 .J
.A
.5
';; 0
-+-PI/8
1
-a-PI/80
0
.c ~
== 0.5 .:.::
---
--6--DB
-
0 -4
-3
4
3 -2 -1 0 1 2 Variation in Dynamic Loads Duration (hr)
Fig. 4.16(b), kWh cost ofPVRAS at 95% Rel. & load duration variations
Remote Area Systems Performance (90% Rei. - DSCS)
2 -+-PI/8 -a-PI/80
ui 1.5 .J .5 ';; 0
1
0
.c
~ 0.5
--6--DB
~
~
-
-
0 -4
-3
-2
-1
0
1
2
3
4
Variation in Dynamic Loads Duration (hr)
Fig. 4.16(c), kWh cost ofPVRAS at 90% Rel. & load duration variations
For reliability less than 100% the DB-system results in minimum kWh cost over very wide load duration changes from the nominal value ( 4hr) while the PVBD can result in comparable cost especially at load duration increase in the positive direction. At 100% reliability, DB-system can result in very high kWh cost as load duration decreases.
100
4.5.2. Effect of load peak changes on optimum kWh cost Figure 4.17(a), (b) and (c) show the comparison between the generated kWh cost from the different remote area systems ifthe load peak varies in a range (-20 to 20)%. The results show that, for optimum operating conditions, at 100% reliability, the best choice is the PVBD system for both peak load and load duration change while at reliability level less than 100% (up to 99%) the DB system is the best choice. However DB choice can be the first choice for higher levels of dynamic load duration's. Remote Area systems Perfonnance (100% Rei ..PSCS)
2.5
w _j
-
..-
2
-....
.5 1.5 I I)
0
0 .c
1
::.:r: 0.5
--+-PVB ----PVB -.a.-DB
0
*"..... 0
I
*" "?
0
*"
1.()
*"~
*" 0 N
Variation in load peaks
Fig. 4.17(a), kWh cost ofPVRAS at 100% Rei. & load peaks variations.
Remote area Systems Perfonnance (95% Rei.- PSCS)
2.5 --+-PVB
. w
----PVBD
2
-.a.-DB
_j
.5 1.5
Vi 0
0 .c
1
::.:r: 0.5 0 0~
0
NI
0~
1.()
..... I
*"..... 0
I
0
*"
*"
1.()
1.()
I
0~
0 .....
*"..... 1.()
0~
0
N
Variation in Load Peaks
Fig. 4.17(b), kWh cost ofPVRAS at 95% Rei. & load peaks variations.
101
For distance change still DB is the best choice concerning kWh cost for reliability level less than 100%, while at 100% reliability level it is the PVBD system that result in minimum kWh cost.
Remote Area Systems Performance (90% Ret.- PSCS)
2.5 -+-PIIB --PIIBD -.-DB
2
u.i .J .5 1.5 ..... 1/)
0
(.)
.c
;:: ..:.::
A.
_A
1
-
0.5 0 0~
0 NI
0~
.
.c1
~
20
1/)
>.
U)
10 0
0 -4
-3
-2
-1
0
1
2
3
4
Variation in Dynamic Loads Duration (hr)
Fig. 4.18(c), kWh cost and LOLP ofPVRAS at 90% Rel. & load duration variations, (at Fixed components size).
However, at reliability less than 100%, both PVBD- and DB-systems results in comparable rebust minimum kWh cost, yet, PVBD-system shows no deviation from desired LOLP where as DB-system results in high deterioration in LOLP as load duration increases from the nominal value. Hence, PVBD-system results in the most robust behaiviour of all three systems as far as kWh cost and LOLP are concerned as load duration is changed from the nominal value.
4.5.4. Effect of load peak changes on system LOLP and kWh cost Figure 4.19(a), (b) and (c) show the comparison between the generated kWh cost and the LOLP of the different remote area systems if the load peaks varies in a range (-20 to 20) % of its rated value, at stationary predetermined optimal system components size. The results show that the PV-battery-Diesel generator hybrid system has the minimum kWh cost ifhigh degree of reliability (100%) is required, but on the hand it has a medium degree of reliability, since the reliable system is the Diesel-battery system, but it has too expensive kWh cost. If 95% degree of reliability is required, the most economical and reliable system is the Diesel-battery hybrid system. If 90% degree of reliability is
104
required, the Diesel-battery hybrid system has a minimum kWh cost, but the PV-battery stand-alone system is the most reliable system.
Remote Area Systems Performance (100% Rei-PSLU)
3
-+- kVVhFVB
kVVhFVBD --.- kVVhDB -4)-- LOLPFVB -a- LOLPFVBD -1:r- LOLPDB
. 2.5
w ...i .5
~
-
2
'Iii 1.5
0 ()
.c
~
1 0.5 0 ~ 0
0 NI
~ 0 1.0
......I
*0 ......
*1.0
0
I
~ 0
~ 0 1.0
0
N
I
Variation in Load Peaks
Fig. 4.19(a), kWh cost and LOLP ofPVRAS at 100% Rel. & load peaks variations, (at Fixed components size).
Remote Area Systems Performance (95% Rei.-PSLU)
*1.0 ......
*0 ....-
I
I
*1.0
0
I
*1.0
*0 ..-
0~
0 N
Variation in Load Peaks
Fig. 4.19(b), kWh cost and LOLP ofPVRAS at 95% Rel. & load peaks variations, (at Fixed components size).
105
Remote Area Systems Performance (90% Rei.-PSLU)
3
50
'(ft. I()
..-I
'(ft. 0
0~
0
I()
0~
I()
~ 0
0
I ..-I Variation in Load Peaks
~ 0
0 N
Fig. 4.19(c), kWh cost and LOLP ofPVRAS at 90% Rel. & load peaks variations, (at Fixed components size).
At 100% reliability subject to changes in dynamic load peak of± 20%, PVBDsystem results in constant kWh cost with least value as compared with other two hybrid and stand-alone systems, however, it results in deterioration rate in LOLP higher than that corresponding to DB-system. For reliability levels less than 100%, both DB- and PVBD-systems result in comparable constant kWh cost of value much less than that corresponding to PVBsystem, morever, PVBD-system results in much higher rate of deterioration in LOLP as dynamic load peaks increases from its nominal value than that corresponding to DBsystem.
4.6. General Conclusions
DB design is very sensitive to reliability design level decrease of 1% in this level from 100% to 99% can result in much important behavior concerning kWh cost and sensitivity to change in location distance for nominal design value. However, LOLP sensitivity is the worst of all cases concerning change in dynamic load duration, but it is better than that for other systems concerning peak load change.
106
Therefore if 100% reliability is a must, then PVBD should be used. Also for long distances this decision would be the sound one. As load duration decreases a dramatic increase in kWh cost would result (L.E. 8 at -4hrs) with equal reduction in Diesel size for 100% reliability case. However, reducing reliability level to 95% the kWh cost increase is highly reduced with equal reduction in Diesel size. At 100% reliability, using DB system results in approximately same reduced kWh cost as that corresponding to PVBD as load duration is increased. While at nominal conditions DB results in 2 L.E/kWh while for PVBD it is 0.6 L.E/kWh, therefore in dynamic load duration increases it is better to use PVBD. However reducing reliability to 99% this logic is reversed and selecting DB is more economical.
107
Chapter-S
Pltotovoltaic Remote Area Systems Operation Reliability
5.1. Introduction The unreliability of the world's artifacts account for large amount ofwasted time and money and the endangerment of human life. As the consequences of this type
of unreliable behavior become more serious, so the interest in reliability and
the desire for more reliable products increases. A Mohamed, et al., have reported in /37/ (1992) that, many systems consist of
subsystems, which
in tum consist of components. Failure of any one of these
components may cause failure of the entire system. The performance of a system for a mission under various environmental conditions is critical in many military and industrial applications. Due to the increasing demand for highly reliable, safer, and cheaper systems, two techniques for achieving high system reliability have been identified: 1. Using redundancy, and
2. Exercising a component reliability improvement program.
The
generally
applicable
methods
for
achieving high reliability of all
systems include the following steps: 1. Proper choice
of configurations
and parts during the design phase to
achieve the expected level of reliability. 2. Failure modes and effects analysis (FMEA); timely FMEA and redesign effort during the development phase can minimize the chances of subsystem and system level failure. 3. Adequate testing and inspection. 4. Good quality control throughout product manufacturing and installation. 5. Monitoring and proper failure reporting, e.g. an adequate test programme and timely dissemination of information. 6. Configuration or procedural changes based on performance and trend data. 7. Fault tolerant
systems
usage
(redundancy
technique).
108
or
parallel processing
The
first four steps above dictate the inherent reliability designed into
the equipment or for The
an entire system. The two before the last determine the potential
achieving high reliability and availability of the power plant and subsystem. last
one
provides
an
approximately
absolute degree of reliability and
availability. Based on the monitoring results, certain changes can be implemented in the
system configuration or operational mode in an attempt to extend the life of
components or to prevent further premature failures. It must be emphasized that reliability
cannot be tested into a system or added in any way if the design itself is
inadequate to begin with /38/(1992). As with the practical problems, a balance must
be
struck between an
acceptable degree of reliability in products performance and the cost. The important question is therefore posed about every item where reliability is of interest "is it reliable enough" ?, This inevitably leads to the conclusion that to achieve an economic balance, reliability itself must be defined and hence used in a way that it becomes a measurable quantity.
5.2. Reliability Definition and Relationships Reliability is defined as that characteristic of an
item expressed by the
probability that it will perform its required function in the desired manner under all relevant
conditions and on the occasions or during the time intervals when it is
required so to perform. Also the reliability can defined as: "The reliability of a system is simply the probability that the system (or an item of
equipment)
will
complete
its intended mission successfully". More
precisely, we can define it as the probability that the system will work as designed for a specified duration of time.
An important aspect of reliability, contained in the above definition, is to see how probabilistic concepts may be applied to events occurring in the time domain. We shall see later how this is developed. The probability that an item of equipment will work successfully P(t) 1s described by: P(t) = e -A- .t
5.1
109
where; /...,
total failure rates of parts in the equipment,
t
total time of successful operation of the equipment.
The reciprocal of failure rate is mean time between failure (MTBF), or MTBF= 1//...,
5.2
5.2.1. Reliability of systems with n-items in series If a system consists of several subsystems or items of equipment in series, each having its own probability, the total reliability is equal to the product of all individual probabilities. Thus, if a system is made up oftwo items of equipment having PI and P 2 probabilities as shown in Figure 5.1, the system success probability (reliability) will be PI
*
P 2. This extends to an equipment with n serial parts that are all used at 100%
duty cycle, the system reliability Rs is :
Rs =PI * P2 * P3 .... * Pn
n
R8
= IT P. . 1
1=
5.3
1
where; P
is the item reliability,
n
is the number of items in series.
A
p2
PI
~
B
Fig. 5. I, A system made up of two subsystems or parts in series.
It can be seen that as more items are added in series (Figure 5. 1), the system
reliability decreases since the individual probabilities are all less than unity. Substitution of Eq. ( 5.1) into Eq. ( 5. 3), the "n" series systems reliability can be expressed as follow: 5.4
110
-
J
l: A. t [ n R =e i=1 z s
5.5
The reliability of a system, therefore, can be found by summing up the failure rates of the component parts. This holds true for a non-redundant system as illustrated in Figure 5 .1.
5.2.2. Reliability of systems with m-items in parallel For a system that has a redundancy as shown in Figure 5.2 (for two parallel items) in which block P 1 or P 2 can successfully transfer power or signal from point A to B, the reliability of this system Rs is:
Rs
= 1- (1- ~)(1- ~)
5.6
PI B
A
Pz Fig. 5.2, A system made up oftwo redundant paths in parallel.
The failure rate for a redundant pair can be calculated by an approximately using the expansion ofthe basic reliability equation (Eq. 5.1).
P(t)=e-A t 5.7 P(t) =: 1- At
5.8
If q is defined as the probability of item failure, then, 5.9
q=1-P=:A.t
Then usmg Eq.(5.8) for Pi in eq.(5.6), the reliability of a system with redundancy (for two parallel items) becomes: 5.10
111
5.11
A combination of Eq. ( 5. 8) and ( 5.11) shows that a parallel redundant path can be replaced by an equivalent single path or block that has an effective failure rate: 5.12 and its effective MTBF is: 1
5.13
MTBFiff=e A,
eff
For the redundant systems with m items, the system reliability can be expressed as follows:
m(1-P. ) R =l-IT s i=l l
5.14
where; m is the number of parallel items.
If the m items are identical, the system effective failure rate cad be deduced as equations (5.7- 5.12) as follows:
Aeff =Am* tm-l
5.15
5.2.3. Reliability of series-parallel systems Figure 5.3, show the general series-parallel systems configuration, where the number of series stages is "n" and the number of parallel items is "m".
:& .. .. . . .. ..
.. ..
·--------~ Stage 1
Stage 2
Stage 3
Stage n
Fig. 5.3, Series-Parallel system configuration 112
The system reliability can be expressed by:
Rs
n[
IT
(1- m. (1-P. .) i=1 }=1 lj
= IT
J
5.16
5.3. Reliability Design and Optimization The considered two general situations where an economic balance must be struck between the system reliability and the financial investment are listed bellow. Each is a very elementary general
appreciation
example
which is intended quite simply to provide a
of how reliability engineering can effect system design
/39/(1981), /40/(1995). The optimal design configuration of a series-parallel system which give a maximum reliability are studied subject to two constraints; budgetary constraint and safety constraint.
5.3.1. Under budgetary constraint. In this case the, the design problem, according to this mathematical model, is to choose suitable values for then & m (series stages and parallel items), such that C ::;; B and R is a maximum value. The cost of the system, C is given by:
n mi C
= 2: 2:
ciJ~B
5.17
i=1 }=1 where; B
is the budget limit,
c
is the item cost.
N
is the number of system components which equal to n*mi
5.3.2. Under safety constraints. This technique is used to compensate the effect of spurious signals which may be picked up and
triggering
off the operation of a system component, and
hence setting other system stages into operation or nonopenration. If the probability of a component in the first stage picking up a spurious signal is Pi (item reliability), then the probability of a component not picking up the
113
signal is qi
=
1-Pi (item failure).
The overall system probability to operating due
to a spurious signal when there is no controlled input signal is:
n [ (1- m. rr ri (1-lj.) ] c.s . 1 . 1 I]
1=
5.18
j=
where; Pij
is the reliability of item j in stage i,
S
is the minimum required degree of safety,
n
is the number of series stages,
mi
is the number of parallel items in stage i.
5.4. Application of Fuzzy Logic to Systems Reliability
J.B. Bowles and C. Pelaez have reported in /40/(1995) that, in the case of reliability , uncertainty is also due to the fact that since failures are relatively rare events (typically only a few per million hours of operation), collecting enough data of which to base a statistical
"probability of failure" is a closely and difficult
undertaking, and the relevance of the data to any particular system, as well as its validity, is often questionable. Furthermore, especially early in the design, the item whose probability of failure is needed often does not exist and it must be estimated based on engineering judgment or experience from similar items. Extrapolating those failure probabilities through statistical methods to calculate a system level reliability only increases the uncertainty. By allowing imprecision and approximate analysis, fuzzy set theory and possibility theory, collectively referred to as fuzzy logic, help to restore integrity to reliability analysis by allowing uncertainty and not forcing precision where it is not possible. In reality, many systems performance degrades rather than or before they suffer outright failures, and they have a range of states in which the system is partially working. In such situations a fuzzy set may better define the event whose probability is of interest. For parallel systems; for example, The
system works as at least one
component is operational. If, as is traditionally assumed, components are independent and the system is either working or failed, the system reliability is defined as Eq. (5.14)
114
5.14
Fuzzy logic provides an intuitively appealing way of handling this uncertainty by treating the items probability of failures or reliability's as a fuzzy number. This allows the analyst to
specify a range of values with an associated possibility
distribution for the failure probabilities. If we associate a triangular membership function with the interval, we assume that the analyst has more confidence that the actual parameter lies near the center of the interval than at the edges, and then the system reliability equation could be reexpressed as illustrated in Eq. 5.14':
m R8 =1-_IT (1-~) 1=1
5.14'
This logic could be applied on the series systems or series-parallel systems, and then the system reliability equations can be rewritten as:
n
a) for series systems,
Rs= IT~ i=1
5.3'
IT. 1
n [ (1- m. (1- ~.) ] b) for series-parallel systems, Rs = IT . 1 1=
]=
y
5.16'
The system reliability can also be evaluated using linguistic descriptions of the components failure probabilities. Such descriptions are often used as guidelines for estimating numerical reliabilities. A system analyzed is evaluated using the linguistic rule and corresponding fuzzy arithmetic operations to yield the fuzzy system reliability. The linguistic function for each item reliability (or item probability failure) can be defined as shown in Table 5.1. If desired, the fuzzy result can be defuzzi:fied to give the system reliability rule value. The defuzzi:fication process creates a single assessment from the fuzzy conclusion set expressing how risky the design is, so that corrective action can be prioritized. Several techniques have been developed, one common technique is the Weighted Mean ofMaximum method. This technique averages the points of maximal possibility of each fuzzy conclusion, weighted by their degrees of truth:
115
•
Support value of Low
=
4,
•
Support value ofModerate
=
6,
•
Support value ofHigh
=
8.
Table. 5.1, The linguistic function of item reliability. ]. QUfuitita~iv~:f@uift:Prol.la~ili!; Remote
Failure is unlikely
-
=
:c .!!!
&!
•0-20 40
5 Redundancy
Operating Hours (hr)
Fig. 5. 17, PVBD system, results of sensitivity analysis on system with redundant components at different degrees of redundancy.
.80-100
100
I5J 60-80
80
040-60
~ 0
>-
=
m20-40
60
·0-20
:c .!!!
&! 5
Operating Hours (hr)
~ ~
1 Redundancy
Fig. 5. 18, PVB system, results of sensitivity analysis on system with redundant components at different degrees ofredundancy.
The results in Figure 5.18 show that the effectiveness of improving the system reliability using the redundant components is present at the second degree of redundancy for the PVB system, since the system reliability improved by app. 10%. Also the previous problem in PV-Battery-Diesel hybrid system which is the system
139
degradation growth faster ifthe operating hours is more than 10000 hours. For the DB system the results are illustrated by Figure 5.19. 1180-100
100
[360-80
80
040-60 fl!ll20-40 11110-20
5 1 Redundancy
Operating Hours (hr)
Fig. 5.19, Diesel-battery hybrid system, results of sensitivity analysis on system with redundant components at different degrees ofredundancy. The results in Figure 5.19 show that the Diesel-battery hybrid system reliability. The results show that the system reliability is increased by app. 12% using the redundant components
for both the inverter and controller with m=2. Also the
previous problem in PV-Battery-Diesel hybrid system which is the system degradation decaying faster if the operating hours is more than 10000 hours. The general conclusion for the redundancy technique usage to improve the system reliability can be summarized in that; the redundancy is effective till18,000 hrs of operation. After that the system reliability may decreased although the redundant components are used as listed bellow for the different systems:
a) PVBD system at m = 2, 22,000 hrs, the reliability decreased to 3.0% from 4.2% at m=1, • at m = 3, 20,000 hrs, the reliability decreased to 3.8% from 5.6% at m=1, • at m = 4, 18,000 hrs, the reliability decreased to 8.4% from 10.0% at m=1, • at m = 5, 18,000 hrs, the reliability decreased to 7.9% from 10.0% at m=1,
b) PVB system • at m = 2, 18,000 hrs, the reliability decreased to 5.8% from 6.2% at m=1, • at m = 2, 18,000 hrs, the reliability decreased to 5.5% from 6.2% at m=1, • at m = 2, 18,000 hrs, the reliability decreased to 5.2% from 6.2% at m=1,
140
• at m = 2, 18,000 hrs, the reliability decreased to 4.9% from 6.2% at m=1,
c) DB system, • form= 2- 5, 18,000 hrs, the reliability decreased to 0.2% from 0.3% at m=l.
5.1 0. 4. Effectiveness of preventive maintenance and redundancy on the system reliability. The different remote area systems reliability are studied including the different degrees of critical system components redundancy and the effective of the preventive maintenance, and the results are shown in the following charts.
PVBD Sys. Rei. w Preventive Main. & Redundancy= 2 100 80
~ 0
»
= 60
:c .!!!
~
E 40 Gl en » fl) 20
-
0
Operating Hours (hr)
PVB Sys. Rei. w Preventive Main. & Redundancy= 2 100
~ 0
80
»
=
:c .!!! Cii
60
0::
E 40
Gl
en » fl)
20 0 0
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ..-- ..-- ..-- ..-Operating Hours (hr)
141
..--
..--
..--
..-
..--
8
§ ~
~
System Reliabilty% 1\.)
0
~
0
~
~
System Reliabilty %
.... 8
II
1\.)
0
I
;:,
1-rj
20000
"tJ
"tJ
< DJ
40000
'
190000 200000
~ "tJ
@
~ cr.
:::J
~
:!: I» :::J
.
Qo
::0 ~
c. c :::J c. I» :::J
~
II N
DB Sys. Rei. w Preventive Main. & Redundancy= 4 100
~ 0
80
>.
~
:c .!!!
60
&! E 40
Ill f/1
>.
t/)
20 0
Operating Hours (hr)
Fig. 5.21, The systems reliability including the quad redundancy degree.
Figures 5.20 and 5.21 show the effectiveness of using the redundancy technique beside the preventive maintenance on the system reliability. In all cases, the system reliability are studied based on the standard system components failure rate (Tables 5.2 and 5.3), 10,000 hrs operation interval and 1.1 times for the factor a. It is cleared that the system reliability has a good improvement as well as the
critical system components degree of redundancy increased. For the three proposed systems, the system availability lies between 99.9 and 100% availability.
5.11. Conclusions
In this chapter, the reliability relationships are expressed for single items, items m senes connections and redundant items (items in parallel connections). The equations show that system reliability depends upon the items or subsystems MTBF' s and the operating hours. The general remote area systems configuration with relevant components MTBF' s are shown. For each system, the fault tree for system failure to deliver power to load is deduced and the system characteristic equations are expressed as a function of the system operating hours. For each system, the fault tree for system failure to deliver power to load is deduced and the system characteristic equations are expressed as a
143
function of the system operating hours.
The remote area systems characteristics
equations are expressed as a function of the factor cr. The results show that cr variations the has a marginal effect on RAS reliability, where at cr = 70%, the system reliability's has a very small improvement compared to cr = 30% case. The results show also that, all the proposed remote area systems reliability is sensitive to the operating hours value, as the operating hours increases, system reliability would decrease. For standard components MTBF, the reliability curves show that, the PV-battery-Diesel generator hybrid system is the most reliable system compared to the other two systems, but in general all the three systems are sensitive to the operating hours since it varies in a negative exponential functions. The systems availability's are also studied and the results show that after more than 25 years of operation, the PVB system has 99.7 % availability, PVBD system has 99% availability and the DB system has 96% availability. Two methods are studied to improve the system reliability, by using more reliable components or by using a redundancy of 100% or higher for the critical system items; inverter and system controller: 1. By using an inverter and controller ofMTBF 50000 hrs (2.85 times higher than the standard), the PVBD; for example, has a reliability improvement from 40% at the standard components MTBF (17520 hrs) to 80% at 50,000hrs MTBF's for 5000 hrs of operation.
2. On the other hand, by using a redundant components, the results show that, as redundancy increases reliability decreased at slower rate with increasing the operating hours, e. g. at redundancy of 5: 1 reliability decreased from 100% to 97.7% after
5000 hrs of operation, however beyond 10,000 operating hours
reliability continue decreasing but at a rate higher than before 10,000 hrs interval. Increasing operating hours further than 15,000 hrs redundancy will result in worse deterioration in reliability than the case without redundancy. Consider regular preventive maintenance will result in a further improvement in the system reliability. However further study is needed to optimize the tradeoff between the expected increase in maintenance cost and the resulting improvement in system reliability.
144
The remote area systems reliability's
and availability's are also studied
including the effectiveness of preventive maintenance with and without the redundancy technique usage. The results show that the systems reliabilities have a good improvement as well as the redundancy degrees increases beside the preventive maintenance
usage,
and the systems availabilities have approximately 100%
availability's in all cases of redundancy degrees. But, in our study, the remote area systems are studied only with constant maintenance interval and constant factor a, if the maintenance interval is changed the factor a should changed too, and this results a change in the kWh cost. A trade-off should be present to estimate the optimal maintenance interval and relevant factor a which gives a minimum kWh cost. This part should be thought about in future work.
145
Chapter 6 Review of Present Hybrid PV System Management 6.1 Introduction
The system controller in stand-alone systems may be simple or complex, depending on the power level and the extent of autonomous operation. As the power level goes up, the number of sensing and control functions generally increases, thereby decreasing the system reliability. Power conditioning requirements are more severe for stand-alone systems because of the need for effective battery charge control and discharge protection. Series regulation devices add to the overall system cost, and should be avoided, especially on high power plants, if possible. A critical operational control consideration for stand-alone with storage battery plants is how to prevent the battery from: 1. Repeated deep discharges cycle (more than one cycle per day), 2. Exceeding the specified DOD limit for normal operation, and 3. Overcharging (gassing state). A hybrid system therefore needs a dedicated process logic controller for realtime monitoring and control functions to optimize the overall operation of the system, considering: •
Available power from the PV array,
•
Battery operating constraints,
•
Operating constraints of the auxiliary power source, and
•
Load management criteria. For low power stand-alone PV applications, only DC loads should be selected
because an inverter and hence its losses, can be eliminated. Individual loads are characterized by their input voltage and power requirements. DC loads consist of resistive, constant current, constant voltage, or constant power in applied voltage range. Typical de loads are lamps, radios, stereo, TV, refrigerators, battery charging, and motors. The voltage and current inputs to motors vary according to the mechanical torque requirements of the driven loads. The transient in-rush current (4 6 times the steady state values) of motors must be considered in the system sizing procedure /2/(1992). In the proposed system sizing program, not only the in-rush
146
current of the motor is included, but also the Diesel generator preheating time (5 minutes).
6. 2. Goals of control strategies in PV systems H. Plus and D. Sauer have reported in /46/(1996) that, the control strategy connects all system components and should ensure an optimal cooperation aiming at a reliable and economical operation. This means: • minimizing loss ofload probability, • maximizing solar output energy, • minimizing back-up generator running time (ifthere is one), • maximizing battery lifetime (avoiding deep states of charge, high corrosion voltages, neutralizing acid stratification, regular complete charting), • minimizing the operation costs. There are numerous requirements to the control strategy, but several of them are contradictory and so an optimal trade-offbetween these requirements is needed.
6.3. Present Control Strategies R. Kaiser, et al. have reported in /22/(1996) that, the performance of an operation control system of PV remote area systems can be described as optimum when the temporal sequence of measures (the "strategy") is determined such that the underlying goals are reached (e.g. low auxiliary energy consumption, optimum use of solar-generated electricity, conservative battery operation, satisfied users). The system controller provides supervisory control of all power system elements as well as battery protection. The battery charging sequence is inherently current limited by the capacity of the power system's components. Charging is a critical operation which directly affects the useful life of the battery. The primary objective of a charge control system is to charge the battery efficiently while avoiding the detrimental effects of excessive overcharging. From a system standpoint, excessive overcharging is generally undesirable, since it increases the cell temperature and represents an efficiency loss. A lack of adequate provision to terminate battery discharging can cause battery failure or critical its life. The following configurations ofthe basic stand-alone systems with their control strategy's illustrated in Figures 6.1 (a - 1) are arranged in the order of simple to more complex.
147
In the following control strategies the load is normally powered by the system otherwise the controller take another decision to disconnect it during the critical operation periods such as minimum battery SOC. 6.3.1. PVB with Shunt Regulator Configuration "A"
~
! Array String 1
oo
.. .. ............................... ..
Shunt Regulator
: :
;
Array String N
0
Voltage and
de Ioads
:~rature
Relay Controller 1· ·
I
.. . . .............
Battery ··
I
{)
Fig.6.1(a), PVB with Shunt Regulator system block diagram.
In this configuration, the battery charge/discharge controller is present via relay switching. This arrangement is commonly used in small stand-alone with simple battery chargers which provide both charge and discharge protection functions. The control strategy for this configuration is shown in the following chart.
available means it is able to power the load and charge the batte
es
Disconnect the load from the battery
Charge the battery
Monitor battery
soc
Disconnect the PVarrayto Stop charging.
Fig. 6.1(b), PVB with Shunt Regulator system control strategy. 148
6.3.2. PVB with Trickle Charging Technique Configuration "B"
Array String 1
oo
Array String N
..--------------··--
ac loads
Battery Controller
Fig.6.1(c), PVB with Trickle Charging system block diagram.
This configuration is used in large PV plants. The battery charge controller is present via voltage regulator by ON/OFF control with trickle charging resistor. This type of battery charge control has been used on large PV plants, but it is not recommended, especially for large batteries due to the losses in the power resistor. The control strategy for such systems is illustrated in the following.
Yes
Monitor battery
soc
Disconnect the load from the battery
Charge the battery Charge the battery directly from the through the trickle PV array charging resistor
Fig.6.1(d), PVB with Trickle Charging system control strategy.
149
6.3.3. PVB with de-de Converter Configuration "C" de-de Converter
ac loads Array String 1
0
o
Array String N
---------------- Battery
Fig.6.1(e), PVB with de-de Converter system block diagram.
In this configuration the battery charge voltage regulation is present via a series converter. The converter is used primarily for battery charge control with or without maximum power tracking; also, the converter may be either the buck or buckboost type. The control for this type is present in the following chart.
Monitor battery SOC
Stop the Inverter unit and disconnect the load
Charge the battery through the de-de converter, if the battery reach 95% SOC, eliminate the battery charging current to a minimum value
Fig.6.1(f), PVB with de-de Converter system control strategy.
150
6.3.4. PVB with Array Strings on/off Control Configuration "D" de Bus
acloads Array String 0 1
o
Array String N
Battery
Fig.6.1(g), PVB with Array Strings on/off Control system block diagram.
In this configuration, the battery charge voltage regulation is present by PV array strings control. This type of regulation has been used very successfully on medium to high power plants for de bus voltage levels up to 500Vde. The control strategy is shown as follows:
Monitor the battery SOC
Stop the Inverter unit and disconnect the load
Regulate the charging current by ON/OFF the PV array strings, if the battery reach 95% SOC, eliminate the charging current
Fig.6.1(h), PVB with Array Strings on/offControl system control strategy.
151
6.3.5. PVB with Partial Shunt Switching Capability Configuration "E"
l Inverter
~
*
PV String 1 "upper part"
00 PV StringN
"upper part"
IController PV String 1 "lower part"
0 0
partial shunt
-l Battery J
acloads
PV StringN "lower part" ~-:
partial shunt
'··- ......................
111--:
------------ --
Fig.6.1(i), PVB with Partial Shunt Switching system block diagram. In this configuration, the battery charge voltage control is present via partial shunt (Linear or Digital) regulation. This configuration is useful for medium- to highpower PV stand-alone systems. Note that the partial shunt can be a linear analog regulator or an ON/OFF switch (digital), and the system controller can be analog or digital
(i.e.,~
board, JlC or PC)
Monitor the battery
soc
Regulate charging by ON/OFF the partial-shunt sub-
Stop the Inverter & disconnect the load
Connect all the main & partial PV strings
Fig.6.1G), PVB with Partial Shunt Switching system control strategy. 152
6.3. 6. PVB with Diesel generator Backup Hybrid System Configuration "F" Inverter Il I ac 1oads Backup charge ................ :,;;jiil ~ ~-controller ... .._ Auxiliary PV Batteries ~ Rectifier Power Array I Source
I
I
r--
r..... :
Fig.6.1(k), PVB with Diesel generator Backup Hybrid System block diagram. This diagram illustrates an optimum configuration of a stand-alone PV hybrid system with an Auxiliary Power Source such as a small Diesel or Wind Generator. The control strategy of such hybrid systems with back-up source is illustrated in the following:
No
Monitor the battery Stop Diesel
soc
Diesel Regulate the charging current
Stop charging or charge with min. charging current
Fig.6.1(1), PVB with Diesel generator Backup Hybrid System control strategy. 153
The main key features of this configuration can be listed as follows: • Permits better battery charge maintenance at low rates which result in higher capacity acceptance, • Allows much utilization ofPV energy as compared to one without the rectifier, • Requires significantly lower PV array size due to the presence of the Diesel which could cover the load tasks in the critical operating periods such as minimum battery SOC; thus, lower cost.
Minimizing of the back-up generator running time means the battery is not completely charged often enough and therefore battery aging is accelerated. High voltages cause gassing and therefore neutralization of the acid stratification, and corrosion is accelerated. Common control strategies are realized through a constant end of charge voltage (sometimes temperature corrected), a constant end of discharge voltage (sometimes current corrected) and thresholds for the start and stop ofthe back-up generator. In reality complete charging of the battery with the back-up generator is avoided to increase efficiency of the back-up generator and to decrease long running times. To avoid these problems variable thresholds for all parameters of the control strategy are introduced and the end of discharge and back-up generator thresholds are given by the state of charge and not any longer by battery voltage. In the ac PV systems, a conventional fixed frequency 50-Hz inverter is used to feed ac loads continuously 24 hrs per day. The advantages of such systems is that offthe-shelf household ac appliances or equipment can be used as consumers.
6. 4. PV Pilot Plants The application of the previous different control strategies as stand-alone technique or a combination between different techniques are used in many PV pilot plants. In the following, a brief illustration of some PV pilot plants configurations and their relevant control strategies are discussed.
6.4.1. NASA (USA) PV Pilot Plant/41(1984) The NASA PV pilot plant is a PV-battery-Diesel generator hybrid power system to power the
telecommunication subsystem located in remote area. A
simplified schematic of the PV power system and loads is shown in Figure 6.2. The 154
major components of the power system are the system controller, PV array, battery, inverter, battery charger and the engine-generator. The primary mode of power system operation is for the array and battery to provide de power to the loads. The engine-generator serves as a backup power source in either oftwo methods. In the first method, the engine-generator and battery charger charge the batteries and power the loads concurrently. In the second method, the engine-generator powers the load directly.
r------,---,~--+--l
I
I
I
L...-."f,
~
~ERY) ------
