2013; MTI, 2014, Ebay, 2014). 171. Table 4.1: ..... MATLAB software, version. R2012b and implemented on a laptop computer (Window 8 Pro, 64-bits operating ...
MODELLING AND OPTIMIZATION OF A HYBRID ENERGY SYSTEM FOR GSM BASE TRANSCEIVER STATION SITES IN EMERGING CITIES
BY
OKUNDAMIYA, MICHAEL STEPHEN B.Eng., M.Eng, MNSE, R. Engr. (COREN)
MAT. NO.: PG/ENG0515052
DEPARTMENT OF ELECTRICAL/ELECTRONIC ENGINEERING FACULTY OF ENGINEERING UNIVERSITY OF BENIN BENIN CITY
FEBRUARY, 2015
MODELLING AND OPTIMIZATION OF A HYBRID ENERGY SYSTEM FOR GSM BASE TRANSCEIVER STATION SITES IN EMERGING CITIES
BY
OKUNDAMIYA, MICHAEL STEPHEN B.Eng., M.Eng, MNSE, R. Engr. (COREN)
MAT. NO.: PG/ENG0515052
A THESIS SUBMITTED TO THE SCHOOL OF POSTGRADUATE STUDIES IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE AWARD OF THE DEGREE OF DOCTOR OF PHILOSOPHY (Ph.D) IN ELECTRONIC AND TELECOMMUNICATION, DEPARTMENT OF ELECTRICAL/ELECTRONIC ENGINEERING FACULTY OF ENGINEERING UNIVERSITY OF BENIN BENIN CITY
FEBRUARY, 2015
CERTIFICATION This is to certify that this is the original work of Engr. Okundamiya Michael Stephen (PG/ENG0515052), submitted in partial fulfilment of requirements for the award of the Degree
of
Doctor
of
Philosophy
(Electronic/Telecommunication
option)
Electrical/Electronic Engineering, University of Benin, Benin City.
------------ Signed --------------
---------- 27/02/2015 ---------
Prof. (Mrs.) J. O. Emagbetere
Date
Chief - Supervisor
------------ Signed --------------
---------- 27/02/2015 ---------
Engr. Prof. E. A. Ogujor
Date
Co – Supervisor
------------ Signed --------------
---------- 02/03/2015 ---------
Engr. Prof. S. O. Igbinovia
Date
Head of Department
ii
in
CERTIFICATION OF THESIS ON PLAGIARISM We the undersigned attest and declare that the thesis of Engr. Okundamiya, Michael Stephen Tilted Modelling and Optimization of a Hybrid Energy System for GSM Base Transceiver Station Sites in Emerging Cities has successfully passed the anti-plagiarism test and does not violate any copyright regulations.
----- Signed (27/02/2015) ---Prof. (Mrs.) J. O. Emagbetere
--- Signed (27/02/2015) -Engr. Prof. E. A. Ogujor
Chief Supervisor/Sign & Date
Co-Supervisor/Sign & Date
----- Signed (02/03/2015) ---Engr. Prof. S. O. Igbinovia HOD/Sign & Date
ii
DEDICATION I dedicate this research work to the Almighty God, who is gracious and powerful in all things. God, who makes the impossible to become possible, gave me life and wisdom to accomplish my goal.
iii
ACKNOWLEDGEMENTS I am most grateful to God Almighty for His infinite love, wisdom, grace, understanding and divine favour given me throughout the course of study. I also express my immense gratitude to my supervisors, Prof. (Mrs.) J. O. Emagbetere and Engr. Prof. E. A. Ogujor for their prodigious guidance, comments and encouragement throughout the course of this research. In addition, my sincere appreciation goes to Prof. F.O. Edeko and his research team (Electronic/Telecommunication Research Team) and other members of staff for their valuable comments and suggestions in this work. Thanks to my dearly beloved wife and angel, Mrs. Oghogho Okundamiya, for her unfailing love, concern and understanding throughout the period of this work. I cherish and appreciate you dearly. To my parents, Pastor and Mrs. J. A. Okundamiya, I wholeheartedly appreciate you for your immense support spiritually, financially and otherwise. Finally, I acknowledge the financial support received from Ambrose Alli University through the Tertiary Education Trust (TET) Fund for Academic Staff Training and Development.
iv
ABSTRACT This study presents the modelling and optimization of a Hybrid Energy System (HES) for GSM Base Transceiver Station (BTS) sites in emerging cities. The aim is to ensure reliable and cost-effective power supply, considering the availability, dynamism and viability of energy sources. Theoretical approach is applied in the modelling, simulation and validation of the developed HES, which consists of the utility grid, wind and solar photovoltaic (PV) as primary energy sources incorporating a super-capacitor/battery storage and power conversion unit. The complexity in optimizing continuous variables of the HES informed the use of a hybrid Genetic Algorithm and Pattern Search (h-GAPS) technique. The optimization problem is treated as a single objective function by considering all objectives in terms of cost while constraining the HES to satisfy the load demand safely according to the reliability criteria defined by the energy management strategy. The h-GAPS based optimization model simulated for the peripheral node GSM BTS sites in Abuja, Benin City, Enugu, Ikeja, Maiduguri and Sokoto utilized long-term (22-years) meteorological data sets collected from the Nigerian Meteorological agency and the National Aeronautics and Space Administration. The performance index of various developed and existing energy systems is evaluated based on economy or Cost of Energy (COE), power system reliability, energy throughput, and emission reduction targets. Simulation results showed that Sokoto is the most favourable site for utilizing the proposed HES. Abuja and Benin City are the least favourable locations for utilizing the grid-connected (Grid/PV/Wind) and the off-grid (PV/Wind) configurations respectively. The optimum size v
of grid-connected HES consisting of 2 kW wind turbine, 7.09 m2 PV array inclined at 150, 0.053 kWh super-capacitor and 10.8 kWh (48V, 225Ah) battery banks, and 1,484.60 kWh of energy drawn from the grid per annum enabled reliable (negligible power loss) and costeffective energy supply in Sokoto. The off-grid configuration reduced the COE by 72.81% (N24.75 to N6.73 per kWh) but with larger PV array size (12.68 m2) and reliability of 99.02%. In comparison with current practice of using grid/diesel systems, the proposed offgrid configuration has the best performance index, with an average energy throughput of 0.076 kWh per naira, in Nigeria. A reliable and cost effective energy option will not only reduce the per-unit cost of mobile services in Nigeria, but also reduce the greenhouse gas emission level from GSM BTS sites by an average of 98.34% thereby making the environment much more friendly and safe. This research would be useful for mobile service providers, consultants, regulatory agencies, policy makers, and the society.
vi
TABLE OF CONTENTS TITLE PAGE
i
CERTIFICATION
ii
CERTIFICATION OF THESIS ON PLAGIARISM
iii
DEDICATION
iv
ACKNOWLEDGEMENTS
v
ABSTRACT
vi
LIST OF FIGURES
xiii
LIST OF TABLES
xvii
NOMENCLATURES
xx
CHAPTER ONE: GENERAL INTRODUCTION
1
1.1
Background to the Study
1
1.2
Statement of the Problem
6
1.3
Aim
9
1.4
Objectives of the Study
9
1.5
Methodology
9
1.6
Scope of the Study
11
1.7
Significance of the Study
12
1.8
Arrangement of the Thesis
13
CHAPTER TWO: LITERATURE REVIEW
15
2.1
Overview of GSM Base Transceiver Station Sites
15
2.2
Energy crises in Nigeria
18
2.3
Nigerian Grid Electricity System
21
2.3.1
Electricity Tariff System
24
2.3.2
Electricity Consumers Classification
25
vii
2.3.3 2.4
Environmental Implications
28
Renewable Energy Conversion System
31
2.4.1
Wind Energy Conversion System
32
2.4.2
Photovoltaic Conversion System
46
2.5
Energy Storage System
64
2.6
Optimization
68
2.7
2.8
2.6.1
Meteorological Data for Simulation
69
2.6.2
Statistical Analysis
71
2.6.3
Criteria for Hybrid Energy System Optimizations
73
2.6.4
Optimum Sizing Methods for Hybrid Energy System
80
2.6.5
Control Strategies for Energy Flow and Management
91
2.6.6
State-of-the-art in Optimization Methods for Hybrid Energy Systems 92
Research Area Climate
100
2.7.1
Sokoto
101
2.7.2
Maiduguri
102
2.7.3
Abuja
102
2.7.4
Ikeja
103
2.7.5
Enugu
103
2.7.6
Benin City
104
Summary
104
CHAPTER THREE: METHODOLOGY
108
3.1
Development of the Hybrid Energy System Model
108
3.2
Modelling of the Hybrid Energy System Components
111
3.2.1
Modelling of the Grid Energy Supply System (GESS)
111
3.2.2
Modelling of the Wind Energy Conversion System
114
viii
3.3
3.4
3.2.3
Modelling of the Photovoltaic Conversion System
118
3.2.4
Modelling of the Power Electronics (Conversion Unit)
123
3.2.5
Modelling of the Energy Storage Unit
124
Energy Management Strategy
129
3.3.1
Control Design
129
3.3.2
Operation Strategy
132
3.3.3
System Energy Characteristics
134
3.3.4
System Reliability Considerations
136
System Techno-Economic Analysis
137
3.4.1
Basic Considerations
138
3.4.2
Economic Analysis
140
3.4.3
Technical Analysis
146
3.4.4
Total Cost Analysis
147
3.4.5
Techno-Economic Viability
148
3.5
Environmental Impact Assessment
148
3.6
Estimation of Energy Consumption of GSM BTS Site
150
3.7
Data Collection and Analysis
151
3.7.1
Load Data
151
3.7.2
Wind Turbine Characteristic Curve
152
3.7.3
Meteorological Data
152
3.8
Model Performance Evaluation
155
3.8.1
Evaluation of various Global Solar Radiation Models for Nigeria
155
3.8.2
Evaluation of various Diffuse Solar Radiation Models for Nigeria
158
3.8.3
Determination of the Optimum Tilt Angle of a PV Array Oriented Due South in Nigeria
159
ix
3.8.4
Calibration and Validation of the Proposed Wind Energy Conversion System Model
3.9
160
Optimization Procedure
161
3.9.1
Formulation of the Optimization Problem
161
3.9.2
Optimization of the Proposed Hybrid Energy System
163
3.10 Design of Simulation Model 3.10.1
168
Case Studies: Process Simulation and Application
CHAPTER FOUR: RESULTS, DISCUSSION AND FINDINGS
169 173
4.1
Results
173
4.2
Validation of Developed Estimation Models
204
4.3
Validation of Tilt Angles
209
4.4
Performance Evaluation of Proposed Hybrid Energy System for GSM BTS Sites 4.4.1
211
Comparison of Proposed Energy System and Existing Energy Systems
214
4.4.2
System Environmental Impact
227
4.4.3
Land Requirement for Implementing the Proposed Hybrid Energy System for GSM BTS Sites
229
4.5
Findings
230
4.6
Contributions to Knowledge
234
CHAPTER FIVE: CONCLUSION AND RECOMMENDATIONS
235
5.1
Conclusion
235
5.2
Recommendations
237
5.2.1
238
Further Research
LIST OF RESEARCH PUBLICATIONS
240
REFERENCES
242 x
APPENDICES
280
Appendix A: Main Technical Specifications of Hybrid Energy System Components
280
Appendix B: Meteorological and Load Data
285
Appendix C: MATLAB Script for Model Calibration and Validation
297
Appendix D: MATLAB Script for Determining the Optimum Tilt Angles of PV Array 305 Appendix E: MATLAB Main Scripts for Optimization and Operation Control of Proposed Hybrid Energy System
305
xi
LIST OF FIGURES Figure 1.1: Trends of subscribers‘ base and teledensity in the period (2001 – September 2014) in Nigeria (NCC, 2014).
2
Figure 1.2: Market shares of GSM network operators as in September 2014 in Nigeria (NCC, 2014).
2
Figure 2.1: Funding of the Nigerian power sector in the last three decades (Tallapragada, 2009; Erik, 2011)
19
Figure 2.2: Trends of the Nigerian population within the last three decades (1980 – 2011)
20
Figure 2.3: Average duration of power access and outage of the Nigerian grid electricity system (UNDP-GEF, 2013)
23
Figure 2.4: Average duration of power access per day of the Nigerian grid electricity system (UNDP-GEF, 2013)
23
Figure 2.5: Variation of cost of grid-supplied electricity to a typical GSM BTS site for different locations across Nigeria.
28
Figure 2.6: Share of different anthropogenic GHG emissions, in total emissions in 2004, expressed in terms of CO2-eq (IPCC, 2007).
29
Figure 2.7: Comparison of the cubic law fitted WTG characteristics and that of the typical WTG supplied by the manufacturer for Hummer H3.1-1kW WTG (AHDC, 2013)
44
Figure 2.8: Effect of varying altitude on air density (Okundamiya and Nzeako, 2013) Figure 2.9: Map of Nigeria showing the study locations.
45 101
Figure 3.1: Architecture of the proposed hybrid energy system for electricity supply to GSM BTS sites
108 xii
Figure 3.2: Control design of proposed hybrid energy system
130
Figure 3.3: Control design of proposed micro-grid (stand-alone HES)
132
Figure 3.4: Optimization and operation control model for HES
165
Figure 3.5: GA toolbox parameter settings for the simulation model
168
Figure 4.1: Simulated minute load profile for an outdoor GSM BTS (S/4/4/4) site in Nigeria.
173
Figure 4.2: The Nigerian grid supply voltage profile simulated for a period of 1 year (a) Voltage magnitude (b) Power access/outage frequency.
174
Figure 4.3: Comparison of the measured and estimated monthly average daily global solar radiations on a horizontal surface (in the periods of 1984 – 2005) for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively
177
Figure 4.4: Comparison of performance indices for various global solar radiation models: (a) r-value, (b) RMSE, (c) MBE and (d) MABE.
178
Figure 4.5: Comparison of the monthly relative percentage error (RPE) of global solar radiation estimates from
(a) Angstrom–Prescott (1940),
(b) Badescu
(1999), (c) Chen et al. (2004), (d) El-Metwally (2004), (e) Falayi et al. (2008), and (f) Present study [Eq. (3.22)].
179
Figure 4.6: Comparison of the measured and estimated monthly average daily diffuse solar radiations (1984–2005) for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively
181
Figure 4.7: Comparison of performance indices for various diffuse solar radiation models: (a) r-value, (b) RMSE, (c) MBE and (d) MABE. Figure 4.8: A comparison of the monthly relative percentage error (RPE) of diffuse solar radiation estimates from (a) Page (1964), (b) Liu-Jordan (1960), (c)
xiii
183
Butt et al. (2010), (d) Karakoti et al. (2011), (e) Present study (Eq. 3.23), and (f) Present study (Eq. 3.24).
184
Figure 4.9: Comparison of annual total solar irradiance on a horizontal and tilted surface for various locations in Nigeria
189
Figure 4.10: A comparison of annual minutely global irradiance at annual optimum tilt angle for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu and (f) Benin City.
190
Figure 4.11: Comparison of the proposed WTG power profile and the manufacturer supplied power characteristics for (a) H3.1-1kW, (b) H3.8-2kW, (c) H4.63kW, and (d) H6.4-5kW WTGs respectively.
193
Figure 4.12: Adjusted minutely wind speed data to the GSM BTS tower height of 25 m for a typical year for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively
194
Figure 4.13: Hourly energy profile for Grid-PV/Wind HES in Sokoto: (a) Total energy generation, (b) Energy drawn from grid, (c) Solar energy generation, and (d) Wind energy generation.
199
Figure 4.14: SOC of SC/battery bank of proposed grid-PV/Wind energy system for a simulation period of one year for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively.
200
Figure 4.15: SOC of SC/battery bank of proposed stand-alone PV/Wind energy system for a simulation period of one year for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively.
201
Figure 4.16: Electrical characteristics of proposed stand-alone PV/Wind energy system for a typical day (February 28) for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively.
xiv
202
Figure 4.17: Electrical characteristics of proposed grid- PV/Wind HES for a typical day (September 21) for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively.
203
Figure 4.18: Comparison of the power supply reliability of proposed hybrid and conventional energy systems
215
Figure 4.19: Comparison of the economic performance of proposed hybrid and conventional energy systems
218
Figure 4.20: Comparison of the performance viability of proposed and conventional energy systems
220
Figure 4.21: Variation of energy performance index with power supply reliability for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City.
225
Figure 4.22: Comparison of the growth in cost saving and population till 2020
xv
227
LIST OF TABLES Table 2.1: The main characteristic of the Nigerian grid electricity system (UNDP-GEF, 2013).
24
Table 2.2: Variation of electricity tariff for mobile telecommunication companies for study locations (NERC, 2012)
27
Table 2.3: Global Warming Potential (IPCC, 1996)
30
Table 2.4: Weibull shape and scale factors for study locations in Nigeria (Ahmed et al., 2013)
35
Table 2.5: Power law exponents for different locations in Nigeria (Okundamiya and Nzeako, 2013)
41
Table 2.6: GHG emission factors for the grid electricity from various sources (Nandi and Ghosh, 2010; LGOP, 2010)
80
Table 3.1: Geographical classification and coordinates of selected sites in Nigeria.
154
Table 3.2: Economic specifications of components for optimization of the proposed hybrid energy system (Nandi and Ghosh, 2010; AHDC, 2013; SEDC, 2013; MTI, 2014, Ebay, 2014)
171
Table 4.1: Calibration results of various global solar radiation models (using group-1 data sets for a period of 22-years) along with R2 and t-stat values
175
Table 4.2: Validation results of various global solar radiation models using group-2 data sets for a period of 22-years
176
Table 4.3: Calibration results of different diffuse solar radiation models (using group-1 data sets for a period of 22-years) along with R2 and t-stat values
180
Table 4.4: Validation results of different diffuse solar radiation models using group-2 data sets for a period of 22-years xvi
182
Table 4.5: Result of analysis (based on HDKR model) of influence of annual-based optimum tilt angles for selected areas in Nigeria
185
Table 4.6: Result of analysis (based on HDKR model) of influence of seasonal-based optimum tilt angles for selected locations in Nigeria
186
Table 4.7: Result of analysis (based on HDKR model) of influence of monthly-based optimum tilt angles for selected locations in Nigeria
187
Table 4.8: Comparison of different optimization methods (using HDKR model) for tracking solar radiations in Nigeria
188
Table 4.9: Calibration results of proposed WTG model [Eq. (3.19)] using manufacturer‘s supplied data.
191
Table 4.10: Validation results for proposed WTG model
192
Table 4.11: Simulation results for proposed Grid-PV/Wind hybrid energy system for considered locations
195
Table 4.12: Simulation results for proposed stand-alone PV/wind hybrid energy system for studied locations
196
Table 4.13: The annual total energy composition of the developed Grid-PV/Wind HES
197
Table 4.14: The annual total energy composition of the developed stand-alone PV/Wind HES
198
Table 4.15: Comparison of Energy Composition of Grid-PV/Wind HES by fraction for different sites
216
Table 4.16: Comparison of Energy Composition of stand-alone PV/Wind HES by fraction for different sites
216
Table 4.17: Comparison of the costs of energy (COE) of various energy systems
217
Table 4.18: Comparison of proposed system techno-economic viability for various reliability limits
222 xvii
Table 4.19: Comparison of environmental impact of conventional and proposed energy systems
228
xviii
NOMENCLATURES ∆T
Temperature difference (oC)
A
Area (m2)
apm
Maximum coefficient of rated performance
apr
Coefficient of performance at rated wind speed
B
Temperature lapse rate (K m-1)
bi
Model parameters/regression coefficients
Cˆ
Monthly average daily total Cloud cover during daytime observations (octa)
c
Present cost penalty per unit size (N unit-size-1)
cj
Present cost of component j per unit size (N unit-size-1)
ccj
Cost coefficient of component j per unit size (N unit-size-1)
CO2-eq
Carbon footprint (t)
D, G
Coefficients
E
Total energy (kWh yr-1)
E (τ)
Energy at time τ (kWh)
ef,
Emission factors (g kWh-1)
xix
em,
~ fm
Emission (t)
Modulating function
g
Gravitational acceleration (m s-2)
H
Monthly average daily global radiation on a horizontal surface (kWh m-2 day-1)
h
Hour (h)
Hc
Monthly average clear sky daily global radiation on a horizontal surface (kWh m-2 day-1)
HD
Monthly
average
daily
diffuse
radiation
on
a
horizontal
surface
(kWh m-2 day-1) Ho
Monthly average daily extraterrestrial radiation on a horizontal surface (kWh m-2 day-1)
I
Average hourly global solar radiation on a horizontal surface (kW m-2)
I0
Hourly extraterrestrial radiation on a horizontal surface (kW m-2)
Ib
Average hourly beam solar radiation on a horizontal surface (kW m-2)
Id
Average hourly diffuse solar radiation on a horizontal surface (kW m-2)
It
Hourly global irradiance incident on a tilted surface (kW m-2)
k
Shape factor
K
Monthly average daily diffuse coefficient xx
KD
Monthly average daily diffuse fraction
KT
Monthly average daily clear index
L
Lifetime (yr)
l
Scale factor (m s-1)
m
Months of the year
min
Minutes
N
Number
nf
Noise factor
ns
Outcomes of supply
ol
Moving average function at lag l
P
Power (kW)
pdf
Probability density functions
Ps
Pressure (Pa)
Py
Probability
Q
Quantity
r
Coefficient of correlation
R
Gas Constant (J kg-1 K-1)
xxi
R2
Coefficient of determination
rd
Self-discharge rate (% day-1)
ȓe
Escalation rate
R g
Geometric ratio
RH
Monthly average daily relative humidity (%)
ȓi
Interest rate
ȓf
Inflation rate
S
Monthly average sunshine duration (h)
So
Monthly average daylight sunshine duration (h)
Sz
Size
T
Air temperature (oC)
T*
Absolute temperature (K)
tcp
Temperature coefficient of power (% 0C -1)
t-stat
T-statistic test
uh
Coefficient of heat transfer/loss to surroundings (kW m-2 °C -1)
ul
Autocorrelation function at lag l
v
Wind speed (m s-1)
xxii
v
Standardized wind speed (m s-1)
V
Voltage (V)
Xp
Expectation
yr
Year
z
Altitude (m)
Z, B
Matrix
Greek letters µ
Average
ħf
Horizon brightening factor
ƛi
Anisotropy index
α
Solar absorbance
β
Surface inclination or tilt angle (o)
γ
Azimuth or surface orientation (o)
δ
Solar declination (o)
ε
Noise function
η
Efficiency
θz
Zenith angle of incidence (o) xxiii
λ
Power law exponent
ξ
Solar transmittance of any cover over the PV array
ρ
Air density (kg m-3)
ρg
Ground reflectance or Albedo
σ
Standard deviation
σ2
Variance
τ
Simulation time (s)
ϕ
Latitude (o)
ψη
Wavelet function
ω
Hour angle (o)
ωs
Sunset Hour Angle (o)
Subscripts ac
Alternating current
ah
Anemometer height
ann
Annual
ave
Average
bat
Battery
xxiv
bb
Battery bank
bf
Battery float-life
cf
Coupling factor
ci
Cut-in
cl
Cell
co
Cut-out
con
Converter
cr
Critical
d
Demand
dc
Direct current
def
Deficit
ec
Economic
en
Environmental
es
Energy storage
exc
Excess
fc
Fixed charge
ga
Grid access
xxv
gc
Grid consumption
ge
Grid electricity
gec
Grid energy contribution
grd
Grid
gs
Grid supply
hh
Hub height
hor
Horizontal
hs
Hybrid system supply
ic
Input of converter
inc
Inclination
ini
Initial
int
Interconnection
inv
Inverter
lt
Lifetime
max
Maximum
mea
Measured
min
Minimum
xxvi
mo
Mode of operation
mod
PV Module
mp
Maximum power
norm
Normalized
op
Operation
out
Output
p
Power
pk
Peak
pred
Predicted
R
Ratio/Relative
rat
Rated
re
Renewable energy
rec
Renewable energy contribution
rel
Reliability
rep
Replacement
ret
Rectifier
sal
Salvage
xxvii
sc
SC bank
se
Solar energy
sg
Solar generation
sta
Stabilizer
STC
Standard Test Conditions
sv
Salvage value
sys
System
te
Technical
tec
Total energy contribution
tp
Throughput
un
Unmet
we
Wind energy
wec
Wind energy contribution
wg
Wind generation
wt
Wind turbine
xxviii
Acronyms 3G
Third Generations
AC
Alternate Current
ACO
Ant Colony Optimization
ANN
Artificial Neural Network
AR
Autoregressive
ARMA
Autoregressive Moving Average
BTS
Base Transceiver Station
CER
Commission for Energy Regulation
COE
Costs of Energy
COI
Cost of Investment
CRF
Capital Recovery Factor
D
Decision variable
DC
Direct Current
DE
Differential Evolution
DML
Digital Mobile Licenses
DOD
Depth of Discharge
xxix
ECN
Energy Commission of Nigeria
EPSR
Electricity Power Sector Reform
erf
Error function
ES
Energy Storage
FME
Nigeria Federal Ministry of Environment
GA
Genetic Algorithm
GEF
Global Environment Facility
GESS
Grid Electricity Supply System
GHG
Greenhouse Gas
GMS
Global Mobile Statistics
GSM
Global System for Mobile communication
GUI
Graphic User Interface
GWP
Global Warming Potential
HDRK
Hay-Davies-Klucher-Reindl
HES
Hybrid Energy Systems
HESS
Hybrid Energy Storage System
HEV
Hybrid Electric Vehicle
h-GAPS
hybrid Genetic Algorithm and Pattern Search xxx
HOGA
Hybrid Optimization by Genetic Algorithms
HOMER
Hybrid Optimization Model for Electric Renewable
HYBRID2
Hybrid power system simulation model
IITA
International Institute of Tropical Agriculture
IPCC
Intergovernmental Panel on Climate Change
ISA
International Standard Atmosphere
LCC
Life-Cycle Cost
LCE
Levelized Cost of Energy
LCS
Levelized Cost of System
LLP
Loss of Load Probability
LLPS
Loss of Load Power Supply
LOLH
Loss of Load Hours
LOLP
Loss of Load Probability
LPSP
Loss of Power Supply Probability
MA
Moving average
MABE
Mean Absolute Bias Error
MATLAB
Matrix Laboratory
xxxi
MBE
Mean Bias Error
MDGs
Millennium Development Goals
MOEA
Multi-Objective Evolutionary Algorithm
MTN
Mobile Telephone Network
MTS
Mobile Telecommunications Services
MYTO
Multi-Year Tariff Order
NASA
National Aeronautics and Space Administration
NCC
Nigerian Communications Commission
NCEEC
National Centre for Energy Efficiency and Conservation
NERC
Nigerian Electricity Regulatory Commission
NESI
Nigerian Electricity Supply Industry
NET
Nigerian Electricity Tariff
NIMET
Nigerian Meteorological agency
NITEL
Nigerian Telecommunications Limited
NOCT
Nominal Operating Cell Temperature
NP
Non-deterministic Polynomial-time
NPC
Net Present Cost
xxxii
NREL
National Renewable Energy Laboratory
OM
Operation and Maintenance
pCj
Total present cost of component j in system life
PHCN
Power Holding Company of Nigeria
PS
Pattern Search
PSO
Particle Swarm Optimization
PV
Photovoltaic
PVCS
Photovoltaic Conversion System
PW
Present Worth
REC
Renewable Energy Contribution
RERL
Renewable Energy Research Laboratory
RET
Renewable Energy Technologies
RMSE
Root Mean Square Error
RPE
Relative Percentage Error
SC
Super-Capacitor
SEC
Solar Energy Contribution
SHS
Solar Home System
xxxiii
SOC
State of Charge
SOV
State of Voltage
SOVR
Safe Operating Voltage Region
SPEA
Strength Pareto Evolutionary Algorithm
SPL
System Performance Level
TCN
Transmission Company of Nigeria
TEC
Total Energy Contribution
TPV
Total Present Value
TRNSYS
Transient energy System Simulation tool
TS
Tabu Search
TMY
Typical Meteorological Year
UNDP
United Nations Development Programme
WCED
World Commission on Environment Development
WECS
Wind Energy Conversion System
WTG
Wind Turbine Generator
xxxiv
Units/formula C
Capacitance
CH4
Methane
CO2
Carbon-dioxide
exp
Exponential
F
Farad
HFCs
Hydro fluorocarbons
J
Joules
K
Kelvin
k
Kilo
kg
Kilo-gram
kV
Kilo-volts
kW
Kilo-watt
kWh
Kilo-watt-hour
M
Mega
m
Metre
N
Naira
xxxv
N2O
Nitrous Oxide
NOx
Nitrogen-Oxides
Pa
Pascal
PFCs
Per fluorocarbons
s
Seconds
SF6
Sulphur hexafluoride
SO2
Sulphur-dioxide
t
Tonnes
US$
US dollars
xxxvi
CHAPTER ONE 1.0 1.1
INTRODUCTION Background to the Study
The history of the Nigerian Telecommunication Industry can be divided into three phases (Okere,
2012).
The
first
(sluggish)
phase
began
from
1985,
when
Nigerian
Telecommunications Limited (NITEL) was established till the time when the Nigerian Communications Commission (NCC) was established in 1992. The second (fast) period spanned through 1992 to 2000. This period witnessed deregulation agenda that led to the launch of the National Policy on Telecommunications. The third (supersonic) phase commenced in 2001 and continued to the present day. Mobile cellular services started on the Nigerian market in 1993 with the introduction of Mobile Telecommunications Services (MTS). The MTS was a joint-venture between NITEL and Digital Telecommunications of Atlanta. The two companies offered voice services, voicemail, and paging, from three switches (in Abuja, Enugu, and Lagos), with a joint subscriber base of 12,500. The auction for Digital Mobile Licenses (DML) conducted by NCC in January 2001 brought about the emergence of three mobile operators. These companies are Econet Wireless now Airtel, Mobile Telecommunications Network (MTN) and MTel (a subsidiary of NITEL) while a fourth DML was issued to Globacom in 2002 (Baez et al., 2010). In order to foster market competition a fifth Mobile License with GSM spectrum was awarded to Etisalat, in 2012. Through the award of DML and subsequent introduction of unified access service licenses in February 2006 (which replaced all existing licenses), NCC facilitated a phenomenal expansion of subscriber from about half a million connected lines in 2000 to over one hundred and thirty-four million connected lines as in September 2014 as shown in 1
Figure 1.1 (NCC, 2014). In addition, several licensed operators (including Airtel, Etisalat, Globacom, and MTN) have launched third-generation (3G) services in Nigeria. Figure 1.2 shows the market shares of Global System for Mobile communications (GSM) network
140
100
126
90
112
80
96
70
84
60
70
50
56
40
42
30
28
20
14
10
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year Subscribers
Teledensity
Subscribers (in millions)
operators in Nigeria as in September 2014 (NCC, 2014).
0
Teledensity
Figure 1.1: Trends of subscribers‘ base and teledensity in the period (2001 – September 2014) in Nigeria (NCC, 2014).
19.8%
Airtel
44.2% 15.1%
Etisalat Globacom
20.9%
MTN
Figure 1.2: Market shares of GSM network operators as in September 2014 in Nigeria (NCC, 2014). 2
The mobile telecommunication industry is one of the fastest growing sectors of the global economy. The ever-increasing demand for mobile telecommunication services in developing countries is the main driving force for this significant economic growth. Nigeria currently has African‘s largest mobile telecommunication market with a teledensity of over 96 telephone lines per 100 people as in September 2014 (GMS, 2013; NCC, 2014). The growth in the Nigerian telecommunications sector since the inception of the GSM in 2001 is unmatched by any other sector. It has recorded a phenomenal growth both in terms of subscribers‘ base and infrastructural development in the country. The rapid growth of mobile telecommunications creates a number of problems such as network congestion and poor quality of service delivery. These problems are fast eroding the gains of the Nigerian mobile telecommunication sector (Emagbetere, 2008; Ajanaku, 2013). The operating companies are unable to increase their networks fast enough to meet the ever growing demand by subscribers in most parts of Nigeria due to lack of reliable utility grid and the cost implication of a supplementary energy source. Nigeria‘s grid electricity supply is characterized by high unreliability index (Ogujor, 2007). Most GSM base transceiver station (BTS) sites in Nigeria rely heavily on the use of fossil-fuelled generators either as supplements to the utility grid or exclusively in remote locations. The operation and maintenance of fossil-fuelled generators accounts for about 78% of the total cost of operations (equivalent to about 35% of the total cost of ownership) of the GSM BTS sites (Adegoke and Babalola, 2011). The use of fossil-powered solution at GSM BTS sites presents a number of economic, logistic, and environmental challenges (Martinez-Diaz et al., 2013). Studies by Kovats et al. (2005), VandeWeghe and Kennedy (2007) and Rahmstorf (2008) have shown that the earth‘s climatic change is a result of increasing concentrations of greenhouse gases (GHG) being released primarily from fossil fuel combustion into the 3
atmosphere. The higher the unreliability of grid power supply, the higher the demand for fossil-powered solution. Besides, the current and future demand patterns of energy are not sustainable (Oyedepo, 2012). Energy sustainability is of immense importance given the pervasiveness of energy use, the significance in economic development and standard of living, and the significant impacts that energy processes and systems have had, and will continue to have, on the environment (Oyedepo, 2012). The World Commission on Environment Development (WCED) classified the sustainability of energy generation system in three dimensions: economic, social and environmental (WCED, 1987). Sustainable energy option provides accessible, affordable and reliable energy services that improve the socio-economic and environmental standards within the overall developmental context of the society while recognizing equitable distribution (Davidson, 2002). The growing need for energy sustainability has made alternative energy sources a promising research area. The world trend in recent times is to find a clean and nondepleting source of energy, with little or no adverse effects on the environment. The increasing interest in energy saving and environmental protection has led to the extensive utilization of green (renewable) energy sources (Oliveira, 2007; Vine, 2009; Evans et al., 2010). The green energy resources and technologies are essential components of sustainable development mainly because of the following reasons (Oyedepo, 2012). First, they are more eco-friendly than other sources/technologies as such the extensive utilization of the green option can help in making the environment more green and safe. Second, they are nonexhaustible and if properly utilized in the appropriate application they can supply reliable and sustainable power almost indefinitely. Third, they favour system decentralization and local
4
solutions that are somewhat independent of the utility grid. This property enhances the flexibility of the green energy system, providing enormous benefits to small isolated populations. Moreover, the prospects of green technologies in sustainable energy development in Nigeria are enormous (Oyedepo, 2012). Among various technologies for green energy generation in the literature, wind and solar energy are commonly used because they are both technically and environmentally viable options (Zhou et al., 2010). In addition, they are ubiquitous and freely available. Nevertheless, energy storage devices used in cancelling out unpredicted power fluctuations, stabilizing voltage and frequency, and improving power quality form an integral part of the renewable energy system since it is highly dependent on weather conditions (Okundamiya et al., 2014a). Typical storage devices are battery bank, Super-Capacitor (SC) bank and fuel cell system (Wang and Nehrir, 2008; Fang et al., 2011). Hybrid Energy Systems (HESs) are widely used in recent times because it combines different power sources to maximize each source's strengths while compensating for the others' shortcomings (Klychev et al., 2007). Modelling of a Hybrid Energy System (HES) is a promising approach necessary for the efficient design, control, and management of energy resources. In addition, it is an essential step in the system design process before any phase of optimum sizing (Zhou et al., 2010). The modelling approach differs from one article to another, and there is no standard method for providing reliable results. However, the information regarding the operating performance of the system components play a significant role in determining the optimum control strategy for the system. Optimum control strategy is required for efficient management of the power flow between the different energy sources, storage device, and the distribution for reliably satisfying the energy demand (Kornelakis and
5
Marinakis, 2010). The control allows proper exchange of energy among system components, thereby enhancing the system‘s performance at optimum cost. The hybrid energy system if optimally designed and deployed for GSM BTS sites, can be more reliable and cost effective than any single energy source. A reliable and cost-effective energy solution can result in the global expansion of GSM BTS sites to meet the ever growing demand by subscribers in emerging cities. The modelling of reliable, cost-effective and greener alternative energy system is a necessary step taken in the right direction, to addressing the energy related problems and thereby making the environment much more friendly and safe. The system will not only enhance the quality of service delivery but also could increase the mobile market penetration and reduce the cost of cellular mobile services in emerging cities. 1.2
Statement of the Problem
Nigeria has been experiencing an extreme electricity shortage for over two decades. Only about 40% of Nigerians are connected to the national electric power network (utility grid) (Ogujor and Orobor, 2010). In addition, the utility grid is characterized by high unreliability index. The power supply reliability varies from 39 to 66 % with an average duration of power access between two power outages of 4.5 h per day. At any period when grid power is available, the supply voltage fluctuates mostly between 160V and 205V (UNDP-GEF, 2013). The inadequacy of the utility grid has consistently led to load shedding, with adverse effects on domestic, commercial and industrial activities (Oseni, 2011). As a result, most entrepreneurs have resorted to using fossil-powered sources. The socio-economic and environmental implications associated with the use of fossil-fuelled generators can be very alarming. Compared to the utility grid, the cost of electric power 6
generation from fossil-fuel generators is significantly higher. The operation and maintenance of diesel generators accounts for about 78% of the total cost of operations of the GSM BTS sites (Adegoke and Babalola, 2011). Consequently, the use of fossil-fuel generator as an alternative source of electric power for GSM BTS sites increases the unit cost of cellular mobile services in emerging cities. Besides, the environmental consequences of harnessing and utilizing fossil fuels are assuming alarming proportions (Kovats et al., 2005; VandeWeghe and Kennedy, 2007; Rahmstorf, 2008). The most threatening of these problems, perhaps, is the greenhouse effect, where certain pollutants prevent the sun‘s radiation from escaping from the atmosphere. This problem has the potential of creating dangerous climate changes, with devastating effects on certain species and the ecosystem. A typical example of such impact is the recent flooding in Nigeria, which claimed lives, displaced citizens and destroyed properties estimated worth over billions of naira. The hybrid energy systems are capable of providing the needed energy in the cellular mobile telecommunication sectors, but critical issues such as reliability and the enabling technologies are yet to be resolved (Zhou et al., 2010; ElNozahy and Salama, 2013). The dynamic interaction between the utility grid, green energy sources, and energy demand can lead to system instability. Instability can severly affect the reliability of a power system. It is worthy of note that the operational specifications of green energy systems are location dependent. In addition, green energy technologies have higher initial cost of investment compared to conventional (fossil) technology. As a result, potential investors are at a cross road on the choice of system design configuration, optimum specifications, capacity projections and the techno-economic implications. The key performance indicators that could influence investor‘s decision on the most suitable technology or its combinations are increasingly lacking in the literature. In addition, green energy solutions are not commonly used in GSM 7
BTS sites in developing cities presently. Consequently, the quest for better alternative (hybrid) energy systems and the enabling technology for a sustainable economic development prompted this present research. There are several methods and tools in articles reviewed for sizing HES. The assumption of an ideal grid distribution system and the subsequent neglect of a grid supply model commonly characterized grid-connected hybrid system design analysis. The computer-based model such as the Hybrid Optimization Model for Electric Renewables (HOMER) is a frequently used simulation tool for designing HES (NREL, 2010). HOMER failed to account for uncertainties, such as power outages and unreliable supply, which characterizes the grid electricity distribution in the emerging world. In other words, HOMER assumes that the utility grid always supplies the energy demand at any given time when renewable energy is unavailable (or when there is a shortfall). However, for regions characterized by high unreliability index, such as Nigeria, the grid electricity is not always available (Ogujor, 2007). A robust model to account for the uncertainties is required in determining the correct optimum configurations of such an energy system. Moreover, accurate prediction of the renewable energy potential of a given place depends on accurate determination of the intensity of the renewable resources at that location. So far, the global solar radiation data are not measured at the 45 meteorological stations in Nigeria. Within the limits of articles discussed no article modelled and analyzed the grid-connected hybrid (wind-photovoltaic) system, considering the uniqueness of the grid electricity supply, for developing regions. Consequently, the gap in knowledge has motivated the present research.
8
1.3
Aim
This study aims to model and optimize a hybrid energy system for reliable and cost-effective power supply to GSM base transceiver station sites in emerging cities. 1.4
Objectives of the Study
The objectives of this study are stated as follows: 1. Development of estimation models for predicting solar radiation in Nigeria. 2. Modelling the dynamics and operation strategy for hybrid energy sources for GSM BTS sites in Nigeria. 3. Development of performance index for assessing energy generation system. 4. Collection and evaluation of GSM BTS load and meteorological data for Nigeria. 5. Calibration and validation of developed models. 6. Development of the simulation model for optimum sizing of a hybrid energy system. 7. Determination of the economic, technical and environmental viability of the proposed hybrid energy system. 1.5
Methodology
The methods adopted to achieve the set objectives for this study are as follows: 1. Regression theory is applied to develop empirical models for estimating solar radiation on a horizontal surface. 2. Theoretical approach is applied in the modelling, simulation and validation of the designed hybrid energy system, which consists of five units (grid supply system, wind energy conversion system (WECS), solar photovoltaic (PV) conversion system, energy storage system and the energy management system).
9
3. On the basis of the method (2) above, the crucial parameters that have a significant impact on the energy throughput of the system are reliability and cost. Reliability as defined here is the percent of electricity demand that the power system can deliver. Therefore, the proposed energy index is deduced in terms of reliability and energy cost. 4. Power profiles supplied by the Anhui Hummer Dynamo Company Limited (Anhui Province, China) for Hummer WTGs are utililzed for developing the proposed WTG model. Load data of a typical outdoor GSM BTS sites in Benin City were collected from Etisalat, Nigeria. The long-term (22-years) meteorological data used were obtained from two data sources for 37 sites (the 36 states capital cities and Abuja) in Nigeria. The first part collected from the archives of the National Aeronautics and Space Administration (NASA) spanned through 1984 to 2005. The second part was collected from the archives of the Nigerian Meteorological (NIMET) agency, Oshodi, Lagos State, for the period 1991–2012. The data collected are analyzed and evaluated using stochastic methods. 5. The developed models are calibrated using linear and non-linear regression techniques. On the basis of seven statistical indicators (coefficient of correlation, coefficient of determination, mean bias error, mean absolute bias error, relative percentage error, root mean square error, and t-statistic test) the applicability of deduced models for the selected locations are determined. 6. Computer program is written for simulation of the proposed hybrid energy system model, designed to power GSM BTS sites consuming up to 2 kW of power. The program applied the Genetic Algorithms (GA) toolbox in MATLAB software, version R2012b and implemented on a laptop computer (Window 8 Pro, 64-bits operating
10
system with processor speed of 2.8GHz and 4.0GB RAM). The program determines the fixed optimum tilt angles for the installation of the PV array based on the HayDavies-Klucher-Reindl (HDKR) model proposed by Duffie and Beckman (2006). And adjusts the wind speed data to the GSM BTS tower height using power law (Hocaoglu et al., 2009a, b). The formulated optimization problem is solved using a hybrid Genetic Algorithms and Pattern Search (h-GAPS) method. The proposed hybrid energy system simulation is performed with a time step of 1 min, for a period of one year (525,600 min). 7. Sensitivity analysis is performed on the system by varying system design parameters. Four key performance indices (power supply reliability, economy or cost of energy, viability, and emission reduction) are applied to evaluate the overall performance of the proposed hybrid energy system over existing (diesel-only, grid-diesel backup and grid-only) energy generation systems. 1.6
Scope of the Study
This study is limited to the use of three energy sources (utility grid, wind and solar). The utility grid is considered where available because it is the current source of electricity in Nigeria. In contrast, wind and solar are both technically and environmentally viable options. The dynamic nature of these energy sources is considered only for emerging cities. An emerging city is a rapidly growing city, with impressive infrastructure and economic gains. Investors consider such a city attractive because of the potential for high returns. Six areas in Nigeria (Abuja, Benin City, Enugu, Ikeja, Maiduguri, and Sokoto) are selected as case studies. A location is randomly chosen from each of the six geopolitical zones in Nigeria. These sites cut across the different climatic zones in Nigeria. The hybrid energy systems discussed in this study are designed to supply power to outdoor BTS sites consuming 11
up to 2 kW of power continuously. The outdoor BTS sites consume lesser energy compared to traditional (indoor) GSM BTS sites. It is worthy of note that the use of energy efficient infrastructure is essential to attaining green energy solution. 1.7
Significance of the Study
The study focused on the use of the concept of energy sustainability for solving the technoeconomic problem of mobile telecommunication sector in emerging cities. This method is in line with global best practices in mitigating the devastating effects of global warming on the earth. It suggests an optimum design configuration of a hybrid energy system that can efficiently replace the existing fossil-fuelled power system for cheaper and cleaner production of electricity with lower emission targets. The simulation of the proposed hybrid energy system for GSM BTS sites in Nigeria, under different climatic conditions, demonstrates the efficacy of the proposed approach. This study would: 1. Inform decision making on the use of natural renewable energy resources on a much larger scale than what it is presently. 2. Improve the quality of GSM mobile services in Nigeria, by making the operation of GSM BTS sites more reliable. 3. Reduce the unit cost of cellular mobile services (voice and data). 4. Reduce greenhouse gas emissions of mobile telecommunication sites and thereby making the environment greener and safe. 5. Assist the government in developing policy guidelines for the provision of reliable and sustainable energy supply.
12
6. Speed up the implementation of policy guidelines, which would facilitate the enforcement of relevant environmental protection laws for attaining sustainable development in Nigeria. 1.8
Arrangement of the Thesis
The research work embodied in this thesis comprises of five chapters. Chapter 1 has been discussed. Chapter 2 examines the GSM BTS infrastructure and the current electricity situation in Nigeria in addition to, renewable energy technologies and methodologies for optimum sizing of HESs. It considers the prospects of energy resource availability and challenges for the deployment of existing technologies in emerging economies using Nigeria as a case study. It emphasizes on underlying principles of existing theories and concepts in sufficient details to indicate shortcomings in the application of various techniques for Nigeria. The objective is to enable the development and deployment of sustainable energy systems in Nigeria. Chapter 3 provides a detailed description of the methods adopted for modelling the proposed hybrid energy system. The focus is on useful factors or system components and both usually neglected, which are vital in determining accurate optimum system configuration in the modelling of an efficient and reliable energy system for GSM BTS sites in emerging cities. This chapter is organized into ten sections. The first section presents the methods adopted for designing the proposed hybrid energy system. Next, the modelling of the system component units, the energy management strategy, and the techno-economic analysis in determining the system‘s cost effective components that can reliably satisfy the load demand are presented. The environmental impact assessment and the energy consumption model of GSM BTS sites are given in sections five and six respectively. The seventh section discusses the method of
13
data collection and analysis for case study locations. The eighth section evaluates the performance of developed models as well as the calibration and estimation of model parameters and the validation of developed models. Section nine details on the methods and procedures utilized for optimum sizing of proposed hybrid energy system. The chapter ends with a description of the simulation process. Chapter 4 presents an analysis of the results. It compares the performance of proposed hybrid energy system with existing energy generation systems using key performance indices, such as economy, power supply reliability, energy throughput, and emission reduction targets. The findings and contributions to knowledge are highlighted. Chapter 5 gives the conclusion and recommendations for further study. Five appendices are included in the thesis. Appendix A shows the main technical specifications of components while appendix B provides the load and meteorological data used for the study. Appendix C presents the computer program for model calibration and validation. Appendix D gives the computer program for determining the optimum tilts of PV array. The main computer program for optimization and operation control of proposed HES is shown in appendix E.
14
CHAPTER TWO 2.0 2.1
LITERATURE REVIEW Overview of GSM Base Transceiver Station Sites
A recent estimate indicates that there are about 20,000 GSM Base Transceiver Station (BTS) sites in Nigeria with a corresponding annual growth rate of approximately 20% from 2007 to 2012 (Ajanaku, 2013). In other developed countries of the world, for instance, the United Kingdom, there are over 52,500 GSM BTS sites, yet Nigeria‘s population is more than twice as large as that of the United Kingdom. According to the 2011 census (Wikipedia, 2014a; 2014b), the population of Nigeria is 162,471,000.0, extending over a total area of 923,768.0 km2 (with a landmass of 98.6%) compared to the United Kingdom‘s population of 63,181,775.0 with a total area of 243,610.0 km2 (and a landmass of 98.7%). This figure clearly demonstrates the need for a rapid infrastructural development in the Nigerian mobile telecommunication sector. There are two types of BTS sites on cellular mobile networks (Phelan, 2014). The first type exists on the periphery, which can only handle communications for their cell. In contrast, the second category, which is stationed on the core network, can handle communications for multiple cells. A typical GSM BTS site consists of the tower on which the antenna and lighting systems are mounted, and shelter as shown in Plate 2.1. The height of a typical micro-cell tower used for cities (with coverage of 100m – 1km) is 25m. Conversely, the height of macro-cell tower used in remote sites (with coverage of 1 – 35km) varies between 34 – 55m (Emagbetere, 2008). The shelter consists of power amplifier, transceiver, combiner, duplexer, control function, alarm extension system and cooling systems. In terms of energy
15
consumption, the elements of a BTS site consist of two groups (Lorincz et al., 2012). These are radio-frequency equipment and support system. The radio-frequency equipment, which includes power amplifiers and transceivers, serves one or more sectors/cells. The support system consists of Alternate Current/Direct Current (AC/DC) power conversion modules, airconditioning elements, analogue and digital signal processors, and battery backup.
Plate 2.1: Pictorial view of a typical GSM BTS site The radio-frequency consumes over 80% of the total energy consumption (Huawei, 2012). Power amplifier is the largest energy consumer. It has a share of around 65% of the total energy consumption (Chen et al., 2010). Other base station components with significant energy consumption are air-conditioning (17.5%), digital signal processing (10%) and AC/DC conversion systems (7.5%). The key issue is how to reduce the energy consumption of the GSM BTS while guaranteeing quality service delivery and increased service availability for users. The problem has led to the concept of the green-mobile technology, a process that utilizes energy efficiency technology. Investments in energy efficiency could result in long-term gains such as reduced energy consumption, local enhancement of 16
ecosystem and overall economic development (Cristobal, 2011; Oyedepo, 2012). The installation of energy efficient infrastructure is a pre-requisite for attaining ‗greener solution‘; with gradual replacement of an inefficient conversion process at existing sites with an efficient one (Okundamiya et al., 2014b). Nevertheless, the main challenge for investors is the upgrading of the existing infrastructure while remaining profitable, i.e., the replacement of the inefficient energy conversion system with an efficient system (Banos et al., 2011; Oyedepo, 2012). Several authors including Qiang (2008), Roy (2008), Zoican (2008) and Chen et al. (2010) applied various methods to reduce the energy consumption of GSM BTS sites. One of the methods adopted is the use of an advance-power-magnifier technology such as a highfrequency digital power amplifier. A digital pre-distortion technique can be used for cancelling the distortion in the power amplifier, and therefore achieving better linearity (Hirata et al., 2010). Improvements in power amplification and the introduction of intelligent energy saving techniques will not only decrease the power consumption of the GSM BTS site but can also make the site less dependent on air conditioning. Further measures employed by GSM vendors to reduce energy consumption of BTS sites include the design of BTS based on the actual local climate conditions. The eco-friendly BTS use direct and intelligent ventilation as well as heat exchange for heat dissipation in the equipment rooms. The power consumption of a typical eco-friendly GSM BTS is 1.0 kW, but depending on the numbers of carriers, it can consume as much as 2.0 kW (Jianguo et al., 2008). Hassan et al. (2011) examined various techniques for improving the power efficiency of cellular networks such as efficient resource management, low energy spectrum sensing, energy-aware medium access control and routing, and cross-layer design. The objective was
17
to reduce both the operational costs and the environmental effects. The study emphasized on various strategies for reducing the energy consumption of the BTS and recommended the use of green (wind and solar) energy sources for electricity generation. Lack of an enabling technology perhaps is a major hindrance to the use of the renewable energy solution in mobile telecommunication industry in Nigeria presently. Unlike Nigeria, the Indian government is currently implementing a proposal for subsidizing network operators (companies) using renewable sources for powering their telecommunication infrastructure. The subsidy is expected to enable the country achieve her goal of exclusively powering remote GSM BTS sites and 50% reduction of urban diesel-generator-powered GSM BTS sites with green technology by 2015. Bangladesh government has taken similar initiative to install one million Solar Home Systems (SHSs) by 2015 ( ̴ 50 MW) which will prevent about 84,000 tons of CO2 per year (Nandi and Ghosh, 2010). This target will ensure optimum development of renewable energy sources; environmentally sound sustainable energy development causing minimum damage to environment, public and private sector participation in the development and promotion of competition among the entrepreneurs. Chen (2011) suggests that the energy policy maker should create an environment that can motivate the development of clean energy supply and utilization for the achievement of energy policy objectives. 2.2
Energy crises in Nigeria
Energy is a valuable input for economic and social development. Electric power is a fundamental requirement for meeting the Millennium Development Goals, and for improving the living standard of the citizenry. It plays a significant role in the socio-economic and technological development of the country. Specifically, electric power is an important need for developmental services such as, the provision of food, pipe borne water, health care, 18
conducive abode, and quality education among others. It serves as a source of power to the telecommunication industry for the utilization of their equipment, serves as an important tool for lighting, heating and other domestic activities (Oseni, 2011). The Nigerian power sector experienced insufficient funding between the early and late 90s as shown in Figure 2.1. An investigation report on the Nigerian electricity sector presented by Tallapragada (2009) showed that there were underinvestment and poor electricity infrastructure planning during 1981-1999, with an average daily production of 1,750 MW. For over a decade (1989-1999), no new infrastructure was installed. Below 2% of the Transmission Development Plan was implemented during 1995 – 2005. And no transmission line was built from 1988 – 2005. In addition, the existing power infrastructure is mostly in a dysfunctional state. As a result, the growth in electricity generation in Nigeria has not been enough to accommodate the increasing demand for electricity. About 40% of Nigerians have access to electricity (Ogujor and Orobor, 2010). The long bout of poor infrastructural development translates into the present electricity crises besetting the country (Nigeria).
Figure 2.1: Funding of the Nigerian power sector in the last three decades (Tallapragada, 2009; Erik, 2011)
19
Nigeria is Africa‘s most populous country, with a population above 177 million people (IWS, 2014). The trend of the Nigerian population within the last three decades is shown in Figure 2.2 (Wikipedia, 2014a). As observed, Nigeria has witnessed over 128% increase in her population within this period (1980 – 2011). Based on the United Nation‘s projections, Nigeria is one of the eight countries that is expected to account collectively for half of the world's total population increase from 2005–2050 (Wikipedia, 2014a).
Population (in million)
180 150 120 90 60 30 0
1980 1984 1988 1992 1996 2000 2004 2008 2012 Year
Figure 2.2: Trends of the Nigerian population within the last three decades (1980 – 2011) As the country‘s population grows and its economy expands, the demand for electricity multiplies. Despite the importance of electricity in the economy, Nigeria has not been able to generate adequate and reliable electricity to meet its growing demand (Oseni, 2011). The inadequate supply of electricity has consistently led to load shedding, with adverse effects on domestic, commercial and industrial activities. Power unreliability is a major setback to ensuring complete coverage in the telecommunication industry. Most entrepreneurs have resorted to using fossil-fuelled generators either as supplements to the national grid or exclusively in remote sites. The use of these generators in industries has contributed to high
20
cost of energy. The cost of energy production in Nigeria, compared to China, is nine times higher, and this has crippled the industrial sector (Aliyu et al., 2013). 2.3
Nigerian Grid Electricity System
The Nigeria Electricity Supply Company (NESCO), which commenced operations in 1929 in an attempt to connect all parts of the country and ensure secure electricity supply, has undergone so many transformations and reforms. It was renamed National Electric Power Authority (NEPA) in 1972. NEPA was known to have the burden of subsidies and low service quality (Aliyu et al., 2013). The Electric Power Sector Reform (EPSR) Act introduced in 2005, led to the transformation of NEPA into Power Holding Company of Nigeria (PHCN). PHCN was subsequently split into 18 companies, which consist of 6 generators, 11 distributors and one transmission company. The structure of the PHCN was to oversee the activities of the Managing Director/CEOs of the successor company for a period of 5 years, which should be adequate time to enable the companies to be privatized (Sambo et al., 2010). These companies are saddled with the responsibility of carrying out functions relating to the generation, transmission, trading, distribution in addition to bulk supply and resale of electricity in Nigeria (Oseni, 2011). On the other hand, Nigerian Electricity Regulatory Commission (NERC) regulates and monitors the Nigerian Electricity Supply Industry (NESI). Despite several attempts made by the Nigerian Government, in recent times, in the provision of power infrastructure, the performance of the power sector has remained weak in comparison with other developing economies. In 1993, the World Bank assessed the performance of energy development in Nigeria (World Bank, 1993). The purpose was to compare the Nigeria‘s power sector with that of 20 other developing countries around the
21
globe. The study showed that the sector had the highest percentage of system losses (33 – 41%), the lowest generating capacity factor (20%), average revenue ($1.56 kWh-1) and rate of return (8%). Oyedepo (2012) performed a similar study on energy for sustainable development in Nigeria. The study reported the existence of a total installed power generation capacity of 8702 MW (77.73% thermal and 22.27 % hydro), available capacity of 4825 MW (72.93% thermal and 27.07% hydro), and an operational capacity of 3149 MW (68.2% thermal and 31.8% hydro) with an availability factor of 0.55 in Nigeria. The findings, which are based on PHCN 2009 Annual Report, indicate that the utility grid has a downtime of about 11 hours daily. An independent study carried out by Ogujor (2007) reported a value of 87,639 minutes of the annual average total duration of electric power supply interruption to a customer in Nigeria. This index is relatively high as compared to other countries of the world such as USA, Singapore and France, with a value of 88, 61.5 and 52 minutes respectively (Ogujor and Orobor, 2010). A recent study conducted by the United Nations Development Programme (UNDP) with support from other stakeholders showed that the Nigerian electricity grid is not able to ensure a permanent supply of electricity (UNDP-GEF, 2013). The findings of the study are that power outage varies from 34% for Lagos (south-western zone) to 61% for Sokoto and Bauchi (north-western and north-eastern zones respectively), with an overall power access average of 55% across Nigeria as shown in Figure 2.3. The duration of electricity access per household varies from 2 – 24 hours with an average of 13 hours as shown in Figure 2.4 while the power outage duration per household varies from 1 – 15 hours. The supply voltage during power access periods varies from 162 to 242 volts, with an average of 204 volts across Nigeria. The main characteristic of the Nigerian grid electricity system is shown in Table 2.1.
22
100%
80%
61
37
34
36
41
45
63
66
64
59
55
Abuja
Lagos
Enugu
Benin City
NIGERIA
61
60%
40%
20%
39
39
Sokoto
Bauchi
0%
Power Access
Power Outage
Figure 2.3: Average duration of power access and outage of the Nigerian grid electricity system (UNDP-GEF, 2013)
Figure 2.4: Average duration of power access per day of the Nigerian grid electricity system (UNDP-GEF, 2013)
23
Table 2.1: The main characteristic of the Nigerian grid electricity system (UNDP-GEF, 2013). Characteristics
Sokoto
Bauchi
Abuja
Lagos
Enugu
B/City
NIGERIA
Power access (%)
39
39
63
66
64
59
55
Minimum
172
186
178
183
174
162
172
Maximum
230
240
225
228
226
242
242
Average
203
203
208
212
202
172
204
Minimum
2.5
3.5
5.5
9
9
9.5
2.5
Maximum
16
14
21
24
21
23
24
Average
9.5
9.5
15
16
15
14
13
Minimum
2
2
1
0
1
1
0
Maximum
10
12
7
15
8
9
15
Average per outage cycle
5
5
3
4
3
4
4
Minimum
2
2
2
3
3
2
2
Maximum
5
5
13
12
8
11
13
Average
3
3
5
6
5
6
4.5
Average voltage during power access (V)
Number of hours households get electricity per day (h day-1)
Average duration of power outage (h)
Average duration of power access between two power outages (h)
2.3.1 Electricity Tariff System Electricity tariff refers to the amount of money the consumer has to pay for making electric power available to them at their homes. Electricity tariff system takes into account various factors, such as the maximum power demand and the total energy consumed, to calculate the total cost of electricity. There are various tariff systems discussed in the literature (Otokpa and Ajeibi, 2013). However, in response to the prevailing electricity problems in Nigeria, the Nigerian Electricity Regulatory Commission (NERC) developed a Multi-Year Tariff Order (MYTO) 24
based on industrial revenue requirements that will aid pricing of electricity in the most methodical and transparent manner (NERC, 2012). It is multi-year because these tariffs are set over a number of years. The MYTO 1, approved by the Federal Government of Nigeria in April 2008, sets tariffs for electricity customers for a five-year period (2008 – 2013) at a time while providing a 15-year (2008 – 2023) projection on the evolution of tariff with time. The objective was to ensure appropriate pricing for electricity that can encourage investors for sustainable growth in the energy sector. The Nigerian Electricity Regulatory Commission established by section 31 of the Electricity Power Sector Reform (EPSR) Act 2005, regulators and monitors the Nigerian Electricity Supply Industry. It is incumbent on the Regulatory Authority (NERC) to ensure strict adherence to the tariff regime by all industry players, carrying out reviews and calibration of the model annually for five years. A new tariff order (MYTO 2) was adopted in June 1, 2014. 2.3.2 Electricity Consumers Classification Customers, for purpose of charging, are categorized according to common shared characteristics such as connection to a particular voltage level, type or nature of customer (i.e. domestic, commercial etc.), and load factor/specific. In the more established markets, the number of tariff categories typically covers such groups as domestics, small commercial and industrial, larger customers by voltage class, unmetered supplies, time-of-use, and maximum power demand (CER, 2004). The grid electricity customers in Nigeria are classified on the basis of the MYTO as residential, commercial, industrial, special and street lighting customers (NERC, 2012) as discussed below. Residential: This class of energy consumers use their premises exclusively as a residence house, flat, or multi-storeyed building. There are four sub-divisions, viz. R1 (Life-line or 50 25
kWh), R2 (Single and 3-phase), R3 (Low voltage maximum demand or 400/230 V) and R4 (High voltage maximum demand or 11/33 KV). Commercial: This group of energy consumers use their premises for any purpose other than exclusively as a residence or as a factory for manufacturing goods. Therefore, if you do not run a factory or use a building as a place of residence, you fall under this category. Examples are offices, cyber cafes, business centres, salons, etc. The sub-divisions are C1 (Single and 3phase), C2 (Low voltage maximum demand) and C3 (High voltage maximum demand). The mobile telecommunication companies are examples of the Commercial C2 electricity consumers. Industrial: This class of energy consumers use their premises for manufacturing goods including welding and ironmongery. Factories and industries with high power demand are under this category with the following sub-divisions: D1 (Single and 3-phase), D2 (Low voltage maximum demand) and D3 (High voltage maximum demand). Special: These are customers such as agriculture and Agro-allied industries, water boards, religious houses, government and teaching hospitals, government research institutes and educational establishments. The subdivisions are A1 (Single and 3-phase), A2 (Low voltage maximum demand) and A3 (High voltage maximum demand). Street Lighting: This has been classified as S1 (Single and 3-phase). The variation of electricity tariff for mobile telecommunication companies, based on MYTO2 (NERC, 2012), for the study locations in Nigeria is shown in Table 2.2. The grid electricity tarrif presented in Table 2.2 are the revised energy rates with took effect from June 1, 2014.
26
Table 2.2: Variation of electricity tariff for mobile telecommunication companies for study locations (NERC, 2012) Unit Cost of Electricity
Monthly Fixed Charge
(N kWh –1)
(N Month–1)
Sokoto
26.24
21,260.90
Maiduguri
21.11
21,942.90
Abuja
22.08
47,771.54
Ikeja
21.54
23,814.00
Enugu
22.87
22,141.18
Benin City
19.79
34,020.00
Study Locations
Table 2.2 indicates that the cost per kWh of electricity varies from one location to another. The change perhaps is due to the variation in the value of different energy sources. Figure 2.5 shows a comparison of the present cost of grid supplied electricity (COE) per kWh for a typical GSM BTS site for the study locations. The cost computation used the electricity tariff presented in Table 2.2 is based on the following assumptions.
1. A typical GSM BTS site consumes an average of about 2.0 kW. 2. Grid electricity is available for an average of 13 hours daily (UNDP-GEF, 2013).
27
90 80
COE (Naira per kWh)
70 60 50 40 30 20 10 0
Sokoto
Maiduguri
Abuja
Ikeja
Enugu
Benin City
Locations
Figure 2.5: Variation of cost of grid-supplied electricity to a typical GSM BTS site for different locations across Nigeria. 2.3.3 Environmental Implications The inadequacy of the grid power supply has made electricity consumers resort to various forms of power sources to meet their needs. One of such prominent source is the fossil fuelled generator. Fossil fuels combustion has a severe impact on the ecosystem. In recent times, emissions from fossil fuel combustion have added to the concentration of greenhouse gases in the atmosphere causing a trend of global warming (IPCC, 1996; 2007). The earth‘s climatic change is the result of increasing concentrations of greenhouse gases resulting primarily from fossil fuel combustion into the atmosphere (IPCC, 1996; 2007; Kovats et al., 2005). The greenhouse gasses, referred to as Kyoto gases, are either carbon dioxide (CO2) or non-CO2 gases (Pandey et al., 2011). The non-CO2 gases consist of methane (CH4), nitrous oxide (N2O) and fluorinated gases, such as hydrofluorocarbons (HFCs), perfluorocarbons (PFCs) and sulphur hexafluoride (SF6). The percentage contribution (relative share) of 28
greenhouse gases, according to the Intergovernmental Panel on Climate change (IPCC) is presented in Figure 2.6.
Florinated gases 1.1%
N2O 7.9%
CH4 14.3% CO2 (Others) 2.8%
CO2 (Fossil fuel use) 56.6%
CO2 (deforestation, decay of biomass, etc.) 17.3%
Figure 2.6: Share of different anthropogenic GHG emissions, in total emissions in 2004, expressed in terms of CO2-eq (IPCC, 2007). Carbon dioxide accounts for about 76.7% of the total GHG emissions, which typically originates from fossil fuels use, deforestation and decay of biomass. Methane accounts for about 14.3% of GHG emissions. Methane gas sources are primarily from livestock, manure management and rice cultivation in wetlands. Nitrous oxide mainly produced by agricultural land use accounts for about 7.9% of GHG emissions. Fluorinated gases represent about 1.1% of GHG emissions. Fluorinated gases are synthetic gases, and they are produced from industrial processes. However, when fossil fuels such as, diesel and or natural gas, are used for grid electricity generation, the greenhouse gases emitted are CO2, CH4 and N2O (Nandi and Ghosh, 2010).
29
The effect of each type of GHG emission on global warming differs due to the difference in the physical and chemical properties of gases (Pongthanaisawan and Sorapipatana, 2013). Hence, their potential effects, referred as global warming potential (GWP), on the atmospheric temperature differs as shown in Table 2.3. GWP factors represent the ratio of the heat-trapping ability of each greenhouse gas relative to that of carbon dioxide. Table 2.3: Global Warming Potential (IPCC, 1996) GHG Emissions
GWP (g CO2/g substance)
CO2
1
CH4
21
N2O
310
SF6
23,900
The global warming of the earth poses severe threats to the ecosystem, and the most vulnerable is the coastal region (Awosika et al., 1992). A large percentage of Nigeria‘s urban population (for example Benin City, Lagos, and Port Harcourt) lives in coastal cities. In addition, most of the economic activities situated within the coastal zone form the backbone of the national economy. Nigeria is already experiencing adverse impacts of climatic changes on agriculture, power generation and tourism. A typical example is the recent flooding in the country, which claimed lives, displaced citizens and destroyed properties (estimated over billion dollars). The effects could severely affect almost all sectors of the economy negatively if unmitigated. Thus, a clarion call for a sustainable energy solution is a way forward to achieving the Millennium Development Goals in eradicating poverty. Sustainable energy 30
solution is a pre-requisite to achieving better coverage and enhanced quality in the services provided to subscribers by Telecommunication industries in Nigeria. 2.4
Renewable Energy Conversion System
The devastating effect of climate change, which relates to the emission of carbon into the atmosphere, has led to the introduction of the green technology. The goal of the green technology is to attain negligible emissions of carbon dioxide, which constitute by far, the largest part of the emissions of GHG, thereby making the environment much more friendly and safe (Okundamiya and Nzeako, 2013). Reducing energy consumption is the first strategy to reducing the dependence on fossil fuel. A typical approach involves applying energy savings programs focused on energy demand reduction and energy efficiency in both industrial and domestic applications (Vine, 2009). The second strategy consists of extensive utilization of renewable sources for energy systems. Renewable energy technology is capable of alleviating the already overstretched ecosystem; it can supply the energy required for rapid development (Cristobal, 2011). The prospects of renewable energy resources in sustainable energy development in Nigeria are enormous (Oyedepo, 2012). In spite of the availability of abundant renewable energy resources (including, hydro, solar, wind and biomass), such renewable power solutions are not commonly used in GSM BTS sites in Nigeria presently. The under-utilization of the renewable technology in Nigeria may not wholly be due to unawareness of the vast potential of available resources but also on technological constraints such as the enabling technology for deployment (Okundamiya et al., 2014b). The renewable energy system design, usually, integrates renewable energy mixes, such as biomass, wind and solar. Inauspiciously, large area of land, water usage and social impacts often characterize the electricity generation from biomass and this requires further study to
31
verify the techno-economic viability of its power generation (Evans et al., 2010). Consequently, it may be required to shift demand to other power sources, such as wind and solar. Wind and solar energy are ubiquitous and freely available. They are commonly used sources of renewable energy generation because they are both technically and environmentally viable options. 2.4.1 Wind Energy Conversion System Wind is one of the most viable and promising sources of renewable power globally. Accurate estimate of wind speed distribution, selection, operational strategy and management of the wind turbines are essential factors that affect the wind energy potential (Banos et al., 2011). The first steps a utility company considers when deploying wind as a power source is to examine the available wind speed (Gao, 2006). The next step is to modify the wind speed data at anemometer height to wind turbine hub height using appropriate conversion ratio. The adjustment of the wind profile is necessary to account for the effects of wind shear inputs. Moreover, accurate assessment of wind power potential at a site requires detailed knowledge of the wind speeds at different heights (Al-Abbadi and Rehman, 2009; Rehman et al., 2010; Rehman and Al-Abbadi 2010). A literature survey undertaken for assessing the viability of wind energy conversion systems (WECS) showed that limited work is available in this specific field (Zhou et al., 2010). One of the primary interests in WECS relates to the optimal design of the wind turbines. In particular, two problems, usually, considered include optimal design and placement of the wind turbines. Some authors have centred their interest in optimizing the turbine settings in order to maximize their performance. Jensen et al. (2010) provided an overview of techniques applied to the optimum design of wind turbine blades. Hameed et al. (2009) gave
32
an overview of the techniques, methodologies and optimization algorithms developed for monitoring the performance of Wind Turbine Generators (WTGs) and early error detection to avoid catastrophic conditions due to sudden breakdowns. Conversely, the placement of the WTG involves determining the optimum positions of the turbine in order to maximize the power output (Rasuo and Bengin, 2010). Kusiak and Song (2010) proposed a multi-objective evolutionary algorithm for WTG placement based on wind distribution with the aim of maximizing the wind energy captured while minimizing an index, which determines constraint violation. 2.4.1.1 Wind Speed Time Series Model The generation of the synthetic series of wind speed data has traditionally adopted a variety of methods, classified either as physical or statistical, depending on their input data (Carapellucci and Giordano, 2013). Physical models take into consideration several factors, including shelter from obstacles, pressure, temperature, and local surface roughness effects. On the other hand, statistical models establish a relationship between statistical parameters and wind speed data. Compared to physical methods, statistical methods are, usually, simpler and provide accurate results with reduced computational effort (Ernst et al., 2007). Statistical methods based typically on probabilistic and stochastic models for the synthetic generation of wind speed time series. Probabilistic Models The commonly used models for generating independent and identically distributed random numbers are Normal and Weibull probability distribution functions. However, both types are independent of wind speed time series. The wind speed time series is of high dependence. It
33
is essential for accurately estimating the energy output of a wind turbine (Billinton et al., 1996; Carapellucci and Giordano, 2013).
Normal distribution: Hourly average wind speed time series are generated by using a sequence of independent random numbers from a normal distribution. According to Aksoy et al. (2004), the normal probability distribution function is stated as follows.
(v ) 2 f (v ) exp 2 2 2 , 1
(2.1)
where v is the hourly average wind speed (m s-1), µ is the average value of the wind speed (m s-1), and σ is the standard deviation of the wind speed (m s-1). A number of computational methods are available for generating random numbers with normal probability distribution of average (µ) and standard deviation (σ). Weibull distribution: According to Aksoy et al. (2004), the Weibull distribution is defined by the following equation.
f (v )
1 k k 1 v exp k v lk l k
v 0; l , k 0 ,
(2.2)
where v is the hourly average wind speed (m s-1), l is the scale factor (m s-1), and k is the shape factor. These factors can be determined by using either a graphical method or the method of moments. Table 2.4 shows the shape and scale factors for study locations in Nigeria (Ahmed et al., 2013).
34
Table 2.4: Weibull shape and scale factors for study locations in Nigeria (Ahmed et al., 2013) Locations
Abuja
Benin City
Enugu
Ikeja
Maiduguri
Sokoto
k
4.13
10.07
8.12
3.79
5.47
6.02
l (m s-1)
4.12
4.60
5.80
5.91
5.90
7.69
Stochastic Models In stochastic approach various models can be used, including Autoregressive Moving Average (ARMA), Markov chain and the Wavelet-based approach (Aksoy et al., 2004). Autoregressive Moving Average models: Autoregressive Moving Average models are a group of linear stochastic models, classified into three categories. These are the purely autoregressive (AR) models, the moving average (MA) models, and the mixed (ARMA) models that combine autoregressive and moving average processes (Carapellucci and Giordano, 2013). AR processes are useful tools in generating data sets in cases where ―persistency‖ exists. ―Persistency describes the tendency that large values tend to be followed by large values, and small values by small values, so that runs of values of similar magnitude tend to persist throughout the sequence.‖ First-order AR models: The first-order autoregressive [AR(1)] model accommodates the effect of the previous value in the series in which the observed sequence of wind speed data {v1, v2,. . ., vη,. . .} is used to fit a model of the form (Sahin and Sen, 2001):
v
n
u v i 1
i
i
(2.3) 35
where v is the hourly average wind speed (m s-1), u is the autoregressive coefficient, that is, the model parameter, and ε is the normally distributed independent random variable (m s-1). In other words, it is a normal white noise process with average of zero and variance of σ2. A white noise process is a random process of random variables that are uncorrelated, with an average of zero, and a finite variance. The simplest case of Eq. (2.3) is obtained for n = 1, which is also called the Markov model defined by the equation (Dukes and Palutikof, 1995);
v u1 v 1
(2.4)
where v is the standardized (average of zero and unit variance) version of v (m s-1), u1 is the lag-one serial correlation coefficient of the sequence, and ε (m s-1) is the random component in AR(1) which is of normal distribution with average of zero and a variance of 1 – u12. The simulation procedure for the processes is very simple. It requires only a random number of a normal distribution to be generated. Second-order AR model: The second-order autoregressive [AR(2)] model is preferred to AR(1). The AR(2) model is more important in cases where dependence in the data set is obvious, as in the hourly average wind speed data. AR(2) is formalized as follows (Sfetsos, 2000).
v u1 v 1 u 2 v 2 ,
(2.5)
where autoregressive coefficients u1 and u 2 are given by the following equations.
u1 u1 (1 u 2 ) /(1 u12 )
(2.6)
u 2 (u 2 u12 ) /(1 u12 )
(2.7) 36
where u1 is the lag-one autocorrelation coefficient and u2 the lag-two autocorrelation coefficient of the wind speeds time series. The random component in AR(2) is again of the normal distribution, with an average of zero and variance equal to 1 - U 2, where
u12 u 22 2u1u 2 U . 1 u12 2
(2.8)
The ARMA series model is defined by the following equation (Li and Billinton, 1991).
v u1 v 1 u 2 v 2 u n v n o1 1 o 2 2 o m m ,
(2.9)
where u i ( i =1,2,…,n ) and o j ( j =1,2,…,m ) are the auto-regressive and moving average parameters of the model respectively. εη is a typical white noise process with an average of zero and variance of ζ2. As opposed to the probabilistic approach, ARMA models take into account the fact that hourly wind speeds are independent of each other. In other words, they incorporate the chronological nature of the actual wind speed, but with a degree of persistence defined by a typical autocorrelation function (Carapellucci and Giordano, 2013). These models have been applied successfully in several studies (Gao, 2006; Castellanos and Ramesar, 2006; Philippopoulos and Deligiorgi, 2009; Erdem and Shi, 2011). The main disadvantage is the high complexity for estimating the model coefficients, order and parameters. Markov chain model: The application of the Markov chain model involves the discretization of the stochastic process in a number of states and the definition of the probabilities for the inter-state transition. The transition probability matrix of a first-order Markov chain with m states is defined by Eq. (2.10) (Aksoy et al., 2004).
37
Py11 Py 21 Py ... Py m1
Py12
...
Py 22
...
Py ij Py m 2
Py1m Py 2 m Py mm
... ...
(2.10)
where Pyij is the probability of transition from state i to state j. If nij is the total number of hours of observation in state j with the previous state i, the probabilities of transition from state i to state j can be calculated as follows (Aksoy et al., 2004);
Py ij
nij
nij
i, j 1, 2, ..., m .
(2.11)
j
Aksoy et al. (2004) discussed the procedure for generating the simulated hourly average wind speed time series using the Markov chain approach. In first-order Markov chain, the current state of wind speed can only be determined from the known previous state, whereas two previous states are used for determining the current state of wind speed in the second-order Markov chain. Hourly wind speed generated based on Markov chain models is widely studied. The focus is to assess potential improvements achievable by varying the transition matrix order and state size (Shamshad et al., 2005; Hocaoglu et al., 2009a). The main disadvantage of this method is the possible loss of information with respect to recorded data. This problem is related with the discretization process and the large number of parameters, i.e. the inter-state transition probabilities (Papaefthymiou and Klockl, 2008). This limitation called for the development of an alternative approach (Carapellucci and Giordano, 2013). Wavelet-based approach: A wavelet is a real or complex-value continuous function with zero mean and finite variance, such as the Haar wavelet defined by the equation (Kitagawa and Nomura, 2003); 38
1 1 0
0 1 / 2 1 / 2 1
(2.12)
otherwise
The Wavelet-based approach was proposed as a non-parametric data generation tool. The idea behind this method is the decomposition of the data sample into its signals which are then reconstructed by randomly adding the signals to generate new data (Aksoy et al., 2004). This technique was first applied to wind speed forecasting by Hunt and Nason (2001), who attempted to model the wind time-series in terms of a multi-scale wavelet decomposition of data collected at a reference site. Detailed description of the wavelet generation algorithm is available in Aksoy et al. (2004). Other Approaches The development of artificial intelligence techniques and other forecasting methods has made a number of alternative approaches accessible for predicting wind data. These include Artificial Neural Network (ANN), fuzzy logic methods, support vector machine and some hybrid methods (Damousis et al., 2004; Mabel and Fernandez, 2008; Fadare, 2010; Gong and Shi, 2010). The common feature of all the methods discussed so far is the need for a set of historical data for calculating their parameters (ARMA models), constructing the transition matrices (Markov chain model) and defining the rules (ANN methods). However, most wind speed databases only provide average and maximum wind speeds on a monthly or yearly basis and lack information on the autocorrelation properties of wind speed data. This led to the development of an alternative method for synthetic generation of hourly wind speed time series, using readily available statistical input data, such as average and maximum wind
39
speed on a monthly or yearly basis (Carapellucci and Giordano, 2013). The objective was to construct a model for synthetically generating hourly wind speed data, adopting a physical– statistical approach. The proposed hourly wind speed model is defined as follows.
1 1 vt l ln erf 2 2
' vnorm
2
1 k
(2.13)
where vη is the wind speed time series (m s-1), l is the scale factor (m s-1), k is the shape '
factor, erf is an error function, vnorm is the normalized time series of the wind speed (m s-1), and µ (m s-1) and ζ (m s-1) is the mean and standard deviation of the observed wind speed respectively. As opposed to the numerous methods reported in the literature, the proposed approach claims to adapt to a different number and type of available input data being able to preserve the different characteristics of the measured wind speed data. 2.4.1.2 Adjustment of Wind Speed Data One of the methods extensively used in literatures for adjusting wind speed data is the power law expressed as stated (Diaf et al., 2007; Ardakani et al., 2009; and Hocaoglu et al., 2009a, b):
v hh
where
z v ah hh z ah
,
(2.14)
is the power law exponent, zhh is the hub height of the WTG (m), zah is the
anemometer height (m), vhh is the wind speed at WTG hub height (m s-1), and vah is the wind speed at anemometer height (m s-1). Bechrakis and Sparis (2000) gave the power law exponent values for different types of terrain. The power law exponent cannot be generalized
40
based on the measurements at one site because wind potential is highly site dependent and local topographical features influence it strongly. However, Okundamiya and Nzeako (2013) applied methods suggested by Al-Abbadi and Rehman (2009), Rehman et al. (2010), and Rehman and Al-Abbadi (2010) to estimate the power exponent (for open land with short grasses) for different regions in Nigeria as shown in Table 2.5. Table 2.5: Power law exponents for different locations in Nigeria (Okundamiya and Nzeako, 2013) Locations
Power law exponents (λ)
Abuja
0.1476
Benin City
0.1481
Katsina
0.1477
2.4.1.3 Wind Turbine Generator Models Wind turbines have different output power curves. As such, there are different models used to describe the performance of wind turbines. Yang and Aydin (2001) carried out a theoretical study on windmill and developed a revised model for the power density and the output power of the WTG determined using the basic WTG model defined by the equation:
Pwt ( )
1 p Awt v3 , 2
(2.15)
where ηp is the power efficiency, ρ is air density (kg m-3), Awt is rotor swept area of a single WTG (m2) and vη is the wind speed at time η (s).
41
Malinga et al. (2003) studied the dynamics and controlled the wind turbine as a distributed resource using the basic WTG design and a relationship for the power efficiency of the WTG proposed by Justus (1978). According to Justus (1978), for any operation pitch angle, a good approximation to ηp as a function of the speed is found using the following equations:
p
2 3 v max v max a pm 1 D1 1 D2 1 v v
3 v rat 3 v
p a pr
vci v v rat ,
v rat v vco ,
(2.16)
(2.17)
where vci is the cut-in wind speed (m s-1), vco is the cut-out wind speed (m s-1), vrat is the rated wind speed (m s-1), apr is the coefficient of performance at the rated wind speed, apm is the maximum coefficient of the estimated performance, vmax is the maximum wind speed (m s-1), and D1 and D2 are coefficients, which can be determined using boundary conditions. The power output characteristics of a WTG are quite different from those of the theoretical model of
Eq. (2.15) as such cannot correctly represent the output power of a WTG under
practical conditions. Several studies have attempted to model the power output of the WTG for practical applications based on assumed power characteristics. Lu et al. (2002) and Koutroulis et al. (2006) assumed that the wind turbine power curves have linear and quadratic characteristics respectively. However, Diaf et al. (2007), Yang et al. (2007), Ardakani et al. (2009), Hocaoglu et al. (2009a, b), Dagdouguia et al. (2010) and Nafeh (2011) considered a cubic characteristic and modelled the output power of a WTG according to the cubic law defined by Eq. (2.18) and estimated the rated power using Eq. (2.19).
42
0 3 3 v vci Pwt (v) Prat 3 3 v rat vci 1
Prat
For v vci & v vco
For vci v v rat
,
(2.18)
For v rat v vco
1 3 p Awt vrat , 2
(2.19)
where vci is the cut-in wind speed (m s-1), vco is the cut-out wind speed (m s-1), vrat is the rated wind speed (m s-1), and vη is the wind speed at time η (s). As indicated by the cubic law model of Eq. (2.18), the WTG provides power (Pwt) when v is higher than vci, produces rated power (Prat) when v varies between vrat and vco, and shut down when v is higher than vco. Although these models give a better prediction of the power output characteristics of a WTG compared to those of the underlying theoretical model of Eq. (2.15), their power output characteristic is only an approximation (as shown in Figure 2.7) to the actual power produced by the WTG. In most cases, the real WTG output power does not vary with a cubic law. Consequently, the WTG model should be developed according to its power output performance curve supplied by the manufacturer.
43
Fitted profile
Typical profile
1.4 1.2 Pw t (kW)
1.0 0.8 0.6 0.4 0.2 0
0
5
10
15
20
-1
v (m s )
Figure 2.7: Comparison of the cubic law fitted WTG characteristics and that of the typical WTG supplied by the manufacturer for Hummer H3.1-1kW WTG (AHDC, 2013) Although, some of the studied WTG models (Yang and Aydin, 2001; Katsigiannis and Geogilarkis, 2008) failed to account for the wind shear, Diaf et al. (2007), Ardakani et al. (2009), and Hocaoglu et al. (2009a, b) used an adjusted wind speed to reflect the wind shear using power law relationship whereas Dagdouguia et al. (2010) applied the logarithmic law resumed by Dufo-Lopez and Bernal-Agustin (2008) and deduced the adjusted wind speed taking the surface roughness of the terrain into consideration. Nevertheless, the logarithmic law cannot represent the wind shear for all conditions, as it is mathematically undefined for periods when the wind speeds at two different heights are similar (Ray et al., 2006). Furthermore, if wind speeds decrease with height, the calculated surface roughness length for that time interval is unrealistically high. The power output of a turbine is a function of air density, swept area and the cube of the wind speed. It also depends strongly on the wind regime as well as on the performance characteristics and the efficiency of the generator (Gao, 2006). These models, often used for the simulation, analysis and optimization of renewable energy generation systems in field applications ignored the effect of altitude, which is
44
essential for accurate estimation of wind energy. Altitude affects air density, which in turn affects the energy production of WTGs. The effect of varying altitude on air density is presented in Figure 2.8. It has been shown that the wind power decreases by a factor of 0.01 (1%) if the WTG is installed at a hub height of about 105m (Okundamiya and Nzeako, 2013).
1.005
1.000
zf
0.995
0.990
0.985
0.980
0
35
70
105 z hh (m)
140
175
200
Figure 2.8: Effect of varying altitude on air density (Okundamiya and Nzeako, 2013) Developing a WTG model requires the consideration of three factors (Gao, 2006). The first factor is the random nature of the site resource, which must be included, in an appropriate model to reflect the chronological characteristics of the wind at the site. The second factor is the relationship between the power output and the site resource. This relationship can be determined using the WTG operational parameters and specifications as provided by the manufacturer. The parameters commonly used are the cut-in wind speed (at which the WTG starts to generate power), the rated wind speed (at which the WTG generates its rated power), and the cut-out wind speed (at which the WTG is shut down for safety reasons). The third factor is the unavailability of the WTG expressed by the forced outage rate. This factor is accounted for using reliability indices (Gao, 2006). These factors directly affect the generator‘s output. 45
2.4.2 Photovoltaic Conversion System In the studies of solar energy, data on solar radiation and its components at a given location are a fundamental input for solar energy applications such as PV design (Besharat et al., 2013). Long-term average daily global irradiation is one of the most important parameters needed in solar applications (Sonmete et al., 2011). In addition, the knowledge of the diffuse solar radiation and its contribution to global solar radiation are essential for efficient assessment of solar energy potential. These data should be reliable and readily available for the design, optimization and evaluation of solar systems for any particular area (AlMohamad, 2004; Khan and Ahmad, 2012). In general, there are two steps in determining the available solar energy when supplying a remote load. The first step involves the determination of the amount of solar radiation that arrives on the earth at the PV panel‘s location. The next step is the actual modelling of the PV array, considering its efficiencies, losses and orientation. Each step requires a model that deals with many variables, and inputs into the second stage of the model utilize the results of the first step. The best way to determine the amount of global solar radiation at any site is to install measuring instruments (such as the pyranometer) at that particular site and to monitor and record its day-to-day recording, but this is really a very tedious and costly exercise (Katiyar and Pandey, 2010). Most developing countries, including Nigeria, are unable to afford the measuring equipments and techniques involved (Agbo, 2013; Besharat et al., 2013). A common practice for regions like Nigeria, where surface measurements (of solar radiation) are sparse, is to rely on available methods of estimations as well as to develop new methods.
46
A number of correlations and methods developed to estimate solar radiations for most weather stations are based meteorological data. The number of correlations published and tested to estimate the monthly average daily global and diffuse solar radiations is relatively high, which makes it difficult to choose the most appropriate method for a particular purpose and site (Al-Mohamad, 2004; Karakoti et al., 2011; Besharat et al., 2013). The selection of an appropriate model from various existing methods should, therefore, be on the basis of input data requirements (ease of obtaining input parameters) and the model accuracy. As a result, models believed to be universally applicable, and those that can be easily calibrated for the study areas are discussed. The aim was to review the available methods in the literature for estimating the monthly average daily and hourly solar (global and diffuse) radiation with the objective of developing the most accurate models for the study locations. 2.4.2.1 Modelling Global Solar Radiation on a Horizontal Surface Besharat et al. (2013) classified the various empirical models available in the literature into four categories (Sunshine-based, Cloud-based, Temperature-based, and other meteorological parameter-based models) based on the employed meteorological parameters. The classification took into consideration the availability of meteorological parameters typically used as inputs into radiation models. This consideration is essential for choosing the proper radiation model in any location. Sunshine-Based Models (Besharat et al., 2013) Sunshine duration is the most commonly used parameter for estimating global solar radiation on a horizontal surface probably for the ease of obtaining reliable sunshine data that are widely available at most weather stations. The most common estimation method in the literature is through the correlation between the monthly average daily clearness index (KT) 47
and the monthly average daily relative sunshine (SR). KT is the ratio of the monthly average daily global radiation (H) to the monthly average daily extraterrestrial radiation (Ho) both expressed on a horizontal surface (kWh m-2 day-1). SR, on the other hand, is the ratio of the monthly average daily sunshine duration (S) to the monthly average daylight duration (So), both expressed in (h). Based on the power of the key input parameter (SR), the sunshine-based models can be classified either as first-order or higher-order models. First-order (Angstrom–Prescott-type) model: Angstrom (1924) proposed the first relation for estimating the monthly average daily global solar radiation based on a correlation between the ratios of average daily global radiation to the corresponding value on a completely clear day. This relation is defined by the equation:
H bo b1 S R HC
(2.20)
where HC is the monthly average clear-sky daily global radiation on a horizontal surface (kWh m-2 day-1), bi is a regression coefficient (empirical constant) and other parameters as previously defined. Prescott (1940) modified the Angstrom relation with a view to resolving the ambiguity characterizing the definition of the HC. The modified Angstrom-type relation is of the form:
K T bo b1 S R .
(2.21)
Several studies (Black et al., 1954; Glower and McCulloch, 1958; Page, 1964; Rietveld, 1978; Akinoglu and Ecevit, 1990; Stone, 1999) applied the modified Angstrom-type (Angstrom-Prescott-type) model for the estimation of the monthly average daily global radiation on a horizontal surface for different regions of the world by determining the
48
empirical constant (bi) of Eq. (2.21) using the meteorological parameters of the given location of study. Among various Angstrom-Prescott-type models available in the literature are those calibrated by Page (1964) and Rietveld (1978), as shown in Eq. (2.22) and Eq. (2.23) respectively, claimed to be universally applicable.
KT 0.23 0.48 S R
(2.22)
KT 0.18 0.62 S R
(2.23)
Higher-order model: The two famous higher-order models applied in literatures are secondorder polynomial (quadratic) and third-order polynomial (cubic) models. Ogelman et al. (1984) expressed the ratio of global to extraterrestrial radiation by a quadratic function of the ratio of sunshine duration of the form:
K T bo b1 S R b2 S R . 2
(2.24)
Several studies in the literature calibrated Eq. (2.24) and determined the regression coefficient for the particular location of interest, for instance, Akinoglu and Ecevit (1990) used meteorological data collected from 58 locations in several countries and proposed the sunshine-based quadratic model:
KT 0.145 0.845 S R 0.280 S R2 .
(2.25)
In addition, Bahel et al. (1987) suggested a sunshine-based cubic relationship of the form: K T bo b1 S R b2 S R 2 b3 S R 3 .
(2.26)
Based on a correlation of meteorological parameters collected from 48 stations around the world with varied meteorological conditions and an extensive distribution of geographical 49
locations the empirical constants have been deduced as bo = 0.16, b1 = 0.87, b2 = - 0.61, and b3 = 0.34. The accuracy of results reported by Akinoglu and Ecevit (1990) and Bahel et al. (1987), believed to be universally applicable is satisfactory. Notwithstanding, Omotayo et al. (2011) evaluated the performance of five sunshine-based models for estimating monthly average daily global solar radiation for three areas (Ibadan, Onne and Kano) in Nigeria. The models evaluated are Black et al. (1954), GlowerMcCulloch (1958), Rietveld (1978), Akinoglu-Ecevit (1990) and Stone (1999). The evaluation used 28 – 35 years solar radiation and sunshine hourly data obtained from International Institute of Tropical Agriculture (IITA) station for studied sites based on three indicators (mean bias error, root mean square error, and t-statistic). The estimates of considered models were significantly different from the measured values. In addition, the considered models may not be applicable to the Nigerian environment. Several studies in literatures calibrated the sunshine-based models for estimating the monthly average daily global radiation on a horizontal surface for different locations in Nigeria. For instance, Akpabio et al. (2005) proposed the second-order relationship of Eq. (2.27) based on 16-years (1984 – 1999) monthly average daily global solar radiation and sunshine duration data collected from IITA station for Onne (latitude 4.77 oN, longitude 7.17 oE) located within the rainforest climate zone of southern Nigeria.
KT 0.147 1.125 S R 1.416 S R2 .
(2.27)
The performance of Eq. (2.27) was compared with four models including Akinoglu and Ecevit (1990) believed to be universally applicable. The evaluation was on the basis on three indicators: mean percentage error, mean bias error and root mean square error. The result
50
showed a higher level of accuracy in the calibration sites as compared to previously published models that claimed to be universally applicable. Cloud-Based Models In the absence of sunshine data, cloud data can be a useful alternative for the estimation of the monthly average daily global solar radiation on a horizontal surface (Supit and Van Kappel, 1998). Badescu (1999) proposed a cloud-based model based on a correlation between the clearness index and the average total cloud cover during daytime observations of the form given by Eq. (2.28) – Eq. (2.30).
KT bo b1 C ,
(2.28)
KT bo b1 C b2 C 2 ,
(2.29)
KT bo b1 C b2 C 2 b3 C 3 ,
(2.30)
where Ĉ is average total cloud cover during daytime observations (octa). The calibration results of the Badescu models for Yard city showed a good degree of accuracy but the simple linear model of Eq. (2.28) has the best performance (Besharat et al., 2013). Temperature-Based Models The air temperature is a characteristic parameter for the estimation of monthly average global solar radiation on a horizontal surface at regions where surface measurements are nonexistence due to the readily available air temperature data at meteorological stations. The
51
standard approach is on the basis of a correlation between the clearance index and the difference in maximum and minimum air temperatures (Besharat et al., 2013) Hargreaves and Samani (1982) first proposed a relation for predicting the global solar radiation on a horizontal surface in terms of the difference between the maximum and minimum air temperatures. The proposed equation is of the form:
K T bo b1 T
0.5
(2.31)
where ∆T (= TMAX – TMIN) is the temperature difference (oC), TMIN and TMAX are the monthly average daily minimum and maximum air temperatures (oC) respectively. The coefficient bo, was fixed at zero while b1 was set to 0.17 for arid and semi-arid regions, 0.16 for internal parts and 0.19 for coastal areas. Allen (1997) suggested employing a self-calibrating coefficient b1, in terms of the relative atmospheric pressure (i.e., ratio of the atmospheric pressure at a site to that at sea level) and introduced the values of 0.17 and 0.20 for interior and coastal regions respectively. The proposed self-calibrating model for estimating the coefficient b1 performs poorly for sites having an elevation above 1500 m. Chen et al. (2004) observed based on a comparison study that the Hargreaves and Allen models are not suitable for the estimation of daily global radiation in China. They proposed a logarithmic relationship between the daily clearness index (KT) and the temperature difference between the maximum and minimum daily air temperature (∆T) for improving the prediction accuracy. The proposed relationship is of the form:
K T bo b1 ln T
(2.32)
52
Hybrid Meteorological Parameter-Based Models The hybrid models combine one or more category-based (sunshine, cloudiness or air temperature) parameter with other available meteorological parameters (such as relative humidity and precipitation) or astronomical parameters (such as solar declination and hour angle), and both. These methods are reported to predict the monthly average global solar radiation on a horizontal surface with a high degree of accuracy but most of their input parameters are not readily available at most locations of interest (Besharat et al., 2013). Based on meteorological data collected for six stations in Egypt, El-Metwally (2004) proposed a relation of the form:
H bo b1 H o b2 TMAX b3 TMIN b4 Cˆ
(2.33)
Falayi et al. (2008) based their study on the correlation between global solar radiation and meteorological parameters using monthly average daily global solar radiations and the corresponding relative sunshine (SR), relative humidity (RH), and average air temperature (TAVE) data obtained from Iseyin, Nigeria. The proposed relation is of the form:
KT bo b1 S R b2 TR b3 TAVE b4 RH
(2.34)
where TR (= TMIN/TMAX) is the monthly average air temperature ratio, RH is the relative humidity (%), and TAVE is the monthly average daily air temperature (oC) 2.4.2.2 Modelling Diffuse Solar Radiation There are several models proposed in the literature, for estimating the monthly average diffuse solar radiation on a horizontal surface. A commonly used approach is on the basis of meteorological parameters, correlating the diffuse fraction as a function of either the 53
clearness index or cloud cover (also referred to as cloud fraction or cloud amount). The advantage of this method is that it requires only one observed or measured input (i.e., global solar radiation or cloud cover). The diffuse solar radiation can be estimated either on a daily or hourly basis. Daily Prediction Models Some of the models used for daily predictions are discussed below. Page model: Page (1964) proposed a widely used linear relationship for diffuse fraction estimation of the form:
K D bo b1 K T ,
(2.35)
where KD (= HD/H) is the monthly average daily diffuse fraction and KT is as previously defined. The model parameters (bi) were estimated in the literature using regression techniques. Iqbal (1979) and Yousif et al. (2013) evaluated the performance of the Page-type model but the results suggested different values of regression coefficients (compared to those initially proposed by Page) for accurate estimation of diffuse solar radiations. The difference is an indication of the variation of regression coefficients with site meteorology. Liu-Jordan model: Liu and Jordan (1960) developed a correlation, expressed by Klein (1977) in the form of a third-order polynomial relationship as of the form: K D bo b1 K T b2 K T2 b3 K T3 .
(2.36)
The model parameters (bi are 1.39, - 4.027, 5.531, and - 3.108 respectively) were obtained using partial regression analysis. The performance validation used experimental measurements (Klein, 1977). 54
Butt et al. model: Butt et al. (2010) investigated the relationship between cloud fraction and diffuse solar radiation, and proposed a linear regression model of the form:
K D bo b1 Cˆ .
(2.37)
The regression coefficients were determined using both ground-based and satellite-derived data collected for two contrasting rainforest sites in western and eastern Amazonian. Prediction results showed good agreement with the observed, but with a seasonal weakness that seems to be aligned with dry season anthropogenic activity, such as biomass burning. The estimation error is likely due to aerosols affecting diffuse proportion, but there are no data of sufficiently high spatial resolution to test the relationship at this stage. Karakoti et al. model: Karakoti et al. (2011) analyzed a group of relations by correlating the diffuse fraction with clearness index and relative sunshine, and then the diffuse coefficient with relative sunshine, using experimental data on global radiation, diffuse radiation and sunshine hours reported by Mani and Rangarajan (1982) for Indian stations. Based on the analysis, they proposed the cubic relationship of Eq. (2.38) for the estimation the diffuse component of the global radiation on a horizontal surface in India. K bo b1 S R b2 S R2 b3 S R3 ,
(2.38)
where K (=HD/Ho) is the monthly average daily diffuse coefficient. The predictive efficiencies of the Karakoti et al. relation in computing diffuse solar radiations on a horizontal surface were compared with the widely used Page proposed relation of Eq. (2.35) for 12 Indian sites. Three statistical parameters (percentage error, mean bias error, and root mean square error) were utilized for the evaluation. The results showed that errors from Page relation are higher compared to Karakoti et al. relation in all studied locations. Based on the
55
result, Page‘s method could be replaced by the Karakoti et al. model for the estimation of diffuse solar radiation in India. Hourly Prediction Models Liu and Jordan (1962) developed a theoretical method for deriving the average hourly solar (diffuse and global) radiation from the average daily global radiation, on the assumption that the atmospheric transmission is constant throughout the day, and this is independent of solar altitude. The ratio of hourly to daily diffuse and global solar radiations is computed respectively by the relations:
Id HD 24
I H 24
cos cos s , s cos s (sin s ) 180
cos cos s , s cos s (sin s ) 180
(2.39)
(2.40)
where I and Id are the average hourly global and diffuse solar radiation on a horizontal surface (both expressed in kW m-2), and ω and ωs are the hour angle and sunset hour angle (both expressed in degrees) respectively. The hourly beam radiation is given by the equation:
Ib I Id .
(2.41)
Using data collected for five stations in the United States of America and the Liu and Jordan‘s curve, Collares-Periera and Rabl (1979) developed an analytical expression for the ratio of hourly to daily global solar radiation, in terms of sunset hour angle as:
56
I G1 G2 cos H 24
cos cos s , s cos s (sin s ) 180
(2.42)
where coefficients G1 and G2 are given by the following equations.
G1 0.409 0.501sin ( s 60) .
(2.43)
G2 0.661 0.4767 sin ( s 60) .
(2.44)
These conversion methods have been tested and verified for many locations of the world (Wan Nik et al., 2012). The accuracy of the results reported by the original authors and those published in reviews are satisfactory. Other conversion methods proposed in the literature include Jain (1984, 1988), Baig et al. (1991) and Kaplanis (2006). Wan Nik et al. (2012) determined the performance accuracy of selected monthly average hourly global solar radiation models, i.e., Collares-Pereira and Rabl model (1979), the Jain model (1984, 1988), the Baig et al. (1991) model, and a new approach to Jain‘s and Baig‘s models by Kaplanis (2006). The evaluation utilized the daily global radiation data collected for three sites in the east coast of Malaysia. The result showed that if only the daily global irradiation is available, one can calculate the monthly average hourly global radiations on a horizontal surface using these models with a good accuracy, but the Collares-Pereira and Rabl model is the most accurate in general for the estimation of the monthly average daily global radiations for all three sites with humid tropical climate. 2.4.2.3 Modelling Total Solar Radiation on a Tilted Surface Photovoltaic arrays are not mounted on a horizontal surface but tilted at some angles. As such, it is necessary to calculate values of total solar radiation on a tilted surface given values
57
for a horizontal surface. The mathematical models used to determine the amount of tilted surface solar radiation are classified into two types, namely isotropic and anisotropic (Jakhrani et al., 2013). Isotropic sky models assume that the intensity of diffuse sky radiation is uniform over the sky dome (Posadillo and Lopez-Luque, 2009). Hottel and Woertz (1942) first assumed that the combination of diffuse and ground reflected radiation is isotropic and that the diffuse radiation from the sky and ground reflected radiation on the tilted surface is the same regardless of orientation. Based on this assumption, the total radiation on the tilted surface is the sum of the beam and the diffuse radiations. Liu and Jordan (1963) improved the HottelWoertz model on the basis that the diffuse radiation is only isotropic, which is given as:
1 cos 1 cos I t I b R g I g Id 2 2 ,
(2.45)
where I, Ib and Id are the average hourly global, beam and diffuse solar radiations on a horizontal surface respectively (kW m-2), It is the hourly total solar radiations on a tilted surface (kW m-2), β is the tilt angle (o), R g is the geometric ratio and ρg is the ground reflectance or albedo. The geometric ratio is the ratio of beam radiation on the tilted surface to beam radiation on a horizontal surface, expressed in terms of the cosine of zenith angles (Alam et al., 2005) as:
R g
cos z , inc
(2.46)
cos z , hor
where θz,inc and θz,hor are the zenith angle of incidence, with respect to the inclined and horizontal surfaces respectively. The most favourable value of Eq. (2.45) usually computed 58
for PV modules is obtained when the solar azimuth angle (γ) is 0° in the northern hemisphere or 180° in the southern hemisphere (Kreith and Goswami, 2011), and Eq. (2.46) expressed as
Rg
sin sin cos cos cos , sin sin cos cos cos
(2.47)
where δ is the solar declination, ϕ is the latitude, β is the surface inclination (tilt) angle, γ is the surface orientation (azimuth) and ω is the hour angle. All angles are expressed in degrees. The anisotropic sky models developed by various authors incorporate the contribution of the horizon brightening components. These models simulate the conditions present during partly cloudy skies to determine the magnitude of clear sky effects. Almost, all authors took the same value of the beam and the reflected part of radiation recommended by Liu and Jordan (1963) and amended only the diffuse radiation part. According to Jakhrani et al. (2013), the commonly used models are Klucher (1979), Reindl et al. (1990a, b), and HDKR models. Klucher (1979) expressed the total solar irradiation on a tilted surface as:
~ 1 cos 1 cos ~ 3 2 3 I t I b Rg I g Id 1 f m sin 1 f m cos sin z 2 2 2
(2.48)
~
where f m is the modulating function defined by
2
~ I f m 1 d I .
(2.49)
Reindl et al. (1990a, b) used the anisotropy index (ƛi) to quantify the diffuse radiation of the form: 59
1 cos 1 cos 3 I t I b I d i R g I g I d 1 i 1 f sin 2 2 2
(2.50)
The horizon brightening factor, ħf is expressed as stated:
f
Ib . I
(2.51)
while the anisotropy index, ƛai is given by the equation:
i
Ib . I0
(2.52)
where I0 is the hourly extraterrestrial radiation on a horizontal surface (kW m-2). Duffie and Beckman (2006) proposed a combination of the Hay and Davies (1980), Klucher (1979) and Reindl et al. (1990a, b) models, referred as the HDKR model is expressed as follows:
1 cos 3 I t R g I b I d i I d 1 i 1 f sin 2 2 1 cos Id g . 2
(2.53)
Different studies examined the performance of various hourly solar irradiance estimation models and based on their findings suggested the most accurate model for estimation of tilted surface radiation. Kudish and Ianetz (1991) examined three empirical models (one isotropic and two anisotropic) for computing total solar radiation on a tilted surface in Israel and found that all three types varied with time, season and location. Kamali et al. (2006) examined eight different models to estimate solar irradiation on tilted surfaces using daily measured solar
60
irradiation data in Iran and suggested Reindl et al. model for computing solar incident irradiation on tilted surfaces. Furthermore, Jakhrani et al. (2013) examined the performance of four tilted surface solar radiation models: Liu and Jordan (1963), Klucher (1979), Reindl et al. (1990) and HDKR (2006) models. On the basis of one-sample statistical test, they found that the Klucher model could be preferred for the estimation of tilted surface radiations in Kuching (Malaysia). 2.4.2.4 Photovoltaic (PV) Models The generated power of the PV array, with respect to the solar radiation, can be calculated (Hocaoglu et al., 2009a) as stated:
PPV pv N pv Amod I t ,
(2.54)
where ηpv is the PV generator efficiency, Npv is the number of PV modules, and Amod is the area of a single PV module (m2). Radziemska and Klugmann (2002), Kerr and Cuevas (2003), and Nishioka et al. (2003) showed the influence of environmental factors on the performance of PV modules. Parida et al. (2011) presented a review on major solar photovoltaic technologies. The study focused on PV system performance, sizing, reliability and control as well as the environmental aspect coupled with a variety of applications. A significant issue of concern is the heat build-up under the PV modules, resulting in the possible structural damage of the panel (if panels are unvented or if heat is unrecovered) and the lower efficiency of most PV modules with increasing temperature (Hollick, 2012). The effect of heating makes cooling required at high illumination conditions such as concentrated sunlight, or tropical conditions. The use of commercially available PV modules combined on top of the transpired collector and the 61
utilization of the unique air balancing features of the transpired collector remove heat from the back of the PV array. This concept allows for heat transfer from most PV modules currently available in the market (Hollick, 2012). Removing the excess heat generated by the PV arrays increases the electrical output. Although some studies (Hocaoglu et al., 2009a; b) neglected the effect of temperature on the PV module by the assumption of a constant PV generator efficiency, Katsigiannis and Georgilarkis (2008) used a derating factor to account for the effect of temperature on the PV system efficiency. Diaf et al. (2007) and Jalilzadeh et al. (2010) applied the energy balance methodology proposed by Duffie and Beckman (2006), to deduce PV generator efficiency in terms of irradiance and ambient temperature. The efficiency of a PV array does not strongly depend on irradiance only, but also on the module temperature (Chokmaviroj et al., 2006). Apeh (2010) focused on the alleviation of acute electrical power generation in Nigeria by use of solar power system. The objective was to determine the viability of the solar power system, either in a stand-alone or grid-connected mode, for a residential customer in Nigeria. Based on a computer program developed, the study estimated the techno-economic benefits for different options in Benin City (Nigeria). Results showed that grid-solar power system is more cost effective and reliable. However, reliable knowledge and understanding of the PV module performance under different operating conditions is of immense importance for correct product selection and accurate prediction of its energy performance (Zhou et al., 2010). The performance of crystalline silicon PV module is a function of the physical variables of the PV module material, temperature of PV module and the solar radiance on the PV module surface. In addition, the performance is highly dependent on its orientation and period of service (Gunerhan and Hepbasli, 2007; Yang and Lu, 2007). The orientation of the PV surface is described using its tilt angle and the azimuth, both relate to the horizontal. PV 62
orientation creates the problem of designing the optimum tilt angle for harvesting solar energy at fixed latitudes. The sizing optimization of PV system is a complex optimization problem. The aim is to achieve sustainable energy at optimum economic cost for the customer, with a reasonably accurate energy supply. Several authors have tackled PV sizing problems using various optimization methods (Katsigiannis and Georgilarkis, 2008; Hocaoglu et al., 2009a; Mellit et al., 2009). 2.4.2.5 Air Temperature Models There are three types of models for predicting air temperature (Bilbao et al., 2002). These are the hourly models developed from daily minimum and maximum air temperature values, stochastic models that link hourly air temperature to monthly average air temperature, and models that link daily average air temperature with daily minimum and maximum values. Erbs et al. (1983) suggested a relation for estimating the monthly average hourly air temperature, using temperature data collected from the United States. Cuomo et al. (1986) studied and analyzed air temperature on a daily basis in the Italian climate. Amato et al. (1989) described the stochastic dynamic types for both air temperature and solar irradiance daily time-series in the Italian climate. Hernandez et al. (1991) developed stochastic models for the predicting the daily minimum air temperatures. Knight et al. (1991) proposed a model to generate hourly air temperature profile with a random component included without introducing discontinuities between the last hour of one day and the first hour of the following day. The model enables the computation of hourly temperature values by an autoregressive method. Heinemann et al. (1996) developed an algorithm for the synthesis of hourly air temperature time-series, which takes into account a monthly average daily temperature pattern. In addition, the study considered the effect of neglecting the random
63
component in hourly temperature data for various solar heat systems. The extra complexities of including the random component of the hourly air temperature are unnecessary (Hollands et al., 1989), although, such part is essential for evaluating heating and cooling loads for passive solar applications (Boland, 1997). Bilbao et al. (2002) carried out a comparative assessment on four air temperature prediction models, i.e., the double cosine (Aguiar, 1996), Erb‘s (1984), Knight et al. (1991) and the standard model proposed by Bilbao et al. (2002). The study used RMSE (Root Mean Square Error), scatter graphs, and cumulative frequency curves to determine the performance of each model based on the European Mediterranean dataset collected for 34 different stations. The results showed that Erbs‘s model gave the overall best performance for estimating average air temperatures in all studied locations. The monthly hourly average diurnal temperature variation introduced by Erbs‘s (1984) is
Tave TAVE T 0.4632 cos (a 3.805) 0.0984 cos(2a 0.360)
0.0168 cos(3a 0.822) 0.0138 cos(4a 3.513)]
,
(2.55)
where Tave is the hourly monthly average air temperature (°C), TAVE is the daily monthly average air temperature (°C), ΔT is the difference between the daily monthly average maximum and minimum air temperature values (°C), a 360(h 1) / 24, and h is the time of the day (h). 2.5
Energy Storage System
When the power generated by the renewable energy system is greater than the load demand, the surplus energy could be stored in batteries for later use. The stored energy is supplied to the load when there is any deficiency in the power generation of the renewable energy sources. 64
The development of battery behavioural models has been the focus of many researchers over the years. Nguyen et al. (1990) examined the dynamic behaviour of the lead–acid cell during discharge with respect to cold cranking amperage and reserve capacity. Bernardi and Carpenter (1995) developed a mathematical model for lead–acid batteries by adding the oxygen recombination reaction. Kim and Hong (1999) developed a mathematical model for assessing the discharge performance of a flooded lead–acid battery cell. The study adopted the approach suggested by Gu et al. (1987), and the reaction kinetics of the negative electrode incorporated the diffusion–precipitation mechanism studied by Ekdunge and Simonsson (1989). These models are complex in terms of the expressions and number of parameters employed. In addition, many of the parameters are determined through measurement of internal components or processes, or by extensive experimentation. Consequently, these models tend only to assess the theoretical performance of battery designs and are not practical for simulating the performance of an arbitrary battery at arbitrary operating conditions (Zhou et al., 2010). Another standard modelling approach is to design a functional electrical circuit that is equivalent to the battery (Bernieri and Noviello, 1991; Cun et al., 1996). The parts of the circuit can represent the internal components of the battery, e.g. electrode and electrolyte resistances. The accuracy of these models depends upon the number of characterization tests performed to identify the values of the circuit elements (Bernieri and Noviello, 1991). In some cases, compensation factors are required to eliminate the influence of temperature. Cun et al. (1996) recommends re-characterization to account for changes due to battery ageing. Other battery behavioural prediction approaches include charge accumulation and empirical models. Two indices, i.e. the SOC (State of Charge) and the terminal (floating charge) voltage characterize the lead–acid battery (Yang et al., 2007). Piller et al. (2001) introduced 65
different methods for SOC determination. It concluded that the most used technique at this time for all systems is Ah (ampere-hour) counting method. The reason is that the Ah is the most direct and transparent way and quite easily implemented with satisfyingly accurate results for short time applications, especially if used in the range of low to medium SOC. Morgan et al. (1997) studied the performance of battery units in an autonomous hybrid energy system at different temperatures by considering the SOV (State of Voltage) instead of SOC. Models are available in the literature that describe the relationship between the terminal voltage, the current rate and the battery SOC (Chaurey and Deambi, 1992). However, battery charge is a complicated function of the battery‘s operating conditions. Therefore, experimentally determined correction factors are required (Kozaki and Yamazaki, 1997). These models also require considerable experimentation to obtain the parameters that characterize the targeted battery‘s behaviour (Zhou et al., 2010). The two independent factors that limit the lifetime of the battery bank are the lifetime throughput and the battery float life (NREL, 2010). In other words, batteries die either from use or old age. The float life of the battery, usually, specified by the manufacturer, is the maximum length of time that the battery can last before it needs replacement, regardless of how much or how little it is used. This limitation is typically associated with the damage caused by corrosion in the battery, which is strongly affected by temperature. Higher air temperatures are more conducive to corrosion, so a battery installed in warm surroundings would have a shorter float life than one installed in air-conditioned surroundings. The annual battery throughput is the amount of energy that cycle through the battery per anuum (NREL, 2010).
66
Although, small energy systems extensively utilized batteries, their use in medium and high power applications is not feasible (Battista et al., 2006). The use of batteries in supplying the instantaneous energy demand is inappropriate inasmuch as fluctuations negatively affect their lifetimes. The integration of super-capacitors (SC) as part of the energy storage (backup) system to meet the instantaneous power requirement can be a feasible option (Van-Voorden, 2008; Tankari et al., 2011). Super-capacitors have several advantages including a remarkably high energy density as compared to conventional capacitors, long life cycle, temperature stability, require no maintenance, and environmentally-friendly (Fang et al., 2011). In addition, SC can be charged and discharged continuously without degrading as batteries do. This property makes the SC a more suitable candidate as energy storage devices in renewable energy applications. Super-capacitors provide power to the system when there are surges (power bursts) since SC can be charged and discharged immediately. In contrast, batteries provide the bulk energy since they can store and deliver a larger amount of energy over a longer slower time. The storage capacity required can be reduced to a minimum when there is optimum sizing of the energy system at a given site. A rate of prediction at which the energy storage unit charges (discharges) when the generated power is more (less) than the demanded power requires accurate energy storage model. The model should also account for protection against over-voltage and under-voltage (to prevent battery over-charge and over-discharge). Accuracy of the model in optimizing the lifetime and capacity of the energy storage unit is needed. Santucci et al. (2014) modelled a high level controller for the power split between battery and super-capacitor, and formulated different Hybrid Energy Storage System (HESS) management controllers for the increase of battery life expectancy. The objectives were to formulate an innovative model predictive controller for hybrid electric or fully electric 67
vehicle management, specially designed for the reduction of battery wear; assess the performance benefits (i.e. battery life extension and power losses), and compare the energy storage power losses in case of battery only and of HESS, used for a case of Hybrid Electric Vehicle (HEV) along a variety of driving cycles. Results showed the ability of the HESS in reducing battery ageing, and enhancing energy efficiency when the system operates in critical climate conditions. 2.6
Optimization
An optimization problem is one requiring the establishment of the maximum or minimum value of a given function, called the objective function, subject to a number of restrictions, or constraints, placed on the variables concerned. Usually maximized, in practice, is the objective function representing units of the output in a production process, subject to constraints reflecting the availability of resources, machine time, environmental conditions, and so forth (Stroud, 1996). Regardless of its name, optimization does not necessarily mean finding the optimum solution to the problem since it may be unfeasible due to the characteristics of the problem, but an approximate (satisfactory) solution to the problem. In some cases, it may involve the category of Non-deterministic Polynomial-time hard (NPhard) problems, which could lead to computation times that are too high for practical purposes (Banos et al., 2011). Practical application of optimization problems requires efficient and powerful computational algorithms. These algorithms, when run on computers, should numerically solve the mathematical models of the medium as well as large size optimization problems arising from different fields. Conventional optimization methods, such as those summarized in (Momoh et al., 1999a, b; Stroud, 1996) can be applied in handling many classes of optimization
68
problems, but may not be able to solve certain types of objective functions (NP-hard problems). The limitation led to the necessity of developing new solution methods that can overcome these shortcomings. In recent decades, several authors have proposed approximate methods, including artificial intelligence techniques to solve the problems of the conventional optimization techniques (Zhou et al., 2010). Various artificial intelligence methods based on heuristic approaches can provide satisfactory (not necessarily optimal) solutions to significant instances of complex problems quickly. Meta-heuristics, a generalization of heuristics, require few modifications to be adapted for specified application (Gendreau and Potvin, 2010). The meta-heuristic algorithms are commonly classified as the trajectory versus population-based, nature-inspired versus non nature-inspired, and memory-based versus memory-less methods (Banos et al., 2011). In some cases, when the complexity of the problems to solve is so high that even heuristic and meta-heuristic methods are not able to obtain accurate solutions in reasonable runtimes, the parallel processing becomes a fascinating way to obtaining robust solutions in reduced runtimes (Alba, 2005). 2.6.1 Meteorological Data for Simulation The long-term system performance is one of the most important design criteria for hybrid energy systems. Meteorological data containing hourly solar radiation, wind speed, and ambient temperature are required in the performance simulation of these systems. The commonly used data for simulation studies are either the time-series or statistical. Time-Series Meteorological Data The hybrid system‘s behaviour can be determined based on the time-series meteorological input data, which, usually, have a resolution of 1-hour intervals. Baring-Gould et al. (2002) 69
employed time-series meteorological data for feasibility study and design of the hybrid systems. A common drawback of this approach is that it requires significant computational effort. Several research efforts strived to decrease the simulation time, reduce the number of variables used in process simulation or both. A common method is through the development of a predictive model such as the hourly time-series algorithm developed by Celik (2003). This algorithm requires monthly average values of wind speed distribution parameters and solar radiations, for enabling the estimation of system performance using simple wind distribution parameters and thereby eliminating the need for time-series hourly data. However, time-series environmental input data, especially wind data, may not be available for many locations (Zhou et al., 2010). Statistical Meteorological Data Knight et al. (1991) pointed out that the hourly records of meteorological variables for extended periods are not available for many locations. When the measured weather data do not exist for the location, they can be obtained in mainly two ways. Firstly, the appropriate weather data may be synthetically generated from the monthly-average values of the meteorological data (Gansler et al., 1994). Secondly, the weather data may be extrapolated from a nearby site by making necessary adjustments (Wahab and Essa, 1998). Generation of wind speed, solar radiation, and temperature data, was the objective of several studies discussed in the literature. Typical Meteorological Year (TMY) is one of the most common synthetic weather data sequences used in wind and solar energy systems simulations (Zhou et al., 2010). Hourly TMY weather data, usually, consists of 12 months of hourly data. The average value for each month is selected from the long-term weather data as being the best representative of that particular month or is generated from several years of weather data
70
which would yield the same statistics (such as the average solar radiation and clearness index) as those of several years‘ data (Zhou et al., 2010). The most popular method for deriving the TMY data, firstly developed by Hall et al. (1979), is an empirical approach selecting individual months from different years. The Hall et al. model applied the statistical method proposed by Filkenstein–Schafer (1971). Yang and Lu (2004) developed a local TMY for solar and wind power application and evaluation for various cities by considering different weighting factors of meteorological parameters. Other studies (Elhadidy, 2002; Elhadidy and Shaahid, 2004; Nema et al., 2010; Liu et al., 2010; Nafeh, 2011; Fulzele and Dutt, 2012; Martinez-Díaz et al., 2013) assessed the performance and viability of various hybrid energy systems based on statistical meteorological data. 2.6.2 Statistical Analysis The performance evaluation and validation of the various estimation models studied in the literature was determined on the basis of one or more statistical indicators. The commonly used statistics are the Coefficient of Correlation (r), Coefficient of Determination (R2), Mean Bias Error (MBE), Mean Absolute Bias Error (MABE), Relative Percentage Error (RPE), Root Mean Square Error (RMSE), and the t-statistic (t-stat) test indicators. The r-value determines the linear relationship between the measured (observed) and predicted values. It is expressed as (Wan Nik et al., 2012; Besharat et al., 2013): N
(Q , pred pred )(Q ,mea mea )
r
1
N N 2 2 ( Q ) , pred pred (Q , mea mea ) 1 1
71
(2.56)
where Qpred and Qmea are the predicted and measured values at time η respectively, while µpred and µmea are the respective average values within the entire simulation time (η = 1, 2, . . ., N). The R2 statistic defined by Eq. (2.57) gives the percentage of the variation of the dependent variable in connection with the explanatory (independent) variables. N
(Q ,mea Q , pred ) 2
R 2 1
1 N
(Q ,mea mea ) 2
.
(2.57)
1
The MBE, MABE, RMSE, and RPE defined in the following equations (Besharat et al., 2013) are common error terms often used in comparing models.
1 N (Q , pred Q ,mea ) . N 1
(2.58)
1 N MABE (Q , pred Q ,mea ) . N t 1
(2.59)
MBE
1 N RMSE (Q , pred Q ,mea ) 2 . N 1 Q , pred Q ,mea 100 RPE . Q , mea
(2.60)
(2.61)
For better data modelling, these error indicators should be closer to zero, but r and R2 should approach to one as closely as possible (Wan Nik et al., 2012; Besharat et al., 2013).
72
Although these tools, i.e., Eq. (2.56) – Eq. (2.61), are commonly used indicators, the t-stat (determined in terms of MBE and RMSE) has satisfactory accuracy for analyzing data predicted as it determines whether the model‘s estimate is statistically significant at a particular confidence level or not. The t-stat is expressed as (Stone, 1993):
t stat
m 1MBE 2 2 2 RMSE MBE
.
(2.62)
To determine whether a model‘s estimates are statistically significant, one has to determine the critical t-stat (i.e., tcr(χ/2) at the χ level of significance and (m - 1)o of freedom, where m refers to the number of data rows, i.e., the 12 months of the year) from standard statistical tables. The model‘s estimates to be judged are statistically significant if the calculated t-stat lies between the interval defined by – tcr and tcr (acceptance region under the reduced normal distribution curve). 2.6.3 Criteria for Hybrid Energy System Optimizations The selection of an optimum combination for a hybrid renewable energy system that will satisfy the load demand is carried out on the basis of power supply reliability and system lifecycle cost, irrespective of the sizing and optimization technique used (Zhou et al., 2010). While the expected reliability from such hybrid system constitutes an important criterion in optimization, the cost of the system is the governing factor, unless an unlimited budget is available. A hybrid renewable energy system can attain an optimum solution when there is a good compromise between these criteria (system reliability and cost). Conversely, when assessing hybrid power systems incorporating conventional energy or fossil-fuelled sources such as diesel generators or utility grids reduction in pollutant emissions is another essential factor, typically considered. 73
2.6.3.1 Power Supply Reliability Analysis Power supply reliability analysis is an important step in hybrid energy system design process because the intermittent nature of energy sources (solar radiation, wind speed, utility grid) highly influences the energy production from such systems. There are a number of methods used in literatures to calculate the power supply reliability of the hybrid systems. The most technical methods widely used for the evaluation of power system reliability include the LLP (Loss of Load Probability), LLPS (Loss of Load Power Supply), LOLP (Loss of Load Probability), SPL (System Performance Level), LOLH (Loss of Load Hours) and the LPSP (Loss of Power Supply Probability) method, but the LPSP is the most popular and frequently used method (Diaf et al., 2008). The LPSP is the probability that an insufficient power supply results when the hybrid system is not able to satisfy the load demand (Bin et al., 2003). The design of a reliable hybrid energy system can be achieved by using the LPSP as the key design parameter. Two approaches (chronological simulations and probabilistic methods) exist for the application of the LPSP in designing a hybrid power system. The method of chronological simulations is computationally burdensome and requires the availability of data spanning a particular period. On the other hand, probabilistic techniques incorporate the fluctuating nature of the resource and the load, thus eliminating the need for time-series data. According to Bin et al. (2003), the loss of power supply probability is defined as: N
Pdef ( ) .
LPSP
1 N
Pd ( ) .
,
(2.63)
1
74
where Pd (η) and Pdef (η) are the load power and the deficit in load power at time η, respectively, while N is the operation time. 2.6.3.2 System Cost Analysis There are different economic criteria for evaluating the viability of hybrid energy systems, such as the Net Present Cost (NPC) and Levelized Cost of Energy (LCE) concepts (Zhou et al., 2010). The NPC is defined as the total present value of a time series of cash flows. It consists of the initial cost of all the system components, the cost of any component replacements that occur within the project lifetime and the cost of maintenance. The system life span considered, usually, is the life of the PV array, which is the component with a longer lifespan. It takes into account any salvage costs, which is the value remained in a part of the system at the end of the project lifetime. A more detailed description of its calculation can be found in (DufoLopez and Bernal-Agustin, 2005; Bernal-Agustin et al., 2006). The HOMER computer-based model uses the total NPC to represent the life-cycle cost of the system. It assumes that all prices escalate at the same rate and takes the ‗‗annual real interest rate‖ rather than the ‗‗nominal interest rate‖. This method allows inflation to be factored out of the analysis (Dalton et al., 2008). Extensively used as an objective term to evaluate the hybrid energy system configurations is the concept of LCE (Deshmukh and Deshmukh, 2008). The LCE is defined as the ratio of the total annualized cost of the system to the annual electricity delivered by the system mathematically expressed as (Athanasia and Anastassios, 2000)
75
LCE
COI ann, sys E ann, sys
TPV CRF . E ann, sys
(2.64)
COIann,sys is the total annualized cost, Eann,sys is the total annual energy output of the system (kWh yr-1), and TPV and CRF are the total present value of actual cost of all system components and the capital recovery factor, respectively, which can be expressed as follows (Athanasia and Anastassios, 2000):
TPV
j
pC j
,
(2.65)
j 1
CRF
ri (1 ri ) LN (1 ri ) LN 1
,
(2.66)
where, pCj is the total present cost of component j in system life, ri is the annual interest rate and LN is the lifespan of the system (yr). Nafeh (2011) defined the annualized cost of investment of a hybrid system in terms of the present worth (PW) as stated in Eq. (2.67).
COI ann, sys
1 LN
COI j, ini COI j, rep( PW ) COI j, om( PW ) COI j, sal ( PW ) , j
(2.67)
j 1
where, COIini is the total initial Cost of Investment, COIrep (PW) is the total PW of replacement cost, COIOM (PW) is the total PW of annual Operation and Maintenance (OM) cost, COIsal (PW) is the PW of all salvage value, and j is the total number of hybrid system‘s component unit. Other economic approaches, such as the LCC (Life-Cycle Cost) and LCS (Levelized Cost of the System) are also widely used (Valente and Almeida, 1998; Yang et al., 2008).
76
2.6.3.3 Environmental Impact Analysis The reduction of pollutant emissions in hybrid power systems consisting of a mix of renewable and fossil fuelled sources imposes additional constraints on the design of such systems. In practice, the design of such hybrid system, considers at least two objectives (costs and either reliability or pollutant emissions) simultaneously. The task of simultaneously minimizing cost and emissions, is considered to be in conflict, since a reduction in design costs implies a rise in pollutant emissions and vice versa (Pelet et al., 2005; Bernal-Agustin et al., 2006; Dufo-Lopez and Bernal-Agustin, 2008; Zhou et al., 2010). The assertion that the cost of renewable energy is higher than conventional energy is consensus of the literature reviewed in this study. In other words, a reduction in renewable energy can lead to a corresponding increase in the conventional energy in attaining optimum configuration. Therefore, the task of getting good results in such (multi-objective) problems is complicated. Given the complexity of the multi-objective problems, because of the large number of variables that are usually considered and the mathematical models applied, classic optimization techniques may consume excessive computer processing time. In addition, it may not be capable of taking into account all the characteristics associated to the posed problem. The multi-objective system design usually search for the configuration, the control or both that yields the lowest total cost through the useful life of the installation while the pollutant emissions are calculated after obtaining the design that minimizes costs (Muselli et al., 2000; Elhadidy, 2002; Dufo-Lopez and Bernal-Agustin, 2005). Other multi-objective systems design (Garcia and Weisser, 2006; Mohammed and Koivo, 2010) included the environmental issues associated with this type of installations during the design process by economically evaluating and including them as part of the costs objective function. These designs simplify 77
the task by reducing the problem to a single objective function. This mapping of costs to emissions is subjective and decisively influences the results of the model. In addition, other methods including the Multi-Objective Evolutionary Algorithm (MOEA) and Strength Pareto Evolutionary Algorithm (SPEA) are being applied in multi-objective design work (BernalAgustin et al., 2006; Zhou et al., 2010). Estimation of Greenhouse Gas Emissions In order to alleviate the pending calamity of global warming, it is necessary to evaluate the carbon footprint from different sectors of the economy since all sectors can reduce GHG emissions through appropriate actions. Carbon footprint is the quantity of greenhouse gas expressed in terms of CO2 equivalent (CO2-eq), emitted into the atmosphere by an individual, organization, process, product, or event from within a specified boundary. Other terms used (as a synonym of carbon footprint) in the available literature are embodied carbon, embedded carbon, carbon content, carbon flows, a virtual carbon, GHG footprint, and climate footprint. In essence, there is little uniformity in the definitions of carbon footprint within the available literature (Pandey et al., 2011). Several studies and methods for carbon footprint calculation suggest the inclusion of other GHG as well, apart from only CO2 (Ferris et al., 2007; Garg and Dornfeld 2008; Johnson, 2008; Kelly et al., 2009; Edgar and Peters, 2009; Brown et al., 2009). If the quantification of carbon includes all GHG originating from within the boundaries, the term ―climate footprint‖ would be a better indicator for comprehensive GHG accounting. Carbon footprint helps in emission management and evaluation of mitigation measures, as well as to compare natural versus anthropogenic impacts on the environment. It acts as an indicator of the impact of lifestyle of a citizen of a country. Estimates of carbon footprint 78
from a given sector of the economy are essential, in order to study more about the associated environmental impact, and to monitor consequences of actions and everyday practices. In addition, managers and policy makers could exploit such information in focusing their efforts on the most influential sectors. The information could assist in monitoring the environmental impacts of the entire industry. Monitoring of the entire sector is necessary to capture any increase in total emission and it gives the basis for suggestions of mitigation measures (Rahman and Khondaker, 2012). Pandey et al. (2011) reviewed a wide range of literatures available in the field of carbon footprinting, involving governments, institutions, households and organizations. In spite of prevailing differences among the calculations, the CO2-eq mass based on 100 years GWP has been accepted as reporting unit of carbon footprint probably due to convenient calculations and extensive acceptance. To convert emissions of non-CO2 gases to units of CO2-eq, the emissions of each gas in units of mass, e.g., metric tons (t) are multiplied by the appropriate GWP factors given in Table 2.3. Table 2.6 shows the greenhouse gases emission factor for the grid electricity from various sources (Nandi and Ghosh, 2010; LGOP, 2010).
79
Table 2.6: GHG emission factors for the grid electricity from various sources (Nandi and Ghosh, 2010; LGOP, 2010) CO2 emission factor
CH4 emission factor
N2O emission factor
(tCO2 MWh-1)
(tCH4 MWh-1)
(tN2O MWh-1)
Natural gas
0.4520
0.00300
0.00100
Large hydro
0.0000
0.00000
0.00000
Diesel
0.8970
0.00200
0.00200
Combined
0.6605
0.00005
0.00001
Fuel type
2.6.4 Optimum Sizing Methods for Hybrid Energy System According to Banos et al. (2011), optimum sizing of renewable energy systems are, usually, achieved using optimization methods. The optimum sizing method can help to guarantee the lowest investment with full use of the system components so that the hybrid system can work at the optimum conditions in terms of investment and power supply reliability. The various optimum sizing methods for hybrid power systems are classified as follows (Zhou et al., 2010): simulation and optimization software and optimization methods. The optimization methods are categorized into two subgroups, viz. conventional and artificial intelligence methods. 2.6.4.1 Simulation and Optimization Software Several software tools are available for designing of hybrid systems, such as Hybrid power system simulation model (HYBRID2), Transient energy System simulation tool (TRNSYS), Hybrid Optimization Model for Electric Renewables (HOMER), HYBRIDS and Hybrid 80
Optimization by Genetic Algorithms (HOGA) (Green and Manwell, 1995; Turcotte et al., 2001; Dufo-Lopez et al., 2005; NREL, 2010). The Renewable Energy Research Laboratory (RERL) of the University of Massachusetts developed HYBRID2. It is a computer model for hybrid systems design and operation simulation; the simulation is very precise, as it can define time intervals from 10 minutes to 1 hour. This software uses a combined time series/probabilistic method to account for longterm/short-term predictions. The probabilistic approach used is based on probability density functions (pdf), such as the Normal, Gaussian and the Weibull distribution. These functions are used in each time step to find expected maximum and minimum values, the fraction of time in which those values may be within a certain range, and the amount of energy that may be required or available in the range. TRNSYS originally developed by the National Renewable Energy Laboratory (NREL), to simulate thermal systems has incorporated PV systems to simulate hybrid systems; but it cannot optimize the system (Turcotte et al., 2001). Consequently, NREL developed another computer model, HOMER. It is a time-step simulator using hourly load and environmental data inputs for renewable energy system assessment. This model could assist in the design of hybrid energy systems as well as to facilitate comparison of power generation technologies based on the net present cost for a given set of constraints and sensitivity variables. It requires inputs on component types, their numbers, costs, efficiency, and so forth. The user must enter the parameters for the optimization by choosing the different combinations for PV array power, wind turbine power, the battery power and the inverter power. There is no optimization between different types of battery. This program uses the kinetic battery model (Manwell and McGowan, 1993) and the program does not optimize the state of charge set
81
point. In addition, it requires excessively high computation time when the number of design parameters is very high (Ekren and Ekren, 2010). HYBRIDS, a commercially available application produced by Solaris Homes, are a Microsoft Excel spreadsheet-based renewable energy system assessment application and design tool. It assesses the technical potential of renewable energy system for a given configuration by determining the potential renewable fraction and evaluating the economic viability based on the systems net present cost. It also requires the daily-average load and environmental data estimated for each month of the year. HYBRIDS‘ is comprehensive in terms of renewable energy system variables and the level of detail required and necessitates a higher level of knowledge of renewable energy system configurations than HOMER. It enables the user improves the design skills for renewable energy system through its application. However, it can only simulate one configuration at a time. In addition, it does not provide an optimized configuration. The NREL recommends optimizing the system with HOMER and then, once the optimum system is obtained, improving the design using HYBRID2 (Zhou et al., 2010). HOGA, developed by the Electrical Engineering Department of the University of Zaragoza (Spain), is a hybrid system optimization program that applies the genetic algorithms (GA) to design the sizing and management control of a hybrid energy system (Dufo-Lopez et al., 2005). The program, developed in C++ calculates the optimal configuration of the system. The optimization, which can be either mono-objective or multi-objective, is carried out using 1-hour intervals, during which all parameters remained constant. HOGA offers to perform an evaluation of all possible combinations, both for components and control variable strategies.
82
2.6.4.2 Conventional Optimization Methods The conventional or traditional optimization methods, such as graphical construction method, the probabilistic approach, and iterative approach can be used for sizing of hybrid energy system. Borowy and Salameh (1996) and Markvart (1996) used graphical methods for optimum sizing of hybrid energy system consisting of wind, PV and battery bank. This approach entirely neglects other important factors (such as the PV module slope angle and the wind turbine installation height) as only two parameters (either PV and battery, or PV and wind turbine) are included in the optimization process. Probabilistic approaches of sizing hybrid power system utilized by Karaki et al. (1999) and Tina et al. (2006) account for the effect of the solar radiation and wind speed‘s variability in the system design. However, it cannot represent the dynamic changing performance of the hybrid system. For iterative optimization process, minimization of the system cost is implemented either by linearly changing the values of the corresponding decision variables or employing linear programming techniques, resulting in suboptimal solutions and increased computational effort requirements (Yang et al., 2007). Furthermore, it, usually, does not optimize the PV module slope angle and wind turbine installation heights which also highly affect both, the resulting energy production and system costs. In general, the conventional optimization methods are derivative-based techniques (Momoh et al., 1999a, b). These methods, although are reliable and have proven their effectiveness in handling many classes of optimization problems, can encounter difficulties such as getting trapped in local minimum, increasing computational complexity, and not being applicable to solve certain types of objective functions (NP-hard problems). As the number of optimization variables increase, the number of simulations also increases exponentially, with a corresponding increase in time and effort required. It is, therefore, very important for 83
designers to find a feasible optimization technique for designing the optimum system configurations quickly and accurately. Using feasible optimization method, optimum configurations which meet the load requirement can be obtained (Yang et al., 2007; 2008). 2.6.4.3 Artificial Intelligence Method Artificial intelligence describes the ability of a machine or artefact to perform similar kinds of functions that characterize human thought (Mellit et al., 2009). Artificial intelligence is a rapidly growing tool that can overcome most of the limitations found in derivative-based techniques (Zhou et al., 2010). There are several global optimization algorithms, which applies artificial intelligence methods for sizing hybrid systems. These are, usually, population-based heuristics (also called general-purpose) algorithms because of their applicability to a wide range of problems. Standard global optimization algorithms include the Genetic Algorithms (GA), Tabu Search (TS), Ant Colony Optimization (ACO), Differential Evolution (DE) and Particle Swarm Optimization (PSO). A brief overview of the concepts and operational principles of these algorithms is presented below. Genetic Algorithms The genetic algorithms follow the principles of natural genetics and natural selection for the determination of the search and optimization procedures, which works on the principle of survival of the fittest. Genetic algorithm, unlike strict mathematical methods, has the apparent ability to adapt to non-linearity and discontinuities commonly found in power systems (Goldberg, 2002). It begins its search from a randomly generated population of designs that evolve over successive generations (iterations), eliminating the need for a user-supplied starting point. Some key features are the use of objective function information to guide the search; coding of the parameters used to calculate the objective function in guiding the 84
search; search through many points in the solution space at one time; and the use of probabilistic rules in moving from one set of solutions (a population) to the next. According to Williams and Crossley (1998), the GA employs three operators to propagate its population from one generation to another. The first operator is the ―selection or reproduction‖ operator that mimics the principle of ―survival of the fittest‖. The essential idea in reproduction is to select strings of above-average fitness from the existing population and insert their multiple copies in the mating pool, in a probabilistic manner. This results in the selection of existing solutions (with better than average fitness) that can act as parents for the next generation. The second operator is the ―crossover or recombination‖ operator, which mimics mating in biological populations. The crossover operator propagates features of exemplary surviving designs from the current population into the future population, which have better fitness value on the average. Crossovers consist of swapping of chromosome bits from two superior quality chromosomes while mutations can be random changes to the chromosomes of individual solutions. The last operator is ―mutation‖, which promotes diversity in population characteristics. The mutation operator allows for global search of the design space and prevents the algorithm from being trapped in a local minimum. While crossover aims at recombining parts of good substrings from good parent strings towards creating a better offspring, mutation, on the other hand, alters a single child string locally towards creating an excellent child string. Consequently, over a number of generations (iterations), desirable traits (design characteristics) will evolve and remain in the genome composition of the population (set of design solutions generated each iteration) over traits with weak undesirable characteristics.
85
The extensive use of GA for solving complex-design optimization problems is due to its capability in handling both discrete and continuous variables and non-linear objective and constraint functions, without requiring gradient information. The ability is traceable to its intuitiveness, ease of implementation, and the practical ability to solving real, non-linear and mixed-integer optimization problems that are typical of complex engineering systems. Conversely, its expensive computational cost is a major disadvantage. Tabu Search The tabu search algorithm first proposed by Glover (1986) uses the ―move‖ operation to determine the region of any given solution. The ability to escape from the local optimum and the occurrence of cycles, which, usually, cause the simple descent algorithms to terminate characterises its operation. It obtains a finite-size list of forbidden moves using tabu moves. The basic assumption is that the sub-optimal points, where the basic greatest-descent algorithm stops, can be better starting points with respect to random restarts, if care is taken so that the local minimum does not become attractors of the dynamics included by the algorithm, and that limit cycles do not occur. The two main components of the tabu search algorithm are the tabu list restrictions and the aspiration criteria of the solution associated with these restrictions (Kim et al., 2001). The tabu list restrictions directly state a given change of movements or indirectly as a set of logical relationships or linear inequalities. Tabu lists are managed by recording moves in the order in which moves are made. If a new feature enters into the tabu list, the oldest one is released from the tabu list. The proper choice of the tabu list size is critical to the success of the algorithm, and it depends on the specified problem. Aspiration criteria can override the tabu restrictions, i.e., if a particular move is forbidden, the aspiration criteria, when satisfied,
86
can reactivate this move. The appropriate use of such measures can be highly significant for enabling a tabu search method to reach its best performance level. The most widely used aspiration criterion removes a tabu classification from a trial move when a move yields a solution better than the best so far obtained. Ant Colony Optimization (ACO) The ant colony algorithm introduced by Dorigo et al. (1991) is a swarm-inspired probabilistic method for solving computational problems. Several problems have emerged from diversifying the original idea to solve a wide range of NP-hard problems, drawing on various aspects of the behaviour of ants. The ant colony algorithm has inspired many researchers to provide solutions to various combinatorial optimization problems such as travelling salesman problem (Dorigo et al., 1996), routing problem (Schooderwoerd et al., 1997) and many other NP-hard problems in which the values for discrete variables are found to optimize an objective function. The significant features of the ACO are positive feedback, distributed computation and incremental construction of solutions. Nevertheless, it suffers several drawbacks. For example, the algorithm is experimental rather than theoretical – this makes theoretical analysis difficult. Probability distribution changes by iteration, although convergence is guaranteed, the time to convergence is uncertain, and the order of random decisions is dependent (Selvi and Umarani, 2010). In addition, its application to the continuous domain is not straightforward since it was originally proposed for discrete optimization problems. Differential Evolution The adaptation scheme of the differential evolution algorithm, introduced by Storn and Price (1995), ensures that the mutation increments are automatically scaled to the correct 87
magnitude. It uses a tournament selection where the offspring vectors compete against one of their parents for reproduction. It controls the speed and robustness of the search (a lower value increases the rate of convergence, as well as, the risk of being stuck at a local optimum). The crossover is a complementary method for DE. It aims at reinforcing the prior successes by generating the offspring vectors out of the object vectors. The objective function evaluates the newly created vector and the corresponding value compared with the target vector. The next generation keeps the best-fit vector. Every generation evaluates the optimal parameter vector in order to track the progress made throughout the minimization process, thus making the DE elitist method (Bakare and Krost, 2012a; 2012b). According to Storn and Price (1995), the design principles of the DE are easy structured; robust and easy to use; operate on floating point format with high precision; effective for integer, discrete and mixed-parameter optimization; handles non-differentiable, noisy and/or time dependent objective functions; and practical for non-linear constraint optimization problems with penalty functions. As a robust and powerful, adaptive tool for solving search and optimization problems, they have been proposed for solving various power system problems. Particle Swarm Optimization The particle swarm algorithm is a population-based stochastic optimization approach, which belongs to the class of evolutionary computation for solving global optimization problems. Kennedy and Eberhart (1995) first introduced the PSO algorithm as an optimization method. It is straightforward in concept, easy to implement and computationally efficient. It applies virtually to any problem expressed in terms of an objective function. Eslami et al. (2012) surveyed the state-of-art in the PSO algorithm. The study concluded that the PSO has
88
gathered considerable interest from the natural computing research community. It provides quick and efficient optimization of complex multi-dimensional search spaces. Clearly, the algorithm shines for its simplicity and for the ease with which it can be adapted to different application domains and hybridized with other techniques The main drawbacks of the PSO, as stated by Lovbjerg (2002), are the problem-dependent performance and premature character. This dependence commonly caused by parameters setting, for example, assigning different parameter settings to particle swarm algorithm, will result in high-performance difference. In general, no single parameter setting exists which can be applied to all problems and performs dominantly better than other parameter settings. A standard way to deal with this problem is to use self-adaptive parameters. Several authors (Tsou and MacNish, 2003; Hassan et al., 2004; Ratnaweera et al., 2004) have successfully applied self-adaptation in particle swarm algorithm. The premature nature could lead to convergence at a local minimum. Although the particle swarm algorithm converges to an optimum much faster than other evolutionary algorithms, it, usually, cannot improve the quality of the solutions as the number of iterations increases (Angeline, 1988a; b). In essence, PSO, usually, suffers from premature convergence when optimizing large multi-modal problems. The particle swarm algorithm shares many similarities with evolutionary computation techniques such as GA. The system initializes with a population of random solutions and searches for optimum by updating generations. However, unlike the GA, it has no evolution (genetic) operators such as crossover and mutation. The potential solutions, called particles, fly through the problem space by following the current optimum particles. They also have
89
memory, which is necessary to the algorithm. Compared to GA, the advantages of particle swarm algorithm are that PSO is easy to implement, and there are few parameters to adjust. Pattern Search Method The Pattern Search (PS) algorithm is an evolutionary technique that is suitable to solve a variety of optimization problems, which lie outside the scope of the standard optimization methods. The method is very simple in concept, easy to implement and computationally efficient. And it does not require any information about the gradient of the objective function at hand, while searching for an optimum solution. Unlike other heuristic algorithms, such as genetic algorithms, pattern search possesses a flexible and well-balanced operator, which enhances and adapts the global and fine tune local search (Al-Sumait et al., 2007). The algorithm proceeds by establishing a set of points called mesh, around the given point X0. This point X0 could be the initial starting point supplied by the user or it can be computed from the previous step of the algorithm. The mesh is formed by adding the current point to a scalar multiple of a set of vectors called a pattern (Mahapatra et al., 2014). If a point in the mesh is found to improve the objective function at the current point, the new point becomes the current point at the next iteration. The algorithm polls the mesh points by computing the fitness function values until it finds one whose value is smaller than the fitness function value of X0. If such point exists, then the poll is successful and the algorithm sets this point equal to X1; otherwise, the poll is successful. The current mesh size is doubled, after a successful poll so that the mesh size at the next iteration is larger, and vice versa. The pattern search optimization algorithm repeats the illustrated steps until it finds the optimum solution for the minimization of the fitness function. The algorithm stops when any of the following conditions is satisfied (Al-Sumait et al., 2007): 90
1. The mesh size is less than the mesh tolerance. 2. The number of iterations performed by the algorithm reaches a predefined value. 3. The total number of objective function evaluations performed by the algorithm reaches a pre-set maximum number of function evaluations. 4. The distance between the point found at one successful poll and the point found at the next successful poll is less than a set tolerance. 5. The change in the objective function from one successful poll to the next successful poll is less than a function tolerance. Other artificial intelligence methods commonly applied for optimum sizing of hybrid energy system in literatures includes Artificial Neural Network (ANN) and fuzzy logic (Zhou et al., 2010). 2.6.5 Control Strategies for Energy Flow and Management The dynamic interaction between different energy sources and the load demand can lead to critical problems of stability and power quality, which are not very common in conventional power systems. In addition, the energy management strategy can affect the operation, reliability, cost and lifetime of the proposed hybrid system (Abedi et al., 2012). Therefore, the management and control of energy distribution throughout the proposed hybrid system to assure continuous power supply for the load demand is an essential part of the optimization process. Several authors applied different conventional techniques for controlling various hybrid energy systems. Some authors, such as Park et al. (2004) presented the power compensation system for controlling energy flow through hybrid energy system according to load demand. Valenciaga and Puleston (2005) developed controller for hybrid power systems. The 91
controller designed utilized sliding mode control methods for controlling the hybrid system (Beltran, 2007). In general, the conventional approach utilizes power electronics based DC– DC converter for maximum energy extract from the wind and solar energy resources and control of the complete hybrid system (Reddy and Agarwal, 2007). Besides the conventional method, other advanced controlling techniques based on either fuzzy logic or genetic algorithm exists, for efficient control of energy flow and proper management of the power delivered to the load (Chedid et al., 2000; Senjyu et al., 2006). ElShater et al. (2001) discussed the energy flow and control of a hybrid solar–wind–fuel system. Fuzzy Logic control method was employed to achieve maximum power tracking for both wind and solar energies and to provide maximum power to a fixed DC voltage bus, and controls each of the three energy sources to produce energy at optimum efficiency. DufoLopez and Bernal-Agustin (2005) implemented a genetic algorithm for the design and control of a hybrid solar-diesel system. Senjyu et al. (2006) used genetic algorithm for controlling power generation and distribution, and determined the optimal configuration for a hybrid wind-solar generating systems in residence. Abedi et al. (2012) used the Differential Evolution Algorithm (DEA) accompanied with fuzzy logic method for the optimal power management and design of hybrid energy system consisting of PV arrays, wind turbines, fuel cells, electrolysers, hydrogen tanks, batteries and diesel generators. 2.6.6 State-of-the-art in Optimization Methods for Hybrid Energy Systems Hybrid renewable energy systems are becoming popular for remote power generation applications due to the advances in renewable energy technologies. There has been outstanding interest in optimizing the design and management of hybrid power systems, in order to achieve energy balance between the maximum energy captured and consumed
92
energy (Kalantar and Mousavi, 2010). Nevertheless, the fluctuating renewable energy supplies, load demands and the non-linear characteristics of some components complicate the design of hybrid systems. In addition, the overall assessment of hybrid systems that incorporate renewable and conventional energy sources depends on economic and environmental (such as energy costs and pollutant emissions, respectively) criteria, which are often conflicting objectives. Utilization of Simulation Software Various researchers around the world have made different attempts to solving the problems of hybrid energy systems by using various simulations and optimization software discussed in this literature review. Barley et al. (1997) applied HYBRID2 software in conjunction with a simplified time-series model reported in the analysis of a hybrid wind/photovoltaic system for providing electricity to about one-third of the non-grid connected households in Inner Mongolia. The system configuration was determined based on a subjective trade-off between the cost of the system and the percent unmet load. The result showed that the addition of PV to the wind-only system, in conjunction with an increase of battery capacity, reduced the unmet load by over 75%, with a cost increase of only 22%. HYBRID2 simulates hybrid systems with remarkably high precision calculations, but it does not optimize the system. Dufo-Lopez et al. (2005) applied HOGA program to develop the sizing and management control of a PV-Diesel system. The optimal structure described precisely as the number/type of PV panels, the number battery, the inverter and the diesel generator power. Besides, the study specified the optimal control strategy of the system with its parameters, the total net present value of the system and the different relative costs (such as the fuel cost), and the number of running hours for the diesel generator per year. The computational results
93
demonstrated the economic advantages of the PV-hybrid system. Nevertheless, this approach is, usually, not feasible, since the number of components and strategies may be large, thus resulting in calculation times of the order of months, or even years. However, the environmental issues associated to this type of installations (comprising diesel generator) were not taken into account during the design process. This omission has a significant effect on the optimum design configuration. HOMER computer-based model is the most widely used simulation software for the design of hybrid energy systems in the literature. Several authors including Balamurugan et al. (2009), Nandi and Ghosh (2009), Nema et al. (2010), and Fulzele and Dutt (2012) utilized HOMER for planning and sizing of energy systems and determined the optimum configurations for different combination of off-grid hybrid power systems. Unlike other simulation software, HOMER allows for comparison with different design options based on technical and economic merits, as a result, several studies (Liu et al., 2010; Nandi and Ghosh, 2009; Supriya and Siddarthan, 2011; Yousif, 2012; Talebhagh and Kareghar, 2012; Teoh et al., 2012) have used this tool for the design, management and sizing of grid-connected hybrid power systems. System components sizing with HOMER assumed many simplifications and analysis requires information on resources, financial constraints and control methods. For instance, in grid-connected applications it considered reliable and steady grid supply, and established grid electricity purchased as the sum of net energy generation (the difference between energy produced and load demand) during periods when energy generated is less than the amount demanded. This assumption has a significant impact on the accuracy of results deduced from HOMER, especially within the emerging world where grid electricity is not reliable.
94
Utilization of Optimization Methods Several researchers have also utilized various optimization techniques discussed in the literature for optimum sizing of different hybrid energy systems for different locations around the world. Chedid and Rahman (1997) employed a linear programming technique for optimum sizing of a hybrid wind-solar power system with minimum average production cost of electricity while meeting the load requirements in a reliable manner. Gupta et al. (2009) developed a mixed integer linear mathematical programming model to determine the optimum configuration, including the assessment of the economic penetration levels of photovoltaic array area. The methodology adopted incorporated a constant cost (cost/unit) for each of the proposed resource (small hydro, biomass, PV modules, wind turbine, diesel generator) in the cost objective function so that the ones with lesser unit costs share the greater percentage of the total energy demand. Several authors used heuristic methods to tackle the optimal mix problems by taking into account not only the renewable energy resources, but also the technological characterization, incentives and economic parameters (such as installed and maintenance costs). Katsigiannis and Georgilarkis (2008) utilized the tabu search algorithm (Glover, 1990) for optimal sizing of a small, isolated hybrid power system (with a maximum annual value of electric load of 100kW) using computer codes developed in MATLAB in Chania region of Greece. Though the implementation of the tabu search algorithm is advantageous in terms of its simplicity, the execution time of over 42 minutes is a considerable disadvantage: the convergence time is considerably high. Fung et al. (1993) applied simulated annealing techniques for sizing an off-grid hybrid power system. This approach consists of a random search method similar to the iterative
95
improvement algorithm, but it may accept inferior solutions based on a probabilistic measure. Starting from an initial system design, the total cost was calculated and in the next iteration, the new unit sizing components were perturbed, while satisfying all the constraints. The new solution was only accepted if the new cost was lower compared to previous solution. Ekren and Ekren (2010) also utilized the simulated annealing algorithm in optimizing a photovoltaic/wind hybrid system with battery storage. The objective was to determine the optimal configuration of a hybrid system with minimum runtime as compared to the response surface methodology (Ekren and Ekren, 2008). The study utilized historical hourly average solar radiation and wind speed data and random measurement of the GSM BTS load situated on the campus area in Turkey. Simulation results indicated that the heuristic tool converges in a reasonable runtime and the hybrid system can effectively power the GSM BTS site. However, the study neglected the seasonal variation of the energy demand, which is an essential factor for accurate determination of the optimum configuration of the power system. The application of the particle swarm algorithm to a wide range of search and optimization problems exist in the literature. Kaviani et al. (2009) optimized a hybrid wind-photovoltaicfuel cell production system using particle swarm algorithm, with the objective of minimizing the annual cost of the hybrid system subject to a reliable supply for the demand. Hakimi and Moghaddas-Tafreshi (2009) demonstrated the ability of the particle swarm algorithm in optimizing the total cost of a stand-alone hybrid power system in Kahnouj area, south-east Iran. The system composed of fuel cells, wind units, electrolysers, a reformer, anaerobic reactor and hydrogen tanks. The system used biomass as an available energy resource, in supplying the load demanded. The result showed that the imposed constraints of the problem were satisfied at optimal cost. Hakimi et al. (2011) then used PSO for sizing a hybrid power system by minimizing the total cost of the system and satisfying the requirement of the 96
residential area. Previous studies including Angeline (1988a, b), Hu and Eberhart (2002), Lovbjerg (2002), Tsou and MacNish (2003), and Ratnaweera et al. (2004) have also used PSO to solve related problems. Nevertheless, based on the available literature reviewed in the field of economic dispatch using PSO, Mahor et al. (2009) concluded that further improvement in PSO algorithms are required, as present versions of PSO have slower convergence at later stage. In addition, the PSO is unable to provide the optimum solution for real-time scheduling problems. Several authors have used the GA for optimal sizing of renewable energy applications. Koutroulis et al. (2006) proposed a methodology to calculate, among a list of commercially available system devices, the optimal number and type of units to ensure a year operation with a minimized system cost, subject to the constraint that the load energy requirements were completely covered, resulting in zero load rejection. The primary objective was the minimization function, which was implemented using genetic algorithms. From a set of individuals which were composed by the system components, the algorithm chose the best combination in the evolution process so that after the maximum iterations defined by the user, the best set of components with the lowest cost were chosen. Yang et al. (2008) also developed an optimization technique with genetic algorithms but, instead of attaining to reach zero load rejection, they used the LPSP and the annualized cost of system concepts. The decision variables included in the optimization process were, the PV module number, wind turbine number, battery number and for a more complex and complete algorithm process, the PV slope angle as well as the wind turbine installation height. This study extended by Yang et al. (2009) supplied the required power for a telecommunication relay station from renewable energy sources on a remote island (Dalajia Island) along the south-east coast of China. The design considers for the normal operation of the telecommunication station GSM 97
base station RBS2206, energy consumption of 1.3 kW (24 V AC) and 0.2 kW (24 V DC) for microwave communication. The study reported based on the one year time-series field data studied that an optimum configuration consisting of 120 kWh (5000 Ah, 24 V) battery capacity, 12 kW WTG, and 7.8 kW PV array inclined at 29.5o enabled good system behaviours and satisfactory performance. Ould et al. (2010) proposed a multi-objective GA for sizing a hybrid solar-wind-battery system with the goal of minimizing the annualized system cost and the loss of power supply probability. Like in Koutroulis et al. (2006) and Yang et al. (2008; 2009) the results showed the robustness of the genetic algorithm in finding global optimum solutions. Other authors in the literature including Hassan et al. (2004), Xu et al. (2005), Senjyu et al. (2007), Lagorse et al. (2009) and Nafeh (2011) applied the genetic algorithms. The search for the global optimum system configuration with relative computational simplicity led to the use of advanced artificial intelligence or hybrid search (combination of different heuristics) techniques. Cai et al. (2009) proposed an approach for supporting community scale renewable energy management. The methodology was on the basis of interval linear, two-stage, and superiority-inferiority based fuzzy-stochastic programming. This allowed the simulation process of unit sizing to incorporate complexities and multiple uncertainties. The aim was to develop a method applied to a case of long-term renewable energy planning for three communities. The results proved the pre-objectives of the work, by generating decision alternatives and thus helping decision makers to identify their desired policies under various economic-reliability system constraints. Abedi et al. (2012) proposed the use of the differential evolution algorithm accompanied, with fuzzy technique, to reduce the overall cost, unmet load, and fuel emission of a mixed-integer non-linear multi-objective optimization problem of a hybrid energy system. This method was tested on a system with 98
some widely used generators in off-grid systems, including wind turbines, PV panels, fuel cells, electrolysers, hydrogen tanks, batteries, and diesel generators. The optimum solution, including design parameters of system components and the monthly parameters of energy management strategy for adapting climatic changes during a year have been obtained. In order to achieve effective power utilization from renewable energy sources, the optimal monthly tilt angles of PV panels and the optimal tower height for wind turbines are calculated. Their results demonstrate the ability of the proposed method for hybrid energy systems A literature survey undertaken in this study for reviewing the system design and performance assessments of energy systems indicates that limited work is available for grid-connected applications. Some authors for example, Kornelakis and Koutroulis (2009) analyzed the optimization of PV grid-connected systems as follows. Given a list of commercially available system devices, they selected the optimum number and type, and the optimum values of the PV module installation details, to maximize the total net economic benefit achieved during the system‘s operational lifetime. Mohammed and Koivo (2010) analyzed and claimed to have attained the online optimal power management for a grid-connected hybrid energy system, capable of mutual exchange of energy with the grid, comprising wind turbine, microturbine, diesel generator, PV array, fuel cell, and battery storage. They minimized cost by selecting the corresponding systems-configuration and/or operation strategy. The study assumed the optimization problem as a single objective function, and considered all objectives such as fuel emissions in terms of cost. Pollutant emissions represented in the form of cost, and limited by their inclusion in the cost function made the results depended on the cost coefficient assigned to the emissions. The fuel emissions were factorized into three main gas components and values for each gas separately presented. Nayeripour and Hoseintabar 99
(2011) proposed a comprehensive, dynamic modelling and power management of hybrid power generation including renewable power generation and energy storage system in gridconnected applications. The foregoing studies ignored the grid model in grid-connected applications and assumed a reliable and steady grid supply. This study observes that more research and developmental effort is needed for improving the performance of hybrid energy systems, by establishing techniques for accurately predicting their output and reliably integrating the utility grid with other renewable or power generation sources. 2.7
Research Area Climate
The climatic condition of a given location influences the renewable energy; hence it is site/location dependent. Nigeria is located within the Equator and the Tropic of Cancer, with main latitude and longitude of 10°N and 8°E respectively. Although the latitude of Nigeria falls within the tropical zone, its climatic conditions are not entirely tropical in nature. It varies in most parts of the country. For example, the climatic condition is arid in the north, tropical at the centre and equatorial in the south. Figure 2.9 is the map of Nigeria showing the selected locations chosen for this study, representing the different climatic conditions. These locations, representing the six geopolitical zones, should effectively tackle the energy related problems of the GSM BTS sites in Nigeria.
100
Study locations
Figure 2.9: Map of Nigeria showing the study locations. 2.7.1 Sokoto Sokoto City is the capital city of Sokoto State located in the north-western zone of Nigeria. It lies between latitude 13.06°N and longitude 5.25°E. The entire area is very hot, but maximum daytime temperatures for most of the year are generally below 40 °C. The warmest months are February to April when daytime temperatures can exceed 45 °C. The rainy season is from June to October during which showers are a daily occurrence. The showers rarely last long and are a far cry from the regular torrential rain known in wet tropical regions. From late October to February, during the cold season, the climate is dominated by the ‗Harmattan‘ wind blowing Sahara dust over the land. The dust dims the sunlight thereby lowering temperatures significantly. The long-term average monthly daily values varies from 17.1°C to 27.3°C (for minimum air temperature) and 31.2°C to 41.1°C (for maximum air temperature), with corresponding values in the range of 16.5–79.3 % (relative humidity), and wind speed 8–14.6 m s-1, measured at 10 m above the surface of the earth (NIMET, 2013). 101
2.7.2 Maiduguri Maiduguri is the capital city of Borno State located in the north-eastern zone of Nigeria. It lies between latitude 12.0 °N and longitude 13.33 °E. The climate of Maiduguri is generally semi-arid, with moderate variations in temperatures. The average monthly maximum temperature is highest prior to and during the onset of the rains in April and the lowest during the peak rained period of August (Mala et al., 2012). The long-term (1991 – 2012) monthly average daily maximum air temperature ranges from 31.3 °C to 41.5 °C with a corresponding minimum values within the range of 12.7 °C to 26.7 °C. The corresponding values for the relative humidity and wind speed are in the range of 23.4 – 84.8 %, and 7.5 – 13.9 m s-1 respectively (NIMET, 2013). 2.7.3 Abuja Abuja, Nigeria‘s capital city is located in the north-central zone of Nigeria. It lies between latitude 9.08 °N and longitude 7.53 °E. Based on the Koppen climate classification, Abuja features as tropical savannah (wet and dry) climate and experiences three weather conditions annually. This includes a warm, humid rainy, and a blistering dry season. The rainy season begins from April and ends in October, when daytime temperatures reach 28 – 30 °C and night-time lows hover around 22–23 °C. The rainy season peaks in September, during which there is abundant rainfall in the form of heavy downpours with an annual rainfall of about 1500 mm. Dry season, from December to March, is hot and dry. A continental tropical air mass laden with dust from the Sahara Desert prevails throughout this period. In between these two periods, there is a brief interlude of ―Harmattan‖ occasioned by the north-east trade wind, with the main feature of dust haze, intensified coldness, and dryness. The average monthly maximum air temperature ranges from 28.7 °C to 37.5 °C with a corresponding
102
minimum values within the range of 18.0 °C to 24.5 °C (NIMET, 2013). The monthly average relative humidity varies from 38–87 %, while the monthly average daily wind speed varies from 3.8 – 5.6 m s-1. 2.7.4 Ikeja Ikeja is the capital city of Lagos State located between latitude 6.58 °N and longitude 3.33 °E in the south-western zone of Nigeria. This area has a tropical wet and dry climate and experiences two rainy seasons, with the heaviest rains falling from April to July and a weaker rainy season in October and November. There is a brief relatively dry spell in August and September and a longer dry season from December to March. Monthly rainfall between May and July averages over 400 mm, while in August and September it is down to 200 mm and in December as low as 25 mm. The main dry season accompanied by ―Harmattan‖ winds from the Sahara Desert, between December and early February can be quite strong. The long-term average monthly daily values varies from 28.2 °C to 32.2 °C (for maximum air temperature) and 24.4 °C to 27.0 °C (for minimum air temperature), with corresponding values in the range of 75 – 83 % (relative humidity), and 3.9 – 5.3 m s-1 (NIMET, 2013). 2.7.5 Enugu Enugu is the capital city of Enugu State located between latitude 6.5 °N and longitude 7.5 °E in the south-eastern zone of Nigeria. The climate is humid tropics with two seasons (rainy and dry). The rainy seasons begins in April and lasts till October. Double maxima, with the first maximum in June and the second in September also characterized the climate. There is thus a ―little dry season‖ in-between known as ‗August Break‘ brought about by the seasonal north and southward movement of the Inter-Tropical Convergence Zone. Annual rainfall varies from 1,990 to 2,200 mm. The long-term monthly average daily values for minimum 103
and maximum air temperatures vary from 19.7 °C to 24.7 °C and 29.6 °C to 35.5 °C respectively. The corresponding values for the relative humidity and wind speed are the range of 55 – 85 %, and 5 – 9 m s-1 respectively (NIMET, 2013). 2.7.6 Benin City Benin City is the capital of Edo State located between latitude 6.45 °N and longitude 3.40 °E in the south-south zone of Nigeria. The area characterized as tropical rain forest experiences two distinct seasons: the rainy and the dry seasons. The rainy season spans from April to October with temperature ranges from 25 °C to 29 °C. The rainy season peaks in July with a brief dry spell (i.e., break) in August. The average rainfall during this period is about 1500 – 2000 mm with an annual rainfall of about 2074 mm. The dry season spans from November to March with temperature ranges from 24 °C to 28 °C significantly marked by the cool ―harmattan‖ dusty haze from the north-east trade winds. However, the average rainfall during this season is much lower (usually no more than 300 mm) compared to the rainy season. The long-term (1991 – 2012) average monthly daily minimum air temperature ranges from 22.7 °C to 24.6 °C with a corresponding maximum values within the range of 28.3 °C to 34.7°C. The annual monthly air temperature within this period varies from 25.5 °C to 29.4 °C. The corresponding values for the relative humidity and wind speed are the range of 74– 91 %, and 3.5–5.4 m s-1 respectively (NIMET, 2013). 2.8
Summary
The significance of energy cannot be overemphasized, as it is essential in virtually all sectors of the economy. As part of research efforts in solving the present electricity related problems in the mobile telecommunication sectors of developing countries, this chapter comprehensively reviews the existing energy infrastructure and underlying principles of 104
various technologies. It indicates in sufficient details shortcomings in the development and application of theories and concepts for deployment using Nigeria as a case study. The assessment of the renewable energy technologies is essential for effective design, planning, and management of renewable energy system as this could encourage extensive and optimum utilization of the renewable energy resources in virtually all sectors of the economy. This study considers the mobile telecommunication sector because it is expanding and most GSM BTS sites in the country rely on the use of fossil-fuelled generators either as supplements to the utility grid or exclusively in remote sites neglecting both the socioeconomic and the environmental implications, which could be very alarming. The use of different energy sources allows for improved system efficiency and reliability of the energy supply and reduces the energy storage requirements compared to systems comprising a single renewable energy source. The combined use of different energy sources depends on the enabling technology. The literatures surveyed shows that wind and solar are commonly used energy sources in recent years, because they are environmentally friendly and technically viable options. Wind and solar are promising power generating sources, due to their availability and topological advantages for local power generations especially in remote locations. However, in order to efficiently and economically utilize the different energy resources for a hybrid energy system it is necessary to apply an optimum sizing method. Optimum sizing of hybrid energy system is performed either by using simulation software or optimization methods. Among various simulation software available in literatures, HOMER is the most common and widely used tool for evaluating the performance of hybrid energy systems. The extensive use of HOMER is perhaps due to ease of availability of software and the simplicity of its operation and use. Nevertheless, none of the studied software can optimize the configuration for a grid-connected hybrid system, considering the 105
random nature of the utility grid in the emerging world. In addition, the user can not intuitively adjust/select the appropriate parameters/components for a system, as the algorithms and calculations are neither visible nor accessible. Consequently, there is a need for an alternative method of designing the optimum configuration of hybrid energy systems. A thorough literature survey undertaken in this study indicates that the modelling, analysis and optimization of energy system is essential for effective design, planning, and management of a sustainable energy system. The modelling approach differs from one article to another and there is no standard approach that can give reliable results. Among the various optimization techniques, the heuristic optimization tools, particularly genetic algorithm is frequently used for optimum sizing of hybrid energy systems probably for its simplicity and ability to produce satisfactory results. In addition, the dynamic interaction between the various energy sources and energy demand can lead to system instability and this could reduce the reliability and quality of the power supplied. Thus energy control strategy is important in the optimal design and efficient utilization of hybrid energy systems, as it could influence the available power supply and the overall lifetime of the system components. Some studies applied the raw weather data for one year and some have used the statistical weather data for prediction after imposing the probability density function. These data are commonly used for assessing the performance of energy generation system. Nevertheless, detailed analysis of renewable energy resources and resource allocation based on load demand is essential for the optimum design of hybrid energy system. Accurate sizing of the hybrid energy system in terms of economical analysis and energy requirements is critical. Therefore, finding the optimum system configuration involves deciding on the mix of energy
106
resources, the size or quality of each component and the power management strategy the system should use, with the minimum economic (cash) and socio-technical cost. The sizing and optimization technique used until now, neglect the need for a grid supply model in grid-connected applications. The common practice is to assume an ideal grid distribution system. As earlier mentioned, for a developing country where grid electricity is not always available, a model to account for uncertainties such as power outages, unstable and unsteady supply is essential in determining the true optimal configurations of such an energy system. Furthermore, no single article within the reach of the literature reviewed, modelled, optimized and analyzed the grid-connected hybrid (wind-photovoltaic) power generation system considering the uniqueness of the utility grid in an emerging country. Consequently, this lack in the literature about modelling and optimization of a grid-connected hybrid system justifies the need for the present research.
107
CHAPTER THREE 3.0 3.1
METHODOLOGY Development of the Hybrid Energy System Model
This study applied theoretical approach in the modelling, simulation and validation of the developed system, as shown in Figure 3.1. The availability, dynamism and technical viability of energy resources in Nigeria, informed the choice of the proposed system architecture. The diesel generator unit was included only for the purpose of comparison.
GSM BTS site Grid energy supply system (GESS) Wind energy conversion system (WECS) Photovoltaic conversion system (PVCS)
Energy conversion and management system
Super-capacitor/Battery bank Diesel generator unit (for comparison only) Energy transfer
Control signal
Figure 3.1: Architecture of the proposed hybrid energy system for electricity supply to GSM BTS sites The grid model utilized a probability prediction technique–uniform distribution function, due to the random nature of the Nigerian electric power grid. The WECS modelling applied stochastic techniques –piecewise third order polynomial using power profiles of wind turbine
108
generators (WTGs) supplied by the manufacturer. The reason is that wind turbines have different power output performance curves. As stated in the literature, a WTG model should be developed according to its power output performance curve supplied by the manufacturer. The stochastic method allows the random nature, which reflects the chronological characteristics of the site resources, such as wind, to be modelled. The wind speed data collected are adjusted to the GSM BTS tower height using power law. The energy balance method proposed by Duffie and Beckman (2006) is applied for modelling the effect of air temperature on the power generated by the PV system. The fixed optimum tilt angles for installating the PV array is determined using the Hay-Davies-Klucher-Reindl (HDKR) model proposed by Duffie and Beckman (2006). The energy storage model includes the super-capacitor and battery models, which are connected in parallel to the DC bus. In particular, two DC-DC converters, which connect the DC bus to the battery, allow bidirectional transfer of power. The hybrid (super-capacitor/battery) energy model described as a function of the state of charge of super-capacitor/battery is defined in terms of equilibrium potential, charge/discharge rate and storage capacity. The hybrid energy storage system is chosen due to its ability in reducing battery ageing and the improvement in energy efficiency when the system operates in critical climate conditions (Santucci et al., 2014). Moreover, the use of super-capacitor/battery as a storage device in HES has already been experimentally validated by Van-Voorden (2008). The energy conversion and management system consists of power electronics and supervisory controllers. Power electronics were included for power conversion and stabilizing unpredicted voltage and frequency fluctuations. The supervisory controller enables effective management of power flow for improved stability. The controller coordinates all
109
power electronics. Control signals are defined by the controller based on the technical constraints specified by the operation strategy. The control ensures the generation of proper power level between power sources, and the distribution for reliably satisfying the energy demand. The optimization problem is treated as a single objective problem by considering all objectives (economic and technical) in terms of cost of energy. The formulated optimization problem is solved using a hybrid Genetic Algorithms and Pattern Search (h-GAPS) method. The h-GAPS based technique combines the genetic and the pattern search algorithms in the search step. It is well-known that the stochastic population-based algorithms like Genetic Algorithm (GA) are good at identifying promising areas of the search space (exploration). The GA has the ability to adapt to nonlinearities and discontinuities found in power systems. The exceptional ability for solving complex-design optimization problems is due to its intuitiveness, ease of implementation, and capability in handling both discrete and continuous variables, and non-linear objective and constraints functions, without requiring gradient information that is typical of complex engineering systems. In contrast, the Pattern Search (PS) method specifically is a more coordinate search method, which guarantees convergence to stationary points from arbitrary starting points. In other words, pattern search method is useful at improving approximations to the minimum (exploitation). The hybridization of global and local optimizers (GA and PS) can provide a more efficient trade-off between exploration and exploitation of the search space. The hybridization can help to guarantee that the global optimum solution is selected.
110
3.2
Modelling of the Hybrid Energy System Components
The architecture of the HES shown in Figure 3.1 is made up of three primary sources of energy (utility grid, wind and solar), energy conversion and management system, and energy storage system. And the proposed hybrid energy system consists of five unit components. These are the Grid Energy Supply System (GESS) unit, Wind Energy Conversion System (WECS) unit, Photovoltaic Conversion System (PVCS) unit, Power electronic (conversion) unit, and Energy Storage (ES) unit. 3.2.1 Modelling of the Grid Energy Supply System (GESS) In order to account for the erratic nature (irregularity) of the Nigerian electric power grid supply, the outcome of the grid supply was likened to the outcome of a coin, i.e., randomly. If the number of possible outcomes of supply (ns) is expected for a defined period η (= 1, 2, . . . , N), representing each minute of the year, i.e., the time step is minutely, then the probability of grid supply can be represented by the uniform-distribution function as stated (Stroud, 1996):
Py gs ( ) pdf ( ; , 2 )
1
2
. exp
1 2
2
,
(3.1)
where, pdf is the probability density function, µ is the average and is the standard deviation. The expectation of the grid supply at time η is Xp gs ( ) ns Py gs ( ) ,
where,
(3.2)
ns Py gs ( ) is the absolute value of ns Py gs ( ) . If Vave is the average grid-supply
voltage (V), then the average grid-supply voltage profile (series) is
111
Vave ( ) Vave Xp gs ( )
(3.3)
Vgs ( ) Vave ns Py gs ( )
(3.4)
Under real conditions, the grid-supply voltage varies from the average. That is; there are variations, for instance, across the minutes of the day and days of the year, from these average values. To account for these variations, let‘s assume that the average voltage profile varies with a minutely random (noise) factor m and a daily random factor d . The real-time grid-supply voltage at anytime η becomes Vgs ( ) Vave ns Py gs ( ) (1 m d )
(3.5)
where, m and d are randomly drawn from the normal-distribution (Gaussian) function with average of zero and standard deviation equal to the minute noise input value and the daily noise input value respectively. The minutely and daily random factor can be expressed as follows:
m d
1 2
. exp
x
2
(3.6)
2
where, x (= η/nf ) is the standardized normal variant, and nf is the minutely or daily noise factor. A minutely and a daily noise factor of 0.15 and 0.20 respectively, and an average gridsupply voltage of 204 V are used in this study for simulating the grid voltage profile. The values assumed are based on the findings of the UNDP-GEF (2013) draft report on the main characteristic of the Nigerian grid electricity system discussed in the literature (see section 2.3).
112
The normal distribution is the most prominent studied probability distribution. This is because the normal distribution arises from the central limit theorem, which states that ―under mild conditions the mean of a large number of random variables drawn from the same distribution distributes approximately normally, irrespective of the form of the original distribution.‖ Hence, it has exceptionally wide application in, for example, sampling. In addition, the normal distribution is very tractable analytically, i.e., a large number of results involving this distribution can be derived in explicit form. The real-time grid supply voltage profile for a reliable grid can be deduced, from the longterm values, using stochastic methods. However, where such analysis is not readily available, grid-supply voltage at any time η can be approximated by the equation: V gs ( ) Vave (1 m d ) .
(3.7)
For safety reasons, the grid-supply voltage must lie within the Safe Operating Voltage Region (SOVR) of the GSM BTS site before it can effectively power the GSM BTS site. If the SOVR of the GSM BTS site varies between Vmin and Vmax, the DC grid power (kW) available to the GSM BTS site is expressed as follows: Pl ( ) Pga ( ) sta ret 0
for Vsta,min V gs ( ) Vsta,max
,
(3.8)
else
where, Pl (kW) is the required load capacity that can respond to sudden rise in power demand of the GSM BTS site, grid-supply,
sta is the efficiency of the voltage stabilizer used in stabilizing the
ret is the rectifying efficiency, Vsta,min and Vsta,max are the minimum and
maximum input voltage of the voltage stabilizer (V) respectively. Equation (3.8) indicates 113
that the grid-power supply is only available to the GSM BTS site when its supply voltage is within the usable voltage range without causing any adverse effect. The energy (kWh) drawn from the grid at any time η is given by the equation:
E ge ( )
Dmo Pga ( ) 60 Pgs ( )
,
(3.9)
60
where Pgs is the grid-power supplied to the GSM BTS site (kW), ∆η is the time step (considered here as 1 min), and Dmo is the decision variable that specifies the mode of operation. Dmo is set to the default value of zero for stand-alone or off-grid operation and one for grid-connected operation. The power rating (size) of the stabilizer is defined by the equation:
Pr ,sta
Pgs,max
(3.10)
sta
where Pgs,max is the maximum minute grid-power (kW) supply during the period N, and Pr,sta is rounded to the greater integer (kW) for safety reason. 3.2.2 Modelling of the Wind Energy Conversion System As discussed in the literature, modelling of the effect of altitude is essential for accurate prediction of wind energy conversion systems. The reason is that altitude affects air density, which in turn affects the output of Wind Turbine Generators (WTGs).
114
Effects of Altitude on WTG Power Output From the ideal gas law (Huffman, 1999; Laugier and Garai, 2007), air density is defined as:
Ps , RT *
(3.11)
where ρ is air density (kg m-3), Ps is the pressure (Pa), R is the gas constant (J kg-1 K-1), and T* is the absolute temperature (K). Using the 1976 International Standard Atmosphere (ISA) standard sea level conditions (i.e., standard pressure, Pso = 101.325 kPa, and standard temperature, To* = 288.15 K), the ISA air density, that is, the density of dry air (Laugier and Garai 2007), ρo is
o
Ps o . RTo *
(3.12)
Air density ratio, that is, the ratio of the actual air density, Eq. (3.11) to ISA air density, Eq. (3.12) is
Ps To * . o Ps o T *
(3.13)
As a result, altitude affects both pressure and temperature. Using the ISA simplifying assumption (up to an altitude of 11,000 m), temperature decreases linearly with altitude according to the equation (Trani, 2013): T* = To* − Bzhh,
(3.14)
where, B is the temperature lapse rate (K m-1), and zhh is the altitude (m). The air pressure depends on the altitude according to the equation (Trani, 2013): 115
Bz hh Ps Ps o 1 To *
g / RB
(3.15)
,
where, g is the gravitational acceleration (m s-2). Substituting Eq. (3.14) and Eq. (3.15) into Eq. (3.13),
To * Bz 1 hh o To * Bz hh To *
g / RB
.
To * To * Bz hh
(3.16)
1 g / RB
Equation (3.16) shows that the air density ratio is only a function of altitude, zhh, since B, g, R, and To* are all constant. Denoting the air density ratio by the altitude factor zf, Eq. (3.16) can be simplified as follows:
zf
To * To * Bz hh
1 g / RB
.
(3.17)
z f o .
(3.18)
Proposed Wind Energy Conversion System (WECS) Model The study suggested that the WTG should be mounted on the GSM BTS tower due to land constraint as well as to save additional cost of installing a self-supporting WTG tower. Hence, the proposed WECS consists of only one WTG. However, four different sizes (Pwtj = 1, 2, 3, and 5 kW) of wind turbines described in Table A1 (appendix A) were considered. The purpose is to ensure efficient utilization of the available wind energy for the study sites.
116
The output power (kW) of a WTG at time η can only be computed accurately by using its characteristic curve expressed as follows:
0 2 3 a1 a11v hh ( ) a12v hh ( ) a13v hh ( ) a a v ( ) a v 2 ( ) a v 3 ( ) 22 hh 23 hh 2 21 hh 2 3 a3 a31v hh ( ) a32v hh ( ) a33v hh ( ) Pwt outj (v hh ) . . . 2 3 a n a n1v hh ( ) a n 2 v hh ( ) a n 3 v hh ( ) 0
for v hh ( ) vci for vci v hh ( ) v1 for v1 v hh ( ) v 2 for v 2 v hh ( ) v3 ,
(3.19)
for v n 1 v hh ( ) vco for v hh ( ) vco
where, vci is the cut-in wind speed (m s-1), vco is the cut-out wind speed (m s-1), vhh is the wind speed (m s-1) adjusted to GSM BTS tower height, and aij is the piecewise polynomial coefficient which can be determined using MATLAB curve fitting tool. Equation (3.19), the characteristic of the WTG is calibrated by fitting the practical output characteristic curve using least squares method. In order to guarantee the fitting accuracy, a minimum of three 3rd order polynomial expressions are used. Including the effect of altitude on air density on the output of the WTG as well as unavailability due to maintenance of wind turbine, the proposed WECS model under real temperature and pressure becomes
Pwecs ( ) z f N wt j Pwt outj (1 FOR ) ,
(3.20)
where zf is the altitude factor, αFOR is the forced outage rate of the WTG – with a typical value of 4% (Gao, 2006), and (1– αFOR) is the probability of the WTG being operational. Forced outage rate is one of the methods used in power system engineering to determine the 117
probability of unavailability of a generating unit at some distant time in the future, for instance, due to maintenance. Nwtj specifies the number and type (rated capacity) of turbine generators to be selected. The proposed sizing model discussed in section 3.9 is used to select the rated capacity of the wind turbine (Pwtj) that can optimize the wind energy of a given site. The wind energy generated (kWh) by the HES at time η is stated as:
E wg ( )
Pwecs . 60
(3.21)
where, ∆η (= 1 min) is the time step. 3.2.3 Modelling of the Photovoltaic Conversion System The power generated by the PV array depends on a number of factors, such as the operational constraints of the solar cell, the solar array arrangement, and the atmospheric conditions at the site (location) at a given time. Correct prediction of the power generated by the PV array, requires accurate determination of the intensity of the solar irradiation at the particular location and PV array temperature (Chokmaviroj et al., 2006; Duffie and Beckman, 2006). In order to determine the best model for estimating the monthly average daily solar radiation on a horizontal surface in Nigeria, besides the established methods discussed in the literature, which are believed to be universally applicable, the following regression relationships were considered.
KT bo b1 S R b2 TR b3 TMAX b4 C , K D bo b1 K T b2 KT2 ,
(3.22)
(3.23)
118
K D bo b1 K T b2 C ,
(3.24)
where KT is the monthly average daily clearness index, KD is the monthly average daily diffuse fraction, TR is the monthly average daily air temperature ratio, TMAX is the monthly average daily maximum air temperatures (oC), Ĉ is average total cloud cover during daytime observations (octa), and bi is the model parameters (coefficients) determined using regression techniques. The proposed relations (Eq. (3.22) – Eq. (3.24)) were deduced on the basis of various correlations between solar radiation and other meteorological parameters (using monthly average daily data sets discussed in section 3.7.3). Clouds play a crucial role in the transfer of energy between the surface and the atmosphere. The inclusion of cloud fraction could significantly improve the accuracy of solar radiation prediction models (Muneer and Munawwar, 2006). The method for determining the amount of solar radiation that arrives on the earth at the PV array‘s location is discussed in section 3.8. PV Cell Temperature Based on the energy balance approach proposed by Duffie and Beckman (2006), the method for estimating the PV cell temperature deduced in terms of global irradiance and air temperature is given as (Diaf et al., 2007):
Tcl ( ) Tave ( ) I t ( ) uh
,
(3.25)
where Tcl is the minutely average PV cell temperature (°C), Tave is the minutely average air temperature (°C), It is the minutely average global irradiance incident on a tilted surface (kW m-2), α is the solar absorbance of the PV array, ξ is the solar transmittance of the cover 119
over the PV array, and uh is the coefficient of heat transfer (lost) to the surroundings (kW m-2 °C-1). It is a difficult task to measure the value of α ξ/uh directly instead manufacturers report the Nominal Operating Cell Temperature (NOCT) (Chedid and Saliba, 1996; Duffie and Beckman, 2006). The NOCT is the cell temperature, which results at an incident radiation of 0.8 kW m-2 and air temperature of 20 0C, and α ξ/uh is computed using the equation (Jalilzadeh et al., 2010): uh
where Tcl,
Tcl, NOCT Tave, NOCT
NOCT
I t , NOCT
,
(3.26)
is the nominal operating cell temperature, which ranges from 40–70 0C
(Shakyaa et al., 2005), Tave, NOCT is the air temperature at which the NOCT is defined (20 °C) and It, NOCT is the solar radiation at which the NOCT is defined (0.8 kW m-2). Substituting Eq. (3.26) into Eq. (3.25), the cell temperature becomes: Tcl ,NOCT 20 . Tcl ( ) Tave ( ) I t ( ) 0.8
(3.27)
In the absence of measured data, the average minutely air temperature Tave (η), was estimated from the monthly average daily air temperature (NIMET, 2013) using the Erbs‘ model (Erbs et al., 1983). Photovoltaic Conversion System Model The power generated (kW) by the PV array at time τ, with respect to solar radiation, can be calculated using the equation (Hocaoglu et al., 2009a):
120
PPV ( ) N pv pv cf Amod I t ( ),
(3.28)
where pv is the PV generator efficiency, ηcf is the efficiency of coupling, Npv is the number of PV modules, Amod is the area of a single PV module (m2), and It is the minutely global irradiance incident on the tilted PV (kW m-2). When a perfect maximum power point tracking device is used, the PV array efficiency varies linearly with temperature according to (Habib et al., 1999):
PV mp, STC 1 t po Tcl Tcl, STC ,
(3.29)
where ηmp, STC is the maximum power efficiency under standard test conditions, Tcl, STC is the cell temperature under standard test conditions (oC) and tpo is the temperature coefficient of power (efficiency) expressed in percentage per degree (% 0C-1). The temperature coefficient of power is typically negative, meaning that the efficiency of the PV array decreases with increasing cell temperature. Its magnitude ranges from 0.4–0.6 % 0C-1 for silicon cells (Diaf et al., 2007). The power generated (kW) by the PVCS at any time η is estimated as follows:
Ppvcs ( ) mp , STC cf N pv Amod I t ( ) 1 t po Tcl ( ) Tc , STC , mp , STC cf Sz pv I t ( ) 1 t po Tcl ( ) Tc , STC .
(3.30)
where Szpv (= Npv Amod) is the total area (m2) of the PV array, which is determined by the optimization model discussed in section 3.9. As suggested by Duffie and Beckman (2006), the average value for temperature coefficient of efficiency, tpo is assumed to be - 0.5% 0C-1.
121
The main technical specifications of a typical PV module considered are given in Table A2 (appendix A). By dissipating the heat from the PV array and lowering the operating temperature, the effect of temperature on the PV array could be reduced. Under such conditions, the PV array efficiency of Eq. (3.30) becomes the reference efficiency at standard test condition and the generated power of the PVCS reduces to Eq. (3.31).
Ppvcs ( ) mp, STC cf Sz pv It ( ) .
(3.31)
The solar energy produced (kWh) by the HES at any time η is given by the equation:
E sg ( )
Ppvcs ( ) . (3.32)
60
where, ∆η (= 1 min) is the time step. At any time η, the hybrid power generated in DC (kW) defined by Eq. (3.33) supplies the energy demand, and charges the SC/battery (sb) bank provided the constraints imposed by the proposed energy management strategy is satisfied, else the generated power (i.e., excess generation, which is not required by the GSM BTS site) is dumped. Phg ( ) Pgs ( ) Pwecs ( ) Ppvcs ( )
(3.33)
Pgs ( ) Prg ( )
where Pgs, Pwecs, Ppvcs, and Prg, are power supplied by the grid, wind energy conversion system, photovoltaic conversion system, and renewable energy, respectively. All supplied powers are expressed in kilo-watts (kW).
122
The hybrid energy generation at time η is defined by the following equation. E hg ( ) E ge ( ) E wg ( ) E sg ( )
(3.34)
E ge ( ) E rg ( )
where Ege, Ewg, Esg, Erg, and Ehg, are the grid, wind, solar, renewable, and hybrid (total) energy production respectively, all energy values are expressed in kilo-watt-hour (kWh). 3.2.4 Modelling of the Power Electronics (Conversion Unit) The energy conversion and management system uses power electronics, which consist of power converters (inverter and the rectifier), to convert energy levels (from DC to AC or AC to DC respectively). In order for the inverter to adequately meet the desired GSM AC load demand Pd(ac) (η) at time η, its input power (kW) is modelled as follows:
Pinv ( )
where
Pd ( ac) ( )
inv
,
(3.35)
inv is the inverter efficiency and Pd(ac) is an AC power demand (kW). The power
rating (size) of the inverter (kW) is given by Eq. (3.36).
Pr ,inv Pinv,max
(3.36)
where Pinv,max is the maximum inverter input power drawn during the entire period N. The power rating (Pr,inv) is rounded to the greater integer (kW) for safety reason. The main technical specifications of a typical power inverter are described in Table A3 (appendix A).
123
The rectifier input power (kW) is expressed as stated:
Pret ( )
where
Pl ( ) ret
(3.37)
ret is the rectifier efficiency and Pl is the required load that can response to any
sudden rise in power demand. The power rating (size) of the rectifier (kW) is expressed as:
Pr ,ret
Dmo Pl ,max
ret
(3.38)
,
where Pl,max is the maximum required load demand during the entire period N and Dmo is the decision variable, which specifies the mode of grid operation. 3.2.5 Modelling of the Energy Storage Unit The energy storage unit consists of the super-capacitor and battery banks, which are connected in parallel with the DC/DC converters. Super-Capacitor (SC) Bank The nominal energy storage capacity (kWh) of an SC module is expressed as follows:
Esup
CVsc2 , 2 3.6 10 6
(3.39)
where C and Vsc are the rated capacitance (F) and voltage (V) of the SC module respectively. The total number of SC modules required in the SC bank is given by Eq. (3.40).
124
N sc
Sz sc , Esup
(3.40)
where, Szsc is the size or total energy storage capacity (kWh) of the SC bank and Nsc is rounded to the greater integer for safety reason. To enable the SC bank respond to peak transient energy demand (Epk), the value of Szsc must be greater than or equal to Epk. The size of the SC bank chosen is given by equation:
Sz sc
Pd ,max . 60 sb
,
(3.41)
where Δη is the simulation step time (= 1 min) and ηsb is the SC/battery charge/discharge efficiency. If the operating voltage of the selected SC module is different from that of the battery bank, it is the task of the designer to connect the SC modules in series-parallel such that they operate on the required voltage (which is 48 V in present study). Battery Bank Since the operational voltage of the studied BTS equipment and consequently the inverter is 48 V, it is necessary to develop the optimum value of the total nominal voltage of the battery bank (Szbb) such that the series-parallel battery connection operates on 48 V. If Vop is the DC operation voltage of the BTS equipment, Vbat is the voltage rating or nominal voltage of a single battery (V), Ns-bat and Np-bat are the number of batteries in series and parallel connections respectively, Nbb is the total number of batteries in the battery bank, and Ebat is the nominal energy storage capacity (kWh) of a single battery selected, the total storage capacity of the battery bank is defined by the following equations.
Szbb N s bat N p bat Ebat .
(3.42) 125
Szbb
Vop Vbat
N p bat Ebat .
(3.43)
Szbb N bb Ebat .
(3.44)
The energy management system described in section 3.3 determines the design value of Szbb, and Nbb is selected such that Np-bat is an integer. The battery bank life (yr) is determined by the equation: N bb Elt ,tp Lbb Lbf , E ann,tp
, min
(3.45)
where Lbb is the battery bank life (yr), Nbb is the number of batteries in the battery bank, Elt,tp is the lifetime throughput of a single battery (kWh), Eann,tp is the annual battery throughput (kWh yr-1), and Lbf is the battery float life (yr). The annual battery throughput is the total energy supplied by the battery bank throughout one year. In other words, it is energy that flows out of the battery bank during periods when energy demanded is more than the total energy generated by the hybrid system. The annual battery throughput is deduced as follows: N
E ann,tp Ebb ( ) 1
(3.46) Ed Ehg
Hybrid (Super-Capacitor/Battery) Energy Storage Model The size or total energy storage capacity of the hybrid energy storage (SC/battery) unit, Szsb, is the summation of the total energy storage capacity of both the super-capacitor and battery banks expressed as follows:
126
Sz sb Sz sc Szbb ,
(3.47)
where Szsb is expressed in (kWh). The energy that flows into (out of) the SC/battery bank is composed of the surplus (deficit) energy produced by the HES. In other words, the input power of the SC/battery bank Psb (kW) can be positive (or negative) depending on the charge (or discharge) mode of operation. Power will flow into the SC/battery bank when the power generated exceeds the desired load demand (Pnet > 0) and will flow out of the SC/battery bank when the power generated is less than the desired load demand (Pnet < 0). At any time η, the state of charge of the SC/battery bank, SOCsb (η) is related to the previous state of charge SOCsb (η– 1), and to the energy production and consumption situation of the system during the time from η – 1 to η. During energy exchange (charge/discharge) process, when net power Pnet (kW) flows into (or out of) the storage unit (i.e., when Pnet ≠ 0), the available SC/battery state of charge at time η is expressed as:
SOCsb ( ) SOCsb ( 1)
( sb Phg ( ) Pd ( )).
SOCsb ( ) SOCsb ( 1)
60 Sz sb
(3.48)
Pnet ( ) . 60 Sz sb ,
SOCsb ( 1)
,
(3.49)
E net ( ) Sz sb
where Δη is the simulation step time (previously defined), ηsb is the SC/battery charge/discharge efficiency, Pd is the total power demand by the GSM BTS site (kW), Phg is the total power generated by the HES (kW), and Szsb is the decision variable or total energy 127
storage capacity (kWh) of the SC/battery bank. The net energy difference at any time (Enet (τ), expressed in kilowatt-hour) determines the charge/discharge provided the constraints imposed by the proposed energy management system are satisfied. The depth of discharge (DOD) of the SC/battery bank at time η is expressed as:
DODsb ( ) 1 SOCsb ( ) .
(3.50)
To prolong the battery lifetime (i.e., to prevent over-charging or over-discharging of the batteries) the SC/battery SOC at time η is subject to the constraint: SOCsb,min SOCsb ( ) SOCsb,max ,
(3.51)
where SOCsb,min and SOCsb,max are the minimum and maximum permissible SOC of the SC/battery respectively. The minimum SOC of the SC/battery is defined by the equation: SOCsb,min 1 DODsb,max ,
(3.52)
where, DODsb,max is the maximum permissible DOD of the SC/battery. The total energy capacity (kWh) of the SC/battery bank at time η is given by the equation:
rd E sb ( ) 1 . Sz sb . SOCsb ( ) , 24 60
(3.53)
where rd is the self-discharge rate of the battery. Manufacturers‘ documentation give a selfdischarge rate of 25% over six months for a storage temperature of 20 0C (Diaf et al., 2007), and this is equivalent to the self-discharge rate of (25 /182 =) 0.14 % day-1. The technical
128
details of the selected SC and battery are shown in Tables A4 and A5 (appendix A) respectively. 3.3
Energy Management Strategy
The focus of the control strategy is for effective management of the energy flow for improved stability and power supply reliability of the proposed system. It ensures the generation of the proper power level between the three sources of power and the distribution for reliably satisfying the energy demand at any time η. This study assumes that the Peak Power Trackers (PPTs) and the proposed control algorithm are ideal (lossless). In addition, the conversion (inverter and rectifier) efficiency and the stabilizer efficiency are constant. The simulation step time ∆η, is equal to 1 min and total output power of the HES Phs is assumed constant during ∆η. 3.3.1 Control Design The proposed HES model is shown in Figure 3.2. The design is such that only the green energy sources contribute in the charging process of the energy storage unit at any time η provided the technical constraints imposed on the system are satisfied. The Peak Power Tracker (PPT) determines the annual fixed tilt angle that can keep the PV array operating at the optimum power points. The supervisory controller coordinates all power electronics and regulates the system for reliable operation. Control signals are defined by the controller based on the constraints specified by proposed operation strategy discussed in section 3.3.2. The net power (difference between generation and supply) is controlled through the energy storage system using converters. The excess energy generated by the system, if any, is drawn by the dump load. The aim is to ensure proper functioning of the power conversion unit.
129
Utility grid
Wind resource
Solar resource
GESS
WECS
PVCS
GSM BTS site Auxiliary load
DC/DC
AC/DC
PPT controller BTS Ppvcs Dump load
Pwecs
Pgs DC bus
Prg Pb
Pnet Phg
Charge controller
Psc
Battery bank SC bank
Pout Pd
Pd (dc)
Supervisory controller DC/AC
Pd (ac)
Figure 3.2: Control design of proposed hybrid energy system The proposed HES operates while the SC/battery bank is either in the passive (idle) or active (charging/discharging) modes of operation provided the constraints imposed by the energy management system (see section 3.3.2) are satisfied. The SC/battery bank is said to be in the passive mode when Enet (τ) = 0, i.e., when the total power generation at time η is equal to the total energy drawn by the load at η. Under this condition, SOCsb (η) = SOC sb (η-1). In the active mode, when SOCsb (η) > SOCsb,max, the supervisory controller disconnects the SC/battery from charging while switching on the dump load and dump excess energy based on imposed technical constraint. Conversely, when SOCsb (η) = SOCsb,
min,
the supervisory
controller disconnects the supplied load from the SC/battery bank. The switching operation of 130
the supervisory controller can be implemented via a micro-controller that is programmed based on the operation strategy described in section 3.3.2. The developed HES model of Figure 3.2 can operate either in the stand-alone or gridconnected mode. If selected to operate in stand-alone mode of operation, the decision variable (Dmo), which determines the mode of operation is set to the default value of zero. Under this condition, Ege determined by Eq. (3.9) becomes zero (which is equivalent to disconnecting the system from the grid). Based on the design, parameters for stabilizer and rectifier (Pr,sta and Pr,ret) defined by Eq. (3.10) and Eq. (3.38) respectively, become zero, and Figure 3.2 reduces to Figure 3.3. The stand-alone or off-grid design of Figure 3.3 is also referred to as the hybrid green (renewable) energy system since only green energy sources are used for electricity generation. Similarly, the hybrid energy generated by proposed system (shown in Eq. (3.34)) reduces to Eq. (3.54). E hg ( ) E wg ( ) E sg ( ) E rg ( )
,
(3.54)
where Ewg, Esg, and Erg, are the wind, solar, and renewable (green) energy production respectively, all energy values are expressed in kilo-watt-hour (kWh). The proposed HES design (Figure 3.3) can be referred to as a microgrid, since the HES can intelligently manage interconnected loads and energy resources and is capable of operating in independently, from the utility grid.
131
Wind resource
Solar resource
WECS
PVCS
DC/DC
PPT controller
GSM BTS site Auxiliary load
Ppvcs
BTS
Pwecs Prg Dump load Pb
Pnet Charge controller
Battery bank
Psc
SC bank
Pout Pd
Pd (dc)
Supervisory controller
DC/AC
Pd (ac)
Figure 3.3: Control design of proposed microgrid (stand-alone HES) 3.3.2 Operation Strategy The constraints imposed on the system operation are as follows: 1. If grid energy is available, i.e., if [Ege(τ) > 0] then supply the energy demand [Eout(τ) = Ed(τ), where Eout is the output energy
supplied]
and
check
the
state
of
charge
of
the
SC/battery bank. Afterward, if [SOCsb(τ) > SOCsb,max] then stop charging
the
SC/battery
bank,
calculate
the
energy
balance
[using Ebal(τ) = Erg(τ) – (Szsb×(SOCsb,max-SOCsb(τ-1)))/ηsb] and set SOCsb (τ) = SOCsb,max; else, set Ebal to zero. Afterwards, if [Ebal(τ)
≤
El(τ)]
then
compute
the
actual
energy
drawn
(purchased) from the grid [using Egrd,net(τ) = El(τ) - Ebal(τ),
132
where El is the required load capacity that can response to sudden rise in energy demand] and set Ebal to zero; else, do not purchase
energy
from
the
grid,
re-compute
energy
balance
[using Ebal(τ) = Ebal(τ) –El(τ)]; check, if [Ebal(τ) ≥ Ewg(τ)] then turn off the WTG at time τ and compute the excess energy generation of the system [using Eexc(τ) = Ebal(τ) – Ewg(τ)]; else, set the Eexc(τ) = Ebal(τ). Switch on the dump load and dump excess
energy
supply.
All
energy
values
at
time
τ
are
expressed in kWh. 2. Else,
i.e.,
when
grid
energy
if [SOCsb,min < SOCsb(τ) || SOCsb(τ)
is
unavailable
check;
≤ SOCsb,max] then supply the
energy demand [i.e., Eout(τ) = Ed(τ)] with the SC/battery bank working in any of the three modes of operation depending on the magnitude of net energy generation Enet(τ). Set dump and excess energy supply to zero. Elseif [SOCsb,min < SOCsb(τ) || SOCsb(τ) > SOCsb,max] then supply the power demand [i.e., Eout(τ) = Ed(τ)], set SOCsb (τ) = SOCsb,max, disconnect the SC/battery bank from source and turn off the WTG at time τ, with the SC/battery bank working in the discharging mode of operation. Calculate the energy balance [using Ebal(τ) = Erg(τ) – El(τ) (Szsb×(SOCsb,max - SOCsb(τ-1)))/ηsb] and compute the excess energy supply [using Eexc(τ) = Ebal(τ) – Ewg(τ)]. Switch on the dump load
and
dump
supply
load
and
energy
supplied
excess set
energy SOCsb(τ)
[using
Eout(τ)
renewable energy generation].
133
supply. =
Else,
SOCsb(τ-1), =
Erg(τ),
disconnect and
where
compute Erg
is
the the the
3.3.3 System Energy Characteristics The electrical characteristic of proposed hybrid energy system was determined in terms of the energy contributions/efficiencies of different energy sources that make up the hybrid system. The contributions of various energy sources utilized for estimating the power supply reliability of the proposed hybrid energy system are discussed as follows: Energy Contribution If at any time η the output energy supplied by the proposed hybrid system to the load is Eout (η), the DC energy produced (generated) by the WECS and PVCS are Ewg (η) and Esg (η) respectively, and the DC energy drawn from the utility grid is Egrd,net (η), the total DC energy produced by the hybrid energy system is given by the following equation:
Ehs ( ) Egrd,net ( ) Ewg ( ) Esg ( ) .
(3.55)
The wind energy contribution is defined by Eq. (3.56). N
E wec
N
E wg ( ) . Eout ( )
1
1
N
E
1
.
(3.56)
( )
hs
Solar energy contribution is defined by the given equation: N
Esec
N
E sg ( ) . Eout ( )
1
1
N
E
1
hs
.
(3.57)
( )
134
Renewable energy contribution is defined as the sum of wind and solar contributions, which is stated as follows:
Erec Ewec Esec
(3.58)
The grid energy contribution and the total energy supplied are defined by Eq. (3.59) and Eq. (3.60) respectively. N
E gec
N
E grd, net ( ) . Eout ( )
1
1
N
E
1
Etec
N
E 1
out
hs
.
( )
( ) .
(3.59)
(3.60)
All energy values produced at time η are expressed in kilo-watt-hour (kWh). Energy Efficiency of Various Units The efficiency of the individual unit is defined in terms of the ratio of energy contribution to the total HES. The efficiency of various energy technologies is given as follows: 1. Wind energy efficiency is deduced from the equation:
we
E wec . Etec
(3.61)
2. Solar energy efficiency is estimated using the equation:
se
Esec . Etec
(3.62)
135
3. Renewable energy efficiency is computed using Eq. (3.63).
re
E rec . Etec
(3.63)
4. Grid electricity efficiency is expressed as follows:
ge
E gec Etec
,
(3.64)
where, Etec is the total annual energy contribution of the HES, and Ewec, Esec, Erec, and Egec are annual energy contributions from wind energy conversion system, PV conversion system, renewable sources, and utility grid respectively. All energy contributions are expressed in kilo-watt-hour per year (kWh yr-1). 3.3.4 System Reliability Considerations Reliability as defined here is the percent of electricity demand that the power system can deliver. The technical model for the system reliability developed is based on the loss of power supply probability (LPSP) technique proposed in Bin et al. (2003), which is stated as: N
LPSP
E
1 N
E
1
( )
def
d
( ) ,
N
1
E
1
out
N
E
1
d
(3.65)
( )
( )
where Eout (η), Ed (η) and Edef (η) are the output, demand and deficit energy supplied at time η, ∆η is the simulation step equal to one min, and N is the total simulation time (considered as N
136
= 525,600 min, which is equivalent to one year). All energy values are expressed in kilo-watthour. The reliability (efficiency) of the proposed system is given by the equation: N
rel
E
1 N
out
E 1
d
( )
( )
.
(3.66)
1 LPSP
The annual unmet or deficit energy and the percent of energy not supplied by the HES are given by Eq. (3.67) and Eq. (3.68) respectively. N
Eun LPSP Ed ( )
(3.67)
E 100 * LPSP
(3.68)
1
un
The annual excess energy and the percent of excess energy supplied by the HES are given by Eq. (3.69) and Eq. (3.70) respectively.
Eexc
E exc
3.4
N
Eexc ( ) .
(3.69)
Eexc . Etec
(3.70)
1
System Techno-Economic Analysis
This section outlines the criteria utilized in the economic and technical cost analysis for optimum sizing of the proposed HES. The economic cost includes all necessary investments 137
that are expected to be spent during the period in which the HES is considered at work. It consists of initial capital, replacement cost, tax and operation and maintenance cost of all components. The technical cost relates to the reliability costs implications of the system. The cost includes the cost penalty for supply shortages (unmet load) and cost of development of energy management system or simulation program. 3.4.1 Basic Considerations Based on Eq. (2.67) the annualized cost of investment of a hybrid system in terms of the present worth (PW) is given as:
COI sys, ann
1 LN
COI j
j 1
j , ini
COI j , rep( pw) COI j , om( pw) COI j , sv( pw)
(3.71)
where, LN (yr) is the life span of the project, COIini is the total initial cost of investment, COIrep (pw) is the total PW of replacement cost, COIom (pw) is the total PW of annual operation and maintenance (OM) cost, COIsv
(pw)
is the PW of all salvage value, and j is the total
number of hybrid system component units. If the lifetime Lj (yr) of component j is shorter than that of the project (i.e., Lj ˂ LN); it might be necessary to purchase additional component j before the end of the project life span. Given that cj (N unit-1) and com,j (N unit-1 yr-1) are the present total cost and the annual operation and maintenance cost per unit sizing (design) parameter of component j respectively, Xj (=LN /Lj, rounded to the greater integer) is the number of times component j is needed, and that salvage value of component j is assumed to decrease linearly from cj to svj when component j operates along its lifetime Lj, the components of Eq. (3.71) is calculated as follows (Sweelem and Nafeh, 2006):
138
Xj
x
1 . L j
COI j , ini COI j , rep ( pw) c j Sz j . f ( e,i ) x 1
,
(3.72)
c j Sz j . q xe j LN
COI j , om ( pw) com, j . Sz j . f ( ey,i ) y 1
,
(3.73)
com, j . Sz j . q om Xj1 x.L LN COI j , sv ( pw) Sz j sv jP . f ( j ,i ) sv j . f ( j ,i ) j x 1 , Sz j ( sv jP . q LN sv j . q xf j )
(3.74)
where Szj (unit) is the sizing (design) variable, f(e,i) = (1+re)/(1+ri), f(f,i) = (1+rf)/(1+ri), and ri, rf and re are the annual interest, inflation and escalation rates (considered as 8%, 10% and 7%) respectively. The assumed inflation rate is the average value deduced from the estimates collected from the archives of Central Bank of Nigeria (CBN) from 2011 – 2013 (CBN, 2014). The annual interest rate is determined based on the 2014 interest rates of banks while the escalation rate is computed based on random sampling of prices. Inflation involves average change as it is concerned with a group of goods and services. In contrast, escalation deals with the persistent rise in the price of specific goods and services. Therefore, escalation rates, which affects the actual costs and revenues that will be realized for a project, must be accounted for in economic evaluation of investment. Escalation rates are normally based on index numbers. An index number measures the relative change in price, quantity, value, or some other item of interest from one time period to another.
The coefficient for salvage value of component j at the end of the project (i.e., at any time before the end of component j life span) is defined by Nafeh (2011) as follows:
139
c j sv j sv j p c j L j
.ny ,
(3.75)
where ny indicates the number of years of operation between the installation of the last component j and the end of the project life span. As observed, if the lifetime of the component is equal to or greater than the project lifetime (i.e., Lj ≥ LN), Xj = 1, and no additional purchase of component j is required, that is, Eq. (3.72) and Eq. (3.74) reduces to Eq. (3.76) and Eq. (3.77) respectively.
COI j , ini COI j rep ( pw) c j Sz j .
(3.76)
COI j , sv( PW ) Sz j sv j . q LN .
(3.77)
The PW calculation of each component of the proposed HES is treated in a similar manner throughout the economic and technical analysis. 3.4.2 Economic Analysis The economic Cost of Investment (COI) on the proposed HES is defined by the equation:
COI sys, e
COI j
j 1
(3.78)
j
where, j is the number of component units of the proposed power system, COIj is the total investment of each component unit j. The proposed system component units of Figure 3.1 are the grid energy supply system, wind energy conversion system, photovoltaic conversion system, energy storage system, and power conversion units respectively.
140
Cost of the Grid Energy Supply Unit Based on Eq. (3.71), the total cost of investment on the grid energy supply unit is given by the equation:
COI gs COI gs, ini COI gs, rep ( pw) COI gs, om ( pw) COI gs, sv ( pw)
(3.79)
where COIgs,ini is the initial cost of grid energy supply unit, COIgs,rep (pw) is the total PW of replacement cost of grid supply unit, COIgs,om (pw) is the total PW of annual OM cost of the grid supply unit, and COIgs,sv (pw) is the PW of all salvage value of the grid supply unit. All costs are expressed in Naira (N). The initial cost of the grid supply unit refers to the interconnection charge and the capital cost of voltage stabilizer, for grid voltage stabilization and the rectifier. The interconnection charge is a one-time (fixed) fee charged for allowing the micro-grid to be connected to the grid. The replacement cost of the grid is always zero. However, due to the installation of the voltage stabilizer (with lifetime, Lsta < LN) and rectifier (with lifetime, Lret < LN) in the grid supply unit, the total PW of the initial and replacement cost of investments in grid supply unit is stated as follows: X sta
x
1 . Lsta
COI gs, ini COI gs, rep ( pw) COI grd, int c sta Pr , sta . f ( e,i ) x 1
X ret
x
1 . Lret
c ret Pr ,ret . f ( e,i ) x 1
, (3.80)
COI grd, int c sta Pr , sta . q sta c ret Pr ,ret . q ret
where COIgrd,int is microgrid interconnection charge (N), csta is the capital cost of the stabilizer per kW (N kW-1), Pr,sta is the rated power of the stabilizer (kW), Pr,ret is the rated power of the rectifier (kW), Lsta and Lret (usually shorter than LN) are the lifetime of the stabilizer and rectifier (yr), Xsta (= LN /Lsta) and Xret (= LN / Lret) rounded to the greater integer,
141
are the number of times the stabilizer and rectifier should be purchased respectively during the project life span. The grid OM cost is equal to the annual cost (energy cost, fixed charge/metre maintenance cost and tax) of energy drawn from the grid and the annual operation and maintenance cost of the stabilizer and rectifier. The PW of all annual operation and maintenance cost is given by the equation: LN
LN
y 1
y 1
COI gs, om ( pw) com, sta . Pr , sta . f ( ey,i ) com, ret . Pr ,ret . f ( ey,i ) 1.05 LN (COI fc c ge E grd,net )
,
com, sta . Pr , sta . q om com, ret . Pr ,ret . q om 1.05 LN (COI fc c ge E grd,net )
(3.81) where, com,sta and com,ret are the annual operation and maintenance cost of stabilizer and rectifier respectively (in N kW-1 yr-1), COIfc is the annual fixed charge or cost of metre maintenance (N yr-1), cge is the grid electricity price (N kWh-1), and Egrd,net is the annual energy drawn from the grid (kWh yr-1). The factor, 1.05, accounts for the 5% charged as Value Added Tax (VAT) on the cost of energy drawn from the grid in Nigeria. The salvage value is expressed asfollows: X sta 1 COI gs, sal ( pw) Pr , sta svstaP . f ( Lj ,Ni ) svsta . f ( xj ,.iL)sta x 1 Pr , sta . svstaP . q LN Pr ,ret svretP . q LN
X ret 1 Pr ,ret svret . f ( Lj ,Ni ) svret . f ( xj ,.iL)ret P x 1
, (3.82)
where svsta and svret (considered here as zero) are the salvage value per unit size of the stabilizer and rectifier (in N kW-1) respectively at the end of their useful lifespan.
142
Simplifying Eq. (3.79), the total cost of investment on the grid energy supply unit is given the following equations: COI gs COI grd, int 1.05 LN (COI fc c ge E grd,net ) (c sta . q sta com, sta . qom svstaP . q LN ) Pr , sta
(3.83)
(c ret . q ret com, ret . qom svretP . q LN ) Pr ,ret . Dmo
COI gs cc gs Dmo
(3.84)
where, Dmo is the decision variable that specifies the mode of HES operation and ccgs, the cost coefficient of the grid supply unit is: cc gs COI grd, int 1.05 LN (COI fc c ge E grd,net ) (c sta . q sta com, sta . qom svstaP . q LN ) Pr , sta
(3.85)
(c ret . q ret com, ret . qom svretP . q LN ) Pr ,ret
Cost of the Wind Energy Conversion System Based on Eq. (3.71), the total cost of investment on the WECS unit can be expressed as:
COI wecs COI wecs, ini COI wecs, rep ( pw) COI wecs, om( pw) COI wecs, sv ( pw)
(3.86)
where COIwecs,ini is the initial cost of WECS unit, COIwecs,rep (pw) and COIwecs,om (pw), are the total PW of replacement cost of investments and annual OM cost of WECS unit respectively, and COIwecs,sv (pw) is the PW of all salvage value of the WECS unit. The decision variable of the WECS is the number of WTGs selected, which is considered as one. If the lifetime of the wind turbine Lwt is less than LN, the total cost of the WECS is estimated by the equation:
143
COI wecs c wt j . q wt com, wt j . qom ( svwtPj . q LN svwt j . q wtxf ) cc wt j
,
(3.87)
where, cwtj (N unit-1) is the current cost per unit of WTG j, com,wtj (N unit-1 yr-1) is the annual operation and maintenance cost per unit WTG j, svwtpj (N unit-1) and svwtj (N unit-1) are the salvage value per unit of WTG j at the end of project and at the end of the WTG j lifetime respectively, and ccwtj is the cost coefficient of the WECS with WTG j unit. Cost of the Photovoltaic Conversion System If the project life span (LN) is assumed the same as that of the PV array (Lpv), then the replacement cost of the PV array is negligible. Therefore, the total cost of investment on the PV system is defined by the equation:
COI pvcs ( c pv com, pv . qom sv pv . q LN ). Sz pv cc pv . Sz pv
,
(3.88)
where cpv is the initial cost per square meter of the PV array (N m-2), com,pv is the annual operation and maintenance cost per square meter of the PV array (N m-2 yr-1), svpv is the salvage value per square meter of PV array (N m-2), and ccpv, the cost coefficient of the PV array defined as follows:
cc pv c pv com, pv . qom sv pv . q LN .
(3.89)
Cost of the Power Inverter The design variable in the case of conversion unit is the rated power or size of the inverter (Pr,inv in kW). Since the lifetime of the power electronics is usually lesser than the project
144
lifespan, as in the case of the stabilizer, the total cost of investment on the inversion unit is given by Eq. (3.90).
COI inv ( cinv qinv com, inv . qom svinvp . q LN ). Pr ,inv cc inv . Pr ,inv
(3.90)
where, cinv (N kW-1) is the capital cost of the power inverter, com,inv (N kW-1 yr-1) is the annual operation and maintenance cost of the inverter, svinvp (N kW-1) is the corresponding salvage values at the end of the project. The cost coefficient ccinv per unit size of the power inverter defined as follows:
ccinv cinv qinv com, inv . qom svinvp . q LN
(3.91)
Cost of the Energy Storage Unit The decision variable in the case of the energy storage unit is the capacitance or number of the SC modules Nsc and the energy storage capacity or number of batteries in the battery bank, Nbb. In renewable energy applications, the lifetime of the storage unit is usually shorter than that of the project. Assuming that the salvage values of super-capacitors and batteries at the end of their lifetime are negligible, the total cost of the energy storage unit can be determined by the following equation:
COI es (c sc . q sc com, sc . q om svscP . q LN ) . N sc (cbb . qbb com, bb . q om svbbP . q LN ). N bb
,
(3.92)
cc sc . N sc cc bb . N bb where, csc (N unit-1) is the capital cost per SC module, cbb (N unit-1) is the capital cost per battery, com,sc (N yr-1 module-1) and com,bb (N yr-1 unit-1) are the annual operation and 145
maintenance cost of the SC module
and battery respectively, svscp (N unit-1) and svbbp
(N unit-1) are salvage value of the SC and battery banks at the end of the project life span, and ccsc and ccbb are the cost coefficient per unit size of the SC bank and battery bank defined respectively as follows:
cc sc csc q sc com, sc . qom svscp . q LN
(3.93)
ccbb cbb qbb com, bb . qom svbbp . q LN
(3.94)
The total economic cost of investment on the HES is simplified by the equation:
COI sye, ec
COI j
j 1
j
.
(3.95)
cc gs Dmo cc wt . Sz wt cc pv . Sz pv cbb N bb cc sc N sc cc inv Pr ,inv
3.4.3 Technical Analysis The technical cost of proposed HES defined in terms of the cost penalty on deficit energy supply is given by the equation: COI sys, te c ge . LN
N
E
1
def
( ) C sys, mgt
c ge . LN E def ,ann C sys, mgt
(3.96)
c ge . LN . LPSP. E d ,ann C sys, mgt
where, LPSP is the loss of power supply probability, c ge is the cost penalty per kilo-watthour of energy (N kWh-1), Edef,ann and Ed,ann are the total annual deficit energy supplied and demand energy (kWh yr-1) respectively. Csys,mgt is the cost of developing the energy management system or simulation program. Equation (3.96) can be simplified as follows:
146
COI sys, te cc def LPSP Csys, mgt ,
(3.97)
where ccdef, the cost coefficient on deficit energy supplied is stated as follows:
ccdef c ge . LN .Ed ,ann .
(3.98)
3.4.4 Total Cost Analysis The total cost of investment on the proposed HES is the sum of the economic and technical cost defined by the equation:
COI sys, tot C sys, ec C sys, te cc gs Dmo cc wt . Sz wt cc pv . Sz pv cc inv . Pr ,inv . cc sc . N sc cbb . N bb cc def LPSP C sys, mgt
(3.99)
The cost of energy per kWh [COEkWh (N kWh-1)] is expressed as follows:
COEkWh
COI sys, tot N
LN . Eout ( )
COI sys, tot
(3.100)
LN . E sys,ann
1
COEkWh
cc gs Dmo cc wt . Sz wt cc pv . Sz pv cc inv Pr ,inv cc sc N sc cbb N bb cc def LPSP C sys, mgt E sys,tot
(3.101) where Esys,ann and Esys,tot (= LN Esys,ann) are the annual and total energy supplied by the HES to the GSM BTS site, in kWh yr-1 and kWh respectively. Rearranging Eq. (3.101), the costs of energy per kWh is simplified by the equation: COEkWh cc gs Dmo cc wt . Sz wt cc pv . Sz pv
cc inv Pr ,inv cc sc N sc cbb N bb cc def LPSP C sys, mgt
147
,
(3.102)
,
where Dmo is the decision variable that specifies the mode of HES operation, and ψ (kWh-1) is the reciprocal of the total energy supplied by the HES during the project life span, which is equal 1/Esys,tot. 3.4.5 Techno-Economic Viability The key parameters, which have a significant impact on the energy throughput of the hybrid system, are the power supply reliability and Cost of Energy (COE). The various sizing methods discussed in the literature try to establish a compromise between these factors (techno-economic) based on defined constraints, usually specified by the designer. Therefore, this study specifies the performance index of the energy system based on the core parameters (reliability and COE). The techno-economic viability of the proposed power system is defined by the equation:
Ke
rel COEkWh
(3.103)
where, ηrel is the power supply reliability of the hybrid energy system, which is defined in terms of the LPSP (as ηrel = 1-LPSP). This study proposed an Energy index (Ke), measured in kWh per Naira as a Key Performance Index (KPI) for assessing the techno-economic viability of a hybrid energy system. The proposed KPI is an expression of the energy throughput of the system. It specifies the amount of energy that the system can produce per Naira investment. The higher the KPI, the better is a system. 3.5
Environmental Impact Assessment
This section deals with the environmental impact assessment of the proposed HES on the environment. The environmental analysis relates to the environmental implication resulting from pollutant emissions into the environment. The green energy sources, such as wind and 148
sun, are environmentally friendly with negligible GHG emissions, as such the emissions from the proposed HES into the environment is the summation of the CO2, CH4, and N2O emissions produced from the grid electricity supply. The total carbon footprint (GHG emissions) from proposed HES is given by the equation (Nandi and Ghost, 2010):
emsys, ghg emsys, CO2 emsys, CH 4 emsys, N2 O
(3.104)
where, emsys,CO2 emsys,CH4 and emsys,N2O are the system total emission resulting from the release of CO2, CH4 and N2O respectively into the environment as given in Table 2.6. The system total emissions from CO2 combustion (tCO2) is defined by NREL (2010) as follows: em sys, CO2
e f , CO2 . LN . E grd,net
(3.105)
10 3
where, ef,CO2 is the emission factor of CO2 (tCO2 MWh-1), Egrd,net is the annual energy drawn from the grid (kWh yr-1), and LN is the project life span (yr). Similarly, the system total emissions from CH4 combustion (tCH4) and N2O combustion (tN2O) respectively, are determined by the following equations. em sys, CH4
e f , CH4 . LN . E grd,net
em sys, N 2 O
e f , N 2 O . LN . E grd,net
10 3
10 3
,
(3.106)
,
(3.107)
where, ef,CH4 and ef,N2O are the emission factor of CH4 (tCH4 MWh-1) and N2O (tN2O MWh-1) respectively.
149
The system total GHG emissions, which is the summation of individual emissions expressed in terms of CO2-eq, is defined by IPCC (1996) and Nandi and Ghost (2010) as: em sys, ghg
LN E grd,net (e f , CO2 . GWPCO2 e f , CH4 . GWPCH4 e f , N 2 O . GWPN 2 O ) 10 3
,
(3.108)
where, GWPi is the global warming potential of gas i given in literature (see Table 2.3). The environmental cost implication (total cost penalty for the emission of GHG) is defined by Eq. (3.109). COI sye, en cco2 emsys, ghg ,
(3.109)
where, cco2 (N tCO2-eq-1) is the cost penalty per tCO2-eq released into the environment. 3.6
Estimation of Energy Consumption of GSM BTS Site
The electrical power drawn by the GSM BTS site can be synthesized from the specified typical average load profile. If PD is the typical average power demand of the GSM BTS site, then, under real conditions, there are variations from these average value across the minutes of the day and days of the year due to the operational conditions (such as the varying call rates) and ageing of components. In order to account for these variations, it is assumed that PD varies with a minutely random (noise) factor m , and a daily random factor d . The minute load profile of the GSM BTS site is deduced as:
Pd ( ) ( Pdc ( ) Pac ( ) / inv ) (1 m d ) PD ( ) (1 m d )
,
(3.110)
where, Pdc (kW) is the DC power drawn by the BTS equipment, Pac (kW) is the AC power drawn by the auxiliary equipment, PD (kW) is the typical power drawn by the BTS site, ηinv is
150
the inversion efficiency, and δm and δd are the minutely and the daily noise input values, respectively, derived from normal distribution with an average of zero and standard deviation equal to the noise factor. A minute noise factor of 0.04 and a daily noise factor of 0.06 are assumed in this study. The energy (kWh) drawn by the load (GSM BTS site) at any time η is given by the equation:
E d ( )
Pd ( ) . 60 ,
(3.111)
where, ∆η is the simulation step equal to 1 min. Real time electric load tends to jump around randomly. As a result, an operating reserve is needed to prevent the power system from going down in case of a sudden rise in energy demand (Nandi and Ghost, 2010). The required load capacity to enable any sudden rise in energy demand is stated in terms of the minutely energy demand as follows:
El ( ) Ed ( ) Eor ( ) .
(3.112)
where, Eor (kWh) is the energy reserve, which is assumed to be 10% of the minutely energy demand. 3.7
Data Collection and Analysis
The data collected were analyzed, evaluated and discussed in the following sub-sections. 3.7.1 Load Data The power consumption data of components/equipments of a typical out-door GSM base station (S/4/4/4) collected from Etisalat Nigeria are shown in Table B1 (appendix B). These data were used to estimate the typical minute load profile for a typical year. The load profile 151
is adjusted using the stochastic method defined by Eq. (3.110) to account for the real-time variations. 3.7.2 Wind Turbine Characteristic Curve The power profiles (output characteristic curves) of WTGs used were supplied by Anhui Hummer Dynamo Company (AHDC) Ltd., Anhui Province, China; manufacturer of Hummer WTGs. AHDC provided power profiles, as shown in Table B2 (appendix B), for four different sizes of WTGs (H3.1-1 kW, H3.8-2 kW, H4.6-3 kW, and H6.4-5 kW). 3.7.3 Meteorological Data The monthly average daily meteorological data used were collected from two data sources for 37 sites in Nigeria. The sites selected consist of the 36 capital cities and the Federal Capital Territory, Abuja. The first part was downloaded from the archives of the National Aeronautics and Space Administration (NASA), accessed from the internet (http://eosweb.larc.nasa.gov/; accessed November 20, 2012). These data consist of the monthly average daily global and diffuse solar radiation on a horizontal surface and the corresponding cloud cover, relative humidity, and maximum and minimum air temperatures for a period of 22-years (1984–2005). NASA derives these data sets from a variety of earth-observing satellites and reanalysis research programs, which provide reliable meteorological resource data over regions where surface measurements are scarce or nonexistent. In addition, the satellite-derived data provide two unique features: the data are global and contiguous in time (NASA, 2012). For simplicity, the 22-years monthly average daily values of data obtained from NASA are presented in Tables B3–B8 (appendix B).
152
The second part of data was collected from the archives of the Nigerian Meteorological (NIMET) agency, Oshodi, Lagos State, for same locations. It consists of the ground-based monthly average daily measurements of sunshine duration and wind speed (measured at a height of 10 m above sea level) for a period of 22-years (1991–2012). These data were checked for inconsistencies, the purpose was to eliminate incomplete measures. It is worthy of note that some of the data sets obtained from NIMET consist of missing (unrecorded or unmeasured) values. The technical specifications of the Nigerian Meteorological stations are shown in Table B9 (appendix B) while the retained monthly average daily ground-based data sets obtained from NIMET are shown in Tables B10 and B11 (appendix B). The retained ground-based data sets and those collected from NASA were integrated to form the final data sets used in this study. The final data sets for the 37 cities considered were classified into two groups. Firstly, the site is randomly chosen from each of the six geopolitical zones. The data sets from the remaining (31) sites were combined to form regional data sets (i.e., northern and southern Nigeria) as shown in Figures B1 – B4 (appendix B). The combined (group-1) data sets were used for calibration of solar radiation model parameters as well as to develop generalized correlations, which can be used for estimating the daily global and diffuse solar radiation on a horizontal surface for each of the six geopolitical zones in Nigeria. Thereafter, the individual data sets randomly selected from each geopolitical zone (referred to as group-2 data sets) were used for performance evaluation and validation of various solar radiation models for the six geopolitical zones in Nigeria. Table 3.1 gives a summary of the geographical classification of Nigeria, indicating the coordinates of the 31 sites (cities) used for model development or calibration and the six randomly selected sites (highlighted in bold fonts) used for evaluating the developed and existing radiation models discussed in section 3.8. 153
Table 3.1: Geographical classification and coordinates of selected sites in Nigeria. Site
Latitude (oN)
Longitude (oE)
Altitude (m)
1.
Birnin Kebbi
12.46
4.20
271
2.
Dutse
11.70
9.34
480
Gusau
12.16
6.67
415
Kaduna
10.50
7.35
575
5.
Kano
12.03
8.50
457
6.
Katsina
12.99
7.61
492
7.
Sokoto
13.06
5.25
331
8.
Bauchi
10.31
9.85
559
9.
Damaturu
11.75
11.96
397
Gombe
10.28
11.17
416
Jalingo
8.90
11.37
476
12.
Maiduguri
12.00
13.33
313
13.
Yola
9.20
12.50
430
14.
Abuja
9.08
7.53
484
15.
Ilorin
8.50
4.58
274
Jos
9.93
8.88
587
Lafia
8.49
8.52
374
18.
Lokoja
7.81
6.74
216
19.
Makurdi
7.73
8.54
249
20.
Minna
9.61
6.55
319
21.
Abeokuta
7.16
3.35
156
22.
Ado-Ekiti
7.63
5.22
233
Akure
7.25
5.20
233
Ibadan
7.37
3.97
183
25.
Ikeja
6.58
3.33
73
26.
Oshogbo
7.77
4.56
223
27.
Abakaliki
6.33
8.10
277
Awka
6.21
7.07
183
S/N
Geopolitical zone
3. 4.
10. 11.
North-western (NW)
North-eastern (NE)
16. 17.
23. 24.
North-central (NC)
South-western (SW)
28. 29.
South-eastern (SE)
Enugu
6.50
7.50
183
30.
Owerri
5.48
7.03
176
31.
Umuahia
5.53
7.48
176
32.
Asaba
6.2
6.73
159
33.
Benin City
6.34
5.52
135
Calabar
4.95
8.33
246
Port Harcourt
4.67
7.17
117
36.
Uyo
5.05
7.93
176
37.
Yenagoa
4.93
6.26
83
34. 35.
South-south (SS)
Source: Internet (http://eosweb.larc.nasa.gov/, http://www.distancesfrom.com/Latitude-Longitude.aspx, http://www.punchng.com/news/constitution-six-geopolitical-zones-divide-north-south/; accessed on November 20, 2012).
154
3.8
Model Performance Evaluation
The methods adopted for evaluating the performance accuracy of developed models (such as solar radiation and wind turbine generator), which constitute inputs to the proposed hybrid energy system for Nigerian terrain are discussed in this section. The models were calibrated using linear, multi-linear or non-linear regression techniques. On the basis of widely used statistical indices, (coefficient of correlation, coefficient of determination, mean bias error, mean absolute bias error, relative percentage error, root mean square error, and t-statistic test) the applicability of different methods of estimation were determined. The assessment compared the performance of the developed solar radiation models with the widely used models discussed in the literature. The objective is to determine the most suitable model for estimating the monthly average daily global and diffuse solar radiation on a horizontal surface for Nigeria. Furthermore, it assessed the different methods for determining the optimum tilt angles for harvesting solar energy in Nigeria. 3.8.1 Evaluation of various Global Solar Radiation Models for Nigeria A general survey of various global solar radiation models in the literature indicates that most estimation methods are based on solar radiation data reaching the earth‘s surface, which depends on the climatic conditions of the particular location (Besharat et al., 2013). Global solar radiation models based on commonly measured meteorological parameters (sunshine, temperature, cloudiness, and relative humidity) were considered. A model is selected each from the sunshine-based, cloud-based, temperature-based, and three models from the hybrid-parameter-based models. The choice of model selection is on the basis of the model prediction accuracy and the ease of obtaining input parameters. As a result, this study considered only models believed to be universally applicable with readily 155
calibration parameters. The models considered were those proposed by Angstrom-Prescott (1940), Badescu (1999), Chen et al. (2004), El-Metwally (2004), and Falayi et al. (2008). In addition, the predictive accuracy of the deduced multivariable regression model discussed in section 3.2.3 is assessed. The input data or parameters (sunshine duration, air temperature, and cloud cover) of the deduced multivariable regression model were measured at Nigerian stations. These six global solar radiation models studied are shown in Eq. (2.21), Eq. (2.28), Eq. (2.32), Eq. (2.33), Eq. (2.34), and Eq. (3.22), respectively. Method of Calibration Given the data points of the form xi,yi for i = 1, 2, . . . , m set of points (where m corresponds to the 12 months of the year), and that yi depends linearly on xi, then, for accurate prediction of the set points, it is necessary to find the slope bo and y-intercept b1 of a line that best fit these data set points, which can be deduced from the equation:
yi bo b1 xi .
(3.113)
If xi are n independent variables (that is, x1i, x2i, . . ., xni), then, Eq. (3.113) can be expressed as:
yi bo b1 x1i b2 x2i . . . b1 xni ,
(3.114)
where n is the number of independent variables that can minimize the sum of squares of the difference between the observed and the estimated data lines. Based on the selected models the maximum number of independent variables n (such as sunshine ratio, maximum air temperature, relative humidity, etc.) for accurately forecasting the monthly average daily global solar radiation data on a horizontal surface is 4, and Eq. (3.114) reduces to Eq. (3.115). 156
yi bo b1 x1i b2 x2i b3 x3i b4 x4i .
(3.115)
The matrix form of Eq. (3.115) is stated as follows:
y1 1 . . . . . . y i 1
x11
x 21
x31
.
.
.
. . x1i
. . x 2i
. . x 3i
x 41 . . . x 4i
b0 b 1 b2 . b3 b 4
(3.116)
i.e.,
Y Z B,
(3.117)
where Y is i ×1 column matrix, Z is i × 5 matrix and B is 5 × 1 matrix. To solve for Z ,
it is necessary to transform Z to a square matrix (since i ≠ 5) as follows:
Z T Y (Z T Z ) B ,
(3.118)
where Z T is the transpose of Z .
T 1 If (Z Z ) exists then Eq. (3.118) can be expressed as follows:
B (Z T Z ) 1 (Z T Y ).
(3.119)
The solution of Eq. (3.119) is a matrix of empirical constants. For each studied model, Y is the dependent variable equivalent to either KT or KD; x1, . . ., x4 represents other
157
meteorological parameters used for modelling, and B ≡ [bo b1 b2 b3 b4] are the calibrated model parameters. A computer program written in MATLAB (see appendix C) based on Eq. (3.125), was used to compute the empirical constants of the considered solar radiation models. The calibration process utilized the grouped meteorological data sets while the performance validation used the individual average data sets values collected for each of the selected city. The performance evaluation and validation of the considered models for estimating the global solar radiation was determined on the basis of seven statistical indicators (discussed in section 2.6.2). The calibration and validation results are presented in chapter 4. 3.8.2 Evaluation of various Diffuse Solar Radiation Models for Nigeria Among various diffuse solar radiation models, the four established models reviewed in the literature were considered. These considered models proposed by Page (1964), Liu-Jordan (1960), Butt et al. (2010), and Karakoti et al. (2011) are shown in Eq. (2.35), Eq. (2.36), Eq. (2.37) and Eq. (2.38), respectively. In addition to the four models considered, which are believed to be universally applicable, the proposed regression models of Eq. (3.23) and Eq. (3.24) were considered for estimating the monthly average daily diffuse solar radiation on a horizontal surface in Nigeria. The available meteorological parameters informed the choice of the selected models for comparison. Also considered is the ability of these models to generate data from limited average values and the accuracy (quality) of their results. The estimation accuracy of the established diffuse radiation models reported by the original authors and those published in reviews are satisfactory. 158
The method of calibration, performance evaluation and validation are similar to those discussed in section 3.8.1. The calibration and validation results are presented in chapter 4. 3.8.3 Determination of the Optimum Tilt Angle of a PV Array Oriented Due South in Nigeria This section examines the influence of orientation on a south-facing photovoltaic surface for the study areas. The objective is to determine the optimum tilt angles for harvesting solar energy in Nigeria. The average daily solar radiation (global and diffuse) estimates on a horizontal surface for the study sites (see sections 3.8.1 and 3.8.2) are converted to the hourly average using relations proposed by Liu and Jordan (1962) as shown in Eq. (2.39) and Eq. (2.40). The minute values of solar irradiance are determined from the corresponding average hourly values on the assumption that the hourly solar radiation is constant. The total solar irradiance on the tilted PV array was computed from the hourly horizontal values using the Hay-Davies-Klucher-Reindl (HDKR) model (Duffie and Beckman, 2006). In order to achieve maximum global solar radiation on the PV array, three optimization methods (monthly-based, seasonal-based and annual-based) were implemented. The monthly-based optimization method used a fixed monthly average tilt angle, while the seasonal and the annual methods utilized a fixed seasonal average tilt and fixed annual tilt angles respectively. These methods are widely used globally for investigating the effect of orientation and determining of the optimum tilt angles of a photovoltaic surface (Jafarkazemi and Saadabadi, 2013). A computer program (see appendix D), written in MATLAB programming language computes the global solar irradiance on the tilted PV array. The program varied the angle of 159
inclination from 00 to 900 with a step of 10 to determining the optimum tilt angles for harvesting solar energy for the study areas in Nigeria. The results of the analysis of total global solar radiation for the different optimization techniques at optimum tilt angles for selected sites in Nigeria are presented and compared with values recommended in the literature as shown in chapter 4. 3.8.4 Calibration and Validation of the Proposed Wind Energy Conversion System Model This study applied the MATLAB curve fitting tool for calibrating the proposed WECS model of Eq. (3.19) using the output characteristics curves of Hummer WTGs supplied by the manufacturer (AHDC, 2013). The available wind resource (wind speed data) informed the choice of selected WTGs. The ability of selected WTGs to generate power from limited average wind speed data was also considered. The accuracy of the results reported by the manufacturers and those published in reviews is satisfactory. The maximum capacity of considered WTG is limited to 5 kW since it is intended that the WTG be mounted on the GSM BTS tower. The performance validation of the proposed WTG model for estimating wind energy generation is determined on the basis of six statistical indicators. The validation process utilized wind speed data based on the manufacturers supplied data set points. The validation results are presented in chapter 4. In addition, present study applied the HOMER software, using the Weibull shape factors (k) given in Table 2.4, to convert the available monthly daily average ground-based wind speed data, measured at anemometer height of 10 m as shown in
160
Table B11 (appendix B) for study locations to minutely average values for every minute in one year as shown in Figure B5 (appendix B). 3.9
Optimization Procedure
This section focuses on the optimization of the proposed hybrid energy system for efficient and reliable power supply to GSM BTS sites in emerging cities using Nigeria as a case study. The section begins with the formulation of the optimization problem followed by the actual design of the optimum sizing and operation strategy for effective management of energy resources. 3.9.1 Formulation of the Optimization Problem A primary concern in the design of the proposed hybrid energy system is to determine the size of each component participating in a hybrid system so that the load can be effectively satisfied under techno-economic restriction. The optimization problem is treated as a single objective problem by considering all objectives (both technical and economic) in terms of cost. The constraints that ought to be satisfied (see section 3.3.2), while minimizing the objective or cost function ensure that the load is reliably served. The extent of the system reliability in this study, measured by the LPSP, defines the long-term average fraction of the load not supplied by a hybrid system. An LPSP of 0 means the load can always be satisfied; while LPSP of 1 means that the load can never be satisfied. Consequently, the system components are subject to minimizing the total cost of the hybrid power system while ensuring that the load is served according to proposed energy management strategy.
161
Objective (Fitness) Function The objective of this study is to minimize the cost of energy (COE) of the proposed hybrid energy system, subject to reliable operation. Therefore, the cost of energy is used as an objective or fitness function. The objective or fitness function is defined by Eq. (3.120) and Eq. (3.121). Minimize the COE:
min
COEkWh ( Sz wt , Sz pv , N bb ) (cc gs Dmo cc wt . Sz wt cc pv . Sz pv ccinv Pr ,inv cc sc N sc cbb N bb cc def LPSP C sys, mgt )
.
(3.120)
Subject to:
LPSP 0.10 ,
(3.121)
where Szwt, Szpv, Nbb, are design variables to be sized by the proposed optimization model (described in section 3.9.2), which represent the required power capacity (kW) of the WTG, area (m2) of PV array, and the number of batteries in battery bank respectively. Ψ (kWh-1) is the reciprocal of the total energy drawn by the GSM BTS site from the proposed HES during the project life span, which is equal 1/Esys,tot and ccj is the cost coefficient of component j. Dmo, with a default value of zero for stand-alone and one for grid-connected operation, is a decision variable that specifies the mode of HES operation. The loss of power supply probability (LPSP) is computed using Eq. (3.65) based on the proposed energy management strategy described in section 3.3. It should be noted that the required or rated capacity of the WTG selected for optimum performance is determined by the proposed sizing model using codes given appendix E6.
162
Given that Nj,min and Nj,max are the minimum and maximum acceptable values of the decision or sizing variable Nj, the decision or sizing variables of Eq. (3.120) are determined based on the following bounds:
Sz wt,min Sz wt Sz wt,max Sz pv,min Sz pv Sz pv,max . N bb,min N bb N bb max
(3.122)
The minimum and maximum bounds chosen for Szwt are 1 kW and 5kW respectively, which corresponds to the minimum and maximum capacities of considered wind turbine generators. It is assumed that the total landmass for installing a PV array should not exceed one-fourth (1/4) of a plot of land measuring 462.10 m2. Based on this assumption, the minimum and maximum limits for the PV array are set to 1 m2 and 115 m2 respectively. Since the operation voltage of the BTS site considered is 48V, and a 6V battery is selected, a minimum of eight batteries connected in series are required to form a string of 48V battery bank. 3.9.2 Optimization of the Proposed Hybrid Energy System The hybrid Genetic Algorithm and Pattern Search (h-GAPS) technique is applied to solve the optimum design specification problem with hybrid energy resources under techno-economic restriction by efficient use of available energy resources to achieve maximum profits. It is well-known that the stochastic population-based algorithms like genetic algorithm are good at identifying promising areas of the search space (exploration). In contrast, the pattern search (PS) algorithm specifically is a more coordinate search method, which guarantees convergence to stationary points from arbitrary starting points. In addition, pattern search algorithm is useful at improving approximations to the minimum (exploitation). Thus, 163
hybridization of the GA and PS can provide a more efficient trade-off between exploration and exploitation of the search space. The hybridization can help to guarantee that the global optimum solution is selected. Figure 3.4 shows the proposed optimization and operational control model for HES.
164
Input weather/load data and technical/economic parameters of hybrid energy system
Start
Compute solar radiation on tilted PV array using fixed annual best tilt angles and store values. No
SOCsb (τ) > SOCsb,max
Modify wind speed data to GSM BTS tower height
?
Yes
Select site and operation mode
Stop SC/battery charging, set SOCsb (τ) = SOCsbmax and compute Ebal (τ)
Ebal (τ) = 0
GA parameter setting and initialization
No SOCsb (τ) > SOCsb,max
No
Ebal (τ) ≤ Ed (τ)
A
τ=1
?
?
B Yes
Yes Compute Ed (τ), Ehg (τ), Ege (τ), Esg (τ), Ewg (τ), Erg (τ) & SOCsb (τ)
Compute Egrd,net and set Ebal (τ) = 0
Re-compute Ebal (τ)
Stop SC/battery charging, set SOCsb (τ) = SOCsbmax and compute Ebal
Yes Ege (τ) > 0
No
?
Ebal (τ) ≥ Ewg (τ) ?
Turn off the WTG and dump excess energy
No Yes
SOCsb (τ) ≥ SOCsb,min
Yes
?
Turn off the WTG and dump excess energy
No Eexc (τ) = 0 Stop discharging SC/Battery, compute SOCsb (τ) and Eout (τ)
Dump Eexc (τ) = Ebal (τ)
Supply energy demand
Evaluate fitness function Mutation
No Convergent ? Yes Store result Stop Start
Selection
Crossover
C
Measure individual fitness value, COE (j)
Calculate the LPSP
iter = 0
Construct pattern vectors & create mesh points
Set starting points, Xo
Read initial values from GA
No Termination reached ?
Set mesh size & factor, max. number of evaluation function & iteration
τ = 525,600 ? No
iter = iter + 1
B
Yes
Evaluate fitness function
Half mesh size Yes
Double mesh size
End
Figure 3.4: Optimization and Operation Control Model for HES 165
A
No Poll successful ?
Yes Save design values & evaluate system‘s performance
τ=τ+1
C
Based on the estimated load profile of the GSM BTS site (discussed in section 3.6) and the available weather data measured at the study sites (see appendix B), the fixed optimum tilt angles for harnessing solar energy are determined using the method discussed in section 3.8.3, while the wind speed data were adjusted, based on power law (see Eq. (2.14)), to the GSM BTS tower height for accurately determining the available wind energy. The proposed h-GAPS based approach, shown in Figure 3.4 begins the optimum sizing of the HES using the GA, which uses three operators (selection, crossover, and mutation). The GA begins with the evaluation of the initial system design configuration or chromosome (either chosen or set by default) to determine if they provide reliable power (defined in terms of LPSP) to the load otherwise new chromosome is determined. An arrangement is considered feasible if the power system reliability is greater than or equal to 90%. Although a no load rejection ratio is desirable, and LPSP set to vary within 0 and 10% is considered here for typical applications. Next, the GA determines the evaluation qualified configuration with the best (lowest COE) value for a pre-specified number of generations or when a criterion that determines the convergence is satisfied. The Pattern search algorithm begins at the initial point X0 that is given as a starting point by the GA. At the first iteration, with a scalar =1 called mesh size, the pattern vectors are constructed as: [0 1], [1 0], [-1 0] and [0 -1]. The pattern vectors are added to the initial point X0 to compute the mesh points as: X0 + [0 1], X0 +[1 0], X0 + [-1 0] and X0 +[0 -1]. The algorithm computes the fitness function at the mesh points in the same order. The algorithm polls the mesh points by computing the fitness function values until it finds one whose value is smaller than the fitness function value of X0. If such point exists, then the poll is successful and the algorithm sets this point equal to X1; otherwise, the poll is unsuccessful. The current 166
mesh size is doubled (i.e., multiplied by an expansion factor = 2), after a successful poll so that the mesh size at the next iteration is larger, and vice versa. After a successful poll, for instance, the algorithm doubles the current mesh size and steps to the second iteration. The mesh points at the second iteration are: X1 + 2 *[0 1], X1 + 2 * [1 0], X1 + 2 * [-1 0] and X1 + 2 * [0 -1]. The algorithm polls the mesh points until it finds one whose value is smaller than the fitness function value of X1. The first such point it finds is called X2. And the algorithm doubles the current mesh size to get a mesh size of 4 at the third iteration. If the third iteration poll, with mesh size = 4, ends up being unsuccessful, the algorithm does not change the current point at the next iteration (i.e., X3 = X2). At the next iteration, after an unsuccessful poll, the algorithm divides the current mesh size (= 4) by 2, so that the algorithm polls with a smaller mesh size. The point from the previous iteration is replaced by a better point, if any. The illustrated procedure is repeated until the algorithm finds the optimum solution for the minimization of the fitness function. The algorithm stops when a predefined condition or a criterion that determines convergence is satisfied Unlike other methods, the proposed h-GAPS based approach of Figure 3.4 is capable of minimizing the fitness (COE) function of the GSM BTS site while constraining it for safely satisfying the load demand according to the reliability criteria defined by the proposed energy management strategy. The proposed energy management maintains the best compromise between system cost and reliability to guarantee continuous and reliable power supply. If the system can not efficiently supply the required energy demand, the deficit demand can be purchased from the stand-by diesel generating system via imposed cost penalty.
167
3.10
Design of Simulation Model
A computer program was written for simulation of the proposed hybrid energy system model. The program applied the GA toolbox in MATLAB software (version R2012b) and was implemented on a laptop computer (Window 8 Pro, 64-bits operating system with processor speed of 2.8GHz and 4.0GB RAM). The GA tool contains the elitist approach, which ensures that the best individual of a generation is copied to the next generation without any changes being made to it. The design utilized default GA toolbox parameter settings with the following changes. The ―creation function‖ was set to ―feasible population‖, the ―initial range function‖ varied from ―1; 10‖ to ―1; 40‖ for various locations of study, mutation function set to ―adaptive feasible‖ and hybrid function set to ―pattern search‖ as shown in Figure 3.5.
Figure 3.5: GA toolbox parameter settings for the simulation model
168
The individuals of the considered optimization problem contain three variables (or genes), which are Szpv, Szwt and Nbb. Their bounds are entered directly in the dedicated positions of the GA toolbox as shown in Figure 3.5. A maximum rating of 5 kW WTG was chosen for technical reasons as it is proposed that the WTG be mounted on the GSM BTS tower. Dmo is set to the default value of zero for stand-alone and one for grid-connected mode of operation. Four M-files written in MATLAB codes (see appendix E) in conjunction with the GA toolbox (shown in Figure 3.5) are used for simulation of the proposed optimization and process control model of the hybrid energy system. The M-files consist of a central file (obj.m) and three sub-files. The first sub-file (input_data.m) inputs all required data for the process simulation. The second sub-file (cost_coe.m) computes all cost coefficients for the fitness function while the third sub-file (cstrategy.m) computes the non-linear constraints for satisfactory performance. The central file (obj.m) in conjunction with the GA toolbox combines all sub-files, computes the fitness function based on proposed optimization and process control model of Figure 3.4 and determines the optimum configuration for selected location/operation mode. An additional M-file (disp_results.m) is also created to display the optimization and performance analysis results. The MATLAB simulation scripts for the proposed hybrid energy system are presented in appendix E (E1 – E6). 3.10.1 Case Studies: Process Simulation and Application The proposed optimization and operation control model for a hybrid energy system was simulated for a typical GSM BTS site in Nigeria. The overall process simulation utilized the adjusted typical load profile of a GSM S/4/4/4 out-door base station site. The adjusted load profile, shown in Figure 3.6, which accounts for real-time variations (as discussed in section
169
3.6) is considered for six locations (Abuja, Benin City, Enugu, Ikeja, Maiduguri and Sokoto) representing the six geopolitical zones in Nigeria. The study locations were carefully chosen to cut across the different climatic zones in Nigeria. The purpose is to ensure that the energy related problems of the GSM BTS sites are actually addressed in Nigeria. This study applied synthesized minutely (minutes) solar radiations and wind speed data discussed in sections 3.8.3 and 3.8.4 respectively for simulation of proposed hybrid power system. Table 3.2 shows the economic specifications of components that constitute inputs of the proposed hybrid power system sizing. The project life span is considered to be the same as that of the PV module, which has a longer life span of 25 years. The cost of developing the energy management or simulation software (Csys,mgt) is taken to be N 100,000.00. The reason is that the present research is sponsored by the Federal Government of Nigeria through the Tertiary Education Trust Fund for Academic Staff Training and Development (TETFund AST & D). In addition, if the simulation software is made available to the public at a subsidized rate of N 100,000.00, it can facilitate the implementation of the proposed HES. The annual fixed charges for connecting to the utility grid and the unit price of energy drawn from the grid are given in the literature (see Table 2.2). However, a micro-grid interconnection fee of N 100,000.00 is considered. The estimated social cost (N 5,920.00 per ton of CO2 emission), given by the United States is used to cost carbon emissions. Although, an independent research carried out by Moore and Diaz (2015) showed that the cost carbon emissions is not as previously estimated, but N 35,200.00 per ton of CO2 emission, the finding is still a subject of discourse and it has not been globally accepted.
170
Table 3.2: Economic specifications of components for optimization of the proposed hybrid energy system (Nandi and Ghosh, 2010; AHDC, 2013; SEDC, 2013; MTI, 2014, Ebay, 2014) Specifications Micro-grid components
Model
Units
Life time
yr
Purchase cost per unit size
Installation cost per unit size
Operation and maintenance cost, com,j
Salvage value, svj
Naira per unit
Naira per unit
Naira per unit per annum
Naira per unit
H3.1-1 kW
15
200,000.0
20,000.0
2,000.0
4,800.0
H3.8-2 kW
15
314,560.0
31,456.0
3,145.6
7,864.0
H4.6-3 kW
15
826,240.0
82,624.0
8,262.4
20,656.0
H6.4-5 kW
15
1,139,680.0
113,968.0
11,396.8
28,492.8
PV module
CNSDPV 150 Monocrystalline
25
37,600.0
3,760.0
38.4
1,880.0
Battery
USB US-250
Float life: 10 (Lifetime throughput: 845 kWh)
32,000.0
3,200.0
160.0
0
Inverter
MLP-1000 (1kW)
15
40,000.0
4,000.0
192.0
0
Rectifier
Typical (1kW)
15
40,000.0
4,000.0
192.0
0
Stabilizer
Century CVR-TUB 5000 VA
15
2,560.0
256.0
128.0
0
Supercapacitor
Maxwell Technologies BM0D0165 P048BXX
10
264,000.0
2,640.0
Wind turbine
0
0
Note: The conversion is based on an exchange rate of US$1 ≈ N160.0 (Central Bank of Nigeria, accessed April 21, 2014).
The capital and maintenance costs of a typical diesel generator (utilized for comparison study) are N 45,000.00 per kW and N 20.00 per hour respectively with a service life of 15,000 h (Trazouei et al., 2013; Girma, 2013). The fuel price varies from one location to another (considered here as N 160.00 – N 174.00 per litre) owing to additional cost of 171
transportation of the fuel that varies from one part of the country to another. It is assumed that the cost implication for not supplying a kWh load demand at any hour (i.e., cost implication per unit unmet energy demand) is equal to the COE per kWh for diesel-generator system. The COE per kWh for the diesel-generator system is determined using the Hybrid Optimization Model for Energy Renewables (HOMER).
As discussed in the literature, HOMER is the most widely used software (computer model) with a high degree of accuracy for the design and analysis of stand-alone energy systems. For simplicity, the environmental cost implication of the diesel power generated is neglected in COE analysis. In other words, the cost penalty for emission of pollutant gases by the diesel generator into the atmosphere is neglected. The components of the energy cost are the capital cost, replacement cost, and the operation and maintenance costs of the diesel generation system. The design configuration for a loss of power supply to GSM BTS sites below 10% (LPSP < 0.10) intended for this study is presented, discussed and compared with traditional approaches in the following chapter.
172
CHAPTER FOUR 4.0 4.1
RESULTS, DISCUSSION AND FINDINGS Results
The simulated load profile of the studied outdoor GSM BTS site for a simulation period of one year (525,600 min) is shown in Figure 4.1.
2.6 2.4 Pd (kW)
2.2 2.0 1.8 1.6 1.4 0
175,200
350,400
525,600
Time (mins)
Figure 4.1: Simulated minute load profile for an outdoor GSM BTS (S/4/4/4) site in Nigeria. The estimated voltage profile of the Nigerian electric power grid, computed using Eq. (3.5), is shown in Figure 4.2.
173
300
Voltage (V)
250 200 150 100 50 0 0
1460
(a)
2920
4380
5840
7300
8760
Time (h)
Power access/outage (h)
4200 3500 2800 2100 1400 700 0 (b)
0
50
100 150 Voltage (V)
200
250
Figure 4.2: The Nigerian grid supply voltage profile simulated for a period of 1 year (a) Voltage magnitude (b) Power access/outage frequency. The calibration and validation results of the six global solar radiation models studied (discussed in section 3.8.1) for different locations in Nigeria are shown in Tables 4.1 and 4.2. Based on the established regression coefficients shown in Table 4.1, the monthly average
174
daily global solar radiations (H) for each city selected from the six geopolitical zones in Nigeria were estimated as shown in Figure 4.3. Table 4.1: Calibration results of various global solar radiation models (using group-1 data sets for a period of 22-years) along with R2 and t-stat values
Region
Global solar radiation models
b1
b2
b3
b4
R2
t-stat
0.1540
0.6901
0
0
0
0.9412
0.0011
Badescu (1999)
0.7334
-0.0393
0
0
0
0.8111
0.0023
Chen et al. (2004)
0.2116
0.1730
0
0
0
0.8882
0.0029
El-Metwally (2004)
-4.3764
0.6900
0.0611
0.1559
-0.4985
0.9722
0.0012
Falayi et al. (2008)
-0.1674
0.6286
0.5482
0.0036
-0.0022
0.9874
0.0085
-0.1104
0.1370
0.3899
0.0155
-0.0365
0.9734
0.026
0.1984
0.6855
0
0
0
0.7673
0.0016
Badescu (1999)
0.7598
-0.0551
0
0
0
0.8806
0.0048
Chen et al. (2004)
-0.0231
0.3026
0
0
0
0.8931
0.0026
El-Metwally (2004)
-8.2997
0.7735
0.0747
0.2751
-0.5822
0.9939
0.0026
Falayi et al. (2008)
0.7017
0.2673
-1.1291
0.0218
0.0003
0.9849
0.218
-0.4904
0.0839
0.3215
0.0302
-0.0338
0.9931
0.0148
(1940)
Nigeria
Present study [Eq. (3.22)] Angstrom-Prescott (1940)
Southern Nigeria
Statistics
bo
Angstrom-Prescott
Northern
Regression coefficients
Present study [Eq. (3.22)]
175
Table 4.2: Validation results of various global solar radiation models using group-2 data sets for a period of 22-years Site
Error terms
Angstrom Badescu -Prescott (1999) (1940)
Chen et al. (2004)
ElMetwally (2004)
Falayi et al. (2008)
r
0.5788
0.9306
0.7925
0.9243
0.8246
0.9170
0.5788
0.9306
RMSE
0.6145
0.4816
0.4351
0.2821
0.4511
0.2646
0.2646
0.6145
MBE
-0.3547
-0.3133
-0.2256
-0.0333
-0.2776
-0.0598
0.0333
0.3547
MABE
0.5087
0.3453
0.3057
0.2367
0.4111
0.2074
0.2074
0.5087
r
0.8866
0.8558
0.8855
0.8577
0.9635
0.9035
0.8558
0.9635
Maiduguri RMSE (NE) MBE
0.3293
0.3936
0.3122
0.4306
0.2297
0.3414
0.2297
0.4306
-0.1592
-0.0899
-0.0829
0.2032
0.0010
0.1452
0.0010
0.2032
MABE
0.2707
0.3392
0.2810
0.3478
0.1714
0.2742
0.1714
0.3478
r
0.9042
0.9159
0.9023
0.9505
0.9471
0.9732
0.9023
0.9732
RMSE
0.3735
0.3560
0.3311
0.2281
0.3222
0.1986
0.1986
0.3735
MBE
-0.2142
0.1052
-0.0530
-0.0889
-0.2362
-0.1226
0.0530
0.2362
MABE
0.2875
0.2786
0.2851
0.1721
0.2859
0.1530
0.1530
0.2875
r
0.6896
0.8870
0.9021
0.9745
0.9264
0.9697
0.6896
0.9745
Ikeja
RMSE
0.4745
0.3434
1.7431
0.6550
0.8862
0.2150
0.2150
1.7431
(SW)
MBE
-0.1659
0.2149
-1.7249
0.6398
-0.8573
0.1607
0.1607
1.7249
MABE
0.3641
0.3029
1.7249
0.6398
0.8573
0.1607
0.1607
1.7249
r
0.8592
0.9023
0.9083
0.9941
0.9820
0.9914
0.8592
0.9941
Enugu
RMSE
0.3848
0.3281
0.4182
0.2436
0.3535
0.1237
0.1237
0.4182
(SE)
MBE
0.1414
-0.1924
0.3277
-0.2324
0.3061
-0.0467
0.0467
0.3277
MABE
0.3393
0.2452
0.3329
0.2324
0.3061
0.1075
0.1075
0.3393
r
0.8186
0.9246
0.9066
0.9884
0.9787
0.9879
0.8186
0.9884
Benin City
RMSE
0.4225
0.2937
0.4565
0.1958
0.2527
0.2066
0.1958
0.4565
(SS)
MBE
-0.1014
0.1299
0.3470
0.1579
0.2092
0.1746
0.1014
0.3470
MABE
0.3303
0.2698
0.3470
0.1683
0.2285
0.1746
0.1683
0.3470
(Zone)
Sokoto (NW)
Abuja (NC)
*
Present Range of values study (magnitude) [Eq . (3.22)] Min. Max.
Error terms (RMSE, MBE and MABE) are expressed in kWh m–2 day–1
176
8
6
6
H (kWh m -2 day-1)
H (kWh m -2 day-1)
8
4
2
2
3
4
5
6 7 8 Months
8
8
6
6
4
2
0 1 (c)
2
3
4
5
6 7 8 Months
H (kWh m -2 day-1)
H (kWh m -2 day-1)
6
4
5
6 7 8 Months
9 10 11 12
2
3
4
5
6 7 8 Months
9 10 11 12
2
6
4
22 2
2
(e)
1
2
3
20
1
18
2 16
3
4
4
H Measured H Measured 0 1 2 3 (1940) 4 5 6 7 8 9 10 11 12 5 6 7 8 9 10 11 Angstrom-prescott 12 (f) Angstrom-prescott (1940) 18 Months Months Badescu (1999) Badescu (1999) H Measured Chen et al. (2004) Chen 16 et al. (2004) Angstrom-prescott (1940) El-Metwally (2004) El-Metwally (2004) 14 Badescu (1999) Falayi et al. (2008) Falayi et al. (2008) 5 Chen 6 et 7al. (2004) 8 9 10 11 12 Present study [Eq.(3.21)] Present 12 study [Eq.(3.22)] Months (2008) El-Metwally H (MJ m-2 day-1)
220
H (MJ m-2 day-1)
H (kWh m -2 day -1) (f)
4
20
4
0
3
8
6
2
2
4
0 1 (d)
9 10 11 12
8
8
2
0 1 (b)
9 10 11 12
H (kWh m -2 day-1)
H (kWh m -2 day-1)
0 1 (a)
4
Figure of theetmeasured and estimated monthly average daily global solar 14 4.3: ComparisonFalayi al. (2008) Present surface study [Eq.(3.21)] radiations on a horizontal (in the10 periods 0 10 20of 1984 – 2005) for (a) Sokoto,
12
Monthsand (f) Benin City respectively (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu,
10
0 10 20 Months
177
For ease of comparison, the validation results of studied global solar radiation models for the six geopolitical zones (Table 4.2) are also presented in Figure 4.4, while the magnitude of errors from estimating the monthly daily global solar radiation for the six geopolitical zones in Nigeria is shown in Figure 4.5.
2.0
1.0 RMSE (kWh m -2 day -1)
0.9 0.8 r-value
0.7 0.6 0.5 0.4 0.3 0.2
1.6 1.2 0.8 0.4
0.1 0 (a)
Sokoto Maiduguri
Abuja Ikeja Locations
0
Enugu Benin City
(b)
Maiduguri
Abuja Ikeja Locations
Enugu Benin City
Sokoto
Maiduguri
Abuja
Enugu Benin City
2.0
0.8
MABE (kWh m -2 day -1)
0.4 MBE (kWh m -2 day -1)
Sokoto
-0 -0.4 -0.8 -1.2
1.6 1.2 0.8 0.4
-1.6 Sokoto Maiduguri
Abuja Ikeja Locations
MABE (kWh m -2 day -1)
(c)
MABE (kWh m
day )
2.0 1.6 1.2
Figure 4.4:
2.0
Enugu Benin City
0 (d)
1.6
Angstrom-prescott (1940) 1.2 Badescu (1999) Chen et al. (2004) 0.8 El-Metwally (2004) Falayi et al. (2008) Comparison of performance 0.4(3.22)] Present study [Eq.
Ikeja
Locations Angstrom-prescott (1940) Badescu (1999) Chen et al. (2004) El-Metwally (2004) Falayi et al. (2008) Present study [Eq. (3.22)]
indices for various global solar radiation models:
0.8
(a) r-value, (b) RMSE, (c) MBE and (d) MABE. 0
0.4 0
(d)
(d)
Sokoto
Maiduguri
Abuja Ikeja Locations
Sokoto
Maiduguri
Enugu Benin City
178
Abuja Ikeja Locations
Enugu Benin City
40
30
30
20
20
10
10
RPE (%)
RPE (%)
40
0 -10
-10
-20
-20
-30
-30
-40
1
2
3
4
5
(a)
6 7 8 Months
-40
9 10 11 12
1
2
3
4
5
6 7 8 Months
9 10 11 12
1
2
3
4
5
6 7 8 Months
9 10 11 12
1
2
3
4
5
6 7 8 Months
(b)
40
40
30
30
20
20
10
10
RPE (%)
RPE (%)
0
0 -10 -20
0 -10 -20
-30
-30
-40 1
3
4
5
6 7 8 Months
-40
9 10 11 12
(d)
40
40
30
30
20
20
10
10
RPE (%)
RPE (%)
(c)
2
0 -10
0 -10
-20
-20
-30
-30
-40
1
2
3
4
(e)
5
6 7 8 Months
Sokoto
-40
9 10 11 12
(f )
Maiduguri
Abuja
Ikeja
9 10 11 12
Enugu
Benin City
45
Figure 4.5: Comparison of the monthly relative percentage error (RPE) of global solar 30
%
radiation estimates from (a) Angstrom–Prescott (1940), (b) Badescu (1999), (c) Chen et al. 15 0 (d) El-Metwally (2004), (e) Falayi et al. (2008), and (f) Present study [Eq. (3.22)]. (2004), -15 -30 -45
179 (ff) 1
2
3
4
5
6 7 Months
8
9
10
11
12
The calibration results of the six diffuse solar radiation models studied (discussed in section 3.3.2) along with their R2 and t-stat (used to determine the statistical significance or accuracy of the calibrated parameters (bi) for different locations in Nigeria) are shown in Table 4.3. Based on the established regression coefficients shown in Table 4.3, the monthly average daily diffuse solar radiations (Hd) for each city selected from the six geopolitical zones in Nigeria were estimated as shown in Figure 4.6. Table 4.3: Calibration results of different diffuse solar radiation models (using group-1 data sets for a period of 22-years) along with R2 and t-stat values Region
Northern
Studied diffuse
Regression coefficients
Statistics
radiation models
bo
b1
b2
b3
R2
t-stat
Page (1964)
1.0430
-1.2287
0
0
0.9859
0.0065
Liu-Jordan (1960)
7.5894
-37.2990
65.7417
-39.636
0.9919
0.0104
Butt et al. (2010)
0.1321
0.0508
0
0
0.8849
0.0280
Karakoti et al. (2011)
-0.1703
1.9835
-3.1048
1.3632
0.9515
0.0026
0.7942
-0.3350
-0.7921
0
0.9869
0.0073
0.8322
-0.9545
0.0133
0
0.9973
0.0423
Page (1964)
1.0424
-1.2460
0
0
0.9915
0.0150
Liu-Jordan (1960)
2.3860
-10.1075
19.1918
-13.651
0.9931
0.0093
Butt et al. (2010)
0.0859
0.0706
0
0
0.9223
0.0788
Karakoti et al. (2011)
0.2557
-0.4385
1.5743
-1.8683
0.4267
0.0041
0.9547
-0.8682
-0.3964
0
0.9919
0.0186
0.8391
-0.9912
0
0
0.9972
0.0098
Nigeria Present study [Eq. (3.23)] Present study [Eq. (3.24)]
Southern Nigeria
Present study [Eq. (3.23)] Present study [Eq. (3.24)]
180
3.0
3.0
2.5 Hd (kWh m -2 day-1)
Hd (kWh m -2 day-1)
2.5 2.0 1.5 1.0
2.0 1.5 1.0 0.5
0.5 0
1 2
3 4
5 6 7 8 Months
0
9 10 11 12
(b)
3.0
3.0
2.5
2.5 Hd (kWh m -2 day-1)
Hd (kWh m -2 day-1)
(a)
2.0 1.5 1.0
1 2
3 4
5 6 7 8 Months
1 2
3 4
5 6 7 8 Months
9 10 11 12
1.0
(d)
10 2.5
2.5 Hd (kWh m -2 day-1)
3.0
2.0
2.0
8
1.5
1.5
6 1.0
1.0 0.5
0.5
4 0
2
0
1 2
3 4
5 6 7 8 Months
2
Page (1964)
Karakoti et al. (2011)
0
9 10 11 12
4
Hd Measured
Butt et al(2010) 3.0
9 10 11 12
1.5
3.0
(e)
5 6 7 8 Months
2.0
0
9 10 11 12
d
-1 m-2 d) Hd (MJ m -2 )day-1) H (kWh
(c)
3 4
0.5
0.5 0
1 2
(f)
1
2
3
4
5
6 8 Months Liu-Jordan (1960)
Present study [Eq. (3.23)]
6 7 8 Months
9 10 11 12
10
12
Butt et al (2010)
Present study [Eq. (3.24)]
Figure 4.6: Comparison of the measured and estimated monthly average daily diffuse solar radiations (1984–2005) for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, 2.5
and (f) Benin City respectively 181
day -1)
2.0
Karakoti e
The performance indices used for validating different diffuse solar radiation models are shown in Table 4.4 and Figures 4.7 and 4.8. Table 4.4: Validation results of different diffuse solar radiation models using group-2 data sets for a period of 22-years Site
Error terms
Page (1964)
LiuJordan (1960)
Butt et al. Karakoti Present Present Range (2010) et al. study study (magnitude) (2011) [Eq. (3.23)] [Eq. (3.24)] Min. Max.
r
0.9502
0.8285
0.8636
0.9040
0.9323
0.9955
0.8285
0.9955
RMSE
0.1288
0.2948
0.2702
0.1958
0.1503
0.0489
0.0489
0.2948
MBE
-0.0764
-0.1972
0.1676
0.0829
-0.0955
-0.0310
0.0310
0.1972
MABE
0.1039
0.2122
0.1927
0.1692
0.1152
0.0419
0.0419
0.2122
r
0.9490
0.8372
0.9005
0.9288
0.9331
0.9940
0.8372
0.9940
RMSE
0.1156
0.2275
0.2110
0.1281
0.1293
0.0600
0.0600
0.2275
MBE
-0.0557
-0.1079
0.0096
0.0055
-0.0612
-0.0490
0.0055
0.1079
MABE
0.0964
0.1677
0.1890
0.1060
0.1042
0.0490
0.0490
0.1890
r
0.9937
0.9753
0.8866
0.8746
0.9941
0.9974
0.8746
0.9974
RMSE
0.0430
0.1246
0.1826
0.1783
0.0459
0.0312
0.0312
0.1826
MBE
-0.0084
0.0102
-0.0529
0.0177
-0.0209
-0.0163
0.0084
0.0529
MABE
0.0312
0.0807
0.1483
0.1355
0.0356
0.0239
0.0239
0.1483
r
0.9882
0.9867
0.9089
0.8863
0.9876
0.9957
0.8863
0.9957
Ikeja
RMSE
0.0419
0.0500
0.1792
0.2145
0.0412
0.0243
0.0243
0.2145
(SW)
MBE
-0.0190
-0.0299
0.1025
-0.1286
-0.0196
0.0047
0.0047
0.1286
MABE
0.0323
0.0423
0.1398
0.1475
0.0348
0.0206
0.0206
0.1475
r
0.9831
0.9908
0.9259
0.8529
0.9844
0.9955
0.8529
0.9955
Enugu
RMSE
0.0451
0.0427
0.1861
0.1315
0.0422
0.0436
0.0422
0.1861
(SE)
MBE
-0.0098
-0.0050
-0.1362
0.0145
-0.0073
-0.0395
0.0050
0.1362
MABE
0.0378
0.0347
0.1619
0.1015
0.0363
0.0395
0.0347
0.1619
(Zone)
Sokoto (NW)
Maiduguri (NE)
Abuja (NC)
r
0.9870
0.9872
0.9480
0.8945
0.9875
0.9935
0.8945
0.9935
Benin City
RMSE
0.0399
0.0361
0.1530
0.0932
0.0376
0.0352
0.0352
0.1530
(SS)
MBE
-0.0086
-0.0011
-0.0811
0.0062
-0.0090
-0.0255
0.0011
0.0811
MABE
0.0330
0.0274
0.1375
0.0739
0.0312
0.0305
0.0274
0.1375
*
Error terms (RMSE, MBE, and MABE) are expressed in kWh m–2 day–1
182
1.0 0.9 RMSE (kWh m -2 day -1)
0.3
0.8 0.7 r-value
0.6 0.5 0.4 0.3 0.2
0.2
0.1
0.1 0 (a)
Sokoto
Maiduguri
Abuja Ikeja Locations
0
Enugu Benin City
(b)
Sokoto
Maiduguri
Abuja Ikeja Locations
Enugu Benin City
0
-0.1
-0.2
MABE (kWh m -2 day -1)
(c)
0.2
0.1
0 Sokoto
Maiduguri
0.2
Figure 4.7:
Abuja Ikeja Locations
Enugu Benin City
(d)
0.2
Page (1964) Liu-Jordan (1960) Butt et al. (2010) 0.1et al. (2011) Karakoti Present study [Eq. (3.23)] Comparison ofstudy performance indices Present [Eq. (3.24)]
(1964)Abuja SokotoPage Maiduguri Ikeja Locations Liu-Jordan (1960)
Enugu Benin City
Butt et al. (2010) Karakoti et al. (2011) Present study [Eq. (3.23)] Present study [Eq. (3.24)]
for various diffuse solar radiation models:
0.1
(a) r-value, (b) RMSE, (c) MBE and (d) MABE. 0
(d)
0 (d)
MABE (kWh m -2 day -1)
0.1
MABE (kWh m -2 day -1)
MBE (kWh m -2 day -1)
0.2
Sokoto
Maiduguri
Abuja Ikeja Locations
Sokoto
Maiduguri
Enugu Benin City
183
Abuja Ikeja Locations
Enugu Benin City
30
30
20
20
10
10
RPE(%)
40
0 -10
0 -10
-20
-20
-30
-30
-40
1
4
5
6 7 8 Months
-40
9 10 11 12
(b)
40
40
30
30
20
20
10
10
0 -10
-30
-30 1
2
3
4
5
6 7 8 Months
-40
9 10 11 12
(d)
40
40
30
30
20
20
10
10
0 -10
-30
-30
(e)
2
3
4
5
6 7 8 Months
Sokoto
-40
9 10 11 12
Maiduguri
4
5
6 7 8 Months
9 10 11 12
1
2
3
4
5
6 7 8 Months
9 10 11 12
1
2
3
4
5
6 7 8 Months
9 10 11 12
-10 -20
1
3
0
-20
-40
2
-10 -20
-40
1
0
-20
(c)
RPE(%)
3
RPE(%)
RPE(%)
(a)
2
RPE(%)
RPE(%)
40
(f)
Abuja
Ikeja
Enugu
Benin City
45
Figure 4.8: A comparison of the monthly relative percentage error (RPE) of diffuse solar 30
radiation estimates from (a) Page (1964), (b) Liu-Jordan (1960), (c) Butt et al. (2010), %
15
(d) Karakoti et al. (2011), (e) Present study (Eq. 3.23), and (f) Present study (Eq. 3.24). 0
-15 -30 -45
184 (ff) 1
2
3
4
5
6 7 Months
8
9
10
11
12
Tables 4.5 – 4.7 show the results of the analysis of total global solar radiation for the different optimization techniques at optimum tilt angles for locations in Nigeria used in this study. These results were deduced using the Hay-Davies-Klucher-Reindl (HDKR) model (Duffie and Beckman, 2006) discussed in the literature (see Eq. (2.53)). Table 4.5: Result of analysis (based on HDKR model) of influence of annual-based optimum tilt angles for selected areas in Nigeria Sokoto
Maiduguri
Abuja
Ikeja
Enugu
Benin City
Months Global Irradiance (kWh m-2 month-1) January
187.5
190.2
208.0
184.5
201.2
198.6
February
192.5
189.8
192.2
167.8
179.4
178.0
March
218.5
227.8
192.0
172.5
170.1
168.2
April
207.6
211.9
157.9
154.1
148.1
149.8
May
198.2
198.7
149.3
146.6
140.4
141.4
June
172.5
169.4
131.1
125.7
119.4
119.1
July
164.0
159.1
120.5
118.5
111.3
108.8
August
167.0
159.6
123.3
121.7
113.0
109.6
September
183.1
179.0
138.6
124.5
122.1
119.3
October
222.2
217.1
170.6
147.6
143.4
143.6
November
215.0
207.5
200.8
167.8
167.6
167.4
December
195.5
186.5
216.4
185.0
193.9
188.4
2323.7
2296.5
2000.6
1816.2
1810.1
1792.2
15.0
14.0
15.0
11.0
12.0
12.0
ϕ+2
ϕ+2
ϕ+6
ϕ + 4.4
ϕ + 5.5
ϕ + 5.7
Total (kWh m-2 yr-1) Optimum
Tilt
o
angle, β ( ) Opt. tilt in terms of latitude, ϕ (o)
185
Table 4.6: Result of analysis (based on HDKR model) of influence of seasonal-based optimum tilt angles for selected locations in Nigeria Sokoto
Maiduguri
Abuja
Ikeja
Enugu
Benin City
Months Global Irradiance (kWh m-2 month-1) January
211.4
213.9
227.8
202.6
220.2
216.4
February
203.8
200.8
198.7
174.0
185.3
183.7
March
208.6
218.4
181.8
164.5
161.4
160.3
April
215.7
220.1
166.4
160.1
154.7
156.4
May
219.6
218.9
166.8
158.5
153.3
154.2
June
219.6
218.9
166.8
158.5
153.3
154.2
July
183.9
176.8
134.9
128.3
121.8
118.2
August
177.3
168.9
131.7
127.7
119.1
115.2
September
180.9
177.5
139.4
125.2
123.1
120.3
October
227.4
222.1
171.0
148.3
143.3
143.6
November
239.1
229.7
217.0
181.1
180.1
179.3
December
224.5
213.1
241.2
206.7
215.9
208.5
Total (kWh m-2 yr-1)
2489.4
2452.0
2126.6
1914.4
1910.9
1888.0
Optimum Tilt angle (o) Dry season 37.0
35.0
34.0
30.0
31.0
30.0
0
0
0
0
0
0
(October – March) Wet season (April - September)
186
Table 4.7: Result of analysis (based on HDKR model) of influence of monthly-based optimum tilt angles for selected locations in Nigeria Sokoto
Maiduguri
Months
Abuja
Ikeja
Enugu
Benin City
Global Irradiance (kWh m-2 month-1)
January
213.2
216.3
229.5
204.5
222.0
218.4
February
203.9
200.8
199.0
174.1
185.5
183.8
March
218.8
228.1
192.0
172.5
170.2
168.3
April
215.7
220.1
166.4
160.1
154.7
156.4
May
219.6
218.9
166.8
158.5
153.3
154.2
June
197.2
191.7
150.0
137.5
132.8
132.0
July
183.9
176.8
134.9
128.3
121.8
118.2
August
177.3
168.9
131.7
127.7
119.1
115.2
September
183.5
179.5
139.9
125.3
123.2
120.4
October
229.0
223.3
172.8
149.6
145.0
145.0
November
240.2
231.1
217.7
181.8
180.6
180.0
December
228.0
217.0
244.7
210.2
219.1
212.0
Total (kWh m-2 yr-1)
2510.3
2472.5
2145.5
1930.1
1927.1
1903.9
Optimum tilt angles (o) January
45.0
44.0
41.0
38.0
39.0
38.0
February
35.0
34.0
31.0
28.0
28.0
28.0
March
18.0
17.0
14.0
11.0
11.0
10.0
April
0
0
0
0
0
0
May
0
0
0
0
0
0
June
0
0
0
0
0
0
July
0
0
0
0
0
0
August
0
0
0
0
0
0
September
11.0
9.0
6.0
3.0
3.0
3.0
October
30.0
29.0
25.0
21.0
21.0
21.0
November
43.0
42.0
39.0
36.0
36.0
35.0
December
48.0
46.0
44.0
41.0
41.0
41.0
187
Table 4.8 and Figure 4.9 show the comparison of the annual total global solar radiation for various techniques. Table 4.8: Comparison of different optimization methods (using HDKR model) for tracking solar radiations in Nigeria Sokoto
Maiduguri
Abuja
Ikeja
Enugu
Benin City
Global Irradiance (kWh m-2 month-1) Latitude (o)
13.06
12.00
9.08
6.58
6.50
6.34
Longitude (o)
5.25
13.33
7.53
3.33
7.50
5.52
2255.9
2239.1
1943.4
1788.6
1777.5
1760.5
2323.7
2296.5
2000.6
1816.2
1810.1
1792.2
3.0
2.6
2.9
1.5
1.8
1.8
2489.4
2452.0
2126.6
1914.4
1910.9
1888.0
10.4
9.5
9.4
7.0
7.5
7.2
2510.3
2472.5
2145.5
1930.1
1927.1
1903.9
11.3
10.4
10.4
7.9
8.4
8.1
Horizontal Total It (kWh m-2 yr-1) Annual-based tilt Total It (kWh m-2 yr-1) Increase (%) Seasonal-based tilt Total It (kWh m-2 yr-1) Increase (%) Monthly-based tilt Total It (kWh m-2 yr-1) Increase (%)
188
Global Irradiance (kWh m-2 yr-1)
3000 2500 2000
Horizontal Tilted (Annually) Tilted (Seasonally) Tilted (Monthly)
1500 1000 500 0
Sokoto
Maiduguri
Abuja Ikeja Locations
Enugu Benin City
Figure 4.9: Comparison of annual total solar irradiance on a horizontal and tilted surface for various locations in Nigeria The estimated (using annual optimum tilt angles given in Table 4.5) minutely solar irradiations are shown in Figure 4.10. These radiation data are used as input data for the simulation of the proposed hybrid energy system.
189
1.6
1.2
1.2 It (kW m -2)
It (kW m -2)
1.6
0.8
0.8
0.4
0.4
0 175,200 350,400 Minutes (min)
525,600
(b)
1.6
1.6
1.2
1.2 It (kW m -2)
It (kW m -2)
0
0
(a)
0.8
0
175,200 350,400 Minutes (min)
525,600
0
175,200 350,400 Minutes (min)
525,600
0
175,200 350,400 Minutes (min)
0.8
(d)
1.6
1.6
1.2
1.2 It (kW m -2)
It (kW m -2)
525,600
0
0
0.8
0.8
0.4
0.4
0
0 (e)
175,200 350,400 Minutes (min)
0.4
0.4
(c)
0
0
175,200 350,400 Minutes (min)
525,600
(f)
525,600
Figure 4.10: A comparison of annual minutely global irradiance at annual optimum tilt angle for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu and (f) Benin City.
190
The calibration result of the proposed WTG model of Eq. (3.19), deduced using the manufacturer‘s supplied power profile as discussed in section 3.8.4 is shown in Table 4.9, while Table 4.10 shows the validation results. Table 4.9: Calibration results of proposed WTG model [Eq. (3.19)] using manufacturer‘s supplied data. Model-
Parameters
Capacity
(m s-1)
H3.1-1kW
H3.8-2kW
H4.6-3kW
H6.4-5kW
i
ai
ai1
ai2
ai3
vci = 3, v1 = 9,
1
0.03433
-0.0596
0.0216
-0.000333
v2 = 13
2
-3.6800
1.15
-0.0925
0.00250
vco = 25
3
1.6890
-0.1117
0.00706
-0.000136
vci = 3, v1 = 8,
1
-0.3356
0.2489
-0.05177
0.006472
v2 = 13
2
-8.902
2.752
-0.2314
0.006537
vco = 25
3
1.994
0.015
-0.0003333
0
vci = 3, v1 = 8,
1
0.1007
-0.1779
0.06548
-0.00111
v2 = 12
2
-27.0
7.819
-0.6762
0.01958
vco = 25
3
1.119
0.3578
-0.0192
0.000372
vci = 3, v1 = 8,
1
-0.65
0.3674
-0.03143
0.004722
v2 = 13
2
-13.7
2.971
-0.1159
0
vco = 25
3
4.985
0.0375
-0.0008333
0
191
Table 4.10: Validation results for proposed WTG model Model-Capacity Statistics H3.1-1kW
H3.8-2kW
H4.6-3kW
H6.4-5kW
r
1.0000
1.0000
1.0000
0.9999
R2
0.9999
0.9999
0.9999
0.9998
t-stat
0.0640
0.6306
0.0560
0.0918
RMSE (kW)
0.0052
0.0105
0.0374
0.0374
MBE (kW)
-0.0001
-0.0020
-0.0003
-0.0010
MABE (kW)
0.0030
0.0084
0.0125
0.0241
A comparison of the fitted power profile (using least-squares method) and the manufacturer supplied WTG power characteristics are also shown in Figure 4.11, to validate the predictive accuracy of the proposed model of Eq. (3.19). Figure 4.12 shows the adjusted wind speed profile at the GSM BTS tower height for various locations in Nigeria.
192
2.5 WTG Output power (kW)
WTG Output power (kW)
1.2
0.8
0.4
0
2.0 1.5 1.0 0.5 0
0
2
4
6
(a)
8 10 12 14 16 18 20 22 -1
Wind speed (m s )
(b)
WTG Output power (kW)
WTG Output (kW)(kW) powerpower WTG Output
3.0
8
2.5 2.0 6 1.5
4 1.0 0.5
2 0
2
4
(c)
0
(d)
2
4
6
8 10 12 14 16 18 20
Wind speed (m s -1) 6
3.5
0
0
6
8 10 12 14 16 18 20 -1
Wind speed (m s )
0
2
4
6
8
5 4 3 2 1 0
(d)
0
2
6
8 10 12 14 16 18 20
Wind speed (m s -1)
10 12 Wind speed (m s -1)
Real power profile
4
14
16
18
20
Fitted power profile
Figure 4.11: Comparison of the proposed WTG power profile and the manufacturer supplied power characteristics for (a) H3.1-1kW, (b) H3.8-2kW, (c) H4.6-3kW, and (d) H6.4-5kW WTGs respectively.
193
25
15 Wind speed (m s -1)
Wind speed (m s -1)
20 15 10 5 0
0
525,600
20
15
15
10
5
0
(c)
175,200 350,400 Time (min)
0
175,200 350,400 Time (min)
0
175,200 350,400 Time (min)
525,600
175,200 350,400 Time (min)
525,600
(b)
20
0
5
0 (d)
20
20
15
15
10
5
0 (e)
0
175,200 350,400 Time (min)
525,600
10
525,600
Wind speed (m s -1)
Wind speed (m s -1)
175,200 350,400 Time (min)
5
Wind speed (m s -1)
Wind speed (m s -1)
(a)
10
10
5
0
525,600
0
(f)
Figure 4.12: Adjusted minutely wind speed data to the GSM BTS tower height of 25 m for a typical year for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively
194
Tables 4.11 and 4.12 show the simulation results for the proposed hybrid energy system for a loss of power supply probability of less than 10% (LPSP < 0.10), under the same load demand profile (for a simulation period of 1 year) for the studied locations in Nigeria. Table 4.11: Simulation results for proposed Grid-PV/Wind hybrid energy system for considered locations Optimum size Site (Zone)
Performance indices
Egrd,net (kWh/yr)
GHG emission
Szpv
Szwt
Szbb
COE
(m2)
(kW)
(kWh)
(N/kWh)
LPSP
ηrel
Energy index, Ke (kWh/N)
(t CO2eq/yr)
Sokoto (NW)
1,484.60
7.09
2
10.8
24.75
0.0002
0.9998
0.0404
0.9919
Maiduguri
3,270.62
15.05
3
10.8
31.17
0.0047
0.9953
0.0319
2.2942
4,751.20
25.45
5
21.6
56.24
0.0026
0.9974
0.0177
3.2263
5,271.66
35.85
5
21.6
40.36
0.0044
0.9956
0.0247
3.6166
3,308.24
20.76
3
10.8
32.56
0.0070
0.9930
0.0305
2.3784
5,497.89
33.84
5
21.6
47.74
0.0048
0.9952
0.0208
3.7786
38.80
0.0039
0.9961
0.0277
2.7143
(NE) Abuja (NC) Ikeja (SW) Enugu (SE) Benin City (SS) Average
3,930.70
Note: The computation is based on an exchange rate of US$1 ≈ N160.0 (Central Bank of Nigeria, accessed April 21, 2014).
195
Table 4.12: Simulation results for proposed stand-alone PV/wind hybrid energy system for studied locations Optimum size Site (Zone)
Performance indices
Szpv
Szwt
Szbb
COE
(m2)
(kW)
(kWh)
(N/kWh)
LPSP
ηrel
Energy index, Ke
GHG Emission (t CO2-eq/yr)
(kWh/N)
Sokoto (NW)
12.68
2
10.80
6.73
0.0098
0.9902
0.1471
0.1458
Maiduguri
11.98
5
10.80
10.82
0.0178
0.9822
0.0907
0.4596
41.10
5
54.00
19.26
0.0162
0.9838
0.0511
0.4193
59.50
5
54.00
21.45
0.0200
0.9800
0.0457
0.5167
23.79
5
21.60
12.82
0.0126
0.9874
0.0770
0.3244
69.14
5
54.00
22.81
0.0234
0.9766
0.0428
0.6047
15.65
0.0166
0.9834
0.0757
0.4117
(NE) Abuja (NC) Ikeja (SW) Enugu (SE) Benin City (SS) Average
Note: The computation is based on an exchange rate of US$1 ≈ N160.0 (Central Bank of Nigeria, accessed April 21, 2014)
The design specifications/capacity of power electronic components and super-capacitor bank are; Pr,sta = 3 kW, Pr,ret = 3 kW, Pr,inv = 1 kW, and Szsc = 0.0528 kWh. The specifications of components selected are as follows. Hummer WTGs of DC current with vci = 3.0 m s-1, vco = 25.0 m s-1. The capacities of the three WTGs selected are Pr1 = 2 kW, Pmax1 = 2.20 kW, Pr2 = 3 kW, Pmax2 = 3.30 kW, and Pr3 = 5 kW, Pmax3 = 5.60 kW respectively. CNSDPV 150 modules of Pmax = 150 Wp, Amod = 0.991 m2, Vdc = 24V, and TC,STC = 25 oC. USB US-250 battery with Ebat= 1.35 kWh (i.e., 225 Ah at a nominal battery voltage of 6 V). A module of 196
BM0D0165/P048BXX super-capacitor with Esc = 0.0528 kWh (i.e., 165 F SC at a nominal rated voltage of 48 V). The technical details of components are described in appendix A. The annual total energy composition of the developed grid-PV/Wind and stand-alone PV/Wind hybrid energy systems are compared in Tables 4.13 and 4.14 respectively for various cities in Nigeria. Taking Sokoto as an example, the hourly energy profile is shown in Figure 4.13. Table 4.13: The annual total energy composition of the developed Grid-PV/Wind HES Energy drawn from grid, Egrdnet
Solar energy generation, Esg
Wind energy generation, Ewg
Excess energy generation, Eexc
Total Fraction (kWh yr-1) (%)
Total (kWh yr-1)
Total (kWh yr-1)
Fraction (%)
Total Fraction (kWh yr-1) (%)
Sokoto
1,485
6.95
2,440
12.70
15,439
80.33
579
3.49
Maiduguri
3,271
13.19
4,971
22.28
14,293
64.06
2,098
12.66
Abuja
4,751
18.06
9,969
42.11
9,369
39.57
2,694
16.26
Ikeja
5,272
20.99
9,351
41.37
8,409
37.20
2,709
16.35
Enugu
3,308
13.20
5,652
25.05
13,775
61.05
1,772
10.70
Benin City
5,498
22.67
8,708
39.89
8,069
36.96
2,274
13.72
Average
3,931
15.84
6,849
30.57
11,559
53.20
2,021
12.20
Site Fraction (%)
197
Table 4.14: The annual total energy composition of the developed stand-alone PV/Wind HES Solar energy generation, Esg
Wind energy generation, Ewg
Excess energy generation, Eexc
Total (kWh yr-1)
Fraction (%)
Total (kWh yr-1)
Fraction (%)
Total (kWh yr-1)
Sokoto
4,370
21.84
15,439
77.18
0
0.00
Maiduguri
3,957
14.64
22,597
83.58
0
0.00
Abuja
16,099
62.19
9,369
36.19
4,615
27.85
Ikeja
15,520
63.56
8,409
34.44
4,091
24.69
6,477
22.59
21,840
76.16
433
2.61
Benin City
17,792
67.19
8,069
30.47
6,074
36.65
Average
10,703
42.00
14,287
56.34
2,536
15.30
Site
Enugu
198
Fraction (%)
6 5
E (kWh)
4
tot
3 2 1 0
(a)
0
1280
2560
3840
5120
6400
7680
8760
3840 5120 Time (h)
6400
7680
8760
Time (h) 6
Egrdnet (kWh)
5 4 3 2 1 0
(b)
0
1280
2560
6 5
3
sg
E (kWh)
4
2 1 0 0
(c)
1280
2560
3840 5120 Time (h)
6400
7680
8760
1280
2560
3840 5120 Time (h)
6400
7680
8760
6
4 3
wg
E (kWh)
5
2 1
(d)
0
0
Figure 4.13: Hourly energy profile for Grid-PV/Wind HES in Sokoto: (a) Total energy generation, (b) Energy drawn from grid, (c) Solar energy generation, and (d) Wind energy generation. 199
The state of charge (SOC) of the SC/battery bank of the proposed grid-PV/Wind and stand-
1.0
1.0
0.8
0.8
0.6
0.6
SOCsb
SOCsb
alone PV/Wind configurations are shown in Figures 4.14 and 4.15 respectively.
0.4
0.4 0.2
0.2 0
0
175,200
350,400
0
525,600
(b)
Time (min)
0.8
0.8
0.6
0.6
0.4
0 (c)
0
175,200
350,400
525,600
350,400
525,600
350,400
525,600
0.4
0
525,600
(d)
Time (min)
0
175,200 Time (min)
1.0 0.8
0.6
0.6
SOCsb
0.8
0.4
0.4
0.2
0.2
0
350,400
0.2
1.0
(e)
175,200 Time (min)
1.0
0.2
SOCsb
0
1.0
SOCsb
SOCsb
(a)
0
175,200
350,400
525,600
Time (min)
0 (f)
0
175,200 Time (min)
Figure 4.14: SOC of SC/battery bank of proposed grid-PV/Wind energy system for a simulation period of one year for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively.
200
1.0
0.8
0.8
0.6
0.6
SOCsb
SOCsb
1.0
0.4
0.4
0.2 0 (a)
0.2
0
175,200
350,400
0
525,600
(b)
0
1.0
0.8
0.8
0.6
0.6
0.4
0
350,400
0.2
0
175,200
350,400
0
525,600
(d)
0
175,200
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
0
175,200
350,400
350,400
525,600
350,400
525,600
Time (min)
SOCsb
SOCsb
Time (min)
(e)
525,600
0.4
0.2
(c)
175,200 Time (min)
1.0
SOCsb
SOCsb
Time (min)
525,600
Time (min)
0 (f)
0
175,200 Time (min)
Figure 4.15: SOC of SC/battery bank of proposed stand-alone PV/Wind energy system for a simulation period of one year for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively. In order to actually study the operation and control performance of the proposed HES, two typical days (February 28 and September 21) randomly chosen within the simulation period were considered for the stand-alone PV/Wind and grid-PV/Wind configurations respectively, as shown in Figures 4.16 and 4.17. For ease of illustration (considering the large simulation time steps of 525,600 min), the average hourly energy values of various sources are presented. 201
7
6
6
5
5 Energy (kWh)
Energy (kWh)
7
4 3 2
0 1,393 1,397 1,401 1,405 1,409 1,413 1,417 (a) Time (h)
2
0 1,393 1,397 1,401 1,405 1,409 1,413 1,417 (b) Time (h)
7
7
6
6
5 7
5 Energy (kWh)
Energy (kWh)
3
1
1
4 3 6 2 1 5
Energy (kWh)
7
6 Energy (kWh)
2
1
2 1
2
7
3
4 3
3
0 1,393 1,397 1,401 1,405 1,409 1,413 1,417 (d) Time (h)
6 5
4
1
0 1,393 4 1,397 1,401 1,405 1,409 1,413 1,417 (c) Time (h)
Energy (kWh)
4
0
1,393
1,397
1,401
Epv
4 3 2
1,405 1 Time (h) 0
0 1,393 1,397 1,401 1,405 1,409 1,413 1,417 (e) Time (h)
Egrdnet
5
(f)
Ew t
1,409
1,413
1,417
1,393 1,397 1,401 1,405 1,409 1,413 1,417 Time (h)
Ed
Esup
Ehs
Figure 4.16: Electrical characteristics of proposed stand-alone PV/Wind energy system for a typical day (February 28) for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively.
202
7
6
6
5
5 Energy (kWh)
Energy (kWh)
7
4 3 2
2
0 6313
7
6
6
5 7
5 Energy (kWh)
7
4 3 6 2 1
5
Energy (kWh)
6317 6321 6325 6329 Time (h)
6333
6337
(d)
6333
6337
4 3 2
0 6313
6317 6321 6325 6329 Time (h)
6333
6337
7
3
6
6
5 2
5
4 3 1 2 1 0
6317 6321 6325 6329 Time (h)
1
0 6313 4
7
Energy (kWh)
(b)
Energy (kWh)
Energy (kWh)
0 6,313 6,317 6,321 6,325 6,329 6,333 6,337 (a) Time (h)
(e)
3
1
1
(c)
4
1,393
0 6313
1,397
6317 6321 6325 6329 Time (h)
Egrdnet
1,401 6333
3 2 1
1,405 Time (h) 0 6337 (f)
Epv
4
Ew t
6313
1,409
1,413
6317 6321 6325 6329 Time (h)
Ed
Esup
1,417 6333
6337
Ehs
Figure 4.17: Electrical characteristics of proposed grid- PV/Wind HES for a typical day (September 21) for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City respectively.
203
4.2
Validation of Developed Estimation Models
There is a variation of the energy demand of the GSM BTS site within the range of 1.3 – 2.5 kW as shown in Figure 4.1. The fluctuation is an indication of the real-time variations in the operational conditions of the GSM BTS site due to varying call rates, components ageing, and period of use of auxiliary components (such as lighting and cooling systems). It is clear from the grid voltage characteristic of Figure 4.2(a) that the proposed model of Eq. (3.5) can mimic the erratic nature of the Nigerian power grid. The power outage/access characteristic shown in Figure 4.2(b) provides the total duration of annual power supply interruption of 4345 hours (with supply voltage ≤ 25 V) in Nigeria. The result shows that the grid system would experience a downtime of (4345/8760 =) 49.6%. At any time, when electrical power is available, the supply voltage is unstable. It varies within the range of 150 – 250 V as shown in Figure 4.2(b). The variation consists of 6, 741, 3645 and 23 hours of power access with supply voltage in the range of 150 – 174.99, 175 – 199.99, 200–224.99 and 225–250 volts respectively. Hence, there is electrical power access (supply) for a total duration of 4415 hours per year (with an average of 12.1 hours per day). A comparison of the simulated value of 12.1 hours (12 hours 6 minutes) of daily power access to the utility grid with the experimentally determined average value of 13 hours reported in the literature (UNDP-GEF, 2013) gives an estimation error of 6.9%, which is considered reasonable. A thorough study of the calibration results of the twelve studied solar radiation models (consisting of six global and six diffuse solar radiation models as shown in Tables 4.1 and 4.3 respectively) shows that there is a significant variation of the estimated regression coefficients (model parameters) from one location to the other. The variation of regression coefficients of studied models shows the dependence of solar radiation on site meteorology. As discussed in the literature (see section 2.6.2), R2 and t-stat indices are two important 204
statistical tools used for determining the accuracy of deduced model parameters for solar radiation estimation. For higher estimation accuracy, the computed R2-values should approach unity as closely as possible while the t-stat values should be less than the critical tstat value (1.96). Analysis of results of six global solar radiation models studied (Table 4.1) shows that methods proposed by El-Metwally (2004), Falayi et al.(2008), and the newly developed multivariable regression model have higher R2-values (0.9722 – 0.9939) compared to methods proposed by Angstrom-Prescott (1940), Badescu (1999), and Chen et al. (2004) (0.7673 – 0.9412). The results suggest that the use of more than one meteorological parameter for modelling global solar radiation could yield better estimation accuracy compared to the use of a single parameter. Conversely, a very close to unity R2-values (0.9859–0.9973) of Page and Liu-Jordan‘s model as well as the newly developed regression models of Eq. (3.23) and Eq. (3.24), shown in Table 4.3, shows a more satisfactory agreement of their diffuse solar radiation estimates with the observed values as compared to Butt et al. (2010) and Karakoti‘s estimates. Nevertheless, the t-stat values of the twelve considered models, computed at the 95% confidence level, are significantly lower than the corresponding critical t-stat (tc-stat) value. In order to determine the best models for estimating the global and diffuse solar radiations in Nigeria, five error indices, viz. Coefficient of Correlation (r), Root Mean Square Error (RMSE), Mean Bias Error (MBE) and Mean Absolute Bias Error (MABE) as shown in Figures 4.4 and 4.7 (for global and diffuse solar radiation models respectively), and the monthly Relative Percentage Error (RPE), shown in Figures 4.5 and 4.8 (for global and diffuse solar radiation models respectively), were utilized to estimate the magnitude of errors from the twelve studied models for the six geopolitical zones in Nigeria. The test of the correlation coefficient (Table 4.4) shows that the newly deduced multivariable regression 205
model of Eq. (3.24) gives the highest r-values in the range of 0.9935 – 0.9974 for the entire studied locations in Nigeria. This result suggests that the newly developed multivariable regression model provides the best diffuse solar radiation estimates for the studied sites in Nigeria. As noticed that the calculated values of error indices from the twelve studied models (Tables 4.2 and 4.4, and Figures 4.4b-d, 4.5, 4.7b-d and 4.8) vary from one place to another. This is perhaps due to seasonal variations of the solar radiation caused apparently by the degree of cloud cover, atmospheric dust, and presence of water vapour and Ozone in the atmosphere which differs from one place to another. The magnitude of estimation errors for diffuse radiation models are higher in the northern compared to the southern sites (Figure 4.8). The difference perhaps is due to the presence of dusts particles (that affects diffuse fraction), which are higher in arid sites due to the influence of the Sahara desert. As earlier mentioned, the climatic conditions change in most parts of Nigeria. The highest RMSE values for global solar radiation models of 0.6145 kWh m–2 day–1 and 1.7431 kWh m–2 day–1 as shown in Table 4.2 and Figure 4.4(b), were produced by Angstrom-Prescott (1940) and Chen et al. (2004) models in the north-western and south-western parts of the country (Nigeria) respectively, but the newly developed multivariable regression model of Eq (3.22) provided the lowest range of RMSE values (0.1273 – 0.3414 kWh m–2 day–1) throughout the studied locations. Similarly, the RMSE values for diffuse solar radiation models (Table 4.4 and Figure 4.7b), vary from 0.0243 to 0.2948 (kWh m-2 day-1), but the newly developed multivariable regression model of Eq. (3.24) produced the lowest range of values (0.0243– 0.0600 kWh m-2 day-1). Therefore, based on the RMSE index, the newly developed multivariable global and diffuse solar radiation models of Eq. (3.22) and Eq. (3.24) produced the best global and diffuse solar radiation estimates, respectively for selected areas in Nigeria. 206
Based on the MBE index, negative and positive MBE values indicate under-estimation and over-estimation respectively, of the particular term. The MBE values computed for the studied global solar radiation models, except for Chen et al. (2004), El-Metwally (2004), Falayi et al. (2008) models with considerable under-estimation/over-estimation in the southwestern region, are in the acceptable range (< ± 0.3547 kWh m-2 day-1) as shown in Table 4.2. Analysis of the MBE index for diffuse radiation models considered (Table 4.4 and Figure 4.7c) showed that the Liu-Jordan‘s estimates are lower than the observed values for arid sites (Sokoto and Maiduguri). Both Butt et al. (2010) and Karakoti et al. (2011) estimates vary between under-estimation and over-estimation for some locations in southern and northern regions. However, Page (1964) as well as the developed relations of Eq. (3.23) and Eq. (3.24) produced negligible error (under-estimation, over-estimation or both) for the considered areas. Small MBE values are desirable, but over-estimation of a particular observation may cancel under-estimation in a separate observation. The test of MABE gives a more reliable index compared to MBE. It provides information on the long-term performance of the considered models. According to the MABE index (Tables 4.2 and 4.4, and Figures 4.4d and 4.7d), the newly established multivariable regression models of Eq. (3.22) and (3.23) have the best long-term performance estimation with values in the range of 0.1075 – 0.2742 and 0.0206 – 0.0490 (kWh m-2 day-1) respectively. A comparison of the relative percentage error of the observed and estimated values of global solar radiation for the six considered models as shown in Figure 4.5, shows that the relative percentage error for each month produced by newly developed global solar radiation model of Eq. (3.22) rarely exceeds ± 10%, but others fall within ± 20% except in the south-western zone where Chen et al. (2004) model gives a significant under-estimation. A similar comparison of diffuse radiation models shown in Figure 4.8, shows that the Relative 207
Percentage Error (RPE) for each month produced by the newly developed multivariable regression model of Eq. (3.24) rarely exceeds ± 5%. The RPE values from Page‘s model and the newly developed regression model of Eq. (3.23) fall within ± 10% in southern Nigeria, but estimates from Liu-Jordan (1960), Butt et al. (2010), and Karakoti‘s model exceed ± 20%. The RPE index further confirmed the excellent agreement between the measured and predicted values of global and diffuse solar radiations produced by the newly established multivariable regression models of Eq. (3.22) and Eq. (3.24) respectively. However, it is worth mentioning that the high-performance accuracy of the newly established multivariable regression models of Eq. (3.22) and Eq. (3.24) for estimating the global and diffuse solar radiations in Nigeria could be traceable to the inclusion of cloud cover, as an additional input parameter. As earlier mentioned, cloud plays a crucial role in the transfer of energy between the surface and the atmosphere. In addition, the results justify the suggestion made by Muneer and Munawwar (2006) that the inclusion of cloud cover improves the prediction accuracy of radiation models. This study, therefore, recommends the use of the newly developed multivariable regression models of Eq. (3.22) and Eq. (3.24), based on regression coefficients presented in Tables 4.1 and 4.3, for estimating future or past values of the monthly average daily global and diffuse solar radiations respectively on a horizontal surface in Nigeria, and at other regions with similar climatic conditions. The performance validation of the proposed WTG model of Eq. (3.19) for estimating the wind energy generation was determined on the basis of six statistical indicators as shown in Table 4.10. A unity and a very close to unity (0.9999) r and R2-values of the proposed WTG model show that the proposed model has an excellent mapping accuracy in the prediction of wind energy. The negative MBE values shown in Table 4.10 suggest under-estimation but are negligible, for instance, a power error of 0.0020 kW (2 W) compared to the rated power of 208
2 kW. The proposed model‘s estimates are statistically significant since the calculated t-stat is less than critical t-stat (1.96). A comparison of the fitted power profile (using least-squares method) and the manufacturer provided WTG power characteristics shown in Figure 4.11, also validate the predictive accuracy of the proposed WTG model of Eq. (3.19). Hence, it can be used for estimating the energy supplied by the wind energy conversion system. The wind speed data adjusted to the GSM BTS tower height of 25 m for study locations (shown in Figure 4.12) is considerably greater than the cut-in wind speed for selected WTGs. This implies that the WECS would operate (generate power) for most period of the year (simulation time), but with significant variation in their energy as wind profile (Figure 4.12) varies from one place to another. In addition, the wind profile shown in Figure 4.12 indicates a greater viability of wind energy potential for Sokoto, Maiduguri and Enugu compared to Abuja, Ikeja, and Benin City. 4.3
Validation of Tilt Angles
As discussed in the literature, tilt angle and orientation are two important factors that affect the efficiency of solar surfaces. From the evaluation results, given in Tables 4.5 – 4.7, it is clear that the optimum tilt angles for harvesting solar energy depends on the geography and time of use of the photovoltaic surface. In addition, the results indicate that global solar irradiance varies with geographical locations, and it increases with increasing latitudes. A comparative analysis of results from various optimization methods presented in Table 4.8 and Figure 4.9 show that the use of a fixed annual average optimum tilt angle improves the global solar radiation in Sokoto, Maiduguri, Abuja, Ikeja, Enugu and Benin City, by 3.0 %, 2.6 %, 2.9 %, 1.5 %, 1.8%, and 1.8 %, respectively, while the use of seasonal average optimum tilt angles improves the total global solar radiation for Sokoto, Maiduguri, Abuja, Ikeja, Enugu and Benin City by 10.4 %, 9.5 %, 9.5 %, 7.0 %, 7.5 % and 7.2 % respectively. The use of the 209
monthly average tilt angle gives the highest increase of 11.3 %, 10.4 %, 10.4%, 7.9 %, 8.4 % and 8.1 % respectively. The fixed optimum tilt angle for a south-facing PV system in Sokoto, Maiduguri, Abuja, Ikeja, Enugu and Benin City are found to be 150, 140, 150, 110, 120 and 120 respectively as shown in Table 4.5. The fixed optimum tilt angle is greater than the local latitude (ϕ) of the areas selected in this study. It varies from (ϕ + 2) degrees, in the Sokoto and Maiduguri, to approximately (ϕ + 6) degrees, in the Abuja. The performance of the PV system can be optimized for considered sites, using the seasonal-based sizing techniques as shown in Table 4.6, if the PV surface is positioned horizontally (β = 00) between April and August, and inclined at optimum tilt angles of 31–37 degrees during dry season (September to March). The monthly-based optimum tilt angles, shown in Table 4.7, vary from 0 – 48 degrees across the areas considered in this study. It is worth mentioning that although the monthly-based approach yields the highest annual global solar radiation, followed by the seasonal-based method (with less than 1% energy loss, which may be ignored as shown in Table 4.8), the use of these methods (monthly and seasonal based) involves additional cost of installing a solar tracking system. However, for technical and economic reasons, considering the technology and cost of installing a solar tracking system, it may not be feasible to change the tilt angle of the PV array. Based on this assumption, the use of the fixed optimum tilt angles, shown in Table 4.5 is recommended for harvesting solar energy in Nigeria. According to Benghanem (2011), both the orientation and tilt angles have significant effects on the magnitude of the solar radiation reaching the surface of a PV array. For solar energy applications in the northern hemisphere, such as in Nigeria, the optimum orientation of the PV surface is shown to be that of due south (Ahmad and Tiwari, 2009; Calabro, 2009). As a rule of thumb, PV arrays are usually positioned at a tilt angle approximately equal to the 210
local latitude (in degrees) of the site and facing south (Mehleri et al., 2010). However, based on various theoretical models proposed by researchers, the optimum tilt angle could vary from ϕ to ϕ ± 25 degrees (Balouktsis et al., 1987; Lewis, 1987; Ulgen, 2006; Chang, 2009; Pavlovic et al., 2010, Oloketuyi et al., 2013). The positive and negative signs represent the northern and southern hemisphere respectively. Qiu and Riffat (2003) suggested that the tilt angle of a PV surface set within the optimum tilt angle of ± 100 is an acceptable practice. Therefore the recommended fixed optimum tilt angles deduced from this study using the HDKR model given by Eq. (2.53), which varies from ϕ + 2 degrees to approximately ϕ + 6 degrees (see Table 4.5) for harnessing optimum solar energy for selected locations in Nigeria is considered reasonable. Figure 4.10 shows the time variation of irradiance for each minute of the year. It is pertinent to note that the solar radiation reaching the earth‘s surface follows an oblique path length in the early morning and the late afternoon. The result of this oblique incidence through the atmosphere is a greater atmospheric attenuation and lesser intensity of solar radiation (Ahmad, 1989). 4.4
Performance Evaluation of Proposed Hybrid Energy System for GSM BTS Sites
A study of results presented in Tables 4.11 and 4.12 show the variation of results of the HES from one part of the country to the other, indicating the dependence of renewable sources on climatic conditions. The variation justifies the need for determining the optimum design specification of the renewable energy sources for any given location as no two locations could perform optimally under same design specification. However, it is worth mentioning that the fixed design specifications (capacity allocation) for the power electronic components (i.e., stabilizer, inverter and rectifier) is as a result of the same power profile of the GSM S/4/4/4 out-door base station site utilized for all areas for ease of comparison. In addition, it is 211
intended in this study that the SC bank responds to the transient power requirement of the GSM BTS sites. As a result a fixed total storage capacity of 0.0528 kWh, which can responds to the peak load demand is chosen. The electrical properties of the grid-PV/Wind system are compared in Table 4.13. The energy characteristics vary for different sites (cities). The largest proportion of energy is from wind. The highest wind energy input occurs in Sokoto (15,439 kWh per year and accounts for 80.33% of the total energy drawn by the GSM BTS site). The increase in wind power generation leads to the least energy drawn from the grid of 1,485.0 kWh per year in Sokoto. In contrast, Benin City has the least wind electricity contribution (8,069.0 kWh per year and accounts for 36.96% of the total energy drawn by the GSM BTS site). Consequently, grid electricity (5,498.0 kWh per year) is mostly purchased in Benin City. The amount of energy drawn from the grid tends to decrease from the southern (Benin City) to northern (Sokoto) part of Nigeria with a corresponding increase in the renewable energy contribution. The decrease is obvious owing to the dependence of renewable resources on climatic conditions. As observed, the energy drawn from the grid to satisfy the power demand of the GSM BTS sites varies from an average of 6.95 % (in Sokoto) to 22.67 % (in Benin City). It implies that, the renewable energy contributes an average of 76.85 % (Benin City) to 93.03 % (Sokoto). The larger percentage contribution of renewable sources throughout the studied locations for the grid-PV/Wind system is a reflection of the vast resource availability, and the techno-economic viability of the renewable resources for a sustainable electricity supply to mobile telecommunication sites in Nigeria. The wind energy generation of the PV/Wind system, shown in Table 4.14, is mainly limited in the southern part of the country by the sharp decrease in the annual average wind speed. In 212
other words, as the wind energy potential tends to decrease from the northern (Sokoto) to the southern (Benin City) part of Nigeria, a greater percentage of solar energy is required to reliably satisfy the energy demand. The increase in the observed wind energy in Enugu (south-eastern zone) is an indication of the impact of the north-east trade wind that blows from Sahara desert. Analysis of results of the state of charge of energy storage system throughout the simulation period, as shown in Figures 4.14 and 4.15 for grid-connected and stand-alone configurations respectively, shows that the proposed energy management strategy constrains the SOC of the SC/battery at any time to fall within the defined minimum and maximum acceptable limits of 40% and 98% respectively. The imposed constraint ensures a longer lifetime for the SC/battery banks as it can prevent over-discharging and over-charging of the storage system. A study of the operation and control performance of the proposed HES shows that during the early and late hours of the day when solar energy is unavailable, the stand-alone PV/Wind system‘s energy production (as shown in Figure 4.16) is considerably lower than the energy demand in most areas. During these periods, the energy storage provides the deficit energy needed to balance the energy demand requirement. At about noon, i.e., between 7 and 17:00 hours of the day when solar energy is available the hybrid system‘s energy production is considerably higher than the energy demand throughout the studied locations. The surplus energy is stored to be used in periods of short supply. On the other hand, the operation of the grid-PV/Wind system (see Figure 4.17) shows that the grid electricity supply is erratic. During periods when the renewable energy production is less than the energy demand, the net (deficit) energy generated is drawn from the grid if available; else, the energy stored in the SC/battery bank can be used during periods of short supply. These results validate the reliable operation performance of the proposed HES in Nigeria. 213
4.4.1 Comparison of Proposed Energy System and Existing Energy Systems This study considered four performance indices to determine the applicability of the proposed power generation system over existing traditional or conventional power generation systems. The performance indices utilized are the system power supply reliability, economy (cost of energy), viability, and the emissions reduction target. Power Supply Reliability Figure 4.18 shows a comparison of the power supply reliability of the conventional and proposed hybrid energy systems. As noticed, the power supply reliability of the different energy systems varies from the worst case scenario of 50.4% (for the grid-only or utility grid system) to the best case of about 99.99% that can be realized from both grid-diesel backup and diesel-only energy systems. Powering GSM BTS sites with the utility grid would imply that the BTS would have a downtime, as a result of the unavailability of power supply, for an average of 11.9 hours daily (see Figure 4.2b). The high rate of the grid power unavailability suggests why the mobile telecommunication sector resorts to the use of a diesel generator power system, either as a backup to the utility grid or exclusively in remote locations.
214
Power supply reliability
1.0 0.8 0.6 0.4 0.2 0
Sokoto
Maiduguri
Abuja
Ikeja
Enugu
Benin City
Locations
Diesel-only
Grid-Diesel
Grid-only
Grid-PV/Wind
PV/Wind
Figure 4.18: Comparison of the power supply reliability of proposed hybrid and conventional energy systems The power supply reliability of proposed hybrid energy systems varies from 97.66 – 99.98 %, an improvement of 93.77 – 98.37 % of the power supply reliability of the utility grid supply system. The proposed hybrid energy system can attain 99.999% power supply reliability, but with increased COE. As a result, the best compromise between system cost and reliability is determined, through purchased energy to account for supply losses if any, in order to establish the optimum design specification. Based on the optimum design specifications, the loss of power supply, which are purchased from a standby generating system, are negligible (0.02 – 0.69 %) as shown in Table 4.15 for grid-PV/Wind energy system while those of the hybrid green (stand-alone PV/Wind) energy system falls within acceptable limits (0.98 – 2.34 %) as shown in Table 4.16.
215
Table 4.15: Comparison of Energy Composition of Grid-PV/Wind HES by fraction for different sites Site
Renewable fraction (%)
Deficit fraction* (%)
6.95
93.03
0.02
Maiduguri
13.19
86.34
0.47
Abuja
18.06
81.67
0.26
Ikeja
20.99
78.57
0.44
Enugu
13.20
86.11
0.69
Benin City
22.67
76.85
0.48
Average
15.84
83.76
0.39
Sokoto
Grid fraction (%)
Table 4.16: Comparison of Energy Composition of stand-alone PV/Wind HES by fraction for different sites Site
Grid fraction (%)
Renewable fraction (%)
Deficit fraction* (%)
Sokoto
21.84
77.18
0.98
Maiduguri
14.64
83.58
1.78
Abuja
62.19
36.19
1.62
Ikeja
63.56
34.44
2.00
Enugu
22.59
76.16
1.26
Benin City
67.19
30.47
2.34
Average
42.00
56.34
1.66
* Note: The deficit fraction shows the pecent of energy demand to be purchased from the standby diesel generating plant in order to guarantee power supply reliability in the order of nine (≥ 99.99%)
It is worthy of note that although the proposed hybrid energy system supply reliability varies from 97.66 – 99.98%, the opportunity cost for the system downtime (in supplying deficit power) is accounted for in the proposed system COE by the inclusion of cost penalty 216
equivalent to COE of diesel-only system. In other words, the proposed HES can attain supply reliability of 99.999% by generating 97.66 –99.98 % of energy demand and purchasing 0.21 – 2.34 % of energy demand from the diesel-only conventional system (as shown in Tables 4.15 and 4.16). System Economic Performance A comparison of the economic performance of the conventional and proposed HES deduced based on objective function of Eq. (3.120) is shown in Table 4.17 and Figure 4.19. Table 4.17: Comparison of the Costs of Energy (COE) of various energy systems Cost of Energy Options (N kWh-1)
Site (Zone)
Diesel-Only
Grid-Diesel
Grid-Only
Grid-PV/Wind
PV/Wind
Sokoto (NW)
100.15
80.82
60.79
24.75
6.73
Maiduguri (NE)
99.92
78.55
56.45
31.17
10.82
Abuja (NC)
98.32
97.89
97.09
56.24
19.26
Ikeja (SW)
96.95
78.70
59.77
40.36
21.45
Enugu (SE)
99.69
79.50
58.60
32.56
12.82
Benin City (SS)
97.63
85.89
73.59
47.74
22.81
Average
98.78
83.56
67.71
38.80
15.65
217
COE (Naira per kWh)
100 80 60 40 20 0
Sokoto Diesel-only
Maiduguri
Abuja Ikeja Locations
Grid-Diesel
Grid-only
Enugu Grid-PV/Wind
Benin City PV/Wind
Figure 4.19: Comparison of the economic performance of proposed hybrid and conventional energy systems The variation in the cost of energy (shown in Table 4.17 and Figure 4.19) production by the conventional (diesel-only, grid-diesel backup and grid-only) power systems results mainly from the cost variation of the diesel (fuel) consumed owing to additional cost of transportation as well as the variation of grid electricity prices, which varies from one part of the country to another. The larger percentage of renewable contribution within selected areas in Nigeria has a significant improvement in the overall system cost of energy supplied by the proposed system. For grid-connected configurations, it is observed that the proposed gridPV/wind system gives the lowest COE (N 24.75 – N 56.24 per kWh; given an exchange rate1 of US$ 1 to ₦ 160.0) with an average value of N 38.80 per kWh while the grid-diesel backup energy system gives the highest cost of energy (N 78.55 – N 97.89 per kWh) with an average of N 83.56 per kWh for considered areas. Consequently, the use of the proposed grid-
1
Central Bank of Nigeria (CBN): accessed April 21, 2014.
218
PV/wind system could reduce the COE of grid-diesel backup energy systems by 53.57 %. For stand-alone configurations, the average COE of the proposed hybrid PV/Wind energy system (N 15.65 per kWh) is significantly lower compared to the diesel-only power system (N 98.79 per kWh). This figure amounts to an average reduction of 84.16% in the COE for off-grid GSM BTS sites. In terms of grid versus off-grid (stand-alone) application, the proposed hybrid green (PV/Wind) system provides the best overall reduction in cost of 84.16% (compared to traditional diesel-only), 81.27 % (compared to traditional grid-diesel backup) and 59.66 % (compared to proposed grid-PV/Wind) energy systems. System Performance Viability It is observed from this study that an increase in the power supply reliability of energy production system could lead to a corresponding increase in the COE. For instance, an increase in the power supply reliability of the utility grid in Sokoto from 50.4 – 99.999 %, by incorporating diesel generating system as backup, results in an increase in the COE from N 60.79 to N 80.82 per kWh. In other words, the objective of enhancing the power supply reliability of energy production system tends to conflict with that of cost minimization. To ensure the best compromise between reliability and cost, this study proposes a novel key performance index (Energy index, Ke). It is an expression of the performance viability of the energy production system as it gives the energy throughput of the system. The energy throughput of the system indicates the amount of energy a system can produce per unit investment. The higher the energy throughput of an energy system the better the system is (in terms of reduced cost and increased reliability). A comparison of the performance viability (Energy index) of the conventional and proposed energy systems is shown in Figure 4.20. A thorough study of the Energy index shows that the 219
energy throughputs of the conventional generation systems are significantly lower than proposed HES.
Ke (kWh per Naira)
0.15 0.12 0.09 0.06 0.03 0
Sokoto Diesel-only
Maiduguri
Abuja Ikeja Locations
Grid-Diesel
Grid-only
Enugu Grid-PV/Wind
Benin City PV/Wind
Figure 4.20: Comparison of the performance viability of proposed and conventional energy systems The conventional (diesel-only, grid-diesel, and grid-only) energy systems have an average of 0.0101, 0.0120 and 0.0076 kWh per Naira respectively. The grid-PV/Wind provides a better energy throughput of 0.0177 – 0.0404 kWh per Naira with an average of 0.0277 kWh per Naira, while the stand-alone (PV/Wind or hybrid green) energy system gives the overall best energy throughput in the range of 0.0428 – 0.1471 kWh per Naira with an average of 0.0758 kWh per Naira. By a comparison with the current practice of using diesel generator as backup to the utility grid, the use of the proposed hybrid system can give an average performance improvement in the range of 130 – 530 % per unit investment in the GSM BTS sites within the studied locations. The larger energy throughput of the proposed system compared to the existing technology is a justification of the techno-economic viability of the wind and solar resources for a sustainable electricity supply to mobile telecommunication sites in the country. 220
To further demonstrate the techno-economic viability of the green energy technology and the significance of the energy performance index (Ke) proposed for the evaluation of energy generation system, sensitivity analysis was performed by varying the loss of power supply probability for proposed hybrid energy system. Table 4.18 shows the comparison of proposed system techno-economy for the loss of power supply not exceeding 10 %.
221
Table 4.18: Comparison of proposed system techno-economic viability for various reliability limits Studied locations
LPSP
ηrel
COE (Naira per kWh)
Ke (kWh per Naira)
0.0930
0.9071
22.74
0.0399
0.0856
0.9144
21.08
0.0434
0.0724
0.9276
20.21
0.0459
0.0669
0.9331
19.87
0.0470
0.0536
0.9464
19.04
0.0497
0.0396
0.9604
9.54
0.1007
0.0231
0.9769
7.95
0.1229
0.0131
0.9869
7.13
0.1384
0.0098
0.9902
6.73
0.1471
0.0000
1.0000
8.88
0.1126
0.0932
0.9068
14.89
0.0609
0.0876
0.9124
14.61
0.0625
0.0780
0.9220
14.14
0.0652
0.0656
0.9344
12.70
0.0736
0.0522
0.9478
12.17
0.0779
0.0433
0.9567
11.86
0.0807
0.0379
0.9621
11.70
0.0822
0.0212
0.9788
11.15
0.0878
0.0178
0.9822
10.82
0.0908
0.0000
1.0000
12.31
0.0812
0.0849
0.9151
24.36
0.0376
0.0741
0.9259
22.28
0.0416
0.0652
0.9348
22.25
0.0420
0.0559
0.9441
21.40
0.0441
0.0451
0.9549
20.83
0.0458
0.0351
0.9649
20.34
0.0474
0.0266
0.9734
20.12
0.0484
0.0187
0.9813
19.53
0.0502
0.0162
0.9838
19.26
0.0511
0.0000
1.0000
21.66
0.0462
Sokoto
Maiduguri
Abuja
Note: The computation is based on an exchange rate of US$1 ≈ N160.0 (Central Bank of Nigeria, accessed April 21, 2014)
222
Table 4.18: Continuation Studied locations
LPSP
ηrel
COE (Naira per kWh)
Ke (kWh per Naira)
0.0808
0.9192
25.92
0.0355
0.0734
0.9266
25.47
0.0364
0.0668
0.9332
25.07
0.0372
0.0514
0.9486
23.99
0.0395
0.0456
0.9544
23.42
0.0408
0.0370
0.9630
22.67
0.0425
0.0223
0.9777
21.60
0.0453
0.0200
0.9800
21.45
0.0457
0.0077
0.9923
22.37
0.0444
0.0000
1.0000
28.00
0.0357
0.0887
0.9113
15.68
0.0581
0.0744
0.9256
15.04
0.0615
0.0639
0.9361
14.64
0.0639
0.0561
0.9439
13.54
0.0697
0.0473
0.9527
13.40
0.0711
0.0375
0.9625
13.52
0.0712
0.0238
0.9762
13.13
0.0743
0.0126
0.9874
12.82
0.0770
0.0078
0.9922
14.94
0.0664
0.0000
1.0000
16.09
0.0622
0.0940
0.9060
27.21
0.0333
0.0775
0.9225
25.65
0.0360
0.0644
0.9356
24.96
0.0375
0.0539
0.9461
24.46
0.0387
0.0436
0.9564
24.04
0.0398
0.0340
0.9660
23.36
0.0414
0.0234
0.9766
22.81
0.0428
0.0126
0.9874
22.84
0.0432
0.0093
0.9907
23.49
0.0422
0.0000
1.0000
30.12
0.0332
Ikeja
Enugu
Benin City
Note: The computation is based on an exchange rate of US$1 ≈ N160.0 (Central Bank of Nigeria, accessed April 21, 2014)
223
As observed, the cost of energy tends to decrease with an increase in power supply reliability of the energy system up to a certain (optimum) point, which shows the best compromise between system cost and reliability. Beyond the optimum point, an increase in power supply reliability can result in a corresponding increase in the cost of the system. The variation of energy performance index with power supply reliability is shown in Figure 4.21. Given a desired tolerance of 10% unavailability for a typical application, the highest ke-values suggest the best compromise between the cost of energy and power supply reliability. An energy system designed to operate at this optimum (break-away) point yields the best technoeconomic performance. Based on the ke-values shown in Table 4.18, a system designed with a downtime (LPSP) of 0.0098 (in Sokoto), 0.0178 (in Maiduguri), 0.0162 (in Abuja), 0.0200 (in Ikeja), 0.0126 (in Enugu) and a downtime of 0.0234 (in Benin City) gives the overall best improvement of the system energy throughput of 0.1471, 0.0907, 0.0511, 0.0457, 0.0770, and 0.0428 kWh per Naira investment respectively, compared to systems designed never to go down. These results agree with the proposed optimum design specifications given in Table 4.12 for the considered locations in Nigeria.
224
0.15 0.15
0.12 Ke (kWh per Naira)
Ke (kWh per Naira)
0.12 0.09 0.06
0 0.90
0.94 0.96 Reliability
0.98
1.00
(b)
0 0.90
0.15
0.15
0.12
0.12 Ke (kWh per Naira)
Ke (kWh per Naira)
0.92
0.09 0.06 0.03
(c)
0 0.90
0.94 0.96 Reliability
0.98
0.92
0.94 0.96 Reliability
0.98
0.92
0.94 0.96 Reliability
1.00
0.09 0.06
0.92
0.94 0.96 Reliability
0.98
1.00
(d)
0 0.90
0.15
0.15
0.12
0.12
0.09 0.06
1.00
0.09 0.06 0.03
0.03
(e)
0.92
0.03
Ke (kWh per Naira)
Ke (kWh per Naira)
0.06 0.03
0.03
(a)
0.09
0 0.9
0.92
0.94 0.96 Reliability
0.98
1.00
(f)
0 0.90
0.98
1.00
Figure 4.21: Variation of energy performance index with power supply reliability for (a) Sokoto, (b) Maiduguri, (c) Abuja, (d) Ikeja, (e) Enugu, and (f) Benin City.
225
The average energy throughput of 0.076 kWh per Naira realized from the proposed PV/Wind system is equivalent to an annual cost of energy drawn of N 4,360,789,473.68 (with an average of N 218,039.47 per site) required for running about 20,000 GSM BTS sites in Nigeria. On the other hand, the average energy throughput (0.012 kWh per Naira investment) of the conventional grid-diesel energy system is equivalent to an annual cost of energy drawn of N 27,618,333,333.33. With the investment of the gains in infrastructural development, an additional increase of 6% in the annual infrastructural growth rate could be achieved. This corresponds to an annual additional profit of N 23,257,500,000.00, and N 361,556,000,000.0 by 2020 in Nigerian mobile telecommunications industry owing to improvement in the proposed PV/Wind system compared to the current scenario of the grid-diesel energy system. The cost estimation is based on the estimated annual energy consumption (16,571 kWh per annum) of a typical GSM S/4/4/4 out-door BTS site in Nigeria, and at an annual population growth rate of 2.5%. Given a projected number of about 80,000 GSM BTS sites in Nigeria by 2020, the corresponding number of subscribers per cell site should be in the neighbourhood of 2500 as this figure is expected to improve the quality of service delivery of Nigerian mobile telecommunications industry based on global best practice. The projection is in line with the country‘s target (Vision 20:2020) of becoming one of the twenty world‘s leading economy in the year 2020. With an overall gain of estimated ₦ 361.6 billion, the Nigerian mobile telecommunications industry would be rightly positioned towards fulfilling her social obligations to the host communities. A comparison of the growth in cost saving and population till year 2020 is shown in Figure 4.22.
226
100
210
80
195
60
180
40
165
20
150
2012
2013
2014
2015
2016
Population
2017
2018
2019
2020
Cost saving (in Billion Naira)
Population (in million)
225
0
Cost saving
Figure 4.22: Comparison of the growth in cost saving and population till 2020 4.4.2 System Environmental Impact Table 4.19 shows the results of the environmental impact assessment of the conventional and the proposed HES. It is observed that the diesel-only energy system has the highest annual GHG emissions of 25.83 tCO2-eq compared to the lower emissions of 0.9919 – 3.7786 tCO2-eq from the grid-PV/Wind system, while emissions from stand-alone PV/Wind energy systems are negligible (0.2535 – 0.6047 tCO2-eq). It should be noted that the pollutant emissions from the stand-alone PV/Wind system are due to purchased energy from the diesel generation system during periods of deficit supply in order to guarantee over 99.999% supply reliability.
227
Table 4.19: Comparison of environmental impact of conventional and proposed energy systems Annual Carbon credit
Emissions Studied locations
Sokoto
Maiduguri
Abuja
Ikeja
Enugu
Benin City
Model
Emission penalty (Naira/yr)
CO2 (tCO2 /yr)
CH4 (tCH4/yr)
N2O
Diesel-only
14.8642
0.0331
0.0331
25.8342
0
Grid-diesel
12.9217
0.0168
0.0171
18.4886
7.346
28.434
109,452.22
GridPV/Wind
0.9835
0.0000
0.0001
0.9919
24.842
96.161
5,871.97
PV/Wind
0.1458
0.0003
0.0003
0.2535
25.581
99.019
1,500.46
Diesel-only
14.8642
0.0331
0.0331
25.8342
0
Grid-diesel
12.9217
0.0168
0.0171
18.4886
7.346
28.434
109,452.22
GridPV/Wind
2.2295
0.0002
0.0003
2.2942
23.540
91.120
13,581.55
PV/Wind
0.2644
0.0006
0.0006
0.4596
25.375
98.221
2,720.81
Diesel-only
14.8642
0.0331
0.0331
25.8342
0
0
152,938.46
Grid-diesel
12.9217
0.0168
0.0171
18.4886
7.346
28.434
109,452.22
GridPV/Wind
3.1775
0.0001
0.0003
3.2263
22.608
87.511
19,099.73
PV/Wind
0.2412
0.0005
0.0005
0.4193
25.415
98.377
2,481.98
Diesel-only
14.8642
0.0331
0.0331
25.8342
0
0
152,938.46
Grid-diesel
12.9217
0.0168
0.0171
18.4886
7.346
28.434
109,452.22
GridPV/Wind
3.5468
0.0002
0.0004
3.6166
22.218
86.001
21,410.20
PV/Wind
0.2973
0.0007
0.0007
0.5167
25.317
98.000
3,058.94
Diesel-only
14.8642
0.0331
0.0331
25.8342
0
0
152,938.46
Grid-diesel
12.9217
0.0168
0.0171
18.4886
7.346
28.434
109,452.22
GridPV/Wind
2.2884
0.0003
0.0004
2.3784
23.456
90.794
14,080.00
PV/Wind
0.1867
0.0004
0.0004
0.3244
25.510
98.744
1,920.49
Diesel-only
14.8642
0.0331
0.0331
25.8342
0
0
152,938.46
Grid-diesel
12.9217
0.0168
0.0171
18.4886
7.346
28.434
109,452.22
GridPV/Wind
3.7029
0.0002
0.0004
3.7786
22.056
85.374
22,369.07
PV/Wind
0.3479
0.0008
0.0008
0.6047
25.230
97.659
3,579.62
(tN2O/yr)
228
GHG (t CO2eq/yr)
(t CO2eq /yr)
%/yr
0
0
152,938.46
152,938.46
In terms of GHG emission reduction, the diesel-only powered system has zero annual carbon credits (carbon reduction), grid-diesel system has an annual average credit of 7.35 tCO2-eq (28.43 % carbon reduction), grid-PV/Wind system has an average of 23.12 tCO2-eq (89.49 % carbon reduction), while stand-alone PV/Wind system has the highest annual reduction of 25.40 tCO2-eq (98.34 % carbon reduction). In other words, the use of the proposed hybrid green or stand-alone PV/Wind system is expected to reduce the annual GHG emissions from GSM BTS sites in Nigeria by 98.34 % (equivalent to 508,093.33 tCO2-eq yr–1) thereby making the environment much more friendly and safe. It is worth mentioning that although photovoltaic conversion systems are considered clean, these systems have secondary emission of pollutant gases, since fossil fuels are used in the production of the Silicon cells that constitute the photovoltaic conversion systems. 4.4.3 Land Requirement for Implementing the Proposed Hybrid Energy System for GSM BTS Sites The maximum landmass (area) required for implementing the PV array is 69.14 m2, which is less than one-sixth (1/6) of the total landmass of a plot of land measuring 462.10 m2. The result shows that even with one-fourth (1/4) of a plot of land (115.53 m2), the implementation of proposed stand-alone hybrid energy system is feasible. However, for existing GSM BTS sites where the landmass is about one-sixth of a plot of land (77.02 m2), the use of standalone option may not be feasible, unless there is possibility of site expansion or extension. Where site expansion or extension is possible, the cost of additional land requirement (not considered here) would reduce the range of economic or cost benefits of the proposed PV/Wind hybrid system. As a result, the availability of necessary landmass is a determining factor for selecting the proposed HES mode of operation.
229
4.5
Findings
The following findings are drawn from the discussion of the results presented in the foregoing sections as well as from the experience gained during the course of present research. 1. The inclusion of cloud cover as part of input parameters for estimating solar radiation improves the predictive accuracy of solar radiation models for Nigerian terrain. Cloud plays a crucial role in the transfer of energy between the surface and the atmosphere. The results also justify the suggestion made by Muneer and Munawwar (2006) that the inclusion of cloud cover could improve the prediction accuracy of solar radiation models. 2. The simulation of the dynamic behaviour of different energy sources, which constitute the proposed HES shows that the Nigerian power grid system can only deliver electric power for an average of 12.1 hours (12 hours, 6 minutes) daily. At any time, when electrical power is available, the supply voltage is unstable. It fluctuates between 150 V and 250 V, but for over 80% of the time (power supply periods) the variation of supply voltage is within 200 – 225 V. The viability of wind energy conversion system is higher in Sokoto, Maiduguri and Enugu compared to Abuja, Ikeja, and Benin City. The optimum tilt angles for harvesting solar energy depends on the geography and time of use of the photovoltaic surface. When the PV array is oriented due south, the fixed optimum tilt angles for harvesting solar energy in Nigeria are found to vary from 11 – 15 degrees (specifically, 11o for Ikeja, 12o for Benin City and Enugu, 14o for Maiduguri and 15o for Abuja and Sokoto). The deduced tilt angles are greater than the local latitudes (ϕ) of selected areas in Nigeria. If it is possible to use a solar tracking 230
system for adjusting the orientation of the PV array (considering the cost and technology for the installation), optimum tilt angles would change from 0 – 48o across the locations considered in this study. 3. The power supply reliability and Cost of Energy (COE) are two important parameters that have a significant impact on the energy throughput of the system. The proposed photovoltaic–wind hybrid system designed with a downtime (loss of power supply probability) of 0.0098 (in Sokoto), 0.0178 (in Maiduguri), 0.0162 (in Abuja), 0.0200 (in Ikeja), 0.0126 (in Enugu) and 0.0234 (in Benin City) gives the overall best improvement of the system energy throughput of 0.1471, 0.0907, 0.0511, 0.0457, 0.0770, and 0.0428 kWh per Naira investment respectively, compared to systems designed never to go down. 4. Solar radiation increases with increasing latitudes while higher wind speed was observed in arid locations (extreme north). 5. The simulation results from the twelve solar radiation models studied are statistically significant at the 95% confidence level, but their predictive ability varies from one site to another. The highest error found in the southern part of Nigeria was produced by the Chen et al. (2004) model, with monthly relative percentage error exceeding ± 30%. The proposed multivariable regression model described by Eq. (3.22) has the overall best accuracy for estimating the daily global solar radiation on a horizontal surface. It has the least RMSE and MABE of 0.1237–0.3414 and 0.1075–0.2742 kWh m–2 day–1 respectively, and the least monthly RPE of ± 10% for the study sites. The proposed multivariable regression model defined by Eq. (3.24) has the overall best accuracy for estimating the daily diffuse radiation on a horizontal surface. It has the least RMSE and MABE of 231
0.0234–0.0600 and 0.0206–0.0490 kWh m-2 day-1 respectively and the least monthly RPE of ± 5% for the study sites. 6. Simulation results for the proposed HES show that the optimum configuration depends on the operation mode and the geographical location. The grid-connected (Grid/PV/Wind) configuration has an optimum size of between 2 and 5 kW wind turbines, 1.07 – 5.43 kW (7.09 – 35.85 m2) PV array inclined at 11 – 15 degrees and 0.0528 kWh super-capacitor and 10.8 – 21.6 kWh battery bank in addition to 1,484.60 – 5,497.89 kWh of energy drawn from the grid per annum. The use of the PV/Wind system reduces the COE of Grid/PV/Wind system by 59.66% (from N 38.80 to N 15.65 per kWh) but with larger size of PV array and battery banks in the range of 11.98 – 69.14 m2 and 10.8 – 54.0 kWh respectively. Sokoto, located in north-western Nigeria, is the most favourable of all locations considered for utilizing both the Grid/PV/Wind and PV/Wind generation systems. On the other hand, Abuja (located in north-central region of Nigeria) and Benin City (located in south-south zone of Nigeria) are the least favourable sites for implementing the Grid/PV/Wind and stand-alone PV/Wind energy systems respectively. 7. The utility grid system has the worst case power supply reliability scenarios of 50.4% in Nigeria. Powering mobile telecommunication sites with the utility grid would imply that the GSM BTS would have a downtime, as a result of unavailability of power supply, for an average of 11.9 hours (11 hours, 54 minutes) daily. The use of diesel generator power systems as backup to the electricity grid system can improve the power supply reliability to over 99.99%, but with an increase (> 20 %) in the COE and GHG emissions into the environment per kWh energy drawn by the GSM BTS site. The application of the grid-connected and stand-alone HES for mobile telecommunication sites can 232
improve the quality of GSM mobile services in Nigeria, by making the operation of base stations sites more reliable and cost effective (cheaper). Results showed that depending on the mode of operation, the cost of the conventional Grid/Diesel energy generation per kWh for powering mobile telecommunication sites can reduce by between 53.57 % (for Grid/PV/Wind) and 81.27 % (for PV/Wind) without compromising the power supply reliability. The PV/Wind system has the highest carbon reduction credits of 98.34%, which corresponds to 508,903.33 tCO2-eq yr–1. The lower the GHG emissions (i.e., the ‗greener‘ the energy solution), the cleaner is the environment. The objective of cost minimization does not conflict with that of emission reduction as it is accepted by Pelet et al. (2005), Bernal-Agustin et al. (2006), Dufo-Lopez and Bernal-Agustin (2008), Zhou et al. (2010) and other authors discussed in the literature. The simulation results showed that the proposed HES if deployed in Nigeria would reduce both the cost of energy and greenhouse gas emissions. The deviation from the general norm is perhaps due to the relatively expensive grid electricity as compared to renewable sources and the high unreliability index that characterized grid electricity system in Nigeria. It is worthy of mention that the cost of the renewable energy has significantly reduced over the years due to improvement in renewable technologies. The stand-alone PV/Wind system gives the best performance index, with an average 0.0758 kWh per Naira. This corresponds to an additional annual economic benefit of ₦ 23.26 billion, and ₦ 361.56 billion by year 2020 in Nigerian mobile telecommunications sector, given an estimated annual energy consumption (16,571 kWh yr-1) for a typical GSM BTS site on the assumption of an additional average increase of 6% in infrastructural development and a population growth rate of 2.5% in Nigeria. 233
4.6
Contributions to Knowledge
The study has contributed to knowledge in the following ways:
1. Empirical models for predicting the monthly average daily global and diffuse solar radiations on a horizontal surface for Nigeria were developed. 2. A stochastic model, which accounts for uncertainties such as unreliability and power outages characterizing the grid electricity supply in Nigeria, was developed. 3. A novel performance index (Energy index) for determining the techno-economic viability of energy generating system was developed. 4. An optimum and operation control model for both grid-connected and stand-alone hybrid energy systems that can efficiently replace the existing conventional fossil fuelled generators for cheaper and cleaner production of electricity with lower annual emission targets was developed.
234
CHAPTER FIVE 5.0 5.1
CONCLUSION AND RECOMMENDATIONS Conclusion
The rapid growth of the mobile telecommunication sector of many emerging cities creates a number of problems such as network congestion and poor quality of service delivery. The mobile telecommunication operators are unable to expand their networks fast enough to meet the ever growing demand by subscribers in most parts of the emerging world. The reason is due to lack of a reliable utility grid and the cost implication of a supplementary energy source. In addition, the use of fossil-powered solution has a number of economic, logistic, and environmental problems. Access to clean and stable power supply is essential in actualizing Nigeria‘s quest for joining the league of the twenty most industrious nations by the year 2020 (vision 20:2020). A thorough review of literatures showed that although the hybrid energy systems are capable of providing the needed energy for sustainable economic development in the mobile telecommunication sectors, critical issues such as reliability and the enabling technologies must be resolved. In addition, potential investors should be aware of the optimum design configuration, capacity projections and the techno-economic implications of the renewable technology as this can enable the investor decide on the most suitable technology or system. This study developed of an optimum and operation control model for both grid-connected and stand-alone HES that will actually replace the existing conventional fossil fuelled generators for cheaper and cleaner power supply to GSM BTS sites in emerging cities. The method adopted is in line with global best practices in mitigating the devastating effects of global warming on the earth. The developed model accounts for uncertainties such as power 235
outages and unreliability, which characterizes grid electricity in developing nations, by incorporating a probability-based prediction method according to the uniform distribution function. The practical ability of the genetic algorithms to solving continuous variables and non-linear objective and constraint functions commonly found in power systems without requiring gradient information informed the choice of a h-GAPS based optimization method. The deduced model was calibrated and validated, using 22-years meteorological data sets collected from the Nigerian Meteorological agency, Oshodi and National Aeronautic and Space Administration for 37 locations across Nigeria. The results showed that the stand-alone PV/Wind hybrid system should be utilized for existing and new GSM BTS sites across Nigeria. However, where the available landmass is about 77.02 m2 and the possibility of site expansion, or extension is not economically feasible, the: 1. Stand-alone PV/Wind hybrid system should be utilized for existing and new GSM BTS sites in arid areas (Sokoto and Maiduguri), while 2. Grid-PV/Wind hybrid system should be utilized for existing GSM BTS sites in other locations (Abuja, Benin City, Enugu, and Ikeja). 3. If the application of either proposed stand-alone or grid-connected hybrid energy system is not economically feasible for any location due to additional land constraint, then the diesel generator should be integrated as a part of the hybrid energy system. However, the total energy drawn from a diesel generator should be reduced to acceptable limits due to environmental hazards resulting from fossil combustion. The concept of the energy throughput of the system as a Key Performance Index (KPI) for determining the overall performance characteristic of a power system was proposed. The KPI is an expression of the techno-economic viability of an energy system measured in terms of 236
per unit cost of energy supplied. The developed stand-alone PV/Wind system gives an average improvement of energy throughput of 0.0758 kWh per Naira as compared to existing energy systems, such as utility grid (0.0076 kWh per Naira) and the grid-diesel system (0.0120 kWh per Naira). The results showed the excellent performance of the developed PV/Wind over the existing systems, yet with considerable excess capacity generation (an average of 15.30 % of energy demand) that could improve the lives of the host community. Moreover, the use of renewable (solar and wind) sources reduces the dependence on the utility grid, reduces load shedding and thereby improving power supply reliability. In conclusion, this study would: 1. improve the quality of GSM mobile services in Nigeria, by making the operation of GSM BTS sites more reliable; 2. reduce the per unit cost of cellular mobile services (voice and data); 3. reduce greenhouse gas emissions of mobile telecommunication sites by 98.34% and thereby making the environment much more friendly and safe; 4. inform decision making on the use of wind and solar resources on a much larger scale than what it is presently, 5. help the government in the development of policy guidelines for the provision of reliable and sustainable energy supply. 5.2
Recommendations
As Nigeria is already witnessing a tremendous growth in the mobile telecommunication sector, mobile operators should consider the green-mobile infrastructure. Green technology infrastructure consumes lesser energy compared to traditional GSM BTS sites. The use of
237
energy efficient technology would in turn reduce the larger land area needed for the installation of larger optimum operation size of the hybrid green power system. The government, on the other hand, through appropriate agencies and legislation should overhaul the telecommunication sector and develop policies that can encourage extensive utilization of energy efficiency and conservation method. Such policies should include the use of green-mobile base stations, i.e., installing energy efficient infrastructure at new sites and the gradual replacement of an inefficient conversion process with an efficient one. The policy framework should specify the modality for the replacement of all inefficient conversion process with efficient ones. In addition, there should be regulatory policies that can speed up the implementation of policy guidelines to enable the enforcement of the relevant environmental protection laws for enhancing sustainable development in Nigeria. Notwithstanding, adequate funding is essential for motivating renewable and sustainable energy research in Nigeria. Furthermore, the legislation and enabling conditions for electricity sale-back to the national grid should be put in perspective. The gains from electricity sale-back can also encourage extensive utilization of renewable resources in other sectors of the economy and thereby making the environment much more friendly and safe. 5.2.1 Further Research The development of an energy optimization map for Nigeria that incorporates both conventional energy sources (utility grid and diesel generator) and non-conventional (alternative) energy sources (including wind, solar, fuel cell), and hybrid storage system, considering the economy and feasibility of land as well as pollutant emissions as additional
238
constraints in determining the optimum capacity allocation of hybrid energy sources for a given location.
239
LIST OF RESEARCH PUBLICATIONS ISI-Indexed Journals Okundamiya, M. S., Emagbetere, J. O., Ogujor, E. A., 2014. Assessment of Renewable Energy Technology and a Case of Sustainable Energy in Mobile Telecommunication Sector, The Scientific World Journal, vol. 2014, article ID 947281, 13 pages (Impact Factor: 1.73): http://dx.doi.org/10.1155/2014/947281. Okundamiya, M. S., Emagbetere, J. O., Ogujor, E. A., 2014. Design and Control Strategy for a Hybrid Green Energy System for Mobile Telecommunication Sites, Journal of Power Sources, 257: 335-343 (Impact Factor: 4.68): http://dx.doi.org/10.1016/j.jpowsour.2014.-01.121 Okundamiya, M. S., Emagbetere, J.O., Ogujor, E. A., 2015. Techno-Economic Analysis of a Grid-Connected Hybrid Energy System for Developing Regions, Iranica Journal of Energy and Environment, 6(4): 243–254, http://www.ijee.net/Journal/ijee/vol6/no4/1.pdf. Okundamiya, M. S., Emagbetere, J.O., Ogujor, E. A., 2016. Evaluation of Various Global Solar Radiation Models for Nigeria, International Journal of Green Energy, 13(5): 505 – 512 (Impact Factor: 1.469): http://www.tandfonline.com/doi/abs/10.1080/15435075-.2014.968921 Okundamiya, M. S., Emagbetere, J.O., Ogujor, E.A. Modelling and Optimization of Hybrid Energy Systems Considering Economy, Reliability, and Emission, under review
240
Conference Proceedings Okundamiya, M. S., Emagbetere, J.O., Ogujor, E.A., 2014. Assessment of six daily diffuse solar radiation models for Nigeria, in proceeding of the 4th International Workshop on Computer Science and Engineering (WCSE 2014), August 22-23, Dubai, UAE: http://www.sciei.org/wcse/index.htm.
241
REFERENCES Abedi, S., Alimardani, A., Gharehpetian, G. B., Riahy, G. H., Hosseinian, S. H., 2012. A Comprehensive Method for Optimal Power Management and Design of Hybrid RESBased Autonomous Energy Systems, Renewable and Sustainable Energy Reviews, 16 (3): 1577–1587. Adegoke, A. S., Babalola, I.T., 2011. Quality of Service of GSM Telephone System in Nigeria, American Journal of Scientific and Industrial Research, 2(5): 707 – 712. Agbo, S., 2013. Evaluation of the Regression Parameters of the Angstrom-Page Model for predicting Global Solar Radiation, Journal of Energy in Southern Africa, 24 (2): 46 – 49. Aguiar, R., 1996. Synthetic Series of Meteorological Parameters, Ph.D Thesis, Lisbon University, Lisbon, Portugal. AHDC (Anhui Hummer Dynamo Company), 2013, Hummer Specification Sheets, Available from: http://www.hummerwindenergy.com/products [Accessed: April 18, 2013]. Ahmad, F., 1989. Solar Radiation Studies at Karachi, Pakistan, Ph.D Dissertation, University of Karachi, Karachi, Pakistan. Ahmad, M. J., Tiwari, G. N., 2009. Optimization of Tilt Angle for Solar Collector to Receive Maximum Radiation, Open Renewable Energy Journal, 2: 19 – 24. Ahmed, A., Bello, A. A., Habou, D., 2013. An Evaluation of Wind Energy Potential in the Northern and Southern Regions of Nigeria on the Basis of Weibull and Rayleigh Models, American Journal of Energy Engineering, 1(3): 37-42. Ajanaku, L., 2013. Long Road to Quality Service, April 15, Available from: http://thenationonlineng.net/new/issues/long-road-to-quality-service/ [Accessed: July 15, 2013]. 242
Akinoglu, B. G., Ecevit, A., 1990. A Further Comparison and Discussion of Sunshine Based Models to Estimate Global Solar Radiation, Energy, 15:865–872. Akpabio, L. E., Udo, S. O., Etuk, S. E., 2005. Modelling Global Solar Radiation for a Tropical Location: Onne, Nigeria, Turkish Journal of Physics, 29: 63 – 68. Aksoy, H., Toprak, Z. F., Aytek, A., Unal, N. E., 2004. Stochastic Generation of Hourly Mean Wind Speed Data, Reneweable Energy, 29:2111–2131. Al-Abbadi, N. M., Rehman, S., 2009. Wind Speed and Wind Power Characteristics for Gassim, Saudi Arabia, International Journal of Green Energy, 6(2): 201-217. Alam, M. S., Saha, S. K., Chowdhury, M. A. K., Saifuzzaman, M., Rahman, M., 2005. Simulation of Solar Radiation System, American Journal of Applied Sciences, 2(4): 751 – 758. Alba, E., 2005. Parallel Meta-Heuristics: a New Class of Algorithms, Wiley and Sons. Aliyu, A. S., Ramli, A.T., Saleh, M. A., 2013. Nigeria Electricity Crisis: Power Generation Capacity Expansion and Environmental Ramifications, Energy, 61: 354 – 367. Allen, R., 1997. Self-Calibrating Method for Estimating Solar Radiation from Air Temperatures, Journal of Hydrologic Engineering, 2: 56–67 Al-Mohamad, A., 2004. Global, Direct and Diffuse Solar Radiation in Syria, Applied Energy, 79, 191 – 200. Al-Sumait, J. S., Al-Othman, A. K., Sykulski J. K., 2007. Application of Pattern Search Method to Power System Valve-Point Economic Load Dispatch, Electrical Power and Energy Systems, 29, 720 – 730.
243
Amato, U., Cuomo, V., Fontana, F., Serio, F. C., 1989. Statistical Predictability and Parametric Models of Daily Ambient Temperature and Solar Irradiance: An Analysis in the Italian climate, Journal of Applied Meteorology, 28, 711–721. Angeline, P. J., 1998a. Evolutionary Optimization versus Particle Swarm Optimization: Philosophy and Performance Differences, in proceedings of the 7th Annual Conference on Evolutionary Programming, San Diego, USA, pp. 601-610. Angeline, P. J., 1998b. Using Selection to improve Particle Swarm Optimization, The IEEE International Conference on Evolutionary Computation Proceedings Anchorage, AK, USA, pp: 84-89. Angstrom, A., 1924. Solar and Terrestrial Radiation, Quarterly Journal of Royal Meteorological Society, 50: 121–126. Apeh, S. T., 2010. Alleviating Inadequate Electrical Power Generation in Nigeria through Mini Solar Power System, Ph.D Thesis, University of Benin, Benin City, Nigeria Ardakani, F. J., Dehkordi, G. H. R., Abedi, M., 2009. Optimal Sizing of a Stand-Alone Hybrid System for South-West of Iran-Case Study, in proceedings of the 24th International Power System Conference, pp. 1 – 8, Tehran – Iran Athanasia, A. L., Anastassios, D. P., 2000. The economics of PV stand-alone residential households: a case study for various European and Mediterranean locations, International Journal of Solar Energy Materials and Solar Cells, 62, 411 – 427. Awosika, L. F., French, G. T., Nicholls, R. T., Ibe, C. E., 1992. The Impacts of Sea Level Rise on the Coastline of Nigeria‘, in Global Climate Change and the Rising Challenge of the Sea: proceedings of the IPCC Workshop at Margarita, ed. J O‘Callaghan, Island, Venezuela.
244
Badescu, V., 1999. Correlations to Estimate Monthly Mean Daily Solar Global Irradiation: Application to Romania, Energy, 24: 883–893. Baez, G., Kechiche, B., Boguszewska, S., 2010. The Impact of Mobile Services in Nigeria: How Mobile Technologies are Transforming Economic and Social Activities, Pyramid Research, pp. 1 – 150. Bahel, V., Bakhsh, H., Srinivasan, R., 1987. A Correlation for Estimation of Global Solar Radiation, Energy, 12:131–135. Baig, A., Achter, P., Mufti, A., 1991. A Novel Approach to Estimate the Clear Day Global Radiation, Renewable Energy, 1: 119 – 123 Bakare, G. A., Krost, G., 2012a. Comparative Application of Differential Evolution and Particle Swarm Techniques to Reactive Power and Voltage Control, http://rtpis.org/documents/mypaper/RTPIS_publication_1291225494.pdf, [Accessed: April 20, 2012] Bakare, G. A., Krost, G., 2012b. Differential Evolution Approach for Reactive Power Optimization of Nigerian Grid System, http://rtpis.org/documents/mypaper/RTPIS_publication_1284584799.pdf [Accessed: April 20, 2012]. Balamurugan, P., Ashok, S., Jose, T. L., 2009. Optimal Operation of Biomass/Wind/PV Hybrid Energy System for Rural Areas, International Journal of Green Energy, 6(1): 104 – 116. Balouktsis, A., Tsanakas, D., Vachtsevanos, G., 1987. On the Optimum Tilt Angle of a Photovoltaic Array, International Journal of Sustainable Energy, 5(3): 153- 169. Banos, R., Manzano- Agugliano, F., Montoya, F. G., Gil, C., Alcayde, A., Gomez, J., 2011. Optimization Methods Applied to Renewable and Sustainable Energy: a Review, Renewable and Sustainable Energy Review, 15: 1753 – 1766.
245
Baring-Gould, E. I, Manwell, J. F, Van Dijk, V., 2002. Hybrid 2, National Renewable Energy Labouratory. Barley, D., Lew, D., Flowers, L., 1997. Sizing Wind/Photovoltaic Hybrids for Households in inner Mongolia, Colorado: National Renewable Energy Labouratory. Battista, H. D., Mantz, R. J., Garellia, F., 2006. Power Conditioning for a Wind-Hydrogen Energy System, Journal of Power Sources, 155: 478 - 486 Bechrakis, D. A., Sparis, P. D., 2000. Simulation of the Wind Speeds at Different Heights Using Artificial Neural Networks, Wind Engineering, 24 (2): 127-136. Beltran, B., 2007. Sliding Mode Power Control of Variable-Speed Wind, In: Electric Machines and Drives Conference, 2; 943–948 Benghanem, M, 2011. Optimization of Tilt Angle for Solar Panel: Case Study for Madinah, Saudi Arabia, Applied Energy, 88: 1427–1433. Bernal-Agustin, J. L., Dufo-Lopez, R., Rivas-Ascaso, D. M., 2006. Design of Isolated Hybrid Systems Minimizing Costs and Pollutant Emissions, Renewable Energy, 31(14): 2227–2244. Bernardi, D. M, Carpenter, M. K., 1995. A mathematical Model of the Oxygen Recombination Lead–Acid Cell, Journal of Electrochemical Society, 142(8):2631– 2641 Bernieri, A., Noviello, E. I., 1991. A Method for Lead–Acid Battery Performance Prediction, In: proceedings of the 10th IASTED International Symposium on Modelling Identification and Control, 436–441. Besharat, F., Dehghan, A. A., Faghih, A. R., 2013. Empirical Models for Estimating Global Solar Radiation: A Review and Case Study. Renewable and Sustainable Energy Reviews, 21:798–821. 246
Bilbao, J., de-Miguel, A. H., Kambezidis, H. D., 2002. Air Temperature Model Evaluation in the North Mediterranean Belt Area, Journal of Applied Meteorology, 41: 872 – 884 Billinton, R., Chen, H., Ghajar, R., 1996. Time-Series Models for Reliability Evaluation of Power Systems Including Wind Energy, Microelectronics Reliability, 36(9): 12531261. Bin, A., Hongxing, Y., Shen, H., Xianbo, L., 2003. Computer Aided Design for PV/Wind Hybrid System, Renewable Energy, 28, 1491 – 1512. Black, T. N., Bonython, C. W., Prescott, J. A., 1954. Solar Radiation and Duration of Sunshine, Quarterly Journal of Royal Meteorological Society, 80: 231–235. Boland, J., 1997. The Importance of the Stochastic Components of Climatic Variable in Simulating the Thermal Behaviour of Domestic Dwellings, Solar Energy, 60: 359 – 370. Borowy, B. S., Salameh, Z. M., 1996. Methodology for Optimally Sizing the Combination of a Battery Bank and PV Array in a Wind/PV Hybrid System, IEEE Transactions on Energy Conversion, 11(2): 367 –373. Brown, M. A, Southworth, F., Sarzynski, A., 2009. The Geography of Metropolitan Carbon Footprints, Policy and Society, 27: 285–304. Butt, N., New, M., Malhi, Y., Costa, A. C. L., Oliveira, P., Silva-Espejo, J. E., 2010. Diffuse Radiation and Cloud Fraction Relationships in Two Contrasting Amazonian Rainforest Sites, Agricultural and Forest Meteorology, 150: 361–368. Cai, Y. P., Huang, G. H., Tan, Q., Yang, Z. F., 2009. Planning of Community-Scale Renewable Energy Management Systems in a Mixed Stochastic and Fuzzy Environment, Renewable Energy, 34(7):1833– 1847.
247
Calabro, E., 2009. Determining Optimum Tilt Angles of Photovoltaic Panels at Typical North-Tropical Latitudes, Journal of Renewable and Sustainable Energy, 1(033104): 6, doi: 10.1063/1.3148272. Carapellucci, R., Giordano, L., 2013. A Methodology for the Synthetic Generation of Hourly Wind Speed Time Series Based on Some Known Aggregate Input Data, Applied Energy, 101, 541–550 Castellanos, F., Ramesar, V. I., 2006. Characterization and Estimation of Wind Energy Resources Using Autoregressive Modelling and Probability Density Functions, Wind Energy, 30(1): 1–14. CBN (Central
Bank of Nigeria), 2014. Data and
Statistics, Availiable from:
http://www.cenbank.org/rates/inflrates.asp [Accessed: April 21, 2014]. Celik, A., 2003. A Simplified Model for Estimating the Monthly Performance of Autonomous Wind Energy Systems with Battery Storage, Renewable Energy, 28(4): 561 – 572. CER (Commission for Energy Regulation), 2004. Electricity Tariff Structure Review: International Comparisons, An Information Paper, CER/04/101, 46 pp., March 9, 2004. Chang, Y. P., 2009. Optimal Design of Discrete-Value Tilt Angle of PV Using Sequential Neural-Network Approximation and Orthogonal Array, Expert Systems with Applications: an International Journal, 36(3): 6010 – 6018 Chaurey, C., Deambi, S., 1992. Battery Storage for PV Power Systems: An Overview, Renewable Energy, 2(3): 227 –235. Chedid, R. B., Karaki, S. H., El-Chamali, C., 2000. Adaptive Fuzzy Control for Wind–Diesel Weak Power System, IEEE Transactions on Energy Conversion, 15(1): 71 – 78. 248
Chedid, R. B., Saliba, Y., 1996. Optimization and Control of Autonomous Renewable Energy Systems, International Journal of Energy Research, 20: 609 - 624. Chedid, R., Rahman, S., 1997. Unit Sizing and Control of Hybrid Wind-Solar Power Systems, IEEE Transaction on Energy Conversion, 12(1): 79 - 85. Chen, C-C., 2011. An Analytical Framework for Energy Policy Evaluation, Renewable Energy, 36(10): 2694 - 2702. Chen, R., Ersi, K., Yang, J., Lu, S., Zhao, W., 2004. Validation of Five Global Radiation Models with Measured Daily Data in China, Energy Conversion and Management, 45: 1759 –1769 Chen, T., Haesik, H., Yang, Y., 2010. Energy Efficiency Metrics for Green Wireless Communications, in proceedings of International Conference on Wireless Communications and Signal Processing, Suzhou, China, 21–23 October, p. 1–6. Chokmaviroj, S., Wattanapong, R., Suchart, Y., 2006. Performance of a 500 kWp Grid Connected Photovoltaic System at Mae Hong Son Province, Thailand, Renewable Energy, 31 (1): 19 – 28. Collres-Pereira, M., Rabl, A., 1979. The Average Distribution of Solar Radiation Correlations Between Diffuse and Hemispherical and Between Daily and Hourly Insolation Values, Solar Energy, 22 (2): 155 – 164. Correia, L. M., Zeller, D., Blume, O., Ferling, D., Jading, Y., Godor, I., Auer, G., van der Perre, L., 2010. Challenges and Enabling Technologies for Energy Aware Mobile Radio Networks, IEEE Communications Magazine, 48: 66 – 72 Cristobal, J. R. S., 2011. Multi-Criteria Data Envelopment Analysis Model to Evaluate the Efficiency of the Renewable Energy Technologies, Renewable Energy, 36(10): 2742 2746. 249
Cun, J. P., Fiorina, J. N., Fraisse, M., Mabboux, H., 1996. The Experience of UPS Company in Advanced Battery Monitoring, In: INTELEC, Boston, USA; p. 22 – 25 Cuomo, V., Fontana, F., Serio, C., 1986. Behaviour of Ambient Temperature on Daily Basis in Italian climate, Rev. Phys. Appl., 21: 211 – 218. Dagdouguia, H., Minciardia, R. Ouammia, A., Robbaa, M., Sacilea, R., 2010. A Dynamic Optimization Model for Smart Micro-Grid: Integration of a Mix of Renewable Resources for a Green Building, in proceedings of International Congress on Environmental Modelling and Software Society (ICEMSS2010): Modelling for Environment’s Sake, Fifth Biennial Meeting, Ottawa, Canada. Dalton, G. J., Lockington, D. A., Baldock, T. E., 2008. Feasibility Analysis of Stand-Alone Renewable Energy Supply Options for a Large Hotel, Renewable Energy, 33(7): 1475 –90. Damousis, I. G., Alexiadis, M. C., Theocharis, J. B., Dokopoulos, P. S., 2004. A Fuzzy Model for Wind Speed Prediction and Power Generation in Wind Parks Using Spatial Correlation, IEEE Transaction on Energy Conversion, 19(2): 352 – 361. Davidson, O., 2002. Sustainable Energy and Climate Change: African Perspectives, in: Davidson, O and Sparks D. eds. developing energy solutions for climate change, South African research at Energy Development & Research Centre (EDRC), Cape Town, South Africa, p. 145 – 152. Deshmukh, M. L., Deshmukh, S. S., 2008. Modelling of Hybrid Renewable Energy Systems, Renewable and Sustainable Energy Reviews, 2(1): 235 – 249 Diaf, S., Notton, G., Belhamel, M., Haddadi, M., Louche, A., 2007. A Methodology for Optimal Sizing of Autonomous Hybrid PV/Wind System, International Journal of Energy Policy, 35: 5708–5718. 250
Diaf, S., Notton, G., Belhamel, M., Haddadi, M., Louche, A., 2008. Design and TechnoEconomical Optimization for Hybrid PV/Wind System Under Various Meteorological Conditions, Applied Energy, 85(10): 968 – 987. Dorigo, M., Maniezzo, V., Colorni, A., 1991. Positive Feedback as a Search Strategy, Technical Report 91- 016, Dipartimento di Elettronica, Politecnico di Milano, IT, pp: 91-106. Dorigo, M., Maniezzo, V., Colorni, A., 1996. Ant System: Optimization by a Colony of Cooperating Agents, IEEE Transactions on Systems, Mans, and Cybernetics, 1 (26): 29 – 41. Duffie, J. A., Beckman, W. A., 2006. Solar Engineering of Thermal Processes, 3rd Edition, John Wiley and Sons, New York, NY. Dufo-Lopez, R., Bernal-Agustin, J. L., 2005. Design and Control Strategies of PV-Diesel Systems Using Genetic Algorithms, Solar Energy, 79(1): 33 - 46. Dufo-Lopez, R., Bernal-Agustin, J. L., 2008. Multi-Objective Design of PV–Wind–Diesel– Hydrogen–Battery Systems, Renewable Energy, 33(12): 2559 – 2572. Dufo-Lopez, R., Bernal-Agustin, J. L., Javier, C., 2006. Optimization of Control Strategies for Stand-Alone Renewable Energy Systems with Hydrogen Storage, Renewable Energy, 31 Dukes, M. D. G., Palutikof, J. P., 1995. Estimation of Extreme Wind Speeds with Very Long Return Periods, Journal of Applied Meteorology, 34: 1950 – 1961. Ebay, 2014. Available from: http://www.ebay.com/itm/Maxwell-3000-Farad-UltracapacitorSupercapacitor-Super-Ultra-Capacitor-NEW-/301100054670 [Accessed: January 15, 2014].
251
Edgar, G. H., Peters, G. P., 2009. Carbon Footprint of Nations: A Global, Trade Linked Analysis, Environmental Science and Technology, 43: 6414 – 6420. Ekdunge, P., Simonsson, D., 1989. The Discharge Behaviour of the Porous Lead Electrode in the Lead–Acid Battery, I: Experimental Investigations, Journal of Applied Electrochemicals 19(2): 127 – 135. Ekren, O., Ekren, B. Y., 2008. Size Optimization of a PV/Wind Hybrid Energy Conversion System with Battery Storage Using Response Surface Methodology, Applied Energy, 85: 1086 – 1101. Ekren, O., Ekren, B. Y., 2010. Size Optimization of a PV/Wind Hybrid Energy Conversion System with Battery Storage Using Simulated Annealing, Applied Energy, 87: 592 – 598. Elhadidy, M. A., 2002. Performance Evaluation of Hybrid (Wind/Solar/Diesel) Power Systems, Renewable Energy, 26(3): 401 – 413. Elhadidy, M. A., Shaahid, S. M., 2004. Promoting Applications of Hybrid (Wind +Photovoltaic+Diesel+Battery) Power Systems in Hot Regions, Renewable Energy, 29(4): 517 – 528. El-Metwally, M., 2004. Simple New Methods To Estimate Global Solar Radiation Based on Meteorological Data in Egypt, Atmospheric Research, 69: 217–239. ElNozahy, M. S., Salama, M. M. A., 2013. Technical Impacts of Grid-Connected Photovoltaic Systems on Electrical Networks—a Review, Journal of Renewable and Sustainable Energy 5: 032702 (2013); doi: 10.1063/1.4808264. El-Shater, T. F, Eskander, M., El-Hagry, M., 2001. Hybrid Photovoltaic-Fuel Cell System Design and Simulation, in: 36th Intersociety Energy Conversion Engineering Conference, pp. 112 –121. 252
Emagbetere, J. O., 2008. Evaluation of the Performance Characteristic of GSM Network in Nigerian Rural Environment: a Case Study of NIFOR in Edo State and Oghara in Delta State, Ph.D Thesis, University of Benin, Benin City, Nigeria Erbs, D. G., S. A. Klein, Beckman, W. A., 1983. Estimation of Degree-Day and Ambient Temperature Bin Data from Monthly-Average Temperatures, American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) Journal, 25: 60– 65. Erdem, E., Shi, J., 2011. ARMA Based Approaches for Forecasting the Tuple of Wind Speed and Direction, Applied Energy, 88: 1405 – 1414. Erik, F., 2011. The Nigerian Power Sector: a Case Study of Power Sector Reform and the Role
of
PPP,
p.
14.
World
Bank;
November
2011.
Available
from:
http://www.ifc.org/wps/wcm/connect/d96709004956ca9a925abe849537832d/6_Distri bution+and+Generation+PPPs+in+the+Power+Sector+Nigeria+A+Case+Study+of+P ower+Sector+Reform+and+the+Role+of+PPP+-.pdf?MOD=AJPERES
[Accessed:
April 21, 2014]. Ernst, B., Oakleaf, B., Ahlstrom, M. L., Lange, M., Moehrlen, C., Lange, B., et al., 2007. Predicting the Wind-Models and Methods of Wind Forecasting for Utility Operations Planning, IEEE Power Energy Magazine, 5(6): 78 – 89 Eslami, M., Shareef, H., Khajehzadeh, M., Mohamed, A., 2012. A Survey of the State of the Art in Particle Swarm Optimization, Research Journal of Applied Sciences, Engineering and Technology, 4(9): 1181 - 1197. Etisalat Nigeria Limited, Benin City Office, Sapele Road, Benin City.
253
Evans, A., Strezor, V., Evans, T. J., 2010. Sustainability Considerations for Electricity Generation from Biomass, Renewable and Sustainable Energy Review, 14(5): 1419 – 1427. Fadare, D., 2010. The Application of Artificial Neural Networks to Mapping of Wind Speed Profile for Energy Application in Nigeria, Applied Energy, 87: 934 – 942. Falayi, E. O., Adepitan, J. O., Rabiu, A. B., 2008. Empirical Models for the Correlation of Global Solar Radiation with Meteorological Data for Iseyin, Nigeria, International Journal of Physical Sciences, 3 (9): 210 - 216. Fang, X., Kutkut, N., Shen, J., Batarseh, I., 2011. Analysis of Generalized Parallel-Series Ultra-Capacitor Shift Circuits for Energy Storage Systems, Renewable Energy, 36(10): 2599 - 2604. Fard, H. H., Moghaddas-Tafreshi, S. M., Hakimi, S. M., 2013. Optimization of GridConnected Micro-Grid Consisting of PV/FC/UC with Considered Frequency Control, Turkish Journal of Electrical Engineering & Computer Sciences, pp. 1-16, doi:10.3906/elk-1111-28. Ferris, L., Best, J., Scholefield, H., Marques, J., Sawada, B., Nemetz, P., 2007, Carbon neutrality and University of British Colombia: A first glance’, University of British Colombia, Colombia. Filkenstein, J. M., Schafer, R. E., 1971. Improved Goodness to Fit Tests, Biometrical, 58: 641–645. Fulzele, J. B., Dutt, S., 2012. Optimum Planning of Hybrid Renewable Energy System Using HOMER, International Journal of Electrical and Computer Engineering, 2(1): 68 – 74.
254
Fung, C. C., Ho, S. C. Y., Nayar, C. V., 1993. Optimization of a Hybrid Energy System Using Simulated Annealing Technique, in proceedings, IEEE Region 10 Conference on Computer, Communication, Control and Power Engineering, TENCON ´93, 5: 235–238, October 1993. Doi: 10.1109/TENCON.1993.320626. Gansler, R. A., Klein, S. A., Beckman, W. A., 1994. Assessment of Accuracy of Generated Meteorological Data for Use in Solar Energy Simulation Studies, Solar Energy, 53(3): 279 – 287. Gao, Y., 2006. Adequacy Assessment of Electric Power Systems Incorporating Wind and Solar Energy, M.Sc Thesis submitted to Electrical Engineering Department, University of Saskatchewan, Canada. Garcia, R. S., Weisser, D. A., 2006. Wind-Diesel System with Hydrogen Storage: Joint Optimization of Design and Dispatch, Renewable Energy, 31: 2296 – 2320. Garg, S., Dornfeld, D., 2008. An Indigenous Application for Estimating Carbon Footprint of Academia Library Based on Life Cycle Assessment, University of California, Berkeley: Laboratory for Manufacturing and Sustainability, Available from: http://escholarship-.org/uc/item/8zp825mq. [Accessed: March 12, 2012] Gendreau, M., Potvin, J. Y., 2010. Handbook of Meta-Heuristics, International Series in Operations Research and Management Science, Springer, 2010. Girma, Z., 2013. Technical and economic Assessment of Solar PV/Diesel Hybrid Power System for Rural School Electrification in Ethiopia, International Journal of Renewable Energy Research, 3(3): 735 – 744. Glover, F., 1986. Future Paths for Integer Programming and Links to Artificial Intelligence, Computers and Operations Research, 13: 533 - 549. Glover, F., 1990. Tabu search - Part II, ORSA Journal on Computing 2, 4-32. 255
Glower, J., McGulloch, J. S. G., 1958. The empirical Relation Between Solar Radiation and Hours of Sunshine, Quarterly Journal of the Royal Meteorological Society, 84: 172 GMS (Global mobile statistics), 2013 Part A: Mobile Subscribers; Handset Market Share; Mobile Operators, 2013. Available from: http://mobithinking.com/mobile-marketingtools/latest-mobile-stats/a [Accessed: September 15, 2013]. Goldberg, D. E., 2002. Genetic Algorithms in Search Optimization and Appurtenance Learning Reading, Addison-Wesley. Gong, L., Shi, J., 2010. On Comparing three Artificial Neural Networks for Wind Speed Forecasting, Applied Energy, 87: 2313 – 2320. Green, H. J., Manwell, J., 1995. HYBRID2 – A versatile Model of the Performance of Hybrid Power Systems, in proceedings of WindPower’95, Washington DC, March 27 - 30 Gu, H., Nguyen, T. V., White, R. E., 1987. A Mathematical Model of a Lead–Acid Cell: Discharge, Rest, and Charge, Journal of Electrochemical Society, 134(12): 2953 – 2960. Gunerhan, H., Hepbasli, A., 2007. Determination of the Optimum Tilt Angle of Solar Collectors for Building Applications, Building and Environment 42 (2): 779 - 783. Gupta, A., Saini R. P., Sharma, M. P., 2009. Steady-State Modelling of Hybrid Energy System, in 3rd International Conference on Electrical Engineering, ICEE ’09, 1–10, April 2009, Doi: 10.1109/ICEE.2009.5173176. Habib, M. A., Said, S., El-Hadidy, M. A., Al-Zaharna, I., 1999. Optimization Procedure of a Hybrid Photovoltaic-Wind Energy System, International Journal of Energy, 24: 919929.
256
Hakimi, S. M., Moghaddas-Tafreshi, S. M., 2009. Optimal Sizing of a Stand-Alone Hybrid Power System via Particle Swarm Optimisation for Kahnouj Area in South-East of Iran, Renewable Energy, 34 (7): 1855 – 1862. Hakimi, S. M., Moghaddas-Tafreshi, S. M., HassanzadehFard, H., 2011. Optimal Sizing of Reliable Hybrid Renewable Energy System Considered Various Load Types, Journal of Renewable Sustainable Energy, 3: 062701. Hall, I. J., Prairie R. R., Anderson H. E., Boes, E. C., 1979. Generation of Typical Meteorological Years for 26 Solmet Stations, American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) Transaction, 85(2): 507– 517. Hameed, Z., Hong, Y. S., Chu, Y. M., Ahu, S. H., Song, C. K., 2009. Condition Monitoring and Fault Detection of Wind Turbines and Related Algorithms: a Review, Renewable and Sustainable Energy Review, 13 (1): 1 - 39. Hargreaves, G. H., Samani, Z. A., 1982. Estimating Potential Evapotranspiration, Journal of Irrigation and Drainage Engineering, 108(IR3): 223 – 230. Hasan, Z., Boostanimehr, H., Bhargava, V. K., 2011. Green Cellular Networks: a Survey, Some Research Issues and Challenges, IEEE Communications Surveys and Tutorials, 13(4): 524 – 540. Hassan, R., Cohanim, B., de Weck, O., Vente, G., 2004. A Comparison of Particle Swarm Optimization and the Genetic Algorithm, American Institute of Aeronautics and Astronautics Hay, J. E., Davies, J. A., 1980. Calculation of Solar Radiation Incident on an Incline Surface, First Canadian solar radiation data workshop, April 17-19, Toronto, Ontario, Canada, Ministry of Supply and Services, Canada. 257
Heinemann, D., Langer, C., Schumacher, J., 1996. Synthesis of Hourly Ambient Temperature Time Series Correlated with Solar Radiation. Proc. EuroSun’96 Conf., Freiburg, Germany, ISES-Europe, 1518–1523. Hernandez, E., Garcıa, R., Teso, M. T., 1991: Minimum Temperature Forecasting by Stochastic Techniques: an Evidence of the Heat Island Effect. Mausam, 41, 161–166. Hirata, H., Totani, K., Maehata, T., Shimura, T., Take, M., Jurokawa, Y., Onishi, M., Ada, Y., Hirata, Y., 2010. Development of High Efficiency Amplifier for Cellular Base Stations, SEI Tech. Rev. 2010, 47–52. Hocaoglu, F.O., Gerek, O. N., Kurban, M., 2009a. The Effect of Model Generated Solar Radiation Data Usage in Hybrid (Wind–PV) Sizing Studies, Energy Conversion and Management, 50: 2956–2963. Hocaoglu, F.O., Gerek, O. N., Kurban, M., 2009b. A Novel Hybrid (Wind–Photovoltaic) System Sizing Procedure, Solar Energy, 83: 2019 – 2028. Hollands, G. T., D‘Andrea, L. T., Morrison, I. D., 1989. Effect of Random Fluctuations in Ambient Air Temperature on Solar System Performance, Solar Energy, 42: 335–338 Hollick,
J.,
2012.
PV
thermal
systems
-
capturing
the
untapped
energy,
http://solarwall.com/media/images-articles/ASESPaper-PVThermalSystemstheUntappe [Accessed: April 10, 2012]. Hottel, H. C., Woertz, B. B., 1942. Performance of Flat-Plate Solar Heat Collectors, Transactions on American Soociety of Mechanical Engineers (ASME), 64: 91. Hu, X., Eberhart, R.C., 2002. Adaptive Particle Swarm Optimisation: Detection and Response to Dynamic Systems. In Proceedings of congress on Evolutionary Computation, Hawaii, USA, pp: 1666-1670.
258
Huawei Technologies, 2012. Save Energy and Reduce Emissions to Achieve Sustainable Development and Improve Corporate Competitiveness, Huawei Whitepaper For Saving Energy And Reducing Emissions, pp. 1 – 20, China, 2012. Huffman, G. D., 1999. Using the Ideal Gas Law and Heat Release Models to Demonstrate Timing in Spark and Compression Ignition Engines, International Journal of Mechanical Engineering Education 28(4): 279 – 296. Hunt, K., Nason, G. P., 2001. Wind Speed Modeling and Short Term Prediction Using Wavelets, Wind Energy, 25(1): 55 – 61. IPCC, 1996. Revised IPCC Guidelines for National Greenhouse Gas Inventories, Bracknell, IPCC/OECD/IEA, UK Meteorological Office IPCC, 2007. Climate Change 2007: Mitigation of Climate Change, Contribution of Working Group III to the Fourth Assessment Report of the IPCC, Geneva, Switzerland. Iqbal, M., 1979. A study of Canadian Diffuse and Total Solar Radiation Data, Part I. Monthly Average Daily Horizontal Radiation, Solar Energy, 23 (2): 169 – 173. IWS (Internet World statistics), 2014. Available from: http://www.internetworldstats.com/africa.htm, [Accessed: April 21, 2014]. Jafarkazemi, F., Saadabadi, S.A., 2013.
Optimum Tilt Angle and Orientation of Solar
Surfaces in Abu Dhabi, UAE, Renewable Energy, 56: 44 – 49. Jain, P.C. 1988. Estimation of Monthly Average Hourly Global and Diffuse Irradiation, Solar Wind Technology, 5: 7–14. Jain, P.C., 1984. Comparison of Techniques for the Estimation of Daily Global Irradiation and a New Technique for the Estimation of Global Irradiation, Solar Wind Technology, 1: 123–134. 259
Jakhrani, A. Q., Samo, S. R., Rigit, A. R. H., Kamboh, S. A., 2013. Selection of Models for Calculation of Incident Solar Radiation on Tilted Surfaces, World Applied Sciences Journal, 22(9): 1334 - 1343. Jalilzadeh, S., Kord, H., Rohani, A., 2010. Optimization and Techno–Economic Analysis of Autonomous Photovoltaic/Fuel Cell Energy System, ECTI Transaction on Electrical Engineering, Electronics and Communications, February, 8(1): 118 – 125. Jensen, P. M., Puri, A. S., Dear, J. P., Bruneer, J., Morris, A., 2010. Investigating the Impact of Non-Linear Geometrical Effects on Wind Turbine Blades – Part 1: Current Status of the Design and Test Methods and Future Challenges in Design Optimization, Wind Energy, doi: 10.1002/we.415. Jianguo, Z., Qiang, Z., Yu, G., Yingwen, P., 2008. China Mobile in Action, Huawei Communicate, December, 31 – 35. Johnson, E., 2008. Disagreement over Carbon Footprints: a Comparison of Electric and LPG Forklifts, Energy Policy, 36: 1569 – 1573. Justus, C. G., 1978. Winds and Wind System Performance, Franklin Institute Press, Philadelphia, PA. Kalantar, S. M., Mousavi, G., 2010. Dynamic Behaviour of a Stand-Alone Hybrid Power Generation System Turbine, Micro-Turbine, Solar Array and Battery Storage, Applied Energy, 87 (10): 3051 – 3064. Kamali, G.A., Moradi, I., Khalili, A., 2006. Estimating Solar Radiation on Tilted Surfaces with Various Orientations: a Case Study in Karaj, Iran, Theory and Applications of Climatology, 84: 235 - 241. Kaplanis, S., 2006. New methodologies to Estimate the Hourly Global Solar Radiation: Comparisons with Existing Models. Renewable Energy, 31: 781 – 790. 260
Karaki, S. H., Chedid, R. B., Ramadan, R., 1999. Probabilistic Performance Assessment of Autonomous Solar–Wind Energy Conversion Systems, IEEE Transactions on Energy Conversion, 14: 766 – 772. Karakoti, I., Pande, B., Pandey, K., 2011. Evaluation of Different Diffuse Radiation Models for Indian Stations and Predicting the Best Fit Model, Renewable and Sustainable Energy Reviews, 15(5): 2378–2384. Katiyar, A. K., Pandey, C. K., 2010. Simple Correlation for Estimating the Global Solar Radiation on Horizontal Surfaces in India, Energy, 35(12): 5043–5048. Katsigiannis, Y. A., Georgilarkis, P. S., 2008. Optimal Sizing of Small Isolated Hybrid Power Systems Using Tabu Search, Journal of Optoelectronics and Materials, 10(5): 1241 – 1245. Kaviani, A. K., Riahy, G. H., Kouhsari, S. H. M., 2009. Optimal Design of a Reliable Hydrogen-Based Stand-Alone Wind-PV Generating System, Considering Component Outages, Renewable Energy, 34 (11): 2380 – 2390. Kelly, L. M., Shepson, P. B., Strim, B. P., Karion, A., Sweeney, C., Gurney, K. R., 2009. Aircraft-Based Measurements of the Carbon Footprint of Indianapolis, Environmental Science and Technology, 43: 7816 – 7823. Kennedy, J., Eberhart, R., 1995. Particle Swarm Optimization, in proceedings of the IEEE International Conference on Neural Networks, Piscataway, 1942-1948. Kerr, M. J., Cuevas, A., 2003. Generalized Analysis of the Illumination Intensity vs. Open Circuit Voltage of PV Modules, Solar Energy, 76(1–3): 263–267. Khan, M. M., Ahmad, M. J., 2012. Estimation of Global Solar Radiation Using Clear Sky Radiation in Yemen, Journal of Engineering Science and Technology Review, 5 (2): 12-19. 261
Kim, H., Mun, K. J., Park, J. H., Hwang, G., 2001. Application of Real-Type Tabu Search in Function Optimization Problems, IEEE International Symposium, ISIE, Pusan, South Korea, 1: 613 - 618, June 2001. Kim, S. C., Hong, W. H., 1999. Analysis of the Discharge Performance of a Flooded Lead/Acid Cell Using Mathematical Modeling, Journal of Power Sources, 77(1): 74 – 82. Kitagawa, T., Nomura, T., 2003. A Wavelet-Based Method to Generate Artificial Wind Fluctuation Data, Journal of Wind Engineering and Industrial Aerodynamics, 91: 943 – 964. Klein, S. A., 1977. Calculation of Monthly Average Insolation on Tilted Surface, Solar Energy, 19: 325 – 329. Klucher, T. M., 1979. Evaluation of Models to Predict Insolation on Tilted Surfaces, Solar Energy, 23(2): 111 - 114. Klychev, S. I., Mukhammadiev, M. M., Zakhidov, R., Potaenko, K. D., 2007. Technical and Economic Conditions for Creation of Combined Solar-Wind Power Plants, Applied Energy, 43 (4): 214 – 217. Knight, K. M., Klein, S. A., Duffie, J. A., 1991. A Methodology for the Synthesis of Hourly Weather Data, Solar Energy, 46: 109 – 120. Kornelakis, A., Koutroulis, E., 2009. Methodology for the Design Optimization and the Economic Analysis of Grid-Connected Photovoltaic, IET Transactions on Renewable Power Generation, 3 (4): 476 – 492. Kornelakis, A., Marinakis, Y., 2010. Contribution of Optimal Sizing of Grid-Connected PVSystems Using PSO, Renewable Energy, 35 (6): 1333 – 1341.
262
Koutroulis, E., Kolokotsa, D., Potirakis, A., Kalaitzakis, K., 2006. Methodology for Optimal Sizing Of Stand-Alone Photovoltaic/Wind Generator Systems Using Genetic Algorithms, Solar Energy, 80(9): 1072 - 1088. Kovats, R. S., Campbell-Lendrum, D., Matthies, F., 2005. Climate Change and Human Health: Estimating Avoidable Deaths and Disease. Risk Analysis, 25(6): 1409 - 1418. Kozaki, M., Yamazaki, T., 1997. Remaining Battery Capacity Meter and Method for Computing Remaining Capacity, United States Patent No 5691078, November 1997. Kreith, F., Goswami, D.Y., 2011. Principles of Sustainable Energy, CRC Press, 46: 373-420, Boca Raton, FL, USA. Kudish, A.I., Ianetz, A., 1991. Evaluation of the Relative Ability of Three Models, the Isotropic, Klucher and Hay, to Predict the Global Radiation on a Tilted Surface in Beer Sheva, Israel, Energy Conversion and Management, 32: 387-394. Kusiak, A., Song, Z., 2010. Design of Wind Farm Layout for Maximum Wind Energy Capture, Renewable Energy, 35 (3): 685 – 694. Lagorse, J., Paire, D., Miraoui, A., 2009. Sizing Optimization of a Stand-Alone Street Lighting System, Powered by a Hybrid System Using Fuel Cell, PV and Battery, Renewable Energy, 34(3): 683 – 691. Laugier, A., Garai, J., 2007. Derivation of the Ideal Gas Law, Journal of Chemical Education 84(11): 1832 – 1833. Lewis, G., 1987. Optimum Tilt of Solar Collector, Solar and Wind Technology, 4(3): 407 410. LGOP, 2010. The Climate Registry, Local Government Operations Protocol (LGOP), Version 1.1, May 2010, 229 p. 263
Li, W., Billinton, R., 1991. Effect of Bus Load Uncertainty and Correlation in Composite System Adequacy Evaluation, IEEE Transaction on Power Systems, 6(4): 1522 1528. Liu, B. Y. H., Jordan, R. C., 1960. The Interrelationship and Characteristic Distribution of Direct, Diffuse and Total Radiation, Solar Energy, 4(3): 1–19. Liu, B. Y. H., Jordan, R. C., 1962. Daily Insolation on Surfaces Tilted Towards Equator, Transaction on American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE), 526 – 541. Liu, B. Y. H., Jordan, R. C., 1963. The Long-Term Average Performance of Flat-Plate Solar Energy Collectors, Solar Energy, 7(2): 53-74. Liu, G., Rasul, M. G., Amanullah, M. T. O., Khan, M. M. K., 2010. Economic and Environmental Modelling of a Photovoltaic-Wind-Grid Hybrid Power System in Hot Arid Australia, International Journal of Thermal & Environmental Engineering 1(1):15-22. Lorincz, J., Garma, T., Petrovic, G., 2012. Measurements and Modelling of Base Station Power Consumption Under Real Traffic Loads, Sensors, 12: 4281 - 4310. Lovbjerg, M., 2002. Improving Particle Swarm Optimization by Hybridisation of Stochastic Search Heuristics and Self-Organized Criticality, M.Sc. Thesis, University of Aarhus. Lu, L., Yang, H., Burnett, J., 2002. Investigation on Wind Power Potential on Hong Kong Islands-An Analysis of Wind Power and Wind Turbine Characteristics, International Journal of Renewable Energy, 27(1): 1-12. Mabel, M. C., Fernandez, E., 2008. Analysis of Wind Power Generation and Prediction Using ANN: a Case Study. Renewable Energy, 33(5): 986 – 992.
264
Mahapatra, S., Panda, S., Swain, S. C., 2014. A Hybrid Firefly Algorithm and Pattern Search Technique for SSSC Based Power Oscillation Damping Controller Design, Ain Shams Engineering Journal, 5(4): 1177–1188. Mahor, A., Prascad, V., Rangnekar, S., 2009. Economic Dispatch Using Particle Swarm Optimization: a Review, Renewable and Sustainable Energy Review, 13(8): 2134 – 2141. Mala, M., Nyanganji, J. K., Mukhtar, A., 2012. Gully Development Along River Ngaddabul Floodplain of Maiduguri, Borno State, Nigeria, Journal of Environmental Issues and Agriculture in Developing Countries, 4(1): 45-53, April 2012. Malinga, B., Sneckenberger, J. E., Feliachi, A., 2003. Modelling and Control of a Wind Turbine as a Distributed Resource, Paper Presented at the IEEE proceedings of the 35th Southeastern Symposium on System Theory, March 16–18, Morgantown, WV, USA, 108 – 112. Mani, A., Rangarajan, S., 1982. Solar Radiation over India, New Delhi: Allied Publishers, Pvt. Ltd. Manwell, J. F., McGowan, J. G., 1993. Lead Acid Battery Storage Model for Hybrid Energy Systems, Solar Energy, 50: 399 - 405. Markvart, T., 1996. Sizing of Hybrid PV–Wind Energy Systems, Solar Energy, 59(4): 277– 281. Martinez-Diaz, M., Villafafila-Robles, R., Montesinos-Miracle, D., Sudrià-Andreu, A., 2013. Study of Optimization Design Criteria for Stand-Alone Hybrid Renewable Power Systems, International Conference on Renewable Energies and Power Quality (ICREPQ’13), Bilbao (Spain), March 20 - 22, 2013.
265
Mehleri, E. D., Zervas, P. L., Sarimveis, H., Palyvos, J. A., Markatos, N. C., 2010. Determination of the Optimal Tilt Angle and Orientation for Solar Photovoltaic Arrays, Renewable Energy, 35: 2468 – 2475. Mellit, A., Kalogirou, S. A., Hontoria, L., Shaari, S., 2009. Artificial Intelligence Techniques for Sizing Photovoltaic Systems: a Review, Renewable and Sustainable Energy Review, 13 (2): 406 – 419. Mohammed, F. A., Koivo, H. N., 2010. System Modelling and Online Optimal Management of Micro-Grid Using Mesh Adaptive Direct Search, International Journal of Electrical Power and Energy Systems, 32: 398 – 407. Momoh, J. A., Adapa, R., El-Hawary, M., 1999a. A Review of Selected Optimal Power Flow Literature to 1993. I: Nonlinear and Quadratic Programming Approaches, IEEE Transaction on Power System, 14: 96-104. Momoh, J. A., El-Hawary, M., Adapa, R., 1999b. A Review of Selected Optimal Power Flow Literature to 1993, II: Newton, Linear Programming and Interior Point Methods, IEEE Transaction on Power System, 14: 105-111. Moore, F., Diaz, D., 2015. A New Study Calculating the Economic Impacts of Climate Change Shows the Costs of Carbon Dioxide Emissions have been Underestimated, Nature Climate Change. Available from:
http://news.stanford.edu/news/2015-
/january/-emissions-social-costs-011215.html [Accessed on February 2, 2015]. Morgan, T. R., Marshall, R. H., Brinkworth, B. J., 1997. ARES-A Refined Simulation Programme for the Sizing and Optimization of Autonomous Hybrid Energy Systems, Solar Energy, 59(4–6): 205 – 215. MTI (Maxwell Technologies Incoporated), 2014. 3888 Calle Fortunada, San Diego CA 92123,
USA.
Available
from: 266
http://www.maxwell.com/ultracapacitors/,
http://www.richardsonrfpd.com/Pages/Product-Details.aspx?productId=1100547 [Accessed: December 12, 2013]. Muneer, T., Munawwar, S., 2006. Potential for Improvement in Estimation of Solar Diffuse Radiation, Energy Conversion and Management, 47: 68–86. Muselli, M., Notton, G., Poggi, P., Louche, A., 2000. PV-Hybrid Power Systems Sizing Incorporating Battery Storage: an Analysis via Simulation Calculations, Renewable Energy, 20(1): 1–7. Nafeh, A. A., 2011. Optimal Economical Sizing of a PV-Wind Hybrid Energy System Using Genetic Algorithm, International Journal of Green Energy, 8(1): 25 - 43. Nandi, S. K., Ghosh, H. R., 2009. A Wind-PV-Battery Hybrid Power System at Sitakunda in Bangladesh, Energy Policy, 37: 3659 - 3664. Nandi, S. K., Ghosh, H. R., 2010. Prospect of Wind-PV-Battery Hybrid Power System as an Alternative to Grid Extension in Bangladesh, Energy, 35: 3040 – 3047. NASA (National Aeronautics and Space Administration), 2012. Surface Meteorology and Solar Energy Data and Information, Available from: http://eosweb.larc.nasa.gov/sse [Accessed: November 20, 2012]. Nayeripour, M., Hoseintabar, M., 2011. A Comprehensive Dynamic Modelling of GridConnected Hybrid Renewable Power Generation and Storage System, International Journal of Modelling and Optimization, 1(1): 62 – 65. NCC (Nigerian Communications Commission), 2014. Industry Data, Available from: http://www.ncc.gov.ng/index.php?option=com_content&id=68&Itemid=70 [Accessed: January 28, 2015].
267
Nema, P., Nema, R. K., Rangnekar, S., 2010. PV-Solar/Wind Hybrid Energy System for GSM/CDMA Type Mobile Telephony Base Station, International Journal of Energy and Environment, 1(2): 359 – 366. NERC (Nigerian Electricity Regulatory Commission), 2012. Multi-Year Tariff Order MYTO) II, Nigerian Electricity Regulatory Commission. Available from: http://www.nercng.org/index.php/myto-2/rates, [Accessed: August 12, 2014]. Nguyen, T. V., White, R. E., Gu, H., 1990. The Effects of Separator Design on the Discharge Performance of a Starved Lead–Acid Cell, Journal of Electrochemical Society, 137(10): 2998 – 3004. NIMET, 2013. Archives of Nigerian Meteorological (NIMET) agency, Oshodi, Lagos State, Nigeria. Nishioka, K., Hatayama, T., Uraoka, Y., Fuyuki, T., Hagihara, R., Watanabe, M., 2003. Field-Test Analysis of PV System Output Characteristics Focusing on Module Temperature, Solar Energy Materials and PV Modules, 75(3–4): 665 – 671. NREL (National Renewable Energy Laboratory), 2010. Hybrid Optimization Model for Electric
Renewables
(HOMER)
Software,
Available
from:
http://www.nrel.gov/intern-ational/tool/HOMER/homer.html [Accessed: January 12, 2014]. Ogelman, H., Ecevit, A., Tasdemiroglu, E., 1984. A New Method for Estimating Solar Radiation from Bright Sunshine Data, Solar Energy, 33: 619 – 625. Ogujor, E. A., 2007. Reliability Assessment of Electric Power Distribution, Ph.D Thesis, University of Benin, Benin City, Nigeria
268
Ogujor, E. A., Orobor, N. P., 2010. SAIDI Minimization through Conventional Energy Resources: Opportunities and Challenges, Journal of Economics and Engineering, 54 – 6. Okere, C., 2012. Recounting the Gains of GSM in Nigeria, August 6, 2012, Available from: http://www.voiceofnigeria.org/comentary/recounting-the-gains-of-gsm-innigeria.html [Accessed: September 29, 2013]. Okundamiya, M. S., Akpaida, V. O. A., Omatahunde, B. E., 2014a. Optimization of a Hybrid Energy System for Reliable Operation of Automated Teller Machines, Journal of Emerging Trends in Engineering and Applied Sciences, vol. 5, no. 8, pp. 153 -158. Okundamiya, M. S., Emagbetere, J. O., Ogujor, E. A., 2014b. Assessment of Renewable Energy Technology and a Case of Sustainable Energy in Mobile Telecommunication Sector, The Scientific World Journal, vol. 2014, article ID 947281, 13 pages. Okundamiya, M. S., Nzeako, A. N., 2011. Influence of Orientation on the Performance of a Photovoltaic Conversion System in Nigeria, Research Journal of Applied Sciences, Engineering and Technology, 3(12): 1386 – 1392. Okundamiya, M. S., Nzeako, A. N., 2013. Model for Optimal Sizing of a Wind Energy Conversion System for Green-Mobile Applications, International Journal of Green Energy, 10 (2): 205 - 218. Oliveira, C., 2007. The Energy Shift: Towards a Renewable Future, International Journal of Low Carbon Technologies, 2(3): 289 – 299. Oloketuyi, S. I., Oyewola, M. O., Odesola, I. I., 2013. Determination of Optimum Tilt Angles for Solar Collectors in Low-Latitude Tropical Region, International Journal of Energy and Environmental Engineering, 4: 29, doi: 10.1186/2251-6832-4-29.
269
Omotayo, F. S., Oguntunde, P. G., Olufayo, A. A., 2011. Evaluation of Solar Radiation Models for Predictive Purpose and Application in Agriculture, Water and Energy, M.Eng. Thesis, Federal University of Technology, Akure, Nigeria. Available from: http://hdl.handle.net/123456789/2237 [Accessed: November 13, 2013]. Oseni, M.O., 2011. An Analysis of the Power Sector Performances in Nigeria, Renewable and Sustainable Energy Reviews, 15: 4765 – 4774. Otokpa, D. O., Ajeibi, B. C., 2013. Electric Energy Tariffs in a Deregulated Economy for Self Reliance and Sustainable Development in Nigeria, International Journal of Scientific & Engineering Research, 4(6): 3046 – 3054. Ould, B., Sambou, V., Ndiaye, P. A., Kebe, C. M. F., Ndongo, M., 2010. Optimal Design of a Hybrid Solar-Wind-Battery System Using Minimization of the Annualized Cost System and the Minimization of the Loss of Power Supply Probability (LPSP), Renewable Energy, 35(10): 2388 – 2390. Oyedepo, S. O., 2012. On Energy for Sustainable Developments in Nigeria, Renewable and Sustainable Energy Reviews, 16: 2583– 2598. Page, J. K., 1964. The Estimation of Monthly Mean Values of Daily Total Short-Wave Radiation on Vertical and Inclined Surfaces from Sunshine Records For Latitude 40oN – 40oS, in Proceeding of the UN Conference on New Sources of Energy, Paper S/ 98. Pandey, D., Agrawal, M., Pandey, J. S., 2011. Carbon Footprint: Current Methods of Estimation,
Environmental
Monitoring
Assessment,
178:135–160.
Doi:
10.1007/s10661-010-1678-y. Papaefthymiou, G., Klockl, B., 2008. MCMC for Wind Power Simulation, IEEE Transaction on Energy Conversion, 23(1): 234 - 240. 270
Parida, B., Iniyan, S., Goic, R., 2011. A Review of Solar Photovoltaic Technologies, Renewable and Sustainable Energy Reviews, 15: 1625 – 1636. Park, S. J., Kang, B. B., Yoon, J. P., Cha, S., Lim, J. Y., 2004. A Study on the Stand-Alone Operating or Photovoltaic Wind Power Hybrid Generation System, in: 35th Annual IEEE Power Electronics Specialists Conference, 2095 – 2099. Pavlovic, T., Pavlovic, Z. L., Pantic, L., Kostic, J., 2010. Determining Optimum Tilt Angles and Orientations of Photovoltaic Panels in Nis Serbia, Contemporary Materials, 2: 151–156. Pelet, X., Favrat, D., Leyland, G., 2005. Multi-Objective Optimization of Integrated Energy Systems for Remote Communities Considering Economics and CO2 Emissions, International Journal of Thermal Science, 44(12): 1180 – 1189. Phelan, S., 2014. Using Atmospheric Pressure Tendency to Optimize Battery Charging in Off-Grid Hybrid Wind-Diesel Systems for Telecoms, Ph.D Thesis, Dublin City University, Dublin. Philippopoulos, K., Deligiorgi, D., 2009. Statistical Simulation of Wind Speed in Athens, Greece Based on Weibull and ARMA Models, International Journal of Energy and Environment, 3(4): 151 - 158. Piller, S., Perrin, M., Jossen, A., 2001. Methods for State-of-Charge Determination and their Applications, Journal of Power Sources, 96(1): 113 – 120. Pongthanaisawan, J., Sorapipatana, C., 2013, Greenhouse Gas Emissions from Thailand‘s Transport Sector: Trends and Mitigation Options, Applied Energy, 101: 288–298. Posadillo, R., Lopez-Luque, R., 2009. Evaluation of the Performance of Three Diffuse Hourly Models on Tilted Surfaces According to the Utilizability Concept, Energy Conversion and Management, 50(9): 2324 - 2330. 271
Prescott, J. A., 1940. Evaporation from Water Surface in Relation to Solar Radiation, Transactions of the Royal Society of Australia, 46: 114 - 118. Qiang, Z., 2008. Creating Green Sites, Huawei Communicate, December, 45: 36 – 38. Qiu, G., Riffat, S. B., 2003. Optimum Tilt Angle of Solar Collectors and its Impact on Performance, International Journal of Ambient Energy, 24 (1): 13 – 20. Radziemska, E., Klugmann, E., 2002. Thermally Affected Parameters of the Current–Voltage Characteristics of Silicon Photocell, Energy Conversion and Management, 43(14): 1889 – 1900. Rahman, S. M., Khondaker, A. N., 2012. Mitigation Measures to Reduce Greenhouse Gas Emissions and Enhance Carbon Capture and Storage in Saudi Arabia, Renewable and Sustainable Energy Reviews, 16: 2446 – 2460 Rahmstorf, S., 2008. Global Warming: Looking beyond Kyoto. Brookings Institution Press, Ch. Anthropogenic Climate Change: Revisiting the Facts, 34–53. Rasuo, B. P., Bengin, A. L., 2010. Optimization of Wind Farm Layout, PME Transactions, 38: 107 – 114. Ratnaweera, A., Halgamuge, S. K., Watson, H.C., 2004. Self-Organizing Hierarchical Particle Swarm Optimizer with Time-Varying Acceleration Coefficients, IEEE Transaction on Evolutionary Computation, 8: 240 - 255. Ray, M. L., Rogers, A. L., McGowan, J. G., 2006. Analysis of Wind Shear Models and Trends in Different Terrain, in Proceedings of American Wind Energy Association Wind power, Pittsburgh, PA, USA; June 2006. Reddy, K. N., Agarwal, V., 2007. Utility Interactive Hybrid Distributed Generation Scheme with Compensation Feature, IEEE Transactions on Energy Conversion, 22(3): 666– 673. 272
Rehman, S., Ahmad, A., Al-Hadhrami, L. M., 2010. Detailed Analysis of a 550-MW Installed Capacity Wind Farm in Saudi Arabia, International Journal of Green Energy, 7(4): 410 - 421 Rehman, S., Al-Abbadi, N. M., 2010. Wind Power Characteristics on the North West Coast of Saudi Arabia, Energy & Environment, 20(8)/21(1): 1257 - 1270. Reindl, D. T., Beckman, W. A., Duffie, J. A., 1990a. Diffuse Fraction Correlations, Solar Energy, 45: 1. Reindl, D. T., Beckman, W. A., Duffie, J. A., 1990b. Performances Evaluation of Hourly Tilted Surface Radiation Models, Solar Energy, 45: 9. Rietveld, M., 1978. A New Method for Estimating the Regression Coefficients in the Formula Relating Solar Radiation to Sunshine, Agricultural Meteorology, 19:243–52. Roy, S. N., 2008. Energy logic: A Road Map to Reducing Energy Consumption in Telecommunications Networks, in proceedings of Emerson Network Power, IEEE 30th Telecommunications Energy Conference (INTELEC), San Diego, CA, USA, 14– 18 September 2008; pp. 1–9. Sahin, A. D., Sen, Z., 2001. First-Order Markov Chain Approach to Wind Speed Modelling, Journal of Wind Engineering and Industrial Aerodynamics, 89: 263 – 269. Sambo, A. A., Garba, B., Magaji, M. M., Electricity Generation and the Present Challenges in the Nigerian Power Sector, Paper 70 presented at the 2010 World Energy Congress of the World Energy Council, Montreal, Canada. Santucci, A., Sorniotti, A., Lekakou, C., 2014. Power split Strategies for Hybrid Energy Storage Systems for Vehicular Applications, Journal of Power Sources, 258: 395 407
273
Schoonderwoerd, R., Holland, O., Bruten, J., Rothkrantz, L., 1997. Ant-Based Load Balancing in Telecommunications Networks, Adaptive Behaviour, 5 (2): 169 – 207. SEDC (Solar Energy DC), 2013. Technical Data Sheet and Price Lists, Available on: http://cdn.shopify.com/s/files/1/0408/8521/files/130_150w_mono_solar_panels.pdf?2 491, http://solarenergydc.com/collections/solar-panels-pv/products/solar-panel-150w [Accessed: December 21, 2013]. Selvi, V., Umarani, R., 2010. Comparative Analysis of Ant Colony and Particle Swarm Optimization Techniques, International Journal of Computer Applications, 5 (4): 1–6. Senjyu, T., Hayashi, D., Urasaki, N., Funabashi, T., 2006. Optimum Configuration for Renewable Generating Systems in Residence Using Genetic Algorithm, IEEE Transactions on Energy Conversion, 21(1): 459 – 467. Senjyu, T., Hayashi, D., Yana, A., Urasaki, N., Funabashi, T., 2007. Optimal Configuration of Power Generating Systems in Isolated Island with Renewable Energy, Renewable Energy, 32(11): 2111 – 2118. Sfetsos, A. A., 2000. Comparison of Various Forecasting Techniques Applied to Mean Hourly Wind Speed Time Series, Renewable Energy, 21: 23 – 35. Shakyaa, B. D., Ayea, L., Musgraveb, P., 2005. Technical Feasibility and Financial Analysis of Hybrid Wind-Photovoltaic System with Hydrogen Storage for Cooma, International Journal of Hydrogen Energy, 30: 9 - 20. Shamshad, A., Bawadi, M. A., Wan-Hussin, W. A., Majid, T. A., Sanusi, S. M., 2005. First and Second Order Markov Chain Models for Synthetic Generation of Wind Speed Time Series. Energy, 30: 693–708.
274
Sonmete, M. H., Ertekin, C., Menges, H. O., Hacıseferogullari, H., Evrendilek, F., 2011. Assessing Monthly Average Solar Radiation Models: a Comparative Case Study in Turkey, Environmental Monitoring Assessment, 175: 251 – 277. Stone, R. J., 1993. Improved Statistical Procedure for the Evaluation of Solar Radiation Estimation Models, Solar Energy, 51(4): 289 – 291. Stone, R. J., 1999. A New Sunshine-Based Global Solar Radiation Estimation Model for Trinidad, West Indian Journal of Engineering, 22(1): 42 – 49, July 1999. Storn, R., Price, K., 1995. Differential Evolution–a Simple and Efficient Adaptive Scheme For Global Optimization Over Continuous Spaces, Technical Report (TR-95-012), March 1995. Stroud, K. A., 1996. Further Engineering Mathematics: Programmes and Problems, 3rd Edition, Macmillan Press Ltd., London. Supit, I., Van Kappel, R. R., 1998. A Simple Method to Estimate Global Radiation, Solar Energy, 63: 147 – 160. Supriya, C. S., Siddarthan, M., 2011. Optimization and Sizing of a Grid-Connected Hybrid PV-Wind Energy System, International Journal of Engineering Science and Technology, 3(5): 4297 – 4323. Sweelem, E. A., Nafeh, A. A., 2006. A Linear Programming Technique for Optimally Sizing a PV-Diesel-Battery Hybrid Energy System, Al-Azhar University Engineering Journal 9(3): 705 – 716. Talebhagh, S., Kareghar, H. K., 2012. Applying Grid-Connected Photovoltaic Systems as Supplement to Public Electricity Network in Rural Area of Iran, International Journal of Advanced Renewable Energy Research, 1(7): 419- 428.
275
Tallapragada, P. V. S. N., 2009, Nigeria‘s Electricity Sector: Electricity and Gas Pricing Barriers, International Association for Energy Economics (IAEE) Energy Forum, 1st Quarter, 2009. Tankari, M. A., Camara, M. B., Dakyo, B., Nichita, C., 2011. Ultra-Capacitors and Batteries Integration for Power Fluctuations Mitigation in Wind-Photovoltaic-Diesel Hybrid System, International Journal of Renewable Energy Research, 1(2): 86 – 95. Teoh, W. Y., Khu, S. Y., Tan, C. W., Hii, I. H., Cheu, K. W., 2012. Technoeconomic and Carbon Emission Analysis for a Grid-Connected Photovoltaic System in Malacca, International Scholarly Research Network (ISRN) Renewable Energy, vol. 2012, article ID 745020, 8 pages, 2012. Tina, G., Gagliano, S., Raiti, S., 2006. Hybrid Solar-Wind Power System Probabilistic Modelling for Long-Term Performance Assessment, Solar Energy, 80(5): 578 – 588. Trani, A. A., 2013. Analysis of Air Transportation Systems: Fundamentals of Aircraft Performance (1), Virginia Tech-Air Transportation Systems Laboratory, p. 131, 2013. Available from: http://atsl-g4-733.cee.vt.edu/courses/cee5614/cee5614_pub/Aircraft_perf_notes1_2013.pdf [Accessed: April 21, 2014]. Trazouei, S. L., Tarazouei, F. L., Ghiamy, M., 2013. Optimal Design of a Hybrid Solar Wind-Diesel Power System for Rural Electrification Using Imperialist Competitive Algorithm, International Journal of Renewable Energy Research, 3(2): 403 – 411. Tsou, D., MacNish, C., 2003. Adaptive Particle Swarm Optimization for High-Dimensional Highly Convex Search Spaces, in proceedings of IEEE Congress on Evolutionary Computation 2003 (CEC 2003), Canbella, Australia, pp: 783-789.
276
Turcotte, D., Ross, M., Sheriff, F., 2001. Photovoltaic Hybrid System Sizing and Simulation Tools: Status and Needs, in proceeding of Photovoltaic Horizon: Workshop on the Photovoltaic Hybrid Systems, September 10, 2001. Ulgen, K., 2006. Optimum Tilt Angle for Solar Collectors, Energy Sources: Part A, 28 (13): 1171 - 1180. UNDP-GEF (United Nations Development Programme- Global Environment Facility), 2013. End-Use Metering Campaign For Residential Houses In Nigeria, in Metering Campaign Report, Draft Version, p.148, September, 2013. Available from: http://www.ng.undp.org/content/dam/nigeria/docs/Sustainable%20Development/UND P_NG_sustaindev_Metering_Campaign.pdf [Accessed: May 2, 2014]. Valenciaga, F., Puleston, P. F., 2005. Supervisor Control for a Stand-Alone Hybrid Generation System Using Wind and Photovoltaic Energy, IEEE Transactions on Energy Conversion, 20(2): 398 – 440. Valente, L. C. G., Almeida, S. C. A. D., 1998. Economic Analysis of a Diesel-Photovoltaic Hybrid System for Decentralized Power Generation in Northern Brazil, Energy, 23(4): 317 – 323. VandeWeghe, J. R., Kennedy, C., 2007. A Spatial Analysis of Residential Greenhouse Gas Emissions in the Toronto Census Metropolitan Area, Journal of Industrial Ecology, 11(2): 133 – 144. Van-Voorden, M. A., 2008. Power Balancing in Autonomous Renewable Energy Systems, Sustainable Energy: Extraction, Conversion and Use (SENECU), TU Delft, Netherlands, 150 – 159. Vine, E., 2009. Breaking Down Silos: The Integration of Energy Efficiency, Renewable Energy, Demand Response and Climate Change, Energy Efficiency, 1: 49 – 63. 277
Wahab, M. A., Essa, K. S. M., 1998. Extrapolation of Solar Irradiation Measurements: Case Study over Egypt, Renewable Energy, 14(1–4):229–39. Wang,
C.,
Nehrir,
M.
H.,
2008.
Power
Management
of
a
Stand-Alone
Wind/Photovoltaic/Fuel Cell Energy System, IEEE Transactions on Energy Conversion, 23 (3): 957 - 967, September 2008. Wan-Nik, W. B., Ibrahim, M. Z., Samo, K. B., Muzathik, A. M., 2012. Monthly Mean Hourly Global Solar Radiation Estimation, Solar Energy, 86: 379 – 387. WCED (World Commission on Environment Development), 1987. Our Common Future, World Commission on Environment Development. Wikipedia, 2014a. Available from: http://en.wikipedia.org/wiki/Nigeria [Accessed: April 17, 2014]. Wikipedia, 2014b. Available from: http://en.wikipedia.org/wiki/United_Kingdom [Accessed: April 17, 2014]. Williams, E. A., Crossley, W. A., 1998. Empirically-Derived Population Size and Mutation Rate Guidelines for a Genetic Algorithm with Uniform Crossover, Soft Computing in Engineering Design and Manufacturing, Chawdhry, P. K., Roy, R., Pant, R. K., (editors), Springer-Verlag, pp. 163 - 172. World Bank, 1993. Issues and Options in the Energy Sector, Report No. 11672-UNI, Nigeria: World Bank; 1993. Xu, D., Kang, L., Chang, L., Cao, B., 2005. Optimal Sizing of Stand-Alone Hybrid Wind/PV Power Systems Using Genetic Algorithms, in: Canadian Conference on Electrical and Computer Engineering, pp. 1722 - 1725. Yang H. X., Lu, L., 2004. Study on Typical Meteorological Years and their Effect on Building Energy and Renewable Energy Simulations, American Society of Heating, 278
Refrigerating and Air-Conditioning Engineers (ASHRAE) Transaction, 110(2): 424 – 431. Yang, H. X., Lu, L., 2007. The Optimum Tilt Angles and Orientations of PV Claddings for Building Integrated Photovoltaic Applications, Journal of Solar Energy Engineering, 129: 253 – 255. Yang, H. X., Lu, L., Zhou, W., 2007. A Novel Optimization Sizing Model for Hybrid Solar– Wind Power Generation System, Solar Energy, 81(1): 76 – 84. Yang, H. X., Zhou, W., Lou, C. Z., 2009. Optimal Design and Techno-Economic Analysis of a Hybrid Solar–Wind Power Generation System, Applied Energy, 86: 163 – 169. Yang, H. X., Zhou, W., Lu, L., Fang, Z. H., 2008. Optimal Sizing Method for Stand-Alone Hybrid Solar–Wind System with Loss of Power Supply Probability Technology by Using Genetic Algorithm, Solar Energy, 82(4): 354 - 367. Yang, W. J., Aydin, O., 2001. Wind Energy–Hydrogen Storage Hybrid Power Generation, International Journal of Energy Research, 25: 449 – 463. Yousif, C., Quecedo, G. O., Santos, J. B., 2013. Comparison of Solar Radiation in Marsaxlokk, Malta and Valladolid, Spain, Renewable Energy, 49: 203 – 206. Yousif, E., 2012. A Study of a Grid-Connected Photovoltaic Household System in Amman and the Effect of the Incentive Tariff on the Economic Feasibility, International Journal of Applied Science and Technology, 2(2): 100- 105, February 2012. Zhou, W., Lou, C., Lu, L., Yang, H., 2010. Current Status of Research on Optimum Sizing of Stand-Alone Hybrid Solar-Wind Power Generation Systems, Applied Energy, 87 (2): 380 – 389. Zoican, S., 2008. The Role of Programmable Digital Signal Processors for 3G Mobile Communications Systems, ACTA Tech. NAPOC. 2008, 49: 49– 56. 279
APPENDICES Appendix A: Main Technical Specifications of Hybrid Energy System Components Table A1: Technical Specifications of Wind Turbine Generators Parameter Model
Specifications H3.1-1kW
H3.8-2kW
H4.6-3kW
H6.4-5kW
Rated power (kW)
1.0
2.0
3.0
5.0
Maximum output power (kW)
1.2
2.2
3.3
5.6
Battery bank voltage, VDC (V)
60
120
180
240
110/220
110/220
110/220
110/220/380
Cut-in wind speed (m s-1)
3
3
3
3
Rated wind speed (m s-1)
10
11
10
12
Rated rotating rate (r min-1)
500
450
265
240
Blade diameter (m)
3.1
3.8
4.9
6.4
Generator weight (kg)
15
25
72
147
System output voltage, VAC (V)
Wind energy utilizing ratio, Cp Over speed regulation Energy monitoring/control
0.45
0.4
Passive furling tail design
Auto Yawing Siemens PLC controller with touch screen
LED display
Working wind speed (m s-1)
3 - 25
Survival wind speed (m s-1)
50
Generator efficiency
> 0.8
Generator type (Alternator)
Permanent Magnet Alternator
Configuration
3 Blades, Horizontal axis, Upwind
Blade material
Fibre glass reinforced composite
Shutting down method
Manual + Automatic Electromagnetism braking
Source: Internet (http://www.hummerwindenergy.com/products; accessed on April 18, 2013).
280
Table A2: Technical Specifications of PV Module
Source: Internet (http://cdn.shopify.com/s/files/1/0408/8521/files/130-150w_mono_solar_panels.pdf?2491; accessed on June 19, 2014).
281
Table A3: Technical Specifications of Inverter Parameter Model
Input
Specifications MAILI Pass Power
MLP-1000
DC Voltage (V)
12/ 24/ 48
DC Voltage range (V)
9.5~16 19~32 38~64
No Load Current Draw (A)
< 0.3A
Efficiency (%)
>90
DC Connector
Cables With Clips or Car Adaptor
AC Voltage (V)
100, 110, 120, 220, 230, 240
Continuous Power (kW)
1.0
Surge Power (kW)
2.0
Waveform
Pure Sine Wave
Frequency (Hz)
50 / 60
THD (%)
