scalable dc microgrids for rural electrification
TRANSCRIPT
i
Scalable DC Microgrids for Rural
Electrification
A Dissertation
Presented by
MASHOOD NASIR
In partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
in Electrical Engineering
Supervisor: Hassan Abbas Khan (LUMS)
Co-supervisor: Nauman Ahmad Zaffar (LUMS)
Syed Babar Ali School of Sciences and Engineering
Lahore University of Management Sciences (LUMS)
Lahore, Pakistan.
iii
Acknowledgement
The author expresses his gratitude, appreciation and sincere thanks
to his supervisor Dr. Hassan Abbas khan and co-supervisor Prof.
Nauman Ahmad Zaffar for their supervision of this research work.
Their advice, guidance and assistance, both technical and financial
in conducting this research and the preparation of this thesis are
thankfully acknowledged.
The author is extremely thankful to his foreign supervisor Prof.
Josep M. Guerrero and co-supervisor Juan C.Vasquez, who guided
and helped him conducting part of this research during his stay at
Aaalborg University (AAU), Denmark.
The author also thanks the advisory committee members including
Dr. Naveed-ul-Hassan, Dr. Jahangir Ikram for their suggestions
and advice. The author would also lie thank to external evaluator
Dr. Tauseef Tauqeer and internal evaluator Dr. Naveed Arshad for
their constructive remarks.
Assistance provided by the colleagues at EPS cluster, lab staff
including M. Farman Ali and Secretarial staff of the SBASSE is
thankfully acknowledged.
The author also acknowledges his parents, brothers and sister for
their prayers, constant encouragement and loving support
throughout his study tenure. Last but not the least; author would
like to thank his beloved wife for always being a constant source of
inspiration and support during this work.
iv
Abstract
Access to electricity is one of the key factors indicating the socio-economic status
of any community. Reliable and adequate provision of electricity is mandatory for
improved standards of living including better health, education, transport, agriculture and
employment opportunities. Unfortunately, according to International Energy Agency,
over 1.1 billion people around the world lack access to any electricity out of which 85
percent reside in rural areas of developing world. Electrification of these remote rural
communities through national grid interconnection is not economically feasible for many
developing countries due to high cost associated with the development of generation,
transmission and distribution infrastructure. Alternatively, DC microgrids implemented
with distributed generation and low voltage distribution are becoming very popular for
low cost rural electrification. However, current implementations are largely suboptimal
due to high distribution losses associated with their centralized architecture and their
inability to support high power community loads. In this work, a novel distributed DC
microgrid architecture which allows a scalable approach with minimal upfront investment
to fulfill rural electricity needs along with the provision of higher powers for communal
loads and beyond subsistence provisioning of electrical power is proposed. The
architecture is capable to work entirely on solar energy with power delivery capability to
individual consumers and added inherent ability to integrate resources to power up larger
loads for communal/commercial applications. The proposed microgrid architecture
consists of a cluster of multiple nanogrids (households), where each nanogrid has its own
PV generation and battery storage along with bi-directional connectivity to the microgrid.
Thus, each nanogrid can work independently in islanded mode along with the provision
of sharing its resources with the community through the bidirectional converter. In the
proposed architecture, the bi-directional power flow capability is implemented through a
modified flyback converter. A decentralized control methodology is also proposed to
ensure a communication-less, yet coordinated control among the distributed resources in
multiple nanogrids. The microgrid is evaluated for optimal distribution voltage level,
conductor size and interconnection scheme between nanogrids using Newton-Raphson
analysis modified for DC power flow. Various scenarios for power sharing among the
contributing nanogrids and communal load power allocation are analyzed from operation
and control prospective to validate the architecture and its performance. Further, an
optimal framework for the planning of distributed generation and storage resources in
each nanogrid with respect to time varying profiles of region-specific temperature and
irradiance is also presented to ensure the better resource utilization. A scaled version of
the proposed architecture is implemented on hardware, while the efficacy of control
methodology is validated on MATLAB/Simulink and hardware in loop facilities at
microgrid laboratory in Aalborg University. The proposed distributed architecture along
with decentralized control can be considered as a promising solution for the future rural
electrification implementations in developing regions.
v
List of Publications from thesis work
Journal Articles
[J1]. M. Nasir, H. A. Khan, A. Hussain, L. Mateen, and N. A. Zaffar, "Solar PV-Based
Scalable DC Microgrid for Rural Electrification in Developing Regions," IEEE
Transactions on Sustainable Energy, vol. 9, pp. 390-399, 2018.
http://ieeexplore.ieee.org/document/8002658/
[J2]. M. Nasir, S. Iqbal, and H. A. Khan, "Optimal Planning and Design of Low-Voltage
Low-Power Solar DC Microgrids," IEEE Transactions on Power Systems, vol. 33, issue
33, pp. 2919-2928, 2018.
http://ieeexplore.ieee.org/document/8052118/
[J3]. M. Nasir, Z. Jin, Hassan A. Khan, N. Zaffar, J.C. Vasquez, J. M. Guerrero, "A
Decentralized Control Architecture applied to DC Nanogrid Clusters for Rural
Electrification in Developing Regions." IEEE Transactions on Power Electronics, 2018.
https://ieeexplore.ieee.org/document/8341807/
[J4]. H. A. Khan, H. F. Ahmad, N. A. Zaffar, M. F. Nadeem and M. Nasir, "Decentralized
Electric Power Delivery Model for Rural Electrification in Pakistan, ", Under Revision,
Energy Policy.
Conference Proceedings
[C1]. M. Nasir, N. A. Zaffar, and H. A. Khan, "Analysis on central and distributed
architectures of solar powered DC microgrids," in Clemson University Power Systems
Conference (PSC), 2016, pp. 1-6.
http://ieeexplore.ieee.org/document/7462817/
[C2]. M. S. M. Hamza, S. Fazal, M. Nasir and H. Khan, "Design and Analysis of Solar PV
based Low-Power Low-Voltage DC Microgrid Architectures for Rural Electrification,"
IEEE PES General Meeting, Energizing a More Secure, Resiliant and Adaptable Grid,
Chicago, IL USA, 2017. http://ieeexplore.ieee.org/document/8274134/
[C3]. M. Nasir, H. A. Khan, Z. Jin amd J. M. Guerrero, "Dual–loop Control Strategy applied
to PV/battery based Islanded DC microgrids for Swarm Electrification of Developing
Regions," IET, Renewable Power Generation conference (RPG), Copenhagen, Denmark
2018.
Working Papers
M. Nasir, H. A. Khan, N. Zaffar, J.C. Vasquez, J. M. Guerrero, "Scalable Solar DC
Microgrids – A way forward to revolutionize the electrification architecture of
developing communities." IEEE Electrification Magazine
M. Nasir, S. Iqbal and H. A. Khan, "Optimal Pacement of Generation and Storage
Resources in Clustered DC Microgrids, ", Renewable Energy
vi
Contents 1 Motivations and Background ................................................................................................ 1
1.1. Need for Rural Electrification .......................................................................................... 1
1.2. Conventional Schemes for Rural Electrification.............................................................. 2
1.2.1. Electrification via Utility Extension and National Grid Interconnection ............. 3
1.2.2. Electrification via Standalone Solar Home Systems ............................................ 5
1.2.3. Electrification via Microgrids .............................................................................. 6
1.3. Need for Viable Microgrid Architectures for Rural Electrification ................................. 7
1.3.1. Suitable Generation Technology for Microgrid based Rural Electrification ....... 8
1.3.2. Suitable Mode of Distribution and Utilization for Microgrid Based Rural
Electrification ........................................................................................................................... 9
2 Architecture of Solar PV based DC Microgrid ................................................................. 11
2.1. Central Generation Central Storage Architecture .......................................................... 11
2.2. Central Generation Distributed Storage Architecture .................................................... 14
2.3. Other Distributed Architectures of DC Microgrid in Literature .................................... 14
2.4. Proposed Distributed Generation Distributed Storage Architecture .............................. 15
2.4.1. Model of Nanogrid ............................................................................................. 18
2.4.1.1. DC–DC MPPT Converter .................................................................................. 18
2.4.1.2. Bidirectional Flyback Converter ....................................................................... 19
2.4.2. Model of a Village and Microgrid Scheme of Interconnection ......................... 21
3 Power Flow Analysis of Solar PV based DC Microgrid ................................................... 23
3.1. Newton-Raphson Method Modified for Power Flow Analysis of DC Microgrid ......... 23
3.2. Case Study Parameters and Simulation Results ............................................................. 26
3.2.1. CGCSA without Communal load ...................................................................... 27
3.2.2. CGCSA with Communal load............................................................................ 28
3.2.3. DGDSA with Radial Scheme of Interconnection .............................................. 29
3.2.4. DGDSA with Ring-Main Scheme of Interconnection ....................................... 30
3.2.5. DGDSA with Communal Load and Ring-Main Scheme of Interconnection ..... 31
3.3. Summary of Power Flow Comparison between CGCSA and DGDSA ......................... 32
3.4. Power Flow Comparison between Radial and Ring - Main scheme of interconnection
for DGDSA of Microgrid........................................................................................................... 34
3.4.1. Typical Load Comparison between Radial and Ring- Main Scheme of DGDSA
34
3.4.2. Peak Load Comparison between Radial and Ring- Main Scheme of DGDSA.. 35
vii
3.5. Hardware Setup, Parameters and Results for Power flow Analysis of DC Microgrid .. 36
3.6. General Conclusions Drawn from Power Flow Analysis of DC Microgrids ................. 38
4 Decentralized Control of Solar PV- based DC Microgrid ................................................ 40
4.1. Need for a Decentralized and Communication-less Control Strategy for DC Microgrid
based Rural Electrification ......................................................................................................... 40
4.2. Hysteretic Voltage Droop Algorithm for Distributed Control of DGDSA .................... 41
4.3. Hardware Implementation Results for Hysteresis based Voltage Droop method For
DGDSA ...................................................................................................................................... 44
4.4. Limitations of Hysteresis based Voltage Droop Method and Need of a Robust Adaptive
Control ....................................................................................................................................... 46
4.5. Distributed Architecture and Control Objectives ........................................................... 48
4.6. Proposed Adaptive Decentralized Control Scheme ....................................................... 50
4.6.1. Power Stage and Control Scheme for an Individual Nanogrid .......................... 50
4.6.1.1. Multi-mode Adaptive Control Scheme for Bidirectional Converter Integrated
with DC bus............................................................................................................................ 51
4.6.1.2. Scheme for Switching between MPPT and Current Control Modes for the
Converter Integrated with PV Panel ...................................................................................... 54
4.7. Results, Discussions and Conclusions for Adaptive Decentralized Control .................. 55
4.7.1. Simulation Results for Decentralized Control ................................................... 55
4.7.1.1. All nanogrids are within specified thresholds of SOC ....................................... 55
4.7.1.2. All nanogrids are within specified thresholds of SOC except one which is below
minimum threshold of SOC .................................................................................................... 58
4.7.1.3. All nanogrids are within specified thresholds of SOC except one which is above
maximum threshold of SOC ................................................................................................... 60
4.7.1.4. Multi-mode switching of an individual nanogrid ............................................... 62
4.7.1.5. All nanogrids are above maximum threshold of SOC and surplus PV power is
available 63
4.7.2. Experimental Results for the Validation of Proposed Adaptive Algorithm for
Conv2i 66
4.7.2.1. All nanogrids are within specified thresholds of SOC ....................................... 69
4.7.2.2. All nanogrids are within specified thresholds of SOC except one which is above
maximum threshold of SOC ................................................................................................... 70
4.7.2.3. Multi-mode switching of an individual Nanogrid .............................................. 71
4.7.3. General Conclusions drawn from the results of Decentralized Control ............. 73
5 Optimal Planning and Design of Low-Voltage Low-Power Solar DC Microgrids ........ 74
5.1. Need for Optimal Planning and Design of Solar PV based DC Microgrids .................. 74
5.2. Common Village Orientations ....................................................................................... 75
viii
5.3. Proposed Distribution Architectures for Centralized Linear DC Microgrids ................ 77
5.3.1. Linearly Distributed C- Architecture ................................................................. 77
5.3.2. Linearly Distributed O- Architecture ................................................................. 78
5.4. Energy Balance Model for Centralized DC Microgrids ................................................ 78
5.5. System Model Formulation for Optimal Component Sizing of Centralized DC
Microgrids .................................................................................................................................. 81
5.6. Results and Discussion for Optimal Sizing of Centralized DC Microgrids ................... 84
5.7. Optimal Component Sizing for Distributed Generation and Distributed Storage
Architecture based DC Microgrids ............................................................................................ 94
5.8. Optimal Component Sizing Comparison between Centralized and DGDSA based DC
Microgrid ................................................................................................................................... 96
6 Conclusions and Future Work ............................................................................................ 99
6.1. Conclusions .................................................................................................................... 99
6.2. Potential Challenges in Practical Deployments of Decentralized Microgrids ............. 102
6.3. Cost and Affordability Evaluation of the Decentralized Microgrid System ................ 103
6.3.1. Survey Data .............................................................................................................. 103
6.3.2. Survey Results ......................................................................................................... 104
6.3.3. Cost Analysis Model for Economic Analysis .......................................................... 105
6.3.4. Results of Economic Analysis ................................................................................. 108
6.4. Future work .................................................................................................................. 110
ix
List of Figures
Figure 1. 1 Schematics for Electrification through National Grid Interconnection ......................... 4
Figure 1. 2 Schematic Diagram of a PV- based Solar Home system (SHS) .................................... 5
Figure 1.3 Schematic Diagram for Islanded Microgrid based Electrification ................................. 7
Figure 2. 1 Central Generation Central Storage Architecture (CGCSA) of PV based DC
Microgrid ....................................................................................................................................... 12
Figure 2. 2 Conceptual Diagram of Distributed Generation Distributed Storage Architecture
(DGDSA) of DC Microgrid with Contributing Nanogrids and Communal Load ......................... 17
Figure 2. 3 DC-DC boost Converter with MPPT/Current Control for Desired Voltage
Converrsion .................................................................................................................................... 18
Figure 2. 4 Modified Switch Realization of Flyback Converter enabling Bidirectional Power Flow
among the Contributing Nanogrids ................................................................................................ 20
Figure 2. 5 Proposed Architecture of the Radial Schemes of Interconnection with Elaborated
Single-Unit Design......................................................................................................................... 21
Figure 2. 6 Proposed Architecture of the Ring Main Schemes of Interconnection ....................... 22
Figure 3. 1 Power Network for Newton-Raphson method modified for DC Power Flow Analysis
....................................................................................................................................................... 24
Figure 3. 2 Typical % Voltage drop and Efficiency for CGCSA with Peak Load and Far End
Placement ....................................................................................................................................... 27
Figure 3. 3 Typical % Voltage Drop and Efficiency for CGCSA with Peak Load and Central
Placement ....................................................................................................................................... 28
Figure 3. 4 % Voltage drop and Efficiency for CGCSA with Communal Load and Central
Placement ....................................................................................................................................... 29
Figure 3. 5 Percentage Voltage Drop and Efficiency at Different Voltages and Different
Conductor Sizes for Typical Load sharing with Radial Scheme of Interconnection (Common
Load Sharing Radial DC Microgrid) ............................................................................................. 30
Figure 3. 6 Percentage Voltage Drop and Efficiency at Different Voltages and Different
Conductor Sizes for Peak Load sharing with Radial Scheme of Interconnection (Peak Load
Sharing Radial DC Microgrid) ....................................................................................................... 30
Figure 3. 7 Percentage Voltage Drop and Efficiency at Different Voltages and Different
Conductor Sizes for Peak Load sharing with Ring-Main Scheme of Interconnection. (Peak Load
Sharing Ring-Main DC Microgrid) ............................................................................................... 31
Figure 3. 8 Percentage Line Losses and Efficiency at Different Voltages and Different Conductor
Sizes for Communal Load Case with Ring-Main Scheme of Interconnection .............................. 32
Figure 3. 9 Hardware Implementation of Scaled down version for Power Flow Analysis ............ 36
Figure 3. 10 Measured v/s Simulated %Voltage Drops Results at 120V, 230V, 325V and 400V
for a) DGDSA and b) CGCSA .................................................................................................... 37
x
Figure 3. 11 Simulated v/s Measured Results for Normalized Line Losses in DGDSA with Radial
and Ring-Main Schemes of Interconnection .................................................................................. 37
Figure 4. 1 Hysteretic-based Distributed Voltage Droop Control Algorithm ................................ 44
Figure 4. 2 Implementation of the DGDSA Microgrid Hardware through the Integrations of
Nanogrids ....................................................................................................................................... 45
Figure 4. 3 Results of hardware implementation of typical voltage variations of the microgrid in
various power sharing scenarios. ................................................................................................... 46
Figure 4. 4 A Cluster of Multiple Nanogrids Interconnected via DC Bus Formulating the DGDSA
of PV/battery based DC Microgrid ................................................................................................ 48
Figure 4. 5 Power Electronic Interface and Control Schemes for Converters Employed in An
Individual Nanogrid Unit ............................................................................................................... 51
Figure 4. 6 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L,
I2L, I3
L and I4
L) (left Y -axis) in case 1 (simulation results) ............................................................ 56
Figure 4. 7 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 1 (simulation results) .................................. 57
Figure 4. 8 DC DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids
(I1L, I2
L, I3
L and I4
L) (left Y-axis) in case 2 (simulation results) ...................................................... 59
Figure 4. 9 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 1 (simulation results) .................................. 59
Figure 4. 10 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L,
I2L, I3
L and I4
L) (left Y-axis) in case 2 (simulation results) ............................................................. 61
Figure 4. 11 DC bus voltage VB profile (right Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 2 (simulation results) .................................. 61
Figure 4. 12 Nanogrid 1 SOC1 variation in the various thresholds ranges (left Y-axis) and
associated current sharing among the contributing nanogrids in case 3 (right Y- axis) (simulation
results) ............................................................................................................................................ 63
Figure 4. 13 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L,
I2L, I3
L and I4
L) (left Y-axis) in case 4 (simulation results) ............................................................. 64
Figure 4. 14 Power generated by PV panels in nanogrid 1 P1PV
(righy Y-axis) and output current
I1in of conv21(left Y-axis) in case 5 (simulation results) ............................................................... 64
Figure 4. 15 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 4 (simulation results) .................................. 65
Figure 4. 16 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 4 (simulation results) .................................. 66
Figure 4. 17 Schematics of experimental setup at microgrid laboratory ...................................... 68
Figure 4. 18 Hardware setup for practical measurements .............................................................. 68
Figure 4. 19 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L,
I2Land I3
L ) (left Y-axis) in case 1 (measured results) ................................................................... 69
Figure 4. 20 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 1 (measured results) .................................... 70
Figure 4. 21 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L,
I2Land I3
L ) (left Y-axis) in case 2 (measured results) ................................................................... 70
xi
Figure 4. 22 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 2 (measured results) .................................... 71
Figure 4. 23 Nanogrid 1 SOC1 variations in the various threshold ranges (left Y-axis) and
associated current sharing among the contributing nanogrids (right Y- axis) in case 3 (measured
results) ............................................................................................................................................ 72
Figure 5. 1 Topological Diagram of Linearly Distributed C-architecture with PV Generation and
Power Processing and Storage Units (PPSU) ................................................................................ 77
Figure 5. 2 Topological Diagram of Linearly Distributed O-architecture with two (PPSU‘s) ...... 78
Figure 5. 3 System Diagram for Energy Flow in C-architecture with N Houses ........................... 81
Figure 5.4 System Diagram for Energy Flow in O-architecture with N Houses ........................... 82
Figure 5. 5 Distribution Efficiency, η (right y-axis and Worst Voltage Dip, VD (left y-axis) for C-
architecture and O-architecture with 5W Loading at Different Gauge Sizes, and Different Voltage
Levels. ............................................................................................................................................ 87
Figure 5. 6 Optimal Selection of Conductor Size with 5W Power Provisions at Different Voltage
Levels and Distribution Architectures. .......................................................................................... 87
Figure 5. 7 Optimal PV Panel Sizing of the System at 5W Loading (Case 1). Kindly, note that
each data point on the figure represents a region shown in table 1. For instance, the first data point
(from left) is for Bihar where PSH is 4.992 and so on. (Followed in all subsequent figures) ....... 88
Figure 5. 8 Optimal Battery Sizing of the System at 5W Loading on left Y- axis and Irradiance
Volatility Factor on Right Y- axis (Case 1) ................................................................................... 89
Figure 5. 9 Optimal Installation Cost of the System at 5W Loading (Case 1) .............................. 89
Figure 5. 10 Distribution Efficiency ηD and Worst Voltage Dip VD for C-architecture and O-
architecture with 10W Loading at Different Gauge Sizes, and Different Voltage Levels ............. 91
Figure 5. 11 Optimal Selection of Conductor Size with 10 W Power Provisions based upon the
Relative Cost of Distribution at Different Voltage Levels and Distribution Architectures ........... 92
Figure 5. 12 Optimal Installation Cost of the System (case 2) ...................................................... 93
Figure 5. 13 Optimal Sizing for 24V, 5W, O- Configuration Represented on Map ...................... 93
Figure 5. 14 Optimal PV Panel Sizing for N houses in DGDSA at 10W Loading ........................ 95
Figure 5. 15 Optimal Battery Sizing for N houses in DGDSA at 10W Loading ........................... 96
Figure 5. 16 Optimal System Cost for N houses in DGDSA at 10W Loading .............................. 96
Figure 5. 17 Comparative Results of Optimal PV Sizing for Centralized and DGDSA based DC
Microgrid ....................................................................................................................................... 97
Figure 5. 18 Comparative Results of Optimal Battery Sizing for Centralized and DGDSA based
DC Microgrid ................................................................................................................................. 97
Figure 5. 19 Comparative Results of Optimal PV Sizing for Centralized and DGDSA based DC
Microgrid ....................................................................................................................................... 98
Figure 6. 1 Life Time Operation Cost Break-up of a Distributed Microgrid ............................... 110
xii
List of Tables
Table 1. 1 Detailed Comparisons between AC and DC Microgrids .............................................. 10
Table 3. 1 Communal Load Comparison between CGCSA and DGDSA ..................................... 33
Table 3. 2 Typical Load Comparison between Radial and Ring-Main DC Microgrid ................. 34
Table 3. 3 Peak Load Comparison between Radial and Ring-Main DC Microgrid ..................... 35
Table 4. 1 Parameters of Simulated Case Study ........................................................................... 58
Table 4. 2 Parameters of Experimental Case Study ....................................................................... 67
Table 5.1 Specific Regions for Analysis with Irradiance Profiles ................................................... 88
Table 6. 1 Services offered and their prices by plan .................................................................... 104
Table 6. 2 Average Prices Willing to be Paid for Each Level of Service .................................... 104
Table 6. 3 Relative and Absolute Differences in Reported Prices with Respect to Base Plan
(Lights Only) Standard Error in Parenthesis ................................................................................ 105
Table 6. 4 Estimated Cost of decentralized Solar Generation Implementations through DGDSA
..................................................................................................................................................... 109
1
1 Motivations and Background
This chapter discusses the motivations behind addressing the critical problem of
rural electrification and the background work that has been done so far on this
challenging issue. The chapter sequentially emphasizes the need for rural electrification
by highlighting the social and economic impacts associated with the access to electricity.
The role of technological advancements leading towards the development of a sustainable
solution that is economically feasible and efficient in operation is discussed. The chapter
ends with the highlights of shortcomings in the existing systems for rural electrification
and possible room for improvement from architecture, operation and control prospective.
1.1. Need for Rural Electrification
―Access to energy is absolutely fundamental in the struggle against poverty‖
-Rachel Kyte, World Bank Vice President (2013).
Electricity is one of the most impactful forms of energy that has revolutionized
the society. All major technological advancements in modern society can be attributed to
electricity. From our daily lives to industries and from agricultural fields to offices, the
stimulating role of electricity is undeniable. Reliable access to electricity and its
consumption rates are therefore considered as the key indexes for the socio-economic
status of any community. The significant availability of electricity, even at very basic
levels, is extremely crucial for human well-being and social resources development.
Unavailability of electricity hampers basic human rights like access to clean water, health
care units and schooling facilities, therefore, severely affects the quality of life and results
in higher poverty levels.
Unfortunately, over 1.1 billion people throughout the world that constitute nearly
16% of the global population lack access to electricity [1]. These mainly include
approximately 634 million in Africa, 512 million in developing Asia, 22 million in Latin
America and 18 million in the Middle East. It is also estimated that around 85% of the
people lacking access of electricity are the residents of rural areas [1]. If we analyze the
case of Pakistan in particular, International Energy Agency (IEA) statistics of 2017
2
shows over 51 million people without any access to electricity [1]. Therefore, the main
focus of this work is to come up with a viable and sustainable solution for rural
electrification in Pakistan and beyond that can play a key role in alleviating poverty.
The inhabitants of un-electrified regions are deprived of the basic facilities of
electricity driven heating, air-conditioning and water supply systems. In most of these
under-privileged areas, schooling and health care facilities are virtually absent due to
non-availability of electricity. They have to largely rely on unhealthy resources, like
wood and biomass for cooking purposes. According to a report by National Geographic,
cook stove smoke is extremely life threatening and around 3.5 million people die each
year due to the respiratory diseases caused by indoor pollution of wood/biomass based
stoves (approximately three times of mortality rate caused by malaria and 2.3 times of
mortality rate caused by HIV/AIDS [2]. Alternatively, kerosene oil is being largely used
by the inhabitants of these under-privileged areas for cooking and even for lighting
purposes, which also has many documented adverse effects on individuals as well as on
environment [3]. The substantial provision of electricity to these inhabitants can not only
reduce alarming fatality rates but can also contribute for improved standards of living
including better health, education, agricultural, industrial and employment opportunities
[4, 5]. In addition, electrification of these regions through green and environment-friendly
energy resources will help in reducing climate change and deforestation rates [6].
Along with social benefits, there are remarkable business opportunities in the
energy markets of these developing regions due to global focus on energy poverty
eradication and associated initiatives, e.g. sustainable energy for all (SE4ALL), and
‗Lightning Africa‘ [7]. Since, human development, economic stability, and social growth
of these regions is coupled with the access to electricity, therefore, rural electrification is
the need of the hour to attain the socio-economic benefits associated with the easy access
and reliable availability of electricity.
1.2. Conventional Schemes for Rural Electrification
There is a worldwide focus on electrification of developing rural areas as evident
by United Nations (UN) sustainable development goals (SDG). In particular, SDG- 7
aims to ensure universal access to affordable, reliable, sustainable and modern energy
3
services for all by 2030 [1, 8, 9]. As a result of these efforts, over 1 billion people have
been given access to electricity since 2000 among which around 220 million were
provided access to electricity between 2010 to 2012 [8, 9]. These efforts are more
pronounced in developing Asia, where around 870 million people have gained access to
electricity since 2000. Also, for the very first time in recent years, electrification rates in
Africa have become on par with the growing population [8, 9]. The main source of
electrification during this all course has been the extension of utility and national grid
interconnection of these remote villages with large dependence on fossil fuel. However,
with the constant depletion of fossil fuel, increasing awareness about their hazardous
impact on environment and rapidly decreasing prices of renewable energy technologies,
there is a paradigm shift towards the adoption of environment friendly renewable energy
resources for off-grid rural electrification. Over the last five years, a considerable trend
has been seen towards renewable based decentralized rural electrification and
approximately 6% of the new access connections are based upon renewable energy
resources [9].
Although these efforts are pronounced, however, due to consistent increase in
population, and limited potential of conventional electrification schemes, today, the
number of people without electricity are more than what it were in 2000 [8, 9]. Existing
schemes are not sufficiently capable to achieve the global objective of universal energy
access. A proportionate increase in electrification rate can be achieved only with properly
planned financing and policy commitments along with adoption of technologically
advanced sustainable technologies on a broader scale. Moreover, innovative business
models to finance the energy access need to be adopted for the considerable growth of
rural electrification in the coming years. A detailed overview of the conventional
electrification schemes adopted by developing countries, along with their pros and cons is
necessary for the understanding of their limited potential for the growing needs of rural
electrification.
1.2.1. Electrification via Utility Extension and National Grid Interconnection
Predominantly, electrification via laying three phase transmission lines and
interconnection with national grid has been widely adopted for electrification. According
4
to a report by international energy agency, world energy outlook (IEA, WEO), 70 percent
of the new electricity connections, during 2000-2016, were provided through grid
expansion [9]. As a standout example of an emerging economy, China, has given
electricity access to 900 million people from 1949 till date in two phases. In the first
phase, 97% population was given access to electricity till the end of 20th
century, out of
which four-fifth of the rural population was electrified through grid extension [10]. The
schematic for grid extension- based electrification is shown in figure 1.1.
Electrification via centralized generation and subsequent transmission and
distribution involves efficient transformation of voltages from one level to another,
allowing power to be carried long distances at high voltages. At destinations, the
electricity is then converted to the low voltages appropriate for use in homes and
businesses [11]. This technology is relevantly mature, therefore, remained as the main
choice for rural electrification for years. However, limited funds to construct new large
power plants and high cost of long distance transmission lines (over 1 million USD/km
[12]) are some of the constraints on developing economies to meet the ever growing
energy demands for remotely located people without access to the grid-electricity.
Moreover, the conventional source of fossil fuel in this central generation based grid
expansions causes carbon emissions and is hazardous for environment. The transmission
of power over long distance lines causes considerable amount of energy losses and power
quality significantly deteriorates due to inductive and capacitive natures of transmission
lines [13]. Looking over these limitations, central generation based grid expansions may
not be the most optimized choice for rural electrification of developing regions.
Figure 1. 1 Schematics for Electrification through National Grid Interconnection
Conventional Central AC Generation
Power Transformer
Distribution Transformer
Distribution Line
Electrified Village
25-1000 km
House 1 House N
1-10 km
5
1.2.2. Electrification via Standalone Solar Home Systems
Rural electrification via grid expansion requires the deployments of mega projects
including building new power plants and long distance transmission lines. For developing
and under-developed economies, these large scale developments are generally
constrained by the limitation of funding resources. Alternatively, various standalone solar
home systems (SHS) have been incorporated as a stop-gap measure to provide rural
residents with basic electricity in the last decade [14, 15]. These systems generally
provide between a few watts to a few tens of watts enough to run one or two LED‘s along
with a mobile charging unit for an average rural house. The schematic diagram of a PV
based SHS is shown in figure 1.2. As a standout example, in Bangladesh alone, 3 million
SHS were installed by 2014 and this is growing day by day [16]. Infrastructure
development company (IDCOL) by government of Bangladesh has reported the
installation of 4.12 million SHS in the remote areas up to May, 2017 through which 18
million people i.e. 12% of total population has been given access to electricity. The
projected target of IDCOL is to install 6 million SHS by 2021 [17]. The SHS technology
is cost effective and relatively easy to deploy in comparison to grid extension alternative,
however, these standalone solutions are suboptimal, as without resource sharing, they do
not take advantage of electricity usage diversity at a village scale. Moreover, they have
limited electrification capabilities and are not feasible for something demanding like,
water filtration plants/ irrigation pumps, school computing loads or health care units for a
village. Therefore, such schemes cannot provide electricity beyond subsistence level
living and cannot contribute for the significant improvement in terms of quality of life.
Figure 1. 2 Schematic Diagram of a PV- based Solar Home system (SHS)
Roof Mounted PV Panel
SOLAR HOME SYSTEM(SHS)
Household LoadCharge Controller
Battery
6
1.2.3. Electrification via Microgrids
Although SHS provides a low upfront cost and relatively simpler off grid
electrification solution, there are several limitations to this approach. It cannot support
larger loads due to prohibitively large solar panels and storage requirement for rural
occupants in the developing regions. Even with the smaller systems, the LCOE is
generally high due to lack of resource sharing capabilities. Alternatively,
Wind/solar/fossil fuel based islanded microgrids are becoming very popular for rural
electrification of developing regions due to their ability to support electrification beyond
substance level living [18]. Based upon the type of generation and distribution microgrids
may be classified as renewable/non-renewable AC/DC microgrids. The schematic
diagram of an islanded microgrid based electrification solution is shown in figure 1.3.
According to a report by United Nation Foundation, 760 MW of microgrid
capacity has been installed for rural electrification till 2013 and this capacity is increasing
day by day [19]. A very successful commercial scale PV-based islanded microgrid is the
Mera Gao Power (MGP) in India that involves central PV generation and central battery
storage with distribution at 24V DC to subscribing houses. The subscribers of MGP may
consume up to 5W of DC electricity (enough to power an LED light and a mobile-phone
charging point). It is reported that MGP has over 0.1 million subscribing households
spread across 400 villages [20, 21]. However, it is only limited to provide electricity for 8
hours in a day. Moreover, the implementation is incapable to support high power
community load. This and other similar implementations [22, 23] use centralized
generation and storage and therefore have considerable distribution losses due to
concentrated generation and storage at a village scale. Moreover, such systems are not
scalable and have to be designed according to the peak power requirements, leading
towards lower utilization factor due to usage diversity. While, these microgrids have been
piloted in several regions [11, 16, 23, 24], the viability/sustainability of such large
systems is not feasible due to limited paying capacity for many rural occupants in
developing regions. This generally results in large subsidies given by government or
donor agencies to make these viable.
7
Figure 1.3 Schematic Diagram for Islanded Microgrid based Electrification
These limitations of conventional microgrid based electrification schemes can be
substantially overcome by the optimal selection of microgrid architecture and generation
technology.
1.3. Need for Viable Microgrid Architectures for Rural Electrification
Although the conventional schemes for rural electrification are being largely
deployed as a stop gap measure for energy poverty eradication, however, owing to their
limited potential, such schemes are not sufficient for wide scale deployment to achieve
the global objectives of SDG-7. With the growing population and associated
electrification requirements, there is the need of a robust, technologically advanced,
economically feasible, financially viable and widely adoptable electrification solution
that can grow in a bottom-up manner and support micro-financing for enhanced rates of
electrification. In order to advance towards more efficient topologies for RE, it is
important to review the recent technological advancements in microgrids which will lead
towards a more suitable candidate for future rural electrification deployments. Two major
aspects i.e. a) generation technology (renewable or non-renewable), b) type of
distribution (AC or DC) are critical in this regard and discussed in the sub-sections
below. Another critical factor is the architecture of DC microgrid with regards to
generation and storage resources placement. Concentrated placement of resources in
existing deployments generally limits the electrification capabilities from distribution loss
and scalability prospective and is discussed further in the next chapter.
Centralized Renewable/non-renewable Generation, Battery Storage and Charge Controller Electrified Village
8
1.3.1. Suitable Generation Technology for Microgrid based Rural Electrification
A microgrid can be either AC or DC and is considered as a highly reliable
medium of generation and distribution of electrical energy by connecting distributed
generation (DG) units and critical loads in close proximity [25-30]. Microgrids may have
various renewable and non-renewable sources such as fuel cells, photovoltaic systems,
small diesel generators, wind turbines, and micro-turbines [25, 30-32]. Conventional
resources of generation including fossil fuel based generation in particular diesel based
generation systems result in carbon emission and are not considered as an attractive
solution for electrification due to their adverse effects on environment. Moreover, the
levelized cost of electricity and operation cost for such diesel based electricity generation
systems are higher and unviable for low-income communities.
Over last two decades, the renewable and alternate energy technologies has
gained world-wide interest as an effective alternative to reduce the dependence on fossil
fuels and to avoid their adverse effects on climate change [33, 34]. Therefore, renewable
energy resources, in particular wind and solar energy generation are being largely
adopted by microgrid practitioners due to their green and environment friendly nature
[35-37]. Among all other renewable technologies, installations based upon solar energy
extraction using Photovoltaic (PV) systems are more successful due to natural availability
of sunlight, relatively simpler schemes of installation, environment-friendly nature and
noise-free operation [38, 39]. The consistent reduction in PV panel prices, Feed-in-Tarrifs
(FiT) and favorable governmental policies to incorporate renewable energy resources
have also encouraged domestic consumer to invest in this technology to contribute
towards sustainable electricity generation. Therefore, due to green nature, abundant
availability of solar energy in most of the non-electrified areas (above 5.5 kWhr/m2/day
for most of the regions) [14, 15], and constantly diminishing panel prices, solar PV
microgrids are highly suitable for remote areas electrification [24]. Also, battery
technology has become mature and allowing deeper discharges and longer life at a
lowering cost [40], therefore, PV/battery based microgrids can be considered as optimal
choice for future electrification projects.
9
1.3.2. Suitable Mode of Distribution and Utilization for Microgrid Based Rural
Electrification
Depending upon the mode of generation, distribution and utilization, microgrids
may be classified either as AC microgrids or DC microgrids. Several researches between
the comparison of AC and DC microgrids has been presented in literature [41-43]. Due to
their inherent simplicity, higher power quality, enhanced efficiency and straight forward
controllability, DC microgrids are preferred over AC microgrids for rural electrification
applications [41-47]. A detailed tabular comparison between the design, implementation
and operational characteristics of AC and DC microgrid is shown in Table 1.
Solar photovoltaic (PV) produces DC, batteries store DC and most modern loads
are now DC, which allows local power generation and distribution through DC
microgrids with source closely matching the load profile. Compared to traditional AC
distribution, DC microgrids are significantly more efficient due to no DC-AC or AC-DC
conversion when implemented with distributed generation (DG). These systems have
end-to-end efficiency of around 80% (for DC loads) compared to AC microgrids which
are less than 60% efficient [42, 48]. Along with higher efficiency, DC microgrids and
associated distribution has the inherent advantage of less conductor usage for distributing
the same amount of peak power in comparison to AC distribution. Therefore, cost
associated with distribution conductors can be substantially reduced using DC
distribution [49]. Also, DC distribution is more resilient from power quality issues and its
reliability is relatively higher in comparison to AC distribution [50]. These all factors
make PV/Battery based DC microgrids as an optimal choice for rural electrification
applications.
PV/battery based DC microgrids can be further optimized from architecture
prospective and this is further highlighted in next chapter.
10
Table 1. 1 Detailed Comparisons between AC and DC Microgrids
Factors Sub Factors DC Microgrid AC Microgrid
Energy
Efficiency
Distribution losses Lower [41, 51] Higher [41, 51]
Conversion losses Lower due to inherent DC
nature of the loads [52, 53]
Higher due to AC/DC
conversions [52, 53]
Overall Efficiency 85-77% for DC loads,
<63% for AC loads [54].
< 60% with DC
generation [54].
Conductor
Usage
Power carrying capability
Higher for the same
conductor thickness [49].
Lower for the same
conductor thickness [49].
Cu mass usage 33% higher than AC [49]. 33% lower than DC [49].
Cost
Conductor cost Lower [49]. Higher [49].
Protection cost Higher due to electronic
relays [55].
Lower due to mechanical
switches/ relays [55, 56].
Protection Response time Higher [56]. Lower [56].
Cost Higher due to non-
availability of zero crossing
Lower due to zero
crossing availability [57].
Reliability
Critical load handling
capability
Higher (data centres and
UPS systems ) [50].
Lower [50].
Pulsed load handling
capability
Higher [47, 58]. Lower [47, 58].
generator synchronization Simple [59-61]. Complex [59-61].
Power
Quality
Transient stability Higher [47, 61]. Lower [47, 61].
Power quality to loads Better [42, 61]. Lower [42, 61].
Data
Analysis and
computation
Planning, operation and
Control studies
Simple due to involvement
of real numbers [42].
Complex due to
imaginary numbers [42].
11
2 Architecture of Solar PV based DC
Microgrid
As discussed in chapter 1, solar PV based microgrids provide an efficient and
potentially cost-effective rural electrification solution; however, there are several issues
which must be addressed for their widespread deployments. The limited electrification
capabilities of existing PV based DC microgrid architectures in terms of their low
distribution efficiency, inability to handle large power communal loads, non-modularity
in their structure and limited scalability are elaborated in this chapter. Based upon the
highlighted limitations, a scalable DC microgrid architecture having inherent advantages
of higher efficiency, modular scalability, simplified communication-less control and
efficient aggregation of power for larger household or communal loads is detailed for
future electrification implementations.
2.1. Central Generation Central Storage Architecture
Figure 2.1 shows the topological diagram of a typical centralized DC microgrid
architecture used for many rural electrification implementations. Such an architecture in
which generation (PV panels) and storage (batteries) are placed on a central location is
referred as central generation central storage architecture (CGCSA). CGCSA has a
unidirectional flow of power from a central location with solar PV generation and storage
to households. The load for community center consisting of school computing or health
care unit may also be powered from the central distribution line, however, generation
capacity has to be designed as per over all peak load requirements. A single DC-DC boost
converter is required for maximum power point tracking (MPPT) of PV panels and
stepping up the voltage to microgrid distribution voltage level. At the consumer end
another DC-DC converter is required to step down the microgrid voltage level to
household devices level.
12
Figure 2. 1 Central Generation Central Storage Architecture (CGCSA) of PV based DC Microgrid
Prominent practical implementations for rural electrification through CGCSA of
PV/battery based DC microgrids include micro-solar plants in Chhattisgarh, by
Chhattisgarh renewable energy development agency (CREDA) in India [22, 23].
CREDA has deployed 576 PV based DC microgrids with cumulative capacity of
2.15MW, serving around 31000 customers in remote areas [19]. MGP discussed in the
previous chapter, subscribed by 0.1 million consumers, is also based upon CGCSA [20,
23]. Similarly, in 2012, Uttar Pradesh and Renewable Energy Development Agency
(UPNEDA), installed 1 kW DC microgrids in 11 districts covering around 4,000 houses
[62]. The Jabula project in Cape Town, South Africa is another successful model, where
Zonke Energy installed a PV/battery based DC micro-grid (750WP) to serves nine
families residing in informal settings with basic electricity [58].
In all the above mentioned practical deployments, centralized architecture of
PV/battery based DC microgrid is being used. This energy is then delivered to
subscribing households via distribution conductors and therefore, distribution losses are
associated with the delivery of energy. The distribution losses in this architecture depend
upon the distribution voltage level and size of mass produced conductor used for
distribution. Generally, line losses reduce at higher distribution voltages and the wider
PV
PanelsDC- DC Converter
Battery
StorageDC- DC Converter
House 1
LoadDC-DC
Converter
House 2
LoadDC-DC
Converter
House n
LoadDC-DC
Converter
Communal
LoadDC-DC
Converter
DC- DC
Converter
Central Generation Central Storage Point
13
area conductor size, while the system exhibit lower efficiency and higher line losses at
lower distribution voltage and lower conductor area used for distribution. The central
positioning of the resources is generally beneficial from the perspective of control, where
overall generation and storage level (state of charge) are reliably monitored. However,
this results in higher distribution losses and rigidity in terms of future expansions [15].
Moreover, powering a high power communal load will substantially enhance the
distribution losses in the path of power flow.
Another draw-back associated with CGCSA is that its generation and storage
capacity has to be designed as per peak power requirements of the load, thereby
increasing the upfront capital cost of installation. In such a topology, the advantage of
usage diversity cannot be extracted. For instance power provisioning to high power
communal loads including water filtration plant, computing load of a school or load of
medical equipment in a health care unit results in a substantial increase in the required
capacity and associated cost of the installation. Moreover, at the day time, when there is
enough production by the PV panels and lighting load requirement at houses is
comparatively negligible, the excessive power generated by the PV panels cannot be
utilized optimally after the storage system has fully charged. Thereby, excessive power
will be wasted within panels making the overall scheme essentially sub-optimal in terms
of resource utilization.
Considering the example of ―Mera Gao Power‖ (MGP) in India, which provides
only 5 W of DC power to each subscribing house, with a limit of 0.2 amps—enough to
power two LED lights and a mobile phone-charging point [20, 23]. Although ‗small
power‘ is beautiful, it is unable to drive high power community loads [63]. Due to very
limited power supply, such a scheme is unlikely to alleviate poverty in rural areas or
contribute to significant improvements in their socio-economic circumstances [16]. If
such a central generation central storage architecture (CGCSA) is implemented for high
power loads for households (50 W or higher), the losses associated with the distribution
of energy are significantly higher, thereby making the scheme unviable. The framework
for detailed loss analysis based upon Newton – Raphson method modified for DC power
flow will be presented in the Chapter 3.
14
2.2. Central Generation Distributed Storage Architecture
Madduri et al. [54, 64] presented a central generation distributed storage
architecture for rural electrification. It has been shown that by distributing the storage
system at individual consumer nodes will result in reduced distribution losses, while the
distributed power may be intelligently stored or consumed at the load end using
household power management units (PMU). The provision of energy storage at local
houses results in higher efficiency compared to CGCSA, but this architecture is still
suboptimal in two respects: a) central PV generation requires a higher upfront cost
because a large nameplate capacity is required for the solar panel at the outset, resulting
in a cost barrier; and b) distribution to distant houses causes significant system losses.
Moreover, the presented architecture uses central PV generation and is unable to pool for
communal loads effectively without resource sharing capability. Therefore, architecture
with minimum possible distribution losses capable to integrate its resources in a scalable
manner is required to have an optimal solution for impactful rural electrification.
2.3. Other Distributed Architectures of DC Microgrid in Literature
Another PV based ad-hoc partially distributed DC microgrid architecture for rural
electrification proposed by Wardah et al. [65] integrates the power needed for several
consumers (up to 20) into a single generator unit. However, the overall distribution is at
48 V, which renders it impractical for the requirements of larger households or a
community level load due to higher distribution losses. Moreover, in this architecture,
peer to peer electricity sharing was enabled by GSM based communication between
power management units (PMU) of generating modules (houses having PV generation,
battery storage and local load) and consuming modules (houses having only local loads
without any generation or storage facilities). The advantages of distributed architecture
are mainly reduction in distribution losses and modularity in structure. However,
coordination among the distributed resources and control for power sharing becomes
extremely challenging. Several strategies for hierarchical and supervisory and droop
control of DC microgrids have been proposed in [8, 66-70]. However, these require an
extra layer of monitoring, sensing and communication, which in turn enhances the cost
and complexity of the system. For rural electrification purposes, such a complex, high
15
cost and communication based distributed architectures is not preferable due to
constraints of limited funding.
What is needed instead is an architecture that can scale in power for individual
homes and also support large (kilowatt level) loads, such as water pumps and
refrigeration units, for communal use. The architecture should be scalable and built
through bottom-up approach so that it can enable micro-financing for successful business
model as well as wide adaptability. The architecture must have (a) resource sharing
capability and (b) potential of higher powers even with limited roof-top PV production
and (c) ability to fulfill the community load demands. Such systems can in principle work
as a primary grid or in parallel with the surrounding AC grid. In this thesis, one such PV
based DC microgrid system is proposed that allows a scalable approach with minimal
upfront investment to run the electricity needs along with the provision of higher powers
for a communal load.
Therefore, for rural electrification purposes, architecture having low cost
deployment lower distribution and conversion losses, higher end to end efficiency, high
reliability, scalability along with modularity in structure, resource sharing feature and
capability to drive high power communal load is highly desirable. Architecture with these
characteristics is a true rural electrification architecture that can provide beyond
subsistence level power provisioning and can genuinely contribute towards the socio-
economic uplift of the society and an architecture with these characteristics can
substantially enhance the electrification rates to achieve the global objectives of SDG-7.
2.4. Proposed Distributed Generation Distributed Storage Architecture
In light of the limitations on (a) distribution efficiency, (b) power supply for
household-level loads, (c) provision of communal loads, (d) rigidity in future expansion,
and (e) requirements for extensive communication based control techniques in the above-
mentioned architectures for a DC microgrid, we propose a solar PV based scalable,
distributed generation and distributed storage architecture (DGDSA) with a novel
resource (power)-sharing provision among the distributed resources (see Fig. 2.2). The
architecture has the built-in advantages of (a) higher efficiency because of distributed
generation and distributed storage, (b) modular scalability for future expansion, (c)
16
efficient aggregation of power for larger loads even with limited roof-top PV, (d) delivery
to communal entities as rural schools and basic health units by pooling power from
individual household units without dedicated (large) generation, (e) reliable and
simplified control through the hysteresis based voltage droop method (implemented
through a localized controller without the need for central, adaptive, or supervisory
control, and a reduction in extensive communication requirements. Furthermore, the
distributed nature of the proposed DGDSA makes it independently scalable in its
planning and operation [71].
To the best of our knowledge, none of the existing architectures provides this
level of scalability because the requirements of these architectures for upfront generation
are generally higher. For instance, the architecture developed by Madduri et al. [54, 64]
features central generation, which requires a relatively larger generation capacity at the
outset for village-scale electrification. Similarly, the architecture proposed by Wardah et
al. [65] has cluster-/neighborhood-level distributed generation to cater to the needs of
multiple houses and, therefore, requires a larger number of subscribing houses for the
system to be viable. Other architectures [20, 23] also use central generation requiring a
certain minimum number of subscribers for proper utilization. On the contrary, in our
proposed architecture, even a single system installed (panel, power processing unit, and
battery) at a house can initiate the operation of the DGDSA based DC microgrid, which
can then grow in a scalable manner as neighbors are added to the grid.
The electrical energy architecture proposed in this thesis is a type of microgrid
(figure. 2.2) – a small interconnected self-sustaining electrical generation, distribution
and utilization system. A household is referred as a nanogrid which is the basic building
block for the entire system. A primary task is to establish an efficient mechanism via DC
distributed generation (DG), to channel excess energy between connected nodes.
17
Figure 2. 2 Conceptual Diagram of Distributed Generation Distributed Storage Architecture
(DGDSA) of DC Microgrid with Contributing Nanogrids and Communal Load
For instance, in figure 2.2, PV panels at rooftop (home ‗a‘) should be able to
provide energy to a neighboring house (home ‗b‘) if the PV produced power is not being
utilized in home ‗a‘ and vice versa. Similarly, if both home ‗a‘, home ‗b‘ up to home ‗n‘
have surplus power then there must be a mechanism to allow this power to be utilized at
communal loads. Inherently, PV panels continuously provide power during the presence
of sunlight and if not utilized properly, this power is wasted within the panel. Therefore,
in a communal setting, the surplus DC power produced by all the panels must also be
utilized by communal loads such as water pumping for drinking/irrigation, medical
equipment in basic health units, lighting and computing loads in school etc. Such large
communal loads otherwise (standalone basis) are often very expensive and unsustainable
in rural scenarios of developing countries. Although solar energy is taken as the primary
source however other sources both renewable and non-renewable can be integrated at a
common coupling point. In addition, integration with utility grid could also be possible
allowing bidirectional flow of power. However, integration of utility AC grid or other
sources is out of scope for the current implementation as this work primarily focuses on
the implementation and operation of the microgrid itself in off-grid communities.
18
2.4.1. Model of Nanogrid
A nanogrid is a basic building block and integral part of DGDSA of microgrid
that integrates its resources in a scalable manner into the community. Each
house/nanogrid has its own generation in the form of a roof-mounted solar PV panel, its
own battery storage and a few DC loads. Therefore, each nanogrid has the capability to
work independently in islanded mode or in conjunction with the other nanogrids in the
architecture in power sharing mode. The bidirectional flow of power is controlled via
power electronic converters referred to as central power processing units (CPPUs). A
CPPU contains a microcontroller along with a maximum power point tracking (MPPT)
based DC-DC converter and a bidirectional flyback converter.
2.4.1.1. DC–DC MPPT Converter
The output power of a PV panel is a non-linear function of temperature and
incident irradiance [72]. MPPT techniques are employed to extract the maximum power
from the available solar energy. Various schemes for MPPT under uniform and non-
uniform irradiance have been discussed in the literature [73, 74]. In this article, the
perturb and observe (P & O) algorithm is employed due to its simplicity and low
computational complexity [73]. The conversion ratio of the DC-to-DC converter is
adjusted such that its output voltage is suitable for supplying power to the load and
charging the battery. Based on the time-varying values of the output voltage and current
of the PV panel, the controller adjusts the duty cycle of the converter to obtain the desired
voltage conversion with current control or MPPT control for all operating conditions
Figure 2. 3 DC-DC boost Converter with MPPT/Current Control for Desired Voltage Converrsion
V
A
Micro
Controller
Solar
PanelVoltage
Signal
Current
Signal
OP
-Am
p
C
19
The continuous conduction mode (CCM) governing equations of DC-DC boost
converter for the output voltage gain M(D), current ripple across the inductor ΔIL and
voltage ripple across the capacitor ΔVC are given by (2.1), (2.2) and (2.33) respectively
[75].
DV
VDM
out
in
1
1)( (2.1)
)( sDTL
VL
I in (2.2)
)(2
sDTRC
VcV out (2.3)
Where, Vin is the input voltage, Vout is the output voltage, Ts is the time period
based on switching frequency of the converter, L is the value of inductance to have the
desirable current ripple C is the value of output capacitor to have a desirable ripple in
output voltage and D is the duty cycle of the converter. The converter switches its mode
of operation when excess power is being generated and is not useable for battery charging
or neighborhood sharing. The switching of DC-DC boost converter between MPPT
control and current mode control based upon the external grid state and internal nanogrid
state (state of generation and storage) is discussed in details in Chapter 4.
2.4.1.2. Bidirectional Flyback Converter
A bidirectional flyback converter is employed to enable the resource sharing feature, as
it allows for the transfer of power from nanogrids to the microgrid, and vice versa. The
bidirectional power flow in the proposed flyback converter is attained through modified
switch realization, i.e. replacing the diode of conventional flyback converter with another
controlled mosfet switch. The switch position is also changed to ensure that the source is
grounded without affecting the continuity of the circuit. This allows optimum gate driver
circuit design without the requirement of complex bootstrapping circuit [76].
Flyback converter has inherent advantages of simple design and less component usage
over other types of buck-boost converters, therefore, highly suitable for DC microgrid
applications. Along with the higher conversion ratio, it also allows the use of inherent
magnetizing inductance Lm of the flyback transformer, thus mitigating the need of extra
20
inductor for converter energy transformation. Also, it provides isolation between multiple
nanogrids to ensure reliability in architecture, in case if fault occurs on the grid side or
nanogrid mosfet becomes short circuit. Bi-directional switch realization of flyback
converter is shown in figure. 2.4.
The continuous conduction mode (CCM) governing equations of flyback
converter for the output voltage gain Mf(D) and transformer DC component of
magnetizing current Im are given by (4) and (5)
D
Dn
V
VDM
out
inf
1)( (2.4)
'RD
nVIm (2.5)
Where, Vin is the input voltage, Vout is the output voltage, n is the turn ratio of
transformer and D is the duty cycle of the converter. Multi-mode adaptive control of
isolated bi-directional converter ensuring coordinated resource sharing among the
contributing naogrids is detailed in chapter 4.
Figure 2. 4 Modified Switch Realization of Flyback Converter enabling Bidirectional Power Flow
among the Contributing Nanogrids
V V
120
V D
C M
icro
grid
Transformer
Mosfet
Switches
12V
DC
Batte
ry
Voltage
Signal
OP-Amp
Micro
ControllerVoltage
Signal
21
2.4.2. Model of a Village and Microgrid Scheme of Interconnection
Depending upon the structure, a typical village containing n houses is divided into
x segments with n/x houses per segment, as shown in figures 2.5 and 2.6. Power is
supplied to the load in each household via a flyback converter that, along with the
resistance of the supplying wire, is modeled as a constant power bus and represented by a
distributor resistance. The interconnection resistance between two consecutive nanogrids
is modeled as feeder resistance.
Two interconnection schemes are considered and shown in figure 2.5 and 2.6. Fig.
2.5 shows the radial interconnection of nanogrids that lowers the cost of the conductor in
the system design. However uneven loading, non-uniform voltage distribution, high-
voltage dips at the rear end, and subsequent reliability issues render radial schemes a
relatively poor choice for the optimal distribution of power [77, 78]. Therefore, to address
these issues, the ring main scheme of interconnection is proposed (figure 2.6). It uses an
extra layer of conductors (dashed lines) to connect feeders at the periphery of radial
architecture in a ring main fashion. Thus, at the cost of extra conductors, higher
efficiency and increased reliability are achieved, even at comparatively low distribution
voltages.
Figure 2. 5 Proposed Architecture of the Radial Schemes of Interconnection with Elaborated Single-
Unit Design
x-1(n/x)
1 2 n/x (n/x)+1
Segment 4 Segment x
Segment 1 Segment 2
Segment 3
(n/x)+2 2(n/x)
2(n/x)+1 3(n/x)
3(n/x)+1 4(n/x) n
Battery/
PV Panel
Household
Load
Distributor
Feeder
Bus 1
Bus 2
Single Cell in a Segment
(Nanogrid)
22
Figure 2. 6 Proposed Architecture of the Ring Main Schemes of Interconnection
. Using the values of the feeder and the distributor resistances, based on the scheme
of the interconnection and topological configuration of a village, a conductance matrix G
can be calculated to model it. For a village with n houses, G is of the order of 2n × 2n, as
each house contains two buses: 1) a load bus at the interconnection of the distributor
resistance and the load bus, and 2) another bus at the interconnection of the feeder and
the distributor resistances (figures 2.5 and 2.6). Thus, elements of the conductance
matrices Gij and G can be written in terms of individual conductance gij between any
arbitrary buses i and j, where i may vary from 1 to 2n:
ji ;
i; 2
1
ij
ij
ij
g
jn
ij
j
g
G
(2.6)
nnRG
G
G
G
G
nnn
n
n
22;
G G
G G
G G
2,2 2n,21,2
2,222 21
2,11211
(2.7)
The conductance matrices developed here will be used for power flow analysis
modified for DC power flow in presented chapter 3 for the assessment of central and
distributed architectures of DC microgrids from voltage dips, distribution efficiency and
distribution losses prospective at various LVDC voltage levels and conductor sizes.
x-1(n/x)
1 2 n/x (n/x)+1
Segment 4 Segment x
Segment 1 Segment 2
Segment 3
(n/x)+2 2(n/x)
2(n/x)+1 3(n/x)
3(n/x)+1 4(n/x) n
23
3 Power Flow Analysis of Solar PV
based DC Microgrid
In order to ascertain distribution losses, distribution efficiency and voltage drop in
various power provisioning scenarios of PV based DC microgrid operation, Newton-
Raphson method has been modified for DC power flow analysis. Based upon the
presented method, a comparative analysis of CGCSA and DGDSA with and without
communal load has been performed. In case of CGCSA analysis, optimal placement of
central generation and storage point is discussed in details, while in case of DGDSA
various power sharing scenarios have been evaluated to ascertain the efficacy of proposed
architecture.
3.1. Newton-Raphson Method Modified for Power Flow Analysis of DC
Microgrid
Power flow analysis is conducted through Newton-Raphson method and its application
is modified for our DC microgrid [79]. This is essential to ascertain various critical
elements of the proposed system such as total line losses, efficiency and maximal voltage
drops [77]. These parameters are used as an indicator for the selection of optimal
distribution voltage for DC microgrid. Conventionally, iterative algorithms such as Gauss
Seidel method, Newton–Raphson method and Fast Decoupled method are being used for
the load flow analysis of AC systems [77, 80]. Conventional AC power flow algorithms
cannot be directly applied for DC systems and have to be modified [81]. For our design,
load flow analysis is performed through a modified Newton-Raphson method discussed
in this chapter [79, 82]. This reduces the computational complexity in comparison to
conventional AC load flow analysis which involves complex Jacobian matrix calculation
for the determination of reactive power and voltage angles.
For DC microgrid operation, the interconnections between multiple nodes are
modelled as a power network shown in figure 3.1. Consider the case where there total n
busses and bus i corresponds to each interconnection point between neighboring houses
Each ith
household bus is connected to jth
household bus through a connecting wire with
24
its resistance modelled as a lumped conductance gij. Where, i and j may vary from 1 to n.
gi1, gi2 and gin are the respective value of the conductance between bus i and j.
Using KVL, current in bus i, Ii may be calculated by (3.1) [79]
n
jjiji VGI
1
(3.1)
Where, Gij is the conductance matrix of the power system and is given in terms of
individual conductance gij between bus i and j as given by (3.2).
jifor
ifor 1
ijg
jn
jijg
ijG
(3.2)
Instantaneous power at ith
bus Pi can be calculated as
n
jijjiiii GVVIVP
1
(3.3)
Using Taylor series expansion while neglecting higher order terms yields (3.4)
knV
kV
nV
knP
V
knP
nV
kP
V
kP
knP
kP
:
2
....
2
:
2 ....
:
2
2
:
2 (3.4)
Figure 3. 1 Power Network for Newton-Raphson method modified for DC Power Flow Analysis
25
Where, ΔP is the difference in scheduled power Psch and calculated power Pi
using (3.4) for k iterations. Desired voltage level of LVDC on which power flow analysis
has to be performed, is taken as reference for slack bus. Initial estimate of voltages for
generator busses are slightly higher than reference voltage and for load busses these are
assumed slightly lower than reference voltage. Using (3.4), change in voltages ΔV and
corresponding bus voltage V for k iterations are found until the difference between
scheduled and calculated power becomes negligible. After convergence, bus voltage V is
used to calculate the power of slack bus using (3.3). By using the converged value of
voltage at each bus, associated line losses LLg, percentage line losses %LLg, voltage drops
VDg, percentage voltage dip %VDg, and efficiency η for the dc microgrid are calculated
as given by (3.5) – (3.9).
n
i
n
jijjjiiijg VVVVVVGLL
1 12
1
(3.5)
n
ii
g
g
P
LLLL
1
0
%100 %
(3.6)
gLL%100
(3.7)
minmax VVVDg
(3.8)
max
minmax %100*
V
VV%VDg
(3.9)
Where, Vmax and Vmin are the maximum and minimum values of voltage at any bus after
kth
iteration. The methodology for power flow analysis is valid for DC microgrids and
will be used for the calculation of distribution losses, distribution efficiency and voltage
drops for CGCSA and DGDSA for various mass produced conductor sizes and voltage
levels. Based upon the methodology presented here, various power provisioning
scenarios will be evaluated for DGDSA, while, for CGCSA, effect of placement of
central generation and storage resources at various possible locations in the voltage will
also be discussed in details. Therefore, the presented Newton Raphson method modified
26
for power flow analysis will serve as the basis for microgrid architecture classification
from the prospective of distribution losses.
3.2. Case Study Parameters and Simulation Results
Voltage level is one of the prime factors affecting the operation, control, performance,
protection and safety of the microgrid based distribution systems [83-85]. Feasibility
studies of viable voltage levels for low voltage DC distribution in commercial and
residential systems are presented in [85, 86]. Due to the available applications and
devices, 12V, 48V, 120V, 230V, 325V and 380V are considered as viable voltages for
low voltage direct current (LVDC) distribution, where 12V is the most common voltage
for market available lead-acid batteries and DC loads including LED‘s and DC fans, 48V
is generally used in telecommunication sector, 120V rms and 230V rms are used as
standard distribution voltage in many countries, 325 is equivalent to the capacitor filtered
rectified voltage produced in 220V rms AC systems and 380V is generally employed for
DC distribution [83-85, 87]. They may be further categorized as very low, medium and
high levels of LVDC distribution.
For a given amount of power to be distributed, higher voltages yield higher efficiency
and lower line losses. While, lower voltages result in better safety aspects and reliable
protection against short circuit faults. In case of coupling with the public grid, safety and
protection issues are of more prime importance. Therefore, an optimal selection of
voltage level ensures an acceptable balance between safety, reliability and efficiency. In
CGCSA, distribution losses for 12 and 48V are high for driving considerable amount of
load i.e. 40W for 40 houses and solution does not converge, therefore, for CGCSA, these
two voltage levels are not considered.
The proposed architecture has been modeled around a typical village in developing
world containing 40 houses. All the houses are interconnected to DC microgrid via
‗feeders‘ and ‗distributors‘, where, the distance between consecutive houses (feeder) and
the length of the internal wiring (distributor) was 20 m, consistent with the situation in
rural settlements in developing countries as described by Varshney et al. [88]. Each
house is capable of driving 40W of DC load including lighting, fan and charging loads.
The rated power provisioning is in accordance with the market availability of DC loads
27
with three to four lighting bulbs (~4W each), one DC fan (~14 W) and one to two mobile
charging unit (~4 W). Similarly, for the village under analysis a communal load of up to
500W is considered for water filtration plant/pump for drinking purposes. Different
distribution conductors with wire gauge areas of 0.2 mm2 (local market name 3.0-
0.029‖), 0.45 mm2 (local market name 7.0-0.029¬‖), 2.5 mm
2, 6 mm
2, and 7.5 mm
2 were
considered for analysis. Using (3.6) – (3.9), %LLg, %η and %VDg are calculated for the
following scenarios.
3.2.1. CGCSA without Communal load
The CGCSA of a PV based DC microgrid is shown in figure 2.1. For the analysis
in this section peak load scenario is considered, where every house is demanding 40W
power. The optimal placement for generation and storage units are found by running an
iterative loop for all possible generation busses. System efficiency against each bus
placement is recorded and bus with highest efficiency is selected. Figure 3.2 shows the
results for %VDg and %η for different voltage levels and different mass produced
conductors, when generation and storage units are placed at one of the starting or ending
points of the corner segments (Refer to figure 2.5).
Figure 3. 2 Typical % Voltage drop and Efficiency for CGCSA with Peak Load and Far End
Placement
0 1 2 3 4 5 6 7 8
75
80
85
90
95
100
Eff
icie
ncy
(%
)
Conductor Area (mm2)
400V
325V
230V
120V
28
Figure 3. 3 Typical % Voltage Drop and Efficiency for CGCSA with Peak Load and Central
Placement
From figure 3.2, it can be observed that at higher voltages, distribution efficiency
is higher while voltage drops are relatively lower, while thin conductors are not suitable
at low voltages, resulting in very high voltage drop. Voltage drop is a direct measure of
the converter requirements in the system; therefore, voltage drop generally higher than
20% is not suitable for reliable operation of the microgrid.
Figure 3.3 shows the results for %VDg and %η, for different voltage levels and
different mass produced conductors, when generation and storage units are placed at
center of segment 3 (Refer to figure 2.5) . From figure 3.3, it may be concluded that
higher efficiencies and lower voltage drops are achieved by placing generation and
storage unit at central location as compared to far end region placement.
3.2.2. CGCSA with Communal load
Figure 3.4 shows the results for %VDg and %η for communal load scenario, in
which each of the houses is restricted to 30W, while 400W is supplied to communal load.
This may be a typical scenario where during first half of a day, the power is allocated for
computing school for children with limited power delivered to households. The optimal
location for communal load on the microgird is found by running an iterative loop for
placement at all possible load busses and comparing system efficiency. The bus with
highest efficiency is selected which in our proposed case of CGCSA is located at the
center point of the central segment (see figure 2.5).
0 1 2 3 4 5 6 7 8
76
80
84
88
92
96
100
Eff
icie
ncy
(%
)
Conductor Area (mm2)
400V
325V
230V
120V
29
Figure 3. 4 % Voltage drop and Efficiency for CGCSA with Communal Load and Central Placement
3.2.3. DGDSA with Radial Scheme of Interconnection
The DGDSA of a PV based DC microgrid is shown in figure 3.5. Typical and
peak load scenarios are evaluated for DGDSA with radial scheme of interconnection. In
typical load scenario, the power sharing between houses and grid is taken as ±20% (a
house can demand 20% more or supply 20% of its rated power). Power sharing is
randomly scheduled at various household between ±8 W i.e. 20% of rated power, therefor
power losses and voltage drops are very limited. Results for percentage voltage drop and
efficiency (figure 3.5) show that with moderate size conductor such as 2.5 mm2 the loss is
low for 120V and higher voltages with efficiencies above 99%.
The situation changes significantly when power flow provision is kept higher to a
household. Under this peak load scenario, a house can consume twice of its rated power
and it can supply all available power to the grid. In this scenario, power is randomly
scheduled either +40W or -40W at each household bus. Results show that efficiency is
lower for low voltages and a significantly higher cost of thick conductor must be incurred
for this topology to achieve lower line losses as shown in figure 3.6.
0 1 2 3 4 5 6 7 8
75
80
85
90
95
100
Eff
icie
ncy
(%
)
Conductor Area (mm2)
400V
325V
230V
120V
30
Figure 3. 5 Percentage Voltage Drop and Efficiency at Different Voltages and Different Conductor
Sizes for Typical Load sharing with Radial Scheme of Interconnection (Common Load Sharing
Radial DC Microgrid)
Figure 3. 6 Percentage Voltage Drop and Efficiency at Different Voltages and Different Conductor
Sizes for Peak Load sharing with Radial Scheme of Interconnection (Peak Load Sharing Radial DC
Microgrid)
3.2.4. DGDSA with Ring-Main Scheme of Interconnection
In order to mitigate issues with Radial DGDSA, we proposed a more efficient and
reliable ring-main topology (refer to figure 2.5). This topology significantly improves the
efficiencies at peak load sharing in comparison to radial topology at the cost of extra
conductors in ring main topology which is used to connect the end nodes as shown in
figure 2.6.
0 1 2 3 4 5 6 7 898.4
98.6
98.8
99.0
99.2
99.4
99.6
99.8
100.0
Eff
icic
en
cy
(%
)
Conductor Area (mm2)
48V
120V
240V
325V
400V
0 1 2 3 4 5 6 7 8
50
60
70
80
90
100E
ffic
ice
nc
y (
%)
Conductor Area (mm2)
48V
120V
240V
325V
400V
31
Figure 3. 7 Percentage Voltage Drop and Efficiency at Different Voltages and Different Conductor
Sizes for Peak Load sharing with Ring-Main Scheme of Interconnection. (Peak Load Sharing Ring-
Main DC Microgrid)
Ring-main interconnection topology also provides necessary redundancy to
ensure service operation in grid disconnection due to a fault or broken wires. The results
for peak load sharing for ring-main scheme have been shown in figure 3.3-3.7. From the
comparison of figures 3.6 and 3.7, it is evident that ring-main topology has higher
efficiency and less voltage drops in comparison to radial scheme of interconnection. For
instance, in case of peak load sharing at 120V and 2.5mm2 conductor size, ring-main
scheme exhibits higher efficiency (95.18%) as compared to radial scheme (92.66%).
Similarly, maximal voltage drops which are a direct measure of converter requirements,
are lower for ring-main scheme of interconnection in comparison to radial scheme for all
voltage levels and conductor sizes under considerations. The tabular comparison between
radial and ring-main DGDSA power flow will be presented in the subsequent sections.
3.2.5. DGDSA with Communal Load and Ring-Main Scheme of Interconnection
In order to analyze the performance of DGDSA under communal load scenario, with
ring-main scheme of interconnection among the contributing nanogrids, the power
scheduling has been adjusted such that each and every household/ nanogrid contribute
10W for communal load. Therefore, an aggregation of resultant power drives a
communal load of 400W.
0 1 2 3 4 5 6 7 850
60
70
80
90
100
Eff
icie
nc
y (
%)
Conductor Area (mm2)
48V
120V
240V
325V
400V
32
Figure 3. 8 Percentage Line Losses and Efficiency at Different Voltages and Different Conductor
Sizes for Communal Load Case with Ring-Main Scheme of Interconnection
This is the main advantage of such a nanogrid based scheme that dedicated power
resources need not be planned/installed for driving communal loads, including the
pumping load for water filtration plant or irrigation system, computing load of a school or
medical equipment load of a hospital. At day time when enough sunlight is available and
each household is producing more than its local requirements, power may be pooled up
for community load purposes. The power electronic interface and associated control
schemes for this coordinated power sharing is discussed in details in chapter 4. The
analysis performed using Newton – Raphson method modified for DC power flow
showed that for the selected parameters, the distribution losses were less than 3% even at
a communal load of 400 W as shown in figure 3.8. Also voltage drops are within limited
range of ±15 % even at lower voltage levels of LVDC i.e. 48V and 120V.
3.3. Summary of Power Flow Comparison between CGCSA and DGDSA
Due to distributed placement of resources in DGDSA, distribution losses
associated with the delivery of power are relatively lower. In those scenarios, when each
household is relying on its local resources without any exchange from the neighboring
households, distribution losses are negligible. Even, in the case of typical power sharing
among households, distribution efficiency is above 98% for all voltage levels and
conductor sizes.
0 1 2 3 4 5 6 7 8
60
70
80
90
100
Eff
icie
ncy (
%)
Conductor Area (mm2)
48V
120V
240V
325V
400V
33
In worst possible scenarios, when all the households are exchanging their rated
power, distribution losses are still less than the case of CGCSA. Therefore, due to its
inherent distribution of resources near to local load, DGDSA has better distribution
efficiency in comparison to CGCSA of DC microgrid for all of its operation modes. In
order to make a comparison between both the architectures, they are compared on the
same level of system loading. A typical conductor mass of 2.5 mm2 cross section is
selected.
Table 3.1 show the comparison between the system parameters (Percentage line
loss, Percentage voltage drop and efficiency) of CGCSA and DGDSA for communal load
scenario, where 400W power is supplied to communal load. From the results summarized
in table 3.1, it may be concluded that for uniform system loading conditions, DGDSA has
comparatively higher efficiencies, lower line losses and lower voltage drops in
comparison to CGCSA due to distribution of resources, usage diversity and mutual
resource sharing capabilities.
Table 3. 1 Communal Load Comparison between CGCSA and DGDSA
Voltage
Level
(V)
Cond.
Area
(mm2)
CGCSA
DGDSA
LLg (%) VDg
(%)
η
(%)
LLg
(%)
VDg
(%)
η
(%)
120 2.5 8.11 8.86 91.89 3.30 3.52 96.70
230 2.5 2.38 2.53 97.62 0.80 0.84 99.20
325 2.5 1.33 1.43 98.67 0.44 0.45 99.56
400 2.5 0.89 0.96 99.11 0.29 0.30 99.71
34
3.4. Power Flow Comparison between Radial and Ring - Main scheme of
interconnection for DGDSA of Microgrid
As evident from results, DGDSA has superior characteristics in terms of higher
distribution efficiency, lower distribution losses and lower worst voltage drops in
comparison to CGCSA. Further, at the cost of extra conductor, distribution losses and
worst voltage can further be reduced and system reliability can be increased at the same
time using ring- main scheme of interconnection. Therefore, a comparison between radial
and ring-main scheme has been illustrated to develop the superiority of ring-main scheme
over radial scheme for the same level of system loading and parametric conditions. For
the purpose of comparative analysis following cases are considered.
3.4.1. Typical Load Comparison between Radial and Ring- Main Scheme of DGDSA
Table 3.2 illustrates the comparative results of system parameters (Percentage line
loss, Percentage voltage drop and efficiency) for typical case of power sharing, in which
each house is either supplying or demanding 20% of its rated power, at conductor size of
2.5 mm2. From tabular comparison (Table 3.2), it is evident that ring-main scheme is far
more superior in terms of higher efficiency, lower distribution losses and lower voltage
drops in comparison to radial scheme.
Table 3. 2 Typical Load Comparison between Radial and Ring-Main DC Microgrid
Voltage
(V)
Area
(mm2)
Radial Microgrid
Ring-Main
Microgrid
LLg
(%)
VDg
(%)
η
(%)
LLg
(%)
VDg
(%)
η
(%)
48 2.5 0.79 1.89 99.21 0.58 1.04 99.42
120 2.5 0.13 0.30 99.82 0.09 0.17 99.91
400 2.5 0.02 0.03 99.95 0.02 0.01 99.99
35
3.4.2. Peak Load Comparison between Radial and Ring- Main Scheme of DGDSA
Table 3.3 illustrates the comparative results of system parameters (Percentage line
loss, Percentage voltage drop and efficiency) for the case of peak power sharing, in which
each house is either supplying or demanding 100% of its rated power, at conductor size
of 2.5 mm2. From tabular comparison (Table 3.2), it is evident from table that ring-main
scheme is far more superior in terms of higher efficiency, lower distribution losses and
lower voltage drops in comparison to radial scheme. Even at very low voltage level of
LVDC i.e. 48V, distribution efficiency is above 82% at peak power sharing, in
comparison to radial scheme where distribution efficiency is 76% only (6 % gain in
efficiency at the cost of extra conductor). Since lower voltages of LVDC i.e. 48V and
120V are considered safe for indirect touch and requires less protection in comparison to
higher voltage levels, i.e. 230V, 320V and 380V, therefore, with the ring-main scheme of
interconnection, lower voltages can be used for distribution without significant reduction
in the distribution efficiency. Therefore, ring-main scheme provides ensure reliability
along with higher distribution efficiency with minimum converter requirements to keep
the voltage dips minimum at the rare end of the microgrid.
Table 3. 3 Peak Load Comparison between Radial and Ring-Main DC Microgrid
Voltage
(V)
Area
(mm2)
Radial Microgrid
Ring-Main
Microgrid
LLg
(%)
VDg
(%)
η
(%)
LLg
(%)
VDg
(%)
η
(%)
48 2.5 23.5 34.4 76.41 17.9 27.1 82.1
120 2.5 7.34 10.9 92.66 4.82 7.38 95.18
400 2.5 0.83 1.23 99.17 0.51 0.77 99.49
36
3.5. Hardware Setup, Parameters and Results for Power flow Analysis of DC
Microgrid
Scaled model of the microgrid is implemented as shown in figure 3.9. The
proposed model that originally includes 40 houses is scaled down to four houses, where
each house may either generate or consume the requisite amount of power under various
scenarios. The generation capacity of each house was implemented via power supplies
(ESCORT EPS3030T) and consumption capability through DC load banks (LABTECH
LEMSPL) available in the laboratory.
In case of DGDSA as show in figure 3.9, two houses are producing net power,
while remaining two houses are consuming net power. Similar arrangement is made for
CGCSA, where only one house generates net power and rest of three houses will
consume power. Peak load operation for DGDSA and CGCSA is evaluated and measured
results are found in agreement with the simulated results. The measured v/s simulated
results of percentage voltage drops for CGCSA and DGDSA for scaled down version are
shown in figure 3.10. Difference between simulated and measured results accounts for
the excessive resistive losses at the joints and terminal points.
Figure 3. 9 Hardware Implementation of Scaled down version for Power Flow Analysis
37
Figure 3. 10 Measured v/s Simulated %Voltage Drops Results at 120V, 230V, 325V and 400V for
a) DGDSA and b) CGCSA
Both radial DGDSA and ring-main DGDSA interconnection schemes are tested at
the specified voltage levels with appropriate conductor size of 2.5mm2 and results are
compared with simulation outcomes as shown in figure 3.11. Simulated and measured
line losses are normalized with respect to highest value of the losses in corresponding
scheme. Results show that there is a difference between simulated and hardware results.
Difference between simulated and measured results accounts for the excessive resistive
losses at the joints and terminal points in hardware setup. Moreover, it may be verified
from the results that the losses (both simulated and measured) in radial scheme of
interconnection are higher than ring-main scheme at all the specified voltage levels.
Figure 3. 11 Simulated v/s Measured Results for Normalized Line Losses in DGDSA with Radial and
Ring-Main Schemes of Interconnection
100 150 200 250 300 350 4000.00
0.02
0.04
0.06
Vo
lta
ge
Dro
p (
%)
Voltage (V)
Simulated Results
Measured Results
100 150 200 250 300 350 4000.00
0.05
0.10
0.15
0.20
Vo
ltag
e D
rop
(%
)
Voltage (V)
Simulateded Results
Measured Results
48 120 230 325 4000.0
0.2
0.4
0.6
0.8
1.0
No
rma
lize
d L
ine
Lo
sse
s
Voltage Levels (V)
Radial Scheme
Measured
Simulated
48 120 230 325 400
0.0
0.2
0.4
0.6
0.8
1.0
Norm
aliz
ed L
ine L
osses
Voltage Levels (V)
Ring Main Scheme
Measured
Simulated
38
3.6. General Conclusions Drawn from Power Flow Analysis of DC Microgrids
It is clear from figure. 3.2 to figure 3.11 that at higher voltages, line losses are less
while efficiencies are higher. For a kW range power to be distributed, higher voltages
yield higher efficiency and lower line losses. While, lower voltages result in better safety
aspects and reliable protection against short circuit faults. Safety is a critical concern and
at higher voltage levels, the system requires sophisticated and expensive protection
equipment. This limits our choice of voltage level depending on the amount of resources
available for the DC-Grid. It should be noted that DC breakers are more sophisticated and
cumbersome to manage in case of short-circuits or unwanted transients which may
require disconnection of the DC grid from nanogrids. For instance, in disconnection, the
arching is largely limited in AC due to zero crossing of the 50Hz frequency unlike a DC
grid. In DC systems the arcs may be higher and could potentially be hazardous.
In general lower voltages i.e. 24V and 48V do not have additional safety
concerns. The possibility of over current and associated over-heating hazards exist in
case of short circuits and such short circuits may be avoided using fuse protection after
the DC power source. For a practical system this must be added as an additional level of
safety. According to IEC 60950-1, voltage levels lower than 60V DC are considered safe
for direct touch and potentially do not create any electric shock hazard. Similarly,
according to NEC standards (NFPA 70), DC power sources having voltage levels lower
than 60V DC and maximum power capability of 100VA are referred as class 2 DC power
sources and are capable to eliminate electric shock hazards as well as fire protection
hazards. Emerge Alliance; an organization working on DC standards for buildings has
also nominated 24V DC as safe voltage against electrical shock/fire hazard with the
source capability to provide current less than 4.1A [89-91]. Therefore, in general, from
safety point of view 24V and 48V are relatively more safe levels of LVDC in comparison
to 120V 230V and 380V DC. However, voltage levels lower than 48V are not suitable for
distribution as voltage drop is much higher than 15%. Therefore, from system safety
point of view 48V distribution with a reasonable conductor can be considered as a choice
for future practical installations of DGDSA.
39
Moreover, as we move from 48V to higher voltage levels, losses are reduced and
efficiency is increased but this particular trend is not uniform as listed in Table 3.3 to
Table 3.3. As we move from 48 V to 120 V, there is a significant reduction in losses and
increase in the efficiency. The associated protection and safety requirements are not
excessive [92, 93]. Moving from 120V to 400V, percentage reduction in losses is not
significant for a typical conductor gauge of 2.5 mm2. However, system complexity,
protection and safety requirements increase significantly. Also, the distributed storage
and generation allows small current to flow on the grid and the arc lengths and arc energy
could be easily manageable even with 120V DC, in comparison to 240V AC [93].
Moreover, voltages lower than 120V are considered safe for indirect touch and require no
extra grounding and protective conductors [92]. While distribution at 120V is generally
less efficient than 400V, our proposed ring main interconnection between feeders and
microgrid allows minimal and loss adds necessary redundancy for a reliable but efficient
solution. 120V can therefore be considered as another optimal medium level voltage that
provides the desired trade-off between the safety aspects and efficiency, while keeping
voltage drops and line losses within acceptable range. For this particular village and
household load settings, selecting 120V voltage level will result in safer overall system
without a significant loss in the system efficiency. Therefore, for the considered
specifications of the village, 120 V using a conductor of area 2.5 mm2 and the ring main
scheme of interconnection is optimum for the operation of the microgrid. For different
load specifications for other villages, the proposed analysis may yield optimum voltage
levels others than 120 V, depending on the trade-off between losses and the cost of the
protective equipment.
40
4 Decentralized Control of Solar PV-
based DC Microgrid
Due to distribution of resources in DGDSA of DC microgrid, the coordination
among the dispersed generation and storage resources in a decentralized and
communication-less manner is extremely challenging task for the controlled operation of
DC microgrid. Therefore, this chapter is dedicated to the communication-less and
decentralized control of clusters of nanogrid. A hysteresis based voltage droop method,
implemented through a localized controller without the need for central, adaptive, or
supervisory control has been presented with hardware implementation. The limitations
with the perturbation based voltage droop method and possible enhancements in terms of
resource availability based coordinated control have been addressed through an adaptive
decentralized controller. Various possible scenarios of DC microgrid operation are
simulated and efficacy of the proposed adaptive control is validated through hardware in
loop (HIL) experimentation at Microgrid Laboratory in Aalborg University.
4.1. Need for a Decentralized and Communication-less Control Strategy for DC
Microgrid based Rural Electrification
PV/battery based islanded DC microgrids are becoming very popular for the
electrification of developing regions. Almost all of the existing deployments use CGCSA
of DC microgrid, where PV generation and battery storage is kept at a centralized
location. CGCSA based implementation are relatively simpler from installation, control
and operation prospective, however, they lack modularity and significant distribution
losses are associated with their delivery of energy. Alternatively, various distributed
architectures for PV/battery based islanded DC microgrids have been proposed in
literature. The advantages of distributed architecture are mainly reduction in distribution
losses and modularity in structure. However, coordination among the distributed
resources and control for power sharing becomes extremely challenging. Several
strategies for hierarchical and supervisory control of DC microgrids have been proposed
in [8, 66-68]. However, these require an extra layer of sensing and communication, which
41
in turn enhances the cost and complexity of the system. For rural electrification purposes,
such a complex, high cost and communication based distributed architecture is not
preferable due to constraints of limited funding. Therefore, a communication-less and
decentralized control strategy for the stable operation of a highly distributed architecture
of DC microgrid (DGDSA) is highly desirable as it has the capability to combine the
advantage of both of the existing architectures i.e. a) lower distribution losses, b)
scalability and modularity, and c) simplified, robust and coordinated control [71].
4.2. Hysteretic Voltage Droop Algorithm for Distributed Control of DGDSA
Energy balance for an ideal DGDSA based village (figures 2.5 and 2.6) containing n
households with distributed PV generation Pi PV
(t) at each household i at any time t is
given by (4.1), based upon the constraints given in (4.2)
n
i
B
i
n
i
L
i
n
i
PV
i tCtttPtP t111
i
SOC
(4.1)
tiSOCiSOCiSOC
nRSOC
BV
LP
PVP
; minmin
1
,,,
(4.2)
where, CiB(t) is the household battery capacity, SOC is the state-of-charge of the battery,
SOCi max
(t) and SOCi min
(t) are the allowable limits on the battery‘s SOC, and PiL(t) is the
household load power connected to the battery bus drawing Ii(t)L for i
th household within
time interval Δt. For the grid to operate at rated voltage VG
rated with net current IG(t), the
batteries can either deliver power to the grid or take power from it using the bidirectional
flyback converter. Therefore, (4.1) can be written in terms of battery current IiB(t), which
may have either a positive or a negative value depending on the state of the grid and
subject to the constraints in (4.4)
ttIVttIVttIVn
i
B
i
B
i
n
i
L
i
B
i
n
i
G
i
G
rated 111
.
(4.3)
; , 0 max,min,max,
tIIIIIB
i
B
i
B
i
L
i
L
i
(4.4)
42
where, IiL,max
is the maximum value of the permissible load current at each house, and
IiB,max
and IiB,min
are the limits on the charging and discharging currents of the battery.
Based on the duty cycle control of the flyback converter, power must be
channeled from the microgrid to nanogrid or vice versa. If the flyback converter operates
at critical duty dicric
(given by (4.5)), there is zero power sharing between the microgrid
and the battery of household i. To set the direction and magnitude of power flow from
microgrid to nanogrid, a positive perturbation in duty Δdi is applied. Thus, the flyback
converter channels the power from the microgrid to the load bus when operating above
critical duty dicric
+Δdi. Below the critical duty dicric
-Δdi, the bidirectional flyback
converter ensures the flow of power from battery to microgrid. Thus, the energy stored in
the battery may be transferred back to the grid using the negative perturbation in critical
duty.
G
rated
B
i
B
it
cric
iNVtV
tVd
(4.5)
For stable microgrid operation, the duty of each flyback converter is adjusted such that
it produces VG
rated at the microgrid; therefore, (4.5) may be written as follows, subject to
the constraints in (4.4):
111
n
i
tB
itL
i
i
tin
i
tG
i IItd
dNI
(4.6)
The stable operation of the microgrid is defined by the hysteresis in grid voltage such that
VG
min ≤ VG
rated ≤ VG
max , where, VG
min and VG
max are the minimum and maximum values of
the grid voltage, respectively, dictated by the hysteresis generally maintained at ±2% of
the rated grid voltage. Using the power balance of (4.6), an algorithm is formulated
(shown in Figure 4.1) for generalized microgrid operation based on the duty cycle control
of the flyback converter of each household. The perturbation applied in the duty cycle
adjusts the direction and amount of allowable power shared between the nanogrids and
the microgrid, hence maintaining power flow such that the microgrid is always stable.
As an example, if each household has enough availability of resources, i.e. either
PV generation or battery storage is enough to operate the household load, each flyback
43
converter will operate on its critical duty, such that net power exchange to or from the
microgrid is zero and grid voltage is stable with in the permissible range. If the grid has
excess power available, its voltage will start rising and after approaching the maximum
allowable limit, the positive perturbation in duty of individual nanogrids will channelize
the excess power toward the battery storage of the nanogrid or towards an elastic load. In
this situation, the voltage of the microgrid decreases in proportion to the net power
transfer. Our algorithm ensures that if the grid voltage drops below VG
min, the direction of
power flow is reversed using a negative perturbation in the duty of the flyback converter
to maintain balance in voltage. Thus, a negative perturbation applied above the critical
duty channelizes the stored power of the battery toward the microgrid to increase its
voltage above VG
min to VG
max.
Thus, hysteresis based voltage droop control determines the required perturbation
in the duty and the associated amount of power flow between microgrid and nanogrid
while keeping the voltage within the hysteretic limit, hence ensuring the stability of the
scheme throughout its operation. Because of this distributed control structure, each
nanogrid is responsible for the stable operation of the microgrid. Therefore, the need for a
central controller and a costly communication interface is obviated in the proposed
architecture. Further, due to decentralized control, the hysteresis based voltage droop
control also renders the proposed DGDSA highly scalable in terms of future expansions.
44
Figure 4. 1 Hysteretic-based Distributed Voltage Droop Control Algorithm
4.3. Hardware Implementation Results for Hysteresis based Voltage Droop
method For DGDSA
The distribution voltage of the grid VG
rated was 120 V while the household load
distribution and storage voltage VB was 12 V. The setup for the integration of three
atomic nanogrids into the microgrid is shown in Figure 4.2. The DC microgrid was
implemented using a large capacitor (5000 μF), the states of charging and discharging of
which were continuously monitored. House 1 (H1) supplied constant power, and was
modeled using a DC power supply (ESCORT EPS3030T). House 2 (H2) was modeled
using a four-quadrant bipolar power supply that could act as a power source or sink.
House 3 (H3) was modeled using by a battery along with a bidirectional flyback
converter.
Start
Measure: VG(t),ViB(t),SOCi
B(t),IiB(t)
Set di(t)=dicric(t)
VG(t) > VGmin
SOCiB(t)<SOCi
max
IiB(t)<Ii
B,max
di(t+1)=di(t)+Δdi di(t+1)=di(t)
SOCiB(t)>SOCi
min
IiB(t)>Ii
B,min
di(t+1)=di(t)-Δdi di(t+1)=di(t)
VG(t) < VGmax
Set IiL=0
YES
YES
YES
YES
YES
YES
No
No
No
No
No
No
Connect Load0<Ii
L IiL,max
ORDisconnect MPPT
45
Typical voltage variations of the microgrid in various power sharing scenarios are
shown region wise in figure 4.3. In ―Region 1,‖ houses 2 and house 3 used power from
the grid while house 1 supplied to the grid. Based on the algorithm in Figure 4.2, when
the power supplied by house 1 was less than the power being taken by the other two
houses, the voltage of the grid decreased. As the voltage dropped below the specified
lower threshold VG
min=117.5V, the loads of both houses were turned off. In ―Region 2,‖
power from the battery bank in house 3, house 1, and house 2 started charging the grid
again to increase its voltage above 120 V. When the voltage was above the hysteretic
threshold of the grid, set to 122.5 V, the loads were turned on again per the proposed
algorithm. In ―Region 3,‖ houses 1 and 2 supplied power to charge the battery of house 3.
The battery voltage ―VB‖ is also constantly monitored during charging and discharging as
shown in Figure 9. The state of the system was not kept fixed at 120 V; rather, hysteresis
was kept at around the upper and lower cut-off limits, i.e., VG
max=122.5 and VG
min=117.5
V. Therefore, a balanced load and bidirectional flow of power were maintained
throughout the operation to ensure the stability of the grid.
Figure 4. 2 Implementation of the DGDSA Microgrid Hardware through the Integrations of
Nanogrids
46
Figure 4. 3 Results of hardware implementation of typical voltage variations of the microgrid in
various power sharing scenarios.
4.4. Limitations of Hysteresis based Voltage Droop Method and Need of a Robust
Adaptive Control
The hysteretic based voltage droop control scheme presented above allows the
simplified operation and control of DGDSA without any need of a central controller.
However the hysteretic based voltage droop algorithm presented in [71] depends upon the
perturbations in duty cycle. Therefore, a very small perturbation in duty makes the
dynamics of system very slow to achieve the desired power sharing, while a higher
perturbation in duty cycle may lead to instability. Moreover, in such a scheme, resource
sharing capability among the distributed resources is uncoordinated i.e. all nanogrids
share or demand uniform amount of power regardless of their current states generation
and storage.
Xiaonan et al. [94] developed an adaptive dual loop droop control (inner current
loop and outer voltage loop) on the basis of state of charge (SOC) balancing for
distributed storage resources in DC microgrid applications. This adaptive droop considers
Grid Voltage
Battery Voltage
117
118
119
120
121
122
123
Hx= House Supplying Power
Hx= House Consuming Power
H1,H2,H3
Region 4
H1,H2,H3
Region 2
H1,H2,H3
Region 3
Time
Gri
d V
olt
ag
e (V
)
H1,H2,H3
Region 1
11.0
11.5
12.0
12.5
13.0
13.5
14.0
14.5
15.0
Lower cut off Limit Ba
tter
y V
olt
ag
e (V
)
Upper cut off Limit
47
power sharing proportional to the battery SOC index during power supply mode (battery
discharge mode). However, it does not consider power sharing in proportional to the SOC
index during charging mode of the battery. Therefore, all batteries get charged with the
same power independent of their state of charge or resource availability for battery
charging. If such a scheme is applied on DGDSA of DC microgrid presented in [71]
having local loads, there will be redundant distribution losses for un-wanted SOC
balancing. Ideally, in such architectures, it is desirable that if SOC is above a certain
threshold, it must be maintained to that level rather than undesired balancing. Moreover,
Zheming et al. [95] showed that the VI dual loop droop control exhibit slower dynamics
in comparison to IV droop, therefore, it cannot achieve fast power sharing among the
distributed resources.
Therefore, in order to rectify the limitations of above mentioned decentralized
control schemes, we present an adaptive IV droop method for the decentralized control of
a PV based DGDSA of DC microgrid suitable for rural electrification. The resource
sharing among the contributing nanogrids is kept in proportion to the availability of
resources for both operation modes i.e. during supply and demand of the power to or
from the microgrid (charging and discharging of the battery). This power sharing
proportional to resource availability is achieved by using an adaptive IV droop algorithm
which may adopt multiple modes based upon the local measurement of DC bus voltage
and SOC of the battery. Moreover, the proposed control scheme ensures fast dynamics
and is capable to deal with the extreme operating conditions by synchronizing PV
generation capability of individual nanogrids with the local load requirements and grid
stability conditions through a local controller which may shift its modes of operation
between MPPT mode and current control mode. Since, the proposed control scheme
relies on the local measurements of load current, PV generation, battery state of charge
(SOC) and DC bus voltage; therefore, does not require communication for the
coordinated power sharing among the contributing nanogrids. Thus, with the proposed
adaptive control scheme, PV based DGDSA combines the advantages of both of the
existing architectures i.e. a) scalability, b) modularity, c) lower distribution losses and d)
robust, coordinated and communication-less decentralized control. Thus, it can be
48
considered as an ideal candidate for future deployments of rural electrification projects in
developing regions.
4.5. Distributed Architecture and Control Objectives
The combination of PV generation, battery storage, local DC loads and DC-DC
converters in an individual household formulates a nanogrid. Local generation and local
storage allows the nanogrid to work independently even if the grid is unavailable and has
many practical advantages compared to central generation based systems. Therefore,
DGDSA of DC microgrid can be considered as a cluster of multiple nanogrids that are
interconnected via a DC-link in figure 4.4.
Figure 4. 4 A Cluster of Multiple Nanogrids Interconnected via DC Bus Formulating the DGDSA of
PV/battery based DC Microgrid
DC
BU
S
Community Load
Nanogrid 1/ Household 1
Nanogrid 2/ Household 2
Nanogrid N/ Household N
Circuitbreaker
Step DownConverter
DC Loads Battery
BidirectionalConverter
Roof Mounted PV Panel
Step DownConverter
DC Loads Battery
BidirectionalConverter
Roof Mounted PV Panel
Step DownConverter
DC Loads Battery
BidirectionalConverter
Roof Mounted PV Panel
Circuitbreaker
Circuitbreaker
Circuitbreaker
49
An individual nanogrid is therefore considered a basic building block, whose
modular replication and subsequent DC-link integration yields scalability in the
architecture. Each nanogrid operates independently when it is self-sufficient in its
resources and the resource sharing among multiple nanogrids is enabled only when an
individual nanogrid has either access or deficiency of resources. Therefore, energy losses
with the distribution of energy in DGDSA are limited in comparison to centralized
architectures, where generated energy has to be distributed all the way from centralized
generation point to individual households [71, 79]. Further, DGDSA has the capability to
aggregate power from multiple nanogrids for driving community loads, which is
otherwise expensive and unsustainable in limited rural electrification projects [71, 79].
For stable operation of the grid, DC bus voltage VB must be maintained to a rated
value Vref with some allowed fluctuation in bus voltage ΔVB for all possible operating
conditions. The other control objective is to minimize the overall distribution losses,
while maintaining a coordinated resource sharing among the nanogrids without any
additional physical layer of communication. Based upon the availability of distributed
resources, following cases of operation may arise:
a. Each nanogrid is self-sufficient in its resources i.e. PV generation/battery cushion is
in accordance with household load requirements, and any exchange of power among
the contributing microgrids is not desirable to minimize the distribution losses.
b. Although each nanogrid is self-sufficient in its resources, but there is a communal
load demand on the microgrid. In this case it is desireable that each individual
nanogrid contributes power for communal load operation in proportion to its
resources availability.
c. Out of N nanogrids, K nanogrids are self-sufficient while N-K nanogrids are deficient
in resources i.e. their household load requirements are higher from their PV
generation/battery capacity. In this case it is desireable that K self-sufficient
microgrids share their resources with the remaining N-K resource deficient
microgrids in a coordinated fashion such that the nanogrid with highest resource
availability should supply more power in comparison to the rest of self-sufficient
nanogrids and the nanogrid with the highest resource deficiency should receive more
power in comparison to the rest of deficient nanogrids.
50
d. All the nanogrids are generating more power than their local requirements i.e. excess
power is available after fulfilling household load requirements and battery capacity.
Although the frequent occurrence of this situation can be largely avoided by
optimally designing the distributed resources (PV generation and battery storage) in
accordance with the regional irradiance and temperature profiles as discussed in [96].
Still a single occurrence of this situation may instigate grid instability and voltage
rises above the allowed tolerance. In this case, it is desirable to culminate the PV
generation and synchronize it with household load requirements.
e. All nanogrids are deficient in resources and they start demanding power, which may
result in grid voltage drop below specified tolerance and subsequent instability. In
this situation it is highly desirable that all household loads are shed and there is no
power sharing with the common DC bus, until the batteries are recharged again when
PV resources are available.
4.6. Proposed Adaptive Decentralized Control Scheme
The desired control objectives for the DGDSA of DC microgrid are realized through a
communication-less scheme having decentralized control for individual nanogrids. The
SOC based thresholds along with the information of bus voltage dictate the mode of
control for individual nanogrids, such that the overall microgrid is stable and is achieving
coordinating power sharing among distributed resources during all of its possible
operation modes [97].
4.6.1. Power Stage and Control Scheme for an Individual Nanogrid
Power electronic interface for the formulation of an individual nanogrid along
with its control scheme is shown in Figure 4.5. There are two DC-DC converters in each
nanogrid. Converter 1 is an isolated bidirectional converter and is responsible for
controlled power sharing among nanogrids through interconnected DC bus. Converter 2
is a step down converter and is responsible for optimal power extraction from PV panels.
Battery acts as buffer between converter 1 and converter 2 and is responsible to keep the
voltage fixed at the local bus to which household load is connected.
51
Figure 4. 5 Power Electronic Interface and Control Schemes for Converters Employed in An
Individual Nanogrid Unit [97]
4.6.1.1. Multi-mode Adaptive Control Scheme for Bidirectional Converter Integrated
with DC bus
For each nanogrid i, control mode for its bus interfaced converter Conv1i is determined
by an adaptive controller on the basis of bus voltage VB and state of charge of its battery
SOCi. The SOCi of the battery is approximated by a simple Columb counting method, as
governed by (4.7) and is based upon the ideal energy balance at ith
local bus given by
(4.8):
Measure and Initialize SOCi, VB , Vref
SOCi<SOCmin
VB VL
Set Iref=1/Rd*(VL-VB)
Set Ii
ref=Irated*[(SOCi/SOCmin)-1]
SOCi<SOCmax
VB Vref
Set Iref=kc*(Vref-VB)
Set Iref=kd*(Vref-VB)
VB > VH
Set Iref=Irated*[(SOCi-SOCmax)/(100-SOCmax)]
Iiin
Iiload
IiL
VB
Di
VB
di
Control for Conv2i
`
Multi-Mode Adaptive Algorithm for Conv1i
No No
NoNo No
Yes
Yes Yes Yes
Iiload
Set Iref=1/Rd*(VH-VB)
Iiref
PWM
ViPV
MPPT Control
IiPV
Iiin
PI Controller
Iiload
SOCi>SOCmax
andVB VH
SOCi
No
Mode Switching for Conv2i= Signal Flow
= Power Flow
Yes
SOCi
I iref
-Ira
ted
SOCmin
VB
I iref
I rate
d
VL
1/RdVB
I iref
I rate
d
Vref
-Ira
ted
SOCmin
SOCmax
SOCmax
SOCminVB
I iref
I rate
d
VH
1/Rd
SOCi
I iref
I rate
d
100SOCmax
DC Loads Battery
SOCi)
Conv1iConv2i
VB
IiL
PWM
PI Controller
Control for Conv1i
SOCi
Vref
Multi-ModeAdaptive Algorithm
(Vib,
nanogridlocal bus
DC Bus
a) Power stage of ith nanogrid
b) Control stage of ith nanogrid
c) Multi-mode adaptive algorithm used in control stage of conv1 at ith nanogrid
PV Panel Isolated Bidirectional Converter
Buck Converter
52
tdIIIVC
SOCtSOC
T
L
i
load
i
in
i
b
i
i
ii 0
1
0 (4.7)
tdIIIVttPttPttP
T
L
i
load
i
in
i
b
i
L
i
load
i
PV
i 0
(4.8)
Where, SOCi(0) is the initial state of charge for the battery at ith
nanogrid, Ci is its
rated energy capacity (Wh), Iiin
is the current provided by PV panels after buck converter
(Conv2i), Ilload
is the current demanded by household DC loads, IiL is the current supplied
by the nanogrid to the DC bus, PiPV
(t) is the power generated by PV panel at time t whose
rated capacity is PPV
(Wp), Piload
(t) is the power demanded by household at time t whose
rated load capacity is Pload
(W) and Vib(t) is the time varying voltage of the battery whose
rated voltage is Vb. By convention IiL and Pi
L values are positive, when current and power
is being supplied by the nanogrid to the DC bus and negative when current and power is
being demanded by the nanogrid for household load or battery charging. SOCi of the
battery is considered as the resource availability index in this distributed structure. In
order to ensure the coordinated operation along with enhanced battery life time, upper
and lower threshold on the battery state of charge are defined as SOCmax and SOCmin. A
value of SOCi below SOCmin indicates that individual nanogrid is deficient in resources
and any further discharge below this point will badly deteriorate the battery life. So,
individual household loads must be shut down with the help of a relay and it must start
absorbing power to achieve the minimum sustainability level i.e. SOCmin. An inner loop
current control is used to govern the control of the converter through PI controller such
that the deficient nanogrid may demand power in accordance with its extent of
deficiency. In order to maintain the voltage of the microgrid within the stable higher and
lower limits i.e. between VH and VL (generally taken ± 2 to ±5 % of the rated bus voltage
Vref), following droop conditions are employed for SOCi < SOCmin
LB
LB
min
VV if N][1,i ; 1
VV if N][1,i ; 1
BL
d
ref
i
i
rated
ref
i
VVR
I
SOC
SOCII
(4.9)
53
Where, Irated is the rated charging current for the battery, specified by manufacturer
datasheet and Rd is the virtual droop resistance based upon the power ratings of the
converter.
Similarly, a value of SOCi higher than SOCmax indicates that nanogrid has very
high resource availability and needs to supply power to the neighboring nanogrids based
upon the acceptability condition of the microgrid dictated by its upper acceptable voltage
limit VH. Therefore, for SOCi > SOCmax
HB
HB
max
max
VV if N][1,i ; 1
VV if N][1,i ; 100
BH
d
ref
i
i
rated
ref
i
VVR
I
SOC
SOCSOCII
(4.10)
In between SOCmax and SOCmin thresholds, the control of the converter is dictated by
SOC based adaptive IV droop function. For VB higher than Vref, net power supplying
nanogrids are more than the net power demanding nanogrids, therefore, for this condition;
ith
nanogrid needs to absorb power to keep the microgrid stable. However, in order to
make this a coordinated power sharing, the charging droop coefficient Kc has been defined
as a function of droop resistance Rd and SOCi, such that among all the nanogrids in this
range, the nanogrid with relatively lower state of charge absorbs higher amount of power
in comparison to the other nanogrids. The droop coefficient in this range has been varied
linearly between 0.5 Rd to Rd from SOCmin to SOCmax. Other linear and non- non-linear
variations of droop function can also be considered, such that it does not violate the
stability conditions for microgrid operation [66]. This trend is also shown in Figure 4.5
(c).
minmax
min2
1,
SOCSOC
SOCSOC
RSOCRK
i
d
idc (4.11)
Similarly, for VB lower than Vref, each individual naogrid has to supply power in
accordance to its resource availability index. For this range discharging droop co-efficient
Kd as a function of SOCi and droop resistance Rd is given by (4.12) and is also shown in
54
Figure 4.5 (c). In this case, the nanogrid with highest resource availability (SOC) will
supply more in comparison to the nanogrid having relatively lower SOC.
minmax
min1
1,
SOCSOC
SOCSOC
RSOCRK
i
d
idd (4.12)
Based upon (5) and (6), the SOC based I-V adaptive droop control in the range of SOCmin
≤ SOCi ≤ SOCmax is shown in Figure 4.5 (c) and expressed as follows:
refB
refB
VV if N][1,i ;
VV if N][1,i ;
Brefc
ref
i
Brefd
ref
i
VVKI
VVKI
(4.13)
If all the converters are within this range of operation, there will be zero current
sharing among the contributing nanogrids. Therefore, DGDSA implementation with this
type of control will offer minimum distribution losses.
An adaptive algorithm for the calculation of Iiref
based upon (4.9) - (4.13) and local
measurements of VB and SOCi dictates the mode of operation as shown in Figure 4.5(c).
An inner loop current control is then used to control the current of Conv1i through PI
controller that generates the duty cycle Di given by (4.14), where, kp and ki are the
proportional and integral constants for PI controller respectively.
t
L
i
ref
ii
L
i
ref
ipi dtIIkIIkD0
(4.14)
4.6.1.2. Scheme for Switching between MPPT and Current Control Modes for the
Converter Integrated with PV Panel
The buck converter of each nanogrid (Conv2i) at the output of PV panel is
responsible for optimal battery charging. Maximum power point tracking (MPPT) control
is widely used in PV based systems for the extraction of maximum power out of incident
solar energy. Various schemes for MPPT under uniform and non-uniform irradiance have
been discussed in the literature [73, 74]. In this article, the perturb and observe algorithm
is employed due to its simplicity and low computational complexity [73]. The algorithm
processes PV panel voltage ViPV
and current IiPV
to generate duty cycle di for maximum
power extraction from PV panel at a given solar irradiance. In most of its operation range
55
Conv2i will operate in MPPT mode however, based upon the measurements of SOCi and
VB, Conv2i may shift its operation from MPPT mode to inner loop current control mode
such that it culminates its power generation from MPPT to household load current
requirements Iiload
only. Thus, for SOCi > SOCmax and VB ≥ VH, Conv2i will operate in
inner loop current control mode through a PI controller that will generate duty cycle di
given by:
t
in
i
load
ii
in
i
load
ipi dtIIkIIkd0
'' (4.15)
where, kP’ and ki
’ are proportional and integral constants of PI controllers employed for the
control of conv2i.
4.7. Results, Discussions and Conclusions for Adaptive Decentralized Control
For the validation of proposed scheme various test cases are analyzed via
simulations and hardware in loop (HIL) facilities in the laboratory.
4.7.1. Simulation Results for Decentralized Control
Simulations are carried out on MATLAB/Simulink using physical models of the
converters and control schematic shown in Figure 4.5 (c). Various parameters for
simulation are shown in Table I. In order to have a better illustration of results, PiPV
(t) is
assumed equal to Piload
(t) for test case 4.7.1.1 to 4.7.1.3.
4.7.1.1. All nanogrids are within specified thresholds of SOC
In this scenario, the batteries of all nanogrids are assumed to be within specified
thresholds of SOC i.e. 4 3, 2, 1,i; maxmin SOCSOCSOC i.
This case is evaluated with and without communal load and results for variations
in bus voltage, current sharing among contributing nanogrids and accelerated simulations
(0.5 hr) for SOCi are shown in figures 4.6 and 4.7 respectively. It can be seen that after
starting transient, if there is no communal load, current sharing among the nanogrids is
almost zero, i.e. each nanogrid is working independently, without supplying or
demanding power from DC bus. So, their SOC‘s remain constant in this region and
56
distribution losses are zero, despite load requirements of each household is being
fulfilled.
At t= 0.025 s, a communal load of 500 W is applied due to which voltage of the
DC bus drops from 48 V to 47.3 V and each nanogrid starts contributing for communal
load based upon its availability index i.e. according to its SOCi value. Therefore, all
nanogrids are supplying power (being discharged) based upon the modified droop Kd(Rd,
SOCi) given by equation (4.12). Consequently, the nanogrid with highest SOC,
contributes more towards communal load and its SOC decreases at a rapid slope in
comparison to other nanogrids in the system. (ΔSOC1 = 1.92% in comparison to ΔSOC4
= 2.52% at the end of simulation).
Figure 4. 6 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L, I2
L,
I3L and I4
L) (left Y -axis) in case 1 (simulation results)
57
Figure 4. 7 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 1 (simulation results)
Also, it is worth noting that upon a step change on bus loading (communal load
application), the response time of proposed control scheme is very fast and system
achieves new steady state rapidly. Thus the proposed control method exhibits fast
dynamics due to I-V droop method which uses only one controller delay in comparison to
V-I voltage droop which uses two controller delays resulting in relatively slower system
dynamics. The detailed parameters for the simulated case studies are tabulated in Table I.
It is worth mentioning here, that switching frequency fsw is kept 10 kHz in the simulation
case studies. Generally, higher frequencies may result in lower filter requirements in the
converter operation. The selection of fsw in this particular case study is in accordance with
the compatibility of our dSpace based hardware in loop (HIL) setup discussed below.
Power level for individual households is kept 200 W in this case study, which is
relatively higher in comparison to previous case studies discussed in chapter 3, where we
considered power levels in the range of 30-40W. The selection of higher power level is
made to validate the efficacy of proposed control scheme for higher power levels for
improved livelihood level village scale electrification.
58
Table 4. 1 Parameters of Simulated Case Study [97]
Description of the Parameter Symbol Value Description of the
Parameter Symbol Value
No. of Nanogrids/ households N 4 Maximum threshold of
battery SOC SOCmax 80%
DC bus capacitance CB 10mF Minimum threshold of
battery SOC SOCmin 30%
Inductance of each Conv1i L1 500μH Reference voltage for DC
bus Vref 48V
Switching frequency for
Conv1i and Conv2i fsw 10kHz Initial Voltage of DC bus VB0 24V
Rated power of each PV panel PPV
500Wp Lower limit on DC bus
voltage VL 45.6V
Rated household load Pload
200W Higher limit on DC bus
voltage VH 50.4
Battery capacity for each
nanogrid C 2400Wh
Droop Coefficient for
Conv1i Rd 0.218Ω
Rated Charging current for the
battery Irated 10A
Proportional and integral
parameters (Conv1i) kp ,ki
0.33,
15
Rated voltage of each battery Vb 24V
Proportional and integral
parameters (Conv2i) kp
’ ,ki
’ 0.5, 50
4.7.1.2. All nanogrids are within specified thresholds of SOC except one which is below
minimum threshold of SOC
In this scenario, the batteries of three nanogrids are assumed to be within
specified thresholds of SOC, while battery of fourth nanogrid is assumed to be below
minimum threshold of SOC, i.e. min4maxmin ; 3 2, 1,i; SOCSOCSOCSOCSOC i
Results for bus voltage profile, current sharing among contributing nanogrids and
accelerated simulations (1 hr) for SOCi are shown in Figures 4.8 and 4.9 respectively. It
can be observed from these figures that the nanogrid 4 which is deficient in resources is
demanding current as dictated by (4.9). Therefore, nanogrid 4 is absorbing power with a
negative value of I4L=-5.1A
(charging current), and its SOC4 is increasing with a high
slope.
59
Figure 4. 8 DC DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L,
I2L, I3
L and I4
L) (left Y-axis) in case 2 (simulation results)
Figure 4. 9 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 1 (simulation results)
All other nanogrids are supplying power (being discharged) based upon the
modified droop Kd(Rd, SOCi) given by equation (4.12). The modified droop ensures the
coordinated power sharing such that the nanogrid having higher SOC supply more
current in comparison to nanogrid having relatively lower SOC. For instance in figure
4.8, nanogrid 1 with SOC1=35% is supplying 1.2A, nanogrid 2 with SOC2=55% is
supplying 1.6A and nanogrid 3 with highest SOC3=75% is supplying 2A respectively.
Thus resource sharing is set in proportional to resource availability without any physical
60
layer of communication among the contributing resource and decentralized,
communication-less, yet, coordinated control is realized with the proposed scheme. It is
worth noting that sum of three supplying currents is lower than the absorbing current in
microgrid 4. This is in accordance with the energy balance in (4.7) and (4.8) as the
batteries of three supplying nanogrids are at relatively higher voltage in comparison to
demanding microgrid whose voltage is way below rated voltage due to very low value of
SOC.
Figure 4.9 shows the accelerated simulations for one hour SOCi variations in
batteries of four nanogrids. The battery with higher SOC is being discharged at a rapid
rate in comparison to the batteries having relatively lower SOC as evident by the
observation that changes in SOCi values from the start till end of the simulation are
ΔSOC1=2.09%, ΔSOC1=2.61% and ΔSOC1=3.1% respectively (ΔSOC1 < ΔSOC2<
ΔSOC3). This is in accordance to the desired objectives of the control scheme such that
the battery with highest resource availability i.e. battery 1, is supplying more in
comparison to other two supplying batteries which are also supplying in proportion to
their resource availability.
4.7.1.3. All nanogrids are within specified thresholds of SOC except one which is above
maximum threshold of SOC
In this scenario, the batteries of three nanogrids are assumed to be within
specified thresholds of SOC, while battery of fourth nanogrid is above maximum
threshold of SOC, i.e. max4maxmin ; 3 2, 1,i; SOCSOCSOCSOCSOC i .
Therefore, this case evaluates the operation of microgrid with three of the contributing
nanogrids (1, 2, 3) deficient in resources, while fourth one have enough resource
availability to support the resource deficient nanogrids.
Results for bus voltage profile, current sharing among contributing nanogrids and
accelerated simulations (1 hr) for SOCi is shown in Figures. 4.10 and 4.11 respectively.
Since the initial SOC4(0)
is above threshold i.e. 90%, therefore, in this scenario, nanogrid
4 is supplying power as dictated by equation (4.10) with I1L=4.98A, while other three are
absorbing power (their batteries are being charged) based upon the modified droop Kc(Rd,
SOCi) given by equation (4.11).
61
Figure 4. 10 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L, I2
L,
I3L and I4
L) (left Y-axis) in case 2 (simulation results)
Figure 4. 11 DC bus voltage VB profile (right Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 2 (simulation results)
It can be observed from Figures 4.10 and 4.11 that power sharing via modified droop
ensures resource distribution based upon the availability index. Therefore, nanogrid with
initial SOC3(0)
=75% (highest SOC and highest resource availability) is being charged
with the lowest current I3L = -1.28A in comparison to nanogrid with SOC2
(0)= 55% and
nanogrid with SOC2(0)
= 35% which are being charged at I2L=-1.73A and I3
L =-2.18A
respectively. Moreover, the changes in SOCi from start till end of the simulation are also
in accordance with the modified droop given by equation (4.11) and shown in Figure 4.5
(c), such that ΔSOC1=0.96%, ΔSOC1=0.49% and ΔSOC1=0.2% respectively (ΔSOC1 <
ΔSOC2< ΔSOC3).
62
4.7.1.4. Multi-mode switching of an individual nanogrid
In order to realize the working of an individual nanogrid in all possible threshold
ranges and to visualize the multi-mode switching based upon the SOC thresholds,
nanogrids 2,3 and 4 are considered to be working within specified maximum and
minimum thresholds of SOC with SOC2<SOC3<SOC4 while, nanogrid 1 is considered
below threshold in the start of simulation. It is assumed that PV power produced within
the first three nanogrids is in accordance with their household load requirements; while
incident irradiance and associated PV power produced within nanogrid 1 is higher than
its household load requirements. Therefore, based upon the energy balance given in (4.7)
and (4.8), SOC1 will increase from values below SOCmin to values above SOCmax,
Consequently, Conv11 will switch its operating modes accordingly.
Figure 4.12 shows the variations in current sharing among contributing nanogrids
(I1L, I2
L, I3
L and I4
L) based upon the accelerated SOC variations of an individual nanogrid
(SOC1). Accelerated SOC variations at nanogrid 1 are achieved by considering reduced
battery capacity (C/5) and high incident irradiance (1000W/m2). It can be observed that
when SOC1<SOCmin, nanogrid 1 is demanding current with negative value of I1L as
dictated by equation (4.12). Current demanded by naogrid 1 I1L
decreases as SOC
increases and becomes almost zero upon reaching SOC1 =SOCmin=30%.
It is worth noting that within this range of operation, the current supplying
capability of the remaining three microgrids is governed by the modified discharging
droop Kd(Rd, SOCi) (given by equation (6) and its visual representation is also shown in
Figure 4.5 (c), such that nanogrid 4 having highest SOC is supplying maximum current,
while nanogrid 2, having lowest SOC is supplying lower current. In mid operation range,
i.e. within specified limits of thresholds, all nanogrids are sharing zero current, therefore,
in this range distribution losses are comparatively negligible. Also, the inter-mode
transition is very fast and smooth with the proposed strategy. For SOC1>SOCmax,
nanogrid starts supplying current in accordance with (4.10) and value of I1
L keeps on
increasing with increase in SOC1. Therefore, out of current demanding nanogrids,
nanogrid 2 with the lowest SOC is demanding the highest current while nanogrid 4 with
the highest SOC is demanding the lowest amount of current.
63
Figure 4. 12 Nanogrid 1 SOC1 variation in the various thresholds ranges (left Y-axis) and associated
current sharing among the contributing nanogrids in case 3 (right Y- axis) (simulation results)
4.7.1.5. All nanogrids are above maximum threshold of SOC and surplus PV power is
available
In this scenario, critical conditions of islanded operation are evaluated by
considering that all the nanogrids are above maximum threshold and surplus PV power is
available due to high incident irradiance (1000W/m2) i.e. 4, 3 2, 1,i; max SOCSOCi .
Each naogrid will tend to supply power to the DC bus based upon the equation (4.10),
therefore, its voltage will rise until it reaches to VH. At VH, the proposed droop function
will reduce the current supply to zero and will try to keep the voltages fixed at VH. Since,
the batteries are already above maximum threshold, therefore, any local PV generation
PiPV
, higher than local household requirements Piload
will overcharge the battery and
cause DC bus voltage to rise above the maximum limit VH, thus instigating instability in
the system. At this point, the control schematic of conv2i changes its control from MPPT
to inner loop current control mode as shown in Figure 4.5 (c). Therefore, IV droop
control mode (constant droop coefficient Rd) of Conv1i stabilizes the DC bus voltage at
VH and Conv2i ensures stability by culminating generation capability of each nanogrid
according to the load requirements at individual household level.
64
Figure 4.13 shows that when DC bus voltage is below maximum threshold VH,
each nanogrid contributes for current according to its SOCi. Once the voltage reaches to
VH, current contribution from each nanogrid becomes zero, and further rise in voltage is
restricted to VH. Before attaining VH, each Conv2i is operating in MPPT mode, thus
extracting maximum power (500 W at incident irradiance of 1000 W/m2). However, once
DC bus voltage attains its maximum value VH, the PV generation is limited according to
household load requirements.
Figure 4. 13 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L, I2
L,
I3L and I4
L) (left Y-axis) in case 4 (simulation results)
Figure 4. 14 Power generated by PV panels in nanogrid 1 P1PV
(righy Y-axis) and output current I1in
of conv21(left Y-axis) in case 5 (simulation results)
65
Figure 4. 15 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 4 (simulation results)
This is shown in figure 4.14, where Conv21 of nanogrid 1 is working in MPPT
(P&O) mode and generating power around 500W in the start of simulation. At t=0.027s,
VB reaches to its maximum allowable limit, therefore, Conv2i shifts is control from MPPT
to current control mode, therefore, the output current of conv2i i.e. I1in
coincides with load
current I1PV
waveform as shown in Figure 4.14. This has been also shown in figure 4.15
where, SOCi of each converter is increasing due to PV generation higher than load
requirements, when VB is below VH. After VB becomes equal to VH, due to change in
control mode of Conv21 and associated limited PV generation, the SOC of the battery
does not rise any further and becomes constant onwards.
4.7.1.6. All nanogrids are below threshold of SOC and PV generation is not available
In this case the batteries of all nanogrids are assumed to be below threshold level
and PV generation is not available i.e. 4, 3 2, 1,i; min SOCSOCi . This is the typical case
at night or in cloudy days when no sunlight is available for PV power production.
Since PV generation is not available and all the batteries are already blow
minimum threshold SOCmin, therefore, any local load demand can further discharge
batteries and cause DC bus voltage to collapse below minimum threshold level VL.
66
Figure 4. 16 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 4 (simulation results)
Therefore, all the local loads are turned off in this condition through a relay and
DC bus voltage is limited to lower threshold of voltage VL through IV droop with
constant droop coefficient as shown in figure 4.5 (c). Thus, any further power sharing
among the contributing nanogrids is restricted to maintain the bus voltage level and
battery SOCi level of individual batteries as shown in figure 4.16. This condition is
maintained until PV irradiance and associated PV generation is available again to charge
the batteries above SOCmin. Thus the proposed control scheme ensures stability even in
the critical conditions of operation, while maintaining the voltage across the grid at lower
threshold limit.
4.7.2. Experimental Results for the Validation of Proposed Adaptive Algorithm for Conv2i
In order to realize various power sharing scenarios during normal operation of
microgrid and to validate the proposed decentralized control scheme, an hardware in loop
(HIL) experimentation is conducted using Danfoss converters and dSpace RTI 1006
platform in Aalborg University (AAU), microgrid laboratory capable to perform real time
data acquisition and control operations [98].
The functioning of adaptive algorithm for the control of Conv1i (shown in Fig. 2.5
(c)) is evaluated using HIL experimentation whose schematics and hardware setup is
shown in figures 4.17 and 4.18 respectively. PV power is emulated using power supply
and battery model is emulated using (4.7) and (4.8). Since functioning of Conv2i is to
67
ensure optimal PV generation in normal and critical conditions of operation and in the
current setup PV power is being emulated, therefore, control of Conv2i is not
implemented for experimentation. Based upon the output of emulated battery model i.e.
based upon SOCi, adaptive algorithm adjusts the mode of operation for each individual
converter. Various parameters of experimentations are further detailed in Table 4.2.
Table 4. 2 Parameters of Experimental Case Study [97]
Description of the
Parameter Symbol Value
Description of the
Parameter Symbol Value
No. of Nanogrids/
households N 3
Maximum threshold of
battery SOC SOCmax 80%
DC bus capacitance CB 3.3mF Minimum threshold of
battery SOC SOCmin 30%
Inductance of each
Conv1i L1 8.6H
Reference voltage for DC
bus Vref 48V
Stray resistance for
Inductors ri 0.1Ω Initial Voltage of DC bus VB0 24V
Switching frequency
for Conv1i fsw 10kHz
Lower limit on DC bus
voltage VL 45.6V
Rated power of each
PV panel P
PV 500Wp
Higher limit on DC bus
voltage VH 50.4V
Description of the
Parameter Symbol Value
Description of the
Parameter Symbol Value
Rated household load Pload
200W Proportional and integral
parameters (Conv1i) kp ,ki
0.02,
0.1
Battery capacity for
each nanogrid C 2400Wh
Droop Coefficient for
Conv1i Rd 0.25Ω
Rated charging current
for battery Irated 5A
68
Figure 4. 17 Schematics of experimental setup at microgrid laboratory
Figure 4. 18 Hardware setup for practical measurements
PWM
1
ME
AS
UR
EME
NT
1
ME
AS
UR
EME
NT
2
ME
AS
UR
EME
NT
3
S1
S4
DC
PO
WER
SU
PPLY
Communal Load
C
PWM
2
C
PWM
3
S3
C
Conv11Conv12Conv13
Nanaogrid 3 Nanaogrid 2 Nanaogrid 1
io1vBio2io3
S2
Multi-ModeAdaptive Algorithm
PiPVPi
loadVB*ioi
Emulated Battery Model
SOCio
vBvB
SOCi
dSPACE
Bidirectional Converters
Monitoring Platform
Oscilloscope
DC Load Bank
69
4.7.2.1. All nanogrids are within specified thresholds of SOC
In this scenario, the batteries of all nanogrids are assumed to be within specified
thresholds of SOC i.e. 3 2, 1,i; maxmin SOCSOCSOC i.This case is evaluated with
and without communal load of 135W and results for variations in bus voltage, current
sharing among contributing nanogrids and accelerated simulations (1 hr) for SOCi are
shown in figures 4.19 and 4.20 respectively. Measured results are in accordance with the
simulation results as without communal load, the current sharing among the contributing
nanogrids is almost zero (slightly higher than zero due to ESR of individual capacitors,
which otherwise was zero in case of simulation result due to ideal capacitor) and on
application of communal load the current sharing is in proportional to SOCi value. For
instance, battery of nanogrid 1 with initial SOC10=35% is supplying 0.79 A, battery of
nanogrid 2 with initial SOC20=55% is supplying 1.05 A and the battery of nanogrid 3
having initial SOC30=75% is supplying 1.33 A for communal load application. The
change in SOC from start till end of the simulation is in accordance with the SOC
availability i.e. ΔSOC1 = 0.49 %, ΔSOC2 = 0.66 % and ΔSOC3 = 0.84 %. Also the initial
transition and transition from no load to communal load scenario is fast and smooth as
shown in figuress 4.19 and 4.20 respectively.
Figure 4. 19 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L,
I2Land I3
L ) (left Y-axis) in case 1 (measured results)
70
Figure 4. 20 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 1 (measured results)
4.7.2.2. All nanogrids are within specified thresholds of SOC except one which is above
maximum threshold of SOC
In this scenario, the batteries of three nanogrids are assumed to be within
specified thresholds of SOC, while battery of fourth nanogrid is above maximum
threshold, i.e. max1maxmin ; 3 2,i; SOCSOCSOCSOCSOC i . Results for bus voltage
profile, current sharing among contributing nanogrids and accelerated simulations (1
hour) for SOCi are shown in figures 4.21 and 4.22 respectively.
Figure 4. 21 DC bus voltage VB profile (righy Y-axis) and current sharing among nanogrids (I1L,
I2Land I3
L ) (left Y-axis) in case 2 (measured results)
71
Figure 4. 22 DC bus voltage VB profile (righy Y-axis) and battery SOC for contributing nanogrids
(SOC1, SOC2, SOC3 and SOC4) (left Y-axis) in case 2 (measured results)
Results verify that the nanogrid 1 having SOC higher than maximum threshold is the
supplying nanogrid while remaining two nanogrids demand according to their resource
availability i.e. nanogrid 2 with higher value of initial SOC20=60% is absorbing
relatively lower current in comparison to nanogrid 3 having higher value of initial
SOC20=40%. Therefore, change in SOC for absorbing nanogrids from start till end of the
simulation is in accordance with resource availability i.e. ΔSOC2 = 0.95% and ΔSOC3 =
1.2% with (ΔSOC3 > ΔSOC2).
4.7.2.3. Multi-mode switching of an individual Nanogrid
Nanogrids 2 and 3 are considered to be working within specified maximum and
minimum thresholds of SOC with SOC2<SOC3 while, nanogrid 1 is considered below
threshold in the start of simulation. It is assumed that PV power produced within
nanogrids 2 and 3 is in accordance with their household load, while PV power produced
within nanogrid 1 is higher than its household load requirements. Therefore, based upon
the emulated model of battery, SOC1 will increase from values below SOCmin to values
above SOCmax, and Conv11 will switch its operating modes accordingly.
Figure 4.23 shows the variations in current sharing among contributing nanogrids
(I1L, I2
L, and I3
L) based upon the accelerated SOC variations of an individual nanogrid
(SOC1). Accelerated SOC variations at nanogrid 1 are achieved by considering reduced
72
battery capacity (C/10). From Fig. 4.23 it can be observed that for region SOC1 < SOCmin
, nanogrid 1 is demanding current with negative value of I1L and nanogrid 2 and 3 are
supplying in proportion to their SOC, therefore, battery of nanogrid 3 having initial
SOC3(0)
=60% is supplying more current in this region in comparison to nanogrid 2 having
SOC2(0)
=40%. This is in accordance with the simulation results shown in Fig. 5 and IV
droop function as shown in Fig. 4.5 (c). The slope of droop increases with SOC in this
particular region as shown by the arrow in Fig. 4.23 which is in accordance with equation
discharging droop coefficient Kd(SOCi, Rd) given by equation (4.12). Similarly for
intermediate region, when all the batteries are within their limits of threshold, the current
contribution from each nanogrid becomes zero; therefore, it also validates our
consideration of almost zero distribution losses in the range of SOCmin≤SOCi≤SOCmax.
Finally, in the region when SOCi > SOCmax, nanogrid 1 start supplying current with
positive value of I1L, while nanogrid 2 and nanogrid 3 absorb power in proportion to their
resource deficiency.
Current sharing is controlled by charging droop coefficient Kc(SOCi, Rd) given by
equation (4.11) such that nanogrid 3 having SOC3(0)
= 60% is absorbing less current in
this region in comparison to nanogrid 2 having SOC2(0)
=40%. Therefore, the slope of
droop increases with the decreasing SOC as shown in simulation results in figure. 4.12.
Figure 4. 23 Nanogrid 1 SOC1 variations in the various threshold ranges (left Y-axis) and associated
current sharing among the contributing nanogrids (right Y- axis) in case 3 (measured results)
73
4.7.3. General Conclusions drawn from the results of Decentralized Control
An adaptive IV droop method for the decentralized control of a PV/Battery based
distributed architecture of an islanded DC microgrid is presented and its validity is
demonstrated with simulations and hardware in loop experimentation. The stability of
islanded microgrid in critical operation conditions is ensured via controlled
synchronization between generation resources and load requirements. The proposed
control method is highly suitable for the rural electrification of developing regions
because it (i) enables coordinated distribution of generation and storage resources at a
village scale, (ii) reduces distribution losses associated with delivery of energy between
generation and load end; (iii) decentralized controllability omits the need of central
controller and associated costly communication infrastructure, and (iv) enables resource
sharing among the community to extract the benefit of usage diversity at a village scale.
Results have also shown that adaptive IV droop algorithm enables fast and smooth
transitions among various modes of microgrid operation based upon the resource
availability in individual households of the village. Therefore, the implementation of
proposed control method on PV/battery based DGDSA of islanded DC microgrid will
enable high efficiency and better resource utilization in future rural electrification
implementations.
74
5 Optimal Planning and Design of
Low-Voltage Low-Power Solar DC
Microgrids
Along with technical improvements in the structure and operation of new
microgrids, it is also important to optimize the overall system through right selection of
system components. Since major existing commercial scale DC microgrid
implementations involve central placement of resources, therefore, it is important to first
evaluate optimal planning of distribution architecture and sizing of various system
components such as solar panels, batteries and distribution conductors to minimize the
overall system cost. Therefore, this chapter is dedicated to develop a framework for
optimal planning and design of low-power low-voltage DC microgrids for minimum
upfront cost. The analysis is based on a) region-specific irradiance and temperature
profiles, b) distribution loss analysis and c) optimum component sizing (storage,
conductor and PV panel) requirements based upon an energy balance model for a 24-hr
operation. Based upon the presented framework, the merits of tailoring distribution
architecture for maximizing the system utility in the planning of future microgrid
deployments are analyzed [99]. The presented framework is valid for existing central
architectures as well as distributed generation distributed storage architecture presented in
this work [96].
5.1. Need for Optimal Planning and Design of Solar PV based DC Microgrids
Although low-voltage, low-power solar PV-based DC microgrid systems are
becoming popular for off-grid rural electrification, however, formal analysis on optimal
component sizing and loss evaluation is not addressed in the literature. Generic microgrid
systems planned without taking the regional characteristics into consideration are
significantly oversized and are not a good fit for all environments and regions. Thus,
there is a need to plan efficient distribution schemes based upon the detailed loss analysis
75
along with quantification of optimal system sizing incorporating local conditions for
overall cost minimization and enhanced system utilization. Various critical parameters
that affect the optimal sizing of system components needs to be identified and analyzed
for optimal system assessment. The component sizing in a solar based DC microgrid is a
function of overall power requirements at each household, regional profile of temperature
and irradiance and distribution losses incurred in the path of power flow from source end
to utilization end. In order to analyze the impact of distribution architecture on optimal
component sizing, distribution losses to actual structure of the village and spatial
distribution of houses in it must be related. Therefore, in order to design optimal
microgrid systems, quantification of distribution losses with respect to distribution
architectures in line with common settlements must be considered.
Kindly note that the focus in this chapter is on lowest cost topologies for basic
rural electrification. Therefore, power levels of individual households for the optimal
framework analysis are kept 5W and 10W, in accordance to the typical commercially
available microgrid solution e.g. Mera Gao Power (MGP) in India, which supplies 5 W of
electricity to the subscribing households. High power microgrids (with household
provisions for several hundred watts) are generally unviable (due to high up front cost
and lack of micro-financing options) and not commonly implemented for self-sustained
rural electrification. Therefore, we restrict our analysis to low-power and low voltage DC
microgrids which are considered safe for direct touch and potentially do not create any
electric shock or fire hazards when implemented at lower voltage levels (24V and 48V)
with 100W name plate capacity. Importantly, the methodology presented in this work, for
the optimal component sizing and optimal distribution architecture planning, is generic
and equally applicable for higher power DC microgrids. However, lower voltages (24V
and 48V) are not suitable for very high power system, as distribution losses would be
very high for low distribution voltages.
5.2. Common Village Orientations
In order to design an efficient power distribution architecture that ensures the
optimal power flow from source end to load end, it is important to analyze the spatial
76
distribution (orientation) of houses commonly found in villages across the developing
countries. Typically, two main arrangements of houses are found:
a) Linear arrangement in which houses are generally situated alongside a central
street/road.
b) Clustered arrangement in which houses are situated in independent fields or in
clusters of multiple huts/homes [100]. Some other similar orientations also occur such as
Zulu‘s in Southern Africa where people built houses shaped like beehives. They built their
houses in a circular fenced compound forming a cluster [101].
Northern Africa and Namibia show highly clustered settlement and population
distribution patterns [102]. However, in Asia, primarily in South East Asian countries such
as India, Pakistan and Bangladesh, the most common forms of rural settlements is linear
arrangement of houses which are situated across roads in order to facilitate access to
infrastructure facilities, markets and resources [88, 103]. In order to electrify these
villages, generally standard radial systems are installed irrespective of the structure of
village and orientation of houses. However, the distribution efficiency of these systems
can be significantly enhanced by taking the structure of village into account and fitting the
right microgrid distribution architectures on it.
In this work, a system model for these linear architectures is developed and
optimization framework for planning and design of low-power and low-voltage DC
microgrids situated in these linear settlements is formulated. In a clustered non-linear
architecture, houses may not be in close proximity and may exist in clusters which are
wide apart. Therefore, an additional concern of PV resources placement along with power
processing and storage unit (PPSU) placement arises because it directly affects the
distribution losses and associated optimal sizing calculations. Due to inclusion of an
additional objective i.e. optimal placement of PV generation and storage unit, the overall
objective function becomes a combinatorial optimization problem. The work on the
development of framework for optimal sizing of the villages with clustered orientations is
under process and its findings will be shared with the community in future contributions.
77
5.3. Proposed Distribution Architectures for Centralized Linear DC Microgrids
Typically, in order to electrify the linear villages with conventional centralized DC
microgrids, standard radial systems with single generation and storage hub are installed
[20, 23, 104, 105]. However, the distribution efficiency of these systems can be
significantly enhanced by considering a second generation hub if the provision is
available. Based upon the structure of village and spatial distribution of houses, we
propose two broad power distribution architectures. The proposed architectures cover
almost all types of village configuration discussed in section 5.2. Therefore, we consider
linear distribution microgrid architecture with up to two generation and storage hubs.
These proposed structures can then be classified into C- distribution architecture and O-
distribution architecture depending upon the number of generation and storage hubs.
5.3.1. Linearly Distributed C- Architecture
The visual representation for linearly distributed C- architecture is shown in
figure 5.1. Distribution conductors are laid in a linear manner while generation and power
processing and storage units (PPSU) are placed at the start of a village, thus formulate a
C-like structure and is termed as linearly distributed C-architecture. This village
architecture is a simplistic model of villages found commonly in India [20].
Figure 5. 1 Topological Diagram of Linearly Distributed C-architecture with PV Generation and
Power Processing and Storage Units (PPSU) [96, 99]
PPSU
PV Panels
78
5.3.2. Linearly Distributed O- Architecture
Based upon the availability of the land, PV generation unit and PPSU may be
located at both ends of the central street. Thus, such a structure in which conductors are
laid in a linear manner, interconnecting generation and storage at both ends of the house
load formulates a linearly distributed O-architecture DC microgrid (figure 5.2). While
keeping the overall generation and storage capacity the same, the introduction of this
architecture having two similar generation and similar storage units on system enhances
its overall efficiency as discussed in subsequent sections.
Figure 5. 2 Topological Diagram of Linearly Distributed O-architecture with two (PPSU’s)[96, 99]
5.4. Energy Balance Model for Centralized DC Microgrids
In order to design an optimal system, it is important to analyze the system in
terms of load requirements, converter requirements, supply availability, storage size and
loss analysis. In this section we evaluate various loss elements in the operation of
microgrid. These losses mainly include i) panel losses in the PV output due to low
irradiance, soiling and mismatch, ii) Degradation in PV output due to temperature, iii)
DC/DC converter losses, iv) battery charge/discharge cycle losses and v) distribution
losses.
The output produced by PV panels is a function of incident irradiance and
temperature. However, due to panel losses and temperature degradations, output of PV
panel is generally lower than its name plate capacity. Low irradiance losses generally
vary linearly with the peak sun hours (PSH) of the incident irradiance. A mathematical
model to quantify low irradiance loss has been presented in [106]. The degradation in
PPSU
PV Panels
PPSU
PV Panels
79
output characteristics of PV Panels due to soiling and mismatch of cells and temperature,
discussed in [107, 108], are also incorporated. Therefore, considering panel losses and the
temperature degradation effect, S(t) is given by (5.1)
tIAtS tcmPL . (5.1)
Where, Itc is the temperature compensated irradiance and depend upon I(t), and Tcell and
is given by (5.3) [109].
tItTtT ambcell *01875.0 (5.2)
tITtI celltc 0045.0251 (5.3)
The output from PV panels is processed through MPPT DC/DC converter which
incurs losses in PPSU along with further DC/DC losses at the distribution panel of each
house (see figures 5.3 and 5.4). The efficiency of a typical DC-DC converter varies
significantly at low power levels (percentage loading). However, the converter losses can
be linearized and its efficiency can be considered as a constant for majority of its
operational range [110]. For instance Kolar et al. [110] shows that a typical DC-DC
converter exhibits slight efficiency variations (±2%), during its operation in between 25
to 95 percent of its rated output power loading. Therefore, the rating of CPPU converter
is approximated to a fixed value. For variable loading scenarios, detailed analysis on
converter losses may be included using (5.8) and (5.14) for accurate sizing estimations.
Battery charging/discharging cycle efficiency (ηB) depends upon the battery technology
and manufacturer specifications. For the system evaluation, lead acid battery is
considered, while its charging/discharging efficiency is considered constant for the
simplicity of analysis. Distribution efficiency is assessed by calculating distribution
losses in the system through Newton-Raphson Method modified for DC power flow
analysis [79]. Ploss(t) accounts for distribution losses and is a function of a) distribution
configuration in a village, b) distribution voltage level and c) permitted load levels to
each household and must be critically analyze. In order to quantify distribution losses, an
N-house village is modeled as a combination of interconnection resistance of the laid
conductors as shown in figures 5.3 and 5.4 respectively. For each configuration (C or O)
of the village, a unique conductance matrix G can be formulated depending upon the
80
spatial distribution between house and length of conductor laid and is given by (5.4) and
(5.5).
NN
NNN
N
N
RG
G
G
G
G
;
G G
G G
G G
N21
22221
11211
(5.4)
y x;
x; 1
xy
N
yxy
xy
g
ygG
(5.5)
This G-matrix can be used to calculate Ploss and maximal voltage dip (VD) as
discussed in [79] and are given by (5.6) and (5.7).
N
xxyyyxx
N
yxyloss VVVVVVGP
1 12
1
(5.6)
minmax
xx VVVD (5.7)
For energy balance model, we consider a typical village microgrid orientation
consisting of N houses electrified via solar PV panels having maximum power generation
capacity Smax (kWp) and a battery storage system having energy capacity CB (kWh). Solar
PV generation S(t) will vary with time depending upon the input irradiance and ambient
temperature, therefore, battery state of charge and associated state of energy EB(t) will
also vary with time depending upon S(t) and load requirements. Load demand of each
house i at any time is given by Pi(t).Therefore, for any time interval Δt, the balance of
energy at microgrid is given by (5.8)
t 1
tPttPtEttS loss
N
iiiBBMP (5.8)
Where, ηMP is the efficiency of DC/DC converter employed at PPSU responsible for
maximum power point tracking, battery charging and maintenance of distribution voltage
and ηi is the efficiency of DC/DC converter employed at the distribution panel of each
house. Since battery may take energy from PV panels or may supply power to the load at
any time t depending upon the net energy flux in the battery, ΔEB(t) can be positive or
81
negative. ΔEB(t) will be negative in durations of no solar power generation and the stored
energy in the battery will be used to meet the load demand.
5.5. System Model Formulation for Optimal Component Sizing of Centralized
DC Microgrids
In order to optimally size various system components including PV generation
capacity Smax (kWp), battery Storage capacity CB (kWh) and conductor size X (AWG) for
minimum cost of installation, we consider the energy flow diagrams for C- and O-
distribution architectures (as dictated by (5.8)) as shown in figures 5.3 and 5.4
respectively.
Figure 5. 3 System Diagram for Energy Flow in C-architecture with N Houses
1 2 3 N/2
EB(t)Battery
PV output
(N/2)+1 (N/2)+2 (N/2)+3 N
S(t)
ηMP
ηB
ESL(t)
E BL(
t)
g12 g23 gxygxy
gxy gxy gxy
η1 η2 η3 ηN/2
ηN
P1(t) P2(t) P3(t) PN/2(t)
PN(t)
ESB(t)
gxy
82
Figure 5.4 System Diagram for Energy Flow in O-architecture with N Houses
5.5.1. Objective function
The overall optimization problem is therefore, the minimization of objective
function (5.9), subjected to the constraints defined by (5.10) - (5.17).
XtStCT
tB
CSEEEE XBBBLSLSB
3max21
1) , , , , ,( min
,max
(5.9)
5.5.2. Inequality Constraints:
Non-negativity constraints on decision variables
T ; 0 ,0 ,0 ,0 ,0 max SCEEE BBLSLSB (5.10)
Since the battery lifetime is dependent upon its depth of discharge (DOD),
therefore, energy level of the battery EB(t) at any time t is not allowed to go below its
minimum energy level dictated by allowable minimum state of charge SOCmin.
T; min BB CSOCE (5.11)
5.5.3. Equality Constraints
1 2 N/2
EB(t)Battery
PV output
(N/2)+1 (N/2)+2
S(t)
ηMP
ηB
E BL(
t)
g12gxy
gxy
η1 η2 ηN/2
ηN
P1(t) P2(t) PN/2(t)
PN(t)
ESL(t)
ESB(t)
N
EB(t)Battery
PV output
S(t)
ηB
ESB(t)
ηMP
E BL(
t)
ESL(t)
gxy
gxy gxy
83
The constraints on generated solar energy are dictated by (5.12) as the generated
energy can either be used to supply load or to charge the battery including the losses
encountered in the path of power flow.
T; tEtEtS BLSLMP (5.12)
Constraints on battery energy are given by the net balance of influx and out-flux of
energy and are given by (5.16)
T; 1 tEtEtEtEE BLSBBBBB (5.13)
The constraints on load are defined by (5.14) such that load demand must always be
fulfilled either through battery or solar PV output. For constant household power loading
scenario as discussed in the current scope of the work, converter efficiency ηi is
considered constant. Therefore, linearization of load constraints does not result in any
significant loss of accuracy. However, in case of variable household loading, non-linear
accurate models for converter efficiency need to be included for accurate estimations
[110].
T;
11 1
tEtEtPttP
BLSL
T
t
loss
T
t
N
ii
i
(5.14)
The optimal PV size Smax is determined by the maximum output power produced by PV
array. Similarly, optimal battery size CB is determined by the maximum energy state
attained by the battery, therefore, (5.15) and (5.16) dictate the equality constraints
associated with the objective function for the optimal sizing of PV and battery
respectively.
T; )max(max SS (5.15)
T; )max( BB EC (5.16)
A range of possible conductor sizes for DC distribution are considered from
American wire gauge (AWG) table [48]. Therefore, constraints on Conductor size are
given by (5.17).
AWGAWGAWGAWGAWGAWGAWGX 16 14 12 10 8 6 4 ,,,,,, (5.17)
84
For considered range of operation, the stated objective function along with
equality and inequality constraints exhibit linearity as shown by (5.9)–(5.17). Therefore,
optimization problem is written in standard linear form and is solved through linprog
function in MATLAB.
5.6. Results and Discussion for Optimal Sizing of Centralized DC Microgrids
For the current scope of work, we consider a typical linear village structure in
South East Asia having (typically) 40 houses and distance between two consecutive
houses is (typically) 10m. We consider both C- and O- architectures along with two
distribution voltages of 24V and 48V. Optimal selection of PV panel size, battery storage
capacity and conductor size is performed for 5W (case 1) and 10W (case 2) provision and
is based upon problem formulated in section 5.6.
The idea is to account for the impact of instantaneous fluctuations in PV
generation (based upon timed fluctuations in instant irradiance and temperature) and
losses on system sizing, therefore, selection of time base is critical. Selection of smaller
time base results in better resolution to account for instantaneous changes, however,
requires more computational resources for data processing. The proposed methodology is
generic and can be applied for any time base, based upon the availability of solar
irradiance data, load profile and computational resources for optimization problem
solving. For current optimization problem, discrete time interval of 1 hour with constant
load demand for each house is used as time base for all the calculations. Hourly
variations in the irradiance and temperature data are taken from NREL [49]. According
to the presented optimization problem optimal PV, battery and conductor sizing is
determined.
PV panel shading losses are considered 8% for both cases. Although the converter
losses are quadratic in nature but due to fix load assumption and for simplicity in
optimization problem formulation, DC-DC converter losses for CPPU as well as
individual household distribution panel converter are considered 10% for both cases.
Battery cycle efficiency ηB is considered 95%, and is assumed constant for simplicity of
analysis. The cost of panels per watt-peak ‗w1‘is taken as 600$/kWp (including the cost
of mounting frame) and lead-acid battery cost per kWh ‗w2‘ is taken as 120$/kWh and
85
gauge sizes cost is taken at 1671, 1305, 947, 588, 358, 250 and 204 $ for 4, 6, 8, 10, 12,
14 and 16 AWG, respectively [50, 111, 112]. While there are variations in these costs but
we took most commonly found prices at which sourcing is readily available. For overall
system cost calculation, fixed cost of 300$ has been included which accounts for CPPU
and converter cost.
5.6.1. Case 1: 24-7 5W Supply to 40 Houses for a 365-day Operation
Based upon the analysis presented in section 5.5, the distribution efficiency, ηD
and worst voltage dips, VD calculated for following cases are plotted in figure. 5.5.
i. C-24V (C-architecture with 24V distribution)
ii. O-24V (O-architecture with 48V distribution)
iii. C-48V (C-architecture with 48V distribution)
iv. O-48V (C-architecture with 48V distribution)
VD is critical in terms of power electronic converter requirement at each
subscribing household. While, generally, a 20% input voltage variation capability is
allowed in most power electronic converters, the performance is optimal close to the
rated input voltages. For this particular analysis, we limit the optimum component
selection at 20% variations in the grid voltages. It is also interesting to note that O-24V
and C-48V have similar voltage dips and efficiencies for all wire sizes. This is because of
the uniform loading and equal generation at both ends of the O- architecture. In case of
non-uniform loading and unequal generation, the characteristics of C-24V and O-48V
will not necessarily resemble, therefore, detailed distribution loss analysis has to be
performed as discussed in [79]. Along with the selection of PV panel and battery size,
one important parameter is the optimization of the conductor size. The cost of conductor
increases with its decreasing gauge thickness and vice versa while an opposite trend for
system distribution efficiency is observed. Therefore, in order to analyze this effect in
terms of cost, figure. 5.6 (based on 365-day study on the solar irradiance data of Bihar,
India) is plotted. The cost of distribution losses is calculated by taking the difference
between overall system cost with distribution losses at a particular AWG and the overall
system cost with ideal conductor having zero distribution losses.
86
From figure.5.6 it can be seen that conductor cost decreases with the increase in
gauge size while the cost associated with the distribution losses increases at higher gauge
values. Thereby, our optimization problem calculates the optimal point such that overall
cost of the distribution losses and upfront cost of conductor is minimal. For instance, for
C-24V distribution architecture, 10AWG is optimum compared to its operation at
12AWG where its operation would be less efficient and sub-optimal from cost
perspective. Alternatively, at 8AWG the system will be more efficient but at the resulting
cost would be higher compared to the optimal value at 10AWG. In the current case study,
the two cost functions i.e. cost of distribution losses and cost of conductor intersect due to
similar scale. However, in general (for all cases), it may not be the case. The objective is
to find the minima rather than the intersection for calculating the optimal conductor size
and there may not be an intersection point for certain cases where cost of conductor is
considerably higher than cost of distribution losses and vice versa. For further analysis of
the impact of region-specific data i.e. time varying incident irradiance and temperature
profiles on the cost and sizing of the system, results of a 365-day study for multiple
locations in India as shown in Table 5.1, depicting optimal panel sizing, optimal battery
capacity and overall optimal installation cost are shown in figures 5.7, 5.8 and 5.9
respectively. These areas are selected based upon their spatial distribution on the map and
variation in annually averaged, daily PSH.
87
Figure 5. 5 Distribution Efficiency, η (right y-axis and Worst Voltage Dip, VD (left y-axis) for C-
architecture and O-architecture with 5W Loading at Different Gauge Sizes, and Different Voltage
Levels.
Figure 5. 6 Optimal Selection of Conductor Size with 5W Power Provisions at Different Voltage
Levels and Distribution Architectures.
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
5
10
15
20
25
30
Ma
x V
oltag
e d
ip (
%)
Wire Size (AWG)
C-24V
O-24V
C-48V
O-48V
4 6 8 10 12 14 1670
75
80
85
90
95
100
Dis
trib
uti
on E
ffic
iency
(%
)
4 6 8 10 12 14 16
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Co
st
(x1
00
0)
$
Size of Distribution Conductor (AWG)
Conductor Cost
C-24V Distribution Cost
O-24V Distribution Cost
C-48V Distribution Cost
O-48V Distribution Cost
88
Table 5.1 Specific Regions for Analysis with Irradiance Profiles
Regions Annually Averaged PSH Regions Annually Averaged PSH
1. Bihar 4.992 7. Jhansi 5.457
2. Dehli 5.035 8. Indore 5.493
3. Kolkata 5.063 9. Mumbai 5.604
4. Rajpur 5.280 10. Aurangabad 5.664
5. Ranchi 5.328 11. Jhodapur 5.808
6. Banglore 5.400
Figure 5. 7 Optimal PV Panel Sizing of the System at 5W Loading (Case 1). Kindly, note that each
data point on the figure represents a region shown in table 1. For instance, the first data point (from
left) is for Bihar where PSH is 4.992 and so on. (Followed in all subsequent figures)
5.0 5.2 5.4 5.6 5.81.2
1.3
1.4
1.5
1.6
1.7
PV
Siz
ing (
KW
p)
Average Daily Peak Sunlight Hours (PSH)
C-24V
O-24V
C-48V
O-48V
89
Figure 5. 8 Optimal Battery Sizing of the System at 5W Loading on left Y- axis and Irradiance
Volatility Factor on Right Y- axis (Case 1)
Figure 5. 9 Optimal Installation Cost of the System at 5W Loading (Case 1)
From Fig. 5.7-5.9, a few important observations can be made:
1) Optimal PV sizing varies approximately linearly with the average daily PSH as shown
in figure 5.7.
2) Average daily PSH are not a direct measure for optimal battery sizing (figure 5.8).
5.0 5.2 5.4 5.6 5.84
5
6
7
8
9
10
11
12
13
14
15
16
17
18 C-24V
O-24V
C-48
O-48V
IVF
Average Daily Peak Sunlight Hours (PSH)
-10
-5
0
5
10
15
Irra
dia
nce
Vo
latilit
y F
acto
r (%
)
Ba
tte
ry S
izin
g (
KW
h)
5.0 5.2 5.4 5.6 5.8
1.8
2.0
2.2
2.4
2.6
2.8
3.0
3.2
3.4
Optim
al C
ost (x
1000 $
)
Average Daily Peak Sunlight Hours (PSH)
C-24V
O-24V
C-48V
O-48V
90
3) As the battery cost constitutes a predominant portion of the overall cost of the system,
the overall cost of the system also does not vary linearly with the daily PSH (figure 5.9).
Figure 5.8 and table 5.1 verify that ―optimal battery sizing is not directly related
with the average PSH and the information of average PSH only is not sufficient for
optimal sizing of battery storage requirements.‖ Rather, a detailed analysis on the
variations pattern of PSH is needed to calculate the optimal battery storage requirements.
For instance, in figure 5.8, for all four cases, the battery requirements are lower for
PSH=5.3 in comparison to PSH=5.4 and PSH=5.5. Thus, if a proper investigation is not
made for daily variations in PSH, it may leads to incorrect sizing estimations. Therefore,
a detailed analysis must be conducted to find the optimal battery sizing incorporating the
volatility of irradiance. In order to analyze this further an irradiance volatility factor
(IVF) has been defined, which accounts for the number of dips from the mean
‗temperature compensated irradiance‘ and their corresponding extent of deviation
normalized over mean irradiance.
DN
i
i
tc
m
tcm
tc
IINI
IVF1
1 (5.18)
Where, N is the total number of temperature compensated irradiance recordings
out of which ND numbers of recordings are below mean temperature compensated
irradiance (Itcm) and (Itc
i) is the temperature compensated irradiance of i
th dip. Using
(5.18), IVF has been calculated for various regions under observation and it is found that
battery sizing varies in direct proportion with IVF as shown in figure 5.8. The value of
IVF depends on number of dips and their extent of deviation from the mean intensity.
IVF directly impacts the battery sizing and in turn the overall cost of the system. For
instance, in figure 5.8, it can be seen that Kolkota with a relatively low PSH 5.06 (table
5.1) compared to Banglore (PSH 5.4) requires lower battery size. This is due to the
reason that Kolkota has lower IVF (volatility factor) compared to Banglore which
directly translates into reduction in the battery size. Similarly, regions with higher IVF
require higher battery sizing despite of possibility of having high mean irradiance.
Therefore, the sizing of a PV system must incorporate the volatility of irradiance along
with the mean values for PSH.
91
5.6.2. Case 2: 24-7 10W Supply to 40 Houses for 365-days Operation
The most critical aspect is the calculation of optimal sizing for higher power
(10W) is the worst voltage dip along with distribution efficiency of each distribution
topology (see figure 5.10). With the increase in power provisioning, the distribution
losses (I2R) increase significantly compared to case 1 (5W household load). However,
like case 1, C-48V and O-24V have very similar voltage dips and overall distribution
efficiencies. Moreover, distribution at lower voltage level i.e. 24V with C-distribution
architecture becomes practically infeasible due to higher losses and higher voltage dips.
The increased distribution losses can be compensated by selecting a thicker
conductor which will increase the cost of the system. Therefore, it becomes even more
critical to optimally size the conductor by taking the capital and relative cost of
distribution into account. The proposed optimization framework therefore, enables the
optimal selection of conductor size based upon the trade-off between cost of the
conductor and its relative cost of distribution (see figure 5.11). From the comparison of
figures 5.6 and 5.11, it can be seen that with the higher power provision, the optimal
conductor selection has been shifted to lower gauge size (thicker conductor).
Figure 5. 10 Distribution Efficiency ηD and Worst Voltage Dip VD for C-architecture and O-
architecture with 10W Loading at Different Gauge Sizes, and Different Voltage Levels
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
5
10
15
20
25
30
C-24V
O-24V
C-48V
O-48V
Max V
oltage d
ip (
%)
4 6 8 10 12 14 1670
75
80
85
90
95
100
Dis
trib
uti
on E
ffic
iency
(%
)
Wire Size (AWG)
92
Figure 5. 11 Optimal Selection of Conductor Size with 10 W Power Provisions based upon the
Relative Cost of Distribution at Different Voltage Levels and Distribution Architectures
For instance, the optimal conductor sizes for 5W operation (figure 5.6) are 10,
12, 12 and 16 AWG for C-24, O-24, C-48 and O-48 configurations respectively. While
for 10W provision, the gauge sizes are reduced to 6, 10, 10 and 12 AWG for C-24, O-24,
C-48 and O-48 configurations respectively. Therefore, the proposed optimization
framework has adjusted the conductor size such that overall cost of the losses has been
compensated by the selection of a thicker conductor while keeping the overall cost of the
system as minimum. Figure 5.12 shows the overall system optimal cost for various
regions (with reference to their PSH). It is important to note that due to the planning
flexibility in the current framework, the overall cost of the system in high power
provision follows the same trend with relatively higher values in comparison to case 1.
Further it has been observed that the overall panel and battery requirements for case 2
also follow a similar trend as compared with case 1 (results for case 2 are not shown
here).
4 6 8 10 12 14 160.0
0.5
1.0
1.5
2.0
Conductor Cost
C-24V Distribution Cost
O-24V Distribution Cost
C-48V Distribution Cost
O-48V Distribution CostC
ost (x
1000)$
Size of Distribution Conductor (AWG)
93
Figure 5. 12 Optimal Installation Cost of the System (case 2)
Based upon the analysis, figure 5.13 shows the spatial distribution of various
regions on the map and associated optimal sizing requirements for PV panel (kWp),
battery sizing (kWh) and optimal system cost. The case for 24V, O-configuration with
5W loading is represented on the map as a sample case. Optimal data map for other
configurations and voltage levels may also be drawn by using the presented analysis.
Figure 5. 13 Optimal Sizing for 24V, 5W, O- Configuration Represented on Map
5.0 5.2 5.4 5.6 5.83.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
Optim
al C
ost (x
1000 $
)
Average Daily Peak Sunlight Hours (PSH)
C-24V
O-24V
C-48V
O-48V
94
5.7. Optimal Component Sizing for Distributed Generation and Distributed
Storage Architecture based DC Microgrids
In case of DGDSA based DC microgrids, the overall energy balance model can be
considered same as that presented for centralized microgrids. The difference in sizing
results primarily due to lower distribution losses in the distributed architecture which
reduces the need of PV generation and battery storage in individual nanogrids. Therefore,
energy balance model for DGDSA of DC microgrid build around a village having N
houses is given by
t 11
,1
tPttPtEttS loss
N
iii
N
iiBB
N
iiMP (5.19)
Where, ηMP is the efficiency of DC/DC converter employed for MPPT of PV panels, ηi is
the efficiency of DC/DC converter employed at the distribution panel of each house, Si(t)
is the time varying solar power generation, ΔEB,i(t) is the time varying change in energy
state of the battery and Pi(t) is the household load demand at individual household i, in
between any time interval Δt.
Due to assumption of constant battery and DC/DC converter efficiency, equation
(5.19) for DGDSA behaves exactly in same manner as equation (5.8) for centralized
architecture, if battery capacity and PV generation capacity on all N houses is considered
identical. Therefore, the optimal component sizing framework presented for centralized
architectures is equally applicable for DGDSA with the aforementioned assumptions. The
only factor which creates the difference is the distribution losses Ploss(t) which is
considerably lower in DGDSA as compared to centralized architecture as discussed in
Chapter 3. Thereby, using the framework presented in section 5.6 and the distribution
loss calculation model for DGDSA with ring-main scheme of interconnection presented
in chapter 3, the overall sizing of N households (PV sizing, battery sizing and conductor
sizing) is calculated. Sample results for 10W rated power at each household for various
voltage levels and various regions under considerations in table 5.1 are shown in figure
5.16-5.18 respectively. It should be noted here that for distribution loss calculation for
DGDSA operation, peak power sharing scenario, where each household may share
±100% of its rated power has been considered.
95
Figures 5.14-5.16 show the optimal PV sizing (kWp), optimal battery sizing
(kWh) and optimal system cost ($) for N houses in DGDSA at 10W rated power with
various voltage levels and various irradiance levels (regions under consideration given in
Table 5.1) respectively. It should be noted that at low voltages i.e. at 48V, the distribution
losses are relatively higher, therefore, PV sizing, battery sizing and system cost is
relatively higher. At higher voltages 120V and above, due to limited power delivery, i.e.
10 W only, the distribution losses are virtually absent due to distributed nature of the
architecture even at peak power sharing scenario, therefore, the associated sizing values
on higher voltages overlap with each other. Moreover, the trend for optimal PV sizing,
optimal battery sizing and optimal system cost is the same as for centralized architecture
of DC microgrid i.e. optimal PV sizing is directly proportional to incident irradiance,
while the optimal values of battery sizing and system sizing depends upon the incident
irradiance pattern, i.e. its mean value, number of dips and their extent of deviation from
mean value, therefore, IVF is a true measure for optimal battery sizing and optimal
system cost for DGDSA based electrification as well.
Figure 5. 14 Optimal PV Panel Sizing for N houses in DGDSA at 10W Loading
5.0 5.2 5.4 5.6 5.8
2.4k
2.7k
3.0k
400 V
325 V
240 V
120 V
48 V
Average Daily Peak Sunlight Hours (PSH)
PV
Siz
ing
(K
WP)
2.4k
2.7k
3.0k
96
Figure 5. 15 Optimal Battery Sizing for N houses in DGDSA at 10W Loading
Figure 5. 16 Optimal System Cost for N houses in DGDSA at 10W Loading
5.8. Optimal Component Sizing Comparison between Centralized and DGDSA
based DC Microgrid
In order to have comparison between centralized architecture of DC microgrid
having all of its generation/storage resources placed at a central location and the proposed
DGDSA from optimal component sizing prospective, we have considered the case of
5.0 5.2 5.4 5.6 5.8
6.0k
8.0k
10.0k
12.0k
14.0k
Battery
Siz
ing (
Whr)
400 V
325 V
240 V
120 V
48 V6.0k
8.0k
10.0k
12.0k
14.0k
Average Daily Peak Sunlight Hours (PSH)
5.0 5.2 5.4 5.6 5.8
2.7k
3.0k
3.3k
3.6k
3.9k
4.2k
Op
tim
al C
ost
(x1
$)
400 V
325 V
240 V
120 V
48 V
Average Daily Peak Sunlight Hours (PSH)
2.7k
3.0k
3.3k
3.6k
3.9k
4.2k
97
10W power delivery at each household with 48V distribution voltage. The framework
described in section 5.6 is used to calculate the optimal component sizing for both
centralized and DGDSA based DC microgrids at various irradiance levels under
consideration. The comparative results for optimal PV sizing (kWp), optimal battery
sizing (Whr) and optimal system cost at various irradiance levels have been plotted in
figures 5.17-5.19 respectively. Others system specification and village specifications are
same as discussed in previous cases.
Figure 5. 17 Comparative Results of Optimal PV Sizing for Centralized and DGDSA based DC
Microgrid
Figure 5. 18 Comparative Results of Optimal Battery Sizing for Centralized and DGDSA based DC
Microgrid
5.0 5.2 5.4 5.6 5.8
2.1k
2.4k
2.7k
3.0k
3.3k
3.6k
DGDSA
Centralized
Average Daily Peak Sunlight Hours (PSH)
PV
Siz
ing
(K
WP)
2.1k
2.4k
2.7k
3.0k
3.3k
3.6k
5.0 5.2 5.4 5.6 5.8
6.0k
9.0k
12.0k
15.0k
18.0k
DGDSA
Centralized
Average Daily Peak Sunlight Hours (PSH)
Battery
Siz
ing (
Whr)
6.0k
9.0k
12.0k
15.0k
18.0k
98
Figure 5. 19 Comparative Results of Optimal PV Sizing for Centralized and DGDSA based DC
Microgrid
From figures 5.16-5.19 it may be concluded that due to relatively lower
distribution losses in DGDSA, the optimal component sizing requirements are relatively
lower in DGDSA in comparison to centralized architecture. Therefore, from optimal
system planning and design prospective, DGDSA is a better choice for future rural
electrification.
5.0 5.2 5.4 5.6 5.8
2.8k
3.2k
3.6k
4.0k
4.4k
4.8k
5.2k
DGDSA
Centralized
Average Daily Peak Sunlight Hours (PSH)
Optim
al C
ost (x
1$)
2.8k
3.2k
3.6k
4.0k
4.4k
4.8k
5.2k
E2
99
6 Conclusions and Future Work
Based upon the work presented in previous chapters, general conclusions are
drawn in this chapter. Moreover, the challenges associated with the practical
implementation of the proposed microgrid architecture are also highlighted. Although the
work presented in this thesis conducts the technical investigation, the economic aspects
associated with the installation, operation and maintenance needs to be investigated.
Since it is envisioned that the proposed architecture will have practical implementation in
developing countries, therefore, a framework for economic analysis is currently being
worked out to ensure wide adoptability of the proposed technology. This aspect is further
highlighted in this chapter.
6.1. Conclusions
General findings from the perspective of architecture, control and optimal planning are
detailed and guidelines for future implementations are outlined in this section.
Distributed Generation and Distributed Storage Architecture of DC Microgrid is an
Optimum Technical Choice for Rural Electrification
This work presents a novel DC microgrid architecture for rural electrification with
emphasis on power provisioning beyond subsistence level [71]. The proposed microgrid
is formulated through the interconnection of multiple nanogrids, where an individual
nanogrid is the integral part of the overall microgrid structure. Each nanogrid has its own
PV generation, battery storage and also has the capability of bidirectional power
exchange with neighboring microgrids. Resource sharing allows the consumers to pool
up power for running high power community loads. The proposed microgrid built
through bottom-up cluster of multiple nanogrids is therefore scalable and also has the
capability to extract the benefit of usage diversity due to its resource sharing nature. The
distributed nature of PV generation and battery storage resources result in lower
distribution losses in comparison to existing centralized architectures of DC microgrids
for rural electrification. Using the Newton-Raphson method for modified DC power flow
100
analysis, various power provisioning scenarios for nanogrids and communal load are
evaluated. The validation of the proposed methodology is also done through a scaled
down version of hardware. Based upon the results, it has been concluded that by
increasing the distribution voltage level, distribution efficiency can be enhanced;
however, higher voltages result in enhanced system complexity form protection and
safety point of view. Although lower voltages are considered more safe and do not
require sophisticated schemes for safety and protection, but considerable distribution
losses are associated with power delivery at lower voltages. The choice for optimal
distribution voltage level can be made based upon the trade-off between the cost of
distribution losses, distribution conductor and protection equipment. Central architectures
incur very high distribution losses at high power delivery, therefore, are limited to low
power applications and are not suitable for high power community load operations. The
results of analyses show that the proposed distributed storage architecture can enhance
distribution efficiency by approximately 5% more than other LVDC architectures. The
analysis showed that for the selected parameters, the distribution losses were less than 3%
even at a communal load of 400 W.
Decentralized Control is an Excellent Candidate Solution for Communication-less
Coordination among the Resources in Distributed Microgrids
Centralized architectures are relatively simpler from installation, control and
operation prospective, however, they lack modularity and significant distribution losses
are associated with their delivery of energy. Distributed architectures on the other hand
are scalable, modular and have lower distribution losses; however, they require
sophisticated control techniques involving communication among the distributed
resources for their stable operation. Alternately, in this thesis, a communication-less
control strategy for the stable operation of a PV/battery-based highly distributed
architecture of DC microgrid is presented. A decentralized control scheme is developed
for the communication-less, yet coordinated control among the distributed resources
using multi-mode adaptive IV droop method, relying on local information of DC bus
voltage and battery SOC at individual nanogrids. The efficacy of the proposed control
scheme is validated for various possible power sharing scenarios using simulations on
101
MATLAB/Simulink and hardware in loop facilities at microgrid laboratory in Aalborg
University. The proposed control method is highly suitable for the rural electrification of
developing regions because it (i) enables coordinated distribution of generation and
storage resources at a village scale, (ii) decentralized controllability omits the need of
central controller and associated costly communication infrastructure, and (iii) enables
resource sharing among the community to extract the benefit of usage diversity at a
village scale.
Optimal Planning and Design of DC Microgrid sizing can significantly reduce the
system costs
In this thesis, a planning framework is developed to find the optimal component
sizing for various components of a DC microgrid ensuring minimum cost of installation
[113]. To analyze the impact of distribution architecture on optimal component sizing,
distribution losses are related to actual structure of the village and spatial distribution of
houses in it. For centralized microgrids, two possible architectures of DC distribution (O-
architecture and C-architecture) have been analyzed for power flow and associated
distribution losses. It has been identified that optimal sizing of the system components is
highly dependent upon region-specific time varying profile of irradiance and temperature
throughout the year. The proposed framework allows optimal selection of 1) solar panel
sizing, 2) storage size and 3) conductor size for optimal cost solution. The selection of
distribution voltage and the distributed architecture is ultimately the choice for the
microgrid installers/planners and is influenced by power converter preferences and
financial models. But, importantly, the framework established in this work for optimal
system topology (in multiple constraint scenarios) will be highly useful in planning for
new optimal low-voltage systems along with efficiency enhancements of many existing
systems by retrofitting distribution architectures. Also, it has been found that distributed
generation, distributed storage architecture results in lower distribution losses, lower
system sizing and lower installation cost for same power delivery in comparison to
centralized architecture.
102
6.2. Potential Challenges in Practical Deployments of Decentralized Microgrids
Although the proposed architecture allows for the efficient utilization of
distributed resources in a highly scalable manner, some challenges associated with larger
deployments persist. High-level distribution of resources poses a challenge with respect
to safety and protection due to the increased likelihood of short circuit contribution from
multiple paths within the microgrid. Therefore, future large-scale practical
implementations must include an intelligent protection scheme capable of real-time load
flow and short circuit analysis for adaptive relay settings (not part of the scope of this
study). Further work on the islanded operation of the microgrid is also important to
ensure optimum isolation from the grid for enhanced safety.
From the perspective of control, although hysteretic-based voltage droop control
can reduce multiple layers of sensing and control in the DGDSA, for large-scale
implementations, additional sensing and hierarchical control layers may be added to
ensure the enhanced stability of the microgrid over a wide range of operations.
From practical implementation prospective, there can be potential challenges for
the distributed placement of resources. Space barriers along with the maintenance of
converters and cleaning of PV panels at individual households are some of the practical
challenges that need to be addressed for successful practical implementations. Bi-
directional power flow metering and theft monitoring issues must also be considered for
future installations.
From economics point of view, such a distributed model is highly suitable for
micro-financing opportunities for private investors / public-private partnerships and many
successful models exist in practical as well as in literature highlighting the role of micro
financing in energy access [114]. The participation of the private sector has grown in
recent years [115] but more needs to be done. For instance, experience of past rural
electrification projects in South Asia suggests subsidized micro-financing schemes for
local communities along with public/private partnerships for the successful
implementation of such projects [116, 117]. Therefore, Technical innovations must be
coupled with suitable business models to ensure wide uptake of energy-related initiatives.
103
6.3. Cost and Affordability Evaluation of the Decentralized Microgrid System
Although rural electrification through PV based DC microgrids has been widely
adopted as a successful solution for providing basic electrification in off-grid
communities throughout the world, however, in context of Pakistan, there exists some
potential barriers related to sophisticated demand and policy implementation that needs to
be considered for wide deployment of distributed microgrids. Therefore, a thorough
economic analysis along with a suitable business model for its wide uptake must be
investigated. The work on a demand based decentralized electrification model in
collaboration with International growth Center (IGC), UK is under progress. Focus of the
project is to assess ‗willingness to pay‘ for certain electricity services and how consumers
would behave when the actual system is rolled out. The work in this aspect is still
ongoing and findings of the project including financial model to support micro financing
in the sector of energy along with associated energy policy recommendations will be
shared with academic community in due course. A brief highlight of the work under
progress is discussed in the following sub-sections.
6.3.1. Survey Data
A household census was carried out across 7 villages in the Multan district of
Punjab, Pakistan. The surveys covered all 140 households in the area that were
designated as off-grid, i.e. they are not covered by the national electricity grid. Other than
basic household characteristics, the survey elicited willingness to pay for different
bundles of electricity services, to gauge the demand for such services. Finally, the survey
elicited whether households would be willing to pay for three different levels of
electricity service, provided through microgrids. Levels of service were chosen to
replicate services provided by traditional microgrids, such as Mera Gao Power, and those
made available through the use of distributed architecture of microgrid. The prices at
which these services were offered were randomized between three rate plans, which
presented increasing prices for each level of service. Due to the close proximity of
households inside each village, plans were randomly allocated at the village level, instead
of the household level. Table 6.1 provides details of the level of services and their
respective prices per month under each price plan.
104
Table 6. 1 Services offered and their prices by plan
Level of service Prices
Price plan 1
(USD/month)
Price plan 2
(USD/month)
Price plan 3
(USD/month)
1 24/7 provision of high quality light (3 LED
lights), and a mobile charging point.
1.5 2.5 4.5
2 Services in Bundle 1 and a Fan. 3.0 5.0 9.0
3 Services in Bundle 2 and shared communal
load (water pump).
4.5 7.5 10.5
6.3.2. Survey Results
Table 6.2 presents summary statistics for the reported prices for each level of
service, and also separates it by price plan. We find that not only is demand for the level
of services significant in our sample villages but demand for higher levels of service is
significantly more than for basic electrification. Table 6.3 reports the absolute and
relative difference in prices respondents reported they were willing to pay with respect to
the base level of electrification (lights only), where, absolute difference is the arithmetic
difference in prices of two packages (Py – Px), and relative difference is their ration
(Py/Px).
Table 6. 2 Average Prices Willing to be Paid for Each Level of Service
Average Price/month in USD (standard error)
Number of
Households
Lights Only Lights and Fan Lights, fan and
communal load
Overall 122 1.87
(0.98)
3.40
(0.16)
4.35
(0.17)
Price plan 11 22 1.5
(0)
3.0
(0)
4.5
(0)
Price plan 2 56 1.53
(0.69)
2.94
(0.12)
3.8
(0.12)
Price Plan 3 44 2.48
(0.23)
4.19
(0.41)
4.97
(0.44)
1All respondents in Village 2 accepted the quoted prices under plan 1, yielding a standard error of 0.
105
Table 6. 3 Relative and Absolute Differences in Reported Prices with Respect to Base Plan (Lights
Only) Standard Error in Parenthesis
Number of
Observations
Absolute difference Relative Difference
Lights and Fan v.
Lights
122 1.53
(0.12)
1.97
(0.67)
Lights, fan and
communal load vs.
Lights
120 2.48
(0.12)
2.57
(0.07)
We find that there is significant difference for all services beyond those provided
by traditional microgrids (lights and mobile charging). Households are willing to pay
almost twice as much for the addition of a fan, and about 2.5 times more for both a fan
and communal load. Additionally, the marginal change between the two highest levels of
service is also found to be significant, with household willing to pay and additional Rs.
94 for a shared village water pump (communal load).
6.3.3. Cost Analysis Model for Economic Analysis
Given that the average willingness to pay for basic electrification in our sample is
around 1.87$/month for basic electrification and 3.40$/month for the additional provision
of a fan, we need to ensure the economic viability of the proposed decentralized scheme
in comparison with conventional systems. Therefore, a cost analysis framework is
developed for economic assessment. For a typical village with N households and
allowable power provision of Ph watts per household for T hours and a communal load of
Pc watts for t hours, the total number of required energy units at the output Eo of system
is given by (6.1).
tPUTNPE cdho (6.1)
Where, Ud is the usage diversity factor. Ud captures the inter-household usage
diversity in energy expenditure at any given time. Intuitively, it captures the fact that the
power demand of each household will be less than or equal to allowable provision and
106
will be different from other households in the village. It is typically assumed that Ud < 1,
i.e. on average households are consuming below their allowed provision. For example, it
is reasonable to assume from a design perspective that in the day time when sunlight is
available, lighting load at each household will be reduced. Similarly, in winters the fan
loads will be reduced due to seasonal variation in temperature and associated cooling
requirements. Therefore, panel requirements are reduced accordingly and are quantified
in (6.1). The capability to extract the benefit of usage diversity is precisely the reason
microgrids are more efficient that standalone systems.
Given the energy requirement of a system, system characteristics such as solar PV
panel sizing required to produce the requisite amount of power must be calculated. These
must take into account the incident irradiance, effect of temperature degradation and
various losses including wiring losses, converter losses, and storage losses during
charging/discharging cycle losses. To account for various energy losses and degradation
in a system, the amount of energy needed to be produced, , is given by (6.2).
DCBT
P
EE
0
(6.2)
Where, and are degradation in efficiency due to temperature, storage
(or battery) inefficiency, converter inefficiency and line losses, respectively. The
resulting panel size needed at per household can be expressed by (6.3).
NE
EP
p
PV
(6.3)
Where, ̅ is the average value of peak sunlight hours for the particular region.
Similarly, for first order cost calculation model for a battery system, the battery energy
capacity is determined by the total energy that battery has to supply when the sun is
not available along with the extra energy that is dissipated during charging/discharging.
Moreover, to extend the battery life, generally there is a limit on minimum discharging
state (%), which again tends to increase the required battery capacity. The
overall battery requirement for a microgrid then is given by (6.4).
107
B
hB
N
SOCPEPSHE
min124
(6.4)
In addition to storage and solar panel costs, there are other costs which include the
cost of converters, system protection equipment and conductor (wiring) length. The price
of converters is generally proportional to their power processing requirements, loading
levels and current carrying capacity. Therefore, for simplicity, it can be taken as a fixed
percentage, , of the total cost of PV panel. Similarly, the cost of protection is
proportional to the power loading level and short circuit current capacity, therefore, in the
current analysis it can be taken as a fixed percentage as well ( ). Finally, the cost of
total conductor length ( ) is given by C3 (USD/m). Considering PV panel cost as C1
(USD/watt) and battery cost as C2 (USD/watt), total system upfront cost CU is given by
(6.5).
PBPVCU lCECPCC 11 321 (6.5)
Generally, life time of a solar panel is 25 years, while the life time of battery can be taken
as years and life time of power electronic converters is given by years. Therefore,
for a typical 25 year system, the operation and maintenance cost, along with the number
of battery and power converters replacement is calculated and added with the capital cost
to find the overall lifetime cost of the system. The total lifetime cost CLT of the system
over its lifetime is then given by (6.6).
PCPV
c
B
B
ULT PCN
ECN
CC
11
251
2512
(6.6)
Further, it is important to evaluate the effective levelized cost of electricity (LCOE)
(USD/kWh) of system, given by equation (6.7), which effectively calculates the cost of
each unit produced by the microgrid over its 25 year operation in comparison to the
lifecycle costs.
)25)(1000)(365(O
LT
E
CLCOE
(6.7)
108
As, Eo is the energy produced per day which is multiplied by 365 (days in a year) and 25
(operational lifetime of the system). 1000 (in the denominator) gives LCOE in price per
kWh as kWh is standard unit for electricity production/consumption. The presented cost
model is applied for the electrification of a typical village having 40 houses, 30W rated
power at each house with ± 100% flexibility in power provision i.e. each house may
consume up to 60W (double of its rated power) of electricity, or may sell 30W of
electricity at a given time. The rated power provisioning is in accordance with the market
availability of DC loads with three lighting bulbs (~4W each), one DC fan (~14 W) and
one mobile charging unit (~4 W). Similarly, for the village under analysis a communal
load of up to 500W is considered for water filtration plant/pump for drinking purposes.
The household operation is for 24 hours while communal load operation is considered for
6 hours per day. The value of average peak sunlight hours ̅ for the typical village is
assumed to be 6 hours per day, i.e. (6 hours of standard daylight on average over the
year).
6.3.4. Results of Economic Analysis
The proposed first-order cost analysis model is applied on the village with
specifications discussed above to calculate system sizing requirements and associated
costs. Usage diversity factor is considered 0.3, which approximates lack of simultaneous
loading for all households at all the time. The costs are taken as followings:
PV panel price =0.8 $/kWp [118, 119],
battery price =1.05$ $/Wh [111] (Lead acid battery),
distribution conductor cost for the village is $500 and converter cost factor =0.3
and protection cost factor = 0.05 [120].
Therefore, considering all these factors, LCOE, along with flat rate tariff plan for
the proposed and existing schemes of electrification is calculated over 25 year project
life. For 6 years ROI plan, one battery replacement is considered, for 9 years ROI plan,
one battery and one charge controller replacement is considered, while for 12 years ROI
plan 2 batteries and one charge controller replacement is considered. Table 6.4 presents
the estimated costs of all three levels of services including and compares them to
109
alternative implementations. Kindly note that these are typical costs and the prices may
vary from one region to another. There may well be additional costs for some newer
aspects of efficient power processing (power electronics) in distributed microgrids. Solar
panels, storage and distribution prices used are standard wholesaler‘s rates, correct for the
month of December 2017.
Calculations show that solar power, in particular decentralized microgrids, present
a viable alternative to grid electricity even for loads beyond high quality light. Load
sharing allows the system to even provide electricity for a communal load for a negligible
increase in price. A major component to the cost of any solar system is the cost of
storage. Batteries are both expensive, have short life spans and are inefficient. However,
recent developments in battery technology suggest that the overall cost of such a system
is likely to come down in future. As an example, for a case of a decentralized microgrid,
we show that the cost of storage highly dominates the overall cost of the system in its
lifespan of 25 years (figure 6.1), at current market prices.
Table 6. 4 Estimated Cost of decentralized Solar Generation Implementations through DGDSA
Scenario Load per house
(24/7 provision
to subscribers)
Capital
Cost
(USD)
Effective
LCOE
(USD)
Capital +
25 years
O&M Cost
(USD)
Subscription Charges Per user per
Month for payback in (USD/Month)
3 years 6
years
9
years
12
years
Traditional
microgrid
(e..g. MGP)
1 light + mobile
charging unit
(5W).
2117 0.11 4775 1.47 0.91 0.66 0.62
Decentralized
Microgrid
3 Lights,
1 fan, charging
unit
(30W)
9563 0.10 26149 6.65 4.45 3.30 2.70
3 Lights,
1 fan, charging
unit and
Communal load
(30W +500W)
10000 0.09 26787 6.95
4.73 3.50 3.25
Standalone
Production
and
Consumption
(No grid)
3 Lights,
1 fan, charging
unit
(30W)
10828 0.11 29300 7.50 5.15 3.50 3.30
3 Lights,
1 fan, charging
unit and
Communal load
(30W + 500W)
11658 0.11 31304 8.10 5.6 4.0 3.80
110
Figure 6. 1 Life Time Operation Cost Break-up of a Distributed Microgrid
From the highlights of economic analysis that distributed microgrid architecture
presented in this thesis has a potential to create a micro-energy economy with a viable,
scalable and profitable microgrid model providing opportunities for entrepreneurs to
enter the market and also provide new sources of employment.
6.4. Future work
Our findings indicate that distributed solar microgrids present a promising route
to rural electrification, especially in areas where grid expansion may be prohibitively
expensive. They have the capacity and scalability to provide electricity beyond those
offered by traditional microgrids. In countries like Pakistan, that are already facing major
crises in supply, they present a low cost solution to not just the distribution problem, but
also the problem of generation. The next steps in this agenda would be the roll out pilots
of distributed microgrids in the surveyed areas to evaluate the performance of the system
along with its uptake in various regions. Further, a natural area for expansion would be to
study the appropriateness of microgrids as a source of back up electricity in on grid areas,
experiencing high number of rolling black outs. Along with that following are the key
16%
57%
14%
4% 4%
5%
25 years Operation Cost of the Distributed Microgrid
PV Panel Cost
Battery Storage Cost
Controller+ Converter Cost
Conductors Cost
others
Protection Cost
111
area that needs to be researched for successful practical deployment of the proposed
distributed microgrids.
In order to enable energy trading among multiple nanogrids, there must be a
mechanism to monitor energy transactions among neighboring nanogrids.
Although energy trade mechanism will formulate a local energy market and will
be helpful for empowering rural inhabitants, however, it will require a
communication layer at neighborhood levels to ensure monitoring of energy
exchange.
A key challenge for the successful implementation of such a distributed system
will be the development of theft monitoring mechanism which will be based upon
remote monitoring and communication system to ensure provisioned power flow.
Since battery storage system is the prime cost element in such systems, therefore,
a mechanism must be incorporated for battery state of charge, depth of discharge
and state of health monitoring for the maximization of battery life and preventive
maintenance.
The proposed architecture must be robust; therefore, a distributed, yet coordinated
and selective protection scheme must be developed to ensure the reliability of the
system as well as longevity of the power electronic equipment.
A scheme for optimal dispatch of the distributed resources based upon time
varying household and communal load requirements needs to be developed for
efficient resource utilization.
112
Bibliography
[1] World Energy Outlook (WEO, 2017), Electricity Access Database [Online].
Available:
https://www.iea.org/publications/freepublications/publication/WEO2017SpecialR
eport_EnergyAccessOutlook.pdf
[2] M. Lavelle. (2012, Cookstove Smoke is ―Largest Environmental Threat,‖.
Available: http://energyblog.nationalgeographic.com/2012/12/13/cookstove-
smoke-is-largest-environmental-threat-global-health-study-finds/
[3] N. L. Lam, K. R. Smith, A. Gauthier, and M. N. Bates, "Kerosene: a review of
household uses and their hazards in low-and middle-income countries," Journal of
Toxicology and Environmental Health, Part B, vol. 15, pp. 396-432, 2012.
[4] J. Peters and M. Sievert, "Impacts of rural electrification revisited–the African
context," Journal of Development Effectiveness, vol. 8, pp. 327-345, 2016.
[5] K. Reiche, A. Covarrubias, and E. Martinot, "Expanding electricity access to
remote areas: off-grid rural electrification in developing countries," Fuel, vol. 1,
p. 1.4, 2000.
[6] A. M. Tanner and A. L. Johnston, "The Impact of Rural Electric Access on
Deforestation Rates," World Development, vol. 94, pp. 174-185, 2017.
[7] F. Yang and M. Yang, "Rural electrification in sub-Saharan Africa with
innovative energy policy and new financing models," Mitigation and Adaptation
Strategies for Global Change, pp. 1-20, 2017.
[8] J. Chi, W. Peng, X. Jianfang, T. Yi, and C. Fook Hoong, "Implementation of
Hierarchical Control in DC Microgrids," Industrial Electronics, IEEE
Transactions on, vol. 61, pp. 4032-4042, 2014.
[9] World Energy Outlook (WEO, 2016), Electricity Access Database [Online].
Available:
http://www.worldenergyoutlook.org/resources/energydevelopment/energyaccessd
atabase/.
[10] G. He and D. G. Victor, "Experiences and lessons from China‘s success in
providing electricity for all," Resources, Conservation and Recycling, vol. 122,
pp. 335-338, 2017.
[11] Rajendra Singh and K. Shenai. (2014, DC Microgrids and the Virtues of Local
Electricity. IEEE Spectrum. Available: http://spectrum.ieee.org/green-
113
tech/buildings/dc-microgrids-and-the-virtues-of-local-
electricity/?utm_source=energywise&utm_medium=email&utm_campaign=0212
14
[12] P. Brinckerhoff, "Electricity Transmission Costing Study," IET, 2012.
[13] M. Nasir and A. A. Bhatti, "A Micro Computer-Based Methodology for Distance
Protection on Long UHV Transmission Lines Using Symmetrical Components,"
in Power and Energy Engineering Conference (APPEEC), 2012 Asia-Pacific,
2012, pp. 1-6.
[14] M. N. Qaiser, M. Usama, B. Ahmad, M. A. Tariq, and H. A. Khan, "Low Cost,
Robust and Efficient Implementation of MPPT based Buck-Boost Converter for
Off-grid PV Applications," presented at the 40th IEEE Photovoltaic Specialists
Conference, Denver Colorado, USA, 2014.
[15] H. A. Khan and S. Pervaiz, "Technological review on solar PV in Pakistan:
Scope, practices and recommendations for optimized system design," Renewable
and Sustainable Energy Reviews, vol. 23, pp. 147-154, 2013.
[16] S. R. Khandker, H. A. Samad, Z. K. Sadeque, M. Asaduzzaman, M. Yunus, and
A. E. Haque, Surge in solar-powered homes: Experience in off-grid rural
Bangladesh: World Bank Publications, 2014.
[17] Solar Home System programme by Infrastucture Development Company
(IDCOL), Bangladesh [Online]. Available: http://idcol.org/home/solar
[18] I. Khan, Y. Xu, H. Sun, and V. Bhattacharjee, "Distributed Optimal Reactive
Power Control of Power Systems," IEEE Access, vol. 6, pp. 7100-7111, 2018.
[19] D. Schnitzer, D. S. Lounsbury, J. P. Carvallo, R. Deshmukh, J. Apt, and D. M.
Kammen, "Microgrids for Rural Electrification: A critical review of best practices
based on seven case studies," United Nations Foundation, 2014.
[20] D. Palit and G. K. Sarangi, "Renewable energy based mini-grids for enhancing
electricity access: Experiences and lessons from India," in International
Conference and Utility Exhibition on Green Energy for Sustainable Development
(ICUE),\ 19-21 March 2014, pp. 1-8.
[21] J. Urpelainen, "Energy poverty and perceptions of solar power in marginalized
communities: Survey evidence from Uttar Pradesh, India," Renewable Energy,
vol. 85, pp. 534-539, 2016.
[22] S. Mishra and O. Ray, "Advances in nanogrid technology and its integration into
rural electrification in India," in Power Electronics Conference (IPEC-Hiroshima
2014-ECCE-ASIA), 2014 International, 2014, pp. 2707-2713.
114
[23] D. Palit, G. K. Sarangi, and P. Krithika, "Energising Rural India Using
Distributed Generation: The Case of Solar Mini-Grids in Chhattisgarh State,
India," in Mini-Grids for Rural Electrification of Developing Countries, ed:
Springer, 2014, pp. 313-342.
[24] S. Chakrabarti and S. Chakrabarti, "Rural electrification programme with solar
energy in remote region–a case study in an island," Energy Policy, vol. 30, pp. 33-
42, 2002.
[25] H. Jiayi, J. Chuanwen, and X. Rong, "A review on distributed energy resources
and MicroGrid," Renewable and Sustainable Energy Reviews, vol. 12, pp. 2472-
2483, 12// 2008.
[26] R. H. Lasseter, "MicroGrids," in Power Engineering Society Winter Meeting,
2002. IEEE, 2002, pp. 305-308 vol.1.
[27] N. Hatziargyriou, H. Asano, R. Iravani, and C. Marnay, "Microgrids," Power and
Energy Magazine, IEEE, vol. 5, pp. 78-94, 2007.
[28] B. S. Hartono, Y. Budiyanto, and R. Setiabudy, "Review of microgrid
technology," in QiR (Quality in Research), 2013 International Conference on, pp.
127-132.
[29] R. H. Lasseter and P. Paigi, "Microgrid: a conceptual solution," in Power
Electronics Specialists Conference, 2004. PESC 04. 2004 IEEE 35th Annual,
2004, pp. 4285-4290 Vol.6.
[30] M. Soshinskaya, W. H. Crijns-Graus, J. M. Guerrero, and J. C. Vasquez,
"Microgrids: Experiences, barriers and success factors," Renewable and
Sustainable Energy Reviews, vol. 40, pp. 659-672, 2014.
[31] J. B. Almada, R. P. S. Leao, F. F. D. Montenegro, S. S. V. Miranda, and R. F.
Sampaio, "Modeling and simulation of a microgrid with multiple energy
resources," in EUROCON, 2013 IEEE, 2013, pp. 1150-1157.
[32] L. Xuan and S. Bin, "Microgrids — an integration of renewable energy
technologies," in Electricity Distribution, 2008. CICED 2008. China
International Conference on, 2008, pp. 1-7.
[33] A. S. Rana, M. Nasir, and H. A. Khan, "String level optimisation on grid-tied
solar PV systems to reduce partial shading loss," IET Renewable Power
Generation, 2017.
[34] M. Nasir, A. A. Bhatti, and W. T. Toor, "Model formulation and design of an
efficient control algorithm for fuel cell power system," in Power Engineering,
115
Energy and Electrical Drives (POWERENG), 2013 Fourth International
Conference on, 2013, pp. 806-811.
[35] M. Nasir and H. A. Khan, "Solar photovoltaic integrated building scale hybrid
AC/DC microgrid," 2016.
[36] K. Siraj, H. Siraj, and M. Nasir, "Modeling and control of a doubly fed induction
generator for grid integrated wind turbine," in Power Electronics and Motion
Control Conference and Exposition (PEMC), 2014 16th International, 2014, pp.
901-906.
[37] F. Zia, M. Nasir, and A. A. Bhatti, "Optimization methods for constrained
stochastic wind power economic dispatch," in Power Engineering and
Optimization Conference (PEOCO), 2013 IEEE 7th International, 2013, pp. 129-
133.
[38] J. Peters and M. Sievert, "Impacts of rural electrification revisited: The African
context," Ruhr Economic Papers 3867886377, 2015.
[39] N. Sasidharan, J. G. Singh, and W. Ongsakul, "An approach for an efficient
hybrid AC/DC solar powered Homegrid system based on the load characteristics
of home appliances," Energy and Buildings, vol. 108, pp. 23-35, 2015.
[40] N. Bashir, H. S. Sardar, M. Nasir, N. U. Hassan, and H. A. Khan, "Lifetime
maximization of lead-acid batteries in small scale UPS and distributed generation
systems," in PowerTech, 2017 IEEE Manchester, 2017, pp. 1-6.
[41] F. Dastgeer and A. Kalam, "Efficiency comparison of DC and AC distribution
systems for distributed generation," in Power Engineering Conference, 2009.
AUPEC 2009. Australasian Universities, 2009, pp. 1-5.
[42] J. J. Justo, F. Mwasilu, J. Lee, and J.-W. Jung, "AC-microgrids versus DC-
microgrids with distributed energy resources: A review," Renewable and
Sustainable Energy Reviews, vol. 24, pp. 387-405, 2013.
[43] H. E. Gelani, M. Nasir, F. Dastgeer, and H. Hussain, "Efficiency Comparison of
Alternating Current (AC) and Direct Current (DC) Distribution System at
Residential Level with Load Characterization and Daily Load Variation."
[44] P. A. Madduri, J. Rosa, S. R. Sanders, E. Brewer, and M. Podolsky, "Design and
verification of smart and scalable DC microgrids for emerging regions," in
Energy Conversion Congress and Exposition (ECCE), 2013 IEEE, 2013, pp. 73-
79.
[45] D. J. Hammerstrom, "AC versus DC distribution systemsdid we get it right?," in
Power Engineering Society General Meeting, 2007. IEEE, 2007, pp. 1-5.
116
[46] D. Nilsson and A. Sannino, "Efficiency analysis of low-and medium-voltage DC
distribution systems," in Power Engineering Society General Meeting, 2004.
IEEE, 2004, pp. 2315-2321.
[47] H. Kakigano, Y. Miura, and T. Ise, "Low-Voltage Bipolar-Type DC Microgrid for
Super High Quality Distribution," Power Electronics, IEEE Transactions on, vol.
25, pp. 3066-3075, 2010.
[48] D. Soto and V. Modi, "Simulations of Efficiency Improvements Using Measured
Microgrid Data," in Global Humanitarian Technology Conference (GHTC), 2012
IEEE, 2012, pp. 369-374.
[49] W. A. Simon Willems, Stijn De Jonge, Dries Haeseldonckx, Pepijn van
Willigenburg, Johan Woudstra and Harry Stokman, "Sustainable Impact and
Standardization of a DC Micro Grid," ed. ‖ Proceedings of Ecodesign 2013
International Symposium, 2013-12-06.
[50] D. J. Becker and B. J. Sonnenberg, "DC microgrids in buildings and data centers,"
in Telecommunications Energy Conference (INTELEC), 2011 IEEE 33rd
International, 2011, pp. 1-7.
[51] (2001, Table US-1. Electricity consumption by end use in U.S. households, 2001,
Energy Information Administration
[52] A. C. Brent and D. E. Rogers, "Renewable rural electrification: Sustainability
assessment of mini-hybrid off-grid technological systems in the African context,"
Renewable Energy, vol. 35, pp. 257-265, 2010.
[53] J. W. Stevens and G. P. Corey, "A study of lead-acid battery efficiency near top-
of-charge and the impact on PV system design," in Photovoltaic Specialists
Conference, 1996., Conference Record of the Twenty Fifth IEEE, 1996, pp. 1485-
1488.
[54] P. A. Madduri, J. Rosa, S. R. Sanders, E. A. Brewer, and M. Podolsky, "Design
and verification of smart and scalable DC microgrids for emerging regions," in
Energy Conversion Congress and Exposition (ECCE), 2013 IEEE, 2013, pp. 73-
79.
[55] S. B. Tennakoon and P. M. McEwan, "Short-circuit interruption performance of
thyristor circuit breakers," in Applied Power Electronics Conference and
Exposition, 1994. APEC '94. Conference Proceedings 1994., Ninth Annual, 1994,
pp. 832-838 vol.2.
117
[56] M. E. Baran and N. R. Mahajan, "Overcurrent Protection on Voltage-Source-
Converter-Based Multiterminal DC Distribution Systems," Power Delivery, IEEE
Transactions on, vol. 22, pp. 406-412, 2007.
[57] M. Wang, M. Abedrabbo, W. Leterme, D. Van Hertem, C. Spallarossa, S.
Oukaili, I. Grammatikos, and K. Kuroda, "A Review on AC and DC Protection
Equipment and Technologies: Towards Multivendor Solution," 2017.
[58] A. Mohamed, V. Salehi, and O. Mohammed, "Real-Time Energy Management
Algorithm for Mitigation of Pulse Loads in Hybrid Microgrids," Smart Grid,
IEEE Transactions on, vol. 3, pp. 1911-1922, 2012.
[59] H. Kakigano, M. Nomura, and T. Ise, "Loss evaluation of DC distribution for
residential houses compared with AC system," in Power Electronics Conference
(IPEC), 2010 International, 2010, pp. 480-486.
[60] D. Salomonsson, L. Soder, and A. Sannino, "Protection of Low-Voltage DC
Microgrids," Power Delivery, IEEE Transactions on, vol. 24, pp. 1045-1053,
2009.
[61] A. R. Mohammad Ali Tavakkoli, Houshang Hassibi, "Simulation and Analysis of
a Compact Electronic Infrastructure for DC Micro-Grid: Necessity and
Challenges," Smart Grid and Renewable Energy, vol. Vol.3 pp. pp. 73-82., May
2012.
[62] A. K. Srivastava, "Solar minigrids in rural areas of Uttar Pradesh," Akshay Urja,
vol. 4, pp. 16-17, 2013.
[63] F. Pearce. World's poor need grid power, not just solar panels. Energy and Fuels
2980, 2014.
[64] P. A. Madduri, J. Poon, J. Rosa, M. Podolsky, E. Brewer, and S. R. Sanders,
"Scalable DC Microgrids for Rural Electrification in Emerging Regions," IEEE
Journal of Emerging and Selected Topics in Power Electronics, vol. PP, pp. 1-1,
2016.
[65] W. Inam, D. Strawser, K. K. Afridi, R. J. Ram, and D. J. Perreault, "Architecture
and system analysis of microgrids with peer-to-peer electricity sharing to create a
marketplace which enables energy access," in Power Electronics and ECCE Asia
(ICPE-ECCE Asia), 2015 9th International Conference on, 2015, pp. 464-469.
[66] T. Dragičević, J. M. Guerrero, J. C. Vasquez, and D. Škrlec, "Supervisory control
of an adaptive-droop regulated DC microgrid with battery management
capability," IEEE Transactions on Power Electronics, vol. 29, pp. 695-706, 2014.
118
[67] Q. Shafiee, T. Dragičević, J. C. Vasquez, and J. M. Guerrero, "Hierarchical
control for multiple DC-microgrids clusters," IEEE Transactions on Energy
Conversion, vol. 29, pp. 922-933, 2014.
[68] J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. De Vicuña, and M. Castilla,
"Hierarchical control of droop-controlled AC and DC microgrids—A general
approach toward standardization," IEEE Transactions on Industrial Electronics,
vol. 58, pp. 158-172, 2011.
[69] I. Khan, Y. Xu, S. Kar, and H. Sun, "Compressive Sensing-Based Optimal
Reactive Power Control of a Multi-Area Power System," IEEE Access, vol. 5, pp.
23576-23588, 2017.
[70] I. Khan, Z. Li, Y. Xu, and W. Gu, "Distributed control algorithm for optimal
reactive power control in power grids," International Journal of Electrical Power
& Energy Systems, vol. 83, pp. 505-513, 2016.
[71] M. Nasir, H. A. Khan, A. Hussain, L. Mateen, and N. A. Zaffar, "Solar PV-Based
Scalable DC Microgrid for Rural Electrification in Developing Regions," IEEE
Transactions on Sustainable Energy, vol. 9, pp. 390-399, 2018.
[72] C. S. Solanki, Solar photovoltaics: fundamentals, technologies and applications:
PHI Learning Pvt. Ltd., 2011.
[73] B. Subudhi and R. Pradhan, "A comparative study on maximum power point
tracking techniques for photovoltaic power systems," Sustainable Energy, IEEE
transactions on, vol. 4, pp. 89-98, 2013.
[74] M. Nasir and M. F. Zia, "Global maximum power point tracking algorithm for
photovoltaic systems under partial shading conditions," in Power Electronics and
Motion Control Conference and Exposition (PEMC), 2014 16th International,
2014, pp. 667-672.
[75] R. W. Erickson and D. Maksimovic, Fundamentals of power electronics: Springer
Science & Business Media, 2007.
[76] R. Pasonen, "Model of Bi-directional Flyback Converter for Hybrid AC/DC
Distribution System," International Journal of Power Electronics and Drive
Systems (IJPEDS), vol. 3, pp. 444-449, 2013.
[77] H. Saadat, Power system analysis: WCB/McGraw-Hill, 1999.
[78] V. Mehta and R. Mehta, Principles of power system: S. Chand, 1982.
119
[79] M. Nasir, N. A. Zaffar, and H. A. Khan, "Analysis on central and distributed
architectures of solar powered DC microgrids," in 2016 Clemson University
Power Systems Conference (PSC), 2016, pp. 1-6.
[80] B. Stott and O. Alsaç, "Fast decoupled load flow," power apparatus and systems,
ieee transactions on, pp. 859-869, 1974.
[81] G.-X. Luo and A. Semlyen, "Efficient load flow for large weakly meshed
networks," Power Systems, IEEE Transactions on, vol. 5, pp. 1309-1316, 1990.
[82] R. Farooq, L. Mateen, M. Ahmad, S. Q. Akbar, H. A. Khan, and N. A. Zaffar,
"Smart DC microgrids: Modeling and power flow analysis of a DC Microgrid for
off-grid and weak-grid connected communities," in Power and Energy
Engineering Conference (APPEEC), 2014 IEEE PES Asia-Pacific, 2014, pp. 1-6.
[83] S. Anand and B. Fernandes, "Optimal voltage level for DC microgrids," in
IECON 2010-36th Annual Conference on IEEE Industrial Electronics Society,
2010, pp. 3034-3039.
[84] M. J. Sarker, B. Asare-Bediako, J. Slootweg, W. Kling, and B. Alipuria, "DC
micro-grid with distributed generation for rural electrification," in Universities
Power Engineering Conference (UPEC), 2012 47th International, 2012, pp. 1-6.
[85] A. Sannino, G. Postiglione, and M. H. Bollen, "Feasibility of a DC network for
commercial facilities," Industry Applications, IEEE Transactions on, vol. 39, pp.
1499-1507, 2003.
[86] K. Engelen, E. L. Shun, P. Vermeyen, I. Pardon, R. D'hulst, J. Driesen, and R.
Belmans, "The feasibility of small-scale residential DC distribution systems," in
IEEE Industrial Electronics, IECON 2006-32nd Annual Conference on, 2006, pp.
2618-2623.
[87] A. Pratt, P. Kumar, and T. V. Aldridge, "Evaluation of 400V DC distribution in
telco and data centers to improve energy efficiency," in Telecommunications
Energy Conference, 2007. INTELEC 2007. 29th International, 2007, pp. 32-39.
[88] K. R. Varshney, G. H. Chen, B. Abelson, K. Nowocin, V. Sakhrani, L. Xu, and B.
L. Spatocco, "Targeting villages for rural development using satellite image
analysis," Big Data, vol. 3, pp. 41-53, 2015.
[89] B. Patterson, "LVDC: from concept to practice: an EMerge perspective," 2015.
[90] D. Salomonsson, L. Soder, and A. Sannino, "Protection of low-voltage DC
microgrids," IEEE Transactions on Power Delivery, vol. 24, pp. 1045-1053,
2009.
120
[91] N. F. P. Association, NFPA 70: National Electrical Code:
NationalFireProtectionAssoc, 2011.
[92] I. Marx, "Combining the best of both worlds," Industry Applications Magazine,
IEEE, vol. 16, pp. 30-34, 2010.
[93] A. F. X. Welsch, "DC Installation, Different Hazards Regarding Thermal Effects
and Electrical Shock," IEC LVDC Workshop, University of Applied Science,
Regensburg, 2011.
[94] X. Lu, K. Sun, J. M. Guerrero, J. C. Vasquez, and L. Huang, "State-of-charge
balance using adaptive droop control for distributed energy storage systems in DC
microgrid applications," IEEE Transactions on Industrial Electronics, vol. 61, pp.
2804-2815, 2014.
[95] Z. Jin, L. Meng, and J. M. Guerrero, "Comparative admittance-based analysis for
different droop control approaches in DC microgrids," in DC Microgrids
(ICDCM), 2017 IEEE Second International Conference on, 2017, pp. 515-522.
[96] M. Nasir, S. Iqbal, and H. A. Khan, "Optimal Planning and Design of Low-
Voltage Low-Power Solar DC Microgrids," IEEE Transactions on Power
Systems, vol. 33, pp. 2919-2928, 2018.
[97] M. Nasir, Z. Jin, H. Khan, N. Zaffar, J. Vasquez, and J. M. Guerrero, "A
Decentralized Control Architecture applied to DC Nanogrid Clusters for Rural
Electrification in Developing Regions," IEEE Transactions on Power Electronics,
2018.
[98] L. Meng, M. Savaghebi, F. Andrade, J. C. Vasquez, J. M. Guerrero, and M.
Graells, "Microgrid central controller development and hierarchical control
implementation in the intelligent microgrid lab of Aalborg University," in Applied
Power Electronics Conference and Exposition (APEC), 2015 IEEE, 2015, pp.
2585-2592.
[99] M. Hamza, M. Shehroz, S. Fazal, M. Nasir, and H. A. Khan, "Design and analysis
of solar PV based low-power low-voltage DC microgrid architectures for rural
electrification," in Power & Energy Society General Meeting, 2017 IEEE, 2017,
pp. 1-5.
[100] A. R. Desai, Rural sociology in India: Popular Prakashan, 1994.
[101] T. Odeyale and T. Adekunle, "Innovative and sustainable local material in
traditional African architecture–Socio cultural dimension," 2008.
121
[102] C. Linard, M. Gilbert, R. W. Snow, A. M. Noor, and A. J. Tatem, "Population
distribution, settlement patterns and accessibility across Africa in 2010," PloS
one, vol. 7, p. e31743, 2012.
[103] M. Silberfein, Rural settlement structure and African development: Westview
Press, 1998.
[104] S. C. Bhattacharyya and D. Palit, "Mini-grid based off-grid electrification to
enhance electricity access in developing countries: What policies may be
required?," Energy Policy, vol. 94, pp. 166-178, 2016.
[105] M. M. H. Sajeeb, A. Rahman, and S. Arif, "Feasibility analysis of solar DC Nano
grid for off grid rural Bangladesh," in Green Energy and Technology (ICGET),
2015 3rd International Conference on, 2015, pp. 1-5.
[106] S. Pervaiz and H. A. Khan, "Low irradiance loss quantification in c-Si panels for
photovoltaic systems," Journal of Renewable and Sustainable Energy, vol. 7, p.
013129, 2015.
[107] M. R. Maghami, H. Hizam, C. Gomes, M. A. Radzi, M. I. Rezadad, and S.
Hajighorbani, "Power loss due to soiling on solar panel: A review," Renewable
and Sustainable Energy Reviews, vol. 59, pp. 1307-1316, 2016.
[108] E. Forniés, F. Naranjo, M. Mazo, and F. Ruiz, "The influence of mismatch of
solar cells on relative power loss of photovoltaic modules," Solar Energy, vol. 97,
pp. 39-47, 2013.
[109] C. S. Solanki, Solar photovoltaics: fundamentals, technologies and applications:
PHI Learning Pvt. Ltd., 2015.
[110] J. Kolar, F. Krismer, Y. Lobsiger, J. Muhlethaler, T. Nussbaumer, and J.
Minibock, "Extreme efficiency power electronics," in Integrated Power
Electronics Systems (CIPS), 2012 7th International Conference on, 2012, pp. 1-
22.
[111] S. Matteson and E. Williams, "Residual learning rates in lead-acid batteries:
Effects on emerging technologies," Energy Policy, vol. 85, pp. 71-79, 2015/10/01/
2015.
[112] S. M. Schoenung and W. V. Hassenzahl, "Long-vs. short-term energy storage
technologies analysis: a life-cycle cost study: a study for the DOE energy storage
systems program," Sandia National Laboratories2003.
[113] M. Nasir, S. Iqbal, and H. A. Khan, "Optimal Planning and Design of Low-
Voltage Low-Power Solar DC Microgrids," IEEE Transactions on Power
Systems, 2017.
122
[114] S. C. Bhattacharyya, "Financing energy access and off-grid electrification: A
review of status, options and challenges," Renewable and Sustainable Energy
Reviews, vol. 20, pp. 462-472, 2013.
[115] N. J. Williams, P. Jaramillo, J. Taneja, and T. S. Ustun, "Enabling private sector
investment in microgrid-based rural electrification in developing countries: A
review," Renewable and Sustainable Energy Reviews, vol. 52, pp. 1268-1281,
2015.
[116] D. Palit, "Solar energy programs for rural electrification: Experiences and lessons
from South Asia," Energy for Sustainable Development, vol. 17, pp. 270-279,
2013.
[117] A. Newcombe and E. K. Ackom, "Sustainable solar home systems model:
Applying lessons from Bangladesh to Myanmar's rural poor," Energy for
Sustainable Development, vol. 38, pp. 21-33, 2017.
[118] Q. Bao, T. Honda, S. El Ferik, M. M. Shaukat, and M. C. Yang, "Understanding
the role of visual appeal in consumer preference for residential solar panels,"
Renewable Energy, vol. 113, pp. 1569-1579, 2017/12/01/ 2017.
[119] J. Palm, "Household installation of solar panels – Motives and barriers in a 10-
year perspective," Energy Policy, vol. 113, pp. 1-8, 2018/02/01/ 2018.
[120] K. A. W. Horowitz, R. Fu, T. Silverman, M. Woodhouse, X. Sun, and M. A.
Alam, "An Analysis of the Cost and Performance of Photovoltaic Systems as a
Function of Module Area,"; National Renewable Energy Lab. (NREL), Golden,
CO (United States) NREL/TP-6A20-67006 United States 10.2172/1351153
NREL English, 2017.