2019

10 Years of ISGT!
Innovation for a Flexible and Resilient Grid
REGISTER NOW
2019 Innovative Smart Grid Technologies (ISGT 2019)
18-21 February 2019 | Grand Hyatt Washington, Washington DC

Join experts from around the world as we gather at the ISGT North America Conference to
share experiences and lessons learned, and to raise awareness and understanding of the latest
concepts, applications, and technologies.

ISGT 2019 provides a forum to discuss the latest issues, trends, and emerging and innovative
technologies for grid modernization in the face of challenges of a rapidly changing environment
resulting from the dramatic increase in deployments of renewable and Distributed Energy
Resources (DERs) and the emergence of new business and operating concepts and services.
The topics discovered include prosumers, microgrids, aggregators, distribution markets, and
platforms for the generation, commercialization, and management of electricity.

The conference is organized along three tracks:

• Transmission and distribution systems planning and analysis
• Technology applications and supporting tools for transmission and
distribution system operations
• Market and policy considerations in facilitating innovation and enabling
a flexible and resilient grid

For more information visit: ieee-isgt.org

Digital Object Identifier 10.1109/MELE.2018.2880355

 IEEE Vol. 6, No. 4 december 2018 ISSN 2325-5987

www.ieee-pes.org/

MAGAZINE

f e at u r e s

6 The Long Road to

Community Microgrids
48 Microgrid Field Trials
in Sweden
Adapting to the necessary Expanding the electric
changes for renewable infrastructure in the
energy implementation. village of Simris.
Efraín O’Neill-Carrillo, Isaac Jordán, Henning Wilms, Dominik Mildt,
Agustín Irizarry-Rivera, Marco Cupelli, Antonello Monti,
and Rafael Cintrón Peder Kjellen, Thomas Fischer,
Demijan Panic, Michael Hirst,
Eugenio Scionti, Sebastian Schwarz,
18 An Enterprise Systems
Engineering Approach
Paul Kessler, and Luis Hernández

to Electrification
Looking at the bigger picture
63 Scalable Solar
dc Microgrids Cover image: Bringing power to all communities
with the help of microgrids.
through life-cycle analysis On the path to revolutionizing
of community microgrids: A case the electrification architecture Cover image: Paula Bronstein/IEEE Spectrum

study in Papua New Guinea. of developing communities.

Alexander Anderson and Mashood Nasir, Hassan Abbas Khan,
Siddharth Suryanarayanan Nauman Ahmad Zaffar, Juan C. Vasquez,
Mission Statement: IEEE Electrification Maga-
and Josep M. Guerrero zine is dedicated to disseminating information on
32 Solar Energy, dc Distribution, all matters related to microgrids onboard electric
and Microgrids
Ensuring quality power
73 Standalone Integrated
Power Electronics Systems
vehicles, ships, trains, planes, and off-grid applica-
tions. Microgrids refer to an electric network in a
car, a ship, a plane or an electric train, which has a
in rural India. Applications for off-grid limited number of sources and multiple loads. Off-
Ashok Jhunjhunwala and Prabhjot Kaur grid applications include small scale electricity sup-
rural locations. ply in areas away from high voltage power net-
David Schumacher, Omid Beik, works. Feature articles focus on advanced con-
40 Electric Vehicles
in India
and Ali Emadi cepts, technologies, and practices associated with
all aspects of electrification in the transportation
and off-grid sectors from a technical perspective in
A novel approach to scale
electrification.
83 Solar Generation,
Storage, and Electric
synergy with nontechnical areas such as business,
environmental, and social concerns.
Ashok Jhunjhunwala, Prabhjot Kaur,
and Sushant Mutagekar Vehicles in Power Grids IEEE Electrification Magazine (ISSN 2325-5897) (IEMECM)
is published quarterly by the Institute of Electrical and
Challenges and solutions ­Electronics ­Engineers, Inc. Headquarters: 3 Park Avenue,
with coordinated control 17th Floor, New York, NY 10016-5997 USA. Responsibility
at the residential level. for the contents rests upon the authors and not upon the
IEEE, the Society, or its members. IEEE Operations Center
Faeza Hafiz, Anderson Rodrigo (for orders, subscriptions, address changes): 445 Hoes
de Queiroz, and Iqbal Husain Lane, Piscataway, NJ 08854 USA. Telephone: +1 732 981
0060, +1 800 678 4333. Individual copies: IEEE members
US$20.00 (first copy only), nonmembers US$133.00 per
copy. Subscription Rates: Society members included with
d e pa r t m e n t s & c o l u m n s membership dues. Subscription rates available upon
request. Copyright and reprint permissions: Abstracting is
permitted with credit to the source. Libraries are permitted
to photocopy beyond the limits of U.S. Copyright law for the
private use of patrons 1) those post-1977 articles that
2 ABOUT THIS ISSUE Public Inter-Metro
carry a code at the bottom of the first page, provided the
91 dates ahead per-copy fee indicated in the code is paid through the
dc Fast Charge
Public in Metro Areas
Level 1 and 2 and dc Fast Charge Copyright Clearance Center, 222 Rosewood Drive, Danvers,
92 Newsfeed Fleets
Level 1 and 2 and dc Fast Charge
MA 01923 USA; 2) pre-1978 articles without fee. For other
copying, reprint, or republication permission, write
Workplace

94 viewpoint Multifamily Home Level 1 and 2

102 technology leaders Single-Family Home
The charging pyramid representing
necessary relative abundance
of EV charging systems.
Page 96.
Copyrights and Permissions Department, IEEE Operations
Center, 445 Hoes Lane, Piscataway, NJ 08854 USA.
Level 1 and 2
Copyright © 2018 by the Institute of Electrical and
Electronics Engineers, Inc. All rights reserved. Postmaster:
Level 1 and 2 Send address changes to IEEE Electrificaton Magazine, IEEE
Operations Center, 445 Hoes Lane, Piscataway, NJ 08854
USA. Canadian GST #125634188 Printed in U.S.A.

Promoting Sustainable Forestry

Digital Object Identifier 10.1109/MELE.2017.2788974 SFI-01681

 Edi t orial board Chris Searles
about this issue Iqbal Husain Editor, Electric Vehicles
Editor-in-Chief BAE Batteries
North Carolina State Wisconsin, USA
University Chris.Searles@
North Carolina, USA baebatteriesusa.com
ihusain2@ncsu.edu
Silva Hiti
Tamas Ruzsanyi Editor, Electric Vehicles

The Socially
Editor, Electric Trains Faraday Future
Ganz-Skoda California, USA
Hungary silva.hiti@ff.com
tamas.ruzsanyi@ieee.org

Responsible
Eduard Muljadi
Eduardo Pilo de la Fuente Editor, Electric Vehicles
Editor, Electric Trains Auburn University
EPRail Research Alabama, USA

Microgrid and Consulting

Spain
eduardo.pilo@eprail.com
mze0018@auburn.edu

Jose Conrado Martinez
Editor, Electric Trains
Directcion de Estrategia
By Rob Cuzner y Desarrollo
Spain
jcmartinez@adif.es
Suryanarayana Doolla
Editor, Microgrid
Indian Institute of
­Technology Bombay
oday, many factors are driving significant
Syed A. Hossain
Editor, Electric Planes
GE Aviation
Ohio, USA
syed.hossain@ge.com
Kaushik Rajashekara
Editor, Electric Planes
University of Houston
Texas, USA
ksraja@central.uh.edu
Babak Nahid-Mobarakeh

T
India Editor, Electric Planes
shifts in electrical-energy delivery paradigms. If you are suryad@iitb.ac.in University of Lorraine
tuned in to what microgrids are all about, you will recog- France
Mohammad Babak.nahid@ieee.org
nize their role as part of the solution to so many of the challenges Shahidehpour
we face today: extreme weather, vulnerable and aging electrical Editor, Microgrid Bulent Sarlioglu
Illinois Institute Editor, Electric Planes
infrastructures, changing transportation needs, and socioeconomic of Technology University of
disparities. Whereas the microgrid first emerged as a means for Illinois, USA Wisconsin-Madison
ms@iit.edu Wisconsin, USA
integrating distributed renewable resources into an existing electri-
bulent@engr.wisc.edu
cal-grid infrastructure, the fact that the microgrid can operate Steve Pullins
independently from the grid as a power and energy delivery sys- Editor, Microgrid IEEE Periodicals
GridIntellect Magazines
tem, tailored to the needs of local users, is becoming increasingly Tennessee, USA ­Depar tment
important. Combining the concepts of microgrid, smart grid, and steve@gridintellect.com 445 Hoes Lane,
shared-energy resources together with a localized need for Piscataway, NJ 08854 USA
Antonello Monti
dependable, secure, and resilient power delivery gives you a com- Editor, Microgrid Debby Nowicki
RWTH Aachen Managing Editor
munity microgrid. If you recognize that there is a human need at
Germany
the receiving end of microgrid power, you have a socially responsi- Geri Krolin-Taylor
amonti@eonerc
Senior Managing Editor
ble microgrid, one with a priority to provide critical needs that are .rwth-aachen.de
Janet Dudar
unique to the people being serviced. Highlighted by recent events, Marta Molinas Senior Art Director
such as the effects of Hurricane Maria on Puerto Rico, it is becom- Editor, Electric Ships
Norwegian University of Gail A. Schnitzer
ing universally apparent that the availability of reliable electrical Associate Art Director
Science and Technology
energy is a basic human right, the lack thereof resulting in devas- Norway Theresa L. Smith
tating human costs. Since 2013, past issues of IEEE Electrification marta.molinas@ntnu.no Production Coordinator
Magazine have highlighted many aspects of the community Herb Ginn Felicia Spagnoli
microgrid, from the benefits of interconnectivity, to rural electrifi- Editor, Electric Ships Advertising Production
University of Manager
cation, to transactive energy. This issue revisits the salient aspects South Carolina
of community microgrids today: how they will be effectively South Carolina, USA Peter M. Tuohy
ginnhl@cec.sc.edu Production Director
deployed; how they will be architected; how they will be sustained;
Kevin Lisankie
and, most importantly, how the societies full of individuals with Robert Cuzner Editorial Services Director
real human needs interact with them. Editor, Electric Ships
University of Dawn M. Melley
This issue begins with “The Long Road to Community Micro­­ ­Wisconsin-Milwaukee Staff Director,
grids” by Efraín O’Neill-Carrillo, Isaac Jordán, Agustín Irizarry-Rivera, Wisconsin, USA Publishing Operations
and Rafael Cintrón, researchers at the University of Puerto Rico-­ Cuzner@uwm.edu
advertising sal es
Mayaguez (UPRM). This article highlights the vast challenges that Khwaja Rahman Erik Henson
Editor, Electric Vehicles Naylor Association Solutions
General Motors Tel: +1 352 333 3443
Digital Object Identifier 10.1109/MELE.2018.2871238 Michigan, USA Fax: +1 352 331 3525
Date of publication: 16 November 2018 khwaja.rahman@gm.com ehenson@naylor.com

2 I EEE E l e c t ri f i c a t i on M a gaz ine / DECember 2018 Digital Object Identifier 10.1109/MELE.2017.2788975

result when governmental policies and institutions drive
remote and developing areas to adopt the centralized, hierar-
chal infrastructure for electrification. A historical perspective is
provided as to how Puerto Rico arrived at its present electric Today’s ADVANCED flexible
infrastructure, which integrated all electric power systems electrical insulation just got better.
provided to the island under a single public entity, the Puerto
Rico Electric Power Authority (PREPA). Advocates for the intro-
duction of renewables into the centralized grid were met with Introducing
resistance until 2007, when net-metering laws were intro- ELAN-Film® HT-200 Insulation
duced. Since then, Puerto Rico has demonstrated the econom-
• Excellent for use as wire wrap and layer
ic feasibility of rooftop photovoltaic (PV) systems as the sole insulation in transformers and lifting
means of renewable production. However, the PREPA-man- magnets NEW
aged Puerto Rico utility grid provides a stark illustration of the • Unique ratio of PAI/PET provides
outstanding electrical, mechanical,
problems associated with top-down electrical infrastructures. and thermal properties
These problems include the incompatibility of locally pro- • Higher temperature 200oC rating
duced solar with a grid structure based on the assumption of • 2 mil thickness is thin and flexible

constant power output and the inability of an electrical grid • Meets IPC Specification 4202

designed for an industrial economy to adapt to the reality of

Puerto Rico’s 48% loss of industrial demand since 2006.
A series of impregnating resins A superior magnet wire coating
Attempts by the PREPA to address energy restructuring that provide environmental and designed for use in inverter
came too little, too late, and in 2017, the company filed for electrical protection in all types grade magnet wire.
of electrical apparatus.
A Heritage of Innovation Since 1919
bankruptcy. Finally, the devastating impact of Hurricane
Maria on 20 September 2017 brought worldwide attention to ELANFilm@altana.com Elan-Film.com
the human impacts of a poorly planned, centrally managed ELANTAS PDG, Inc. •Second
5200 North Second
ELANTAS PDG Inc. • 5200 North Street • Street • St. Louis
St. Louis, MO MO 63147
63147
grid. Despite resistance, UPRM led the movement toward 314-621-5700 •ELANTAS PDG Olean
info.elantas.pdg • 1405 Buffalo
@altana.com Street • Olean NY 14780
• www.elantas.com

community-based distributed energy resources for more

than a decade and achieved success in developing stakehold-
ers in this effort at the community level. Today, Puerto Rico is
poised to lead the world in the deployment of distributed
community-based, renewable-drive microgrids. UPRM has Are You Moving?
played a significant role in preparing for this day. This article Don’t miss an issue of this magazine—
describes the feasibility studies, viability considerations, and update your contact information now!
proposed architectural implementations that will enable Update your information by:
Puerto Rico’s transition to a community microgrid-based E-MAIL: address-change@ieee.org
infrastructure for electric power delivery. PHONE: +1 800 678 4333 in the United States
On another front, IEEE Smart Village has been a catalyst or +1 732 981 0060 outside
for ensuring that developing parts of the world do not make the United States
the mistake of adopting top-down, centralized approaches as If you require additional assistance
their electrical infrastructures develop. “An Enterprise Sys- regarding your IEEE mailings,
tems Engineering Approach to Electrification” by Alexander visit the IEEE Support Center
Anderson and Siddharth Suryanarayanan documents a case at supportcenter.ieee.org.
study in Papua, New Guinea, demonstrating the impact of
community-based electrification on the development of sus- IEEE publication labels are printed six to eight weeks
in advance of the shipment date, so please allow sufficient
tainable electrical infrastructures and increased prosperity. time for your publications to arrive at your new address.

This article highlights the energy poverty that affects more

than 1 billion people worldwide who still lack access to elec-
tricity. A community system-of-systems enterprise is
described, showing how the community-based microgrid for
electrification must be part of a hierarchy spanning social,
economic, and technical levels.
This approach is far superior to traditional government-led
developmental planning based on the assumption of, for
instance, a future industrialization and productivity mimick- © ISTOCKPHOTO.COM/BRIANAJACKSON

ing the norms of developed countries. With this approach, the

IEEE Elec trific ation Magazine / D EC e mb e r 201 8 3

 about this issue
underlying foundation of supporting
services and sustainable partnerships
holds up the basic needs for survival,
leading to growth enabled by the
reduction of inequities that informs
future economic growth, industrial
innovation, and infrastructure growth.
The ultimate goal is the elimination of
poverty. The first priority of this
approach is the undergirding com-
munity enterprise of organizations,
entrepreneurs, processes, systems,
technologies, stakeholders, and com-
munity members to keep the system
moving forward, adapting to changing
needs. At the core is demonstrated
scalable microgrid topologies for de­­
veloping communities, which IEEE
Smart Village pioneered. With Papua
New Guinea as a backdrop, this article
describes in detail the systems-engi-
neering approach needed to propel
and sustain community-based microgrid
initiatives of the future.
India has been at the forefront of
the battle on energy poverty, where
community dc microgrid installations
in rural areas have outpaced the gov-
ernment’s effort to tie remote villages
into the national grid. “Solar Energy, dc
Distribution, and Microgrids” by Ashok
Jhunjhunwala and Prabhjot Kaur of
the Indian Institute of Technology (IIT),
Madras, highlights the challenges of
electrifying villages that are isolated
from more developed areas by diffi-
cult, sometimes treacherous, terrain.
These villages are on mountains, in
deserts, located remotely on islands,
and across rivers. IIT Madras pio-
neered the use of low-cost, scalable,
and distributed solar dc systems that
can be easily deployed and configured
to the needs of community users.
Compared with many parts of the
world, the installed systems have sig-
nificantly lower power- and energy-
sourcing capacity.
However, it has been demonstrated
that these 48-Vdc distribution systems
can be well matched to homes in
these remote communities, and their
deployment has resulted in signifi-
cant strides toward the reduction of
4 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018
energy-poverty effects. This article de­­
scribes the efforts by IIT Madras to fur-
ther develop its 48-Vdc systems as
power-dense, modular, scalable, plug-
and-play systems with constituent
parts that are transported through
harsh terrain. Unburdened by the
need to support an entrenched infra-
structure of ac-fed appliances and fix-
tures, the rural Indian household has
also provided a fertile ground for mar-
ket growth in dc-powered appliances.
Shifting paradigms demonstrated in
energy, informing the world how to
adapt to changing, potentially more-
efficient power and energy delivery
systems of service are explored.
“Electric Vehicles in India,” by
Ashok Jhunjhunwala, Prabhjot Kaur,
and Sushant Mutagekar, provides a
unique perspective on the integration
of electric vehicles (EVs) into a devel-
oping and dynamic electrical infra-
structure. India is on track to adopt
the EV as the preferred form of land
transportation by the majority of its
population, particularly in large cities.
This trend is becoming imperative
considering the levels of air pollution
in cities and the contributing effects
on population health and green gases.
The Indian government has recog-
nized the need to take action. This
article describes the country’s unique
situation regarding vehicle composi-
tion and density. India also faces sig-
nificant infrastructure challenges,
including how to provide an EV-charg-
ing infrastructure commensurate to
the shear demand. The country’s strat-
egy to develop an EV-charging infra-
structure that will meet present needs
and that is sustainable for the future is
described in detail.
Shifting to the application of local-
ly targeted, distributed microgrids
to developed parts of the world, “Mi­­
crogrid Field Trials in Sweden,” by
Henning Wilms, Dominik Mildt, Marco
Cupelli, Antonello Monti, Peder Kjel-
len, Thomas Fischer, Demijan Panic,
Michael Hirst, Eugenio Scionti, Sebas-
tian Schwarz, Paul Kessler, and Luis
Hernández, of the E.ON, addresses the
challenge of networking DERs to
accommodate the rapid development
and growth of renewable-energy sys-
tems. In Germany, it has become clear
that energy systems need to become
more networked to meet climate objec-
tives. However, the present microgrid
hierarchical control architecture, which
is stratified into a pyramidal structure
of component level, primary control,
secondary control, and tertiary control
based on time-response horizons con-
trolled by a distribution system opera-
tor, is becoming a limiting factor to the
growth of renewable-energy systems.
Many challenges are expected in
the networking of power and energy
delivery systems that, if not ad­­dressed,
will result in reductions in system
resiliency and bottlenecks to capacity.
This article describes the implementa-
tion of the Simris energy system by the
E.ON that was initiated as a technical
pilot trial to address the expected chal-
lenges associated with the growth of
renewable energy systems. The instal-
lation is located in the small village of
Simris in the southern area of Sweden
and demonstrates integration of a
local community microgrid with an
existing grid combined with a diversi-
ty of renewable- and storage-energy
sources, an intelligent energy man-
agement system, protective relay de­­
vices, and a demand-response platform.
This article describes in detail the
constituent parts of the system, ad­­
aptive and predictive controls, and
measured performance during transi­­
tional scenarios.
The remainder of the issue ad­­
dresses the technologies that will
be needed to manage the growth in
distributed, community microgrids.
“Scalable Solar dc Microgrids,” by Mas-
hood Nasir, Hassan Abbas Khan, Nau-
man Ahmad Zaffar, Juan C. Vasquez,
and Josep M. Guerrero, describes how
to optimize and extend the dc mi­­cro­
grid concept to larger systems. An
often neglected consideration of dis-
tribution efficiency as a function of
the number of households and distri-
bution voltage levels is presented in
the article. Various distribution archi- Juan C. Vasquez, and Josep M. Guerre- and users of distributed energy re­­
tectures are considered as well as ro, explains the role of the prosumer sources can help to define a new col-
challenges to economic viability, provi- in energy transactions within the laborative economy. We are excited to
sioning, and deployment. “Standalone community-based microgrid. Here, provide this special issue of IEEE Elec-
Integrated Power Electronics Systems,” the way forward to resiliency and sus- trification Magazine and are certain
by David Schumacher, Omid Beik, and tainability can be solidified as the that the readers will be convinced
Ali Emadi, takes a detailed look at the human factors and responses to the that community microgrids are a way
implementation of portable, off-grid microgrid are considered. In this way, forward to a better world!
microgrid architecture. These systems energy sharing between the owners 
serve as power and energy manage-
ment and delivery systems suitable
for connection to nonconventional
power distributions applied to remote
community microgrids. Low-voltage
dc, higher-voltage dc, low-voltage
ac, and higher-voltage ac interfacing
con­­figurations are compared. Deploy-
ability is also considered, taking into
account an array of commercially
available building-block components
and systems.
The article “Solar Generation, Stor-
THE IDEAL CONTACTORS
age, and Electric Vehicles in Power FOR SOLAR APPLICATIONS
Grids,” by Faeza Hafiz, Anderson
Rodrigo de Quieroz, and Iqbal Husain,
addresses the need for and imple-
mentation of integrated power and
energy systems for the residential
application. This article describes
how a household solar PV-storage
system, integrated with homeowner
plug-in EVs with system-coordinated
control at the residential level, can
provide energy security and benefits
to the overall grid. While not address-
ing specifically the community mi­­
crogrid, this article emphasizes the
fact that successful implementation Developed for use in solar combiner boxes and inverters
of microgrids is a simultaneous bot-
toms-up/top-down problem. Com- The smallest UL508 recognized contactors rated for
high current switching
munity microgrids will not be viable,
particularly in developed countries Rated for switching 175 amps at 1000Vdc and
like the United States, until there are a continuous duty up to 350 amps
commercially viable solutions avail-
able to the user. Featuring GIGAVAC’s EPIC® sealed switching technology
This issue is rounded out by the for superior protection against harsh environments
“Viewpoint” and “Technology Leaders”
GIGAVAC’s dual coil design means low coil power
columns. “Electrification and Chal- consumption and no EMI emission or crosstalk
lenging Our Preconceptions,” by Peter
O’Connor and Michael Jacobs, contin-
ues the discourse on societal impacts. Fo r c o m p l e t e s p e c s , v i s i t u s o n l i n e
“Peer-to-Peer Energy Market for Com-
munity Microgrids,” by Pablo Baez- o r c a l l +1.8 0 5.6 8 4.8 4 01

w w w. g i ga va c.c o m
Gonzalez, Enrique Rodriguez-Diaz,
 By Efraín O’Neill-Carrillo, Isaac Jordán,
Agustín Irizarry-Rivera, and Rafael Cintrón

The Long Road

to Community
Microgrids
Adapting to the necessary icrogrids offer tremendous oppor-

M
tunities in isolated areas or developing
changes for renewable countries that are unable to establish a tra-

energy implementation. ditional power infrastructure. However,

Digital Object Identifier 10.1109/MELE.2018.2871239
Date of publication: 16 November 2018
microgrids face vast challenges in places
where the dominant model is based on a centralized, hier-
archical infrastructure with policies and institutions
developed to support such infrastructure. Renewable ener-
gy researchers at the University of Puerto Rico–Mayaguez

6 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018 2325-5987/18©2018IEEE

 (UPRM) have traveled a long road advocating
distributed energy resources (DERs) and com-
munity microgrids.
Overview of Electric
Energy in Puerto Rico
In 1941, the Puerto Rico Water Resources
Authority (AFF, its acronym in Spanish) was
created as a public power company to plan,
design, construct, operate, and maintain
Puerto Rico’s electric infrastructure. As part of
a plan to take many Puerto Ricans out of the
dire conditions they lived in the 1930s, Anto-
nio Lucchetti proposed to integrate all of the
electric power systems in Puerto Rico. War-
time circumstances enabled the final acquisi-
tion of private power companies, which
helped the AFF complete its mission: the
electrification of Puerto Rico.
The AFF accomplished its founding mis-
mountain: ©istockphoto.com/alejandrophotography,
solar panels—image licensed by ingram publishing
sion in the 1970s. While its name changed to
the Puerto Rico Electric Power Authority
(PREPA) in 1979, its mission stayed the same.
As of December 2018, PREPA is a vertically
integrated utility and is one of the largest
public power companies in the United States,
and remains the only provider of retail elec-
tricity on the island (Figure 1). It has more
than 2,400 mi of transmission lines (230 and
115 kV), 51 115-kV transmission centers, 283
subtransmission substations (38 kV), and
over 30,000 mi of distribution lines (13.2, 8.32,
7.2, and 4.16 kV). New technologies, practices, and oppor-
tunities were missed because of the lack of a renewed
mission to face the challenges of the latter quarter of the
20th century and begin an ordered
and comprehensive transforma-
tion of the electric infrastructure,
its business structure, and custom-
er service. On the other hand, 248 MW
Caonillas Cambalache
there was no holistic planning or
integration of energy strategies
and technologies in Puerto Rico,
220 MW
and changes in energy policy direc-
Mayaguez
tions (due to excessive partisan inter-
ventions in PREPA) were an obstacle Garzas I and II
for decades.
An area of disagreement in the
last 20 years between PREPA’s Yauco I and II
Costa Sur
management and electricity users
1,032 MW
has been the use of renewable
Oil (6, 2, or Diesel)
sources to produce electricity. In Hydro (100 MW)
spite of PREPA’s narrow planning Coal
Natural Gas
vision, a net-metering law was
enacted in 2007, and a renewable
portfolio standard became law in Figure 1. The installed generating capacity in Puerto Rico. (Image courtesy of Dr. Agustín Irizarry.)
2010. In 2008, researchers at UPRM quantified the poten-
tial for electricity generated from local renewable energy
sources. They found that the best renewable resource
with commercially available technology in Puerto Rico is
the sun (Figure 2). Figure 3 shows the potential for gener-
ating solar electricity using photovoltaic (PV) panels on
residential rooftops. All of the electric energy in Puerto
Rico could come from using PV systems in 65% of resi-
dential rooftops once we obtain an efficient, cost-effec-
tive, and environmentally sound technology to store
electric energy for use at night. Figure 4 shows results
from another UPRM study [funded by the U.S. Depart-
ment of Energy (DOE)] proving that residential rooftop PV
systems already had an average cost of between 11 and
12 cents per kWh in 2013 (assuming an installed cost of
US$3 per watt, net metering, and four hours of “peak
sun”). In contrast to most U.S. jurisdictions, rooftop PV
systems with net metering have been economically fea-
sible in Puerto Rico since at least 2010, reaching grid pari-
ty near that time.
A key challenge is to turn the potential for renewable
energy into a reality through a safe and reliable grid con-
nection. The differing opinions regarding this matter result
from the fact that traditional power systems were built and
operated based on the assumptions of that time. Conse-
quently many electric companies around the world (includ-
ing PREPA) have usually reacted negatively to renewable
energy, believing that it complicates conventional power
system operation. Despite these challenges, distributed
generation continues to grow in Puerto Rico, furthered by
the passing of Act 133 in 2016, which permits solar commu-
nities and microgrids on the island.
PREPA’s mission was finally updated in 2014 to reflect a
new mandate toward a sustainable energy future. A new
regulator, the Puerto Rico Energy Commission, was created
728 MW
Dos Bocas 840 MW
Palo Seco
San Juan
Rio Blanco
Toro Negro I and II
ecoEléctrica Aguirre AES
507 MW 454 MW
1,534 MW
Not Shown: 386 MW from Smaller Units Distributed
Around the Island
Installed Capacity: 5,839 MW (3,443 MW in the South)
IEEE Elec trific ation Magazine / D ec emb er 201 8 7
 Toa Baja
Quebradillas Hatillo Barceloneta Vega Baja Carolina
Manati Vega Alta Cataño Canovanas
Dorado
Aguadilla
Rincón Isabela San Loiza
Arecibo

Guaynabo
Juan

Bayamón
Moca Camuy
Aguada San Florida Toa Alta Trujillo Río Luquillo
Sebastián Morovis Grande
Corozal Alto Fajardo
Añasco Ciales
Lares Utuado Naranjito Aguas Gurabo
Ceiba
Las Marías Orocovis Comerio Buenas Juncos Naguabo
Mayagüez Jayuya Barranquitas Caguas
Hormigueros Maricao Adjuntas Cidra San
Villalba Aibonito Humacao
San Cayey Lorenzo
Yauco Coamo Yabucoa
Germán Ponce
Cabo Juana
Las
Rojo Lajas Díaz Santa Salinas
Piedras
Guánica Isabel Guayama
Maunabo
Sabana Guayanilla Peñuelas Patillas
Grande Arroyo

1,495 → 3.4 h 1,952 → 4.5 h 2,408 → 5.5 h

1,343 → 3.1 h 1,800 → 4.1 h 2,256 → 5.2 h
1,191 → 2.7 h 1,648 → 3.8 h 2,104 → 4.8 h

Figure 2. The annual average insolation in kWh/m2 and peak sun hours.

in 2014. Also in that year, PREPA began a restructuring pro-

40
cess to address the financial problems that had been
35
Annual GWh Generated

evident since the early 2000s. The existing electric infra-

30
structure was built for an industrial economy, but the loss
25 of 1,000 industrial clients left Puerto Rico with a generation
20 overcapacity. The decade between 2007 and 2017 saw a
15 48% reduction in industrial demand (after its 2006 peak), a
10 13% loss in the residential sector (after its 2005 peak), and a
5 10% loss in commercial demand (after its 2007 peak). In
0 2017, PREPA filed for bankruptcy protection under Title III
0 10 20 30 40 50 60 70 80 90 100 of the Puerto Rico Oversight, Management, and Economic
Available Rooftop Area (%)
Stability Act, a U.S. law. One option for overcoming PREPA’s
Figure 3. The potential for electricity production from residential roof- difficult circumstances is to facilitate the use of DERs,
top PV systems. which present an opportunity for local socioeconomic
development—if priority is given to
renewable energy in a hybrid sys-
tem that enables solar communities
Levelized Cost of Energy (LCOE), US$/kWh and microgrids.
(Net Metering, 20 Years, 1% Annual Degradation)
US$0.25
0.22
Renewable-Driven Microgrids
US$0.20 EI Yunque 0.19 0.18
0.16 In Puerto Rico, approximately 70% of
0.15
US$/kWh

US$0.15 0.13 0.13 the population lives in areas with an

0.17 0.15 0.13 0.12 0.11
US$0.10 0.12 excellent solar resource (Figure 2).
0.11 0.10
0.09 0.09 0.08 Hence, rooftop PV systems in Puerto
US$0.05 Adjuntas, Canóvanas,
Cabo Rojo Sur
Mayaguez Norte Guánica Rico have greater potential when
US$0.00 compared to other renewable re­­
2.5 3 3.5 4 4.5 5 5.5
sources. PV systems have some
Peak Sun Hours
drawbacks, however, such as the
LCOE 4 US$/W LCOE 3 US$/W fluctuation of output power, which
depends on weather conditions. One
Figure 4. The cost per kilowatthour of residential rooftop PV systems. way to increase the penetration of

8 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

PV generation is to add energy storage One option for per group (12.8 kWh per house). Loads
devices. With storage and a dedicated were simulated with one of three
control system, PV generators can be overcoming PREPA’s demand profiles: 100 houses used
transitioned to serve the role of an 834  kWh per day, 50 houses used
active generator and provide more
difficult circum­­ 918 kWh per day, and 50 houses used
flexibility for system operators and stances is to 627 kWh per day. Each grouping of
users. This approach is limited by the houses had a different combination of
economic and environmental issues facilitate the use profiles and was not uniformly dis-
related to existing storage technolo- tributed. Load reductions represented
gies. Another way to increase the use
of DERs, which present DR actions. When load reduction was
of rooftop PV systems is by managing an opportunity for triggered, the 834- and 918-kWh pro-
the demand through demand-response files were reduced by 33%, while the
(DR) strategies, which can be used to local socioeconomic 627-kWh profile was reduced by 25%.
match load profiles to available renew- Demanding a constant block of
able energy production.
development—if electric energy from the utility mini-
It is well known that many commu- priority is given to mizes the impact resulting from the
nity-led efforts in Puerto Rico have interconnection of renewable sources.
yielded better results when dealing renewable energy The goal is to overcome two main
with local problems than top-down objections from utilities regarding the
approaches. There are many well-orga-
in a hybrid system penetration of renewable energy: fluc-
nized communities in Puerto Rico that that enables solar tuation in power output and unbal-
share a common, albeit local, vision of anced energy generation (mainly from
their future and use this vision to communities and PV) that produces energy only during
inspire the actions of many local volun- the day. At the transmission level,
teers eager to improve their communi-
microgrids. where the stability of the whole sys-
ties. They put their community first, tem is at stake and varying power may
ahead of other considerations such as cause a significant variable operation-
political party bickering. Thus, combining these ongoing al cost, electric utilities often penalize connected clients if
community-based efforts with DERs (including DR) could they violate the LF agreement in the service contract:
enable local energy options in support of socioeconomic
development and community well-being. In the near future, average load
LF = .
this could lead to well-planned community microgrids in maximum load in given period
Puerto Rico.
LF in a solar-community microgrid would greatly
Technical Feasibility Study depend on the mix of installed solar-generation capacity,
of a Community Microgrid storage, and demand peaks. Moreover, without a good
A zero or low-net-energy community microgrid is one that resource-management plan, achieving a high LF could be
has sufficient resources to locally manage its demand, difficult and/or costly. Simulation scenarios were used to
thereby minimizing its impact on the power grid. A study study the behavior of a microgrid with different levels of
of the possibility for a community microgrid serving an PV capacity and weather conditions, indicating that differ-
actual community in southern Puerto Rico divided the ent approaches to mitigate and improve LF should be pur-
community into 20 groups of ten houses, each with PV sys- sued depending on the conditions and available resources.
tems and energy storage. PV systems and energy storage On sunny days, the state of charge of the energy storage
were simulated as aggregated systems for each grouping of could reach 100%; if this happens, the storage would be
ten houses, as shown in Figure 5. The battery’s state of disconnected and the excess renewable energy injected
charge was modeled by adding or subtracting the energy into the grid, thus drastically reducing the LF. One
sent or received by the microgrid and obtaining output approach to mitigate the inevitable LF reduction is to
decisions from a fuzzy logic controller. The microgrid was charge the battery at a slower rate and reduce the amount
grid connected, and the power it delivered could not of energy coming from the grid. Of course, an LF near one
exceed a certain amount using the highest possible load would be impossible because of the grid reduction that
factor (LF) (i.e., near constant demand). The grid was mod- occurs from charging the battery more slowly; however, it
eled as an infinite bus. is preferable to disconnecting the battery altogether,
Each grouping of ten houses was connected to a which would send excess renewable energy production to
75-kVA distribution transformer. The aggregated PV sys- the grid, thereby drastically lowering the LF.
tem varied from 10 to 20 kW per group (1 to 2 kW per Table 1 shows how the LF is affected when grid depen-
house), and the aggregated battery storage was 128 kWh dence is decreased and the installed renewable-generation

IEEE Elec trific ation Magazine / D ec emb er 201 8 9

10
Isec_8
Solar Panel Battery House 1 House 2 House 3
+

Pole-Mounted PV_I_8
Transformer
Battery_I_8 Load4_I_8
Load1_I_8 Load2_I_8 Load3_I_8

2
1
+V
3 –

I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

Vsec_8

Battery_Ah1 IL1_8 IL2_8 IL3_8 IL4_8

Ipv_8 Ibattery_8

Load 5_I_8 Load 6_I_8 Load 7_I_8 Load 8_I_8 Load 9_I_8 Load 10_I_8 Load CC

IL 9_8 IL CC
IL 6_8 IL 7_8 IL 8_8 IL 10_8
IL 5_8

Figure 5. One group of houses on the microgrid. (Image modeling courtesy of Isaac Jordan using Simulink.)
capacity increased, while the amount During a good managing the energy demand coming
of storage remains the same. However, from the grid by considering the mi­­
the degree to which it is affected insolation day and crogrid’s resources, while the second
varies tremendously according to focuses on managing internal demand.
weather conditions; in sunny scenar-
if the batteries are In a solar-community microgrid, man-
ios, the LF is reduced as one moves charged or nearly aging the demand from the grid could
toward more renewable-generation bring economic benefits; the microgrid
capacity and/or less grid dependence charged, the operator could determine, based on the
on zero net. Nevertheless, there are status of microgrid resources, the best
exceptions, e.g., in cases 2 and 3, the
microgrid operator strategy for operating the microgrid
system has a slightly better LF when can demand less reliably and economically. For example,
it has more solar-­generation capaci- during a good insolation day and if the
ty in­­stalled, which means that case 3 energy from the grid batteries are charged or nearly charged,
is the desirable operating point for the microgrid operator can demand
the microgrid. Case 3 has an LF simi-
during times when less energy from the grid during times
lar to case 1 (0.89 versus 0.91) and has the energy prices when the energy prices are high.
more renewable-generation capacity Taking as an example the time-of-
in­­stalled. In a cloudy scenario, LF is are high. use rate for its simplicity, cases 9 and
more dependent on the mix, and one 10 with DR simulated a change in peak
can see many variations that are demand, i.e., when energy prices are
reduced significantly as zero net is approached. higher (see Table 1). In case 9, the microgrid needs
In the scenarios, LF improves as more renewable capaci- 10 kW of power per group of ten houses from the grid off
ty is installed, while requiring the same amount of energy peak. When the night peak is reached, the microgrid
from the grid. When the required grid energy is reduced, reduces its demand to 8 kW per group; the same scenario
cloudy days have a more severe impact than sunny days if occurs in case 10, but with their respective boundaries.
the installed renewable capacity increases. The gap increase In contrast to a sunny day, for a cloudy-day scenarios, no
between sunny days and cloudy days makes microgrid demand reduction is made because not enough energy is
operation more difficult, especially in cloudy-day scenarios. produced to completely charge the battery; therefore, the LF
It would be possible to require more energy from the grid on and renewable generation on cloudy days in Table 1 cor­
cloudy days and less on sunny days, but the microgrid oper- respond to a nondemand grid-reduction scenario. This
ation cost will rise. Additionally, this solution moves in the demand-side management strategy can help tackle the big
opposite direction of zero-net energy. Overall, the microgrid energy gap problem that exists between sunny days and
operation maintains a good LF
on sunny days and cloudy days
when DR is used. On sunny days, Table 1. The LF and renewable generation for sunny and
an LF improvement of more than cloudy days.
0.5 is easily achieved, as com-
Secondary Installed Sunny/Cloudy
pared to scenarios with no DR
Case ­Demand Goal PV ­Capacity Sunny/ ­Renewable
and no energy storage. ­Number per Group per Group Cloudy LF Generation (%)
Going to net-zero is a real chal-
lenge because the LF is reduced 1 10 kW 10 kW 0.91 0.92 27 14
regardless of the scenario, as 2 7.5 kW 10 kW 0.86 0.38 27 14
shown in Table 1. Case 5 is the 3 7.5 kW 15 kW 0.89 0.67 40 22
only scenario in the community
4 6.5 kW 15 kW 0.81 0.33 40 22
microgrid that generated more
than half its electric energy from 5 6.5 kW 20 kW 0.77 0.51 53 29
renewables when power from 6 No DR, no 10 kW 0.33 0.38 27 14
the grid was at its lowest (PV-in­­ ­energy storage
stallation capacity is the highest
7 No DR, no 15 kW 0.27 0.35 40 22
of all, and additional renew- ­energy storage
able generation occurs only on a
8 No DR, no 20 kW 0.21 0.32 53 29
sunny day).
­energy storage
There are essentially two per-
spectives of demand-side man- 9 8–10 kW 10 kW 0.85 0.92 27 14
agement in a solar-community 10 7–8 kW 15 kW 0.85 0.85 40 22
microgrid. The first focuses on

IEEE Elec trific ation Magazine / D ec emb er 201 8 11

 cloudy days. From the perspective of the microgrid communities should make them lower; these findings are
user, this is only a demand-side management strategy; summarized in Table 2. Solar PV panels, pure sine wave
microgrid users did not respond to a price change (the inverters, charge controllers, and general materials (e.g.,
microgrid operator did), so, from the utility’s perspective, conductors, boxes, connectors, tubing, and screws) cost
the client (the microgrid as a whole) responds to a price approximately US$0.71/W, US$0.18/W, US$0.07/W, and
change during peak hours, thereby demanding less energy. US$0.45/W, respectively. The total cost of the materials and
Thus, from a utility perspective, the microgrid is a DR user. equipment (not including batteries) is US$1.41/W.
This same line of thought can be implemented for other Table 2 also shows a range of costs for design, installa-
types of electric rates. This work motivated UPRM research- tion, and certification. These figures come from a set of
ers to continue advocating for DERs and, eventually, actual quotes obtained over the same time period. Note
microgrids in Puerto Rico. that the design and certification cost ranges US$1–2.20/W.
The installation cost has an even wider range, US$0.40–
Cost Estimates 1.50/W. The amount does include the installation cost of
UPRM researchers studied the residential rooftop PV costs batteries. The capital cost of batteries is discussed later in
up to October 2017. The analysis focused on buying equip- this article. The resulting cost is shown in Figure 6.
ment and materials for a solar PV system in the range of The lower (green) curve in Figure 6 shows the cost of
1–5 kW. Because these are retail costs, bulk purchasing for the system were it to be installed by the owner. The mid-
dle (red) curve is the installed cost of the system, with a
low-end of US$1.40/W for design, installation, and certifi-
Table 2. The actual cost for a residential cation. The upper (blue) curve is the installed cost of the
rooftop PV system (1–5 kW). system, with a high end of US$3.70/W for design, installa-
tion, and certification. Many PV installers or companies
Cost in Dollars follow the red curve (or even lower), proving the existence
Component/Task per Watt (US$/W)
of a very competitive local PV market in Puerto Rico, yield-
PV panels 0.71 ing economic benefits that stay in the particular commu-
Inverter 0.18 nity. In contrast, large power plants burn fossil fuels,
which have an environmental impact and many fewer
Charge controller 0.07 economic benefits. Figure 6 also shows the downward
Balance of system 0.45 trend in rooftop PV costs when compared to the 2013
Subtotal (do it yourself) 1.41 costs presented in Figure 4.
For stand-alone (and, eventually, microgrid) operations,
Installation (estimate) 0.4 to 1.5
storage is needed. Lead-acid flooded batteries can be pur-
Design, permitting (estimate) 1–2.2 chased for US$100/kWh. For the system size considered in
Total 2.81 to 5.11 the analysis, a 10-kWh battery storage subsystem was
selected; these batteries, operated at 50% discharge and

LCOE, US$/kWh
(Net Metering, 20 Years, 1% Annual Degradation)
30
28.5
25
19.2
20
15.7
14
¢/kWh

15
10.6
10 7.9 7.7
5.3
5
3.9
0
2.5 3 3.5 4 4.5 5 5.5 6
Peak Sun Hours

1.41 US$/W 2.81 US$/W 5.11 US$/W

Figure 6. The PV costs as a function of location in Puerto Rico, not including the cost of batteries.

12 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

have an estimated life of three years. Because most PV
panels, inverters, and charge controllers have a warranty
of at least 20 years, we assume that in 21 years seven bat-
tery changes will be needed, resulting in a high-end cost
of US$7,000 over the life of the system. Battery cost is
expected to continue declining, and assuming a “round-
trip” efficiency of 90%, the cost of storing electric energy is
approximately US$0.18/kWh.
Thus, the total cost of electricity in this stand-alone
system ranges from lower than US$0.23 to US$0.29/kWh. At
US$75/bbl, oil electricity from the grid costs US$0.277/kWh.
Note that this last amount was calculated using PREPA’s
outdated basic tariff of US$0.05/kWh (residential), which
has not been evaluated since 1989. The actual basic tariff
should range US$0.08–0.10/kWh, creating grid parity with
a stand-alone solar PV system that includes batteries.
Furthermore, because these are purely cost estimates,
they do not include or account for sustainability or resil-
iency metrics/credits that would tilt the balance in favor
of renewable-based systems. Puerto Rico is, therefore,
ready to begin its transition to an electric infrastructure
that favors local DERs, a move that would also support
local socioeconomic activity and provide increased resil-
iency in facing future natural disasters.
and financial issues. Every community is different; there-
fore, the process of implementing microgrids can be cum-
bersome for an electric industry accustomed to technical
standards and regulatory certainty. As part of social con-
siderations, citizens must be engaged through community
training and capacity building to help facilitate the transi-
tion to a more distributed electric sector. Furthermore,
there is a need within the utility industry for workforce
development that addresses not only technical matters,
but also the broader social issues related to distributed
energy generation and microgrids.
UPRM is a core member of the Center for Grid Engi-
neering Education (GridEd), a DOE-sponsored project led
by the Electric Power Research Institute. GridEd spon-
sored two colloquia on solar communities held in April
2017, one in Ponce (Southern Puerto Rico) and another
in Bayamón (Northern Puerto Rico) (Figure 10). Roughly
Community
Center

Transition to Community Microgrids

Community microgrids could be established if only a few
houses have rooftop PV systems; if those individuals agree
to share the responsibilities and benefits, they could cre-
ate a solar community (Figure 7). As this solar community
grows, and if enough DERs are locally available, a commu-
nity microgrid could be established. The flowchart in Fig-
ure 8 illustrates one possible process.
This process has been studied at UPRM since 2009,
yielding three master’s theses and various undergraduate
capstone design projects. During the spring 2018, four under- Figure 7. A three-house solar community and the community center
graduate students further developed the community building.
microgrid presented in the feasibility
study discussed previously. They
increased the number of houses to
238, updated distribution transform- As more people in the community
ers with the actual capacity of each, Only a few houses install solar systems, the
have PV installed. distribution system limit is
and improved load profiles so that reached.
they were closer to actual communi-
ty-demand profiles. The students
also completed initial work on pro-
tection and communication issues
related to the microgrid. Figure 9
shows the updated system for the
proposed community microgrid. Solar community: Existing Community microgrid:
PV systems and aggregated more investment in
demand are coordinated to an energy storage and
Capacity Building and management system.
increase solar energy use.
Workforce Development
The social dimensions of renewable
energy and microgrids are even
more difficult than the technical Figure 8. A diagram of the movement toward a community microgrid.

IEEE Elec trific ation Magazine / D ec emb er 201 8 13

 Substation
Transformer
EPS 1.3 mi

38–4.16 kV

578 ft 640 ft

177 ft
345 ft
T2 T 20 T 14
× 16 × 16 × 20
275.25 kWh 588 ft 256 kWh
245.75 kWh 75 kVa

319 ft
50 kVa 50 kVa

393 ft
T4 T8 T 17
× 10* ×8
× 12
131.25 kWh 209.5 kWh 209.5 kWh
25 kVa
626 ft

75 kVa 75 kVa

325 ft

432 ft
T1 T 10 T 18
× 14 ×5 ×8
223.25 kWh 107 kWh 128.5 kWh
571 ft

75 kVa 50 kVa

518 ft
50 kVa

605 ft
T3 T9 T 16
× 11 × 17 × 17
171.25 kWh 245.75 kWh
588 ft

50 kVa 236 kWh

352 ft
50 kVa

819 ft
50 kVa
T 19 T 11 T 15
× 10 × 11 ×5
244 kWh 187.5 kWh 76.5 kWh
679 ft

75 kVa 75 kVa

536 ft
75 kVa
100 ft

T6 T 12 T 13
× 16 ×7 ×6

246 kWh 114.5 kWh

720 ft

50 kVa 97.25 kWh 75 kVa 75 kVa

T5
× 22
358.75 kWh
544 ft

75 kVa
T7
×8
144.75 kWh 50 kVa

Figure 9. The updated system information for community microgrid analysis, including a connection to the electric power system (EPS).

(a) (b)

Figure 10. (a) An open discussion during the first colloquim and (b) a poster presentation during the second colloquim. (Photos courtesy of
Efraín O’Neill-Carrillo.)

14 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

150 attendees participated in these community-engage- researchers have been collaborating with that community
ment and outreach events. Participants were mostly in various capacity-building activities. These include semi-
community leaders, although, for each meeting, key nars open to the community and two series of workshops
expert collaborators (between 12 and 20) were also for community high schoolers and young adults. Through
invited. The main purpose of the colloquia was to pres- that collaboration between an actual community and
ent the concept of solar communities, its potential local UPRM, the process described in Figure 8 has been imple-
socioeconomic benefits, and its technical challenges. mented, and the first PV system has been installed in the
The work was structured around written assignments community’s common building (Figure 12). It serves not
participants completed as members of groups. For the only as an electric energy source but also as an educa-
final exercise, each group was asked to envision mem- tional tool and a step toward a solar community and
bers’ ideal solar community and shared their thoughts
in a written poster
The most important question asked of participants 1% 1%
concerned the management of the solar community.
Figure 11 shows that community participants over- 7%
whelmingly supported either themselves (39%) or the 13%
community in collaboration with the utility (39%). Results
from the expert collaborators participating in Ponce and
Bayamón were also interesting; in Ponce, most of the col-
39%
laborators favored a shared management between the
community and utility (60%), while, in Bayamón, 43% of
the collaborators favored the community as the sole
39%
manager, and 29% favored a collaboration between com-
munity and developer. These results might reflect the
regional nature of the collaborators’ experiences. One
thing is certain: there is strong agreement on the central
role the community must play in the establishment of a
solar community. These findings are more important now Utility Private Developer
than ever before, given the slow recovery of electric ser- Community Community and Utility
vice in many rural, isolated communities following Hurri- Community and Community, Utility and
cane María in 2017. Developer Academia, Developer
The feasibility study described earlier was based on
information obtained by undergraduate students in a Figure 11. The community leaders’ preferences regarding ownership
capstone course during spring 2015. Since then, UPRM of solar-community systems.

(a) (b)

Figure 12. The beginning of a community microgrid: (a) the rooftop solar panels and (b) the rest of the rooftop PV system. (Photos courtesy of
Efraín O’Neill-Carrillo.)

IEEE Elec trific ation Magazine / D ec emb er 201 8 15


microgrid. In May 2018, the commu-
nity secured funds to acquire more
rooftop PV systems that will be man-
aged as a solar community. Support
from UPRM researchers was a vital
component in the community’s sus-
tainability journey.
As part of a university-based work­­
force development program, Dr. Fabio
Andrade and his team developed a
microgrid test bed at UPRM to teach
students and communities about
the challenges and opportunities of
microgrids (Figure 13). The main
components of the test bed are a hard­­
ware-in-the-loop interface, four 2.2-kW
inverters, and controllable loads. Some
of the experiments/demonstrations
performed include droop control and
virtual impedance, control of islanded
microgrids, and DR. The test bed is a
practical tool for studying the is­­sues
related to micro­­grid im­­plementation in
Puerto Rico.
Figure 13. A microgrid test bed at the UPRM. (Photo courtesy of
Dr. Fabio Andrade.)
(a)
Figure 14. (a)–(c) The Oasis of Light relief effort after Hurricane María. [Photo (a) courtesy of Jonathan Castillo, photo (b) courtesy of Efraín
O’Neill-Carrillo, and photo (c) courtesy of Rafael Rosario.]
16 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018
Lessons Learned from
As part of a Hurricane María
university-based On 20 September 2017, Hurricane
María, the most devastating hurricane
workforce in the territory’s modern history, hit
development Puerto Rico and forever changed the
lives of more than 3 million residents.
program, Dr. Fabio The hurricane proved that the central-
ized electric power model is insuffi-
Andrade and his cient for facing natural disasters in the
team developed a Caribbean. The system failed for sever-
al reasons, e.g., a centralized electric
microgrid test bed system, years of neglect in grid main-
tenance, and the sheer force from a
at UPRM to teach hurricane that missed Category 5 sta-
students and tus by one mile. The local prepared-
ness and response to disasters must
communities about change. The terrible aftermath of Maria
presents opportunities to further DERs,
the challenges and solar communities, and microgrids in
opportunities of Puerto Rico, especially in rural and re­­
mote areas that had no power for
microgrids. months after the storm.
UPRM researchers and students,
with support from the IEEE (in partic-
ular the IEEE Engineering Projects in Community Service
program) and projects such as GridEd, designed, tested,
and deployed as part of an “Oasis of Light” effort at four
locations in Puerto Rico that had no electricity after Hurri-
cane María hit. Each oasis had roughly 1 kW of solar pan-
els and approximately 4 kWh of storage (Figure 14),
providing each community with enough power to charge
small electronic devices and a small refrigerator for stor-
ing medicine. In addition to providing some relief, each
oasis also served as an opportunity to educate Puerto
Rico’s residents on renewable energy, DERs, and microgrids.
Basic information about energy use and renewables was
displayed at each oasis.
The new reality of Puerto Rico’s electric sector is one
based on diminished electric demand, an increased
(b) (c)
emphasis on efficiency, and a strong These findings are
interest in using renewable energy.
The existing model, based on large more important now
fossil-fuel power plants and passive
users, is insufficient for addressing
than ever before,
the challenges facing Puerto Rico’s given the slow
electric infrastructure. The Oasis ini-
tiative showed that there is a better, recovery of electric
more resilient alternative to central-
ized power systems, i.e., a hybrid sys-
service in many
tem with DERs (which would yield a rural, isolated
more sustainable and resilient elec-
tric infrastructure) and, even­­tually, communities
microgrids. As of August 2018, the
continuous efforts of advocates for
following Hurricane
DERs yielded the first two communi- María in 2017.
ty microgrids and the first microgrid
regulation published by the Puerto
Rico Energy Commission (the local regulator).
Conclusions
The road ahead will be difficult, and all of the stakehold-
ers should recognize that renewable energy sources are
different from conventional electric power. If renewable
energy use is to be maximized, the grid needs to change.
This change must encompass not only integrating
renewables (which assumes that the grid remains
essentially the same) but also a new grid. This grid
would provide the flexibility needed to use renewable
energy as the first option, combined with aggressive
conservation and efficiency initiatives as well as active
participation from users (e.g., through DR programs).
Thus, the business model of any organization that
wants to be involved in electric energy in Puerto Rico
must be expanded from merely selling electricity to
“providing energy services” in support of increased use
of renewable energy.
It took the destruction of half the transmission system
and most of the main distribution lines to reach consen-
sus that Puerto Rico needs a significant transformation of
its electric infrastructure. This should be a lesson to the
utility industry, in general, to not wait for extreme circum-
stances (e.g., natural disasters or financial distress) to eval-
uate resiliency options, embrace more aggressive customer
engagement, and pursue alternative ways to plan, design,
build, and operate the electric infrastructure.
The industry must look for a more holistic approach to
planning, with an understanding that the challenges go
beyond technical and economic matters and are usually
context dependent. If available, distributed energy options
would bring about not only local socioeconomic develop-
ment but also a new “supporting” role for the central, large
power plants and a new “service provider” business model
for utilities. These changes will require new procedures for
training the workforce and also new
levels of user engagement. Partner-
ships among academia, industry, and
communities are essential in this new
era and would further local initiatives
such as solar communities and com-
munity microgrids in support of
increased sustainability and resilience.
For Further Reading
A. A. Irizarry, J. A. Colucci, and E. O’Neill-
Carrillo. (2009). Achievable renewable
energy targets for Puerto Rico’s renew-
able energy portfolio standard. Energy
Affairs Administration. Puerto Rico.
[Online]. Available: https://bibliotecale
galambiental.files.wordpress.com/2013/12/
achievable-renewable-energy-targets-
fo-p-r.pdf
E. O’Neill-Carrillo, A. Figueroa, and A.
Irizarry. (2013). Improved permitting and interconnection pro-
cesses for rooftop PV systems. Energy Affairs Administra-
tion. Puerto Rico. [Online]. Available: http://prsolar.ece
.uprm.edu/docs/Rooftop-PV-Systems-Book.pdf
A. Skibell. (2016, May 2). How a stubborn utility and aging
grid added to island’s woes. E&E News. [Online]. Available:
http://www.eenews.net/stories/1060036577
E. O’Neill-Carrillo, R. Santiago, Z. Méndez, H. Vega, J.
Mussa, and J. Rentas, “Capstone design projects as foundation
for a solar community,” in Proc. 47th ASEE/IEEE Frontiers in Edu-
cation Conf., 2017, pp. 1–9.
G. A. Carrión, R. A. Cintrón, M. A. Rodríguez, W. E. Sanabria,
R. Reyes, E. O’Neill-Carrillo “Community microgrids to in­­
crease local resiliency,” in Proc. IEEE Int. Symp. Technology and
Society (ISTAS), Nov. 2018.
R. Reyes and E. O’Neill-Carrillo, “Optimal use of distribut-
ed resources to control energy variances in microgrids,” in
Proc. 7th World Conf. Photovoltaic Energy Conversion, June 2018.
[Online]. Available: https://peer.asee.org/27008.
E. O’Neill-Carrillo, A. A. Irizarry-Rivera, C. Ortiz, M. Pérez-
Lugo. “The role of engineers as policy entrepreneurs toward
energy transformations,” in Proc. 123rd ASEE Annu. Conf., 2016.
doi: 10.18260/p.27008.
E. O’Neill-Carrillo, “Energy policies in Puerto Rico and
their impact on the likelihood of a resilient and sustainable
electric power infrastructure,” J. Center Puerto Rican Studies, vol.
30, no. 3, Nov. 2018.
M. Galluci, “Rebuilding Puerto Rico’s power grid,” IEEE
Spectr., vol. 55, no. 5, pp. 30–38, 2018.
Biographies
Efraín O’Neill-Carrillo (efrain.oneill@upr.edu) is with the
University of Puerto Rico-Mayaguez.
Isaac Jordán (isaac.jordan@upr.edu) is with the Univer-
sity of Puerto Rico-Mayaguez.
Agustín Irizarry-Rivera (agustin@ece.uprm.edu) is
with the University of Puerto Rico-Mayaguez.
Rafael Cintrón (rafael.cintron@upr.edu) is with the Uni-
versity of Puerto Rico-Mayaguez.

IEEE Elec trific ation Magazine / D ec emb er 201 8 17
 By Alexander Anderson
and Siddharth Suryanarayanan

lectrification of remote communities worldwide represents a key

e
­necessity for sustainable development and advancement of the 17 United Nations (UN)
Sustainable Development Goals (SDGs). Additionally, it is a prerequisite to the creation
of numerous other infrastructure and economic systems, including agriculture, health
care, education, clean water, sanitation, transportation, and telecommunications. With
over 1 billion people still lacking access to electricity, finding new methods to provide safe, clean, reli-
able, and affordable energy to off-grid communities deserves to be a dynamic area of research. This is
the reason that so many papers have been written on the design, optimization, and construction of
electrification microgrids.
However, traditional approaches to power-system design have focused on cost and reliability cri-
teria and do not provide a sufficiently broad view of the profound impact of electrification. The

Looking at the bigger picture through life-cycle analysis

of community microgrids: A case study in Papua New Guinea.

An Enterprise
Systems
Engineering
Approach to
Electrification
Digital Object Identifier 10.1109/MELE.2018.2871240
Date of publication: 16 November 2018

18 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018 2325-5987/18©2018IEEE

installation of a single microgrid is a life-changing experi-
ence for thousands of people, including the residents who
receive direct electricity service and many others who
benefit from better education, new economic opportuni-
ties, incidental job creation, and other critical infrastructure
systems enabled by electricity. Numerous socioeconomic
factors, which span the engineering design process in
terms of both scale and scope, determine whether the
power system will be able to provide these benefits and
operate sustainably.
For an electrification program to succeed, the project
team must work with the community to satisfy people’s
needs directly, be sensitive to local environmental con-
straints, mitigate possible risks, and have a plan in place for
at least ten years of sustainable operations and mainte-
nance. These considerations extend beyond the technical
lightbulb—©istockphoto.com/domdeen,
backround—footage firm, inc.
and optimization problems typically addressed in
microgrid design. In the highlands of Papua New Guinea
(PNG), for example, the designers of an IEEE Smart Village
(ISV) regional electrification program are demonstrating
how these various requirements can be addressed through
a systems engineering life-cycle approach.
Papua New Guinea Case Study
Located just 300 km north of Australia, PNG is notable for
its astonishing number of languages—no fewer than
840—among its population of 8.1 million living in several
thousand separate communities. Despite strong growth
in the oil, mining, and agricultural commodity sectors,
82% of the population survives on subsistence agriculture
in remote villages. The widespread lack of critical infra-
structure has stymied growth throughout the country:
only 41% of the population has access to proper sanita-
tion, 31% to clean water, 10% to electricity, and 3% to
Internet connectivity.
PNG Power Ltd., the national electricity provider, is
expanding the power grid by installing small hydroelectric
and thermal power plants, but progress has been limited
by the country’s rugged terrain and lack of supporting
infrastructure. Three islanded networks and 19 diesel
microgrids account for the country’s total generation
capacity of 580 MW, composed of 300 MW controlled by
PNG Power and 280 MW owned by independent power
producers. On the mainland, PNG is served by the Ramu
System in the highlands and the Port Moresby System in
the capital city, as depicted in Figure 1. Despite plans to
expand the Ramu System by increasing the capacity of
the Ramu hydro plant from 75 to 273 MW and upgrading a
few smaller units, it is unlikely that the distribution sys-
tem will extend farther than several kilometers from the
Highlands Highway connecting Mount Hagen to Lae. Ser-
vice to the few customers connected to the grid is highly
unreliable, as illustrated in Figure 2, with some customer
outages lasting weeks. For this reason, islanded renew-
ables-based electrification microgrids remain the most
viable method for providing safe, affordable, and reliable
electricity to all the residents of PNG.
About 30 km east of Mt. Hagen is the Madan community,
which is the center of a capacity-building program by PNG
Community Transformation Centre (CTC), Inc. [a local non-
governmental organization (NGO)] and Transform Interna-
tional to provide access to electricity, intranet connections,
education, safe water, proper sanitation, and community
empowerment programs. Within the initial 10-km project
radius, there are seven tribes with an estimated 44,000 peo-
ple and 35 schools with an average of 275 students each.
Most schools lack critical resources, including safe drinking
water, electricity, sanitation, desks, books, teaching supplies,
and access to the government curriculum.
IEEE Elec trific ation Magazine / d ec emb er 201 8 19
 The Madan CTC p rog ra m is
based on 15 years of capacity build-
ing focused initially on the critical
health-care needs of PNG, which
has the highest infant and child
mortality rate in the Pacific. To ad­­
Ramu System:
dress these needs, in 2007, retired
Nameplate Capacity: 151 MW
Available Capacity: 70 MW surgeon Larry Hull and the commu-
Customers: 26,000 nity built the Madan Medical Clinic
Wabag and Birthing Center, which is entire-
Madang ly off-grid. Each year, the center and
Kudjip Gusap
its staff accommodate over 10,000
Sing Sing
Mount Hagen patient visits, administer 5,000 vac-
Mendi Goroka cinations, and deliver hundreds
Kainantu
of births. In 2013, the community
Erap
Lae launched the current sustainable
Papua New Guinea development program, which has
Population: 8.1 Million made great progress in creating the
community capacity to build, install,
maintain, operate, and expand in­­
frastructure systems. A series of
Rotary Global grants has supported
the construction of a set of rainwa-
Port Moresby System: ter harvesting and distribution sys-
Nameplate Capacity: 170 MW
tems that provide a million liters of
Available Capacity: 102 MW
Customers: 43,000 Port Moresby clean water a year and safe sani­
tation to 5,000 people in the com-
munity (Figure 3). The Madan CTC
program has also provided over
175,000 textbooks to local schools
Substation with Thermal Generation
Substation with Hydro Generation and established three new commu-
Transmission Substation nity centers that function as hubs
Queensland 132-kV Transmission Line for ongoing adult literacy, women’s
Australia 66-kV Transmission Line empowerment, and vocational train-
22-kV Transmission Line ing programs. In collaboration with
Figure 1. The Port Moresby and Ramu power systems, which provide service to the capital and ISV, the program will also provide
highlands. Not shown is the 42-MW system powering the island of New Britain. electricity, community intranet

System Average Interruption Frequency Index: System Average Interruption Duration Index:
265 Interruptions/Customer/Year 13,490 min/Customer/Year

Generation Other
Other 51% Distribution 12% Generation
3% 19% 33%

Distribution
16%

Generation or
Generation or Transmission
Transmission Transmission
Transmission 6%
11% 30%
19%

Figure 2. Ramu System reliability statistics for Lae, a seaport on the east coast of PNG. Roughly 60% of generation-related interruptions are
caused by system underfrequency events. The reliability figures for customers in the highlands are even lower. (Data courtesy of the Japan
­International Cooperation Agency Study Team.)

20 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

capability, digital education, and entrepreneurial opportu-
nities, with regional expansion planned to reach all 700,000
residents of the provinces of Jiwaka and Western Highlands.
ISV Approach to Sustainability and Scalability
ISV is the IEEE’s member-led, not-for-profit, humanitarian
outreach program and one of the IEEE Foundation’s four
priority initiatives. ISV enables community entrepreneurs
to empower their communities through capacity-build-
ing projects focused on the three pillars of energy, educa-
tion, and enterprise. With a focus on field implementation
of the SDGs, ISV has created a network of experienced
entrepreneurs and village leaders engaged in community
microutility infrastructure projects, humanitarian technol-
ogy transfer, community-based education, and holistically
sustainable enterprise. ISV aims to empower 50 million
people by 2025, with plans for expansion in Africa, India,
South America, and Asia.
ISV, through bottom-up and top-down participatory
development strategies, has been acting as the catalyst for
socioeconomic and technological change with eight years
of successful projects. Each new project builds upon an
on-the-ground network of entrepreneurs and village lead-
ers who have previously led community infrastructure
projects addressing other SDGs, including clean water,
proper sanitation, affordable health care, quality educa-
tion, sustainable agriculture, and gender equality. For each
community, lack of access to safe, clean, secure, and
affordable sources of electricity has limited economic
growth and stalled efforts to eradicate poverty.
The ISV development strategy is based on establishing
long-term collaboration with in-country community entre-
preneurs who, in turn, are expected to build relationships
at both the local and national level with financial, social,
and governmental institutions. This approach eliminates
conflicts of interest with national utilities, raises awareness
and support for ISV initiatives, enables creation of comple-
mentary infrastructure, and pro-
motes opportunities for increasing
the prosperity of villages surviving
on fewer than US$2 per day of per
capita income. Open sharing and
standardization of technology, busi-
ness models, education, vocational
training, community-wide enter-
prise, and ap­­proaches to funds lever-
aging have allowed the ISV model
to be implemented worldwide, reach-
ing over 100,000 people in 2017 with
a 50% annual growth rate.
Achieving Sustainability
Success of electrification projects
extends far beyond the design, in­­
Figure 3. Community members gather to celebrate the construction of a new water and sanita-
stallation, and commissioning of a tion facility at Papen Elementary School in the Madan community. [Photo courtesy of PNG CTC,
microgrid. To provide real benefits Inc. (Aarlie Hull).]
to a community, the power system must be complement-
ed by an equally complex array of social infrastructure
and community-based organizations so that the system
and its components are properly maintained, operated,
and eventually replaced. Prior to the start of the design
process, a network of local stakeholders, customer groups,
financial mechanisms, and community organizations
must be established. During the installation and commis-
sioning phase, the community must be an equal partici-
pant so that core knowledge is transferred to full-time
staff through train-the-trainer processes.
In creating a holistically sustainable electrification pro-
gram, only 20% of a project is installing technology; the
remaining 80% is building trust-based community rela-
tionships, providing inclusive education, and developing
holistically sustainable social businesses. Due to the com-
plexity of the process required to create a truly sustainable
project, advanced design and project management tools
are needed. Those can be found in the discipline of sys-
tems engineering.
Systems Engineering
Systems engineering is a broad discipline that not only
includes elements of electrical, mechanical, and civil engi-
neering but also integrates technical design with the
social, management, human, regulatory, and business
domains. It provides the holistic perspective needed to
guide analysis, design, testing, integration, and deploy-
ment of complex systems formed from numerous interre-
lated components working together to achieve a common
goal—arguably a description of any power system.
With systems engineering, traditional project manage-
ment’s focus on defining success in terms of scope, sched-
ule, and budget is expanded to include other factors such
as performance, cost, customer satisfaction, stakeholder
requirements, business opportunities, and individual tech-
nical attributes. These must be balanced through tradeoffs
IEEE Elec trific ation Magazine / d ec emb er 201 8 21
 to achieve the best results. At the same time, systems engi-
neering extends the engineering design process to include
client needs, use cases, operational scenarios, technologi-
cal maturity, risk analysis, functional requirements, perfor-
mance specifications, subsystem interfaces, production,
deployment, operations, maintenance, and disposal. This
sequence of analyses and considerations leads to the
development of a systems engineering life cycle, of which
the project management life cycle is a subset.
A number of standards have been developed to model
the activities involved in systems engineering, including
MIL-499B, IEEE-1220, EIA-632, and ISO-15288. The first
three focus on individual processes and systems analysis/
control and so are of less relevance to the current discus-
sion. However, ISO-15288 integrates technical processes
with a set of enterprise-management and project-plan-
ning tasks. Although these elements are not presented in
sequential order, they form a foundation for creating a life
cycle for electrification systems.
Community System-of-Systems Enterprise
Although a microgrid typically represents a single system,
it actually forms part of a system-of-systems (SoS) hierar-
chy spanning social, economic, and technical levels, not
unlike Maslow’s hierarchy of needs. Arguably, the final
mission of every electrification project is to improve the
quality of life for the residents of the community, as repre-
sented by community enterprise. This abstraction inte-
grates all the elements that contribute to eradicating
poverty in a community and can be expressed through
the 17 UN SDGs, as depicted in Figure 4.
More specifically, community enterprise represents the
organizations, entrepreneurs, processes, systems, technol-
ogies, stakeholders, community members, and other re­­
sources that contribute to the holistically sustainable
Community
Enterprise
and Empowerment
SoS
Providing Infrastructure,
Education, and
Entrepreneurial Opportunities
Multiple Businesses and Projects
Working Together in a Single Operational
Area and SDG Advancement Mission
Individual Projects, Businesses, and
Infrastructure Systems
Physical Electrical, Mechanical, Hydraulic,
and Communications Components
Operational, Market, and Physical Environments
Figure 4. A hierarchy of elements forming the community infrastructure SoS including a map showing the UN SDGs corresponding to each sys-
tem level (i.e., system, subsystem, and component).
22 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018
development of the community. The scale of community
enterprise extends far beyond electricity service. The cre-
ation of reliable access to electricity enables information
communications technology (ICT) and other critical infra-
structure that will transform the previously remote com-
munity into a regional hub that attracts new businesses,
education programs, and further investment. The goal of
community empowerment and capacity building drives
the enterprise development and management strategy to
include other infrastructure, services, business opportuni-
ties, training, ICT investments, and operations.
A community infrastructure program is composed of a
series of partially interdependent projects and systems
that form an SoS and cooperate to support and promote
sustainable economic growth and social empowerment
throughout the community. The second layer of the enter-
prise systems hierarchy includes the interfaces and col-
laboration among a microgrid system and other systems,
such as those related to water, sanitation, electrical, infor-
mation technology, and wireless telecommunications. The
lower layers of the hierarchy are made up of multiple busi-
nesses that may provide the same service (such as multi-
ple islanded microgrids or service kiosks), individual
systems, and their respective components.
The SoS structure for a community electrification and
infrastructure can be explained in terms of the seven char-
acteristics of SoS originally outlined by Sage and Cuppan.
xx Operational independence: An electrification microgrid
forms the foundation for other infrastructure systems
and community services that are independent and
individually useful.
xx Managerial independence: The microgrid and other
community systems typically are operated and man-
aged independently. The entities responsible for the
design, construction, operation, and maintenance of
 various systems (including those for electricity, water,
or communications technologies) may be unrelated
but contribute toward the overall mission of the SoS.
xx Geographic distribution: The systems typically are dispersed
over a wide region of a community. Through the advance-
ment of communications access and Internet of Things
humanitarian technologies, the exchange of information
among systems is becoming increasingly common.
xx Emergent behavior: The overall goal of the SoS (to enable
sustainable development and capacity building) is not
related to the direct functions of an individual microgrid
or other system.
xx Evolutionary development: The community SoS evolves
continuously, with each new system not only
changing the structure, mission, and role of the SoS
but also enabling the addition and modification of
other systems.
xx Self-organization: The SoS represents a dynamic orga-
nization structure that responds in an agile manner
to its environment and the objectives of the commu-
nity program.
xx Adaptation: The entire community enterprise (at the
pinnacle of the SoS) is itself responsive to changing
community needs, technological acceptance, success
of current projects, stakeholder inputs, and an array of
socioeconomic factors.
In the case of the Madan program, the highest level of
community-based infrastructure enterprise is organized
and guided by the Madan CTC management authority as
part of PNG CTC, Inc. (the local nonprofit responsible for
construction, maintenance, and managerial oversight of
the program) and small entrepreneurial businesses sup-
ported by the new infrastructure. The Madan CTC is addi-
tionally responsible for providing community members,
sponsors, and other stakeholders with a forum to estab-
lish the continuous feedback loop necessary to keep the
community enterprise SoS agile and adaptive.
The community center and maintenance monitoring staff
operate at the second level, ensuring proper coordination of
operations and maintenance of the systems. At the third level
are the individual systems and their operators who run small
entrepreneurial businesses providing safe drinking water,
offering charging services for cell phones, operating portable
battery kits, collecting waste, and distributing tablets and
computers to community workshop participants at schools
and community centers. Finally, at the lowest two levels are
the individual solar-power, digital classroom, community
intranet, water, and sanitation subsystems and the individual
components of each system, such as solar panels, batteries,
computers, and wireless routers.
The Engineering Life Cycle
of Engineering Systems
Several life-cycle models have been created for the devel-
opment of defense technologies, software, and other com-
plex projects, including the waterfall, spiral, agile, rapid
prototyping, and incremental models. Although models
developed for other applications share many of the same
phases as those of an electrification program, most mod-
els first emphasize the development of specifications and
then integration and testing, because these phases are
essential for the successful deployment and production of
new technologies and software.
However, these phases are of less importance for the suc-
cess of community electrification programs. Few microgrid
projects implement unproven, cutting-edge technologies (as
are required for defense systems) but rather focus on the
design and operation of robust, durable, and reliable off-the-
shelf generation, energy storage, and controller components
that have already been integrated and tested by the manu-
facturer. Likewise, the significant differences between
va­rious custom software systems are not included in the
functional and performance requirements between one
microgrid and another. The complexity of community elec-
trification lies in attaining the microgrid’s financial, techni-
cal, and operational sustainability so that it can successfully
reach its 20-year design lifespan.
To address the need for a simple description of the pro-
cess to establish the required technical, financial, and or­­
ganizational support subsystems, we propose a new
electrification systems engineering life cycle. This model
targets project planners, pro bono consultants, and in-
country engineers seeking to collaborate with a series of
communities to establish a community-based infrastruc-
ture program capable of regional scaling. The proposed life
cycle follows a spiral pattern, as shown in Figure 5, due to
the cyclical nature of building successive infrastructure pro-
grams within a single community and expanding these to
additional communities. Each phase of the life cycle will be
discussed and illustrated using the Madan community
empowerment program.
Needs Analysis and Concept Development
The needs-analysis process is possibly the most impor-
tant step for understanding the required project scope,
establishing a relationship with the community based on
mutual trust, and creating the foundation for a successful
regional program. This phase comprises far more than a
simple survey of how many people in a village lack elec-
tricity, what they are willing to pay, and how much kerosene
they burn. It is a comprehensive process that includes
the following:
xx establishing relations with community leaders
xx identifying potential local technical contributors
xx categorizing key stakeholders
xx recognizing factors that have limited electricity access
xx researching publications and records of prior project
successes and failures
xx evaluating availability of supporting ICT
xx assessing access to education and vocational training
xx gauging the familiarity of the community with mi­­
crogrid technologies
IEEE Elec trific ation Magazine / d ec emb er 201 8 23
 Community-Based
Validated Management
Community Organization
Development
Model
In-Field Preliminary System
Demonstration Services, Siting,
Electrification, Dep
Evaluation loy Sizing
Education, and eri
ng Plan ment
ine gn ning
Entrepreneurial g
En esi Pilot
Components D
Project Project
Management

Op ecis sis
Results

tim ion
Plan Functional and

D aly
An
ati nd

iza
Microgrid and Performance
rific n a

tio
on

ICT Design Specifications

Ve ratio

n
Specifications Community
eg

Needs Defined
Int

Infrastructure and
Capacity
Business Plans
Building
Fully Integrated Production and Operational and Concept
Hardware and Deployment Electrification
Performance

io y
Enterprise

at it
Organization Specifications

lid un
Requirements

n
Focus Area

Va mm
Structure
Pro eplo

o
C
Need
D
du yme

Analy
ctio nt

sis Concept
Development
na

Operations and
Maintenance
nd

Documentation Candidate
System Enterprise System
Remaining Capabilities Concepts
Community
Commissioned Op Needs
and eratio
Operational Su ns
Systems ppo
rt
Repeat Cycle with New Community
or New Infrastructure System

Community Enterprise
SoS

Figure 5. A spiral of community-based infrastructure: this system life-cycle model has been developed to address the challenges and consider-
ations involved in electrification projects.

Figure 6. Madan community members, who had never seen a tablet

computer before, gather around a project volunteer conducting a site Figure 7. A typical household in the PNG highlands. [Photo courtesy
survey. [Photo courtesy of PNG CTC, Inc. (Aarlie Hull).] of PNG CTC, Inc. (Maureen Yalde).]

appraising past records of maintenance, operation,

xx mapping existing assets for infrastructure and shared
xx
and materials availability community services
compiling community income-generation sources
xx selecting candidate sites for preliminary in-field dem-
xx
and local access to financing onstration systems.
estimating new jobs and entrepreneurial opportuni-
xx A detailed assessment covering the technical, social,
ties that will be created by electrification geographical, financial, cultural, and organizational aspects

24 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

 Electric Service
ICT Service
Microentrepreneurs Customers Environment
Energy Input
Weather
Education
Wear/Damage
Vocational Training Payment Field Support
Job Placement

Microutility Electrification Enterprise System

Delivered Equipment
Warranty Service
Microgrid
Training Operations Equipment

Suppliers

Program Engineering
Finance Management Design Specs
Manufacturing Specs

Loan Repayments
Project Proposals Needs Feedback Legal Requirements Design
Grants
Status Reports Validation Project Authorization Advice
Investments

Funding Agencies Community Leaders Regulators Pro Bono Consultants

Figure 8. A context diagram depicting some of the electrification enterprise system’s interfaces with its internal subsystems and external
stakeholders.

of the community enables the development of a project

baseline justifying the need in both qualitative and quan- Operational Requirements: Why
titative terms (Figures 6 and 7). Concluding the needs-
analysis phase is an operations analysis that defines the Functional/Performance
general approach, estimates the value delivered to the Requirements: What and How Much
community, and lists operational objectives for the planned
electrification program.
The subsequent concept-development phase focuses
on converting the identified community needs into a set Electrification
of candidate enterprise system concepts. The first process Enterprise
is functional analysis, which identifies those new prod- SoS
ucts, services, or features that would be delivered to the Operational Operational
community by the electrification plan being considered. Concept: Scenarios:
The goal is to translate the operational objectives previ- Who and How Where and When
ously identified into a list of functions assigned to subsys-
tems. The analysis also identifies operational interfaces Figure 9. The triumvirate of conceptual design is a useful tool for
defining the concept of operations and system requirements as
both between internal subsystems and with external answers to seven simple interrogatives: why, what, how much, who,
stakeholders. These interfaces can be summarized in a how, where, and when.
context diagram, depicted in Figure 8.
A complete functional analysis enables the identifica- much, who, how, where, and when—and form the overall
tion of the operational, functional, performance, and phys- electrification enterprise system concept. It is important
ical requirements for the system, leading to the creation of that community leaders and key stakeholders be closely
a detailed concept of operations. A useful systems-engi- involved in this process to ensure that the resulting con-
neering tool for this process is the triumvirate of concep- cept of operations is relevant to community needs, cultur-
tual design, illustrated in Figure 9. The operational context, ally acceptable, and actually feasible.
operational scenarios, and system requirements are In the case of the Madan project, the needs-analysis pro-
answers to seven simple interrogatives—why, what, how cess (conducted in 2015) resulted in three parallel operational

IEEE Elec trific ation Magazine / d ec emb er 201 8 25

 Table 1. The concept development results for electrification of the Madan community
center, one of four parallel operational concepts established for the program.
Functional requirements Scalable, robust systems to provide reliable electricity for
Light-emitting diode (LED) lighting
■ 

Digital classroom technologies (computers, projectors, cameras, digital content servers,

■ 

Android tablets, printers, Wi-Fi routers)

Paid cell-phone charging
■ 

Charging of portable battery kits for home electrification

■ 

Operational requirements Sufficient lighting for operations after dusk and during periods of heavy rain
■ 

Delivery of power with sufficient quality and capacity for information technology equipment
■ 

and electronics for digital classrooms

Ability to store sufficient energy to provide at least two days of power during monsoons
■ 

Sufficient capacity for cell-phone charging

■ 

Ability to expand the system to allow charging of portable battery kits

■ 

Ability to power both ac and dc loads

■ 

Performance requirements Ability to power at least ten LED light bulbs per room with minimum luminosity
■ 

of 300 lm each
Ability to power a computer, projector, Wi-Fi modem, and intranet receiver with a maximum
■ 

power consumption of 50 W each

Ability to operate the system at maximum power for at least 12 h/day
■ 

Ability to charge at least 30 Android tablets with a maximum power use of 10 W for
■ 

use 3 h/day
Ability to charge at least 100 cell phones per day with a cell-phone battery capacity of 2 Ah
■ 

each
Ability to charge up to five laptops per day with a battery capacity of 50 Ah each
■ 

Portable Battery Kit Charging for

Rural Home Electrification

PBK PBK
=
=
PBK PBK

Power for Digital Classroom to Support

Community Education and Training

~
Smart
Smart
= Meter

Power for Cell Phone

Charging at Payment Kiosk
=
=

Figure 10. The concept of operations for electric service at the three community centers. PBK: portable battery kit.

26 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

Figure 11. Madan microutility managers with engineering volunteers
from the United States and Australia drafting system requirements.
[Photo courtesy of PNG CTC, Inc. (Aarlie Hull).]
concepts for the first set of electrification systems designed to
serve as technology-demonstration platforms and provide
maximum initial social impact. The operational concepts
focused on the electrification of an existing shared communi-
ty infrastructure, beginning with the following:
1) formation of digital learning and empowerment hubs
at the three Rotary community centers
2) creation of electricity and education access for the 35
schools within the project radius
3) electrification of the Madan Medical Clinic for vaccine
refrigeration and power of critical medical equipment
4) improvement of electric supply reliability at the com-
munity coffee mill, which provides fair employment
to 800 workers and was certified as the world’s fourth
greenest coffee operation by Rainforest Alliance.
The operational concept created for the Madan com-
munity consisted of a community-based organization fol-
lowing the proven enterprise model of a rural electric
cooperative that would be responsible for stakeholder
management, customer relations, and the commission-
ing, operation, and maintenance of the electrification
infrastructure. The local microutility would also be respon-
sible for training community members as both staff and
customers. The community organization would continue
Table 2. Select optimization objectives for microgrid planning to maximize capacity
building, community empowerment, and socioeconomic impact.
Technical Economic
Reliability Fuel/operations cost
Safety Installation cost
Technical maturity Maintenance cost
Energy efficiency Revenue/profit
Renewable resource Monthly cost to customer
potential
Scalability Payback period
Equipment lifespan Financing
Figure 12. Validation review with community leaders, local teachers,
installers, and other stakeholders. [Photo courtesy of PNG CTC, Inc.
(Aarlie Hull).]
to manage the Madan water, sanitation, and education
infrastructure as separate components of its microutility
project portfolio. Program management, oversight, and
funding would be provided through Transform Interna-
tional, an NGO based in the United States and Canada.
Revenue from residential and commercial electric service,
battery-charging services, and technology delivery fees
assessed from schools receiving electrification/digital edu-
cation packages would pay for staffing, operations, mainte-
nance, upgrades, and expansion.
The operational concept and functional analysis of the
micro-utility were subsequently translated into a series of
functional, operational, and performance requirements,
which are summarized for the community center deploy-
ment strategy in Table 1. The associated concept of opera-
tions is depicted in Figure 10.
Validation, Optimization, and Planning
The next set of phases in the enterprise systems life cycle
brings the initiative from concept to an actionable plan for
deployment and operation of the electrification systems. The
third phase, community validation, evaluates and verifies
whether the proposed concept of operations and require-
ments are not only feasible but also likely to satisfy the
Environmental Social
Greenhouse gas emissions Social acceptability
Land use Ease of maintenance
Renewables utilization Ease of training
Component toxicity Jobs creation
Other environmental degradation Community services provided
Noise Energy policy
Support of other nearby Relevance to community
infrastructure needs
IEEE Elec trific ation Magazine / d ec emb er 201 8 27
 The validation process must involve
Table 3. Configurations of the initial set of microgrid
all local community stakeholders to
installations in the Madan community.
ensure that the concept will be rele-
Photovoltaic ESS vant, acceptable, and practical. Fig-
Installation Capacity Capacity Voltage Site ures  11 and 12 show a local review
Community 2 kW 5 kWh 24 Vdc meeting with stakeholders. If commu-
centers nity leaders have some ICT access (e.g.,
3G cellular service), then it is possible
to expand the validation phase into a
continuous feedback process through
e-mail and video-conferencing soft-
Medical 5 kW 10 kWh 48 Vdc; ware. Hosting virtual meetings can
clinic 220 Vac, 1z enable the involvement of a global
team of pro bono consultants, provide
answers to site-specific questions
without the need for international trav-
el by experts, and reduce the delay in
Madan mill 20 kW 50 kWh 48 Vdc; receiving feedback and project-status
and busi- 415 Vac, updates from months to merely hours.
ness center 3z (T) The next phase in the life cycle
addresses preliminary optimization of
the system and decision analysis for
the electric services to be provided as
Pilot 0.5 kW 1 kWh 12 Vdc well as initial microgrid siting and siz-
schools ing issues. From the SoS standpoint,
the broad community impact and
empowerment opportunities created
by electrification require a much
Standalone 2.2 kW 5 kWh 24 Vdc broader perspective than that offered
Sunblazer by traditional power system optimiza-
solar charg- tion approaches based on levelized
ing kiosks cost of energy and system reliability.
Fortunately, systems engineering
offers a set of tools that have been
adapted to modeling, simulation, opti-
[Site photos courtesy of PNG CTC, Inc. (Aarlie Hull); Sunblazer photo courtesy mization, and decision analysis of
Nextek Power Systems (Wayne Gutschow).] complex systems with both quantita-
tive and qualitative attributes.
The appropriate siting and sizing
Table 4. A sample risk matrix for the Madan project. of distributed energy resources in a
Impact community microgrid are required to
ensure secure, economic, and reliable
Likelihood Low Medium High operations. Unlike traditional radi-
High Monsoon, earth- Random vandalism Theft of electrical al distribution feeders, electrification
quake and ICT equipment microgrids often have a meshed net-
work topology with frequent power-
Medium Late delivery Inability to build ICT Inability of com-
support systems for flow reversals depending on variations
from suppliers munity to maintain
operations system in renewable generation and load. Poor
siting of distributed generators (DGs)
Low Misuse of technology Inability to hire/train
and energy storage systems (ESSs) can
qualified staff
result in higher network losses, un­­
satisfactory voltage profiles, and poor
needs of the community and the sustainable development generator performance due to the unique topology and
objectives of the project. This phase generates an initial renewable resource distribution of every site. Optimal DG
infrastructure and enterprise development plan, as well as a and ESS sizing is also critical to achieving a balance
fully defined community capacity-building concept. between system reliability and installation cost.

28 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

 –– Existing Grid
–– Existing 220-V Line
–– Planned Expansion

Figure 13. The EmpowerPack electrification education system, devel-

oped by the authors, provides a portable, instantly deployable system
for schools, medical clinics, or community centers with the latest ICT
and power technologies. [Photo courtesy of EmpowerPack Social Pur-
pose Corporation (Alexander Anderson).]

Optimization for microgrid planning often focuses on

solving an economic dispatch or unit commitment prob- 1 km
lem over a daily, weekly, or seasonal cycle using historical
data and predictions of renewable generation and load
Figure 15. The planned interconnection of Madan pilot demonstra-
growth. Multiple software tools, such as HOMER, have tion systems and initial distribution system expansion.
been developed to handle these planning tasks. However,
these methods do not include nonquantitative measures popular MCDM methods are based on utility theory and
of system performance that affect the sustainability and include multiattribute utility theory, the analytical hierar-
scalability of community electrification programs. chy process, and the simple multiattribute rating technique.
Multicriteria decision-making (MCDM) methods are Utility theory allows the relation between the costs and
evaluation tools used in systems engineering for tradeoff benefits (or “utility”) of a particular alternative to be
analyses and selection of system configurations using both expressed in terms of multiple objectives, criteria, and pref-
quantitative and qualitative decision criteria. The most erence rankings. The amount of preference given to a par-
ticular attribute is expressed in
terms of a utility function that var-
ies between zero and one. Utility
Solar Input functions and MCDM methods
Wall enable a wide variety of technical,
Laser Projector USB 3.0 Charger Input
­e conomic, environmental, and
On/Off
RJ45 Switches social objectives to be included in
Quick the optimization and decision-mak-
Charge 3.0 ing processes. Table 2 presents a
HDMI Thin-Film short list of criteria that can be used
Solar Panels
100–200 W in evaluating candidate micro­­grid
LED Lights
(2,000 lm) 5.5 × 2.1-mm
configuration alternatives.
Military-Grade 12-V Output Completion of the preliminary
Education Tablet Active Stylus optimization process leads to the
Up to 4-TB SSD or 15-TB HDD Waterproof Case creation (or revision) of a detailed
project-management plan that will
500-Wh Lithium Battery with guide the project through the re­­
MPPT Charge Controller
maining stages of design, produc-
tion, installation, commissioning,
Figure 14. An EmpowerPack standard solar ICT kit for a single classroom is capable of bringing and training. An effective planning
the latest digital education technologies to off-grid settings. USB: Universal Serial Bus; SSD: solid-
state drive; HDD: hard disk drive; MPPT: maximum power point tracking; HDMI: High-Definition document will include many of the
Multimedia Interface. elements of a traditional project

IEEE Elec trific ation Magazine / d ec emb er 201 8 29

 Ancillary Services TransCO

Scheduling Coordinator

Community
Independent System Multimicrogrid Operator
Operator

Power Exchange Marketer

Microgrid 1

Sunblazer GenCO
DistCO Aggregator
Operator

Residential Business
PBK PBK
Customer Customer
Operator Operator
Residential Business
Customer Customer
Customer Customer
Customer Customer
Residential Business
Customer Customer
Customer Customer

Microgrid n

Sunblazer
DistCO Aggregator GenCO
Operator

Residential Business
PBK PBK
Customer Customer
Operator Operator
Residential Business
Customer Customer
Customer Customer
Customer Customer
Residential Business
Customer Customer
Customer Customer

— Power Sold by DistCo — Power and AS Bought by DistCo — Power Sold by GenCo — Money Flow

Figure 16. The conceptual framework for a future multimicrogrid transactive market in PNG. TransCO: transmission company; DistCO: distribution
company; GenCO: generating company; AS: ancillary services; PBK: portable battery kit.

30 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

management plan, such as a project scope, schedule, bud-
get, statement of work, business case, deliverables to the
community, work breakdown structure, organization break-
down structure, risk management plan, supplier manage-
ment plan, and communications plan.
For the Madan project, a set of detailed optimization
studies using the optimization objectives listed in Table 2
were performed for each of the parallel microgrid dem-
onstration projects at the three Madan community cen-
ters, eight pilot schools, the medical clinic, and the coffee
mill. The selected configurations for each system are sum-
marized in Table 3.
Demonstration, Design, and Integration
The focus of the next three phases of the electrification
enterprise system’s life cycle is risk mitigation. Total sys-
tem failure shortly after a successful “intervention” in a
community is unfortunately a very common occurrence
worldwide. Such breakdowns can be attributed to failures
in earlier phases of the project, such as incomplete needs
analyses, lack of community engagement, and the absence
of an organizational structure to operate and maintain the
commissioned system. However, even if all the planning
and validation steps are followed, a vast array of risks still
threaten successful project deployment and long-term
sustainability. Table 4 presents a sample risk matrix with a
few of the risk events considered for the first phase of the
Madan electrification project.
One of the most effective tools for risk management in
sustainable development is the deployment of a series of
small-scale demonstration systems that enable evaluation
of candidate technologies in the actual community under
conditions impossible to simulate. Development, deploy-
ment, and evaluation of scaled prototypes of the actual
systems and critical components are essential to gauge
whether the defined system concepts and preliminary
designs will be technically feasible, cost-effective, and
socially acceptable.
Consequently, a series of demonstration systems will be
constructed in Madan prior to the design and deployment
of the full electrification program in the 10-km project radi-
us. These demonstration systems will be located at the
three community centers and test the effectiveness of por-
table battery kits, wired distribution, prepaid smart meters,
and other technologies. At the same time, the ability of
local schools to operate digital classroom technologies will
be evaluated using the innovative EmpowerPack digital
electrification education system developed by the authors
(Figures 13 and 14). Each of the EmpowerPack systems will
provide instantly deployable capabilities to equip schools,
community centers, and other community facilities with
solar power, computers, Internet connectivity, ways to
access government curricula, adult vocational training
resources, and means for local content development.
After the successful deployment and evaluation of the
demonstration systems in fall 2018, a full engineering
design will be created for interconnecting nearby demon-
stration systems, electrifying all remaining schools, and
deploying home electrification systems throughout the
community with a combination of wired distribution sys-
tems and portable battery kits. Figure 15 shows a map of
the first multimicrogrid interconnection in Madan. All of
the elements will subsequently be assimilated into the
community-based microutility enterprise system, which
will be responsible for final integration and validation of
all system functions.
Toward Regional Deployment
and Sustainable Operations
The last set of phases in the electrification enterprise life
cycle focuses on the transition of the program from a set
of pilot demonstration projects into a cohesive regional
infrastructure SoS. The process involves an even wider
range of stakeholders and considerations, including the
establishment of dedicated equipment supply chains,
staff training programs, operating procedures, mainte-
nance schedules, customer relations teams, and financial
mechanisms. As the Madan CTC progresses toward this
phase as an adaptive enterprise system, changes in even
the microutility organization structure may be necessary,
such as the transition from its current rural electricity
cooperative model to a community-based independent
system operator with transactive market mechanisms
capable of coordinating a broad network of interconnected
islanded and grid-tied microgrids. Figure 16 presents a
conceptual framework for this future model.
For Further Reading
R. Podmore, R. Larsen, H. Louie, N. Johnson, and S. Saha, “Fuel-
ing sustainability: The exponential impact of empowering
off-grid communities,” IEEE Electrific. Mag., vol. 4, no. 1, pp.
11–17, 2016.
A. Kossiakoff, W. Sweet, S. Seymour, and S. Biemerthe, Sys-
tems Engineering Principles and Practice, 2nd ed. Hoboken, NJ:
Wiley, 2011.
A. Anderson, S. Suryanarayanan, and R. Podmore, “Capac-
ity optimization of a community microgrid for rural electrifi-
cation,” in Proc. IEEE PES PowerAfrica, Accra, Ghana, 2017, pp.
423–428.
A. Sage and C. Cuppan, “On the systems engineering and
management of systems of systems and federations of sys-
tems,” Inform. Knowledge Syst. Manage., vol. 2, no. 4, pp. 324–345,
2001.
Japan International Cooperation Agency. (2016). Power
development master plan of Ramu power system. [Online].
Available: http://open_jicareport.jica.go.jp/pdf/12266508_01.pdf
Biographies
Alexander Anderson (alexander.a.anderson@ieee.org)
is chair of Partner Engagement at the IEEE Smart Village
and is with EmpowerPack Social Purpose, North Bend,
Washington.
Siddharth Suryanarayanan (Sid.Suryanarayanan@
colostate.edu) is with Colorado State University, Fort Collins.

IEEE Elec trific ation Magazine / d ec emb er 201 8 31
 By Ashok Jhunjhunwala and Prabhjot Kaur
Solar
Energy, dc
Distribution,
and
Microgrids
Ensuring quality power in rural India.
s the Leisang village in Manipur
a
received electricity in April 2018, the prime
minister of India announced that every
Indian village is now electrified. A village is
considered electrified in India when 10% of
its homes receive electricity. However, the number of vil-
lage homes that have electricity has now reached 84%,
with some 41 million households still without power. This
village-electrification program has been going on for
many years. Until a few years back, there was a large
shortage of power with power demand exceeding supply,
leading to no/a weak push to extend the grid to the village.
Over the last few years, when more coal power plants
became operational, the supply strengthened. At the same
time, the energy costs for the solar and wind power-
attained grid parity, enabling the addition of significant
renewable power. Simultaneously, the power grid was
Digital Object Identifier 10.1109/MELE.2018.2871277
Date of publication: 16 November 2018
32 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018
strengthened, and India attained a single national grid on
31 December 2013, such that power generated in surplus
areas could be transported to deficit regions. All of this
helped in surplus power generation because the demand
did not pick up much in recent years. The social obligation
to extend the electric grid to each village and then to at
least 10% of its homes no longer had a fundamental bot-
tleneck. The target-driven approach of the prime minis-
ter’s office helped to expedite the effort.
However, the power grid could not reach a significant
number of villages, especially in far-flung areas, deserts,
mountains, and difficult terrains. A rooftop solar system
to provide electricity to these off-grid homes was a viable
solution. Early efforts to put a solar light in each home
were not acceptable to many villagers as means of electri-
fication. They wanted a minimum of several lights, a fan, a
cell phone charger, and a TV point. A conventional rooftop
solar system with ac output was too lossy and expensive.
Furthermore, these systems would not connect to the grid
if and when the grid could reach the village. A solar-dc
2325-5987/18©2018IEEE

system evolved in India, which will be discussed later.
Such solar-dc systems became the basis of powering these
remote villages where the grid could not be extended.
This article begins with stories of how the solar-dc
systems were deployed in deserts and mountains to pro-
vide electric power, describes the next frontier of such
home systems, and discusses the limitations of such off-
grid systems as well as the existing grid-connected sys-
tems in rural areas in India. Village-level microgrids
appear to have some answers to these limitations. The
article concludes with a discussion on efforts to build
such microgrids and the importance of distributed ener-
gy resources (DERs).
Off-Grid Homes in Difficult Terrain
India has a few million homes for which grid connections
will be very difficult and expensive. The villages are on
mountains, in deserts, located remotely on islands, and
across rivers. The best immediate option is to provide dis-
tributed solar-dc systems.
©istockphoto.com/pulpitis
The Indian Institute of Technology, Madras (IIT Madras),
had taken up the development of a 48-V solar-dc system
catering to not only off-grid homes but also to the ones
with grid connection. Because power outages (power cuts)
are still common, varying from 30 min/day to 14 h/day in
many rural areas, many homes use a battery-backed sys-
tem known as an inverter. IIT Madras developed a solar
system called Inverterless-500, with output power supplied
in dc form rather than converting it to an ac form. The sys-
tem interfaces directly with a solar panel varying in size
between 125 and 500 watt peak [W(p)]. The grid input
(when the grid is available) is also up to 500 W, converted to
48 Vdc, and combined with solar output. The combined
output can charge a battery and provide a 48-Vdc distribu-
tion line within homes. When solar energy or the grid can-
not supply the power output required, the battery output
pitches in. The system is designed to minimize conversion
losses. When solar energy directly powers the load, losses
are approximately 3%, whereas losses rise to 10% when
solar power is first stored in the battery and later supplied
IEEE Elec trific ation Magazine / D EC e mb e r 201 8 33
 Table 1. Energy consumption for a typical rural household
with essential dc appliances and fixtures.
Average Power
Appliance Number Consumption (W) Daytime (h) Nighttime (h) Day Energy (Wh) Night Energy (Wh)
Fan 1 25 6 6 150 150
Lights 5 5 2 8 50 200
Cell phone 2 5 1 1 10 10
TV 1 25 5 5 125 125
Induction stove 1 400 1 1 400 400
Refrigerator 1 25 14 10 350 250
Mixer/churner 1 100 0.5 0 50 0
Cooler 1 50 8 4 400 200
Total 1,535 1,335

The daytime is assumed to be 14 h (5 a.m.–7 p.m.), and the night is assumed to be 10 h (7 p.m.–5 a.m.).

to the load. The grid-to-load losses

are limited to 6% and through the
battery it goes up to 15% when lithi-
um (Li)-ion battery storage is used.
Furthermore, as shown in Table 1 in
the article “Solar-dc Microgrid for
Indian Homes” (IEEE Electrification
Magazine, June 2016), the power
usage goes down significantly when
dc-powered dc appliances are used.
The net result is that a 125-W solar
panel can power a small home in­­­
cluding multiple lights, a fan, a
TV, and a cell phone charger when
used with a battery of 1.25 kWh.
The solar systems become relative-
ly inexpensive, each costing approx-
imately US$400, and includes solar
panels, the Inverterless-500, a bat-
tery, fans, lights, and a cell phone
charger. The size and weight are
small too. The systems can be car-
ried to difficult terrain and have
been used to provide power in homes
in remote areas in some of the states
of India including Assam, Manipur,
Jammu and Kashmir, Rajasthan,
and Madhya Pradesh.
The transportation and deploy-
ment of these solar-dc systems
were nevertheless a challenge. The
state of Rajasthan had desert areas
with scattered homes. Normal
Figure 1. The solar systems deployed in the villages of the 16 districts of Assam: Barpeta, Bon-
gaigon, Cachar, Dhemaji, Dhubri, Dibrugarh, Dima Hasao, Goalpara, Hailakandi, Jorhat, Kamrup, vehicles could not be taken to those
Karbi Anglong, Lakhimpur, Nalbar, Sonitpur, and Tinsukia. (Image courtesy of Bing Maps.) areas; tractors and camels were

34 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018

used in these instances. The challenge of Assam
was its villages scattered across the length and
breadth of the state, as shown in Figure 1. The ter-
rain was either hilly (the Himalayan mountain
region) or contained river banks (Brahmaputra
regions). Multiple modes of transportation were
required to reach a village. Roads are mud with no
tarring, making it difficult to travel during rains
because they become muddy and slushy. Trucks
could get stuck in the slush, slide, and fall off the
mountain. Material was moved in small trucks,
and tractors were used where the incline was very
steep. The last lap almost always involved head
loading/hand carrying of material. This part some-
times took up to four days. Manipur involved sim-
ilar challenges. Figure 2 shows the poles to mount
solar panels being carried up the mountains in
Manipur, and Figure 3 includes the transportation
of solar panels and the Inverterless-500 system. Figure 2. Workers carrying the poles to install the solar panels in off-grid village
homes in Meghalaya.
River bank terrains required multiple modes
of transportation to reach villages, including
small trucks to the river bank, buffalo carts from small Despite all the difficulties involved in taking the mate-
trucks to the ferry, and the ferry to across the river. When rial to the villages, there is a great joy when a home is
the river depth becomes very shallow or the bed dries up in electrified and fans and cell phone chargers are enabled.
the summer, material is hand-carried across the river. All of Beneficiaries are identified, and their eligibility for electric-
this involves multiple loading/unloading of materials. In ity connections along with applicant photographs and
Figure 4, the material is being carried across the river and identity proof are registered on the spot, as shown in Fig-
on buffalo carts in Assam. Multiple loading/unloading of ure 5. The great reward is changing the lives of the people
solar materials makes the work proceed very slowly. getting electricity.

Figure 3. The solar panels and inverterless system being transported to villages through the hills of the Himalayas.

IEEE Elec trific ation Magazine / D EC e mb e r 201 8 35

 (b)

(a) (c)

Figure 4. The materials being transported by river to remote villages in Assam on boats and buffalo carts: (a) being loaded onto the boat,
(b) shipped to destinations, and (c) unloaded.

The Approach Has Some Limitations is as important as the ability to provide power. The chal-
There is little doubt that this combined approach of taking lenge now will be uninterrupted electricity supply at
the grid to the villages and using decentralized solar sys- affordable tariffs.
tems, where taking the grid becomes very difficult, is Even though India has surplus power, the power distri-
an important milestone in electrifying India. But the bution companies (DISCOMs) of most states experience
approach has its limitations. The quality of power-supply losses and have a large debt. Writing off debt in the past has
not helped, as the losses quickly
build back up. Power theft contrib-
utes to a significant extent toward
the Indian DISCOMs large transmis-
sion and distribution losses. Con-
versely, the income of a large section
of the Indian population is rather
low. Even though the political com-
pulsion keeps the tariffs low, most
low-income homes consume very
little as they cannot afford higher
power bills. The problem becomes
more severe in rural areas, where
most people have low incomes.
Transmission losses are higher, the
voltage in the rural grid often falls to
an unacceptable level, and people
often default on their power bills.
Therefore, the DISCOMs limit the
power they provide on the rural grid
(a) (b) using power cuts as high as 14 h/
day. The quality of the power supply
Figure 5. (a) The delivered materials at a village home and (b) the installed system in a home. to rural areas is severely limited.

36 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018

 The poor quality and low affordability implies that
1)  rural households cannot expand the use of electricity
and 2) rural industry cannot prosper. The off-grid rooftop
deployment using solar-dc has similar dilemmas. A 125-W
solar panel may give effective energy (taking into account
all losses) of little more than 500 Wh/day. This may be ade-
quate for home lighting, a fan, cell phone charging, and TV
usage for a limited time when low-power dc appliances
and fixtures are used. Let us look at these in more detail.
Next Frontier for Rural Household
Rural households in India are looking forward to some of
the conveniences that urban India enjoys. They would
like to have a small refrigerator, a mixer, a butter churner
(in many parts of the country), and an air cooler. Al­­
though it is not on the radar of most homes, a very
important need is having an electric cooker. Today, most
low-income rural households cook using biomass com-
bustion. The smoke emitted is a se­­rious health hazard
and impacts mostly women and children. Their lives
would be very different if a pollution-free electric cook
stove was available. The cost of the appliance may not be
a serious bottleneck but the cost of quality and reliable
electricity limits the usage.
Rural Industry
Most people practice agriculture in rural areas. The pro-
ductivity is not very high, and the number of people
employed is large. No other employment is available
because there is hardly any industry in rural India. Young
men are migrating to urban areas but, the rest of the pop-
ulation continue to be highly underemployed in agricul-
ture. For rural industry to take off, the buying and selling
of goods and raw materials, transportation, and electricity
are the biggest problems. With mobile phones reaching
most rural areas, the buying and selling problem has
been eased. As roads to villages increasingly improve,
auto-rickshaws are taking care of the transportation
need. The biggest bottleneck is quality electricity. The low
voltage and power cuts make the electricity-dependent
rural industry unviable. Low affordability adds to the
problem. Rural industry depends on a diesel generator for
electricity, costing about four to five times that of grid-
based electricity.
An Alternative Approach: Solar-dc to Microgrids
To make electricity economically viable for low-income
people in rural India, both for home usage as well as rural
industry, a different approach is required. It would include
the following:
xxdevelopment of highly energy-efficient appliances
and fixtures
xxconnecting the solar-dc system to the electric grid
xxa new kind of rural microgrid enabling the sharing of
power.
Let us look at these in more details.
Highly Energy-Efficient Appliances and Fixtures
Reducing the power consumed by an appliance is essential
to help reduce the electricity costs for a rural low-income
consumer. Making the equipment dc powered is an impor-
tant step, as most appliances inherently need only dc
power. An induction motor-based ceiling fan, widely used in
India and deployed in off-grid homes, would consume
72 W of ac power at full speed. The power consumption of
the fans was reduced to merely 30 W at full speed, when a
48-Vdc-powered brushless dc motor (BLDC) motor-based fan
was used. In a similar manner, a dc-powered dc motor-based
refrigerator is being developed, so that the average power
consumed will go down substantially. An air cooler is a big
blower of air through a wet honeycomb or wood-wool
sheet into a home; it works very well in a dry and hot cli-
mate. The company Zazen has developed a dc-powered
BLDC air cooler, which consumes only 60 W at peak speed,
as opposed to 125 W for the equivalent ac air cooler. A
switched-reluctance motor-based 48-Vdc food mixer/
grinder consumes a peak power of 150 W, as opposed to
350 W for an equivalent ac mixer. A similar butter churn-
er is also being developed. Pictures of some dc appliances
used in our system package are shown in Figure 6.
An induction cooker is one of the most important dc
equipment used in rural areas. A 48-Vdc-powered, energy-
efficient dc induction stove is being developed that, even
at 500-W peak power, can do most of the cooking. If the
power is increased to 700 W, the induction stove becomes
an excellent cooking stove. Of course, the energy con-
sumed for cooking is not going to be too small. It may take
approximately 700–1,000 Wh/day to cook for four people
in a rural home. Also, the cooking is not limited to daytime
when the sun is present. Therefore, at least a part of this
energy may have to be stored. Another type of equipment
that would be highly relevant to rural India is a water
pump. There are several BLDC motor pumps today. But
even the induction motor pumps with a variable frequen-
cy drive yield excellent result.
dc Fan
dc Tubelight
14 in
dc Induction Stove
dc Air Cooler
dc Refrigerator
dc Bulb dc TV
Figure 6. The dc appliances.
IEEE Elec trific ation Magazine / D EC e mb e r 201 8 37
 The importance of such energy-efficient appliances and
fixtures cannot be understated. They will quite often be
used when grid power is not there and would depend on
solar power and power through batteries. The latter power
source is quite expensive and needs to be minimized. The
best way to do this is via dc-powered appliances/fixtures.
Grid-Connected Solar-dc Systems
As appliances/fixtures get added to homes, and especially
the induction stove, 125- or 200-W solar panels will not be
sufficient to provide the required energy. Fortunately, the
Inverterless-500 can use up to a 500-W solar panel. Increas-
ing the size of the solar panel would increase the cost but
not be too large an amount. The 1.25-kWh Li-ion battery
that was used with the system would not be enough. One
could use a larger battery (e.g., 2.5 kWh) but that would add
substantially to the costs. What are the options?
The first option would be to connect the solar-dc sys-
tem to the grid. Even if the grid is unreliable and available
only for about 10 h/day, the 250- to 375-W(p) solar-dc sys-
tem with the 1.25-kWh battery may be adequate to provide
all the power required by the home. Table 1 provides the
power consumed by different appliances and fixtures
48 Vdc
during the day (e.g., 14 h) and at night (10 h). The energy
required in this example is 1,535 Wh during the day and
1,335 Wh at night. A 375-W(p) solar panel will deliver
1,500 Wh of energy in most of India, considering the con-
version and cable losses. Thus, a 375-W(p) solar panel and
close to 1.4 kWh taken from the grid would adequately
meet the requirement. A 1.25-kWh battery will be enough
even with 14-h power cut. If the 250-W(p) solar panel is
used, the consumption from the grid would be larger, and
a larger battery may have to be used.
The second option can be used when the grid is absent
in the location. One of the limitations of the stand-alone
solar-dc system at individual homes is that the home has
to consume all the energy, minus the losses, that the solar
system generates. It has a battery to store excess energy or
take deficit energy on some specific days but, the small
storage will overflow or underflow quickly. If one home has
excess energy on most days and a nearby home has deficit
energy, the two cannot share. If the nearby homes are con-
nected to each other, the excess energy can be shared with
a deficit-energy home whenever required. This will also
reduce the storage required at each home and, therefore,
reduce the costs. Microgrids that provide such connections
become important.
Here, we present two types of
rural microgrids. Figure 7 illustrates
the first one. Here, solar-dc systems
in individual homes are connected to
each other on a 380-Vdc microgrid.

48 Vdc

48 Vdc

48 Vdc

The 380-Vdc connection is bidirec-

tional. The solar-dc system at the
homes can take in energy from or
give out energy to the microgrid, thus
380 Vdc sharing energy between homes. If a
home uses an appliance on some
days for a longer time, it can draw
Figure 7. A 380-Vdc microgrid connecting the solar-dc system at each home. power from other homes in the grid
that have excess stored power. The
control for such a system is decen-
tralized. Based on the state of the
battery, the home is placed in a sur-
plus mode, when power can be
drawn from it, or in a deficit mode,
48 Vdc

48 Vdc

48 Vdc

48 Vdc

when it can draw power from the

microgrid. The state surplus or defi-
cit can be changed by the local con-
troller any time. The controller can
also limit the maximum current
380 Vdc that a home can draw or deliver
dynamically. When a home in the
deficit state attempts to draw power
from the microgrid, all homes in the
Storage surplus state will contribute. If none
of the homes are in the surplus
Figure 8. The microgrid with centralized power-generation sources like wind and solar and central- state, the home will not be able to
ized storage. draw power.

38 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018

 The second type of microgrid does much more than
sharing. Although it connects the solar-dc systems of each
home to a 380-Vdc microgrid that enables power sharing
between homes, this microgrid has power generation
sources and storage. As shown in Figure 8, the centralized
power generation sources could be wind and/or solar. The
centralized storage can be large and store excess ener-
gy from a home or from centralized sources. Also, it is possi-
ble to connect the power grid to this microgrid when
available. The grid itself is 380 Vdc.
It is envisaged that the solar-dc system in each home as
well as the centralized power generation and power
storage has communication capability [e.g., General Pack-
et Radio Service (GPRS)]. Through this, each source, storage,
and consumption point share the data with a centralized
power management system (PMS). The PMS can decide on
the extent of power to be stored at any time at each home
and centralized storage as well as which node will transmit
and which will receive the power, while sharing power. If
there is an overall shortage of power, the PMS can inform
the individual node to restrict usage.
Microgrid in Jharkhand Using the DERs
A microgrid with the DERs, like that described in Figure 8, is
being built in the state of Jharkhand in India jointly by IIT
Madras and ABB India Pvt Limited and is supported by the
Ministry of Human Resources and Development. The current
power supply to all homes, industries, and water pumps in
the village is provided by such a microgrid. The only differ-
ence is that the village grid is not 380 Vdc but three-phase
440 Vac, commonly used in India. The centralized source of
power is solar. The electric grid is also connected to the
microgrid with a two-way power transfer capability between
them. The electric grid today is highly unreliable and has
power cuts in excess of 10 h/day. The distribution at each
home and industry is a combination of 48 Vdc and 230 Vac
(on two separate circuits). There is a provision of homes
shutting down their ac supply, in case of an overall power
shortage. With the new microgrid, many homes will have
their own solar system and battery, making the power gener-
ation distributed. As mentioned previously, the load manage-
ment is an inherent part of this experiment.
Yet another pilot is likely to occur at Andaman and
Nicobar Island of India as an Indo-Sweden project, sup-
ported by the Department of Science and Technology,
Government of India. This pilot will use a 380-Vdc microgrid
for the village and 48 Vdc for power distribution within
homes and industries. It will have centralized wind and
solar generation, in addition to centralized storage. Each
home and centralized resource will be manageable. The
strategy of managing the DER and loads will evolve to
provide optimal utilization.
Conclusion
The rooftop solar-dc systems for homes and small offices
was a departure from the 100-year-old practice of using ac
within homes. There was, however, a strong logic: the solar
photovoltaic system produces dc power, storage (batteries)
takes in and out only dc power, and most appliances and
fixtures are dc powered. The ac-to-dc conversion, and vice
versa, amounted to large losses, especially when the power
levels used were small. The dc distribution at homes
reduced the losses and, therefore, reduced the size of the
solar panel and batteries used as well as the overall costs.
The solar-dc system contributed significantly to getting
electricity to every village in India by last April.
While this met the immediate needs of low-income
rural homes, it was not enough to provide quality power
for rural homes or rural industry in the future. The power
grid is nonexistent or highly unreliable in several places.
The dc standalone homes do not share power with each
other, and rural industry still depends on expensive
power from a diesel generator. A different approach was
needed to provide reliable power to India’s remote and
rural areas. In this article, two novel microgrids are pro-
posed to connect the decentralized resources and loads
as possible means to overcome this issue. The first is a
simple 380-V microgrid so that solar-dc homes can share
the excess power with each other when needed. The sec-
ond type of microgrid does more than sharing, as the dc
microgrid itself has power-generation sources and stor-
age. These robust solutions are expected to provide power
to rural India. These microgrids are being installed in
Jharkhand and the Andaman Islands over the coming
year and will help pave the way for future solutions for
rural India.
For Further Reading
S. Shah. (2017, Mar. 14). Why Indian solar process will fall further,
by how much and when. Green World Investor. [Online]. Available:
http://www.greenworldinvestor.com/2017/03/14/why-indian-
solar-prices-will-fall-further-by-how-much-and-when/
A. Jhunjhunwala, A. Lolla, and P. Kaur, “Solar-dc microgrid
for indian homes: Transforming power scenario,” IEEE Electrif-
ic. Mag., vol. 4, no. 2, pp. 10–19, June 2016.
A. Jain, A. Ramji, and S. Ashraf, Innovations for the Last Mile:
Evaluating Prices, Performance and Perceptions for Decentralised
Solar dc Electrification. New Delhi, India: Council on Energy,
Environment and Water, 2017.
P. Kaur, S. Jain, and A. Jhunjhunwala, “Solar-dc deploy-
ment experience in off-grid and near off-grid homes:
Economics, technology and policy analysis,” in Proc. IEEE
Int. Conf. dc Micro­­grids, Atlanta, GA, 2015. doi: 10.1109/
ICDCM.2015.7152004.
Guidelines for 48v ELVDC Distribution System, Bureau of Indi-
an Standard Draft ETD50 (11294), 2017.
Biographies
Ashok Jhunjhunwala (ashok@tenet.res.in) is with the
Indian Institute of Technology Madras, Chennai.
Prabhjot Kaur (prabhjot@tenet.res.in) is the chief exec-
utive officer of the Centre of Battery Engineering and Elec-
trical Vehicles at the Indian Institute of Technology,
Madras, Chennai.

IEEE Elec trific ation Magazine / D EC e mb e r 201 8 39
 By Ashok Jhunjhunwala, Prabhjot Kaur,
and Sushant Mutagekar

Electric Vehicles
in India
A novel approach to scale electrification.

ver the last few years,

O
electric vehicles (EVs) have
captured the imagination of
people in many parts of the
world. Approximately 1.1 mil-
lion passenger EVs (cars) were sold in 2017,
up by about 57% from the previous years.
China contributed 600,000 vehicles, the
United States had 200,000 and Europe
125,000. EV sales in Norway constituted
50% of all vehicle sales. Several nations
have announced that their vehicles will be
fully electric by 2025, 2030, or 2040. General
Motors, Ford, Toyota, Volkswagen, and oth-
ers demonstrated their EV ambitions by
making major EV announcements, while
Chinese automakers like BAIC and Chan-
gan announced they will sell only EVs after
2025. According to Bloomberg, the global
EV sales will grow by 40% in 2018. U.S.
sales are expected to exceed 300,000 units,
and European sales should reach around
400,000, with Germany as the leader. China
will lead the way in four-wheeled vehicle
as well as electric bus sales. Beijing has
committed to completely switch over its
taxi fleet of around 70,000 vehicles by 2020.
Moreover, by the end of 2018, charging
infrastructure is expected to constitute
almost 700,000 stations.

Digital Object Identifier 10.1109/MELE.2018.2871278

Date of publication: 16 November 2018
©istockphoto.com/Petmal

40 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018 2325-5987/18©2018IEEE

 India has recognized that EVs are the future of road
transportation. However, even with EVs’ much higher
energy efficiency and significantly fewer moving parts
(and, thus, significantly higher reliability), customers
often avoid purchasing EVs today only because of their
higher costs. Because lithium (Li)-ion battery prices are
falling rapidly, it is a matter of only a few years before
EVs become a preferred vehicle in India. So far, only a
lone company, Mahindra Rewa, sells a small number of
cars every year. Only in 2017 did the industry and gov-
ernment take note that, if they failed to act, they would
lose the race, and the country would be flooded with
imported EVs. Besides, in most large cities, India’s air
quality is incredibly poor. Petrol/diesel vehicles con-
tribute significantly to such pollution, and EVs can pro-
vide the answer.
Industry, academia, and research and design (R&D) per-
sonnel (with some government support) got together in
2017 and created a task force to figure out a solution. The
group realized that EVs were being promoted all over the
world with large government support and subsidies. In
fact, subsidies in the United States, Europe, and China
range from 30% to 40% of the total cost. However, the Indi-
an government was not in a position to provide large sub-
sidies, even though some tax concessions and limited
incentives could have been possible. Making any signifi-
cant headway in such a situation looked like an impossible
task. However, the task force did not give up and persisted
in finding a solution for the Indian context.
India’s Unique Situation
India’s vehicle composition is very different from that in
many other parts of the world. Table 1 shows the annual
sales of different kinds of vehicles over the last six years. It
is obvious that two-wheelers dominate the Indian automo-
bile sector (a typical traffic scenario on the road is shown in
Figure 1). Three-wheeler taxis (called autos) operate all over
India and carry a large number of passengers. The com-
mercial vehicle segment including buses and trucks rose
to 856,000 units during 2017–2018. Buses represent a
IEEE Elec trific ation Magazine / d ec emb er 201 8 41
 Table 1. Automobile sales trends (data from the Society of Indian Automobile Manufacturers).
Category 2011–2012 2012–2013 2013–2014 2014–2015 2015–2016 2016–2017
Passenger vehicles 2,629,839 2,665,015 2,503,509 2,601,236 2,789,208 3,046,727
Commercial vehicles 809,499 793,211 632,851 614,948 685,704 714,232
Three-wheelers 513,281 538,290 480,085 532,626 538,208 511,658
Two-wheelers 13,409,150 13,797,185 14,806,778 15,975,561 16,455,851 17,589,511
Total 17,361,769 17,793,701 18,423,223 19,724,371 20,468,971 21,862,128

and 150,000 and is not affordable at higher

prices. The same is the case of four-wheeled
passenger vehicles. Though the number of these
vehicles is also growing substantially, 28% of
them cost below 0.5 million and 56% cost
between 0.5 and 1 million. Only 16% of the
vehicles are sold at a price exceeding 1 million
(US$15,000), as shown in Table 2.
There are other ways Indian vehicles differ
from those used elsewhere in the world. Most
vehicles in Indian cities are driven at low speeds,
averaging under 25 km/h, so the vehicles have to
be designed to be energy efficient at such speeds.
Furthermore, they rarely travel long distances. A
privately owned two-wheeler would typically
travel 20–30 km, and a four-wheeler would travel
Figure 1. The typical traffic pattern on Indian roads (Sardar Patel Road, Chennai, India).
about 30–40  km/day. Besides, the ambient tem-
perature in most parts of the country is over 35 °C
and can exceed 45 °C on many days. One needs to
understand the impact of these temperatures on
Table 2. Costs of passenger vehicles
the life cycles of EV batteries when they are being
(four-wheelers) in India.
charged as well as discharged.
Cars Sold in India (%) EV charging infrastructure is a major ex­­
Price Range 2015–2016 2016–2017 2017–2018 pense everywhere in the world. Who would
build such an expensive infrastructure in India,
Below 500,000 28.02 28.85 27.43
especially because it is unlikely to financial­­
0.5–1 million 55.49 54.96 56.48 ly break even for a long time to come? In India,
interest rates on capital hover over 10%, making
1–1.5 million 15.29 15.23 14.65
it difficult to invest and receive returns over the
Above 1.5 million 1.20 0.96 1.43 long run.
Most importantly, because India’s EV program
Source: The sale figures and costs are compiled from the Society of Indian Automobile Manufacturers
and selling market prices in India. will get no subsidy (or only a very limited one), the
country needs to determine how its EV strategy
can evlove so that it will not require a substantial
significant part of this total and provide public transport for financial subsidy from the government.
a large segment of India’s population. Therefore, if India It was clear to the task force that it must evolve an ap­­
could start its EV program with two-wheeled and three- proach different from that adopted elsewhere and consider
wheeled public vehicles, it would make a large impact, both India’s uniqueness as a strength. Stakeholders must be
socially as well as in terms of the environment. City buses courageous in the ways they innovate and come up with
could follow. approaches that make commercial sense in India today.
The second point that needs to be understood is the
low cost and affordability of such vehicles in India. Two- India’s Strategy
wheelers mostly retail between 40,000 and 100,000 Given the constraints/opportunities discussed previously,
(dominated by low-speed scooters with a price lower than India’s EV strategy evolved by focusing on
55,000). A three-wheeler auto sells for between 130,000 1) the energy efficiency of EVs

42 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

 2) adding battery swapping as an option to charging and
developing a charging and swapping infrastructure
3) an end-to-end battery ecosystem from materials to
battery pack
4) the generation of demand, especially with electric
public transport.
The strategy resulted in a unique approach for India’s EV
ramp-up. We will illustrate this with an example of an
electric auto widely used for public transport.

Energy Efficiency Enhancement

The focus was to minimize the use of energy (watt-hours)
per kilometer of travel. A typical electric three-wheeler
auto [like the one shown in Figure 2(a)] consumed 80 Wh/
km on Indian roads in early 2017. Because this was con- (a)
sidered excessive, a goal was established to reduce
consumption to 45 Wh/km. It then appeared to be an
impossible task. Brushless dc electric motors or switched
reluctance motors were designed to replace induction
motors. Tires were improved to lower rolling resistance,
and attempts were made to reduce the weight of the vehi-
cle. Finally, creating better vehicle aerodynamics helped in
enhancing energy efficiency.
Over the last ten months, most auto manufacturers
have reduced their products’ energy consumption to
below 52 Wh/km. The 35% reduction in energy usage
means that the battery size required to travel a certain
distance decreases by 35%. Because the battery dominates
the costs of the EV, this reduction is substantial, cutting
(b)
the subsidy required. More can be accomplished in the
future. Distributed motors will be one way to go. The strat- Figure 2. (a) An electric auto (the most widely used example of a
egy of enhancing energy efficiency is paying dividends in general three-wheeler) and (b) an electric rickshaw.
all kinds of EVs. In the case of city buses,
the energy requirement has been reduced
by 40%.

Battery-Swapping Options
Even though enhanced energy efficiency
brought the vehicle’s capital cost down,
some subsidy was still required if EVs
were to compete with petrol-based autos.
The cost for a battery that provides a
desirable range would still be substantial.
What if one used a much smaller battery
size, say, a third of what would provide an
acceptable range? This requires fast-
charging several times in a day, and the
low-cost batteries used could not be
charged fully in less than an hour. Waiting
an hour for batteries to charge is unac-
ceptable. What if one could simply swap
the discharged battery with a charged
one? Figure 3 presents the easy swapping (a) (b)
mechanism developed for a three-wheel-
er and an example of chargers used for Figure 3. The swapping of batteries with (a) one of the batteries being taken out of a
such three-wheeler batteries. The waiting three-wheeler and (b) a charger used for these modular, swappable small batteries.

IEEE Elec trific ation Magazine / d ec emb er 201 8 43

 time would now decrease to a few
minutes—quite acceptable to the
auto driver. The engineering challeng-
es involved in handling and fastening
batteries and using appropriate long-
life connectors were addressed. One
big advantage is that batteries no lon-
ger need to be fast-charged in the
vehicle, especially at high ambient
temperatures. The vehicle could be
ready to leave within minutes. The
swapped-out battery can be taken to
a conditioned indoor environment
and charged in about 2 h. This would
preserve the life cycles of even low-
cost battery chemistries. The batteries
would no longer be sold with the ve­­
hicles; instead, a separate business, referred to as an ener-
gy operator (EO), would purchase the battery, charge it, and
lease it to a vehicle at different outlets. An EO would have
to set up a large number of swapping stations so that a
vehicle could easily access one when needed.
For an EO’s battery-swapping venture to be successful, it
is desirable that a common battery be used in different
manufacturers’ autos. The automobile manufacturers in
India got together and determined this common battery,
confining themselves to the establishment of battery ener-
gy capacity (in kWh) and minimum number of life cycles
as well as the maximum size and the maximum weight.
These manufacturers also defined the connector and the
communication protocols the battery will use to talk to a
vehicle and to a charger, but they left the chemistry to the
battery manufacturers. Thus, as newer chemistries emerge,
batteries incorporating these can be used. While this
description is for battery swapping in a three-wheeler, it
also works for buses, two-wheelers (e-scooters), and four-
wheelers (passenger vehicles), as discussed later.
Hitherto, it was understood that the charging infra-
structure is a precondition for EVs to work, but battery
swapping adds a new dimension. Because battery swap-
ping is economically viable, businesses can set up battery
swapping and the required battery-charging infrastruc-
ture. This would enable EV usage to take off. Vehicle charg-
ing can then be added by the same business that sets up
the battery-swapping infrastructure based on require-
ments. The difficult problem of setting up charging infra-
structure is largely resolved.
Digitalization of the Battery-Swapping Operation
The battery-swapping operation, involving charging and
swapping, payment for different services, and perfor-
mance monitoring, is quite complex. This is simplified by
the digitalization of the whole process, assisted by mobile
telephony. The EO would charge for the leased battery
based on kilowatt-hour usage and the term of the bat-
tery’s lease. It is important that customers not be able to
44 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018
It is important that charge the swappable battery them-
selves and, instead, would return it as
customers not be soon as it is significantly discharged.
To ensure this, locked smart batteries
able to charge the were designed. Such a battery cannot
swappable battery be charged except by an EO-autho-
rized charger and can be discharged
themselves and, only by the vehicle for which it is
leased. This is accomplished by dis-
instead, would return connecting the battery from the ter-
it as soon as it is minals/connectors internally and
allowing a communication protocol
significantly to be on only at the beginning. The
battery first communicates with the
discharged. authorized charger or vehicle, and
encrypted tokens are passed for
authentication (similar to a block-
chain). Upon authentication, the battery is connected to its
terminals, allowing the input and output of energy.
Furthermore, the battery communicates with the vehi-
cle controller and stores (in its battery management sys-
tem) complete data about vehicle usage, including speed
and acceleration every second and the amount of energy
used. It also stores information concerning the battery’s
state of charge, the state of balance of cells, and the tem-
perature of each cell. A vehicle-to-battery protocol is
defined to enable this as well as the authentication. Simi-
larly, a battery-to-charger communication protocol is
defined. After authentication, the charger picks up all the
stored information from the battery and sends it to the
cloud. While charging the battery, it also receives infor-
mation on cell balance, cell voltage, and currents and
temperature. There is an option of adding a global posi-
tioning system to the vehicle and recording the position-
ing information in the cloud. All these data are then
processed to determine an individual’s driving habits, the
vehicle’s performance (especially its energy efficiency),
and the battery behavior during charging and discharg-
ing. The latter will help to ensure that the battery is used
optimally and has a long life.
The careful design of the battery-swapping and com-
munication protocols ensures that the battery-swapping
business becomes viable. At the same time, the user bene-
fits in comparison to using a petrol vehicle.
Battery Ecosystem
The battery is the key subsystem that makes EV use possi-
ble and dominates the cost of EVs. A nation cannot trans-
verse to full electric mobility without building a sound
battery ecosystem. India’s low affordability implies that it
should choose the lowest-cost battery. Table 3 provides dif-
ferent Li-ion battery options available today. Note that Li
cobalt oxide (LiCoO2)/graphite, nickel cobalt aluminum
oxide (NCA)/graphite, and nickel manganese cobalt (NMC)/
graphite are not only the lowest-cost options, but they
have the highest specific energy in terms of watt-hour per
kilogram. This makes them the light-
est among the Li-ion cells. These cells,
however, have moderate cycles unless
silica is added to the graphite anode.
Others used ceramic separators to
enhance the number of life cycles. It
is expected that some variations of
these chemistries are likely to have
specific energy in excess of 400 Wh/kg
and cell costs of about US$80/kWh in
the future.
The other drawback for these cells
is their poor high-temperature behav-
ior and safety aspects. Careful ther-
mal design and the design of a battery
management system needs to ad­­
dress these weaknesses. An alterna-
tive is NMC/lithium titnate (LiTO),
which has excellent life cycles, can
withstand high temperatures, and is safe. The problem of
these cells is the weight and cost. Unless the battery used
is very small and charged very frequently, the NMC/LiTO
option is not attractive.
The lithium iron phosphate (LiFP) cells fall in between.
They have a moderate cost, even though they are safer
and have slightly higher life cycles than NMC. However,
the cells have a theoretical limit of 160 Wh/kg for their
specific energy. China has set a target for all EVs to have
350 Wh/kg by 2020, 400 Wh/kg by 2025, and 500 Wh/kg by
2030 (China Association of Automobile Manufacturers,
Beijing). Most of the world uses NMC/graphite, and the
future in the near term clearly suggests NMC/graphite.
Therefore, India opted for NMC/graphite as the cell of
choice and went on to build the battery ecosystem. This
ecosystem includes making battery packs using the cells
(30%–35% added value), cell manufacturing (25%–30%
added value), and securing materials and chemicals
(about 40%–45% added value). Over the last couple of
years, India has mastered
xx cell-to-pack manufacturing involving quality thermal
design to ensure that packs will work in Indian
The country has temperature conditions quality me­­
chanical design ensuring that each
begun to recycle cell has the right pressure a battery
management system ensuring bal-
existing batteries, anced charging and discharging of
mastering recovery cells and guaranteeing that no cell
has thermal runaway.
of 90%–95% of such India does not have the technology
to manufacture commercial Li-ion
materials from used battery cells that can compete with
batteries in an the rest of the world’s. The approach
here is to invite international compa-
environmentally nies that have the best cells. The most
complex part is getting the materials.
friendly manner with India does not have the resources
zero effluents. needed. Therefore, the country has
begun to recycle existing batteries,
mastering recovery of 90%–95% of such
materials from used batteries in an
environmentally friendly manner with zero effluents. The
task force recognized that India’s Li-ion battery material
strategy must be based on such urban mining.
Demand Generation Strategy
The final aspect of India’s EV strategy focused on creating
some early demand for vehicles and subsystems so that
the EV industry could grow. The demand generation strat-
egy the task force determined is based on public vehicles.
The task force would work with the government to create
quantity requirements for three-wheelers, four-wheel
taxis, and public intracity buses. At the same time, it would
persuade the government offices to start using only EVs.
This early demand generation is spurring the industry to
move forward rapidly. Table 4 provides the industry eco-
system that has emerged for EVs over the last year.
Electric Autos
As discussed in the sections “Energy Efficiency Enhance-
ment” and “Battery-Swapping Options,” the auto is a low-
cost taxi, used widely not only in big cities but also in
most of India’s small towns. The vehicle could easily be

Table 3. The battery cell options available today.

Li-Ion Cell LiCoO2/Graphite LiFP/LiTO
Chemistry or NCA/Graphite NMC/Graphite LiFP/Graphite NMC/LiTO ­(Niobium Doped)
Specific energy 150–300 150–300 90–120 (150 with 60–100 50–80
(Wh/kg) silica in anode)
Charge/discharge 0.5C/1C 1C/1C (2C with 1C/2C (4C with 4C/4C 5C/10C
rate silica in anode) silica in anode)
Life cycles 1,000 2,000 (8,000 3,000 (4,000 with 10,000 20,000
with silica) silica)
Safety Cell < 55 °C Cell < 55 °C Safer Safest Safest
Cell costs/kWh US$120 US$145 US$225 US$500 High
The C-rate is a measure of the rate at which the battery is charged/discharged relative to its maximum capacity.

IEEE Elec trific ation Magazine / d ec emb er 201 8 45

 Table 4. Some industries that have committed to scaling EVs and EV subsystems
manufacturing and services in India.
EVs: Ashok Leyland, Tata Motors, Mahindra Electric, Eicher, Bajaj, Kinetic, Lohia, Electrotherm,
Goenka, Hero-Eco, Okinawa, Ather, Avon Cycles, TVS Motors, and Mahindra and Mahindra
Li-ion battery and recycling: Exide, Amar Raja, Exicom, ACME, Grintech, Greenfuel, Ion Batteries, Attero, and Sun Mobility
EOs: Essel Infra, Sun mobility, BPCL, NTPC, PGCIL, and Kerala DISCOM
Chargers and motors: Exicom, TVS Motors, Consulneowatt, Valeo, Compageautomation, most state governments,
and state transport units
BPCL: Bharat Petroleum Corporation Ltd.; NTPC: National Thermal Power Corporation Ltd.; PGCIL: Power Grid Corporation of India Ltd.

(a) (b) (c)

Figure 4. Passenger vehicles in rural India (a village near Red Hills, Chennai, India). (a) A rural Indian three-wheeler bike and (b) and (c) tractors
being used to transport people and materials.

The autos were sold without a battery at a

cost similar to that of petrol autos. The
batteries would be purchased by an EO,
who would set up charging and swapping
stations and lease the batteries. The eco-
system is ready to scale.
E-rickshaws (like the one shown in Fig-
ure 2) are low-speed versions of e-autos.
Wider-body autos are used extensively in
rural areas (as in Figure 4) and carry a larger
number of passengers. Cargo autos are used
to carry goods within cities. They can all be
electrified using the approach described pre-
viously. Making them energy efficient and
defining standard batteries would be the key.

Figure 5. An overloaded bus on an Indian road (Adyar, Chennai, India).
converted to electric, but, when used with a sufficient-
sized Li-ion battery, the capital cost is too high. The task
force met with a group of auto manufacturers to work
together on
1) enhancing the vehicles’ energy efficiency from 80 to
52 Wh/km
2) defining a common and smaller battery to provide a
minimum range of 50 km.
Electric Buses
The 9- and 12-m buses are used extensive-
ly in most of India’s large cities to carry
intercity traffic. A typical bus route is in about 25–30 km.
Because of high-traffic routes, most of these buses move
slowly (averaging 15 km/h making about eight to ten trips
per day). India is adopting a novel approach to electrify
such buses. First, the energy efficiency buses are en­­
hanced from 1,600 Wh/km to about 950 Wh/km when air
conditioning is not used. Next, to keep the costs low, a bat-
tery of 55 kWh was standardized. This ensures a 35-km

46 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

travel distance even when air con- India opted for
ditioning is used and the battery is
near the end of life. This battery is NMC/graphite as
swapped at both ends of the journey
(see h t t p s ://www.youtube.com/
the cell of choice and
watch?v=8ibDFMa4JMA). A con­­ went on to build the
sequence of this is that the battery
weight is only about 600 kg. The lower battery ecosystem.
weight means that the bus can carry
more, which is important in a country
like India, where buses are often overloaded (as shown
in Figure 5).
Personal Vehicles (Two-Wheelers
and Four-Wheelers)
As discussed so far, the use of battery swapping in addi-
tion to fast-charging has given a new boost to EV use in
India. But how does one make personal vehicles economi-
cally viable in the absence of a subsidy? The task force has
come up with a unique strategy.
It is illustrated with a low-cost petrol car costing about
400,000. If this vehicle is redesigned as an EV with a low-
cost NMC/graphite battery of, say, 100-km range, it should be
possible to sell it at 475,000. However, if a battery with a
200-km range is used, the cost would reach 650,000, making
it too expensive for those. Thus, a 100-km vehicle range is the
only option. Now, typical users of such a car drive under 100 km
90%–95% of the time. On those days, overnight charging of
the vehicle is adequate. The difficulty is the other 5%–10% of
days. The user must find an available fast-charger in the day-
time and wait there for an hour to charge the vehicle. This
wait is not acceptable, and so the user may not switch from a
petrol vehicle to an EV. The attempt to charge faster will also
severely impact the life of the low-cost battery.
So what is the option? What if the vehicle is designed
to have two compartments, each with a capacity for a
100-km-range battery: one compartment containing the
fixed battery and the second left empty. On 90%–95% of
days, overnight charging of the fixed battery is sufficient.
On the day the user needs to travel a longer distance, he
or she drives to a petrol station that provides charged
100-km-range batteries, referred to as range extension (RE)
batteries. Such a battery could be mounted on the vehi-
cle in about 4 min, and the vehicle would then have
an additional 100-km range. If a vehicle needs to travel
even further than 200 km, the used RE battery could be
swapped at another petrol station, extending the range
to 300 km. The vehicle can go on and on by further swap-
ping when needed. There is no range limitation, and
fast-charging infrastructure is not even required. But, if
fast-charging infrastructure is available, then the user
would have an option of either fast-charging batteries
(waiting for an hour) or swapping.
The same approach can be used with two-wheelers.
E-scooter manufacturers have gotten together and de­­
fined a standard RE battery. The e-scooter has slots for two
batteries. One is fixed, typically
providing a 50-km range, and the RE
battery (with another 50-km range) is
swapped in when required. These vehi-
cles are being tested today and will
soon be launched into the market. Be­­
cause two-wheelers dominate the Indi-
an auto market, the switch over to
electric is expected to happen rapidly.
Conclusion
As discussed in this article, India has chosen to take a
novel approach for its EV program, recognizing that EVs
are important but also considering the obstacles of low
affordability and lack of a large government subsidy. Its
approach to EVs will, therefore, not be consonant with
that in the rest of the world. Thus, battery swapping has
been added to public charging. Users no longer have to
wait for expensive charging infrastructure to be built.
Businesses would set up a battery-swapping infrastruc-
ture because it would make economic sense. They will add
appropriate public slow- and fast-charging, where there is
a demand. India has decided to carry out urban mining to
secure battery materials. It recognizes that the higher effi-
ciency of drivetrains brings down the cost of an EV. It
needs to carry out R&D to develop the most efficient elec-
tric drivetrain including innovative motor and controller
design and other power-electronic subsystems. India’s
approach is more in tune with India’s economy. Its large
market gives it a chance to establish this alternate
approach. The approach may be useful not just to India
but probably to 70% of the world, where similar affordabil-
ity exists. Only time will tell.
For Further Reading
Bloomberg, “Electric vehicles: 10 things to watch for 2018,”
Bloomberg New Energy Finance, 2018.
Society of Indian Automobile Manufacturers. (2018). Publi-
cations and reports. [Online]. Available: http://www.siamindia
.com/
U.S. Department of Energy. (2016, June 6). Overview of the
DOE VTO advanced battery R&D program. [Online]. Available:
https://www.energy.gov/sites/prod/files/2016/06/f32/es000_
howell_2016_o_web.pdf
A. Jhunjhunwala. (2017, Dec. 1). Understanding the EV ele-
phant. [Online]. Available: http://electric-vehicles-in-india
.blogspot.in
Biographies
Ashok Jhunjhunwala (ashok@tenet.res.in) is with the
Indian Institute of Technology Madras, Chennai.
Prabhjot Kaur (prabhjot@tenet.res.in) is with the Cen-
tre of Battery Engineering and Electrical Vehicles, Indian
Institute of Technology Madras, Chennai.
Sushant Mutagekar (sushantmm.iitm@gmail.com) is
with the Indian Institute of Technology Madras, Chennai.

IEEE Elec trific ation Magazine / d ec emb er 201 8 47
 By Henning Wilms, Dominik Mildt, Marco Cupelli,
Antonello Monti, Peder Kjellen, Thomas Fischer,
Demijan Panic, Michael Hirst, Eugenio Scionti,
Sebastian Schwarz, Paul Kessler, and Luis Hernández

Microgrid Field
Trials in Sweden
Expanding the electric infrastructure
in the village of Simris.

icrogrids (MGs) are electri­­

M
city-­distribution systems con­
taining loads and distributed
energy resources (DERs) that
can be operated in a con­
trolled, coordinated way, either while
connected to the main power network
and/or while islanded. MGs have been
around for decades in the energy sys­
tem. They can be commonly found
in critical infrastructure buil­­ding
complexes such as military bases,
hospitals, and data centers. With the
challenges of the new energy world,
network operators are starting to
look into the technological solutions
used in MGs (e.g., MG controllers) as an
alternative to their conventional net­
work technologies.

Microgrids Status Quo

For modern microgrid operation, a general
hierarchical control structure has been devel­
oped that separates control tasks into different
levels and time horizons. Concepts are derived from
the hierarchical control in transmission grids and adjust­
ed for the needs of distribution system operators (DSOs) and
MGs in particular (Bidram and Davoudi 2012). Just as in classic
transmission systems, control is conceptually divided into zero, primary,
secondary, and tertiary levels. Time constant, resource interconnection, and
©wikimedia commons

Digital Object Identifier 10.1109/MELE.2018.2871295

Date of publication: 16 November 2018

48 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018 2325-5987/18©2018IEEE

communication needs increase along with the proposed layers. Figure 1 shows a simplified overview of
the tasks of different control levels and an approximation of their respective time frames.

Lower-Level Control
The component control level comprises the internal control loops and protection of the DERs. This
includes the control of classical generators and power electronic devices. Primary control provides ampli­
tude and frequency references for those inner loops, e.g., using droop control to stabilize the network and
prevent the circulation of active and reactive power. Secondary control ensures that frequency and volt­
age deviations are kept within predefined limits and safeguard appropriate steady-state behavior. An
overview of the related approaches can be found in the study by Guerrero et al.

Higher-Level Control
The tertiary level of control is usually the highest level of technical control in MGs and works on time
frames between several minutes and hours. Specifically, the tertiary level fulfills the tasks of steady-state
operational planning, power flow control, and optimal usage of all the DERs. For this purpose, active and
reactive power set points are calculated for all actively controlled DERs, based on the expectations about
future behavior. An energy management system (EMS) is used in this stage to find the optimum refer­
ences under a combination of economic and technical objectives. An overview of some EMS
strategies is provided in the study by Minchala-Avila et al. An important aspect of the
EMSs in MGs is the inherent uncertainty of the regarded problem. Not only is the
behavior of renewable energy sources (RESs) such as wind generators and
photovoltaics (PVs) not perfectly known for the future; the same holds
true for the actual electric and thermal demands of consumers. Those
tend to be particularly critical in the MG scenario, as the load con­
stituted by a single consumer is more difficult to predict than
larger load aggregations. The forecast of the operationally rel­
evant parameters is therefore an indispensable aspect of
high-level MG control.
DSOs do not always utilize all the functionalities
technically offered by an MG, e.g., switching between
islanded and/or interconnected mode; hence, net­
work operators talk about local energy systems (LESs).
In this context, MGs can be seen as a special sub­
group of LESs.

Renewable Energy Source Projections in

Europe: Uncertainty Scenario for the
Distribution System Operators
RESs are a major component of the energy transition
in Europe. The member states have set a new target
goal for the European Union (EU) to cover at least 32%
of gross final energy consumption from renewables by
2030. In the long term (2050), the share of the RESs should
exceed 80%.
The rapid development of RESs induces high investments
in distribution networks as the design of most of these networks
(designed and deployed decades ago) was not meant for and did
not foresee the massive connection of DERs. Figure 2 illustrates this
rapid evolution of the power system.
In some European countries, network tariffs paid by customers in
areas with a high penetration of renewables have risen to be the highest values
of all national regions (up to 50% higher than the national average price), whereas
electricity is consumed elsewhere in the country or beyond. Because of the current DSO-
regulated business model, the costs of renewable energy stay local, but the benefits do not. This
means that local communities in rural areas do not benefit from RES development due to very high
network costs.

IEEE Elec trific ation Magazine / D ec emb er 201 8 49

 The role of the DSO is to distribute power to connected
customers while keeping the power quality within the
compliance limits. Increasing network tariffs are caused by
the high investment costs linked to the RES integration as
new connections can overload the existing distribution
assets, disrupt the voltage values at the distribution line, or
a combination of both. If these problems are not mitigated
by the DSO, the protection devices installed at the distribu­
tion network, which are meant to protect the system, can
cause unwanted local blackout situations.
Because of the nature of the DSO’s business, regulation
defines the solutions that the DSO can use for mitigat­
ing network problems. Currently, most of the European
DSOs are bounded to invest only in copper (conductor and
substation upgrade) or conventional voltage regulation
devices (e.g., capacitor banks) instead of using the poten­
tial offered by LES flexibility and control. For this reason,
high network investment cost in countries with the obli­
gation to connect renewables regardless of the location is
a common occurrence, despite the uncertain development
of the RES in such local systems.
Typically, the short-term action of the DSOs is to exam­
ine which types of bottlenecks, including network conges­
tion, and/or voltage issues may occur. DSOs are required
to plan appropriate network upgrades to remove such bot­
tlenecks in accordance with national regulations. The

• Energy Management
Hours
• System Optimization
Tertiary • Optimal Dispatch
Control • Active/Reactive Power Reference Generation

Minutes
Secondary • Voltage and Frequency Restoration
Control • Real/Reactive Power Set Point Adjustment

Seconds

• Voltage and Frequency Stabilization

Primary Control • Droop Control
• Active Load Sharing
Milliseconds

• Inner Control Loops

Component Level
• Module Control
Control
Microseconds • Local Protection

Figure 1. The MG hierarchical control architecture.

The Energy System Needs to Become More Distributed and Renewable to Meet the Climate Objectives (e.g., Germany):
Yesterday Today 2050

6 Million Plus Renewable Power Plants2

Approximately 400 Large 1.6 Million Renewable
Conventional Power Plants Power Plants1 20 Million Plus EV Charging Points3
45 Million Plus Smart Meters4
1
Bundesnetzagentur: “EEG in Zahlen 2016,” 2017
2Bundesministerium für Wirtschaft und Energie: “Die Energie der Zukunft–Vierter Monitoring Bericht zur Energiewende,” 2015
3Bundesministerium für Umwelt, Naturschutz, Bau und Reaktorsicherheit: “Erarbeitung einer fachlichen Strategie zur
Energieversorgung des Verkehrs bis zum Jahr 2050–Endbericht,” 2016
4Assumption that all German households possess smart metering infrastructure in 2050.

Figure 2. The power system evolution.

50 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

entire process (planning, permitting, and upgrading
deployment) takes time, typically two to five years in
most of the European member states. Sizing the network
upgrade faces huge uncertainties when it comes to fore­
casting future additional renewable requirements and
ranking the bottleneck risks. In some cases, the uncertain­
ty is so high that the network planners do not know
whether the network capacity should be upgraded by
three times the current capacity or not at all.
As new market incentives for renewable energy are pub­
lished, their market uptake is highly unpredictable, which
causes further uncertainty for the decision to upgrade. For
example, in one project, new regulation from 2017 incentiv­
ized the installation of PV farms up to 500 kW and conse­
quentially promoted customer investment in PV farms to be
connected to a 24-km medium-voltage line. The first con­
nection requests to this line have shown that the cost of a
grid upgrade to mitigate the voltage disruption created by
two distributed generation assets would be at least three
times higher than the total investment value of the PV
farms themselves. According to current regulation, this cost
would then be distributed among all the customers con­
nected in the region, leading to higher grid cost in the area.
Moreover, uncertainty is expected to increase, as a wide
range of 0–50 additional PV farms could possibly be added to
the grid during the validity period of the regulatory incen­
tive. Planning an upgrade to mitigate the problem caused by
the first two PV farms (reactive approach) might lead to an
undersized solution that will need further upgrades in the
short to medium term. In contrast, planning for a number of
connections that has not been confirmed yet (proactive
approach) might lead to an oversized system (stranded
investment), which will also generate higher grid fees.
Modular systems are an appropriate answer to mitigate
the inherent uncertainty given by the market uptake of the
RESs. After studies were done in the aforementioned proj­
ect, it was realized that the use of some specific building
blocks of an MG system, e.g., the battery energy storage
system (BESS) and the power system controller (PCS),
would be up to four time less costly than the alternative
conventional grid upgrade when used for mitigating the
electricity network disruptions such as voltage deviations
caused by newly connected RES plants in a rural area. In
addition, the nature of an LES and its modular approach
would enable a faster and more cost-appropriate response
of the DSO to the connection request of its customers.
Types of Local Energy Systems
LESs can either be connected to existing distribution net­
works or deployed as off-grid systems. The former re­­
presents solutions for local energy communities where
network operators already manage the public grid,
referred to as grid-connected LESs. Grid-connected LESs
have the opportunity to operate in an islanded mode,
meaning that they can be electrically disconnected from
the distribution grid. The latter solutions are primarily
designed for remote destinations, where no public grid is
available, referred to as off-grid LESs. Off-grid LESs are
characterized by genuine standalone networks, i.e., island­
ed networks. One must additionally distinguish between
private and public MGs as follows:
xx Grid-connected private LES: The LES is operated by a
small or low-voltage DSO while satisfying the network
codes agreed with the connected DSO. This means
that the LES operator must satisfy the regulatory obli­
gations for itself and its LES customers.
xx Grid-connected public LES: The LES is operated by the
connected DSO directly as a service, but possibly with
different regulatory obligations for the LES.
xx Off-grid private LES: The LES is operated by a small or
low-voltage DSO with or without regulation.
xx Off-grid public LES: The LES is operated as a utility
with or without regulation.
The E.ON project commissioned in the village of Sim­
ris, Sweden, is an example of a grid-connected public LES
with the opportunity to operate in an islanded mode.
The Simris Project
In 2015, E.ON decided to start the design and development
of a technical pilot trial, motivated by the expected chal­
lenges associated with the increased integration of the
RESs. This trial demonstrated that an electrical system can
host a penetration of up to 100% of power obtained from
renewable sources (PVs and a wind turbine) by using field-
proven and market-available innovative technologies.
For this pilot project, the small village of Simris in south­
ern Sweden was connected and disconnected from the main
grid in a seamless way while being sourced at times solely by
renewably energy coming from a local wind turbine, a PV
farm, and rooftop PV installations of the local households.
The LES Simris was created on top of an already exist­
ing grid. Within the Simris area, the existing 10-kV dis­­
tribution system that forms the LES comprises seven
substations of 10/0.4 kV, of which five supply 150 custom­
ers. The feed-in point to the LES is via a 20/10-kV substa­
tion from a 10-kV bay. The bay is equipped with a power
breaker, voltage and current transformers, and relay pro­
tection devices.
In addition to the existing grid and RESs, the main
assets forming the LES are as follows:
xx an intelligent EMS communicating with all power pro­
duction units so that the LES delivers electricity with­
in the conventional power-quality limits
xx a main BESS of 333 kWh/800 kW, which operates as
the grid-forming unit and is in charge of the instanta­
neous balancing of the MG
xx a state-of-the-art secondary substation (DER substa­
tion), built next to the existing wind turbine and PV
plant and fully equipped with all the necessary com­
munication, protective relay, and control system de­­
vices needed to run the LES
xx a bio-diesel backup generator of 480-kW-rated power
IEEE Elec trific ation Magazine / D ec emb er 201 8 51
 Table 1. The overview on the installed assets and the systems at Simris.

System Description Manufacturers

Central battery 333-kWh capacity, 800-kW charging and discharging power Samsung, Loccioni, and TDE Macno
EMS Controls the LES assets and holds load-balancing Encorp
­responsibilities
Wind turbine 500-kW nominal power Enercon
PV plant PV plant size of 442 kWp SolarWorld and SMA Solar
Backup generator 480-kW, 4,500-L bio-diesel tank Scania and Coromatic
Substations Interconnects the LES assets, customers, and the main grid Holtab
DSR platform Enables the access to distributed flexibility sources by ICONICS
­steering smart technologies
Heat pumps New heat pumps Nibe
Retrofit heat pumps Ngenic
Water boiler Retrofit device MClimate
Residential PV and battery PV: 10 kWp Fronius Solar
Battery: 9.6-kWh capacity, 6.4-kW charging and
­discharging power
E-mobility Smart electric vehicle charging station Ensto
Smart meter Residential metering device Comsel
Peer-to-peer platform Visualizes customer energy data Lumenaza
Local measurement units Measurement of power, voltage, current, frequency, and Janitza
other physical parameters

The Simris EMS in charge of the load-balancing res­­

Figure 3. The Simris energy system. (Photo courtesy of E.ON.)
a demand-side response (DSR) platform, which steers
xx
smart technologies (e.g., heat pumps, hot tap-water
boilers, and electric vehicle charging stations) for sup­
porting and balancing the MG.
Table 1 provides an overview of all the assets and energy
systems present at Simris. Furthermore, Figure 3 shows
the Simris energy system comprising the local wind tur­
bine, the PV farm, and the containers hosting the BESS
and the backup generator.
The Microgrid Controller:
Energy Management System
The safe and reliable operation of the LES is realizable
with a three-layer energy management approach, as
shown in Figure 4.
ponsibilities consists of industrial field-proven hardware
(provided by Encorp) located in the BESS shelter, DER sub­
station, and Simris primary substation. In each location,
the EMS has access to relevant information, such as mea­
surements from current and voltage transformers and
status information from power breakers and disconnec­
tors. In addition, the EMS has direct control over the power
breaker at the point of common coupling (PCC) so that it
can command the island mode by switching off the break­
er when certain requirements are fulfilled.
RES generation is also controlled by the EMS, which can
demand the RESs to be curtailed if the BESS state of
charge (SoC) reaches certain levels. During times of high
demand and low generation, the EMS can demand the
combustion engine-driven backup generator to be activat­
ed and extend the time in the island mode.
The EMS controls the operational mode of the battery
system and sets the target voltage and frequency when
operating as an island. An overview of the components
and their dependencies is illustrated in Figure 5.
The Simris EMS can operate in different modes de­­
pending on the desired function, as presented later.
Virtual Island Mode
In the virtual island mode, the EMS controls the active
and reactive power set points to the BESS such that no

52 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

 Goal Control Responsible Technology Competencies Required

Optimization and Nice to

Seconds/ Optimized Scheduling All Steerable Loads
Forecasting, e.g., Load Have
Minutes
EMS Supplier

Seconds Local Balancing Wind Turbine, PV,

Local Control
and Diesel Generation

Battery System or High Power and Local Fast Control Critical/

Milliseconds Frequency Regulation High Energy Storage Assets e.g., PCS Mandatory
PCS Supplier

Figure 4. The three-layer energy-management approach.

exchange occurs over the PCC. In other words, in the virtu­ power are near zero. The system is then ready to be island­
al island mode, the PCS controller running in current- ed, and the EMS triggers the circuit breaker to open. The
source-inverter mode operates as a grid-following unit. It battery system does not change the operating mode but
is also possible to change the “exchange” set point over assumes the responsibility for managing the island volt­
the PCC to any value, within the capabilities of the battery age and frequency. The EMS updates the voltage and
system and restrictions of the power system. Thus, the frequency targets when islanded to maintain constant
MG can either consume or inject active and reactive 10.7-kV/50-Hz power, as these will vary along the BESS
power to the overlying grid, depending on current con­ droop curves with changing active and reactive power.
ditions and the state of the grid.
These types of ancillary services
could be of great value in future Connection to E.ON 22-kV Grid
complex power systems. Balancing
could be managed locally with sev­
22 kV
eral distributed LESs during peak
load hours. In this scenario, the E.ON Control Room
10 kV
LESs provide anciliary services to
EMS PCC
the overlying grid by adapting their
power consumption and genera­
tion. This would increase the over­
all grid-hosting capacity by keeping 150 Customers
the voltage levels within limits, Across Five
despite the fast, varying power Secondary
flows given by the intermittance of Substations
the RESs.

Intentional Island Mode

In the intentional island mode, the
PCS controller running in the volt­
480 kW EMS
age-source-inverter mode operates
G 500 kW
as the grid-forming unit in droop
for both power as a function of
frequency and reactive power as
EMS 442 kW
a function of voltage and allows E.ON Control
(Main)
the BESS to seamlessly transition Room
between the grid-connected and BESS
Controller
islanded operating modes. To phys­
ically island the system, the EMS
measures the real and reactive
BMS
power across the PCC and sets the
voltage and frequency targets for
the BESS such that real and reactive Figure 5. The system and component overview in Simris.

IEEE Elec trific ation Magazine / D ec emb er 201 8 53

 50.2
Frequency (Hz) Frequency (Hz)
EMS Trimming
Frequency Back to 50 Hz BESS Power (kW)
50.1 Backup Generator Power (kW)
183-mHz Customer Consumption (kW)
Increase in Wind Turbine Plus PV Power (kW)
50 Frequency 224-mHz
During Backup Drop in
Generator Frequency
49.9 Onloading EMS Trimming
Frequency Back to
50 Hz
49.8 550
11:06:00 11:07:00 11:08:00 11:09:00 11:10:00 11:11:00 11:12:00 11:13:00 11:14:00
Power (kW)

Inertia from Backup Generator Delivering 275

412-kW Step Due to Power in Response to Frequency Drop
Wind Turbine Tripping
Backup Generator Starting 0
and Going to Full Power Over
37 s (9.3 kW/s)
Customer Load Is
–275
Near Constant Over
Test Period
–550

Figure 6. The frequency and powers during the backup generator starting and the wind turbine tripping.

However, the control in the EMS is relatively slow, such

50.2 that a power imbalance allows the frequency and voltage
50.15 to transiently change before being trimmed back.
50.1 To reconnect to the grid, the EMS controls the voltage
Frequency (Hz)

50.05 and frequency to match the grid voltage and frequency,

50 i.e., it synchronizes the systems. Once within prescribed
49.95 limits, the EMS closes the circuit breaker.
49.9
In Figure 6, the system was physically islanded, operat­
49.85
ing on power from the wind turbine, the PV farm, the back­
49.8
up generator, and the BESS. At 11:08:30, the BESS SoC was
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
24:00

low, triggering the EMS to start the backup generator,

12 April 2018 which ran up to a controlled 350  kW. During the power
(a)
increase, the BESS allowed the frequency to increase along
its droop curve, and the EMS subsequently trimmed the
50.2 frequency back to nominal. At 11:11:40, the wind turbine
50.15 tripped, and the BESS allowed the frequency to reduce
50.1
Frequency (Hz)

along its droop curve. During this frequency reduction, the

50.05 backup generator power contribution increased slightly
50
due to the system inertia. The EMS subsequently updated
49.95
the target frequency to return the system to a nominal
49.9
50 Hz. Customer supply was maintained within statutory
49.85
limits throughout.
49.8
Longer-term testing comparing the frequency, voltage,
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
24:00

and total harmonic distortion of the system when is­­­

12 April 2018
landed to the same parameters for the local Nordic
power system shows that the LES had better power qual­
(b)
ity when islanded than when connected to the grid. Fig­
Figure 7. The frequency when islanded. (a) The intentional island ures 7–9 show the comparison for islanding testing on
mode and (b) the Nordic power system. 12 April 2018, were the system was islanded from 8:00 a.m.

54 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

 10.9 10.9
10.8 10.8
Voltage (kV)

Voltage (kV)
10.7 10.7
10.6 10.6
10.5 10.5
10.4 10.4
10.3 10.3
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
24:00

00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
24:00
12 April 2018 12 April 2018
(a) (b)

Figure 8. The voltage when islanded. (a) The intentional island mode and (b) the Nordic power system.

to 8:00 p.m. All figures show the results from the LES
when in the intentional island mode in (a) and the Nor­
dic power system in (b).
Battery Energy Storage Systems
In the Simris project, the BESS is a core element of the LES.
When islanded, it acts as the grid-forming unit and sets
both the voltage amplitude and frequency. The BESS
selected for the project is characterized by a rated power
of 800 kW and an energy capacity of 333  kWh and
Demand-Side Response
A final balancing element connected to the EMS is the
DSR platform. This platform allows various distribut­
ed sources of flexibility to be connected at the custom­
er’s premises.
The platform is capable of the following:
xx deliver a real-time data platform for monitoring
and control
Total Harmonic Distortion (%)

includes PCS functions to allow for islanding. 1.6

The BESS includes three key control components con­ 1.5
trolling the system operation:
1.4
1) the battery management system, responsible for the
SoC estimations and enforcing the battery system limits 1.3
2) the PCS inverter control, responsible for the high- 1.2
speed switching of the power electronics to manage 1.1
the different operating modes
1
3) the battery system controller, including a state machine
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
24:00
for different operating modes and manages transi­
tions between states. 12 April 2018
A summary of the key components is shown in Figure 10. (a)
The BESS control system state flow diagram is shown
Total Harmonic Distortion (%)

1.6
in Figure 11.
The different states are requested by the EMS, and the 1.5
BESS controller then implements the mode to the PCS. 1.4
The previously described control structure allows the 1.3
high-speed operation to be managed at a local level (invert­
1.2
er switching at 2.5 kHz), with the BESS controller control­
ling the local interfaces and operating modes and the EMS 1.1
managing the system demands at around the 1-Hz com­ 1
00:00
01:00
02:00
03:00
04:00
05:00
06:00
07:00
08:00
09:00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
24:00

munication update.
Short-circuit operation was tested by introducing sin­
gle- and three-phase faults to a low-voltage circuit pro­ 12 April 2018
tected by 63-A fuses while islanded. Figure 12 shows the (b)
system voltage for a three-phase fault while operating Line 1 Line 2 Line 3
on the BESS only, where the battery system terminal
voltage was reduced to 86% during fault clearance and Figure 9. The total harmonic distortion when islanded. (a) The inten-
the nominal voltage was restored within 200 ms. tional island mode and (b) the Nordic power system.

IEEE Elec trific ation Magazine / D ec emb er 201 8 55

 Encorp
Microgrid Other Communication
EMS
Assets Power
controller

Battery System (Samsung SDI) PCS Transformer

Master Loccioni
BMS Controller

BMS

TDE Macno

BMS
375 V : 10,700 V

TDE Macno

BMS

Seven Units Five Units

Figure 10. The overview of the BESS components. BMS: battery management system.

110
Alarm
(% of Nominal Voltage)

Grid Voltage
105

Resetting 100
Voltage

95
Disconnected 90
85
Connecting Disconnecting 80
0.2 0.3 0.4 0.5 0.6
Time (s)
Connected
Figure 12. The voltage during short circuit testing.
Stopping
simplify integration of new asset types to expand the
xx
Running Csi Islanding Running Vsi asset base
visualize key performance indicators and monitor
xx
energy consumption, steering signals, and user inter­
action (peer-to-peer platform).
Synchronizing As shown in Figure 13, the DSR technologies used in
the Simris project are residential PVs and battery solu­
Figure 11. The BESS state controller. tions, steerable heat pumps, hot tap-water boilers, and an
e-mobility charging station.
provide control loops to optimize energy usage in the LES
xx
offer real-time connectivity to assets and external
xx Photovoltaics and Battery Systems
platforms PVs and battery systems enable the customers to be
provide open connectivity to extend services to the sys­
xx more self-reliant and reduce feed-in from the grid. The
tem operator and flexibility marketplaces rooftop-installed PV panels produce energy in the daytime,

56 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

and the battery allows excess energy to be stored. If the
overall load demand exceeds the PV generation, the bat­
tery is discharged, allowing the stored energy to be used
for auxiliary grid-balancing services. Due to the increasing
market penetration for this solution and the versatility of
the battery, it is expected to become a vital part of the
future energy system.
One general concern regarding battery storage in MGs is
how much flexibility it will be able to provide. At a brief
glance, the PV and battery system seems to be a great
source of flexibility for the LES. However, when the LES is
characterized by an excess of energy, due to high power
generation in the central PV power plant, the batteries are
likely to be charged by household solar-power generation
alone since the residential PV would also provide the max­
imum output. Hence, there might be little additional stor­
age capacity left to use for grid-balancing services through
the DSR. Likewise, a lack of solar-power production yields
an energy deficiency in the system, while, at the same
time, it is unlikely for the batteries to have stored enough
energy to support the grid during those times.
One of the DSR strategies is based on residential battery
power control. Every 5 min, a set
point is sent from the DSR platform
to each of the residential batteries.
Intelligence
An example of one of the control ap­­­
Module
proaches can be seen in Figure 14,
where the residential battery power
set points to be sent out by the DSR
platform are a result of the current
main battery’s SoC. The control strat­
egy is intended to support the oper­
ation of the main battery in case it
suffers from low or high SoC levels.
In the former case, the residential Smart
batteries are requested to discharge; Meters
in the latter case, the residential bat­
teries are requested to charge. Like­
wise, i.e., between a 30 and 70% Figure 13. The DSR platform.
main-battery SoC deadband, the
residential batteries can operate in
their usual operation modes as de­­ Maximum
fined by the customers. Charging Power
period of time without the indoor temperature dropping
more than 1 °C from the initial temperature. For these rea­
sons, electrical heat pumps offer an opportunity to help
balance the power demand and supply in the grid. An esti­
mation of the overall Swedish flexibility potential in heat
pumps is in the size of gigawatts, thus providing an excel­
lent foundation for scalability.
Heat pumps can provide both negative and positive
flexibility since it is possible to both increase and decrease
heating demand. However, the available flexibility largely
depends on the outdoor temperature, as represented in
Table 2. The specific temperatures depend on the heat-
pump technology installed and other factors such as
wind, sun, heat generation from the internal electrical
loads and people.
During warm weather, there is a theoretical potential for
flexibility, but as temperatures exceed 15 °C, the heat pump
is often turned off because there is enough thermal energy
to keep a comfortable indoor climate. At colder tempera­
tures, approximately –15 °C and below, the heat pump most
likely operates at maximum capacity, unless it is overdi­
mensioned, and therefore it is only possible to decrease the
Peer-to-Peer
MG Controller
Platform
DSR Platform
PV + Hot Electric
Heat
Battery Tap-Water Vehicle
Pumps
Systems Boilers Charging

Deadband
Residential Battery Power

Heat Pumps Charge

Electrical heat pumps are a com­
mon source of heating for Swedish Main Battery SoC
households and use high amounts
0 10 20 30 40 50 60 70 80 90 100
of electricity as compared to other (%)
appliances in a residence. In Swe­
den, approximately 60% of the total Discharge
household energy consumption is
allocated to meet heating demands. Maximum
In addition, due to the thermal Discharging Power
inertia of buildings, it is possible to
turn a heat pump off for a long Figure 14. The residential battery control.

IEEE Elec trific ation Magazine / D ec emb er 201 8 57


Table 2. The flexibility potential based on the
outdoor temperature.
Outdoor Temperature Flexibility Potential
215 °C No flexibility
0 °C Increase/decrease heat demand
1-15 °C Decrease heat demand
power output. In between, it is possible to both increase
and decrease the demand. By how much this can be done
depends on the current outdoor temperature.
Rather than a signal based on the power in kilowatts,
the DSR platform creates a steering signal in percentages,
based on the main battery SoC in a range between 100%
(maximum power increase) and –100% (maximum power
decrease). To ensure that customer comfort is always
maintained, a boundary control is implemented, disabling
the DSR control if the indoor temperature deviates by
more than ! 1 °C from the customer-set indoor comfort
temperature range.
Hot Tap-Water Boilers
Hot tap-water boilers offer an inexpensive opportunity
to increase flexibility in a household. However, since the
energy is stored as heat and cannot be converted back
to electricity, it is less versatile than a battery. During
excess energy production in the LES, the water tanks
can be heated, reducing the need to heat water at other
times such as when the system has a deficit of energy
production. One might also think about the option to
stop the hot tap-water boilers from heating as usual, in
a way to not disturb customer comfort. However, since
the heating patterns of a boiler tend to be highly spo­
radic, the risk of a negative impact on customer com­
fort, by not having hot water when needed, is considered
100% of
Maximum
Charging Current
Charging Current
66% of
Maximum
Charging Current
33% of or 21 A
Maximum
Charging Current
or 10 A
No
Charging
0 20 50 80
Main Battery SoC (%)
Figure 15. The electric vehicle charging control.
58 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018
within the project, for which reason a remote control
will only manage heating water during excess energy
generation. The introduced control system, therefore,
never affects the customer’s comfort.
By monitoring household consumption patterns, the
hot tap-water boiler device will know when and how
much water will be consumed and will minimize the boil­
er’s water tank heat losses by preheating it just in time.
For this purpose, temperature measurements to the water
tank’s inlet and outlet pipes are necessary. The local hot-
tap-water boiler’s steering device will receive steering sig­
nals from the DSR platform using smart meter technology,
and thus the water tanks will be managed by remotely
setting the heating mode.
Electric Vehicles
Electric vehicles connected to a smart charging station
can be a source of flexibility to support the penetration of
renewable energy, by controlling the steerable electric
vehicle supply equipment (EVSE) current output in res­­
ponse to the fluctuating balancing requirements of the
system. The EVSE is integrated in the project at Simris and
connected to the DSR platform. The DSR platform is able
to send periodic steering signals to the EVSE via a back
end, limiting the amount of energy the charging station is
providing over time. A graphical view of the control logic
for this example is shown in Figure 15.
RWTH’s Contribution to the Project
with the Study and Simulation
of an Advanced Control Strategy
RWTH Aachen University closely collaborated with the
E.ON team in defining alternative ways to control the MG
systems. The project activities of RWTH Aachen University
are aimed at providing a long-term evaluation of the EMS
optimization for MGs in general. A special focus is the abil­
ity to switch to and continuously stay in the islanded
operation mode. Furthermore, the effects of the active
control on the predictability of the MG behavior are exam­
ined. This includes the following aspects:
xx Developing a central EMS and showing its viability for
MGs other than the Simris LES but based on some of
the Simris current measures. In particular, the full inclu­
sion of an exact relaxation of the power-flow equations
or 32 A is used to make best use of reactive-power capabilities
and take equipment constraints into account.
xx Comparing the effects of different operational objectives
on a larger-scale simulation based on the developed test
system and the EMS. A whole year of simulated opera­
tion, based on measurements obtained from the Simris
field trial site, is evaluated. This also allows us to show
some tradeoffs to be made and critical disadvantages in
100 the focus on certain objectives.
xx Performing a comprehensive analysis of the maxi­
mum possible time frame for the islanding operation
under different operational objectives.
 Developing data science-based forecasting methods to
xx
predict the behavior of the generation and loads in
an MG.
xx Analyzing the behavior of an MG and its EMS when it is
only seen as a single point of load connected at the
PCC. The main focus of the analysis is to show how the
effects of active control by an EMS deteriorates with
reduced forecasting accuracy. It is analyzed how differ­
ent operational objectives also result in different behav­
ioral patterns, which pose a varying degree of difficulty
in the prediction.
One of the possible approaches for the MG control is a
hierarchical control structure as used in the Simris project.
The optimization problem is solved by a single entity for
all parameters. Therefore, all relevant data have to be
gathered by the central system, and schedules are then
distributed to all entities after the computation is finished.
The RWTH Aachen University considers these controls’
capabilities and their master/slave communication model.
A general form of an optimization-based EMS is
described by a set of generic equations. A desired objective
function C is minimized over a certain period T with a
time-step length Tt considering the system state x (t) at a
time t (e.g., system voltages, current, stored energy, and
others), the inputs u (t) at a time, which in our case are the
real- and reactive-power references and set points, and the
cost or fitness f (x (t), u (t)) at the time t. This objective func­
tion can be subject to various boundary conditions includ­
ing the following:
xx model equations (e.g., power-flow equations, equip­
ment constraints, and so on)
xx energy limits
xx power limits
xx voltage limits
T
argmin u C (x (t), u (t)) = / f (x (t), u (t))
t=1
In EMSs, a simple distinction can be made between sub­
classes of convex problems solved with classic-, gradient-,
or Hessian-based methods and nonclassic methods that
may not be convex and include heuristics. If peak shaving
of, e.g., power consumption is considered ­(Diekerhof et al.
2017), objectives tend to be quadratic, leading to a quadratic
problem. Recently, second-order cone programming (Yan
et al. 2017) and semidefinite programming (Peng and Low
2015) formulations have also become popular, as they allow
the exact relaxation of grid constraints for radial networks.
Another important aspect is whether binary variables (or
integer variables, in more general terms) are considered.
These require additional mixed-integer programming solv­
ing methods that increase computation time and make
distribution of the control problem challenging to solve in a
limited time horizon. Mixed-integer formulations are often
used because they allow an easy inclusion of varying
behavior patterns for single resources. This becomes
relevant for the inclusion of generator startup constraints,
piece-wise linear approximations of nonlinear cost func­
tions, or reconfigurable switches in the network.
An optimization-based EMS can be operated both in an
open- and closed-loop form referring to control theory ter­
minology. Early approaches usually adopt the concept of
day-ahead scheduling, i.e., a single optimization and imple­
mentation of set points, based on the prediction for a whole
day (Chaouachi et al. 2013). Most concurrent approaches,
however, make use of a receding horizon implementation,
which includes a feedback loop for MG states and reopti­
mizes the behavior at every time step. The optimization is
performed for a discrete time-step length that requires
averaging of behavior for that period, forecast values of
future behavior, and possibly simplified models.
The proposed developed controller uses the concept of
model predictive control (MPC) that has become an inter­
esting approach for energy management purposes in MGs
in recent years. MPC provides set points for the energy stor­
age charging/discharging power as well as set points for the
curtailment of renewables and their reactive-power output.
The state of the system consists of the SoC of the energy
storage devices. One time step within the calculations is
1 h. MPC is designed to minimize the exchange of energy
with the main grid, covering demand of the customers and
fulfilling the energy balance. This approach could potential­
ly prolong the islanding mode of the MG but was shown to
have an adverse effect on islanding based on actual imple­
mentation (Mildt et al. 2018).
Model Predictive Control
MPC is an advanced control technique for multivariable
control problems. MPC uses an internal system model to
generate predictions of its future behavior. The designed
MPC finds an optimal plan for running the assets (by
defining control set points) for each time step over the
prediction horizon, starting at the current time step. The
optimization is based on predictions of the upcoming
demand, production from RESs, and prediction for the
state of the system. A general depiction of the methodolo­
gy is presented in Figure 16.
For the optimization, only the first control input is
implemented, and subsequently the time horizon is shift­
ed by one time step. At the next time step, the new state of
the system is measured or estimated, and a new optimi­
zation is performed while considering the new informa­
tion. This approach is called a rolling horizon control.
The control horizon can be chosen to be lower than
the prediction horizon within the formulation to deal
with the time lag of computational-intensive calculations.
If the control horizon is set to be lower than the predic­
tion horizon, it could be assumed that the last calculated
control variables remain the same until the end of the
prediction horizon. The predictions needed for the appli­
cation of MPC could be assumed as perfect in the MPC
statement, i.e., not affected by errors, as the consistent
IEEE Elec trific ation Magazine / D ec emb er 201 8 59
 set point recalculation will already provide a means of
disturbance rejection. Extensions to robust or stochastic
optimization approaches (Diekerhof et al. 2018) are, how­
ever, possible.
In the current LES Simris activities and beyond, RWTH
Aachen University is evaluating a combination of the fol­
lowing objective functions (compare Table 3):
xxthe minimization of the summed costs of all assets, J c
xxthe minimization of the energy losses, J l
xxthe maximization of local energy use to satisfy the
local self-consumption desire of certain local energy
systems, J p
xxthe minimization of the grid-connected time while
avoiding consecutive connection and disconnection
from the grid, J d .
These four objective functions are evaluated in
three cases:
Past Forecast Horizon
Measured Behavior
Predicted System Behavior
Case A: Economic value—minimizing the operation
x 
costs for the MG, which combines the operation cost
minimization and the loss minimization.
x Case B: Local use of energy, which is a weighted
function of the exchanges with the power grid and
the losses.
x Case C: Potential islanding time (PIT) maximization,
which sets the PIT as a primary objective whereas the
secondary objective is the pure cost minimization.
The evaluation of one year of operation in a test (Fig­
ure 17) shows how the different operational strategies
impact operation costs, energy exchange, and line loss­
es. Obviously, direct minimization of operational cost
(case A) or energy exchange with the main grid (case B)
allowed the best results concerning their respective
objective. Yet, it is interesting to note how maximization
of the PIT (case C) achieves similar results in terms of
energy exchange with the main
grid, compared with a cost opti­
mization, while incurring signifi­
Future cantly higher energy losses. This
indicates high currents, specifical­
ly fast charging and discharging
events of energy storage in prepa­
ration of an islanding event, as
further explained by Mildt et al.

Implemented Input u (k ) Resource Forecast

For applying the control strategy as
Past Inputs described previously, PV and wind
Optimized Inputs production as well as the consumer
load needs to be forecast.
k
To create forecasts for the un­­
Current Timestep k+1 k+2 k+3 k+4 k+5 ... k+N
known variables, the RWTH Aachen
Time
University applies state-of-the-
art recurrent neural network in a
Figure 16. The MPC prediction and control horizon. encoder-decoder setup to derive
these values. This RNN setup (Fig­
ure 18) uses historical values in the
Table 3. Objective functions evaluated by RWTH Aachen encoder as a starting point to pro­
University. duce an outlook into the future in
Operational Cost Minimization Loss Minimization the decoder. RNNs are based on
long short-term memory cells that
T T
possess cyclic, self-feeding connec­
J c = / cost of power import at time t Jl = / /P i, j,loss
t=1 t = 1 (i, j) ! f tions. These cycles are capable of
T
capturing time-series dependen­
- / revenue of power export at time t
t=1 cies and their dynamic behavior
T

+ / cost of all assets at time t over a sequence of time steps. They

t=1 include exogenous features at each
forecast time step to further im­­
Minimization of Power Exchange with the Maximization of Possible
Main Grid Islanding Time After tisland prove prediction accuracy. For the
upcoming 24  h, this type of fore­
T T
casting algorithm is capable of pre­
J p = / Pimp (t) - Pexp (t) Jd = / d PCC (t)
t=1 t = t island dicting wind and PV production as
s.t. d PCC (t + 1) $ d PCC (t) well as electrical loads. Exogenous
variables for these forecasts comprise

60 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

 meteorological values in applicable combinations, e.g.,
× 104
10 temperature, air pressure, solar irradiation, wind speed
and direction, cloud coverage, and so on. They are, collec­
Cumulative Cost (€)

tively, assumed to be known for the forecast time steps,

suggesting the availability of highly accurate meteorologi­
5 cal forecasts. In a real-life application, the forecast accura­
cy of the desired target variables is expected to degrade
slightly with ever-so-slightly imperfect meteorological
forecasts. Furthermore, the developed approach considers
0
0 2,000 4,000 6,000 8,000 calendar variables, such as weekday, time of day, month,
Time (h) national holiday, and others, to account for consumer
(a) behavior for the electrical load forecasts.
Figure 18 shows the setup of the RNN architecture
250
Energy Exchange (MWh)

(Wilms et al. 2018). As shown, an encoder-decoder setup

200 is used that takes the historical sequence to encode this
within a so-called hidden state (a collection of memory
150
vectors), which then provides the starting point for the
100 decoder that derives the forecasts. In Figure 18, l pres­
ents labels (i.e., target values), y indicates the forecast
50
(output) value, and f  denotes the respective set of exoge­
0 nous variables.
0 2,000 4,000 6,000 8,000
The encoder as well as the decoder uses a combination
Time (h)
(b)
of the label (forecast value) of the previous time step and
the features of the current time step. This lagged use of
8 the previous time step for inferring the current time
step also focuses the forecasting algorithm to exploit the
Line Loss (MWh)

6
strong auto-correlation that is inherent of the time-series
of question.
4
Using this algorithm, the RWTH Aachen University was
2 able to produce reasonably accurate results for 24-h fore­
casts on synthetic load data from Simris. Afterward, these
0 forecasts were used as inputs to the MPC. It turned out
0 2,000 4,000 6,000 8,000
that the evolved forecasting architecture outperforms
Time (h)
similarly employed encoder-decoder RNN-based forecast­
(c)
ing techniques.
A B C On this basis, the simulation results of RWTH Aachen
University showed that the combination of a carefully
Figure 17. The annual cumulation of (a) operational cost, (b) energy
exchange with the main grid, and (c) line losses employing operation designed MPC approach, as described previously, in com­
strategies A–C. bination with a tailored forecast algorithm could provide

y : Outputs yt yt +1 yt +2 yt +3
f : Features, i.e., Exogenous Inputs
Encoder Hidden Decoder
State
Cell Cell Cell Cell Cell Cell Cell Cell

lt–4ft–3 lt–4ft–2 lt–4ft–1 lt–1ft eos eos yt ft+1 yt +1ft+2 yt +2ft+3

Historic Values Forecasted Values

Figure 18. The setup of the forecasting algorithm.

IEEE Elec trific ation Magazine / D ec emb er 201 8 61

 an improvement (self-consumption increase) of more
than 10% (Bogdanovic et al. 2018). Theses initial simulation
results will be validated with tests performed during the
remainder of the project.
Conclusions
E.ON and its partners believe that LESs could become an
important and powerful opportunity for the energy indus­
try. LESs could guarantee a cost-effective integration of
renewables in those cases where conventional solutions
are clearly not the appropriate ones. The work on LESs
continues, and the development of future solutions is
being built upon the lessons learned in the first years in
that area.
New regulation coming from the European Commis­
sion’s Clean Energy Package for all Europeans (published
in 2016) will provide an opportunity for the European
member states to define a national regulatory framework
that will allow LESs to become the technical enablers for
the creation of local energy communities. Such local
energy communities will support the transition toward a
European energy system that combines sustainability
and security of supply with competitiveness and citizens
at its core.
Acknowledgments
We gratefully acknowledge the funding received from
the European Union’s Horizon 2020 research and innova­
tion program for project InterFlex under Grant 731289 as
well as our suppliers and main partners Encorp, Loccioni,
and Holtab.
For Further Reading
A. Bidram and A. Davoudi, “Hierarchical structure of mi­­
crogrids control system,” IEEE Trans. Smart Grid, vol. 3, no. 4, pp.
1963–1976, May 2012.
J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. de Vicuna, and
M. Castilla, “Hierarchical control of droop-controlled AC and
DC microgrids: A general approach towards standardiza­
tion,” IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 158–172, Jan.
2011.
L. I. Minchala-Avila, L. E. Garza-Castan, A. Vargas-Martnez,
and Y. Zhang, “A review of optimal control techniques applied
to the energy management and control of microgrids,” Proce-
dia Computer Sci., vol. 52, suppl. C, pp. 780–787, 2015.
M. Diekerhof, F. Peterssen, and A. Monti, “Hierarchical dis­
tributed robust optimization for demand response services,”
IEEE Trans. Smart Grid, vol. PP, no. 99, pp. 1, 2017. doi: 10.1109/
TSG.2017.2701821.
D. Yan, P. E. I. Wei, C. Naishi, G. E. Xianjun, and X. Hao, “Real-
time microgrid economic dispatch based on model predictive
control strategy,” J. Modern Power Syst. Clean Energy, vol. 5, no. 5,
pp. 787–796.
Q. Peng and S. H. Low, “Distributed algorithm for optimal
power flow on an unbalanced radial network,” in Proc. 54th
IEEE Conf. Decision and Control (CDC), 2015, pp. 6915–6920.
A. Chaouachi, R. M. Kamel, R. Andoulsi, and K. Nagasaka,
“Multiobjective intelligent energy management for a micro­
grid,” IEEE Trans. Ind. Electron., vol. 60, no. 4, pp. 1688–1699,
Apr. 2013.
62 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018
H. Wilms, M. Cupelli, and A. Monti, “Combining auto-
regression with exogenous variables in sequence-to-sequence
recurrent neural networks for short-term load forecasting,”
in Proc. IEEE Int. Conf. Industrial Informatics (INDIN), Porto, Por­
tugal, 2018, pp. 673–679.
M. Diekerhof, S. Schwarz, and A. Monti, “Demand-side
management—Recent aspects and challenges of optimization
for an efficient and robust demand-side management,” in Clas-
sical and Recent Aspects of Power System Optimization, A. F. Zobaa,
S. H. E. Abdel Aleem, and A. Y. Abdelaziz, Eds. Amsterdam, The
Netherlands: Elsevier, 2018, pp. 331–360.
M. Bogdanovic, H. Wilms, M. Cupelli, M. Hirst, L. Hernán­
dez and A. Monti. “Interflex—Simris—Technical management
of a grid-connected microgrid that can run in an islanded
mode with 100% renewable generation,” in Proc. CIRED Work-
shop, Ljubljana, Slovenia, 2018, pp. 1–4.
D. Mildt, M. Cupelli, A. Monti, and R. Kubo, “Islanding time
evaluation in residential microgrids,” in Proc. IEEE PES-IAS
Power Africa Conf., Cape Town, South Africa, 2018.
T. Logenthiran, D. Srinivasan, A. M. Khambadkone, and H.
N. Aung, “Multiagent system for real-time operation of a
microgrid in real-time digital simulator,” IEEE Trans. Smart
Grid, vol. 3, no. 2, pp. 925–933, June 2012.
The European Commission. (2016). Clean energy for all
Europeans. Brussels, Belgium. Tech. Rep. COM(2016) 860 final.
[Online]. Available: https://ec.europa.eu/transparency/regdoc/
rep/1/2016/EN/COM-2016-860-F1-EN-MAIN.PDF
Biographies
Henning Wilms (hwilms@eonerc.rwth-aachen.de) is with
the E.ON Energy Research Center, RWTH Aachen Universi­
ty, Germany.
Dominik Mildt (dmildt@eonerc.rwth-aachen.de) is with
the E.ON Energy Research Center, RWTH Aachen Universi­
ty, Germany.
Marco Cupelli (mcupelli@eonerc.rwth-aachen.de) is
with the E.ON Energy Research Center, RWTH Aachen Uni­
versity, Germany.
Antonello Monti (amonti@eonerc.rwth-aachen.de) is
with the E.ON Energy Research Center, RWTH Aachen Uni­
versity, Germany.
Peder Kjellen (peder.kjellen@eon.se) is with E.ON,
Malmö, Sweden.
Thomas Fischer (thomas.fischer1@eon.com) is with
E.ON, Essen, Germany.
Demijan Panic (demijan.panic@eon.se) is with E.ON,
Malmö, Sweden.
Michael Hirst (mike.hirst@eon.com) is with E.ON,
Essen, Germany.
Eugenio Scionti (eugenio.scionti@eon.com) is with E.ON,
Essen, Germany.
Sebastian Schwarz (sschwarz@eonerc.rwth-aachen.de)
is with the E.ON Energy Research Center, RWTH Aachen
University, Germany.
Paul Kessler (paul.kessler@eon.com) is with E.ON,
Essen, Germany.
Luis Hernández (luis.hernandez@eon.com) is with
E.ON, Essen, Germany.

By Mashood Nasir, Hassan Abbas Khan, Nauman Ahmad Zaffar,
Juan C. Vasquez, and Josep M. Guerrero
Scalable Solar
dc Microgrids
here is a worldwide focus on the
T
electrification of developing regions, as is
evident from the sustainable develop-
ment goals (SDGs) of the United Nations.
In particular, the aim of SDG 7 is to ensure
universal access to affordable, reliable, sustainable, and
modern energy services for all by 2030. Because of these
sustained efforts, more than 1 billion people worldwide
have gained access to electricity since 2000. During
this course, the electrification architecture of developing
Digital Object Identifier 10.1109/MELE.2018.2871297
Date of publication: 16 November 2018
2325-5987/18©2018IEEE
On the path to
revolutionizing
the electrification
architecture
of developing
communities.
footage frim, inc.
regions has taken different forms, ranging from exten-
sive utility grid extensions to limited off-grid solutions.
Off-grid solutions generally offer cost-effective and
lower up-front cost alternatives in comparison to utility
grid extensions and are deemed more suitable for
developing economies. In developing economies, these
off-grid electrification solutions have evolved from in­­
dividual solar home-based systems to community-based
microgrids in pursuit of achieving higher efficiency and
reliability on a village scale. Community grids responsi-
ble for the electrification of developing regions are fur-
ther categorized based on architecture, type of generation,
and mode distribution.
IEEE Elec trific ation Magazine / d ec emb er 201 8 63
 The natural abundance of solar
resources in most of the developing
regions in Southeast Asia and Africa,
coupled with the diminishing costs of
solar photovoltaic (PV) panels, advance-
ments in the battery industry, and the
advent of power electronics technology,
has made solar PV generation an attrac-
tive alternative to conventional electrici-
ty generation. Compared to traditional
ac distribution-based microgrids, dc
microgrids, when implanted with dc
generation, distribution, and loads, ex­­
hibit significantly higher efficiency due
to the omission of unwanted ac-to-dc or dc-to-ac conver-
sions. Due to these advancements, PV-based dc microgrids
have paved the way in practical deployments and are
regarded as the most optimal electrification solution for
developing economies.
Despite all of these advancements, the world is not
currently on track to meet the global objectives of SDG 7,
and today the number of people without electricity is
more than what it was in 2000. Although the convention-
al schemes for rural electrification are being largely
deployed as a stopgap measure for energy poverty eradi-
cation, owing to their limited potential, these schemes
are not sufficient. With the growing population and asso-
ciated electrification requirements, a highly robust,
systematically efficient, technologically advanced, eco-
nomically feasible, and widely adoptable electrification
solution is needed that can be scaled in a bottom-up
manner and can support microfinancing for enhanced
rates of electrification. This article highlights the need for
rural electrification and provides an overview of various
schemes to do so. A detailed analysis of various architec-
tures of solar PV-based dc microgrids existing in practice
Figure 1. A primary school in Naran Valley, Pakistan, without access to electricity and
basic education facilities.
64 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018
Reliable access to or literature is presented with their
respective pros and cons. These archi-
electricity and its tectures mainly include centralized
architectures and partially distributed
consumption rates and highly distributed architectures
are key indicators as discussed in subsequent sections.
The analysis concludes that scalable
for the socioeco- solar dc microgrids have the tendency
to offer a ­viable solution for future
nomic standing of rural electrification implementations
any community. and can be regarded as a way forward
to achieve the objective of universal
electricity access.
The Need for Rural Electrification
Reliable access to electricity and its consumption rates are
key indicators for the socioeconomic standing of any com-
munity. The significant availability of electricity, even at
very basic levels, is extremely crucial for human well-
being and social resources development. On the contrary,
the lack of electricity hampers basic human rights like
access to clean water, health-care delivery, education facil-
ities, and proper lighting, thereby enhancing poverty and
significantly deteriorating the quality of life. According to
the International Energy Agency, around 1 billion people
throughout the world, which constitutes nearly 13% of the
global population, lack access to electricity. It is also esti-
mated that around 87% of the people lacking access to
electricity are the residents of rural areas. According to the
statistics of the United Nations Department of Economic
and Social Affairs, around 90% of children in Sub-Saharan
Africa are studying in unelectrified schooling facilities,
and another 27% of village schools in developing Asia,
including India and Pakistan, do not have access to elec-
tricity. One such remote school near the valley of Naran
in Pakistan is shown in Figure 1, where students of a pri-
mary school are deprived of the basic edu-
cation facilities, such as access to well-lit,
heated, and ventilated classrooms. More-
over, basic computing and printing facili-
ties considered key to advanced learning
are virtually absent due to the unavailabili-
ty of electricity.
The access to water is the same for
these developing regions, because nearly
1 billion people living in these developing
communities do not have access to clean,
safe water for drinking and irrigation pur-
poses. According to an estimate by the
United Nations, due to the unavailability
of electricity-driven water pumps, the
inhabitants of Sub-Saharan Africa alone
spend 40 billion hours per year collecting
water, which is equivalent to France’s
entire annual worth of workforce labor.
Also, due to the lack of electricity, more
than 1 billion people worldwide rely on unhealthy
resources, such as kerosene for lighting and wood-based
stoves for cooking purposes. According to estimates by the
World Bank, breathing kerosene fumes is highly carcino-
genic and as dangerous as smoking two packs of ciga-
rettes each day. According to a report by National Geographic,
cookstove smoke is extremely life threatening, and
around 3.5 million people die each year due to the respira-
tory diseases caused by indoor pollution of wood/bio-
mass-based stoves (approximately three times the
mortality rate caused by malaria and 2.3 times the mortali-
ty rate caused by HIV/AIDS).
The provision of electricity to these inhabitants could
help reduce high fatality rates and also contribute to
improved standards of living, in­­cluding better health and
education, and could also provide more agricultural, indus-
trial, and employment opportunities. In addition, the electri-
fication of these regions through green and environmentally
friendly energy re­­sources will help in reducing climate
change and deforestation rates. Along with social benefits,
there are remarkable business opportunities in the energy
markets of these developing regions due to the global
focus on energy poverty eradication and associated initia-
tives, such as the programs Sustainable Energy for All and
Lighting Africa. Because the human development, eco-
nomic stability, and social growth of these regions are cou-
pled with access to electricity, electrification of these
developing regions is the highest priority and will enable
the socioeconomic benefits associated with the easy
access and reliable availability of electricity.
Current Status and Brief Overview of the
Existing Practices for Rural Electrification
More than 1 billion people have gained access to electrici-
ty since 2000, with around 220 million between 2010 and
2012. The number of people attaining power has been
accelerating since 2010 to around 118 million each year.
These efforts are more pronounced in developing Asia,
where around 870 million people have been provided with
electricity since 2000, with India as a major contributor for
providing access to roughly 30 million people each year in
between 2010 and 2016. For the very first time in recent
years, electrification rates in Africa are on par with the
growing population. Remarkable efforts in Bangladesh,
Ethiopia, Kenya, and Tanzania have led to expanded elec-
tricity access by at least 3% of those populations annually
between 2010 and 2016.
The main source of electrification during this period has
been the extension of utility and national grid interconnec-
tion of these remote villages, with a large dependence on
fossil fuel. Electrification via grid expansion requires the
deployments of mega projects, including building new
power plants and laying long-distance transmission lines.
For developing and underdeveloped economies, these large-
scale developments are generally constrained by the limita-
tions of funding resources.
There has also been a paradigm shift toward the adop-
tion of environmentally friendly renewable energy resourc-
es for off-grid rural electrification. This shift in preference
is due to a number of factors, such as
xx the constant depletion of fossil fuel
xx increasing awareness about the hazardous impact of
fossil fuel on the environment
xx rapidly decreasing prices of re­­newable energy tech-
nologies
xx advancement in microgrid technology offering a cost-
effective alternative in comparison to grid extension.
During the last five years, a positive trend has been
seen in renewable-based off-grid rural electrification, and
approximately 6% of the new access connections are based
on renewable energy resources. Out of all renewable ener-
gy resources, solar energy has become more appealing due
to the natural availability in most of the underdeveloped
areas (most regions in Southeast Asia and Africa receive
abundant sunlight, i.e., above 5.5 kWh/m 2 /day) and grad-
ually decreasing prices of PV panels and batteries.
The adoption of solar energy for off-grid electrification
has evolved from stand-alone solar home systems (SHSs)
to microgrid-based implementations. SHSs, such as the
one shown in Figure 2, have been implemented in many
developing countries as a stopgap measure to provide resi-
dents with basic electricity. These schemes are highly effi-
cient because the generation, distribution, and use all are
in dc form, and there are no losses in the form of ac-to-dc
or dc-to-ac conversions. Also, the design is simpler and
cost-effective for a limited level of electrification. These
systems generally provide between a few watts to a few
tens of watts, enough to run one or two light-emitting
diodes (LEDs) along with a mobile charging unit and a dc
fan for an average rural house. As a standout example, in
Bangladesh alone, 3 million SHSs were installed by 2014,
Bidirectional Unidirectional
Power Power
Battery dc Loads
Figure 2. A schematic diagram of a PV-based SHS.
IEEE Elec trific ation Magazine / d ec emb er 201 8 65
 and this is growing daily. The infra-
structure Development Company
(IDCOL), run by the government of
Bangladesh, has reported the installa-
tion of 4.12 million SHSs in remote
areas through May 2017, from which
18 million people, i.e., 12% of the total
population, have been given access to
electricity. The projected target of
IDCOL is to install 6 million SHSs by
2021. The SHS technology is cost-ef­­
fective and relatively easy to deploy in
comparison to the grid extension alter-
native. However, these stand-alone
solutions are suboptimal, as, without
resource sharing, they do not take
advantage of electricity-usage diversi-
ty at a village scale. If the power pro-
duced by an individual household is
higher than its local requirements,
then extra power is wasted after the
battery is fully recharged. Moreover,
due to the unavailability of a resource-sharing feature,
these schemes have limited electrification capabilities and
are not feasible for something demanding like water fil-
tration 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 to the significant improvement in
terms of quality of life.
The Advent of Solar PV-based dc Microgrids
Although an SHS provides a low up-front cost and rela-
tively simpler off-grid electrification solution, there are
several limitations to this approach. It cannot support
larger community loads due to prohibitively large solar
panels and storage requirements for rural occupants in
developing regions. Even with the smaller systems, the
levelized cost of electricity (LCOE) is generally high due to
lack of resource-sharing capabilities. Alternatively, wind/
solar/fossil fuel-based islanded microgrids are becoming
very popular for the rural electrification of developing
regions due to their ability to support electrification beyond
subsistence-level living and extract the benefit of usage
diversity at a village scale. Three major aspects of microgrid
design are critical and need to be optimized for making
them the best fit for the electrification of remote commu-
nities: 1) generation technology, 2) mode of distribution,
and 3) architecture for the placement of generation and
storage resources.
Conventional resources of generation, including fossil
fuel-based generation and, in particular, diesel-based gen-
eration systems, result in carbon emissions and are not
considered an attractive solution for electrification due to
their adverse effects on the environment. Moreover, the
LCOE and operation costs for such diesel-based electricity
66 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018
Due to their generation systems are higher and
not viable for low-income communi-
inherent simplicity, ties. During the last two decades, re­­
newable energy technology has gained
higher power worldwide interest as an effective
quality, enhanced alternative to reduce dependence on
fossil fuels and avoid their adverse
efficiency, and effects on climate change. Therefore,
renewable energy resources, especial-
straightforward ly wind and solar energy generation,
controllability, are being adopted by microgrid practi-
tioners due to their green and envi-
dc microgrids are ronmentally friendly nature. Among
all other renewable technologies,
preferred over installations based upon solar energy
ac microgrids for extraction using PV systems are more
successful because of the natural
rural electrification availability of sunlight, relatively sim-
pler schemes of installation, environ-
applications. mentally friendly nature, and noise-free
operation. The consistent reductions
in PV panel prices, feed-in tariffs, and
favorable governmental policies to incorporate renewable
energy resources have also encouraged the domestic con-
sumer to invest in this technology to contribute toward
sustainable electricity generation. Battery technology has
matured and produced batteries with deeper discharge
and longer life at lower cost, and, therefore, PV/battery-
based microgrids can be considered as an optimal choice
for future electrification projects.
Solar PV systems produces dc, batteries store dc, and
most modern loads are now dc, which allows for local
power generation and distribution through dc microgrids
with the source closely matching the load profile. Com-
pared to traditional ac distribution, dc microgrids are signif-
icantly more efficient due to no dc–ac or ac–dc conversion
when implemented with distributed generation (DG). These
systems have an end-to-end efficiency of around 80% (for
dc loads) compared to ac microgrids, with an efficiency of
lower than 60%. Along with higher efficiency, dc microgrids
and the associated distribution have the inherent advan-
tage of less conductor usage for distributing the same
amount of peak power in comparison to ac distribution.
Therefore, the cost associated with distribution conductors
can be substantially reduced using dc distribution. The dc
distribution is more resilient from power quality issues, and
its reliability is relatively higher in comparison to ac distri-
bution. Therefore, due to their inherent simplicity, higher
power quality, enhanced efficiency, and straightforward
controllability, dc microgrids are preferred over ac microgrids
for rural electrification applications. These factors all favor
PV/battery-based dc microgrids for practical deployments,
and they are regarded as an optimal choice for rural electri-
fication applications.
Since their inception, solar dc microgrids have been
regarded as a game changer for transforming the power
scenario of remote communities. It is Power provisioning
believed that government-subsidized
and public–private partnership-based to high-power
solar microgrids have the potential
to do much more than provide basic
communal loads
electricity. These community microgrids results in a
have the abilty to create opportunities
for business and em­­ployment by pow- substantial increase
ering schools, medical care units, water
filtration plants, agriculture pumps,
in the required
telecom towers, and many other micro- capacity and
enterprises. These universal objectives
can be attained only through the associated cost of
proper tailoring, design, and selec-
tion of a suitable microgrid architec-
the installation.
ture that favors a local energy economy
with higher degrees of scalability and
adaptability. An overview of the existing architectures in
practice or in literature along with their pros and cons is
discussed in the following sections.
Centralized Architecture of Solar dc Microgrid
Figure 3 shows the topological diagram of a typical cen-
tralized dc microgrid architecture used for many rural
electrification implementations. Such an architecture in
which generation (PV panels) and storage (batteries)
resources are placed at a central location is referred to as a
central generation central storage architecture (CGCSA).
CGCSA has a unidirectional flow of power from a central
location with solar PV generation and storage to house-
holds. A single dc–dc boost converter is required for the
maximum power-point tracking of PV output and step-
ping 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.
Prominent practical implementations for rural electri-
fication through a CGCSA of solar dc microgrids include
microsolar plants in Chhattisgarh, India, by the Chhattis-
garh Renewable Energy Development Agency (CREDA),
India. CREDA has deployed 576 solar-based dc microgrids
with a cumulative capacity of 2.15 MW, serving around
31,000 customers in remote areas. Another very success-
ful commercial-scale solar microgrid is Mera Gao Power
(MGP), India, which involves central PV generation and
central battery storage with distribution at 24 Vdc to sub-
scribing houses. The subscribers of MGP may consume
up to 5 W of dc electricity (enough to power an LED light
and a mobile-phone-charging point) for 8 h in a day. It is
reported that MGP has more than 0.1 million subscribing
households spread across 400 villages. Similarly, in 2012,
the Uttar Pradesh New and Renewable Energy Develop-
ment Agency installed 1-kW dc microgrids in 11 districts
covering around 4,000 houses. The Jabula project in Cape
Town, South Africa, is another successful model, where
Zonke Energy installed a PV/battery-based centralized
solar dc microgrid (750 WP) to serves
nine families residing in informal
settings with basic electricity.
In all of the previously mentioned
practical deployments, the central-
ized architecture of a PV/battery-
based dc microgrid is used, in which
energy is generated and stored at a
centralized location. This energy is
then delivered to subscribing house-
holds via distribution conductors,
and, therefore, distribution losses are
associated with the delivery of ener-
gy. The distribution losses in this
architecture depend upon the num-
ber of subscribers, power levels to be
distributed, distribution voltage level,
and size of the mass-produced conductor used for distri-
bution. Generally, line losses reduce at higher distribu-
tion voltages and a wider area conductor size, whereas
the system exhibits lower distribution efficiency and
higher line losses with an increasing number of con-
nected households, higher power levels to be delivered
at individual households, 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, which
become more apparent at higher power levels due to the
increasing number of subscribers and household power
or community load requirements. Therefore, powering a
high-power communal load is not feasible in centralized
architecture from a distribution losses perspective.
dc Converter
PV Array
Battery Banks
dc Loads
Figure 3. A typical centralized dc microgrid architecture.
IEEE Elec trific ation Magazine / d ec emb er 201 8 67
 Figure 4 shows the distribution effi-
ciency variations of a typical village in
South Asia with the variations in
power delivery to individual house-
holds (W), a number of connected
households (N), and voltage levels at a
typical distribution conductor size of
AWG-2. It is evident from the figure
that distribution efficiency decreases
drastically (at lower voltage levels that
are considered safe for more village-
level electrification) with increasing
power levels, thereby making the
application of community load practi-
cally infeasible in the centralized
architecture. Considering the example
of MGP in India, which provides only
5 W of dc power at 24 V to each subscribing house, with a
limit of 0.2  A, enough to power two LED lights and a
mobile-phone-charging point. Although small power is
beautiful, it is unable to drive high-power community loads
and contribute to the uplift of the society. Due to the very
limited power supply, such a scheme is unlikely to alleviate
poverty in rural areas or contribute to significant improve-
ments in socioeconomic circumstances.
With the increase in population or number of subscrib-
ing households, a CGCSA offers rigidity in terms of future
expansions due to the nonscalable and nonmodular nature
of power processing equipment (dc converters). Generation
and storage resources are generally nonscalable due to the
requirements of high-cost synchronization equipment.
Therefore, such architectures require central planning of
the resources with a top-down approach. The major draw-
back associated with the top-down planning of a CGCSA is
that its generation and storage capacity have to be
designed as per peak power requirements of the load,
The distribution thereby increasing the up-front capital
cost of installation. Power provisioning
losses will occur to high-power communal loads,
including a water filtration plant, com-
only when there is puting load of a school, or load of
a need for power medical equipment in a health-care
unit, e.g., results in a substantial
sharing among increase in the required capacity and
associated cost of the installation.
multiple households Such centralized schemes cannot
or when there is incorporate microfinancing for wide-
scale deployments. Moreover, the use
a communal load factor of resources is generally lower
due to centralized planning require-
demand. ments. During the daytime, e.g., when
there is enough production by the PV
panels and the lighting load require-
ment for houses is comparatively negligible, the excess
power generated by the PV panels cannot be used optimal-
ly after the storage system has fully charged. Thereby,
excess power will be wasted within panels, making the
overall scheme essentially suboptimal in terms of resource
use. Distribution efficiency, nonscalability, lower use factor,
and higher up-front cost requirements are the major limi-
tations in narrowing down the scope of centralized archi-
tecture for global energy access realization.
Distributed Architectures of a Solar dc Microgrid
To minimize the limitations of a centralized architecture,
researchers are working on scalable architectures that
are either partially or completely distributed from the
perspective of PV generation and battery storage resource
placement at a village scale. Distributed and scalable
architectures with a bottom-up approach can enable the
organic growth of microgrids and, therefore, can poten-
tially empower local communities for sustainable

AWG-2, 380 V AWG-2, N = 40

100 100
Distribution Efficiency (%)

Distribution Efficiency (%)

98 80

96 60
24 V
94 40 48 V
120 V
N = 40 230 V
92 N = 80 20
325 V
N = 120 380 V
90 0
0 100 200 300 400 500 0 100 200 300 400 500
Household Load (W) Household Power (W)
(a) (b)

Figure 4. The distribution efficiency variations of a CGCSA with power provisioning variations at an individual household (W) (a) as a function of
the number of subscribing households (N) and (b) as a function of distribution voltage level (V).

68 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

development. Due to their distributed nature, they have
minimum distribution losses in the path and can provide
higher power deliveries at individual households. With
resource sharing and power-aggregation features, they
can sustain high-power community loads, without dedi-
cated power generation. One such architecture is the
CGDSA, as shown in Figure 5. Distributing the battery
storage system at individual consumer nodes will result
in reduced distribution losses, and the distributed power
may be intelligently stored or consumed at the load end
using household power management units (PMUs). This
will impart scalability to the overall structure in terms of
future capacity enhancement. Furthermore, the provi-
sion of energy storage at local houses also results in
higher efficiency compared to a CGCSA. The communica-
tion among distributed storage resources at various
households can be done through the Global System for
Mobile Communication (GSM), and resources at individ-
ual households can be pooled together for communal
load application. However, the presented architecture
uses centralized PV generation because of the higher
distribution losses associated with the distribution of
generated energy.
The distribution losses associated with the generated
energy can be further reduced using either partially dis-
tributed architecture (PDA) or highly distributed archi-
tecture termed as DG distributed storage architecture
(DGDSA) of the solar dc microgrid. In the PDA of the solar
dc microgrid, the consumer and generation modules are
distributed throughout the village, thus formulating a
PDA (see Figure 6). The generating modules have PV gen-
eration and battery storage resources at their disposal
that can be used to power up their local load as well as
dc
Solar PV Converter
Bidirectional Power
Unidirectional Power
Battery dc Loads
Figure 5. The CGDSA of a solar dc microgrid.
supply power to consuming nodes. Therefore, partial
distribution of both generation and storage resources
and peer-to-peer electricity sharing through GSM-based
communication between PMUs 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) result
in the reduction of distribution losses. The advantages of
the PDA are mainly a reduction in distribution losses
and modularity in a structure that allows scalability for
future capacity enhancements. Power from generating
modules can be pooled up for community load purpos-
es, thereby making PDA a suitable architecture for eco-
nomic uplift of the society.
The distribution of generation and storage resources
results in a reduction of overall distribution losses and
also imparts scalability to the architecture. The DGDSA of
the solar dc microgrid having PV generation and battery
storage at each individual house is shown in Figure 7. Each
household has its own PV generation, battery storage, and
local dc loads. Further, bidirectional power flow capability
is available in each household through bidirectional dc
converters to supply or demand power based upon the
local conditions of generation storage and use. In such a
system, each household can work independently as well
as share resources with the community. PV generation and
battery storage resources are designed according to local
load profile, and most of the time, each household in
DGDSA will be operating in islanded mode. Consequently,
the losses associated with the distribution of energy from
generation point to use point will be minimized while all
of the household load requirements will also be fulfilled
simultaneously. The distribution losses will occur only
Bidirectional Power
Unidirectional Power
Battery dc Loads
Generator Module
Consumer Module
dc Loads
Figure 6. The PDA of a solar dc microgrid.
IEEE Elec trific ation Magazine / d ec emb er 201 8 69
 when there is a need for power sharing among multiple
households or when there is a communal load demand.
The architecture has the built-in advantages of 1) a
higher efficiency because of DG and distributed storage, 2)
modular scalability for future expansion, 3) efficient aggre-
gation of power for larger loads even with limited rooftop
PVs, and 4) delivery to such communal entities as rural
schools and basic health units by pooling power from indi-
vidual household units without dedicated (large) genera-
tion. Furthermore, the distributed nature of the proposed
Communal Loads
DGDSA makes it independently scalable in its planning
Bidirectional Power and operation. Therefore, such a highly distributed archi-
Unidirectional Power tecture does not require centralized resource planning.
Rather, it can be planned in a bottom-up manner such
that more and more subscribers can be added within the
architecture without excessive modifications/replace-
ments in the existing structure. The coordinated resource
Battery dc Loads sharing among dispersed generation and storage resourc-
es, therefore, formulates a swarm of energy that has a high-
er potential than the uncoordinated centralized resources in
the microgrid structure.
From the sustainability perspective, distributed archi-
tectures are highly suitable for microfinancing opportuni-
Figure 7. The DGDSA of a solar dc microgrid. ties for private investors/public–private partnerships.
Distributed architecture also has the potential to create an
energy economy in the village through regulated energy
transactions among multiple households. This will not
TABLE 1. The in-practice solutions for the only result in business opportunities but also will open
electrification of developing regions. many horizons for local employment, e.g., for bill collec-
Type of Solution Option tion and maintenance of distributed PV resources. There-
fore, along with the capability to bring socioeconomic uplift
Utility (national) grid A in terms of communal load driving capability, distributed
Stand-alone solar B architectures have the potential to empower the local
population by creating indigenous business and employ-
Diesel generators C
ment opportunities.
Centralized solar microgrids D
Distributed solar microgrids E Comparison Matrix
The electrification architecture of developing communities
has taken a different form since the beginning of the rural
electrification era and the focus of governments and inter-
national organizations on universal energy access. These
TABLE 2. The typical power provisioning levels include electrification via a utility grid, SHSs, diesel genera-
for village-scale electrification. tor-based electrification, and solar microgrid-based electri-
Typical Power fication as tabulated in Table 1. Each electrification solution
Provisioning has some salient characteristics associated with it. Eco-
(Levels) Details nomic viability is one of the key concerns for developing
economies with constrained funding resources. The ability
1 Light/mobile phone charging up to
8 h/day to provide higher amounts of power for subscribing house-
holds as well as for community load applications is anoth-
2 24/7 light/mobile charging
er key feature that quantifies the potential of the scheme
3 Light(s), mobile charging, and house for contributing toward socioeconomic uplift. Various pos-
loads (fans and so forth) sible power provision levels for village-scale electrification
4 Light(s), mobile charging, fan(s), and have been tabulated in Table 2. Figure 8 shows the map-
larger communal loads ping of various electrification solutions from the per-
spective of economic viability and power provision
5 All loads (including industrial)
capability. It is apparent from the discussion of various

70 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

architectures that distributed solar dc microgrids exhibit
higher levels of scalability and can be designed with a bot-
tom-up approach, resulting in lower up-front cost require-
ments and higher economic viability for developing
D
regions. Also, lower distribution losses in the path of power E
delivery and bidirectional resource-sharing capabilities

Economic Viability
enable the architecture to support higher power loads for
B
community benefits. A brief comparison of other salient
characteristics of various electrification schemes, including
scalability, modularity, use efficiency, the potential for
energy microeconomy, the potential for poverty alleviation, C
and legal challenges, is shown in Table 3. Cost and afford- A
ability are the main drivers for adaptability; a cost compar-
ison of various architectures for a village of 40 households 1 2 3 4 5
is highlighted in Table 4. These village-level microgrids are Power Provisioning (Levels)
generally sustained through microfinancing or private sec-
tor investment, and a subscription-based cost analysis Figure 8. The mapping of various architectures from power provision-
is also listed in Table 4. Monthly charges are calculated ing and economic viability perspectives.
for paying back up-front and opera-
tion and maintenance (O&M) costs
including battery replacement for a TABLE 3. The characterization of various in-practice solutions
period of 25 years. for the electrification of developing regions.
Other Aspects A B C D E
Potential Challenges in
Practical Deployments of Scalability Low Low Low Low High
Distributed Microgrids Modularity Medium Low Low High High
Although the scalable architecture
Use efficiency High Low Low Medium High
allows for the efficient use of distrib-
uted resources in a highly scalable Communal loads High Low High Low High
manner, some challenges associated The potential for energy Low Medium High High High
with larger deployments persist. microeconomy
From a practical implementation
The potential for poverty High Medium Low Low High
perspective, there can be potential alleviation
challenges for the distributed place-
Legal challenges Low Medium Low Medium Medium
ment of resources. Space barriers

TABLE 4. The estimated cost of various architectures of dc microgrids for a

village of 40 households.
Subscription Charges per
User per Month for Payback
Capital Plus (in US$/Month)
Capital 25 Years
Load per House Cost O&M Cost Three Six Nine
Architecture (24/7 Provision to Subscribers) (US$) (US$) Years Years Years
CGCSA One light and mobile-phone-charging unit 2,020 4,550 1.4 0.8 0.6
(5 W)
DGDSA Three lights, one fan, charging unit (30 W 9,110 24,900 6.3 4.2 3.1
per house) 
Three lights, one fan, charging unit, and 9,525 25,510 6.6 4.5 3.3
communal load (30 W per house + 500 W)
Stand-alone Three lights, one fan, charging unit (30 W 10,310 27,905 7.1 4.9 3.3
production and per house)
consumption
(no microgrid) Three lights, one fan, charging unit, and 11,100 29,815 7.7 5.3 3.8
communal load (30 W + 500 W)

IEEE Elec trific ation Magazine / d ec emb er 201 8 71

 along with the maintenance of con-
verters and cleaning of PV panels at
individual households are some of the
practical challenges that need to be
addressed for successful practical
implementations. Bidirectional power
flow metering and theft monitoring
issues must also be considered for
future installations. The high-level dis-
tribution 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 also include an intelligent protec-
tion scheme capable of real-time load
flow and short circuit analysis for adap­­
tive relay settings.
From an economic point of view, such a distributed
model is highly suitable for microfinancing opportunities for
private investors/public–private partnerships. To enable
energy trading among multiple households, there must be a
mechanism to monitor energy transactions among neigh-
boring houses. Although an energy trade mechanism will
formulate a local energy market and be helpful for empow-
ering rural inhabitants, it will require an information and
communication layer at neighborhood levels to ensure
monitoring the exchange of energy. A practical solution to
these technical challenges along with successful business
models and financing solutions may enable a wide uptake
of the distributed solar microgrids for achieving the enhanced
level of energy access in the coming years.
Conclusions
The conventional schemes of electrification are limited in
their potential either due to economic constraints or their
inability to sustain high-power loads. Alternatively, scalable
solar dc microgrids designed through a bottom-up approach
offer a financially viable solution along with the ability to
sustain high-power loads for community benefits. There-
fore, scalable architectures of solar dc microgrids don’t just
provide access to basic electricity; they also have the poten-
tial to act as a catalyst for economic growth and improved
livelihoods. A proportionate and meaningful electrification
of developing regions can be achieved through implement-
ing scalable solar dc microgrids coupled with financing and
policy commitments on a broader scale. Thus, scalable solar
dc microgrids with a distributed nature and bottom-up
design can be regarded as a way forward to realize the glob-
al objectives of universal energy access.
Acknowledgments
We would like to acknowledge the support of Kiran Siraj,
a Ph.D. candidate in the Department of Electrical Engineering
at Lahore University of Management Sciences, Pakistan, for
72 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018
Scalable solar dc her editorial contributions in terms of
graphic de­­signs for this publication.
microgrids with a
For Further Reading
distributed nature World Energy Outlook. (2017). Electricity
and bottom-up access database. [Online]. Available:
https://www.iea.org/publications/
design can be freepublications/publication/WEO-
2017SpecialReport_EnergyAccess​Out-
regarded as a way look.pdf
D. Palit, G. K. Sarangi, and P. Krithika,
forward to realize “Energising rural India using distribut-
ed generation: The case of solar mini-
the global objectives grids in Chhattisgarh State, India,” in
Mini-Grids for Rural Electrification of Devel-
of universal oping Countries, S. C. Bhattacharyya and
D. Palit, Eds. New York: Springer-Verlag,
energy access. 2014, pp. 313–342.
J. Urpelainen, “Energy poverty and
perceptions of solar power in marginal-
ized communities: Survey evidence from
Uttar Pradesh, India,” Renewable Energy, vol. 85, pp. 534–539,
Jan. 2016.
P. A. Madduri, J. Poon, J. Rosa, M. Podolsky, E. A. Brewer, and
S. R. Sanders, “Scalable dc microgrids for rural electrification
in emerging regions,” IEEE J. Emerg. Sel. Topics Power Electron.,
vol. 4, no. 4, pp. 1195–1205, 2016.
W. Inam, D. Strawser, K. K. Afridi, R. J. Ram, and D. J. Per-
reault, “Architecture and system analysis of microgrids with
peer-to-peer electricity sharing to create a marketplace which
enables energy access,” in Proc. 2015 9th Int. Conf. Power Electron-
ics and ECCE Asia (ICPE-ECCE Asia), pp. 464–469.
M. Nasir, N. A. Zaffar, and H. A. Khan, “Analysis on central
and distributed architectures of solar powered dc micro­­
grids,” in Proc. 2016 Clemson University Power Systems Conf.
(PSC), pp. 1–6.
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 Trans. Sustain. Energy, vol. 9, no. 1,
pp. 390–399, 2018.
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 Trans. Power Electron., Apr. 2018. doi: 10.1109/
TPEL.2018.2828538.
H. A. Khan, H. F. Ahmad, M. Nasir, M. F. Nadeem, and N. A.
Zaffar, “Decentralised electric power delivery for rural elec-
trification in Pakistan,” Energy Policy, vol. 120, pp. 312–323,
Sept. 2018.
Biographies
Mashood Nasir (mashood.nasir@lums.edu.pk) is with the
Lahore University of Management Sciences, Pakistan.
Hassan Abbas Khan (Hassan.khan@lums.edu.pk) is with
the Lahore University of Management Sciences, Pakistan.
Nauman Ahmad Zaffar (nauman.zaffar@lums.edu.pk)
is with the Lahore University of Management Sciences,
Pakistan.
Juan C. Vasquez (juq@et.aau.dk) is with Aalborg Univer-
sity, Denmark.
Josep M. Guerrero (joz@et.aau.dk) is with Aalborg Uni-
versity, Denmark.

 By David Schumacher, Omid Beik, and Ali Emadi

Standalone
Integrated Power
Electronics Systems
Applications for off-grid rural locations.

©istockphoto.com/fotomem

ntuition would assume that the number of

I
Digital Object Identifier 10.1109/MELE.2018.2871318
Date of publication: 16 November 2018
people without energy access would decrease over time.
However, it has been estimated that the number of peo-
ple without access to electricity will continue to increase
over the coming years, as shown in Table 1. In Africa, for

2325-5987/18©2018IEEE IEEE Elec trific ation Magazine / D ec emb er 201 8 73

 Table 1. World population without access to electricity by region (in millions).

2009 2030

Share of Share of
Rural Urban Population Rural Urban Population
Africa 466 121 58% 539 107 42%
  Sub-Saharan Africa 465 121 69% 538 107 49%
Developing Asia 595 81 19% 327 49 9%
  China 8 0 1% 0 0 0%
  India 268 21 25% 145 9 10%
  Rest 319 60 36% 181 40 16%
Latin America 26 4 7% 8 2 2%
Middle East 19 2 11% 5 0 2%
Developing countries 1,106 208 25% 879 157 16%
World* 1,109 208 19% 879 157 12%

Note: * Includes countries in the Organisation for Economic Cooperation and Development and Eastern Europe/Eurasia.

Area Out of Threshold

PV Under Threshold
Diesel Under Threshold
Grid Connection Area Out of
Threshold
Diesel and PV Under
PV Under
Grid and PV Threshold
Grid and Diesel
Grid, Diesel, and PV

Figure 1. The geographical distribution of technologies with elec-

tricity costs lower than 25c€/kWh and a conservative assumption
on grid extension. PV: photovoltaic. (Photo used with permission Figure 2. An economic comparison of diesel versus PV. (Photo used
from Szabó et al.) with permission from Szabó et al.)

example, there are almost 600 million people currently liv-
ing without access, with most living in rural areas.
Rural electrification continues to influence developing
countries around the world. Traditionally, rural electrifica-
tion is powered by either grid connection; distributed energy,
such as solar, wind, or hydro; or through the use of biofuels
like wood, oil, or diesel, with diesel being the most common
off-grid option for rural locations. Distributed energy has
advantages over the traditional options, and, in reality, the
decision usually comes down to cost and time. However,
renewable energy solutions would be most ideal because
they are sustainable and help to fight climate change.
Luckily, renewable energy sources, such as solar and
wind, are becoming more popular due to reduced costs
and advancing technologies, such as battery technology,
as well as more efficient power electronic converters. For
example, the cost of solar energy has fallen by 80% and
could continue to decrease another 60% by the end of the
current decade, and the efficiency of solar energy systems
has also increased over the past 40 years.

74 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

 These improvements strengthen the
case for distributed and renewable energy

Figure 3. The single bus topologies. Note: Red lines indicate low dc voltage, green lines indicate high dc voltage, and blue lines indicate high ac voltage. (a) The topology for a single ac bus in which all the
HVac
sources, specifically for off-grid applica-

Load
tions. In Africa, 60% of new energy generation
is estimated to be from off-grid solutions,
with an estimated 730 million people cur-

dc/ac
=/∼
rently relying on biomass for their electrici-
ty needs. Thus, the potential output for

HV Bus
switching to a renewable distributed alter-
native, like solar and wind, can have a great

Isolated dc/dc
Isolated dc/dc

Isolated dc/dc
Bidirectional

Bidirectional
impact on the energy developments around

dc/dc
dc/dc
=/=

=/=
the world, specifically for developing coun-

or

or
tries. For small rural communities, there are

(c)
several reasons why alternative of energy
sources are preferred over grid-connected

Source

Source

Load

ESS

ESS
HV

HV
LV
solutions. The infrastructure costs and time

LV

LV
required for grid extension, because of road
and tower building and gathering of
required materials, is costly and time con-

HVac
Load
suming compared with diesel and solar.
And even those connected to the grid do
not always receive reliable energy because
of a weak grid and power outages. Diesel is

dc/ac
=/∼
the most common option used in some
rural areas for electricity generation be­­
Isolated
dc/dc

cause of its reliability and cost. However,

there are downsides. Diesel is a fossil fuel

LV Bus
and contributes to climate change. For indi-
viduals to use diesel, methods of trans-
Isolated dc/dc

Isolated dc/dc
Bidirectional

Bidirectional
porting and extracting it can be time
dc/dc

dc/dc

dc/dc
=/=
=/=

=/=

or
or

consuming, especially in remote locations.

This time can be spent on education or

(b)
contributing to family income. Therefore,
Source

an alternative energy option that addresses

Source

Load

ESS

ESS
HV

sources are connected individually, (b) a single LVdc bus, and (c) a single HVdc bus.
HV
LV

LV

LV
these issues will be more attractive. Com-
pared to grid extension and diesel, distrib-
uted energy has higher initial costs and
lower energy densities, but it is sustainable,
HVac
Load

quick to implement, low maintenance, and

does not require large transportation costs.
Figure 1 shows different energy methods
ac Bus

suitable for each location in Africa on a cost

Bidirectional

basis, and Figure 2 shows the cost compari-

son for solar and diesel. On the basis of Fig-
dc/ac
dc/ac

dc/ac

=/∼
=/∼
=/∼

ures 1 and 2, solar is a suitable energy

source on a cost basis in many areas and
indicates its viability versus the current
Isolated dc/dc
Isolated dc/dc
Isolated dc/dc

energy options.
Bidirectional
dc/dc
dc/dc

=/=
=/=

or
or

DC Conversion Systems
(a)

Typically, for off-grid systems, there are three

different topologies: dc, ac, or hybrid sys-
Source

Source

tems. For dc systems, there is a main dc bus

ESS

ESS
Load

HV
HV

LV
LV

LV

where both of the sources, loads, and energy

storage systems (ESSs) all implement dc elec-
tricity. For ac systems, it is exactly the same

IEEE Elec trific ation Magazine / D ec emb er 201 8 75

 The topologies shown in Figure 3
are further broken down by isolated
dc/dc
24–40-V Solar LV and nonisolated configurations,
Panel > 1 kW Source =/=
which are dependent on the volt-
age level of each component within
dc/dc
1 × 12–24 V, LV the system. Generally, low-voltage
100 W Load =/= dc/ac
HVac (LV) systems require galvanic isola-
=/∼ Load tion from high-voltage (HV) sys-
dc/dc
LV Cell tems. Figure 3(a) shows the topology
3 × 5 V, 1 A =/= 1 × 1 kW, 230 V,
Phone for a single ac bus in which all the
50 Hz
Bidirectional sources are connected individually
LV dc/dc either through isolated or nonisolat-
48-V Battery =/=
ESS ed converters, depending on the
voltage level. If a main dc bus is
48-V 400-V desired, this topology shifts into a
LV Bus HV Bus single LVdc bus [Figure 3(b)] and a
single HVdc bus [Figure 3(c)].
Figure 4. A proposed standalone off-grid integrated power converter system. Topologies presented in Figure 3(b)
and (c) are more common than that
as dc but with ac buses. And for hybrid systems, both dc shown in Figure 3(a), when only one ac load is present
and ac sources, ESSs, and loads are used. The selection because the ESS can be connected to the dc source and dc
between these architectures is dependent on the applica- load through the same dc bus. Note that the dc loads are
tion requirements. Since solar is the dominant choice for assumed to be at low voltages, i.e., 5, 12, 24, or 48 V, while
an energy system architecture as shown in Figures 1 and 2, the ac loads are assumed to be at high voltages, i.e., 120,
a dc-sourced system will be the focus of the topology dis- 220, 230, or 240 V. The ESS is required to have bidirectional
cussion in this article. Therefore, the possible topologies for operation because of charging and discharging demand,
an off-grid rural application are presented in Figure 3. and its voltage is dependent on the connection type.

Table 2. Portable off-grid products and applications.

EnerPlex’s Generatr 1200 Solar Generator Wagan Tech Solar ePower Cube 1500

■ 1,231-Wh, 83.2-Ah, 12-V Li battery ■ Solar controller, 20 A

Input:
■ 4# solar panel dc jack 16–34 V, 120 W maximum
■ Wall charger dc jack 19 V, 120 W maximum
■ 2# Anderson jacks 12 V, 120 W maximum
■ 1# 100-Ah, 12-V, 15-A maximum continuous absorbent
glass mat (AGM)/Gel hybrid battery
Input:
■ 2# 100–240-Vac charger 50/60 Hz

■ 5# 16-W solar panels (80 W total), 17% efficiency,

Output:
■ 
20 A maximum
USB, 12-V standard ac wall outlet
■ 3# USB 2.4 A/12 V–10 A, 120 W maximum total Output:
■ 12-V Anderson jack 10 A, 120 W maximum total ■ 2# 5-V USB outlet, 2.1 A maximum

■ 19 V, 6.3 A, 120 W maximum total ■ 2# 115-Vac, 60-Hz 1.5-kW continuous 3.6-kW peak

■ 3# 100-Vac 60-Hz pure-sine wave 1,000 W maximum ac output, modified sine wave, 90% efficiency
­total ­continuous (1,100 maximum over loading) ■ 2# 12-Vdc outlet

Goal Zero Yeti 1250 Solar Generator Renogy’s Lycan PowerBox

■ 1,200-Wh, 100-Ah, 12-V AGM lead-acid battery ■ 24-Ah, 1,075-Wh, 12-V lead-acid battery
■ 20-A MPPT 1.2-kW pure sine wave inverter
Input:
■ 
120-Vac wall charger (72 W) Input:
■ 
12-V car charger (30 W) ■ Solar panel

■ 
8-mm solar 16–48 V, 10 A, 160 W maximum ■ Wall outlet

■ 
Power pole 16–48 V, 20 A, 240 W maximum ■ 12-V car charger
Output: Output:
■ 3# 5-V, 2.1-A USB outlets ■ 2# 12 Vdc

■ 8-mm 12 V, 10 A, 120 W ■ 3# 110 Vac

■ 12-V, 33-A power pole port ■ 4# 2.4-A USB

■ 3# 110-Vac, 60-Hz, 1,200-W pure sine wave

76 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

 Since the topologies presented here are an off-grid
standalone application, there will be no grid connection,
and the ESS is charged using the source. Structures pre-
sented in Figure 3(b) and (c) are ideal for topologies with
only one ac load, while the topology shown in Figure 3 is
suitable for multiple ac loads and sources. Therefore, for
a rural off-grid standalone system, the converter shown
in Figure 4 is proposed. The  proposed topology imple-
ments an integrated converter system that can be used
to power
1) one 100-W, 12–24-Vdc load
2) one 1,000-W, 230-V, and 50-Hz ac load
3) three 5-V and 1-A cell phone charging ports.
Note that 12- or 24-V batteries can also be connected to
the 12–24-Vdc load, but ideally, this connection is for light-
emitting diode lighting or a 12–24-V load. The existing off-
grid applications that achieve similar tasks as the proposed
topology fall into portable and stationary systems. Porta-
ble systems utilize battery banks to power the loads and a
conversion system that uses a 12-Vdc to 120-Vac inverter
for either pure sine or modified sine wave output. The
input charging capability in the existing portable systems
is low, so the charge times are long where usually the
battery needs to be charged for the system to run the out-
puts because the input power is not enough to operate all
the loads at full power. Some of the portable products
that are primarily designed for camping purposes are list-
ed in Table 2.
Stationary systems are more variable in the literature
and usually customized for each application. Many range
Table 3. Stationary off-grid products and applications.
Sunblazer II Unit
■ MPPT solar charge controller
Input:
■ 6# 300-W solar panels
■ 4# 24–48-V AGM/gel batteries
■ 1-kW ac input charger (wind/diesel) backup power
Output:
■ 10–20# 12-V, 9–18-Ah universal portable battery units
■ 1.5 kW ac 230 V, 50/60 Hz
Leading Edge Power Box
■ 1,200-Wh, 360-Ah/24-V battery with 24 V/(10, 25, 50,
or 70 A charger)
Input:
■ 2# 140-W solar panels
■ 300–450-W wind power
■ 187–265-V, 16-A, 45–65-Hz ac input
Output:
■ 
12 V, 9 A, and 24 V, 6 A dc load
■ 
230 V, 16 A ac output
from trailer systems, storage containers, carts, or systems
installed on houses or buildings. Table 3 lists some of the
commercially available stationary products. The Sunblazer
II unit and the Smart Village Universal Charge Controller
shown in Figure 5 have a similar topology to the proposed
system in Figure 4. For the Smart Village Universal Charge
Controller, the demo board’s solar input is from 17 to 54 V
with a maximum power of 200 W. The demo board is used
to charge a 12-V sealed lead acid battery with a maximum
input and output charge current of 10 and 16.6 A, respec-
tively. The Sunblazer unit II consists of six 300-W solar
panels, four 24/48-V batteries (absorbent glass mat/gel), a
solar charge controller using maximum power point
tracking (MPPT), 10/20 charging channels for 12-V, 9–18-Ah
universal portable battery units, and the option to have
both a 1,000-W ac input charger and a 1,500-W ac output
load at 230 Vac, 50 Hz, or 60 Hz. It is mainly used to charge
multiple battery packs for approximately 500 people.
Compared with the portable option, the stationary system
can achieve higher power, and packaging and mass are
not as much of a concern.
Integrated Solar Conversion System
On the basis of the proposed topology in Figure 4, the
implemented standalone off-grid power electronic sys-
tem for rural applications is shown in Figure 6 with its
main specifications listed in Table 4. The proposed topol-
ogy is suitable for a home system. However, the system
can be used as a battery charging station similar to the
Sunblazer II unit. For a battery system connection,
Smart Village Universal Charge Controller
■ MPPT solar charge controller
Input:
■ 18–24-V solar panel
■ 45–55-V solar panel
■ 50–60-Vdc microgrid
■ 24-V Sunblazer-Lite or storage
■ 
110/220 ac microgrid
Output:
■ 
12-V home battery
■ 
12-V appliances (fan, lights, and so on)
Other Products/Manufacturers
■ Tesla PowerWall
■ SMA products, e.g., Sunny Boy 3.0–5.0, Sunny Island
4.4 M
■ 
Schneider electric products, e.g., Xantrex XW
■ SolarEdge, e.g., SE3000H
■ Enphase Energy, e.g., Enphase ac battery, Enphase IQ
6 microinverter
■ ABB, e.g., MGS100
IEEE Elec trific ation Magazine / D ec emb er 201 8 77
 charging and discharging, an off-the-shelf bidirectional
dc/dc converter and control system can be used, as
schematically depicted in Figure 6. The proposed sys-
tem includes
1) a three-phase input boost converter
2) a single-phase buck converter to supply a dc load
3) a phase-shifted full-bridge (PSFB) converter
4) a single-phase full-bridge inverter.
The boost converter increases the input voltage from 24–40
to 48–60 V, which will be referred to as the LVdc link. The
basic single-phase boost converter can be used. However,
moving to a higher number of phases helps with reducing
the phase current, input and output current, and voltage
ripples. The dc load is powered from the LVdc link using a
buck converter, which converts the 48–60 to 12–24 V at the
load side, as most common dc loads are in the range of
5–24 V. The power flow is from the LVdc link to the dc load,
in the range of 100 W. Therefore, a nonisolated buck-type
converter is chosen for simplicity and cost perspective. The
system allows connection of a three universal serial bus
(USB) ports, e.g., for cell phones, because they are very com-
mon. For cell phone charging, an off-the-shelf converter is

Solar
Sunblazer dc Microgrid
PV Home
24 Vdc 18–24 Vdc 50–70 Vdc

Port C1 Port C2
Universal Charge Controller

Buck

Buck

Port C3 Battery Port Load Port

ac Microgrid Rectifier 12–24-V Battery Loads

(a) (b)

Figure 5. (a) A smart village universal charge controller. Note: Red indicates ac voltage, blue is dc voltage, gray indicates controller, and purple
signifies converter. (b) A SunBlazer II unit.

LVdc Link HVdc Link

Three-Phase Boost Converter PSFB

L5
L1 D1 D2 D3 S4 S6 D5 D7 S9 S11

L2 L6 ac Load
C1 C3
L3
Solar Input

C4 Rac
+
– D6 D8 S10
S1 S2 S3 S5 S7 S12

Single-Phase Inverter
Off-the-Shelf
Battery Bidirectional S8
dc/dc Converter L4

USB D4 C2 Rdc dc Load

Off-the-Shelf
USB USB Isolated dc/dc
Ports Converter Buck Converter
USB

Figure 6. The proposed standalone off-grid integrated power converter system for rural applications. USB: universal serial bus.

78 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

used to convert the LVdc link voltage down to 5 V. The LVdc flyback and forward topologies utilize only one quadrant
link also powers the single-phase full-bridge inverter employ- of the B–H curve as compared to half- and full-bridge con-
ing the PSFB converter. The PSFB converter implements an verters that use two-quadrant B–H characteristics. There-
eight-turn transformer at 50 kHz to achieve the required a fore, for the same output power, the single-quadrant
350–400-V HVdc link. The HVdc link is input to a single-phase topologies will require the core to be double the size. The
full-bridge inverter that powers a 230-V, 50-Hz, and 1-kW ac switching devices also have to withstand larger currents
because most countries use 230 Vac for their appliances. and voltages compared to the half and full bridge. The

Zero Voltage Switching

and Output Voltage Table 4. Proposed converter specification.
Considerations
Converters Boost Buck PSFB Inverter
The PSFB converter is a suitable iso-
lated topology for the proposed Input voltage (Vdc) 24–40 48–60 48–60 330–450
application because of its facilitated Output voltage (V) 48–60 dc 12–24 dc 330–450 dc 230 ac rms
ability to achieve zero voltage switch­­
Maximum output inductor 40 10 10 40
ing (ZVS) compared to other isolat-
current ripple (%)
ed topologies, i.e., flyback, forward,
push–pull, half bridge, and full Maximum output voltage 6 6 5 5
ripple (%)
bridge. The transformers in the

Eight Turns Eight Turns

10 10
Output Filter Inductance (µH)

Output Filter Inductance (µH)

Out of Range
50 50
ZVS Condition
90 90
130 130
170 Out of Range 170
210 210
250 250
ZVS Condition Vo Condition
290 290
330 330
370 370
400 400
5 10 15 20 25 30 35 40 45 50 5 10 15 20 25 30 35 40 45 50
Nine Turns Nine Turns
10 10
Output Filter Inductance (µH)

Output Filter Inductance (µH)

Out of Range
50 50
90 90
130 130 ZVS Condition
Out of Range
170 170
210 210
250 250
290 ZVS Condition 290
330 330 Vo Condition
370 370
400 400
5 10 15 20 25 30 35 40 45 50 5 10 15 20 25 30 35 40 45 50
Ten Turns Ten Turns
10 10
Output Filter Inductance (µH)

Output Filter Inductance (µH)

Out of Range
50 50
90 90
130 Out of Range 130
170 170
210 210 ZVS Condition
250 250
290 ZVS Condition 290
330 330
370 370 Vo Condition
400 400
5 10 15 20 25 30 35 40 45 50 5 10 15 20 25 30 35 40 45 50
Leakage Inductance (µH) Leakage Inductance (µH)
(a) (b)

Figure 7. (a) and (b) The phase-shift conditions for varying turns, load power, leakage inductance, and output filter inductance.

IEEE Elec trific ation Magazine / D ec emb er 201 8 79

 push–pull is an improvement, but it
Table 5. Integrated converter components.
suffers from flux imbalance in the
Converters core and also has larger voltages on
the switching devices compared to
Components Type Boost Buck PSFB Inverter
the half and full bridge. When com-
AGP4233-103ME Inductor #6 0 0 0 paring the half bridge to the full
IPP083N10N5 MOSFET #3 #1 0 0 bridge, the full bridge is more ad­­
vantageous because it has lower
IDW30E65D1 Diode #3 #1 #4 0
conduction losses. For the same out-
MKP1848C61060JK2 Capacitor #5 #1 0 0 put power, it achieves half the con-
1140-101K-RC Inductor 0 #1 0 0 duction losses since the currents are
FDP045N10A MOSFET 0 0 0 twice as large in the half bridge,
#4
which can be even more advanta-
T1000DC-1-8 Transformer 0 0 #1 0
geous as the power level increases.
1140-151K-RC Inductor 0 0 #2 0 There are a few improvements
AIRD-03-270 Inductor 0 0 #1 0 that can be made to the traditional
full-bridge converter, which usually
LLS2W221MELA Capacitor 0 0 #8 0
come down to ZVS or zero current
DENO-23-0001 Inductor 0 0 0 #1 switching. There have been many
940C12P1K-F Capacitor 0 0 0 #1 converters implementing some
STP28N65M2 MOSFET 0 0 0 #4 form of resonance, with the sim-
plest being the PSFB converter. The
MOSFET: Metal–oxide–semiconductor field-effect transistor; x: Number of components used. phase shift between two legs of
the PSFB converter has a range of
0–180°; hence, if a specific group of
25 PSFB parameters results in the
Voltage (V)

20 phase shift out of this range, then

those parameters cannot be used.
15
Figure 7 shows a comparison be­­
0 0.5 1 1.5 2 2.5 3 tween the ZVS and the output volt-
Time (s) age for a different number of turns
(a) and for 250- and 1,000-W output,
while the output filter inductance
Power (kW)

1
and leakage inductance are varied
0.5 from 10 to 400 µH and from 1 to
50 µH, respectively. The areas in yel-
0
0 0.5 1 1.5 2 2.5 3 low are out of the range of 0–180°,
Time (s) blue shows the areas where the
(b) output voltage constraint has lower
values than the ZVS constraint, and
Voltage (V)

40 orange shows the locations where

30 the ZVS constraint has lower values
than the output voltage constraint.
20
Lower phase-shift values mean
0 0.5 1 1.5 2 2.5 3
Time (s) that it is harder to achieve either
(c) the output voltage or ZVS. These
conditions indicate the highest the
100
Power (W)

phase shift can be. For instance, if

80 the phase shift is 50°, any value
60 below 50° will achieve either ZVS or
the output voltage. Therefore, if the
0 0.5 1 1.5 2 2.5 3
Time (s)
ZVS condition has the phase shift
being lower than 50° and the out-
(d)
put voltage phase shift is lower
Figure 8. The prescribed set of conditions applied to the converter: (a) buck converter output than 30°, then any value below 30°
voltage, (b) ac load power, (c) boost converter input voltage, and (d) buck converter output power. will guarantee that both ZVS and

80 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018

 60 56
VLVdc VLVdc
Voltage (V)

Voltage (V)
55
55
54
50 53
0 0.5 1 1.5 2 2.5 3
Time (s)
(a)
30 13
Voltage (V)

Voltage (V)
20
12
10
VOBuck VOBuck
0 11
0 0.5 1 1.5 2 2.5 3
Time (s)
(b)

380 360
VHVdc VHVdc
Voltage (V)

Voltage (V)
360 355
350
340 345
320 340
0 0.5 1 1.5 2 2.5 3
Time (s)
(c)

650 VAC 650 VAC

Voltage (V)

Voltage (V)

325 325
0 0
–325 –325
–650 –650
0 0.5 1 1.5 2 2.5 3
Time (s)
(d)
2,000 PinBoost 2,000 PinBoost
Power (W)

Power (W)

1,500 1,500
1,000 1,000
500 500
0 0
0 0.5 1 1.5 2 2.5 3
Time (s)
(e)
120 120
POBuck POBuck
Power (W)

Power (W)

100 100
80 80
60 60
40 40
0 0.5 1 1.5 2 2.5 3
Time (s)
(f)
2,000 2,000
Pac Pac
Power (W)

Power (W)

1,500 Inst 1,500 Inst

1,000 AVG AVG
1,000
500 500
0 0
0 0.5 1 1.5 2 2.5 3
Time (s)
(g)

Figure 9. The integrated converter-simulated waveforms: (a) LVdc link voltage, (b) buck converter output voltage, (c) HVdc link voltage, (d) ac
load voltage, (e) boost converter input power, (f) buck converter output power, and (g) ac load power.

IEEE Elec trific ation Magazine / D ec emb er 201 8 81

 Inverter Input Voltage
Inverter Input Current
dc Input
Load Voltage
Load Current
LVdc
Power
Supply
(a)
Figure 10. (a) The experimental results for 1,000-W, 230-V, and 50-Hz ac load and (b) the experimental test setup.
the output voltage are achieved. If the phase-shift value is
between 30 and 50°, it means that ZVS will be achieved
but the desired output voltage will not. Because this
design is trying to achieve both ZVS and the output volt-
age, the lower phase-shift value out of the two is the one
that needs to be targeted for both conditions to be met.
Figure 7 indicates which condition has the smaller
phase-shift value, showing which condition takes prece-
dence over the other. ZVS is harder to achieve at light
loads, and as seen from Figure 7, there is no blue coloring
for the 250-W case. It can also be seen that as the number
of turns increases, the blue area decreases. By an increase
in the number of turns, the output voltage will be easier to
achieve, which in turn means that the phase-shift values
will be higher. Table 5 lists the parameters that were
selected for the proposed converter system.
Discussion
To demonstrate the feasibility and effectiveness of the
topology presented in Figure 6, a prescribed set of condi-
tions, shown in Figure 8, is applied to the integrated con-
verter over a time frame of 3 s with the LVdc link and
HVdc link voltages clamped at 55 and 350 V, respectively.
Over this time frame, the buck converter output voltage
and power are changed, along with the input source volt-
age and output ac load power, as can be seen in Figure 8.
From these applied changes, the integrated converter
waveforms at a few specific areas are shown in Figure 9.
The right-hand side shows a zoomed-in version of the
waveforms on the left, and the purple dash lines indicate
from where these zoomed-in figures are taken.
From Figure 9, it is clear that the controller, through a
closed-loop control scheme, is able to maintain the de­­sired
prescribed variables (voltages and power) over the entire
simulated operating range. The HVdc link voltage has an
ac ripple of 3%, i.e., below the 5% threshold, which is
acceptable. The buck converter output power is able to
meet the desired 50 and 100 W at both 12 and 24 V, while
82 I EEE E l e c t r i f i c a t i on M a gaz ine / December 2018
Buck PSFB
Boost
Inverter
Integrated Control Unit
(b)
the ac load power is successfully maintained at 250 and
1,000 W at 230 V and 50 Hz, as shown in Figure 9. The
inverter output voltage is maintained at 230 V rms over the
entire operating range, and the total harmonic distortion
over the entire simulated operating range is 0.3%, which is
well below the required 5% threshold for distributed ener-
gy systems. Figure 10 shows the experimental results for
an ac load of 1,000 W that are in a good agreement with
the simulation, confirming the simulation models, proce-
dures, and the practicality and feasibility of the system.
Note that the total conversion system efficiency is mea-
sured at around 90% at the rated load condition.
For Further Reading
IRENA, “Africa 2030: Roadmap for a Renewable Energy
Future,” International Renewable Energy Agency, Abu
Dhabi, 2017.
R. Podmore, R. Larsen, H. Louie, N. Johnson, and S. Saha,
“Fueling sustainability: The exponential impact of empower-
ing off-grid communities,” IEEE Electrific. Mag., vol. 4, no. 1, pp.
11–17, 2016.
H. Louie, E. O’Grady, V. Van Acker, S. Szablya, N. P. Kumar,
and R. Podmore, “Rural off-grid electricity service in Sub-
Saharan Africa,” IEEE Electrific. Mag., vol. 3, no. 1, pp. 7–15, 2015.
S. Szabó, K. Bódis, T. Huld, and M. Moner-Girona, “Energy
solutions in rural Africa: Mapping electrification costs of dis-
tributed solar and diesel generation versus grid extension,”
Environ. Res. Lett., vol. 6, no. 3, pp. 34002, 2011.
L. V. Bellinaso, C. D. Schwertner, and L. Michels, “Price-
based power management of off-grid photovoltaic systems
with centralised dc bus,” IET Renew. Power Gener., vol. 10, no. 8,
pp. 1132–1139, 2016.
Biographies
David Schumacher (schumadr@mcmaster.ca) is with
McMaster University, Hamilton, Ontario, Canada.
Omid Beik (omid.beik@ieee.org) is with Magna Pow-
ertrain Inc., Toronto, Ontario, Canada.
Ali Emadi (emadi@mcmaster.ca) is with McMaster Uni-
versity, Hamilton, Ontario, Canada.

By Faeza Hafiz, Anderson Rodrigo de Queiroz, and Iqbal Husain

footage firm, inc.

Solar Generation,
Storage, and
Electric Vehicles
in Power Grids
Challenges and solutions olar energy is an abundant renewable

s
energy source that is available all around the
with coordinated control world every day. Each hour, the solar rays that

at the residential level.
Digital Object Identifier 10.1109/MELE.2018.2871319
Date of publication: 16 November 2018
reach our Earth (if properly converted to electrici-
ty and other forms of energy) represent more
than the total energy consumption of the entire human race
over the course of one year. Wind energy is another important
renewable resource available in large amounts every day.
These two renewable energy sources are attracting significant

2325-5987/18©2018IEEE IEEE Elec trific ation Magazine / D EC e mb e r 201 8 83

 investment as countries seek technology cost reductions
to aid sustainability.
Improvements in the conversion and use of larger frac-
tions of solar and wind energy would help in terms of sup-
ply security and satisfying societal energy needs for
decades and even centuries to come without adding high
levels of carbon dioxide and other greenhouse gases to the
environment. Therefore, in several energy markets (such as
in Brazil, Norway, Germany, England, and Australia), gov-
ernments and regulators are emphasizing the need to
increase the deployment of solar and wind energy sources
as well as other measures to enhance monitoring, pricing,
and control mechanisms in modern power grids.
The use of these renewable generation sources by them-
selves in power grids generally raises issues related to sup-
ply uncertainty, mismatch with system peak demand, and
voltage and frequency regulation. For example, imagine the
simple case of a household with installed solar panels that
are used to produce electricity. In the daytime, the sun may
be shining, and electricity can be produced without prob-
lems. But during this period, the household need for elec-
tricity is typically lower because the users are away. This
simple example can be extended to more complex and
interconnected issues as the level of penetration of renew-
able energy resources increases.
In modern power grids, the use of devices ensuring the
proper use of renewable energy sources is key as invest-
ments in wind and solar grow. Static batteries and other
kinds of energy-storage devices are encouraged in these
systems, and their employment helps to shift energy
usage over time. Storage technologies are attracting the
attention of investors, and there are plenty of successful
applications of different storage devices alongside renewable
energy sources. Government agencies and other regulators
are also invested in exploring new opportunities for energy
storage in energy markets. For example, in the United
States, recently completed and ongoing projects can be
found in the states of Massachusetts, California, North Car-
olina, Texas, and others (Figure 1).
Along with advances in modern power grids, the transpor-
tation sector is also significantly modernizing. New technolo-
gies are focused on developing environmentally friendly
means of transport, such as plug-in electric vehicles (PEVs).
PEV technology introduces a new link between the power and
transportation infrastructures. These vehicles require electrici-
ty from the grid for battery charging, thus achieving the auton-
omy to transport people from place to place. But if the PEV is
connected to the grid, the electricity stored in its battery can
also be used to satisfy other household electricity demands.
Because of PEVs’ bidirectional capabilities, these transporta-
tion vehicles represent at once a load on the grid and a source
of electricity for the grid, depending on when it is needed.
Because the use of solar energy panels, energy storage,
and PEVs is ramping up, homeowners with both forms of
storage can control and manage their equipment’s opera-
tion in a coordinated way to reduce electricity/transporta-
tion costs and minimize other problems. The goal of this
article is to offer an overview of solar PV generation, bat-
tery storage, and PEVs and highlight the benefits associat-
ed with the coordinated use of these technologies in
energy management at the household level.
Present and Future of Solar Generation
at the Residential Level
According to the Solar Energy Industry Association, 2,227 MW
of rooftop solar was installed in 2017 in the United States.

4%
42%
9% 5% 36%
13% 14%
33%
3% 18%
14% 23%
1% 22%
26%
15% 23%
–11% 13% 15% 26%
–4% 20% 12% 12%
22% 3% 16% 12% 8%
22% 7% 13%
6%
17% 9%
–15% 18% 14%
2% Change
Cha
9% 10% 3% (% of Total
Tot Sales)
0% 7% 11% or
o More
1–10%
5%
0 or Negative
N

Figure 1. The change in the rooftop solar potential 2008–2016. (Image courtesy of the National Renewable Energy Laboratory, www.nrel.gov/
docs/fy16osti/65298.pdf.)

84 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018

Based on research by the National Renewable Energy Labo-
ratory (NREL), rooftop solar could one day produce almost
40% of our electricity using the average efficiency of solar
panels installed in 2015. The research report stated that
there are over 8 billion m2 of suitable roofs in the United
States. It is possible to generate about 1,400 TWh of elec-
tricity each year, and two-thirds of this would come from
small residential buildings. The analysis found that Califor-
nia could supply approximately 74% of its total electricity
usage by covering its buildings with solar panels, while
Wyoming (with fewer sunny hours) would achieve about
14% of its needs and New England around half.
In addition to the available potential generation capacity, demand and the amount of available solar PV energy
the reduction in the cost of solar photovoltaic (PV) genera-
tion over the past five years is creating a significant shift on observed solar curtailment exceeding 30% of the solar pro-
the supply side of power grids. According to recent esti-
mates by the International Renewable Energy Agency, most
homeowners were investing between US$2.70 and US$3.60
per watt to install solar PV in 2018. Furthermore, solar PV
costs are expected to be cut in half by 2020, as reported by
the same agency. The total price of solar PV panel installa-
tion varies from country to country and, in the United States,
from state to state: in U.S. states, local governments and
utilities offer rebates and other tax incentives that reduce
solar generation investment costs. For example, a Green-
tech Media (GTM) research analysis showed that home-
owners can save, on average. approximately US$38,000 in
Portland, Oregon; US$63,000 in ­Boston; and US$90,000 in
Los Angeles over 20 years, considering incentives. Currently,
North Carolina homeowners can often spend fewer than
US$2 per watt on installation.
In a household, the capacity of solar PV installation
depends on the customer’s needs and preferences and also
affects electricity generation and overall cost. The size of
the customer’s PV panels has an impact on the existing grid
infrastructure. Safety, reliability, and operational issues may sign several contracts with peak-power generation plants
not be evident at lower penetration levels of solar power,
but higher penetration levels can create reliability challeng-
es for the grid. These issues are not generally observed at
the individual household level; as
penetration levels increase, however,
30,000
the effect is seen in the overall dis-
tribution circuit operation or even
further upstream in the regional dis­­ 25,000
tribution or transmission substation.
Net Load (MW)
The reliability issue also plays a sig-
nificant role in microgrids and isolat- 20,000
ed power grids.
For example, Hawaii is a proto-
type of the integration challenges 15,000
associated with residential rooftop
solar PV generation: in 2017, energy
10,000
production exceeded 75% of the
typical daytime gross electricity
demand in the system on several is­­ Figure 2. Load–solar PV curves (so-called duck-belly curves) from California. (Image courtesy of
lands. The state’s high PV penetration California ISO.)
has already created a real and unpredictable threat of
large-scale overvoltage situations. Transient overvoltage
events in Hawaii’s grid have resulted in damage to grid
equipment and customers’ home electronics. In such situ-
ations, advanced solar PV inverters must be used to curtail
solar generation produced at the household level and
maintain outputs that will not affect system stability.
California is experiencing a sudden and excessive rise
in power when rooftop solar systems are available online.
In 2013, the California Independent System Operator (ISO)
published a chart, now commonly known as the duck-belly
curve (Figure 2), which shows the difference in electricity
throughout the day. On 11 March 2017, the California ISO
duction for an hour to maintain grid stability. On 4 March
2018, the solar peak generation reached approximately
50% of the total peak demand in California, which intensi-
fied the curtailment level.
Curtailing renewable energies is counterproductive and
inhibits the achievement of environmental and economic
goals. A large amount of carbon-free electricity production
lost because of curtailment could otherwise be used to
maintain the reliability and stability of the grid. In this cir-
cumstance, the net load drops faster during the daytime,
and the afternoon-to-evening ramp-up grows rapidly steep-
er when there is no solar generation available. The problem
is at its worst in springtime, when the sun shines longer
each day, and fewer people are running their air condition-
ers because of temperatures being in the comfort zone.
In December 2017, electricity demand in California ramped
up by approximately 13 GW, which represents 50% of the
increase in demand, within three hours during the eve-
ning period. In situations like this, utility companies must
prepare themselves to satisfy the peak demand and also
that will operate during the period to supply the addition-
al demand. Thus, utilities are working with stakeholders
and state leaders to innovate solutions to the rising trend
2012 (Actual)
Increased
2013 (Actual) Ramp
2015
Potential
2017
Overgeneration
2020
0 3 6 9 12 15 18 21 24
Time (h)
IEEE Elec trific ation Magazine / D EC e mb e r 201 8 85
 of renewable curtailment and sudden load ramp prob-
lems. Energy-storage devices, along with solar PV panels
at the residential level and PEVs with their storage capa-
bilities, are considered emerging solutions to offset
these problems.
Solar PV-Storage Hybrid Systems
at the Household Level
Energy-storage devices can be used to store the excess supply
and so address some current grid management issues and
overcome barriers during the transition to a greener, more
cost-effective, and more efficient grid. The stored energy can
be used later when demand is higher and available renew-
able energy is smaller. The energy in storage devices can also
provide support for voltage and frequency regulation, conges-
tion relief, energy arbitrage, and network upgrade deferral at
the transmission and distribution levels. On the generation
side, storage paraphernalia can be utilized to provide synthet-
ic inertia to generators, peak capacity deferral, spinning
reserve, flexible ramping, and frequency regulation.
Unfortunately, in the past, significant cost and value barri-
ers were obstacles to the deployment of combined solar PV
and energy-storage systems. But with the development of dif-
ferent storage technologies (more specifically, static lithium-
ion batteries), the per-unit cost of devices is decreasing and
their lifetime is improving. As a result, residential-level stor-
age usage has recently started to grow. A significant uptrend
took place last year when investments in residential energy
storage scaled up. Global annual residential-level energy-stor-
age deployment is expected to scale up from 94.9 MW in 2016
to 3,773.3 MW in 2025, as reported by Navigant Research.
According to recent data, in the United States, nine grid-
connected home energy-storage systems were deployed
per day through the third quarter of 2017, totaling 4.2 MW
(amounting to a growth of 202% over the year). In Hawaii,
which has the highest electricity rates in the nation, solar
PV storage systems are making economic sense under the
state’s Customer Self-Supply program. In California, the
Self-Generation Incentive Program, with its consideration
for residential storage systems, has helped to drive growth.
5
4 Gas and Electric TOU rate structure is indicated in Figure 4.
Power (kW)
3 The heuristic approach (heuristic control) and DDDP- and
2
1 trol strategy reduced electricity purchase costs per day be­­
0 cause of the uncertainty of solar generation (Figure 5). The
0 4 8 12 16 20 24
Time (h)
Household Load Demand
Forecast Solar Power
Figure 3. The household and solar generation profiles for a summer day.
86 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018
Other states and utilities are trying out models to reduce
the cost of batteries and power electronics over time.
The residential segment is expected to contribute about
10% of the storage market in 2018, and this value is expect-
ed to grow to 38% by 2022, as predicted by GTM. In most
cases, homeowners are adding storage devices for backup
power or monetary savings, while utilities encourage the
deployment of the technology to mitigate the effects of
high solar penetration on the grid. Today, U.S. residential
energy storage is concentrated in a few markets with the
right mix application and regulatory environment, such as
Hawaii, California, and the Northeast region.
Storage Device Control at the Household Level
Heuristic approaches are commonly applied for storage
control, where the storage is charged when electricity
demand is lower than solar PV generation and discharged
when solar PV generation is lower than demand. As most
customers pay electricity bills by following cost structures
based on a time-of-use (TOU) rate, the reduction of elec-
tricity purchase costs is possible by controlling the storage
device’s charge/discharge scheduling by using dynamic
programming (DP)-based advanced control strategies. DP
approaches ensure that the global optimal path is reached
considering all of the possible paths, which is exhaustive
(for more details, see Hafiz et al. 2018).
A specific algorithm for DP is deterministic dual DP
(DDDP), which considers a deterministic forward path. For
systems involving stochasticity, such as those with solar
PV generation and household load usage, stochastic dual
DP (SDDP) methods can be used. SDDP helps to avoid DP’s
curse-of-dimensionality problem by constructing an
approximation of the future cost function with piecewise
linear functions. For convergence, a stopping criterion
based on the desired level of a variable, such as the stor-
age state of charge, is considered. These DP-based control
strategies will always outperform heuristic approaches.
The impact of such advanced storage control strategies
is shown in Figures 3–6 for a single-household energy-use
case study (for more details, see Hafiz et al. 2017). A solar
PV installed capacity of 3 kW, along with a 4-kWh energy-
storage device capacity, is considered for the home. The
household load and solar generation profiles considered
for the case study are shown in Figure 3, and the Pacific
SDDP-based control strategies were applied, focusing on
reducing electricity purchase costs. The SDDP-based con-
use of solar energy and peak savings also improved with
the application of SDDP (Figure 6).
Increasing Deployment of PEVs
At the residential level, another increasing form of energy
storage is PEVs. Such vehicles use electricity to charge the
 10
9.8

Purchase Cost (US$)

TOU Rate (US$/kWh)
Off Peak 9.8
0.3 Partial Peak 9.6
Peak 9.4

Electricity
9.4
0.2 9.2
9.0 8.9
0.1 8.8
8.6
0 8.4
0 4 8 12 16 20 24 Heuristic DDDP SDDP
Time (h) Control

Figure 5. The electricity purchase costs per day (in U.S. dollars) for
Figure 4. The TOU rate for summer. different control strategies.

on-board batteries instead of fuel (gasoline, ethanol, or

diesel), unless the vehicle is a plug-in hybrid vehicle with
100 97.3
a limited zero-emissions range. As mentioned previously, 87.4 88.1
PEVs also have the potential to transfer power to the grid 80
and so alleviate peak power demand and provide ancillary 60 48.1 48.7
services in a way similar to other energy-storage devices. 40 29.6
These PEV services are often said to have vehicle-to-grid 20
power transfer capabilities. An emerging issue with PEVs
0
is that, when a large number of vehicles are simultane- Heuristic Control DDDP SDDP
ously connected to the power grid, they may cause a sig-
PV Usage per Day (%)
nificant threat to its quality and stability. Peak Hour Savings per Day (%)
According to a recent NREL report, the number of PEV
users in California will grow to approximately 1.3 million by
Figure 6. The improvements in solar PV generation usage and peak
2025, and annual sales of PEVs in 2025 will increase by hours savings.
more than 7% compared to 2017. Connecting and charging
the large number of PEVs at the household level will signifi-
cantly increase the system’s electricity demand within a 19.7
Summer Winter
Purchase Cost (US$)

short period of time and require reconfiguration planning 20

14.3 14.1
for the power distribution infrastructure. Based on the Elec- 15
9.2
Electricity

tric Vehicle Infrastructure Projection toolbox developed by 8.5 8.2

NREL, residential charging demand will be almost 900 MW
at 8:00 p.m. each day by 2025 in California. On weekdays,
the aggregate demand from all charging types (commercial
and residential customers) will face an increase of approxi-
mately 500 MW between 4:00 and 7:00 p.m., with a maxi-
mum demand of nearly 1,000 MW before 8:00 p.m.
PEV Charge/Discharge Control
at the Household Level
The distribution system’s modification and upgrading
costs can be alleviated by coordinating the charging behav-
ior of PEVs using either centralized or decentralized control
schemes. To apply these techniques, an effective commu-
nication system is required, but owner privacy is an impor-
tant issue that must also be addressed. The storage in the
PEV provides various opportunities at the household level,
such as electricity purchase cost minimization, voltage reg-
ulation, and load regulation. In this regard, a controller can
regulate the PEV’s charge/discharge actions considering
different objectives. Typical owners would like to reduce
their electricity purchase cost per day by controlling their
PEV’s storage device. If homeowners can be assured that
they will have the required amount of charge in the PEV
10
5
0
Heuristic DDDP SDDP
Control
Figure 7. The electricity purchase costs per day in U.S. dollars for
different PEV control strategies in different seasons.
when leaving home, then, with an effective control strate-
gy, the electricity purchase cost per day can be reduced.
Daily electricity purchase costs with different control
strategies for different seasons are shown in Figure 7. For
this case study, the PEV considered has a battery capacity
of 85 kWh with a 10-kW charger. It can be seen that, if elec-
tricity demand uncertainty is considered when controlling
PEV storage, then electricity purchase costs are lower com-
pared to the deterministic control approach based on the
DDDP and heuristic control methods for all seasons.
Coordinated Control of Solar PV Generation,
Storage, and PEVs at the Residential Level
As PV-based storage and PEV-deployed households are
expected to increase in the next decade, both types of

IEEE Elec trific ation Magazine / D EC e mb e r 201 8 87

 storage devices can be used by employing a coordinated
control algorithm at the residential level that addresses
security issues and communication requirements. Coordi-
nated control of the two storage devices can provide bene-
fits to both the customer and the utility. Customers will be
able to control the available storage devices in their house-
hold according to their requirements and preferences,
thus providing independence, an important factor for vir-
tually everyone. The system overview with coordinated
control is shown in Figure 8.
For coordinated control, a mathematical optimization
model is possible, with the representation of the manage-
ment problem through decision variables, parameters,
objective function, and constraints. As an example, this
model can be tailored to represent the owner’s preference
and behavior. Following the modeling step, data related to
electricity demand, solar PV generation, storage device
characteristics, and other factors need to be considered. If
the goal is to control the storage devices for a particular day,
then day-ahead data on these parameters need to be fore-
cast. Based on these forecasts and optimal control deci-
sions, the actions or operating policies for the storage
devices on a particular day are defined. Depending on the
characteristics of the overall optimization model, this can
be handled by using off-the-shelf methods to solve linear
programming, mixed-integer linear programming, quadrat-
ic programming, and DP. All of these methods can be appli-
cable if electricity demand and solar PV generation are
considered deterministic parameters for a particular day.
But, in the real problem, both solar PV generation and
electricity demand are uncertain parameters. Also, for the
transactive energy market, electricity per unit cost is
uncertain. Thus, to ensure accurate coordination of
system resources, these uncertainty sources must be con-
sidered when developing a robust and realistic coordinat-
ed control scheme.
Stochastic models can be optimized through model
predictive control, the stochastic gradient descent method,
stochastic DP, SDDP, approximate DP, and machine learn-
ing and mechanisms. In all of these cases, it is crucial to
precisely model and represent the uncertainties associated
with the random parameters. Based on available electricity
demand and solar PV generation data, it is a good practice
to compute the possible deviations around forecast values
and integrate that information with the optimization pro-
cess. However, in real-time implementations for control-
ling storage devices’ inverters, rolling-­window time
horizons and rule-based control methods can be applied to
achieve better management results.
Benefits of Coordinating Resources
to the Customers
One of the benefits of combined PV–battery hybrid systems
and PEV battery usage at the residential level is energy
security during distribution- or transmission-level power
outages. If the grid experiences an outage for a few hours,
as in the case of bad weather or rolling blackouts due to
higher load demand in the grid, homeowners with PV–bat-
tery hybrids and PEVs can control energy usage and storage
in a coordinated way, considering priority, availability, and
outage duration. They also can be used to maintain power
quality by smoothing out voltage fluctuation and voltage
swell/sag, compensating for voltage harmonics distortion,
and mitigating voltage unbalance. Moreover, households
equipped with solar panels, along with energy storage and
PEVs, can be protected from many grid vulnerabilities.
The increase in a household’s electricity demand typi-
cally occurs when the power generated by solar PV drops

Solar PV Generation

Coordinated
Controller
Storage
House Utility Grid

Customer Benefits: Utility Companies Benefits:

• Reducing Power Outages
PEV • Minimizing Electricity Purchase Cost
• Participation in Transactive Energy
• Independent Energy Usage
• Load Regulation
• Reliability
• Reduction of Backfeed of Power
• Deferral of Assets Replacement
and Network Upgrade
• Maintaining Power Quality

Figure 8. A system overview with coordinated control at the residential level and the benefits at different levels.

88 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018

off at about the same time that PEVs plug into the grid as

Purchase Costs (US$)

users return home from work. To mitigate rising demand,
utility companies need to produce or purchase electricity 20

Electricity
at higher costs from more expensive power generation 15
units, which impacts market electricity prices. House- 10
holds with generation and storage devices can avoid elec- 5
tricity purchases from the grid during peak times and use 0
Summer Fall Winter Spring
the energy in the storage devices to minimize electricity
purchase costs. Coordinated control of available storage Heuristic Control
devices can reduce electricity purchase costs per day SDDP-Based Control on PV Storage and
compared to individually controlled methods. The elec- Heuristic Control on PEV Storage
Heuristic Control on PV Storage and
tricity purchase cost reduction achieved through coordi-
SDDP-Based Control on PEV
nated control between these two storage methods for Coordinated Control
different seasons is shown in Figure 9, based on a case
study (for more details, see Hafiz et al. 2018). Figure 9. The electricity purchase cost-per-day reduction for different
seasons.
Benefits for the Overall Power Grid of
Coordinating Resources
Spreading out household energy production/usage during 10
Household Profile
the day to prevent large spikes in electricity demand from
Average Power
the grid is highly beneficial to the utility company. While 8
Power (kW)
Regulated Profile
reduced electricity demand during daytime may lead to
overvoltage problems, a sudden demand increase during 6
the night may cause low-voltage problems. It is thus help-
ful for utility companies if the overall load of a particular 4
day is regulated so that any sudden decline of load or
ramp-up of electricity demand can be avoided. Coordinat- 4 8 12 16 20 24
ed PV–storage hybrids and PEVs can be used to regulate Time (h)
and maintain a daily average household load demand that
Figure 10. The regulated profile after control.
will help flatten the belly of the duck-belly curve, shown in
Figure 2, when many household power demands are flat-
tened with the coordinated control method.
3
The utility incentive mechanism can be developed to
2.5
encourage customers to use coordinated control to
Variance

2
maintain the average daily load in their household. The
1.5
impact of load regulation with the coordinated control
1
strategy is shown in Figure 10 for the household with the
0.5
same capacities of a PV–storage hybrid system and a PEV
0
as shown in Figure 9. If both storage devices are con-
No PEV and

Heuristic
Controlled

Only PEV
Controlled

Coordinated
Control
PV

trolled in coordination, then the electricity demand vari-

ance reduces when compared to its value obtained by
controlling them individually (Figure 11). The act of shift-
ing demand at the household level may be viewed as a
key alternative to alleviate issues in the distribution sys-
tem while allowing homeowners to often avoid expen-
sive peak load charges.
As distributed renewable energy resources increase in
the system, one scenario may be that, after a certain level
is reached, utilities will start to experience considerable
losses in revenue. As a result, consumers without genera-
tion might need to pay more to make up the difference.
Recent research recommends pursuing the idea of trans-
active energy to help mitigate such issues. Transactive
energy will allow prosumers to buy and sell electricity and
ensure the dynamic balance of supply and demand. Based
on this scheme, end-user participation can be encouraged
Figure 11. A comparison of different methods for load variation.
using various negotiation and bidding strategies for the
customers’ available capacity and resources until agree-
ment with the utility is reached. By coordinating the avail-
able storage devices owned by households, customers can
engage in this transactive energy market.
Another emerging issue is how to properly compensate
the utilities for providing their assets and providing for the
participation of customers in the grid and the transactive
energy market. Therefore, novel smart-pricing mecha-
nisms that take into account the complexity associated
with this environment are necessary.

IEEE Elec trific ation Magazine / D EC e mb e r 201 8 89

 An additional problem for the distribution grid is asso-
ciated with backfeeding solar PV generation and the
potential voltage violations. To avoid backfeeding, either
solar PV generation is curtailed by controlling the inverter,
or storage devices are used to store the energy surplus.
But, if the PV-based storage reaches its saturation level,
the energy surplus should be curtailed. Instead of energy
curtailment, the PEV’s storage device can be used for charg-
ing if it is connected to the household circuit during the
pertinent time period. Therefore, with coordinated control
in the system formed by solar PV generation, PV storage,
and PEVs, excess energy can be absorbed to avoid energy
curtailment, reduce reverse power flow, and be used at
appropriate times to optimize the use of the available
renewable resources.
System stability and frequency regulation are two
other aspects that must be properly ensured in power
grids. Utility companies control frequency through auto-
matic generation controls to adjust the grid frequency
based on a reference level. But, because of the frequent
variation of distributed energy resources, batteries, and
PEVs in the grid, the distribution-level frequency can expe-
rience significant oscillation and violation. At the distribu-
tion level, electricity demand and distributed generation
can be used to regulate frequency. Based on a defined fre-
quency signal, prosumers and consumers that interact in
the system can control the actions in their storage devices
and so limit frequency deviations, providing the grid with
a type of ancillary service. If homeowners can control
their storage devices in a coordinated way to maintain ref-
erence frequency, a fast response to utility frequency con-
trol can be achieved.
As the number of PEVs on the road increases, their
impact on the power grid infrastructure is likely to intensi-
fy. While there are relatively few PEVs presently circulating
in the United States, if a small number of them (e.g., eight
to 15 vehicles) are clustered in a given area dependent on
a single distribution transformer, the additional load
might reduce the life expectancy of that transformer by
up to 50%. Despite being a transportation mode that is
mostly not harmful to the environment, an escalated
deployment of these vehicles may produce negative
impacts on distribution system assets, such as distribu-
tion primary feeders, transformers, and secondary distri-
bution lines.
As a result of these impacts, modification and upgrad-
ing of distribution system components may be required.
This can be achieved by increasing distribution transform-
er sizes and adding new lines to the existing system, both
expensive solutions. Similarly, because of the distributed
generation during daytime, it is more frequent to observe
the switching of existing online tap-changers and capaci-
tor banks, which will reduce the lifetime of these compo-
nents. However, upgrades in the distribution network can
be deferred or postponed if resources are properly coordi-
nated to satisfy the needs at the residential level.
90 I EEE E l e c t r i f i c a t i on M a gaz ine / DECember 2018
Conclusions
The task of preparing the grid for the seamless integration of
solar generation, static storage, and PEVs while supporting
customer preferences and creating benefits for utility com-
panies is technically and economically challenging. Coordi-
nated control between these devices at the residential level is
important to both the customer and utility in several differ-
ent ways. Some previous studies in this area analyzed using
PEVs and evaluated possible reductions of solar PV genera-
tion at the grid level for large-scale penetration, but very few
studies have considered the design of control strategies to
properly facilitate coordination among them.
Several concerns arise, as simultaneous deployment of
solar PV generation, storage, and PEVs at the residential
level is likely to increase within the next decade. This arti-
cle compared different control strategies for various objec-
tives, highlighting the significance of coordinated control
approaches at the residential level for renewables, storage,
and PEVs. Coordinated control was shown to better pro-
vide benefits for customers and the distribution system.
Acknowledgments
We appreciate the funding support from the FREEDM Sys-
tems Center, Raleigh, North Carolina, and also thank the
National Science Foundation for support under grant
CyberSEES-1442909.
For Further Reading
F. Hafiz, D. Lubkeman, I. Husain, and P. Fajri, “Energy storage
management strategy based on dynamic programming and
optimal sizing of PV panel-storage capacity for a residential
system,” in Proc. IEEE Transmission and Distribution Conf. Exposi-
tion, 2018, pp. 1–9.
F. Hafiz, A. R. Queiroz, and I. Husain, “Multi-stage stochas-
tic optimization for a PV-storage hybrid unit in a household,”
in Proc. IEEE Industry Applications Society Annu. Meeting, 2017,
pp. 1–7.
F. Hafiz, A. R. Queiroz, P. Fajri, and I. Husain, “Charge sched-
uling of a plug-in electric vehicle considering load demand
uncertainty based on multi-stage stochastic optimization,” in
Proc. North American Power Symp., 2017, pp. 1–6.
F. Hafiz, A. R. Queiroz, and I. Husain, “Coordinated control
of PEV and PV-based storage system under generation and
load uncertainties,” in Proc. IEEE Industry Applications Society
Annu. Meeting, 2018, pp. 1–5.
F. Hafiz, P. Fajri, and I. Husain, “Load regulation of a smart
household with PV-storage and electric vehicle by dynamic
programming successive algorithm technique,” in Proc. Power
and Energy System General Meeting, 2016, pp.1–5.
Biographies
Faeza Hafiz (fhafiz@ncsu.edu) is with the FREEDM Sys-
tems Center, North Carolina State University, Raleigh.
Anderson Rodrigo de Queiroz (adequeiroz@nccu.edu) is
with the School of Business, North Carolina Central Uni-
versity, Durham.
Iqbal Husain (ihusain2@ncsu.edu) is with the FREEDM
Systems Center, North Carolina State University, Raleigh.

 dat e s a h e a d
2018
10–1 3 D E C E MB E R
SPEC 2018: IEEE Fourth Southern Power Electronics Confer-
ence, Singapore, http://spec2018.sg/
18–2 1 D E C E MB E R
PEDES 2018: IEEE International Conference on Power Elec-
tronics, Drives, and Energy Systems, Chennai, India, contact
Dr. Arun Karuppaswamy, akp@ee.iitm.ac.in, http://www
.ee.iitm.ac.in/PEDES2018/
2019
17–2 0 F E B R U ARY
ISGT 2019: IEEE PES Innovative Smart Grid Technologies
Conference, Washington, D.C., United States, contact
2019ISGT@ieee.org, http://ieee-isgt.org/
4–8 MAR C H
ESW 2019: IEEE IAS Electrical Safety Workshop, Jacksonville,
Florida, United States, http://www.ewh.ieee.org/cmte/ias-
esw/
17–2 1 MAR C H
APEC 2019: IEEE Applied Power Electronics Conference and
Exposition, Anaheim, California, United States, http://www
.apec-conf.org/
28 A P R IL – 1 M AY
REPC 2019: IEEE Rural Electric Power Conference, Blooming-
ton, Minnesota, United States, http://sites.ieee.org/repc-
2018/2019-save-the-date/
28 A P R IL – 2 M AY
IAS/PCA 2019: IEEE IAS PCA Cement Industry Conference,
St. Louis, Missouri, United States, http://cementconference
.org/
5–8 MAY
I&CPS 2019: IEEE IAS 55th Industrial and Commercial Power
Systems Technical Conference, Calgary, Alberta, Canada
11–1 5 MAY
IEMDC 2019: IEEE International Conference on Electric
Machines and Drives, San Diego, California, United States,
contact Dr. Avoki Omekanda, avoki.omekanda@ieee.org,
http://iemdc-conference.org/
Digital Object Identifier 10.1109/MELE.2018.2871320
Date of publication: 16 November 2018
2 0 –2 3 MAY
ICDCM 2019: IEEE Third International Conference on DC
Microgrids, Matsue, Shimane, Japan, http://power.aitech
.ac.jp/ICDCM2019/
2 1 –2 4 MAY
ISGT Asia 2019: IEEE PES Innovative Smart Grid Technolo-
gies Conference Asia, Chengdu, China, contact Prof. Qi
Huang, hwong@uestc.edu.cn, http://sites.ieee.org/isgt-
asia-2019/
2 7 –3 1 MAY
ICPE 2019–ECCE Asia: 10th International Conference on
Power Electronics and ECCE Asia, Busan, South Korea,
http://www.icpe2019.org/
1 6 –2 0 JUNE
PPFIC 2019: IEEE IAS Pulp, Paper, and Forest Industries Con-
ference, Jacksonville, Florida, United States
1 9 –2 1 JUNE
ITEC 2019: IEEE Transportation Electrification Conference
and Expo, Novi, Michigan, United States, https://itec-conf
.com/
4 –8 A UGUS T
GM 2019: IEEE PES General Meeting, Atlanta, Georgia, United
States, contact Matt Stryjewski, matthew.stryjewski@ieee
.org, http://pes-gm.org/2019/
1 3 –1 6 A UGUS T
ESTS 2019: IEEE Electric Ship Technologies Symposium,
Arlington, Virginia, United States, contact Dr. Scott Sudhoff,
sudhoff@purdue.edu, https://ests19.mit.edu/
For more information on upcoming conferences, please
visit the following websites:
IEEE Power & Energy Society: https://www.ieee-pes.org/
meetings-and-conferences/conference-calendar
IEEE Transportation Electrification Community: https://tec
.ieee.org/conferences-workshops
IEEE Industry Applications Society: https://ias.ieee.org/
events-conferences/conference-schedule.html
IEEE Power Electronics Society: https://www.ieee-pels.org/
conferences/all-upcoming-conferences

IEEE Electrific ation Magazine / d ec emb er 201 8 91
 Newsfeed

Dartmouth’s Formula Hybrid

Wins Coveted Award

he race car design com­­

T petition Formula Hybrid has

been awarded one of the
most coveted prizes in engineering
education, the Accreditation Board
for Engineering & Technology Innova­
tion Award. Launched by the Thayer
School of Engineering at Dartmouth,
Hanover, New Hampshire, in 2007, the
annual Formula Hybrid competition
challenges teams of undergraduate
and graduate engineering students to
collaboratively design and construct a
fuel-efficient hybrid or an electric-
only vehicle. Student teams work
throughout the academic year to plan,
design, and build their cars before
bringing them to the four-day compe­
tition in May at the New Hampshire
Motor Speedway.
More than 3,500 students from 80 Students preparing their car for technical inspection.
colleges and universities have partic­
ipated in Formula Hybrid since its
beginning. For many students, For­
mula Hybrid is their first experience
collaborating across disciplines, from
computer science to mechanical and
electrical engineering.
As a result of the competition, a
number of institutions have devel­
oped Formula Hybrid-specific curricu­
la to encourage cross-disciplinary col­
laboration. Formula Hybrid is part of
the Society of Automotive Engineers
(SAE) Collegiate Design Series and
regarded as one of its most complex
and dynamic competitions.

Digital Object Identifier 10.1109/MELE.2018.2871321

Date of publication: 16 November 2018 Students accessing data from a dynamic event.

92 I EEE E l e c t ri f i c a t i on M a gaz ine / DECember 2018

 Teams demonstrate their creativi­
ty and engineering skills as they
compete in aspects of vehicle design,
acceleration, handling, energy effi­
ciency, and endurance. Each year,
more than 100 volunteers, including
engineering professionals from spon­
sors such as Fiat Chrysler, Ford, Gen­
eral Motors, LG Chem, BAE Systems,
IEEE, SAE International, and Toyota,
support the students as mentors and
judges. After watching students in
action, many of the sponsors go on to
offer internships and jobs to Formula
Statistics for the Formula Hybrid competition. Hybrid participants.

IEEE Power & Energy Society

Scholarship Plus Initiative
ince 2012, Sch­­weitzer
s Engineering Laboratories,
Inc. (SEL) has consistently
invested in the IEEE Power & Energy
Society (PES) Scholarship Plus Initia­
tive, enabling the selection and rec­
ognition of nearly 100 SEL Meritori­
ous Scholars. SEL, its founder Dr.
Schweitzer, and his wife Beatriz are
passionate about supporting the
future talent of the power and ener­
gy industry. A pipeline of qualified
and work-ready power engineers is
(From left) Frank Lambert (PES president-elect), Miriam Sanders (SEL), Saifur Rahman (PES presi-
being generated as a result of their dent), and Karen Bartleson (2017 IEEE president) at the IEEE PES General Meeting.
historic support.
PES and the IEEE Foundation were
honored to recognize them recently 1,500 scholarships. The program con­
during the 2018 PES General Meeting at tinues to have a positive impact with­
the Visionary Level, celebrating SEL as in the industry by ensuring that more
the largest overall donor to the IEEE students engage and seek a career
PES Scholarship Plus Initiative. and assists in connecting them with path within the power industry.
The IEEE PES Scholarship Plus Ini­ relevant and meaningful career You can learn more about how
tiative awards highly qualified elec­ experiences. The program’s goal is to Scholarship Plus is successfully
trical engineering undergraduate increase the number of well-quali­ addressing the continued shortage of
students with multiyear scholarships fied, entry-level engineers to the power engineers in the power and energy
and energy industry. industry by visiting the IEEE PES
Since the program’s launch in 2011, Scholarship Plus website (www.ee-
Digital Object Identifier 10.1109/MELE.2018.2871322
Date of publication: 16 November 2018 more than 900 students have received scholarship.org).

IEEE Elec trific ation Magazine / D EC e mb e r 201 8 93

 V iI e
v EwWp
PoO iI n
N tT
Electrification and Challenging
Our Preconceptions
By Peter O’Connor and Michael Jacobs
t is difficult to over-
I state the impact of elec-
trification on society. The
earliest experiments advanced our
understandings of the physical world.
Electricity enabled the telegraph and
telephone, revolutionizing the trans-
mission of information. Electricity
has continued to transform commu-
nications with television, satellites,
the Internet, and cell phones. The
inventions of Thomas Edison and
many others transformed house-
hold life and entertainment. In
industry, electric lights improved
safety by replacing gas lamps. Indus-
trial motors and task machines
improved safety and efficiency by
replacing steam engines and belt-
driven tools. The electrolytic produc-
tion of aluminum and other materials
reduced costs. Overall, the U.S. Nation-
al Academy of Engineering consid-
ered electrification as the greatest
engineering accomplishment of the
20th century.
These energy transitions are not
solely the province of historians.
They continue today. Converting
more end uses from the direct ap­­
plication of fuel to the use of elec-
tricity has the potential to reduce
costs, reduce pollution, and impro­­­
ve performance.
Digital Object Identifier 10.1109/MELE.2018.2871324
Date of publication: 16 November 2018
94 I EEE E l e c t ri f i c a t i on M a gaz ine / december 2018
The Case for Electrification in developed countries and from
The environmental benefits of electri-
fication are often highlighted. Elec-
superior performance. These perfor-
mance advantages include cleanli-
tricity at the point of use does not ness and quiet for electric lawn tools,
produce carbon dioxide (CO 2) emis- superior acceleration and a smoother
sions, criteria pollutants such as sul- ride for EVs, and greater safety or
fur oxides or nitrogen oxides, or controllability in some applications.
particulate matter. While there may The remaining direct-energy end
be emissions from the generation of uses are those that are portable with
electricity, there are three primary high energy demand (such as vehi-
reasons why electri- cles) and for heating
fication may still be The U.S. National (stoves, furnaces, and
beneficial. First, power water heaters). Bat-
Academy of
plant emissions are tery technology con-
concentrated at a rel- Engineering considered tinues to advance,
atively small number electrification as the making EVs more
of facilities, where greatest engineering competitive on cost
they can more easily and range. Technolo-
accomplishment of
be measured and con- gies such as induc-
trolled. Second, the the 20th century. tive stovetops and
greater efficiency of high-efficiency heat
electric technologies pumps improve elec-
and the efficiencies of large-scale tricity’s appeal for other end uses.
power plants means that less energy Electrification is not simply fuel
may be required for the electric option switching. Oil lamps in the 19th cen-
even after considering transmission tury shifted from whale oil to coal oil
losses [this is generally true for electric to kerosene, but they were essentially
vehicles (EVs), as shown by the Union the same technology. Electric lights,
of Concerned Scientists (UCS)]. Third, even if ultimately powered by coal or
electricity can be produced from a wide oil, offered a fundamentally different
array of sources, many of them with experience. They could be used in
few or no emissions. Renewable energy places and applications that oil
has been on a remarkable rise in the lamps could not. Changing the con-
past two decades, and smart load man- version technology created a revolu-
agement can help mitigate problems tion in lighting, in business models,
caused by intermittency. and in the distribution of energy.
Electrification also benefits from Only those willing to look beyond the
the extensive existing infrastructure way things had always been done
2325-5987/18©2018IEEE
were poised to take advantage of
this. It is useful to consider how elec-
trification of other end uses might
challenge our preconceptions.
Supplies from Wind and
Solar Can Drive a Change
in Our Thinking
Along with the increased demand of
electricity for end uses, we see an
increasing use of renewable energy in
regions that are favorable to the growth
of wind and solar. These resources are
being developed in competition with
fossil fuel-based generation. Present­­
ly, the Southwest Power Pool, which
serves 11 states in the U.S. great plains
region, has sufficient wind generation
installed that instantaneous wind pro-
duction re­­ached 60% of the simultane-
ous electricity demand on 16 March
2018. For Texas, grid manager Electric
Reliability Council of Texas’ planners
projected solar development to com-
pletely dominate the construction of
new generation for years 2017–2031.
These levels of renewables offer an
electrification opportunity as periods
of high renewable energy output can
create apparent surpluses and periods
with very low wholesale energy pric-
es. Adding electric loads to fill such
periods will be socially and economi-
cally responsible. Various programs to
address the duck curve (Figure 1) in
California are reminiscent of utility
load-building and load-control pro-
grams of past de­­cades. Attractive
electrification strategies that use
abundant renewable energy supplies
but that was only ever a means to an One way to incorporate these
end. The true goal of efficiency efforts resources into the grid would be to
during the 1970s oil crises was reduc- encourage energy consumption at
ing expenditures on energy. The goal predictable times of renewable ener-
in recent decades has been reducing gy surplus with an approach such as
energy-related pollution. As Jevons time-of-use (TOU) rates or during
pointed out in the more unpredictable
mid-1800s, if a soci- Energy efficiency real-time events with
ety were actually run- programs like real-
will still enable lower
ning out of energy, time pricing or Pacif-
great­­e r efficiency costs, greater EV ic Gas and Electric’s
would not be a solu- driving distances, Excess Supply pilot.
tion. For­tu­nately, the and a reduced need Energy efficiency
United States is not will still enable lower
for infrastructure
in danger of running costs, greater EV driv-
out of energy. Primary expansion. ing distances, and a
wind and solar ener- reduced need for in­­
gy are exceedingly frastructure expan-
abundant as are uranium and other sion. But in some cases, an option that
resources. As the conversion technol- uses the least energy may not be as
ogy has improved, the actual amount advantageous as one that uses more
of wind and solar electricity has in­­ energy but incorporates storage.
creased so much that, at times, these
resources produce more than a local EVs Are Different
grid needs, even driving wholesale EVs present several challenges to
prices into the negative. A future of conventional thinking. Some re­­search
abundant clean energy requires us to efforts for EVs focus on making them
change our thinking on what it function as much like internal com-
means to be efficient. What would it bustion vehicles as possible. For ex­­
mean if electricity had effectively no ample, some companies developed
marginal cost? Consider the impacts battery-swapping technology so that
of zero marginal cost reproduction of an EV could be refueled in under
music and news on those sectors at 5 min, which is comparable to go­­
the turn of the 21st century. ing to the gas station. This was not a
30,000
25,000
System Load (MW)

to displace fossil fuel deserve great-

20,000
er attention.
Electrifying end uses and changing 15,000
energy supplies require thinking about
energy in off-peak seasons and off 10,000
hours, and not exclusively about load
5,000 Load Net Load
management in a few peak hours. A
future of abundant wind and solar 0
electricity may cause us to rethink the
.

.
m

m
a.

a.

a.

p.

p.

p.

purpose of energy efficiency.

0

00

00

00

00
:0

:0
4:

8:

4:

8:
12

12

A World of Abundant Energy

Energy efficiency might be character- Figure 1. The duck curve caused by solar power in California on 1 May 2018. The net load is
ized as an effort to use less energy, the remaining load after considering wind and solar generation. (Data courtesy of CAISO.)

IEEE Elec trific ation Magazine / d ec emb er 201 8 95

 viewpoint
particularly popular option. Others
have tried to develop business mod-
els based on operating EV chargers as
a stand-alone business, just like gas
stations. However, there is no guar-
antee that the same business model
will be viable.
Very few people have a gas station
at home, while a great many have an
EV charger—even a simple 120-V
outlet in the garage will replenish
typical daily driving needs overnight.
And a 240-V outlet for faster charging
can be simple to install. As a result,
EV drivers have a very different expe-
rience. They do not patronize a busi-
ness once a week to buy fuel for their
vehicle, and they charge them at
home the vast majority of time.
When they need a fast charge on a
long road trip, they might stop for a
half hour to an hour at a station paid
for by a car company, a state or utility,
or a grant of some kind. These char-
gers are often at shopping centers,
where a driver might spend that time
getting a meal; current fast chargers
supply around 50–150 kW. On a multi-
day road trip, an EV driver may select
a hotel because it has EV charging on
site. This is called destination charging
and could be a level 1 or level 2 sys-
tem. The distribution of charging
Public Inter-Metro
Public in Metro Areas
Fleets
Workplace
Multifamily Home
Single-Family Home
Figure 2. The charging pyramid representing necessary relative abundance of EV charging
systems. (Data courtesy of the New York State Energy Research and Development Authority.)
96 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018
infrastructure likely to be needed is
shown in Figure 2.
Although some companies have
announced their intentions to build
networks of even higher-powered
chargers (150–350 kW) that would
make charging an EV much more
like fueling a car, the current surge in
EV growth has occurred in a world in
which those experiences are quite
different. Many existing vehicles
have the battery capacity for a week’s
worth of driving but also have the
opportunity to charge every night.
So, they are sipping more than slurp-
ing electricity.
EVs also require rethinking distri-
bution system management. Because
of the significant load of a level 2 home
charger (often 7.2 kW or even higher),
there is considerable interest among
utilities in various strategies to ensure
that a large number of EVs on one
feeder do not overload the system.
The EV transition is occurring just
ahead of another set of related chang-
es. Shared autonomous EVs would rev-
olutionize transportation and present
different charging needs. Is society
ready for such a transition? It is too
early to say for sure, but it may be worth
considering what those charging needs
might be. One can envision charg­ing
dc Fast Charge
Level 1 and 2 and dc Fast Charge
Level 1 and 2 and dc Fast Charge
Level 1 and 2
Level 1 and 2
Level 1 and 2
hub stations with multiple high-pow-
ered chargers, possibly backed up with
stationary storage, located close to
major transportation routes. In the near
term, such stations could be used for
delivery or transit fleets.
Electrification Can Aid
Integration of Renewables
Traditional power system operations
rely on generator capacity reserves
and the ability to control the fuel con-
sumed in those generators. Power
plant responses to adjust to changes
in the power system are familiar to
engineers, although exact definitions
of the responses can vary. Examples
of power system controls include the
management of supply and demand
balance, frequency response, voltage
regulation, and reserved power for
the replacement of lost generation.
While several kinds of demand man-
agement are available to address
these conditions, most grid managers
view any manipulation of consumer
demand as a special case and pro-
foundly limited.
The needs for reserves and flexibil-
ity to maintain reliability is integral to
all electric power systems, regardless
of the presence of variable renewables.
No generator is 100% reliable, and
power systems must be continuously
prepared to respond if a large genera-
tor or transmission facility suddenly
fails. Exact rules vary from region to
region, but such contingency reserve
requirements are typically based on
the size of the largest generator. Each
balancing area or reserve-sharing pool
must keep enough spinning and nons-
pinning reserve ready to replace power
when a generator fails.
Most common strategies assume
that conventional generators pro-
vide these services, while changes
in the demand and the output from
variable renewable resources come
and go. However, careful analyses of
grid o p e ra t i o n s w i t h va r i abl e
generation and forward-looking
modeling of higher levels of wind
and solar show that initial assumptions
about variability and associated
costs are not proven. The increased
coordi­­nation and expansion of bal-
ancing areas, changes in scheduling
practices, and the use of renewable
energy forecasts have demonstrated
benefits in reducing anticipated costs
of integrating wind and solar. Much
has been learned as grid operators
have adopted strategies for integrating
renewables, with attention focused on
using the capacity available from cur-
tailed re­­newable generation as a
source of services when the ratio of
renewables to conventional genera-
tion is greatest.
The electric industry has gained
confidence regarding renewables
integration. John Moura, director of
reliability assessment and system
analysis at the North American Elec-
tric Reliability Corp., said in March
2018, “Variable resources can be reli-
ably integrated, but they need to be
cautiously planned and operated.”
“You can have 30%, 40%, 80% renew-
able resources, you just have to plan
and operate the system correctly,” he
said. However, the challenge of meet-
ing daily, seasonal, and annual peak
demands never ceases. Quick-starting
and fast-ramping resources become
more significant contributors to meet-
ing the daily peak following sunset
and the de­­cline of solar generation.
Electrification Offers
Load Flexibility
The growth in electrification offers a
larger set of solutions at the same
time the industry sees growth in wind
and solar generation. Meeting the
challenge of the combined daily in­­
crease in demand with the rise and
decline in solar generation (Figure 1)
calls for the development of new
thinking on both supply and demand
resources. Technology changes, in
both supply and demand, create new
pressures and new opportunities. The
changing supply mix creates more
variability in bulk power resources,
with implications for operators man-
aging the grid and for marginal prices.
Multiple factors, from economics to consumers and grid operations. Just
regulations to asset age, have in­­ as the hours with renewable energy
creased the use of renewable genera- surpluses allow lower-cost substitu-
tion and decreased steam generation. tion for fossil fuel in the dispatch of
The displacement of energy from generation, the same is possible in
synchronous generators with wind end uses. One of the trends in electri-
and solar energy production reduces fication is the need for flexible uses,
the online capacity of conventional or flexiwatts.
generation. More on–off cycling of Beneficial electrification is a term
thermal units and reliance on quick- used by various state agencies and
start gas units are observed in re­­gions national organizations. The National
with greater amounts Rural Electric Cooper-
of variable genera- Careful analyses of ative A s s o c i a t i o n
tion. The same condi- notes, “The wise use
grid operations with
tions are projected in of electricity, benefi-
studies of high-re­­ variable generation cial electrification,
newables scenarios. and forward-looking has s p a r k e d wide-
The expanded use of modeling of higher spread rethinking of
zero-fuel re­­newable policies that encour-
levels of wind and
generation and more age or mandate less
quick-starting fossil solar show that initial electricity use and
fuel units will create a assumptions about promote infrastruc-
more dynamic daily variability and ture planning. Ad­­
pattern of changes in vancements in electric
associated costs are
the marginal cost of technologies contin-
energy on a power not proven. ue to create new op­­
system than a system portunities to use
with fossil fuel as electricity as a substi-
the marginal fuel in all hours. tute for on-site fossil fuels like natu-
In various U.S. regions with orga- ral gas, propane, gasoline, and fuel oil,
nized wholesale electricity markets with increased efficiency and control.”
providing real-time prices, high levels The opportunities for consumer-
of renewable energy output already facing programs to communicate the
cause periods of zero-price energy. benefits of flexible electrification uses
Generally, this is a sign of oversupply. and then operate the end-use equip-
Any market with a supply surplus that ment are limited only by the imagina-
can be sold cheaply will lead to re­­ tion and availability of new metering
sponses to make more use of that technology and Internet connections
supply (i.e., further electrification) but to loads. Traditional approaches for
also adjustments to reduce the sur- offering lower rates, with off-peak
plus. An important change in thinking rates available to customers that par-
among power system operators is rec- ticipate in a load control program, are
ognizing that this is a signal for greater widespread. Customers agreeing to
flexibility from conventional genera- interruptible service to an end use
tion. As discussed next, more flexible that is not sensitive to when it is
operating assumptions and reliance powered pay lower rates.
on nonconventional generation for An example of this in rural areas is
reserves can alleviate the oversupply. applied to water pumping. During
peak times, farmers’ irrigation sys-
Loads with Flexibility tems are interrupted for approximate-
Increasing Roles ly 3 h, three or four times a month.
Where electrification strategies use Farm customers with off-peak rates
this pattern of lower system margin- are reported to save 20% on their elec-
al prices, benefits are increased for tric bills. Similar strategies for more
IEEE Elec trific ation Magazine / d ec emb er 201 8 97
 viewpoint

types of end uses and with the coor- transportation and would be sitting depressed the market for grid servic-
dination linked to renewable ener- idle most of the time. es. Even though vehicles and flexible
gy production have already been This solution is technologically fea- loads can provide many of the same
deployed. sible. Another solution, providing fewer services as dedicated storage, only
Beginning in 2010, New Brunswick benefits but also lower implementation one type of flexible one-way load is
Power illustrated an approach based on costs, is smart charging, varying the actually classified as energy storage
wind integration. Its PowerShift Atlantic one-way flow of po­­wer into the vehi- (ice systems for commercial air con-
program demonstrated cle to provide grid ser- ditioning). Flexible loads will com-
how Canadian Mari- The expanded use of vices such as demand pete against each other and against
time utilities used heat zero-fuel renewable re­­sponse or frequen- dedicated storage and flexible gener-
storage from residen- cy regulation. This is ation to constrain the prices for reli-
generation and more
tial and commercial sim­­ilar to water heat- ability services.
loads. “New Brunswick quick-starting fossil ers or space heaters, As a result, many utilities see
Power is uniquely fuel units will create a in that electricity does TOU rates as the best near-term
positioned because we more dynamic daily not flow back to the option for monetizing the value of
have 250,000 hot water grid. Therefore, it does load flexibility. This approach is well
pattern of changes in
heaters rented from us not require intercon- established and carries low transac-
by our customers,” the marginal cost of nection agreements. tion costs. Existing EVs can already
said Program Director energy on a power Smart charging and schedule charging. Numerous utili-
Michel Losier. system than a system vehicle-to-grid ex­­ ties offer TOU rates for residential
Cooperatives and change together are customers, with some limiting it to
with fossil fuel as
municipal utilities have considered vehicle- households that have EVs and oth-
long histories of con- the marginal fuel grid integration (VGI) ers trying out EV-only TOU rates. San
trols on electric water in all hours. strategies. Even with Diego Gas & Electric is conducting a
heaters. New technolo- its relatively lower pilot with day-ahead hourly prices.
gy companies have ex­­ implementation cost, Apps can automatically calculate
panded the idea to offer frequency smart charging has not been cost- the optimal time to charge, given
regulation from electric resistance water effective in most pilots. There are sever- the need for a full charge by a speci-
heaters. Water heaters with communi- al reasons for this. fied time.
cation interfaces and heat pumps xx Soft costs such as market partici- TOU rates can mitigate the most
instead of resistance elements demon- pation fees have a proportionate- significant impact of variable renew-
strate new opportunities. ly greater impact on small-scale able energy to date: the duck curve.
pilot projects. This might entail a low-rate block in
Experience with xx Rate structures in many areas the middle of the day, to be taken
Load Flexibility incentivize EVs to charge outside advantage of by residential water
The load flexibility of EVs, water of peak times, essentially creating heaters, commercial air conditioners,
heaters, and space heaters can pro- preemptive demand response and workplace chargers for EVs.
vide significant value to regional rather than market-participating TOU plans do have some transac-
electricity grids. Challenges remain demand response. tion costs, such as the installation of
in monetizing this value. xx VGI requires custom software smart meters. This has already been
Scientists described the value of integration with legacy systems done in many regions of the country,
EVs to a renewables-heavy grid well of the utility or grid operator. and even in those regions, there
before there was any significant xx Rules and technologies for mar- would be an additional cost to add a
number of such vehicles on the road ket participation were developed second smart meter solely for the EV.
or significant amount of variable for a relatively small number of As a result, EV-only TOU plans are not
renewable energy. Anticipating both large systems. heavily subscribed. California is
the problem of intermittency and a xx The flexibility of the existing grid exploring the use of meters embed-
potential solution, Dr. Willett Kemp- leads to low prices for grid services. ded in the EVs or charging stations
ton outlined how vehicle-to-grid Increasing the market penetration for submetering, allowing EV-only
power exchange could allow vehicles of renewables has not increased the TOU rates without installing a second
to function as grid batteries; the prices for services such as demand utility meter. This requires changing
major advantage over dedicated response or frequency regulation. our thinking about the way things are
grid batteries would be that the California, the site of many VGI pilots, done. While normally a utility must
vehicles were already purchased for also has a storage mandate that has use a utility-grade meter, there could

98 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

be a viable argument for allowing
submetering with a different system.
Even if not as precise as a utility
meter, as long as the error was nor-
mally distributed, submetering should
not be a disadvantage to either the
customer or the utility.
Studies of the value of load flexibil-
ity, like those of energy storage,
emphasize the need to stack multiple
value streams. One potentially valu-
able service is the deferral of distribu-
tion system upgrades. If many EV
chargers were added to a residential
neighborhood circuit and no attempt
was made to manage the charging,
the circuit could quickly become
overloaded. The utility would need to
upgrade the local transformer (and,
with enough such systems, the sub-
station). However, if charging was
managed in a way to reduce the
instantaneous load from all these
systems—such as by having them
communicate to stagger their charg-
ing—the utility would not need to
upgrade the distribution network as
soon. This deferral has a monetary
value. But because the utility can go
to a regulatory body and request
approval to include needed distribu-
tion upgrades in its rate base, there
is often no impetus to employ load
management solutions. The Brook-
lyn-Queens Demand Management
pilot is an exception. If a utility could
quantify the approximate value of
managed EV charging, it could offer
an incentive to customers who par-
ticipate in such a program.
A recurring concern with man-
aged charging of EVs is that a vehicle
may not be charged when its driver
needs it to be. None of the stakehold-
er in these programs would benefit
from such an experience: not the
utility, the charging station operator,
nor the vehicle manufacturer. Ac­­
cordingly, driver needs are the high-
est priority in any managed-charging
program. Prioritizing driver needs
has meant that, in some cases, pro-
grams were unable to obtain the nec-
essary load to participate in demand
response or frequency regulation
events. This is a consequence of the
small size of pilots to date. A fre-
quency-regulation provider could
better predict the availability of a
fleet of 100,000 vehicles than a fleet
of 100 vehicles.
There is not enough renewable
energy on the grid to necessitate large
amounts of demand flexibility. When
such a need does arise, it will show
itself in the prices of grid services
such as demand response and fre-
quency regulation. EVs are not plen-
tiful enough to provide
significant demand flex-
ibility, but their numbers
One potentially
are growing ra­­pidly. If valuable service
large amounts of de­­ is the deferral of
mand flexibility become
distribution system
necessary to manage
the variability of renew-
upgrades.
ables, then EVs, water
heaters, and air condi-
tioners can do that at a relatively low
cost. They will compete with each
other and with dedicated storage and
flexible generation. This will drive
down the cost of incorporating re­­
newables into the grid. That is a great
thing for the electricity system,
although it is a challenge for those
seeking to develop a business model
based on demand flexibility.
Pathways to More Renewable
Energy with Electrification
Renewable energy goals are the most
common and influential state and
private sector renewable energy poli-
cy. These define a requirement for
a percentage of energy used to be
provided by renewable generation.
This policy framework adjusts the
amount of renewable energy pro-
cured to keep up with electrification
and overall growth in electricity use
on the system.
An analysis by UCS of the 50%
renewable portfolio standard (RPS) in
California demonstrates the impor-
tance of using electric end uses for
grid-support services. Predictions
were circulating that significant
IEEE Elec trific ation Magazine / d ec emb er 201 8
amounts of wind, and especially
solar, would be curtailed in a high-
renewables scenario for California.
These analyses did not make chang-
es in grid planning and operations. To
estimate how advanced demand
re­­sponse and electricity storage can
reduce greenhouse gas emissions
and production costs, UCS conduct-
ed power system modeling using
PLEXOS simulation software.
The UCS model used more than
two years of hourly data on wind and
solar power production, electricity
demand, require-
ments for follow-
ing reserves, and
generator outages.
The model repre-
sented the flexi­
bility of individual
conventional gas-
fired power plants
(42 combined-cycle
gas units with a total capacity of
15.1  GW and 124 combustion tur-
bine/peaker units with a total capaci-
ty of 6.7 GW). This represented an
up-to-date projection of the Califor-
nia Independent System Operator
(CAISO) gas fleet in 2024. The model
included six following and contin-
gency reserve products load following
up, load following down, regulation
up, regulation down, spinning, and
nonspinning, with each modeled as a
single CAISO-wide product.
Without any changes in thinking
and operations, this study simply
repeated the findings of others. As the
fraction of energy provided by re­­
newable generation increases from
25%, grid operators’ practices would
expand the amount of curtailed re­­
newable energy. In the 50% RPS base
scenario, substantial levels of gas gen-
eration and renewable curtailment
coincide during the daytime hours.
During these hours, the amount of
curtailed renewable energy could
replace up to 6 GW of electricity from
gas, if the grid services provided by
natural gas facilities could be provid-
ed by other sources.
99
 viewpoint
So by changing the assumptions electricity. Consider a water heater, a
about grid operations, UCS was able to potential area for electrification with
show a means to increase renewable heat pump technologies, and a source
energy utilization. It added 1  GW of of advanced demand response. The
advanced demand response, 1 GW of purpose and service of the water heater
additional electricity storage, and the is to keep a tank of water hot by staying
ability to export 1 GW (net) of electric- within a certain temperature range. If
ity in any hour. This a water heater is in the
enabled large cuts middle of its accept-
Many devices and
in renewable curtail- able t e m p e ra t u re
ment, greenhouse gas processes can provide range and not actively
emissions, and pro- advanced demand heating, it could 1) not
duction costs. Com- response by changing operate because its
pared with the 50% temperature is al­­ready
when they consume
RPS base s c e n a r i o, adequate or 2) begin
this amount of non- electricity. heating in re­­sponse to
generation flexibili- a signal from the grid
ty reduces renewable operator to increase
curtailment by 70% (from 4.8 to 1.4%), demand, thereby producing a reserve
CO 2 emissions by 5.6% (or 2.3 mil- product such as regulation down or load
lion metric tons of CO 2 per year), following down.
and production costs by 4.4% (or The UCS study provided indica-
US$205 million per year). To achieve tions of value and need of grid ser-
high penetrations of re­­newable vices by pricing the services. By
e n ergy, it will be critical to have increasing the amount of nongenera-
smart end uses of electricity that can tion flexibility, the share of both
provide many grid reliability ser- upward and downward reserves sup-
vices that are produced by conven- plied by nonfossil resources also grew.
tional natural gas plants today. However, the downward reserves
issued by these sources were more
How Electrification Can central to reducing curtailment than
Increase Grid Reserves and the upward reserves.
Create System-Wide Benefits Without substantial amounts of ad­­
Increasing end-use electrification ditional nongeneration flexibility, the
with flexible demand will provide price of downward reserves was much
multiple benefits to the power sys- higher than that of upward reserves.
tem, delivering greater utilization of The very high price of providing down-
available renewable resources. These ward reserves—especially during hours
benefits can be measured in objec- of frequent curtailment—shows that
tive terms beyond the direct substitu- doing so during these hours is difficult.
tion of fuels at the end use. During hours of frequent curtailment,
Controllable electric end uses can the price of downward reserves consis-
­provide online reserves without si­­ tently ex­­ceeded the US$100/MWh cost
multaneously generating electricity. of re­­newable curtailment, suggesting
This is why this category of resources that the provision of these reserves is
is more effective than natural gas causing curtailment. Downward re­­serves
generation in reducing renewable cur- are a driving factor in renewable cur-
tailment, greenhouse gas emissions, tailment; providing enough nonfossil
and production costs. sources of flexibility can largely address
Many devices and processes can the need for downward reserves.
provide advanced demand response Controllable end uses could be
by changing when they consume aggregated to provide grid-scale
100 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018
demand response in CAISO markets.
This would need to occur in a way
that meets operational requirements
for providing reserve capacity, includ-
ing sustained response over a given
time. Electricity storage devices or
other forms of advanced demand
response—such as EV charging, air
conditioning, and water pumping—
could provide reserves in an analo-
gous manner.
Conclusions
The electrification of additional ener-
gy end uses can reduce pollution,
allow greater utilization of existing
grid assets, put downward pressure
on rates, and provide superior perfor-
mance and enhanced quality of life. It
also has the potential to transform
society, just as the telegraph, electric
light, personal computer, and cell
phone did. All of these transitions
had the need for creative thinking in
common. New technologies are not
just like the ones they replaced, save
for fuel-switching. They have differ-
ent capabilities and enable chang-
es in the way things are done. This
might entail a broader use of TOU
rates for residential customers, allow-
ing submetering from nonutility
meters embedded in vehicles, vehicle
manufacturer investment in charging
infrastructure, or ISO consideration of
demand-side technologies for reli-
ability services.
Energy transitions require us to
rethink our preconceptions and re­­
consider much of what we take for
granted. Although we cannot say
exactly what impacts these transi-
tions will have, they require us to
imagine and prepare.
Biographies
Peter O’Connor (poconnor@plugin
america.org) is with Plug In America.
Michael Jacobs (mjacobs@ucsusa
.org) is with the Union of Concerned
Scientists.

 IF YOU ARE READING THIS...
YOU BELONG WITH IEEE POWER & ENERGY SOCIETY
Our local chapter activities, live events, educational workshops and webinars connect you
with your peers while keeping you informed of the latest technical developments.

PES members enjoy...

• Subscriptions and Discounts to IEEE PES Publications

• Discount Registration to IEEE PES Conferences and Live Events
• Premium access to our PES Resource Center where you will find over 1,400 tutorials,
videos, technical publications, archived webinars and presentations.

Plus, when you become a member, you’ll receive complimentary membership in your
local chapter and numerous opportunities to contribute to the future of power & energy.

Join the global leaders in power and energy who are bringing More Power to the Future™

Learn more: www.ieee-pes.org

Digital Object Identifier 10.1109/MELE.2018.2880356

 technology leaders
Peer-to-Peer Energy Market
for Community Microgrids
By Pablo Baez-Gonzalez, Enrique Rodriguez-Diaz, Juan C. Vasquez, and Josep M. Guerrero
n the last decade, new
i energy policies have boost­
ed the integration of re­­
newable energy sources (RESs) into
the electrical power system in an
attempt to reduce fossil fuel con­
sumption and move toward a cleaner
and more sustainable energy system.
A good example of this is the 2020
and 2030 energy strategy of the Euro­
pean Union. The established goals of
this policy require a 40% cut in green­
house gas emissions and at least a
27% increase of shared renewable ener­
gy consumption and a 27% increase in
energy savings.
The introduction of distributed
generation systems and the im­­
proved efficiency of energy storage
systems (ESSs) have brought about a
change in the paradigm of the energy
market. Energy delivery paradigms
have moved from a vertical unidirec­
tional generation-distribution-con­
sumption scheme to a much more
horizontal one, in which flows are
bidirectional between and within lev­
els. With this change, the definition
of consumer becomes obsolete and is
replaced by that of prosumer. Howev­
er, because of the historical architec­
ture of the electricity network, it is
the distributors who still have the
power to define the conditions under
Digital Object Identifier 10.1109/MELE.2018.2871326
Date of publication: 16 November 2018
102 I EEE E l e c t ri f i c a t i on M a gaz ine / december 2018
which prosumers can sell energy. In
fact, to date, they can only carry out
these transactions with the distrib­
utor itself and not with other pro­
sumers, even if they are in the same
geographical vicinity.
The inability to trade freely with
other prosumers can also be consid­
ered one of the causes for the emer­
gence and expansion of microgrids.
Microgrid-based electrical power sys­
tems ease the integration of RESs and
ESSs together with loads at the con­
sumption level, aiming to increase shares
of renewable energy generation, reli­
ability, and efficiency (Figure 1).
Reliability is potentially increased
by the microgrid concept definition
because a microgrid can operate isolat­
ed from the main power grid. There­
fore, when there is a grid fault, the
local load can be supplied locally from
the RESs and ESSs.
Energy distribution efficiency is
inherently increased by bringing gener­
ation closer to consumption and, there­
fore, reducing distribution losses in the
main power grid. On top of this, to fur­
ther increase the electricity distribution
efficiency, dc-based electrical power
systems can be used. Overall, dc distri­
bution systems are inherently simpler
(e.g., there is no need for frequency
synchronization) and more efficient
(e.g., there is no reactive power loading
the lines, and slightly higher voltages
can be used for the same isolation
capability of the conductors).
In our vision, energy becomes a
commodity for prosumers, who can
use it for their own consumption and
also trade their surplus, either in real
time or on a deferred basis if they
have storage systems. Extending the
marketing possibilities for the sur­
plus energy of distributed producers
could make a significant contribution
to encourage households and build­
ing owners to install this type of
renewable energy microgenerators.
The resulting increase in aggregate
supply, combined with free price
competition among the different pro­
ducers, should result in a reduction
in the average energy price within
the (micro)grid.
This behavior can be observed in
other markets, where the emergence
of digital platforms that enable direct
trading between end users has dis­
placed companies with a traditional
reseller role. In particular, several
reports claim that the emergence of
ride-sourcing companies—such as
Uber and Cabify—has brought both
direct (e.g., cuts in the average price
per unit of traveled distance) and indi­
rect (e.g., a reduction in alcohol-relat­
ed traffic accidents) benefits for users.
Similar studies show how the rise of
lodging platforms—such as Airbnb or
HomeAway—have also led to a reduc­
tion of around 10% (nonuniformly dis­
tributed between segments) in hotel
prices, which benefits not only the
users of those applications but all
2325-5987/18©2018IEEE
 RES

Household 1

Household 2

Household n
Psharedisc,2

Psharedisc,n
Psharech,2

Psharech,n
Utility Grid Psharech,1
Pgridabs,1 Ppv,1

ac Bus
Pgridinj,1 Load Manager
Psharedisc,1 Pbat
dc Bus
Data Bus Psl,1 Pnsl,1 Community
Energy
Input Profiles Management Battery

Controllable System
Scheduled Noncontrollable
Loads
Variables Loads
ESS

Figure 1. A microgrid-based community grid.

users. If the collaborative economy
has already shaken up sectors such as
transportation or lodging, what is
blocking its entry into the energy sec­
tor? And, more importantly, how can
these obstacles be overcome?
Can a Collaborative Energy
Economy Be Defined?
The definition of collaborative econo­
my is not yet consolidated, and this
term is often inconsistently replaced
by others such as sharing economy,
collaborative consumption, access
economy, or circular economy. In
fact, researchers consider that some
or all of these concepts could be
particular cases of two- or multi-
sided markets or platforms, although
there is no agreed definition for the
latter either.
From greater to less restriction,
the conditions that have been pro­
posed as sufficient for the exis­
tence of two- or multisidedness in
a market are as follows:
xx The existence of two-way or
one-way network effects: the
relationship b etween the
number of market participants
in each side and the total ben­ Figure 2. A conceptual mapping of sharing platforms.
efits obtained by that side isWhatever their name, collabora­
not proportional. tive consumption platforms can be
Price nonneutrality: the differ­
xx classified according to several criteria:
ence in what the platform charg­
xx by the type of product they trade
es to each participant according1)  recirculation of goods (market
to the side to which it belongs for secondhand or surplus
is important. goods)
Direct interaction: although the
xx 2)  an increase in the use of as­­
platform is necessary for match­ sets (market of productive
ing, end users retain control over factors)
the key terms of the transaction.
3)  exchange of services or labor
Affiliation to the platform: all par­
xx (productive time market)
ties have a fee for belonging to
xx by the established hierarchy among
the platform. the interactivity
1)   peer to peer (P2P)
2)   business to consumer (B2C)
3)  government to government
P2P
(in general, between public
administrations)
Commercial xxfor the orientation of its activity
True Sharing
P2P sharing 1)   for profit
(Sameride)
(Uber)
2)   not for profit.
NFP FP For any of the three types, tak­
ing into account the other classifi­
Commercial cation criteria, a matrix (Figure 2)
Empty Set B2C has been established that charac­
(Zipcar)
terizes the platform on which these
products are traded. The obvious
B2C question at this point is in which
quadrant would a platform be locat­
ed that would allow for the exchange

IEEE Electrific ation Magazine / d ec emb er 201 8 103

 technology leaders
of energy between prosumers within
the current distribution network? The
answer is important in both 1) defin­
ing the structure and functioning of
the platform itself and 2) the policies
for regulating it, with the aim of maxi­
mizing its in­­tended positive effects:
xx promoting the deployment of
renewable energy generation and
storage microsystems
xx reducing the average cost of ener­
gy to the end user.
The assumption is that a platform
in which energy (productive factor) is
exchanged cannot be anything other
than profit oriented (i.e., producers
receive monetary compensation for
the energy they provide), so it would
be in the right half of the matrix. To
carry out the agreed transactions, it is
necessary to have a physical network
on which to transfer the energy. It is
worth asking who must provide the
infrastructure that allows these prof­
itable transactions and under what
economic regulatory conditions will
this be feasible.
The most plausible option is to
use the distribution network main­
tained by a distribution network
operator (DNO), if one exists. To do
this, the DNO must also be interested
in enabling and/or participating in
a platform that will allow its sub­
scribers to exchange energy directly
among themselves.
An obvious argument against this is
the economic one. Usually, the DNO is
also the owner of the network and,
therefore, would bear the necessary
investment to adapt the technological
architecture of the network to allow
transfers between end users. Any addi­
tional costs of network op­­eration and
degradation associated with such
transfers should also, in principle, be
borne by the DNO. Therefore, the dis­
tributor can demand a percentage of
the profits obtained in each transac­
tion. If this fee is not sufficient to make
the adaptation and/or operation of
the network beneficial, the DNO may
choose to not enable the platform.
104 I EEE E l e c t ri f i c a t i on M a gaz ine / december 2018
However, the distributor may
also be interested in deploying and
using the platform for its own ben­
efit, for one or more of the follow­
ing reasons:
xx Regulation: direct transfers bet­­
ween end users, properly gov­
erned by the DNO, can contribute
to the stabilization and control of
the network in certain situations.
xx Economics: to meet peak de­­
mand, the distributor must pur­
chase additional energy from a
primary producer, usually at high
cost. Transfers between end users
could help to minimize both the
number of demand peaks and
their cost to DNO.
xx Anti-isolation: in the long term, if
the cost of a residential-based
microgrid [e.g., photovoltaics
(PVs) plus battery] for self-con­
sumption starts to be competi­
tive in terms of cost, many users
may choose to disconnect indefi­
nitely from the distribution net­
work to operate in isolated mode.
Conversely, the possibility of
doing business by selling the
excess production to other con­
sumers is an argument in favor
of not isolating oneself from
the network.
xx Future investments: in a hierarchi­
cal distribution network, increased
demand over time may force the
distributor to expand the capaci­
ty of the transmission network or
transforming equipment. Allowing
energy trading between end users
downstream of the area trans­
former could reduce the changes
needed. The rationale behind this
statement is that increased de­­
mand is met by increasing connec­
tivity between terminals in the
capillary network, rather than by
expanding the capacity of the lines
and nodes of the distributor’s
transport network.
The second, much less common,
option is for users themselves to
deploy and maintain the network. In
these cases, either because of geo­
graphical isolation (islands, rural
population clusters, and so on) or for
political reasons (to have a certain
degree of autonomy from the compa­
nies and/or the government), it is the
prosumers who share ownership of
the (micro)grid, operating it coopera­
tively. Of course, such a network may
have one (or several) points of con­
nection to interface with the public
network, given it exists.
In this case, the users receive all
the benefits obtained in each trans­
action, but the maintenance and
operation costs are also shared pro­
portionally among all network mem­
bers. Cooperation may not be limited
to costs but also extend to sharing
local production or storage systems.
For example, homeowners on an
island may choose to install commu­
nity wind turbines and battery banks.
In this case, the wind turbines partic­
ipate in the market by selling their
production in bundles to the best
offering prosumers. Similarly, the
storage system is another player that
sells storage capacity to the highest
bidder. This type of cooperative
microgrid may require the presence
of a microgrid manager, a centralized
entity that regulates the interactions
among end users, in this case, each
of the sources/loads comprised with­
in the microgrid.
In either of the two cases, assum­
ing that the appropriate conditions
are in place for the distributor (or,
alternatively, for the microgrid man­
ager) to enable an energy exchange
platform, what role should it adopt?
As one more peer (perhaps primus
inter pares) giving rise to a P2P sys­
tem? Or as a mere manager/interme­
diary as in a B2C model?
Benefits of P2P
A P2P system is a self-organizing sys­
tem of equal, autonomous entities
(peers) that aims for the shared usage
of distributed resources in a networked
environment avoiding central services.
P2P systems are a special type of dis­
tributed system and share a set of
issues that researchers have
addressed for years: security, equa­
nimity, anonymity, fault tolerance,
scalability, distributed processing, and
coordination. However, P2P structures
also present advantages: symmetric
role of peers, ease of scalability, het­
erogeneity of network members, dis­­
tributed control, and the ability to
function in dynamic environments.
Several of these advantages are
suitable for the case of the electrical
distribution system, seen as a grid of
microgrids. There are already some
proposals in the literature for what is
to be called the peerification of the
electricity market as well as some
successful business initiatives. In our
view, some of these proposals suffer
from two fundamental problems.
xx The assumption of altruism: to
consider common welfare as a
distribution optimization mecha­
nism for the energy surpluses of
individual prosumers is some­
what naive. Although there is a
growing environmental aware­
ness in society, it seems unrea­
sonable that producers would be
willing to accept a centralized
manager deciding to whom and
at what price to sell their surplus
Agent 1 Joins
the Market
Events
1,1
Tacc
e1
DDA
TDDA
1,1
Tneg
Tneg
e1
CDA
Figure 3. A transfer comparison for the same agent. Discrete-time double auction (DDA) versus continuous double auction (CDA). The qualifier
“double” indicates that both buyers and sellers can make offers to buy (bids) or sell (asks), respectively.
production, even if it maximizes
social welfare.
xxUse of discrete-time markets: those
works that do raise the establish­
ment of virtual markets for the
negotiation of transactions usually
use discrete-time auction models
such as those used for determining
the price of energy in wholesale
markets. In these kinds of markets,
while the transferring time is con­
tinuous, the auction timeline goes
through three states.
1) The bid submission period
begins immediately after each
instant of market session open­
ing. Any peer interested in trad­
ing must make an offer to buy
or sell from this moment on
and always before closing.
2) After each instant of session
closing, the matching proce­
dure takes place, during which
it is calculated who buys and
who sells and at what equilibri­
um price.
3) Once matching is complete, the
auction level remains inactive
until the opening at the next
session.
This implies that, since a peer
joins the market, it has to wait until
the next market session opens plus
the submission period and the match­
Agent 1 Leaves
the Market
1,2
Tacc
e2
TDDA TDDA TDDA
1,3
1,2 Tneg
e2 e3
IEEE Electrific ation Magazine / d ec emb er 201 8
ing period before it can start an ener­
gy transfer, assuming its offer was
accepted. The same applies after the
completion of each transfer. There­
fore, there exists a tradeoff bet­­ween
the perfect efficiency that discrete
auctions show in terms of potential
profit extraction and the maximi­
zation that continuous auctions
show in the use of potential trans­
fer time (Figure 3).
There is another issue that some­
how links markets using discrete-
time auctions with centralized P2P
systems and continuous auctions
with decentralized ones. When using
discrete models, an entity is needed
to carry out the process of matching
supply and demand. Such an entity
must be neutral (i.e., not participate
in the auction) because if it were not,
it could take advantage of the com­
plete information it has on the bids of
other traders for its own benefit. It
would seem logical that a DNO
should play this role; but in this case,
it is either unable to participate as a
trader, or, if it does, the condition of
anonymity and equanimity of the P2P
systems is not met. If, on the other
hand, a third party is chosen, it could
demand a fixed fee or a percentage
of the profits for the matching to be
carried out.
Traded
Energy
t
t
TDDA TDDA
t
Agent 1
105
 technology leaders
A Double Auction Market for
P2P Energy Trading Among
Prosumers
A CDA based market is proposed in
which each prosumer can alternate,
but not simultaneously, between the
roles of seller and buyer of energy. Sub­
mitted offers, whether they are bids or
asks, should include the following:
xxThe amount of power that the
seller (buyer) is bidding (asking)
for, which means that the auc­
tion allows multiunit trading.
xxThe price for that energy package.
xxInformation on the location of the
peer within the network topology
to accept the offer, the other peers
participating in the market must
be able to estimate potential trans­
mission losses by making use of
the position published by the bid­
der/asker and its own location
within the network. This informa­
tion is used to prefilter the mar­
ket, causing traders who are too
far away from the offering peer to
exclude themselves from that
offer (because transmission loss­
es would make the transaction
unprofitable).
0.13
0.12
0.11
Price (€/kWh)
0.1
0.09
0.08
0.07
06:00
Figure 4. An example of one-day price and market evolution.
106 I EEE E l e c t ri f i c a t i on M a gaz ine / december 2018
If, once the losses have been cal­
culated, the amount and price are
within any peer’s range of acceptabili­
ty, the transaction can be closed, and
the transfer of energy is immediately
initiated. This system includes the
right of the offering trader to termi­
nate an ongoing transfer if it receives
a better offer. However, the acceptor
retains the right of first refusal to
match, if it so desires, the new offer.
This would promote market conver­
gence toward its theoretical equilibri­
um price while ensuring that bidders/
askers, who take a proactive role
within the market, are not trapped in
suboptimal transactions.
If automating the participation of
different peers in the market is re­­
quired, autonomous trading agents
can be incorporated into peers’ ener­
gy management systems to
xx (re)define the market role of
the peer
xx adapt the offering prices accord­
ing to market evolution.
For this purpose, each agent car­
ries out a private energy assessment
for its corresponding peer based on
its estimated short-term consump­
Evolution of Dealing Price and Private Valuations
09:00 12:00
Time
tion profile and on the cost that the
consumed energy would have if it
were purchased from the utility (i.e.,
each peer estimates its average ener­
gy cost for the next TT period). From
this private valuation and the mini­
mum and maximum relative gain lim­
its (in the most generic case g min 2 L tx
and g max " 3), both selling and buying
price ranges are defined. These rang­
es are used to calculate the probabili­
ty of closing a transaction by acting as
buyer and seller based on market
transaction history. Finally, the role
with the highest probability of suc­
cess is chosen. To carry out the auto­
matic price adaptation, each trading
agent can implement any of the
autonomous adaptation mechanisms
existing in the literature. Figure 4
shows the intraday price evolution for
energy transactions in an example
CDA based market.
In summary, the establishment of
CDA-based markets for P2P energy
trading between prosumers makes it
more attractive for people to invest
in their own renewable energy pro­
duction and storage systems, thus
contributing to the expansion and
Closed Bids
Closed Asks
Buyers’ PVs
Sellers’ PVs
15:00 18:00
 generalization of the energy system.
At the same time, the possibility of
buying and selling energy surpluses
from distributed local production
allows a reduction in the average cost
of energy to those local consumers,
compared to the average cost if they
kept the utility as the only alternative
for the acquisition of energy.
For Further Reading
B. Brandherm, J. Baus, and J. Frey, “Peer
energy cloud—civil marketplace for
trading renewable energies,” in Proc.
IEEE 2012 Eighth Int. Conf. Intelligent Envi-
ronments, June 2012, pp. 375–378.
C. Codagnone and B. Martens.
(2016). “Scoping the sharing economy:
origins, definitions, impact and regu­
latory issues,” Institute for Prospective
Technological Studies Digital Econo­
my Working Paper 2016/01, JRC100369.
[Online]. Available: https://ssrn.com/
abstract=2783662 or http://dx.doi
.org/10.2139/ssrn.2783662
C. Giotitsas, A. Pazaitis, and V. Kosta­
kis, “A peer-to-peer approach to energy
production,” Technol. Soc., vol. 42, pp.
28–38, Aug. 2015.
A. Q. Huang, M. L. Crow, G. T. Heydt, J.
P. Zheng, and S. J. Dale, “The Future
Renewable Electric Energy Delivery and
Management (FREEDM) system: the
energy Internet,” Proc. IEEE, vol. 99, no. 1,
pp. 133–148, Jan. 2011.
E. Rodriguez-Diaz, J. C. Vasquez, and
J. M. Guerrero, “Potential energy savings
by using direct current for residential
applications: A Danish household
study case,” in Proc. IEEE 2017 Second Int.
Conf. DC Microgrids (ICDCM), June 2017,
pp. 547–552.
E. Rodriguez-Diaz, J. C. Vasquez, and
J. M. Guerrero, “Intelligent DC homes in
future sustainable energy systems:
When efficiency and intelligence work
together,” IEEE Consum. Electron. Mag.,
vol. 5, no. 1, pp. 74–80, Jan. 2016.
Biographies
Pablo Baez-Gonzalez (pbaez@us.es) is
with the Universidad de Sevilla, Spain.
Enrique Rodriguez-Diaz (erd@et
.aau.dk) is with Aalborg University,
Denmark.
Juan C. Vasquez (juq@et.aau.dk) is
with Aalborg University, Denmark.
Josep M. Guerrero (joz@et.aau.dk)
is with Aalborg University, Denmark.


IEEE POWER AND ENERGY

TECHNOLOGY SYSTEMS
JOURNAL

The IEEE Power and Energy Technology Systems Journal

is a peer-reviewed technical journal containing
practice-oriented articles.

Focus areas include the development, planning, design,

construction, maintenance, installation and operation of
equipment, structures, materials and power systems.

This is a fully electronic, open access, peer-reviewed journal

whose articles are accessed via the Internet. Both PES members
and non-members can access content without restriction (no cost).

Visit www.ieee-pes.org/pets-j to learn more

Digital Object Identifier 10.1109/MELE.2018.2880357

IEEE Electrific ation Magazine / d ec emb er 201 8 107

 ad index & sales offices

The Advertisers Index contained in this issue is compiled as a service to our readers and advertisers: the
publisher is not liable for errors or omissions although every effort is made to ensure its accuracy. Be
sure to let our advertisers know you found them through IEEE Electrification Magazine.

Advertiser Pag e U R L Phone

ELANTAS PDG, Inc. 3 Elan-Film.com

GIGAVAC 5 www.gigavac.com +1 805 684 8401

IEEE Electrification Magazine Representative

Erik Henson
Naylor Association Solutions
Tel: +1 352 333 3443
Fax: +1 352 331 3525
ehenson@naylor.com

Digital Object Identifier 10.1109/MELE.2017.2770698

Time to update your license?

IEEE PES

RESOURCE
CENTER
Continuing Education Credits

Need to train yourself in a specific area?

Video Tutorials and Webinars
EVERYTHING YOU NEED IN ONE PLACE
Unable attend a PES conference?
Videos and Presentations

POWER & Want access to in-depth technical materials?

ENERGY
CONTENT Complete Technical Reports

IEEE PES – More Power to the Future TM

The IEEE PES Resource Center is the most extensive library in the world devoted exclusively to the power and energy
industry. You don’t have to be a PES member to access the Resource Center’s vast array of content available to assist in
your research, presentations, or academic programs, and benefit your professional development!

PES Member Access: Most materials in the PES Non-PES Member Access: Discounted prices are
Resource Center are free to members. Members log-in listed for each product and non-members can use
with your username and password to download content. the shopping cart to purchase.

Visit the IEEE PES Resource Center: resourcecenter.ieee-pes.org

Digital Object Identifier 10.1109/MELE.2018.2880314

108 I EEE E l e c t r i f i c a t i on M a gaz ine / december 2018

 ESMO 2019
NEW HORIZONS
BOUNDLESS
POSSIBILITIES
Early Bird Registration Now Open
June 24-27, 2019 | Columbus, Ohio | ieee-esmo.com

Where Engineering Principles

Meet Field Realities
ESMO 2019 is set to be the industry’s largest “how to” event in North America. ESMO serves electric utility
professionals, contractors, and consultants who are interested in hands-on solutions for the safe engineering,
construction, operation and maintenance of the world’s power delivery systems. The four day event features
two days of outdoor field demonstrations and exhibits and a two-day technical conference program with
indoor exhibits.
IEEE Power and Energy Society and American Electric Power have partnered together and pooled their
resources to bring you this unique event where manufacturers and service providers connect with those who
engineer, design, operate and maintain the electrical network. PES and AEP envision this conference to be
the premier Safety event in the industry and have engaged T&D World to help make that happen.

For Information & Program Updates, Visit: ieee-esmo.com.

For Exhibit & Supporter Information, Contact:

Sam Posa | Exhibit Sales | 800.525.9154 | sam.posa@informa.com
Steve Lach | Midwest, Canada | 708.460.5925 | steve.lach@informa.com
Matt Wintringham | Eastern US | 508.277.8397 | matthew.wintringham@informa.com

General Information, Contact:

Digital Object Identifier 10.1109/MELE.2018.2880312 Eriks Surmanis | e.surmanis@pdc-cables.com
Digital Object Identifier 10.1109/MELE.2018.2880313