Renewable and Sustainable Energy Reviews 51 (2015) 648–661

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser

A review on electric vehicles interacting with renewable energy

in smart grid
Liansheng Liu a,b, Fanxin Kong b, Xue Liu b, Yu Peng a,n, Qinglong Wang b
a
School of Electrical Engineering and Automation, Harbin Institute of Technology, Harbin, China
b
School of Computer Science, McGill University, Montreal, Quebec, Canada

ar t ic l e i nf o a b s t r a c t

Article history: Electric vehicles (EVs) represent one of the most promising technologies to green the transportation
Received 22 November 2014 systems. An important issue is that high penetration of EVs brings heavy electricity demand to the power
Received in revised form grid. One effective way to alleviate the impact is to integrate local power generation such as renewable
17 April 2015
energy sources (RESs) into charging infrastructure. Because of the intermittent and indispatchable
Accepted 29 June 2015
nature of RESs, it becomes very challenging to coordinate EVs charging with other grid load and
Available online 15 July 2015
renewable generation. In this paper, EVs charging problem in the presence of smart grid technologies is
Keywords: investigated, and the interaction with renewable energy is reviewed. An overview about EVs and RESs is
Electric vehicle first presented, which mainly introduces major types of EVs and renewable energy estimation methods.
Renewable energy sources
Then, according to the objectives, the existing research works are divided into three categories: cost-
Smart grid
aware, efficiency-aware, and emissions-aware interactions between EVs and RESs. Each category's
Charging cost
Energy management discussion includes the description of core ideas, summarization of solutions, and comparison between
Emissions different works. Finally, some key open issues about EVs interacting with RESs are given and some
possible solutions are also discussed.
& 2015 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
2. Overview of electric vehicles and renewable energy sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
2.1. Electric vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
2.1.1. Battery electric vehicles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
2.1.2. Plug-in hybrid electric vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
2.1.3. Battery chargers for EVs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
2.2. Renewable energy sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
2.2.1. Wind energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
2.2.2. Solar energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
3. Cost-aware EVs interacting with RESs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
3.1. Minimizing operational cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
3.2. Reducing electricity generation cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
3.3. Increasing profit or benefit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
3.4. Minimizing charging cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
3.5. Other cost-aware works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
4. Efficiency-aware EVs interacting with RESs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
4.1. Application in microgrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655
4.2. Maximizing RESs utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 656
4.3. Other efficiency-aware works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
5. Emissions-aware EVs interacting with RESs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658
6. Open issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659

n
Corresponding author.
E-mail addresses: [email protected] (L. Liu), [email protected] (F. Kong), [email protected] (X. Liu), [email protected] (Y. Peng),
[email protected] (Q. Wang).

http://dx.doi.org/10.1016/j.rser.2015.06.036
1364-0321/& 2015 Elsevier Ltd. All rights reserved.
 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661 649

6.1. Large scale RESs integration into the grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659

6.2. EVs battery utilization for RESs interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
6.3. Feasibility of V2G supporting RESs interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659

1. Introduction
utilization. Most of them leverage EVs batteries as energy buffer-
ing, where charging EVs batteries when RESs production is surplus
Electric vehicles (EVs), especially plug-in electric vehicles, are
and discharging EVs batteries when the power supply is in
gaining more and more popularity due to the fewer emissions and
shortage.
lower oil dependency. The number of EVs in the globe would be
(3) Emissions-aware EVs interacting with RESs: These research
over 35 million by 2022 [1], and EVs sales per year in U.S. would
works discuss how to leverage the integration of EVs and RESs into
reach 2.4 million by 2024 [2]. However, deep penetration of EVs
the electric grid to further reduce emissions.
comes with a large magnitude of charging demand and causes
According to the above summarization, the interaction of EVs
heavy impact on the power grid such as branch and transformer
with RESs has been broadly studied. The objective of this paper
congestion [3]. One effective solution to mitigate the impact is to
aims at reviewing the related technologies in the presence of SG.
associate local power generation such as renewable energy
For any research, the objective is important and requires different
sources (RESs) with the charging infrastructure. For example,
methodologies. Following this note, three typical objectives of the
SolarCity provides solar power system for Tesla EVs charging [4];
interaction are investigated in this paper.
SunPower supports Nissan Leaf charging with solar power [5].
The rest of this review is organized as follows. Section 2
Charging EVs with RESs further decreases greenhouse gas
provides background for EVs and RESs. Sections 3–5 investigate
emissions and even reduces charging cost. However, RESs such
research works that belong to each category. Section 6 discusses
as solar and wind energy significantly vary over time because of
several essential open issues for EVs interacting with RESs pro-
the great dependence on weather conditions. They are usually
blem. Section 7 concludes the review.
characterized by intermittency and indispatchability [6]. These
characteristics make it very challenging to coordinate EVs charging
with other grid load and renewable generation. On the other hand,
smart grid (SG) technologies enable two-way communications 2. Overview of electric vehicles and renewable energy sources
between the power grid and power consumers such as EVs. In the
context of SG, the power grid can embrace a large scale of EVs with A general introduction of both EVs and RESs is addressed in this
intelligent charging coordination. Along this line, many recent section. For EVs, two types of plug-in EVs (PEVs), which are
research efforts have studied EVs interacting with RESs in the SG prevalent in current market, are introduced in detail, as well as
environment. This paper is confined to a detailed investigation of the corresponding battery chargers. For RESs, wind and solar
these works, and the researches from 2008 to 2014 are compre- energy are mainly studied, as well as the models applied for the
hensively classified and reviewed. It makes this review comple- prediction of electricity generation.
ment or different from other EVs coordination reviews [7–9].
Mwasilu et al. [7] describe a future grid system model and review 2.1. Electric vehicles
EVs interaction with it. Richardson [8] investigates the economic,
environmental and grid impacts of EVs. Su et al. [9] present a EVs are different from conventional vehicles by using electric
comprehensive overview of charging, communication and man- motors for propulsion, while the latter utilize an internal combus-
agement infrastructure. tion engine (ICE) for the same purpose. The electricity energy for
According to the objectives, these EVs and RESs coordination driving motors comes from three main sources: (1) generation
works are grouped into three categories, as shown in Table 1: plants, (2) on-board generators, and (3) on-board energy storage.
(1) Cost-aware EVs interacting with RESs: The research works in The third type of source is employed by most EVs. In recent years,
this category focus on reducing operational cost, electricity gen- much attention from industry and academia has been drawn by
eration cost, and charging cost , or how to increase profit for EVs with plug-in feature (e.g., PEVs), which is also the main focus
service provider. of this paper. Specifically, two types of PEVs are studied, i.e.,
(2) Efficiency-aware EVs interacting with RESs: The focus of this battery electric vehicles (BEVs) and plug-in hybrid electric vehicles
category is how to improve the efficiency of renewable energy (PHEVs).

Table 1
Related works in EVs charging interacting with RESs.

Category Related works

Cost-aware EVs charging interacting Zhang and Chen [10], Chen and Duan [11], Tan et al. [12], Verzijlbergh et al. [13], Zhou et al. [14], Shaaban and El-Saadany [15],
with RESs Aunedi and Strbac [16], Derakhshandeh et al. [17], Jin et al. [18], Hu et al. [19], Li et al. [20], Vayá and Andersson [21], Vasirani
et al. [22], Gunter et al. [23], Bouallaga et al. [24], Saber and Venayagamoorthy [25], Fernandes et al. [26], Khodayar et al. [27],
Dallinger and Wietschel [28], Zhao et al. [29], Goonewardena and Le [30], Chen and Tong [31], Ghofrani et al. [32], Wang et al.
[33], Saber and Venayagamoorthy [34], Kristoffersen et al. [35], Li et al. [36], Shen and Cui [37], Caramanis and Foster [38]
Efficiency-aware EVs charging Tushar et al. [39], Schuller and Hoeffer [40], Fazelpour et al. [41], Yi et al. [42], Gottwalt et al. [43], Wu et al. [44], Strnad et al.
interacting with RESs [45], Damiano et al. [46], Borba et al. [47], Zhu et al. [48], Li et al. [49], Battistelli et al. [50], Mets et al. [51], Pantoš [52]
Emissions-aware EVs charging Shaaban and El-Saadany [15], Aunedi and Strbac [16], Saber and Venayagamoorthy [25], Saber and Venayagamoorthy [34], Patil
interacting with RESs et al. [53], Shen and Cui [54], Saber and Venayagamoorthy [55], Lund and Kempton [56]
650 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661

2.1.1. Battery electric vehicles combine both power sources to operate as efficiently as possible.
The typical architecture of BEVs is illustrated in Fig. 1, which is Meanwhile, the controller can monitor the battery SOC level and
composed of three major parts (i.e., electric motor, rechargeable maintain it within a predetermined band.
battery, and controller). The electric motor utilizes rechargeable
battery as the energy source to generate propulsion [57]. A two- 2.1.3. Battery chargers for EVs
quadrant controller is only capable of managing the power Most PEVs utilize Li-ion battery as energy storage due to its
supplied to the electric motor which drives the vehicle to higher energy density, longer life span, and less environmental
move forward or backward. Meanwhile, a controller which can influence. However, it is not a simple issue to charge Li-ion
also support regenerative braking is known as a four-quadrant batteries, delicate control of voltage and current output of the
controller. charger is required. Otherwise, a large voltage and current fluctua-
Another important part of a BEV is the inverter, which is not tion may damage the battery. Therefore, advanced battery charger
shown in Fig. 1. An inverter is for converting DC power (the design can largely accelerate the development of PEVs.
electricity stored in the battery) to alternating current (AC) power, Currently, PEVs battery chargers are categorized as on-board or
since most of the electric motors equipped by EVs are AC motors, off-board. An on-board charger usually has restricted power level
which have simple schemes and low cost. due to its weight, space and cost limits [60]. Meanwhile, these
limits do not hold for an off-board charger. Besides, PEVs battery
2.1.2. Plug-in hybrid electric vehicles chargers can also be differentiated by whether their supporting
According to different types of powertrains (or drivetrains), power flow directions are unidirectional (charging power is only
hybrid electric vehicles can be classified into three categories: from the charger to the battery) or bidirectional (power can be
(1) parallel hybrid, (2) series hybrid, and (3) power-split hybrid. drawn from and also injected back to the grid) [61,62]. Usually, a
Among these, the type of parallel hybrid is commonly adopted, unidirectional charger has lower hardware requirements and
with its architecture shown in Fig. 2. simpler interconnection compared with bidirectional chargers
PHEVs are usually equipped with both an electric motor and an [63]. At present, most PEVs are equipped with single-phase
additional ICE for propulsion [59]. This mixed propulsion system chargers [64], which are commonly used for Level 1 and Level
enables PHEVs to be driven in two modes: charge depleting (CD) 2 charging [65]. However, there have been many researches
mode and charge sustaining (CS) mode. When operated in CD conducted based on the assumptions that bidirectional chargers
mode, PHEVs mainly draw energy from on-board battery packs. If are available as well as the technique of vehicle-to-grid (V2G) has
the battery state of charge (SOC) has been depleted to a pre- been realized [16,19,21,22,25–27,32,34,35,37,46,50,55]. V2G tech-
determined level, PHEVs will switch to CS mode and utilize the ICE nique envisions that electric vehicle can inject power back to
system for further propulsion. When operated in CS mode, PHEVs the grid.

2.2. Renewable energy sources

The increasing awareness of environmental crisis and fossil

energy depletion drives people to search for other promising
alternatives, among which RESs stand out for environmental
friendly and sustainable properties. In this section, background
overview of wind and solar energy (two most common types of
RESs), as well as several corresponding electricity generation
models are provided.

2.2.1. Wind energy

Wind energy is playing a more and more crucial role for
supplying renewable energy. According to the Global Wind Energy
Council, worldwide wind power capacity reached 318,137 MW in
2013 with an increase of nearly 200,000 MW in the past five years
Fig. 1. The architecture of BEVs [58]. [66]. The European Wind Energy Association has set a goal for
satisfying 23% of European electricity demand with wind energy
by 2030 [67]. To improve the efficiency of electricity generated by
wind energy, it is important to choose beneficial locations for
wind farms.
The first step is to evaluate the quantity and quality of wind
energy [68], which is directly related to wind speed. An approach
to evaluate wind speed is to measure the wind with testing
equipments at a certain height (e.g., 15 m), which is commonly
much lower than the actual height of wind turbines. To solve this,
Peterson and Hennessey [69] propose the 1/7 wind profile power
law to assess the wind speed for a turbine at specific height (e.g.,
90 m):


a
v h
¼ ; ð1Þ
vr hr
where v refers to the wind speed at height of h; vr indicates the
already-known wind speed at a reference height of hr; a is the
Fig. 2. The architecture of parallel hybrid EVs [63]. coefficient and its normal value varies from 0.1 to 0.4 [70].
 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661 651

The quantity of usable electricity generated by wind energy is electricity by solid-state semiconductor that converts sunlight into
greatly influenced by meteorological factors such as wind speed electricity directly [75]. In recent years, the development of PV
and air density, as estimated by solar technique has outrun that of CSP. According to the report of
Solar Energy Industries Association, U.S. solar industry has
P ¼ 12 αρAV 3 ; ð2Þ
installed 4751 MW capacity of new PV solar panel in 2013 [76].
where α refers to the Albert Betz constant, ρ indicates the air Solar cell is the basic element in PV solar panels and works
density in kilograms per cubic meters, A represents the swept area efficiently at low ambient temperature. The power efficiency
of wind turbine in square meters, and V is the wind speed in (ηsolar) of a solar cell varies negligibly as the temperature grows.
meters per second [71]. For example, the decreasing rate of ηsolar is 0:2–0:5% for the
Additionally, factors including parameters of wind turbine and temperature increase of 1 1C. In a practical situation, solar cell
season changes can also influence the electricity generation by manufacturers usually adopt heat insulation technique to prevent
wind energy. Due to technical limitations, wind turbines can only large temperature fluctuations. Accordingly, the variance of ηsolar is
convert partial wind energy into electricity. The power output of a relatively small as the ambient temperature varies. This explains
wind turbine varies with different wind speeds and can be the reason why many researchers assume the efficiency of solar
expressed by [72] cell as constant.
8 Apart from ηsolar, the electricity output (psolar(s)) is also deter-
> v r V in
> 0;
>
> mined by the intensity of solar irradiation (s in W/m2) and the area
< gðvÞ; V in o v o V rate
PðvÞ ¼ ð3Þ of aggregated solar cells (a in m2) and can be predicted by
>
> P rate ; v Z V rate
>
> P solar ðsÞ ¼ ηsolar sa: ð9Þ
: 0; v Z V out ;
For PV solar panels, the electricity output can be represented by
where v refers to the wind speed, Vin indicates the cut-in speed,
P PV ðtÞ ¼ AβμðtÞ; ð10Þ
which is the minimum wind speed at which the wind turbine
starts to rotate and generate electricity. Vout represents the cut-out where A refers to the area of a PV solar panel, β indicates the
speed, which is the maximum value that the machine is allowed to efficiency of the PV sloar panel, and μðtÞ is the solar insolation, of
work. Prate is the rated power, Vrate is the rated wind speed at which the value can be chosen according to the statistic of
which the rated power of the turbine can reach. g(v) is the power authorities (e.g., [34] adopt the solar insolation data from Solar
curve between the cut-in speed and the rated speed [73]. Radiation Research Laboratory of National Renewable Energy
Another two methods which can be applied to predict wind Laboratory).
speed and wind power have been proposed in [29] and [44] In order to improve PV technique application, a key issue is the
respectively. Zhao et al. [29] employ Rayleigh distribution to modeling of solar energy. Khatib et al. [77] survey the linear,
model wind speed as follows: nonlinear, artificial intelligence models for solar energy prediction.
The most straightforward models for predicting global and diffuse
V 2
=2σ 2
f v ðV Þ ¼ eV ; ð4Þ solar energy are linear models. A commonly used global solar
σ2
energy linear model is
where V is the wind speed and σ is the scale parameter. The wind
power output can be predicted by ET S
¼ a þb ; ð11Þ
8 Eextra So
< 0;
> V r vin ; V Z vout
where ET refers to the global solar energy, Eextra indicates the
WðVÞ ¼ aV þ b; vin r V r vr ð5Þ
>
: P rate ; extraterrestrial solar energy, S represents the day length, So is the
v oV rv ; r out
number of shining hours, a and b are coefficients of the model. A
commonly used diffused solar energy linear model can be
ωr
a¼ ; ð6Þ expressed by
vr  vin
Ed
vin  ωr ¼ a þ bK T ; ð12Þ
b¼ ; ð7Þ ET
vr  vin
where Ed refers to the diffuse solar energy, ET indicates the global
where vin, vr, vout, and ωr are, respectively, the cut-in speed, the solar energy, KT represents the clearness index, a and b are
rated speed, the cut-out speed, and the power output for the rated coefficients of the model.
speed. Wu et al. [44] provide another model to estimate the wind
power:
8 3. Cost-aware EVs interacting with RESs
> 0; vf o vci ; vf 4 vco
>
>
> v3  v3
<
f ci In this section, research efforts with objectives of developing
P fwind ¼ P r ; vci r f o vr ð8Þ
>
> v3r  v3ci cost-aware interaction between EVs with RESs are investigated in
>
>
:P ; vr r vf ovco ; detail. The cost-aware interaction can be interpreted differently
r
according to different perspectives. For charging service providers,
where vf is the wind speed, vci is the cut-in speed, vco is the cut- it means to reduce operational cost and increase profit. While for
out speed, vr is the nominal wind speed and Pr is the nominal charging service consumers (EVs owners), it indicates the desire to
power of wind power generator. lower battery charging cost. For the grid, it can include electricity
generation cost, system lifecycle cost, and energy
2.2.2. Solar energy transmission cost.
Solar energy can be converted into electricity by concentrating
solar power (CSP) plants or photovoltaic (PV) solar panels. CSP 3.1. Minimizing operational cost
plants utilize mirrors or lenses to focus sunlight, creating suffi-
ciently high temperature to drive traditional steam turbines or The operational cost is the concern of charging providers which
engines to generate electricity [74], while PV solar panel generates can be understood from two perspectives. One situation is that the
652 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661

interaction of EVs with RESs is taken into account in particular shortage, EVs batteries and BSS can discharge as electricity source.
scenario (e.g., the microgrid). The problem of minimizing opera- The basic idea is to schedule the aggregator of EVs, BSS, and loads
tional cost can be considered from the perspective of such system. to minimize the operational cost of microgrid. They first apply a
The other is that the charging providers only consider how to price-intensive model to coordinate the charging of EVs and BSS in
minimize their operating costs. To model the minimizing opera- the grid-connected mode. In this way, the cost of EVs is minimized
tional cost, the general mathematic methodology is to formulate and the profit of BSS is maximized. Compared to [10], Chen and
an optimizing problem, such as nonlinear programming and Duan [11] only consider PV solar power system and aim at
mixed integer programming (MIP). The formulated problems can minimizing the total cost of microgrid. They utilize the radial basis
be solved by evolutionary algorithm (e.g., particle swarm optimi- function network (RBFN) technique to forecast the power output
zation), control strategy (e.g., fuzzy control), etc. The summary of of PV solar system. RBFN is one kind of online 24-h PV solar power
related research efforts is listed in Table 2. output forecasting method and is introduced for more details in
For system operational cost, one typical application scenario is [79]. The performance of RBFN is evaluated by the mean absolute
in the microgrid, as depicted in Fig. 3. A microgrid can leverage percentage error (MAPE), which is given by
information and communication technologies to manage electri-
city generators, energy storage devices, and loads [78]. More 1XN
j P fi  P ai j
MAPE ¼ n100%; ð13Þ
specifically, a microgrid can be divided into several types of Ni¼1 P ai
subsystems, which includes wind power generator, PV solar power
generator, distributed generators, battery energy storage system where N is the total number of forecasting data, Pfi is the forecasted
(BESS), home EV charging appliance and EVs charging stations, value, Pai is the actual value, and i is the index of the dataset. They
while the microgrid controller is responsible for coordinating all formulate the minimal total cost of microgrid by using the model
its subsystems and scheduling power delivery. Besides, a micro- of unit commitment problem. Genetic algorithm is applied to solve
grid can also exchange electricity with the power grid. Several the optimization problem.
research efforts have been conducted to reduce the system Derakhshandeh et al. [17] study how to minimize the system
operational cost of a microgrid with EVs interacting with RESs operational cost of an industrial microgrid (IMG). An IMG is
[10,11,17]. described as using combined heat and power (CHP) system to
The microgrid in [10] is composed of wind power, PV solar generate electricity. Therefore, the cost formulation in [17] is
power, distributed energy generators, and battery swapping sta- different from those in [10,11] by containing costs related to the
tion (BSS). When the electricity generated by RESs is surplus, EVs CHP system (the costs of electricity generation and heat produc-
batteries and BSS will be charged. If the power demand is in tion by using boilers). According to dynamic optimal power flow,

Table 2
Minimizing operational cost.

Research work Objective Formulation Solution Evaluation

Zhang and Chen [10] Minimize operational cost Nonlinear programming Fuzzy control Case study
Chen and Duan [11] Minimize operational cost Nonlinear programming Genetic algorithm Case study
Tan et al. [12] Minimize operational cost Nonlinear programming ADMM Numerical simulations
Aunedi and Strbac [16] Minimize operational cost Mixed-integer linear programming FICO Xpress Trace-based simulations
Derakhshandeh et al. [17] Minimize operational cost Nonlinear programming GAMS Numerical simulations
Saber and Venayagamoorthy [25] Minimize operational cost Nonlinear programming Particle Swarm Optimization Numerical simulations
Fernandes et al. [26] Minimize operational cost Mixed integer programming – Case study
Khodayar et al. [27] Minimize operational cost Mixed integer programming – Case study
Wang et al. [33] Minimize operational cost Nonlinear programming – Case study
Saber and Venayagamoorthy [34] Minimize operational cost Nonlinear programming Particle Swarm Optimization Numerical simulations

Fig. 3. EVs interacting with RESs in microgrid.

 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661 653

the authors formulate the problem as a nonlinear programming namely the base cost and the cost incurred by demand fluctuation.
problem, and solve it by using a general algebraic modeling system These two costs are further represented by base electricity price
(GAMS) software. and fluctuation electricity price. The distributed renewable gen-
As mentioned in [11], the operational cost of power system can eration is estimated by using a short-term prediction, with
be modeled based on the unit commitment (UC) problem. UC Gaussian noise added as the prediction error. The authors assume
problem aims at minimizing the power system operational cost the availability of V2G (EVs can store energy or sell electricity back
under the constraints of power demand and generator. The basic to the utility), and formulate the operational cost as a centralized
definition of UC problem is given by [80] optimization problem. This centralized problem is further decen-
tralized by alternating direction method of multipliers (ADMM),
N X
X T
and converted to a augmented Lagrangian function. The augmen-
OC ¼ ½FC itðP it Þ þ MC it ðP it Þ þ ST it þ SDit ; ð14Þ
i¼1t ¼1
ted Lagrangian function is solved by leveraging Gauss–Seidel
algorithm.
where OC is the system operational cost, FC it ðP it Þ is the fuel cost,
MC it ðP it Þ is the maintenance cost, STit is the startup cost, and SDit is 3.2. Reducing electricity generation cost
the shut-down cost. Besides the application in [11], the system
operational cost formulation based on UC problem has also been The electricity generation cost has direct influences on both
applied in [16,25–27,33,34]. EVs charging providers and consumers. To reduce the electricity
Saber and Venayagamoorthy [34] integrate both wind and solar generation cost, the interaction between EVs and RESs has been
energy into the power supply system. Wind power output is studied intensively by researchers. The researches have employed
calculated by Eq. (2), as given in Section 2.2.1, while PV solar mathematical models including optimal power flow and knapsack
power output is calculated by Eq. (10), as given in Section 2.2.2. problem, as well as the corresponding solutions including dynamic
Meanwhile, EVs can function as loads, energy storages or energy programming and particle swarm optimization.
sources. The authors consider the overall operational cost consist- Li et al. [36] develop a hierarchical control algorithm to reduce
ing of fuel cost, start-up cost and penalty emission cost, and adopt electricity generation cost, via scheduling EVs charging demands
particle swarm optimization (PSO) to solve the problem. Saber and to accommodate fluctuation of electricity generation by wind
Venayagamoorthy further expand their work in [25]. They for- energy. This hierarchical control algorithm is divided into three
mulate the optimization problem based on UC problem, taking levels with different time scales. The top-level controller solves the
into account the existence of uncertainty. Discrete probability scheduling of wind energy and conventional energy utilization
distribution sets are used to express the uncertainties of loads with an hour scale. The middle-level controller schedule of EVs
demands, wind generation, solar generation, and EVs’ availability. charging demands with a sub-hour scale. The bottom-level con-
Based on these uncertainties, the authors propose a set of possible troller utilizes the power grid frequency deviation to conduct real
scenarios, which is further utilized by the objective function. time control of EVs charging. The scheduling of conventional
Similar to [34], PSO is utilized to solve the nonlinear problem. energy and wind energy usage is first modeled as a Knapsack
The mix-integer form has been employed in both [26] and [27] problem. Then, the authors adopt dynamic programming as the
to formulate the optimization problem as mixed-integer problems. solution.
Fernandes et al. [26] propose a mid-term operation model, which Li et al. extend their works [36] through fully exploring the
contains two stages. In the first stage, operational cost is mini- synergy between EVs and wind energy generation [20]. In both
mized with the time scale of a single day of the year. In the second [36] and [20], the frequency dynamics is used as a feedback to
stage, the previously obtained schedule (by solving the optimiza- control EVs charging in real time. To be more specific, when there
tion problem) is revised according to unit outages, which are is a mismatch between generation and load, the frequency of AC
simulated by Monte Carlo method. Similar to the work in [25], power will deviate from the nominal value. However, the fre-
Khodayar et al. [27] consider the existence of uncertainty. A quency dynamics in [20] is approximated by the rational
truncated normal distribution is used to represent EVs energy dynamics. In [20], the authors formulate the generation costs as
consumption and loads forecast. Meanwhile, a Weibull distribu- composed of energy cost and reserve cost. They also provide
tion is utilized to simulate the variation of wind speed. The comparison between the costs with various EVs penetration levels.
effectiveness of the proposed approach is evaluated via two case The comparison result shows that, with a high EVs penetration,
studies of a 6-bus power system and a 118-bus power system. the cost can be further reduced by 20%.
Aunedi and Strbac [16] formulate a day-ahead multi-unit Vayá and Andersson [21] aim at minimizing generation cost
commitment problem by mixed-integer linear programming, from a different perspective. In their work, EVs batteries are
including wind power generators. The wind output fluctuation utilized as distributed storages to accommodate RESs fluctuations.
and demand forecast errors are considered in the formulation The fleet of EVs is modeled as a virtual storage of which the SOC
constrains. The mobility requirements of EVs are based on the depends on individual driving patterns. They formulate the
transport statistics of UK. These requirements are converted into problem as an optimal power flow (OPF) problem, which integrate
electricity demand. Then, EVs are integrated into the formulated a probabilistic model of wind power generation. The fluctuations
problem in two manners, which include uncontrolled charging of wind power generation can be compensated by EVs charging
and controlled charging. Wang et al. [33] also utilize UC model, flexibility. In the simulation setup, the authors utilize the Swiss
considering wind power. They focus on the Illinois power system transmission system. For OPF, they leverage information collected
and assume four kinds of EVs charing scenarios: unconstrained from major power plants, as well as data of wind power genera-
charing, 3-h delayed constrained charging, smart charging, and tion and forecast taken from the National Renewable Energy
smart charging with demand response. Wind power is estimated Laboratory.
by using the historic data from National Renewable Energy Zhao et al. [29] develop a novel economic dispatch (ED) model
Laboratory and four general geographic regions are selected. to minimize the generation cost. The authors assume that some
In [12], the objective is to minimize the operational cost of the EVs will voluntarily register themselves as dispatchable EVs, thus
utility company. In the proposed scenario, EVs and renewable can realize the V2G functions. The ED problem is solved by interior
distributed generators, as well as demand response management point algorithm based on PSO. In the evaluation part, the authors
scheme are considered. The operational cost consists of two parts, simulate the charging and discharging behaviors of EVs via Monte
654 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661

Carlo method. The simulation results show that the aggregated where ωh is the wind energy in hour h, ω^h is the predicted mean
loads of EVs and V2G power output satisfy the normal distribution. value, and errw is the zero-mean random variable. The real-time
The probability distribution of wind power has been introduced in electricity price is forecasted by price forecasting technique.
Section 2.2. They further use an IEEE 188-bus test system and four Linearization method is used to solve the stochastic programming
well-known benchmarks to validate the effectiveness of their in this paper. From the numerical simulations, the effectiveness of
algorithm. the proposed method is evaluated.
Shen and Cui [37] investigate the influence of EVs and RESs on Chen and Tong [31] propose an architecture of energy manage-
reducing electricity cost. Both wind and solar energy are discussed ment system (iEMS) to maximize the operating profit of service
in this article. The output of wind and PV solar power are provider. They use a deterministic model to obtain arbitrary
calculated by using the formulas given in Section 2.2. The cost is arrivals, charging requests, and service deadlines of EVs. An online
implemented by non-linear programming. PSO is leveraged to scheduling algorithm called TAGS is proposed, which is based on
solve the formulated function. They utilize the data collected from threshold admission and greedy scheduling. The dispatcher in
Melbourne in Australia and carry out simulations of two scenarios: iEMS manages a mixture of multiple energy sources, including
only EVs and EVs with RESs together. The conventional power renewable energy, purchased electricity, and locally stored energy.
system is used as baseline. By observing the results, it can be
concluded that the interaction of EVs and RESs on reducing 3.4. Minimizing charging cost
electricity cost is obvious.
For EVs owners, the charging cost is one of their primary
concerns. If the fluctuation of RESs generations and EVs charging
3.3. Increasing profit or benefit demands can be scheduled appropriately, the cost for EVs charging
can be effectively reduced. To formulate the problems of minimiz-
For EVs charging providers, they aim at increasing their profit ing charging cost, researchers have utilized quadratic program-
or benefit for providing charging services. Renewable energy can ming, stochastic programming, etc. In their models, the unce-
be utilized to help realize this objective. It can be considered from rtainty of renewable energy is carefully addressed. Meanwhile,
the perspective of increasing investment benefit or the manage- several methods have been successfully applied to solve the
ment of electricity deliver. The models that are used in this optimization problem, such as Lyapunov optimization and graph
category include linear programming and stochastic program- search algorithm.
ming. To solve the formulations, simplex method and linearization Verzijlbergh et al. [13] and Hu et al. [19] formulate the
method have been applied. optimization problem by utilizing quadratic programming. Both
Zhou et al. [14] consider to increase investment benefits from papers take wind power as the only energy source for EVs
demand side management (DSM) perspective by using EVs as charging. In [13], the problem is viewed from the perspective of
flexible demand. Among various active network managements price-response demand. In the case of having high penetration of
(ANMs), the autonomous regional active network management RESs, the fluctuation of RESs can influence the electricity price in
system is formulated by linear programming problem. In the various ways. The authors investigate several potential approaches
proposed ANM with DSM, renewable energy utilization and in terms of aligning EVs responsive demand, increasing the share
investment benefit are improved. They carry out a 47-node net- of RESs, and considering the capacity of distribution grid. Then
work to analyze the benefit in numerical value. The authors also they focus on the congestion issue in the distributed power system
consider the scenario of wind generation reaching saturation. The to realize the optimal charging strategy. Furthermore, three con-
load profile is predicted after the wind power staying the same gestion mechanisms are proposed, i.e. dynamic grid tariff, advance
quantity. They carry out the increasing investment benefit inves- capacity allocation and distribution grid market.
tigation by minimizing the cost of integrating DSM with ANM. The Hu et al. [19] consider EVs as controllable loads of which the
uncertainty of wholesale electricity is also taken into account in batteries can be charged or discharged, in order to achieve
their method. In the evaluation results, it can be observed that minimum charging cost. The aggregated batteries are modeled
7.9% improvement of utilizing renewable energy can be achieved as a storage system to accommodate the fluctuation of wind power
and the network operational benefit can be increased significantly. generation. The authors adopt the Danish power system in a case
Vasirani et al. [22] propose an effective scheme for wind energy study, which has a 27% penetration of wind power. They utilize the
providers to improve the profits. To achieve this objective, the Danish National Transport Survey Data to estimate the driving
authors study the feasibility of using EVs as a storage medium. behaviors of EVs owners. Through data analysis, they claim that
Meanwhile, they adopt the virtual power plant (VPP) of wind farm, 50% of EVs are always in the idle state while connected to the
which can be regarded as a single entity composed of some wind power grid. The economic benefits are evaluated in both spot
power producers and EVs in the electricity market. The revenue market and regulation market. By observing the simulation results,
earned by a VPP is modeled by linear programming. This problem EVs owners are able to get 30.0 kEUR for providing regulation
is solved by the classical simplex method in their work. In their service in weekdays during winter when the wind speeds are high.
case study, the authors use real wind power generation data and Stochastic programming method is adopted to handle uncer-
electricity prices, as well as collect wind speed data and wind tainty in [18,38]. The uncertainty results from inaccurate predic-
turbines parameters from a wind farm in Spain. tions of both EV charging demands and the electricity generated
Goonewardena and Le [30] formulate the charging of EVs and from RESs. Jin et al. [18] aim at making the best use of RESs for EVs
wind energy by stochastic programming. The key components of charging, and formulate a stochastic optimization problem based
SG are discussed in this article, including bulk and real-time on queuing model. The authors are the first to utilize Lyapunov
purchase of electricity, penalties of selling back electricity to the optimization to solve the cost-aware EV charging problem. Their
grid operator, and the flexible demand of EVs. An aggregator is simulation results show that Lyapunov optimization can bring
utilized to coordinate EVs charging in a controlled manner. The minimum charging cost. Caramanis and Foster [38] assume that
objective is to maximize the profit of the aggregator. The stochas- every EV is equipped with smart device. A hierarchical decision
tic modeling of wind energy is given by making methodology is proposed to control EVs charging. In the
evaluation, the scenario is assumed to take place at Texas where
ωh ¼ ω^h þ err w ; ð15Þ the amount of electricity generated from wind farm is substantial.
 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661 655

The simulation results show that the coordination of EVs charging Bouallaga et al. [24] utilize fuzzy logic supervisor control to
demands and RESs generation can result in reduced cost. In reduce energy transmission cost of the distribution system opera-
addition, their method can also help mitigate distribution network tor. Fuzzy logic supervisor is suitable for those systems which
congestion. cannot perfectly modeled. The objective is achieved by coordinat-
Dallinger and Wietschel [28] aim at investigating the effective- ing EVs charging (avoiding EVs recharge during peak periods) and
ness of utilizing EVs to improve the integration of RESs into the wind power generation. The core of the control method is to shave
grid, while achieving charging cost minimization. In their work, EVs peak power demand. The estimated EV charging demand is
EVs are taken as mobile storages. The mobility is estimated by calculated by the number of EVs, plugged-in time and char-
using a stochastic approach, and represented by three parameters, ging time.
i.e., the probability of traveling on a certain day, the probability of Ghofrani et al. [32] propose an energy management method to
traveling within a specific time slot, and the probability of EVs minimize the cost of conventional generation required to supple-
driving range. The authors evaluate the contribution of EVs on ment the wind power generation. This is achieved by coordinating
balancing RESs (including wind and solar energy) generation EV charging and discharging operations, in order to maximize the
fluctuation in a constructed scenario in Germany. The charing cost utilization of wind power generation. Their method is based on a
is minimized by using a graph search algorithm, which chooses genetic algorithm to find the optimal coordination of EVs charging
the optimal charging and discharging periods when EVs are and discharging periods. The input of the algorithm is based on
plugged into the grid. deterministic models of both the wind power generation and EV
Kristoffersen et al. [35] carry out a case study in Danish power charging loads. Three case studies are carried out to evaluate the
system, which has a high penetration of wind power (approxi- effectiveness of their algorithm, including fixed charging instants,
mately 20% [81]). In this article, an aggregator of EVs is assumed to fixed percentage of EVs, and controllable decision variables.
manage the participation of a fleet of EVs into the electricity
market. The aggregator can trade electricity with the spot market
on the behalf of EVs owners, given that EVs can be leveraged to be 4. Efficiency-aware EVs interacting with RESs
charged or discharged. If the aggregator is a price-taker, the
problem is formulated by linear programming. Otherwise, the The efficient utilization of renewable energy can reduce the
optimization is formulated as a quadratic programming problem. consumption of fossil energy significantly. This can bring many
In their case study, EVs driving patterns are derived from the benefits, e.g., reducing air contamination and increasing cost
clustering of survey data in Western Denmark. The electricity performance for consumers. The interaction of EVs with RESs
market price and its variation are predicted by using his- can achieve this objective at a certain degree. The research efforts
torical data. reviewed in this section are based on three following objectives:
(1) maximizing RESs utilization, (2) optimizing energy dispatch,
and (3) optimizing energy management. The efficiency-aware
3.5. Other cost-aware works objectives can be formulated as optimization problems, such as
linear programming and convex programming. The current solu-
The aforementioned research efforts are categorized into four tions for those formulated problems of this category include
aforementioned classes, while each subsection shares the same optimization tools (e.g., IBM CPLEX) and evolutionary algorithm
cost-aware objective. However, there are also other valuable works (e.g., genetic algorithm). The related formulations and solutions
which have different goals, such as minimizing system cost and are listed in Table 3.
reducing transmission cost. Similarly, these related works are
modeled by mixed-integer nonlinear programming, linear pro- 4.1. Application in microgrid
gramming, etc., and solved by logic control, genetic algorithm, etc.
Shaaban and El-Saadany [15] consider minimizing the system The research efforts of EVs interacting with RESs related to
cost consisting of the upgrades cost and energy losses cost. The efficiency-aware objective in microgrid have been carried out in
methodology is to design an optimal plan for the distribution [39,44,45,48]. Tushar et al. [39] develop an optimal scheduling for
networks to accommodate EVs penetration. The planning problem joint control electricity consumption of appliances and EVs. The
is formulated by mixed-integer nonlinear programming. In their objective is to achieve optimal utilization of electricity generated
model, the planning horizon is composed of 12 months. They by RESs and minimize the amount of electricity from the external
propose an approach based on non-dominated sorting genetic grid. The microgrid considered only includes wind power, PV solar
algorithm (NDSGA) to solve the planning problem. In their work, power, a central controller, home appliances, and EVs. It can be
three kinds of distributed generation (DG) are covered, including operated in either self-sustained mode or a mode interacting with
natural-gas-based DG, wind-based DG, and photo-voltaic-based the macrogrid. The scheduling problem is formulated as a mixed
DG. A 38-bus 12.66-kV system is utilized in a case study. The EV integer linear programming (MILP) problem. During the renewable
charging model is obtained from the Department of Transporta- power generation hours, EVs can be regarded as energy storage;
tion in Great Britain. From their evaluation results, it can be during the high electricity demand period, EVs will provide
concluded that a significant reduction of system costs can be additional electricity. They use a Markov chain state transition
achieved by applying their method. probability to predict the generation of wind power and solar
Gunter et al. [23] aim at minimizing the lifecycle cost when the power in next 24 h. Home appliances are differentiated according
system is integrated with EVs and RESs. The lifecycle cost is to the different requirements, e.g., some require strict scheduling,
composed of initial capital costs and operating cost. The initial while other's scheduling is more flexible. The arrival of EVs in
capital costs can be further divided into renewable distribution microgrid is assumed to follow a Poisson process. The duration of
generation cost and storage unit cost. The authors propose three charging EVs at home can be random, in that EVs with various
models for this optimization problem: (1) linear programming amount of demanded energy are distributed randomly. The
model, (2) search-based model, and (3) MIP model. These models formulated model includes continuous and discrete decision vari-
are all applied to a case study of residential charging. Within ables and is solved by IBM CPLEX MILP solver.
different scenarios, it can be observed that PV solar distributed Different from Tushar et al. [39], Wu et al. [44] consider wind
generator is at least irradiation of 720 W/m. power as the single resource to interact with EVs and aim at
656 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661

Table 3
Efficiency-aware works in EVs interacting with RESs.

Research work Objective Formulation Solution Evaluation

Tushar et al. [39] Minimize imported electricity MILP IBM CPLEX Numerical simulations
Schuller and Hoeffer [40] Maximize utilization Linear programming – Trace-based simulations
Fazelpour et al. [41] Minimize power loss – Genetic algorithms Numerical simulations
Yi et al. [42] Minimize loss energy Bipartite graph Heuristic Trace-based simulations
Gottwalt et al. [43] Maximize utilization Mixed-integer programming IBM CPLEX Trace-based simulations
Wu et al. [44] Optimize dispatch Statistics model Heuristic Numerical simulations
Strnad et al. [45] Maximize utilization – FICO Xpress Numerical simulations
Damiano et al. [46] Enhance exploitation – Heuristic Numerical simulations
Borba et al. [47] Maximize utilization Mixed-integer programming MESSAGE tool Trace-based simulations
Zhu et al. [48] Maximize utilization Semi-Markov decision process Heuristic Numerical simulations
Li et al. [49] Maximize utilization Stochastic programming GAMS Numerical simulations
Battistelli et al. [50] Optimize energy management Linear programming CPLEX Numerical simulations
Mets et al. [51] Maximize utilization Convex optimization Dual decomposition Numerical simulations
Pantoš [52] Maximize utilization Linear programming Heuristic Numerical simulations

optimizing the utilization efficiency of wind power in microgrid. solar panel power, current remaining battery energy, the number
Based on the control flexibility of EVs charging and discharging of EVs charged by PV solar panel and the number of EVs charged
behaviors, they propose three coordinated energy dispatching by BESS. Then, the system dynamics is characterized by the state
methods for EVs interacting with wind power. Their model covers transition probabilities. Linear programming technology is adop-
the factors of wind power generation volatility as well as the ted to solve the SMDP model.
distribution of SOC of individual vehicle. These three approaches
are valley searching dispatching, interruptible dispatching and 4.2. Maximizing RESs utilization
variable-rate dispatching. EVs driving patterns are portrayed by
departure time, arrival time and driving distance using statistic Gottwalt et al. [43] regard EVs charging as a type of flexible
model from family ICE vehicles. Departure times and arrival times load, which can be used to maximize RESs utilization. They use an
are modeled using normal distribution. Driving distance is mod- aggregator to coordinate the EVs charging behavior and satisfy the
eled using logarithmic normal distribution. SOC is used to repre- demand from residential household simultaneously. Only the
sent the residual available energy of EV. The model for wind power behavior of the EVs owned by full-time employees is modeled
generation is depicted in Section 2.2. They utilize Latin Hypercube via using the empirical driving profiles. They obtain the residential
Sampling technique to generate evaluation scenarios. From the household demand from a German dataset. The electricity source
simulation results, it can be seen that the three proposed methods consisted of renewable energy and conventional generator. The
can efficiently improve the matching performance. latter is initiated only when the demanded energy for charging
Strnad et al. [45] integrate thermal units, as well as a PV solar EVs and household exceed the available amount from RESs. Hence,
plant, BESS, a diesel power generator, and EVs charging stations if the demand of EVs can be coordinated to the available genera-
into their considered microgrid model. They aim at maximizing tion from RESs, the conventional generation will be minimized.
the utilization of electricity generated by the PV solar plant and Gottwalt et al. formulate this as a mixed-integer programming
thermal units, which is used for charging EVs or BESS when there problem. They also assume that the knowledge of future RESs, EVs
is energy generated electricity. In this way, the impact on the traveling behavior as well as a one-week household load is
microgrid caused by EVs charging can be alleviated. The genera- available. To demonstrate the benefits of coordinated charging,
tion schedule of the diesel power generator depends on its they first evaluate the uncoordinated charging which is operated
production cost. To be more specific, when microgrid confronts a in a fast manner. Then, the coordinated method is simulated in
power supply shortage, the diesel generator is started only when detail. The duration of evaluation is 52 weeks. The proposed
the generation cost is lower than the cost of purchasing the same mixed-integer programming problem is solved by IBM ILOG
amount of electricity from the grid. The authors utilize the power CPLEX 12.4.
system of the Faculty of Electrical Engineering and Computing in The aforementioned work is one kind of response to residential
Zagreb as a case study. They provide a simulation covering one electricity demand. In contrast, Mets et al. [51] focus on the
winter and one summer workday with 15 min interval. In this demand from charging EVs to maximize the RESs utilization. They
scenario, 8 working hours are enough for EVs charging even using adopt a distributed demand side management (DSM) algorithm to
a slow charging rate, and this amount of electricity can be coordinate EVs charging demand and wind energy supply. In this
sufficiently supplied by RES. way, the renewable energy is utilized to the most extent. Their
Zhu et al. [48] propose charging policies to maximize RESs method is to ensure that energy supply matches energy consump-
utilization in microgrid. They consider the RESs states, battery tion within a certain period. The power system of the objective is
states, and EVs number and formulate the optimization problem composed of EVs owners and a coordinator. In order to maintain
by semi-Markov decision process (SMDP). Although they only the balance, EVs owners can change their demand. If the supply
consider one PV solar panel in their model, they claim that the does not demand, an imbalance cost is inevitable. Thus, the
model can be extended to other sources of renewable energy, such original RESs maximization problem is transformed into imbal-
as wind and hydro energy. As described in Tushar et al. [39], the ance cost minimization problem. The latter is formulated as a
authors utilize continuous Markov chain to model the solar convex optimization problem. To solve this, dual decomposition
radiation and linear method to model the battery states, which method is adopted. Constraints of the objective function are
is used to obtain the remaining energy store in battery. As they moved by applying the method of Lagrange multipliers. To
take EVs arrival and departure rate as Poisson distribution, they evaluate the proposed method, the authors carry out a four-
use the decomposition property of Poisson process to get the week case study(for what). The simulation results show that 73%
transition probabilities. The system state consisted of current PV of energy demand can be satisfied in the most ideal scenario,
 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661 657

which is impractical in real world situation. Meanwhile, through procedure, it is first to assess the stochastic input data, including
using their distributed algorithm, 68% of energy demand can be expected electricity from RESs energy, electricity prices in market
supplied. In the usual business scenario, only 40% of energy and energy requirements of EVs. The electricity from RESs and
demand can be supplied. electricity prices are predicted from the past data. The driving
Different from a one-week period setting in [43], Damiano et al. patterns of existing EVs are obtained from historical transportation
[46] aim at improving RESs utilization within each one-hour dataset. First, for new EVs, the driving patterns are obtained by
period. EVs can comprise an energy storage system. To best utilize using Monte Carlo simulation. EVs with similar driving patterns
such energy storage system, the authors propose a management are clustered in order to decrease the problem complexity. Then
strategy. First, the EVs mobility is modeled to get the number of they model the RESs energy optimization problem as a linear
EVs, which can be used for storing energy. Then the charging and programming problem. The proposed optimization method is
discharging behavior of these EVs can be coordinated within every evaluated in three case studies to demonstrate its effectiveness.
hour. The mobility requirements of EVs are considered in advance. In this paper, the authors only focus on maximizing RESs, despite
According to the classification of EVs status, they propose an that there is a tradeoff between maintaining more RESs and
energy model to predict the interaction capability of EVs fleet with decreasing the charging cost.
the grid for each hour. Three kinds of EVs status are presented: on
the road, parking EVs with and without connection to the grid. The
status of EVs can be transformed between these three statuses 4.3. Other efficiency-aware works
either directly or indirectly. Then, each status is estimated the
average state of charging. Finally, the management strategy of EVs, According to weekly driving profiles of Germany, Schuller and
which function as electric storage, is introduced and is used to Hoeffer [40] propose an optimization approach for individual EV
exploit a given RESs energy system. To evaluate the model, a charging to maximize RESs utilization. The period of EVs charging
simulation is carried out in Sardinia Island electricity power is determined by choosing times with the highest renewable
system. The simulation results validate the effectiveness of their energy generation. In this paper, the authors assume that EVs
management strategy. can be notified if there is available RES. In such a way, EVs owners
Li et al. [49] also study the model of using EVs as energy storage can shift their charging demand to the highest generation of
in SG to interact with RESs. They integrate the uncertainties of renewable energy. They study 6466 empirical individual driving
RESs, power price, and users demand into this work. Moreover, profiles and cluster the profiles into two groups (full-time
thermal power plants, wind power as the single renewable employee and retire). The groups are differentiated by the total
resource, and diesel generators are also considered. On one hand, driving distance of EVs owners. Then, they carry out a benchmark
EVs can act as energy storage when there are surplus in RESs. One simulation which is in an uncoordinated manner to get the base
the other hand, EVs can also be the energy source when there is level of utilized RESs for EVs. The load and RESs generation data of
shortage of electricity supply. The authors utilize stochastic pro- 2011 from the eastern part of Germany are analyzed to determine
gramming method to model this problem in two cases. The first the relative share of RESs. The time interval is 15 min. Based on the
case is to consider the power management without integration of above analyzed results, EVs owners can choose charging times
EVs, while the other has EVs involved. The models of two cases are when the share of RESs generation is highest. A linear program-
implemented using GAMS. With a proper control of charging and ming approach is utilized to model individual EV optimization.
discharging process, the energy cost can be effectively decreased. From the simulation results, it can be seen that the most important
The authors divide the wind power into six levels, which is not factors from driving profiles are charging location and weekly
realistic due to that wind power is continuous. Moreover, they driving distance. The wind share can be increased by at least 80%
only consider wind power as a single renewable resource, while by employees groups. By contrast, the wind share can be increased
PV solar power is also prevalent in other researches. by at least 103%. The results are encouraging for researchers and
Battistelli et al. [50] aim at optimizing energy management in the authors propose their future work that includes the individual
small electricity energy systems (SEES) with V2G technology and economic implication.
RESs utilization. The SEES is in realistic conditions. They propose a Fazelpour et al. [41] solve several optimization problems about
linear model to assess the contribution of V2G technology from EVs interacting with RESs and the outcomes can be used to
EVs in supporting energy management. Since a single EV can consider as the principles for EVs parking lot. Compared to
impact the grid negligibly, they instead using an aggregated model previous works in [82,83], minimization of the power loss,
consisted of many EVs. The uncertainties of EVs are modeled by a optimizing RESs, the satisfaction of users and the energy con-
deterministic program. The probability distributions of uncertain- straints of the power grid are considered simultaneously. These
ties are not required. The model of aggregated EVs is built upon factors are considered as the fundamental conditions for con-
several assumptions, including homogeneous batteries with same structing EVs parking lots. First, they optimize the size and
efficiency, negligible batteries loss, and unchangeable capacity. location of distributed generators subjecting to power loss mini-
They assume homogeneous batteries due to that different types of mization. Adding a new distributed electricity system to current
batteries can impact the capacity of aggregated EVs directly. The power grid may cause voltage disturbances and fluctuations. This
energy management model is formulated as a linear program. The problem is formulated with the objective of minimizing power
problems are solved by using CPLEX. In the case study, the loss. The authors utilize genetic algorithm to solve this problem. In
numerical simulation results show the effectiveness of V2G on order to satisfy the size of distributed generation, they choose a
management of energy resources in a SEES. However, if different hybrid renewable energy system which includes PV solar power,
types of EVs batteries are considered in the assumptions, it will be wind power and diesel generation at the optimum sites. The
more practical in a realistic scenario. objective of considering hybrid renewable energy is to minimize
Pantoš [52] presents a stochastic optimization algorithm to the energy cost with respect to the variation of renewable energy.
maximize RESs utilization. In order to capture the stochastic Finally, the charging rate is tuned to provide the maximal fair
nature of vehicles transportation, they use Monte Carlo simulation energy delivery. EVs and infrastructure are assumed to be capable
to model the uncertainty in many scenarios. Uncertainties of RESs of establishing a two-way communication, in order to identify SOC
are solved by statistical methods. They assume that EVs charging of arriving EVs, the amount of EVs being charged, starting and
can be coordinated. In the initial process of optimization ending time of charging, etc.
658 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661

Yi et al. [42] study the problem how to route RESs to energy the carbon footprint derived from purchasing electricity from the
station when some routing paths are clogged. They propose a new grid, which produce the carbon footprint as a by-product, the
infrastructure which is constituted of EVs and EVs charging emissions from the distributed generation units, and the emissions
stations and name it as mobile electrical grid (MEG). It is reduced by replacing conventional vehicles with EVs. Thus, the
composed of three parts: energy generation, energy transportation emission formulation is as follows:
and energy consumption. EVs are charged by RESs plants and
f ðΩÞ ¼ Egrid þ EDG EEVs ; ð16Þ
moved to charging stations. Then, EVs discharge electricity to
charging stations. They present an electric bus system as an where Ω denotes the vector of decision variables. Egrid, EDG, and
application. The optimization objective in this application is to EEVs indicate the above three categories of emissions. Wind and PV
minimize the hop number of energy routing when some paths are solar distributed generations are taken into account. The final
clogged by traffic jam. They use bipartite graph to formulate this objective is to accommodate EVs in conjunction with RESs in the
problem in a typical Manhattan district. The application scenario is period of multi-year. By observing the simulation results, the
suitable for an electric bus company to transfer renewable energy effectiveness of the proposed algorithm is validated.
to charging stations in downtown area. Then, the energy can be Compared to emissions-aware work in [15], Aunedi and Strbac
forwarded to another charging station by another bus. Given that [16] evaluate the impact of EVs and RESs interacting on reducing
every bus station can also act as a charging station. The authors emissions from the perspective of transportation sector instead of
utilize two algorithms to compute minimal energy metric routing. the distribution system. A nationally representative fleet of EVs is
These two algorithms are evaluated by real transporting data sets taken into consideration. The penetration level of EVs is assumed
from Manhattan bus lines in New York and the bus map of the to increase from 0% to 100% with 10% increment. As summarized
Pioneer Valley Transit Authority. By observing the simulation in Section 3.1, the primary objective is to reduce the system
results, it can be concluded that their method is effective to help operational cost. The reduction of emissions is measured by the
route renewable energy. ability of integrating renewable energy into the power grid.
Compared to Schuller and Hoeffer [40], Borba et al. [47] only According to the evaluation results, the conclusion is that there
consider the wind power in northern Brazil. They aim at evaluat- are obvious reductions of emissions when EVs penetration level is
ing the possibility of regularizing the wind energy imbalance by less than 40%. As the level of EVs penetration increases, the
the expanding EVs. In this paper, the authors point out that the emission reduction benefits of their method have the tendency
total installed wind power capacity in Brazil is expected to be of converging.
4.0 GW by 2013, mainly in the northern area. If wind turbines are Patil et al. [53] develop a dynamic programming-based frame-
installed with inflexible power plants, e.g., nuclear station hydro- work for power management of PHEVs and charging control.
electric power plants, they will have impacts between power These two problems are related to PHEVs driving on the road
generation supply and users demand. They also provide a case and charging from the power grid. The framework is used to
study in the exactly same situation. The authors argue that minimize CO2 emissions. They present the quantification of total
integrating wind power into power systems will bring many CO2 reduction by charging PHEVs with wind energy. Three 24-h
impacts. The impacts are presented from local and system terms. CO2 traces of per unit energy are chosen from a one-year data set.
They also introduce the impacts of EVs on power system. This In the model, the demand from grid is divided with the scale of
work provides a typical example for EVs to assist integrating wind one-hour. The objective is to optimize PHEV charging and power
power into the power grid. The power sector and wind energy in management for one day. Bellman equation which is the dynamic
Brazil are described in detail. Then, the high penetration of wind programming is used to optimize the two control problems. The
power is simulated by MESSAGE which is a MIP model. It can be amount of CO2 produced by fuel usage and grid charging are both
used to formulate and evaluate alternative strategies for supplying considered, as well as the impact of different SOC conditions. From
energy [84]. The mathematical principle in MESSAGE is to achieve the evaluation results, it can be seen that their method can result
the optimal solution with some constrains which restrict all in at least 2% CO2 reduction when the wind speed is low.
possible solutions in a defined feasible region. From the simulation Saber and Venayagamoorthy publish three emissions reduction
results, it can be observed that 1.5 million EVs in 2030 recharged related works from 2010 to 2012 [25,34,55]. These works have the
overnight for half the year. In this paper, the authors do not same formulation for emissions, given by
consider storage system and smart technologies which can be
εci ðP i ðtÞÞ ¼ αi þ βi P i ðtÞ þ γ i P 2i ðtÞ; ð17Þ
used to reduce cost.
where Pi(t) is the output power of ith unit at time t, αi, βi, and γi are
emissions coefficients of unit i. Then both emissions optimization
5. Emissions-aware EVs interacting with RESs and cost optimization are taken into consideration. PSO is
employed to solve this optimization problem. The authors empha-
One of the most appreciated features of EVs is their low size three different aspects in each of these works. In [25], the
emission. In this section, research efforts aiming at reducing emphasis is how to deal with uncertainty. Discrete probability
emissions through utilizing the interaction between EVs and RESs distribution is used for modeling load demand, wind resources,
are investigated. Although there are some overlaps with previous solar resources, and vehicles distribution. The emphasis in [34] is
sections, this section needs to be separated since the research how to handle a large number of EVs to reduce cost and emissions.
works with emission-aware objective are fundamentally different Refs. [34] and [55] do not share the same perspective. The
from research works with cost-aware or efficiency-aware objec- emissions of power systems are formulated by the same function.
tive. The mathematical models employed in these works are In [34], they formulate the problem from a system perspective.
similar to those adopted in Sections 3 and 4, which is to formulate The transport industry is considered as one source of emissions.
an optimization problem. Accordingly, the solutions include evo- However, in [55], the problem is formulated from the perspective
lutionary algorithm (e.g., PSO) and optimization tool (e.g., FICO of unit commitment and does not include the emissions from
Xpress). transport sector.
One sub-objective in the work of Shaaban and EI-Saadany [15] Shen and Cui [54] study the impact of EVs and RESs on
is to minimize emissions by interacting EVs and RESs in the reducing emissions of electricity, through manipulating the char-
distribution system. The emissions are composed of three parts: ging and discharging schedules of EVs. Their output forms of PV
 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661 659

solar system and wind turbine are the same as introduced in lithium. So, the useful life span of battery will be decreased. EVs
Section 2.2, while the formulation of emissions is the same as the owners will be careful to consider whether to take part in
work in [34]. The difference is that the average travel distance in integrating RESs. This problem may be solved by two possible
this paper is obtained by a log-normal distribution. The authors methods. One is to tune the tradeoff between battery useful life
evaluate their method in the scenario of two systems, i.e. one and integrating with RESs. Researchers can provide strategy to
includes EVs only and the other has both EVs and RESs. Their play the game to bring benefit for EVs owners. The other method
simulation results demonstrate the effectiveness of this method. is to propose new battery technique to prolong useful life.
Lund and Kempton [56] adopt EnergyPLAN model to evaluate
the impact of EVs with V2G on CO2 emission reduction. Energy- 6.3. Feasibility of V2G supporting RESs interaction
PLAN is an analysis tool which can be used to analyze regional or
national system. One important input of EnergyPLAN is the In Sections 3–5 many research works are based on V2G
distribution of transportation demand. The other is the amount technology, such as the works in [16,19,21,22,25–27,32,34,
of traveling V2G cars. Two national energy systems are modeled 35,37,46,50,55]. However, most of the EVs produced do not
using this tool. One considers combined heat and power, while the support V2G technology in the current market. The primary goal
other is similarly sized country without combined heat and power. of introducing EVs into transportation sector is to replace the ICE
They take the wind penetration within the range from 0% to 100%, vehicles. To make V2G feasible, there are some problems need to
and use Danish statistics to model transportation demands. The be considered in detail. Firstly, in order to utilize V2G function, EVs
evaluation results show that EVs interacting with RESs have need to be at least equipped with bidirectional power converters
significantly effect on reducing CO2 reduction. and advanced communications device. Secondly, charging utility
also needs to be equipped with smart charging point that can
communicate with the control unit of EVs. Thirdly, as the number
6. Open issues
of EVs increases, the control strategy of grid operator will become
more and more complicate. Some EVs owners may not choose to
In this section, three essential open issues shared by different
take part in V2G application though their EVs can support V2G.
objectives of EVs interacting with RESs are provided.
The uncertainty of usable EVs may not be estimated accurately.
Fourthly, EVs are mostly used as energy storage or energy source
6.1. Large scale RESs integration into the grid for V2G. Because of the mobility of EVs, the distribution cannot be
predicted accurately in advance. This may bring congestion pro-
The difference between RESs generation facilities and conven- blem to schedule energy among EVs, RESs, and charging station by
tional power plants is significant. Take wind turbine as example, power distribution network, as the problems presented in [13,38].
its generation system is based on power electronics. For conven- Finally, it needs to charge and discharge EVs battery frequently.
tional power plants, synchronous generators are more usually This will reduce the useful life cycle of battery, as presented in
employed. This difference between RESs and conventional gen- Section 6.2. All of these problems need to be solved to make V2G
erator gives rise to new problems when integrating RESs into feasible.
power grid. These problems can be viewed in two angles. From the
perspective of local power system, the problems include voltage
control, fault current, harmonic distortion and flicker [85]. As for 7. Conclusion
macrogrid system, the impacts include imbalance between load
and generation, reactive power generation, and the reduced The penetration of EVs is reshaping the transportation system.
frequency control [86]. There are same issues for PV solar power Many positive impacts can be expected from the growing number
system interacting with macrogrid system. However, these ques- of EVs, including lower fossil oil dependency and fewer carbon
tions will not bring challenges to power systems if the participa- emissions. At meantime, charging a large scale of EVs causes
tion of RESs is in the small scale. As RESs are more utilized, these significant impact to the power grid. A promising way to mitigate
problems will become severer. Söder and Ackermann [87] present this impact is to take advantage of the smart grid technologies and
that high level of wind power penetration into the existing power exploit local power generations such as renewable energy. This
systems may require step-by-step redesign of the power system review right investigates the existing works of this thread.
and operation approaches. In the above reviewed articles, Chen Specially, the related works are divided into three categories,
and Tong [31], Ghofrani et al. [32], and Borba et al. [47] have been including cost-aware charging, the efficiency-aware charging,
carried out to investigate the large scale RESs penetration inter- and the emissions-aware charging. For each category of works,
acting with EVs. One promising direction is to propose more the summarization of the focused problems and adopted solutions
practical methods to realize large scale RESs integration into to is provided. The assessments and comparisons among them are
the grid. also illustrated. Finally, several key open issues and possible
solutions are also discussed. It is expected that this review can
6.2. EVs battery utilization for RESs interaction help researchers and engineers better understand the state-of-
the-art and contribute more to this field.
The research works [16,19,21,25,35,37,46,50] investigated in
this paper leverage EVs battery pack as energy storage. If the
electricity generated by RESs is surplus, it can be buffered in EVs Acknowledgments
battery. When the power supply is in shortage, EVs battery can
discharge to supply electricity. However, this will increase the This work was supported by the China Scholarship Council (No.
charging-discharging cycle of battery. Nearly all EVs in the present 201206120149).
market utilize Li-ion battery. As the number of charging-
discharging cycle increases, the power and capacity of Li-ion References
battery will fade vastly. The factors that influence power include
loss of lithium and impendence rise. The capacity fading results [1] Tanaka N. Technology roadmap: electric and plug-in hybrid electric vehicles.
from loss of active material (graphite exfoliation) and loss of International energy agency. Technical report; 2011.
660 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661

[2] Block D, Harrison J, Brooker P. Electric vehicle sales for 2014 and future [33] Wang J, Liu C, Ton D, Zhou Y, Kim J, Vyas A. Impact of plug-in hybrid electric
projections. Florida Solar Energy Center; 2015. [Online] Available: 〈http://fsec. vehicles on power systems with demand response and wind power. Energy
ucf.edu/en/publications/pdf/FSEC-CR-1998-15.pdf〉. Policy 2011;39:4016–21.
[3] Lopes JAP, Soares FJ, Almeida PMR. Integration of electric vehicles in the [34] Saber AY, Venayagamoorthy GK. Plug-in vehicles and renewable energy
electric power system. Proc IEEE 2011;99:168–83. sources for cost and emission reductions. IEEE Trans Ind Electron
[4] Power your Tesla with solar fuel; 2014. [Online] Available: 〈http://solar. 2011;58:1229–38.
solarcity.com/promotions/tesla/〉. [35] Kristoffersen TK, Capion K, Meibom P. Optimal charging of electric drive
[5] Nissan LEAF charges up with SunPower; 2014. [Online] Available: 〈http://us. vehicles in a market environment. Appl Energy 2011;88:1940–8.
sunpower.com/partnerships/nissan-leaf-electric-vehicle/. [36] Li C-T, Ahn C, Peng H, Sun J. Integration of plug-in electric vehicle charging and
[6] Resch G, Held A, Faber T, Panzer C, Toro F, Haas R. Potentials and prospects for wind energy scheduling on electricity grid. In: IEEE innovative smart grid
renewable energies at global scale. Energy Policy 2008;36:4048–56. technologies (ISGT); 2011. p. 1–7.
[7] Mwasilu F, Justo JJ, Kim E-K, Do TD, Jung J-W. Electric vehicles and smart grid [37] Shen W, Cui X. Impact of electric vehicles and renewable energy systems on
interaction: a review on vehicle to grid and renewable energy sources cost and emission of electricity. In: Proceedings of IEEE 7th conference on
integration. Renew Sustain Energy Rev 2014;34:501–16. industrial electronics and applications; 2011. p. 120–5.
[8] Richardson DB. Electric vehicles and the electric grid: a review of modeling [38] Caramanis M, Foster JM. Management of electric vehicle charging to mitigate
approaches, Impacts, and renewable energy integration. Renew Sustain renewable generation intermittency and distribution network congestion. In:
Energy Rev 2013;19:247–54.
Proceedings of the 48th IEEE conference on decision and control (CDC); 2009.
[9] Su W, Eichi H, Zeng W, Chow M-Y. A survey on the electrification of
p. 4717–22.
transportation in a smart grid environment. IEEE Trans Ind Inf 2012;8:1–10.
[39] Tushar MHK, Assi C, Maier M, Uddin MF. Smart microgrids: optimal joint
[10] Zhang M, Chen J. The energy management and optimized operation of electric
scheduling for electric vehicles and home appliances. IEEE Trans Smart Grid
vehicles based on microgrid. IEEE Trans Power Deliv 2014;29:1427–35.
2014;5:239–50.
[11] Chen C, Duan S. Optimal integration of plug-in hybrid electric vehicles in
[40] Schuller A, Hoeffer J. Assessing the impact of EV mobility patterns on
microgrids. IEEE Trans Ind Inf 2014;10:1917–26.
renewable energy oriented charging strategies. Energy Proc 2014;46:32–9.
[12] Tan Z, Yang P, Nehorai A. An optimal and distributed demand response
[41] Fazelpour F, Vafaeipour M, Rahbari O, Rosen MA. Intelligent optimization to
strategy with electric vehicles in the smart grid. IEEE Trans Smart Grid
integrate a plug-in hybrid electric vehicle smart parking lot with renewable
2014;5:861–9.
[13] Verzijlbergh RA, de Vries LJ, Lukszo Z. Renewable energy sources and energy resources and enhance grid characteristics. Energy Convers Manag
responsive demand. Do we need congestion management in the distribution 2014;77:250–61.
grid? IEEE Trans Power Syst 2014:1–10. [42] Yi P, Zhu T, Lin G, Zhang Q. Routing renewable energy using electric vehicles in
[14] Zhou L, Li F, Gu C, Hu Z, le Blond S. Cost/benefit assessment of a smart mobile electrical grid. In: Proceedings of mobile ad-hoc and sensor systems
distribution system with intelligent electric vehicle charging. IEEE Trans Smart (MASS); 2013. p. 19–27.
Grid 2014;5:839–47. [43] Gottwalt S, Schuller A, Flath C, Schmeck H, Weinhardt C. Assessing load
[15] Shaaban MF, El-Saadany E. Accommodating high penetrations of PEVs and flexibility in smart grids: Electric vehicles for renewable energy integration.
renewable DG considering uncertainties in distribution systems. IEEE Trans In: IEEE power and energy society general meeting (PES); 2013. p. 1–5.
Power Syst 2014;29:259–70. [44] Wu T, Yang Q, Bao Z, Yan W. Coordinated energy dispatching in microgrid
[16] Aunedi M, Strbac G. Efficient system integration of wind generation through with wind power generation and plug-in electric vehicles. IEEE Trans Smart
smart charging of electric vehicles. In: Proceedings of 8th international Grid 2013;4:1453–63.
conference and exhibition on ecological vehicles and renewable energies [45] Strnad I, Skrlec D, Tomisa T. A model for the efficient use of electricity
(EVER); 2013. p. 1–12. produced from renewable energy sources for electric vehicle charging. In:
[17] Derakhshandeh S, Masoum AS, Deilami S, Masoum MA, Hamedani Golshan M. Proceedings of 2013 4th international youth conference on energy (IYCE);
Coordination of generation scheduling with PEVs charging in industrial 2013. p. 1–8.
microgrids. IEEE Trans Power Syst 2013;28:3451–61. [46] Damiano A, Marongiu I, Porru M, Serpi A. Electric vehicle energy storage
[18] Jin C, Sheng X, Ghosh P. Energy efficient algorithms for electric vehicle management for renewable energy sources exploitation. In Proceedings of
charging with intermittent renewable energy sources. In: IEEE power and IEEE electric vehicle conference (IEVC); 2012. p. 1–8.
energy society general meeting (PES); 2013. p. 1–5. [47] Borba SM, Szklo A, Schaeffer R. Plug-in hybrid electric vehicles as a way to
[19] Hu W, Su C, Chen Z, Bak-Jensen B. Optimal operation of plug-in electric maximize the integration of variable renewable energy in power systems: the
vehicles in power systems with high wind power penetrations. IEEE Trans case of wind generation in northeastern Brazil. Energy 2012;37:469–81.
Sustain Energy 2013;4:577–85. [48] Zhu L, Yu FR, Ning B, Tang T. Optimal charging control for electric vehicles in
[20] Li C-T, Ahn C, Peng H, Sun J. Synergistic control of plug-in vehicle charging and smart microgrids with renewable energy sources. In: Proceedings of IEEE 75th
wind power scheduling. IEEE Trans Power Syst 2013;28:1113–21. conference on vehicular technology conference (VTC Spring); 2012. p. 1–5.
[21] Vayá MG, Andersson G. Integrating renewable energy forecast uncertainty in [49] Li Y, Kaewpuang R, Wang P, Niyato D, Han Z. An energy efficient solution:
smart-charging approaches for plug-in electric vehicles. In: IEEE Grenoble integrating plug-in hybrid electric vehicle in smart grid with renewable
PowerTech (POWERTECH); 2013. p. 1–6. energy. In: Proceedings of IEEE workshop on computer communications;
[22] Vasirani M, Kota R, Cavalcante RL, Ossowski S, Jennings NR. An agent-based 2012. p. 73–8.
approach to virtual power plants of wind power generators and electric [50] Battistelli C, Baringo L, Conejo A. Optimal energy management of small electric
vehicles. IEEE Trans Smart Grid 2013;4:1314–22. energy systems including V2G facilities and renewable energy sources. Electr
[23] Gunter SJ, Afridi KK, Perreault DJ. Optimal design of grid-connected PEV Power Syst Res 2012;92:50–9.
charging systems with integrated distributed resources. IEEE Trans Smart Grid [51] Mets K, De Turck F, Develder C. Distributed smart charging of electric vehicles
2013;4:956–67.
for balancing wind energy. In: Proceedings of IEEE 3rd international con-
[24] Bouallaga A, Merdassi A, Davigny A, Robyns B, Courtecuisse V. Minimization of
ference on smart grid communications (SmartGridComm); 2012. p. 133–8.
energy transmission cost and CO2 emissions using coordination of electric
[52] Pantoš M. Stochastic optimal charging of electric-drive vehicles with renew-
vehicle and wind power (W2V). In: IEEE PowerTech (POWERTECH); 2013. p.
able energy. Energy 2011;36:6567–76.
1–6.
[53] Patil R, Kelly JC, Filipi Z, Fathy H. A framework for the integrated optimization
[25] Saber AY, Venayagamoorthy GK. Resource scheduling under uncertainty in a
of charging and power management in plug-in hybrid electric vehicles. In:
smart grid with renewables and plug-in vehicles. IEEE Syst J 2012;6:103–9.
Proceedings of 2012 American control conference (ACC); 2012. p. 1327–34.
[26] Fernandes C, Frías P, Latorre JM. Impact of vehicle-to-grid on power system
[54] Shen W, Cui X. Impact of electric vehicles and renewable energy systems on
operation costs: the Spanish case study. Appl Energy 2012;96:194–202.
[27] Khodayar ME, Wu L, Shahidehpour M. Hourly coordination of electric vehicle cost and emission of electricity. In: Proceedings of the 7th IEEE conference on
operation and volatile wind power generation in SCUC. IEEE Trans Smart Grid industrial electronics and applications (ICIEA); 2012. p. 120–5.
2012;3:1271–9. [55] Saber AY, Venayagamoorthy GK. Intelligent unit commitment with vehicle-to-
[28] Dallinger D, Wietschel M. Grid integration of intermittent renewable energy grid: a cost-emission optimization. J Power Sour 2010;195:898–911.
sources using price-responsive plug-in electric vehicles. Renew Sustain [56] Lund H, Kempton W. Integration of renewable energy into the transport and
Energy Rev 2012;16:3370–82. electricity sectors through V2G. Energy Policy 2008;36:3578–87.
[29] Zhao J, Wen F, Dong ZY, Xue Y, Wong KP. Optimal dispatch of electric vehicles [57] Battery electric vehicle; 2014. [Online] Available: 〈http://en.wikipedia.org/
and wind power using enhanced particle swarm optimization. IEEE Trans Ind wiki/Battery_electric_vehicle〉.
Inf 2012;8:889–99. [58] James L, John L. Electric vehicle technology explained. Chichester, United
[30] Goonewardena M, Le LB. Charging of electric vehicles utilizing random wind: Kingdom: John Wiley & Sons; 2012.
A stochastic optimization approach. In: IEEE globecom workshops (GC [59] Curtin R, Shrago Y, Mikkelsen J. Plug-in hybrid electric vehicles. Reuters/
Wkshps); 2012. p. 1520–5. Surveys of Consumers, Michigan: University of Michigan; 2009.
[31] Chen S, Tong L. IEMS for large scale charging of electric vehicles: architecture [60] Haghbin S, Khan K, Lundmark S, Alakula M, Carlson O, Leksell M, et al.
and optimal online scheduling. In: Proceedings of IEEE 3rd international Integrated chargers for EV's and PHEV's: examples and new solutions. In:
conference on smart grid communications (SmartGridComm); 2012. p. 629– Proceedings of XIX international conference on electrical machines (ICEM);
34. 2010. p. 1–6.
[32] Ghofrani M, Arabali A, Etezadi-Amoli M. Electric drive vehicle to grid [61] Yilmaz M, Krein PT. Review of battery charger topologies, charging power
synergies with large scale wind resources. In: IEEE power and energy society levels, and infrastructure for plug-in electric and hybrid vehicles. IEEE Trans
general meeting; 2012. p. 1–6. Power Electron 2013;28:2151–69.
 L. Liu et al. / Renewable and Sustainable Energy Reviews 51 (2015) 648–661 661

[62] Lee Y-J, Khaligh A, Emadi A. Advanced integrated bidirectional AC/DC and DC/ [76] Solar energy facts: 2013 year in review; 2014. [Online] Available: 〈http://www.
DC converter for plug-in hybrid electric vehicles. IEEE Trans Veh Technol seia.org/sites/default/files/YIR%202013%20SMI%20Fact%20Sheet.pdf〉.
2009;58:3970–80. [77] Khatib T, Mohamed A, Sopian K. A review of solar energy modeling
[63] Fasugba MA, Krein PT. Gaining vehicle-to-grid benefits with unidirectional techniques. Renew Sustain Energy Rev 2012;16:2864–9.
electric and plug-in hybrid vehicle chargers. In: Proceedings of IEEE con- [78] Bertocco M, Giorgi G, Narduzzi C, Tramarin F. A case for IEEE Std. 1451 in smart
ference on vehicle power and propulsion (VPPC); 2011. p. 1–6. microgrid environments. In: Proceedings of IEEE international conference on
[64] Du Y, Lukic S, Jacobson B, Huang A. Review of high power isolated bi- smart measurements for future grids (SMFG); 2011. p. 124–9.
directional DC-DC converters for PHEV/EV DC charging infrastructure. In: [79] Chen C, Duan S, Cai T, Liu B. Online 24-h solar power forecasting based on
Energy conversion congress and exposition (ECCE); 2011. p. 553–60. weather type classification using artificial neural network. Solar Energy
[65] Singh B, Singh BN, Chandra A, Al-Haddad K, Pandey A, Kothari DP. A review of 2011;85:2856–70.
single-phase improved power quality AC–DC converters. IEEE Trans Ind [80] Padhy NP. Unit commitment-a bibliographical survey. IEEE Trans Power Syst
Electron 2003;50:962–81.
2004;19:1196–205.
[66] Global wind energy council; 2014. [Online] Available: 〈http://www.gwec.net/
[81] O'Connell N, Wu Q, Østergaard J, Nielsen AH, Cha ST, Ding Y. Day-ahead tariffs
global-figures/wind-energy-global-status〉.
for the alleviation of distribution grid congestion from electric vehicles. Electr
[67] Chen Z, Blaabjerg F. Wind farm—a power source in future power systems.
Power Syst Res 2012;92:106–14.
Renew Sustain Energy Rev 2009;13:1288–300.
[82] Vasirani M, Ossowski S. A proportional share allocation mechanism for
[68] Kong F, Dong C, Liu X, Zeng H. Blowing hard is not all we want: quantity vs
quality of wind power in the smart grid. In: Proceedings of the 33rd IEEE coordination of plug-in electric vehicle charging. Eng Appl Artif Intell
conference on computer communications; 2014, p. 2813–21. 2013;26:1185–97.
[69] Peterson EW, Hennessey Jr JP. On the use of power laws for estimates of wind [83] Su W, Chow M-Y. Performance evaluation of a PHEV parking station using
power potential. J Appl Meteorol 1978;17:390–4. particle swarm optimization. In: IEEE power and energy society general
[70] Masters GM. Renewable and efficient electric power systems. Hoboken, New meeting; 2011. p. 1–6.
Jersey: John Wiley & Sons; 2013. [84] Keppo I, Strubegger M. Short term decisions for long term problems—the
[71] Mabel MC, Fernandez E. Estimation of energy yield from wind farms using effect of foresight on model based energy systems analysis. Energy
artificial neural networks. IEEE Trans Energy Convers 2009;24:459–64. 2010;35:2033–42.
[72] Kong F, Dong C, Liu X, Zeng H. Quantity vs quality: optimal harvesting wind [85] Anaya-Lara O, Jenkins N, Ekanayake J, Cartwright P, Hughes M. Wind energy
power for the smart grid. Proc IEEE; 2014;102:1762–76. generation: modelling and control. Chichester, United Kingdom: John Wiley &
[73] Kong F, Liu XA. Survey on green-energy-aware power management for Sons; 2011.
datacenters. ACM Comput Surv 2015;47:30. [86] Fernandez R, Mantz R, Battaiotto P. Impact of wind farms on a power system.
[74] Concentrating solar power; 2014. [Online] Available: 〈http://www.seia.org/ An eigenvalue analysis approach. Renew Energy 2007;32:1676–88.
sites/default/files/cspfactsheet-120223144940-phpapp01.pdf〉. [87] Söder L, Ackermann T. Wind power in power systems: an introduction. Wind
[75] Photovoltaic solar technology; 2014. [Online] Available: 〈http://www.seia.org/ Power Power Syst 2005:25–47.
sites/default/files/resources/photovoltaic-solar-technology_0.pdf〉.