Energy 140 (2017) 1182e1197

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

The feasibility of solar parking lots for electric vehicles

Raquel Figueiredo*, Pedro Nunes, Miguel C. Brito
Instituto Dom Luiz, Faculdade de Ci^
encias, Universidade de Lisboa, 1749-016 Lisboa, Portugal

a r t i c l e i n f o a b s t r a c t

Article history: The direct coupling of solar electricity with electric vehicles is very topical due to their many co-benefits,
Received 27 April 2017 such as reducing the electricity flow into and from the grid. This work illustrates this solution using a
Received in revised form park-and-ride lot outside Lisbon to study its solar energy potential to charge electric vehicles. The solar
28 August 2017
resource, its exploitation and coordination with the vehicles under different charging approaches are
Accepted 6 September 2017
discussed in detail, using methods extendible to other cases. A genetic-algorithm optimisation was
Available online 7 September 2017
performed to maximise the solar energy generation. To ascertain the techno-economic feasibility of the
concept, several sets of assumptions are analysed, including the use of energy storage and smart
Keywords:
Solar parking lots
charging. For current market conditions, the payback time is found to be 14 years; a modest financial
Electric vehicles public incentive would significantly improve the project economics e making the payback drop to 7
EV charging years e enabling conditions to put this inescapable and green approach at the forefront.
Photovoltaic energy © 2017 Elsevier Ltd. All rights reserved.
Sustainable energy systems
Feasibility study

1. Introduction
The growing global electricity consumption e the Interna-
tional Energy Agency estimates a 30% increase by 2040 [1] e
creates concern on the sources of electricity production for the
future. In order to not compromise the objectives of the COP21
agreement, renewable energy is seen as the main solution. The
large amount of solar energy reaching the earth e 101 PWh per
hour [2], which is about the same as the world annual final en-
ergy demand [3] e combined with the price competitiveness of
photovoltaics (PV) is enhancing the role of this technology as part
of that solution.
Alongside, the transport sector is also of concern, since it con-
tributes significantly to air pollution and greenhouse gases (GHG)
emissions. The replacement of internal combustion engine (ICE)
vehicles for the much more efficient battery electric vehicles (EVs)
is one of the possible measures to overcome this. However, EV
charging must be based on a clean, fossil-free electricity [4].
In both PV and EV cases, the grid integration is an issue: their
emerging penetration may jeopardise the grid reliability, causing
stabilisation and flexibility problems. The main disadvantage of
* Corresponding author. FCUL, DEGGE, Ed. C8, 1749-016 Campo Grande, Portugal.
E-mail address: rvfigueiredo@fc.ul.pt (R. Figueiredo).
PV generation is its variability and uncertainty [5]. As for the EVs,
they could cause a boost in electricity demand and overload the
grid, degrade power quality and destabilise the electric system
[6]. One way to avoid these issues is by charging EVs directly from
PV. This integration should be carefully planned, using, for
example, scheduled charging plans to coordinate both technolo-
gies [7].
By doing this, on the one hand EVs could be used as storage
devices and dispatchable loads, helping the grid maintaining the
supply-demand balance while allowing a larger renewable
penetration [8]. On the other hand, PV production could also
enable a larger penetration of EVs when they charge from it. This
way there is no significant net-load increase in demand. There-
fore, the appropriate charging strategy can allow not only a larger
amount of solar energy into the power generation system, but
also a higher EV penetration, improving at the same time the grid
flexibility.
Decentralised photovoltaic power plants to satisfy the EV
charging demand could improve the grid operation, while
decreasing GHG emissions not only by themselves but also by
supporting EVs adoption through infrastructures creation. This can
be done by means of electric vehicles solar parking lots (EVSPLs),
i.e., solar parking lots to charge EVs. This way the grid does not have
to integrate a large photovoltaic capacity nor does it need a big
reinforcement to satisfy the EVs' demand. The drivers' preferences
align with this, since they regard EVs and solar panels as

https://doi.org/10.1016/j.energy.2017.09.024
0360-5442/© 2017 Elsevier Ltd. All rights reserved.
 R. Figueiredo et al. / Energy 140 (2017) 1182e1197
complementary [9].
The interest for EVs solar charging stations has been emerging in
the literature, which was thoroughly reviewed in Ref. [10]; other
reviews of technological aspects are provided in Refs. [11] and [12].
Below prior recent research on the topic is summarised.
1.1. Prior recent work
One of the most important objects of the (mostly simulation)
studies concerns charging strategies, which may be of two types:
uncontrolled, where the EVs start charging immediately when
connected to a charger; or controlled, where the charging is
scheduled according to certain criteria, such as energy price or
available renewable energy. The latter implies the use of an opti-
misation algorithm with a certain objective function. Some studies
address both strategies from a price perspective, as in Ref. [13],
using a multi-agent approach; it was found that controlled
charging adherents were responsive to charge with renewables
when that choice was given. Ref. [14] shows that, to reach the same
environmental target, uncontrolled charging requires a higher
storage power and energy capacity than controlled smart charging.
A controlled charging plan works according to what is defined as
objective in the analytical model. The most common models use a
stochastic approach to model the EV charging (see Table 1). Often
the optimisation function is the minimisation of costs (for the
parking operator or EV owners) or maximisation of the use of
renewable electricity.
Costs can be minimised in several ways, such as taking advan-
tage of time-of-use tariffs, as in Ref. [15]. Ref. [16] compares de-
mand response approaches in what regards grid stability,
concluding that an approach oriented to time-of-use prices leads to
reduced costs to the aggregator.
As to the maximisation of renewable energy use, Refs. [17] and
[18] conclude that storage greatly reduces grid dependency. In
Ref. [19] the authors use an optimised PV power extraction tech-
nique to get the most advantage of the solar charging. Power
sources other than photovoltaics are also considered, for example
wind in Ref. [20].
The solar charging stations can be either off- or on-grid. In
Ref. [21] these two types of solar systems are optimised according
Table 1
Summary of the literature.
Characteristics of some solar charging stations' studies
Grid connection Off-grid
On-grid
Charging strategies Uncontrolled charging
Controlled charging Models
Objective function
V2G
Ancillary services
Other power/energy sources Diesel
Wind
Stationary energy storage
Fuel cell
Environmental analysis
Economic analysis
1183
to CO2 emission reduction costs; it is concluded that the cheapest
alternative is the smaller off-grid solution, but the grid-connected
option, bigger, reduces the overall emissions the most. A similar
comparison was performed in Ref. [22], but optimising the system
lifecycle cost; the results showed that a grid-connected station with
storage using a supply mix of on-site PV and diesel generated
electricity was the most economical option, while the renewable
off-grid system was found to be the most expensive.
An on-grid EVSPL may help the grid by behaving as dispatchable
loads, feeding energy into it (vehicle-to-grid, abbreviated V2G) and
providing ancillary services. From this viewpoint, Ref. [23] presents
a fuzzy control method to optimize the schedule of EVs charging.
Ref. [24] presents an EV charging scheduling aiming to comply with
the grid capacity constraints.
A microgrid framework is also used to address EV charging
plans, often considering several energy sources, as in Ref. [25]. The
operational optimisation of microgrids is studied cost-wise in
Ref. [26], for minimising the fuel consumption on a remote island;
in Ref. [27], considering the impact of using different EV charging
strategies; and in Ref. [28], investigating at what extent the V2G
concept can minimise its cost.
The location and sizing of charging stations have also been
addressed, since they can be determinant for the projects feasibility
and its grid impact. Refs. [29] and [30] address this issue from the
grid stability viewpoint, using an evolutionary-based optimisation
technique with positive results.
Besides the energy viewpoint, the environmental and economic
angles are also addressed occasionally. The environmental
perspective is taken in a limited number of EVSPL studies, such as in
Refs. [21] and [22]. The economic angle has been explored under
different approaches, emphasising the standpoint of the operator
[13], the EV user [27] or the overall grid operation [25]. Generally,
these studies show a win-win economic situation for the parties
involved on V2G services. The literature shows that EVSPLs are not
yet financially attractive, presenting payback times between 17 and
26.5 years, mostly due to low revenues and high investment costs
[31].
Table 1 makes a summary of the literature referenced above,
categorising the studies according to their distinctive
characteristics.
[21,22]
[13e30]
[13,14,18,21e23,26,27]
Stochastic [16,21,25,27,28]
Fuzzy control [23]
Multi-agent System [13]
Profits maximisation [13,15]
Renewable electricity use maximisation [13e15,17e20]
Costs minimisation [13,15,16,24,25,27e30]
GHG minimisation [21]
[14,16,19,20,25,28e30]
[13,16,23,29,30]
[16,22,25e28]
[16,20,24,25,27e30]
[14,17,18,21,22,26,27]
[25,27,28]
[21,22]
[13,15,16,21,22,24e30]
1184 R. Figueiredo et al. / Energy 140 (2017) 1182e1197
1.2. Scope of this work
The focus of the literature regarding solar powered parking lots
for electric vehicles has been the energy analysis and EV charging
strategies, mainly using simulated case studies. The use of actual
occupation data for an extensive period e which provides a
detailed realistic input of occupancy patterns often missed with
synthesised datasets e may contribute to a better insight into the
potential benefits of this type of solution, enabling a careful eco-
nomic and financial analysis.
The present study fills this gap by analysing different business
models of solar PV charging to apply to an existing non-solar pri-
vate-owned park-and-ride facility, i.e., a parking lot with public
transport connections to serve commuters. Its core focus is not only
the understanding of the correspondence of EV charging and solar
electricity production, but also the comparison and discussion of
possible business models in order to draw conclusions about their
viability. It proposes a not yet existing comprehensive and
straightforward energy-economic approach to solar parking lots for
electric vehicles.
The paper is structured as follows: Section 2 presents the
methods and its application to a case study. In Section 3 the results
are shown and discussed, and the conclusions are summed up in
Section 4.
2. Approach
This section presents the methods used in this work (Section
2.1), followed by their application to a real park-and-ride facility
(Section 2.2). Several assumptions were made (as explained in
Sections 2.2.1 and 2.2.2), which are combined in different scenarios
(Section 2.2.3). The assumptions are specific to the case under
study, but any others may be adopted.
2.1. Methods
This section presents in detail the methods adopted for this
work.
2.1.1. Solar resource
The energy production at hour i (Eprod (i), in Wh) considers the
global incident irradiance on the modules plan (Gtilted (i), in W/m2),
the area of the modules (Amodules, in m2) and the overall efficiency of
the photovoltaic system (hPV system (i)). It is given by the following
expression:
Eprod ðiÞ ¼ Gtilted ðiÞ  Amodules hPV system ðiÞ
DC/AC Inverters = ~
AC/AC unidirectional chargers ~ ~
Fig. 1. Parking lot system architecture.
The annual PV energy generation (Eannual in Wh) is the sum of all
the hourly energy production throughout the year:
X
8760
Eannual ¼ Eprod ðiÞ (2)
i¼1
The calculation of Gtilted and hPV system are explained below and in
the next subsection.
To calculate Gtilted, a typical meteorological year (TMY) weather
data file for the site can be used [32]. These files, which are freely
available, contain hourly statistical data of several meteorological
parameters for an entire year, such as solar global irradiance, hor-
izontal diffuse irradiance (Dh, in W/m2), direct normal irradiance
(DNI, in W/m2), temperature, etc. [33]. The following expressions
make use of DNI and Dh to calculate the different irradiance com-
ponents (in W/m2) reaching the modules plan: direct (Btilted),
diffuse (Dtilted) and ground-reflected (Rtilted) [34,35].
Btilted ðiÞ ¼ DNI ðiÞ  cosðAOI ðiÞÞ (3)
Dtilted ðiÞ ¼ Dh ðiÞ  ðð1 þ cosðbÞÞ=2Þ (4)
Rtilted ðiÞ ¼ ðDNI ðiÞ  senðaðiÞÞ þ Dh ðiÞÞ  r  ðð1
 cosðbÞÞ=2Þ (5)
Here, AOI is the angle of incidence of the solar radiation on the
module surface [ ]; b is the tilt angle of the module [ ], to be
determined e see Sections 2.2.2.1.1 and 2.2.2.1.2; a is the solar
altitude [ ]; and r is the ground-reflectivity, which for asphalt may
be considered as 0.2 [36]. The parameters AOI and a depend on the
geographical coordinates and are calculated using established solar
geometry methods [35,37,38].
The sum of the previous components gives Gtilted.
Gtilted ðiÞ ¼ Btilted ðiÞ þ Dtilted ðiÞ þ Rtilted ðiÞ (6)
2.1.2. System energy production
The system design is showed in Fig. 1. Its main components are:
(1) the photovoltaic system, assumed to be fixed; it includes a set of
DC/AC inverters; (2) the EV chargers; (3) a Control Center to
manage the system and energy exchanges, according to the energy
management strategy, explained in Section 2.1.3; (4) the electric
grid, which trades energy unidirectionally with the parking lot (it
provides electricity when required); and (5) the adjoining facilities,
(1)
e.g. train station's facilities, which receive photovoltaic energy in
Control Center
Adjoining facilities
~
~
~ ~
 R. Figueiredo et al. / Energy 140 (2017) 1182e1197
surplus.
The overall system efficiency, hPV system , is based on a set of in-
dividual technical specifications of these components and is given
by the following expression:
hPV system ðiÞ ¼ hmodules ðiÞ  hinverters  hcables  hchargers
The efficiencies of the inverters (hinverters ), cables (hcables ) and
chargers (hchargers ) are considered constant, and the modules' effi-
ciency (hmodules ðiÞ) depends on the ambient temperature. When
the EV chargers' efficiency is already included in the EV con-
sumption, as often is (see Section 2.1.3), hchargers should be assumed
as 1 in Eq. (7), to avoid double counting [39].
The temperature is given by Eq. (8), and it is a product of the
modules nominal efficiency (hnominal ) and the temperature related
efficiency (hTamb ðiÞ), which decreases as the ambient temperature
rises according to Eq. (9) and Eq. (10) [40].
hmodules ðiÞ ¼ hnominal  hTamb ðiÞ
Tmodule ðiÞ ¼ ðNOCT  TambNOCT Þ=GNOCT  Gtilted ðiÞ þ Tamb ðiÞ
hTamb ðiÞ ¼ ð100%  TC PMPP Þ  DTamb ðiÞ
Here, Tmodule ðiÞ is the module temperature at hour i [K]; NOCT is the
Normal Operating Cell Temperature (NOCT) [K]; TambNOCT ¼ 20  C is
the ambient temperature for NOCT conditions, [40];
GNOCT ¼ 800 W/m2 is the solar irradiance for the same conditions;
Tamb ðiÞ is the ambient temperature [K]; DTamb ðiÞ is the difference
between the temperatures of the environment and the module, and
TC PMPP is the temperature coefficient at the maximum power
point, i.e., the efficiency degradation factor with temperature of the
module, in %/K, given in the technical datasheet.
2.1.3. Energy management
The average daily distance driven by a typical car owner varies
between countries, and even between regions, but generally is
under 40 km [41]; in Portugal, for example, it is 35 km [42]. This
study assumes that the batteries of the cars are topped up ac-
cording to the average energy needs. Using the same example, in
Portugal each vehicle needs to charge daily 6 kWh on average,
assuming an EV energy consumption of 0.171 kWh/km [39]. Given
the large number of clients, it is assumed that fluctuations in energy
requirements between individual vehicles offset each other.
Two types of charging are simulated: uncontrolled and
controlled. In both approaches, after achieving the period necessary
to complete the daily charge of each vehicle ( tcharge , Eq. (11)), the
vehicle stops charging and its demand becomes null. The energy
supply options are the PV system and the grid, and in some cases
also a stationary storage battery. In the uncontrolled charging, the
EVs charging starts as soon as they arrive at the park.
Edaily demand
tcharge ¼
Pcharger
where Pcharger is the charging rate, and Edaily demand is the daily
energy demand [kWh].
The controlled charging assumes that the number of vehicles
charging at every hour, NEV ðiÞ, considers the number of vehicles still
charging from the previous hour and is defined by the following
expression:
1185
Eprod ðiÞ  CEV ði  1Þ
NEV ðiÞ ¼ (12)
Pcharger
Here, Eprod ðiÞ is the PV energy production at a certain hour and
(7) CEV ði  1Þ is the power demand in the hour i correspondent to the
EVs that started charging previously and are still doing it.
The energy management of the park is hereafter briefly
described. The photovoltaic generation is primarily used for sup-
plying the EV charging, and in the case of surplus it is stored in the
central battery of the EVSPL. Finally, if there is no EV demand to
supply and if the storage battery is completely charged, the
generated electricity is sold to the adjoining facilities. On the other
hand, if neither the PV generation nor the storage battery are
enough to supply the EV load, grid purchases are required. The
energy management strategy is summarised in Fig. 2 for both
controlled and uncontrolled charging plans. This method is applied
to seasonal, working and weekend mean days of the parking lot, as
(8) Section 3.1 shows.
2.1.4. Economic feasibility
The economic balance is based on the revenues and costs during
(9)
the project, which may be assumed to last 25 years, which is
generally the lifetime warranty of PV modules [43]. This section
(10) presents the analytics applied to the economic component, and also
the common assumptions to all the scenarios tested. The assump-
tions that are scenario dependant (PV investment costs, etc.) are
explained in Section 2.2.2.4. It is only considered the PV-EV
component of the parking lot, and it is treated as if it is indepen-
dent of the other components (parking space rent, etc.), i.e., the cost
of the operation already in place is considered to be sunk.
The nominal value, i.e. cash flow, of energy traded between the
park and the grid/facilities on the year n of the project (the in-
vestment year corresponds to n ¼ 0), V(n), is calculated according to
the following expression:
V ðnÞ ¼ EðnÞ  c  ð1 þ iÞn (13)
where EðnÞ is the amount of energy traded during that year; c is the
energy unit price; i is the inflation rate. Equation (13) is used to
calculate both the grid sales and the facilities purchases to the park.
The EV demand does not change from year to year of the project,
but both the sales and purchases are dependent on the annual
energy production, Eprod , which varies with the modules degrada-
tion, as in Eq. (14).
Eprod ðnÞ ¼ Eprod ð1Þ  ð1  aPV Þn1 (14)
where aPV is the degradation rate of the modules energy yield per
year.
Four different well established financial indicators are used to
determine the viability of the project: net present value (NPV),
discounted payback time, internal rate of return (IRR) and dis-
counted return on investment (ROI) [44]. The NPV represents the
cumulative discounted balance of revenues and expenses for every
year n of the project.
Eq. (15) is used to calculate the NPV (in V), where t is the project
(11)
duration and r is the nominal discount rate.
X
n¼t
net cash flow ðnÞ
NPV ¼  Investment (15)
n¼1
ð1 þ rÞn
The discounted payback time indicator (abbreviated below as
‘payback’) shows the elapsed years required until the NPV achieves
a positive value, i.e., the required time until the revenues equal the
1186 R. Figueiredo et al. / Energy 140 (2017) 1182e1197
Fig. 2. Energy management diagram.
costs (both discounted) and the initial investment. It is calculated
through Eq. (16).
 
cumulative cash flowdisc: ðYÞ
discounted payback ¼ Y þ  
 (16)
cash flow ðY þ 1Þ
disc:
where Y is the elapsed years since the beginning of the project
correspondent to the last year with a negative cumulative dis-
counted cash flow [years], and disc. stands for ‘discounted’.
The IRR is the discount rate that nullifies the NPV. The IRR is
determined using the discount rate adopted, and therefore the
decision rule to invest is if the IRR is above this rate [45].
Finally, the discounted ROI (abbreviated below as ‘ROI’) is the
number of times that the investor receives the amount initially
invested. The ROI is calculated through Eq. (17), where it should be
considered the discounted cumulative at the end of the project net
benefits and expenses, and the investment.
P during the day, but also throughout the year.
ðrevenues  expensesÞdisc:
discounted ROI ¼ P (17)
expensesdisc: þ investment
2.2. Case study
To test the methods above, the commuter parking lot from the
train station Pragal in Almada (38.66ºN, 9.18ºW), Portugal, a city
located south of Lisbon across the river Tagus (Fig. 3), was chosen.
This location has one of the highest solar irradiation potential in
Europe, with an average daily irradiation of 5.15 kWh/m2 [46].
This station is located in a municipality of 174,030 inhabitants
[47], and serves passengers traveling on the working days between
Almada and the capital, on a train line accounting for 25.6 million
trips every year [48].
For the study, the parking lot occupation data from the entire
year of 2014 was gathered. It consists of hourly records of the
number of arrivals and departures for both monthly subscriptions
and daily tickets, the two payment types that exist. The park has a
capacity for 1045 cars and is open every day from 7 a.m. to
midnight. Each parking space uses around 12.2 m2 of area, which
accounts for about 12,700 m2 of available area for car shading
(about half of the total parking lot area) [49]. The layout is mainly of
perpendicular parking spots oriented towards the southeast
(north-based azimuth: 150 ).
The occupation pattern of a park-and-ride facility is dependent
on the users' working schedule, and so it follows a bell curve on
weekdays. The average occupation rate of the parking lot is 23%,
and its maximum occupation is 694 vehicles, i.e., about two-thirds
of the park capacity. Fig. 4 shows the average profile of the weekly
occupation rate for the two weeks with maximum and minimum
occupation along the year (in March and in December, respectively)
and for the entire year. As it was expected, during the weekends the
occupation is significantly lower. The least occupied week is not
representative of the park use, as it was observed during Christmas.
The comparison between the two extreme weeks and the differ-
ence in occupation between workdays and the weekend show that
the occupation varies a lot in different timescales, i.e., not only
2.2.1. General assumptions
To test the methods, several assumptions had to be made. This
section describes those that are common to all the scenarios (Sec-
tion 2.2.3).
The climate data was obtained through the TMY file for Lisbon
[32]. The surface of the parking lot is asphalt, thus the ground-
reflectivity is assumed as 0.2 [36] (see Section 2.1.1). The tech-
nical components that were assumed for the system and their
technical data are summarised in Table 2.
One common assumption, unless otherwise stated, is that the
EVSPL occupation is only of EVs. The daily electricity demand of
each vehicle is considered to be 6 kWh (see Section 2.1.3). The
charging rate (Pcharger ) is assumed as 3.7 kW, corresponding to Level
1 AC Charging. The charging infrastructures are sized using the
observed maximum occupation of the parking lot.
The project lifetime is 25 years (similar to the PV modules
typical lifetime [43]). Since batteries typically have a lifetime of 10
years [53], shorter than the project, the investment should account
for their replacement, and so a second battery had to be accounted
in the investment. Due to technology improvements, it has been
considered that the second battery would last 15 years.
 R. Figueiredo et al. / Energy 140 (2017) 1182e1197 1187

Fig. 3. The parking lot and the metropolitan area where it is located.

0.7 they are assumed to be constant (in real terms) throughout the
project. The capital investment is considered to cover all the
Maximum occupation fraction

0.6
hardware, such as the photovoltaics modules, metallic structures,
0.5
charging stations, inverters, storage battery, cables, and labour.
0.4 It is also assumed that the EVs owners pay 0.25 V/kWh to charge
0.3
their cars, which is less than the average charging fee in Europe
(about 0.35 V/kWh, according to [59]). Even though home charging
0.2 could be cheaper (about 0.17 V/kWh [60]), charging at the park is
0.1 still attractive, e.g. due to convenience or the avoidance of the need
to have access to a charging station at home e a public one might
0
Monday Tuesday Wednesday Thursday Friday Saturday Sunday not exist and a private one is expensive and sometimes is not
Highest occupation Lowest occupation Average occupation possible to install.
It is also assumed that the EVSPL sells energy to the adjoining
Fig. 4. Occupation profile during the most and least occupied weeks and the average of
train station's facilities for the same price that the facility is
the entire year.
currently paying to the utility, which is better (0.072 V/kWh) than
the price at which the park sells energy to the grid (0.050 V/kWh)
In the end of the components life, their residual value is [61]. This selling is actually virtual for the case studied, since the
assumed to be 20% of the initial value [54]. Table 3 shows the in- operator of the train and parking facilities is the same. Additionally,
vestment costs' for the PV modules, stationary battery (see Section the parking lot purchases energy from the grid at the same price it
2.2.2.2 for the details on energy storage) and charging stations, both sells to the facilities (again, 0.072 V/kWh).
for 2015 and 2050 (2050 prices are given due to a delayed invest- Regarding annual expenses, the EVSPL operational and main-
ment and gradual implementation scenarios e see Sections tenance cost (cleaning, repairs, etc.) is considered to be 1% of the
2.2.2.4.2 and 2.2.2.4.3). investment, and the managing cost (administration, customer
As for the energy purchase prices between the stakeholders, support, etc.) is assumed to be 5% of the annual revenues. The
1188 R. Figueiredo et al. / Energy 140 (2017) 1182e1197

Table 2
System technical details.

Modules [43] Sunmodule Plus SW 285 mono

Technology Monocrystalline silicon

Maximum power Standard Test Conditions e STD (STC - 1000 W/m2; 25  C; AM 1.5) 285 Wp
NOCT (800 W/m2; NOCT ¼ 46  C; AM 1.5) 213.1 Wp
Module area 1.67 m2
Nominal efficiency 17.17%
Temperature coefficient at the maximum power point, TC_PMPP 0.41%/K
Degradation rate [50] 0.8%

Inverter efficiency [51] 98.8%

Cable efficiency [52] 95%

Table 3 2.2.2.1. Parking lot orientation. The planning of a PV installation

Investment costs' for the PV modules, lithium-ion battery and charging stations.
Investment costs
Year
2015
PV modules [V/kWp] [55] 550
Lithium-ion battery [V/kWh] [56] 469
Charging stations [57,58]
Charger [V/unit] 1600
Aggregator [V/unit] 2500
charging management (Control Center in Fig. 1) is considered to be
done by the charging stations control system, and so its cost is part
of the charging stations cost.
Finally, the nominal discount rate (r) and the inflation rate (i, to
determine future cash flows) are considered to be 7% [45] and 2%
[62], respectively.
Table 4 summarised this data, as well as some other relevant
parameters.
2.2.2. Scenarios assumptions
Based on the conditions and methods explained above, the
analysis and assessment of the EVSPL project viability are per-
formed for a set of additional assumptions and principles, which
are then combined to form several scenarios. Then, the scenarios
are tested one by one and compared. This section presents and
explains each one of these assumptions.
The criteria to establish the assumptions and subsequently the
scenarios are to reflect: (1) the actual conditions of the existent
parking lot; (2) different conditions that other similar parking lots
might have, such as a different geographical orientation; (3) pre-
sent and future technology and respective adoption level, such as
EV market penetration; (4) simple conditions as well as more
evolved ones that might turn the project viable, such as simple and
more complex charging strategies and plans.
must account for a modules layout that ensures minimum mutual
shadowing. In this case study, two logical layouts arise: (1) modules
southeast orientated, which is the park orientation; and (2) mod-
2050 ules south orientated, which maximises the modules specific en-
ergy yield. Both cases were simulated in Sketchup using the plug-in
246
168 Skelion [63], to determine the number of modules to be installed,
assuming standard modules measuring 1.67 m of width by 1 m of
806 height. The two possible layouts are shown in Fig. 5.
1260
The southeast orientation enables the installation of 3480
modules, while the south one allows 2200 modules, i.e., 37% fewer
modules, a considerable difference. To understand what would be
the best configuration, PVGIS [46] was used to perform a simple
estimation of the annual energy yield. The layout with the highest
energy generation was chosen for the rest of the study. The results
showed 33% lower energy generation for the south orientation; in
this case a better orientation does not compensate for fewer
modules. For this reason, the first layout (on top of Fig. 5) is used for
the rest of the study, allowing for about 5800 m2 of photovoltaic
area, referred as Amodules.
The parking lot geographical orientation, ceteris paribus, de-
termines the solar energy harvest. Two different orientations were
tested, as explained below.
2.2.2.1.1. PV layout I. The ‘PV layout I’ corresponds to the actual
orientation of the case study, which is southeast (see Section 2.2).
Using Excel Solver, a genetic-algorithm's optimisation was per-
formed having as objective function the maximisation of the yearly
energy production by varying the modules' tilt (the azimuth was
fixed to southeast). This kind of technique uses mating, crossover
and mutation to find the best solution, and is often used to solve
solar energy problems [64]. The population size was of 200 and the
mutation rate was assumed as 7.5%. The tilt range was constrained
between 26 and 38 , to shorten the optimisation time. The opti-
misation takes less than 5 min to run with an Intel Core i3 pro-
cessor. The optimum tilt was found to be 33 , and as can be seen in
Table 5, this layout provides an energy generation of 1299 MWh/
year.

Table 4
Energy prices and other assumptions.

Energy purchase prices EVs to EVSPL 0.25 V/kWh

EVSPL to the grid 0.072 V/kWh
Train station facilities to EVSPL 0.072 V/kWh

Lifetime Project and PV modules [43] 25 years

Stationary energy storage [53] 10 (1st) and 15 (2nd) years

Economic assumptions Residual value [54] 20%

Operational and maintenance cost 1% of the investment
Managing cost 5% of the annual revenues
Nominal discount rate, r [45] 7%
Inflation rate, i [62] 2%
 R. Figueiredo et al. / Energy 140 (2017) 1182e1197 1189

50 m

50 m

Fig. 5. Layout of the PV installation orientated southeast (above) and south (below).

Table 5
Characteristics and production of the photovoltaic installation.

Orientation Azimuth (º) Modules' tilt (º) # of PV modules Rated power [MWp] Energy generation [MWh/year] Energy generation per area [MWh/m2]

Southeast 150 33 3480 1.0 1299 0.22

South 182 32 1491 0.25

Table 6
Summary and description of the analysed scenarios in this work.

Scenarios Parking lot orientation Charging plan Other assumptions Energy Economic feasibility
storage [kWh]
PV layout I PV layout II Controlled Uncontrolled Delayed Gradual Base Flat fee Grid support
implentation deployment

A0 - Reference (uncontrolled w/ storage) x x 1,700 x

A1 - Optimal orientation x x 1,800 x
A2 - Delayed implementation x x x 1,700 x
A3 - Uncontrolled charging w/o storage x x - x
A4 - Controlled charging w/o storage x x - x
A5 - Gradual deployment x - - x - x

B0 - Flat fee charging x x 1,700 x

C0 - Frequency’s secondary regulation x x 1,700 x

2.2.2.1.2. PV layout II. Since in a parking lot yet in its design i.e., as if the park could be rotated to it, named ‘PV layout II’. The
phase the geographical orientation might possibly be chosen, it was optimisation procedure was as explained above, with the difference
decided to test an orientation corresponding to the optimum one, that the azimuth was also a variable. A boundary of þ15 and 15
1190 R. Figueiredo et al. / Energy 140 (2017) 1182e1197

Assumptions

Parking lot orientation Charging plan Other assumptions

PV layout I EV controlled charging Present

• Southeast orientation • Maximum EV solar charging The project is implemented in the
• Objective function: optimum • Variables: hourly arrivals, present
capacity factor tcharging, PV production
• Variables:
Delayed implementation
EV uncontrolled charging It considers that the EVSPL is
PV layout II
• EVs start to charge immediately implemented in the future,
• Variable orientation
as they arrive considering price learning curves
• Objective function: optimum
• Variables: hourly arrivals, tcharging for PV and battery until 2050
capacity factor
• Variables: and azimuth
Economic feasibility Gradual deployment
Energy storage The project and its structures are
implemented in time steps
Base
Revenues are the energy sales to
Central battery
EVs and to the facilities
• The surplus of PV production is
stored for use during peak hours Flat fee
• Sizing is based on the average Unlimited charge by means of a
PV excess during working days fixed monthly payment

Grid support
Frequency’s secondary regulation
provided by the park to the grid

Fig. 6. Assumptions summary.

1200 1600
a. b.
1400
1000
1200
800
Power [kW]

1000
Power [kW]

600 800

600
400
400
200
200

0 0
1 3 5 7 9 11 13 15 17 19 21 23 1 3 5 7 9 11 13 15 17 19 21 23
Daytime [h] Daytime [h]
PV gen. - Winter PV gen. - Spring PV gen. EV Load - Average day
PV gen. - Summer PV gen. - Autumn EV Load - Working day EV Load - Weekend
EV Load

Fig. 7. a: Average profiles of seasonal PV production and annual EV load; b: average profiles of EV load (working day and weekend) and annual PV power.

around south was imposed to the possible azimuth range of values.
The optimum tilt and north-based azimuth were found to be 32
and 182 , respectively, enabling an energy generation of
1491 MWh/year.
The PV generation results for both approaches are presented in
the following resulted in Table 5. Both options enable the installa-
tion of 3480 photovoltaic modules (see Section 2.2.2.1), however
with optimal orientation the annual energy generation increases
around 14.8%.
2.2.2.2. Energy storage. Since the EV load does not perfectly follow
the PV production, an energy storage device acting as an energy
buffer between the PV system and the EVs can increase the solar EV
charging. It stores the surplus PV production, using it to charge the
EVs when there is not enough sunshine. To simulate this, a lithium-
ion battery e the most promising battery technology for renewable
energy applications [65] e was considered, sized according to the
total PV excess of energy, counted hourly, that occurs on average
during the working days plus the energy losses in a charge/
discharge cycle (shadowed area in Fig. 7b). This was considered to
be the appropriate storage size to maximise the solar EV driven
fraction, i.e., the distance driven by the cars with solar electricity
(see Section 3.1.1); if the criteria were different, e.g. techno-
economic, the storage size could also be optimised using the
same procedure of Section 2.2.2.1.1.
The round-trip efficiency of the battery was considered 92.5%
[53]. Its size varies with the scenario (see Table 6 at the end of this
section, where the battery sizing presented is already
 R. Figueiredo et al. / Energy 140 (2017) 1182e1197 1191

Table 7 2.2.2.4.1. Present. The EVSPL as described is assumed to be

Average of daily correlations between the PV production and the occupation profiles. totally implemented in the present. This scenario assumes that all
Averaged daily correlations cars parking at the EVSPL are electric, as if it was an EVSPL exclusive
Day of the week Working day Weekend
for EVs implemented today. The charging infrastructure is sized
52% 36% according to the maximum annual occupation of the park.
2.2.2.4.2. Delayed implementation. The project feasibility is
Seasonal Winter Spring Summer Autumn
59% 45% 31% 50% studied for the future, i.e., analysing its implementation in the
future by considering the beginning of the project in each one of the
Annual 47%
years until 2050. For that, it was considered the modules', batteries'
and EV chargers' expected cost decrease until 2050, Table 3. For the
charging stations, it was assumed that their cost follows the trend
correspondent to its useful energy capacity). of the EV's expected costs [57], using as start value the commercial
current unit price presented in Table 3 for each charger and
2.2.2.3. Charging plan. As explained in Section 2.1.3, two charging aggregator (the latter is needed for each groups of ten chargers
plans were considered, uncontrolled and controlled. The charging [58]), to which a 20% discount is applied due to a large volume
schedule is according to the order of arrival. In both plans it is purchase.
assumed that the grid connection and the capacity of the distri- Finally, the calendar year in which the payback time achieves for
bution network are sufficient to meet all the EVs demand. the first time an attractive payback time for a typical investor e
2.2.2.3.1. Uncontrolled charging. The uncontrolled charging as- assumed as 7 years [31] e was identified to be 2030, which was
sumes that when a vehicle arrives at the park it starts immediately used for the results having this assumption.
charging. Primarily, the EVs are fuelled by the PV, and by the central 2.2.2.4.3. Gradual deployment. The gradual deployment of the
battery if the scenario considers it; when that is not sufficient, project corresponds to an approach in which the EVSPL structures
energy is imported from the grid. are gradually installed following the increase of EVs' market share
2.2.2.3.2. Controlled charging. The controlled scheduling fol- [67]. The EV energy demand assumed for each year is directly
lows only one criterion: to maximise solar EV charging. This is proportional to the corresponding EV market penetration,
achieved using a simple algorithm that manages the charging of the assuming an evolution which considers 50% of EV share in the 2050
vehicles according to the PV energy availability (see Section 2.1.3). car stock. It represents a level where the EVSPL is always able (in
The short-term forecast of PV production is increasingly more real time) to charge the EVs with 100% on-site generated electricity,
precise [66], and so it is assumed that the PV energy generation for and so the uncontrolled and controlled approaches lead to the same
each day is known in advance, as well the EV hourly occupation. As results.
before, when the PV energy is not enough, energy is imported from The investment in the charging stations and PV modules follows
the grid. the same learning curves of the ‘Delayed implementation’
assumption (Section 2.2.2.4.2). The total PV installation was
2.2.2.4. Other assumptions. Since there are different deployment assumed to take place in the project beginning, since it showed
strategies, the following alternative project calendar options were better results than its gradual implementation, mainly due to the
considered. large energy sale to the facilities from the project beginning when
the number of EVs is very low. Through the project, these energy
sales to the facilities decrease as the number of charging EVs
Table 8
EV demand and PV production (Scenario A0).
increases.
When assuming a gradual deployment of the project, an expo-
Averaged daily EV demand and PV production [kWh]
nential trend line is applied to calculate the differential residual
EVs' demand PV production PV net balance values of the equipment, depending on the year it was installed.
Annual mean day 2805 3558 753
2.2.2.5. Economic feasibility. The perspective taken is that of the
Winter 2723 2445 279
Spring 2967 3528 560 park operator. As described in Section 2.1.3, it is assumed that the
Summer 2745 4208 1462 EVSPL sells primarily its PV production to the EVs; its surplus, if
Autumn 2768 4009 1241 exists, is firstly stored in the central battery (if the scenario con-
Working day 3883 3480 353 siders it) and ultimately sold to the train station's facilities. The
Weekend 222 3754 3532 revenues and expenses are calculated as described in Section 2.1.4.
These parameters vary with the assumptions.
2.2.2.5.1. Base. It is considered that the revenues come from the
Table 9 sale of energy to the EVs owners, to the grid and to the train sta-
EV charging sources for the uncontrolled charging approach with storage (Scenario tion's facilities. The expenses consist of the energy purchase from
A0).
the grid, the capital investment, and the operational, maintenance
Averaged daily EV charging sources by its sources [kWh] and managing costs of the parking lot.
EV charging 2.2.2.5.2. Flat fee. The flat fee charging considers the same
sources of revenues and expenses as the Base case, but it assumes
Directly from PV From the battery From the grid
two types of customers: (1) customers that pay a monthly fee of
Annual mean day 2025 780 0 V35 and are able to unlimitedly charge their vehicles; (2) clients
Winter 1378 1066 279 that pay a fixed one time charging fee of V1.5 plus the energy
Spring 2054 913 0 consumed. These values are additional to the regular parking fees.
Summer 2413 332 0
Their distribution is according to the distribution of monthly pay-
Autumn 2099 669 0
ments and daily tickets that the parking lot has presently, which is
Working day 2203 1327 353 of 76% and 24%, respectively. The monthly flat fee subscribers are
Weekend 222 0 0
considered heavy users because, since they do not pay for extra
1192 R. Figueiredo et al. / Energy 140 (2017) 1182e1197
fuel, they charge their cars 50% more than those with daily tickets
(9 kWh and 6 kWh, respectively).
2.2.2.5.3. Grid support. The V2G approach offers an opportunity
for the EV facilities to provide ancillary services to the grid, since
they represent controllable power capacity [8]. This business model
adds to the Base case the opportunity to offer secondary frequency
regulation to the grid, being this a remunerated service. This type of
regulation contributes to maintain the grid voltage stable, and its
provision does not imply significant energy trading. It considers as
bidirectional the EVSL-grid power flow. The contracts are based on
the power capacity offered during a certain period of time by the
facility [68]. The power capacity offered by the EVSPL was assumed
as being equivalent to the number of EVs parked multiplied by the
connection power of each one (3.7 kW). To that, it is applied the
current secondary frequency regulation price paid by the grid,
which is of 25.6 V/MW/h [69].
This is considered as an additional revenue of the parking lot
(compared to the Base model), besides the energy sales.
2.2.2.6. Summary. A summary of all the previous assumptions is
presented below, Fig. 6.
2.2.3. Scenarios
Several scenarios were created concerning the EVSPL operation
and financial model; they are presented in Table 6.
The main concern when configuring the scenarios was to show
practical alternatives for an investor. To keep the study constrained,
not all the possible combinations of assumptions were tested.
Instead, the scenarios A0 to A5, based on the Base financial model,
were constructed as a sequential improvement on the project
viability. The other ones (B0 and C0) are representative of alter-
native financing options, and were tested having as basis the
Reference scenario. A scenario considering both controlled EV
charging and storage was discarded because of redundancy, as the
aim of both systems is the same.
3. Results and discussion
The results and their discussion are divided between the energy
and the financial analyses. For the sake of containment, the energy
analysis is shown just for the most illustrative scenarios. The
financial analysis is shown for all the scenarios since it is the one
determinant for the EVSPL feasibility assessment.
1200
1000
800
Power [kW]
600
400
200
0
1 2 3 4 5 6 7 8
PV gen.
Controlled charging
Fig. 8. Average uncontrolled and controlled EV load and PV generation profiles.
3.1. Energy analysis
Below the results for the scenarios considered to be the most
representative, i.e., scenario A0, representing the uncontrolled
charging approach with energy storage, and scenario A4, corre-
sponding to the controlled charging approach without energy
storage are presented.
The correlation between the PV production and the park occu-
pation along the days is an important parameter for the profit-
ability of an EVSPL, because it translates to what extent it is possible
to directly solar-charge the EVs with on-site generated solar energy.
For each day, the correlation between the parking lot occupation
and the PV production was determined, enabling the calculation of
the average correlation factor for several representative periods of
the year, Table 7.
As expected, the working days show a far better correlation than
weekends, since the weekend occupation is flatter during the day,
and thus farther from the PV production profile. As for the seasons,
the worst correlation is in the summer: since the park is southeast
oriented, it produces significantly more during the mornings in the
summer (Fig. 7a), which results in a more asymmetric PV profile. In
contrast, the production profile during the winter is more sym-
metrical (Fig. 7b) and therefore it relates better to the occupation
(see Fig. 4). As it was expected, the average correlation for the
whole year is between these values.
3.1.1. Uncontrolled EV charging plan
Despite of the importance of the occupation rate in the parking
lot and its correlation with the photovoltaic production, the actual
demand that the project must support is not correspondent to
every vehicle in the park, but only from those charging. The
charging plan dictates the EV load profile, which is shown in Fig. 7a
for the ’Uncontrolled’ approach, along with several representative
PV production profiles.
For the scenario A0, with storage, the energy demand, produc-
tion and surplus PV production is stated in Table 8 for the afore-
mentioned representative days.
Table 9 presents the fraction of the EV demand supplied by each
energy source, i.e., directly by the PV production, the battery and
the electric grid.
All periods of the year provide enough PV energy to satisfy the
EV demand, except the winter. In the summer, 87.9% of the EVs'
demand is directly satisfied by PV generation, which means that the
battery in this period is mostly used to store energy produced
during the weekends; it is also in this period that most of the
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Daytime [h]
Uncontrolled charging
Controlled charging - Working days
 R. Figueiredo et al. / Energy 140 (2017) 1182e1197
energy selling to the facilities occurs, after the battery is full. During
winter, the direct solar EV charge is of 50.6%. For the whole year, it is
of 72.1%, wherein battery supplies 27.8% of the EV load (this cor-
responds to the charging load that surpasses the PV generation
between 8 a.m. and midday). The working and winter days are the
only ones showing PV generation deficit, thus requiring grid pur-
chases (9.1% and 10.2%, respectively).
The energy storage, as described in Section 2.2.2.2, was sized in
order to maximise solar EV charging in a typical working day, and
so its size corresponds to the shadowed area in Fig. 7b, which
represents the amount of energy that the PV system cannot fulfil
directly (1700 kWh). The use of the energy storage is crucial to
allow a high solar EV driven fraction; on average, in a working day
the solar EV driven fraction increases from 56.7% without storage to
90.1% with it.
3.1.2. Controlled EV charging plan
The controlled charging approach without storage (Scenario A4)
allows a much better coincidence between the EV load and the PV
generation profiles, as it can be seen in Fig. 8. The figure also shows
the uncontrolled charging profile, to better perceive the differ-
ences. Even though both charging plans represent the same daily
EV demand and PV generation (Table 8), the controlled charging
allows less grid or battery dependence by peak shaving the EV load
according to the PV generation.
Table 10, analogous to Table 9, quantifies the EV charging by
source. As in scenario A0, the grid is mostly required during the
working winter days. The small amount of grid energy required in
the summer (on average 15 kWh/day) is due to a few vehicles that
arrive late in the day at the parking lot, when the PV generation is
no longer able to supply them.
Even though in scenario A0 the working days EV demand is
much larger than the PV generation until the middle of the day, the
grid dependence in scenario A4, without that difference, increases
about 67.4%. This is because in scenario A0 the early morning PV
generation, a period during which the park is still unoccupied, is
stored for later use, whereas in scenario A4 it is sold to the facilities.
As a result, in A4 the grid has to supply the late afternoon EV
charging. Even though the controlled charging in the working days
increases the energy purchases to the grid (shadowed area in
Fig. 8), the direct solar EV charge is much higher, as it can be seen
when comparing Figs. 7b and 8. Replacing the energy storage by a
controlled charging decreases slightly the solar EV driven fraction,
from 90.1% (see Section 3.1.1) to 84.8%, showing that this scenario
does not compromise significantly the objectives, even though it
avoids a heavy investment on storage.
3.2. Economic feasibility analysis
The financial analysis is performed first for the scenarios based
Table 10
EV charging sources for the controlled charging approach w/o storage (Scenario A4).
Averaged daily EV charging by its sources [kWh]
EV charging
Directly from PV From the grid
Annual mean day 2805
Winter 2441
Spring 2967
Summer 2731
Autumn 2768
Working day 3292
Weekend 222
1193
on the reference case (A0-A5), because they all have the same
revenues base, and then for the scenarios B0 and C0, based on a
different revenue model.
Fig. 9 shows the discounted cumulative revenues, expenses and
profit for scenarios A0-A5.
Comparing scenarios A1 and A2 with A0, the three scenarios
that consider energy storage, it is noteworthy the 35% reduction on
the investment costs in A2, which refers to a delayed project
implementation for 2030, as explained above. As for the results
differences between A1 and A0, the optimal orientation from sce-
nario A1, as opposed of the southeast from A0 (they only differ in
this regard), allows for fewer grid purchases, since the energy yield
is higher, and thus higher profits (þ11%).
The scenarios without energy storage (A3 onwards) show clearly
lower investments and O&M costs, due to the absence of the bat-
tery related expenses, and better overall results, except A5 (the
reason is addressed below). As expected, when an uncontrolled
charging without battery is assumed (A3), the grid dependence of
the EVSPL and the energy export to the facilities increase signifi-
cantly. However, the profits show an improvement of 140% over A0,
meaning that the use of energy storage allows for a greater fraction
of solar EV charge, but financially is not a good option.
Compared to the uncontrolled plan without storage (A3), the
controlled charging (A4) allows about the same profits (þ1%), 65%
less grid dependence and 40% less energy export to the facilities. If
compared to A0, A4 has 142% better overall operational results,
even though grid imports increase (þ13%). This is caused by the EVs
that arrive late, when PV is not enough to their supply.
Scenario A5 shows the lowest investment costs, because it does
not consider storage and its gradual deployment allows the
investor to install the components at cheaper prices as the time
goes on. The lower EV load avoids the need to import energy from
the grid. Compared to A0, A5 exports more energy to the facilities
by a factor of three and has 44% lower profits. The significant lower
sales to EVs (less 88% than in scenario A0) are the main reason for
this improvement.
The results of the scenarios with alternative revenues models
(B0 and C0) are presented in Fig. 10, along with the results for the
base case, for comparison.
In A0 and C0, the main source of revenues is the energy sold to
the EVs, even though in C0 there are extra revenues coming from
the grid support (they represent 5% of the total income). In B0 the
daily and monthly fees are the main revenues source. Even though
there is an extra income source from A0 to B0, the grid imports also
increase (by a factor of four), since it is assumed that the monthly
fee subscribers charge more. Hence, the project's profit suffers a
strong decrease of 85%. Scenario C0 increases its income, which
leads to 28% higher profits. As for the energy export to the facilities,
they do not differ much between A0, B0 and C0.
The financial indicators concerning all the scenarios are pre-
sented in Table 11.
It should start to be noted that all the scenarios are viable for all
the indicators, being the payback time between 7 and 24 years e
which is in agreement with the literature, e.g. Refs. [70] and [31], in
which the payback time is 4e26 years and 17e26.5 years,
respectively.
The base case scenario, A0, presents a NPV of 0.87 MV and a
payback of c. a. 16 years, which may be seen as average results.
0 These results only improve slightly in scenario A1 (e.g., the NPV
289 raises c. a. 10%), due to the optimal PV layout configuration
0 considered in the latter. In contrast, scenarios A2, A3 and A4 pre-
15
sent all much better indicators than the predecessors, which can be
0
attributed to the absence of energy storage and the investment
591 implied.
0
Scenario A4, which considers a controlled charging plan, shows
1194 R. Figueiredo et al. / Energy 140 (2017) 1182e1197

4
Energy sales to EVs
3
2.08 2.10 Energy sales to train station
1.85
2 facilities
0.87 0.96
0.48 Grid purchases
1
[M€]

0 O&M&Management

-1
Investment
-2
Profit
-3

-4
A0 A1 A2 A3 A4 A5

Fig. 9. Revenues, expenses and profit for scenarios A0-A5 for the project lifetime. Positive values are revenues and negative are expenses.

the best results. They are however marginally better than the re-
sults of scenario A3, which is similar but considering uncontrolled
charging. This was expected, since the prices of buying energy from
the grid and selling energy to the train facilities e both the trans-
actions happen more in A3 e are considered to be the same, and so
the small differences in the results may be attributed to different
efficiencies of directly using the electricity to charge the vehicles
and send it and use it at the facilities. As for A5, considering gradual
deployment, its NPV is the lowest of the scenarios based on the
reference revenues model (A0-A5), but its IRR is better than some
of them (scenarios A0 and A1); this is due to different patterns of
cash flows e in A5, the initial energy sales to EVs are lower and to
the train facilities higher, due to a higher PV surplus than in A0 and
A1.
As for the scenarios considering alternative revenues models,
scenario B0, the one that considers a monthly flat fee, presents the
worst results, being profitable just by a small margin, since it has a
low NPV and an IRR slightly above 7%. This is due to the assumption
that the cars charge 50% more with this plan, and so it seems that
under these conditions the flat fee model is not appealing. How-
ever, a flat fee business model can lead to different occupancy
patterns and rates, since it may be seen as a marketing strategy to
get new customers. As for scenario C0, similar to A0 but with
additional revenues from supporting the grid, it shows a 27.5%
bigger NPV and a 1.5-year lower payback than the latter, which is
indicative that the grid support improves significantly the project
profitability.
Since an investor in this kind of project is greatly driven by a
short payback time, desirably less than 10 years, and typically 5e7
years [31], this puts scenarios A2, A3 and A4 as the appealing ones,
with paybacks of 7.3e7.4 years. Moreover, those scenarios also
come with better NPV, IRR and ROI.
3.2.1. Sensitivity analysis
The EV public charging infrastructures have been widely sup-
ported by the central state or local governments [71]. This support
can be subsidies and lower taxes, benefiting all the parties
involved: from the users perspective, it allows them to have access
to a wider EV charging network; from the promotor perspective, it
allows lower investment and costs [72]; from the state perspective,
the support of cleaner initiatives helps to achieve international
commitments regarding GHG emissions, promoting societal well-
being and economic development.
Examples of such policies are in: Portugal, where the govern-
ment supported in 2010 the installation of a network of 1300
charging stations spread across the country [71] and, more recently,
its update and extension [73]; USA, where it is available a federal
tax credit of 30% of the charging infrastructures costs [74]; Japan,
where governmental incentives cover two-thirds of the total cost of
the charging stations (it allowed them to be the first country having

4 Energy sales to EVs

3 Energy sales to train station

facilities
2
1.11 Energy daily/monthly fixed fees
0.87
1
0.13 Grid support
[M€]

0
Grid purchases
-1
O&M&Management
-2
Investment
-3
Profit
-4

-5
A0 B0 C0

Fig. 10. Revenues and expenses for the main revenues models. Positive values are revenues and negative are expenses.
 R. Figueiredo et al. / Energy 140 (2017) 1182e1197 1195

Table 11
Financial indicators.

Scenarios
A0 A1 A2 A3 A4 A5 B0 C0
NPV (M€) 0.87 0.96 1.85 2.08 2.10 0.48 0.13 1.11
discounted Payback (years) 16.1 15.5 7.4 7.4 7.3 14.3 24.3 14.5
IRR 10.08% 10.38% 17.36% 18.44% 18.53% 11.30% 7.04% 11.02%
discounted ROI (times) 1.07 1.13 1.59 1.53 1.75 1.25 0.76 1.17

16 24% market conditions it would exceed 14 years. Financially, the use of

14 21% energy storage is not yet viable, due to its high cost, but may be in
the future, as it becomes cheaper. The provision of ancillary services
Payback time [years]

12 18%
to the grid, such as frequency regulation, may improve significantly
10 15%
the project profitability. In addition, a financial incentive of 40% of

IRR [%]
8 12%
6 9%
4 6%
2 3%
0 0%
0% 10% 20% 30% 40% 50%
Subsidies [% of investment]
Payback time IRR
Fig. 11. Sensitivity analysis on the subsidies of a solar parking lot, showing the payback
time and IRR.
more EV charging points than gas stations) [75].
These examples reflect the importance of government support
in EV adoption. This support can include subsidies to park operators
to install charging stations [31].
Since the more appealing scenarios may not be feasible in the
present, since they assume as electric all cars parking at the EVSPL,
an analysis to what extent a public co-financing of the initial in-
vestment could impact on the EVSPL project attractiveness under
current conditions was performed. And so, taking the scenario A5,
the most realistic one, a sensitivity analysis was done to a public
support from 0% to 50% of the total investment. Fig. 11 shows the
results for the IRR and payback time.
It can be seen that the IRR rises from 11.3%, which corresponds
to the scenario A5 with no subsidies, to around 22.5%, for the case of
50% subsidised. As for the payback time, it decreases from 14 years
to 6 years. If one considers that the top encouraging payback time
for an investor is of 7 years [31], a subsidy of 40% of the initial in-
vestment is required to make the EVSPL financially appealing. In
this case, the IRR would be of 19%.
4. Conclusions
This work presented the potential and viability of car parking
lots to solar-charge electric vehicles, using a park-and-ride facility
as a case study. Different operating and business models were
designed and tested.
A good correlation between the photovoltaic generation and the
parking lot's occupation is crucial to directly use photovoltaics for
EV charging; for this park-and-ride, that correlation is of 52% on
working days. A controlled charging plan is found to create less
dependence on the grid as opposed to store solar electricity and
using it later through an uncontrolled plan.
The economic feasibility aimed to ascertain the most suitable
business model for an investor. Considering a discount rate of 7%, all
the scenarios show viability, although some by a small margin. The
payback time is found to be as low as 7 years, although for current
the initial investment is required to reduce the payback time to 7
years, in line with what a private investor expects. The allocation of
such subsidies would thus have a large impact on the profitability
of this business, promoting the deployment of electric vehicles
solar parking lots.
This study intends to be demonstrative of the energy and
financial potential of a solar parking lot to charge EVs. It presents a
general methodology extendible to other cases. For further studies,
it would be interesting to consider the uncertainty of the PV gen-
eration (as here it is assumed that the PV generation profile for the
day is known) and exploit other business models, such as including
revenues coming from commercial advertising on the EVs charging
stations. Also, testing this methodology with parking facilities other
than park-and-ride, and therefore with potentially very different
occupation profiles, would be of interest. Additionally, the case
study taken here is a typical park-and-ride facility. Different results
are expected for projects of other dimensions, which should
improve with the project size due to economies of scale, and in
other regions, where the solar resource and financial conditions
may be different.
Aknowledgements
The authors gratefully acknowledge the financial support by the
MIT Portugal Program on Sustainable Energy Systems (PD/BD/
114174/2016), the Portuguese Science and Technology Foundation,
grant PD/BD/114174/2016, project UID/GEO/50019/2013, and proj-
ect SusCity MITP-TB/CS/0026/2013.
References
[1] International Energy Agency. World energy outlook 2016-executive summary.
2016. Paris.
[2] International Energy Agency. Technology roadmap solar photovoltaic energy.
2014. Paris.
[3] International Energy Agency. IEA sankey diagram. 2014. https://www.iea.org/
Sankey/#?c¼World&s¼Final consumption. [Accessed 28 June 2017].
[4] Coffman M, Bernstein P, Wee S. Integrating electric vehicles and residential solar
PV. Transp Policy 2017;53:30e8. https://doi.org/10.1016/j.tranpol.2016.08.008.
[5] IEA. Snapshot of global PV markets. 2014. p. 2015.
[6] Poullikkas A. Sustainable options for electric vehicle technologies. Renew Sustain
Energy Rev 2015;41:1277e87. https://doi.org/10.1016/j.rser.2014.09.016.
[7] Fattori F, Anglani N, Muliere G. Combining photovoltaic energy with electric
vehicles, smart charging and vehicle-to-grid. Sol Energy 2014;110:438e51.
https://doi.org/10.1016/j.solener.2014.09.034.
[8] Honarmand M, Zakariazadeh A, Jadid S. Integrated scheduling of renewable
generation and electric vehicles parking lot in a smart microgrid. Energy
Convers Manag 2014;86:745e55.
[9] Delmas MA, Kahn ME, Locke SL. The private and social consequences of
purchasing an electric vehicle and solar panels: evidence from California. Res
Econ 2017;71:225e35. https://doi.org/10.1016/j.rie.2016.12.002.
[10] Nunes P, Figueiredo R, Brito MC. The use of parking lots to solar-charge
electric vehicles. Renew Sustain Energy Rev 2016;66:679e93. https://
doi.org/10.1016/j.rser.2016.08.015.
1196 R. Figueiredo et al. / Energy 140 (2017) 1182e1197
[11] Bhatti AR, Salam Z, Aziz MJBA, Yee KP, Ashique RH. Electric vehicles charging
using photovoltaic: status and technological review. Renew Sustain Energy
Rev 2016;54:34e47. https://doi.org/10.1016/j.rser.2015.09.091.
[12] Ashique RH, Salam Z. Bin Abdul Aziz MJ, Bhatti AR. Integrated photovoltaic-
grid dc fast charging system for electric vehicle: a review of the architec-
ture and control. Renew Sustain Energy Rev 2017;69:1243e57. https://
doi.org/10.1016/j.rser.2016.11.245.
[13] Xydas E, Marmaras C, Cipcigan LM. A multi-agent based scheduling algorithm
for adaptive electric vehicles charging. Appl Energy 2016;177:354e65.
https://doi.org/10.1016/j.apenergy.2016.05.034.
[14] Forrest KE, Tarroja B, Zhang L, Shaffer B, Samuelsen S. Charging a renewable
future: the impact of electric vehicle charging intelligence on energy storage
requirements to meet renewable portfolio standards. J Power Sources
2016;336:63e74. https://doi.org/10.1016/j.jpowsour.2016.10.048.
[15] Liu N, Zou F, Wang L, Wang C, Chen Z, Chen Q. Online energy management of
PV-assisted charging station under time-of-use pricing. Electr Power Syst Res
2016;137:76e85. https://doi.org/10.1016/j.epsr.2016.04.002.
[16] Nezamoddini N, Wang Y. Risk management and participation planning of
electric vehicles in smart grids for demand response. Energy 2016;116:
836e50. https://doi.org/10.1016/j.energy.2016.10.002.
[17] García-Trivi P, Torreglosa JP, Fernandez-Ramírez LM, Jurado F. Control and
operation of power sources in a medium-voltage direct- current microgrid for
an electric vehicle fast charging station with a photovoltaic and a battery
energy storage system. Energy 2016;115:38e48. https://doi.org/10.1016/
j.energy.2016.08.099.
[18] Chandra Mouli GR, Bauer P, Zeman M. System design for a solar powered
electric vehicle charging station for workplaces. Appl Energy 2016;168:
434e43. https://doi.org/10.1016/j.apenergy.2016.01.110.
[19] Fathabadi H. Novel solar powered electric vehicle charging station with the
capability of vehicle-to-grid. Sol Energy 2017;142:136e43. https://doi.org/
10.1016/j.solener.2016.11.037.
[20] Fathabadi H. Novel grid-connected solar/wind powered electric vehicle
charging station with vehicle-to-grid technology. Energy 2017;132:1e11.
https://doi.org/10.1016/j.energy.2017.04.161.
[21] Karan E, Asadi S, Ntaimo L. A stochastic optimization approach to reduce
greenhouse gas emissions from buildings and transportation. Energy
2016;106:367e77. https://doi.org/10.1016/j.energy.2016.03.076.
[22] Hafez O, Bhattacharya K. Optimal design of electric vehicle charging stations
considering various energy resources. Renew Energy 2017;107:576e89.
https://doi.org/10.1016/j.renene.2017.01.066.
[23] Falahati S, Taher SA, Shahidehpour M. A new smart charging method for EVs
for frequency control of smart grid. Int J Electr Power Energy Syst 2016;83:
458e69. https://doi.org/10.1016/j.ijepes.2016.04.039.
[24] Usman M, Knapen L, Yasar AUH, Vanrompay Y, Bellemans T, Janssens D, et al.
A coordinated framework for optimized charging of EV fleet in smart grid.
Procedia Comput Sci 2016;94:332e9. https://doi.org/10.1016/
j.procs.2016.08.049.
[25] Kavousi-Fard A, Khodaei A. Efficient integration of plug-in electric vehicles via
reconfigurable microgrids. Energy 2016;111:653e63. https://doi.org/10.1016/
j.energy.2016.06.018.
[26] Bhatti AR, Salam Z, Ashique RH. Electric vehicle charging using photovoltaic
based microgrid for remote islands. Energy Procedia 2016;103:213e8. https://
doi.org/10.1016/j.egypro.2016.11.275.
[27] Kamankesh H, Agelidis VG, Kavousi-Fard A. Optimal scheduling of renewable
micro-grids considering plug-in hybrid electric vehicle charging demand.
Energy 2016;100:285e97. https://doi.org/10.1016/j.energy.2016.01.063.
[28] Tabatabaee S, Mortazavi SS, Niknam T. Stochastic scheduling of local distri-
bution systems considering high penetration of plug-in electric vehicles and
renewable energy sources. Energy 2017;121:480e90. https://doi.org/10.1016/
j.energy.2016.12.115.
[29] Rahbari O, Vafaeipour M, Omar N, Rosen MA, Hegazy O, Timmermans J-M,
et al. An optimal versatile control approach for plug-in electric vehicles to
integrate renewable energy sources and smart grids. Energy 2017;134:
1053e67. https://doi.org/10.1016/j.energy.2017.06.007.
[30] Mozafar MR, Moradi MH, Amini MH. A simultaneous approach for optimal
allocation of renewable energy sources and electric vehicle charging stations
in smart grids based on improved GA-PSO algorithm. Sustain Cities Soc
2017;32:627e37. https://doi.org/10.1016/j.scs.2017.05.007.
[31] Robinson J, Brase G, Griswold W, Jackson C, Erickson L. Business models for
solar powered charging stations to develop infrastructure for electric vehicles.
Sustainability 2014;6:7358e87. https://doi.org/10.3390/su6107358.
[32] LNEG - National laboratory of energy and geology n.d. http://www.lneg.pt/
iedt/unidades/25/(Accessed 8 December 2014).
[33] Marion W, Urban K. User's manual for TMY2sdtypical meteorological years.
Gold Co Natl Renew Energy Lab 1995:55.
[34] Gilman PSAM. Photovoltaic model technical reference SAM photovoltaic
model technical reference. 2015.
[35] Noorian AM, Moradi I, Kamali GA. Evaluation of 12 models to estimate hourly
diffuse irradiation on inclined surfaces. Renew Energy 2008;33:1406e12.
https://doi.org/10.1016/j.renene.2007.06.027.
[36] Santamouris M. Environmental design of urban buildings - an integrated
approach. London: Earthscan; 2006.
[37] Benkaciali S, Gairaa K. Comparative study of two models to estimate solar
radiation on an inclined surface the radiometric station was installed on the
roof of Solar Radiation Laboratory of. Rev Des Energies Renouvelables
2012;15:219e28.
[38] Azimuth Angle | PVEducation. PVEducation n.d. http://pveducation.org/
pvcdrom/properties-of-sunlight/azimuth-angle (Accessed 16 July 2015).
[39] Howey DA, Martinez-Botas RF, Cussons B, Lytton L. Comparative measure-
ments of the energy consumption of 51 electric, hybrid and internal com-
bustion engine vehicles. Transp Res Part D 2011;16:459e64. https://doi.org/
10.1016/j.trd.2011.04.001.
[40] Fesharaki VJ, Dehghani M, Fesharaki JJ, Tavasoli H. The effect of temperature
on photovoltaic Cell efficiency. Emerg. Trends energy conserv. Tehran: ETEC;
2011. p. 20e1.
[41] Pasaoglu G, Fiorello D, Martino A, Zani L, Zubaryeva A, Thiel C. Travel patterns
and the potential use of electric cars e results from a direct survey in six
European countries *. Technol Forecast Soc Chang 2014;87:51e9. https://
doi.org/10.1016/j.techfore.2013.10.018.
[42] INE. Estatísticas dos transportes. 2009. p. 2010.
[43] Solar World. Sunmodule plus SW 260-290 mono - data sheet. 2015.
[44] Mourin ~ o H. Metodos de avaliaç~ ao de projectos de investimento Me todos de
avaliaç~ao de projectos de investimento. 2012.
[45] Hirth L. The market value of variable renewables. The effect of solar wind
power variability on their relative price. Energy Econ 2013;38:218e36.
https://doi.org/10.1016/j.eneco.2013.02.004.
[46] Photovoltaic geographical information system (PVGIS) - European Commis-
sion. JRC’s Inst Energy Transp n.d. http://re.jrc.ec.europa.eu/pvgis/info/faq.htm
(Accessed 19 September 2016).
[47] Ca^mara Municipal de Almada. Almada num Minuto n.d. http://www.m-
almada.pt/xportal/xmain?
xpid¼cmav2&xpgid¼genericPage&genericContentPage_
qry¼BOUI¼5771022&actualmenu¼5770956 (Accessed 13 June 2016).
[48] Fertagus. QUEM SOMOS n.d. https://www.fertagus.pt/pt/testee-quem-somos/
quem-somos (Accessed 13 June 2016).
[49] Regulamento de exploraça ~o de Parques. Fertagus n.d. https://www.fertagus.
pt/pt/regulamento-de-exploracao-de-parques (Accessed 27 August 2015).
[50] Skoczek A, Sample T, Dunlop ED. The results of performance measurements of
field-aged crystalline silicon photovoltaic modules. Prog Photovoltaics Res
Appl 2009;17:227e40. https://doi.org/10.1002/pip.
[51] Sunny tripower inverters. Wind sun powering futur n.d. http://www.
windandsun.co.uk/products/Inverters/SMA-Inverters/Sunny-Tripower-
Inverters#10066 (Accessed 22 July 2015).
[52] Mayrbaurl RM, Camo S. Guidelines for inspection and strength evaluation of
suspension bridge parallel-wire cables. 2004. Washington.
[53] Tesla PowerWall Specs. Tesla power mot. 2016. https://www.teslamotors.
com/en_EU/powerwall?redirect¼no. [Accessed 1 January 2016].
[54] Mayr Florian. How to determine meaningful, comparable costs of energy
storage |. Energy Storage News; 2016. https://www.energy-storage.news/
blogs/how-to-determine-meaningful-comparable-cost-of-energy-storage.
[Accessed 10 November 2016].
[55] Mayer JN. Current and future cost of photovoltaics. 2015.
[56] Bronski P, Creyts J, Guccione L, Madrazo M, Mandel J, Rader B, et al. The
Economics of Grid Defection: When and where distributed solar generation
plus storage competes with traditional utility service. Boulder. 2014. https://
doi.org/10.1016/j.enpol.2015.03.005.
[57] German J. Electric Vehicles : performance, cost, penetration. CCAP MAIN-Latin
Am. Reg. Dialogue. The International Council on Clean Transportation - ICCT;
2014.
[58] MCC16-Kiosk with satellites. Magnum Cap n.d. http://magnumcap.com/index.
php/contacts?id¼24 (Accessed 31 July 2015).
[59] Electric charging in Europe. New motion n.d. https://www.thenewmotion.
com/en/products/electric-charging-in-europe/(Accessed 20 August 2016).
[60] Tarifas Transito rias de Venda a Clientes Finais em Portugal Continental em.
2017. p. 2017.
[61] Entidade Reguladora dos Serviços Energe ticos (ERSE). Tarifas e preços para a
energia eletrica e outros serviços em 2015 e par^ ametros para o período de
regulaça~o 2015-2017. 2014.
[62] Inflation forecast. OECD data. 2016. https://data.oecd.org/price/inflation-
forecast.htm#indicator-chart. [Accessed 2 September 2016].
[63] Sketchup Skelion solar design plugin renewable energy. n.d. http://skelion.
com/(Accessed 17 September 2016).
[64] Ban~ os R, Manzano-Agugliaro F, Montoya FG, Gil C, Alcayde A, Go mez J.
Optimization methods applied to renewable and sustainable energy: a review
2011;15. https://doi.org/10.1016/j.rser.2010.12.008.
[65] Diouf B, Pode R. Potential of lithium-ion batteries in renewable energy. Renew
Energy 2015;76:375e80. https://doi.org/10.1016/j.renene.2014.11.058.
[66] Qamar Raza Muhammad, Nadarajah Mithulananthan, Ekanayake Chandima.
On recent advances in PV output power forecast. Sol Energy 2016;136:
125e44. https://doi.org/10.1016/j.solener.2016.06.073.
[67] Nunes P, Farias T, Brito MC. Day charging electric vehicles with excess solar
electricity for a sustainable energy system. Energy 2015;80:263e74. https://
doi.org/10.1016/j.energy.2014.11.069.
[68] Chukwu UC, Mahajan SM. V2G parking lot with pv rooftop for capacity
enhancement of a distribution system. IEEE Trans Sustain Energy 2014;5:
119e27. https://doi.org/10.1109/TSTE.2013.2274601.
[69] Preço da Banda de Regulaça ~o Secunda ria. Sist Inf Mercados Energ 2014. http://
www.mercado.ren.pt/PT/Electr/InfoMercado/InfOp/BandaSecundaria/
Paginas/Preco_BandaSec.aspx. [Accessed 29 August 2016].
[70] Tulpule PJ, Marano V, Yurkovich S, Rizzoni G. Economic and environmental
 R. Figueiredo et al. / Energy 140 (2017) 1182e1197 1197

impacts of a PV powered workplace parking garage charging station. Appl
Energy 2013;108:323e32.
[71] REDE MOBI.E e Definiç~ ao e Pontos de Carregamento n.d. https://www.mobie.
pt/a-rede-mobi.e (Accessed 28 January 2015).
[72] Yang T, Long R, Li W, Rehman S. Innovative application of the publiceprivate
partnership model to the electric vehicle charging infrastructure in China.
Sustainability 2016;8:738. https://doi.org/10.3390/su8080738.
[73] República Portuguesa. Comunicado do Conselho de Ministros do Ambiente de
8 de junho de 2016 n.d. http://www.portugal.gov.pt/pt/o-governo/cm/
comunicados/20160608-cm-com.aspx (Accessed 16 August 2016).
[74] Tax Credit for Electric Vehicle Stations - ChargePoint. Charg point. 2016.
http://www.chargepoint.com/station-incentives/. [Accessed 29 August 2016].
[75] McCurry Justin. Japan now has more electric car charge points than petrol
stations. World news; 2016. The Guardian, https://www.theguardian.com/
world/2016/may/10/japan-electric-car-charge-points-petrol-stations.
[Accessed 29 August 2016].