See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326085027

EVLibSim: A Tool for the Simulation of Electric Vehicles’ Charging Stations

Using the EVLib Library

Article  in  Simulation Modelling Practice and Theory · June 2018

DOI: 10.1016/j.simpat.2018.06.007

CITATIONS READS

10 833

4 authors:

Emmanouil S. Rigas Sotiris Karapostolakis

Aristotle University of Thessaloniki Aristotle University of Thessaloniki
25 PUBLICATIONS   312 CITATIONS    4 PUBLICATIONS   20 CITATIONS   

SEE PROFILE SEE PROFILE

Nick Bassiliades Sarvapali D. Ramchurn

Aristotle University of Thessaloniki University of Southampton
271 PUBLICATIONS   3,425 CITATIONS    180 PUBLICATIONS   6,448 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Catalogue of System and Software Properties View project

DEiXTo View project

All content following this page was uploaded by Nick Bassiliades on 07 November 2018.

The user has requested enhancement of the downloaded file.

 EVLibSim: A Tool for the Simulation of Electric
Vehicles’ Charging Stations Using the EVLib Library
Emmanouil S. Rigasa , Sotiris Karapostolakisa , Nick Bassiliadesa , Sarvapali
D. Ramchurnb
a
Aristotle University of Thessaloniki, Thessaloniki , 54124, Thessaloniki, Greece
{erigas, skarapost, nbassili}@csd.auth.gr
b
Electronics and Computer Science, University of Southampton, Southampton,
SO17 1BJ, UK,
sdr1@soton.ac.uk

Abstract
Electric Vehicles (EVs) are considered an efficient alternative to internal
combustion engined ones, aiming to reduce global CO2 emissions. In the
last years, EVs are entering the market in an increasing pace. In contrast
to conventional cars, EVs have a more complicated recharging procedure.
Thus, the development of tools for the efficient simulation of the charging
of large numbers of EVs is critical. In this vein, EVLibSim is a tool for
the simulation of EV activities at a charging station level. EVLibSim unifies
EVLib’s primary functions such as the charging and dis-charging of batteries,
battery swapping, as well as parking/inductive charging. EVLib is a Java
library that provides a simple, yet efficient framework for the management
of a number of Electric Vehicle (EV) activities, at a charging station level,
within a Smart Grid. EVLibSim provides a great variety of configuration
options such as the types and number of chargers, the available energy, the
waiting queues, etc. Furthermore, through plots and overview dashboards
each user can supervise the operation of the tool in real time. Both EVLib’s
and EVLibSim’s efficiency and scalability have been tested in realistic sce-
narios, while the correctness and safety of the code have been verified using
state of the art tools. Finally, the user experience of the EVLibSim has been
evaluated and improved through a detailed user evaluation.
Keywords: Electric Vehicle, Java Library, Simulation,
Charging, Dis-charging, Battery Swap, Parking, Smart Grid

Preprint submitted to Simulation Modelling Practice and Theory June 29, 2018
1. Introduction
Electric vehicles (EVs) are entering our daily lives fast. It is estimated
that more than 1 million EVs have already been deployed worldwide, while
the target set by the International Energy Agency is 30% market share for
fully electric vehicles by 2030 [18]. However, in order to ensure that the large-
scale deployment of EVs results in a significant reduction of CO2 emissions, it
is important that they are charged using energy from renewable sources (e.g.,
wind, solar). Crucially, given the intermittency of these sources, mechanisms
as part of a Smart Grid [14], need to be developed to ensure the smooth
integration of such sources in our energy systems. EVs could potentially
help by storing energy when there is a surplus, and feed this energy back to
the grid when there is demand for it [20].
Indeed, the ability of EVs to store energy while being used for trans-
portation [23] represents an enormous potential to make energy systems more
efficient. On the one hand, given that vehicles drive only for a small percent-
age of the day, and considering the fact that EVs are equipped with large
batteries, they could be used as storage devices when parked (i.e., as part
of Vehicle-to-Grid (V2G) schemes [20]), and thus dramatically increase the
storage capacity of the network. On the other hand, given that large numbers
of EVs need to charge on a daily basis, (e.g., 40% of EV owners in Califor-
nia travel daily further than the range of their fully charged battery [24]) if
EVs charge as and when needed, they may overload the network. For this
reason, new scheduling mechanisms are required to be able to manage the
charging of EVs –Grid-to-Vehicle (G2V)– while considering the constraints
of the distribution networks within which EVs need to charge [29]. Similarly,
battery swap schemes which can be used instead of simple battery charge
and can minimize waiting times at the charging stations [27], [28], must also
be efficiently designed and smoothly integrated to the grid.
Advanced Artificial Intelligence (AI) techniques (e.g., mathematical pro-
gramming based optimization, electronic markets and coalition formation)
have been proven to be efficient in solving EV-related problems, and a large
number of such solutions already exists [30]. However, one of the key findings
of [30] is that interoperability between various technologies and techniques
is missing, and it is vital for successful large scale EV deployments. Thus,
there is a need EV technologies to be able to work seamlessly and efficiently
together. Different types of chargers should be able to work with all EV
models, and data exchanged between entities (EVs, charging points, network

2
operators) should have an understandable by all format and meaning.
EVLib library tries to solve such problems, as it provides simple, yet
efficient, tools for the management of the basic EV activities (i.e., charging,
dis-charging, battery swap) in a charging station level. On top of this, the
simulation framework called EVLibSim applies all major features of EVLib.
EVLibSim unifies the Smart Grid benefits along with electric vehicles needs
and characteristics, giving the users the ability to simulate many operations
of electric vehicles such as charging or discharging, alongside the use of some
advanced Smart Grid techniques, such as pricing policies.
Another observation of [30] is that most of the research works in the
Electric Vehicle and the Smart Grid sector are developed from scratch by
the researchers. This leads to works that are difficult to be re-developed and
cross-evaluated by other researchers in the community. The EVLib library
aims to be used as a base for the development of sophisticated EV-related
algorithms. In this way, the core structure of the code will be the same
and only the logic will have to be programmed separately. Thus, the inter-
operability between various works will increase. Moreover, the EVLibSim
tool can be useful in two main ways: 1) The tool can receive as input an
EVs’ charging or discharging plan, execute the plan and then provide the
results in csv files. In this way, the output of many different algorithms
can be given to the tool and the results can then be easily evaluated and
compared to each other. This is crucial in the direction of comparing dif-
ferent algorithms, which as discussed previously, is missing in the literature.
2) Due to the wide range of researchers dealing with EV-related problems
(e.g., computer scientists, electrical engineers, civil engineers, mechanical en-
gineers) the EVLibSim tool gives them the opportunity to construct and
compare different settings without the need to write code and develop their
own implementations. This can engage more people in doing research in this
interdisciplinary sector. Apart from the scientific sector, EVLibSim can help
real world deployments of EV charging stations being simulated and evalu-
ated with different configurations. This can have significant economic gains
for the charging stations owners.
The rest of the paper is organized as follows: Section 2 presents a review of
a number of vehicle-related simulation frameworks. Section 3.1 presents the
mathematical formulation of the setting and Section 3.2 presents an overview
of EVLib as well as a detailed description of its main classes and functions.
Moreover, Section 4 introduces the EVLibSim simulation tool and describes
its functionality. Section 5 provides details about the code security and safety

3
testing of EVLib and EVLibSim and the user evaluation of EVLibSim. At
the same time, Section 6 presents a detailed usage example of EVLibSim.
Finally, Section 7 concludes and Section 8 provides ideas for future work.

2. Related Work
After a thorough study of the existing literature we realized that a num-
ber of frameworks concerning the simulation of different aspects of vehi-
cle travelling and modelling exist. We distinguish two major categories: 1)
Transportation-related simulation tools and 2) EV-related simulation tools
and we present representative works of both categories in what follows.

2.1. Transportation-related simulation tools

SUMO [21] is an open traffic simulation suite. SUMO allows modelling
of intermodal traffic systems including road vehicles, public transport and
pedestrians. It contains a number of supporting tools which handle tasks
such as route finding, visualization, network import and emission calculation.
SUMO can be enhanced with custom models and provides various APIs to
remotely control the simulation.
Moreover, MatSim [17] is a framework for large-scale, agent-based trans-
port simulations. Each agent has a transport demand represented by a chain
of activities it has to perform during one day at different times and locations.
The decisions on how to travel between places to perform these activities are
planned before the mobility simulation. It is an open source project devel-
oped in Java. A user can design a detailed network including millions of
vehicles. MatSim provides the ability for demand-modelling in parallel with
a controller to start, stop, re-plan or iteratively run the simulations. MatSim
has been used in a number of occasions such as in [25] for transport en-
ergy demand modelling, or in [36] to increase the transferability of transport
demand models.
In a slightly different domain, VIVUS (Virtual Intelligent Vehicle Ur-
ban Simulator) [15] simulates vehicles and sensors, taking into consideration
their physical structure, while applying artificial intelligence algorithms such
as platoon solutions [13] and obstacle avoidance techniques [16]. The urban
vehicle simulator is based on two main engines: one for physics simulation
and one for 3D immersion in a topological environment. These engines al-
low a precise simulation of vehicle dynamics and a wide range of physical
sensors. This simulator has been successfully used as a prototyping tool for

4
sensor designing and positioning. Experiments presented in this paper show
simulation results close to reality. The differences with the real conditions are
essentially due to the precision of the world topology and the model chosen
for the wheel/road contact.
Finally, battery autonomy is the major pursuit of [32]. They propose
an on-board assembling energy system, particularly a photovoltaic system
(PVS). A PVS includes a solar array, batteries, regulator and load. A number
of variables (e.g., radiance and ambient temperature) are essential inputs to
estimate the power production of a PVS. A mobile system is in continuous
movement, therefore an additional analysis in changing conditions, due to
the variability of the solar exposure, had to be done. They present ways of
measuring solar radiation, which define the power production of the system.
The simulation tool was implemented using Matlab Simulink. Given that
the path followed by a mobile PVS is known, the Simulink tool calculates
both the energy produced by PVS and the energy consumption.

2.2. EV-related simulation tools

AVL CRUISE [1] is a tool to analyze fuel efficiency, vehicle emissions,
energy transmission modeling, controller design, and vehicle electrification.
In the case of EV design, the most important part is the controller design to
determine the optimal operating point for minimizing energy consumption. It
facilitates this task in a MATLAB/Simulink environment. In a similar vein,
EMCAS [4] (Electricity Market Complex Adaptive System) is an agent-based
simulation tool to analyze complex power systems. EMCAS can simulate
electric vehicle technologies, renewable energy generation subsystems and
their integration, as well as thermal power generation. It is used to analyze
the external technical, operational and economical impacts on the electricity
sector where the simulation is carried out on an hourly basis over a user-
specified timescale.
Moreover, Caspoc [2] simulates electric power generation, conversion, dis-
tribution, power electronics and mechatronics systems. CASPOC is also use-
ful for the design and analysis of electric vehicles. It can model the entire
drive cycle (e.g., gearbox, brakes, wheel slip, etc.) of an EV, the electric drive
control, and thermal models of electric machines. It can also help to design
and analyze grid connections from renewable energy sources (e.g. solar and
wind) and the distribution grid.
FASTSim [5] is a vehicle simulation tool to analyze and design conven-
tional vehicles and EVs. FASTSim evaluates the impact of technology up-

5
grades on EV performance, battery lifetime and overall efficiency. For exam-
ple, it can calculate the most cost-effective battery size for EVs. It deals with
the vehicle powertrain, regenerative braking, energy management strategy,
cost estimation, and battery life.
The tools presented so far are not designed solely for EV modelling and
simulation, but this is a subset of their features. However, tools related
exclusively to EVs exist. For example, [10] presents a framework for mod-
elling EVs’ operations and services. The authors developed a simulation
platform that is based on the SUMO (vehicular traffic simulator) [21] and
OMNET++ (network simulator) [33] and can simulate system and communi-
cation protocols for better usage of city-wide mobile services. They evaluated
the operations of EVs on a large-scale scenario, by analyzing the effective-
ness of the charging reservation process and the impact on the Smart Grid.
The second contribution is the Mobile Application Zoo sandbox. Electric
Mobility-related mobile applications can be seamlessly combined within the
simulation framework in order to be tested on virtual environments before
their distribution in real situations.
In addition, [12] study the use of EVs as a distributed energy resource
(DER). They propose a fleet aggregator model using a stochastic formulation
of DER-CAM [3], an optimization tool used to address DER investment and
scheduling problems. This is used to assess the impact of EV interconnections
on optimal DER solutions considering uncertainty in EV driving schedules.
Results indicate that EVs can have a significant impact on DER investments.
Furthermore, results suggest that uncertainty in driving schedules has little
impact on total energy costs. In a similar vein, [35] propose a framework for
microgrid planning with EV charging demand to find the most economical
configuration through which to maximally utilize renewable generation. The
algorithm uses a renewable generation-following EV charging scheme and
HOMER [6]. Through simulations, they show that the microgrid constructed
by the proposed algorithm reduces the investment cost and CO2 emission.
Moreover, V2G-Sim [7] (Vehicle-to-Grid Simulator) addresses vehicle-to-
grid integration by incorporating driver-specific sub-modules, charging sys-
tems and vehicle powertrains. The major applications are: automotive (e.g.,
battery, vehicle powertrain), electricity grid (future grid, charging and dis-
charging, renewable generation, electricity market etc.), policy and business
(e.g., regulatory, economic), and end users (e.g., V2G-Sim mobile app). It
addresses the potential barriers and uncertainties in vehicle-to-grid integra-
tion through a systematic quantitative method. It predicts the real-world

6
scenario (drive system, trip distance and time) of an EV and analyzes the
impact or opportunity of these vehicles to the electricity grid. To analyze im-
pact on the grid, it models the charging and discharging control approach and
driving behavior of an individual EV. V2G-Sim designs the vehicle-to-grid
energy exchange system and analyzes it in second-by-second time steps.
In addition, [11] demonstrates step-by-step mathematical calculations and
simulation modeling of EVs. The simulation model is equipped with recuper-
ation capability of EV. Its performance has been investigated by simulation
results for various vehicle velocity inputs and summarized in terms of vehicle
range and Battery State of Charge (SOC). Furthermore, vehicle range and
SOC are evaluated and summarized with variation of input parameters at
the end of this paper.
In a similar vein, [8] proposes a modelling of electric and parallel-hybrid
electric vehicle using Matlab/Simulink environment and the authors focus
on different aspects of the vehicle such as engine power, type and size of the
battery or weight and try to observe how changes can affect the performance
and the distance travelled. The model was simulated in order to obtain the
electric vehicle’s autonomy. Through the use of a Geographic Information
System together with a mathematic algorithm based on genetic algorithms
the planning of charging stations was obtained aiming to minimize cost and
maximize service for the customers.
In a slightly different domain, where the aim is the optimal placement of
charging stations, [34] presents a simulation model that uses mathematical
optimization techniques to determine the optimal location of electric vehicle
charging facilities aiming to maximize their use by privately owned EVs.
Applying this model to the central-Ohio region, the authors demonstrate that
a combination of level-one and level-two chargers is preferable to level-two
chargers only. They further explore interactions between the optimization
criterion used and the budget available and they show that although the
optimal location is sensitive to the specific optimization criterion considered,
overall service levels are less sensitive to the optimization strategy.
Finally, in [9] the proposed model of an electric vehicle charging station
is suitable for the fast DC charging of multiple electric vehicles. The station
consists of a single grid-connected inverter with a DC bus where the electric
vehicles are connected. The control of the individual electric vehicle charging
processes is decentralized, while a separate central control deals with the
power transfer from the AC grid to the DC bus. The electric power exchange
does not rely on communication links between the station and vehicles, and

7
a smooth transition to vehicle-to-grid mode is also possible. Simulations are
performed in Matlab/Simulink to illustrate the behavior of the station.
Table 1 summarizes the tools presented in this subsection and compares
them based on a set of criteria. After studying and comparing these tools
and considering the findings in [22], we argue that a simulation tool regarding
EV activities at a charging station level that considers all the aspects of both
the EV and its interconnection to the smart grid is missing. Thus, we argue
that EVLibSim is an innovative tool and greatly important for the research
community, as well as for the industry.
In the next section, the problem formulation as well as the EVLib library
that supports the creation, configuration and handling of such a charging
station are presented.

8
 Cit. Vehicle Charging V2G Electricity Electricity Renewable Commercial/ Technology
mod- station pricing schedul- energy free used
elling/ mod- ing
analysis elling/
analysis
EVLibSim Modelling Yes Yes Yes Partially Yes Free Java
[1] AVL Yes No Yes No No No Commercial MATLAB/
CRUISE Simulink
[2] Yes No Yes Yes No Yes Commercial No info
CASPOC
[4] EM- No No Yes Yes Yes Yes Commercial No info
CAS
[5] FAST- Yes No Yes No No No Free Excel and
9

Sim Python
[10] Yes No No Yes Yes Yes Free OMNET++
and SUMO
[12] Yes No Yes Yes Yes Yes Free DER-CAM
[35] Modelling No No Yes Yes Yes Free (re- HOMER
stricted)
[7] V2G- Modelling No Yes Yes Yes Yes Free (re- No info
Sim stricted)
[34] Modelling Yes No No Yes No Free No info
[8] Yes Yes No Yes No No Free MATLAB/
Simulink
[9] Yes Yes Yes No Yes No Free MATLAB/
Simulink
[11] Yes No No No Yes No Free MATLAB/
Simulink
Table 1: Comparison of EV modelling / simulation tools
3. Setting Formulation and the EVLib Library
In this section, we describe the mathematical formulation of the problem
and later we describe how this is translated into the EVLib library.

3.1. Setting formulation

In our setting, we study the modeling and simulation of an EV charging
station. The station holds charging infrastructure, as well as electric energy
and the EVs announce their requests for charging or discharging over time.
Later, the station decides on whether it will service each EV based on energy
and charger availability. A large number of parameters (e.g., the types and
numbers of chargers, as well as the energy sources and the available energy
from each source) can be configured based on the scenario that the user needs
to simulate. The time t ∈ T ⊆ R is continuous and is measured in seconds
or minutes and the energy e ∈ R is measured in Wh or kWh.
We denote a charging station i ∈ I ⊆ N. Each charging station has a
number of slow chargers si ∈ Si ⊆ N, fast chargers fi ∈ Fi ⊆ N, battery
exchange handlers ehi ∈ EHi ⊆ N, inductive chargers indi ∈ IN Di ⊆ N,
dischargers di ∈ Di ⊆ N and parking slots prki ∈ P RKi ⊆ N. In more
detail, there are two main categories of EV chargers, namely the slow and
the fast chargers. The slow chargers can fully charge an EV’s battery in
6-8 hours, while the fast chargers in around 1 hour. However, fast charging
increases the load on the grid and reduces the life-time of the battery. Apart
from the traditional chargers that use a cable, inductive charging also exists.
In this case, the EV simply stops on top of an inductive charger and the
energy transfer takes place contactless. In addition, battery exchange is an
efficient alternative to battery charging, where instead of charging the EV’s
battery, a fully charged one is swapped into the EV. This technique leads
to very fast recharging, but it is costly and raises the problem of battery
ownership, as once an EV unloads the battery it may never get the same
battery again. In order to support this function, the station has a number
of batteries bi ∈ Bi ⊆ N. Each battery has a maximum capacity emax bi and a
current charge level ecur
bi . Finally, the dischargers support the discharging of
an EV’s battery when there is excess energy in it.
Now, the slow chargers have a charging rate srate i ∈ R, the fast chargers
rate rate
fi ∈ R (i.e., si < fi ), the inductive chargers indrate
rate
i ∈ R, the dis-
rate
chargers have a discharging rate di ∈ R and the battery exchange handlers

10
have an exchange duration ehdur i ∈ R. In addition, we denote binary vari-
ables avlbsi ,t ∈ {0, 1}, avlbfi ,t ∈ {0, 1}, avlbehi ,t ∈ {0, 1}, avlbindi ,t ∈ {0, 1},
avlbdi ,t ∈ {0, 1}, avlbprki ,t ∈ {0, 1} that capture whether the equivalent charg-
ers, exchange handlers, inductive chargers, dischargers, or parking slots are
available at time point t.
Moreover, the charging station has available energy e ∈ E for EV charging
from a set of one or more of the following energy sources: Solar energy esol i,t ∈
geo
R, wind energy ewind i,t ∈ R, wave energy ewave
i,t ∈ R, geothermal energy e i,t ∈
hydro
R, hydroelectric energy ei,t ∈ R, and energy from other non-renewable
nonR
sources ei,t ∈ R. Note that available energy may vary over time based
on EV consumption. Finally, a price for each charging and discharging type
and per energy unit is defined: pi,t,s ∈ R for slow charging, pi,t,f ∈ R for fast
charging, pi,t,eh ∈ R for battery exchange, pi,t,ind ∈ R for inductive charging,
and pi,t,d ∈ R for discharging.
Apart from the charging station, a number of EVs j ∈ J ⊆ N also exist.
Each EV has a battery capacity ecap j ∈ R, a remaining amount of energy
rem dem
ej,t ∈ N, an energy demand ej,t ∈ N and a request for a particular type
of charging or discharging, as well as a maximum waiting time tmax i for it
to be serviced. The acceptance of the charging or discharging of an EV
depends on the availability of energy and chargers or dischargers. Once an
EV makes a request to the station, an equivalent event evnt ∈ EV N T ⊆
N is created (i.e., evnti,j,s , evnti,j,f , evnti,j,d , evnti,j,eh , evnti,j,ind , evnti,j,p ).
For the handling of waiting events, a number of equivalent lists levnt ∈
LEV N T exist (i.e., levnti,s , levnti,f , levnti,d , levnti,eh , levnti,ind , levnti,p ).
For each event in any list, a waiting time twait levnt can be calculated based on
the number of events in the list and the execution time of each one. The
lists operate in a FIFO (First In First Out) manner. However, the user has
the ability to manually override the order. Finally, each event has a current
state evntstatus
i,j,d ∈ {arrived, accepted, rejected}.
In the next section, we describe the EVLib library.

3.2. The EVLib library

EVLib1 is implemented in the Java programming language and its main
goal is to simulate the charging, discharging, battery swap and inductive
charging functions in the level of a single charging station i [19]. There

1
http://intelligence.csd.auth.gr/files/systems/EVLib/evlib.jar

11
are four main functions, as well as a number of secondary ones, while each
function is executed in 2 phases, namely the pre-processing and the execution
phase. Figure 1 provides the library’s detailed class diagram.

Figure 1: EVLib Class Diagram

3.2.1. EVLib’s main functions

The main functions are the following:
Charging (G2V technology): There are 2 types of charging depending
on the charging time, namely the fast and the slow charging. Equiv-
alently, slow si and fast fi chargers exist. It is possible to define the
rate for each type of charging (i.e., srate
i and firate ). The pre-processing
phase requires a request for an empty charger and available energy. If
the pre-processing phase is successful, the execution phase can begin.

12
 Battery swap: The pre-processing phase searches for a battery with enough
range ecur dem
bi ≥ ej,t and a battery swapping handler ehi to be available
in the charging station. If those requirements are met, the execution
function can be called and the battery is swapped into the EV.

Discharge (V2G technology): Similarly to a charging event, the pre-

processing phase makes a request for an empty discharger di . If this
phase is successful then the execution can start.

Parking/Inductive charging (G2V technology): The parking function

can be merged with the inductive charging. Each parking slot is able
to support inductive charging. The user either demands to park, or to
park and charge inductively. The pre-processing method searches for
an available parking slot prki and an inductive charger indi . If there
is one, then looks for available energy in case of inductive charging. A
successful pre-processing phase leads to calling the execution method.
Algorithm 1 describes the steps for the handling of a slow charging event.
Initially the charging station receives the request for charging from an EV
j. Once the request has been received, the equivalent charging event evnti,j,s
is created and its state is set to arrived (line 1). After the event has been
created, the station checks whether the type of charger needed for the re-
quested charging (i.e., slow charger in this case) exists. If no such charger
exists, then the event is rejected (lines 2-3). If the necessary infrastructure
exists, the charging time tchrg
i,j is calculated based on the energy demand edemj,t
and the charging rate srate
i of the charger. If the charging time is longer
than the maximum time the EV can stay at the station, the event is rejected
(lines 4-8). Then the availability of charger and energy to start the charg-
ing immediately is checked (lines 9-12). If both charger and energy exist,
then the price to pay for the charging pj is calculated and the event state
is set to accepted (lines 13-17). If no charger and/or no energy is available
for immediate charging, then the scheduling of the event for later charging
is considered. Initially, an available charger for time tchrg
i and for the time
max max
window tcur + tj (tcur is the current time point and tj is the latest time
point EV j can be at the charging station) is searched. If a charger available
for at least the number of time points needed to complete the charging is
found, then a search for the availability of energy takes place. If enough
energy is found, then the charging can take place, the event is added to a
waiting list levnti,s and the event state is set to accepted (lines 18-30). If no

13
charger or no energy can be found, then the event is rejected (lines 31-34).
Finally, the state of the event is returned.
Regarding the complexity of Algorithm 1, the most time consuming ac-
tivities are the search of available chargers and energy. Note that the search
is a simple access to a number of cells in the equivalent tables. Given the
flow of actions as described in Algorithm 1, the case where the searches take
place more times is when an EV is scheduled for later charging. In this case,
lines 1, 2, 4-12, 20-30 are activated. The cost is calculated as follows:
Cost = 1+1+1+1+1+tchrg chrg max
i,j +ti,j ×|E|+1+tj ×2+1+1+tmax
j ×|E|+1+1+1 (1)

Note that the term tmax

j × 2 refers to the lines 21 and 22. Given that the lines
with cost equal to 1 do not play a significant role in the overall complexity,
Equation 1 becomes:
Cost = (1 + |E|) × tchrg max
i,j + (2 + |E|) × tj (2)

All the parameters in Equation 2 do not scale and are upper bounded and for
this reason the worst case complexity of Algorithm 1 is O(1). The algorithms
and equivalent complexity analysis for the rest of the event types are similar
and are skipped. Note that in the case of discharging, no need for energy
exists and in the case of battery swapping, instead of available energy, an
available battery must be found.

3.2.2. EVLib’s secondary functions

The library also supports a number of secondary functions: The cre-
ation of multiple charging stations, as well as the creation and integration
of a charger si or fi , discharger di , battery swapper ehi , parking slot prki
or inductive charger indi in every station. Additional operations are the
recharging of batteries bi which are later to be swapped into EVs, as well as
the ability to add new batteries to the storage for the seamless operation of
the battery exchange process to be achieved. Finally, the cost of the charging
pi,t,s or pi,t,f , discharging pi,t,d and battery swapping pi,t,eh can be calculated
based on various cost types (e.g. energy cost) defined by the user.
During the creation of the charging station i, 4 waiting lists are created.
A list for the charging events which want slow charging levnti,s , a list for
the charging events which want fast charging levnti,f , a list for the discharg-
ing events levnti,d , and a list for the vehicles waiting for battery exchange
levnti,eh . A list can be managed in 2 ways: Either automatically by the li-
brary using the default settings, or programmatically by the user. Moreover,

14
Algorithm 1 Charging event handling algorithm
Require: ecap rem dem max
j , ej,t , ej,t , tj , esol wind wave geo hydro
i,t , ei,t , ei,t , ei,t , ei,t , enonR
i,t , avlbsi ,t .
state
1: Create charging event evnti,j,s and set event state evnti,j,d to arrived
2: if (|Si | = ∅) then {No slow chargers exist in the charging station }
3: Set event evnti,j,s state evntstate
i,j,d to rejected

4: {Calculate charging time}

5: tchrg
i,j = edem
j,t /si
rate

6: {If the charging time is larger than the maximum time the EV can be at
the station, the event is rejected }
chrg
7: if (tcur + ti,j > tcur + tmax
j ) then
8: Set event evnti,j,s state evntstate
i,j,d to rejected
chrg
9: {Search available charger
P in avlbsi ,t for the duration ti,j of the event}
10: Calculate sumchrg = t:t≥tcur ,t≤tcur +tchrg (avlbsi ,t )
i,j
chrg
11: {Search available energy in all sources based on demand edem j,t for ti,j }
Calculate sumenrg = t:t≥tcur ,t≤tcur +tchrg (esol
P
12:
i,j
i,t ). Repeat for all sources
13: {If charger and energy exist, start immediate charging}
14: if (sumchrg ≥ tchrg
i AND sumenrg ≥ edem j,t ) then
15: {Calculate price to pay}
16: pj = pi,t,s × edem
j,t
17: Set event evnti,j,s state evntstatei,j,d to accepted
18: {Schedule for later charging}
19: else
chrg
20: {Search charger in avlbsi P ,t for the duration ti,j for tcur + tmax
j }
21: Calculate sumchrg = t:t≥tcur ,t≤tcur +tmax (avlbsi ,t )
j
22: Add the time points the charger is available to set τchrgavlb
23: {If available charger exists}
24: if (sumchrg ≥ tchrgi ) then
25: {Search energy for the time P the charger is available}
26: Calculate sumenrg = t∈tchrgavlb (esol i,t ). Repeat for all sources
27: {If available energy exists}
28: if (sumenrg ≥ edem j,t ) then
29: Set event evnti,j,s state evntstate i,j,d to accepted
30: Add event to the waiting list levnti,s
31: else
32: Set event evnti,j,s state evntstate i,j,d to rejected

33: else
34: Set event evnti,j,s state evntstate
i,j,d to rejected

35: Return evntstate 15

i,j,d
each time an event is added to any list, an expected maximum waiting time
twait
levnt is calculated, and once an event is executed these times are updated.
Another useful feature is the plethora of energy sources for the supply
of energy to the charging station, which are available through the use of an
energy warehouse. The library includes 6 different relevant classes, where
geo
each class describes a different energy source (i.e., esol wind wave
i,t , ei,t , ei,t , ei,t ,
ehydro
i,t , enonR
i,t ). The amount of available energy in the warehouse is updated
in a regular basis. For this update there are 2 options: In the first one, the
user is responsible for calling the energy storage update method. At each
invocation only one energy unit is withdrawn from the total energy amount
of each energy source, inserting it to the station’s energy warehouse. In
the second option, the user must initially enable the automatic update of
the energy storage and then define a time space. Then the energy update
storage method is executed automatically at specified time moments.
The waiting lists give the ability to the charging station to execute some
events, following a priority order based on the time of arrival at the station.
An event is inserted to the waiting list during the pre-processing phase. As
described earlier, the mechanism is the following: The method looks for an
empty charger/discharger/exchange handler. In case of not finding one, the
method calculates the time the event should wait. The next step is the
comparison of this time with the maximum time tmax i the event can actually
wait. If the calculated waiting time is less than the time the vehicle can
wait, the event enters the waiting list and the status is set to ”wait”. In
any other case, the event’s status is set to ”nonExecutable”, meaning that
the execution is not possible. A method returning the waiting time twait evnt of
each event is provided. Additionally, for each kind of event a method for the
remaining time tremevnt is provided to constantly inform the users of the state
of each event.
Pricing policy for the charging function is a typical Smart Grid feature.
Two different options are provided. One with fixed time space for each price
and one with varying time space. By time space we actually refer to the
time period that a specific price is valid. A method for the definition of the
current prices for slow pi,t,s or fast pi,t,f charging, discharging pi,t,d , battery
swapping pi,t,eh and inductive charging pi,t,ind is included.
ChargingEvent, DisChargingEvent and ParkingEvent classes store all the
created events. Based on this, the library is equipped with a method which
returns all this information along with some descriptive statistics regarding

16
the operation of the charging station all gathered in a text file.
Synchronizing and scheduling a group of charging events is another fea-
ture of EVLib. The user can give, through a text file, a plan of events for
execution committing a predefined number of chargers. Each charging is car-
ried out in parts or in whole. There is a detailed description of this feature
in Section 4.
In the following subsections the main classes of the EVLib library are
presented.

3.2.3. ChargingStation
ChargingStation class describes a charging station and is a central class
in EVLib. Technically speaking, this class is the corner stone of the library
since everything is attached to it, or is strongly connected with it as shown in
Figure 1. The class has several methods, the most important of them being:
1. updateQueue(ChargingEvent event) inserts a charging event to
the corresponding waiting list.
2. getChargers() returns an array with all the chargers of the charging
station.
3. getWaitingTime(String kind) returns the waiting time for a specific
type of event. The values of the argument are: ”fast” for fast charging,
”slow” for slow charging, ”exchange” for the battery exchange opera-
tion, ”discharging” for the discharging function.
4. assignCharger(ChargingEvent event) assigns a charger to the given
event.
5. insertEnergySource(EnergySource source) inserts one of the al-
ready predefined energy sources to the charging station.

3.2.4. Charger
This class represents a charger of a charging station. Charger class’ main
methods are:
1. startCharger() starts the execution of the linked charging event.
2. handleQueueEvents() executes the first charging event in the wait-
ing list.
Note that the DisCharger, ExchangeHandler and ParkingSlot classes operate
similarly to the Charger class.

17
3.2.5. ChargingEvent
ChargingEvent class is responsible for the representation of a charging
event. The main methods of this class are:
1. preProcessing() method which executes the preliminary process be-
fore the charging.
2. execution() is responsible for starting the execution of the event.

3.2.6. EnergySource
The library contains a set of renewable and non-renewable energy sources
that can provide energy to the charging station. Specifically, there are 5
sub-classes each of them representing a renewable source and 1 sub-class
for the energy from non-renewable sources. All sub-classes have the same
interface and inherit this abstract class. Each of the 6 classes that inherit
EnergySource class contain an energyAmount field, which stores the energy
that is going to be inserted after each update.
Moreover, EnergySource class has 2 methods:
1. popAmount() returns the first energy amount to be given in the
charging station.
2. insertAmount(double amount) inserts an energy amount to the
energy source list of amounts.
PricingPolicy Pricing policy class provides two different types. Primarily,
a pricing policy with a fixed time space for each price and the second one
with changing time spaces.
Some of the most noticeable methods are:
1. setTimeSpace(long timeSpace) sets the time space when it is fixed.
2. getDurationOfPolicy() returns the duration of the pricing policy.
3. setSpecificPrice(int index, double price) modifies a price.
4. getSpecificPrice(int index) returns the price in the ”index” posi-
tion, or 0 if ”index” is greater than the number of the policy’s prices.

3.2.7. WaitingList
WaitingList class is responsible for keeping the events that have not yet
been assigned to any charger, discharger or exchange handler during the pre-
processing phase and ensuring the right order of them. Technically speaking,
waiting list operates similar to the FIFO Queue data structure. Events are

18
ordered based on their arrival time. It contains a variety of methods, pro-
viding in this way the tools for the handling of a set of events.
WaitingList class includes the following functions:
1. add(T object) inserts an object in the list.
2. get(int index) returns the object in the ”index” position.
3. getSize() returns the number of objects in the list.
4. moveFirst() removes the first object and returns it back.
5. takeFirst() returns the first object of the waiting list.
In the next section, the EVLibSim simulation environment, which makes
use of all EVLib’s classes and methods, is presented.

4. EVLibSim
EVLibSim2 simulation tool, aims to unify EVLib’s functions and present
them in a simple, yet efficient and comprehensive user interface. EVLibSim
provides the ability to configure a charging station depending on the user’s
demands, along with the configuration of its environment taking advantage
of all EVLib’s features. The user can initialize a charging, dis-charging,
battery swapping or parking event at any time. The simulation tool operates
constantly waiting for a new event to be scheduled for execution or an action
to be processed. Each time a new EV arrives, Algorithm 1 (or equivalent
algorithms for the other event types) is called. Given that the complexity
of this algorithm is O(1), the complexity for the simulation tool is O(n)
where n is the number of EVs that have already arrived to the station. In
the meantime, all the proper tools for the management of energy-related
activities are provided. Another aspect of the tool is that the user can keep
logs of all executed events utilizing the information for further processing
and data analysis. Finally, EVLibSim uses real-time plots for monitoring all
the running processes in the station.

4.1. Main View

Through the start screen of the tool the user can have access to all the ba-
sic functionality (see also Figure 2). The creation of a new charging station,
addition of a new pricing policy, construction of any kind of event and man-
agement of energy activities are the main features in the central panel. In

2
http://intelligence.csd.auth.gr/files/systems/EVLib/evlibsim.jar

19
Figures 3, 4, 5 we can see the provided choices of the 3 buttons of start screen.
The user can also navigate using the menu bar, where all the operations are
also included.

Figure 2: Start screen of EVLibSim

Figure 3: Options of ”Sta- Figure 4: Options of ”New Figure 5: Options of ”En-

tion“ button Event“ button ergy“ button

The “Station” selection provides 3 options: Creating a new charging sta-

tion, attaching a new pricing policy to a running station and modifying the
running station. The “New charging station” button leads to a new window

20
with the fields shown in Figure 6. The top fields concern the number of
chargers, dischargers, battery handlers, parking slots and the types of energy
sources linked to the station. The middle and lower ones refer to the basic
configuration of the charging station, such as the rates of the various func-
tions (charging rate, discharging rate, inductive charging rate), the prices of
energy units for each basic operation, the handling of the waiting lists, the
mode of energy storage update and the duration of the battery swapping
function. The modification of a station is similar to the creation but the
fields related to the infrastructure of the station are deactivated. The user
cannot change the number of chargers, dischargers, battery swapping han-
dlers or parking slots because some of them may be occupied. The insertion
and deletion of them can be performed through separate menu functions.

Figure 6: Creation of a new charging station

The “New Event” option leads to a new window for the creation of a new
event. The window has 4 basic fields: 1) name of the driver, 2) brand of
the vehicle, 3) capacity of the battery and 4) remaining amount of energy.
A user can select the vehicle’s brand through a choice box providing only
electric vehicles as choices. The capacity always needs to be greater than the
remaining amount. Next, depending on the kind of event, there are fields for

21
the proper construction of the event. The user should always be aware of
the remaining energy in the station for charging and available batteries for
battery swapping. Otherwise, the events will not be executed.
The selection of the “Energy” button opens a new window to manage
the activities related to the sources and the available energy in the charging
station (see also Figure 7). Each field corresponds to a specific kind of avail-
able energy source. The energy source types that are available for a specific
station are marked with an asterisk. An amount of energy can be inserted
only to those sources that were selected during the creation of the station.
When adding an amount, it is recorded to the storage of each source, but
not yet to the station’s warehouse. The energy storage update method is
responsible for this. The energy storage update is made either directly or it
can be scheduled for a later time. Selecting “Direct”, the energy amounts are
added immediately to the station’s warehouse. Selecting “Scheduled”, the
transfer will be implemented in one of the next scheduled updates. At each
update only one energy unit from each source is transferred to the station.
Moreover, the option to add or remove an energy source from the charging
station is provided.
The sorting of the energy sources is another option included in the “En-
ergy” selection. It provides the ability to select which energy source will be
used first to provide energy during a charging. By selecting this option, a
window, similar to the one for energy addition, opens. The user needs to de-
fine a number for each field that signifies the order. The range of the values
is 1 to the number of attached energy sources.
Beyond the start screen’s operations, on the left part of the screen two
panels exist (see Figure 2). The “Prices” panel informs the users regarding
the current prices per energy unit for the charging, discharging, the induc-
tive charging and the price of battery swapping operation. The “Wait” panel
presents the waiting time for each operation based on the state of the equiv-
alent lists.
Moreover, there is an output box located at the bottom of the screen.
It shows messages about the completion of an event which typically consists
of the event type and ID, followed by the name of the driver, the brand of
the vehicle, the station and the “OK” keyword. A plan execution provides
some extra information, such as a message for the completion of a specific
charger’s plan. Furthermore, the end of all the chargers’ plans signifies the
completion of the plan, leading to a message display. Notice that EVLib
library does not allow the simultaneous application of more than one “Plan

22
 Figure 7: Addition of energy

execution” functions.

4.2. Menu Bar Functionality

The menu bar provides the complete functionality of the simulation tool.
Beyond the categories with the features, the tool has several options to con-
trol the simulation environment, through the use of selection boxes as shown
in Figure 2 in the upper right corner. The first one is for controlling the run-
ning charging station. It contains all the created stations, giving the ability
to select one of them. By saying running we mean that every action in the
EVLibSim concerns that station. The next selection box refers to the time
metering unit. The user can choose between two units, the second and the
minute. Finally, the last one is for the energy unit. The available energy
units are the Wh and the kWh. The metering unit can be updated at any
time and the values in all related fields and operations are updated auto-
matically. Additional functions provided from the menu bar are described in
what follows.
The user can save all the previously created stations with their amounts
of energy and attached batteries, as well as all the events’ history with their
progress and load all these whenever he/she wants using “Save” and “Save
as...” operations. Additionally, he/she can start over the operation of the

23
tool using the “New” feature, at any time. The “Events report” operation
is responsible for exporting all the created events since the beginning of a
session. “Station report” gives an analytical report of a selected charging
station, which contains information for the state of the charging station.
Both types of reports are text files.
The user can also enable or disable the suggestions box showing up in
every event’s creation window from the “Suggestions” selection. Suggestions
box refers to some information being given by the tool, trying to consult the
users to select the best charging station. The recommendations are catego-
rized based on energy, prices and waiting times. The main idea is to sort the
energy, prices, and waiting times of each station. Recommendations include
only the first five stations.
The supervision of the different actions taking place in the station is
provided through an overview window with 4 plots as shown in Figure 8.
The upper plots show the energy partition and the infrastructure of the
station respectively, while the lower charts show the number of the running
events and the state of the waiting lists respectively.

Figure 8: Overview plots of a charging station

24
 The ability to insert or delete a new charger, discharger, battery handler,
a new parking slot or a new battery for the efficient operation of the battery
swapping function is also provided through “Station” menu. For the chargers,
only the supported type of charging needs to be defined. For a battery
creation, three fields need to be completed, namely the battery capacity, the
remaining energy in the battery and the upper number of battery’s chargings.
The battery swapping function demands the charging of the attached to the
station batteries.
Pricing policy is a solution for balancing the excessive demand and work-
load across a set of charging stations [29]. By selecting this function a pop-up
window is shown, giving two choices. The “Fixed space” selection denotes
that each price will be valid for a fixed time space. In contrast, the “Chang-
ing space” choice means that every price’s time space will be different. In
both cases, after the user’s initial choice, a window with two fields is dis-
played. For the fixed time space option, the user must define the time space
and then a set of prices. For the changing time space option, the user should
set the time spaces in the first field and the prices in the second one. There
is no restriction regarding the number of prices the user sets. After the end
of pricing policy’s duration, the charging price returns to the default value.
In Section 6.1 there is a detailed example of the implementation of this op-
eration.
The “Plan execution” selection is for the simulation of a set of events the
scheduling plan of which has been calculated externally of the EVLibSim.
Events use a predefined set of chargers, that will be shown as occupied until
the end of execution of this set of events.

Figure 9: Plan execution template - example

Plan execution is defined through a text file as the one displayed in Figure
9. At each line, symbols or numbers are separated with comma. Initially,

25
we define the charging events and then the chargers’ plans. Every charging
is executed using only one charger. The symbol “ch” denotes a charging,
followed by the energy to be received. At the end of the charging events, the
plan of every charger follows. Symbol “de” signifies the start of a charger
plan. Inside the line of a charger plan there are also some other symbols.
Symbol “ch” means that a part of a charging starts. The number shows the
exact charging from the initial list with the charging events. Symbol “int”
implies an interruption and its duration follows after the comma. In our case,
the plans that tested by our simulation tool were calculated from an already
existing scheduling algorithm [31].

Figure 10: Summary of chargings

4.3. Toolbox
At the right hand side of EVLibSim a toolbar is located and one of its
choices is the ability to provide information about the energy in the station.
By selecting it, a window is shown containing all the connected, to the charg-
ing station, energy sources and their remaining energy units. Moreover, the

26
toolbar includes 5 event selections. Each of them corresponds to a different
type of event, namely fast charging events, slow charging events, discharging,
battery swaps, and parking events. Each selection displays the state of all
events of the same type in a table-like form (see Figure 10), where every
line represents a different event. The event’s displayed information is the
received energy, the waiting time of the vehicle, the cost, the station they
were serviced, the time of their execution and the remaining time for the
completion.

5. User Evaluation and Testing

In this section we present the evaluation of the EVLib library and EVLib-
Sim simulation environment.

5.1. Code security and safety testing

EVLib and EVLibSim were checked for errors regarding the code devel-
opment, errors about the view of the code, or security holes that may exist
using state of the art tools. In so doing, FindBugs3 and CheckStyle4 were
utilized. FinbBugs is a tool which carries out static testing on Java code.
Alongside, CheckStyle reports all cases which do not follow patterns of proper
programming, related to the view and logical development of code. In ad-
dition to these programs, a number of unit tests, related to all classes were
developed using JUnit 5. All the errors that were discovered were corrected.

5.2. Evaluation of the EVLibSim

The evaluation phase of the EVLibSim included a questionnaire5 with 8
questions of varying difficulty that was assigned to 21 persons from the IT
sector. Each question required a task to be executed using one of EVLibSim’s
features. The user had to rate from 1 (very difficult)-to-5 (very easy) the
difficulty of the task. Comments about the corresponding task could be freely
added by the users. In the end, the user could also add general comments
about the tool as well as suggestions for changes and / or new functionality.
The questionnaire had two parts. The first part included 4 questions
about the background of every respondent. The questions referred to the

3
http://findbugs.sourceforge.net/
4
http://checkstyle.sourceforge.net/
5
https://goo.gl/forms/9GYAeK0hK5sI0qZU2

27
 14.3% 4.8% 9.5% 4.8%
24-30 BSc
18-24 MSc
47.5% 30-40 47.6% PhD
40+ 38.1% Other
33.4%

Figure 11: Age and education of participants

4.8% 4.8%
9.4% Student
2-5 4.8%
14.3% Engineer
0-1
Researcher
61.9% 6-10 42.9%
IT-Consultant
10+
38.1% Other
19%

Figure 12: Years of experience and profession of participants

age, education, profession and experience of the participants. The answers

are presented in Figures 11 and 12.

28
 Group Group Group
Task Overall
1 2 3
Create a new charging station. 4 3.5 3.5 3.7
Add an amount of energy to
3.8 4.2 3.7 3.9
an existing energy source.
Update/modify an existing
4 3.7 3.4 3.7
charging station.
Create an (charging) event. 3.8 3.8 3.7 3.8
Using the overview buttons
on the right-hand side of the
tool, evaluate the available
4.2 4 4 4
information and how useful
they are in understanding the
current state of the station.
Evaluate your understanding
regarding the waiting queue
3.4 4 3.2 4
of events and how it relates to
the waiting time panel.
Construct a new pricing
policy using both options.
Evaluate your understanding
2.7 2.8 3 2.8
of the difference between fixed
time space and changing time
space.
Export the information
related to all events to a csv
(comma separated value) file, 4.5 4.5 4 4.4
using the button inside the
File menu.
Table 2: Questionnaire’s tasks with the average answers.

The second part included the tasks/questions about the simulation tool.
Table 2 presents the questionnaire’s tasks and the average grade for each
task. In conducting this evaluation, we followed the Delphi method [26] and
we separated our set of participants to 3 groups. After the end of each group’s
evaluation, the tool was improved based on their comments and the updated
version was provided to the next group.

29
 The first group completed the questionnaire having the initial version.
This version did not show tips, the rates of each (charging or discharging)
function at the charging station creation panel had to be manually defined
by the user, events’ history export was not available and the user had to
define the time he/she could wait until serviced. Among the most crucial
suggestions from the first set of participants were the addition of color code in
events’ history panels, tips for better understanding of the terms and time not
to be measured in milliseconds but in a more comprehensible unit i.e., seconds
or minutes. The proposed features were taken into account by making the
corresponding improvements to the tool. The field for waiting time at every
event’s creation panel was removed, a choice box for time measuring units
having as options seconds or minutes was added in the menu and tips were
added for each field in the tool that needs information to be entered by the
user. The rest of the requested features were also implemented.
The second group of participants evaluated the improved version of EVLib-
Sim. The users of this phase outlined two big problems at the modification
of the charging station and at the sorting of energy sources. The suggested
improvements were the addition of the feature to restart the application, the
addition of a home button and the addition of a choice box for energy mea-
surement units. Furthermore, users had an inconvenience to comprehend the
construction of a pricing policy. All the user suggestions were implemented
along with the addition of an explanatory text for the pricing policy function.
The third set of respondents evaluated the most improved edition of
EVLibSim. Most of the comments had to do with the pricing policy feature,
the weakness to understand the terms at the charging station creation panel
(i.e., charging rate) and the range in which the required values vary. Some
other participants proposed the addition of an example of how to construct
a pricing policy, the vehicle brand field to be substituted with a choice box
containing only electric vehicles brands, more informative error messages and
a better positioning in the screen for the selection of changing the running
charging station.
To sum up, the most important reported issues were the following:
1. Better explanation of the values required in the charging station cre-
ation panel.
2. Provision of a more explanatory example to facilitate the pricing policy
feature.
3. Update of all fields in the modification of a charging station panel.

30
 4. Difficulty in understanding waiting time field in association with left
panel that shows waiting times.
5. Exporting of events’ log.
6. Difficulty to comprehend the energy storage update options.
7. Provision of a manual/video with examples.
8. Insufficient display time of tips.
9. Better explanatory messages in case of fields completion errors.
10. Moving the choice box for charging station in an easier to read position
in the screen.
In the latest revision of the tool, we positioned the choice box for selecting
charging station in the menu bar and we provided an example for constructing
a pricing policy at the top of each creation panel. In addition, we altered
the brand field with a choice box giving as options only electric vehicles
brands. At the charging station creation panel, the rate fields (charging rate,
discharging rate, inductive charging rate) were replaced by choice boxes with
values that are valid in real stations.
In table 2 the average values of users’ answers are presented. Users’
ratings did not increase among the 3 groups as long as new features and
corrections were inserted. The lowest average rating was for the creation of
a pricing policy followed by the creation of a new charging station and its
modification. The main reason for this is the requirement to comprehend
and provide values for a lot of new terms. In contrast, exporting reports has
the highest average rating.
The user evaluation results are considered satisfactory. The participants
were interested in testing an innovative application and to learn new terms
concerning electric vehicles. Positive comments include the wide range of
functions, appropriate presentation and the ability to export reports on the
operation of EVLibSim. In contrast, most users complained about the lack of
documentation for some novel terms (i.e., charging rate or remaining energy
in battery) and had difficulties to construct a pricing policy. Through their
comments we had the opportunity to enhance EVLibSim. We believe that the
version that came out of the suggestions is more compact and user-friendly.

6. Using EVLibSim
In this section, examples related to the use and performance of EVLibSim
are provided.

31
6.1. Example of Use
The objective of this section is the presentation of EVLibSim’s basic
functions, in order to provide guidelines for the proper use of the simulation
tool and the required execution order of some activities.

6.1.1. Charging Station

The integration of a charging station is the most significant function in
EVLibSim. To create a new charging station, the user should select “Sta-
tion” → “New charging station” from the main panel or the menu bar. In
Figure 13 an example configuration of a charging station is presented. The
fees refer to the cost of the Wh or kWh for each operation, except for the
battery swapping function where the cost is for the whole process. The rates
(charging, discharging and inductive) define the energy that is transferred
per hour, and determine the duration of the corresponding function. The
linked energy sources are wind, wave and solar.

Figure 13: Charging station example

6.1.2. Energy Addition

The addition of energy is the second required action. Energy is necessary
for charging operations. At each update, an amount of energy is selected

32
from each associated source with the charging station.
To add energy, we select “Energy” → “Add energy” from either the main
screen or the menu bar. Figure 14 presents an example of this selection,
where 30 kWh of Solar energy, 30 kWh of Wind and 20 kWh of Wave energy
become available.

Figure 14: Energy addition example

6.1.3. Charging Event

Charging is implemented either by selecting “New event” → “New charg-
ing” from the main screen or from the menu bar. Figure 15 shows the given
values for a charging event. The vehicle’s brand is Tesla Model S, the bat-
tery capacity is 50 kWh, the remaining energy in the battery is 1.5 kWh, the
requested amount of energy is 15 kWh to be charged using fast charging.

Figure 15: Charging event example

6.1.4. Pricing Policy

The integration of a pricing policy function is there to regulate the con-
gestion of a charging station. To use this feature, we select “Station” →
“New charging pricing policy” from either the main screen or the menu bar.
In Figure 16, such a policy with 6 values, each of them lasting for 15 minutes,
is presented.

33
 Figure 16: Pricing policy example

6.2. Usage scenarios

This section presents a comparison between a number of common usage
scenarios, in order to better comprehend EVLibSim’s functionality. The
scenarios focus on the way the charging station handles a number of charging
events when there is not enough energy or available chargers to complete all
events. The state of waiting queues, energy sources and running charging
events are presented through graphs.
The configuration of the charging station, as well as the amount of avail-
able energy is the same for all scenarios. The charging station’s configuration
is the one shown in Figure 13. Regarding the energy, we assume that the
station has stored 50 kWh of solar energy, 50 kWh of wind energy and 20
kWh of wave energy.

Slow Fast Requested Energy Sources

Scenario
Events Events Energy Order
1 2 2 56 Wave, Solar, Wind
2 5 5 96 Solar, Wind, Wave
3 5 5 215 Wind, Solar, Wave
Table 3: Scenarios of EVLibSim

Table 3 shows the first 3 examined scenarios. According to scenario 1, the

first slow charging requests 15 kWh and the second 17 kWh. The first fast
charging event requests 14 kWh and the second one 10 kWh. Scenario 2 has
5 slow charging events and 7 fast charging events. For all charging events the
requested energy is 8 kWh. In scenario 3 the order and the requested energy
measured in kWh are the following: slow(20), fast(20), slow(20), fast(20),
slow(30), fast(20), slow(15), slow(20), fast(25), fast(25). Figures 17 and 18

34
demonstrate the number of charging events in the waiting lists and the re-
maining energy amounts of each energy source, after the insertion of all
charging events at each occasion.

60
3 50
3 44
2 2 40
2
20
1 20 14

0 0 0 0 0 0 0 0
0 0

Scenario1 Scenario2 Scenario3 Scenario1 Scenario2 Scenario3

Slow Fast Solar Wind Wave

Figure 17: Waiting queues state Figure 18: Energy sources state

The first result to be outlined is that the required energy is committed

as soon as the event starts to be executed. In scenario 2, the total requested
energy from all events is 96 kWh. The total charging station’s energy is 120
kWh, meaning that there is enough energy for all charging events. Consid-
ering that every charging event demands 8 kWh and all running chargers
are 7 that means 56 kWh of energy, which is the result of the total energy
minus the sum of remaining wind and wave energy amounts shown in Fig-
ure 18. Therefore, the charging station provided energy only to those events
that charge at this moment. The remaining 40 kWh that is the total re-
quired energy amount of waiting lists’ events will be committed when events’
execution starts.
It is interesting to note that the number of events in the waiting lists
is always 0, or less or equal to the total number of events minus the total
number of chargers in the station. In scenario 1, the number of slow and
fast chargers are greater than the number of slow and fast charging events
respectively. Thus, 0 events are in the waiting lists as shown in Figure 17. In
scenario 2, the number of charging events is 12, while the number of chargers
are 7, meaning that the events in the waiting list are 5. Figure 17 outlines
this difference.

35
 Scenario 3 corresponds to the third occasion where the number of waiting
charging events is smaller than the total number of events minus the total
number of chargers. Observing the created events, it is clear that the charging
station’s energy has been totally consumed after the insertion of the third
fast charging event. Until here, all the slow chargers are occupied, but one
fast charger is available. The next 2 slow charging events enter the waiting
list, while the next 2 fast chargings do not. Although there is an available
fast charger, the events are rejected and declared as non executable. The
reason is the unavailability of energy to charge.

Scenario 4 5 6 7 8 9
Slow chargings 8 17 32 30 25 50
Fast chargings 12 13 15 25 55 50
Non executable
14 27 35 49 60 85
chargings
Energy(kWh)
demand per 20 40 10 20 6 8
charging
Table 4: Scenarios where there is not enough provided energy

Scenario 3 revealed that the remaining energy is a crucial factor that de-
fines charging events’ execution. In order the escalation of this phenomenon
to be examined, 6 extra scenarios were created, while the number of charg-
ers (3 slow chargers and 4 fast chargers) and remaining energy (120 kW) in
the charging station are unchanged. Each scenario has a greater number of
events than the previous one. Table 4 shows the number of slow charging
events, the number of fast charging events, the number of non executable
charging events and the requested energy for every scenario.
Some limitations are introduced. The energy demand of every charging
event at a specific scenario is the same so as the charging time of all events
to be the same. In scenario 4, the input sequence is first a fast charging and
then a slow charging and so on, while in scenario 7 it is the opposite. In
scenario 5, the first action is the addition of 3 slow chargings consuming all
the provided energy. The input sequence is important because the requested
energy is committed only when the event starts charging. In addition, in
these 3 scenarios the product of the total number of chargers per charging
event’s energy requirement is greater than the charging station’s remaining

36
energy which will be consumed before all chargers become occupied. Figure
19 demonstrates the total number of charging events that has been executed
for each scenario. Figure 20 shows waiting lists’ state after the insertion of
all events.

20
Executed charging events

15

12

8
6

3
1
4 5 6 7 8 9
Scenarios

Figure 19: Charging events that have been executed

It is important to note, that the number of events that will be executed

is always the (rounded if not integer) quotient of the total remaining energy
in the charging station to the requested energy per charging event. If the
number is rounded, the last executed event receives just a proportion of its
requested energy. In scenarios 4, 6 and 8, an increase in the number of
charging events to be executed is observed. In these cases, the descending
requested energy per charging event makes the difference.
Figure 20 provides some interesting information. In scenarios 4 and 7, the
requested energy per charging event is the same in both cases. In scenario
4, the input sequence consists of pairs of a fast charging event followed by
a slow charging event. After the insertion of the third slow charging event,
the number of available slow chargers and the remaining energy have been
depleted. All remaining slow charging events enter the waiting list. On the

37
 60
51
50 47

40

30 29
27
22
20
14
11
10
5
0 0 0 0
0

−10
4 5 6 7 8 9

Slow chargings Fast chargings

Figure 20: Waiting lists’ state

other hand, for every subsequent fast charging there is an available charger,
but not enough energy. These vehicles will not be charged and are declared
as non executable. The same occurs for scenario 7. The waiting list for slow
charging events contains the remaining slow charging events, while the fast
charging waiting list is empty.
To sum up, assuming that there is not enough remaining energy in the
charging station two cases exist: In case there is an available charger, then
the event does not enter the respective waiting list and will never be executed.
Instead, in case there is not any available charger, the event is inserted or
remains in the waiting queue. Through these execution scenarios, we aim
to outline the importance of EVLibSim in studying the performance of a
station given different demand schemes and available resources. In this way, a
charging station’s owner can better configure the station based on prediction
of EV demand.

38
7. Conclusions
As mentioned earlier in the text, Electric Vehicles is a continuous growing
sector that is strongly related to the Smart Grid and the extended use of
renewable energy sources. The bidirectional communication between EVs
and the Smart Grid creates great opportunities for the IT and the power
systems sectors. EVLibSim provides a simple, yet efficient interface for the
management of all major EV-related activities such as the charging and dis-
charging of batteries, as well as the battery swapping and parking, while
using energy from renewable sources.
In developing the EVLib and the EVLibSim, a major research challenge
was to design and develop both tools in such a way to be able to cover a
wide range of technologies involved in charging/discharging of EVs. This is
crucial so as to provide with a tool that would be able to cover a wide range
of EV-related applications and be useful for a large number of researchers.
Another consideration was the data that the tool will produce as output, as
well as the format of them. This is important so as to assist researchers in
easily comparing their work with others in the literature.
From a technical point of view, and regarding the EVLib library the chal-
lenge was to cover the larger possible set of EV-related entities and functions,
while at the same time to provide a simple interface for the users. In terms of
the simulation tool EVLibSim, the main challenge was to design a tool that
provides the correct balance between the availability of the necessary features
to construct and configure an EV charging station, while at the same time it
is not over-constraint and can be useful in many different applications. More-
over, given that the EVLibSim uses the classes and functions of the EVLib,
we tried to detach the simulation tool from the basic methodology, so as any
changes to functions of the library to have little affect to the operation of the
simulator. Later, the management of the parallel execution of many events
was another issue to be taken care of. In this case, we used a multi-thread
application in order to achieve a close to reality handling of events.
The main modelling consideration was on deciding on the main actors
of the model. These are the ”Charging Station”, the ”EV”, the ”chargers”,
”dis-chargers”, ”inductive chargers” and ”battery swappers”, the ”Energy”,
and the ”Events”. After selecting the main actors, the attributes of them
should also be selected. For example, a charger has a ”charging rate”, or an
energy source has a number of available ”energy units”. In addition, we had
to consider whether we would include the modelling of the electricity grid.

39
Given that the tool focuses on the charging station level, we decided to model
only the basics that are needed for the operation of the stations such as the
available energy per time point. The detailed modelling and simulation of
the electricity grid is out of the scope of this work.
Another aspect of both tools is the extendability of them. Initially, the
EVLib was developed. As discussed earlier, this library contains several
classes that model all major EV-related actors such as the Charging Station,
the EV, the Energy Source amongst others. Every class contains a large
number of methods that handle different functions. The library itself is writ-
ten in Java, it is opensource and has a detailed description of each class and
function using Java docs. Thus, it can easily be used and extended to fit the
needs of each user. As far as the EVLibSim is concerned, the functionality
of the tool is based on the EVLib. For every option in the UI one or more
EVLib methods are called and return the equivalent results. However, the
UI is detached from the logic of the methods called. For example, the energy
source that is being used to charge and EV can be selected with many dif-
ferent criteria, but the UI will present the result no matter which selection
procedure has been used. The detachment of the simulation methodology
from the basic operation of the simulation is an important contribution as
the simulation can easily be re-used in case the library changes. The UI of the
simulation tool was developed using Java FX which has rich documentation,
and any extension can take place is a relatively straight forward manner.
The tool itself is also opensource.
Finally, the decisions taken, and the knowledge obtained in modelling the
EV charging station can be useful in other domains as well. For example,
insights can be taken for the modelling and simulation of a smart building
operating within a smart grid, or the simulation of electricity-powered un-
manned aerial vehicles (UAVs). Both domains, although at first hand seem
different, have many features in common, as in all cases the central actor of
the system is an electric appliance needing to be powered or charged.
In the next section, ideas for future work are presented.

8. Future work
Concerning future improvements of the tool, congestion management can
be faced up by using algorithms for optimal distribution of electric vehicles
to the available charging points [29]. Also, the logs of charging and discharg-
ing events that EVLibSim keeps can be exploited using statistical packages.

40
EVLibSim would be able to provide better prices for frequent users. More-
over, for every charging station the energy demand and the number of energy
storage updates can be better measured.
Future work will also focus on the integration of more advanced AI tech-
niques, such as electronic markets, agent-based negotiation and coalition
formation to further expand the library’s and EVLibSim’s capabilities [30].
Finally, mechanisms and techniques for managing uncertainty and increase
fault tolerance will also be studied.

9. References
[1] Avl cruise. https://www.avl.com/cruise. Accessed: 05/06/2018.

[2] Caspoc. https://www.integratedsoft.com/Products/caspoc.aspx. Ac-

cessed: 05/06/2018.

[3] Der-cam. https://building-microgrid.lbl.gov/projects/der-cam. Ac-

cessed: 15/06/2018.

[4] Emcas. https://ceeesa.es.anl.gov/projects/emcas.html. Accessed:

05/06/2018.

[5] Future automotive systems technology simulator.

https://www.nrel.gov/transportation/fastsim.html. Accessed:
05/06/2018.

[6] Homer. https://www.homerenergy.com/index.html. Accessed:

15/06/2018.

[7] V2g-sim. http://v2gsim.lbl.gov/home. Accessed: 05/06/2018.

[8] Susana Alegre, Juan V Mı́guez, and José Carpio. Modelling of electric
and parallel-hybrid electric vehicle using matlab/simulink environment
and planning of charging stations through a geographic information sys-
tem and genetic algorithms. Renewable and Sustainable Energy Reviews,
74:1020–1027, 2017.

[9] A. Arancibia and K. Strunz. Modeling of an electric vehicle charging

station for fast dc charging. In 2012 IEEE International Electric Vehicle
Conference, pages 1–6, March 2012.

41
[10] L. Bedogni, L. Bononi, M. Di Felice, A. D’Elia, R. Mock, F. Morandi,
S. Rondelli, T. Salmon Cinotti, and F. Vergari. An integrated simula-
tion framework to model electric vehicle operations and services. IEEE
Transactions on Vehicular Technology, 65(8):5900–5917, Aug 2016.

[11] Aalok Bhatt. Planning and application of electric vehicle with

matlab /simulink
R .
R In Power Electronics, Drives and Energy Systems
(PEDES), 2016 IEEE International Conference on, pages 1–6. IEEE,
2016.

[12] G. Cardoso, M. Stadler, M.C. Bozchalui, R. Sharma, C. Marnay,

A. Barbosa-Pvoa, and P. Ferro. Optimal investment and scheduling of
distributed energy resources with uncertainty in electric vehicle driving
schedules. Energy, 64:17 – 30, 2014.

[13] Jean-Michel Contet, Franck Gechter, Pablo Gruer, and Abderrafiaa

Koukam. Bending Virtual Spring-Damper: A Solution to Improve Lo-
cal Platoon Control, pages 601–610. Springer Berlin Heidelberg, Berlin,
Heidelberg, 2009.

[14] H. Farhangi. The path of the smart grid. Power and Energy Magazine,
IEEE, 8(1):18–28, January 2010.

[15] Franck Gechter, Jean-Michel Contet, Stephane Galland, Olivier Lam-

otte, and Abderafia Koukam. Virtual intelligent vehicle urban simula-
tor: Application to vehicle platoon evaluation. Simulation Modelling
Practice and Theory, 24(Supplement C):103 – 114, 2012.

[16] Franck Gechter, Jean-Michel Contet, Pablo Gruer, and Abderrafiaa

Koukam. Car-driving assistance using organization measurement of re-
active multi-agent system. Procedia Computer Science, 1(1):317 – 325,
2010. ICCS 2010.

[17] Andreas Horni, Kai Nagel, and Kay W Axhausen. The Multi-Agent
Transport Simulation MATSim. Ubiquity Press, United Kingdom, 2016.

[18] IEA. Global ev outlook. Technical report, 2017.

[19] Sotiris Karapostolakis, Emmanouil S. Rigas, Nick Bassiliades, and Sar-

vapali D. Ramchurn. Evlib: A library for the management of the electric
vehicles in the smart grid. In Proceedings of the 9th Hellenic Conference

42
 on Artificial Intelligence, SETN ’16, pages 13:1–13:4, New York, NY,
USA, 2016. ACM.

[20] Willett Kempton and Jasna Tomic. Vehicle-to-grid power fundamen-

tals: Calculating capacity and net revenue. Journal of Power Sources,
144(1):268 – 279, 2005.

[21] Daniel Krajzewicz, Jakob Erdmann, Michael Behrisch, and Laura

Bieker. Recent development and applications of SUMO - Simulation
of Urban MObility. International Journal On Advances in Systems and
Measurements, 5(3&4):128–138, December 2012.

[22] Khizir Mahmud and Graham E Town. A review of computer tools for
modeling electric vehicle energy requirements and their impact on power
distribution networks. Applied Energy, 172:337–359, 2016.

[23] W. J. Mitchel, C. E. Borroni-Bird, and L. D. Burns. Reinventing the

automobile: Personal urban mobility for the 21st century. MIT Press,
2010.

[24] Michael A Nicholas, Gil Tal, and Justin Woodjack. California statewide
charging survey: What do drivers want? 92nd Annual Meeting of the
Transportation Research Board, 2013.

[25] T. Novosel, L. Perkovi, M. Ban, H. Keko, T. Pukec, G. Krajai, and

N. Dui. Agent based modelling and energy planning utilization of mat-
sim for transport energy demand modelling. Energy, 92:466 – 475, 2015.
Sustainable Development of Energy, Water and Environment Systems.

[26] Chitu Okoli and Suzanne Pawlowski. The delphi method as a research
tool: An example, design considerations and applications. 42:15–29, 12
2004.

[27] Emmanouil S. Rigas, Sarvapali D. Ramchurn, and Nick Bassiliades. Al-

gorithms for electric vehicle scheduling in mobility-on-demand schemes.
In Intelligent Transportation Systems (ITSC), 2015 IEEE 18th Interna-
tional Conference on, pages 1339–1344, Sept 2015.

[28] Emmanouil S. Rigas, Sarvapali D. Ramchurn, and Nick Bassiliades.

Algorithms for electric vehicle scheduling in large-scale mobility-on-
demand schemes. Artificial Intelligence, 262:248 – 278, 2018.

43
 [29] Emmanouil S. Rigas, Sarvapali D. Ramchurn, Nick Bassiliades, and
George Koutitas. Congestion management for urban ev charging sys-
tems. In Smart Grid Communications (SmartGridComm), 2013 IEEE
International Conference on, pages 121–126, 2013.

[30] E.S. Rigas, S.D. Ramchurn, and N. Bassiliades. Managing electric vehi-
cles in the smart grid using artificial intelligence: A survey. Intelligent
Transportation Systems, IEEE Transactions on, 16(4):1619–1635, Aug
2015.

[31] Andreas Seitaridis, Emmanouil S Rigas, Nick Bassiliades, and Sarva-

pali D Ramchurn. Towards an agent-based negotiation scheme for
scheduling electric vehicles charging. In Multi-Agent Systems and Agree-
ment Technologies, pages 157–171. Springer, 2015.

[32] Giuseppe Marco Tina and Cristina Ventura. Simulation tool for en-
ergy management of photovoltaic systems in electric vehicles. Energy
Conversion and Management, 78(Supplement C):851 – 861, 2014.

[33] Andrs Varga and Rudolf Hornig. An overview of the omnet++ simula-
tion environment, 01 2008.

[34] Xiaomin Xi, Ramteen Sioshansi, and Vincenzo Marano. Simulation–

optimization model for location of a public electric vehicle charging
infrastructure. Transportation Research Part D: Transport and Envi-
ronment, 22:60–69, 2013.

[35] Sung-Guk Yoon and Seok-Gu Kang. Economic microgrid planning al-
gorithm with electric vehicle charging demands. Energies, 10(10), 2017.

[36] Dominik Ziemke, Kai Nagel, and Chandra Bhat. Integrating cemdap
and matsim to increase the transferability of transport demand models.
Transportation Research Record: Journal of the Transportation Research
Board, (2493):117–125, 2015.

44

View publication stats