Master Thesis: Impact of The Electric Vehicle On The Electric System

Master Thesis
EG201X
Impact of the Electric Vehicle on the Electric System
Mélaine ROUSSELLE
2009
Supervisors at RTE
Vincent Lefieux
Emmanuel Neau
Thibault Prévost
Supervisor at KTH
Lennart Söder
Impact of the Electric Vehicle on the Electric System July-November 2009
Since few years the electric vehicles draw the attention. The battery technology’s continual
improvements and incentives from the authorities guarantee them an assured future with a fast
and considerable development. Some figures are forecasted by 2020: one or two millions
electric vehicles. This could mean a huge increase in electricity consumption. The
consequences on the total energy consumption have already been analyzed, however studies
on the impact on the load curve remain scarce. In this context, this master thesis focuses on
the impact of the electric vehicles’ charge on the French load curve and the production and
consumption balancing. The grid issues are not studied here.
The first part presents the state of the art in battery and electric vehicle technology, as well as
battery charge’s characteristics. Besides the French traffic data are studied, in order to build
the most realistic model as possible.
The second part explains the principle of the modeling. A general and flexible model for the
French fleet is built. This model can simulate the times of charge of a vehicle, and the state of
charge of its battery, in order to get a total load curve for several days for an electric fleet.
And finally, the different load curves are analyzed as well as their impacts on the electric
balancing system and the possible solutions to lower these impacts. As the peak load for the
electric vehicles’ charge corresponds to the total consumption peak in the evening according
to the most probable scenario, services have to be implemented to reduce or put back the
overload due to the electric vehicles. Those services (tariff signals regulations, battery
management system, cut-off injunctions) have been tested. A model for grid injections during
peak hours has also been developed, in order to help the electric system, and erase the electric
vehicles’ impact on the load curve. Studies for the optimum load, the reserves and the margins
are also suggested as well as other possible studies, such as EV’s carbon footprint.
The reflections and strategies developed in this master thesis don’t reflect RTE’s policy and
don’t commit the company.
Mélaine Rousselle - RTE-DMA – KTH 1

I would like to thank my supervisors at RTE, Emmanuel Neau, Thibault Prévost and Vincent
Lefieux for their patience during the development of the master thesis subject from a distance,
between Sweden and France, and for trusting me to carry out this study. I thank them for their
valued help and advice. I am specially thankful to Emmanuel for his guidance and for all the
time he devoted to answering to my questions.
I wish to thank the entire group for welcoming me and integrating me inside the team.
Finally, I am thankful to my examiner and supervisor at KTH, Lennart Söder who agreed to
supervise and review my work.

ABSTRACT .............................................................................................................................. 1
ACKNOWLEDGMENTS ....................................................................................................... 2
LIST OF FIGURES ................................................................................................................. 5
LIST OF TABLES ................................................................................................................... 7
GLOSSARY AND ABBREVIATIONS.................................................................................. 8
1. INTRODUCTION............................................................................................................ 9
1.1. PRESENTATION OF RTE............................................................................................... 9
1.2. BACKGROUND ............................................................................................................. 9
1.3. PURPOSE AND AIM .................................................................................................... 10
2. STATE OF THE ART AND TRAFFIC DATA........................................................... 11
2.1. ELECTRIC VEHICLE’S AND PLUG-IN HYBRID VEHICLE’S CONCEPT ........................... 11
2.2. BATTERY ................................................................................................................... 12
2.2.1. State of the Art.................................................................................................. 12
2.2.2. Lithium Batteries’ Principle............................................................................. 13
2.2.3. Charge of a Lithium Battery............................................................................. 15
2.3. CHARGE FACILITIES .................................................................................................. 17
2.4. VEHICLES CHARACTERISTICS .................................................................................... 18
2.5. TRAFFIC DATA .......................................................................................................... 20
3. MODELLING OF A PEVS’ FLEET ........................................................................... 23
3.1. CHARACTERISTICS OF THE FLEET’S MODEL .............................................................. 23
3.2. SIMULATION PROCESS ............................................................................................... 26
3.2.1. Principle ........................................................................................................... 26
3.2.2. Changes of State............................................................................................... 27
3.2.3. Algorithm.......................................................................................................... 29
3.3. MONTE CARLO ANALYSIS ......................................................................................... 31
3.4. EXTRAPOLATION ....................................................................................................... 33
3.5. REFERENCE SCENARIO .............................................................................................. 34
3.6. SENSITIVITY ANALYSIS ............................................................................................. 37
3.6.1. Impact of the Battery’s Initial State of Charge ................................................ 37
3.6.2. Impact of the EV/PHEV Distribution ............................................................... 40
3.6.3. Impact of the CV/PV Distribution .................................................................... 40
3.6.4. Impact of the Temperature ............................................................................... 41
3.6.5. Impact of the Available Points of Charge Distribution.................................... 42
3.6.6. Assumption of a Systematic Charge ................................................................. 46
3.6.7. Battery Exchange ............................................................................................. 48
3.6.8. Impact of the Other Parameters....................................................................... 49
3.6.9. Assumptions’ Impact Recap and Retained Reference Scenario....................... 49
3.7. PEVS’ “NATURAL” CHARGE CURVE ........................................................................ 50
4. IMPACTS ON THE FRENCH ELECTRIC SYSTEM.............................................. 52
4.1. FUTURE “NATURAL” CONSUMPTION CURVES ........................................................... 52
4.1.1. Background ...................................................................................................... 52

4.1.2. Consumption by 2015, 2020 and 2025............................................................. 53

4.1.3. Consumption with PEVs................................................................................... 53
4.2. TARIFF SIGNALS ........................................................................................................ 57
4.2.1. Tariff Signals’ Principle................................................................................... 57
4.2.2. Load Curves and Total Consumption Curves with Tariff Signals ................... 58
4.3. BATTERY MANAGEMENT SYSTEM............................................................................. 61
4.3.1. Battery Management System Principle ............................................................ 61
4.3.2. Load Curves and Total Consumption Curves with BMS.................................. 63
4.4. BATTERY MANAGEMENT SYSTEM AND TARIFF SIGNAL ............................................ 68
4.5. INJECTIONS INTO THE GRID ....................................................................................... 71
4.6. OTHER SOLUTIONS .................................................................................................... 75
4.7. EXTREME SCENARIOS................................................................................................ 77
4.7.1. The Worst Case ................................................................................................ 77
4.7.2. The Theoretical Optimum Cases ...................................................................... 78
4.8. RESERVES AND MARGINS .......................................................................................... 82
4.8.1. Reserves............................................................................................................ 82
4.8.2. Margins ............................................................................................................ 84
5. CO2 FOOTPRINT.......................................................................................................... 86
CONCLUSIONS..................................................................................................................... 89
MAIN ASPECTS OF THE STUDY .............................................................................................. 89
PERSPECTIVES ....................................................................................................................... 90
APPENDIX ............................................................................................................................. 91

Figure 1 - Drivetrain for Plug-in Hybrid Electric Vehicles (PHEVs) [Axsen, Burke, Kurani
(2008)].............................................................................................................................. 11
Figure 2 - Principle of a lithium-ion battery [Anderson (2008)].............................................. 13
Figure 3 - Experience curve for NiMH and Li-Ion batteries [Kromer et al (2007)] ................ 14
Figure 4 - Specific power versus specific energy, and battery possible applications [BERR
and Department for transport (2008)] .............................................................................. 14
Figure 5 - Charging Characteristics at various currents (charge capacity, charge current,
charge voltage) [AA portable power corporation] ........................................................... 16
Figure 6 - Load for the different types of charge ..................................................................... 16
Figure 7 - Electric consumption for EVs [Perrin, Renault (2009)].......................................... 19
Figure 8 - Vehicles’ trips distribution for Lille’s area ............................................................. 21
Figure 9 - Fleet characteristics for the simulation.................................................................... 24
Figure 10 - Possible states and transitions for a vehicle .......................................................... 27
Figure 11 - Simulation’s Flowchart ......................................................................................... 30
Figure 12 - Consumed power and load for three vehicles........................................................ 31
Figure 13 - Load, mean power and box-and-whisker plot: 100 simulations, 5000 vehicles ... 32
Figure 14 - Mean powers and Relative Standard Deviations with different numbers of
simulations ....................................................................................................................... 33
Figure 15 - Mean Power per vehicle and Relative Standard Deviations with 100 simulations34
Figure 16 - Load with an initial SOC of 100% for all the vehicles, for 3 days........................ 38
Figure 17 - Mean powers and Relative Standard Deviations for different initial SOC ........... 39
Figure 18 - SOC distribution function...................................................................................... 39
Figure 19 - Mean powers and Relative Standard Deviations for four scenarios (EV/PHEV) . 40
Figure 20 - Mean Powers and Relative Standard Deviations for six different PV/CV
distributions...................................................................................................................... 41
Figure 21 - Mean powers and Relative Standard Deviations for three temperature’s scenarios
.......................................................................................................................................... 42
Figure 22 - Mean powers and Relative Standard Deviations for three different fast charge
points’ distribution ........................................................................................................... 43
Figure 23 - Mean powers and Relative Standard Deviations for three different standard charge
points’ distributions for PVs ............................................................................................ 44
Figure 24 - Mean powers and Relative Standard Deviations for three different standard charge
points’ distributions for CVs ............................................................................................ 45
Figure 25 - Mean Powers and Relative Standard Deviations for three scenarios for Fast charge
points’ distribution for CVs.............................................................................................. 46
Figure 26 - Mean Power for systematic and not systematic charges during five days ............ 47
Figure 27 - Relative Standard Deviations with and without systematic charge....................... 47
Figure 28 - Load from one simulation with battery exchange ................................................. 48
Figure 29 - Mean load and relative standard deviation with battery exchange........................ 49
Figure 30 - Mean power for the reference scenario ................................................................. 51
Figure 31 - Consumption curves for a winter and summer workday....................................... 52
Figure 32 - Load in 2009, 2015, 2020 and 2025 ...................................................................... 53
Figure 33 - PEVs’ and electricity demand’s evolution [RTE (2009)] ..................................... 54
Figure 34 - Total load curve with PEVs in 2015...................................................................... 54
Figure 36 - Dynamics of the load curve with and without PEVs in 2020................................ 55


Figure 38 - Dynamics of the load curve with and without PEVs in 2025................................ 56
Figure 39 - Tariff Signal’s principle ........................................................................................ 57
Figure 40 - Load and consumed power for two vehicles ......................................................... 58
Figure 41 - Mean Powers and Relative Standard Deviations for three different scenarios for
tariff signals...................................................................................................................... 59
Figure 42 - Total load in 2020 with PEVs and tariff signals (40%)......................................... 59
Figure 43 - Dynamics of the load curve with and without PEVs in 2020, with tariff signals . 60
Figure 44 - Total load in 2020 with PEVs and tariff signals (60%)......................................... 61
Figure 45 - Battery Management System’s principle............................................................... 62
Figure 46 - Power for two vehicles with and without BMS..................................................... 63
Figure 47 - Load with different BMSs and without BMS........................................................ 63
Figure 48 - Mean powers and relative standard deviations for three scenarios for BMS’s
distribution ....................................................................................................................... 64
Figure 49 - Total load in 2020 with PEVs and BMS (40%) .................................................... 65
Figure 50 - Dynamics of the load curve with and without PEVs in 2020, with BMS (40%) .. 65
Figure 51 - Total load in 2020 with PEVs and BMS (80%) .................................................... 66
Figure 52 - Mean powers and relative standard deviations for different distribution for BMS
bis ..................................................................................................................................... 67
Figure 53 - Total load in 2020 with PEVs and BMS bis (80%)............................................... 67
Figure 54 - Load with different regulations ............................................................................. 68
Figure 55 - Mean powers and relative standard deviations for five different distribution for
BMS combined with tariff signals ................................................................................... 69
Figure 56 - Total load in 2020 with PEVs and BMS & Tariff Signals (40%) ......................... 70
Figure 57 - Total load in 2020 with PEVs and BMS & Tariff Signals (80%) ......................... 70
Figure 58 - Mean powers with and without injection .............................................................. 71
Figure 59 - Mean power and Mean injected power ................................................................. 72
Figure 60 - Total load curve in 2020 with PEVs and injection into the grid (scenario 1) ....... 73
Figure 61 - Total load curve in 2020 with PEVs and injection into the grid (scenario 13) ..... 74
Figure 62 - Dynamics of the load curve with and without PEVs with injections .................... 74
Figure 63 - Cut off injunction’s principle ................................................................................ 75
Figure 64 - Mean required time to charge................................................................................ 76
Figure 65 - Mean Available Energy......................................................................................... 76
Figure 66 - Mean load with fast charge point at anytime of the day........................................ 77
Figure 67 - Consumption curves in 2020 with 100% of the fleet having access to fast charge
points at anytime of the day ............................................................................................. 78
Figure 68 - Optimum consumption profile .............................................................................. 79
Figure 69 - Load profiles for optimum case in winter and summer......................................... 79
Figure 70 - Classes of load curves of PEVs according to their required charge time.............. 80
Figure 71 - Optimal load placement......................................................................................... 80
Figure 72 - Load profiles and load ........................................................................................... 81
Figure 73 - Consumption curve with one empirical simulation............................................... 81
Figure 74 - Mean Available Energy......................................................................................... 83
Figure 75 - Margin’s sizing principle....................................................................................... 84
Figure 76 - Mean power for EVs and different slots for CO2 emissions ................................. 86
Figure 77 - Initialisation of a fleet............................................................................................ 92
Figure 78 - Transition Matrix’s Flowchart............................................................................... 93
Figure 79 - Power Block’s Flowchart ...................................................................................... 94
Figure 80 - Number of running vehicles (fleet with 5000 vehicles) ........................................ 94

Table 1 - PHEVs and EVs battery requirement [Kalhammer et. al. (2007)] ........................... 12
Table 2 - Price and safety for actual and potential vehicle batteries presented at a 2007
industry forum [BERR and Department for transport (2008)]......................................... 12
Table 3 - Advantages and disadvantages of a lithium-ion battery [Soncini, RTE (2009)] ...... 13
Table 4 - Different types of charge .......................................................................................... 15
Table 5 - Type of charge according to the location & needed charge points [Renault (2009)]18
Table 6 - Electric consumption for PHEVs in city driving conditions [Hillebrand, Duoba
(2007)].............................................................................................................................. 19
Table 7 - Electric consumption for PHEVs in highway driving conditions [Hillebrand, Duoba
(2007)].............................................................................................................................. 19
Table 8 - French fleet in traffic in 2008 [MEEDDAT/SOeS (2009)] ...................................... 20
Table 9 - Commuting distances [Baccaïni, Sémécurbe, Thomas, INSEE (2007)] .................. 20
Table 10 - Number of daily trips [Lille Métropole (2006)] ..................................................... 21
Table 11 - Number of trips per day in Lille, Lyon and France[CERTU (2008)]..................... 21
Table 12 - Simulation characteristics ....................................................................................... 34
Table 13 - Allocation of PVs’ characteristics .......................................................................... 35
Table 14 - Allocation of CVs’ characteristics .......................................................................... 35
Table 15 - Allocation of number of trips.................................................................................. 36
Table 16 - Allocation of trip length for PVs ............................................................................ 36
Table 17 - Allocation of trip length for CVs ............................................................................ 36
Table 18 - Allocation of consumption...................................................................................... 36
Table 19 - Allocation of departure times ................................................................................. 36
Table 20 - Influence description............................................................................................... 37
Table 21 - Different tested scenarios for the initial SOC impact ............................................. 38
Table 22 - Different distributions for EVs and PHEVs ........................................................... 40
Table 23 - Different PV/CV distributions ................................................................................ 41
Table 24 - Three scenarios for the temperature’s impact......................................................... 42
Table 25 - Fast charge points’ distributions ............................................................................. 42
Table 26 - Standard Charge points’ distributions for PVs ....................................................... 43
Table 27 - Standard Charge points’ distributions for CVs....................................................... 44
Table 28 - Fast Charge points’ distributions for CVs .............................................................. 45
Table 29 - Parameters’ influence recap.................................................................................... 50
Table 30 - Parameters’ influence recap.................................................................................... 50
Table 31 - Three scenarios for tariff signals distribution ......................................................... 58
Table 32 - Three scenarios for BMS’s distribution.................................................................. 64
Table 33 - Different scenarios for BMS bis ............................................................................. 66
Table 34 - Five different scenarios for BMS & Tariff signals distributions ............................ 68
Table 35 - Different scenarios for injection into the grid......................................................... 72
Table 36 - Energy per day, and CO2 emissions........................................................................ 87
Table 37 - Characteristics of the EVs and PHEVs to come ..................................................... 91

BMS Battery Management System

CC Constant Current
CV Commercial Vehicle
DoD Depth of Discharge
DSM Demand Side Management
EV Electric Vehicle
HEV Hybrid Electric Vehicle
Li Lithium
NiMH Nickel-Metal Hydride
PEV Plug-in Electric Vehicle
PHEV Plug-in Hybrid Electric Vehicle
PV Private Vehicle
SOC State of Charge
TSO Transmission System Operator
UCTE Union for the Co-ordination of Transmission of Electricity
V2G Vehicle-to-Grid

1.1. Presentation of RTE
RTE (“Réseau de Transport de l’Electricité”) is responsible for transmitting electricity from

generation plants to local electricity distribution operators. This company is the only
Transmission System Operator (TSO) in France.
RTE was created in 2000 following the European law opening the European market up to
competition, and became a subsidiary company of EDF (“Electricité de France”) in 2005.
RTE is one of the largest European Transmission System Operator (TSO) with more than
100,000 km of lines. This company employs around 8,400 people, and its turnover in 2008
was around 4,200 M€.
RTE’s goals, defined through a public service contract between the State and the company,
are controlled by the Commission of Energy Regulation (CRE).
RTE has a public service mission and must operate, maintain and develop the French
electricity transport network at the best cost, while reducing its environmental impact. This
company guarantees all users a fair and non-discriminating access to the grid, and preserves
the freedom of all market actors. RTE aims to develop interconnection capacities, in
cooperation with the other European TSOs.
It must secure the balance between production and consumption at anytime (from long to
short terms and in real time), and assure the safety of the electric system operation. It must
also secure supply and alert public authorities in case of disruption risk [rte-france.com].
That is why RTE cares about the plug-in electric vehicles’ development.
DMA (Method and Support Department) is a R&D department of RTE’s. This department
carries out prospective studies, provides expertise on the electric network mechanisms, and
develops tools for the study and the development of the network. Therefore this master thesis
has been carried out in this department.
1.2. Background
People become more and more aware of the impact of their lifestyle, their mobility on the
environment. The trend is not to have bigger and bigger cars anymore, but less consuming
vehicles. Furthermore, in France, governmental incentives contribute to direct French people
towards less emitting vehicles. Hybrid Electric Vehicles (HEVs) enjoy a real success, but
their electric energy still comes from gasoline, and their battery ranges are really limited.
That’s why vehicles manufacturers focus more and more on Plug-in Electric Vehicles (PEVs)
including electric vehicles (EVs) and plug-in hybrid vehicles (PHEVs).
We currently cannot say if we will face a very high demand in the next couples of years, or if
we will have a slow increase in the vehicles market share. However, many studies foresee
around one million Plug-in Electric Vehicles (PEVs) in 2020. This development of electric
vehicles depends a lot on the improvements in battery technology, on the PEVs price, and the
battery range. The development of recharge facilities, the gas price evolution, or incentives
from the government are other important factors for the PEV’s future.
The PEVs’ carbon footprint depends on the electricity sources. The PEVs can reduce our
country’s dependence on the petroleum importations. Besides it is important to analyze the

electric system impacts in order to know accurately if the planned generation capacity will be
sufficient to supply the PEVs’ demands during the peak hours.
Lots of studies have been carried out on the impact of the electric vehicles on the total energy
demand. But the real impact on the consumption curves is still not well analyzed.
1.3. Purpose and Aim
The aim of this master thesis is to study the impact of PEVs’ development on the French load
curve.
It is important to know the consequences of introducing a large number of vehicles connected
to the grid. According to the time they are plugged in, to the power they require, they can
cause the resizing of reserves, if the installed capacity is not sufficient enough to keep up the
vehicles demand, or they can require the addition of new generation capacity.
We can imagine that the users will charge their vehicles when coming back to home, while
switching on their electric heaters, or the air-conditioning. Will it have an impact on the
production and consumption balancing system? Yes, probably but what will be the amplitude
of this impact? From how many vehicles can it have an impact?
The purpose of this master thesis is to try to answer to these questions, and try to analyze
solutions to reduce a possible impact.
With the announced characteristics for batteries and PEVs and assuming that the PEVs will
replace the conventional vehicles whose utilization is already well known, it is possible to
study such an impact.
That is why we will first study the state of the art of the battery technology and PEV, and
define the charge characteristics for a PEV.
After analyzing the French traffic data, a model for PEV’s fleet is built in order to simulate,
with Matlab, the behavior of a certain number of PEVs. With this model, it will be possible to
know when the EV/PHEV users will charge their vehicles and the state of charge (SOC) of
their battery. A sensitivity analysis is performed.
In a second part, with the load curves resulting from different scenarios, we will study their
impact on the total French load curve, on the reserves, and try to find solutions to remedy for
possible problems. A CO2 study is also performed.
The study is limited to workdays since the consumption is more important during workdays,
and thus the impact might be more important during those days.
Some assumptions are made on the battery technology improvement and on the type of
charges. Simplifying assumptions are used to describe the users’ behavior.
The impact on the electric network (such as overload of lines) is out of scope of this master
thesis.

! "
Many data are available on the state of the art of the battery technology, on PEV and traffic
data. General documents on the introduction of PEVs in a fleet, or on the vehicle-to-grid
concept are disposable, however real analysis on the impact of introducing PEVs on the
electric system are quite scarce.
2.1. Electric Vehicle’s and Plug-in Hybrid Vehicle’s Concept
In 1899, the first car was electric. But then, due to the gasoline’s easy transport and storage,
the internal combustion engine eclipsed the electric generator to power the vehicles’ wheels.
However, in our current environmental context, Electric Vehicles (EVs) and Plug-in Hybrid
Vehicles (PHEVs) become once again trendy. Car manufacturers are developing new EVs
with higher battery range, and PHEVs are derived from current hybrid vehicles to increase the
“all electric” range. The industries aim at launching these vehicles in the next couple of years.
PHEVs combine operational aspects of both electric vehicles (EVs) and hybrid electric
vehicles (HEVs). Similar to an EV, a PHEV can store energy in an onboard battery for use
during daily driving and recharge the battery from the electric grid. However PHEVs also
have internal combustion engines used for propulsion when the required speed is more
important, or when the state of charge of the battery is not sufficient.
Two main types of PHEVs exist: series or parallel drive train. The principle of both designs is
explained on Figure 1.
Figure 1 - Drivetrain for Plug-in Hybrid Electric Vehicles (PHEVs) [Axsen, Burke, Kurani (2008)]
In the first design, the gasoline is used to drive an electric generator who powers the wheels,
or recharges the battery. In the parallel design, the wheels rotate thanks to either the gasoline
engine, either the electric generator.
EVs don’t have an internal combustion engine, therefore they need a greater battery size than
PHEVs.

However, the EVs and PHEVs will have to present some characteristics to be attractive for
the costumers. The whole problem consists of an optimisation between the battery size and
weight, the battery price and the battery range.
Table 1 - PHEVs and EVs battery requirement [Kalhammer et. al. (2007)]
Minimum
Power electric Energy
Max Peak power cost
density start density
weight (kg) (kW)
(W/kg) capacity (Wh/kg)
($/kWh)
(kWh)
PHEV 120 65;50 540;400 6;12 50;75 300
EV 250 50-100 200;400 25;40 100;160 150
Table 1 presents the energy storage requirements for PHEVs and EVs prepared for the
California Air Resources Board on the energy storage requirements for HEVs, PHEVs and
EVs. The figures are deduced from the constraints set up by requirements on the weight, the
cost, the battery range but are not current data for the batteries.
2.2. Battery
2.2.1. State of the Art
The future of the EV and the PHEV depends a lot on the battery technology development. The
lithium-ion technology is the most promising technology for the future EVs and PHEVs.
Indeed, the energy density of a Lithium battery is much higher than the other current
technologies, and this is why this type of battery is highly used for portable devices (digital
cameras, laptops, cell phones…).
Table 2 - Price and safety for actual and potential vehicle batteries presented at a 2007 industry forum
[BERR and Department for transport (2008)]
Li-M- Na-NiCl2
Li-ion NiMH lead-acid
Polymer (Zebra)
Energy density
75-120 100-120 50-70 100-120 20-30
(Wh/kg)
Power density
1000-3000 200-250 1000-1500 180 200-500
(W/kg)
Cost ($/kWh) 1000-2000 ? 1000 600 100-200
300-800
lifetime (valve
(cycles, 100% 1000-3000 ? 2000 1000 regulated
DoD) lead acid
battery)
no
temperature single
Issues safety, cost commercial
limitations supplier
product
The main constraints for EVs and PHEVs batteries are the cost, the safety and the
performance.

2.2.2. Lithium Batteries’ Principle
Figure 2 - Principle of a lithium-ion battery [Anderson (2008)]

The negative electrode (anode) of Li-Ion battery cell is graphite and the positive electrode
(cathode) is a lithium-bearing metal compound. Li-ion batteries have a good cycling aptitude
due to the stable electrode structure: charging and discharging involves exchange of lithium
ions between the electrodes via the electrolyte [saftbatteries.com].
While charging, the electrons are pumped from the cathode to the anode. Positively charged
lithium ions move from the cathode through the separator via the electrolyte to the anode.
While discharging, positively charged lithium ions move from the anode through the separator
via the electrolyte to the cathode. Electrons move through the external load from the anode to
the cathode, resulting in a current that provides power to the load [Anderson (2008)].
For a problem of safety, the charge intensity and the charge voltage must be correctly adapted
for a Li-ion battery. An important overload can induce the destruction of the battery. Indeed
an overload can produce hydrogen and cause the battery explosion.
Intern protection systems against temperature rises or overload are generally implemented in
the Li-ion batteries [Soncini, RTE (2009)].
Table 3 - Advantages and disadvantages of a lithium-ion battery [Soncini, RTE (2009)]
Advantages Disadvantages
no memory effect explosion risk

Energy density higher than the
other technologies for a lower essential raw material
weigh
advanced technology deep cycles are inadvisable
can have different shapes cost

Figure 3 - Experience curve for NiMH and Li-Ion batteries [Kromer et al (2007)]
NiMH is a mature battery technology who has reached a technological advancement’s end.
HEV applications have resulted in reduced costs due to economies of scale, but little
advancement remains for further performance or cost improvements.
Li-Ion battery will clearly be the dominating technology in the future developpment. Indeed,
there is a growing consensus that Li-ion batteries offer the most promising combination of
energy storage capacity with power (Figure 4) [BERR and Department for transport (2008)].
There are numerous opportunities of cost reduction thanks to economies of scale.
Additional cost reductions will come from the use of lower-cost materials, process
improvements, and improved yields [Anderson (2008)].
Figure 4 - Specific power versus specific energy, and battery possible applications [BERR and Department
for transport (2008)]

Within twenty years, we can have an evolution of performance of 10%. This means that a
battery of 20 kWh in 2010 will have a capacity of 22kWh in 2020, with the same weight and
same volume [MEEDDAT, SGT, Saft (2009)].
2.2.3. Charge of a Lithium Battery
Currently, we can charge a vehicle with two types of charge: conductive and inductive charge.
Inductive charging is one kind of short distance wireless energy transfer. The conductive
charge requires direct electrical contact between the battery and the charger. The load with
inductive charge might be a little bit higher than a conductive load since there might be more
losses. In this report, we will focus on the conductive charge, since it is the type of charge that
is more likely to be developed.
We can charge a Lithium Ion battery with different current. The different terms used in this
report are expressed in Table 4.
Table 4 - Different types of charge
load charge time

slow charge < 6kVA > 4 hours
standard charge between 6 kVA and 22 kVA between 1 and 4 hours
fast charge between 43 kVA and 125 kVA less than 1 hours
We can see on Figure 5 the charge current, voltage and capacity of a battery cell. We can
observe, that we first have a constant current phase (CC phase), then, when the capacity is
approximately equal to 80%, the voltage becomes constant, and the current starts to decrease
(Constant voltage (CV) phase). On the figure, we can see two different charges: one at
0.1CmA (Capacity milliamp: the current is equal to 0.1 times the capacity of the battery),
which we could call slow charge, and one at 0.5 CmA, which we could call standard charge.

Figure 5 - Charging Characteristics at various currents (charge capacity, charge current, charge voltage)
[AA portable power corporation]
With these charge’s characteristics, we could deduce that at the end of the charge, the load
curve looks like an exponential curve. However, in a concern of simplicity for the simulation,
we consider that a normalized charge curve is a constant curve, with two different levels.
Indeed, it is better for the battery to be charged with a constant power. The charge curve
model used for the simulation is represented on Figure 6. This is an assumption for this
model, but it can be adapted according to the battery manufacturers’ announcements.
Figure 6 - Load for the different types of charge

Slow charge
Slow charge will be the most used type of charge because it is already available for a majority
of consumers with current domestic installation. In 90 to 95% of the cases the charge will be

done with power under 6kVA, during the night [MEEDDAT, SGT (2009)]. This type of
charge can last 7,5 hours for a totally depleted battery.
Standard charge
It is the type of charge that can be used at workplace, or on long-duration car parks. It can last
2.5 hours [MEEDDAT, SGT (2009)].
Fast charge
A battery of 20 kWh charged with a power greater than 10 kW can endure severe damages.
There is a possible compromise so that the battery can bear a fast charge, but this has an
impact on the battery lifetime. But future improvements on the battery lithium-ion will allow
extending the fast charge acceptability domain [MEEDDAT, SGT, Saft (2009)]. This charge
lasts only 30 minutes.
Battery exchange
First, the battery exchange possibility will only be on an experimental phase. Renault plans to
use battery exchange stations in a couple of years’ time frame. However, the batteries in these
stations would be mostly charged with a fast charge in order to avoid having a large amount
of stocked batteries [MEEDDAT, SGT (2009)].
2.3. Charge Facilities
The electrical mobility development depends a lot on the available charge infrastructure.
Definitely, nobody will buy a PEV, if he cannot charge it at home. Besides, it is more likely
that people will only buy EVs if their daily covered distance is shorter than the battery range.
So if people use them properly and don’t forget to recharge it at home during the night, they
will probably not even need public charge points when they use their vehicle in their usual
way. However public charge points can be necessary to reassure the users, or when they need
to do a special trip.
Only 43% of the French population have access to a private parking during the night [Lille
Métropole (2006)].
There are currently around 80 free public charging points in Paris for PEVs and 30 in Paris’
suburbs [http://www.paris.fr]. There are at total of around 200 points in France (July 2009).
The French government will devote 1.5 billions euro for the installation of 400 000 electric
plugs by 2020 on the public roads and parking lots [enerzine.com].
Moreover, only 41% of the labour force has access to a parking space at workplace [Gantelet,
Sareco (2003)].
On Table 5, we can see the possible number of needed charge points according to the
location, figures set up by Renault [Renault (2009)]

Table 5 - Type of charge according to the location & needed charge points [Renault (2009)]
Charge points for 100
Place Charge EV in 2020
home, collective housing Domestic charge 120
workplace Normal charge 30
Shopping parking places, gas stations Fast charge 10
Short time parking places Fast charge 2,5
Recently, the largest European energy supply companies and vehicle manufacturers agreed on
the basic parameters for the charging connection for electric cars, in order to have a
standardisation of the plug in Europe [mennekes.de].
2.4. Vehicles Characteristics
The load resulting from the PEVs’ charges depends on the vehicles’ consumption, the battery
size. Data have been collected on this subject.
Battery range and size for EV

Many sources (EDF, Renault…) foresee that the future battery range will be greater than 150
km for EVs. The battery size for EVs should be between 20 kWh (most common, for
domestic use) and 30kWh (for commercial use).
Battery range and size for PHEV

The battery range for a PHEV can fluctuate from 40km to 60km [Sénat (2005)]. For instance,
the Ford Escape will have a battery of around 10kWh, for Renault, it will be around 15kWh.
The Toyota Prius will certainly have a battery size of 9kWh, for Chevrolet Volt it’s around 16
kWh.
EVs engine efficiency

It’s difficult to interpret the announced power consumption for EVs. For [BERR &
Department for transport (2008)], the consumption by 2010 will be 16 kWh/100km, 13
kWh/100km by 2020 to decrease finally until 11 kWh/100km. But these consumptions
certainly don’t take into account the auxiliaries. For the car manufacturer Renault, the
consumption by 2010 with the auxiliaries and losses will be around 22 kWh/100km, and 18
kWh/100km by 2020 [Perrin, Renault (2009)].
Furthermore, this consumption depends also on the type of use (urban, road, highway…).

Figure 7 - Electric consumption for EVs [Perrin, Renault (2009)]
PHEVs engine efficiency

On Table 6 and Table 7 we can see the consumption of three different PHEVs in city driving
conditions (Table 6) and on highway driving conditions (Table 7).The battery is much more
used when running in the city.
Table 6 - Electric consumption for PHEVs in city driving conditions [Hillebrand, Duoba (2007)]
City driving Average

Charge Usable
conditions - cycle Battery
Depleting Energy
charge depletion usage
range km (kWh)
Operation (Wh/km)
kokam Hymotion Prius 79 38,6 2,9

A123 Hymotion Prius 81 40,2 3,1
EnergyCS Prius 82 54,7 4,5
Table 7 - Electric consumption for PHEVs in highway driving conditions [Hillebrand, Duoba (2007)]
highway driving Average

Charge Usable
conditions - cycle Battery
Depleting Energy
charge depletion usage
range km (kWh)
Operation (Wh/km)
kokam Hymotion Prius 61 52,8 3,2

A123 Hymotion Prius 71 49,6 3,4
EnergyCS Prius 64 80 4,9
Charge-depleting refers to a mode of vehicle operation when all the used energy comes from
the battery pack. EVs operate only in this mode. Most PHEVs operate in charge-depleting
mode at startup, and switch to charge-sustaining mode after the battery has reached its
minimum state of charge (SOC), when the vehicle' s all-electric range is over.

2.5. Traffic Data
In order to build a model of a French fleet, it is important to know users’ behaviour (daily
covered distance, number of trips, and hours of their trips…). Therefore statistical data on
traffic have been studied.
Number of vehicles
Table 8 - French fleet in traffic in 2008 [MEEDDAT/SOeS (2009)]
Number of
French fleet in traffic in 2008 vehicles (1000
vehicles)
Private vehicles 30 775
gasoline 14 438
diesel 16 337
Commercial vehicles 5 725
gasoline 809
diesel 4 916
Heavy vehicles 638
Heavy trucks 554
buses 84
Total vehicles number 37 138
The current evolution rate of the French fleet is around 1 % [DDRPE (2008)].
But we can assume by 2025 a yearly average growth rate for the private vehicles fleet of
0.86% [MTET/SESP (2007)].
Around 77% of French people live in urban and suburban areas [Science et Décision (2006)].
Nowadays, around 30% households have two cars, and around 19% have none [IFEN (2007)].
In 2020, we can consider that around 40% household will have more than one car.
Around 70% of people go to work by car [Science et Décision (2006)].
We can assume that the PEV will be a second vehicle for the households, used for the
commuting trips.
Distance and number of trips

Table 9 presents the average and median trip distance for a private car.
Table 9 - Commuting distances [Baccaïni, Sémécurbe, Thomas, INSEE (2007)]
Commuting distance (data 2004) Road Distance (km)

Average Median
Urban poles 23,6 5,8
Suburban cities 30,5 13,6
Rural spaces 28,4 10,2
whole 25,9 7,9
The average number of trips per day is 3.5/day [Science et Décision (2006)].

Table 10 - Number of daily trips [Lille Métropole Table 11 - Number of trips per day in Lille, Lyon
(2006)] and France[CERTU (2008)]
number of car trips
Number of daily trips for data 2006
/day/inhabitants
Lille's inhabitants
Lille 2,1
11 and more 2% Lyon 1,6
9 or 10 4% France 2,2
7 or 8 9%
5 or 6 20%
3 or 4 31%
1 or 2 24%
0 9%
Travel hours
The times of the day with the biggest traffic is from 7:00 to 9:30 and it reaches its peak
between 8:00 and 8:30, and from 16:00 to 20:00 with a maximum between 18:00 and 18:30
(the traffic remains high until 21:30 then decreases until midnight).
From 12:00 to 14:00: 50% of the traffic is caused by commuting trips [Science et Décision
(2006)].
Figure 8 - Vehicles’ trips distribution for Lille’s area

On Figure 8 we can see the vehicles’ trips distribution for Lille area. However, this graph
could be generalised to big cities.
Average daily driving distance in urban areas for a captive fleet

The average daily mileage of a commercial vehicle in urban area is 98km/day [DAEI-SESP –
Samarcande (2005)].
Average daily driving distance in rural areas of a captive fleet

The average daily mileage of a commercial vehicle in non-urban area is 250 to 350 km/day
[DAEI-SESP – Samarcande (2005)].

# $ %&' (
The previous data are used in order to build a model of a French fleet. The reasoning is valid
for any country by adapting the data. The model will then be simulated with Monte Carlo
method in order to have mean charge curves for a typical day.
3.1. Characteristics of the Fleet’s Model
To create a representative French fleet, the total number of vehicles was split up into several
classes (see Appendix Figure 77).
First the purpose of use of the vehicle is considered: Commercial Vehicle (CV) or Private
Vehicle (PV); then if it is a plug-in-vehicle (PHEV) or an electric vehicle (EV), and finally, if
the vehicle is used in urban/suburban areas or in rural areas.
Then, for each vehicle, a random number is allocated for the battery size, speed, consumption,
departure times, trips length, and number of trips, initial state, initial State of Charge (SOC),
and the available type of charge at a given period of the day.
The different characteristics are allocated according to the global data, so that the fleet’s
characteristics represent as much as possible the collected data. The initialisation of a fleet
with the allocation of the different characteristics is represented on Figure 9.

Initialisation of a fleet
Utilisation allocation : PV/CV

(% of the fleet according to a
normal distribution)
Type allocation: EV/PHEV

(% of the fleet according to a
normal distribution)
Geographical allocation : Rural/Urban

(% of the fleet according to a normal
distribution)
Battery Size
(Normal distribution)
TripLength
(Normal distribution for CV, log normal distribution
for PV)
Number of Trips (for each period of the day)

(different fixed number)
Speed
(Fixed number)
Consumption
(fixed number)
Consumption = Consumption * temperature coefficient
Available Type of Charge (for each location)
Trip Length
(reduced for some EVs)
Departure Time (for each trip)

(normal distribution)
Initial Battery State of Charge
Figure 9 - Fleet characteristics for the simulation

The battery size for each vehicle is a random number, allocated according to a normal
distribution, with a mean value depending on the type of the PEV (EV/PHEV). However, we
cannot totally deplete the battery, so the usable battery size is lower than the battery size
announced by the manufacturers. So we have to multiply the battery size by a coefficient
representing the usable percentage of the battery.
The trip length is also a random number but it depends on the use of the vehicle (CV/PV),
and on the geographical location (urban/rural). It is assumed here that all the trips covered
during the day by one vehicle have the same length.
We have allocated the length trip according to a log normal distribution for PVs. Indeed the
collected data for private vehicles give us a median and a mean trip length. Given that the
median is much lower than the mean value, a log normal distribution is more appropriate.
σ2
µ+
If X has a log normal distribution of parameters and , then its expectation value is e 2
and its median is e µ . For the trip length, the collected data give us the median m and the
mean value m . So the different parameters are given by:
µ = ln(m)
σ = 2(ln(m ) − ln(m) )
For CVs, the trip length is allocated according to a normal distribution. Indeed, a median
value is not available in the collected data.
For each period of the day (morning, midday, afternoon, evening), a number of trips to cover
is given to the vehicle. It is assumed here, that except for midday, the vehicle has only one trip
to cover for a given period. For midday, it can be two trips.
The speed is not a random number, but depends on the location (urban or rural) and the use of
the vehicle. This data is used in order to know the arrival time of the vehicle.
For each place where the vehicle could be during the day (home, workplace, any place at
midday), the vehicle has a corresponding available type of charge.
For each period of the day, it is allocated to the vehicle a departure time for its trip. Except
for midday, the vehicle has one departure time. Given that the vehicle can cover two trips at
midday, it has two departure times for the both possible trips. For CVs, the departure times
can also correspond to time to charge. Indeed, as the entire work day of a CV is represented
by only one or two trips, the vehicle cannot start to recharge as soon as it parks, otherwise
some vehicles could start to charge at 10am for instance whereas they are considered to still
be working.
The consumption is not a random number, but its value depends on the geographical location
and on the use of the vehicle. For PHEVs the consumption can be modified if the battery
range is not sufficient to run on electricity for the whole day. Indeed, if the battery range is
not sufficient, the user will run more on gasoline, and then we can consider that the electricity
consumption can be decreased. However, it is less decreased for urban PHEVs than for rural
PHEVs, given that the speed and the trip length in urban area are lower than in rural area.

Moreover, after allocating a trip length, we reduce it according to the battery range for EV if
the user has to cover more kilometres than its battery range. Indeed, we can assume that an
EV’s user only buys an EV with a battery range greater than his daily covered distance.
For the simulation, we need to have an initial state for the vehicle (“parked”, “running” or
“charging”), and an initial state of charge (SOC) for the battery.
The consumption of a PEV depends much more on the temperature than an internal
combustion vehicle. Indeed, the energy for the heating or for the air-conditioning is directly
taken from the battery for a PEV. Besides, when the air is cooler, the electric motor consumes
more than usually. Thus the temperature has a real impact on the consumption and we have to
multiply the consumption by a coefficient representing the temperature impact.
The temperature can also have an impact on the battery’s efficiency. So when the vehicle is
charging, we have to multiply the load by a certain coefficient representing the impact of the
temperature on the charge’s efficiency.
The charge and discharge efficiency is taken equal to 100% in this model. (Nowadays the
charging and discharging efficiency is around 94% [Andersson, Elofsson (2009)])
Only workdays are studied in this master thesis. During weekends and holidays, users’
behaviors are really different. As the EV might be a second vehicle, it might not be used at all
during the week ends. But PHEVs can certainly be used, and then the distances covered are
really different than during work days.
It is assumed that the PEV’s user charges systematically his battery when he has an available
charge point. This assumption will be analysed later on. Some other assumptions (such as the
percentage of vehicles having access to a fast charge point for instance) will also be studied
further on.
3.2. Simulation Process

3.2.1. Principle
For the simulation, we consider that the vehicle can take three different states: “parked”,
“running” and “charging”. At each step of the simulation, the vehicle will have to decide
according to probabilities equal to 0 or 1 to change of state or not.
We assume that a parked vehicle can start running, or start charging, or obviously stay parked.
A charging vehicle can start running (even if the charge is not over, the user can decide to
unplug his vehicle and drive), or keep on charging. We consider that even if the battery is full,
the vehicle stays in the state “charging” with the load equal to 0.
A running vehicle can keep on driving if the battery is not fully depleted, or if the trip is not
over. Otherwise, the user parks the vehicle.
We consider in this model that the vehicle cannot charge directly after running. It first has to
go through the state “parked”. If the user wants to plug his car as soon as he parks his car, the
vehicle will go through the state “parked” only for one step, which is in this simulation quite a
small time.
We consider that the vehicle cannot go from the state “charging” to the state “parked”. Indeed
it is quite unlikely that the PEV’s user will come to unplug the vehicle once the charge is

over. Moreover when the charge is over, as the load is equal to 0, there is no difference
between the state “charging” and the state “parked”.
All the different states and possible transitions are summed up on Figure 10.
pR,P
pP,P pR,R
number of trips =
parked running
number of trips + 1
SOC, remaining distance

p *,R
pC,C
pP,C p *,R
charging number of trips =

number of trips + 1
SOC
Figure 10 - Possible states and transitions for a vehicle
3.2.2. Changes of State
The change of state is decided through probabilities equal to 0 or 1. These probabilities are
described below.
• Probability to go from the state “parked” or “charging” to the state “running” : p*,R
p*,R depends on the probability that the user can drive (pSOC), that it is time to drive (ptime) and
that the user wants to drive (pwants to drive).
ptime: probability that it is time to drive.

For each period of the day, each vehicle has a departure time. So if the departure time
for the period given by the simulation corresponds to the time of the simulation, ptime is
equal to 1. Otherwise, ptime is equal to 0.
pSOC: probability that the user can drive

For EVs, we assume that the needed SOC is the battery range corresponding to the
total covered distance during the whole day. For CVs who can charge at midday, the
needed SOC is the battery range corresponding to the total trip length for the morning.
If the needed SOC is lower than the SOC, pSOC is equal to 1.
For PHEVs, the state of the battery is not a condition to start running or not. So pSOC is
always equal to 1 for PHEV.
pwants to drive : probability that the user wants to drive.

For the period of the day given by the simulation, we know the number of trips for
each vehicle. So if the vehicle has not done its trip for the given period, pwants to drive is
equal to 1. Otherwise, pwants to drive is equal to 0.
Finally p*,R = ptime . ptime . p wantstodrive . p SOC
• Probability that the vehicle stays in the state « running » : pR,R

If the user still has kilometres to cover, and if the battery is not fully depleted for EVs, pR,R is
equal to 1.
• Probability to go from the state “running” to the state “parked” : pR,P

p R,P = 1 − p R,R
• Probability to go from the state « parked » to the state « charging »: pP,C

pP,C depends on the probability that the user can charge (pTC) and could depend on the SOC
(pSOC). But in this simulation we consider that the user will charge whenever he can
(“opportunist charge”). Besides for CVs, it depends on the time of charge (ptimecharge).
pTC : probability that the user can charge.

Given that for each place, we know the available type of charge, and since the location of the
vehicle is given by the simulation, we can deduce pTC .
pSOC : probability that the user wants to charge. Assuming that the user charges
whenever he can, pSOC is always equal to 1.
ptimecharge: probability that it is time to charge.
For PVs, ptimecharge is always equal to 1, as we consider that they charge whenever they can.
CVs have a time to charge for midday and for the late afternoon. So if the time to charge for
midday and the afternoon corresponds to the time of the simulation, ptimecharge is equal to 1,
otherwise, ptime is equal to 0. For the other periods of the day, ptimecharge is equal to 1.
p P ,C = pTC . p SOC . ptimech arg e
However, the priority is given to the trip. So if pP,C is equal to 1, and so is p*,R , then we
change pP,C to 0.
• Probability to remain in the state “charging”: pC,C

pC ,C = 1 − p*,R
• Probability to remain in the state « parked » : pP,P

p P,P = 1 − p P , R − p P ,C
We can then write the transition matrix T:

p C ,C p*,R 0
T= 0 p R,R p R,P
p P ,C p*,R p P,P
Then, the new state of the vehicle (staten) can be determined by:
staten = staten-1 .T

The flowchart of the transition matrix’s calculation is in the Appendix (Figure 78).
3.2.3. Algorithm
At each step of the simulation, the program determines the new state of each vehicle,
according to the process explained before. Besides it calculates the needed power for each
vehicle according to the SOC of the battery and the type of charge (slow, standard, fast).
In this flowchart (Figure 11) we can see that the fleet created as explained before is used. We
initialise the different parameters such as the time, the initial state of the vehicle, the
completed number of trips and the location of the vehicle. Then the different parameters are
calculated or deduced as explained in the flowchart. The number of trips depends on the
period of the day, as well as the departure time and the Boolean. In this simulation, the
Boolean is always equal to true (this Boolean will be used later on). The available type of
charge depends on the vehicle’s location (“place”). Then the load (P(time)) can be calculated.
The power block is in the Appendix (Figure 79). The load is calculated thanks to the model
described on Figure 6. It is assumed that battery’s improvements allow fast charge without
severe damages.
The new state of charge of the battery (SOC) can be deduced. The simulation step
(“SimulStep” in the flowchart) is taken equal to 10 minutes.
And then, according to the new SOC, the time, the number of trips for the period, we change
the state of the vehicle, with the transition matrix. The transition matrix (T) is calculated in the
Transition’s block (in Appendix Figure 78).
If the vehicle starts running, we increase the number of completed trips, and we change the
location of the vehicle, according to the corresponding period of the day and the number of
trips for the period.
Then the time can be raised, and the simulation continues until the simulation period is
reached (“SimulPeriod”). The result of the simulation is P, the load for each step of the
simulation. And this simulation is done for each vehicle.
The simulation is performed with Matlab.

F le e t's
ch a ra cte ristics
In itia lisa tio n

tim e = 0
sta te 0 = p a rke d
p la ce = h o m e
co m p le te d n u m b e r o f trip s = 0
p e rio d o f th e d a y
N u m b e r o f trip s B o o le a n = tru e D e p a rtu re tim e
A va ila b le T yp e o f C h a rg e
P o w e r B lo ck
P (tim e )
S O C = S O C + e fficie n cy co e fficie n t * P (tim e ) * S im u lS te p / B a tte ry S ize
T ra n sitio n 's
B lo ck
S ta te t = S ta te t-1 *T
No
S ta te t =
ru n n in g ?
S ta te t-1 =
No ch a rg in g o r
p a rke d ?
C o m p le te d n u m b e r o f trip s = C o m p le te d n u m b e r o f trip s+ 1
P la ce
tim e = tim e + S im u lT im e
tim e < S im u l No
P e rio d
Yes
End
R e su lt: P
Figure 11 - Simulation’s Flowchart

Figure 12 - Consumed power and load for three vehicles

On Figure 12, we can see the result of a simulation with three vehicles. The vehicle 1 is a
private vehicle. We can see that it starts its trip around 6am, and then it arrives at work and
can start charging with a standard charge. Then this vehicle moves at midday, comes back to
work at 14:00 and charges. It finally goes back home at 15:00 and charges at home with a
slow charge. Vehicle 2 is a commercial PHEV. It starts its long trip at 6am. We can see that
the vehicle hasn’t enough battery so it also has to use gasoline (gross power is electric power
and the equivalent in electric power of gasoline power). Given that this long trip represents an
entire day of work, it can only start charging around 20:00. Vehicle 3 starts its trip around
9am, and comes back home around 16pm and directly starts charging with a slow charge.
This is the result of a simulation of a fleet of only three vehicles. The idea is to simulate a
fleet with a larger number of vehicles, and sum up the load to have a total load curve for a
large fleet.
3.3. Monte Carlo Analysis
The load curve resulting from one simulation can change a lot from another simulation given
that for each simulation the characteristics are reallocated to the fleet. That is why it is more
appropriate to study the mean value of many simulations. The standard deviation is also of
interest since it can give a measure of the variability or the dispersion of the mean value.
The idea is to simulate a large enough number of observations for a same scenario (that is to
say with the same assumptions), but with a new random fleet for each simulation. And then
we can calculate the expectation value of the load thanks to the random observations of the
load.
We know that if we have n independent observations, P1,… Pn, of the load P, then the mean
value of these observations:

1 n
P= Pi
n i =1
is an estimate of E[P].
In the same way we can determine an estimate of the variance and the median for each step of
the simulation. This is particularly interesting for the peaks in the morning, at lunch time and
in the evening.
The mean load curves represent normalized mean power, that is to say the mean load divided
by the number of vehicles. So this power is still in kW, but we have to multiply it by the
number of vehicles to represent the load for an entire fleet.
Figure 13 - Load, mean power and box-and-whisker plot: 100 simulations, 5000 vehicles
The load presented on Figure 13 is the power for 5000 vehicles divided by the number of
vehicles. We can see that for the peak at 20:00, the different simulations are quite different.
The box-and-whisker plots for the Monte Carlo simulation are also represented to show the
dispersion and outliers. At peak time, 25% of the simulations are under 0.96 kW/vehicle and
25% are higher than 1.20 kW/vehicle.
Then the influence of the number of simulations has been tested.

Figure 14 - Mean powers and Relative Standard Deviations with different numbers of simulations
We can see on Figure 14 that no matter the number of simulations, the mean powers are
identical. The relative standard deviations have also been plotted.
The relative standard deviation is the absolute value of the coefficient of variation. The
coefficient of variation (CV) is a normalized measure of dispersion of a probability
distribution. It is defined as the ratio of the standard deviation ( ) to the mean value ( ).
The relative standard deviations with 500 and 1000 scenarios are identical. The relative
standard deviations for 50 and 100 simulations are less smooth than for a higher number of
simulations. However, they are similar enough (less than 2% of difference). Since 100
simulations are representative enough and given that the simulation time for 500 and 1000
simulations are too important, only 100 simulations will be performed for the other analysis.
3.4. Extrapolation
Given the important required time to simulate a fleet with a high number of vehicles, another
approach is preferred to simulate a fleet of one million vehicles. The simulations will be
performed with a limited but representative number of vehicles and then the results will be
extrapolated in order to have the final load curve.
That is why we have first studied the correlation of the curves with 100, 500, 1 000, 5 000,
and 10 000 vehicles in order to identify the limit of reliability. And then the load curve for a
larger fleet can be deduced by extrapolation of a load curve with a representative number.

Figure 15 - Mean Power per vehicle and Relative Standard Deviations with 100 simulations
We can see on Figure 15 that the mean values of 100 simulations with five different numbers
of vehicles are very similar. Then the relative standard deviation for each step of the
simulations has been plotted. The relative standard deviations of the 100 simulations with the
five different scenarios (Figure 15) are more important when the number of vehicles is
smaller. We can see that from 5000 vehicles, the relative standard deviations converged. We
can conclude that a fleet with 5000 vehicles is representative enough. Therefore, all the
simulations will be performed with 5000 vehicles.
3.5. Reference Scenario
For the simulations, a reference scenario for the different fleet’s characteristics has to be
retained. The data of the characteristics are described below, and many of them come from the
bibliography study. Some of them will be discussed after.
Simulation step, simulation period, number of vehicles

Table 12 - Simulation characteristics
Number of
Simulation period Simulation Step
vehicles
5000 30 hours 10 min
Some characteristics for PVs

The assumed percentages of PVs/CVs, EVs/PHEVs are random numbers drawn from normal
distribution.
The battery size is a random number for each vehicle according to a normal distribution.

Table 13 - Allocation of PVs’ characteristics

Geographical Battery size Available Charge
EV/PHEV Speed (km/h)
Characteristics PV area (kWh) points
EV PHEV Urban Rural Urban Rural PHEV EV Work Lunch Home
Mean value 80% 20% 80% 80% 20% 50 70 9 25 20% 5% 100%
Standard
10% 10% 10% 10% 0 0 3 5 0 0 0
deviation
random
number for
normal normal each vehicle
distribution distribution according to
a normal
distribution
The available type of charge at work is a standard charge, at lunch it is fast charge, and at
home, slow charge. These figures are assumptions.
Some characteristics for CVs

Table 14 - Allocation of CVs’ characteristics
Geographical Battery size Available points of

EV/PHEV Speed (km/h)
Characteristics CV area (kWh) charge
EV PHEV urban rural Urban Rural PHEV EV Work Lunch Home
Mean value 20% 80% 20% 70% 30% 30 40 9 25 0% 20% 100%
Standard
10% 10% 10% 10% 0 0 3 5 0 0 0
deviation
random
number for
normal normal each vehicle
distribution distribution according to
a normal
distribution
The available type of charge at lunch is standard charge, and at home, it is slow charge. We
consider that CVs has no charge point during work time.
Number of trips
For the PVs’ number of trips, we consider that we have three possibilities:
-the PV goes to work in the morning, moves at midday (2 trips), and comes back home
in the afternoon
-the PV goes to work in the morning, and comes back home in the afternoon
-the PV doesn’t move at all
For CVs, we consider that we have four possibilities:

-the CV has a trip in the morning, and one in the afternoon (only those vehicles can
charge with a standard charge if it is available at midday)
-the CV has only one trip in the morning that can last until the end of the day
-the CV has only one trip in the afternoon
-the CV doesn’t move for the entire day
It is assumed that the vehicles don’t move during the evening or the night.

All these possibilities are allocated with a percentage.

Table 15 - Allocation of number of trips
Number of trips PVs CVs

Morning 1 1 0 1 1 0 0
Midday 2 0 0 1 0 1 0
Afternoon 1 1 0 0 0 0 0
Evening 0 0 0 0 0 0 0
% of vehicles 30% 70% 0% 50% 40% 5% 5%
Trip length
Table 16 - Allocation of trip length for PVs Table 17 - Allocation of trip length for CVs
Trip Length PVs Trip Length CVs

(km) (km)
location Urban Rural location Urban Rural
mean value 26 28 mean value 100 100
median 8 10 standard
5 10
log normal distribution deviation
normal distribution
The trip length can then be reduced, as explained before.
Consumption
Table 18 - Allocation of consumption
Use PVs CVs
location Urban Rural Urban Rural
Consumption
12 18 15 20
(kWh/100km)
The electric consumption can then be reduced with a coefficient for some PHEVs as
explained before.
This coefficient of consumption for urban vehicle is 70%, and 50% for rural vehicles.
Departure time
All departure times (morning, midday, afternoon, evening) are random numbers according to
a normal distribution.
Table 19 - Allocation of departure times
PVs'mean PVs standard CVs standard
Departure time CVs'mean value
value deviation deviation
Morning 08:00 02:00 08:00 02:00
11:30 12:00 (charge time)
Midday 01:00 01:00
13:00 13:30
Afternoon 17:00 02:00 17:00 (charge time) 02:00

For CVs, the time allocated for midday (the first one) is a time to charge for CVs who can
charge at lunch time, and the allocated time for the afternoon, is also a time to charge.
If the vehicle has to start a new trip while driving, the departure time is increased until the
vehicle stops, and then can leave once again if the SOC allows it.
Initial SOC and State

We assume that all the vehicles have an initial SOC of 100% at 0:00.
We consider that all the vehicles are parked at the beginning of the simulation.
Other parameters
In the reference scenario we consider that the temperature has no impact on the electric
consumption and the battery’s efficiency.
3.6. Sensitivity Analysis
For some parameters, there are no available data. For instance, we don’t know if there will be
more EVs or PHEVs for a private use. So the idea is to simulate a fleet several times by
changing only one parameter each time, in order to see if the parameter has an influence on
the total load curve. The influence of the parameter will be described as the terms expressed
in Table 20.
Table 20 - Influence description
Percentage of change
Impact from the reference
scenario
small <30%
medium <50%
important <80%
huge >80%
3.6.1. Impact of the Battery’s Initial State of Charge
The initial SOC could influence the stability of the curve on several days.

Figure 16 - Load with an initial SOC of 100% for all the vehicles, for 3 days
On Figure 16 we can observe the load per vehicle for three days, with an initial SOC of 100%
for all the vehicles.
We can see that the curve is unchanged for the two following days, it is stable. However the
entire fleet with a totally full battery might not be realistic. Therefore several values for the
initial SOC have been tested, with a simulation period representing several days.
Table 21 - Different tested scenarios for the initial SOC impact
Initial SOC distribution
scenario 1 30% 100%
scenario 2 50% 100%
scenario 3 80% 100%
scenario 4 100% 100%
100% 70%
scenario 5
70% 30%
All the other characteristics for the fleet are identical for each scenario.
The five different scenarios are simulated 100 times, in order to compare the mean values of
each scenario, and the relative standard deviations.
Given that the vehicles have time to charge their battery before leaving for their morning trip,
the curves with different initial SOC are similar (Figure 17).

Figure 17 - Mean powers and Relative Standard Deviations for different initial SOC
We can see on Figure 17 that from 6:00am the first day, the curves for mean powers and
relative standard deviations are strictly identical and stable. So after one day of simulation, the
initial SOC has no impact on the load curve.
We can conclude that we can allocate the most realistic initial SOC for our simulation, and
simply simulate our fleet on 30 hours.
Figure 18 - SOC distribution function
On Figure 18, the initial SOC was 100% for 70% of the vehicles, and 70% for 30% of the
vehicles. The mean value of the SOC after two days of simulation, at midnight is 93%. The
mean value of the initial SOC was 91%.
We can assume that a fleet with 70% of the vehicles with an initial SOC of 100%, and 30%
with an initial SOC of 70% is then realistic enough.
Therefore, for all the other simulations, the initial SOC will be as in scenario 5 and the studied
period will be from 6am to 6am the next day.

3.6.2. Impact of the EV/PHEV Distribution
We currently cannot say if more EVs or PHEVs will be sold. Given that these two types of
vehicles have different battery size, it can have an impact on the load.
We will test four different scenarios, with different distributions for EVs and PHEVs. The
other characteristics are unchanged. The four scenarios are described hereafter (Table 22)
Table 22 - Different distributions for EVs and PHEVs
PV (80%) CV (20%) Total
Type PHEV EV PHEV EV PHEV EV
scenario 1 0% 80% 0% 20% 0% 100%
scenario 2 80% 0% 20% 0% 100% 0%
scenario 3 16% 64% 16% 4% 32% 68%
scenario 4 (reference) 64% 16% 4% 16% 68% 32%
A Monte Carlo analysis is performed. The four mean values for the different scenarios are
plotted on Figure 19, from 6am to 6am the next day (as explained in the paragraph on the
initial SOC impact). Scenarios 1 and 2 correspond to two extreme scenarios. In the first one,
all the vehicles are EVs and in the second one, all the vehicles are PHEVs. Obviously, since
the PHEVs’ batteries are smaller, the load for scenario 2 is less important than for the other
scenarios. However, the load for scenarios 3 and 4 are quite similar.
Figure 19 - Mean powers and Relative Standard Deviations for four scenarios (EV/PHEV)
The mean values vary between 1 30% kW/vehicle. We can say that the EV/PHEV’s
distribution has a medium impact on the total load.
However, the relative standard deviation can change with 60% at 20:00, and 25% at midday.
So this parameter has an important impact on the relative standard deviation at 20:00 and a
small impact at midday.
3.6.3. Impact of the CV/PV Distribution
We know that the CVs represent today 20% of the French fleet. However we currently cannot
say if this distribution will be the same for the PEVs. Six different scenarios have been tested.

The CV/PV’s distribution is changed, but also the EV/PHEV’s distribution, since they are
much correlated.
Table 23 - Different PV/CV distributions
PV EV PHEV CV EV PHEV
scenario 1 100% 100% 0% 0% 100% 0%
scenario 2 100% 0% 100% 0% 0% 100%
scenario 3 0% 100% 0% 100% 100% 0%
scenario 4 0% 0% 100% 100% 0% 100%
scenario 5 20% 20% 80% 80% 80% 20%
scenario 6 (reference) 80% 20% 80% 20% 80% 20%
Figure 20 - Mean Powers and Relative Standard Deviations for six different PV/CV distributions
The highest curve on Figure 20 represents the scenario 3, that is to say all the vehicles are
commercial (longer distances than for private vehicles) and totally electric (greater batteries).
The lowest curve represents the scenario 2: all the vehicles are private (shorter distances) and
plug-in hybrid (smaller batteries). These two curves represent the minimum and maximum
load we can have according to the PV/CV’s and EV/PHEV’s distributions.
We can see that the load curves are approximately 1 120% kW/vehicle or 1-50% kW/vehicle
around 20:00. The load curve is around 0.2±50% kW/vehicle at 12:00.
So the PV/CV can have a huge impact on the load at 20:00 and a medium impact at 12:00.
The relative standard deviations are quite similar (except for scenario 3&4 before 12:00, when
the mean power is almost equal to zero).
3.6.4. Impact of the Temperature
As explained before, the temperature can have an impact on the charge of the vehicles, and on
their electric consumption. Three different scenarios are then tested, in order to evaluate the
temperature’s impact on the load curve.

Table 24 - Three scenarios for the temperature’s impact

Temperature
coefficient
scenario 1 (reference) 1
scenario 2 1,2
scenario 3 0,8
The vehicles’ electric consumptions are multiplied by the temperature coefficients.
Figure 21 - Mean powers and Relative Standard Deviations for three temperature’s scenarios
We can see on Figure 21, that the mean power per vehicle is almost multiplied by 1.1 at the peak hour, for
scenario 2 (cold temperatures, winter days), and by 0.9 for scenario 3 (high temperature, summer days).
The relative standard deviations (
Figure 21) are also really similar. So the temperature has a small impact on the load curves
and on the relative standard deviations.
3.6.5. Impact of the Available Points of Charge Distribution
Fast Charge points for PVs
We don’t really know how many fast charge points will be available and if we can use it with
complete safety for the battery’s state. So the percentage of vehicles having access to a fast
charge point is unknown. That is why three different scenarios are tested, with different
percentages of private vehicles having access to a fast charge point at midday.
Table 25 - Fast charge points’ distributions
Use PV CV
Lunch:
Available type of Work: Standard Lunch: Fast Home: Slow Home: Slow
Standard
charge Charge Charge Charge Charge
Charge
scenario 1
20% 5% 100% 20% 100%
(reference)
scenario 3 20% 50% 100% 20% 100%
scenario 4 20% 100% 100% 20% 100%

However, only the PVs moving at midday can have access to a fast charge point. Besides,
they can start to charge to a fast charge point only if their battery’s SOC is lower than 80%.
Figure 22 - Mean powers and Relative Standard Deviations for three different fast charge points’
distribution
We can see on Figure 22 that the mean powers, corresponding to the three scenarios, are
similar, except around 12:00. At 12:00, the load is around 0.2±40% kW/vehicle, but still
represents less than 25% of the peak load at 20:00. The relative standard deviation is much
more important (60%) for scenario 1 at midday.
The fast charge points’ access has a medium impact on the load curve at midday, but a small
impact at 20:00.
Standard Charge points for PVs
Some companies can decide to help the PEVs’ development and install charge points at work
place. But we currently cannot say how many charge points will be available for PEVs.
In the three different scenarios presented hereafter, the percentage of PVs having access to a
standard charge point is changed.
Table 26 - Standard Charge points’ distributions for PVs
Use PV CV
Available Lunch:
Work: Standard Lunch: Fast Home: Slow Home: Slow
Type of Standard
Charge Charge Charge Charge
Charge Charge
scenario 1 0% 5% 100% 20% 100%
scenario 2
20% 5% 100% 20% 100%
(reference)
scenario 3 100% 5% 100% 20% 100%

Figure 23 - Mean powers and Relative Standard Deviations for three different standard charge points’
distributions for PVs
We can see (Figure 22) that in scenario 3 (all the PVs have standard charge points), the loads
during the morning and the afternoon represent 50% of the load at 20:00 whereas in scenario
2 (20% of PVs have standard charge points), it represents less than 10% of the peak load. The
load during the morning can vary from 0kW to 0.4 kW according to the standard charge
distribution at work for PVs. So this repartition has a huge impact on the total load during the
afternoon and the morning. The peak at 20:00 can be 30% less important. It has a medium
impact at 20:00.
The relative standard deviation is less important (50%) for scenario 1 during the evening. It is
obviously more important during the morning and the afternoon, since the standard deviation
is divided by a mean power equal to zero.
Standard Charge points for CVs
It is assumed in the reference scenario that some commercial vehicles can have access to a
standard charge point at midday. However the percentage of CVs that can charge at midday
with such a charge is obviously unknown. The three different scenarios, described in Table
27, have different percentages of CVs having access to a standard charge point at midday.
Table 27 - Standard Charge points’ distributions for CVs
Use PV CV
Available Lunch:
Work: Standard Lunch: Fast Home: Slow Home: Slow
Type of Standard
Charge Charge Charge Charge
Charge Charge
scenario 1
20% 5% 100% 20% 100%
(reference)
scenario 2 20% 5% 100% 0% 100%
scenario 3 20% 5% 100% 100% 100%

Figure 24 - Mean powers and Relative Standard Deviations for three different standard charge points’
distributions for CVs
The standard charge points’ distribution has a small impact on the load for the evening.
However, the mean load around 1pm for scenario 3 (all the CVs with standard charge points)
can represent 35% of the peak load. For the other scenario it is around 10% of the peak load.
The load around 1-2pm is around 0.1+250% kW/vehicle or 0.1-50% kW/vehicle. Besides, the
relative standard deviation of scenario 3 represents 150% of the relative standard deviation of
scenario 1.
So the standard charge points’ distribution has a huge impact on the load and the relative
standard deviation at 12:00.
Fast Charge points for CVs
It was assumed in the reference scenario that some CVs can have access to a standard charge
point at midday. But we can assume that instead of having access to standard charge points at
midday, the CVs can have access to a fast charge point. The scenarios are described hereafter
in Table 28.
Table 28 - Fast Charge points’ distributions for CVs
Use PV CV
Available Work: Standard Lunch: Fast Home: Slow Lunch: Fast Home: Slow
Type of Charge Charge Charge Charge Charge
Charge
scenario 1 20% 5% 100% 20% 100%
scenario 7 20% 5% 100% 0% 100%
scenario 8 20% 5% 100% 100% 100%

Figure 25 - Mean Powers and Relative Standard Deviations for three scenarios for Fast charge points’
distribution for CVs
The fast charge points’ distribution has only a small impact on the peak at 20:00. However,
for scenario 8 (all the CVs with fast charge points), the peak at 12:00 represents around 45%
of the peak at 20:00, whereas for the two other scenarios, it is less than 10%. If we consider
scenario 1 as a reference scenario, the load at 12:00 can increase with 300%, or decrease with
50%. So the CVs’ fast charge points’ distribution has a huge impact on the midday peak.
Besides the relative standard deviation can increase with 30%, or decrease with 40% at
midday. The impact is important then.
Another possible consequence is that an easy access for CVs to fast charge point at midday
could have an impact on the development of the number of vehicles.
3.6.6. Assumption of a Systematic Charge
For all the results presented before, we assumed that the PEV’s user charges his battery as
soon as he can, each time he has access to a charge point. However, the user’s behavior will
mainly depends on the batteries manufacturers’ advices.
We know that the battery’s span of life represents a maximum number of cycles. However,
deep discharges are not advisable, but the user doesn’t care about it if he doesn’t own the
battery. And users can be much stressed to run out of battery.
That is why we assumed that the users charge whenever they can, even if the battery is still
full enough to run on it during an entire day.
However, we can imagine that some users will charge only if they have not enough electricity
to run on the battery until the end of the day. And during the night, they will charge only if the
battery cannot cover the whole day’s trips.

Figure 26 - Mean Power for systematic and not systematic charges during five days
We can observe that the first day, the total load is less important than the other days when the
charge is not systematic at 20:00; however this is due to the allocated initial SOC. But then it
oscillates around the load for a systematic charge. Besides, the load during the day is much
less important than for a systematic charge. Indeed, the users can drive on their battery for the
whole day, so they wait for the evening to charge their battery. At 20:00, the assumption of a
non systematic charge as a small impact on the load. During the day, the load can decrease of
70% from the reference scenario (systematic charge). So the impact during the day is
important.
Figure 27 - Relative Standard Deviations with and without systematic charge

During the morning and the afternoon, the relative standard deviation is 10% higher than with
a systematic charge, that is to say an increase of 60%. At 20:00, it is almost the same. So the

impact on the relative standard deviation of a non systematic charge is small at 20:00 and
important during the day.
3.6.7. Battery Exchange
Previously, it was assumed that the user charges his battery when he has this possibility, and
when he has time to do so. However, we could assume a development of battery exchange
stations and the user doesn’t charge his battery during the day anymore, but exchanges it
whenever he needs (project Better Place in Israel and Denmark
[http://www.betterplace.com]).
It is possible to simulate such a project with the previous model assuming that all the vehicles
have access to a fast charge point. Indeed, it is assumed here that the battery exchange stations
will recharge the batteries with fast charge in order to avoid having a large number of stocked
batteries. When the client comes to exchange his battery, the station starts charging it directly
with a fast charge.
The covered distance for the whole day is not reduced anymore according to the battery range
as it was explained before since the driver can exchange his battery in the numerous exchange
stations whenever he needs. Nonetheless the trip length for EV is reduced so that the user can
cover at least one trip. They exchange their battery only if the battery’s SOC is lower than
30%.
However they keep on charging their battery at home with slow charge whatever their
battery’s SOC is. Indeed they don’t care about the battery life span, since they don’t own the
battery.
We can assume that this simulation is valid for EVs and PHEVs. PHEVs will certainly not
exchange their battery however, if these exchange stations are available, we can assume that
they will offer fast charge points for PHEVs. So instead of exchanging their battery, they will
charge it with a fast charge, thus it is exactly the same simulation as for EVs.
Figure 28 - Load from one simulation with battery exchange

Given the short duration of a fast charge, the load with the battery exchange simulation is less
smooth than with the reference scenario.

Figure 29 - Mean load and relative standard deviation with battery exchange
We can see that the load is much more important during the afternoon than it was without
battery exchange (300% higher). The load is just a little bit lower during the evening.
The relative standard deviation is higher during the afternoon (20%) than without battery
exchange, but lower with 20% during the evening. The impact on the relative standard
deviation of the battery exchange assumption is huge during the morning and small during the
rest of the day.
The impact of the assumption of a battery exchange is small at 20:00 and huge at midday for
the load. Besides, the required energy is approximately equal to 9 kWh in the case of battery
exchange, whereas in the reference case, the required energy is around 8 kWh.
But this simulation doesn’t take into account the fact that if such infrastructures are
developed, more users will buy PEVs. Indeed the battery range constraints would be reduced
since it will be easy to exchange his battery when he needs it. So the drivers with longer
distances could buy PEVs, and PEVs could be used for other trips than commuting trips. So
the electric fleet could be more important.
3.6.8. Impact of the Other Parameters
The other parameters can also have an impact on the load curve. The load is much more
spread out for instance if the departure times (and then charge time) are outstretched.
Moreover the impact of the initial SOC can also depend on the departure time distribution.
The battery size determines the load’s level, as well as the trip length. However it is not
possible to check as many scenarios as this would require in the time allocated for this master
thesis. That is why the more realistic values for these parameters have been allocated. These
values are based on the bibliographical study.
3.6.9. Assumptions’ Impact Recap and Retained Reference

Scenario
Extreme values have been taken for different parameters. The impacts of the parameters on
the load curve are summed up in the tables hereafter.

Table 29 - Parameters’ influence recap

PVs fast CVs fast
CVs standard
Impact charge charge EV/PHEV CV/PV Temperature
charge points
points points
20:00 small small small medium huge small
Mean power
12:00 medium huge huge small medium small
Relative 20:00 small small small important small small
standard
deviation 12:00 important huge huge small huge small
Table 30 - Parameters’ influence recap

PVs standard Non systematic Battery exchange
Impact
charge points charge
morning and
huge important huge
Mean power afternoon
20:00 medium small small
Relative morning and

huge important huge
standard afternoon
deviation
20:00 medium small small
We can imagine that the impacts at 20:00 will have more consequences on the load curve than
at 12:00. So in this case, the assumption on PV/CV’s distribution is important. However, in
the reference scenario was allocated a distribution according to the current division between
CVs and PVs. So this distribution can be kept as in the reference scenario.
Therefore, all the fleet’s characteristics described in the reference scenario are kept, except for
the initial SOC. 70% of the fleet will have 100% of SOC, and 30% of the vehicles will have
70% of SOC.
In the reference scenario, it is assumed that the users have no trip during the evening. Only
few of them have access to a charge point during the day. The fast charges are assumed to be
possible without damages thanks to battery improvements. The users charge systematically
their vehicle. This scenario corresponds only to a workday.
3.7. PEVs’ “Natural” Charge Curve
The word “natural” means in this context without any regulation. If no incentives come from
the authorities, the load curve resulting from the PEVs’ charge might look like on Figure 30.
With the reference scenario, the mean load is around 1 kW/vehicle at 20:00.

Figure 30 - Mean power for the reference scenario
The loads resulting from different kinds of charge (slow, standard and fast charge) are also
represented. The fast charge is almost invisible since very few users can have access to a fast
charge point at midday and thanks to a smoothing effect due to the short duration of this
charge. The standard charge is mostly used during the morning and the afternoon, and the
slow charge is only used during the evening and the night. The total consumed energy per
vehicle is around 8 kWh/day.
This type of curve has then to be added to a consumption curve in order to know the possible
impacts on the electric balancing system.

) * ! !&
4.1. Future “Natural” Consumption Curves

4.1.1. Background
Nowadays, our society is strongly dependent on electricity. Hence, the electricity power
system has to be highly reliable. In order to ensure the security of the electricity network, the
balance between production and consumption must be kept all the time. Since it is not easily
and economically possible to store electricity during a period of low demand to use it during a
period of high demand, France’s electricity generation capacity has to be higher than the peak
loads.
The electricity consumption fluctuates with the time, according to the users’ needs. These
needs follow the economical and home’s activities which depend on the seasons’ cadence.
The consumption is higher during the day than during the night, during workdays than week-
ends, in winter than in summer. It can vary a lot with the temperature, due to air-conditioning
and especially heating.
Two examples of a French load curve are presented on Figure 31.
The winter day is on January 20th- 21st 2009 (Tuesday and Wednesday). The summer day is
on June 24th-25th 2009 (Wednesday and Thursday). These two curves correspond to work
days.
Figure 31 - Consumption curves for a winter and summer workday
The load increases rapidly from 5:00am due to the private activities’ increase, as well as the
increase of rail, industry or service activities. Then in the late afternoon, it starts increasing
again because of the rail transport consumption peak and the residential consumption is added
to the end of the day activities. We can also observe a peak at 23:00. This is due to electric
water heaters. These devices are automatically set on by a tariff signal system. Those signals

are spread out between 22:00 and 24:00, but the simultaneous operation of those electric
water heaters induces a load rise.
4.1.2. Consumption by 2015, 2020 and 2025
The load curves evolve naturally with the yearly energy consumption’s increase. However the
load curves are slightly distorted with the years because some uses grow in different ways,
different uses or regulations can appear. For instance, new European regulations for lighting,
the heat pumps’ development, and of course the PEVs’ charge can modify a lot the future load
curves.
The profiles used for this analysis come from RTE’s studies. The consumptions by 2015,
2020 and 2025 were studied according to the evolution, apparition of some uses.
Figure 32 - Load in 2009, 2015, 2020 and 2025

On Figure 32 examples of load curves in 2015, 2020 and 2025 without PEVs charge are
plotted for a workday. The black curves represent the load for a day in January and June 2009.
They are the same curve as Figure 31. The future consumption increases much more for the
morning than for the peaks in the evening. However, those curves correspond only to curve’s
profile, and represent a given day, so a precise comparison of these curves with a curve from
2009 wouldn’t be appropriate.
4.1.3. Consumption with PEVs
RTE has studied two different scenarios for 2015, 2020, 2025. In a reference scenario, the
number of vehicles is around 500,000 PEVs in 2015, 1 million in 2020 and 2.7 millions in
2025. According to a DSM scenario (Demand Side Management),, the number of vehicle in
2015 can be 1 million, 3.5 million in 2020, and 6.7 millions in 2025 in France [RTE (2009)].
These are figures used as assumptions for official long term studies (Figure 33).

Figure 33 - PEVs’ and electricity demand’s evolution [RTE (2009)]

The figures below correspond to mean consumption curves. The mean values of
consumptions during workdays (Tuesday, Wednesday, and Thursday) in January and in June
have been calculated for 2015, 2020 and 2025. The mean loads due to PEVs have been
multiplied by the forecasted number of vehicles and added to these consumption curves. For
the winter day, the temperature coefficient for the PEV’s consumption is taken equal to 1.2.
Figure 34 - Total load curve with PEVs in 2015
We can see (Figure 34) that with 0.5 or 1 million vehicles in 2015, the PEV’s impact is really
small. Indeed, at the peak at 20:00, we “only” add 1,000 MW with 1 million vehicles.

However, we begin to see an impact in 2020 with 1 and 3.5 millions vehicles. The peak
increases with 4% with 3.5 millions vehicles in 2020 in winter. It can increase with 5% at
18:00 in summer.
Figure 36 - Dynamics of the load curve with and without PEVs in 2020
We can see on Figure 36, that the dynamics of the curves can be changed between 3 and 6
pm. It is increased with 150MW/10min in summer at 17:00, or with 160MW/10min in winter.
In 2025, the profile of the load curve is totally distorted with the introduction of PEVs.
Indeed, we can see on Figure 37, that for a summer day, the load was more important during
the morning. But with PEVs, there is a peak of consumption at 18:00. The maximum load was

66 GW during the morning in summer. But with 6.7 millions vehicles, the maximum load
would be of 68 GW at 17:00. The load increases with approximately 10% at 18:00 in winter
and in summer.
The impact on the dynamics of the load curve is also huge. Indeed at 18:00, the dynamics was
around 100MW/10min summer. But with PEVs, the dynamics is approximately equal to
400MW/10min. The load increases instead of remaining constant. In winter the dynamics of
the load curve increase also with 300MW/10min.
Figure 38 - Dynamics of the load curve with and without PEVs in 2025
Obviously, the curves presented are only assumed mean loads but this can give an idea of the
impact on the load curve in general. The impact in 2015 seems to be very small given the low
number of PEVs in traffic. However we can observe an increase of the load peak from 1

million vehicles in 2020. In 2025, 2 to 6 GW can be added to the consumption at peak hours,
therefore, the curves can be distorted, and the dynamics increased. Thus the margins might
need to be increased. In the next parts, we will try to simulate some possible solution in order
to decrease the PEVs’ impact during peak hours.
4.2. Tariff Signals
4.2.1. Tariff Signals’ Principle
In order to avoid that all the users charge their vehicles when coming back to home, when the
French electricity consumption reaches a peak, we can imagine that some users can use
electricity meters. These meters can have relays to turn off nonessential equipment, such as
electric water heaters, or in our case, the charge of the battery during peak hours. This
solution already exists and is quite simple. Nowadays, 40% of French consumers have a tariff
signals regulation. However this system can create load peaks if the tariff signals are
synchronous for many consumers. That is why the signal tariffs are distributed on several
hours [http: //www.erdfdistribution.fr].
The principle of the tariff signals is explained on Figure 39.
Public Low
network voltage
P (kW)
t0
P (kW)
Tariff Signal
t0
Low
voltage
P (kW)
Tariff Signal t0
t0 PEV connection time
total load without tariff signal

Total load
load with one tariff signal
load with two tariff signals
off-peak
Peak period
period
Figure 39 - Tariff Signal’s principle

To simulate a possible tariff signal, it is allocated at each vehicle a different time slot. When
we are not on the given time slot while in the late afternoon, evening or night, the charge of
the battery is turned off.

Some of them have no tariff signal at all. The others are equally distributed between the six
following time slots: 21:00 to 5:00, 21:30 to 5:30, 22:00 to 6:00, …, 23:30 to 7:30
They are allowed to charge during these time slots. In the flowchart (Figure 11), the box
‘Boolean = true’ is changed. The Boolean is true only during off peak hours for the vehicles
having tariff signal regulation.
During daytime, we assume that there is no tariff signal (some areas use tariff signals during
daytime but as we expect that it won’t be efficient for PEVs, we choose only night period for
tariff signal in the simulation).
Charge
Consumed power when

running
Figure 40 - Load and consumed power for two vehicles
We can see on Figure 40 that the user of vehicle 1 starts his trip at 5:30, moves at 10:30 and
13:00. Then he comes back home at 16:30 and charges directly his battery. Vehicle 1 has no
tariff signal.
The user of vehicle 2 goes to work at 6:00. He charges his battery when he arrives at
workplace, and then comes back home at 16:30. He is under restraint of a tariff signal, so his
battery only begins to charge at 21:00 in this case.
4.2.2. Load Curves and Total Consumption Curves with Tariff

Signals
Different scenarios with different percentages of PEVs having tariff signals have been tested
(Table 31).
Table 31 - Three scenarios for tariff signals distribution
no regulation tariff signals

scenario 1 60% 40%
scenario 2 80% 20%
scenario 5 20% 80%
On Figure 41 we can see the mean powers and the relative standard deviations for each
scenario.

Figure 41 - Mean Powers and Relative Standard Deviations for three different scenarios for tariff signals
We can observe that for scenario 2, the peak load at 20:00 is only 0.8 kW/vehicle that is to say
around 70% of the peak without regulation. For scenario 1, the peak is put back at midnight,
with approximately 0.9 kW/vehicle (80% of the peak load without regulation). However for
scenario 5, the peak is higher (130% of the peak load without regulation). But this peak is put
back at midnight. Actually, the peak is higher because the charge times are concentrated on a
smaller period (21:00 to 23:30) than without tariff signal, since the arrival times are much
more spread out (see Appendix Figure 80).
The relative standard deviations are quite similar.
So the percentage of vehicles having tariff signals has a huge impact on the load curve.
On Figure 42 has been plotted the mean consumption loads in 2020 with PEVs with 40% of
them with tariff signals.
Figure 42 - Total load in 2020 with PEVs and tariff signals (40%)

We can see that the load is more important during the night than it was without tariff signals,
and the load peak at 6pm is less important with tariff signals. The impact of the PEVs is
decreased with 1 GW.
Figure 43 - Dynamics of the load curve with and without PEVs in 2020, with tariff signals
The dynamics of the load curve with and without PEVs are more similar with tariff signals
than they were without tariff signal. However, the charge is still over before the moment when
the total consumption is less important.
However, if only 40% of the vehicle’s users have tariff signals regulation, the impact on the
load curve is still present.
On Figure 44 a mean load curve with 80% vehicles having a tariff signals regulation has been
plotted. We can see that on this example, the impact of the PEVs’ charge is not visible at peak
hours. But in summer the peak load at 22:00 is much more important.
We can see that the load due to the PEVs can be put back after the peak with the tariff signals
regulation. However, we can have another peak at 22:00, when all the vehicles are charging.
This is due to the tariff signals distribution, and the time slots allocated by the tariff signals.
However, 80% having tariff signals regulation is a high percentage hard to achieve.
Other solutions will then be tested in order to see if they can be more efficient.

Figure 44 - Total load in 2020 with PEVs and tariff signals (60%)
4.3. Battery Management System
4.3.1. Battery Management System Principle
Battery Management Systems (BMS) can have different significations. But in our case, we
suppose that such a system can monitor the charge of the battery so that the user is certain to
have a fully charged battery for a scheduled departure time. In our model, the charge of the
battery is then cut in several periods during the available time to charge.

Public Low
network voltage
P (kW)
t0 t1
P (kW)
t0 t1
Low
voltage
P (kW)
t0 t1
t1 next trip's time
total load without BMS

Total load
Load with BMS
total load with BMS
Figure 45 - Battery Management System’s principle

Given that the battery needs 2.5 hours to charge from 80% to 100%, we are sure that 2.5 hours
before leaving, the battery will have at least a SOC of 80%.
In the simulation, the block ‘Boolean = true’ is changed: the Boolean is equal to true during
certain periods in the time slot (from arrival time to departure time) for the vehicles having
BMS.
Different BMSs have been simulated. In the first simulation, the number of periods in which
the charge is cut is fixed. In another version, the number of periods can vary from 2 to 12. In a
third version, the period can be fixed to one hour, and the number of period can be deduced
from it. In a last version, the battery keep on charging during the allocated time, but with a
lower power, so that the battery is full only at the end of the allocated period.

Charge
Consumed power when

running
Figure 46 - Power for two vehicles with and without BMS

We can see on Figure 46 two vehicles’ needed or consumed power. The first one has no BMS,
and the second one has a BMS with a fixed period length (one hour).
Vehicle 1 starts its trip at 9:00, and comes back home at 18:00. Then the battery directly starts
charging, without BMS.
Vehicle 2 leaves at 9:00 and starts coming back at 15:00. Then we can see that the charge of
its battery is spread out during the night, from its arrival until the time to leave in the morning.
4.3.2. Load Curves and Total Consumption Curves with BMS
Figure 47 - Load with different BMSs and without BMS

We have simulated a fleet with different BMSs for 40% of the vehicles, and the same fleet
without BMS.

We can observe on Figure 47 that the load is much more spread out during the night. We still
have a consumption peak at 20:00, but this peak is 25% lower than the peak without BMS.
Besides, we can see that the load with BMS with a fixed number of periods, or a variable
number of periods, the curves are not smooth. But the curves with BMS, for a same fleet are
similar.
We have then simulated the fleet with different distributions for BMS. The simulations were
performed with a BMS with a fixed number of periods.
Table 32 - Three scenarios for BMS’s distribution
no
regulation BMS
scenario 1 0% 100%
scenario 2 50% 50%
scenario 3 20% 80%
Figure 48 - Mean powers and relative standard deviations for three scenarios for BMS’s distribution
We can see on Figure 48 that the peak load at 20:00 can be decrease with 40% (if all the
vehicles have a BMS). The peaks are still at 20:00, except for scenario 1 whose peak is put
back at 22:00. The relative standard deviations are similar.
A mean load has been added to consumption curves in 2020. 40% of the vehicles have a
BMS.

Figure 49 - Total load in 2020 with PEVs and BMS (40%)
We can see that the load is more spread out than without BMS, but in this example, the
advantages of the BMS is not clearly visible since it decreases the PEVs’ impact at the peak
hours with only less than 1GW.
Figure 50 - Dynamics of the load curve with and without PEVs in 2020, with BMS (40%)
The dynamics can be 100MW/10min higher than without PEVs’ charge.

Figure 51 - Total load in 2020 with PEVs and BMS (80%)

We can see on Figure 51 that even with 80% of the fleet having a BMS, the impact on the
load curve is still well present. This reduces the impact with approximately 2 GW but still
have an impact around 2 GW (with 3.5 million vehicles).
We can imagine another BMS system that can smooth the consumption curve, filling the
hollow. In other words, the vehicles would be allowed to charge only from 0:00 to 5:00, when
the consumption curve is the lowest. But this depends on the time the vehicles need to charge.
We can determine when the vehicle should start charging in order to have a full battery at a
given time. The time when the charge has to be over (tend) is distributed among the vehicles
having this regulation. The six different possibilities for tend are: 3:30, 4:00, 4:30, 5:00, 5:30
and 6:00. Then the time when the charge has to begin (tstart) can be deduced from the required
time to charge.
In the simulation, the “Boolean block” is changed. The Boolean is only equal to true when the
time is between tstart and tend for the vehicles having the BMS regulation when the vehicles are
at home. For the other ones, the Boolean is always equal to true.
We can imagine that the TSO sends a signal to notify when the charge should end, and the
BMS calculates when the charge has to begin.
Table 33 - Different scenarios for BMS bis
no regulation BMS bis

scenario 1 60% 40%
scenario 5 20% 80%

Figure 52 - Mean powers and relative standard deviations for different distribution for BMS bis
We can observe that with scenario 1 (40% with this BMS), the peak lasts a longer time, and is
lower than without BMS (around 0.65 kW/vehicle). If 80% of the fleet have this BMS
(scenario 5), the peak can be as high as without regulation (1 kW/vehicle) but this peak is
around 3am. The relative standard deviations are quite similar, except for scenario 5, the
relative standard deviation can be 5% higher between 9pm and midnight.
Figure 53 - Total load in 2020 with PEVs and BMS bis (80%)
We can see here an extreme scenario, where 80% of the vehicles have a BMS. The load is
only from 22:00 to 5:00. It fills in the hollow, but not correctly. Besides, we don’t have the
phenomenon of another peak after the peak hours; the load is more spread out during off peak
hours.
But the solution is still not optimum.

4.4. Battery Management System and Tariff Signal
The idea is to combine the advantage of the tariff signals (to put back the peak) with the
advantage of the BMS (lower the peak). The BMS can be performed only during off peak
hours.
In the simulation, the box ‘Boolean = true’ is changed: the Boolean is equal to true during a
certain period in the time slot allocated by the tariff signals for the vehicles having both
regulations.
On Figure 54, we can see a simulation for a fleet without regulation, a fleet with 80% BMS, a
fleet with 80% tariff signals, and a fleet with 38% with tariff signals combined with BMS and
42% with only tariff signals.
Figure 54 - Load with different regulations

We can observe that the load peak with BMS combined with tariff signals is indeed put back
to midnight; it is lower than with only tariff signal, but still a bit higher than without any
regulation.
Five different scenarios with different distribution for the regulations have been tested.
Table 34 - Five different scenarios for BMS & Tariff signals distributions
BMS
No combined Tariff
regulation with tariff signals only
signals
Scenario 1 0% 100% 0%
Scenario 2 100% 0% 0%
Scenario 3 20% 80% 0%
Scenario 4 20% 40% 40%
Scenario 5 0% 80% 20%

Figure 55 - Mean powers and relative standard deviations for five different distribution for BMS
combined with tariff signals
We can observe that the load curves with regulation are not smooth, but oscillate. In this case,
the oscillations are not with high amplitude, however this could create some problems for the
system is the load varies rapidly with high amplitude. In this case, BMS with a fixed number
of periods was used.
The impact of the number of periods on the oscillations was studied. The worst case was with
8 periods, that is to say, the cases represented on Figure 55. The amplitude of the oscillations
can be around 0.2 kW/vehicle. Therefore, with 1 million vehicles, this amplitude can be of
200 MW.
The principle of the BMS will have to take into account that in some cases, it can create some
oscillations, if the vehicles have similar behaviours.
The load at 20:00 for scenarios 1 and 5 is equal to 0. That is why the relative standard
deviations are so high. However, the load with regulations are quite similar (scenario 2
corresponds to no regulation).
So the impact on the mean load curve of introducing a regulation or not is huge. However the
impact of the different distributions for the regulation is small.
A load curve resulting from a fleet’s simulation has been added to a consumption curve in
2020 (Figure 56).

Figure 56 - Total load in 2020 with PEVs and BMS & Tariff Signals (40%)
To obtain these curves, 20% of the fleet have BMS combined with tariff signals, 20% only
tariff signals, and 60% have no regulation at all.
At 18:00, the load increases only with 2 GW instead of 3GW. The PEVs’ charge stops around
midnight without regulation, whereas it stops only around 3am or 6am with regulation.
Therefore, despite the regulations, the peak load is still increased at 18:00.
Figure 57 - Total load in 2020 with PEVs and BMS & Tariff Signals (80%)
If 80% of the fleet has the two combined regulations, the PEV’s impact in 2020 is much
smaller. But once again 80% is a very optimistic figure.

In the next part, we will study if the PEVs can help the system during peak hours.
4.5. Injections into the Grid
A PEV can also be seen as a mean to help the electric system when needed. Indeed, as a PEV
can store energy, it can release it to the grid at peak load in order to decrease the peak. Some
complex systems have been studied, where a PEV can support the electric grid to provide
many services, including maintaining the grid frequency. For instance, the PEV charges when
the frequency is too high, and releases energy to the grid when the frequency is too low. This
is called Vehicle-to-Grid (V2G). The study of such a complex V2G system is out of scope of
this master thesis. However a small model of injection into the grid during peak hours has
been developed.
The principle is that a PEV with a tariff signal regulation or a BMS combined with tariff
signals will inject energy into the grid during peak hours in the evening. Indeed if the vehicle
is connected during peak hours in the evening, and if it has a tariff signal regulation, this
vehicle won’t charge. But the idea is to use this connected vehicle to support the system
during the peak hours. So if the battery’s SOC is high enough, the vehicle will start to deliver
energy to the grid, until the SOC reaches a minimum level. This simulation is possible if we
consider that the users plug systematically their vehicles. Indeed if we assume that a PEV’s
user plugs his vehicle only when the battery’s SOC is not sufficient to cover the next trips,
then we can guess that the vehicle won’t inject into the grid since the SOC won’t be high
enough. So in this context, the assumption of a systematic charge or not has an important
impact on the load curve.
The efficiency of the cycle is taken equal to 100%.
A Monte Carlo analysis has been performed, in order to know the mean load when there are
injections into the grid, and the mean injected power. The reference scenario is used in this
analysis, with 40% having tariff signals or BMS combined with tariff signals. If the battery’s
SOC is more than 50%, then the vehicle can inject 1kW into the grid.
Figure 58 - Mean powers with and without injection

We can see (Figure 58) the load minus the injection into the grid. The peak is put back at 0:00
as expected, thanks to the BMS combined with tariff signals advantages. Besides, this peak is
lower than the peak without injection and regulation at 20:00. Obviously, the peak is higher
than without injection, since batteries are more depleted due to the injections.
Different scenarios with different percentages of vehicles allowed to inject into the network
were then studied.
Table 35 - Different scenarios for injection into the grid
BMS
No combined Tariff Injected
SOC limit
regulation with tariff signals only power (kW)
signals
scenario 1
60% 20% 20% 50% 1
(reference)
scenario 9 0% 50% 50% 10% 3
scenario 11 0% 50% 50% 10% 1
scenario 13 0% 50% 50% 30% 3
Scenarios 9, 11, and 13 are extreme scenarios. There will certainly never be 100% of the fleet
with regulation, and the SOC limit is too low for scenario 9 and 11. However, these scenarios
were simulated in order to know the maximum power that could be injected into the grid.
Figure 59 - Mean power and Mean injected power
We can see that with an extreme scenario such as scenario 9, the load can increase until 2
kW/vehicle at midnight.
On Figure 59 is represented the mean injected power per vehicle. We can see that it can reach
0.1 kW/vehicle at 20:00, that is to say 10% of the load at 20:00 with scenario 1, which is the
most likely scenario. It could increase until 50% of the peak load, and maximum to 90%.
On Figure 60 is represented a mean load curve with 40% of PEVs who can inject energy into
the grid during peak hours.

Figure 60 - Total load curve in 2020 with PEVs and injection into the grid (scenario 1)
We can see that the load that was higher than 90 GW without injection in winter, is now kept
under 90 GW. The impact of one million vehicles is almost half less visible during peak
hours. Compared to tariff signals, BMS and BMS combined with tariff signals, the injection
combined with BMS & tariff signals is more efficient at the peak hours. However, with only
40% of vehicles who can inject into the grid, the overload due to PEVs is still present at the
load peak.

Figure 61 - Total load curve in 2020 with PEVs and injection into the grid (scenario 13)
With an extreme scenario such as scenario 13, the total consumption curve would look like
Figure 61. The total load is decreased at actual peak hours. But another peak is introduced in
the system. Thus it’s clearly not a good solution. Besides around 21:30, the dynamics of the
curve becomes really important (Figure 62), and the total load curve is much higher than it
would be without injections. This abrupt increase of the load can require more important
margins’ sizes. So the injections into the grid can be helpful only if the number of vehicles
injecting into the grid is reasonable.
Figure 62 - Dynamics of the load curve with and without PEVs with injections

4.6. Other Solutions
A solution of cut-off injunctions can also be imagined in order to decrease the impact of the
PEVs at the load peak. The TSO can send down regulation signals to an aggregator. This
aggregator can control the PEVs and prevent them from charging when receiving such a
signal. Only PEVs that won’t be needed for a while can answer to this order. The principle of
such a regulation is explained on Figure 63.
Public Low
network voltage
P (kW)
t0
P (kW) t0
Low
voltage
P (kW)
t0
down
regulation
Information
Aggregator
Total load
cut-off
injunction
flexible
not flexible
total load without tariff signal
EV's load curve with cut-off injunction
Total load curve with cut-off injunctions
demand control system
Figure 63 - Cut off injunction’s principle

The aim of the simulations is to know the possible duration of the cut-off injunctions, and the
power that could be cut-off without disturbing the final charge.
Thanks to the previous simulations, we have got some information on the PEVs’ fleet. We
know the total consumed energy (around 8kWh/vehicle/day) and the distribution of the
required time to charge (Figure 64).
We can see (Figure 64) that more than 30% of the fleet need more than 5 hours to charge.
More than 40% need between 4 and 5 hours. And less than 30% need less than 4 hours.

Figure 64 - Mean required time to charge

A Monte Carlo analysis has been performed to know the number of vehicles connected at
home in order to know the available power. We assume that the connected vehicles draw off
around 3kW. Therefore, on Figure 65, we can see that between 21:00 and 3:00, 3kW can be
cut off.
Figure 65 - Mean Available Energy

Thus it is possible to send cut-off injunctions to an aggregator for instance, and this
aggregator will order cut off to the vehicles according to the different data described above,
and the departure time for each vehicle.
For a future work, a program simulating down regulation signals and the resulting charge of
the vehicles could be done.
Obviously, the different presented regulations depend on the social acceptance and utilisation.
Indeed this could induce some possible constraints. Besides the manufacturers need the total
buyers’ acceptance to provide such services.

4.7. Extreme Scenarios
The study of the impact on the load curve wouldn’t be complete without the research of the
worst and optimum cases for the electric system.
4.7.1. The Worst Case
We can imagine that the worst solution would consist of fast charges during peak hours.
Therefore, the model has been simulated with a scenario that allows fast charges at anyplace
and anytime of the day. So even during the evening, the vehicles charge with a fast charge
point. This scenario is not realistic on an infrastructure point of view but it can give a good
idea of the worst case for the electric system.
Figure 66 - Mean load with fast charge point at anytime of the day
We can see that with this scenario (Figure 66), we have now two peaks. One at midday with
0.8 kW/vehicle and one at 18:00 with 1.1 kW/vehicle. The evening peak is sooner than the
reference scenario, and synchronous with the evening consumption peak. The peak load at
midday is also synchronous with the consumption peak at midday in summer.
On Figure 67, we can see consumption curves in 2020 with this scenario. With 3.5 millions
vehicles, the consumption can be increased with approximately 4 GW at 18:00. At midday,
the consumption can increase with 3 GW. We have no charge at all during the night when the
consumption is minimal.

Figure 67 - Consumption curves in 2020 with 100% of the fleet having access to fast charge points at
anytime of the day
Thus, this is the kind of scenario that we should avoid for the electric system. On the contrary,
we should try to reach an optimum solution for the PEV’s charge.
4.7.2. The Theoretical Optimum Cases
4.7.2.1.Optimisation of the Load Curves
The solutions presented before don’t “fill in the hollow” correctly. The idea would be to
achieve to get a consumption curve as in Figure 68, with a constant minimum power during
off peak hours.

Figure 68 - Optimum consumption profile

The load profile can be deduced from the optimum consumption curve (Figure 69).
Obviously, given that the consumption curves are not identical for a winter day and a summer
day, the optimum load profiles are different according to the season.
Figure 69 - Load profiles for optimum case in winter and summer

To get this profile, we can calculate:
-Econs: the total energy from 0:00 to 6:00 without PEV’s charge
-EPEV: the energy consumed by PEVs at home (from 16:00 to 6:00)
Then E, the total energy is equal to: E = E cons + E PEV
We want P, the total power to be constant from 0:00 to 6:00.
E
So P (t ) =
6
And then the load can be deduced by: PPEV (t ) = P(t ) + Pcons (t )
with Pcons, the consumed power without PEV’s charge.

But then, with this reference load profile, we want to optimize the difference between this
reference and the real load.
We actually want to optimize the charge placement.
We can assume that the fleet is divided into three different parts, according to the required
charge time (Figure 64).
Figure 70 - Classes of load curves of PEVs according to their required charge time
And these different charge blocks have to be placed into the optimum load profile in order to
minimize the area between the load profile, and the area formed by the blocks, under the
constraints of the departure times.
Figure 71 - Optimal load placement

Let P(t) be the optimal limit for the load, and ci(t) the load of a vehicle i.
The function that we want to minimize is:
P(t ) − ci (t )
t i
Subject to conditions on the end of the charge time
An illustration of empirical placement of the charge is given in Figure 72. We can see that it is
hard to achieve the optimum profile due to the long required time to charge (more than 6
hours for the totally depleted batteries). So it is difficult to get a thin load curve similar to the
profiles. The result on the consumption curve is given on Figure 73.
However, this placement depends a lot on the assumption of the required time to charge a
totally depleted battery. It is assumed in this master thesis that the required time is equal to
7:30. But if in the next years this time is reduced, then the problem can be entirely changed.

Figure 72 - Load profiles and load
Figure 73 - Consumption curve with one empirical simulation

The consumption is not as constant as we would like it to be during off peak hours.
However, this optimisation should be considered in the Demand Side Management context.
Indeed we should not only consider the PEVs’ charge but also other utilisations such as
electric water heaters’ consumption that could be optimized. And then the solution for the
optimisation of the PEVs’ charge would be different.
4.7.2.2.Implementation
The optimum solution would need to be implemented in the PEVs’ fleet. The result of this
optimum charge placement could be analysed in order to deduce a distribution law for the
beginning of the charge according to the required time to charge. Then this law could be used
in real time in order to decide the charge’s start.

This regulation could be done in two different ways:

-centralized regulation: an aggregator could collect the information on the PEVs
(number of connected vehicles, required energy, required time to charge, time when the user
wants the charge to be over…) and optimize the beginning of the charge sending signals to
the PEV to start the charge.
-on board regulation (charge management system in the vehicle): this regulation
wouldn’t require as important exchanged data flows as with the centralized regulation. The
distribution laws for the placement of the charge could be implemented in the vehicle system.
But this solution requires the car manufacturers’ participation.
Therefore, this part gives only an idea of the worst case, of an optimum consumption curve,
and a possible optimisation that could be performed. But this optimisation needs to take into
account other utilisations’ optimisation.
4.8. Reserves and Margins
It is necessary for the electric system safety to have power reserves and to be able to use them
in an efficient way. A correct reserve’s sizing is really important to guarantee the balance
between consumption and production at each time.
4.8.1. Reserves
It is important to check in real time the effective availability of the power reserves. To
maintain the grid frequency, these reserves can be called up according to different time
constants. There are different levels of reserves, and each reserve can be reconstituted by the
level above. The first two reserves levels are handled and restored by the automatic primary
and secondary controls. The tertiary control is handled manually by the dispatchers.
The primary control aims at rapidly restoring (few seconds) the balance between load and
production. Frequency system equipments installed on some power plants can change their
generation (increase the generation when the frequency decreases for instance). It operates
until the balance between consumption and generation is reached and the frequency is stable
again. However the new frequency differs from the reference frequency.
The secondary control aims to restore the initial frequency (50Hz) and to bring the
international exchange to their scheduled values. This can be done by changing the power
generation set point of some power plants.
The tertiary control consists in changing the production programs of some power plants
through the balancing mechanism in order to restore the secondary reserves, and in some
cases the primary reserves. This is done in order to be ready for a new possible unforeseen
event [Mémento- RTE (2004)].
Primary reserve is the total of the primary reserve of each power plant participating in the
primary control. The primary reserve of a power plant is the power allocated to the primary
control. The UCTE (Union for the Co-ordination of Transmission of Electricity) stipulates
that France should have a primary reserve of 700 MW at anytime.

The secondary reserve is the reserves of each power plant participating in the secondary
control. RTE determines the needed secondary reserve’s level for each 30 minutes step. It can
be equal to a value advised by the UCTE or higher when the demand’s gradient is important.
The tertiary reserve is the power that can be mobilized in less than 30 minutes. This reserve is
constituted by the bids of the balance responsible players on the balancing mechanism. The
tertiary reserve can be split up in two parts: the fast tertiary reserve and the complementary
tertiary reserve. The fast tertiary reserve can be mobilized in less than 15 minutes, during at
least one hour and can be activated at least twice a day. The minimum value for this reserve is
1,000 MW. The complementary tertiary reserve can be mobilized between 15 minutes and 30
minutes, for a duration of at least six hours and can be activated at least once a day. The
minimum value for this reserve is 500 MW and its role is to restore the fast tertiary reserve.
There is also another type of reserve (‘réserve différée’). It corresponds to the power that can
be mobilized within more than 30 minutes. This reserve aims to restore the tertiary reserve
[memento RTE].
PEVs can be a potential help for the system, with their battery storage. Indeed we can imagine
that the batteries participate to the tertiary control, since they can be seen as a potential fast
tertiary reserve. If the PEV is equipped with V2G it can provide energy when the system
needs it.
That is why the PEVs’ potential for the reserves has been studied. The idea is to know what is
the mean available energy and power at anytime from the PEVs.
A Monte Carlo analysis has then been performed. At anytime, the energy contained in the
vehicles connected at home has been analysed.
Besides, it is possible to know the power that can be injected into the grid if we assume that if
they are connected, they can inject 3 kW into the grid with Figure 65.
Figure 74 - Mean Available Energy
We can see that from 3pm to midnight, 2 kWh to 13 kWh can be available. If we assume that
only 50% of the battery can be used, that means that 1 kWh to 6.5 kWh can be available. That
is to say that they can provide 3 kW during approximately 20 minutes at 3pm to 2 hours at
midnight. So if 10% of a fleet with 1 million vehicles participate in this primary control, 300

MW can be available during 20 minutes to 2 hours. At current peak hour (6pm), 300 MW are
available during approximately 40 min.
4.8.2. Margins
At t0, the margin for a later time t1 is the difference between the available generation given at
t0 for t1 and the forecasted demand at t0 for t1. In other words, the margin is the total of the
secondary, tertiary reserves, the “reserve différée” and the joint UCTE’s reserves.
The required margin must respect a defined risk level. It depends on the demand and
generation level, on the power plants’ reliability, on the hazard on the forecasted demand…
Every day, RTE determines the margin for different terms for the next day. The day before
and at anytime during the day, RTE must control that the margin is always higher than the
required level [memento RTE].
The PEVs’ charge introduces more important dynamics on the consumption. Therefore if we
want to keep the same risk level, the margins might need to be changed.
Figure 75 - Margin’s sizing principle

If we know the distribution of the hazards on the production Production, on the consumption
Consumption and on the wind power production Wind, we can deduce the required

margins’ size. Indeed, with these distributions for a given term i, if we accept a risk of 1% on
the hazard, then we can deduce the minimum size of the margin in MW for this term i.

+ , *
This paragraph doesn’t aim at calculating the exact PEV’s CO2 emission, but at showing that
the electric vehicles are cleaner when they are charged during the night.
The electricity CO2 footprint depends on the sources, and on the power plants. The utilization
of the power plants depends on the time of the day. Therefore the PEVs’ CO2 footprint
depends on the time when PEVs are charged.
In France, base electricity emits 45 gCO2/kWh. French electricity production emits in average
85 gCO2/kWh, and European electricity production emits 400 gCO2/kWh in average
[ADEME (2009)].
We assume here that the day can be divided into three different parts:
-base electricity from 0:00 to 5:00 (45 gCO2/kWh)
-mean French electricity from 5:00 to 17:00 and 20:00 to 0:00 (85 gCO2/kWh)
-mean European electricity from 17:00 to 20:00 (400 gCO2/kWh). It is assumed that
during peak hours, we can import electricity from European countries.
These different time slots for CO2 emission are represented on Figure 76.
Given the load curve for EVs and the mean daily covered distance, we can deduce the CO2
emission per kilometer.
This type of calculation for CO2 footprint doesn’t take into account the week-ends. Indeed it
is the CO2 emission during workdays. But we can assume that during week-ends and
holidays, as the consumption is lower, France won’t import European electricity and the used
power plants will be nuclear, wind power and hydro power plants. Then the EV’s CO2
emissions are lower during week-ends and holidays.
Fleets with only EVs and with the different regulations have been simulated.
The mean daily covered distance is 58km/day.
Figure 76 - Mean power for EVs and different slots for CO2 emissions

Table 36 - Energy per day, and CO2 emissions
Mix Mix
European Value (with 45g/kWh from (with 85g/kWh from
0:00 to 5:00) 0:00 to 5:00)
(gCO2/km)
(gCO2/km) (gCO2/km)
No regulation 68 32 33
Tariff Signals (40%) 68 24 26
Tariff Signals (80%) 68 18* 18
BMS bis (40%) 68 24 26
BMS bis (80%) 68 18* 18
BMS & Tariff Signals

68 24 25
(40%)
BMS & Tariff Signals
68 18* 18
(80%)
Injection (40%) 68 21 24
Injection (100%) 68 - -
*: with 85g/kWh from 0:00 to 5:00 because of the addition of a new utilization
The mean energy consumed by EVs per day is around 10 kWh/day. The energy with the
regulations are similar to the consumed energy without regulation since the regulations only
spreads out the load during the day. With the injections, the total energy corresponds to the
consumed energy minus the injections into the grid. So we obviously find the same the results
as without regulation.
The first column corresponds to the CO2 emissions if we take the mean European value for
CO2 emissions for the whole day. Obviously, since we consider in this case that the CO2
emissions don’t depend on the period of the day, and given that the energies are identical with
the regulations, the CO2 emissions per kilometers are identical. The EVs emit around 68
gCO2/km then.
The “tank to wheel” average CO2 emission in 2008 in France for internal combustion engines
is 140 g/km (161g/km for “well to wheel”). The European Union goal is to reach 95 gCO2/km
by 2020 (110 g/km for “well to wheel”) [ADEME (2009)]. If we consider the mean European
CO2 emissions per kW, EVs CO2 emissions can be 40% lower than the European goal for
conventional vehicles (well to wheel emissions).
Nonetheless if we consider the CO2 emissions per period of the day, the EVs’ carbon footprint
without regulation is around 32 gCO2/km. With regulations or injections for 40% of the fleet,
the EVs’ CO2 footprint is around 24 gCO2/km. So the regulations can decrease the EV’s
carbon footprint with 25%.
Besides, if 80% of the fleet, or more, have a regulation, or can inject into the grid, we can
assumed that a new non negligible utilization is added, then during the night, we consider the
French mean value for CO2 emissions and not the base value as before. However, the CO2

footprint with regulation is still lower than without regulation with 40% (around 18
gCO2/km).
If the entire fleet can inject into the grid (corresponds to the line “injection 100%”) the CO2
footprint is negative (-6 gCO2/km). However, we cannot deduce from this negative value that
if the vehicle covers longer distances then the carbon footprint is better. Indeed in this case,
the vehicle injects less power into the grid, and then its carbon footprint is less interesting. So
in the case of injection, we should not consider the emission per kilometer, since the batteries
can be seen as a mean of electricity storage in order to smooth the consumption curve.
In the last column, we consider that during the night we don’t use base electricity anymore
because of the EV’s charge, but the France mean CO2 emissions per kW. In this case, EVs’
CO2 footprint without regulation is around 33 gCO2/km. With 40% of the fleet having a
regulation, it is decreased to approximately 26 gCO2/km, or 24 gCO2/km if 40% of the fleet
inject into the grid.
Therefore we can see that the PEV’s carbon footprint is much less important than a
conventional vehicle’s one.
Besides the figures used for these calculations are official figures for carbon emissions due to
electricity production. Nonetheless, in a recent study [PwC, Enerpresse (2009)], the mean
European carbon emissions has been calculated at 350 gCO2/kWh. Moreover, we can assume
that in the next years these figures will be less important due to the introduction of renewable
energy. Thus the PEV’s carbon footprint can be even more interesting.
Furthermore we can see that if we consider the CO2 emissions per kW according to the period
of the day, EVs CO2 footprint can be much less important with the different regulations than
without regulation.
Therefore, the regulations are not only beneficial for the TSO, but also for the EV’s carbon
footprint. The regulations contribute to the PEVs’ cleanness.

Main Aspects of the Study
This master thesis aimed at evaluating the possible impact of the plug-in electric vehicles’
charge on the balancing electric system by 2020-2025. A study on the battery technology’s
and vehicles’ state of the art has been performed. Traffic data were collected in order to
constitute the most realistic vehicles fleet. Then a fleet’s model has been built to simulate the
charge’s times, the battery’s SOC and the needed power. A sensitivity analysis has been
performed in order to study the impact of some hypothesis on the results. This general and
flexible model can be used for different studies.
The impacts of the different assumptions on the fleet have been studied. However the PEVs’
load curve always results in a peak at 20:00, when the total consumption curve is already
high. The examples of total consumption curves, with different numbers of PEVs in 2015,
2020 and 2025 show the possible impact on the load curve. The load peaks are increased with
around 3 GW in 2020. Some solutions have been studied in this master thesis in order to limit
this impact. Tariff signals that already exist for electric water heaters for instance have been
simulated. The charge is put back during the night when the consumption is less important.
Battery management system that spreads out the charge of the vehicle during the whole night
has also been modelled. This way, the peak load is less important at 20:00. These two possible
regulations have been combined in order to benefit from the two advantages given by the two
different regulations. However the feasibility of these regulations depend on the user’s
behaviour. It can be efficient only if a high enough number of users decide to use it. If most of
the fleet uses these regulations, the impact of the PEVs’ charge can be invisible. But the
utilisation of these regulations depend on social and human factors. The manufacturers won’t
develop such systems without the total social acceptance.
A small model with possible injections into the grid during peak hours has been performed.
The results of this system are interesting if enough vehicles can participate, since injections
into the grid can erase the one million PEVs’ charge effect on the consumption. However it
can increase a lot the dynamics of the consumption curve around 9:00pm and put back the
peak. PEVs can also be seen as a potential help for the reserves and have an impact on the
margins’ sizing.
The problem of optimisation to get an optimum load curve for the electric system has also
been suggested.
Besides, an environmental analysis has also been performed in order to know the PEVs’
impact on the CO2 emissions. If the PEVs charge during the night the CO2 emissions are
totally different than when they charge at 6-8 pm. The times of charge determine the PEVs’
cleanness and this analysis shows that the possible regulations have a positive impact on the
carbon footprint as well as on the electric balancing system.
Therefore if no incentive is given to the PEVs’ users the impact on the load curve, and thus on
the carbon footprint can be important. This study shows that regulations have to be taken.
Many solutions are possible to decrease the electric system’s impact, help the system and
improve the PEVs’ cleanness.

Perspectives
Some assumptions for the simulations are quite rough but give a good idea of the possible
impact on the consumption curves. In a future works a more complex simulation with fewer
assumptions could be performed. The study of a PEV’s load curve during week-end and
holiday would complete this analysis.
An economic model would have to be studied in order to know if PEV’s owner will have
some interests in injecting power into the grid. The impact on the battery is not well known,
but we know that it shortens the battery’s life. Therefore the financial contribution would have
to cover the battery’s damages. A precise simulation of a smart grid would also be of interest.
Fleets of Vehicle-to-Grid vehicles can act as energy storage systems. This can increase the
efficiency and reliability, and will allow for more use of intermittent renewable energy as
wind power in the future.
The impact on the network was out of scope of this master thesis. The analysis of the PEVs’
dispersion in France, and simulating a fleet for each big city in France would give a good idea
of what could be the impact on the French network.
An optimisation of the charge placement could be done in order to get a constant consumption
curve during off peak hours. But this optimisation has to be seen in the Demand Side
Management context, and take into account the other flexible utilisations that could be
optimized.
An investigation of a possible coupling of solar power with PEV’s charge could also be
performed.

** -
Table 37 - Characteristics of the EVs and PHEVs to come

battery top
Vehicle's EV Type of battery
Company size speed Launching
Name /PHEV battery range (km)
(kWh) (km/h)
2011
Kangoo Be ~100 (but (for commercial
Bop when kangoo)
Renault EV Li-ion 15 130
[renault.com] released,
~160 km) 2012
(for urban cars)
Imiev
[media.mitsubi april 2010 (in
shi- Mitsubishi EV Li-ion 16 160 130 Europe, partnership
motors.com] with PSA)
Leaf
[nissan- laminated
Nissan EV 24 160 140 2011 in Europe
zeroemission.c Li-Ion
om]
2010 (not to sell,
LMP + super
Blue Car Bolloré- but to rent, already
EV capacitors 30 250 130
[batscap.com] Pininfarina 5000 bookings in
(batscap)
Europe)
Ampera
60 (all production begins
[opel- Opel PHEV Li-ion (GM) 16
electric) in 2011
ampera.com]
Smart fortwo large scale
Smart -
ed EV Li-ion ~100 commercialization
Daimler
[smart.com] in 2012
Li-ion
500 vehicles in
Th!nk city (Enerdel,
Th!ink EV 28,3 180 100 Spain, end of 2009,
[think.no] A123
beginning of 2010
systems
Ener1: Li-Ion
[volvocars.com
Volvo PHEV (in demons- 11,3 50 by 2012
]
-tration fleet)
Focus BEV
[media.ford.co Ford EV Li-Ion 23 ~160 2011? / 2012?
m]
Fiorina Micro
Fiat /
Vett EV Li-Ion 17 or 22 70 or 100 on the market
Newteon
[newteon.com]
Twin Drive
Seat PHEV Li-ion 50 100 not until 2014
ecomotive
Prius Toyota PHEV 20-30

battery top
Vehicle's EV Type of battery
Company size speed Launching
Name /PHEV battery range (km)
(kWh) (km/h)
??(no partnership
signed with a car
SVE
Cleanova 2 manufacturer to
(Dassault- EV Li-Ion 210 130
[dassault.fr] produce them
Heuliez)
yet=> uncertain
future)
Eclectic
Solar
[www.venturi.f Venturi 7 On the market
EV
r]
Tesla
Roadster Tesla 390 (244
EV Li-Ion
[www.teslamot Motors miles)
ors.com]
Mini E
BMW EV Li-ion 35 168 153
[www.mini.fr]
This list is obviously not exhaustive and was made in July 2009. Given that the car
manufacturers regularly announce launchings of new vehicles, this list would need to be
updated every month. But this gives a good idea of the characteristics of the vehicles to come.
Fleet of N
vehicles
80% 20%
PV CV
20% 80% 80% 20%
EV PHEV EV PHEV
20% 80%
Rural Urban
Battery Size
Trip Length
Number of Trips (for each period of the day)
Speed
Available Type of Charge (for each location)
Departure Time (for each trip)
Initial State of Charge
Figure 77 - Initialisation of a fleet

Remaining distance until charge
needed SOC = Consumption / Battery Size* Remaining distance until charge
State =
No
Running ?
Yes
Remaining Distance = Trip Length Remaining Distance = Remaining Distance -Speed*SimulStep
EV ? No
Yes
pRR = remaining distance > Trip length & SOC >0 pRR= remaining distance > Trip Length
PSOC = needed SOC <SOC PSOC = 1
PRP = 1-PRR
PWants to drive
Ptime
P*R = PSOC *PWants to drive * Ptime
PCC = 1-P*R
Ptime charge
PTC
PPC = PTC * Ptime charge
PPC =
P*R ?
Yes
PPC = 0
PPP = 1- P*R - PPC
Figure 78 - Transition Matrix’s Flowchart

For the transition’s block, we have to calculate the remaining distance until the next charge,
and the corresponding battery that the vehicle will need.

If the vehicle is in the state “running”, we calculate the remaining distance to travel. And then
the different probabilities explained before can be calculated in order to get T, the transition
matrix.
State
State =
No
Charging ?
Yes
State = Boolean
Running? No
No
Boolean =
True ?
Yes
Yes
P = -Consumption * Speed P=0 P = Pnom*Battery Size
Figure 79 - Power Block’s Flowchart

Pnom is the nominal power, as described on Figure 6.
Figure 80 - Number of running vehicles (fleet with 5000 vehicles)

Rechargeable Li-Ion Button Battery Serial LIR2032AA AA

AA portable power corporation
portable power corporation
Les transports électriques en France : un développement
ADEME (2009) nécessaire sous contrainte - ADEME&VOUS n°21 – Stratégie et
Etudes – July 2009
Status and Trends in the HEV/PHEV/EV Battery Industry – D.
Anderson (2008)
Anderson – Rocky Mountain Institute – summer 2008
Plug-in Hybrid Electric Vehicle as Control Power – Case
Andersson, Elofsson (2009) Studies of Sweden and Germany – Master Thesis – Sara-Linnea
Andersson and Anna Elofsson -2009
Goals and State of Technology circa 2008 – J. Axsen, A. Burke,
Axsen, Burke, Kurani (2008) K. Kurani – Institute of Transportation Studies – University of
California – May 2008
Les déplacements domicile-travail amplifié par la
Baccaïni, Sémécurbe, Thomas,
périurbanisation – B. Baccaïni, F. Sémécurbe, G. Thomas –
INSEE (2007)
INSEE – march 2007
Bluecar specifications -
batscap.com http://www.batscap.com/actualites/communiques/BlueCar_tech
nic-04-2007.pdf - Accessed August 5, 2009
Investigation into the Scope for the Transport Sector to Switch
BERR & Department for transport
to Electric Vehicle and Plug-in Hybrid Vehicle – BERR and
(2008)
Department for transport – October 2008
Better Place- http://www.betterplace.com - Accessed November
http://www.betterplace.com
22, 2009
Note de Synthèse CERTU – Mobilité et transports – Les recueils

CERTU (2008)
de données sur la mobilité urbaine – CERTU – February 2008
Le transport léger – Situation et perspectives – DAEI-SESP –

DAEI-SESP – Samarcande (2005)
Samarcande –juin 2005
Cleanova Presentation -
dassault.fr http://www.dassault.fr/filiale.php?docid=82 - Accessed
November 22, 2009
Développement du véhicule électrique particulier – Frédérique
DDRPE (2008)
Decré – RTE-DDRPE – June 2008
Enerzine- Actualité - http://www.enerzine.com/15/8417+plan-
enerzine.com borloo-pour-des-vehicules-electriques-et-hybrides+.html -
Accessed October 2, 2009
ERDF - http ://www.erdfdistribution.fr – Accessed November
http ://www.erdfdistribution.fr
22, 2009

Quel lien entre les conditions de stationnement et le choix du

Gantelet, Sareco (2003)
lieu de résidence ? – E. Gantelet – Sareco - 2003
PHEV Laboratory Testing and Analysis – D. Hillebrand, M.

Hillebrand, Duoba ( 2007)
Duoba – Argonne National Laboratory – June 2007
Les pratiques environnementales des français en 2005 – Les

IFEN (2007)
dossiers IFEN – December 2007
Status and Prospects for Zero Emissions Vehicle Technology, F.

Kalhammer et. al. (2007)
R. Kalhammer et. al., CARB Report, April 2007
Kromer et al (2007) Kromer et al (2007), & RMI Analysis
Enquête sur les déplacements des habitants de Lille métropole –

Lille Métropole (2006)
Lille Métropole – 2006
Rapport du sous groupe de travail « Normalisation-
Standardisation » du groupe de travail « Infrastructure de charge
MEEDDAT, SGT (2009)
pour véhicules électriques et hybrides rechargeables » -
MEEDDAT – June 2009
Les transports en 2008 (premiers résultats)-Version provisoire
MEEDDAT/SOeS (2009) du 46ième rapport de la commission des comptes des transports
de la Nation – MEEDDAT/SOeS - March 2009
Ford Battery Electric Vehicule -
media.ford.com http://media.ford.com/images/10031/electrification_1.pdf -
Accessed November 22, 2009
Mémento- RTE (2004) Mémento de la surêté du système électrique – RTE - 2004
Mitsubishi Motors- I-Miev - http://media.mitsubishi-

media.mitsubishi-motors.com motors.com/pressrelease/e/products/detail1940.html-Accessed
November 22, 2009
Mennekes website - http://www.mennekes.de/web-Accessed
mennekes.de
November 22, 2009
MTET/SESP (2007) La demande de transport en 2025 - MTET/SESP –May 2007
Fiat Fiorino Characteristics -

http://www.newteon.com/www2/fr/Vehicules/Transport_de_ma
newteon.com
rchandises/Fiat_Fiorino_Electrique_Micro-Vett_-
_version_Cargo_20080414348/- Accessed November 22, 2009
Nissan – Zero Emission - http://www.nissan-
nissan-zeroemission.com zeroemission.com/EN/index.html - Accessed November 22,
2009

Opel Ampera - http://www.opel-ampera.com/francais/-

opel-ampera.com
Accessed November 22, 2009
Bornes de recharge pour véhicules électriques -
http://www.paris.fr/portail/deplacements/Portal.lut?page_id=381
http://www.paris.fr
&document_type_id=5&document_id=11370&portlet_id=1199
- Accessed November 22, 2009
Les voitures électriques : enjeux techniques et perspectives
Perrin, Renault (2009) d’une nouvelle mobilité respectueuse de l’environnement – J.
Perrin – Renault – February 2009
PricewaterhouseCoopers - http://www.pwc.fr/le-facteur-
PwC, Enerpresse (2009) carbone-europeen-des-electriciens-a-atteint-son-niveau-le-plus-
bas-depuis-2001.html - November 2009
Véhicule Electrique : recharge batterie et infrastructure. La
Renault (2009)
charge rapide – Renault – 2009
Renault - http://www.renault.com/FR/CAPECO2/VEHICULE-
renault.com ELECTRIQUE/Pages/vehicule-electrique.aspx- - Accessed
November 22, 2009
Bilan prévisionnel de l’équilibre offre-demande de l’électricité
RTE (2009)
en France – RTE - 2009
rte-france.com RTE - http://www.rte-france.com- Accessed November 2, 2009
Saft- Lithium Ion Battery -

saftbatteries.com http://www.saftbatteries.com/Technologies_Lithium_Liion_301/
Language/en-US/Default.aspx - Accessed November 22, 2009
Les transports urbains en France : des solutions techniques mais
Science et Décision (2006) une gestion politique est nécessaire – Science et Décision –
October 2006
Rapport sur la définition et implication de concept de voitures
Sénat (2005) propres – Députés C. Cabal, C. Gatignol – Sénat – December
2005
Smart for two - http://www.smart.com - Accessed November
smart.com
22, 2009
Les différentes technologies de stockage d’énergie – Camille

Soncini, RTE (2009)
Soncini – RTE – May 2009
Think – EV technology and Innovation -
think.no http://www.think.no/think/content/view/full/264 - Accessed
November 22, 2009
Volvo -
volvocars.com http://www.volvocars.com/intl/corporation/NewsEvents/News/P
ages/default.aspx?item=81 - Accessed August 18, 2009

Master Thesis: Impact of The Electric Vehicle On The Electric System

Uploaded by

Copyright:

Available Formats

Master Thesis: Impact of The Electric Vehicle On The Electric System

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Master Thesis: Impact of The Electric Vehicle On The Electric System

Uploaded by

Copyright:

Available Formats

Master Thesis

Impact of the Electric Vehicle on the Electric System

Mélaine Rousselle - RTE-DMA – KTH 1

Mélaine Rousselle - RTE-DMA – KTH 2

Mélaine Rousselle - RTE-DMA – KTH 3

4.1.2. Consumption by 2015, 2020 and 2025............................................................. 53

Mélaine Rousselle - RTE-DMA – KTH 4

Mélaine Rousselle - RTE-DMA – KTH 5

Figure 37 - Total load curve with PEVs in 2025...................................................................... 56

Mélaine Rousselle - RTE-DMA – KTH 6

Mélaine Rousselle - RTE-DMA – KTH 7

BMS Battery Management System

Mélaine Rousselle - RTE-DMA – KTH 8

1.1. Presentation of RTE

RTE (“Réseau de Transport de l’Electricité”) is responsible for transmitting electricity from

Mélaine Rousselle - RTE-DMA – KTH 9

1.3. Purpose and Aim

Mélaine Rousselle - RTE-DMA – KTH 10

2.1. Electric Vehicle’s and Plug-in Hybrid Vehicle’s Concept

Mélaine Rousselle - RTE-DMA – KTH 11

Mélaine Rousselle - RTE-DMA – KTH 12

2.2.2. Lithium Batteries’ Principle

Figure 2 - Principle of a lithium-ion battery [Anderson (2008)]

no memory effect explosion risk

advanced technology deep cycles are inadvisable

can have different shapes cost

Mélaine Rousselle - RTE-DMA – KTH 13

Mélaine Rousselle - RTE-DMA – KTH 14

2.2.3. Charge of a Lithium Battery

Table 4 - Different types of charge

load charge time

Mélaine Rousselle - RTE-DMA – KTH 15

Figure 6 - Load for the different types of charge

Mélaine Rousselle - RTE-DMA – KTH 16

2.3. Charge Facilities

Mélaine Rousselle - RTE-DMA – KTH 17

2.4. Vehicles Characteristics

Battery range and size for EV

Battery range and size for PHEV

EVs engine efficiency

Mélaine Rousselle - RTE-DMA – KTH 18

Figure 7 - Electric consumption for EVs [Perrin, Renault (2009)]

PHEVs engine efficiency

City driving Average

kokam Hymotion Prius 79 38,6 2,9

highway driving Average

kokam Hymotion Prius 61 52,8 3,2

Mélaine Rousselle - RTE-DMA – KTH 19

2.5. Traffic Data

Table 8 - French fleet in traffic in 2008 [MEEDDAT/SOeS (2009)]

Distance and number of trips

Commuting distance (data 2004) Road Distance (km)

Mélaine Rousselle - RTE-DMA – KTH 20

Figure 8 - Vehicles’ trips distribution for Lille’s area

Average daily driving distance in urban areas for a captive fleet

Average daily driving distance in rural areas of a captive fleet

Mélaine Rousselle - RTE-DMA – KTH 21

Mélaine Rousselle - RTE-DMA – KTH 22

3.1. Characteristics of the Fleet’s Model