DUAL TARIFF CHARGING CONTROL
FOR LARGE EV FLEETS
Evangelos L. Karfopoulos, Despina Koukoula and Nikos Hatziargyriou
National Technical University of Athens, Department of Electrical and Computer Engineering, 15773, Athens, Greece
Keywords: Electric Vehicles, Energy Management, Multiagent Systems, Intelligent Agents.
Abstract: In the forthcoming years, a significant deployment of Electric Vehicle (EV) technologies, plug-in hybrid
and pure battery EVs, is expected. Uncontrolled charging can affect significantly the normal operation of
the power system and result in premature grid reinforcements. Dual-tariff scheme can be effective provided
that EV uptake is not high. This paper presents an enhanced dual-tariff distributed EV management
approach for handling large EV fleets. The proposed management scheme allocates efficiently the EV
charging demand during the low energy price period achieving a “valley filling” affect.
1 INTRODUCTION
In the forthcoming years, a significant deployment
of Electric Vehicle (EV) technologies, plug-in
hybrid and pure battery EVs, is expected. This is
indicated by several prospective studies for instance
EPRI&NRDC (2007), Valentine-Urbschat &
Bernhart (2009), Electrification Coalition (2009),
IEA (2009), National Academy of Sciences USA
(2009) and The Royal Academy of Engineering UK
(2010). Even though there still exist significant
technological and economic barriers mainly related
to the storage technologies, the future of electric
vehicles (EVs) seems to be rather promising
considering their energy efficiency and
environmental advantages compared to the
conventional transportation.
The integration of plug-in electric vehicles into
electric power systems can be envisaged into three
phases 0. In Phase I, the major objective is to
facilitate the EV uptake. In this phase EVs will be
regarded as mere additional loads like any other
conventional load. As the EV penetration ratios
become more significant the network operation is
likely to be affected. Thus, in Phase II, potential
charging control concepts should be developed to
manage the additional EV load. This will enable the
emerge of new business models such as the one of
EV Supplier-Aggregators (EVS/A) which will be
responsible for managing large EV fleets, either in
private or public charging places, through bilateral
contracts. In Phase III, a very optimistic and long-
term scenario has been identified by introducing the
bidirectional exchange of power between the grid
(Grid-to-Vehicle operation, G2V) and the vehicle
battery (Vehicle-to-Grid operation, V2G).
This paper focuses on the second development
stage (Phase II) of electric vehicles and aims to
investigate the deployment of conventional demand
management concepts, such as the dual-tariff one,
for controlling EV charging. The scope is to
envisage the possible network impediments that may
prevent the implementation of the dual-tariff
charging scheme and propose an enhanced dual-
tariff EV management approach which enables a
more efficient EV integration regarding the network
operation. The proposed EV management scheme is
a decentralised, price-based control which allocates
the EV demand among the low energy price hours in
a ‘valley filling’ concept. The operational
framework for the implementation of the enhanced
dual-tariff EV charging control is base on the Multi-
Agent System (MAS) technology.
In Section 2, the energy requirements that fulfil
EV charging needs during Phases I and II are
identified considering different penetration level
scenarios. The resulted charging profiles enable the
estimation of the grid impact of EV deployment. In
Section 3, the proposed enhanced dual-tariff EV
management scheme is analytically described.
Conclusions are drawn in Section VI.
121
L. Karfopoulos E., Koukoula D. and Hatziargyriou N..
DUAL TARIFF CHARGING CONTROL FOR LARGE EV FLEETS.
DOI: 10.5220/0003978801210125
In Proceedings of the 1st International Conference on Smart Grids and Green IT Systems (SMARTGREENS-2012), pages 121-125
ISBN: 978-989-8565-09-9
Copyright
c
 2012 SCITEPRESS (Science and Technology Publications, Lda.)

2 IDENTIFYING EV CHARGING
NEEDS
The integration of EVs into power systems is
expected to increase the system demand due to the
charging needs of EVs. The amount of this
additional EV demand depends mainly on the EV
penetration level and the EV owner’s driving profile
(travel distance, battery consumption). The time of
EV plug-in and the available type of charging
(Modes1, 2 & 3) (Bending et al., 2010)-0, State of
the art charging infrast] defines the grid impact of
the additional EV demand.
Figure 1 presents the simulation tool developed
to identify the additional EV demand in the first and
the second EV deployment stages. The inputs of
this model are described below:
 EV deployment scenarios:
Different number of
EVs can be used regarding different
penetration (high, middle, low) scenarios.
 Classification of EVs:
The plug-in EV
technologies can be classified into two general
categories:
- the plug-in hybrid EV (PHEV) and
- the pure battery EV (BEV)
BEV can be further subdivided into further
subcategories according to their technical
characteristics as:
-
L7e
: small city purpose vehicles
-
M1
: 4-seater passenger vehicles
-
N1
:carriage of goods with a maximum
laden mass of less than 3,500 kg
-
N2
: maximum laden mass of 3,500 kg
to 12,000 kg for commercial
purposes
 Trip Length:
This parameter describes the
daily distance covered by an EV between two
successive charging cycles and thus the
corresponding amount of charging energy.
 Battery consumption:
Average energy
consumption over travelled distance (kWh/km)
can be considered per each EV type.
 EV demand:
The total battery demand
requested from the grid, considering the losses
of the charger’s power electronics (15%).
 Availability of charging:
The charging can be
initialised once (home charging) or multiple
times per day (home/work/commercial
charging).
 Time of plug-in
: The time of EV plug-in
defines the initialisation of the charging period,
especially in case of uncontrolled charging.
 Charging Scheme
: It depends on the EV
deployment stage.
- Uncontrolled charging:
It is the plug and
play connection of Electric Vehicles into
the grid which happens after the last trip of
each day or when a charging point is
available and there is no need for mobility.
- Dual-tariff Charging:
It's a market way of
controlling energy demand in order to
promote energy demand in off-peak hours.
Figure 1: Simulation Model identifying EV demand.
For the purpose of EV demand analysis, three
hypothetical EV fleets of 250, 500 and 1000EVs,
(for low, middle and high EV deployment
respectively) have been considered. Only pure
battery EVs have been considered for simplicity and
it is assumed that L7e=1%, M1=88%, N1=10%,
N2=1% as a percentage of the total fleet. Moreover,
the travel distance of EVs has been simulated by an
exponential probabilistic density function with mean
value 40km. Different average battery consumptions
over travelled distance per pure battery EV type has
been assumed: L7e=0.13, M1=0.16, N1=0.24,
N2=0.8 kW/Km.
SMARTGREENS2012-1stInternationalConferenceonSmartGridsandGreenITSystems
122

Figures 2 and 3 present the results of the
simulations, exploring the tool of Figure 1, for home
charging (Mode 1 charging), considering different
EV deployment scenarios.
Figure 2(a) presents the EV demand of
uncontrolled home charging assuming that the EV
plug-in time is defined by a normal distribution with
mean value at 8:00pm. Figure 2(b) shows the impact
of the additional EV demand in the typical daily
winter load diagram of a Greek urban network
[Merge 3.1]. In case of uncontrolled home charging
strategy, the EV load is coincided with the increased
domestic consumption increasing the daily load peak
by 5.8%, 12.53% and 26.42% for the respective EV
penetration levels.
(a)
(b)
Figure 2: EV demand for uncontrolled charging.
Figure 3(a) presents the EV demand when a
dual-tariff market model. The winter low energy
pricing period in Greece is between 11pm-7am.
Dual-tariff charging is more effective than the
uncontrolled charging since it enables the shifting of
the EV demand from high loading hours to off-peak
ones namely valley-hours. However, there is an
instantaneous increase of the EV load verified at the
beginning of the low energy price period. In case of
1000EVS, a new system load peak occurs due to EV
demand, which is increased 10.76% compared to the
initial one. This might provoke several technical
problems in some networks especially to those that
operate in strained conditions and the EV
deployment is high. Premature grid investments
would be inevitable.
(a)
(b)
Figure 3: EV demand for dual-tariff charging.
The allocation of the EV demand among the low
energy price hours under a “valley filling” concept
could enable a more efficient network operation
increasing simultaneously the maximum allowable
EV integration levels without considering network
reinforcements.
3 A MULTI-AGENT SYSTEM EV
MANAGEMENT CONCEPT
This section presents a decentralised, multi-agent
system solution for the coordinated charging of
plug-in EVs aiming to avoid the instantaneous
increase of the EV demand when low energy price
period starts. The schematic overview of the
proposed multi-agent system is depicted in Figure
40.
DSO agent is responsible for monitoring and
ensuring the secure and reliable operation of the
distribution network.
EVS/A agent is a new business model 0 which is
expected to emerge during the second EV
0
250
500
750
1000
1250
1500
1750
2000
2250
123456789101112131415161718192021222324
Demand(MW)
Hours
EVDemand
250EVs 500EVs 1000EVs
0
1
2
3
4
5
6
7
8
9
10
123456789101112131415161718192021222324
Demand(MW)
Hours
DailySystemLoad Diagram
Non‐EV case 250EVs 500EVs 1000EVs
0
500
1000
1500
2000
2500
3000
3500
4000
4500
1 2 3 4 5 6 7 8 9 101112131415161718192021222324
Demand(MW)
Hours
EVDemand
250EVs 500EVs 1000EVs
0
1
2
3
4
5
6
7
8
9
123456789101112131415161718192021222324
Demand(MW)
Hours
DailySystemLoad Diagram
Non‐EV case 250EVs 500EVs 1000EVs
DUALTARIFFCHARGINGCONTROLFORLARGEEVFLEETS
123

Figure 4: Agent coordination structure.
deployment stage and will be responsible for
purchasing energy from electricity market and
managing large EV fleets. An EVS/A agent can
offer additional load management services
incentivized by the DSO agent contributing to a
more efficient, in terms of losses, voltage drops, grid
loading, operation of the grid.
Local EV/A agents are located at the secondary
substations (MV/LV). They are responsible for
aggregating the demand of the EVs parked at a
specific geographical area. The aggregated profiles
are communicated to the EVS/A.
Vehicle Controller (VC) agents represent the
EVs to the upstream network. The VC agents have
the ability to act autonomously and the intelligence
to take decisions that fulfil individual goals.
The multi-agent EV charging management
concept is a price based one. The EVS/A agent is
responsible for defining hourly series of prices
which are directly proportional to the grid area
loading. The prices set by EVS/A are not the actual
market prices, but pricing signals reflecting that
higher non-EV and EV demand results in higher
market prices.
Based on EVS/A’s pricing policy, the VC agents
define their charging strategy aiming to minimise
their virtual charging cost. The objective function of
a VC agent can be found in 0. According to this
formula, the best charging strategy of every VC
agent is balanced between the charging price and the
cost of deviating from the average charging strategy.
Thus, VC agents are less aggressive in charging the
cheaper valley hours.
The coordination mechanism consists of the
following steps:
1. DSO provides EVS/A with the hourly available
power for charging without exceeding the
domestic area’s peak load, for the whole low
energy price period.
2. EVS/A defines the pricing policy in accordance
to the grid area loading.
3. VC agents communicate their best charging
response to the local EV/A
4. Local EV/A agents aggregate the charging
demand of a small geographical area and
provide one load profile to EVS/A.
5. EVS/A agent evaluates the VC agents’ response
and reschedules its pricing policy. It also
provides the hourly average value of the EV
charging demand.
6. Steps 3-5 are repeated until no better charging
response can be achieved.
Figure 5 presents the EVS/A pricing strategy for
the example of Figure 3. Figure 6 shows the total
non-EV and EV demand after each round. It can be
seen that at each round, the charging instants with
smaller non-EV demand are assigned to a larger
individual EV charging and average charging rate.
After a few oscillations, the presented multi-agent
coordination mechanism provides a “valley filling”
effect during the low energy price period defined by
the specific dual-tariff scheme.
Figure 5: EVS/A pricing strategy.
Figure 6: Total demand.
0
1
23:00 23:59
1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00
10:00 11:00 12:00 13:00
PriceSignal
Hour
EVS/ApricingStrategy
1stRound 2ndRound 3rdRound
3
3.5
4
4.5
5
5.5
6
6.5
7
23:00 23:59
1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00
10:00 11:00 12:00 13:00
Demand(MW)
Hour
Total Demand
Non_EV 1stRound 2ndRound 3rdRound
SMARTGREENS2012-1stInternationalConferenceonSmartGridsandGreenITSystems
124

4 CONCLUSIONS
The implementation of current simple dual tariff
schemes results in instantaneous increase of the EV
demand at the beginning of the low energy price
period. In case of high EV deployment levels, the
additional EV demand can affect the operation of
some networks, mainly those which are already
strained. This issue can be handled by the presented
enhanced dual-tariff multi-agent EV management
approach. In this approach, a new business model of
EVS/A has been introduced being responsible for
purchasing power from the market and managing
EVs demand. EVS/A agent coordinates the EV
charging operation by providing virtual pricing
incentives to VC agents. The VC agents act
autonomously trying to satisfy their individual goals
considering the EVS/A pricing policy. The tracking
parameter within VC agent’s objective function
enables the effective allocation of EV demand
during off-peak hours achieving a “valley filling”.
ACKNOWLEDGEMENTS
This work was partially supported by the European
Commission within the framework of the 7th
Framework Programme, EC Project MERGE-
Mobile Energy Resources in Grids of Electricity
under the Grant Agreement: 241399.
REFERENCES
I. Momber, M. Rivier, R. Cossent, J. T. Saraiva, K.
Kanellopoulos, P. Andrianesis, MERGE Deliverable
D5.2: “Identification of regulatory issues regarding
market design and network regulation to efficiently
integrate electric vehicles in electricity grids”, pp 14-
17, October 2011. [Online]. Available:http://www.ev-
merge.eu/images/stories/uploads/MERGE_WP5_%20
D52final.pdf
S. Bending, M. Ferdowsi, S. Channon, K. Strunz, MERGE
Deliverable D1.1, “Specifications for EV-Grid
interfacing, communication and smart metering
technologies, including traffic patterns and human
behaviour descriptions”, August 2010. [ONLINE].
Available:http://www.ev-merge.eu/images/stories/
uploads/MERGE_WP1_D1.1.pdf
A. M. Foley, I. J. Winning, B. P. Ό Gallachόir, “State of
the art in Electric Vehicle Charging Infrastructure”,
IEEE Vehicle Power and Propulsion Conference,
Lille, pp. 1-6, Sept 2010.
E. L. Karfopoulos, P. Papadopoulos, S. Skarvelis-
Kazakos, Iñaki Grau, L. M. Cipcigan, N.
Hatziargyriou, N.Jenkins, “Introducing Electric
Vehicles in the Microgrids Concept”, 16th
International Conference on Intelligent System
Applications to Power Systems, ISAP11, Heraklion,
Sept. 2011.
Zhongjing Ma, Callaway, D., Hiskens, I., “Decentralised
charging control for large population of plug-in
electric vehicles: Application of the Nash certainty
equivalce principle”, in IEEE International Control
applications (CCA) Conference, Yokohama, Sept.
2010.
APPENDIX
Evangelos L. Karfopoulos was born in Athens, Greece
in 1982. He received his diploma in Electrical and
Computer Engineering from NTUA, where he is now a
PhD Candidate. His research interests include
optimization of power system operation, distributed
generation, RES and Microgrids, Electric Vehicle
management, multi-agent system controls.
Despina Koukoula was born in Athens. She received the
diploma in Electrical and Computer Engineering from
NTUA in 2008. She is PhD candidate at Electrical and
Computers Engineering Department of NTUA. Her
research interests include dispersed generation, artificial
intelligence techniques in power systems and computer
applications in liberalized energy markets.
Nikos D. Hatziargyriou was born in Athens, Greece. He
is professor at the Power Division of the Electrical and
Computer Engineering Department of NTUA. Since 2007
he is Deputy CEO of the Public Power Corporation (PPC)
in Greece, responsible for Transmission and Distribution
Networks, island DNO and the Center of Testing,
Research and Prototyping. He is Fellow Member of IEEE,
past- Chair of the Power System Dynamic Performance
Committee, member of CIGRE, Convener of SCC6,
member of the BoD of EURELECTRIC and member of
the EU Advisory Council of the Technology Platform on
SmartGrids. His research interests include Smartgrids,
Distributed Energy Resources, Microgrids, Renewable
Energy Sources and Power System Security.
DUALTARIFFCHARGINGCONTROLFORLARGEEVFLEETS
125