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

net/publication/224190677

Grid interactive charging control for plug-in electric vehicles

Conference Paper in Conference Record - IEEE Conference on Intelligent Transportation Systems · October 2010
DOI: 10.1109/ITSC.2010.5625073 · Source: IEEE Xplore

CITATIONS READS
28 1,489

4 authors, including:

Filipe Soares Pedro Rocha Almeida

Institute for Systems and Computer Engineering, Technology and Science (INESC TEC) Smarter Grid Solutions Ltd
130 PUBLICATIONS 3,850 CITATIONS 46 PUBLICATIONS 2,497 CITATIONS

SEE PROFILE SEE PROFILE

Joao Abel Peças Lopes

Institute for Systems and Computer Engineering, Technology and Science (INESC TEC)
325 PUBLICATIONS 15,433 CITATIONS

SEE PROFILE

All content following this page was uploaded by Joao Abel Peças Lopes on 29 May 2014.

The user has requested enhancement of the downloaded file.

2010 13th International IEEE MB5.1
Annual Conference on Intelligent Transportation Systems
Madeira Island, Portugal, September 19-22, 2010

Grid Interactive Charging Control for

Plug-in Electric Vehicles
R. J. Rei, F. J. Soares, Student Member, IEEE, P. M. Rocha Almeida, Student Member, IEEE,
J. A. Peças Lopes, Senior Member, IEEE

Low Voltage level (LV) have a considerably high Resistance

Abstract—This paper describes technical solutions to be to Inductance ratio and are operated using a radial
adopted by Electric Vehicles (EV) battery grid interfaces, in configuration, some problems are likely to appear with the
order to get the provision of ancillary services to the grid. The deployment of EV: the lines around the feeding points tend
developed solution exploits an EV battery charge control
approach based in a grid cooperative response to frequency
to reach high congestion levels, while electrically farthest
and/or voltage deviations. To be effective, those cooperative buses have a tendency to face voltage drops, losses might
actions must result from both centralized/coordinated increase and voltage unbalances can also occur [1].
commands and local/individual EV charger response to the grid The most obvious way to solve the above presented
behaviour. In a scenario characterized by a massive deployment problems is making large investments to reinforce the grid’s
of EV, the adoption of such a solution allows an improvement infrastructures, in order to allow all EV owners to charge
on the power system dynamic behaviour, namely in small island
grids. their vehicles at free will. By adopting this solution EV could
be seen as a “threat” by Distribution System Operators
Index Terms—Charging Management, Droop Control, (DSO), who want to avoid bulky investments at all cost.
Electric Vehicle, Improved Power Quality Converters. In order to break up with these technical/economic
barriers, advanced management and control strategies must
I. INTRODUCTION be implemented for EV charging. This way, instead of being

O VER half of the world’s oil consumption is nowadays

allocated to the transportation sector, and a large
portion is used by road vehicles. Despite the predictable
a burden for the grid, EV will act as an active element in the
network, capable of helping in the control of the grid
operating conditions. While the predictable increase on the
depletion of fossil fuel reserves, the oil consumption has electric energy consumption at peak hours can be easily
increased over time, as well as the related CO2 emissions, tackled, merely, by shifting the EV charging towards the
being of utmost importance to implement measures to break valley hours, the solution for problems like high branches’
this cycle. Given these circumstances, the change towards an congestion levels or abnormal voltage profiles demand more
electricity mobility paradigm is becoming more and more sophisticated strategies. These sophisticated approaches face
needed, given that this shift will bring clear advantages for EV batteries as dynamic loads, able to vary their charging
cities air quality and for the global warming problematic [1]. rates, and within an even bolder generalization, they can be
At first look, taking an Electric Vehicle (EV) as a simple exploited as storage devices with the capability of injecting
load during battery charge, it would represent a large amount power into the grid, i.e. featuring Vehicle-to-Grid (V2G).
of electric power consumed, approaching in some cases the The successful implementation of those advanced
power of a typical domestic household at peak load. From management and control strategies relies on the development
the distribution grid point of view, it is clear that massive of EV grid electronic interfaces that must be able to perform
integration of EV will lead to substantial increase on electric grid voltage and frequency measurements, allowing an
energy demand and will raise load values at peak hours. automated response to detected deviations. Communication
Additionally, as the majority of the distribution grids at the functionalities should be available in order to enable the
broadcast of local data and measurements to central entities,
allowing also the reception of actualized control parameters.
This work was supported in part by Fundação para a Ciência e
Tecnologia under Grants SFRH/BD/48491/2008, SFRH/BD/47973/2008, Inspired by the MicroGrid (MG) hierarchical management
within the framework of the Project with the Reference concept with communication infrastructures [2] and looking
MIT-Pt/SES-GI/0008/2008 and by Fundo de Apoio à Inovação (Ministério into the EV deployment context, the aforementioned central
da Economia, da Inovação e do Desenvolvimento) within the framework of
the Project REIVE – Redes Eléctricas Inteligentes com Veículos Eléctricos. entities can be seen as EV aggregators [3] with adequate
The authors are with Instituto de Engenharia de Sistemas e control capabilities that allow defining coordinated and
Computadores do Porto (INESC Porto) and with the MIT Portugal Program optimized charging profiles for all dependant EV.
on Sustainable Energy Systems from Faculdade de Engenharia da
Universidade do Porto (FEUP), Porto, Portugal.(e-mail:[email protected],
The role of central units, regarding the control of EV
[email protected]; [email protected], [email protected];). charge, can be outlined, for a LV radial grid without

978-1-4244-7658-9/10/$26.00 ©2010 IEEE 386

distributed generation, presenting voltage drop problems. In assumed that the converter will connect to a three-phase
such a grid, the exclusive automatic and independent 400V, 50 Hz grid, with standard power quality requirements.
response of EV battery chargers to local voltage deviations,
B. Battery Interface
will lead to recurrent penalization of EV connected at the
farthest buses from the feeding point. As these buses present Among the major types of chemistries, Li-ion type appears
lower voltages, small charging rates will be possible. to be the election battery for the expected Plug-In Hybrid EV
In the extension of the coordinated control features, EV and EV to be in the market in a near future [4]. Nevertheless,
aggregators can collect information about the EV willing to due to the lack of detailed information on Li-ion batteries,
provide the above described system services and sell them in the NIMH chemistry type was chosen for this study, as these
the electricity market. The envisaged synergetic approach batteries are currently used in some Hybrid EV and much
between EV and the grid can, not only improve the operating information is already available. The battery pack model
conditions, but also increase the number of electric vehicles used in this study is based in the Toyota Prius NHW11,
that can be safely integrated into distribution networks. which includes 38 modules of 6 Panasonic HHR650D cells,
In related several studies, the detailed models of power with 6.5 Ah and 1.2 V each, all of them in series, resulting in
electronic converters were replaced by models with simpler total 273.6 V and 1778 Wh (the largest of the Prius packs).
dynamic, based on their main control functions, as those are The NIMH battery model was included in the simulation
more adequate, when the goal is to perform macro scale platform using the SymPowerSystem block. The Simulink
studies, with a large number of power electronic converters battery model was presented and validated in [5], using data
operating simultaneously. However, the practical feasibility available on HHR650D cells. Model limitations, related to
of the lower detail models must be confirmed from the ageing or temperature effects, do not have relevant impact on
viewpoint of the power electronic converter control. To do the objectives of this work.
so, suitable electronic converter architecture must be chosen, The main concern of the charging system, on top of any
as well as a proper control approach, in order to achieve grid requirements, is to preserve the EV battery life,
satisfactory converter stability, robustness, and power quality implying that recommended charging methods must be
specifications. In this stage, a transient analysis focused on followed imperatively. Hence, the battery charger should
the converter detailed model should be done, in order to not always superimpose suitable charging rates and other
undervalue commutation inherent side effects, like harmonic constraints defined by the manufacturer, over the upstream
distortion, displacement factor and response delays. references generated by the grid interactive charging
Afterwards, upper control layers, based in measurements of management module, described later in section IV. The most
frequency or voltage deviations, should be applied to the important specifications provided by the NIMH battery
converter/battery system. The overall performance can be manufacturer are the following:
evaluated by integrating the full controlled model in a small --Standard charging: 0.1 C (0.65 A), requiring 16 hours;
islanded grid and analysing the converter dynamic behaviour --Fast charging: 1 C (6.5 A), requiring 1.2 hours
when load variations occur. (ranging from 0,5 C to 1 C is possible);
This work describes an integrated control solution --Fast charge transition restoration current: current is
developed for an EV battery grid interface and includes a limited within 0.2 C to 0.3 C as cell voltage is below 0.8 V;
study of the proposed power electronic converter control --Trickle charging current: 0.033 C to 0.05 C might be
behaviour, within a grid islanded operation scenario. All the applied after rapid charge or in deeply discharge conditions.
simulations were performed using the software Several timers for each charging stage and stop charge
Matlab/Simulink. triggering events (like maximum voltage cell), were also
defined (temperature related events were not considered, as
II. POWER ELECTRONIC CONVERTER SELECTION previously mentioned) [6], [7].
Power electronic converters are designed to efficiently C. Power Electronic Converter
interface electric systems with different voltage, current and The chosen converter should be an AC/DC current
frequency needs. Regarding EV, the power converter will controlled charger circuit, with a topology that allows the
manage the energy flow between the distribution grid and the bidirectional power flow between the grid and the battery.
battery, so each side requirements will be used as guidelines From the battery point of view, the converter should charge
in the converter selection process. with a controlled constant current, thus small current ripple
A. Grid Interface will be desirable. From the grid point of view the converter
must be a three-phase device, presenting phase currents with
As a first approach to the problem, this work considers a
low Total Harmonic Distortion (THD) and close to unity
three-phase grid connection, instead of a single-phase. This
displacement factor. Thus, the selection of an Improved
solution contributes to avoid voltage unbalances by reducing
Power Quality Converter (IPQC) is the better option. The
phase currents, as EV charging at its nominal rate represents
IPQC converters use solid-state switching devices like
a large amount of power being absorbed from the grid. It is
Insulated Gate Bipolar Transistors (IGBT) and reactive

387
 elements like inductors in the AC side and capacitors in the A. AC Current Control
DC side (helping to limit AC currents and DC voltages Nonlinear and time variant structures, like power
ripples). Hence, to achieve the mentioned specifications, a converters, restrain the use of linear control strategies and,
bidirectional three-phase and three-level diode clamped for that, it is appropriate to apply sliding mode control.
converter topology was adopted [8], [12]. This circuit can be By using the dq rotating reference frame (with grid
seen as a boost converter and so, looking toward the grid angular velocity input), the balanced symmetrical three-
nominal voltage and the battery voltage working range, a phase set of AC input currents can be represented with a null
step-down three-phase transformer was used in the grid quadrature projection (as unity displacement factor is
connection side, providing also galvanic insulation. required). The output of the upstream DC control loop is the
reference for the direct current on the dq frame, which will
control the amplitude of AC currents phasors. The grid
frequency estimation was done using the straightforward
zero-crossing method described in [9].
As aforementioned, each Vbj voltage can assume three
different discrete values, so 33 = 27 combinations for Vbj are
possible. Assuming that the capacitor voltages are equalized,
Vc1 = Vc2 = V/2, Vbj possible values can be referred to V
as -½, 0 and ½. Then ab0/αβ0 transformation can be applied
Fig. 1. Three-phase three-level diode clamped converter with IGBT Sji. to the voltage vectors [Vb1; Vb2; Vb3], resulting in 27
C1=C2=10 mF and L=8 mH.
different vectors in the αβ space. Doing the necessary
In Fig. 1 the Vj voltages sources are the secondary y-y transformations to the αβ0 frame, α and β errors between
step-down transformer voltages. Each converter leg can be references and measured AC currents can be computed. The
emulated by a three position switch as shown in the grey represented Vbj vectors define five different values in β axis
zone. Therefore, each Vbj voltage can assume three different and nine different values in α axis. However, when selecting
values, available from the capacitor divider. Those three a single α or β value, a maximum of five different values are
level voltages implicitly reduce the input currents distortions, defined in the orthogonal component. Hence, current errors
which is the main advantage of this converter. In the next can be quantified by means of a sum of four hysteretic
equations [10], γ j is the switch function for converter leg j: comparators output (λ_α and λ_β) with maximum and
 Vc1 se γ j = 1 ⇔ S j1 , S j2 ON ∧ S j3 , S j4 OFF minimum saturation levels of -½ and ½ (thus, outputting -2, -
 1, 0, 1 and 2). The dead-band of each comparator is centred
Vb j =  0 se γ j = 0 ⇔ S j2 , S j3 ON ∧ S j1, S j4 OFF , j = 1,2,3 (1)
− Vc se γ = −1 ⇔ S , S ON ∧ S , S OFF
 2 j j3 j4 j1 j2 in zero and has a width of ε,1.1×ε, 1.2×ε and 1.3×ε (ε is the
γj γj biggest error assumed as zero). The reduction of ε parameter
Vb j = ( )
γ j + 1 Vc1 +
24243
( )
1 − γ j Vc 2 = Γ1 j Vc1 + Γ2 jVc 2
24243
(2)
(related to the ripple of the AC currents) is limited by the
1 1
Γ1 j ∈{0;1} Γ2 j ∈{−1;0}
maximum switch frequency of the IGBT.
Considering a sliding surface defined as S(ei_α,β,t) =
III. CONVERTER CONTROL k(iα,β_ref - iα,β) = 0, finite switching frequency will imply
The converter will be controlled in current mode. To act S(ei_α,β,t) ≠ 0, so there is a certain dynamic error tending to
in the output current of the battery, an internal AC current zero, resulting in AC currents ripple. To assure stability of
control loop and an external DC current control loop are the system under sliding mode control, sliding mode
used, as shown in the Fig. 2. existence and reaching conditions must be fulfilled. The
 latter establishes that enough command action (Vbj value)
must be available, in order to handle perturbation effects,
while the former establishes that in sliding mode:
S (ei , t ) S& (ei , t ) < 0 (if S (ei , t ) > 0 → S& (ei , t ) < 0 , S (ei , t ) will decrease to

zero; if S (ei , t ) < 0 → S& (ei , t ) > 0 , S (ei , t ) will increase to zero) [10].
Using the last condition, if S (ei , t ) > 0 , iα , β must increase:
Quantification Error :

 d iα , β ref d iα , β  di (3)
 −  < 0 ⇔ α,β > 0
 dt dt  dt

Vb_

Thus, according to the Fig. 1 definitions, a vector with

component

negative value Vγ,β must be applied. Reciprocal logic can be

applied if S(ei_α,β,t) < 0, resulting in the choice of a vector
Fig. 2. AC and DC (grey area) currents controls. Representation of Vbj with positive Vγ,β value, with suitable magnitude. According
voltages in αβ frame and vector selection strategy. to the error quantification [λ_α, λ_β], the vectors can be

388
selected as shown in Fig. 2. However for some error IV. GRID INTERACTIVE CHARGING MANAGEMENT
combinations, no vector seems suitable (for [-2,-2] the In order to provide ancillary services to the distribution
nearest vector is the 3, and this vector will be selected also grid, the proposed EV charging process must respond to
for [-1,-2]). Variable switching frequency is obtained as a frequency and/or voltage variations. Frequency is an
consequence of the appliance of sliding mode control, given instantaneous indication of the power balance, so if the
that switching will occur at the needed moment. frequency drops, it will be desirable to decrease the charging
As shown in Fig. 2, univocal vector selection is not always rate of connected EV (reverse observation is valid, mutatis
possible (e.g. both 13 and 26 vectors can be selected for mutandis). EV charge reduction can also help to correct
error [1, 1]). The resulting degree of freedom can be used for eventual voltage sags that might appear when a considerable
capacitors equalization, as can be concluded from (4). number of EV are charging simultaneously in the farthest
dVc k Γk1 i1 + Γk 2 i2 + Γk 3 i3 I buses from the feeding point.
= − o , k is the capacitor index (4)
dt Ck Ck The EV charging current, as function of the grid
frequency/voltage deviations, will then assume a
Vector 26 has Γ1 = (0, 0, 0) and if Io > 0 then dVc1/dt < 0 proportional nature, with positive slopes outside a dead-band
and capacitor C1 discharges; vector 13 has Γ2 = (0, 0, 0) and centred at nominal frequency/voltage and saturating at the
if Io > 0 then dVc2/dt < 0 and capacitor C2 discharges. Each considered absolute maximum and minimum values. This
of the 27 represented vectors in αβ space correspond to a kind of proportional control is called “droop control” [11],
unique combination of the three γj switching functions, so [12]. Fig. 4 presents an example for frequency and voltage
vectors can be translated in ON/OFF signals for IGBT gates. droop. For this case, the dead-band was defined
B. DC Current Control symmetrically around the nominal frequency/voltage and,
within this range, the charging rate is kept constant and equal
The application of the 123/dq0 transformation to equation
to a set-point value, so the autonomous control is immune to
(4) parameters, results in the following expression for C1:
frequency/voltage variations inside. Droop saturation
Γ1d = Γ1α ⋅ cos(θ ) + Γ1β ⋅ sin(θ ) maximum (1 C) and minimum (0) values must be defined to
dVc1 Γ1d I
= id − o , with Γ1α = 2 / 3 ⋅ (Γ11 − 1 / 2 ⋅ Γ12 − 1 / 2 ⋅ Γ13 ) (5)
dt C1 C1 occur, respectively, above and below the grid admissible
Γ1β = 2 / 3 ⋅ (− 3 / 2 ⋅ Γ12 + 3 / 2 ⋅ Γ13 )
maximum and minimum frequency/voltage values.
200 210 220 230 240 250 260
By using the sliding mode control, AC currents follow the 7
Voltage
reference values with some delay (Td), related to the 6.5 = 1C
dead-band
switching frequency, 6
-5% +5%
1st Order Term
exp Taylor Series
id ( s) = id _ ref ( s) ⋅ e − sTd    → id ( s) = 1 /(1 + s ⋅ Td ) ⋅ id _ ref ( s) (6) 5
Charging Current [A]

The voltage output dynamic can be neglected when 4

compared to the dynamic of the currents and, if so,

expression (5) is simplified to Io = Γ1d×id. Additionally, 3
Frequency
assuming unity displacement factor (iq=0) and that converter dead-band
2
losses are neglected, the input and output powers are the
same and equal to V×Io=Vd× id, where Vd is the direct 1
-0.2% +0.2%

component of the three-phase input voltages. Therefore, a

simplified model for the converter, is presented in Fig. 3. 0
48,5 49 49,5 50 50,5 51 51,5
Grid Frequency [Hz]
Fig. 4. Frequency and voltage droops : definition of dead-bands and set-
point of 4A for the charging rate
Fig. 3. DC current control loop with simplified converter. Under this approach, the charging rate of EV will be
In order to obtain a zero static error, integral action is adjusted taking into account the frequency droop or voltage
needed, thus a PI controller was selected. The zero of this droop output.
controller was selected in order to cancel the high frequency Despite slight propagation effects, frequency is a global
pole, which is related to the dynamics of the converter AC scope variable, as any deviation is simultaneously and
currents. Therefore Tz=Td=1/(2πfs), where fs is the switching equally transmitted to all points of the grid. On the other
frequency. Finally, manual fine-tuning of PI parameters was hand, voltage is a local variable which is very dependent on
performed. local load profiles. So, for a particular EV charging control,
The control strategies presented in this section are widely local voltage conditions will always prevail over frequency
used with well proven performance, in many actual variations, given that voltage can only be corrected by EV
electronic converter industrial applications. connected locally, while frequency problems might be solved

389
with the contribution of any EV connected to the system. For this set of simulations, constant 6.5 A = 1C rapid
Additionally to some parameters and set-points, in order to charging current is imposed to the battery pack, with 2.45%
manage the coexistence of both droops, the central units can SOC, in order to point out the action of the charger control in
send a boolean variable (depicted as “voltage droop priority assuring the battery requirements. Other basic control goals
signal” in Fig.5), determining the EV behaviour. If the defined above will also be checked. The related effects of the
predetermined default value is sent, the addressee EV will CB switching events are shown in Fig. 7.
react only to voltage variations. If the complementary value The voltage drop near 1.9 s of simulation time occurs, not
is sent, the voltage droop is active, but the frequency droop as consequence of any CB switching events, but as a result of
control is responsive to frequency deviations, only as long as the charging rate increase reported in Fig. 8.
the local measured voltage is within its droop dead-band. 8
Charging Current - Battery Requirements

In conclusion, the EV charging controller can react locally Cell Voltage < 0.8 V

Charge Current [A]

6
Rapid
Rapid Charge
and autonomously to frequency and voltage measurements, 4
Transiton
Voltage
Charge
Current
1 C = 6.5 A
Restorarion
based in the defined droops. The signals received (Fig. 5) Current
0.2 C = 1.3 A
2
from central control units should take into account some Charge Current
Current Io Reference

previous inputs collected periodically from connected EV: 0

0 0.5 1 1.5 2
Time [s]
2.5 3 3.5 4

batteries Sate of Charge (SOC), user defined period of Fig. 8. Charger Control action to meet battery charge rate requirements.
charge and desired SOC at the disconnection time, etc.
The upstream reference for the charging current is set to
6.5 A, however, the charge controller only outputs this value
when SOC increases so that the cell voltage rises above 0.8
V. Until that, reduced current is applied (rapid charge
transition voltage restoration current). A fast step response
without oscillations was obtained, thus pointing to a good
Fig. 5. “Droop Control” block outputs will be subject to constraints dynamic response. Steady state error is negligible and the
imposed by the “Charge Control”. charging current presents small ripple. Thus, good tracking
behaviour is obtained and immunity to voltage and frequency
V. SIMULATION MODEL AND RESULTS disturbances is verified.
The evaluation of the performance of the described In Fig. 9 are represented the waveforms of the line
control and management strategies is tested in a small currents in the grid side of the converter transformer. The
islanded grid represented in Fig.6. currents have a sinusoidal shape, with a frequency of 50 Hz,
and form a three-phase symmetric and balanced system.
Small currents switching ripple results from the high
switching frequency combined with filtering effect of the line
inductors. Thus, currents THD values are below the 5% grid
standard limit. The second graph of Fig. 9 shows a tiny or
Fig. 6. Grid model used for EV charger test. absent displacement angle, that results in a close to unity
The grid has a single synchronous machine generator with displacement factor (the converter behaves almost like a
its own PI controller, in order to perform primary and resistive variable load).
secondary frequency control; the machine has also voltage 4
Grid Side Input AC Currents

regulation. The circuit-breaker CB will switch on a load with

AC Currents[A]

considerable value. CB “switch on” and “switch off” events 0

-2
occur at 0.5 s and 2.5 s, respectively.
-4
1.8 1.82 1.84 1.86 1.88 1.9 1.92 1.94 1.96 1.98 2
Time [s]
Phase Shift Between Line Current and Line Voltage - Displacement Factor near to unity
A. Constant Rate Charging 3
Line Voltage
400

Line Current
AC Line Current

AC Line Voltage

Load Switch On (t=0.5s) and Switch Off (t=2.5s) - Frequency Variation and Estimation 1.5 200
51
Frequency Estimation 0 0
Frequency Variation
Frequency [Hz]

50.5
-1.5 -200

50 -3 -400
1.8 1.82 1.84 1.86 1.88 1.9 1.92 1.94 1.96 1.98 2
Time [s]
49.5
Fig. 9. Three phase currents and phase shift.
49
0 0.5 1 1.5 2 2.5 3 3.5 4
Time [s]

360
Load Switch On (t=0.5s) and Switch Off (t=2.5s) - Line Voltage magnitude Estimation B. Grid Interactive Charging Management
Line Voltage Magnitude Estimation
350
Line Voltage For this set of simulations, battery pack SOC was set in
Line Voltage [V]

340
Voltage drop due to
330 charging rate rising 10%, thus the battery charge controller doesn’t restrain the
320
310 upstream current reference output from the frequency droop
300
290
0 0.5 1 1.5 2 2.5 3 3.5 4
control presented in Fig. 4. Voltage droop was disabled.
Time [s]
As shown in Fig. 10, after a transient initial overshoot, the
Fig. 7. Frequency and voltage grid values and respective measurements.

390
 charge current tracks, with delay of tens of ms, the frequency VI. CONCLUSION
variation according to the above mentioned frequency droop. In this paper, the technical feasibility of mentioned EV
The charging rate is immune to frequency variations within charging management strategies is confirmed by means of
the defined dead-band. Data cross-checking between Fig. 11 electronic power converter selection and application of
and Fig. 4 provide also consistent results (e.g. for 49.5 Hz, a proper control techniques. Besides the single application of
near to 2A charging current is obtained, while for 50.5 Hz it the frequency droop, a method to handle the coexistence of
is obtained 5A). voltage and frequency droops was also developed.
Three-Phase Sinusoidal Currents Magnitude Contour
Currently, it is being developed a single-phase model to
conduct simulations similar to those performed in this paper,
Grid Input AC Currents[A]

as single-phase connections are much more common than

0
three-phase ones. The work being presently developed will
not only contemplate EV load controllability, but also its
-5
0 0.5 1 1.5 2
Time [s]
2.5 3 3.5 4
capability to provide power to the grid (V2G concept). EV
51
Interactive Charge Management - Frequency Droop Control (Set-Point: 4A; Dead-Band: [49.9;50.1] Hz)
Droop Action Frequency deviation
control strategies will be enhanced in order to improve EV
50.75 7
potential to provide ancillary services to the grid [14].

Charge Current [A]

Charge Current
Frequency [Hz]

50.5 6
50.25 5
50 Dead Band Set-Point 4
49.75 3
49.5 2 REFERENCES
49.25 1
Droop Action
49
0 0.5 1 1.5 2 2.5 3 3.5
0
4 [1] J. A. Peças Lopes, F. J. Soares, P. M. Rocha Almeida, Patrícia C.
Time [s]
Baptista, Carla M. Silva, Tiago L. Farias, “Quantification of
Fig. 10. Frequency droop control – charging rate and resulting AC Technical Impacts and Environmental Benefits of Electric Vehicles
currents. Integration on Electricity Grids”, ELECTROMOTION 2009, Lille,
France, July, 2009.
[2] J. A. Peças Lopes, C. L. Moreira, A. G. Madureira, “Defining Control
The magnitude of AC currents is shaped according to the
Strategies for MicroGrids Islanded Operation”, IEEE Transactions
charging current values. On Power Systems, vol. 21, Issue. 2, pp. 916-924, May, 2006.
In Fig. 11 it is depicted the default behaviour of the [3] C. Guille and G. Gross, "A conceptual framework for the vehicle-to-
control management, using both frequency and voltage grid (V2G) implementation", Energy Policy, vol. In Press, Corrected
Proof.
droops. So when the measured voltage is within its droop [4] F. Nemry, G. Leduc, A. Muñoz, “Plug-in Hybrid and Battery-Electric
dead-band, the frequency droop control is active and can Vehicles: State of the research and development and comparative
react to frequency deviations. Otherwise, only voltage droop analysis of energy and cost efficiency”, JRC Technical Notes - Annex
I, October, 2009.
is active, reacting to eventual voltage variations. [5] Olivier Tremblay, Louis-A. Dessaint, Abdel-Illah Dekkiche, “A
Generic Battery Model for the Dynamic Simulation of Hybrid Electric
260
Interactive Charge Management - Voltage Droop Control (Set-Point: 4A; Dead-Band: 230 ± 5%)
8 Vehicles”, Vehicle Power and Propulsion Conference, Arlington,
Charge Current Voltage Frequency
7 Texas, USA, pp. 284-289, September, 2007.
Grid Line Voltage [V]

252.5
Droop Action Droop
Charge Current [A]

Line Voltage
245 Active 6
237.5 5 [6] Panasonic Nickel Metal Hydride Batteries Technical Handbook.
Voltage Dead Band
230 4 [7] Nickel Metal Hydride Batteries: Individual Datasheet HH650D.
222.5 3
215 2 [8] Bhim Singh, Brij N. Singh, Ambrish Chandra, Kamal Al-Haddad,
207.5
200
Frequency Droop Active
1
0
Ashish Pandey, Dwarka P. Kothari, “A Review of Three-Phase
0 0.5 1 1.5 2 2.5 3 3.5 4
Time [s] Improved Power Quality AC-DC Converters”, IEEE Trans. Ind.
51
Interactive Charge Management - Frequency Droop Control (Set-Point: 4A; Dead-Band: 50 ± 0.2%) Electron, vol. 51, Issue 3, pp. 641-660, June, 2004.
Frequency Droop Frequency
50.75
Action Charge Current
7 [9] V. Backmutsky, V. Zmudikov, “Accurate Frequency Estimation in
Charge Current [A]

6
Frequency [Hz]

50.5
50.25 5
Power Systems by DSP”, Proceedings of 18th IEEE Conference in
50 4 Israel, Tel-Aviv, Israel, pp. 5.2.4/1 – 5.2.4/5, May, 1995.
49.75 3
49.5 2 [10] Muhammad H. Rashid, “Power Electronics Handbook”, Academic
49.25 1 Press, 2nd ed., ch. 34, 2006, pp.955-993.
49 0
0 0.5 1 1.5 2
Time [s]
2.5 3 3.5 4
[11] J. A. Peças Lopes, P. M. Rocha Almeida, F. J. Soares, N.
Hatziargyriou, “Electric Vehicles Grid Integration Under the
Fig. 11. Frequency and voltage droops action – Frequency droop enabled. MicroGrid Concept”, Smart Grids and Mobility - Smart Grids and
Mobility, Würzburg, Germany, June, 2009.
If the “voltage droop priority signal” is 1, only the voltage [12] J. A. Peças Lopes, Silvan A. Polenz, C. L. Moreira, Rachid
droop is enabled, as shown in Fig. 12, after 2.5 s. Cherkaoui, "Identification of control and management strategies for
LV unbalanced microgrids with plugged-in electric vehicles." Electric
Power Systems Research, Vol.80, Issue 8, pp. 898-906, August, 2010.
Grid Line Voltage Magnitude [V]

Interactive Charge Management - Voltage Droop Control (Priority Droop Signal Action)
260
Charge Current Voltage
8 [13] Sharanya Jaganathan, Wenzhong Gao, “Battery Charging Power
252.5 7
Electronics Converter and Control for Plug-in Hybrid Electric
Charge Current [A]

Line Voltage Droop Action

245 6
237.5 Voltage Dead-Band 5 Vehicle”, Vehicle Power and Propulsion Conference, Dearborn,
230 4
222.5 3 Michigan, USA, pp. 440-447, September, 2009.
207.5
215 2
1
[14] J. A. Peças Lopes, P. M. Rocha Almeida, F. J. Soares, “Using
200
0 0.5 1 1.5 2 2.5 3 3.5
0
4
Vehicle-to-Grid to Maximize the Integration of Intermittent
Time [s]
Renewable Energy Resources in Islanded Electric Grids”, ICCEP
Fig. 12. Frequency and voltage droops action – Frequency droop disabled. 2009 - International Conference On Clean Electrical Power
Renewable Energy Resources Impact, Capri, Italy, June, 2009.

391

View publication stats