You are on page 1of 8

IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO.

4, JULY 2014 1815

Impact of Electric Vehicle On-Board Single-Phase


Charging Strategies on a Flemish Residential Grid
Niels Leemput, Student Member, IEEE, Frederik Geth, Student Member, IEEE,
Juan Van Roy, Student Member, IEEE, Annelies Delnooz, Jeroen Büscher, Member, IEEE, and
Johan Driesen, Senior Member, IEEE

Abstract—This paper quantifies the impact of single-phase are determined by the departure time of the next trip and re-
on-board charging strategies for electric vehicles (EVs) in a quired range. This flexibility can be used to include an addi-
case study of a heavily loaded unbalanced Flemish three-phase tional objective besides the mobility objective, to determine a
low-voltage residential grid. Voltage droop charging and EV-based
peak shaving, which do not need communication with the dis-
unique charging pattern.
tribution grid, are modeled and the results are compared. The The scope of this paper is to assess the residential grid impact
grid voltages are analyzed according to the probabilistic and of EV charging strategies. An unbalanced three-phase load
deterministic limits of the EN50160 standard, for a 100% EV flow analysis is performed. A realistic residential feeder with
penetration rate. The impact on the EV user comfort is evaluated real household power profiles is modeled. Uncontrolled and
in terms of charging time and electrically driven distances. The on-board controlled charging strategies are compared.
chosen voltage droop charging eliminates critical voltages below
0.85 pu and reduces voltage unbalance, with a limited impact The paper is structured as follows. Section II discusses the
on the total charging time. EV-based peak shaving makes the literature on coordinated charging, the resulting impact on the
grid fully compliant with EN50160 and avoids the need for an residential distribution grid and locally controlled EV charging
infrastructure upgrade. The electrically driven distances are not strategies. In Section III, the scenario and modeling are dis-
influenced by the charging strategies. cussed. The results are discussed in Section IV, followed by the
Index Terms—Coordinated charging, electric vehicles, smart conclusions in Section V.
grids.
II. BACKGROUND
I. INTRODUCTION
A. Coordinated Charging

E LECTRIC vehicles (EVs) are introduced into the market


nowadays. The electric powertrain allows to reduce the
consumption of greenhouse gas emitting fossil fuels [1]. Fur-
An extensive amount of research has already been performed
on coordinated charging of EVs [5]. Typically, it is concluded
thermore, the absence of tailpipe emissions reduces the local that coordinated charging strategies can reduce the impact on
concentrations of pollutants harmful to human health [1]. the power system.
Both battery electric vehicles (BEVs) and plug-in hybrid Charging can be coordinated to integrate a higher share of
electric vehicles (PHEVs) are charged through the electric intermittent renewable energy in the power system, thereby re-
power system. An increasing number of EVs increasingly ducing the emissions related to electric power production [6]–[8].
impact the power system, e.g., peak power demand, load Also, the charging flexibility of EVs can be used to provide grid
profile and voltage magnitude deviations [2]. The residential services, e.g., congestion management [9], frequency regulation
low-voltage grid impact may be significant, due to the simul- [10] and imbalance reduction [11]. The charging flexibility can
taneity between the residential power peak and the plugging also be used to minimize the charging cost, by anticipating on the
in of EVs when arriving at home, which starts the charging variable electricity prices on the wholesale markets [12]–[14].
process for uncoordinated charging [3]. These charging strategies require a substantial penetration
There is a certain flexibility for EV charging, due to the long rate on the regional/national level to have a noticeable impact,
standstill time and the relatively low distances driven [4], which which would justify the required investments in smart grid in-
frastructure [15]. Therefore, widespread coordination is not ex-
Manuscript received June 28, 2013; revised November 07, 2013, January 07,
pected to occur in the near-term future. Furthermore, high local
2014; accepted February 20, 2014. Date of current version June 18, 2014.This EV concentrations, which impacts the local distribution grid,
work was supported by the European Institute of Innovation and Technology may occur before a significant penetration rate occurs on the
through the EV City project organized by the KIC InnoEnergy. The work of N. regional/national level. Because it is expected that EVs will
Leemput was supported by a Ph.D. grant from the Institute for the Promotion
of Innovation through Science and Technology in Flanders (IWT-Vlaanderen). mainly be charged at home, the residential grid impact can be
The work of J. Van Roy was supported by a VITO doctoral scholarship. KU substantial in areas with many EVs.
Leuven and VITO are partners of EnergyVille, 3600 Genk, Belgium. Paper no.
TSG-00481-2013. B. Residential Grid Impact
N. Leemput, F. Geth, J. Van Roy, J. Büscher, and J. Driesen are with the De-
partment of Electrical Engineering (ESAT), Division Electa, KU Leuven, 3001 Residential EV charging impacts different aspects of the dis-
Leuven, Belgium (e-mail: niels.leemput@esat.kuleuven.be).
A. Delnooz is with Flemish Institute for Technological Research (VITO),
tribution grid: power profile, voltage magnitude deviations, and
2400 Mol, Belgium. voltage unbalance. The impact of coordinated charging on these
Digital Object Identifier 10.1109/TSG.2014.2307897 aspects are discussed in this section.

1949-3053 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1816 IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 2014

1) Power Profile: EV charging influences the power profile the EV and the distribution grid infrastructure, e.g., the voltage
of the distribution system [3], which impacts the aging of trans- magnitude and the EV departure time.
formers and cables. The aging is a strong function of the hot If the voltage magnitude is above 0.90 pu and below 1.10 pu,
spot temperature, which depends on the transformer and cable the value is within the normal operating conditions [23]. If EV
conductor currents [16], [17]. Thermal effects significantly in- chargers are given responsibility to support voltage standards,
fluence the technical lifetime of transformers and cables caused charging is stopped below 0.85 pu, which is the lower voltage
by the accelerated aging of the insulation material. limit in the measurement procedure of [23]. Therefore, control
EV charging could be controlled in a way to smoothen out the of the EV charger power is to take place between 0.85 and 0.90
power profile, e.g., peak shaving on grid level [18], [19]. Also, pu. Voltage droop charging, as discussed in [29] and [30], can
the power peak could be taken into account as a constraint in be used for this purpose.
cost-minimizing coordination strategies [20], [21]. These pro- The EV user could inform the EV charger about the next
posed strategies require communication between the EVs and departure time through an on-board communication interface.
the distribution grid. Thus, in case of malicious or absent com- Then, the EV charging power rating can be reduced based on the
munication, these strategies are ineffective. required charging energy and the time until the next departure.
2) Voltage Magnitude Profile: The distribution voltage As a result, the EV charger power peak is reduced, while still
magnitude should stay within a regulated range, to ensure fully charging the EV before the next departure. To the author’s
that electric appliances are operated in a safe and satisfactory knowledge, the combination of such an EV-based peak shaving
manner [22]. The European EN50160 standard specifies that strategy and voltage droop enabled chargers has not been dis-
the 10 min mean rms voltage deviation should not exceed cussed in the literature.
10%, measured on a weekly base [23]. For under voltage, a Voltage dependent charging and EV-based peak shaving are
wider range is allowed in the measurement procedure: 15% used separately and combined in this paper. The impact on the
to 10% for 5% of time. EV user comfort is discussed in terms of charge duration in-
The impact of uncoordinated EV charging on the voltage crease and electrically driven distances for a fleet with differ-
magnitude is discussed in the literature [24]. Peak shaving, as entiated vehicle categories. The commonly used single-phase
discussed in the previous subsection, reduces the simultaneity of charging power rating of 3.3 kW is assumed. A typical Flemish
household and EV power demand, which has a beneficial impact urban low voltage grid model is used. The voltage constraints
on the voltage deviations [18]. The reduction of voltage devia- are evaluated according to the probabilistic and deterministic
tions could also be considered as a coordination objective [3]. limits of the EN50160 standard, for a 100% EV penetration rate.
However, both strategies require communication, which makes The analysis is performed for a half-year simulation horizon
them ineffective in case of absent or malicious coordination. with a 1-min time resolution.
Furthermore, the nodal voltage is influenced by the power pro-
file of the individual households, which is highly stochastic and III. SCENARIO AND MODELING
unpredictable [22]. Therefore, many of the proposed algorithms A. Residential Low-Voltage Grid
are used for benchmarking.
3) Voltage Unbalance Profile: The uneven distribution of The feeder topology is based upon a real urban LV feeder
single-phase loads [25] and asymmetric conductor config- with a TT grounding arrangement located in Flanders, Belgium.
urations [22] lead to unequal voltage drops in the different The information is provided by a Flemish DSO, a partner in the
phases of the distribution grid. An excessively high unbalance EVCity project [31]. The 39 residential loads are grid con-
negatively impacts the power system, as well as electric and nected through 29 nodes , as illustrated in Fig. 1. The distance
electronic end-user appliances [22]. In European standards, the between these nodes and the transformer varies between 310
voltage unbalance factor VUF is defined as the magnitude of and 550 m. All households are assumed to have a single-phase
the ratio of the negative sequence voltage to the posi- grid connection between one of the three phases
tive sequence voltage . The and the common neutral conductor ; the nominal neutral-to-
EN50160 standard specifies that the 10-min mean rms value of phase voltage is 230 V.
the VUF should be below 2% for 95% of time, measured on a Cable parameters are taken from the design specifications of
weekly base [23]. the standard for underground distribution cables, NBN C33-322
The impact of EV charging on the VUF is not to be ne- [32]. Cable type EIAJB 1 kV mm is used for
glected, as discussed in [26]. For three-phase EV charging, the the main feeder. Cable type EXVB 1 kV 4 16 mm is used to
power offtake from each phase could be varied to compensate connect the household supply terminals with the main feeder,
voltage unbalance. Methods being proposed for the injection of except at node 28, where cable type EXVB 1 kV 4 35 mm
power by three-phase grid inverters of photovoltaic installations is used, because of the high load that is connected there. These
[27], [28], could be adapted for three-phase EV charging. For cables are between 5.3 and 14.8 m in length.
single-phase EV charging, the charger has no local three-phase The main feeder is connected to a 400 kVA transformer, to-
voltage measurement. Therefore, it has no information on the gether with five parallel feeders that have a length of 250 m,
VUF, only on the voltage magnitude of the phase to which the which are modeled in a simplified manner here. The cable type
EV is connected. is the same as for the main feeder, and the load connected to the
end of each of the feeders is equal to the aggregated load of the
main feeder. In this way, the 400-kVA transformer is serving
C. Locally Controlled EV Charging
the equivalent of 234 residential loads, which is in line with
Several local parameters can be used to control the EV the DSO’s deployment strategy. The transformer impedance of
charging process without the need for communication between is derived from [33].
LEEMPUT et al.: IMPACT OF EV ON-BOARD SINGLE-PHASE CHARGING STRATEGIES 1817

Fig. 1. Schematic overview of the residential low voltage feeder topology with 29 connection nodes serving 39 households .

B. Residential Load and Generation 2) Battery Model: The battery model from [38] is used. The
evolution of the energy content and state of charge at
A stochastically representative set of 39 single-phase
each 1-min time step is calculated as follows:
Flemish household electric power consumption profiles was
sampled in 2008 [34], with a 15-min time resolution. The (1)
profiles are self-identified in the survey as within an urban
environment. Only active power consumption was measured, (2)
so reactive power consumption is neglected in the simulations.
The households are alternatingly connected between the three with the self-discharge rate of the battery during each time
phases and the neutral conductor. step , which is chosen at 3% of the nominal battery capacity
The total annual residential photovoltaic (PV) energy produc- per month. The nominal battery capacities for the three
tion in Flanders amounts to 10% of the total annual residential vehicle categories are respectively 10, 15, and 20 kWh. and
energy consumption [35]. Because these installations are not are the charging and discharging power during each time
uniformly distributed, clusters with a high PV penetration rate step . Obviously, an EV cannot be simultaneously charging
occur. Therefore, a PV penetration rate of 35% is assumed here: and discharging .
14 houses are assigned a PV installation and their location is il- The charging and discharging efficiency and are 95%,
lustrated in Fig. 1. which are typical values for lithium-ion batteries. The usable
The PV power profiles are scaled versions of a measurement battery capacity is limited to 80% of the nominal capacity
on an installation of the KU Leuven, with a 15-min resolution. , to extend the battery cycle life
The profile is scaled to match the annual energy production to [39]. The EVs are modeled as extended range electric vehicles
the annual household energy consumption at the selected loca- (EREVs), in which the combustion engine is only used when
tions, considering a maximum inverter power rating of 5 kVA the battery is depleted. Therefore, even if the battery is depleted,
imposed by regulations [36]. all mobility requirements are met.
3) Charging Model: It is assumed that Mode 3 charging in-
C. EV Charging Load frastructure is used [40], which is directly connected to the grid
through a dedicated power circuit. The connection between the
The set of EV charging profiles are based on Flemish charging infrastructure and the EV occurs through a specific
mobility behavior and are generated with a 1-min time resolu- plug/socket, that also handles the communication on the allowed
tion , as discussed in [4]. Three sub-models are used to create charge current rating. Also circuit braking, fault current and
the charging load: a mobility behavior model, a battery model, over-current protection are located in the off-board charging in-
and a charge behavior model. Each household is as- frastructure.
signed an . According to IEC 61851 [40], the current and voltage rating
1) Mobility Behavior Model: The mobility profiles of each for Mode 3 charging range from a single-phase 16 A/230 V con-
EV is created with the mobility simulation tool discussed in nection up to a three-phase 63 A/400 V connection. The com-
[4]. This tool uses statistical data on Flemish transportation be- monly used charging power rating of 3.3 kW is assumed here.
havior, to create a realistic driving pattern for each vehicle. For This rating can be delivered through a single-phase grid con-
each vehicle, it is known when it is driving, standing still, and nection, which results in a charging current of 16 A at 90% of
also where it is when standing still: at home, at work or at an- in continental Europe [23].
other location. The EVs are assumed to be grid-connected at home and at the
The diversity of vehicle types and vehicle fuels within the workplace, whenever they are standing still at these locations
fleet, and the resulting variations in specific power consump- for more than 15 min. For shorter standstill times, it is assumed
tion and yearly driven distances are taken into account. Three that the EV user will not make the effort to connect the vehicle
vehicle types are modeled: subcompact, midsize and large ve- to the grid, because of the limited SOC increase during such a
hicles. The specific power consumption results of [4] are in- short time span.
creased with 15%, to take into account the impact of external
factors, e.g., ambient temperature, wind, altitude, road grade, D. Controlled Charging Cases
and surface [37]. Different charging cases are discussed here, as summarized
The resulting specific power consumption for each vehicle in Table I. These cases are compared to the case without EVs,
category is respectively 0.181, 0.213, and 0.302 kWh/km. Given i.e., case 0.
the average vehicle speed in Flanders of 42 km/h, the average 1) Case 1: The first case represents uncoordinated charging.
battery discharging power for each vehicle category is, re- The EVs start charging immediately when they are grid con-
spectively, 7.6, 8.9 and 12.7 kW during driving. nected for more than 15 min, until they are fully charged or
1818 IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 2014

TABLE I In case 2a, a constant power behavior is used .


SUMMARY OF THE CHARGING CASES Case 2b uses a voltage droop behavior to determine , as illus-
trated in Fig. 2. The droop behavior determines the maximum
charging power

(6)

(7)

e.g., if the voltage magnitude is 0.875 pu, the maximum


charging power is limited to 1.65 kW, i.e., 50% of 3.3 kW.
If the charging strategy already limits the charging power
to 1 kW, the charging process will not be influenced. If the
voltage droop influences the charging process, the charging
power will be adapted in the following time step to get the EV
fully charged by the next departure, given the charging power
limit of 3.3 kW.
Fig. 2. Piece-wise linear voltage droop charging behavior .
E. Load Flow Analysis

leave for the next trip. Thus, the charging flexibility is com- A three-phase unbalanced load flow algorithm is imple-
pletely used to maximize the user comfort, because the battery mented in MATLAB. Unbalanced loads are taken into account,
is charged as soon as possible. as well as the resulting zero-point shift due to currents running
In case 1a, the charging power is equal to 3.3 kW, the through the common neutral conductor . The backward–for-
battery charger has as a constant power behavior under all cir- ward sweep technique is used, because of the radial layout of
cumstances. In case 1b voltage droop charging is used [29]. The the grid [42].
charging power depends on the voltage magnitude at the phase The problem is converged when the worst time step voltage
and connection node where the EV is grid connected: error is below 0.1 V. A supplementary convergence crite-
rion is used for the EV loads: the worst time step power error
(3) should be below 33 W, which is 1% of the charging power
rating. All loads are modeled with unity power factor. The loads
where is the voltage magnitude at phase of grid con- are modeled as constant power loads for case 0, 1a and 2a. For
nection node on time step . The range between the deter- case 1b and 2b, the voltage dependent EV charging behavior, as
ministic and probabilistic voltage limit of the EN50160 stan- illustrated in Fig. 2, is implemented within the backward–for-
dard measurement procedure is fully used here to adapt the EV ward sweep.
charging power. linearly increases from 0 to 1 at
voltage magnitudes between 0.85 and 0.9 pu, as illustrated in F. Simulation Specifications
Fig. 2. This will increase the charging time when the voltage The simulations are conducted for 25 randomly selected
magnitude drops below 0.9 pu. weeks out of the proposed one-year scenario, resulting in a
EV chargers make use of an active rectifier, which measures number of time steps . For each 1 min time step
the voltage waveform to control the charging current waveform. , the evolution of the SOC is calculated as formulated in (1)
The charging power can be adapted within a wide range, which and (2).
is required to stay below the maximum charging current that is The resulting residential EV charging profile is added to the
allowed by the Mode 3 charging infrastructure. Thus, voltage residential load and generation profile at the respective grid
droop could be implemented in current Mode 3 compliant EV node. These load and generation profiles are kept constant
chargers. during their 15-min period. The resulting nodal load profiles
2) Case 2: For this case, the charging power is dependent are used to conduct the load flow simulation. EVs are also
on the time until the next departure and . The charged at the workplace, but this is not within this residential
charging power set point is equal to the minimum power grid. Therefore, the workplace charging profiles are not added
that is needed to get the EV battery fully charged during . to the residential load and generation profiles.
To avoid a too low efficiency at partial load, the charging
power set point lower limit is 25% of 3.3 kW, i.e., 825 W: IV. RESULTS
(4) A. Charging Behavior
The EV charging simultaneity is defined as the number of
(5) EVs that are charging simultaneously at the feeder, on which
each of the 39 households has an EV. The simultaneity is quan-
tified in Table II for the different cases. For all cases, the max-
e.g., if an EV is standing still for 10 h until the next departure, imum charging simultaneity is below 30 EVs, and even below
and 10 kWh is needed to recharge the battery, is 1 kW. 20 EVs for case 1a and 1b. During more than 50% of time, at
LEEMPUT et al.: IMPACT OF EV ON-BOARD SINGLE-PHASE CHARGING STRATEGIES 1819

TABLE II
EV CHARGING SIMULTANEITY

Fig. 4. Charging simultaneity for case 1a (black curve) and 1b (gray area).

TABLE III 1b than for case 1a. This is due to the charging time increase
EV CHARGING TIME
that occurs if the droop behavior is active, which compensates
for the charging power reduction.
For three households , each on a different
phase, the voltage magnitude and EV charging profile are illus-
trated in Fig. 5 for case 1a (black curve) and 1b (gray area). The
EVs at and reduce their charging power in case
1b, because the voltage magnitude goes below 0.9 pu. The re-
sulting charging time increase and the substantial impact on the
voltage magnitude can be observed.
The voltage droop behavior of all EVs that face voltage mag-
nitudes below 0.9 pu during charging influences all nodal volt-
ages on the feeder. This can be seen at , where the voltage
profile changes, while the charging profile of the EVs is iden-
tical. Also, there are changes in the voltage profile when the EV
is not charging, due to the voltage droop behavior of other EVs
that are charging.
Fig. 3. Duration increase per EV of cumulative charge events for case 1b versus
case 1a. C. Power Profile
The maximum transformer load , delivered energy ,
least 3 EVs are charging in case 1a and 1b, and 9 EVs in case 2a
and grid losses are summarized in Table IV, for both posi-
and 2b. The higher simultaneity for case 2a and 2b is due to the
tive (pos) and negative (neg) values. The negative power values
spreading of the charging process over the full standstill time at
occurs at times when instantaneous PV power production ex-
home.
ceeds power consumption in the LV grid. This is due to the
The impact of the different charging cases on the total EV
difference in timing of PV power production and household
charging time is summarized in Table III. The charging time
power consumption. The load on each conductor is defined
increase in case 1b versus 1a is less than 5% for all vehicles
as follows:
in the fleet. The charging time for case 2a and 2b are identical
and substantially longer than for case 1a and 1b, since both fully (8)
make use of the standstill time at home.
The cumulative charging time increase for each charge event
per EV, as illustrated in Fig. 3, remains below 7.2, 16.2, 30.8, where and are the active and reactive grid power
and 72.8% for, respectively, 85, 90, 95, and 100% of all charging consumption for each conductor on each time step. is the
actions. This illustrates the significant impact of voltage droop
residential load, including PV power production, is the
behavior on the charge duration at some moments, but a limited
active power consumption of an EV during charging.
impact most of the times.
is maximum of the sum of the load on each conductor
The minimum, mean and maximum utility factor (UF, the
. and are de-
fraction of distance driven on battery energy) in the fleet are,
fined as follows:
respectively, 57.4, 70.8, and 85.5%. The UF remains identical
in all cases for each EV in the fleet. This indicates that there is
sufficient charging flexibility to implement voltage dependent
charging behavior, even in this constrained grid situation.
B. Voltage Droop Charging Behavior (9)
To illustrate the voltage droop charging behavior, a four hour (10)
period between 22.30 and 02.30 on one specific day is discussed
in detail for case 1a and 1b.
The charging simultaneity for case 1a (black curve) and 1b where only contains the energy the households effectively
(gray area) are illustrated in Fig. 4. The simultaneity varies be- consume (pos) and produce (neg), the grid losses are not of in-
tween 6 and 18 EVs for both cases, and is slightly higher for case terest for the residential end user.
1820 IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 2014

Fig. 5. Voltage magnitude profiles and charging power profiles at for case 1a (black curve) and case 1b (gray area).

TABLE IV TABLE V
IMPACT OF THE CHARGING CASES ON POWER PROFILE IMPACT OF THE CHARGING CASES ON VOLTAGE MAGNITUDE

Compared to case 0, is more than doubled for case 1a (12)


and 1b. As a result, the transformer is loaded above its nominal
rating if case 1 charging is applied. The increase is significantly
less for case 2a and 2b. Because is, respectively, at 93.8 Case 0 is compliant with EN50160, the minimum voltage
and 91.0% of the transformer rating for case 2a and 2b, a trans- is just below 0.9 pu. Voltages below 0.9 pu occur for only
former upgrade is not required. When comparing case 1a and 0.2% of time in one week at one location.
1b, it can be concluded that the presence of voltage droop be- For case 1a, the grid is not compliant with the EN50160 stan-
havior only has a limited impact on peak power. dard. is 0.75 pu and the voltage magnitude goes below
approximately doubles due to EV charging. The differ- 0.85 pu in each of the 25 weeks of the simulation, affecting 37
ence in power production and consumption between case 1 and households.
2 is due to a higher self-consumption for case 2, because of the For case 1b, is above 0.85 pu. However, the grid is still
increased simultaneity between EV charging and PV power pro- not compliant with EN50160. The voltage magnitude is below
duction. However, the limited difference indicates that the si- 0.9 pu in 8 weeks of the simulation for up to 10.4% of time per
multaneity remains low. week, affecting 13 households.
for case 1 is approximately the fourfold compared to For case 2a and 2b, the voltage magnitude is compliant with
case 0, due to the increase in peak power and energy consump- EN50160. is above 0.85 pu and the voltage magnitude is
tion. for case 2 is approximately 25% lower than for case between 0.9 and 0.85 pu for a maximum of 1.4% of time per
1 due to the lower peak power values. week.

D. Voltage Magnitude Profile E. Voltage Unbalance Factor


The impact of EV charging on the voltage magnitude is sum- Voltage unbalance compliance with EN50160 is calculated as
marized in Table V. The voltage magnitude stays below 1.1 pu follows:
for all cases. Voltage magnitude compliance with EN50160 is
calculated as follows: %
for % of all 10 min intervals of any week (14)
for % of all min intervals of any week where is the voltage unbalance factor at node on time
(11) step . The results are summarized in Table VI. The VUF is
LEEMPUT et al.: IMPACT OF EV ON-BOARD SINGLE-PHASE CHARGING STRATEGIES 1821

TABLE VI pletely. Furthermore, the functionality of Mode 3 charging in-


IMPACT OF THE CHARGING CASES ON VUF frastructure is to be explored, e.g., an adaptation of the charging
power by the infrastructure.

REFERENCES
[1] E. Knipping and M. Duvall, “Environmental assessment of plug-in hy-
brid electric vehicles, volume 2: United States air quality analysis based
on AEO-2006 assumptions for 2030 Palo Alto, CA, Jul. 23, 2007, Tech.
Rep. 1015326, EPRI.
[2] J. A. P. Lopes, F. J. Soares, and P. M. R. Almeida, “Integration of
electric vehicles in the electric power system,” Proc. IEEE, vol. 99,
no. 1, pp. 168–183, Jan. 2011.
compliant with EN50160 when there are no EVs charging, as [3] K. Clement-Nyns and J. Driesen, “The impact of charging plug-in hy-
the maximum value is below 2%. brid electric vehicles on a residential distribution grid,” IEEE Trans.
For case 1a, VUF is above 2% for up to 7.7% of time per Power Syst., vol. 25, no. 1, pp. 371–380, Feb. 2010.
[4] J. Van Roy, N. Leemput, S. De Breucker, F. Geth, P. Tant, and J.
week, in 7 of the 25 weeks. This affects 24 of the 39 households, Driesen, “An availability analysis and energy consumption model for
in up to 7 weeks for individual households. For case 1b, VUF a flemish fleet of electric vehicles,” in Proc. Eur. Electric Veh. Congr.,
values are compliant with the standard. VUF only exceeds 2% Brussels, Belgium, 2011, pp. 1–12.
[5] N. Leemput, J. Van Roy, F. Geth, P. Tant, B. Claessens, and J. Driesen,
for a maximum of 4.5% of time per week. Thus, voltage droop “Comparative analysis of coordination strategies for electric vehicles,”
behavior substantially reduces VUF. in Proc. IEEE Power Energy Soc. Innovative Smart Grid Technol.
For both case 2a and 2b, the VUF exceeds 2% for a maximum Conf. Eur., Manchester, U.K., 2011, pp. 1–8.
of 0.3% of time per week, which is compliant with EN50160. [6] W. Kempton and J. Tomić, “Vehicle-to-grid power implementation:
From stabilizing the grid to supporting large-scale renewable energy,”
This is significantly lower than for case 1a and 1b, due to the J. Power Sources, vol. 144, no. 1, pp. 280–294, Jun. 2005.
increased charging simultaneity and lower peak power values. [7] A. Y. Saber and G. K. Venayagamoorthy, “Plug-in vehicles and renew-
able energy sources for cost and emission reductions,” IEEE Trans. Ind.
Electron., vol. 58, no. 4, pp. 1229–1238, Apr. 2011.
V. CONCLUSION [8] M. D. Gallus, R. La Fauci, and G. Andersson, “Investigating PHEV
The residential grid impact of EV charging can be substan- wind balancing capabilities using heuristics and model predictive con-
trol,” in Proc. IEEE Power Energy Soc. General Meeting, Minneapolis,
tially reduced with on-board strategies that do not require com- MN, USA, pp. 1–8.
munication between the EVs and the distribution grid. Voltage [9] T. Tran-Quoc, M. Braun, J. Marti, C. Kieny, N. Hadjsaid, and S. Bacha,
droop charging and on-board peak shaving both can be imple- “Using control capabilities of DER to participate in distribution system
mented within the EVs. operation,” IEEE Power Tech., pp. 561–566, Lausanne, Switzerland.
[10] H. Sekyung, H. Soohee, and K. Sezaki, “Development of an optimal ve-
Voltage droop charging eliminates EV-induced voltage mag- hicle-to-grid aggregator for frequency regulation,” IEEE Trans. Smart
nitudes below 0.85 pu. VUF values higher than 2% occur for Grid, vol. 1, no. 1, pp. 65–72, Jun. 2010.
less than 5% of time per week. However, because the voltage [11] S. Vandael, K. De Craemer, N. Boucké, T. Holvoet, and G. Deconinck,
“Decentralized coordination of plug-in hybrid vehicles for imbalance
magnitude is below 0.9 pu for more than 5% of time per week, reduction in a smart grid,” in Proc. 10th Int. Conf. Autonomous Agents
the grid is not compliant with the EN50160 standard. On-board and Multiagent Syst., Taipei, Taiwan, 2011, pp. 803–810.
peak shaving makes the grid fully compliant with EN50160, and [12] S. Acha, T. C. Green, and N. Shah, “Optimal charging strategies of
the peak power remains below the transformer rating. As a re- electric vehicles in the UK power market,” in Proc. IEEE Power En-
ergy Soc. Innovative Smart Grid Technol., Anaheim, CA, USA, 2011,
sult, there is no need for grid infrastructure reinforcements. pp. 1–8.
For all cases, the voltage droop behavior reduces EV-induced [13] N. Rotering and M. Ilic, “Optimal charge control of plug-in hybrid elec-
voltage deviations and voltage unbalance. The combination of tric vehicles in deregulated electricity markets,” IEEE Trans. Power
Syst., vol. 26, no. 3, pp. 1021–1029, Nov. 2010.
both strategies enables the integration of EVs prior to the wide-
[14] M. Doostizadeh, M. Khanabadi, A. Esmaeilian, and M. Mohs-
spread availability of coordination. Once available, EV-based eninezhad, “Optimal energy management of a retailer with smart
peak shaving can be bypassed, while the voltage droop enabled metering and plug-in hybrid electric vehicle,” in Proc. 10th Int. Conf.
charger provides an effective fallback mechanism in case of ma- Environ. Elect. Eng., Rome, Italy, 2011, pp. 1–5.
[15] T. P. Lyon, M. Michelin, A. Jongejan, and T. Leahy, “Is “smart
licious coordination. charging” policy for electric vehicles worthwhile?,” Energy Policy,
The electrically driven fraction remains identical to uncon- vol. 41, pp. 259–268, Feb. 2012.
trolled charging for both strategies. However, the impact on the [16] Q. Gong, S. Midlam-Mohler, V. Marano, and G. Rizzoni, “Study of
charging time is different. For voltage droop charging, there is PEV charging on residential distribution transformer life,” IEEE Trans.
Smart Grid, vol. 3, no. 1, pp. 404–412, Mar. 2012.
a significant impact at some moments, but a limited impact at [17] C. Dang, J.-L. Parpal, and J.-P. Crine, “Electrical aging of extruded
most of the time. Therefore, the total charging time is only im- dielectric cables: Review of existing theories and data,” IEEE Trans.
pacted marginally. For EV-based peak shaving, the charging Dielect. Elect. Insul., vol. 3, no. 2, pp. 237–247, Apr. 1996.
[18] N. Leemput, F. Geth, B. Claessens, J. Van Roy, R. Ponnette, and J.
time increases significantly because the full standstill time at Driesen, “A case study of coordinated electric vehicle charging for
home and at the workplace is used. peak shaving on a low voltage grid,” in Proc. IEEE Power Energy Soc.
Droop parameters can be optimized to support other objec- Innovative Smart Grid Technol. Conf. Eur., Berlin, Germany, Oct. 6–9,
tives, e.g., decreasing the grid losses and the frequency of oc- 1012, pp. 1–7.
[19] S. Shao, T. Zhang, M. Pipattanasomporn, and S. Rahman, “Impact of
currence of voltages below 0.9 pu. Methodologies to determine TOU rates on distribution load shapes in a smart grid with PHEV pen-
the optimal configuration and behavior of DERs [43] could be etration,” in Proc. IEEE PES Transmiss.Distribut. Conf. Expo., New
adapted for this problem. Orleans, LA, USA, pp. 1–6.
[20] S. Vandael, B. Claessens, M. Hommelberg, T. Holvoet, and G. Decon-
Future work will include other impact factors for the EV user,
inck, “A scalable three-step approach for demand side management of
e.g., to quantify the loss in comfort of non-fully charged EVs, plug-in hybrid vehicles,” IEEE Trans. Smart Grid, vol. 4, no. 2, pp.
as UF and the charging time increase may not reflect this com- 720–728, Jun. 2013.
1822 IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 4, JULY 2014

[21] K. De Craemer, S. Vandael, B. Claessens, and G. Deconinck, “An Niels Leemput (S’11) received the M.Sc. degree in
event-driven dual coordination mechanism for demand side manage- electrical engineering, with specialization in energy,
ment of PHEVs,” IEEE Trans. Smart Grid, vol. 5, no. 2, pp. 751–760, from the KU Leuven, Leuven, Belgium, in 2010. He
Mar. 2014. is currently pursuing the Ph.D. degree at KU Leuven.
[22] H. L. Willis, Power Distribution Planning Reference Book. New Currently, he is working as a Research Assistant
York: Marcel Dekker Inc., 1997. with the division ESAT-ELECT. He is funded by
[23] Voltage Characteristics of Electricity Supplied by Public Electricity IWT-Vlaanderen and is enrolled in the KIC InnoEn-
Networks, CENELEC Std. EN 50 160, Jul. 2010. ergy Ph.D. School. His research interests include the
[24] A. S. Masoum, S. Dei, P. S. Moses, and A. A. Siada, “Voltage profile grid integration of charging infrastructure for electric
and THD distortion of residential network with high penetration of vehicles, and the grid-coupling of electric vehicles.
plug-in electrical vehicles,” in Proc. IEEE Power Energy Society
Innovative Smart Grid Technol. Conf. Eur., Gothenburg, Sweden,
pp. 1–6. Frederik Geth (S’10) received the M.Sc. degree in
[25] A. von Jouanne and B. Banerjee, “Assessment of voltage unbalance,” electrical engineering from the KU Leuven, Leuven,
IEEE Trans. Power Del., vol. 16, no. 4, pp. 782–790, Oct. 2001. Belgium, in 2009. He is currenlty pursuing the Ph.D.
[26] F. Shahnia, A. Ghosh, G. Ledwich, and F. Zare, “Predicting voltage degree at KU Leuven.
unbalance impacts of plug-in electric vehicles penetration in residential Currently, he is working as a Research Assistant
low-voltage distribution networks,” Elect. Power Compon. Syst., vol. with the division ESAT-ELECTA. He is enrolled in
41, no. 16, pp. 1594–1616, Oct. 2013. the KIC InnoEnergy Ph.D. School. His research in-
[27] S. Weckx, C. Gonzalez, P. Vingerhoets, and J. Driesen, “Phase terests include optimal storage integration in distri-
switching and phase balancing to cope with a massive photovoltaic bution grids, batteries for electric vehicles, and con-
penetration,” in Proc. CIRED, Stockholm, Sweden, pp. 1–4. trolling the impact of the charging currents of electric
[28] M. B. Delghavi and A. Yazdani, “A unified control strategy for elec- vehicles on the grid.
tronically interfaced distributed energy resources,” IEEE Trans. Power
Del., vol. 27, no. 2, pp. 803–812, Apr. 2012.
[29] F. Geth, N. Leemput, J. Van Roy, J. Büscher, R. Ponnette, and Juan Van Roy (S’09) received the M.Sc. degree in
J. Driesen, “Voltage droop charging of electric vehicles in a residen- electrical engineering, with specialization in energy,
tial distribution feeder,” in Proc. IEEE Power Energy Soc. Innovative from the KU Leuven, Leuven, Belgium, in 2010. He
Smart Grid Technol. Conf. Eur., Berlin, Germany, 2012, pp. 1–8. is currently pursuing the Ph.D. degree at KU Leuven.
[30] S. Huang, J. R. Pillai, and P. Thøgersen;, “Voltage support from electric Currently, he is working as a Research Assistant
vehicles in distribution grid,” in Proc. 15th Eur. Conf. Power Electron. with the division ESAT-ELECT . He is funded
Applicat., Lille, France, 2013, pp. 1–8. through doctoral scholarship of the Flemish Institute
[31] A. Delnooz, D. Six, C. Mol, and E. Gielen, “State-of-the-art in business for Technological Research (VITO) and is enrolled
models for charging services: The EVCITY approach,” presented at in the KIC InnoEnergy Ph.D. School. His research
the European Electric Vehicles Conf., Brussels, Belgium, Nov. 19–22, interests include integrating and controlling the
2012. charging of electric vehicles and their interaction
[32] Kabels Voor Ondergrondse Aanleg, Met Synthetische Isolatie En Ver- with thermal energy flows in buildings.
sterkte Mantel (Type 1 kV), NBN Std. C 33-322, 1975.
[33] R. Belmans, G. Deconinck, and J. Driesen, “Deel IV: Transforma-
toren,” Elekrtische Energie. Deel 1. Leuven, Belgium, ch. 4, pp. Annelies Delnooz received the M.Sc. degree in ap-
269–270, 2010, ACCO, sec. 2.2.3. plied economics, specialization policy management,
[34] W. Labeeuw and G. Deconinck, “Customer sampling in a smart grid from Hasselt University, Diepenbeek, Belgium, in
pilot,” in Proc. IEEE Power Energy Soc. General Meeting, San Diego, 2010.
CA, USA, pp. 1–7. Since November 2010, she has been active as a
[35] ODE: De Ciifers, Oct. 2013 [Online]. Available: http://www.ode.be/ Researcher at the Flemish Institute for Technological
zonnestroom/de-cijfers Research (VITO), Mol, Belgium. She is mainly in-
[36] Specifieke Technische Voorschriften Voor Decentrale Productie-Instal- volved in research with regard to the energy markets
laties die in Parallel Werken met het Distributienet, May 2009, Syner- in an international context and elaborating business
grid regulation C10/11, Rev. 12. case calculations regarding VPPs, electric vehicles,
[37] “Fuel economy labeling of motor vehicles: Revisions to improve cal- and other smart grid applications.
culation of fuel economy estimates,” U.S. Environmental Protection
Agency, Tech. Rep. EPA420-R-06-017, Dec. 2006.
[38] J. Tant, F. Geth, D. Six, and J. Driesen, “Multiobjective battery storage Jeroen Büscher (M’11) received the M.Sc. degree
to improve pv integration in residential distribution grids,” IEEE Trans. from the University of Ghent, Ghent, Belgium,
Sustain. Energy, vol. 4, no. 1, pp. 182–191, Jan. 2013. in 2005 and Ph.D. degree in physics from the
[39] J. Dogger, B. Roossien, and F. Nieuwenhout, “Characterization of Katholieke Universiteit Leuven (KU Leuven),
Li-ion batteries for intelligent management of distributed grid-con- Leuven, Belgium, in 2010.
nected storage,” IEEE Trans. Energy Convers., vol. 26, no. 1, pp. In 2011, he joined the ESAT-ELECTA division of
256–263, Mar. 2011. the KU Leuven as a Postdoctoral Researcher. His in-
[40] IEC, 2010, Electric Vehicle Conductive Charging System—Part 1: terests include the smart grid integration of charging
General Requirements. Geneva, Switzerland, (IEC 61851-1), Int. infrastructure for electric vehicles and the modeling
Electrotechnical Commission. of electrical energy systems.
[41] M. Braun et al., “Is the distribution grid ready to accept large-scale
photovoltaic deployment? State of the art, progress, and future
prospects,” Prog. Photovolt: Res. Appl., vol. 20, no. 6, pp. 681–697,
Sep. 2012. Johan Driesen (S’93–M’97–SM’12) received the
[42] C. Cheng and D. Shirmohammadi, “A three-phase power flow method M.Sc. and Ph.D. degrees in electrical engineering
for real-time distribution system analysis,” IEEE Trans. Power Syst., from the KU Leuven, Belgium, in 1996 and 2000,
vol. 10, no. 2, pp. 671–679, May 1995. respectively.
[43] A. Keane, L. F. Ochoa, C. L. T. Borges, G. W. Ault, A. D. Alarcon-Ro- Currently, he is a Professor with the KU Leuven
driguez, R. Currie, F. Pilo, C. Dent, and G. P. Harrison, “State-of- and teaches power electronics and electric drives. In
the-art techniques and challenges ahead for distributed generation plan- 2000, he was with the Imperial College of Science,
ning and optimization,” IEEE Trans. Power Syst., vol. 28, no. 2, pp. Technology and Medicine, London, U.K. In 2002,
1493–1502, May 2013. he was with the University of California, Berkeley.
Currently, he conducts research on distributed gener-
ation, power electronics, and its applications.

You might also like