206 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 26, NO.

1, FEBRUARY 2011

Assessment of the Impact of Plug-in Electric

Vehicles on Distribution Networks
Luis Pieltain Fernández, Tomás Gómez San Román, Senior Member, IEEE, Rafael Cossent,
Carlos Mateo Domingo, and Pablo Frías

Abstract—Plug-in electric vehicles (PEVs) present environ- become technically possible. This means that the battery, while
mental and energy security advantages versus conventional being connected to the grid, will also be able to inject energy into
gasoline vehicles. In the near future, the number of plug-in electric
it. In this sense, future PEVs will behave sometimes as electric
vehicles will likely grow significantly in the world. Despite the
aforementioned advantages, the connection of PEV to the power loads and some other times as generators, i.e., they will operate
grid poses a series of new challenges for electric utilities. This as energy storage devices [1], [2]. Notwithstanding, their av-
paper proposes a comprehensive approach for evaluating the im- erage net power flow will be of consumption, as with electric
pact of different levels of PEV penetration on distribution network loads.
investment and incremental energy losses. The proposed approach
is based on the use of a large-scale distribution planning model The EU Electricity Directive [3] states that distribution
which is used to analyze two real distribution areas. Obtained system operators (DSOs) must be at least legally and func-
results show that depending on the charging strategies, investment tionally unbundled from the remaining activities, generation
costs can increase up to 15% of total actual distribution network and retail, in the electricity sector. They are responsible for
investment costs, and energy losses can increase up to 40% in
off-peak hours for a scenario with 60% of total vehicles being
providing network services for the connection of electricity
PEV. consumers and distributed generation. As PEVs are developed,
DSOs will also become responsible for providing the required
Index Terms—Distribution investment, distribution network
network upgrades to supply the PEV charging points. In ad-
planning, electricity distribution, network energy losses, plug-in
electric vehicles. dition, similarly to distributed generation (DG), DSOs would
have no control over the location of future PEV charging points
or stations, and no direct control over periods and frequency
I. INTRODUCTION of PEV charging. As a consequence, the network expansion
planning problem will become more complicated due to the
higher degree of uncertainty.
T HE fast technological development in the automotive
sector, together with the growing environmental concerns
and the increase of oil prices, have triggered the appearance
Recently, some research work on the integration of PEVs into
the electricity system has been performed. References [4]–[6]
are focused on the impact assessment of PEVs in the electricity
of vehicles with diversified energy sources. This is the case
generation system. Generation dispatch, operational reserves,
of plug-in electric vehicles (PEVs). Basically there are two
and the integration of higher volumes of intermittent wind gen-
models: pure battery electric vehicles (BEVs) and plug-in
eration are studied depending on the presence of PEVs. Other
hybrid electric vehicles (PHEVs) which essentially work with a
studies assess the impact of PEVs on the generation mix and
combination of two power sources, i.e., batteries and gasoline.
The latter are an extended version of current hybrid electric emission reductions in future scenarios in different areas
vehicles including a battery with larger autonomy and able or countries [7], [8].
to connect to the grid to be charged. PHEVs might be a fast Furthermore, several works and reports about new business
models for a successful integration of PEVs in the electricity
and temporary solution until more advanced EVs technologies
become mature. sector have been published. Public charging points and charging
The first generations of PEVs are expected to be connected stations infrastructures can be developed by new players, PEV
to the grid just for charging the battery. However, as the new energy commercialization is subject to competition, and aggre-
technology matures, the concept of vehicle-to-grid (V2G) will gators may exploit the V2G concept providing ancillary services
or simply trading energy in the day-ahead market. Some of these
new business models are intimately related with a successful
Manuscript received December 18, 2009; revised March 05, 2010; accepted
April 18, 2010. First published May 24, 2010; current version published January
deployment of smart-grids with intelligent meters and bidirec-
21, 2011. Paper no. TPWRS-00985-2009. tional communication systems [9], [10].
L. Pieltain Fernández is with ENAGAS S.A., Madrid, Spain (e-mail: Finally, with regard to the impact of PEVs on distribution net-
[email protected]).
T. Gómez San Román, C. Mateo, R. Cossent, and P. Frías are with the Insti- works, some interesting works that deal with this problem can
tuto de Investigación Tecnológica of Universidad Pontificia Comillas, Madrid, be found [11]–[15]. The approach used in these previous works
Spain. is based on the detailed analysis of small-scale distribution net-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. works. On the contrary, the approach presented in this paper is
Digital Object Identifier 10.1109/TPWRS.2010.2049133 unique because it obtains results for large-scale real distribution
0885-8950/$26.00 © 2010 IEEE
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2025 at 14:15:21 UTC from IEEE Xplore. Restrictions apply.
 PIELTAIN FERNÁNDEZ et al.: ASSESSMENT OF THE IMPACT OF PLUG-IN ELECTRIC VEHICLES ON DISTRIBUTION NETWORKS 207

areas. This paper analyzes an urban area with more than 6000
low-voltage residential customers, and an industrial and resi-
dential area with over 61 000 customers.
A large-scale distribution network planning model has been
used for the calculation of the required network investment with
different future levels of PEV penetration. This model is also
known as a Reference Network Model. These models are used
by regulators when setting DSOs revenues for a regulatory pe-
riod of several years ahead. They are used to assess the ex-ante
efficient investments DSOs have to make in order to connect the
expected new loads and distributed generators to the distribution
network [16]–[21].
The remainder of this paper is organized as follows. In
Section II, a general overview of the developed assessment
approach is presented. In Section III, the description of the two Fig. 1. Base case distribution network in area A.
real distribution areas that have been selected to evaluate the
impact of PEV penetration is provided. Section IV describes
the three scenarios of PEV penetration and the vehicle charging 3) Calculation of network reinforcements and incremental en-
patterns in peak and off-peak hours. Section V presents the basic ergy losses. For each PEV penetration scenario, the op-
features of the large-scale distribution planning model used for eration of the electric distribution network is analyzed at
identifying and quantifying required network reinforcements. peak hours and at off-peak hours. The number of con-
Section VI analyzes the obtained results for the considered case nected PEVs and modes of charging are different for peak
studies. Finally, conclusions are drawn in Section VII. and off-peak hours. The large-scale planning distribution
model computes quasi-optimal reinforcements to the base
II. ASSESSMENT APPROACH case network in case they are needed to connect and to
supply the load required by PEVs. In addition, this model
This section presents an overview of the developed approach also computes incremental energy losses with respect to
to assess distribution network investment and incremental en- the base case due to PEV charging.
ergy losses in real distribution areas for different penetration These steps are described in more detail in the following sec-
levels of PEVs. tions.
The approach consists of a series of sequential steps.
1) Base case definition. Each distribution area is defined by III. DESCRIPTION OF THE DISTRIBUTION AREAS
its size and location. The total population and the location
and demand of consumers are also known. In each dis- Two real distribution areas have been selected to illustrate the
tribution area, assuming that no PEVs are connected yet, application of the proposed assessment approach. In this sec-
an optimal planned network is designed from scratch to tion, the main features of each region together with the con-
supply actual loads and to connect actual distributed gen- structed base case networks are described. Further numerical
eration meeting quality-of-service requirements. That is data will be provided in Section VI.
the base case distribution network. To build this network,
a large-scale (greenfield) distribution planning model is A. Residential Urban Area (A)
used. This model is described in more detail in Section V. Area A is a region of in which three residential towns
2) PEV scenarios definition. In this step, different scenarios are included. The base case distribution network is represented
of penetration of PEVs are defined. Each scenario is char- in Fig. 1. The only supply point is a high voltage (HV) substa-
acterized by a number of PEVs expressed as a percentage tion whose capacity is assumed to be large enough to supply
of the total population of vehicles in the area. Three sce- the required load in all scenarios. The medium voltage (MV)
narios, representing PEV penetration levels of 35%, 51%, network is composed of 20-kV underground feeders that con-
and 62%, have been modeled. These penetration levels nect the substation with medium/low voltage (MV/LV) trans-
would be achieved by 2020, 2030, and 2050, respectively, formation centers. Low voltage (LV) cables connect LV elec-
according to [7]. Other sources provide different penetra- tricity consumers to MV/LV transformers. There are 6121 LV
tion speeds [2]. However, the aim of this research is to ob- electricity consumers with a contracted demand of 34.21 MW,
tain results for a set of diverse scenarios representing dif- and 15 MV customers with a contracted demand of 37.54 MW.
ferent potential situations. In addition, two different modes There are also connected some DG units, solar photovoltaic,
of PEV charging are considered: normal and fast. Charging and micro combined heat and power (CHP), but the installed
points, including new charging stations, are located ran- capacity is so small that they have almost no impact on the net-
domly but at the same coordinates as actual electricity con- work design.
sumers. Finally, a behavior pattern of charging and driving The total population of vehicles owned by residents in the
hours is also assumed for PEVs. Charging modes and PEV area has been estimated in 3676 with a ratio of 0.6 vehicles per
scenarios are defined in more detail in Section IV. home or LV supply point.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2025 at 14:15:21 UTC from IEEE Xplore. Restrictions apply.
 208 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 26, NO. 1, FEBRUARY 2011

TABLE I
CHARACTERIZATION OF DIFFERENT PHEV AND BEVS BATTERIES

Fast charging means that just in 1 h (1C) or in half an hour (2C),

Fig. 2. Base case distribution network in area B. the battery can be completely charged.
From the point of view of network infrastructure require-
ments, the charging power or demand is the relevant parameter.
B. Industrial and Residential Area (B)
In this study, two types of charging points are considered.
Area B is a region of which includes 22 towns and • Individual charging points: located in residential or
more than 52 200 residents. Most of LV loads are connected parking areas designed for normal or slow charging rates.
mainly within towns, whereas there are several hundreds MV Due to the moderate required power, which according to
loads and a few HV loads that are connected in industrial areas Table I range from 1.6 kW to 9.6 kW, these charging points
in the outskirts of towns. are directly connected to the LV distribution network.
The base case network for this distribution area is repre- • Charging stations: similarly to current gas stations, they
sented in Fig. 2. This network is supplied through an HV comprise several connection points for fast charging. Fast
transmission substation. The HV subtransmission network is charging stations will be connected to MV distribution net-
composed of 132-kV and 45-kV overhead lines that supply works.
power to the HV/MV substations. The 15-kV overhead feeders The possibility of V2G has been also considered. Under this
departing from those substations supply MV loads and MV/LV mode, a percentage of the total PEVs are modeled as injecting
transformers within the towns. LV lines supply LV consumers energy into the grid at peak hours when electricity prices are
but they are not represented in Fig. 2. Contracted demand is higher assuming that time of use tariffs are implemented. The
280.31 MW at LV, and 114.65 MW at MV. considered rated power for this operation mode goes from 3 to
The population of vehicles in the area has been estimated in 10 kW.
28 626 assuming a ratio of vehicles per supply point slightly
lower than in area A because this area is more extensive and B. PEV Penetration Scenarios
presents lower load concentration. Different prospective and technology roadmap studies have
been recently developed on this topic [1], [2]. The level of un-
IV. PLUG-IN ELECTRIC VEHICLES SCENARIOS certainty regarding how fast this technology will penetrate the
In this section, the three PEV penetration scenarios selected market is high. According to EPRI [7] under a moderate evolu-
for the analysis are defined. First the different types of consid- tion by 2020, 35% of total vehicles will be PEVs. This figure
ered PEVs are described. will increase up to 51% between 2030 and 2040, and it is esti-
mated that 62% of total vehicles will be electric by 2050.
A. Types of PEVs and Charging Points Therefore, three PEV penetration scenarios are considered in
PEVs will be connected to the distribution network in specif- this study.
ically designed charging points. The batteries used by PEVs are • Scenario 1: 35% of total vehicles are assumed to be PEV.
characterized by its storage capacity and by the power required PHEV 30, 40, and 60 models with slow and fast charging
to charge them in a predefined period of time. (1C) possibilities are considered. In addition, 10% of the
In Table I, different types of PHEVs batteries together with PEV connected at peak hours can operate in V2G mode.
one BEV model, according to MIT [22], USABC, and EPRI • Scenario 2: 51% of total vehicles are assumed to be PEV.
studies [23], are shown. In PHEV models, the notation PHEV In addition to PHEVs, BEVs are also included. As it can be
X is used to indicate that the autonomy as electric vehicle is X observed in Table I, the required charging power for BEVs
miles. As it can be observed in the table, the required storage is much larger than for PHEVs.
capacity increases as the autonomy of the vehicle is higher. De- • Scenario 3: 62% of total vehicles are assumed to be PEV.
pending on the charging rate, more or less power will be re- Fast charging (2C) is considered only in this scenario.
quired to achieve full charge. Charging rates are specified as In all these scenarios, the total amount of vehicles is con-
0.2C, 1C, or 2C. Normal or slow charging rate is represented by sidered to remain unchanged. The scenarios are constructed se-
0.2C. This means that to reach the full storage capacity when quentially. This means that PEVs adopted for scenario 1 are kept
the battery is totally empty 5 h are required, i.e., . in scenarios 2 and 3, and so on.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2025 at 14:15:21 UTC from IEEE Xplore. Restrictions apply.
 PIELTAIN FERNÁNDEZ et al.: ASSESSMENT OF THE IMPACT OF PLUG-IN ELECTRIC VEHICLES ON DISTRIBUTION NETWORKS 209

TABLE II practices would require more investment for connecting future

CONNECTION AND CHARGING PATTERNS FOR PEAK AND OFF-PEAK HOURS customers than other that has made network reinforcements
and its actual network presents higher loadability margins. The
use of adapted networks instead of actual networks solves this
problem.
The used large-scale distribution planning model is based on
heuristic planning algorithms together with a geographic in-
formation system (GIS). The objective function minimizes the
one-off investment costs and the present value of energy losses
and maintenance costs throughout the planning period given by
a number of years. As a result, all the network components,
lines, and substations are designed, and their technical and eco-
nomic parameters are obtained. The model provides: invest-
On the other hand, load is considered static. In this way, the ment, maintenance, and energy losses costs, detailed by voltage
individualized effect of increasing EVs is analyzed. level and type of network installation. In addition, a geograph-
ical representation of the network layout is obtained, as shown
C. PEV Connection and Charging Patterns in Figs. 1 and 2 [25].
Once the base case distribution network has been obtained,
As it was explained in Section II, power flows in the distri- the additional network requirements associated with each PEV
bution network will be analyzed at peak hours when the con- penetration scenario are determined. In order to calculate those
sumers demand is higher and at off-peak hours when it is ex- network reinforcements in each PEV scenario, the large-scale
pected that most of PEV charging will take place. It is assumed distribution planning model is used, taking as starting point the
that time-of-use tariffs or even real-time pricing with smart me- base case distribution network. A similar study carried out for
ters will be adopted for PEV charging. assessing the impact of distributed generation on distribution
In addition to the number and type of PEV in each scenario, networks can be found in [21].
it is needed to define their connection and charging patterns at The charging points with the associated required power are
peak and off-peak hours. Table II summarizes them. located randomly in the distribution area but always at the same
According to Table II, it is assumed that 85% of the PEVs location as existing load points. Two power flow analyses are
are charging at off-peak hours, and most of them in a slow or carried out for each scenario. One represents the charging situ-
normal charge mode. At peak hours, it is assumed that 40% of ation at peak hours, and the another one at off-peak hours ac-
the PEVs are connected, 90% are charging, some of them in a cording to Table II.
fast charging mode, and 10% are injecting power into the grid. The planning algorithm finds the minimum reinforcement in-
This approximation is based on studies reported in [7]. vestment required for connecting and charging the PEVs in each
operating condition and each scenario. The same results pre-
V. LARGE-SCALE DISTRIBUTION PLANNING MODEL viously commented are provided for incremental installations
A large-scale distribution planning model has been used first and incremental investment and maintenance and energy losses
to build the optimal base case distribution network to supply the costs. A cost is assigned to energy losses in order to avoid sub-
demand in the two areas of interest. Second, the base case net- investment in the network.
work is taken as a reference to assess the impact of PEV pen-
etration scenarios on network reinforcements and incremental VI. CASE STUDIES
energy losses. Several different studies have been carried out by applying the
The base case network is designed as an optimally adapted explained assessment approach to compute network investment
network, i.e., minimum investment and operational costs, requirements and incremental energy losses in the two consid-
which are required to supply actual electricity customers ered distribution areas.
meeting voltage drop and capacity constraints together with In the following, the main obtained results are presented. In
continuity of supply requirements [16], [17]. Some of the the Appendix, a more detailed description of the input data is
authors of the paper have been working in this type of models included.
for the Spanish regulator, who uses the concept of optimally
adapted network as a benchmark when calculating costs of A. Incremental Investment at Peak Hours
regulated distribution companies [18]–[20], [24]. In Fig. 3(a) and (b), the incremental investments needed in
The use of adapted networks as reference provides more peak hours to integrate the PEVs corresponding to the three sce-
robustness to the results since these are independent from the narios considered in distribution areas A and B are shown.
particular situation in which the actual network is. This is The incremental investment is disaggregated for the different
somewhat similar to the case when regulators determine the network installations in both areas (see Sections III-A and B):
allowed investment costs of DSOs [24]. In this case, adapted LV lines, MV/LV transformers, MV feeders, HV/MV substa-
networks allow regulators to overcome asymmetries of infor- tions, and HV lines. In addition, the incremental investment is
mation. For example, under incentive regulation, a DSO with expressed as a percentage of the investment cost of the base case
an actual congested network due to profitable underinvestment distribution network without PEVs taken as reference.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2025 at 14:15:21 UTC from IEEE Xplore. Restrictions apply.
 210 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 26, NO. 1, FEBRUARY 2011

Fig. 4. Incremental investment in peak hours versus charging simultaneity fac-

tors.

In this case study, it has been assumed that the simultaneity

factor for PEVs is equal to 1. This means that the charging of
all PEVs coincides in time.

B. Investment at Peak Hours With Intelligent

Charging Strategies
In this case study, the effects of applying intelligent strategies
to avoid PEV charging simultaneously during peak hours have
been analyzed. The results are presented in Fig. 4.
In Fig. 4, two different cases are represented for areas A and
B. The cases with charging simultaneity factors equal to 1 cor-
respond with the assumption made in the previous section.
It can be observed how simultaneity factors lower than 1 due
Fig. 3. (a) Incremental investment and maintenance costs in area A. (b) Incre-
mental investment and maintenance costs in area B.
to smart charging strategies and random behavior, especially for
fast charging modes, decrease the need for network reinforce-
ments. In this case, we have considered the following simul-
TABLE III taneity factors: 0.8 for slow charge (0.2C), 0.15 for fast charge
SIMULTANEITY FACTORS FOR PLANNING DISTRIBUTION NETWORKS
(1C), and 0.07 for the very fast charge mode (2C). To compute
those values, we have considered the time required for each PEV
to fully charge its battery and assumed that peak hours take place
between 16 and 21 h (5 h) per day because these are mainly res-
idential areas . Simultaneity factors are calculated as the proba-
bility of coincidence assuming that the charge can be uniformly
In relative terms, the incremental investment in LV network distributed along the total available charging period.
for area A and MV/LV transformers in area B are the most im-
portant. C. Incremental Investment at Off-Peak Hours
Once we obtained the different costs for the three scenarios The different PEV penetration scenarios with the charging
(35%, 51%, and 62%), a linear approximation to the evolution mode at off-peak hours (85% of the total PEV connected, 95%
of the investment needs from 0% to 62% share of PEVs was normal charging, and 5% fast charging) shown in Table II were
obtained with the least-squares method. The slope of the linear modeled and analyzed with the large-scale distribution planning
approximation mentioned before corresponds to the total invest- model.
ment needs expressed per PEV, which are: 6310 /PEV in area In both areas A and B, the network designed to supply the
A and 201 /PEV in area B. peak demand can accommodate the connection and charging of
The required investment for the same level of PEV penetra- PEV with no need for network reinforcements.
tion is higher in area A, which has a higher density of population This leads to think that there can be a margin to move
than in area B and a fully underground network. In addition, in- charging periods from peak to off-peak hours in order to reduce
stallation costs for new distribution facilities according to the the investment requirements obtained previously for peak hours
data supplied by DSOs are much higher in area A than in area (Section IV-A). By shifting charging from peak to off-peak
B; see the Appendix for details. hours, the required investment initially calculated at peak hours
Concerning network planning, peak load is the most signifi- decreases (see Fig. 5). However, if you increase charging at
cant parameter. In the case of electricity consumers, peak load is off-peak hours, new requirements of investment can be needed
obtained by applying a simultaneity factor to the contracted de- also at off-peak hours. The carried out study decreases the
mand. For instance, in Spain, the simultaneity factors in Table III investment needed at peak hours by shifting charging until
are typically used. the limit in which investment at off-peak hours would be also
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2025 at 14:15:21 UTC from IEEE Xplore. Restrictions apply.
 PIELTAIN FERNÁNDEZ et al.: ASSESSMENT OF THE IMPACT OF PLUG-IN ELECTRIC VEHICLES ON DISTRIBUTION NETWORKS 211

the base load with a 62% share of PEVs). Incremental losses

at peak hours are up to 13%. They are lower than at off-peak
hours because of two reasons: 1) the number of charging PEVs
is smaller, and 2) the grid loads are at their maximum values.
The presented results were obtained assuming a single load
condition for off-peak hours and another one for peak hours.
A more refined approach to calculate energy losses variations
using time series of 24 h for representative days would be
adopted in further research.
More details on results of investment needs and losses evolu-
Fig. 5. Savings obtained by moving battery charging from peak to off-peak tion can be found in [26].
hours.
VII. CONCLUSION
In this paper, an assessment approach for identifying and
quantifying the impacts of different levels of plug-in electric
vehicles penetration on distribution network investments and
incremental energy losses has been presented.
Two large-scale distribution areas with several voltage levels
have been deeply analyzed. Most previous publications focused
on small single voltage level test cases. This type of analysis
can only be made by the use of recently developed large-scale
distribution planning models [27]. Three scenarios with 35%,
51%, and 62% of PEV penetration have been analyzed for each
area.
Fig. 6. Incremental energy losses in off-peak hours.
The obtained results show that, with the estimated levels of
driving and charging patterns in peak hours, the required net-
work reinforcements can reach values up to 19% of total actual
required. As conclusion, the investment required at peak hours network costs in a situation without PEVs. The required invest-
decreases. ment is higher in the urban area with high load density. If PEV
In Fig. 5, it is first represented for both areas the initial invest- smart charging strategies are implemented, hence decreasing
ment (expressed as a percentage of the base case investments) charging simultaneity factors, up to 60%–70% of the required
that is required without shifting PEVs charging from peak to incremental investment can be avoided. In addition, if strategies
off-peak (see Fig. 3). Then, the final investment is represented, are defined so that some of the PEVs that were charged at peak
which, as it can be appreciated, has been reduced. Finally, this hours are charged at off-peak hours instead, up to 5%–35% of
difference in the investment is translated into savings, which are the required investment can be also avoided.
expressed as a percentage over the initial incremental invest- In the case studies analyzed, despite the fact that most of
ment (without charge shifting). the considered PEVs were assumed to be charging at off-peak
With a PEV share of 62%, in order to reduce the investment hours, from the point of view of the distribution network, in-
needs by moving battery charging from peak to off-peak hours, vestments will follow those results obtained at the peak hours
up to 636 kW (around 187 PEVs) can be moved to off-peak scenarios.
hours in area A and up to 19 492 kW (around 5730 PEVs) in Regarding energy losses, in the maximum penetration sce-
area B. Savings are lower in area A because the strategy is more nario, they could increase up to 40% of actual values in off-peak
limited. hours when most of PEVs would be in a charging mode.
It should be remarked that all the referred investment costs are
D. Incremental Energy Losses
related to new installations or reinforcements in the distribution
The impact of the connection of PEVs on energy losses has network that should be carried out by the DSO. This study does
been analyzed in off-peak hours where most of the PEVs are not consider the costs of charging points or stations themselves.
connected and charging batteries (see Table II). These costs would be incurred by the promoters or owners of
Energy losses increase notably with respect to the reference these infrastructures.
situation at off-peak hours with no PEV connected. In Fig. 6, the On the other hand, despite the fact that the presented results
incremental energy losses are represented. It is observed how are case dependent, two different extreme situations were se-
they increase with the level of PEV penetration. In addition, the lected: 1) an urban area with underground cables and high load
incremental variation is slightly larger in area A than in area B. density and 2) a rural area in a more extended region with highly
As it was mentioned in the case of network investments, this dispersed loads; and results were expressed in relative terms (in
effect is due to the fact that load density is higher in area A than percent). Therefore the two case studies together with the dif-
in B. ferent PEV penetration levels provide a framework of compara-
Incremental losses at off-peak hours are very high because in tive results that can serve as a valuable reference for other poten-
proportion, the amount of added load is high (around 16% of tial situations in network topologies or PEV penetration levels.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2025 at 14:15:21 UTC from IEEE Xplore. Restrictions apply.
 212 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 26, NO. 1, FEBRUARY 2011

TABLE IV TABLE V
NUMBER OF CUSTOMERS AND PEVS AND CONSUMED/GENERATED POWERS TYPES OF LINES/CABLES (O—OVERHEAD; U—UNDERGROUND)

TABLE VI
TYPES OF HV/MV SUBSTATIONS AND MV/LV TRANSFORMING CENTERS

Future research will be focused on analyzing methods to rep-

resent the uncertainty of driving and charging PEV patterns and
its potential impact on network design. In addition, the design
and implementation of smart charging strategies avoiding si-
multaneity of charging periods is another interesting research
topic. Smart strategies could be implemented through smart me-
ters including hourly energy prices and other price signals sent
by DSOs depending on network congestion situations. Finally,
in future scenarios, the combined effects of DG and PEV on net-
work requirements should be investigated.
Voltages at all buses were limited to of rated values.
APPENDIX The present value of annual costs is computed through a given
In Table IV, the main input data for electricity consumers weighted average cost of capital (WACC). Values used were
and PEVs in the two considered distribution areas are shown. within the interval 5.0%–8.5%. Useful life of assets was fixed
The number of customers and the number of charging PEVs are to 30–40 years for lines and transformers and ten years for pro-
given for each scenario and for each operating condition: peak tection and control devices. Energy losses were valued at 36-66
and off-peak hours. In addition, the total power consumption, /MWh.
loads and charging PEVs, and injected power, in case of V2G,
are specified for each network voltage level. ACKNOWLEDGMENT
It can be observed that the PHEV 40 is the most common The authors would like to thank Mr. D. Trebolle from Union
type of PEV considered in the study. This assumption is based Fenosa (Spain) and Ms. B. Doersam from MVV (Germany)
on [4] where it is justified that, in average, people drive less than their collaboration for providing the two distribution areas’ data
40 miles a day. used in this study in the context of the European Project IM-
In Tables V and VI, the relevant parameters for lines/cables, PROGRES
and HV/MV substations and MV/LV transforming centers taken
in this study are shown. REFERENCES
As it can be seen from Table V, in area A, only underground
[1] M. Valentine-Urbschat and W. Bernhart, Powertrain 2020-The Future
cables are available. In addition, investment costs are signifi- Drives Electric Roland Berger Strategic Consultants Report, Sep. 2009,
cantly higher in area A. pp. 1–95. [Online]. Available: http://www.rolandberger.com/.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2025 at 14:15:21 UTC from IEEE Xplore. Restrictions apply.
 PIELTAIN FERNÁNDEZ et al.: ASSESSMENT OF THE IMPACT OF PLUG-IN ELECTRIC VEHICLES ON DISTRIBUTION NETWORKS 213

[2] Electrification Coalition. Electrification Roadmap: Revolutionizing [23] J. Axsen, A. Burke, and K. Kurani, Batteries for Plug-in Hybrid Electric
Transportation and Achieving Energy Security, Nov. 2009. [Online]. Vehicles (PHEVs): Goals and the State of Technology Circa 2008, Inst.
Available: http://www.electrificationcoalition.org/. Transportation Studies, Univ. California, 2008.
[3] Directive 2009/72/EC Concerning Common Rules for the Internal [24] Remuneration of Electricity Distribution Activities in Spain, Feb. 2008.
Market in Electricity and Repealing Directive 2003/54/EC, European [Online]. Available: http://www.cne.es/ (in Spanish)
Communities, 2009. [25] J. P. Peco, “Model for electricity distribution network planning in
[4] W. Kempton and J. Tomic, “Vehicle-to-grid power implementation: a geographical area of service,” Ph.D. dissertation, Univ. Pontificia
From stabilizing the grid to supporting large- scale renewable energy,” Comillas, Madrid, Spain, 2001.
J. Power Sources, vol. 144, pp. 280–294, 2005. [26] L. Pieltain Fernández, “Impact of plug-in hybrid electric vehi-
[5] P. Denholm and W. Short, An Evaluation of Utility System Impacts cles (PHEVs) on power systems,” M.Sc. thesis, Univ. Pontificia
and Benefits of Optimally Dispatched Plug-in Hybrid Electric Vehicles, Comillas, Madrid, Spain, 2009. [Online]. Available: http://www.iit.up-
NREL National Renewable Energy Lab., 2006. comillas.es/docs/TM-09-008.pdf.
[6] A. Ramos, L. Olmos, J. M. Latorre, and I. J. Perez-Arriaga, “Modeling [27] C. Mateo, T. Gomez, A. Sanchez, J. Peco, and A. Candela, “A ref-
medium term hydroelectric system operation with large-scale penetra- erence network model for large-scale distribution planning with auto-
tion of intermittent generation,” in Proc. XIV Latin and Iberian Conf. matic street map generation,” IEEE Trans. Power Syst., submitted for
Operations Research (CLAIO 2008), Cartagena de Indias, Colombia, publication.
Sep. 9–12, 2008.
[7] Environmental Assessment of Plug-in Hybrid Electric Vehicles, Na-
tionwide Greenhouse Gas Emissions, 1015325, Fin. Rep., EPRI and
NRDC, 2007, vol. 1, pp. 1–56.
[8] R. W. McCarthy, C. Yang, and J. M. Ogden, “Interactions between
electric-drive vehicles and the power sector in California,” presented at
the EVS24 Int. Battery, Hybrid and Fuel Cell Electric Vehicle Symp., Luis Pieltain Fernández was born in Ferrol, Spain, in 1985. He received the
Paper UCDavis UCD-ITS-RR-09–11, Stavanger, Norway, May 13–16, Industrial Engineer degree in electricity and the M.Sc. degree in electric power
2009, pp. 1–14. industry from the School of Engineering, Universidad Pontificia Comillas,
[9] D. Pudjianto, C. Ramsay, and G. Strbac, “The FENIX vision: The Madrid, Spain, in 2008 and 2009, respectively.
virtual power plant and system integration of distributed energy re- He was on a scholarship in the Research Institute of Technology at the UPCO
sources,” in FENIX Deliverable 1.4.0, 2007. [Online]. Available: http:// (IIT-UPCO), where he did his M.Sc. thesis on the impact of PEVs on power
www.fenix-project.org. systems and worked as a researcher in training between December 2008 and
[10] D. Six, E. Peeters, M. Hommelberg, C. Batlle, J. Su, F. Bouffard, R. July 2009. He is now with ENAGAS, SA in Madrid.
Belhomme, M. Sebastian, P. Lang, C. Yuen, J. Jimeno, A. Vicino, W.
Fritz, R. Cerero, G. Valtorta, and D. Hirst, “ADDRESS internal report
task 1.1-analysis of ADDRESS concepts and proposed architecture;
first contextual scenarios,” in ADDRESS, Jan. 30, 2009. Tomás Gómez San Román (M’88–SM’09) received the Ph.D. degree in indus-
[11] K. Clement-Nyns, E. Haesen, and J. Driesen, “Coordinated charging trial engineering from the Universidad Politécnica, Madrid, Spain, in 1989.
of multiple plug-in hybrid electric vehicles in residential distribution He is a Professor of electrical engineering at the Engineering School of Uni-
grids,” in Proc. IEEE/PES Power Systems Conf. Exhib., Seattle, WA, versidad Pontificia Comillas (UPCo), Madrid. He has wide industrial experience
Mar. 15–18, 2009. in joint research projects in the field of electric energy systems. His areas of in-
[12] K. Clement-Nyns, E. Haesen, and J. Driesen, “Stochastic analysis of terest are the operation and planning of transmission and distribution systems,
the impact of plug-in hybrid electric vehicles on the distribution grid,” power quality assessment and regulation, and economic and regulatory issues
in Proc. 20th Int. Conf. Electricity Distribution (CIRED), Prague, in the electrical power sector.
Czech Republic, Jun. 8–11, 2009.
[13] K. Clement-Nyns, E. Haesen, and J. Driesen, “The impact of charging
plug-in hybrid electric vehicles on a residential distribution grid,” IEEE
Trans. Power Syst., vol. 25, no. 1, pp. 371–380, Feb. 2010. Rafael Cossent received the Industrial Engineering degree in electrical engi-
[14] J. A. P. Lopes, F. J. Soares, P. M. R. Almeida, P. C. Baptista, C. M. neering from the Universidad Pontificia Comillas-ICAI, Madrid, Spain, in 2007,
Silva, and T. L. Farias, “Quantification of technical impacts and envi- where he is pursuing the Ph.D. degree in industrial and computer engineering.
ronmental benefits of electric vehicles integration on electricity grids,” He is currently an Assistant Researcher at the Instituto de Investigación Tec-
in Proc. Electromotion Conf., Lille, France, Jul. 2009. nológica at the Universidad Pontificia de Comillas. His areas of interest com-
[15] J. A. P. Lopes, F. J. Soares, and P. M. R. Almeida, “Identifying man- prise the integration of renewable energy sources, distributed generation, and
agement procedures to deal with connection of electric vehicles in the the regulation of distribution utilities.
grid,” in Proc. IEEE Power Tech Conf., Bucharest, Romania, Jul. 2009.
[16] J. Román, T. Gómez, A. Muñoz, and J. Peco, “Regulation of distri-
bution network business,” IEEE Trans. Power Del., vol. 14, no. 2, pp.
662–669, Apr. 1999.
[17] J. Peco and T. Gómez, “A model for planning large scale MV networks Carlos Mateo Domingo received the Ph.D. degree in industrial and computer
considering uncertainty in reliability modeling,” in Proc. 6th Int. Conf. engineering from the Universidad Pontificia Comillas, Madrid, Spain, in 2007,
Probabilistic Methods Applied to Power Systems, Madeira, Portugal, with a thesis on modes of elastic wave propagation in electrical cables.
Sep. 25–28, 2000. He is currently a Researcher at the Instituto de Investigación Tecnológica
[18] V. Levi, G. Strbac, and R. Allan, “Assessment of performance-driven (IIT) in the School of Engineering (ICAI) of the Universidad Pontificia
investment strategies of distribution systems using reference net- Comillas. He has worked in national and international projects related to
works,” presented at the IEE Proc. Generation, Transmission and regulation, distribution planning, and quality of service. His research interests
Distribution, 2005. include ultrasound and electricity distribution network models.
[19] C. J. Wallnerström and L. Bertling, “Investigation of the robustness
of the Swedish network performance assessment model,” IEEE Trans.
Power Syst., vol. 23, no. 2, pp. 773–780, May 2008.
[20] A. Navarro and H. Rudnick, “Large-scale distribution planning—part Pablo Frías received the M.S. degree and the Ph.D. degree in electrical engi-
I: simultaneous network and transformer optimization,” IEEE Trans. neering from the Universidad Pontificia Comillas, Madrid, Spain, in 2001 and
Power Syst., vol. 24, no. 2, pp. 744–751, May 2009. 2008, respectively.
[21] R. Cossent, T. Gómez, L. Olmos, C. Mateo, and P. Frías, “Assessing He is currently a Researcher at the Instituto de Investigación Tecnológica
the impact of distributed generation on distribution network costs,” in (IIT) at the Universidad Pontificia Comillas, where he also teaches at the Power
Proc. 6th Int. Conf. European Energy Market, Leuven, Belgium, 2009, System Department of the Engineering School (ICAI). He has participated in
pp. 586–593. many international research projects and in several consultancy projects with
[22] M. A. Kromer, Electric Powertrains: Opportunities and Challenges electricity utilities in Spain and South America. His interests are electric ma-
in the US Light-Duty Vehicle Fleet. Cambridge, MA: Massachusetts chines, ancillary services in power systems, and the integration of distributed
Inst. Technol. (MIT), 2007. generation in power systems.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2025 at 14:15:21 UTC from IEEE Xplore. Restrictions apply.