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

net/publication/277841627

Grid harmonic impact of multiple electric vehicle fast charging

Article  in  Electric Power Systems Research · October 2015

DOI: 10.1016/j.epsr.2015.05.012

CITATIONS READS

52 1,095

4 authors, including:

Alexandre Lucas Evangelos Kotsakis

Technical University of Lisbon European Commission
22 PUBLICATIONS   246 CITATIONS    42 PUBLICATIONS   246 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

North Sea Transnational Grid View project

SGILAB View project

All content following this page was uploaded by Evangelos Kotsakis on 19 February 2019.

The user has requested enhancement of the downloaded file.

 Electric Power Systems Research 127 (2015) 13–21

Contents lists available at ScienceDirect

Electric Power Systems Research

journal homepage: www.elsevier.com/locate/epsr

Grid harmonic impact of multiple electric vehicle fast charging

Alexandre Lucas ∗ , Fausto Bonavitacola, Evangelos Kotsakis, Gianluca Fulli
European Commission, JRC, Institute for Energy and Transport, PO Box 2, 1755 ZG Petten, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: Fast charging is perceived by users as a preferred method for extending the average daily mobility of
Received 16 December 2014 electric vehicles (EV). The rated power of fast chargers, their expected operation during peak hours, and
Received in revised form 11 May 2015 clustering in designated stations, raise significant concerns. On one hand it raises concerns about standard
Accepted 20 May 2015
requirements for power quality, especially harmonic distortion due to the use of power electronics con-
necting to high loads, typically ranging from 18 to 24 kW h. On the other hand, infrastructure dimension-
Keywords:
ing and design limitations for those investing in such facilities need to be considered. Four sets of measure-
Electric vehicles
ments were performed during the complete charging cycle of an EV, and individual harmonic’s amplitude
Harmonics
Power quality and phase angles behaviour were analysed. In addition, the voltage and current total harmonic distortion
Interoperability (THD) and Total Demand Distortion (TDD) were calculated and the results compared with the IEEE519,
Fast chargers IEC 61000/EN50160 standards. Additionally, two vehicles being fast charged while connected to the same
Phase angles feeder were simulated and an analysis was carried out on how the harmonic phase angles would relate.
The study concluded that the use of TDD was a better indicator than THD, since the former uses the max-
imum current (IL ) and the latter uses the fundamental current, sometimes misleading conclusions, hence
it is suggested it should be included in IEC/EN standard updates. Voltage THD and TDD for the charger
analysed, were within the standard’s limits of 1.2% and 12% respectively, however individual harmonics
(11th and 13th) failed to comply with the 5.5% limit in IEEE 519 (5% and 3% respectively in IEC61000).
Phase angles tended to have preferential range differences from the fundamental wave. It was found that
the average difference between the same harmonic order phase angles was lower than 90◦ , meaning that
when more than one vehicle is connected to the same feeder the amplitudes will add. Since the limits
are dependable on the upstream short circuit current (ISC ), if the number of vehicles increases (i.e. IL ), the
standard limits will decrease and eventually be exceeded. The harmonic limitation is hence the primary
binding condition, certainly before the power limitation. The initial limit to the number of chargers is not
the power capacity of the upstream power circuit but the harmonic limits for electricity pollution.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction electricity flow system, poses new challenges. These include the
need for new operational strategies, models or simulation tools,
The paradigm change from centralised unidirectional electric- and better infrastructure design (technological development and
ity flow from plants to consumers into a distributed bidirectional adaptation to a wider distributed grid). The interaction between
distributed agents including smart houses or electric vehicles (EV),
and the electricity grid has prompted the interest of researchers,
industry and policy makers. These interactions between technical
Abbreviations: Ax , X axes component of vector A; Ay , Y axes component of vector
A; Bx , X axes component of vector B; By , Y axes component of vector B; EV, electric agents are supported by social/economic agents like prosumers,
vehicle; E(), phase angle mean value; I1 , fundamental current; Ih , individual current retailers, Distributed System Operators (DSO) or service compa-
harmonic order; IL , maximum demand load current at PCC; Iref , reference current; ISC , nies, which are crucial to drive the interactions at a technical
maximum short-circuit current at PCC; J, load; PCC, point of common coupling; PQ,
level. Standardisation requirements and a better understanding
power quality; PWHC, partial weighted harmonic current; R, resulting amplitude of
a vector; RX , resulting amplitude of vector X axes component; RY , resulting amplitude
of interoperability phenomena, especially the impacts that large
of vector Y axes component; RSCE, short-circuit ratio; SOC, state of charge; TDD, total scale integration of distributed agents may have, are subjects
demand distortion; THD, total harmonic distortion; THDI , current total harmonic of interest for industry, utilities, regulatory and policy making
distortion; THDV , voltage total harmonic distortion; Z, impedance; Uh˙ , resultant organisations. Among others, interoperability studies focus on
vector of a harmonic order;  h , phase angle of a vector related to the fundamental;
enabling universal connectivity, batteries and EV interaction with
(), phase angle standard deviation.
∗ Corresponding author. Tel.: +351961741327. the grid and other agents, and identifying gaps in standards or
E-mail address: [email protected] (A. Lucas). technologies.

http://dx.doi.org/10.1016/j.epsr.2015.05.012
0378-7796/© 2015 Elsevier B.V. All rights reserved.
14 A. Lucas et al. / Electric Power Systems Research 127 (2015) 13–21

The electricity industry has recommended Mode 3 EV charging common that only the harmonics amplitude levels are observed,
(quick, single-phase or three-phase options) (IEC 61851) [1] as the as the utilities are required to keep the harmonics levels under a
preferred solution for all types of locations, making fast charging given limit. Authors in Ref. [20] however note that if the diversity
an important area to be addressed. Topics for the analysis by which of chargers is not taken into account, the harmonic problems could
EV charging impacts distribution networks can be listed as volt- be overestimated. One of the first papers in this area was actually
age regulation, harmonic distortion levels, unbalances, additional presented by Ref. [21] where multiple different EV chargers in the
losses and transformers loss of lifetime. Battery chargers for EVs network have been observed. It is reported, that 10% smaller har-
employ nonlinear switching devices which may result in signifi- monic current magnitudes were observed compared to the simple
cant harmonic voltage and currents injected into the distribution summing of magnitudes. Another study [22] applies a methodology
system. Literature reports different findings regarding the impact which accounts for diversity of SOC and initial charging moments
on power quality from EVs. Some authors state that distribution in California. The results indicate that accounting for variation in
infrastructure can have limitations with EV charging supply even start-time and SOC in the analysis leads to reduced estimates of
for relatively low EV penetration levels [2–5]. Other studies [6–11], harmonic current injection. Authors argue that traditional methods
suggest that low EV penetration levels, with normal charging rates, do not account for these variations. Researchers show that from the
will have acceptably low harmonic levels and voltage variations, point of view of the substation transformer, the impact of EV’s is
however fast charging rates could cause significant voltage har- mainly one of power and energy, rather than harmonics. Analysis
monics and losses. Most of the studies tend to focus only on current with real and imaginary components for each harmonic has been
harmonics, addressing as the main concern residential and normal described in Ref. [7] which analyses 20 kW h charges and reports a
chargers, as they are expected to have higher penetration. There is THDI over 40% at the connection point. The 11 kV medium-voltage
however, a very limited number of studies which analyse both volt- network was simulated with 36 chargers, each at power level of
age and current harmonics focusing on fast charging specifically 8.2 kV A, which makes it difficult to observe the total cancellation
performed in a cluster of chargers connected to the same feeder effect.
[12,13]. The motivation of the studies tends to be the same, high There is still a lack of studies focusing on clustering fast chargers
expected impact on energy demand and expected usage during and the impacts on both THDI and THDV (voltage total harmonic
peak hours. distortion) referring specifically to fast charging. It is critical to
Considerable literature focuses on the distribution networks study the phase angles in order to understand how the amplitude
especially residential networks [14,15], where EV charging could of the harmonics measured will sum when considered part of a
bring an additional severe power electronic load and associated cluster (connected to the same feeder). The main goal of this study
power quality issues. The European standard for public power sup- is to clarify the following questions: (i) investigate the voltage and
ply EN 50160 [16], sets conditions for: voltage magnitude variation, current Total Harmonic Distortion (THD) caused by fast charging
voltage harmonics, inter-harmonic voltage and voltage unbalance one single electric vehicle and standard limits compliance. (ii) how
among others. All loads that are connected to the power network does the THD caused by fast charger/EV load vary along the charg-
must provide a sufficiently low effect on the network, so that it ing cycle if at all?, and (iii) does the THD and TDD caused by charging
does not cause a violation of the power supply conditions stated two EVs simultaneously with the same fast charger decrease due
in this standard. This means that the EV chargers, once connected to phase cancellation?
to a public network, must not influence the network operation to
the extent that can cause deviation from the standard. In gen-
eral, EV chargers have to fulfil requirements for loads that can 2. Theoretical background
be connected to an electric power network as described by the
electromagnetic compatibility IEC 61000 series standards. These Harmonic distortion is a deviation of the current or voltage
standards set the emission levels, including the harmonic currents waveform from a perfect sinusoidal shape. In the case of nonlin-
or power factor that a charger is allowed to have. The standards ear loads, such as EV charge controllers, current distortion is very
applied to the low-power EV chargers are IEC 61000-3-2 [17] and common due to the need of power electronics switches to con-
IEC 61000-3-4 [18], which set limits to the harmonic emissions vert power from AC to DC form. Introduction of these currents
generated by the charger. Another study based on practical mea- into the distribution system can distort the utility supply voltage
surements of charging commercial EVs [8] presents a maximum and overload expensive electrical distribution equipment. In order
THDI of 17.3% for a level III charger at the end of the charge, to prevent harmonics from negatively affecting the utility supply,
and maximum of 19.2% for Level I and II also at the end. This standards such as IEC 61000-3-12 [23]/2-4 [24] or IEEE Standard
publication acknowledges that TDD use would improve the con- 519-1992 [25], were established with the goal of developing rec-
clusions regarding the distortion impact. Results from a case study ommended practices and requirements for harmonic control in
in Portugal [9] found a THDI of 11.6%, during the constant charg- electrical power systems. These standards, widely adopted by the
ing stage in a fast charging station integrated in a commercial industrial and research communities, describe the problems that
facility. unmitigated harmonic current distortion may cause within elec-
A typical distribution network has a large number of different trical systems as well as the degree to which harmonics can be
non-linear loads connected to it. Authors in Ref. [19] argue that tolerated by a given system. Utilities are obliged to provide power
adding EV chargers from different manufacturers may result in a quality whose limits among others depend on the level of voltage
variety of different harmonic patterns. The diversity of the patterns connection. End users on the other hand, are responsible for not
may lead to notable harmonic cancellation. This effect occurs when degrading the voltage of the utility by drawing significant nonlinear
harmonics with different phase angles provide a sum in the magni- or distorted currents. Power quality, specifically harmonic impact
tude that is smaller than the individual harmonics magnitudes. It is in Points of Common Coupling (PCC) is a subject of interest to both
however rather complicated to evaluate this effect. Authors in Ref. parties. The PCC with the consumer/utility interface is the closest
[10] studying low voltage nonlinear loads also suggest that cancel- point on the utility side of the customer’s service where another
lation is more probable as the number of consumers and appliances utility customer is or could be supplied. The goal of applying the
increase. It has also been indicated that harmonic cancellation is harmonic limits specified in the standards is to prevent one cus-
more expected at higher harmonic orders, which can then account tomer from causing harmonic distortions to another customer or
for the relatively minor THDI decrease. In most papers, it is rather the utility.
 A. Lucas et al. / Electric Power Systems Research 127 (2015) 13–21 15

two vectors A and B in Eqs. (7)–(8) and (9)–(10), where A1 and A2 are
the vector’s amplitude and  1 and  2 are the corresponding angles:

Ax = A1 cos 1 (7)

Ay = A1 sin 1 (8)

Bx = A2 cos 2 (9)
Fig. 1. Simplification example of two loads connection to the same grid feeder.
By = A2 sin 2 (10)

2.1. Harmonics After obtaining the RX (by adding the X and Y components), the
resulting amplitude (R) in Eq. (11) and angle ( R ) in Eq. (12) are
Harmonics topic theory is a well-covered subject. In practical given by
terms however, it is difficult to assess the phase angles from each 
harmonic order, and therefore to make valid assumptions regarding R= RX2 + RY2 (11)
the way they add up when multiple devices interact. Typically,
probabilistic approaches are made [20,26], and often studies will
R 
Y
R = tan−1 (12)
treat the vector summation considering high uncertainty or the RX
worst case scenario [27]. For this reason literature results and con-
clusions often differ. For example in Fig. 1, two loads J1 , J2 connected 2.2. Standard IEEE 519, IEC 61000 and EN 50160
to a grid with the impedance Zh are considered.
The vectors U in Fig. 1, with harmonic voltage order h, will sum IEEE 519-1992 and IEC 61000-3-12/2-4 are respectively the
(Uh˙ ) according to Eq. (1), where  is the phase angle related to the American and International standards which apply to the case
fundamental. under study. Both discuss the impacts that harmonic distortion
   can have on distribution assets, particularly transformers, power
Uh˙ = 2 + U 2 + 2U U
Uh1 h2 h1 h2 cos h2 − h1 (1) cables, capacitors, metering, relaying and switch gear. Both propose
limits for voltage and current distortions and limits for individual
However if the angles are unknown and if no probability func- frequencies. IEEE 519 presents the voltage limits, and still makes
tion exists for , one can use the properties of a uniform distribution a clear distinction between THDI and TDD concepts. EN 50160
to deduce the probability of a conservative summation by upper presents the main voltage level parameters and their permissible
and lower deviation phase angles establishing as shown in Eq. (2), deviation ranges at the customer’s PCC in public low voltage and
the limit between 0 and  where, medium voltage electricity distribution systems. However the load
  1
current is not relevant to EN 50160. Regarding the actual current
f  = ,  ∈ [0, ] (2) harmonic limits, the European standards are the same as within

IEC, hence only the latter will be referred to from this point. Table 1
The following equations shows the expected mean value shows the Voltage THD limits for different voltage levels. Similarly
obtained by Table 2 shows the limits for the TDD and individual harmonics
according to each voltage level. Tables 3 and 4 show the same limits

     set in IEC 61000-3-12/2-4 respectively.
E  = f  d = (3) ISC /IL ratio shows the relative size of the load compared to the
2
0 utility system. Power systems in a given point (under a linearity
hypothesis) can be transformed into a Thevenin equivalent with the
and the corresponding standard deviation in the following equa- related impedance. The short circuit, which may also be expressed
tions is in short-circuit power (SCP) in MVA, at that point “quantifies” the

 equivalent impedance of the network. If it is high (i.e. low SCP)
      
2
 
  =   2 f  d − E  = √ (4)
2 3 Table 1
0 Voltage distortion limits set in IEEE 519-1992.

Bus voltage at PCC Individual voltage Total voltage

This can give the upper and lower phase angles estimations as distortion (%) distortion THD (%)
follows in the following equations:
69 kV and below 3 5
      1
 69.001 kV through 161 kV 1.5 2.5
o
upper =E  +  = 1+ √ → 141.96 , p = 78.86% 161.001 kV and above 1 1.5
2 3
(5)
Table 2
Maximum harmonic current distortion in percent of IL set in IEEE 519-1992.
    
 1

Individual harmonic order (odd harmonics)
lower = E  +   = 1+ √ → 38.04o , p = 21.13%
2 3 ISC /IL <11 11 ≤ h < 17 17 ≤ h < 23 23 ≤ h < 35 35 ≤ h TDD
(6)
<20 4.0 2.0 1.5 0.6 0.3 5.0
20 < 50 7.0 3.5 2.5 1.0 0.5 8.0
50 < 100 10.0 4.5 4.0 1.5 0.7 12.0
In the event the statistical distribution or exact angles are 100 < 1000 12.0 5.5 5.0 2.0 1.0 15.0
known, they will add together when their difference is below 90◦ >1000 15.0 7.0 6.0 2.5 1.4 20.0
(add perfectly if 0◦ ) or subtract if below (and cancel each other if IL —Max. demand load current and ISC —short circuit current. TDD—total demand
180◦ ). Moreover the resulting angle of each sum can be calculated distortion, harmonic current distortion in % of maximum demand load current. Even
analytically by decomposition of the X and Y components. Consider harmonics are limited to 25% of odd harmonic limits above.
16 A. Lucas et al. / Electric Power Systems Research 127 (2015) 13–21

Table 3
Maximum harmonic current distortion in percent of IL set in IEC 61000-3-12.

Minimum RSCE Admissible individual harmonic current Ih /Iref (%) Admissible harmonic parameters (%)

I5 I7 I11 I13 THC/Iref PWHC/Iref

33 10.7 7.2 3.1 2 13 22

66 14 9 5 3 16 25
120 19 12 7 4 22 28
250 31 20 12 7 37 38
≥350 40 25 15 10 48 46

The relative values of even harmonics up to order 12 shall not exceed 16/h%. Even harmonics above order 12 are taken into account in THC and PWHC in the same way as
odd order harmonics. Linear interpolation between successive RSCE values is permitted. RSCE —short-circuit ratio; Ih —harmonic current component; Iref —reference current;
THC—total harmonic current; PWHC—partial weighted harmonic current.

Table 4
Voltage distortion limits set in IEC 61000 2-4.

Harmonic order n Class 1 ␮n [%] Class 2 ␮n [%] Class 3 ␮n [%]

(non multiples of 3)

5; 7 3; 3 6; 5 8; 7
11; 13; 17 3; 3; 2 3.5; 3; 2 5; 4.5; 4
THDV 5% 8% 10%

Class 1: compatibility level lower than public (laboratory instrumentation, some

protection equipment, etc.). Class 2: compatibility level equal to public (any equip-
ment designed for supply from public networks). Class 3: compatibility level higher
than public (equipment in the presence of welding machines, rapidly varying loads,
large converters, etc.).

the network is “weak” and the voltage is affected by the (har-

monic) currents; if it is high (infinite), the impedance is zero and
the network is strong and the voltage is not affected. It is therefore
necessary to calculate or to measure the short circuit current (ISC ) at
the PCC where the measurements are intended. TDD is very similar
to THD, except for the denominator as shown in Eqs. (13) and (14).
In TDD, harmonics are expressed as % of IL (maximum demand load
current) whereas THD express harmonic content as % of I1 (fun-
damental current). For the IL it is recommended to consider the
maximum averaged current of at least a 15–30 min interval of the Fig. 2. Simplified single line diagram of upstream electricity grid, from PCC to MV
last 6 months for a given consumer. level.

I22 + I32 + I42 + I52 + · · ·
THDI = (13) lights). The balance of the system was verified successfully. The 4
I1 sets of measurements were performed from different low states of
charge 8%, 7%, 5%, and 10% respectively, to 100% SOC in a maximum
I22 + I32 + I42 + I52 + · · ·
TDD = (14) of 42 min.
IL
Before starting the measurements, in addition to phase
Usually the THD is measured first and then a comparison is made sequence verification, an initial verification procedure was per-
to the limits, if there is a problem, then the TDD is calculated. It is formed. This was intended to mitigate the fact that not all
rarely needed to convert to the TDD and % of IL , which is why the measurements were recorded at the same time, and that no volt-
THD concept is much better known. age control source was used. A set of three files were recorded per
measurement: (i) no load, (ii) only with the charger connected and
2.3. Test design (iii) with load. This was intended to verify: frequency fluctuation,
voltage fluctuation, THDV with no load and THDV only with the fast
Four measurements were performed on a full electric vehicle charger connected.
using a fast charger. As a measurement device the Fluke 437 Series The upstream grid representation is shown in Fig. 2, as well as
II [28], 400 Hz Power Quality and Energy Analyser was used, set the point (PCC) where the measurements were performed. The PCC
with 0.25 s time step data acquisition. The harmonics were reg- in theory will often be at the medium voltage level, i.e. the primary
istered until 2500 Hz. The resolution and accuracy of the THD for of the distribution transformer serving the users, irrespective of
both voltage and current is 0.1% and ±2.5% of the reading respec- transformer ownership or the location of the metering system. In
tively, whereas for the phase angles it is 1◦ with an accuracy of practical terms however it is often more secure or accessible to
±n × 1◦ (where n is the harmonic order). A commercial fast charger perform such measurements on the transformer secondary, as is
[29] was connected to a max 63 A outlet, 230 V, 50 Hz at one end the case presented in this analysis. The present study considers that
and at the other, with the chargers manufacturer’s cable, to a full a distinction of consumers would be done at the PCC point shown in
electric vehicle VW model E-Up, with an 18.7 kW h battery pack. Fig. 2. An impedance meter (Impmeter 2) was used to record the ISC
The vehicle was discharged over random driving cycles and the at the PCC and with each IL from each measurement the standard
different measurements were performed on different days. The lab- limits were identified.
oratory temperature was approximately 25 ◦ C in all measurements. To accomplish the first and second goals of this research, after
All loads inside the car were disconnected (air conditioning, radio, the measurements were performed, PowerLog 4.2 software [28]
 A. Lucas et al. / Electric Power Systems Research 127 (2015) 13–21 17

Fig. 3. Current behaviour during a charging cycle (L1 ).

Fig. 4. THDI and TDD during a complete charging cycle (L1 ).

was used to import and verify the reading. The data from the four
sets of measurements were then exported to spreadsheets. Values
of THD were observed and compared with TDD, the ARMS cur-
rent during charging cycle was registered and all shown in the
result and discussion chapter. From the absolute and relative val-
ues, a comparison with the standard limits was possible. It was also
important to obtain these values in order to simulate the THD and
TDD with one and two electric vehicles working together. Since
the system is balanced, the analysis was only performed in one
phase. To pursue the third challenge of this study, apart from the
amplitudes of the harmonics, also their phase angles were exam-
ined. Using the Crystal-ball excel add-in, the time series of each
angle and phase were submitted to a curve fit calculator (based
on Anderson–Darling statistical test). The phase angles from each Fig. 5. THDV during a complete charging cycle (L1 ).

frequency were analysed and their corresponding statistical distri-

butions were analysed regarding their range differences. After this distinguished during the last 15 min corresponding to 77%–100%
analysis, a simulation was carried out with the goal of obtaining SOC where the current starts decreasing, making the THD reach a
the corresponding TDD resultant from charging one vehicle or two maximum of 41% and TDD drop to 0%. The graphic in Figs. 3 and 4
vehicles simultaneously. As inputs for the simulation, the complete corresponds to just over 40 min of charge.
data set of approximately 40 min was considered from two differ- Values of both TDD and THDI are coherent with others presented
ent load measurements from one single phase. Using Crystal-ball, found in literature [8,9]. TDD calculations were based on the max-
both absolute values of amplitudes, and the phase angles statistical imum current of 67.5 A demand during the charging cycle (even
distributions were used to apply Eq. (1) to the harmonics. By using though a peak of 76.7 A was recorded, it only lasted 40 s and there-
Eq. (8) the two TDD were found and compared. fore was neglected). Maximum value of TDD was 13.12% and for
the THDI it was 47%. The readings enhance the need to separate the
3. Results and discussion analysis using the fundamental and the maximum demand current.
Regarding the THDV , Fig. 5 presents the complete behaviour during
3.1. Results from measurements the charging cycle for phase L1 .
The time evolution is coherent with the current one as shown
Measurements were taken for approximately 40 min with 0.25 s in Fig. 3. The voltage starts decreasing at approximately 75% SOC,
time steps, generating over 9000 records for each unit. A simplifica- which also happens with the current, meaning that this stage has a
tion of 7 events during the charging cycle is given in Tables A1–A4 lower power being fed into the battery (coherent with lithium-ion
of the Annex. All the odd harmonics until the 25th are shown, as cells technology charging curve). The lower power in this stage is
well as the THDI of each phase in the corresponding minute. In all the reason why the last 12 min only charge about 15% of the charge.
measurements it is possible verify a high predominance of the 3rd, The Voltage THD in the initial stage reaches 1.26% distortion and it
5th, 7th, 9th, 11th, 13th harmonics. Furthermore, higher variabil- stabilises 1.16% during the constant stage, dropping to 0.7% at the
ity in the phase angles is observable in the 3rd and 9th harmonic final stage, which are all below the 5% limit [25]. The harmonic his-
(Figs. A1–A6), whereas the other odd harmonics present a rela- togram for both voltage and current can be observed in Fig. 6. Even
tively narrow angle range. The THDI tends to increase at the end of harmonics were excluded due to their low contribution and only
the charging cycle, which can be explained by the decrease of the harmonics until the 25th are shown, since above this their values
current at the end of the cycle, as shown in Fig. 3. This however, are negligible. One can verify that the odd harmonics are the most
can be misleading if only the THDI is considered, since it consid- significant especially the 3rd, 5th, 7th, 9th, 11th and 13th. Their
ers the fundamental current as reference. The THDI can reach as lower or higher relative values as shown in the picture do not reveal
high as 40% in the 4th measurement shown in Table A6, however the total importance of their individual analysis, since their limits
the current which it refers to is below 10 A instead of 67.5 A during as set by the standards become lower as the frequency increases.
most of the charging, where the TDHI is approximately 11%–12%. For this reason all odd harmonics until the 25th were analysed. For
For this reason, the analysis of the TDD which considers the maxi- the analysis of the phase angle range, each odd harmonic frequency
mum current instead of the fundamental is the most elucidative of was compared with the other measurements by phase.
that which is under study. From Figs. 3 and 4 three distinct stages An example is given in Fig. 7 for the 9th harmonic and phase
can be observed: First during the first 2–3 min of charging with very 1, which even though it is not one of the highest impacting the
high THDI peaking to more than 45% while TDD starts low, followed distortion, from the data analysed it is the one which presents the
by a second stage with a constant behaviour where TDD and THD highest variations in terms of angle range (at this harmonic order
present very close values of 11.5% to 12.5%. A third stage can be the accuracy is ±9◦ ). The range however, is not uniform and actually
18 A. Lucas et al. / Electric Power Systems Research 127 (2015) 13–21

Fig. 6. Voltage and current harmonic histogram.

Fig. 7. Analysis of the 9th harmonic phase angle range (a) 1st, (b) 2nd, (c) 3rd and (d) 4th measurement for L1 .

tends to have a higher density (in terms of event number) around even with higher angle variability, the mean values from their com-
an average. This means that a statistical distribution can be drawn mon referential (the fundamental) vary between 128◦ and 158◦ and
from all the harmonics and ranges. from each other their difference will be below 90◦ . This means an
Other harmonic frequencies are shown in Figs. A1–A6 in the amplitude summation is possible most of the time.
Annex, and these also show preferential range angles.
Comparing the statistical distributions between the 4 measure- 3.2. Comparison with the standard limits
ments of all harmonics will provide the probability of two angles
being within a range, i.e. if their amplitudes will tend to sum or As it was verified that the current and the THDI changed during
subtract. Computing all events in Crystal-ball software, a distribu- the charging cycle, for the most even analysis the TDD will be used,
tion can be drawn with the corresponding mean value and standard hence for the sake of the research analysis a higher focus will be
deviation. This was performed for all the significant harmonics (odd given to the IEEE 519 standard, but it is also applicable to IEC stan-
harmonics until the 25th). Fig. 8 shows the statistical distribution dards. Apart from the TDD limit values presented in the standard,
of the 4 measurements when analysing the 9th harmonic variation individual harmonics must also comply with the limits. From the
shown in Fig. 7. harmonic histogram in Fig. 6, it can be verified that some harmonics
In this example the best fit distribution for the first and third such as the 11th could be out of the limits as set by both IEEE 519 and
measurements of the 9th harmonic correspond to a Logistic Distri- IEC 61000-3-12/2-4 standards. To identify the limits in Tables 1–4
bution. Likewise the second and fourth measurements correspond the ISC value was measured in the General Low Voltage Main Cabi-
to a Minimum Extreme Distribution. As can be observed from Fig. 8, net (PCC under analysis). Values ranged from 8080 A to 8480 A. The
 A. Lucas et al. / Electric Power Systems Research 127 (2015) 13–21 19

Fig. 8. Example of the 9th harmonic phase angles range statistical distributions for the four measurements.

Fig. 9. Forecast of individual 11th, 13th, 23rd and 25th harmonics distortion.

ISC /IL was calculated considering the most unfavourable scenario: Fig. 5 shows the voltage distortion being within the limit of 5%, and
ISC = 8.08 kA; IL = 67.5 A → Ratio = 119.7 hence complying with the standard.
According to Tables 2 and 3, the TDD and THD limits are 15%
and 16% respectively with the corresponding individual harmonic 3.3. Simulating two electric vehicles
limits. Using the first measurement as an example, Crystal-ball tool
was used to consider all values of the distortion during the cycle. Simulating two vehicles will allow verifying if in fact there is
The corresponding distributions regarding the individual impact of no tendency for phase cancellation, meaning that TDD average
the harmonics were obtained, shown only for 11th, 13th, 23rd and value should suffer negligible change. Moreover, adding another
25th in Fig. 9. Is it possible that those harmonics may be in violation fast charger/vehicle to the system will show how the harmonic
of the IEEE 519 and IEC 61000 3-12. Fig. 9 shows that the limits of limits of the standards to be consider change. To simulate two vehi-
5.5% from Table 2 or 5% and 3% from Table 3 are exceeded by the cles connected to the same feeder (both working at 67.5ARMS), two
11th and 13th, and even though less probable, the 23rd and 25th measurement cycles (1 and 2, Tables A1–A2 of the Annex) were con-
may break the limit of 2% as well. sidered. These two measurements represent two different vehicles
As seen in Fig. 4, TDD complies with the standard, but stays charging, each with its own initial SOC, phase angles, and harmonic
above 12% most of the charging cycle. It is important to point out readings. All odd harmonics until 25th and corresponding data from
that the ratio ISC /IL was close to, and just below 100. This would a complete charging cycle were considered in Crystal-ball as inputs.
cause the limits of the TDD to be 12% and in this case the charger Since the values of the phase angles are not always constant, the
would fail to meet these requirements as well. Regarding the THDV , whole range was inputted as the best fit statistical distribution and
20 A. Lucas et al. / Electric Power Systems Research 127 (2015) 13–21

Fig. 10. TDD for 1 and 2 electric vehicle simulation.

ran for 10,000 trials. All other harmonics, not analysed individu- Regarding the resulting assessment with the simulation of two
ally, were considered and have a fix amplitude of 6.1 A (to reach vehicles, it was verified that neither a synchronisation nor a ran-
the measured real TDD of 12%) and that their angles would add as dom behaviour occurred in phase angles. Instead the phase angles
well. The TDD was forecast for 1 and 2 vehicles and the results are tended to have preferential angle difference to the fundamental
shown in Fig. 10. wave. From their statistical distribution study, one can observe that
The simulation results with one and two vehicles report very the differences between the same harmonic order (of each mea-
close values with a mean of 12% for the TDD. As expected, with surement) are lower than 90◦ , which means that with the same
two vehicles the mean value slightly decreased (11.7%) since the charger type they will add, suggesting that there is an upper limit
amplitudes do not sum up perfectly. Furthermore the standard to the number of vehicles to be considered allowable in the system.
deviation seems to decrease when two vehicles are considered. This Should the number of EVs increase, i.e. IL , the standard limits will
means that if the number of vehicles increase, the TDD will tend to decrease, reaching a point where harmonic limits are exceeded. It
be the same or to have a slightly decrease. Once again, this hap- is therefore dependent on the robustness of the systems in terms of
pens because preferential phase angles will make the amplitudes the foreseen short circuit current and the amount of current drawn
on average to add up ( < 90) but not perfectly (0◦ ). Furthermore by the vehicle cluster. Such types of infrastructure may need dedi-
as the current IL also increases linearly with the EV number, the cated systems depending on the robustness (if short-circuit values
distortion remains the same. However, if the ISC is maintained this are low) of the upstream system. Apart from the amplitude value
will eventually cause the violation of the limits set in the IEEE-519 of the harmonics, which may indeed differ between manufactur-
standard as observed in Figs. 9 and 4. If more vehicles are con- ers (IGBTs, active filtering, etc.), the phase angle behaviour in other
nected this will not increasing the harmonic impact, however the chargers should be verified.
IL value will increase, making the ratio of ISC /IL decrease and caus-
ing the harmonic limits to be exceeded eventually (if infrastructure
Appendix A. Supplementary data
is unchanged). It is hence, extremely advisable that sufficiently
high short-circuit power should be available at the interconnection
Supplementary data associated with this article can be found, in
point.
the online version, at http://dx.doi.org/10.1016/j.epsr.2015.05.012

4. Conclusions
References
Three goals were set for this research, first to investigate the
[1] IEC, IEC 61851-1. Electric Vehicle Conductive Charging System—Part 1: General
total voltage and current harmonic distortion impact caused by fast Requirements, IEC, 2010.
charging an electric vehicle and maintaining standard limit com- [2] P. Richardson, D. Flynn, A. Keane, Impact assessment of varying penetrations
pliance. Second, understand how the total harmonic distortions of electric vehicles on low voltage distribution systems, in: IEEE PES General
Meeting, 2010, pp. 1–6.
caused by fast charger/EV load vary through the charging cycle. [3] V. Tikka, J. Lassila, J. Haakana, J. Partanen, Case study of the effects of electric
Finally, simulate if the distortions caused by charging more than vehicle charging on grid loads in an urban area, in: Second IEEE PES Interna-
one EV with the same charger model simultaneously, will decrease tional Conference and Exhibition on Innovative Smart Grid Technologies, 2011,
pp. 1–7.
due to phase cancellation. THDV , THDI and TDD were calculated
[4] F. Musavi, M. Edington, W. Eberle, W.G. Dunford, S. Member, Evaluation and
reporting the former at 1.2% and the latter two at 12% impacts efficiency comparison of front end AC–DC plug-in hybrid charger topologies,
respectively. For the charger under consideration, during the con- IEEE Trans. Smart Grid 4 (1) (2012) 413–421.
[5] F. Lambert, A. Georgia, L.B. Suggs, G. Power, C. John, B. Sisco, E. Morales, V.P.
stant cycle stage, total values complied with the IEEE 519 and
Company, Residential harmonic loads and EV charging, in: Power Engineering
61000-3-12/2-4 standard limits. However, individual harmonics Society Winter Meeting. 2 (C), 2001, pp. 803–808.
failed to do so, mostly due to the 11th and 13th orders which are [6] M. Basu, K. Gaughan, E. Coyle, Harmonic distortion caused by EV battery charg-
likely to exceed the 5.5% limit in IEEE 519 (5% and 3% respectively ers in the distribution systems network and its remedy harmonic distortion
caused by EV battery chargers in the distribution, in: 39th Int. Universities
in IEC61000 2-4). Furthermore, also the 23rd and 25th harmonics Power Engineering Conferences, 2004, pp. 869–873.
even though less probable, may be in violation of their own individ- [7] E. Lo, D. Sustanto, C.C. Fok, Harmonic load flow study for electric vehicle charg-
ual limits. From the results of the experiment, it can be seen that the ers, in: Int. Conference on Power Electronics and Drive Systems, PEDS’99, July,
1999, 1999, pp. 495–500.
primary limitation of the number of chargers/vehicles in a cluster [8] N. Zimmerman, R. Bass, Impacts of electric vehicle charging on electric power
is not the power capacity of the upstream power transformers, but distribution systems, in: OTREC-SS-731, September, 2013, Portland, OR, 2013,
the harmonic limits for electricity pollution. For the standards limit pp. 1–53.
[9] N. Melo, F. Mira, A. de Almeida, J. Delgado, Integration of PEV in Portuguese
analysis the TDD should be applied instead of the THD since there distribution grid: analysis of harmonic current emissions in charging points,
is a variation of the current during the cycle, and hence an oppor- in: 11th International Conference on Electrical Power Quality and Utilisation,
tunity for the European standard to improve in future versions. 2011, pp. 1–6.
 A. Lucas et al. / Electric Power Systems Research 127 (2015) 13–21 21

[10] K. Kim, C.S. Song, G. Byeon, H. Jung, H. Kim, G. Jang, Power demand and total har- [20] P.T. Staats, W.M. Grady, A. Arapostathis, R.S. Thallam, A statistical analysis of
monic distortion analysis for an EV charging station concept utilizing a battery the effect of electric vehicle battery charging on distribution system harmonic
energy storage system, J. Electr. Eng. Technol. 8 (5) (2013) 1234–1242. voltages, IEEE Trans. Power Deliv. 13 (2) (1998) 640–646.
[11] A.H. Foosnæs, A.N. Jensen, N.C. Nordentoft, Report: Case Studies of Grid Impacts [21] J.A. Orr, A.E. Emanuel, D.J. Pileggi, Current harmonics, voltage distortion,
of Fast Charging, Edison Consortium, D 4.4.3 & D 4.4.4 Copenhagen, September, and powers associated with electric vehicle battery chargers distributed
2011, pp. 1–62. on the residential power system, IEEE Trans. Ind. Appl. 20 (4) (1984)
[12] C.H. Dharmakeerthi, N. Mithulananthan, T.K. Saha, Impact of electric vehicle 727–734.
fast charging on power system voltage stability, Int. J. Electr. Power Energy [22] Electric Power Research Institute, The Harmonic Impact of Electric Vehicle
Syst. 57 (2014) 241–249. Battery Charging, TR-108540, Electric Power Research Institute, Palo Alto, CA,
[13] V.L. Nguyen, T. Tran-Quoc, S. Bacha, Harmonic distortion mitigation for electric August, 1997, pp. 1–184.
vehicle fast charging systems, in: PowerTech (POWERTECH), 16–20 June, 2013, [23] IEC, IEC 61000-3-12: Electromagnetic Compatibility (EMC)—Part 3–12:
2013, pp. 1–6. Limits—Limits for Harmonic Currents Produced by Equipment Connected to
[14] M.A.S. Masoum, P.S. Moses, S. Deilami, Load management in smart grids consid- Public Low-Voltage Systems with Input Current > 16 A and ≤75 A per Phase,
ering harmonic distortion and transformer derating, in: Innovative Smart Grid 2011.
Technologies (ISGT), 19–21 January, 2010, 2010, pp. 1–7. [24] IEC, IEC 61000-2-4, Electromagnetic Compatibility (EMC)—Part 2–4:
[15] C. Jiang, S. Member, R. Torquato, D. Salles, Method to assess the power-quality Environment—Compatibility Levels in Industrial Plants for Low-frequency
impact of plug-in electric vehicles, IEEE Trans. Power Deliv. 29 (2) (2014) Conducted Disturbances, 2014.
958–965. [25] IEEE, IEEE 519-1992, IEEE Recommended Practices and Requirements for Har-
[16] EN, Voltage characteristics in public distribution systems, in: EN 50160, Euro- monics Control in Electric Power Systems (ANSI), IEEE, New York, NY, 1992.
pean Standard, 2010. [26] J.H. Enslin, P.J.M. Heskes, Harmonic interaction between large number of dis-
[17] IEC, IEC 61000-3-2 ed3.0:2005, Electromagnetic Compatibility (EMC)—Part tributed power inverters and the distribution network, IEEE Trans. Power
3–2: Limits–Limits for Harmonic Current Emissions (Equipment Input Current Electron. 19 (6) (2004) 1586–1593.
<=16 A per phase), 2005. [27] Y. Xiao, X. Yang, A grid harmonic summation method based on the probability
[18] IECTS, IECTS 61000-3-4 ed1.01998, Electromagnetic Compatibility (EMC)—Part assessment of harmonic phase angles, in: Proc. 14th Int. Conf. Harmon. Qual.
3–4 Limits—Limitation of Emission of Harmonic Currents in Lowvoltage Power Power (ICHQP), 26–29, September, 2010, 2010, pp. 1–6.
Supply Systems for Equipment with Rated Current Greater Than 16 A, 1998. [28] Fluke Corporation, User manual for Fluke series 434-II/435-II/437-II rev.1, EU
[19] C.H. Dharmakeerthi, N. Mithulananthan, T.K. Saha, Overview of the impacts of 2012, pp. 174.
plug-in electric vehicles on the power grid, in: Innovative Smart Grid Technolo- [29] EFACEC, Technical brochure for quick charges station model Q45 - Overview.
gies Asia (ISGT) IEEE PES, 13–16 November, 2011, 2011, pp. 1–8. EFACEC Communication Office, Porto, 2012, pp. 2.

View publication stats