You are on page 1of 10

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

net/publication/273392210

Magnetic Field Exposure Assessment in Electric Vehicles

Article  in  IEEE Transactions on Electromagnetic Compatibility · February 2015


DOI: 10.1109/TEMC.2014.2359687

CITATIONS READS
21 3,298

6 authors, including:

Andrea Vassilev Alain Ferber


Cea Leti SINTEF
13 PUBLICATIONS   160 CITATIONS    33 PUBLICATIONS   161 CITATIONS   

SEE PROFILE SEE PROFILE

Christof Jan Wehrmann Olivier Pinaud


Technische Universität Braunschweig G2Elab
4 PUBLICATIONS   23 CITATIONS    7 PUBLICATIONS   27 CITATIONS   

SEE PROFILE SEE PROFILE

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

EC - FAR View project

Flexible pipes subsea monitoring View project

All content following this page was uploaded by Alain Ferber on 06 April 2015.

The user has requested enhancement of the downloaded file.


This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY 1

Magnetic Field Exposure Assessment


in Electric Vehicles
Andrea Vassilev, Alain Ferber, Christof Wehrmann, Olivier Pinaud, Meinhard Schilling,
and Alastair R. Ruddle, Member, IEEE

Abstract—This article describes a study of magnetic field expo- (EVs), which include battery powered, hybrid, and fuel cell
sure in electric vehicles (EVs). The magnetic field inside eight dif- variants.
ferent EVs (including battery, hybrid, plug-in hybrid, and fuel cell The International Commission on Non-ionizing Radiation
types) with different motor technologies (brushed direct current,
permanent magnet synchronous, and induction) were measured Protection (ICNIRP) has recommended limits for exposure to
at frequencies up to 10 MHz. Three vehicles with conventional static [1] and time-varying [2] EMFs. These limits aim to pro-
powertrains were also investigated for comparison. The measure- vide protection from well-established acute physiological ef-
ment protocol and the results of the measurement campaign are fects of EMF exposure, which include electro-stimulation of
described, and various magnetic field sources are identified. As the nerves (relevant for frequencies from 1 Hz to 10 MHz) and heat-
measurements show a complex broadband frequency spectrum, an
exposure calculation was performed using the ICNIRP “weighted ing of body tissues (for frequencies from 100 kHz to 300 GHz).
peak” approach. Results for the measured EVs showed that the The recommendations relating to electro-stimulation effects,
exposure reached 20% of the ICNIRP 2010 reference levels for which are the primary interest in this study, were revised by
general public exposure near to the battery and in the vicinity of ICNIRP in 2010 [3].
the feet during vehicle start-up, but was less than 2% at head height The exposure limits are actually specified in terms of in-body
for the front passenger position. Maximum exposures of the order
of 10% of the ICNIRP 2010 reference levels were obtained for the quantities that are not easy to determine. Consequently, ICNIRP
cars with conventional powertrains. has derived more readily measureable field reference levels from
the exposure limits. It is considered that if the exposure envi-
Index Terms—Electric vehicle, human exposure, hybrid vehicle,
magnetic field. ronment complies with the field reference levels then it can be
assumed that the exposure limits will not be breached. Exceed-
ing the field reference levels does not necessarily mean that the
I. INTRODUCTION exposure limits are also breached, but it is deemed that more de-
UBLIC expectations to move toward the electrification of tailed investigation is required in order to establish compliance
P road transport are driven by a multitude of factors and con-
cerns, which include climate change, primary energy depen-
with the exposure limits.
In situations of simultaneous exposure to fields of differ-
dence, and public health. On the other hand, there is widespread ent frequencies, these exposures are considered to be additive
public concern regarding the possible adverse effects of elec- in their effects [2], [3]. Thus, the more frequencies that are
tromagnetic fields (EMF), particularly low-frequency magnetic present, the lower the levels that can be tolerated for any of
fields. The occupants of vehicles with electric powertrains will them relative to the field reference levels. Furthermore, the in-
be exposed to low-frequency magnetic fields arising from cur- fluence of other fields that may be present in the environment
rents flowing in the high-voltage power network, traction bat- also impact on what can be tolerated from equipment generating
teries, and associated devices such as inverters and electrical fields that people may be exposed to. This is very different to
machines. Thus, there is a need to properly assess the level the evaluation of electromagnetic emissions against equipment
of magnetic field exposure that may result in electric vehicles EMC requirements, where compliance with the limits at each
frequency is considered independently of all other frequencies,
Manuscript received September 2, 2013; revised April 14, 2014; accepted and independently of the other equipment that may be present
August 29, 2014. This work was supported in part by the European Commu- in the intended operating environment.
nity’s Framework Programme (FP7/2007-2013) under grant agreement number
265772.
The field reference levels vary with respect to frequency and
A. Vassilev is with the CEA LETI, Grenoble 38054, France (e-mail: andrea. are valid only for pure sinusoidal signals. In the case of nonsinu-
vassilev@cea.fr). soidal exposures, ICNIRP has proposed two exposure criteria
A. Ferber is with the SINTEF, Oslo 0314, Norway (e-mail: alain.ferber@
sintef.no).
that should remain below 100% to avoid undesirable electro-
C. Wehrmann and M. Schilling are with the Technische Universität stimulation effects. The first one (which also applies in the case
Braunschweig, Braunschweig 38106, Germany (e-mail: c.wehrmann@tu-bs.de; of separate sinusoidal sources at multiple frequencies) consists
m.schilling@tu-bs.de).
O. Pinaud is with the G2ELAB, Grenoble 38402, France (e-mail: olivier.
in adding the ratios of the different spectral component (SC)
Pinaud@g2elab.grenoble-inp.fr). magnitudes of the field in the environment to the field refer-
A. R. Ruddle is with the MIRA Limited, Nuneaton CV10 0TU, U.K. (e-mail: ence levels. A similar approach is also described in the IEEE
alastair.ruddle@mira.co.uk).
Color versions of one or more of the figures in this paper are available online
standards relating to human exposure [4], [5], although in these
at http://ieeexplore.ieee.org. documents the frequency range for the evaluation of electro-
Digital Object Identifier 10.1109/TEMC.2014.2359687 stimulation threats is up to 5 MHz.

0018-9375 © 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.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY

The main drawback of this criterion for nonsinusoidal signals TABLE I


SUMMARY OF MEASURED VEHICLES
is that it assumes that all frequency components add in phase.
This assumption is acceptable for signals with a limited number
Car code Powertrain type Electric motor technology Power (kW) Energy (kWh)
of SC, but for broadband signals this procedure could lead to
an unnecessarily conservative exposure assessment. In [6] it is EV#1 Battery Brushed dc 11 10
shown that for a measured field spectrum with SC that are all EV#2 Plug-in hybrid Permanent magnet 30 5
EV#3 Small hybrid Induction 10 0.7
less than 1% of the ICNIRP 1998 [2] reference levels for the EV#4 Battery Permanent magnet 10 14
general public, the first criterion would suggest that the exposure EV#5 Battery 3-phase asynchronous 34 24.5
is 634%. Relative to the ICNIRP 2010 reference levels the SC EV#6 Battery Permanent magnet 40 24
EV#7 Fuel cell Permanent magnet 100 1.4
are all less than 0.1% of the general public recommendations, EV#8 Battery Permanent magnet 35 16
but the first criterion would suggest that the exposure reaches CV#1 Gasoline – 75 –
99% of the general public recommendations. CV#2 Gasoline – 66 –
CV#3 Diesel – 125 –
For these reasons a second criterion is defined by ICNIRP
(the “weighted peak” approach [3]) in which a time-varying
exposure measure is derived from an inverse Fourier transform
that take into account the phase of each SC. This approach is
more general and can be applied for every kind of magnetic field weighted peak approach). Section II outlines the measurement
waveform. Using this approach, [6] indicates that the maximum protocol and Section III summarizes the main sources of mag-
value of the time-varying exposure measure is less than 20% of netic field in vehicles. The exposure calculations are presented in
the ICNIRP 1998 reference levels for the general public, and Section IV, and Section V summarizes the conclusions.
less than 5% of the ICNIRP 2010 field reference levels.
Published data concerning field exposure in EVs is somewhat
scarce, and not always appropriate for assessment against the II. MEASUREMENT PROTOCOL
ICNIRP field reference levels. For example, the research con-
ducted in [7] characterized the magnetic fields associated with A. Measurement Setup
various form of transportation. The results showed that EVs have 1) Sensors: Four kinds of sensors were used:
a magnetic field spectrum roughly similar to conventional ve- 1) Low-frequency (LF) 3-axis magnetic field sensors (Flux-
hicles (CVs), but no exposure assessment was calculated. This gate magnetometers):
study also found no evidence for measureable electric fields a) Bartington MAG-03 (bandwidth 0–3 kHz; range:
associated with vehicles in the band 5 Hz to 3 kHz. ±100, ±250 or ± 500 μT; noise < 10 pT/Hz);
In a more recent paper [8], it is reported that magnetic fields b) Sensys FGM 3-D/100 (0–1.2 kHz).
measured in a hybrid car are much lower than the ICNIRP 1998 2) High-frequency (HF) 3-axis magnetic field sensors:
field reference levels, but no results taking into account addi- a) Narda EHP-50D (5 Hz to 100 kHz);
tive effects were presented. Maximum fields of 15 μT (in the b) Spectran NF-5035 (1 Hz to 10 MHz).
band 0.01–5 kHz) were recorded in a hybrid bus at the seat lo- 3) Current sensor Fluke i310s (range 0–300 A, 0–20 kHz)
cated closest to power cables during acceleration, deceleration, used to establish the correlation between the magnetic
and driving at 25 km/h [9], but no information regarding the field and the current flowing in the high-voltage power
time variation of the magnetic field is provided. network.
Two hybrid cars were measured in [10], which indicates that 4) Acceleration sensor (STMicroelectronics, model
the exposure could reach 80% of the ICNIRP 1998 reference LIS3LV02DL, range ±6 g), used to establish the
levels for the general public in the vicinity of the passenger’s feet relationship between acceleration and power flow.
during braking and acceleration. In [11] the second criterion was 2) Acquisition system: The LF sensors and the current sen-
employed to evaluate measurements at 12 points that represent sor were connected to an OROS analyzer (model name OR36)
various body parts from head to foot for front and rear occupant whose main functions were to:
location in five cars with motor powers ranging from 15 to 1) low-pass filter the signals (cut-off frequency 2 kHz);
147 kW. Time-domain measurements were carried out over 1 s 2) sample the signals (sampling frequency 5.12 kHz);
periods under constant drive conditions for frequencies from 3) provide concurrent recording of the signals.
1 Hz to 100 kHz. Exposure relative to the ICNIRP 1998 general The HF sensors had their own sampling and recording system.
public reference levels was reported to reach a maximum value 3) Mannequin and Sensor Positions: In order to mount the
of around 15% in the vicinity of the front occupants’ feet. These different magnetic sensors and to ensure reproducible position-
authors also report maximum exposure measures approaching ing, a nonmagnetic mannequin was developed (see Fig. 1), in-
20% of ICNIRP 1998 general public levels at head height in a spired by [7] and [13]. As tests were made while driving the cars,
hybrid bus [12]. the mannequin was installed on the front passenger seat. The
This paper describes the measurement of eight different EVs mannequin was equipped with three LF and three HF sensors
and three CVs, with characteristics as outlined in Table I, as located near the head, seat, and foot regions. A fourth LF mag-
well as the calculation of realistic exposure measures from netic field sensor was placed above the main battery in the trunk.
these measurements using the second criterion (i.e., the ICNIRP The current sensor was placed on one of the battery cables.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

VASSILEV et al.: MAGNETIC FIELD EXPOSURE ASSESSMENT IN ELECTRIC VEHICLES 3

Fig. 1. In-vehicle measurement setup.

Fig. 3. Spectrogram showing evidence of two 50 Hz high-voltage power lines.

lem with these tests is that there are several external magnetic
perturbations that could influence the results. The approaches
used for identifying these perturbations, and then performing
the tests, are described later.
1) Identifying external magnetic perturbations: There are
several sources of external magnetic perturbation in the en-
vironment, such as stationary or moving ferromagnetic masses
(e.g., manhole covers, railway lines, other cars), as well as 50 Hz
power distribution equipment (high-voltage transmission lines,
power transformers etc.).
These initial tests were carried out within a specific area with
restricted access (for example the CEA complex in Grenoble)
because the number of other cars was then limited. A specific
Fig. 2. Magnetic flux density spectrum due to steering pump of EV#1. driving route was defined in this controlled area and the car
repeated the journey several times under different driving con-
ditions (low and high speed, acceleration and deceleration). If
B. Laboratory Tests the results contained a magnetic field feature that occurred ev-
For some of the EVs, laboratory tests were carried out in ery time at the same place, it was concluded that this field was
order to characterize the magnetic field emissions of all of the due to an external perturbation. For example, a spectrogram ob-
on-board electrical equipment. The car was raised on a lift in tained from a sensor located on the floor of EV#1 is illustrated
order to be able to place sensors below the car; everything was in Fig. 3, which shows a clear permanent 50 Hz signal with two
switched off except for the item under test. The magnetic field strong features located at around 30 and 100 s (on the vertical
was then measured and this process repeated for every other time axis). These features are probably due to two high-voltage
item of equipment. This procedure is quite time consuming but 50 Hz power lines passing under the driving route.
very informative. For example, it revealed that a very specific Once the external magnetic perturbations were identified, fur-
harmonic spectrum is due to the steering pump of EV#1. This is ther tests were performed on a normal road.
illustrated in Fig. 2, which shows the root sum square (RSS) of 2) On Road Tests: On road tests were performed which in-
the Fourier transforms obtained from the waveforms recorded volved driving with maximum acceleration and deceleration in
for each of the three orthogonal components of the magnetic order to ensure maximum positive (traction) and negative (re-
field. generative brake) currents. A straight road is better because in
this case the magnetic fields due to the earth and due to the in-
duced magnetization of the car are constant during the test. For
C. Outside Tests
the first few cars, measurements were also carried out on a steep
Outside tests are the most important because they are the most slope and at high speed on a highway. However, it appeared
representative of real-world driving conditions. The main prob- that these two driving conditions did not generate higher fields
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY

Fig. 4. Correlation between current and vertical magnetic flux density mea-
sured above rear pack battery on EV#1. Fig. 5. Spectrogram (0.4–20 Hz) from a sensor close to a wheel of EV#1.

than for acceleration and deceleration on the flat road. For the ICNIRP general public limits for frequencies below 1 Hz [2].
remaining cars, therefore, the on road tests were performed only Nonetheless, traction current transients with SC above 1 Hz may
on a flat road and at moderate speeds (0–60 km/h). contribute significantly to the in-vehicle magnetic field exposure
[6].
III. IDENTIFICATION OF MAIN SOURCES OF MAGNETIC FIELD
In this section, the main sources of magnetic field are listed, B. Wheels
grouped by frequency content and presented in order of increas- The permanent magnetization of steel belted tires is a well-
ing frequency. known source of in-vehicle magnetic fields (see [14], [15]). Our
measurements show that this phenomenon is responsible for a
A. Traction Currents magnetic field inside the car of up to 2 μT at the wheel frequency
It is well known that a current flowing through a wire or a fw (which ranges from 0 to 20 Hz for speeds ranging from 0
loop generates magnetic field. Our measurements show that, to 130 km/h). There are also several harmonics present at lower
for most of the cars, the high-voltage power network acts as a intensities. The spectrogram shown in Fig. 5 (in dBμT) was
current loop. Moreover, the magnetic field close to the battery obtained from a sensor placed near the passenger’s foot location
could be important: the ratio between the induced magnetic in EV#1, close to the front right-hand wheel. The electric motor
field and the traction current was found to be in the range 0.2– was switched off and the car was manually pushed at low speed
1 μT/A, depending on the car. Therefore, if the traction current (4–5 km/h) corresponding to fw 1 Hz. This clearly shows a
has variations up to ±300 A, the magnetic field could also have fundamental signal 1μT at fw as well as the higher harmonics
variations of up to ±300 μT. An example is shown in Fig. 4, f2 , f3 , f4 , f8 , and f16 with decreasing magnitudes. As the
where the vertical component of the magnetic field at each point ground was not flat, it was difficult to maintain a constant speed,
in time is plotted versus the corresponding current. In this case, which is why the fundamental frequency fluctuates over time.
the traction current varies from −100 A to + 200 A whereas
the magnetic flux density is in the range −40 μT to +100 μT; C. Internal Combustion Engine
hence the field–current ratio is of the order of 0.5 μT/A.
Measurements on the hybrid car EV#2 show that there is a
The magnetic field due to traction currents also presents a
correlation between the rotational frequency fm (varying up
high spatial variability. In EV#1, for example, the magnetic
to 100 Hz) of the internal combustion and the magnetic field
field in the vicinity of the passenger’s foot ranges from 30 to
signals. This could be due to the motion of the pistons. The
130 μT within distances of a few tens of cm. This is due to
magnitude of the associated magnetic signal was 150 nT at
the fact that when two cables carrying opposite currents are
frequency 2fm in this case.
close together (as is the case in the central tunnel of EV#1),
the resulting magnetic field is minimized. But when the cables
diverge (as is the case in the engine bay of EV#1), the field can D. Specific Equipment
increase significantly. Specific equipment of the car may also generate magnetic
Although this field is important, it is far below the ICNIRP fields. The magnetic field emissions from the power steering
general public limit of 40 mT for static fields [1] and below the pump (500 W, 12 V) of EV#1 have already been reported
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

VASSILEV et al.: MAGNETIC FIELD EXPOSURE ASSESSMENT IN ELECTRIC VEHICLES 5

Fig. 7. Magnetic flux density spectrum (0–100 kHz) measured from EV#3.

TABLE II
SUMMARY OF IDENTIFIED SOURCES

Fig. 6. Spectrogram (0.5–1 kHz) due to the steering pump of EV#1. Sources Maximum field Frequency range Cars involved

Traction currents 100–300 μT 0–10 kHz EV#1 to EV#8


Wheels 0.1–2 μT 0–20 Hz All EVs and CVs
(see Fig. 2). This equipment can generate fields up to 1 μT Internal combustion engine 50–150 nT 0–200 Hz EV#2, CV#1 to CV#3
inside the car at frequencies in the range 0.5–1 kHz (see Fig. 6). Steering pump 1 μT 0.5–1 kHz EV#1
Regenerative brake 1 μT 0–400 Hz EV#2
For EV#2, just after an accelerator pedal release, a wide-band Electrical coupling 0.5 μT 0–600 Hz EV#3
magnetic signal of 1μT at frequencies up to 400 Hz was ob- Inverter 20 nT 7–9 kHz EV#1 to EV#4
served on the sensor above the battery. It is believed that this Inverter 60 nT 0.2–10 MHz EV#1 to EV#4
signal may be due to the regenerative brake. On EV#3, an un-
explained coupling between an equipment working at constant
frequencies (50 and 100 Hz) and the motor (whose fundamental
frequency f1 is proportional to the wheel frequency, such that
f1 = 4fw ) was found to lead to fields of the order of 0.5 μT at
frequencies up to 600 Hz.

E. Inverter
A power inverter is an electrical device that converts direct
current (dc) to alternating current (ac). In EV, it is typically
characterized by a switching frequency of around 10 kHz. To
deal with these frequencies, it is necessary to use the HF sensors.
These measurements were carried out for only four of the cars
(EV#1–EV#4). Below 200 kHz, the field level was less than
20 nT; the maximum values observed in this frequency range
(see Fig. 7) were at
1) 7–9 kHz, probably due to the inverter;
2) 16 kHz, due to the LF sensor electronics.
Between 200 kHz and 10 MHz, the field level is a little
Fig. 8. Summary of magnetic sources and ICNIRP 2010 [3] reference levels
higher but still low (less than 60 nT). A harmonic spectrum can for magnetic flux density.
be observed, but it is difficult to establish its origin.

F. Summary this envelope is decreasing with increasing frequency. Most of


The contributions from the different sources above 1 Hz that the magnetic field sources (wheels, brake, and steering pump)
were identified are summarized in Table II and in Fig. 8. In this produce frequencies ranging between a few Hz and 1 kHz, with
figure, each source is represented by a horizontal solid (blue or magnitudes of 0.1–2 μT. At frequencies above a few kHz, the
black) colored line. The vertical position indicates the maximum magnetic field is less than 100 nT and only the inverter was
field measured, while the length indicates the frequency range. identified as a source from the vehicles.
The dotted red line connects the different sources and forms Compared to the ICNIRP 2010 reference levels defined for
a spectral envelope of the measurements. It can be seen that the general public and for pure sinusoidal signals [3], which
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY

related to the frequency dependence of BR (f ):


W F (f, fr ) = BR (fr )/BR (f ). (2)

The weighted magnetic field waveform can then be calculated


as

n

Bw e,f r (t) = W Fj 2Bj . cos (2πtfj + θj + ϕj ) (3)
j =p

where W Fj represent the peak amplitude of weighting func-


tion at frequency fj and the terms ϕj represent the corre-
sponding phase values. The parameters ϕj are specified as zero
when BR (fj ) is constant, π/2 when BR (fj ) ∝ 1/f , π when
BR (fj ) ∝ 1/f 2 , and −π/2 when BR (fj ) ∝ f (see Appendix
A of [3]).
ICNIRP recommends that the summation goes from 1 Hz
(corresponding to j = p) to 10 MHz (corresponding to j = n).
Fig. 9. Amplitude and phase of weighting function, WF. In this paper it was found that the magnetic field levels decreased
rapidly for frequencies above 1 kHz. Therefore, the exposure
sum ranges from 1 Hz to 2 kHz, and only the signals from
the LF sensors are used. The weighted signal is then simply
is plotted in Fig. 8 using dashed red line, the envelope of the compared to the reference level at frequency fr .
magnetic field emissions is roughly two decades lower than the √
general public levels. However, this does not mean that the order Bw e,f r (t) < 2BR (fr ), for t ∈ [t1 ; t2 ] . (4)
of magnitude of the exposure is 1% of the reference levels. As Dividing by the right term of the inequality, we obtain a
already noted in Section I), it is necessary to take into account time-varying exposure function c(t), independent of fr that is
the additive effects of all frequencies from all sources in order required to be <1 in order to guarantee compliance with the
to assess the exposure (see Section IV). basic restrictions:
 
f j =2 kHz 
  B 
IV. EXPOSURE CALCULATIONS c(t) =   j
. cos(2πtfj + θj + ϕj ) ≤ 1,
B
 f j =1 Hz R ,j 
A. Weighted Peak Approach for Nonsinusoidal Exposures
The weighted peak exposure criterion defined by ICNIRP in for t ∈ [t1 ; t2 ] . (5)
the case of nonsinusoidal exposures (see [3] and [16]) is appli- Electro-stimulation effects are instantaneous, with the result
cable for arbitrary magnetic field waveforms B(t). The general that ICNIRP recommend that the basic restrictions (and by im-
idea is: plication, the field reference levels) should not be time averaged.
1) defining a reference frequency fr ; For convenience, therefore, the time-varying exposure measure
2) computing from B(t) a weighted signal Bw e,f r (t) that can be summarized in terms of its maximum instantaneous value
takes into account the frequency dependence of the mag- EC (for exposure criterion):
netic field reference level BR (f );
3) comparing this weighting signal to the reference level EC = max {c(t)} , for t ∈ [t1 ; t2 ] . (6)
BR (fr ) at reference frequency.
If it is assumed that B(t) is periodic on the measured interval B. Practical Implementation
[t1 ; t2 ], then B(t) may be approximated by the n spectral com- The ICNIRP recommendations give very few details con-
ponents (SC) derived from B(t) by means of a discrete Fourier cerning the practical implementation of this calculation. For
transform. Thus, example, it is not clear how to:
1) choose t1 and t2 ;

n

B(t) = 2Bj . cos (2πtfj + θj ) (1) 2) compute the SC in case of a nonperiodic signal;
j =1 3) deal with the three components of a magnetic field;
4) select a sampling rate for digitizing the signals.
where j is the summation index of the SC (including the dc For the first point, the criterion has been calculated with dif-
component), Bj represent the RMS amplitudes of the SC at ferent bounds t1 and t2 in order to verify that the criterion does
frequency fj , and θj correspond to the phases of the SC. not change.
The weighting function WF is defined (see Fig. 9) such that For the second point, the fact that the signal may not be peri-
the gain is inversely proportional to the chosen magnetic field odic can lead to spectrum leakage, i.e., signal energy smearing
reference levels BR (f ) and the corresponding phase values are out in the frequency domain. Another consequence is that the
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

VASSILEV et al.: MAGNETIC FIELD EXPOSURE ASSESSMENT IN ELECTRIC VEHICLES 7

TABLE III
SUMMARY OF EXPOSURE RESULTS

Car code Maximum exposure Position

EV#1 14.3% Front passenger foot


EV#2 17.8% Above rear battery
EV#3 7.9% Above rear battery
EV#4 5.9% Rear passenger foot
EV#5 4% Front passenger seat
EV#6 3.2% Front passenger seat
EV#7 2.1% Front passenger seat
EV#8 2.7% Front passenger foot
CV#1 2.7% Front passenger seat
CV#2 9% Left rear passenger seat
CV#3 10% Left rear passenger seat

temporal function c(t) is perturbed near its bounds (at t1 and t2 ).


One way to deal with this problem is to use short-time Fourier
transform (STFT) instead of a simple FFT. This solution has
been successfully tested, but it is quite time consuming. A sim- Fig. 10. Example of 3-D magnetic field signal from EV#1.
pler solution is to discard a small region around each bound
(e.g., q = 2% of signal length l) when calculating the exposure
measure:

EC = max {c(t)} , for t ∈ [t1 + ql; t2 − ql] . (7)

For the third point, it is assumed that for each of the M


sampled field component the weighted field waveforms and their
associated time-varying exposure measures ci (tk ) are computed
according to (3) and (5), respectively. For each time sample tk ,
the root sum square (RSS) value then is calculated from the three
orthogonal components in order to obtain the multidimensional
exposure C(tk ), where

M

C(tk ) =  (ci (tk ))2 . (8)
i=1

Finally, the worst case instantaneous exposure for the entire


magnetic field is obtained by applying (7) to C(tk ) to obtain the
maximum value. Fig. 11. Component and combined exposure measures, based on ICNIRP
2010 magnetic field reference levels for the general public.

C. Results
For each car, simple drive cycles were investigated includ- For the CV examples, the highest values were around 10% of
ing acceleration and deceleration phases. The exposure was the ICNIRP 2010 general public reference levels, and were also
computed as outlined earlier from measurements at the four linked with vehicle start-up and braking events.
positions: head, seat, foot, and above the battery. The maxi- As an example, Fig. 10 shows the whole magnetic field wave-
mum exposure levels and locations results are summarized in form for a short journey with EV#1, as recorded by a sensor
Table III. located in in the vicinity of the passenger’s foot. It can be seen
The main results for the EV examples are as follows. that there is a sharp peak in the magnetic field waveforms on
1) The highest values, between 14% and 18% of the ICNIRP the X and Z axes when the vehicle is switched on (around time
2010 general public reference levels, appear on EV#1 and 15 s).
EV#2, when the engine is switched on. The exposure measures ci (tk ) obtained from the three orthog-
2) For every EV, the highest values were reached near the onal field component waveforms are shown in Fig. 11, along
battery and the foot of the driver or passenger. with the corresponding RSS value C(tk ). A sharp peak can also
3) The maximum exposure at head-height for the front pas- be seen at the same time (at around 15 s) and the peak exposure
senger was found to be 1.5% of the ICNIRP 2010 general reaches almost 15%. Finally, Fig. 12 shows the detail around
public reference levels. the peak of the X component of the field and the corresponding
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY

For the cars with conventional powertrains, maximum expo-


sures of the order 10% were obtained, depending on the position
of the sensor.

ACKNOWLEDGMENT
The authors are grateful for the support and contribu-
tions of other members of the EM-Safety project consor-
tium, from Centro Ricerche Fiat (Italy), Leibniz University
of Hanover (Germany), IPM (Italy), MIRA (UK), Prysmian
(Italy), TAMAG Iberica (Spain), and the University of Torino
(Italy). Further information can be found on the project website
(www.sintef.no/Projectweb/EM-Safety).

REFERENCES
[1] ICNIRP, “Guidelines on limits for exposure to static magnetic fields,”
Health Phys., vol. 96, no. 4, pp. 504–514, Apr. 2009.
[2] ICNIRP, “Guidelines for limiting exposure to time-varying electric, mag-
netic, and electromagnetic fields (up to 300 GHz),” Health Phys., vol. 74,
Fig. 12. Detail of X component of magnetic field waveform and the corre-
no. 4, pp. 494–522, Apr. 1998.
sponding exposure measure.
[3] ICNIRP, “Guidelines for limiting exposure to time-varying electric and
magnetic fields (1 Hz to 100 kHz),” Health Phys., vol. 99, no. 6,
pp. 818–836, Dec. 2010.
[4] IEEE standard for safety levels with respect to human exposure to elec-
tromagnetic fields, 0–3 kHz, IEEE Standard C95.6-2002, Oct. 23, 2002.
exposure measure cX (tk ). It can be seen that the rapidly ris- [5] IEEE standard for safety levels with respect to human exposure to electro-
ing edge in the X component of the magnetic field waveform magnetic fields, 3 kHz to 300 GHz, IEEE Standard C95.1-2005 Apr. 19,
2005.
is a major contributor (at 10%) to the overall exposure (see [6] A. R. Ruddle, L. Low, and A. Vassilev, “Evaluating low frequency mag-
Fig. 11). netic field exposure from traction current transients in electric vehicles,”
in Proc. 12th Int. Eur. Symp. Electromagn. Compat., Brugge, Belgium,
Sep. 2013, pp. 78–83.
[7] F. M. Dietrich and W. L. Jacobs, “Survey and assessment of electric
V. CONCLUSION and magnetic field public exposure in the transportation environment,”
US Department of Transportation, Federal Railroad Administration, Rep.
The magnetic field of eight different EVs (including battery PB99-130908, Mar. 1999.
powered, hybrid, and fuel cell variants) with different types of [8] M. N. Halgamuge, C. D. Abeyrathne, and P. Mendis, “Measurement and
motor technology (brushed dc, permanent magnet synchronous, analysis of electromagnetic fields from trams, trains and hybrid cars,”
Radiation Protection Dosimetry, vol. 141, no. 3, pp. 255–268, 2010.
and induction) were measured in the frequency range from dc [9] M. Pous, A. Atienza, and F. Silva, “EMI radiated characterization of an
to 10 MHz. Three vehicles with conventional powertrains were hybrid bus,” in Proc. 10th Int. Eur. Electromagn. Compat. Symp., York,
also investigated for comparison purposes. A suitable measure- U.K., Sep. 2011, pp. 208–213.
[10] E. Karabetsos, E. Kalampaliki, G. Tsanidis, D. Koutounidis, N. Skam-
ment protocol was developed, based on the use of a mannequin nakis, T. Kyritsi, A. Yalofas, “EMF measurements in hybrid technology
to ensure reproducible positioning of the field sensors. cars”, presented at 6th Int. Workshop Biol. Effects Electromagn. Fields,
The main magnetic sources of the EV were identified: at very Bodrum, Turkey, Oct. 2010.
[11] G. Schmid, R. Überbacher, and P. Göth, “ELF and LF magnetic field
low frequency (<1 Hz), magnetic fields of several hundreds exposure in hybrid- and electric cars,” in Proc. Bioelectromagnetics Conf.,
of μT can be encountered mainly due to traction currents and Davos, Switzerland, Jun. 2009, pp. 9–3.
induced magnetization effects. Most of the magnetic sources [12] R. Überbacher, G. Schmid, and P. Göth, “ELF magnetic field exposure
during an inner-city hybrid bus ride,” in Proc. Bioelectromagnetics Conf.,
(wheels, regenerative brake, steering pump, and combustion en- Davos, Switzerland, Jun. 2009, pp. P–23.
gine) are found at frequencies between a few Hz and 1 kHz, with [13] A. Brecher, D. R. Disk, D. Fugate, W. Jacobs, A. Joshi, A. Kupferman, R.
magnitudes between 0.1 and 2 μT. Above 1 kHz, the magnetic Mauri, and P. Valihura, “Electromagnetic field characteristics of the tran-
srapid TR08 maglev system,” US Department of Transportation, Federal
field is less than 100 nT and only the inverter was identified as Railroad Administration, Rep. DOT-VNTSC-FRA-02-11, May 2002.
a source from the vehicle. [14] S. Milham, J. B. Hatfield, and R. Tell, “Magnetic fields from steel-belted
As the measurements showed a complex broadband fre- radial tires: Implications for epidemiologic studies,” Bioelectromagnetics,
vol. 20, no. 7, pp. 440–445, 1999.
quency spectrum, an exposure calculation was performed ac- [15] S. Stankowski, A. Kessi, O. Bécheiraz, K. Meier-Engel, M. Meie, “Low
cording to the ICNIRP 2010 recommendations [3]. It was found frequency magnetic fields induced by car tire magnetization,” Health
that the highest exposure for the EVs that were tested was less Phys., vol. 90, no. 2, pp. 148–153, Feb. 2006.
[16] K. Jokela, “Restricting exposure to pulsed and broadband magnetic fields,”
than 20% of the field reference levels for general public expo- Health Phys., vol. 79, no. 4, pp. 373–388, 2000.
sure, and corresponded to vehicle start-up. For all of the cars in
the EV sample, the highest values were found near the battery
and in the vicinity of the feet of the driver or front passenger. Andrea Vassilev received the mechanical engineering degree, in 1996, from
Ecole Centrale de Lyon, France.
The maximum exposure at head height for the front passenger From 1996 to 2001, he was with PSA Peugeot Citroen, as a Mechanical Sim-
head was found to be less than 2% of the ICNIRP 2010 field ulation Engineer. Since 2001, he has been with CEA-LETI, Grenoble, France,
reference levels for general public exposure. a research institute for electronics and information technologies.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

VASSILEV et al.: MAGNETIC FIELD EXPOSURE ASSESSMENT IN ELECTRIC VEHICLES 9

Alain Ferber received the M. Sc. degree in Cybernetics from The Norwegian Meinhard Schilling received the diploma in physics, in 1989, and the Ph.D.
Institute of Technology, in 1976. degree (Dr. rer. nat), in 1992, both from the Universität Hamburg, Hamburg,
He is currently a Senior Scientist in the Department of Optical Measurement Germany.
Systems and Data Analysis at SINTEF, Oslo, Norway. In 1998 after habilitation he was with the University Hamburg as Privat-
dozent. Since 2001, he has been a Full Professor and Head of the Institut für
Elektrische Messtechnik und Grundlagen der Elektrotechnik, at the Technische
Universität Braunschweig.

Christof Wehrmann received the M.Sc. degree in electrical engineering, from


the Technische Universität (TU) Braunschweig, Braunschweig, Germany, in
2011. He is currently working toward the Ph.D. degree at Institut für Elektrische
Messtechnik und Grundlagen der Elektrotechnik at the TU Braunschweig. Alastair R. Ruddle (M’10) received the B.Sc. degree in physics from the Uni-
Since 2011, he has been with the Institut für Elektrische Messtechnik und versity of Bristol, Bristol, U.K. and the Ph.D. degree in electronic and electrical
Grundlagen der Elektrotechnik at the TU Braunschweig as a Research Assistant. engineering from the University of Loughborough, Loughborough, U.K.
Since 1996 he has been a computational electromagnetics specialist with
MIRA Limited, Nuneaton, U.K., an automotive research and technology orga-
nization. Prior to this he carried out research in the defense and power industries.

Olivier Pinaud received the electrical engineering degree from the Grenoble
Institute of Technology, Grenoble, France, in 2010. Since 2011, he has been
working toward the Ph.D. degree on electromagnetic computation at Grenoble
Electrical Engineering Laboratory (G2ELAB), France.

View publication stats

You might also like