batteries

Article
Fast Characterization Method for Modeling Battery
Relaxation Voltage
An Li 1,† , Serge Pelissier 2, *, Pascal Venet 3 and Philippe Gyan 4
1 SERT Metal of VESUVIUS Group, 69150 Décines, France; airblelee@gmail.com
2 Univ Lyon, IFSTTAR-AME-LTE, 69675 Bron Cedex, France
3 Univ Lyon, Lyon 1 University, Ampere laboratory UMR CNRS 5005, 69622 Villeurbanne Cedex, France;
pascal.venet@univ-lyon1.fr
4 Renault, 78280 Guyancourt, France; philippe.gyan@renault.com
* Correspondence: serge.pelissier@ifsttar.fr; Tel.: +33-472-14-24-99
† Former Ph.D. Student Funded by Renault at Ampere Laboratory, Lyon 1 University,
69622 Villeurbanne Cedex, France.

Academic Editor: Juan Carlos Álvarez Antón

Received: 29 January 2016; Accepted: 23 March 2016; Published: 6 April 2016

Abstract: After the end of a charging or discharging sequence, the battery voltage keeps evolving
towards a finite value, during hours or even days, although no current is exchanged with the
battery. This corresponds to the battery relaxation. In the context of electric vehicles (EV), a good
measurement of the voltage at rest allows an accurate estimation of the battery state of charge (SoC).
The characterization of the battery voltage at different levels of SoC after the full relaxation would be
very time consuming. In this paper, a fast method to extrapolate long relaxation voltage is proposed.
It needs only one complete measurement of relaxation at one given SoC and could give accurate
voltage estimation at other states of charge from short and partial measurement. This generic method
was validated on three different cells and could be easily extended to any type of battery.

Keywords: batteries; relaxation; characterization; modeling

1. Introduction
The battery is a challenging energy source for commercial electric vehicles (EV) in today’s
automobile market. As there is not as much on-board energy in EVs as in internal combustion engine
vehicles, estimation of the remaining driving range is a critical issue for EVs. The remaining energy of
the battery pack is related to its state of charge (SoC), which can be estimated by various techniques
used alone or in combination: coulomb-metric measurement, identification by observers such as
Kalman filter, and comparison between the open circuit voltages (OCV) deduced from measured
voltage and its OCV-SoC curve. In the context of determining the SoC from estimation of the OCV,
precise determination of different OCVs for different SoCs are essential to estimate accurately the SoC
of the battery at any given time. However, EVs are frequently parked for long periods (e.g., a day,
a night, or a weekend), which results in a long relaxation condition for the battery. Phenomena inside
the battery, such as polarization and mass transfer, have very slow dynamics and occur during hours of
relaxation phase, causing voltage variations. For example, Figure 1 shows the typical voltage response
of a battery following a discharge current pulse. The relaxation corresponds to the phase after a period
of discharge, during which there is no current and the battery voltage tends towards a steady state.
The voltage measured at the end of the relaxation can be considered as the OCV value at this SoC.
Figure 1 also shows two other phases of the battery voltage. One is the phase of voltage drop, which
occurs when the current level changes. Another one is the phase of forced regime during which the
battery current is not zero.

Batteries 2016, 2, 7; doi:10.3390/batteries2020007 www.mdpi.com/journal/batteries

Batteries 2016, 2, 7 2 of 15
Batteries 2016, 2, 7 2 of 15

which occurs when the current level changes. Another one is the phase of forced regime during
which occurs
Batteries 2016, 2, 7 when the current level changes. Another one is the phase of forced regime during
2 of 15
which the battery current is not zero.
which the battery current is not zero.
Discharge
Courant current
en décharge
3 Discharge
Courant current
en décharge
3 I≠0
I≠0

(A)
(A)
2

(A)
(A)
Courant
2

Current
Courant
Current
1
1
I=0
0 I=0
00 200 400 600 800 1000 1200 1400 1600
0 200 400 600 Temps
Time (s)
800 (s) 1000 1200 1400 1600

2 Voltage drop Temps

Time
Battery
Tension (s)
la(s)
de voltage
batterie
2 Voltage drop Battery
Tension de voltage
la batterie
3.34
(V)
(V)
(V) 3.34
3.32
Relaxation 1
(V)
Voltage

1
Tension
3.32
Relaxation
Voltage
Tension

3.3
3.3 Urelax_init
3.28 Urelax_init 4
3.28 Voltage drop 4
800Voltage
1000 drop
0
00 200 400 600 1200 1400 1600
3 200
Forced regime 400 0 600 Temps (s) 1000
800 (s)
Time 1200 1400 1600
3 Forced regime Temps
Time (s)(s)

Figure 1. Voltage response of a battery

battery afterafter discharge.
discharge.
Figure 1. Voltage response of a battery after discharge.

Relaxation time has a significant impact on the measured voltage, which is considered to be the
Relaxation time has a significant impact on the measured voltage, which is considered to be the
OCV
OCV at at the
the end
end of
of relaxation.
relaxation. Figure
Figure 2 shows
shows thethe measurement
measurement results
results of
of OCV
OCV atat different
different SoCs
SoCs and
and
Figure 22 shows the measurement
after different relaxation
after different relaxation times
relaxation times of a Li-ion battery. The curves measured after discharge
times of a Li-ion battery. The curves measured after discharge and charge and charge
after
are different. The
The presence of gaps between the two curves is called called hysteresis.
hysteresis. When the relaxation
are different. The presence of gaps between the two curves is When the relaxation
time increases, the value of relaxation
time increases, the value of relaxation voltage measured in discharge increases and
relaxation voltage measured in discharge increases and the
the one
one in
in charge
charge
decreases. The longer
longer the
the relaxation
relaxation time, the more the measurement of the battery voltage is close to
decreases. The longer the relaxation time, the more the measurement of the battery voltage is close to
the real OCV. In fact, the battery voltage still
the real OCV. In fact, the battery voltage still evolves
evolves beyond
beyond aa few
few hours.
hours. In
In [1],
[1], it
it was
was stated
stated that
that it
it
may take over 24 h before the battery voltage stabilizes.
may take over 24 h before the battery voltage stabilizes.
3.35
3.35
After
After
charge
3.33
3.33 charge
(V)

3.31
(V)

3.31
OCV
OCV

3.29 0.5 h
3.29 After 0.5 h
After 1h
discharge 1h
3.27 discharge 3h
3.27 3h
12 h
12 h
3.25
3.25 30 40 50 60 70 80
30 40 50 60
SoC (%) 70 80
SoC (%)
Figure 2. Open circuit voltage (OCV)-state of charge (SoC) curves of a Li-ion battery LiFePO4
Figure 2. Open
Open circuit
circuit voltage (OCV)-state of
of charge
charge (SoC)
(SoC) curves
curves of
of aa Li-ion
Li-ion battery
battery LiFePO
LiFePO4
Figure 2.
technology obtained aftervoltage (OCV)-state
different relaxation times. 4
technology obtained after different relaxation times.
technology obtained after different relaxation times.
If the voltage variation during the relaxation is not taken into account in OCV estimation, error
If the voltage variation during the relaxation is not taken into account in OCV estimation, error
If the
will be voltage variation
introduced during theofrelaxation
into the estimation SoC fromisthe notvoltage.
taken into
For account
example,inas OCV estimation,
shown in Figureerror
3, if
will be introduced into the estimation of SoC from the voltage. For example, as shown in Figure 3, if
will be introduced
the OCV-SoC curveinto the estimation
approximated from of the
SoCvoltage
from the voltage.after
measured For example,
1 h is usedasas
shown in Figurefor
the reference 3,
the OCV-SoC curve approximated from the voltage measured after 1 h is used as the reference for
if the OCV-SoC curve approximated from the voltage measured after
estimating the SoC from the voltage measurement, the voltage increase during relaxation after1 h is used as the reference
estimating the SoC from the voltage measurement, the voltage increase during relaxation after
for estimating
discharge the SoC
between from24the
1 h and voltage
h will measurement,
introduce the voltage
an overestimation increase
of the SoC. during relaxation after
discharge between 1 h and 24 h will introduce an overestimation of the SoC.
discharge between 1 h andof
The characterization 24the
h will introduce
voltage change an during
overestimation of the SoC.
long relaxation (e.g., 24 h) requires a long
The characterization of the voltage change during long relaxation (e.g., 24 h) requires a long
Thetime.
testing characterization of theisvoltage
As the relaxation dependent change on during
the SoC, long
therelaxation (e.g., 24
measurements h) requires
must a long
be repeated at
testing time. As the relaxation is dependent on the SoC, the measurements must be repeated at
testing
differenttime. As Yet,
SoCs. the relaxation is dependent
such extensive on the SoC,
testing would put athe measurements
heavy must planning,
load on testing be repeated at different
and increase
different SoCs. Yet, such extensive testing would put a heavy load on testing planning, and increase
SoCs. Yet, such
the testing extensive
costs. testinginteresting
It is therefore would putto a heavy
reduceload on testing
or adapt planning, andtime
the measurement increase the testing
of relaxation to
the testing costs. It is therefore interesting to reduce or adapt the measurement time of relaxation to
costs. It is therefore
the necessary interesting
desired precision.to However,
reduce or adapt with athe measurement
reduced time of time,
measurement relaxation to the change
long-term necessary
in
the necessary desired precision. However, with a reduced measurement time, long-term change in
desired
voltage precision.
cannot beHowever,
observed.withThe aproblem
reduced to measurement time, long-term
solve is to reduce change intime
the measurement voltage cannot be
of relaxation
voltage cannot be observed. The problem to solve is to reduce the measurement time of relaxation
observed. The problem
while maintaining to solve isoftothe
the accuracy reduce
voltagethefinal
measurement
value. time of relaxation while maintaining the
while maintaining the accuracy of the voltage final value.
accuracy of the voltage final value.
Batteries 2016, 2, 7 3 of 15
Batteries 2016, 2, 7 3 of 15

Voltage
Batteries 2016, 2, 7 increase during 3 of 15
OCV relaxation between 1 h and OCV measured 24 h after discharge
OCV
24 hVoltage x to ∇) during
(from increase
relaxation between 1 h and OCV measured 24 h after discharge
24 h (from x to ∇)
∆
x OCV measured 1 h after discharge
∆
x OCV measured 1 h after discharge
Over-estimation of the SoC
Over-estimation of the SoC

SoC
SoC
Figure
Figure 3.
3. Illustration
Illustration of
of the
the overestimation
overestimation of
of SoC
SoC caused
caused by
by long
long relaxation.
relaxation.
Figure 3. Illustration of the overestimation of SoC caused by long relaxation.

In the literature, the problem of reducing the measurement time of relaxation has seldom been
In theIn literature,
the literature, thethe problemofofreducing
problem reducing the measurement
measurement time
time of of
relaxation
relaxationhas has
seldom
seldombeen been
addressed. In [2],
addressed. a method
In [2], a method of of
rapid
rapidtesting
testingwaswas proposed
proposed forgetting
getting thethe OCV-SoC curve. However,
addressed. In [2], a method of rapid testing was proposedfor for getting OCV-SoC
the OCV-SoC curve. However,
curve. However,
slow slow
variation of the
variation of voltage
the voltage during
during thethe relaxation
relaxation phaseisisnot
phase notfully
fully taken
taken into
into consideration
consideration in in this
this
slow variation of the voltage during the relaxation phase is not fully taken into consideration in this
method. Several
method. extrapolation
Several extrapolation methods
methodsofofthe the battery relaxationvoltage
battery relaxation voltage areare proposed
proposed in two
in two patentspatents
method. Several extrapolation methods of the battery relaxation voltage are proposed in two patents
and one
and paper:
one paper:Patents
Patents[3,4] used
[3,4] usedanananalytical
analytical relaxation
relaxation law lawtotoestimate
estimate thethe
OCV OCVand and
SoC SoCof theof the
and one paper: Patents [3,4] used an analytical relaxation law to estimate the OCV and SoC of the
battery
battery when when
the the battery
battery is is notininsteady
not steady state.
state. Reference
Reference[5] [5]used
usedanan original
originalequivalent circuit
equivalent circuit
battery when
model of the batterywith
relaxation is not in steadyvarying
parameters state. Reference
over time. [5] used analloriginal
However, these equivalent
proposed circuitdomodel
methods
model of relaxation with parameters varying over time. However, all these proposed methods do
of relaxation with parameters
not conventional
use conventional varying
electrical circuitsover time.
and may
may be However, all for these proposed methods do not In use
not use electrical circuits and be problematic
problematic for embedded
embedded implementation.
implementation. In
conventional
[6], a fast electrical
extrapolation circuits and ismay
method be problematic
proposed to estimatefor embedded
OCV-SoC implementation.
by using In [6], a fast
6 min of relaxation;
[6], a fast extrapolation method is proposed to estimate OCV-SoC by using 6 min of relaxation;
extrapolation
however, the method
error of is estimation
proposedistostill estimate
4 mV apartOCV-SoC
from the byOCVusing 6 minfrom
obtained of relaxation; however,
5 h of relaxation. In the
however, the error of estimation is still 4 mV apart from the OCV obtained from 5 h of relaxation. In
error[7], a similar method
of estimation is stillwas
4 mVusedapart
with from
10 min of OCV
the relaxation with an
obtained fromerror5 hofofthe order of 5 mV.
relaxation. Asawe
In [7], similar
[7], acould
similarsee method
further inwas
this used
paper, with
an 10 min
error of 2 of relaxation
mV in OCV-SoC with
curveancould
errorcause
of thean order of 5 error
important mV. in As we
method was used with 10 min of relaxation with an error of the order of 5 mV. As we could see further
couldSoCseeestimation.
further in this paper, an error of 2 mV in OCV-SoC curve could cause an important error in
in this paper, an error of 2 mV in OCV-SoC curve could cause an important error in SoC estimation.
SoC estimation.
In this paper, a new method to characterize long relaxation voltage is proposed with an error of
In this paper, a new method to characterize long relaxation voltage is proposed with an error of
In this1paper,
about mV. This a newmethodmethoduses to
thecharacterize
extrapolationlong relaxation voltage
of conventional is proposed
electric circuit response with
andan error of
takes
about 1 mV. This method uses the extrapolation of conventional electric circuit response and takes into
into consideration the slow voltage variation during relaxation.
about 1 mV. This method uses the extrapolation of conventional electric circuit response and takes The principle of the method,
consideration the slow voltage variation during relaxation. The principle of
h)the
for method, illustrated
into illustrated
consideration in Figurethe 4, comprises:
slow voltage(i) variation
a full relaxation
during measurement
relaxation.(e.g., The24principle a single
of theSoC to
method,
in Figure 4, comprises:
calibrate the method; (i) a several
(ii) full relaxation
shorter measurement
relaxations (e.g., (e.g.,
1 h, 5 24
h) h) for a single
measured at SoC considered
various to calibrate the
illustrated in Figure 4, comprises: (i) a full relaxation measurement (e.g., 24 h) for a single SoC to
method;
SoCs;(ii)and
several
(iii) shorter relaxationsto(e.g.,
an extrapolation 1 h, 5 h)
a longer measured
duration (the atsame
variousas in considered
i) for eachSoCs; and (iii) an
previous
calibrate the method; (ii) several shorter relaxations (e.g., 1 h, 5 h) measured at various considered
extrapolation to a longer duration (the same as in i) for each previous measurement.
measurement.
SoCs; and (iii) an extrapolation to a longer duration (the same as in i) for each previous
measurement. Extrapolation of the
relaxation (e.g., to 24 h)
Slower behavior of
Battery voltage

Extrapolation of the the battery

relaxation (e.g., to 24 h)
Slower behavior of
Battery voltage

the battery
≈

Short measurement of the

≈

relaxation (e.g., 1 h, 5 h, …)
Battery model

Short measurement of the

relaxation (e.g., 1 h, 5 h, …)
Battery model

Time

Figure 4. Principle of the proposed method.

Figure 4. Principle of the proposed method.

Time of relaxation by
This paper is organized as follows. Section 2 gives a mathematical description
This paper
using is organized
an electrical model. as follows.
Section Section a2 gives
3 describes study aofmathematical description
the long relaxation, which of relaxation
reveals the by
Figure 4. Principle of the proposed method.
using an electrical model. Section 3 describes a study of the long relaxation, which reveals the possibility
This paper is organized as follows. Section 2 gives a mathematical description of relaxation by
using an electrical model. Section 3 describes a study of the long relaxation, which reveals the
Batteries 2016, 2, 7 4 of 15

of extrapolating the slow voltage behavior of the relaxation from a short and partial measurement.
Batteries 2016, 2, 7 4 of 15
Section 4 gives details on the proposed extrapolation method. The validation of the proposed method
through both simulations
possibility and experiments
of extrapolating is discussed
the slow voltage in the
behavior of Section 5; Section
relaxation from a 6short
concludes the paper.
and partial
measurement. Section 4 gives details on the proposed extrapolation method. The validation of the
2. proposedDescription
Mathematical method through
of both simulations and experiments is discussed in Section 5; Section 6
Relaxation
concludes the paper.
Several types of models can be used to represent the temporal behavior of the battery
2. Mathematical Description of Relaxation
(e.g., electrochemical models [8], equivalent circuit model [9,10], black box model [11]. In this paper,
an equivalent Several
circuittypes
modelof models can be
is chosen to used
study to represent the temporal
the relaxation behavior
because it is of the battery
simple, fast,(e.g.,
robust, with
electrochemical models [8], equivalent circuit model [9,10], black box model [11]. In this paper, an
a good compromise of accuracy/simplicity, suitable for the application of electrified vehicles, and is
equivalent circuit model is chosen to study the relaxation because it is simple, fast, robust, with a
easily integrated into simulation
good compromise software. The
of accuracy/simplicity, model
suitable for used to represent
the application the dynamic
of electrified vehicles, behavior
and is of the
battery is shown in Figure
easily integrated 5.simulation
into This model is composed
software. The modelof: (1)toarepresent
used voltagethe source
dynamicto represent
behavior of the
the OCV to a
given SoC;battery is shown in Figure
(2) a resistance Rs to 5.represent
This modelthe
is composed
voltage of: (1) a and
drop; voltage
(3)source
severalto represent
Ri //Ci the OCV toin series to
circuits
a given SoC; (2) a resistance Rs to represent the voltage drop; and (3) several Ri//Ci circuits in series to
represent the forced regime and relaxation. All these parameters of the model are mainly dependent
represent the forced regime and relaxation. All these parameters of the model are mainly dependent
on the SoC,onthe
the temperature, the current
SoC, the temperature, (amplitude
the current (amplitude and direction),
and direction), andand the aging.
the aging.

3
2 C1 Cn
Rs
i(t) i(t)

us(t) R1 Rn
ubatt(t) +
u1(t) un(t) ubatt(t)
OCV
1

Figure 5. Thevenin model.

Figure 5. Thevenin model.
During the relaxation phase, as the current is zero, the voltage drop across the resistance Rs is
zero.
During the However, voltage
relaxation in individual
phase, as theRcurrent
i//Ci circuits
is is still dependent
zero, on time.
the voltage dropByacross
denoting Ui_Relax
the the
resistance Rs is
voltage drop on Ri//Ci at the beginning of relaxation, the voltage of the battery during relaxation can
zero. However, voltage in individual Ri //Ci circuits is still dependent on time. By denoting Ui_Relax
be expressed by:
the voltage drop on Ri //Ci at the beginning of relaxation, the voltage of the battery during relaxation
n  −
t - t0 
can be expressed by:  τ i_Relax 
ubatt (t ) = U Relax_init + U
i =1
i_Relax ⋅ 1 − e
¨


(1)

t ´ t0 
˛
n ´
´ e τi_Relax
ÿ
where URelax_init is u
the initial
ptq “battery
U voltage`at theU
beginning¨ ˚
of 1relaxation (see (4)‹in Figure 1) which (1)
batt Relax_init ˝ i_Relax ‚
can be measured directly; t0 is the moment when the relaxation starts; and τi_Relax is the time constant
i “1
of Ri//Ci circuit during relaxation with τ1_Relax < τ2_Relax < ··· < τn_Relax.
The OCV expression is given by:
where URelax_init is the initial battery voltage at the beginning n
of relaxation (see (4) in Figure 1) which
can be measured directly; t0 is the moment OCVwhen
= U Relax_init 
the +relaxation
i =1
U i_Relax starts; and τi_Relax is the time
(2) constant
of Ri //Ci circuit during relaxation with τ1_Relax < τ2_Relax < ¨ ¨ ¨ < τn _Relax .
The OCVFinally, the unknown
expression parameters
is given by: to be identified for a relaxation are the number of Ri//Ci (n),
the voltage drops on each circuit Ri//Ci at the beginning of relaxation (Ui_Relax), and the time constant
of each circuit Ri//Ci during relaxation (τi_Relax). ÿn
In this work, the parametersOCVof the
“model depend`only on
URelax_init the SoC. All experiments presented in
Ui_Relax (2)
this paper were realized at an ambient temperature of 25i“°C 1 and with new cells.
The structure of the model in Figure 5 is valid for both forced regime and relaxation phase.
However,
Finally, the valuesparameters
the unknown of parameters to
arebe
different in eachfor
identified phase. The work ofare
a relaxation thisthe
paper focuses only
number
(n), of Ri //Ci
on the relaxation phase.
the voltage drops on each circuit Ri //Ci at the beginning of relaxation (Ui_Relax ), and the time constant
of each circuit Ri //Ci during relaxation (τi_Relax ).
In this work, the parameters of the model depend only on the SoC. All experiments presented in
this paper were realized at an ambient temperature of 25 ˝ C and with new cells.
The structure of the model in Figure 5 is valid for both forced regime and relaxation phase.
However, the values of parameters are different in each phase. The work of this paper focuses only on
the relaxation phase.
Batteries 2016, 2, 7 5 of 15

Batteries 2016, 2, 7 5 of 15
3. Study of a Long Relaxation
3. Study of a Long Relaxation
3.1. Test of a Long Relaxation
3.1. Test of a Long Relaxation
A relaxation experiment was carried out on an A123 26650 LiFePO4 cell whose full capacity,
A relaxation experiment was carried out on an A123 26650 LiFePO4 cell whose full capacity,
nominal voltage, and cutoff voltage are 2.3 Ah, 3.3 V, and 2 V, respectively [12]. The cell was previously
nominal voltage, and cutoff voltage are 2.3 Ah, 3.3 V, and 2 V, respectively [12]. The cell was
charged to 100%
previously SoC through
charged to 100% theSoCmethod
through oftheconstant
method of current/constant voltage, voltage,
constant current/constant and wasanddischarged
was
to 55% SoC by to
discharged a constant
55% SoC by current (1C).current
a constant After 1 h ofAfter
(1C). rest,1 ah current
of rest, aprofile
current was applied
profile to discharge
was applied to
the cell from 55%
discharge SoC
the cell to 45%
from SoC.toThen,
55% SoC a rest
45% SoC. period
Then, a restof 24 h of
period was24 imposed to measure
h was imposed the the
to measure voltage
voltage
evolution evolution
during during relaxation,
relaxation, as shownasin shown
Figurein Figure
6. The6.cellThewas
cell was placed
placed in ainthermal
a thermalchamber
chamber with
ambient temperature regulated to 25 C during all experiments. The minimum sampling timetime
with ambient temperature regulated
˝ to 25 °C during all experiments. The minimum sampling used for
used for
relaxation relaxation
is 0.1 s. is 0.1 s.

Courant (Idch>0,Ich<0)
Discharge current
20
I (A) (A)

15
Current

10

0
0 5 10 15 20 25
Time (h)(h)
Temps
Tension de la batterie
Battery voltageA123 2.3Ah
3.6
(V)

3.4
Ubatt (V)
Voltage

3.2

3 Long relaxation of 24 h
2.8
0 5 10 15 20 25
Time (h)
Temps (h)
Courant (Idch>0,Ich<0)
Discharge current
20
I (A) (A)

15
TensionBattery
de la batterie
voltageA123 2.3Ah
Current

3.305
10
(V)
Ubatt (V)

5
3.3 x
Voltage

0
0 0.01 0.02 0.03 0.04 0.05 0.06 x
Time
Temps(h)(h)
A TensionBattery
de la batterie
voltageA123 2.3Ah
3.295
B C
(V)

3.3 x 3.29
Ubatt (V)

1 5 10 15 20 24
3.2
Voltage

3.1
Temps
Time (h)(h)
3
2.9
0 0.01 0.02 0.03 0.04 0.05 0.06
Temps(h)
Time (h)

∆SoC = 10% (from 55% to 45%)

Figure 6. Voltage measurement during a relaxation period of 24 h at 45% SoC for an A123 cell.
Figure 6. Voltage measurement during a relaxation period of 24 h at 45% SoC for an A123 cell.

Figure 6 shows that the voltage during relaxation evolves over several hours. The variation of
Figure
voltage 6beyond
shows1that the voltage
h appears during
negligible when relaxation
comparedevolves over several
to the battery voltage. hours.
Indeed,The variation of
the voltage
voltage beyond
variation 1 h appears
during relaxationnegligible
between 1 hwhen and 24compared to the battery
h is approximately voltage.points
2 mV (between Indeed, theCvoltage
B and in
Figure
variation 6). Despite
during this small
relaxation change,
between 1hitsand
determination is important for
24 h is approximately batteries
2 mV with flat
(between OCV-SoC
points B and C in
Figurecurves because this
6). Despite a difference of only aits
small change, few millivolts mayiscause
determination a significant
important error in SoC
for batteries withestimation.
flat OCV-SoC
curves because a difference of only a few millivolts may cause a significant error inpart
This applies to the tested cell, whose OCV-SoC curve has a substantially horizontal SoCbetween
estimation.
40% SoC and 60% SoC. Figure 7 provides the SoC estimation error introduced by a difference of 2
This applies to the tested cell, whose OCV-SoC curve has a substantially horizontal part between 40%
mV for the case of Figure 6. Points A, B, and C in Figure 7 correspond to those in Figure 6. Points A
SoC and 60% SoC. Figure 7 provides the SoC estimation error introduced by a difference of 2 mV for
and B correspond, respectively, to the OCVs of 55% and 45% SoC measured at 1 h after the
the case of Figure
discharge; 6. CPoints
point A, B, and
corresponds C in
to the OCVFigure 7 correspond
of 45% SoC measured to those
at 24 in Figure
h after 6. Points
discharge. If A
theand B
correspond,
OCV-SoC curve measured at 1 h after discharge is used as the reference to estimate the SoC frompoint
respectively, to the OCVs of 55% and 45% SoC measured at 1 h after the discharge; a C
corresponds to the OCV ofthe
voltage measurement, 45% SoC measured
voltage evolution at 24 h after
during discharge.
relaxation betweenIf the
1 hOCV-SoC
and 24 h atcurve
45%measured
SoC
at 1 h(from
afterPoint
discharge is used
B to Point asintroduce
C) will the reference to of
an error estimate
about 4% the SoC
into thefrom a voltage measurement, the
SoC estimation.
voltage evolution during relaxation between 1 h and 24 h at 45% SoC (from Point B to Point C) will
introduce an error of about 4% into the SoC estimation.
Batteries 2016, 2, 7 6 of 15
Batteries 2016, 2, 7 6 of 15

Batteries 2016, 2, 7 6 of 15

Figure 7. SoC estimation error introduced by voltage evolution during relaxation on the cell of Figure 6.
Figure 7. SoC estimation error introduced by voltage evolution during relaxation on the cell of Figure 6.
3.2. Identification of the Relaxation
3.2. Identification
Figure 7.ofSoC
theestimation
Relaxation errorresults
introduced byidentifications
voltage evolution
This section presents the of two of during relaxation
the relaxation on the cell
voltage of Figure
shown 6.
in Figure
6. These
This section results
presentsrevealthetheresults
possibility
of twoof extrapolating
identifications the ofslowthevoltage
relaxationbehavior of theshown
voltage relaxation
in Figure 6.
3.2.from
Identification
a short andof partial
the Relaxation
measurement.
These results reveal the possibility of extrapolating the slow voltage behavior of the relaxation from a
Thesection
first identification
presents theisresults
to identify theidentifications
parameters of aofmodel with four voltage
Ri//Ci circuits (nin
= 4Figure
in
short andThis partial measurement. of two the relaxation shown
Equation (1)) by using the first hour of relaxation voltage. The second
6. These results reveal the possibility of extrapolating the slow voltage behavior of the relaxation identification is to identify the
The first identification
parameters of a partial
model withis tofive
identify the parameters
Ri//Ci circuits (n = 5 in Equation of a model(1)) bywith
usingfour Ri //Cduring
the voltage i circuits
24 h(n = 4 in
from a short and measurement.
Equation (1)) by using the first hour of relaxation voltage. The second identification is to identify the
of relaxation.
The first identification is to identify the parameters of a model with four Ri//Ci circuits (n = 4 in
parameters firstRihour
The ause
of (1))
model of models
with with
five four and
//C five circuits
iofcircuits R5 i//C
(n =voltage.ini, respectively,
Equation forby
(1)) theusing
first hour
the and for the
voltage 24
during
Equation by using the relaxation The second identification is to identify the 24 h
h of relaxation is a compromise between model complexity and identification accuracy. In the
of relaxation.
parameters of a model with five Ri//Ci circuits (n = 5 in Equation (1)) by using the voltage during 24 h
literature, many authors use two or three Ri//Ci circuits for the model [10,13–15]. However, these
The use of models with four and five circuits Ri //Ci , respectively, for the first hour and for the
of relaxation.
models are not intended to accurately represent at the same time the fast behavior related to charge
24 h of transfer
relaxation
The use ofis a compromise
models with four and between model
five circuits Ri//Ccomplexity
i, respectively, and foridentification
and the slow behavior related to diffusion effect (e.g., during relaxation). As we can see in
the first hour and accuracy.
for the 24 In the
h of
literature,
Figurerelaxation
many is
authors
8, which a compromise
use two
represents between
or three
the root mean square model
Ri //C(RMS) complexity
i circuits errorforfor theand identification
model [10,13–15].
the estimation accuracy. In
However, these
of the 24 h relaxation the
modelsliterature,
voltage many
are notversus
intended authors
the number useof two
to accurately or represent
circuits three Ri//Cat
for different i circuits for the model [10,13–15]. However, these
the same
battery time thethe
technologies, fast
usebehavior related
of five Ri//C i circuitsto charge
models
transfer andare
allows us not
the toslow intended
obtain to accurately
an error
behavior reading
relatedofto represent
less than 1%.
diffusion at the same
effect timeduring
(e.g., the fastrelaxation).
behavior related As we to charge
can see in
transfer and the slow behavior related to diffusion effect (e.g., during relaxation). As we can see in
Figure 8, which represents the root RMS meanforsquare
errors estimation(RMS) error
of 24error for the estimation of the 24 h relaxation
h relaxation
Figure 8, which represents the root mean
versusfornumber
square (RMS)
of R //Cbattery
for the voltage
circuits technologies,
estimation of the 24 h relaxation
for different cellsthe use of five R //C circuits
voltage versus the number of circuits different i i
voltage versus the12 number of circuits for different battery technologies, the use of five Ri//Ci i circuits i
allows us tous
allows obtain
to obtain an error
an errorreading
reading ofofless
lessthan
than1%. 1%. A123 (LiFePO )
4
10 Kokam (NMC)
RMS errors for estimation of 24 h relaxation voltage
Saft (NiMH)
versus number of R //C circuits for different cells
i i
12 8
RMS error (%)

A123 (LiFePO )
4
10 6 Kokam (NMC)
Saft (NiMH)

84
RMS error (%)

2
6
1
0
4 1 2 3 4 5 6 7 8 9 10
Number of R //C circuits
i i

2
Figure 8. Average root mean square (RMS) error for the estimation of the 24 h relaxation voltage vs.
the number 1of Ri//Ci circuits for A123 2.3 Ah LiFePO4 cell, Kokam 1.5 Ah nickel manganese cobalt
(NMC) cell [16],
0 and Saft 3 Ah nickel metal hydride (NiMH) cell [17].
1 2 3 4 5 6 7 8 9 10
Number of R //C circuits
i i

Figure 8. Average root mean square (RMS) error for the estimation of the 24 h relaxation voltage vs.
Figure 8. Average root mean square (RMS) error for the estimation of the 24 h relaxation voltage vs.
the number of Ri//Ci circuits for A123 2.3 Ah LiFePO4 cell, Kokam 1.5 Ah nickel manganese cobalt
the number of R[16],
(NMC) cell i //C i circuits
and fornickel
Saft 3 Ah A123metal
2.3 Ah LiFePO
hydride 4 cell,cell
(NiMH) Kokam
[17]. 1.5 Ah nickel manganese cobalt
(NMC) cell [16], and Saft 3 Ah nickel metal hydride (NiMH) cell [17].
Batteries 2016, 2, 7 7 of 15
Batteries 2016, 2, 7 7 of 15

Identification of
Identification of unknown
unknown parameters
parameters in
in Equation
Equation (1)
(1) is
is carried
carried out
out with
with the least squares
algorithm implemented in Matlab. The identification results are shown in Figure 9.
Batteries 2016, 2, 7 7 of 15

Parameters Model with 4 Ri//Ci

Identification of unknown parameters in Equation (1) is carried out with the least squares
identified with U (mV) U2_Relax(mV) U3_Relax(mV) U4_Relax(mV)
algorithm implemented in Matlab. 1_Relax
The identification results are shown in Figure 9.
the first hour of
30.8 134.7 31.6 5.7
the relaxation
Parameters Model
τ1_Relax(s) with 4τ2_Relax
Ri//C(s)i τ3_Relax(s) τ4_Relax(s)
identified with U1.0
1_Relax(mV) U2_Relax
19 (mV) U3_Relax110
(mV) U4_Relax(mV)
1600
the first hour of
4 30.8 134.7 31.6 5.7
the relaxation
τ U
i =1
i_Relax
1_Relax(s)
= 0.2027V
τ (s)
2_Relax
OCV =(s)
τ3_Relax 3.2986V
τ4_Relax(s)
1.0 19 110 1600
Parameters Model
 Uwith 5 Ri//Ci
4

i_Relax = 0.2027V OCV = 3.2986V

identified with i =1
U1_Relax(mV) U2_Relax(mV) U3_Relax(mV) U4_Relax(mV) U5_Relax(mV)
24 hours of the
Parameters 30.7 Model with 5 Ri//Ci
135.2 32.3 5.0 1.2
relaxation
identified with
U1_Relax(mV)
24 hours of the τ1_Relax (s) τU2_Relax (mV)
2_Relax(s) U3_Relax(mV)
τ3_Relax(s) U4_Relax(mV)
τ4_Relax(s)U5_Relax(mV)
τ5_Relax(s)
relaxation 30.7 135.2 32.3 5.0 1.2
1.0 19 120 3100 33000
τ1_Relax(s) τ2_Relax(s) τ3_Relax(s) τ4_Relax(s) τ5_Relax(s)
5

 U1.0
i =1
i_Relax = 0.2044V
19 120= 3.3002V3100
OCV 33000
5

U
i =1
i_Relax = 0.2044V OCV = 3.3002V
Figure
Figure 9.
9. Results
Results of
of identifications.
identifications.
Figure 9. Results of identifications.
For the results of time constants (τi_Relax), the difference between the two models lies mainly in
For theForresults of time
the results constants
of time (τi_Relax
constants (τi_Relax),
), the difference
the difference between
between the the
two two models models lies mainly
lies mainly in in
the last time constants, as highlighted in Figure 9. The first three time constants of the two models
the last the
timelast time constants,
constants, as highlighted
as highlighted ininFigure
Figure9. 9. The
The first
firstthree time
three timeconstants
constants of theoftwothemodels
two models
are of the same
aresame order
of theorder of value.
same order This implies
of value. that, when therelaxation
relaxation time is longer,
Ri//CRi R i//Ci circuits
are of the of value. ThisThis implies
implies that,when
that, when the
the relaxationtime timeis longer,
is longer, circuits
i //C i circuits
with greater time constants
with greater needneed
time constants to be added
to be addedtotothe the model
model ininorder
orderto to represent
represent the dynamics
the slow slow dynamics
with greater time constants need to be added to the model in order to represent the slow dynamics of
of relaxation voltage
of relaxation with with
voltage good accuracy.
good accuracy.FigureFigure 10 presentsthethe
10 presents simulation
simulation of voltage
of voltage duringduring
a a
relaxation voltagephase
relaxation
withofgood
24 h
accuracy.
by using
Figure
the two
10 presents
models in
the9.simulation
Figure The model
of voltage
with four R
during a relaxation
i//Ci circuits is
relaxation phase of 24 h by using the two models in Figure 9. The model with four Ri//Ci circuits is
phase ofinsufficient
24 h by using the two models
to adequately representintheFigure
entire9.relaxation
The model with lack Rofi //C
four i circuits is insufficient
insufficient to adequately represent the entire relaxation due duetotothethe lack aoftime constant
a time with
constant with
to adequately represent the entire relaxation due to the lack of a time constant with higher value.
higher value.
higher value.

Figure 10. Simulation of the voltage during 24 h of relaxation with the two model parameters in
Figure 10. Simulation of the voltage during 24 h of relaxation with the two model parameters in
Figure 9.
Figure 9.
For the results of the voltage drops (Ui_Relax) in Figure 9, U4_Relax and U5_Relax are much smaller
Figure 10. Simulation of the voltage during 24 h of relaxation with the two model parameters in
than other Ui_Relax. In Table 1, a comparison between the values of Ui_Relax of the two models shows
Figure 9.results of the voltage drops (Ui_Relax ) in Figure 9, U4_Relax and U5_Relax are much smaller
For the
that there is a difference of 1.6 mV between the sums of Ui_Relax, which leads to a difference of 1.6 mV
than other Ui_Relax . In Table 1, a comparison between the values of Ui_Relax of the two models shows
in the estimation of OCV according to Equation (2) for the two models. The difference is not only
that there is aresults
Forrelated
the difference
to the value
ofof1.6
of the mV between
voltage drops
U5_Relax which
the
(U
is 1.2
sums
i_Relax
mV, ) inof
but
Urelated
Figure
also i_Relax9,,towhich
4_Relaxleads
Uother Uand
to a difference
U5_Relax are muchof 1.6
i_Relax. This difference of 1.6
mV in
smaller
the estimation
than other
mV U of OCV
i_Relax.inIn
results according
Table 1,
a voltage to Equation
a comparison
difference (2)
at the endbetween for the two models.
the values
of 24 h between The
of Ui_Relax
the measured difference
of the
voltage andtwois not only related
models shows
the estimated
to the
that value
there
oneisusing Uthe
aofdifference
model
5_Relax which
ofwith is 1.2
1.6four
mV mV,
Rbetween
i//C butthe
i circuits.also related
If sums
this of Uto
difference other Ui_Relax
in ,voltage
i_Relax which leads
could . be
This a difference
toestimated, mV
of 1.6 mV
it would
results
in beapossible
in voltageof
the estimation todifference
reduce
OCV the length
at theof
according endtheEquation
to experiment.
of 24 h between
(2) for thethe measured
two models. voltageThe and the estimated
difference one
is not only
using the model with four R
related to the value of U5_Relaxi which//C i circuits. If this difference in voltage could be
is 1.2 mV, but also related to other Ui_Relax. This difference of 1.6estimated, it would be
possible
mV to in
results reduce the length
a voltage of theatexperiment.
difference the end of 24 h between the measured voltage and the estimated
one using the model with four Ri//Ci circuits. If this difference in voltage could be estimated, it would
be possible to reduce the length of the experiment.
Batteries 2016, 2, 7 8 of 15

Table 1. Comparison of the Ui_Relax for the two models.

Batteries 2016, 2, 7 8 of 15
5
ř
Models U 1_Relax (mV) U (mV) U (mV) U (mV) U 5_Relax (mV) Ui_Relax (mV)
Table 1. 2_Relax 3_Relax
Comparison of the 4_Relax
Ui_Relax for the two models. i“1
Model with four Ri //Ci 30.8 134.7 31.6 5.7 0 5 202.8
Model with five Ri //Ci
Models
Difference (mV)
30.7
U1_Relax (mV)
´0.1
135.2
U2_Relax (mV)
0.5
32.3
U3_Relax (mV)
0.7
5.0
U4_Relax (mV)
´0.7
1.2
U5_Relax (mV)
1.2 
i =1
U i_Relax204.4
(mV)
1.6

Model with four Ri//Ci 30.8 134.7 31.6 5.7 0 202.8

Model with fiveofRithe
3.3. Extrapolation //Ci Relaxation
30.7 135.2 32.3 5.0 1.2 204.4
Difference (mV) −0.1 0.5 0.7 −0.7 1.2 1.6
This difference in voltage can be estimated by using a technique that transforms the voltage
3.3. Extrapolation
measurement from aoflinear-based
the Relaxation time scale to a logarithmic time scale. Figure 11 shows such a
transformation of the measured
This difference in voltage voltage
can befor the relaxation
estimated by using of a24technique
h. The voltage estimated
that transforms thebyvoltage
the model
with four Ri //Ci isfrom
measurement also shown in the same
a linear-based figure.
time scale to In Figure 11a, time
a logarithmic the divergence
scale. Figure between
11 shows thesuch
measured
a
voltagetransformation
and the estimated of the measured
one startsvoltage
at around for the
2.2 relaxation of 24 h. The voltage
h, which corresponds to the endestimated
of the by the of
validity
modeltimewithconstant
four Ri//C(τ i is also )shown in the same figure. In Figure 11a, the divergence between the
the fourth 4_Relax of the model with four Ri //Ci . The end of the validity, denoted as
∆tval , is equal to five times the value ofone
measured voltage and the estimated starts because
τ4_Relax at aroundafter
2.2 h,awhich corresponds
time equal to the, end
to 5τ4_Relax the of the
influence
validity of the fourth time constant (τ4_Relax) of the ´ 5 model with four Ri//Ci. The end of the validity,
of this time constant is considered negligible as e = 0.7%. The voltage difference at the end of 24 h
denoted as ∆tval, is equal to five times the value of τ4_Relax because after a time equal to 5τ4_Relax, the
(denoted by ∆UDiff in Figure 11a) between the measured and the estimated voltage is the one we want
influence of this time constant is considered negligible as e−5 = 0.7%. The voltage difference at the end
to estimate. In Figure
of 24 h (denoted by11b,
ΔUDiffwith time 11a)
in Figure in abetween
logarithmic scale, the
the measured andmeasured
the estimatedvoltage
voltagebeyond 2.2 h is
is the one
close to a straight line, as indicated by the black dotted line. The voltage difference
we want to estimate. In Figure 11b, with time in a logarithmic scale, the measured voltage beyond at the end of 24 h
between2.2 hthe straight
is close to a line andline,
straight the as
estimated
indicatedvoltage of the
by the black model
dotted with
line. Thefour Ri //C
voltage i can thus
difference beend
at the used to
of 24∆U
estimate Diff . Thethe
h between slope of the
straight line
line andcould be determined
the estimated voltage of bythe
a linear
model regression
with four Riusing
//Ci canonly
thusabesmall
part ofused
thetomeasured
estimate ΔU Diff. Thebeyond
voltage slope of 2.2
the h,
lineascould be determined
indicated in Figure by11b.
a linear
Theregression
only problemusing toonly a is
solve
small part of the measured voltage beyond 2.2 h, as indicated in Figure
to determine the added duration needed beyond 2.2 h to identify the line, as shown in Figure 11b. 11b. The only problem to
solve is to determine
The advantage the addedscale
of the logarithmic duration
is toneeded
enablebeyond 2.2 h to
estimation ofidentify
this addedthe line, as shown
duration in Figure
needed. Indeed,
11b. The advantage of the logarithmic scale is to enable estimation of this added duration needed.
as the objective is to estimate the value of ∆UDiff between the four Ri //Ci model’s response after 24 h
Indeed, as the objective is to estimate the value of ∆UDiff between the four Ri//Ci model’s response
and the real measures, the linear shape of signals in the logarithmic scale (Figure 11b) is better than in
after 24 h and the real measures, the linear shape of signals in the logarithmic scale (Figure 11b) is
the linear
betterscale
than (Figure 11a),scale
in the linear as explained
(Figure 11a),in as
theexplained
following in part of the text.
the following part of the text.

Tension
Voltage de relaxation
of relaxation à 45% de
measured at SoC
45%et sesand
SoC estimations
its estimation
3.302

3.3
(V)
Tension(V)

∆UDiff
x
Voltage

3.298

3.296
∆tval = 2.2 h Voltage measured
Tension mesuréeduring 24 relaxation
pendant h of relaxation
de 24h
Estimation of model
Estimation du modèle 4 Ri//C
with avec i identified
4R//C identifiéwith
avec1 1h
h of
derelaxation
relaxation

1 5 10 15 20 25
Time
Temps (h)(h)
(a)

Time in logarithmic scale

Tension
Voltage de relaxation
of relaxation à 45% de at
measured SoC et ses
45% SoCestimations
and its estimation
3.302
Added duration
3.3
Voltage (V)

x
Tension

3.298
2.2 h
Voltage measured
Tension mesuréeduring 24 relaxation
pendant h of relaxation
de 24h
3.296
Estimation of model
Estimation du modèle 4 Ri//C
with avec i identified
4R//C identifiéwith
avec11h
h of
derelaxation
relaxation
0 1
10 10
Time
Temps in logarithmic
à l'échelle scale(h)(h)
logarithmique
(b)

Figure 11. Long relaxation (24 h) at 45% SoC of an A123 cell with: (a) linear scale on time axis and (b)
Figure 11. Long relaxation (24 h) at 45% SoC of an A123 cell with: (a) linear scale on time axis and
a logarithmic scale.
(b) a logarithmic scale.
Batteries 2016, 2, 7 9 of 15

In short, the study of relaxation in this section reveals the possibility of extrapolating the slow
behavior of the relaxation by using the observation that some part of the relaxation voltage is close
to a straight line in logarithmic time scale, as shown in Figure 11. Based on the result of this study,
an extrapolation method is elaborated. The method allows us to reconstruct a new model with five
Ri //Ci from a short and partial measurement of relaxation voltage. The new model is capable of
correctly estimating the slow relaxation voltage to 24 h. With this method, it is then possible to reduce
the relaxation measurement to a shorter duration. The details of this method are discussed in the
next section.

4. Extrapolation Method of Relaxation

4.1. Principle of the Method

The proposed method allows the characterization of the long relaxation battery voltage at different
SoCs with a reduced time. For the sake of simplicity and practicality, the characterization for the
relaxation of 24 h is taken as an example to explain the method. Furthermore, 24 h of relaxation is
well adapted to automotive applications, where daily uses are regular. However, the method can be
transposed to characterize relaxation with longer or shorter duration. The relaxation time can vary
greatly with different technologies of cells and with the temperature of operation. The method includes
two phases:
The first phase is to measure the battery voltage during 24 h of relaxation at an intermediate SoC
(e.g., 45% SoC). With this measurement, a minimum duration (∆tmin , less than 24 h) is determined to
measure the relaxation at other SoCs.
The second phase is to measure at other SoCs the battery relaxation voltage during the minimum
duration defined in Step 1. With these measurements, a model with five Ri //Ci can be reconstructed.
For each SoC tested, the parameters (Ui_Relax and τi_Relax ) of the first four Ri //Ci are identified
by using the measurement during the first hour of relaxation, like the example shown in Figure 9.
The parameters (U5_Relax and τ5_Relax ) of the fifth R//C circuit (R5 //C5 ) can be estimated by using the
relaxation voltage measured after the first hour and the extrapolation at 24 h of the signal measured
during ∆tmin .
Once the values of Ui_Relax and τi_Relax are obtained, OCV can be calculated from the measurement
of URelax_init and Equation (2).
The determinations of the minimum duration in the first phase and the five Ri //Ci in the second
phase are detailed as follows.

4.2. Determination of the Minimum Duration

The minimum duration ∆tmin is calculated as:

∆tmin “ ∆tval ` ∆tadded (3)

where ∆tval is the end of the validity of the fourth time constant (τ4_Relax ) of the model with four
Ri //Ci . τ4_Relax and other parameters of the model with four Ri //Ci are determined by using the
first hour of relaxation. ∆tadded is the added duration needed to identify the slope of the line, as in
Figure 11b. To determine the added duration needed just to identify the slope, a study is needed on the
relationship between the measurement duration added beyond ∆tval and the identification accuracy of
the line slope. The study is conducted in several steps, as illustrated in Figure 12 by using the data in
Figure 11.
To overcome the influence of measurement noise on the voltage, the model with five Ri //Ci
identified with relaxation of 24 h was used (green curve in Figure 12).
After transforming the time into a logarithmic scale, an added duration beyond ∆tval (2.2 h in
the example of Figure 12) is used to identify the slope of the line. The error ∆UError at the end of 24 h
Batteries 2016, 2, 7 10 of 15
Batteries 2016, 2, 7 10 of 15

between the identified

By repeating theline and theofestimation
calculation the error ΔUof the model with five Ri //Ci isthe
Error with different added durations,
used to assess
evolution of the
accuracy of the identified line slope.
the error versus the added duration can be obtained.

Step 1 :
Voltage of relaxation measured at 45% SoC and its estimation
3.302

Voltage (V)
3.3
Batteries 2016, 2, 7 10 of 15
3.298

By repeating the calculation of the error

Voltage
Tension ΔU
measured
mesurée Error with
during
pendant 24 h ofdifferent
relaxation de 24h added durations, the evolution of
relaxation
3.296 Estimation du
can beofobtained.
Estimation
the error versus the added duration model
modèlewith
avec5R i//Ci identified
5R//C with 24
identifié avec 24hhde
of relaxation

1 5 10 15 20 25
Time (h)
Step 1 :
Voltage of relaxation measured at 45% SoC and its estimation
3.302 Time on a logarithmic scale
Voltage (V)

3.3
Step 2 : Line
Added duration identified
3.302 3.298

Voltage mesurée
Tension measuredpendant
during relaxation
24 h of relaxation
de 24h
3.296 Estimation du
Estimation of model
modèlewith
avec5R i//Ci identified
5R//C with 24
identifié avec 24hhde
of relaxation ∆UError
Voltage ((V))

3.3

3.298
1 5
xTime (h)
10 15 20 25

Portion used to
∆tval
3.296
Time on a logarithmicidentify
scale the line
= 2.2 h
Step 2 : 1 5 10 Line
15 20 25
Time in logarithmic scale (h)
Added duration identified
3.302

∆UError
Voltage ((V))

3.3

3.298 Step 3 : x
Portion used to
∆tval identify the line
|∆UError|

3.296 = 2.2 h
1 5 10 15 20 25
Time in logarithmic scale (h)

Step 3 : Added duration

|∆UError|

Figure 12. Illustration of steps to investigate the relationship between the added duration beyond 2.2
Figure 12. Illustration of steps to investigate the relationship between the added duration beyond 2.2 h
h and the identification accuracy of the line slope.
and the identification accuracy of the line slope.
Figure 13 shows the evolution of the error vs. the value of the added duration for the relaxation
Added duration
By repeating
of Figure 12. Inthe calculation
Figure of the
13, the error ∆Uerror ∆UErroras
Error decreases with different
the added addedincreases.
duration durations, the evolution
By choosing an of
acceptable
the error value
versusFigure
the 12. of
added theduration
error,
Illustration thecan
of steps necessary
to duration
be obtained.
investigate of ∆tbetween
the relationship added canthebe determined.
added For example,
duration beyond 2.2 to
achieve13
Figure ahshows
level ofidentification
and the precision
the evolution equaloftoof
accuracy 0.5 mV,slope.
theerror
the line the duration
vs. the valueadded
of theisadded
equal to 3 h. In this
duration case,relaxation
for the the
minimum
of Figure 12. In duration
Figure ∆t min is equal to 5.2 h (2.2 h + 3 h) according to Equation (3). This means that for
13, the ∆UofError
Figure the error decreases as the added
addedduration increases. By choosing
the relaxation at13other
shows
SoCs, evolution
only 5.2 h the error
of vs. the value
measurement of the
is needed insteadduration for
of 24 h. the relaxation
an acceptable value
of Figure 12. of the error,
In Figure 13, thethe necessary
error duration
∆UError decreases as theof ∆tadded
added canincreases.
duration be determined.
By choosingFor
an example,
to achieveacceptable
a level of value of the error,
precision |Δthe
equal U|∆U
to
Ecart
|
0.5en
necessary
| fonciton
mV,
v.sduration
the
Added de
of la
duration durée
∆tadded
duration can
added
forajoutée
be determined.
the is equal For3example,
to h. In to case, the
this
Error
achieve a level of precision equal to 0.5 mV, the duration added is equal to 3 h. In this case, the
minimum duration ∆tmin is equal to 5.2 pour hrelaxation
la
(2.2relaxation
h + 3 h) at àaccording
45%SoC
45% de SoC to Equation (3). This means that for
minimum duration ∆tmin1.5 is equal to 5.2 h (2.2 h + 3 h) according to Equation (3). This means that for
the relaxation at otheratSoCs,
the relaxation only only
other SoCs, 5.2 h5.2of measurement
h of measurement isisneeded
needed instead
instead of 24 of
h. 24 h.
(mV)

1 |Δ U|∆U | en fonciton de la durée ajoutée

Ecart | v.s Added duration for the
Error
pourrelaxation
la relaxation
at à45%
45%SoC
de SoC
Error|
Ecart

1.5
0.5
|∆U
(mV)

1
0
Error|
Ecart

0 3 5 10 15 20 25
0.5 Durée
Addedajoutée au-delà
duration de h2.2h
after 2.2 (h) (h)
|∆U

Figure 13. Error ∆UError vs. the added duration ∆tadded.

0
0 3 5 10 15 20 25
Durée
Addedajoutée au-delà
duration de h2.2h
after 2.2 (h) (h)
Figure 13. Error ∆UError vs. the added duration ∆tadded.
Figure 13. Error ∆UError vs. the added duration ∆tadded .
Batteries 2016, 2, 7 11 of 15

4.3. Determination
Batteries 2016, 2, 7 of the Parameters in R5 //C5 11 of 15
Batteries 2016, 2, 7 11 of 15
Two unknown parameters
4.3. Determination (τ5_Relax
of the Parameters //C5 U5_Relax ) of the five Ri //Ci circuit need to be identified for
in R5and
every4.3. Determination
relaxation measured of the Parameters in R5//C5
with the minimum duration. The time constant τ5_Relax is equal to the one
Two unknown parameters (τ5_Relax and U5_Relax) of the five Ri//Ci circuit need to be identified for
identifiedTwoin the first phase
unknown with 24 (τ
parameters h5_Relax
of relaxation. of the is
and U5_Relaxτ)5_Relax supposed
five to beneed
Ri//Ci circuit constant for all other
to be identified for SoCs.
every relaxation measured with the minimum duration. The time constant τ5_Relax is equal to the one
everyverified
This was relaxation by measured
the tests with the minimum
described as followsduration.
for The time
different constant
battery τ5_Relax is equal
technologies. The tosame
the one tests of
identified in the first phase with 24 h of relaxation. τ5_Relax is supposed to be constant for all other
identified
relaxations of in h
24 theasfirst
in phase with
Section 3 24 hperformed
were of relaxation.on τ5_Relax different
three is supposedcellsto (LiFePO
be constant , for
NMC, all other
and NiMH
SoCs. This was verified by the tests described as follows for different battery technologies. 4 The same
SoCs.
cells)tests This
at different was verified
SoCs. of Whileby the tests described as follows for different battery technologies. The same
of relaxations 24 h comparing
as in Sectionthe values
3 were of the time
performed constants
on three identified
different for the4,model
cells (LiFePO NMC, with
tests of relaxations of 24 h as in Section 3 were performed on three different cells (LiFePO4, NMC,
five Rand
i //CNiMH
i , a logarithmic
cells) at distribution
different SoCs. of
While time constants
comparing the was found
values of thefor the
time three
constants cells. Figures
identified for14–16
and NiMH cells) at different SoCs. While comparing the values of the time constants identified for
the the
present model with of
results five Ri//C
the i, a constants
time logarithmicidentified
distribution forofthe
timemodel
constants
withwas fivefound
Ri //C fori .the
The three cells.show
results
the model with five Ri//Ci, a logarithmic distribution of time constants was found for the three cells.
Figures
that the 14–16 present
distribution the results
of τi_Relax of the time constants
in a logarithmic identified for the
scale is substantially model
linear. with
Thewith time five Ri//Ci. The
constants τi_Relax
Figures 14–16 present the results of the time constants identified for the model five Ri//Ci. The
results show that the distribution of τi_Relax in a logarithmic scale is substantially linear. The time
results show that the distribution of τi_Relax in a logarithmic scale is substantially linear. The time of i.
for different SoCs and different cells also have the same magnitudes of value for the same index
constants τi_Relax for different SoCs and different cells also have the same magnitudes of value for the
For example,
constantsthe values
τi_Relax of τ5_Relax
for different SoCs forand
different
differentSoCscellsand
alsodifferent
have the cells
same are very close
magnitudes of (around
value for 38,000
the s).
same index of i. For example, the values of τ5_Relax for different SoCs and different cells are very close
The magnitudes
same index ofofi.values of
For example, different time
the values constants
of τ5_Relax forare: τ
different =
SoCs
1_Relax 1.3
ands,different
τ2_Relax
(around 38,000 s). The magnitudes of values of different time constants are: τ1_Relax = 1.3 s, τ2_Relax = 17
= 17
cells s,
are τvery =
close
3_Relax 220 s,
(around
τ4_Relax = 2900 38,000
s and s).τThe magnitudes
5_Relax = 38,000 of
s values
(«11 of
h). different time constants are: τ 1_Relax = 1.3 s, τ2_Relax = 17
s, τ3_Relax = 220 s, τ4_Relax = 2900 s and τ5_Relax = 38,000 s (≈11 h).
s, τ3_Relax = 220 s, τ4_Relax = 2900 s and τ5_Relax = 38,000 s (≈11 h).
|τ i_Relax| of relaxation at different SoC (A123)
|τ i_Relax| of relaxation at different SoC (A123)
i5 Relax
10i5 Relax
10
(s)(s)

4
104
scale

10
scale

3
103
logarithmic

10
logarithmic

2
102
10

1 after discharge
85% après décharge
101
| in

85% after
after discharge
55% après
après décharge
discharge
décharge
10
| in

55% after
after discharge
25% après décharge
discharge
|τi_Relax

après décharge
25% after discharge
|τi_Relax

0 25% après
après décharge
after charge
recharge
100
25% after
after charge
55% après
après recharge
charge
recharge
10
55% after
after charge
85% après
après recharge
charge
recharge
-1 85% après
after charge
recharge
10-1
10 1 2 3 4 5
1 2 Value3of i 4 5
Value of i
Figure 14. Time constants τi_Relax identified of the model with five Ri//Ci for different SoCs of an A123
Figure 14. Time
Figure constants
14. Time τi_Relaxidentified
τi_Relax
constants identified of the model
of the modelwith fiveRR
withfive i //C
i//C for different
i fori different SoCsSoCs
of anof an A123
A123
2.3 Ah LiFePO4 cell.
2.3 Ah
2.3LiFePO 4 cell.
Ah LiFePO 4 cell.

|τ i_Relax| of relaxation at different SoC (Kokam)

|τ i_Relax
5
| of relaxation at different SoC (Kokam)
105
10
(s)(s)

4
104
scale

10
scale

3
103
logarithmic

10
logarithmic

2
102
10

1 85% après
85% after discharge
décharge
101
| in

85%
85%
55% after
after discharge
55%après
après décharge
discharge
décharge
10
| in

55%
55% after
after discharge
25%après décharge
|τi_Relax

25% après discharge

décharge
|τi_Relax

0 25%
25%
25% after
after discharge
25%après
après décharge
charge
recharge
100
10 25%
25%
55% after
after charge
55%après
après recharge
charge
recharge
55%
55%
85% after
after charge
85%après
après recharge
charge
recharge
-1
10-1 85% après
85% after charge
recharge
10 1 2 3 4 5
1 2 Value3 of i 4 5
Value of i
Figure 15. Time constants τi_Relax identified of the model with five Ri//Ci for different SoCs of a Kokam
Figure 15. Time constants τi_Relax identified of the model with five Ri//Ci for different SoCs of a Kokam
Figure AhTime
1.515. NMCconstants
cell. τi_Relax identified of the model with five Ri //Ci for different SoCs of a Kokam
1.5 Ah NMC cell.
1.5 Ah NMC cell.
Batteries 2016, 2, 7 12 of 15
Batteries 2016, 2, 7 12 of 15

|τ i_Relax| of relaxation at different SoC (Saft )

5
ea
10

|τi_Relax| in logarithmic scale (s)

4
10

3
10

2
10

1 85%
85%après décharge
after discharge
10
55%
55%après décharge
after discharge
25%
25%après décharge
after discharge
0
10 25%après
25% after charge
recharge
55%après
55% after charge
recharge
85%
85%après recharge
after charge
-1
10
1 2 3 4 5
Value of i
Figure 16. Time constants τi_Relax identified of the model with five Ri//Ci for different SoCs of a Saft 3
Time
Figure 16.Ah constants τi_Relax identified of the model with five Ri //Ci for different SoCs of a Saft 3
NiMH cell.
Ah NiMH cell.
Another unknown parameter U5_Relax of the five Ri//Ci circuit is calculated by the following
expression:
Another unknown parameter U5_Relax of the five Ri //Ci circuit is calculated by the
'
ΔU Diff
following expression: U 5_Relax =
 − Relax 
Δt
 τ (3)
1 − e 5_Relax
1 
∆UDiff
U5_Relax “ ¨  
 (4)
∆tRelax
˛
´
where ∆tRelax is the total duration of the relaxation ˚ (hereτ24 h) and ∆U′Diff is the estimated value for
5_Relax ‹
1 ´ e
∆UDiff (Figure 11). In the second phase, the voltage measured during relaxation for every SoC is
˝ ‚
limited to the minimum duration, which is determined previously in the first phase (e.g., 5.2 h). The
first hour of the measurement is used to identify the parameters of the model with four Ri//Ci. The
where ∆tRelax is the total
measurement duration
after the end of theofvalidity
the relaxation
of the fourth(here 24 h) and
time constant (∆tval)∆U
1
is used isidentify
Diff to the estimated
a line value
for ∆UDiffin(Figure 11).scale,
logarithmic In the second in
as illustrated phase,
Figure the voltage
12. ∆U′ measured
Diff is equal during
to the voltage relaxation
difference forofevery SoC
at the end
is limited 24
tohthe
between the estimation
minimum duration,of thewhich
model with four Ri//Ci andpreviously
is determined the identifiedinline.theAtfirst
this phase
point, the
(e.g., 5.2 h).
interest of the linear shape of the curve is obvious as it facilitates the computation of ∆U′Diff and
The first hour of the measurement is used to identify the parameters of the model with four Ri //Ci .
finally of U5_Relax.
The measurement after the end of the validity of the fourth time constant (∆tval ) is used to identify a
line in logarithmic scale,
4.4. Discussion as Values
on the of Ui_Relaxin Figure 12. ∆U1 Diff is equal to the voltage difference at the end
illustrated
of 24 h between Thethe estimation
values ofthe
of Ui_Relax for theA123
model with
cell at different Ri //C
fourvalues andarethe
of iSoC identified
given in Table 2.line.
It canAt
be this point,
seen
the interest of that
the U i_Relax varies
linear shape when SoCcurve
of the goes from 25 to 85. Consequently,
is obvious the the
as it facilitates same of ∆U1 Diff and
set of Ui_Relax values
computation
cannot be used for all SoCs. If the same set of values was used to estimate the relaxation voltage for
finally of U 5_Relax .
all SoCs, an important error in voltage estimation would be generated at the end of 24 h. For
example, from Table 3, if values of 45% SoC are used, the error in the voltage estimation would be
4.4. Discussion on the Values of U mV for 85% SoC and 222.1 mV − 205.0 mV = 17.1 mV for 25% SoC. As
205.0 mV − 181.9 mV = 23.1i_Relax
mentioned
The values of in
UFigure 7, such an error in voltage may generate an important error in the estimation of
i_Relax for the A123 cell at different values of SoC are given in Table 2. It can
SoC.
be seen that Ui_Relax varies when SoC goes from 25 to 85. Consequently, the same set of Ui_Relax
values cannot be used for all 2.SoCs.
Table IfUthe
Values of same
i_Relax for set cell
the A123 of atvalues
differentwas
valuesused
of SoC.to estimate the relaxation
voltage for all SoCs, an important error in voltage estimation would
SoC U1_Relax (mV) U2_Relax (mV) U3_Relax (mV) U4_Relax (mV) U5_Relax (mV) be generated at the end of 24 h.
ΣUi_Relax (mV)
For example, from85% Table
24.72, if values116.7of 45% SoC 29.9 are used, 9.1the error 1.5
in the voltage 181.9 estimation would
45% 30.8 134.7 31.6 5.7 2.3 205.0
be 205.0 mV ´ 25%181.9 mV34.6
= 23.1 mV for
134.4
85% SoC
39.6
and 222.1
8.1
mV ´ 205.0
5.4
mV = 17.1 mV for 25% SoC.
222.1
As mentioned in Figure 7, such an error in voltage may generate an important error in the estimation
of SoC.

Table 2. Values of Ui_Relax for the A123 cell at different values of SoC.

SoC U 1_Relax (mV) U 2_Relax (mV) U 3_Relax (mV) U 4_Relax (mV) U 5_Relax (mV) ΣUi_Relax (mV)
85% 24.7 116.7 29.9 9.1 1.5 181.9
45% 30.8 134.7 31.6 5.7 2.3 205.0
25% 34.6 134.4 39.6 8.1 5.4 222.1
Batteries 2016, 2, 7 13 of 15

5. Validation
Batteries 2016,Test
2, 7 13 of 15

To evaluate the
5. Validation Test performance of the proposed method, a validation test was performed by using
three relaxations of 24 h measured on the A123 cell. First, with the relaxation of 24 h at SoC of 45%
To evaluate the performance of the proposed method, a validation test was performed by using
(Figure 11) and by choosing 0.5 mV as the acceptable accuracy, the minimum measurement duration
three relaxations of 24 h measured on the A123 cell. First, with the relaxation of 24 h at SoC of 45%
(∆tmin(Figure
) is determined to be 5.20.5h mV
11) and by choosing (Figure
as the13). Second,
acceptable with the
accuracy, measurements
the minimum of the duration
measurement first 5.2 h of
relaxation at SoC of 85%, 45%, and 25%, three models with
(∆tmin) is determined to be 5.2 h (Figure 13). Second, with the measurements five R i //C i wereof the first 5.2 h of The
reconstructed.
performance
relaxationofatthe
SoC reconstructed
of 85%, 45%,models
and 25%, to three
estimatemodelsrelaxations
with fiveofR24i//Ch is shown
i were in FigureThe
reconstructed. 17. The
errorsperformance
at the end of ofthe
24 reconstructed
h between the estimation
models of the
to estimate reconstructed
relaxations of 24 h model
is shown and the measurement
in Figure 17. The
errors
are ´0.8 mVat the end of
at SoC 24 h ´0.2
85%, between
mVtheatestimation
SoC 45% of the0.1
and reconstructed
mV at SoC model
25%.and theresult
The measurement
shows thatare the
−0.8 mV at SoC 85%, −0.2 mV at SoC 45% and 0.1 mV at SoC 25%.
reconstructed models can accurately estimate the long relaxation during 24 h. Thus, to obtain The result shows that the the
reconstructed models can accurately estimate the long relaxation during
relaxations of 24 h at these three SoCs with the proposed method, it is only necessary to perform 24 h. Thus, to obtain the
relaxations of 24 h at these three SoCs with the proposed method, it is only necessary to perform a
a measurement of 24 h of relaxation at 45% SoC and two measurements of 5.2 h at 85% and 25%
measurement of 24 h of relaxation at 45% SoC and two measurements of 5.2 h at 85% and 25% SoC,
SoC, in total 34.4 h of measurement. Compared to the conventional method, which requires 72 h of
in total 34.4 h of measurement. Compared to the conventional method, which requires 72 h of
measurement
measurement(e.g.,(e.g.,
three measurements
three measurements ofof2424h hfor
forthe
thethree
three SoCs), the proposed
SoCs), the proposedmethodmethod saves
saves 37.6 h,
37.6
whichh,represents a gain of
which represents 52%of
a gain in52%
measurement
in measurementtime, time,
without considering
without considering the setting of initial
the setting SoC for
of initial
each test.
SoC for each test.

Voltage Tension de relaxation

of relaxation measuredà 85%
at de
85%SoC et ses
SoC andestimations
its estimations
3.347

3.346
Tension (V)

3.345
Voltage

3.344

3.343 Voltage
Tensionmeasured during 24relaxation
mesurée pendant h of relaxation
24h
3.342 Model
Modèlewith 4 Ridentifiée
4R//C i//Ci identified
avecwith 1 hrelaxation
1h de of relaxation
Modèlewith
Model 5R//C
5 Rreconstruit avec5.2
i//Ci rebuilt with 5.2h derelaxation
h of relaxation
3.341
1 5 10 15 20 25
Time
Temps(h)(h)
(a)
Voltage Tension de relaxation
of relaxation measuredà 45%
at de
45%SoC et ses
SoC andestimations
its estimations
3.302

3.3
(V)
Tension(V)
Voltage

3.298
Voltage measured
Tension during 24relaxation
mesurée pendant h of relaxation
24h
Model with
Modèle 4Ridentifiée
4R//C i//Ci identified
avecwith
1h 1deh relaxation
of relaxation
3.296 Modèle 5R//C
Model with 5Rreconstruit avec5.2
5.2h
h ofderelaxation
relaxation
i//Ci rebuilt with

1 5 10 15 20 25
Time
Temps(h)(h)
(b)
Voltage Tension de relaxation
of relaxation measuredà 25%
at de
25%SoC et ses
SoC andestimations
its estimations
3.265
(V)
Tension(V)

3.26
Voltage

3.255 Voltage
Tensionmeasured during 24relaxation
mesurée pendant h of relaxation
24h
Model
Modèlewith
4R//C i//Ci identified
4 Ridentifiée avecwith 1 hrelaxation
1h de of relaxation
Modèlewith
Model 5R//C
5 Rreconstruit avec5.2
i//Ci rebuilt with 5.2h derelaxation
h of relaxation
3.25
1 5 10 15 20 25
Temps
Time (h)(h)
(c)
Figure 17. Performance of the model with five Ri//Ci reconstructed with 5.2 h of relaxation at SoC of:
Figure 17. Performance of the model with five Ri //Ci reconstructed with 5.2 h of relaxation at SoC of:
(a) 85%; (b) 45%; and (c) 25% for A123 cell.
(a) 85%; (b) 45%; and (c) 25% for A123 cell.
Validation tests were also conducted on two other technologies (NMC and NiMH). For
ValidationFigure
example, tests 18 presents
were the result for
also conducted onestimation
two otherof relaxation voltage
technologies (NMCatand 25%NiMH).
SoC forFor
the example,
two
cells. By using the same acceptable accuracy as the A123 cell (0.5 mV), the minimum measurement
Figure 18 presents the result for estimation of relaxation voltage at 25% SoC for the two cells. By using
duration
the same (∆tmin) is
acceptable determined
accuracy to be
as the 5.1 h
A123 for(0.5
cell the mV),
Kokamthecell and 7.8 hmeasurement
minimum for the Saft cell. The accurate
duration (∆t ) is min
Batteries 2016, 2, 7 14 of 15
Batteries 2016, 2, 7 14 of 15

determined to be 5.1
estimation result in hFigure
for the18
Kokam
showscell andthe
that 7.8proposed
h for the Saft cell. The
method accurate
is also estimation
applicable result
to these twoin
Figure 18 shows that the proposed method is also applicable to these two technologies.
technologies.

Voltage of relaxation measured at 25% SoC and its estimations (Kokam)

3.685
Voltage (V)

3.675

Voltage measured during 24 h of relaxation

3.665 Model with 4 R //C identified with 1 h of relaxation
i i
Model with 5 R //C rebuilt with 5.1 h of relaxation
i i
3.655
1 5 10 15 20 25
Time (h)
(a)
Voltage of relaxation measured at 25% SoC and its estimations (Saft)
1.295
Voltage (V)

1.285

Voltage measured during 24 h of relaxation

1.275 Model with 4 R //C identified with 1 h of relaxation
i i
Model with 5 R //C rebuilt with 7.8 h of relaxation
i i
1.265
1 5 10 15 20 25
Time (h)
(b)
Figure 18.
18. Performance
Performance of of the
the model
modelwithwithfive
fiveRR//C
i//Ci reconstructed at 25% SoC for: (a) a Kokam cell
Figure i i reconstructed at 25% SoC for: (a) a Kokam cell
with 5.1 h of relaxation; and (b) a Saft cell with 7.8 hh of
with 5.1 h of relaxation; and (b) a Saft cell with 7.8 of relaxation.
relaxation.

6. Conclusions
6. Conclusions
A fast characterization method to determine the full relaxation response of a battery (e.g., 24 h)
A fast characterization method to determine the full relaxation response of a battery (e.g., 24 h)
has been proposed in this paper. This method allows for reconstructing an electric circuit model that
has been proposed in this paper. This method allows for reconstructing an electric circuit model that is
is capable of correctly estimating the voltage relaxation. Only one complete measurement of the
capable of correctly estimating the voltage relaxation. Only one complete measurement of the voltage
voltage relaxation at one SoC is needed to calibrate the method. For other SoCs, only a shorter
relaxation at one SoC is needed to calibrate the method. For other SoCs, only a shorter measurement
measurement duration (e.g., about 5 h) is necessary. When compared with a step-by-step charging
duration (e.g., about 5 h) is necessary. When compared with a step-by-step charging and discharging
and discharging protocol (ΔSoC = 5% with 3 h of rest at each step) to determine the OCV (SoC) curve
protocol (∆SoC = 5% with 3 h of rest at each step) to determine the OCV (SoC) curve like in [1],
like in [1], the present method has the following advantages: besides giving the OCV (SoC) curve it
the present method has the following advantages: besides giving the OCV (SoC) curve it enables us to
enables us to compute the variation of the output voltage during relaxation. Furthermore, if one
compute the variation of the output voltage during relaxation. Furthermore, if one wants to estimate
wants to estimate 24 h of relaxation with a step-by-step protocol, it could last 40 × 24 h (20 SoC levels
24 h of relaxation with a step-by-step protocol, it could last 40 ˆ 24 h (20 SoC levels during charging
during charging and 20 SOC levels during discharging), which represents 40 days of experiments.
and 20 SOC levels during discharging), which represents 40 days of experiments. With our method,
With our method, only one measurement of 24 h is necessary plus 40 measurements of 5 h, which
only one measurement of 24 h is necessary plus 40 measurements of 5 h, which represents only nine
represents only nine days of works. This method could save a lot of characterization time, which is
days of works. This method could save a lot of characterization time, which is very useful when tests
very useful when tests are required at different temperatures and aging states. When the method is
are required at different temperatures and aging states. When the method is applied for relaxation
applied for relaxation after charge and discharge, the OCVs obtained could also bring out the
after charge and discharge, the OCVs obtained could also bring out the hysteresis. Furthermore, it is
hysteresis. Furthermore, it is generic so it could easily be extended to any technology of battery and
generic so it could easily be extended to any technology of battery and adapted to longer or shorter
adapted to longer or shorter relaxation durations.
relaxation durations.
This method could be applied to calculate the SoC of batteries in EVs from measurements of
This method could be applied to calculate the SoC of batteries in EVs from measurements of OCV.
OCV. In that case, the curve OCV (SoC) stored in the on-board computer and used to calculate the
In that case, the curve OCV (SoC) stored in the on-board computer and used to calculate the SoC is
SoC is usually determined after a specified relaxation time (e.g., 24 h). The proposed model of
usually determined after a specified relaxation time (e.g., 24 h). The proposed model of relaxation
relaxation voltage can be used to accurately predict the OCV and thus the SoC at the specified time
voltage can be used to accurately predict the OCV and thus the SoC at the specified time without
without waiting for it. On the other hand, when the curve OCV (SoC) stored in the on-board
waiting for it. On the other hand, when the curve OCV (SoC) stored in the on-board computer is
computer is determined after a short relaxation time (e.g., 1 h), the proposed model of relaxation
determined after a short relaxation time (e.g., 1 h), the proposed model of relaxation voltage can be
voltage can be used to go back in time. For example, when the vehicle has been parked for a longer
time (e.g., 24 h), the value of OCV at 1 h, and thus the SoC can be determined from the measured
value of OCV at 24 h.
Batteries 2016, 2, 7 15 of 15

used to go back in time. For example, when the vehicle has been parked for a longer time (e.g., 24 h),
the value of OCV at 1 h, and thus the SoC can be determined from the measured value of OCV at 24 h.

Acknowledgments: This research was supported by the National French Association for Technological
Research (ANRT).
Author Contributions: An Li, Serge Pelissier, Pascal Venet and Philippe Gyan conceived and designed the
experiments; An Li performed the experiments; An Li, Serge Pelissier, Pascal Venet and Philippe Gyan analyzed
the data; An Li, Serge Pelissier and Pascal Venet wrote the paper.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Roscher, M.A.; Sauer, D.U. Dynamic electric behavior and open-circuit-voltage modeling of LiFePO4 -based
lithium ion secondary batteries. J. Power Sources 2011, 196, 331–336. [CrossRef]
2. Abu-Sharkh, S.; Doerffel, D. Rapid test and non-linear model characterization of solid-state lithium-ion
batteries. J. Power Sources 2004, 130, 266–274. [CrossRef]
3. Bergveld, H.; Pop, V.; Notten, P.H.L. Apparatus and Method for Determination of the State of Charge of
a Battery When the Battery Is Not in Equilibrium. U.S. Patent 20100036627 A1, 11 February 2010.
4. Driemeyer-Franco, A.; Lucea, M.; Saint-Marcoux, A.; Bui-Van, A.-L.; Perichon, P. Système et
Procédé Correspondant D’estimation de L’état de Charge d’une Batterie. Patent WO 2013171425 A1,
21 November 2013. (In French)
5. Pei, L.; Wang, T.; Lu, R.; Zhu, C. Development of a voltage relaxation model for rapid open-circuit voltage
prediction in lithium-ion batteries. J. Power Sources 2014, 253, 412–418. [CrossRef]
6. Petzl, M.; Danzer, M.A. Advancements in OCV measurement and analysis for lithium-ion batteries.
IEEE Trans. Energy Convers. 2013, 28, 675–681. [CrossRef]
7. Baronti, F.; Zamboni, W.; Roncella, R.; Saletti, R.; Spagnuolo, G. Open-Circuit Voltage Measurement of
Lithium-Iron-Phosphate Batteries. In Proceedings of the IEEE Instrumentation and Measurement Technology
Conference (I2MTC), Taipei, Taiwan, 23–26 May 2015; pp. 1711–1716.
8. Chaturvedi, N.A.; Klein, R.; Christensen, J.; Ahmed, J.; Kojic, A. Algorithms for advanced
battery-management systems. IEEE Control Syst. Mag. 2010, 30, 49–68. [CrossRef]
9. Urbain, M.; Hinaje, M.; Raël, S.; Davat, B.; Desprez, P. Energetical modeling of lithium ion batteries including
electrode porosity effects. IEEE Trans. Energy Convers. 2010, 25, 862–872. [CrossRef]
10. Chen, M.; Rincon-Mora, G.A. Accurate Electrical Battery Model Capable of predicting runtime and I-V
performance. IEEE Trans. Energy Convers. 2006, 21, 504–511. [CrossRef]
11. Sarvi, M.; Masoum, M.A.S. A Neural Network Model for Ni-Cd Batteries. In Proceedings of the 43rd
International Universities Power Engineering Conference (UPEC), Padova, Italy, 1–4 September 2008;
pp. 1–5.
12. A123 Systems. Li-Ion 26650 Cylindrical Cell, 2011. Available online: http://www.a123systems.com (accessed
on 14 May 2011).
13. Zhang, J.; Ci, S.; Sharif, H.; Alahmad, M. Modeling discharge behavior of multicell battery. IEEE Trans.
Energy Convers. 2010, 25, 1133–1141. [CrossRef]
14. Kroeze, R.; Krein, P.T. Electrical Battery Model for Use in Dynamic Electric Vehicle Simulation. In Proceedings
of the IEEE Power Electronics Specialists Conference, Rhodes, Greece, 15–19 June 2008; pp. 1336–1342.
15. Hu, Y.; Yurkovich, S.; Guezennec, Y.; Yurkovich, B.J. Electro-thermal battery model identification for
automotive applications. J. Power Sources 2011, 196, 449–457. [CrossRef]
16. Kokam. Li-Ion SLB603870H Cell, 2011. Available online: http://www.kokam.com (accessed on 14 May 2011).
17. Saft. NiMH VHCS3500 Cell, 2011. Available online: http://www.saftbatteries.com (accessed on
14 May 2011).

© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons by Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).