applied

sciences
Article
Comparative Research on RC Equivalent Circuit
Models for Lithium-Ion Batteries of Electric Vehicles
Lijun Zhang 1, * ID
, Hui Peng 1 , Zhansheng Ning 1 , Zhongqiang Mu 1 and Changyan Sun 2
1 National Center for Materials Service Safety, University of Science and Technology Beijing,
Beijing 100083, China; peng111hui@126.com (H.P.); ustb41140007@163.com (Z.N.);
s20161186@xs.ustb.edu.cn (Z.M.)
2 School of Chemistry and Biological Engineering, University of Science and Technology Beijing,
Beijing 100083, China; cysun@ustb.edu.cn
* Correspondence: ljzhang@ustb.edu.cn; Tel.: +86-10-6232-1017

Received: 18 August 2017; Accepted: 25 September 2017; Published: 28 September 2017

Abstract: Equivalent circuit models are a hot research topic in the field of lithium-ion batteries for
electric vehicles, and scholars have proposed a variety of equivalent circuit models, from simple
to complex. On one hand, a simple model cannot simulate the dynamic characteristics of batteries;
on the other hand, it is difficult to apply a complex model to a real-time system. At present,
there are few systematic comparative studies on equivalent circuit models of lithium-ion batteries.
The representative first-order resistor-capacitor (RC) model and second-order RC model commonly
used in the literature are studied comparatively in this paper. Firstly, the parameters of the two
models are identified experimentally; secondly, the simulation model is built in Matlab/Simulink
environment, and finally the output precision of these two models is verified by the actual data.
The results show that in the constant current condition, the maximum error of the first-order RC model
is 1.65% and the maximum error for the second-order RC model is 1.22%. In urban dynamometer
driving schedule (UDDS) condition, the maximum error of the first-order RC model is 1.88%, and for
the second-order RC model the maximum error is 1.69%. This is of great instructional significance
to the application in practical battery management systems for the equivalent circuit model of
lithium-ion batteries of electric vehicles.

Keywords: lithium-ion batteries; equivalent circuit model; electric vehicles; parameter identification;
battery management system

1. Introduction
Lithium-ion batteries have many advantages, such as high working voltage, small volume,
light weight, long cycle life, environmental friendliness, weak memory effect, wide working
temperature range, low self-discharge rate, and so on [1–4]. Lithium-ion batteries have been widely
used in mobile phones, digital cameras, laptops, other portable consumer electronics, and renewable
energy storage [5,6]. With the continuous development and improvement of science and technology,
the application range of lithium-ion batteries has further widened to the electric power, aerospace,
military, and other critical fields [7,8]. Lithium-ion batteries have become core components of energy
supply for many critical devices or systems, and are often critical to the reliability and functionality of
the overall system [3,8]. However, consequences of battery failure could range from inconvenience to
catastrophic failure; for example, NASA’s Mars Global Surveyor stopped running due to battery failure
in November 2006. The reliability of lithium-ion battery systems has yet to be improved. Recently,
lithium-ion batteries have reached a degree of implementation that enabled their use in stringent
automotive applications; for example, the Nissan LEAF rolled off the assembly line in 2010 as well
as the Tesla Roadster in 2008 [9]. Automotive lithium-ion batteries usually consist of hundreds or

Appl. Sci. 2017, 7, 1002; doi:10.3390/app7101002 www.mdpi.com/journal/applsci

Appl. Sci. 2017, 7, 1002 2 of 16

even thousands of single cells connected in series and parallel so they have high power and energy
density coupled with such problems as safety, reliability, and uniformity. At present, an efficient
and reliable battery management system (BMS) has become the key to guaranteeing the reliable and
safe operation of lithium-ion batteries [10,11]; there are many aspects to the process of developing a
battery management system, such as requirement analysis, modeling and simulation, control strategy
research, and on-line hardware test, which all need a capable model to identify the characteristics of
the lithium-ion batteries.
In order to predict the behavior of the battery, many different models have been established,
which can be generally divided into two categories: electrochemical models and equivalent circuit
models. Although the electrochemical model accounts for the main underlying physics in the
battery dynamics in terms of the main electrochemical parameters and variables [12,13], the on-site
accuracy is not high when it is applied in vehicle control systems due to the complexity of the
model, and a large number of design and electrochemical kinetic parameters need to be obtained
first. Therefore, the electrochemical mechanism model is usually used to understand the reaction
process inside the battery, and to guide the design and manufacture of the battery. In an equivalent
circuit model (ECM), resistances, capacitances, and voltage sources are used to describe charging
and discharging processes, and the model is built in frequency- or time-domain [14,15], in which
each component has clear physical meaning, simple mathematical expressions, is more intuitive,
and is easy to handle. It has been widely applied in the estimation of the state-of-charge (SOC),
the state-of-health (SOH) and the state-of-function (SOF) for lithium-ion batteries. The literature [16]
based on the first-order RC (resistor-capacitor) model of lithium-ion battery short-circuit detection was
explored. The literature [17] used a first-order RC equivalent circuit model to estimate the resistance of
lithium-ion batteries and the open circuit voltage. Reference [18] used the first-order RC equivalent
circuit model for the modeling and simulation of lithium-ion battery packs. A first-order RC equivalent
circuit model was created and validated for a particular battery using a MATLAB (the MathWork,
Inc., Natick, MA, USA) Simulink in terms of the internal temperature distribution, the open circuit
voltage, the heat generation, and the internal resistance in References [19,20]. A SOC estimation
method based on the first-order RC equivalent circuit model of lithium-ion batteries was studied
in [21–23]. The studies in [24–26] explored a SOC estimation method based on the second-order RC
equivalent circuit model of lithium-ion batteries. References [26–28] studied SOH estimation based
on second-order RC equivalent circuit models of a lithium-ion battery. Additionally, Reference [29]
used the equivalent circuit model of a second-order RC model for lithium-ion battery modeling. It can
be seen that the SOC and SOH estimation algorithms based on the equivalent circuit model are the
most important class of estimation algorithms in battery management systems. Many scholars have
proposed a variety of equivalent circuit models, from simple to complex. However, there are few
systematic comparative studies on equivalent circuit models of lithium-ion batteries, and simple
models are not able to adequately simulate dynamic characteristics of batteries, while it is difficult
to apply complex ones to real-time systems. The choice between these models is a trade-off among
modeling complexity, precision, and computational cost. Therefore, the first-order RC model and
the second-order RC model commonly used in the literature are studied comparatively in this paper
because of their simplicity and relative accuracy. In this paper, firstly, the parameters of the above two
models are identified by experimental methods individually, and then the simulation models are built
in the Matlab/Simulink environment, and finally the precision of results for the two models is verified
and discussed comparatively with the actual data.

2. Experiments and Methods

2.1. ECM for Lithium-Ion Batteries

Equivalent circuit models use a circuit consisting of voltage sources, resistors, and capacitors
to simulate the dynamic characteristics of batteries [14,15], thus describing the relationship between
Appl. Sci. 2017, 7, 1002 3 of 16

voltage and current shown in the battery operation. In this paper, a first-order RC model3 of
Appl. Sci. 2017, 7, 1002
and 16
a
second-order
Appl. Sci. 2017,RC model are comparatively researched, as they refer mostly to the equivalent
7, 1002 3 ofcircuit
16

models
models used
used ininthetheliterature.
literature.On Onone onehand,
hand, thethe two kinds
kinds of ofmodels
modelsdo doaagood
goodjob jobofofreflecting
reflecting thethe
models used in the literature. On one hand, the two kinds of models do a good job of reflecting the
dynamic
dynamic and andstatic
staticcharacteristics
characteristics of oflithium-ion
lithium-ion batteries,
batteries,and
andononthe theother
otherhand
handthe thecomplexity
complexityofofthe
dynamic and static characteristics of lithium-ion batteries, and on the other hand the complexity of
models is also is
the models appropriate,
also appropriate, easy toeasy implement in engineering,
to implement and it isand
in engineering, easyit tois implement parameter
easy to implement
the models is also appropriate, easy to implement in engineering, and it is easy to implement
parameter identification
identification with high precision.
parameter with high precision.
identification with high precision.
AAschematic
A schematic diagram
diagram
schematic diagram ofof
ofthe
the first-order
thefirst-order
first-order RC RC
RC model
model isis shown
isshown
shownin in Figure
inFigure
Figure 1a,
1a,1a,where
where
where the
thethe controllable
controllable
controllable
voltage
voltage VOC V
voltage V denotes
denotes
OC
OC denotes
the
the open open circuit
circuitcircuit
the open voltage voltage of the lithium-ion
of the lithium-ion
voltage battery
battery battery
of the lithium-ion which usuallywhich
which varies usually varies
usuallynonlinearly
varies
nonlinearly
withnonlinearly
SOC, R0 denoteswith SOC,
with SOC, R
theRohmic
0 denotes
0 denotes
the ohmic
resistance
the ohmic resistance
ofresistance of the
the lithium-ion lithium-ion
battery which
of the lithium-ion battery
battery which
describes describes the
the electrolyte
which describes the
electrolyte
resistance and
electrolyte resistance
connection
resistance and connection
andresistance
connectionof resistance
the battery,
resistance of the battery,
R1battery,
of the denotes RR11the
denotes thepolarization
polarization
polarization
denotes the resistance,
resistance, C1C1
C1 denotes
resistance,
denotes
denotes polarization
polarization capacitance,
capacitance, II denotes
denotes the
the current
current flowing
flowing through
through
polarization capacitance, I denotes the current flowing through the load which can be directly measured the
the load
load which
which can
can bebe directly
directly
measured
from currentfrom
measured fromcurrent
sensor, current
and U sensor,
sensor,
denotesand
andthe U
U denotes
denotes
terminal the
the terminal
terminal
voltage voltage
ofvoltage
the battery ofthe
of the battery
battery
which which
which
can canbebedirectly
can
be directly directly
measured
measured from a voltage sensor. The parallel RC network describes the nonlinear polarization
frommeasured
a voltagefrom sensor. a voltage
The parallelsensor.RC The paralleldescribes
network RC network describes the
the nonlinear nonlinear response
polarization polarization of the
response
response ofofthe
the lithium-ion
lithium-ion battery,
battery, and
and II is
is positive for discharging
dischargingand andnegative
negativefor forcharging.
charging.
lithium-ion battery, and I is positive for discharging and negative for charging.
The
Theschematic
schematicdiagramdiagramof of the
the second-order
second-order RC model model is is shown
shownin inFigure
Figure1b, 1b,where
whereVV ,R
OCOC ,R , I,
0, 0I,
The schematic diagram of the second-order RC model is shown in Figure 1b, where VOC , R0 , I,
and
andUUhave havethe thesame
samemeanings
meaningsas as in
in the
the first-order model, except
first-order RC model, exceptthatthatthere
thereare aretwo
twoparallel
parallelRC RC
and U have the same meanings as in the first-order RC model, except that there are two parallel RC
networks
networkstotodescribe
describethe thenonlinear
nonlinear polarization
polarization response
response of of the
the lithium-ion
lithium-ionbattery.
battery.
networks to describe the nonlinear polarization response of the lithium-ion battery.

Figure 1. The first-order

1. The resistor–capacitor
(RC) (a)(RC) (a) and second-order RC (b) circuit
equivalent
Figure
Figure 1. first-order resistor–capacitor
The first-order resistor–capacitor and second-order
(RC) RC (b) equivalent
(a) and second-order models.
RC (b) equivalent
circuit models.
circuit models.
2.2. Battery Testing Bench
2.2. Battery Testing Bench
2.2. Battery Testing Bench
To complete
To complete thethe
model
modelparameter
parameter identification, the testing
identification, the testingbench
benchshown
shownin in Figure
Figure 2 was
2 was
To
established. complete
It is the
composedmodel parameter identification, the testing bench shown in Figure 2 was
established. It is composedofofaaNEWARE
NEWARE (the NEWARECo.,
(the NEWARE Co.,Limited,
Limited,Shenzhen,
Shenzhen, China)
China) battery
battery
established.
tester system It is (BTS4000)
4000 composed with
of a NEWARE
eight (the NEWARE
independent Co., an
channels, Limited, Shenzhen,
intermediate China)and
machine, battery
a host
tester system 4000 (BTS4000) with eight independent channels, an intermediate machine, and a host
tester system
computer on 4000 NEWARE
which (BTS4000) BTS
with Software
eight independent
v7.5.6 andchannels,
MATLAB anR2010a
intermediate
are machine, and a host
installed.
computer on which NEWARE BTS Software v7.5.6 and MATLAB R2010a are installed.
computer on which NEWARE BTS Software v7.5.6 and MATLAB R2010a are installed.

Figure 2. NEWARE BTS4000 battery testing bench.

Figure 2. NEWARE BTS4000 battery testing bench.
Figure 2. NEWARE BTS4000 battery testing bench.
Appl. Sci. 2017, 7, 1002 4 of 16

The battery tester is used to load the programmed current profiles on the battery cells within a
range of voltage 0–10 V and current −6 to 6 A, and the measuring errors of its voltage and current
sensors are both less than 0.5%. The intermediate machine is mainly responsible for the network
connection, receiving the control command from the host computer, controlling the lower battery
cycler, and uploading the real-time test data. The host computer is employed to control and monitor
the cycler via Ethernet cables, as well as to store the voltage and current acquired by the sensors.
The tested battery cells are representative 18650 cylindrical cells whose specifications mainly include
2350 mAh nominal capacity, 3.7 V operating voltage, 4.2 V charging limit voltage, 2.7 V discharge limit
voltage, and 10 A maximum continuous discharging current. During the test, the battery current and
the voltage were recorded every second, while they were recorded every 100 milliseconds during the
hybrid pulse power characterization (HPPC) test. The important parameters of the experiment are
summarized in Table 1.

Table 1. The important parameters of the experimental scheme.

Sampling Frequency
Cell Cell Maximum
BTS4000 BTS4000 BTS4000 Cell Limit
Nominal Continuous Constant
Voltage Current Measuring Voltage HPPC
Capacity Discharging Current
Range (V) Range (A) Errors (V) Test (Hz)
(mAh) Current (A) Discharge (Hz)
0–10 −6 to 6 0.5% 2350 10 2.7–4.2 1 10

2.3. Model Parameter Identification

2.3.1. Open Circuit Voltage

The open circuit voltage (OCV) of the battery is the stable voltage value of the battery when the
battery is left in the open circuit condition [11]. Regarding the battery after being charged, the battery
terminal voltage will gradually decline to a stable value when it is left in the open circuit condition;
regarding the battery after discharge, the battery terminal voltage will gradually rise to a stable value
after the load is removed. The electromotive force of the battery is basically equal to the open circuit
voltage of the battery, while the battery electromotive force is one of the metrics used to measure the
amount of energy stored in the battery. Thus, there is a certain relationship between the battery OCV
and the battery SOC [22]. There are a few ways to obtain OCV, in which the stationary method is a
direct method and is relatively more accurate. To obtain the relationship between the battery OCV and
SOC in the stationary method, the test procedures are performed as follows [30]:

(1) Calibrate the battery capacity. The battery is fully charged with the standard charging method,
in which the battery is charged using a constant current phase of 2.35 A (1C) to 4.2 V followed by a
constant voltage phase of 4.2 V until the current is reduced to 0.02 A. Then, the battery is discharged
with a constant current phase of 2.35 A to its discharge cut-off voltage 2.7 V. The experiment is
repeated until the difference between the discharge capacity of each measurement does not exceed
2%, and then the measured capacity is deemed to be the actual capacity of the battery.
(2) The battery is fully charged with the standard charging method described by Step (1) and then the battery is
left in the open-circuit condition to rest for 4 h to achieve electrochemical and heat equilibrium [9].
(3) The battery is discharged with a constant current 2.35 A for 6 minutes (i.e., discharging by 10% of
the capacity) and then the battery is left in the open-circuit condition to rest for 4 h to achieve
electrochemical and heat equilibrium.
(4) Step (3) is repeated eight times.
(5) The battery is discharged with a constant current 2.35 A by 3 min (i.e., discharging by 10% of the
capacity) and then the battery is left in the open-circuit condition for 4 h to achieve electrochemical
and heat equilibrium.
(6) Step (5) is repeated four times.
 Appl.Sci.
Appl. Sci.2017,
2017,7,7,1002
1002 5 5ofof1616

In the tests, the measured voltages at the end of each standby stage are regarded as the final
In the tests, the measured voltages at the end of each standby stage are regarded as the final
open-circuit voltage. The variation of OCV with SOC obtained by the experimental method above is
open-circuit voltage. The variation of OCV with SOC obtained by the experimental method above is
shown in Figure 3.
shown in Figure 3.

4.2

4
OCV/V

3.8

3.6

3.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
SOC
SOC
Figure 3. The varying curve of open circuit voltage (OCV) with state-of-charge (SOC).
Figure 3. The varying curve of open circuit voltage (OCV) with state-of-charge (SOC).

2.3.2. Ohmic Resistance

2.3.2. Ohmic Resistance
In general, the methods used to measure the battery resistance are the power-off method, the
In general, the methods used to measure the battery resistance are the power-off method, the step
step method, and the electrochemical impedance spectroscopy (EIS) method. With the EIS method,
method, and the electrochemical impedance spectroscopy (EIS) method. With the EIS method,
the battery is excited by a series of alternating current (AC) signals of different frequency and then
the battery is excited by a series of alternating current (AC) signals of different frequency and then
through analyzing the relation between the input excitation signal and the battery voltage output
through analyzing the relation between the input excitation signal and the battery voltage output
response, and the battery impedance characteristic can be obtained. The principle of the power-off
response, and the battery impedance characteristic can be obtained. The principle of the power-off
method and the step method is similar, in which the difference between ohm polarization response
method and the step method is similar, in which the difference between ohm polarization response time
time and other polarization response time is utilized to identify the different impedance. When the
and other polarization response time is utilized to identify the different impedance. When the current
current has a sudden change, the voltage drop part caused by ohmic polarization changes
has a sudden change, the voltage drop part caused by ohmic polarization changes instantaneously and
instantaneously and then the voltage drop part by other polarization completes the transient process
then the voltage drop part by other polarization completes the transient process with the approximation
with the approximation exponential until the voltage returns to a steady-state voltage value.
exponential until the voltage returns to a steady-state voltage value. Assuming that the voltage before
Assuming that the voltage before the current changes is U0 and the voltage after the moment the
the current changes is U0 and the voltage after the moment the current changes is U1 , the formula for
current changes is U1, the formula for calculating the ohmic resistance is as follows:
calculating the ohmic resistance is as follows:
U −U
R 0 (SOC) = 1 0
(1)
U1 − ΔIU0
R0 (SOC) = (1)
∆I
Referring to the “Freedom CAR Battery Test Manual” hybrid pulse power characterization test
(HPPC test) [31],
Referring this“Freedom
to the paper adopts
CARthe step Test
Battery method to obtain
Manual” the pulse
hybrid ohmicpower
resistance, the polarization
characterization test
resistance,
(HPPC and this
test) [31], polarization capacitance
paper adopts from the
the step method different
to obtain SOCs. resistance,
the ohmic The hybrid the pulse power
polarization
characterization
resistance, test profile is capacitance
and polarization shown in Figure from4. the different SOCs. The hybrid pulse power
To obtain the
characterization testbattery
profile ohmic
is shownresistance,
in Figurepolarization
4. resistance, and polarization capacitance at
the different
To obtainSOCs, tests ohmic
the battery are performed as polarization
resistance, follows: resistance, and polarization capacitance at the
different SOCs, tests are performed as follows:
(1) Calibrate the battery capacity as described in Section 2.3.1.
(2) Calibrate
(1) Charge the battery capacity
the battery fully with the standard
as described charging
in Section 2.3.1. method the same as described in
Sectionthe
(2) Charge 2.2.battery fully with the standard charging method the same as described in Section 2.2.
(3) The HPPC
(3) The HPPC test testisisperformed
performedandandthen
thenthethebattery
batteryisisleft
leftin
inthe
theopen-circuit
open-circuitcondition
conditiontotorest
restfor
for
4 h to achieve electrochemical and heat equilibrium
4 h to achieve electrochemical and heat equilibrium [9]. [9].
(4)
(4) TheThebattery
batteryisisdischarged
dischargedwithwith a constant
a constant current
current 2.352.35 A for
A for 6 minutes
6 minutes andand
thenthen the battery
the battery is leftis
left in the open-circuit condition to rest for 4 h to achieve electrochemical and
in the open-circuit condition to rest for 4 h to achieve electrochemical and heat equilibrium. heat equilibrium.
(5) Steps (3) and (4) are respectively repeated eight times alternately.
(5) Steps (3) and (4) are respectively repeated eight times alternately.
(6) The battery is discharged with a constant current 2.35 A for 3 min and then the battery is left in
(6) The battery is discharged with a constant current 2.35 A for 3 min and then the battery is left in
the open-circuit condition to rest for 4 h to achieve electrochemical and heat equilibrium.
the open-circuit condition to rest for 4 h to achieve electrochemical and heat equilibrium.
(7) The HPPC test is performed and then the battery is left in the open-circuit condition to rest for
4 h to achieve electrochemical and heat equilibrium.
Appl. Sci. 2017, 7, 1002 6 of 16

Appl. The
(7) Sci. 2017,
HPPC 7, 1002
test is performed and then the battery is left in the open-circuit condition to rest6 of
for16
Appl. Sci. 2017, 7, 1002 6 of 16
4 h to achieve electrochemical and heat equilibrium.
(8) Steps (6) and (7) are respectively repeated four times alternately.
(8) Steps (6)
(8) Steps (6) and
and (7)
(7) are
are respectively
respectively repeated
repeated four
four times
times alternately.
alternately.

2
2
1
1
Current/A

0
Current/A

0
-1
-1
-2
-2
-3
-3 0 10 20 30 40 50 60
0 10 20 Time/s
30 40 50 60
Time/s
Figure 4. Hybrid pulse power characterization (HPPC) test current profile.
Figure 4.
Figure 4. Hybrid
Hybrid pulse
pulse power
power characterization
characterization (HPPC)
(HPPC) test
test current
current profile.
profile.
From the above experimental steps, it can be seen that the HPPC test can be implemented along
From
withFrom the above
the open experimental
circuit steps,
voltage teststeps, it can beSection
described seen that the According
HPPC test can be implemented along
the above experimental it canbybe seen that 2.3.1.
the HPPC to the
test can above experimental
be implemented along
with the open
procedures, circuit voltage test described by Section 2.3.1. According to the above experimental
with the openthe voltage
circuit response
voltage curves of lithium-ion
test described by Section battery excited byto
2.3.1. According HPPC test atexperimental
the above different SOC
procedures,
can be the voltage
obtained. The response
results are curves in
shown of Figure
lithium-ion
5. battery
Then, the excitedresistance
ohmic by HPPCat test at different
different SOCsSOC
can
procedures, the voltage response curves of lithium-ion battery excited by HPPC test at different SOC
can be obtained.
be obtained The results
by calculating are shown in Figure 5. Then, the ohmic resistance at different SOCs can
can be obtained. The results Equation
are shown(1). The results
in Figure are the
5. Then, shown in Figure
ohmic 6. at different SOCs can be
resistance
be obtained by calculating Equation (1). The results are shown in Figure 6.
obtained by calculating Equation (1). The results are shown in Figure 6.

SOC=0.9
4.6 SOC=0.9
SOC=0.6
4.6 SOC=0.6
SOC=0.4
4.4
4.4 SOC=0.4
SOC=0.2
Voltage/V

4.2 SOC=0.2
SOC=0.05
Voltage/V

4.2 SOC=0.05
4
4
3.8
3.8
3.6
3.6
3.4
3.4
3.2
3.2
0 10 20 30 40 50 60
0 10 20 30
SOC 40 50 60
SOC
Figure5.5.Voltage
Figure Voltageresponse
responsecurves
curvesby
byHPPC
HPPCtest
testatatdifferent
differentSOC.
SOC.
Figure 5. Voltage response curves by HPPC test at different SOC.
Appl.
Appl.Sci. 2017,7,7,1002
Sci.2017, 1002 7 7ofof1616

80

Ohmic resistance/m 70

60

50

40
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
SOC
SOC
Figure 6. The varying curve of ohmic resistance with SOC.
Figure 6. The varying curve of ohmic resistance with SOC.

2.3.3. Polarization Resistance and Polarization Capacitance

2.3.3. Polarization Resistance and Polarization Capacitance
As discussed in Section 3.3.2, for the polarization process the terminal voltage completes the
As discussed in Section 2.3.2, for the polarization process the terminal voltage completes the
transient process in an exponential way approximately when the current sees a sudden change.
transient process in an exponential way approximately when the current sees a sudden change.
According to the Kirchhoff laws, for the first-order RC model the variation law of the terminal
According to the Kirchhoff laws, for the first-order RC model the variation law of the terminal voltage
voltage can be ruled by the following formula:
can be ruled by the following formula:
U(t) = VOC  IR1e  t/τ (2)
U(t) = VOC − IR1 e−t/τ (2)
where τ = R 1C 1 represents the time constant.
τ = R1 C1 the
whereReplacing represents the time
coefficients constant.(2) and rewriting the form as:
of Equation
Replacing the coefficients of Equation (2) and rewriting the form as:
U(t) = VOC  ae  bt (3)
U(t) = VOC − ae−bt (3)
Compared to the two formulas above, the parameter values can be obtained as follows:
Compared to the two formulas above, the R
parameter
1 = a / I
values can be obtained as follows: (4)

R = a/I (4)
C1 1= 1/  R1b  (5)
C1 = 1/(R1 b) (5)
According to the method described above, the polarization resistance and polarization capacitance
of the first-order
According toRC
themodel
methodcan be obtained
described by the
above, the least square method
polarization as shown
resistance in Table 2. capacitance
and polarization
of the first-order RC model can be obtained by the least square method as shown in Table 2.
Table 2. Polarization resistance and polarization capacitance values of lithium-ion batteries for the
first-order
Table RC model.resistance and polarization capacitance values of lithium-ion batteries for the
2. Polarization
first-order RC model.
SOC τ1/s R1/Ω C1/103 F
0 56.2430 0.0382 1.4743 3
SOC τ1 /s R1 /Ω C1 /10 F
0.05 42.8082 0.0263 1.6265
0 56.2430 0.0382 1.4743
0.05
0.1 51.3875
42.8082
0.0226
0.0263
2.2758
1.6265
0.1 0.15 47.5964
51.3875 0.0244
0.0226 1.9491
2.2758
0.15 0.2 47.5964
55.9597 0.0244
0.0237 1.9491
2.3622
0.2 55.9597 0.0237 2.3622
0.3 34.7826 0.0203 1.7126
0.3 34.7826 0.0203 1.7126
0.4 0.4 35.8938
35.8938 0.0204
0.0204 1.7612
1.7612
0.5 0.5 41.9287
41.9287 0.0211
0.0211 1.9919
1.9919
0.6 0.6 37.5657
37.5657 0.0267
0.0267 1.4049
1.4049
0.7 40.6504 0.0242 1.6798
0.8 0.7 40.6504
38.5654 0.0242
0.0272 1.6798
1.4178
0.9 0.8 38.5654
37.3134 0.0272
0.0235 1.4178
1.5878
1 0.9 46.2321
37.3134 0.0240
0.0235 1.9287
1.5878
1 46.2321 0.0240 1.9287
Appl. Sci. 2017, 7, 1002 8 of 16

Similarly, for the second-order RC model, the variation law of the terminal voltage can be ruled
by the following equation:
U(t) = VOC − IR1 e−t/τ1 − IR2 e−t/τ2 (6)

where τ1 = R1 C1 , τ2 = R2 C2 represent the time constant.

Replacing the coefficients of Equation (6) and rewriting the equation as:

U(t) = VOC − ae−bt − ce−dt (7)

Compared to coefficients of the two equations above, the parameter values can be obtained
as follows:
R1 = a/I (8)

R2 = c/I (9)

C1 = 1/(R1 b) (10)

C2 = 1/(R2 d) (11)

According to the method described above, then the polarization resistance and polarization
capacitance of the second-order RC model can be obtained by the least square method as shown
in Table 3.

Table 3. The polarization resistance and polarization capacitance values of lithium-ion batteries for the
second-order RC model.

SOC τ1 /s R1 /Ω C1 /103 F τ2 /s R2 /Ω C2 /104 F

0 1.4769 0.0334 0.0442 51.5996 0.0169 0.3044
0.05 5.5741 0.0051 1.0871 62.0732 0.0091 0.6801
0.1 5.2938 0.0041 1.2881 66.0502 0.0085 0.7804
0.15 7.8370 0.0043 1.8020 81.9001 0.0079 1.0314
0.2 5.3763 0.0040 1.3375 69.9301 0.0091 0.7719
0.3 18.7512 0.0072 2.6151 86.1204 0.0023 3.5084
0.4 15.5497 0.0045 3.4769 69.7350 0.0048 1.4490
0.5 2.9833 0.0025 1.1805 45.3309 0.0086 0.5300
0.6 7.1276 0.0047 1.5090 56.7859 0.0087 0.6500
0.7 6.7249 0.0052 1.2954 60.4230 0.0102 0.5897
0.8 4.6404 0.0047 0.9950 47.1254 0.0102 0.4634
0.9 5.3220 0.0049 1.0819 89.7116 0.0433 0.8196
1 10.8684 0.0043 2.5064 93.8967 0.0070 1.3357

3. Model Simulation
After obtaining the required model parameters, the simulation model of the lithium-ion battery
can be established in Matlab/Simulink. The simulations of the established first-order RC and the
second-order RC models are shown in Figures 7 and 8, respectively. It can be seen that the simulation
model is mainly composed of three sub-modules: the SOC calculation module, the circuit parameter
updating module, and the terminal output voltage calculation module. The specific sub-modules are
shown in Figures 9 and 10. Figure 9 shows the SOC calculation module based on current time integral
method, the discharge capacity is obtained by integrating the discharge current with time, and then
the residual capacity could be obtained by using the nominal capacity minus the discharge capacity.
Finally, the SOC value can be obtained by the standardized calculation. Figure 10 shows the terminal
output voltage calculation module based on Kirchhoff laws, and the output voltage can be obtained
by the voltage difference between the OCV and the resistor capacitor. The simulation model has two
inputs: the load current profile and the initial SOC of the battery. The load current profile is input
through the program, so the load current can be set to any form. The maximum value of SOC is set
Appl. Sci. 2017, 7, 1002 9 of 16
simulation model
Appl. Sci. 2017, 7, 1002has two inputs: the load current profile and the initial SOC of the battery. The load 9 of 16
voltage profile
current can be obtained
is input by the voltage
through difference
the program, sobetween
the loadthe OCV and
current can the resistor
be set to anycapacitor.
form. The The
voltage can
simulation
maximum be obtained
model
value ofhas
SOCtwoisby the
set tovoltage
inputs: the
1, load
and difference
thecurrent
minimum between
profile andthe
value 0,OCV
isthe so asand
initial SOC theofresistor
to preventthe the capacitor.
battery.
batteryThefromThe
load
simulation
current profile
overcharge model has two
andisover-discharge. inputs:
input through the the load current
program,
Because of the profile
so relativelyand the
the load current initial SOC
can be
short cycle of the
set to
time, battery.
theany The
form. The
influence load
of
current
Appl.
maximum
temperature profile
Sci. 2017,value1002is
7,change of input
SOC isthrough
on the set to 1,the
output andprogram,
voltagethe of
minimumso thevalue
a lithium-ion loadbattery
iscurrent ascan
0, so is be setFinally,
to prevent
ignored. to
theany form.
battery
the ofThe
9from
actual16
maximum
overcharge
voltage curvevalue
and of SOC is with
over-discharge.
is compared set tothe
1, and
Because theofminimum
simulation the
curve value
the is
relatively
of 0, so cycle
short
model as verify
to to prevent
time,thethe theinfluence
battery
accuracy of from
of
the
overcharge
temperature
model. The and over-discharge.
changeresults
onvalue
thewilloutput Because
voltage of
of the relatively
a lithium-ion short
battery cycle time,
is ignored.and the
Finally, influence
the actual of
to 1, and thedetailed
minimum is be discussed
0, so in the
as to prevent nextbattery
the section. from overcharge over-discharge.
temperature
voltage curve change on thewith
is compared outputthe voltage
simulationof a curve
lithium-ion the battery is ignored. Finally, theof actual
Because of the relatively short cycle time, the influence ofoftemperature
model to verifyonthe
change theaccuracy
output voltage the
voltage
model. curve
The detailed is compared
results with the simulation
will be Finally,
discussed curve of the model to verify the accuracy of the
of a lithium-ion battery is ignored. theinactual
the next section.
voltage curve is compared with the simulation
model. The detailed results will be discussed in the next section.
curve of the model to verify the accuracy of the model. The detailed results will be discussed in the
next section.

Figure 7. The simulation model of the first-order RC model for Simulink simulation.

Figure 7. The simulation model of the first-order RC model for Simulink simulation.
Figure7.7.The
Figure Thesimulation
simulationmodel
modelofofthe
thefirst-order
first-orderRC
RCmodel
modelfor
forSimulink
Simulinksimulation.
simulation.

Figure 8. The simulation model of the second-order RC model for Simulink simulation.

Figure 8. The simulation model of the second-order RC model for Simulink simulation.
Figure 8. The simulation model of the second-order RC model for Simulink simulation.
Figure 8. The simulation model of the second-order RC model for Simulink simulation.

Figure
Figure 9.
9. The
The simulation
simulation model
model of
of the
the SOC
SOC calculation
calculation for
for Simulink
Simulink simulation.
simulation.

Figure 9. The simulation model of the SOC calculation for Simulink simulation.
Figure 9. The simulation model of the SOC calculation for Simulink simulation.
Appl. Sci. 2017, 7, 1002 10 of 16
Appl. Sci. 2017, 7, 1002 10 of 16

Figure 10. The terminal voltage calculation submodule of the second-order RC model for
Figure 10. The terminal voltage calculation submodule of the second-order RC model for
Simulink simulation.
Simulink simulation.

4. Results and Discussion

4. Results and Discussion
4.1. Constant Current Discharging
4.1. Constant Current Discharging
In order to verify the accuracy of the models under the simple operating conditions,
In order to verify the accuracy of the models under the simple operating conditions, constant
constant current discharging tests were performed. First of all, the tested battery was fully charged
current discharging tests were performed. First of all, the tested battery was fully charged according
according to the standard charging method and then the battery was left in the open-circuit condition to
to the standard charging method and then the battery was left in the open-circuit condition to rest
rest for 4 h to achieve electrochemical and heat equilibrium. Subsequently, the battery was discharged
for 4 h to achieve electrochemical and heat equilibrium. Subsequently, the battery was discharged at
at a constant current 1C (2.35 A) to its discharging cut-off voltage 2.7 V. For the simulation models,
a constant current 1C (2.35 A) to its discharging cut-off voltage 2.7 V. For the simulation models, first
first the initial SOC was set to 1 and the magnitude of the input current to 2.35 A, and then simulation
the initial SOC was set to 1 and the magnitude of the input current to 2.35 A, and then simulation
models were run. Simulation results of the first-order RC model are shown in Figure 11, where the
models were run. Simulation results of the first-order RC model are shown in Figure 11, where the
red dotted line represents the simulation output voltage and the solid blue line represents the actual
red dotted line represents the simulation output voltage and the solid blue line represents the actual
voltage value. Similarly, the simulation results of the second-order RC model are shown in Figure 12,
voltage value. Similarly, the simulation results of the second-order RC model are shown in Figure 12,
where the red dotted line represents the simulation output voltage, and the solid blue line represents
where the red dotted line represents the simulation output voltage, and the solid blue line represents
the actual voltage value. It can be seen from the figures that the simulated voltage could basically
the actual voltage value. It can be seen from the figures that the simulated voltage could basically
follow the battery terminal voltage changes in the constant current discharging process. However, at
follow the battery terminal voltage changes in the constant current discharging process. However, at
the beginning and end stages of discharging, the error was relatively larger; this is at the beginning
the beginning and end stages of discharging, the error was relatively larger; this is at the beginning
stage and end stage of discharging, lithium-ion batteries are deepening gradually in the polarization
stage and end stage of discharging, lithium-ion batteries are deepening gradually in the polarization
process, and the electrochemical reaction taking place in the battery tends to be unbalanced; the two
process, and the electrochemical reaction taking place in the battery tends to be unbalanced; the two
factors above can lead to the change of parameters in a relatively large range, resulting in a relatively
factors above can lead to the change of parameters in a relatively large range, resulting in a relatively
large increase in output error of the models. For better comparison, the output error of the first-order
large increase in output error of the models. For better comparison, the output error of the first-order
RC model and second-order RC model is analyzed as shown in Figure 13, in which it can be seen that
RC model and second-order RC model is analyzed as shown in Figure 13, in which it can be seen
the second-order RC model has a smaller output error than the first-order RC model in general.
that the second-order RC model has a smaller output error than the first-order RC model in general.
 Appl. Sci. 2017, 7, 1002 11 of 16
Appl. Sci. 2017, 7, 1002 11 of 16
Appl. Sci. 2017,
2017, 7,
7, 1002
1002 11 of 16

Voltage/V
Voltage/V
Voltage/V

Figure 11. The simulation output of the first-order RC model for 1C constant current discharging.
Figure 11. The simulation output of the first-order RC model for 1C constant current discharging.
Figure
Figure 11.
11. The
The simulation
simulation output
output of
of the
the first-order
first-order RC model for 1C constant current discharging.
Voltage/V
Voltage/V
Voltage/V

Figure 12. The simulation output of the second-order RC model for 1C constant current discharging.
Figure 12.
Figure 12. The simulation
simulation output
output of
of the
the second-order
second-order RC
RC model
model for
for 1C
1C constant
constantcurrent
currentdischarging.
discharging.
Figure 12. The simulation output of the second-order RC model for 1C constant current discharging.
error/V
error/V
error/V
Output
Output
Output

Figure
Figure13.13. Output errors of
of the
the first-order
first-order and
and second-order
second-order RC
RC models
models for
for 1C
1Cconstant
constant
Figure discharging.
current 13. Output errors of the first-order and second-order RC models for 1C constant
current discharging.
Figure 13. Output errors of the first-order and second-order RC models for 1C constant
current discharging.
current discharging.
4.2.Urban
4.2. UrbanDynamometer
DynamometerDriving
DrivingSchedule
ScheduleCycle
Cycle
4.2. Urban Dynamometer Driving Schedule Cycle
4.2. Urban Dynamometer
Theurban
urban Driving Schedule
dynamometer drivingCycle
schedule (UDDS) cycle describes thedriving
city bus driving
The dynamometer driving schedule (UDDS) cycle describes the city bus conditions
The urban dynamometer driving schedule (UDDS) cycle describes the city bus driving
inconditions
theThe
United in
urban the
States, United
and was States,
dynamometer developedandby
driving was developed
the United
schedule Statesbyrenewable
(UDDS) the United
cycle States
energy
describes renewable
laboratory
the city energy
to adapt
bus to
driving
conditions in the United States, and was developed by the United States renewable energy
laboratory
the city bus to adapt
driving to
cycle the city bus
conditions. Todriving
simulate cycle
real conditions.
battery To simulate
operation real
scenarios,
conditions in the United States, and was developed by the United States renewable energy a battery
current operation
sequence
laboratory to adapt to the city bus driving cycle conditions. To simulate real battery operation
laboratory to adapt to the city bus driving cycle conditions. To simulate real battery operation
Appl. Sci. 2017, 7, 1002 12 of 16
Appl. Sci. 2017, 7, 1002 12 of 16
Appl. Sci. 2017, 7, 1002 12 of 16
scenarios, a current sequence was extracted from a hybrid electric vehicle (HEV) battery pack under
scenarios, a current sequence was extracted from a hybrid electric vehicle (HEV) battery pack under
UDDS cycle, which is widely used in the battery model evaluation. In order to verify the accuracy
UDDS cycle, which is widely used in the battery model evaluation. In order to verify the accuracy
of the
was modelsfrom
extracted under UDDS cycle,
a hybrid electricthe following
vehicle (HEV) tests werepack
battery performed. First ofcycle,
under UDDS all, the tested
which battery
is widely
of the models under UDDS cycle, the following tests were performed. First of all, the tested battery
was in
used fully
thecharged according
battery model to the standard
evaluation. In order to charging
verify the method
accuracyandofthen the battery
the models under wasUDDSleft in the
cycle,
was fully charged according to the standard charging method and then the battery was left in the
open-circuit
the followingcondition
tests weretoperformed.
rest for 4 hFirst
to achieve electrochemical
of all, the tested batteryand washeat
fullyequilibrium. Subsequently,
charged according to the
open-circuit condition to rest for 4 h to achieve electrochemical and heat equilibrium. Subsequently,
the obtained
standard current
charging sequence
method and was
thenloaded on the
the battery was tested
left inbattery through the
the open-circuit battery tester.
condition to rest For
for 4the
h
the obtained current sequence was loaded on the tested battery through the battery tester. For the
simulation
to models, the initial
achieve electrochemical andSOCheatwas first set toSubsequently,
equilibrium. 1, and then thethe obtained
obtained current
current sequence
sequence waswasset
simulation models, the initial SOC was first set to 1, and then the obtained current sequence was set
as the input
loaded on thecurrent
tested file running
battery the simulation
through the batterymodels. Thethe
tester. For simulation
simulationresults of thethe
models, first-order
initial SOC RC
as the input current file running the simulation models. The simulation results of the first-order RC
model
was firstare
setillustrated
to 1, and then in Figure 14, where
the obtained the sequence
current dotted red wasline
setrepresents
as the input thecurrent
simulation output
file running
model are illustrated in Figure 14, where the dotted red line represents the simulation output
voltage
the and the
simulation solid blue
models. The line represents
simulation theof
results actual voltage value.
the first-order Similarly,
RC model the simulation
are illustrated in Figureresults
14,
voltage and the solid blue line represents the actual voltage value. Similarly, the simulation results
of the the
where second-order
dotted red RC linemodel are illustrated
represents in Figure
the simulation output15,voltage
where theanddotted
the solidredblue
lineline
represents
representsthe
of the second-order RC model are illustrated in Figure 15, where the dotted red line represents the
simulation
the outputvalue.
actual voltage voltage and thethe
Similarly, solid blue lineresults
simulation represents
of the the actual voltage.
second-order It canare
RC model beillustrated
seen from
simulation output voltage and the solid blue line represents the actual voltage. It can be seen from
the
in figures
Figure 15,that
where under UDDSred
the dotted cyclelinethe simulating
represents the voltage
simulation of the bothvoltage
output models can
and thebasically
solid blue follow
line
the figures that under UDDS cycle the simulating voltage of the both models can basically follow
the battery
represents theterminal voltageIt can
actual voltage. variation
be seentrend
from theexcept for that
figures some subtle
under UDDS peakcycleshifts, while the
the simulating
the battery terminal voltage variation trend except for some subtle peak shifts, while the
second-order
voltage RC model
of the both modelshas canabasically
better performance
follow the batterythan the first-order
terminal voltageRCvariation
model intrend tracking
exceptsome
for
second-order RC model has a better performance than the first-order RC model in tracking some
subtlesubtle
some peakpeakshifts. Obviously,
shifts, while thethis is because RC
second-order the model
second-order RC model
has a better has a stronger
performance capability
than the first-order
subtle peak shifts. Obviously, this is because the second-order RC model has a stronger capability
to describe
RC model inthe nonlinear
tracking somepolarization
subtle peak response with twothis
shifts. Obviously, RCisnetworks
because theinstead of one RC
second-order RCnetwork
model
to describe the nonlinear polarization response with two RC networks instead of one RC network
compared to the first-order RC model.
has a stronger capability to describe the nonlinear polarization response with two RC networks instead
compared to the first-order RC model.
of one RC network compared to the first-order RC model.
Voltage/V
Voltage/V

Figure 14. The simulation output of the first-order RC model for the urban dynamometer driving
Figure 14.
Figure 14. The
The simulation
simulation output
output of
of the
the first-order
first-order RC
RC model
model for
for the
the urban
urban dynamometer
dynamometer driving
driving
schedule (UDDS) cycle.
schedule (UDDS)
schedule (UDDS) cycle.
cycle.
Voltage/V
Voltage/V

Figure15.
Figure 15. The
The simulation
simulation output
outputof
ofthe
thesecond-order
second-orderRC
RCmodel
modelfor
forthe
theUDDS
UDDScycle.
cycle.
Figure 15. The simulation output of the second-order RC model for the UDDS cycle.
 Appl.Sci.
Appl. Sci.2017,
2017,7,7,1002
1002 1313ofof1616

For a better comparison, the output error of the first-order RC model and second-order RC
For
model are a better comparison,
analyzed the output
as illustrated error of
in Figure 16.the
It first-order
is seen that RCthe
model and second-order
second-order RC model RC has
model
less
are analyzed as illustrated in Figure 16. It is seen that the second-order
output error than the first-order RC model in general, while the maximum absolute errors of both RC model has less output
error than
models the
are first-order
less than 0.06RC model
V (i.e., in general,
relative while
error less than the maximum
2%). It can alsoabsolute
be found errors
thatof
theboth models
output error
are less than 0.06 V (i.e., relative error less than 2%). It can also be found that
of the first-order model tended to be partial to the positive semi axis while that of the second-orderthe output error of the
first-order model tended to be partial to the positive semi axis while that
model tended to be partial to the negative semi axis, which is caused by modeling itself on the of the second-order model
tended
process to of
bethe
partial to the negative
parameters semi axis,
identification. which
The is caused
results can alsoby modeling
prove the itself onof
ability thethe
process of the
two models
parameters identification. The results can also
above to capture battery dynamic behaviors in some sense. prove the ability of the two models above to capture
battery dynamic behaviors in some sense.

the 1st-order RC model

0.06 the 2nd-order RC model
Output error/V

0.03

-0.03

-0.06

0 500 1000 1500 2000 2500 3000 3500 4000 4500

Time/s

Figure16.
Figure 16.Output
Outputerrors
errorsofofthe
thefirst-order
first-orderand
andsecond-order
second-orderRC
RCmodel
modelfor
forUDDS
UDDScycle.
cycle.

4.3.Comparative
4.3. ComparativeAnalysis
Analysis

InInorder
ordertotobetter
betterevaluate
evaluatethe
theaccuracy
accuracyofofthethetwo
twomodels,
models,their
theiroutput
outputerror
errorwas
wasquantified
quantifiedbyby
calculating the maximum absolute error, the maximum relative error, and the
calculating the maximum absolute error, the maximum relative error, and the root mean square (RMS) root mean square
(RMS)
error, error, respectively,
respectively, as shown as shown
in Table in Table
4 for 4 for theRC
the first-order first-order
model and RCinmodel
Table 5and
for in
theTable 5 for the
second-order
RCsecond-order
model. Under RC 1Cmodel. Under
constant 1C constant
current current
discharge, discharge,
for the for RC
first-order the model,
first-order
the RC
rootmodel, the root
mean square
mean square error was 0.0221 V, the maximum relative error was 1.65%; for
error was 0.0221 V, the maximum relative error was 1.65%; for the second-order RC model, the root the second-order RC
model, the root mean square error was 0.0156 V, the maximum relative error
mean square error was 0.0156 V, the maximum relative error was 1.22%. Under UDDS cycle, for thewas 1.22%. Under UDDS
cycle, for RC
first-order the model,
first-order RC model,
the root the root
mean square mean
error wassquare
0.0298 error
V, thewas 0.0298 V,
maximum the maximum
relative error was relative
1.88%;
error was 1.88%; and for the second-order RC model, the root mean
and for the second-order RC model, the root mean square error was 0.0282 V, the maximumsquare error was 0.0282 V, the
relative
maximum
error relativerespectively.
was −1.69%, error was −1.69%, respectively.

Table4.4.The
Table Theoutput
outputerror
errorofofthe
thefirst-order
first-orderand
andthe
thesecond-order
second-ordermodel
modelfor
for1C
1Cconstant
constantcurrent
current
discharging. Root mean square (RMS): root mean square.
discharging. Root mean square (RMS): root mean square.
Model Type Maximum Absolute Error (V) Maximum Relative Error RMS Error (V)
Model Type Maximum Absolute Error (V) Maximum Relative Error RMS Error (V)
The first-order RC model 0.0610 1.65% 0.0221
The first-order RC model
The second-order RC model 0.0610
0.0452 1.65%
1.22% 0.0221
0.0156
The second-order RC model 0.0452 1.22% 0.0156

It can be seen from the two tables that the maximum relative error of the first-order RC model
andItthe
cansecond-order
be seen from RC the two
modeltables thatthe
under thetwo
maximum relative error
representative of the conditions
operating first-order RC
weremodel
bothand
less
the second-order
than 2%, which RC canmodel under
generally the two
satisfy therepresentative operating conditions
precision requirements were bothcalculation.
in the engineering less than 2%, By
which can generally satisfy the precision requirements in the engineering calculation.
longitudinal comparison, it can be found that the output error of two models had a certain decrease,By longitudinal
comparison,
but not to the it great
can be found
extent that theinoutput
expected errorconditions
the simple of two models had athe
compared certain
actualdecrease,
complexbut not to
conditions
the great extent expected in the simple conditions compared the actual complex
(UDDS), which may be due to measurement errors of VOC. By transverse comparison, the output conditions (UDDS),
which
error may
of thebesecond-order
due to measurement
RC model errors of V
in the OC .conditions
two By transverse
had comparison, the output
a certain decrease errorchange.
but little of the
second-order
This may beRC duemodel
to thein the two conditions
increase had a of
in the number certain decreaseresulting
parameters, but little in
change. This may
the increase of be
the
due to the increase in the number of parameters, resulting in the increase of the uncertainty
uncertainty factors of identification process; thus, the overall accuracy of the second-order RC model factors of
did not provide great improvement over the first-order RC model.
Appl. Sci. 2017, 7, 1002 14 of 16

identification process; thus, the overall accuracy of the second-order RC model did not provide great
improvement over the first-order RC model.

Table 5. The output error of the first-order and the second-order models for the UDDS cycle.

Model Type Maximum Absolute Error (V) Maximum Relative Error RMS Error (V)
The first-order RC model 0.0695 1.88% 0.0298
The second-order RC model −0.0628 −1.69% 0.0282

The discharge curve has been used to predict capacity fading of lithium-ion cells, which can
reflect the discharge cut-off point, discharge plateau, and capacity of the cell. In this paper, the area
surrounded by the discharge curve is used to describe the similarity of the discharge curve, which is
defined as follows: Z
A = Udt (12)

It can be seen from Table 6 that the area surrounded by the discharge curve of the first-order
model is much larger than the second-order model, and therefore the second-order model is more
accurate than the first-order model in terms of UDDS profile.

Table 6. The area surrounded by the discharge curve of the first-order and the second-order models for
UDDS cycle.

Model Type The Area Surrounded by the Discharge Curve (Vs)

The first-order RC model 15,896
The second-order RC model 15,875
Experimental data curve 15,873

5. Conclusions
The lithium-ion battery models are generally divided into electrochemical mechanism models
and equivalent circuit models. The representative first-order RC model and second-order RC model
commonly used in the literature are studied comparatively in this paper, and the following main
conclusions are achieved:
(1) The ECM has a very large advantage in computation time in contrast to electrochemical models;
the simulation time based on the electrochemical model was as long as 9600 s [13], while the
simulation time based on equivalent circuit model in this paper was negligible.
(2) The maximum relative errors of the two RC models under the two representative operating
conditions were all less than 2%, which can generally satisfy the precision requirements for the
practical engineering calculation, such as algorithms based on ECM for advanced BMSs.
(3) The second-order RC model improved the output error in contrast to the first-order RC
model in both simple and complex discharging conditions, but did not improve much.
Therefore, for ordinary applications such as portable consumer electronics, the first-order RC
model could be the preferred choice. However, for stringent applications such as automotive
and aerospace, the second-order RC model could be the preferred choice. The results are of great
instructional significance to the application in practical control systems for the equivalent circuit
modeling of lithium-ion batteries.

Acknowledgments: This work was financially supported by the National Key Research and Development
Program of China (No. 2016YFF0203804), the National Natural Science Foundation of China (No. 51775037),
and the Fundamental Research Funds for Central Universities of China (No. FRF-TP-14-061A2). Also, thanks to
Zachary Omariba for his English editing.
Author Contributions: Lijun Zhang, Hui Peng and Changyan Sun conceived and designed the experiments;
Hui Peng, Zhansheng Ning and Zhongqiang Mu performed the experiments; Hui Peng and Zhansheng Ning
analyzed the data; and Lijun Zhang and Hui Peng wrote the paper.
Appl. Sci. 2017, 7, 1002 15 of 16

Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
VOC open circuit voltage (V)
R0 ohmic resistance (Ω)
R1 polarization resistance (Ω)
C1 polarization capacitance (F)
I current (A)
U the terminal voltage (V)
U0 the voltage before the current changes (V)
U1 the voltage after the moment the current changes (V)
τ the time constant
Ω ohmic
AC alternating current
Ah ampere-hour
BMS battery management system
ECM equivalent circuit model
EIS electrochemical impedance spectroscopy
HEV hybrid electric vehicle
HPPC hybrid pulse power characterization
LiFePO4 lithium iron phosphate
MATLAB MATrix LABoratory
OCV open circuit voltage
RC resistance capacitance
RMS root mean square
SOC state-of-charge
SOH state-of-health
SOF state-of-function
UDDS urban dynamometer driving schedule

References
1. Jaguemont, J.; Boulon, L.; Dubé, Y. A comprehensive review of lithium-ion batteries used in hybrid and
electric vehicles at cold temperatures. Appl. Energy 2016, 164, 99–114. [CrossRef]
2. Wu, L.; Fu, X.; Guan, Y. Review of the remaining useful life prognostics of vehicle lithium-ion batteries using
data-driven methodologies. Appl. Sci. 2016, 6, 166. [CrossRef]
3. Sbarufatti, C.; Corbetta, M.; Giglio, M.; Cadini, F. Adaptive prognosis of lithium-ion batteries based on the combination
of particle filters and radial basis function neural networks. J. Power Sources 2017, 344, 128–140. [CrossRef]
4. Chaturvedi, N.A.; Klein, R.; Christensen, J. Algorithms for advanced battery-management systems.
IEEE Control Syst. 2010, 30, 49–68. [CrossRef]
5. Worwood, D.; Kellner, Q.; Wojtala, M.; Widanage, W.D.; McGlen, R.; Greenwood, D.; Marco, J. A new
approach to the internal thermal management of cylindrical battery cells for automotive applications.
J. Power Sources 2017, 346, 151–166. [CrossRef]
6. Brand, M.J.; Schuster, S.F.; Bach, T. Effects of vibrations and shocks on lithium-ion cells. J. Power Sources 2015,
288, 62–69. [CrossRef]
7. Liu, X.; He, Y.; Zheng, X.; Zhang, J.; Zeng, G. A new state-of-charge estimation method for electric vehicle lithium-ion
batteries based on multiple input parameter fitting model. Int. J. Energy Res. 2017, 41, 1265–1276. [CrossRef]
8. Hannan, M.A.; Hoque, M.M.; Mohamed, A.; Ayob, A. Review of energy storage systems for electric vehicle
applications: Issues and challenges. Renew. Sustain. Energy Rev. 2017, 69, 771–789. [CrossRef]
9. Barai, A.; Chouchelamane, G.H.; Guo, Y.; McGordon, A.; Jennings, P. A study on the impact of lithium-ion
cell relaxation on electrochemical impedance spectroscopy. J. Power Sources 2015, 280, 74–80. [CrossRef]
10. Lu, L.; Han, X.; Li, J. A review on the key issues for lithium-ion battery management in electric vehicles.
J. Power Sources 2013, 226, 272–288. [CrossRef]
Appl. Sci. 2017, 7, 1002 16 of 16

11. Wang, D.; Zhao, Y.; Yang, F.; Tsui, K.-L. Nonlinear-drifted Brownian motion with multiple hidden states
for remaining useful life prediction of rechargeable batteries. Mech. Syst. Signal Process. 2017, 93, 531–544.
[CrossRef]
12. Nikolian, A.; Firouz, Y.; Gopalakrishnan, R.; Timmermans, J.M.; Omar, N.; van den Bossche, P.; van Mierlo, J.
Lithium ion batteries—Development of advanced electrical equivalent circuit models for nickel manganese
cobalt lithium-ion. Energies 2016, 9, 360. [CrossRef]
13. Chu, Z.; Feng, X.; Lu, L.; Li, J.; Han, X.; Ouyang, M. Non-destructive fast charging algorithm of lithium-ion
batteries based on the control-oriented electrochemical model. Appl. Energy 2017, in press. [CrossRef]
14. Zhang, X.; Lu, J.; Yuan, S.; Yang, J.; Zhou, X. A novel method for identification of lithium-ion battery equivalent
circuit model parameters considering electrochemical properties. J. Power Sources 2017, 345, 21–29. [CrossRef]
15. Wang, Q.K.; He, Y.J.; Shen, J.N.; Ma, Z.F.; Zhong, G.B. A unified modeling framework for lithium-ion
batteries: An artificial neural network based thermal coupled equivalent circuit model approach. Energy
2017, 138, 118–132. [CrossRef]
16. Feng, X.; Weng, C.; Ouyang, M.; Sun, J. Online internal short circuit detection for a large format lithium ion
battery. Appl. Energy 2016, 161, 168–180. [CrossRef]
17. Chiang, Y.H.; Sean, W.Y.; Ke, J.C. Online estimation of internal resistance and open-circuit voltage of
lithium-ion batteries in electric vehicles. J. Power Sources 2011, 196, 3921–3932. [CrossRef]
18. Mathew, M.; Kong, Q.H.; McGrory, J.; Fowler, M. Simulation of lithium ion battery replacement in a battery
pack for application in electric vehicles. J. Power Sources 2017, 349, 94–104. [CrossRef]
19. Panchal, S.; Dincer, I.; Agelin-Chaab, M.; Fraser, R.; Fowler, M. Experimental and simulated temperature
variations in a LiFePO 4–20 Ah battery during discharge process. Appl. Energy 2016, 180, 504–515. [CrossRef]
20. Panchal, S.; Mcgrory, J.; Kong, J.; Fraser, R.; Fowler, M.; Dincer, I.; Agelin-Chaab, M. Cycling degradation
testing and analysis of a LiFePO4 battery at actual conditions. Int. J. Energy Res. 2017. [CrossRef]
21. Mu, H.; Xiong, R.; Zheng, H.; Chang, Y.; Chen, Z. A novel fractional order model based state-of-charge
estimation method for lithium-ion battery. Appl. Energy 2017, in press. [CrossRef]
22. Dang, X.; Yan, L.; Xu, K.; Wu, X.; Jiang, H.; Sun, H. Open-circuit voltage-based state of charge estimation of
lithium-ion battery using dual neural network fusion battery model. Electrochim. Acta 2016, 188, 356–366. [CrossRef]
23. Dang, X.; Yan, L.; Jiang, H.; Wu, X.; Sun, H. Open-circuit voltage-based state of charge estimation of
lithium-ion power battery by combining controlled auto-regressive and moving average modeling with
feedforward-feedback compensation method. Int. J. Electr. Power Energy Syst. 2017, 90, 27–36. [CrossRef]
24. Tian, Y.; Li, D.; Tian, J.; Xia, B. State of charge estimation of lithium-ion batteries using an optimal adaptive
gain nonlinear observer. Electrochim. Acta 2017, 225, 225–234. [CrossRef]
25. Li, J.; Barillas, J.K.; Guenther, C.; Danzer, M.A. A comparative study of state of charge estimation algorithms
for LiFePO4 batteries used in electric vehicles. J. Power Sources 2013, 230, 244–250. [CrossRef]
26. Du, J.; Liu, Z.; Wang, Y.; Wen, C. An adaptive sliding mode observer for lithium-ion battery state of charge
and state of health estimation in electric vehicles. Control Eng. Pract. 2016, 54, 81–90. [CrossRef]
27. Tong, S.; Klein, M.P.; Park, J.W. On-line optimization of battery open circuit voltage for improved
state-of-charge and state-of-health estimation. J. Power Sources 2015, 293, 416–428. [CrossRef]
28. Remmlinger, J.; Buchholz, M.; Meiler, M.; Bernreuter, P.; Dietmayer, K. State-of-health monitoring of
lithium-ion batteries in electric vehicles by on-board internal resistance estimation. J. Power Sources 2011, 196,
5357–5363. [CrossRef]
29. Xia, B.; Sun, Z.; Zhang, R.; Lao, Z. A Cubature Particle Filter Algorithm to Estimate the State of the Charge of
Lithium-Ion Batteries Based on a Second-Order Equivalent Circuit Model. Energies 2017, 10, 457. [CrossRef]
30. He, W.; Williard, N.; Chen, C.; Pecht, M. State of charge estimation for electric vehicle batteries using
unscented kalman filtering. Microelectron. Reliab. 2013, 53, 840–847. [CrossRef]
31. Schweiger, H.G.; Obeidi, O.; Komesker, O.; Raschke, A.; Schiemann, M.; Zehner, C.; Birke, P. Comparison
of several methods for determining the internal resistance of lithium ion cells. Sensors 2010, 10, 5604–5625.
[CrossRef] [PubMed]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).