Detailed Modelling Procedure for Lithium-ion

Battery Using Thevenin Equivalent

Lokesh Gurjer Poonam Chaudhary H. K. Verma
Dept. of Electrical Engineering Dept. of Electrical Engineering Dept. of Electrical Engineering
Shri G. S. Institute of Technology and Shri G. S. Institute of Technology and Shri G. S. Institute of Technology and
Science, Indore, India Science, Indore, India Science, Indore, India
[email protected] [email protected] [email protected]

Abstract—This paper presents a comprehensive procedure for transportation. The separator is made up of micro porous
modelling of Lithium-ion batteries. The proposed modelling membrane which prevents short circuit between anode and
procedure adopts Thevenin equivalent circuit of both first and cathode by providing passage to only Li-ions.
second orders in order to incorporate the dynamic
characteristics of Li-ion batteries. The model parameter The state of the art for Li-ion modelling boasts of
extraction process is done through the use of pulse discharge three methods, electrochemical model, mathematical model
(PD) and pulse charge (PC) tests and the procedure is discussed and battery run time model / equivalent circuit model [9].
in detail. The effect of self-discharging is ignored due to its very Electrochemical model requires several partial differential
large time constant. In order to verify the accuracy of the model, equations of higher order to be solved, hence is exhaustive
comparison of simulation results with experimental data is
demonstrated for three temperatures, 0°C, 25°C and 45°C. The
in approach. Mathematical models introduces significant
modelling technique for the battery would specially find errors which may not be acceptable in some applications[5].
applications in electric vehicle based simulations where the load They also involve large simulation times.
current is highly pulsating and the battery gets very frequently Equivalent circuit models are of three types, Impedance
charged and discharged. based model, Thevenin based model and battery run time
model. The difference between the impedance based model
Keyword- SOC, Li-ion, pulse charge, pulse discharge, equivalent and the Thevenin based model lies in their parameter
circuit model, Thevenin equivalent extraction procedure. Impedance based model uses
electrochemical impedance spectroscopy [6-8]while
I. INTRODUCTION Thevenin based models use current discharge and current
charge pulses for parameter extraction.
The rising global temperature of the earth is a serious issue
which has gained evident attention in the last few decades.
Out of the many other causes, one of the major culprits is
surging CO2 emission. According to report of NASA [1], the
atmospheric CO2 level has seen a sharp rise from 300 ppm
to405 ppm in the last six decades. In urban areas, the issue
can be resolved significantly by replacing internal combustion
engine (ICE) cars with zero emission vehicles i.e. electric
vehicles (EVs) [2]. Along with EVs comes the demand for
rechargeable batteries as well. Considering EVs, Lithium-ion
batteries are leading the market due to various advantages
like, high power density, high energy density, high terminal
voltage, no memory effect and lower self-discharging Fig 1. Two RC based thevenin equivelent circuit model
properties [3, 4]. Since the load conditions are highly non-
linear, in order to accurately verify the performance of
charging techniques, a detailed model of the battery is II. THEVENIN EQUIVALENT CIRCUIT
required. Accurate electrical modelling of the battery requires
knowledge of battery chemistry and dynamics. Thevenin equivalent circuit for a single Li-ion cell is shown
in Fig 1. The equivalent circuit consists of open circuit
Out of the four major components of Li-ion battery cell, the voltage, Vocv, internal resistance or ohmic resistance, R0, and
cathode material is made up of lithium metal oxide which is one or two parallel RC circuits. All these parameters are a
of layered or tunneled structure in order to have high energy function of state of charge (SOC), Temperature and ageing.
capacity [3]. Different cathode materials generally used are The accuracy of the model depends upon number of RC
LiMnO2, LiCoO2 and LiNiO2 because of their high oxidation branches as well. Typically one-three RC branches are used.
potential of approximately 4V. The anode is made up of Increment of the order causes accuracy at the cost of
carbon or pyrolytic carbon which is capable of reversible Li- complexity. However, generally only first and second order
ion reservoir [2]. The Electrolyte consists of Lithium salt and equivalent circuits are sufficient to model the fast and
organic solvents and provides mode for lithium-ions
978-1-5386-8158-9/19/$31.00 ©2019 IEEE
medium scale battery dynamics. This is the reason only first Calculation of Open Circuit Voltage: At the end of every
and second order models are included in the paper. relaxation period or rest period, it is observed that voltage is
All the parameters of the Thevenin equivalent circuit shown not dependent on previous SOC level, as shown in Fig 3.
in Fig 1, and the related terminologies are discussed below,

State of Charge (SOC):SOC is defined as the ratio of

capacity still available in cell (c) to nominal capacity of cell
(CQ). SOC is determined in this paper through coulomb-
counting method. IfSOC0is initial SOC of battery andib is
battery current, then,

(1)

Open Circuit Voltage (OCV), Vocv:OCV of the cell is the

open circuit voltage, Vocv, under steady state conditions or
chemical equilibrium conditions. The voltage is variable and
needs to be determined constantly through pulse discharging
(PD) and pulse charging (PC) tests [9].

Internal Resistance, Ro: The large electrolyte resistance is

represented with an internal resistor, Ro. Resistances of
other components such as electrodes and separator are
included in internal resistance Ro [8].

RC Branch: Two parallel pairs of RC consisting of RP1, RP2,

CP1 and CP2, model the transient behavior of the battery.RP1
and CP1describe the fast dynamics like surface effect and
reaction kinetics on the electrodes. RP2 and CP2 represent the
slow dynamic nature of the battery. The time constant is in
the order of hours and represents the processes in the
organic electrolyte and active material [8].
Fig. 2 Flow Chart of Modeling Approach of Lithium-ion battery cell

III. PARAMETER EXTRACTION

In this paper a detailed design procedure is proposed to

estimate the above mentioned parameters, as shown in the
flow chart of Fig. 2. The parameters for the equivalent
model are extracted using the data obtained from
experiments performed under controlled conditions. The
parameters are dependent upon both state of charge (SOC)
and temperature. Hence, two dimensional lookup table is
generated to model the SOC and temperature dependence of
the parameters.

Experimental Data:
The experimental data is acquired from CALCE [10]. The
data for PD and PC tests is collected for temperatures of
0°C, 25°C and 45°C. For performing the PD test, the
authors have charged the battery initially to full (100%
SOC) using positive current pulse. Next, negative constant
current (0.5C) pulses with relaxation duration of 10%
SOCare fed to the battery cell after a time gap of 2hr. Fig 3 Pulse Discharge response of battery cell.
Terminal voltage in response to the PD test is shown in Fig
3. Therefore, in order to calculate the steady state value of
In Fig.3. Time period,t1,is duration of discharging pulse and OCV,voltage at a given value of SOC is observed.
time, t2, is voltage relaxation period. During the time period, Hysteresis effects are ignored for easing the process of
t1, all parameter of the cell vary with SOC but in period, t 2, mapping. Final OCV is obtained by averaging the OCV
or relaxation period, SOC is constant. values obtained from both charging and discharging
characteristics. OCV-SOC characteristics are common for 1
both one RC and two RC models. C p1 
R p1
,
2
Cp2  (8)
Rp 2
.
(9)

Table 1 Specification of lithium ion battery

Property Cell Unit

Nominal capacity 2000 mAh
Nominal voltage 3.6 V
Max voltage 4.2 V
Cutoff voltage 2.5 V
Operating Temp.Max 0-50 °C
charge current 4 A
Std. charge current 1 A
Max discharge current 22 A
Fig. 4. Discharging current pulse response.
Std discharge current 4 A

Calculation of Ro & Rp: From Fig.4 Initially Ro is Where Cp (single RC model), Cp1&Cp2 (double RC model )
calculated using [8], are polarized capacitors.

V (2) SIMULATION RESULTS

RO 
I
whereV is instantaneous voltage drop and I is magnitude Li-ion battery with specifications shown in Table. 1. is used to
of current pulse. In this case Equation (2) is rewritten as, develop the simulation model. The equivalent circuit shown in
Fig. 5. Is used for simulation.
V2  V3 (3)
RO 
I
and value of polarized resistances, Rp (single RC model), Rp1
and Rp2 (double RC model ) is calculated using,

Vocv  V2 (4)
Rp 
I ,
V1  V2 (5)
R p1 
I ,
Vocv  V1 (6)
Rp 2 
I .

Estimation of Cp: The time constant for RC cell is

calculated using the technique of curve fitting for time
interval, t2.Finally values of capacitors are calculated using
Equations (7)-(9),
Fig.5.one RC based Thevenin equivalent circuit model.
 (7)
Cp 
Rp
,
Both simulation and experimental results are shown in Fig. Fig.7. PD test at 0.5C, 10%duty cycle and 2 hour rest period at 25°C (a)
Experimental and simulated voltage are compare (b) voltage error for both
6. – Fig. 11.
model at 45°C

(a)

(a)
(b)

Fig.8. PD test at 0.5C, 10%duty cycle and 2 hour rest period at 0°C (a)
Experimental and simulated voltage are compare (b) voltage error for both
model at 0°C.

(b)

Fig.6. PD test at 0.5C, 10%duty cycle and 2 hour rest period at25°C (a)
Experimental and simulated voltage are compare (b) voltage error for both
model at25°C.

(a)

(b) Fig.9. PC test at 0.5C, 10% duty cycle and 2 hour rest period at 25°C (a)
Experimental and simulated voltage are compare (b) voltage error for both
model at 25°C.
 It is observed that the simulation results match very closely
with experimental result. The maximum value ofMPE
(maximum percentage error) is 6.2 % and 5.2% for one RC
and two RC models respectivily at 25°C.
For one RC model, the mean absolute percentage error is
0.47%,RMSPE (Root mean square percentage error)is
0.89% and MAPE(maximum absolute percentage error) is
4.3%.
For two RC model, MAPE is 0.45% and RMAPE is 0.78%
and maximum MPE is 1.9 %.

Table 2 ERROR CHARACTRISTICS AT DIFFERENT TEMPRATURE,

SOC RANGE 10-100%

(a)
Temp. Test Max APE MAPE(%) RMSPE
At rate (%) (%)
0.5C One Two One Two One Two
RC RC RC RC RC RC
0°C PD 3.62 2.8 0.26 0.25 1.58 0.96
PD 4.3 1.9 0.47 0.45 0.89 0.78
25°C PC 1.56 0.88 0.36 0.06 1.81 0.60
(b) PD 1.9 0.55 0.51 0.11 0.72 0.50
45°C PC 1.13 1.1 0.14 0.07 0.69 0.29
Fig.10. PC test at 0.5C, 10% duty cycle and 2 hour rest period at 45°C (a) 25°C DST 10.3 6.72 1.94 0.9 6.66 1.77
Experimental and simulated voltage are compare (b) voltage error for both
model at 45°C

CONCLUSION AND FUTURE WORK

This paper has proposed a detailed design procedure for

modelling of Li-ion battery. The model used is based on
Thevenin equivalent circuit. Parameter extraction process for
the Equivalent model is a complicated process which is
simplified upto a certain level by the authors. Validation of
the proposed procedure is done through comparison of
simulation and experimental data. It is observed that for the
model will be quite useful for pulsed constant current
charging and discharging applications like electric vehicles.
For future work, the authors will focus to include the capacity
fading occurring as an effect ofbattery aging.

REFERENCES

(a) [1] Global climate change- https://climate.nasa.gov/vital-signs/carbon-

dioxide.
[2] XIAPOENG Chen, Thanh TU Vo ''An overview of lithium-ion batteries
for electric vehicles'' Conference on power& Energy IEEE 2012.
[3]Kularatna Nihal ''Energy Storage Devices for Electronic Systems''
ELSEVIER publication.
[4] Gilbert M. Masters “Renewable and efficient Electric power system’’ A
JOHN WILEY & SONS, INC., PUBLICATION.
[5] M.S. Whittngham, “ Lithium batteries and cathode materials”, Chem.
Rev. 2004, 104, pp.4271-4301.
[6] A.Seaman, T.-S. Dao, J. McPhee, “A survey of mathematics-based
(b) equivalent-circuit and electrochemical battery models for hybrid and electric
vehicle simulation”,J. Power Sources, 256 (2014), pp. 410-423.
Fig.11. DST test at different amplitude of current, 10% at 25°C (a) [7] Z. Deng, Z. Zhang et al., “Electrochemical Impedance Spectroscopy
Experimental and simulated voltage are compare at 25°C(b) voltage error Study of a Lithium/Sulfur Battery: Modeling and Analysis of Capacity
at 25°C Fading,” J. Electrochemical. Soc., vol. 160, no. 4, pp. A553–A558, Jan.
2013.
[8] Chen M and Rincon-Mora GA. “Accurate electrical battery model capable
of predicting runtime and I-V performance.” IEEE Trans Energy Converse
vol 21.no.2. pp.504-511. June. 2006.
[9] M. Einhorn, F. V. Conte et al., “Comparison, selection, and
parameterization of electrical battery models for automotive applications,”
IEEE Trans. Power Electron., vol. 28, no. 3, pp. 1429–1437, 2013.
[10] Center for advanced life cycle engineering
(CALCE),https://web.calce.umd.edu/batteries/data.htm
[11] Jaguemout,J. et al ''Characterization and Modeling of a hybrid electric
vehicle Lithium-ion battery pack at Low temprature'' IEEE Trans.Veh.
Technol.vol65, Jan. 2016.
[12] A. Hentunen, T. Lehmuspelto, and J. Suomela, “Time-Domain
Parameter Extraction Method for Thevenin-Equivalent Circuit Battery
Models,” IEEE Trans. Energy Convers., vol. 29, no. 3, pp. 558–566, 2014.
[13] S. Buller, M. Thele, R. W. De Doncker, and E. Karden, “Impedance
based simulation models of super capacitors and Li-ion batteries for power
electronic applications,” in Conf. Rec. 2003 Industry Applications Conf., vol.
3, pp. 1596016000.