Parameter Estimation of Li-ion Battery

based on Equivalent Circuit Modelling

Sachinkumar R. Patel Sushant Khade
R & D Battery CTO
Tork Motors Tork Motors
Pune,Maharashtra Pune,Maharashtra
[email protected] [email protected]

Abstract—Most modern technology, including provide these estimations. Since the SOC of the
Electric Vehicles (EVs) & energy storage device are battery cannot be evaluated directly, other methods
powered by Lithium ion Batteries (LiB). The advan- have been suggested. Model-based estimation of
tages of this kind of battery over other battery types
include greater energy density & reliability . They the state of charge is one of these methods and it
must be managed using a battery management system is popular in application due to its high accuracy
(BMS), which ensures their safety and improves [10] , [8]. The ECM is one of the cell models
their performance under normal circumstances, for that can be utilized to carry out such an activity.
them to function properly. Accurate estimation of the ECMs simulate the functioning of a lithium-ion
State of Charge (SOC) and State of Health (SOH)
is very critical in EVs. The lithium-ion battery is cell by offering an electrical circuit that displays
a highly time-invariant, nonlinear, and complicated cell like behavior. The ECM’s parameter values are
electrochemical system, making this task extremely fine-tuned using data acquired from laboratory tests
difficult. In this article, we provide a cell model on actual cells [12], [1]. This adjustment process
that may be applied to the process of estimating the aims to precisely align the current and voltage
charge. This model relies on an electrical approach
known as the Equivalent Circuit Model (ECM), in characteristics of the model with those observed
which we construct an electrical circuit with the in real cells. Any disparity between the model’s
same behavior as the actual cell. We will use cutting- predictions and the actual behavior exhibited by
edge methodologies to establish the model parameters the cells is classified as a modeling inaccuracy or
using a set of laboratory data. The estimation of the error [9], [7].
SOC, SOH and other internal parameters will use
these variables. The objective of this research is to create an
Index Terms—Li-ion, Cell Modelling, State of extremely precise ECM for batteries that can ac-
Charge, State of Health commodate a diverse range of temperature con-
ditions, load profiles and SOC Range [11]. The
I. I NTRODUCTION developed model is based on an Electric ECM 1st
In EVs, the battery system is the most crucial Order Thévenin Topology (Fig. 1). The lithium-ion
energy storage component. These days the adoption chemistry identified for this analysis was the Nickel
of high-capacity lithium-ion batteries as a battery Manganese Cobalt (NMC) 5 Ah. The characteris-
energy storage system is the main trend in the tics of these cell are discussed in Table 1.
field of electric vehicles. These batteries have many A wide variety of temperatures were determined
advantages including high power density, light in order to have an in-depth understanding of the
weight, long life, and temperature stability [2], [4]. behavior of the cells, ranging from -5◦ C, 5◦ C,
Real-time monitoring of these batteries SOC and 15◦ C, 25◦ C, 35◦ C, 45◦ C, 55◦ C. The model is vali-
SOH is necessary to increase their useful life and dated at these temperatures under dynamic profiles
safety. BMS that regulate, enhance, and protect such as vehicle drive cycle. In section II, discuss
the battery carry out this function. Estimating a about pre required cell level test before cell mod-
variety of crucial quantities including SOC and elling and all tests are performed at different tem-
SOH of the cell, available power and available perature and different SOC. After pre-required cell
energy is one of a BMS key jobs [6], [5]. Models level test, next step is identify cell parameter which
that accurately capture the dynamic behavior of is discuss in section III. In section IV, detailed
the cell are necessary for the best techniques to overview of LiBs cell modelling equation (1RC)
Fig. 1: First order Thévenin equivalent circuit
topology to model a lithium-ion battery

TABLE I: NMC Cell Specification

Specification Unit Value
Cell Format - 21700
Dimensions mm 21.3 x 70.6
Fig. 2: NEWARE BTS-4000
Weight gm 68.2
Nominal Capacity Ah 4.9
Energy Wh 17.7
Gravimetric Energy Density Wh / kg 260
Volumetric Energy Density Wh / l 704
Gravimetric Power Density kW / kg 1.01
Volumetric Energy Density kW /l 2.73
Maximum Discharge C Rate - 3
Maximum Charge C Rate - 1
Operating Temperature Range ◦C -20 to 60

along with parameter estimation like Open Circuit

Voltage (OCV), Temperature, Terminal Voltage,
SOC and RC parameter [3], [13].

II. P RE R EQUIRED C ELL L EVEL T EST

Pre required cell level test is very important

to find cell capacity, SOC-OCV relationship and
cell internal parameter which is very useful for
cell modelling. The battery testing equipment Fig. 3: NEWARE Environmental Chamber
NEWARE BTS-4000 and Neware Enviromental
chamber is used for this testing.

A. Capacity Test
Purpose of the test is a determine discharge
capacity under different temperature.
1) Set environmental chamber e.g 25◦ C
2) CC Discharge up to 2.5V at C/5
3) Rest 1 hour
4) CC-CV Charge up to 4.2V, C/2 Charging
Current and C/20 cut off current
5) Rest 1 hour
6) CC Discharge up to 2.5V at C/5 and record Fig. 4: Discharge Capacity Test at Different Dif-
discharge capacity ferent Temperature
7) Repeat step 1 to 6 for different temperature.
B. OCV-SOC Test
Purpose of the test is a find relationship between
OCV and SOC at different temperature.
1) Set environmental chamber e.g 25◦ C
2) CCCV charged up to 4.2V and C/20 cut off
current
3) Rest for 1 hr
4) Reduce 5% SOC using C/5 constant current
5) Rest 3 hr and record cell voltage
6) Repeat step 4 to 5 until 0% SOC
7) Repeat step 1 to 6 for different different
temperature.

Fig. 6: The Complete Voltage and Current Profile

of HPPC Test

Fig. 5: SOC-OCV at Different Different Tempera-

ture

C. The Hybrid pulse power characterization Test

HPPC test is a battery performance test method
proposed by the manual freedom test manual for
power assist hybrid EV. Through the experiments
effective identification of model parameters can be
performed.
1) CCCV charged up to 4.2V and C/20 cut off
current e.g 25◦ C
2) Rest 1 hr
3) 1C Discharge plus for 10 sec
4) Rest 3 min
5) 1C Charge plus for 10 sec
6) Rest 3 min
7) Reduce 5% SOC using C/5 constant current Fig. 7: The Single Cycle Voltage and Current
8) Rest 1 hr Profile of HPPC Test
9) Repeat step 3 to 8 until 0% SOC.
10) Repeat step 1 to 9 for different different
temperature. III. A DVANCE PARAMETER I DENTIFICATION
D. Cycle Life Test The internal resistance R0 , polarization resis-
Purpose of the test is to find out a battery tance R1 , polarization capacitance C1 and time
degradation. The test procedure is same as capacity constant tau1 for discharge and charge are all
test and repeat continues until battery capacity is calculated through the HPPC characterization test.
≤ 80% of initial capacity. Perform SOC-OCV and The calculated parameters are then forwarded to
HPPC test at every 20 cycle and record all the data the 1st Order Thévenin ECM topology equations
which is useful for SOH estimation. which calculates the estimation parameter of the
 Fig. 8: Cycle Life Test Fig. 11: Optimized parameter value of Polarization
Resistance (R1 )

cell based on a specific current load profile.

Fig. 12: Optimized parameter value of Polarization

Fig. 9: Equation of parameter identification for 1st Capacitance (C1 )
Order Thévenin ECM

For parameter identification we use above equa- IV. M ODEL D EVELOPMENT

tion (Fig.9) and HPPC test were performed from For the development of the ECM models, a
-5◦ C to 55◦ C in 10◦ C increments. The main is- suitable methodology has been developed in order
sues encountered were related with the parameter to achieve this, as shown in Fig. 13. The work-
identification results of the nonlinear behaviours at flow can be separated in the following steps: (i)
very high (100 to 95%) and low (20 to 0%) SOC Optimal Cell Identification, (ii) Pre required cell
regions. level test, (iii) Parameter identification, (iv) ECM
Parameterization to finally arrive to the final step
(v) Complete ECM Battery Model. In workflow
point (i) to (iv) already discuss in previous section,
now we focus on parameter estimation. In this
section we derive a equation of SOC estimation,
Temperature estimation, Power loss estimation and
Terminal voltage estimation.

A. SOC Estimation
The SOC is one of the most important pa-
rameters for batteries. SOC for a battery can be
expressed as the relationship between its present
Fig. 10: Optimized parameter value of Ohmic Re- capacity and the manufacturer-specified nominal
sistance (R0 ) capacity, which signifies the battery’s maximum
charge storage capacity.
 C. Cell Temperature Estimation
The temperature estimation is very important
parameter for ECM because temperature is directly
affected on thermal runaway, SOC estimation and
SOH estimation. In this model we consider natural
convection and for temperature estimation we use
equation 4.
Z
hA∆T + Ploss
∆T = (4)
mCp
h : Conevactive Heat Transfer Coefficent; A :
Area of cell in m3 ; Ploss : Total Power Loss; m :
Mass of cell in Kg;
Cp : Cell Specific Heat Capacity J/(Kg.K)
D. Terminal Voltage Estimation
Based on Kirchhoff’s law, the state equation and
output equation of the first-order RC ECM can be
obtained according to eq 5.
Up
+ CiLp

 U̇p = − Rp C p
Fig. 13: Complete workflow for the development (5)
and validation of a lithium ion ECM model Ut = UOCV − iL R0 − Up


Where UOCV is the battery open-circuit voltage

related to SOC, Up represents the potential differ-
ence across RC network, Ut is the battery terminal
Q(t)
SOC(t) = (1) voltage, iL refers to the battery current, R0 repre-
Qn sents the ohmic internal resistance of the battery,
and Rp and Cp are the polarization resistance and
In this paper we focus on coulomb counting
polarization capacitance, respectively.
method. This method measures the discharging cur-
rent of a battery and integrates the discharging cur-
rent over time in order to estimate SOC. Coulomb
counting method is done to estimate the SOC(t),
which is estimated from the discharging current,
I(t) and previously estimated SOC values, SOC(t-
1). SOC is calculated by the following equation:

I(t)
SOC(t) = SOC(t − 1) + ∆t (2)
Qn

B. Power Loss Estimation

For the battery to operate in a safe manner, the
battery power losses must be calculated. Determin-
ing the power losses will be crucial for selecting
Fig. 14: Schematic diagram of the first-order RC
the battery’s cooling system and keeping the accu-
equivalent circuit
mulator in the ideal temperature range extending
the battery’s lifetime. Total Power Loss is addition
of Power loss due to R0 and Power loss due to R1 V. R ESULT AND D ISCUSSION
.
To give a better idea of the capabilities of our
ECM model, we present some modeling results in
Ploss = I 2 R0 + I 2 R1 (3) this section. True data was collected from CAN
analyzer, which is connected with Vehicle Control
Unit (VCU) CAN bus and we compare estimated
result with true value.

Fig. 17: Estimated ohmic resistance

Fig. 15: Current profile at cell level

Figure 15 shows that current profile of though

out vehicle running operation. We can see that
average cell current is around 10 A, 7A & 5A
respectively sports, city and eco mode. This cur-
rent profile we can used in ECM modelling for
parameter estimation.
Fig. 18: Estimated polarization resistance

Fig. 16: Estimated power loss Fig. 19: Estimated polarization capacitance

Figure 16 shows that estimated power loss value

and this value directly depend on current value and
internal ohmic and polarization resistance value so
we can see that high power loss in sports mode be-
cause in this region current and resistance value is
high (Higher SOC region) also we can see that after
10000 sec current value almost constant but power
loss increase because that time our resistance value
is high. (Lower SOC region).
Figures 17,18 and 19 shows that ohmic re-
sistance, polarization resistance and polarization Fig. 20: Estimated open circuit voltage
capacitance value respectively. we can see that re-
sistance value high during lower and higher soc. we
can also see that initially polarization capacitance Figure 20 shows that result of estimated open
value is low at higher SOC, then increasing with circuit voltage and this estimation is very useful
temperature and soc value. for SOC and SOH estimation. For SOH estimation,
OCV graph trend changed with respect to ageing
and based on OCV value at particular SOC and
temperature we can estimate SOH value.

Fig. 23: Comparison of estimated and true terminal

voltage

Fig. 21: Comparison of estimated and true temper-

ature

Figure 21 shows that result of comparison of

estimated and true temperature. Initially error value
is low but after some period of time error value
is increasing because in this case, we are using
passive cooling material inside our battery pack but
while ECM modelling we consider only single cell
with natural convection.

Fig. 24: Error of estimated and true terminal volt-

age wrt time

Fig. 22: Comparison of estimated and true SOC

Figure 22 shows that comparison of true

SOC and estimated SOC using coulomb counting
method. We can see that error value is relatively
high in the lower soc because of non linear be- Fig. 25: Error of estimated and true terminal volt-
haviour of NMC cell in this region. age wrt SOC
Figure 23 shows that result of comparison of
true and estimated terminal voltage. We can see
that graph trend of estimated and true values are
almost same only we can observe some offset in positive 0.1 V. The root-mean-square difference
between estimated and true value. This offset is between the actual results and the model results
relatively high in lower soc region because of non was 98.65 mV in this case.
linear behaviour of cell. Figure 25 shows that result of relationship be-
Figure 24 shows that result of error between true tween terminal voltage error and SOC. As we see
and estimated value. As we see in figure 24, the in figure error value relatively high in lower soc
range of the error is between negative 0.3 V and region because of non linear behaviour of cell.
 VI. C ONCLUSION AND F UTURE WORK [11] A. Darwin Jose Raju, S. Solai Manohar, and A. An-
nie Steffy Beula. A behavior modelling and analysis of
In this paper we built ECM model using 1 RC lithium ion battery. In 2023 2nd International Conference
branch, our model can describe the behavior of the for Innovation in Technology (INOCON), pages 1–5, 2023.
cell. Using a set of data collected from cell testing, [12] Jussi Sihvo, Tomi Roinila, and Daniel-Ioan Stroe. Novel
fitting algorithm for parametrization of equivalent circuit
we simulated our model and we compared the model of li-ion battery from broadband impedance mea-
predicted terminal voltage and measured terminal surements. IEEE Transactions on Industrial Electronics,
voltage and we did find that our model predicts 68(6):4916–4926, 2021.
[13] Geetika Vennam and Avimanyu Sahoo. Simultaneous
terminal voltage very well. For future work, we state and parameter estimation of lithium-ion battery: An
work on 2 RC model with major focus on lower observer based approach. In 2019 American Control
and higher soc region where we can reduce the Conference (ACC), pages 4485–4490, 2019.
error and also we work on improvement of SOC
estimation using Extended Kalman Filter (EKF)
method.
R EFERENCES
[1] H.P.G.J. Beelen, H.J. Bergveld, and M.C.F. Donkers. On
experiment design for parameter estimation of equivalent-
circuit battery models. In 2018 IEEE Conference on Con-
trol Technology and Applications (CCTA), pages 1526–
1531, 2018.
[2] Zhaohui Cen, Pierre Kubiak, and Ilias Belharouak. Online
parameter estimation/tracking for lithium-ion battery rc
model. In 2016 International Renewable and Sustainable
Energy Conference (IRSEC), pages 936–940, 2016.
[3] Deya Das, Shrivatsal Sharma, Malay Jana, and Suman
Basu. A modified equivalent circuit model to capture
low temperature effects in li ion batteries. In 2019 IEEE
Transportation Electrification Conference (ITEC-India),
pages 1–6. IEEE, 2019.
[4] Simone Fasolato and Davide M. Raimondo. Observability
analysis of a li-ion cell equivalent circuit model based
on interval arithmetic. In 2022 IEEE Vehicle Power and
Propulsion Conference (VPPC), pages 1–7, 2022.
[5] Guangming Liu, Languang Lu, Hong Fu, Jianfeng Hua,
Jianqiu Li, Minggao Ouyang, Yanjing Wang, Shan Xue,
and Ping Chen. A comparative study of equivalent circuit
models and enhanced equivalent circuit models of lithium-
ion batteries with different model structures. In 2014 IEEE
Conference and Expo Transportation Electrification Asia-
Pacific (ITEC Asia-Pacific), pages 1–6, 2014.
[6] ER-RAKIBI Marwane. Lithium-ion battery modeling
using equivalent circuit model. International Journal of
Engineering and Applied Physics, 1(1):48–60, 2021.
[7] Arpita Mondal, Aurobinda Routray, Sreeraj Puravankara,
and Debanjana Pahari. On-line parameter identification
and state of charge estimation of li-ion battery. In
2020 IEEE 17th India Council International Conference
(INDICON), pages 1–7, 2020.
[8] Raul Nemes, Sorina Ciornei, Mircea Ruba, Horia Hedesiu,
and Claudia Martis. Modeling and simulation of first-order
li-ion battery cell with experimental validation. In 2019
8th International Conference on Modern Power Systems
(MPS), pages 1–6. IEEE, 2019.
[9] Raul Octavian Nemes, Sorina Maria CIORNEI, RUBA
Mircea, and Claudia MARTIS. Parameters identification
using experimental measurements for equivalent circuit
lithium-ion cell models. In 2019 11th International
Symposium on Advanced Topics in Electrical Engineering
(ATEE), pages 1–6. IEEE, 2019.
[10] Alexandros Nikolian, Joris Jaguemont, Joris De Hoog,
Shovon Goutam, Noshin Omar, Peter Van Den Bossche,
and Joeri Van Mierlo. Complete cell-level lithium-ion
electrical ecm model for different chemistries (nmc, lfp,
lto) and temperatures (- 5 c to 45 c)–optimized modelling
techniques. International Journal of Electrical Power &
Energy Systems, 98:133–146, 2018.