Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Modeling and Simulation of Battery Management System (BMS) for

Electric Vehicles

Ravi P Bhovi1, Ranjith A C2, Sachin K M3 and Kariyappa B S4.

1
Dept of Electronics and Communication, RV College of Engineering, India,
2
Dept of Electronics and Communication, RV College of Engineering, India,
3
Dept of Electronics and Communication, RV College of Engineering, India
3
Dept of Electronics and Communication, RV College of Engineering, India
1
[email protected], [email protected], [email protected] and
3
[email protected]

Abstract: Electric cars have evolved into a game-changing technology in recent years. A Battery
Management System (BMS) is the most significant aspect of an Electric Vehicle (EV) in the automotive
sector since it is regarded the brain of the battery pack. Lithium-ion batteries have a large capacity for
energy storage. The BMS is in charge of controlling the battery packs in electric vehicles. The major role of
the BMS is to accurately monitor the battery’s status, which assures dependable operation and prolongs
battery performance. The BMS’ principal job is to keep track, estimate, and balance the battery pack’s cells.
The major goals of this work is to keep track of battery characteristics, estimate SoC using three distinct
approaches, and balance cells. Coulomb Counting, Extended Kalman Filter, and Unscented Kalman Filter
are the three algorithms that will be implemented. Current is used as an input parameter to implement the
coulomb counting method. In contrast to voltage and temperature, the current value is taken into account by
the Extended and Unscented Kalman Filters. To calculate the state transition and measurement update
matrix, these parameters are considered. This matrix will then be used to calculate SoC. Results of all the
algorithms will be comparatively analyzed.
MATLAB R2020a software is used for the simulation of different algorithms and SoC calculation. Three
states of BMS are considered and they are Discharging phase, the Standby/resting phase, and the Charging
phase. At the beginning of the Simulation, the SoC values of the cells were 80%. At the end of simulation
maximum values of SoC of Coulomb counting, Extended Kalman Filter (EKF), and Unscented Kalman Filter
(UKF) reached are 100%, 98.74% and 98.46% respectively. After SoC Estimation, Cell balancing is also
performed over 6 cells of the battery pack.

Keywords: Electric Vehicle (EV), State-of-Charge (SOC), Unscented Kalman Filter (UKF), Extended
Kalman Filter (EKF), Battery Pack Current, Coulomb Counting (CC), State-of- Health (SOH), Battery
Management System (BMS), Cell balancing.

1. Introduction
The electrifying of transportation systems and the need to make electric products portable are driving up global usage of
electric batteries, that will account for the great majority of battery needs in terms of total power storage. With the
increasing demand for batteries, the level of quality must be sustained. Several factors influence battery performance,
including ambient temperature, charging, and battery composition. The state of charge (SoC) of a battery is
acknowledged as a critical metric. Its correct estimation can lengthen the battery’s life cycle, improve battery
performance, and improve the system’s safety and dependability. It’s indeed, however, impossible to estimate SoC
directly. The fundamental issue is that direct calculations are erroneous due to the console’s nonlinearity, temperature,
and time-dependent properties of the battery. Different techniques have been created to quantify SoC, and researchers all
around the globe are continually working on new approaches. The state of charge (SoC) and cell balance are critical
indications for determining a battery’s usefulness and capabilities. The SoC and Cell balancing work together to offer a
state of function, a battery overview, and a comprehensive perspective of the pack’s capabilities. Along with these
indicators, Cell balancing and Charge limit calculation is also important factors in case of battery health and efficient
performance.

Volume 23, Issue 6, June - 2021 Page -805

 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

2. Battery Management System

In electric and hybrid cars, the Battery Management System is crucial to attaining battery performance and extending battery
life. Electric vehicles have become more popular as a result of government regulations limiting CO2 emissions and encouraging
emission-free transportation. Electric vehicles’ primary drawbacks include restricted driving range, high price, battery difficulties,
and inconsistent charging. Lithium- ion batteries were introduced to solve this problem. Battery packs have been embraced for
mobile electronic gadgets and electric cars due to high efficient energy densities. They’ve become more popular in a variety of
applications, including small mobility vehicles, trucks/buses, and industrial gear. The inappropriate utilization of battery leads to
electric shock and fire. A Battery Management System (BMS) is necessary to use battery packs effectively and safely.
A BMS may be thought of as the brain of a battery pack, monitoring pack current, cell voltage, cell temperatures, and determining
available energy throughout the cells, state of charge (SOC), and status of health (SOH) in order to maintain cells within safe operating
limits and extend cell life. The increased reliance on battery packs for power storage has highlighted the significance of battery
management systems (BMSs) that can assure optimal performance, reliable handling, and long lifespan under a variety of charge-
discharge and atmospheric circumstances across sectors. Engineers create feedback and supervisory control algorithms to construct a
BMS that meets these goals. Keep an eye on the voltage and temperature of the cells.
The proposed BMS model in this paper has two major components namely BMS Controller and Battery plant. The battery plant has
a 6 cells module, pre-charge circuit, charger, and inverter. Along with this, BMS Model is composed of a charge limit block, state
machine, SOC estimation block, and cell balancing circuit model. The battery parameters like input and output current for each
cell, cell voltages, cell temperatures and pack current values are extracted from the battery module through the appropriate sensors.
These parameter values will be given to the controller as inputs. Taking these inputs, the controller calculates the input and output charge
limits, provides the SOC information to the user, and also sends the cell balancing command to the battery plant.

Figure 1. BMS Model overview [3]

The BMS Model overview with inputs and outputs is depicted in Figure 1.

3. State-Of-Charge Estimation

The proportion of charge existing inside a rechargeable battery is known as the State-of-Charge (SoC). The discharge rate and temperature
have an impact on the State of Charge. The SOC is a percentage assessment of the amount of available charge in a battery at a given
moment in time and expressed it in %. The SOC tells the user how long the battery may last before it has to be charged or updated.
Battery parameters such as temperature and voltage may be directly monitored using proper instruments. Because of its non-linearity,
SOC cannot be calculated directly and must be estimated. Because there is no appropriate sensor to measure the state of charge, it is
difficult to calculate. SOC is a state parameter that varies dynamically throughout time. As a result, a variety of mathematical and analytical
methods have been developed and are being used. As a result, a variety of mathematical and analytical meth- ods have been developed
and are being used. The charge of a battery degrades when the active material on the elec- trodes wears away. Coulomb Counting,
Unscented Kalman Filter(UKF), and Extended Kalman Filter(EKF) are the sug- gested SOC estimation methods in this study.

Volume 23, Issue 6, June - 2021 Page -806

 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735
A. Coulomb Counting Method
Ampere-hour counting and current integration are other names for this procedure. This is also the most frequent method for
calculating the SoC. To determine SoC values, it uses battery current data that have been mathematically integrated over time.
The remaining capacity is then calculated using the coulomb counting method, which accumulates the charge moved in and out
of the battery. The accuracy of this approach is mostly determined on a precise measurement of the battery current and a
precise estimate of the initial SoC. The State-of-Charge of a battery may be computed by integrating the charging and
discharging currents during the working periods with a known capability, which may be originally determined by the operating
parameters. During the charging and discharging processes, heat is lost. In addition to self-discharging, these losses result in
accumulated mistakes. These aspects should be considered for a more accurate and exact SoC assessment of the battery.
Furthermore, the SoC must be re-calibrated on a regular basis, and the decline of the discharging charge should be taken into
account with a more precise estimation. To solve the drawbacks of the coulomb counting approach and increase its estimation
accuracy, an upgraded coulomb counting technique has been devised for calculating the SoC. As a result, the SoC estimation
becomes more exact and accurate. The proportion of the releasable regarding sustainability to the nominal battery is the state
of charge.

B. Extended Kalman Filter

The Kalman filter is a method for estimating a battery’s SoC. It may also be utilised to figure out what a dynamic system’s inner states
are. Under process and measurement noise, the Kalman filter provides an efficient state estimator for linear systems. The Extended
Kalman filter is a nonlinear system’s version of the Kalman filter. Extended Kalman filter linearizes functions based on their mean and
covariance. There are two phases to the estimating process. The initial step is to update the time in order to estimate the states. This type
of estimating is referred to as a priori estimate. The states from the previous iteration, an input signal sample, and the process covariance
matrix are used as input signals in this stage. The second stage is to update the measurement, which is done through feedback. This stage
is known as a posteriori estimate and is symbolised by the carat symbol. It employs inaccuracy in calculating output signal to correct a
priori prediction of the states.

C. Unscented Kalman Filter

The Unscented Kalman filter, unlike the EKF, doesn’t really linearize state-space equations. Instead, it uses a nonlinear Unscented
Transformation (UT), in which the mean and error covariance are computed and updated repeatedly, to produce sigma points for states.
All sigma points are then transmitted via nonlinear model functions, providing an a priori approximation of the states and output signal.
Depending on their statistics, the mean and covariance of those variables are determined [5].

4. Cell Balancing

Figure 2. Balancing of cells with different states of charge.

Cell balancing is a critical feature of every BMS since it ensures that each cell has the same amount of charge and so maximizes the
overall energy available in the pack. The energy of the entire battery pack is restricted by the block with the lowest charge. All blocks
require equal amount of charge in order to supply its maximum energy capacity. As shown in figure 2, battery pack capacity is limited
to the weakest cell (figure a), A pack for balancing (figure b), after recharging the balanced battery pack, the total capacity available
(figure c).
The circuit to illustrate the passive balancing is shown in the Figure 3 Passive balancing approaches work by removing charge from
the highest cells until the charge of the lowest cells is equal. This is accomplished by discharging the extra charge through resistors that
are switched by transistors in their active state. Methods for passive balancing can be used in two ways. The first method is to use a
fixed shunt resistor [7], and the second method is to use a controllable shunt resistor [10]. Although it does not appear to be as elegant

Volume 23, Issue 6, June - 2021 Page -807

 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735
as active balancing since energy is spent in heat, there are number of reasons why this technique has become the preferred method of
most BMS, including: Simple design, low-cost parts, simple installation, and expandable design. Passive balancing [11] can be used
with a cooling system to maintain the battery pack temperature under control.

Figure 3. Passive Balancing using resistors to dissipate additional energy

Figure 4. Lithium Cell

5. Simulation
In this section, simulation of SOC estimation and cell balancing is discussed. A battery pack of 6 lithium cells is considered for
the process. A typical lithium cell configuration is shown in fig 4. Using this cell configuration, the battery parameter like Voltage,
Current and temperature of each cell is extracted using appropriate sensors. The 6 cells configuration is shown in fig 5. This
model shows a single battery module. A single battery module consists of six individual cells connected in series. The reason for
making these modules is to increase the efficiency of the system and to simulate different configurations. It can be observed
that each cell is equipped with an individual thermal element that gives out the temperature of the individual cells. Along with
the temperature, voltage and current are also noted of each cell. This helps to calculate SoC accurately and reduces the chances
for junk value. The State of Charge estimation is done using three meth- ods namely, the Coulomb Counting method, the
Unscented Kalman filter & the Extended Kalman Filter. In Coulomb Counting method current and temperature from the battery
plant is given as inputs, whereas Kalman filters accept cell voltages as inputs and predict the SOC values. The CC configuration
is shown in fig 6. Coulomb Counting method basically deals with the current owing out of the cells to calculate SoC. This could
be illustrated with the expression given in equation 2. The two versions of the Kalman filter are shown in fig 7. Both versions are
modeled in Simulink R2020a. Cell voltages.

Volume 23, Issue 6, June - 2021 Page -808

 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 5. Six cells battery pack

Figure 6. Simulink Model of Coulomb Counting method

are given as input to them. The non-linear Kalman filter versions Extended Kalman Filter and Unscented Kalman Filter
rely on the unit cell model to forecast a terminal voltage arising from a current input, thus predicting the internal cell
states. Both the methods have two common steps involved in estimating SoC: State update and Measurement update.
Both the methods take cell voltage, state transition function, and measurement function as input. Similarly, state transition
and measurement function take previous SoC, temperature, and current to evaluate the block result. All the outputs are
transferred from the output port to the BMS info port.

Figure 7. Simulink model of Kalman Filters (UKF & EKF)

current are shown graphically and the values are tabulated in the following subsections. Also, the comparison of three SOC
estimation methods and cell balancing results are discussed.

Volume 23, Issue 6, June - 2021 Page -809

 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

6. Results and Analysis

After Simulating the BMS model, the battery parameters such as cell voltages, cell temperatures, and battery pack.

A. Cell Voltages

Figure 8. Cell voltages

Figure 8 shows the voltage across each cell of the battery pack will be exactly the same during any mode of operation. The current
flowing in and out of the battery causes the cell voltage to fluctuate. Because the model is set up with a slight SoC imbalance
and the BMS is in discharge mode, cell voltages are slightly fluctuating at the start of the simulation. As a result of the balancing,
the values converge towards one another at the end of the simulation. The Table 1 gives a detailed description of graph.
During Discharging phase the Cell voltage of the battery pack gradually decreased from 3.8V to 3.48V. During Standby Cell
voltage is showing 3.6V. When BMS is at the beginning of the Charging phase then the Cell voltage starts increasing.

Table 1. Cell Voltages Based on BMS States

Serial no. State Obtained

Voltage[V]
1 Beginning of 3.8
Discharge
2 End of 3.48
Discharge
3 During Standby 3.6
4 Beginning of 4.021
Charging
5 End of Charging 4.171
6 Max. Voltage 4.208
Value

B. Cell Temperature
When the battery is charging or discharging, the temperature generated in the Battery model is roughly proportional to the
current through it. The temperature traces reflect a massive variation between the hottest and coldest cells. Figure 9 shows the cell
temperatures of the battery pack based on BMS states accordingly. Because it is thermally insulated on one side, cell 1 gets
considerably hotter than cell 6. Active thermal regulation is preferred due to the significant temperature difference between the
edge cells. Table 2 gives a detailed description of the graph.

Volume 23, Issue 6, June - 2021 Page -810

Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Figure 9. Cell temperatures

Table 2: Cell 1 Temperatures Based on BMS States

Serial no. State Obtained

Temperature[K]
1 Beginning of 288.1
Discharge
2 End of 297.1
Discharge
3 During Standby 295.7
4 Beginning of 297.3
Charging
5 End of Charging 297.6
6 Max. Temperature 302.8
Value

C. Pack Current

Figure 10. Battery pack current

Volume 23, Issue 6, June - 2021 Page -811

 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

Fig 10 shows the variation of cell current in different modes of BMS. The current limit is calculated based on the maximum
resistance of the battery cell. Initially, the current fluctuates during the discharge state due to variation of factors across load.
During standby state, the graph goes to zero. Then later during the charging state the current increases and reach its peak value as
cell voltage reaches its maximum value. With the help of a built-in current limiting circuit, the current decreases, and a sudden
increase of current is avoided. The Table 3 gives a detailed description of graph.

Table 3: Cell Current Based on BMS States

Serial no. State Obtained

Current[A]
1 Beginning of 0
Discharge
2 End of -19.16
Discharge
3 During Standby 0
4 Beginning of 30
Charging
5 Median of Charging 10.28
6 End of Charging 0

D. State of Charge (SOC)

Figure 11. State of Charge

In Fig 11 the graph shows the tracks of three different SoC methods that are used in our work. The yellow colour track indicates
the Coulomb Counting method, the Blue one is of the Unscented Kalman filter, and the orange trace indicates the SoC estimation
of the Extended Kalman filter. At the beginning of discharging phase, the SoC value of cells is 80%. As the BMS state proceeds
to the Standby state then SoC values of cells vary for different methods. It can be concluded that EKF is more accurate and precise
in estimation. Table 4 gives a detailed description of the graph. EKF performed better than the UKF by recovering from the
initial error.

Table 4: Comparison of Soc Estimation Methods Based On BMS States

Serial no. State Coulomb EKF[%] UKF[%]

Counting[%]
1 Beginning of 80 80 80
Discharge
2 End of 52.96 49.31 48.1
Discharge
3 During Standby 52.45 48.7 43.6

Volume 23, Issue 6, June - 2021 Page -812

 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735

4 Beginning of 60.3 56.5 55.5

Charging
5 End of Charging 100 98.62 98.46

E. Cell Balancing
Cell balancing is a very important factor in BMS. Fig 12 depicts the cell balancing function of the BMS. The traces of 6 cells
are is shown in the graph.

Figure 12. Cell balancing

F. BMS State
The Complete simulation of BMS is divided into three states as shown in Fig 13, they are
• Discharging
• Standby
• Charging

Figure 13. BMS State

All the output graphs are plotted with reference to these states only. First, the BMS goes into discharge state then standby state
later and finally goes into Charging state. Almost all the parameters measured are unstable during the discharge state and become
zero or low during the standby state and during the charging state the graphs try to attain a stable state with time.

7. Conclusion
Volume 23, Issue 6, June - 2021 Page -813
 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735
The BMS Cell voltage, cell temperature, Cell balancing and SoC estimation is performed using MATLAB. Comparison of the
three algorithms is done with the help of output graphs simulated in Simulink Scope viewer. From the SoC graph, we can see that
the initial SoC value of the battery pack is 75%, but the Estimators were initiated to 80% to evaluate their potential to recover. At
the end of the charging state, the maximum values of SoC of Coulomb counting, EKF, and UKF reached are 100%, 98.74%, and
98.46% respectively. Practically SoC can’t reach 100% but from the values obtained, coulomb counting has reached 100% due to
the accumulation of error from the current sensor as it is an open loop method. Whereas two versions of the Kalman filter are
closed-loop and did not reach 100% by eliminating the error. It can be observed that UKF recovers from the error much earlier
than EKF by dropping to a minimum value of 48.7% and retains it until standby mode is finished. Whereas EKF drops to a
minimum value of 47.6%, thereby taking more time to recover from the error. It can be concluded that Extended Kalman Filter
estimates more accurately in the discharging phase, while over a substantially extensive range of SoC values, EKF gives a finer
SoC estimation.

References
[1] H. Ren, Y. Zhao, S. Chen, and T. Wang, “Design and implementation of a battery management system with active charge
balance based on the soc and soh online estimation,” Energy, vol. 166, Oct. 2018. doi: 10.1016/j.energy.2018.10.133.
[2] K. W. E. Cheng, B. P. Divakar, H. Wu, K. Ding, and H. F. Ho, “Battery- management system (bms) and soc development
for electrical vehicles,” IEEE Transactions on Vehicular Technology, vol. 60, no. 1, pp. 76–88, 2011. doi:
10.1109/TVT.2010. 2089647.
[3] B. Kumar, N. Khare, and P. K. Chaturvedi, “Advanced battery man- agement system using matlab/simulink,” in 2015 IEEE
International Telecommunications Energy Conference (INTELEC), 2015, pp. 1–6. doi: 10.1109/INTLEC.2015.7572447.
[4] B. Balasingam, M. Ahmed, and K. Pattipati, “Battery management systems—challenges and some solutions,” Energies, vol.
13, p. 2825, Jun. 2020. doi: 10.3390/en13112825.
[5] X. Liu, X. Deng, Y. He, X. Zheng, and G. Zeng, “A dynamic state- of-charge estimation method for electric vehicle lithium-
ion batteries,” Energies, vol. 13, p. 121, Dec. 2019. doi: 10.3390/en13010121.
[6] M. Ali, A. Zafar, S. Nengroo, S. Hussain, M. Alvi, and H.-J. Kim, “Towards a smarter battery management system for electric
vehicle applications: A critical review of lithium-ion battery state of charge esti- mation,” Energies, vol. 12, p. 446, Jan. 2019.
doi: 10.3390/en12030446.
[7] T. Xiao, X. Shi, B. Zhou, and X. Wang, “Comparative study of ekf and ukf for soc estimation of lithium-ion batteries,” in
2019 IEEE Innovative Smart Grid Technologies - Asia (ISGT Asia), 2019, pp. 1570–1575. doi: 10 . 1109 / ISGT - Asia .
2019.8880915.
[8] Z. B. Omariba, L. Zhang, and D. Sun, “Review of battery cell balancing methodologies for optimizing battery pack
performance in electric vehicles,” IEEE Access, vol. 7, pp. 129 335–129 352, 2019. doi: 10.1109/ACCESS.2019.2940090.
[9] M. Khalid, S. S. Sheikh, A. K. Janjua, and H. A. Khalid, “Performance validation of electric vehicle’s battery management
system under state of charge estimation for lithium-ion battery,” in 2018 International Conference on Computing, Electronic
and Electrical Engineering (ICE Cube), 2018, pp. 1–5. doi: 10.1109/ICECUBE. 2018.8610969.
[10] M. Daowd, N. Omar, P. Van Den Bossche, and J. Van Mierlo, “Passive and active battery balancing comparison based on
matlab simulation,” in 2011 IEEE Vehicle Power and Propulsion Conference, 2011, pp. 1–7. doi: 10 . 1109 / VPPC . 2011
. 6043010.
[11] V. Vaideeswaran, S. Bhuvanesh, and M. Devasena, “Battery management systems for electric vehicles using lithium ion
batteries,” in 2019 Innovations in Power and Advanced Computing Technologies (i-PACT), vol. 1, 2019, pp. 1–9. doi:
10.1109/ i-PACT44901.2019.8959965.
[12] W. Yan, B. Zhang, G. Zhao, S. Tang, G. Niu, and X. Wang, “A battery management system with a lebesgue-sampling-
based extended kalman filter,” IEEE Transactions on Industrial Electronics, vol. 66, no. 4, pp. 3227–3236, 2019. doi:
10.1109/TIE. 2018.2842782.
[13] R. Xiong, Y. Zhang, J. Wang, H. He, S. Peng, and M. Pecht, “Lithium- ion battery health prognosis based on a real battery
management system used in electric vehicles,” IEEE Transactions on Vehicular Technology, vol. 68, no. 5, pp. 4110–4121,
2019. doi: 10.1109/TVT.2018.2864688.
[14] M. Corno and G. Pozzato, “Active adaptive battery aging management for electric vehicles,” IEEE Transactions on
Vehicular Technology, vol. 69, no. 1, pp. 258–269, 2020. doi: 10.1109/TVT.2019.2940033.
[15] H. Karlsen, T. Dong, Z. Yang, and R. Carvalho, “Temperature- dependence in battery management systems for electric
vehicles: Chal- lenges, criteria, and solutions,” IEEE Access, vol. 7, pp. 142 203–142 213, 2019. doi: 10 . 1109 / ACCESS .
2019 . 2943558.
[16] M. Cherukuri and M. Kanthi, “Battery management system design for electric vehicle,” in 2019 IEEE International
Conference on Distributed Computing, VLSI, Electrical Circuits and Robotics (DISCOVER), 2019, pp. 1–6. doi: 10 . 1109
/ DISCOVER47552.2019.9008088.
[17] S. Kivrak, T. Ozer, Y. O ˘guz, and E. Erken, “Battery management system imple- ¨ mentation with the passive control
method using mosfet as a load,” Measurement and Control, vol. 53, p. 002 029 401 988 340, Dec. 2019. doi:
10.1177/0020294019883401.
[18] Y. Xing, E. Ma, K.-L. Tsui, and M. Pecht, “Battery management systems in electric and hybrid vehicles,” Energies, vol.
4, Dec. 2011. doi: 10.3390/en4111840.
Volume 23, Issue 6, June - 2021 Page -814
 Journal of University of Shanghai for Science and Technology ISSN: 1007-6735
[19] h. Hongwen, H. Qin, X. Sun, and Y. Shui, “Comparison study on the battery soc estimation with ekf and ukf algorithms,”
Energies, vol. 6, pp. 5088–5100, Oct. 2013. doi: 10.3390/en6105088.

[20] P. Salehen, M. Su’ait, H. Razali, and K. Sopian, “Battery management systems (bms) optimization for electric vehicles
(evs) in malaysia,” vol. 1831, Apr. 2017, p. 020 032. doi: 10.1063/1.4981173.

Volume 23, Issue 6, June - 2021 Page -815