KEC Journal of Science and Engineering, Vol.

8 Issue 1, August 2024 23

Received Date: 26th December, 2023

Revision Date: 14th May, 2024
Accepted Date: 14th July, 2024

Active Cell Balancing Approach for Efficient Battery Management System

Aakriti Sharma1, Anusha Adhikari2*, Bibek Adhikari3, Pratik Pandey4, Indraman Tamrakar5
1
Dept. of Electrical Engineering, Pulchowk Campus, IOE, TU. Email: [email protected]
Dept. of Electrical Engineering, Pulchowk Campus, IOE, TU. Email: [email protected]
2

3
Dept. of Electrical Engineering, Pulchowk Campus, IOE, TU. Email: [email protected]
4
Dept. of Electrical Engineering, Pulchowk Campus, IOE, TU. Email: [email protected]
5
Professor, Dept. of Electrical Engineering, Pulchowk Campus, IOE, TU. Email: [email protected]

Abstract—With the increasing concern for EV and Designing a battery pack that is both safe and capable of
its battery management system, the effective life and providing adequate energy is an extremely challenging
performance of li-ion battery is of utmost importance. task. Typically, a battery pack needs to supply direct current
The battery cells should be frequently equalized in order (DC) voltage in the range of hundreds of volts and deliver
to increase the lifetime of battery pack. The conventional power in the range of hundreds of kilowatts to the vehicle's
method of cell balancing is passive cell balancing where drive train. To meet these voltage and power requirements,
excess energy from cell is dissipated in the form of heat a considerable number of battery cells are interconnected
until all the cells are equally charged, causing thermal both in parallel and in series. Although these cells are
issues and efficiency in the battery pack. This paper considered identical however the cells in a battery pack
proposes a novel approach for improving the efficiency exhibit mismatches in parameters due to manufacturing
of battery management systems through active cell defects and aging [2]. These fluctuations result in a decrease
balancing using the Kalman filter algorithm thereby in the effective capacity of the battery pack. As a result, it
overcoming the drawbacks of passive cell balancing. is almost always necessary to have a battery management
The goal of this paper is to develop a system that can system (BMS) with an external balancing circuit in order
extend the lifespan of batteries by ensuring that each cell to fully utilize the energy capacity of all the individual
is charged and discharged evenly. The proposed system cells. Additional circuits used for battery balancing are
includes an active balancing circuit that uses the Kalman generally divided into two categories: passive balancing
filter algorithm to estimate the state of each battery cell and active balancing. [3]. Passive balancing is characterized
and determine the optimal charging and discharging as a dissipative procedure, whereas active balancing is
currents. acknowledged as a non-dissipative procedure. In passive
balancing circuits, a shunt resistor is employed to convert
Keywords— State of charge, Lithium-ion battery, Kalman
the energy of a cell into heat energy, thereby safeguarding
filter, Cell balancing, Battery management system.
the cells against overcharging. On the other hand, active
Introduction balancing involves the direct transfer of energy to or from
the cell by utilizing DC/DC converters or other methods of
With the concern of whole world being shifted towards
power transportation.
environmental pollution due to exhaust emissions from
automobiles and rising price of fuels the necessity of Hence by implementing proposed active balancing circuit
deployment of Electric Vehicles have been pronounced. the safety, durability, proper charge and discharge of
One of the most important aspects that has enabled EVs battery packs, optimum utilization of available energy can
to stand as a strong candidate of future of transportation is be enhanced.
the revolution in its battery management system that has
Proposed Scheme
recently been experienced. But still there are lots of areas
in battery management system itself which can be worked
upon to improve its efficiency and reliability [1].
For the reliable and efficient operation of a battery
management system, the importance of accurate estimation
of State of Charge (SOC) and State of Health (SOH) of
battery cannot be underemphasized. SOC is a measure of
available capacity of a battery relative to its fully charged
state while SOH is the indicative of aging level of battery.
It quantifies the energy that battery can store now in a fully
charged state compared to when it was manufactured.
Fig 1. Proposed Block Diagram
* Corresponding Author
24 KEC Journal of Science and Engineering, Vol. 8 Issue 1, August 2024

Figure 1 shows the block diagram of the proposed scheme. models on modern computers that inherently work with
The state of charge (SOC) of each cell is determined using limited numerical representation and discrete processes. The
extended kalman filter algorithm. The controller senses the process of converting the continuous time model to a discrete
SOC of each cell and sends signal to the active cell balancing time state space involved utilizing closed form discretization
circuit to operate accordingly there by transferring the charge formulas on the relevant matrices and vectors from the
from cell with higher SOC to lower SOC and eventually continuous time model. This resulted in the creation of the
equalizing the charge in each cell of a battery pack. discrete state and output equations which are as follows:
A. SOC estimation using extended Kalman filter = = (6)
algorithm:
While SOC can be estimated using various techniques, the
= I (7)
extended Kalman filter is used to estimate the SOC. This is
a repetitive process where the SOC is estimated accounting
for the various noise and errors encountered with the
Now it is ready to apply the Kalman filter algorithm with the
instruments and estimations. It is started by figuring out the
developed state space models.
various property of a battery and their dependencies. Figure
2 shows the equivalent circuit model of a battery which is State space model
designed using a lumped parameter model. (8)
(9)
Where, and are independent, zero mean Gaussian noise
processes of covariance matrices Σw and Σv respectively.

Definitions: Ak = xk=x̂ k+, Ck= xk=x̂ k-

Initialization: for k=0, set x̂ 0+= E[x0], Σ+x̃ ,0=E[(x0-E[x0])(x0-E[x0])T] (10)
Computation: For k= 1, 2, 3…….. compute;
Fig 2 Dual Polarity Model
Time Update:
By utilizing Kirchhoff's Voltage Law (KVL) to analyse
the entire circuit loop, we derive the subsequent equation x̂ k-= f(x̂ k-1+, uk-1 ) (11)
representing the terminal voltage of the circuit. This equation P x̃,k = Ak-1 P
- +
(Ak-1) + Pw
T
(12)
x̃,k
is dependent on the internal components of the circuit.
Measurement Update:
𝑉𝑇=𝑉𝑂𝐶(𝑠𝑜𝑐)−𝑅𝑠𝐼−𝑉1−𝑉2 (1)
Lk = P-x̃,kCkT[Ck P-x̃,k CkT+ Σv]-1 (13)
By applying KCL to both RC branches, the following
equations are derived: x̂ k+ = x̂ k-+Lk[yk-g(x̂k-, uk)] (14)

(2) P x̃,k=(I-LkCk) P x̃,k

+ -
(15)
B. Principle of Operation of Buck Boost Converter
Figure 3 shows the circuit diagram of Buck boost converter It
(3) is a type of DC-DC converter whose output voltage is either
The SOC of the battery has its relationship with the current less or greater than input voltage. When the output voltage
of the circuit by the following equation (4). is less than input, it performs as buck converter and if the
output is greater than input, it performs as boost converter.
(4)
By using the above equations, the state space model is
developed as shown in equation (5):

=
(5)

Fig 3. Buck Boost Converter

Although the continuous time model is a crucial step in
system estimation, it is much simpler to utilize discrete time
 KEC Journal of Science and Engineering, Vol. 8 Issue 1, August 2024 25

Case I: Switch is ON, Diode is OFF to transfer the energy to the Cell N-1 through the path of
While the switch is in the ON state, the inductor stores the diode DN-1. In similar manner the energy is transferred from
charge and the current does not flow through load due to the bottom cell to the top cell if the controller detects that
reverse biasing of the diode. bottom cell has to transfer its energy from bottom cell to top
through switch SN-1 .
Simulation Results

Fig 5. Switch is ON, Diode is OFF

Case II: Switch is OFF, Diode is ON

Fig 6 Switch is OFF, Diode is ON

In this operating mode, when the MOSFET switch is open, the
Fig 7. Plot of actual SOC and estimated SOC vs time
polarity of inductor current reverses causing it to discharge.
The energy is then dissipated in the load resistance, which In the figure 7, both estimated SOC and actual SOC are
helps to keep the current flowing in the same direction through plotted together. This figure shows that the plot of estimated
the load. Furthermore, the inductor now acts as a source in SOC vs time is same as the plot of actual SOC vs time
addition to the input source, resulting in an increase in the which indicates that extended kalman filter algorithm is
output voltage. successfully implemented in matlab to estimate the state of
C. Active Cell Balancing Circuit charge of the cell.

Fig 6. Active cell balancing circuit diagram using dc-dc converter

Figure 6 shows the circuit diagram of Active Cell Balancing
system. Here, the controller senses the SOC imbalance
between cells and distinguishes the cell from where the Fig 8. Simulation of active cell balancing circuit in matlab using buck
charge transfer should take place. The controller sends the boost converter
output signal to turn on the switch of the cells. In the above Figure 8 shows the Matlab simulation model of active cell
diagram if the controller senses that top cell N needs to balancing circuit with buck-boost converter. The upper
transfer its energy to bottom cell N-1, it sends signal to switch battery was set to 23% SOC and the lower battery was set
SN. The signal sent will be a PWM of certain frequency and to 20% SOC. After simulation, the final balanced SOC was
duty cycle. The inductor starts to store energy and after the obtained as 21.39% in upper cell and 21.4% in lower cell
energy stored in the inductor reaches a maximum value the which is almost equal to the average value of SOC contained
signal sent to the switch SN is turned off. After the switch initially in each cell. Hence, the equalization of charge in
is turned off the inductor voltage reverses in the direction each cell has been successfully carried out using the active
and diode DN-1 becomes forward biased. The inductor starts cell balancing circuit.
26 KEC Journal of Science and Engineering, Vol. 8 Issue 1, August 2024

The following observations are made varying the parameters hazards. However, it also highlighted the need for careful
of different components: consideration of the specific requirements of battery systems
TABLE 1 and applications in order to determine the most appropriate
VARYING THE VALUE OF L AND KEEPING PERIOD =1.5S AND active cell balancing system.
50% DUTY CYCLE
References
L(inductance) Time taken to Final SOC (%) [1] F. Zhang, "Modeling and control of a modular battery
in H balance in sec management system for lithium-ion battery packs,"
1 423 21.45 University of Colorado at Boulder, 2017.
0.5 228 21.4 [2] Cell-Balancing Techniques: Theory and Implementation,
2013.
0.1 80 21.02
[3] H. Sugumar , "Overview of cell balancing methods for Li-
0.01 39 20.16 ion battery technology," Energy storage, p. 12, 2020.
0.001 34 21.5 [4] Qays, M. O. Y. Buswig, M. L. Hossian, M. M. Rahman and
A. Abu-Siada, "Active cell balancing control strategy for
TABLE 2
parallelly connected LiFePO 4 batteries," CSEE Journal of
VARYING THE VALUE OF PERIOD KEEPING L=0.5H AND Power and Energy Systems 7.1, pp. 86-92, 2020.
DUTY CYCLE 50% [5] Reynaud, J. C. Carrejo, O. Gantet, P. Aloisi, B. Estibals
and C. Alonso, "Active balancing circuit for advanced
Time taken to balance lithium-ion batteries used in photovoltaic application," in
Period (s) Final SOC (%)
in sec Proceedings of International Conference on Renewable
Energies and Power Quality, Spain, 2010.
1 329 21.44
[6] Sreejyothi, R. Khammampati, K. Chenchireddy, S. A. Sydu,
1.5 228 21.4 V. Kumar and W. Sultana , "Bidirectional Battery Charger
2 187 21.36 Circuit using Buck/Boost Converter," in 6th International
Conference on Electronics, Communication and Aerospace
2.5 143 21.34 Technology, 2022.
TABLE 3 [7] Y. Jin, Z. Wenchun, L. Zhixin, L. Baolong and L. Liu,
"Modeling and simulation of lithium-ion battery considering
VARYING THE DUTY CYCLE KEEPING PERIOD=1.5S AND L=0.5H
the effect of charge-discharge state," Journal of Physics:
Duty cycle (%) Time taken to balance Final SOC (%) Conference Series, vol. 1907.No.1, 2021.
in sec [8] J. Dorsey and F. Valdez, "Extended kalman filtering of state
and sensor bias estimation of li-ion battery model".
30 594 21.45
[9] T. Duraisamy and D. Kaliyaperumal, "Active cell
40 340 21.43 balancing for electric vehicle battery management system,"
International Journal of Power Electronics and Drive
50 228 21.4 Systems, p. 571, 2020.
60 72 21.2 [10] "robu.in," [Online]. Available: https://robu.in/introduction-
to-buck-boost-converter/. [Accessed 24 April 2023].
70 51 20.93

Hence the overall result shows that there is a trade-off

between the time taken to balance and the final balanced
SOC.
Conclusions
The paper was focused on the topic of active cell balancing
which is a technique used to balance the charge levels of
individual cells in a battery pack. Upon the completion
of project, an active balancing circuit was designed and
simulation of circuit was done to obtain the required result.
A number of methods were studied for SOC estimation
of individual cell, out of which extended Kalman filter
method was adopted because of its accuracy in estimating
non-linear parameters. Overall, the project was successful
in demonstrating the effectiveness of active cell balancing
in improving battery performance and reducing safety