Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

CHAPTER 1
INTRODUCTION
While several recent studies have focused on eliminating the imbalance of energy stored in
series-connected battery cells, very little attention has been given to balancing the energy
stored in parallel-connected battery cells. As such, this paper aims at presenting a new
balancing approach for parallel LiFePO4 battery cells. In this regard, a Backpropagation
Neural Network (BPNN) based technique is employed to develop a Battery Management
System (BMS) that can assess the charging status of all cells and control its operations
through a DC/DC Buck-Boost converter. Simulation results demonstrate the effectiveness of
the proposed approach in balancing the energy stored in parallel-connected battery cells in
which the state of charge (SoC) estimation error is found to be only 1.15%.

In the past decades, smart grids (SGs) and electric vehicles (EVs) have become popular
industry trends around the globe due to their cost effectiveness in the long run. For both
technologies, a Battery Energy Storage System (BESS) will play a significant role for grid
stabilization [1]–[3]. However, a BESS has to overcome several challenges, including cell to
cell factorial delinquencies and cell performance over its entire operational life. Several
Battery Management Systems (BMS) have been proposed to observe the cell parameters, i.e.,
current, voltage, state of charge (SoC) and temperatures to maintain the reliability of the cells
[4], [5]. The primary role of a BMS is to protect the cells from over-charging and/or deep-
discharging and hence extending the battery-life cycle. Usually, cells are predisposed in a
battery pack with a minimum parameter variation for the sake of smooth and long operations.

Fig 1.1represents the different levelled state of charges for battery cells

Fig. 1.1 Different leveled state-of-charges for battery-cells

Technical seminar Report Dept. Of EEE 2023-24 Page 1

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

Fig 1.2 represents Current circulation among the parallel connected cells due to imbalanced
state-of-charge, The performance of a battery pack is significantly impacted due to imbalance
in the energy stored among its cells. Generally, only 10%-20% to 60%-70% SoC is permitted
for a cell to maintain its performance. Outside this range, the cells exhibit over-charge and
depth of discharge (DoD) [6]–[8].

Fig.1.2 Current circulation among the parallel connected cells due to imbalanced state-of-charge

The performance of a battery pack is significantly impacted due to imbalance in the energy
stored among its cells. Generally, only 10%-20% to 60%-70% SoC is permitted for a cell to
maintain its performance. Outside this range, the cells exhibit over-charge and depth of
discharge (DoD) [6]–[8]. Fig.1.1 shows five battery cells of different levels of SoC. Upon
connecting these cells in series/parallel, the cells with the highest SoC become over-charged
during the charging period while the batteries with the lowest SoC experience deep-discharge
during the discharging period. Several recent studies proposed different approaches to reduce
the imbalance of the SoC among series-connected cell strings [9]–[11]. However, not much
attention was given to balancing the SoC of parallel-connected cells as shown in Fig.1.2.
Parallel cells should have the same terminal voltage, despite voltage variations which
ultimately initiate circulating current and SoC variations among the cells [12], [13]

Subsequently, this paper presents a new approach to eliminate the unbalancing factors of
parallel allied Lithium Ion Battery (LIB) cells. Reference [14] investigated the performance
of parallel-connected cell strings and proposed a method to balance the different levels of
SoC. Reference [15] considered normal battery module capacity and charge voltage shift for
parallel-connected LiFePO4 cells and calculated the battery pack capacity using a proposed

Technical seminar Report Dept. Of EEE 2023-24 Page 2

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

BMS. The proposed method resulted in a standard integration difference in the range of 0%-
0.35%. Reference [16] tested the cells internal resistance and its impact on the battery life
cycle for parallel-connected Lithium-Ion battery (LIB) cells. Results concluded that only 20%
of the internal resistance variation could reduce 40% of the battery life cycle.

Reference [17] classified cell balancing topologies into two categories: passive and active cell
balancing. In the case of the passive cell balancing technique, current flows through a resistor
which results in energy loss. In an active cell balancing method, energy is allocated equally
among the battery cells. Thus, power losses can be reduced in many ways, including an active
charge equalization scheme for battery cells [18], [19]. Usually, cell terminal voltage, open-
circuit voltage (OCV) and SoC parameters are required for active cell balancing techniques.
However, the problem of fluctuation of cell terminal voltage cannot be expressly
accomplished for the equalization structure.

A minor difference of OCV among the cells causes a significant variation of the system
performance as OCV is directly related to SoC. In general, LIB comprises a flat terminal
voltage in the range of 20% to 80% of the SoC. As shown in Fig.1.3, the voltage variation in
this range is less than 0.09 V. Such slight voltage variation is hard to be used to persuade a
proper charging and discharging process. Consequently, this paper focuses on active cell
balancing tactics for online SoC assessment. Furthermore, a Backpropagation Neural
Network (BPNN) is widely used for SoC estimation due to its self-handling characteristics.
BMS observes the condition of the cells through assessing their SoC and issues a
charging/discharging/islanding signal via a DC/DC buck-boost converter. In consequence,
BPNN is adopted in this paper to develop a proper BMS for parallel-connected cells.

Fig1.3 the graph of OCV vs SOC for LiFePO4

Fig. 1.3: OCV vs SOC for LiFePO4

Technical seminar Report Dept. Of EEE 2023-24 Page 3

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

CHAPTER 2

PROPOSED METHODOLOGY
2.1:Cell Modelling

A Thevenin equivalent circuit model (ECM) is an effective way to describe the internal
characteristics of the battery cell. To avoid calculation complexity, the first-order Thevenin
model shown in Fig.2.1is proposed in this paper. In this figure, R1//C1 models the time-
delayed polarization resistance and capacitance, charge transfer activation and diffusion
concentration [19], [20]. The OCV (VOC) of the equivalent circuit model is:

VOC = VR + VP + Vout (1)

Vout = (VR + VP) − VOC (2)

Fig. 2.1: First-order equivalent circuit model.

Here, R and VR = iR is the ohmic resistance and ohmic voltage drop respectively, which is
related to the cell temperature, operating current and SoC. VP is the polarized voltage for
R1//C1 and Vout is the terminal or output voltage. The variations of the charging and
discharging conditions are determined as follows.

(3)

Here, the functioning polarization is written off from the equivalent capacity C1 = t/R1 of
R1//C1; where t is the polarization time factor. The aforementioned elements can be easily
identified using pulse tests [21]

Technical seminar Report Dept. Of EEE 2023-24 Page 4

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

2.2 SOC Estimation

For cell strings, reliable SoC estimation is important to confirm its protection scheme and
avoid over-charging and deep-discharging. However, accurate SoC estimation is quite
challenging due to the nonlinear electrochemical reactions. Among existing SoC estimation
techniques, Artificial Neural Network (ANN) especially BPNN shown in Fig.2.2 is widely
used due to its self-handling characteristics [22], [23].

Fig.2.2 represents the BPNN structure for SOC estimation,

Fig.2.2: BPNN structure for SOC estimation

As can be seen in Fig. 2.3, the proposed SoC method in this paper employs the current,
voltage and temperature of each cell to estimate the SoC of individual cells by using (4) to (7)
to minimize the Root Mean Square Error (RMSE) of the estimated SoC, BPNN structural
weights are adjusted to achieve the desired minimum error [24].

Fig. 2.3, represents the SoC estimation for individual cells,

Fig. 2.3: SoC estimation for individual cells.

Technical seminar Report Dept. Of EEE 2023-24 Page 5

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

(4)

Here, netjn is the input of the hidden layer neuron n, the input to the hidden neuron is ai, and
win is the weight between input neuron j to hidden neuron n, bias of the hidden neuron n is pn.

(5)

(6)
where hn is the applied tangent activation function of the hidden layer neuron n, total input of
the output neuron S is nety, hm contains the value for hidden neuron to output neuron S, gm
shares weight for hidden neuron m to output neuron y, and q is the bias of the output layer
neuron y. The Sigmoid function is the implemented activation function for the output layer
neuron. Output function S can be calculated from:

(7)

2.3 Buck-boost Converter

A circuit diagram of the implemented synchronous Buck-Boost converter is shown in

Fig.2.5[25], [26]. For standardizing the energy conveyance of cell sequences, four MOSFET
switches (M1-M4) are involved in the DC/DC converter. The converter can be utilized as a
Buck, Boost or Buck-Boost converter. The converter is set-back to buck and boost mode
procedure according to the anticipated cell voltage. To carry out the Buck mode, M3 is
permanently turned off while M4 is turned on. On the other hand, the operation of M1 and
M2 are subject to a governor formula. In Fig.2.5, the source power charges the inductor by
closing and opening the MOSFET switches M1 and M2 respectively. The capacitor delivers
the output current at charging duration. M2 and M1would be closed and opened respectively
during the inductor discharging period. For charging the load, the inductor provides reserved
energy to the capacitor. The average terminal voltage V Ter = DVSC, can be attained by
adjusting the duty cycle D to be less than 1.

Technical seminar Report Dept. Of EEE 2023-24 Page 6

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

Fig. 2.4,represents the four switched DC/DC Buck-Boost converter,

Fig.2.4:Four switched DC/DC Buck-Boost converter.

Conversely, during Boost mode action, MOSFET switches M1 and M2 are permanently
turned on and off respectively, while M3 and M4 are closed and opened continuously
following the inductor charging status. This action is inverted during the discharging period.
Average output voltage Vout is 1/ (1 − D) V SC while the output voltage during the Buck-
Boost operation is Vout = D/(1−D) VSC. For this mode, MOSFET switches M1 and M3 are
turned off while M2 and M4 are turned on to charge the inductor. This action is reversed
during inductor discharging.

2.4 Proposed Cell Balancing Technique

Battery pack capacity can be increased by connecting cell strings in parallel or in series.
Theoretically, the terminal voltages of individual parallel cells should be identical [18], [26].
To maintain the reliability of the battery pack, charging and discharging cycles are triggered
as soon as a cell SoC reaches a pre-defined cut-off region, even if its terminal voltage is the
same as the other cells. An active cell balancing scheme is developed to diminish the SoC
deviation among the cells. BMS estimates the SoC from each cell and issues a balancing
signal through the DC/DC synchronous Buck-Boost converter shown in Fig. 2.4. In this
technique, the average SoC (S¯) of the parallel-connected cells is compared to the SoC of
each cell (Si). If the difference is more or less than a predefined value (S thr), BMS commands
the converter to proceed with the charging/discharging action as per the equation below:

(8)

Technical seminar Report Dept. Of EEE 2023-24 Page 7

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

CHAPTER 3

SIMULATION AND DISCUSSION

The proposed active cell-balancing topology, along with the battery equivalent circuit model,
as shown in Fig.3.1 is simulated in the MATLAB/SIMULINK environment to validate the
effectiveness of the proposed BMS. A LiFePO4 cell is employed to assess the proposed
active cell balancing methodology. Three parallel-connected A123 Li-iron phosphate
ANR26650 M1 cells, with nominal voltage of 3.3 V and 2.3 Ah capacity of which the OCV
vs SoC, are presented in literature as shown in Fig.1.3 [27]. The collected battery data is
trained using the MATLAB/Simulink toolbox.

Fig. 3.1, represents the active charge equalization structural design,

Fig.3.1. Active charge equalization structural design

As can be seen in Fig.3.2, the overall regression for Training, Validation, Testing and Overall
regression (R) is over 99%.

Fig. 3.2 (a) Training Fig.3.3(b) Validation

Technical seminar Report Dept. Of EEE 2023-24 Page 8

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

Fig.3.4(c) Testing Fig.3.5 (d) Overall regression for Soc estimation.

Consequently, the actual and estimation SoC are almost similar as can be seen from Fig. 10

Fig.3.6 Actual and Estimated Battery SoC

Fig. 11 shows 1.15% RMSE for the SoC estimation using the proposed BPNN. Compared
with the estimation methods published in literature that include the Invariant-Imbedding
Method (IIM), Extended Kalman Filter (EKF) and Unscented Kalman filter (UKF), the
estimation error of the proposed BPNN in this paper is the lowest, as can be observed from
Table I.

Technical seminar Report Dept. Of EEE 2023-24 Page 9

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

Fig.3.7 RMSE of the BPNN proposed model.

Table I: Units comparison of SoC estimated methodologies

To validate the robustness of the proposed balancing technique, three cells are assumed to
initially exhibit different SoC; 54%, 52%, and 50%. The proposed BMS analyses these SoC
levels and issues a signal to the DC/DC Buck-Boost converter to either charge or discharge
its individual stored energy in order to achieve SoC balance among the three cells. Fig. 12 to
Fig. 14 illustrate the effectiveness of the proposed active cell balancing technique through the
cells’ voltages, currents and SoC without and with the implementation of the proposed
methodology.

Technical seminar Report Dept. Of EEE 2023-24 Page 10

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

Fig.3.8 Cells voltages (a) without the Fig.3.9 Cells voltages (b) with the application of
application of the proposed method the proposed method.

Fig. 3.8(a) shows that although the terminal voltage for the three parallel cells should be the
same, due to different SoC, the terminal voltages are different without the implementation of
any BMS scheme which may result in circulating current among the cells as can be observed
in Fig. 3.9(a). Employing the proposed BMS will bring cell voltages to almost the same level
as can be seen in Fig. 3.10(b) which can limit the circulating current as can be seen in
Fig.3.11(b). Fig. 3.12shows the SoC before and after the implementation of the proposed
technique. Before the implementation, the three cells exhibit imbalance to escalate the SoC
levels of 54% 52%, and 50%, as can be shown in Fig. 3.12(a). With the implementation of
the proposed technique, the SoC of the three cells become closer and rise up to 78% as shown
in Fig. 3.13(b). The detailed flowchart of the proposed topology is shown in Fig.3.14.

Fig. 3.10. Cells currents (a) without the Fig.3.11. Cells currents (b) with the application

Technical seminar Report Dept. Of EEE 2023-24 Page 11

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

application of the proposed method of the proposed method

Fig. 3.12Cell SoC (a) without the application Fig. 3.13Cell SoC (b) with the application of
of the proposed method. the proposed method.

FLOWCHART OF PROPOSED SCHEME:

The flowchart of the proposed scheme is represented,

Fig.3.14 Flowchart of the proposed active cell balancing scheme.

Technical seminar Report Dept. Of EEE 2023-24 Page 12

Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

CHAPTER 4

CONCLUSION
An active cell balancing methodology based on adaptive BPNN to estimate the state of
charge of battery cells is presented and implemented on LiFePO4 battery cells connected in
parallel. The proposed BMS assesses the SoC level of each cell and drives an activation
signal to the DC/DC Buck-Boost converter to equalize the SoC of all cells. The activation
signal triggers charging, discharging or islanding cell modes succeeding the direction of
BMS. The key conclusions of the obtained results are as follows:

• The proposed BMS can minimize the SoC estimation error to 1.15% which surpasses the
accuracy of other estimation models published in literature.

• The proposed active cell balancing approach eliminates the SoC discrepancies among the
connected cells, eliminating the circulating current among parallel connected cells.

• The proposed technique is simple and easy to implement within parallel connected cells and
can be modified for series connected cells as well. This technique is essential to maintain the
reliability of series/parallel connected cells and improve its performance by lengthening its
operational life cycle.

• If the proposed scheme fails to moderate the charging levels, and SoC of a battery string, it
is suggested to replace the weakest cell of the battery pack. Further research to validate the
proposed technique will be conducted experimentally Training

to prove its practical feasibility.

Technical seminar Report Dept. Of EEE 2023-24 Page 13

 Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

REFERENCES
[1] M. J. M. Al Essa, “Management of charging cycles for grid-connected energy storage
batteries,” Journal of Energy Storage, vol. 18, pp. 380– 388, Aug. 2018.
[2] C. A. Correa-Florez, A. Gerossier, A. Michiorri, and G. Kariniotakis, “Stochastic operation of
home energy management systems including battery cycling,” Applied Energy, vol. 225, pp.
1205–1218, Sep. 2018.
[3] E. Hossain, M. Zawad, K. H. Rakibul Islam, and M. Q. Akash, “Design a novel controller for
stability analysis of microgrid by managing controllable load using load shaving and load
shifting techniques; and optimizing cost analysis for energy storage system,” International
Journal of Renewable Energy Research, vol. 6, no. 3, pp. 772–786, Jan. 2016.
[4] W. Sung and J. Lee, “Improved capacity estimation technique for the battery management
systems of electric vehicles using the fixed-point iteration method,” Computers & Chemical
Engineering, vol. 117, pp. 283–290, Sep. 2018.
[5] G. N. Xu, X. W. Du, Z. J. Li, X. J. Zhang, M. X. Zheng, Y. Miao, Y. Gao, and Q. S. Liu,
“Reliability design of battery management system for power battery,” Microelectronics
Reliability, vol. 88–90, pp. 1286– 1292, Sep. 2018.
[6] E. Wikner and T. Thiringer, “Extending battery lifetime by avoiding high SOC,” Applied
Sciences, vol. 8, no. 10, pp. 1825, Oct. 2018.
[7] S. Ci, N. Lin, and D. L. Wu, “Reconfigurable battery techniques and systems: a survey,” IEEE
Access, vol. 4, pp. 1175–1189, Mar. 2016.
[8] Q. Ouyang, J. Chen, J. Zheng, and H. Z. Fang, “Optimal cell-to-cell balancing topology design
for serially connected lithium-ion battery packs,” IEEE Transactions on Sustainable Energy,
vol. 9, no. 1, pp. 350–360, Jan. 2018.
[9] X. D. Cui, W. X. Shen, Y. L. Zhang, C. G. Hu, and J. C. Zheng, “Novel active LiFePO4 battery
balancing method based on chargeable and dischargeable capacity,” Computers & Chemical
Engineering, vol. 97, pp. 27–35, Feb. 2017.
[10] M. Caspar, T. Eiler, and S. Hohmann, “Systematic comparison of active balancing: a model-
based quantitative analysis,” IEEE Transactions on Vehicular Technology, vol. 67, no. 2, pp.
920–934, Feb. 2018

[11] Y. M. Ye and K. W. E. Cheng, “Analysis and design of zero-current switching switched-

capacitor cell balancing circuit for series-connected battery/supercapacitor,” IEEE
Transactions on Vehicular Technology, vol. 67, no. 2, pp. 948–955, Feb. 2018.

[12] T. Bruen and J. Marco, “Modelling and experimental evaluation of parallel connected lithium
ion cells for an electric vehicle battery system,” Journal of Power Sources, vol. 310, pp. 91–
101, Apr. 2016. [13] T. Grun, K. Stella, and O. Wollersheim, “Influence of circuit design ¨ on
load distribution and performance of parallel-connected Lithium ion cells for photovoltaic
home storage systems,” Journal of Energy Storage, vol. 17, pp. 367–382, Jun. 2018.

[14] M. Dubarry, A. Devie, and B. Y. Liaw, “Cell-balancing currents in parallel strings of a battery
system,” Journal of Power Sources, vol. 321, pp. 36–46, Jul. 2016.

[15] L. M. Wang, Y. Cheng, and X. L. Zhao, “A LiFePO4 battery pack capacity estimation
approach considering in-parallel cell safety in electric vehicles,” Applied Energy, vol. 142,
pp. 293–302, Mar. 2015.

Technical seminar Report Dept. Of EEE 2023-24 Page 14

 Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

[16] R. Gogoana, M. B. Pinson, M. Z. Bazant, and S. E. Sarma, “Internal resistance matching for
parallel-connected lithium-ion cells and impacts on battery pack cycle life,” Journal of Power
Sources, vol. 252, pp. 8–13, Apr. 2014.

[17] V. Knap, D. I. Stroe, A. E. Christensen, K. Propp, A. Fotouhi, D. J. Auger, E. Schaltz, and R.

Teodorescu, “Self-balancing feature of Lithium-Sulfur batteries,” Journal of Power Sources,
vol. 372, pp. 245– 251, Dec. 2017.

[18] H. B. Ren, Y. Z. Zhao, S. Z. Chen, and T. P. Wang, “Design and implementation of a battery
management system with active charge balance based on the SOC and SOH online
estimation,” Energy, vol. 166, pp. 908–917, Jan. 2019.

[19] W. P. Diao, N. Xue, V. Bhattacharjee, J. C. Jiang, O. Karabasoglu, and M. Pecht, “Active

battery cell equalization based on residual available energy maximization,” Applied Energy,
vol. 210, pp. 690–698, Jan. 2018.

[20] Q. Q. Wang, J. Wang, P. J. Zhao, J. Q. Kang, F. Yan, and C. Q. Du, “Correlation between the
model accuracy and model-based SOC estimation,” Electrochimica Acta, vol. 228, pp. 146–
159, Feb. 2017. [21] M. Bottiger, M. Paulitschke, and T. Bocklisch, “Systematic
experimental ¨ pulse test investigation for parameter identification of an equivalent based
lithium-ion battery model,” Energy Procedia, vol. 135, pp. 337– 346, Oct. 2017.

[22] M. A. Hannan, M. S. H. Lipu, A. Hussain, M. H. Saad, and A. Ayob, “Neural network

approach for estimating state of charge of lithium-ion battery using backtracking search
algorithm,” IEEE Access, vol. 6, pp. 10069–10079, Jan. 2018.

[23] A. B. Ahmad, C. A. Ooi, D. Ishak, and J. Teh, “State-of-charge balancing control for
ON/OFF-line internal cells using hybrid modular multi-level converter and parallel modular
dual l-bridge in a grid-scale battery energy storage system,” IEEE Access, vol. 7, pp. 131–
147, Dec. 2018.

[24] O. Qays, Y. Buswig, and M. Anyi, “Active cell balancing control method for series-connected
lithium-ion battery,” International Journal of Innovative Technology and Exploring
Engineering, vol. 8, no. 9, pp. 2424–2430, Jul. 2019.

[25] J. K. Shiau and C. W. Ma, “Li-Ion battery charging with a buckboost power converter for a
solar powered battery management system,” Energies, vol. 6, no. 3, pp. 1669–1699, Mar.
2013.

[26] F. Q. Chang, F. Roemer, M. Baumann, and M. Lienkamp, “Modelling and evaluation of

battery packs with different numbers of paralleled cells,” World Electric Vehicle Journal, vol.
9, no. 1, pp. 8, Jun. 2018.

[27] Z. C. Gao, C. S. Chin, W. L. Woo, and J. B. Jia, “Integrated equivalent circuit and thermal
model for simulation of temperature-dependent LiFePO4 battery in actual embedded
application,” Energies, vol. 10, no. 1, pp. 85, Jan. 2017.

[28] G. Z. Dong, J. W. Wei, C. B. Zhang, and Z. H. Chen, “Online state of charge estimation and
open circuit voltage hysteresis modeling of LiFePO4 battery using invariant imbedding
method,” Applied Energy, vol. 162, pp. 163–171, Jan. 2016.

Technical seminar Report Dept. Of EEE 2023-24 Page 15

 Active cell balancing control strategy for parallelly connected LiFeP04 Batteries

[29] X. J. Dang, L. Yan, H. Jiang, X. R. Wu, and H. X. Sun, “Open-circuit voltage-based state of
charge estimation of lithium-ion power battery by combining controlled auto-regressive and
moving average modelling with feedforward-feedback compensation method,” International
Journal of Electrical Power & Energy Systems, vol. 90, pp. 27–36, Sep. 2017.

[30] C. Campestrini, S. Kosch, and A. Jossen, “Influence of change in open circuit voltage on the
state of charge estimation with an extended Kalman filter,” Journal of Energy Storage, vol.
12, pp. 149–156, Aug. 2017.

[31] W. D. Wang, X. T. Wang, C. L. Xiang, C. Wei, and Y. L. Zhao, “Unscented kalman filter-
based battery SOC estimation and peak power prediction method for power distribution of
hybrid electric vehicles,” IEEE Access, vol. 6, pp. 35957–35965, Jun. 2018.

Technical seminar Report Dept. Of EEE 2023-24 Page 16