Development and Verification of the Equilibrium Strategy for Batteries in Electric Vehicles

Rui Xiong and Yanzhou Duan

(National Engineering Laboratory for Electric Vehicles, School of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, China)

Under the current technology level, the battery of the electric vehicle must meet the demand of the capacity and voltage through a large amount of cells in series and parallel connection. However, because of the limit of the level of manufacturing, the capacity, internal resistance and other parameters between cells may be different even the type, the size and the batch are the same. The imbalance will make the performance of the battery group can not achieve the desired level. Some cells may overcharge or over discharge, which will result in energy loss and reduce service life[1].Therefore, it is necessary to develop a proper equilibrium strategy to improve the inconsistency of the batteries in time.

At present, the research of equilibrium strategy mainly focus on the selection of the equilibrium criterions. The equilibrium criterion determines when the equilibrium is turned on. It is chosen mainly from the open circuit voltage (OCV), terminal voltage, internal resistance, SOC and some other parameters. In these parameters, the OCV can only be used offline. The internal resistances of batteries are also difficult to measure in actual operation. And because of the existence of polarization resistance, the internal resistance of the same battery varies greatly under different working conditions. Therefore, the internal resistance is also not suitable as a criterion of equalization.

The relatively mature equilibrium criterions include battery terminal voltage and SOC at present. Because of the advantages of small computation and easy parameter acquisition, the strategy based on terminal voltage is easier to implement than SOC.However, when the vehicle is running, the battery voltage fluctuates greatly, so the terminal voltage can not reflect the change of the battery state well. Compared with the voltage, the SOC can be a good reflection of the remaining capacity of the battery. The equilibrium strategy that bases on the SOC can effectively manage the capacity of the individual cells, which ensures that the capacity of the battery pack can be utilized maximally[2].However, the lack of SOC estimation accuracy and the complexity of the algorithm are the main limiting factors.

In general, the current equilibrium strategies based on the single equilibrium criterion have some limitations.So a hybrid equilibrium strategy based on SOC and terminal voltage is proposed. An HIL platform is also been constructed to verify the equilibrium strategy. Compared with the traditional equilibrium strategy based on a single criterion, this equilibrium strategy can avoid the overcharge and over-discharge of cells and ensure that the battery pack can release the available capacity as much as possible.

In Section 1, the equilibrium strategy is described in detail. An on-line parameter identification and SOC determining method is introduced. In Section 2, based on MPC5644A and LTC6804-1, the HIL platform is constructed. Then, equilibrium strategy is been verified under CCCV and DST conditions. Finally, the conclusions are drawn in Section 4.

1 Equilibrium Strategy and SOC Emission

1.1 Equilibrium strategy

The current equalization circuit includes active equalization and passive equalization circuit. Active equalization is achieved by energy transfer with big equalizing current and no energy loss, but it has the complex structure and poor stability. Passive equalization is achieved by converting the excess energy into heat energy. The equilibrium strategy in this paper is designed based on the passive circuit mainly for the following reasons:

①The initial consistency of the battery pack performs well. When the batteries are packed, the cells with good consistency should be selected for group, and the inconsistent of batteries forms actually slowly. Although the equilibrium current of the passive equalization is small, equalizing regularly to eliminate the small differences timely can also meet the consistency requirements of packs.

②Passive equalization has several advantages, such as good reliability, simple structure and low cost.

③Energy loss has little effect on actual use. On the one hand, the equalization commonly happens during charging rather than discharging. During charging, the lost energy is from the external energy source, so it has no effect on the performance of the pack. On the other hand, this part of the energy loss is negligible for the entire battery pack.

Therefore, the equilibrium strategy based on passive equalization circuit designed in this paper is as follows.

① During CCCV charging, the terminal voltage is used as an equilibrium criterion. The traditional equilibrium system, which uses SOC as the only equilibrium criterion, has the following questions: 1) When the battery voltage is higher than 4 V, SOC is more than 80%. The current sampling error is about 40 mA. But the current in the end of CCCV charging process is only about tens of milliamperes.So the relative error of current sampling and SOC estimation error increase. 2) When SOC is high, the voltage rising rate becomes large. Besides,because of aging, the difference in capacity between cells increases. In these cases, if cells are equalized based on SOC at the end of the CCCV charging, the voltage difference between cells may be too large. The three cases mentioned above will lead that some voltages ofcells are over the cut-off voltage. So during the CCCV charging, the terminal voltage should be used as the equalization criteria.

②Under the dynamic conditions, when the SOC is between 10% and 80%, the SOC is used as the equilibrium criterion. The SOC estimation method used in this paper has high accuracy when SOC is in the range of 10% to 80%. Therefore, within this range, choosing SOC as the equilibrium criterion can effectively reflect the battery inconsistency.

③When the voltage is below 3.6 V, the pack is no longer equalized. When the voltage is lower than 3.6 V, the SOC is lower than 10%, and a lot of SOC estimation methods have large error in this range relatively[3]. Besides, because of the low power at this time, if passive equalization is turned on, the battery is likely to be over discharged.

Fig.1 Equilibrium strategy

1.2 SOC estimation

According to this paper, only the terminal voltage, current and temperature can be obtained by measuring. Besides, the sampling interval is 2 s. Therefore, the parameter identification and SOC determining method can only be carried out based on these three inputs at the large sampling interval. Ref.[4] proposes a model-based online parameter identification method. This method uses the equivalent circuit model to identify the OCV, and then obtains the SOC based on the one-to-one correspondence between the OCV and SOC. This method is applicable to the condition that only the terminal voltage, current and temperature can be measured. However, this method requires that the sampling interval be small. It ignores the change in the OCV of the battery during the sampling interval and considers that the OCV of the adjacent two sampling points are almost equal. The experimental results show that this method has higher accuracy at small sampling interval, but the error of SOC estimation will exceed 10% when the sampling interval is large. So it is not applicable to the equalization system designed in this paper.

In this paper, an improved on-line parameter identification and SOC determining method based on Thevenin model is adopted. Compared with the traditional method, this method can still guarantee high accuracy under different sampling interval. It can quickly converge to the correct value when the sampling error occurs at the initial moment or when it is disturbed.The specific content of this method is as follows[5].

For the Thevenin model shown in Fig.2, based on Kirchhoff’s Law, we can get

UOCV=Ut+iLR0+U1

(1)

(2)

whereUOCVis the OCV,Utis the terminal voltage,iLis the load current,R0is the equivalent ohmic resistance,U1is the voltage across the RC network,R1is the resistance in the RC network,C1is the captance in the RC network.

Fig.2 Thevenin model

Then Laplace transformation and bilinear transformation are carried out based on Eq.(1) and Eq.(2), which is shown as follows.

Laplace transform,

(3)

(4)

(5)

Bilinear transformation,

(6)

Define

(7)

Define

M(k)=UOCV(k)-a1UOCV(k-1)
Ut(k)=M(k)+a1Ut(k-1)+a2iL(k)+
a3iL(k-1)

(8)

Define

and

(9)

Based on Eq.(9), the battery parameter identification process is as follows:

y(k)=φ(k)θ(k)+e(k)

(10)

(11)

(k)=(k-1)+K(k)[y(k)-φ(k)(k-1)]

(12)

(13)

In view ofM(k)=UOCV(k)-a1UOCV(k-1), we have

UOCV(k)=M(k)+a1UOCV(k-1)

(14)

It is found thatM(k),a1are all known by recursive least squares (RLS).Therefore, it is found that the OCV can be calculated by knowing the initial OCVUOCV(0) at any time.According to the one-to-one correspondence between the OCV and SOC, the SOC can be finally obtained.

On the whole, the improved on-line parameter identification and SOC determining method can meet the requirements of the estimation accuracy at different sampling interval. At the same time, the calculation is moderate for microcontrollers.

2 HIL Platform

Based on the discussion of Section 1, we need to design passive equalization hardware circuit. In order to reduce the complexity of the hardware circuit and improve the overall integration of components, LTC6804-1 is eventually selected as the core chip of the equalization and acquisition circuit. This chip can collect the voltage of 12 batteries at the same time[6].

Take the 12th cell as an example, the equalization circuit used in this paper is shown in Fig.3. This part of circuit is connected in parallel with the 12th battery. The CELL12 and CELL11 are respectively connected to the positive electrode and the negative electrode of the 12th battery. The three wires on the right are connected to the C12, C11 and S12 pins of the LTC6804-1 chip respectively. The core of this circuit is R2, which is a 33 Ω dissipative resistor. The LTC6804-1 collects the voltage of the 12th battery through the inputs of the C12 and the C11. When it is necessary to equalize according to the strategy, the chip will turn on the transistor by controlling S12 pin. Then, the dissipation resistance will be connected with the 12th battery in parallel. The current flows through the resistance so that excess electricity of the 12th battery will be converted into heat.

Fig.3 Equalization circuit

In order to control the LTC6804-1 chip and send the data collected to the host computer, it is necessary to design a matching control circuit. The MPC5644A is chosen as the core chip and the power module, communication module and other peripheral circuits are designed.

The structure of the entire equalization experimental platform is shown in Fig.4.

Fig.4 Equalization experimental platform

3 Experimental Verification

In order to shorten the experimental time, improve the equalization effect and ensure the safety of the experiment, Lishen 18 650, which has small capacity, is selected in the experiment. The standard capacity of each cell is about 2.5 A·h, the lower cut-off voltage is 3 V, and the upper cut-off voltage is 4.2 V.

3.1 Battery characteristic experiment

Before the equalization experiment, it is necessary to do characteristic experiments to get the corresponding relation between the OCV and SOC, which is used in the SOC determining method. Battery characteristic experiments include capacity experiment and OCV experiment.

3.1.1Capacity experiment

The capacity experiment is used to measure the charge and discharge capacity of the battery. This experiment is carried out at room temperature about 25 ℃.The procedure is shown in Tab.1. The experimental results are shown in Tab.2.

3.1.2OCV experiment

In order to obtain the OCV-SOC fitting curve, the OCV experiment is required. The procedure is shown in Tab.3.Since the rest time is long enough, the terminal voltage measured after resting for 2 h is approximately the OCV. The OCV-SOC curve can be obtained by fitting the OCV got from each sampling points.

Tab.1 Capacity experimental procedure

Tab.2 Capacity of cells

Tab.3 OCV experimental procedure

3.2 Equilibrium experiment and result analysis

Based on this platform, equilibrium strategy is been verified under CCCV and DST conditions.

3.2.1CCCV experiment

The cells in series are equalized based on terminal voltages under CCCV condition.

The voltage curves of each cell at the end of CCCV charging is shown in Fig.5. The voltage rising rate of each cell varies a lot when the voltages are over about 3.85 V. When the constant current charging is converted to the constant voltage charging, the voltage change becomes very small. As the passive equalization continues, the terminal voltages of batteries gradually converge during the constant voltage charging time.

Fig.5 Equilibrium voltage curves of 12 cells under CCCV charging condition

The initial and final terminal voltages of the 12 cells are shown in Tab.4.

Tab.4 Terminal voltages before and after equalization under CCCV charging condition

The initial voltage difference between two batteries reaches a maximum of 100 mV. After CCCV charging, the difference between cells is almost zero. So the voltage inconsistency is eliminated. The actual equalization time is about two hours and the equilibrium effect is expected.

3.2.2DST experiment

In order to test the SOC-based equilibrium strategy, this paper uses DST condition for experimental verification. Firstly, the SOC calibration of each cell is carried out according to the results of capacity experiment. The maximum SOC difference between cells is 10%.Then, the battery is equalized based on SOC under DST condition. The experimental process is shown in Fig.6. The initial and final SOC of each cell are shown in Tab.5.

Fig.6 Equilibrium SOC curves of 12 cells in DST condition

CellnumberInitialSOC/%FinalSOC/%180502805038251482515845368453786538865398854108854119056129055

After almost 15 DST cycles, we can see that the inconsistency of SOC between cells has been greatly improved. At the beginning, the maximum SOC difference between cells reaches 10%. After equalization, the maximum difference is 6%, which decreases by 4%.

The equilibrium current of this system is about 100 mA. For batteries with large capacity, the experimental process may be a little longer and the equilibrium effect is not obvious enough. However, in fact, as long as the regular equalization is guaranteed, the inconsistency of each cell can be kept within a controllable range, so there is no need for large current.

4 Conclusion

A hybrid equilibrium strategy based on decision combing battery SOC and voltage has been proposed. This equilibrium strategy avoids the limitations of traditional strategies based on single equilibrium criterion. It can choose different equilibrium criterion according to different working conditions. Under CCCV condition, it is equalized based on terminal voltage to avoid the overcharge and over-discharge of each cell. When the batteries are in dynamic condition, it chooses SOC as an equilibrium criterion. The HIL platform has high acquisition accuracy. The error of voltage acquisition is about 1mV, the current acquisition error is about 0.1% and SOC estimation error is less than 4%. Combined with the passive equali-zation circuit, this hybrid equilibrium strategy has been verified under CCCV and DST conditions. Experimental results indicate that the strategy can achieve a good equalization effect when the initial voltage or SOC is not consistent.

[1] Wang Z, Sun F, Lin C. Analysis of the influence of inconsistency on the service life of power battery pack [J]. Transactions of Beijing Institute of Technology, 2006, 26(7):577-580. (in Chinese)

[2] Chen Y, Liu X, Cui Y, et al. A multi-winding transformer cell-to-cell active equalization method for lithium-ion batteries with reduced number of driving circuits [J]. IEEE Transactions on Power Electronics, 2016, 31(7):4916-4929.

[3] Li S, Zhang C. Research on prediction method of lithium-ion battery charge state [J]. Transactions of Beijing Institute of Technology,2012,32(2):125-129,145. (in Chinese)

[4] He H, Zhang X, Xiong R, et al. Online model-based estimation of state-of-charge and open-circuit voltage of lithium-ion batteries in electric vehicles[J]. Energy, 2012, 39(1):310-318.

[5] Li Z, Xiong R, He H. An improved battery on-line parameter identification and state-of-charge determining method [J]. Energy Procedia, 2016, 103:381-386.

[6] Li C. The voltage acquisition and processing based on LTC6804 [R]. Phuket,Thailand:Advanced Science and Industry Research Center,2014.