Researcvh On The Safety Of LiFePO₄ Battery Pack

With the increasing urgency of both energy crisis and environmental pollution, electric vehicles (EVs) and hybrid electric vehicles (HEVs) have been researched and promoted. Electric vehicle technology is considered one of the most promising technologies in the future of the automobile industry. However, failures or accidents caused by battery packs remain obstacles to the development of EVs and HEVs. The safety of battery packs, a core technique of EVs or HEVs, should be researched to ensure the overall safety of the vehicles and to accelerate the industrialized process of EVs and HEVs.Focused on the protection technique for battery packs, the main contributions of the dissertation are listed as follows:1. Development of test bench and simulation platform for battery performance testingTo measure the characteristics of battery packs and to provide data support for the subsequent researches, a battery performance test bench is developed. The test bench comprises a constant temperature-humidity test chamber, a battery performance tester and an in-parallel battery cell parameters monitoring system. The in-parallel battery cell parameters monitoring system consists of a main control module, several current measurement modules and several voltage/temperature measurement modules. The main control module is equipped with CAN buses and an U interface to receive, store and upload battery parameters which are tested or calculated by the current measurement modules and the voltage/temperature measurement modules. The voltage/temperature modules are used to measure battery cell voltage and temperature, as well as to calculate battery cell internal resistance. The current measurement modules are used to measure battery cell current, battery pack voltage and current.Based on the Matlab/Simscape platform, a battery cell model is developed. The module contains the open circuit voltage description element, the ohmic internal resistance description element, the polarization resistance and capacity description element. The battery cell model is validated with a constant current charge/discharge test and a pulse current discharge test. The simulations agree well with the experiments and the maximum difference is within 1%. Then, a battery module model and a battery pack model containing eight battery cells are developed and validated by tests. Results show that the developed battery cell model can be used to assemble different types of battery pack models.2. Influence of battery integration on battery pack performanceThe constant current discharge tests show that there are remarkable differences in the terminal voltage of the battery cells within a LiFePO4 battery module. And the input impedance of battery voltage monitoring system (IIoBVMS) also causes different current among each battery module.Taken the inter-cell connecting plates as resistors, an in-parallel connected battery module model is established based on the battery cell model. The effect of inter-cell connecting plate resistance (ICCPR) on the battery module performance is simulated. Results indicate that the ICCPR causes unevenly current flowing among the battery cells. The battery cell directly connected to the battery module posts (BMP) is the first one reaching its end-of-discharge (EOD) voltage. Also, it presents the lowest terminal voltage and state-of-charge (SOC) during the discharge process. The battery cell directly connected to the BMP goes into deep discharge state more easily. Therefore, it performs higher aging rate. The aging of the battery cell causes over-discharge of the adjacent battery cells. The reasonable ratio of the ICCPR to the battery ohmic internal resistance (OIR) is discussed for different average currents and different numbers of battery cells, to guarantee the maximum SOC evaluation error within a target value of 0.05.A "parallel-series" topology battery pack model containing the battery voltage monitoring system is proposed and validated. The effects of IIoBVMS on the battery pack performance are analyzed. Results indicate that the leakage current remains at the level of 10 μA even if the IIoBVMS reaches MQ level, which causes SOC inconsistency among battery modules. The maximum SOC deviation is near 0.03 for an HEV when the cycle number reaches 2000. A safety management strategy to correct the measured battery pack SOC is proposed. A method is presented to calculate the SOC correction coefficient according to the cycle number and the average charge or discharge current.3. Battery cell inconsistency on battery pack performance and battery pack capacity estimationThe current flowing though each battery cell in a battery module is different due to the resistance inconsistency. The charge or discharge rate of each battery cell is mildly different, which denigrates the capacity of the battery module. An in-parallel battery module model is developed for analyzing the effect of battery cell inconsistency on the performance of a battery module. Results indicate that the SOC of a battery module cannot characterize the SOC of ALL the internal battery cells. When the battery management system (BMS) controls the end-of-charge (EOC) time according to the SOC of a battery module, some internal battery cells are over-charged. The battery module EOC voltage and the battery pack capacity should be re-rated to guarantee the safety of ALL individual battery cells.It is proposed that the safety EOC voltage can be set according to the number of battery cells and the applied charge current of the battery module. For a battery module containing n battery cells, the safety EOC voltage lies in the condition when the battery module includes ONLY one normal battery cell and ALL the capacity of the other battery cells is 0.8n-1/n-1 times of the rated capacity.A new approach is proposed to calculate a battery pack capacity considering in-parallel battery cell safety. In the approach, the "normal battery module" capacity and charge voltage shift are evaluated by a mathematical transform method. The "normal battery module" SOC is estimated according to the relationship deduced from the standard charge voltage-to-SOC curve. Then, the battery pack capacity is calculated according to the SOC and capacity of the "normal battery module". Experimental results show that battery pack capacity estimation difference between the proposed method and the standard current integration method is within 0.35%.4. Battery voltage and available power prediction based on improved Dynamic Matrix Control algorithmsA new method that linearizes RC equivalent circuit model and predicts battery voltage and available power according to improved Dynamic Matrix Control (DMC) algorithm is proposed. The bench test results indicate that a first RC block equivalent circuit model could be used to describe the dynamic characteristics of a battery under testing condition. However, lacking of long time constant RC modules, there is a sample deviation in the open-circuit voltage identified and that measured. The long time constant RC modules play major roles in the steady state. As an n-RC blocks equivalent circuit model is highly nonlinear, a new method to predict the open-circuit voltage is presented, in which the time constants are determined based on the binary encoding method. Given the battery parameters identified and the end-of-discharge (EOD) voltage permitted, the maximum continuous discharge current in the future time (At seconds) can be calculated by the dichotomy method and the improved DMC algorithm. Then, the discharge power available at this time is determined.Testing results from an HEV equipped with 320 V/66 Ah battery pack and an EV equipped with 72 V/100 Ah battery pack show that the battery pack voltage predicted is in good agreement with that measured, the maximum difference is within 3.7% and 0.8%, respectively. Fixing the time constant to a numeric value, satisfactory results can still be achieved.5. On-board battery pack state of health estimationThe differential voltage analysis (DVA) method and the incremental capacity analysis (ICA) method can describe the battery aging state according to the feature point of the differential voltage (DV) curve directly. Four LiFePO4 batteries in different aging states are adopted to perform cycle charging processes. The cycle DV curves and the cycle incremental capacity (IC) curves are calculated by the proposed center least square method. The characteristic of the IC curves and DV curves of one battery under different cycles are analyzed. Then, a new method is proposed to estimate SOH based on the feature point or the transform coefficient of the DV curve. And the method is verified by other three battery cycle DV curves.The results show that the SOHs of the four battery cells can be evaluated with an error bound no more than 2% according to the capacity difference between two feature points, associated with the stage phenomenon in lithium intercalation process, in DV curves. Transforming the DV curve to be consistent with the initial DV curve provided by the manufactory, the battery SOH can be estimated according to the transform coefficient with the error bound is within 2.5%. This method can be applied to the battery cells which are unable to be charged to the second feature point. The battery pack SOH monitoring framework according to the feature point and the transform coefficient of the DV curve is presented.