The battery system plays an important role in renewable energy systems and electric vehicles. Since the battery system consists of multiple series-connected cells, over-charging or over-discharging can occur due to the heterogeneous characteristics of the cells. Thus, an effective equalizer is essential to maintain the cells' homogeneous performance, and thus extend their lifespan. Energy regenerative equalization systems provide different advantages in terms of capacity, speed, and efficiency. However, their performance depends on the initial energy distribution of the cells. On the other hand, there is no unified assessment method to compare the performances of equalizers. This thesis deals with the performance, effectiveness, and flexibility of the modular battery system. The main purpose of the research in this thesis is to develop a modular structure of the equalizer which can mitigate the inconsistency between cells and modules.
Firstly, a switch-matrix capacitor equalizer (SMC-E) is proposed for transferring energy between two cells directly. With the switch-matrix structure, a single capacitor is needed, thereby reducing the size of the equalizer. In order to achieve the shortest equalization time, an optimal pairing algorithm is introduced. The equalization process is divided into the scanning step and the pattern-holding step. Because the voltage deviation between two cells in the different pairs of cells is heterogeneous, their balancing currents are uneven. By scanning all possible balancing currents from all cell pairs, the optimal equalization pattern can be obtained, which has the highest balancing current. It is observed that only the pair of the highest voltage cell and the lowest voltage cell can have the highest compensating current. Thus, the optimal equalization pattern is the shortest way of transferring energy between the cells. The optimal equalization pattern is maintained for a period, Th, before another current scanning step is activated. The optimal equalization pattern changes dynamically during the equalization process as the cell voltages change. By repeating the equalization cycle, the energy is transferred between the cells as fast as possible. In addition, design instructions for operational and circuit parameters are provided. The hardware experimental results of the equalization verified the performance of the SMC-E in terms of equalization capability, speed, and stability. The cell-inconsistency is mitigated in all test scenarios, even if the energy distribution of the cells is different.
For a fair performance comparison for different equalizers in a short period of time, a simulation method based on a unified average model (UA-model) is proposed. By counting the amount of charge that flows out and into the cells, the equalization process of two-cells can be emulated by the UA-model. Since switching elements are suppressed in circuits, simulation based on the UA-model can be performed with a large sampling time, thus reducing the overall simulation time. In comparison with the RTSS simulation, the operational profiles of the UA and RTSS models have a homogeneous characteristics. The simulation method based on the UA-model has advantages as a platform of evaluation to fairly compare the performances of equalizers. The different equalizer topologies and designs can be tested under different initial test scenarios within a short period. Therefore, the design of the SMC-E may be evaluated and optimized.
The SMC-E can overcome the impact of initial energy distribution dependency, but the equalization time becomes longer as the number of cells increases. Thus, the fundamental analysis of equalizer performance at cell count is provided. To reduce equalization time, the modular structure of the SMC-E is available for both serial and parallel connected modules. In series-connected modules, the hybrid strategy is proposed to inherit the benefits of SMC-E, where several SMC-E units are activated at the same time. Once the cell voltages inside each module are equalized, the module equalization function is triggered to balance the module voltages. After two separate processes, the voltages of all cells are equalized in a small gap. The RTSS test results show a 13.5% shorter equalization time than the conventional method, where a single SMC-E is implemented for the whole cell string.
In addition, the same modular SMC-E design is applied to parallel-connected battery modules. In the parallel connection, the uneven distribution of current between the branches and the inrush current during the hot swap process are serious problems. Thus, two equalization strategies are proposed for SMC-E to address them in IDLE and non-IDLE modes. In IDLE mode, the modules are balanced by the same energy exchange scheme as the modules connected in series. On the other hand, the energy levels of the modules are balanced during the charging or discharging processes by the current distribution scheme. Based on the test results on RTSS, the modules are equalized in the 4.82% SOC difference in the IDLE mode without the inrush current of the self-balancing effect. In addition, the modules can be equalized within 1% SOC difference in the charging and the discharging processes by the current distribution scheme. Thus, all modules are fully charged or fully discharged at the same time. The inrush current during the hot swap process is also mitigated since the branch currents can be adjusted by the SMC-E and the current distribution scheme.