As part of the global effort to combat climate change, electric vehicles (EVs) are gaining popularity, even for long-haul commercial transportation. A battery pack is a critical component of an EV, and it contains several modules with many series- and parallel-connected electrochemical cells. Strict safety and operational limits are enforced on the cell-level to ensure safe operation of the battery pack. However, variations in the electrochemical properties among the cells in the pack causes some cells to reach the safety and operational limits faster than others. This limits the total power, and over time, the energy delivered and the lifetime of the battery pack. Maximizing the energy delivered by the battery pack (potentially also improving the battery pack’s lifetime) can be achieved by increasing cell-level control, and battery-integrated modular multilevel converters (BI-MMC) is presented as a solution. A BI-MMC has several series-connected DC-to-AC converters, commonly called submodules (SM), and every SM has a small battery pack with a few series-connected cells, thus allowing for increased cell-level control. The focus of the thesis is the evaluation of different control, design, and modulation techniques of BI-MMCs and compare the efficiency against a state-of-the-art SiC MOSFET-based 2-level inverter for a 40-ton long-haul commercial vehicle.
The first contribution is an efficiency evaluation of BI-MMC circuit topologies against the SiC two-level inverter. BI-MMCs with full-bridge SMs have higher efficiency than with half-bridge SMs. Compared to the 2-level inverter, full-bridge SM based BI-MMCs have similar efficiencies. 6-phase BI-MMCs have higher efficiency than both 3-phase BI-MMCs and the 2-level inverter, but with higher cost due to the increased number of SMs. Full-bridge SM based BI-MMCs with 5 to 6-series connected cells per SM have the highest efficiency.
The second contribution is a design optimization of the SM DC-link capacitors and the MOSFET switching frequency to maximize the total efficiency. The optimization revealed that selecting the MOSFET switching frequency close to the resonant frequency of the SM DC-link capacitors and the SM-battery decreases the total efficiency.
The optimization and evaluation considered a phase-shifted carrier-based modulation technique, but the total system efficiency can be further increased by using low-switching frequency modulation techniques. The third contribution is the impact of the nearest level modulation (NLM) on the efficiency of the BI-MMCs. NLM results in lower semiconductor losses than phase-shifted carrier-based modulation, but with higher battery losses. Direct implementations of NLM result in unequal utilization of the SM-batteries and is mitigated by cycling the SM duty cycles at the fundamental frequency, which results in lower total losses than cycling the SM duty cycles at lower frequencies.
Distributed control is preferred for MMCs with numerous SMs, where a central control unit performs the converter control, and the switching and modulation are carried out by the SM control units. The fourth contribution is the concept of reconstructing the SM reference signals by increasing the sample frequency of the SM control units relative to the central control unit, which improves the controllability of the SMs.
Conventional EV battery packs are directly connected to the fast charger, but in a BI-MMC, the battery and the inverter are integrated, potentially increasing the DC charging capabilities. The fifth contribution is the derivation and investigation of the maximum DC charging power of the BI-MMCs with the same SM semiconductor losses during traction. The investigation shows that most BI-MMCs have a maximum DC charging power of about 1 MW.