Multilevel Converters for Battery Energy Storage: How Many Levels and Why?

This work explores the potential benefits of cascaded H-bridge multilevel converters in low-voltage applications, particularly grid-attached battery energy storage systems (BESS). While some benefits of these are discussed in literature, this work seeks to create practical, quantitative models for system performance in terms of a number of key performance parameters. These models are then used to find the trends in these performance parameters with an increasingly high order converter, starting to answer the question of how many levels are best. The system performance parameters modelled are power loss, thermal performance and reliability. Wherever practical models and assumptions are validated, be that experimentally or through comparison with existing methods – this work includes a number of experimental series. The resulting trends explored highlight a number of interesting trends, principally: total power loss can be much lower, particularly at high switching frequencies; system thermal performance can be much improved owing to more efficient heatsink utilisation; and due to these thermal benefits, the system reliability based on switching device failure does not suffer as one might expect, and can in fact be higher under some conditions. The investigation also considers the use of cutting-edge switching device technology, such as gallium nitride power transistors, which a multilevel converter enables the use of, and in turn can significantly reduce power dissipation and increase switching frequency. Overall, the work adds new arguments in favour of multilevel converters in such applications and lowers the barrier to practical implementation by answering a number of questions a designer would likely ask. The key novel contributions of this work are the results of the trends that were found in terms of converter power loss, system thermal performance and switching device reliability with respect to multilevel converter order – with the methodologies created for these being somewhat novel in their own right. Along the way, however, other novel work was conducted including: an experimental investigation in to the accuracy of voltage-capacitance curves provided by manufacturers; experimental derivation of relationships for predicting MOSFET body diode performance from readily available device parameters; analysis showing the potential impact of GaN devices on converter efficiency; an experimental validation of GaN device gate turn-on energy; creation and validation of empirical relationship for predicting how heatsink performance varies with more devices of a smaller size; as well as an exploration of whether the extreme small size of some modern power transistors could lead to unexpected thermal cycling issues.