Global electricity usage has been increasing exponentially over the last 40 years and with the current pace of population and economic growth, the same trend will continue in the coming years. However, the current electricity generation is predominantly fossil fuels based, which makes the overall process unsustainable and polluting, and is contributing directly to the menace of global warming. The need of the hour is to transition to renewable and sustainable sources of electricity generation such as solar, wind etc.

However, the major bottleneck in the large-scale implementation of RESs is the intermittent availability and their integration into existing AC grids as the RESs are predominantly DC sources. In order to tackle the issue of intermittency various energy storage solutions are being developed. Further, as a solution to the issue of integrating RESs into the current electricity distribution system, DC distribution grids are being developed.

The DC distribution architecture is of two types, Unipolar and Bipolar distribution systems characterized by the number of wires used for power transmission; 2 wires for unipolar and 3 wires for Bipolar. Out of the two, Bipolar systems have inherent advantages of flexibility, stability and efficiency over unipolar systems.

Current thesis focuses on series connected switch configuration of voltage balancing converter for bipolar DC power distribution systems.

For this thesis, various topologies of voltage balancers for bipolar dc distribution systems were studied and a buck boost-based series switch-connected voltage balancer topology was chosen for the final design. For this topology, using MATLAB the parameterization and optimization of magnetic components of the converter were performed. The optimized configuration was then modeled and simulated using LTSpice.

Post this, a comparison between different methods adopted to account for the unbalanced voltage sharing across the series-connected switch configuration of the balancing converter owing to the non-linearities present in the circuit was done. Based on this study, the most prominent of the methods is then integrated to the LTSpice model.