RELIABILITY ORIENTED POWER ELECTRONICS DESIGN FOR DC MICROGRIDS

Date

2025-8-15

Author

Nyamhanza, Wayne

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

237
views

0
downloads

Cite This

In modern power electric systems, power electronic converters constitute the backbone of DC microgrids, serving as interfaces between photovoltaic panels, wind turbines, battery storage systems, and grid-connected systems. The vulnerability of power converters stems from components in the power electronic converters. Power modules experience continuous thermal and mechanical stress during the time of operation, which manifests as repetitive junction temperature fluctuations, leading to progressive degradation of bond wires, solder joints. The literature shows that, reliability assessment and thermal management can extend converter life once the dominant failure modes are understood. However, such reliability modelling approaches for each converter type are often done in isolation, without an integrated system perspective. This proposed study made use of an electrothermal model for power loss, thermal prediction, a PINN thermal model surrogate, and lifetime modelling for a multi-source DC Microgrid. MATLAB Simulink and Python featuring PyTorch and DeepXde environments are used for the electrical and thermal simulations featuring a set of mission profiles from ambient, solar irradiance, wind speed, and SOC. Validated against Foster/Cauer models and experimental data with an RMSE of 1.2873 - 0.0581 °C, the PINN framework achieves >2°C accuracy improvements in junction temperature (Tj) prediction across four critical converters. Mission-profile-driven analysis reveals distinct thermal stress patterns. Wind converters show severe thermal cycling, and ESS endures stress from daily charge/discharge cycles. Lifetime consumption analysis via Miner’s rule and Bayerer/Coffin-Manson-Arrhenius models identifies bond wire fatigue (60–75% dominance) as the critical failure mode, with B10 lifetimes spanning 9.97 years (wind) to 19 years (grid). With PINN, damage drivers show high-frequency thermal cycles (>50°C) contribute >60% annual damage via Miner’s rule, while Monte Carlo simulations (5,000 iterations) confirm probabilistic lifetimes (5th–95th percentile: 7.85–37.7 years) and design sensitivity review ±18% lifetime variance from Tj uncertainty. A mitigation strategy leveraging advanced thermal management to address reliability, particularly for the PV converter, demonstrated an improvement in Tj management.

Subject Keywords

Damage Accumulation, DC Microgrid, PINN, Reliability, Junction Temperature.

URI

https://hdl.handle.net/11511/116198

Collections

Northern Cyprus Campus, Thesis