Accelerating Switching Model-Based Simulation Through Parallel Computing

Power electronics converters become an enabler for future power and energy system. High-fidelity power electronics simulation can be compute-intensive and time-consuming because of the higher switching frequency (i.e., switching actions per second) of converter using advanced power semiconductors and an increased number of converters applied in the system, such as renewable energy with energy storage and electrified transportation. Unlike the switching model, average modeling can efficiently alleviate the computational burden, and adequately represent converters’ behavior for the controller design, but sacrifice the resolution for switching ripples/harmonics, which are also crucial for power electronics design, operation, and reliability studies. This digest proposes to accelerate the switching model-based simulation by leveraging a fast but low-fidelity average model plus parallel computing technique, entitled the average-to-switching (A2S) method. First, in comparison with today’s sequentially-computing approach, the basic concept of the A2S method by introducing parallel computing is presented. Then, the detailed methodology of the proposed A2S approach is described with the derivation of the algorithm suitable for paralleling computing. Finally, a case study with a widely applied two-level voltage source converter is performed and a comparison among these methods is summarized. It is observed that this proposed method performed >5 times speedup than the benchmark while maintaining an r-square value of the results > 0.99.