Detailed Equivalent Model for Parallel CPU-GPU EMT-type Simulation of Solid-State Transformer

Solid-state transformer (SST) enables the efficient interface between AC grids and the DC distribution networks. Detailed modeling of numerous components in the power converter system requires massive computational resources. This paper proposes a numerically efficient decoupled detailed equivalent model (DEM) of the SST for efficient electromagnetic transient (EMT) simulation. The proposed model can accurately represent the fast dynamics of the converter system under normal, transient, and IGBT-blocked states. In addition, a switching interpolation technique is employed to ensure compensation of intra-time-step switching events and convergence of the simulation results with relatively large time step-sizes. To further accelerate the EMT simulation, a parallel simulation algorithm is developed to utilize the computational resources of both central and graphics processing units (CPU-GPU). The accuracy of the proposed DEM is validated against the detailed model (DM) of SST system, while the efficiency of the parallel algorithm is assessed by comparison with a CPU-only implementation.