Multicore methods to accelerate ship power system simulations

Abstract

Two methods to partition and parallelize the simulation of large-scale shipboard power systems on multicore computes are demonstrated. The first method is node tearing, which is used for offline simulation. The second is the state-space nodal method, which is used for real-time simulation. Both methods are benchmarked against MATLAB/Simulink 2012b for speed and accuracy The simulation model is a notional shipboard power system having characteristics of AC-radial, 450 V, 60 Hz, three-phase, delta-ungrounded power system. The parallel simulation results show speedups in excess of one order of magnitude and general agreement in accuracy.