ERROR ESTIMATORS FOR ADAPTIVE SCHEDULING ALGORITHM FOR SERIAL CO-SIMULATION

In serial co-simulation, the subsystems are solved and exchange their variables subsequently. For this reason, the solution sequence of the subsystems plays an important role. Prior work has demonstrated the effect of solution sequence on the accuracy of serial co-simulation and proposed an adaptive scheduling algorithm that predicts when to change the solution sequence using global input estimation error. This paper introduces another mechanism that considers physical causality to determine the computational causality in the co-simulation. We benchmark this method against input estimation error and static solution sequences in adaptive scheduling algorithm in serial co-simulations of a linear problem. The resulting method generally avoids the worst solution sequence and mitigates the risk of co-simulation being unstable if one of the sequences produces a greater error.

Automated summary

In serial co-simulation, the sequence of subsystem solution plays a crucial role. Previous research has shown the impact of solution sequence and proposed an adaptive scheduling algorithm based on global input estimation error. This paper introduces a mechanism considering physical causality to determine computational causality in the co-simulation. Benchmark tests against input estimation error and static solution sequences demonstrate that the resulting method generally avoids the worst solution sequence and mitigates the risk of co-simulation instability.