Data-Oriented Parallel-in-Time Electromagnetic Transient Simulation of AC–DC Grid

An electromagnetic transient (EMT) simulation framework is developed by integrating the Parallel-in-Time algorithm with a data-oriented Entity–Component–System (ECS) architecture. The proposed method aims to enhance computational efficiency for EMT problems by jointly exploiting temporal concurrency and memory locality on multi-core CPU platforms. The ECS paradigm reorganizes component data into cache-friendly, structure-of-arrays layouts, reducing memory access latency and enabling efficient multithreaded execution. On top of this, the Parareal algorithm introduces time-parallelism by decomposing the simulation horizon into macro segments for concurrent fine-grid integration and coarse–fine correction iterations. A delay-consistent segmentation strategy is introduced to accommodate traveling-wave transmission-line effects without sacrificing parallel scalability or numerical accuracy. The framework is implemented in Rust based on the Bevy engine and evaluated using a modified IEEE 118-bus AC-DC network. Numerical validation confirms agreement with PSCAD/EMTDC® reference waveforms under diverse scenarios, including synchronous generator ramp-up, three-phase fault switching, line energization, and MMC DC-link dynamics. Performance benchmarks reveal that optimal integration parameter tuning can yield significant speedups: a maximum runtime reduction of 44.1% is achieved at δt= 5 μs with Δt= 80 μs, while a 38.5% gain is recorded at δt= 1 μs with Δt= 25 μs, where δtand Δtdenote the fine and coarse time-step in the parallel-in-time framework. These results highlight the practical advantages of combining time-parallel algorithms with data-oriented simulation structures for accelerating EMT simulations under realistic grid conditions.