Decoupled Space-Time Parallel Solver Integrating Parareal and MATE for Scalable Transient Stability Simulation

Read full text Summarise with VIRAT
Authors: Francis C Joseph, Gurunath Gurrala
Cited by: 0
References: 30

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper presents a decoupled space-time parallel solver that integrates the Parareal algorithm (parallelin-time) with the Multi-Area Thevenin Equivalent (MATE) method (parallel-in-space) for simulating the transient stability of large-scale power systems. A shared-memory implementation of MATE, exploiting both spatial and task-level parallelism, is incorporated into the Master-Worker (PM) and Distributed (PD) Parareal paradigms. The hybrid solver concurrently employs two high-performance computing (HPC) frameworks-OpenMP for MATE and message passing interface (MPI) for Parareal-to achieve scalability across both domains. Two implementation strategies are examined: homogeneous configurations with equal MATE partitions and heterogeneous configurations with unequal partitions. Simulation results on large systems with detailed generator and composite load models demonstrate that homogeneous scheduling complements Parareal, achieving nearly linear speedup while preserving MATE's expected performance. Consequently, the overall speedup of the hybrid solver approximates the product of the individual MATE and Parareal speedups. Heterogeneous scheduling offers performance benefits when uniform resource allocation is infeasible, allowing for flexible processor deployment across diverse computing environments.

Fields of Study & Subjects

Power System Optimization and Stability Parallel Computing and Optimization Techniques Real-time simulation and control systems

Publication details

Publication date
20 March 2026
Language
EN
Document type
Preprint