Parallel primal‐dual interior point method for the solution of dynamic optimal power flow. Issue 4 (30th December 2022)

• Record Type: Journal Article
• Authors: Sundermann, Rylee; Abhyankar, Shrirang G.; Zhang, Hong; Kimn, Jung‐Han; Hansen, Timothy M.
• Is Part Of: IET generation, transmission & distribution. Volume 17:Issue 4(2023)
• Page Start: 811
• Page End: 820
• Publication Date: 2022-12-30
• Subjects: power control -- power generation dispatch -- parallel programming -- power system analysis computing -- power system computation -- power system economics; Electric power production -- Periodicals; Electric power transmission -- Periodicals; Electric power distribution -- Periodicals; 621.3105
• DOI: 10.1049/gtd2.12708
• Languages: English
• ISSNs: 1751-8687

Abstract

Abstract: This work presents a novel solution for accelerating the dynamic optimal power flow using a distributed‐memory parallelization approach. Unlike other two‐stage relaxation‐based approaches (such as ADMM), the proposed approach constructs the entire dynamic optimal power flow problem in parallel and solves it using a parallel primal‐dual interior point method with an iterative Krylov subspace linear solver with a block‐Jacobi preconditioning scheme. The parallel primal‐dual interior point method has been implemented in the open‐source portable, extensible toolkit for scientific computation (PETSc) library. The formulation, implementation, and numerical results on multicore computers to demonstrate the performance of the proposed approach on medium‐ to large‐scale networks with varying time horizons are presented. The results show that a significant speedup is achieved by using a block‐Jacobi preconditioner with an iterative Krylov subspace method for solving the dynamic optimal power flow problems. Abstract : We developed a fully parallel primal‐dual interior point method (PDIPM) for solving the multiperiod dynamic optimal power flow (DOPF) problem. The method was used with a block‐Jacobi preconditioner to overcome the memory bottleneck on the large size DOPF problem. We released the parallel PDIPM in the open‐source library PETSc to benefit scientific community.