Gauss-Seidel Accelerated: Implementing Flow Solvers on Field Programmable Gate Arrays
Non-linear steady-state power flow solvers have typically relied on the Newton-Raphson method to efficiently compute solutions on today's computer systems. Field Programmable Gate Array (FPGA) devices, which have recently been integrated into high-performance computers by major computer system vendors, offer an opportunity to significantly increase the performance of power flow solvers. However, only some algorithms are suitable for an FPGA implementation. The Gauss-Seidel method of solving the AC power flow problem is an excellent example of such an opportunity. In this paper we discuss algorithmic design considerations, optimization, implementation, and performance results of the implementation of the Gauss-Seidel method running on a Silicon Graphics Inc. Altix-350 computer equipped with a Xilinx Virtex II 6000 FPGA.
- Research Organization:
- Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC05-76RL01830
- OSTI ID:
- 893666
- Report Number(s):
- PNNL-SA-47743; TRN: US200625%%371
- Resource Relation:
- Conference: 2006 IEEE Power Engineering Society General Meeting, 18-22 June 2006, 5 pages
- Country of Publication:
- United States
- Language:
- English
Similar Records
A parallel Gauss-Seidel algorithm for sparse power system matrices
Two-Stage Gauss-Seidel Preconditioners and Smoothers for Krylov Solvers on a GPU Cluster: Preprint