TY - GEN
T1 - Towards high performance discrete-event simulations of smart electric grids
AU - Perumalla, Kalyan S.
AU - Nutaro, James J.
AU - Yoginath, Srikanth B.
PY - 2011
Y1 - 2011
N2 - Future electric grid technology is envisioned on the notion of a smart grid in which responsive end-user devices play an integral part of the transmission and distribution control systems. Detailed simulation is often the primary choice in analyzing small network designs, and the only choice in analyzing large-scale electric network designs. Here, we identify and articulate the high-performance computing needs underlying high-resolution discrete event simulation of smart electric grid operation large network scenarios such as the entire Eastern Interconnect. We focus on the simulator's most computationally intensive operation, namely, the dynamic numerical solution for the electric grid state, for both time-integration as well as event-detection. We explore solution approaches using general-purpose dense and sparse solvers, and propose a scalable solver specialized for the sparse structures of actual electric networks. Based on experiments with an implementation in the THYME simulator, we identify performance issues and possible solution approaches for smart grid experimentation in the large.
AB - Future electric grid technology is envisioned on the notion of a smart grid in which responsive end-user devices play an integral part of the transmission and distribution control systems. Detailed simulation is often the primary choice in analyzing small network designs, and the only choice in analyzing large-scale electric network designs. Here, we identify and articulate the high-performance computing needs underlying high-resolution discrete event simulation of smart electric grid operation large network scenarios such as the entire Eastern Interconnect. We focus on the simulator's most computationally intensive operation, namely, the dynamic numerical solution for the electric grid state, for both time-integration as well as event-detection. We explore solution approaches using general-purpose dense and sparse solvers, and propose a scalable solver specialized for the sparse structures of actual electric networks. Based on experiments with an implementation in the THYME simulator, we identify performance issues and possible solution approaches for smart grid experimentation in the large.
KW - discrete event
KW - parallel computing
KW - smart grid
KW - sparse solver
UR - http://www.scopus.com/inward/record.url?scp=84857288729&partnerID=8YFLogxK
U2 - 10.1145/2096123.2096135
DO - 10.1145/2096123.2096135
M3 - Conference contribution
AN - SCOPUS:84857288729
SN - 9781450310611
T3 - HiPCNA-PG'11 - Proceedings of the 1st International Workshop on High Performance Computing, Networking and Analytics for the Power Grid, Co-located with SC'11
SP - 51
EP - 58
BT - HiPCNA-PG'11 - Proceedings of the 1st International Workshop on High Performance Computing, Networking and Analytics for the Power Grid, Co-located with SC'11
T2 - 1st International Workshop on High Performance Computing, Networking and Analytics for the Power Grid, HiPCNA-PG'11, Co-located with SC'11
Y2 - 13 November 2011 through 13 November 2011
ER -