National Laboratories Study Finds Exascale Supercomputer Capable of Predicting Power Grid Failures

Researchers from four national laboratories have explored the potential of an exascale supercomputer to help grid operators predict and prevent power disruption events caused by different weather events.

The study used Oak Ridge National Laboratory’s Frontier supercomputer to run the HiOp high-performance computing software from Lawrence Livermore National Laboratory, enabling simulations that helped identify over 100,000 possible grid failures or contingencies and weather scenarios within 20 minutes.

With commodity computing hardware, system operators only consider up to 100 contingencies and at least five weather scenarios, limiting the capability of grid operators to address potential failures, LLNL said.

Cosmin Petra, computational mathematician and principal investigator for LLNL, said the project aimed at demonstrating the capability of exascale computers to solve grid-control issues in a manner that aligns with the existing practices of power grid operators. Currently, the decision-making process for handling emergency grid situations involves humans who may not identify optimal ways to keep the grid operational under different renewable energy forecasts, he added.

National Renewable Energy Laboratory and the Pacific Northwest National Laboratory supported the study, which was conducted as part of the Exascale Computing Project’s ExaSGD project.