The National Renewable Energy Laboratory announced that in 2024, NREL’s Kestrel high performance computing system powered more than 425 energy research projects across 13 funding areas.
Kestrel is NREL’s newest HPC system, delivering roughly 56 petaflops of computing power to accelerate energy research, including through the use of artificial intelligence and machine learning to open new avenues in energy research, materials science, forecasting, and other areas.
Kestrel is an HPE system located at NREL’s Energy Systems Integration Facility HPC data center. Last summer, NREL and HPE completed installation of an additional 132 GPU nodes — each hosting four NVIDIA H100 GPU processors — to the 2,314 existing central processing unit (CPU) nodes with two Intel Sapphire Rapids processors.
NREL’s Advanced Computing Annual Report for FY 2024 examines the contributions of advanced computing. A highlight describes how Questaal — a suite of electronic structure software designed to answer basic questions about chemical and solid-state systems at the atomic level — solves quantum physics equations and simplifies computationally intensive processes while retaining high fidelity. Questaal’s high fidelity enables it to resolve many properties where prior theories have been inadequate, showing the ability to answer key science questions in a wide range of studies of chemical and materials systems. More in the highlight: Questaal Software Resolves Complex Physics, Helping Researchers Solve Complex Problems.
Another key project discussed in the annual report focuses on developing cost-effective methods for breaking down plant cell walls to gain a deeper understanding of biomass at a molecular level.
Funded by the Bioenergy Technologies Office, researchers used a multidisciplinary approach to study how polymers are structured in Populus wood. They then used Kestrel to create molecular models of the lignocellulosic biopolymer assemblies. The models helped researchers identify which biopolymer interactions are responsible for the chemo-mechanical resilience of biomass. Read more in the report highlight: First Macromolecular Model of Woody Plant Cell Walls Will Improve Biomass Conversion.
