Energy-Efficient Advanced Computing System Operations

For energy research, NREL leads the advancement of high-performance computing (HPC), cloud computing, data storage, and energy-efficient system operations.

NREL's award-winning, most energy-efficient data center in the world is a living laboratory where researchers advance energy HPC best practices for energy research. Research done in concert with operating the HPC capability enables a reduction in energy use within NREL’s HPC Data Center and across the industry. Examples include enabling parameter estimation for complex systems, physics-informed machine learning, and understanding how to use multiple model fidelities together to enable optimization under uncertainty, visualization of complex data, co-simulation, and distributed control.

High-Performance Computing Systems

We operate systems with thousands to hundreds of thousands of processing cores, very high-performance networks, and thousands of petabytes of data storage space for hosting the models and simulations of many researchers and engineers. Customized systems—tailored to differentiated needs of NREL, the U.S. Department of Energy (DOE), and partners—include hardware-in-the-loop, machine learning, and high-throughput materials discovery and maximize benefit of hardware and software resources necessary to advance energy technologies using:

• Applications – Implementations and strategies for using large, highly parallel computer resources—from interacting with users on system use to improving the mechanisms for using HPC systems
  
  
• Parallel performance – Improvements to computational science and engineering research by using hundreds to tens of thousands of processing cores all applied to solving a single problem
  
  
• Software architecture – Design and optimization of software to take advantage of large hardware resources to address research and analysis goals.

NREL's modeling, simulation, and hardware-in-the-loop testing build on experimentation to expand understanding of system interactions, accelerating bench-scale discoveries and process innovations to market-relevant, pilot-scale energy integration and production solutions.

Cloud Computing Services

We provide cloud computing services as an enterprise solution for enabling scalability and reliability of data science, computing, application development, data management and sharing, and analysis innovation. On-premises/cloud hybrid configurations embrace virtualized and containerized computing, support varied and data-centric workloads, and enable tight integration between experiment and computing. Self-service cloud services enable researcher access, data sharing, and experimentation at multiple scales and locations.

Big Data and Data Storage

As bodies of data grow, traditional storage and processing options quickly become obsolete. Big data needs optimal infrastructure. At NREL, reliable access to robust data sources is a foundational capability that supports labwide research and attracts critical public- and private-sector partnerships. We manage data storage systems to enable the integration of complex and large-volume experiments with HPC resources.

NREL leverages in-house expertise in this field to enable parameter estimation for complex systems, physics-informed machine learning, multiple model fidelity optimization, optimization under uncertainty, visualization of complex energy data, co-simulation, distributed control, and inverse design.

Computing Operations and Facilities

We manage the process of allocating portions of the DOE Office of Energy Efficiency and Renewable Energy’s computational capabilities to projects, create and maintain user accounts, and provide assistance to help users utilize the computational capabilities. Learn more about accessing NREL's HPC systems as a user.

NREL HPC experts manage systems and operations in the HPC Data Center and the HPC User Facility.

Featured Projects

Simulating High Electrification Futures for the U.S. Grid

NREL leads the collaborative Electrification Futures Study to examine demand-side and supply-side electrification scenarios in the U.S. energy system. The most recent grid simulations are for the year 2050. NREL's high-performance computing system generated hourly unit commitment and economic dispatch models to examine the operability of systems serving the electrified loads.

High-Performance Computing-Powered Experimentation

NREL's Energy Systems Integration Facility is part of the on-campus network that provides 20 MW of experimentation space, which leverages the HPC systems. Advanced Research on Integrated Energy Systems is a virtual research platform that provides unprecedented scalability (regional and metropolitan) to model dynamics of hundreds to millions of devices within complex energy system.

Publications

Advanced Computing, Data Science, and Artificial Intelligence Research Opportunities for Energy-Focused Transportation Science, NREL Technical Report (2021)

Using Thermosyphon Hybrid Cooling System To Optimize Data Center Water Efficiency, DOE Fact Sheet (2019)

Contact