Computing Applications

National Lab Works to Cap, Reduce Carbon

Argonne National Laboratory will use supercomputers to help companies advance clean energy technologies with ultra-fidelity simulations and data analysis.

Credit: Shutterstock AI wind turbines above a manufacturing operation, illustration

The first U.S. exascale-era supercomputer at the Argonne National Laboratory—named Aurora—along with its petaflop-era cohorts Polaris and Theta, will participate in the pioneering U.S. Department of Energy (DoE) program High Performance Computing for Energy Innovation.

The four most recently announced funding winners in the program are the Ford Motor Company, Capstone Green Energy, Lakril Technologies Corporation, and Power Systems Manufacturing.

Supercomputer simulations by Argonne National Laboratory in collaboration with these industrial companies are intended to improve manufacturing efficiency and to explore new materials for clean energy applications, according to Michael Papka, director of the Department of Energy’s High Performance Computing for Energy Innovation program and the Argonne Leadership Computing Facility (ALCF), where Aurora is located. Each of the four companies will use up to $3 million in funding to explore the feasibility of proposed breakthrough ideas by using the exascale supercomputers to quickly evaluate their new concepts with ultra-fidelity simulations.

According to Papka, Argonne intentionally chose a diversity of applications to demonstrate how its supercomputer investments can help reduce carbon emissions while simultaneously improving new products.

Ford aims to explore how to improve the solid-electrolyte interface (SEI) film that forms on the anodes of electric vehicle (EV) batteries. Capstone Green Energy will use supercomputer simulations to help convert its micro-turbines to run on clean-burning hydrogen, rather than natural gas. Lakril Technologies will pursue new catalysts to enabling its transition from natural gas, propane, and landfill gas to sustainable biomass feedstocks. Power Systems Manufacturing (PSM) will simulate converted turbines to co-fire capabilities with renewable gases such as hydrogen, which could lead to a net reduction of carbon output for power-grid applications.

In 2022, Oak Ridge National Laboratory announced the world’s first exascale supercomputer, the Frontier 1.2 ExaFlops per second supercomputer, which as of November 2023 still leads the Top500 Supercomputer List of the world’s fastest supercomputers. However, Argonne National Laboratory anticipates its 2 ExaFlops per second Aurora (due for completion in 2024) will depose Frontier from the leadership spot in the ranking. The Aurora credits its almost-doubling of Frontier’s speed to its Hewlett Packard Enterprise (HPE) Cray EX architecture using Intel Exascale Compute Blades containing Xeon 2.4GHz CPUs (central processing units) and Intel Max Series Data-Center GPUs (graphic processing units). For reference, the Polaris supercomputer achieves 44-PetaFlops using AMD 2.8GHz Milan processors and Nvidia A100 GPUs. Theta achieves 11.7-PetaFlops using 1.3-GHz Intel Xeon Phi processors and Nvidia A100 GPUs.

Besides the exascale supercomputer hardware, Papka emphasizes the “deep supercomputer expertise” of ALCF researchers also will be allocated to each of the four projects, to maximize their potential successes in the DoE/industry cooperative effort. The aim of the collaboration is to provide the exascale high-performance computing (HPC) resources needed by private industry to clean up energy generation and usage in the U.S.

The human expert assigned to lead the lab’s cooperative efforts with Ford Motor Company is DoE Office of Science user facility member Subramanian Sankaranarayanan, a group leader at Argonne’s Center for Nanoscale Materials specializing in improving battery manufacturing efficiency.

According to Sankaranarayanan, in collaboration with Ford researchers, he will use Argonne supercomputers to model, in nanoscale detail never before possible, the formation of the solid-electrolyte interphase (SEI) in a lithium ion battery. The SEI is a solid film that forms on the initial charging of a liquid-electrolyte lithium-ion battery. Sankaranarayanan says the crucial mechanical structure of the SEI’s initial formation determines the longevity and stability of long-cycling lithium batteries for electric vehicles (EVs). To get the mechanical structure just right, industry currently spends as much as a week or more to form a stable, long-lasting SEI.

Sankaranarayanan and Ford engineers hope to perform ultra-detailed modeling of the SEI formation process to increase their understanding of why it takes so long to form a viable SEI. The function and formation of an excellent SEI is currently the subject of debate among scientists, but with the lightning speed of exascale computer simulations, Sankaranarayanan hopes to put to rest the open theoretical questions and to devise a quicker method for forming excellent SEIs, reducing the cost of lithium-ion batteries and increasing the acceptance of EVs, which will reduce carbon emissions.

Capstone Green Energy’s C200 microturbine system for power and heat generation is already a green technology, but the company plans to work with Argonne research scientist Debolina Dasgupta to redesign, optimize, and refit the C200 to be even greener by running completely on hydrogen fuel. Capstone’s need for exascale supercomputers comes in part from the computational fluid dynamics simulations required to assess and compare designs to ultimately produce a configuration that can optimize the engine’s design for pure hydrogen combustion and fuel injection. This new modeling effort will help Capstone make the switch to pure hydrogen fuel, leading to reduced nitrogen oxide emissions.

“These new HPC resources will be used for running multiple low- and high-fidelity computational fluid dynamics simulations for the hydrogen fuel injector to determine an optimal design,” said Dasgupta. “Hydrogen behaves differently compared to natural gas, the current fuel used in combined heat and power systems. The fuel and air mixing determines the combustion and emissions behavior of such systems. An optimal injector design should be capable of maintaining a stable flame for the operability map of the engine with low nitrogen oxide emissions. This will be ensured through efficient mixing and burning of the fuel/air mixture. Thus, both stability and reduced emissions are our goals to be achieved through a new fuel injector design and its integration with the internal combustion engine.”

Working with Chicago-based Lakril Technologies, Argonne computational scientist John Low is engaged in removing excess carbon from paint, coatings, adhesives, and super-absorbent polymers. Using Argonne’s HPC resources, Low and Lakril engineers aim to improve the performance of the catalyst that enables the transition to sustainable biomass feedstocks.

Argonne’s supercomputers will allow “ab initio” (from first principles) simulations, rather than the approximate simulations performed today.  We “will perform ab initio Metadynamics (AIM) at reaction conditions for the five catalyst steps involved,” said Low. “Metadynamics enhances the sampling of many portions of phase space and works well in combination with ab initio methods to study processes in which changes in electronic structure play a dominant role.”

Power Systems Manufacturing provides advanced technology that enables gas turbines to co-fire with renewable gases such as hydrogen, leading to a net reduction of carbon output in power grid applications. Flashback, however, is a common problem in hydrogen-fueled combustion systems (in which the flame passes backwards from the combustion chamber toward the raw gas holding tanks). PSM, working with Argonne computational scientist Shashi Aithal, will use the lab’s HPC resources to run ultra-detailed simulations of engine redesigns that reduce the risk of flashbacks.

“Co-firing is a term used to describe incorporating a secondary fuel with a primary fuel utilizing the same combustion equipment. Our ultimate goal is understanding how to use 100% hydrogen in gas turbines without causing flashback,” said Aithal. “But the intermediate goal is to understand how to eliminate flashback with higher and higher hydrogen percentages, thereby gradually converting our carbon-emitting co-fire equipment into greener, hydrogen-burning units, thereby systematically lowering carbon emissions.”

The High Performance Computing for Energy Innovation program is sponsored by DoE’s offices of Advanced Materials and Manufacturing Technologies and Industrial Efficiency and Decarbonization within the Energy Efficiency and Renewable Energy Office and the Office of Fossil Energy and Carbon Management.

R. Colin Johnson is a Kyoto Prize Fellow who ​​has worked as a technology journalist ​for two decades.

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More