Supercomputer code can help capture carbon and reduce global warming

July 19, 2022 — More than a third of carbon dioxide (CO2) the emissions that contribute to global warming in the United States come from power plants. What if scientists could capture CO2 and store it where it would not contribute to climate change? Several technologies aim to do just that, and exascale computing can help.

National Energy Technology Laboratory scientist Jordan Musser leads ECP’s MFIX-Exa subproject.

Carbon dioxide is a greenhouse gas produced by burning fossil fuels that contributes to global warming. Like many multiphase flow devices, one of the biggest challenges in deploying carbon capture technologies is scaling up designs from the lab to industrial scales. The U.S. Department of Energy’s (DOE) Exascale Computing Project (ECP) software subproject MFIX-Exa helps achieve the necessary scaling.

“We are developing the tools to allow scientists and engineers to impact large-scale chemical reactors for a variety of industries,” said Jordan Musser, MFIX-Exa’s principal investigator. “It’s a set of tools that allows computer modeling to determine what’s going on inside gas-solid flow reactors. But we need this next-generation computing capacity” because of the complexity of the calculations required.

Track billions of particles

This type of modeling simultaneously tracks billions of individual particles to simulate gas-solid fluxes typical of a power plant. Standard computing usually can’t do this, Musser said. With MFIX-Exa, scientists can provide information to engineers when, for example, a chemical loop reactor is not performing well. And, he added, by using exascale computing, they can recognize when things are going wrong much faster than in a lab or experimental setting.

Musser, a scientist at DOE’s National Energy Technology Laboratory (NETL), is a mechanical engineer and applied mathematician who has always had an interest in modeling physical processes. He began his career at NETL as a Fellow of the Oak Ridge Institute of Science and Education in 2009. Recipient of the Presidential Early Career Award for Scientists and Engineers, the highest honor given by the U.S. government to early-career researchers , Musser combines his engineering know-how with computational mathematics to solve real-world problems. Weiqun Zhang and Andrew Myers, scientists at the DOE’s Lawrence Berkeley National Laboratory in California, and William Fullmer, a NETL research engineer, are also crucial in developing this code.

Unequal distribution challenge met

The MFIX-Exa team had to overcome several challenges to optimize the code for exascale, but one of the biggest is that local particle concentrations can vary dramatically in space and time, making it difficult efficient use of HPC resources. The team tackled this challenge using a dual-grid approach that separates the fluid and particle calculations, allowing the work of the particles to be periodically rebalanced in the system. The team rewrote the physical models from the older MFIX code and ported it to faster modern GPUs, successfully testing the code on supercomputers at DOE’s Oak Ridge National Laboratory (ORNL).

The code aims to simulate commercial chemical loop reactors, using technology to reduce CO2 emissions through capture and storage. The MFIX-Exa code calculates the fluid dynamics in these types of reactors and can simulate large-scale commercial reactors during their design phase. This allows engineers to prototype reactors with a high-fidelity model and diagnose problems before a plant is built.

The code “allows us to look inside the corrosive environment of these reactors and see how the process behaves,” Musser said. An extension of the older MFIX primarily used for lab-scale devices, MFIX-Exa will allow for an increase in the size, speed, and accuracy of problems on exascale computers, like Frontier, over the course of the next decade, Musser said.

Exascale computing can help reduce risk

At ORNL, Frontier recently became the first supercomputer to achieve exascale, with 1.1 exaflops of performance and a threshold of one quintillion calculations per second. The system will allow researchers to develop technologies critical to the country’s energy, economic and national security missions, helping solve problems of critical importance to the nation that lacked realistic solutions just five years ago.

Carbon capture and storage technologies such as chemical loop reactors require high performance computing to provide designers with data to make informed decisions before building and testing such a system.

“Exascale computing can model these technologies to optimize these systems, which will reduce the risk of failure,” Musser said.

MFIX-Exa is not isolated from chemical loop reactors and carbon capture technologies. The code can also help analyze a wide variety of engineering devices that work with a mixture of gases and solids, including devices found in the pharmaceutical, steel and cement industries, for example, Musser said. As exascale computing expands, these industries could benefit from code originally intended for carbon capture and storage.

This research is part of the DOE-led Exascale Computing Initiative (ECI), a partnership between the DOE Office of Science and the National Nuclear Security Administration. The Exascale Computing Project (ECP), launched in 2016, brings together research, development, and deployment activities as part of an exascale computing ecosystem capable of ensuring sustainable exascale computing capability for the nation.


Source: Lawrence Bernard for the exascale calculation project

Teresa H. Sadler