News

Expanse supercomputer. Credit: SDSC Communications

Solar panels and wind turbines increasingly dot the landscape, but the future of clean energy may well depend on how smoothly we burn hydrogen. Yet as anyone who’s lit a gas grill or fireplace knows, igniting a flame can be a bit tricky. Imagine how complex that process can be in commercial and industrial applications. Thanks to U.S. National Science Foundation (NSF) ACCESS allocations, University of California San Diego researchers have taken a major step toward taming hydrogen flames with highly precise simulations that could reshape how we design tomorrow’s zero-carbon gas turbines.

Led by Antonio L. Sánchez, a professor in the UC San Diego Jacobs School of Engineering Department of Mechanical and Aerospace Engineering, the team recently simulated hydrogen combustion to better understand how to increase safety when the chemical element is utilized by industry. Using ACCESS allocations on the Expanse system at the San Diego Supercomputer Center (SDSC) at UC San Diego’s School of Computing, Information and Data Sciences, the researchers modelled swirling jets of nitrogen-diluted hydrogen injected into hot, high-pressure air — conditions meant to mimic real-world gas-turbine combustors used in commercial and military jets, naval vessels and electric power plants, along with gas and oil industries. Their goal: figure out what flow conditions allow a so-called “lifted flame” (one hovering safely above the fuel injector instead of clinging to it or blowing off) to behave stably and reliably. The significance is profound. Hydrogen burns clean, but its very speed and intensity make it unstable unless carefully managed.

“What we’re doing with these Expanse-generated models is providing design engineers with a physics-based roadmap for how hydrogen can behave in real turbine environments,” explained Sánchez, principal researcher with the Jacobs School of Engineering Department of Mechanical and Aerospace Engineering. “When hydrogen flames go unstable, they can damage hardware, reduce efficiency and/or cost fuel-based operations enormous time and money.”

Tracking Every Spark in the Flame

One of the key innovations in the study, published in the Combustion and Flame journal, was the comparison of two Expanse-generated chemistry models. The researchers first used a full 20-step detailed oxidation mechanism, modelling the combustion process through a series of approximately twenty separate chemical reactions that showed exactly how hydrogen turns into water.

Instead of jumping straight from hydrogen and oxygen to water in one step, the model included every tiny stage in between — including the creation and destruction of short-lived particles called radicals (for example, hydrogen [H], oxygen [O] and OH — a combination of the two). These radicals appeared like “chemical sparks” that briefly formed and disappeared in millionths of a second, driving the flame’s chain reactions forward.

Tracking all of those steps gave the scientists a very precise picture of how the flame behaves — its speed, temperature and stability. This kind of simulation was extremely detailed and time-consuming to run. But, supercomputers like the SDSC Expanse system made it possible — with researchers formally dubbing the process the “20-Step San Diego Mechanism.”

The Shortcut That Works

With the second model, the team tested a simpler, faster simulation — one that combines all those reactions into just a single overall step.

This streamlined version assumes that the short-lived “in-between” molecules (called intermediates) stay at steady levels instead of constantly changing. Under the high-pressure conditions inside a gas turbine, that’s a reasonable shortcut, as those intermediates react so quickly that their amounts show negligible fluctuation.

The payoff? This simpler model produced nearly identical results to the full, detailed chemistry — but ran much faster on Expanse. That means engineers can now study flame stability and turbine performance more efficiently, without sacrificing accuracy.

“By showing that our reduced-chemistry model holds up, we open the door to rapid prototyping of burner designs,” Sánchez said. “That’s a major step forward for clean-hydrogen deployment.”

Swirl combustor ignites hydrogen molecules. Credit: SDSC, AI-generated image

Finding the Sweet Spot: How Flames Stay Stable

There are two key factors that determine whether a hydrogen flame stays steady or goes out:

• The Swirl Number (Sw): how much incoming fuel and air are made to spin as they enter the combustion chamber. That spinning motion creates a small, stable “bubble” of recirculating gas — like a tiny tornado — that can help the flame stay anchored instead of blowing away.
• The Damköhler Number (Da): a way to compare how quickly the fuel burns to how fast the air and fuel are moving through the system.

By adjusting these two conditions in their computer models, the scientists found there’s a “sweet spot” where the flame remains stable. If the swirl is too weak or the gas moves too quickly, the flame can blow out. If the reaction is too fast or the flow too slow, the flame can creep backward toward the fuel injector, increasing heating rates that may damage the injector rim. Sanchez said that the work also showed that this internal swirling “vortex bubble” — the zone where the flow reverses and mixes — is crucial for keeping hydrogen flames steady in gas turbines, a finding that could guide safer and more efficient engine designs.

Because many existing gas turbines were designed for natural gas and not pure hydrogen, this kind of insight matters: it helps engineers redesign injectors, swirlers and combustion chambers to handle hydrogen’s high reactivity and flame temperature safely and efficiently.

Supercomputing Power Behind the Science

SDSC's Expanse enabled the team to explore many combinations of swirl and flow conditions at high resolution — something that would have been impossible on smaller computing systems. The researchers ran their simulations under smooth-flowing, controlled conditions — known as laminar flow — rather than the highly turbulent, chaotic flow found inside full-size jet or power turbines. Even so, the team says the main lessons still apply — to keep hydrogen flames steady and efficient, engineers must carefully balance swirl strength, reaction speed, and chemical modeling methods.

“Large-scale compute resources let us explore a wide range of conditions in minutes or hours instead of weeks,” Sánchez commented. “That means engineering risks are reduced, and clean‐fuel timelines move up.”

From Research to Real-World Impact

For industry, the implications are clear: a validated “roadmap” for hydrogen flame behavior means fewer prototypes, less trial and error, and faster pathways to commercial hydrogen combustion units. For society, it means accelerated progress toward zero-carbon power generation — and cleaner air for everyone.

“Hydrogen will be an essential fuel if we’re serious about decarbonization,” Sánchez said. “Our job is to make sure it can burn safely, reliably, and efficiently in the systems of the future.”

Computational work on Expanse was supported by NSF ACCESS (allocation nos. MCH230034 and MCH230047).