August 21, 2026
Cheyenne, Wyoming – Big Blue Technologies (BBT) has partnered with the National Laboratory of the Rockies (NLR) and Hewlett Packard Enterprise (HPE) to apply high-performance computing to one of the hardest problems in bringing primary magnesium production back to the United States: understanding exactly what happens inside an industrial smelter that no one can see into.
The collaboration, recently featured in HPE’s Digital Game Changers series, used HPE’s Kestrel supercomputer, an HPE Cray Supercomputing EX system, to build high-fidelity simulations of BBT’s magnesium smelting and condensation process. The work is directly informing the design of BBT’s commercial demonstration facility now under construction in Cheyenne, which is expected to produce up to 1,000 tons of magnesium metal per year.
A Critical Mineral With No Domestic Supply
Magnesium is essential to aluminum alloying, lightweight automotive and aerospace components, titanium production, and defense applications, the United States has no primary magnesium production, despite abundant domestic ore reserves.
BBT was founded to close that gap, but the company’s leadership has been clear from the start that proving the underlying chemistry is only part of the challenge. Scaling a smelter from pilot to commercial size safely and cost-effectively requires understanding conditions inside the reactor that simply cannot be measured directly with physical instrumentation.
Simulating What Can’t Be Observed
Working with NLR researcher Rajneesh Chaudhary, BBT’s team modeled the full physics of its process within the complex geometry of its smelter and condenser. Some simulations ran continuously for two days at a time once the models stabilized, a workload that required the scale of Kestrel’s computing capacity.
The simulations delivered concrete engineering insight the company is applying directly to its scale-up design:
- Heat loss mapping: identifying where energy was escaping around the central electrode and eliminating localized insulation and electrode hotspots that had been destabilizing the reactor.
- Electric arc furnace parameters: modeling voltage, amperage, and power against BBT’s highly conductive slag to guide the design of the scaled-up smelter.
- Vapor and condenser dynamics: tracing how magnesium vapor moves from the smelting zone into the condenser, revealing flow paths that affect both product yield and energy efficiency.
With the smelter accounting for roughly 70% of the energy demand of BBT’s overall production process, even modest efficiency gains identified through the modeling translate directly into lower operating costs and improved commercial viability.
From Simulation to Steel in the Ground
That computational work is now shaping a real facility. BBT has broken ground on its commercial demonstration plant in Cheyenne, Wyoming, engineered using insights from both the NLR simulations and lessons from the company’s pilot campaign, which successfully produced commercial-grade magnesium ingots.
Looking further ahead, BBT estimates that a full-scale, 20,000-ton-per-year commercial plant built on the ATR process would represent roughly a $150 million investment, underscoring why de-risking the scale-up through simulation, rather than trial and error at commercial scale, matters so much to the company’s capital efficiency.
Read the full HPE Digital Game Changers feature, “Bringing Primary Magnesium Production Back to the US,” at hpe.com.
