Democratizing supercomputing via the cloud

A service making supercomputing affordable to more engineers is here, and it points to a new era of exponentially expanding computing power for aerospace simulation. Yes, it’s a bold statement, but I think the technology promises to make access to high-power computing comparable to the transformation the personal computer created.

Luminary Cloud aims to revolutionize engineering workflows with cloud-based software as a service (SaaS) for simulation and analysis. SaaS isn’t new but coupling it with Nvidia graphics processing units (GPUs) developed for highly detailed video game rendering allows immense parallel processing and accelerated data crunching. Additionally, the platform is web-based and software independent, so it can take any CAD file, any meshes if they exist from other workflows, and produce a simulation 100x faster than the competition.

“A single simulation that takes six hours for legacy vendors on computer-based resources, we can do in a minute or two, with resources of GPUs in the cloud,” says Dr. Juan Alonso, founder and director of Stanford’s Aerospace Design Laboratory (ADL) and co-founder of Luminary Cloud.

“We’re consumption-based, pay-as-you use, so no limits on users, no limits on the number of GPUs you want to use,” Alonso tells me.

Luminary recently emerged from beta testing, with customers including Piper Aircraft, Joby Aviation, and Flighthouse Engineering, plus automotive, sporting goods, and industrial equipment designers.

“With unlimited resources in the cloud, we can run many simulations simultaneously, producing a lot more useful and actionable information for engineers to do their job,” Alonso continues. This ability to do all these simulations can democratize simulation.

“Engineers don’t use simulation today because they don’t feel qualified or trained to do it. We believe the level of automation, the level of intelligence we’re building into the tools, the fast turnaround time, makes our tools accessible to engineers who are going to make the key design choices,” he explains. “We’re allowing all that engineering brainpower to come up with new ideas, test them quickly and accurately, with some guarantees, and make design decisions that lead in the right direction.”

Luminary’s artificial intelligence (AI)-based engineering design copilot, Lumi AI, reduces the time engineers spend in setup and simulation, so they can spend more time analyzing and optimizing.

“People spend an inordinate amount of time cleaning CAD so it can be ready for simulation. AI large language models can free them from spending hours clicking around, fixing geometry before the analysis,” Alonso says.

Also, it’s possible to share the simulation for collaborative input, view it, comment on it, even run additional calculations, so a suggestion can quickly be turned into an action item for the engineering team.

Alonso adds, “We’re hoping to enable new ways of embedding simulation into engineering, or to look for more alternatives, and reduce risks or the number of physical prototypes needed.”

I believe easy access to iterative simulation bodes well for the future of aerospace engineering. – Eric