NASA geeks code new tricks to model rocket plumes and avoid a lunar dust-up

NASA researchers have developed tools to simulate how rocket engines disrupt the lunar surface in light of plans for newer and larger landers as part of missions to return to the Moon.

The Artemis program is NASA's project to return humans to the surface of Earth's natural satellite, and the landers involved will be larger and equipped with more powerful engines than those used in the Apollo program half a century ago.

This means that the mission risks associated with their operation during landing and liftoff are significantly greater, according to the space agency. The rocket engines blast supersonic plumes of hot gas toward the surface and the intense forces kick up dust and may eject rocks or other debris at high speeds from the regolith – the loose, unconsolidated layer at the surface.

Naturally, there are fears that this can cause hazards such as obstructed vision for the astronauts with dust clouds that might interfere with navigation and the scientific instruments, or even cause damage to the lander itself.

These hot supersonic plumes can also erode the surface under the lander, and although the Apollo era landers did not carve out craters, NASA said it is unclear how much the larger landers planned for Artemis missions will erode the surface and whether they might cause cratering, potentially posing a risk to the lander's stability and the astronauts on board.

To better understand what NASA has termed plume-surface interactions (PSI), researchers at the Marshall Space Flight Center in Huntsville, Alabama, have developed new software tools to predict "PSI environments" for NASA projects and missions, including the Human Landing System and future Mars landers.

The Human Landing System is actually more than one lander, developed for NASA by commercial space operators. The first crewed landing is expected to use a Starship HLS being built by Elon Musk's SpaceX, while another lander is being built by Jeff Bezos' Blue Origin for a follow-on mission.

• RIP: Frank Borman, NASA commander of first Moon mission
• NASA's Lucy probe scores a threefer as it flies by first target in 12-year mission
• NASA gasping for ideas to extract oxygen from Moon dirt
• Bids for ISS demolition rights are now open, NASA declares

According to NASA, the Marshall team recently produced a simulation of the Apollo 12 lander engine plumes interacting with the surface of the Moon, and the predicted erosion closely matched what happened during landing.

An animation released by NASA depicts the last half-minute of descent before engine cut-off, but illustrates the predicted forces exerted by plumes on a flat computational surface, with fluctuating radial patterns showing the intensity of the predicted shear stress.

NASA said the simulations were run on the Pleiades supercomputer at its Advanced Supercomputing facility at the Ames Research Center in Silicon Valley. Built in 2008 by SGI, Pleiades is now considered quite an old system and originally featured Intel "Harpertown" Xeon processors, but was periodically updated and was still ranked as the 32nd most powerful computer on the TOP500 list in 2019.

According to NASA, Pleiades comprises 10,410 nodes in 152 racks interconnected by InfiniBand, with a total of 228,572 CPU cores and 921 TB memory.

The tools developed by the Marshall team allow researchers to determine how to best meet simulation and time requirements for each project by varying model fidelity, NASA claimed.

The highest fidelity tool is said to be the Gas Granular Flow Solver (Loci/GGFS) that models gas-particle multi-phase interactions to "predict regolith cratering and ejection of particles" into the area surrounding the lander.

At its highest fidelity, this can model microscopic regolith particle interactions with a particle size or shape distribution that replicates actual regolith, but NASA said that to make most effective use of the compute resources, it is currently run using only one to three equivalent particle sizes and shapes.

"A high-fidelity PSI model is extremely computationally intensive," stated Jeff West, Team Lead for Computational Fluid Dynamics at the NASA Marshall Space Flight Center.

The simulations can require meshes with over 200 million cells, leading to runtimes on the order of weeks, and generating terabytes of data.

West said that the toolset can support NASA's needs today with current agency compute power, but "we have strategically designed our tools to grow incrementally in fidelity as NASA's computing capacities increase." ®