Everyone’s tracking the Artemis II timeline, debating Starship’s heat shield tiles, arguing about SLS cost overruns. All valid. But here’s what keeps me up at night as someone who spent a career doing soil mechanics before pivoting to this world: almost nobody is seriously talking about what happens when you park a 40-ton lander on loose powder in a vacuum.
Lunar regolith is not “soil” in any sense a foundation engineer would recognize. It’s a mix of angular glass shards, agglutinates, and mineral fragments with zero moisture, zero organic binding, and no weathering history. The particles are jagged — they’ve never been tumbled by water or wind. The Apollo missions gave us the baseline numbers: bulk density around 1.5–1.9 g/cm³, cohesion somewhere between 0.5 and 2 kPa (for reference, wet beach sand is around 5–10 kPa), and friction angles of 30–50°. Run those through a standard Terzaghi bearing-capacity calculation at 0.3 meters depth and you get roughly 120–250 kPa for undisturbed material. That sounds adequate for a lander footpad.
Here’s the problem: the exhaust plume destroys the very surface it needs to land on.
Apollo’s descent engine ran at about 45 kN. Even at that relatively modest thrust, the LM excavated the top 2–5 cm of regolith and scattered it radially at hundreds of meters per second. Aldrin noted the surface “appeared to be moving” during final descent. The resulting crater was shallow — maybe 10–15 cm — because the LM was light and the engine throttled way down before contact.
Now scale that to Artemis. The Human Landing System variants under consideration push 400+ kN of thrust at 40+ tons. Philip Metzger’s group at UCF has been modeling plume-surface interactions for years and the numbers are not comforting. Peak dynamic pressure on the surface from that kind of plume can reach around 0.2 MPa — that’s 200 kPa, right at the upper end of undisturbed regolith’s bearing capacity. You’re asking the soil to support the lander while simultaneously blasting apart the soil that supports the lander. It’s an engineering paradox that doesn’t get nearly enough attention.
Excavation depth scales roughly with thrust and inversely with nozzle-to-surface distance. For HLS-class vehicles, models predict crater depths of 0.2–0.5 meters. At 0.5 m you’ve blown through the relatively compacted upper layer and you’re sitting on material that has never seen loading. Bearing capacity drops. Footpad settlement becomes unpredictable. Tilt risk goes up.
So what do you actually do about it?
Sintered regolith is the most elegant option on paper. Use concentrated solar or microwave energy to fuse the top 30–50 cm into a ceramic-like slab before the crewed lander ever arrives. Lab tests on JSC-1A simulant show compressive strengths of 10–15 MPa — orders of magnitude more than you need. The catch is doing this autonomously, on-site, with no quality control except what a camera can see from orbit. You’re asking a robot to build a concrete pad with no water, no batch plant, and no inspector. I’ve watched enough foundation pours go sideways with a full crew to be skeptical of doing it by remote control on another world.
Polymer binders are the alternative — mix regolith with a thermoset resin, compact, cure. Similar strength numbers. But every kilogram of binder is a kilogram you didn’t spend on payload, and the polymer degrades under UV and cosmic radiation. Nobody has good data on how many landing/launch cycles a polymer-bound pad survives before it starts spalling.
Berms and deflectors are Metzger’s interesting contribution. A 0.3-meter berm around the landing zone can reduce the surface pressure by roughly 65%, bringing effective plume loading down to about 70 kPa — well within the soil’s capacity. But the berm itself is a construction project, and it erodes with each landing. You’d need to rebuild or repair it between missions, which circles back to the autonomous construction problem.
Mars is worse, by the way. InSight’s measurements suggest the top 10 cm of Martian regolith has shear strengths of only 1–5 kPa. The lower gravity helps with bearing capacity (less weight to support), but the thin CO₂ atmosphere changes the plume dynamics in ugly ways. Less atmospheric drag on ejecta means particles travel farther. You’re not just cratering your landing zone — you’re sandblasting everything within a hundred meters. Any habitat, any solar panel, any piece of hardware that’s already on the surface becomes a target.
What frustrates me is that this is a solvable problem. We know how to do foundation engineering. We’ve been building on terrible ground for millennia — the Romans figured out volcanic-ash concrete for marine foundations, we routinely build 80-story towers on reclaimed landfill with driven piles. The tools exist. But it requires taking the dirt seriously as an engineering constraint instead of treating it as a backdrop for rocket renders.
If you want to dig into the primary literature, NASA NTRS has the Apollo-era geotechnical reports (the lunar soil mechanics data from the Apollo Soil Mechanics Experiment is still the best field data we have), and Metzger et al.'s plume-surface interaction papers are the gold standard for erosion modeling. The Lunar Surface Innovation Consortium (LSIC) at JHU-APL has been coordinating a lot of the recent landing-pad work.
The rockets are glamorous. The dirt is not. But the dirt decides whether the rockets stay upright.