To the Drawing Board

In the February 11, 2020 edition of “The Space Show,” Robert Zubrin shared what he learned in a visit to the “shipyard” at Boca Chica. He spent time talking with Musk and with SpaceX engineers, and there were a variety of key takeaways. Most important, Starship is expected to cost $5-20 million per unit, when mass produced. We hinted at that price point based on the September 2019 update; it’s vastly lower than the original estimates. Significantly, that number is based on Boca Chica literally being a shipyard—SpaceX intends to churn out these ships rapidly. Musk suggested two per week, although Zubrin was skeptical, opining they might start at one a month and reach one per week. Even at the rate of one per month, though, the numbers are significantly higher (and costs significantly lower) that what we’ve been planning. “Mass produced” includes lots of tankers and presumably lots of cash-producing suborbital passenger ships. But, as we pondered in the Day 20 update, it also suggests the probability of many vehicles on a one-way mission, with just enough returning to assure we can bring people back to Earth, for all the reasons outlined earlier. In February, Earth orbit by 2021 seemed possible to Zubrin, assuming no significant issues with Super Heavy development, not yet underway in any significant sense. Since then, a “high bay” structure suitable for stacking a Super Heavy has nearly completed at Boca Chica. SpaceX is now hinting at an early 2021 date for the first Super Heavy flight, even as the first 15 km hop is getting close.

In the Space Show interview, the item I personally found most interesting was a nugget from Zubrin suggesting SpaceX wasn’t yet seriously looking at the ISRU problem. A recent “What About It” episode on YouTube suggests SpaceX may be using its Florida location to work out some of the technology needed for propellant production. (Episode 120, linked below.)

But, in the Space Show, Zubrin stated he had specifically asked whether Musk understood the size of the solar power fields necessary to power the ISRU plants. When Zubrin described it as multiple football fields (which we already established), Musk essentially shrugged and said, “Fine,” with an implicit, “Then we send shiploads of solar panels.” Exactly what we derived, but perhaps we aren’t reverse engineering what SpaceX has already figured out, after all. That’s exciting. In the conversation, Zubrin indicated the ISRU plant might be able to get by on 600 to 1,000 kilowatts. This is about one-third to one-half of the number we came up with, but we don’t have deep knowledge of how well his ISRU design scales. Regardless, that’s a good crosscheck that our numbers aren’t wildly off.

Zubrin also suggested that the first crew might be on the order of 20 to 50. Although it wasn’t clear whether that was his own number or Musk’s, the same figure has since been quoted by others. As outlined earlier, that seems high.

Where does this leave us?

Regardless of how many hulls it takes, eventually, a bare-bones vehicle will make it to orbit and return. Once that occurs, SpaceX will have a choice: work on a heavy lift option, delivering cargo to Earth orbit; or work on life support, which could support the #DearMoon mission and support suborbital, revenue-producing flights. SpaceX might take both tracks. Time will tell, but there are implications for each.

Although the suborbital option lets us perfect life support and high-rate landings, passenger-rating is orders of magnitude beyond human-rating. Musk has stated as much, suggesting “hundreds” of unmanned flights before the first humans fly on a Starship. On the other hand, demonstrating Starship as a heavy-lift vehicle might finally get the community to adapt its future plans to the reality of Starship’s existence. For that reason alone, a cargo-to-LEO version is much more compelling. Every launch requires a landing, but every launch has the option of orbital refilling, whether to reach a higher orbit for payload delivery or simply for practice after delivering a primary payload. SpaceX made this clear in its March release of a “Starship User Guide,” available on its website, https://www.spacex.com/sites/spacex/files/starship_users_guide_v1.pdf

The animations have suggested it; let’s go there. If you haven’t seen the James Bond film “You Only Live Twice,” it’s mandatory viewing for Starship aficionados. The giant clamshell door of a cargo Starship practically begs for a remake of the film’s opening sequence. The possibilities here should scare the daylights out of every competitor—as the user guide points out, they include not just multiple satellites, but huge observatories, orbital cleanup, or simply transferring operational satellites to new orbits.

In his Space Show interview, Zubrin pointed out that landing a Starship on the moon was likely impractical, because the size of the vehicle means the Raptor engines will dig a crater so large the ship itself will tip over. He went on to discuss the need for a purpose-built, smaller lunar lander, at least until landing pads could be built. Not mentioned in the Space Show, but brought up a couple months later in an interview with "What About It,", is that every proposed component from every reference architecture could delivered by a cargo Starship. (Part 1 of the interview linked below.)

Need a LEM in lunar orbit? Send it in a cargo Starship. Need the Lunar Orbital Gateway Tollbooth that Zubrin hates? Forget every other launch vehicle. If Lockheed Martin can put the thing together, send it safely cocooned inside a reusable cargo Starship. ULA won’t like that, but see “reality,” above. NASA seems to be warming to SpaceX’s capabilities, announcing that SpaceX’s “Dragon XL” will deliver cargo to the Lunar Orbital Gateway and selecting SpaceX as a contender for delivering lunar cargo. Since that interview, the concerns about Raptor engines having been apparently solved with large landing thrusters halfway up the rocket. NASA also selected SpaceX as one of three organizations to begin development work on lunar cargo delivery. SpaceX has made clear the plan for those lunar freighters is a one-way mission from Earth, so there’s a lot of architecture that needs a ride.

Need a fuel depot for your LEM? Okay, let’s make the Lunar Orbital Gateway an actual fuel depot, but don’t send the fuel from the parched lunar surface; send it using reusable tankers from a place where water covers 70% of the planet’s surface. If you’re making your own propellant using wind (and solar?) power at the launch site, the economics really start to change.

Worried about space debris from your defunct satellite? Send a Starship to go collect it. (Shotwell recently suggested as much.) Of course, you’re probably not going to launch a Starship just to bring back junk, but the transportation industry is very familiar with the concept of “backhaul.” You make your money delivering freight, but then you need to get your container/freight car/semi-trailer back to the starting point. Any load on the return trip is better than none, even if the return trip still doesn’t break even. Alternatively, maybe you don’t bring those satellite back and instead make the mission dedicated to cleanup—with a 100 ton payload and refillable propellant, how many tethers could you take to orbit for deorbiting defunct satellites?

Your hundred million dollar satellite malfunctioned after launch? Let’s go get it. It needs refueling, or new solar panels? Can we pencil you in for an appointment next week?

For all practical purposes, the limitations on mass to orbit simply vanish. At $5-20M per launch (and Musk now hinting at $2M), with 100 tons to orbit per launch, the problem isn’t how to afford it. The problem is how to get your idea from drawing board to flight-ready as fast as possible, and no worrying about shaving a few kilograms.

For planetary exploration, we’re in a similar boat. Stop worrying about landing a lightweight rover on Mars, or a lightweight science instrument. Instead of Mars Insight’s tiny little mole, we’re talking about an industrial-grade drillship. We could send half a dozen freighters to half a dozen potential landing sites, fully aware that five of them will be abandoned in place when that site isn’t selected. We could also finally send that subsurface radar imager Zubrin described in “Drilling Operations” by aerobraking into Mars orbit. And if we’re not yet ready to send astronauts in the early Earth-Mars launch windows, how about aerobraking into Mars orbit and doing a close recon of Phobos? For that matter, why not just deliver a dedicated Phobos lander?

In Day 20, we concluded it was too soon to rework the whole architecture. At this point, we can see enough to posit that our six-ship cycling architecture probably doesn’t need to be that literal. We will still need the capability to bring crew and cargo back to Earth, but it’s starting to look as if any cargo we might need is simply a matter of ordering up another MFR. The Martian Colonist Youtube channel analyzed Musk’s public discussions and reports the plan is two freighters in the initial launch window, followed by three freighters and two MCCS in the second, with that same combination in the third. Pretty close to what we’ve already derived, give or take a ship.

Every additional ship can bring another hundred tons of cargo at a cost of, say, $10M. That’s $100 per kilogram. For comparison, this is about the same cost per kilogram to build an aircraft carrier. Expensive? Sure. But not so expensive that a major nation state can’t build a ship that displaces 100,000 tons. Correction: for a superpower, a dozen such ships. We aren’t talking about an aircraft carrier, though, which is good—those cost $10 billion. But we’re also not talking about the Kon Tiki. We’re talking about the Mayflower.

In comparison, the original Interplanetary Transport System advanced spaceflight like Shuttle compared to Apollo—impressive, but still vastly expensive. This new design is the Maersk Viking compared to the NS Savannah; literally an economic game-changer.

This realization brings us to an important shift. Up until now, we’ve been broadly discussing all the things SpaceX is going to do. We’ve informed this discussion with the things SpaceX has said in public, and made a variety of—hopefully reasonable—assumptions describing a likely overall architecture. But, as we said earlier, that’s all stage setting. Next, we need to move to what we’re going to do.

We have a basic timeline, a general list of resources, overall sizing, and a personnel flow. We have precursor ships on the surface, and now realize we can probably assume any reasonable cargo we might need. We also have a set of assumptions about things that must occur, and which tasks have priority claim on cargo space (mass and volume) and human labor once it arrives. Change any of this, and what unfolds after necessarily changes, too. Overall, though, we have a workable mission architecture to start from.

SpaceX has put us on Mars. Now, we have to live there.

The Pioneer Astronautics paper, “Drilling Operations to Support Human Mars Missions,” that we read on Day 22 describes an array of orbital and surface instruments looking for water. Since 1998, when the paper was written, a multitude of missions have explored Mars, and NASA has even adopted a “follow the water” approach to exploration. Some cursory research finds a variety of papers supporting the “Drilling Operations” assumption that liquid water is likely available in some areas within a kilometer of the surface. However, it doesn’t appear that we’ve really obtained widespread data about the likely depth of liquid water. NASA missions planned for the near term (e.g. Mars 2020) are still proceeding with a non-Starship architecture—bigger than their predecessors, but still orders of magnitude less than we can propose, and not sufficient to give us the answers we need about the location and depth of Martian aquifers.

It’s not clear whether a Starship can actually aerobrake into Mars orbit, although we discussed that possibility earlier. If it is possible, then delivering the orbital instruments proposed in “Drilling Operations” should be a near-term goal. In particular, a ground-penetrating radar is called for. That might be two satellites delivered by Starship, two Starships configured with the necessary equipment, or one Starship with the transmitter built-in and a separate receiver deployed after achieving Mars orbit. We’ll set that aside, other than to note that it’s needed early, so it’s an excellent candidate for the first or second set of precursor missions.

We also need a comprehensive suite of ground-based seismic instruments. We’ve already discussed automated setup of our solar power field; now we have another candidate for early cargo. Once the power field is operating, even before we start our “strip mining” demonstrations, we might consider the water-seeking operations described in the paper. Instead of the four landers called for in the paper, our tractors can deploy four (or more) ground instruments. And, instead of the lightweight “mole” on board NASA’s InSight lander, we can use something far more robust—after all, we’re talking about using hydraulic hammers for strip mining operations, so pounding a few sensor suites a couple meters into the ground shouldn’t pose too much of a problem. Finally, we can assemble a series of mortar-based charges into a platform that will fit on our trailer and haul it to a safe distance before firing them.

Identifying subsurface water would be game-changing for follow-on operations, so much so that we should spend a few moments considering changes to our original design architecture. At the very least, Finity’s End needs to bring the seismic instruments we’ve just described. It would also be very useful to get Finity’s End on the ground and searching before Constellation lands. And, if it turns out Finity’s End is in a location without an underground aquifer, Constellation should land at a predetermined alternate site. Connie, too, needs at least minimal seismic instruments, even if we keep it as a MCCS, with life support. But, if we have to make a trade, we should seriously consider converting Constellation from a MCCS to a Mars Freighter, so we have the option of diverting to an alternate site. We won’t replan the architecture for now, but we’ll put this marker down for serious consideration.

If we’re able to identify subsurface water at our landing site, we also have a new challenge—the second launch window already has three ships—Nostromo, our ISRU plant; Serenity, hauling solar panels for Nostromo; and Heart of Gold, the first manned ship. If there’s any way to pull it off, we want that drillship, too, since that makes the whole strip-mining-for-ice/water unnecessary.

Given the changes to make Starship’s design more affordable, this seems like a plausible addition. We’d now have four ships in the second launch window. The biggest problem, in all likelihood, isn’t a Starship hull. Rather, it’s the amount of specialized design work we now need. Musk has already stated SpaceX is working on propellant plants, and there’s an obvious need for solar panels to power it. If we’re sending people, then we need a ship with life support and supplies. But, if we’re now going to expand into well drilling, that’s not exactly a SpaceX specialty. Like the container cranes earlier, the drillship seems a likely candidate to bring in outside help.

The type of specialized drilling we’re talking about isn’t the sort of work a water well driller is going to be familiar with. But, it’s well within the skillset of any major offshore oil well operator—much simpler, by comparison, than an earthbound drillship. We joked earlier about using “Deepwater Horizon” for the refinery ship. It’s unlikely anyone will reuse that one, even though it has a nice ring to it. However, in addition to bringing in outside expertise, this also seems like a good candidate for outside investment. In particular, a major company might well be interested in financially supporting the design work in exchange for naming rights. Companies pay tens of millions to name stadiums...why not tens of millions in investment for the drillship to be named “Maersk Viking” or “Exxon Ares”? That would only require SpaceX to provide the hull, outsourcing the rest. In particular, the outside company could also handle the logistics of training the drillers, possibly even selecting them.

Again referencing the Pioneer Astronautics paper, that plan called for a three-person drilling crew. Would it be better to find three oil well drillers with the right skillset to become astronauts for five years, or spend months to years training astronauts to drill a single, deep water well? “Drilling Operations” suggests the former, but assumed there would already be a large amount of infrastructure before they went. This will be a single well, completed in weeks to months. Our drillers will need a variety of other skillsets, so the decision doesn’t seem settled.

For now, let’s call the drillship “Deepwater Pioneer,” recognizing the Pioneer Astronautics work that might make it a reality. We’ll hold off on making it part of our reference architecture, for now. But that doesn’t mean we won’t advocate for it.

If we were reasonably sure there’s liquid water within a few hundred meters of the surface, then bringing along well drilling equipment would be worth the substantial mass required to obtain it. A fair amount of research has been done on this topic; for an excellent primer, see “Drilling Operations to Support Human Mars Missions” (Frankie, Tarzian, and Lowther: 1998, Pioneer Astronautics) for a discussion of one proposed approach, along with some of the challenges. That paper assumes a fair amount of well drilling knowledge, so let me provide a couple introductory items I learned from more Earthbound resources before sending you off for this homework. The site https://www.drillyourownwell.com is dedicated to just what it sounds like—hand-drilling very shallow irrigation wells with extremely simple equipment. Alas, we discover the most critical ingredient for water well drilling is...water. As the drill bit cuts through the ground, water is used as a “drilling fluid” to wash the cuttings up to the surface. Water can also be mixed with bentonite to make “drilling mud,” which is more efficient, but it still needs water.

A quick side note here—recall the movie, “Armageddon,” in which a team of professional oil well drillers, including Bruce Willis, [spoiler alert] save the planet by becoming erstwhile astronauts, drilling a hole into an asteroid, and dropping a nuke down the shaft. They didn’t use any drilling fluid in the movie, which raises the question—how did they get the cuttings out of the hole? They could have used something as simple as pressurized air, since the asteroid had almost no gravity, but the story didn’t consider the problem at all. We don’t get the luxury of Hollywood writing.

The Pioneer Astronautics paper proposes using liquid CO2 as a drilling fluid. This requires that the well be pressurized, but it turns out that’s not unheard of. The proposed well also uses a non-rotating shaft with a hydraulically powered rotating drill head. At Drill Your Own Well, there’s a great video of a rotating shaft, but a little online research reveals that deep drilling often uses a stationary shaft and then has a complicated, hydraulically powered drill bit on the business end of the well. In Pioneer’s “Drilling Operations,” liquid CO2 is used as the hydraulic fluid to power the drill bit as well as being the drilling fluid. CO2 requires about five atmospheres to maintain liquid form, or about 75 psi. The proposed well uses a four inch borehole, minimizing the amount of force we need to counterbalance. (Warning: researching well drilling online is likely to cost you as much time as using Google/Bing maps and wondering, “Where does that railroad track go?”)

A final introductory aspect is that drilling mud is important to balance the weight of the drill string—the mud provides buoyancy for the drill pipe and drilling head.

The paper is available at https://marspapers.org, but unfortunately does not include any of the figures or tables referenced by the text. (I asked Pioneer Astronautics for a complete copy, but received no response.) Still, with careful reading, we can get a pretty good idea of what the authors were thinking 20 years ago under a very different reference architecture. To avoid too much repetition, the rest of this post assumes you’ve read it. It’s worth the time.

Welcome back. A couple items immediately come to mind from the Pioneer Astronautics paper. First, a Starship-based architecture makes this approach eminently feasible, even early on. The aft cargo pods of a Starship seem like a perfect place for a wellhead, and the overall mass of a Starship means the vehicle can handle the 10,000 psi expected of a “kick.” With a four inch borehole, the kick will produce about 62 tons of force—a Starship logs in at 100+ tons. Of course, we’ll need a better understanding of how kick forces operate in 0.4 gee, but we seem to be in the ballpark.

Further, the power requirements for the proposed drilling operation are actually quite feasible with solar power given the enormous capacity of Starship. We should also be able to bring more bits than proposed, more pumps, and more tools—also the sort of equipment that would likely be quite happy in aft cargo compartments. The proposed plan budgets just over seven tons for drill pipe, so we can easily bring more, and potentially trade more mass for greater corrosion resistance, if desired. And, in addition to the capacity of our primary drill ship, drill pipe certainly seems like the sort of non-complicated cargo that a bare-bones Starship freighter might haul.

There is the issue of how to connect the wellhead to our drilling platform. Starship’s cargo bay should be reconfigurable to support the operations described in “Drilling Operations,” to include pressurized operations, crown block and draw works, additional drill pipe, etc. The mud pump and baffles would be set up on the surface, since the cuttings will need to be periodically dumped. The one problem is that Starship has a pair of very large tanks between the aft cargo compartments and the main cargo compartment, so the drill ship will need a special configuration. It seems relatively straightforward to modify the tanks with a six inch (or so) raceway from top to bottom to allow the drill string to pass through. There doesn’t seem to be a particular reason why the drilling equipment can’t simply be positioned in line with one of the aft cargo compartments, thus minimizing changes to the engine configuration.

The second item immediately obvious from the Pioneer Astronautics paper is that we first need to establish the presence of liquid water deeper below the surface. The “strip mining” option seems to be something we can verify as a precursor mission, which means it’s automatically available as a backup. But, before we send a Starship (or two) with well drilling equipment, we really need to be sure there’s water down there. That means more than a couple seismic instruments; we’ll need a whole suite of gear. That’s up next.

Earlier, we lamented the loss of Red Dragon and the precursor missions that might have been. But, perhaps we lamented too soon. Musk has been pretty clear that the first robotic ships on Mars will do things that are essentially precursor missions, just using a full-size ship instead of a Red Dragon. That makes our precursor missions a lot more interesting.

For starters, we want to keep risk manageable, and we only have so much time to program autonomous actions into the rovers. It seems likely, for example, that once those solar power fields are set up, the rovers won’t spend a lot of time amongst the panels clearing off dust. Too much can go wrong. We might need a contingency plan here in case of a major dust storm, but let’s set that aside for now.

This leaves the autonomous tractors available for exploring the local area, using a variety of scientific instruments—whatever we can fit on the trailer, and maybe even on custom trailers. We have two years, more or less, and can recharge the vehicles at will. (We might even want to equip the trailers with a solar panel once they’re done hauling cargo for now. Then, as we tow them around on excursions, we have an emergency charging capability to get back to the ships.) We might consider bringing the tractors back on board during Martian winter to minimize thermal cycling. It might also be worth bringing one on board in the event of a major dust storm. (We’d probably want at least one available on the surface for contingency ops.)

We should also ponder what sort of equipment would be appropriate for the aft cargo compartments. (In the new design, it looks like there are still three.) The upside of these compartments is that we can probably just drop things onto the Martian surface without breaking them. The downside is that they’re right next to the engines, so the vibrations are going to be insane. We’ll note these items as we go along, but we can identify one right now.

If it can survive the vibration environment, one piece of equipment we’d really like on the surface quickly is a suite of seismic instruments, so we can take advantage of the next ship’s landing to gather data about the local geology. Raptor engines blasting the surface and a solid thump of a couple hundred tons ought to beat a little seismic explosive charge. For redundancy, a twin set of instruments would be very desirable.

We want to understand the local geology for two reasons. First, it may be unlikely, but we’d really like to know if there are any voids that could threaten follow-on landings. The last thing we need is a beautiful 4K video from Constellation showing Heart of Gold breaking through the surface and tipping over on landing. Second, a major task for the first ships is to verify the presence of water/ice.

Musk has been explicit about this, but without any specifics. We have a couple options here. First is figuring out how much water is present as ice in the soil just below the surface, a few centimeters or so. Second is figuring out whether there’s actually liquid water farther below the surface.

For the first, returning to https://planetaryprotection.arc.nasa.gov/file_download/90/29-Sanders.Mars.ISRU.PP_Sanders.V2.pdf, slides 10 and 11 illustrate the sort of “small” prospecting gear we might use. And https://www.nasa.gov/sites/default/files/atoms/files/mars_water_isru_planning-hangout-5-24-16.pdf slides 11 and 12 give us a sense of how large an area, and how much depth, we might need to excavate. The presentation shows multi-centimeter depth, not multi-meter depth. This is extremely promising, since that suggests we simply need front-end loaders to get the “ore” we need to process.

Thinking about how this might work with the robotic precursor ships, our first observation is that we literally need a front-end loader, not a tractor with a robot arm. Specifically, we’re suggesting an industrial-grade vehicle, not anything remotely like a Curiosity or the Phoenix lander. Although it can be lighter construction than would be needed on Earth, at the low end, we’re looking at something from the Northern Tool & Equipment line (see: tractors); at the high end, the Caterpillar 900 series. This will be an electric vehicle, of course, and we’ll need to determine whether hydraulics can be an option at Mars temperatures. Since it will be electric, we probably want to drag a power cable out to the work area so it can recharge in between loads. That presumably means we also need a small robot arm on the loader so it can plug itself in. This has the added advantage that we can put the loader to dual use, picking up things we want to look at more closely, for example, a scientifically interesting rock. We might just look at it with the loader’s cameras, or we might bring it back for a true science rover to examine.

We’ll be loading some sort of (electric) dump truck, which will haul our cargo back to the ships for analysis. Eventually, they’ll be hauling this ore to a processing point next to Nostromo, but for now, we have a stationary lab that will determine the water content. We won’t actually do anything else with the product, yet, other than dump it in a convenient pile. Nevertheless, we probably want two dump trucks, so we can get a better sense of timing. The loader scrapes off a few centimeters of soil with each pass, loading the truck. The truck heads back for analysis. Meantime, a second truck is arriving at the loading point. The two trucks cycle between the loader and lab, primarily to let us verify the process, to include timing. For full-scale propellant production, should there be two trucks per loader? Three? Four? Keep in mind both loader and trucks will need periodic recharge. It seems the best way to be sure about this will be to first test the arrangement on Earth, and then verify it, day after day, on Mars. Of critical importance—determining whether the loader has enough power to actually scrape off those few centimeters, or whether the surface material is too hard. If it’s too hard, we’ll need some alternatives. For now, let’s assume the loaders will be successful, even if we have to take smaller bites. Watching the road and subdivision construction near my house, it certainly seems there are ways to mechanically break up hard soil. We can reexamine this assumption later.

Accessing ice will likely require removing a lot more overburden, which was also examined in our references above. Before we start digging deep, we’ll want to take lots of those surface samples to be confident this landing site can provide water from the soil. After that, then we can explore whether the site also has ice, and how deep it might be. Confirming ice would let us radically change our approach, dramatically reducing the amount of bulk ore to be processed—as long as it’s not too deep. See slides 10-13 in the “Mars ISRU hangout” pdf.

But, as long as we’re digging, we might consider whether we can combine activities. We have two years, after all. Once we have power to the ships, and we’ve established that the site has accessible water, what’s next? While there are plenty of possibilities for robotic ships, there’s one overriding concern for the crewed ships that will follow: radiation protection. Specifically, we might as well see whether we can dig a deep enough, large enough trench to place a hab brought by the first crewed missions. We could then process the removed material and use it to cover the hab. In the digging process, we will identify exactly what we’ll find a few meters under the surface, and incidentally bring tears of joy to planetary geologists everywhere.

Digging more than a few centimeters deep will likely require a jackhammer attachment on the loader. If you’re searching, try “hydraulic hammer,” but that reminds us of our earlier question about what sort of hydraulics will work at Mars temperatures. We’ll set the hydraulics concern aside for now, just noting that we need some way of breaking up the surface. Zubrin recommends explosive charges, which might also be a good thing to test out before we have people around.

Arguably, though, a better solution is to drill for water. We’ll talk about that next.

We’ve identified that goal #1 after landing is getting enough power to the ships to keep them healthy. We sketched out a preliminary design for unloading a tractor, trailer, and sets of solar panels. Now, let’s dig a little deeper. This is still work that SpaceX will need to do, but having a general idea of what it needs to look like will inform our designs.

The first question is, “How much power does a ship need if there’s no crew?” A NASA Spaceflight Forum study (https://www.nasaspaceflight.com/2018/08/evolution-big-falcon-rocket/6/) suggested a 100 kW figure for a ship with crew, but since power is so critical, using this figure for every ship gives us a valuable safety margin. Absent better information, we’ll use that. Recall our NSP D6M panels provide approximately 214W apiece. (See https://www.mitsubishielectricsolar.com/images/uploads/documents/specs/NSP_350W_355W_360W_lo_res.pdf) With dimensions of 1m x 2m x 35mm, they mass 23 kg each. On Earth, they would weigh about 50 pounds. On Mars, they’ll weigh 19 pounds—technically, pounds force, or lbf. That’s 86N, but I’m betting that’s a meaningless number even for those of you in metric countries. (An interesting aside—since Mars has only 0.38 Earth gravity, what units will permanent Mars residents use when they talk about how much something “weighs”? On Earth, we say something is X kilograms, but we unconsciously factor in 9.81 m/sec2, so we have a sense of what 23 kilograms “means.” It won’t mean the same thing on Mars.)

These panels will need to be placed autonomously, which will mean something between a robot arm and crane. This won’t be an ISS-style Canadarm; rather, something more like an industrial robot crossed with a backhoe. With a requirement to only move a 19 lb. panel, this can still be a fairly lightweight setup. A spreader seems to be the most practical grappling solution; for only one panel, that can be a lot lighter than our ship’s crane. The arm will need the ability to switch between the spreader and other types of tools.

Sticking with the D6M panels, we’ll need about 467 of these to provide 100 kW power. Of course, that’s just average irradiance, so we’ll be looking for margin. We’re using these panels to charge batteries, initially aboard the ships and eventually within the base. A couple of interesting article at https://www.washingtonpost.com/news/energy-environment/wp/2015/05/01/what-backing-up-your-home-with-teslas-battery-might-be-like/?noredirect=on&utm_term=.04359f447eba and https://en.wikipedia.org/wiki/Tesla_Powerwall#cite_note-wp2015-05-13 provide some useful specifications.

First, it appears we want 400-450V for charging Tesla-manufactured lithium-ion batteries. Conveniently, the D6M’s can be wired in series to obtain a variety of voltages up to about 1000V, or in parallel to increase current. Twelve in series will give us 450V at 5 amps. (We’ll round down in all these power calculations.) This “voltage segment” package will amount to 276 kg, or 100 “Mars pounds” on the trailer. We’ll put a junction box at the start of each voltage segment. For deployment, let’s assume rows with 5 voltage segments of 12 panels each, making the row 60 panels long. Nine of those rows gives 540 panels and 115 kW. Each 12 panel voltage segment is wired in parallel across the 9 rows, so we have 108 panels providing 45 amps at 450V per power segment. The five power segments feed the ship independently. This provides redundancy, and means our junction boxes, cabling, etc. only need to handle 450V at 45 amps constant power.

To store this power, a Tesla Powerpack 2 is spec’d at 200 kW. It’s only rated to -30C, but that just means it needs to be aboard the ship or in a partially heated part of the base when it comes time to offload. More importantly, it’s only 1.6 tons and 2.2m x 0.8m x 1.3m. That’s 616 Mars pounds, which shouldn’t present a serious problem for our crane, when the time comes. For redundancy, let’s initially plan on two for the ship, with each ship bringing two more for the base. (Until it is time to offload, we probably want the base Powerpacks also tied into the ship grid.)

540 panels masses 12.4 tons...which is a reminder just how much flexibility we have with the scale of these ships. Adding in four Powerpacks is another 6.4 tons. (Technically, the ship will need battery storage regardless, but it’s more conservative to count them all as cargo.)

We previously outlined generally how we would deploy these panels. With a few more details worked out regarding number of panels and layout, we can return to the deployment scheme. First, for reasons we’ll get into in a later post, I’m going to recommend sizing the trailer at 2m x 3m. Second, it seems that we’ll significantly simplify the deployment process if each voltage segment is pre-wired in series. A stack of twelve 1m x 2m panels, the junction box, and associated cabling is a reasonably sized bundle that should easily fit on our trailer. Stop #1 for the tractor trailer is the ship’s fixed fin, which seems the perfect location for the power connections. The tractor unhooks, rolls around back, grabs the power cable, and plugs it into the ship. After reconnecting to the trailer, it heads out to the designated location for the field, unrolling the main power cable as it goes.

An open question is how far the power field should be from the ship. We may learn enough for planning purposes from the Starship hopper tests, but regardless, my sense is that the fields should initially be located close, to minimize autonomous, battery-sucking driving. Once additional ships arrive with crew, we have years to relocate them to a safe distance from launch exhaust.

So, the tractor unrolls the main cable to a distance of, say, 50m. That’s far enough to let equipment pass easily between the ship and power field, but close enough to minimize driving distance. It swings right, stops, and unloads the junction box using the robot arm. (This might require another unhook, drive around, rehook.) It then unloads the first panel. Viewed from the rear, the panel is set down on the left side of the trailer, away from the ship. If we use a tripod support, the left side of the panel has small feet to grip the soil. As the panel lifts away from the one underneath it, the support leg on the panel’s right side pops open under spring pressure. (It will probably require some type of protection at the end so the leg won’t scratch the underlying panel.) To the panel’s rear, it’s connected to the junction box via power cable. From the forward end of the panel, a power cable connects it to the rear of the next panel down.

From this description, the distance between panels will largely be driven by the slack needed in the series-connecting power cables. We’ll need creativity in routing the cables so they come off the trailer cleanly, and we may need the trailer’s left rear twistlock post to be removable so it doesn’t foul the power lines.

From here, it’s lather, rinse, repeat. With creativity, we might be able to carry two 12-packs of panels and two junction boxes on each load. Regardless, when the trailer is empty, it heads back to the ship to receive the next load. With this design, the only cable connections that need to be made are between each row’s junction box.

We briefly dove to this very detailed level for a reason—even this level, which took me a couple weeks iterating to work out, is actually just scratching the surface of our design. Just to identify a single detail we still need to work out, those power cables are going to need insulation that won’t deteriorate under high ultraviolet light and will stay flexible at -60C. Such materials are available, but they aren’t exactly off the shelf. http://www.appstate.edu/~clementsjs/polymerproperties/plastics_low_temp.pdf gives a short introduction.

I’ve used Mitsubishi solar panels in these calculations because their specs are readily available online. That’s certainly a possibility, but it seems a lot more likely that Musk will want to contract with Tesla, which now owns Solar City. “Likely,” but not certain. One obvious concern is that Tesla is a public company, and SpaceX may be looking for a deal. I could imagine a third party wanting the cachet of providing a Mars (or Moon) solar power field, and being willing to buy in so low that Musk could be worried about the reaction of his Tesla stockholders and the SEC if he countered. Ultimately, though, none of that matters. Someone will build the panels, junction boxes, cables, etc. They’ll need to operate at lower temperatures than Earth, and be suitable for autonomous deployment. But the specs we’ve seen pretty much tell us what the art of the possible looks like, so the mass and volume constraints should only get better. Where we’re at seems reasonable as a first cut.

Similarly, this assumes SpaceX buys Tesla battery technology, which again has a certain logic, but it’s certainly not the only option. For starters, liquid metal batteries scale extremely well, are low-cost, very safe, and the “waste” heat they generate is anything but waste on Mars. Check

for an easy introduction to this technology. For now, we’ll continue with PowerPacks only because we have readily available specs to use.

We do have one major, unexamined issue: how much power reserve do we really need in the event of a months-long dust storm? Short answer, “a lot more than days.” We’ll come back to this problem, but as a sneak peek, our answer will revolve around utilizing Nostromo’s immense power field and storage banks.