Gray Dragon on Falcon Heavy

I often bring up the idea that the role SLS & Orion perform in taking crew to lunar orbit could be performed by Falcon Heavy launching a modified lunar Dragon with a crew of four, and I'm often met with skepticism from certain quarters of spaceflight fandom.

Recently I've gotten frustrated that instead of framing their objections in the form of "how would that work with consideration X", many of the skeptics come out of the gate with phrasings along the lines of "that wouldn't work because X", seemingly not even considering the possibility that I and others have already accounted for X, Y, Z, and a number of other factors before bringing this up.

Tired of trying to convince people one at a time, I've decided to lay it out here. I won't hit all of the details I've considered, but hopefully enough to convince most people that this is actually feasible.

To be clear, I think of this as an alternate history story or an idea for a secondary or even tertiary backup, since I think that at this point, solutions for replacing SLS+Orion that involve Starship are far better, but it's an idea worth giving serious consideration.

Gray Who?

To start out, a bit of background on the "Gray Dragon" name. From what I've heard from Garrett Reisman (engineer, former astronaut, former Director of Space Operations for SpaceX), Gray Dragon was never a SpaceX project; the name comes from a 2020 opinion piece by Robert Zubrin and Homer Hickham. Reisman has said, though, that Dragon was designed with BLEO (beyond low Earth orbit) operations in mind, and this is also reflected in some of the phrasing on the SpaceX page for Dragon ("capable of carrying up to 7 passengers to and from Earth orbit, and beyond"), as well as SpaceX's short-lived ambitions for sending two private space tourists on a flight around the Moon on Dragon.

But Falcon Heavy is not crew rated! 

The Ars article also mentions that Falcon Heavy is not crew-rated, which is described as meaning the launch vehicle doesn't have some safety factors that NASA requires for human flight. More precisely, though, it means that it hasn't been verified by NASA as being safe enough for human flight. That crew certification would likely take a few flights, some time, and some money. But few if any changes to the vehicle are likely to be required, given how similar Falcon Heavy is to the crew-rated Falcon 9. Potential hurdles might be lower structural margins on the center core and risk factors associated with the higher engine count. However, considering that SpaceX achieved 1:276 predicted LOC (loss of crew) risk across the Demo-2 mission (ascent would have been a fraction of that, and that was Dragon's first crewed flight) *without* taking risk reduction from the launch abort system into account, and given their willingness to launch the most expensive of customer payloads on Falcon Heavy, there's a good chance that FH+Dragon could be shown to have low enough LOC risk without mitigation, through detailed analyses of vehicle failure modes, their likelihoods, and probabilities of successful abort in each scenario.

Do we have to fly crew on FH?

There are other interesting scenarios involving combining Falcon 9 and Falcon Heavy launches (crew on one, an Earth Departure Stage on the other). Gray Dragon wouldn't have been dead in the water if Falcon Heavy couldn't be certified or the mass budget couldn't close for some reason. In fact, F9+FH scenarios allow for not only reusable boosters, but also for more practical LLOR (low lunar orbit rendezvous) schemes. Like the kind that Apollo benefitted from, and Constellation would have been benefitted from, but SLS+Orion can't do. There are even ways to work out a lander from the Falcon and Dragon parts bins.

But all that would require more development and is outside the scope of this post, which is really about envisioning a close analogue to SLS+Orion and the capabilities they provide.

So, what would it take? 

In the wake of the Zubrin/Hickham op-ed, Eric Berger of Ars Technica asked Garrett Reisman what it would take for Dragon to perform a lunar mission. Berger's summary of Reisman's answer was:

"Traveling beyond low Earth orbit would therefore require some substantial but feasible changes to the spacecraft, Reismann said. Dragon’s [navigation] system works through GPS, so it would need a new communications and navigation system. In terms of radiation, he said, addressing this for astronauts is relatively straightforward, but hardening electronics would require some work. The heat shield could be made capable of returning from the Moon relatively easily, Reismann said. Additional consumables for a longer journey would take up interior volume."

(For a more recent perspective from Garrett Reisman on the future of human spaceflight in general, check out my interview of him on YouTube)

How much mass would that all add?

Zubrin cites 9.5t as the mass of Crew Dragon, but this is its mass after the trunk has been jettisoned and the de-orbit burn has been completed. The mass with the trunk and before the de-orbit burn is 12.5t. That 12.5t figure was for Demo-2 before departing the station -- not a completely full propellant load, since some propellant would have been used to dock with the station, but I figure that Gray Dragon wouldn't need to launch with a full propellant load (lunar orbit entry and exit will be performed by a different propulsion system). Demo-2 was also carrying only two, instead of four astronauts, but I vaguely recall hearing there were mass simulators for the two missing astronauts, so I won't worry about those extra ~0.2t. I figure 12.5t is a reasonable baseline on which to start adding extra pieces.

First, let's add an extra week of consumables (not as much as Orion would have, but enough for a trip to lunar orbit and back, with some modest margins given the starting point). If you pack the consumables efficiently, with no water recycling but reasonable water rationing, we're looking at 0.2-0.3t of for a crew of four for an added week (this would be about 7-10kg/day/astronaut for water, food, oxygen, and consumable CO2 scrubbers).

Next the communications and navigations systems and the radiation hardening of equipment. Given the sizes of small lunar and interplanetary probes these days, 0.1t is likely more than enough for all of those changes. For reference, Artemis 1 will be carrying along a fleet of 6U cubesats, each weighing ~1/7 of that, inclusive of science payload, propulsion, and power.

Then let's look at radiation protection for the crew. Background radiation on a short 7-10 day transit and landing mission is not enough to cause a serious health risk. A bigger risk is from the possibility of an intense solar storm. Artemis's primary strategy for dealing with this is building a storm shelter by putting stuff already in the capsule for other purposes (like seats) inside of bags and hiding behind them. The secondary strategy is with radiation vests. These strategies have negligible mass impact, but let's budget 0.1t to be safe.

We're now at ~13t, and need to beef up the thermal protection system. Per Reisman and SpaceX, the main shield is actually sufficient for high energy reentries. What needs to be uprated is the side TPS. Conservatively, adding 2cm of PICA-density ablatives to the sides would add ~0.2t to the vehicle mass. (PICA-X is used on the main shield, not the sides, and ~2cm of it wears out on LEO reentry; the windward facing nose gets far hotter than the sides on reentry).

What about getting to NRHO and back?

Now we have a 13.2t vehicle that can perform a lunar flyby, but doesn't have the delta v to get in and out of lunar orbit. For that, let's add a propulsion module integrated into the empty volume of the trunk. NASA baselines 450m/s for the delta v requirement of moving between TLI/TEI and NRHO, or 900m/s roundtrip. It's possible to do it with substantially less delta v given a slow transit (4-5 days each way), but I'll target 900m/s, noting that we still have extra performance margin from the propellant onboard the capsule.

The easiest solution is to use pressure-fed storable propellants with either an off-the-shelf engine like the AJ-10 (that Orion will use) or an engine SpaceX might develop in-house based on SuperDraco (would need a vacuum optimized nozzle, as well as a lower chamber pressure to minimize tankage & pressurant system mass). The AJ-10 on Delta-K achieved an isp of 320s (3143Ns/kg), though higher values are possible. With 320s isp, 1.5% residuals, and a conservative 15% structural coefficient (very close to the structural coefficient of the Delta K of the 80s, includes the mass of the engine), we're looking at a propulsion module that is 5.6t wet with close to 4.7t useable propellant, bringing us to a total mass of 18.8t.

Would that propulsion module fit in the trunk?

The trunk certainly has more than enough space, with a volume of 37 cubic meters. And it's long enough to fit an engine like AJ-10. But trying to fit tankage and an engine in the typical tandem configuration is awkward. One potential solution is to use a number of smaller, shorter engines. Another is to put a number of tanks (likely four), in pairs of opposing tanks (i.e. fuel tanks at 0 and 180 degrees, oxidizer at 90 and 270) around one central engine, with pressurant tanks fitted in the remaining spaces. Toroidal tanks are also a possibility, but arguably more complicated.

In short, yes, there are ways to do it. 

Isn't that too heavy?

We have an 18.8t payload (give or take). Can Falcon Heavy send that towards the Moon?

The answer is yes, but marginally. Contrary to NASA figures of ~15t to C3=0 (in the fully expended configuration), SpaceX on its website claims 16.8t to TMI and 26.7t to GTO for Falcon Heavy. This tracks with independent estimates of the upper stage's burnout mass as being near 4.5t, for GTO at LEO+2.4-2.5km/s and TMI at LEO+3.6-3.8km/s, suggesting 20t is achievable for LEO+3.2km/s, which is enough for TLI (which can be done with as little as LEO+3.08km/s).

However, this requires leaving very little in the way of residuals in the propellant tanks and probably leaves no margin for the likes of engine out or underperformance scenarios. The NASA figures probably have healthy margins baked in. These margins may seem to be necessary for human spaceflight, but actually with abort capability at any point during launch and TLI, the risk is mainly to the mission, not to the crew. Also, we're leaving relatively tight but not insignificant margins here, with 20+t capability vs. 18.8t payload mass budget; mass growth is typical, but here we have a lot of known quantities as well as conservative values for some components and unexplored opportunities for mass reduction*.

One other factor to bear in mind with regards to FH performance is that maximizing payload might require a trajectory that's not suitable for crewed flight. Too lofted of a trajectory could increase g loads to unacceptable limits during abort, but given the high thrust of the Falcon second stage. Too shallow of a trajectory might also have some disadvantages I'm not clear on (possibly reducing separation from LV components during abort? -- that one isn't likely to be a deal-breaker). Maximum payload would also require not throttling down to reduce g loading on ascent, but this would mean briefly peaking at around 5g before the center core shuts down and around 4.2g before the second stage shuts down -- generally acceptable levels for human spaceflight. It might be worth bringing the peak down to 4.5g or even 4g -- this wouldn't dramatically impact payload performance.

*One important opportunity to consider for mass reduction is the propellant load on Dragon. For the pad abort, Dragon was apparently loaded with ~1.4t (3060 lbs) of propellant (archived EA FONSI -- page 8), but for the in flight abort (page 20; credit to Jeff Vader for showing me this) was loaded with ~2.6t (5650 lbs) of which ~1.1t (2400 lbs) would remain after abort. The same document also says the vehicle is ~7.7t (17,000 lbs) dry -- it seems probable this just refers to the capsule, and not the trunk. This would suggest that after the de-orbit burn and trunk separation, Demo-2 had ~1.9t (4200 lbs) worth of astronauts, (mass simulators/cargo/consumables?), and propellant & pressurant (for RCS (reaction control system), reserve, and residuals), which would make sense. As I understand, RCS and LAS on Dragon use the same propellant tanks. So there might be on the order of 1t of propellant load reduction available for Gray Dragon, since the LAS (launch abort system) would only really need to be used during launch to orbit (Orion jettisons its LAS well before orbit and relies on the AJ-10 for in-space abort modes); for nominal launches, once in orbit, abort propellant could be used for RCS. This would mean around 11.5t baseline and around 17.4t wet mass for Gray Dragon (give or take several hundred kgs).

What about power and thermal management?

It's not an accident that Reisman didn't mention changes to those systems. The LEO environment is actually a more challenging one from both a power generation and thermal management perspective. The transit between Earth and the Moon is comparable to the daylight half of a low Earth orbit, except that the spacecraft doesn't need to reject heat from the major IR and reflected sunlight source that is the Earth and doesn't need to charge batteries to use in the night half of the orbit.

What about cargo?

I often run into the argument that Orion will be used to build Gateway and can take other cargo to NRHO. This is true. However, SpaceX is also contracted to build Gateway, taking the first two modules on a dedicated Falcon Heavy launch, and also contracted to take cargo to Gateway with their Dragon XL proposal.

It's true that modules other than HALO & PPE (Falcon Heavy's manifest) don't have their own propulsion systems and need something like Orion to brake them into NRHO and dock them to Gateway. However, "like Orion" doesn't need to be all that similar to Orion. It could be Dragon XL or a tug based on it; it could be a tug developed by a third party (a number of companies are currently working on orbital tugs for the private sector); it could even be HLS Starship or a different Starship variant, both of which have plenty of spare capacity. There are only a few non-HALO/PPE modules to launch, and there's no rush. They are not expected to be launched before Artemis IV.

How long would Gray Dragon take to develop?

Most of the physical changes to Dragon aren't that much more complicated than the cupola that SpaceX designed, built, validated, and flew in a matter of months for Inspiration 4. Long distance navigation and communications could be bought off the shelf and integrated. Side TPS augmentation and extra consumables could likely be added at least as easily as a cupola. Radiation hardening the electronics and building a propulsion module would be the exceptions, but even those could be surprisingly quick, given that there are off-the-shelf solutions for rad-hardened electronics, and small pressure-fed hypergolic engines small enough to 3D print are relatively easy, and SpaceX already has plenty of experience with them.

It would of course take time to work with NASA to certify the craft and the launch vehicle for Artemis. All in all, I would expect it to take 3-5 years from work-start to crewed flight, building on the foundation of a functional Falcon Heavy and a functional (or near functional, thinking alt-history) Dragon 2. If work started now, it could be ready to fly crew in the 2025 to 2027 timeframe.

But the better question is how long would it have taken? If work had started in mid- or late 2018, some months after Falcon Heavy and some months before Dragon 2 first flew, even if it had taken an extra year to develop (4-6 to crewed flight), Gray Dragon may have already flown an uncrewed demo by now (Artemis I, which we are still waiting for as of March '22) and would likely be flying crew (Artemis II) sometime between this year and the end of 2024. 2024, by the way, is the  current no-earlier-than timeframe for Artemis II.

How much would it have cost?

According to Elon Musk, SpaceX developed Falcon Heavy out of pocket for somewhat over half a billion dollars, building on Falcon 9. NASA paid SpaceX $3.1B to develop Dragon 2 (which ended up being an all-new craft compared to Dragon 1) and provide 8 flights, including 7 crewed flights, 6 of which were considered operational. NASA recently contracted SpaceX for an additional 3 flights for $900M. Given those data points and the extent of the changes required for Gray Dragon, I'd expect that a development contract that included one uncrewed (Artemis I) and one crewed (Artemis II) demonstration flight would likely cost NASA in the range of $1-3B.

By comparison, SLS alone has cost $23B to date to develop, and Orion had already cost NASA $23.7B in 2020 dollars over a year ago. But that's money already spent. We have to look at the opportunity cost. We spend around $4.5B *every year* on SLS, Orion, and EGS (exploration ground systems in support of SLS+Orion), meaning I would expect Artemis I and Artemis II to have been completed not only sooner with Gray Dragon but for a total cost to the taxpayer that would be equivalent to 3-9 months of spending on SLS/Orion & associated systems -- a period of time that is comparable to the total delay that has been incurred in Artemis I's projected launch date since the SLS green run a year ago.

I'll let that sink in for a bit.

But wait, there's more

Now, putting aside cost and schedule, let's consider exploration opportunities. Right now, SLS is the bottleneck for lunar mission cadence. We're targeting approximately one a year over the next decade. By contrast, FH+Gray Dragon could easily double or quadruple that tempo within 1-3 years of its first flight, without SpaceX breaking a sweat or NASA paying more than we currently pay for SLS alone.

Q&A

How would the extra mass affect abort? -- asked by Jeff V

The propulsion module would be jettisoned, so the aborted mass wouldn't really be affected. In addition to a mechanism for jettisoning the propulsion module and ensuring the trunk slides past it cleanly, this might require a mechanism connecting the propulsion module to the upper stage (released on stage separation), so that the upper stage and the propulsion module don't collide as Dragon is pulling away. This would also keep the structural loads on the trunk in check during ascent.

Could you use a stretched Falcon upper stage instead of a propulsion module for LOI and TEI? -- asked by Nicholas B

Yes, barely (or almost), but there are a few drawbacks.

Going with SpaceX's 63.8t payload claim to LEO and a (very dry) 4.5t burnout mass estimate, increasing the wet mass of the upper stage by ~50t would let you get a gross mass (including the mass of the upper stage) of ~20.5 through TLI, LOI, and TEI, if you could manage zero boiloff.

That increase would add 1.5-2t to the minimal-residual burnout mass of the upper stage, and you'd likely need to budget another 0.5-1t of insulation, making for a ~7t upper at a minimum, leaving 13.5t for Dragon. (We had budgeted 13.2t, or 12.2t with a low prop load).

Leaving even just half a percent more residual propellant in the tanks would cost ~0.8t and make it infeasible. Since we're relying on the upper stage to get back, counting on running it bone dry is not really an option.

The other problem is that we'd be relying on a single pump-fed engine to get back. This would likely be considered too risky.

One way to get around these limitations would be to use Dragon's onboard propellant to do the TEI burn, and use the stretched upper stage just for LOI. The Dracos' thrust is too low, and the SuperDracos have low isp and point off-axis. Factoring in cosine losses, using the SuperDracos, you'd need ~2.5t for the 450s budgeted for the TEI burn. This is basically the full propellant load and would leave almost nothing for course correction and RCS use during reentry.

This could be solved by increasing the size of the propellant tanks or using a new set of smaller engines with vacuum-optimized nozzles pointing axially in the direction that is "up" when on the pad. With a conservative 300s isp, TEI could be managed with a more reasonable ~1.9t of propellant, or ~1.8t with 320s isp.

How much would it help with mass margins to add a hydrolox 3rd stage? -- asked by SpaceHaxx

If you sized so that the combined mass of hydrolox stage and the payload roughly matched FHE's LEO payload, and you used the 465s RL10B-2, with a burnout mass of around 4t for that stage, you could get around 27t to TLI.

If you insulated it well enough for the coast, you could also use it for the LOI burn and use the Dragon modifications mentioned in the previous answer to do the TEI burn, without a separate propulsion module. This would in theory enable over 10t of cargo (or an ISS module) to be carried along (though the aerodynamics of a bulky hydrolox 3rd stage + cargo + Dragon on top of the already tall and skinny Falcon Heavy may or may not be tractable).

Bonus Alt History

If you enjoyed that foray into FH+Dragon saying "anything you can do..." to SLS+Orion, consider an even more surprising possibility: what if we had married the best aspects of Apollo and Soyuz, modernized the hardware and trimmed some of the fat, and put together a lightweight 2-crew-to-the-surface solution?

Could we realistically have launched such a contraption (lander included) on a single SLS flight? Could we have launched it on Falcon Heavy? Find out