To The Red Planet with the Kuiper Launch Coalition and Hypergols
Ten days ago Amazon announced they had made a deal to purchase a huge amount of launch capacity to launch their Kuiper constellation. Not including the 9 previously bought Atlas V launches, we're looking at roughly 2000 tons of VLEO launch capacity with 38 Vulcan-Centaur 6, 18 Ariane 64, and 12 New Glenn launches (with an option for 15 more).
I recently claimed that this would be enough to launch multiple crewed Mars landing missions using storable hypergolic propellants, and I was asked to show my work. So here it is:
The Goal
I will choose a relatively small but generally acceptable crew size of 4 and try to fit three long-stay conjunction class missions within a ~2 kiloton upmass budget.
The first mission would be an uncrewed demo in 2033 that would include landing components of a Mars base. It would be followed by two crewed missions, in 2035 and 2037, that would also bring more cargo.
The crew would travel to Mars in a modular Mars Transit Vehicle (MTV) composed of a small Earth return vehicle (ERV), and a habitation element with independent power and RCS. The MTV would powerbrake into a highly elliptical Mars orbit (HEMO).
At around the same time a Mars descent/ascent vehicle (MDAV, not to be confused with Lockheed Martin's reusable MADV proposal) would aerobrake into HEMO. The MDAV would be composed of a lander (aeroshell + retropulsive final landing), a hypergolic ascent stage, and a detachable crew cabin.
The MTV and MDAV would rendezvous, and all 4 crew would transfer to the MDAV to descend to the surface.
Separately, an ascent transfer element would aerobrake into low Mars orbit (LMO), and cargo landers would land base components and supplies.
At the end of the surface mission, the Mars ascent vehicle (MAV) (ascent stage of the MDAV) would take crew to LMO and rendezvous with the ascent transfer element, which would take the MAV's crew cabin to HEMO to rendezvous with the MTV. After a crew transfer, the MTV would return to Earth.
The crew capsule of the ERV would reenter and land with crew, while the transit hab and the ERV's service module would be safely disposed of in the atmosphere.
Each synod, modular propulsion and payload components would be launched to LEO, and then orbit raised in batches to highly elliptical Earth orbit (HEEO) using some of the hypergolic propulsion elements to minimize orbit keeping requirements while waiting for the Mars transit window. The components would be assembled into several Mars-bound vehicles before doing the final TMI burn.
Propulsion Elements
Since the context is the launch capacity of a contract fulfilled with A64, VC6, and NG, it would make sense to have standardized propulsion elements that are in the neighborhood of 20 tons each, with A64 and VC6 launching them one at a time to the highest orbits they can, and NG potentially launching two at a time.
I specified that this mission would use all hypergolic propellants (after launching to Earth orbit). Let's get more specific.
Hypergolic propellants are those that spontaneously ignite when their components come together. This enables simpler and more reliable ignition with virtually unlimited restarts.
In practice, hypergolic propellants also tend to be storable in liquid form near room temperature. This makes them easier to keep liquid in space, as well (in LEO, LMO, and interplanetary space), and enables staging components in Earth orbit (and leaving components in LMO & HEMO) over a long period of time without worrying about boil off.
They are also generally associated with pressure-fed engines like the AJ-10, which are simpler and typically more reliable than pump-fed ones, and bladder tanks, which enable startup in 0g without ullage motors.
Here, I want to lean on the storable nature of hypergolics and take advantage of their simple and reliable multi-start capabilities. For the sake of this performance I will baseline pump-fed engines, which provide higher isp and allow much lower structural coefficients, leading to dramatically lower LEO mass for this kind of mission.
Some prior examples of pump-fed hypergolic engines are the S5.92 flown on Fregat and the ground-tested Aestus 2 / RS-72, which was apparently capable of ~340s isp with a twr of ~40:1.(Another source (pg. 6) shows slightly lower isp of a little under 339s isp and twr of ~35:1 for the RS-72, while showing >340s is feasible for similar engines). Both of these engines are gas generators and are relatively simple as far as pump fed engines go.
Similarly I'll eschew bladder tanks for performance reasons and rely on RCS (needed for rendezvous & docking, anyway) for ullage burns.
As far as structural coefficient (dry mass to wet mass ratio) goes, the F9 second stage seems to have a structural coefficient well under 5%, while the single engine Centaur III has a structural coefficient of ~9.7%. Our target mass of ~20t is comparable to a Centaur III, while F9S2 is much more massive. Larger stages do tend to have lower structural coefficients, but a bigger reason for the Centaur III's much higher structural coefficient is that hydrolox is not nearly as dense as kerolox. Hypergolics like NTO/MMH are denser than both, and unlike hydrolox don't need any intertank insulation.
While I think that a much lower structural coefficient is possible, I will baseline with a conservative 7%. Compared to the 9.7% of SEC (single engine Centaur), for roughly the same gross mass, we're looking at ~1/4 of the tank volume & commensurately smaller pressurant system, and unlike the Centaur III, these propulsion elements don't have to bear the structural loads of a payload during launch.
For all propulsion elements, I'm also factoring in 1.5% propellant residuals. Lower values with better performance are certainly possible.
And I will use an isp of 340s.
Landing Elements
For the landing elements, it might make sense to have multiple sizes to fine tune the amount of downmass, or at least larger elements to
b) better take advantage of NG's 7m diameter, and
All landing elements would be pre-packed cargo in aeroshells with TPS and retrothrusters for final landing (whether sky crane or traditional thrusters on the bottom), with the exception of the MDAV, where crew would enter into a crew cabin after MOI (Mars orbit insertion).
The landers could be made to fit inside the New Glenn's fairing or fly without a fairing, with a base either flush with NG's 7m diameter, or extending out a bit more to 8-10m.
For all of the landers, I'm using what I believe is an achievable (pg. 55-67) payload mass fraction of 45% for a lowland mission (the remaining 55% being reserved for the TPS, aeroshell, and retropulsive landing system).
While the reference vehicle in the above thesis paper has a higher retrothruster isp of 379s, it also uses a descent propulsion stage structural mass fraction (pg. 16) of .65 (structural coefficient of 39%), which I think is higher than it needs to be (high twr solids with low isp could provide similar or better performance), as well as an aeroshell mass fraction of .68, which also leaves a lot of room for improvement.
For the cabin of the MDAV, which is a detachable component that will be moved from LMO to HEMO with a transfer stage, I've budgeted 2.5t (including several hundred kg for crew and consumables). Compare this to the 1.3t (pg.3) of the Soyuz orbital module, which includes 2 hatches and can provide life support for a crew of three for 10 days, or the 2.15t dry mass of the Apollo Lunar Module descent stage, which not only provided lunar habitation for two astronauts, but included the dry mass of propulsion elements (engine, prop tanks & pressurant system, RCS).
If 2.5t sounds too low, consider that a 2.8m diameter spherical shell would have a pressurized volume of over 11m^3 (more than Dragon) and with 20kg/m^2 of wall area density (enough for a double wall with 3mm of aluminum in each layer plus some padding) would only mass around half a ton, leaving ~2t (more than enough) for hatches, astros, suits, consumables, power, comms, ECLSS, and miscellaneous other items.
For the sake of simplicity, I'm treating the MDAV as a lander whose payload is an ascent stage with a 2.5t payload, 4.2km/s delta v capability w/ 7% structural coefficient, 1.5% propellant reserve, and 340s isp. For the same ascent vehicle mass & performance, higher structural coefficients (up to ~11% for each stage) are allowable with a 2 stage ascent vehicle.
This yields a 11.5t gross MAV, and a 25.5t MDAV (including entry/descent/landing elements).
Transfer Element
I'm baselining LMO as 300km x 300km orbit, and HEMO as 300km x 100,000km -- this elliptical orbit has a period of 8 days and 8 hours, short enough that it could be possible to have more than one launch & rendezvous window on the return.
It takes ~1.33km/s to move from that LMO to that HEMO. The maximum delta v required to align the longitude of the ascending node and the true anomaly with those of the MTV's orbit is ~0.34m/s. Accounting for orbit-keeping in LMO, rendezvous, and reserves, a 1.9km/s delta v budget should be adequate for the transfer element.
MTV
The Mars Transit Vehicle would be composed of a habitation module and an ERV (Earth Return Vehicle), with a crew capsule and service module.
On ascent, the ERV would rendezvous with the hab in either LEO or HEEO (highly elliptical Earth orbit). On return, the crew could close the hatch just hours before reentry.
I would not choose to use Orion as an ERV, since it's overkill for getting a crew of 4 to and from near Earth space. We could use a Dragon or Starliner modified with beefed up TPS and electronics capable of safely transiting the Van Allen Belts and deep space, but it would be more mass efficient to go with an even lighter vehicle without optional seating for 7 or integrated abort motors (abort tower instead).
Dragon is 12.5t with the trunk, ~2.6t of propellant, and integrated abort motors, and modifying it for BLEO operations wouldn't add very much mass (not counting propulsion that's budgeted separately).
I've budgeted 10t for an optimized ERV, but it would be possible to go a lot lower. The 3-person Apollo Command Module was ~5.6t, including ~1.5t of electronics, and it was mainly just missing power components relative to a full-fledged ERV w/ minimal propulsion.
For sizing the transit hab, I'll reference the Salyut 7 space station, which at one point supported a 3-person expedition (EO-3) for 237 days in its 90m^3 of habitable volume + 5m^3 in the Soyuz orbital module (we can count the ERV as being comparable to the Soyuz reentry module). For a crew of 4, this scales to ~125m^3 of habitable volume. Accounting for consumables and ECLSS, 300m^3 of pressurized volume would be more than enough.
A 10m long, 6.2m diameter cylinder would provide about that much volume and would fit neatly inside a New Glenn fairing (pg 59). At a generous 30kg/m^2, the main structure of such a hab module would mass at ~7.7t. I'm budgeting 20t for the hab (not counting consumables), which leaves over 12t for ECLSS, power, radiation shelter, and other equipment.
To reduce radiation exposure, the crew could sleep in a radiation shelter, the primary structure could be made largely of plastics, and consumables and propellant could be arranged to provide additional shielding.
Even without decent shielding, the health risks are borderline. We don't know enough to say how big the risk is with any precision, but on the high end of estimates, it would roughly double an astronaut lifetime risk of dying from cancer, while the most likely figure is almost an order of magnitude lower.
And unfortunately, dramatically reducing GCR exposure (by say well over a factor of 2) using shielding requires an impractically large amount of material.
It seems like an acceptable compromise on the admittedly very fuzzy cost/benefit curve to not provide heavy shielding for most of the MTV, but to cut the one-way transit time to 6-7 months, have the crew spend 1/3 of their time in a rad shelter, and use a baseline of ~3g/cm^2 of plastic plus a bunch of consumables.
Reducing time in space by maximizing time on the surface (contrary to short-stay opposition class missions NASA is studying) is also very beneficial for reducing risk. The planet's shadow cuts GCR by around half, and the atmosphere provides further dramatic reductions, with on the order of 20-30g/cm^2 of atmospheric mass in the lowlands (and substantially more effective shielding for radiation coming in at angles that aren't normal to the surface).
It's also easier to provide additional shielding on the surface, since the bulk of radiation will come from a much smaller range of directions, and there's regolith that can be utilized.
Storing most waste products around the hull and/or the radiation shelter in between propulsive maneuvers can help retain some of the effectiveness of consumable-based radiation shielding as consumables are used up, without impacting the mass budget.
Consumables
The primary consumables are water, air (oxygen & CO2 scrubbers), and food, in descending order of mass and ease of recycling. If we weren't using any recycling, we would want to budget 4-5kg of water per person per day (for drinking, hydrating meals, and sanitary uses), a bit under 1kg of oxygen, a bit over 1kg of LiOH for scrubbing CO2, and a bit under 2kg of food. Roughly 8-9kg per astronaut per day.
We can pretty easily reclaim most of the water and it's actually possible to end up with a water surplus when factoring in water present in not-fully-dehydrated food (fully dehydrated food would be <1kg/day not including packaging, based on caloric density & fiber content) and water produced when food is metabolized. ~5kg/day is achievable with decent water recycling. And since water is 89% oxygen, it's also a great way to store oxygen.
At this scale, there are ground-based electrolyzers that mass 750kg (presumably, space-optimized electrolyzers could be made lighter) that can generate 2.15kg/day of hydrogen (split ~19.3kg/day of water -- sufficient for waste water recycling and O2 generation for 4 astronauts) using 6.3kW of power. With fuel cells reforming ~3/4 of the H2 & O2 back into water, it'd be more like 5kW. With 25% PV efficiency, that needs a bit over 30m^2 of solar panels at 1.5AU, about the solar panel area of a single Starlink satellite. So, I'd expect this all to fit comfortably within the mass budgets of the MTV hab and the surface hab. (The surface hab would of course need substantially more solar generation capacity and batteries to support water recycling, if it's not using nuclear power, since it won't have 24/7 access to sunlight.)
Using reusable CO2 scrubbers, which NASA is already planning on for Mars missions, it should be possible to save another ~2kg/astro/day (from saving on consumable scrubbers and extracting O2 from CO2), enabling a total consumption rate of ~3kg/astro/day.
With better water recycling, ~2kg/astro/day should be feasible. Even without synthesizing or growing food, with thoroughly dehydrated and lightly packaged food, under 1kg/astro/day is theoretically possible (though infeasible in the near term), as ~400g/day of carbs & proteins + ~50g/day of fat will provide ~2000kcal, while other non-water components of food seem to account for <100g/day in a typical diet based on what comes out in feces & urine.
I will go with 5kg/astro/day for a middle of the road figure that leaves some margin for reserves.
Dates, durations, and delta v
To plan more precisely for these hypothetical missions, I used the University of Madrid Space Dynamics Group's porkchop plotting app.
Based on the porkchop plots, I picked out the following departure dates & travel times with associated surface stay durations:
The MOI burn is done at a periares of around 100km, and includes a periares raise to 300km at apoares. Similarly, the TEI burn starts with lowering the periares back down to 100km. A low TWR could slightly increase these delta v requirements.
As an example, in 2035 we have an inbound travel time of 190 days, meaning we need 3.8t of consumables for the return trip, so we have ~33.8t of payload during the MOI and TEI burns of the MTV. Using the simplified MOI and TEI payload fractions, we get a mass to TMI of ~71.3t for the MTV with its propulsion modules.
Using standard 20t propulsion modules, we get a mass to TMI of ~71.6t to cover the same delta v of 2237m/s. It ends up needing two propulsion modules, one underfueled by ~2t (presumably having spent those 2t to complete the TMI burn). We're only off by 0.3t in TMI mass. For 2033, we're off by 0.7t. On the order of 1%.
On the Earth side, it's pretty conservative. Chaining propulsion modules can increase TMI payload by over 10%. Oberth losses are generally greater with low twr (a longer chain of propulsion modules will have lower twr), but by doing multiple short apogee-raising burns near perigee, they can be brought down to negligible levels. And since we have a lot of vehicle diameter to work with, we can actually make these propulsion modules very squat. Combined with their low thrust, it shouldn't be hard to make a long chain of them stiff enough without adding much mass.
It makes sense to frontload the cargo deliveries so that the first crewed mission has access to as much of the equipment as possible. This also aligns well with the lower TMI delta v requirements for the '33 and '35 launch windows.
This means that the first crew would have 60t worth of equipment (habs, rovers, power sources, scientific equipment, emergency supplies) to work with, in addition to standard duration consumables. In the event that the crew is not able to leave the surface, they would need to stay an extra synod and would be resupplied. The amount of consumables needed to tide them over until the supply ship arrives is on the order of 5t and can comfortably fit within that 60t budget as emergency supplies.
The second crew would have an additional 10t budget for new equipment and replacements.
Here's what the landed payload and gross lander masses would look like (in tons):
Both VC6 and Ariane 64 are also anticipating substantial upgrades in LEO performance, with Centaur V-LEO
There are a few contingencies to examine.
1. What happens if the crew is unable to descend to the surface or if they have to leave early for some reason? They'll be stuck in Mars orbit waiting for their return window with virtually no remaining consumables.
One way to solve this is to bring an additional full surface-stay's worth of consumables to HEMO and dump the excess before the TEI burn.
2. What happens if the crew is unable to rendezvous with the transfer element / the transfer element fails / the transfer element fails to launch or aerobrake into Martian orbit?
One easy mitigation is to send an extra transfer element on the '33 mission as a backup for '35 and '37.
Factoring in both of these changes, we get the following masses:
No Transfer Element
Another option for reducing the risk associated with the transfer element is to get rid of it altogether and launch directly to HEMO. For this, I'll budget 5.6km/s (LMO w/ margin + 1.4km/s) and use a 2 stage ascent vehicle, with an 8% structural coefficient for each stage. This yields a 18t MAV translating to a 40t MDAV, compared to 25.5t + 3.9t for the MDAV + TE approach.
Here's what it would come out to (with the extra consumables from before):
I honestly don't think this mission profile makes sense in a world with Starship in development.
But I do think it's more practical than opposition class missions, as well as long stay missions with a similarly limited scope and small crew that still rely on novel in-space propulsion technologies (like nuclear thermal propulsion) to reduce mass to orbit requirements, as if space launch hasn't gotten much cheaper than when those solutions were first proposed.
I would expect the total launch cost for these 3 missions to be on the order of $10B using existing non-SpaceX commercial launchers -- about the equivalent of 4 years worth of SLS spending, and the propulsion modules to add on the order of another $5B (plus or minus a few billion), including development.
It might raise a few eyebrows to talk about putting using much toxic hypergolic propellant, but consider that the total hypergolic propellant for all three missions would be comparable to the amount used in three Proton launches (a bit less, actually), or about one per mission. And it's arguably safer to take it to orbit in relatively small batches than trying to light it up on the ground.
Now, one might ask, why not save some on launch by using cryogenics or even nuclear propulsion? Well, we'd be looking at maybe $1-$2B in launch savings per synod (or about half that per year), but how much would we be spending on cryogenic propulsion modules that can last for months (or even up to 2 years) in Earth orbit? How much would we spend on NTP development? And how much risk would a more technically complex solution come with? I'll leave it to the reader to try to answer those questions.
Starship promises extremely large interplanetary transport capacities at low unit cost. It's technically very ambitious, but by rolling multiple functions into a single spacecraft developed by a famously cost-efficient company, it even has a good shot at coming in at lower development cost than more technically conservative solutions.
But if it fails, we still have paths to affordable small scale missions using commercial heavy-lift capabilities.
Comments
Post a Comment