Ok, picking up where the first part left off, we’ve reviewed the background, methodology, and some general observations on the methodology from our RAAN-agnostic 3-burn departure paper. In this part, I want to go over an interplanetary launch campaign concept that demonstrates how such a technique could be used to enable exciting interplanetary exploration missions that would be hard to perform otherwise. To do so, I’ll introduce the Interplanetary Blitz campaign, describe some of the key elements of the campaign, and then go into the results we found. Finally I’ll wrap up with some lessons learned, and areas for future research.
2024 Interplanetary Blitz
In order to illustrate the power of this RAAN-agnostic 3-burn departure methodology, we developed an interplanetary mission campaign that showed how a single ISS-coorbital depot could enable a series of interplanetary “smallsat” missions to four planets (Mercury, Venus, Mars, Jupiter), the Moon, and four NEOs (2007 XB23, 2008 EV5, 2001 QJ142, and 2009 HC) over an approximately five month period between late August 2024 and late January 2025. The main point of this exercise was to show that a single depot in a convenient location could support a rapid-fire series of interplanetary missions, without having any time to phase its orbit to align for each mission, and without excessive penalties for doing so.
A secondary but very important point of this exercise was to illustrate how a LEO micro-depot could enable dedicated smallsat launch vehicles, which currently can barely deliver payloads to LEO, to send those payloads practically anywhere in the solar system, at price-points that rival even semi-reusable larger launch vehicles.
Key Elements of the Interplanetary Blitz
This interplanetary blitz involves four key elements:
- The LEO propellant micro-depot
- One or more dedicated smallsat launchers
- A long-duration storable bipropellant kick stage
- A long-duration storable bipropellant lunar lander stage
We’ll take them in order. First, for this scenario, we used a LEO micro-depot concept that I’ve been noodling on for some proposals over the past few years. The basic concept is a low-cost, single-launch depot that can be used primarily for refueling dedicated smallsat launcher upper stages and storable bipropellant kick states. This depot is actually capable enough that it could support much larger missions, and ones including LOX/LH2 stages, but its primary near-term mission would be allowing dedicated smallsat launch stages to access much higher energy destinations, with much larger payloads, than they otherwise could.
Artist’s Conception of a LEO Propellant Depot[note]I’m including this picture again, both because it’s awesome, and because I can[/note] (Credit: Brian Versteeg)
We didn’t really go into the technical concept for this depot in the paper, but because many of you may be curious, here’s the thinking behind some key features:
- The micro-depot is comprised of a depot kit (the conical section and everything to the left) attached to a repurposed Centaur V upper stage. This kit is designed to be able to ride as a secondary payload on an ISS Cygnus mission with approximately 10-12 tonnes of initial propellant on-board.
- The LOX and LH2 tanks of the Centaur V are repurposed after arriving on orbit–the LOX tank is reused as a primary depot LOX tank, allowing for the storage of over 50 mT of LOX onboard. The LH2 would probably be used to chill the LOX, helium, and other propellants down as far as possible, to extend how long the depot could last with minimal boiloff. In theory, the depot could be used for LOX/LH2 missions, but only if the LH2 was used quickly–in this configuration, and in LEO, it’s not designed to be able to store LH2 without boiloff for super long periods of time.
- I show a conical sunshield, similar to previous ones studied by ULA, though I’m not positive at 400km if that trades better than just a bunch of layers of MLI. If you’re at too low of an altitude, the earth starts peeking into the conical sunshield, dramatically lowering its effectiveness. I haven’t created a spreadsheet yet that allows you to figure out what altitude you have to be around a given body with a given sunshield half-angle to make sure it works.
- The main body of the depot kit is the kerosene tank, which is inside the cylindrical part of the depot kit. In this particular design concept, I’m assuming that the kerosene tank is derived from a Centaur III LOX tank, and that the structures surrounding it are derived from Centaur III and V forward bulkhead elements. Reusing structural elements like that can sometimes enable lower development costs while still providing a reasonably efficient structure. The total capacity of LOX/Kerosene in this depot, assuming 50% boiloff margin on the LOX was somewhere around 40mT, which is enough to support dozens of interplanetary smallsat missions.
- Hidden inside the conical section are the helium tanks and the plumbing that control the main Centaur V tanks after it has been repurposed as a depot element. The helium tanks would probably be kept right up against the Centaur V hydrogen tank, with insulation isolating them from the warmer parts of the depot, so they can be stored at cryogenic temperatures where the helium is higher density. It may be possible if desired to actually store the helium at low pressure inside the Centaur V hydrogen tank, if all of the LH2 is vented previously, but that depends on the depot CONOPS and how often it receives LH2 deliveries, and how often it has customers that need LH2.
- The tanks around the outside of the Kerosene tank on the cylindrical section can be used for storable propellants for the kick stages and landers. This could be your traditional storable bipropellants (hydrazine/NTO), or any of the green variants being developed by companies such as Bradford, DSI, and Tesseract. I can’t remember the exact dimensions on these tanks, but there are eight of them, and I think they’re approximately 2-3 cubic meters each, so this may actually be able to support several missions worth of kick stage propellant.
- The solar arrays are oversized for just powering the main depot functions (robot arms, rendezvous/prox-ops sensors, spacecraft control avionics, so it may be possible to have power available either for running a small cryo-cooler, or for things like converting water electrolytically back to hydrogen and oxygen, or into hydrogen peroxide and hydrogen.
- There are shown six “Sticky Boom” style capture arms and two refueling arms. The Sticky Boom arms are scaled-up versions of the ones Altius is developing for its Bulldog satellite servicing vehicles, and they use our magnetic grappling technology for grappling the target stages. These capture arms have reaches in excess of 10m, which can allow a relatively non-agile upper stage to rendezvous close enough to the station to allow the arms to capture the target and damp out relative rates. A trio of arms is used to allow for a parallel robotic structure, which is much stiffer than a series connection. The fueling arm is a more traditional 6DOF arm that can also support depot maintenance. Two sets of capture/fueling arms are provided to enable handling fueling operations with two stages simultaneously.
- One last detail is that the reason I selected an ISS coorbital location was to take advantage of crew/cargo traffic to the station. Most ISS crew/cargo vehicles launch on vehicles with a lot of excess mass capacity that ends up going to waste. I’m not sure exactly how much wasted propellant there is on say a typical Dragon flight, or on Cygnus flight where you add a few extra solids, but in theory the numbers I’ve run suggest you might be able to get a better deal on this leftover prop than you could by buying a dedicated F9R or FH reusable launch, while still being economically interesting to the ISS crew/cargo launch companies. It is true that NASA isn’t intending to keep the ISS operating indefinitely, but I wouldn’t be surprised if commercial ISS replacements initially start in a similar orbit, which could create a similar dynamic.
Ok, that’s a lot more details on the LEO micro-depot, but I wanted to share some of my thinking, since we didn’t get a chance to go into it much in the paper.
Dedicated Smallsat Launcher Upper Stage
The next element is the smallsat launcher upper stages. For this paper, we focused on Virgin Orbit’s LauncherOne, as it is in the middle of the size range for the more credible smallsat launcher capabilities. But there’s no particular reason you couldn’t use a Rocket Lab Electron, or a Firefly Alpha, etc. It should be noted that while these stages are not currently designed for rendezvous operations, we think there are a few credible paths forward that can require minimum modifications to the stages themselves. The Rendezvous/Prox-Ops (RPO) sensors would be on the depot itself, with some added avionics to receive commands from the depot and translate them into maneuvers that could be performed either by the kick stage, by upgraded RCS on the stages themselves, or even by using the engine purges as a sort of ghetto cold-gas thruster to augment their 3dof steering RCS thrusters. Additionally the stage would need some DogTag grappling interfaces, and fill/drain ports designed for in-space refueling.
Since LauncherOne’s full stage performance specs aren’t yet currently available, we derived them from a mix of publicly available data. Here’s the numbers we used for LauncherOne performance:
- Propellant mass: 2415kg
- Dry mass: 329kg
- Stage Isp: 325s
- Payload to ISS-like LEO: 475kg
It should be reiterated that LauncherOne by itself has almost no payload capacity beyond LEO. You could add a large storable kick stage and launch small payloads beyond LEO (likely <100kg net payload), but by refueling the upper stage and a small storable kick stage, the payloads are a lot closer to the full LEO capacity for not dramatically higher costs.
Storable Bipropellant Kick Stage
The next element in the scenario is a long-duration storable bipropellant kick stage. Because most rocket upper stages are not designed for missions much longer than even an hour, we only used the refueled/recharged upper stage to perform the first burn of the 3-burn maneuver–the apogee raise into the near escape-velocity highly elliptical phasing orbit. But there are still at least a pair of burns that need to happen after this–the final interplanetary injection burn that happens at periapsis of the final phasing loop, and any plane change and/or perturbation correction maneuvers that need to happen at apogee of the orbit. For these burns, we assumed the use of a storable kick stage, though in theory this could also be performed by a storable propulsion system integral to the spacecraft. Because the delta-V required for each of these missions is different, we assumed a sort of modular “dial-a-stage” for these calculations, based on Isp and structural fraction estimates from two companies developing storable bipropellant kick stages for smallsat launchers, Tesseract and Deep Space Industries. We took an average Isp, and used the worst number on the provided structural fraction curves, even though for larger stages this is probably excessively conservative. We assumed that the stages were infinitely stretchable with the structural fraction defining how much dry mass was associated with the desired propellant mass.
Here’s the specs we used for the paper:
- Kick Stage Isp: 310s
- Kick Stage Structural Fraction: 0.25
We feel these are pretty darned conservative, and could be readily improved on with additional development. Also for most of these scenarios, we assumed the kick stage would be launched empty, and filled-up in LEO at the depot, though for most of the lower-energy missions launching the kick stage prefueled would probably impact the delivered payload by less than 20%.
Storable Bi-Propellant Lunar Lander
For the one lunar landing mission, we needed specs for a lunar lander, so we just took the same approach as the kick stage, but assumed a worse structural fraction:
- Lander Isp: 310s
- Lander Structural Fraction: 0.40
This may be overly pessimistic of a structural fraction, but we ran with it for conservatism sake, to cover things like landing gear, landing sensors, etc. In the lunar scenario, we assumed this was launched dry and filled-up at the depot.
Destinations and Departure Schedules
The following table shows the departure order and departure C3s used for this campaign:
Interplanetary Blitz Targets and Depature Conditions
|Trajectory Target||Earth Departure Date||Departure C3 (km²/s²)
|Jupiter||22 Aug 2024||86.9
|Mercury||15 Sep 2024||54.9
|Moon||26 Sep 2024||-1.99
|Mars||07 Oct 2024||11.12
|2007 XB23||02 Nov 2024||0.38
|2008 EV5||05 Nov 2024||2.15
|Venus||28 Nov 2024||11.32
|2001 QJ142||02 Jan 2025||0.65
|2009 HC||29 Jan 2025||0.31
A couple of quick notes on this table before going on to the results:
- The departure windows weren’t optimized much for arrival C3, and the missions were simulated as though they were flybys, though as you’ll see from the performance specs, it would be quite possible with most of the payloads to include propulsion or aerocapture capabilities in the payload available to turn these into orbiters or lander missions.
- All of the trajectories, including the Jupiter and Mercury missions assumed a direct trajectory without using any intermediate gravity assists. As can be seen for Jupiter this is a very high energy mission, requiring more than 6km/s of delta-V from LEO. Going to a Venus or Venus/Earth gravity assist would likely increase the payload to close to same range as is seen for Venus missions, at the cost of added mission time and complexity.
- For the lunar landing, the C3 is negative, since the TLI burn is lower than escape velocity. For this mission though there is an additional 827m/s for lunar orbit insertion performed by the kick stage, and then 2150m/s of delta-V for the lander stage.
- It should be noted that the closest two missions are only 3 days apart, which while ambitious should theoretically be possible–LauncherOne was designed for surge capacity, and the depot operations themselves should only take a few hours. But doing back-to-back missions like that would be a sight to behold.
- It should also be noted that while we picked 2024 for the blitz window, that there’s a similar season where the planets align in a similar manner in 2026. We picked 2024 as that was the soonest we thought a depot capability could realistically be available.
The missions in this campaign all used a fairly similar CONOPS:
- The mission stack–launcher upper stage, dry kick stage, dry lander stage (if used), and science payload/spacecraft are launched into LEO in preparation for rendezvousing with the depot. For air-launched missions, this can be readily done as a single-orbit rendezvous mission. For ground launch, you’d still want to limit the number of orbits prior to rendezvous due to the short stage life.
- The upper stage would then perform maneuvers to rendezvous with the depot, which grapples and secures the mission stack.
- The depot would then refuel the upper stage with enough propellant for the mission, recharge batteries if necessary, and fuel the kick stage and lander (if present).
- The launcher upper stage would then depart the depot, and at a specific pre-planned time, would perform an apogee raise maneuver to place the kick stage, lander stage, and payload into a highly-eliptical phasing orbit. The launcher stage would separate as soon as this burn is completed, keeping the amount of time the stage needs to operate at a minimum.
- At each apogee during the phasing orbit, the kick stage will perform required plane change and perturbation correction burns. For some of these near-escape orbits, lunar and solar perturbations can move the perigee around that these maneuvers are necessary to avoid either hitting the atmosphere, or having the perigee raised high enough to negatively impact the final burn. These burns tend to be pretty modest in delta-V (typically <100m/s total).
- At the end of the final phasing loop, the kick stage would then perform the final injection burn that sends the payload into interplanetary space. In this scenario, for all missions other than the lunar landing mission, we assumed the kick stage would then be jettisoned.
- For the lunar landing mission, the kick stage would stay attached to perform the lunar orbit insertion maneuver when the stack arrives at lunar orbit, inserting the lander into a low (250km) lunar polar orbit, at which point the kick stage would be jettisoned. The lander would then perform the descent to the lunar surface.
There are tons of variations on the theme that could’ve been used, and I didn’t have time to create one of those cool mission CONOPS illustrations, but this gives the general idea of the approach used.
Interplanetary Blitz Results
Using the 3-burn methodology described in the previous section, we were able to complete all nine of the missions from an ISS-coorbital depot. The following table provides a summary of key results, including providing some numbers on the delta-V and trip-time penalties incurred for using the 3-burn maneuver, the total delivered payload, and the amount of propellant that would need to be loaded at the LEO micro-depot:
Interplanetary Blitz Payload Results
|Destination||# of Phasing Loops||Total ΔV Penalty (m/s)||Flight-time Penalty (days)||Total Net Payload (kg)||LauncherOne Prop (kg)||Storable Prop (kg)
Here are some key takeaways from these results:
- As can be seen from this campaign, even though the depot was often pretty poorly aligned at the departure date for a single-burn maneuver, the delta-V penalties for using the approach were very modest–less than 100m/s, and the trip time penalties were all less than one month.
- Even to an extremely high-C3 trajectory (the Jupiter direct trajectory), this method still provides a pretty substantial net payload. For the high-C3 missions there are several ways that could improve the total net payload, including using flyby trajectories, adding an additional kick stage to split the injection burn into two segments, using a higher-Isp kick stage (like a LOX/Methane storable, or a cryo-cooled LOX/LH2 stage), doing the mission as a interplanetary boost followed by a low-thrust/high-Isp SEP mission, etc. But the fact that using this methodology LauncherOne could send over 90kg on a Jupiter direct trajectory is pretty crazy when you think about it.
- Even with the really poor structural fraction assumed for the lunar lander, we’re still talking almost 120kg of net payload to the lunar surface using this approach. Which is pretty impressive when you think about it.
- The Mars payload is around 1/2-2/3 of the injection mass of the Mars Insight lander. Which was launched on an Atlas V launch vehicle, which is around 10x bigger than LauncherOne, and has one of the highest performance upper stages in history.
- For the lunar mission, you don’t really need to use the 3-burn departure approach unless you’re either a) trying to rendezvous with a depot or other facility in lunar orbit, in which case you could probably increase the net payload to the lunar surface pretty dramatically, or b) if you were trying to land at a very specific local lunar time. Otherwise, the Moon has optimal 1-burn departure opportunities every 7-10 days from an ISS-coorbital depot.
- Most of the burns had the best payload with a single phasing loop, though for about half of the trajectories, the difference between one loop and three loops was in the noise (<1-2kg). Many of the trajectories had lower delta-V penalties on the 2 or 3 phasing loop options, but had less delta-V provided by the upper stage, which is the most efficient from an Isp and structural fraction standpoint.
- While we didn’t analyze it, it’s pretty clear that a long-lived, high-Isp/high-pmf stage like ACES or Centaur III with IVF would be pretty amazing, since you could have the stage itself perform all three burns.
- For most of the asteroid missions, the kick stage is really small, and similar enough in size that you could probably make a “one-size fits all” kick stage for asteroidal missions using this 3-burn departure methodology. And it would be a pretty small stage–less than 50kg wet.
- Most of the missions didn’t require refilling the LauncherOne stage much more than about halfway. Only the very high C3 missions to Jupiter and the lunar surface needed a full LauncherOne.
- For Rocketlab Electron payloads my gut suggests that you could multiply these results by ~50% (since it’s about half the payload to LEO but similarish performance), and for Firefly Alpha, multiply approximately 2x (since it’s twice as big). It wouldn’t be hard though to run the numbers for a different launch vehicle using the data in the paper.
- Performing all nine of these missions would require ~15.6mT of LOX/Kero (not counting boiloff losses) and about 2.3mT of storable bipropellant. This is about the amount of net payload that could be launched on the first depot launch, as a secondary payload to Cygnus, if you added 3-4 solids to the Vulcan/Centaur (call it ~$40M in net depot launch costs if you include a $10M payment to NGIS to entice them to use Vulcan instead of Antares).
- We didn’t go into the economics of the concept, but if the depot cost $100M to develop, this means you could do the depot development, launch, commissioning, and all nine of these missions for a total cost of less than $250M. If you assume the asteroid missions all use the same spacecraft design/payloads, you could do this complete blitz for less than the cost of a single NASA Discovery mission if you could keep the spacecraft design/fabrication costs below about $40M/each. The previous calculations we did a few years ago suggested that it would be possible to make a decent profit off of 2-4 missions per year, and a price point of around $25M for a dedicated deep space LauncherOne mission, and around $15M for a dedicated deep space Electron mission. This is much cheaper than buying a whole Falcon 9 mission anytime in the foreseeable future, and to many of these destinations, if you want to go at all, you’re unlikely to get many secondary payloads who can use your trajectory, so your main alternative would be buying a full Falcon 9.
Conclusions and Next Steps
I think with this paper, we’ve successfully retired concerns about LEO depots being a viable platform for enabling deep-space missions. While we would still like to run some analyses showing what the worst-case delta-V or trip-time penalties look like even if your depot RAAN has the pessimal alignment for a given mission, the data from the Interplanetary Blitz campaign preliminarily suggests that the penalties are likely pretty minor. The possibility of enabling 100-400kg class payloads to be sent almost anywhere in the solar system for launch costs in the <$25M range could be a game changer for the interplanetary community. While I don't think this would replace what NASA does with flagship, New Frontiers, or Discovery class missions, lowering the cost of a useful interplanetary mission into the $50-60M range could enable more space agencies and non-space agencies to participate in interplanetary science, enable more frequent visits to destinations that don't get enough love currently (Venus, Neptune/Uranus, etc), enable companies that would like to launch smallsat-class MEO or GEO (or lunar or Martian) telecoms relay constellations.