An Updated Propellant Depot Taxonomy Part V: Human Spaceflight Fixed Depots (Low-Orbit)

For this next depot taxonomy post, we’ll finally be talking about what people usually think about when they hear the term orbital propellant depot — larger, cryogenic refueling facilities, focused on enabling large-scale human spaceflight missions, performed by a diverse variety of users, and going to/from a wide variety of destinations. The idea for such propellant depots for enabling interplanetary human spaceflight dates back to at least 1928 with the writings of Guido von Pirquet1.

Concept illustration for a LEO Depot and reusable space tug (Credit: Cislunar Space Development Company)

This blog post will be focused on what I call “Low-Orbit” human spaceflight depots. These are depots located near the lowest stable orbit around a planetary body. This is the first place at which you can realistically refuel or switch vehicles on your way from a planetary body, and is the last place at which you can refuel or switch vehicles on your way down. As a firm believer in the idea of refueling early and refueling often, low-orbit depots are an important piece of infrastructure for any planetary system humanity wants to travel to/from regularly. A later blog post in the series will talk about “High-Orbit” depots — depots operating in fixed locations2 located out near the edge of a planetary system’s gravitational sphere of influence, and will include a discussion of where those types of depots might make sense.

Before we jump into the weeds about low-orbit human spaceflight depots, I did want to address a recent train of thought I’ve seen that suggests that just using tankers and directly refueling a vehicle is superior to having a depot involved. While this could easily be the topic of a series of its own, I wanted to briefly highlight a few of the biggest advantages I can see of having a depot vs just using direct refueling with tankers:

  • Flexibility: A depot, properly designed, with published, standardized grappling, refueling, and power-data interfaces, can be agnostic about who it gets propellant from and who it sells propellant to. Depots can quickly take advantage of whatever the cheapest source of propellant is at a given time (RLVs, ISRU, propellantless launch, buying leftover capacity from other missions going to destinations near a depot, atmospheric gathering, etc), and can easily service both smaller missions and bigger ones. Tanker-based approaches tend to be a lot less adaptable, typically being optimized for one or two specific vehicles that needs refueling.
  • Robustness: With a fixed installation, that only has to be launched once for a long mission lifetime, you can afford to throw way more resources (dry mass, volume, and power) at making rendezvous, prox-ops, docking (RPOD), and manipulation as safe and reliable as possible. This could include beacons, larger more capable (and/or redundant) relative navigation sensors and comms, longer reach capture robotics that minimize the dry mass requirements on tanker and client vehicle alike, etc.
    • For tankers, on the other hand, you want to minimize parasitic dry mass that has to be launched every time, and for the departing vehicle you want to minimize mass you have to carry through large, high delta-V in-space maneuvers. You could in theory carry a nicer RPOD kit on your departing vehicle that you jettison before leaving LEO, but now you’re amortizing that mass and cost over a much smaller number of missions, or starting to add more complexity than just doing a depot.
    • Another question to ponder is with vehicles that require large numbers of refueling events per mission, is it best to have the client vehicle handle all of those docking maneuvers with its (by definition) less-capable RPOD capabilities? With a depot, each tanker and each client vehicle only has to perform one mission-critical RPOD/refueling operation per mission, whereas with direct refueling via tankers, the client vehicle would typically have to perform a larger number of mission-critical RPOD events. A depot would also have to handle the same larger number of RPOD/refueling events, but as mentioned before, can throw more resources at making these as reliable and safe as possible.
    • Longer refueling cycles from using a direct-tanker refueling approach also increases the MMOD3 risk to the departing vehicle, which by definition can’t afford to throw as many resources at MMOD protection as a depot can.
    • A corollary to this is that depots make the most sense if you use them in a way that offloads as much of the refueling-unique hardware/software as possible from the delivery and client sides of the system to the depot. Ideally a tanker would be a minimally modified upper stage4, and client vehicles would also have similarly minimalistic hardware needed to be grappled and receive propellant. If you’re doing your delivery vehicle or your client vehicle in a way that makes you question the utility of a depot, that may be a hint that you’re doing something wrong.
  • Non-Integerality: Yes I may have made up that word, but the point is that tankers tend to come in integer quantities. Unless you always design your departing vehicles to use only integer quantities of tankers, you’ll almost always end up having wasted propellant. This is especially true given how launch vehicles, and in-space vehicles tend to increase performance and get upgraded over time, and don’t necessarily upgrade at the same rates. If you only ever had a monopoly/monopsony situation where you only had one tanker provider, and only one vehicle needing tankers, you might be able to keep tanker size locked to an integer fraction of the amount of propellant needed, but in reality that isn’t going to happen. So a tanker-based system is always going to end up wasting propellants, and this is even more the case when you have diverse customer vehicle and delivery vehicle sizes. The more diversity you have in your space transportation ecosystem, the more a depot makes sense.

There are probably other good arguments I’m glossing over, but long-story short, unless you’re interested in a boring monoculture world where only one type of in-space transportation system exists, depots make a lot of sense. So, without further ado, let’s jump into some of the taxonomical considerations of human spaceflight low-orbit depots.

Lockheed Martin concept for an orbital Mars Base Camp with refueling capabilities for their reusable Mars landers in Low Mars Orbit (credit: Lockheed Martin)

Human Spaceflight Low Orbit Depots

Application: Refueling large transfer stages or in-space transports for ferrying people and cargo between LEO, the Moon, Mars, Venus, and other destinations of interest.

Location: As discussed earlier, this blog post is focused on depots located near the lowest practical stable orbit around a planetary body. Details vary for different planetary bodies, as described below5:

  • For Earth, the low-orbit depot location is in LEO, ideally in the lowest inclination that still lets you hit required departure asymptotes and maximize the number of economically useful launch sites to send propellant, people, cargo, and materials to/from the depot6. You probably only need one or two such low-orbit depots, though if you get to high-enough earth departure throughput there may eventually make sense to spread more depots out at different RAANs7. Likely for a first human spaceflight depot, as with the previously discussed smallsat launcher depots, you’ll want to locate it in LEO near other human-occupied facilities like ISS — far enough away to be safe, but close enough to conveniently move between each other, ideally within one work shift8.
  • For a low-orbit depot around the Moon, this would likely be a polar or near-polar LLO9, though due to the very slow rotation and practically zero J2 perturbation10, if you have a lot of non-polar surface sites, you eventually may want multiple smaller depots in equally RAAN-spaced near-polar LLO planes, and maybe one in an equatorial orbit. If you’re trying to do lunar surface missions, having your depot in LLO makes way more sense than in a higher orbit like NRHO, for reasons I should probably go into in another blog post.
  • For Mars this would also likely be a LMO orbit, with an inclination high enough to be able to access any points of interest on the surface, while still being low enough to minimize delta-V penalties11, and keep the nodal precession rate fast enough to minimize phasing orbit time for three-burn departures. You’ll likely also have to put some thought into perturbations from Phobos and Deimos12.
  • For Venus, the extremely low rotation speed and therefor very low J2 pertubation may require you to do multiple smaller depots in similar inclinations but equally RAAN-spaced planes, as you won’t pass over a given point on the surface very frequently, and the very slow nodal precession rate could potentially require very very long phasing orbits for a 3-burn departure. Venus has a deep enough gravity well that you do want to refuel in LVO coming to/from, but it’s not trivial from an orbital dynamics standpoint13.

Size: As big as you can practically get away with — ideally you’d want this depot to be at least 2x the propellant capacity as whatever the largest vehicle you’re refueling. So somewhere in the 100-2000mT range, or even bigger14. Early versions will want to be single-launch if possible, in many cases repurposing at least one of the main propellant tanks from the stage that delivered them to their destination as one of the depot tanks15. Eventually, it may be possible to do multi-tank depots, but if you can do a single launch depot big enough for refueling two missions, you may be better off making more than one depot instead of trying to make the depot super big.

Propellant Types: For low depots, you’re primarily going to be dealing with large transfer stages (Centaur V, Starship, New Glenn Upper stage), which typically use LOX, and either Methane or LH2 for the fuel. Most of these use autogenous pressurization, and use the main propellants for RCS. So most of the depot will be for LOX, LH2, and/or Methane.

  • For Mars or Venus you may eventually also want to store liquid CO for some applications, since it’s an easier ISRU propellant, but that remains TBD.
  • A lunar low-orbit depot may also want to stock storable propellants, depending on what lander propellants end up being most popular16.
  • You may eventually also want to store some secondary fluids (Helium or Neon for active cooling loops, life support consumables like water, air, etc), but you may not explicitly need a depot for that function.

Some Other Considerations for Human Spaceflight Low-Orbit Depots

  • As mentioned before, human spaceflight depots want to be designed in a way to enable offloading as much of the RPO and docking/or berthing from the vehicles they’re servicing. The less parasitic dry mass that tankers and clients have to lug around on every mission, the better.
  • Storing cryogens in low orbits tends to be hard — you have a warm planet blocking half of the sky. So launching the propellant in a subcooled state or even partially frozen (i.e. slushy propellants) can help a lot. Also a lot should be done to minimize heat leaks between the cold part of the depot and any hot sections (habitation, power, etc). If you can’t get to zero boiloff, LH2 is a great thermal sponge, and can be used to chill other propellants, and intercept heat from heat sources before being vented. You may have to vent some hydrogen boiloff, but if you’re smart, you can use that hydrogen boiloff on the way out to eliminate boiloff issues for everything else.
  • These depots are also big debris targets17. Deployable MMOD/MLI18 solutions could be very helpful to avoid a puncture, which would probably be very hard to patch. Since these depots are fixed, and are typically only performing stationkeeping maneuvers, it may be possible to augment their MLI/MMOD protection over time using in-space assembly/manufacturing techniques19.
  • Especially for low-orbit depots around the Moon/Mars/Venus, there may be a benefit to having some temporary habitation/shelter collocated with the depot, especially if you’re supporting multiple sites, as a search and rescue option during exploration phases, and as a stop-over point to act as a buffer between different sizes of transportation between planets and between the depot and the surface.
  • Over time you may want to add in other facilities such as dry docks for assembling, and repairing/maintaining large in-space vehicles/structures, habitation facilities, etc. But they should probably be coorbital, near the depot, not attached (as that will make cryo thermal management all the harder, and the depot is a big hazardous work location, where you should probably minimize the amount of time people spend in close proximity to it). This could be done in two ways — coorbital facilities, spaced where the safe time to travel from one to the other is as short as practical (definitely less than an 8hr work shift if at all possible, and much closer if possible), or by having the two facilities connected by a connecting tether or other structure that includes elevator facilities.
    • If your depot facility starts wanting to have permanent collocated habitation (say for in-space assembly/repair/maintenance of in-space stages), and having the two be coorbital doesn’t work, you’re likely going to want to keep the people separated as far away from propellant tanks as possible, both to minimize heat-leak into the tanks, but also to minimize hazard to the people20.
    • For larger depots, and ones where people will be there more, putting some thought into spatially separating the fuels and oxidizers more could be a good idea. In rockets you often can’t do much to keep the two separated, and many use a common bulkhead, but in a fixed facility it’s more of a possibility. Having fuel and oxidizer that close together for long periods of time is somewhat tempting fate — you kind of have to do it for high efficiency rockets, but there’s something to be said for having your fuel and oxidizer many meters apart for a long-duration facility.
  • One area of disagreement I have with other depot advocates is whether propellants should be shipped to a depot as cryogenic propellants (LOX/LH2/LCH4), or if you should ship them as something more storable like water and CO2, and have the depot itself have large-scale electrolyzing, separation, and propellant refrigeration systems. My concern is that while in theory very large solar arrays could be done in space, combining large flexible structures like multi-MW solar arrays and radiators with a facility that sees a lot of docking, propellant slosh, etc seems like a bad idea from a structural dynamics standpoint. Also depots with very large power generation and heat rejection capabilities are likely to come later in the process, since they’ll almost certainly require multiple launches and in-space construction.

Anyhow, I probably could go on, but as with the previous parts of this series, I am only trying to scratch the surface with considerations and operating details, as I introduce each new type of depot. This definitely isn’t the last you’ll hear from me on the topic.

Next-Up An Updated Propellant Depot Taxonomy Part VI: Roving Depots

Posted in Blue Origin, Commercial Space, Launch Vehicles, Lunar Commerce, Lunar Exploration and Development, Mars, NEOs, Orbital Dynamics, Propellant Depots, RLV Markets, Space Development, Space Exploration, Space Transportation, SpaceX, ULA, Venus | Tagged , , , , , , , | 13 Comments

An Updated Propellant Depot Taxonomy Part IV: Smallsat Launcher Refueling Depots

After a brief hiatus, I’d like to continue this series with the type of depot I’ve spent the most time thinking about over the last several years, and the first type of depot that really fits the mold of what people typically think of when they hear the word propellant depot–depots focused on refueling smallsat launcher upper stages. Before I get into the details of expected characteristics and considerations of this type of depot, I’d like to give some backstory on what led me to thinking of this type of depot1.

For those of you who haven’t been neck deep in the politics around propellant depots over the past decade and a half, suffice it to say, large-scale depots focused on human spaceflight missions have excited more than their fair share of political opposition. Because of that, I started looking at other situations where cryogenic propellant depots could make economic sense independent of any change in the political dynamics around depots for human spaceflight2. Of several potential economically-viable depot markets, the one that I became most interested was in using depots to support sending small satellites to destinations beyond LEO.

Over the past two decades, there has been a steady and now rapid increase in the utilization of satellites smaller than 500kg (“smallsats”) in LEO, driven in part by the continuing miniaturization of electronics and sensors, new manufacturing techniques, and increasing rideshare and dedicated launch capabilities. We’re now also starting to see interest in sending smallsats further afield, with some companies doing GEO smallsats for serving smaller GEO telecom markets and Bringing into Use applications3, and even several groups working on interplanetary smallsats missions to the Moon, Mars, Venus, and beyond. One of the biggest challenges with beyond LEO smallsat missions is that almost all of the options for getting smallsats to destinations beyond LEO suck.

Specifically, here are some of the current options for launching smallsats beyond LEO and some of their limitations:

  • Rideshare: If you’re going to a popular enough orbit, sometimes you can hitch a ride as part of a bigger mission. Unfortunately, as a secondary payload, you have little control over the timing, you can only go to places where others are going (or get dropped off along the way)4, and there are often lots of restrictions and added scrutiny of secondary payload propulsion systems. If you’re going someplace unpopular, rideshare may at best modestly reduce the amount of delta-V your spacecraft has to produce.
  • Buy a Bigger Flight: You could also just secure a larger flight than you need, like say on a Falcon 9, and try to sell the rest of the space. But then you’re stuck herding cats, and if one of them isn’t ready on time, you may have to foot more of the bill if you can’t afford to delay the mission5. And selling the rest of the space only works if you can find an orbit you can drop people off in along the way that they actually want to go to, which can add additional mission constraints to an already complicated mission.
  • Make a High Delta-V Smallsat: You could also just try to make a really high performance smallsat, maybe with staging, drop tanks, and/or an electric propulsion system. But in most of these cases, your propulsion system now dominates your satellite, the amount of net usable payload may be very sensitive to even modest mass growth in your propulsion system, and in the case of EP systems, you may dramatically add to the amount of time it takes to get to your destination.
  • Fly on a Dedicated Smallsat Launcher with a Third Stage: If your payload is small enough, many of the dedicated smallsat launchers are now either offering or contemplating the use of a small chemical or EP third stage. RocketLab for instance can send 15-40kg net to destinations beyond LEO. But the cost in $/kg delivered to the destination can be >$400k/kg.

From talking with at least a few developers of beyond LEO smallsats, what would be nice would be a way to launch your satellite on a dedicated mission that was reasonably right-sized for your spacecraft, with a propulsion system that could get you to your desired orbit as quickly as possible, while minimizing the propulsion requirements on your satellite. And that’s where Smallsat launcher refueling depots come in.

I’m including this Artist’s Conception of a Smallsat Launcher Refueling Propellant Depot Again, Because It’s Awesome, and Because I Can (Credit: Brian Versteeg)

Smallsat Launcher Refueling Depots

Application: Refueling the upper stages, kick stages, and payloads launched by smallsat launchers, for sending dedicated, on-demand missions beyond LEO to MEO, GEO, Cislunar space, or interplanetary destinations.

Location: LEO, likely a singular station (or a small number of stations), ideally at a low altitude (<500km), mid-inclination, and near where other missions are going to. My current favorite location is in an ISS-trailing ~400km x 51.6 degree orbit, trailing as closely as NASA will allow (hopefully <200km behind).

  • The moderate inclination is to maximize the range of interplanetary departure declinations that can be hit6, without being too high of inclination for practically getting to GEO or equatorial MEO destinations.
  • For deep space departures, you’d prefer to have your perigee at departure be as low as possible, to maximize the benefit of the oberth effect. Also for most smallsat launchers the LEO payload falls off somewhat rapidly. But on the other hand you don’t want to be so low that you’re having to waste a lot of propellant on stationkeeping for your depot. 400km like ISS is a reasonable tradeoff.
  • Picking an orbit where many other missions are going to, like an ISS-like orbit, potentially enables buying excess propellant from those other missions. As it is, most missions to the ISS massively underutilize the mass capacity of the launch vehicles, meaning there is potential for buying leftover propellant from commercial crew/cargo launches. Because the primary customer has paid for the whole mission, selling this leftover propellant would be pure profit for the launch operator, potentially enabling pretty interesting price points.

Size: At least 5-10mT capacity, maybe up to 20-40mT on the high end.

  • You want something small enough to launch on a single launch, either as a secondary payload on one of the larger launch vehicles, or as a dedicated launch on one of the larger smallsat launchers (Relativity, Firefly, ABL, etc.)
  • You’d like enough propellant capacity to handle at least 2-3 missions with your largest customer, because mission demand may not be well synchronized with when you can get propellant especially if you’re buying excess propellant from ISS missions.
  • Ideally you want to be bigger than the excess propellant capacity of say a Falcon 9 Dragon mission to ISS, or an Atlas or Vulcan mission with other crew/cargo vehicles, so you can buy as much propellant as possible when its available.

Propellant Types: LOX plus Kerosene and maybe Methane for the upper stages, some form of storable bipropellant for the kick stages and payloads, and helium for pressurization.

  • Most of the existing smallsat launchers use LOX plus a hydrocarbon propellant (mostly Kerosene, but with a few looking at Methane) for their main stages, and some form of storable bipropellant combo (many using HTP plus some sort of hydrocarbon) for the kick stages. Exactly which storable propellant combos get settled on will likely be driven by which companies first start taking this type of depot most seriously.
  • Even though LH2 is almost certainly not a propellant smallsat launcher customers are likely to buy anytime soon, buying some leftover LH2 from a Vulcan or other LOX/LH2 upper stage might still be useful as an expendable coolant to supercool the other cryogens, potentially eliminating cryogenic boiloff potentially without requiring active cooling if you can get LH2 frequently enough.
  • For the helium, it might be worth trying to recover the helium from the upper stages being refueled, removing any oxidizer/fuel trace impurities, and recompressing it. In that way you wouldn’t actually need much helium other than to make up for losses and cold-gas usage on the upper stages. The helium you’ll likely want to try to keep as cold as possible, to make storage easier, so you likely want to have it in close thermal contact with the LOX or methane tanks. though for safety reasons you might not want the helium tanks inside the LOX tanks.

Other Considerations for Smallsat Launcher Refueling Depots

  • You’ll almost certainly want to design the depot to last as long as reasonably possible. This will likely drive you to make the depot robotically serviceable with deliberate modularity for likely wear components.
  • This class of depots will almost certainly be designed for purely robotic operation, without any human habitation capabilities. Though you might want to make the serviceability designed for both robotic and manual servicing. Maybe.
  • You’ll almost certainly want to have your depot repurpose at least one of the main propellant tanks of the upper stage that launches the depot, once the depot has been delivered to orbit. As I’ve shown in previous papers, this is a great way to get free depot tankage capacity. If the depot was launched as a secondary payload on Vulcan, you could get up into the ~40mT capacity for LOX and Kerosene if you wanted to.
  • You’ll almost certainly want to have a fairly capable capture and manipulation robotics capability, with RPO sensors on-board. The goal is to offload as much requirements-wise from the customer upper stages/kick stages to the depot as possible. My personal preference has been to see if the customer smallsat launcher upper stages/kick stages can maneuver their stack adequately to do a drift-by near-rendezvous7 close enough that one or more deployable capture arms can magnetically grapple and retract the stacks — ie avoiding trying to make the smallsat launcher stacks capable of full RPO/docking maneuvers, while also trying to see if you can avoid the depot having to do rendezvous maneuvers in nominal cases. The depot is probably the much heavier of the two vehicles/stacks, so it’s more efficient to have the smaller vehicle move, but you want to do it with the least mods to a stock upper stage–ideally just grapple and refueling ports, and an upper stage capable of being remotely guided by the depot8.
  • Most upper stages and kick stages have RCS thrusters that enable at least 3-axis attitude control. Most upper stage engines also likely have a cold gas purge function to blow excess propellant out of the engine prior to a relight (to avoid the risk of a hard start). It may be possible to use such a cold gas purge with the main engine to provide enough axial thrust, and with a sufficiently tight minimum impulse bit, to enable the close flyby rendezvous maneuvers. This is something I haven’t had the budget to simulate in detail, but it’s my preferred approach.
  • I should probably do another blog post at some point about the work Altius has done on cryogenic and storable refueling for rocket upper stages. Long-story short, our approach is focused on minimal modifications to the T-0 umbilical ports that the stages already typically have to have to enable reconnecting umbilicals on orbit, combined with poseable individual-fluid transfer lines, because you’re going to be servicing a wide range of customers where you might be able to get standardized T-0 quick disconnect interfaces, but you’re almost certainly not going to be able to standardize things at the umbilical plate level (since not all stage use the same propellant combinations). More on that some other day.
  • Alternately, if such a close rendezvous approach isn’t feasible, another option would be to use a servicing tug to grapple the smallsat launcher stack and tow it to the depot. Depending on where the OOS ecosystem is by the time such a depot exists, this may be a relatively simple operation that can be “bought by the drink” by a servicer that also services other clients.
  • Michael Loucks, John Carrico, and I wrote an AAS paper that I discussed in a pair of previous blog posts, on the orbital dynamics of using such a depot. Some key takeaways from that study that are worth repeating in this blog post include:
    • For most beyond LEO missions, you’re going to want to refuel both the launcher 2nd stage and a storable propellant kick stage. The 2nd stage does your boost to a highly elliptical orbit, and the kick stage does the rest (plane changes at apogee, circularization burns if going to MEO/GEO, injection burns if going to lunar or deep space destinations).
    • For payloads that themselves don’t need much refueling, such a system was shown to enable you to send >80% of your LEO payload for the smallsat launcher onto a TLI, TMI, or TVI trajectory, and you don’t even need to top the second stage up all the way typically. So for RocketLab, if you could refuel and reuse their 2nd stage and their Photon stage9, you could send on the order of 250kg into an interplanetary trajectory this way, instead of the ~15-25kg they can send currently. With Virgin Orbit you’d be talking up to ~400kg, and for Firefly you’d be talking about nearly a ton.
  • One last consideration is the economics of such missions. Since your depot is buying its propellant wholesale, in bulk from larger, more efficient rockets, the depot enables much lower $/kg costs for interplanetary missions than you would think. I need to redo the analysis with updated details, and a good set of eyes looking at assumptions, but when I ran the numbers previously, you could run a pretty profitable depot where the depot mission was 1.5-2x the cost of a normal LEO launch by the smallsat launcher. But since using a depot increases the amount of beyond LEO payload you can launch with a dedicated smallsat launcher by typically a factor of 10x over what you could do with just putting a kick stage on without LEO refueling, my previous calculations were suggesting interplanetary missions at a price point around $50k/kg for many of the smallsat launchers was feasible, and with total mission prices much less than half the cost of a reusable Falcon 9. As I said earlier, I need to rerun the numbers, but I think those would be in an interesting price range for customers.

Anyhow, while we could definitely go on and dig way, way deeper into the technical weeds on how to do the depot, how to make an upper stage compatible with such a depot, how you’d do the rendezvous/prox ops for getting to the depot, and the economics of such a depot, hopefully you can see why this concept is potentially very powerful for enabling affordable, dedicated beyond LEO smallsat missions.

[Edit 11/16/2020: I completely buried the lede, and totally forgot to mention the fact that my company is working with Eta Space, a Florida-based cryogenic propellant management startup on their LOXSAT-1 flight demo under NASA’s Tipping Point Technologies program. This 9-month flight demonstration, which is set to launch on a Rocket Lab Electron vehicle in 2023, will demonstrate a suite of cryogenic fluid storage, management, and transfer technologies, including using a version of Altius’s cryogenic refueling coupler to transfer LOX between two tanks on orbit. Eta Space’s planned follow-on depot, LOXSAT-2, would be a LOX-Kerosene depot, potentially of the very kind described in this article. They’re targeting a 2025 timeframe for fielding this depot, and I’ll be supporting the Eta Space team on conversations with interested partners/customers, and development of design and mission concepts for LOXSAT-2. After years of talking about propellant depots, I’m personally really excited to be supporting Eta Space and NASA on this project.]

Next-Up An Updated Propellant Depot Taxonomy Part V: Human Spaceflight Fixed Depots (Low-Orbit)

Posted in Launch Vehicles, Lunar Exploration and Development, Orbital Dynamics, Propellant Depots, Space Exploration, Space Transportation, SpaceX, ULA, Venus | Tagged , , , , , , | 6 Comments

Blog Links Updated

Ben Brockert in comments pointed out that with the server switch, now might be a good time to clean out and refresh the blog links, seeing as how about half of them were for companies that had been out of business for a while or blogs that were now defunct. Hopefully you find the new links interesting, if they’re ones you’re not familiar with already.

Posted in Administrivia | Leave a comment

Blog Migration Completed

I mentioned in a blog post last month that we’d be migrating the domain name registrar and site hosting. We’ve now completed the process, and as far as I can tell everything migrated correctly. If any of you see any issues, let me know, so we can get them fixed.

And once again, I’d like to thank Michael Mealling for hosting us for the last 14 years. I’m glad I’m in a position to start paying our own way.

Posted in Administrivia | 2 Comments

An Updated Propellant Depot Taxonomy Part III: GEO Depots

Of the six depot types I’ll be describing in this series, GEO Depots are probably the least fully-baked of the concepts for me, mostly because I’ve only had limited involvement in the traditional GEO satellite world. But I wanted to share a few preliminary thoughts about how one would do depots in support of activities in the Geostationary belt before moving on to talk about more traditional depots. And before getting into the specific characteristics I think GEO depots will likely have, I wanted to share some background thoughts on design drivers for GEO depots, to walk you through my logic on where I think things will go.

Background on Geostationary Orbit

First off, for those of you newer to spaceflight, what is the Geostationary Belt1? Basically, when people talk about GEO they’re typically talking about a circular equatorial earth orbit whose altitude (~35,786km) is such that its orbital period is exactly one day, meaning that to an observer on earth, the satellite always stays in the exact same position in the sky. Sir Arthur Clarke came up with the idea of using such orbits for telecommunications in the 1940s, and today there are hundreds of GEO satellites, providing not just communications, but also weather observation, and other commercial, scientific, and military functions.

Animation of a geostationary satellite
Illustration of Geostationary Orbit (from the NASA Basics of Space Flight page)

Satellites in GEO experience various orbital perturbations2, that require stationkeeping maneuvers to prevent the satellite from slowly drifting out of position. These stationkeeping maneuvers amount to ~52m/s per year of delta-V requirements. Satellites are designed with enough propellant to not only get to GEO, but to perform stationkeeping for a certain amount of time, and then at end of life boost themselves up to a “graveyard orbit” that’s typically ~200km above GEO.

One nice thing about GEO compared to LEO is that pretty much everything is in the same plane, heading in the same direction, all moving at about the same velocity3. Which means that moving between different GEO satellites never requires costly inclination changes. You basically raise or lower your orbit into an orbit with a slightly different orbital period and either catch up with the new satellite, or let it catch up with you. But to give you an idea of scale4, an object in the graveyard orbit, 200km higher than GEO, will take ~10min longer to complete an orbit, which means that over the course of ~140days, the GEO satellite will “lap” a spacecraft in the graveyard orbit5. I’ll get into why this matters later.

Another characteristic of GEO is that most GEO satellites are pretty big. Some GEO satellites are bigger than a school bus, can weigh several tonnes, and cost hundreds of millions of dollars. There are some groups like our friends at Astranis that are trying to develop GEO smallsats (~350kg in their case), but the vast majority of GEO satellites, whether commercial, civil, or military, are over 1 tonne.

Because of the very expensive nature of satellites in GEO, there has been a lot of interest in servicing GEO satellites, primarily to extend their life. Most GEO birds are designed with ~15yrs of propellant on-board, and normally when they run out, they have to be boosted to a graveyard orbit to avoid becoming a “zombiesat” that endangers other GEO operators. However, in many cases by the time the propellant starts running low, the satellite may still be economically useful. Maybe it’s been transferred from a higher value orbital slot to a lower value one (one owned by a less populous or less well developed country), but in many cases, the satellite can still be producing millions or dollars per year of revenue. Because of these realities, several companies have proposed servicer vehicles that could fly up to a satellite that’s almost out of fuel, dock with it, and then either refuel it or provide “jet pack” services where the servicer takes over stationkeeping maneuvers. Our friends at Northrop Grumman Space Logistics Services finally pulled off the world’s first successful commercial satellite servicing mission just this year, with their MEV-1 vehicle. Other players in the field include our friends at Astroscale/Effective Space6, as well as our friends at Maxar7. Since GEO satellite aren’t typically designed for servicing, these missions have focused on leveraging structural features on GEO satellites (liquid apogee motor nozzles, or the separation system hardware left on the satellites after they’re released by their launch vehicles) to mechanically grapple the satellites. Initial servicing missions are focused on what I was calling “jet pack” services where after grappling with the client satellite, they take over stationkeeping and other propulsion requirements. But most of the players have plans to move to fancier services, such as using a servicer to clip on a Mission Extension Pod to provide the jetpack services without tying up the more expensive servicer, or using robotic manipulators to fix stuck deployable structures, etc.

Illustration of a MEV providing stationkeeping services to a client satellite (Credit: Northrop Grumman Space Logistics)

One final attribute of GEO is how things get there. Most GEO satellites are launched on a rocket into a Geostationary Transfer Orbit (GTO), which has a perigee down in LEO, and an apogee up at or above GEO8. Occasionally, mostly for military GEO satellites, the rocket will also then perform a second burn at apogee to circularize the orbit and drop the satellite off directly in GEO, but for most commercial satellites, the satellite itself provides the ~1.5km/s worth of maneuvers to raise perigee, lower inclination, and circularize in GEO. Some satellites do this quickly, in a small number of chemical propulsion burns9, others take months slowly spiraling out using more efficient solar electric propulsion.

What I think This Means for GEO Depots

My first opinion when I started looking at GEO depots, was that because it was a lot easier to get around GEO without using lots of propellant, that maybe it would make sense to have a single aggregated depot, located in a circular orbit somewhat above GEO. But then several factors made me reconsider that as I dug deeper:

  • First, at 200km above GEO, you’re talking a very long time for the depot to make it’s lap around the GEO belt. Servicers are expensive, so you don’t want to have to wait forever to replenish/resupply. You can cut into that time by having your depot higher than GEO (where the relative orbital periods mean that it takes less time to do a lap around the belt), but the round trip delta-V requirements start growing pretty quickly on you. Including a rendezvous with the depot, you’d be at over 60m/s of round-trip dV with a depot 800km above the GEO belt, but it would still take over 1 month for it to do a lap, which means you could be waiting for weeks. If media reports that Intelsat is paying Northrop Grumman $13M/yr for MEV jetpack services are correct, that means that waiting a month could be costing you $1M worth of revenue.
  • Also, because most missions, and thus most rideshare opportunities go to GTO, not GEO, you’re going to have to use a decent amount of propulsion to get any propellant or supplies (spare extension pods, tools, etc) up to the depot10. Sending a tug down to pick things up in GTO and bring them back to GEO may be possible, but that’s a lot of dV 11, and for low-thrust, potentially several months. I haven’t done the detailed trades, so I can’t be sure, but I think this suggests that in many cases you’ll want the depot deliveries to be self-propelled to some extent.
  • Also, because no GEO satellites are currently designed for refueling, most of what the depot(s) would be storing will be propellant, extension pods, tools, and other supplies for servicers themselves. That means that almost all depot customers in GEO will be satellite servicing vehicles with full rendezvous, proximity operations, and docking (RPOD) capabilities, and in many cases with sophisticated servicing robotics.
  • Lastly, all GEO servicers I’ve seen use either storable chemical bipropellants or storable electric propulsion gases. Nobody tends to use cryogenic propellants or other things that need careful thermal design or active cooling.

To me, all of these things undermine the case for a unitary depot, and push me in the direction of a more distributed depot arrangement, like in LEO. Basically, you don’t want to have to take a long time to get to/from a depot, so having more than one of them spread throughout GEO makes sense. Since you likely will need on-board propulsion anyway to get your supplies into the depot orbit, it’s harder to have a dumb tanker that supplies a more sophisticated unitary storage facility. Since your customers all have to have RPOD capabilities, and most have robotics, there’s less need for that on the depot side, so once again less benefit to a centralized depot that can amortize its robotics/RPOD capabilities over lots of customers. Lastly, since the propellant mostly doesn’t need much propellant conditioning, there’s not a big advantage for storing it in a larger quantity like there would be if the propellant needed was cryogenic.

Anyhow, so based on all of that, here’s my best guess at what I think we’ll see for GEO depots.

GEO Depot Characteristics and Considerations

Application: Propellant for refueling GEO servicers and “extension pods”12, spare extension pods, and tools for servicers.

Location: In a circular orbit at moderately higher altitude than GEO (likely 100-400km over GEO), spread out fairly evenly to minimize the wait time before a depot has drifted close enough to boost to and rendezvous with.

  • Note that for propellants, one depot location is like another, but for specific tools or pods, it may be important to have the depot preposition itself in the right orbital position relative to the servicer, rather than just going to any old GEO depot.
  • For payloads launched via PODS or similar GEO-insertion opportunities where the material may not have its own propulsion, it may need a servicer to capture it, and drag it up to a safe operating altitude for temporary storage if it’s not needed right away.

Depot Size: Likely ESPA class (180-400kg). Most rideshare opportunities to GTO are via ESPAs or similar adapters. PODS-delivered options may be more in the 100-150kg range.

  • As described elsewhere, I see these depots as likely being one or more propellant tanks, attached to a propulsion system, with some basic spacecraft bus functionality. They likely wouldn’t have RPOD, but would likely have grapple fixtures, and possibly some servicing ports for attaching tools, pods, or transfering propellant, or attaching dumb payloads that need to be attached to something capable of stationkeeping.
  • It’s possible that the propulsion/bus system that delivers the payloads from GTO could be some sort of deployer tug like what Rocketlab is doing for Photon or what Spaceflight is doing with their SHERPA vehicle. Which means that in theory, the system might also be delivering some smaller GEO satellites along the way to getting into a depot parking orbit.

Propellant Types: Storable chemical propellants and EP propellants. Unlike LEO, there’s a lot more standardization of propellant types in GEO. Most GEO satellites, once they’re all the way in GEO use either something in the Hydrazine family as a monopropellant, or Xenon as a EP propellant.

  • Though as with all other depot types, the propellant type chosen is going to be driven by what clients (in this case servicers) are looking for. Xenon seems like a likely first bet for most customers, though Hydrazine or one of its variants (MMH or UDMH) might also be first.
  • In addition to just the propellants, the depots would also likely be storing tools, “pods”, and other hardware supplies.

Other Characteristics/Considerations:

  • If PODS or other direct GEO insertion options result in significant numbers of deliveries of dumb cargo pallets or tanks, it could be beneficial to haul them up and attach them to one of the self-propelled depots, because in that case the PODS themselves would lack the stationkeeping and other functionality you’d get from a self-propelled tanker.
  • But given the tradeoff between lapping time and round-trip dV to depots, I think you’ll still want to keep things fairly small overall, since you’ll want to have a large number (maybe eventually more than a dozen) smaller depots.
  • Because GEO is so hard to reach from any place you could likely get propellant from, I don’t see the likelihood of reusable tankers being used to fill up a permanent depot tank, I really think you’ll be talking about expendable tankage for the most part.
  • However, the tankage is a lot more likely to last for more than one refueling — since there’s a lot of variation in sizes of servicers, so most tanks will likely be involved in multiple refueling. And if a tank starts getting relatively low, it might be worth transferring a little from one tank to another tanker/depot before retiring the empty tank.
  • Because these depots would already be operating in a safe GEO graveyard orbit, they won’t need further disposal delta-V. Though since the use of graveyard orbits may not always and forever be best practices for GEO disposal, having grapple fixture on-board could be a good idea to enable future disposal missions — but you’d probably want those anyway just for making refueling operations easier.

As I said up-front, of all the depot concepts, this is the one I feel least definite about. I’d love to hear other people’s thoughts, but I tried to do my best to put together some logic and rationale for how I think things would turn out, and why.

Next-Up An Updated Propellant Depot Taxonomy Part IV: Smallsat Launcher Refueling Depots

Posted in Orbital Dynamics, Propellant Depots, Satellite Servicing, Space Transportation | Tagged , , , , , | 6 Comments

An Updated Propellant Depot Taxonomy Part II: Distributed LEO Nano-Depots

For this first type of propellant depot, we have an idea that several of us seem to have independently realized. My company is developing a family of servicing tugs focused on servicing customers in Low Earth Orbit. We realized pretty early on that a lot of the economics of satellite servicing depended on how many missions a given servicer could complete over its lifetime. Much like with reusable launch vehicles, the more missions you could spread a servicer/tug’s replacement cost over, the lower the cost per mission. Which leads you really quickly to the idea of refueling. On the one hand you could try and make your servicer capable of doing a ton of missions off of a single tank, but that quickly bloats your vehicle, and you end up spending most of your propellant lugging around the propellant for future missions1. So it was obvious that to maximize the economics of LEO servicers you want to be able to refuel early and refuel often. Last year as part of an AFRL SBIR Phase I contract, we worked with SpaceWorks to create an economic model for LEO servicing with our Bulldog vehicle2, to try and figure out how small of satellites we could economically service3. One of the big takeaways from this research is that for LEO servicing, the ideal way to do propellant depots isn’t to have one big centralized depot that everything goes to, but to have a disaggregated constellation of small propellant pods that you could position at strategic places near where your servicer was operating, to minimize the amount of propellant needed to move back and forth between the depot and your clients. Our friends at OrbitFab independently came up with the same concept that I’m calling Distributed LEO Nano-Depots.

Conceptual illustration of OrbitFab’s Distributed LEO Nano-Depot architecture

Distributed LEO Nano-Depots

Application: Refueling servicer/tugs and other satellites in LEO and eventually other areas.

Location: LEO orbits near where servicing and/or refueling needs to take place (e.g. Sun Synchronous Orbit, in planes near constellations, etc.)

  • As mentioned earlier, SSO4 is a good place to start focusing on for satellite servicing because there are a small number of planes, all with approximately the same inclination5, with all of the satellites heading in the same direction at the same speed. To avoid a long discussion, this all makes it easier for servicers to maneuver around between clients without having to perform a lot of propellant-intensive inclination change maneuvers. SSO is also where a lot of satellites go because it’s useful for many kinds of earth observation applications, and because of this SSO is also where a lot of the biggest and most capable LEO satellites exist.
  • Eventually you’d also want distributed nano-depots placed in planes near other large constellations, as they begin to become more interested in serviceability (especially including backup post-mission disposal services, and refueling). Right now a lot of these constellations believe that they can get their replacement cost so low that there’s no way servicing could ever be useful for them. But space is the only domain where anybody would be dumb enough to treat an object that takes several hundred thousand to several million dollars to replace as a throw-away item.

Depot Size: Cubesat scale (likely 16U) up to ESPA class (up to ~180kg)

  • The size of the depot would likely be driven by both the propellant tank size of servicers and client satellites, but also by convenient sizes that can be purchased as “excess capacity”.
  • [Edit 9/17: I realized that I didn’t really clearly explain the configuration I had in mind. I think that most of these Nanodepots will basically be one or more tanks packed into a Cubesat body or ESPA-class satellite body, with the minimal avionics needed to maintain attitude, perform collision avoidance/deorbit maneuvers, and maybe perform simple relocation maneuvers, and a grappling fixture and a fuel interface. To me less like a tiny space station, and more like an expendable “smart” propellant canister. I’ll get into more in later posts about what situations where a more permanent facility that people normally think of when you use the term depot makes sense.]

Propellant Types: Storable monpropellants or bipropellants (e.g. Hydrazine, MMH, UDHM, HAN, ADN, NTO, MON, HTP, non-cryogenic hydrocarbons), working fluids for electrothermal systems (water, ammonia, etc.), and electric propulsion propellants (e.g. noble gases like Xenon and Krypton, sublimable solids like Iodine or Bismuth, liquids like mercury or ionic fluids, etc.), and maybe pressurants (neutral gases like nitrogen or helium).

  • Most servicers, short-range (intra-LEO) tugs, and satellites use some form of storable chemical or electric propulsion. Unfortunately while there is some commonality among larger, more traditional spacecraft6, smaller commercial missions have not really settled on one or two most common propellant types. Frankly almost all of them have at least one serious drawback (toxic, corrosive, carcinogenic, expensive, unstable, low performance, easy to freeze, hard to light reliably, etc, etc.). Picking which propellants to focus on for a nano-depot constellation is going to be challenging.
  • My guess is that the way this will evolve will depend on the propellant choices of whoever starts doing LEO servicing seriously first, combined with the propellant choices of whichever constellations start buying refueling services first. Once you have a large enough customer to justify a nano-depot constellation, there are now strong incentives/network effects for future groups that want to benefit from refueling to pick what’s already being used by others.

Other Key Characteristics/Considerations for Distributed LEO Nano-Depots:

  • In many cases these nano-depots may actually be single-use tanks. I haven’t run the numbers yet, but I’m skeptical it will make economic sense to collect empty nano-depots, move them back to some central refueling place, refuel them off of bigger tanks, and then move them back out again. That said, I reserve the right to change my opinion if I run the numbers and they suggest that could actually work.
  • Ideally you’d want to scatter these as close to the planes and altitudes where they would be used. Though at the same time, these will likely be wanting to be launched as cheaply as possible, which means they’ll likely almost always be launched on a rideshare basis. This may mean that they’ll need either some modest on-board propulsion capabilities, or they may want to be tugged to a final location after launch, if feasible.
  • Since we’re talking about a large number of objects, you’ll almost certainly want them to have at least enough maneuverability to dodge potential conjunctions with non-maneuverable pieces of debris7, and to safely dispose of a tanker once it has transferred as much propellant as possible to the servicer client.
  • While in theory you could make each of these tankers capable of performing rendezvous, proximity operations, and docking (RPOD) maneuvers with services or client satellites, I’m really skeptical things will optimize in this direction. Especially if it turns out that most tanks only get used once, my opinion is that you’re almost certainly going to want to offload as much of the cost and complexity of RPOD to the reusable element in the system (the servicers), so you can minimize the complexity on the tankers and the client satellites.
  • Eventually, similar concepts will likely want to be copied in other domains with similar orbital dynamics, where you have a large number of satellites and/or constellations distributed out between multiple planes (e.g. MEO, eventually Mars orbit, etc.)

My guess is this class of depot is going to be the first one that gets built, even if relatively speaking they look nothing like what most people would think of when they think of orbital propellant depots. If you’d like to learn more about the concept (and get their opinions rather than just mine), I’d suggest going over to the OrbitFab website, and poking around their resources and whitepapers.

Next Up: An Updated Propellant Depot Taxonomy Part III: GEO Depots

Posted in Altius Space Machines, Commercial Space, Propellant Depots, Satellite Servicing | Tagged , , , | 7 Comments

An Updated Propellant Depot Taxonomy

After way too many years in the wilderness, the concept of orbital propellant depots and in-space refueling in general, are finally beginning to be taken seriously again in public circles. A couple of examples include: Elon’s announcement that he is building his Mars architecture around RLVs and propellant depots, NASA’s baselining of in-space refueling as part of their Artemis lunar return program, even the most recent NASA Tipping Points solicitation1 had Cryogenic Fluid Management Technology Demonstration (active cooling, transfer, and pressure management) as one of its three topics. There are also now several startups out there explicitly focused on orbital propellant depots including my friends at OrbitFab, my friend Dallas Bienhoff’s company Cislunar Space Development Company, and also my startup, Altius Space Machines2.

Those of you who’ve been following this blog over the years have probably seen a lot of my previous thoughts on the topic. But I’ve been realizing that there are now a lot of new people becoming interested in the topic, and during some recent conversations on Twitter, I realized that it might be helpful to share some of my thoughts on the different types of propellant depots, and key considerations for each type of depot (things like where you’d likely put them, what sort of propellants they’d likely contain, how big they’d likely be, what you’d use them for, etc).

Instead of doing what I often do, and trying to cram six blog posts into one, I’m going to release a series of blog posts over the next week or two about the six main types of orbital propellant depots I’ve been able to think of so far3:

I’m sure there are probably more categories than that, but I figured that it was worth at least sharing some of my thoughts about these different types of depots, and their similarities and differences.

Next Up: An Updated Propellant Depot Taxonomy Part II: Distributed LEO Nano-Depots

Posted in Commercial Space, ISRU, Lunar Commerce, Lunar Exploration and Development, Mars, NASA, Orbital Dynamics, Propellant Depots, Space Development, Space Exploration, Space Transportation, Venus | Leave a comment

Blog Maintenance

Hey everyone, I just wanted to share a quick note that over the next day or so, we’ll be migrating our domain name registrar and hosting service. In theory, you shouldn’t actually notice anything on this or my other blog 1. The urls aren’t changing, we’ll still be a WordPress site, we’re just changing who we’re paying for hosting things.

I just wanted to publicly thank Michael Mealling, one of my fellow cofounders at Masten Space Systems, who has been handling, and paying for, the hosting for these two blogs since the beginning. I figure now that I was able to cash out a little of my Altius stock when Voyager majority-acquired us last year, that I can afford to start paying my own way around here.

Posted in Administrivia | Leave a comment

How much mass can we put in orbit before running into atmospheric constraints?

In January, Elon Musk mused that the design goal for Starship was 3 flights per day, for about 1000 flights per year per Starship (assuming for the sake of simplicity he’s talking about a whole stack) with a payload of about 100 tons to LEO. That’s 10 Starships flying a total of 10,000 flights per year to reach 1 megaton in orbit. He points to 100 megatons annually to orbit (100 Starships per year for 10 years… 1 million Starship flights per year) as the goal.

Building 100 Starships/year gets to 1000 in 10 years or 100 megatons/year or maybe around 100k people per Earth-Mars orbital sync

Just how big is that? Is it realistic?

Note: I decided to make a couple images to illustrate the difference between 1 Megaton IMLEO/year (corresponding to roughly 1000 larger Starships heading off to Mars each synod) and 100 Megaton IMLEO/year (roughly 100,000 larger Starships heading off to Mars each synod). There are no stars in these images, just Starships:
1000 Starships doing trans-Mars-insertion simultaneously:

1000 Starships doing TMI
(alternately, for 100,000 Starships, they might do this once per orbit for a week straight)

Going by rough figures, there’s about 3 billion tons of methane (in the form of 4 trillion cubic meters of natural gas) produced every year. Each Starship takes about 1000 tons of methane (very rough numbers…). So you’d need 1 billion tons of methane every year to reach Elon’s 100 Megaton-to-orbit goal.

1 billion tons of methane is approximately 0.75 billion tons of carbon. About 5 billion tons of carbon is absorbed total by land and sea every year (Candela and Carlson 2017), so if all the world was doing as far as carbon emissions was launching SpaceX’s rockets, this wouldn’t be a problem. 1 gigaton of payload to LEO (so about 7.5 billion tons of carbon), however, WOULD be beyond the current ability of the land and sea to absorb carbon, slightly less than the current global carbon emissions from human civilization.

That puts current Starship launch efficiency to about 580MJ/kg, compared to the absolute minimum of ~32MJ/kg, or about 5% efficient. This is remarkably efficient if you think about it, but it’s nowhere near the efficiency that’s *possible*. Stretch estimates for what SpaceX hopes to eventually achieve with Starship might be 2320 tons of methane for 380 tons of payload (propellant in this case), given the 2016 ITS tanker figures (from “making_life_multiplanetary_2016.pdf“). They had a higher O:F mixture ratio (3.8), higher Isp, and lower dry masses. That’s just under 10% efficiency. Significantly better. That would mean half as much methane would be needed, perhaps lowering the carbon emissions to 4 or so gigatons of carbon for 1 gigaton to orbit, below the 5 gigatons the land and sea can absorb… Back to the 100 megaton goal, that’s 340MJ/kg times 100 billion kilograms… divided by about 31 billion seconds in a year, and you’re talking about a terawatt of methane per year.

Of course, you could switch from fossil methane to CO2-direct-air-captured methane. That’s about 50% efficient, so it’d take about 2 terawatts average per to produce. Or, given about a 20% capacity factor, about 10 terawatts of solar nameplate capacity.

But I think they’re potentially still leaving energy on their payload. (See previous posts) If they operate at yet higher O:F, even deeply oxygen-rich, for the first stage, they can get closer to an optimum Isp for early in flight. They can switch to a near-stoich hydrolox upper stage. Maybe we continue making advances in structural materials. Maybe there’s a small launch assist in the beginning (at least getting the vehicle to an altitude where vacuum-optimized first stage engines are feasible). Or they use gigantic expansion nozzles on the upper stage; higher chamber pressures; adjustable Isp. I can imagine achieving 20% or even 30% efficiency. Perhaps 100MJ/kg could be achieved without miracles. That lowers to about 300Gigawatts of chemical energy per year. Hydrogen may be more efficient to make (75%?), so maybe 400Gigawatts of average electricity per year… The US electric grid produces about 475Gigawatts average, so for the first time we’re below the US’s electrical output to power Elon’s 100 megaton/year dream.

However… Pumping all that water vapor up in the atmosphere could cause problems, too. But let’s say we avoid that somehow. There are other problems:

NOx emissions (nitrous oxide and similar) in the high atmosphere cause several problems. One is breaking down the ozone layer. Another is acid rain (although this is also part of the normal nitrate cycle on Earth, where lightning fixes nitrogen into the soil). Another is greenhouse effects… NOx emissions are approximately 250 times worse than CO2, pound-for-pound. It’s estimated that the Space Shuttle produced about 5% (but perhaps up to like 15%) of its reentering mass in NOx emissions (see: Global atmospheric response to emissions from a proposed reusable space launch system). Air-breathing rockets would make this worse by also producing NOx on the way up (see Skylon). We currently emit about 13 megatons of NOx every year from burning fossil fuels (compared to another 8 megatons annually from lightning). Considering the current fairly high dry mass of Starship, there’s basically about a 1-to-1 ratio of Starship mass reentered to payload delivered. So 1 megaton of payload would produce about 50,000kg of NOx. Not nothing, but not a showstopper. 100megatons, however, would produce about 5 megatons of NOx emissions… almost half of what we already make, but could be even higher, if the higher estimates of reentry NOx production are accurate (I don’t think they are).

However, I think we can do much better. The 2016 ITS tanker had a propellant payload to reentry mass ratio of about 4, reducing the amount of NOx production for 100 Megatons by a factor of 4 again. And we can maybe do better by changing the staging situation… Because NOx production and reentry temperature are really non-linear, there would be very little NOx production from a reentering 1st or 2nd stage in a 3-stage-to-LEO rocket. Falcon Heavy, in expendable mode, has an upper stage dry mass of around 4.5 tons (guesstimate from, and a payload of about 63.8 tons. That puts the ratio at about 14! Over an order of magnitude better than first-generation-Starship. Maybe knock that back to 10:1 for a reusable upper stage (but still using really advanced structures) for a launch vehicle optimized for this constraint, and we could be talking only 500,000tons of NOx per year. MUCH more manageable. To equal the current 20 megatons tons (combined human and lightning) NOx per year, we can reenter about 400 megatons of material, or launch (with an upper stage empty mass to payload ratio of 10) about 4 gigatons.

It may also be possible to scrub NOx from the atmosphere. This concept (backup link: by Ming et al) suggests using a solar tower to help scrub NOx from the atmosphere and generate solar electricity at the same time. At really high launch rates, that might be necessary. In fact, any plan to use space resources to “deindustrialize” Earth (like Bezos and O’Neillians like to mention) would have to deal with the problem of (re)entry of massive amounts of material to Earth and the NOx emissions that causes.

It is also possible to pump Ozone into the stratosphere or maybe even suppress lightning to compensate.

Thinking long-term, what is the ultimate limit to ability to launch stuff with rockets, of any type? Current anthropogenic global warming from the greenhouse effect from fossil fuel emissions is much larger than, say, fundamental waste heat from any energy usage whatsover. Waste heat is on the order of 18 TW (same as primary energy usage), with global warming effect from fossil gas emissions (and land use changes) about 100 times that, so about 1-2 Petawatts. If we take current global warming level to be the ultimate limit that we could safely pursue long-term, then human society could grow to use approximately 100 times as much energy as it does right now relying on fossil fuels, or about 1-2 Petawatts. If all of that was used for chemical rockets with each kg of payload into LEO requiring 100MJ/kg, then we could get about 300 gigatons of payload into LEO per year before producing too much waste heat. Maybe with perfect launch systems, about 1 trillion tons per year.

So there are a lot of constraints. We, long-term, probably want to off-load much of that into space. That means maybe using solar-electric propulsion eventually. Before we get much beyond 1 megaton per year, I hope we’re looking seriously at scaling up solar electric propulsion, asteroid mining for propellant, and tethers. Using rotovators (discussed elsewhere on this blog), we could drastically reduce the amount of energy needed to be expended on Earth to launch payloads. And maybe just important (at that scale), we don’t need to use the atmosphere to slow down payloads to the surface of the Earth, either. Tethers combined with megastructures ~100km tall would allow payloads to be launched at higher efficiency and returned to Earth without massive aerobraking… in fact, even reducing reentry from 7.8km/s to 5.5km/s using a modest rotovator would halve the orbital energy input into the atmosphere and probably would non-linearly reduce NOx emissions as well.

Humans move on the order of 50-100 gigatons of material per year (with trucks, bulldozers, etc). See:,%2C%20glaciers%2C%20oceans%20or%20wind.
That’s more than all the sediment moved by all the rivers of the world every year.
To move that much of material *to space* every year would require some clever thinking, but wouldn’t be impossible. It’d just take on the order of 100 Terawatts. Current solar cell prices are just 5.5 cents per watt… if they were illuminated constantly, that’s less than $10 trillion dollars worth of solar cells to move more material into space than all the material that all of humanity moves anywhere every year. And we might not even have to cook ourselves to do it.

However, Musk’s 100 Megatons to LEO every year would use up about a third of the world’s annual natural gas production. Might want to move beyond fossil fuels (and maybe optimize launch vehicle efficiency) if we’re going to really launch that much stuff…

And to continue the crude visual thought experiment from earlier, this is what it might look like if those 100 Megatons IMLEO were used to send 100,000 Starships to Mars at once each synod, with a total power output of approximately 2 Petawatts for 8 minutes, maybe even regionally outshining the Sun for a few minutes:

Posted in Uncategorized | 24 Comments

Adding an Earth-sized magnetic field to Mars

Mars has only the ancient remnants of a magnetic field. What little chunks of field it does have (imprinted into magnetic rocks) are regional in scale and do nothing at all for radiation shielding (I once calculated this). Additionally, over a long enough timescale (tens of millions of years), the solar wind will erode the atmosphere of a terraformed Mars. So, let’s just get on with replacing the lost magnetic field.

To do this, we aren’t going to do something silly like restart the core. We’re going to rely on a tried and true existing technology: cryogenic superconductors. Just put a superconducting ring around the equator of Mars. Turns out, this wouldn’t even cost that much to build. I will be using Magnesium Diboride (MgB2) because it’s cheap, has pretty good performance (critical temperature of 39K, critical current of like 105 A/cm2 at 5K), and both Boron and Magnesium are known to exist fairly commonly on Mars. (Magnesium is super common, and Boron has been found in clays at >150ppm concentrations–according to this paper–and probably exists in far higher concentrations somewhere since Mars once had a quite active water cycle. I’ll assume that by the time the residents of Mars want to do this sort of thing, the costs of extracting these minerals will be similar to that of Earth (maybe a poor assumption, but why would you do this unless there are like millions of people on Mars, at a minimum?). We’ll also assume that because of the ridiculous scale being operated at, and because MgB2 is a pretty easy superconductor to make (just need to heat the mixture of Magnesium and Boron powders), the cost of actually building the ring will be some single-digit multiple of the raw costs of the material.

Okay. So just how big is the Earth’s magnetic field? We’ll use its total energy (when approximated as current on a sphere) to estimate what we’ll need as far as current in the ring. According to this, the Earth’s magnetic field stores about 1026 erg, or 1019J (roughly half the energy the world uses in a year). The energy stored in an inductor is just:
(according to Wikipedia)
Where L is inductance and I is the current in the ring.

To calculate the inductance L of a ring of radius R with wire radius a and number of turns N, we use the approximation:
L \approx N^2 R \mu_0 \mu_r \left ( \textup{ln}\left ( \frac{8 R}{a} \right ) - 2\right )

(from here)

Since N=1 and we’ll conservatively (very conservatively) say the relative permeability  \mu_r = 1, and since the current I is related to the critical current density Jc such that: I=J_c \pi a^2, we can write the equation as:

E\,=\,\frac{1}{2}\,R \mu_0 \left ( \textup{ln}\left ( \frac{8 R}{a} \right ) - 2\right )\cdot\left (J_c \pi a^2\right )^{2}
If we let a=.42m, R=rMars, and Jc = 105 A/cm2:*r_Mars*(mu_0)*(ln(8*r_Mars/(.42m))-2)*(10^5A/cm^2*pi*(.42m)^2)^2 = 1.047*10^19 Joules, when we only needed 10^19 J to equal the same energy as Earth’s magnetic field.

Given the density of MgB2 is 2.57g/cc (source), the mass of the superconductor is:*pi*r_Mars*pi*(42cm)^2*2.57g/cc or about 3*1010kg, 30 million tons, almost have of which is boron. The Earth mines about 4 million tons of Boron a year, so the Earth produces enough boron to build that thing in about 4 years (we’ll mine this on Mars, of course). Pretty reasonable, considering we’re doing some pretty hardcore terraforming, here.

Given a price of about 10USD/kg for Boron (just spitballing here, since Ferroboron is half boron by molarity and 15-20% boron by mass and is 1-2USD/kg… of course, boron ore is much cheaper) and like 2USD/kg for magnesium metal (just look up the spot prices for Mg and Ferroboron), so about USD6/kg of bulk MgB2.

This whole thing would cost about 180B USD in raw materials but would store about 3 trillion kWh for a ridiculously low price per kWh of storage (like 6 cents/kWh! For storage that can be reused!). Of course, there is also insulation and cooling, plus some method to inject power into the ring.


NOTE: Some Japanese researchers recently (last 2-3 years I think?) published a paper about the thing I am proposing here. I can’t find the paper now, but I assume they did I better job than I did. Also, I don’t really buy into Jim Greene’s L1 magnetosphere, since the solar wind does actually shoot straight out from the sun but actually follows the spirals of the Interplanetary Magnetic Field, so I’m pretty sure solar wind would hit Mars because the shadow of a big magnetosphere at L1 would miss Mars.

Posted in Uncategorized | 7 Comments