The Space Reviewin association with SpaceNews

Deep Space Gateway
An illustration of NASA’s proposed Deep Space Gateway, a human-tended outpost in cislunar space that serves as a step towards human missions to Mars in the agency’s current plans. (credit: NASA)

An alternative architecture for deep space exploration using SLS and Orion

Bookmark and Share

From the beginning, the “Journey to Mars” program architecture as a whole, and the design and execution of the Space Launch System and Orion programs in particular, have been dogged by controversy stemming from a sense held by many that there was no one at the wheel. No concrete goal or plan governing the consequential long-term decisions being made as the program advanced. Now, with many of these decisions made, and major program elements committed, NASA has unveiled an exploration architecture which doesn’t square with these commitments.

The SLS was designed ostensibly with the purpose of facilitating Mars exploration, but without any public and articulated plan for what this would mean.

As a result, it accomplishes little exploration despite very large expense and initial mass in low Earth orbit (IMLEO) over a long period of time, and endangers astronaut safety unnecessarily. However, a “revisioning” of the same architecture, which plays to the strengths of the key technologies involved (SLS, Orion, and electric propulsion), can resolve these issues, and at this early stage in the process could be undertaken with little disruption.

NASA has a rocket and a plan, but are they right for each other?

The SLS was designed ostensibly with the purpose of facilitating Mars exploration, but without any public and articulated plan for what this would mean. Why was the SLS designed with no commitment to achieve a particular flight rate that would be required for an articulated program? Why were there four versions of the same launcher, some flown as few as once? Why 70, 100, and 130 tons to LEO, and not other numbers? Why did the Orion capsule have provisions for six, and not a different number? Why 21 days and not longer or shorter? Why was Orion designed to reenter from lunar orbit and not a Mars-origin interplanetary trajectory? What, in sum, was the plan, and how were the major elements of program hardware designed against the plan?

It’s not clear that these program elements were designed against a specific plan, but designed they have been, and now, an exploration plan must be designed against them. The SLS has many disadvantages: high ongoing maintenance cost, high cost per launch, low flight rate, and very large sacrifices of payload when launching into higher energy orbits. The SLS also requires very heavy sacrifices in cargo mass and volume when cargo is co-manifested with a crewed Orion spacecraft. But the SLS has advantages as well: very large payload mass to LEO as well as a very large monolithic cargo fairing. Any plan should be formulated to take maximal advantage of these strengths, and not to tread on these weaknesses.

It’s not clear that these program elements were designed against a specific plan, but designed they have been, and now, an exploration plan must be designed against them.

Now, in an extremely welcome gesture, NASA has at last described an articulated plan for using the SLS in crewed Mars exploration. It recently published a manifest of 12 SLS launches with their destinations, cargoes, approximate launch dates, and mass budgets, leading from here to a crewed Mars mission in the early 2030s, at approximately the 2033 launch window. This mission would not be a Mars landing, but a Mars orbit mission, possibly with a Phobos and/or Deimos landing included.

Ultimately I will critique this plan and argue that it makes no sense and does not square with the design decisions made in the development of SLS and Orion. But first, it will serve to talk about the plan, and mention its many positive points. First, it’s specific. It is a real thing we can understand and analyze and discuss. Second, it is workable. Although it relies on all the pieces to work, there is nothing insane or undoable there. It could happen. And third, it’s not set in stone yet. None of the key pieces of cargo in the manifest, other than the SLS and Orion themselves, have yet been specified in such a way that they can’t change. If it’s decided that modifications to the architecture would improve it—and I will argue that they would—these changes can and should be adopted, and these improvements realized.

With this in mind, I will proceed to describe the current plan, and then critique it.

The current plan

The plan involves a total of 12 SLS launches, each with its own cargo, separated into four phases:

  1. The first two would be the EM-1 uncrewed test flight (possibly being changed to a crewed test flight), and the Europa Clipper probe being launched on a direct injection to Jupiter. Both of these previously-described missions are testing and scientific flights with their own objectives and do not directly contribute to the exploration architecture. This means that about five to six years remain before the first exploration architecture launch.
  2. Following this are four crewed SLS Block 1B launches occurring annually between approximately 2022 and 2026 that would visit a Near Rectilinear Halo Orbit of the Moon. They assemble a small space station called the Deep Space Gateway out of four pieces, each weighing about ten tons and launched as co-manifested payloads (CMP) in the interstage between the Exploration Upper Stage and the Orion service module. These four pieces—a 40-kilowatt solar electric propulsion (SEP) tug, a habitation module, a logistics module, and an airlock docking adapter—would fit the Deep Space Gateway for maneuvering itself to different lunar orbits, playing host to crews, and docking with visiting Orion craft and interplanetary spacecraft.
  3. Following the completion of the Deep Space Gateway, two SLS launches in 2027 will assemble and provision an SEP-driven spacecraft called the Deep Space Transport. The core of the DST will be a monolithic 41-ton module launched to TLI by a single Block 1B cargo flight. The DST will feature an integrated SEP drive with a power level of approximately 150 kilowatts. An additional 10 tons of logistics payloads in a CMP configuration will travel in a Block 1B crew flight, which will dock the DST to the DSG, install and configure the CMP cargo, and ready the DST before returning to Earth.
  4. Then, two missions will be carried out using the DST. Each will involve two SLS launches, first a cargo launch carrying 40 to 45 tons of fuel and logistics to resupply the DST, and then a crew launch with the crew for the mission and about 13 tons of CMP logistical supplies, after which the DST, under SEP power, will proceed on its mission before returning to the DSG for the next mission. The two missions will be first a long-term habitation checkout in lunar orbit beginning in 2029, followed in the early 2030s by the first crewed mission to the vicinity of Mars.

Now, let’s review what these launches accomplish, independent of the particular sequence of missions.

  1. They assemble a 40-ton DSG and place it in orbit around the Moon.
  2. They assemble a 50-ton DST and place it in orbit around the Moon.
  3. They provision the DST with 54 tons of supplies and place a crew, with an Orion spacecraft and service module, in it for its first mission.
  4. They reprovision the DST with another 58 tons of supplies and another crew and Orion spacecraft and service module, for its second mission.

Ten SLS launches over a twelve year period, with a total of 1,100 tons of IMLEO, to place about 250 tons of cargo and crew vehicles in lunar orbit.

Using a hundred-ton rocket to launch ten-ton modules

I would like to present a couple major critiques of this mission architecture, centered on five major questions.

1. Why does the Deep Space Gateway need to be assembled from a series of small modules in crewed flights? Why can’t it be launched intact in a cargo flight?

The decision to launch directly to TLI cuts the available mass of translunar cargo literally in half, from about 80 tons per launch of SLS Block 1B to 41 tons.

The Deep Space Gateway, as presented, is assembled from four ten-ton modules, each delivered by a crewed mission. The crew are necessary, one assumes, in order to perform the assembly sequence. While this notion is debatable (the experience of Mir, and the Russian segment of the ISS, have argued persuasively that crew are not necessarily needed for rendezvous and assembly), it’s at least somewhat defensible. But with the DSG weighing 40 tons, and the SLS Block 1B Cargo capable of throwing 41 tons to TLI in one shot, why not throw the DSG to TLI in one launch, with no assembly, and therefore no assembly crew, needed? This would launch the DSG in one launch instead of four, giving the schedule about three extra years and three extra launches worth of breathing room.

Alternatively, if the DSG is to be made up of small ten-ton modules, they could be placed in lunar orbit for uncrewed docking by a smaller, cheaper rocket like the Falcon Heavy or possibly the Delta IV Heavy, without using the SLS at all.

2. Why does the Deep Space Gateway need to be assembled in lunar orbit? Why can’t it be assembled in LEO and moved to lunar orbit with SEP?

With the Deep Space Gateway being assembled by crewed flights, of course those flights need to inject directly to the Moon on top of the SLS. With a crew on board, it would be impossible to take a year or more to spiral out to the Moon using solar electric propulsion. But if the gateway is assembled in LEO, it could move to lunar orbit on SEP without a crew on board.

The Deep Space Gateway, as described, has the ability to maneuver itself with its onboard SEP capacity, and will be launched and assembled years before it will be used in lunar orbit. With all this time, why not assemble it in LEO and maneuver it to lunar space with SEP? This would involve a much smaller sacrifice in mass. The SLS sacrifices about 60 percent of its launch capacity by launching to TLI instead of LEO. With SEP, at a specific impulse of 4,200 seconds, a cargo can maneuver from LEO all the way to Earth-Moon L-2 point, for instance, with a delta-v of about 9 kilometers a second, while burning only 20 percent of its weight in fuel. The decision to launch directly to TLI cuts the available mass of translunar cargo literally in half, from about 80 tons per launch of SLS Block 1B to 41 tons.

Indeed, this option could even be combined with the monolithic construction proposal above. In this option, a monolithic Deep Space Gateway weighing about 80 tons, along with about 20 tons of xenon propellant for its SEP module, would be launched into LEO with a single launch of SLS Block 1B Cargo, and slowly maneuver on SEP to lunar space over the course of a year or more. This would not only enable the DSG to be launched in one flight instead of four, but also enable it to be doubled in mass.

Although this imposes some restrictions on the mission architecture, such as the use of SEP and the necessity to launch the DSG and the DST a year or more before they are used, both of these two elements are already involved in the manifest as already described. Since these sacrifices are already being made, why not get more value out of them?

This would involve a slight change in design. Right now, the plan is for the DSG to have a SEP module with a power of about 40 kilowatts. However, assuming that the SEP module has the ISP and efficiency of the NEXT module currently in final checkouts, a module of about 400 kilowatts power would be needed to lift an 80-ton module to the Moon over a two-year period. If something similar to VASIMR is used, with a higher specific impulse of around 5,000 seconds but lower thrust per unit power, less fuel would be needed but the power required would increase to on the order of 600 kilowatts.

Either of these options would be a significant increase in the power, size, and mass of the SEP module relative to the current plan, but this would also make it more capable of maneuvers for a wider variety of mission profiles throughout the solar system. The need for this additional system mass, both for power generation (solar or nuclear) and engines, would be compensated by the extra mass budget afforded by the more efficient ascent to lunar space.

At the moment, there is nothing in the plan for the Deep Space Gateway to do, other than play host to the crews that assemble it.

The SEP module could be integral to the cargo (DSG, DST, etc.) being lifted, and carry it on its journey through the solar system, or it could be a reusable tug that, now without cargo, would go back down to LEO and wait to rendezvous with another payload and fuel tank for its next journey.

3. What is the Deep Space Gateway for? Why is it needed?

As we have seen, the Deep Space Gateway can be launched to LEO fully fueled in one piece and moved to the Moon using SEP, or injected to the Moon in one piece, cutting its assembly from four crewed flights of the SLS to one cargo flight. But is it even needed?

At the moment, there is nothing in the plan for the Deep Space Gateway to do, other than play host to the crews that assemble it. By the time the first crew arrives to the Deep Space Transport, the vehicle is already assembled and provisioned. The crew arrives and departs on the DST, and when they return, visits the DSG only as a waypoint on the journey back to Earth.

Why can’t we eliminate the DSG entirely? Why can’t the crew simply dock with the DST and take off on their mission, and then, on their return, park the DST in lunar orbit, undock, and use Orion and its service module to return to Earth? What is the Deep Space Gateway for?

4. Why does the DST need to be a singular piece of hardware that is reused for multiple missions? Wouldn’t it be easier to design, and easier to improve on the basis of flight experience, if it were used once? Wouldn’t it be safer to send astronauts into deep space in a new DST, rather than a used one that has been in deep space for a decade?

As we have seen, doing away with the presumption that our hardware must be assembled by crewed flights in lunar orbit, and instead launching it as LEO cargo and moving it to the Moon using SEP, allows approximately an eightfold improvement in the amount of cargo launched to the lunar jumping-off point with each launch of the SLS. This would allow about 80 tons of cargo to lunar space with SLS Block 1B Cargo, or about 104 tons with SLS Block 2 Cargo. With these kind of figures, a complete and provisioned Deep Space Transport spacecraft could be lofted in one shot. Even with the plan as written, it’s possible for a provisioned DST to be launched to lunar space in two pieces, which dock under remote control and automation, and are met by a crewed Orion spacecraft shortly prior to departure.

This would mean that a DST could be positioned in lunar orbit in only one or two launches of the SLS. Since it will not be complicated to assemble, this means it’s quite feasible to assemble a brand new DST for each deep space mission, at least initially. This has the disadvantage of requiring the additional construction and lift of DST modules, but several advantages.

One advantage is that it would allow the DST design to evolve as requirements change and as lessons are learned, rather than making the design unalterable after its first iteration. Right now the DST is scheduled to undergo a long-duration test in lunar orbit with a crew aboard, but there is no provision in the current plan to make use of any lessons learned. If flaws are identified in the life support or other systems on the DST, there is no plan for altering the design to account for this other than whatever maintenance could be done by the mission crew when they arrived to depart on their mission! Instead, launching a new DST for each deep space mission would allow the DST design to be altered as needed to account for lessons learned from each mission.

A second advantage is that it would give the crew the assurance of a brand new, fully functional vehicle instead of a used one. Every spacecraft has a finite life, and every long-duration spacecraft (Skylab, Salyut, Mir, and the ISS) has suffered degradation of various functions over the course of its time in service. The ISS, Mir, and other long-duration spacecraft have shown that long duration is possible, but also that it presents complications. When we make the decision to irrevocably commit a team of astronauts to a three-year deep space mission for the first time, surely it would be preferable to send them in a brand new spacecraft, every part of which is functioning optimally, rather than one that has been in space hibernating for up to six years and which has already been taxed by a long-endurance deep space mission.

A crewed deep space mission should take advantage of the flexibility afforded by a generous mass budget, rather than making itself a barebones, shoestring affair.

Finally, making explicit provision for producing more DST modules, rather than creating one bespoke product, would allow us to eventually expand to conduct more than one deep space mission at once. With each mission requiring only a small number of launches of the SLS (and possibly pieces of such missions launched by other rockets), we might be able to carry out multiple deep space missions simultaneously once the SLS launch cadence is established and once the development expense of this overall architecture is behind us.

We shouldn’t design the DST to be disposable, and we should plan for it to return to cislunar space under power and be refitted and reused. But we also should make sure the DST is an industrial product produced on an ongoing basis, not a bespoke artifact made once and never again.

5. These mass budgets are plausible, but are they optimal? Could changing the architecture in some of these ways enable an increased mass budget on the same IMLEO which would make the mission better?

The current architecture is based on the notion that a monolithic 41-ton DST, outfitted with one 10-ton CMP, a 41-ton logistics and refueling package, and another 10-ton CMP, is the correct package for a deep space mission. But given the rather odd set of constraints into which NASA has shoehorned itself in formulating this launch manifest, it is fair to ask whether this mass budget of about 60 to 100 tons Earth departure mass was formulated for the purposes of carrying out the mission, or for the purposes of meeting the restrictions imposed by their architecture choices. Perhaps 150 tons is a better figure, or maybe 200 tons.

Perhaps the crew should have access to two redundant habitat modules with an airlock between them, to address critical failures. Maybe there is a need for excursion craft (e.g. Phobos and Deimos landers) which aren’t mass budgeted. Perhaps heavy scientific experiments (e.g. ISRU experiments, robotic Mars landers for teleoperation, communication satellites, telescopes) are warranted which aren’t mass budgeted.

A crewed deep space mission should take advantage of the flexibility afforded by a generous mass budget, rather than making itself a barebones, shoestring affair. This may both enhance the impact the mission will offer, and allow monetary economies, because not every possible opportunity to save weight by spending money will need to be taken up.

Summing up: a different SLS exploration plan

So, put it all together, and what do we have? If NASA were to accept these recommendations and reconfigure the SLS manifest along these lines, it would reorganize its crewed deep space efforts along the following lines:

1. Immediately cancel the Deep Space Gateway and all plans to use the SLS to assemble deep space infrastructure with ten-ton co-manifested modules around the Moon. Don’t waste time or money on using a hundred-ton rocket to launch ten-ton modules.

The current proposal for the SLS launch manifest is somewhat incoherent, because it does not play to the strengths of the SLS and Orion system that it uses to accomplish crewed exploration, and does not make heavy enough use of the major R&D elements to which it does commit.

2. Immediately start an intensive program of accelerated research and development to push the current long-term plans for multi-hundred-kilowatt SEP, and/or space nuclear reactors in the multi-hundred-kilowatt range, toward space readiness as fast as possible. The goal should be to have such a system deployed for an all-up uncrewed test flight lifting heavy cargo to the Moon in the mid-2020s. This would include either the firm design requirement that such a system be fueled with argon gas, or a firm decision to initiate procurement for large amounts of xenon gas (e.g. 20 tons per year) on an ongoing basis beginning in the 2020s. This is undoubtedly a serious development challenge, but with 15 years remaining until the target departure window, it is tractable. In fact, with both VASIMR and NEXT getting closer to the right power, efficiency, and duration figures, electrical power will probably play a larger role in the process than propulsion itself.

3. Immediately begin a serious design for a large, monolithic DST that uses the volume enabled by the 8.4-meter fairing of the cargo variant of the SLS, and is in the 80-ton range, inclusive of its power and electric engines but not of fuel. Plan to launch an early prototype of this DST for a long-endurance crewed LEO test in the mid-2020s.

4. As these efforts begin to mature, begin serious planning for what the science payload of early deep space missions will look like, and which missions should be undertaken. This might include crewed landers for low-gravity bodies (Phobos, Deimos, Ceres, near Earth objects). It might include both heavy instruments and robotic telepresence landers.

The SLS launch manifest would look somewhat different under this plan. It would include, in the early-to-mid-2020s timeframe, one or more independent test flights along each of the following three lines of research and development:

1. Short-duration test flights for Orion and its service module in lunar space, similar to the current EM-2 proposal.

2. Long-duration ECLSS trials with early-draft DST habitation systems in LEO, with the DST launched by SLS Cargo flights and the crew launched on commercial crew vehicles such as the Dragon v2 and Boeing CST-100.

3. Uncrewed test flights of progressively more powerful solar and nuclear electric propulsion systems, to eventually include a full-scale ascent of 80 tons to the Moon under electric power.

These three lines of development would then converge, in the late 2020s or early 2030s, in the initiation of crewed deep space missions uniting one or two heavy SLS cargos sent from LEO to lunar space by means of electric propulsion with an Orion spacecraft sent directly to the Moon with the SLS, without the involvement of other on-orbit infrastructure. These missions could include a variety of targets, including Mars, Venus, NEOs, and main belt asteroids like Ceres, Vesta, Pallas, and Psyche, within the same general parameters, and utilizing the same core DST habitation and propulsion module designs.

The current proposal for the SLS launch manifest is somewhat incoherent, because it does not play to the strengths of the SLS and Orion system that it uses to accomplish crewed exploration, and does not make heavy enough use of the major R&D elements to which it does commit. A similar proposal along the same lines, using the strengths of the SLS, Orion, and heavy electric propulsion without trying to use them for purposes for which they aren’t suited, would equip the space agency to accomplish much more capable crewed deep space missions, with greater crew safety, in the same time frame.