Part of my series on misconceptions in space journalism.
About a year ago I wrote a blog on lunar exploration architectures. Like the articles written since, I was motivated to convert my confusion and frustration on the subject into a more positive opportunity to explore and explain. The Mars exploration counterpart has been overdue.
Like Lunar exploration concepts, there are many different approaches and I think it’s a fair starting point to assume that they are all locally optimal given their own set of (rarely explicitly articulated) assumptions and goals. This blog is a good faith exercise in reverse engineering these concepts and practicing backwards reasoning.
Let’s begin by reiterating some introduction from the Lunar architectures blog.
Design convergence and divergence as a sign of product sector maturity
When considering mature engineered products we often take design convergence for granted. For example, most modern SUVs, smart phones, and jet planes look basically the same. This wasn’t always the case. Whenever a new kind of thing is invented, it typically takes a few iterations for the customer and engineer to even understand what exactly it is that they want.
An instructive example is in the emerging electric car space. Starting from the same set of axioms:
– Batteries are very expensive
– Electric car drive trains are underdeveloped
– Customers will want good value
A typical electric car c. 2005 was a golf cart. The founders of Tesla Motors realized that exactly the same set of axioms could derive a product which, if anything, delivered better value and was the Roadster.
Crucially, only one of these visions represented a product which could grow the market segment!
Space exploration is still an immature area in terms of product development, so there is a natural degree of divergence in architecture and vision. I don’t have a time machine so I can’t predict exactly which architecture or vision will win out, but I can at least demonstrate a process that gives rise to each vision.
Typical space program requirements
The overarching requirement of any large scale, government funded space exploration mission is “Program success on budget and on schedule.” Of course, any project that costs more than a billion dollars, in any industry, has a 50-90% chance of never being delivered. For more in this vein, check out Edward Merrow’s tome on the subject.
To deliver program success on budget and schedule, project managers develop a common sense list of subsidiary requirements, such as:
– Minimize cost
– Minimize weight
– Minimize fuel
– Minimize risk to astronauts
– Minimize programmatic risk (e.g. getting canceled)
– Emplace infrastructure for a future series of missions
– Avoid technological dead ends (e.g. expendable vs reusable rockets)
– Avoid cost of new rocket (to avoid sticker shock, since rocket development is considered to be expensive)
– Placate constituencies (to avoid sniping)
Some of these requirements are easier to measure. Others are effectively beyond the project manager’s control. Still others give very different answers depending on the order of priority, such as the electric car example above. In a sentence, this varied order and weighting of common sense requirements is what gives rise to such different architectures in an immature product space.
Δv in the Earth-Mars space
Δv is the most important measure of “distance” when it comes to launching missions into space. It measures the change in velocity, and therefore fuel, required to travel from one place to another. The actual physical distance is relatively unimportant.
The hardest part of any mission is leaving Earth, which requires a Δv of 9.3 km/s just to get into low Earth orbit (LEO). LEO to escape (C3 in fancy parlance) is an additional 3.4 km/s. From here one final kick is required to go to another planet, and that is given by the graph above. Generally for Mars we say that during the launch window we can get to Mars with a minimum of 0.6 km/s, a relatively small addition on top of the launch requirement! In practice, things are a little more complicated.
The graph is a “pork chop plot”, so named because the shapes of the launch windows resemble a pork chop. The horizontal axis shows launch date, while the vertical axis shows time of flight. Thus we see that in August 2020 there’s a launch opportunity to Mars that takes 6 months or 18 months, and NASA is using it to launch the 2020 rover “Perseverance”. For Earth and Mars, we get another launch window roughly every 26 months. Launching outside this window is possible, but may require enormously more fuel and/or take a really long time to get there.
There are also launch windows for returning from Mars to Earth, and these are given in the lower half of the graph. A common question regards the narrow bands that cut through each launch window – these are due to orbital plane mismatches.
One final detail is that the pork chops for Earth-Mars have the additional requirement for “free return”, where a spacecraft that aborted landing would arrive back at Earth with no further fuel burns. This is a big nice-to-have for human missions, since the alternative is floating far from any planet for centuries in the event of an aborted landing.
Looking at different architectures
I’m going to categorize different exploration concepts chronologically so the reader can follow the evolution of thought on the topic.
Published in 1948, “The Mars Project” is the first technically comprehensive Mars exploration concept. The author, Werner von Braun, was the architect of the Nazi V2 rocket bomb and, by 1948, had settled in the US with many of his captured colleagues. Ten years later, the launch of Sputnik once again elevated his area of expertise and he became instrumental in the development of the Apollo program, which culminated in the 1969 Moon landing.
With a clear field and very little known about Mars, von Braun had the freedom to invent. His architecture called for 70 men in 10 spacecraft landing on the smooth(ish) polar caps in space planes, building a crawler to travel to the equator, and assembling a rocket for the return. Each spaceship would require over 5 million tons of hypergolic fuels.
The Mariner missions to Mars in the early 1960s discovered that the atmosphere was too thin and the polar ice too lumpy to land on skis. The Mars Project was a compelling point design showing that human Mars exploration was, in 1948, as technically feasible as a Lunar landing within a near-term technological horizon.
For those of us who love space history counterfactuals, there are few better explorations than Stephen Baxter’s book “Voyage”, which lays out how Apollo-derived architecture could be used to explore Mars. Here is one of a number of really lovely demonstrations in Kerbal Space Program.
The Apollo program was a “multi path” program in that nearly every technology was being developed in multiple, parallel paths. When it became clear one would work, the others were often stopped so effort could focus elsewhere. But exploring Mars would require a lot of these options to be started up again.
First up, soup up the puny and weak Saturn V rocket by strapping on some solid fuel boosters, which were actually tested! Next, replace the second and third stage J-2 engines with nuclear rockets. Assemble a spacecraft in orbit consisting of a bigger Skylab and four or five launches of just more fuel, plus some kind of landing system.
The landing system is a much larger version of the Apollo command module, designed to enter and land on Mars, and contains a Mars Ascent Vehicle. Numerous studies have been done on landers of this style and all suffer from packaging problems and low cargo weight.
When all is said and done, a mission like this can land 3 astronauts on Mars for a couple of weeks, and requires about six launches of an upgraded Saturn V. It moves the Mars Project about 20 years into the future and works in the universe that contains a Mars with its actual physical characteristics, but it doesn’t lend itself to extended missions or mass cargo delivery.
Space Exploration Initiative, or the “90 day report”
An initiative begun in 1989 by President George H.W. Bush, the SEI is the most comprehensive Space Shuttle-based Mars exploration plan. It proposed a comprehensive, “check all the boxes” exploration plan and had a price tag to match. Despite satisfying every constituency in the universe, no-one wanted to pay for it and it quickly died in the Clinton administration.
The SEI was focused on Lunar base construction, but the Mars exploration plan was very similar to the Apollo-derived architecture above, except that the loss of the Saturn V meant that instead of a mere five or six Saturn V launches, the SEI required on-orbit assembly at a purpose-built “Space Station Freedom”, which in turn would require hundreds if not thousands of Shuttle launches to deliver various components.
Part of the reason for the huge cost was that on-orbit assembly is enormously more expensive than on Earth, and modularity actually increases complexity and reduces margins.
So while an Apollo-derived mission might have cost $10-20b per mission, the SEI was initially costed at $500b, and we all know that that sort of money is reserved strictly for nation (un)building in the Middle East.
As explained in my blog on modular space stations, the fundamental problem here is that the Shuttle is the wrong tool for the job. No-one can bear the thought of throwing away this (at the time) shiny new rocket launch system, but if the mission is building Moon bases and Mars exploration ships, it would be cheaper to start again from scratch. Build a bigger rocket!
So far all the Mars missions architectures have shown us ways to spend impossibly huge sums of money getting a tiny number of humans to Mars for a few days. After the underwhelming delivery of the SEI in 1989, the underlying questions began to shift.
If we want to launch people to Mars quickly and cheaply, how? Why is it so expensive? What does a successful mission look like?
Prior Mars exploration architectures are expensive because they bring all the fuel from Earth. So all the fuel to return to Earth has to be brought from Earth. This is a scaling up of the Lunar orbit rendezvous architecture used in Apollo. Prior to this, a direct ascent architecture for the Moon had been studied because it avoided the complexity of docking in orbit!
Obviously, a direct launch to Mars and back with all the fuel produced on Earth would be even more prohibitive, so there’s a good reason to leave the fuel needed to get home in Mars orbit. But what if most of that mass could be made on Mars? Instead of being on the Δv summit, the Mars surface would be a new reset, since fuel can be made there from water and the ambient CO2 atmosphere.
Another thing Mars Direct does right is build a new, purpose-designed launch vehicle that looks a lot like the SLS, along with a high performance vacuum stage. It also places a crew of 4-6 on the surface for 500 days, not 14.
Mars Direct does a lot of things. It focuses on Mars surface science. It gets a crew of 6 on the surface for 500 days – a total of 3000 person surface days – with only two heavy lift launches. And it does so without space stations and other cruft. I don’t think it’s an exaggeration to say that it improves on the Mars time per dollar value by a factor of a thousand or so.
Mars Direct makes a lot of sense for exploration missions. Its blunt body landers, however, can never deliver more than about 10T of cargo to the surface and so they place a strict lower limit on cargo delivery cost. One expendable heavy lift launch per 10T of cargo is much cheaper than previous estimates, but the cost and timeline ramps up when trying to scale up to a base or city, and it doesn’t have an obvious path to improving this.
One other major weakness of Mars Direct is that the Earth Return Vehicle is really tiny for an 8 month journey through space. In many ways, getting humans from Mars back to Earth is a really tough problem in any of these architectures. While Earth is, gravitationally, much harder to launch from it has the advantage that the air is breathable and launch services exist.
Cyclers, studied and heavily promoted by Buzz Aldrin, offer potential savings in Mars exploration missions. The idea is to build a big heavy space station, but to fly it between Mars and Earth continually without having to refuel. At each end, a smaller, faster shuttle spacecraft docks with the cycler. In this way, the cycler provides living space and life support for the 8-12 month journey between the planets.
Cyclers are neat from a mathematical perspective. In particular, a cycler that’s able to use every launch window has to aggressively precess its argument of perihelion through a series of gravitational assists and maybe even aerobraking. All very cool!
On the other hand, they don’t solve several other problems. Cyclers require more Δv to reach and this high speed deep space docking becomes mission critical. They also don’t provide any intrinsic way to solve the Mars Entry, Descent, and Landing problem that is, in many ways, the hardest part of the mission. And they don’t solve the Mars Ascent Vehicle problem. Indeed, in all the prior architectures very little attention is given to the problem of getting back.
Many of my readers will have seen The Martian and may want to see its architecture referenced here. I am a huge fan of this book/film and wrote a technical commentary a few years ago. The author, Andy Weir, has explained in public that his architecture was his own technically literate attempt to solve some of the hard problems, which I describe in depth in my book on Mars exploration.
In summary, it contains a giant spaceship with nuclear electric propulsion (the Hermes), a Mars Descent Vehicle, a Mars Ascent Vehicle (MAV), and numerous smaller robotic cargo delivery landers of some kind. The book doesn’t give a perfectly consistent picture of any of these, but the descent vehicle is small, and gets the human crew to the base site without exceeding their G limits. The cargo landers deliver lots of cargo in small chunks, though presumably with a less comfortable ride. Somehow the much larger MAV is also landed nearby, and refuels itself much as in Mars Direct.
The Hermes is a lot like a cycler. It has a rotating section for gravity, but its propulsion system allows it to insert into Mars orbit, and to leave again when it’s time to go home.
Large space nuclear power systems are not physically impossibly but have never been built or flown. There is a common perception that public tolerance for nuclear-propelled rockets has been steadily diminishing since the demise of Project Orion. I personally think that nuclear rockets will have a renaissance but it will be a result, rather than a critical path enabler, of increased human exploration into deep space.
The Hermes is also a large modular space station with all attendant issues.
While The Martian provides a plausible technical architecture for 6 consecutive short duration surface missions it has no path to longer duration missions or mass cargo transport for city building.
Journey to Mars/Artemis
Where does that leave us in 2020? We have a robust and successful robotic exploration program but the human program is fairly precarious. In just the last year we’ve heard reports of a Lunar gateway-style space station in Mars orbit. A similar concept was briefly popular in the Obama years when exploration of Mars’ moons was considered as an attempt to meet the mandate to explore an asteroid. Like the Lunar gateway, it sidesteps the fundamental lack of the technology to actually land humans on the surface, and in a number of ways is a solution looking for a problem.
Various teams from the major aerospace contractors have been pushing concepts for Mars exploration but it all feels fairly half-hearted. Neither NASA, nor LM, nor Boeing really have a compelling vision for Mars exploration. Why? How? What? Part of this is a reaction to almost two decades of programmatic whiplash between the Moon and Mars, and part of it is a reaction to an organization that has gotten out of the habit of being rapidly effective.
Which brings us to SpaceX. I’ve written extensively about Starship before, so here I will focus instead on the architecture. SpaceX has a compelling vision for Mars exploration. They want to achieve not just a boot print on the surface, but a self-sufficient city. Looked at this way, many architectural choices slide into focus. The emphasis is no longer a desperate and daring sallying forth and retreat, but a sustained program to build capacity and deliver cargo. Less Hillary and Tenzing on Everest, more D-Day or the Berlin Airlift.
Clearly, SpaceX had to take Mars Direct’s 1000x improvements and run with them. In-situ resource utilization for fuel? Yes, and!
Recall that Mars Direct still requires expendable heavy lift launches to operate. In the usual plan, an SLS-like vehicle is required at a production rate of one per year. Instead, SpaceX has built a rocket factory in Texas that aims to build 50 Starships per year.
Recall that Mars Direct’s blunt body entry vehicles could deliver about 10T of cargo per lander. SpaceX is using a lifting body entry profile to increase this mass to 100T and beyond.
Recall that Mars Direct required the development of a high performance vacuum rocket stage to throw the payload to Mars? With 10x the cargo, even that will not do, so instead SpaceX will refill the Starship in LEO with about six more launches. That could get expensive, so both stages of the Starship will be fully and rapidly reusable.
Even if the Starships are flown hundreds of times, their initial construction costs have to be amortized, so they are being built using the cheapest manufacturing techniques. Welded stainless steel instead of aluminium or composites. If it’s good enough for the Centaur upper stage, it’s good enough for Starship.
In 1990, Zubrin and Baker estimated that Mars Direct could be developed for $20b and each mission flown for $2b. The current Starship price target is $5m per vehicle, and each can fly around 10x as much cargo as the Mars Direct plan. So while Mars Direct was a 1000x cheaper than its predecessors, SpaceX’s architecture is about 1000x cheaper again. Now we’re getting somewhere!
In summary, the SpaceX architecture calls for overwhelming, ongoing, and increasing cargo capacity at a low and decreasing cost, all under one development program requiring only two vehicles with system commonality and one kind of engine. If you can think of a way to make it simpler, faster, and cheaper, I’m sure they’d love to know!
I hope this summary has given the reader some context on the evolution of visions for Mars exploration. I look forward to revisiting it in 2030 and seeing what has changed.
As I concluded my analysis of Lunar exploration architectures:
Let us go forward into the future in a spirit of constructive dialogue, technical competency and results-focused optimism.