In my talk on the commercial space panel at the Mars Society Convention on Saturday afternoon, one of the things I said that we needed was “gas stations on the road to Mars.”

I’d never really thought about it much, because I don’t think about Mars much, but I wondered just how advantageous it was, so I decided to take a break from the space safety stuff and play with the idea in a spreadsheet. I haven’t triple checked the numbers yet, but what I found blew me away. For a hundred metric tons of payload delivered to Mars orbit in the vicinity of Phobos, I looked at three mission scenarios: departure from LEO direct to Mars, departure from EML1 direct to Mars, and departure from EML1 with a fuel stop half way. For the latter scenario, you can deliver a ton of payload to Phobos’s orbit for less than a ton of propellant. For the EML1 scenario without refueling, it takes over ten times as much. For the LEO scenario, it takes over twenty times as much. That is, for the refueling scenario, you start with an IMEML1 (Initial Mass in EML1) of about a hundred and fifty tons, two thirds of which is payload bound to Mars (the total mission propellant needed is twice as much as you start with, because you pick up the other half, about 43 tons, on the way). For the non-refueling case in EML1, the IMEML1 is over a million tonskilograms, and for the LEO case, IMLEO is almost two-and-a-half million tonskilograms.

Such is the power of the rocket equation, and refueling.

What’s the catch? There are two. First, of course, it assumes the delivery of the propellants to EML1 and the gas-station orbit. If it were done chemically, the savings would go away, but it could be done electrically. The other catch is that it obviously both complicates and increases the time of the mission. I’d recommend it for cargo only. The gas station orbit I chose was at a distance of 1.25 AU, which turns out to exactly split the delta V between earth departure and Mars arrival, which means that you just refill the tanks at the gas station. That orbit turns out to have a synodic period with earth of about three and a half years, and with Mars of about five and a half years. But with the savings, you could afford to put a dozen stations in that orbit, which would give you an opportunity every month, and then you could take more of the savings and do faster trajectories, so you don’t have to wait for Hohmann line ups.

If anyone wants to check my numbers, I ignored the moon’s gravity, just doing a departure from EML1 starting at EML1 velocity with patched conics (same thing going into Mars orbit to co-orbit with Phobos). I used a stage fraction (ratio of propellant to total stage weight) of 0.85 and an ISP of 460 seconds (lower ISPs would just make the numbers even worse for the non-refueling scenario). I also didn’t do any aerobraking at Mars, so the numbers would work the same in either direction. So if PR wants to sell water, that’s the location for the gas station — 1.25 AU.

[Update a few minutes later]

Just looking at the difference between LEO and EML1, it seems insane to not use propellant depots with electric propulsion.

[Update a while later]

Let me put this another way. This implies that you could deliver a hundred metric tons (that is, 80% of a Saturn V payload) to Mars orbit in four Falcon Heavy launches. Two launches for electric depots (one of which has cryo propulsion, probably RS-68 class if you can start it in space, or a bunch of RL-10s if you don’t mind the gravity loss) and two for the payload. Less than half a billion in transportation costs to LEO. Leaves a lot of money left over for the other stuff.

[Late afternoon update]

OK, I found an error in the spreadsheet. As I note at the new post, the propellant advantage is not as dramatic as I indicate above, but it is still significant.