Eric Berger thinks it’ll be a Dragon XL:
SpaceX is already developing the XL, which is an expendable, cargo-only version of Dragon meant for the lunar gateway. So it’s not totally implausible.
I still lean toward the plain Dragon, though. There’s enough room in the trunk for the propellant.
Dragon has the SuperDraco thrusters, whereas I expect the XL does not. These are typically only used for launch escape–but the ISS will need a significant push! It weighs around 400,000 kg and needs 60 m/s of delta-V to deorbit. I don’t know what timescale they’re looking at–but if they want a reentry into a specific location, I think it needs to be a relatively small part of an orbit. Let’s say a 10 minute burn (an orbit is 90 minutes). That’s 0.1 m/s^2 (1% of a gee), and 40 kN of force.
A SuperDraco has 71 kN of thrust but can throttle down to 20%. And I’m assuming they’ll want to fire two for symmetry. Two of them throttled to 28% gives the required 40 kN force. The standard Draco maneuvering thrusters are only 400 N each–you’d need 100 of them! Even if I’m pretty far off on the timescale, it’s still not enough.
One sorta funny bit from the article:
The bidding process for the US Deorbit Vehicle was opaque, but there are a few intriguing clues. Initially, the contract was offered as a hybrid. NASA’s original documents said the “design” portion of the contract would be cost-plus and the development portion firm-fixed-price. Then a couple of things happened. Perhaps because there were not that many bidders (one source suggested to Ars that SpaceX did not even bid initially), NASA modified the process to allow flexibility on the contracting mechanism.
…
And notably, the award is entirely based on a firm-fixed-price contract, which is SpaceX’s preferred way of working with NASA.
So Boeing refuses to bid anything as firm-fixed-price, whereas SpaceX refuses to bid anything except firm-fixed-price.
Boeing of course loves cost-plus. It requires an army of accountants to verify that every nut and bolt is accounted for and they aren’t overcharging the government by a penny. And a smaller army of accountants to keep track of the first army. Of course, all of this gets charged to the government.
SpaceX seems unwilling to increase the number of beancounters to support bidding on cost-plus.