COA 1: the first crewed mission to Mars happens after a precursor ship has arrived and generated the propellant for the return trip. This isn't the SpaceX plan as presented, but is a more conservative approach, which is why we'll talk about it first.
First, this plan means the first big ships to go will be unmanned, and at least one will carry an ISRU propellant plant. It will need solar panels to power the plant, and robotic equipment to gather ice or water-impregnated soil to bring back to the ISRU plant. That also requires robotic means to unload and set up this cargo. Musk specifically called for two cargo ships going in the launch window prior to the manned mission, so in this COA, they might do more than preposition supplies. Two ships give available cargo roughly equivalent to what we would have had with the original Mars Colonial Transport--about 300 tons.
There are two possibilities here. The ISRU ship (or ships) could be purpose-built, or could reuse the developmental ships. The developmental ships will have a lot of mileage on them by this time, and life support for 24 people. They would probably have no cargo offload capability during development, so would need to be retrofitted. That's probably a small effort, but it this raises an interesting question. Does the ISRU plant need to be offloaded from the lander at all? For starters, we'll need a place to store the fuel and LOX; a Mars vehicle's tanks seem like they're about the right size... If we assume the first "big" ship carries the ISRU plant, then we'll need to take all the steps necessary to ensure its tanks can store the propellant long-term. This is important, since Musk emphasized cryogenic propellants don't agree with carbon fiber tanks. This might be the one case where tank liners are worth the cost and complexity. In 2016, Musk indicated there's a plan for this, using "invar" alloy. He followed up in his 2017 Reddit, indicating they'd figured out a way to prevent even LOX infiltrating the carbon fiber.
The need for tank liners could well be the determining factor whether a developmental ship makes a one-way trip to Mars, or whether a purpose-built ship will be needed. Either way, making it into a "refinery ship" means it won't be coming back to Earth any time soon. Refinery Ship 1 has a long-term mission on Mars.
Making the ISRU plants part of the ship allows them to be completely integrated and tested on Earth, without the need to disassemble them for transport and reassemble upon arrival. "Plants" is deliberately plural. Local propellant production enables everything else, so instead of one big plant, we really want two smaller ones on a ship. In addition to redundancy during unmanned operations, it provides an automatic 100% spare parts capacity once there are people to available to make repairs.
This is so critical that a second Refinery Ship will almost certainly follow within the next few launch windows, regardless of which COA we ultimately select. We're likely to discover some adjustment is desirable to the plant's design, and we'll need more refining and storage capacity once more than one ship travels per launch window. Splitting the ISRU plants into smaller components also allows experimenting with different performance adjustments, since a second set is available. For now, let's assume the ISRU plants are installed on a purpose-built freighter. Then, it seems reasonable that the other precursor cargo ship be a MCCS. The MCCS will carry all the other supplies we'll need, including backups for the cargo carried on the Refinery Ship. It will also have life support, which will operate for the two years before the next launch window. In that way, we have an opportunity to verify reliability, failure modes, etc. To add interest, how about bringing along some plants that we can monitor from Earth via camera? Maybe some trees that will eventually be planted in Martian soil? That gives the life support an immediate purpose, and adds a small emotional aspect, in addition to the science.
The next problem is, how will any of the ships be powered during their stay on Mars? The presentation showed solar panels while enroute. Will these be designed to extend while on the surface? That seems inherently dangerous for a reusable ship, since dust in the retraction mechanism could cause jams, and could even cause problems with re-extension once on the way home. Additionally, an extra extend/retract cycle on every mission seems likely to increase between-flight maintenance. The ISS has also shown how finicky large solar panels can be when it comes to retraction. This isn't a real problem for the Refinery ship, but Musk's comment about a "solar panel field" powering the ISRU plant suggests extending the in-space panels isn't the plan.
Also, if the propellant plant isn't part of the ship, as in the SpaceX presentations, then it needs that external field anyway. How does that get deployed?
Both of these questions lead to a third design parameter, the size of the ISRU plants. There will need to be solar panels for power, deployment mechanisms, etc. Those take some fraction of the available cargo mass. How much is needed? And, by extension, how much of the cargo mass remains available for the ISRU plants?
At this point, our answer will be completely arbitrary, just because we need to start sizing things. Let's say the ISRU plants take up about two-thirds of the available 150 ton capacity of the Refinery Ship. That's 50 tons each, leaving the remaining 50 tons available for supporting cargo, plus 150 tons aboard the MCCS. As we go through the other various COAs, we'll keep the ISRU plants sized at 50 tons each, to make COA comparison simpler. We may quickly discover this size doesn't work, but too many variables makes it impossible to solve our simultaneous equations.
So, what does our available 200 tons consist of?
Using robotic deployment, we'll need a crane on the ships to unload the equipment, a rover to carry the solar panels, and a rover with some type of arm to set up the panels. In addition, we need to figure out how to mine water, which means breaking up soil and loading it into some sort of transport. That implies at least some sort of "jackhammer" on a rover--we see those as a back hoe attachment--a "front end loader," a "bulldozer," and a "dump truck." We'll probably need two of each rover type and two cranes, for redundancy--if we can't unload, set up, and maintain the solar panel field, the ship dies. We also need to figure out how to find water-bearing minerals, mine them, and refine them--autonomously. The bulldozer/dump truck seem simple enough, given the availability of self-driving cars, but unloading from a cargo bay 30 meters off the ground, power field setup, maintenance, setting up a mechanism to get ice/soil 30 meters up to the ISRU plant, actually finding water-bearing soil to break up, and maintaining the entire operation autonomously for at least two years... At first blush, this seems problematic.
This also puts into question the whole SpaceX idea of sending multiple cargo ships to Mars first. Almost every presentation has simply assumed this will happen. But, that just expands the problem of autonomous robotic operations, in addition to increasing the up-front cost of sending ships that can't yet get home. And, as episode 1 of NatGeo's Mars dramatizes, prepositioned supplies aren't necessarily something the crewed ship should count on--what if a malfunction causes you to miss the landing zone? Further research may show this approach is completely feasible, but at this point, there are a lot of questions. We should at least consider whether there are other alternatives.