Even before the 2017 IAC presentation, the need for a mostly-cargo version of MCT/BFR was clear. At the very least, the first few ships might as well be designed to carry a couple dozen people and a lot of cargo. They will probably spend their whole lives in that capacity.
We've established our mission, but it bears repeating: before large groups of people can go to Mars, there needs to be someplace to receive them. Musk's 2016 presentation illustrates that pretty well--the hatches open, the colonists look out on Mars vistas--but those vistas are barren. I couldn't help but think, "I hope they remembered to build a base before they sent 100 people." My guess is that the ship will actually position its hatch so the passengers look out on a natural vista, mostly because early the early outpost won't have much capability to make things pretty. Think Antarctic base and lose some of the dreamy vision. So, let's assume the camera pans around to reveal the outpost we're going to design here.
To keep it simple, even assuming a colonist doesn't mass more than 100 kilograms, they're still going to need a lot more than 1.4 tons of support equipment each if they're going to set up shop on Mars. Shotwell also confirmed this thought in 2014, without elaborating. The Mayflower's 102 passengers brought only 180 tons of provisions, but they didn't need to worry about air.
What will be needed? Let's start a list:
--Equipment to make oxygen
--Equipment to scrub carbon dioxide
--A pressurized habitat
--Equipment to recycle water
--Equipment to obtain water (for crops and human consumption, and for making fuel)
--To become self-sufficient, enough pressurized volume to grow crops. It takes about an acre to feed a person on Earth. Martian soil is closer to the Sahara Desert than a Nebraska cornfield, but the colonists will have 100% control over their crops--water supply, fertilization, bacteria, etc. Well, they'll have that control once the water, fertilizer, bacteria, etc. are available, anyway.
--Until we're self-sufficient, lots of food. Assuming supply ships every two years, an initial thought is there should always be enough food to handle the loss of the supply ships for an entire cycle. (I assume you saw "The Martian.")
--A place to take shelter from solar radiation storms. That poses some interesting problems for those crops.
--Equipment to produce electricity.
--Equipment to repair everything from shelters to spacesuits to rovers. The colonists will need an awesome set of tools. (Bonus points if you get the reference.) Initially, this will require lots of spare parts; eventually, we'll want equipment to make the simpler parts. 3D printing and 3D metal printing leap to mind; both require feedstock.
--Cranes mounted on rovers to move the big stuff.
--We'll also need a propellant plant, so cargo ships can make the return trip to Earth, critical for them to bring more supplies.
That's a lot of cargo. Let's talk about that last one first.
Ultimately, this whole process doesn't work if the ships can't get back to Earth. So, one of the early things we need is in-situ-resource utilization (ISRU) to make propellant (oxygen and methane, both of which will need to be cooled to liquid form). Just considering safety, this probably has to be the first thing we need.
It's hard to imagine a manned mission to Mars that doesn't have the ability to get back if something goes wrong. SpaceX has based its architecture on orbital refueling, reusability, and ISRU propellant manufacturing on Mars. So, the big question to be asked is whether a crewed vehicle could really go to Mars before its return propellant supply is available. In the 2017 IAC presentation, Musk says "yes." To be blunt, I'm not yet convinced. This is key, because the answer will drive our early architecture more than any other. And this isn't only a SpaceX decision. If the US government thinks SpaceX is going to create a serious long term problem (read: a bunch of dead astronauts/colonists), they'll simply deny the launch license. But regardless, even if the ship goes to Mars without return propellant being available, making it available will be a top priority.
Ultimately, we (or SpaceX) may decide we want option "A," but realize that for a variety of reasons, we must settle for option "B." In the planning business, those are alternative courses of action (COAs). We need to identify which COAs are acceptable from a safety standpoint and feasible from a technical standpoint.
We need to identify the COAs available and carefully examine each in turn. But, we should also establish some design parameters that will probably hold true across all COAs. That will let us focus the analysis on the important differences. (This is also a basic planning premise.)
First, it looks very much like there will effectively be four types of ship, not just three. Tanker, cargo, and manned versions were identified in Musk's presentation, but the manned versions will likely be "mostly cargo" in the first edition. These will probably have been the developmental ships anyway, so presumably, only enough life support was installed for proving system reliability. (Those lunar orbital trips.) Assuming a crew size of 12 on each of 2 ships, we specify life support on each for 24--enought to sustain the entire crew of both ships. The exact number entirely arbitrary, but this makes the math simpler, matches Musk's comments, and is high enough that a variety of equipment can be tested out. We call these "mostly cargo" ships to distinguish them from later "mostly passenger" ships expected later. These are different than the unmanned tankers or unmanned cargo ships.
Let's give the various "ships" different class names, for convenience. Calling everything "BFR" is going to get confusing, so let's just borrow from earlier terms. To identify the unmanned cargo version, let's call it the Mars Freighter (MFR). That acronym might also have an amusing alternate meaning. The original mostly-passenger version was the Mars Colonial Transport, MCT. So, let's make the new manned-but-mostly-cargo versions the Mars Colonial Cargo Ship, MCCS. (Better options welcomed.)
Round numbers, the "mostly cargo" option gives us somewhere around 150 tons to Mars. It appears the primary difference between freighter and crewed ships is what's carried in the "passenger" area. For freighters, that will include lower density stuff. It's clear that if we have access to a big ship, the first few need to be freighters. (Musk and Shotwell agree.) That will affect the overall program cost, since they will never generate revenue from passengers, but if we assume the first couple of ships were developmental, then they simply become the cost of doing business. For most new airliners, the first few are never intended for commercial delivery. They prove out the design and are then parked, put in a museum, or spend their careers in specialty service for the manufacturer.
Let's also establish that even for crewed ships, we're willing to accept a slower trip for more cargo, as long as we're only taking a couple dozen people. If the ship is going unmanned, it should take the lowest energy trajectory available, to maximize cargo.
The "people-versus-cargo" question is largely one of volume, not mass. Realistically, freighter and the mostly-cargo MCCS versions of the ship will carry more cargo than the later MCT simply because we'll accept a longer travel time, but we'll also be able to utilize the pressurized volume of the personnel section. That's good, because it means the design changes required are minimal to accommodate high-volume cargo. If you've ever seen a cargo version of an airliner, you get the idea. Take out the seats, roll in the cargo. (If you haven't seen one I must recommend an airshow. Or Bing. Try "747 freighter.")
The second design parameter is safety margin. Standard parameters for human-rated systems specify a 2.0 safety factor. There's no compelling reason to go below this, and plenty of reasons to go above it, where practical. This is incredibly dangerous stuff. Our architecture needs to ensure that problems don't become tragedies, and tragedies don't become catastrophes. No one in Europe knew or cared that half the Mayflower's passengers and crew died the first winter. That won't be the case here. However, this constraint will mostly drive the order of systems, equipment, and people arriving. We can get the safety factor through redundancy and spare capacity versus over-engineering.
Okay, that should get us started. Let's sketch out some COAs.