I'm working something out in my head out loud, so this post is probably going to be a little high on noise side of the signal-to-noise ratio, even by my standards.
I'm pretty confident that I'm not going to run into plumbing issues. I'm also certain that any plumbing issues I run into in the future are more than addressed by terrestrial plumbing systems, with the occasional tweak to account for differences in the weight of water.
The relief of my thought experiments in Making Toilets Flush in Artificial Gravity is that flat decks are possible. With flat decks I can carry out the vision I had that this entire construction process is somewhere between assembling a ship and a skyscraper. (And believe it or not, the two are not that different.)
So with all of that resolved I can contemplate the design of the ship itself. Not, how it's built, but what it does. Carrying 3000 people (okay 1000 people and 2000 of their decedents) to another star system is a nice trick. But what are they going to do once they get there? Is this a scientific mission where we load the ship up with space probes, collect our data, and go home? Is this a mission to build a forward operating base for humanity to leap further out into the stars? Are these pioneers who are going to settle in this new star system permanently?
Or... is this mission going to be a mess of competing priorities.
One thing that won't happen is this ship stopping in another system, topping off the tank, and then taking off again to another system. While the ship could do that once or twice if need be, there is a limit to how far the ship will travel before it starts falling apart. Even with proper care and maintenance, steel structures only last between 50 years (a Nuclear Aircraft Carrier) to 100+ years (Steel Skyscaper.) And consulting some handy engineering tables we can see that there are critical components of the structure and infrastructure that have have design life well below 50 years. It's no good having a steel structure that lasts 200 years if the plumbing that makes it habitable starts springing leaks after 60.
So for our thought experiment, let is envision that the vessel has a design life of 100 years, with a major renovation every 25 years. And by "major rennovation" I mean strip the vessel back to the bare structure and rebuild all of the wiring, plumbing, walls, etc.
The problem is that the ship can take 20+ years to arrive at its destination. (At least 20+ years of ship time, more than a century could pass back home.) To make a return trip the ship will have to go through one of these mid-life revitalizations, while not necessarily upon arrival, certainly before starting the trip back.
So the "Arrive, plant a flag, launch a probe, and leave" sort of mission is immediately out of the question. This ship is going to have to build some critical infrastructure in the remote star system. A staging base where the crew can live outside of the ship while it's either being scrapped or renovated.
Renovating the ship with people living on board is a very very bad idea. Firstly, construction requires a lot of welding, painting, sawing through pipes and cables, and so on. The probability of a screw up causing a major power outage, flood, fire, chemical spill, or structural collapse is 100%. The ship also requires either thrusting or rotation to maintain gravity. You aren't going to be doing major renovations with the engines on. For rotation, the vessel needs to be balanced. If an accident drains a tank or dislodges a major chunk of structure the imbalance could lead to the entire ship shaking itself apart. Plus, construction could benefit from gravity being completely off. Which... is not very conducive to life on board.
So one of the considerations that is immediately apparent is that the ship needs to be as easy to disassemble as it was to assemble. It also needs to facilite the removal of the agricultural plots. For planted fields like wheat or vegetables or cotten, moving the field is as simple as slapping a cover on the dirt and transporting it via fork lift and elevator off the ship to where it would be placed in the colony. Trees and other perennials are a slightly different challenge. But one that could be overcome by cultivating them in pots, pruning them down to size before transport, and making allowances for height and width in your elevator and transport design.
Inhabited spaces will end up moving just like people move from one office to another, or one house to another. Take everything out, stuff it in a shipping container, transport that container to the destination, unpack.
Factories will vary in moving complexity by how large the capital equipment involved is. For a machine shop you just move the equipment like you were moving out of a house. Granted, a 1000 lb milling machine you would use more than a hand-truck to transport. For a large metal stamping plant, or a large industrial furnace you start to think about building them inside of a module that can be unbolted from the ship, towed with a tug ship, and bolted onto the colony. This approach would also work for fuel tanks and major ship infrastructure like the reactor room and the engines.
How much of the ship is modular, and how much is allowed to conform to the oddball structure of the ship (In-Situ) is a balancing act.
Modular construction is pretty wasteful on both space and material. A module has be be built far stronger than if it was simply incorporated into the ship's own structure. It has to withstand the rigors of transport. Modules also have to have their own plumbing and power systems. Surrounding areas of the ship need to be built around how that modules connects, disconnects. And you may lose useful volume to accomodate how the module is removed and replaced.
In-Situ structures are very efficient in materials and use of space. In-Situ structures can be tweaked to balance other external requirements. (Like, in the case of rotation, things like weight distribution.) But when if comes time to move a In-Situ structure your only option is tear it apart and scavenge what parts are still useful, and then build a new one in the new location.
Engine rooms and reactors make a pretty clear argument for modular construction. They work or do not work as a whole. You generally want to make their removal and replacement as simple as possible. They need to work on applications beyond the operations on the ship itself. Spare reactors on the ship may power your colony while it is getting started. If the vessel is no longer using its engines, they can serve as the propulsion systems for support craft.
Larger factories are also a pretty clear cut application of modular construction. A heavy water plant requires a lot of heavy equipment that needs to be aligned and balanced. The melting furnace for a steel plant also needs a lot of large components precisely aligned. Half of one of these plants isn't useful. And odds are the plumbing and electrical systems in one of these plants is going to age at about the same rate as the working components.
A kitchen, on the other hand, is a very poor application of modular construction. Appliances regularly wear out. Everything needs to be unbolted and removed periodically for cleaning. Most devices need to be level and hooked up to power, water, and/or gas. Precision alignment not required. Kitchens also need to be reconfigurable as the needs or tastes of the population shift.
It may sound like I'm quibbling, but there are very important design tradeoffs that correct answer depends very much on the answers to these questions I raise.
For instance: do we have a gravity ring, or do we just rotate the entire ship? There are pros and cons.
Rotating the entire ship makes construction simple. But it means that we have to commit to an external shipyard for maintenance. It is physically impossible to swap an engine, for instance, and keep the ship in balance. Stopping the rotation means turning off the gravity.
A gravity ring means that different parts of the ship can rotate, or not rotate, at different speeds. But they are a plumbing, wiring, and logistical transport nightmare. Also, because there are more moving parts, we limit the design life of the vessel down to the expected lifespan of the bearings that attach the gravity ring to the spine.
But how long do bearings last? We have two things going for us here actually. One is that the rotational speed of our gravity ring is extremely low (1.33 RPM). Most bearing calculations define anything less than 100 RPM as low speed. There are also bearing designs such as Tapered Roller Bearings that are specially designed to handle both radial AND axial loads.
Bearing lifetime is generally measured in how many revolutions it will perform before failure. For this calculation, let us do a worst case scenario for our bearing. Let us cap the design life of our ship at 50 years. In transit between systems, we would keep the gravity ring rolling at a low speed, just to keep it from ceasing after decades of disuse. Let us say, once around every hour, so 0.017 RPM. When the ship is not thrusting, the gravity ring must be in rotation to maintain our life systems. During those times it rotates at 1.33 rpm.
So 15 years at 0.017RPM + 35 years at 1.3 RPM gives us a design life of (RPM*YEARS*24*60*325.25)= 24617557.8 revolutions. I've been working the numbers from a specification sheets of industrial bearing makers. And it seems that they already manufacture bearings that would fit the requirements of the vessel. There are industrial applications where slinging millions of pounds of force is already a requirement. And for bearings in the range the ship would need, specs sheets more or less assume that with proper lubrication and keeping loads within the spec sheets, bearings such as these are expected to last for 90,000,000 revolutions. Which gives us a safety factor of 3.7.
(That information is buried in the footnotes on the bottom of this website.)
Actually... like my post on plumbing, I was expecting there to be some sort of mathematical drama. But... there isn't. We now have a harder cap on the design life of ships. But that design life is well into the plausibly useful range of 128.7 years at full rotation. With space journeys actually extending the life of the vessel.
128 years is plenty of time to travel to a distant system, establish a manufacturing and agricultural base, and even return to Sol in the same ship. And that 128 years is beyond the expected lifetime of the steel plating that makes on the structure of the ship itself. (Yes, the outside of the ship is exposed to the pristine vacuum of space, but the interior is still wet, and bathed in oxygen like on Earth.)
At this point we can just eliminate designs that require the entire vessel to be rotating at once. It is inevitable that at some point the vessel will need to halt rotation on the engine section while the engines themselves are off. Engine swaps and Refueling are two cases I can think of off of the bat. I am also realizing that certain industrial portions of the vessel will need to rotate independently of the habitat. The heavy water refinery, for instance, may benefit from operating at a much higher rotational speed for some phases of operation, but may need to halt rotation for others.
If we can design each of these rings to operate independently, we can eliminate the need for anything other than crew or bulk cargo to pass between them. That eliminates the exotic pipe and cable connectors that would have an extremely short life. The proviso is that we are going to have to invest in smaller fusion reactors to power each of the modules independently.
How will we fuel all of these reactor scattered around the ship? Actually that is pretty simple. The most stable way to store heavy water long term is in the form of ice. I estimate that between all of the reactors on the ship, the vessel will go through 121 metric tons of fuel per day. The mass ratio of Deuterium water to Lithium metal is 8:3. Lithium metal, while you have to keep it away from water, is fairly trivial to transport. And to transport the heavy ice, we can chop it into meter sized cubes (or whatever shape is most practical) and tug it around as if it was a meter sized crate of a solid material. Ice has the strength of concrete. You can be pretty rough with it.
Delivering the fuel is even easier. Each reactor could have a chute leading up to the spine of the vessel. Have some sort of cargo lifter lug the block of ice from fuel storage area to the top of the chute, and drop it in. Well... at least while the vessel is in rotation. When the vessel is in thrust, we will need some sort of ramrod to slide the blocks of ice down the chute horizontally. And we'll need to build in a ramp in that direction to keep the ice down.
I have in my head a vision that the ideal shape for storing and transporting cargo on the vessel will actually be an extruded hexagon. Like cubes, hexagons can line up next to one another and occupy 100% of the volume of a space. Unlike cubes, you can fit a larger volume though a round opening.
And... come to think of it, with the fuel storage area located at the top of the vessel, we don't even need an elevator! We just need something to start the block of ice moving. In thrust mode the accelleration of the ship would be sufficient. In rotation mode we would employ some sort of mechanical plunger. The ice or lithium would travel down a tube. The chute that is intending to recieve would deploy a piston/stopper/ramp device that would deflect the falling cargo.
Aside from the piston, the only "moving" part is the block of fuel. And if the size of the fuel pellet is designed to match the inner diameter of the shaft, friction will control the speed of the pellet. If a pellet gets stuck, no problem. Warm the inner walls of the shaft, and parts of the pellet in contact with the shaft will melt. The hexagon shape means that only 6 points are in contact with the wall. And if a cube crushes itself on contact with the piston? No worries, the debris is accelerated laterally towards the opening of the chute. For longer drops we add teeth into the shaft to control the spin that the puck picks up, as well as introduce friction to counter gravity. Those teeth will need to be retractable because with the engines NOT thrusting, we have the problem that our puck's only accelleration is at the top of the chute.
While we can control the direction of fuel flow, we still have the problem of other bulk cargo. We also have the problem of that during thrust we have to overcome gravity, and with no thrust, we have the problem of microgravity. One approach is to just ignore one of those states of operation. For instance, if the vessel is in thrust, we simply don't support moving bulk cargo between the rings. People can move up and down ladders. But if there is a need to haul something massive, that massive thing just better be where it needed to be in the first place. The habitat is designed to make transporting material between the spheres easy. There is ample room in the spheres to stage 20 years worth of material for construction, fabrication, and agriculture. The only system that needs regular infusions of matter are the power reactors, and we can feed them through gravity.
For transporting bulk cargo in zero-g, we can get by with an amply sized truck like device that moves along a geared track. You pack a container, the container is winched up or transported by elevator. The truck grabs the container, takes it to the ring it is destined for, and releases.
With all of that in mind, the rough design of the overall ship is emerging: