On modular space stations

 A lot of people brought up specific advantages of the ISS and limitations of Mir in this response to this tweet by @deltaIV9250 questioning ISS architectural choices and wondering about a more Mir-like station. But a modular space station does not have to look like Mir or have the same limitations as Mir.

Imagine for a second a 8.4m long x 4.2m diameter module (~Destiny-sized). Its barrel section will have a surface area of ~110m2.

Say it has a 100m2[1] solar panel wrapped most of the way around it and deployed in orbit off to one side. At a modest 15% panel efficiency it'd have a peak power of >20kW, and even w/ single-axis tracking[2] and factoring in degradation over time and losses from battery roundtrip, it could average >5kW.

[1]This is slightly less than the solar panel area of a single Starlink mini v2 satellite, btw -- it's not going to be super heavy -- less than half a ton.

[2]One option is vehicle roll w/ panels showing slim profile in flight direction; another is single axis swivel. 1/pi average solar exposure.

--

In the meantime, with high emissivity & high solar reflectivity white paint, just ~50m2 of conformal radiator surface[3] is enough to dissipate upwards of 8kW at below room temp (no heat pump, just coolant circulation).

[3]The best location for radiator surface is on the side facing away from the Earth so you have minimal exposure to IR and reflected sunlight coming from the Earth. You'll have IR sources like the solar panels and nearby modules in the configuration described below, but the solar panels can be thermally isolated w/ a relatively short and lightweight MLI panel if needed (segmented or flexible -- folded or wrapped <1/6 of the way around the module for launch) and the interference from such reflectors and nearby modules can be conservatively approximated by modeling the radiators as a 4x8m surface facing the sun. At a steady 280K and 1/pi average solar expsosure, a coat of PSBN (w/ 0.15 solar absorptance, 0.92 thermal emissivity -- https://acktar.com/thermal-control-coatings/) will average 256W/m2 of net dissipation, or ~8.2kW for the flat 32m2 approximation of our ~50m2 curved panel.

--

Now, suppose this module has indepedent propulsion (ion engines for high delta v + cold gas for docking/undocking) and 3 extendable large diameter docking ports -- one on each end and one on the side opposite from where the solar panel(s) attach. A even number of these can then self-assemble into a double barrel structure (2 modules wide, many modules long) w/ solar panels on either side, none shading another, with the whole structure presenting a slim profile in the direction of flight, and with the ability to isolate any single module while retaining the ability to traverse the full station.

And with docking ports that are extendable & retractable[4], you can remove/reposition/replace any given module independently w/o disconnecting other modules. So not only does the station not need any EVAs for assembly, you also have a path to keeping the entire station fresh indefinitely w/ module replacements without any EVAs or bifurcation.

[4]Flexible material between the port and the main body section w/ tension lines and servos (substantially easier than whole inflatable modules). This would come at the expense of rigidity, but it'd be possible to more than make up for the loss of rigidity by using several docking bolts near the outer diameter of the modules. These docking bolts could also be used unaltered with the unpressurized modules. 

--

To get ISS-level power, you'd need 14-18 such modules. Make it 18-20 for good measure. Some of these you'll want to be unpressurized for external experiments (they can still share much of the bones of a standard pressurized module), and all gathered on one end (ideally the forward end, so vehicles docking at the aft end are shielded by the station from the worst of MMOD). Say you have 14 pressurized and 4-6 unpressurized modules. That gives you a pressurized volume of >1400m3 (more than the ISS) and a pressurized length of 60m (less than the ISS for easy movement), and a total span of ~35x85m -- more compact than the ISS. And with slim cross section in the direction of flight, it'd have far less drag area, as well as lower MMOD exposure. This would make a potentially lower operating altitude more feasible (and more stable), as well, which would mean lower energy cargo and crew transits.

Destiny was <15t. You could add the required components for ~5t, conservatively -- 0.5t solar panels, 0.5t batteries[5] & converters, ~1t propulsion[6], 1t radiators and pumps[7], leaving a generous 2t[8] for the third docking port and retractability mods for all 3 ports. This would keep these modules within the single launch capacity of the Delta IV Heavy and Proton, and it would have even be possible to keep them within the single launch capacities of Atlas V, Ariane 5, H-IIB, the Space Shuttle, and Falcon 9R, especially if a lower inclination orbit (excluding Russia) had been chosen[9]. So, <400t for the whole station (somewhat less than ISS), more power and more volume than the ISS, with 20 deployment flights w/ LV flexibility and no EVAs for assembly. Compare to 27 Shuttle + 2 Proton + 2 Soyuz assembly flights before Shuttle retirement (pre-Nauka).

[5]100kWh per module gets you 2000 cycles over ~10 years at ~80% average power; usable well past that, also replaceable on a shorter schedule if smaller batteries are desired. Some have suggested that a distributed power architecture would pose a major power distribution challenge. This is not true; in fact, I'd argue the opposite is the case. Average transmission path will be shorter, which means you can use lighter gauge wiring or lower voltage. And hooking up one or more common buses is not really a greater difficulty than hooking up power from a central source to each module. And if we want to avoid the complexity and inaccessibility of an automated power connection in the docking mechanism, it'd be possible to electrically connect modules after docking with a power cord hooked up across each doorway, manually, internally, easily, and mostly out of the way (mostly behind a conduit). 

[6]400kg of 2500s isp pro will buy a 20t module ~500m/s of delta v -- 25m/s of stationkeeping budget (way more than needed with this design at 400km altitude) for 20 years, <400W power consumption at 50% efficiency; 15 degree thruster angle so modules can fire without getting in each other's way would only be ~3.5% cosine losses. 1t can cover argon or krypton tankage (forget xenon), electric thrusters, and cold gas COPVs and thrusters. CMGs could be kept in the unpressurized section with indepedent module flight relying on thruster-based attitude control.

[7]16kg/m2 over 50m2 of radiator area would leave ~200kg for pumps and reservoirs and is fairly conservative; for comparison, ISS's radiator assemblies each weigh 1082kg including deployment mechanisms and base structure and have a double-sided radiator area of 139.5m2 (https://www.jstor.org/stable/44612020) -- 15.5kg/m2 for single-sided area including deployment mechanism.
And it might actually be preferable to just circulate cabin air through channels running near the outer wall, with many redundant loops that would allow shutting off sections if/when leaks develop. A 10K temperature drop with a 8kW load would need about 0.6m3/s of air flow -- roughly the equivalent of 20 computer fans or one ceiling fan. That temperature drop would also allow keeping relative humidity <60% without active cooling and would fit with the 280K radiator temperature. This would architecturally be the equivalent of moving the internal heat exchangers and water separators to the outside of the main cabin and deleting all loops with liquid coolants. At these short distances, aluminum will make for a lightweight heat sink between the air or coolant channels and the outer wall. 1kg/m2 of aluminum could carry the average heat flux with a temperature drop of ~1K for 2cm of separation from the channels to the outer wall, which would then serve as both radiator area and the bumper layer of a whipple shield. (Distributed architectures can make simpler solutions possible...)
For situations where the thermal load is greater than 8kW for a particular module, air circulation between modules would be adequate to make up the difference (can just circulate air through the station with well-placed fans -- don't need special fluid hoses and connections).

[8] What about other components of the life support system, you might ask. Destiny came with a good chunk of life support -- it was one of international side modules with CO2 scrubbers. And later, two racks were temporarily added (later moved to another module) for water reclamation. One could retain distributed life support systems, or try to have every module have fully independent life support with smaller units. Either way, we're not looking at much of a change in life support mass or volume per module. The equipment isn't very large (relative to a module), and much of it could scale down reasonably well.

[9]Even though the Shuttle was limited to 16t of payload to ISS orbit (16t is still feasible, btw, for this module size), going to a lower inclination would have pushed that over 20t, and lower deployment altitude (from where EP could raise to station orbit) could have been used to increase payload capacity even further. You could have still had mixed crews on two stations, btw, even if a US station had been kept separate from a Russian replacement for the Mir.

--

You might ask, what about the impact volume of impact of adding all that hardware? The radiators and solar panels can be added to the exterior w/ little to no impact on pressurized volume, while keeping the modules below the 4.6m payload envelope[10] diameter of most, if not all, of the LVs listed above, including the Shuttle and Delta IV. The rest of the equipment (by bulk, mainly COPVs for electric and cold gas propulsion + a relatively small volume of batteries) could fit within a single digit number of cubic meters (even w/ inefficient space usage that left dead space around spherical tanks) -- i.e. < 10% of pressurized volume.

Then there are operational questions, starting with where do visiting spacecraft dock? A few of the modules would have an APAS port (likely on the Earth-facing side), in addition to APAS adapters on the unused end ports of each of the two aft-end modules. The aft end would be the preferable location for long term lifeboats [11], while the Earth-facing ports would be ideal for cargo resupply and temporary use during crew transfers.

What about airlocks? The two unused ports on the forward end of the station would allow egress into the unpressurized section of the station, and either a part or the entirety of each of those forward-end pressurized modules could be used as an airlock. That location would also provide relatively convenient access to the top and bottom sides of the pressurized length of the station. The pressurized modules could ofc be fitted with handholds and tether anchor points.

What about an exterior robot arm? How about multiple robots that launch with and reside in the unpressurized modules, and have multiple arms so they can traverse the station using handholds/anchor points while having arms free for work. They'd come back to their storage area to charge their batteries. 5-10kWh worth of batteries could buy a lot of useful work in 0g. But you can also pack more like 100kWh per robot if you really want.

[10]Having extendable ports would allow the side ports to be recessed for launch and fit within the barrel diameter.

[11]Dragon and Soyuz sized vehicles could fit side by side at the aft end, whereas it'd be a bit tight but workable for two Starliners (extra spacing provided by extendable docking ports between the two barrels, as well as mating adapters that provide an off-axis APAS port), while a docked Shuttle would tend to preclude a second vehicle there.