I certainly don't know why one would develop NTR and then not use it for the most intensive application either.
As for your use of e of pi's spreadsheet--I don't see how one can realistically compare the mission architectures without adding in a comparison of the relative costs of putting X tonnage at LEO versus L2 in the first place. Clearly, assuming no use of extraterrestrial resources for any of the components, the latter costs more than the former, especially factoring in the fact that maintaining a human labor force out beyond the Moon is going to be both more costly and more risky than doing the assembly in LEO.
None of the scenarios you are considering seem to rely on developing Luna as a resource source--and if they did, you'd have to factor in the cost of developing Lunar infrastructure. It might not be necessary to charge the full cost of a Moon base or three, mining and processing there, plus maglev or rocket launchers built of Lunar resources to the Mars mission alone, since presumably if this were done then there would be other uses to put the Lunar stuff to besides so the cost would be shared. But one can't assume this stuff is free either; not only does boosting ice or metal up from Luna to L2 cost on an ongoing basis, the huge investment up front has to be somewhat factored in.
Given your desire to see space weaponization, the lack of consideration of permanent extraterran colonies for these purposes seems a little strange--though I suppose it is a bit early for these dividends to be paying off even if you retcon them in.
I think your notion that space weapons are "inevitable" is very strange considering that we've pretty much avoided them hitherto OTL! I did note they were not, as e of pi was saying elsewhere, nuclear weapons, but rather kinetic impact systems. If anything that points all the more strongly to a Lunar source since all they'd need is raw mass, no need to find and refine fissionable ores. But a TL with space weapons being developed is fraught with all sorts of major deviations from OTL that complicate things a lot, whereas the example of OTL makes it clear that a ban on such systems is indeed sustainable. Well, I guess it is a legitimate question whether such a ban is sustainable when one is doing things like missions to Mars; such grand projects present the opportunity to cheat on weapons ban treaties by smuggling large systems out of sight hidden in the ostensibly peaceful launches. There are ways around that; the easiest one being using the ban to discourage the grand missions in the first place. But I don't think one can reasonably say the main reason we didn't do early Mars missions was fear of upsetting the balance of terror in space; if we had been determined to spend the money on a Mars shot or three then even in a tense Cold War period it would be quite possible for NASA to operate with transparency that made it evident we weren't sneaking up a battery of weapons. This would put some pressure on the Soviets or any third parties to be equally transparent.
Anyway, your choice, hope you deal with the consequences carefully.
So now then--given that the entire set of masses you ask us to compare are all launched from Earth ultimately, I suppose it is reasonable, given your stipulation that the choice to assemble and launch from L2 is made and fixed for all options, to compare between the L2 based options only. Whatever it costs to get all that junk up there in the first place is assumed to have been spent. Then of course it falls to us to look at the numbers, and pick the one with the lowest mass, since the cost of sending X kilograms to L2 is going to dominate the cost of the materials on Earth.
Sort of. Some materials are more problematic than others to handle! Hydrogen for instance is notoriously tricky to keep stored. Should we presume that at this early date, sufficient investment has been made in whatever it takes to assure that a given quantity of hydrogen can be stored indefinitely with no significant bleed-off, or shall we be a bit more realistic and figure that liquid hydrogen has a shelf life, that some of it must always be boiling off? In that case, getting your necessary hydrogen propellants to the assembled craft becomes a problem of timing; you have to wait to the last minute, and/or launch more hydrogen than you plan to use, the rest being a budget for boil-off while you wait to use the useful stuff.
Nuclear Thermal systems use a lot less reaction mass in total to get a given delta-V, but 100 percent of what they do use is hydrogen. (One could use other fluids, but the ISP advantage over chemical systems vanishes, leaving one with only liabilities). Thus, they wind up using more hydrogen for a given delta-V on a given mass than a good hydrogen-oxygen chemical rocket would use, the latter disposing of a much greater mass that is however mostly oxygen. I suppose that might be one reason you'd consider using chemical cryogenic rockets for the later burns in the mission, since these would be happening many months or even years after launch, and keeping hydrogen stored until then could be impossible. But if so, one can't use the highly efficient hydrogen-oxygen rockets the table assumes! The fuel would have to be something more long-term storable, like methane, and the ISP would be lowered accordingly.
So the all-nuclear option assumes that one can keep hydrogen stored with minimal losses for years. I don't think that's too unrealistic, especially given the use of the nuke reactor as a bimodal power source to drive active re-condensation cycles to rechill the hydrogen.
My feeling is, given that nuke rockets have been developed at all, and that the problem of storing their hydrogen reaction mass has been licked, then the nuclear option is the one to take all the way.
Given that is that due attention has been paid to minimizing unnecessary radiation exposure of the crew! Any nuclear plant of any kind is going to raise the total exposure somewhat; it is a question of whether that exposure is a large or small addition to the cumulative effects of cosmic ray and solar radiation exposure. In other contexts, I've thought it perhaps clever to use a nuclear booster to launch a craft on an interplanetary trajectory, then brake and return that engine for reuse (I guess that is the meaning of your "space tug" line items?) while the ship goes on separated from the hot core. Either it has another nuke rocket that has not been activated and so is not radiologically very "hot" yet, or else rely on aerobraking or other strategies at the destination. For this Mars mission, the latter is not an option; we'll need some kind of rocket to brake into Mars orbit, and then another one to get back to Earth from there.
Again, if perhaps we have not licked the hydrogen storage problem, those two could be chemical rockets using a fuel other than hydrogen, at a great cost in mission mass of course, while we use mass-intensive strategies to store hydrogen for the "boosters" that launch the thing from Earth orbit (whichever Earth orbit, counting L2 as such) and at the end of the mission intercept and brake the returning package.
But I suppose the problem of hydrogen storage can be solved and if so, nuclear all the way would be the way to go--from whichever base.
I suspect that a realistic Mars manned mission would include exploration and (for later missions anyway) possible in situ use of Mars's moons. Given our state of knowledge today, I'd focus on Deimos, since Phobos apparently is depleted of volatiles. These moonlets are so small that their surface gravities are quite feeble, so that a large sprawling interplanetary space ship can reasonably land on them with very little delta-V and thrust. This would bring the craft into direct contact with a substantial mass, thus cutting radiation exposure in half. The regolith of these bodies, I would guess, is very lightly compacted, and moving masses of it is relatively easy, so it would be possible to tunnel deeply into it, creating caverns in which 5 or ten meters of regolith "above" cuts the other half, creating low-radiation refuges. If Deimos at any rate has got extractable volatiles (and tunneling to locate them is relatively easy) one might be able to count on in situ replenishment of hydrogen for the return voyage! The first expedition obviously can't count on that, but verifying that this is the case can transform follow-up missions, and leave the infrastructure in place to support these missions. Given that you have these expeditions in the 1980s, presumably the relative merits of the two satellites would be less known, and a lot of time might be wasted on Phobos--though since it is closer to Mars, some investment there is justified anyway.
For Mars itself, I guess the IMIS plans you work with did not rely on any in situ uses whatsoever. But a carefully considered plan should I suspect at least plan on it for the purpose of launching people off of Mars's surface. Landing is not so difficult; I gather from previous conversations with e of pi that aerobraking can reliably bring any vehicle down to a few hundred meters/sec terminal speed, and modest rockets can accomplish the rest, so there is little need to carry a lot of fuel for that purpose. The trick is getting back up again! For that, it is well known now that a small mass investment of hydrogen can be transformed, using the carbon dioxide in the atmosphere, to a large mass of launch propellant.
The catch is, it takes power and time to do the conversion. In situ launch fuel requires a single craft that descends to be stuck on the surface a considerable time while the ascent fuel stocks are being processed. Another way to go would be to send down an unmanned lander, presumably the ascent craft, to start acquiring these stocks while the expedition concentrates on exploring and possibly developing the moons, then send down the manned lander for a period of Martian surface investigation before returning to the deep space craft, then launching back to Earth.
I have to ask, why is it necessary for the craft returning to Earth to be brought into Earth orbit, in LEO or at L2, at all? Would it not be possible to have a return capsule that aerobrakes and lands on Earth, a la Apollo, while any extra modules needed for habitation during the long flight back are simply abandoned to burn up in the atmosphere? If there is concern that Martian organisms might have contaminated these modules, then it would be possible with relatively little propellant to instead divert the disposable modules so they miss Earth and are catapulted into Solar orbit again. Of course if there are such concerns, allowing the manned capsule to return presumably with the samples from which such organisms have leaked is also a grave risk! Somehow though I doubt the major reason to try to keep the returning craft in orbit is based on this worry. If we resolve that the astronauts are going to return to Earth eventually, and that Terran labs want the samples, they have to come down somewhere sooner or later. The act of reentry will pretty well sterilize the exterior and the recovery team can take measures to keep the capsule contained and recover the crew to suitable isolation environments where they and their cargo can be examined very carefully.
One might want to recover the engines used to boost the return vehicle Earthward from Mars. Presuming these to be nuclear, it would be possible for the main manned craft to separate and the two sections to distance themselves from each other, then if the nuke engines have a reserve of hydrogen, they can rocket themselves alone into a recovery orbit around Earth, to be examined and refurbished eventually. Or they can simply be abandoned, again remaining in orbit around the Sun or crashed into the Moon.
I finally think you need to reexamine the whole question of launching from L2, or anyway show your work whereby you conclude that overall it is more economical than simply assembling and launching in LEO. You might be right about this, but I don't see it. I do realize that there are trajectories from LEO to L2 that use remarkably little delta-V. (This path is economical but slow; I would not recommend sending astronauts on it, but it would be excellent for mass cargo including of course supplies for human assembly workers and mission crews).
I have to wonder at your figure (well, e of pi's) of just 1000 m/sec delta-V from L2 to Mars! I do realize that Luna's orbital speed of 1000 m/sec is already 70 percent of escape velocity from the Earth/Moon system at Lunar distance, and that L2 has the same orbital period while being at a greater distance from Earth, and hence the additional impulse (plus extra to escape the Moon itself, but that will be a small addition) to escape L2 and with it Earth's pull entirely is going to be pretty low, a few hundred meters/sec. But by the same token of being almost out of Earth's potential well already, we don't get much of an Oberth effect benefit in achieving the additional delta-V needed to wind up on even a minimal Hohmann orbit to Mars, let alone the further oomph needed to get there faster if so desired. If escape from Earth can be had for (to pull up a number by guesswork) 250 m/sec, this seems to imply that a transfer orbit to Mars would cost only 750 m/sec more. Well, gosh, I know it isn't tremendous but I think it is rather more than that! Earth's orbital speed is a hair under 30 km/sec. Adding 1000 m/sec to that raises the kinetic energy by a bit under 7 percent, which is to say it lowers the magnitude of the potential energy deficit relative to escape from the Sun to 47 percent (versus 50 for our approximately circular orbit at 1 AU) which would correspond to a circular orbit a bit over 7 percent more than 1 AU, or a Hohmann transfer orbit going from one to 1.145 AU, or almost 22 million kilometers farther out from the Sun. But that falls far short of Mars orbit!
This makes me wonder if it is actually a two-burn maneuver; one burn takes the craft onto a path that falls past the Moon to fall in toward Earth on a long ellipse that brings it to near escape velocity near but safely above the atmosphere, at about 11,000 m/sec, and then most of the 1000 m/sec delta V is applied, to bring it up to say exactly 12,000 at an altitude where circular orbital speed would be say 7850--that would send it on a hyperbolic path away from Earth that would asymptotically fall to a residual recession speed of 4556 m/sec--which is I think considerably more speed than a Hohmann transfer to Mars would generally require! That would swing out to nearly 2 AUs, and even at aphelion Mars is closer to the Sun than that, so yeah, I suppose this is what the L2 launch actually looks like, a slingshot maneuver dropped from L2 to Earth, taking full advantage of the Oberth effect.
But while we can see that much of the investment in delta-V required to boost packets of material from LEO to L2 would clearly be recovered by this launch maneuver, it is not so clear to me that we actually do come out ahead. Maybe, maybe not, it depends on more detailed and accurate calculations than I bothered with just now. It does seem clear to me that if there is any advantage, it will be a marginal one.
Whereas the drawbacks of working from an L2 assembly point are considerable:
1) Need to develop adequate infrastructure from as far beyond the Moon as a geosynch satellite orbits above Earth, to support the lives of assembly crew;
2) radiation exposure; not only is all work being done out where Earth's magnetosphere is negligible and so everyone is exposed to full Solar radiation, in order to get people there to work on things they have to be launched through the Van Allen belts! To be sure this was a minimal problem for Apollo since the Moon's inclination allows one to pass north or south of the lower, high-intensity belt, while the outer belt is relatively mild and one passes it quickly. Still, all this is extra exposure compared to simply operating below Earth's main belt, in LEO, where the magnetic field offers ongoing protection.
3) probably the most decisive disadvantage--launch windows. The basic launch window for a given transfer orbit attainable by a given rocket system is given by the relative locations of the two planets involved. But then the exact moment of launch is given by the object to be launched own position in its parking orbit. For a craft assembled in LEO in an orbit inclined so as to be in the plane that averages closest to Mars's (pretty much to say, in the plane of the ecliptic) it approaches the right angular position for boosting to Mars about every 90-100 minutes, depending on the orbit. But a craft based at L2 only would be at the right angle once a month! This problem applies no matter which way we launch the thing, whether it dumbly needs to add several thousand meters/sec for a direct ascent from L2 to the Hohmann path to Mars, or if we boldly dive down to Earth and maneuver at perigee as e of pi is surely assuming to be catapulted onto the course. Either way, when the optimum alignment of planets occurs, the L2 launch might have to hold off for weeks, or launch weeks prematurely, waiting for the Moon to get to the right place.
I very strongly recommend that a spacecraft of this type, using high-thrust engines of ISP under the tens of thousands of seconds, should simply be assembled in LEO, and that the apparent high cost of boosting it to TMI from there versus L2 would still have to be paid piecemeal to send the components to L2.
So best launch from LEO, where assembly is less difficult and the launch opportunities are more reliable; the economy of staging via L2 looks like an illusion to me.
It would be different if a sufficient portion of the material used for the craft is from the Moon, and if the high costs of getting Lunar material are less than those involved in sending up Terran material. Even then though we'd still have the launch window problem to work around.