I'm not sure you're right about the throttled behavior, though. My copy of RPA is spitting out that at 80% throttle, the SSME-35 engine would still produce 395s Isp at sea level, compared to 399s full-throttle. Thus, the 80% thrust from 5 engines has about the performance and propellant consumption you were assuming from 4 engines.
We are almost certainly using different versions of RPA. Mine is free and lacks all sorts of desirable things such as accounting for different modes of pumping the chamber, factoring in cooling, etc. Basically it just assumes all engines are pressure fed with a meter area throat. I've noticed I might be able to adjust the temperature of the propellants fed in but I've never tried to fool with that (so when I run numbers for hydrogen peroxide it assumes the input temperature is 288 K rather than 275, for instance--I've wondered if I could get figures for an atomic rocket with hydrogen monopropellant by setting the input temperature high enough; not sure if that would work in this version of the software or not/
But even so, by golly you are right...even at 69 expansion there is very very little difference in the "throttled performance" estimate at SL between 60 percent and 100. In this worst case it comes to under 10 percent
However at 87.2 percent, at standard SSME expansion of 69, the curve does indicate flow separation. Throttling to 80 percent with a normal SSME would only involve less than 5 percent reduction in Isp, but we'd have those vibrations from chaotic air intrusion fingers.
Of course the key here is lowering the expansion ratio, which costs us some performance in thin air but gains it where we need it most, at SL.
Switching over to that, the simple software I have says there is no flow separation at all, even at 60 percent throttle. Isp is lower at 80 percent but only down to 97.25 that at full throttle. (I am not giving absolute figures it gives me because these are theoretical and overoptimistic, for actual performance I rely on "Estimated Delivered Performance" which matches up well with published data and in this case full throttle Isp would be 394.11 sec. Thus at 80 percent I'd expect 383.26, a difference of about 11 sec. I noticed your propellant masses are indeed on the high side, which makes sense, because of this little hit and because throttling down more for Max-Q minimization would involve a bit more Isp deterioration--but by then we are in much thinner air, maybe the shock wave wake offsets the pressure more too, and in vacuum there is hardly any Isp variation due to throttling to speak of.
It seems then that the SSME operates at such a high core pressure, near 200 atmospheres, that even high expansion leaves it plenty of margin to shove sea level air aside with only minor impediment. The standard 69 expansion has the exhaust down to 1/5 atmosphere at nozzle exit, while 35 expansion firing into vacuum is close to half an atmosphere (these both at full throttle of course). The combination of lower expansion and the high pressure core is what keeps the throttled Isp curve so close to flat I guess.
This helps vindicate the high tech approach of modifying such a fancy engine as the SSME for the booster job, as opposed to what I favored--adapting the F-1A by downsizing and derating it. If we need a spare engine for engine out on the boosters as well as the Orbiter/OPAM core, this high pressure core approach is strongly favored. The Isp curve for lower pressure core engines like the F-1A would be a lot steeper and flow separation at lower throttle a major concern. Of course simply using hydrogen for both sets simplifies operations and made it possible to use the identical cores for both versions which simplifies refurbishment operations, and of course amortizes them over a lot more engines in the pipeline (8/11 of which have undergone a relatively short burn for lighter average task load in necessary repairs, though checkup must be the same for all). I didn't find any major advantage in overall thrust or other parameters trying to switch the SSMEs to other fuels; my main objection to hydrogen fuel was the tremendous tank volume, which aside from construction costs, involves aerodynamic issues in the boost phase.
How much of a performance hit comes from the volume difference between the LBR and SRB?
As noted in a previous post, we're actually using the 5-engine pod version from the report, which allows all engines to run at 80% (improving engine life) and preserving throttle-up for nominal performance even in the event of a single engine out almost directly off the pad--a dramatic improvement in Shuttle abort modes!
Derating the engines for durability is something I did mention, though in context of alternative fuels and different engine designs. I'd guess it is not linear--that is, if all 5 engines run at 80 percent, we get more than a 25 percent extension in seconds of burn before a given refurbishment is necessary. Well, I suppose that would be true of some forms of wear and tear, and maybe others just go with burn time period, so it would be a mix. At any rate everything should be well back from red lines. And there is still that emergency 9 percent super-throttle available on all 11 engines too.
The pod weight is a bit high in the paper (~37 tons per pod), but we're trusting the study, the additional engine(s) over the orbital pod, and the various floatation and recovery gear.
And comes amazingly close to my 70 tonne all up (per booster) dry estimate, based on aiming for a certain delta-V target.
Is this 37 tonnes well in line with the 3-engine OPAM? The OPAM has a tougher job of course, it has to put the payload of Shuttle C into orbit, then brake itself for descent and endure orbital reentry. The booster engine pods just have to survive modest supersonic/hypersonic braking and then splash down, but with 5 engines instead of 3. So we might guess the OPAM must be more than 25 tonnes, but possibly not more than 35.
Michoud was tooled for up to 24 full external tanks/year with their existing tooling, so producing more tanks comes at relatively small capital cost.
OK, capital cost is handled with built in surplus capability--though insofar as these improvements lower launch costs, we'd expect an uptick in demand that might make Michoud a choke point.
But how much does each tank cost? Not all the costs are amortized! I suppose the materials cost might be modest. That leaves the actual costs involved in assembling each one for a single use (except for ETs that are destined to be sent to orbit to be repurposed there, but I think it is clear by now such tanks need a fair amount of customizing--widen the intertank separation, install hatches or anyway easy cutouts to place them, install RCS thrusters, etc, never mind pre-fitting floors and walls, and so on). The material and labor costs per tank, setting the amortization at naught at first, should be compared to any first cost increases to make the booster tanks reusable, and then refurbishment per launch. Indeed we take a hit with amortization if we don't need a new tank for each booster to be launched--but if the price of a launch can go low enough, better amortizing the investment of infrastructure and launch crews at Canaveral and Vandenberg, Michoud will offset lack of work for new booster tanks with increased rates of orders for new ETs.
Having looked at the use of one of your LRBs as a booster for a two-stage LEO stick rocket, I guess I need to withdraw the suggestion of an RLB where the ET returns to Earth along with the SSMEs. If the OPAM masses say 33 tonnes and an ultralight ET 27, that would leave 90 tonnes for the payload module (though there are non-payload mass requirements there too, shrouds and so on). But we can't recover that, just the OPAM. To make the ET reusable too, we'd have to raise the mass--I suppose making it out of steel might come close to doubling it versus the aluminum-lithium alloy ultralight tank, and it would still need some sort of TPS, probably lighter than people might guess between the load being "fluffy" and the heat tolerance of a steel tank, but still substantial, plus maneuvering fins and terminal landing braking of some kind. The 90 tonne payload package gets whittled down fast; if we can get by with as little as 30 tonnes (after all, much of the OPAM mass can be trimmed, if the ET is included, to partially compensate for greater mass there) the payload section is down to 60 tonnes. Whereas I am pretty sure using one LRB as a booster stage will put up more than 30 tonnes! Two LBR stick launches would be superior in capacity to one RLV launch,using the same boosters, while making the stick second stage recoverable might be very feasible at modest payload cost.
Therefore we might not have nearly as many full scale STS-2 launches, going over to many LRB/reusable upper stage launches. But each mega-scale 90 tonne launch we do will require an expendable tank. And many of these, a customized ET to be used as a structure in LEO or beyond.
Second, as long as you're recovering from the water, one of the most important things is keeping the engines out of the water (not as critical perhaps for something simpler like an H-1 or a pump-fed stage, but for an SSME...best to keep the turbo machinery dry and salt-free for ease of maintenance). This is easier with a pod floating "nose down" in the water like a small boat than a full booster stage floating on its side in the water.
But if the LRB is fully reusable, what we can do is first of all trim it to brake sideways, a la Starship--this requires fins on nose and tail, the large ones on the tail of course. With those fins the thing comes down in "skydiver" horizontal attitude, lowering terminal speeds below 100 m/sec. We can lower than more on near approach to the water with parachutes, two sets, nose and tail, the latter again larger than the former because of the engines biasing the center of mass back. Now, just before we hit, we inflate a pair of balloon floats on the tail or farther forward, between tail and CM, and another single bumper balloon on the nose tip, and then release the nose parachute. (We can avoid losing it by having a line to the rear cluster of parachutes). Now the nose swings down, the tail being better braked, and the nose balloon slams into the water as a shock absorber--we design it to deflate at maximum pressure, dipping the nose tip in the sea. The tail is now coming down as a lever, partially buoyed by the nose floating, so the tail balloon floats hit at a moderated speed and handily prevent the actual stage tail from getting a dunking. It rides on the water as a tripod supported beam, nose awash, tail well above the wave crests and out of the lower heavier spray zone. The engine nozzles certainly can take a moderate amount of salt spray. To protect the more delicate stuff above the level of the nozzle throats, we can have inflatable collars, simpler and lighter than clamshells, presumably the engines are gimbaled to a fixed stowage neutral angle and thus pre-tailored inflatable shapes, that don't have to handle anything worse than residual warmth of the engines, can fit pretty snugly. We might want these on the booster pod only version anyway since even a rather long engine pod won't be able to lift the engines terribly far above sea level. Float gas volume could be less for superior clearance above the water in the full recovered version too. Now the recovery boat can just secure the tail end to the boat stern, put a spray shroud over the engines, and either drag the stage back with the nose being dragged through the water, or to protect it better and perhaps lower drag, some floats can be affixed to it.
since I know you're fond of the idea @Shevek23 -- even most contemplated airships)
Well, there seems little reason a reusable booster would mass much more than the LRBs here, so 60-80 tonnes. Your booster engine capsule is already on the heavy side of possible helicopter lift. Whereas such airships as the USN ZRS designs USS Akron and Macon already had useful lifts in this ballpark; Hindenburg or Graf Zeppelin "II" could nearly do it as built and operated. These are big rigids of course; for a single airship to fish out both of two boosters each massing 80 tonnes would require 160 tonnes useful lift. No airship constructed in real life would quite do it but they come close and modest lengthening to 300 meters/1000 feet was planned by various concerns, such as the British imperial air service contemplated in the 1920s; their facilities were designed for anticipated "thousand footers."
Using 1990s on the drawing board technology, a variation of the Zeppelin NT design would seem entirely feasible and would fit within some hangars already extant--though these (the Akron Airdock, the former Naval hangar at Sunnyvale California now absorbed into Ames NASA campus; perhaps the Lakehurst hangar in New Jersey, and maybe Cardington in Britain, as well as the Zeppelin hangar in Brazil) have all been repurposed and cluttered up. The basic NT design involves three contoured longitudinal keels braced by an inner prism of struts triangulated with tension lines, the whole thing fitted inside a blimp envelope to form a semirigid design. A variation on this ought to work fine stretched to 300 meter length, and could handily handle cargoes much exceeding 100 tonnes.
It isn't extant but it is fairly conservative. We'd want new hangars convenient to Canaveral and Vandenberg, and maybe another one at Michoud if we want to use the airships to hasten transport of assembled boosters and ETs to Canaveral. If designed for 200 tonnes useful lift, one could haul a pair of LRBs stowed in a streamlined (air-pressurized, per blimp operations) lower bay and the ET as a sling load outside. We would need to allow some volume expansion room hence a bit longer and greater diameter to fly from Michoud to Vandenberg within CONUS, taking the Gadsden Purchase route for the lowest passes over the continental divide. But you tell me how it was proposed OTL to get an ET from Michoud to Vandenberg? Was it going to go in a Super Guppy type airplane of some kind? That might work since the ETs were so light. Blimps have built in ability for their helium to expand some 33 percent, mainly for trim reasons rather than to rise 10,000 feet, the problems the old rigids had clearing the Divide even on the southern route had to do with their not being designed to rise terribly high above sea level. If the high CONUS route to the west coast were ruled out, the narrow low isthmus in Mexico is available if we can trust crossing Mexican airspace, or if not, the Panama Canal zone is presumably secure enough, and if ETs would not go on airplanes by OTL plans the Panama Canal route is the only one I suppose the ET could take to Vandenberg anyway. With the airship it goes a lot faster than by barge, and barring severe weather, smoother.
No extant design, nor any in history, would be suitable for recovering even single LRBs in the 50-90 tonne range, true. But anyway airships no bigger than some historic rigids, including conservatively a rigid rather than pressure ship design modified for sling load lifting and securing in a cradle, could be designed conservatively in the case of the rigid. No pressure ship big enough has flown, but I think extrapolating a design like the NT is pretty straightforward.
A big cost item would be constructing the hangars, a minimum of three, for Michoud, Canaveral and Vandenberg at least. Once upon a time, there was talk, in the days when NASA closed its eyes and assumed the STS would meet all launch cadence and price reduction targets as designed, of creating a new launch site in south Texas, about where SpaceX is operating today. And delerious levels of success might suggest a few more bases--a section at Kourou for subtropical inclination launches and maybe expanding Kodiak or some other far northern point, in Canada or Europe, for very high inclination launches such as sun-synchronous orbit. We'd want hangars at each launch site.
Meanwhile, I have to admit fishing the stages out of the water with boats seems to work well enough, at realistic launch cadances.
It's worth noting the VAB height limits are quite generous. Though the Shuttle (and particularly our LRB-fitted Shuttle and Shuttle-C) are tight on the doors i, they literally could roll out Shuttle without opening the door all the way:
The VAB doors can accommodate a launch vehicle up to 120 m tall and roughly 23 m wide.
Lesson learned; here I have been assuming the Shuttle did crowd the limits, but of course I should have remembered Saturn V was a much taller stack.
Plenty of room then for supersized Shuttle-derived systems. All liquid fueled, they would be lightweight brought to the pad empty; the major limit would seem to be the magnitude of net thrust blast the extant launching pads can handle, historically set by the Saturn V.
However, Silverbird isn't dealing with the use of OMS during boost or for circularization (part of why per @TimothyC's messing around with it, it doesn't do a great job modeling Shuttle with the default assumptions for that vehicle, either).
I have been assuming the figures are close for when one chooses the "historic vehicles" Shuttle options, which requires one to name the particular Orbiter.
To try to get a realistic match in custom user-defined vehicles, I specify two boosters on a two stage design. Stage one is full SSME thrust, 6510 kN, vacuum Isp 452, with the ET as the stage dry mass. Second stage is the Orbiter, assuming 4 tonnes of OMS propellant with the OMS as the engine--now I wonder if I underestimated the thrust by using thrust for one OMS when there are actually two. Using historic figures for Columbia mass minus the 4 tonnes propellant, and IIRC 53 or so kN for Columbia's own OMS thrust, and knowing exact figures for the SRB burnout mass and mass of the grain consumed, I kludged around with the inferred booster vacuum Isp, which multiplies by 9.81*grain mass/123 seconds burn to give a thrust in KN--wrong of course versus the real world, because SRBs burn in dense atmosphere and with variable burn rates, starting high and gradually dropping. Mass flow is not constant and vacuum conditions are never achieved, but the inferred Isp where the payload matches what the historic vehicle program reports a given Shuttle could have delivered to a test orbit, I suppose the Silverbird model with the side boosters having the specified performance will be about right.
It might be better if I did it as a three stage, with the second stage having zero dry mass; this represents two OMS burns, one upon ET separation and the other being the final circularization.And it is a major blooper if I am using just half the available Orbiter thrust--though actually such near-free-fall maneuvers should give similar results whether longer burn at lower thrust or shorter at higher.
The way I did it, the inferred constant thrust SRB equivalent would have thrust 11,100 kN and Isp 277 sec both in vacuum. Naturally this kludges right over such fine points as lowering thrust to minimize Max-Q and so on.