Most designs must provide at least the min net positive suction head (10-100 psi) for the engine pump, whatever that is. A few tank designs are for pressure-fed engines, something like 2-3 times the engine chamber pressure. That overpressure (tank pressures anywhere from 1000 to 5000 psi) is for the pressure drop or drops through the flow regulator.
Internal pressure is difficult to design for, and always ends up being heavy, for anything but simple circular cross sections, period. The only feasible conformal design is “lobed”, parts of circular sections “stitched” together into an approximation of the noncircular space to be occupied. In effect, you build the equivalent of an air mattress.
This sort of thing is still today far easier to do with metals and welding than with composites, because of the inherent difficulties with composite joints (especially at higher pressures). How to reliably design the joints between composite components, or between composite and metal components, is another topic for another time.
Especially with hydrogen, the inherent porosity of a composite requires some sort of nonporous liner, even if it is a paint layer, no matter how “wet” your layup is (and “wet” layups are much heavier). This zero-porosity liner requirement raises composite panel weight significantly. There is no way around that dilemma.
The best structural joints with composites are fully “glassed-in” joints. Unless you have a conformal tank with lobes big enough for a man to crawl inside, this can be difficult or impossible to do.
With current material technologies, all of the preceding facts of life are why I personally favor metals and welding for conformal tankage.
Liquid hydrogen tanks down here on Earth are usually made of 300-series stainless steel, and for very good reasons. These tanks can be filled and refilled for decades without cracking, in spite of the super-cold propellant-induced thermal stress cycles, cycling that is excruciatingly severe. The new lithium-aluminum alloys now favored for rocket stage tank construction seem to work for liquid hydrogen quite well. I am not familiar with that material, since it is so new, and I am not.
The effects of repeated fill cycles may cause aluminum-lithium to crack from fatigue (something inherent with all metals, and notorious with aluminum alloys). That sort of difficulty is not something you “run into” with one-shot throwaway stages. But, if it doesn’t crack with repeated use, or there are lots of cycles available before it does crack, then reusable tank structures are possible with it. We’ll see.
Lobed construction with metals does not require the extensive use of doublers (except near ends where the strain mismatches), unless you badly design the shapes. Ideally, you join segments of cylinders together, with a properly-perforated (again, a whole other topic) linear web-wall at the joint. The only “trick” is to eliminate all bending by your chosen geometry. These panels are butt-welded to a three-way joint piece at every joint line. That joint piece has a cross section that looks like a three-prong grass burr, radiused down in the “groins” at the base of each prong. That radius need only be at most a little larger than the panel thickness.
Done successfully, you have a tank only a few percent heavier than a cylinder of the same volume, but not heavier by factors. It will be at least a little bit heavier, that is inevitable. That’s simply the price you must pay for the shape you want. Update 10-7-13: for the same panel thicknesses and weights as cylindrical construction, a lower-bound estimate of the weight growth factor is the perimeter length ratio, computed from cross-section views.
That joint piece could be made by extrusion. Each leg of the joint piece is the same thickness as the panel that joins to it, and must match the tangent at the edge of the panel. Only through the cross section of the three-way joint piece is the effective thickness about twice that of the panels. This shape’s stress distribution has been checked with 3-D finite element analysis: it works fine. There will be a tiny amount of shear yielding down in each “groin” line, but only on the first pressurization cycle.
You could use wire-feed welding to assemble the tank, but you have to put a slightly-bulging weld bead on both sides. That means a man must be able to crawl inside each lobe, and be able to weld inside there. Wire feed welding works very well with aluminum, though.
If the tank were a stainless steel like D6ac or 4130, you can electron beam-weld right through from one side, with nearly-perfect weld strength efficiency. Weld the joint pieces to the web walls, then weld the outer shell panels to the joint pieces, all from the outside. Then proof test. The reject rate should be low, once your process is defined. One-side electron beam-welding of steel is a well-proven industrial technology for the mass production of solid rocket motor cases.
About a quarter century ago, I proposed exactly that electron beam-welded, stainless steel lobed design, for a small conformal-case solid rocket motor case, to meet a seemingly-idiotic shape requirement for a weapon project (that ended up never flying). The customer expected to see some version of elliptical designs proposed, which simply do not work unless they are extremely heavy. Preconceptions clouded his ability to see a lightweight solution that would work.
Questioning the assumptions you would otherwise start with, has been the most powerful tool in my engineering arsenal for nearly 4 decades now.
Mastering non-conformal tank technology is required for any winged or lifting-body spaceplane, whether it be one stage or two. Chemical, nuclear, or something not yet invented, this requirement still holds for best-storing whatever propellant is required, within the odd spaces inside the vehicle.
Metals we know how to handle in this application, composites not so very much. And that plus politics is really why the X-33 program ended up unsuccessful. It really should have started with the metal tanks in the first place, but the folks working on it were seduced by the higher strength-to-weight ratio of composites. In a pressure tank situation, those advantages tend to evaporate in the harsh light of all the other design issues. That picture of things hasn’t changed, and probably never will.
After reading this, some of you may wonder why I haven’t been “snapped-up” by a Boeing or a Lockheed-Martin. The answer is simple: I’m old, and old guys are more expensive.
Having the wide-ranging cross-disciplinary experiences of an old guy on your team, may well help guide you very cost-effectively to the “right” solution for your project. But, the way R&D is funded by the government in this country, project success is not required. So labor cost is the only factor considered in government contractor hiring. Few-to-none of us old guys get hired.
That has led to a widely-unrecognized lost-art problem. The engineering project team is supposed to consist of a mix of old guys and young guys. The old guys pass on to the young ones that engineering art that was never written down. It wasn’t written down, primarily because the company didn’t want to pay for writing it down. That art is about 40% of engineering practice in aerospace work.
If there are no old guys on the team, no art gets passed down. Which lack neatly explains why different outfits keep reinventing all the same wheels, and why progress with flying machines has slowed in the last few decades.
Few are learning from industry history anymore, because those who knew that history are largely no longer there. You cannot get that kind of knowledge from a college classroom, it is dirty-fingernails workplace experience, pure and simple.