the "Spencer Launcher"

A couple of months back, Ed Kyle wrote:
>>Henry Spencer wrote:
>> Of course, if preservation of jobs at Rocketdyne was not a priority,
>> you *could* just use a cluster of RL10s...
>
> Keep going with that idea. Replace the SRB booster with a stretched Zenit
> first stage, powered either by an RD-171 or by two RD-180s, and top it
> with the new "Spencer" cluster-RL10B second stage...

A launcher with such a noble name :-) ought to be thought out from scratch,
rather than as a derivative of somebody else's hack job. :-)

So I thought about it, in my copious spare time (which is why this followup
has taken a while to appear). And I came up with an idea that seems to be
of some interest. (Note that I also choose a different name for it.)


Assumptions

Of course, what you get out of the engineering design process depends a
whole lot on what assumptions go in. I'm assuming:

1. I'm not allowed to mess with the overall architecture. Notably, no
serious orbital assembly is allowed -- although a lunar mission is allowed
to dock two payloads in LEO -- and the launchers are expendable. Stupid
assumption, of course, because it's a stupid architecture...

2. Launchers have to be "shuttle-derived" in at least some loose sense,
as a matter of salesmanship. EELV derivatives are ground-ruled out, as
are Saturn V revivals and clean-*** designs.

3. However, preservation of jobs at existing shuttle contractors is *not*
a priority. (If it is, you get NASA's existing designs.)

4. No crucial foreign components.

5. The time scale does not permit developing new engines.

6. Launch is from KSC LC-39, possibly with modifications.

Assumption #5 is actually a serious headache. The US really needs a big
high-Isp kerosene engine that doesn't run afoul of #4. The RS-27A is the
best currently on hand, and it's just not good enough.

It *does* make sense to base tankage on the shuttle ET. That's a big,
light tank, with a fairly large diameter, with production tooling already
in place and operating.


Architecture

Consider costs. For low launch rates -- and there's no real prospect of
seriously high launch rates here -- costs are dominated by development
cost and by the "standing army", the salaries of the people needed to do a
launch. Altering the launch rate has essentially *no* effect on annual
cost. Launching fewer times saves you almost nothing; launching more
times costs almost nothing. (The shuttle and Titan IV are examples of
this.) Some contemplation of this yields an interesting conclusion:

Build One Launcher, Not Two!

There is no possible way you can justify the added development costs of
the CLV, or the extra standing army needed to operate it. It's *cheaper*
to put the capsule plus a tub of ballast on a heavy launcher than to build
a medium launcher just to carry the capsule. (There's even a development
path here: the Block 1 heavy launcher can have a substantial performance
shortfall without hampering early CEV operations, provided you can make it
back with incremental improvements in a Block 2 design.)


Launcher Concept

I fiddled around with this a bit, and here's what I came up with.

The first stage is a shuttle ET, the modern aluminum-lithium version,
about 26t of tank (all masses in metric tons) holding about 720t of
propellant. The LOX tank at the top needs to lose the pointed nose, in
favor of a blunter tank end with an interstage ring extending up, but that
shouldn't be at all hard -- we can use the tooling currently used to
build the *bottom* of the LOX tank. With no SRBs, a fair bit of heavy
structure can be deleted; I assume that suffices to cover the interstage
ring's mass.

At the bottom of the tank, we put seven RS-68s, a ring of six around a
seventh. The engines are about 48t, and throw in another 10t for a thrust
structure. The RS-68's sea-level Isp is 357s, and its vacuum Isp is 409s;
I assume an average of 380s. Takeoff thrust/weight is 1.25, about the
same as the Saturn V.

The last mass property of interest is residual propellant at the end of
the burn. NASA's designs seem to assume the orthodox rule of thumb, 1%...
which is ridiculous. The S-IVB specified 0.25%, and typically achieved
even better. I'm specifying 0.25%, or about 2t of residuals.

On top of this is the second stage... another shuttle ET! Same capacity
and mass, same blunted nose. This one has only two RS-68s (14t), with a
smaller thrust structure (5t). Here the Isp is the full vacuum 409s. The
thrust/weight at ignition is about 0.8, which is okay for an upper stage.
Same 0.25% residuals. Needs an altitude-start variant of the RS-68, but
that should be pretty trivial -- it's a much simpler and less cranky
engine than the SSME.

On top is 100t of payload. A fairing, if needed, is charged against
payload.

This is kind of a tall stack, but it fits. A stock ET is 154ft tall.
These stages lack the pointed nose, but have engines sticking down, so
call that even, making two stages about 310ft high. That leaves about
100ft -- nearly 4x the rocket diameter -- for payload while still clearing
the VAB transom.

The first stage yields about 2.1km/s of delta-V, and the second about
7.1km/s, putting that 100t of payload into low LEO (none of this business
of getting it almost there). The total of 9.2km/s arguably is slightly
low, but if needed, we can probably pick up 0.2-0.3km/s by putting a
nozzle extension on the second-stage engines -- the RS-68 is optimized for
first-stage use and has a rather low expansion ratio.

The first-stage tank mass may be a little optimistic, since it may need
some beefing-up to cope with the heavy load on top, plus the bending loads
of the longer stack. However, extra structural mass in the first stage
has little effect on performance.


Payloads

100t is not as much as NASA's heavy launcher. However, because we keep
the idea of docking two payloads, we have *200t* to use for the lunar
mission. We do it much like the final Apollo EOR concepts.

One launch carries the TLI stage, fully loaded. It's the same 8.4m
diameter as the lower stages, but rather shorter. Knock off, say, 2t for
a nose fairing, jettisoned on the way up. Scaling the mass numbers, but
discounting structure a bit for light loads, we get about 92t of
propellant in 3t of tank. Here we might want to use a common bulkhead
rather than an intertank. A 1t thrust structure carries seven RL10Bs,
which weigh about 2t total. The initial T/W is low, about 0.4, but that's
okay for an orbital-maneuvering stage, and the RL10B Isp is 462.4s, better
than even an SSME and much better than a J-2S.

This stage can't quite boost 100t of payload to escape, but it comes
pretty close. Either cut the TLI payload to around 90t (we have to deduct
something for its fairing anyway), or (better) have the payload supply the
last little bit of boost.

Docking the two together in LEO might seem a little tricky. Remember,
though, that the lunar lander needs a downward view for landing. Once in
orbit, the capsule separates, turns around, and docks to the lander, like
Apollo. Then the docking to the TLI stage is flown from the *lander*
cockpit. (Actually, it might be better to do this as berthing rather than
docking -- have a small robot arm reach out and grab the TLI stage, and
ease the two together slowly under precise control.)

For LEO operations, there's lots of extra mass available, even with NASA's
elephantine 20t capsule... not forgetting that the second stage, which is
roughly a shuttle ET, also reaches the initial parking orbit, and could be
useful. (If not, it gets deorbited the same way the Skylab S-IVBs were
deorbited, by propellant dumps through the engines.)

The capsule can be made the full 8.4m diameter, which gives a lot more
room inside, and permits drastically shrinking, perhaps even eliminating,
the expendable service module.

Finally, if you really want to optimize and you think it's worth the extra
hassles of dealing with another configuration... For light-load LEO work
like station crew rotations, if the capsule (with its service module, if
any) can supply about 1km/s for orbit insertion, you can dispense with the
second stage entirely. Yes, with a light load the first stage is a
near-SSTO; in practice, you'd want to delete three or four of the RS-68s
from such single-stage configurations, to save mass and avoid overly-high
dynamic pressures. (Caveat: this scheme works only if the first-stage
structural mass doesn't grow much, because *here* it really matters.)


Conclusion

Okay, the same-size first and second stages are a bit inelegant. It'd be
better if the first stage was LOX/kerosene, in which case it would have
several times the mass of the second, and would contribute substantially
more delta-V. (Running the numbers on that, with revived F-1s as the
first-stage engines, is left as an exercise for the reader. :-)) And a
larger diameter would be better. But given the constraints, it actually
works out pretty well, at least at this back-of-the-envelope level.

However, it does seem a bit egotistical to attach my own name to it. Hmm,
long and thin, and given the color of the ET insulation (which will still
be needed on the LH2 tanks), I think I'll call it the Brown Bess. Still
an antiquated weapon(*), but a lot better than a stick. :-)

(* For those who don't recognize the reference: the "Brown Bess" was
[slang for] the standard British Army muzzle-loading musket of the 18th
and early 19th century. )

This design uses only existing engines -- no J-2S revival needed -- and no
solids. The first two stages structurally are simple derivatives of the
shuttle ET. The TLI stage is all new (except for the engines), although
its tanks can use some of the ET tooling, and it's only needed for lunar
missions so it can be scheduled a bit later. The SSME and SRB production
lines can be closed. And we get nearly twice the TLI payload of NASA's
design, and several times the LEO payload, at lower overall cost. What's
not to like? :-)
--
spsystems.net is temporarily off the air; | Henry Spencer
mail to henry at zoo.utoronto.ca instead. | henry@xxxxxxxxxxxxx
.

Quantcast