P.S. For Ray, here's a question: why would a capsule version of OSP
or CEV require as many lines of code as a shuttle? The GNC should be
a lot simpler, especially on a LEO-version (start with a modest
computer for CEV-L, replace the Pentium with a Pentium Pro for CEV-S
(S=Selene for the moon shots), and go with an Itanium for CEV-M
(Mars); processor names chosen for familiar analogy rather than as an
actual design point). In addition, many of those lines of code should
already exist (GEO transfer stage guidance, for instance).

ELCSS should also be closer to "off the shelf" now that we have
experience with Soyuz, Apollo, and Shuttle designs; the ISS designs
are also useful input, but would be overkill on a 4-day flight. Space
suit designs might also have given us engineering data that would help
with a compact modular unit (here's an opportunity for a reusable
component in an expendable airframe).

Modelling a capsule's aerodynamics and heat transfer should be simpler
than a winged design, thus saving CFD and wind tunnel costs.

So many systems should be easier to design and/or manufacture on a
capsule CEV that I would expect to see big savings from adding up all
the smaller savings. Do we lose all that in system integration costs?
Would they really be as bad or worse than the SI for the shuttle?

Tnx

/dps

You would think so. However, there're numerous studies going back to the
days of the RLV Subpanel of the NASA/DOD Aeronautics and Astronautics
Coordinating Board (1965) and the "Integral Launch and Reentry Vehicle"
(ILRV, 1968) work that preceeded the shuttle Phase A effort (1969-70)
indicating that the mode of reentry (capsule/parachute, lifting
bodies/runway, or shuttle orbiters/runway) is not a strong driver of
development cost. NASA spent about $20B (current dollars) to develop and
manufacture five orbiters. Of this, $14B was spent on engineering
development and for Enterprise, Columbia and Challenger. About $2B was spent
for each of the last three orbiters, Discovery, Atlantis and Endeavour.

For Apollo, the CSM cost was $22B (current dollars) for engineering
development and for 12 Block I vehicles, 23 Block IIs and 20 boilplate
units.

When I worked on the DC-X/XA program at McDonnell Douglas in the early
1990s, we made a lot of PR noise about semi-automated software development
tools like Matrix-X. And I'm sure that there are better tools now. But
flight computer hardware and software typically account for 5-10% of total
development cost. So saving 10 or 20% on this cost doesn't change the bottom
line significantly. As far as "off-the-shelf" hardware, I'm not aware of
any that could be used in a new vehicle like the CEV without significant
modification. Spacecraft and launch vehicle designers and program managers
have enough problems without trying to shoehorn "alien" hardware into their
designs.