By the way, 80 or less column posts are usually appreciated.
Format, format, format!

Charles Talleyrand wrote:
"Richard Schumacher" wrote:
The important point is that propellant is cheap,
cheap, cheap, and loss of a launcher is as expensive as all hell.

Best reliability calls for a completely reuseable single stage
launcher, with engines of such a size and number that, if at any
time one of them must be shut down, the others throttle up to
compensate, and you just keep going.

This is not obvious.

In a world where engines explode upon failure, having exactly
one engine per stage is best.

Engines actually rarely explode on failure; going back
through the history of flight failures shows almost exclusively
systems failure followed by shutdown, or accidental shutdown,
without any uncontained failure. It's not unknown but is a lot
rarer than 'graceful' shutdowns.

Cost and complexity constraints as well as reliability
analysis do argue for single engines per stage on expendables,
and five or more on reusables with abort-to-orbit (fewer if
abort-to-ground is ok).

In a world where there are finite development dollars,
and those dollars can buy reliability, having
one TYPE of engine per vehicle is best.

That does have its limits. It's great on SSTO and Stage
and a Half, and some TSTO concepts. It sucks on three or
more stage vehicles where the GLOW of the first stage may be
a hundred or more times the GLOW of the last stage...

It has been suggested to me in private email that the correct
answer to this delima is to have one engine
FAMILY, but with multiple engine sizes per family.

Hard to do that; engines don't scale very well in terms of keeping
similar design and construction. Similar operating concept and
specifications? Sure. But the parts won't be descended or
related very closely.

One reasonable exception is truncated versus long nozzles...
*that* isn't such a big change. You can keep the exact same
pumps, combustion chamber, injector, etc.

The answers to some of these questions vary significantly when
you look at serious RLV operability and seriously far out BDB
designs, as optimizations start to pull in unexpected ways.


-george william herbert