Index Home About Blog
From: Henry Spencer <>
Subject: Re: Rocket Engine Test Q
Date: Mon, 17 Jun 1996 20:25:33 GMT

> >  How are rocket engines with high expansion ratio nozzles tested?
> >
> >  You can't test them at sea level without getting flow detachment; if
> >you enclose them in a vacum chamber, and run your engine for any length
> >of time, it won't be a vacum chamber anymore, etc.
> well given the USAF Phillips labs have a fairly large vacuum test
> chamber, i would bet the problem was solved with large pumps.

Yes and no.  The rocket engine itself makes a dandy pump.  Once it's
started, a properly-shaped exhaust duct downstream of it suffices.  For
the start, you may need to pump the duct down beforehand, but you don't
need high vacuum (or even particularly high pumping capability, if your
start sequence is quick).  The RL10 was developed that way. 

Another approach, as others have mentioned, is to build the engine with
a short main nozzle plus an extension, with the main nozzle sized to
be testable at sea-level pressure, and the extension using some type
of cooling that doesn't interact much with the rest of the engine.
You still need *some* vacuum testing, but it can be much reduced.  This
is what the Japanese did for the LE-7.

Still a third approach is to just live with the flow separation.  If the
nozzle is reasonably strong, the only big problem is that asymmetric
separation causes unwanted thrust vectoring.  You can live with this by
just tying the engine down well.  If you need to test gimballing, or have
some other problem with the unwanted vectoring, you can put a water-cooled
collar up inside the nozzle to trip separation symmetrically.  Again, you
may need a bit of vacuum testing to verify vacuum operation, but the bulk
of the testing can be done at sea level.  The J-2 developers did this. 
If we feared danger, mankind would never           |       Henry Spencer
go to space.                  --Ellison S. Onizuka |

From: Henry Spencer <>
Subject: Re: engine reliability (was Re: Venturestar turnaround...)
Date: Mon, 27 Jan 1997 18:21:02 GMT

In article <5ch9kd$ip5$>,
Dwayne Allen Day <> wrote:
>: In that context, a rocket with a substantial number of engines can easily
>: tolerate a single failure at most times in its flight...
>This relies upon the assumption that the success rate for the engines
>individually is the same as the success rate for the engines stuck
>together.  This should not be taken for granted...

True.  Not only are there possibilities for interactions between them, but
there are common systems like the tankage which can fail (although the
engines are much the likeliest trouble spot, because they're more complex
and are under more stress).

>Wasn't there a problem
>with one of the early Saturn V flights due to "sympathetic vibration" or
>some other such term which essentially meant that the five big F-1s
>produced enough vibration to damage the fuel lines to one of them?

You're probably thinking of the second test flight -- Apollo 6 -- when two
J-2 igniter fuel lines (one on one of the second-stage engines, one on the
third-stage engine) ruptured because of vibration induced by flow through
the lines.  While there was a first-stage Pogo oscillation on that same
flight, which also needed solving, it is thought to have been unrelated to
the line ruptures.  The ruptures were fully reproducible on the ground,
without any external vibration, once the key environmental condition (run
the test in a vacuum chamber) was understood.  It must have been just the
luck of the draw, slightly different flow rates through the lines, that
the second flight got hit twice and the first not at all -- when the
ground tests hit the right flow rate for flow-induced vibration, in a
vacuum, the lines failed every time.

The F-1s themselves ran perfectly despite the Pogo oscillation.  While
nobody is sure why the Pogo oscillation didn't make itself obvious until
the second flight (it was detectable in the first-flight data, once the
engineers knew what to look for), there is one plausible speculation:
while both tests carried Apollo spacecraft with dummy LMs, the dummy LM
on the first flight was just a lump of ballast, while the one used on
the second flight made a much more careful attempt at simulating the
mass distribution of the real LM, and maybe this was significant.

Speaking more generally, it's true that interactions have to be taken
into account.  Still, for purposes of calculation this can be considered
to be just yet another independent failure source.

>That, indeed, was one of the problems with
>the N-1--there was never a full test of the first stage on the ground.

As I recall, the testing of the individual engines was also a bit skimpy.
"We don't care.  We don't have to.  You'll buy     |       Henry Spencer
whatever we ship, so why bother?  We're Microsoft."|

Index Home About Blog