Index Home About Blog
From: Henry Spencer <>
Subject: Re: Saturn V Payload Question
Date: Sun, 30 Mar 1997 06:40:19 GMT

In article <RGpVXMAXkAPzEwP$>,
Kit  <> wrote:
>>J-2 could restart multiple times as long as the helium supply held out and
>>the Auxillary Propulsion System could seat the propellants in the tanks.
>I'm sure I cannot be the only lurker who would welcome a few words
>explaining how you start and restart a rocket engine and some of the
>problems involved.

Well, the details of the start sequence for the engine proper tend to be
black magic; even the usual books on rocket-engine design don't talk about
this much.  However, the fundamentals are straightforward.

You need to supply the engine with fuel and oxidizer, at suitable
pressures.  The pressure means pressurizing the tanks, sometimes with
vaporized propellant, sometimes with a separate gas like helium.  If
you're using vaporized propellant, you may need to use something else
temporarily during the startup sequence, until engine heat is available
for vaporization.

Making sure you get propellant and not pressurizing gas coming down the
pipes is tricky.  The simplest approach is just to use an auxiliary rocket
system, small enough that *it* can use techniques like bladders separating
gas and liquid, to provide enough acceleration to settle the liquids into
the bottoms of the tanks.

Liquid-hydrogen engines may need some initial small flow of hydrogen to
chill down their plumbing, so the hydrogen doesn't boil when it hits
warm pipes.  This generally isn't necessary with anything else, but with
hydrogen you usually need it.

If the engines are pump-fed, as all big engines are, you need to spin the
turbopumps up to speed.  They are usually powered by tapping off a small
fraction of the propellants and burning them separately, but depending on
details, that may not work until after the pumps are running, so you often
need some other source of turbine gas temporarily to get things started.

Then you have to introduce the propellants into the chamber in a precisely
timed sequence, and ignite them if they won't do that by themselves.
Single-burn engines often use chemical igniters, but this gets awkward for
multi-burn engines, and spark igniters (or self-igniting propellants) are
preferred despite having complications of their own.

Finally, at shutdown time, you may need to blow the plumbing clean of
leftover propellants and combustion products with helium or some such,
if you want to be able to do a restart later.

There are a hundred possible complications, but that's the basics.
Committees do harm merely by existing.             |       Henry Spencer
                           -- Freeman Dyson        |

From: (Henry Spencer)
Subject: Re: Starting a rocket in zero-gee
Date: Thu, 14 Jan 1999 04:57:58 GMT

In article <369bfebe$0$>,
Jim McCauley  <> wrote:
>...How do you start a liquid-fueled rocket engine
>in zero-gee, particularly when the fuel/oxydizer tanks are only partly

In a word, it's a headache.  But there are solutions.

The indirect solution, almost invariably used when restarting *large*
rockets, is to have a smaller rocket apply acceleration -- either
continuously (more or less; regular pulsed burns work about as well as
steady ones) or at least for a few seconds beforehand -- to settle the

Often the settling ("ullage") rockets are full-fledged rockets in their
own right, but sometimes one can get adequate settling by lesser means.
For example, the S-IVB kept its propellants settled by using its LH2
boiloff, vented through special directional vents, to provide a continuous
slight acceleration.  For that matter, it's possible in principle to get
adequate settling (although without a choice of settling direction) just
using air drag, if you're in a relatively low orbit.

Of course, if you're using honest-to-God ullage rockets, that just defers
the question -- now the ullage rockets have the same problem.  If you're
only going to need one ullage burn, you can just use solid rockets.  But
for multiple burns, you probably want to use liquids.  So...

In a spin-stabilized vehicle, the spin supplies settling, although perhaps
not in the direction you want.  Without spin...

The definitive solution is to separate tank-pressurizing gas from the
liquid with a flexible bladder or diaphragm, so the liquid is confined and
can't wander around the tank.  This is used quite extensively in small
rocket systems, and in ullage rockets for large ones.  There are several
problems, most notably the difficulty of finding bladder materials which
are adequately flexible, compatible with the propellants, and impermeable
to both propellant and pressurizing gas.

Particularly for long-duration operations, an alternative is to use
capillary action to confine at least *some* of the fuel, enough to get the
engine started and apply settling forces to the rest.  For example, just
putting a mesh dome over the tank outlet will trap some of the liquid.
Such designs can become quite complex, including solid or perforated
baffles, narrow channels, etc., as well as mesh surfaces.  (Some of the
Mariners were particularly messy cases, because their propulsion systems
were upside-down at launch!  As were the Apollo LM tanks during SM engine

If you are pressurizing the liquid with its own vapor -- not uncommon for
cryogenic fluids -- then it's possible to design engines that can run on
a gas/liquid mixture, at low thrust, long enough to settle the liquids.
(This can be tricky, especially for pump-fed engines, but it's possible
even for them.)

In principle you can run ullage rockets on propellants stored as gases,
but the pressure tanks get very heavy very quickly.  There have been
designs which store one ullage burn worth of propellant as gas, and refill
the ullage tanks by boiling some of the liquid during the main engine burn
(which sets a minimum burn length, sometimes a problem).  I don't *think*
this has ever been done for ullage rockets in an operational system,
(although the boil-enough-for-one scheme was used for turbine spin-up in
the J-2 restart system).

Finally, for cryogenic fluids stored roughly at their boiling point, you
can use a phase-change heat exchanger to extract a small flow of gas, and
use that in ullage rockets.  Run the exiting gas/liquid mixture through a
pressure drop, and then through a heat exchanger bonded to the tank wall.
Lowering the pressure also lowers the boiling point, so bringing the
exiting flow to tank temperature will reliably boil any liquid in it...
which cools the tank while you're at it.

There are some other weird methods that are usable in principle, like
using electrostatic forces to move liquids around, but I think the above
is the full list of things that have been done in practice.

>In zero-gee, wouldn't the fuel tend to float around in globs or stick
>to the tank walls in random ways?

Usually the most stable configuration is liquid lining the walls with a
big gas bubble in the center, but much depends on detail, and on the
presence of small forces like air drag.  And it can take a long time for
the fluid to reach the most stable configuration if it starts out in a
different one.

>I presume that there's just one tank
>outlet, presumably at the "bottom" (with respect to the normal line of

True in essence although not in detail -- multi-engine rockets sometimes
have multiple tank outlets, but in general they're all on the bottom.
The good old days                   |  Henry Spencer
weren't.                            |      (aka

From: Henry Spencer <>
Subject: Re: AGENA docking rocket (Project Gemini, c1960s)
Date: Mon, 10 Jun 1996 16:49:47 GMT

In article <4oj7ui$> (Dwayne Allen Day) writes:
>[ullage rockets] ...But if you think about it, this 
>is a less than ideal solution, since you only have so many of these 
>things and they get used up every time you start the engines...

Actually, ullage rockets can themselves be liquid-fuel rockets, with a
restart capability.  You might think this just defers the problem, but
there are solutions to propellant settling which work for small rockets
but don't scale up well, like putting a flexible diaphragm between the
pressurant gas and the propellant.

For example, the S-IVB used solid ullage rockets for its initial burn
(to put the Apollo stack in orbit), but it also had a liquid-fuel
attitude-control system for use in parking orbit, and that was used to
settle the main propellants for TLI.

>...the rather brilliant idea of putting a mesh screen 
>down near the engine inside the tank.  It allows propellant to flow 
>through to the nozzle, but surface tension (I think) keeps it from 
>floating away from the nozzle...

Actually, there are a variety of such schemes, but Dwayne has captured
the essence fairly well.  Yes, surface tension is used to retain liquid
within the screen.  It's not perfect; in particular, with fluids like
LH2 which are at their boiling point, you can get gas generated *inside*
the screen by boiloff, and that complicates life.  That sort of thing
is still a research topic, but for storable liquids the technique
is in routine use, under buzzwords like "propellant management devices".

>Therefore you always have fuel at the 
>nozzle ready to ignite.  Now I don't know the specifics beyond that, like 
>how you get the fuel the slight further distance to the igniters once you 
>open the valve (it could be under slight pressure beneath that screen)...

The tanks are pressurized, so the gas pressure in the tank forces the
liquid into the engines once you open the valves.  That's easy; the only
tricky part is making sure you don't get some of the gas forced down the
plumbing too.  This isn't too hard, with PMDs, if the flow rate is low,
like for attitude control.  For big burns that suck a lot of fuel fast,
it's harder, and the common approach is to use PMDs to get things started
and rely on the burn itself settling the fuel before gas starts to get
into the plumbing.  (Another question for big burns is getting your pumps
started, of course.)
If we feared danger, mankind would never           |       Henry Spencer
go to space.                  --Ellison S. Onizuka |

Index Home About Blog