$64 Billion and seventeen years to land on the moon. What's wrong with this picture?

In article ,
(Henry Spencer) wrote:

. . . since the engine is fighting less back pressure.

Ugh, the prof I had in undergrad for Compressible Fluid Dynamics and
Aerospace Propulsion HATED that term - the grimace on his face when
someone slipped up and used it was priceless. You're bringing back fond
memories, Henry! ;-)

--
Herb Schaltegger, B.S., J.D.
Reformed Aerospace Engineer
Columbia Loss FAQ:
http://www.io.com/~o_m/columbia_loss_faq_x.html

In article , says...

snip

The F-1 had horrible stability problems and motivated a good part of the
combustion-stability research done in the 60s. It was deliberately built
as the most straightforward possible scale-up of smaller LOX/kerosene
engines, in the belief that sheer size would be a big enough problem.
Unfortunately, stability generally deteriorates as size grows. It
appears that the size and design of the injector elements determines
the frequencies at which instability can occur, and the key question is
whether the chamber is large enough for any of those frequencies to
resonate within it. The F-1 chamber was large enough for a lot of
resonances... and the injector designers did not have much performance
margin to give away, which greatly complicated improving stability.

And they didn't have supercomputers that let them design an injector
plate in CAD/CAM and then run it through simulated burns to get an idea
of whether or not a given modification would fix the current problem or
create ten others.

Murray & Cox have an excellent chapter on the F1 instability problem --
it includes the fact that, by the end of the process, the F1 designers
were deliberately setting off explosive charges within the combustion
chamber while the engines were being tested, to *create* instabilities,
testing whether or not the engine would recover from them properly. And
there is the probably apocryphal story of the guy in charge of the
injector plate design, who supposedly got so frustrated with the
instabilities and the lack of data on what was causing them that he
started taking F1 injector plates and drilling holes in them at random,
then testing the plates on the test stand, just to see what would
happen...

Doug

Doug... wrote:
And they didn't have supercomputers that let them design an injector
plate in CAD/CAM and then run it through simulated burns to get an idea
of whether or not a given modification would fix the current problem or
create ten others.

Wouldn't have mattered if they *did* have the supercomputers because
they *didn't* have the data to feed the simulators.

That's the important part about these simulators folks often forget...
They only work as predictive or diagnostic tools in fields that are a)
non-chaotic (if not outright deterministic) and b) fairly well
understood. The F-1 combustion instability was neither, which is why
they tried more than a few random solutions on the basis of "can't
hurt, might work". In the end (IIRC) when they found one solution
that seemed to work, it was improved incrementally and empirically.

Given that it only took a few weeks to run a (far more complex) H-bomb
simulation on a primitive computer over a decade before, one does
wonder just how much computer power it would have needed in the mid
60's. (Had the problem domain been sufficiently understood.) The
main reason supercomputers are used for many problems is to obtain
useful results in a reasonable period of time, not the complexity of
the problem being simulated.

D.
--
Touch-twice life. Eat. Drink. Laugh.

(Derek Lyons) writes:
The
main reason supercomputers are used for many problems is to obtain
useful results in a reasonable period of time, not the complexity of
the problem being simulated.

Yes and no. Due to limited memory on non-supercomputers, the
performance curve is nonlinear. As problem size increases
(i.e. memory required to solve), solve times increase exponentially
once the problem grows so big it can't fit into RAM. The issue here
is that without enough RAM, you end up swapping to disk. The more you
swap to disk, the worse it gets since disk storage is far slower than
RAM. Because of this, some large problems really do require a
supercomputer.

We've got some HP computers with, I believe, 16 or 32 processors to
solve these sorts of problems (not sure about RAM). If you go to HP's
web site, you'll see that you can buy 128 processor machines with 1TB
of RAM (1024 GB). That's far more RAM than your typical high end PC
(which max out in the single digit GB range). These machines *rock*
for problems that require massive amounts of RAM. ;-)

There are problems that are "large" because they take a great deal of
compute time to solve, but can still fit in a "regular" computer's
RAM. For this class of problem, what you say is absolutely true.

Jeff
--
Remove "no" and "spam" from email address to reply.
If it says "This is not spam!", it's surely a lie.

(Henry Spencer) writes:

Right. Actually it can be either diameter or length; to first
order, what matters is what's called the "L*" of the chamber, which
is the chamber volume divided by the throat area. Throat conditions
don't depend all that strongly on chamber design, so other things
being equal, the L* tells you how long the gas stays in the chamber.
As noted above, you do pay a price for a high L*.

So would this explain why the russians went with multiple chambers on
their rocket motors? To keep down the L* ratio to a manageable
number, while also limiting the excess weight by only having one set
of pumps for both chambers?

Chambers are typically cylindrical, in fact, with the injector being
the top, and a taper to the throat at the bottom. Cylinders are
easier to make and simplify some issues, even though they are not
precisely optimal as pressure vessels.

Do they ever use multiple injectors? I assume there's one each for
the oxidiser and the fuel, but how about using multiple injectors
for each?

Typically there is only one injector, but that's a complex structure
with a considerable number of orifices for each propellant. In a
sizable engine, typically the injector will have at least several
dozen "elements", each of which is a set of two or more orifices
arranged so that the liquid jets impinge on each other (to speed
atomization and vaporization). (I neglect some less-orthodox
options for simplicity of explanation.)

So would the injector really be considered to not so much be a single
orifice pair, but more accurately the top of the combustion chamber,
with the throat being the bottom?

Quite so. Indeed, one of the big tradeoffs in injector design is
performance vs. stability. For example, having each fuel jet
impinge on an oxidizer jet gives rapid mixing and tends to improve
performance (or reduce the L* needed to get a given performance),
but it's bad for stability, probably because a pressure disturbance
affects the two jets differently (they will typically have different
densities and probably different velocities) and hence tends to mess
up the impingement. Having fuel impinge on fuel and oxidizer
impinge on oxidizer delays mixing but helps stability.

I really wish I could see some pictures of an injector, it would be
interesting to see how it all looks, just to get a better idea of how
it all works.

So having a fuel and oxidizer which have the same density could make
things easier, since the mixing would be simpler to have happen in a
quicker manner (smaller L*). But in the case of LOX and Hydrogen,
the large disparity in the densities of the fluids seems to have
worked out by forcing the designers to think out of the box. How has
this work affected later LOX/H2 engine design?

How stable was the F-1, J-2, RL-10 and other such engines?

The RL10 and J-2 were generally well behaved, perhaps because their (then)
unorthodox fuel, LH2, encouraged unorthodox injector designs that seem to
be unusually stable.

How did the design of the RL10 and the J-2 influence the design of the
SSME engine and it's injectors? I know the SSME is considered to be
high performance in some aspects, but a bear from a maint point of
view. So how much performance impact in terms of ISP would an
injector have over the total performance of an engine?

The F-1 had horrible stability problems and motivated a good part of
the combustion-stability research done in the 60s. It was
deliberately built as the most straightforward possible scale-up of
smaller LOX/kerosene engines, in the belief that sheer size would be
a big enough problem. Unfortunately, stability generally
deteriorates as size grows. It appears that the size and design of
the injector elements determines the frequencies at which
instability can occur, and the key question is whether the chamber
is large enough for any of those frequencies to resonate within it.
The F-1 chamber was large enough for a lot of resonances... and the
injector designers did not have much performance margin to give
away, which greatly complicated improving stability.

This brings up an interesting question to me. If the F-1 had more
performance margin than original, what else could they have done to
reduce the stability problems? It doesn't seem like reducing the L*
by increasing the chamber size would have helped much. Would they
have just gone to a less complete mixing strategy, accepting the lower
performance from incomplete mixing instead? That seems to be (to a
first order approximation) the only three tradeoffs you can make he

- smaller L*
- less mixing
- different mixture ratios of the Oxidizer and Fuel.

Most rocket engines aren't throttlable, and thrust is pretty much
constant once the ignition transient is over. (One exception: it
rises as the outside air pressure drops off, since the engine is
fighting less back pressure.) There is a little bit of
irregularity, which shows up as noise and vibration, but not a lot.

But isn't the SSME throttlable down to something like 80% of full? I
know it starts off at a supposed 104% of max, which has always bugged
me, since it seems that they're just pushing the performance at the
expense of maintenance. Commercial jet engines seems to always be
de-rated to get longer runnings times, since pulling an engine is
really expensive, but a little loss in absolute performance (thrust)
can actually gain you in terms of fuel costs as well. I'm not
explaining this well at all I'm sure.

The usual manifestation of instability, by the way, is not just
rough operation, but hot gas very rapidly burning through the
chamber wall. Rocket engines, even those that are cooled by
circulating fuel or oxidizer through the wall before injecting it,
rely very heavily on maintaining a layer of relatively cool gas next
to the wall, to reduce wall heating. Instability in the chamber
tends to mix things up and disrupt the cool gas layer, and the power
handled in a rocket engine is so enormous that the wall fails very
quickly when fully exposed to it.

This raises the $64,000 question to me as well. So while mixing is
good, you don't want the mixing to be too energetic, since that
disrupts the boundary layer. So how do they get this boundary layer
in the first place? Do they have fuel and/or oxidizer just pouring
down the sides (the outer ring of the injector?) so that it's
combustion won't happen until either it's outside the throat or into
the bell of the engine?

This is where some good diagrams/pictures of how they work would be
good. I've just got to get down to the Air & Space museum and take a
look at their cut-away engines I seem to recall them having.

Unfortunately, there are books about this stuff for the nontechnical
layman, and books for the rocket engineer, but not much in between.

Bummers. What's the most accessible rocket engineering book, Murray &
Cox's "Ignition"? I know that's a hard one to fine, and god knows
pages of equations make me swoon. I'm much more of a look at the
graph to understand how an equation works type of guy.

As usual Henry, you've been a font of knowledge, thanks!

John

In article ,
John Stoffel wrote:
...the L* tells you how long the gas stays in the chamber.
As noted above, you do pay a price for a high L*.

So would this explain why the russians went with multiple chambers on
their rocket motors?

No, that was a straight stability issue: Glushko was having instability
problems in his attempts to scale up his engines, and using four smaller
chambers rather than one big one was an expedient way around that, for a
program that couldn't wait.

Using a single pump set for multiple chambers is something that hasn't
been done much in the West, partly because Western designers historically
were not keen on engine clustering and preferred to build big single
engines. But there *are* a few examples of it -- in particular, certain
versions of Atlas had some of the booster-engine pump hardware shared
between the two booster engines.

(NB, there is disagreement about whether four chambers with one pump set
is considered one engine or four.)

Typically there is only one injector, but that's a complex structure
with a considerable number of orifices for each propellant. In a
sizable engine, typically the injector will have at least several
dozen "elements"...

So would the injector really be considered to not so much be a single
orifice pair, but more accurately the top of the combustion chamber,
with the throat being the bottom?

Basically correct... with the reservation that the injector isn't *always*
the top of the chamber. The Viking engine (on the old Arianes) used an
injector that was a ring around the perimeter of the chamber, and in
"pintle" engines, the injector is a structure sticking down into the
chamber from top center.

I really wish I could see some pictures of an injector, it would be
interesting to see how it all looks, just to get a better idea of how
it all works.

Generally it's a big plate with a complex pattern of many holes in it.
You really need to look at diagrams to appreciate how a particular
injector works; things like the slant of the orifices aren't obvious just
from a photo.

So having a fuel and oxidizer which have the same density could make
things easier, since the mixing would be simpler to have happen in a
quicker manner (smaller L*).

It would help a little, but the options are limited enough that other
issues generally drive propellant selection. In particular, both LOX and
hydrogen peroxide are significantly denser than essentially all fuels,
and there is just nothing to be done about it.

But in the case of LOX and Hydrogen,
the large disparity in the densities of the fluids seems to have
worked out by forcing the designers to think out of the box. How has
this work affected later LOX/H2 engine design?

For LOX/LH2, it's pretty universal to use "coaxial" injector elements,
where each element is a nested pair of tubes, with LOX coming through the
inner tube surrounded by hydrogen from the outer one. Note that I didn't
say "LH2" -- the hydrogen is typically a gas at injection time, because
it's been used to cool the chamber. So each LOX jet emerges in the middle
of a stream of relatively hot, fast-moving hydrogen gas. This is really
good for both atomization and mixing. It also helps with stability -- at
least some kinds of instability are droplet-based phenomena which occur
only if both propellants are liquid at injection.

The RL10 and J-2 were generally well behaved, perhaps because their (then)
unorthodox fuel, LH2, encouraged unorthodox injector designs that seem to
be unusually stable.

How did the design of the RL10 and the J-2 influence the design of the
SSME engine and it's injectors?

Very heavily. The SSME injector is close to a straight copy of that used
in the earlier engines. Some differences, of course, but same general
design concept. (Other areas, notably the pump system, differ a lot.)

...So how much performance impact in terms of ISP would an
injector have over the total performance of an engine?

What injector design (plus related details like L*) mostly determines is
(broadly speaking) the combustion efficiency of the engine -- how close it
comes to the Isp it theoretically ought to have. *That* is determined by
other factors, most notably the energy content of the propellants, the gas
properties of the combustion products, and how much the nozzle expands the
exhaust gas.

The SSME's high Isp comes from a highly energetic propellant combination
producing gas with good properties, plus a high-expansion nozzle (made
possible, in an engine that has to operate at sea level, by a very high
chamber pressure... which caused all sorts of problems).

This brings up an interesting question to me. If the F-1 had more
performance margin than original, what else could they have done to
reduce the stability problems?

Mostly, they could have accepted easily-found stability improvements that
happened to hurt performance a bit. For example, delaying the propellant
mixing a bit, until the droplets have more of a chance to vaporize, is
generally good for stability, but obviously hurts combustion efficiency
unless you lengthen the chamber to give more time for mixing.

A more specific example is that adding baffles sticking out from the
injector face, essentially breaking the injector face up into sections,
helps stability (given the right baffle configuration), by sheltering the
vaporization/mixing region just downstream of the injector from pressure
disturbances in the lower chamber. But the baffles are in a very harsh
environment, and they probably have to be cooled, e.g. by feeding a bit of
fuel (and only fuel) out through holes in them... which makes it harder
for that fuel to mix with oxidizer. The F-1 absolutely needed baffles,
but the early baffle-cooling schemes took 9s off the Isp. Slow, painful
improvements in baffle design and general injector performance got the
performance back.

It doesn't seem like reducing the L*
by increasing the chamber size would have helped much.

Careful he increasing the chamber size *increases* the L* (it's
chamber volume divided by throat area).

Most rocket engines aren't throttlable, and thrust is pretty much
constant once the ignition transient is over...

But isn't the SSME throttlable down to something like 80% of full?

Correct. The SSME is somewhat unusual, in this as in other ways. The F-1
and the J-2 both had only one throttle setting, wide open (although
experimental variants had some throttling capability).

Throttling capability was more common in Russian engines, mind you.

know it starts off at a supposed 104% of max, which has always bugged
me, since it seems that they're just pushing the performance at the
expense of maintenance. Commercial jet engines seems to always be
de-rated to get longer runnings times...

The main engine on a Delta II -- one of the most reliable expendable
launchers around -- runs at 153% of its original design thrust. (Maybe
higher now, that's a slightly old number.) The Delta IV's upper stage
uses an RL10 running at over 150% of the original design thrust. Any
second production batch of the Saturn V would almost certainly have used
the F-1A, which ran at 120% of the original F-1 thrust and was pretty
much flight-qualified by 1970.

There is nothing scary about improving an engine's performance as
experience builds up and incremental upgrades are made. This is
absolutely routine; even many commercial jet engines run at far higher
thrust than the original version of the same engine. Whether it was a
good idea to push the SSME thrust up without significant hardware changes
is a separate question...

...Rocket engines, even those that are cooled by
circulating fuel or oxidizer through the wall before injecting it,
rely very heavily on maintaining a layer of relatively cool gas next
to the wall...

This raises the $64,000 question to me as well. So while mixing is
good, you don't want the mixing to be too energetic, since that
disrupts the boundary layer.

You want very good mixing in *most* of the chamber, and little or none
right next to the wall. This is definitely a complication, with potential
for serious performance impact.

So how do they get this boundary layer
in the first place? Do they have fuel and/or oxidizer just pouring
down the sides (the outer ring of the injector?) so that it's
combustion won't happen until either it's outside the throat or into
the bell of the engine?

It's normal for the outermost ring of injector orifices to be all fuel, no
oxidizer, to set up a fuel-vapor "curtain" around the flame. (You can
also do that with oxidizer, but this has been less popular for several
reasons.) There is *some* mixing, but yes, this does delay or even
prevent combustion of some of that fuel, so you do see an Isp loss.

(Combustion pretty much has to be in the chamber, by the way. Once past
the throat, the gas temperature is dropping very rapidly and the gas is
moving very quickly, so there is essentially no chance for further
combustion before leaving the nozzle.)

Unfortunately, there are books about this stuff for the nontechnical
layman, and books for the rocket engineer, but not much in between.

Bummers. What's the most accessible rocket engineering book, Murray &
Cox's "Ignition"?

(That's actually mixing up two books -- Clark's "Ignition!" and Murray &
Cox's "Apollo".)

"Ignition!" is a readable and entertaining (!) book, but it's about the
history of rocket propellants rather than about rocket engineering per se,
although there is obviously some connection.

Turner's "Rocket and spacecraft propulsion" may be worth reading as a
gentle introduction (told from a European perspective). It's mildly
technical in spots but might be a good start.

The classic rocket-engineering text is Sutton's "Rocket Propulsion
Elements", now in its 7th edition (with a new co-author whose name I
forget right now).

...pages of equations make me swoon. I'm much more of a look at the
graph to understand how an equation works type of guy.

One book which followed that approach was Max Hunter's "Thrust Into Space".
But if you think "Ignition!" is hard to find, TIS is almost impossible.
--
MOST launched 30 June; science observations running | Henry Spencer
since Oct; first surprises seen; papers pending. |

John Stoffel wrote:
[...]
Unfortunately, there are books about this stuff for the nontechnical
layman, and books for the rocket engineer, but not much in between.

Bummers. What's the most accessible rocket engineering book, Murray &
Cox's "Ignition"? I know that's a hard one to fine, and god knows
pages of equations make me swoon. I'm much more of a look at the
graph to understand how an equation works type of guy.

John --

here's a short excerpt from my sci.space.tech booklist:

--------------------------------------------------------------------------------
A: Hill & Peterson
T: Mechanics and Thermodynamics of Propulsion
P: [publication data TBD]
Votes For: sstezel
Votes Against:
--------------------------------------------------------------------------------
A: Hyder et al
T: Spacecraft Power Technologies
P: [publication data TBD]
Votes For: chrisw
Votes Against:
--------------------------------------------------------------------------------
A: Tajmar
T: Advanced Space Propulsion Systems
P: [publication data TBD]
Votes For: --
Votes Against: chrisw
--------------------------------------------------------------------------------
A: Constantine & Cain
T: Hydrogen Peroxide Handbook
P: [publication data TBD] (R-6931, AD819081)
Votes For: lowther
Votes Against:
--------------------------------------------------------------------------------


The list is based on information I've gathered from the newsgroups,
including portions of lists other people have posted. I'll try to
post it here, but I put it up in sst most often, because the list is
oriented towards the technical side. (I'm starting a ssh list, but
most people are more familiar with those entries, so I'm not
hurrying).

/dps

Dave,

Thanks for the list, I've saved off a copy and I'll keep it for
reference.

John

(Henry Spencer) wrote:
But the baffles are in a very harsh
environment, and they probably have to be cooled, e.g. by feeding a bit of
fuel (and only fuel) out through holes in them... which makes it harder
for that fuel to mix with oxidizer.

Why only fuel?

D.
--
Touch-twice life. Eat. Drink. Laugh.

In article ,
(Derek Lyons) wrote:

(Henry Spencer) wrote:
But the baffles are in a very harsh
environment, and they probably have to be cooled, e.g. by feeding a bit of
fuel (and only fuel) out through holes in them... which makes it harder
for that fuel to mix with oxidizer.

Why only fuel?

D.

I would suspect because lox has a pretty low specific heat and you're
using the fluid as a cooling medium - that, and having oxidizer in
contact with your presumably-exceedingly-hot baffle would result in
rather rapid oxidation/corrosion/erosion of your baffle material.

But hey, it's been a really long time since my rocket propulsion classes
and I haven't thought about this stuff since then.(*)

*I think Henry thinks too much for any one human. Either: a) he's
really 73 different people all posting using the same account; b) he's
one man who uses time-warp technology to learn as much as 73 people all
at the same time; c) he's another of Pat's atom-brained zombies, but
just a really, really smart one! ;-)

--
Herb Schaltegger, B.S., J.D.
Reformed Aerospace Engineer
Columbia Loss FAQ:
http://www.io.com/~o_m/columbia_loss_faq_x.html