August 6 and 10, 2002 Meeting Notes
On Tuesday, we tested a new biprop injector design. There was some interesting discussion on
aRocket about it:
We are going to be trying a new injector with our peroxide / kerosene engine tomorrow. Our 2" cat pack engine has so far achieved good combustion efficiency with ethane (when it lights...), but while the
kerosene lights instantly, the Isp has only been slightly better than monoprop mode, indicating very poor combustion.
The two injectors that we have tried so far are a single point stream that shoots directly across below the catalyst pack with about a 0.030" orifice, and a 2" annular ring with a 0.005" gap (orifice
is upstream). The ring was mostly for gaseous ethane injection, because the required tolerance for decent kerosene velocity would be difficult to machine, and
would probably change too much with thermal expansion anyway.
The single point injection delivered better results with a smaller jet on our initial 1" cat pack motors than on the 2"
motors, so we were sort of resigned to having to make more injection points with smaller holes. Some recent discussion about the virtues of fuel mixing in high
speed gas streams made me consider that perhaps we just need to make the cat pack discharge faster moving when we are injecting the fuel.
Our current biprop engines have a 16:1 contraction ratio from the cat pack to the nozzle throat. Our monoprop engines use 9:1. but we shrank the throats on the biprops to increase L* in a given length
At 16:1, the gas moving past the injector isn't moving all that fast. I have made a new injector that necks down underneath the cat
pack from 2" to 1", which will have the gas moving four times as fast at the single injection point, and have the opposite wall closer, which
may also have a positive effect.
There are two things I am pondering:
Should I care about smoothly transitioning back to the 2" chamber diameter? On this injector, I just left it an abrupt transition. The turbulence after the transition will probably be good for mixing, but there may be some absolute pressure loss, although with the temperature tripling after the transition with the resulting volume expansion, I
don't have a good intuitive feel for pressure losses in that situation.
How fast is fast enough? In theory, I suppose you could neck down at the injection point to a diameter even smaller than the nozzle
throat, and have sonic velocity gas flowing by, because combustion will increase the gas volume before the real nozzle. I suppose you would want to stay below the speed where compressible flow starts giving irreversible
losses. The cutoff profile probably ties into this, where a smooth profile with nothing but subsonic flow might be better suited to extreme contractions.
Making a venturi does increase motor length, but if it allows big motors to be made with only a small number of large injection points, it could be a notable benefit. I am distinctly leery about tons of
tiny injector holes, for several reasons.
When I worked at Marquardt I worked on a very similar project. It was a 3"ID, 1000 lfb 85% Peroxide / JP8 thruster. After reading your description of the low C* efficiency, it brought back memories of how difficult development of our thruster was.
There are many things you can do to increase the combustion efficiency. Some of which you already mentioned.
> The two injectors that we have tried so far are a single point
> stream that shoots directly across below the catalyst pack
We also found this method of injection results in very poor mixing.
> the required tolerance for decent kerosene velocity would be
> difficult to machine, and would probably change too much with
> thermal expansion anyway.
Thermal expansion won't be a problem. Constant fuel flow will keep the injector passages at a reasonable temperature.
> The single point injection delivered better results with a smaller jet
> on our initial 1" cat pack motors than on the 2" motors, so we
> were sort of resigned to having to make more injection
> points with smaller holes.
If the jets are too large, they don't break up enough. If the jets are too small, they don't have enough penetration. We found that impinging the larger fuel streams dramatically increased C* efficiency. This is much more effective than simply
using more, smaller diameter jets.
> recent discussion about the virtues of fuel mixing in high speed gas
> streams made me consider that perhaps we just need to make the
> cat pack discharge faster moving when we are injecting the fuel.
This is true, but try to keep the mach number below 0.4. Otherwise there will be significant viscous losses.
> Our current biprop engines have a 16:1 contraction ratio from the
> cat pack to the nozzle throat. Our monoprop engines use
> 9:1. but we shrank the throats on the biprops to increase
> L* in a given length motor.
> At 16:1, the gas moving past the injector isn't moving all that fast.
> I have made a new injector that necks down underneath the cat
> pack from 2" to 1", which will have the gas moving four
> times as fast at the single injection point, and have the
> opposite wall closer, which may also have a positive effect.
Design to a chamber gas velocity of Mach 0.2 for optimum
> There are two things I am pondering:
> Should I care about smoothly transitioning back to the
Not if you are at Mach 0.4 or lower.
> On this injector, I just left it an abrupt transition.
This works as a boundary layer tripping device. The pressure drop won't be significant if the Mach number is low enough but it significantly enhances mixing of the boundary layer. OTOH your chamber will
also experience a higher heat transfer rate.
> Making a venturi does increase motor length, but if it allows big
> motors to be made with only a small number of large
> injection points, it could be a notable benefit.
The benefits are not worth the added effort unless the gas velocity is very high. Mach 0.6+.
> I am distinctly leery about tons of tiny injector
> for several reasons.
Use the larger diameter impinging fuel streams and you will obtain much greater efficiency.
Another alternative is to design the combustion chamber similar to that of a ramjet combustion chamber and use a fuel atomizer with a flame holder. As built, it weighed more than the others, but we achieved
nearly 96% C* efficiency with this combustion chamber design.
> John Carmack
On Mon, 5 Aug 2002, John Carmack wrote:
> Should I care about smoothly transitioning back to the
> diameter? On this injector, I just left it an
That is essentially what the Brits did on their peroxide/kerosene engines: the fuel injector was a plate across the chamber, with big holes (they didn't even round the edges) for the gas from the catalyst pack to pass through, and a ring of small fuel orifices around each gas
hole. There were eight holes, each (by eye) no more than 1/8 the diameter of the plate, so the total gas-flow area was only a small fraction of the chamber cross-section, with no attempt at smooth transitions.
They did find it necessary to drill some small gas holes in the parts of the plate which didn't have fuel circulation, so that a small flow of decomposed peroxide would keep the downstream flame off the plate. You might watch out for heating problems on your injector base.
> How fast is fast enough? In theory, I suppose you could neck
> down at the injection point to a diameter even smaller than the
> nozzle throat, and have sonic velocity gas flowing by...
At least some of the non-peroxide work I referred to in earlier mail did that, I think, but it's probably serious overkill.
From left to right: venturi injector, single point injector,
annular injector (hard to see the ring), copper gasket, standard retaining
plate, experimental retaining plate, clamp ring, 2 catalyst chamber, 2
nozzle, 2 aluminum biprop chamber (water flood cooled).
All tests with the one liter of peroxide, 1/4" ball
valve on the peroxide side, temperature was over 95 F, and all tests were at
roughly 250 psi regulated tank pressure.
We tried the ethane first, to see if the venturi injector
would help it light more reliably, but still no luck with that:
With the old (recompressed loose pack) bench test motor:
We switched to one of the new highly compressed catalyst
packs for future tests.
On the first run, we forgot to put the ethane jet back in,
so it was running with about 6x the ethane it was supposed to have. Not surprisingly, it didnt light.
We ran it again with the proper jet:
We gave up on the ethane, and had much better luck with
kerosene. Quoted Isp figures here are oxidizer
Isp, since we dont have a flow meter on the kerosene. True Isp will be 15% to 20% lower than this,
given a 7:1 optimal O/F ratio.
Theoretical peak Isp for 90% peroxide and kerosene with a 200 psi
chamber pressure expanding to atmospheric is around 215.
We increased the peroxide jet size, hoping that the highly
compressed pack wouldnt run rough when the peroxide flow is cut down in biprop
34 lbf thrust, smooth
47 lbf thrust, a little rough in biprop mode
47 lbf thrust, a little rough in biprop mode
We wanted to see what it looked like leaning out the
39 lbf thrust
That was as lean as we could run the jet, and it was still
burning smoothly, although it was obviously less powerful from the noise level.
We then started running it rich:
We were out of time, so we stopped there. Given the relationship between these oxidizer
Isp and real Isp figures, we were probably past the true Isp peak, at around 164
seconds. This was somewhat better than
we had gotten with the ethane on the rare occasion that we got it to ignite,
and a drastic improvement over the previous kerosene injectors, so the high
speed flow at the injection point is definitely a big improvement.
From 250 psi tankage, the best we can realistically hope for
is a 200 Isp at sea level, and even that is a bit optimistic. The low design pressure is based on light
tanks and radiatively cooled engines, but if we wind up with a different
cooling strategy and custom tanks, we could increase the chamber pressure and
get another 10% or so.
Our radiatively cooled chamber is finally ready for testing.
From left to right: uncoated TZM chamber, silicide coated
TZM, burned-through aluminum chamber, current aluminum chamber.
This has the classic R512 silicide coating used by NASA
since the late 60s for oxidation protection of refractory metals. The standard material for radiative chambers
is usually C-103 columbium, which we may switch to in future engines. TZM (molybdenum) has somewhat better strength
at temperature, but has a ductile-to-brittle transition temperature of only 60
degrees F after thermal cycling, so it isnt considered a good material for
thrusters used in space. It shouldnt
be a problem for us.
Radiative engines trade expensive materials for simplicity
of operation. The raw material was $44
/ lb from H.C. Stark, and we contracted the machining out, because it was a
bitch to turn. The coating was twice as
expensive as the base fabrication, but it should only scale weakly with
increasing size, being mostly labor and process based. The coating looks like paint primer, and has
a slightly rough texture to it. We are
being careful about scratching it, although it does have a metallic bond to the
base metal, and is somewhat self healing.
Our target of 200 psi chamber pressure is on the high side
for a radiatively cooled engine, so we added some additional drilled holes to
the injector to let some unburned catalyst exhaust provide a film cooling
blanket around the outside, and we made sure we were at a rich O/F ratio.
We had an interesting set of issues with our testing.
Things obviously got very hot when the engine had been
running at an orange glow for ten or twenty seconds. We started insulating our fuel solenoid after the first couple
We were expecting to warped the brass retaining ring, which
did happen, and contributed to bending the retaining bolts. We started putting
the bolt heads (socket head) on the chamber side and the nuts on the catalyst
pack side, which helps some with rigidity.
We need to make a stainless steel retaining ring.
The brass injector did not seem to suffer from the firings,
but it is a really heavy hunk of metal.
The injector should generally be ok, being cooled by the catalyst
gasses on the side opposite the flame.
The radiative chamber has an interesting benefit in that you
can see where combustion is taking place by how quickly it heats up to orange
heat. Our injector is clearly getting a
lot more fuel on the bottom side, opposite the injection point, even with the
faster venturi mixing. The next
injector we make will have two opposite impinging streams, and an even smaller
venturi. It looks like we still have 15%
combustion performance or so to pick up, and this may help.
The completely unexpected problem we had was that we couldnt
get a good seal between the injector and the TZM chamber. All of our brass and aluminum chambers have
either been bolted directly against a brass injector, or just had a copper
flange gasket in between, and we have not had any leakage problems. Every run we had resulted in various degrees
of gas leaking out past the chamber, and eventually bursting into flames. We tried one of our copper gaskets, then two
different Cotronics high temperature sealants, but it never held a seal. The high temp thread sealant worked best,
holding a seal for the first five seconds or so of the burn, then letting
things leak after that. The combination
of the rough surface texture of the coating, and the drastic thermal issues are
making this challenging. We are going
to try some alumina-silica gasketing next, then move to an energized metal
o-ring if necessary.
We started testing with the rich jetting, which should be
nice and safe:
It ran rough in biprop mode, meaning that our highly
compressed pack had lost its exceptional smoothness, and seemed to be reverting
to the same characteristics as our previous packs.
A little rough on the first fire, but smooth on the second.
Acceptable burn, but thrust is down to 39 lbf, which means
chamber pressure is down 25%. We dont
mind at the moment, as it is less stress on the chamber.
We tried a few runs with the ethane, still with no ignition:
We went back to kerosene (still fighting the sealing
problems over all these runs):
Rough in biprop mode.
Still a bit rough.
The pack seems to be worse than the old one for smoothness now.
(although this is very low flow rate)
Still moderately rough.
We had graduated our kerosene sightglass, so we could tell that this
flowed about 600 ml of kerosene for the 2000 ml of peroxide, which, at 0.81
kerosene density and 1.40 peroxide density, is a 5.8 O/F ratio, moderately
We did have over 60 seconds of hot fire on the chamber
without any problems, but the sealing problems and roughness keep it from being
a completely satisfactory day of tests.
(amusing shot of us blowing out the burning rocket engine
at the end
The seated lander is completely reassembled, and ready for
testing. We have moved away from
aluminum AN fittings where possible, because we have started noticing pitting
in the flares on fittings that have been exposed to peroxide for long
periods. We attempt to use 316 SS for
everything now, but we still have a couple larger custom fabricated parts (tank
thread adapters and the main distribution manifold) that are aluminum.
I need to find a supplier of stainless steel male NPT to
swivel female AN/JIC fittings, which we use between our valves and engines,
mated with a ground down AN male fitting holding a restrictor jet. All our other fittings are now stainless.
We added some additional foam bars underneath the wire rope
isolators to give it a little more cushion on landing, but they may well get
ripped off with any side speed. I think
the best bet is going to be getting some large wire rope isolators made with
custom wire sizing, so we can get a larger stroke with the same basic weights.
Joseph built us a nice cradle for the tube vehicle. We can use this for working on the vehicle
in the shop, and for transporting it to launch sites. It is large enough to handle our next two planned vehicles.
Our new parachute arrived, but the spun nosecones
havent. I am getting quotes for
filament winding nose cones and tail cones for 2 diameter vehicles