Aug 4, 2005 notes
Despite not having time to do an update for a while, we have
been steadily working
When we last worked with it, the setup showed what seemed to
be a valve lash problem flow would begin when the high pressure ball valve
reached 15% open, but it wouldnt shut off until it was closed all the way back
to 5%. Since we had fabricated our own
actuator to valve adapter, we thought we might have allowed too much lash into
the coupling. We built a new mount using
helical beam couplers with zero lash, but that turned
out not to help. The coupling seems tighter,
with the valve following every little jitter of the actuator, but the flow
behavior seems to be an aspect of the seals in the ball valve, not the linkage
between the actuator and the valve.
This cracking problem is only really an issue at very low flow
rates, so we were able to do some flow tests at roughly the performance levels
that our single-man space shot vehicle will use. With a single large nitrogen bottle feeding
the servo regulator, we did the following test:
2700 psi initial bottle pressure
60 gallons of water at 230 psi and
215 gpm flow rate
1800 psi final bottle pressure
2 plumbing, 1 valve
The small fittings at the bottle valve became the limiting
factor as the pressure dropped below about 2200 psi,
with the servo valve eventually going wide open and still not quite being able
to keep up. Our flight vehicle pressurant tanks will manifold directly out of bottle necks
with a -10 fitting, so they wont become flow limited at all. When our new 36 hemispheres arrive, we will
be welding up the full tankage and pressurization
system for the big vehicle and doing water flow tests in preparation for
testing a 5,000 lbf class engine.
Speaking of spheres, here are a couple pictures of the tear
area on the burst one:
I was recently looking at the table in Sutton regarding
losses due to small chamber to throat contraction ratios, and they weren't as
significant as I had remembered them. A chamber with no contraction ratio
at all will lose 20% of its thrust due to pressure losses from accelerating
gasses in the straight section, but the Isp
loss is only 1.5%. The text mentions "throatless
rockets" being used in some missile applications to minimize chamber
length and dry mass at the expense of Isp. The text doesn't say if these were liquids or
solids, but I assume they were solids.
However, this does open up the question of building liquid engines like
that. If L* remained constant, you would have an extremely long engine
that would probably be impossible to cool, but I could imagine the
accelerating, high speed flow could reduce required combustion stay times
significantly. A 1.5% Isp
loss is utterly meaningless for our purposes, so a configuration that traded
that for fabrication benefits could be quite useful.
We fired a few crude throatless
lox / ethanol chambers, and the results were surprisingly encouraging. With a very crude injector (a spray nozzle
for the lox and four straight horizontal jets for the ethanol), we measured a
190 Isp from a 12" long
straight pipe combustion chamber. It melted in a couple seconds, but this
was still very impressive. With a 3:1 expansion
cone added, performance should increase about 15% to around 220 Isp. That would be right at
theoretical values, and MUCH better than we have been seeing in our engines so
Side note: it turns out that our flow distribution to cooling
channels and injector ports has been Really Bad with our previous
designs. The test that demonstrated this dramatically for us was to cut
off the top of an engine so the cooling channels were exposed, and flow water
through it. With our original manifolds,
there was a 2:1 difference in height between the highest and lowest flowing
channels, and the high point
moved around the engine as flow rates changed.
We are now using taper milled manifolds that maintain a constant flow
velocity around the ring, and flow rates are essentially the same. Unfortunately, this doesnt seemed to have helped the engines in any perceptible way.
Our second revision to the engine didnt work out so well. We wanted to incorporate two of our new
tapered flow manifolds with the same injection points we have been using on the
other engines, but I wasnt about to try machining them out of stainless to
weld onto expendable pipe stages.
Instead, I machined the engine top parts out of brass, and we tried
brazing them together and onto a section of straight pipe.
This didnt work out for two independent reasons. Most obviously, because the brass top section
had a smaller ID than the stainless pipe below it, the step section served as
an excellent flameholder, and the pipe burned itself
off right under the injector before the rest of the pipe even started
glowing. Second, the burn-off wasnt
completely even, so we flowed water through the injectors and found out that
almost half of the LOX holes werent flowing anything at all. We cut the engine up and found that brazing
flux had snuck in and plugged most of them up.
Watching the disposable chambers glow red hot gives us lots
of good information on the evenness of our burning and the required L* of the
engine, but we decided to go ahead and make a regen
cooled pipe engine, because we could just make the entire thing out of
aluminum. A 2 aluminum pipe with only
light external machining can be a slip fit inside a 2.5 tube section, which
lets us make these engines without any boring at all.
The first test engine was a bit shorter than the
expendable ones, with a 12 total engine pipe length, and 9.5 of cooling
channels below the injector. 20 cooling
channels of 3/16 width, tapering from 0.030 deep at the nozzle end to 0.060
deep at the injection point.
wasnt good. There was about 30% less L*
on this motor, but it may turn out that our crappy, cobbled together injector
for the expendable engine was actually a lot better than the 20 hole version we
moved to. We have been using very low
velocity injectors because we had a lot of smooth running engines even without
high pressure drop, and I feared that relying on the pressure drop for
stability would give us problems when deep throttling, However, the hand-made injector did
have four very high velocity streams impinging reasonably accurately, and that
may have been the key to the performance, more so than the straight tube nature
of the engine.
Since we didnt have much else to do with the first regen tube engine, we made a couple more full throttle
runs, then did a run without any ethyl silicate mixed in the fuel. At about the 25 second point, fuel flow rose
and the engine note changed. We had
burned through one of the cooling channels.
That was the type of back-to-back confirmation that I was looking for,
showing that the ethyl silicate really is making a difference, and we arent
just putting it in as a random hope. We
are currently using 1 oz of ethyl silicate per gallon of ethanol.
The next test engine has much higher injection velocity, and 3 more chamber length. The great thing about these engines is that
it only takes me two nights to machine the parts, so we can test two engines a
week if necessary.
We got to see a new failure mode on this one the internal
chamber buckled, then melted through. We
have been making the outer jacket a tight press fit over the inner chamber, but
the thermal expansion of the chamber, coupled with the extra pressure drop, apparently
caused the inner chamber to buckle instead of stretching the jacket.
The next test engine will use a pressure drop intermediate
between the last two, use a looser fit outer jacket (inner machined to 2.345)
to allow a bit of thermal expansion, and will bring the propellant impingement
points farther out towards the side. I
used to have issues with wandering drill bits during injector drilling, but now
I am spotting everything with a larger diameter carbide spotting drill, and
manually applying cutting fluid to every hole as the mill runs. The 1/32 holes in the last engine came out
If this line of tube engine development works out, we can
make a 5,000 lbf engine with very little more effort
than the test engine.
Hold down test
We swapped out the valve actuators on the vehicle from 1.5
second to 0.5 second speeds. On the
previous vehicles we had moved to slower valves to reduce the bobbing at hover
effect that we got from the overshoots due to latency between moving the valve
and vehicle acceleration, but now that all the motor drives use PWM for variable
speed, I can control this much better in software. Having valves that shut off fast is good in
general 1.5 seconds feels like a very long time when you notice something is
on fire and you want to shut down the engine.
We moved the flow meters from the test stand to the vehicle
in preparation for calibrating throttled mixture ratios in-situ. This involved some contorted plumbing, and we
are still having some leakage problems.
I finally got around to finding a supplier (Aircraft Spruce stocks them)
of the little conical seal caps you can put on AN
fittings to improve sealing, so hopefully that will help. Plumbing 1 and over is much more difficult
to seal than smaller stuff. We are
moving more and more towards welding practically everything together. It is tempting to move to elastomer
seals for the fuel side sometimes.
Another thing we have started to do is put some red food
coloring in our ethanol when it has been mixed with ethyl silicate, so we dont
accidentally forget it, mix too much in, or confuse fuel drained out of tanks
with pure fuel. This has the side
benefit of leaving evidence where fuel leaks have been, even if the ethanol has
We welded the gimbal arms on the
latest converging / diverging engine, and mounted it up under the vehicle. We fired up the engine and gimbaled it around
for a few seconds, which should be more heat load than the vehicle will see
during liftoff and landing on an actual flight.
There was a little bit of fire on the bottom of the legs silica
insulation, which we think was some of the RTV that holds it on burning, but it
could also have just been methanol from the engine that came out during the
initial very rich throttle up. The main
heat shield and the rest of the legs were cooler than we expected, we probably
worried more than necessary.
The one surprise was that the vehicle clearly pulled the
support chains taut, which we didnt expect at the pressure we were running. With propellant, the vehicle was over 400
pounds, and at the pressure we ran it, the engine should have made under 350 pounds of thrust.
My working theory is that the big flat heat shield allowed the vehicle
to get some ground effect lift since it was only 3 off the ground.
I spent a while double checking the GPS integration with the
new electronics box orientation, and getting the startup self-test functioning
with the new electronics layout. Since
the exact same electronics box runs both the test stand and the vehicle, we
have had a few problems with forgetting to add the command line parameter to
specify the vehicle valve calibrations.
To fix that, I made a stub connector for one of the sensor values that
isnt used on the vehicle (the load cell sensor) that just ties the signal line
to ground, so I read a hard zero from that channel. I now use that to fail the self test if the
state of that line doesnt match the specified configuration.
The vehicle is ready to fly, as soon as we are comfortable
with the throttling and reliability of our engines. Worst case, we can just make an engine with
really crappy Isp and still
do our flights while we are figuring out how to make a reliable, higher Isp engine.