September 27, 2003 Update
Mixed monoprop progress
I skipped last week’s update partly because I was out of
town all day Sunday when I usually write them, and partly because I was
frustrated at the results of the day’s testing. Things are better now.
We finally had all the pieces together to do high quality
engine development – high purity 50% rocket grade peroxide from FMC, high
purity methanol, laser cut spreading plates and pack plates form Global
Stencil, and a big batch of fresh rolled-foil catalyst from Catalytic Products.
The catalyst we had the most success with so far was a 200
pore-per-square-inch rolled foil corrugated catalyst. The 5.5” diameter catalyst was originally 3.5” deep, but used in
that form it provided little activity, with most of the propellant being
channeled straight through the pores by the gas from reaction at the pore
edges. When we cut the catalyst into
three sections a bit over 1” deep and interspersed a couple screens between
them to break up the channeling, we finally got some good runs. However, the length of the runs was always
limited – our standard test run of 9.6 liters was always starting to cloud up
at the end, and looking up at the catalyst through the nozzle with a mirror
showed areas that were no longer red hot.
There was also some degradation in general reactivity over the span of
many runs, which may have been due to poisoning or stripping.
Because it was a relatively long lead time item, I
speculatively ordered a lot of catalyst, hoping that we would get the
combination right and be able to build a complete set of engines for the big
vehicle without having to reorder. I
ordered 12 1” thick sections 5.5” diameter catalysts with 400 PPI, and one
speculative 12” diameter, 2” thick 600 PPI catalyst that we could try using in
our big engine.
http://media.armadilloaerospace.com/2003_09_27/packParts.jpg
The engines start with a spreading plate at the top with
250+ tiny holes. In previous engines,
we had always added some spacers at the top to give more volume for liquid to
spread out before getting to the spreading plate, but we were hoping to fit 2 x
1” deep catalysts with all the various screens and pack plates inside our
existing engine shells, so we were tight enough on space that we went without
any extra spacers. There is a slope at
the top of our chambers, and the plate flowed fine in early tests, but this has
proved problematic because after welding and running the engine, the spreading
plate gets some warping in it, and it can pop up right into the top of the
engine, almost closing off the flow. In
the future, we are welding positive standoffs to the plate to keep it from
closing off at the top, but a little extra space is probably still a good idea.
We have found that it takes 10 screens (20 mesh 304
stainless) after the spreading plate to completely even out the flow across a
high-density monolith. With less
screens, you can clearly see a grid of hotter spots during the preheat,
indicating uneven distribution. We might
be able to reduce this by pushing for even more tiny laser cut holes in the
spreading plate, or using higher density screens.
After the screens goes the first catalyst, followed by a
stainless screen and one of our cruciform pack plates welded in to support it.
We did an open-catalyst test run with a single 1” thick
catalyst first. We observed a few
interesting things: we preheated the
pack to dull-red heat, then when we started the propellant flow there was uncatalyzed
liquid raining out fairly evenly from the entire pack, but the pack bottom
significantly increased in temperature to bright orange, then cooled down to
black as the liquid rain increased.
This largely confirms that there is a hot gas boundary layer in the vertical
channels, because liquid below 250 F was streaming out of nearly 2000 F
channels that were increasing in temperature, rather than being cooled by the
passing liquid.
We added the second 1” catalyst brick, another screen, and
another pack plate. It didn’t quite fit
in the engine shell, and had to be welded a bit above flush. This test run still had liquid streaming out
, but it stayed hotter longer before cooling down. We learned several things:
even though this had more catalyst surface area than our previous
configuration (2” deep by 400 PPI vs 3.5” deep by 200 PPI), it catalyzed worse,
allowing liquid to flow through even a freshly preheated engine. The 0.030” thick pack plates warped severely
under thermal expansion when heated, such that the middle section was providing
no support to the catalyst above it, making it rely on transferring the forces
through the welded edges alone. The 304
screen above the catalyst pack melted badly.
We had seen a few small holes burned in the screens in some of our
previous tests, but we weren’t sure if it was from the firing, or a preheat
that got a little out of range. We saw
it melt away on the video this time, so there is no question. Our temperatures are higher now because we
are using the peroxide from FMC that is really exactly 50%, instead of the
warehoused Solvay food grade that was actually closer to 40%. The screen was there to provide a little
escape for the catalyst pores occluded by the bars on the pack plates, but it
isn’t all that critical.
http://media.armadilloaerospace.com/2003_09_27/meltedScreen.jpg
While this was still performing worse than the old
configuration in open-catalyst form, we went ahead and bolted a nozzle to it
and gave it a try. It didn’t work very
well.
We welded up an extension to hold a third catalyst brick and
did some testing, then added a fourth brick.
Each test got harder to properly preheat, but did run somewhat longer
before giving out. I think it is clear
that the pack simply cools from the top to the bottom during a run, with the
hot catalyst giving up some heat to the propellant flowing past it, but the
propellant not reacting and producing heat until it has flowed some distance
down the pack, so the only way to get heat back up in the pack is by conduction
axially along the couple-mills-thick metal foil, which doesn’t get very far.
With four catalyst bricks we did make our best runs ever to
that point, hitting 145 s Isp with 11 second runs at about 300 lbf, but there
were always signs that it was about to go away at the end, and our last test of
the day clouded up and gushed liquid after only a few seconds, which we
attributed to not getting the entire catalyst pack properly preheated.
http://media.armadilloaerospace.com/2003_09_27/quadPack.jpg
We assembled the big 12” engine with the single 600 PPI catalyst,
in hopes that the even finer pore size would avoid the channeling, but to our
disappointment but not great surprise, there was lots of straight-through
channeling even at fairly low flow rates.
The conclusion we are forced to is that straight through monolithic catalysts
just suck for monopropellant use. I was
under the impression that some catalysts of this design were used in hydrazine
thrusters, but they are just for attitude thrusters, so they may be of much
lower flow rate for a given catalyst volume than we are dealing with. They have room-temperature active catalysts,
so quenching also isn’t an issue.
http://media.armadilloaerospace.com/2003_09_27/bigRoll.jpg
http://media.armadilloaerospace.com/2003_09_27/weldingBigSpreader.jpg
http://media.armadilloaerospace.com/2003_09_27/bigFlowTest.jpg
If we made a really deep (multiple layer) and wide catalyst
pack of the current designs, we could probably do some useful vehicle testing
with 15 second burns, but there is no way it can deal with the 80+ second burn
we need for the X-Prize flight profiles.
On Tuesday, we tried a different approach. We took the chamber with the Bete
impingement atomization nozzles that we had been trying to get liquid catalyst
injection / torch ignition to work with, and plumbed it up to just spray
atomized propellant down onto the extension with the bottom two catalyst bricks
from Saturday’s tests. The hope was
that the fine spray would combust immediately on the top of the catalyst, and
the open area above the catalyst would allow some turbulent convection to keep
the heat of combustion at the top, instead of migrating only down the
pack. It didn’t work.
We only had a few more things we could think of to try. We had found a local business (Galco) that
does electron beam metal deposition, which we were planning to have platinum
coat some of the stainless screens, but after seeing the screens melt inside
our engine, we canceled that idea.
Screens would positively stop any axial channeling through the pack, but
they probably wouldn’t help the quenching much. I was looking around for 316 screen discs, but in examining other
options, I thought about using nickel screens.
Nickel has the drawback for us that we can’t use nitric acid to clean
the catalyst pack if there are any nickel components, because it dissolves it
pretty fast. However, it does have a
high melting point, it has better thermal conductivity than stainless, which
might help the quenching, and it is easier to cut if we have to make the discs
ourselves. After considering nickel, I
realized that we could just use the nickel foam that we already have on hand. The silver plated nickel foam was always far
and away our most active catalyst for 90% peroxide, but we had two significant
problems with it: the silver stripped
off, limiting the life to only 100 seconds of firing or less, and the foam didn’t
have enough structural strength to keep from getting progressively squashed
under the backpressure. Our current
practice of welding in support plates in several places through the pack
relieves the total compression felt by each part of the catalyst, and the
e-beam people tell us that platinum deposited on a nickel substrate Is Going To
Stick (much better than on stainless), so this may be an exciting option. Unfortunately, the raw platinum material
only showed up on Friday, so we didn’t have this to test on Saturday.
The other major direction we had was to try and get heat
into the propellant before it got to the catalyst, so the catalyst wouldn’t be
giving up so much heat just to raise the liquid temperature. A regeneratively cooled motor would do this,
but our regen chamber for the 5.5” motors is made out of aluminum, so it wouldn’t
be able to survive the preheating process with no liquid flowing through
it. We talked about just brazing or
welding some coils of tubing to the outside of the stainless engine chamber,
but I wasn’t confident we would get enough heat transfer.
What we finally tried was fabricating milled plates to fit
on the top and bottom of the monolithic catalysts so that propellant is
directed down the outside quarter inch of the pack, then pack up the next half
inch, then finally down the central four inches. This back-and-forth flow allows full combustion heat to conduct
back to a cooler section of the catalyst by just going through the thin walls,
rather than along them.
I made a couple test plates in aluminum, but I had an
absolutely miserable time making the final 316 SS plates. My order for new carbide end mills didn’t
make it in by Friday, so I was stuck with a couple used mills that broke one
after another during the fabrication, eventually leaving me using an extended
length end mill designed for aluminum, my very last 3/8” end mill, to do the
final cuts. I also had to change all of
my depth plunges to sloped entries, because even when the mill didn’t jam when
plunging into the 316, it often pushed the end mill a bit up into the collet,
changing the depth between cuts. I
finally finished at 7 am, got some sleep, and headed in to the shop.
http://media.armadilloaerospace.com/2003_09_27/plates.jpg
http://media.armadilloaerospace.com/2003_09_27/stacked.jpg
http://media.armadilloaerospace.com/2003_09_27/assembled.jpg
We had some issues with getting even flow distribution
through the outer annulus on the top plate after welding it in, but we got it
reasonably close. We later had some of
the some warping-up-to-the-top problem that we saw with the spreading
plates. We should have added standoffs
to the top of the plate as well.
We put two catalyst bricks in after the top plate, without
anything between them. The pores don’t
line up, so it should still break up the channeling. Previously, we had turned an entry relief into the engine
chambers and pressed the slightly-oversize catalyst bricks into the chamber
with the hydraulic press, a press plate, and some persuasion with a screwdriver
at the edges. Today we found that a
combination of sticking the catalyst in the freezer and torch heating the
chamber allows them to slide right in with finger pressure, a big improvement.
The bottom plate welded on with a little bit of room to
spare. In hindsight, we should have
added a 0.010” spacer above the top plate to hold it off from the roof a little
more.
In the open-catalyst test, there was liquid escaping past
the bottom plate guide wall into the center, which wasn’t too surprising
without any kind of seal, and the catalyst darkened somewhat towards the
center, likely due to the cruciforms on the plates, but the big news was that
most of the catalyst didn’t cool off during the run.
We put a nozzle on it and did a test run. It started ok, but rapidly took on a cloudy
plume. However, it stayed in this state
for almost 20 seconds without degrading farther. This seemed to show that it was operating in a steady state after
the extra heat at the top and side of the chamber was sucked out by the
propellant.
We added an extension to the chamber and packed it with the shredded-metal
catalyst that we tried (very unsuccessfully) a couple months ago. The primary purpose is to act as a
flame-holder to let the rest of the propellant combust along with the hot
gasses from the main catalyst pack, but the extra catalyst certainly doesn’t
hurt.
We made the first run a repeat of the previous test, and it
worked absolutely perfectly. There wasn’t
even a hint of cloud even at startup, and the exhaust was completely clear the
entire run. The chamber extension got
bright red very rapidly, to the point where we were actually a little concerned
that it might come apart with a high pressure run.
We hose-clamped a thermocouple to the outside of the chamber
and did a run with the same amount of propellant, at a higher pressure. The thermocouple peaked at 1100 F, which
means the metal was somewhat hotter. This
is still acceptable.
We then did our largest-ever mixed-monoprop run, at five
gallons of propellant and regulated high pressure. It worked perfectly through the entire run, over 15 seconds at just
under 400 lbf. The thermocouple
temperature did stabilize below 1100 F.
http://media.armadilloaerospace.com/2003_09_27/extendedChamberBurn.mpg
We took the engine apart to see if we could get by with a
much shorter extension chamber. The loose
catalyst bale had been compressed down to half the original height, which is
obvious in the video by looking at the heating pattern, so it clearly doesn’t
need to be that huge. Much of the
powdery catalyst coating had also been worn off already.
We made a few tests with the short extension, with one run
scrubbed due to a bad preheat, one run that went fine, but may have been
clouding at the end, and then a nozzleless experiment. We need to do more tests at this size, but
first we need to work out a production technique for initiating the preheat,
since our manual torching of the bottom or sides of the engine is hackish and
not very consistent. We have some good
ideas to try next week.
Big Vehicle Landing Gear
The X-Prize configuration is probably not going to have
landing gear, but powered vertical landing is still our future direction, and
we may yet be forced into using it just to minimize the maximum possible range
the vehicle can travel under any failure.
We also really want to repeat some of the DC-X flight tests with the big
vehicle. Big wire rope isolators aren’t
the most mass efficient designs, but they provide three axis energy absorption,
and can handle very high heat.
The mounting and bracing of the big wire rope isolators is
pretty clunky, and we would have done it differently if we had all the pieces
at the start, but it should be ok for a while.
We are a bit concerned about the loads going into the tank through the
strut mounts, which are only ten square inches each, but the tank is pretty
damn strong. The next one we build, I
want to just run the struts up the side of the tank and bond them there, which
has a much more benign failure mode – shear off instead of punch into the tank.
We are finding ourselves limited by the main beam height in
our shop when working on the big vehicle.
We recently added a big eye bolt plate right in the middle of the
vehicle top bulkhead just so we could have another foot and a half of clearance
over hanging the vehicle under cables from the side points. We can now just barely get our hanging scale
above the vehicle, which showed we have a current weight of 1460 pounds after
welding all of our bracing and mounting points around the bottom and mounting
all the (small) engines and plumbing.
The bigger engines and valves for X-Prize flights are somewhat
compensated for by removing the landing gear, but there is still a fair amount of
mass in the drogue cannon, main chute, and cabin gear. It still looks reasonable to come in at 1800
pounds dry, plus 600 pounds of passengers / ballast.
We did a couple drops of the vehicle from a little bit off
the ground to see how it behaves. The
isolators are sized more for a full-up 2400 lb vehicle, so there is a lot of spring
in them with the current weight, but we are satisfied with how they work.
http://media.armadilloaerospace.com/2003_09_27/mountedIsolators.jpg
http://media.armadilloaerospace.com/2003_09_27/lifting.jpg
http://media.armadilloaerospace.com/2003_09_27/upright.jpg
http://media.armadilloaerospace.com/2003_09_27/shortDrop.mpg
News
