June 21, 2003 Notes

 

Big Engine Mounts

 

We debated the exact engine mounting strategy for the big vehicle for some time, but we settled on a simple solution that wastes a bit of weight, but works out conveniently.  The initial test flights will be with 5.5” diameter engines with 2” diameter throats, but the final engines will be much larger.  We built mounting plates that clamp to the bottom of the tank manway flange on the inside, and are supported by two ¾” thick studs on the outside.  To attach to the tank, we cut ¾” coupling nuts at an angle, and welded them to 3” by 3” squares of perforated metal.  These squares will be bonded to the tank with epoxy / chopped glass fiber.  The engines will be canted with angled shims above the mounting plate, just like the smaller vehicle.

 

http://media.armadilloaerospace.com/2003_06_21/base.jpg

 

I learned some lessons machining the mounting plates.  Since I had to make four of them (and will probably have to make more when we crash the vehicle…), I wanted to let the CNC program run completely automated, which meant no tool changes on our mill.  I was a little surprised when the 0.5” and 0.75” diameter holes bored with a 3/8” end mill came out distinctly non-circular.  Tool deflection was a bigger issue than I expected – none of the previous parts I had made actually required tight tolerances like that.  I wound up having to cut the holes in two passes and at a very slow rate, but in the end it did work out exactly how I wanted, where I could just clamp a plate on the mill, start the program, and come back to a finished plate.  I am going to buy a bunch pf short, stubby end mills to minimize deflection for hole milling in the future, but bringing the spindle down to right over the plate does make work piece clamping more difficult.

 

 

More 50% Firings

 

We learned a lot, but we don’t have a solid solution yet.

 

We received another 2” thick, 6” square block of impregnated ceramic catalyst from CPI, as well as a 4” thick block, and an uncoated 2” thick block.  It was fairly easy to rough-cut the block to cylindrical shape with a band saw, then get it to final shape on a bench grinder.  We saved the scraps for various tests.

 

We got a new supply of nitric acid for catalyst cleaning, so we tried to clean the catalyst we had been using for the earlier tests.  Even after significant soaks at fairly high concentrations, it never regained the activity that the brand new catalyst displayed.  At the time, we thought there might be some poisoning that the nitric wasn’t removing, but based on later tests, we now think that there is a degree of mechanical stripping of active catalyst.

 

We found a local supplier of 50% concentration food grade peroxide, and purchased a drum.  In drum quantities, food grade was $1/lb, compared to $0.50/lb for technical grade.  We are still looking for a supplier of semiconductor grade at 50% concentration.  I finally got around to ordering a TDS (total dissolved solids) meter for testing peroxide, so I will report some numbers next week.

 

To test the poisoning of the catalysts, we took two chips of the catalyst and let them each soak in 100 ml of 50% peroxide, one technical grade, the other food grade.  They both caused lots of bubbling, but the food grade sample cooked all the peroxide off faster.  We then placed them in another 100 ml each, and the difference grew more pronounced.  On initial immersion, the temperature of the food grade sample was higher, and when we checked on it again an hour later, the food grade sample was reduced to completely still water, while the technical grade sample was still slowly foaming away with some peroxide left to decompose.  We removed the samples and did eye dropper tests of fresh food grade peroxide on both, which showed the one that had been immersed in technical grade to be noticeably slower to react.  They both still catalyze immediately, but the technical grade sample isn’t nearly as vigorous.  We should have contrasted it with a completely fresh chip of catalyst, but we didn’t think about it at the time.

 

We fabricated a big combustion chamber to sit between the catalyst and the nozzle, in hopes that extra volume would get a more consistent performance, and also to give us an extra half inch to put inert screens between the spreading plate and the catalyst.  After all of our tests, we are now of the opinion that post-catalyst combustion volume is of no value with this combination.  In the few references we have seen to 50% peroxide work, it is mentioned that the reaction is not self-sustaining without the presence of a catalyst, which would also argue against the value of post-catalyst space.

 

http://media.armadilloaerospace.com/2003_06_21chamber.jpg

 

We added 16 inert stainless screens above the catalyst to fill out the space.  All of our test firings were very smooth today, even with no explicit metering orifice to provide an injector pressure drop, which is visible as the thrust curves dropping linearly with tank pressure.  This was a large difference from the incredible chuffing that we had on our first successful tests, but we aren’t sure if it is just because of the higher flow rates, or because of the inert screens at the top of the pack.

 

We had some crud on the top flange, so we had a leak there on most of our tests.  It may be a permanent gouge, so we will probably use a copper gasket for the next test.

 

These firings were much louder than normal monoprop firings in the 100 – 200 lbf range, but the Isp only calculated out to be in the 80-100 range, which is a little puzzling. There may be a core of high velocity gasses, with a jacket of unreacted, low velocity gasses bringing down the Isp.  In theory, the Isp should be around 150 even at very low chamber pressure.  We aren’t expecting any good Isp until the burn stays cloud-free the entire time, but the noise level was unexpected for the low performance.

 

We did runs at both 200psi and 400psi tank pressure, with basically similar results.  For long runs, we could get it to start with a nice clear exhaust plume, but after a few seconds it would cloud up.  Interestingly, the thrust continued to follow the tank pressure, without any inflection as the exhaust clouded.  We have seen behavior like that before on 90% monoprop motors without any metering jets that had catalyst deterioration, as the Isp drops, the flow increases to compensate for it.

 

http://media.armadilloaerospace.com/2003_06_21/50firing.mpg

 

One supposition was that as the engine heated up, the expansion allowed more flow to bypass around the side of the catalyst.  We removed two of the screens at the top and added a Teflon encapsulated silicone O-ring.  We barely put any crush pressure on it, because we were unsure of the strength of the ceramic catalyst.  The O-ring didn’t improve the engine behavior at all, and when we opened it up, the catalyst web had crushed some under compression, shearing off vertical fragments.  If we need to seal against the catalyst, we will probably have to cast a solid ceramic outer layer around the porous catalyst to give it enough strength to compress the O-ring.  I was a little surprised that we didn’t melt the O-ring at all with our torch preheating before the run.  This implies that the top of the pack isn’t getting all that hot, even when we make the bottom of the pack glow red hot.

 

We tried adjusting the methanol content to be notably richer, but that run was much worse, with no flame at all, and lots of wet clouds.

 

Someone mentioned that the food grade peroxide drum had some fine print on it that said “40% to 60% concentration” under the big 50% marking, so we decided to test the density to see if we might be working with notably weaker peroxide.  The food grade peroxide came out to 1.15 g/cc, and the technical grade came out to 1.17 g/cc.  We should have explicitly measured the temperature, but it was probably about 85 degrees F, or 30 degrees C.   The book value for 50% peroxide at 25 C is 1.1914 g/cc, and the value for 40% is 1.1487 g/cc.  The food grade peroxide is only about 41% concentration, while the technical grade is about 46%.  Someone probably got a raise for making it company policy to dilute to the lower limit…  We are certainly going to check each drum we buy in the future, because a 60% mixture could be dangerous, as well as skewing the O/F ratio.

 

Density table from p 199 of Hydrogen Peroxide by Schlumb, Satterfield, and Wentworth:

 

Wt % H2O2    0C                   25C

0                      0.9998             0.9971

5                      1.0193             1.0145

10                    1.0393             1.0324

15                    1.0598             1.0507

20                    1.0804             1.0694

25                    1.1014             1.0885

30                    1.1226             1.1081

35                    1.1441             1.1282

40                    1.1661             1.1487

45                    1.1883             1.1698

50                    1.2110             1.1914

55                    1.2342             1.2137

60                    1.2579             1.2364

65                    1.2822             1.2592

70                    1.3071             1.2839

75                    1.3326             1.3086

80                    1.3589             1.3339

85                    1.3858             1.3600

90                    1.4136             1.3867

95                    1.4421             1.4142

100                  1.4709             1.4425

 

Since the peroxide wasn’t as concentrated as expected, all of our mixture ratios were running on the rich side, which hasn’t helped, but probably isn’t the core of our problem.  The fact that the runs can start clean, then go cloudy implies that the catalyst pack may need to be at a higher temperature than the steady state reaction temperature, and only functions initially due to the torch heating.  A longer pack may or may not help with this, we have seen very odd behavior with 90% peroxide and silver screen catalysts, where a pack of a certain length would never catalyze completely even on tiny amounts of flow, but adding another quarter inch of catalyst all of a sudden allowed huge amounts to catalyze perfectly.  We can also experiment with higher volume propane torches for preheating, and possibly a flow-through the top heating arrangement that would give much more thorough heating.

 

Another thing we noticed on disassembly was that the new catalyst was less active than the first pack (that we had cleaned with nitric acid), so we may still be experiencing some poisoning from the food grade peroxide.  We do believe that there is some degree of mechanical stripping of the catalyst going on as well as a degree of poisoning.  The fresh catalyst probably has some poorly bonded bits of catalyst that offer high surface area initially, but are rapidly washed away under high pressure liquid flow, leaving the strongly bonded catalyst with less total surface area.  We could test this by doing running five gallons of distilled water through a new pack instead of peroxide, and seeing if the pack then has reduced activity compared to its corner scraps.

 

Future tests:

 

See if nitric acid cleaning the second pack brings it back to the activity level of the first pack that was cleaned.  If true, it means that the food grade peroxide is still poisoning to some degree.

 

Make a test run with proper O/F ratio for our 40% peroxide to see if that behaves any better.

 

Fabricate a chamber that will let us test with the 4” thick catalyst.

 

Investigate the rolled metal foil catalysts.

 

Investigate more thorough pre-heating options.