February 4 – 15 Notes

 

Sorry for skipping an update last weekend, ran out of time…

 

Propellant experiments

 

We spent some time very cautiously investigating a potential propellant combination.

 

A Russian rocket engineer that had worked at Energomach sent me a document with his thoughts on X-Prize vehicles.  He covered a lot of ground, some that I agreed with, and some that I didn’t.  One particular thing he mentioned, however, was quite interesting:

 

Propellant: unitary fuel  mixture of 50% water solution of hydrogen peroxide (H2O2) and 8% of ethanol. It is well-forgotten unitary fuel developed by Germans in WWII and containing the oxidizer (H2O2) and fuel (ethanol) in one. When it burns, H2O2 is decomposed by catalyzer producing the mixture of water steam and oxygen at temperature about 450C that ignites ethanol. It was designed for gas-generators, but it can also be used for rocket engines. It provides the theoretical specific impulse of I = 205 (kg thrust/kg sec) at 30/1 ratio. Practically, counting all thrust chamber losses; it could be about 180  190 (at 30/1 ratio). It is the steachiometric combination, so the exhaust contains mostly water steam and just a few percents of carbon dioxide. Gas temperature inside of thrust chamber is about 800  850C (about 1000  1100K). The fuel is stable, decomposing at 150C and freezing at  30C. This fuel is much better than pure H2O2 that can provide the specific impulse of 140  150 only. The propellant and exhaust are environment friendly and do not contain poisonous and dangerous components.

 

One fact in here is clearly wrong: 50% peroxide decomposes to a temperature of basically 100C, failing to boil all the water.  Also, mixtures of high concentration peroxide and alcohols are very very dangerous explosives (we have tested them, and the detonations are truly something to give one pause), but with that much water to de-sensitize it, it might be a viable propellant.  A little looking around dug up this passage in “Ignition!”:

 

At any rate, peroxide is still used as a low-energy monopropellant, and will probably continue to be used in applications where its high freezing point isn't a disadvantage.

 

One such application is as a propellant for torpedoes. (After all, the ocean is a pretty good thermostat!) Here it is decomposed to oxygen and superheated steam, the hot gasses spin the turbines which operate the propellers, and the torpedo is on its way.  But here a little complication sets in.  If you're firing at a surface ship, the oxygen in the turbine exhaust will bubble to the surface, leaving a nice visible wake, which nor only gives the intended victim a chance to dodge, but also tells him where you are.  BECCO (Buffalo Electrochemical Co.) came up with an ingenious solution in 1954.  They added enough tetrahydrofuran or diethylene glycol (other fuels could have been used) to the peroxide to use up the oxygen, letting the reaction go stoichiometrically to water and carbon dioxide.  The water (steam) is naturally no problem, and CO2, as anybody knows who's ever opened a can of beer, will dissolve in water with the help of a little pressure.  That solved the wake problem, but made the stuff fearfully explosive, and brought the combustion temperatures up to a level which would take out the turbine blades.  So BECCO added enough water to the mixture to bring the chamber temperature down to 1800F, which the turbine blades could tolerate, and the water dilution reduced the explosion hazard to an acceptable level.

 

We thought it was worth a try, so we prepped one of our tiny 1” cat pack engines and biprop test chambers for some very very cautious experiments.

 

It turns out that our catalyst packs cannot effectively decompose 50% peroxide.  Even with a very low flow rate, most of the liquid comes out with very little decomposition.  Even with perfect catalyzation, 50% would still leave the water content liquid, but there should be quite a bit of obvious gas flow as well.  If a little shot of peroxide was pulsed into the engine, it would fully cook off eventually, but any kind of a flow would quench it.  This is rather to be expected with so much water taking up valuable catalyst-contact surface area.  Liquid catalysts can work much more effectively with lower concentration peroxide, but then you have all the plumbing of a biprop, with only the performance of a monoprop.

 

Since auto-ignition didn’t seem to be at all viable (which we rather expected), we tried a couple solid propellant ignitor slugs to see if we could get combustion going.  We didn’t have any luck with a couple tests, and we are scared enough of the dangers that we aren’t going to pursue it any farther.

 

Also on the peroxide front, just for completeness, we made contact with Solvay about supplying high concentration peroxide.  They used to produce 85%, but after our local sales rep checked around, she found that they do not have the capability at all anymore.  FMC and Degussa seem to be the only large scale producers in the world.

 

 

Hatch and Nose

 

We got the hatch reinforcement welded to the side of the cabin, and the crush cone mounted.  We also have the hatch sealing lip and the hatch itself ready, but we probably won’t install those until we have the full sized tank to mount the cabin on for pressure testing.  The hatch reinforcement needs to be strong enough that the hatch isn’t a buckling point in any axis, which basically means it has to weigh nearly three times as much as whatever you cut out of the base cone for the hatch.  We used ¼” thick plate for the reinforcement.

 

http://media.armadilloaerospace.com/2003_02_16/crushConeInstalled.jpg

http://media.armadilloaerospace.com/2003_02_16/hatchAndCone.jpg

 

Working with the big vehicle is starting to get challenging because of the size.  We are moving into a new facility in a couple weeks, which will allow us to set up some dedicated areas with scaffolding and multiple hoists for manipulating the vehicle.

 

The crush cone is 0.050” thick aluminum (except for the 12” hemisphere nose, which is 0.125”), secured to the 0.125” thick cabin cone with 31 ¼-20 nutserts, a couple inches above the main pilot bulkhead.  The cone weighs 37 pounds, and should be the only thing discarded between vehicle flights.

 

 

Big Crush Test

 

We did our full size, full weight drop test this Saturday.  To drop a 2400+ lb load, we built a custom electric release mechanism.  We needed to do this for drogue and main releases on the full size vehicle anyway, so this was a good opportunity to test them out.  We used the same trunk release mechanisms we are using for the small vehicle drogue releases, but we built a “mousetrap” load amplifier that gives a 15:1 lever arm.  The small releases were able to handle 850 pounds without breaking (although they did need 24 volts to release at that weight), so a pair of amplifiers like this are more than adequate for the worst abort drogue loading on our vehicle design.

 

http://media.armadilloaerospace.com/2003_02_16/mousetrap.jpg

 

The design worked perfectly through all the tests.  This one is set up to hang between two D-rings for crane or helicopter drop tests, but the vehicle ones will be machined integrally into the bottom tank flange closure.

 

We mounted the 50g accelerometer on a small piece of foam with hot glue, and stuck it to the pilot bulkhead in the vehicle.  The hope was that a little bit of isolation would remove the large shock vibrations that we saw in the small vehicle drop tests.

 

We had spent a while discussing ways to rig up something to drop the vehicle from a sufficient height, but we finally decided to just rent some heavy equipment for the day.  Joseph has experience working with it, and the several hundred dollar price was well worth it.

 

We welded four very strong D-rings onto the inside of the boilerplate tank end we have the cabin mounted on, and used high strength ½” chain for everything.  We used 50 pound sandbags for ballast, loaded into the tank end before the drop test, with the total weight measured by a scale above the latch release.

 

The drop test was done with 2400 pounds, with the nose ten feet above the ground, for a 0.79 second drop time and an impact speed of 25 fps.  These are ballpark figures for our X-Prize vehicle’s landing configuration.

 

Prep:

http://media.armadilloaerospace.com/2003_02_16/dropPrep.jpg

http://media.armadilloaerospace.com/2003_02_16/dropPrep2.jpg

 

The video:

http://media.armadilloaerospace.com/2003_02_16/bigDropTest.mpg

 

Aftermath:

http://media.armadilloaerospace.com/2003_02_16/crushedCone.jpg

 

Basically, everything worked perfectly, except for one thing – we didn’t have enough pressure vents, so the cone acted like an air-spring as well as crumpling.  At about 1/3 volume crushed, the inside pressure may have been as high as 7 psi, which equates to over four tons on the back of the pilot bulkhead.  Our flox fillets popped off, and the entire bulkhead shot to the bottom.  You can see this in the video, as the crumpling finally changes to a buckling, then the bulkhead pops up visibly through the hatch.  We had a large area of epoxy bond holding the bulkhead in, but pushing it down towards the wider part of the cone put it in tension, and it obviously wasn’t enough.  Surprisingly, the bulkhead isn’t damaged at all, this 2” honeycomb paneling is some tough stuff.  We are going to do a more aggressive surface prep on the aluminum when we bond it back in, and we are also going to weld some custom brackets above and below the bulkhead to mechanically hold it in next time.  We are going to add several 1” diameter vents in the crush cone, so it shouldn’t see an overpressure anyway, but we are going to triple-fix the problem.

 

Accelerometer data:

http://media.armadilloaerospace.com/2003_02_16/theBigDropAccels.gif

 

The accelerometer still had the back and forth vibrations, so I’m still not sure if they are a sensor issue or a true reading.  If I scroll back in the data, there is a nice, clear transition from 1G to 0G when the release is triggered, then the drop time before the first impact.  The initial impact to squash the top dome and start the cone crumpling is higher, then the cone crumples at a fairly constant rate, then the bulkhead pressurizes up and pops, absorbing most of the remaining energy, then the cone tips over and bounces once.

 

We are going to re-test in the same configuration without the overpressure, as soon as a new crush cone is fabricated.  It should continue the even crush quite a bit farther without the air-spring effect, and we need to make sure it still absorbs all the energy.

 

Drogue Ejection

 

We have our drogue ejection system for the small vehicle basically worked out.  One of our priorities is to be completely pyro free, and also minimize the unique consumables for flights. After some discussion on the aRocket list, Rick Weber graciously offered to fabricate a system for us that would use the little 12 gram CO2 cartridges.  He did quite a bit of design and experimentation work, but in the interest of getting us something quickly, he built a simple adapter that punctures the CO2 cartridges and converts to 1/8” NPT, which was plumbed through a solenoid.  He tested it in a piston of the size we were going to use, and sent it off to us.

 

When it arrived, we hastily put together a piston out of some HPR components Phil had lying around, and tried it out.  Our initial results were very poor – the drogue didn’t get out of the tube at all.  We tried both orientations for the CO2 canister, to see if the liquid vs gas CO2 injection mattered, but it didn’t.  We finally hooked a nitrogen bottle up to the solenoid, and found that even at 600 psi, it still didn’t punt the drogue to our satisfaction.  The issue was that Rick’s test used a machined piston with an O-ring, which sealed well enough to build up pressure.  Our slip-fit piston just let most of the gas leak out without building up pressure.

 

We finally swapped out the little solenoid Rick had provided for the largest one we had, and this time 200 psi nitrogen popped the drogue out, and 600 psi did it with some authority.  It is possible that the large solenoid would have let the CO2 cartridge pop the drogue, but we didn’t test that combination.  We are still going to make a good piston out of aluminum with O-rings, but since we aren’t very weight constrained, we have decided to go ahead and just use a small paintball tank filled with nitrogen for the ejection.  I have a quick connect on the bottom bulkhead for this, so we can just pressurize the ejection tank before we fill the main peroxide tank.  There is a pressure transducer under the little nitrogen tank, so we can monitor any leakage, and I can have a warning light to check before launch.

 

http://media.armadilloaerospace.com/2003_02_16/bottomBulkhead.jpg

 

http://media.armadilloaerospace.com/2003_02_16/drogueEjection.mpg

 

 

Electronics bulkhead

 

The new electronics bulkhead is ready for hover-testing the vehicle, but we are still going to move to more capable actuator boards and power supplies before free flight testing.

 

http://media.armadilloaerospace.com/2003_02_16/electronicsBulkhead.jpg

 

Battery lugs are soldered on, and the main system has dual batteries.  We aren’t going to have the same issue that crashed us last time.

 

I am still having some issues with the compact flash based drive system.  The Ampro bios sometimes needs a second power cycle to get it to boot from the CF, and, more disturbing, I seem to get an occasional error while doing disk access.  My previous attempts to use CF would always give errors to the point that I couldn’t get my Linux system installed, but moving to a Synchrotech CF card (instead of the Kingston cards I had) let me get things running.  However, I had some corruption during my first installation, and now on the second installation (after reformatting with a full check for bad blocks) I still get the compiler crapping out every once in a while with a Signal 11.  If anyone has any relevant advice, please let me know.  I am probably going to go back to the IDEFLASH drive that never gave me any problems, but it is a shame to stack another board when the Ampro has the CF slot built in.

 

One issue I ran into this time was that using 50k ohm resistors for the voltage divider that is necessary to monitor a 12v battery on a +/- 10v ADC didn’t give a strong enough signal, giving weird values that changed with the connection of other channels.  Finding some 1k resistors fixed the problem.