April 2 and 6, 2002 Meeting Notes

 

In attendance:

 

John Carmack

Phil Eaton

Russ Blink

Joseph LaGrave

 

Tuesday

 

Last week’s biprop testing had eaten a section of the inner web out of the brass catalyst retaining pack, so I made a couple out of stainless for testing. Cutting them apart was a big hassle, so I added a dozen stainless spreading plates and retaining plates to our CNC order with DynaTurn.

 

Russ and Phil had our current water cooled chamber nickel plated over the brass, in preparation for running it in regenerative form with peroxide instead of water providing the cooling.

 

The cat pack that was borderline on Saturday took a really long time to completely clear today, so we decided to rebuild it with more doubled up silver screens, which seems to be working well now. The total count is now 70 silver screens and 50 stainless screens, plus a few spacers.

 

We fired with kerosene and the new pack without any problems, but we couldn’t get the ethane to light again, for reasons we do not understand.

 

We cut the new rotor blades.

 

Saturday

 

We began using new test stand software with several new features that make our lives easier: dynamic bias zero, remote pulsing, constant display of pressure and thrust, and the ability to take multiple logs before exiting. I need to buy a big A/D system so we can start measuring a lot more things: peroxide flow rates, kerosene flow rates, coolant temperature, wall temperature, etc.

 

We did all the bonding on new rotor blades, we will be drilling the bolt holes next week. The blades are set at 20 degrees of pitch this time, up from 10 degrees last week. This will be higher than optimal for static testing, but more appropriate if we do any vertical acceleration.

 

The fat fan blade extrusions from a different supplier arrived. They have over a 2” bar hole, which would make the hub attachment trivial, but the extrusions aren’t completely straight. We may ask about getting a hand-selected set with better characteristics.

 

The big set of tests today was working up towards regenerative cooling on the test engine.

 

We started out by calculating the peroxide flow rate that we have for a long, clean biprop run (500 ml in 12 seconds), and turned the water down until it matched. Our normal coolant flow for long runs was higher than the equivalent peroxide flow. We don’t have temperature instrumentation, so we ran the water exit hose behind the trailer and let someone with a carefully calibrated gloved finger tell us how hot it was getting. On a monoprop run, the water got about as hot as a hot water faucet. With a biprop run, the water did get to the point of having some bubbles in it, showing some boiling.

 

We made the plumbing to run the peroxide through the cooling jacket and back to the cat pack, and pressure tested it with water. The higher pressure coolant should suppress boiling a fair amount over the dump-to-ambient water cooling.

 

We set up with everyone behind the concrete wall, watching the engine on a closed circuit monitor, and very slowly worked our way up with monoprop runs going through the cooling jacket, first with 200 ml of peroxide, then 400, then 600, then 1000. Total thrust is somewhat lower, due to the pressure drop in the cooled chamber, but it does increase slightly as it pulls more heat out of the chamber. Based on the discoloration, the cat pack is clearly getting hot earlier in the cat pack with the preheated peroxide. We could probably get by with a smaller cat pack if we didn’t mind a cloudy run until it warmed up.

 

The next run, we briefly pulsed the kerosene on a couple times. Nothing exploded.

 

The next run, we held the kerosene on for a while. After about four seconds, the flame got rough, and we cut the kerosene. It continued for the rest of the run on monoprop without a problem.

 

We repeated this on the next run, and the behavior was the same. After four seconds of hot fire, things got rough. We re-pulsed the kerosene later in the run, but it was still rough. We assume that we are seeing some boiling in the cooling channels.

 

http://media.armadilloaerospace.com/misc/FirstRegen.mpg

media.armadilloaerospace.com/2002_04_06/regenerative.xls

 

We took the engine apart to look at the state of the catalyst pack. Some literature has claimed that 90% peroxide will melt silver packs if it has been preheated by being used as a coolant, and that 85% should be the limit for that application. There was some silver plating under the cat pack, and a few of the bottom silver wires had a semi-melted look, but none of the 32 mesh screen wires were melted through. This melting at the bottom was probably due to the rough running pushing biprop combustion back into the pack. We have seen this when using a pressure transducer in the chamber – rough running pushes way more heat in odd directions than smooth running does. It is too early to say conclusively, but it doesn’t look like the pure silver screens are going to have a real problem. Maybe the reported problem was dealing with plated screens that lost their adhesion at temperatures below the actual melting point.

 

Cooling Analysis

 

We aren’t absolutely positive that there are no stagnant spots in the cooling jacket which might boil easier, but we do know for sure that the coolant is getting plenty hot, so we should assume that it is a raw heat transfer problem.

 

Heat transfer is basically proportional to chamber pressure, times chamber temperature minus wall temperature, times chamber surface area.

 

The cooling problem appears to be independent of pressure in a given engine, because chamber pressure, and therefore heat transfer, is directly proportional to peroxide flow, which is proportional to heat absorption. That means that high pressure engines and throttled engines shouldn’t put more heat into the coolant, although higher pressure would demand more conductivity out of the wall material. I had been carrying around assumptions to the contrary for a while.

 

Flame Temperature

 

Flame temperature is determined by peroxide concentration, fuel ratio, and to a minor extent by peroxide inlet temperature. We are already running very rich on the kerosene, so we can’t reduce heating much more that way. We had been considering options with lower peroxide concentration, and we had identified some fuel mixtures (containing ether and some other low flash point additives) that would probably still auto ignite at the lower cat pack exit temperatures, but considering all the problems we have had lighting ethane even with 90%, I am getting dubious about autoignition at any lower temperatures. We might try diluting to 85% and seeing if the kerosene still lights. We might be able to ignite with the peroxide circulating through the cooling channel, even if it wouldn’t when water cooled, because there will be a little temperature boost.

 

Putting 5% water in the combustion chamber can drop the flame temperatures a fair amount.

 

Surface Area

 

The required chamber volume is the throat area times L*, the characteristic length, which is an empirically derived value for each basic type of engine. Our first biprop engine only had an L* of 17, and it clearly was not enough. The current engine has an L* of 30, which is at the low end of what is recommended. For a given shape combustion chamber, surface area will be proportional to the volume to 2/3 power. When you double the throat area (which doubles the thrust), the chamber volume must also double, but the surface area only increases by 2 to the 2/3 power, or 1.59. Because the total coolant flow is proportional to throat size, big engines will have a lower total amount of heat absorbed per unit of coolant.

 

Surface area to volume for a cylinder is lowest when the height is equal to the diameter, which would be about 1.6” for this engine to keep the same L*. That would only be an improvement of about 10% over what we have now. A true spherical chamber of 1.84” diameter would be an additional 13% less surface area for the same volume.

 

We could shrink our chamber volume a little, and make it the optimal dimensions to cut maybe 20% of the heat transfer. Or, we could make a much bigger engine and get really significant reductions in cooling load.

 

Fuel Cooling

 

We could attempt to use some of the fuel as an additional coolant. At best, this would be a 10% benefit, given the much larger flow of peroxide, and the worse thermal characteristics of kerosene. We could circulate the kerosene once around the chamber, but the conventional wisdom is that with normal kerosene (as opposed to RP-1), you would get coking. Almost certainly not worth it. Right now, the fuel injector is a separate ring between the cat pack and the cooled chamber. When we make a new chamber with the fuel injector integral, there will be better heat transfer to it, so there will probably be some small benefit as the kerosene impinges on the opposite wall and draws heat from it.

 

We certainly don’t have enough fuel flow at this engine size to think about film cooling, and even in much larger engines, it doesn’t make as much sense for peroxide.

 

Chamber Materials

 

A high thermal conductivity material, like copper or aluminum, will transfer heat a farther distance, so there is less chance of a localized hot spot causing a burn through. The maximum allowable distance from flame to coolant probably doesn’t change with scale, which is a benefit for small engines, allowing straight cooling cylinders, instead of having to closely follow the nozzle profile. The allowable distance would also decrease with increased chamber pressure.

 

However, a high thermal conductivity material will put much more chamber heat into the coolant.

 

Our current chamber is nickel plated brass, which is sort of middle of the road. Going to aluminum would make for a very lightweight engine, but we would transfer lots more heat out, which we clearly can’t afford at this size engine. Going to stainless steel might significantly reduce our heat load, but we might wind up melting the throat out of the engine unless we changed how we route the coolant.

 

By http://www.aksteel.com/markets/pdf/316_316L_Data_Bulletin.pdf , 316 SS has a thermal conductivity of 9.4 BTU/hr/ft^2/degF

360 brass has 67 BTU/hr/ft^2/ft/degF

Most aluminums have between 80 and 125 BTU/hr/ft^2/ft/degF.

 

Something like jet-hot coating on the inside of the chamber would prevent a large amount of the heat transfer, but the conventional wisdom is that ceramic based coatings don’t last inside combustion chambers. We could try it out on the current engine pretty easily.

 

Summary For Future Work

 

We could try burning 85% peroxide.

 

We could try getting the inside of the current chamber coated.

 

We could build a larger biprop engine to take advantage of the scaling law.

 

We could build a cooled chamber out of stainless steel.