I took out the setup for some initial testing. This is using the manual controls, not the controls which will implement precise sequencing. First steps were to connect it all, and open the pressure on the LP tank(but not the inlet to the accumulator), and use the leak detect fluid to check for leaks on the joints I couldn't test before. No leaks found. (Overall, the 1/4" connections didn't seem to have any issues, it was the 3/4" ones that were harder)
Testing was done in a wide open area with a slight breeze.
At this point, the accumulator has atmospheric air because I used a standard air compressor for leak testing. This is problematic because adding propane will result in a mixture of propane and oxygen. This means that the propane could ignite inside the accumulator and explode. Normally this isn't possible because with just propane and no oxygen, ignition cannot happen. One of the reasons the air fitting was added was to purge the tank with an inert gas, such as CO2, Nitrogen or welding shielding gas. Though commonly available at welding shops, brew-n-grow places and beverage supply houses, I don't yet have a tank of this, mainly because I need to purchase a tank and regulator.
Instead, given the location, I purged with the propane itself. I did this by filling the accumulator, then venting it with the ignitor off so ignition wouldn't happen. I did this a couple times, waiting a bit between each for the propane to dissipate. In the future I will need to get a tank of gas to purge, because I'll want to purge the propane out of the accumulator for long term storage and transport.
I turned on the HSI and gave it a couple minutes to heat, and let a test fire go, nothing. But second try was a success! (I guess the HSI wasn't hot enough the first time)
This picture was taken in daytime, and is a 1/160sec exposure just manually pressing the shutter release a moment after hearing a helper pushed the button. That is, this is not a long duration exposure. The top of the flame is 15-20ft high.
I spent a while experimenting with different timing, different pressure, and sequencing the two barrels. The second barrel prelaunched without ignition definitely helps create more fireball at lower pressure, and a better effect. I never got above 35psi as the LP tank was down to that by the time I got up to that pressure; it was mostly empty at the start anyway.
It is a very impressive display, and definitely puts out quite a bit of heat from 20ft away. Running the calculations, given the Cv value of the valve used(7.6), the specific gravity of propane(1.52), and the average energy density of propane(2516 BTU/ft^3 as a gas), a 1 second burst is about 6500 BTU. This is obviously a burst type device, but that would equate to 23,000,000 BTU per hour. (A 2500ft^2 home might have a 100,000 BTU furnace) Flow through the regulator is estimated at a bit less than 1/10 of that, so it would take 10-15 seconds to repressurize the accumulator after a 1 second burst. During testing, it didn't seem quite that long to repressurize, but it was at least reasonably close.
An LP tank usually cannot sustain that sort of flow because the propane will freeze in the tank. After about 15 min of testing (maybe 80 bursts), it did start to freeze as rocking the tank revealed the tell tail signs of slush inside. The outside of the tank was forming ice. This was probably exacerbated because the tank was very low. Some other large projects using this type of system will actually heat the tank with a hot water bath to prevent this, though in my use case it shouldn't be a problem. It may just limit how much testing I can do in a given session.
I will continue working on the sequencing controller, and get the LP tank refilled for testing next time to determine the desired timing, and trying out higher pressure.