I'm currently working on updating the controls for an oxygen-acetylene process for Union Carbide. It uses a gun that is fueled with oxygen and acetylene and that triggers a pulsing flame with a spark plug driven from a mechanical magneto that also controls a valve that admits powdered carbide. The pulses of carbide are liquified by the explosion and deposited in the molten state onto blades for jet engines in a continuous coating process. The process is very violent.
Oxy-acetylene scares the hell out of me, but I was actually surprised at how simply they deal with it. The gun is located in a sealed room, sort of like a walk-in cooler, that has ventillation at either end that constantly recirculates air in one side and out the other (like a paint booth) and back to the roof where it is just exhausted into the air. The fan motors are only about 10 hp.
The blades have to be set up in a fixture that moves them around while the gun coats them. The worker wears a respirator while in the booth, but only because of the carbide dust. Apparently the fuel gases are not a big concern. When I was troubleshooting the new program, I didn't even have to wear a repirator since we didn't run it with powder until the very end.
When the blade is ready, the operator strikes a mushroom button near the exit door as she leaves and a buzzer sounds. The operator then has fifteen seconds to get out of the room and seal the door. The system is then "armed". After the coating process is finished, the magneto is grounded out to stop the flame, and the fuel gases are shut off. But a purge gas (nitrogen) is then introduced which flushes out the gun barrel for fifteen seconds. After this she must wait another ten seconds while the ventillator fans run some more before she can open the door. Then everything is fine and she can walk right into the room.
The controls I was replacing were from 1968 (relay logic!), so they have been using this process successfully for a long time with no problems.
Hope this helps.
Don
Kansas City