I've been doing research to understand the potential risk of flammable refrigerants for our heat pumps, such as R290 we've discussed here. The automotive industry is phasing in R1234yf which is mildly flammable. Daimler originally said they wouldn't use it in their cars because they could repeat fires and a corrosive acid cloud during their testing. Other manufacturers were satisfied with their tests, the SAE and others decided the risk was so small even airbags posed much more risk, and Daimler has since backed down.
So just how much extra risk does a flammable refrigerant pose? A small leak will never amount to a fire risk. A catastrophic leak even with a nonflammable refrigerant still has a fire risk because of the aerosolized oil cloud that escapes with the refrigerant. I found a few documented cases of oil fires happening this way, including one in the report below. Anyone who has ever loosened (accidentally) a gauge set hose that had not been evacuated knows the cloud I mean.
I've personally witnessed an AC system fire with a flammable refrigerant, but it was uncertain whether the refrigerant was to blame. The disintegrating compressor caused a hot spot and the system was full of air instead of all refrigerant.
THIS report was done after Australia allowed flammable refrigerants in vehicles and many in the USA put flammable refrigerants such as OZ12 in their R12 vehicle systems.
They found that the average automotive charge of R290 (most have 1 to 2 tons cooling capacity) was about 200g, less than half what you are carrying around in that green cylinder for your portable grill. They compared the number of predicted refrigerant fires (49 per year in Australia alone) to the number of actual fires: zero in both countries through multiple years.
They found that small leaks were easier to detect and drivers would address them because flammable refrigerants have odorants. They found that even if 200g is released in a catastrophic 1 second release, the gas is hard to ignite due to its velocity and concentration (ever try to light a torch while the valve is wide open?) and it is diluted below the level of flammability within 10 seconds. Even though mobile systems, unlike stationary systems, can be catastrophically damaged in a crash actual risks are clearly much lower than expected and far lower than many other things we should worry about.
Nations are now approving flammable refrigerants in larger quantities, allowing "those who must comply with the IBC" more equipment choices. Gasoline is more dangerous than R290 as I detailed
IN THIS POST. The 315 cubic feet I described is the equivalent volume of a cramped 5x8' equipment room, not a large room or basement.
We've lived with gasoline's extensive use and spillage for a century and think nothing of putting a car, or a lawnmower, or a gas can in the garage. Me? I'll take the most effective refrigerant. The risks of flammable refrigerants are small and overblown.