You wrote elsewhere that when welding, the resulting gas has a composition within the explosive limits. Did all accidents happen when welding, or also when cutting?
Maybe the gas composition changes from one sheet side to the other, when you cut or weld steel. More oxygen at the torch side, but behind the steel sheet, iron has used up the oxygen, and molten iron droplets make hydrogen by water contact. And maybe cutting slowly leaves an excess of oxygen, while cutting as fast as possible leaves little oxygen. Can this be tested on the ground?
Yes, rotting underwater produces methane, more so in quiet waters like dam lakes, normally not in open shallow seas where waves dissolve oxygen. In a closed volume like a boat hull, methane can accumulate, and oxygen brought by the torch lets it explode. If detailed accident reports exist, they might tell if the explosion was near a closed volume.
Hey the chemists, can rusting produce hydrogen? Fe to Fe3O4 and hydroxides is said to happen only if H2O dissolves some O2, but does the reaction decompose some H2O too? Or maybe a subsequent reaction, say from Fe3O4 to Fe2O3? Problem is, this hydrogen production would be very slow, and hydrogen leaks away so easily.