Explosions at a vent stack
A common vent system was used to connect vent lines from liquid hydrogen storage vessels to the vent stack (which was lit flaring the gas). Low order detonations were observed around the vent stack which continued when the vent valves were closed. Oxygen concentrations measured in the vent system peaked at about 6% at which time an explosion would occur. The oxygen concentration would then drop to zero. These events continued until the hand valves in series with the rupture discs were closed. On subsequent inspection it was found that the rupture discs had blown in backwards probably by the initial explosion. [Ordin, NASA (1974)]

A flange covering a part of the vent system was removed which resulted in air into the stack. A chimney effect was in existence when the vent was filled with low density hydrogen. Air mixed with the hydrogen and was ignited by the flare. Although the vent valves were closed, hydrogen continued to be supplied to the vent system due to the failure of the rupture discs in the first explosion. Why the flange was removed is not told, but it suggests a ROOT CAUSE related to lack of management of change, and/or possible installation mishap.