Fire at a flare stack of a chemical plant

A fire developed at the top of the vent stack of a during a violent thunderstorm, in a chemical plant (Seveso-classified).
The vent stack was used to discharge hydrogen produced by the restarting of chlorine electrolysis cells. The fire was ignited despite steam and nitrogen were specifically injected to prevent it from occurring.
The site’s firefighters cooled the stack while the technicians turned off the electrolysis cells to cut off the hydrogen feeding the flames.
The electrolysis cells had been restarted following a power outage that had occurred earlier, at the start of the storm. The fire was extinguished after approximately 3 hours, and the electrolysis unit was restarted.

INITIATING CAUSE was the lightning bolt which released enough energy to ignite the hydrogen reach mixture out of the flare stack.

flare stack, electrolyser cells

A bnormal weather conditions: a thunderstorm had caused a power outage and the stopping of the electrolysis cells one hour before the incident.

1:20 Power outage and restart of the electrolyser
2:45 Lightning bolt ignited the hydrogen at the top of he flare stack
3:10 The site’s firefighters started cooling the flare stack.
4:57 The fire was extinguished
5:10 the Electrolyser was restarted

The same event occurred already five years earlier. At that time, corrective measures had been adopted, among which the dilution of the hydrogen flow, to reduce the probability of ignition.
However, this measure did not altogether eliminate the risk of ignition, because the hydrogen remained above the minimum ignition concentration, as demonstrated by the fact that a lightning was sufficient to start a fire.

The report does not mention the lesson learnt in this second event.
One obvious follow-up action would be the extension of the measures against consequences of lightning to the whole year.
Another action would aim at improving them. The assessment of the hydrogen concentrations in and at the exit of the flare stack and consequently performing a risk assessment to ascertain the need for modification of the venting conditions, including the steam/nitrogen dilution process.
(See also HIAD_797, a 3rd incident caused by lightning).

The corrective actions taken five years earlier were:
1) Installation of a velocity-seal on the stack to avoid the possibility of a backfire in case of de-pressurisation of the stack;
2) Reduction of the diameter at the exit of the flare stack to increase flow velocity;
3) Improvement on the second flare stack to limit possible heat flow on the neighbouring cooling system of the chlorine.
Finally, steam and nitrogen are injected to dilute the hydrogen stream and raise the minimum ignition energy of the hydrogen-oxygen mixture.