Explosion in an ammonia production plant

The leak point was identified by the trace left on the gasket; looking from the channel side, the leak was located at about the 5 o’clock direction when the top of the flange is taken as 12 o’clock.

The direct cause of the incident was the failure to adequately isolate hydrogen from the section of the system which was undergoing repairing works.
Despite a successful purging and the negative results of hydrogen detection, the detection procedure was not designed in such a way to reveal hazardous concentration of hydrogen able to develop an explosive mixture with air during opening to air.

The root cause was the inadequate management system, which failed to identify the hazard and to take the necessary preventive measures.

hydrogen pipe

The plant had been partially shut down for repairing. Radiographic examination indicated that internal cracking of the pipe was more extensive than the outward
manifestation of it, and required the partial shutdown of the plant to allow for grinding and welding works.
The preparatory actions were not able to achieve the dry conditions necessary for the grinding and welding work. Additional measures were adopted, affecting also the sections still containing gases.

The ammonia production plant operated on natural gas feedstock.
As explained by Nightingale (see References), the section of plant affected by the incident was transferring gas from the low temperature shift converter (LTS) to the carbon dioxide removal section. Heat exchangers in the section provided boiler feedwater pre-heat heat for Benfield solution regeneration. In doing so, the gas was cooled to below its dew point, and condensate was collected and removed in a catch pot. The crack to be repaired was at a point very near the Benfield regeneration.

A welder was slightly injured.
No damage to the plant.

According to Nightingale (1989), the very complex mechanisms that led to this incident had not been foreseen. Even after its occurrence, the explanation was only found after two full days of intensive investigation.
Among other findings related to the failure to ensure hydrogen tightness, the investigation explained why the gas measurements performed at the end of the purging could not detect hydrogen: the nitrogen flow was enough to reduce hydrogen concentration below the detection threshold of the instrument.

As corrective measure, the instructions for the isolations of the process during maintenance were improved, adopting positive (slip plates) isolations for hazardous materials.