Explosion on an ammonia reactor of a fertilizers plant.

An explosion, accompanied by abundant black smoke, occurred at an ammonia production plant. An ignited leak in the unit spread to the pressure line carrying synthesis gas (mixture of nitrogen N2, H2 and NH3 already synthesised).
Operators on site at that moment operated the emergency stop and evacuated the unit to the fall-back zone (reinforced control room), due the risk of a toxic leak.
The plant operator triggered the emergency plan around 20 minutes later. Local fire brigades took over from the in-house firefighters who were watering the accident unit with three water spears to prevent the spread of the fire.
The synthesis gas circuit was flashed with nitrogen to remove the hydrogen. All components in the ammonia peoduction unit were stopped, while a lower secondary explosion occurred. The operator closed the water networks of the site to avoid pollution of the nearby river by the extinction water. For lack of fuel, the fire went down gradually around one hour after its start.

The INITIATING CAUSE could have been one of two options:
(1) A high-pressure rupture of a valve carrying synthesis gas at 200 bar near a compressor of the unit, causing a burst perceived as an explosion,
or
(2) A gas leak from the same valve, followed by an unconfined vapour cloud explosion.

After loss of confinement, the ignited hydrogen spread around the affected pipe.
The ROOT CAUSE is unknown. One year before, the same unit had experienced the failure of a pipe carrying steam. The reason for that event had been identified in a steel not suitable for that operation conditions.

pipe, compressor

An accident due to pneumatic bursting on a pressure steam pipe had occurred on the same unit in June 2010 (ARIA 38831).

DESCIPTION OF THE SITE
The plant was specialised in the manufacture of fertilisers from four raw materials: natural gas, ammonia, sulfuric acid, and phosphoric acid.
The four main manufacturing units were:
− an ammonia manufacturing unit with a capacity of 1,200 t/day;
− two nitric acid manufacturing units with a capacity of 3,000 t/day;
− an ammonium nitrate manufacturing unit with a capacity of 2,100 t/day;
− a specialty fertilizer (NS/NP) manufacturing unit with a capacity of 2,000 t/day.

The hydrogen required for the production of ammonia was produced from the natural gas by Steam Methane Reforming.

The 4 injuries were minor. Following an emotional shock, 4 employees who worked on or near the area shortly before the explosion felt bad at the end of the intervention and were examined in the infirmary of the site.
The blast of the explosion damaged a cinderblock wall close to the compressor. The sheet metal roof of the room was reached by the thermal effects of the fire. The compressor was also damaged as a result of its abrupt shutdown and overheating of its oil-free axles. The supply of NH3 was vital for the synthesis of different nitrogen fertilizers, the site lost for several weeks 40% of its production capacity (several million Euros loss). As this plant supplied a large part of the French nitrate nitrogen fertilizer market, the price of these fertilizers increased significantly during this period.

There was no noticeable damage to the environment. Nevertheless, the characteristic ammonia (NH3) odours were perceptible off site. Toxicity measurements (NH3) were performed in the unit and around the site by fixed and mobile sensors. While downwind around the site the measurements did not reveal any dangerous concentration; 200 ppm was measured in the machine room of the accident unit, well below the toxicity threshold of 345 ppm for a 1 hour exposure). the NH3 concentrations were zero at the accident unit which was put in cold stop. Half hour later toxicity was back to zero.

A press release from the prefecture announcing the smells, the absence of toxic effects and the control of the accident was sent at 10:40, while an automatic telephone message informed the mayors of local communities. The emergency plan was raised at 10:55. The operator held a press conference with the prefecture for the benefit of the numerous local and national media who visited the site; information and photos of the accident taken by local residents were circulating on the Internet since 10 am. The management of the site apologized to the residents for the inconvenience caused by the accident. Firefighters left around 11:50.

The detailed accidental sequence:
8:50 Explosion, accompanied by abundant black smoke. The seven-men operation team on the site activates the emergency stop and evacuates the unit because of the risk of toxic leak, moving to the fall-back zone (reinforced control room). Some employees of neighbouring units ran to hide in their vehicles while others attend calmly to the subsequent fire.

9:15 The plant operator triggers the emergency plan around, alerts the public rescue service and the crisis unit of the prefecture, which was already operational, thanks to an ongoing emergency drill.
The Firefighters mobilise 70 men, 25 vehicles and a mobile chemical intervention cell (CMIC).). They take over from the in-house firefighters who water the accident unit with 3 spears to prevent the spread of the fire.
In the control room, the production team flushes the synthesis gas circuit with nitrogen in order to remove the H2 and stops the machines in the workshop while a lower secondary explosion is heard.
The plant operator closes the water networks of the site to avoid pollution of the river Seine by the extinction water.

9:55 the fire dies out due to lack of fuel

9:50 The gendarmerie and the inspection of IC arrived on the site.
Toxicity measurements are performed in the unit and around the site by fixed and mobile sensors. Measurements carried downwind (10 km/h) around the site did not reveal a dangerous concentration; however, 200 ppm was measured at 10:07 to the machine room of the accident unit (toxicity threshold of 345 ppm for a 1 hour exposure). Characteristic NH3 odours were perceptible off site.

10:40 A press release from the prefecture announcing the smells, the absence of toxic effects and the control of the accident is sent, while an automatic telephone message informed the mayors of local communities.

10:55 the emergency plan is called off.

11:25, the NH3 concentrations are zero at the level of the accident unit, which has been put in cold stop. The operator holds a press conference with the prefecture for the benefit of the numerous local and national media who visited the site; information and photos of the accident taken by local residents were circulating on the Internet since 10 am. The management of the site apologized to the residents for the inconvenience caused by the accident.

11:50.Firefighters leave the site

This specific accident did not generate lessons, at least they are not provided in the ARIA report, and in absence of detail repeated to the technical reason for the leak or rupture of the pipe, HIAD cannot perform its own analysis.

In 2010, one before this incident, another type of incident had brought to a long stop of the ammonia production unit, followed by a detailed analysis of the causes. It was not a hydrogen-related one; a high-pressure failure of a stem container was at that time the initiating cause. The investigation had found shortcomings in material safety document control and traceability. The company could not find the original pipe construction files. The documentation was starting after 25 years since the first commissioning.

Upon request of the authority, the operator performed an inventory of pipes likely to contain non-compliant steels, or at risks of hydrogen (syngas) cracking corrosion. An analysis identifies 65 critical pieces of equipment, which brought to almost 700 material analysed using a portable X-ray analyser (sections of pipes, welds and equipment such as bosses, tees, elbows, etc.). Critical location such bended and curves were also assessed by magnetic particle inspection methods.

This work suggests (but cannot completely exclude) that the failure of a high-pressure pipe on the hydrogen recycle line one year after this thorough analysis and assessment work was not related with material defect or ageing, but possibly to failure in some connection.

A more general lesson learnt is related to the sharing of description of a safety-related event in the public domain. Almost always an event is captured and filed by a repository or a database soon after its occurrence, when the inspection has not even started, or not yet provided conclusions on root causes and recommendations for corrective actions. The investigation works could take several years, and is very seldom re-emerging I the public domain, especially if a legal process is involved. This lowers dramatically the chance to achieve a useful return of experience able to inform the technical community.