Explosion in a foundry

Coke oven gas was released from a pipe during maintenance works. The pipeline was supplying burners of the coke oven with coke oven gas (65% hydrogen, 25% CH4, 3-6% CO). Following the release, a delayed explosion occurred.

The main reason for the maintenance procedure called “gas-stop” was the need to execute some works on the water piping network. This procedure is carried out very seldom because it is difficult and because it is very difficult to stop the oven, which was made of refractory bricks which could be irreversible destructed if their temperature would go below 600°C.
The stop was planned for 10 h. Therefore, employees decided to do some additional works on the gas pipe network and other cleaning works. The gasometer used to store coke oven gas had been repaired the previous week and safety checks were due. The gas release occurred when subcontractors were replacing a measuring diaphragm by a blind flange on the gas network, to be able to isolate a pipe section.

The leak was estimated to release 1 Nm3/s and lasted for 5 to 10 minutes before the explosion occurred. Therefore 300 to 600 Nm3 (125 to 250 kg) of coke gas was released.

The explosion caused the death of 3 subcontractors, 39 injured, among whose 13 seriously burned and material damage.

[Note of HIAD event validator: the composition of the gas involved in this event is debatable. eMARS report an accidental release of coke gas. The English version of ARIA contains the same information as eMARS. The French version of ARIA, however, provides more technical information and reports that during the maintenance stop, the coke gas must be replaced by a mixture of natural gas and air, to feed the burners and keep oven at temperature. For doing this, it is necessary to isolate a section of the coke gas supply network. According to this version, it remains unclear if the released gas was coke gas or its replacement, the ternary gas. HIAD version assumes eMARS report, but caution is due when drawing hydrogen-specific conclusion on this event.]

The INITIATING cause was a release from a gas line when opening a flange. The pipe section to which the valve was belonging had not been properly purged (or not purged at all).

The ROOT CAUSE was the absence of a safety culture. There were no written procedures to inert the gas line, and line purging / isolation was a difficult exercise. On top of that, there was no coordination between the teams working at different duties on the same unit. Probably a contributing factor was the time pressure, due to the short time slot available for all the maintenance works.

pipeline, coke oven, gasometer, metals processing

The planned works for the day were: replace some valves on the water and gas piping network, performing safety checks on the control system of the gasometer , cleaning a transport belt, cleaning some other equipment and also replace a measuring diaphragm by a blind flange to be able to isolate a piece of the gas pipe.
A subcontractor specialised on gas piping works was chosen to do this.

3 fatalities and 26 injured.
The total release was estimated at 125 to 250 kg of gas,

The operator was found liable in civil court as the employer. Of the seven defendants, the coking plant manager was acquitted, and two engineers and two foremen received suspended sentences for three years. Two workers were sentenced to three months in prison and fined €250.

The purging procedure gone wrong was foreseeing:
• The closure of an upstream hydraulic valve.
• The closure of a downstream manual handwheel valve.
• the installation of a N2 injection point immediately after the hydraulic valve (internal network at 6 bar) and an N2 outlet before the manual handwheel valve.
• Checking for the absence of gas at the N2 outlet.
After that, the solid gasket could be installed by unbolting the flange. The explosion occurred during this operation.
The investigation determined that the hydraulic valve was not leak-proof at pressures exceeding 1 bar. It could not withstand the pressure of the ternary gas at 2 bar. No nitrogen outlet was installed on the line to measure the effectiveness of the purge. Witness statements indicate that the gas measurement check was performed at the nitrogen injection inlet, not the outlet, using a detector functioning only in presence of oxygen. Furthermore, the purging time was too short. Finally, workers individuals were not wearing fire-resistant clothing.

The incident highlighted the importance of:
- a safety culture permeating the whole organisation, from management down to the relationship with contractors.
- Written procedures to ensure effective purge of gas lines.
- Working procedures agreed with contractors, which can be followed and supervised in the field.
- A maintenance plan for valves.
- Hazardous area classification, to exclude ignition sources at places with risks for flammable gas leaks.