Leak from a hydrogen compressor

A leak occurred on hydrogen compressor in a plant of production of renewable hydrogen.
A pipe on the condensate purge circuit of the compressor ruptured, releasing hydrogen and producing a noise peak detected by the area's ultrasonic leak detector located in the building. 3 minutes later, the fixed hydrogen detectors in the area measured a hydrogen concentration in the air above 10% of the LEL, and the plant automatically went into safety mode.
The production unit was shut down, the compressor was isolated, the facilities were depressurised, and forced ventilation remained activated. Approximately half hour later, the detectors were no longer detecting presence of hydrogen in the production area.
The next day, the operator replaced the leaking the ruptured pipe and its connection on the compressor's condensate drain circuit and inspected similar equipment, such as other condensate drainpipes on the other compression stages, before restarting operations.

The INITIATING CAUSE was the rupture of a pipe on the compressor.

When the drain valves on the condensate circuit of one of the compressor stages were opened as scheduled, hydrogen circulated through the leaking pipe, leading to its release into the premises. The gas tank supplying the compressor (probably a buffer tank between the production unit and the compressor) gradually emptied into the premises, while the other production equipment was isolated when the site was put into safety mode.

Since the safety system worked as designed, and nothing is known on the reason for the rupture, its inspection and maintenance schedules, etc. the ROT CAUSE can be attributed to a generic material failure.

pipe, compressor

DESCRIPTION OF THE FACILITY
The productioncalacity of the plant was 300 kg of hydrogen per day.
The nominal pressure of compressor for final storage and delivery was 350 bar.

No injury, no environmental impact.
Property damages liited to the loss of hydrogen (unknown quantity) andthe repalcement of a pipe.

In this incident, the safety system in place worked as designed: the hydrogen detection system of the facility detected the hydrogen concentration increase and triggered it shutdown.

However, it took 3 minutes for the detector in the room to detect hydrogen above 10% LEL. This could be an indication that the approach to the choice of locations of the hydrogen safety detectors in the buildings could be reviewed an improved. The detectors should be deployed to trigger a high-concentration level timely enough to intervene and avoid formation of large quantity of explosive atmosphere inside a building.

There are also uncertainties in the description of the accidental sequence, which make difficult the full understanding of the causes and the deduction of a useful lesson learnt. ARIA report does not clarify well the reason of the first event, i.e. the rupture of the condensate purge system. Moreover, it is unknown if the compressor was operating during the ongoing purging operation. Finally, the system description (connections, valves) and its quantitative data (stored hydrogen quantity and pressure, leaked amount) are not provided.
The most probable sequence is:
(1) Hydrogen leaking into the condensate purge pipe.
(2) Explosion in the condensate pipe, leading to its rupture.
(3) Hydrogen release into the premises.

Following the incident, the operator:
(1) checked the tightness and sealing of the condensate drain circuit on all compression stages;
(2) assessed the possibility of isolating the tank supplying the compressor and the compressor itself when the site was put into safety mode.