Leaks from a hydrogen compressor

This incident event consists in two leaks occurred in the same week on a hydrogen compressor of a hydrogen refuelling station.
After filling a fuel cell vehicle (FCV) with hydrogen at a compressed hydrogen filling station, the pressure in the hydrogen storage dropped, triggering the automatic intervention of the compressor to re-establish the nominal pressure.
This resulted in hydrogen leaks from the compressor's fifth-stage cylinder head on the 24 of October and from the third-stage cylinder head on the 31 of October.

The cause of the leaks may have been the twisting of the O-ring due to pressure pulsation (24/10) or damage of the O-ring due its use beyond its service life (31/10).

The INITIATING CAUSE was in both cases the loos of confinement caused by the degradation of O-rings located in the head of the compressor heads.

The compressor manufacturer recommended that the compressor be disassembled, inspected, and opened for maintenance after three years or after 1,000 operation hours after the completion of the test run.

The first O-ring which leaked had not yet reached the end of the prescribed lifetime. It was found twisted, but the twist disappeared with time. The investigator concluded that the pressure cycles occurring in the compressor head (from 45 to 82 MPa) were responsible for the twist.

The second O-ring which leaked had scratches on the surface, no cracks or twists. This O-ring had arrived almost at the end of its lifetime and had never been replaced since installation. It is probable that the scratches were the consequences of a too long service life.
The ROOT CAUSE was in first instance the use of not-optimal materials for the harsh condition experienced in the compressor heads, worsened by the fact that the inspection and maintenance should have been able to detect a deviation from the original state of the rings.

compressor, O-ring

DESCRIPTION OF THE FACILITY
The HRS had a Production Capacity of 7,896 m3/day of hydrogen.
The stages which leaked had a nominal pressure of 24.1 MPa (3rd Stage) and 82 MPa (5th Stage)
Operating Temperature: 3rd Stage = 180°C / 5th stage 160°C
The O-rings were made of the types G60 and P36 and mande of Fluororubber

Considering the extermely low hydrogen contentration measured, this event borders to a near miss.

These two incidents are almost a near-miss: the detection alarm was set at a very low concentration of hydrogen in air (in the range of 120 to 160 ppm). Since operations were stopped immediately after, it can be assumed that hydrogen amount released was very low and by far below the threshold necessary to form an explosive mixture in the compressor room. HIAD classifies this type of leaks as incident, because (I) every emission of hydrogen is unwanted event, and (ii) it implies stop of operation, with consequent financial losses, including the costs of investigations.

Moreover, the events provided important considerations:
(1) The General High-Pressure Gas Safety Regulations require alarms set at 10,000 ppm or less. Alarms for gas leak detection set at less than 10,000 ppm of hydrogen in air meet this legal requirement. The KHK notes in its report (see References) that alarm settings for equipment installed at compressed hydrogen fuelling stations vary among operators. Since a technical basis to identify a leak event as an incident is missing, and if an event is classified as (high-pressure) incident or not depends only on administrative consideration. It can be added, that choice of a effective alarm able to prevent escalation depends on many factors, not only intrinsic to the equipment, such as the type and number of detectors, their location, the place where the equipment are installed, the ventilation available, other equipment nearby, etc. This explain why it is challenging to agree on a general alarm setting.
(2) The sealing materials used in compressed hydrogen fuelling stations were still under development in the period in which these leaks occurred, and are still it now, some years later. Manufacturers must be aware of this when designing and manufacturing, and users must be aware of this when operating and managing the equipment. Therefore, efforts must be made to constantly obtain and utilize the latest information. Information on accidents and malfunctions should be shared widely with those involved in compressed hydrogen fuelling stations, not just inside the affected plant operator, and should be used to contribute to accident prevention.
(3) Incidents and malfunctions that occur at compressed hydrogen stations often involve causes that cannot be easily solved technically. As a result, it may not be possible to immediately implement permanent measures, and it may take time to verify whether the permanent measures being considered are effective. Therefore, until permanent measures can be implemented, it is important that manufacturers and users work together to implement on-site temporary measures such as strengthening monitoring systems.

The following emergency measures were taken:
• To replace the O-rings with a new one.
• To reduce the time planned for the replacement of the new O-rings (200 h) in respect to their nominal lifetime.

On the long term, these additional permanent measures were considered:
1. To change the O-ring material and dimensions. However, several development and tests did not show the expected improvement in performance.
2. Change the Cylinder Head Structure, mainly by using two O-rings in place of one, with two different roles. The tests with this new design showed the expected improvements and were implemented.
3. The compressor's overhaul inspection and maintenance cycle (including O-ring replacement) were made more frequent.