Explosion on a hydrogen compressor a space centre

A hydrogen-air explosion occurred outside a hydrogen compressor. Gaseous hydrogen was released from the vent stack due to a premature activation of a relief valve. The flammable atmosphere ignited and exploded, followed by a fire causing considerable damage. Following explosion, the shut-off valves were closed and system vented.

The compressor unit included two relief valves downstream of a pressure regulator which had the role to reduce pressure from 5000 to 3000 psig (from 345 to 207 barg). The relief valve were downstream of regulator, in the 3000 psig system.

The valves were sized to handle substantially different flows and one was designed for another program.
The relief valve seemed to have opened prematurely when the system pressure was increasing from 2700 to 2900 psig (from 186 to 200 bar). The lack of accuracy of the 5000 psig gauge used to control the regulator control, combined with the tolerance of the relief valve settings (±3 percent) was sufficient to open the relief valve.
Large flows were vented due to the large relief valve. In addition, the vent stack cap was designed so that the vented gases were released in a horizontal rather than vertical direction.

The INITATING CAUSE was the premature activation of a relief valve.
The ROOT CAUSES were a combination of the use of inappropriate safety equipment and bad safe design (venting pipe horizontally places, no isolation valve between storage and compressor). It seems that part of the equipment was in place for another project, what indicate also bad (project) management.

PRV, gauge, vent, Compressor

Some of the valves had been designed for another project.

Only a qualitative statement of "considerable damage"

The original emergency strategy of the company was to let the tank bleed. Eventually, the company called the local fire department. The onsite building, a neighbouring business and a residence were evacuated, and police blocked nearby streets. Major concern was the possibility to produce ignition source able to ignite the leaking hydrogen.

The incident initiated with the premature opening of a relief valve at 2900 psig, while the activation pressure was probably 3000 psig. This was (partially) attributed to a lack of accuracy of the gauge measuring pressures.
The premature opening of a pressure relief device is not an uncommon event. However, the consequences of this event could have been reduced with a better unit design:
(1) The horizontal venting of the vent duck created a horizontal flame probably causing damage to the neighbouring equipment (see event HIAD_310 for further cosndieration on good venting design)+E302.
(2) The amount of hydrogen contained in a compressor is usually limited. Moreover, if it had been purged at shutdown, nothing would have happened by re-starting. But even if was left under pressure, the considerable damage caused by the venting was very probably caused by the release of the hydrogen contained in other equipment connected to the compressor. As H2TOOLS properly remarks, a backflow prevention device (for example a non-return valve) could have prevented this release.
Moreover, the NASA report mentions the fact that the valves were sized to handle substantially different flows and were designed for another program. This calls for the responsibility of the management to ensure that the correct equipment is used, properly maintained and calibrated, and that an overall risk assessment and its verification are performed before starting operation.