Release of a pressure release valve

A hydrogen safety sensor went into alarm at a hydrogen demonstration project facility. The early morning temperature was near freezing and there was a trace of precipitation on exposed surfaces.
Upon further inspection following the sensor alarm, a 3,500 psig (approximately 24 MPa) stationary storage vent was found to be releasing hydrogen at the vent stack through the Pressure Relief Device.
The alarming sensor was at an adjacent building but responded to hydrogen 20 yards (18 m) downwind from the vent location. The vent release location was 10 feet (3 m) above ground level. No ignition was detected.
[Source: Burgess et al., Technical Report NREL/TP-5400-67381]

Review of pressure data did not show a pressure above the PRD set point. Freezing temperatures and moisture are suspected as contributing factors to spurious opening of the valve. Once the valve popped open, the cold temperatures of the escaping gas could have caused the valve to freeze in the open position.
Freezing and ice forming are regularly reported as causes for the malfunctioning of safety devices. Inspection and maintenance should be designed according to the related risks.

PRD, vent stack, CGH2 storage

The early morning temperature was near freezing and there was a trace of precipitation on exposed surfaces.

The investigation included removal of the valve from the system and shipment to the valve manufacturer for inspection. A failure investigation was conducted to determine the root cause. This investigation involved a review of operations data, inspection of PRDs and system components, and shipment of the valve back to the manufacturer for further evaluation. The post-test manufacturer’s evaluation included testing the operation of the valve to the set pressure. The valve set point was shown to be within the ±3% ASME tolerance. Manufacturer valve testing under indoor environmental conditions showed that the valve was able to show leak tight operation and opened within the ±3% ASME tolerance [ASME BPVC Section VIII Division 1 Paragraph UG-126 (d)]. Note that the valve was inspected under laboratory conditions and no attempt was made to recreate the cold outdoor ambient conditions that the valve experienced during failure.
Disassembly inspection showed that all internal parts were within the manufacturer’s specification. Discussion between operator and the manufacturer led to the conclusion that contamination (or possibly freeze/thaw moisture) was a possible cause for PRD failure. In this scenario, the valve is able to open even though the pressure is below the set point. The flowing gas provides additional opening force that can cause the valve to actuate to a full open position. The expanding gas creates a low temperature that causes the valve to stick in place either through thermal contraction or condensation freezing. The relief valve is frozen open and thus not able to reseat as the pressure drops in the storage vessel. The high flow rate achieved during the pressure venting is able to remove any evidence of the contamination that caused the initial valve opening.

Following the investigation, the following were proposed:
(1) Consider the removal of PRDs from stationary storage where appropriate. the quantitative risk associated with a PRD failure has been compared against the risk of not having a PRD in the case of stationary hydrogen storage. It has been argued that the overall risk is lower with no PRD on the stationary storage.
(2) PRD predictive maintenance program. As part of the corrective action surrounding this PRD failure, the R&D facility operator initiated a pilot program to test and recertify compressed gas PRDs across the lab.