Leak from a hydrogen compressor
The leak occurred on the compressor of a hydrogen testing laboratory. The location of the leak in the cylinder cap of one of the ultra-high-pressure hydrogen compressor unit. The hydrogen gas detector installed above the compressor detected a hydrogen leak of 2,000 ppm and triggered the automatically shutdown of the hydrogen compressor and the closing of the shutoff valve. After hydrogen concentration had fell below 2,000 ppm, an employee entered the compressor room and using a portable hydrogen gas detector identified the leak on the leak port in the cylinder cap. The compressor had been inspected approximately one month before. The investigation found that the reason of the leak was the failure of a check valve. The valve had been replaced during the inspection with one made of steel with slightly different mechanical characteristics, with lower mechanical performance in a hydrogen environment.

The INITATING CAUSE was the rupture of a check valve of the compressor. The check valve had been replaced after the overhaul inspection. The new one was made of the same material as the previous product, 416 SS (equivalent to JIS SUS416, a martensitic, free-cutting stainless steel with a relatively high sulphur content for improved machinability). However, the material had undergone a modified heat treatment process. The change in heat treatment process was made by the check valve producer without the equipment manufacturer's approval. The heat treatment had brought to these changes: (1) The hardness of the modified product was lower than that of the previous product, nevertheless it still met the equipment manufacturer's hardness requirements. (2) The tensile strength of the check valve material, estimated from the hardness requirement, was in the 1,200-1,300 MPa range. It is known that tensile strengths exceeding 1,200 MPa increase the hydrogen sensitivity of mechanical properties such as reduction in area. The tensile strength of the modified product, estimated from hardness, is approximately 200 MPa lower than that of the previous product. (3) The check valves were made of 416 SS, which has high hydrogen sensitivity. A change in the heat treatment process resulted in finer inclusions (MnS) segregating at the grain boundaries compared to conventional products, reducing toughness and presumably leading to grain boundary fracture due to high-pressure hydrogen gas. (See Figures 3-5. ) Based on these finding, the ROOT CAUSE can be identified in a design failure. The original material chosen for the valve materials was already a martensitic steel known for its hydrogen-sensibility, the material of he replaced valve worsened the risk of ruptures. Moreover, the lack of communication from the part producer and the equipment manufacturer did not allow a possible independent check. This regards a lack of proper management of the supply chain. Finally, a lack of effective communication management caused an additional failure. Since the failure of the hydrogen booster occurred on October 23 was still under investigation, the information regarding a wrong material choice was not shared within the site. The day after, another hydrogen gas leak occurred due to the same check valve failure in another hydrogen booster.