A hydrogen leak at a 70 MPa dispenser

This event occurred at the hydrogen 70 MPa dispenser of a refuelling station during the testing phase.
Just after the start of the refuelling test, at a pressure of approximately 13 MPa, the hydrogen gas leak detection alarm inside the main body of the dispenser went off at 4% of the LEL, triggering the to shut down of the whole the hydrogen refuelling station. Using a portable hydrogen gas detector to identify the leak location, workers discovered that the leak had originated from a joint inside the dispenser. There were no injuries or property damage.

The leak occurred during the first filling operation after a safety inspection of the hydrogen station. The leak location was at a threaded joint on the direct filling compressor side, immediately upstream of a hand valve.
This is the chronological summary of the actions performed before the leak occurred:
(1) 13 days before the leak, during a regular inspection, an engineering company subcontractor performed an airtight test on the dispenser piping, but only from the downstream side of the direct-filling compressor to the hand valve. The test was done with hydrogen at 82 MPa for 15 minutes. No leak was found.
(2) The day after, the dispenser manufacturer retightened some of the fittings inside the dispenser in, but again only downstream of the hand valve for maintenance inside the dispenser. This retightening work that was not specified in the work instructions. It is believed that the fitting in question loosened, when another fitting was being retightened. The fittings in question were not retightened.
(3) After the retightening, the company conducted an airtightness test, but due to a misunderstanding of the scope of the airtightness test, the fittings in question and the retightened fittings were not included.
(4) 7 days later, an airtightness test was conducted during a safety inspection. Due to an operation error of the engineering company, the hand valve remained close and the hydrogen was not supplied from the pressure vessel to the joint in question.

The INITIATING CAUSE was a loose threaded joint inside the dispenser

The ROOT CAUSE was a series of mistake which did not allowed to detect the leak during maintenance inspections and tightness tests. This can be attributed to a lack of clarity in the testing procedure and shortcoming in the checklist of the operation to be performed.

threated joint

The leak occurred during the first refuelling test following a safety isnpection (joints tightening and tightness tests).

DESCRIPTION OF THE FACILITY
The HRS had two dispensers: one at 35 MPa and at 70 MPa. The latter was equipped with a hydrogen pre-cooling device. The HRS had capacity of type of 53,856 m³/day. The KHK reprot does not specify the type of the (onsite?) hydrogen source. A compressor was delivering hydrogen at 40 MPa to a storage system serving the 35 MPa dispenser. A second compressor was delivering hydrogen to a second 82 MPa storage system, serving the 70 MPa dispenser. A third compressor was installed, able to provide hydrogen directly to the 70 MPa dispenser from the 42 MPa storage system. It is this last one which was in use during the leak.

The in-dispenser hydrogen detection system was of diffusion type, activating at 1/4 of 4% LEL.

This is case at the border between a minor incident and a near miss. The safety systems in place worked as designed with no consequence to human or installation, except the time lost during the shut-down and the investigation

(modified from the original in the KHK report)
This is case at the border between a minor incident and a near miss. The safety systems in place worked as designed with no consequence to human or installation, except the time lost during the shut-down and the investigation. Nevertheless, the case provides a list of lessons which can be applied to a much broader technical area:

(1) Threaded joints maintain airtightness at contact surfaces which are not the threads their self. If both ends of a connecting pipe are threaded joints, tightening one side may cause the separation the contact surface of the joint on the other side, with consequent leakage. To prevent this, proper tightening management of the threaded joints on both sides is required, and the gas tightness must be tested and confirmed by a test.
(2) At hydrogen filling stations two types of joints are present: screw-type and flange-type joints. From both types of joints, leaks may occur. It must be noted also that there are more screw-type joints inside the body of hydrogen filling station dispensers compared to CNG and LPG filling stations. O ensure that all of them are maintained and tested, work procedures and checklists must clearly specify the scope of the work, and identify all section by their number of valves, etc.
(3) When multiple operators (e.g. contractors) are working on the same high-pressure gas facility, it is necessary to clearly define the scope of work in advance using a flow chart, or similar and to share it among the operators. Moreover, a plant manager should agree on who-does-what and be present to supervise the works.
(4) Some additional diagnostic tools may assist in preventing misunderstanding and mistakes. In this specific case, because a connection valve had not been open, not hydrogen under pressure was present in the section under testing. A pressure gauge was installed in that section would have allowed to detect a low-pressure state. However, the addition of more connections my also increase the probability of leakage, so that a careful overall risk assessment has to be performed when introducing changes.

In the work specifications, it was decided that care should be taken not to touch any piping or equipment outside the scope of work, and that this should be clearly indicated before the inspection.
(1) In the work specifications, the work scope was clarified by specifying the sections using the tag numbers of valves, etc. In addition, it was decided that the airtightness test at the boundary of the work scope would be carried out in the presence of both the upstream and downstream companies.
(2) A procedure manual for airtight testing and a valve operation checklist were created.