L2 leak during tanking

The event occurred at a stationary storage of liquid hydrogen. A vent line leaked hydrogen during filling. The leak location a crack in a weld of the line. The leak was attributed to a cracked weld on a hydrogen vent line that consisted of a double wall aluminium piping with slotted spacers between the inner and outer line to provide a hydrogen gas blanket for insulation.

The weld that failed was repaired using a "clamshell" over the area of the failed weld to continue operations. It was later replaced with a new half shell piping section.

After operations, the clamshell repair was excised from the double wall piping for further analysis of the failed weld.

The INITIATING CAUSE was the cracking of a weld, which released liquid or cryogenic hydrogen.
The origin of the failure was identified by the post-incident investigation in an inadequate welding process, characterised by the choice of an inappropriate of the fillermaterial and by a weld execution which induced critical defects which lowered the mechanical performance of the component at the weld location.
The crack occurred at a stress transient induced by thermal gradients from a difference of almost 100°C between the top and the bottom of piping during hydrogen transfer.

vent line,
cryogenic transfer line,
tLH2 tank

ALUMINIUM ALLOYS INVOLVED
In this plant, vent lines were made of an aluminium alloy belonging to the 5000 series,
alloyed with magnesium, with high corrosion resistance and strength.
Al 5083 composition: Mg 4.4; Mn 0.7; Cr 0.15

Alloy 4043, containing silicium (5.2%) is often used for filling, but it is not recommended as filler with 5083 as base material, because:
• The formation of magnesium silicide which induce brittleness.
• Cracking upon solidification, due to the dilution of the base metal into the weld pool.
• Reduction of the high corrosion resistance of the base material due to galvanic corrosion at the boundary between the weld and the base
[see Wipikedia: https://en.wikipedia.org/wiki/Aluminium_alloy#]

Several industrial and national standards are available for the correct choice of welding materials for each of the aluminium alloys.
In the US, the industrial standard AWS D1.2 - “Structural Welding Code – Aluminum” (among others) recommends fillers matching the high magnesium content of the 5083.
In the European Union, the Part 4 0f the EN standard EN 1011 provides recommendations for welding of aluminium alloys.

The H2TOOLS report of this event does not provide information on the filling process, the characteristic of the tanks involved (storage capacity, hydrogen flow, the filling process). They cannot be deduced either from the context, because industrial sector, application and reason for the ‘tanking’ are also unknown.

Nevertheless, the report is extremely interesting because it describes in detail the root cause analysis process which brought to the identification of the failure causes of the weld. The post-incident investigation used a fault-tree analysis approach identified five paths, i.e. 5 different potential failure mechanisms:
1. a manufacturing defect of the welding,
2. its ageing,
3. a wrong choice of materials or components,
4. overstress,
5. design error.
The analysis identified a manufacturing defect as the most probable root cause, consisting of 5 elements: (i) the section of an incorrect filler rod, (ii) an insufficient fusion in the weld, (iii) insufficient weld penetration, (iv) incorrect power level setting and (v) incorrect heat level applied during welding.
The investigators underpinned these conclusions based on the following evidences:
Laboratory chemical analysis confirmed that improper filler metal (4043 Al) was used to weld the 5083 aluminium alloy vent pipe. Standards do not recommend the 5083/4043 combination when welding aluminium alloys.
Non-destructive examination of the welds showed many critical defects, compared to other vent line welds.
Fractography analysis found that the failure had failed due tensile overload originating at the bottom surface of the vent line, and that it was probably the effect of one single overload event.

To identify additional weak points on the vent line, all other similar welds were assessed by means of visual inspection, conductivity tests (to identify the aluminium alloy class), helium leak tests, combined x-rays and ultrasonic non-destructive evaluations, and silicon etch tests to identify weld material.

Moreover, gauges were installed on the vent lines, which provided operative data for the assessment of weld stress. This allowed the identification of the welds which required clamshell repairs. Thanks to this, weld defect growth could now be monitored at periodic intervals.