Threaded connection rupture on a pressure gauge line

Hydrogen gas leaked from a broken pipe joint connected to a pressure gauge mounted on a hydro desulphurisation test unit.
The leak was indirectly detected as reactor pressure drop. The plant manager shut down the desulphurisation reactor

The leaking joint was a threaded connection of the two-compression-ring type. The back ferrule, which fixed the gauge pipe, was subjected to higher stress than the front ferrule. Excessive tightening generated cracks on the inner surface of the pipe near the ferrule section.

The INITIATING CAUSE was overstress of the pipe section involved in the joint assembly, combined with corrosion-assisted crack growth.
From post-incident chemical analysis found deposits containing chlorine and presumed that chlorine damaged the protective oxide film naturally protecting steel surfaces, leading to accelerated corrosion of cracks.

The failure occurred at a pressure gauge connection, a location without process fluid flow. This explains why it was not included in routine inspection items. Therefore, the equipment had been in service for 1 years, but the internal condition of the joint had never been examined. The ROOT CAUSE could be assigned tentatively to small glitches in the installation procedure and in the assessment of the lifetime of he component.

ferrule,
Swagelok-type connection,
gauge

The unit was a prototype design and was in steady state operation, nevertheless testing the effect of reduced pressure on the light oil feed. These testing conditions did not have any influnce on the failure of the component.

THE DETAILED CHRONOLOGICAL SEQUENCE OF THE EVENT:
3 Jan (Sun) 10:10 The operator (safety officer) started a preliminary sulfuration step in the test unit to evaluate performance of a prototype under reduced pressure light oil feed conditions.

31 Jan (Mon) 10:30 The operator completed the preliminary step and began the evaluation of the reduced pressure light oil feed condition.

7 Feb (Mon) 
• 20:00 The operator performed a routine patrol and confirmed no abnormality in the test unit.
• 20:20 The distributed control system (DCS) issued an alarm indicating a pressure drop in the reactor. The operator went to the site and observed the reactor inlet pressure falling from the operating pressure to 2 MPa (continually decreasing).
• ≈ 20:25 The operator reported the situation to the section manager. The manager ordered a stop of the test unit, pressure reduction, and temperature reduction as safety measures.
• ≈ 20:25 22:10 The operator stopped the test unit, reduced pressure and temperature, and then confirmed fluid leakage from the pressure gauge connection joint (reducer) downstream of the reactor.
• 22:10 The section chief placed a 119 emergency call.

9 Feb (Wed) The operator removed the pressure gauge downstream of the reactor and confirmed rupture of the reducer (see Fig. 1 2).

The joint which failed in this event was a two compression ring type threaded joint. Gas tightness is achieved by the compression of two ‘ferrules’ on the pipes to be connected. Especially the back ferrule, which fixes the tube, is subjected to higher stress than the front ferrule and transmits these stresses to the tube.
During the assembly of this joint, excessive tightening generated cracks on the inner surface near the ferrule portions.

Leading manufacturers of this type of joint provide guidelines on how to assemble their fitting, including the correct choice of the pipe materials and detailed instruction on how to close manually the nuts and the quantity of turns to fix the threaded assembly. They do not assign, however, a quantitative torque value, so that the fixing operation contains a certain margin of personal interpretation. This could be improved by plant operators by prescribing a certain torque value with a dynamometric wrench. After installation, it could be also possible to check with a gap inspection gauge if the nut-to-nut distance is correct. Finally, manufacturers offer also on-site training.

The regular internal inspection of this type of joints could result rather tedious and/or unsuitable for complex installations. They imply also the replacement of parts. Possible alternatives could consider pressure tests combined with an external inspection and replacement at regular base.

A second lesson from this small incident is that must be taken for steel in a chlorine environment, because stainless steel is vulnerable to chlorine induced oxide film breakdown. Moreover, additional element can contribute to a reduced lifetime of stressed material in presence of chlorine: machining, forming, or other processes can alter grain size, shape, and composition segregation, potentially reducing strength and corrosion resistance compared with the as produced material.

(1) An inspection of this type of joints was introduced: it requested, ever 3 years, to open the pressure gauge connection, to inspect the inner surface, to clean any deposits and to replace the joint if cracks, defects or corrosion were found.
(2) Guidelines on how to properly tighten this type of joints were issued, aiming at precenting material and component damage. They included a “one point lesson” sheet to ensure technique transfer.
(3) Trainings by external instructors were organised on equipment maintenance and upkeep.
(4) The entire line downstream of the reactor pressure gauge for all units of the plant was replaced.