Case Study: Failed relief valve investigation submitted to our metallurgical testing laboratory.

One (1) failed relief valve from a 4” natural gas line was submitted to our metallurgical testing laboratory for a steel failure investigation. Reportedly, the failure occurred through the circumferential seal weld in the valve body, after approximately two years of operation. It was also reported that the valve was fabricated from CF8M stainless steel and that the relief line experienced service vibrations induced by a compressor.

Our client further noted that the seal weld was deposited over a threaded connection between two mating sections of the valve body, for the purpose of hermeticity and to prevent loosening of the threads. No further background information was available. We were requested to determine the mechanism responsible for the relief valve failure and identify any contributing detrimental material or welding related factors.

1. The opinion of our metallurgical testing laboratory based on the performed tests and examinations is that the relief valve failure occurred by a reverse bending fatigue mechanism. Fatigue cracking initiated at the center of the weld bead, which was deposited over the mating surfaces of two body sections, joined by a threaded connection. The weld joint geometry formed a plane of stress concentration through the center of the face, creating favorable conditions for initiation and propagation of fatigue cracking.

2. The initiation of the circumferential crack was attributed to the compressor induced service vibrations and with the associated cyclic bending stresses that exceeded the fatigue strength of the deposited weld bead. As a result, the valve assembly failed prematurely through the relatively thin (~.040”) weld bead cross-section.

3. SEM examination of the exposed crack surface revealed macroscopic features (ratchet marks at the face of the weld bead) and microscopic features (arrays of fine striations). The observed surface characteristics positively identified the failure mechanism as metal fatigue.

4. Metallographic examination of the failure region at the middle of the visible crack span revealed that the cracking propagated through the whole thickness (~.040”) of the deposited weld bead. The weld bead showed low penetration, which is common for seal welds, and contained no evidence of weld defects, such as porosity, slag inclusions, overlaps or undercuts.

5. Both the weld and the valve body showed dendritic microstructures typical of solidified filler metal and cast CF8M stainless steel, respectively. No evidence was observed of pre-existing steel defects, excessive nonmetallic inclusions, or any other detrimental base metal conditions that could have contributed to the failure.

6. Chemical analysis confirmed the valve body material as CF8M stainless steel.

7. Since no material or welding related defects were detected in the examined valve, it is our opinion that the operating stress conditions exceeded the fatigue rating of the supplied part. To alleviate similar valve failures in the future, we respectfully recommend reviewing the vibration induced service load conditions and upgrading the load bearing capacity of the replacement part. This can be achieved by selecting parts with an increased load bearing cross section of the valve body and the seal weld. If an upgrade of the part is impractical, another alternative is appropriate vibration management of the gas line. Dampening the vibrations will reduce the cyclic stress level and increase the useful fatigue life of the valve.

1. Visual and stereoscopic examination of the submitted relief valve revealed that the fracture occurred through the center of the circumferential seal weld (see Photos 1 – 2). The crack extended approximately 110° on the circumference, with the remaining bead surface apparently intact.

2. No evidence was observed of porosity, slag inclusions or undercut at the weld toes. The surface of the weld showed no evidence of mechanical damage or corrosion attack.

3. The valve body was sectioned through the crack termination points on the face of the weld, and the mating crack surfaces were separated for examination by scanning electron microscopy (SEM). In addition, weld cross-sections were removed from (a) the middle of the externally visible crack span; (b) the 90° location from the middle of the crack span; and (c) the 180° location from the middle of the crack span. All collected specimens underwent further examinations, described in the following sections of this report.

1. SEM examination of the exposed crack surface revealed macroscopic features in the form of multiple radial ledges (ratchet marks) located at the face of the weld bead (see Photo 3, arrows). These features are commonly associated with fatigue crack initiation sites.

2. Examination at high magnification revealed microscopic, evenly spaced, faint crack progression marks, known as fatigue striations (see Photo 4, arrows).

3. The observed macroscopic and microscopic features positively identified the failure mechanism as metal fatigue.

1. Metallographic examination of the failure region in the middle of the visible crack span revealed that the cracking propagated through the whole thickness (~.040”) of the deposited weld bead (see Photo 5). The 90° and 180° locations on the weld circumference contained no detectable cracks (see Photos 6 – 7).

2. The weld metal exhibited a dendritic microstructure typical of solidified filler metal. Neither examined weld cross-section contained any detectable porosity, slag inclusions, or evidence of overlapping or undercutting.

3. The valve body (base metal) revealed a dendritic duplex microstructure consisting of an austenitic matrix and multiple pools of delta-ferrite. These microstructural features are typical of cast CF8M stainless steel. No evidence was observed of pre-existing steel defects, excessive nonmetallic inclusions, or any other detrimental base metal conditions that could have contributed to the failure (see Photos 8 – 10).