Piping and Pressure Vessel Failures Associated with Secondary Stresses

Keeping pressure equipment operating reliably requires a clear understanding of its mechanical operating conditions, such as loads and load cycles. Primary loads/stresses, such as those associated with dead and live weights, and internal pressure, are easy to envisage. However, secondary stresses, often related to welding, displacements, changes in operational temperature, and deformation, are not as easily visualized. Yet these easily overlooked stresses result in costly failures of piping and pressure vessels. This article illustrates the importance of secondary stresses by presenting several failures resulting from them:

• A post-weld heat-treated (PWHT) caustic service NPS 1 carbon steel pipe failure.
• A high-temperature austenitic radiant tube assembly failure.
• A high-temperature boiler feed water service, environmental stress cracking failure.

Introductory Secondary Stress Example

A secondary stress is “developed by the constraint of adjacent parts or by self-constraint of a structure.”. The concept is abstract and illustrated below.

While testing the pressure capacity of a pipe fitting, strain gauges were instrumented along the spool, which included the fitting welded to the pipe and two non-PWHT welds between the spool and the sealing flanges (Figure 1). The welds between the fitting and the pipe had been PWHT. The applied hydrostatic pressure was high because the objective was to assess the fitting’s yielding pressure. Once the hydrostatic test was completed, the strain on the two non-PWHT welds was negative. The pressure had made the non-PWHT welds yield, and once it was removed, the elongated welds were compressed within the spool.

Figure 1. Spool tested during a hydrostatic test.

This compressive field is known in the pipeline industry, where hydrotests have been reported to reduce the stress-corrosion crack propagation rates. The rate reduction is associated with compressive residual stresses in front of the crack tips.

Secondary Stresses Influence Stress Corrosion Cracks (SCC)

Caustic Service NPS 1 Carbon Steel Pipe Failures

Several caustic service flanges to reducer socket welds leaked after less than five years of service. They operated at 200°F (93°C) and 42 psi (290 kPa). As per the NACE Caustic Soda Service Chart, the piping had been post-weld heat-treated. However, the deposited weld metal (DWM) had intergranular cracks characteristic of caustic SCC. See Figures 2 and 3.

The stress that caused the failure was likely due to mechanical bending of the small-diameter branch. Branch connection welds are subject to high mechanical pipe-bending loads. These secondary stresses make small-diameter connection welds, even after PWHT, susceptible to SCC.

Figure 2. On the left, a leaking flange-to-reducer piping socket weld. On the right, the socket weld crack radiates from the weld root.

Figure 3. Intergranular crack in the deposited weld metal.

High-Temperature Austenitic Radiant Tube Assembly Cracks

A radiant tube assembly leaked after only three years of service. It comprised a 0.75-inch OD 316L nipple, a 304H olet, and a 4-inch OD 25Cr-35Ni radiant tube. The nipple assembly was bent with respect to the radiant tube (see Figures 4 and 5). Penetrant testing (PT) identified at least two cracks on the outer diameter (OD).

Figure 4. Leaking radiant tube assembly. On the right, the cracks identified with penetrant testing are indicated.

Figure 5. The nipple assembly was bent.

The assembly operated at ~1300°F (704°C) and 35 psig (241 kPa). It carried an ethane/propane mix. A sister facility had experienced leaks from cracks in a similar assembly.

The nipple-to-olet weld crack identified with PT was perpendicular to the nipple axis. A cross-section revealed that the nipple-to-olet joint was a lap weld. It also showed the crack to run from the inside diameter (ID) geometric notch between the nipple and the olet toward the OD of the assembly. The straight oxide-lined straight crack (with minor dagger-shaped secondary cracks) had an appearance characteristic of thermal fatigue (see Figure 6). The ID surfaces of the various components were heavily carburized (see Figure 7). This crack initiated on the ID.

Figure 6. The nipple to olet weld was cracked.

Figure 7. Heavily carburized ID surface.

The olet-to-tube weld cracks identified with PT had an orientation perpendicular to the radiant tube’s axis. After cutting open the radiant tube, the cracks were visibly longer on the OD than on the ID (see Figure 8). During fabrication, the olet was placed on top of the tube and joined to the tube with a through-thickness fillet weld. The fractured surfaces had fan-like ridges characteristic of fatigue. The fatigue cracks were associated with the olet being pushed inward into the tube. The bend in the assembly is also consistent with the olet being pushed inward with the thermal expansion of the various joints (see Figure 9). This crack initiated on the ID.

Figure 8. Olet to radiant tube joint cracks.

Figure 9. Stresses that caused the radiant tube cracks. The olet was pushed into the radiant tube.

A third important crack could not be identified externally; it developed on the olet ID at the contact corner between the olet and the nipple (see Figures 10 and 11). It also had the oxide-lined straight crack, dagger-shape appearance characteristic of thermal fatigue. In addition, it had a thin filament between the cracked surfaces, likely associated with stress-relaxation cracking.

Figure 10. Stresses that caused the radiant tube cracks. The olet was pushed into the radiant tube.

Figure 11. Contact corner crack between the olet and the nipple.

According to ASME B31.3 “Process Piping,” from fabrication, these socket weld components must have a gap prior to welding of approximately 0.0625 inch (1.5 mm). However, from the extensive nipple bending shown in Figure 5, a much larger gap than that stated would have been needed to prevent interference between the nipple and the olet.

All these ID-initiated cracks are associated with high-temperature thermal expansion and thermal cycling. The thermal expansion resulted in longitudinal (instead of hoop) stresses and interference that are not easy to account for. These secondary stresses reduced the typical 20-year life of these components to three years. The owner replaced socket welds with butt welds and the sock-olet with an extruded outlet. The gaps between parts were modified to accommodate thermal expansion.

High-Temperature Boiler Feed Water Service Environmental Stress Cracking Failure

A large tube-and-shell heat exchanger developed through-wall cracking in the shell-side welds after three years of service. Subsequently, after repairs, the shell welds continued to develop cracks over the next few years. During one turnaround, an ultrasonic (UT) inspection found cracks in virtually all shell-side welds, which magnetic particle testing (MT) confirmed on extracted samples (see Figure 12). All of these welds had to be repaired to allow the exchanger to operate long enough to design and build a replacement. The cracking evaluation and repairs took 131 days, during which time the entire plant was offline.

Figure 12. Crack indications in the welds.

The exchanger was large in diameter and heavy-walled. The shell side contained treated ~465°F boiler feed water (BFW), and the tube side had a chemical process reaction stream. The exchanger shell was constructed of SA 543, Type B, Class 1, quench-and-tempered alloy steel (HY80), and the nozzles were the forging equivalent. The welds were completed using 110 ksi UTS filler metals. The welds were not PWHT because it was — and still is — optional in the ASME code, and ASME warned of temper embrittlement if PWHT was performed improperly.

The cracking analysis found that cracks initiated mainly at pits on the BFW-exposed surface and progressed along an intergranular path following prior austenite grain boundaries in the weld zones (see Figure 13). Only the DWM and heat-affected zones (HAZ) had cracks; the base material (BM) was free of cracks. The metallurgical analysis determined that the cracking was the result of SCC in the presence of high-temperature water.

Figure 13. Multiple, branched intergranular cracks in the welds.

Since the BM did not crack, the driving force behind the cracking was suspected to be very high stresses in the welds. Finite element modelling and direct strain measurements (ASTM E837) found that the residual weld stresses were as high as the yield strength of the material. The literature confirmed this can be the case. These secondary stresses, combined with the mechanical stresses from the operating pressure, result in highly stressed welds, making them susceptible to SCC. This type of failure was also found in the nuclear power industry, where turbine blades made of similar materials also cracked due to SCC. David et al. presented a graph in their paper showing the relationship between the total stress required to initiate SCC in various environments, including this scenario, which is close to “condensing steam” (see Figure 14).

Figure 14. Graph from David et al showing the stress relationship to SCC cracking initiation in various environments.

The problem was solved by changing the material of construction and using PWHT’d welds. These changes resulted in lower operating stresses, but more significantly, much lower secondary weld stresses.

Concluding Remarks

The various cracked components detailed had service lives typically reduced from more than 20 years to less than 5 years. In all instances, the cracks initiated from the ID and could not be identified easily visually or with OD surface inspection methods. For services with potential fatigue cycles, SCC, and time-dependent material degradation and elongation, overlooking them is costly since their effects control service life.