A leak was discovered at a Nitrate plant on a 3.5” outer diameter (OD) stainless steel (SS) pipe carrying Ammonium Nitrate (AN). High readings on the steam conductivity monitoring program triggered an investigation, which resulted in finding that AN leaked into the steam jacket.
By Ibrahim M. Gadala, Ph.D., P.Eng.
The pipe was made from 304L Stainless Steel (SS) manufactured to the ASTM A312 specification and was situated in 5” carrier pipe containing 750 kPa steam at ~175ºC. The failed AN line operated at ~1013 kPa between 144ºC and 153ºC. The internal surface of the failed section is shown in Figure 1. Cutouts of the failed section, an intact equivalent nearby, and the carrier pipe were submitted for failure analysis. This article summarizes the scope, methods, results, and interesting findings from the failure analysis conducted.
1. Scope
The author was requested to conduct an analysis to identify the cause(s) of failure. The thorough scope of work spanned three phases, summarized in the following steps: (1) detailed visual examinations (2) analysis of chemical composition of scale from surfaces, (3) microscopic and chemical examinations of cracks and materials/scale therein (4) comparing material microstructures at the failure location to intact location, (5) fractography, and (6) hardness measurements as per NACE MR0175/ISO 15156 at “bad” and “good” areas. In this article, only results from steps (1)-(5) are presented, but discussions and conclusions are based on findings from all steps (i.e., 1-6).
Figure 1: Overall View of Internal Surface of Failed Section of AN Pipe after Splitting in the Longitudinal Direction (Left); Zoomed View of the Linear Crack and Bulge at Failure, Seen from the Root of the Weld (Right).
2. Examination
2.1. Visual Examinations
Visual examinations of the failed and intact samples were conducted at first. In the failed sample, evidence of a bulge and a slight “fish-mouth” rupture appearance from the external surface indicated the presence of plastic deformation. The internal surfaces were uniformly covered with a dark well-adhered scale. A linear groove was evident on one side of the weld root along the full length of the sample, with a non-penetrating crack along it (Figure 2, Left). No visual evidence was found to indicate that flow direction contributed to the failure. The intact sample was also uniformly covered with a dark well-adhered scale (Figure 2, Right). However, the weld visibly appeared in good condition from the OD and no cracks were observed anywhere on the external surface.
2.2. Chemical Composition of Scale
X-ray diffraction (XRD) and energy-dispersive X-ray spectroscopy (EDS) analyses were conducted on the internal corrosion product/scale and deposits taken from the failed “bad” and intact “good” samples. The general classes of anionic components of the compounds found were oxides/hydroxides (O, O2, OH), nitrate (NO3), and sulfide (S). All could be reasonably attributed to corrosion products of aqueous internal media, remnants of AN, or the base 304L SS metal. The cationic elements of the compounds found were Fe, Ni, Cr, Si, NH4, Zn, Ti, and Al. These are also consistent with the SS base metal, inclusions (even though not to specification of 304L SS), or remnants of the AN internal environment.
Figure 2: One Side {A} of the Failure Location {B: not shown} with Groove and Extended Crack Along Weld Root (Left); Internal Scale and the Weld of the Intact “Good” Section (Right).
2.3. Metallographic/Microstructural Examinations
A cross-section for each sample was prepared for metallographic/micro-structural examination. Microscopic examination on the failed sample showed a crack initiating from the internal diameter (ID) at the groove in the weld root, and partially penetrating approximately half the wall thickness (WT) as seen in Figure 3. The pipe’s WT was noticeably smaller on the crack-side of the weld compared to the opposite un-cracked side, localized to what appears to be “necking”. Also, the crack did not start on the fusion line between the weld and base metals, indicating the unlikelihood of weld lack of fusion as the root cause. Finally, it was seen that secondary cracks and the main crack tip itself propagated up until the fusion line, most likely due to hardness variations as will be discussed in upcoming sections.
Microscopic examination on the intact sample also showed a crack initiating from the ID at the groove in the weld root. This one partially penetrated only 10% of the WT as seen in Figure 4. Again, the pipe’s WT was noticeably smaller on the crack-side of the weld compared to the opposite un-cracked side. Similarly, the crack did not start at the line of fusion at the root, but secondary cracks and the main crack tip itself halt at this line. The crack in the good sample was visibly thinner (hairline) versus that of the bad sample.
Microstructural analysis revealed that the base material of both micros exhibited a relatively uniform single-phase austenitic microstructure in all areas. Twins in a few grains suggest the material was cold-rolled as a part of the manufacturing process1. Annealing twins, formed as a consequence of the recrystallization of deformed cubic-close packed metals such as austenitic iron in 304L SS2, are typically more abundant than what was seen in these samples. This indicates the material was not annealed prior to manufacturing or welding. Nonetheless, the microstructure seen across all areas is what’s expected for 304L SS, and the lack of microstructural differences between “bad” and “good” areas suggests that microstructure was not a contributor to the failure.
2.4. Fractography
Figure 5 shows low-magnification stereoscopic and high-magnification Scanning Electron Microscope (SEM) images taken of the fracture surface. The images indicate fatigue crack growth in a cleavage region of the fracture face closer to the ID, where clear striations appear axially. Striations are characteristic of fatigue failures in pipes where crack propagation occurs intermittently under cyclic pressure creating cyclic hoop stresses. Closer to the OD, microvoid features were seen in the protruded “shear lip/bevel” type region of the fracture face. Microvoid formation and coalescence is characteristic of ductile overload at the final stage of fracture.
Figure 3: Microscopic View of Crack Near Failure Location Starting at the Root of the Weld and Propagating in the Base Metal – 1.6X Magnification (Top); Zoomed Image of Left – 200X Magnification (Bottom).
Figure 4: Detailed View of Cross-Section from “Good” Sample Showing Crack Starting at the Root of the Weld and Propagating Through the Base Metal – 200X Magnification.
3. Discussion
Based on material evidence shown here (and other evidence/details not shown), fatigue from temperature-pressure cycles was found to be the main contributing factor to the failure:
- Fatigue results in a single, non-branched, transgranular crack perpendicular to the surface, consistent with what was found at the failed location.
- Striations in the axial direction were clearly visible in the cleavage region of the fracture face closer to the ID, characteristic of fatigue failures.
- The failure location was down-stream of a pump which creates pulsations in the pressure output3. The cyclic stresses could also be a result of thermal variations in the steam jacketing system.
- No significant corrosion products were found on the fracture faces or embedded in the cracks. Thus, a contribution of corrosion to the fatigue process (i.e., corrosion-fatigue) is considered minimal or nonexistent.
- A groove was present at the weld toe before the crack initiated. The internal surfaces of the groove developed a very tenacious corrosion product/scale, consistent with the groove pre-existing the crack and pre-dating the line’s service (i.e., from manufacturing). Several points could explain/elaborate on this:
- A pre-existing manufacturing artifact of the butt-weld end preparation prior to the actual welding.
- An internal overpressure event (e.g., from plugging), which plastically deformed the area near the weld seam, could leave behind the groove. This is less likely than the previous option since it would be localized to the specific overstressed areas.
- Issues with manufacturing were further corroborated by the existence of a filler material (weld metal) in the welds analyzed. This is not compliant with Clause 6.1.3 of ASTM A312 which states that “welded pipe shall be made using an automatic welding process with no addition of filler material”.
- The groove near at the weld toe created the necessary stress concentrations to initiate the fatigue crack. In calculations using the ASME B31.3 code for pressure piping, based on an allowable stress value of 16.7 ksi (115 MPa) for 304L SS pipes and tubes at 300ºF (150ºC), crack tip stresses were shown to be greater than typical fatigue threshold values, hence fatigue crack propagation was possible.
Figure 5: Stereoscopic 1.0X Magnification of Part of the Fracture Surface (Top Left); Overview SEM Image of Same Area (Top Right); Zoom of Cleavage Area {A} Therein, Showing Flat Ridges and Secondary Cracks (Bottom Left); Zoom of Ductile Overload Area {B} Therein, Showing Microvoid Features (Bottom Right).
4. Summary and Recommendations
Results from the examination presented in this article indicate the 3” 304L SS pipe carrying AN solution failed due to fatigue. This is supported by striations on the fracture face seen under low and high magnification microscopy, showing intermittent crack growth with cyclic operating conditions. Cracking initiated at a pre-existing groove near the weld toe in the pipe and most likely propagated due to the pipe’s exposure to temperature-pressure variations in service. The groove was shown to most likely preexist service as a manufacturing defect, a trend further corroborated by the presence of a filler material (weld metal) in both samples being noncompliant with the ASTM A312 specification the pipes were constructed to. Evidence indicates this groove created the necessary stress concentrations to initiate the fatigue crack based on ASME B31.1 pressure piping code calculations.
Possible remedial actions which can help mitigate such failures in the future include:
- Replace the existing filler-welded SS line with seamless variants compliant with ASTM A312.
- Ensure no stress concentrators exist in newly procured pipes prior to installation, as per applicable codes.
- Dampen pressure fluctuations at their source by installing a pulsation dampener downstream of the pumps.
- Ensure overpressure protection devices are installed and periodically checked and maintained to prevent surges in pressure above acceptable design limits.
- Post-weld heat-treat (PWHT) welds to enhance their ductility against fatigue and other forms of cracking.
- Periodically inspect representative locations along a cyclically loaded line according to a risk-based inspection (RBI) program specifically addressing fatigue/corrosion-fatigue.
References
1. http://www.pennstainless.com/stainless-grades/300-series-stainless-steel/304l-stainless-steel- 2/
2. B. B. Rath, M. A. Imam and C. S. Pande, Mater. Phys. Mech. 1 (2000) 61.
3. https://www.pumpsandsystems.com/peristaltic-pumps/april-2014-sizing-pulsation-dampeners- critical-effectiveness
Acknowledgements
This investigation was conducted during the author’s role as a Materials Engineer and Failure Analyst at Skystone International Inc. (an Acuren Company). The author would like to acknowledge Prakash Dodia for conducting the laboratory work and Dr. Alex Tatarov for technical guidance and review.