INVESTIGATION INTO THE FAILURE OF A 16 inch PIPE BEND IN A NOX GAS PIPELINE AT A NITRIC ACID PLANT

J.C.Guild: Southern African Institute of Welding, Johannesburg, South Africa

Key Words: Pipe, Bend, Failure Investigation, Sensitization, Knife-line corrosion, Weld, Intergranular corrosion

ABSTRACT

The investigation into the failure of a Titanium stabilized (Type 321) stainless steel pipe bend, in NOX gas service, is described. A metallurgical examination revealed that the materials used appear to be acceptable to the relevant specifications, but that the failure was due to “knife-line” corrosion adjacent to the welds. This occurred due to the presence of condensed Nitric / Nitrous acid in the line. Recommendations are made to monitor the rest of the line.

1. INTRODUCTION

A catastrophic failure of a bend occurred in a NOX gas pipeline at a Nitric Acid plant. The bend that failed formed part of a 406 mm (16") OD, 3.0 mm nominal wall thickness pipeline between a cooler condenser and tail gas heater. The 90° bend was fabricated from two formed plates welded longitudinally. The longitudinal welds lay along the inside and outside radii respectively. The bend joins a horizontal pipe section from the tail gas heater at one end and an inclined 45° bend into the cooler condenser at the other end.

A section of the pipe comprising the entire failed bend was blown clear from the remainder of the line as a result of the burst. This section was sent to the author with a request to determine the cause of failure.

The line carries NOX gas at a temperature of 60°C and at a pressure of 9 bar. The material of construction of the pipe bend was Werkstoff Nr. 1.4541, an equivalent to ASME SA-312 TP321 titanium stabilised austenitic stainless steel. This particular line had been in service for 18 years. It was reported that the line had not been inspected for several years.

2. VISUAL EXAMINATION

2.1 Inspection Of The Burst Section

The section of the pipe bend that was blown away from the line was severely deformed by impact damage (Figure 1).

Examination indicated that the bend had ruptured along the inside radius longitudinal weld causing the bend to flare open and then tear alongside the circumferential welds on either end. The flaring of the pipe section was pivoted about the axis of the outside radius longitudinal weld. The corroded fracture face of the outside surface of the failed longitudinal weld is shown in Figure 2. This shows that there was hardly any fresh metal exposed alongside the weld after failure.

The length of the fracture along the failed longitudinal weld was approximately 550 mm between the T-joints formed with the circumferential welds on either end of the section. From the one T-joint for approximately 124 mm, the amount of exposed fresh metal along the longitudinal weld was in the region 0.5 mm with respect to wall thickness. For another 220 mm, the amount of exposed fresh metal increased gradually to a peak of 2 mm and then gradually decreased back to 0.5 mm. Over the remaining section of 206 mm, the freshly exposed metal was in the region of 0.5 mm up to the opposite T-joint.

All of the fresh fracture face, that could be examined, was typical of a ductile shear mechanism.

The longitudinal weld on the outside radius opposite the failed weld in the inside radius displayed a smooth regular profile, which is indicative of an automatic welding process having been used to fabricate the bend. Cracking alongside this weld was observed. Cracking alongside a section of a circumferential weld was also noted. The circumferential welds were larger and more irregular in nature than the longitudinal welds, indicating manual welds.

The manufacturer's markings were visible on the failed section. These markings, which were observed on the outside of the failed pipe section, gave an indication of the material of construction. The material indicated on the pipe section was Werkstoff Nr. 1.4541, an equivalent of type 321 stainless steel, and the wall thickness was indicated to be 3.2 mm.

The actual wall thickness of the bend was measured to be 2.80 mm adjacent the failed weld and again 2.80 mm at a section well away from the failure. This represents a loss of 0.024 mm/year over the pipeline service period.

2.2 On-Site Inspection Of The Remaining Two Ends Of The Line

The two mating ends from where the burst section blew away were also examined. Ductile shear fracture occurred around the circumference of both ends of the failure with the fracture generally found to have run alongside the circumferential welds (Figure 3). A repair weld was observed on the circumferential weld of the straight section.

3. MICROSCOPIC EXAMINATION

Samples were taken from the longitudinal and circumferential welds at various positions for examination. General, shallow intergranular corrosion of the parent metal was noted on all the samples examined.

3.1 Failed Longitudinal Weld

The cross section of this weld revealed an austenitic stainless steel parent metal, and a dendritic austenite/ferrite weld metal. Titanium carbides were observed in both the parent metal and the weld metal. The presence of these carbides was indicative of a stabilised grade of stainless steel.

Oxidation of the fracture face alongside the weld metal was observed. The fracture itself was observed to have propagated alongside the weld metal in the region of the fusion line. Plastic deformation of the weld metal at the corner cap of the weld was observed (Figure 4). This region of plastic deformation was where final failure occurred leaving behind freshly exposed metal. Along the fracture face, some parent metal was observed to have remained bonded to the weld metal. In this region intergranular corrosion was observed.

On the other side of the weld, an intergranular corrosion crack was observed to have initiated at the inside surface of the pipe in the region of the fusion line (Figure 5).

Evidence of some grain boundary sensitisation and grain boundary ferrite was observed in the region of the crack shown in Figure 5. Preferential corrosion of the ferrite phase in the weld filler metal on the inside of the pipe surface was also observed.

3.2 Outside Radius Longitudinal Weld Opposite The Failed Longitudinal Weld

The microstructure of the weld and parent metal were similar in structure to the failed weld. lntergranular corrosion of the surface of the parent metal was observed. The onset of knife-line attack or intergranular corrosion at the fusion boundary on the inside diameter of the pipe was noted.

Grain boundary ferrite in the HAZ adjacent the fusion boundary was observed. Evidence of grain boundary sensitisation was noted in the region of knife-line attack.

3.3 Section Of A Circumferential Weld

The weld bead was observed to be larger and more irregular than the longitudinal weld beads. However, the microstructure was similar to that of the longitudinal welds. Once again titanium carbides were observed in both the parent metal and weld filler metal. Figure 6 shows the preferential corrosion of the ferrite phase in the weld filler metal.

lntergranular corrosion of the surface of the parent metal was observed. No significant knife-line attack was observed on this sample. Also, there was no evidence of significant grain boundary sensitization, but grain boundary ferrite in the HAZ adjacent the fusion lines was noted.

4. CHEMICAL COMPOSITION

The chemical composition of the material of construction of the pipe bend was determined (Table 1).

ELEMENT	PIPE SECTION (WT. %)	ASME SA 312 TP321	WERKSTOFF NR. 1.4541
Carbon	0.049	0.08 max	0.08 max
Manganese	1.51	2.00 max	2.00 max
Phosphorus	0.030	0.040 max	0.045 max
Sulphur	0.009	0.030 max	0.030 max
Silicon	0.56	0.75 max	1.00 max
Nickel	10.0	9.00 – 13.0	9.00 – 12.00
Chromium	17.8	17.0 – 20.0	17.00 – 19.00
Molybdenum	0.04	-	-
Titanium	0.296	5x%C min/0.70 max	5x%C min/0.80 max
Copper	0.16	-	-
Niobium	£0.01
Aluminium	0.036	-	-
Vanadium	0.068	-	-

Table 1: Chemical analysis of the failed pipe section.

The results confirm the material of construction of the failed pipe bend was a titanium stabilised type 321 stainless steel.

The weld metal used for the failed longitudinal weld was also analysed, the results of which are shown in Table 2.

ELEMENT	WELD METAL (WT. %)	ASME SFA-5.9 ER321
Carbon	-	0.08 max
Manganese	1.55	1.0 – 2.5
Phosphorus	-	0.03 max
Sulphur	-	0.03 max
Silicon	0.74	0.30 – 0.65
Nickel	8.84	9.00 – 10.5
Chromium	19.2	18.5 – 20.5
Molybdenum	0.11	0.75 max
Titanium	0.11	9x%C min./1.0 max.
Copper	0.48	0.75 max.
Cobalt	0.30	-
Aluminium	£0.01	-
Vanadium	0.037	-

Table 2: Chemical analysis of the failed longitudinal seam weld metal.

The chemical analysis results obtained suggest that the weld was a type 321 stainless steel filler metal.

5. DISCUSSION

The evidence of this investigation indicates that the pipe bend ruptured by a plastic collapse mechanism. This was caused by the wall thickness of the bend having been reduced to the extent that the remaining ligament was unable to resist the stress induced by the pressure in the line. The applicable pipeline specification indicates that required wall thickness for pipes and bends should be 3.0 mm and 4.0 mm respectively. In the event it appears that bends with 3.2 mm wall thickness were used. However, this does not appear significant as design calculations indicate that the minimum allowable wall thickness (incorporating safety factors) for pipe sections and bends in 321 type stainless steel is 1.9 mm. Examination of the fracture faces indicates that when failure occurred a significant portion of the longitudinal weld on the inner radius had been reduced to an effective wall thickness of about 0.5 mm.

General thinning of the bend material had occurred in service but this was not significant. The cause of the failure was found to be knife-line corrosion attack at the fusion boundary of the longitudinal weld on the inner radius. Knife-line corrosion of 321 type stainless steels is a well known phenomenon in nitric acid plant applications. It is a form of preferential intergranular corrosion by nitric/nitrous acid which occurs immediately adjacent to welds and is attributable to sensitisation of the stainless steel parent material in the vicinity of the fusion boundary. Preferential corrosion of grain boundary ferrite also appears to have been a contributory factor in the failure.

Titanium is alloyed in 321 stainless steel to 'stabilise' the carbon content by the formation of titanium carbides in the steel. This counteracts sensitisation which can occur, notably in a weld heat affected zone (HAZ), as a result of grain boundary chromium carbide precipitation. However, near the fusion boundary titanium carbides are redissolved putting carbon back into solid solution. Subsequent cooling through the temperature range 850 – 450°C, or the heat effect of second or subsequent weld passes can lead to chromium carbide precipitation depleting grain boundary areas of chromium and rendering the material susceptible to intergranular corrosion attack. Attack by nitric/nitrous acids can occur in normal operation, if gas condensation occurs, or at plant shutdowns. In this case it has been confirmed by process personnel that condensate will be present in this line. Various factors can be influential in the severity of knife-line damage. These include the material susceptibility which is influenced by its processing and fabrication history, the frequency and duration of exposure to acid formation, acid concentrations and temperatures, with condensing and re-boiling conditions being particularly severe. In addition the positioning of welds can be significant.

The low carbon, unstabilised grade 304L type stainless steel is less susceptible (but not immune) to HAZ intergranular corrosion effects and should be considered as an alternative material of construction in this type of application. However, care should be taken to assess the merits of each case of material substitution.

Inspection of plant equipment for knife-line attack is difficult since the various non-destructive testing techniques all have limitations in their capability of providing definitive information. Where access is available visual examination, perhaps assisted by dye penetrant testing can be used but will not provide a definitive measure of the severity of attack. Shear wave ultrasonic testing is not suitable for examination of welds in thin gauge stainless steel. Compression wave methods may provide information if weld caps can be ground off but this would also not be easily recommended on thin gauge material. Radiographic examination of welds can be used to confirm the presence of knife-line corrosion and will give some indication of the severity of attack. However, destructive sectioning to provide samples for microscopical assessment should be used to support radiography where this is possible.

5. CONCLUSION AND RECOMMENDATIONS

The pipe bend failure was caused by knife-line corrosion adjacent to a longitudinal seam weld. Local wall thinning occurred and the remaining wall thickness was unable to sustain the pressure in the pipe line.

An appropriate on-going inspection program should be incorporated into plant maintenance programs. This program needs to be tailored to suit each item or type of plant equipment.