A forged pressure vessel made from high temperature austenitic steel X8Cr-Ni-MoVNb 16 13 K (DIN 1.4988) failed. The widest part of the burst had fine cracks on the internal wall running longitudinally. When the internal wall was cleaned, numerous even finer cracks were exposed. On the fracture surfaces in this region an irregularly formed zone was visible in the direction of the internal wall and a fibrous oriented fracture zone towards the external wall. The fracture was typical of stress-corrosion cracking in austenitic steels. Vanadium trichloride was present and tensile stresses were of necessity set up by the internal pressure. Stress-corrosion cracking does not occur if one of the basic requirements is lacking. Because the chloride agent and tensile stresses were inevitably present, the only possible way to prevent future reoccurrence is to forge the entire pressure vessel from a material immune to stress-corrosion cracking or to use interchangeable linings of such a material. A nickel alloy could be considered.

This article discusses the effect of using unsuitable alloys, metallurgical discontinuities, fabrication practices, and stress raisers on the failure of a pressure vessel. It provides information on pressure vessels made of composite materials and their welding practices. The article explains the failure of pressure vessels with emphasis on stress-corrosion cracking, hydrogen embrittlement, brittle and ductile fractures, creep and stress rupture, and fatigue with examples.

The flanged bearing bush carrying the drive shaft of a feed pump suddenly fractured after about two years of service. The chemical composition was normal for high chromium ledeburitic cast steel, which was corrosion and wear resistant as well as refractory. For unknown reasons the rotating shaft came into direct contact with the flange. Mechanical friction caused a rise in temperature on both contact surfaces. This mutual contact lasted long enough for the temperature in the contact zone to exceed 1200 deg C, at which the flange material became softened or molten. As a result, considerable structural changes took place on the inner wall of the flange. Thermal stresses and excessive mechanical loads due to smearing of the flange material then led to fracture of the flange.

A Ti-6Al-4V alloy pressure vessel failed during a proof-pressure test, fracturing along the center girth weld. The girth joints were welded with the automatic gas tungsten arc process utilizing an auxiliary trailing shield attached to the welding torch to provide inert-gas shielding for the exterior surface of the weld. A segmented backup ring with a gas channel was used inside the vessel to shield the weld root. The pressure vessel failed due to contamination of the fusion zone by oxygen, which resulted when the gas shielding the root face of the weld was diluted by air that leaked into the gas channel. Thermal stresses cracked the embrittled weld, exposing the crack surfaces to oxidation before cooling. One of these cracks caused a stress concentration so severe that failure of the vessel wall during the proof test was inevitable. A sealing system at the split-line region of the segmented backup ring was provided, and a fine-mesh stainless steel screen diffuser was incorporated in the channel section of the backup ring to prevent air from leaking in. A titanium alloy color chart was furnished to permit correlation of weld-zone discoloration with the degree of atmospheric contamination.

During autofrettage of a thick-wall steel pressure vessel, a crack developed through the wall of the component. Certain forged pressure vessels are subjected to autofrettage during their manufacture to induce residual compressive stresses at locations where fatigue cracks may initiate. The results of the autofrettage process, which creates a state of plastic strain in the material, is an increase in the fatigue life of the component. Analysis (visual inspection, 50x/500x unetched micrographs, and electron microprobe analysis) supports the conclusion that the fracture toughness of the steel was exceeded, and failure through the wall occurred because of the following reason: the high level of iron oxide found is highly abnormal in vacuum-degassed steels. Included matter of this nature (exogenous) most likely resulted from scale worked into the surface during forging. Therefore, it is understandable that failure occurred during autofrettage when the section containing these defects was subjected to plastic strains. Because the inclusions were sizable, hard, and extremely irregular, this region would effect substantial stress concentration. No recommendations were made.

A type 321 stainless steel (AMS 5570) pressure-tube assembly that contained a brazed reinforcing liner leaked during a pressure test. Fluorescent liquid-penetrant inspection revealed a circumferential crack extended approximately 180 deg around the tube parallel to the fillet of the brazed joint. The presence of multiple origin cracks was indicated on the inside surface of a fractured portion of the crack surface. The cracks had originated adjacent to the braze joining the tube and the reinforcing liner and propagated through the wall to the outer surface. The residues on the inner surface of the tube were identified as fluorides from the brazing flux by chemical analysis. The nature of the crack, potential for corrosion due to residual fluorides and residual swaging stress in the tube prior to brazing, confirmed that failure of the tube end was due to stress-corrosion cracking. Stress relief treatment of tube before brazing and immediate cleaning of brazing residual fluorides was recommended to avoid failure.

A pressure probe assembly comprised of type 347 stainless steel housing, brazed with AMS 4772D filler metal to the pressure probe, failed due to detachment of a rectangular segment from the housing. The presence of a large brazing metal devoid region in the pressure probe-housing joint was revealed by visual examination. Fatigue marks, emanating from multiple crack origins on the inside surface of the housing at the brazed joint were revealed by further study of the fracture. A poor metallurgical bond was confirmed by the presence of large irregular voids, flux trapped braze metal and separation between braze and housing.

A portable propane container with a name-plate soldered onto it exploded in service. When the vessel was inspected afterwards, it was found to have developed a crack in the top end plate. A portion of the end plate cut out to include the midlength and one termination of the crack was examined microscopically. This revealed that the crack was associated with intergranular penetration by molten metal. The microstructure in general was indicative of a good-quality mild steel. It was evident from that solder that was responsible for the penetration and that fused brass from the hand wheel had not played any part. Tensile stress was present at the time of the failure sufficiently high to enable solder penetration to take place. The use of soft solder as a medium for attaching name-plates directly on to stressed steel parts is not recommended. It would be preferable to use a welded-on patch plate or to employ one of the high-strength, non-metallic adhesives.

Following a fracture mechanics “fitness-for-purpose” analysis of petroleum industry cold service pressure vessels, using the British Standard PD 6493, it was realized that an analogous approach could be used for the failure analysis of a similar pressure vessel dome which had failed in service some years previously. The failed pressure vessel, with a diam of 2.5 m and several meters tall, had been made of 12 mm thick IZETT steel plate of the same type and heat treatment as used in the earlier fitness-for-purpose already measured. Examination of the fracture surfaces suggested, from fatigue striations manifested by SEM, that the vessel was subject to significant fatigue cracking, which was probably corrosion assisted. From COD measurements at the operating temperature of -130 deg C (-202 deg F), and a finite stress analysis, a fracture mechanics evaluation using BS PD6493 yielded realistic critical flaw sizes (in the range 51 to 150 mm). These sizes were consistent with the limited fracture surface observations and such flaws could well have been present in the vessel dome prior to catastrophic failure. For similar pressure vessels, an inspection program based on a leak-before-break philosophy was consequently regarded as acceptable.

A large pressure vessel that had been in service as a hydrogen sulfide (H2S) absorber developed cracks and began leaking at a nozzle. The vessel contained a 20% aqueous solution of potassium hydroxide (KOH), potassium carbonate (K2CO3), and arsenic. The vessel wall was manufactured of ASTM A516, grade 70, low-carbon steel plate. A steel angle had been formed into a ring was continuously welded to the inside wall of the vessel. The groove formed by the junction of the lower tray-support weld and the top part of the weld around the nozzle was found to have a crack. Pits and scale near the crack origin were revealed by microscopic examination and cracking was found to be transgranular. Periods of corrosion alternated with sudden instances of cleavage, under a tensile load, along preferred slip planes were interpreted during examination with a microscope. It was concluded that the combination of the residual plus operating stresses and the amount of KOH present would have caused stress corrosion as a result of caustic embrittlement. It was recommended that the tray support should be installed higher on the vessel wall to prevent coincidence of the lower tray-support weld with the nozzle weld.

During periodic inspection of the tubes of a reformer furnace, a soapy water leak test with the tubes pressurized with nitrogen was being carried out by site personnel in a manner contrary to the policy of the organization when one of the tubes suddenly disintegrated with explosive violence. The tube approximately 30 ft. long by 6 in. diam, was constructed of three spun cast sections butt welded together. The material specified for tubes for this service was basically a 25% chromium, 20% nickel, cast stainless steel containing 0.4% carbon to optimize creep resistance. Failure initiated in the region of the tube where the dark fracture surface and columnar grain structure were evident. These features indicated the presence of a defective zone or progressive cracking which had occurred during service. Microscopic examination of sections through the zone revealed extensive creep cracking. The cracking was intergranular and followed the interdendritic columnar structure adjacent to the outer surface.

A large pressure vessel designed for use in an ammonia plant failed during hydrostatic testing. It was fabricated from ten Mn-Cr-Ni-Mo-V steel plates which were rolled and welded to form ten cylindrical shell sections and three forgings of similar composition. The fracture surfaces were metallographically examined to be typical for brittle steel fracture and associated with the circumferential weld that joined the flange forging to the first shell section. Featureless facets in the HAZ were observed and were revealed to be the fracture-initiation sites. Pronounced banding in the structure of the flange forging was revealed by examination. A greater susceptibility to cracking was interpreted from the higher hardenability found within the bands. Stress relief was concluded to have not been performed at the specified temperature level (by hardness and impact tests) which caused the formation of hard spots. The mode of crack propagation was established by microstructural examination to be transgranular cleavage. It was concluded that failure of the pressure vessel stemmed from the formation of transverse fabrication cracks in the HAZ fostered by the presence of hard spots. It was recommended that normalizing and tempering temperatures be modified and a revised forging practice explored.

A spherical carbon steel fixed-catalyst bed reactor, fabricated from French steel A42C-3S, approximately equivalent to ASTM A201 grade B, failed after 20 years of service while in a standby condition. The unit was found to contain primarily hydrogen at the time of failure. The vessel had a type 304 stainless steel shroud around the catalyst bed as protection against the overheating that was possible if the gas bypassed the bed through the refractory material. The failure was observed to have begun at the toe of the shroud-support ring weld. The ring was found to have a number of small cracks at the root of the weld. The cleavage mode of fracture was confirmed by SEM. The presence of extensive secondary cracking and twinning (Neumann bands) where the fracture followed the line of the shroud-support ring was revealed by metallography. It was revealed by refinery maintenance records that the ring had been removed for hydrotest and welded without any postweld heat treatment. The final cause of failure was concluded to be cracking that developed during the installation of the new shroud ring. Stress-relief heat treatments were recommended to be performed to reduce residual-stress levels after welding.

In January 1965, a Reaction Control System (RCS) pressure vessel (titanium alloy Ti-6Al-4V) on an Apollo spacecraft cracked in six adjacent locations. It used N2O4 for vehicle attitude control through roll, pitch, and yaw engines, and was protected from the N2O4 by a Teflon positive expulsion bladder. Investigation (visual inspection, pressure testing of 10 similar vessels, and chemical testing of the N2O4) supported the conclusion that the failure was due to stress corrosion from the N2O4, and specifically from a specification change in the military specification MIL-P-26539. Recommendations included revising the specification to require a minimum NO content of 0.6%.

An external tank pressure/vent valve regulates the external tank fuel feed system, which transfers fuel under pressure to the internal tanks of the aircraft. A dual-position valve was found to be sticking at the intermediate positions. Also, service air check valves located on the incoming lines contained poppets that were being stuck in a closed or partially closed position because of suspected corrosion product. Residue taken from the check valve poppet and from the dual-position valve was chemically analyzed. Chloride was present in both samples. It was suspected that moisture entering the service air lines left a chloride-containing compound upon evaporation within the air check valves and pressure/vent assembly. This compound subsequently reacted with the anodized, dichromate sealed check valve housing to lock the check valve poppets in a closed or partially closed position, decreasing the actual pressure being supplied to the pressure/vent valve. It was recommended that an inspection be conducted to ensure that the service air check valves are operating properly prior to removal and servicing of the pressure/vent valve assembly. It was also recommended that dry-film lubricant be checked to ensure that it meets specifications for the pressure/vent valve assembly.

A high-pressure steam pipe specified to be P22 low-alloy steel failed after 25 years of service. Located at the end of the steam line, the pipe reportedly received no steam flow during normal service. Visual examination of the failed pipe section revealed a window fracture that appeared brittle in nature. Specimens from the fracture area and from an area well away from the fracture were examined metallographically and chemically analyzed. Results indicated that the pipe had failed by hydrogen damage that resulted in brittle fracture. Chemical analysis indicated that the pipe material was 1020 carbon steel, not P22. The misapplication of pipe material was considered to be a contributing factor. Position of the pipe within the system caused the localized damage.

A cracked, martensitic stainless steel, low-pressure turbine blade from a 623 MW turbine generator was found to exhibit fatigue cracks during a routine turbine inspection. The blade was cracked at the first notch of the fir tree and the cracks initiated at pits induced by chloride attack. Examination of the blade microstructure at the fracture origins revealed oxide-filled pits and transgranular cracks. The oxide filled cracks appeared to have originated at small surface pits and probably propagated in a fatigue or corrosion-fatigue fracture mode. It was recommended that the sources of the chlorides be eliminated and that the remaining blades be inspected at regular maintenance intervals for evidence of cracking.

A carbon steel (ASTM A515 grade 70) pressure vessel failed by brittle fracture while being hydro tested in the fabricating shop. The fracture origin was a small crack at a welding arc strike associated with the toe of a nozzle weld. A fracture mechanics calculation indicated that this imperfection, although small, initiated fracture because of the local geometry and stress conditions and the low toughness of the steel. It was recommended that (1) the probability of flaws be reduced by welding over or grinding out arc strikes, (2) the local stresses be lowered by post weld stress relief and improved weld toe geometry, and (3) toughness be improved by specifying fine-grain steel and/ or by normalizing.

A white cast iron water-line plug in a fire sprinkler systems split during leak repair. Examination revealed no material flaws, fatigue, or excessive corrosion. The plug head exhibited signs of excessive loads used in attempts to force the plug farther into the pipe. The evidence obtained indicated that the failure resulted from human error.

Field metallography and replication were performed on a type 316 stainless steel column in diglycol amine vacuum service to determine the cause of visible OD pitting on the column in several areas above the insulation support rings. The examination revealed transgranular stress-corrosion cracking beneath the pitted areas on the OD. The likely cause of the cracking was chloride stress corrosion, with chlorides deriving from the marine atmosphere and concentrating under the insulation around the support rings. A complete insulation evaluation, including repair or replacement, was recommended to prevent chloride buildup. Painting of the steel surface with an epoxy-phenolic or epoxy-coal tar was also suggested.