A large diameter steel pipe reinforced by stiffening rings with saddle supports was subjected to thermal cycling as the system was started up, operated, and shut down. The pipe functioned as an emission control exhaust duct from a furnace and was designed originally using lengths of rolled and welded COR-TEN steel plate butt welded together on site. The pipe sustained local buckling and cracking, then fractured during the first five months of operation. Failure was due to low cycle fatigue and fast fracture caused by differential thermal expansion stresses. Thermal lag between the stiffening rings welded to the outside of the pipe and the pipe wall itself resulted in large radial and axial thermal stresses at the welds. Redundant tied down saddle supports in each segment of pipe between expansion joints restrained pipe arching due to circumferential temperature variations, producing large axial thermal bending stresses. Thermal cycling of the system initiated fatigue cracks at the stiffener rings. When the critical crack size was reached, fast fracture occurred. The system was redesigned by eliminating the redundant restraints and by modifying the stiffener rings to permit free radial thermal breathing of the pipe.

A femoral knee implant was returned to the casting vendor for analysis after exhibiting poor bond strength between the cast substrate and a sintered porous coating. Both the coating and the substrate were manufactured from a cobalt-chromium-molybdenum alloy. Metallographic analysis indicated that a decarburized layer existed on all surfaces of the casting, which prevented bonding during the sintering thermal cycle. Bead-to-bead bonding within the coating appeared sufficient, and no decarburized layer was present on the bead surfaces. It was concluded that the decarburization did not occur during the sintering thermal cycle. It was recommended that the prosthetic manufacturer investigate atmosphere controls for all thermal cycles prior to coating.

An open electrical circuit was found between plated through-holes in a six-layer printed circuit board after thermal cycling. The copper plating was very thin in the failure area but did make an electrical contact during initial testing. During thermal cycling, differential z-expansion between the epoxy board and copper caused the thin plating to crack. During electrical testing of a four-layer circuit board, an open electrical circuit was found between the plated through-holes. Plating discontinuity was caused by poor drilling using a dull drill with improper speed (rpm) and/or feed rate as was observed by nonuniform plating and nodule formation in the plated layer. In a third example, an open electrical circuit was found in a six-layer board between two adjacent plated through-holes. A plating void was on one side of the conductor joining the two holes. Continuity was found when tested from one side of the board but lost when tested from the other. In a fourth case, an open circuit found between a plated through-hole and contact pad on a six-layer printed circuit board was caused by an etching defect.

Examination of the header of the third superheater of a boiler producing 150 t/h of steam at 525 deg C and 118 kPa, disclosed extensive internal cracking at the connection to the tube joining this to a safety valve. Cracking was observed within the tube and in the thickness of the shell wall itself. The boiler had been in operation for approximately 160,000 h and was shut down for inspection when the cracking was detected. The material involved was 2.25 Cr, 1 Mo steel, and the unit had been subjected to 115 shutdowns. Initiation of the cracks was attributed to thermal shock, caused by the periodic return of condensate along the long connecting line (some 9 m long). Propagation of the cracks was due to thermal cycling, together with periodic pressure cycles, producing growth by low cycle fatigue. This was aided by corrosion within the cracks and by the wedging action caused by corrosion deposits at their tips. The failure suggests control of dissolved solids in the boiler feedwater may have been inadequate.

A return bend (made from ASTM A213, grade T11, ferritic steel) from a triolefin-unit heater ruptured after two years in service. The unit operated at 2410 kPa, with a hydrocarbon feed stream (85% propylene) entering at 260 to 290 deg C and leaving at 425 to 480 deg C. The fracture was found to terminate at the welds that joined the bend to the pipeline. A high concentration of both small and large inclusions was exhibited by the metallographic examination of the steel near the fracture. Branched cracks similar to those produced by stress corrosion of steel were observed in a section through the fireside edge of the fracture surface. Scale was observed over most of the crack path which acted as a stress raiser. The effect of the oxide was magnified during thermal cycles because of differential thermal expansion, with the steel having a greater expansion coefficient than the scale. It was recommended that material that is intended for critical applications where failure cannot be tolerated should be non-destructively examined.

A root cause failure analysis was performed on a vaporizer coil removed from a horizontal forced circulation vaporizer. The carbon steel coil was wound in a right-hand helix with a coil centerline diameter of about 2 m. The vaporizer was gas fired and used Dowtherm A as the heat transfer fluid. Design conditions are based on annular fluid flow to cool the coil wall. NDE, metallographic and fractographic examinations were performed. Numerous, circumferentially oriented, OD initiating cracks were found near the crown for two coils near the non-fired end of the vaporizer. The cracking was confined to the inner diameter of the vaporizer coil at positions from 4:00 to 7:00. The cracking was characterized as transgranular and the fracture surface had beach marks. The failure mechanism was thermal fatigue. The heat transfer calculation predicted that dryout of the coil would occur for coils at the non-fired end of the vaporizer during low flow transients. Dryout results in rapid increase in the tube wall temperature. Thermal cycling of the coil is completed by liquid quenching resulting from resumption of normal flow rates and the return to annular flow. The probable root cause of failure was low flow transient operation.

After some 87,000 h of operation, failure took place in the bend of a steam pipe connecting a coil of the third superheater of a steam generator to the outlet steam collector. The unit operated at 538 deg C and 135 kPa, producing 400 t/h of steam. The 2.25Cr-1Mo steel pipe in which failure took place was 50.8 mm in diam with a nominal wall thickness of 8 mm. It connected to the AISI 321 superheater tube by means of a butt weld and was one of 46 such parallel connecting tubes. The Cr-Mo tubing was situated outside the heat transfer zone of the superheater. The overall sequence of failure involved overheating of the Cr-Mo outlet tubes, heavy oxidation, oxide cracking on thermal cycling, thermal fatigue cracking plus oxidation, creep-controlled crack growth, and rapid plastic deformation and rupture. This failure was indicative of excess temperature of the steam coming from the heat transfer zone of the coil. It showed that many damage mechanisms may combine in the transition from fracture initiation to final failure. The presence of grain boundary sliding as an indication of creep damage was useful in the characterization of the stress level as high and showed that the process of creep was not operative throughout the life of the equipment.

An oil scavenge pump was found to have failed when a protective shear neck fractured during the start of a jet engine. Visual inspection revealed that the driven gear in one of the bearing compartments was frozen as was the corresponding drive gear. Spacer wear and thermal discoloration (particularly on the driven gear) were also observed. The gears were made from 32Cr-Mo-V13 steel, hardened and nitrided to 750 to 950 HV. Micrographic inspection of the gear teeth revealed microstructural changes that, in context, appear to be the result of friction heating. The spacers consist of Cu alloy (AMS4845) bushings force fit into AA2024-T3 Al alloy spacing elements. It was found that uncontrolled fit interference between the two components had led to Cu alloy overstress. Thermal cycling under operating conditions yielded the material. The dilation was directed inward to the shaft, however, because the bushing had only a few microns of clearance. The effect caused the oil to squeeze out, resulting in metal-to-metal contact, and ultimately failure.

Corrosion failure occurred in a titanium clad tubesheet because of a corrosive tube-side gas-liquid mixture leaking through fatigue cracks in the seal welds at tube-to-tubesheet joints. The tubesheet was a carbon steel plate clad with titanium on the tube side face. The seal weld cracks were initiated by cyclic stress imposed by exchanger tubes. The gas-liquid mixture passed through cracks under tube-side pressure, resulting in severe corrosion of the steel backing plate. The failure started with the loosening of the expanded tube-to-tubesheet joints. Loose joints allowed the exchanger tubes to impose load on seal welds and the shell side cooling water entered the crevice between the tubesheet and the tubes. The cooling water in the crevice caused galvanic reaction and embrittlement of seal welds. Brittle crack opening and crack propagation in seal welds occurred due to the cyclic stress imposed by the tubes. The cyclic stress arised from the thermal cycling of the heat exchanger. The possible effects of material properties on the failure of the tubesheet are discussed.

Grate bars in the traveling grate indurators in several taconite processing units suffered excessive corrosion following a conversion from acid to fluxed pellet production procedures. The campaign life of the HH grade cast stainless steel bars was reduced from more than 7 years to approximately 9 months. Several corroded grate bars were examined metallographically and by electron microscopy to determine the causes of the accelerated corrosion. Chemical and X-ray diffraction analyses were also conducted, along with simulation tests to assess the role of alkali chlorides in the corrosion process. The basic cause of degradation was found to be hot corrosion caused by the deposition of alkali sulfates and chlorides. However this degradation may have been aggravated by thermal cycling and abrasion. The source of the salt was impurities in the flux. Two potential solutions were proposed: modification of the processing parameters to reduce the salt deposition and / or change of bar materials to a more resistant alloy.

Several failures occurred in 64-mm schedule 80 type 304 stainless steel (ASME SA-312, grade TP304) piping in a steam-plant heat-exchanger system near tee fittings at which cool water returning from the heat exchanger was combined with hot water from a bypass. Various portions of the piping were subjected to temperatures ranging from 29 to 288 deg C. Each of the failures were revealed to consist of transgranular cracking in and/or close to the circumferential butt weld joining the tee fitting to the downstream pipe leg, where the hot bypass water mixed with the cool return water. The transgranular cracks suggested that thermal fatigue was a more likely cause of failure than SCC. It was concluded by temperature measurements that circumferential temperature gradients, in combination with inadequate flexibility in the piping system as a whole, had caused the failures. The tee fitting was redesigned to alleviate the thermal stress pattern.

Type 347 stainless steel moderator circuit branch piping in a pressurized hot water reactor was experiencing frequent leakage. Investigation of the problem involved failure analysis of leaking pipe specimens, analytical stress analysis, and determination of “leak-before-break” conditions using fracture mechanics and thermal fatigue simulation tests. Failure analysis indicated that cracking had been initiated by thermal fatigue. Data from the analysis were used in making the leak-before-break predictions. It was determined that the cracks could grow to two-thirds of the circumferential length of the pipe without catastrophic failure. A thin stainless steel sleeve was inserted in the branch pipe to resolve the problem.