The connecting end of two forged medium-carbon steel rods used in an application in which they were subjected to severe low-frequency loading failed in service. The fractures extended completely through the connecting end. The surface hardness of the rods was found to be lower than specifications. The fractures were revealed to be in areas of the transition regions that had been rough ground to remove flash along the parting line. The presence of beach marks, indicating fatigue failure, was revealed by examination. The fracture origin was confirmed by the location and curvature of beach marks to be the rough ground surface. An incipient crack 9.5 mm along with several other cracks on one of the fractured rods was revealed by liquid penetration examination. Metallographic examination of the fractured rods indicated a banded structure consisting of zones of ferrite and pearlite. It was established that the incipient cracks found in liquid-penetrant inspection had originated at the surface in the banded region, in areas of ferrite where this constituent had been visibly deformed by grinding. Closer control on the microstructure, hardness of the forgings and smooth finish in critical area was recommended.

Two outboard main-wheel halves (aluminum alloy 2014-T6 forged) from a commercial aircraft were removed from service because of failure. One wheel half was in service for 54 days and had made 130 landings (about 1046 roll km, or 650 roll mi) when crack indications were discovered during eddy-current testing. The flange on the second wheel half failed after only 31 landings, when about 46 cm (18 in.) of the flange broke off as the aircraft was taxiing. Stains on the fracture surfaces were used to determine when cracking was initiated. The analysis (visual inspection, liquid penetrant inspection, and micrographs with deep etching in aqueous 20% sodium hydroxide) supported the conclusion that failure on both wheel halves was by fatigue caused by a forging defect resulting from abnormal transverse grain flow. The crack in the first wheel half occurred during service, and the surfaces became oxidized. Because the fracture surface of the second wheel half had chromic acid stains, it was obvious that the forging defect was open to the surface during anodizing. No recommendations were made except to notify the manufacturer.

A 680,000 kg (750 ton) per day ammonia unit was shut down following a fire near the outlet of the waste heat exchanger. The fire had resulted from leakage of ammonia from the type 316 stainless steel outlet piping. The outlet piping immediately downstream from the waste heat exchanger consisted of a flange made from a casting, and a reducing cone, a short length of pipe, and a 90 deg elbow, all made of 13 mm thick plate. A liner wrapped with insulation was welded to the smaller end of the reducing cone. All of the piping up to the flange was wrapped with insulation. Investigation (visual inspection, 10x unetched images, liquid-penetrant inspection, and chemical analysis of the insulation) supported the conclusion that the failure occurred in the area of the flange-to-cone weld by SCC as the result of aqueous chlorides leached from the insulation around the liner by condensate. Recommendations included eliminating the chlorides from the system, maintaining the temperature of the outlet stream above the dewpoint at all times, or that replacing the type 316 stainless steel with an alloy such as Incoloy 800 that is more resistant to chloride attack.

While undergoing vibration testing, a type 347 stainless steel inlet header for a fuel-to-air heat exchanger cracked in the header tube adjacent to the weld bead between the tube and header duct. Investigation (visual inspection and liquid penetrant inspection) supported the conclusion that the crack in the header tube was the result of a stress concentration at the toe of the weld joining a doubler collar to the tube. The stress concentration was caused by undercutting from poor welding technique and an unfavorable joint design that did not permit a good fit-up. Recommendations included manufacturing the doubler collar so that it could be placed in intimate contact with the header duct, and a revised weld procedure was recommended to result in a smaller, controlled, homogeneous weld joint with less distortion.

A type 321 stainless steel bellows expansion joint on a 17-cm (6 in.) OD inlet line (347 stainless) in a gas-turbine test facility cracked during operation. The line carried high-purity nitrogen gas at 1034 kPa (150 psi) with a flow rate of 5.4 to 8.2 kg/s (12 to 18 lb/s). Cracking occurred in welded joints and in unwelded portions of the bellows. The bellows were made by forming the convolution halves from stainless steel sheet, then welding the convolutions together. Evidence from visual examination, liquid penetrant inspection chemical analysis, hardness tests, and metallographic examination of sections etched with Vilella's reagent supports the conclusions that failure of the bellows occurred by intergranular fatigue cracking. Secondary degrading effects on the piping existed as well. Recommendations included the acceptability of Type 321 stainless steel (provided open-cycle testing does not result in surface oxidation and crevices) Although type 347 stainless steel would be better, and Inconel 600 would be an even better choice. Welds would also need modified processing for reheating and annealing. Prevention of oil leakage into the system would minimize carburization of the piping and bellows.

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 forged aluminum alloy 2014-T6 catapult-hook attachment fitting (anodized by the chromic acid process to protect it from corrosion) from a naval aircraft broke in service. Spectrographic analysis, visual examination, microscopic examination, and tensile analysis showed minute cracks on the inside surface of a bearing hole, and small areas of pitting corrosion were visible on the exterior surface of the fitting. The analysis also revealed a small number of rosettes, suggestive of eutectic melting, in an otherwise normal structure. These examinations and analyses support the conclusion that the presence of chromic acid stain on the fracture surface proved that the forging had cracked before anodizing. This suggest that the crack initiated during straightening, either after machining or after heat treatment. The structure and composition of the alloy appear to have been acceptable. Ductility was acceptable so rosettes found in the microstructure are believed to have been nondamaging. Had they contributed to the failure, the ductility would have been very low. The recommendations included inspection for cracks and revising the manufacturing process to include a fluorescent liquid-penetrant inspection before anodizing, because chromic acid destroys the penetrant. This inspection would reduce the possibility of cracked parts being used in service.

Nondestructive testing (NDT), also known as nondestructive evaluation (NDE), includes various techniques to characterize materials without damage. This article focuses on the typical NDE techniques that may be considered when conducting a failure investigation. The article begins with discussion about the concept of the probability of detection (POD), on which the statistical reliability of crack detection is based. The coverage includes the various methods of surface inspection, including visual-examination tools, scanning technology in dimensional metrology, and the common methods of detecting surface discontinuities by magnetic-particle inspection, liquid penetrant inspection, and eddy-current testing. The major NDE methods for internal (volumetric) inspection in failure analysis also are described.

Two suction rolls at the first press section of a 25 ft. wide paper machine developed cracks within two years of service. The rolls were austenitic stainless steel castings made of ASTM A 351 Grade CF8M alloy containing molybdenum. The rolls were exposed to slightly acidic white water (pH approximately 4.7) containing chlorides (45 ppm). Visual and liquid penetrant inspections of the rolls revealed extensive cracking at the roll inside surface. The cracks penetrated more than 30 percent of the wall thickness and a few cracks were several inches long. The cracks were preferentially oriented along the roll length and primarily at the roll inside surface. Field metallographic examination showed significant grain boundary chromium-carbide precipitation and intergranular corrosion. The roll failures were attributed to chromium depletion along the grain boundaries (sensitization) resulting from slow cooling of the casting to avoid large residual stresses. The roll manufacturer recommended a proprietary ferritic/austenitic stainless steel as the replacement material for the rolls.

Liquid penetrant inspection of an ASTM A296 grade CA-15 residual heat removal pump impeller from a nuclear plant revealed a crack like indication that approximated the outer contour of the wear ring. Examination of a section containing the crack and three sections from near the main crack indication revealed that the failure was caused by hot cracking related to original weld repairs performed on the impeller casting.

The Pandia digester is a long cylindrical vessel which uses alkaline sulfite liquor to cook sawdust for pulping. The inlet cone was fabricated from AISI 304L stainless steel with E308 welds. Typical liquor concentration was approximately 80% NaOH, 20% Na2SO3 with chloride concentrations at 2 grams per liter. The operating pressures in the inlet cone were up to 1.2 MN/sq m (170 psig). The inlet cone had developed leaks within a year of service. Liquid penetrant inspection showed significant through-wall cracking near the fillet welds joining the bottom flange and side wall and the butt welds. Metallographic specimens were prepared from the welds to examine the microstructure and nature of the cracks. The cooking liquor at the inlet cone contained over ppm chlorides and was aggressive to 304 stainless steel. The cracking was identified as chloride-induced SCC. The inlet cone was replaced with an Inconel clad carbon steel inlet cone to combat the SCC.

A stainless steel pipe transferring hot white liquor solution of sodium hydroxide and sodium sulfite, developed leaks adjacent to the welds within four years of service. The stainless steel pipe was AISI type 304 and welded with E308 weld electrodes. The service temperature was 190 deg C (375 deg F) and the solution contained approximately 700 ppm chlorides. Liquid penetrant inspection of the pipeline showed the leaks were numerous and confined adjacent to the welds. A metallographic specimen from the circumferential weld showed the cracks initiated at the inside surface. In addition to the base metal, SCC also had initiated at a notch at the weld root due to improper welding procedures. Failure was attributed to chloride-induced SCC with secondary contributory factors, including improper welding procedures. It was recommended that the pipeline be replaced with a material more resistant to SCC. The candidate materials are commercial grade unalloyed titanium or Inconel 600, which have superior resistance to SCC compared to austenitic stainless steels.

A fluorescent liquid-penetrant inspection of an experimental stator vane of a first-stage axial compressor revealed the presence of a longitudinal crack over 50 mm (2 in.) long at the edge of a resistance seam weld. The vane was made of titanium alloy Ti-6Al-4V (AMS 4911). The crack was opened by fracturing the vane. The crack surface displayed fatigue beach marks emanating from the seam-weld interface. Both the leading-edge and trailing-edge seam welds exhibited weld-metal expulsions up to 3.6 mm (0.14 in.) in length. Metallographic examination confirmed that metal expulsion from the resistance welds was generally present. The stator vane failed by a fatigue crack that initiated at internal surface discontinuities caused by metal expulsion from the resistance seam weld used in fabricating the vane. Expulsion of metal from seam welds should be eliminated by a slight reduction in welding current to reduce the temperature, by an increase in the electrode force, or both.

A fuel-nozzle-support assembly showed transverse indications after fluorescent liquid-penetrant inspection of a repair-welded area at a fillet on the front side of the support neck adjacent to the mounting flange. Visual examination disclosed an irregular crack. The crack through the neck was sectioned; examination showed that the crack had extended through the repair weld. The crack had followed an intergranular path. The crack was opened, and binocular-microscope examination of the fracture surface showed that the surface contained dendrites with discolored oxide films that were typical of exposure to air when very hot. Several additional subsurface cracks, typical of hot tears, were observed in and near the weld. There had been too much local heat input in making the repair weld. The result was localized thermal contraction and hot tearing. The cracking of the repair weld was attributed to unfavorable welding practice that accentuated thermal contraction stresses and caused hot tearing. Recommendations involved use of a small-diameter welding electrode, a lower heat input, and deposition in shallow layers that could be effectively peened between passes to minimize internal stress.

An air bottle, machined from a solid block of aluminum alloy 2219-T852, displayed liquid-penetrant crack indications after assembly welding. The air bottle was machined to rough shape, a 3.8 mm (0.15 in.) wall thickness cylindrical cup with a 19 mm (3/4 in.) wall thickness integral boss on one side. After annealing, hot spinning, annealing a second time, and tack welding a port fitting, the assembly was torch preheated to 120 to 150 deg C (250 to 300 deg F). The port fitting was then welded in place. Final full heat treatment to the T62 temper was followed by machining, testing, and inspection. The crack indications were found only on one side of the boss and on the lower portion of the hot-spun dome region. The metallographic specimens revealed triangular voids and severe intergranular cracks. The cracks displayed the glossy surfaces typical of melted and resolidified material. The localized cracks in the air bottle were from grain-boundary eutectic melting caused by local torch overheating used in preparation for assembly welding of a port fitting. A change in design was scheduled to semiautomatic welding without the use of preheating for the joining of the port fitting for the dome opening.

A missile detached from a Navy fighter jet during a routine landing on an aircraft carrier deck because of a faulty missile launcher detent spring. Visual inspection of Inconel 718 detent spring assembly revealed that four of the nine spring leafs comprising the assembly were plastically deformed while two of the deformed leafs did not meet minimal hardness or tensile requirements. Liquid penetrant testing revealed no cracks or other surface discontinuities on the leaf springs. Material sectioned from the soft spring leafs was heat-treated according to specifications in the laboratory. The resultant increase in mechanical properties of the re-heat-treated material indicated that the original heat treatment was not performed correctly. The failure was attributed to improper heat treatment. Recommendations focused on more stringent quality control of the heat-treat operations.

Eddy current inspection was performed on a vertical evaporator unit (that contained 180 tubes) used in a chemical processing plant. It was advised that the tube material was type 316 stainless steel. The shell-side fluid was condensate and gaseous methylene chloride, while the tube-side fluid was contaminated liquid methylene chloride. More than 100 tubes exhibiting severe outer surface pitting and cracklike indications near each tube sheet were revealed during eddy current inspection. It was observed that the indications correlated with rust-stained, pitted, and cracked areas on the outer surfaces. The cracking was revealed by metallographic examination to have initiated from the outer surface, frequently at pits, and penetrated the tube wall in a transgranular, branching fashion. The crack features were characteristic of chloride stress-corrosion cracking. A change in tube material was recommended to avoid future failures and loss of service.

Several compressor disks in military fighter and trainer aircraft gas turbine engines cracked prematurely in the bolt hole regions. The disks were made of precipitation-hardened AM355 martensitic stainless steel. Experimental and analytical work was performed on specimens from the fifth-stage compressor disk (judged to be the most crack-prone disk in the compressor) to determine the cause of the failures. Failure was attributed to high-strain low-cycle fatigue during service. It was also determined that the cyclic engine usage assumed in the original life calculations had been under estimated, which led to low-cycle fatigue cracking earlier than expected. Fracture mechanics analysis of the disks was carried out to assess their damage tolerance and to predict safe inspection intervals.

Parts of 21Cr-6Ni-9Mn stainless steel that had been forged at about 815 deg C (1500 deg F) were gas tungsten arc welded. During postweld inspection, cracks were found in the HAZs of the welds. Welding had been done using a copper fixture that contacted the steel in the area of the HAZ on each side of the weld but did not extend under the tungsten arc. In SEM examination, the cracks appeared to be intergranular and extended to a depth of approximately 1.3 mm (0.05 in.). The crack appearance suggested that the surface temperature of the HAZ could have melted a film of copper on the fixture surface and that this could have penetrated the stainless steel in the presence of tensile thermal-contraction stresses. The cracks in the weldments were a form of liquid-metal embrittlement caused by contact with superficially melted copper from the fixture and subsequent grain-boundary attack of the stainless steel in an area under residual tensile stress. The copper for the fixtures was replaced by aluminum. No further cracking was encountered.

One of the coils in the radiant section of a primary reformer furnace used in an ammonia plant was found leaking. The bottom of one of seven outlet headers (made of ASME SA-452, grade TP316H, stainless steel) was revealed during examination to be ruptured. It was revealed by metallurgical examination that it had failed as a result of intergranular fissuring and oxidation (creep rupture). The ruptured area revealed that the header had failed by conventional long-time creep rupture as a result of exposure to operating temperatures probably between 900 and 955 deg C. Three samples from different sections (ruptured area, slightly bulged but nonruptured area and visually sound metal) were inspected. The presence of pinhead-size intergranular fissures throughout the cross sections of the latter two samples was observed. An ultrasonic attenuation method was employed to investigate the remaining headers. All headers were revealed by ultrasonic readings to be in an advanced stage of creep rupture and no areas were found to be fissured to a degree that they needed immediate replacement. As a conclusion, the furnace was deemed serviceable and it was established that in the absence of local hot spots, the headers would survive for a reasonable period of time.