Rotor blades in the compressor section of a J79 engine had failed. Optical, stereoscopic, microhardness testing, and SEM examinations were conducted to determine the cause. The blades were made of STS403 and were used uncoated. They were damaged over an extensive area, from the 15th through the 17th compressor stages, as were stator vanes and casing sections. The fractured surface of the 17th blade showed multiple origins along with secondary cracking and extensive propagation that preceded separation. The metallographic analysis of the microstructure suggested work hardening. Based on the results, the cause of the fractured blade was high-amplitude fatigue due to severe stall. After normal engine usage of five months, the blade fractured sending fragments throughout the combustion and turbine sections.

Copper electrical feedthrough pins used in a bolting application in a refrigeration compressor had functioned without failure for years of production and thousands of units. When some of the pins began to fail, an investigation was conducted to determine the cause. Visual examination revealed that the observed fractures were mixed brittle intergranular with ductile microvoid dimples. An extensive analysis of failed samples combined with a process of elimination indicated that the fractures were due to stress-corrosion cracking caused by an unidentified chemical species within the sealed compressor chamber. A unique combination of applied stress, residual stress, stress riser, and grain size helped isolate the failure mechanism to a single production lot of material.

Initially, two vertical double-acting two-stage compressors delivering chlorine gas at a pressure of 100 psi appeared to be running satisfactorily. About six months later the LP piston-rod of the No. 2 compressor failed due to burning, the compressor being worked double-acting at the time. About five months later, the HP piston rod of the No. 1 compressor failed in a similar manner. Specimens for microscopic examination were cut from the rod in the region of the failure and from the extreme end that had been situated above the piston and hence not subjected to an appreciable rise in temperature. The material was a steel in the normalized condition with a 0.35% C content. It appears probable that deficient lubrication of the gland resulted in overheating of the rod due to friction. The presence of a sprayed-metal coating was probably an additional factor in promoting failure, as it would present to the gas a surface area considerably greater than that of a homogeneous material.

A failed crosshead of an industrial compressor was examined using optical and SEM. The crosshead was an ASTM A148 grade 105-85 steel casting. On the basis of the observations reported and available background information, it was concluded that the failure began with the initiation of cracks at slag inclusions and sharp fillets in weld-repair areas in the casting. The weld-repair procedures were unsatisfactory. The cracks propagated in a fatigue mode. he casting quality was judged unacceptable because of the presence of excessive shrinkage porosity. It was recommended that crosshead castings be properly inspected before machining. Revision of foundry practice to reduce or eliminate porosity was also recommended.

Tube sheets (found to be copper alloy C46400, or naval brass, and 5 cm (2 in.) thick) of an air compressor aftercooler were found to be cracked and leaking approximately 12 to 14 months after they had been retubed. Most of the tube sheets had been retubed several times previously because of unrelated tube failures. Sanitary (chlorinated) well water was generally used in the system, although filtered process make-up water (river water) containing ammonia was occasionally used. Investigation (visual inspection, chemical analysis, mercurous nitrate testing, unetched 5X micrographs, and 250X micrographs etched in 10% ammonium persulfate solution) supported the conclusion that the tube sheets failed by SCC as a result of the combined action of internal stresses and a corrosive environment. The internal stresses had been induced by retubing operations, and the environment had become corrosive when ammonia was introduced into the system by the occasional use of process make-up water. Recommendations included making a standard procedure to stress relieve tube sheets before each retubing operation. The stress relieving should be done by heating at 275 deg C (525 deg F) for 30 min and slowly cooling for 3 h to room temperature.

One 49-in. impeller of a two-stage centrifugal air compressor disrupted without warning, causing extensive damage to the casings, the second impeller, and the driving gear box. Prior to the mishap, the machine had run normally, with no indications of abnormal vibration, temperature, or pressure. Initial failure had taken place in the floating dished inlet plate (eye plate) of the first-stage impeller. Failure occurred predominantly by tearing along the lines of rivet holes for the longer blades, these extended for practically the full radial width of the dished plate. Examination of the fractured surfaces showed that failure had been preceded by fatigue cracking. The material from which the dish plate was forged was a Ni-Cr-Mo steel in the oil hardened and tempered condition. Fractographic examination of the surface of the cracks showed striation markings indicative of the progress of fatigue cracks. Failure of the one impeller and the cracking of the others were attributed to “low-cycle high-strain fatigue” due to fluctuating circumferential (hoop) stresses.

An air compressor was installed at a chemical plant in which nitric acid was produced by burning ammonia with air. It was a 5000 hp, 5-stage centrifugal machine running at 6000 rpm, compressing air to 5 atm. Failure of the first stage impeller occurred due to a segment from the back plate becoming detached. On the remaining portion, cracks were visible running between the holes for rivets by which the vanes were attached. Metallographic examination of selected sections from the backplate revealed the material to be in the hardened and tempered condition, and the cracking to have initiated on the internal surface of the plate at the crevice between the plate and the vane. It was evident that the impeller failed by stress-corrosion cracking, which initiated in the crevice between the vanes and back plate and propagated through the plate along the line of the rivets where working stresses would be greatest. The compressor intake was situated in the vicinity of nitric acid pumps which had a history of leakage troubles, and which had evidently given rise to the nitrates found on the impeller.

During testing of compressors under start/stop conditions, several helical suspension springs failed. The ensuing failure investigation showed that the springs failed due to fatigue. The analysis showed that during start/stop testing the springs would undergo both a lateral and axial deflection, greatly increasing the torsional stresses on the spring. To understand the fatigue limits under these test conditions, a bench test was used to establish the fatigue strength of the springs. The bench tests showed that the failed springs had an unacceptable surface texture that reduced the fatigue life. Based on an understanding of the compressor motion, a Monte Carlo model was developed based on a linear damage theory to predict the fatigue life of the springs during start/stop conditions. The results of this model were compared to actual test data. The model showed that the design was marginal even for springs with acceptable surface texture. The model was then used to predict the fatigue life requirements on the bench test such that the reliability goals for the start/stop testing would be met, thus reducing the risk in qualifying the compressor.

A DC-10 in transit from Denver to Chicago experienced failure of the center engine. The titanium compressor disk burst and severed the hydraulics of the plane. Investigation supports the conclusion that the cause of the disk rupture was the presence of a large fatigue crack near the bore emanating from a hard alpha (HA) defect. Such defects can result from occasional upsets during the vacuum melting of titanium. These nitrogen-rich alpha titanium anomalies are brittle and often have associated microcracks and microvoids. A probabilistic damage tolerance approach was recommended to address the anomalies, with the objective of enhancing rotor life management practices. The ongoing work involves the use of fracture mechanics and software (called DARWIN.) optimized for damage tolerant design and analysis of metallic structural components.

Cylinder fatigue can result from abnormal heating in service. Fatigue can be experienced also by piston heads, exhaust valves, and turbosupercharger housings (castings). Pistons from different engines series can sometimes fit, but because of slight design modifications, they may not function properly. Circumferential cracks and fractures near the head-to- barrel junctions have occurred on numerous cylinders of reciprocating piston engines. In most instances, cracks were caused by high cyclic pressures and high temperatures resulting most probably from detonation. At times, fractures or cracks (or both) were also caused by a combination of unfavorable temperature distribution (and possibly excessive pressures around the cylinder barrel), un-nitrided internal surfaces of cylinder barrels, and inadequate thread contours, which caused high stress concentrations at the thread roots. One example of the most common type of cylinder failure is illustrated.

Shaft fracture of a 10 hp squirrel cage motor took place at the driving end just outside the roller bearing and not at an abrupt change of section behind the bearing where it might be expected to occur. A portion of shaft to the right of the fracture was deeply grooved. About a year prior to failure the inner race of the roller bearing became slack on the shaft and the seating was built up by the metal-spray process. The shaft was machined to form a rough thread to provide the requisite mechanical key for the sprayed-on metal. Part of this sprayed-on layer became detached after the fatigue failure occurred. The quality of the welding was poor. Slag inclusions were present adjacent to the sides of the keyway, which had been re-cut shorter than the original one after the welding repair. Failure at the unusual location was caused by the presence of the weld deposit.

The shaft-and-bearing assembly in a freon compressor was subjected to severe pounding and vibration after six years of service. After about one year of service, the compressor had been shut down to replace a bearing seal. One month before the shaft failed, a second seal failure occurred, requiring the collar, spacer sleeve, seal, roller bearing, and lock washer to be replaced. The shaft was made of 4140 steel, heat treated to a hardness of 20 to 26 HRC. The seal, bearing, and lock washer were commercial components. Investigation (visual inspection, 4.5x images, x-ray diffraction, hardness testing, and microscopic exam) supported the conclusion that shaft failure was initiated by fretting between the bearing race and the bearing surface on the shaft because of improper bearing installation. Once clearance was established between the bearing and the shaft, the shaft began pounding on the inner bearing race, causing final failure of the shaft surface. Recommendations included proper fitting of the shaft and bearing race to preventing movement of the bearing on the shaft. Also, the lock washer and locknut must be installed properly.

High-horsepower electric motors were utilized to drive large compressors (made of 4340 steel shafts and gear-type couplings) required in a manufacturing process. The load was transmitted by two keys 180 deg apart. Six of the eight compressor shafts were found cracked in a keyway and one of them fractured after a few months of operation. Visual examination of fractured shaft revealed that the cracks originated from one of the keyways and propagated circumferentially around the shaft. The shaft and coupling slippage was indicated by the upset keys and this type of fracture. The shaft surface both near and in the keyways indicated fretting which greatly reduced the fatigue limit of the shaft metal and initiated fatigue cracks. Fatigue marks were observed on the fractured key. Repetitive impact loading was responsible for propagation of the cracks. The high cyclic bending stresses were caused by misalignment between the electric motor and compressor and were transmitted to the shaft through the geared coupling. Flexible-disk couplings capable of transmitting the required horsepower were installed on the shafts as a corrective measure.

The crankshaft of a compressor fractured through the web remote from the driving end after about three years of service. The fracture ran diagonally across the web into the crankpin. It passed through the centers of two screwed plugs inserted into the web from opposite faces approximately in line with the crankpin center line. The fracture was of the fatigue type, slowly developing cracks having started from opposite sides of each tapped hole and crept across the section. Microstructure of the crankshaft indicated the material was a plain carbon steel, the carbon content being of the order of 0.3%. The failure resulted principally from the stress-raising effects of the screw holes combined with the cracks in the welds. If the screw holes had been left unfilled or if some form of mechanical locking had been used if plugged, failure would have been postponed if not averted.

After about 17 years in service, copper alloy C27000 (yellow brass, 65% Cu) innercooler tubes in an air compressor began leaking cooling water, causing failure and requiring replacement. The tubes were 19 mm in diam and had a wall thickness of 1.3 mm (0.050 in.). The cooling water that flowed through the tubes was generally sanitary (chlorinated) well water; however, treated recirculating water was sometimes used. Analysis (visual inspection, 9x and 75x unetched micrographs, and spectrochemical analysis) showed a thick uniform layer of porous, brittle copper on the inner surface of the tube, extending to a depth of about 0.25 mm (0.010 in.) into the metal, plug-type dezincification extending somewhat deeper into the metal. This supported the conclusion that failure of the tubes was the result of the use of an uninhibited brass that has a high zinc content and therefore is readily susceptible to dezincification. Recommendations included replacing the material with copper alloy C68700 (arsenical aluminum brass), which contains 0.02 to 0.06% As and is highly resistant to dezincification. Copper alloy C44300 (inhibited admiralty metal) could be an alternative selection for this application; however, this alloy is not as resistant to impingement attack as copper alloy C68700.

An aftercooler was of conventional design and fitted with brass tubes through which cooling-water circulated. Air at 100 psi pressure was passed over the outsides of the tubes, entering the vessel near to the upper tubeplate on one side and leaving it by a branch adjacent to the lower tubeplate on the opposite side. After a mishap, the paint had been burned off the upper half of the shell. Internally, most of the tubes were found to be twisted or bent. The casing of the pump used to circulate the cooling water was also found to be cracked after the mishap. All the evidence pointed to the probability that a fire had occurred within the vessel. Some months before the failure, one of the tubes situated towards the center of the nest developed a leak. Owing to the difficulty of inserting a replacement tube, the defective one was scaled by means of a length of screwed rod fitted with nuts and washers at each end. This assembly became loose, thereby allowing air under pressure to enter the waterside of the cooler and expel the water, leading to overheating and ultimately to the damage described.

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.

Macrofractographs of the fracture surface from a multibladed fan showed that cracks started at the corner where bending stress was concentrated and propagated through the blade by fatigue. Peak stress at the monitoring position was less than 10 MPa. To simulate crack growth, the rotor was repeatedly deformed by a hydraulic fatigue tester. Comparison of striations of the failed blade with that of the tested one revealed the failed blade was loaded with more than 30 MPa of stress. These tests confirmed that the rotor and blades had sufficient strength to withstand up to 3x the stress of normal operation. The casing of the fan was vibrated at 10 to 60 Hz. Peak stress easily overcame 30 MPa, which was enough to initiate cracking. The fracture surfaces and starting position were the same as those on the failed fan. It was concluded that the exciting force from an air compressor caused blade failure.

An A-470 steel rotor disk was removed from the high-pressure portion of a steam turbine-powered compressor after nondestructive testing revealed cracks in the shoulder of the disk during a scheduled outage. Samples containing cracks were examined using various methods. Multiple cracks, primarily intergranular were found on the inlet and outlet faces along prior-austenite grain boundaries. The cracks initiated at the surface and propagated inward. Multiple crack branching was observed. Many of the cracks were filled with iron oxide. X-ray photoelectron spectroscopy indicated the presence of sodium on crack surfaces, which is indicative of NaOH-induced stress-corrosion cracking. Failure was attributed to superheater problems that resulted in caustic carryover from the boiler. Two options for disk repair, installing a shrink-fit disk or applying weld buildup, were recommended. Weld repair was chosen, and the rotor was returned to service; it has performed for more than 1 year without further incident.

Several compressor diaphragms from five gas turbines cracked after a short time in service. The vanes were constructed of type 403 stainless steel, and welding was performed using type 309L austenitic stainless steel filler metal. The fractures originated in the weld heat-affected zones of inner and outer shrouds. A complete metallurgical analysis was conducted to determine the cause of failure. It was concluded that the diaphragms had failed by fatigue. Analysis suggests that the welds contained high residual stresses and had not been properly stress relieved. Improper welding techniques may have also contributed to the failures. Use of proper welding techniques, including appropriate prewelding and postwelding heat treatments, was recommended.