NTSB Factual Report

On February 13, 2003, about 2000 eastern standard time, a Cessna 500, N891CA, registered to Mike's Aviation Inc., and operated by Gulf Atlantic Airways, as a Title 14 CFR Part 91 positioning flight, had an uncontained failure of the right engine during the takeoff roll/run, at Fort Lauderdale, Florida. Visual meteorological conditions prevailed, and an instrument flight rules flight plan was filed. The airline transport-rated pilot, and first officer were not injured, and the airplane incurred minor damage. The flight was originating at the time of the incident.

The pilot stated that after refueling, he taxied out to runway 09 at Fort Lauderdale-Hollywood International Airport, Fort Lauderdale, Florida, for a positioning flight to Miami, Florida. He further stated that at the beginning of the takeoff roll, as the engine was spooling up, he heard a bang, and saw the fan fly past the right window. He aborted the takeoff, exited the runway, and shut down the airplane.

Initial postincident investigation revealed that an uncontained failure had occurred to the No. 2 engine, and there was also some damage to the engine pylon, the top of the wing, and the right side of the fuselage. Specifically, the Pratt &Whitney Canada (PWC) JT15D No. 2 turbofan engine, had experienced an uncontained rupture of the high-pressure compressor (HPC), with it having had a penetration of the engine case in the plane of the HPC impeller. There was a large hole in the top of the engine cowling at approximately 1 o'clock aft looking forward, just forward of the thrust reverser clamshell. The missing segment of the HPC impeller was not located.

The engine was removed and shipped to Pratt &Whitney Canada's Bridgeport, West Virginia, facility for teardown and inspection.

On March 11-12, 2003, the No. 2 engine was examined, and the examination revealed that the fan along with a portion of the low-pressure compressor shaft had been liberated during the event. All of the fan blades were in place and intact. Each of the fan blade tips had curled on their corners in the opposite direction of rotation. The fan exit guide vanes were intact and undamaged with no visible evidence of any foreign object damage. The fan case was intact. The No. 1 bearing housing was intact, and the air seal had fractured at the attaching flange. The low-pressure compressor shaft had separated just aft of the gear teeth for the tachometer drive, and the No. 1 ball bearing was intact on the shaft, but had separated from the outer raceway.

The intermediate case had multiple casting fractures. The integral oil tank had multiple casting fractures. The case had completely separated from the gas generator case. The support struts also exhibited multiple fractures.

The gas generator case assembly was penetrated at approximately the 1 o'clock position with the hole measuring 18 inches in length by 6 inches deep. The gas generator assembly and the hot section components were removed as one assembly. The fractured high-pressure compressor impeller was loose inside the case and was lifted out by hand. The high-pressure compressor stub-shaft had fractured from the impeller and was found inside the outer bypass duct. The high-pressure turbine shaft had separated from the impeller attachment flange and was still lodged in the gas generator case.

The combustion chamber and low-pressure turbine support case were not accessed but a visual examination did not reveal any obvious signs of distress. The high-pressure turbine disc and blades had rubbed on the downstream side of the disc face at the firtree location. Five of the high-pressure turbine blades had partial fractures of their tips. The blade shroud segments were intact with signs of light rubbing. The high-pressure turbine shaft was still properly engaged to the high-pressure turbine disc via the cup washer and nut. The front flange had torn away, in the opposite direction of rotation, from the high-pressure compressor impeller.

The low-pressure turbine assembly was removed and was not disassembled. It was still connected via the hollow bolt. The leading edges of the first-stage blades showed evidence of light metal splatter. The downstream side of the second-stage blades did not reveal any visible damage.

The high-pressure compressor impeller had separated from both the high-pressure compressor and high-pressure turbine shafts. A segment of the impeller was liberated and never recovered. The missing segment was approximately the length of 6 full vane airfoils. There were 7-8 full vanes that were missing material down to the hub. There was hard body impact damage to the leading edges of all of the vanes at the inducer and exducer locations. There was peening and a smear of material adjacent to the bore of the aft face. Of the approximately 230-degrees of the remaining portion of the impeller, there was circumferential scoring and rubbing through 60-degrees. There was a score/scratch on the aft face of the impeller in area "K" that was approximately 0.60 inches inboard from the balance flange. The fracture surface on the impeller also coincides with a portion of the scratch/score. There were 2-3 vanes, on each side of where the fracture surface coincided with the scratch/score, which had symmetrical, mirror image shear lips

At low magnification, the fracture surface on the HPC impeller presented two smooth semicircular areas centered on the back face. At higher magnification, small river lines were found emanating from the back face. Away from the back face there were shear lips around the fracture surface. Closer examination of the back face showed the crack initiated at the circumferential scratch/score. This feature appears to be a step on the back face. This step was measured at several locations along the circumference using a Taylor Hobson profilometer with a probe radius of 0.00008 inches. The profile next to the fracture showed a deep outer groove (1,848 µin), located approximately 0.68 inches inboard of the balance rim along with a shallower inner groove. Moving away from the fracture surface, the inner groove deepens while the outer groove becomes shallower. At a position diametrically opposed to the fracture, only the inner groove remains. The crack initiated from the outer groove at its deepest location.

The part was ultrasonically cleaned per CPOP 7308 for 15-20 minutes to remove oil prior to performing the Florescent Penetrant Inspection, and the florescent penetrant inapection showed that the crack extended on both sides of the fracture surface and covered an arc of approximately 180 degrees.

A sample containing the fracture surface and both crack initiation sites was cut from the impeller using electro-discharge machining (wire EDM) for further fractographic examination. At higher magnification and using low angle incident lighting, the fracture surface exhibited several small river lines emanating from the surface at the back face. A similar area was observed at a second crack initiation site. There was not a unique crack initiation site as the crack initiated from several locations along the outer groove. SEM examination showed that next to the surface, the crack propagated along crystallographic planes, consistent with low crack growth rates consistent with stage I fatigue crack growth. A short distance from the surface (about 10 mils), faint striations were observed, consistent with stage II fatigue crack growth. Striation spacing increased as the crack propagated, consistent with propagation under constant stress. The last striations were observed at 352 mils from the surface. A striation count was performed and determined that the number of cycles the crack propagated was between 5,500 and 7,000 cycles.

A cross-section of the impeller along an axial-radial plane revealed that the crack initiated next to a groove. The radius of the groove was approximately 4 mils. No metallographic anomaly was observed at the crack initiation site. The microstructure of the impeller met the requirement of CPW289G. A cross-section of the impeller along an axialcircumferential plane next to the groove showed a shallow deformed layer that was consistent with the machining of the aft face.

According to the engine logbooks, the engine was last been overhauled on December 13, 1996. At that time, the logbook indicated a total engine time was 5,712 hours, with 4,959 total cycles. The subject impeller was installed during this overhaul. The accompanying FAA Form 8130-3 (Airworthiness Approval Tag) listed the total impeller time as 7,311.5 hours and 6,116 cycles. Since that overhaul, the engine accumulated 1,582.1 hours and 1,760 cycles, for a total of 7,876 cycles since new. The cyclic life limit on the impeller is 8,400 cycles.

NTSB Probable Cause

Inadequate quality control following overhaul of the high-pressure compressor (HPC) impeller by other maintenance personnel that resulted in a machining groove on the aft face of the HPC impeller being undetected, which resulted in a fatigue fracture and an uncontained separation of the HPC impeller from the engine during the takeoff roll/run, and the pilot rejecting the takeoff.