NTSB Factual Report

HISTORY OF FLIGHT

On January 25, 2007, at 1647 mountain standard time, a Bombardier CL-600-2B19, N17337, operated by Mesa Airlines, doing business as U.S Airways Express flight 2985, sustained minor damaged when it experienced a fan disk separation while climbing through approximately 24,000 feet (FL240), 8 miles west, southwest of Cripple Creek, Colorado. Day visual meteorological conditions prevailed at the time of the incident. The regularly scheduled domestic passenger flight was being operated under the provisions of Title 14 Code of Federal Regulations (CFR) Part 121 on an instrument flight rules flight plan. The captain, first officer, flight attendant, Federal Aviation Administration (FAA) inspector, and 50 passengers were not injured. The flight departed Denver International Airport (DEN), Denver, Colorado, at 1629 and was en route to Phoenix, Arizona.

According to an interview, subsequent telephone conversations, and a written statement from the flight crew, they were climbing through approximately FL240 when they heard a loud pop and experienced a vibration or buffet. The airspeed began to slow. The crew reported that they were not able to maintain a climb configuration or a level altitude, and initiated a controlled descent. They observed a "no oil pressure" indication for the number one engine along with a maximum indication on the N1. Thrust was reduced to idle power.

The crew reported that they reversed course for a return to DEN, and were able to maintain an altitude of 12,000 feet mean sea level (msl). The flight crew received vectors for a visual approach and landed uneventfully. The crew did not report any airplane controllability issues after the separation.

DAMAGE TO AIRCRAFT

The National Transportation Safety Board investigator-in-charge (IIC) arrived at DEN approximately 0800 on January 26, 2007. The Safety Board IIC, a Safety Board powerplant engineer, a Safety Board structural engineer, the FAA Aviation Accident Investigation (AAI-100) IIC, and representatives from Bombardier, General Electric (GE) Aviation, and Mesa Airlines, examined the airplane and number one engine on January 27 and 28, 2007. Examination of the number one engine revealed that the inlet, fan rotor assembly, fan containment case, and portions of the thrust reverser (TR) were all missing. Examination of the airplane revealed damage to the fuselage, in-line with the plane of rotation of the number one engine fan rotor, and marks on the vertical and horizontal stabilizers.

PERSONNEL INFORMATION

The pilot, age 40, held an airline transport pilot certificate with an airplane multi engine land rating, and a commercial pilot certificate with an airplane single engine land rating. In addition, he held type ratings in the Beech BE-1900 and CL-65. He was issued a first class airman medical certificate on November 2, 2006. The certificate contained no limitations. According to Mesa Airlines, the pilot had logged approximately 11,000 hours total time; 4,000 of which were in the make and model of the incident airplane.

The first officer, age 34, held a commercial pilot certificate with airplane single engine land, multi engine land, and instrument ratings. In addition, he held a type rating in the CL-65, with the limitation of second-in-command privileges only. He was issued a first class airman medical certificate on February 16, 2006. The certificate contained the limitation "must wear corrective lenses." According to Mesa Airlines, the first officer had logged approximately 3,700 hours total time; 2,300 of which were in the make and model of the incident airplane.

AIRCRAFT INFORMATION

The incident airplane, a Bombardier CL-600-2B19 (serial number 7337), was manufactured in 1999. It was registered with the FAA on a standard airworthiness certificate for transport operations. The airplane was powered by two GE CF34-3B1 (serial number 872749 and 872190, number one and two respectively) turbo fan engines.

The airplane was registered to Wells Fargo Bank, operated by Mesa Airlines, and was maintained under a continuous airworthiness inspection program. At the time of the incident, the airframe had 19,629 hours total time, and 15,438 total cycles. The incident engine (No. 1) had 13,627 hours time since new (TSN), and 10,849 total cycles since new (CSN). According to the Engine Service Record, the incident engine was installed on the event airplane on January 10, 2007. Previously, the No. 1 engine had been installed on a company airplane; serial number (SN) 7410.

METEOROLOGICAL CONDITIONS

The closest official weather observation station was Colorado Springs Municipal Airport (COS), Colorado Springs, Colorado, located 32 nautical miles (nm) east, northeast of the incident site. The elevation of the weather observation station was 6,187 feet msl. The routine aviation weather report (METAR) for COS, issued at 1654, reported, winds, 140 degrees at 9 knots, visibility, 10 statute miles; sky condition, clear, temperature 05 degrees Celsius (C); dewpoint, minus 04 degrees C; altimeter, 30.18 inches.

FLIGHT RECORDERS

Cockpit Voice Recorder

The airplane was equipped with a Fairchild Model A 100S 30-minute solid-state cockpit voice recorder (CVR). The CVR was secured and sent to the National Transportation Safety Board's (Safety Board) Audio Laboratory for readout. The CVR had not sustained any heat or structural damage and the audio information was extracted normally from the recorder. The recording consisted of four channels of excellent quality audio information. The arrival and landing back into DEN was captured on the recording. The engine separation occurred prior to the start of the recording. A CVR group was not formed and a transcript was not prepared.

Flight Data Recorder

The airplane was equipped with an L-3 Communications Fairchild Model F1000 digital flight data recorder (DFDR), which recorded 412 parameters of airplane flight information using solid state media as a recording medium. The DFDR was secured and sent to the Safety Board Vehicle Recorders Laboratory for readout. The DFDR had not sustained any heat or structural damage and the information was extracted normally from the recorder. A DFDR group was not formed. For this incident, 82 parameters were checked and validated, and a report was prepared.

Review of the DFDR data showed that the takeoff and climb were normal. At a time(T) of T = 546 seconds (9 minutes and 6 seconds after takeoff and 4 minutes and 9 seconds prior to the fan disk separation) the vibration level for the No. 1 engine fan rotor assembly (N1) started to increase from a relatively stable level around 0.7 mils. At T = 635 seconds, the vibration level reached over 1 mil and continued to steadily increase to 1.5 mils and 1.8 mils at T = 746 seconds and T = 789 seconds, respectively. After T = 789 seconds, the vibration quickly increased and reached its maximum value of 2.65 mils at T = 794 seconds, corresponding to the time just before the fan disk separation.

The fan rotor assembly, just prior to the fan disk separation, had been stable at approximately 95 percent N1 and the airplane was steadily climbing at around 290 knots. At the moment identified as the fan disk separation, the fan rotor speed had increased to 99.6 percent N1 and the airspeed started to decrease; however the airplane continued to climb for another twelve seconds before the pilot initiated a descent. The fan rotor assembly speed reached its peak, 133.94 percent N1 one second after the disk separation, and immediately started to decrease. The fan rotor assembly speed continued to decrease and eventually reach zero percent N1 at T = 879 seconds (1 minute 24 seconds after the fan disk separation). The airplane's airspeed and altitude eventually stabilized at 200 knots and 11,000 feet, respectively as the airplane returned to DEN.

MEDICAL AND PATHOLOGICAL INFORMATION

FAA toxicological tests were not performed on the flight crew. Both crew members submitted specimens for drug screening following the incident. Tests for marijuana, cocaine, amphetamines, opiates, and PCP were all negative.

TESTS AND RESEARCH

Airframe Examination

A detailed examination of the damage to the airplane fuselage was conducted on January 27, 2007. The airplane sustained damage to the fuselage in-line with the plane of rotation of the event engine fan rotor, as well as marks on the vertical tail, the right horizontal stabilizer, and left engine pylon damage. The damage to the airplane fuselage was comprised of denting, one puncture measuring approximately 5 inches by 2 inches, a skin tear measuring approximately 8 inches in length, and several scrape and gouge marks. The left engine pylon was deformed and fractured forward of the forward engine mount, and the forward engine mount exhibited minor scoring damage.

Engine Examination

A detailed on wing examination of the No. 1 engine was conducted on January 27 and 28, 2007, in a United Airlines hangar located at Denver International Airport. Representatives from the Safety Board, FAA, Bombardier Aerospace, and General Electric were present during this examination.

An examination of the incident engine revealed that the inlet, the fan disk, the forward fan case assembly, and the TR assembly were all missing except for a small piece of the aft flange of the fan disk that remained attached to the fan forward shaft, the rear flange of the forward fan case which remained attached to the fan aft case, and four TR blocker doors. Pieces of the Kevlar® wrap, used as the ballistic shrapnel-resistant shielding within the forward case assembly, were found entangled with the remains of the fan case assembly aft flange.

Fan Disk Search and Recovery

On January 31, 2007, a search team comprised of persons from the Safety Board, FAA, and GE, supported by the Teller County Sheriff's Department and an aircraft recovery company, conducted a ground search for engine parts, with an emphasis on recovering the fan disk. A section of the fan disk, several loose fan blades, parts of the engine cowling and TR, the engine spinner, and pieces of the fan containment case were recovered.

The recovered fan disk section represented approximately one third of the entire disk, from pinhole 20 through 28, with the interior ends of pinholes 1 and 19. Five full-length fan blades, one half-span blade, and three blades fractured above the platform remained attached to the disk piece. The search did not recover any additional pieces of the fan disk. The fan disk section was shipped as recovered to the materials laboratory in Washington, D.C. for metallurgical examination.

Metallurgical Examination

The Safety Board materials laboratory metallurgists along with representatives from General Electric examined the fan disk from February 5-7, 2007. A summary of the findings follows.

According to the metallurgical report, examination of the fan disk revealed it had fractured through the blade retention pinholes of 1 and 19. Visual examination of the fracture area intersecting pinhole 19 revealed large areas of gross plastic deformation, bulk material yielding, and highly textured shear lips, all indicative of tearing overstress separations. The fracture intersecting through pinhole 1 exhibited a finely textured flat region located at the aft bore corner of the aft disk and a faint curved crack arrest mark that delineated a transition from the smooth textured region to the rough surrounding region.

A section of the aft fan disk bore containing the smooth textured region was saw cut and examined using a scanning electron microscope (SEM). Fatigue striations were observed coming from the aft bore corner in an area containing a depression and raised material. The fatigue region measured 0.016 inches along the bore and 0.0185 inches at the aft face. The raised material region had a blue-to-straw color, typical of high temperature exposure of the titanium to air, and consistent with localized electrical arc damage (also referred to as 'arc-out' damage) that melts the surrounding area then resolidifies into a smooth raised fusion zone or nugget. Two overlapping fusion zones were noted in the raised material area. The entire arc area measured approximately 0.055 inches radially along the edge of the hub corner, of which, the raised fusion zone accounted for about 0.035 inches. Based on projections from the existing undamaged disk surfaces, the original longitudinal size of the fusion zone was estimated to be between 0.025 inches and 0.030 inches; therefore the entire arc damage area was estimated to be 0.055 inches by 0.050 to 0.060 inches.

The fracture region adjacent to the fusion zone displayed several ratchet mark like features. Propagating from each of the ratchet marks, were closely spaced fatigue striations. The closely spaced fatigue striations propagating from each of the ratchet marks was indicative of multiple fatigue origins at the fusion zone interface. GE produced a curve of striation density versus measured crack length to determine the estimated number of cycles for the crack to propagate to failure. The curve showed a rapid drop in measured striation densities from the fracture origin that, according to GE, is a characteristic consistent with mission (flight) driven low cycle fatigue (LCF) crack propagation mechanisms. An estimated 15,000 cycles was calculated across the fracture to the estimated fatigue crack end starting at the edge of the remaining arc nugget fracture. According to the maintenance records, fan disk SN GEE1486K had accumulated 10,849 CSN at the time the fan disk separation. GE performed an analytically prediction of the striation densities for an initial flaw size of 0.023 x 0.018 inches and created a curve that was overlaid on the measured striation density curve from the failed fan disk. Based on the two curves, GE concluded there that was excellent correlation between the measured and the predicted striation densities and the fatigue crack initiated and propagated very early in the life of the component.

An energy dispersive X-ray spectroscopy (EDS) spectra from the fracture surface was consistent with the specified material, Titanium (Ti) -17. Core microstructure, well away from the arc region, was composed of acicular alpha in a matrix of transformed beta, a structure consistent with the specified processing of the fan disk material.

The Safety Board performed a series of tests to produce exemplar electrical arc damage to compare it with the damage observed on the failed fan disk. Electrical arcs were produced on sections of the failed fan disk using direct current at an arbitrarily chosen 28 volts, with the maximum current flow of 50 amps. A positively charged brass (copper zinc alloy) rod was brought close to the negatively charged fan disk section and two arc indications were produced. The current polarity was then reversed (probe negative) and four more arc indications were produced. During the experiment the brass rod contacted the fan disk section in several occasions.

The positive probe arc indications tended to be somewhat larger in the affected area than the negatively charged probe arcs, 0.07 to 0.08 inch versus 0.04 to 0.05 inch. Furthermore, the central fusion zone region for a positively charged probe arc was depressed with an eroded or washed out appearance whereas for a negatively charged probe the fusion zone had a built up central peak.

ADDITIONAL INFORMATION

Arc-Out Events

Prior to the fan disk separation, there had been 30 previous CF34 fan disks with arc-out damage, some of which with multiple damage sites, but none of which resulted in a fan disk separation. GE first discovered that CF34 fan disks had been arc damaged in June 2000, when an operator reported a crack in the forward tang of a CF34 fan disk. GE initiated a root cause investigation and found that the crack had initiated from a site where electrical arc-out damage was present. The source of the electrical arc-out damage was believed to have been the electro-chemical etch permanent marking procedure used to match mark the fan disk and forward fan shaft during initial engine assembly at GE. GE concluded that the arc-out damage was caused during the electro-chemical etching pro

NTSB Probable Cause

A fatigue fracture, which resulted in the separation of the No.1 engine fan disk. Contributing factors in the accident were the operators' maintenance personnel lack of training to be able to detect arc-out damage on the fan disk, and the fan disk manufacturer's inadequate electro-chemical etch match marking process of the fan disk to forward fan shaft that allowed for the initiation of a fatigue point that was undetected at the time of original manufacture.