On April 30, 2000, at 2142 Eastern Daylight Time, a Canadair CL-600-2B19, N924CA, operated by Comair as flight 5597, received minor damage when it experienced an uncommanded roll, while on approach to Westchester County Airport, White Plains, New York. The 2 certificated airline transport pilots, 1 flight attendant, and 50 passengers were not injured. Visual meteorological conditions prevailed for the scheduled passenger flight that originated from The Greater Cincinnati/Northern Kentucky Regional Airport, Covington, Kentucky. Flight 5597 was operated on an instrument flight rules (IFR) flight plan that was conducted under 14 CFR Part 121. Use your browsers 'back' function to return to synopsisReturn to Query Page
According to interviews with the pilots, the captain was the operating pilot. The pilot had been cleared for a visual approach to runway 34, and the flight was uneventful until the uncommanded roll. After turning final, the wing flaps were selected to move from 30 degrees to 45 degrees. As the wing flaps extended, the airplane performed an uncommanded roll to the right. In addition, the Master Caution light illuminated, and the engine indicating and caution advisory system (EICAS), displayed a flight spoiler caution alert.
The captain reported that he was already hand flying the airplane, when the roll occurred, and he applied correction with the rudder and aileron. In addition, he had to hold the correction, to maintain a wings level attitude. He also checked that the spoiler handle was stowed and full forward, and momentarily moved the ground spoiler ARM/DISARM switch from automatic, to manual, and back to automatic with no change in the situation. He then checked the flight control page of the EICAS display and remembered seeing a flight spoiler extension of about 10 degrees on the left wing and about 20 degrees on the right wing.
The captain reported that the airplane was fully controllable, and added additional power as the airplane continued toward the runway with about 10 extra knots of airspeed. After landing, the airplane was taxied to the gate, where the passengers were disembarked through the airstair door.
The airplane was parked at the gate, and electrical power remained applied. The captain then conducted a walk-around inspection of the airplane. All spoiler panels were found retracted except for the flight spoiler on the left wing, which was extended about 10 degrees.
The airplane was equipped with four spoiler panels on each wing. The outboard panels were referred to as spoilerons, and were located next to the ailerons. Their extension was tied to movement of the cockpit control yoke. The next inboard panel on each wing were the flight spoilers. A lever on the cockpit center pedestal controlled their extension and retraction. The two inboard panels were ground spoilers, and only extended when the airplane was in the ground mode. In addition, once the airplane had transitioned to the ground mode, the spoilerons and flight spoilers would also extend, if the ground spoilers were extended.
Each flight spoiler had two attach lugs, which were attached to independently operated hydraulic power control units (PCU). However, only one PCU was powered at a time. The system would switch over between PCUs after the ground spoilers were stowed, after landing.
Examination of the airplane by mechanics from Comair revealed that the outboard connecting lug between the right wing flight spoiler, and its associated PCU was fractured. Both halves of the failed lug were recovered, and forwarded to the Safety Board Materials Laboratory for testing. Functional testing of the other components from the spoiler system did not reveal any additional problems.
According to the Safety Board Materials Laboratory Factual Report;
"...Examination [of the fractured lug]...revealed that the fatigue originally initiated in the corner between the bore of the lug and the chamfered break edge on the outboard face...As the fracture progressed, secondary fatigue cracks initiated along the bore of the lug...The fatigue fracture extended approximately 0.16 inches toward the outer surface of the lug and approximately 0.31 inches inboard along the bore of the lug. The rest of the fracture face contained overstress fracture features stemming from the terminus of the fatigue fracture...The bore surfaces of all lugs from the accident spoiler rest on the outer race of spherical bearings that are inserted into the holes in the lugs. Examination of the bore surface in the fractured lug revealed fretting damage in the area located adjacent to the fatigue fracture...."
According to the METALS HANDBOOK, VOL 10, Failure Analysis and Prevention:
"FRETTING is a wear phenomenon that occurs between two mating surfaces; it is adhesive in nature, and vibration is its essential causative factor. Usually, fretting is accompanied by corrosion. In general fretting occurs between two tight-fitting surfaces that are subjected to a cyclic, relative motion of extremely small amplitude...Fretted regions are highly sensitive to fatigue cracking. Under fretting conditions fatigue cracks are initiated at very low stresses, well below the fatigue limit of nonfretted specimens....":
"...The relative motion required to produce fretting damage may be quite small. Displacements of 0.03...microns [1.15 X 10-6 inches] are sufficient to cause damage, but the amplitudes of displacement usually seen in service are of the order of a few thousandths of an inch...."
According to the Safety Board Flight Data Recorder Factual Report, the airplane was equipped with a digital flight data recorder (DFDR). The data from the recorder revealed the airplane was descending through 1,300 feet, with the wing flaps extending from 30 degrees to 45 degrees, when it suddenly rolled right about 11 degrees. The ailerons were immediately deflected about 10 degrees, the left spoileron extended about 4 degrees, and the airplane returned to a wings level attitude. With the exception of minor variations, the airplane remained in this configuration until touchdown, about 85 seconds later.
The failed lug was an integral part of the flight spoiler. According to Comair, the right wing flight spoiler was original equipment with the airplane. It had accumulated 16,194.7 flight hours, and 15,261 cycles.
The spoiler had last been inspected on November 10, 1998, during a 2C phase inspection. The inspection of the spoiler consisted of a visual, external inspection of the spoiler attach point, without disassembly. Since the inspection, the part had accumulated an additional 3,469.8 hours.
The right wing spoiler extension was not recorded on the DFDR. According to an engineering report from Bombardier:
"...The floating RH [right hand] FS [flight spoiler] surface event was not recorded on the FDR because it [the FDR] obtains its information from the active PCU, which remains retracted as commanded. It is conceivable that the flight crew obtained the surface information from the FLT/CTRL [flight control] Synoptic page (EICAS) on the Co-Pilot side, which receives its information from the floating standby PCU. The Pilot side FLT/CTRL Synoptic page would show the same information, for the spoilers, as is received by the FDR, and thus would not have shown a surface deployment...."
"...No evidence of an uncommanded deployment of the LH FS exists. Unlike the RH FS, the LH surface remained intact, maintaining the connection between the two PCUs. If the surface deployed to 10 degrees as reported by the pilots, either it would have been recorded on the FDR or, if it were not recorded, it would have had to deploy and retract in the one-second window of the FDR updates. The latter could have resulted in a position error for the active PCU...."
In a later email, a representative of Bombardier amplified the information about where the FDR gets its information. The email stated in part:
"...the connection to the FDR is made via one of the Data Concentrator Units (DCU), No. 1 (by default) or No. 2 should there be a detected fault on DCU 1. DCU 1 always receives the RH [Right Hand] OB [Outboard] Flight Spoiler PCU information whether the PCU is active or in standby mode. The co-pilot's side Synoptic Page would have displayed the deployment via DCU 2. Had there been a detected fault on DCU 1 then the FDR would also have seen the DCU 2 information...."
The email also stated:
"...We believe that the spoiler system posted a STATUS message and not a CAUTION message. Spoiler logic would not cause a Master Caution Switch illumination."
Canadair had issued service bulletin (SB) A601R-57-027, on April 19, 1999, that specified non-destructive testing for the lug. This was based upon two previously known similar failures. According to personnel from Comair, the SB had not been complied with. Personnel from the Federal Aviation Administration (FAA) reported that Comair was not required to comply with the service bulletin.