On July 15, 2007, at 1617 Pacific daylight time, a Cirrus SR22, N254SR, landed uneventfully at the Reno/Tahoe International Airport, Reno, Nevada, following a partial loss of engine power in cruise flight at 17,000 feet mean sea level (msl). The private pilot and the passenger were not injured, and the airplane was not damaged. The airplane was registered to Aquatic Innovations Inc. of Alamo, California, and operated by the pilot under 14 CFR Part 91. Visual meteorological conditions prevailed for the landing, and an instrument flight rules (IFR) flight plan was filed. The flight departed from Concord, California, about 1510 with an intended destination of Alpine, Wyoming.

According to the pilot, the airplane was in level flight at 17,000 feet msl when he heard a "burp" sound. There was a loss of manifold pressure and initially, smoke in the cockpit. The oil pressure and fuel flow indications also dropped. The pilot declared an emergency and received vectors to the Reno Airport. As the airplane descended, the engine continued to lose oil pressure, and the oil pressure warning light came on. The engine lost power completely and the propeller came to a stop when the airplane was on final approach to runway 16L. The pilot landed the airplane without further incident.

The airplane was issued an airworthiness certificate on June 5, 2007, and at the time of the incident, had accumulated 48.2 flight hours. It was powered by a Teledyne Continental Motors (TCM) engine model IO-550-N. The engine was modified by the addition of a Tornado Alley Turbonormalizing System installed at the Cirrus factory in accordance with Supplemental Type Certificates SA10588SC and SE10589SC. The Turbonormalizing System provides 29 inches of manifold pressure up to an altitude of 25,000 feet msl. It utilizes two turbochargers with an absolute manifold pressure controller and a pressure relief valve; these components are manufactured by Kelly Aerospace.

Engine data downloaded from the airplane's Avidyne avionics showed normal engine parameters until about 1 hour and 1 minute into the flight. At this time, there was a loss of manifold pressure from about 29.5 inches to about 16 inches, consistent with a loss of turbo boost. The data indicated that the oil pressure was about 40 pounds per square inch (psi) until shortly after the event at which point it started to fall, reaching a minimum of 15.8 psi at the end of the data.

On July 17 and 18, 2007, the airplane was examined under the supervision of the NTSB investigator-in-charge (IIC) by representatives from the Federal Aviation Administration (FAA), Cirrus Design Corporation, TCM, Kelly Aerospace, and Tornado Alley Turbo. A significant amount of oil was observed on the belly of the airplane that appeared to emanate from the left exhaust stack and extended aft to the empennage. The engine oil level was checked, and the level on the engine oil dipstick was about 2 quarts. The turbochargers were identified as part number (P/N) 466304-003. Examination of the left turbocharger (serial number (S/N) KCN01222) revealed its turbine wheel had separated from the shaft and was missing. Examination of the right turbocharger indicated that it was intact, and the compressor/turbine wheel assembly rotated freely when turned by hand.

According to Kelly Aerospace, this P/N turbocharger is supplied to two customers, Tornado Alley and TCM. The turbochargers sold to Tornado Alley are installed only on TCM IO-550-N engines. The turbochargers sold to TCM are installed on TCM TSIO-520-BE and TSIO-550-A, B, C, E, and G engines.

During the course of the investigation, the NTSB IIC learned of four similar incidents involving this P/N turbocharger where the turbine wheel separated from the shaft. Two of the incidents were reported by Tornado Alley personnel and involved Cirrus SR22 airplanes, and two of the incidents were reported by TCM personnel and involved Columbia 400 airplanes powered by the TCM TSIO-550-C engine.

On May 6, 2007, a Cirrus SR22, N552SR, was on a production test flight near Duluth, Minnesota, when the pilot heard a loud bang and noticed a drop in manifold pressure. An uneventful landing was made, and susequent examination revealed that a turbine wheel had separated from one of the two turbochargers (S/N KCN01105) and was missing. Another incident in May 2007 involved a Cirrus SR22, N188G, which was being flown home from the Cirrus factory in Duluth, when the engine experienced a loss of manifold pressure. The pilot landed without further incident at Spanish Fork, Utah, and a subsequent examination revealed that a turbocharger (S/N KCN01182) turbine wheel had sheared from its shaft, but had not departed from the turbocharger housing. The airplanes involved in the Duluth and Spanish Fork incidents had each accumulated less than 20 flight hours.

On May 25, 2007, a Columbia 400, N1571C, was on a production test flight near Bend, Oregon, when a loss of manifold pressure occurred. The pilot landed uneventfully, and an examination revealed that the turbine wheel of the right turbocharger (S/N KBL00797) had separated from the shaft and lodged in the housing. An earlier incident occurred on October 16, 2006, involving a Columbia 400, registration number unknown, in cruise flight at 19,000 feet msl over Germany, when the manifold pressure dropped to 17 inches and the oil pressure dropped to 15 psi. The pilot landed without further incident, and the right turbocharger was removed and sent to TCM. Examination of the turbocharger (S/N JAL00305) revealed that the turbine wheel had separated from the shaft and was missing. The airplane involved in the Bend incident was on its first flight, and the airplane involved in the Germany incident had approximately 100 hours time in service.

At the direction of the NTSB IIC, Kelly Aerospace personnel disassembled the five failed turbochargers and prepared reports documenting their findings. The report regarding the failed turbocharger from the Reno incident airplane (S/N KCN01222) indicated that the turbine wheel had separated from the turbine shaft just above the turbine journal. The turbine housing, compressor housing and compressor wheel had witness marks of contact through 360 degrees. The journal bearings were extruded to the center housing bore. The compressor journal bearing exhibited outboard axial wear. The turbine journal bearing exhibited inboard axial wear. The thrust bearing axial ramps were worn inboard and outboard. The thrust collar exhibited axial wear inboard and outboard. The backplate assembly bore was worn. The compressor wheel backface perpendicularity to the bore was measured at .001 inch. This measurement exceeded the maximum permissible value of .0003 inch.

The reports regarding the failed turbochargers from the Duluth (S/N KCN01105), Spanish Fork (S/N KCN01182), and Bend (S/N KBL00797) incidents documented findings similar to the failed turbocharger from the Reno incident airplane. Specifically, the compressor wheel backface perpendicularity to the bore was measured at .0008 inch, .0006 inch, and .00525 inch, for the Duluth, Spanish Fork, and Bend turbochargers, respectively. All of these measurements exceeded the maximum permissible value of .0003 inch. Kelly Aerospace concluded that this perpendicularity error was the root cause of the failure of these four turbochargers. The perpendicularity error led to abnormal turbocharger vibration, which resulted in abnormal wear. The abnormal wear allowed the turbine and compressor wheels to contact their respective housings, which resulted in fatigue and failure of the turbine shaft.

The report regarding the failed turbocharger from the Germany incident (S/N JAL00305) described a different discrepancy. During disassembly of this unit, it was discovered that the compressor end journal bearing and snap ring had been omitted at assembly. The report concluded that the turbocharger failed due to this omission.

The fractured pieces of the turbine wheel assemblies from the five turbochargers were sent to the NTSB materials laboratory for examination. No manufacturing defects that would have contributed to the failure of the shafts were found. For details of the examination see the Materials Laboratory Factual Report in the public docket for this accident.

Kelly Aerospace traced the source of the perpendicularity error to a change, made in March 2007, to the machining process for the compressor wheel. (Kelly machines these compressor wheels (P/N 409826-0011) in house, and they are used only in the P/N 466304-003 turbocharger.) The change involved a switch from a two-machine process to a three-machine process. This change was made without formal documentation or coordination with other functional areas; no formal first article inspection for the new process was performed. In April 2007, the machine shop returned to the original two-machine process. The compressor wheels manufactured during the time that the three-machine process was in use were installed in turbochargers produced from March 20 to April 20, 2007. This included the four turbochargers involved in the Reno, Duluth, Spanish Fork, and Bend incidents.

Kelly Service Bulletins No. 026, Revision B, dated July 27, 2007, and No. 027, dated July 25, 2007, were issued identifying the affected turbochargers by serial number and mandating their replacement. On August 8, 2007, the FAA issued airworthiness directive (AD) 2007-16-10 requiring compliance with Kelly Service Bulletins 026 and 027. The AD became effective on August 23, 2007.

Use your browsers 'back' function to return to synopsis
Return to Query Page