N618BB

On September 28, 1996, at 0835 eastern daylight time, a Mitsubishi MU-2B-30, N618BB, was substantially damaged during a precautionary landing at the Ross County Airport (RZT), Chillicothe, Ohio. The certificated airline transport pilot and three passengers were not injured. Instrument meteorological conditions prevailed for the passenger flight that originated at RZT, at 0830. An instrument flight rules flight plan had been filed for the flight conducted under 14 CFR Part 135.

According to a pilot statement, he departed runway 05, destined for Columbus, Ohio. He further stated:

"...had climbed to around 500 feet. I think it was about 30 seconds after I had brought the landing gear up. We experienced a sudden loss of all power on the left engine. I then secured the left engine, then I began to make a left turn back to the airport and determined that the aircraft would not hold airspeed, or altitude. I decided to continue the left turn to runway no. 23. During this time I was having a great deal of control problems. I was attempting to land on the runway, but may have touched down on the grass. I know that I flared over the runway; however, there was a pulling of the plane to the right caused by the right engine beta. This caused the plane to run into the grass, off of the right side of the runway..."

During a telephone interview, the pilot stated that the takeoff was performed with the flaps set at 20 degrees. When the airplane obtained a positive rate of climb, he raised the gear handle. During the climb, with the gear in transit, the airplane's airspeed was about 110 to 115 knots. The airplane was about 500 feet above the ground, and about 1 mile straight out from the departure end of the runway, when he observed the gear handle red light extinguish. He stated that this indicated that the gear was retracted and the gear doors were closed. At that point, with the airspeed at 120 knots, the left engine lost power.

When the engine lost power, he feathered the propeller, lowered the airplane nose to the horizon, and began a left turn. He estimated the airplane's speed was about 115 knots. He left the flaps set at 20 degrees, and noted the vertical speed was showing a descent of about 200 to 300 feet per minute in the turn. He was too low to see the runway, but continued in the direction of the airport, just above the trees. When the airplane cleared the trees he saw the runway. He then lowered the landing gear, and banked the airplane to the right, to align the airplane with the runway. The airplane was about 50 to 100 feet above the runway when he lowered the flaps to 40 degrees.

In a telephone interview, the pilot rated passenger that occupied the copilot's seat stated that shortly after the gear handle was raised, the left engine lost power. He was not sure if the gear had completely retracted when the loss of power occurred. The pilot immediately feathered the left propeller and started a left turn into the dead engine. The passenger stated that the airplane touched down on the runway at an angle, and when the right engine was reversed on the ground, the airplane veered off the right side of the runway.

According to the Ohio State Police report, the airplane came to rest in a drainage ditch about 900 feet beyond the point where it departed the right side of the runway. A post crash fire erupted on the right engine, and was extinguished by the local fire department.

The left engine was removed and shipped to Allied Signal, Phoenix, Arizona. On November 6, 1996, the engine was examined under the supervision of a Federal Aviation Administration Inspector from the Flight Standards District Office, Scottsdale, Arizona. The Inspector's report stated:

"...the most likely cause of the engine failure was the failure of the torque sensor housing. This allowed the disengagement of the direct-drive fuel control gear train, which interrupted the mechanical drive to the fuel pumps and fuel control assemblies resulting in an engine shutdown failure."

The failed parts were sent to the NTSB Materials Laboratory Division for examination. The metallurgist's factual report stated:

"...The fractures on this arm [torque sensor and gear assembly] contained mechanical damage that resulted from the mating fractures rubbing against each other which obliterated the original fractured features...No surface anomalies such as a gouge mark or porosity was found in the origin area. Fatigue propagation was through the entire wall in the general direction indicated by unmarked arrows...No anomalies such as porosity was found in the microstructure..."

A Garrett Turbine Engine Company (Allied Signal) Service Bulletin was issued September 14, 1979, with subsequent revisions issued through 1986. The Service Bulletin (SB) stated that resonant vibrations set up in the torque sensor housing arm resulted in cracking of the housing arm. It also stated that several instances of torque sensor housing arm fractures occurred, and the corrective action was to replace the torque sensor housing and support with a newer design. The SB further stated that the rework of the torque sensor and gear assembly was to be accomplished at an overhaul facility, and recommended that the SB be accomplished at the operator's convenience, but not later than the next access to the affected parts.

The overhaul of the left engine to zero time and zero cycles by the AIResearch Mfg. Company (Garrett Turbine Engine Company), was completed on December 11, 1979. The SB was not complied with at the time of overhaul. On December 15, 1989, at 1,846 hours, a hot section inspection of the engine was completed, which did not require the examination of the torque sensor. At the time of the failure, the engine had accrued about 3,350 hours, 250 hours short of a mandatory overhaul.

According to Allied Signal Minimum Modification Standards, all recommended SBs which were in effect at the time of overhaul would be incorporated into the engine; however, it further stated that only SBs released 30 days prior to the induction of the engine into the overhaul cycle would be applied. An Allied Signal document revealed that the failed engine was received into the overhaul cycle on August 8, 1979, about 35 days before the torque sensor SB was issued. The overhaul was completed 88 days after the SB was issued.

Service Difficulty Reports (SDR) maintained by the Federal Aviation Administration were reviewed. The SDRs revealed that between 1978 and 1985, over 40 malfunctions of torque sensors were reported, of which over 24 resulted in the loss of engine power. All of these were related to the old style torque sensor. Between 1986 and 1996, only 5 cases of torque sensor problems were reported. These also were related to the old style torque sensor.

Engine monitoring had been conducted through an oil analysis program. An oil analysis report of the left engine oil filter, dated September 24, 1996, recommended, "Resample after 25 hours due to minor magnesium."

Weather reported at other airports at the time of the accident were as follows:

Rickenbacker International Airport (LCK), 24 miles north-northeast of RTZ;

At 0819: wind 320 degrees at 6 knots, visibility 3 miles, clouds 1,000 scattered, 3,500 broken, 6,500 overcast.

Port Columbus International (CMH), 34 miles north-northeast of RTZ;

At 0822: winds 290 degrees at 6 knots, visibility 3 miles and rain, clouds 700 few, 1,200 broken, 3,000 overcast.

At 0832: winds 310 degrees at 9 knots, visibility 4 miles and rain, clouds 1,200 scattered, 3,700 overcast.

Ross County Airport Observation taken by the airport manager;

At 0805: winds 290 degrees at 6 knots, visibility 4 miles, ceiling 700 foot overcast.

In a letter submitted by the pilot, dated November 25, 1996, he stated that factors contributing to the accident were, "...Improper training: Flight Safety International & Enterprises both teach 20 [degree] flaps takeoffs as normal procedure for the MU-2B-30, even when the POH/AFM specifically warns that a climb is not assured should an engine fail in this configuration..." He also stated that the Mitsubishi publications were contradictory and vague regarding takeoff procedures and engine failures.

A review of the MU-2B-30 operator's manual revealed that the first half of the manual was the Pilot's Operating Manual (POM), while the second half of the manual was the Airplane Flight Manual (AFM).

The POM contained a section on general twin-engine performance and controllability, which began on page 3-79. The first paragraph indicated that what followed was generic to twin-engine operation, and was not specific to the MU-2. However, after the lead paragraph was a NOTE that referenced the MU-2B-30. It stated that the left engine was the critical engine on the MU-2B. It further stated, "All other information is technically correct." The NOTE was repeated on page 3-88. The POM general twin-engine performance text contained information regarding P-factor and yawing. It also contained the most efficient operating condition with the "slip/skid" ball out of trim, and banking the airplane into the live engine.

A review of the emergency procedures in the AFM revealed that a NOTE on page 3-2 stated, "Single engine climb rates are best attained with wings level by use of rudder to correct for yawing tendency and using the minimum amount of spoiler necessary to maintain lateral control." The AFM procedure was contrary to the POM procedure.

The AFM published an emergency procedure for "Engine Failure After Liftoff - Gear down or in Transit to UP." The procedure stated, "If the engine failure occurs after liftoff, but before the landing gear cycle is fully completed (gear up, doors closed) and continued flight is not possible:

1. Landing Gear............................DOWN 2. Operating Engine.................