Wednesday, March 13, 2013

Solving a U-2 “Mystery”

We discovered a unique use for flight simulators
by Charles S.

In the early 1970s, I was assigned to Wright Field to support human factors research with flight simulation. It involved analysis of new cockpit configurations and equipment to be used in new aircraft under development. We had three reconfigurable simulators (F-111, V-135, F-4) to support real-time analysis of pilot/aircraft interactive control problems.

One day, the Air Force visited our facility and asked us to help with an aircraft accident investigation. A Lockheed U-2 aircraft had just taken off when one of the wing-tip ground support gear failed to eject. Procedure was to circle back around and fly down the runway centerline at a low altitude in level flight while rocking the wings in an attempt to dislodge the gear. As the pilot rocked the wings, the aircraft abruptly rolled into a steep bank and struck the runway in spite of the pilot’s attempt to right the aircraft. The aircraft flight data recorder had been recovered intact from the wreckage. The Air Force wanted us to evaluate the data.

I indicated that we had the capability to attach the flight data recorder to the simulator and play back the recorded flight control positions into the simulator in real time. We could then repeat the flight control positions into the simulator and analyze the aircraft dynamics associated with the pilot/aircraft interactive control, while the flight recorder flew the simulator, repeating the last few minutes of flight. It would require the Air Force obtaining the Lockheed aerodynamic model and mass properties of the U-2 to enable us to program the simulator with the aircraft flight characteristics.

After installing the U-2 flight model, I partitioned the wing aerodynamic model into six sections, using the inner-outer span wise boundary of the wing trailing edges devices as the division between sections (two inboard flaps, two outboard flaps and two ailerons). Each section could then have its own independent set of aerodynamic angles and pressures to determine aerodynamics and moments acting on the aircraft under pilot control.

We initialized the simulator to the aircraft recorded conditions that existed as the aircraft rolled out of the final bank on runway centerline and then let the recorder fly down the runway to repeat the remainder of the flight to impact. We found that even though the aircraft was above the stall speed, the left outboard wing section forward of the left aileron stalled when the pilot input a sharp roll right command, as the aircraft was rolling left. The full down position of the left aileron commanded by the roll right command reduced the angle of attack required for the onset of stall for that section of wing to the stall point, and the right aileron roll right section command was not strong enough to appose the roll moment resulting from the loss of lift on the entire left tip as the aircraft continued to roll left. None of the other five sections of the wing had stalled. The roll progressed quickly and the pilot held the roll right command until impact with the runway approximately 92 degrees of left bank. The simulator repeated the maneuver and struck the runway at 96 degrees of left bank at approximately the same position on the runway.

We programmed the simulator with U-2 flight characteristics.