Boeing 717-200 N482HA

The airplane unexpectedly rolled to the left during the climb to cruise. The takeoff and initial area departure phases were normal. The captain engaged the autopilot at 3,000 feet mean sea level (msl). Passing 8,900 feet msl, the airplane experienced a sudden upset resulting in a rapid roll of 35 degrees to the left. The captain countered with 50 to 60 degrees of right aileron input. The autopilot had been engaged, but disengaged automatically during the pilot control input. The first officer (FO) asked the captain if he wanted some aileron trim. The captain responded that he desired only a small amount, reasoning that he did not want to mask a major problem. He did not feel any vibration, binding, or unusual control feel other than the effort and aileron input required to hold wings level. He remembered seeing the right aileron and right elevator deflected upward on the Configuration synoptic page on the systems display, and thought that the aileron deflection would be consistent with his control input to lower the right wing against a left rolling tendency. He flew the airplane to a landing with approximately 45 degrees yoke deflection required to maintain a level wing attitude. Analysis of the digital flight data recorder (DFDR) data indicated that the aileron trim value was 90 percent of full aileron trim authority prior to pushback, engine start, taxi, and takeoff. The data from the previous inbound flight, which had the same flight crew showed that the trim changed from 10 to 90 percent authority about 3 minutes after the previous flight landed. The trim again shifted following the landing and taxi back of the event flight from 85 percent to 10 percent. The captain said that he felt that the airplane was not tracking properly after takeoff. The airline's debrief of the flight crew indicated that they did not see or recognize the excessive aileron deflection on the cockpit screens before or after the event worsened and resulted in autopilot disengagement. The autopilot was trying to roll out of a turn, and was commanding aileron in a direction opposite to the trim position, which resulted in the aileron torque limiter reducing the aileron authority. With the reduced authority, the aileron position could not keep up with the autopilot command, which triggered a autopilot disconnect. The airline indicated that the Rudder and Aileron check for zero is done during the Cockpit Preparation Checklist, which a crew accomplishes just prior to the Before Starting Engines Checklist. DFDR data about 2/3 of the way through the flight indicated that the trim did respond when the copilot manipulated it. The data also showed that the trim repositioned back to zero at the end of the flight. Thorough post accident tests of the system found that the aileron trim control switches and the aileron trim control actuator unit functioned per design specifications with no mechanical or circuit discrepancies noted. Testing demonstrated that there was no evidence of any uncommanded motion. Tests were also performed on the Aileron and Rudder Trim Control Module (ARTCM) and no faults were found.

Quoted verbatim from the NTSB record.