The fatal crash of the second Leonardo Helicopters (formerly AgustaWestland) civil tiltrotor prototype (AC2), N609AG, on Oct. 30, 2015 at Tronzano Vercellese, Italy, is ascribable basically to the “combination of three factors”: the development of latero-directional oscillations; the inability of the fly-by-wire flight control system (FCS) control laws to maintain controlled flight; and the failure of the engineering flight simulator (SIMRX) to “foresee the event in any way,” according to the final report from Italy's National Agency for Flight Safety (ANSV—Agenzia Nazionale per la Sicurrezza del Volo). The accident aircraft had accumulated 567 hours since first flying in 2006. It took off from the company's production facility at Cascina Costa and crashed at 10:42 a.m. local time while executing a third planned high-speed descent as part of test flight T664. During the descent the aircraft entered uncontrolled flight in a series of lateral-directional oscillations, broke up and caught fire in flight before striking the ground, killing both test pilots.
Difficulty of Recovery
The ANSV said that a combination of ground debris mapping and telemetry data led it to “hypothesize with reasonable certainty” that the aircraft broke up in flight as a result of multiple prop-rotor strikes from excessive blade flapping on the wings as a consequence of excessive yaw angles reached during the fatal dive. This damaged the hydraulic and fuel lines that are positioned along the wing leading edges, precipitating the in-flight fire. The aircraft was equipped with flapping stops, but they were not designed to “contain the effects of the extreme aerodynamic forces generated during the event.” Because of the aerodynamic characteristics of the aircraft and the specific conditions created by the dive, the flying pilot's attempt to counteract the oscillations with a roll-tracking maneuver to level the wings was ineffective, partly because the FCS was designed to “couple” on more axes than the command inputs given on the single axis by the pilot.
Specifically, “Total lateral control resulting from the summation of pilot input and automatic FCS input has an effect on the yaw axis through aerodynamic coupling and feedforward and feedback turn coordination automatically provided by the FCS. Consequently, giving a command in counterphase on the roll axis to dampen the relative oscillations creates an effect on the yaw axis that can be in phase with the yaw oscillations. This occurred during the accident: the correction of the roll oscillation induced, by the control laws of the FCS, a manuever in phase with the oscillations of the yaw axis, generating a divergence of the oscillations.” The ANSV said that the “low frequency and low amplitude nature of the oscillations” made them difficult for the pilots or ground crew to perceive until the roll and yaw “reached excessive levels only a few seconds before loss of control.”
The pilot flying also made rudder-pedal inputs. As explained above, the inputs exacerbated the situation, taking sideslip to maximum values. The tiltrotor entered a dive at the 293-knot design dive speed. AC2 was fitted with a new tapered rear fuselage and redesigned vertical fin with less surface area. During the dive, the aircraft reached 306 knots.
Investigators attempted to recreate the accident flight in the AW609 SIMRX in Philadelphia using the same software and flight conditions, but could not; they came close by inserting algorithms that changed the aerodynamic configuration of the aircraft, but even then the lateral-directional oscillations developed were in a different phase. They did, however, use the exercise to verify the “great difficulty” of recovery to controlled flight under the conditions. The ANSV found the inability to replicate the accident flight in the simulator unremarkable, given “the lack of experimental data obtained previously in the wind tunnel and in-flight evaluations with those speed conditions and relating to the recent modified geometry of the tail fin; this last change was considered conservatively by entering a reduction in the tail fin area into the database and then implementing the computational fluid dynamics.”
The ANSV made several safety recommendations after the accident: more high-speed and complex-flight-condition modeling, verification and wind tunnel testing as part of the AW609 certification process; and verification of the flight control laws in extreme flight conditions, in particular reviewing their effectiveness with regard to pilot inputs and uncommanded coupling effects.
AW609 flight-testing resumed in August last year. AC3 is flying from the company's Philadelphia facility and recently completed testing for flight into known icing. AC4 is under assembly in Philadelphia and is expected to fly next year. AC1 is in Italy undergoing modifications before return to the flight-test program. Leonardo Helicopters expects FAA certification next year.