AW609 crash: final report points to oscillations and flight control laws

Severe latero-directional oscillations during a high-speed dive caused the fatal crash of a Leonardo AW609 during flight-testing in 2015, Italian investigators have found.

AW1260 AW609
The fatal crash of the second AW609 tiltrotor prototype last year raised new questions about this 20-year-old development program. Leonardo Helicopters Photo
Advertisement

The aircraft’s excessive yaw angle forced its proprotors to hit its wings multiple times, damaging the hydraulic and fuel lines, and causing an in-flight breakup and fire. The resulting crash fatally injured test pilots Herb Moran and Pietro Venanzi.

The findings are detailed in the final report on the accident from Italy’s national agency for flight safety (Agenzia Nazionale per la Sicurezza del Volo, or ANSV), in which it also points to two other causes: the AW609’s flight control system (FCS) controls laws, and a project simulator (SIMRX) that “did not foresee the event in any way.”

The ANSV also noted that the accident flight was the first in which such speeds had been reached in the new configuration of a streamlined fuselage in the tail and a reduced tail fin surface.

A nacelle and parts of the left wing of the aircraft.
A nacelle and parts of the left wing of the AW609. ANSV Photo

The aircraft’s wreckage was found in three main parts near the city of Tronzano Vercellese in Italy. The ANSV said the distribution of the debris was coherent with a structural breakup in flight, which then caused an explosion and ballistic trajectory towards the point of impact on the ground.

The accident took place as the aircraft was performing the third high-speed dive of a test flight on Oct. 30, 2015. The pilots commenced the dive with a left 180-degree turn, targeting 293 knots for the maneuver (though the aircraft reached a maximum airspeed of 306 knots as the crew attempted to resolve the ensuing controllability issues). According to the report, the oscillation started on the roll axis following the exit from the turn, about four seconds into the maneuver. Another oscillation, this time in yaw, was added to the initial slight oscillation in roll shortly afterwards.

The original design of the AW609's rear fuselage and tail fin are shown on the left.
The original design of the AW609’s rear fuselage and tail fin are shown on the left. The new design appears on the right. ANSV Image

“The crew did not initially react using inputs to counteract them,” the ANSV states, noting that the oscillating phenomenon had been noticed in previous test flights, but it was considered to be slight and not dangerous, and was believed to be self-damping.

As the crew felt the oscillations increase in magnitude, about 23 seconds into the maneuver, Moran tried to counter them with by “roll tracking” — maneuvering the aircraft on the roll axis — the standard pilot procedure for this type of condition. Noticing a pronounced yaw condition, he then attempted to counter this using his rudder pedals.

Around this time, an amber “QBALTH” message appeared on the EPDU, indicating a problem with the torque balancing ratio.

The ANSV explained that the aircraft’s control laws worked against Moran’s countering maneuvers. “A roll command [in the AW609] is transferred by the control laws into different commands that are sent to the control surfaces that act on the roll (for example: flaperons) and to the differential collective pitch control, that, in this aircraft, regulates yaw,” the ANSV stated in the report. This coupling is to compensate for the expected aerodynamic effect of flaperon control surface motion.

So, despite Moran performing the standard compensating procedure, it served to increase the oscillations.

A few seconds later, the first proprotor came into contact with the leading edge of the right wing “and the aircraft started to become irredeemably uncontrollable.”

The ANSV said the excessive flapping of the proprotors was caused primarily by the sideslip angle reached by the aircraft, that exceeded, by nearly two and a half times, the maximum flight envelope value at the speed of 293 knots (10.5 degrees as opposed to the four degrees maximum allowed).

Advertisement

A similar phenomenon had been found during a flight test on July 17, 2014, when angle of attack, angle of bank, mach number, rate of decent and number of “g” caused an accelerated stall of the aircraft right wing, and a significant sideslip developed due to lateral acceleration.

The situation caused excessive flapping on the right proprotor to the extent that it made light contact with the leading edge of the right wing, but in that instance, the crew was able to maintain control of the aircraft and perform an emergency landing.

Following the 2014 incident, Leonardo established new procedures and limitations in the flight envelope. A new parameter (QBALTH) was added to be continuously monitored; between 0.7 and 1, an amber message appeared on the EPDU, with no crew action required. Above 1, and the message appeared in red, and the test would have to be interrupted and the aircraft smoothly leveled.

During its investigation, the ANSV said that it visited AgustaWestland Philadelphia Corporation to use the aircraft’s flight simulator, but noted that it was “not possible” to reproduce the conditions that occurred during the accident.

“As evidenced by the tests carried out by the ANSV, the simulator demonstrated not being able to faithfully reproduce the dynamics occurred during test flight T664 [the accident flight], reasonably due to the non-representativeness of the aerodynamic set, for the unique and extreme conditions encountered, obtainable in the wind tunnel for the new updated configuration including the tapered rear fuselage and the modified tail fin,” the report states.

“Therefore, the [simulator] was not really able to properly carry out the role of test bench for the control laws and risk reduction.”

In its safety recommendations, the ANSV said the AW609’s control laws should be reviewed in the management of the extreme flight conditions in which the aircraft could possibly fly. “That verification should be addressed to ensure the effectiveness of the flight controls inputs given by the pilot avoiding the possibility of unexpected and un-commanded coupling effects.”

It also called for the mandatory requirement of flight data recorders in experimental aircraft — those on the AW609 were in place solely because Leonardo had chosen to do so, but were central to providing the information the agency needed to piece together the accident.

When reached for comment, Leonardo said it would issue a statement “following complete analysis and review of the report.”

6 thoughts on “AW609 crash: final report points to oscillations and flight control laws

  1. Testing on a ‘new’ type of aircraft should always have a chase plane which should continuously film the testing as an aid to understanding any unexpected event.

    1. Chase planes are usually only used during first flight and during Initial Air Worthiness testing. Since this program has been flying for at least 3 years, it was likely considered an unnecessary cost for “routine” envelope expansion testing. The FDRs would give a much clearer picture if the airframe was properly instrumented.

  2. I find the conclusions of this report to be incomplete, at best, and perhaps incorrect. For the recommendations related to the control system, they seem to peg all the blame on the lat stick/roll rate to DCP path. However, if that compensation was legitimately there to cancel inherent proverse yaw tendencies, then the compensation itself should not have led to the sideslip departure. Maybe the argument is that the compensation is over-mixed, so fair enough, but that is not how it is stated in the report.

    If you look at the data, though, you’ll find that the phase of the oscillatory divergence in sideslip does not correspond to the DCP contributions due to lat stick. Instead, it seems that the QBAL contribution to DCP reduces (and ultimately reverses) the contributions of the normal yaw axis turn coordination. Look at t=26.0 – 27.0 in Figure 19. Lat Axis DCP is heading in the same direction as the Yaw axis DCP component. That means that this compensator is helping with turn coordination, but both of them are opposed by the enormous, low bandwidth, QBAL term. You’ll see a similar characteristic on the front half of each of the sideslip oscillatory half-cycles.

    Bottom line on the control system part: Structural Load Limiting, through that QBAL component, completely eliminated the contributions of the automated turn coordination. By the last oscillation, before things got really bad, the SLL actually CONTRIBUTED to the yaw divergence.

    The report also completely ignored the absence of a rudder in this aircraft’s basic design, and only briefly touched on the inevitable degradation of N-beta due to the smaller vertical stab. Changes to airplane mode flapping controllers, after a previous incident of blade strike on the wing? Unmentioned. Maybe this aircraft doesn’t even have a flapping controller in airplane mode. I’d love to see a fuller set of conclusions, ideally from the prime’s engineering team and not the investigatory board. Guessing we won’t see that in public, though.

  3. Did Bell Boeing V-22 Osprey undergo the same test regime, and if so what did they learn? I wonder if they share their learning experiences with the Italians, or was this test the first of it’s type ever undertaken? I guess it may have been because the flight simulator could not replicate this test.

    1. Several things to consider:

      -V-22 has an H-tail, rather than a T-tail
      -V-22 has rudders, while -609 does not
      -Aside from technical conferences, there isn’t much mechanism or incentive for cross-company technical transfer or lesson sharing. On the contrary, all of these things are proprietary information that companies actively prevent sharing with competitors.

Leave a Reply