Cyclone crash report details events that led to fatal flight control problem
By Chris Thatcher | June 29, 2021
Estimated reading time 10 minutes, 9 seconds.
A previously unidentified software problem related to the automatic flight control system (AFCS) was among the leading causes of a tragic accident that killed four crew members and two passengers of a Royal Canadian Air Force (RCAF) CH-148 Cyclone conducting a routine surface surveillance mission from HMCS Fredericton in the Ionian Sea on April 29, 2020.
The maritime helicopter, callsign Stalker 22, was completing a low altitude maneuver astern of the Royal Canadian Navy frigate as part of Operation Reassurance, Canada’s contribution to NATO, when a phenomenon created by manually overriding the flight director resulted in a high energy descent into the water.
According to the Flight Safety Investigation Report, completed in May and released publicly on June 28, the Cyclone’s AFCS “contained a fly-by-wire flight control law that created a Command Model Attitude Bias Phenomenon — an aircraft objectionable behavior characteristic — that resulted in the pilots’ inability to effect a restorative pitch correction, while overriding the engaged airspeed/pitch axis flight director mode.”
“This phenomenon develops under a very specific and narrow set of circumstances where manual inputs to the primary flight controls override the aircraft’s automation system, referred to as the flight director [FD], while it is engaged and set to fly at a fixed airspeed or pitch attitude,” the Department of National Defence explained in a statement. “The bias that developed in this instance resulted in insufficient aft cyclic controller authority, resulting in a high-energy descent and impact with the water.”
The investigation also found that inadequate crew resource management (CRM), confusing or insufficient information in the aircraft’s standard maneuver manual, and crew training on “how the levels of automation function” all contributed to the tragedy.
“This accident was not the result of one single causal factor but a combination of several,” BGen John Alexander, Director of Flight Safety and the RCAF’s airworthiness investigative authority, wrote in the final report.
The accident claimed the lives of MCpl Matthew Cousins, SLt Abbigail Cowbrough, Capt Kevin Hagen, Capt Brenden MacDonald, Capt Maxime Miron-Morin, and SLt Matthew Pyke.
The crash occurred during the second of three missions planned that day for the ship’s helicopter air detachment. Around 2 p.m. local time, the first aircrew conducted a routine surface surveillance mission that concluded with hoisting of passengers to the ship’s flight deck.
A second aircrew then launched Stalker 22 around 5:30 p.m. Following area surface surveillance, the crew opted to demonstrate a right turn return-to-target (RTT) maneuver for its two passengers, during which the pilot flying the aircraft disengaged the radar altitude hold (RALT) and indicated airspeed hold (IAS) modes from the flight director. The modes were re-engaged upon completion of the maneuver.
More than an hour later, upon return to the Fredericton, Stalker 22 conducted a low altitude pass of the ship from stern to bow on the port side, completed a left 180-degree level turn, and repeated the low pass, again along the port side, from bow to stern. The maneuver was “flown manually by the same pilot at the controls as from the earlier demonstration RTT,” report noted.
Article Continues Below
Once astern of the ship at about 50 feet above ground level (AGL) and 130 knots, the pilot began to execute a left RTT by making an abrupt aft cyclic input. The aircraft began a rapid climb to approximately 410 feet AGL, decelerating to an indicated airspeed of 85 knots. During the climb, the pilot initiated a left turn with cyclic and full left pedal, resulting in a steep bank and high yaw rate to the left.
“At this point, the aircraft began a pitch down to a nose-low accelerating attitude at the same time the cyclic was pulled to the aft stop and torque significantly increased,” the report stated. “Prior to the pitch reversal neither the RALT nor IAS FD functions were deselected by the crew, nor did it disengage automatically via the flight control software logic as the aircraft maintained above 60 knots throughout the maneuver.”
As the aircraft rolled wings level in a nose-down attitude, it continued to accelerate until it impacted the water. The series of events occurred in just 14 seconds.
“The investigation could not determine why the crew did not decouple the FD during the accident maneuver as they had during the first RTT maneuver,” the report said. In an instance of what investigators described as “ineffective CRM,” members of the aircrew did not brief the RTT beforehand and did not verbalize or challenge the manipulation of the primary flight controls “to override the engaged airspeed/pitch flight director modes during the final maneuver.”
“One of the last lines of defense to any accident is the effective use of [CRM]. Open communications where the pilot flying expresses their intent to the crew and the confidence of the other crew members to challenge or question actions that have not been briefed are essential,” Alexander wrote in his summation.
“The crew of Stalker 22 were not communicating most of the required intra-crew calls. This investigation could not, therefore, determine whether all crew members knew the intent of the pilot flying or the level of automation selected. With greater automation on modern aircraft comes a greater need for effective crew resource management to ensure all members understand what has been asked of the automation in order to effectively monitor systems to ensure they are responding appropriately.”
Alexander noted that as the Air Force transitions to new fleets with greater capacity for automation, designers and test authorities must “assume that aircrew will attempt to fly the newer aircraft in the same manner as they did with the legacy aircraft. If this is the starting point for design and testing, then flight testing can certify the full range of foreseeable maneuvers or, at the very least, offer the appropriate cautions and warnings to operators to preclude them from attempting to fly a profile for which the aircraft is not intended.”
The investigators found that RTTs are common practice within the maritime helicopter community, as are manual overrides during flight director coupled operations, even though there are no official standard operating procedures for the maneuver.
According to the report, the Cyclone’s manufacturer, Sikorsky, as well as airworthiness authorities and aircraft operators “were unaware of the Command Model Attitude Bias Phenomenon prior to the accident,” and the aircraft’s publications “contained information that may have been confusing or misleading to operators, contained numerous sections on automation considerations that were incomplete, and lacked maneuver descriptions that would have been required for operational use.”
Article Continues Below
Among the report’s recommendations are amendments to the publications around automation operating intent and employment strategies, as well as “modification to software embedded within the electronic flight control laws.”
The investigation also suggested establishing a working group to review the original Maritime Helicopter Project statement of operating intent, and “to determine the operational requirement of the return-to-target, and similar maneuvers, in the CH-148.”