CEN22FA331 · Columbia Helicopters CH-47D

The public firefighting helicopter was equipped with a 2,600-gallon water bucket attached via a 200-ft long line and a belly cargo hook. On the day of the accident, the pilots had flown about 1.5 flight hours and dropped multiple bucket loads. A witness video of the accident revealed that the helicopter was setting up to dip the water bucket in the river when it ascended and subsequently began a left yaw. Once it had turned about 180° the helicopter suddenly pitched nose down. The left yaw rate increased as the helicopter rapidly descended until it impacted the river.

Postaccident examination of the wreckage revealed no evidence of malfunction or failure of the helicopter’s rotor systems, transmission and drive system, structures, and engines. The witness video corroborated that there was no structural failure before ground impact. Examination of the hydraulic system found no evidence of contamination, loss of fluid, or malfunction that would contribute to a loss of control inflight. Examination of the flight control system linkages showed no evidence of preimpact disconnection or separation. None of the hydraulic actuators of the flight control system exhibited functional hydraulic anomalies. The lee plugs that were found to be missing from the aft swiveling upper boost actuators (UBA) likely blew out from the actuator due to ground impact forces. The yaw integrated lower control actuator’s (ILCA) control rod and lever likely over-traveled upward due to ground impact, resulting in its impact with the yaw ILCA’s main control valve spool. While the jam indicators of multiple hydraulic actuators were extended, their extension was likely due to ground impact forces on the actuators rather than an actual jam of the control valves of multiple actuators during the accident flight. Therefore, no malfunction of the hydraulic system and flight control system was found.

Analysis of the witness video showed no rotor speed anomalies throughout the accident flight. The video also showed the helicopter initially in a coordinated left yaw about the vertical axis as it remained in a mostly level attitude through the 180° turn. Because each UBA’s input is a result of the mixed pitch, roll, yaw, and thrust outputs, more erratic helicopter attitudes would have been expected with anomalous behavior of the UBAs. Therefore, malfunction of the UBAs was unlikely in this accident.

A simulation study showed that the flight control inputs needed for the accident flight profile were not beyond the capability of a normally functioning flight control system. Additionally, the left pedal input was not at its forward limit stop until about 2 to 3 seconds after completion of the 180° left turn; therefore, it was likely that the initial 180° left turn was intentionally commanded by the pilot flying. Accordingly, investigators explored factors that could result in the flight crew’s inability to arrest the left yaw with the absence of a flight control system malfunction or failure.

An iPad used by the flight crew was discovered in the river near the wreckage. The iPad displayed three distinct gouges on its long side and a bend from the back side. Testing with an exemplar helicopter and the iPad showed that it was possible to wedge the iPad between the airframe and the copilot’s pedals in a manner that would produce three distinct gouges corresponding to parts of the airframe and the copilot’s pedal heel slide. Additionally, as force was applied to the right pedal to bring the pedals to neutral, the jammed iPad would press against the copilot’s left pedal adjustment lever, which could result in unintended readjustment of the copilot’s left pedal position, and the iPad would remain jammed. In the wreckage, the copilot’s left pedal was found in the forward-most adjustment position, which was unusual given his height. Testing also found that, when the left pedal was pushed forward, the iPad became unjammed but could migrate further down between the left pedal and the airframe while still preventing the left pedal’s return back to its neutral position. Lastly, the position of the yaw magnetic brake clevis connection was found in a position consistent with the left pedal (in the cockpit controls) being forward of neutral.

Based on the aggregate findings, it is likely that during the 180° left turn, the unsecured iPad migrated into the copilot’s left pedal and became jammed between his left pedal and the airframe, restricting the pedals from moving back to neutral. Testing showed that it would be difficult for the copilot to reach down to move the iPad, and possible for the copilot to bump the cyclic control forward when attempting to reach down, which would result in a sudden nose-down pitch, as seen in the accident video. The helicopter descent continued, likely due to a commanded reduction in the thrust lever. The reduction of thrust from both rotor systems would alleviate the increasing yaw rate as the left pedal could not be returned to neutral. The pilot likely pushed the left pedal further forward to assist the copilot in removing the jammed iPad, but the iPad likely migrated further down between the left pedal and the airframe, putting it further out of reach from the copilot while significantly increasing the helicopter’s left yaw rate.

The operator stated that the flight crew’s use of the iPad was not required during the firefighting flight and that the copilot should have been monitoring the cargo hook load gauge during water pickups. However, it is plausible that the copilot was using the iPad before the water pickup and was holding onto it, allowing for its migration into the pedals during the accident flight. While the operator had a portable electronic device (PED) policy that prohibited the use of cell phones during an operation, it was ambiguous as to whether that policy extended to use of company-issued iPads. Regardless, the security of all items in the cockpit during an operation is necessary to ensure they do not adversely interact with critical systems, such as flight controls.

Verbatim NTSB analysis and probable cause from the NTSB dataset