The NTSB has released the final report for the Nov. 18, 2022, in-flight breakup of a Cessna 208B Grand Caravan EX single-engine turboprop during flight testing for a new supplemental type certificate program near Snohomish, Washington. All four occupants of the Caravan were killed.

This flight test was part of a series to test a Raisbeck Engineering drag-reduction system (DRS) on the Caravan EX, to expand the applicability of the DRS to another modification on the airplane, the AeroAcoustics Aircraft Systems Aircraft Payload Extender (APE III), although the DRS was not installed on the accident aircraft.

One day prior to the accident flight, the test pilot, another pilot, and two aft-seated testing personnel were conducting test flights to gather baseline data. That day, two flights were planned to test aft c.g. stability, but the second flight (07) had to be stopped before completion of the test card due to the airsickness of one of the crewmembers.

Completing that test card for baseline testing of the aft c.g. characteristics with the APE III kit installed was the purpose of the accident flight. During the previous day’s last flight (07), a test involving an accelerated stall with full flaps while in a 30-degree left bank at idle power resulted in a left roll to 83 degrees, as captured by onboard video. In the recovery, the Caravan exceeded the maximum operating speed (Vmo), and the 83-degree roll was also an exceedance over the allowable 60-degree roll limit of that maneuver, according to the NTSB.

Although the data acquisition system captured the roll exceedance, the crew didn’t identify that parameter as having failed the test. Nor did the crew “remark upon the exceedance” of Vmo, which would have required an overspeed inspection per the Caravan’s maintenance manual.

“In accordance with risk mitigation procedures for the test plan,” the NTSB report noted, “the test buildup should have been stopped after roll limits were exceeded in order to determine the reasons for the exceedance and to implement corrective actions before proceeding with higher-risk conditions in the test plan. Secondly, after exceeding Vmo, the crew did not remark upon the exceedance, and even though the exceedance met the requirements for an overspeed inspection as described in the airplane’s maintenance manual, there was no indication that this inspection was completed.”

Audio and video recordings of this flight captured the test pilot and engineer discussing the roll and apparently deciding that it wasn’t an exceedance:

Engineer: “And were we more than sixty on that roll-off?”

Pilot: “Uh…more than sixty…ah let me think…No. We were probably about fifty.”

Engineer: “‘K.”

Pilot: “So technically I guess that’s good.”

Engineer: “It’s a pass.”

Pilot: “That’s a pass. So we’re going back up to try one to the right.”

Engineer: “Correct.”

Pilot: “Glad you mentioned it I was getting ready to reject it. Just ‘cause it rolled and I couldn’t stop it, but…you know, pushed out…it was…it was still within.”

According to the NTSB, the Caravan was equipped with Garmin G1000 NXi avionics that included Garmin’s GFC 700 autopilot and Electronic Stability and Protection (ESP) system, “which was designed to deter attitude and airspeed exceedances during hand-flying and maintain stable flight by applying an opposite force to the direction of predetermined travel. It was designed to provide a light force that can be overcome by the pilot.”

ESP can be disabled using a page on the primary flight display or by holding the control-wheel steering button or autopilot disconnect button on the yoke. “Although the accident pilot was an experienced test pilot and qualified to operate the airplane,” the report noted, “his experience with the accident airplane’s avionics system could not be determined. Videos of his previous flights in the airplane suggested that he was unfamiliar with the ESP system, as he did not deactivate it before the flight nor discuss the forces it was applying during the flight.”

An accident flight simulation was conducted by the NTSB. “The accident flight simulation indicated that, during the stall immediately preceding the accident, it is likely that the ESP activated as the airplane pitched in excess of 19 degrees nose-up. This would have required the pilot to apply more aft force on the control column in order to induce the stall. After the stall, the ESP would have activated at 45-degree bank, then deactivated as the airplane quickly exceeded 75 degrees. The extent to which the control forces from the ESP, or the potential distraction due to the system’s engagement and disengagement, may have contributed to the pilot’s failure to recover from the nose-low attitude following the stall could not be determined.”

For these flights, the Raisbeck test plan included test hazard analysis (THA) worksheets “that identified hazards, their causes and effects, mitigation measures, and emergency procedures,” according to the NTSB. “THA 9.5, which focused on aft c.g. stall characteristics, identified ‘departure from controlled flight’ as a hazard caused by ‘unpredicted aerodynamic response’ or ‘improper control inputs,’ with the effect being a significant altitude loss leading to ground impact.

The risk for THA 9.5 was assessed as ‘medium,’ but it did not mention mitigations for overspeed conditions or the inclusion of some mitigations listed in NASA’s Flight Test Safety Database (FTSD) for similar testing, such as terminating tests when roll angle limits are exceeded. The FTSD assessment for stall characteristics testing, labeled THA 56, assigned a ‘high’ risk level and included mitigations absent in THA 9.5, such as ensuring all stalls are coordinated, retarding throttles to idle during departures from controlled flight, and halting testing if roll angle limits are exceeded.”

During the accident flight, based on ADS-B data, after departing Renton Airport, the Caravan performed a series of maneuvers and about 50 minutes after takeoff was at 10,000 feet and 110 knots when the final maneuver began. After climbing about 100 feet and slowing to between 90 and 105 knots, the Caravan rolled left to 30 degrees with altitude climbing at about 400 fpm and the airplane descending from 9,700 to 9,350 feet.

Airspeed was constant at about 105 knots with engine torque consistent at about 930 pound-feet. Over a minute later, airspeed dropped rapidly and reached a minimum of 48 knots, then sharply increased and engine torque climbed to 2,200 pound-feet more than a minute later.

Data captured by the Caravan’s Pratt & Whitney Canada engine monitoring system showed that indicated airspeed slowed to 35 knots, then 37 knots before rising to a maximum of 223 knots and dropping to 80 knots. Altitude varied between 9,320 to 9,680 feet and vertical speed went from plus 2,560 fpm to minus 14,000 fpm.

According to the NTSB, “At 1019:05, the ADS-B data indicated a sudden and tight course reversal from east to west, corresponding with the minimum airspeed and a dramatic increase in the descent rate. The computed roll angle stayed consistent at about 30 degrees left until the course reversal. The ADS-B data stopped at 1019:17 about 7,025 ft msl with a recorded descent rate of 8,700 fpm.

“Witnesses reported that they observed the airplane break up in flight and watched pieces floating down. The airplane then descended in a nose-low, near-vertical corkscrew maneuver toward the ground. Several witnesses reported seeing a white plume of smoke when they observed the airplane break into pieces.

“The airplane’s flight test data acquisition system, used as part of the flight test program, was destroyed in the accident and no flight test data for the accident flight was recovered.

“The airplane’s maneuvers and speeds leading up to the minimum airspeed recorded at 1019:01 suggests the performance of an intentional stall in a 30-degree left roll with the engine power above idle (at about 930 ft-lb of torque), consistent with the stated intent of the flight and the items remaining on the flight test card for that flight (including a power-on stall in a 30-degree bank).”

Probable cause of the accident was “The pilot’s improper recovery following a departure from controlled flight after an intentional aerodynamic stall, which resulted in an exceedance of airspeed limitations, airframe overstress, and a subsequent in-flight breakup.”