The zero-knots perspective

At 15:26 local time on March 3, 2023, a Bombardier BD-100-1A10 (Challenger 300) taxied onto Runway 2 of the Dillant/Hopkins Airport (EEN) in Keene, New Hampshire. The flight was intended to return three passengers to Virginia’s Leesburg Executive Airport (JYO), from which they’d flown to Keene the day before. As the jet accelerated on its takeoff roll, the co-pilot saw that the right primary flight display (PFD) hadn’t registered increasing airspeed and called on the pilot to abort the takeoff, which he did.

After pulling onto the taxiway, the pilot shut down the left engine. The co-pilot climbed out and found that he hadn’t removed the cover from the right pitot probe after his exterior preflight. He removed the cover, confirmed there was no damage to the probe, and returned to the cockpit. The pilot restarted the left engine.

When he did, a cyan “Rudder Limiter Fault” advisory illuminated on the Crew Alerting System (CAS). This had happened on past flights when the airplane was cold-soaked, most recently just a few days earlier. Running the avionics stall test usually cleared the warning, but this time two attempts failed to do so. The cockpit voice recorder captured the co-pilot offering to call maintenance control, then asking, “Do you want to take off with a rudder limiter fault?” The pilot replied that “It’s advisory only,” and after a brief additional discussion they agreed to continue the flight.

They did not consult the Challenger’s go/no-go reference guide. Had they done so, they would have found the “Rudder Limiter Fault” advisory listed as a no-go item, with no relief provided under its minimum equipment list.

During the ensuing takeoff, the flight director’s pitch attitude command bars did not appear on the PFD after the pilot pushed the takeoff button. He chose to continue the takeoff. At about 80 knots, the co-pilot noticed that the V-speed bugs hadn’t appeared on the airspeed indicator. Relying on memory, he called out V1 at 116 knots, then “rotate.” The jet lifted off at 15:35:27. The autopilot was engaged for their on-course turn and initial climb to 6,000 feet.

Both pilots later recalled that multiple additional cautions illuminated on the CAS during the climb, including “Mach Trim Fail,” “AP Stab Trim Fail,” and “AP Holding Nose Down.” They did not remember the exact order in which they appeared or whether any other messages accompanied them.

Forty-four seconds after takeoff the flight was cleared to FL230. The cockpit voice recorder captured the co-pilot asking the pilot whether he’d prefer a lower altitude. The pilot replied, “No, get the checklist,” without specifying which one. The aural warning of an autopilot disconnect sounded at 15:36:52. The co-pilot asked whether it had failed or been disconnected, and the pilot replied, “I did that.” The flight data recorder showed that he reengaged the autopilot twice more during the climb, each time triggering multiple caution messages on the CAS, while each disconnect was followed by manual adjustment of the horizontal stabilizer trim.

At 15:37:58 the co-pilot said, “I’d just leave the autopilot off” and the pilot replied, “All right….Get the checklist going.” They spent four more minutes trying to reprogram the flight management system (FMS). Eventually the co-pilot said “OK, there we go…they took those,” and the pilot again asked him to “run the checklist.”

The co-pilot brought up the Primary Stabilizer Trim Failure checklist: “the only trim fail checklist in the quick reference, and it seemed to be the root cause of our problem,” he later said. Its only item was turning the stabilizer trim switch on the center console from “Pri” to “Off,” thereby also disconnecting the autopilot. As soon as he called that out and flipped the switch, the Challenger pitched up hard.

The flight data recorder logged a 4-G acceleration as it reached an attitude of 11 degrees nose-up in just one second. The pilot pushed the control column forward with at least 90 pounds of force, according to flight data recorder information, registering a downward acceleration of minus 2.3 G in the jet’s return to a near-level attitude. Once he eased the forward pressure it pitched up again, logging another 4-G acceleration and activating the stall protection system as it reached 20 degrees nose-up.

The last jolt tripped the flight data recorder’s inertial switch, ending the recording; the full extent of the second pitch excursion remains unknown. The cockpit voice recorder recorded multiple stall warnings in the first four seconds after the trim was disconnected. The pilot later told investigators that he “thought it was the most rapid experience I’ve ever had.”

After the pilot regained control using manual trim, a passenger asked, “Can you guys get it on the ground?” Another passenger who had not been buckled in had suffered a serious head injury. (The seat belt sign had remained on throughout.) The crew advised ATC and made an emergency diversion to Bradley International Airport (BDL) in Windsor Locks, Connecticut, landing 17 minutes after the upset. An ambulance transported the passenger to a hospital, where she died of her injuries later that day.

On March 16, 2023, representatives of the NTSB and the unit’s manufacturer downloaded the non-volatile memory from the Challenger’s horizontal stabilizer electronic control unit (HSTECU). It revealed that the mismatch between airspeed readings during the rejected takeoff registered a fault code for disagreement between the two air data computers. This in turn triggered the “Rudder Limiter Fault,” sending that message to the CAS, and also set “Confirmed Mach Valid” to False, disabling Mach trim.

The HSTECU also disabled the stabilizer’s autopilot trim function, producing the “AP Stab Trim Fail” message and requiring the autopilot to compensate for the out-of-trim condition, indicated by the “AP Holding Nose Down” warning. By design, the “Mach Trim Fail” alert on the CAS was inhibited until the weight-on-wheels sensor detected that the jet was airborne and its radar altitude reached 400 feet.

The data did not indicate any malfunction of the HSTECU logic or circuitry or any combination of faults that would have produced a “Pri Stab Trim Fail” warning. The “AP Stab Trim Fail” and “AP Holding Nose Down” checklists both warned that “An abrupt change in control force or an out-of-trim situation may be experienced when disconnecting the autopilot.” Company maintenance control would have advised trying to clear the faults by powering the airplane down and starting it up again. Had that not worked, they’d have grounded it until maintenance staff could troubleshoot it in person.

While interviewing the pilot, the investigator-in-charge acknowledged the advantage of “sitting here at zero knots at a desk two weeks after.” Sitting at zero knots would seem to have been the opportune time to definitively resolve that initial warning even if they had seen it before. Once in the air, things happened very fast.