Classic Aftermath: An Attitude Indicator Fails at the Worst Time

The trip took a little more than three hours. By the time they were nearing Nanaimo, it was dark, and the airport was reporting a 400-foot ceiling and 2.5 miles visibility in light drizzle and mist. The pilot told the controller that he would be making the Runway 16 ILS approach.

A few minutes later, the pilot asked the controller for weather at Vancouver International Airport, opposite Nanaimo on the mainland side of the Georgia Strait. Vancouver was better: 5 miles in mist, 600 broken, 1,200 overcast. The controller also passed on a pilot report from an airplane that had landed at Nanaimo 15 minutes earlier; the pilot had seen the approach lights at minimums, 373 feet above the runway elevation.

At 1803 PST, the controller vectoring the Aerostar observed that it had flown through the localizer and was continuing past it on a southwesterly heading. It was then at 2,100 feet and 140 knots. Aware of high terrain to the southwest, the controller asked the pilot whether he still intended to intercept the localizer; the pilot replied that he did and momentarily lined up before again drifting off to the right.

At 1804:03, the pilot told the controller that he “just had a fail” and requested vectors. The controller initially instructed the pilot to make a “tight” left turn to 090 and then, when the pilot asked the controller to repeat the instruction, changed it to a right turn to 360. The pilot acknowledged the heading but continued past it. The airplane climbed to 2,500 feet and slowed to a groundspeed of 80 knots before descending to 1,800 feet and accelerating to 160 knots.

At 1804:40, the pilot reported that he had lost his attitude indicator. The Aerostar was now climbing again and turning to the right. The pilot requested a heading from the controller, who again gave 360.

The Aerostar reached 2,700 feet and slowed to a groundspeed of 60 knots; even taking prevailing winds into account, the airplane was very close to stalling speed. It continued in a right turn and again began to lose altitude. The controller told the pilot to climb if he could, but the pilot did not respond. The last Mode C return came at 1805:26; the airplane was near the point at which it had originally crossed the localizer, traveling northeastward at 120 knots and 300 feet above the surface. Moments later, as the accident report puts it, “Control was lost.” The airplane crashed on Gabriola Island, just offshore from Nanaimo. All three occupants died. A witness cited in press accounts reported the airplane spiraling down, but that information, suggestive of a stall and spin, did not find its way into the Transportation Safety Board of Canada’s final accident report.

The Aerostar was equipped with a pressure-driven BendixKing KI 256 attitude indicator and dual pneumatic pumps. All were badly damaged in the crash, and the reason for the reported failure could not be determined.

The airplane also had a BendixKing KI 825 horizontal situation indicator. It had briefly failed twice in the past three weeks and was scheduled for repair. The HSI was electrically driven, however, and the chance that it and the pneumatic attitude indicator would fail simultaneously is remote.

The occupants of the front seats were both pilots. The airplane’s owner—a 13,000-hour ATP and instructor with extensive experience in airplanes, helicopters and sailplanes—was in the right seat. In the left was a 320-hour private pilot without an instrument rating who had logged 11 hours of night flying and 29 hours of instrument training under the hood. The accident report says that “the investigation could not determine who was flying the aircraft.” It’s hard to imagine that the right-seat pilot would not have taken at least partial control after the failure of the attitude indicator, although the erratic excursions of speed, heading and altitude suggest otherwise.

Unlike our NTSB, Canada’s TSB does not attempt to determine a “probable cause.” The accident report is very sketchy, omitting (among other useful information) witness accounts, a description of the accident site and toxicological results. It does offer up some boilerplate about “high cognitive workload conditions” and “perceptual bias”—fancy ways of saying “overwhelmed and confused”—before concluding that after losing the gyro horizon, pilots are reduced to depending on “the remaining cockpit displays, communication with other flight crew members, and their own perceptions of motion and orientation.”

One’s own perceptions of motion and orientation inevitably lead to disorientation and loss of control, so you don’t want to depend on them. You have to rely solely on the remaining instruments. Altimeter, vertical speed indicator and airspeed indicator become primary for pitch attitude; look to the turn coordinator and the directional gyro, if it’s still working, for turn rate. If the air is not excessively turbulent, as was most likely the case on the night of the accident, the airplane will maintain its trimmed pitch attitude of its own accord; but the pilot may complicate matters by unconsciously adding a push or pull to roll inputs.

Loss of the AI presents a difficult challenge. In the training and check-ride environment, the challenge is of brief duration and usually focuses on recovery from an unusual attitude. In fact, the right-seat pilot had successfully completed such a check ride a few months earlier. Experience has shown, however, that IFR flying for an extended time by reference to the backup instruments sounds easy but isn’t. One thing that may eventually confuse pilots is that the turn coordinator looks like an AI, but what the behavior of the little airplane actually reflects is a combination of roll and yaw rates. (The now-outmoded turn-and-slip indicator registered only yaw rate—that is, heading change—and had more in common with a DG than with an AI.)

The accident report says nothing about the arrangement of the Aerostar’s instrument panel, but it is often the case that the turn coordinator is placed at the lower left corner, where it would have been hard for the pilot in the right seat to see. Even so, it’s difficult to understand what happened in the final moments. The persistent—and eventually fatal—loss of altitude is particularly baffling in an airplane whose engines, as far as anyone knows, were working fine.

I assume that the airplane owner had put his novice passenger in the left seat to give him some experience with real-world night instrument flying. He probably felt that he could intervene at any time and, if need be, could fly the familiar ILS approach himself from the right seat. The possibility that an attitude indicator failure could occur then, of all times, probably never crossed his mind.

Nor does it occur of most of us, I suspect, to cover the AI (and the DG, if both use the same power source) from time to time and fly a complete ILS approach, including a miss, on partial panel. I’ve never done it—have you?

The most common cause of gyro trouble is vacuum-pump failure. Presumably, most vacuum-pump failures occur in VMC—because more flying is done in VMC than in IMC—but even under the best of circumstances, an equipment failure may occasionally have dire consequences. The NTSB attributed the 2018 crash of a Mooney during a landing approach in VFR conditions to a vacuum-pump failure that distracted the pilot. Equally surprising was the 1999 crash near Albuquerque, New Mexico, of a P210N with three aboard. The airplane was cruising in IMC at FL 220 when both vacuum pumps failed. Despite the airplane having an electrically driven standby attitude indicator, the 42-year-old 1,300-hour pilot lost control while descending—cloud bases were at 13,000 feet—and the airplane broke up in flight. Never underestimate the power of the unexpected.

This story appeared in the December 2020 issue of Flying Magazine