Air Canada Flight 624

Waiting for a landing opportunity in Halifax

Air Canada flight 624 left Toronto for Halifax on the evening of 28 March 2015. The Airbus A320-211, registered C-FTJP, carried 133 passengers and five crew members. The captain flew the aircraft from the left seat while the first officer monitored from the right. They were experienced A320 pilots, but this was their first flight together. Before departure, the forecast for Halifax included snow, drifting snow, gusting wind and visibility of half a statute mile.

The crew selected a localizer approach to runway 05. The other runway was unavailable because of snow accumulation. A localizer gives lateral guidance towards the runway centreline, but this approach did not provide the vertical glidepath guidance associated with a precision instrument landing system. The pilots would manage the descent using a selected flight path angle, then complete the landing visually if sufficient references became available.

Reports during the flight showed worsening visibility, below the half-mile threshold needed to continue beyond the final approach fix under Air Canada's authorization. The pilots arranged to hold near Halifax while waiting for an improvement. They also considered the fuel available for holding and an eventual diversion to Moncton. At one stage, they discussed diverting if the weather failed to improve within approximately twenty minutes.

At 00:16 Atlantic Daylight Time on 29 March, they received a special observation reporting half-mile visibility in snow and drifting snow. The improvement met the applicable minimum, so they left the hold and began the approach. The decision was made in weather that remained difficult: it was dark, the airport was snow-covered and the available forward view was limited.

Preparing a selected descent

During cruise, the pilots had calculated cold-temperature corrections for the approach altitudes. They planned to cross the final approach fix at 2,200 feet above sea level and use a minimum descent altitude of 813 feet. Their calculated descent angle was 3.5 degrees downward. The approach was planned with autopilot guidance followed by a manual landing.

As flight 624 approached Halifax, the crew asked for the runway lights to be at their brightest setting, setting five. The controller said the lights were then on setting four but would be increased in time for landing. In fact, the lights remained at setting four during the approach. The TSB later identified both that unfulfilled expectation and a discrepancy in the preset lighting configuration.

The aircraft intercepted the localizer and was fully configured for landing before the final approach fix. The gear was extended and the flaps were set fully down. Autopilot and autothrust remained engaged. The crew selected flight path angle mode, initially at zero degrees while level, then selected minus 3.5 degrees approximately 0.3 nautical miles before the final approach fix.

The descent began before the fix, which the aircraft crossed at an indicated altitude of approximately 2,170 feet. The autopilot followed the selected angle. However, selecting an angle was not equivalent to capturing a fixed glidepath tied to the runway. The subsequent investigation found that disturbances could shift the actual descent profile and that pilots needed to monitor altitude against distance to identify and correct such a deviation. That check was absent from the way this approach was flown.

The descent profile moves below the intended path

Air Canada's established practice for flight path angle approaches did not require crews, once beyond the final approach fix, to monitor altitude against distance from the threshold or adjust the selected angle. The TSB found that this practice was inconsistent with the operating manuals of both Air Canada and Airbus. Following that practice, the crew did not notice that flight 624 had drifted below and diverged from the intended vertical descent profile.

The aircraft therefore reached the minimum descent altitude farther from the runway than intended. From the cockpit, the approach was still considered stable and consistent with the training the pilots had received. The lateral guidance remained closely aligned with the localizer. Correct alignment with the runway centreline did not establish that the aircraft was at the right height for its distance.

The visual transition was difficult. Blowing snow could momentarily or continuously hide approach lights, runway lights and the precision approach path indicators. There was little surrounding lighting along the final approach path to help the pilots judge height and motion. When the monitoring pilot reported seeing lights only, the aircraft was approximately one nautical mile from the threshold and about 0.35 nautical miles from a lighted facility on the approach path.

The TSB could not determine the exact visibility from the cockpit throughout the final descent. Its analysis considered the cues that might have been available rather than claiming a complete reconstruction of everything the pilots saw. It found that limited and short-lived visual references, possible visual illusions and the dimmer-than-expected lights diminished the crew's ability to recognise that their flight path would bring them to the ground before the runway.

Recognition comes just before impact

Near the end of the approach, both pilots confirmed that they could see some approach lights. The aircraft passed over the lighted facility while continuing towards runway 05. The landing lights, which had been switched off during the approach, were switched on at 00:29:47. The pilot flying disconnected the autopilot, followed rapidly by the automated height calls and the monitoring pilot's instruction to pull up.

The aircraft severed power lines running across the approach path. Those lines supplied utility power to the airport, so the accident also created a power outage that would complicate the response on the ground. About a second before the first ground impact, the captain moved the thrust levers to the takeoff or go-around position and pulled the sidestick fully nose-up.

That response came too late to prevent contact. A left main wheel hit an approach light, and at approximately 00:30 the main landing gear, lower rear fuselage and left engine cowling struck snow-covered ground on the slope before the runway. The report's overall findings place the terrain collision approximately 740 feet short of the threshold. The aircraft then struck the localizer antenna array, remained or became airborne through parts of the sequence, hit the ground twice more and slid along the runway.

It stopped on the left side, approximately 1,900 feet beyond the threshold. The aircraft had been destroyed, but there was no post-impact fire. The entire event had developed from a descent that was too low for the aircraft's distance from the runway. The damaging first contacts occurred before the threshold, rather than after a touchdown on the intended pavement.

Leaving the damaged aircraft

Electrical power was lost during the impact sequence. The cabin lights went out and the emergency lights came on. The tower activated the crash alarm at 00:30:16. There had been no warning of an imminent crash to allow passengers and cabin crew to adopt a brace position before the first impact.

No general evacuation order was given, but passengers near rows seventeen and eighteen opened the four overwing exits and began leaving. The service director opened the forward left door and directed passengers out. The relevant slides deployed. Firefighters arrived approximately two minutes after the crash alarm, while the evacuation was still being completed. All passengers had left the aircraft within five minutes of it stopping.

The report recorded twenty-five injured occupants: one crew member seriously injured, four crew members with minor injuries and twenty passengers with minor injuries. No one died. The cockpit occupants struck the glare shield during the impact sequence. The first officer's shoulder harness inertia reel had an unserviceable automatic locking feature, contributing to his head and serious eye injury. A flight attendant was injured by a coffee brewer that detached because its mounting lock was not correctly engaged.

Those details led the investigation beyond the approach itself into survivability and equipment maintenance. The report also noted that some passengers took carry-on baggage during the evacuation. Its discussion of that behaviour addresses the danger of delaying or obstructing escape; it does not claim that retrieving bags caused this accident. The documented successful evacuation was followed by a further problem: the occupants were outside in harsh winter weather without immediate transport to shelter.

A delayed journey from the runway to shelter

The evacuated occupants gathered approximately two hundred metres behind the aircraft. Many were dressed in shorts, T-shirts or open-toed shoes, suitable for their travel plans but inadequate for standing outside in snow and wind. People with more severe injuries were placed in emergency vehicles. Around 00:42, firefighters confirmed that everyone had evacuated and requested transportation to shelter.

The accident's power outage affected more than terminal lighting. The terminal standby generators failed to sustain power, and the airport operations radio network became unavailable because a network switch was not backed by an uninterruptible power supply. Staff had to coordinate vehicles through individual telephone calls rather than a collective radio broadcast. Airport access gates also required manual opening.

Both Air Canada's and the airport authority's emergency plans assigned passenger transportation responsibility to the airline, but neither identified the airport's Park'N Fly minibuses as available resources for this purpose. The airport had considered using those buses in exercises, yet the transport sequence had not been practised. Arrangements to dispatch them began only after the all-clear message from the firefighters.

Snow-covered roads, poor visibility, confused location information and disrupted communications added delay. Some vehicles initially went to the wrong part of the airport and had to be redirected. Approximately fifty minutes after the aircraft stopped, the remaining passengers had been taken indoors. The TSB separated the rapid arrival of the fire service from the slower provision of transport: a response could meet firefighting requirements and still leave evacuated occupants exposed while shelter arrangements were made.

The investigation identifies approach and response problems

The TSB found that the crew had followed Air Canada's flight path angle practice and did not monitor altitude against distance after the final approach fix. They failed to recognise the resulting deviation below the planned profile. The investigation also identified reliance on the autopilot beyond the minimum descent altitude, reduced lighting brightness, limited visual cues and the likely influence of plan continuation bias as contributing to the late recognition of the unsafe flight path.

Transport Canada had reviewed and approved the airline's manuals and procedures without identifying the discrepancy in vertical-path monitoring requirements. This extended the findings beyond the actions of the two pilots on one snowy night. The approach had been regarded as stable within the training and practice they followed, while the monitoring needed to establish the aircraft's true position relative to the intended profile was missing.

The report examined evacuation communications, passenger transport, restraint equipment and other safety risks. Its discussion of obstructive sleep apnea was a broader medical oversight concern, not an adopted finding that the condition caused the accident. The established sequence was a low approach in difficult visibility, late recognition and a terrain collision before the runway, followed by evacuation and a delayed transfer to shelter.