On 7 October 2014, an Airbus A330-300 (C-GFAF) being operated by Air Canada on a scheduled passenger flight from Frankfurt to Montréal as ACA 875 had flown a daylight ILS approach. After gaining the required visual reference to continue the approach, forward visibility was reduced in a sudden heavy rain shower and the aircraft briefly departed the side of the runway immediately after touchdown. It then regained the runway centreline before completing the landing roll and subsequently taxiing to its assigned parking gate at the terminal. A number of runway edge lights were broken but the aircraft was undamaged except for tyre cuts sufficient to require two wheel changes caused by destructive impact with the edge lights.
An Investigation was carried out by the Canadian TSB. Relevant data was successfully recovered from both the 2 hour CVR and the DFDR. The latter was linked to recorded ADSE data and subsequently the FDR data were synchronised with the CVR data, ATC communication recordings and measurements of marks on and immediately adjacent to the runway. However, contrary to regulatory requirements and the guidelines of the airport operator, Aeroports de Montréal (ADM), their personnel "cleaned the runway and repaired the damaged lights without first consulting TSB investigators". It was noted that "cleaning the runway using jets of water and mechanical brooms erased marks on the runway, thereby depriving investigators of information about the runway excursion (and that) since the quality and quantity of clues lost or altered could not be determined, it was not possible to assess the significance of this (lost) information".
It was found that the Captain had accumulated 17,259 total flying hours which included 427 hours on type and a total of 8,524 hours on "Airbus-type aircraft". The First Officer, who was PF for the investigated flight, had accumulated 8,262 flying hours which included 628 hours on type with no other previous experience on "Airbus type aircraft". They were operating on a single sector duty and the Investigation found no evidence that either pilot was fatigued, although each had taken controlled rest on the flight deck for periods of up to 45 minutes en route.
It was established that during the initial stages of the descent to destination a little under an hour out, the ATIS, which gave VMC and made no mention of convective weather, had been copied. At about 30 nm out in VMC, the crew advised ATC that they had copied the updated ATIS which gave 24R as the only runway in use and did include reference to the presence of Cb whilst also reporting good visibility, light surface winds and no low cloud with no precipitation. When 18 nm north of the airport at approximately 5,000 feet QNH, the crew noted "rather dark conditions north of the airport" which they reported corresponded to the position of strong returns on the aircraft weather radar, although the airport including the landing runway was still clearly visible. As briefed earlier, a Category 1 ILS approach was made to runway 24R and a landing in VMC was "anticipated" with just a very light crosswind from the left. The progress of the obviously poor weather conditions now to the north of the landing runway were visually monitored and the possibility that precipitation and wind shear might be encountered was considered, although not the possibility of a consequent go-around. It was decided that the approach speed would be slightly increased and FLAPS 3 was confirmed as the intended landing configuration.
As the aircraft passed approximately 1,900 feet QNH, ATC gave the spot wind as 280° at 13-18 knots and cleared the aircraft to land, reminding the crew a few seconds later that the runway lighting was out of service as NOTAMed. This latter status required a 50 feet addition to DH and a 400 metres addition to the minimum RVR which had been done but neither addition subsequently prevented the approach continuing to DH or the acquisition of the required visual reference to continue to a landing. The First Officer disconnected the AP passing approximately 900 feet agl and the approach remained stabilised. Passing 600 feet agl, a DHC-8 that had just landed reported experiencing "light wind shear below 400 feet" and in response to a request for a wind check, ATC advised that it was now 300° at 18-24 knots. Still flying with full visual reference, the crew reported having observed a rain shower approaching the middle of the landing runway but still being able to clearly see the entire length of it.
Then at approximately 130 feet agl, the aircraft entered heavy rain and the PF asked for the windshield wipers to be turned on at maximum speed which was done. FDR data showed that the aircraft then oscillated in roll as it crossed the runway threshold at 50 feet agl on centreline and close to the intended approach speed. Idle thrust was selected a 30 feet agl with the aircraft now left of centreline, banked to the left and drifting quickly toward the left edge of the runway as the rain reached its maximum intensity and visual references were degraded. A corrective roll to the right accompanied the initiation of the flare and the right MLG touched down just before the left 400 metres past the threshold and approximately 17 metres left of the centre of the 61 metre-wide runway. The left MLG "followed with a left wheel touchdown 30 metres left of centreline with the aircraft longitudinal axis 8° right of the centreline and drifting left at 2° and one second later, just before touching down, the rear inboard wheel of the left MLG clipped a runway edge light and its left outboard wheel landed on the grass (just as) the Captain ordered a correction to the right". The aircraft then struck two more runway edge lights before the left MLG returned to the runway. The landing roll was completed and the aircraft was taxied to its assigned gate where the crew, who stated that they had been unaware of the excursion, were informed of it.
Tyre marks in the grass which in total extended just over 200 metres. [Reproduced from the Official Report]
The ground track of the aircraft main landing gear during the excursion. [Reproduced from the Official Report]
It was concluded that the strong precipitation just north of the runway, a temporary drop in ground temperature and a change in wind direction and strength all tended to confirm the presence of a downburst from the thunderstorm cell just to the north of the runway at the time the aircraft was landing. However, no reactive or predictive wind shear warnings, which are inactive below 50 feet agl, were generated and the airport did not have a LLWAS. Both the following two aircraft making the same approach successively discontinued their approaches and flew a go around due to the deterioration in weather conditions.
It was noted that Air Canada had decided not to reactivate the rain repellent system on their A330 fleet using the modified repellent introduced in 1998 and had also decided not to take up the available alternative of applying an Airbus-approved hydrophobic windshield coating which provides a similarly-enhanced visibility in heavy rain. It was concluded that it was "not possible to establish with certainty whether the combined use of the rain repellent system and windshield wipers would have enabled the PF to maintain acceptable visibility through the windshield and keep the aircraft on the runway centreline" but it was known that rain both repellent systems and the alternative hydrophobic windshield coatings are effective in reducing the effects of heavy rain on forward visibility. The risk of a lateral runway excursion during landing because of a loss of visual reference, especially on a runway without any runway lighting, was considered to be heightened.
The Investigation noted that the applicable regulations require that runway lighting is on when IMC prevails or during the hours of darkness. Once the content of the SPECI issued four minutes before the flight under investigation touched down indicated that IMC prevailed at the airport, runway 24R should have been withdrawn from use because of the absence of such lighting. The Aerodrome Operator, Aeroports de Montréal (ADM), was responsible for any decision to close a runway or the whole aerodrome but it was found that the agreement with ATC defining the conditions for closure of a runway in the event of inoperative runway lighting did not specify any conditions for daytime closure and so this did not occur. It was concluded by the Investigation that "ADM appeared not to have realised that runway lighting was required both day and night for operation of a runway in IMC".
However, from a flight crew perspective, it was noted that there were no regulations stipulating any requirements for pilots to have runway lighting available in the prevailing conditions and it had been reasonable for the crew to believe that receipt of their landing clearance meant inter alia that "the runway complied with the airport certification standard and that all risk mitigation measures were in place".
In respect of aircraft manual handling immediately prior to the deviation, DFDR data showed that the oscillation in roll as the aircraft crossed the runway threshold had been a consequence of side stick inputs which could be identified as Pilot Induced Oscillations (PIO). Although these had ceased when the aircraft had been 40 feet above the runway, it was still banked 6° to the left and subsequent side stick commands to the right from then on had been insufficient to return the wings to level. It was concluded that the continued left bank combined in the presence of a crosswind component from the right and the effect of a downburst from the nearby thunderstorm cell had resulted in a continued drift to the left with the aircraft very close to the ground.
The question of whether a go around from a very low level would have been a safe option was considered. It was concluded that even in good visibility, commencing a go around would not have been easy whilst returning the aircraft to the centre of the runway very close to the ground in crosswind conditions and that as a result such a manoeuvre would have been "not advisable" under the prevailing conditions. The possibility that the Captain could have taken control was also considered but it was concluded that "even though he had been closely monitoring the progress of the flight", once the aircraft began to deviate left, it would not have been possible for him to "intervene and take the controls in time to make a correction to avoid the runway excursion".
Finally, the Investigation completed its analysis of aircraft operational factors by reviewing the general subject of continuing an approach in the presence of thunderstorms. It was noted that like many other airlines, "Air Canada […] has not developed clear procedures for circumnavigating thunderstorms during approach or landing" nor is this a requirement. Whilst en-route thunderstorm avoidance guidelines are explicit, pilots making an approach to land are expected to use their experience to determine the trajectory of their flight and as a result, "the margin of safety varies from flight to flight depending on the crew". It was considered that the excursion investigated here "shows that despite specialized training, having a detailed landing plan in place and threat management performed in accordance with existing concepts, an experienced crew was unable to counter the factors that suddenly combined at a critical moment in the flight in order to prevent the runway excursion". It was therefore concluded that the occurrence was a demonstration that "landings in the presence of thunderstorms near the runway present an ongoing risk to aviation safety".
The formally-stated Findings as to Causes and Contributing Factors were as follows:
- During an approach in the presence of a thunderstorm, pilot-induced oscillation led to the aircraft being in a left bank as it crossed the runway threshold which, combined with a strong right crosswind, resulted in a rapid drift to the left very close to the ground.
- Once the aircraft crossed the runway threshold, the intensity of the precipitation increased suddenly, such that the pilot flying (PF) had reduced visual references. Under these conditions, the PF did not detect the lateral movement of the aircraft in time to correct the drift before the outboard tires of the left bogie landed in the grass.
- Given the absence of runway lighting in reduced visibility conditions, it was difficult for the pilot flying to detect the lateral movement of the aircraft over the runway and therefore to prevent the runway excursion.
- A lateral wind shear generated by a downburst to the north of the runway suddenly increased the aircraft’s drift to the left during the landing flare.
- Runway 24R was not closed in Instrument Meteorological Conditions, even though the runway lighting was not working. As a result, the runway was not equipped with the lights required to enable crews to clearly distinguish the lateral confines of the runway.
The formally stated Findings as to Risk were as follows:
- If airports are not equipped with a low-level wind shear alert system, crews landing there may not be aware of the presence of downbursts or microbursts, and therefore may be exposed to the risk of approach-and-landing accidents.
- If a crew is unable to verify landing performance in heavy rain conditions involving a risk of hydroplaning, there is an increased risk of runway excursion.
- If the “landing” response to the “minimum” calls reinforces the notion that landing is assured, there is a possibility that preparation for, and the decision to, go-around could be affected, increasing the risk of a landing incident or accident.
- If the rain repellent system is unavailable or not used, there is an increased risk, in heavy rain conditions, that crews will lose the visual references necessary to avoid a runway excursion.
- If a crew does not consider the consequences of multiple threats, there is a risk that pilots will continue a landing under conditions that are not favourable.
- If the aircraft is drifting near the ground and pilots place the aircraft in low-energy landing regime, there is an increased risk of runway excursion.
- If crews are not trained to retake the controls at very low altitudes or during the low energy landing regime, there is a risk that, in the event of a problem, the pilot monitoring will not have time to identify the problem and take the appropriate measures.
- If Transport Canada does not take action to develop the clear standards on avoiding thunderstorms during approach and landing called for in Recommendation A07-01, approaches in the presence of convective weather will continue, exposing aircraft to the multiple, unpredictable hazards associated with thunderstorms.
- If occurrence sites are not preserved, there is a risk that evidence essential to identifying factors that contributed to an occurrence will be lost.
- If dispatch is not aware of an Aircraft Communications, Addressing and Reporting System transmission failure, there is an increased risk that critical flight information is not received by the crew.
Two Other Findings were also identified:
- Although the wind information at Montréal/Pierre Elliott Trudeau International Airport was transmitted to the flight by air traffic control in a timely manner, the information provided did not enable the crew to be fully aware of the rapidly changing weather conditions in the area of the runway.
- The runway excursion was not the result of a premature crab angle reduction manoeuvre, which is often associated with landing incidents in crosswind conditions.
Safety Action taken as a result of the occurrence was noted as having included the introduction of new Flight Operations Manual guidance by Air Canada to:
- ensure clarity on the requirement to go around if a landing is not assured in the touchdown zone, both longitudinally and laterally
- require pilots to commence a go around if the required visual reference is lost after descending below DH, DA, or MDA
- give direction to pilots regarding the lowest usable visibility relative to charted visibility.
The Final Report of the Investigation was authorised for release on 7 March 2017 and it was officially released on 28 March 2017. No Safety Recommendations were made.