Description

On 20 January 2020, a DHC8-300 (LN-WFO) being operated by Widerøe on a scheduled domestic passenger flight from Kristansund to Bergen as WF977 in night IMC encountered moderate and then severe icing conditions en route, during which significant ice shedding from the propellers occurred. As the aircraft continued descent into warmer air, a failure of the left engine at 6,700 feet was attributed by the crew to engine ice ingestion and as expected auto ignition restarted it after a short time. However, the right engine then failed and an emergency was declared, and this time it took longer to restart. Shortly before it did the left engine failed again leaving the aircraft briefly without either engine functioning until the left engine restarted at 4,500 feet. There were no more failures, and the emergency was cancelled prior to landing.   

Investigation

A comprehensive Serious Incident Investigation was carried out by the Norwegian Safety Investigation Authority (NSIA). The CVR was secured after landing but as some time elapsed before this happened, the first 28 minutes of the 73-minute flight was not available. Relevant data from the FDR were also available and data from both recorders were of good quality.  

The Flight Crew

The captain had joined Widerøe as a first officer in 1998 and was promoted to captain in 2006. He had a total of 12,715 hours flying experience of which 2,313 hours were on type, on the DHC8-100, -300 and -400. He had been mainly based at Bergen whilst employed by the airline but had also been based at Tromso. The first officer, who was acting as PF for the flight, had been employed by Widerøe for 18 months and had a total of 1,533 hours flying experience of which 74 hours were on type.

What Happened

No areas of severe icing had been forecast along the route to be flown. After takeoff, the flight was offered a direct routing to the first approach waypoint on the way into Bergen. This was accepted, but took the flight further inland where the precipitation was more intense and the risk of en route icing higher. Cloud was entered at 4,000 feet with the de icing system on. En route, the weather radar was not used but ice began to form on the aircraft around FL140, and a thin layer of ice was seen on the wings. The crew stated that they had expected to pass through this icing layer and climb clear of cloud as they had done on the three preceding flights. However, the icing intensified as they climbed and when they levelled at FL220, the aircraft was still in cloud with considerable ice accretion evident. They stated that they had not noticed changes to the aircraft’s stability or performance other than a slightly lower than normal airspeed which they attributed to accreted ice. This was confirmed by the propellers shedding ice that hit the fuselage and by airframe vibration. According to the airline’s internal investigation report, the experienced cabin attendant had found the situation “dramatic," with abnormally strong shaking, vibrations and loud bangs when ice shed from the propeller and hit the side of the fuselage. 

The crew reported trying different propeller rotational speeds to improve ice shedding and settled for 1,050 rpm before deciding to request descent in the hope of finding increased air temperature (propeller speed was increased to 1,200 rpm later in the descent). This descent was requested at the same time as submitting an AIREP (air report) for what was perceived as moderate icing - the captain stated at interview that he had considered the intensity of the icing to be moderate throughout. However, the Investigation subsequently decided that this view was not consistent with the assembled evidence which pointed to severe icing being a more appropriate description.

In response to their request, the flight was cleared to descend to FL160 and began to do so. Soon after beginning descent, they requested a turn towards the coast (see the map of the flight track below) as they considered this would take them into warmer air over lower terrain. The AP remained engaged and the crew said that they had not considered disengaging it.

The ground track planned (orange) and flown (blue). "BR638" and "Nidgi" are flight plan waypoints and "Bergen flyplass" is Bergen Airport. [Reproduced from the Official Report]

The icing continued even after the aircraft approached FL160, and use of the inspection lights showed that there was a large amount of ice on the wing leading edge, the propeller spinners and underside of the flap fairing but “did not remember seeing ice on or in the engine air inlets". Abnormal amounts of ice accumulated around the left and right windshield wiper arm bases. There was also ice on the lower part of the flight deck windows, which at one point were nearly completely iced over. However, the captain explicitly stated that he had regarded the ice accretion as a nuisance rather than dangerous despite the continuing airframe and substantial ice shedding from the propeller blades. 

Having reached FL160 approximately five minutes after starting the descent, the crew did not wish to stay on a westerly course for too long and requested permission to proceed towards waypoint BR638, an alternative to waypoint NIGDI from either of which an approach to Bergen could be started. Aware that the freezing level was around 6,000 feet, descent to FL100 was obtained in an attempt to reduce the icing, but considerable ice remained despite accreted ice increasingly separating from the aircraft and the perceived intensity of icing decreasing. Another company flight en route northwards from Bergen at FL140 advised via ATC that they had not experienced any icing and so it was decided to stop the descent at FL120. 

The relative severity of icing conditions being experienced as altitude decreased and when the propeller vibrations subsided, the propeller speed was again returned to 900 rpm. Ice was still visible around the wiper arm bases and on the flap fairings and spinners. The flight was cleared to position towards an ILS ‘W’ approach to runway 17 at Bergen and restarted its descent. Ice shedding from the aircraft continued at irregular intervals and as they approached the anticipated freezing level, the quantity of ice being shed noticeably increased. Within a minute or two, power was lost on the left engine. The captain concluded that the engine had flamed out due to ice or water ingestion and that auto-ignition would initiate a relight, which happened after less than a minute. FDR data showed that the engine regained normal power after 25 seconds. However, less than a minute after this recovery, the right engine lost power whilst displaying similar rundown indications to those that the left engine had shown. As the aircraft descended through 5,200 feet, the crew declared an emergency but suspecting the same cause, they waited to see whether the engine would restart automatically which it eventually did on the second attempt. Whilst awaiting this eventually successful restart, the left engine failed again leaving the aircraft completely without engine power for thirteen seconds. As both engines returned to normal operation, the aircraft was passing 4,500 feet and established on the ILS.

The emergency was cancelled prior to landing and the final minutes of the flight were without further event. As the aircraft taxied in after landing, ice was still visible on the leading edge of the wings, the nose, the propeller spinners, and the area between the spinners and the air inlet despite the 8°C surface temperature and the rainy weather. The captain contacted Widerøe’s Flight Operations Duty Manager who instructed the crew to isolate the CVR and decided to remove the crew from active service and have the aircraft towed into a hangar to be checked by maintenance. An inspection of both engines found damage to some of the blades in the first-stage LP compressors of both engines as well as damage to some HP compressor blades in the Number 2 engine - one compressor blade was bent and several others were chipped.

Observations

In respect of the extensive use of the AP when in what the Investigation assessed had been severe icing conditions, it was noted that whilst such operations were not prohibited in the AFM, it did require explicitly that “the autopilot must be disengaged in severe icing." It also states that only if the flight deck side windows are free of ice can it be assumed that the aircraft is not in severe icing conditions.

In respect of the way the crew dealt with the engine failures, it was considered that it had been crucial to the outcome. A different reaction, for example if the pilots had started a go-around or initiated engine failure procedures, could have resulted in a different and much more serious outcome.  

Safety Action taken by Widerøe following this event was based on an internal investigation which made a number of safety recommendations. Those recommendations include development of a comprehensive computer-based training (CBT) programme on flight in icing conditions which covers the recognition of severe icing on the aircraft, the use of weather radar, and special weather phenomena that may lead to icing. Simulator training has also been augmented to cover weather radar use and the auto-ignition system. Several related changes to the company OM "A" and OM "B" to improve procedures for flight in icing conditions were made. Also, an Operational Directive was issued in 2021 to require a tactile inspection of the engine air inlets, which whilst “strongly recommended” by the aircraft manufacturer but was missing from the OM "B" at the time and was not performed prior to the investigated flight. However, the NSIA did not believe that such a check would have made any difference to the sequence of events in connection with this particular flight.

The Main Conclusion of the Investigation was formally documented in the following narrative format:

While climbing after takeoff from Kristiansund, the aircraft inadvertently flew into an area of severe icing conditions, and the crew took corrective action by changing course and altitude. However, they were unable to sufficiently limit the duration of the aircraft’s exposure to severe icing conditions. As a result, ice formed on the aircraft and inside the engine’s air inlets.  

On approach to Bergen, the intended destination, the aircraft lost engine power on the left engine, then on the right engine, and then on the left engine again. The aircraft’s automatic ignition system restarted both engines, but the start-up sequence took time, and the aircraft was completely without engine power for a brief period. The engines flamed out due to ice detaching from the engine’s air inlets. The ice either entered the combustion chamber as slush and water and caused a flameout, or disrupted the airflow into the engine sufficiently to stall it. 

The crew were assessed to have acted professionally in a highly demanding situation and landed the aircraft safely at Bergen.   

Other conclusions included that “the crew had an unclear understanding of the boundary between severe and moderate icing which may have contributed to prolonging the exposure to severe icing conditions” and that “most likely the ice that entered the engines had formed in and around the air inlet during flight”. It was also noted that although the Norwegian ANSP Avinor “has previously considered introducing the use of weather radar, it has concluded that there are technical and financial challenges associated with presenting nationwide weather in the necessary way” but is still currently considering the inclusion of a weather radar element as part of the new en route system being planned for the Oslo TMA.   

Three Safety Recommendations were made as a result of the findings of the Investigation as follows:

• that the Norwegian Civil Aviation Authority implement a project in which Avinor, Norwegian Meteorological Institute and representatives from a suitable airline participate to assess possible solutions on presentation of updated (live) weather information that can be communicated to relevant flights. [2023/01T]
• that Transport Canada assures that De Havilland includes a warning in the Airplane Flight Manual in which users are informed that the engines may flame out if an aircraft inadvertently flies into severe icing conditions. Ice can build up in the air inlet and subsequently detach and enter the engine. [2023/02T]
• that the European Union Aviation Safety Agency clarifies the inadequate definitions as well as the existing inconsistency relating to icing problems and ensure that the results are harmonised with other international authorities. [2023/03T]

The Final Report of the Investigation was completed on 7 February 2023 and published the following day. 

Related Articles

• Ice Formation on Aircraft
• In-Flight Icing
• Aircraft and In Flight Icing Risks
• Aircraft Ice Protection Systems
• Flame Out
• Weather Radar
• Weather Radar and ATC