SF34, en-route, northern North Sea UK, 2014

SF34, en-route, northern North Sea UK, 2014


On 3 October 2014, the crew of a Saab 340 in the cruise at FL150 in day IMC did not recognise that severe icing conditions had been encountered early enough to make a fully-controlled exit from them and although recovery from the subsequent stall was successful, it was achieved in a non-standard manner. The Investigation concluded that although both mountain wave effects and severe icing had contributed to the incident, the latter had been the main cause. Both crew understanding of airframe icing risk and supporting Operator and Manufacturer documentation on the subject were considered deficient.

Event Details
Event Type
Flight Conditions
Flight Details
Type of Flight
Public Transport (Passenger)
Flight Origin
Intended Destination
Take-off Commenced
Flight Airborne
Flight Completed
Phase of Flight
Deficient Crew Knowledge-automation, Deficient Crew Knowledge-performance
Authority Gradient, Inappropriate crew response (automatics), Ineffective Monitoring, Procedural non compliance, Ineffective Monitoring - PIC as PF
Environmental Factors, Aerodynamic Stall
In Flight Airframe Icing
Damage or injury
Non-aircraft damage
Non-occupant Casualties
Number of Non-occupant Fatalities
Number of Occupant Fatalities
Off Airport Landing
Causal Factor Group(s)
Aircraft Operation
Safety Recommendation(s)
None Made
Investigation Type


On 3 October 2014, the crew of a Saab 340B (G-LGNM) on a scheduled domestic passenger flight from Aberdeen to Sumburgh in day Instrument Meteorological Conditions (IMC) were unable to achieve the normal rate of climb or accelerate once at cruising altitude. During an attempted descent, airframe vibration occurred and activation of the stall warning system followed following which a recovery to a safe airspeed was made after which the remainder of the flight to destination proceeded was without further event.


An Investigation was carried out by the UK AAIB. The FDR and CVR were both recovered and successfully downloaded with the former providing data essential to establishing the circumstances which had prevailed.

It was found that the 51 year-old Captain had 9168 hours total flying experience including 6283 hours on type accumulated over 10 years. He was also Line Training Captain for the Operator. The Co-Pilot had 2340 hours total flying experience including 928 hours on type accumulated over 2 years.

It was noted that for both pilots, the flight made had been the third of a four sector morning duty which consisted of two return trips from their Aberdeen base and Sumburgh. During the first two sectors, "little cloud and no significant icing had been encountered" and the pilots, familiar with the route, "were not expecting problems with icing on this sector" and "regarded forecasts of icing conditions and mountain wave effect as routine for that area".

The incident flight, with the Captain as PF was planned to cruise at FL150 and ATC approved climb to FL130. The forecast available to the crew indicated that a cold front, orientated north-east / south-west was expected to pass through their route during the morning with BKN or OVC AC cloud either side of it between 8000 and 10000 feet which would be associated with moderate icing and moderate turbulence. Ahead of the front, the freezing level was estimated as 9000 feet and behind it 6000 feet or lower. A SIGMET was in force for severe mountain waves between FL40 and FL280 over a large area either side of the front because of a strong south-south-westerly airstream, with a maximum vertical speed of 900 fpm expected at 9000 feet. Turbulence within these mountain waves was forecast to be moderate or occasionally severe.

Cloud was entered around FL65 and engine anti-ice and airframe de-ice systems were selected on. Above FL110, the climb rate began to reduce and ice began to accrete on the propeller spinners, prompting the selection of propeller de-icing as the OAT reading reached -5º C. The climb was made with the AP engaged and VS mode selected. Airspeed decreased steadily from "more than 160KIAS, the normal minimum speed for icing" to 145 KIAS, the recommended speed for exiting icing conditions, at which point IAS mode was engaged. Climb Power and the normal 1230 Propeller RPM were maintained and the achieved rate of climb was thereafter very variable, ranging between about 800fpm and "slightly negative".

On reaching FL130 and levelling off still in IMC with the OAT reading now -7º C, the aircraft failed to accelerate beyond 164 KIAS when 180KIAS or greater would be the normal expectation. The Captain decided to "climb the aircraft another 100 ft, then descend back to FL130 using VS mode", which he has seen another pilot do although he had not previously tried it himself. He reported believing that "the angle of attack (AOA) and drag would reduce during descent and the aircraft would accelerate" but the speed reduced to 150 KIAS during the short climb and did not increase in the descent back to FL130. Both pilots reported having initially concluded that they were experiencing moderate icing and it was decided to descend to FL110 to increase forward speed. The AP remained engaged and a 1000 fpm descent in VS mode was selected.

However, when the aircraft pitch attitude remained at about 5º nose-up, the rate of descent was increased to 2500 fpm. Almost immediately, airframe vibration had begun, followed by activation of the stall warning accompanied by stick shaker operation, which disconnected the AP. The Captain, using "a little more force than usual" pitched the aircraft 2º nose-down and "it then responded normally as the speed increased to greater than 190 KIAS". He reported that the vibration had ceased and the aircraft had "seemed to be in trim, without any pitch trim adjustment" so the AP was re-engaged approximately 9 seconds after it had disconnected. Once at FL110, this was maintained as the cruise level and the remainder of the flight proceeded normally.

FDR data showed that after an 11 second descent from FL131 to FL130, the effect of altitude capture mode was to cause pitch attitude to increase and a stall is clearly shown with AoA 8º as the aircraft began the descent from FL130 and buffeting started. Thereafter, whilst the pitch attitude had begun to reduce from its peak of 7/8º, the AoA continued to increase.

An aftercast produced by the UK Meteorological Office indicated that the incident had probably occurred "in the air mass just ahead of the cold front, with a particularly moist layer above the freezing level". It also indicated that the severity of the mountain waves likely to have been encountered by the aircraft at FL130 was likely to have been much less than the forecast maximum - of the order of 100 - 150 fpm. It was also calculated that the aircraft would have encountered "an altocumulus cloud layer about 7,500 ft in depth, with tops at FL150, and with conditions conducive to moderate or severe icing". The water droplets in this cloud at FL130 were considered likely to have been of mid to large size i.e. may well have included droplets over 50 microns in diameter, i.e. meeting the description 'Supercooled Large Droplets' (SLD), which are known to be associated with severe icing conditions.

Aspects of the aircraft pertinent to the airframe de-icing risk were reviewed by the Investigation:

  • It was noted that the aircraft was not fitted with an ice detector and that after stating that ice build up on the windshield wiper arms gives a visual cue of ice accumulation, the AOM then said that "airframe ice can be present without any build-up on the wiper arms.... (so) accumulation of ice shall be monitored on all visible surfaces".
  • It was noted that the aircraft had received a modification to the stall warning system which introduced an 'Ice Speed Function' which if selected will, provided the aircraft has been airborne for at least 6 minutes, reduce the AoA at which the stick shaker operates from 12.1º to 5.9º but would not affect the stick pusher AoA threshold. Had this system - which recognises that an ice-contaminated wing will stall at a lower AoA (and higher airspeed) than a clean one - been active, it would have activated the stall warning at the onset of the first uncommanded roll, some 23 seconds earlier than occurred. However, at the time of the incident, "permission to use the system on Saab 340s had been temporarily withdrawn by EASA".
  • Although the Investigation found that "No tables were provided for the stall speed in JAR-certified icing conditions" it was noted that "applying the likely 10% increase in IAS from flight trials" the stall speed of the incident aircraft would have been 115.5 KIAS, which was 16 knots below the lowest speed reached when stalled.

More generally, it was noted that "the accepted aeronautical terms for describing icing intensity are ‘trace’, ‘light’, ‘moderate’ and ‘severe’ but there is no internationally recognised definition of icing severity". The UK AIP states that for reporting purposes, "moderate icing exists when the rate of accumulation is such that even short encounters are potentially hazardous" with the use of de-icing/anti-icing equipment and/or a change of flight path may be necessary. The same source describes severe icing as existing "when the rate of accumulation is such that de-icing/anti-icing equipment fails to reduce or control the hazard" necessitating immediate change of flight path. The UK Meteorological Office encourages its forecasters "to consider a number of parameters together, including the type and extent of cloud, as well as its depth, temperature and relative humidity" when forecasting moderate or severe icing.

The Standard Operating Procedures (SOPs) of the Operator involved in this incident relevant to the aircraft type involved were considered. It was found that:

  • the manufacturer's AFM and AOM were considered to be the definitive guides to aircraft operation unless specific instructions to the contrary appear in the aircraft type specific volume of the Operations Manual (OM).
  • the aircraft was approved for flight in moderate icing conditions.
  • a graph of liquid water content is published in the OM which distinguishes between "what is considered to be" the maximum size of Supercooled Water Droplets (SWD) for flight in intermittent (40 microns) and continuous (50 microns) icing conditions, the former linked to flight in cumuliform cloud and the latter to flight in stratiform cloud.
  • in the case of "substantial build up of ice on the spinner further aft than normally" increased scanning of the wing upper surface aft of the protected zone is seen, immediate disconnection of the AP (if engaged) and immediate exit of such conditions is necessary. The purpose of the black painted rings on the spinner is not explained.
  • an instruction to disconnect the AP "if there is a significant loss of performance in icing conditions" appears in the AOM but not in the OM.
  • the OM specified the minimum IAS in icing conditions above MSA according to flap setting and angle of bank. For a clean aircraft en route with wings level, this is 160 KIAS. A lower speed is allowable for the purpose of exiting icing conditions.
  • both the OM and AOM state that "if experiencing extreme icing conditions and safe speed and/or climb rate can not be maintained" there should be no hesitation before temporarily setting maximum power "if required to escape from the situation". They also state that "extreme icing conditions do not necessarily imply a large amount of ice" and may equally well be evident from "ice accumulation causing a large impact on performance making airspeed decrease towards the minimum safe speed in icing conditions". However, "no other explanation or definition is provided for the term 'extreme icing'”.
  • the AOM states that Maximum Continuous Power (MCP) is available for two engine operation in 'extreme icing conditions' but is not intended for use during 'normal icing conditions'. The OM uses 'severe icing conditions' in place of 'extreme icing conditions' and notes that MCP (which requires a propeller rpm of 1384, "may be used to ... maintain a safe flying speed". It was noted that the use of MCP at FL130 would have provided approximately 18% more engine power.
  • the OM warns that with ice on the wing, a stall can occur before or at stick pusher activation and in some adverse cases may even be encountered before the annunciation of a stall warning.
  • the OM states that buffeting ahead of a stall may occur at a speed of up to 25% above the applicable ice free stall speed but could be mistaken or hidden by the “vibration” caused by the uneven shedding of ice from the propellers.’
  • the Emergency Checklist states that should unusual vibrations be experienced and there is any doubt as to whether this is pre-stall buffet or an unusual vibration, the response must be a stall recovery.

Much of this was covered in the Operators "Winter Operations Brief 2013/12014" which was no longer current and although a new one had not been published for Winter 204/2015, both pilots stated that they had retained familiarity with the contents of the expired Brief. These contents included reminders that:

  • loss of performance in icing conditions may be indicative of serious airframe icing, even if it is not observable.
  • that ice accretion on the airframe can lead to a stall at speeds up to 30% above comparable clean speeds and result in a stall occurring before activation of the stall protection system.
  • the first indication of an impending icing-induced stall may light buffeting and or an un-commanded roll oscillation.
  • if it is suspected that an icing induced stall has developed or is imminent then the AP must be disconnected, the nose lowered approximately 5 degrees and Maximum Climb Power (MCP) applied.

The Operator noted that although recurrent simulator training covered flight in icing conditions, it "had had not previously been possible to simulate severe icing or the pre-stall buffet".

Previous losses of turboprop aircraft in similar circumstances were noted as including an ATR 72 in the USA in 1994, a Saab 340 in the USA in 2006 and a Saab 340 in Argentina in 2011.

The formally documented Conclusion of the Investigation was as follows:

"The aircraft probably encountered both severe icing conditions and mountain wave effect while climbing. The crew reduced the airspeed to VCLEAN+15, for optimum climb performance, but the propeller rpm and power were not increased to MCP. The co-pilot perceived that the command gradient between himself and the commander was too steep for him to feel comfortable advocating such a procedure, even though he believed it appropriate.

After levelling-off, airspeed initially increased before reducing back towards VCLEAN+15, a speed that was only intended for use when climbing out of icing conditions, and in combination with MCP. It was apparent that the aircraft’s performance was being impaired by ice and it would have been appropriate to set MCP, as well as disengage the autopilot.

Pre-stall buffet was experienced and the recovery was delayed until after the stall warning had activated. Not all the stall recovery vital actions were implemented, although control was regained before a wing drop developed, as had happened in previous Saab 340 stall events."

Safety Action initiated as a result of the investigated event of which the Investigation was aware was noted as:

  • The Aircraft Manufacturer is reviewing the guidance in the AFM and AOM relating to flying the Saab 340B in icing conditions which will include crew actions to be taken on identifying specified icing conditions and will use accepted aeronautical terminology compatible with that used in the UK AIP.
  • The Operator has updated its advice to crews on flying the Saab 340B in icing conditions, introduced improved recurrent training on operation in such conditions and improved access to the AFM and AOM.
  • The EASA has approved a modification which reinstates the Ice Speed Function so that earlier activation of the SPS can occur in icing conditions.

The Final Report was published on 10 September 2015. No Safety Recommendations were made.

Related Articles

Further Reading

SKYbrary Partners:

Safety knowledge contributed by: