SF34, Savonlinna Finland, 2019

SF34, Savonlinna Finland, 2019


On 7 January 2019, a Saab 340B made a late touchdown during light snowfall at night close to the edge of the runway at Savonlinna before veering off and eventually stopping. The Investigation attributed the excursion to flight crew misjudgements when landing but also noted the aircraft operator had a long history of similar investigated events in Scandinavia and had failed to follow its own documented Safety Management System. The Investigation also concluded that there was a significant risk that EU competition rules could indirectly compromise publicly-funded air service contract tendering by discounting the operational safety assessment of tendering organisations.

Event Details
Event Type
Flight Conditions
Flight Details
Type of Flight
Public Transport (Non Revenue)
Take-off Commenced
Flight Airborne
Flight Completed
Phase of Flight
Location - Airport
Inadequate Aircraft Operator Procedures, CVR overwritten, PIC aged 60 or over
Plan Continuation Bias, Procedural non compliance
Surface Friction
Directional Control, Late Touchdown, Off side of Runway
Indicating / Recording Systems
Damage or injury
Aircraft damage
Non-aircraft damage
Non-occupant Casualties
Off Airport Landing
Causal Factor Group(s)
Aircraft Operation
Safety Recommendation(s)
Aircraft Operation
Aircraft Airworthiness
Investigation Type


On 7 January 2019, a Saab 340B (YL-RAF) being operated by RAF-AVIA on an international positioning flight from Riga to Savonlinna as MTL650P touched down late and near the edge of runway 12 after an ILS approach in night IMC before veering off it, striking a significant snowbank and coming to a stop. No occupants were injured but considerable damage was caused to the aircraft and a runway light was broken.

The aircraft in its final resting position. [Reproduced from the Official Report]


An Investigation was carried out by the Finnish Safety Investigation Authority (SIAF). The FDR and CVR were removed from the aircraft and downloaded by the UK AAIB. The FDR data were incomplete and the recording was of poor quality on account of the extremely poor and worn condition of the recorder’s magnetic tape but useful data were eventually extracted. The CVR download did yield data but only from flights prior to the accident flight. When the flight deck was inspected after the accident, it was found that the CVR CB had not been tripped although it was accepted that it was possible that it was in a different position during the flight. It was noted that “the lack of cockpit voice recordings significantly hampered the investigation of the sequence of events”. Voice recordings of all communications from and to the AFISO were also provided.

It was noted that the 61 year-old Captain of the aircraft, who had been PF for the flight had a total of approximately 20,000 hours flying experience since 1982 of which 746 were on type. He had first flown the Saab 340 in the 1990s but then not again until joining RAF-AVIA eighteen months prior to the accident, prior to which he had been an Airbus A320 pilot for approximately ten years. The 34 year-old First Officer had a total 1,057 hours flying experience, all but 220 hours of which were on type.

The positioning flight involved was taking place so that an aircraft would be available to operate the first flight of the year on the Public Service Obligation (PSO) air service from Savonlinna to Helsinki - there had been no service during the Christmas and New Year holidays. This flight was scheduled to depart at 0600 and the 1 hour 45 minute from Riga was due to arrive at 0540. The Latvian aircraft operator, RAF-AVIA, was providing the service on the route as a subcontractor to the company which had been successful in the tendering process which had taken place in late 2017, Maavoima. This Company was a ground handling and logistics company, not an aircraft operator, and “pursuant to EU practices, (it) relied on the fact that Air Operator Certificates and Operating Licences ensure that the (contracted) air operator acts in a safe manner”. It was also noted that “once flight operations commenced, Maavoima supervised the on-time performance of flight schedules (but) their safety management system did not have any procedures or obligation for assessing the airline’s operating safety”.

What Happened

The prevailing weather conditions were normal for winter. The snow clearance team at Savonlinna airport began runway clearance a little over two hours prior to the scheduled departure time of the flight to Helsinki. The runway was confirmed free of ice but there was packed snow from previous clearance work at the edges and new snow was falling throughout the clearance process. After initial clearance was completed, friction was measured but found to be not good enough with further clearance work required. The inbound positioning flight arrived on time but was instructed to begin holding at VALGU until the additional runway clearance had been completed.

The resulting estimated runway friction measurements in thirds in the landing direction (12) were 0.29 (medium to poor), 0.22 (poor) and 0.23 (poor) but using the permitted discretion afforded to the person inspecting the runway to use their judgement to determine the estimated runway friction to be one level worse than indicated by the measured values, the first third of the runway was also reported as having ‘poor’ estimated friction. A 37 metre cleared width out of the 45 metre declared width of the runway between snowbanks which were 40 cm high and four metres inside of the runway edge lights was also notified with that cleared width stated to be completely covered with a 4 mm layer of dry snow. This information was passed to the inbound flight and, when subsequently cleared, it then established on the runway 12 ILS for a flap 20 approach. The surface wind was passed as from 220° degrees at 5 knots which represented a small crosswind component from the right. The required visual reference was acquired prior to DA. FDR data showed that the approach had been stable until reaching the runway threshold. However, following the initiation of the landing flare, the aircraft floated close to the surface for 6/7 seconds before touching down at a speed of 109 knots near the end of the TDZ and on the left side of the runway rather than the (obscured) centreline.

Almost immediately, the left main landing gear hit the left snowbank on the uncleared runway margin and then when the nose gear touched the runway, the left main gear became momentarily airborne again as the right main gear touched the ground followed by the left main gear. The aircraft began to track away from the runway centreline and after approximately 100 metres, all wheels were no longer on the runway cleared width. An attempt to regain it was not successful and resulted in the aircraft skidding sideways till outside the cleared runway width. The left main gear then impacted the much higher snowbank outside the full runway width which caused the aircraft to turn more sharply to the left. The left engine then stalled when snow packed its air intake and the right main gear destroyed a runway edge light. The aircraft finally came to a stop in snow half a metre deep having made a 110° left turn relative to the intended direction of landing. Damage to the aircraft included to the propellers, the nose landing gear assembly and the fuselage which was dented and cracked in various locations. Most of this damage occurred when the aircraft ran into the packed snowbank when the propellers flung snow and pieces of ice at the fuselage.

The ground track of the excursion. [Reproduced from the Official Report]

Flight Crew Performance

It was noted that “investigation of the flight crew’s observations, decision-making and multi-crew cooperation” had been significantly hindered by the absence of CVR data. However, it was concluded that having been awake since the early hours of the morning, the influence of circadian rhythms was likely to have meant, particularly in the case of the Captain, that the pilots would not have been “as alert as possible at the time of the occurrence”. Visual judgement of aircraft position relative to a landing runway just prior to touchdown at night with sideways-blowing snow in the beam of the taxi and landing lights may have created an illusion of sideways motion which is known to be capable of taking a pilot’s attention away from the runway edge lights and increasing the chances of an unnecessary corrective control input which may lead to a touch down at the side of the runway as occurred in this case. It was also noted that the pilots would have been aware that the delay awaiting runway clearance and fiction measurements when there was only 20 minutes between their planned arrival time and the scheduled departure time of the service flight to Helsinki would have already delayed the departure. It was considered likely that overall effects of some or all of these factors would have “increased the pilots’ cognitive stress during the landing, which was already challenging in itself”.

Aircraft Operating Procedures

  • According to the ‘General’ volume of the aircraft operator’s OM, the OM-A, the crosswind limits given in the aircraft type-specific OM-B must be followed during landing. However, it was found that the OM-A also states that if the estimated runway friction is poor, as was the case, then crosswind landings are prohibited. It was observed that “if the estimated runway friction is poor, the OM-A overriding restriction makes it very difficult to comply with the OM-B crosswind limits”.
  • The landing and taxiing lights were selected for the final approach and landing when there was light snowfall, whereas the aircraft operator’s OM-A and OM-B both contained a caution that using lights during heavy snowfall may degrade visibility and cause sensory illusions during landing.

The Aircraft Operator

The Investigation noted that the aircraft operator involved had previously been subject to a number of independently-conducted Serious Incident Investigations in both Finland and adjacent Scandinavian countries. These included a runway excursion on landing at Trollhättan, Sweden in 2018, a dangerous approach to Mariehamn, Finland in 2012 and a risk of ground collision at Helsinki in 2011.

It was decided to examine the operational safety performance of the aircraft operator in some detail and it was found that it had a persistent history of “shortcomings in flight safety” despite considerable attention from the Latvian CAA. This body stated that “the greatest threats to RAF-AVIA’s flight safety included shortcomings in pilot skills and decision making and shortcomings in CRM as well as insufficient support by the airline’s management and shortcomings in supportive procedures regarding flight operations”. It also found that although RAF-AVIA’s documentation included an EASA-compliant SMS, this had little in common with the operational reality. The almost complete absence of any meaningful safety reporting culture or safety performance tracking was attributed to “the time when Latvia was part of the Soviet Union (when) pilots would often avoid occurrence reporting because they could have been punished for doing so”. It was noted that “even though punishment is no longer a part of the airline’s procedures, it has been challenging to mitigate the fears of especially the older pilots”.

The Air Service Tendering Process

The Investigation also examined the context of the investigated accident in respect of the process by which it had been awarded. On request from the City of Savonlinna, the Finnish Transport Agency decided that a scheduled air service linking Savonlinna with Helsinki should be the subject of a Public Service Obligation (PSO) for the period from January 2018 to December 2020. Accordingly a public tender competition which complied with Article 17 of EU Regulation on common rules for the operation of air services in the Community was held and the award to Maavoima was subsequently made solely on the basis that they had submitted the lowest price tender. The tender process did not stipulate any conditions for safe operations because, whilst the applicable EU Regulation did not specifically prohibit setting additional terms, “the Finnish Transport Agency had a strong suspicion that the EU would interpret any safety-related extra conditions as a restriction of competition”. The Agency also noted that historically, competition for public tenders such as the Savonlinna one “has been so fierce that the air carriers that lose public tenders have routinely appealed the contract awards” through the courts which has “resulted in protracted stalemates” and lengthy appeal processes. This experience was stated to have “made the Finnish Transport Agency very sensitive about making absolutely certain that the competitive tendering criteria unquestionably met the EU Regulations” with the consequence that “safety, in reality, has not been a criterion in competitive bidding”.

Flight Recorder Serviceability

The finding that both the CVR and FDR had been dependent on magnetic tape as a recording medium and the quality of recording had in both cases been compromised by the condition of the magnetic tapes was of concern to the Investigation. In the case of the CVR, it was noted that the EASA had prohibited the use of CVRs that use magnetic tape as a recording medium with effect from 1 January 2019 so that the recorder installed in accident aircraft was non-compliant. In the case of the FDR, it was noted that after investigating a 2017 UK Serious Incident, the UK AAIB had recommended to the EASA that they “should set an end date to prohibit the use of flight data recorders that use magnetic tape as a recording medium, to ensure compliance with ICAO Annex 6 from that date”. However, the EASA response to this had been that it “would not be justified because, by 1 January 2019, the proportion of aeroplanes fitted with a magnetic tape FDR was estimated to be close to zero” and stated that “replacing the FDRs that use magnetic tape would most probably be allocated a low priority”.

A previous similar accident: the Investigation noted that a 2013 landing runway excursion involving a Saab 340 Lappeenranta had occurred in similar circumstances to the event under investigation with a similar outcome.

Nine Conclusions from the Investigation in respect of both the cause of the event and the circumstances, both direct and indirect, in which it occurred were summarised as follows, where each factual finding is followed in italics by the conclusion in that respect:

  • The operating licences and air operator licences issued by the EU Member States’ aviation authorities do not guarantee constant and uniform safety levels among air carriers.

Conclusion: In addition to operating licences and air operator licences, competitive tendering for air services requires other practices to verify the safety of airlines.

  • While the EU’s regulations have aimed at ensuring the most open participation in tendering, they may result in overlooking the safety of aviation as one criterion in tendering.

Conclusion: The interpretation of regulations should not result in a situation where qualitative criteria in competitive bidding are discarded because of tendering rules, the risk of challenging a decision or the desire for an uncomplicated process.

  • The purchaser organising the public tender for air services will not necessarily impose any safety-associated criteria because of being cautious about breaking EU competition rules and the court processes launched by losing bidders. Often the price and on-time performance are the tender criteria.

Conclusion: Tenders for air services may not necessarily assess the operators’ safety records at all. Present air service competition rules do not encourage operators to invest in safety.

  • Purchasers of air services do not have suitable and straightforward indicators to assess air carrier safety. Clients and purchasers may also include those that are not deeply familiar with the aviation branch.

Conclusion: It is difficult for purchasers to reliably compare the safety of air carriers.

  • The airline had not completely complied with its own safety management system. Oversight authorities do not always detect the difference between the safety management that operators promise to follow and their real-world practices.

Conclusion: Authority oversight does not always extend to the implementation of operators’ safety management systems or to actual practices.

Conclusion: A go-around is always the safe option if the preconditions for landing are not met or if a safe landing cannot be achieved.

  • The airline’s operational manuals (OM-A and OM-B) were inconsistent concerning maximum crosswind components. The instructions were difficult to follow in practice.

Conclusion: Manuals must be consistent in all respects and user-friendly during the different stages of the flight.

  • Regardless of the alert to the Emergency Response Centre, no information about the airliner accident was relayed to the region’s divisional officer on duty because, owing to the situation assessment, there was no need to deploy rescue service units.

Conclusion: The region’s divisional officer on duty responsible for rescue operations must be sufficiently informed of an accident occurring in the region, even if the situation did not require deploying rescue service units. The situation may change from the onset, requiring the commencement of rescue actions.

  • The Cockpit Voice Recorder had not recorded anything from the flight in question and the earlier recordings that were retrieved from its memory were of extremely poor quality. The recording quality of the FDR, when compared to modern recorders, was poor. The magnetic tape of the FDR was worn, which caused defects in the recording.

Conclusion: The purpose of flight recorders is to make it easier to investigate accidents and incidents so as to improve safety. Aircraft should carry recorders that meet modern-day recording capacity and reliability requirements.

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

  • that the Finnish Transport and Communications Agency explore and instruct how operators’ aviation safety and safety management performance, as criteria, can be used in tendering for air services, taking the EU’s regulations into account. [2019-S54]
  • that the European Commission see to it that a process is created by which it becomes possible to impartially assess operators’ safety management performance and safety levels in tendering for air services. [2019-S58]
  • that the European Union Aviation Safety Agency (EASA) ensure that the audits conducted by the EU Member States on operators also cover the practical functioning and performance of safety management systems. [2019-S59]
  • that the European Union Aviation Safety Agency (EASA) set a deadline for the use of flight data recorders recording on magnetic tape. [2019-S60]

The Final Report was completed on 10 December 2019.

Related Articles

SKYbrary Partners:

Safety knowledge contributed by: