By Captain Chris Seal
The below is the opinion of the author and not necessarily that of the NTSB. Although much is taken from the full NTSB published report, the author bases some insights on his experience as a Pilot and Commander (on the Q400), as a qualified Flying Instructor, Line Trainer, Instrument Examiner, Crew Resource Management Instructor/Facilitator, human-factors safety investigator and risk/assurance professional.
Colgan 3407, Buffalo US, 12 Feb 2009
Bombardier Dash 8 Q400
Fatalities 50 (1 on Ground)
At 2300 ft altitude, during a night/IMC ILS approach to Buffalo Runway 23, the aircraft suffered an upset/LOC-I leading to a stall and it impacted the ground, in a residential area, near the ILS Outer Marker. In level flight, as they turned to intercept the LLZ (and possibly the approaching GS), neither pilot appeared to notice a low/idle Power Lever setting and reducing IAS, possibly selecting landing gear down too early. Neither did they react to the ensuing further speed reduction as the gear travelled until the stall protection systems activated.
On the activation of stick shaker and stick pusher, the actions of the PF were opposite to that which were required. Instead of applying forward neutral elevator and full power, the pilot applied power (not quite full power) BUT CRITICALLY pulled back on the yoke whilst using ailerons in an attempt to level the wings. This pull on the yoke further reduced the IAS and led to a fully-developed stall. The ensuing loss of control led to the crash. The PM raised the flaps without being commanded and then also the gear whilst in the steep spiral descent. The last action was possibly one of desperation.
Note – On the animation there are no actual values for power setting. The powers were not matched but this is not uncommon in the Q400, although it may have led to an amount of asymmetric power during the attempted recovery which could have exacerbated any roll/yaw. However, the effect was less important than the PF’s unexplained aftward nose-up inputs on the flight controls – which ultimately caused the LOC-I. Neither in the animation is there any indication of the proximity of ILS LLZ nor the GS, both of which may have been helpful in understanding the crew’s perception and situational awareness during the intercept.
NTSB Reported Causes:
The probable cause of this accident was the captain’s inappropriate response to the activation of the stick shaker and pusher, which led to LOC-I and an aerodynamic stall – from which the airplane did not recover in time before impact.
Contributing to the accident were:
- The flight crew’s failure to monitor the aircraft’s falling airspeed (despite multiple cues available) – indicating a slow instrument scan by the PF and ineffective monitoring by the PM.
- The flight crew’s failure to adhere to sterile cockpit procedures which led to reduced Situational Awareness (SA) and may have induced the above.
- The captain’s failure to effectively manage the flight by not insisting on less non-pertinent discussions and correct cockpit SOPs that would have improved awareness once below 10,000 ft. The full CVR transcript may indicate an overly ‘relaxed’ cockpit environment.
- Colgan Air’s inadequate procedures and checklists for the Q400 airspeed selection and management during approaches in icing conditions – which meant the crew did not match the aircraft Vref flying speeds with the selection of the anti-icing switches and which caused the low-speed cue (barbers’ pole) on the IAS ‘speed-strip’ on the EADI to be above the briefed approach speed (Vref) and ‘early’ activation of the stall protection systems.
The report focused on over 30 issues associated with:
- Flight crew airspeed selection and management in icing.
- Crew monitoring lapses.
- Pilot professionalism/leadership.
- Crew fatigue due ‘commuting’ and lack of sleeping facilities.
- Generic stall awareness and training procedures.
- Airline management of pilot training records and application of remedial training for those under-performing pilots.
- Federal Aviation Administration (FAA) oversight of Colgan during a period of expansion along with general criticism of pilot selection and monitoring.
- The lack of flight operational assurance programs such as FOQA/FDM and LOSA which may have highlighted issues with some company in-flight procedures.
- Use of personal portable electronic devices on the flight deck which may add to distraction.
- Better dissemination of safety and weather alerts provided to pilots.
The NTSB report was generally very detailed and wide-ranging and addressed the majority of lessons available. There were many findings and recommendations, but perhaps there were a number of unanswered questions or issues that the report did not fully address:
- Why did neither pilot notice the low power setting nor the ensuing low speed and further rapid decrease on the selection of the gear?
- Why did the PF not apply power in mitigation to an early gear selection that he called for?
- On getting a stall warning, why did the PF not immediately push forward with neutral ailerons and apply power?
- Why, instead, did the PF pull back on the yoke and then try to counter the stick pusher?
- Why did the PF try to level the wings by applying aileron inputs when the aircraft was patently in a stalled condition?
- Why did PM raise the flaps without being commanded?
Could the above be explained by the following – either in isolation, or as a linked chain of events and what lessons can we learn from them?
Loss of SA regarding the LLZ/GS
The report correctly pointed to the lack of cockpit discipline and extraneous chat that led to an overly relaxed atmosphere and that key actions were late (checklists and contacting the handling agents done too close to approach) or missed entirely (brief of icing speeds). However, the report could have assessed whether an additional outcome of their reduced SA was that the crew (mainly the PF) misplaced themselves in time and space on intercepting the LLZ. The call that approach was ‘armed’ and the LLZ was ‘alive’ were clearly stated by the PF but there was no mention of the GS being alive. However, the PF called for ‘gear down’ soon after turning on to the final approach track.
This seems like it was too early in the pattern – this is key issue and is important because, generally in Q400 operations, it is often the case that the gear is not selected until just as the aircraft intercepts the glideslope and descends (often enunciated by the PF). Doing this prevents the need to increase power markedly in level flight with the gear down and then taking it all off again in the descent on the GS. It is perhaps possible that the PF got ahead of himself and ordered the gear down BEFORE intercepting the GS (which could have been up to 1-2 miles further ahead) AND, crucially, he did not apply enough power to prevent the rapid speed reduction in level flight towards the stall (added to the higher low-speed speed cue for icing). Neither did the PM notice this (or at least, they did not say anything) – an intervention here could have delayed the selection of the gear (or added the required power) and prevented the LOC-I.
LESSON – As PF, be wary of losing SA with respect to the LLZ and GS intercept (or final descent point), especially if the winds are strong or across. As PM, try to pre-empt this common error – use DME/track distance as well as GS indications as a check of when to lower the gear. Quite often with good ATC vectoring, LLZ and GS intercept can happen at the same time and with a continuous descent so, often, a lot of things can all happen together and catch people out.
Pilots’ Instrument Flying (IF)
Irrespective of the crew’s lack of SA, a basic instrument scan by the PF should have alerted him to the rapidly reducing speed (he was also slow to note an increase in speed in the approach just seconds before – but he did eventually spot it and reduced the power to near idle. The NTSB report drew our attention to the PF’s long struggle to achieve proficiency in instrument flying, especially when manually flying (many of us will empathise with him). However, in this case the AP was flying, so he should have had a good grasp of things – the fact that he didn’t could have been because of a number of things; unfamiliarity with the aircraft type, complacency, distraction or fatigue but probably more likely was a combination of all. There could also be a another underlying reason.
Instrument Flying (IF) is a difficult skill that needs to be taught properly and practiced often. It is possible that the current pilot training programmes across the globe do not give new pilots a proper grounding in the ‘art’ of IF. Furthermore, the introduction of glass cockpits and speed tapes, EADIs/EHSIs and NDs in training aircraft may further have eroded the basic IF scan skills taught on the old ‘T Panel’. As a result of dependency on the automatics, it is possible that across the world, pilots’ IF scans have become lazy and less well practiced. Such a possibility has been highlighted in studies of pilots’ instrument scan techniques using eye-tracking technology. Results show that many pilots do not scan the IAS and vertical speed tapes often enough, nor always do they go through the ‘master indication’ of the ‘attitude dot’ in the centre of the EADI.
The lack of a similar monitoring scan when acting as PM is also apparent when completing checklists, setting radios, configuring the aircraft; quite often the PM obscures their screens by their checklist/documentation. When fatigued, distracted and stressed this shortcoming is likely to be even worse. This could be a better explanation as to why the crew missed the speed reduction. The NTSB report did recommend the inclusion of extra aural and visual cues to alert the pilots. However, this is only plastering over the cracks in the wall – and ensuring pilots are taught and practice better instrument flying and monitoring skills would be more effective.
LESSON – A pilot’s IF scan is unchanged from the days of the ‘T Panel’ – the ‘Master Instrument’ is the attitude indicator in the centre of the panel and you should ALWAYS scan other instruments (including the speed tapes, compass heading, ILS dots, AP modes and power settings) VIA the centre dot of the ADI. As PM, you should also monitor the ac performance as if you were PF and not obscure the EADI/EHSI/ND with your checklist/paperwork – which is why many aircraft often have checklists on the yokes themselves, which do not get in the way!
PF’s Speed/Power Awareness
One other question that could have received more attention in the report was ‘Why did the pilot not apply power after requesting gear down’? The power was near idle until 4 secs after the stall warner activated and the autopilot had disconnected – the levers hadn’t moved for about 15 seconds since the PF noticed he was slightly fast on the intercept heading (that is a long time). In addition to possibly having a lazy scan, there may be yet another angle to this. If you ask any experienced Q400 pilot, they will tell you that the power levers on the Q400 are exceedingly sensitive at the settings required on approach. The smallest of movements of the levers (say, caused by turbulence in cloud) can lead to large changes of the power and the IAS, making accurate flying difficult. Whilst not impossible to do, it is something that a number of Q400 pilots have issues with – so much so that many pilots will take their hands away from the power levers entirely on the approach.
This goes against the widespread belief of many pilots and trainers that the PF should have their hands on the controls throughout the approach. This is to ‘follow through’ the control actions of the automatics and to be ready to take control if the AP/AT disconnects (unless you are an Airbus pilot where this is discouraged because of an unshakeable belief that the Airbus is infallible). However, taking one’s hands from the power levers can create the human-factors risk of ‘pilot-machine disconnect’ which further reduces awareness of the aircraft performance. This is especially so in the Q400 which does not have an auto-throttle/auto-thrust and it would be interesting to know what the PF was taught during his training – both as a whole (and the level of experience of his instructors) and especially during his conversion to the Q400. Ultimately, if he did not have his hand on the power levers and was not scanning the power settings versus IAS as part of his overall scan, then it is no wonder the speed reduction went unchecked.
This is not a criticism of the pilot(s) alone but also on the design of the aircraft, the wider global pilot training system, the regulators and ultimately, the companies for whom both pilots had worked. This could be a clear example of James Reason’s ‘swiss cheese analogy’ – one that set up this crew for failure. (Author’s note – in the Q400 during turbulent approaches, once gear and flaps were set on final approach, many pilots with whom I worked took their hands from the power levers and put their hands around the base of the levers to avoid losing the pilot-machine ‘connection’ – small adjustments were possible without overcontrolling but once close to the ground, a proper grasp of the levers was re-established in preparation for the flare or go-around.)
LESSON – Watch the automatics (including the AP mode annunciations) like a hawk and be ready to take over. Follow through the controls if possible. If you do NOT have an auto-throttle – then being unaware of the position of the power-levers/throttles may kill you.
Stall Awareness Training
It is clear that the configuration/power/speed relationship was not being well-controlled or monitored prior to the stall, so now we come to the PF’s actions once the die was cast for the activation of the stall protection systems. The report says it was unable to ascertain why the PF did the exact opposite of what was required – and continued to do so until impact. The report points to an inappropriate ‘startle response’ as a possible cause but maybe they could have gone further? A startle response of such magnitude should have been preventable by regular and thorough training as defined in national regulations and ICAO Standards, so that as a pilot, one is never totally overwhelmed by events.
Whilst the NTSB did criticise the incorrect focus on minimum height-loss as part of the training, perhaps the report did not go far enough into the standards of pilot training. With an ever-increasing demand for airline pilots (until COVID), the regulators should ensure that a focus on cost savings and expediency does not dilute/erode the delivery of quality training. The NTSB correctly highlighted the undue emphasis on prevention of the stall and minimising height loss which had become the norm. If the stall awareness recovery training that the pilot experienced was either incorrect or not regular enough, again this may help explain the startle; but only if startle was the only reason for the PFs inexplicable actions.
– KNOW THE SYMPTOMS AND RECOVERY ACTIONS OF A REAL STALL. Read the following: FAA (now ICAO) “Airline Upset Prevention and Recovery Training Aid” (AUPRTA) and The Royal Aeronautical Society’s “Aeroplane Upset Recovery Training, History, Core Concepts and Mitigations”. These are MUST READ documents for all commercial pilots and nothing much in them should come as a great surprise to a well-trained pilot. (Links in the References section at the end).
– Contrast and compare the above with all your aircraft flight manuals. Remember that ANY aircraft can stall (including Airbuses) – and the only way to recover is to reattach the airflow over the wings.
– Do UPRT as often as possible – maybe even in an actual aircraft with an approved provider.
– If you do find yourself with the stick shaker and klaxons going off all-around you – DON’T PANIC – deselect the AP/AT and PUSH the controls forward (with ailerons neutral and possibly, power) until the airspeed increases and the stall protection systems have gone quiet – worry about height loss after you have regained control! Trim inputs may be required, so pilots need to make sure they can always reach the switches
Startle’ Response or Intentional Response to Stall Protection Activation
Whilst the NTSB report was correct to highlight the possibility of startle, this still leaves doubt in the minds of some investigators. It is certain that the PF and PM suffered some sort of ‘rude awakening’ because they had patently lost SA regarding the speed trend. Add in some cognitive overload caused by the activation of the stick shaker and AP cut-out, so one could understand a degree of shock/startle. However, there was no conflicting speed/attitude discrepancy (as in AF447), so it is hard to believe that after taking the quickest of looks at the instruments, seeing the low-speed ‘barbers pole’, that the crew would not have quickly recognised all the symptoms of the stall… unless they believed something else was happening.
Generally, people do things because they believe them to be right. This is key! The reports states that all Colgan pilots had received a NASA video brief that explained and demonstrated tailplane icing as part of their company Winter Ops Brief. Tail plane icing occurs because of the induced downwash over a main-plane with flaps extended leading to an increase in AoA of the airflow over a low-tail aircraft, and tail icing makes it possible for the tail to stall before the wings. There is no mention of the Q400 being subject to such a phenomenon (due it being a high T tail ac with de-icing equipment for the wings and the tail) and there is, as a consequence, nothing in the Q400 manuals or conversion training about tail issues.
However, the Captain was new to the Q400 and had come from the Saab 340 which was perhaps more susceptible to tailplane icing. Could seeing the NASA presentation have been misinterpreted by the pilots – almost becoming ‘negative training’ – and caused subliminal predispositions in the pilots’ minds – especially the Captain? The NSTB did not think the PF was trying to recover from a perceived tailplane stall but the pilot’s attempted recovery actions are consistent with this possibility (i.e. pull back on the controls to reduce tailplane AoA and raising the flaps). Could this offer a more-likely scenario? The PF’s previous experience on an aircraft type more-prone to tail-plane icing may have created a pre-disposition to consideration of a tailplane stall which kicked in as result of the startle experienced by the PF.
However, this doesn’t really explain the PM’s actions to raise the flaps without being commanded to do so. This was not fully explored by the NTSB, so, why did she do that? The crew had their first legs of the day cancelled and were in the crew-room for some time, so it is feasible that they may have been discussing possible actions if an aircraft experienced a tail stall when waiting for their departure on the last 2 sectors. Could they have been thinking about the flight to Buffalo and the poor weather forecast? Sadly, we will never know what the crew talked about during pre-flight or during their delay but they were certainly still discussing icing only shortly before the event.
Admittedly, this would have to be regarded as a ‘long leap’ but it is, perhaps, a better explanation of the crew’s actions rather than both completely freezing up or becoming irrational after being startled. But whatever the reason for the bizarre actions of the pilots, the aviation industry has improved and increased awareness of the startle response, pushed for more-professional adherence to better SOPs aligned to FOAQ/FDM processes; all of which are massive steps in the right direction. Also, in the US, tail plane icing discussions/training are only used for aircraft that are known to be susceptible.
LESSON – Pilots should consider ‘what if’ scenarios and what one might do in the event of an unexpected event/attitude/performance – BUT FIRST, pilots MUST do what has been trained in the simulator and mandated in the aircraft flight manuals and checklists – actions should be instinctive! Do you know these actions inside out? Can you instinctively name the correct checklists/call-outs and carry out ALL memory items without fail? If not – you need to practice!
Fatigue AND Illness
The apparent fatigue of the crew was cited as contributing to their reduced performance and lack of SA. It is hard to specify how tired both pilots were but both pilots did not get quality rest the previous night and the airline business was culpable in creating conditions in which such ‘crashing’ in crew-rooms and aircraft seats was accepted practice. Changes to working practices in North America (as result of the report’s recommendations) are most welcome. It is strange to see Europe, with the adoption of EASA’s new FTLs, being less robust over duty limits and practices in some areas.
Furthermore, it is clear from the CVR transcript that the PM had the beginnings of a serious head-cold which would have only further reduced her SA and ability to effectively monitor the captain on the approach – especially as her role was quite intense as they approached the LLZ. The NTSB did not think her illness had any substantial impact but others think it is very likely that sniffing and coughing/sneezing would have been a significant distraction to both pilots and, if the PM had been still trying to clear her ears (which is more difficult/painful in the last 2000 ft above ground), then it is more-likely that her PM role would have suffered and reduced the chance of her picking up the PF’s shortcomings.
LESSON – Get as much QUALITY sleep as possible and never fly with a cold/flu!
Deeper Pause for thought:
- Was the PM ever formally taught HOW to monitor PF actions and orders whilst acting as PM – or was this something learned by osmosis? Most airlines do now attempt to teach better monitoring skills but it might pay for pilots as PM to imagine they are actually flying the aircraft – where should we be heading, at what speed and what configuration? They should try to keep some sort of IF scan going – AND not to obscure the instruments.
- How does one retain situational awareness when IMC and when possibly fatigued/distracted? No briefs/admin/paperwork/chat below 10,000ft. PMs could use the ND more and try to visualise one’s position in the pattern – always ask “What is our next event?”. Think ahead for possible issues (other traffic, terrain, weather build-ups, cross-winds etc.)
- How do you draw someone’s attention to things that are incorrect (rapidly reducing speed)? First check and cross check all instruments make sure it is not your perception that is faulty! Probe gently at first, use one simple phrase “Check speed”. If no reaction, change tone – use ‘Captain’ (drop first names) and state the issue calmly using we/us in assertive but respectful tones: “Captain, we are 10 kts slow and reducing – we need to increase speed/power before…..”). How/when might you take control and should you? Think about this on the ground, so that you are not surprised in the air; trust no-one.
- Do you fly when fatigued and/or with a cold? A professional pilot shouldn’t! Report fatigue through your SMS/FRMS/MOR/ASRS if HR get aggressive!
- What characteristics of your aircraft are less than ideal? Does your company enmesh these into any threat error management processes you might already have in your company SOPs? If not, professional pilots should file a safety report.
- Does the company SMS/ASAP encourage crews to report any shortcomings of the aircraft and SOPs – if it does, do people do so and does change happen quickly enough? Professional pilots use them
- Does your company have a FDM/FOQA programme and are the results shared? Professional pilots read them and add issues to their mental list of threats.
- Does your company have a LOSA programme – do crews behave differently when an observer is aboard (like sometimes happens during a Line Check/Test?) Professional pilots fly every day like it is a check-ride, so they don’t get lazy.
- Is your simulator realistic, especially with respect to flight and handling in icing conditions? Professional pilots report pilot human-factor errors/issues that happen in the simulator – they might happen in the air.
Now, finally ask yourself 3 questions:
- Is the world of aviation any better for the NTSB/FAAs report and recommendations?
- Are all possible barriers and controls in-place and effective?
- Will they prevent a similar accident from happening again?
If the answers to ALL 3 are not all positively ‘YES’, then all pilots could start to think about they might do differently going forward – and even in the event of a recurrence. Please read and digest the references below as a first stop… then cross-refer them to what your company trains in UPRT and the simulator. They should be mutually supportive!
ASRS Air Safety Report System (US equivalent of MOR)
ASAP Air Safety Action Plan (US version of SMS)
CVR Cockpit Voice Recorded
DME Distance Measuring Equipment
EHSI Electronic Horizontal Situation Indicator
EADI Electronic Attitude Direction Indicator
EASA European Air Safety Agency
FAA Federal Aviation Administration
FDM Flight Data Monitoring (EU)
FRMS Fatigue Risk Management System
FOQA Flight Operations Quality Assurance (US equivalent of FDM)
FTLs Flight Time Limitations
GS Glideslope (ILS)
HR Human Resources
IAS Indicated Air Speed
IF Instrument Flying
ILS Instrument Landing System
IMC Instrument Meteorological Conditions
LOC-I Loss of Control – In Flight
LLZ Localiser (ILS)
LOSA Line Orientated Safety Assessment
MOR Mandatory Occurrence Report (EU equivalent of ASRS)
ND Navigation Display
NTSB National Transportation Safety Bureau
PF Pilot Flying
PM Pilot Monitoring
SA Situational Awareness
SMS Safety Management System
SMP Safety Management Plan
SOPs Standard Operating Procedures
UPRT Upset Prevention and Recovery Training
1. NTSB Full Report for Colgan 3407.
2. ICAO/FAA Airline Upset Prevention and Recovery Training Aid (AUPRTA).
3. The Royal Aeronautical Society paper – “Aeroplane Upset Recovery Training, History, Core Concepts and Mitigations”.