[UPDATE: See Comments for reference to NTSB report for similar crash of Atlantic Coast Airlines in 1994]Last week’s aviation news was dominated by the NTSB hearings that sought the cause of the Colgan Air Flight 3407 crash that destroyed a home while en route to Buffalo, NY. Info on crash details was released on the NTSB website and on a YouTube video that shows a reconstruction of the final minutes leading up to the crash. Throughout the hearings, pictures emerged of pilots who were fatigued and lacked key training, a captain who failed to disclose several checkride failures, and of a two-tier industry where regional airlines pay new-hire pilots little more than minimum wage.
When the news initially broke in February, dramatic pictures focused on...
the fire that consumed a house and early speculation was that icing was the cause of the crash. Yet the latest details suggest that the cause of the Flight 3407 crash was something every student pilot learns within the first 15 hours of pilot training: If you get too slow on final, the airplane will stall and crash. The lesson for all pilots is simple. Airspeed is life and even a two-pilot crew, each with thousands of hours of flight experience, cannot repeal this law of nature. What’s less clear is why the pilots let the aircraft stall and we’ll explore that shortly.
The Captain: A Two-Sided Story
My initial direction in a quest to understand the cause of this crash was to get more details about the Captain. Most articles gave few details, but one website went into depth. Marvin Renslow, 47, always dreamed of becoming an airline pilot. Born in a small Iowa town, he left home after high school and began working first at a Howard Johnson’s hotel and later in the reservation center for Piedmont Airlines (now US Air). He earned a 2-year aviation degree and his private pilot license in the early 1990’s. He then worked for American Express as a travel agent and later for Verizon as a sales consultant. Around 2003, he took a company buyout and used the money to further his flying ambitions.
He enrolled at Gulfstream Training Academy to train on the Beech 1900 turboprop. Later he flew as a first officer for Gulfstream International Airlines which, according to Wikipedia.org, at one time required low experience pilots to pay over $32,000 to ride in the right seat of a turboprop in their “pay to work” program, which apparently has since been discontinued. With 625 hours of total experience, he was hired at Colgan Air in 2005.
Two pictures emerge of him as a pilot. "He was always by the book," said Jeff Linquist, a former roommate and a pilot with a private license. "There's a lot of guys out there that do fly by the seat of their pants, but he wasn't one of them." Yet, according to the NTSB, he failed four checkrides, including his instrument airplane, commercial single engine, commercial multi-engine and ATP checkrides.
According to another report, “National Transportation Safety Board records released Tuesday say investigators were told by one training instructor that Flight 3407's captain ‘was slow learning’ the Dash 8-Q400 Bombardier, a twin-engine turboprop. But it said that pilot Marvin Renslow's abilities ‘picked up at the end.’ The training instructor said Renslow struggled to learn the Dash 8's flight management system, a critical computer."
Yet in reflecting on the captain’s background, I’m hard pressed to conclude that any of the failed checkrides or other issues contributed to the crash in a material way. Ultimately, he corrected each of the deficiencies and was qualified to fly the airplane. It seems likely that more fundamental issues—ones that can affect any pilot—were to blame.
Fatigue
While some family members have insisted that the pilots weren’t fatigued, evidence suggests otherwise. The captain traveled to Newark the night before and apparently was logged onto his computer until 3 AM and slept in the crew quarters, which was not intended for sleeping. The first officer took an overnight red eye flight from Seattle and is unlikely to have had a full night’s sleep on the plane. Both pilots were heard yawning shortly before the crash. Given that they were talking about the large amount of ice they had accumulated, it’s unlikely they were bored.
Distraction
In retrospect, every time I’ve made a major error while flying, I later realized that I had become distracted. This crew may have been distracted by the accumulated ice, non-essential chit-chat, and radio calls. The first officer asked shortly before the crash “Is that ice on our windshield?” The captain replied “oh yeah that's the most I've seen—most ice I've seen on the leading edges in a long time. In a while anyway I should say."
I'm guessing the icing may have left the pilots nervous. Just as people walking by a cemetery at night will whistle to ward off the spirits, this crew’s chatter may have been a coping mechanism for dealing with the ice. The first officer said: “I've never seen icing conditions. I've never deiced. I've never seen any—I've never experienced any of that. I don't want to have to experience that and make those kinds of calls. You know I'dve freaked out. I'dve have like seen this much ice and thought oh my gosh we were going to crash."
Less than five minutes later, the aircraft crashed, though not as a result of icing. Yet the discussion may have planted the seed in the captain’s mind that if they were to encounter a problem, it would most likely be the result of icing.
Final Minute: Loss of 60 knots and a Stall
Watching the NTSB’s Colgan Air Crash Animation of Flight 3407 is instructive. Note throughout the entire video that the power is set at flight idle, yet the aircraft is level at 2,300 feet. Then, starting at 22:16:00 local time, 60 seconds before the time of the crash and the end of the video, a chain of events is set in motion which causes the aircraft to rapidly lose airspeed and stall.
Initially the aircraft was flying at 180 knots. Then the captain called for “gear down” just as approach control handed the aircraft off to the tower. The first officer responded “gear’s down” and the captain called for “flaps 15” degrees and for the before landing checklist. In the 30 seconds it took for this to occur, the aircraft slowed to 125 knots and the stick shaker activated, indicating the aircraft was on the verge of a stall.
Sadly, rather than responding properly to a stall by pushing forward on the yoke, the captain pulled back on the yoke, perhaps because he thought that the aircraft was having an icing induced tailplane stall. This exacerbated the stall causing the wings to drop alternately four times before spinning into the ground. The video shows the captain responding properly to each of the wings dropping by applying full opposite rudder, except for the last time, when he left the rudder in the neutral position.
Lessons for Pilots
The lessons for pilots are many. First, airspeed is life. Pilots should closely monitor airspeed throughout the approach to landing. Anytime an aircraft is level and the autopilot is in altitude hold mode, lowering the landing gear and flaps will slow the aircraft. Pilots must compensate by increasing power; otherwise the autopilot will slow the aircraft to a stall. I saw this happen a couple of months ago while giving instrument instruction. While in the clouds and turning onto the final approach segment, the pilot set the power too low and failed to notice the speed decrease. After he failed to respond in a timely fashion, I took over, disconnecting the autopilot and lowering the nose just as the stall horn was starting to come on.
I was surprised to see an experienced pilot make that mistake, but it was a good reminder that we all have to be vigilant during critical phases of flight. Sadly, the pilots of Colgan Air flight 3407, perhaps because of fatigue, distraction by icing or idle chatter, failed to notice their airspeed decrease during a critical 30-second period. It’s a sobering reminder that even good, experienced pilots have to do certain things right every time they fly.
After being an airline captain for 20 years, I would like to interject this scenario on Colgan 3407
The Autopilot was ON, and the Thrust Levers were close to idle during the configuration changes. If at a constant altitude the AP would have held the altitude by trimming and caused the aircraft to slow. While it is doing this, it is trimming the nose up. When at the Stick Shaker, the AP automatically is disconnected there could be a large amount of UP Trim. When it disconnects, and especially if power is introduce there would be an extreme nose up attitude change, and if the pilot is out of the loop he would not react to this and the nose would rise extremely fast. I believe this is what happened, and that "the pilot did not pull the nose up." It went up and he didn't realize what had happened.
When I was at Atlantic Coast Airlines we had a Jetstream 4101, NTSB Docket # DCA94MA027 this Accident occurred Friday, January 07, 1994 in COLUMBUS, OH and the same thing happened. It was almost an identical situation in every way (Please Read it).
As you may have heard during the NTSB hearing, Colgan is NOW teaching its pilots to do AP ON stall training. Atlantic Coast Airline instituted that policy soon after our crash. This is a maneuver that you must be trained on, or you will not react the correct way IF it should happen.
This is a training scenario that would apply to ALL autopilot equipped aircraft.
Posted by: Elbert "Mac" McCormick | May 18, 2009 at 04:09 PM
Hi Max, thanks for the excellent analysis with many various intriguing questions. I know nothing about airline operations but I can comment on the pilot training (or lack thereof) that is seems apparent here. I think, unfortunately, that this accident is only the obvious inevitable result of a pilot training system that is broken. All these "flight academies" with "guaranteed pilot rating" for a fixed price are cheating the system and everyone is pretending it isn't happening. Even major (reputable?) safety oriented magazines enjoy considerable revenue from their advertising while pretending nothing is wrong.
I have been a DPE for years and acknowledge that anyone can bust a check ride for all kinds of reasons. I think five failures surely indicates an unhealthy pattern. It also demonstrates that despite the best effort of examiners to say "no," a determined poor performer can continue to progress. Whether this pilot improved or not will probably go undetermined.
I recently had to disapprove an applicant for a multi-engine commercial evaluation who was absolutely ignorant of the whole body of aviation knowledge that even a private pilot should possess. This was alarming because I knew he had succeeded already in three evaluations where he was dependent upon this same (basic) knowledge. I also knew that if he passed this evaluation he was soon in the right seat of a regional. Maybe he went to another FSDO and passed a week later? The system is only as good as the weakest link unless there is the added redundancy later up the line of someone catching an impostor (someone without the requisite skill and knowledge). Presumably the airline check airmen should filter out this guy (but they are desperate to fill seats up front?)
In the GA world the flight review is the only subsequent "filter" for pilot quality and often is a useless "rubber stamp." Integrity demands every pilot must absolutely police our own and speak up when we see unsafe performance. If you cannot do this directly, your local FAAST rep is a good place to start. (Friends don't let friends fly stupid!)
Posted by: David St. George | May 18, 2009 at 06:57 PM
After studying the cvr transcript I'm puzzled about a couple of things with Flight 3407.
The Dash8-Q400 has an Aerazur Pneumatic De -icing system, Ice Detection system & Windshield Anti-ice controller system.
Why would the windshield have ice on it and why would the captain mention "ice on the leading edges" with the wings protected with this system? What other leading edge would he be talking about?
Also, any comments on the many seconds of silence between cockpit comments? For people who were communicating full time a few seconds earlier, they suddenly just stopped talking when faced with a challenge?
Please forgive me if these questions would be obvious to a pilot, am not one. Thanks for your time.
Posted by: Barry Williams | May 21, 2009 at 11:03 PM
There are many quotations and words of wisdom from which we can summarize important aspects of flight. One of the very first I ever heard, spoken somewhat lightheartedly, was “Let not thy airspeed fall too low, lest the Earth rise up and smite thee!” My very favorite such quotation however, more serious and philosophical, goes something like this… “Flying is not, in and of itself, inherently dangerous. However, to an even greater extent than the sea, it is extremely unforgiving of any carelessness, incapacity, or neglect.”
As a former regional airline captain with thousand of hours flying the Dash-8 (200, not 400), I have followed the NTSB investigation into the Colgan Air accident with great interest... and sadness. Most of that sadness comes from my thoughts of the senseless and needless loss of life for the crew and their passengers, as well as the tragic consequences for their families and friends. Another part is the result of the personal sense of loss I always feel, due to the “camaraderie” among pilots, whenever we lose one of our own. Along with that, it is also always so disappointing when pilots’ failures to realize they have committed the most basic of errors, and disregarded and/or been distracted from following the most basic of operating procedures, combine to result in an accident. I always feel that any failure of any pilot which results in an accident reflects negatively upon us all, and that we all have an obligation and duty to always attempt to study and learn from the mistakes of others lest we repeat those mistakes ourselves. If we happen to also be in the position of flight instructor and/or check airman, or even that of fellow crewmembers, we have an even greater responsibility, and that is to be our brothers’ (and sisters’) keepers as well.
If I may, I would like to make a few points regarding the accident, Max’s article, and some of the comments posted. It seems from the beginning of the NTSB animation video that the captain might have had some issues with airspeed control, and airspeed control should have been very easy in the beginning since they were clean, level, on a vector to the final, and on autopilot. The video starts with the airspeed at 189 KIAS, but it then bleeds down to about 167 KIAS in 20 seconds before he adds in a bit of power to get it back up to around 180 KIAS. At 167 KIAS, they are just 14 knots above the minimum approach speed in icing conditions of 153 KIAS with Flaps 5 (per the Q400 AFM), and they weren’t at Flaps 5… the flaps were still up. Perhaps the autopilot had just finished leveling off from a descent to 2300 MSL, slowing the aircraft in the process, but if the desired speed during vectoring to the final was 180 KIAS then he was late on the power application during the level-off. Anyway, their minimum approach speed at their weight in icing conditions should have been 153 KIAS at Flaps 5, 144 KIAS at Flaps 10 (where they were later incorrectly set by the FO), and 138 KIAS at Flaps 15 (which is where the captain asked for them to be set). These speeds all include a 20-knot increase above the normal speeds as a safety cushion because of the icing conditions (see the NTSB’s Factual Report on this accident).
Approach mode for the autopilot had been armed while on the last vector, and the captain calls for gear down at the same time he says that the localizer is “alive”. It is a pretty standard procedure throughout the industry when on an ILS approach to not lower the gear until beginning to intercept the glideslope (when it’s about a dot above the aircraft’s vertical position) as this helps start the aircraft’s descent to follow it without the airspeed slowing much. Since at this position the glideslope was still well above them, one might wonder why he chose to put the gear down here. It may be that he simply wanted to slow down and configure the aircraft a bit early so as to be “stabilized” once they were on the localizer centerline and before starting their descent on glideslope. It is quite common for pilots with minimal time in a particular aircraft type to do this so as to stay ahead of things and not feel “rushed”, and he only had about 110 total hours in the Dash-8. Shortly thereafter either he or the FO pushed the condition levers full forward to Max, which flattens out the prop pitch on the Dash-8’s huge propeller blades taking them to maximum RPM for the remainder of the approach, so as to be prepared for either a potential max power missed approach climb or maximum drag deceleration (including reverse thrust if desired) on the rollout after landing. He then calls for Flaps 15, but the FO only sets them to Flaps 10.
Making all of these configuration changes increases drag tremendously, and without adding power it’s like throwing out a huge anchor with a short chain. The net result of maintaining altitude while extending the Dash-8’s enormous landing gear, setting the condition levers to max, and extending flaps to 10, all without adding ANY power, is that the airspeed decreased from about 180 KIAS to less than 130 KIAS in 20 seconds and kept on decreasing to as low as 78 KIAS during the stall sequence. Remember, the minimum approach speed at Flaps 10 in icing conditions was supposed to be 144 KIAS, and in normal conditions it would still be 124 KIAS. Contrary to what another commenter said, I don’t believe that the aircraft automatically pitched up when the autopilot disconnected in response to the stick-shaker activation OR when the captain subsequently added power during his attempted recovery. While it is true that the autopilot would have been trimming nose up to maintain altitude during the deceleration, the trim would not have automatically somehow increased pitch even more during the disconnect. Further, while it is true that many aircraft do pitch up when power is significantly increased, the Dash-8 has a very high “T-Tail” with the horizontal stabilizer and elevator in a position where pitch is affected very little by increased thrust airflow with power increases. Again, I am only speaking from my experience in the Dash-8-200, having no experience in the Q400, but they both have that tall “T-Tail” and I’m pretty sure the aircraft are similar in this aerodynamic trait. I also don’t believe this captain responded improperly because he thought he had a tailplane stall. There is absolutely no evidence whatsoever that this was the case, including his never having received any training from Colgan Air about this other than it being mentioned in a video, and his actions didn’t follow the correct procedure for that situation either. I believe the captain DID pull back on the yoke in response to the stick shaker and stick pusher, an action which would seem intuitively correct to a non-pilot, instead of properly recovering as every fixed-wing pilot has been trained to do. Was he simply “surprised” and panicked, responding in a manner contrary to everything he had been taught? We’ll probably never know.
We can discuss ad infinitum what other training might have been helpful in preventing this accident, like recovery from stalls while on autopilot, with stickpusher activation, or including recovery from full, “deep” stalls, and more training certainly never hurts. However, the basic stall recovery procedure is the same for every fixed wing aircraft in the situation in which the crew of Flight 3407 found themselves and EVERY pilot should know it… decrease the angle of attack and increase the power. The fact is that any pilot certification beyond Private only requires recovery from “imminent” stalls… and for good reason. Professional pilots, whether they are Commercial pilots in single-engine Cessnas or 20,000-hour ATPs with a B747 Type Rating, are supposed to be proficient in their knowledge, understanding, planning, and execution of their flight duties. They, their employers, and their fellow crewmembers (by at least reporting observed problems to a training pilot) are supposed to see to it that they ARE proficient. If they are proficient, they should never get the aircraft to the point where an imminent stall, much less a full and/or unrecoverable stall, is even a possibility.
In my experience, the Dash-8 is an unbelievably great performing airplane in every way except perhaps max cruise speed on some models. It’s not the very fastest turboprop around, but it loves to fly… and the -200 model I flew was difficult to stall even when you tried to make it stall. This particular -Q400 aircraft seems to have performed no differently. I watched that video animation over and over again, looking for any problem that might not have been related to improper handling by the crew, and I just can’t find anything at all. The aircraft actually climbs 200 feet during the stall sequence before it starts really descending, recovers from a knife-edge 90-degree right bank, and then keeps only “mushing” even after the FO inexplicably retracts the 10 degrees of flaps that are probably helping the aircraft to stay in the air at that point. Had they simply reduced the angle of attack by getting the yoke forward and then recovered from whatever altitude loss they had allowed in the stall recovery (the ground elevation was less than 700 MSL so they had plenty of altitude to give up), their passengers would have had a wild, scary ride to talk about but everyone would have survived. Better still, had they followed proper procedures, including the FO calling out and insisting on correction of any airspeed deviations of plus or minus 10 knots from normal, perhaps the captain would have realized his error and added power before the airspeed decay from dirtying-up the airplane would ever have gotten to the danger point.
I respectfully submit that the bottom line appears to be that the reason this accident occurred was solely because this crew seems to have done just about everything wrong (except adding power, and perhaps retracting the gear too late to do any good) once the aircraft had slowed into the airspeed “danger zone” related to the stall. Although the captain had about 50% more total flight time than the FO, the FO had seven times the captain’s time in this aircraft type. It truly appears to have been a CREW failure and, as usual, there were multiple links in the accident chain of events. Had they broken any of those links by avoiding any of their errors, it would probably not have occurred. They were certainly both trained and experienced enough that this crew should have avoided this accident.
While we all know that anyone of average intelligence can be as excellent pilot with the right training and the right attitude, I also know from my years as an instructor and captain that some people just don’t have “the right stuff”. I agree with Mr. McCormick’s suggestion that, although it is absolutely true that even the very best pilot can fail a checkride for any one of a myriad of reasons on any given day (they say there are only those who have and those who will), the captain having “failed” six of them in less than 3400 total flight hours is definitely an unhealthy and dangerous pattern. According to the NTSB, he failed his initial instrument airplane checkride in 1991, failed his initial commercial single engine land checkride in 2002, failed his initial commercial multiengine land checkride in 2004, required training to proficiency on his initial proficiency check in the Saab 340 as a FO with Colgan in 2005, failed proficiency training requiring a re-qualification ride in the Saab 340 as a FO with Colgan in 2006 (after over a year of experience as a FO on this aircraft, unsatisfactory tasks included rejected takeoffs, general judgment, landings from a circling approach, oral exam, and non-precision approach… that’s a lot of unsatisfactory areas), and failed his initial ATP and Captain checkride in the Saab 340 with Colgan in 2007. Perhaps the instructors, check airmen, and fellow crewmembers who played a role in the captain’s flying career just weren’t effective enough, or perhaps he was simply not dedicated enough to his own proficiency despite their diligent best efforts. Unfortunately, his life ended sadly and tragically, along with the lives of his crew, his passengers, and an occupant of the home the aircraft hit on the ground, when he failed to successfully handle his final test as an airman.
We pilots are all human and, as living beings, we all make mistakes. While the luck of the draw certainly plays at least a small part, in my experience and belief, the best course to avoid accidents is to unwaveringly dedicate oneself to trying to always pay attention to all of the details and adhere to proper operating procedures. Know your aircraft and its systems well, and be highly suspicious of anything that doesn’t seem right… and seek out the answer as soon as it doesn’t seem right. You’ll still make mistakes, but hopefully you won’t make enough at any one time to complete that chain of actions, inactions, and events that results in an incident or accident… especially one wherein lives are lost. Again… “Flying is not, in and of itself, inherently dangerous. However, to an even greater extent than the sea, it is extremely unforgiving of any carelessness, incapacity, or neglect.”
Posted by: Mike Finkle | June 06, 2009 at 03:18 AM
As a response to Mr. St. George,
Although the aircraft's deice systems DO protect the windshield and wing leading edge from ice buildup, the systems don't protect every single square inch of either. Electric heating elements prevent buildup on and remove ice from the center section of each windshield glass, but the ice still builds up all around the outside edges. The majority of the wing leading edge is also propected, but ice still builds up (and is visible in most areas) immediately above, below, and between the wing deice boot segments.
As far as the silence between comments while on the approach, Federal Aviation Regulation 121.542(b) states, "No flight crewmember may engage in, nor may any pilot in command permit, any activity during a CRITICAL PHASE OF FLIGHT (my caps) which could distract any flight crewmember from the performance of his or her duties or which could interfere in any way with the proper conduct of those duties. Activities such as eating meals, ENGAGING IN NONESSENTIAL CONVERSATIONS WITHIN THE COCKPIT (my caps) and nonessential communications between the cabin and cockpit crews, and reading publications not related to the proper conduct of the flight are not required for the safe operation of the aircraft". Section 121.542(c) states, "For the purposes of this section, critical phases of flight includes all ground operations involving taxi, takeoff and landing, AND ALL OTHER FLIGHT OPERATIONS CONDUCTED BELOW 10,000 FEET (my caps), except cruise flight." This entire section 121.542 is known as the "sterile cockpit rule", which is basically just legally requiring pilots to follow what is obviously good common sense by discussing nothing but necessary information when they are in a flight regime requiring their full attention... like while this crew was on this approach. Unfortunately, they didn't do all of the talking they should have been doing, like making at least a couple of required callouts and acknowledging completion of some items, during this sequence of events. However, the relative silence in this phase of flight IS normal and correct.
Posted by: Mike Finkle | June 06, 2009 at 04:04 AM