Investigators Say FlyDubai Crash Was Caused By An “Illusion”

Filed Under: Media, Other Airlines

On March 18 a FlyDubai Boeing 737 flying from Dubai to Rostov-on-Don crashed shortly after attempting to land, killing all 62 passengers and crew aboard. As we knew from the first day, the weather in Rostov-on-Don was horrible, with limited visibility and gusty winds. However, bad weather in and of itself doesn’t take down planes.


The pilots knew about the bad weather before departure, and loaded plenty of extra fuel in anticipation of some possible missed approaches. Around the same time time as the FlyDubai plane attempted to land, several other planes also had go arounds, and some ended up diverting to other airports.

In the case of FlyDubai, the plane attempted to land once, but had an aborted landing due to the conditions. Then they circled for about 90 minutes hoping for the conditions to improve. Then they came in for a second approach hoping for improved conditions, and they ended up aborting that landing as well. That’s when things got bad, as the plane crashed just moments after the go around.

Below is the footage of the plane going down at what appears to be a nose down angle. It’s horrifying to watch.

As usual, aircraft accident investigations can take years to complete, and intermediate reports are released every so often. Last week the most recent accident report was released, and it rules out a mechanical failure or violent weather as being the cause of the accident.

Popular Mechanics has a great explanation of what happened to the downed FlyDubai plane, according to the most recent accident report. Basically the crash was due to the pilots being disoriented, a phenomenon known as “somatogravic illusion.” Here’s the explanation:

During a go-around after an aborted landing, a plane tends to be lighter than normal since it’s at the end of its flight and has burned up most of its fuel. That means its thrust-to-weight ratio is relatively high, so when the pilot pushes the throttle forward from idle to full thrust the plane accelerates with unusual alacrity. This acceleration pushes pilots back in their seats, which to the inner ear feels exactly the same as tilting upward.

Black-box data show that as the plane started to enter the cloud after the second go-around, the flight crew briefly pushed the controls forward so that its rate of climb decreased, as if the pilots were momentarily disoriented. Then the plane returned to its previous rate of climb. For a few seconds, all was normal. The flight crew members were almost certainly following their instruments, as years of experience had taught them to do. Then, as if suddenly disoriented and unable to believe their instruments were correct, the flight crew pushed the stick far forward.

The pilots probably believed they were preventing the plane from getting too nose-high, which could cause the plane to stall and crash. But in reality they were taking a safe situation and turning it deadly. The lurch downward would have caused them to rise up in their seats as though on a roller-coaster zooming over the top of a hill. By the time they rocketed out of the bottom of the cloud and gained a visual sense of their orientation, they were in a 50 degrees vertical dive at more than 370 mph and just a few seconds from impact. There was no time to pull out.

In conjunction with the other factors involved, including horrible weather and fatigued pilots, that explanation certainly makes sense.

I’ll be curious to see the final accident report…

  1. @Kris – not at all. That was an engine flame out at altitude, followed by autopilot applying max roll, crew failing to establish coordinated flight after disconnecting autopilot and then mistrusting the attitude indicator due to spacial disorientation. By that logic, this acciden is similar to [m]any other caused by spacial disorientation.

  2. Dima, care to explain why you think it’s nonsense? The explanation sounds reasonable, is in accord with the evidence, and has precedent.

  3. This has always been the most likely cause of this accident- Disorientation due to fatigue and somatogravic illusion leading to incorrect pilot control inputs. No stall at any point.

  4. I recently ended a 12 year stint as a professional military pilot, and illusions like the somatagravic can really challenge you up as a pilot since your body is telling you something completely different than the instruments. I have experienced its effects many times, almost always on an initial climb out or on the go after and touch and go landing. It is disorienting, but pilots are trained to deal with it and press on. Unfortunate result, but this is a well known and studied phenomenon. You can simulate something like it using a swivel chair and tilting your head so it is parallel to the ground. Then closing your eyes and spin the chair slowly for 30 seconds or so. Finally, open your eyes and lift your head to a normal position and you will feel very disoriented as your eyes and body disagree on your motion.

  5. So can systems be designed that prevent pilots from angling the nose down under circumstances that are most likely to invoke the somatogravic illusion?

    The somatogravic illusion is the result of our own internal accelerometers. Electronic accelerometers can determine when the effect is likely being experienced. Then, coupled with other information, the computer could prevent the pilot from doing something irrational in those cases.

    Seems to me that this is enough of a problem – several crashes have happened as a result of it – that preventing it should be a priority for plane manufacturers and regulators.

  6. As Justin said, pilots undergo extensive training in the psychological and physiological effects of instrument flying even before they begin their instrument training. In order to get your Private Pilot license, which is VFR only, you have to log a few hours of “hood time” (the hood is a shield worn by the pilot to obscure their view out the windows) including Unusual Attitudes, where the flight instructor has the student close their eyes and then puts the aircraft through a series of steep climbs, banks, and descents in order to disorient the pilot, then have them open their eyes, analyze the instruments, and bring the aircraft back to level flight. In short, this is basic stuff that we’re trained on almost from the beginning.

    I’m not trying to say the findings in the Popular Science article are rubbish. After all, this wouldn’t be the first time seemingly experienced but fatigued pilots killed a bunch of people when they failed basic piloting skills. Colgan Flight 3407 is a prime example. What I find hard to believe is that BOTH pilots simultaneously failed Basic IFR 101 at the same time. More likely there was a total breakdown of communication between PF and PM. I would like to see an analysis of this crash from a Crew Resource Management perspective. I suspect there was a sizable authority gradient between Captain and FO, effectively erecting a brick wall down the middle of the cockpit. Often times such failures can be traced back to corporate culture, and the reports we’ve heard indicate that FlyDubai does not cultivate a safety focused culture…at least as far as their chief pilot is concerned. In other words, I suspect this crash is the result of a much, much bigger problem.

  7. @snic- it is true that airplane avionics are not subject to the confusion that humans are. In fact, one option available to pilots in situations like this is to simply engage the autopilot and let it orient the aircraft correctly. Although, a lot of accidents start with other minor issues which make using the autopilot not recommended, like an avionics fault or sometimes poor weather conditions like icing or turbulence which depend on the limits of the specific system.

    I am sure it is technically possible to have the aircraft ignore pilot commands to prevent mishaps like the one described by the Popular Mechanics article. Many (and probably most) airliners do have EGPWS which knows where the ground is based on a database loaded into the avionics. As far as I know, these systems only give verbal and visual warnings to pilots, but I would imagine that someday a further enhanced layer of safety could be added that would make it impossible to hit the ground… so long as the avionics were functioning correctly.

  8. @Chuck: Read the Popular Mechanics article. It is based on the preliminary investigation by Russian aviation authorities. It is not just speculation.

    @Mark and Justin: Thanks for your answers. I find this fascinating – I didn’t realize that “hood time” was part of pilot training. That’s great that there’s a systematic way to train pilots to ignore what their sensory systems are telling them, and it’s also great that it works 99.999% of the time. It’s the other 0.001% that I’m worried about.

  9. I remember an experience during my instrument training (270 hours PP-ASEL-IA) where I had somehow moved my head up and down from the panel to the charts while executing a pitch maneuver and I could *swear* that we were flying in a 45 degree roll. As I started to correct for that, my instructor patiently said “No. Cross check the instruments. If they cross-check, you are flying correct, so IGNORE YOUR BRAIN.” I really appreciate those experiences, but if I had been at the end of a long flight, I might not have remembered that.

Leave a Reply

If you'd like to participate in the discussion, please adhere to our commenting guidelines. Your email address will not be published. Required fields are marked *