Friends have been emailing to ask why AirAsia 8501 was climbing at 6000 feet minute (BBC News) before crashing. The transport minister said “No passenger or fighter jet would attempt to climb so fast.” What is missing from these articles is that an American thunderstorm can generate updrafts of 4000 to 5000 feet per minute (see this meteorologist’s presentation to an FAA group). A tropical thunderstorm has even more energy and could presumably generate the full 6000′ climb rate even without any pilot action. Of course, it is also possible that the pilots were trying to out-climb the cloud and the rate of climb was a combination of updraft and pilot action. This is suggested by the subhead: “climbed too fast before stalling.” An aerodynamic stall occurs when a pilot pulls the nose up too high relative to the airstream in an attempt to climb, thus creating an “angle of attack” greater than about 16 degrees. This is what happened to Air France 447 (note that the pilots in that incident could have saved the airplane during the first minute or so simply by pushing the stick forward).
You might ask has anything like this ever happened to me? The answer is “sort of”. I once departed Teterboro, New Jersey, which is in some of the world’s busiest airspace and therefore a place in which ATC instructions need to be followed precisely to avoid the risk of collision. I was flying a Cirrus SR20, which doesn’t have the onboard weather radar of a jet. There were some scattered cells of rain in the area and ATC vectored me directly into one. Soon the Cirrus was climbing at 2000 feet per minute, despite having cruise power and pitch angle selected. I pulled the power back and pushed the nose down for a 1500 fpm descent, concentrating on keeping the wings level in the turbulence and the airspeed moderate so as to keep the stress on the airframe low (going fast in turbulence can cause things to bend or crack). The result was a net 500 fpm climb and I told ATC that I was unable to maintain my assigned altitude. Eventually the plane came out the other side and we were never at risk, but it was a sobering reminder that while the pilot has near-complete control of aircraft attitude that is not the same thing as having complete control of the aircraft’s position in the sky. Note that this was in the winter during an unusually warm spell and none of the clouds would have qualified as a true thunderstorm. The AirAsia pilots had a much more capable airplane but also much more challenging and violent weather.
Do a web search for “thunderstorm updraft airplane” and you’ll find a lot of sobering articles on the subject.
Related:
“(note that the pilots in that incident could have saved the airplane at ANY TIME simply by pushing the stick forward)” (emphasis added)
Incorrect. Apparently there was eventually such a loss of airspeed that there was not enough airflow over the control surfaces to get the nose down. The full throttle on the engines also induced a nose up movement, which full nose-down trim *and* stick forward would have needed to overcome. Throttle back, stick (and trim) forward was apparently necessary and even that wasn’t enough after about 18,000 feet. Simulations are apparently a little hazy on when the threshold of “unrecoverable” is for AF 447.
Phil,
Though this is only barely incidental you may enjoy it.
The publication AviationWeek recently interviewed a NASA team working on a ground based copilot project (out of what looks like a glorified janitors closet).
https://www.youtube.com/watch?v=rgdcdj5ietg (5 minutes)
They call it “Super Dispatch Mode” and it resembles your posting on the idea from 2008. http://philip.greenspun.com/flying/ground-monitoring
Unlike your idea though they are developing the concept for airliners as a replacement for a copilot instead of as a supplementation for single pilots as you more reasonably wrote. So it appears the NASA team may be moving out of the utility closet later than they could be.
Google just returned three links about this subject. One is a WSJ piece mentioning a $4M/4yr contract NASA awarded to Rockwell Collins in Dec 2014. The AviationWeek video seems like it is of this. The other two are NASA papers. I was excited when I found the two papers because the possible telemetry technologies which would enable a ground based service like this could be interesting. Unfortunately the papers only blandly cover Crew Resource Management stuff.
(1) http://www.wsj.com/articles/single-pilot-cockpits-floated-in-nasa-study-1418611930
(2) http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20120016028.pdf
(3) http://human-factors.arc.nasa.gov/publications/Lachter_etal_2014_2.pdf
Thanks, Colin, for the correction (I’ll fix the original posting). I hadn’t realized that the analysis you describe had been done. It is kind of amazing that the Air France airplane never entered a spin and yet was still very nearly impossible to recover.
Additionally, with $4M students could probably buy an SR20/22 and route the flight control indications, switch positions and GPS coordinates into a datalink and create On-Star for airplanes. In another world after a copyright and demonstration Cirrus may put up additional money for certification so operators could, like you mentioned, remind pilots to turn their landing light on when they are approaching an airport or whatever. In this world though FAA bureaucracy seems too risk-averse for anything but incremental improvements.
Colin: http://www.vanityfair.com/business/2014/10/air-france-flight-447-crash says that “Though precise modeling was never pursued, the investigators later estimated that this was the last moment, as the airplane dropped through 13,000 feet, when a recovery would theoretically have been possible. The maneuver would have required a perfect pilot to lower the nose at least 30 degrees below the horizon and dive into the descent, accepting a huge altitude loss in order to accelerate to a flying angle of attack, and then rounding out of the dive just above the waves, pulling up with sufficient vigor to keep from exceeding the airplane’s speed limit, yet not so violently as to cause a structural failure.”
The article also indicates that, at least to that point, the airplane did respond to stick forward (the angle of attack would be reduced to the point that the stall warning would sound).
I love William Langewiesche’s writing, especially about flying.
I read Bill Palmer’s book. It went into *excruciating* detail mostly for an aviation audience (whereas Langewiesche is obviously writing for people who like to read about rich people). Palmer apparently did it a bunch of times in the simulator and found some subtleties.
I think everyone would be alive if the joysticks were force-feedback and you knew what the other guy was doing. Many minutes were lost with one pilot trying for full-forward and the other pilot countering his inputs. Maddening. Sleep-deprived captain, possibly violating the eight hour rule, did not help matters.
Colin: I haven’t read Palmer’s book. I don’t think that results from a sim are conclusive. The Level D sims are accurate only within a pretty narrow range of flight parameters. We crashed the CRJ sim during Delta/Comair training a few times. It basically gives up and red-screens if the attitudes and flight control inputs get to be extreme. It is a great trainer for normal procedures (as you would hope, considering the $6 million(?) cost) but not a substitute for unusual attitude and aerobatic training in a real aircraft.
So wait, all this time my wife lied to me when she said turbulence is just like “bumps” on the road!?!
That’s right – the sims are only as good as their model and the model is only as good as the data that goes into constructing the model and nobody can gather any good data for these extreme conditions that are outside the normal flight envelope because you would have to seriously risk life and aircraft in order to gather that data. You can extrapolate or guesstimate how the aircraft would respond but there’s no guaranty that the actual aircraft would really respond in the way that the sim says it would, especially on “close questions” . Maybe the software model says you can pull out of the dive just above wave level but if the model is 1% wrong, maybe you can’t until you are 300′ below sea level (i.e. you can’t do it at all) or maybe you could level it out at 300′ up.