A recent article published by a prominent media company in the U.S. is forwarding the issue of expanded and enhanced simulator training for upset recovery; an issue that the NTSB has recently cited as needing attention.
This recent push being the result of a regional carrier crash in Buffalo, New York where the pilot apparently failed to properly recover from a stalled condition.
While it’s hard to argue with the idea of using better simulator equipment for that type of training, I have to wonder if this truly is the correct fix for the problem at hand. Should we devote our effort to creating precise simulator responses to a myriad of extreme conditions that might lead to a loss of control, or should we spend far less time and money teaching pilots how to 1) avoid those extreme conditions, or 2) recognize what conditions may lead to a loss of control so pilots can halt the progress of a situation before it gets to the point of no-return.
Many of the extreme conditions that lead to loss of control are created by atmospheric disturbances, such as wind shear and thunderstorms. But there are often cases where it might be difficult to avoid extreme weather conditions. The forecasts are not always good enough, and timely information isn’t always available to the pilot.
Luckily, technology being what it is today, these instances are becoming rarer. But even so, the technology has to be used correctly. Even the oldest of onboard weather technology, such as radar, has been misused or misinterpreted, leading to loss of control accidents in thunderstorms.
There is no reason to believe that new technologies in the cockpit will be better used unless pilots are trained to use them differently. In fact, it is easy to imagine how overwhelmed the pilot might be if all of the pertinent weather data were available at any given time. But since we don’t want to purposefully seek out bad weather to train in, simulators are the best place to train for these sorts of upsets.
But what if weather isn’t the cause of an upset? Then loss of control must have something to do with the pilot’s control of the airplane, or a mechanical problem of some sort. In the case of extreme attitude conditions, onboard technology might be able to provide immeasurable assistance in recognizing, and even perhaps assisting with recovery during a potential loss of control event.
As in the Buffalo crash, consider what might have occurred if the airplane had taken the control away from the pilot, who was incorrectly pulling back on the control column whilst the airplane was stalling. That might have saved the lives of the people on that flight.
Is there a downside? Well, automation is a funny thing sometimes. Sometimes it does its thing when the pilot doesn’t want it to do anything! The trick becomes to figure out exactly how that automation is supposed to work, and when it is supposed to engage and “save the flight.”
We’ve gradually accepted more and more automation in the cockpit, but there are limits to what level of control the pilot is willing to give up. So increasing automation may not be the answer to upset recovery.
So what are the alternatives? If not technology, then perhaps we need to look at training. I believe this is that tact that the NTSB is taking in addressing the loss of control problem trend. And what better place to do upset training than on the ground in a simulator? But let’s consider other training that needs to be done prior to simulator training.
How about training pilots to recognize the circumstances that lead to an upset, or loss of control? I’ll bet if you ask a pilot what will happen if the nose gets too high, and the airspeed too low, they’ll respond promptly with the word “stall.”
Everyone knows the inherent dangers of a stall, especially close to the ground, so why are pilots still inadvertently stalling their airplanes? My educated guess is that pilots get distracted from flying and by the time they realize something is amiss, they’re too far out of the control loop to jump in and effectively recover.
If the pilot is not watching what the airplane is doing, he is usually not aware of how things changed to get to that nose high attitude, slow speed, high angle of bank, etc. Sometimes, and I’ll bet most of the time, upsets start with subtle changes to the airplane’s attitude or airspeed that go unnoticed by the pilot(s). That’s a problem that won’t be solved by upset recovery training. Rather, it needs to be solved by training vigilance in the cockpit.
Keeping all of the above in mind, I’m not saying that upset training is a bad idea. Quite to the contrary, I think it’s an excellent idea since it provides a last line of defense if cockpit vigilance should inadvertently break down.
But in a sense, to say that this type of training is the key solution to eliminating loss of control accidents is kind of like saying that car drivers need to have simulator practice on how to regain control after losing control on ice covered roads when driving at too high a speed; clearly because we can’t seem to train drivers to have the good sense to avoid these situations.
One clear problem with large airplane training is that the pilot rarely experiences the full capability of the airplane she is flying. This is partly due because 1) large airplanes are not flown like small airplanes; different mission profiles and passenger comfort considerations, and 2) large airplanes respond differently from small airplanes, especially in the lag between recovery control inputs and actual recovery. As a result, certain maneuvers are not practiced the same way as in a small airplane.
And even when they are practiced for a large airplane, they are practiced in the simulator and seldom if ever done in the actual airplane. This of course requires that the simulator be of high enough fidelity to accurately simulate both the aerodynamic cues, as well as the motion and visual cues of the upset.
Historically, that’s something many small airplane simulators have not done well, especially since most do not have adequate visual or motion cues. Some companies now have simulation devices that are pretty good at simulating these sorts of upset problems with all the appropriate cues, but they are not required to meet current training requirements and they’re quite expensive.
Back to the Buffalo accident. Consider the pilot flying that ill-fated flight. If the pilot were predisposed to the possibility of a tail-plane stall due the ice, and there is ample evidence to believe he might have thought this, then a loss of control might have easily coaxed him into using a tail-plane stall recovery procedure. He pulled back instead of pushing forward, even with the control column stick-shaker on, which should have told him that he needed to push forward. Additional evidence, which the NSTB denies as being relevant to the existence of a tail-plane icing mindset is that the first officer retracted flaps; yet another tail-plane stall recover tactic. Whether or not there was evidence that a tail-plane stall was actually occurring, and there doesn’t appear to be any, these pilots did not respond correctly to the situation.
But would better simulator upset training have fixed that? Perhaps, but I would like to think that the best first step in solving problems like these is to examine why the pilots could not, or did not respond to ample available cues that the airplane was headed for a stalled condition. That’s a situational awareness issue, not a flight skill issue.
I can not say with 100% certainty that I’m sure how to solve all loss of control type problems. But what I can say is that we, as pilots, must consider the events and actions that lead to these types of accidents and out how to eliminate, or at least mitigate errors prior to losing control. Upset training has its place, as does spin recovery training for small airplane pilots, but as your instructor probably once told you, you have to stall the airplane in order to spin it, so don’t stall the airplane.
He or she was teaching you to pay attention to what your airplane is doing so that you never have to get in the position of recovering from an inadvertent spin. So maybe the best approach to take, on the way to doing that simulated upset training, is to teach solid situational awareness of the airplane’s state and the capabilities of the airplane’s response to pilot input. And then if the situation awareness does accidentally break down, the simulator training on upset recovery will kick in and hopefully save the day.
This month’s Pilot Primer is written by Donald Anders Talleur, an Assistant Chief Flight Instructor at the University of Illinois, Institute of Aviation. He holds a joint appointment with the Professional Pilot Division and Human Factors Division. He has been flying since 1984 and in addition to flight instructing since 1990, has worked on numerous research contracts for the FAA, Air Force, Navy, NASA, and Army. He has authored or co-authored over 180 aviation related papers and articles and has an M.S. degree in Engineering Psychology, specializing in Aviation Human Factors.