The tug between human and electronic flying
When we let the electronic systems fly the airplane, we are still flying, if by proxy. That means that a big part of the pilot’s job is to fully understand the computers we use to tell the autopilot what to do. That is not always easy to do and in the worst case can lead to deadly confusion. That puts the operation of the flight control system squarely in the “airmanship” category.
The FAA and the training industry both say that pilots must use automation to cut workload and manage flights safely and efficiently but that they should also maintain manual flying skills for the time when only superior basic airmanship will save the day.
When I was a kid going to Sunday School, I remember the mantra of the religious belief on predestination: You can and you can’t, you will and you won’t, you’ll be damned if you do and damned if you don’t. That seems to also apply to pilots when the question of automatic v. hand flying comes up.
The discussion of safety topics is almost always started because of an airline incident or accident, in this case the Colgan turboprop lost near Buffalo a while back. The crew stalled the airplane and, in effect, spun in. They lost control.
There are other fairly recent airline accidents related to this. One was the Airbus A330 that was stalled at 38,000 feet after an apparent ice-induced failure in the pitot static system and resulting erroneous airspeed information. As the Airbus plunged toward the Atlantic Ocean, the crew never recognized that it was in a stall and kept applying aft stick. It hit the water four minutes after stalling at 38,000 feet.
Another one was a Turkish 737 in Europe. The autopilot was flying the approach and a fault in the system told the autothrottles it was time to land well before it was actually time to land. Power was chopped and the crew didn’t respond to decaying airspeed before it was too late.
A lot of attention is paid to this but it is a relatively small problem in airline operations and is probably related more to aircrews that are not really suited to the role.
Those cases were related to a loss of control, or, a failure in basic airmanship. How else would you describe the situation when a pilot becomes a passenger for the final moments of a flight.
How does general aviation come out in relation to this failure in airmanship? It is much worse than with the airlines.
In 2010, about one third of the fatal accidents in certified fixed-wing airplanes operated under Part 91 of the FARs were loss-of-control accidents. If you want the raw numbers, there were 167 such accidents, total, with 62 of those involving a loss of control. That is a lot.
There are two types of accidents here. The low speed loss of control, often referred to as a stall/spin event, is the dominant one. However, about a third of the accidents involve a high speed loss of control.
I have been following this subject for a long time. Back in the good old days, the stall/spin accidents most often involved airplanes without an excess of power. That meant that the pilot was flying relatively close to a stall all the time. Also, that was back in the day when spins were an order of the day and if an airplane is built to spin well, it will spin to the ground accidentally equally as well. Airplanes with the venerable 65 hp engines, like Cubs, Champs, Taylorcrafts and Luscombes were in the fleet in large numbers and were frequently found in the stall/spin column.
That has largely changed today where just over ten percent of the total loss-of-control accidents are in simple singles. There were about as many twins in these accidents as simple singles.
The majority of this type accident comes in high performance singles. Also, the great majority of high speed losses of control come in retractable singles. The reverse is true in twins where the majority is found in low speed events, usually after a power problem or during training.
Power problems do loom large with almost a third (18 out of 62) power-related. Half of these were twins and a few of those came during simulated engine-out flying.
In both singles and twins, an engine problem on takeoff stands out as a prime contributor to fatal accidents. Three twins and five singles were lost because an engine packed it in during this critical phase of flight.
If pilots losing control after a power failure is a problem today, it will probably be a bigger problem in the future. As airplanes and engines age, they need more expensive maintenance but they usually get less. Reliability is the victim and more and more older airplanes with mechanical problems are showing up in accident reports.
An airframe failure after a high speed loss of control used to be a fairly common occurrence. I found this but two times in 2010. An Archer pilot continued VFR into IFR conditions, lost control and the airframe failed after the airplane accelerated out of the envelope. A Cessna 210 airframe failed after a tussle with some particularly nasty storms. Now more airplanes are reaching the ground in one piece after a high speed loss of control.
Intentional low-level flying, buzzing and low altitude aerobatics are not as pronounced as they used to be and there was only one combination of aerial pranking and a drunk pilot. That is one too many but there used to be more.
There are some pretty simple guidelines to follow to keep those accidents at arm’s length.
A common thread among low speed losses is that the airplane was being maneuvered at low altitude. If a pilot avoids bank angles of over 30 degrees except when practicing steep turns at altitude, the risk fades. A strong awareness of airspeed and angle of attack helps, too.
A common thread among high speed losses is confusion. This can all but paralyze a pilot’s ability to think. The best plan here is to call a time out. You can’t pause the airplane like a simulator but you can cut back to the most basic piloting task while regrouping. At a safe altitude, concentrate on keeping the wings level: Nothing else. If the power was correct and the trim set properly, altitude and airspeed will stay in the ballpark. After a minute or so, some tasks can be added back, the simplest ones first.
Now I want to fast forward to a 2013 accident that illustrates how complex a loss of control can become. This is based on a preliminary NTSB report. I seldom use those and I’ll explain why in the conclusion.
The airplane was a 2001 Piper PA-46-500TP Meridian single-engine turboprop. While well-equipped with a glass cockpit, it did not have the more sophisticated flight control system that came to later-model Meridians.
The pilot’s log showed that he had 2,365.7 hours with 118.3 in actual instrument conditions and 86.3 in simulated instrument conditions. He had 126.9 hours in Meridians including 10.1 hours of instruction. He logged 57.3 hours in the preceding 90 days and 4.3 in the 24 hours prior to the accident.
So, on paper you had a good airplane and a pilot who had checked all the boxes.
The weather was IFR for the departure from Paris, Texas, elevation 547 feet.
Shortly after takeoff, the pilot contacted Fort Worth Center and was told to maintain 5,000 feet. Then the controller told the pilot to climb to and maintain 16,000 feet. The pilot did not acknowledge that instruction and was not heard from again.
At about this time, the airplane was on a southwesterly heading and had reached an altitude of 4,700 feet at a groundspeed of 249 knots. Less than a minute later the airplane was at 5,100 feet and had slowed to 214 knots. It then entered a descending right turn and slowed to 202 knots. Twelve seconds later the airplane continued to turn right, climbed to 5,000 feet, and slowed to 153 knots. Nineteen seconds later the airplane climbed to 5,200 feet and slowed to 115 knots. The last radar return was about eleven seconds later when the airplane was at 4,500 feet at a groundspeed of 110 knots.
One witness reported the sound of an engine backfiring. Another said he heard the sound of an engine that revved up and down about three times before the engine noise just stopped.
The wreckage was consistent with the airplane being in a flat spin prior to impact. No evidence of pre-impact failure had been found when the report was issued.
The event was clearly a loss of control at low speed followed by a flat spin.
Something pretty drastic happened at about the time the controller cleared the flight to 16,000 feet. The behavior of the airplane, as seen on traffic control radar, didn’t reflect anything like even the start of a normal climb out of 5,000 feet toward 16,000 feet.
In a year or two the NTSB will come up with a probable cause. At that time they will reveal any evidence that something in or on the airplane or the pilot himself had a major problem. Or they could just say the pilot lost control for undetermined reasons as they did with a good friend who went from a low flight level to the ground quickly some years ago.
Most loss-of-control accidents do have pretty clear causes but some do not. If no mechanical or physical malfunction is found that means no cause is issued and the matter is left open for speculation, forever.
A leading suspect in a case like this would be some sort of power problem. If so, why did the pilot more or less maintain altitude while the airplane decelerated rapidly into a stall? Could the pilot have done something to cause a power problem?
A flight control system or instrument malfunction that may or may not be revealed in the investigation would have to be a suspect as well. Improper use of the flight control system leading to obviously hopeless confusion is something else to consider. So is pilot incapacitation but that would have led to this chain of events only if the pilot had not been using the autopilot and was hand flying as many pilots still do. That could also explain why the pilot said nothing of a problem even though he had time to do so.
Take your pick or add speculation of your own. And I think you can see from this why I think preliminary reports don’t have much value. They give only bare facts and too much conjecture is required to go any farther.
What do you think?
- From the archives: how valuable are check rides? - July 30, 2019
- From the archives: the 1968 Reading Show - July 2, 2019
- From the archives: Richard Collins goes behind the scenes at Center - June 4, 2019
Has FAA, or anyone else, found many accident cases where an autopilot was fighting the pilot? I’m thinking of situations where either it engaged spontaneously, or would not disengage when a pilot was trying to shut it off?
I think an aspect of this of automation is proper training in its use for standard procedures. In my experience, often the training that occurs is in the airplane with an instructor pushing buttons that the student has forgotten.
Books on using the automation often focus on the setup of radios. However as pilots, we must also setup PFD, MFD, and autopilot. We then need to confirm the integration between the radios, PFI, and autopilot is working as we expected. Often this must be done in a timely basis such as when ATC says “Fly heading 350, plan the ILS approach for airport.” In this example there are actually two transitions, one from what we are doing to flying the heading mode and then setup to anticipate the heading mode to the ILS. Often the transition to the ILS can only be partially setup. One needs to be able to do this is a consistent timely manner to keep up with ATC.
I have seen manuals from an airframe manufacturer that identifies certain conditions of flight and a change to these conditions. The manual then specifies a process to configure avionics to transition from current to new situation.
My sense is that too often we are being taught “The automation may fail; you must be able to fly without the automation.” That is correct. However we also need to be taught “There are many situations where the automation should be used, you must be able to efficiently and effectively setup for the situations and transitions between them.” In my experience that is 99.9% of the cases
One might think such procedure training is unnecessary. Consider all the guides we have for flying such as 6T’s and GUMPS for basic things such as turning and landing. The PFD, autopilot, MFD, and radios are far more complex. I think too often we are making up as we go along and then asking what have we missed.
I’m not IFR qualified, but am familiar with technology overload, which is what I think I hear you saying? The stuff that is supposed to help us, can get in the way of success sometimes? Designed and tested by a squint in a lab, not by a user in the real act of using it?
There’s absolutely no substitute for field testing under actual conditions. McNamara’s meddling with the M-16 comes to mind.
This will be a head scratcher in the absence of more definitive information but your discussion of loss of control accidents is relevant here — we often don’t know what specific events initiate these types of failures, but we do know they follow certain patterns where flights are apparently proceeding in a normal manner — until they don’t!
This accident is similar to the Baron at Topeka, KS (N580EA) in April 2011 and the Pilatus in Florida last year (N950KA). In both cases, the pilot was transitioning from one phase of flight to another in IMC and the aircraft got away from them and everything ends up on the ground a short time later! In these three accidents, the pilots were all relatively new to sophisticated aircraft and had very little time-in-type and even less in IMC in these aircraft. It is one thing to fly a new aircraft under VMC during initial transition training, but an entirely different matter to competently handle things in IMC where all sorts of new systems, equipment and instrument indications have to be mastered. The Texas pilot had some recent experience in the Piper, but there’s no indication whether it was IMC or VMC or the extent of his transition training. If his recent experience was concentrated in VMC, it’s easy to see how he could have become distracted (spatial disorientation) with everything that was going on as he transitioned from initial climb to cruise altitudes. Unless one has a clear mental picture of how to execute each and every maneuver (dare I say a checklist?) and the proper sequence in which to proceed, it would be relatively easy to get things out of order and while trying to regain focus, the airplane goes off in a different direction. If that should happen, the additional mental workload of trying to interpret what is happening with power settings, heading, changes in altitude and communications while trying to correct what you think you’re seeing could easily overwhelm pilots with low time-in-type and inadequate skills for the situation at hand.
Whether any of the preceding is relevant is pure speculation, but I do know from experience that transition training is often less thorough and complete than it needs to be with attention focused on the basics. Seldom does it extend into all realms of flight due to time and cost considerations. In the case of the Baron accident in Topeka, I’d be willing to bet serious money that the transition training program did not include mastery training for executing a back course approach to minimums, transitioning to a missed approach and resetting the navigation equipment for approach to another runway while communicating with ATC and complying with clearances, etc.
There is an old axiom that if areas of training are not satisfactorily mastered during the training process but the student is signed off anyway, the flight environment will, sooner or later, administer the test again in real time. If the test is failed in real time, the Forces of Nature often simply extinguishes the violators by destroying the aircraft and the occupants. Such penalties are imposed without remorse, apologies or appeal, as can be seen in these three accidents (and countless others) — gravity always wins!
Your comments are extremely insightful. Sadly, to often the physical forces win. I am VFR only, but my CFI brother beat into me how to hold a heading and altitude should I venture into the world of IFR. Making a constant rate 180 then could follow one I had achieved full control. I wonder how many pilots don’t figure out when they’ve crossed over the line soon enough to take appropriate action to get ahead of the aircraft again. I would think that if one is not sure of their situation that they immediately go into a urgent mode and go where they know they’re safe. Maybe easier said than done. Surely the old mantra “aviate, communicate, navigate” could be more judiciously applied in many cases. ATC might complicate things, but they can also help solve problems if they are called upon.
Your comments are extremely insightful. Sadly, too often the physical forces win. I am VFR only, but my CFI brother beat into me how to hold a heading and altitude should I venture into the world of IFR. Making a constant rate 180 then could follow once I had achieved full control. I wonder how many pilots don’t figure out when they’ve crossed over the line soon enough to take appropriate action to get ahead of the aircraft again. I would think that if one is not sure of their situation that they immediately go into a urgent mode and go where they know they’re safe. Maybe easier said than done. Surely the old mantra “aviate, communicate, navigate” could be more judiciously applied in many cases. ATC might complicate things, but they can also help solve problems if they are called upon.