Pilots who fly IFR – especially single-pilot IFR – love to debate the merits of George. Not as in Bush, but as in that electronic servant who will fly your airplane any time you want.
Some advocate handing over the manipulation duties as soon as the gear is up on all flights, and letting George have it all the way to short final. Others consider the autopilot a necessary evil, to be used only when absolutely necessary – and turned off as soon as possible. To some its a savior, to others a crutch for the weak. Whats the truth?
To my eye, a crutch that can fail for those who lack skill, a snake waiting to bite those who misuse it, and a tool that can help the pilot who uses it effectively.
There is certainly no question that it can be a crutch. Consider the pilot of a Baron who had recently attempted an instrument proficiency check after which the flight instructor refused to sign him off. The accident report states:
The flight instructor reported that the pilots instrument skills were very weak, he was behind the aircraft and that he certainly could not have passed an instrument flight check – before any malfunctions were introduced. They flew three instrument approaches during the check; however, the pilots performance on instruments was so poor that the flight instructor refused to sign him off for instrument competency. At one point during the flight check, while the pilot was flying by reference to the instruments under a hood, the flight instructor had to take control of the aircraft to prevent loss of control.
Nevertheless, one month later the pilot attempted an IFR flight from Long Beach, Calif., to Santa Barbara. He was cleared for an ILS Runway 7 approach after being vectored to intercept the localizer from the south side, and the last assigned altitude was 2,000 feet msl. Radar data showed the aircraft continued across the localizer toward rising terrain.
The controller tried to vector him back to the localizer but the pilot did not follow the instructions. The controller then issued urgent instructions to turn and climb. The controller said the pilot calmly acknowledged the transmissions, but continued toward rising terrain. The airplane finally turned south, but hit avocado trees on the west side of sloping terrain at about 440 msl while in a steep bank.
The pilots former wife (herself an instrument-rated pilot) recalled instances when he had set the autopilot OBS incorrectly. On those occasions, he had similarly deviated from the localizer at intercept, resisted her suggestion to take over manually, and tried to force the autopilot to fly the approach.
This report certainly makes it appear that the pilot relied heavily on his autopilot to carry the load of maintaining basic aircraft control (attitude/heading/altitude). One can only wonder whether the autopilot was the reason for this reliance, or the result of it.
It is possible that his lack of proficiency was due to his lack of practice in basic instrument flying because he always used the autopilot. It is also possible that he used the autopilot because he lacked those skills in the first place. Either way, the pilot failed to recognize that the combination of lack of skill and excessive reliance on the autopilot could combine to kill him.
There are those who say an autopilot, especially the extremely complex integrated flight control systems in new generation airliners, transform the pilot into a systems manager. There is considerable truth in that statement, and in it, as well, a warning.
The complexity of these systems has reached the point where the traditional pilot training curricula as incorporated in Part 61 of the FARs may be inadequate. Further, the skills and aptitudes that are historically emphasized in pilot training and airline hiring may no longer be sufficient. There have been a disturbing number of airliner incidents (many involving the Airbus series) in which crews were, in effect, reduced to passengers wondering what the airplane would do next.
If one is to be such a systems monitor, one must know how the system is supposed to respond, and exactly what response will result from any particular input. The crew of a Falcon 900 was quite surprised by their airplane in the fall of 1999.
The airplane was descending through 11,400 feet on autopilot when the captain determined it would not level off at the desired 11,000 feet. The first officer, who was in the left seat and flying the plane, grabbed the yoke and pulled to reduce the descent rate. As they reached 11,000, he relaxed the back pressure. The airplane pitched over violently and went through several pitch oscillations from +3.3 to -1.2 gs before the crew regained control. An unrestrained flight attendant was seriously injured.
It turned out that the aircrafts autopilot is not designed to be overridden. It responded to the control pressure by making trim changes, which caused the oscillations. Further, the flight manuals for both the airplane and the autopilot lacked any information on this issue. Nine days after the incident, Falcon sent out a news flash to all operators telling them not to attempt manual control via the yoke when the autopilot is engaged.
While this incident involved a failure of the manufacturer to properly warn the crew, many of the Airbus episodes involved both reactions that were exactly in accordance with the flight manual and crew inputs prohibited by the manual. The problem is that the systems are so complex and their responses sometimes so counterintuitive that the pilots must become computer systems experts as well as aviators.
It is unfortunate that there is not more training on new equipment when its installed in your airplane. Sometimes it seems that its harder to transition to a new coupled autopilot than it is to transition to a new airplane. Sometimes even minor switchology errors can put you in a real bind.
A few years ago, a King Air 200 pilot working solo on an ILS approach to Salt Lake City made one seemingly minor switch error in setting up for the approach. He left the flight director DME switch on DME-2 instead of changing to DME-1. As a result, the system was reading DME off the VORTAC instead of the ILS, and he was running his step-downs off a point 4.7 miles past the ILS DME site.
It appears he figured out something was wrong after he passed over the outer marker 900 feet high, because he dived to capture the glideslope from above, only 1.8 miles from the threshold, 500 feet above DH. He maintained the glide slope for 28 seconds, but the aircraft decelerated the whole time until it stalled 1.3 miles short of the runway.
One might suspect that the pilot was scrambling to get everything organized after the surprise of the outer markers beep-beep-beep about five miles before he expected it, and chopped the throttles to get down. The autopilot managed to capture the glideslope but stalled the airplane trying to stay on glideslope with inadequate power.
Another autopilot gotcha is unfamiliarity with switch location. Even if you know what the switches do, you are still in trouble if you cant find the switch in a hurry.
Another King Air pilot came to grief during a coupled approach to Manchester, N.H., when the localizer bar ran hard right at 100 feet above DH – still in IMC. The airplane rolled hard right and then hard left as the localizer indicator jumped the other way. The plane hit a light pole, but damage was limited to the nose gear assembly and the pilot landed at an airport with better weather. The accident report says the pilot felt for the autopilot disconnect button on the yoke while the maneuvers went on.
This is definitely not the time to be feeling for something that critical – youd best know where it is. If you get a new autopilot system, your old familiarity with the switches has just been overcome by technology. Its time to sit back in the captains chair and learn the switches all over again. Do the blindfold cockpit check.
Before you rip the autopilot out of your panel, remember that there are many things it can do well. A good example of when a properly used autopilot might have made a real difference occurred in a Baron that was cleared for the ILS Runway 27 approach with the weather right down at minimums – indefinite 200 foot ceiling and half a mile visibility in light rain and mist.
All indications were that the approach went routinely, with the highly qualified ATP pilot hand-flying the plane. This is consistent with the later testimony of a business associate of the pilot who reported that she had flown with the pilot on numerous occasions, and he did not fly approaches with the autopilot engaged. Radar data shows the airplane drifted below the glide path, running about 100 feet low at two miles to fly, and the last radar hit showed it 130 feet below glide path. The surviving passenger told investigators that he saw the pilot looking out the left window and then out the front just before the airplane impacted two 40-foot pine trees.
Here we have a single-pilot, night IFR/IMC operation down to minimums. This is clearly an extremely difficult operation that requires the pilot to split his attention between flying the instruments and looking forward to pick up the runway.
Given the weather, the pilot had to know hed be flying right down to DH. This means he had to start looking outside right at the point when precise aircraft control is most critical.
There are two good ways to handle this. One is to have a person in the other front seat who can do the looking while the pilot does the flying. This person doesnt have to be a qualified pilot, but if he isnt, youre adding to the risk by asking him to be your lookout. The pilot flies the approach to minimums and executes the missed approach unless the lookout calls field in sight.
The other method is to use the autopilot to fly the airplane while the pilot divides his attention between monitoring the autopilots operation and looking for the landing environment. The riskiest part of this technique is the transition from approach to missed, because few light plane autopilots have a go-around mode. Nevertheless, shifting from looking out to the power, climb, flaps, gear mode is a lot easier than trying to fly an approach with one eye and look out with the other.
The second great use for an autopilot is as a fatigue prevention device. Ive flown the Baltimore-Dayton run many times in my old autopilot-less Grumman Cheetah – once with a 35-knot headwind that made the leg 4.7 hours. At the end of that flight, I was tired. It is fortunate that flight ended in good VMC, because an instrument approach to minimums would have been a challenge at that point.
On the other hand, just yesterday I logged a 4.6 hour round trip in our Grumman Cougar with its Century III two-axis autopilot. I did the takeoff and climb, handing control over at level-off and taking control back at about 2000 agl on the descent. The difference in workload and fatigue is monumental.
With the autopilot doing the work, youre much more physically and mentally fresh for the approach and landing. You have more attention for tasks like visual lookout. When its chart-finding or map-folding time, youre not juggling with your hands and flying with your knees – a less than satisfactory method of aircraft control.
Changes in routing are also much easier dealt with when you can focus on finding the new route on the chart and lending only a bit to the flight instruments to see that nothing goes amiss while your attention is diverted.
Personally, I usually hand-fly the approaches. My main purpose here is to maintain my own proficiency in case George goes on strike sometime when the weather is lousy. But I must admit that theres a bit of ego involved, too – I get a great deal of satisfaction out of flying an approach to minimums and keeping the needles inside the bulls-eye all the way down despite wind shifts and turbulence.
If I have another pilot in the right seat, I will hand-fly to minimums even in the lowest weather, using basic cockpit resource management techniques of I fly, you look. Again, it allows me to keep up my proficiency.
A couple of times a month, in good VMC, Ill practice a coupled approach to minimums, mainly to keep up my skill at the switchology necessary to make it happen. And then when the chips are down (weather at mins, no help but George), Im ready to use the autopilot to assist me in flying the approach with a modified CRM approach. In that case, George flies, I look.
There are those who suggest that they will only use the autopilot in an emergency situation. One wonders how they are going to remember how to use it if they dont practice. The accident record is very telling of the strange and unusual things of which autopilots are capable.
Those who arent practiced and comfortable in the techniques of getting the autopilot to do what they want and then share the workload with it in a rational manner are likely to be unhappily surprised when they turn to George in a crunch and find him a stranger with strange ways.
Finally, consider the well-known accident involving JFK, Jr. A non-instrument rated pilot in a complex, high-performance single, operating VFR in marginal VMC over water at night has a lot of tasks to juggle.
There are a number of speculations about how the final seconds went – distractions, a need to switch tanks using the tank selector down near his ankle, a last-minute hunt for a radio frequency or runway data. In any case, accident investigators concluded he was not using the autopilot.
One can only wonder whether flipping it on at that critical time would have made the difference.
Also With This Article
Click here to view “Jane, Stop This Crazy Thing.”
-by Ron Levy
Ron Levy is director of the Aviation Sciences Program at the University of Maryland Eastern Shore.