Printer Friendly

Autopilots and IFR: autopilots reduce our workload, but we might not be sharp enough when they fail. Should they be required for IFR? Should they be used as much?

Autopilots are a fabulous workload management tool. In a busy, single-pilot cockpit they quite literally can be a lifesaver. We've come to accept their precision and dependability. There's a growing viewpoint that a functioning autopilot is an essential requirement for all instrument flight--many pilots I know say they would not contemplate IFR without an autopilot.

So consider this scenario: You are cruising above a solid cloud deck, with IMC below and the need to fly an instrument approach. The autopilot fails. Would you declare an emergency? Would you divert to a VMC airport? In other words, if you feel a functioning autopilot is required for IFR, where does that put you when the autopilot dies?


It's a serious question: Should the Federal Air Regulations require an autopilot? Most Part 135 operations must have a working autopilot for IFR flight, or else carry a two-pilot crew. To answer this question for Part 91 ops we need to look at the pros and cons of autopilots ... including how reliable they really are, and what effect autopilot use might have on a pilot.


Autopilots come in three basic forms:

Single-axis autopilot: A single- (or one-) axis autopilot controls airplane heading, using ailerons only. These are commonly called "wing levelers," and as the name implies, they will correct for bank excursions to hold the airplane on a consistent heading. They will not hold pitch or altitude, and may or may not be able to track a navigation signal. Nonetheless they can be a big workload reducer in the single-pilot cockpit. I have about 100 hours in a one-axis equipped Bellanca Super Viking and remember it was a breeze to manually trim up for level flight and point on course with the autopilot. Single-axis autopilots can be lifesavers in disorientation scenarios like inadvertent flight into IMC. For a while both Mooneys and Beech Bonanzas had a full-time wing leveler (Beech called it the "Constant Copilot"; Mooney "PC" for "Positive Control")--and the pilot had to manually override it in order to bank the airplane. Pilots reportedly hated it, crying, "Just let me fly the airplane," and the idea died away. I've not researched the data but some now say airplanes so equipped had a much lower loss-of-control accident rate compared to similar airplanes.

Two-axis autopilot. A two-axis autopilot adds pitch control that can hold attitude, altitude, heading and navigation (including, in some, approach modes). Some pilots mistakenly call these "three-axis" autopilots because with pitch and bank control the autopilot can control flight in everything but yaw ... and airplane stability (not to mention pilot complacency) makes yaw control an afterthought to many. Most general aviation autopilots today are two-axis types.

Three-axis autopilot: Add a yaw damper controlling the rudder to a two-axis autopilot and you've automated flight in all three axes. This is very helpful in some airplane types that suffer from Dutch rolling tendency (a sure way to make back-seat passengers queasy) or are loaded toward the aft end of their c.g. envelope. Often a full three-axis system is merely a two-axis autopilot with an optional yaw damper that can be run independently if the pilot wants to inhibit Dutch roll while hand-flying the airplane. Because they resist rudder control it's imperative that pilots remember to turn off the yaw damper before landing.

Some pilots don't like one-axis autopilots because "you have to watch them all the time." The reality is that all autopilots require continuous monitoring--they're there to help, not act as pilot-in-command. As one of my students once said, an autopilot is an extremely capable, extremely stupid copilot. It does precisely whatever you tell it to do, right or wrong. As reliable as most autopilots themselves are, one of the most common reports to NASA's Aviation Safety Reporting System is an altitude bust because the pilot mis-programmed the autopilot.


Autopilots are driven in one of two ways: a rate-based system, or one that is attitude-based. A rate-based autopilot senses wing leveling through a turn coordinator--the rate of turn. If it senses a turn through the indicator, it corrects as needed. If a turn is commanded, the rate-based system aims for a predetermined rate of turn on the turn coordinator, usually standard-rate or half standard-rate.

Common rate-based autopilots include the Honeywell KAP140 and the S-Tec line. "Glass-cockpit" airplanes that include a rate-based autopilot may have a turn coordinator hidden behind the panel, out of the pilot's view, that drives the autopilot.

Advantages of the rate-based autopilot: It is generally less expensive than attitude-based types. If the airplane's attitude indicator fails and leaves the pilot flying partial panel, a turn coordinator-driven autopilot can be a lifesaver. If the pilot gets into an unusual attitude the turn coordinator won't as readily tumble, so the sensing system is still reliable if the pilot recovers and then engages the autopilot. Disadvantages: Rate-based based autopilots aren't as sensitive as attitude-based systems, so they may "hunt" a little more to remain on a navigation signal. Lose the turn coordinator and the autopilot also fails--if the turn coordinator is hidden in a glass-cockpit airplane, the pilot can't check this during taxi.


Attitude- or position-based autopilots reference the primary attitude indicator, and may also read the heading indicator of an HSI. In turns, attitude-based autopilots generally command a specific bank angle regardless of aircraft speed. Common attitude-based autopilots include the Honeywell KFC150 and 200, and most Century autopilots.

Advantages of attitude-based autopilots: They are more precise than rate-based systems, and make smaller corrections. Disadvantages: If the attitude indicator (or the HSI, if linked) fails, the autopilot will not work properly and in some installations will disengage (and not reengage).

A third type of autopilot just entering the general aviation market is driven by accelerometers and AHRS (attitude/heading reference system) inputs into a glass cockpit.

Garmin's integrated GFC700 is a leading lightplane design in this class. These autopilots are extremely precise--but get "red Xs" on the glass panel and you lose the autopilot also.


Autopilots command control surfaces to move though servos, small electric motors that drive trim tabs or the control surfaces themselves. Autopilots can hold a little control force without trimming--that's why the airplane is often a little out of trim, and tends to nose up or down, when you turn off the autopilot--but will disconnect if the force required reaches a design value, or if bank or pitch becomes extreme.

Strong jolts of turbulence can drive the autopilot to exceed these limits trying to resist, and cause the autopilot to disengage. A little structural icing can freeze up a trim tab, causing the autopilot to try to break it loose--making the airplane lurch when the ice does break, or the autopilot to disconnect (while out of trim) when it reaches its maximum control force input.

If a servo motor seizes the autopilot won't function. More commonly, a short develops in a servo motor and it continues to run after the autopilot makes an adjustment (see the sidebar, "Trim Runaway," on the previous page).

Why concern yourself with the intricacies of what drives your autopilot, or how it controls the airplane? Because autopilots themselves may be very robust, but depending on the type you're flying with they may be rendered inoperative by seemingly independent systems failures like vacuum systems or HSIs, or environmental conditions like turbulence or ice. You can go from everything-working to out-of-trim, partial panel hand-flying in the blink of an eye ... maybe just as you're reaching minimum altitude on a coupled instrument approach. It's important to know what will make the autopilot inoperative besides an autopilot failure alone.


In the early '90s I was a simulator instructor at a brand-name training provider, teaching pilots in Beech piston airplanes. One week I had a pilot in class who had phenomenal experience as an Alaskan bush pilot but who had recently moved south and bought an A36 Bonanza. The insurance company mandated he attend the course.

He knew the airplane well--in fact, he was somewhat upset that our "initial" course concentrated on the airplane systems, because unlike most owners he had made a thorough study of the Pilots Operating Handbook and engine manuals. In the simulator, however (actually a Flight Training Device), he had great difficulty in basic aircraft control and workload management.

In his defense, the FTD was difficult to fly at that stage in its development, but I could tell the difference between control difficulty and a pilot who had severely deficient IFR skills. Unfortunately this pilot fell into the latter category.

His response to my coaching and critique was that I was not letting him use the autopilot fulltime--that he flew hard IFR in busy airspace, and that he turned the autopilot on immediately after gear retraction and used it exclusively until just before touchdown. This, he explained, made him able to better manage the flight, spreading out the workload in IMC and letting him scan for traffic when clear of the clouds. He felt we should teach "pilot-in-the-loop" flying, i.e., fulltime autopilot use with the pilot as a systems manager.

He was so frustrated with our training emphasis on hand-flying (with the autopilot used primarily in distracting situations like briefing for the approach) that he left the program early and did not complete the course. I don't know what effect this had on his insurance, and I could certainly have communicated our stance to him better, but that experience set me thinking about the appropriate use of highly capable autopilots in high-performance, single-pilot airplanes.


In my experience the worst instrument pilots are those who engage the autopilot almost immediately after takeoff, and fly coupled for the entire flight. At the simulator school we called these "gear up, autopilot on" pilots, which I'll now amend to "positive rate, autopilot on" because so many fixed-gear airplanes have highly capable autopilots.

Make one of these pilots hand-fly a straightforward descent, approach setup, approach and missed approach into a hold, and they would almost always reach a point where the workload was too great to meet basic IFR tolerances. Usually we could get the pilot back up to speed in a few hours' training. Problem is, if their first recent experience in flying by hand happens in the airplane in IMC, not in a training environment, the opportunity for retraining may be tragically lost.

I'm not suggesting we forget the autopilot exists, or that all training be hand-flown with no regard for the labor-saving capabilities of autopilots. Instead, I advocate we approach autopilot use intelligently, and remain prepared to hand-fly if something should cause the autopilot to fail. For instance:

* Alternate the way you fly approaches to remain current with fully coupled, flight director (if so equipped) and hand-flown approaches. The less often you fly, the more you should weigh your approaches toward the raw-data hand-flown.

* Use the autopilot in situations that demand your attention, such as mixture leaning, setting up for an approach and VMC in busy areas.

* Don't assume the autopilot will always perform as expected. Scan the mode controller and the flight instruments every few seconds to make sure all's still going as planned.

* When flying an approach to minimums, use the autopilot and monitor it closely. If the autopilot should issue strange commands, excessive rates of descent or unusual intercept angles, turn it off and fly the approach by hand. If you're not immediately comfortable with the result, hand-fly a missed and set up to try again.

* Fly all approaches, navigation tracking and holds logged for purposes of instrument currency by hand--don't count them if they are flown by the autopilot.

* Practice, in a realistic simulator or in the airplane with a knowledgeable instructor or safety pilot, taking over from the autopilot and recovering the hand-flying the airplane in cruise flight, during an approach, and in a missed approach.

* Know all the ways to disengage the autopilot and electric trim.

* Include transition from everything- working to partial-panel flight in those "take-over" scenarios.

* In airplanes without manual trim systems, practice (in a simulator or with a very careful instructor) hand-flying approaches and missed approaches with the trim set too nose-high or too nose-low, as if recovering from a trim runaway.


Should autopilots be required for single-pilot IFR? I think not--history and experience shows an airplane can be flown in IMC without this level of automation, if the pilot is well-trained and does a good job managing workload.

The bigger question is: With high-quality autopilots on board modern single-pilot aircraft, how can pilots maintain the skills necessary to hand- fly the airplane in the event the autopilot disengages?

Until our airplanes have triple-redundant autopilots powered by independent electrical systems and referencing separate attitude or rate indicators, and power multiple trim servos that have the ability to override one another, we have to be ready to hand-fly at any time.

Autopilots are a significant safety addition to the single-pilot cockpit. Most are incredibly reliable. But they can't replace a pilot flying the airplane and there are dozens of failure scenarios.

Keep your skills sharp--focus on hand-flying, partial-panel and workload management--and pick weather based on your ability to safely complete the trip without an autopilot. In other words, never fly anywhere with an autopilot in weather if you cannot immediately take over and recover should it fail. Because it will.

Trim Runaway

A student of mine wrote shortly after his checkout in a five-year-old Bonanza:

"How do pilots keep their hand-flying skills sharp despite the automation? With the newer generation of aircraft being able to fly a fully coupled approach safely, efficiency and reducing workload in SPIFR [single-pilot IFR] is what aircraft makers are striving for.


"But, have we created a whole new generation of 'push-button' pilots, unable to negotiate through the clouds without their autopilot on? General aviation automation has certainly come a very long way ... safety being at the forefront of R&D. Look at the Cirrus. But are newer generation pilots truly pilots or 'flight managers'?

"Flying in the LA basin in IMC with the numerous routing, heading and altitude changes, I personally got to tell you that flying with the autopilot on while in IMC--especially in the very psychotic, busy Los Angeles airspace--is very, very reassuring."

But one day he was in IMC when the pitch trim began to run away, driving down the airplane's nose. Knowing the emergency procedure, he quickly hit the control yoke's trim override, manually trimmed the airplane and pulled the circuit breaker to remove power from the trim system. He continues:

"I got a real taste of what it was like not to have my 'comfort pillow' when I had to maneuver the entire [route] by hand. I asked LA Center for a higher altitude to get over the clouds but was told 'unable' due to traffic in the vicinity.

"I was in and out of IMC for the entire hour, [with] re-routing and vector changes following a SID. That was only half my battle. After holding altitude to plus or minus 100 feet while on IFR flight plan, I had to now negotiate a GPS approach into my home base--I was a bit tired--by hand!

"I recalled my training and flew by the numbers. I broke out well above the 820-foot MDA at my home base and my personal IFR minimums that I set for my home base.

"My goal is to consistently stay sharp by seeking training every six months. I feel that someone who does not fly professionally should stay sharp on those hand skills by seeking professional instructors who insist their students 'fly by hand'.

But what caused the trim runaway?

"At first we thought it was the [trim] servo motor. Further investigation found only a loose screw! The screw that holds the trim to the motor worked its way loose and was found lying next to the motor."

There are any number of scenarios that can result in inability to use an autopilot and they can happen at any time. I've personally had two trim runaways in separate airplanes. Depending on your autopilot's design, dealing with a runaway trim can mean losing its ability to fly the airplane, at least in the pitch axis.

Disconnect Scenarios

* Turbulence great enough to cause a bank, pitch or roll rate that exceeds the autopilot's disconnect criteria, such as from penetrating a towering cumulus cloud.

* A control force strong enough to trigger a disconnect--like a mountain wave, ice accumulation or wind shear.

* Failure of the instrument(s) or the system that powers it, if that instrument drives the rate- or attitude-based autopilot.

* A trim runaway that puts the airplane in an unusual attitude or makes the autopilot exceed control pressure criteria in attempting to correct.

* Pilot activation of the electric trim--the autopilot's logic is that, if the pilot is changing the trim, the autopilot must have commanded an incorrect trim setting.

* Electrical failure.

* Autopilot software or hardware failure.

* Shorts or electrical grounding problems in the trim or associated systems.

* Any AHRS abnormality with an AHRS-driven autopilot.


There are a number of situations requiring the autopilot be off:

* Engine failure in multiengine airplanes. An autopilot generally cannot accurately control the airplane during the transition from twin- to single-engine flight. Once you have feathered the "dead" propeller and trimmed off asymmetric pressures, it may be possible to re-engage the autopilot for a single-engine approach.

* Moderate or greater turbulence, according to most autopilot manufacturers. The autopilot will try to compensate for turbulence and may overstress the airplane. Turbulence may exceed the autopilot's control force limits and shut off the autopilot unexpectedly with the airplane radically out of trim.

* Ice encounters, depending on the autopilot supplement's advice. Like turbulence, ice on trim tabs may result in exceeding control limits and leave you hand-flying an out-of-trim airplane without warning.

* Approaches that state "autopilot coupled approached not authorized" on the approach chart.

* For takeoff and landing.

Tom Turner is a CFII-MEI who frequently writes and lectures on aviation safety.
COPYRIGHT 2007 Belvoir Media Group, LLC
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Turner, Thomas P.
Publication:Aviation Safety
Geographic Code:1USA
Date:May 1, 2007
Previous Article:Ramp rage: aviation's version of road rage can not only lead to doing something stupid, but also to accidents. How it can happen, and what you can do.
Next Article:Fit for flight? There are good reasons pilots must comply with at least some rudimentary medical standards.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters