Hey! Why's the Horizon Over There?

  • E-Mail this Article
  • View Printable Article
  • Text size:

    • A
    • A
    • A

Loss of control accidents following in-flight upsets can be prevented through training and regularly practicing unusual attitude recovery.

It's happened to every pilot at some time or another—suddenly the airplane isn't doing what the pilot intends. Causes vary—his or her attention is diverted while hand flying, or spatial disorientation does its insidious thing or maybe its an upset due to turbulence. The deviation from the desired attitude can range from "Aw, nuts, I let the wing drop a bit," to "HOLY SMOKE! Which way is up?"

Almost invariably the pilot simply fixes the problem. Most of the time it involves a minimum of excitement. Then there are those events that, once the airplane is back to where it should be, it takes a while for the pilot to get her or his pulse back down to double digits. Unfortunately, a few dozen times each year the pilot doesn't get the airplane collected and we read about an in-flight breakup or an airplane that has augured into the ground at high speed.

It's Gonna Roll

Most general aviation airplanes are not stable in roll. Left to their own devices, they will eventually—it may take a while—enter a bank that will get progressively steeper, leading to a spiral and, uninterrupted, an unpleasant ending. Pilots learn from day one to use coordinated controls to pick up a wing that has dropped—it's usually routine. Depending on the aggressiveness of their instructors, they also practice recovering from more aggravated upsets—steep banks with the nose up or down and airspeed high or low.

Instructors comment that when ready for an upset and having practiced recently, pilots do a good job of recovering from some dramatic scenarios—steeply banked, nose down and the airspeed in the yellow as well as nose high and airspeed approaching zero.

Nevertheless, we still average more loss of control accidents each year than we do mid-airs—yet pilots are willing to drop big money on devices to detect traffic but object to buying recurrent training. There are schools that specialize in upset training where a pilot can spend a few days going through exercises that do an excellent job of teaching them to avoid a loss of control accident. In the real world, most pilots won't take such involved training. However, but they can get a lot of protection against the risk of loss of control by reviewing unusual attitude recovery procedures and incorporating unusual attitude recoveries into the next session they have with an instructor.


The goal of unusual attitude recovery technique development has been to come up with something that is one-size-fits-all. A pilot under the tremendous stress of figuring out what direction an upset airplane is pointing, while being forced into the seat by high G loads, amid the horrendous noise level of an airplane going very fast, doesn't need to have to sort through a menu of recovery techniques before applying one.

It's also been learned that pilots under stress have a tendency to pull on the yoke/stick—precisely the wrong thing to do during the initial steps of regaining control following an upset. Most of the time the airplane will be going substantially faster than cruise, so the trim is already trying to pitch the nose up. Pulling isn't going to help the situation. The pull reflex has caused pilots to overload the airframe, causing either the tail to fail downward or the wings upward (it's about 50-50 as to which happens first in an in-flight breakup).

What has resulted – and has been considered standard for several years now—because it works—is a smooth sequence of Analyze, Power, Push, Rudder, Roll, Climb.

The first two control input events should occur almost simultaneously—and some advocate that "Push" should come before "Power." As far as I'm concerned, that's an academic discussion; pilots can do both at once.


When I lived in Michigan, it seemed that every few years a non-instrument-rated pilot would try to cut the corner at the south end of Lake Michigan, heading west, on a hazy summer evening, discover there was no visible horizon and spiral into the lake. Having gotten myself disoriented while flying into the sun in haze, I always had a lot of sympathy for those pilots. I was convinced their bodies were screaming one thing at their brains while their eyes were seeing something else on the instruments. I could hear the shriek of the airframe as the speed approached redline and remember how it felt as my brain seemed to be pressing down into my neck under the G loads as the airplane accelerated in the nose-down, steep spiral.

I read accident reports where pilots lost it in IMC or haze and called ATC to say that they had an instrument failure of some sort—yet the post-accident exam of the instruments showed all were functioning normally. The pilots simply couldn't accept what they were seeing on the panel.

To me, that's a powerful argument for recurrent training that includes having the instructor induce vertigo so the pilot gets to practice how to make the correct control inputs even though her or his body is adamant that the instruments are wrong. Regular practice helps cut down the time it takes to figure out what the airplane is doing so the pilot can start the recovery sequence. 


Once the pilot has determined which way really is up, the first step is to make a power change that will start to improve the situation: if the airspeed is below cruise, go to full power; if airspeed is above cruise, pull the power to idle.  (Some say use Va instead of cruise, but trying to remember what the airplane's weight is at that moment to figure out Va is more workload than is necessary—simple is better when under high stress—cruise speed is a good reference.)

If the airplane is slow and approaching stall, full power buys time to avoid stalling. If the airplane is screaming downward, chopping power begins to stop the bleeding.


Unload the wing. Not zero or negative G—target one G or slightly less. If approaching a stall, it gets the wing away from the critical angle of attack. If the airplane is going fast, the fact that the trim is set for cruise means that it's already pulling the nose up hard and has onset substantial G loading on the airframe.

If you step back to primary training where your instructor told you that the definition of a steep turn is one where the bank tends to increase on its own, the self-perpetuating nature of diving spiral makes sense. In a steep turn, pulling back on the wheel will increase the angle of bank—to maintain the bank angle or decrease it requires opposite aileron. Once the airplane has been upset to a steep bank, the nose will drop, the speed increase and the nose up trim will do nothing but aggravate the situation by increasing the bank as the speed continues to increase.

While it may seem counterproductive to push the nose down when in a diving spiral, it's not—it unloads the wing, getting the airframe away from the G load that will break it. It also allows the pilot to take the next steps, which involve rolling—and you don't want to roll the airplane when it's under high G loads. The wing will withstand fewer Gs in a rolling pull up than it will in a symmetrical pull up due to the twisting loads on the wing in a rolling pull up.


Take a moment to center the ball and keep it there during the remainder of the recovery. The upset may have been induced or made worse by inadvertent rudder deflection—fix it. Do not use just the rudder to roll the airplane back to wings level flight. Any rolling needed should be coordinated.

It took the crash of an airliner on Long Island some years back to emphasize to pilots that the vertical stabilizer can be broken off from ruder deflection at speeds below Va.


With the wing unloaded—one G or less—make an aggressive, coordinated roll to wings level. Continue to resist pulling on the yoke/stick. If you're past 90 degrees of roll, pulling just yanks the nose toward the ground. If you're near stall, pulling won't help. If you're going fast, the trim is doing plenty of pulling for you—you'll need to push, maybe quite hard.


Once the wings are level, establish climb attitude and set maximum climb power. If you're still nose low once you get to wings level, aggressively raise the nose, but not so fast you risk overstressing the airframe (you may be pushing as the nose comes up) or stalling the airplane.

You've been through a high stress situation and are going to need some time to sort out what direction you're going, where you are and how high the airplane is—so it's wise to assume that you've lost significant altitude and default into an aggressive climb rather than level flight when recovering from an upset.


Pilots do a good job of recovering from run of the mill diversions of the airplane from straight and level flight—but VFR into IMC, VFR into sunlit haze, turbulence induced upsets in IMC and wake turbulence are still causing too many accidents each year. Thinking about what you'd do in an upset is a good first step toward handling it if it happens, practicing it during each recurrent dual session with an instructor is even better—and, in my opinion, if you can attend one of the dedicated upset training programs, go for it.

Rick Durden holds an ATP and CFI-AI with type ratings in the Douglas DC-3 and Cessna Citation 500 series. He is the author of The Thinking Pilot's Flight Manual or, How to Survive Flying Little Airplanes and Have a Ball Doing It. Volume I.