Free Upset Recovery Flight Training

The first 2,000 pilots to contact this FAA-funded program can get free upset recovery training—in a real jet.

By Jan W. Steenblik, Technical Editor
Air Line Pilot, February 2003, p. 12

For decades, the leading cause of airline fatalities worldwide has been controlled flight into terrain (CFIT). In recent years, however, thanks in large measure to the joint efforts of aviation authorities, airlines, and a number of aviation safety organizations (including ALPA), plus introduction of enhanced ground proximity warning systems, CFIT, while still a serious safety concern, is no longer the leading cause of airline fatalities.

The new No. 1 killer? Loss-of-control accidents.

You’ve read about the accidents; you’ve tried to imagine what they were like. Maybe you’ve even flown a ground-based flight simulator programmed to reproduce these loss-of-control events. But the sim can’t load you up with realistic g-forces or put the pucker factor in you that comes from knowing you are in a real airplane, 3 miles above the oh-so-hard Earth, rolling out of control, and the airspeed needle heading for the barber pole.

Upset recovery training for pilots has evolved during the past several years from unusual attitude training to selected event training, the Advanced Aircraft Maneuvering Program, and the industrywide Upset Recovery Training Aid.

Now, in Roswell, N.M., famed for its UFO sightings, something strange but wonderful is going on—the next stage in the evolution of upset recovery training: The federal government is picking up the tab for 2,000 airline pilots to undergo free upset recovery flight training—not just in a ground-based simulator, but in a real jet configured as an inflight simulator.

The FAA, using congressionally directed funds, is sponsoring this research study. The congressional directive to conduct the study was the result of leadership that the New Mexico delegation provided. Five organizations have joined together to create the Alliance for Flight Safety Research, which is administering the program through its Flight Safety Training Center (FSTC) at Roswell Industrial Air Center (the Roswell airport).

The prime contractor is CUBRC, which stands for Calspan-UB [University of Buffalo] Research Center. The other partners are Veridian Engineering’s Flight and Aerospace Research Group; the State of New Mexico Department of Transportation, Aviation Safety Branch; the City of Roswell, N.M., Roswell Industrial Air Center; and Eastern New Mexico University, Roswell.

The overall goals of the program are to optimize an inflight simulation-based upset recovery training program, and to begin to reduce the loss-of-control accident rate by training 2,000 airline pilots over a 5-year period.

Regarding the first goal, the study aims to develop an improved inflight training program based on measurements of pilot performance and on pilot feedback, and to develop the next generation of flight simulation technologies for such training. Another important question the study will try to answer is, "How much pilot retraining in upset recovery is required over the long term?"

ALPA’s Upset Recovery Project Team—US Airways Capts. John Cox (ALPA Executive Air Safety Chairman), Terry McVenes, and Robert Sumwalt, plus First Officer Dave Hayes (Northwest) and Capt. Lloyd Beaule (Air Canada Jazz)—have been through the training and participated in its development.

Air Line Pilot joined two ALPA members—First Officers Tom Phillips (US Airways), currently flying the A320, and Alan Moore (Pinnacle), currently flying the CRJ-200—as they went through the 2-day training course in Roswell in early December. The FSTC limits each class to four pilots.

Jim Priest, manager of program development at the FSTC, and Scott Buethe, one of five Veridian demonstration test pilots who take turns teaching the course, welcomed the ALPA group.

The program’s first morning is devoted to classroom discussion of aerodynamics and recovery procedures.

After lunch, Priest briefs one or two pilots to fly the FSTC’s aerobatic Bonanza, and they fly. Buethe briefs the other(s) on the Learjet and guides them through procedures training on a fixed-base ground simulator before heading to the real airplane—one of two unique birds reworked to give the subject pilot in the right seat a fly-by-wire airplane that can be programmed to simulate variable CG locations and stability, and to suffer several kinds of loss-of-control emergencies.

On Day 2, the pilots swap mounts. They spend their spare time practicing upset recoveries in the ground-based sim and reviewing the FSTC’s impressive library of videotapes and printed materials regarding loss-of-control issues.

"You have to understand the theory first and then develop the proper procedures for your airplane and your airline," Buethe told the pilots.

An airplane upset, he said, is defined as "an airplane unintentionally exceeding the parameters normally experienced in line operations or training." Specifically, an upset occurs when pitch exceeds 25 degrees nose up or 10 degrees nose down, or bank angle exceeds 45 degrees.

Causes of upsets include environmental (icing, wake turbulence, and weather turbulence), system anomalies, control failures, pilot inputs (incapacitation, distraction, or flight crew given inadequate or faulty information), or some combination of these factors.

Buethe reviewed basic aircraft aerodynamics, emphasizing aircraft control. Much of the material certainly should be familiar to professional aviators, but he touched on certain aspects of aerodynamics that most pilots don’t think about during routine flying. During an upset, however, understanding some of these basic principles could make the difference between a scary story and a smoking hole.

For example, "corner speed" is the lowest speed at which maximum g-loading on the wings is available, resulting in minimum altitude loss in recovery from a nose-low situation. Corner speed often is the same as maneuvering speed.

Regarding roll control, a balance exists between aileron input, roll inertia (which causes the airplane to overshoot a desired bank angle), and resistance to roll rate (i.e., damping). Ailerons are less effective at higher AOAs and more effective at higher airspeeds.

"Don’t forget that proverse yaw—the opposite of adverse yaw—can be used for recovery in some scenarios," Buethe advised.

Dihedral effect, he noted, increases with AOA, and a T-tail magnifies dihedral effect (i.e., sideslip is more effective on the vertical tail).

The September 1994 crash of USAir Flight 427 near Aliquippa, Pa., involved a rudder hardover and subsequent roll upset involving dihedral effect. "This kind of situation can happen so quickly that, if you don’t react appropriately and immediately, you probably won’t have time to recover," Buethe warned. "The only way to prepare for this is to practice—but if you try to practice this kind of stuff in a ground simulator, you won’t get adequate training."

He added, "You can completely eliminate dihedral effect [the rolling moment induced by sideslip] by unloading the wing to zero g."

Regarding Dutch roll (combined rolling and yawing caused by sideslip) and yaw damper failure, Buethe noted, "You can be a human yaw damper by opposing the yaw rate with rudder pedal pressure—you oppose the rate. In real life, it’s very difficult to do, especially in instrument conditions."

Spoiler control, he said, tends to increase proverse yaw, which increases Dutch roll tendency.

High altitude creates its own effects—the "coffin corner," where the airplane can be caught between Mach overspeed and aerodynamic stall, decreased damping, Mach effects requiring trim changes, and less control power (because of lower indicated airspeed).

High speed can bring on aileron buzz, shift in the aerodynamic center, buffet, aeroelastic effects, and worries about hinge-moment limits in airplanes with hydraulic flight controls.

Low speed reduces control effectiveness, stability, and stall margin; increases dihedral effect; and causes rudder blanking, poor Dutch roll characteristics, and flow separation on different parts of the airplane.

With the review of transport airplane aerodynamics out of the way, the time had come for the real meat of the course: upset recoveries.

"Upsets occur very infrequently," Buethe acknowledged. He added that the "startle factor" may significantly delay a pilot in starting recovery from an upset. "I recommend that pilots say, out loud, ‘I have a flight control problem,’" he advised.

Crew coordination during an upset recovery may require a positive transfer of aircraft control. "The wording should be short, clear, and concise," Buethe cautioned. "The wording I like best came out of the flight test community. The best I’ve heard is, ‘You/I have the flight controls’—that makes it clear exactly what you or the other pilot now controls."

One crew member should be designated to turn off the cockpit automation. "The PNF [pilot not flying] has a lot of work to do," Buethe pointed out. "You should figure out in advance what each pilot can or should do in an upset recovery."

The autopilot and autothrottles should be disconnected at the first sign of a problem. "Whether you should turn off the yaw damper," Buethe said, "depends on whether you think it’s part of the problem."

The pilot flying must quickly but correctly evaluate the airplane’s attitude (pitch and bank angles), velocity vector (airspeed, AOA, and sideslip angle), and altitude (low- vs. high-altitude issues—e.g., avoiding flight into terrain or obstacles at low altitude, and avoiding the "coffin corner" at high altitudes).

"We emphasize use of instruments in recovery because windshields can be unreliable," Buethe stressed. "They can get clouds on them."

The flight crew must crosscheck for erroneous information, checking captain’s vs. first officer’s instruments, plus the standby instruments. Outside visual references may be usable.

The pilot flying should use the aircraft’s gyroscopic instruments to determine aircraft attitude, Buethe said. That includes the pitch ladder on the attitude director/indicator, because the pitch ladder is always displayed at right angles to the horizon.

Citing a B-737 accident in Colombia attributed to instrument failure in IMC, Buethe cautioned, "The important thing is to not get fixated on the little symbols—you need to have the big picture."

To determine the airplane’s velocity vector, the pilot must check airspeed, AOA (inferred via g and airspeed combination, based on experience, plus stickshaker/pusher and control column force), and sideslip angle. Regarding sideslip angle, Buethe acknowledged, "Sometimes it can be difficult to sense sideforce with your body, especially if you’re tensed up."

Nonintuitive factors, he added, "may inhibit your ability to bring the airplane back to straight-and-level flight." These may include unexpected poor handling qualities (e.g., high inertia, high Mach number, or low indicated airspeed), lack of experience in using full control inputs in transport-category airplanes, and such distractions as flying debris, audio warnings and lights, unfamiliar g forces, and improper use of seat belts and/or shoulder straps.

Every pilot should know how to use alternate controls to recover from an upset.

"You can use sideslip to induce roll," Buethe reminded the class. "Asymmetric thrust is your alternate rudder. How does the airplane manufacturer size the rudder? Certification requirements mandate that the rudder be able to counter full asymmetric engine thrust at low airspeeds."

Capt. Cox cautions, however, that, as a result of the ongoing NTSB investigation of the Nov. 12, 2001, inflight breakup and crash of American Airlines Flight 587 in New York, both the Safety Board and ALPA have issued safety bulletins warning that sequential, aggressive, full, and opposite rudder inputs may cause structural damage or failure.

Buethe noted that a lot of airlines are reluctant to endorse the concept of splitting the throttles to control yaw and induce roll. Capt. Cox warns, "Adverse thrust is most effective after the situation is stabilized. You should exercise extreme caution when considering using adverse thrust during a dynamic event."

Crossover speed—the minimum airspeed at which a particular airplane model in a particular configuration can be controlled with full aileron input to counter a rudder hardover—also is important in using sideslip to control roll. Capt. Cox notes, "This is of particular interest to B-737 pilots because the crossover speed for the B-737 is higher than those of other airplane types."

Pitch control can be achieved by banking to bring the nose down. Increasing or decreasing engine thrust symmetrically can also cause a pitching moment, but the effects differ between underwing engines and those mounted above the center of gravity.

Also, "you need to know what the particular effects of extending or retracting spoilers, flaps, and landing gear are in your airplane," Buethe advised. "In many airplanes—the Lear is one of them—extending the landing gear will give you a nose-up pitching moment. In the Lear, it can be as much as 0.5 g."

The horizontal stabilizer and pitch trim also can be used for primary pitch control. So can shifting the center of gravity by moving passengers, fuel, or cabin cargo.

The program covers procedures for specific upset scenarios:

• Nose-high, wings-level—Disengage the autopilot and autothrottle; stop the nose-up pitch rate (using as much as full nose-down elevator); roll to a target of 45 degrees of bank, not to exceed 60 degrees, if necessary; approaching the horizon, level the wings; check airspeed, adjust thrust; adjust pitch attitude.

"You’d be surprised how frequently pilots think they’re against the forward stop [of control column movement]," Buethe noted, "when actually they’re not."

• Nose-low—Recover from the stall, if necessary; roll to the nearest upright bank orientation, using the sky pointer on the attitude display; pull the nose to above the horizon, using trim if necessary; and reduce thrust.

"How hard do you pull?" Buethe asked. "First, you mustn’t hit the ground; over-g may be required. Second, you must never exceed the stall AOA. Third, you must be alert to high-speed buffet.

"Never add thrust to speed up to corner speed," he warned, "especially in a clean jet, because you’ll just blow right through the corner speed. If you’re over speed or expect to hit the ground, add drag." Adds Capt. Cox, "Many jet transport pilots don’t know the corner speed for their airplane."

• Ice-induced roll (ice ridge ahead of aileron; asymmetric ice accumulation)—Reduce AOA by moving pitch control forward as required, increasing airspeed, extending wing flaps, and leveling the wings; set appropriate power; check for ice, and do not retract flaps unless the top of the wing is clear of ice; check the ice protection system and ensure it is functioning symmetrically.

• Ice-induced tailplane stall (characterized by a pitch trim change, pulsing in unboosted controls, and a pitch down, this type of pitch upset can be triggered by extending flaps or by increasing airspeed or power)—To recover, reset flaps; apply nose-up elevator pressure; decrease airspeed; and use power only as needed (high engine power settings may adversely affect response to tailplane stall conditions at high airspeeds in some airplanes).

Physiological effects of an upset include the body’s initial response to stress (i.e., the "startle factor")—an adrenalin rush (the fight-or-flight response), plus an emotional response—and after-effects such as confusion, shaking, and dulling of the senses.

Pilots must be aware of the natural human tendency in stressful situations to block out the world around oneself and to stay with the wrong course of action. "You accept only information that supports your chosen course," Buethe explained.

To deal with the situation, he continued, "Verbalize what you are doing and why. That normalizes the cockpit; it gets other crew members working with, rather than against, you. It also reduces the startle factor for others.

"Avoid escalating the situation," he added. "Shouting makes it worse."

Pilots can prepare for such stressful situations as upsets by exploring "what-if" scenarios. "Building a library of possible solutions," Buethe said, "allows you to select, rather than develop, a solution. It’s a much more efficient problem-solving strategy."

Because F/O Phillips was the more experienced pilot, Priest suggested he practice in the ground-based simulator to fly the Lear first and let F/O Moore fly the Bonanza that afternoon.

"Our Bonanza was a Lufthansa ab initio trainer," Priest explained. "It’s fully aerobatic— +6, -3 gs. Not many were built. Our goal was to come up with a good surrogate for a transport category airplane—limited cockpit visibility and a conventional control yoke. The airplane has a rudder/aileron interconnect, and it has a max roll rate of about 60 degrees per second—about the same as a Boeing 737.

"This is not to make you a good aerobatic pilot; it’s to get you comfortable with basic recovery techniques, g forces, and aerobatic fundamentals. I generally find this to be a straightforward, intuitive experience. We don’t take data or video in the Bonanza."

The Bonanza flight profile includes a brief introduction to stability—control forces and harmony, pitch and roll response, speed stability, and Dutch roll. Then the subject pilot moves on to basic maneuvers, such as lazy 8s, to get a better feel for the airplane.

Next comes g-awareness. Priest demonstrates 2- and 3-g pull-ups (the airplane has a g-meter), then lets the subject pilot attune his or her body to what that amount of positive g feels like. After the pull-up comes the pushover, the goal being to feel that 0.5 g makes one light on the seat cushion but not snug against the seatbelt.

Then it’s on to a stall series, starting with a 1-g, power-off stall ("horn, burble, buffet, break"). Next on the list is an accelerated stall—100 KIAS, 60 degrees of bank, and a smart pull to the aft limit of the yoke. "The airplane won’t have time to talk to us in an accelerated stall," Priest explains.

A loop and an aileron roll are next on the agenda. Finally, the pieces are put together with unusual attitude recoveries. Here, Priest emphasizes the value in unloading to improve roll performance and control pitch rate. The Bonanza session lasts approximately 45 minutes.

Veridian’s two Learjets—and a third in the works—are unique inflight simulators with conventional flight controls on the captain’s side but fly-by-wire (FBW) controls on the copilot’s side. The FBW computers replicate the handling and performance characteristics of a generic large transport. The left-seat pilot can program the FBW computers to introduce various faults or failures that will induce upsets. Either pilot can disconnect the FBW computers at any time and instantly restore the airplane to a conventional Learjet.

The Learjet flight lasts about 75 minutes, and it is a workout indeed. During the climb to the high teens, Buethe hands over the airplane to the subject pilot.

At altitude, he programs the airplane to simulate mid-range CG. The subject pilot tries a 2.5-g pitchup, followed by a 0.5-g pushover to recover. Next comes the same maneuver with the airplane simulating a forward CG—significantly higher pitch control forces required. Again, but with an aft CG—the opposite effect.

The workload jumps dramatically when Buethe programs the airplane to be unstable in pitch. Trying to keep the pitch angle somewhere near level can be done, but it takes constant work and generates a lot of sweat.

On to three upset scenarios without training in the airplane—an aileron hardover, a rudder hardover, and a pitch runaway scenario.

The rest of the flight is devoted to intensive training in these and the other upset scenarios. The controls go hard over or freeze. The brown New Mexico desert and the blue sky roll, pitch, and yaw dizzily; the g-forces rise and fall; the belts go loose and tight.

The final exercise in the Learjet flight is a complete hydraulic failure—the same scenario the flight crew of United Flight 232 faced when they tried to land their crippled DC-10 at Sioux City, Iowa, in July 1989. The airplane can be controlled—after a fashion—with differential thrust and pitch trim.

Descending from the high teens and getting the airplane lined up on final approach seem reasonably easy; trying to get the airplane, wings level, into the touchdown zone at the proper airspeed and sink rate—i.e., to land the airplane—is another matter entirely. This sobering exercise makes you grateful to have flight controls—and determined to know how to recover from an upset even if you lose them.