RethinkingUpsetTrainingUsing simulators to develop upset-recoveryprocedures could lead to negative trainingWilLIAM B. SCOTT/LOS ANGELES

irline pilots who have complet-ed upset recovery training maydiscover that a real aircraft ac-tually responds quite different-lythan their companies'motion-

base simulators, possibly leading toinadvertent lossof control during an ac-tual inflight emergency.

Further, pilots can be trained to userudder inappropriately and actuallycause structural damage or lossof a tailfin by aggressively cycling the rudderduring an upset event.

These and related issuesare being ex-plored bythe Societyof AutomotiveEn-gineers' S7committee, which deals withtransport-category aircraft flight deckand handling qualities standards. Workis ongoing,but the committee expectstoissue its findings in a new SAE Aero-space Information Report this year.

Motion-base simulators' aerodynamic mod-els and "feel" cues, such as g-Ioads (Nz), oftenare not representative of actual aircraft flyingoutside their normal operating envelopes.

A status report on the committee'sprogress was outlined in a technical pa-per written for the Society of Experi-mental Test Pilots' (SETP) 47th sym-posium. Steve Stowe, a captain forDelta Air Lines, and Tom Melody, di-rector of flight operations for BoeingCommercial Airplanes-Long Beach[Calif.], jointly presented the paper toapproximately 900 attendees. Stoweserves as chairman ofthe SAE S7 com-mittee, and Melody is a current mem-ber. Both are former u.S. Air Force testpilots and had tours as USAF TestPilotSchool instructors. Melodywas a test pi-lot on the first flights of then-McDon-nell Douglas MD-88 and MD-ll, andBoeing 717 transports. Stowe has morethan'13,OOOhr. in more than 140aircrafttypes, and is rated on the Boeing 727,737,757,767 and MD-11.

Followingthe loss,in November2001,of American Airlines' Flight 587 due totail fin failure after the Airbus A300 en-countered a mild wake vortex from aBoeing 747, the SAE S7 committeestarted exploringseveral related issues:. When and howrudder shouldbe used

www.AviationNow,comJawst

duringupset recoveriesin largeair trans-ports.. What role ground-based simulatorsplay in developing upset-recovery pro-cedures.. Rudder control system harmony-

how much rudder pedal force translatesto actual rudder deflection. .

Many airlines altered upset-recoverytraining after the AA 587 accident, butmost previously had taught the use of toprudder and full aileron-a "step-and-roll"maneuver-as initial recovery inputs in

use rudder as your initialor primaryrollcontrol, because it's not as easyto con-trol sideslip in big airplanes as it is insmall ones. Inertial effectscause delays,and rudder [response] isn't precise.Youcan build up large sideslip angles, evenafter you take the rudder deflectionout.That can cause what appears to be anabrupt onset of roll."

Aggressiverudder inputs to arrest thatrolling motion, coupled with delayedaircraft response, can lead to large rud-der inputs in the opposite direction,

triggering subse-quent large-am-plitude oscilla-tions in roll andyaw, the commit-tee determined.The resulting ex-cessive sideslip canlead to loss of con-trol, and "over-yawing" might"overstress thevertical tail, evenat airspeeds lessthan the hand-book maneuveringspeed," accordingto the SETP pa-per. An independ-

ent examination by Aviation Week &Space Technology came to similar con-clusions (A W&ST Nov. 25, 2002, p. 44).

Further, after an upset or uncom-manded excursion caused by turbulence,a wake-vortex encounter, flight controlsystem hardover or other trigger mech-anism, "you can find yourself at a highangle-of-attack. Then you'd better bereal careful with the use of rudder. To

depart [from controlled flight], you justhave to stall it and yaw it," Stowe said.

The SAE S7 committee also found

that training simulators are inappropri-ate for developing upset recovery tech-niques, because their aerodynamic mod-els aren't always accurate outside normaloperating envelopes. Airline training de-partments need to be aware that simu-lator fidelity simply isn't good enough toaccurately depict aircraft responses af-ter an upset, Stowe and Melody noted.

For example, one airline had devel-

response to a sharp roll-off (AW&STMar. 27, 1995, p. 43). "That works insmaller airplanes, like a [Beech] T-34-but probably isn't the best initial ma-neuver in a large airplane," Stowe said.

In 1998, aircraft manufacturers Boe-ing, Airbus and then-McDonnell Dou-glas issued recommendations to use fullaileron inputs for primary roll control af-ter a lateral upset, adding rudder only ifthe response to ailerons wasn't quickenough. A joint -manufacturer flight testdepartment bulletin warned: " . . . theamount of rudder needed will depend onthe airplane type and systems. However,at high AOA [angle-of-attack], pilots mustbe extremely careful when using rudderfor assisting lateral control as excessivesideslip can lead to loss of control."However, it appears the test pilots' warn-ings may not have been fully incorporat-ed into airline upset-training curriculums.

Stowe explained, "You don't want to

AVIATIONWEEK & SPACETECHNOLOGY/JANUARY12,2004 39

oped an upset-training profile that in-jected a rapid roll rate into the simu-lator, but artificially inhibited the rollcontrol response-spoiler effective-ness-until the "aircraft" rolledpast 135deg. of bank. "So, they were givingpi-lots the illusion that the only thing thatwould roll the aircraft was rudder-which wasn't the same as the airplane's[actualcontrol mode]," Stowe said."It'sOK to train in a simulator,but don't usethe simulator to develop procedures.And don't validate a profile and proce-dures in the simulator without gettingthe [aircraft]manufacturer involved.Wedon't want [training instructors] saying,'MyideaworkSin the simulator,sowe'regoing to use it in the [actual] airplane.'

"SAE S7 believes that, in general, an[effective]upset recoverywould involvefirst disconnecting the automation (au-topilot and autothrottles), then control-

ling AOA as the first order of business.Then get your velocity vector pointedwhere you want it to go, followedby an-alyzingthe situation," Stowe added.

The committee also is questioning"rudder control harmony" in some air-craft. Most have force-deflection pro-files that closely relate rudder pedalforce and rudder surface travel. In oth-er words, after a pilot exerts enoughpedal pressure to exceed the "breakoutforce" and make the rudder start mov-ing, the resulting control surface de-flection is proportional to rudder pedalforce. However, the committee notedthat, at high speeds, the A300-600'srud-der limiter restricts travel to :!:9.3deg.as a protection feature-but pedal forcerequired to get that much travel is only10lb. abovebreakout force.The aircrafthas a rudder breakout force of about 22lb., but only another 10 lb. of force is

needed to drive the control surface toits full available travel (9.3 deg.).

"That doesn't seem to be an ideal sys-tem-to have two~birds of the forceprofile devoted to o~ercoming friction)

and breakout [forces], then only one-third left to get full [allowable]rudderdeflection," Stowe said. Veridian andNASA Langley experts agreed.

Depending on manufacturers' feed-back, the SAE S7 group may advocatechangingSAE Aerospace Recommend-ed Practice41O4'ssectionon rudder har-mony to promote a better relationshipbetweenbreakout-and maximum-forces.That particularARP is devoted to trans-port aircraft handling qualities and sta-bility and control.

The committee also expects to rec-ommend that manufacturersincorporatemore flight-envelopeprotectionschemesinto flightcontrol systemdesigns. G