FAST / NUMBER 24 FAST / NUMBER 24 19

TTTThe idea for a jointindustry working group to produce an AirplaneUpset Recovery Training Aid* was firstproposed by ATA in June 1996. It was inresponse to increasing interest by the NTSB inaircraft loss of control accidents which, togetherwith Controlled Flight Into Terrain, cause alarge proportion of all accidents. They wereputting a lot of pressure on the FAA to producenew regulations covering this subject.The working group was a voluntary industryinitiative to see what could be done within theexisting regulations to improve the situation.

The joint industry team consisted ofrepresentatives of all sides of industry: aircraftmanufacturers, airlines, governmentalauthorities, and pilots’ unions. It was a goodexample of how the entire industry, designers,users, and regulators can co-operate on safetyissues that are common to everyone. It alsomarked a “first” in showing that the “Big 3”aircraft manufacturers could and will worktogether on technical, non-commercial issues.More than 80 persons coming from all aroundthe world, but principally from the USA,participated from time to time.

The end result of two years work is a trainingpackage including a video and a CD-ROM,giving an airplane upset recovery training aid.This package is on free issue to all ourcustomers, to use as they wish. However, all

members of the joint industry group agreed thatthe package is aimed at preventing loss ofcontrol accidents on conventional aircraft. It isnot aimed at protected Fly-by-Wire aircraft.

There is no need for this type of continuationtraining on protected aircraft, although ageneral knowledge of the principles involved isuseful for every pilot.

The content of the package is not the subjectof this article, but there are a few issues ofgeneral interest which I gained from myexperience as a member of the working groupwhich I would like to mention.

T here is no need

for this type of

continuation training on

protected

fly-by-wire aircraft

THE BEGINNING

The issue of upset training was notnew; major airlines around the world,and in particular in the USA, had al-ready produced Upset RecoveryTraining Programmes, or were usingone produced by another company.Amongst the members of the groupwere training pilots from AmericanAirlines, Delta, and United who werealready running such training pro-grammes in their simulators. Since thiswas essentially seen as a training issue.Initially the Flight Test Departments ofthe three main manufacturers were notinvolved. Airbus was represented byLarry Rockliff, Chief Pilot at AirbusTraining Centre in Miami. Right fromthe beginning there was a conflict be-tween the technical advice given by the

manufacturers’ training pilots and thatexpressed by those of the principal air-lines already practising upset training.They naturally considered themselvesto be the experts on this subject, basedon the many hours of training that theyhad already conducted on a large num-ber of pilots in their simulators.

At the beginning of 1997, the FlightTest Departments were asked to comein to support their training pilots. Fromthen on, the chief test pilots of the threemajor manufacturers became membersof the working group. But the conflictover the different opinions on aircrafthandling and recovery techniques con-tinued for a long time until we finallyachieved agreement at the last meetingin January 1998. The reasons for thesedifferences of opinion are the subject ofthis article.

* The Training Aid itself was

the basis of the article entitled

“AERODYNAMIC PRINCIPLES

OF LARGE AIRCRAFT UPSETS”

that appeared as a Special Edition

of FAST in June 1998.

18

AAAAIIIIRRRRPPPPLLLLAAAANNNNEEEEA test pilotÕs point of view

UUUUPPPPSSSSEEEETTTT RRRREEEECCCCOOOOVVVV EEEERRRRYYYYBy Captain William Wainwright

Chief Test PilotAirbus Industrie

p 16 / 32 1/06/99 9:03 Page 18

FAST / NUMBER 24 21FAST / NUMBER 24

D o not confuse an

approach to the stall and

a full stall. An approach

to stall is controlled

flight. An airplane that is

stalled is out of control

and must be recovered.

THE DIFFERENCESOF OPINION

The differences of opinion were mainlyconcentrated in the following areas:● Procedures versus general advice● Ease of training versus failure cases● Stalling● Use of rudder● Use of simulators.

It is worth saying that there wasnever any difference of opinion be-tween the three test pilots on the group.Although we come from different back-grounds and have worked in differentorganisations with different work cul-tures, we always agreed on our techni-cal advice.

PROCEDURESVERSUS

GENERAL ADVICE

The airlines wanted simplified proce-dures which were common to all air-craft in their fleets and which were easyto teach and easily reproducible. This isunderstandable because everyone is in-terested in having a standard product at

the end of his training programme.And this is what they already had

with the Airplane Upset RecoveryTraining that they were already doing.

For the training managers fromAmerican Airlines, Delta, and United,the only thing necessary was to give

an overall industry approval to theirexisting programmes; they al-

ready worked, because themany pilots that had un-

dergone training allcame out of it with

the same stan-dardised reactionsto the standard

upsets. For them, this was thenecessary proof that their

training programme worked.Where we differed was in our convic-

tion that there is no such thing as astandard upset and our reluctance to en-dorse simplified procedures for recov-ery from an upset.

We wanted a general knowledgebased approach, as opposed to a rulebased one. For this, after proposingsome initial actions, we talk about “ad-ditional techniques which may betried”. This obviously is more diffi-cult to teach.

Where we reached a compromise wasin the order of presenting the variousactions that might be considered to re-cover the situation. For us, the order ofpresentation is for guidance only; it rep-resents a series of options that should

be considered and used as appropriateto the situation. It is not meant to repre-sent rigid procedures that must be fol-lowed in an exact sequence. However,the order can be used in training scenar-ios if a procedural approach is neededfor training.

The airline instructors also wantedprocedures which would apply to all theaircraft in their fleets. This meant thatthey were against certain actions, because they were inappropriate on others. For example, the thrust effectsof underwing-mounted engines werebeing ignored, whereas it has a signifi-cant influence on recovery. Again, wereached a compromise by using the fol-lowing words: “ if altitude permits,flight tests have shown that an effectivemethod to get a nose-down pitch rate isto reduce the power on underwing-mounted engines”.

EASE OF TRAININGVERSUS

FAILURE CASES

The training that was already beingdone, considered upsets as being due tomomentary inattention, with a fully ser-viceable aircraft, that was in trim whenit was upset. We wanted to considerother cases that involve aircraft withtemporarily insufficient control author-ity for easy recovery. This of coursecomplicates the situation, because re-covering an aircraft which is in trim,possessing full control authority andnormal control forces, is not the sameas recovering an aircraft with limitedcontrol available or with unusual con-trol forces.

Thus, for us, an aircraft that isout-of-trim, for whatever reason, shouldbe re-trimmed. Whereas the airline in-structors were against the use of trimbecause of concerns over the possibilityof a pilot overtrimming and of trim run-aways which are particularly likely onsome older aircraft types which are stillin their fleets.

We spent a lot of time discussing theuse of elevator trim and we neverreached agreement. All the major USairlines were adamant on their policy torecover first using “primary controls”which excluded any reference to trim-ming.

Again, a compromise was necessary.What we have done is to talk about us-ing trim if a sustained column force isrequired to obtain the desired responsewhilst mentioning that care must beused to avoid using too much trim.And, the use of trim is not mentioned inthe simplified lists of actions to betaken.

R emember, in an upset

situation, if the airplane

is stalled, it is first

necessary to recover from

the stall before initiating

upset recovery

techniques.

I f altitude permits,

flight tests have shown

that an effective method

to get a nose-down pitch

rate is to reduce the

power on underwing

mounted engines.

20

STALLING

Another aspect that was beingignored in the existing training was thestall. By this I mean the difference be-tween being fully stalled and the ap-proach to the stall. In training, youdo an approach to thestall with a recoveryfrom stick shaker, which is often done byapplying full thrust and maintaining ex-isting pitch attitude in order to recoverwith minimum loss of height. Height can-not be maintained if an aircraft is actuallystalled and should be of secondary impor-tance.

Even those pilots who do stalls onairtests, as might be done after a heavymaintenance check, only do them withgentle decelerations, and they recover im-mediately without penetrating very farbeyond the stalling angle of attack. Thereis a world of difference between beingjust before, or even just at, the stall, andgoing dynamically well into it.

When we started our discussions, thetraining being given in the airlines to re-cover from excessive nose-up pitch atti-tudes emphasised rolling rapidly towards90° of bank. This is fun to do, and it wasnot surprising to find that most of the in-structors doing the training wereex-fighter pilots who had spent a lot oftime performing such manoeuvres in an-other life. The training was beingdone in the same way, with an aircraftstarting in trim with a lot of energy andrecovering while it still had some.However, the technique being taughtonly works if the aircraft is not stalled.

We start our briefing on recovery tech-niques with the following caution:

Recovery techniques assume that theairplane is not stalled. If the airplane isstalled, it is imperative to first recoverfrom the stalled condition before initiat-ing the upset recovery technique.

Do not confuse an approach to the stalland a full stall. An approach to stall iscontrolled flight. An airplane that isstalled is out of control and must be re-covered.

A stall is characterised by any, or acombination of the following:● Buffeting, which could be heavy attimes● Lack of pitch authority● Lack of roll control● Inability to arrest descent rate.

To recover from a stall, the angle of at-tack must be reduced below the stallingangle. Apply nose down pitch control andmaintain it until stall recovery. Undercertain conditions with under-wingmounted engines, it may be necessary toreduce thrust to prevent the angle of attack from continuing to increase.

Remember, in an upset sit-uation, if the airplane isstalled, it is first necessary torecover from the stall beforeinitiating upset recovery tech-niques.

This is something that we arewell aware of in testing, but itwas either being totally ignoredor misunderstood. I consider theinclusion of this note to be one ofour most important contributions.

USE OF RUDDER

We also spent a lot of time dis-cussing the use of rudder. The exist-ing training courses all emphasisedusing rudder for roll control at lowspeeds. It is true that the rudder re-mains effective down to very lowspeeds, and fighter pilots areaccustomed to using itfor “scissor”

e v a -sive ma-

noeuvres whenflying not far from

the stall. But large airlin-ers, with all the inertias that they pos-sess, are not like fighter aircraft. Basedon our experience as test pilots we arevery wary of using rudder close to thestall. It is the best way to provoke a lossof control if not used very carefully,particularly with flaps out.

We finally got the training managersto agree to play down the use of rudderin their existing courses. But we do notsay never use the rudder at low speed.We say that, if necessary, the aileroninputs can be assisted by coordinatedrudder in the direction of the desiredroll. However, we also caution that “ex-cessive rudder can cause excessivesideslip, which could lead to departurefrom controlled flight”.

But why did we have so much diffi-culty in convincing the training pilotsthat it is not a good idea to go kickingthe rudder around at low speed?

Their reply was always the same; butit works in the simulator! This leads meon to my last point.

E xcessive rudder

can cause excessive

sideslip, which could lead

to departure from

controlled flight.

p 16 / 32 1/06/99 9:05 Page 20

FAST / NUMBER 24 23FAST / NUMBER 2422

(based on a simplified model of windtunnel data) or for possible asymmetricstalling of the wings. Also, the rangefor one engine inoperative is much lessthan the range for all engines operatingand linear interpolation is assumed be-tween low and high Mach numbers.Wind tunnel data goes further. For ex-ample, a typical data package wouldcover the areas described in table 2.

In fact, this is a perfectly adequatecoverage to conduct all normal trainingneeds. But it is insufficient to evaluaterecovery techniques from loss of con-trol incidents. Whereas, the trainingmanagers were all in the habit ofdemonstrating the handling characteris-tics beyond the stall; often telling their

trainees that the rudder is farmore effective than aileronand induces less drag and has novices! In short, they were devel-oping handling techniques fromsimulators that were outside theirguaranteed domain.

Simulators can be used for upsettraining, but the training should be con-fined to the normal flight envelope. Forexample, training should stop at thestall warning. They are “ virtual” air-craft and they should not be used to de-velop techniques at the edges of theflight envelope. This is work for test pi-lots and flight test engineers using theirknowledge gained from flight testingthe “ real” aircraft.

C oncentrate everyoneÕs

attention on taking

action early enough to

prevent the occurrence of

loss of control.

S imulators should not be

used to develop

techniques at the edges

of the flight envelope.

Table 1Sideslip Angle of attack

SLATS OUT

● All Engines Operating Around neutral Between 0°and 22°Between + 15° and -15° Between 0° and 12°

● One Engine Inoperative Between +8° and -8 Between 5° and 12°

SLATS IN, LOW MACH

● All Engines Operating Around neutral Between 0° and 12°Between +10° and -10° Between 2° and 9°

● One Engine Inoperative Between +8° and -8° Between 2° and 8°

SLATS IN, HIGH MACH

● All Engines Operating Around neutral Between 0° and 5°Between +5° and -5° Between l° and 3°

● One Engine inoperative Between +2° and -2° Between 1° and 3°

Table 2Sideslip Angle of attack

SLATS OUT From +18° to -18° From -5° to 25°SLATS IN, LOW MACH From +18° to -18° From -5° to 12°SLATS IN, HIGH MACH From + 8° to -8° From -2° to 8°

USE OF SIMULATORS

We manufacturers were very concernedover the types of manoeuvres beingflown in simulators and the conclusionsthat were being drawn from them.Simulators, like any computer system,are only as good as the data that goesinto them. That means the data packagethat is given to the simulator manufac-turer. And we test pilots do not deliber-ately lose control of our aircraft just toget data for the simulator. And evenwhen that happens, one isolated inci-dent does not provide much informa-tion because of the very complicatedequations that govern dynamic manoeu-vres involving non-linear aerodynamicsand inertia effects.

The complete data package includes apart that is drawn from actual flighttests, a part that uses wind tunnel data,

and the restwhich is

pure ex-trapolation.

It should be obvi-ous that firm conclusions

about aircraft behaviour can only bedrawn from the parts of the flight enve-lope that are based on hard data. This infact means being not far from the centreof the flight envelope; the part that isused in normal service. It does notcover the edges of the envelope. Ishould also add that most of the dataactually collected in flight is fromquasi-static manoeuvres. Thus, dy-namic manoeuvring is not very wellrepresented. In fact, a typical data pack-age has flight test data for the areas de-scribed in Table 1.

In other words, you have reasonablecover up to quite high sideslips andquite high angles of attack (AOA), butnot at the same time. Furthermore, thematching between aircraft stalling testsand the simulator concentrates mainlyon the longitudinal axis. This meansthat the simulator model is able to cor-rectly reproduce the stalling speeds andthe pitching behaviour, but fidelity isnot ensured for rolling efficiency

CONCLUSION

It may seem that there is a gulf between the world of testing and that of training,but the message that I would like to get over in this article is that we can alllearn from each others’ experiences and that we should not do things in isola-tion. It is all about working together, which is what we all did when we met toprepare and review this training aid, even though we sometimes had some verylively sessions. And there is one word that crops up frequently: compromise.Life is a compromise, and you always have to search for that ideal point be-tween two extremes which Aristotle called “the golden mean”. By finding suit-able compromise solutions, our two worlds of testing and training were able toresolve their differences and develop something that satisfied everyone.

Of course there are also some points about piloting that were raised duringour discussions which I feel should have a larger audience. They are important,but they should be kept in context. On the whole they are related to recovery ofan aircraft which is already out of control, or is about to be. This is an area inwhich the test pilots have some experience which other pilots do not normallyhave, because the aim of training should be to prevent an aircraft getting intosuch a situation. The end result of all the discussions that took place was to con-centrate everyone’s attention on taking action early enough to prevent the oc-currence of loss of control. We put the emphasis on training within the knownflight envelope, and to avoid going into that part which cannot be guaranteedone hundred per-cent and which may have a negative effect.

In conclusion, we must use each other’s competences in the areas where theyare expert. Of course the training programmes must be designed by training pi-lots, but these training programmes must stay in a reasonable flight envelope.And the test pilots are best qualified to define the flight envelope that should beused. That is what we now have with this joint industry training aid, which is avery good example of how we can all work together in everyone’s interest. n

p 16 / 32 1/06/99 9:05 Page 22