Conference on Aircraft Operating Problems

NASA

SP-83

f_irhhNTl_Ir D IP NTf-_I[Z_ _

AIRCRAFT

OPERATING

PROBLEM,S

LANGLEY RESEARCH CENTER

MAY 10-12, 1965

m

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

NASA SP-83

CONFERENCE ON

AIRCRAFT OPERATING

PROBLEMS

Langley Research Center

May 10-12, 1965

Scientific and Technical Information Division 1 9 6 5 NATIONAL AERONAUTICS AND SPACE ADMINISTRATION Washington, D.C.

For sale by the Clearinghouse for Federal Scientific and Technical Information Springfield, Virginia 22151 - Price $3.00 PREFACE This compilation includes papers presented at the NASA Conference on Aircraft Operating Problems held at the Langley Research Center on May lO-12, 1965. Contributions were made by representatives of the Ames Research Center, the Flight Research Center, and the Langley Research Center of NASA, as well as by representatives of the Federal Aviation Agency.

CONTENTS PREFACE .................................. i GENERAL STUDIES i. RECENT STUDIES OF RUNWAY ROUGHNESS ................. .. 1 By Garland J. Morris and Albert W. Hall_ NASA Langley Research Center By Walter B. Horne and Upshur T. Joyner, NASA Langley Research Center By Roy Steiner, NASA Langley Research Center .

PRELIMINARY STUDY OF STEEP INSTRUMENT APPROACH OF THREE CONVENTIONAL AIRCRAFT ......................

By Albert W. Hall 3 Robert A. Champine 3 and Donald J. McGinley, Jr., NASA Langley Research Center SOME FACTORS AFFECTING FATIGUE OF AIRCRAF_ STRUCTURES ........

5. 37 By Herbert F. Hardrath, NASA Langley Research Center , USE OF VERY WEAK RADIATION SOURCES TO DETERMINE AYRCRAF_ By Fred J. Drinkwater III and Bernard R. Kibort, NASA Ames Research Center A LABORATORY INVESTIGATION OF TURBULENCE DETECTION USING A LASER , 53 By Kent Bourquin and Fred H. Shigemotoj NASA Ames Research Center , 61 SIMULATOR EVALUATION OF A DISPLAY FOR MANUAL ZER0-ZERO LANDINGS .

By Joseph G. Douvillier, Jr., NASA Ames Research Center , FACTORS RELATING TO THE AIRPORT-COMMUNITY NOISE PROBL_M .......

By Harvey H. Hubbard, Jimmy M. Cawthorn, and W. Latham Copeland, NASA Langley Research Center 10.

PRELIMINARY MEAS_S OF TAKE-0FF AND LANDING NOISE FROM A By Carole S. Tanner and Norman J. McLeod, NASA Flight Research Center iii SUBSONIC AIRCRAFT OPERATIONAL EXPERIENCES OF TURBINE-POWERED COMMERCIAL ll.

TRANSPORT AIRPLANES .......................

By Paul A. Hunter and Walter G. Walker, NASA Langley Research Center SIMULATOR STUDIES OF THE DEEP STALL ................

12.

. By Maurice D. White and George E. Cooper, NASA Ames Research Center i13 DEEP-STALL AERODYNAMIC CHARACTERISTICS OF T-TAIL AIRCRAFT ......

13: By Robert T. Taylor and Edward J. Ray, NASA Langley Research Center FLIGHT TESTS RELATED TO JET-TRANSPORT UPSET AND TURBULENT-AIR 14.

PENETRATION ...........................

By William H. Andrews_ Stanley P. Butchart_ Thomas R. Sisk, and Donald L. Hughes, NASA Flight Research Center SIMULATOR INVESTIGATIONS OF THE PROBLEMS OF FLYING A SWEPT-WING 15.

TRANSPORT AIRCRAFT IN HEAVY TURBULENCE .............

By Richard S. Bray and William E. Larsen, NASA Ames Research Center SUPERSONIC AIRCRAFT 16. A SIMULATOR STUDY OF TAKE-OFF CHARACTERISTICS OF PROPOSED SUPERSONIC TRANSPORTS ......................

BY Charles T. Jackson, Jr., and C. Thomas Snyder, NASA Ames Research Center By Thomas A. Toll, NASA Langley Research Center 18. EFFECTS OF THE SUPERSONIC TRANSPORT ON THE AIR TRAFFIC CONTROL SYSTEM ..........................

By Joseph P. 0'Brien and Andrew L. Sluka, Federal Aviation Agency EFFECTS OF THE AIR TRAFFIC CONTROL SYSTIM ON THE 19.

By Norman S. Silsby, Milton D. McLaughlin, and Michael C. Fischer, NASA Langley Research Center iv By Thomas R. Sisk, Kirk S. Irwin, and James M. McKay, NASA Flight Research Center 21.

CONSIDERATION 0FFUEL REQUIREMENTS FOR SUPERSONIC TRANSPORT By Joseph W. Stickle, NASA Langley Research Center 22.

THE EFFECT OF YAW COUPLING IN TUENING MANEUVERS OF LARGE By Walter E. McNeill and Robert C. Innis, NASA Ames Research Center 23. AN ASSESSMENT OF A TITANIUM ALLOY FOR SUPERSONIC TRANSPORT BY George J. Heimerl and Herbert F. Hardrath, NASA Langley Research Center e4.

IMPLICATIONS OF THE EFFECTS OF SURFACE T_MPE_AND By John B. Peterson, Jr., and Albert L. Braslow_ NASA Langley Research Center eS.

By Harry W. Carlson, F. Edward McLean, and Wilbur D. Middleton, NASA Langley Research Center e6.

SIGNIFICANCE OF THE ATMOSPHERE AND AIRCRASVf OPERATIONS By Domenic J. Maglieri and David A. Hilton, NASA Langley Research Center /'_. 1_1_T'1_ A T AXTT A mTf'_'kT .... _,_-_., _. J-_11_,I_ AIRCRAFT eT.

OPERATIONAL EXPERIENCES OF GENERAL AVIATION AIRCRAFT ......... " 257 By Joseph W. Jewel, Jr., and Walter G. Walker, NASA Langley Research Center 28.

BY James S. Bowman, NASA Langley Research Center V/STOL AND _ STOL AIRCRAFT eg.

By John P. Campbell, NASA Langley Research Center 50.

V/STOL AIRCRAFT OPERATION IN THE TERMINAL AREA ............. e81 By John P. Reeder, NASA Langley Research Center V 31. 287 GROUND EFFECTS ON V/STOL AND STOL AIRCRA_ ..............

By Richard E. Kuhn 3 NASA Langley Research Center OPERATIONAL EXPERIENCES WITH TEE X-14A DEFLECTED-JET 32.

VTOL AIRCRAP_ ..........................

_99 By L. Stewart Rolls, NASA Ames Research Center 33. SOME PERFORMANCE AND HANDLING-QUALITIES CONSIDERATIONS FOR OPERATION OF STOL AIRCRAFT ....................

3o9 . By Seth B. Anderson, Hervey C. Quigley, and Robert C. Innis, NASA Ames Research Center 34: LOW-SPEED FLIGHT CHARACTERISTICS OF A POWERED-LI_ JET TRANSPORT By Robert 0. Schade and Harold L. Crane, NASA Langley Research Center vi

1. RECENT STUDIES OF RUNWAY ROUGHNESS

By Garland J. Morris and Albert W. Hall

NASALangley Research Center

SU_WARY

Recent studies of NASA research related to aircraft operating problems on

rough runways are presented. Someof these investigations were conducted coop-

eratively with the airport operators, with the Federal Aviation Agency, and.

with the U.S. Air Force. The studies showthat criteria based on power spec-

tral levels of runway-profile data are not sufficient to define acceptable

levels of runway roughness from the piloting viewpoint. Because of the large

variation in response characteristics between various types of aircraft, a run-

way may be acceptable for someaircraft and unacceptable for others. A crite-

rion for roughness, therefore, should be expressed in terms of aircraft

response - preferably, cockpit acceleration. A criterion suggested is that

the maximum vertical acceleration in the cockpit should not exceed _+0.4gfor

sections of the runway where precise aircraft control is required.

INTRODUCTION

For several years, the National Aeronautics and SpaceAdministration has

conducted research related to aircraft operating problems on rough runways.

Most of the past research efforts have been concerned with the loads problems.

These problems are of concern during all phases of ground operations - taxiing,

take-off, and landing. Recently, the research efforts have been directed more

to the piloting difficulties resulting from rough runways. These problems have

generally occurred during the high-speed portion of the take-off run a_d, occasionally, during the high-speed portion of the landing.

On a rough runway, the cockpit environment is one of decreased instrument

readability, pronounced pitching or plunging of the aircraft, and excessive

accelerations. Such an environment seriously degrades the pilot's ability to

control the airplane precisely during critical phases of the take-off or

landing.

This paper presents a progress report on recent NASA research pertainlng

to the piloting difficulties caused by rough runways. Someof these investiga-

tions were conducted cooperatively with airport operators, with the Federal

Aviation Agency, and with the U.S. Air Force.

DISCUSSION

RoughnessCriteria Based on Profile Data

A considerable amount of work has been done in past years to define the

characteristics of runway roughness. For this purpose, detailed runway eleva-

tion profiles were obtained for many runways. Short-section profiles, typical

of the profiles that have been measured, are presented in figure 1. The three

sections shown(designated runways A, B, and C) cover a range of roughness from

smooth to rough. Each profile contains, in somewhata randommanner, waves of

different wavelengths with the amplitudes generally becoming larger as the

wavelengths increase. The order of magnitude of runway-profile deviations that

can cause appreclableaircraft response is illustrated by the section from run-

way C, which has caused complaints of roughness. One rather well-defined

depression has a distance of about 100 feet from peak to peak with a profile

depression between peaks slightly greater than 0.1 foot.

Past work indicated that a more concise and convenient representation of

the roughness characteristics could be obtained by converting the elevation-

profile measurementsto power spectral plots as shown in figure 2. The ordi-

nate is the power spectral density, which indicates the relative amplitudes of

the roughness corresponding to the wavelengths shownon the abscissa. The

higher a curve is in the figure, the higher the indicated roughness level.

A study of the power spectral curves for a number of runways led to a pro-

posal for power spectral criteria for runway roughness (ref. 1). The criteria

included two power spectral levels, one for "new construction" and one for

"needs repair." The spectral level for the new-constructlon criterion was

established by the power spectral density of runway B which has caused no com-

plaints of roughness. (See figs. 1 and 2.) It was felt that newly constructed

runways would be satisfactory if their power spectrum fell below the line for

new construction. Similarly, repairs were thought to be necessary if the spec-

trum of an existing runway exceeded the needs-repair criterion. Meeting the

new-construction criterion apparently is not an unreasonable task, inasmuch as

recently constructed runway A is much smoother than the new-construction

criterion.

Shortly after attention was directed toward problems involving piloting

difficulties, it becameapparent that the power spectral criterion was not com-

pletely satisfactory to define an acceptable level of roughness for all oper-

atlng problems (ref. 2). This concluslonwas indicated, in one instance, where

two runways having nearly identical power spectral levels caused complaints of

roughness by pilots in one case and caused no complaints in the other case.

The inability of the power spectra to determine the relative level of

roughness in all cases, particularly from the piloting viewpoint, is not too

surprising when the characteristics of a power spectrum are considered to be as

follows: The runway power-spectral-densityvariatlon with wavelength repre-

sents an average roughness over the entire runway length for various wave-

lengths, does not distinguish between manybumpsof small amplitude and a few

bumpsof large amplitude at a given wavelength, and places no importance on

location of roughness along the runway. Because of the averaging and loss-of-

roughness-location information, the power spectral representation does not have

the ability to detect profile defects which might cause piloting problems.

However 3 as originally intended, the runway spectral level is an indication of

average roughness and for this purpose the spectra are still considered to be

useful in analysis of loads and fatigue problems where a large number of run=

ways are to be considered.

Aircraft ResponseCharacteristics

Shortly after the Jet transports were placed in service, it was found that

runways which had been satisfactory for piston-englne aircraft were causing_ complaints by the Jet pilots. Thus, the pilot's impression of the roughness of a given runway Is dependent upon airplane type. Inasmuch as the pilot's impres- sion of runway roughness comes from the manner in which a particular airplane responds to the runway, it is believed that for piloting problems 3 roughness criteria should be expressed in terms of aircraft response in the cockpit.

Figure 3 shows the spread in response characteristics of various types of aircraft ranging from a light Jet trainer to a heavy Jet transport. Each response curve _as obtained by a power spectral analysis of the cockpit accel- eration measured during a constant speed run along a runway. Each response curve presented in this figure was normalized to its own maximum value; there- fore, these curves serve only to illustrate the difference in the low-frequency response modes for the aircraft types shown.

Several years ago, the available data on airplane response to runway and taxiway roughness indicated that practically all significant aircraft response occurred in a rather narrow frequency range of l_to 2 cycles per second. The aircraft types shown in figure 5 have extended this frequency range. The most significant point to note is the trend toward lower frequencies for the larger aircraft. T--hetrend to low response frequencies has been accompanied b_y an increase in runway speeds. These two factors have increased the range of roughness wavelengths of importance.

The relation between speed, frequency, and roughness wavelength is shoe in figure 4 as the variation of wavelength with speed for several frequencies.

At a given frequency, the wavelength increases with increasing speed and, at a given speed, the wavelength increases with decreasing frequency. The lowest resonant frequency for each aircraft and the highest runway speed (take-off speed) determines the maxlmumwavelength which would be expected to exclte any" significant response for that aircraft. For each aircraft, the shaded area encloses wavelengths capable of exciting aircraft response within the frgquency spectrum of the particular aircraft. For example, the Jet tralnerwould ride over 200- or 300-foot-wavelength bumps wlth little or no response. However, these same bumps would be expected to excite considerable response for a jet transport since these wavelengths are within the area shoe for this airplane.

It is this difference in response characteristics for various aircraft types which accounts for the fact that a runway can be satisfactory for some air- craft but cause trouble for others.

The wavelength, speed, and frequency relations are illustrated in figure 5 by actual measurements which were obtained during an investigation with a medium Jet bomber. The small-amplitude bumps shown on the runway profile have about 100-foot wavelengths and amplitudes between O.1 and 0.2 foot. At 75 knots, the lO0-foot wavelengths result in a frequency of 1.2 cps which corresponds to the lowest resonant frequency for this airplane. (See fig. 6.) At lO0 knots, the lO0-foot wavelengths result in a frequency of 1.7 cps. The large difference in acceleration response measured at the two speeds shows the effect of a small frequency change near the resonant frequency (fig. 6). Thus, small changes in aircraf_t response characteristics, or small changes in operating conditions, can have a large effect on the resulting aircraft response.

Runway Repair The problem facing the airport operators when they receive complaints of roughness from pilots is to determine whether the runway actually needs repair and, if so, where the particular areas are that need repair.

At the present time, it is thought that the most direct and practical method of establishing the repair requirements is to relate the magnitude of the cockpit acceleration to the runway profile and to note those sections of the runway which produce excessive accelerations. This approach is illustrated in figure 7- The lower plot shows the runway elevation for a 4000-foot section of the runway. The upper plot shows the maximum cockpit accelerations recorded at various runway locations and at three speeds during tests of a medium Jet bomber.

In order to determine which areas are in need of repair, it is necessary to establish an acceleration level as a basis for separating the satisfactory from the unsatisfactory sections of the runway. Based on correlations of meas- ured accelerations with pilot complaints during a _umber of investigations, it appears that an acceleration level of _+0.4g is approximately the dividing llne between satisfactory and unsatisfactory runways from the pilot's viewpoint.

Consequently, a maxlmumvertical acceleration in the cockpit of -+0.4g is sug- gested as a criterion for assessing whether a runway needs repair and also as a basis for locating the areas in need of repair. This criterion would be most applicable for portions of the runway where precision of control is most crit- ical, such as during the high-speed portion of the take-off run.

"In figure 73 all five of the runway locations identified by the accelera- tion measurements were considered to be in need of repair. This section of the runway is located near the middle of a military runway with the medium Jet bombers generally lifting off in the region between 6500 and 7000 feet; there- fore, the roughness is critical in this section. This investigation was ini- tiated by complaints of a large pitching oscillation near llft-off causing the airplane to become prematurely airborne. During one of the take-offs made for this investigation, a rather severe pitching oscillation of 2½ v double ampli- tude was measured. The areas that caused this oscillation were the same as those located by the ±O.4g criterion during the constant-speed tests. (See fig. 7 at distances along runway of 550% 6000, and 6600 feet.) Repairs to reduce the acceleration level for these three areas would also be expected to eliminate the pitching problem.

CONCLUDINGR_ This paper has presented a report of recent NASA research related to piloting problems caused by runway roughness.

It has been shown that criteria based on power spectral levels of runwaM_ profile data are not sufficient to define acceptable levels of runway roughness from the piloting viewpolnt.

Because of the large variation in response characteristics between various types of aircraft, a runway may be acceptable for some aircraft and unaccept- able for others. A criterion for roughness, therefore, should be expressed in terms of aircraft response - preferably, cockpit acceleration.

A criterion suggested is that the maximum vertical acceleration in the cockpit should not exceed ±0.4g for sections of the runway where precise air- craft control is required.

REFERENCES i. Milwitzky, Benjamin: Study of Taxiing Problems Associated With _nway Roughness. NASA MEM0 2-21-59L, 1959.

2. Coleman 3 Thomas L.; and Hall, Albert W.: Implications of Recent Investiga- tions on Runway Roughness Criteria. AGAED Rept. 416, Jan. 1963.

MEASURED PROFILES TYPICAL SPECTRA OF RUNWAY ELEVATION RUNWAY ELEVATIONp RUNWAY A FT .I /?UNWAY .,_ _SMOOTH) E"\ I I V'l L I I i .oJ _... _._ COMPLAINTS) I POWER • 2_RUNff,"AY B SPECTRAL "1 t I I I I DENSIT_ FT 2 RAD/FT .0001 NEW __.,/_ ,"_ "S F RUNWAY C CONSTRUCTION _,_ _,_ .00001 _'_ "I , I i I i / 0 I00 200 SO0 400 500 "00000500 I00 50 I0 5 OISTANCE ALONG RUNWAY, FT WAVELENGTH, FT Figure i Figure 2 AIRCRAFT RESPONSE TO RUNWAY ROUGHNESS VARIATION OF WAVELENGTH WITH SPEED FOR VARIOUS FREQUENCIES /-JET TRANSPORT / /-MEDIUM JET BOMBER JET TRANSPORT JET TRAINER FREOUENCY, CPS 400 _75 .8 I.C _ .,_,j i_V/--JET TRAINER NORMALIZED -6 _A_KE/OFF SPEED COCKPIT / I ACCELERATION i / I RESPONSE WAVELENGTH.

i _ / .4 FEET 200 I _2 j / _E-OFF SPEED .2 I II ' , I I00 ._..-----_5 ' l • _ • . , "_'_----'_ ""-7 -_ _---_ - - j 0 50 I00 150 200 0 9. 3 4 5 6 7 SPEEO, KNOTS FREQUENCY, CP$ Figure 4 Figure 3 RESPONSE OF A MEDIUM JET BOMBER TO RUNWAY ROUGHNESS VARIATION OF AIRCRAFT RESPONSE WITH RUNWAY DISTANCE AT TWO TEST SPEEDS I.C MEOIUM JET BOMBE.R .. I00 KNOTS COCKPIT

---I=

.E NORMALIZED ACCELERATION, j , COCKPIT 9 UNITS 'E I I _. 75KNOTS ACCELERATION oF.,v_ A =A I1_. /"k RESPONgE F-.- "" ""t" "-i" v" x .4 .I

RUNWAY _, ,, _/_

ELEVATION, O_ _ / \ I v \ jw_ I

FT ..,I ° ,_, V, , ,_

I 2 3 FREQUENC_ CPS

OISTA.CE ALO._ RUNWAY, _

Figure 6 Figure MAXIMUM RESPONSE CAUSED BY PARTICULAR SECTIONS OF RUNWAY SPEED, KNOTS 0 50 [] 75 0 100 COCKPIT NEEDS REPAIR ACCEL., +,4 / g UNITS SATISFACTORY I I I i I I i I0 I !

I i t RUNWAY I ELEVATION, FEET 2 __ I I I I I J I t 0 5,000 6,000 7,000 8,000 9,000 DISTANCE ALONG RUNWAY, FT Fisure 7 _a TRACTION OF PNEUMATIC TIRES ON WET RUNWAYS By Walter B. Horne and Upshur T. Joyner NASA Langley Research Center SUMMARY Recent work on the traction of pneumatic tires on wet runways is discussed, and it is shoe that a loss of tire traction adversely affects cross-wind landings. The effect of runway surface texture is discussed, and a simple method for measuring surface texture is described. A preliminary correlation of tire traction with surface texture is shown. Results of work at Langley Research Center on the use of air jets to improve tire traction on wet or flooded runways indicate that this is a promising approach for alleviating the large losses in tire braking and sideways traction that occur when tire hydro- planing occurs on a flooded runway.

INTRODUCTION The subject of traction of pneumatic tires on wet runways is important because of its effects on the landing, taxiing, and ground handling of air- planes. For example, when the phenomenon of hydroplaning occurs for an aircraft during landing on a flooded runway, tire-ground traction can drop to negligible values, and the pilot is confronted with the more difficult task of keeping the aircraft within the confines of the runway. From figure 1 the results can be seen of a case in which a cross wind caused an airplane to drift off the center of the runway, as shown by the light streaks proceeding from the foreground to the edge of the runway; in this instance, the airplane went off the side of the runway with a number of thousands of feet of runway left for stopping purposes.

The white streaks are caused by high-velocity water, in the tire footprint areas, that scours the runway surface and cleans it of previously deposited , dirt and other contaminants. This type of white streak has been noted for some years and was identified with water action in the footprint region at the Langley landing-loads track several years ago. It should be pointed out that for the incident referred to in figure l, the pilot touched down near the runway center line. He did try the usual procedures but could not control the right- ward drift of the aircraft. Figure 2 indicates why the pilot could not control the direction of the aircraft. At the top of the figure is shoe an airplane making a cross-%Nlnd landing on a flooded runway with the side wind from the left, as indicated at the right side of the figure. The airplane touches doe at a speed above the hydroplaning speed, and tire-ground traction is practically nil. Under this condition of very low traction, the airplane reacts like a sailboat without a keel, tends to weathercock into the wind, and drifts off the right side of the runway because of the cross wind. When an effort is made to slow the aircraft by use of reverse thrust_ the vector diagram shown in figure 2 has to be considered. In this case the side component of reverse thrust is in a direction which tends to augment the side force of the wind and to increase the airplane drift to the right side of the runway.

A cross-wlnd take-off on a flooded runway is not so severe a problem as a cross-wind landing. (See fig. 2.) The aircraft starts from rest with good tlre-ground traction. The traction decreases as speed increases until hydro- planing speed is reached. At this point, traction again is nil and, if the pilot maintains heading as shown by the middle sketch, the cross wind will tend to drift the airplane off the right side of the runway. Drifting during take- off is normally controllable, as shown in the lowest sketch of the figure; the pilot simply yaws the airplane into the wind so that the side component of for- ward thrust, as shown by the lower vector diagram, cancels the cross-wind force, andlthe airplane then travels along the center line of the runway during take -off • This example of the consequences of traction loss points out the need for some method or methods to improve tire traction on wet runways. Recent research on traction at Langley Research Center is so aimed, and several possi- bilities in this area are discussed in this paper.

SYMBOLS calculated hydroplaning speed, knots Vp average friction coefficient WAV locked-wheel friction coefficient WSKID RESULTS ANDDISCUSSION Effect of Runway Surface Texture on Tire Traction Loss of traction is a phenomenon which occurs both in the sideways and braking directions nearly simultaneously and has been studied for several years by investigators to determine the best tire tread for traction on a wet runway.

It has become increasingly obvious to investigators, however, that the runway surface itself is an important factor in determining the wet-runway traction which can be developedby any tire. (See refs. 1 to 3.) In a series of tests, which NASA is now conducting at the Langley landing-loads track in collabora- tionwith the U.S. Air Force to develop a standard groove width and depth for Air Force tires, several different runway surfaces have been installed, all in a single llne, so that every braking test is conducted over each of these surfaces and so that a braking cycle is initiated on each surface. The test- surface arrangement at the track is as follows: The first test surface, sur- face A_ is a steel-troweled concrete surface lO0 feet long that provided a very smooth surface finish similar to a table top. The second, surface B, is a lO

200-foot concrete section that was given a broomed finish to provide a moder-

ately textured surface. The third surface, surface C, is a 300-foot-long sec-

tion of small-aggregate asphalt which provides a medium-textured surface.

Fourth, surface D, is a 300-foot-long section of large-aggregate asphalt that

provides a very coarse-textured surface. Finally, a 200-foot-long refrigerated

section was installed in the track; this section can be flooded and frozen so

that wet ice can be obtained.

Figure 3 showsthe braking friction results obtained on a smooth-tread

32 × 8.8, type VII, aircraft tire during braking runs performed on these test

surfaces when in a water-flooded condition (water depth = 0.1 to 0.2 inch).

The data shownwere obtained for a vertical load of 22 000 pounds and an infla-

tion pressure, acting on the tire, of 290 pounds per square inch. It should_be

pointed out that only the two concrete and the two asphalt surfaces were

flooded; the ice surface during the tests was barely covered by a thin water

film. It can be seen in figure 3 that as ground speed increases, the average

friction coefficients on all surfaces tend toward minimal values at the hydro-

planing speed indicated at 1_4 knots; this is the classic type of traction loss

due to hydroplaning from fluid-denslty effects. At the lower speeds, however,

for which fluld-density effects are negligible, the surface texture becomesall

important. As can be seen from this figure, a surface must be textured in order

to break through the viscous film separating the tire from the pavement even at

speeds as low as 20 knots. In an effort to study the surface texture on a

quantitative basis rather than resort to definition of the surfaces in terms of

words, a method has been devised at the Langley landing-loads track in which a

given quantity of material is worked into the runway voids to determine the

average depth of texture; a known volume of stiff grease is worked into the

runway voids, with a rubber squeegee, within an area of run,ray between two

lines of masking tape. (See fig. 4.) (The rubber in the squeegee is of about

the hardness of tire-tread material. ) After the grease is worked into the run-

way voids over a certain length, between the lines, the area that has been

covered is determined by simply measuring the distance along the lines, which

are a known distance apart. Dividing thls area into the known volume of grease

gives an average depth of texture s which is given in millimeters. Admittedly

this is not an exact basis for determining runway texture, but it is a simple

method and showssome promise. Work along this line has been done by Kummer

and Meyer (ref. l) at the Pennsylvania State University and is being contem-

plated by R. W. Sugg at the Ministry of Aviation in England. The correlation

obtained by using the grease technique for determining runway-surface average

texture depth with wet-braking traction is shoe in figure _. Here, average

friction coefficients obtained on the four flooded surfaces are plotted against

average depth of texture (as determined by the grease technique) for velocities

of 30, 60, and 90 knots. It can be seen that the friction coefficient on the

wet surface coefficient improves at all velocities as the texture is increased

in roughness. Above a certain texture depth, however, as from surface C to sur-

face D, there seemsto be little effect on friction coefficient of increasing

harshness of the texture. This method of measuring runway texture and corre-

lating it with friction coefficient will be pursued at Langley Research Center

by using other surfaces with different tires. Also, it is planned to make

measurements of various runways and roads, both new and used, to determine if,

in fact, this simple method of measuring runway texture can be utilized to

ll

infer from the texture alone, along with previously correlated data between

friction coefficient and texture, whether a given runway-texture depth is below

somevalue which would be considered satisfactory for wet-runway operations.

Use of Air Jets to Improve Tire Traction on Wet Runways

Air-jet research to improve tire traction was first performed at the

Langley Research Center by E. N. Harrin in 1958. The results of this initial

work, performed on a small wheel and belt arrangement_ are reported in refer-

ence 4. Air-jet research at Langley was resumed at the l_nding-loads track

during the summer of 1964 , and someof the results obtained during this inves-

tigation are described.

Figure 6 showsa view of the test fixture and air Jets used in the current

test program. The test tire was a regular 6.50-13 automobile tire inflated to

a pressure of 27 pounds per square inch. Both a smooth-tread and a 4-groove

ribbed-tread tire were tested. Essentially, the air-jet arrangement, as shown

here, was a tandem-nozzle arrangement with the trailing nozzle located about

7_ inches behind the front nozzle and about l0 inches in front of the tire. The

tire in this figure is resting on a glass plate which was located flush with the

concrete surface in the test runway. From beneath this glass plate, pictures

of the tire footprint were taken as the tire traveled across the plate. The

water over the glass plate was colored with a green sea-marker dye to give

better picture contrast. These pictures showedthe very beneficial effect which

can be obtained by completely clearing water from the footprint at speeds well

above the hydroplaning speed. (For example, see fig. 7.) The left portion of

figure 7 showsthe tire in a completely hydroplaning condition when traveling

at a speed of 62.7 mph. The right portion of figure 7 showsthe good contact

of the tire with the runway when the tire is traveling at a speed of 100.9 mph; this beneficial effect was obtained with the use of air jets. In this case,

hydroplaning has been alleviated. In addition, quantitative measurementswere

madeof friction coefficient and hydrodynamic pressure developed between the

tire and test runway surfaces.

The surface of the test runway shown schematically in figure 8 was very

smooth, except for a 52-foot section in the middle which was sandblasted in

order to have a surface texture somewhatmore representative of highways and

runways in use today. For example, the beginning of the sandblasted concrete

surface had an average texture depth of 0.104 millimeter. This value is about

half the texture depth of the broom-finished concrete surface described earlier.

Figure 9 presents values of locked-wheel friction coefficient and hydro-

dynamic pressure plotted against runway distance for one run. These values

were obtained at a carriage speed of 90 knots, almost twice the hydroplaning

speed of 47knots, with the ribbed-tread tire and a runway water depth of

0.3 inch. The runway surface condition is as noted in figure 8; that is, there

were three sections, a smooth concrete section, a sandblasted concrete portion_

and another smooth concrete section. With the air Jet off, it can be seen that

hydrodynamic pressures above 40 pounds per square inch were developed on the

tire footprint and that very low values of _SKID were obtained. But with the

air jet on, there is a reduction in hydrodynamic pressure to near zero on the

tire surface. Note also the great improvement of values of _SKID3especially

on the textured concrete section where the friction coefficient is greater than

0.4. The gradual decrease in WSKID shownhere as the tire traveled over the

sandblasted runway section is due to a nonuniform texture achieved in using the

sandblasting technique.

Figure i0 presents wet-runway cornering-force data as a percent of dry-

runway values plotted against the samerunway distance as in figure 9. The yaw

angle was 5° , the water depth was 0.3 inch_ and the speed was 77 knots, which

was once again greater than the critical hydroplaning speed. Note that with

the air jet off, less than 5 percent of the dry-runway cornering force is

achieved at this water depth; but, with the air jet on, more than 50 percent of

the smooth-tire dry-runway cornering force is obtained in the textured concrete

section, Figure ll summarizesresults, such as those illustrated in figure 9,

obtained with the locked wheel_ the 4-groove ribbed-tread tire, and 0.3 inch of

runway surface water. Here, the values of _SKID of both smooth and textured

runway surfaces are plotted against forward speed in knots. It can be seen that

on the smooth concrete surface there is not muchdifference between the curves

obtained with the air jet off and the air jet on because of the inability of

the air jet to remove the very thin fluid film which adheres to the smooth sur-

face. On the roughened or textured surface_ however, the many and minute sur-

face irregularities puncture this thin surface film_ and the result on friction

coefficient is shownby the curves obtained on the sandblasted concrete. On

this textured runway surface, which is more representative of actual runway

surfaces in use today, much greater values of WSKID are obtained with the air

jet on than with the air jet off.

CONCLUDING R_RKS

Someof the recent work on the traction of pneumatic tires on wet runways

has been discussed and ways in which the loss of tire traction adversely affects

aircraft cross-wind performance have been shown. Runwaysurface texture plays a

most important role in determining the slipperiness of a pavement when wet. A

simple method for determining the average depth of texture of pavements has

been described, and measurementsobtained on four test runway surfaces at the

Langley landing-loads track by using this grease technique give a promising

correlation with tire traction on a wet runway. Preliminary work at Langley

Research Center on the use of air jets to improve traction on wet or flooded

runways has been described and showspromise as a means of alleviating the

large losses in tire braking and sideways traction that develop when hydro-

planing occurs on flooded pavements.

REFERENCES i. K ummer, H. W.; and Meyer, W. E. : Rubber and Tire Friction. Eng. Res. Bull.

B-80, Pennsylvania State Univ., Dec. 1960.

2. Joyner, Upshur T.} Home, Walter B.; and Leland, Trafford J. W. : Investiga- tions on the Ground Performance of Aircraft Relating to Wet Runway Braking and Slush Drag. AGARD Rept. 429, Jan. 1963.

3. Gray, W. E.: Aquaplaning on Runways. J. Roy. Aeron. Soc. (Tech. Notes), VO1. 67, no. 629, May 1%3, pp. 302-304.

4. Karrin, Ezlaslav N. : Investigation of Tandem-Wheel and Air-Jet Arrangements for Improving Braking Friction on Wet Surfaces. NASA TN D-409, 1960.

AIRCRAFT ACCIDENT DUE TO H

YDROPLANING CROSS-WINO LANDINO EFFECT OF CROSS WIND ON LANDING AND TAKE-OFF ON FLOODED RUNWAYS LANDING (TOUCHDOWN ABOVE HYDROPLANING SPEED) WIND WIND

REVERSE THRUST +A

J

Figure 1 L-2411-2 Figure 2

METHOD OF MEASURING RUNWAY SURFACE TEXTURE EFFECT OF SPEED ON BRAKING TRACTION WATER DEPTH=OI TO 0 2 IN W W A Y s"RmcE HYDROPLANING AVERAGE FRICTION COEFFICIENT,

4 - kB

I 'AV I , , 2 -ICE '\ I

--:=----=--

I I ? - - , 1 1 , 0 2 0 40 60 80 100 120 140 160 GROUND SPEED, KNOTS

Figure 3

ARRANGEMENT OF AIR JETS EFFECT OF SURFACE TEXTURE ON AVERAGE

AIRFLOW I 2 7 LWSEC; NOZZLE PRESSURE 3 390 LEIIN *

FRICTION COEFFICIENT GROUND SPEED, D KNOTS C 3 0 AVERAGE FRICTION €0 COEFFICIEN ‘ n v

Figure 6

Figure 5

PATH-CLEARING EFFECTIVENESS OF TANDEM AIR JETS

WATER DEPTH = 0.3 IN.

W I T H O U T AIR JETS WITH AIR JETS (GROUND SPEED = 62.7 MPH’) (GROUND SPEED = 100.9 MPH)

Figure 7 L- 65 -110

EFFECT OF AIR JETS ON BRAKING TRACTION WATER DEPTH =0:5 IN ; GROUND SPEED = 90 KNOTS TEXTURED SMOOTH CONCRETE SMOOTH CONCRETE : I I- CONCRETE SCHEMATIC OF LEVEL RUNWAY SURFACES FRICTION COEFFICIENT, _ " LOCKED-WHEEL SKID "6I A 2 _ 0 _- .... _ ....... I -- AIR JET ON AVERAGE TEXTURE DEPTH, MM ---AIR JET OFF ( _ "_ .o62 .03;" i"_ ,"- ...........

SMOOTH . TEXTUR_ SMOOTH PRESSURE, 2 LB/SQ IN ,¢_.,_-._;]_ -_-CONCRETE RUNWAY ,J/_,_- _L,._

i 'o

0 40 80 120 RUNWAY DISTANCE, FT Figure 8 Figure 9 EFFECT OF GROUND SPEED ON AIR-JET PERFORMANCE IN IMPROVING BRAKING TRACTION EFFECT OF AIR JETS ON CORNERING FORCE WATER DEPTH=0.3 IN WATER DEPTH=0.5 IN; GROUND SPEED = 77 KNOTS TEXTURED CONCRETE I00 ,6[A T ON PERCENT OF _r LOCKED-WHEEL .2 IR JET OFFJ"t:_. ______ __ _- FRICTION COEFFICIENT, DRY CORNERING _" FORCE AT 5 ° I I ! I I

c OOE% o%T# E ----

YAW ANGLE '_SK ID 0 SMOOTH CONCRETE 4C "3 F _/-- AIR JET _X._JET ON .2: I- -. 0 _AIR JET OFF • I IAIR JE ---- I I J

o 2b

0 30 60 90 120 40 60 80 I00 120 140 160 RUNWAY DISTANCE, FT GROUND SPEED, KNOTS Figure lO Figure ll

i?

o SUMMARY OF ATMOSPHERIC TURBULENCE DATA By Roy Steiner NASA Langley Research Center SUMMARY Turbulence data obtained from recent flights of the U-2 airplanes and from the National Severe Storms Project agree in general with the results obtained in previous investigations. The data show that nonstorm turbulence decreases in intensity and amount with increasing altitude. Less than 2 percent of the flight distance was in turbulence at altitudes above 503000 feet. It also appears that the root-mean-square gust velocities of the composite turbulence patches may be 15 to 20 percent lower than previous estimates.

INTRODUCTION In addition to the regular collection of turbulence data from routine air- line operations 3 the NASA collects turbulence information through the use of special flight programs. From long experience in the study of the loads on air- craft in turbulence, it has been found that the airplane responses may be ade- quately defined if the turbulence is represented by two types: nonstormand storm. It has become customary, therefore 3 to classify special flight programs in a similar manner. In this paper, two specialized sampling programs will be discussed. The first concerns nonstorm data obtained with the U-2 airplane at altitudes up to 75,000 feet; the second concerns measurements made in severe storms in Southwestern United States at altitudes up to 403000 feet.

SYMBOLS A airplane response parameter relating rms input and output values an normal acceleration, g units L scale of turbulence, ft N average number of peaks per unit distance which exceed given level of response average number of zero crossings per unit distance with positive slope percent of total flight distance spent in turbulence Wg vertical component of true gust velocity, ft/sec x general response quantity _c composite rms value of gust velocity Sub s c ript s : 1 nonstorm 2 storm SAMPLE SIZES The scope of the nonstorm turbulence data obtained from the U-2 airplane (ref. l) and the B-36 airplane is given in table l(a). The scope of the storm turbulence data (ref. 2) is given in table l(b).

TABLE 1 SCOPE OF TURBULENCE MEASUREMENTS (a) Nonstorm Flight miles Altitude ft To 1953 To 1960 To 1965

(B-36) (u-2) (u-2)

18,900 20,000 to 30,000 66,500 11,800 30,000 to 40,000 24,500 16,200 26,500 40,000 to 50,000 21,500 38,100 49,500 141,700 50,000 to 60,000 105,500 136,400 576,900 60,000 to 70,000 7,000 7,000 70,000 to 75,000 112,500 315,000 Total ..... 820,500 (b) Storm Flight miles Altitude, ft To 1953 To 1965 20,000 to 25,000 2,740 2,950 25,000 to 30,000 2,180 2,400 50 410 30,000 to 35,000 35,000 to 40,000 56O Total .....

4,970 6,320 2O

By 1953, 112,500 miles of flight samples had been obtained with the B-36

airplane at altitudes up to 50,000 feet. Flight operations with the U-2 air-

plane started in 1956, and 315,000 flight miles had been accumulated by 1960.

Since 1960, the high-altitude U-2 sample has more than doubled in size and now

covers 8203500 miles. More than 700,000 miles, or 85 percent, of this distance

is above 50,000 feet. Emphasis is being placed on the higher altitude programs,

since they cover the expected cruising altitudes of the proposed supersonic

transport.

The storm turbulence data obtained by 1953 covered about 5,000 flight

miles, with only 50 miles above altitudes of 30,000 feet. The more recent sam-

ples have increased the sample size to more than 6,000 miles, with approximately

1,000 miles above 30,000 feet.

BASICGUST RELATION

In order to establish a framework for discussing the results of the flight

measurements, it might be well to consider a general relation between the gust

characteristics and the airplane response. (See ref. 3-) This relation, based

on power spectral techniques, is

N = PIN0 e + P2N0 e

This general equation is based on the concept of the response to nonstorm and

storm turbulences as given by the first and second terms, respectively.

Together, the two terms represent the total experience in atmospheric turbu-

lence - the experience for both repeated and limit loads.

The relation is applicable to a given flight condition and yields the

average number of peaks N which exceed a given level of airplane response x.

The given number of peaks is a function of the percent of total flight distance

in nonstorm and storm turbulences (P1 and P2), the corresponding root-mean-

square gust velocities (ac, I and gc,2), and the airplane response parameters

(A and NO).

The center-of-gravity acceleration is a commonlymeasuredresponse param-

eter. Since the derived gust velocity is proportional to the center-of-gravity

acceleration, it may be considered as an airplane response function as well as

a description of the intensity of the atmospheric turbulence. In the following

sections of the paper, the derived gust velocities (a response, x) and the

percent of turbulence for the nonstorm condition will be considered. Also, some

information on the derived gust velocities for the storm condition and an eval-

uation of the root-mean-square gust velocities for the two turbulence conditions

will be considered.

NONSTORMTURIKrLENCE

Intensity of NonstormTurbulence

In figure 13 the distributions of derived gust velocities are presented

for flight of U-2 airplanes at altitudes above 203000 feet. These data are for

routine operations. Routine operations - in contrast to flights that seek out

turbulent conditions - normally use turbulence avoidance procedures to reduce

the flight distance in turbulence. The gust frequency per mile of flight is

plotted against the values of gust velocity. The five distributions correspond

to 103000-foot-altitude brackets from 20,000 to 703000 feet. The larger sam-

p1_ represent approximately 142,000 and 9.77.,000 flight miles for the two high- est altitude brackets.

The curves indicate that the frequency and velocity of the gusts generally decrease in an orderly manner with increasing altitude. The distributions for the two lower altitude brackets are nearly coincident. This correspondence in the turbulence at these altitudes is probably due to flights through the jet stream and the tropopause3 which may vary from 29,000 feet to possibly 403000 feet during the different seasons in the Northern Hemisphere.

In terms of the intensity of the turbulence, the results indicate that the maximum gust velocity experienced per mile of flight for higher altitude brack- ets is on the order of 1/3 or 1/2 that experienced per mile of flight for lower altitudes. The gust velocities 3 however3 are quite low; for example 3 in 13000 miles at an altitude of 903000 feet, a gust velocity of only 4 ft/sec will be exceeded, as compared with 8 ft/sec at the lower altitudes. On the basis of airline data, a comparable value for an altitude of 193000 feet would be 15 ft/sec. Current design values for airplanes are 90 and 66 ft/sec on this scale and would plot far to the right of the figure.

Eighty percent of the total flight distance represented in figure 1 was over the United States. The remaining 20 percent consisted of flights over Western Europe 3 Turkey, and Japan. The sample from Japan was not included in the altitude bracket of 503000 to 603000 feet3 because of two unique patches of turbulence. Figure 2 shows the effect on the distribution of gust veloc- ities if the Japanese sample is included. The distributions for the altitude brackets from 503000 to 60,000 feet and from 203000 to 303000 feet have been included for comparison. The addition of the Japanese sample has shifted the frequency curve upward so that it is more nearly the same as that for the lower altitudes. It is believed that the interaction of moderately severe surface storms 3 the mountainous terrain, and strong Jet streams at an altitude of 353000 feet caused the unique patches of turbulence (ref. 1). The frequency of this unusual combination of conditions, probably over a rather localized area3 and the method of integrating these data into the data for average conditions have not been determined.

Amount of NonstormTurbulence

The percent of the flight distance in nonstorm turbulence is presented in

figure 3 as a function of altitude. The airplane was considered to be in turbu-

lence whenever the accelerometer trace was disturbed and exhibited gust veloc-

ities greater than 2 ft/sec. The solid llne represents a previous estimate

(ref. 4) based on earlier data. The circled data points represent the present

U-2 data. For altitudes between 50,000 and 75,000 feet, the U-2 airplane

encountered turbulence less than 2 percent of the flight distance.

All sets of U-2 data show a peak in percent of turbulence at altitudes

near 35,000 feet. The points for the present sample show a greater percentage

of turbulence than given by previous estimates. Tae prevalence of turbulence

at this altitude is probably due to the high winds and wind shears associated

with Jet streams. However, a note of warning is needed. The flights in the

altitude range from 20,000 feet to the cruise altitude of the U-2 airplane

represent, to someextent, a vertical sounding during climb and descent over

rather localized areas, and the data may not indicate average conditions at

these altitudes over the Northern Hemisphere.

COMPARISON OF DATA WITHPREVIOUS ESTIMATES

Figure 4 summarizes the nonstorm data previously discussed, the data from

the National Severe Storms Project, and earlier estimates for nonstorm and

storm conditions 3 which are shownas the shaded areas for altitudes of 20,000

to 603000 feet. The gust frequency per mile of turbulence is plotted against

the gust velocity. Again, a logarithmic scale is used on the ordinate. The

range of the nonstormU-2 data is shownfor the five altitude brackets ranging

from 203000 to 70,000 feet. Note that a much compressedgust-velocity scale

is used on the abscissa in order to cover both storm and nonstorm gust inten-

sities. The turbulence data from the U-2 airplane flights are generally less

severe than previously obtained data, but not to such an extent as to warrant

a revision of the earlier estimates.

The solid square symbols represent the data for flight at an altitude of

approximately 403000 feet in storms. First impressions might lead to the con-

clusion that the earlier estimates are low. However, the storms in Southwestern

United States are considered to be the most severe storms in the States. Of

these storms, only the ones with sufficient energy to build to altitudes of at

least 403000 feet were investigated. Therefore, the severe-storm data would be

expected to lle above the previously estimated curves. There is not sufficient

Justification, therefore, for revising the estimate at this time.

SPECTRAL REPRESENTATION OFATMOSPKERIC _CE

In recent years considerable effort has been expended in describing atmos-

pheric turbulence in terms of power spectra and in applying power-spectral

techniques to design problems. Typical spectra are presented in figure 5 for a severe-storm condition and for a nonstorm condition. The power spectral density is plotted against reduced frequency, which is given in radians per foot, and against wavelength which is given in feet. Each scale used in this plot is logarithmic. Each spectrum represents a given turbulent area. The spectra are similar in shape but vary in intensity. The square root of the area under the spectrum is the root-mean-square gust velocity given in the figure.

Note that these rms gust velocities vary from about 14 to 4.5 ft/sec for the different meteorological conditions. The spectra for many turbulent patches in each condition would also cover a wide range. For example, in ii storm trav- erses, the rms values varied from 6 to 16 ft/sec. A similar range of rms val- ues would be expected for the nonstorm conditions. The values which describe this vs_r_atlon in the spectra for the nonstorm and storm turbulences are shown in figure 6 as a function of altitude. Earlier estimates of the rms gust veloc- ities of storm and nonstorm turbulences are shown by the solid curves (ref. 4).

These velocities were determined from airplane center-of-gravity response data by use of a rough approximation of the airplane response parameters. Only one degree of freedom was considered in the evaluation. More precise values of the airplane parameters have become available for a few types of airplanes and flight conditions (obtained from Lockheed-California Co.). The reevaluation of a few rms values by use of these revised airplane parameters indicates that the rms true gust velocities for the nonstorm and storm turbulences as shown by the dashed lines may range from 15 to 20 percent lower than the previous estimates.

For consistency, the reevaluation was based on the assumption that the scale of turbulence L is 1,000 feet. Should the scale value be larger, say 2,500 feet, as more recent investigations have indicated, both of the sets of curves would be shifted to the right.

MEASURED AIRPLANE MOTIONS DUE TO SEVERE TURBULENCE The physical behavior of an airplane during flights through two storms is shown in figures 7 and 8. The vertical component of true gust velocity, the normal acceleration 3 and the altitude change of the airplane during flight are given in the figures. These traverses are flights of a fighter-type airplane.

The pilot was instructed to correct only for the major changes in attitude.

Figure 7 indicates that the airplane encountered an updraft with a long gradient distance. The gust velocity changed from 0 to 170 ft/sec in 2,100 feet. The maximum veloeity is approximately 210 ft/sec. Although the velocities are large, the airplane experienced only small acceleration. An altitude change of 1,600 feet, however, was experienced in 3 nautical miles.

At the time the maximum gust velocity was encountered, the pilot had pitched the airplane down 13 ° in an effort to maintain the desired altitude. The ver- tical velocity of the airplane at this time was 80 ft/sec up.

Figure 8 indicates that the airplane was flying in an updraft which reached 70 ft/sec, and then suddenly encountered a downdraft of 90 ft/sec. This change in velocity of 160 ft/sec occurred over a horizontal distance of about 800 feet in contrast to the 2,100 feet shown in figure 7- The airplane normal accelera- tion changed from about 0.Sg to -2g. The change in altitude in this rather severe turbulence was only about 100 feet.

The pilot may believe that he is experiencing large altitude changes in turbulence of this nature. This impression would probably be caused by both the large accelerations and possible instrument response due to the changing angle of airflow over the static source. The large change in altitude illus- trated in figure 7 maY be undetected for a longer period because of the more quiescent flight conditions.

CONCLUDING_ In summary, turbulence data obtainedfrom recent flights agree in general with the results obtained during previous investigations. Nonstorm turbulence decreases both in intensity and amount with increasing altitude. Less than 2 percent of the flight distance was in turbulence at altitudes above 50,000 feet. It now appears that the root-mean-square gust velocities of the composite turbulence patches may be 15 to 20 percent lower than previous publications.

REFERENCES i. Coleman_ Thomas L._ and Steiner, Roy: Atmospheric Turbulence Measurements Obtained From Airplane Operations at Altitudes Between 2%000 and 75,000 Feet for Several Areas in the Northern Hemisphere. NASA TND-548, 1960.

2. Rhyne, Richard H.; and Steiner, Roy: Power Spectral Measurement of Atmos- pheric Turbulence in Severe Storms and Cumulus Clouds. NASA TN D-24693 1964.

3. Houbolt, John C.; Steiner, Roy; and Pratt, Kermit G.: Dynamic Response of Airplanes to Atmospheric Turbulence Including Flight Data on Input and Response. NASA TR R-199, 1964.

4. Press, Harry_ and Steiner, Roy: An Approach to the Problem of Estimating Severe and Repeated Gust Loads for Missile Operations. NACATN 4332, 1958.

INTENSITY OF NONSTORM TURBULENCE INCLUDNG SAMPLE FROM JAPAN INTENSITY OF NONSTORM TURBULENCE i0-1 FLIGHT MILES i0 -I _ ALTITUDE, FT ALTITUDE, FT 18,900 _, _ 20,000 TO 30,000 20,000 TO 30,000 26,500 t:_ --[]--50,000 TO 40,00O

49,500 -_-- #:__ _:_°(,NC,U_NO s_

0 _ ---0- 4O, OOO TO 5O, OOO t41,700 \ "% --_-- 50,00o TO 60,000 !0 -2 576,900 FROM JAPAN) .......... '0-2 _ _ --_-- 60,000 TO 70,000 FREQUENCY PER MILE OF FLIGHT OF FLIGHT 10-3 10-3 0-4 g MR b'l-L,

'°-' t

10-5 4 8 12 16 25 24 O 2,; DERIVED GUST VELOCITY, FT/SEC OER,VEO GUST VELOC,TY, FT,'SEC Figure I Figure 2 NONSTORM AND STORM TURBULENCE PERCENT NONSTORM TURBULENCE

,°'f

80 - X I03 -- PREVIOUS ESTIMATE IoO I _ _i:::_,':_4:i:!:_: U-2 DATA l 1 NSSP DATA o U-2 DATA "tO- O PER MILE OF TURBULENCE 10-2 ALTITUDE, FT 5( 4( 20000 60 OOO FT _.

Io " AZI:,TUDE S_O._ :. NONSTORM lO-50----I0 "2'0 30- 4'0 .50 6(3 2C DERIVED GUST VELOCITY, FT/SEC PERCENT OF FLIGHT DISTANCE IN TURBULENCE Figure Figure 3 INTENSITY OF TURBULENCE WITH ALTITUDE TYPICAL SPECTRA OF TURBULENCE L = 1,000 FT rms GUST VELOCITY, 60 -x 103 FT/SEC / / 1.3.77 104 I_ ['15% TO 20%'] / / STORM 50 '16_--1 REDUCTION ]'_/ / 4.48 POWER 103 ALTITUDE, FT SPECTRAL DENSITY, NON STORM 3O (FT/SEC) 2 ,I RADIANS/FT i02

iI I/

IO I I , I , I J I io / 104 I03 I02 I0 I WAVELENGTH, FT ',7 , / /' ' 4 4 6 8 I0 12 rrns GUST VELOCITY, o- c, FT/SEC REDUCED FREOUENCY. RAD,A.S,FT Figure 6 Figure THUNDERSTORM TURBULENCE DATA THUNDERSTORM TURBULENCE DATA THUNDERSTORM A THUNDERSTORM B TRUE GUST 8C TRUE GUST 80, j 4_¸ _'J'_'_ i VELOCITY, Wg, VELOCITY, wg, FT/SEC FT/SEC O 0 -80 -80 f',, NORMAL ACCEL., 0 NORMAL ACCEL, 0 -,-' '," _' I _" '_ _ _ "_' _ V an, g UNITS a n , g UNITS -2 -2 1,600 1,600[

F

ALTITUDE CHANGE, FT ALTITUDEFTCHANGE, 800L_ 800ri- ,,r_-----t I-- + . L i i o .s ,.o ,.; 21o 215 3b _'_ O 5 1.0 [5 2.0 _._ _O _,S DISTANCE, NAUTICAL MILES °,STANCE, NAUT,CAL M,LES Figure 7 Figure 8 .

PRELIM_ARY STUDY OF STEEP INSTRUMENT APPROACH OF THREE CONVENTIONAL AIRCRAFT By Albert W. Hall, Robert A. Champine, and Donald J. McGinley, Jr.

NASA Langley Research Center SUMMARY Preliminary studies of steep instrument approaches have been conducted with three conventional aircraft. Six-degree approach paths were found to be accept- able. A margin of 3 ° to 4 ° in maximum glide slope was available to allow for flight-path corrections. Control of the lateral-directional axis was the most difficult problem for pilots making either steep or conventional approaches; split-axis control reduced the pilot workload during this operation.

INTRODUCTION The NASA Langley Research Center has recently completed a preliminary study on several aircraft to determine the capability of both aircraft and pilots to make steeper-than-normal instrument approaches. Steep approaches are attractive in that they offer the possibilities of reduced engine noise and reduced air- space requirements.

This paper reports the initial phase of a continuing program and covers studies made with C-47, T-33 (refs. 1 and 2), and DC-8F aircraft. The tests were conducted by making instrument approaches at increasing glide slopes until the maximum glide slope considered suitable for day-to-day operations was estab- lished on the basis of aircraft characteristics and pilot opinion. The scope of the initial studies is indicated in table I which shows for each of the aircraft the total number of approaches, the number of approaches at the maximum opera- tional glide slope, and the number of pilots participating.

SYMBOLS V a approach speed, knots W weight of airplane, lb glide-path angle, deg increment in flight-path angle, deg TEST PROCEDURE AND EQUIPMENT The approaches were flown with the evaluation pilot flying on instruments (simulated IFR, pilot under hood) while a safety pilot maintained visual contact and took control whenever conditions required a termination of the test. The instrument approach started in level flight about 5 to 8 miles from the runway threshold as shown in figure i. The glide slope was intercepted about 4 miles from the runway for all tests. This approach distance was fixed by varying the initial altitude for each glide angle; for example, the 2.5 ° tests were started at an altitude of 900 feet and the 6 ° tests were started at an altitude of 2200 feet.

Special guidance equipment was used which allowed the glide slope to be adjusted to the desired angle by a simple adjustment. This equipment provided straight glide slopes (no bends) with conventional ILS sensitivity (±0.7o). In addition, this equipment could provide curved paths to give flare-path guidance from any desired altitude to touchdown.

The flight-path (glide-slope, flare, and localizer) information was dis- played in the cockpit on the conventional ILS cross-pointer indicator for the C-47 and T-33 aircraft and on a flight director for the DC-8F. The DC-8F was fitted with a split-axis autopilot with which approaches could be made with the lateral-directional axis under autopilot control and with the glide slope under manual control.

DISCUSSION Maximum Operational Glide Slope All approaches were flown at the approach speeds recommended for conven- tional angles except the 9 ° approach for the T-33.

For a constant-speed approach with the airplane in the maximum-drag config- uration, the maximum glide angle occurs with the lowest obtainable thrust_ A qualitative representation of the thrust-required curves is shown in figure 2 for the C-47. As shown in figure 2, the maximum angle obtained with the C-47 was I0 ° with throttles closed and the propellers windmilling. The i0 ° angle is not usable for a constant-speed instrument approach because the pilot cannot get the airplane back to the glide slope from an inadvertent high posi- tion without diving and gaining speed.

For the C-47 airplane, a 6 ° glide slope was determined to be the maximum glide slope suitable for day-to-day use in an instrument approach. At this glide slope, by reducing power, the pilot has a 4° margin available to steepen the flight path without gaining speed. The pilots believed that this margin would be sufficient to take care of gusty conditions, windshear, and other inad- vertent flight-path deviations.

3O

The thrust-required curves for the T-33 are shown in figure 3. Maximum

angles of 9° were flown with the approach speed increased about l0 knots and

with the engines about 57 percent of design rpm. Here again, the pilots picked

6° as the maximum angle suitable for operational use. At the 6° angle, the

engine speed was about 65 percent of design rpm. Because of the poor acceler-

ation characteristics of this engine, the pilots did not want to operate con-

tinuously below this rpm. However, power could be reduced below this level for

short intervals, as needed for flight-path control, so that there was a capa-

bility of steepening the flight-path angle as much as 3° with only a small

increase in aircraft speed.

A maximum operational glide slope for the DC-8F transport was also found to

be 6°. At this angle, the engine speed was about 75 percent of design rpm.

This airplane had turbofan engines and the pilots did not like to reduce the

engine speed below this value_ in fact, they tended to accept a slight airspeed

increase during flight-path corrections rather than reduce power. It should be

pointed out here (as is described subsequently) that the flare path started

at an altitude of 260 feet and, from this point, the pilot had to add power so

that the engine speed was increased as the airplane camenearer the ground.

Flare Paths

Part of the C-47 and T-33 investigation was to determine flare paths suit-

able for transition from the steep slopes to a shallow angle at touchdown.

Early in the C-47 and T-33 programs, it was evident that good approaches could

be flown down to 200 feet; therefore, a more demandingtask of attempting an

instrument landing was initiated. The guidance equipment provided flare paths

as shown in figure 4. The flare path was started at 6000 feet from the touch-

down point for both slopes shown. It should be noted that these flare paths are

long and change slope rather gradually, and thus, allow considerable distance

and time to accomplish the transitions from the glide slope to the landing.

There were only a few cases where the approaches were flown with enough

accuracy that the safety pilot would allow the hooded pilot to complete the

approach to touchdown. However, the safety pilots felt that the approaches,

both at 6° and 2.5 °, were flown with sufficient accuracy to allow visual

landings to be madefrom altitudes as low as 100 feet under good visibility con-

ditions. It should be pointed out that at lO0-foot altitude the glide slopes

and flare paths used in these tests place the airplane in a better position

relative to the desired touchdown than the conventional ILS glide slope shown in

figure 4.

Flare-path geometry was determined for the DC-8Fby extending the 3° flare-

path curvature upward to a point where the 6° slope was tangent to the path.

(See fig. 5-) Instrument touchdownswere not attempted with the DC-8F. The

pilots' task was to simulate instrument flight to lO0-foot altitude and touch

down visually from this position. Here again, the 6° flight-path geometry

appears to be better related to the desired touchdown spot than the conventional

ILS path. In this case, the airplane is in almost the sameposition at lO0-foot

altitude for both the 3° and 6° path. Basedon the conditions of the tests,

clear weather, good visibility, and with flare-path guidance, the pilots felt

that 6° approaches could be flown consistently down to lO0-foot altitudes. The

qualifying conditions of good visibility and special guidance equipment are to

be noted.

Lateral-Directional Control

Another important aspect of this steep approach study was the effect of

lateral-directional control of these aircraft on the steep-approach problem.

The lift, drag, and engine characteristics determined the maximum opera-

tional glide slope for each airplane investigated. Lateral-directional charac-

teristics did not limit the steep-approach capability for any of these aircraft.

However, the instrument approach is a very demandingpilot task, and good lat-

eral handling qualities reduce the pilot workload and result in more precise

flight-path control. This condition is equally true for both the steep and con-

ventional slopes.

The C-47 had the poorest handling qualities of any of the airplanes inves-

tigated. The high lateral-control forces and large wheel displacements made

small heading corrections difficult. It took considerable effort to start the

airplane in the desired direction, and the sameeffort was required to stop it; thus, overshoots were commonplace. Whenapproaches were flown in gusty air, the heading correction becamea very difficult task.

An example of the deterioration of the smooth flight path along the glide

slope after the pilot had to concentrate on a lateral deviation is shown in fig-

ure 6. By using the ordinary ILS cross-pointer display, the pilot had to decide

how muchbank was needed, when to take it out, and then repeat this procedure in

the opposite direction in order to comeout on course with the proper heading.

In this case, rather than ending up on course, he overshot to a greater local-

izer deflection in the opposite direction. The concentration and effort

required to makethese corrections detracted from the glide-path control and

resulted in larger gllde-slope deviations during this time.

The T-33 and DC-8Fhad more desirable handling qualities than the C-47 and

generally the time histories of localizer variation did not show the oscillatory

variations which were found to be typical of the C-47 aircraft.

However, even with good lateral-directional characteristics, the pilots'

workload was still too great to allow both glide-slope and localizer needles to

be flown with the precision required for consistent touehdownswhile under the

hood. It is recognized that a zero-zero touchdown is not a realistic opera-

tional task, particularly with only basic ILS cross-pointer guidance display;

but the extreme difficulty of this task tends to emphasize the differences in

flight-test results, such as those given in table II.

Table II illustrates the improvement in touchdown capability achieved by

eliminating the lateral-directional task. During someof the approaches with

the T-33, the safety pilot controlled the lateral-directional path while the

hooded pilot only had to follow the glide slope and flare path. 0nly one touch-

downwas madeout of 16 approaches with complete control by the hooded pilot.

Six touchdowns were madeout of seven approaches with split-axis control.

The DC-8 had the capability of split-axis autopilot control_ that is, cou-

pled approaches could be madewith autopilot control of the lateral-directional

axis and manual control of the glide slope.

The pilots found this modeof operation to be very effective in giving bet-

ter localizer tracking and in reducing the pilot workload and, thereby, allowing

more effort to be put on glide-path control. Very good approaches were flown

downto lO0 feet. For several years various groups have indicated the desira-

bility of using someform of split-axis autopilot. Someairlines are presently

using this split-axis operation in their fleet as part of the overall program

to reduce ILS minimums.

CONCLUDINGREMARKS

An investigation of steep instrument approaches has indicated the following

results for the C-47, T-33, and DC-8Faircraft:

i. Six-degree glide slopes can be flown with reasonable precision. Under

conditions of good visibility and with special flare-path guidance, these

approaches can be flown to 100-foot altitudes.

2. The maximum glide slope suitable for operational use was 3° to 4° less

than the maximum glide slope attainable. This margin is needed to allow flight-

path corrections to be madewithout increasing speed.

3. Control of the lateral-directional axis was the most difficult problem

for either conventional or steep approaches. Split-axis autopilot control of

this axis relieved the pilot workload considerably and resulted in better

flight-path control of all axes.

REFERENCES

i. Hall, Albert W.; and McGinley, Donald J., Jr.: Flight Investigation of Steep

Instrument Approach Capabilities of a C-47 Airplane Under Manual Control.

_ASA _ _2559, 1965.

2. Hall, Albert W._ and McGinley, Donald J., Jr.: Flight Investigation of Steep Instrument Approach Capabilities of a T-33 Airplane Under Manual Control.

NASA TN D-2775, 1965.

_3 TABLE I INSTRUMENT APPROACHES USED IN EVALUATION NUMBER OF RUNS pER OF AT MAXIMUM OPERATIONAL SLOPE C-47 85 54 5 NASA T-33 107 ,49 5 NASA 2 AIRLINE DC-8F 47 30 TABLE II IMPROVEMENT IN ZERO-ZERO TOUCHDOWN CAPABILITY BY ELIMINATION OF LATERAL-DIRECTIONAL CONTROL TASK T-33 AIRPLANE; GLIDE SLOPES, 2.5 ° AND 6 ° FLIGHT-PATH NUMBER OF NUMBER OF CONTROL APPROACHES TOUCHDOWNS FULL CONTROL BY 16 I HOODED PLLOT SPLIT-AXIS CONTROL 7 6 THRUST REQUIRED FOR C-47 AIRPLANE PATH USED TO INTERCEPT GLIDE SLOPE l r = 0 o Va END OF INTERCEPT ( 3,000 RUNWAY GLIDE SLOPE I I I " """" I THRUST

LT,TUOE ,O00 r-

REQUIRED 1,000

FT %""

I THRUST MARGIN"_ I _ NEEDED FOR \A7._4 o o_- b-'--" _2 5 ] -,o. , ...../%_,H j - "-- I ',= I- _-_--'-t - pROPELLER -_--- 4. MILES } TO 4 MILES OF LEVEL FLIGHT W4NDMILLING SPEED Figure i Figure 2 THRUST REQUIRED FOR T-33 GEAR DOWN; SPEED BRAKES OUT; W=12,000 LB %0OO 7 =0 V__/ 2DO0 THRUST -3_ I / - BO REQUIRED, /q ENGinE SPEED, LB _60 , _i PERCENT OF ipoo _ __._70 MAX. DESIGN -"-_IN_US-ABLE_6n RPM t I _ I I 0 8O I00 120 140 160 SPEED, KNOTS Figure .5 FLARE-PATH GEOMETRY FOR DC-8F FT FLARE-PATH GEOMETRY FOR C-47 AND T-33 ALTITUDE ALTITUDE FT 4O0 50O ITOUCHDOWN ///_o SLOPE FL_E ,y SLOPE I END OF RUNWAY / TOUCHDOWN _// ILS 4O0 ' I / . _LS I I 5 ° FLARE / / END OF ._'" i 2OO 300 I I 6 ° F LAREI //" /'_o SLOP E_ • II .5 SLOPE RUNWAY / _j_//_,_j2 I I I I / iO0 2O0 /// I i I 0 d l _/L....I_ ,''_ T SIMULATED I_FR I00 0 I 2 3 4 5 6 7xlO 3 DISTANCE FROM TOUCHDOWN, FT _ I I 11 L I I I 2 3 4 5 6 7 8xlO 3 DISTANCE FROM TOUCHDOWN, FT Figure 4 Figure 9 EFFECT OF LATERAL DEVIATIONS ON GLIDE-PATH CONTROL C-47_ GLIDE SLOPE =6 ° LOCALIZER DEV IATI ON, DEG TOUCHDOWN 7 FLY LEFT O --_FLARE-"_/ 2.5 _..._ TIME TO FLY RIGHT _ _1 2.5'- GLIDE-SLOPE I DEVIATION, , DEG FLY DOWN _L, ___v , TIME, SEC Figure 6 SOME FACTORS AFFECTING FATIGUE OF AIRCRAft2 STRUCTURES By Herbert F. Hardrath NASA Langley Research Center SUMMARY Several points of interest to the operator of aircraft are reviewed.

Rates of crack propagation tend to increase and residual static strength tends to decrease as the strength of the materials increases. Ground-air-ground and other negative loadings accelerate the damage to certain critical sections so that in first approximation, the fatigue life in these areas should be estimated on the basis of the number of flights, rather than the flight miles or flight hours. Regular inspection for cracks is necessary to assure safe operation of existing flight structures.

I INTRODUCTION The belief that a well-designed aircraft may be operated with no concern about fatigue difficulties is gradually becoming recognized as the myth it is.

The review of some recently developed information may help to place fatigue in proper perspective, particularly for operators; for it is they who_ more than anyone else, must live with the problem.

SOME PARAMETERS OF THE FATIGUE PROBI_4 Several important factors influence the fatigue life of aircraft structures.

Rather obviously, the behavior of the material of which the structure is made, the details of design and manufacture, and the chemical, thermal, fretting, and load environments are the commonly recognized primary factors. Much of the fatigue research carried out over t_e past 100 years has been devoted to estab- lishing methods for anticipating the influence of these factors analytically.

In spite of this effort, current design procedures cannot depend completely on analytical predictions, and considerable testing is used to evaluate particular configurations. Recently, more attention has been devoted to the study of the accumulation of damage under the complex loadings encountered in service, the rate of fatigue-crack propagation, and residual static strength. The present paper deals mainly with these topics.

The observed scatter in fatigue life data under identical loading condi- tions and the varied loads experienced by individual aircraft of a given type raise considerable doubt regarding the predictability of the fatigue llfe of a given aircraft even after full-scale tests have been performed. For example, recent examination of military aircraft of a certain type led to the discovery of cracks which had developed in one-fourth to one-tenth of the time estimated from a full-scale test. The application of a sufficiently high scatter factor to insure against such failures in service would lead to prohibitive weight penalties under present design methods.

How then, can the operator proceed to operate an aircraft with reasonable safety? It is the purpose in the following discussion to consider the steps that can be taken to insure safer operation. The remarks are intended espe- cially for users of aircraft that have parts designed for so-called "safe life" in contrast to the "fail safe" concept.

As indicated previously, the emphasis is on the rate of fatigue-crack propagation and residual static strength; thus, the need for diligent inspec- tion is developed. The way in which fatigue life is affected by the number and severity of loads is discussed.

FATIGUE OF SIMPLE SPECIMENS AND STRUCTURES Fatigue behavior is generally characterized by S-N curves of the type shown in the left side of figure I. Such curves have occupied the attention of a host of investigators over the years and have become the yardstick by which the influence of any of a dozen or more parameters is evaluated. The parameter illustrated in figure i is the effect of stress concentrations.

Many papers have been written to develop empirical or semiempirical relations to relate the S-N curves of unnotched specimens with those of notched or simulated structural specimens.

If one considers the observable propagation of a fatigue crack in a simple unnotched specimen, he will find a behavior such as is indicated by the lower curve on the plot on the right in figure i. However, for the same overall life of a notched specimen, the applied stress is lower, the effective rate of crack propagation is slower, and the critical crack length is longer for these struc- tural parts than for simple specimens. Thus, the crack is present for a larger proportion of the total life in structural parts. In practical situations, cracks are frequently present for well over one-half of the total life. On the one hand, this is fortunate, in that damage may be found by inspection and cor- rected. This fact makes feasible the incorporation of fail-safe characteristics in the design of a structure. On the other hand, such cracks reduce the resid- ual static strength so that the structure becomes more vulnerable to sudden failure under high loads.

CRACK GROWTH The rate of propagation of cracks and their effects on residual strength are getting more attention in recent years and this trend is likely to lead to more satisfactory methods for producing reliable structures. An example of how a systematic study of crack growth and residual strength may lead to an im- proved understanding of fatigue behavior is illustrated in figure 2.

The failures of three sheet panels, each made of a different aluminum alloy, are compared. The panels are all the same width but the thicknesses are proportioned in such a way that the original tensile failing load PU is the same in each case. Each panel contains an initial hole in the center. Each curve in the figure was calculated to illustrate the progress of a fatigue crack P during tests in which each panel is loaded in fatigue at the same proportion --

Pu

of its ultimate tensile strength. The end of the curve represents failure when the panel is subjected to a high static load (60 percent PU) at which time that static load is just sufficient to produce failure. It is quite obvious that the weakest of the three alloys (2024-T3) has the lowest rate of fatigue-crack propagation and tolerates the longest Crack at failure; the strongest material (7178-T6) is at the other end of the scale. With this sort of insight, one can readily understand why structures made of some of the new high strength mate- rials appear much more vulnerable to failure than do those made of moderate strength materials.

The picture is even more dramatic in steels where heat treating to high strength levels S U has become quite common in recent years, particularly for landing gears and actuating mechanisms. The same kind of crack growth as shown in figure 2 is shown in figure 3 for two pieces of steel heat treated to two different strength levels. The lower strength material again exhibits a lower rate of crack growth and tolerates a much longer crack before failure. Note that even this longer crack is only one-fifth as long as that tolerated by the 2024-T3 aluminum alloy in figure 2. The curves represent the behavior that might be observed in tests conducted in a clean laboratory at room temperature.

However, the situation is much worse in the environment encountered in service.

A crack as short as 1/16 inch may propagate to failure under low static loads in just a few hours on a wet day.

The same kind of phenomenon has been responsible for the failure of landing gears that become "cold-soaked" during flight and become wet or even frosty from condensation during descent or landing.

Obviously, the variations of behavior just discussed are partly attribu- table to choice of material and that choice is up to the designer. However, the operator should be aware of the size of crack that may be serious if he owns a structure made of one of these materials.

Perhaps some examples of cracks that have caused difficulties would be of interest. Figure 4 shows a crack in the skin of an aircraft wing. The mate- rial was 7075-T6 aluminum alloy about 1/2 inch thick. Failure occurred at approximately 90 percent of limit load with a crack less than 1 inch long. The disconcerting feature of this crack was that it was not detectable from the out- side and an internal doubler plate covered the inside surface to within 1/4 inch of the tip of the fatigue crack.

Had this wing been made of 2024-T3 aluminum alloy the crack would have grown at a slower rate for at least 2 more inches before failure and thus would have enhanced its chances of being observed before final failure occurred.

This particular crack was observed in a full-scale fatigue test and subsequent inspections in the fleet led to installation of a reinforcement of this loca- tion. Thus, the full-scale test can be a most informative means for identi- fying potential crack locations. It also provides information regarding the rate of fatigue-crack propagation and serves as a test bed for the development of specific crack-detection apparatus and techniques for a given structure.

An example of a manufacturing defect that can lead to fatigue difficulties is given in figure 5. The part shown was taken from the corner of an access cutout in the s_me wing st__Icture in which the previous crack was found. The designer wisely called for a radius of about 3 inches, all rough edges were carefully beveled, but the machinist permitted his milling cutter to dwell, leaving a small indentation. It was from that small defect that the fatigue crack grew. This is probably not representative of failure locations in this particular type of aircraft. Rather, it illustrates the need for thorough repeated inspection of every aircraft to discover cracks starting from manu- facturing or service-induced flaws. Obviously, not every such blemish can be polished out of every aircraft, but inspection must be thorough and regular to prevent trouble. 0nly the operator is in a position to conduct such inspections.

EFFECTS OF GROUND-AIR-GROUND CYCLE ON FATIGUE LIFE It was pointed out previously that fatigue life was related to the number and severity of loads encountered. A few points of interest to the operator should be discussed.

The influence of the so-called ground-air-ground (GAG) cycle has been studied in some detail in recent years. This is the average load cycle that occurs once per flight when the aircraft is lifted off its landing gear during take-off and is then returned to the gear upon landing. The bar graph in fig- ure 6 shows the result of a systematic series of tests in which identical spec- imens were subjected to programs of fatigue loads simulating, in one case_ the gust-load cycles alone and, in the other, the gusts and ground-air-ground cycles. The ground-air-ground cycles caused a decrease in life to less than one-fourth of that for gust-load cycles alone. The influence of the GAG cycle was found to be more severe when it was simulated at appropriate intervals; that is, once per flight, rather than in groups as is often done in simulated service tests.

From these observations one may conclude appropriately that, at least for certain critical sections_ ground-air-ground cycles can produce more damage that all other normal operating loads experienced by transport aircraft. This obser- vation may be rationalized in a qualitative sense by considering that the nega- tive phase of the load cycles systematically removes beneficial residual stresses. The more frequently this happens, as in each flight, the more serious is the effect.

4O

EFFECTS OF FREQUENCY OF OCCURRENCE

Figure 7 is a comparison of the numberof times a given load is encoun-

tered in transport aircraft. In the left plot in the figure the data are

normalized on a per-mile basis and in the right plot, on a per-flight basis.

It should be emphasized that, for this purpose, a flight includes only one

take-off and one landing. The data were accumulated on scheduled airlines

ranging from feeder-line transports to transcontinental jets. The samedata

are used for both plots. Quite obviously the variation in the data is much

smaller when comparedon a per-flight basis.

The trend observed is easily understood when one considers that most maneu-

vers are performed and most turbulence is found at low altitudes. Once the air-

craft reaches cruise altitude, turbulence is much reduced, maneuvers are usually

gentle, and the pilot has freedom to select a route that avoids storms. If one

considers, in addition, the predominant effect of the ground-air-ground cycle,

the variation in the damageproduced per flight is even less than indicated in

figure 7-

Thus, the operator should be aware that, in first approximation, fatigue

life is determined by the number of flights, rather than by flight miles or

flight hours. This fact is of real interest inasmuch as the trend in the air-

line operations seemsto be to use aircraft designed for mediumrange for ever

shorter routes.

Obviously, if the speed_ gross weight_ altitude, or other mission charac-

teristics are changed, as is frequently the case for military aircraft, the

fatigue life will be influenced directly. Probably no factor has been so

potent in this respect as the current trend toward low-level operations. In

somecases, such operation has resulted in an increase of two orders of magni-

tude in the frequency of occurrence of a given load-level per flight. Fatigue

difficulties under such conditions should comeas no surprise.

CONCLUDING REMARKS

The information presented in this paper emphasizes the following points of

interest to the operator of aircraft_

Rates of crack propagation tend to be higher and residual static strength

lower in higher strength materials than in lower strength materials.

Ground-air-ground and other negative phases of the loadings produce more

average damagethan ordinary gust loadings.

In first approximation, fatigue life in somecritical sections is deter-

mined by the number of flights, rather than by the flight miles or flight hours.

Regular inspection for cracks is necessary to insure safer operation of

flight structures.

CRACK GROWTH IN TYPICAL ALUMINUM-ALLOY SHEETS CRACK LENGTH, IN.

P PU FATIGUE BEHAVIOR OF SIMPLE SPECIMENS AND STRUCTURES .2 - LENGTH 7075-T6

l

Oq I I .5 I.OXlO 5 10 3 10 5 iO 7 0 310 410xlO 3 I0 20 CYCLES TO FAILURE CYCLES APPLIED CYCLES OF LOAD Figure i Figure 2 CRACK GROWTH IN HEAT-TREATED AIRCRAFT STEEL CRACK LENGTH, IN. p HEAT TREATED TO 1,2 _ _ SU= 180 TO 200 kSi_T ,.o :2 .8 0 .6 T_HEAT TREATED TO / A FAILURE / SU = 220 TO 240 ksi / ON A COLD / / I I I I 0 25 50 75 lOO X 10 3 CYCLES OF LOAD Figure 3 FATIGUE CRACK AT TOOL MARK FATIGUE CRACK IN 7 0 7 5 - T 6 PLATE

Figure 5 L-59-2980

Figure 4 L-2416-6

IN-FLIGHT ACCELERATIONS EFFECTS O F GROUND-AIR-GROUND CYCLE ON FATIGUE LIFE GUSTS ONLY GUSTS PLUS GAG I L ACCELERATION, UNITS Ax103 L I F E . EQUIVALENT FLIGHTS

Figure 7

Figure 6

, USE OF VERY WEAK RADIATION SOURCES TO DETERMINE AIRCRAFT RUNWAY POSITION By Fred J. Drinkwater III and Bernard R. Kibort NASA Ames Research Center SUMMARY Various methods of providing runway information in the cockpit during the take-off and landing roll have been proposed. The most reliable method has been to use runway distance markers when visible.

Flight tests were used to evaluate the feasibility of using weak radio- active sources to trigger a runway distance counter in the cockpit. The results of these tests indicate that a weak radioactive source would provide a reliable signal by which this indicator could be operated.

INTRODUCTION An investigation was conducted to determine the feasibility of using weak radioactive transmitters to provide position information to the crew during take-off and landing. The transmitters were spaced along the runway and detected by equipment in the aircraft. This information is similar to that which is provided by runway distance markers except that the transmitters are not affected by outside visibility. The need for more precise runway distance information becomes apparent when the basic four-engine jet take-off condition is examined. The airspeed-distance relationship during a take-off is shown in figure i. When all four engines are operating, the acceleration proceeds as shown by the solid curve. There is one point at which, should an engine failure occur, the distances required to come to a complete stop or to continue accel- erating to take-off speed (V2) on three engines are equal. This point on the runway is the critical engine failure point and the associated airspeed reached at this position under normal acceleration conditions is called the critical engine failure speed or basic V 1.

Although the concept of a critical engine failure speed is supposed to account for the sudden loss of an engine, there is no effective way, presently, using speed alone to detect reduced performance resulting from an underthrust engine condition, a dragging brake, drag due to slush or puddles on the runway, or a miscalculation of gross weight for a critical runway length. Under these conditions the critical engine failure speed is reached at a point further along the runway than was expected; thus, speed alone is hazardous to use as a deci- sion point. However, several factors have permitted this lack of information to exist. First, unless the airplane is loaded to the maximum allowable weight for the runway to be used, the refusal condition given the crew is conservative L-4612 45

since this condition is based on the maximum allowable weight for the partic-

ular runway. Secondly, the reliability of jet engines is extremely good, and

lastly, the stopping distances are demonstrated without reverse thrust which

provides an additional margin of safety in an actual emergency.

Obviously, a measurementof aircraft acceleration would provide a satis-

factory solution. The measurementcould be madeeither directly with acceler-

ometers or indirectly by noting the time or distance required to reach a chosen

airspeed. Direct measurementsof acceleration have been used in experimental

take-off monitor systems as noted in references l, 2, and 3- However, none of

these methods are in general use. Time checks of acceleration to a reference

airspeed have been used but the technique is not considered accurate enough by

the pilots; thus, few operators have retained it. Airspeed and distance checks

of take-off acceleration are _ _ the military because _o+ _1_÷o_y runways

have distance markers spaced 1000 feet apart. These markers are useful in day-

light to provide an airspeed-distance decision point even for single-engine

aircraft. However, few civil runways are equipped with distance markers. Since

distance information is an important part of a complete take-off performance

monitor system, several types of wheel rotation pick-offs are being developed

for this purpose, for example, the systems described in references 4, 5, and 6.

The purpose of a take-off monitor system is to provide an indication of

unsatisfactory performance as early as possible. The attendant problems are

several. For example, too early a prediction raises the possibility of unnec-

essary refused take-offs. In addition, the accuracy and reliability of accel-

erometers and wheel pick-offs in this environment are still not adequate and

these systems usually require the inclusion of aircraft gross weight, runway

temperature, runway slope, and altitude into a computation which adds to the

possibility of error. Since these developmental problems have delayed the

introduction of a true take-off performance monitor system, a less ambitious

system is to provide the pilot with runway distance information under all con-

ditions of visibility. Take-off performance is then checked by using airspeed

and distance Just as performance is checked on a runway which has distance

markers. The decision to refuse a take-off can then be madewith the knowledge

that sufficient distance is available for stopping. The reference reports dis-

cuss several ways of obtaining distance during the take-off and this paper pre-

sents still another method.

Preliminary studies indicated that weak radioactive sources spaced along

the runway as distance markers could be readily detected by simple airborne

equipment. The runway installation of sources has advantages such as low ini-

tial cost, simple installation, negligible maintenance, and high reliability

since the runway units required no power and would not be affected by runway

surface conditions. Subsequent to these studies, NASA negotiated a contract

for the design and fabrication of a two-transmitter runway installation and the

airborne components.

RUNWAY POSITION MONITOR EQUIPMENT Runway Installation Gamma radiation transmitters, each of which consisted of a lO-millicurie source of Cesium 137 (Csl37), were installed in containers which shaped the radiation pattern (fig. 2) and were located along the center llne of the run- way. The Cs 137 in the container has a half-life of 30 years, no electrical power or periodic maintenance is required, and the transmitter units are not a source of radio or magnetic interference. The containers were designed to pro- vide security for the small radioactive source and to allow de-activation by the insertion of a lead plug. For this investigation two transmitters were embedded along the center line of the runway on the dry lakebed at Edwards Air Force Base (fig. 3). The transmitters were spaced 200 feet apart, since it had been determined from earlier tests that for the source strength used, 200 feet would be the minimum spacing for separate detection of each source; the spacing could, of course, be greater. For example, lO00-foot spacing would still allow adequate acceleration checks and provide braking distance information. The hardened dry 18_ke surface has radiation shielding characteristics similar to those of concrete. However, the natural background radiation present over the lakebed was several times greater than would be present on a conventional runway.

Airborne Equipment A correlation between the actual aircraft position and the position indi- cated by the radiation sources was determined photographically. The essential airborne components are shown in figure 4. The scintillating probe is a 3-inch thallium-activated sodium iodide crystal, coupled to a 3-inch photomultiplier tube. The largest unit contains the amplifier discriminator, binary counter, timer circuit, logic gates, and an indicator driver. A conventional impulse counter, shown in the figure, was the only component required in the cockpit.

The power used was ll5 VAC 60 cps. Since the large unit was a prototype, no effort was made to reduce its size. In fact, the only units which could not readily be reduced in size were the scintillating probe and photomultiplier tube.

A symbolic diagram showing the complete installation of the airborne equip- ment is shown in figure 5. There was no external probe or antenna. The only requirement was to minimize the mass of aircraft material or structure inter- posed between the scintillating probe and the runway. The recorders indicated in figure 5 were part of the test instrumentation used to obtain a continuous record of the radiation level. The runway distance meter or counter was located between the pilots for coordinated use.

RESULTS AND DISCUSSION The aircraft was flown over the sources in addition to the normal take-off tests. This was done to obtain the higher speed and altitudes required to determine the limiting conditions. A recording made during a typical pass is shown in figure 6. The upper trace is from a count rate meter in counts per minute. The lower signal corresponds to the upper trace but is in a form to operate the cockpit indicator. Each vertical step is 8 counts which are totaled approximately each 1/3 second as indicated by the return to zero on the trace.

The background level is seen to provide 24 to 52 counts before resetting. When a transmitter is approached, it is seen that the count rate increases and when the total exceeds 64, 8 steps up, before the reset time, a signal is passed to the indicator driver. Successive co_uts in excess of 64 are blocked by the and-or gates until the total count drops below 64 during the i/3-second time interval. This return to zero indicates that the beam has been passed and another step is sent to the cockpit indicator by the driver. In this way the mechanical counter is stepped two times for each radiation source, once on entering and once on leaving the beam. The scale on the indicator mounted in the cockpit in the case of this test registered i00 feet for each pulse sent by the indicator driver. However, only the odd lO0-foot indicator pulses were found to measure an accurate distance between the sources. (That is, the dis- tance between the pulses, at the points where the 64 count gate was entered, was proportional to the 200-foot spacing between the sources.) The run shown in figure 6 was made at a measured 155 knots ground speed 35 feet above the center line. As can be seen in figure 6, the signal over each source is readily detect- able and the probability of receiving an inadvertent count or of not counting when over the source at signal levels such as shown in this figure can be esti- mated to be less than one chance in one million. During these tests reliable detections of the sources were achieved at distances of 50 feet and speeds of over 150 knots. If the sources were spaced further apart the speed of passage would have a less significant effect. The source detection at heights of 50 feet was evaluated for possible consideration of the use of this signal to indicate to the pilot his touchdown point after landing.

The radiation level of the source installed in the container was lower than the predicted value because of excessive shielding by the container. This value indicates that improvements can be made in the container design. In addi- tion the hazards associated with the radiation sources can be further reduced by a design which distributes the sources more uniformly across the runway.

The ultimate in distribution would be obtained by painting stripes of an epoxy-type base containing the radiating material. The radiation level of such a stripe could be less than 1 mR per hr at a distance 4 inches from the surface.

This rate can be compared with the radiation level of a luminous dial watch which has a rate between 1.O and 2.0 mR per hr at a distance of 1/2 inch.

Another method would be to install aluminum tubing in slots in the runway. The tubing would contain a homogeneous mixture of the source material and an inac- tive filler which would again reduce the local concentration of radioactive material.

Under the present rules of the Atomic Energy Commission, upon application permission may be obtained to install sources in unrestricted areas if the radiation levels are such that an individual is not likely to receive a radia- tion exposure in excess of 500 mR per yr. A person would have to remain within l-l/2 feet of the source used in this experiment for more than 40 hours per year in order to exceed this exposure. By using the improved installation the distance would have to be in the order of several inches for several hundred hours per year. Since this is unlikely, especially on an active runway, it is considered probable that the U.S. Atomic Energy Commission or the applicable State regulating agencies would approve the use of radioactive sources for the purposes set forth in this report 3 provided satisfactory precautions are taken to secure the sources from accidental or malicious removal from the site.

CONCLUDING REMARKS Operational tests have shown that aircraft distance traveled during the take-off roll can be determined by using relatively simple airborne equipment to count nonhazardous radiation sources located in the runway. The detection of the sources was reliable at heights of up to 50 feet above the runway. With this information provided in the cockpit under all runway conditions, the pilot can not only determine whether his take-off acceleration is satisfactory by using an airspeed-distance check, but also make the decision to continue or refuse the take-off at a point which assures sufficient stopping distance.

REFERENCES i. Kolnick, Joseph J.; and Rind, Emauuel: Investigation of the Characteristics of an Acceleration-Type Take-0ff Indicator in a Large Jet Airplane.

NASA MEM0 4-21-59L, 1959.

2. Morris, Garland J.; and Lina, Lindsay J.: Description and Preliminary Flight Investigation of an Instrument for Detecting Subnormal Acceleration During Take-0ff. NACA TN 3252, 1954.

3. Swindl% Elro M.: Flight Test of Sperry Gyroscope Company Takeoff Monitor in a B-47 Airplane. WADC TN 59-360, U.S. Air Force, Jan. 1960.

4. Various contributors: Towards Safer Take-Offs. Flight Intern. (Letters), vol. 86, no. 2899, Oct. I, 1964, pp. 600-602.

5. Fearnside, K.: Instrument Aids for Take-off. J. Roy. Aeron. Soc., vol. 67, no. 633, Sept. 1963, pp. 544-551.

6. Rush, D. A.: Ground Run Prediction. AGARD Rept. 422, Jan. 1963.

TYPlCAL SOURCE CONTAINER INSTALLATION BASIC ENGINE FAILURE DECISION SPEED SEALED SOURCE, ,LOCK TAKE-OFF SPEED w AL-MlNUM

m

0 0

w 0 LEAD

a w MAXIMUM I n BRAKING P

a COhCRETE

5 4 ENGINE

\ 4 DISTANCE

Figure 1 Figure 2

TEST INSTALLATION AND POSSIBLE OPERATIONAL INSTALLATION

Figure 3

Figure 4 A-33993

AIRBORNE RADIATION DETECTION EQUIPMENT COUNTER RECORDERS \ SCINTILLATING PROBE RADIOACTIVE SOURCE (Os 137) Figure 5 TAKE-OFF RADIATION MEASUREMENTS GROUND SPEED 155 kts ALTITUDE 34.8 ft OFFSET +3ft -- _ _% _-- SOURCE I 2 ° LouNo ; _ ,20- j : F- R i _z _ 72 _- 64 COUNT GATE ...... ..... ' ......

_ 4B- j _ 24 0 r_,L2_.--_-__--J,_;,-'.'_'.C_-Z.-_:Lr'_:'_'LL----;_.q_:-.---JZLC_d.£, I sec TIME, sec Figure 6 A LABORATORY INVESTIGATION OF TURBULENCE DETECTION e USING A lASER By Kent Bourquin and Fred H. Shigemoto NASA Ames Research Center SUMMARY Calculations and experiments have been undertaken to determine the property of the backscattered laser-return signal that undergoes the most significant change in a turbulent region as compared with a nonturbulent region. If the turbulent region is composed mainly of air molecules, the detection of a shift in the frequency property appears to be the most promising. Therefore at this time the main effort at the Ames Research Center is focused on the implementa- tion and laboratory evaluation of a clear-air turbulence detector utilizing a laser source and observing the frequency change.

INTRODUCTION The unexpected penetration of clear-air turbulence has increased greatly in recent years because of the increased number of flights and the lack of any reliable detection methods. Clear-air turbulence presents itself in different forms and has been reported at altitudes from sea level to 100,000 feet where it has been encountered by U-2 and X-15 flights. Turbulence, however, has most often been encountered near the jet stream at altitudes of 20,000 to 40,000 feet.

The region of turbulence is small compared with the overall air volume; it is usually less than 3,000 feet in vertical depth, 20 miles in width, and 50 or more miles in length along the direction of the wind. This localized nature of clear-air turbulence makes it very difficult to predict from normal meteoro- logical observations. Still it would be extremely valuable to be able to detect clear-air turbulence regions and, as yet, there are no suitable devices avail- able. Several different approaches to the detection of turbulence are pres- ently being pursued. Some of these devices rely on the detection of tempera- ture or electric-field gradients in the vicinity of turbulence. Since these effects occur near clear-air turbulence, there exists a limit to the maximum advance warning time available to the pilot to execute a maneuver to avoid the area. A fundamental requirement of a clear-air turbulence detector is that the device must give the pilot enough warning to make avoidance possible.

The success of a conventional radar in detecting storms has led to con- siderations of its use in detecting turbulence. However, the extreme ratio of wavelength to particle size makes its effective use appear extremely doubtful.

An optical radar using a laser appears more promising owing to its much shorter wavelength. Ames Research Center is conducting laboratory experiments in order to determine the feasibility of using an onboard laser as a probe to detect clear-air turbulence.

An investigation is being carried out to determine whether the laser light backscattered from a turbulent region will be modified as compared with light backscattered from a nonturbulent region. Laboratory experiments_ rather than actual onboard experiments_ were chosen in order to correlate the results with a controlled atmosphere.

DISCUSSION The backscatter consists of three phenomena: Rayleigh scattering from gas molecules and microscopic particles that are small compared with the wave- !e_th of light_ Mie scattering from dust particles that are comparable in size to the wavelength of ligh% and refractive scattering due to inhomogeneities of the refractive index. An analytical study of the latter phenomena has revealed it to be insignificant compared with the first two.

Before discussing the results of the experimen% some background material is presented that justifies the choice of a laser system as a possible clear- air turbulence detector. The background material will also include the prop- erties of the laser source and a description of the experimental setup. A laser detection system would be similar to a conventional microwave radar_ except that it would operate at a much higher frequency_ one that is in the visible region. A comparison of the two systems_ Rayleigh scattering being assumed_ shows that there is a 1020 increase in scattered return and a 10 -5 reduction in beam angle owing to the 105 difference in frequency. In order to make a preliminary determination of the feasibility of an optical radar, cal- culations were made of the received signal-noise rati% a region free of high particle content being assumed and, as a result of this assumption 3 the air molecule scattering is predominant. The calculations show that the cur- rently available pulsed lasers are feasible but that the currently available continuous-wave lasers at their best are borderline.

For use of either a pulsed or a continuous-wave laser_ consideration would have to be given to the effect of the laser beam on the eye of occupants in other aircraft that might be located along the laser line of sight. The high- power pulsed laser would certainly require special safety precautions_ whereas the continuous gas laser might be tolerated for short exposures. This area will require further investigation.

A study of the effects of turbulence on the characteristics of backscat- tered radiation requires an understanding of the properties of the laser light contained in the backscattered signal. Important properties to be considered are amplitude_ polarization_ coherence_ and frequency. The amplitude is an indication of the power in the beam. Polarization of the wave refers to the orientation of the electric-field vector in a plane perpendicular to the direction of propagation. The path described by the tip of the electric- field vector in the plane defines the polarization. For example3 a straight- line path would be called linear polarization.

One of the most important laser characteristics is coherence. It is this property which allows the laser to be radiated in a narrow beam and focused into an intense spot. It also causes a granular appearance when the light is incident on a surface. There are two kinds of coherence. One is spatial coherence, the other is temporal coherence (or monochromaticity). These two coherence characteristics in reality form a three-dimensional coherence func- tion where temporal coherence describes phase correlation of the radiation in the direction of propagation, and spatial coherence describes phase correla- tion across the wavefront.

The frequency is of the order of lO 14 cycles per second. The frequency determines the color of the laser light. The laser output can be composed of one or several discrete frequencies 3 depending on the optical cavity length and laser power input.

Laboratory experiments have been designed to determine which properties of the scattered laser-return signal have the most significant change in a turbu- lent region as compared with a nonturbulent region. The experimental equipment is not designed to simulate actual clear-air turbulence phenomena, but only to provide a gross air motion and to introduce aerosol particle concentrations.

This experimental technique provides a means for evaluating the properties of the return signal under controlled conditions. One of the main reasons for not attempting to simulate clear-alr turbulence is that the actual phenomena is not thoroughly known. In particular, the particle content is not known.

Some meteorologists speculate that silt and clay are distributed in the eddies around the jet stream.

Figure 1 shows a photograph of the laboratory setup. The laser output passes through a fused-quartz window and into the test section. The portable enclosure is sealed and bolted to the bulkhead and allows the test section to be evacuated for initial calibration. The ports located on the side of the enclosure allow air flow to be introduced through the ports by an external clean-alr source with a 0.3-micron filter.

Figure 2 shows more detail of the system. A continuous-wave gas laser is located on one end of a 17-foot aluminum T-beam. The output of the laser passes through a fused-quartz window at Brewster's angle, the collimating aper- tures, and the test section, and is finally absorbed. The absorber consists of a cone lined with a smooth gelatin filter material that absorbs or diverts the wide-angle rays into the main section containing a filter. The filter, set at Brewster's angle, causes maximum transmission and minimum reflection of the polarized light; therefore the laser light is transmitted through the filter and rapidly absorbed. The collimator and absorber are required to prevent the laser light from being scattered off the walls of the chamber and into the optical receiver.

The test section is determined by the intersection of the laser beam and the receiver field of view. Figure 3 shows the geometry involved. The detector is a photomultiplier tube which has an S-20 surface.

It should be noted that most of the experimental equipment would not be used in an actual detection system. The main equipment needed in an actual clear-air turbulence detection system would be a laser and an optical receiver.

The other equipment shown in the figures merely permits evaluation of the previously discussed properties in a laboratory environment. The most diffi- cult design requirement was complete absorbance of the laser light at the end of the chamber since the received light at the detector from molecular scat- tering is only i0 -I0 times the initial light power of the laser. Any light reflected from the chamber must be kept well below this value.

The initial experiments verify that when the atmosphere is removed from the chamber, stray light signal is negligible. Under vacuum condition the photomultiplier output is essentially only the noise current of the photo- multiplier. The magnitude of the detected signal is 6.00 millivolts because of the photomultiplier noise curren% 8 miliivoits with the chamber under vacuum_ 350 millivolts with normal room air in the chamber, and i00 volts when a high concentration of water vapor is passed through the test section.

Preliminary experiments and calculations based on Rayleigh scattering have shown that the amplitude property of the returned signal is not signif- icantly modified in a turbulent region. On the other hand_ calculations show that if the turbulent region contained significantly larger particle concentra- tions than the surrounding nonturbulent region_ a change of amplitude could be detected. However_ even in this case_ the amplitude property might not be a reliable property to detec% since absorption and scattering, from clouds and other phenomena other than clear-alr turbulence would affect the amplitude during transmission.

The results_ as expected, have indicated no first-order effects on polar- ization due to turbulence. The effect of polarization was examined by rotating a linear sheet of polaroid in front of the receiver optics. The results showed the return to have polarization identical to that transmitted. However 3 further investigations using different particle sizes with different indices of refraction will be considered wi_h various laser output polarizations.

The spatial coherence of the return wave can be studied by use of an experiment similar to Young's double-slit experiment. If the phase difference between the waves entering the two slits changes, the interference fringe pat- tern will shift according to this change. Investigation of this property has no% as yet, been undertaken by the Ames Research Center.

One of the more promising properties for detection, which calculations show changes significantly in a turbulent region as opposed to a nonturbulent region, is that of frequency shift due to velocity gradients. This frequency shift is an effect analogous to the change in pitch of a moving train whistle as it passes a stationary observer. Detection of this property is complicated by the Brownian motion of molecules due to temperature. The effect of Brownian motion is to spread the frequency spectrum of the return signal from air mole- cules. In a turbulent region 3 the air mass is moving in a random fashion and thus the frequency spectrum is spread still further_ or the whole spectrum is displaced. Most particles are not affected by Brownian motion and backscatter from them will only display a spread in the frequency spectrum due to turbu- lence. If the air mass were moving in only one direction, a shift in frequency _6 would result. Detection is limited to velocity components parallel to the direction of travel. One method for detecting a frequency shift is by photo- mixing a local laser oscillator and the scattered return signal. The use of a local reference oscillator imposes the problem of critical alinement of the reference oscillator and the return signal. Photomixing also required both signals to have essentially the same polarization and spatial coherence. The velocity of the aircraft also causes the Doppler frequency shift to be outside the bandwidth of conventional photomultipliers and the types of photomultlpliers now being developed for laser photomixing would be required.

Another detection method recently proposed uses a laser in an operating mode which contains two frequencies in the output. By photomixing the return scattered signal_ which still contains the two frequencies 3 a frequency shift should be detected. This frequency shift would be proportional to the differ- ence between the two frequencies which is about 1.5 × lO 8 cycles per second instead of being proportional to the single laser frequency of lO i4 cycles per second. The advantage of this technique is that the signal bandwidth is within that of conventional photomultipliers and the bandwidth of the electronics can be narrow; hence, the noise in the system is reduced. Another advantage is there should be no critical alinement or spatial coherence problems, such as I matching with a local oscillator. However 3 there is still the main practical disadvantage of the inadequate power of current gas lasers.

Both methods of Doppler frequency-shift detection are being investigated at the Ames Research Center. Optical equipment is being assembled and elec- tronic spectrum-analysis-amplifier-filter combinations are being designed in order to carry out an experimental program in the test chamber.

CONCLUDING REMARKS Calculations and experiments have been undertaken to determine the property of the backscattered laser-return signal that undergoes the most significant change in a turbulent region as compared with a nonturbulent region. If the turbulent region is composed mainly of air molecules, the detection of a shift in the frequency property appears to be the most promising. Therefore, at this time the main effort at the Ames Research Center is focused on the implementa- tion and laboratory evaluation of a clear-air turbulence detector utilizing a laser source and observing the Doppler frequency change.

Figure 1 A-34386- 4

DETAILED SYSTEM DIAGRAM /BULKHEAD CONTI NUOUS-WAVE GAS LASER L A S E R COLLIMATOR LASER BEAM F I L T E R ENTRANCE WINDOW’

Figure 2

GEOMETRY OF EXPERIMENTAL SETUP AI R FLOW % /TEST SECTION LASER _1/8 in \_\ / _ ........

50 #-_oov RBr..R RECEIVER Figure 3 .

SIMULATOR EVALUATION OF A DISPLAY FOR MANUAL ZER0-ZERO LANDINGS By Joseph G. Douvillier, Jr.

NASA Ames Research Center SUMMARY A research program to study in flight the display, sensor, and computer requirements for airborne, manually piloted zero-zero landing systems is being undertaken at the Ames Research Center. An evaluation has been made, on a flight simulator, of the display which will be used in the flight tests.

Results show that on the simulator the pilots performed at least as well when they used the display for zero-zero approaches and landings as they did when they made simulated ILS approaches followed by simulated VFR breakouts and landings. In some aspects, particularly in the presence of crosswind, perform- ance for the zero-zero display landings was somewhat better than for the ILS-V_R landings.

Eight configurations of the display were studied in an attempt to evaluate the several elements in the display. There was no evident difference among the eight display configurations. The pilots who participated in the simulator program doubt that this result will carry over into flight.

INTRODUCTION The Ames Research Center is conducting a research program on the concept of manually piloted zero-zero landing systems. This paper is a report of some results of the evaluation on a flight simulator of the pilot's display which will be used in the flight program. The particular concept which will be explored is that of an airborne, manually piloted system: airborne, because to be independent of ground installations would allow zero-zero capability to air- plane operators using small, relatively unequipped airports - for examplej to short-haul, commuter airlines or to general aviation operators; manually piloted, because a manual system would likely be simpler than an automatic one and, therefore, probably lighter and more reliable - always two important con- siderations. Furthermore 3 even with an automatic system, provision for manual landings would be desirable as a safety measure. Manual capability, therefore_ is fundamental.

The objectives of the program are: First, to document quantitatively the performance of pilots in making manual zero-zero landings by using a synthesized landing display, compared with their performance in making landings in VFR con- ditions and, second, to determine the degree of accuracy which must be provided in the sensors and the computers as well as in the display. This paper per- tains only to the first objective.

Concurrently with the preparations for the in-flight research, a ground-

based flight simulator was used to develop and make a preliminary evaluation

of a display.

DISCUSSION

The basic display which will be used in the flight tests is shown in

figure 1. Basedon what had been learned in previous display research, the

approach taken was to make the display as VFR-like as possible - in other words,

such that the pilot could use it much as he does the outside world when making

_r_v_ua_1 landings. There would be no numbers explicit in the display and no

director information. The elements of this basic display are shown in figure 1.

There is a horizon line which behaves the sameas the horizon in the conven-

tional attitude gyro. There is a runway, depicted by four lines - like tar

strips across the real runway. The symbolic runway movesabout in the display

and has the sameperspective as would the real runway viewed from the pilot's

position. The velocity circle encloses the point on the ground toward which

the airplane is flying. It is the point of intersection of the total velocity

vector and the ground. There is also the fixed vertical index representing the

nose of the airplane.

It was reasoned that these elements contained all the information a pilot

needs to land his airplane, presented in a way that could be interpreted as

VFR-like as a symbolic display can be. He can discern from it his position in

relation to the runway by the asymmetry of the runway symbol; his path by the

location of the velocity circle; his altitude by the amount of separation

between the horizon line and the runway lines; and his attitude by the dis-

placement and the angle of the horizon line.

However, an early subjective evaluation suggested the need for more infor-

mation. The pilots wanted better lateral-displacement information when rela-

tively far out on the approach and better height and height-rate information

when close to the ground.

For better lateral-displacement information a centerline indicator was

added, as shownin figure 2. It behaves essentially as would the fuselage of

another airplane in front of and slightly below the pilot's own airplane, but

this other airplane is always lined up with the runway centerllne. It has been

nameda "ghost" airplane. Essentially, it gives the pilot a magnified indica-

tion of lateral displacement. It behaves like a lead airplane as seen from the

pilot's position. For example, in figure 2 the airplane is far out on the

approach; the runway symbol is too small for the pilot to perceive any asym-

metry. But he can tell from the ghost airplane that he is off to the left a

bit. He can also tell, from the velocity circle, that he is correcting the

error. He is flying toward a point to the right of the runway. He has to fly

toward the right of the runway if he wants to get lined up. Whenhe is lined

up, he will be right behind the ghost airplane; that is, the ghost airplane will

be between his own airplane and the runway. But the pilot does not want to fly

past the ghost airplane_ so, as he closes on being lined up - as the ghost air-

plane drifts in front of his own airplane - the pilot makes a turn into position

behind the ghost airplane. In other words, he flies toward the ghost airplane

and the runway.

For better height and height-rate information another symbol, a small dash,

was added. This altitude indicator is shown in figure 3. The displacement of

the dash below the velocity circle is proportional to airplane height above the

runway, scaled at 1 inch per lO feet. It does not appear in the display until

the airplane is down to an altitude of about 50 feet. The position and the rate

of change of position of the altitude indicator give the pilot a more compelling

signal than does the perspective of the runway. The dash is somewhatlike the

airplane shadowon the ground. Whenthe airplane is descending, the shadow

comesup to meet the airplane; if the airplane balloons upward, the shadowsinks

away. Whenthe shadowcoincides with the point toward which the airplane is

flying (the velocity circle), the airplane is on the ground.

The scaling of the display is one to one with the real world. That is,

the visual angles associated with a particular element of the display are the

sameas the corresponding visual angles associated with the object in the real

world which the particular element of the display represents. For example, the

visual angle between the horizon line and the runway threshold line in the dis-

play is the sameas the visual angle between the actual horizon and the actual

runway threshold. As a result, the sensitivity of the display changes as the

airplane gets closer to the ground, Just as the effective sensitivity of the

real world changes. The pilot has information to larger scale as he gets close

to the ground - with the display Just as with the real world.

The pilot can, as he chooses, make shallow approaches or steep approaches.

Figure 4 shows a comparison of the display for two angles of approach. In each,

the airplane is at the samedistance from the threshold. In each, it is flying

toward the samepoint on the runway. But in the left display the pilot is

making a more shallow approach than in the right. The different runway per-

spectives show this; but the primary indication is in the velocity circle. Its

distance below the horizon in the display subtends a visual angle equal to the

flight-path angle and is about three times greater in the right display than in

the left.

The experiment was madein such a way that pilots' performance in landing

when they used the display could be comparedwith their performance in landing

when they used a televised facsimile of the visual landing scene in conjunction

with an ILS approach - all on a flight simulator. Not only was the complete

display, as shown in figures 3 and 4, studied but, in order to get a measurement

of the values of the several elements, the display _as tested without the veloc-

ity circle, without the altitude indicator, and without the ghost airplane, each

removed singly and in combination with the other two. In this way the separate

and the interacting effects of these elements on the pilots' performance could

be measured relative to the complete display and to the visual picture. There-

fore, eight configurations were tested of this display, which varied in com-

pleteness from the one in figures 3 and 4 to one containing just the horizon

and the runway lines.

In addition, runs were madeboth with and without crosswlnd so that the

effect of this variable on each of the displays could be examined. The flight

characteristics simulated were those of a Breguet 941 STOL airplane in the landing configuration.

The flight-simulator configuration is such that a televised picture of the simulated visual landing is projected on an 8- by 6-foot screen attached to the front of the simulator cockpit and about 8 feet in front of the pilot as shown in figure 5. The simulator can be rolled, pitched, and translated vertically.

This simulator has been used in several Ames research programs, so the Ames pilots are "well-calibrate_' to its similarities and its dissimilarities to flight.

Figure 6 is a drawing of the view from inside the cab during an oscilloscope-display run. The pilot can see nothing outside the cab. Shown in the figure are the oscilloscope on which the display was generated, and the viewing mirror with the image of the display. During the ILS-VFR runs the dis- play was off. The pilot made an ILS approach down to an altitude of about 550 feet, at which time the projected televised picture of the simulated runway appeared on the screen outside the cockpit, and the pilot made a simulated visual landing by using the picture.

The simulator is fitted with a complete set of cockpit instruments, typi- cal of a large transport airplane. All panel instruments were operating during all runs, so that the pilots could gather whatever information they wanted from them, as they would in an actual airplane.

Although the analysis of the data is not yet complete, enough has been done to discover the trends. The preliminary analysis indicated that, on the average, the pilots did at least equally well at touchdown with any of the dis- plays as they did with the visual simulation. According to significance tests the differences which did appear at touchdown were scatter in the data and not due to any real difference among the displays and the visual simulation.

Not only were data recorded at touchdown but throughout the entire approach. The next three figures (figs. 7, 8, and 9) show average time his- tories during the 30 seconds before touchdown for three of the parameters measured. In these figures the results associated with the eight configurations of the scope display are presented in a shaded band, and the results associated with the ILS-visual runs are shown by a solid line. The results for the scope displays are presented this way not only because the measured differences among them are not large enough to be due, probably, to anything other than experi- mental variation but also because there was no discernible tendency for one scope display to be different from the other scope displays (to be at the high end or the low end of the band). However, on some of the curves there is a tendency for the performance with the visual simulation to be different from performance with the scope displays.

Figure 7 shows rates of descent. The upper curves show rates of descent averaged over all the runs with no crosswind; the lower curves, with crosswind.

As can be seen, there is virtually no difference between the visual runs and the scope-display runs, either with or without crosswind. Moreover_ the crosswind curves are almost the same as the no-crosswind curves; therefore, there appears to be no effect of crosswind on the rate of descent for either the visual sim- ulation or the scope displays.

All this indicates that the pilots controlled their rate of descent in the same way with any of the scope displays as with the ILS-visual simulation, under both no-crosswind and crosswind conditions. The rate of descent was about 13 ft/sec during the approach; they started to break the glide at about lO sec- onds before touchdown, and the rate of sink was about 5 or 6 ft/sec at touch- down. All these values are typical of actual landings with the Breguet 941.

Figure 8 shows time histories of average drift angle. Averages of the runs with no crosswind, the upper curves, show essentially zero drift angle - as might reasonably be expected with no crosswind. However, the averages of the runs with crosswind, the lower curves, show that the pilot flew differently when he used the visual simulation from the way he did when he used any of the scope displays, for which averages tend to group. On the ILS-visual runs, breakout occurred at about 28 seconds before touchdown, so the curve is for the visual portion of the ILS-visual runs. (Of course, the curves for the displays represent only zero-zero conditions.) Obviously, the effect of crosswind tends, during the approach, to be greater when the pilot used the visual simulation than when he used any of the scope displays. This result was not surprising because, in a crosswind, the pilots used a technique with the scope displays that was different from the one they used with the ILS-visual simulation.

This different technique, although it resulted in slightly better perform- ance, was actually necessary because of a shortcoming in the display. The horizontal visual range of the scope display was from i0 ° to -i0 °. The crab angle necessary to correct for the crosswind used was 9 °. Therefore, if the crosswind were corrected for by crabbing, the runway (and the ghost airplane and the velocity circle when they were in the display) would be well over on one edge of the display, almost out of sight. Any small deviation would drive it off the scope. So, with the scope displays, as soon as the approach began the pilots held the heading of the airplane so that the runway stayed essentially in the center of the scope and compensated for the crosswind by holding a wing down. The airplane was more nearly decrabbed throughout the approach with the scope displays than with the visual simulation, and no large last-second correc- tions had to be made.

On the ILS-visual runs the pilots would crab the airplane to compensate for crosswind, until breakout, and then gradually feed in some sideslip. In fact, on the average, they let the drift build up again beginning at about lO seconds before touchdown and then decrabbed during the last 3 seconds. However, at the instant of touchdown, the amount of drift was not large in any case; and the spread is, in fact, statistically not significant.

It is interesting to note that none of the curves indicate a tendency for the pilots to overcompensate for drift. All the end results are on the under- compensating side of zero.

Figure 9 shows a time history of lateral displacement from the runway centerline. The upper curves, the averages over the runs with no crosswind, again show almost no difference, either among scope displays or between scope displays and visual simulation. There is an average error of about 35 feet at breakout on the ILS-visual runs, but the spread at touchdown is about 8 feet.

The lower curves show the effect of crosswind on lateral displacement.

Obviously, the pilots flew differently with the ILS-visual simulation from the way they did with the scope displays. On the ILS-vlsual runs, breakout occurred at about 28 seconds. It appears that the average error at the end of the ILS portion of the approach was about 80 feet, in the direction the crosswind tended to carry the airplane away from the centerllne. At breakout, the pilot detected the misalinement and was able to correct it easily enough by using the televised picture. On the other hand, with the scope displays the pilot was able to stay more nearly alined through most of the approach and did not have to make any appreciable correction. Again, the differences at touchdown were insignificant.

Several other parameters were measured during the experiment, but these have yet to be examined in detail. The brief analysis which has been made suggests that the same trends which were apparent in the rate of descent, the drift angle, and the lateral displacement can be expected: essentially no average difference among the scope displays, and little practical average dif- ference between scope displays and ILS-visual simulation.

It is surprising that there was no difference among the eight configura- tions of the scope displays. One would expect that, as the explicit information provided by the velocity circle, the altitude indicator, and the ghost airplane was removed, the pilots' performance would deteriorate. On the other hand, all the information provided by those three elements is still in the display when it contains only the horizon llne and the four runway lines. It is apparent, from the results and also from the pilots' comments during the experiment, that they learned to use this implicit information. Also, while using the complete dis- play they learned to recognize the behavior of the runway symbol, relative to the horizon and to the scope frame, in successful landings. They could success- fully reproduce that behavior, on the average, even with the three elements missing. However, the pilots do not believe that this would carry over into even the most mildly stressful situation in flight - and perhaps not even into an unfamiliar simulation.

Remember that all the results shown here are averages. It may well be that even though pilots can, on the average, do as well with the minimum scope display as with the most complete, perhaps their performance is less consistent from run to run. Therefore, the data will also be analyzed to show the varia- bility from run to run, with the eight display configurations and the ILS-visual simulation.

The conclusions reached here are tentative, both because they are the results of simulator tests and because the data analysis is not yet complete.

Whether or not they hold up in flight will be finally determined in the research airplane.

The display described in this paper is not hnique. Other similar displays are described in references l, 2, and 3. Reference 3 is a comprehensive dis- cussion of the kinds of information which must be contained in a landing display and the way in which this information should be presented. It contains a com- parison in particular of five existing and proposed displays and makes recom- mendations for flight and simulator test programs. Reference 3 is recommended reading for anyone interested in flight displays in general, and landing dis- plays in particular.

EQUIPMENT FOR FUTURE FLIGHT RESEARCH The flight research part of the program will be conducted in a Convair 340 airplane. This airplane is being prepared as a general-utillty flying labora- tory, especially for conducting research on the required characteristics of sensors, computers, and displays for airborne zero-zero landing systems.

Figure l0 shows briefly how this research airplane is being equipped. The system is basically a distance-measuring system (EMS). Because current state of the art does not permit a completely airborne system, there will be three DMS transponders on the ground at known positions relative to the runway. The air- borne DMS components interrogate the transponders; and the airborne computer, using the DMS signals, calculates the position and rate of change of position of the airplane relative to the runway. These data, together with airplane attitude information fed to the computer from onboard sensors, are used to com- pute the display. The display is generated on a cathode-ray tube (CRT), and the reflection of it in a mirror is viewed by the pilot. When the airplane is near touchdown, the position information from the DMS is augmented with height information from a radar altimeter. References 4 and 5 contain detailed descriptions of this research equipment and its operation.

At present the DMS, the computer, and the display-generating equipment are being tested on the ground and prepared for installation in the airplane. The support equipment has already been installed and checked in flight. When this flying laboratory is completed, Ames will have a flexible research facility for in-flight studies of display concepts as well as of the attendant sensing and generating equipment. The first program to be undertaken on this flying labor- atory is the one which has been described in this paper.

REFERENCES i. Noxon, Paul A.: Microvision. Eclipse-Pioneer Div., Bendix Corp., July 1964.

2. Snodgrass, Reuben P.: Head-Up Display - A New Form of Avionics. Sperry Eng.

Rev., vol. 17, no. 3, Fall 1964, pp. 10-19.

3. Baxter, J. R.; and Workman, J. D.: Review of Projected Displays of Flight Information and Recommendations for Further Development. Human Eng.

Rept. 2, Aeron. Res. Lab., Australian Defense Sci. Service, Aug. 1962.

4. Robinson, Gilbert G.; and Johnson, Norman S.: An Airborne Zero-Zero Landing Laboratm_j. Presented at Inst. Navigation, Low-Cost Navigation Symposium (Los Angeles, Calif.), Nov. 6-7, 1963.

5. Robinson, Gilbert G.; and Johnson, Norman S.: Subsystem Requirements for an Airborne Laboratory To Study Zero-Zero Landing Systems. Presented to 25th Meeting of Flight Mechanics Panel of AGARD (Munich, Germany), Oct. 12-14, 1964.

BASIC ELEMENTS OF THE DISPLAY DISPLAY WITH CENTERLINE INDICATOR CENTERLINE INDICATOR ("GHOST" CIRCLE AIRPLANE) VERTICAL INDEX Figure i Figure 2 THE COMPLETE DISPLAY INCLUDING ALTITUDE INDICATOR TWO VIEWS OF DISPLAY SHOWING DIFFERENT ANGLES OF APPROACH SAME DISTANCE OUT LTITUDE INDICATOR SHALLOW ANGLE STEEP ANGLE Figure 3 Figure 4 THE INTERIOR OF THE SIMULATOR CAB THE AMES TRANSPORT LANDING SIMULATOR 9SCOPE PICTURE /_._ PROJ ECTE D "_ SIMULATOR "_ _ CAB TV Figure 6 Figure 5 TIME HISTORIES OF AVERAGE DRIFT ANGLE DURING TIME HISTORIES OF AVERAGE RATE OF DESCENT DURING LAST 5Osec BEFORE TOUCHDOWN LAST 50 sec BEFORE TOUCHDOWN OF RUNS AVERAGES WITH NO OF RUNS (o I0 CROSSWtND _ - CROSSWlND u.2 WITHNO _ _ --VISUAL SIMUL AT_N AVERAGES ,F- " OF RUNS _ AVERAGES (_ 12 WtTH 4 CROSSWlND OF RUNS _ IO CROSSWIND WITH _ 2 -2 25 20 15 I0 5 0 30 25 20 15 I0 5 0 TIME BEFORE TOUCHOOWN, sec TIME BEFORE TOUCHDOWN, sec Figure 8 Figure 7 7O TIME HISTORIES OF AVERAGE LATERAL DISPLACEMENT FROM RUNWAY CENTERLINE DURING LAST 5Osec BEFORE TOUCHDOWN 4O AVERAGES OF RUNS zW v ...... _ ........ r ........ T , W_TH NO w, 40 _ , CROSSWlND w&c --v,soAL S,.0LAT_ J _= OSSWIND '_ z 80 AVERAGES _m w_ OF RUNS WITH CROSS'WIND °L ' ' '_ 4 -40 ! I L I I L / 30 25 _0 IS I0 5 0 TiME BEFORE TOUCHDOWN, ,SeC Figure 9 AIRBORNE SYSTEM FOR ZERO-ZERO LANDING RESEARCH DMS LINKS DMS RUNWAY DMS RAOAR ALTIMETER i Figure lO FACTORS RELATING TO THE AIRPORT-COMMUNITY NOISE PROBLEM .

By Harvey H. Hubbard, Jimmy M. Cawthorn, and W. LathamCopeland NASA Langley Research Center SUMMARY Factors relating to the airport-community noise problem are discussed herein. The noise problems are associated with aircraft landing approaches, take-offs, and climbouts in communities near commercial airports. The main objectives are to identify the significant factors involved in the problem, to define some of their interrelationships, and to present recent related research information from NASA in-house and contract studies. Changes in the types of aircraft power plants, the aircraft operating procedures, and the community itself are all judged to be important; however, these changes should not be made without understanding the associated human-response implications.

INTRODUCTION Factors relating to the airport-community noise problem which are discussed herein are illustrated in figure 1. These factors relate to noise associated with landing approaches_ take-offs (ground operations), and climbouts in com- munities near commercial airports. The main objectives of this paper are to identify the significant factors involved in the problem, to define some of their interrelationships, and to present recent research information from NASA in-house and contract studies.

The significant factors in the airport-community noise situation are pre- sented in figure 2. At the heart of the problem, as indicated in the center of the figure, is the fact that humans respond unfavorably to noise. Since there is very little possibility of such human responses changing markedly in the near future_ it is important that aircraft operators understand these adverse responses and, particularly, what features of the noise are most objectionable.

There are several means of obtaining an overall solution, all of which fall into the three categories identified by the shaded areas in figure 2. (See also refs. 1 and 2.) Noise sources refer_ of course, to aircraft power plants and involve both the type of engine cycle and its operating conditions. Aircraft operations include such factors as aircraft gross weight_ flight path_ fre- quency of operations, preferential runways, and time of day. Community consid- erations involve such items as size and location of the areas affected, type of activity existing in the community, the social structure, background noise level, and types of construction of the buildings.

The main objective is, of course, to obtain a more favorable human response to the noise. This objective might be achieved by means of any one of the fac- tors in figure 2, or preferably by some combination of them. It should be noted that no physical changes should be made_ however, without an understanding of the associated human-response implications.

There is first a review of some of the recent results of subjective response studies performed under contract by Bolt_ Beranek & Newmanj Inc.

(refs. 3and 4). Then there is a brief discussion relating to each of the physical factors represented by the shaded areas of figure 2.

HUMAN RESPONSE The perceived noise level which is a measure of human response to noise (see refs. 3 and 4) is at present widely used, both in this country and abroad.

One of the basic concepts of this response rating procedure is that high- frequency noise is generally more objectionable than low-frequency noise.

Spectrum shape is thus an important influence in human reaction to noise. This is especially important where discrete frequencies are involved, as illustrated in figure 3 which is taken from unpublished work done under contract to Bolt, Beranek & Newman, Inc. The basic spectrum at the bottom of the figure has a dlscrete-frequency component superposed on the broad-band random noise. Such a discrete frequency might be produced by the compressor of a jet engine and, for the case of the illustration, extends about lOdB above the background (basic) noise level of the octave band in which it resides. The net effect of such a pure tone is to increase the percelved-noise-level value noticeably. For the example shown, the net effect of the single pure tone is the equivalent of a 6 dB across-the-board increase in all octave bands of the basic spectrum, as illustrated in the curve at the top of the figure. Since pure tones have an adverse effect on acceptance, much research effort is being directed toward eliminating such noise components during aircraft operations near airports.

(Decibels are referred to 0.0002 dynes/cm2.)

There has recently become available some information on the effect of time duration of the noise on its acceptability. (See ref. 3.) This effect is illustrated in figure 4 for the flyover-noise situation. The two noise-level time history plots at the bottom of the figure might represent the noise expo- sures at a community observation point for two different flyover conditions, as illustrated in the sketch at the top of the figure. On the left, path A is shown as an example for which the time t (measured between the points that are lO dB down from Zhe maximum levels) is relatively short (about l0 seconds), but the peak noise levels are higher than those for path B on the right where the time is 2t. Laboratory experiments under controlled conditions suggest that the longer exposure time is detrimental. For path B to be equally accept- able, it would be necessary to reduce the peak noise level by about 4. 5 dB to compensate for doubling the time exposure. The two noise-exposure time his- tories shown in the figure are thus Judged to be about equally acceptable.

NOISE REDUCTION AT THE SOURCE Jet Exhaust Noise One of the main objectives of research on exhaust-noise reduction is to find acceptable methods of producing less noise per unit thrust. The fan 3 or bypass, engine offers the possibilities for lower exit velocities and hence lower noise levels. (See ref. 5.) The data of figure 5 are included to show the ranges of perceived noise levels associated with both current and proposed engines. Maximum perceived noise levels at a sideline distance of 200 feet have been estimated for all engines by the methods of references 6 and 7. These levels have been arbitrarily normalized to a thrust of 21_000 pounds for com- parison and, for convenience, have been plotted as a function of bypass flow ratio.

Data for current turbojets are plotted at the zero location on the abscissa scale and fall in the narrow hatched region. Current turbofans are represented by the larger hatched region and are seen to have somewhat lower perceived noise levels. Some,proposed high-bypass-ratio engines are represented by the stippled region at the right, and it is seen that by this means, substantially lower perceived noise levels may be realized. It should be noted that this lower boundary is well defined, but the upper boundary is not well defined.

The vertical extent of the stippled region represents some of the uncertainties regarding the importance of Jet-stream turbulence, combustion noise, compressor noise, and turbine noise for these proposed engines. Research effort is being focused on minimizing the noise from these sources so that the full potential of these engines can be realized.

Compressor Noise Several recent research studies (refs. 8 to 14) have been directed toward ways of minimizing the discrete-frequency compressor noise radiation from inlets. Samples of the results obtained are shown in figure 6. These studies have involved changes to the inlet geometry, the compressor geometry, and the inlet flow conditions. Noise-level reductions (i.e., the differences bet_en those for the modified compared with those for the unmodified configuration in each case) are shown for various azimuth angles in one of the front quadrants.

The use of resonators arranged in a peripheral array around the inlet provided the results shown on the left. For the case where the spacing between the rotor and stator of the front stage was changed, a rather complex pattern of noise reduction was obtained, as indicated by the center diagram. On an inlet for which it was possible to establish choked-flow conditions, the greatest noise reductions were obtained as shown in the right-hand diagram.

The results are summarized briefly in figure 7 by means of a bar graph which illustrates the maximumnoise reductions obtained for each of four dif- ferent approaches. The first three of these approaches represent the results obtained in NASA research studies# whereas the last item represents the results achieved with configurations suitable for application to current aircraft designs. The ranges of noise reductions obtained with resonators, spacing changes_ and inlet flow choking are generally consistent with the results of other known experience. Results have varied from one particular investigation to another, and thus the conclusion is drawn that such results are sensitive to configuration details. The experience with absorptive treatment is also believed to be strongly influenced by the configuration for which the work was done_ since there is reason to believe that configurations having longer duct lengths available for treatment would produce larger noise reductions than those illustrated in the figure.

It is not known whether or not some of these modifications may be success- ftully used in combination. Contract research studies are under way for the purpose of optimizing compressor-noise-reduction procedures for a full-scale engine nacelle.

AIRCRAFT OPERATIONS Landing Approach With regard to aircraft operations, one of the most difficult aspects is related to the landing approach, during which time the airplane flight path is shallow and the aircraft is quite close to the observers on the ground. One obvious scheme to improve this situation would be to use a steeper approach-path angle as shown in figure 8. Some related performance studies are described in paper no. 4 by Hall 3 Champine, and McGinley, and some of the noise-level time histories for a fan-powered transport aircraft are shown in figure 8. The solid curve represents a conventional approach at a 3 ° geometric angle_ whereas the dashed curve applies to a 6 ° approach. The data were obtained at a measuring station about 6900 feet from the end of the runway. An obvious beneficial result of the steep approach is that the maximum noise levels are lower because of the increased distance of the aircraft and the somewhat lower power setting.

There is a trend noted, however, that the aircraft is audible for a somewhat longer total time during approach at the higher angle probably because of the more favorable sound-propagation-path conditions at this higher altitude.

Take-off and Initial Climb There has been much discussion also about the use of steeper initial climb rates in order to increase the distance to the aircraft and thus reduce the noise on the ground. In the plot at the top of figure 9 are shown schemati- cally two altitude-distance profiles. The lower profile is representative of the operation of a current, long-range, fan-powered aircraft. The upper pro- file, on the other hand, represents a hypothetical high-performance aircraft of about the same gross weight but with twice the installed thrust and with advanced-design fan engines.

At the bottom of the figure the plot shows the estimated i00 PNdB contours on the ground for the associated take-off operations. These contours are plotted at appropriate longitudinal and lateral distances with reference to the runway. The reference rectangle of dashed lines is drawn at a 1/2-mile distance in all directions from the runway, and therefore is representative of the size of some commercial airports.

The current airplane 3 because of its lower thrust-to-weight ratio 3 has a longer take-off run_ a slower climb rate 3 and thus an associated ground-noise contour that is relatively long and slender. The hypothetical high-performance aircraf% on the other hand, has a shorter take-off run 3 a more rapid climbout_ and an associated ground-noise pattern that is relatively short in the take-off direction as indicated by the shaded area. A shaded area is included to show the range of results obtainable, depending on the exhaust conditions of the proposed engine. For a high-velocity exit (duct heating), the outer extremes of the area apply, whereas for a high-bypass-ratio engine without duct heating, the inner extremes apply. For this latter condition_ the perceived noise levels above i00 PNdB all occur within the airport boundaries.

COMMUNITY COMPATIBILITY CONSIDERATIONS It is obvious that normal activities in many communities near airports are not, for one reason or another, compatible with the noise exposures from air- port operations. (See ref. 15.) Some changes in the noise characteristics of the power plants or in the manner of aircraft operation, as have been discussed, might be beneficial in reducing noise. Of equal importance to the overall com- patibility problem are possible changes in the community itself to better adapt itself to its environment. Some of the possible considerations in such a com- munity undertaking could be: zoning, tax incentives, financial assistance, building and housing codes, land acquisition, eminent domain, and urban renewal.

Each of these considerations is not discussed, but it should be pointed out that they have already been used successfully to bring about desired changes in some specific situations. A contract study is currently under way for the purpose of gathering together the latest information relative to the application of these procedures to communities with airport noise problems. The use of such instruments by the proper authorities as part of a sound overall plan for com- munity development could be very beneficial and would complement other efforts to minimize noise exposures.

CONCLUDING REMA_I_S Three types of approaches to solving the airport-community noise problem which are closely interrelated with the phenomena of human responses are dis- cussed. These approaches include changes in the types of aircraft power plants, the aircraft operating procedures, and the conmnmity itself. Because there are many facets to this problem 3 a coordinated, multipronged approach to a solution is required.

_CES 1. Tyler, John M. : A New Look at the Aircraft Noise Problem.

EPreprint_ 9]-IB, Soc. Automotive Engrs. 3 Oct. 1964.

2. March, Alan H.; and McPike, A. L.: Noise Levels of Turbojet- and Turbofan- Powered Aircraft. Sound, vol. 2, no. 5, Sept.-Oct. 1963, PP. 8-13.

3. Kryter3 Karl D.; and Pearsons 3 Karl S.: Some Effects of Spectral Content and Duration on Perceived Noise Level. NASA TN D-18733 1963.

4. Kryter, K. D._ and Pearsons, K. S.: Modification of Noy Tables. J. Acous.

Soc. Am. (Letters to Ed.), vol. 363 no. 2, Feb. 19643 pp. 394-397.

5. Kantarges3 George T._ and CawthOrn 3 Jimmy M.: Effects of Temperature on Noise of Bypass Jets as Measured in the Langley Noise Research Facility.

NASA TN D-2378, 1964.

6. Anon.: Definitions and Procedures for Computing the Perceived Noise Level of Aircraft Noise. ARP 8653 Soc. Automotive Engrs. 3 Oct. 153 1964.

7. McPike, A. L. : Recommended Practices for Use in the Measurement and Evalu- ation of Aircraft Neighborhood Noise Levels. FPreprint_ 650216, Soc.

Automotive Engrs., Apr. 1965.

8. Tyler, J.M. and Sofrin, T. G. : Axial Flow Compressor Noise Studies.

_reprint_ _345D, Soc. Automotive Engrs., 1961.

9- Copeland, W. Latham: Inlet Noise Studies for an Axial-Flow Single-Stage Compressor. NASA TN D-2615, 1965.

i0. Bateman 3 David A.; Chang 3 Stephen C.; Hulse, Bruce T.; and Large, John B.: Compressor Noise Research. Tech. Rept. ADS-S1, FAA, Jan. 1965.

ii. Carmichael, R. F.: Analyses of Dynamic Measurements Obtained During the J79-13 Engine Installation Test for the NB-66D Laminar Flow Control Air- craft. Rept. NOR-62-163, Norair Div., Northrop Corp. 3 July 1962.

12. Copeland, William L._ and Crigler, John L.: Rotor-Stator Interaction Noise Studies of a Single-Stage Axial-Flow Research Compressor. NASA paper presented at Sixty-Eighth Meeting of the Acoustical Society of America (Austin 3 Texas), Oct. 21-243 1964.

13. Bragg, S. L.; and Bridge, R.: Noise From Turbojet Compressors. J. Roy.

Aeron. Soc., vol. 683 no. 637, Jan. 1964, pp. 1-10.

14. Advanced Eng. and Technol. Dept., Gen. Elec. Co.: Research Investigation Into Fan Noise Prediction and Alleviation. Contract No. DA 44-177- AMC-72 (T), U.S. Army Transportation Res. Command, Sept. 1964.

15. Anon.: Land Use Planning Relating to Aircraft Noise. Tech. Rept., Bolt, Beranek & Newman, Inc. _ambridge, Mass.l, Oct. 1964.

SIGNIFICANT FACTORS AIRPORT-COMMUNITY NOISE GROUND LANDING APPROACH OPERATIONS CLIMBOUT

L

HUMAN RESPONSES ./ COMMUNITY Figure i Figure 2 EFFECT OF DISCRETE TONE NOISE LEVEL, dB

',ii f

6dB BASICJ T 80 L -- I I I I I I PURE-TONE COM_

iii f

BASIC ,,-_ 80 _ I I I I I I I I 75 300 1200 4800 TO TO TO TO 150 600 2400 9600 FREQUENCY BANDS, CPS Figure 3 EFFECT OF ENGINE BYPASS RATIO ON PERCEIVED NOISE LEVEL ESTIMATED FOR 21,O00-LB THRUST EFFECT OF TIME DURATION 200 FT /B / MAXIMUM OBSERVER PNdB OBSERVER PROPOSED NOISE 140 1 CURRENT LEVEL PATH B PATH A

I N

1201- II0 t ,0% I SECONDARY FLOW RATE i I I I i I I I I I I I PRIMARY FLOW RATE TIME OF EXPOSURE Figure Figure 4 COMPRESSOR NOISE REDUCTION MEASUREMENTS 0 o 0 ° 0 o 1 1 1 (30 0 ° 0 ° 0 I0 20 0 I0 20 0 I0 20 NOISE LEVEL REDUCTION, dB Figure 6 8O LANDING-APPROACH NOISE MEASUREMENTS COMPRESSOR NOISE REDUCTION RESONATORS MEASURING STATI_ /--RUNWAY ///llll/i//,.,y/i///////////////////////////1 -- NOISE LEVEL, J-'_6,900 FT_ dB ROTOR-STATOR SPACING :_oAPPROACH ANGLE INLET CHOKING

,oo[

'.U''OH --"

ABSORPTIVE TREATMENT OI I I I _,1 I I o _ 6 i; 24 16 8 0 8 16 24 TIME, SEC NOISE LEVEL REDUCTION, dB Figure 7 Flgure 8 GROUND NOISE PATTERNS AIRCRAFT GROSS WEIGHT = CONSTANT ALTITUDE - DISTANCE PROFILE 4,000 F "HHHHN_H_/_N/HHHN_.H_H_H_HHH_/HHNHN_/H_H_HHH_/_ IOO PNdB CONTOURS I [- / ,_/_HYPOTH E TICAL LATERAL DISTANCE, ST MILES _-t ..... _ "-CURRENT F / DISTANCE FROM START OF TAKE-OFF ROLL, ST. MILES Figure 9 10. PRELIMINARY MEASUREMENTS OF TAKE-OFF AND LANDING NOISE FROM A NEW INSTRUMENTED RANGE By Carole S. Tanner and Norman J. McLeod NASA Flight Research Center SUMMARY This paper describes the NASA noise-survey instrumentation system pres- ently in use at Edwards Air Force Base, California, and presents preliminary noise data from an F-104 airplane. Also presented are noise measurements of the XB-70 and 707-131B airplanes obtained with essentially the same equipment at another location. The difference between measured noise levels for the XB-70 and 707 is illustrated and comparisons of perceived noise levels are made. The adequacy of noise predictions is discussed briefly.

INTRODUCTION Excessive noise in and near airports has become an increasing problem with the development of commercial jet airplanes. Considerable research on this noise problem has been undertaken in the past but the results have been less than conclusive. Some factors compounding the problem are differences in engine installations, engine types, wind, temperature, humidity, and airplane operating parameters such as flight path and velocity.

The relationship between these factors and noise from advanced engine/airframe configurations presents the airplane designer and operator with serious difficulties since safe and economically feasible operating con- ditions must be combined to obtain acceptable noise levels. In view of the availability of the most advanced airplanes at Edwards Air Force Base, a noise study under actual operational conditions has been undertaken by the NASA Flight Research Center with the assistance of the Air Force Flight Test Cen_er.

This paper describes an acoustic measuring system installed along the main run- way at Edwards Air Force Base and presents some of the initial data obtained from this facility. In addition, XB-70 and 707 noise data obtained with essen- tially the same equipment temporarily installed at Palmdale, California, are presented.

DESCRIPTION OF TEST INSTRUMENTATION Noise data for the F-104 were obtained at Edwards Air Force Base, California; XB-70 and 707 data were obtained at Palmdale, California. The Edwards instrumentation layout is shown in figure 1. Twelve microphone stations are positioned sy_metrically in pairs at 5000-foot intervals along and beyond the runway. Total distance between microphone pairs varies from 310 feet to i000 feet. Each microphone station consists of a i/4-inch microphone/cathode follower unit_ an integrated microphone power supply and line-driving amplifier unit, and a battery operated ac power supply. The noise measured by each microphone is amplified and transmitted via an underground cable system to a tape recorder in the van.

_ne i_strtunentation layout at PaLmdale 3 shown in figure 23 consists of eight NASA microphone stations (circles with numbers) and three microphone sta- tions (circles with letters) provided by North American Aviation. The NASA microphone stations, the same as those used at Edwards, were installed along the 200-foot-wide, 12,000-foot-long runway. The three North American micro- phones were located 3 miles from airplane brake release. Two separate cable systems and recording vans were used.

The Edwards instrumentation for data acquisition and reduction and the Palmdale system were completely electrically and acoustically calibrated. The only deviation was the extrapolation of system response to 31.5 cps. Typical response characteristics shown in figure 3 indicate that system frequency response is flat within 1.8 dB and 2.5 dB for the Edwards and Palmdale systems, respectively. Corrections applied to data obtained with the Edwards instrumen- tation and the eight NASA microphones at Palmdale are for system response and the energy acceptance of the data-reduction filters. Evaluation of these cor- rections indicates that these data are accurate to ±1.5 dB. Unfortunately, complete system and data-reduction corrections are not available for the three North American microphones used at Palmdale; however, the data are considered accurate to within ±3 dB. (Decibels are referenced to 0.0002 dyne/cm2.)

SYSTEM TESTS The Edwards system was placed in operation in December 1964. Initial operation consisted of repeated test flights using a single-engine F-I04 air- plane to determine the extent and effect of space-positioning accuracy and data repeatability. Noise data from the test flights were time correlated with air- plane location which was obtained by using Air Force radar. Evaluation of the test results indicates that the location of the aircraft in relation to the microphones could be determined within the limits of a box having a base of 14 feet, a height of 16 feet, and a length of 50 feet. Since skin tracking was used, larger airplanes may have a larger position error.

The F-104 noise-data repeatability is shown in figure 4. The data are from four consecutive flybys having the minimum ranges between airplane and microphone for the weather conditions shown.

To minimize near-field effects, data from microphone 3 (fig. i) were selected for presentation. The octave band sound pressure levels show excellent repeatability up to 500 cps except for the one point at 250 cps. Data repeata- bility is not as good at the higher frequencies where a scatter of ±3 dB occurs.

Since higher frequencies would be affected by slight changes in engine thrust, the larger variation of those frequencies is attributed in part to changes in engine thrust as gross weight decreased from one flyby to the next. However, the variation in noise at the higher frequencies is not believed to be signif- icant, and since the overall sound pressure levels agree quite well, data repeatability is considered satisfactory for all octave bands.

XB-7OAND 707 NOISE In an effort to provide take-off noise data from the largest supersonic airplane available, preliminary data were obtained from the XB-70.

The data presented were obtained on one of the earlier flights which origi- nated from Palmdale, California. In order to obtain the data at the earliest possible time the Palmdale layout (fig. 2) was used.

In order to help place XB-70 noise in proper perspective, arrangements were made with the Air Transport Association for a 707 fan-jet airliner to operate from the Palmdale runway approximately i hour after the XB-70 take-off.

Noise data presented for these two airplanes are for the take-off profiles and weather conditions shown in figure 5. lllustrated are the actual take-off pro- files of the XB-70 and 707 and representative take-off profiles of the super- sonic transport (SST) obtained from NASA Technical Note D-423 which is an early summary report.

The XB-70 take-off was made with all engines at maximum afterburner thrust, but the programed rate of climb was rather low. The 707 take-off is typical of operation from a commercial airport. It should be noted that the SST will have different engines and higher climb profiles resulting in noise levels lower than those of the XB-70.

The magnitude of the noise levels measured from the XB-70 and 707 for these take-off profiles is illustrated in figures 6 and 7- Figure 6 shows the noise spectra along the runway for the XB-70 and the 707. Data from microphones 3 and 8 (fig. 2) are shown.

The XB-70 maximum spectrum levels occur in the 125 cps band. The spectrum from microphone position 3 had a higher noise level than the spectrum from microphone 8. This difference is between 20 and 2_ dB over all frequencies.

The inverse-square-law attenuation for the difference between the microphone locations is approximately 9 dB; therefore, it is concluded that microphone 3 is in the near field of the XB-70 engines.

The 707 data at the bottom of figure 6 show that the maximum spectrum levels occur in the 12_ cps band. The spectrum level at microphone 3 is higher than the level at microphone 8 and the difference is approximately lO dB. This noise-level difference approaches the inverse-square-law attenuation, which leads to the conclusion that both microphones were in the far field of the 707 fan-jet engines. Far-field noise data are more consistent than near-field data, therefore the difference between noise along the runway is best determined by comparing the spectra from microphone 8 for each airplane.

The XB-70 spectrum is, on the average, i} dB noisier than that for the 707.

This difference is due to the larger number of engines as well as the higher thrust of the XB-70. The remarkable similarity between the spectrum shapes was not expected and more research is needed before the exact reasons for this sim- ilarity can be determined.

The airplane noise at the 3-mile point is also of interest from the stand- point of community response to noise. A direct comparison of the noise spectra of the XB-70 and 707 airplanes at the 3-mile point is shown in figure 7. Spec- tra for both airplanes at similar slant ranges show that the XB-70 noise levels are considerably higher than those of the 707. It is significant that the difference between these two spectra is an average of 25 dB for all fre- quencies, whereas the difference in spectra along the runway was 15 dB (fig. 6).

The larger difference in spectra at the 3-mile point indicates that further research is needed to ascertain the effects of airplane attitude, source size, and airplane motion on the noise produced.

PERCEIVED NOISE LEVELS The noise spectra at the 3-mile point, as well as the spectra along the runway_ can be used to calculate perceived noise levels. The perceived noise level is a judgment of the noisiness or annoyance of a sound and is obtained by weighing the octave band levels with factors determined from the response of the human ear. Perceived-noise-level calculations were made by using the conversion tables in reference 1 and the following formula: where total noisiness_ noys noisiness in noisiest band_ noys sum of noisiness of all bands_ noys ZN N Perceived noise levels for the XB-70 and the 7o7 are shown in figure 8.

The circles represent the positions for which perceived noise levels were cal- culated.

The perceived noise levels for the XB-70 are higher in all cases than are those for the 7o7. The point of interest is the comparison of the magnitude of perceived noise levels from a commercial airliner operated to be quiet with a military airplane whose design and operation did not consider the alleviation of noise as a prime factor. The question of whether or not the XB-70 could be operated in a manner resulting in perceived noise levels comparable with those of the 707 is at present unknown.

NOISE PREDICTIONS The peTceived noise levels shown are calculated from data obtained by a frequency analysis of the measured overall sound pressure levels. However, if no measured data were available, a designer would have to use the methods pre- sented in reference 2 or 3, for example, to predict the overall sound pressure levels. The following table shows a comparison between the predicted and measured overall sound pressure levels for the XB-70: Predicted noise Predicted noise from proposed from Handbook Measured Slant range, Maneuver SAE method, of Noise Control noise, ft (ref. 3), (ref. 2), dB dB dB Take-off roll 141o 121.5 118.5 (1-mile point) Climbout i19.5 (3-mile point) The XB-70 noise prediction is based on a single static turbojet engine with afterburning.

The predicted overall sound pressure level was corrected for six engines and divergence. The results from the table indicate that there are small dif- ferences between the predicted noise levels for both airplane positions. The measured levels are lower than the predicted ones. It should be noted that the predicted results, in this case, were based on effective nozzle area and its associated flow velocity. It would seem that in the final analysis, the designer or operator must presently resort to flight-testing to obtain a true indication of aircraft noise levels and to check on the noise predictions.

CONCLUDING REMARKS The data presented in this paper illustrate some of the problems involved in accurately determining engine noise for flight conditions. The acoustic- noise measuring system operated by the NASA Flight Research Center has been developed to acquire data useful in ascertaining the relative importance of the various parameters involved. Present program plans involve the measurement of noise from advanced aircraft and engine configurations. Data from these studies will aid in the establishment of better prediction methods and will help in fur- ther assessing the relationship between airplane operating parameters and the noise produced.

REFERENCES i. Kryter, K. D.; and Pearsons, K. S.: Modification of Noy Tables. J. Acoust.

Soc. Am., vol. 36_ no. 2_ Feb. 1964, pp. 394-397.

2. Von Gierke, Henning E.: Aircraft Noise Sources. Handbook of Noise Control_ ch. 33, Cyril M. Harris, ed., McGraw-Kill Book Co., Inc., 1957, PP. 33-56.

3. McPike, A. L.: Recommended Practice for Use in the Measurement and Evalua- tion of Aircraft Neighborhood Noise Levels. EPreprint] 650216, Soc. of Automotive Engrs., Apr. 1965.

NOISE-SURVEY INSTRUMENTATION LAYOUT NOISE-SURVEY INSTRUMENTATION LAYOUT EDWARDS MAIN RUNWAY - ELEVATION 2300 FT PALMDALE, CALIF.-ELEVATION 2200 FT OFF NDING "5000 F (TYPIc_ J _ I_ IO00FT

/ \--.OORD,°°+A.S

IOOO FT FROM RUNWAY CENTERLINE jN jN Figure 2 Figure i F-104 DATA REPEATABILITY TYPICAL SYSTEM RESPONSE MICROPHONE 3 EDWARDS SYSTEM MINIMUM RANGE, WEATHER FLYBY SYSTEM RESPONSE, dB 0 I E65 RELATIVE HUMIOITY = 21 PERCENT (ZERO AT 4000 CPS) _...O...._ O 2 585 TEMPERATURE : 57 D F 12OI- O 3 540 WIND VELOCITY = 3 KNOTS 4 600 I I I MAXIMUM 110 • OCTAVE RAND LEVEL, dB . o 6 SOUND PRESSURE I_ L 80 _ @ 1_ o (ZERO AT 1,000 CPS) _ _'--_'_--C_(3_

SYSTE__c il __sYE

[ / OVERALL 31.5 63 125 250 500 10OO 2000 4000 I I I I I I I I I I FREQUENCY, CPS r 31.5 63 125 250 500 I 000 2,00D 4,000 8,0D0 16,O00 FREQUENCY, CRS Figure Figure 3 TAKEOFF NOISE SPECTRA ALONG RUNWAY TAKEOFF PROFILES M'C_OPHONt 3 MAXIMUM 130 _s IRELATIVE HUMIDITy = 30 P_RCENI IEMPERATURE = 45 _ F OCTAVE BAND 120 • _M_CIOPHONE e 1001401 • _'7° SOUND PRESSURE 110 WLNO VELOCITY = S KNOTS LEVEL, dB HEIGHT ABOVE NASA IN D.423 T PROFIL E RUNWAY, FT 1000 1707 FAN JET MAXIMUM 70;'

f OCTAVE SAND 110 • MlCltOP_Ot,/E 3

SOUND PRESSURE 100 • O I _ 2 3 LEVEL, dB 90 DISTANCE FROM BRAKE RELEASE, MILE 80 OVERALL _ 31.5 63 125 250 500 1000 2000 4000 8000 FREQUENCY, CPS Figure Figure 6 PERCEIVED NOISE LEVELS COMPARISON OF Xfi-70 AND 70Z NOISE AT 3-MILE POINT X6-70 I10 MAXIMUM OCTAVE BAND 100 SOUND PRESSURE LEVEL, dB 90 707 107 PNdB _ _ 60 SLANT R_E 1230 FT TO , z _ I I I f r f e_. XB-70 130 PNdB OVERALL 53 106 212 425 850 1700 3400 6800 FREQUENCY, CPS _X 707 101 PNdB B-Z0 125 PNdB 707 100 PNcIB Figure 7 Figure 8 9o ll.

OPERATIONAL EXPERIENCES OF TURBINE-POWERED COMMERCIAL TRANSPORT AIRPLANES By Paul A. Hunter and Walter G. Walker NASA Langley Research Center SUMMARY Recent results obtained from NASA V-G and VGH recorders installed on com- mercial turbine-powered transports have indicated that exceedances of placard speeds appear to have been significantly reduced since the placard speeds have been redefined and changes made in the aural warning. Oscillatory accelera- tions and unusual events_ such as large or rapid departures from the planned flight profile_ occur less frequently. Landing-impact accelerations are higher for turbine transports than for piston transports and vary with operator. The total in-flight acceleration experiences for turbine transports, however_ are not significantly different from those for four-engine piston transports.

INTRODUCTION The NASA V-G/VGH program was initiated on turbine-powered transports con- currently with their introduction into commercial service. A summary of pre- liminary results of the program was presented at meetings with various segments of the aviation industry in 1961. At that time, some concern was expressed in regard to airspeed practices relative to placard speeds, aircraft oscillations, landing impact accelerations, andunusual events. These preliminary results, including information on in-flight accelerations and gust velocities_ were pub- lished in 1962. (See ref. 1.) The purpose of this paper is to present infor- mation obtained since that time.

INSTRUMENTATION AND SCOPE In the program_ information on operational experiences of turbine aircraft is obtained by means of NASA V-G and VGH flight recorders. The V-G recorder (ref. 2) provides an envelope of airspeed and normal acceleration experience, whereas the VGH recorder (ref. 3) provides time histories of airspeed_ normal acceleration_ and altitude.

The scope of the program is indicated in table I which shows the number of turbojet and turboprop airplanes on which VGH and V-G recorders were installed and the number of airlines involved. The turbojet airplanes were the Boeing 707_ Douglas DC-8_ and Convair 880, and the turboprop airplanes were the Lockheed Electra, Vickers Viscount, and Fairchild F-27A. In terms of the accumulated flight time of U.S. commercial turbojet and turboprop aircraft

corresponding to the closing of the various recording programs, the VGHand

V-G samples are rather small, as is indicated by the percentage values shown.

In terms of numberof aircraft, the recorded samples vary from about 4 to

6 percent of the domestic fleet. Despite the small samples in terms of the

fleet, it is believed that the careful selection of the operations sampled

result in data representative of general airline operations.

Most of the programs on the first line of turbine aircraft have been com-

pleted. However, other programs are underway or will be initiated in the near

future on the Sud Caravelle, Boeing 727_ Douglas DC-9, and BAC-111transports.

A total VGHprogram involving about 30 aircraft from both domestic and foreign

airlines is contemplated.

DISCUSSION

Placard Speed Exceedances

Early samples of VGHdata indicated that the turbine-powered transports

were exceeding certain limit speeds more frequently than the piston trans-

ports. (See ref. 4.) The particular limit speeds were: the normal operating

limit speed VNO and the never-exceed speed VNE at which an aural warning

was provided. As a result of continued overspeeds, the VNO speed was rede-

fined as a maximum operating limit speed VMO and an aural warning was pro-

vided within a 6-knot margin above this speed rather than at the higher never-

exceed speed. (See ref. 5.)

Table II showsa comparison of turbojet overspeed experience before and

after the changein aural warning. Before the change, Airline A averaged

61 flights between exceedancesof VNO , whereas Airline B averaged only

2.7 flights between exceedances. For this comparison, a flight is defined as

extending from take-off to landing. The percent of total time spent over VNO

was two one-hundreths and eighteen one-hundreths for Airlines A and B, respec-

tively. Both of these samples showbetter limit speed observance than did some

earlier samples. Most of the overspeeds occurred in descent at the lower alti-

tudes where the limit speeds are determined by dynamlc-pressure considerations

rather than by Mach number.

After the change in aural warning was made, however, no speeds over VMO ,

the maximum operating limit speed, were recorded for either airline. The

sample sizes for these overspeed analyses varied from 200 to 450 hours but for

both airlines the second sample was the larger by about 50 percent. The results

from the samples taken after the change in aural warning are highly encouraging.

Although sampling of speed practices on the older turbine-powered transports is

being phased out, the sampling will be continued on new turbojets in the short-

and medium-rangecategories.

Oscillatory Accelerations

Aircraft oscillations in the longitudinal-lateral modeproved to be one of

the unanticipated aspects of turbine operations. Oscillations seldom occurred

on piston-engine transports but apparently occur on turbine transports because

of decreased damping at the higher speeds and altitudes being flown and because

of increased sophistication of equipment. Figure i showsexamples of two types

of oscillations - one, a constant-amplitude oscillation and the other, a diver-

gent oscillation. The constant-amplitude type of oscillation was found to con-

tinue from several minutes to as muchas several hours. The divergent type of

oscillation generally lasted less than 2 minutes. The acceleration oscillations

shown in these examples have peak-to-peak values of 0.4g for the constant-

amplitude type and 0.8g for the divergent type. Other VGHsamples are available

which showeven greater variations. For example, peak-to-peak oscillatory

accelerations as high as 0.8g for the constant-amplitude type and 1.8g for the

divergent type have been recorded. Variations in the airspeed and altitude were

negligible for the constant-amplitude type of oscillation but reached values of

17 knots and 900 feet, respectively, for the divergent oscillation shownhere.

Every turbine aircraft from which VGHdata have been collected showedoscilla-

tions to exist, the frequency of occurrence varying from about 0.2 to 26 percent

of total flight time. Inasmuch as most of the occurrences of oscillations were

traced to control-system or autopilot components, they were a source of concern

regardless of amplitude. Oscillations showedup in the initial year of opera-

tion of the turbine aircraft but, as a result of a cooperative effort by air-

lines, and airframe and autopilot companiesto improve maintenance procedures

and equipment, recent data samples indicate that the amplitudes and frequency of

occurrence have been considerably reduced. Newairplanes nowbeing phased into

the recording program also experience someoscillations_ but the amplitudes and

frequency of occurrence seen thus far are fairly low.

Landing-Impact Accelerations

Turbine-powered transports have, in general, showna trend towards higher

landing-impact accelerations than the piston transports. This trend has been

confirmed by measurementof sinking speeds madefrom motion pictures of trans-

port landings. In addition, large differences have also been noted between

someoperators of turbine aircraft of the sametype. These differences are

illustrated in figure 2 which showsthe probability of exceeding given values

of landing-impact accelerations for two different operators of the sametype of

turbojet transport. The landing-impact accelerations experienced by Airline K

were about 40 percent higher than those of Airline E. This difference, which

is greater than differences generally found between airplane types, is attrib-

uted to differences in techniques and crew training between the two airlines.

Unusual Events

Routine examinations of VGHrecords in the past have, on occasions, shown

events not normally associated with passenger-airline operations. Examples of

such unusual events, which were reported to the airlines and the aircraft

industry several years agoj were: pltch-up when the aircraft flew at too high an altitude for its weight, collislon-avoldance maneuvers, large acceleration oscillations during landing approach, and an upset in which the aircraft recovered from a spiral dive with 4.4g maximum normal acceleration at an air- speed of about 405 knots.

Figure 3 shows one such unusual event recorded during a routine passenger- carrying flight. The nonlinear airspeed scale results from the nonlinear vari- ation between dynamic pressure and airspeed. The salient point of this time history is the indicated airspeed of lO0 knots reached at an altitude of about 23_000 feet. The pilot apparently reduced altitude and airspeed in order to reduce the airplane response to the turbulence shown by the acceleration trace.

Based on take-off weight and estimated fuel consumption, it appears that the airplane was very near the stall.

The frequency of appearance of such events on VGH records from the older turbine transports has dropped considerably during the past 18 months. During this perlod 3 however# the program on these transports was being largely phased out and it is not known how much this reduced frequency of unusual events is associated with the diminishing number of aircraft in the program.

In-Fllght Accelerations Figure 4 compares the gust, operational maneuver, and check-fllght maneu- ver accelerations for turbojet, four-engine turboprop, and four-englne piston aircraft in the form of cumulative frequency per nautical mile to exceed various values of incremental accelerations. Each curve represents the average of the experience of a number of individual airplanes. Relative to the four-englne piston airplanes, gust accelerations were experienced less often by the turbo- jets and slightly more often by the turboprop aircraft. The operational maneu- ver experience of the three types of transports is essentially the same. Over most of the acceleration range, turbojet and turboprop airplanes show more fre- quent check-flight maneuver accelerations than the piston aircraft.

Figure 5 shows a comparison of total gust and maneuver accelerations for the three types of aircraft. The distributions shown here are a combination of distributions obtained from VGH and from V-G records. As may be seen, the total in-flight accelerations for turbojet, four-engine turboprop, and four- engine piston aircraft are not significantly different.

The relative contributions of acceleration sources to the total number of accelerations equal to or greater than 0.6g for each aircraft type are shown in figure 6. For the piston and turboprop aircraft, the gust accelerations pre- dominate, check-flight maneuver accelerations being a secondary source. For the turbojet aircraft, however, the order is reversed and check-flight maneuver accelerations provide the largest portion of the total accelerations. For all three types, operational maneuver accelerations contribute a relatively small number of accelerations equal to or greater than O.6g. As was indicated earlier, oscillatory accelerations are an additional source for turbine aircraft, but this contribution is negligible at the larger values of incremental acceleration.

Gust Velocities In figure 7, distributions of derived gust velocity (that is, derived from acceleration, airspeed_ altltude_ and weight as indicated in ref. 6) in 10#000-foot altitude intervals for turbine transports are compared with estl- mated distributions (ref. 7) based on past investigations. The estimated dis- tributions represent a model atmosphere widely used to estimate gust-velocity experience. A reduction in gust velocity by a factor of 20 percent had been made in the estimates of reference 7 to account for airplane flexibility. In order that the comparison between the present data and the estimates of refer- ence 7 be more compatible, this 20 percent has been restored to the estimates.

The data are in reasonably good agreement with the estimates made about 7 or 8 years ago.

CONCLUDING REMARKS Recent results from NASA V-G and VGH recorders installed on commercial turblne-powered transports have indicated the following conclusions: Exceedances of placard speeds, considered to be a serious problem in the early period of turbine transport operation_ appear to have been significantly reduced by the redefining of the placard speeds and the changes in the aural warning.

Oscillatory accelerations, which were one of the unanticipated aspects of turbine transport operations_ have become much less frequent but have not been eliminated.

The landing-impact accelerations for turbine transports are, in general, higher than for piston transports and vary substantially between operators.

Unusual events_ such as large departures from the planned flight profile, appear to occur less frequently than in the earlier period of turbine transport operation.

Although the relative contributions of gust and maneuver accelerations to the total in-flight accelerations have changed somewhat from those of piston transports_ the total In-flight acceleration experiences for turbine transports are not significantly different from those for four-englne piston transports.

REFERENCES i. Staff of Langley Airworthiness Branch: Operational Experiences of Turbine- Powered Commercial Transport Airplanes. NASA TN D-1392, 1962.

2. Taback, Israel: The NACA Oil-Damped V-G Recorder. NACA TN 2194, 1950.

3. Richardson, Norman R.: NACA VGH Recorder. NACA TN 2265, 1951.

4. Coleman, Thomas L.; Copp, Martin R.; and Walker, Walter G.: Airspeed Operating Practices of Turbine-Powered Commercial Transport Airplanes.

NASA TN D-744, 1961.

5. Anon.: Airworthiness Standards: Transport Category Airplanes. Federal Aviation Regulation Part 253 Rules Service Co. (Washington, D.C.)3 Feb. l, 1965.

6. Pratt, Kermit G.; and Walker, Walter G.: A Revised Gust-Load Formula and a Re-Evaluation of V-G Data Taken on Civil Transport Airplanes From 1933 to 1950. NACA Rep. 1206, 1954. (Supersedes NACA TN's 2964 by Kermit G.

Pratt and 3041by Walter G. Walker.)

7. Press, Harry; and Steiner, Roy: An Approach to the Problem of Estimating Severe and Repeated Gust Loads for Missile Operations. NACA TN4332, 1958.

T.A3_ I SCOPE OF PROGRAM TURBOJET TURBOPROP VGH NUMBER OF AIRPLANES 22 9 NUMBER OF AIRLINES 8 5 DATA SAMPLE, HR 47,058 12,484 PERCENT OF U.S. AIRLINES FLIGHT TIME 0.88 0.41 V-__GG NUMBER OF AIRPL/_IES 24 I0 NUMBER OF AIRLINES 7 3 DATA SAMPLE, HR 127,796 48,506 PERCENT OF U.S AIRLINES FLIGHT TIME 3.41 1.60 TABLE II SPEED LIMIT EXCEEDANCES FLIGHTS TO EXCEED VNO [ PERCENT TIME OVER VNO AURALWARNING AT VNE 61 / 0.02

=.7 L o.,8

AURAL WARNING AT VMO (VNo) m o -- o LANDING IMPACTS OSCILLATIONS NUMBER OF FLIGHTS ALTITUDE I0°1_'_ AIRLINE O49 ;:'19 ACCELERATION AIRSPEED TIME,MIN ALTITUDE

I '", \

Q 8 Q_'_ ACCELERATION _ AIRSPEED : : : '_F, . .

._. ;, _ _ ,i_ 48 50 52 54 56 58 INCREMENTAL ACCELERATION, g UNITS TIME, MiN Figure 2 Figure i TIME HISTORY OF AN UNUSUAL EVENT ACCELERATION, g UNITS 01 I I _ q I i I q ALTITUDE, 30_C1_ /_ FT 20 L I I I I I I I I I AIRSPEED, 30

_NOTS _oo I- _'--.._

IOOL o' _ _ ; _ ,_ ,!, ,_. ,_ T_ME, M,NOTES Figure 3 ACCELERATION SOURCES TOTAL IN-FLIGHT LOADS AIRPLANE -- TURBOJET 1(3-1 ----- TURBOPROP AIRPLANE ..... RSTON 10-2 OPERATIONAL CHECK FLIGHT 10-3 GUSTS MANEUVERS MANEUVERS _O-I CUMULATIVE I0 -4 FREQUENCY PER NAUTICAL 10-5 10-3 MILE CUMULATIVE FREQUENCY .,'-,-4 1(3-6 PER NAUTICAL IO-2 __ io-7 MILE I0 -_ IO-E _ I0-8 IO-7 I0-9 o i ,._ ,._ o INCREMENTAL ACCELERATION, g UNITS INCREMENTAL ACCELERATION, gUNITS Figure 4 Figure 5 RELATIVE CONTRIBUTIONS TO TOTAL IN-FLIGHT GUST VELOCITY ACCELERATIONS 0 TURBINE INCREMENTAL ACCELERATION _ O.6g I0 TO 20 xlO3FT I/W//_ CHECKFL,GHT MANEWERS I001%0 TO IOxIO3FT i OPER_IONAL MANEUVERS ,o-4_ °o _GUSTS PISTON TURBOPROP TURBOJET CUMULATIVE I q q I I0- I I I I FREQUENCY PER NAUTICAL MILE IOO I 20 TO :30 x IO 3 FT 30 TO 40xlO3FT

.C.T I/

IO-2 OF TOTAL /[_ ACCELERATIONS 401_ 10-6 F Io -8 I i i O 40 810 DERIVED GUST VELOCITY, FT/SEC Figure 6 Figure 7 12. SIMULATOR STUDIES OF THE DEEP STALL By Maurice D. White and George E. Cooper NASA Ames Research Center SUMMARY Simulator studies of the deep-stall problem encountered with modern air- planes are discussed. The results indicate that the basic deep-stall tenden- cies produced by aerodynamic characteristics are augmented by operational con- siderations. Because of control difficulties to be anticipated in the deep stall, it is desirable that adequate safeguards be provided against inadvertent penetrations.

INTRODUCTION Renewed concern has recently developed over an aircraft handling-qualities problem that had received extensive treatment about lO years ago, namely, the pitch-up. Because of some differences in the character of the problem, it has come to be identified differently, either as a "deep-stal_' or a "T-tail" prob- lem. The term "deep stall" refers to the very large angles of attack that are developed after the stalling angle has been exceeded. A popular definition of the term "deep stall" associates it with a locked-in condition where recovery is impossible. In this paper the former more general definition is accepted.

The term "T-tail" refers to the tail arrangement that tends to produce pitching- moment variations that would encourage deep stalls_ the horizontal tail being located high on the vertical tail. The widespread use of this tail arrange- ment on modern airplane designs is one of the factors that prompted the current concern over this problem.

Safeguards against dangerous stall penetrations are included in the requirements of existing civil and military regulations. However, since some testing of these airplanes in the stall is required and since a possibility of inadvertent stall penetration (however remote) doesexist, a further under- standing of the nature of the problem is necessary. Accordingly 3 the National Aeronautics and Space Administration has started a number of investigations to study this problem. The present paper describes some of the results of simu- lator studies that were conducted at the Ames, Langley, and Flight Research Centers.

SYMBOLS lift coefficient C L Cm pitching-moment coefficient

h

altitude, ft

V airspeed, ft/sec

angle of attack, deg e angle of pitch_ deg elevator deflection, deg 5e SIMULATOR STUDIES The variation of the pitching-moment coefficient with angle of attack that is typically identified with deep-stall tendencies is illustrated in figure 1.

On the lower part of the figure is shown a corresponding lift curve. The initiation of unstable slopes is usually identified with the maximum lift coef- ficient. In the stall of an airplane having a pitching-moment curve of this type, the angle of attack tends to increase rapidly as the stall is penetrated as a result of the unstable variation of pitching moments. A further contribu- tion to increased angles of attack results from the accompanying drag increases and lift decreases 3 which produce a descending flight path.

Simulator studies of the deep stall that were performed at the three Centers are indicated in table I and will be described in this paper. As shown in this table, the nature and scope of the various studies were different.

TABLE I.- SIMULATOR STUDIES AT NASA RESEARCH CENTERS Degrees of freedom Mot i on Scope Center Piloted in motion Longitudinal recovery performance with No 3 Flight wide range of variables, including time for initiation of recovery control.

6 Fixed Yes Piloted study of general longitudinal Langley recovery performance with emphasis on control power and pitch-damping effects.

6 Roll Yes General investigation of factors Ames and influencing deep-stall acceptability pitch from pilot's standpoint, longitudi- nal and lateral-dlrectional modes.

In the piloted studies, no consideration was given any stall-warning

effects or stall-prevention effects other than those resulting from the basic

aerodynamic characteristics. Stall tests were conducted as deliberate maneu-

vers from straight, wings-level flight, and no turbulence effects were intro-

duced. However, in the interpretation of the airplane responses, the pilots

would, of course, consider realistic operating conditions.

GENERAL NATURE OFTHEPROBLEM

A great many aerodynamic and operational elements define the deep-stall

problem. Someof the important elements other than the pitching-moment curve

are itemized in table II.

TABLEII.- ELEMENTS OFDEEP-STALL PROBLEM

1. Depth of angle-of-attack penetration

(a) Higher drag

(b) Limited control power

(c) Unsatisfactory lateral-dlrectional characteristics

(d) Delay in recovery control application

2. Steep flight-path angles

(a) Large altitude losses

(b) Steep nose-down attitudes

3. Inadequacy of conventional instrument display

(a) Lack of cues to identify stall entry

(b) Misleading information for timing recovery pull-out

The interrelationships amongthe various elements are complex; somerepresent

causes, someeffects, and someact as both the effect of one element and the

cause of another.

Consider first the depth of angle-of-attack penetration; it has been found

that, with initial entry into the stall, the drag increases and the lift

decreases, and these effects together cause the airplane to descend. This

angle of descent contributes a further increase in angle of attack which fur-

ther accelerates the pitch-up. This effect is illustrated in figure 2 where

the flow-angle relationships before the stall and in the stall are shown.

With penetration to deep-stall angle of attack there frequently occurs a

decrease in longitudinal control power, which compromisesthe ability of the

pilot to arrest the stall penetration and delays recovery. As a result, the

pilot is committed to flight at deep-stall angles of attack for a considerable

time period, during which an altitude loss is suffered, and it may be necessary

lO3 to control unsatisfactory lateral-directional characteristics that occur in many airplanes at these angles of attack.

The depth of stall penetration is also sensitive to delay on the part of the pilot in initiating recovery control, and this problem will be discussed in more detail in a subsequent section.

The remaining elements of the deep-stall problem listed in table II are discussed_rlth the aid of figure 3, which shows two typical deep-stall time histories that were obtained during operations on the Langley simulator. The dashed curves were recorded during the initial experience of a pilot with the deep-stall pitching-moment and control effectiveness characteristics. The solid curves were recorded after the pilot had obtained some familiarity_rlth the problem.

On the first try, as the airplane pitched up, the pilot failed to do more than return the elevator to zero deflection. _ne nose dropped to a horizontal attitude, and the pilot awaited recovery. Recovery seemed to be progressing satisfactorily because the airspeed started to increase. The angle of attack, however, continued to increase to _O ° and thereafter stabilized at about 45 °.

As a result, when the control was pulled back to resume level flight, it served only to maintain the stall.

With practice and understanding of the problem, the pilot applied full nose-down corrective control to initiate recovery, with little effect on the initial attitude changes. This time, however, the pilot held full control, despite the misleading attitude and airspeed indications, and finally recovery was completed.

Note the large altitude losses and the steep nose-down attitudes that were indicated in figure 3 to be required for recovery. Attention has also already been directed to the misleading information that led to a premature pull-out attempt. In the particular case illustrated, the attitude changes at stall entry were large enough to give a strong indication of stall entry. In many cases the attitude change is much smaller, and the cue is correspondingly weaker. In fact, in some cases studied in the Ames investigations, where only minimal pitch-up instabilities existed on the pitching-moment curves, the lack of any warning cue resulted in the airplane drifting into a deep stall unobtru- sively because of only the drag increase.

EFFECTS OF MARGINAL RECOVERY CONTROL One of the most disturbing aspects of the deep stall is the length of time for which the pilot is committed to the stalled region once the stall has been entered. As already discussed a number of different factors contribute to delay in recovery, but one of the most obvious is the fact that, even with full nose-down control, the pitching-moment curve approaches a balance point at deep-stall angles of attack, leaving little pitching-moment margin for recov- ery. This effect is illustrated in figure 4 where, at _ _ 35 ° , a minimal I04

margin of pitching-moment increment is available for recovery. It will be

noted in figure 4 that_ if the curve for full-forward stick crosses into the

region of positive values of Cm, it would be possible for the airplane to

become"locke_' into the stall so that elevator alone cannot effect recovery.

Less extreme cases might still represent serious recovery problems. In order

to define what degree of controllability is required, studies were conducted in

both the Langley and Flight Research Center simulations. The results are indi-

cated in figures 5 and 6.

Figure 5 shows someof the variations in the pitching-moment curves that

were examined in the Flight Research Center studies. These range from curves

for which full-forward stick would be inadequate for recovery if the angle of

attack reached approximately 49° to those for which positive control margin

were available. Figure 6 showsthe results in terms of two parameters, the

maximum angle of attack developed and the altitude lost. For different control

margins these parameters are plotted against another important variable, time

delay in initiating recovery. In the case of curve A, recovery can be accom-

plished by prompt control application but a complete lock-in would occur for a

4-second time delay. For curves B and C, progressively larger stall penetra-

tions and altitude losses with increasing delay in applying recovery control

are evident, but, interestingly, for the data shownthere does not seemto be

much effect of differences in control margin available. It appears to be

important only that somemargin be available.

These specific results were obtained in analog computer studies conducted

at the Flight Research Center. The importance of the second variable, time

delay in initiation of recovery, was also repeatedly emphasizedby pilots who

participated in the studies at Langley and Ames, and these reactions stress the

requirement for adequate cues to identify the stall onset.

STALLPREVENTION BY STABILITYINCREASE PRECEDING PITCH-UP

In operational service it is, of course, basically undesirable to have the

capability of penetrating into a pitch-up that would lead to a deep stall, even

if positive recovery control were provided. This problem is widely recognized,

and attention is being directed to the development of safeguards against such

penetrations. One such safeguard would be the existence of a sharp increase

in static longitudinal stability immediately preceding the pitch-up, and studies

were conducted at Amesto obtain someindication of the stability increase that

would have to be provided.

The first part of the study involved piloted operation for the conditions

represented by the three curves shown in figure 7.

The indications from the tests were that the increase in stability repre-

sentedby the "bump" in curve (b) did not_ of itself, provide a positive enough

indication or warning of the impending pitch-up. It appeared to be partly a

question of the magnitude of the slope changethat occurs in approaching the

bumpfrom lower angles of attack, and partly a question of how deep the bump

was in terms of the amount of control required to pull up through the stable

slope. The condition represented by curve (c) was considered to provide the

minimumamount of warning that would assure no penetration of the pitch-up

region; with the control effectiveness and control-force gradient provided,

this curve required a stick movementfrom 40 to 80 percent of the full aft con-

trol and a control-force change from 20 to 40 pounds to pass the peak of the

bump. One interesting inference of these results was that complete protection

against stall entry may not depend entirely on limiting elevator authority.

This class of stable bumpwas also examined for its ability to protect

against inadvertent deep-stall penetrations due to atmospheric turbulence.

As a part of the analog computer study conducted at the Flight Research Center,

the ai__plane was balanced at the angle of attack corresponding to the knee of

the pitchlng-moment curve (see insert, fig. 8) and then was exposed to turbu-

lence as represented by a vertical-velocity variation measured in flight. This

test was repeated for different levels of turbulence amplitudes and different

shapes of the stable bump, with the results shown in the main part of fig-

ure 8. The ordinate of the main part of figure 8 is an,empirical parameter

(_Cm) X (2k_) that roughly measures the magnitude of the stable bump. (The

increments _C m and _ are identified in the insert in fig. 8.) For a given

turbulence level, the larger values of this parameter should provide more pro-

tection against angle-of-attack reactions that exceed the stable range of

pitching moments. And, of course, with only small values of the parameters

the protection is less, and there is more likelihood of the critical angle of

attack being exceeded to project the airplane into the unstable pitch-up range.

The boundary curve then represents the minimumbumpsize that will prevent

pitch-up into the stall. It is of interest to note that in the analog-computer

studies, the critical part of the turbulence time history that triggered pitch-

ups (of which only a small section is shown) was the sustained updraft that

occurred at the region noted. For the particular curve that had been identi-

fied from the earlier piloted tests as providing adequate protection, the cor-

responding limiting turbulence amplitude would be about 25 percent of the full-

scale turbulence.

FLIGHT-TEST PHILOSOPHY

The simulator studies reported here have provided considerable insight

into the nature of the deep-stall problem but have not provided a schedule for

rating acceptability of a given set of characteristics. On the contrary, it

appears that unless an airplane happens to have unusually mild lateral-

directional and longitudinal problems in the stall, measures are likely to be

required to limit prolonged penetration or exploration of the stall. However,

as noted earlier a certain amount of deliberate stall penetration must be

assumedin development and training, in order to establish and demonstrate

safeguards for operation. To accommodateto this situation somephilosophy of

operation is necessary, and figure 9 illustrates such a philosophy.

The important innovation of this suggested philosophy is that it provide

a positive cue to impending stall, in addition to (and removed from) the usual

stall warning. The locations of the stall warning and positive cue with respect to the angle of attack of deep-stall entry are indicated in the top part of figure 9, with the pitching-moment curve defining the angle-of-attack scale.

Two kinds of airplane operation are of concern, as noted in figure 9. One is standard operational service. For the pilot in this case the stall warning must be demonstrated in training, but positive protection should be provided against deliberate or inadvertent deep-stall entries or penetration. The other kind of operation is test flying for development or certification where the test pilot may explore the deep-stall region.

The stall warnings suggested would be of the same kind for either type of operation - buffeting, perhaps, as a natural form# or a stick shaker as an arti- ficial form. As a positive cue to preclude further angle-of-attack increase, a strong increase in longitudinal stability would be desirable for either oper- ational service or developmental testing. Lacking a natural cue, an artificial positive cue of the nature of a stick pusher is proposed for operational serv- ice. For test flying, the artificial cue could be in the form of an angle-of- attack indicator, so that exploration at higher angles of attack was possible, but under conditions that kept the pilot clearly informed of the status of the stall. Finally, it is fairly obvious that in flight investigations of the deep stall, provision should be made for artificial recovery augmentation if there is any question about the effectiveness of natural methods for recovery.

It should be noted that in the foregoing discussion, the various devices indicated as safeguards are identified as typical and are designated only for illustrative purposes. Other methods or devices could be used and might possibly prove superior to those discussed.

CONCLUSIONS From the simulator studies of the deep-stall problem, the following con- clusions are reached: 1. In airplanes that are identified with deep-stall pitching-moment curves, other aerodynamic and operational factors would tend to prolong the time that the airplane is committed to stalled flight. If the stall is penetrated, the pilot should be given positive cues to recognize it; he should then apply full corrective control promptly and should maintain full control despite misleading information, if the stall duration is to be effectively limited.

2. Adequate safeguards against penetration of the deep stall are needed to avoid the possible severe consequences of prolonged stalled flight. These consequences include large altitude losses, steep nose-down recovery angles, and possible unsatisfactory lateral-directional controlability. It is suggested that these safeguards include an additional positive cue between the stall warning and the actual stall.

i07

3. Attitude and airspeed information were muchless valuable than angle-

of-attack information for detecting stall entry and for completing longitudinal

recovery.

TYPICAL DEEP STALL LONGITUDINAL AERODYNAMICS ANGLES IN A DEEP STALL Cm+ v _ PRESTALL FLIGHT pA_ _._'_,_ !NITIAL STALL V _ i i I / i i PA_ DEEP STAL_L O 10 20 30 40 50 60 FLIGHT ,,,y ANGLE OF ATTACK, deg V _ Figure i Figure 2 AERODYNAMICS OF P,ECOVERY CONTROL TIME HISTORIES OF DEEP STALLS ..... RRST STALL 40F -- AFTER P_CT,CE 201 p, C r J_ _ _//'--_"\\ ...- V / \,\\STICK Art

t_ .......... % _o .--_

\\ STICK FORWARD 6001 20

L I I I I I f\ i _l i_ CL

2oo0 2o 4o 6o 8oloo._"o! .... _ ..... _ TIME, se¢ a;(:J20 i_"/L..._Ji _ ' 0 I0 20 _0 40 50 60 0 20 40 60 80 I00 ANGLE OF ATTACK, deg TIME, sec Figure 4 Figure 3 EFFECT OF CONTROL MARGIN ON RECOVERY STUDIES OF RECOVERY CONTROL MARGIN 4F x io 5 ! A ___.- B / / .tf_C FULL FORWARD CONTROL LOSS, ft 2 ALTITUDE ,_'-_ 01 I I I I o ,,-f-2u.

/ ,-'/P"- \'<',A

c._¢ ,,/y

OF ATTACK, /

ANGLE--'I ' ,'f

/ _ 2D _ t i i t 0 2 4 6 8 RECOVERY DELAY TIME, seC a, d6g Figure 6 Figure INCREASED STABILITY PRECEDING PITCH-UP BOUNDARY FOR PROTECTION AGAINST DEEP STALL IN TURBULENCE VERTICAL '60 [- _, Cm - GUST 8Ok ..... _, ,,4 ^i, U "_..'_ VELOCITY _ ,.p,,It_'_,,_lW'_l j1 1 Irr'_l.l , _ - • v- • - r_ ft/sec eo L _ L-SUSTAINED (FULL SCALE) - | UPDRAFT -16D_ _-- 16 sec + Cm -. 8 STALL HTCH-UP -,6 (,_Cm)(Aa), deg Cm + _.a -.4 -.2 o 2'o ,'o io go 0 I0 20 30 40 50 60 PERCENT FULL-SCALE TURBULENCE AMPLITUDE ANGLE OF ATTACK, deg Figure 8 Figure 7 ii0 SUGGESTED OPERATIONAL PHILOSOPHY FOR AIRPLANES WITH DEEP-STALL TENDENCIES + C m I NORMAL OPERATIONS I WARNING REQ'D POSITIVE(_UE REQ'D RECOVERYAUGMENTOR REQUIRED OPERATION (TYPICAL) (TYPICAL) (TYPICAL) UNMISTAKABLE OPERATIONAL BUFFET STABILITY INCREASE NOT APPLICABLE B SERVICE STICKS H_, K--E'R STICK PUSHER UNMISTAKABLE DEVELOPMENT BUFFET STABILITY INCREASE FLIGHT TAI L PARACHUTE ANGLE OF ATTACK TEST STICK SHAKER INDICATOR Figure 9 iii 13. DEEP-STALL AERODYNAMIC CHARACTERISTICS OF T-TAIL AIRCRAFT By Robert T. Taylor and Edward J. Ray NASA Langley Research Center SUMMARY A wind-tunnel research program has been undertaken by the NASA to study the aerodynamic characteristics of T-tail aircraft at high angles of attack.

The program was designed to show the effects on longitudinal stability and con- trol of several configuration variables.

The results to date do not allow the formulation of general design rules, but the effects of several configuration variables have been noted to have a prime influence on the post-stall characteristics..

An increase in tail size, changes in the location of fuselage-mounted engine nacelles, and reduced fuselage-forebody lift were all found to have a beneficial effect on static longitudinal stability at high angles of attack.

I NTR 0DUCTI ON Recently, both here and abroad, several transport aircraft have appeared in service which employ a horizontal stabilizer mounted on top of the vertical fin. Many others, including executive as well as commercial transports, are in the design or certification stages. Past research on the T-tail problem has given an indication of the general design rules for providing good pitch char- acteristics in the angle-of-attack range prior to wing stall. At angles of attack above the stall angle, however, there is a distinct lack of data on the pitch characteristics of T-tail aircraft.

The NASA has, therefore, undertaken a wind-tunnel research program to investigate the post-stall, or high-angle-of-attack, behavior of transport-type configurations employing the T-tail. Table I shows, in general, the scope of the wind-tunnel tests. The program was designed to study the effects on lon- gitudinal stability and control of many configuration variables, primarily at low speed. The studies have not yet progressed to the point which would allow the formulation of general design rules, but several interesting facts have come to light. It is the purpose of this paper to show the aerodynamic origin of the instabilities at high angles of attack of T-tail airplanes, and this discussion is concerned with only those items and variations which appear underlined in the table.

SYMBOLS

A cross-sectional area

cross-sectional area of basic configuration

ABASIC

b

wing span lift coefficient

CL

pitching-moment coefficient at 0.40_

(cm)

tail contribution to stability Z_Cm,t wing mean aerodynamic chord, in.

tail incidence, deg it change in tail incidence, deg A_ t tail arm, in.

M Mach number S wing reference area tail surface area Z tail height, in.

angle of attack, deg stall angle of attack, deg _STALL DISCUSSION Basic Configuration Figure i shows a line drawing of the model used in the investigation. It was typical of the current T-tail airplanes in that it had a wing with moderate sweepback and aft-mounted engine nacelles. Some pertinent model dimensions are given in terms of the wing mean aerodynamic chord 5 and the wing span b.

The model was constructed so that the configuration changes listed in table I could be readily accomplished. In this discussion the aerodynamic characteris- tics of this model will be used as a standard of comparison and the model will be referred to as the basic configuration.

Effect of Horizontal-Tail Height

Figure 2 shows data for two horizontal-tall heights in terms of the

pitching-moment coefficient as a function of angle of attack. The reference point for pitch data, here and throughout the paper, is 40 percent of the mean aerodynamic chord, which is a typical aft center of gravity for the current T-tail aircraft. The shaded portion of the figure below an angle of attack of 20 ° has received a great deal of attention in the past. The smaller shaded region between l_ ° and 20 ° denotes an angle-of-attack region for which these data are not believed to be applicable tofull-scale aircraft because of the low Reynolds numbers of the tests.

Effect of Tail Incidence At low angles of attack the low horizontal tall contributes much less to stability, because of its position in a region of high downwash behind the wing.

As the angle of attack increases, however, the tail moves below the high down- wash region, becomes more effective, and remains effective to high angles of attack. The high horizontal tail, on the other hand# starts in a region of low downwash at low angles of attack and thus shows greater stability. As the angle of attack approaches the stall angle, the high horizontal tail moves into the high downwash field created by the wing and its contribution to stabil- ity tends to lessen. If the tail location is high enough, this decrease in stability can be delayed until after the stall angle has been exceeded. At very high angles of attack the high horizontal tail becomes even less effective as it penetrates deeper into the alrplane-wake system. The influence of high- angle-of-attack wakes on the airplane pitch-control effectiveness is shown in figure 3. Here pitching-moment coefficient is plotted as a function of angle of attack. The solid curve represents data for the basic configuration trimmed at an angle of attack of lO °. A stable trim point is evident near an angle of attack of 40 °. The dashed curve represents a more positive stabilizer inci- dence (Zkl t = 4.1 o) and serves to illustrate what happens to the control effec- tiveness with increasing angle of attack. If the two curves are compared where the arrows are shown# it can be seen that the amount of control available from a given stabilizer or elevator angle is reduced to a small percentage of its low-llft value due to the wake system at the high angles. The implication is that the amount of control available for recovery from high-angle-of-attack penetrations is seriously limited for some configurations which use the T-tail and that the reduction of control power and stability at high angles of attack is a function of the wake system of the airplane.

Effect of Engine Nacelles In this study of post-stall pltchingproblems the effect of the aft-engine- nacelle placement must be examined. Figure 4 shows the effect of removing the nacelles from the basic configuration. On the left in figure 4 are curves of pitching-moment coefficient against angle of attack for the model with the tall off (upper plot) and with the tail on (lower plot). The nacelles-on curves are solid and the nacelles-off curves are dashed. Notice that the addition of the

nacelles is stabilizing at all angles of attack with the tail off. With the

T-tail on, the addition of the nacelles is destabilizing past an angle of attack

of about 20°. This result suggests the presence of a wake off the nacelles

which masksthe T-tall and reduces its effectiveness; comparison of the tail-

increment curves on the right in the figure showsthis effect more directly.

The tail-increment curves are simply the difference between the tail-on and

tail-off curves and represent a measure of the tail contribution to stability.

There is a break in the nacelles-off curve at the angle of attack at which the

wing-fuselage wake crosses the T-tail. Whenthe nacelles are added, a marked

reduction is seen in the tail contribution. The inference is that at high

angles of attack the T-tall has entered the nacelle wake and the tail contribu-

tion to stability is degraded to the point that the complete configuration

becomesunstable at an angle of attack of about 20° . In addition to this

reduction in stability, the nacelles were found to have a large effect on the

control power available at the higher angles of attack. The effect of the

nacelles appears in the wake system of the aircraft and can affect the tall

only if the tail is mounted high on the vertical fin.

With this information in mind a series of tests was madein which the

nacelles were movedto various locations on the model; results of someof these

tests are shownin figure 5. The small sketches at the top of the figure show

the nacelle locations. The curves again are pitching-moment coefficient as a

function of angle of attack; each complete-configuration (nacelles-on) curve

is solid and each nacelles-off curve, dashed. The difference between the

nacelles-on and the nacelles-off curves is, in general, an indication of the

size of the nacelle wake at the higher angles of attack. The solid curve on

the left showsdata obtained when the nacelles were movedto wing pylons, as

seen in the sketch, at a spanwise location about in line with the tips of the

horizontal stabilizer. A stabilizing influence is apparent prior to stall and

deteriorates between angles of attack of 20° and 30o. This effect at the

higher angles of attack is not fully understood at this point but is presumed

to be due to the nacelle wake, shed forward and above the wing at high angles,

crossing over the tips of the horizontal tail. As a meansof hiding the

nacelles in the aft-fuselage wake they were movedon top of the fuselage close

to the vertical tail. Nacelles added in this location caused essentially no

change in the pitch curve.

The other plots in figure 5 are for nacelle locations which vary only

in longitudinal position; that is_ the nacelle span and vertical position

were unchangedand the nacelles were in a forward_ a basi% and an aft loca-

tion. For the nacelles in the basic location, note again the marked deteri-

oration of the pitch curve past _ = 20° due to the nacelles and a stable

trim point at _ = 40°. With the nacelles in the forward position_ not much

influence is shown on pitch prior to wing stall and a destabilizing trend is

shownafter _ = 20°. However_the nacelle is quickly enveloped in the stalled-

wing wake and near _ = 40° the two curves merge. As the nacelle is moved

aft from the basic location_ notice the heavy stabilizing contribution from

= i0 ° to 20°. The nacelle contribution in this angle range is presumed to

be due to its acting as a low horizontal tail. Beyond _ = 20°, however_ the

pitching-moment-coefficient curve deteriorates as the nacelle wake crosses the

T-tail, and beyond _ = 40° the pitch increment due to the nacelles is even

worse than for the basic configuration. It appears that aft movement of the basic nacelle improves the pitching-moment coefficient at low angles of attack and tends to shed a worse wake at very high angles of attack. Moving the nacelle forward does not appear to have much effect on the data for low angles of attack and improves the pitch curve at very high angles of attack. It should be remembered that no fixed set of criteria exist which will allow the designer to choose_ on the basis of these tests_ the best location for engine nacelles when the T-tail is employed; the data shown are only indicative of some effects which need further exploration. (Wind-tunnel testing at high angles of attack is the only safe way of developing the required stability and control data. ) Effect of Horizontal-Tail Size If the foregoing discussion of nacelle effects may be interpreted as showing the influence of the nacelle-wake size on a fixed size of the horizontal tailj the effect of increasing the size of the T-tail with relation to the basic nacelle wake should be shown. Figure 6 illustrates the effect of increased tail size on the pitching-moment coefficient as a function of angle of attack. On the left in figure 6 data are repeated for the basic configura- tion with the horizontal-tail area of 22 percent of the wing area. The plot on the right in the figure is for the model with the tail size increased to 30 percent of the wing area and the tail span, to 40 percent of the wing span.

The solid curves are for both models trimmed at an angle of attack of lO °. The addition of the larger tall has increased the stability at low angles of attack and has eliminated the stable trim point at _ = 40 ° noted previously for the basic configuration. The dashed curve is added 3 for a fixed incremental inci- dence of 4.1 °, to illustrate the degradation in control effectiveness with increasing angle of attack. Comparing the arrows again shows that increasing tail size has produced a small increase in control power at low angles of attack; at high angles of attack the control power is still reduced to a small percentage of its low-lift value. Whether or not curves such as these are what would be termed acceptable is a subject for further simulator studies.

Effect of Fuselage Size The effect of the fuselage-forebody length and size was also studied during the wind-tunnel program. It has been learned from experience and wind- tunnel tests in the past that the effect of increasing the fuselage-forebody volume willj in general_ decrease the stability. The point in question con- cernlng the T-tail layout has been the influence of the fuselage wake at the T-tail location. Figure 7 shows a line drawing of the basic model and of a version with a small fuselage. The smaller fuselage differed in depth as well as in width and had a cross section which 3 although similar in shape, was only _5 percent as large as that of the basic body. All other dimensions for the two models were identical. Figure 8 shows the effects of fuselage cross- sectional size. Here again are plotted the pitching-moment coefficients for the model with tail off, tail on, and a tail increment, each as a function of the angle of attack. Note that the smaller fuselage, represented by the solid ll7 curve, is more stable with the tail off as well as with the tall on. Note further that the tail-lncrement curves are nearly the same regardless of fuselage size. The reduction of tall contribution at the higher angles of attack is again due to nacelles. Based on the same reasoning as was used pre- viously_ since the tail increment remains relatively constant the fuselage size can be said to have little or no influence at the tall plane. The benefits in pltchlngmoment are primarily due to decreased llft on the fuselage fore- body which in turn is due to decreased cross-sectional size. Fuselage-forebody length has an effect similar to the cross-sectional-size effect shown in this figure - that is, increases in the fuselage-forebody length tend to destabilize the moment curves - and this effect occurs regardless of the location of the horizontal tails.

Effect of Mach Number The data presented thus far were obtained at a Mach number of 0.9_I. Fig- ure 9 shows some results of tests at higher Mach numbers made in the Langley transonic dynamics tunnel. Results are shown as llft and pitching-moment coef- ficients as a function of angle of attack for the basic configuration. It is obvious from the pltchlng-moment-coefflclent curve that the high-angle-of- attack nonlinearities have not disappeared as a result of increased speed; in fact 3 they may be somewhat more severe. The lift coefficient shows the char- acteristic depression in initial stall angle and the associated initial maxi- mum C L. At the higher angles of attack 3 however, an increase of 15 percent in the post-stall maximum llft is noted from a Mach number of 0.21 to 0.86.

CONCLUSION An increase in horizontal-tail size, changes in the location of the fuselage-mounted engine nacelles, and reduced fuselage-forebody lift were all found to have a beneficial effect on static longitudinal stability at high angles of attack.

SCOPE OF THE WIND-TUNNEL RESEARCH PROGRAM • ITEM • VARIATIONS HORIZONTAL TAIL SIZE, VERTICAL POSITION, PLANFORM, AND tNCtDENCE NACELLE SIZE AND LOCATION WING SECTION, STALL CONTROL DEVICES, FLAPS, ASPECT RATIO, AND SWEEP FUSELAGE CROSS-SECTION SIZE, CROSS-SECTION SHAPE, FDREBODY LENGTH MACH NUMBER 0.21 TO 0.90 REYNOLDS NUMBER 0.78 x I0 6 TO 1.8 x I06 EFFECT OF HORIZONTAL-TAIL HEIGHT GENERAL ARRANGEMENT OF BASIC CONFIGURATION i t = O; Z/_ = 378; M = 021 (Cm).40_ .2 4 _'- e STALL NOSE / _ uPo , Y/_/._ .os_I_/-_ \\\

, \

-0 " tO _ 40 50 a, DEG Figure 2 Figure i :]-:]-9 EFFECT OF AFT ENGINE NACELLES EFFECT OF TAIL INCIDENCE i t = O; M = 0.21 _ NACELLES ON M=0.21; BASIC CONFIGURATION f ---- NACELLES OFF '4 [_TA LL/f "2 ITAIL C INCREMENT (Cm).40E .2_-./_',,.._ o .2 LL -.2 Cm) 40"6

(co>,oE-'F/ -°

- .ALTAILON \ _ -.8 I - _0 -I.C a, DEG -" 0 10 20 30 40 50 a, DEG a, DEG Figure 4 Figure 3 EFFECT OF NACELLE LOCATION ON PITCHING-MOMENT COEFFICIENT M=021; it=O --NACELLES ON .... NACELLES OFF BASIC CONFIGURATION -.2 " " a " STALL (Cm) 40_ LL (ISTALL -#, _i ' (t STAL

o

6_3 0 20 40 ' 0 20 40 a, DEG Figure 9 EFFECT OF HORIZONTAL-TAIL SIZE M:021 COMPARISON OF BASIC AND SMALL FUSELAGES -_.55 b _- I F.,,- 40 b_.-_ (Crn).40 _ BASIC .4 St/S_ St/S _===0_ .2 • .- I -,2 \ C_STALL SM_ -A A ABASIC = 055 l -.61 II0 210 3_) 410 . I ilo 210 310 410 5O

.°'°4 t-

a, DEG Figure 6 Figure 7 EFFECT OF FUSELAGE CROSS-SECTIONAL SIZE ---- BASIC FUSELAGE __SMALL FUSELAGE aSTALL_4 TAIL OFF EFFECT OF MACH NUMBER ON LIFT AND PITCH CHARACTERISTICS 40_" z 0 TAIL INCREMENT

< _j47

(Cm) ' -2 F P 1.6 • I I I I I .2 TAIL ON ACm, t 1.2 O / .... ' -.4_- '" _'*_"_ /_ _'_: = 0.21 o l k _= 0.86 .2 CL (Or.) ._ -.2 .8 .4 -.4

_ST,_ : ', i,- _i _

I I I I I I I 61 I I I I / -i'0_) 20 z_) I 0 20 ,40 60 -'60 " 20 40 60 O IO 20 30 40 5u 6O a, DEG a, DEG a, DEG a, DEG Figure 8 Figure 9

14. FLIGHTTESTS RELATED TOJET-TRANSPORT UPSET

ANDTURBULENT-AIR PENETRATION

By William H. Andrews, Stanley P. Butchart,

ThomasR. Sisk, and Donald L. Hughes

NASA Flight Research Center

SUMMARY

A flight program, utilizing a Convair 880 and a Boeing 720 airplane, was

conducted in conjunction with wind-tunnel and simulator programs to study prob-

lems related to jet-transport upsets and operation in a turbulent environment.

During the handling-qualities portion of the program the basic static stability

of the airplanes was considered to be satisfactory and the lateral-directional

dampingwas considered to be marginal without damper augmentation. An evalua-

tion of the longitudinal control system indicated that this system can become

marginal in effectiveness in the high Machnumber and high dynamic-pressure

range of the flight envelope. From the upset and recovery phase of the program

it was apparent that retrimming the stabilizer and spoiler deployment were

valuable tools in effecting a positive recovery; however, if these devices are

to be used safely, it appears that a suitable g-meter should be provided in the

cockpit because the high control forces in recovery tend to N educe the pilot's

sensitivity to the actual acceleration loads. During the turbulence penetra-

tions the pilot noted that the measured vibrations of 4 to 6 cps in the cockpit

considerably disrupted their normal scan pattern and suggested that an improve-

ment should be madein the seat cushion and restraint system. Also it was

observed that the indicator needles on the flight instruments were quite stable

in the turbulent environment.

INTRODUCTION

Within the past few years, the turbine-powered jet-transport airplanes

have experienced a series of incidents which, in the majority of cases, have

been associated with the penetration of a region of turbulent air. The reports

of the flight crews, as well as the analysis of data derived from onboard

crash recorders, have been consistent in that the airplanes were initially

disturbed from stabilized flight to the extent that a period of uncontrolled

flight developed and in several cases the airplanes involved did not recover

stability.

To study this problem, the Flight Research Center of the National

Aeronautics and Space Administration has initiated a flight-test program, in

conjunction with simulator and wind-tunnel studies conducted by other NASA

Research Centers on this subject. The prime objectives of the flight program

were to document the airplane handling characteristics in support of ground-

based simulator studies and to documentthe airplane characteristics and

operational experience during controlled upset maneuvers and flight in a

turbulent environment.

A major portion of the data was derived from flights within the normal

operational boundaries and illustrates someof the salient features related to

the longitudinal control system, longitudinal upset, and lateral-directional

dynamics in turbulence of these airplane types. It is recognized that these

characteristics maybe common knowledge to the majority of those who have a

working relationship with this class of airplane. However, it is believed

that, in consequenceof the series of incidents which have occurred, a

reemphasis of these inherent characteristics, particularly to the operational

pilots, is appropriate.

TESTAIRPLANES

The airplanes used in the investigation were a Convair 880 and a Boeing

720 which were provided by the Federal Aviation Agency Facility at Oklahoma

City, Oklahoma. Both airplanes incorporated standard airline equipment and

were considered to be representative airplanes in the turbine-powered jet-

transport class.

The only deviation from the standard configuration was that related to the

longitudinal control system in the Boeing 720 during the early portion of the

program. At the conclusion of the basic handling-qualities evaluation of this

airplane, several features becameapparent which indicated that the longi-

tudinal control system was not responding as predicted by the manufacturer.

Initially, it was not possible for the pilot to stall the stabilizer jack

screw motor in moving the stabilizer in or out of the trim position except in

the airplane nose-up direction at high dynamic pressure (Mach number, 0.75 at

15000 feet). Furthermore, the data indicated that the predicted maximum ele-

vator deflection was not being obtained during tests which were performed to

establish the maximum elevator available with a critically "mistrimmed" sta-

bilizer (that is, stabilizer at a position other than trim). A detailed

inspection of the cove gaps related to the aerodynamic balance system (shown

in fig. l) indicated that someof the balance-panel seals had been improperly

installed. Atthe completion of the inspection, it was determined that 5 of

the l0 balance panels shown in figure 2 were outside of the specified tolerance

limits of the cove-gap spacing by as muchas 0.28 inch at certain elevator

settings. (TEUsignifies trailing edge up.) As a result of this condition,

flight tests were repeated after the system was reworked to the specified

tolerances and someof the comparative flight results are presented in the

discussion to follow. It should be stated that although there was a measurable

change in the longitudinal wheel-force characteristics, the system modification

did not reveal an apparent change in the stall characteristics of the stabi-

lizer jack screw motor and, as will be shown, the maximum elevator conditions

were only slightly improved.

INSTRUMENTATION

The Convair 880 instrumentation was limited to provide those measurements

necessary to document responses of the airplane, the flight instruments, and

the pilot in turbulence. The instrumentation consisted of two camerasmounted

in the cockpit and at the center of gravity to record the respective pilot,

flight-instrument, and wing response to turbulence. Also included were a

standard NASAairspeed-altitude recorder, three-component linear accelerometers

at the pilot# center-of-gravity, and tail positions, and angular velocity

recorders and accelerometers at the center of gravity.

The instrumentation included in the Boeing 720 airplane was consistent

with

that required for a complete handling-qualities investigation. The meas-

ured

quantities consisted of the following:

Airspeed/altitude

Three-component linear accelerations (pilot, center of gravity,

and tail position)

Angular displacement (pitch and roll)

Angular velocity and accelerations (pitch, roll, and yaw)

Control forces (longitudinal, lateral, and directional)

Control positions (longitudinal, lateral, and directional)

Cockpit and wing cameras (similar to those used in the Convair 880)

Recording accuracies were consistent with standard NASAinstrumentation.

TESTPROGRAM

The flight program was divided into two phases. In the initial phase, a

Convair 880 airplane was flown for a total of 15 hours to record the pilot and

cockpit flight-instrument response and the associated accelerations along the

fuselage during the penetration of thunderstorm turbulence.

The second phase of the program consisted of flying a Boeing 720 airplane

for a total of 40 hours, during which time the basic handling qualities were

recorded for standard flight-testmaneuvers, for conditions of controlled

upsets, and for flights into regions of turbulence. Most of the data presented

herein was derived from this program. Figure 3 presents the variation of altl-

tude with Mach number covered during this program. Data were obtained from

Mach numbers of 0.25 to 0.79 at 15000 feet# from Mach numbers of 0.53 to 0.91 at

35000 feet, and at a Machnumber of 0.75 at 42000 feet. During the controlled

upset maneuvers, the high-speed boundary was penetrated only slightly during

the recovery portion of the maneuver. A limited amount of data was acquired at

18000 feet for comparison with available manufacturers' information and for an

evaluation of the effects of the stabilizer jack screw stall characteristics.

The weight of the airplane ranged from 190,000 to 140,000 pounds and the center

of gravity varied from 16 to 29 percent of the meanaerodynamic chord.

As the investigation of several of the turbulence-assoclated accidents

indicated that the stabilizer may have been in the setting for airplane full

nose down at impactj a significant part of the program was devoted to investi-

gating the longitudinal control characteristics with this trim setting and the

pilot's ability to recover from a longitudinal upset through elevator alone, stabilizer retrinmLing, and manipulation of the spoilers.

DISCUSSION

Consideration of the basic airplane characteristics and the commentswhich

have been obtained from various pilot reports of upset 3 turbulence penetration,

and the associated recoveries indicates a lack of appreciation of the

longitudinal-control-system limitations, particularly in unfamiliar regions of

the flight envelope. Consequently, the next series of figures are shown to

illustrate the various aspects of the system operation and potential limita-

tions, particularly in the high Machnumber and high dynamic pressure range.

Longitudinal Maneuvering Data

Figure 4 presents a typical variation of the stick force and elevator

position for longitudinal control as a function of normal acceleration at an

altitude of 34500 feet and a Machnumber of 0.84. The data illustrated by the

open symbols were obtained with the previously mentioned out-of-tolerance cove

gaps on the elevator-balance panels 3 and the solid symbols present the stick-

force variation indicative of the reduction of force that was obtained when

the clearances were brought within the specified tolerances. The maximum

apparent force reduction was between 18 and 20 pounds for this flight condition

and ranged to as high as 40 pounds in other conditions tested.

The general trend of the data covering the acceleration range between 0.1g

and 2.25g showsthat at the extreme levels of acceleration_ the stick-force

gradient is considerably reduced and the variation of elevator position appears

to be unfavorable and nonlinear.

These nonlinear characteristics are more pronounced at the higher altitude

and Mach numberswhere cruise flight is normally performed. It is realized

that these characteristics are not unusual for a control system of this type.

However, the inherent characteristics do present problems to the pilot3 par-

ticularlywhen he is subjected to extreme levels of acceleration in either

smooth or turbulent air conditions.

Longitudinal Control Effectiveness

Another feature of this longitudinal control system is that related to

the loss of control effectiveness of the tab-actuated elevator in the tran-

sonic speed range. Figure 5 presents the control tab-elevator relationship

at an indicated Mach number M i of 0.54 and 0.84 at an altitude of 35000 feet.

From the data at M i = 0.543 it may be seen that the variation of the elevator tab with elevator deflection is essentially linear and the ratio is approxi- mately 1 to 1 for the traillng-edge-up (TEU) elevator positions. At the condi- tions shoe for the higher Mach number of 0.84 the tab-to-elevator ratio is still about 1 to 1 between elevator deflections of about 5.0 ° TEU and 8 ° TED (trailing edge doE). However, beyond these limits it may be observed that this tab-to-elevator ratio is reduced by about one-half. This loss in control- tab effectiveness may be associated wlth a shock stall_ which may occur on the control-tab_elevator combination as the tab deflection is increased and the local flow becomes supersonic. It is appreciated that this explanation is not new; however, when the overspeed boundary is inadvertently penetrated, it is believed that the pilots operating the aircraft, in many cases, do not have a full appreciation of this influence.

Elevator Available for Recovery In conjunction with the loss in control effectiveness at the high speed, another item of interest is that related to the maximum elevator available for recovery from an upset with a mistrimmed stabilizer. Figure 6 presents the maximum elevator deflection attained during the flight test over the predicted Mach number range of 0.62 to 0.89 at an altitude of 35000 feet. The estimated boundary curve shoe represents the maximum elevator deflection that should be attained with a maximum deflection of the elevator tab. The flight-test data represented by the open and solid symbols show the actual elevator deflection that was obtained with the stabilizer trimmed in the full airplane-nose-down (A.N.D.) position of 3.5 ° and with the pilot applying a maximumpull force as the airplane decelerated from Mi = 0.89 to M i = 0.62.

The open and solid symbols represent the results before and after the elevator-balance-panel cove gaps were corrected to the specified tolerances.

Initially, it may be seen that the out-of-tolerance cove gap cost about 2.0 ° to 2.5 ° in elevator deflection over the speed range. Furthermore, the solid symbols indicate that under the specified trim conditions at M i = 0.89 approximately 8.5 ° of elevator deflection are available for trim maneuvering.

As the speed is reduced to M i = 0.76 the elevator available increases to approximately 10.5o. A major portion of the test data represented by the solid symbols _s obtained with both pilots pulling on their respective control wheels.

The apparent deviation between the estimated and the maximum elevator attained during the flight tests represents a reduction in longitudinal recov- ery control available for the airplane used in this test. From consideration of the fact that the maximum recovery capability of the airplane is dependent upon the elevator deflection per g relationship and the elevator available, a limitation in the elevator available becomes of primary concern. The signifi- cance of this relationship is further illustrated in the discussion of the controlled upset maneuvers.

Recovery by Elevator and Stabilizer

At the completion of the basic handling-qualities evaluation of the air-

plane, a series of controlled upset maneuvers were performed at the test alti-

tudes of 15000 and 35000 feet. The method employed to upset the airplane was

either to push over to the overspeed warning or to execute a stabilizer

"runaway" in either the full-nose-down or the nose-up direction. At the over-

speed warning, the throttle was reduced to idle and the recovery was initiated.

Figure 7 presents a stabilizer runaway performed at 35000 feet at an

initial Machnumber of 0.78. At time 1 second, the stabilizer was activated

in the full-nose-down direction and at 6 seconds, was driven to 3.5° airplane

nose down. In this period of 5 seconds the airplane went from level flight to

approximately -0.2g. At the time 10.7 seconds the throttle was reduced to idle

and at 12 seconds the pilot initiated recovery through the elevator. At time

18 seconds the full elevator available had been applied and a maximum of 1.3g

was attained. At the time of 20 seconds the stabilizer was activated toward

the trim position and as the normal acceleration increased the elevator input

was reduced until a time of 26 seconds# when the full recovery had been

effected and the airplane had reached an acceleration of 2.3g. The elevator

and acceleration traces between the increment of time of 18 and 20 seconds

indicate a loss in elevator effectiveness. The elevator remains essentially

constant in this region while the acceleration decreases from approximately

1. Sg to 1.Og. Also, at the time 18 seconds it is not apparent that the air-

plane rate of descent is at all reduced. However, after the stabilizer is

returned to the trim position, the acceleration increases, the rate of descent

is reduced, and the airplane returns to a level-flight condition.

Elevator-Stabilizer Trade

For the purpose of illustrating the horizontal-stabilizer power and the

effects of a mistrimmed stabilizer, figure 8 presents a summaryof the trades

in elevator and stabilizer for 1 g flight at altitudes of 15000 and 35000 feet.

The ratio of the elevator deflection to stabilizer deflection and of the stick

force to stabilizer deflection are presented over the Mach number range from

0.38 to about 0.85. The ratio of elevator angle to stabilizer angle indicates

that approximately 2° to 2.5 ° of elevator are required for every degree of mis-

trimmed stabilizer over the speed range tested. The corresponding stick forces

required to maintain level flight vary from approximately 15 to 35 pounds per

degree at 35000feet and 18 to 70 pounds per degree at 15000 feet. From these

data it is apparent that the available elevator for maneuvering is considerably

reduced. Also the increased workload associated with the high stick forces

tends to reduce the pilot's sensitivity to the actual accelerations experienced

in maneuvering.

In general, pilots have been taught to minimize stabilizer manipulation

while maneuvering. As shown in the previous time history, once the airplane

has been upset with a corresponding change in stabilizer setting and a recovery

is initiated, the retrimming of the stabilizer can be a valuable tool in

effecting a positive recovery. However, to use the stabilizer intelligently

under these circumstances, the pilot needs someacceleration reference through

a suitable g-meter to prevent overstressing the airplane.

Recovery by Elevator and Spoiler

Figure 9 presents a stabilizer runaway where the initial recovery is

effected through the elevator control and the final recovery is accomplished

through the deflection of the spoiler system. The general operational proce-

dures used in performing the upset were the sameas those employed in the pre-

vious maneuvers. As noted on the previous upset time history, there is a loss

of elevator effectiveness showing up between time 15 and 19.5 seconds. In this

region the elevator remains fixed at approximately 8° while the acceleration

decreases from 1.Sg to 1.1g. Also it maybe observed, through the airspeed and

altitude traces, that the rate of descent has not been checked until after the

spoilers have reached their maximum deflection of about 50°. In this maneuver

the spoiler handle was movedto the full spoiler position of 60° . Regardless

of the fact that only 50o of spoiler deflection were obtained_ because of the

blowdown effect at high speed 3 this much spoiler, in conjunction with the ele-

vator, allowed complete recovery of the airplane to a level-flight condition

while the stabilizer remained in the full nose-down setting.

In similar maneuversperformed with the stabilizer in the trim setting,

there was no problem in recovery with the elevator alone. However, in the

upsets that were performed with the mistrimmed stabilizer, it is obvious that

supplemental recovery control is required if additional elevator deflection

cannot be provided.

Dynamic Lateral-Directional Stability

During the evaluation of the basic handling qualities, the dynamic,

lateral-dlrectional stability characteristics were measured in smooth air in

order that the influence of the Dutch roll modecould be more intelligently

assessed in the flights where turbulence was penetrated. Figure lO presents a

summaryof the dynamic lateral-directional characteristics measured in terms of

the reciprocal of the time to dampto half amplitude over the speed range at an

altitude of 35000 feet. Results are shownfor conditions both with the damper

on and off.

The general observation in the comparison of these damping characteristics

is that the damping without augmentation is quite low and the engagementof the

yaw damper considerably improves the Dutch roll damping. On the basis of these

data and experience acquired in the turbulence flights, it was apparent that

the low level of Dutch roll damping increases the pilot workload when the air-

plane is upset or disturbed by gusts or turbulence, particularly with the yaw

damper inoperative. Also, as has been shownin reference l_ the utilization

of the yaw damper in turbulence, in general, relieves the tail loads by about

50 percent. Consequently, it is believed that the yaw damper should be con-

sidered an essential operational system on all transport airplanes of this type.

Turbulent-Air Penetrations During the turbulent-air portion of the Boeing 720 program, an attempt was made to penetrate either wave turbulence generated by terrain or high-altitude clear-air turbulence. The initial conditions investigated were at the high altitude during cruising flight. Howeverj very little or no turbulence was experienced at the levels between 33000 and 39000 feet. On the other hand, in the region of Denver and Colorado Springs, mountain-wave turbulence, which was classified by the pilots to be light to moderate, was encountered at 22000 feet.

The technique employed in penetrating the turbulent air was to stabilize at an initial indicated airspeed of 390 knots and_ once the turbulence was encoun- tered, to reduce the power and attempt to maintain a constant attitude and an .... _ _ 300 knots.

a_ro_ _ Figure ll presents a 9-second portion of the time histories acquired during three of these tests. The first test was made with the autopilot and yaw damper on, the second was with the autopilot off and the yaw damper on_ and the final test was made with both the autopilot and yaw damper off. The respective normal accelerations show that the peak value ranges from 0.3 to 2.5g at the pilot's station, 0.3 to 2.0g at the center of gravity and 0.1 to 2.6g at the tail. These peak accelerations are of a magnitude similar to those measured from flight recorders of airplanes involved in turbulence incidents; however, the persistence experienced in the illustrated tests was not as long.

In addition to the acceleration, it may be seen that there is an apparent oscillation measured at the pilot's station of approximately 4 to 6 cps. This superimposed motion was noted by the pilots to be quite disconcerting and disrupted their normal scan pattern. These studies indicate the need for improvements in the seat cushion and restraint system to reduce these motions.

The effects of the yaw damper and autopilot may be observed through motion of the pitch, roll, and yawing velocity traces. The progressive decay in Dutch roll damping is apparent as each of the systems is turned off and with both the yaw damper and autopilot off# the Dutch roll, previously discussed, becomes quite pronounced even though the general level of accelerations is consider- ably reduced.

Convair 880 Turbulence Penetration In the Convair 880 test program a series of turbulence flights were accom- plished at an altitude of about 29000 feet and the penetration was into a thun- -derstorm. The flights made through this region of turbulence were made with the autopilot off and the yaw damper on and off. The pilot's task was essen- tially the same as that required on the Boeing 720 airplane. The overall mag- nitude of turbulence was not much greater than that experienced with the Boeing 720; however, because of the persistence of the disturbance, the pilots at times classified the turbulence as moderate to severe. The transverse accelera- tion experienced during these flights was in the range from iO._g and was con- siderably greater than that evidenced in the Boeing 720 airplane. This result, however, could be due to the nature of the turbulence that the airplane was flown through.

C_CID'D_G_S With regard to the basic airplane characteristics, the analysis of the overall static stability characteristics of the airplanes tested appears to be satisfactory and acceptable to the pilots. Considering the inherent charac- teristics of the longitudinal-control system design and the flight envelope where the airplanes are being operated, the system can become marginal in the high Mach number and high dynamic-pressure range. The operational pilot should be thoroughly acquainted 3 either through simulator or in-flight demonstration, with the effectiveness of the longitudinal control system under adverse flight conditions beyond the normal operational boundary. Regarding the lateral- directional dynamics, the data obtained during the test program serve to reem- phasize the low damping of these airplanes, particularly without the operation of the yaw damper system; and the system should be considered, if it is not currently the practice, to be an operational item.

In the upset and recovery category, it is believed that the stabilizer or spoiler system can be effectively employed to assist the pilot in recovering from an upset attitude in the longitudinal mode. However, the increased pilot workload associated with the high control forces tends to reduce the pilots' sensitivity to the increased accelerations during recovery. Consequently, it is believed that a suitable g-meter should be provided in order that all sys- tems available, such as the stabilizer and spoilers, can be safely employed without overstressing the airplane.

From the turbulence data acquired during the test program it was observed that the pilot is subjected to frequency ranges around _ to 6 cps. Also it appeared that an improvement in the seat cushion and restraint system may be essential to reduce the shaking of the pilot and firmly anchor him in his seat. In this regard_ it is believed that the pilot motion in the seat is obviously the main contributor to the fact that his scan pattern is considez- ably disrupted. Finally an observation of the camera data, which were obtained in the cockpit, indicates that the flight instruments and panel are relatively steady and there is no apparent shaking of the indicator needles.

REFERENCE 1. Funk, Jack; and Cooney, T.V.: Some Effects of Yaw Damping on Airplane Motions and Vertical-Tail Loads in Turbulent Air. NASA M_0 2-17-_9L, 1959.

RESULTANT TOLERANCES FROM MISRIGGED CROSS SECTION OF ELEVATOR BALANCE SYSTEM ELEVATOR BALANCE PANELS _L STABILIZER BALANCE PANEL FORWARD-EDGE_2_ J Is ELEVATOR 10 POSIT_ON, 5 TEU 202S_•

OEO

Io COVE SE_ I_ 3 .2 .I 0 1 2 -s .2 .i 1 2 COVE GAP OUT OF TOLERANCE, IN _COVE SHELL Figure 2 Figure i FLIGHT-TEST LIMITS LONGITUDINAL-CONTROL- SYSTEM MANEUVERING CHARACTERISTICS MACH NUMBER : 0.84, ALTITUDE : 34,500 FT SOxlO 3 GROSS WEIGHT, 40 LB PULL 80 _- STICK JBORCE, 0 35 14o,ooo ELEVATOR 40 I 40 o_ -'o''°' o ou_ OF _OLERANCE 3O • IN TOLERANCE 8O INDICATED ALTITUDE, FT 2O %s_,_,_ °R, o -_ DEG 4 0.2 .3 .4 .5 .6 .7 .8 .9 8 ._ ._; i;2 ;;6 21o_.4 NORMAL ACCELERATION, g INDICATED MACH NUMBER Figure 4 Figure 3 LONGITUDINAL-CONTROL EFFECTIVENESS ALTITUDE = 35,000 FT TEU 16 M i = .54 MAXIMUM ELEVATOR AVAILABLE STABILIZER TRIM -- FULL A.N.D.

ELEVATOR O 8 ALTITUDE = 35,000 FT ANGLE, DEG COVE GAP o OUT OF TOLERANCE • IN TOLERANCE 16 ' TEU 24 MAXIMUM CONTROL-TAB TEU 16 2O "_:_C_:_ Mi = .84 ELEVATOR ELEVATOR ANGLE, DEG DEFLECTION DEFLECTION, DEG 1!

1 I I 16 8 D 8 lt6 24 TEU 4 5 .'6 .'7 ._ .; _o ELEVATOR CONTROL TAB, DEG INDICATED MACH NUMBER Figure Figure 6 STABILIZER TRIM RUNAWAY A.N.D.

RECOVERY -- ELEVATOR/STABILIZER IF RECOVERy I STARIUZER _s_zE:__ Fu,,_,_ R i _ ........ v ......

I AND. I I ) APP ED I TO TRIM I I THROTTLELkt I ct4oP _/1 _ I INDICATED 9 t

MACH B_ r_,_[nll ]

NUMBER 17 [ I I I INDICATED 35x103 NORMAL 2 ACCELERATION, 1 g 0 _,,c._LE, lO

AND_ I I J

ELEVATOR TEU 10 POSITION, S DEG S

HOR'ZONTAL AN05_ i I i E I I ! LI__+_ STAB'L,_ER O

POSITION, DEG 5 PULL 150 LONGITUDINAL lOD STICK FORCE, LB SO 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 TIME, SEC Figure ?

LONGITUDINAL TRIM ELEVATOR/STABILIZER TRADE ALTITUDE, FT INCREMENTAL • ELEVATOR INCREMENTAL _3s,ooo STABILIZER I I I I I BO[ STICK FORCE J iS,ODD INCREMENTA'L'40 J.

STABILIZER J INCREMENTAL 60 I- LB/DEG 20 [ 3s,ooo o ._ .:_ .;:, .;_ ,._ INDICATED MACH NUMBER Figure 8 STABILIZER TRIM RUNAWAY A.N.D.

RECOVERY -- ELEVATOR/SPOILER I O_ENE0 INDICATED MACH NUMBER INDICATED ALTITUDE,

t r t

FT NORMAL ACCELERATION, C.G., g ANU PITCH ANGLE, 0 DEG 10 POSITION, 2

SPO,LER TEU"_I_____j___ t t t ! ! !

DEG ELEVATOR TEU I0 POSITION, 5 DEG 0 STABILIZER

HO"ZONTAL-ANUiI____! ! p ! I I I ! I I I I I -_

POSITION, DEG PULL 300 STICK FBORCE, 100 O 8 I 4 TIME, SEC Figure 9 LATERAL-DIRECTIONAL DYNAMICS ALTITUDE = 35,000 FT _.2l,- _ -,Aw DAM,,' oPP o/ YAW DAMPER ON /o 1 .8 [ // TIME TO-DAMP TO - . ONE.HALF AMPLITUDE,.4 j- // I I "_D,CAT_O MAC. N.M_" Figure lO COMPARISON OF AUTOPILOT/YAW DAMPER INFLUENCE IN TURBULENCE AUTOPILOT ON AUTOPILOT OFF AUTOPILOT OFF YAW DAMPER ON YAW DAMPER ON YAW DAMPER OFF ACCELERATION, .e. .

NORMAL _ . --- _ _ ---. _ _ ________ __ ACCELERATION, T_, _" I m.,_J,_ T'_,_ i'-"l C.G., g o _ ='_---_-_------_=_ I I" J I i , --'-_ NORMAL 2 ACCELERATION, _, _ ANU .2_ _ PITCHING VELOCITY, t-----_-I I I J __ ROLL VELOC,Ty, " I--_ _-_-_-_ RAD'AN/_EC. IX___ _ RADIAN/$EC _ _ --V IL---_J • 1 2 3 4 5 O 1 2 3 4 5 O I 2 3 4 5 TIME, 5EC Figure ii 15. SIMULATOR INVESTIGATIONS OF THE PROBLEMS OF FLYING A SWEPT-WING TRANSPORT AIRCP_ IN HEAVY TURBULENCE By Richard S. Bray and William E. Larsen NASA Ames Research Center SUMMARY An investigation of several factors which may contribute to the problem of piloting jet transport aircraft in heavy turbulence was conducted by using a piloted simulator that included the most significant airplane response and cockpit vibrations induced by rough air. Results indicated that the primary fuselage structural frequency contributed significantly to a distracting cock- pit environment, and there was obtained evidence of severely reduced instrument flight proficiency during simulated maneuvering flight in heavy turbulence. It is concluded that the addition of similar rough-air response capabilities to training simulators would be of value in pilot indoctrination in turbulent- flight procedures.

INTRODUCTION During the past 3 or 4 years, jet transport aircraft have been involved in a number of serious incidents resulting from encounters with heavy atmos- pheric turbulence. These experiences fall into two categories: those resulting from encounters with clear-air turbulence that caused structural damage and personal injury, and those that involved encounters with storm turbulence, while in instrument flight, that resulted in partial or complete loss of con- trol of the aircraft. The clear-air turbulence problem indicates the need for meteorological detective work, but the storm turbulence presents a less defined problem and has inspired a review of turbulent-flight procedures among all segments of the airline industry. In response to this general concern, the National Aeronautics and Space Administration instituted a program of investi- gations to examine several factors which might contribute to piloting problems in severe turbulence. Wind-tunnel tests of a transport model at the Langley Research Center and flight tests of a Boeing 720 airplane conducted by personnel of the Flight Research Center were performed for the purpose of augmenting aerodynamic and stability and control data for the swept-wing transport category of airplane. Concurrently, at the Ames Research Center 3 an investigation of rough-air piloting problems was conducted by using a piloted simulator that has unique capabilities regarding creation of the physical environment in the cock- pit during flight in turbulence. This paper summarizes the results and obser- vations obtained in the Ames investigation.

The simulation program was intended to pursue the following objectives:

(a) Examine aircraft response to turbulence

(b) Evaluate handling qualities in turbulence

(c) Study effects of cockpit accelerations on conduct of flight tasks

(d) Evaluate the simulator as a training device

AIRPLANE SIMULATION

Cockpit.- Since manyof the first-hand accounts of incidents in heavy

turbulence stressed the distracting aspects of cockpit accelerations and vibra-

tions, the simulation used in the Amesprogram was designed to utilize a device

known as the height control simulator. The simulator cab is mounted on a ver-

tical track that provides lO0 feet of travel. (See fig. 1.) High-performance

electrical servomotors drive the cab through a cable system. The simulator

cockpit, illustrated in figure 2, employed basic controls and instruments sim-