You are on page 1of 216

Journey in Aeronautical Research

A Career at NASA Langley Research Center

W. Hewitt Phillips

NASA History Office

Office of Policy and Plans
NASA Headquarters
Washington DC 20546

Monographs in Aerospace History

Number 12
November 1998
The use of trademarks or names of manufacturers in this
monograph is for accurate reporting and .does not constitute an
official endorsement, either expressed or implied, of such pro-
ducts or manufacturers by the National Aeronautics and Space
William Hewitt Phillips, January 1945, age 26, then head of the
Stability and Control Branch, Flight Research Division.
Table of Contents

List of Figures ix

CHAPTER1 Introduction I

CHAPTER2 Early Life Through the College Years 3

Parents and Early Years 3

College Years 11

CHAPTER 3 Setting for the Research Work at Langley 17

History 17
Description of Surroundings 17
First Assignments in Flight Research 19

CHAPTER 4 Flying Qualities 21

Flying Qualities Tests 22

Stall Tests 26
Brewster XSBA- l Airplane 27
Short-Period Longitudinal Oscillations 28
Spiral Instability 29
Snaking Oscillations 30
Rudder Deflection Required for Trim 31
Progress in Flying Qualities Research 32
Static Longitudinal and Directional Stability 33

CHAPTER 5 Analytical Studies 37

Graphical Solution of the Quartic Equation 40

Solution of a Nonlinear Problem 41
Computation of Lateral Oscillation Characteristics 44

Monographs in Aerospace History Number 12--Journey in Aeronautical Research v

CHAPTER6 Problems Encountered as a Result of Wartime
Developments 47

The Quest for Reduced Control Forces 48

Flettner Tabs, Servo Tabs, Spring Tabs, and Whirlerons 50

Research on Closely Balanced Controls 53
A Control Surface Hinge-Moment Balancing Mechanism 56
A Spoiler Operated Servo Control 57
The Introduction of Swept Wings 58

CHAPTER7 Effects of World War H on Research Activities 61

Effect of the War on Living and Working Conditions 61

Some Effects of the War on Research Activities 62

CHAPTER 8 New Ideas--The Fruits of Research 67

The First Piloted Simulator Used for Research at Langley 67

Roll Coupling 69
A Method for Determining the Moments and Products of Inertia of
Full-Scale Airplanes 74
Towed Bodies 75

CHAPTER9 Unsteady Lift 77

Spring-Tab Flutter 78
Effect of Unsteady Lift on Snaking Oscillat_ ons 81

CHAPTER 10 An Automatic Aileron Trim Device fc=r Personal

Airplanes 85

Development and Tests of the Trim Device 85

CHAPTER 11 New Areas of Research 91

The Wing-Flow Technique 91

Power Controls and Pilot-Induced Oscillations 97

Human Response Characteristics 101

Landing Studies 102

CHAPTER 12 Measurement of Turbulence in the Atmosphere 105

Analytical Studies of the Accuracy of Turbu ence Data from Flight

Records 106

Measurements of the Spectrum of Atmospheric Turbulence Over a

Large Range of Wavelengths 109

vi Monographs in Aerospace Hi ¢tory Number 12--Journey in Aeronautical Research

CHAPTER 13 Gust Alleviation 113

Background and Analysis of Gust Alleviation 115

Design and Test of a Gust-Alleviated Airplane 121

Analytical Studies of Additional Problems 128
Future Possibilities of Gust Alleviation 129

Review of Work by Ren6 Hirsch 130

Later Studies by Other Investigators 132

CHAPTER 14 Lateral Response to Random Turbulence 133

Modeling the Turbulence 133

Calculation of Response 134

CHAPTER 15 Accident Investigations 137

BAC-I 11 Crash Near Omaha, Nebraska, on August 6, 1966 137

Crash of Convair B-58 Bomber 140

Accidents of the Republic F-105 Thunderchief Airplane 144

Analytical Study of the F-84 Following Accident in High-Speed
Flyby 145

Accidents of the Beech Bonanza Involving In-Flight Structural

Failures 146

CHAPTER 16 NACA Flight Research on the Vought F8U-1

Airplane 151

Results of the Last NACA Flying Qualities Investigation 152

CHAPTER 17 A Couple of Bloopers! 155

Improving on the Wright Brothers 155

Relearning the Laws of Angular Momentum 158

CHAPTER18 Administration of the Langley Research Center 159

Organization of the Center 159

NACA Reports 160
Approval and Support of Research 162

The Committee Organization of the NACA 163

Formal NACA Conferences 164

Preparation of NACA and NASA Talks 165

Meetings of Professional Societies 165
Meetings with Visitors 166

Monographs in Aerospace History Number 12--Journey in Aeronautical Research vii

CHAPTER 19 Dawn of the Space Age 167

References 169

APPENDIX I Early Autobiography 175

APPENDIX II Bibliography of Reports by William Hewitt

Phillips 181

NACA or NASA Reports 181

Papers Other Than NACA or NASA Report_ 184
Unpublished Papers in NASA Langley Research Center
Library 186

APPENDIX III Solution of a High School Geometry Problem 187

APPENDIX IV Solution of Quartic Equation in Terms of Two

Parameters by Use of Charts 191

Index 195

About the Author 201

viii Monographs in Aerospace Hi,:tory Number 12--Journey in Aeronautical Research

List of Figures

Figure 2.1 Scenes from childhood. 4

Figure 2.2 My father's steam-powered torpedo boat destroyer

after restoration. 5

Figure 2.3 Our summer cottage at Long Beach, Gloucester,

Massachusetts. 6

Figure 2.4 Scenes in Belmont, Massachusetts. 7

Figure 2.5 My nonflying scale model airplanes. 8

Figure 2.6 Vultee YA-19 attack bomber with Pratt and Whitney
R-2800 engine installed. 13

Figure 2.7 Mechanics in the Installation Department of Pratt and

Whitney in front of YA-19 airplane. 14

Figure 2.8 Model glider equipped with phugoid damper. 15

Figure 3.1 Airplanes being tested at the NACA Flight Research

Hangar in 1944. 20

Figure 4.1 Pictures of me with two of the airplanes tested in my

first years at the NACA. 23

Figure 4.2 Vought-Sikorsky XF4U-1 airplane. 24

Monographs in Aerospace History Number 12--Journey in Aeronautical Research ix

Figure 4.3 Curtiss P-40 airplane. 24

Figure 4.4 Hawker Hurricane airplane. 25

Figure 4.5 Supermarine SpiOqre airplane. 25

Figure 4.6 North American BT-9B airplane. 27

Figure 4.7 Lockheed 14H transport airplane, a later model of one

of the original 16 airplanes on which Gilruth's flying
qualities requirements were based. 29

Figure 4.8 Grumman XTBF-3 airplane, a torpedo bomber used in

the war in the Pacific. 33

Figure 4.9 Portion of tail area that must be considered ineffective

to obtain agreement between calculated and measured
variation of rudder angle with sideslip angle (low-wing
airplanes). 34

Figure 5.1 Time histories of aileron snaking oscillations as

calculated by my theory. 43

Figure 6.1 Types of control surface aerodynamic balance. 49

Figure 6.2 Diagramatic sketches of aerodynamically operated

servo controls. 51

Figure 6.3 All-movable horizontal tail installed on Curtiss XP-42

airplane. 54

Figure 6.4 Comparison of stick forces in rapid l,ull ups with

Curtiss P-40 airplane and Curtiss XJ_-42 airplane. 55

Figure 6.5 Diagram of operation of vane-type s,,rvo control. 58

Figure 6.6 Photograph of test model of vane-type servo

control. 59

Figure 6.7 Bell L-39 airplane incorporating experimental swept

wing to study low-speed effects of wing sweep. 60

x Monographs in Aerospace Hi¢tory Number 12--Journey in Aeronautical Research

Figure 7.1 Republic P-47D-30. 64

Figure 7.2 Mitsubishi O0 airplane, popularly known as the

Japanese Zero. 64

Figure 8.1 Yaw chair used for studying ability of human pilots to
control lateral oscillations. 68

Figure 8.2 Drop model of the XS-1 airplane. 7!

Figure 8.3 Flight data from drop test of XS-1 model. 71

Figure 8.4 Devices made to illustrate the effect on stability during

rolling motion of the equality or inequality of stiffness
about the pitch and yaw axes. 73

Figure 8.5 Test airplane as set up for determining moment of

inertia and tilt of the principal longitudinal axis of
inertia. 75

Figure 9.1 Diagram showing optimal location of mass-balance

weight to prevent flutter of spring tab. 79

Figure 9.2 Schematic diagram of installation of tab balance

weight ahead of flap hinge line. 79

Figure 9.3 Drawing of balance weight installation in B-34 vertical

tail. 80

Figure 9.4 Photograph of balance weight installation in B-34

vertical tail. 81

Figure 9.5 Amplitude and phase angle of the circulatory lift of

elliptical wings. 83

Figure 9.6 Damping of oscillations of a weathervane computed by

finite-span, quasi-static, and two-dimensional
theories. 84

Figure 10.1 Schematic diagram of automatic aileron trim control

device. 87

Monographs in Aerospace History Number 12--Journey in Aeronautical Research rd

Figure 11.1 Wing-flow model and test apparatus mounted on P-51
wing. 92

Figure 11.2 Drawing of semispan model of rectangular wing with

full-span flap used in wing-flow test_. 93

Figure 11.3 Typical data from wing-flow test. 94

Figure 11.4 Wing-flow model of Vought F7U-1 Cutlass. 96

Figure 11.5 Full-scale Vought F7U-1 Cutlass. 97

Figure 11.6 Simple simulator used to demonstrate to pilots the

possibility of pilot-induced oscillations caused by
characteristics of the power control system. 99

Figure 11.7 Normal acceleration during formation flights of F4U-

4B airplane and F9F-3 airplane with alternate
positions as lead airplane and following airplane;
F4U-4B with and without power controls. 100

Figure 12.1 Error in harmonic amplitude due to finite length of

record for various phase angles of harmonic. 107

Figure 12.2 Plot of amplitude versus phase angle for waves with
various ratios of wavelength to record length. 108

Figure 12.3 Photograph of F9F-3 airplane show, ng special

instrumentation for turbulence meas.trements. 109

Figure 12.4 Power spectrum of gust vertical velocity for

wavelengths of lO feet to 60,O00 feet 110

Figure 13.1 Waterman's airplane incorporating _eings attached to

fuselage with skewed hinges. 114

Figure 13.2 Concept for complete alleviation of vertical gusts

shown by the theoretical analysis. .!21

Figure 13.3 Two-view drawing of the Beech B-18 test airplane

showing the modified control surface s. 122

xii Monographs in Aerospace Hixtory Number 12 Journey in Aeronautical Research

Figure 13.4 Nose boom and angle-of-attack vane installation on
Beech B-18 test airplane. 123

Figure 13.5 Three-quarter rear view of Beech B-18 test airplane

showing elevator control split into three
segments. 124

Figure 13.6 Three-quarter rear view of Beech B-18 test airplane

showing modified flap system with oppositely defected
inboard segment. 124

Figure 13.7 Comparison of flights in light turbulence, basic

airplane, and gust-alleviated airplane. 126

Figure 13.8 Comparison of power spectral densities of normal

acceleration and pitching velocity for basic airplane
and alleviated airplane. 127

Figure 13.9 Comparison of normal acceleration of basic airplane

and alleviated airplane shown on linear scales on two
different types of plots. 127

Figure 13.10 Flight photograph of Reng Hirsch 's first airplane as

equipped with larger engines. 131

Figure 13.11 Ren_ Hirsch in front of his Aerospatial Rallye light

airplane equipped with gust-alleviation system. 131

Figure 14.1 Relations between the power spectrum of rolling gusts

and the point spectrum of turbulence for various values
of the ratio of scale of turbulence to wing span. 135

Figure 15.1 The BAC-111 Transport Airplane. 138

Figure 15.2 Rolling acceleration and rolling velocity of BAC-111

airplane as determined from voice recorder
analysis. 139

Figure 15.3 Model of damaged BAC-111 used for catapult tests to

determine motion following breakup. 140

Figure 15.4 B-58 Hustler bomber airplane. 141

Monographs in Aerospace History Number 12--Journey in Aeronautical Research xiii

Figure 15.5 Beech B-35 Bonanza airplane as tested at
Langley. 147

Figure 15.6 Measured and extrapolated stick force in straight flight

as a function of dynamic pressure for Beech B-35
Bonanza airplane. 148

Figure 15.7 Estimated stick force as a function oj: indicated airspeed

and normal acceleration for Beech B-35 Bonanza
airplane. Trim speed 120 miles per hour. 149

Figure 15.8 Estimated stick force as a function oJ indicated airspeed

and normal acceleration for Beech B-35 Bonanza
airplane. Trim speed 145 miles per hour. 149

Figure 16.1 Vought F8U-1 airplane showing variable-incidence

wing in raised position. 152

Figure 17.1 Sketch of model used to illustrate the use of sweptback

rudder horns to offset centering tenaency of control
springs. 156

Figure 17.2 Fairchild PT-19 airplane used for two-control

experiments. 157

Figure 17.3 Sketch of modified directional contrc I system. 157

Figure 17.4 Sketch of concept for unsuccessful device to apply

oscillating moments to an airplane in flight. 158

Figure 18.1 Organization chart for the Research Department of the

Langley Memorial Aeronautical Lab 9ratory in
1944. 160

Figure 18.2 Instructions to the members of an editorial

committee. 161

Figure 18.3 Certificate of appointment to the NACA Research

Advisory Committee on Control, Gutdance, and
Navigation. 163

_v Monographs in Aerospace Hi vtory Number 12--Journey in Aeronautical Research

CHAPTER1 Introduction

My career to the present has covered these papers are rarely read or even referred
58 years, all at Langley Research Center in to by persons involved in current projects.
Hampton, Virginia. At the start of my work,
The cumulative output of the center and its
the center was called the Langley Memorial
contributions to the development of the air-
Aeronautical Laboratory of the NACA
craft industry, however, are recognized as
(National Advisory Committee for Aeronau-
being substantial, and an overview of this
tics). With the advent of the space program,
history has been published (ref. 1.1). The
it became the Langley Research Center of
present volume gives a different perspective
NASA (National Aeronautics and Space
Administration). In subsequent discussions, on the history of Langley by documenting in
the name of the center will be abbreviated to some detail the experience of my work as an
engineer involved largely with the flight test-
simply Langley. During this entire time, my
ing of full-scale airplanes in the Flight
primary interest has been in research in aero-
Research Division. This volume may also
nautics and in the related problems of space
serve to convey some knowledge of my work
flight. Actually, I have had two parallel
to the general public and to later generations
careers because my interest in aeronautics
of engineers. Though the emphasis in these
started at a much earlier age, about 9, with
discussions is on the technical aspects of the
the building and flying of model airplanes.
work, some autobiographical notes on my
This interest has continued through my life.
own background and education may be of
The model airplane hobby has had a power-
ful influence in contributing to my interest in
aeronautical research. These hobby activities The contents of this volume have been
will not be mentioned in this account except selected with the following considerations in
when they contributed directly to research on mind. First, because most of my more impor-
full-scale airplanes. tant research has been published in the
NACA and NASA reports or in technical
During this period, many advances have journals, all of which are readily available in
occurred in the art and science of aeronau- the NASA libraries, no attempt is made to
tics. My own contributions, like those of present the technical contents of these
other research workers, have largely taken reports. Some of the projects reported will be
the form of published reports or journal arti- mentioned to indicate their background or
cles. Because of the progress in this field, importance. Some other technical work,
interest in research done in earlier years can however, was never published, either
be expected to have declined to the point that because of a lack of general interest or

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 1


because the results were not conclusive. In a equipment and many facilities immediately
few cases, a report that I considered interest- available and with apparently little supervi-
ing was turned down by the editorial com- sion of the technical aspects of the work.
mittee. I will include these items in the dis- There is no doubt that the research environ-
cussion. In addition, I will discuss some ment at Langley was favorable to the devel-
things that I have learned from my research
opment of new ideas while keeping a strong
projects that did not appear in published
focus on the primary objective of aeronauti-
reports. In appendix I, a brief autobiography
cal research. I will insert chapters in the
of my early years, written at the age of 14, is
account at appropriate places to comment on
presented. Appendix II contains a complete
this aspect of the research work.
listing of my professional reports and publi-
The subjects discussed in this autobiography,
My experience at Langley has placed me in a
though not highly technical, require use of
position to discuss not only my personal
some concepts and nomenclature familiar to
activities, but to comment on some aspects of
aeronautical engineers, but possibly unfamil-
the research environment and how it changed
iar to persons working in other fields. To
through the years. Someone reading the early
chapters of this account might wonder how a make the material more readily understood, I
young engineer could become engaged in so will preface some chapters with brief discus-
many projects of great interest for both mili- sions of the background and the terms used
tary and civil aviation with much research in the topic under consideration.

2 Monographs in Aerospace Hi _tory Number 12--Journey in Aeronautical Research

CHAPTER 2 Early Life Through the College

My birth was shortly before the end of World

Parents and Early
War I, and my mother told how she pushed
Years me in my carriage waving a flag on
Armistice Day. Soon after, in 1920, my
I was born on May 31, 1918, in Port
father was sent to the United States by Lord
Sunlight, Cheshire (now Merseyside),
Leverhulme, founder of Lever Brothers
England. This city, a so-called model village
Company, to introduce a new line of cosmet-
across the River Mersey from Liverpool, was
ics. Unfortunately, in 1920 came the postwar
built by the Lever Brothers Company as a
depression, and the business climate was not
place for their workers to live. My father,
right for introducing a new product. My
William Phillips, was employed by this com-
father was given a job at the Lever Brothers
pany. His background was in chemistry and
Plant in Cambridge, Massachusetts and was
prior to working at Lever Brothers, he had
given the choice, after three months, of
been employed in several chemists shops
returning to England or staying in America.
(the British name for drug stores) in which
He elected to stay in America. He returned to
he had obtained a good knowledge of phar-
England to arrange for the movement of the
maceuticals, cosmetics, and perfumes.
family and belongings. We all sailed to the
My mother, Bertha Pugh Phillips, prior to United States and resided first in an apart-
her marriage had been headmistress of the ment on Commonwealth Avenue in Boston,
Evelyn Street School, a large infants school Massachusetts. Later we moved to the first
in Warrington where the students were chil- floor of a two-family house near the
dren in what would, in America, be called Charles River in Watertown, Massachusetts
grades one through three. Her employment (figure 2.1).
in such a position at the early age of about 28
was most unusual in England at that time and My father became the chief perfumer at
indicates her leadership and educational abil- Lever Brothers Company and was responsi-
ity. She had already written articles and a ble for the perfume in the Lever Brothers
book and gained some note in England in the products such as Lux Toilet Soap, Lifebuoy,
field of infant education. My parents, I am and Rinso. Although his work did not
sure, gave considerable thought as to involve mechanical knowledge, his hobby
whether such a promising career should be was in the field of electrical and mechanical
interrupted by marriage. In those days, the devices. He had acquired a metal-turning
wife was expected to become a homemaker lathe (a Drummond lathe, especially
after marriage. intended for hobbyists) while still in England

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 3

Parents and Early Years

FIGURE 2.1. Scenes from

childhood (clockwise
starting in upper left).

(a) With my mother in

England, age 9

(b) Riding my car in

Watertown, MA.
Our house in

(c) With my parents and

baby sister, Hilda
Evelyn Phillips, in
Watertown, MA.

(d) On the Charles

River, where the
steamboat was
tested. Our house in

and had constructed a steam engine Another model made by my father was a
machined from commercially available cast- steam-powered torpedo boat destroyer that
ings. He acquired this interest from his was about four-feet long. One of my earliest
father, Thomas Phillips, who also did lathe memories is the test of this model, when I

work. Thomas Phillips must have had a was five years old, in the Charles River near

remarkable interest and ability in technical our house. The model, pulling a piece of
matters. He started life as an uneducated coal string for its retrieval sailed off with unex-
pected speed and soor_ became a speck in the
miner in Yorkshire, but was taught to read by
distance. The model _ as pulled back against
his wife and continued to improve his knowl-
the full thrust of the engine and shipped
edge through reading and hobbies. My
water over its low sides. It eventually
father, before he died, wrote a brief biogra-
swamped and sank with a spectacular cloud
phy of Thomas Phillips which also gives an of steam about 20 feet from shore. The
insight into my father's own interests and model was brought up from the bottom of
early life. After coming to this country, my the river about a week later, but after that, my
father made an electric generator to be driven father lost interest in the steamboat and it
by the steam engine. The combined model is was kept in our baser _ent for many years. I
still on display in our home. My father was restored the model ;rod equipped it with
also a pianist of moderate ability. radio control a numbtr of years ago, before

4 Monographs in Aerospace Hi _tory Number 12--Journey in Aeronautical Research

Parents and Early Years

my father's death and it is now on display in This review of my childhood is not by any
my living room (figure 2.2). means intended as a complete account of my
activities, but I will mention a number of
Immediately after his test of the steamboat, things I did that illustrate my interest in avia-
tion and some of the factors that influenced
my father became interested in crystal radio
the growth of this interest. The decade of the
sets, which were just becoming popular
about 1923. The construction of successively 1920's was a period of unusual development
more complex radios occupied his interest in aviation with new designs and records for
endurance and speed being reported fre-
for a number of years.
quently in the newspapers. In Watertown, I
had seen the dirigible Shenandoah fly over
My recollections of life in Watertown are
and in Belmont, airplanes like Curtiss
rather sketchy, but I do recall playing with
Jennies were observed, sometimes perform-
wind-up trains and learning to balance on a
ing stunts such as loops and spins.
scooter. My sister, Hilda, was born there.
Lindbergh's flight in May 1927 made a deep
The family made a visit to England during impression on me, as it did on young people
the summer of 1923. In 1924, we moved to a throughout the country. I was particularly
single-family house in Belmont, Massachu- influenced by the suspense caused by the
setts. I started in the public schools there in length of time Lindbergh was in the air with
the second grade at the age of six, having news of the takeoff coming one morning;
been taught to read by my mother. I was sightings of the plane over Newfoundland
therefore a year younger than my classmates that evening; the plane appearing over
throughout my schooling. Ireland the following day; and finally, the

FIGURE 2.2. My father's

steam-powered torpedo
boat destroyer after

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 5

Parents and Early Years

FIGURE 2.3. Our summer

cottage at Long Beach,


landing at Le Bourguet reported in papers the open all summer. This land had been given
day after that. I had never had any idea that to the town by Mr. Underwood, a descendant
an airplane could stay in the air that long. of one of the earliest families in Belmont
Somewhat later, when I was in junior high and head of the Underwood Deviled Ham
school, the Army Air Corps put on a very company. The pool, established in 1906, was
large demonstration in which practically its the first outdoor public swimming pool in the
entire fleet of aircraft was in the air at the United States. I used all these facilities, but
same time. The sky was filled with forma- the aspects that really helped my hobbies
tions of 50 to 100 airplanes such as Keystone were the ability to glice models down the hill
Bombers and DH-4 biplanes. My diary entry and to sail model boats in the pool
for that day says "782 planes flew over." The (figure 2.4).
sound of scores of Liberty motors droning
At the top of the street lived a boy that I
away simultaneously will never be forgotten.
played with whose father, a research doctor
My first model airplanes were paper gliders at Harvard, also was interested in aeronauti-
with about a 5-inch wing span that were pat- cal experiments and had a supply of balsa
terned after Lindbergh's airplane. Our family wood. He built solid balsa gliders of two- to
rented a small cottage at Long Beach, near three-foot span with h rag, slender wings and
Gloucester, Massachusetts, for about a short fuselages. When these models were
month in the summer, with my father coming launched from the top of the playground,
up on weekends (figure 2.3). On rainy and they would glide quite a distance. When they
foggy days, I would fly the paper models in were thrown harder, _hey would do a loop
the cottage where the high, unfinished room and continue the fligh with a series of oscil-
with a peaked roof gave plenty of space for lations. I built a numb,_'r of these gliders.
the models to perform stunts. I learned a lot
My grandmother Phill Lpsin England sent me
from these models, as I have described in an
a pocket line-a-day di uT for 1929. As might
article (ref. 2.1 ).
be expected at the ag_ of 11, my entries in
Back in Belmont, we were fortunate to live this diary were rather brief, but among other
in a house across the street from the Under- things I did make a note of what models I
wood playground that had a large grassy area was building and of the number of airplanes
with a good slope down from the top, a level that flew over each day. This diary also
area, and then another slope down to the bot- started me in the habit, beginning in 1930, of
tom. The playground contained swings, rings getting 5-year line-a-day diaries that I have
for gymnastics, improvised ball fields, and at kept, with some breals, throughout my life.
the bottom, a swimming pool, which was These diaries, which recounted mainly my

6 Monographs in Aerospace Hi_tory Number 12--Journey in Aeronautical Research

Parents and Early Years

FIGURE 2.4. Scenes in

Belmont, Massachusetts.

(a) Our house (top).

(b) Swimming pool

across the street
about 1940 (bottom).

Monographs in Aerospace History Number 12 Journey in Aeronautical Research 7

Parents and Early Years

FIGURE 2.5. My nonflying

scale model airplanes.

(a) Bellanca Skyrocket,

made at age 14 (top).

(b) Pitcairn Autogiro,

made at age 15

8 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Parents and Early Years

hobby and social activities rather than my department store, and the Boston Traveler, a
professional work, are useful in establishing newspaper (ref. 2.2). My mother enrolled me
the dates and sequence of various activities in the club, but I did not immediately take
that I would otherwise have forgotten. part in the activities because at the age of 11
I was too young to travel into Boston by
I had little guidance in my modeling activi- myself on the street cars and subways. I did,
ties, but I did find in the children's section of however, get some information on more cur-
the Belmont Public Library a book entitled rent model designs and materials, such as
Model Airplanes of 1911. I read this book balsa wood strips and sheets. I subscribed to
numerous times, not realizing that the techni- the magazine Model Airplane News, a
cal material in the book was largely incorrect McFadden publication, starting with the first
or that the models described were obsolete. issue in June 1929.
Model airplanes in 1911 were mostly rubber-
On my eleventh birthday I received as a
powered twin pushers that were constructed
of spruce and pine and braced with music present an Ideal Every Boys Airplane, an
wire or bamboo. I made a model like those ingenious but heavy rubber-powered flying
model kit first marketed in 1922. Later, I
pictured in the book. The propellers were
received a Silver Ace, a potentially better
carved from blocks of soft pine obtained by
flyer. The Every Boys Airplane flew about
my father from the carpenter shop at Lever
100 feet, like my twin pusher. I got very sick
Brothers. The wing had a frame of music
wire with ribs soldered in and the canard tail for a few days after building the Silver Ace,
was a piece of cardboard curved to a cam- probably from inhaling the banana oil fumes
in a closed room and I never got it to fly.
bered airfoil shape. The model made a very
Soon, however, I was building balsa models
successful flight in the playground and cov-
of my own design that flew much better. I
ered about 100 feet in stable flight in a
also built small rubber-powered models with
straight line. Naturally, this flight gave me
great encouragement and the incentive to about a seven-inch span that I would fly in
build more models. the living room. I sold some of these models,
now called parlor planes, to my classmates in
junior high school for 25 cents a piece. This
My interests were not solely confined to
price was a real bargain since each model
models. I participated in all the sports
had a hand-carved propeller.
offered by the playground, including sandlot
baseball and swimming in the summer and
By 1932 at the age of 14 I was able to travel
sledding, skiing, and ice hockey in the win- into Boston by myself and started to attend
ter, though I was at this age rather small and regularly the Junior Aviation League meet-
never much of an athlete. By the age of 13, I ings and activities. The club held weekly
had learned to use my father's lathe. I made a
meetings during the winter and monthly con-
solenoid-operated electric motor (still in tests indoors in winter and outdoors in sum-
existence), a cannon that shot a cork when
mer. A building contest was also held each
loaded with a firecracker, and a crude but
year for a specified nonflying scale model. I
workable compressed-air motor. Later, as I
built a Bellanca Skyrocket in 1933 and a
devoted more effort to model airplanes, not
Pitcairn PA-18 Autogiro in 1934 (figure 2.5).
much use was made of the lathe because
model airplanes do not require precision The autogiro, in particular, was a very com-
machined parts. plicated scale model subject. These projects
were very time-consuming, but I learned a
In 1929, a large model airplane club, the lot about full-scale aircraft construction and
Jordan--Traveler Junior Aviation League, about how an autogiro flies. I also built
was started in Boston. The club was spon- indoor and outdoor rubber-powered models
sored by the Jordan Marsh Company, a and by 1934, I was competing on even terms

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 9

Parents and Early Years

with the best flyers in the League. In 1935, I flight was exceeded in 1995 with a flight of
was selected as one of a team of four flyers 63 minutes 54 seconds. A unique situation
from the League who were given all- existed in the early stages of model develop-
expense-paid trips to the National Model ment in that the teenagers building and flying
Airplane Contest in Saint Louis, Missouri. the models knew much more about model
This contest is an annual event called by design and construction than their adult advi-
model enthusiasts the Nationals, or Nats for sors who organized the sport.
short. The winners, after returning from this
trip, were each given one of the newly devel- The Director of the Junior Aviation League
oped small gasoline model engines for pow- at the time of my participation was Willis C.
ering model airplanes. I also was on the team Brown, a manual arts instructor in the
to attend the Nats in Detroit in 1936 and Arlington, Massachusetts schools and an
1938. My main accomplishments in these amateur radio hobbyist. He was always inter-
contests were winning second place in the ested in the technical aspects of model air-
gasoline-powered Texaco event in 1936 with planes and in 1936, he organized a project
a flight of 30 minutes 12 seconds and first for the League members to build a wind tun-
place in the Stout event for indoor stick mod- nel for testing indoor model airplanes. At
els in 1938 with a flight of 21 minutes that time, I was attending MIT and became
56 seconds. I was flying in the Senior cate- the chief participant in the project (ref. 2.3).
gory for flyers under age 21. The Open Class The wind tunnel was unique in design and
for flyers over 21 did not exist in the early had a diameter of 5 feet and a length of 16
days of the League. Nowadays in the large feet with a airspeed ranging from 2.5 to
contests, almost all the flyers are grown-ups 4 feet per second. This wind tunnel required
or senior citizens. Young people are now typ- two years to construct followed by a year
ically more interested in computers and other devoted to testing and research. I learned
hobbies than in model airplanes. This lack of much about aerodynamics and instrumenta-
interest in model airplanes is due partly to tion from this work, particularly since almost
the advanced state of development of models all the problems were encountered a year or
produced by senior citizens through a life- more before my MIT courses gave informa-
time of experience. The lack of interest could tion about the same problems.
also be due to urban growth in cities, which
eliminated suitable flying sites within a rea- In my schooling, a few recollections may be
sonable distance from club activities. mentioned that have _ bearing on my subse-
quent career. In grade school, I was very shy
Building and flying model airplanes, particu- and teachers would have had a hard time
larly indoor models, does involve many tech- detecting any unusual ability. I do have a rec-
nical considerations, almost to the same ollection that by the end of the third grade, I
extent as full-scale airplanes. One of the could remember just about everything that
great incentives for the young people engag- had happened in school up to that time,
ing in this hobby was that the design of mod- something that I think the average student
els was in a stage of rapid development and could not do. This ability started to disappear
the young people in their teens could con- after that time, howev _r. I was very shy and
tribute to this development with their own studious as a child. My marks in grade
efforts. The record flights of indoor models school were just average, but in the first year
increased from about 7 minutes around 1928 of junior high I started to get all A's and this
to 21 minutes in 1938 and has since performance continucd with a few excep-
increased to 30 minutes in 1945 and to tions through high school and college.
52 minutes in 1979. The latter record stood Though neither of my parents knew anything
for 15 years, but was increased to 55 minutes about model airplant_s, they were always
in 1994. The long-awaited goal of a 1-hour very supportive of my interest in this hobby.

10 Monographs in Aerospace Hi_tory Number 12--Journey in Aeronautical Research

College Years

I had a room to use as a workshop and all the

supplies required to fulfill my relatively
small demands. I started at MIT in 1935 while still living at
home to save money, but as a result,
The Belmont schools had some excellent remained shielded from the social life of the

teachers. In the class in ancient history college. In general, the MIT courses were
in junior high school, I had a project to build excellent with the exception of the mathe-
a Ballista, an ancient war machine used matics courses. Perhaps this opinion resulted
from my lack of natural ability in abstract
for bombarding walls. My model was
reasoning. I was always able to visualize
beautifully built and used some parts manu-
solutions, a useful ability for engineering
factured on the lathe. It could be cranked up
problems, but generally contrary to the
and would fire small stones or blocks of
requirements of rigorous mathematics. I had
had very little calculus in high school and the
problems in the physics course at MIT
In high school, the first scientific course was always required the use of calculus tech-
in physics, which I followed easily and once niques two weeks before they were taught in
got in trouble with the teacher for trying to the mathematics courses. I hope that this
correct a mistake that he made. I really scheduling problem has since been corrected
appreciated Euclidean geometry in the senior at MIT. Later in advanced calculus, the need
year in high school, which opened up for the to define a small quantity e "no matter how
small" was not clear to me, nor was it ever
first time the methods of scientific logic. One
explained by the professors. In the problems
particularly difficult problem was given out
encountered, I could visualize what hap-
by the teacher with no expectation that any-
pened as a quantity approached zero.
one could solve it. I managed to give a proof
of the proposition involving 45 steps. I have I was able to do the mathematics problems,
always kept this proof and it is reproduced in but the lack of basic understanding has
appendix IlL This problem was given in the always prevented me from making much use
Mathematical Puzzles section of the MIT of the methods of higher mathematics, for
magazine Technology Review many years example, vector analysis and linear systems
theory, that find many applications in aero-
later. It can be solved by a much simpler and
nautical work. I can recall one problem pre-
more elegant proof, but my brute force
sented in a physics class that I solved using a
approach is equally correct.
method of solution that I had not been taught
previously. The problem had to do with the
I graduated from the Belmont High School
distribution of velocity of a fluid between
in 1935. I was salutatorian and prepared an two parallel plates. The method I used was
address that the teacher thought lacked inter- later taught in the mathematics course in
est or inspiration. She encouraged me to graduate school as the method of undeter-
write about the subject closest to my heart, mined coefficients. The paper was corrected
aviation. I wrote an essay on this subject in by a graduate student who expected the solu-
the style of the Sir Roger de Coverley tion to be obtained by a different method and
Papers, writings that I had admired in who did not recognize that I had "invented" a
English class, and managed to overcome my known mathematical technique.
stage fright enough to present my carefully At MIT, most students took the same courses
rehearsed and memorized talk. The main the first two years and did not start to work
speaker at the graduation, a Belmont lawyer, on their specialty until the junior year. In this
gave a talk very similar to the one that I had year (1937-1938), I took a general theoreti-
originally proposed. cal course called Aeronautical Dynamics

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 11

College Years

under Professor Manfred Rauscher. Unlike cial boundary-layer tunnel in the basement
the later courses which were mostly of a of the aeronautics building. The pressure gra-
practical nature, Professor Rauscher's course dient down the tunnel could be varied as
was devoted entirely to the theories of desired. The data were analyzed to determine
dynamics of rigid bodies and of hydrody- the variation of friction drag coefficient in
namics. Professor Rauscher was a natural the tunnel, especiall} during the transition
born teacher. He taught the course in a very from laminar to turbulent flow. This calcula-
thorough manner so that no steps in deriva- tion was performed using the Gruschwitz
tions were omitted. As a result, the material method, which required graphical integration
and the reasoning behind it were well of the boundary-layer profiles. Professor
established in the students' memories. To Peters was also busy designing the Wright
accommodate the state of the mathematical Brothers Wind Tunnel at MIT and was rarely
knowledge of the undergraduates, however, available for consultation. In addition, he
the entire course was taught without the use was a native of Germany and when WW II
of vector analysis. A follow-up course to broke out, he left to cast his lot with the
show how the same results could have been Germans. Nevertheless, the thesis was com-
obtained with vector analysis techniques pleted and copied on microfiche (ref. 2.4).
would have assisted the students in under- In Germany, Professor Peters designed the
standing material encountered later in more large pelton-wheel-powered wind tunnels in
advanced textbooks, but I never took such a Modane, France that were taken over by the
course or at least not one so clearly pre- French after the war and are still in use.
sented. In the senior year, courses I took
included aircraft structures under Professor In addition to the required work, I devoted
Joseph Newell, aerodynamics under Profes- considerable effort to running tests in the
sor Shatswell Ober, stability and control Junior Aviation League wind tunnel. In 1939,
under Professor Otto Koppen, and automo- I took summer employment at the Pratt and
tive engineering under Professors E. S. and Whitney Aircraft Company in East Hartford,
C. W. Taylor. These professors are men- Connecticut. I was given a job as a draftsman
tioned because they taught a whole genera- in the Installation Department. The main
tion of aeronautical engineers who graduated project in this group was the installation of
and entered the aeronautical industry at the the new R-2800 engine in a Vultee YA-19
time of the tremendous development of avia- attack bomber for its first flight tests. This
tion that occurred during and after World work involved some designing as well as
War II (WW II). These students were very drafting and was finished in time for me to
influential in the development of American get a flight in the airplane. This engine was
aviation during this period. I also took a later used to power many of the military air-
graduate course, Introduction to Theoretical planes in WW II, including the Republic
Physics, which gave me a background in P-47 Thunderbolt fighter and the Consoli-
dynamics problems that required more dated B-24 Liberator _omber. A photograph
advanced techniques than those taught in the of the Vultee attack bc tuber, produced by the
undergraduate years. The objective of the Vultee Aircraft Dividon of the Aviation
Aeronautical course at MIT was to give stu- Manufacturing Corporation with the R-2800
dents a sufficiently broad practical back- engine installed is shown in figure 2.6.
ground so that they could design a complete
A comparison with a photograph of the orig-
airplane or any subassembly thereof. In addi-
inal airplane with a Pratt and Whitney Twin
tion, a bachelor's thesis was required.
Wasp engine develofing 900 horsepower
My thesis was on the subject of boundary would show little a_parent difference. It
layers, under Professor Heinreich Peters. The is remarkable that the R-2800 engine, capa-
objective was to make measurements of the ble of developing 2t J00 horsepower (later
development of the boundary layer in a spe- 2800 horsepower with full supercharging),

12 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

College Years

FIGURE 2.6. Vultee YA-19

attack bomber with Pratt
and Whitney R-2800
engine installed.

would fit in the same cowling as the smaller Unfortunately, by this time, the interest in
engine. In the modified configuration, the the results had disappeared because the anal-
cowling and engine were moved backward ysis applied to radial piston engines, which
about a foot to aid in balancing the airplane have been replaced by gas turbine engines on
with the heavier engine. The crew of most high-speed airplanes.
mechanics in the Installation Department
I was involved in several other projects at
stand in front of the airplane in figure 2.7
Pratt and Whitney, including design of an
along with me in shirt and necktie in the
installation for a proposed in-line vertically
right rear.
opposed engine, design of an installation of
When I first went to work, I had a talk with an R-2800 in a British Vickers Wellington
John M. Tyler, the vibration expert at Pratt bomber, and preparation of an exhibit of a
and Whitney. He presented me with a prob- futuristic engine installation for the New
lem concerning design of the engine mount- York 1939 Worlds Fair. Most all of these
ing pads to decouple vertical and pitching projects were behind schedule and had to be
oscillations of the engine. I worked on this finished during the summer.
problem in the evenings during the summer,
After my work at Pratt and Whitney, I con-
inasmuch as there was not much other activ-
cluded that I was not suited for the work at
ity to occupy me. By the end of the summer,
an industrial concern, with its rushing to
I had performed quite a lot of analysis, but
meet deadlines and its lack of time to study
the final answer appeared to be incorrect.
problems in depth. At MIT, I had become
After discussing these results with Mr. Tyler,
acquainted with the research work at the
I put this work aside, but I was always wor-
NACA and decided to try to work there.
ried about getting the incorrect result. Many
Before leaving MIT, however, I stayed an
years later, after I retired, I got out the prob-
additional year to obtain a master's degree.
lem again and this time obtained the correct
answer. The results were published in a The graduate year included aeronautical
NASA Technical Memorandum (ref. 2.5). courses of a more theoretical nature. In addi-

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 13

College Years

FIGURE 2.7. Mechanics in

the Installation
Department of Pratt and
Whitney in front of
YA-19 airplane. I am at
right rear.

tion, there was a course in instrumentation An option in the graduate year was to per-
by Professor Charles Draper. Dr. Draper was form an independent research project in
already noted for his work on aircraft and addition to the thesis. I was still interested in
engine instrumentation. Later, he became boundary layers and made studies of the
famous for the invention of the inertial navi- effect of air velocity on an electric arc with
gation system and for his work on gyro- the object of using these effects to measure
scopic gun directors. This course perhaps air flow in the boundary layer. I was again
more than any other helped prepare me for entirely on my own and obtained a large 30,
research work because the subjects he taught 000 volt transformer from the electrical engi-
were those that he was working on as neering storeroom to produce the arc. The
research problems. He also brought in some voltage drop across the AC arc was measured
aspects of electronics and physics as well as with an electrostatic ",oltmeter, which inher-
aerodynamics. ently rectifies the At; voltage. This was a
bulky and delicate instrument, which was
The master's degree also required a thesis. I also borrowed from _he electrical engineer-
wrote a thesis Exhaust Gas and Radiator ing storeroom. I was afraid of damaging the
Propulsion under Professor Rauscher, who expensive instrument. I therefore built one of
was also unavailable for much consultation. my own that worked on the same principle.
This thesis could have provided an inspira- My electrostatic votmeter had attracting
tion for the then unknown jet engine, but I plates and chambers made from tin cans and
was discouraged from considering such a suspension construe:ted from the tungsten
developments by statements made in earlier wire used for bracing indoor models in place
courses that metals could not withstand the of the quartz fiber in _:he professional instru-
high temperatures required. This thesis was ment. This constructed instrument worked as
later published, with small changes, in well as the professionally built one. The
the Journal of the Aeronautical Sciences results were put out in a paper that received
(ref. 2.6). a high grade from Professor Draper, but

14 Monographs in Aerospace H story Number 12--Journey in Aeronautical Research

College Years

FIGURE 2.8. Model glider

equipped with phugoid

w_-'G'T"___DA_''_°T___I I._"f'_I fl '_ _ /

8ALAN C,E -_. ' ......


\g -wv,'_,

(#% \l .'li j._MIN.3"EST£_ WITH CQNTI_L


"D_"3PAR -" _-,i_



L AUNCHItJG (DROP I. i/,_'I'_._% _



St,IGGr_rff.O CONY[S1" DESIGN _._


I V_;'_".%, _'->_ ......

y _lIIF_ _ ..... _ _/l_ ALUM rUfF

.. _e_" CONIROL DETAIL .1_--

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 15

College Years

apparently has since been lost. I was fortu- such tests were quite rare. Later, a drawing
nate, probably, to escape electrocution in this of my phugoid damper and an article
project because the transformer stores a large describing its operation was prepared by
amount of energy and could easily kill a per- Herbert K. Weiss. The article was submitted
son. I did receive one pretty sharp shock to the editor of a modeling publication, but
from it. I was quite naive in handling high so far as is known, was never published. A
voltage equipment and should not have been copy of Weiss' drawing of the phugoid
allowed to work alone with it. damper is given in figure 2.8. Incidentally, I
have kept in touch with and often obtained
I also became interested in automatic control
advice from Mr. Weiss throughout my career.
of airplanes. The theoretical methods for
He is a brilliant engineer who became noted
analyzing such systems were just being
for his work in automatic control, missile
developed and were presented in theses by
guidance, and operations research.
two other MIT graduate students, Herbert K.
Weiss and Shih-Nge Lin. I knew from my
My interest in the scientific aspects of model
childhood model glider experience that mod- aviation continued with a series of tests
els with short fuselages and small horizontal
exploring the drag of fuselages, which I
tails would have a poorly damped longitudi- made in an old MIT wind tunnel that dated
nal oscillation (the so-called phugoid oscilla-
from the early 1920's. The tunnel and its bal-
tion). I devised a method using a spring-
ance were copies of the early wind tunnel in
mounted weight and a viscous damper to
the National Physical Laboratories in Ted-
operate the stabilizer in a way that I thought
dington, England. The tunnel was long out-
would damp out this oscillation. I made an
dated for research on full-scale airplanes, but
analysis of the system that indicated favor-
it was quite suitable fi_r tests on outdoor, gas-
able results. The system was then installed in
oline-powered modei airplanes. An article
a model towline glider of about a 40-inch
publishing these resuits was presented in the
wing span with a very tiny stabilizer. When I
magazine Model Airplane News and recently
tried the model, it was difficult to get con-
in the Twenty-Seventh Annual Symposium
vincing results because of the difficulty of
of the National Free Flight Society (ref. 2.7).
towing a glider up and launching it in a wind
in a consistent attitude.
During the graduate year at MIT, I went to
I did, however, feel confident enough to give Boston to take the Civil Service Exam for
a demonstration to Professor Koppen on the Junior Aeronautical Engineer. Though I was
MIT athletic field. The model performed per- anxious to work for the NACA, they were
fectly. On the first launch, with the stabilizer not overly anxious t9 have me. The exam
locked, the model went into a continuous was quite tough. I later learned that the ques-
phugoid oscillation. On the next flight, with tions had been supplied by Eastman Jacobs,
the system operating, the oscillation damped perhaps the most noted aerodynamicist at
out immediately and the model made a Langley. I waited a long time to get the
smooth glide. These results showed the results, but just as I was starting to think
advantages of using model airplanes to study about looking into otaer job opportunities, I
stability problems. Today, such tests are received a notice to n :port for duty. I entered
called dynamic model tests and utilize mod- duty at the NACA at the Langley Memorial
em equipment such as radio control and tele- Laboratory on July 1, 1940 and was assigned
metering to obtain the data. In 1940, however, to the Flight Researct Division.

16 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

CHAPTER 3 Setting for the Research Work at

It is well known to those who worked at the try, and universities and research organiza-
NACA in its early years that little attempt tions. These officials served without
was made to publicize the activities of the compensation and met two or three times a
organization outside the circle of interested year. One of the specified objectives of the
parties in the military services and the aero- committee was to establish a center for aero-
nautical industry. As a result, the general nautical research.
public was largely unaware of the existence
of the Langley Memorial Aeronautical Labo- The Langley Memorial Aeronautical Labora-
ratory or of the work conducted there. Inas- tory was established in 1917 on Langley
much as all the work described herein was Field, already a base for the Army Air Corps
conducted in this location and setting, a brief that had been established in April 1917.
description of the origin and history of the Langley Field, now Langley Air Force Base,
center may be useful in understanding the is located in Hampton, Virginia. Hampton, in
development and progress of the organiza- turn, is on the Virginia Peninsula, which is a
tion. strip of land bounded by the James and York
Rivers that extends down to the entrance of
the Chesapeake Bay.

History By 1920, the first research facilities were in

place and aeronautical research was started
The NACA, or National Advisory Commit- with an initial complement of 4 professionals
tee for Aeronautics, was established by Con- and 11 technicians. Before describing the
gress in 1915 to encourage the development later developments, some information on the
of aviation in the United States. From the
features of this location are presented.
standpoints of both research and industrial
development, progress in aviation in this
country was far behind that in the warring
nations of Europe. The organization estab- Description of
lished by Congress to correct this situation Surroundings
was an independent government agency
reporting directly to the President. The main Many historical areas are located on the
governing body of the organization, often Peninsula, including Jamestown, the first
referred to as the Main Committee of the English settlement in the United States,
NACA, consisted of about 20 men represent- which dates from 1607. Hampton is known
ing the military services, the aircraft indus- as the first permanent English settlement

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 17

Description of Surroundings

in the United States. Other nearby towns life extremely happy and exciting, particu-
dating from colonial times are Williamsburg larly since this was my first experience living
and Yorktown. away from home and without the rigorous
routine of studying at MIT. The small size of
When I arrived at Hampton in 1940, it was a
the Laboratory meant that most of the
town of about 7000 people. The main indus-
employees knew a good number of the staff
tries of the local population were fishing and
personally and the small social groups and
crabbing. Many more people were located in
organizations that had formed made for a
the surrounding area of Elizabeth City
pleasant social life as well as providing for a
County. Hampton was later incorporated as a
good interchange of r,.'search ideas.
city taking in the surrounding area. The other
large city on the Peninsula was Newport To a large extent, the employees made their
News, adjacent to Hampton. It had a much own social activities. The only professional
larger population at that time. Its principal society, the Engineer_ Club, was founded in
industry was the Newport News Shipbuild- 1940 and put on excellent programs. There
ing and Drydock Company. was a very active model airplane club, an
item of importance to me, since this had been
Langley Field was selected as the site of the
research center for the NACA because of the my main hobby before coming to Langley.
The NACA hired about 100 expert model
availability of the airfield; the distance from
builders, starting about 1938. A special Civil
Washington was large enough to avoid politi-
Service Exam for which the only require-
cal interference, but close enough to allow
convenient travel; and the weather was suit- ment was that the applicant should have won
a model airplane contest was used to select
able for flight research. Records had shown
the model builders. Those selected were
that Langley Field had more clear days per
employed in model shops or instrument
year than any other base on the East Coast.
shops, and later many of them rose to high
The early 1940's included the occurrence of positions at Langley. After arriving here,
WW II, a period of tremendous expansion at these model builders _ept up their interest in
Langley and of the aircraft industry in gen- model airplanes by holding regular club
eral. I have prepared an article, "Recollec- meetings and large contests. Langley engi-
tions of Langley Memorial Aero Lab in the neers gave advice to the modelers and offi-
Forties;' that describes the activities during cials took an active part in sponsoring and
this period (ref. 3.1). A summary of some awarding prizes at contests.
aspects of the social life and the surround-
ings at Langley and in the town of Hampton, The NACA Tennis Club had been started in
Virginia, where it is located, may be helpful the 1920's and had six clay courts made by
in setting the stage for the type of work done rolling the local soil, which proved very sat-
during the subsequent years. isfactory for the purp_ ,se.

In 1940, the Langley Memorial Aeronautical A social organization called the "Noble
Laboratory was still quite a small center with Order of the Green Cow" put on dances for
fewer than 750 employees. Early employees the employees. These dances were often held
at the Laboratory have told what an isolated at the Hampton Armory and were very well
and provincial place Hampton was when the attended.
NACA first started operations in the early
1920's. The first few engineers met with con- During the 1930's, qu te a few engineers flew
siderable animosity from the local popula- their own light plaaes. This activity, of
tion and the lack of any social or cultural course, ended with th,: start of the war except
activities made life difficult, especially for for the Civil Air Patral. There were at least
the wives. By 1940, when I arrived, these three private airports in the nearby cities of
conditions had largely disappeared. I found Hampton and Newport News.

18 Monographs in Aerospace H;story Number 12--Journey in Aeronautical Research

First Assignments in Flight Research

Because of the location on a peninsula, water of the wind tunnels have a specific purpose
sports were popular. Many NACA engineers and range of airspeed. A person assigned to a
became members of the Hampton Yacht Club wind tunnel would naturally become a spe-
and engaged in the club activities. cialist in the particular research objectives of
that facility and, if so inclined, would con-
Golf was also available with clubs at
tribute to the theory involved in that phase of
Hampton and Yorktown. The Yorktown golf
research. Flight research, however, involves
course was quite interesting because it was
all types of aeronautical problems. The
located on a Revolutionary War battlefield.
research engineer must have a knowledge of
Later, this course was closed because the
such diverse fields as structures, aerodynam-
Secretary of the Interior, Harold Ickes, ruled
ics, loads, performance, and stability and
that this was not a suitable activity for a
control. As a result, I became a generalist,
national monument. I felt that I learned more
with less than complete knowledge in most
about Revolutionary War history by playing
fields, but better acquainted than most young
golf among the fortifications and redoubts,
engineers with the research work being per-
however, than by reading books on the
formed at all the facilities at the Laboratory.

This seemingly ideal situation at Langley The working conditions were also very desir-
was interrupted by WW II. With the entrance able, both in terms of personnel and facili-
ties. The head of the division when I started
of the country into the war following the
attack on Pearl Harbor in December 1941, work was John W. (Gus) Crowley, Jr. who
the work at the Laboratory took on a new had performed much early flight research
urgency. The work week was increased from and later became Chief of Aerodynamics at
44 to 48 hours, which required a full day of NASA Headquarters. Second in command
work on Saturdays. In addition, the imposi- was Floyd L. Thompson, who became Direc-
tion of gas rationing prevented much travel. tor of Research and later held the positions
Everyone worked very hard during the war. of Associate Director and Director of
The increased work load also required a tre- Langley. The division was divided into sec-
mendous expansion in facilities and person- tions entitled Aircraft Loads, Performance,
nel. Before describing this expansion, how- Helicopters, and Flight Research Maneuvers.
ever, I will review some of my first research The latter title was later changed to Stability
assignments at Langley. and Control. I was assigned to the Flight
Research Maneuvers Section under Robert
R. Gilruth, who had just started work three
years earlier, but was already recognized for
Firsi Assi gnments in his outstanding ability to direct research. He
Flight Research later became director of the Johnson Space
Center during the Mercury, Gemini, and
In my previous acquaintance with the work Apollo programs and the initial stages of the
at Langley, I had learned that most of the Space Shuttle program.
research was performed in wind tunnels.
Certainly my previous experience, particu- The facility available for flight research con-
larly with the Junior Aviation League wind sisted of the Flight Research Hangar, stocked
tunnel and the boundary-layer research of with about 15 to 20 airplanes of all types,
my bachelor's thesis, would have ideally pre- such as light airplanes, transports, and the
pared me for wind-tunnel work. Neverthe- latest military airplanes including both fight-
less, I gladly accepted the assignment to the ers and bombers. Mechanics were available
Flight Research Division. The decision to to maintain the airplanes and to install spe-
place me in this division certainly had a pro- cial research equipment. Recording instru-
found effect on my subsequent career. Most mentation had been developed from the first

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 19

First Assignments in Flight Research

FIGURE 3.1. Airplanes

being tested at the NACA
Flight Research Hangar
in 1944.

days of the NACA and was operated by tech- was taken after the construction of the sec-
nicians skilled in its use and installation. ond hangar.

These airplanes were flown by test pilots.

With this background, I will proceed to
The job of the Flight Research Engineer was
describe some of the early work that I did at
to plan the flight tests, analyze the instrument
Langley. I will not attempt to give a chrono-
data, and prepare reports on the results. I par-
logical account of my activities, which
ticularly enjoyed working in the offices that
would become confusing because of the
were located on the airfield side of the han-
overlapping of many jobs. Instead, I will sin-
gar, which allowed a clear view of the take-
gle out some of the more important catego-
offs and landings of all the military and
ries of research and collect the activities that
research airplanes. Later, however, an addi-
fall under these headings. To make the dis-
tional hangar was built and the offices were
cussions more understandable to persons not
moved to the other side.
trained in aeronautical engineering, I start
A photograph of the airplanes that were with a brief discussion of the history of the
tested in the Flight Research Division in discipline and definitions of the technical
1944 is shown in figure 3.1. This photograph terms used.

20 Monographs in Aerospace H:story Number 12--Journey in Aeronautical Research

CHAPTER 4 Flying Qualities

By "flying qualities" are meant the stability 1904 (ref 4.1). This theory is essentially
and control characteristics of an airplane equivalent to the theory taught to aeronauti-
that have an important bearing on the safety cal students today and was a remarkable
of flight and on the ease of controlling an intellectual achievement considering that at
airplane in steady fight and in maneuvers. the time Bryan developed the theory, he had
The term "handling qualities," which is not even heard of the Wright brothers' first
applicable to other types of vehicles is used flight. Because of the complication of the the-
synonymously with flying qualities when ory and the tedious computations required in
applied to airplanes. its use, it was rarely applied by airplane
designers. Obviously, to fly successfully,
To start the discussion of flying qualities, the
pilotless airplanes had to be dynamically
concept of stability should be understood.
stable. The airplane flown by the Wright
Stability can be defined only when the vehi-
brothers, and most airplanes flown thereaf-
cle is in trim; that is, there are no unbal-
ter, were not stable, but by trial and error,
anced forces or moments acting on the vehi-
designers developed some airplanes that had
cle to cause it to deviate from steady flight. If
satisfactory flying qualities. Many other air-
this condition exists, and if the vehicle is dis-
planes, however, had poor flying qualities,
turbed, stability refers to the tendency of the
which sometimes resulted in crashes.
vehicle to return to the trimmed condition. If
the vehicle initially tends to return to a
Bryan showed that the stability characteris-
trimmed condition, it is said to be statically
tics of airplanes could be separated into
stable. If it continues to approach the
longitudinal and lateral groups with the
trimmed condition without overshooting, the
corresponding motions called modes of
motion is called a subsidence. If the motion
causes the vehicle to overshoot the trimmed motion. These modes of motion were either
aperiodic, which means that the airplane
condition, it may oscillate back and forth.
steadily approaches or diverges from a
If this oscillation damps out, the motion
is called a damped oscillation and the vehi- trimmed condition, or oscillatory, which
means that the airplane oscillates about the
cle is said to be dynamically stable. On the
trim condition. The longitudinal modes of a
other hand, if the motion increases in ampli-
statically stable airplane following a distur-
tude, the vehicle is said to be dynamically
unstable. bance were shown to consist of a long-period
oscillation called the phugoid oscillation,
The theory of stability of airplanes was usually with a period in seconds about one-
worked out by G. H. Bryan in England in quarter of the airspeed in miles per hour and

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 21

Flying Qualities Tests

a short-period oscillation with a period of install instruments to record relevant quanti-

only a few seconds. The lateral motion had ties such as control positions and forces, air-
three modes of motion: an aperiodic mode plane angular velocities, linear accelera-
called the spiral mode that could be a diver- tions, airspeed, and altitude. Then a program
gence or subsidence, a heavily damped ape- of specified flight cor.ditions and maneuvers
riodic mode called the roll subsidence, and a was flown by an experienced test pilot. After
short-period oscillation, usually poorly the flight, data wer_ transcribed from the
damped, called the Dutch roll mode. records and the results were correlated with

pilot opinion. This approach would be con-

Some early airplane designers attempted to sidered routine today, but it was a notable
make airplanes that were dynamically stable, original contribution by Gilruth that took
but it was found that the requirements for advantage of the flight recording instruments
stability conflicted with those for satisfactory
already available at Langley and the variety
flying qualities. Meanwhile, no information of airplanes available for tests under compa-
was available to guide the designer as to just rable conditions.
what characteristics should be incorporated
to provide satisfactory flying qualities. An important quantity in handling qualities
measurements in turns or pull-ups is the
By the 1930's, there was a general feeling variation of control force on the control stick
that airplanes should be dynamically stable, or wheel with the value of acceleration nor-
but some aeronautical engineers were start- mal to the flight direction expressed in g
ing to recognize the conflict between the units. This quantity is usually called the force
requirements for stability and flying quali- per g. This notation will be used in this
ties. To resolve this question, Edward P. monograph.
Warner, who was working as a consultant to
the Douglas Aircraft Company on the design
of the DC-4, a large four-engine transport
Flying Qua[itieL_ Tests ......
airplane, made the first effort in the United
States to write a set of requirements for satis- By the time I started work, Mr. Gilruth had
factory flying qualities. Dr. Warner, a mem- tested about 16 airplanes. These airplanes
ber of the main committee of the NACA, also ranged from light airplanes to the largest air-
requested that a flight study be made to plane then available, the 150-foot span
determine the flying qualities of an airplane Boeing XB-15 bomb_ r. Based on these tests,
along the lines of the suggested require- Gilruth prepared a report, Requirements for
ments. This study was conducted by Hartley Satisfactory Flying Qualities of Airplanes,
A. Soul_ of Langley. Entitled Preliminary that was published as a NACA Technical
Investigation of the Flying Qualities of Air- Report in 1943, but was available in prelimi-
planes, Soul_'s report showed several areas nary form when I arrived at Langley
in which the suggested requirements needed (ref. 4.3). This report served as the basis for
revision and showed the need for more much of the work of the Stability and Con-
research on other types of airplanes trol Section during tht: ensuing years. After a
(ref 4.2). As a result, a program was started period of indoctrination, I was the engineer
by Robert R. Gilruth with Melvin N. Gough in charge of the flying qualifies measure-
as the chief test pilot. It was during the ments. Pictures of me with two of the air-
course of this program that 1 entered planes tested are shown in figure 4.1.
the Flight Research Division and my early
My name first appeared on reports on the fly-
work consisted largely of flying qualities
ing qualities measun.ments of the Vought-
Sikorsky XF4U-1 C( rsair airplane, a Navy
The technique for the study of flying qualities Fighter well-known for its service in WW II
requirements used by Gilruth was first to (figure 4.2).

22 Monographs in Aerospace History Number 12 Journey in Aeronautical Research

Flying Qualities Tests

FIGURE4.1. Pictures of
me with two of the
airplanes tested in my
first years at the NACA.

(a) Curtiss
P-40 fighter
(b) Brewster XSBA-I
scout bomber

This study had been started before I arrived office windows as the Air Force squadrons
and my main contribution was to edit the came in for landings.
text. The first published NACA report for
which I supervised the tests (May 1941) After the start of WW II, the famous British
was on the flying qualities measurements of fighter airplanes, the Hawker Hurricane, (fig-
a Curtiss P-40 airplane (ref. 4.4). This air- ure 4.4) and the Supermarine Spitfire (fig-
plane was the primary U.S. fighter at the start ure 4.5) were obtained for tests at Langley.
of WW II (figure 4.3). Need for some
improvement on the handling qualities of Brief summaries of some of the measured
this airplane was shown by several ground- flying qualities of interest are included in the
looping accidents directly in front of my figure captions of these and other airplanes.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 23

Flying Qualities Tests

FIGURE 4.2. Vought-

Sikorsky XF4U-1
airplane. The flying
qualities of this airplane
arc of interest as one that
was in service before the
publication of Gilruth's
flying qualities
requirements. The
elevator control forces in
turns were found to be
desirably light, but the
aileron forces in high-
speed roils were heavy,
resulting in sluggish
response. The rudder
force variation with
speed for trim in high-
speed dives and strafing
runs was large, resulting
in difficulty in holding
the sights on an aim
point. Control forces in
the carrier approach
condition had an
unstable variation with
speed, a common
condition that does not
have a very adverse
effect on the flying

FIGURE 4.3. Curtiss P-40

airplane. This airplane at
the start of the war was
lightly armed and
underpowercd compared
to British and German
fighters. Its flying
qualities were
satisfactory except for
the usual heavy aileron
control forces. Stalling
characteristics were poor
in some conditions. The
airplane had a
particularly bad
tendency to ground loop,
which was found to be
caused by asymmetrical
stalling of the wing in the
thrce-point attitude. This
problem was cured by
extending the taft-wheel
strut so that the airplane
remained unstalled on
the ground.

24 Monographs in Aerospace H story Number 12--Journey in Aeronautical Research

Flying Qualities Tests

FIGURE 4.4. Hawker

Hurricane airplane. A
heavily armed fighter
airplane noted for its role
in the Battle of Britain,
the Hurricane's flying
qualities were found to
be generally satisfactory.
The most notable
deficiencies were heavy
aileron forces at high
speeds and large friction
in the controls.

FIGURE 4.5. Supermarine

Spitfire airplane. A high-
performance fighter
noted for its role in the
Battle of Britain and
throughout WW II, the
Spitfire had desirably
light elevator control
forces in maneuvers and
near neutral longitudinal
stability. Its greatest
deficiency from the
combat standpoint was
heavy aileron forces and
sluggish roll response at
high speeds.

I published reports on the Hawker Hurricane continual detailed improvements were made
(April 1942) (ref. 4.5) followed shortly by in these fighters. Several research studies
one on the Spitfire. The data obtained in were made on improvements, usually on
these tests served to confirm most of the control systems, and close contact was kept
requirements previously proposed by with the manufacturers through conferences
Gilruth. Other reports followed comparing and preliminary reports.
these results with published data on the
German fighter Mel09 and with U.S. fighter The tests on the high-speed fighters con-
airplanes. During the war, pilots' lives firmed the findings of Gilruth that though all
depended on small differences in perfor- the airplanes exhibited instability in the spi-
mance between the first-line fighters, and ral and phugoid modes of motion, these

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 25

Stall Tests

modes did not concern the pilot because his such as the center-of-gravity location, tail
normal control actions prevented the modes size, and control surface design, and the
from developing to a point that they were resulting flying qualities.
noticeable. That is, the airplanes were spi-
rally unstable, but the rate of divergence was
small enough that it was not discemible to Stall Tests
the pilots. Also, the long-period longitudinal
mode might have been a slow divergence or Stalling characteristics are an important con-
a poorly damped or unstable phugoid oscilla- sideration in studying flying qualities
tion, but the divergence was so slow or the because many accidents have resulted from
oscillation had such a long period that it was poor smiling characteristics. The design fea-
not noticeable in normal flight. The short- tures leading to poor stalling characteristics
period lateral oscillatory mode, the Dutch were at that time, and to some extent still are,
roll, was noticeable but adequately damped poorly understood because stalling is a com-
and the short-period longitudinal mode was plex problem involving separated flow. Sepa-
so well damped that it could not be detected rated flow occurs when the thin layer of
by the pilots. In general, these results applied slowly moving air near the surface of the
to most airplanes of this period and explain wing, called the boundary layer, thickens and
why successful airplanes could be built causes large changes in the external flow.
without the need to consider theoretical pre-
My first assignment at Langley was to ana-
dictions of dynamic stability. On the other
lyze data and plot figures from stall tests of
hand, Gilruth had found that many of the
the North American BT-9B airplane, a trainer
quantities that could be determined without
that had displayed dangerous stalling charac-
the need for complex theories, such as con-
teristics (figure 4.6).
trol deflections and control forces required in
straight flight and maneuvers, trim changes The flight studies, _hich had been started
due to power and flap setting, limits of before I arrived at Langley, were quite
rolling moment due to sideslip, and ade- detailed and included tuft studies, flow field
quacy of the control effectiveness in maneu- surveys in the vicinity of the tail, and tests of
vers, were extremely important to the pilot. slats covering various portions of the span.
The tests on the fighter airplanes showed that By way of explanation, tuft studies refer to
the longitudinal control force gradient in tests made with numerous short pieces of
maneuvers, known as the force per g, was a yarn attached to the wing or tail to show the
very important quantity, whereas the control direction of flow at the surface. If the flow at
force and position variation with speed in the surface is reversed, a stalled condition is
straight flight was of less importance and indicated. Flow field studies required rakes
mainly influenced pilot fatigue on long holding a series of flow-direction and veloc-
flights. These airplanes were found to be ity sensors to determine the flow characteris-
quite satisfactory in most respects, though tics ahead of the tail. Slats are small airfoils
the aileron effectiveness at high speeds was mounted ahead of the leading edge of a sur-
low because of the large control force face to prevent flow s._'paration at the leading
required to deflect the ailerons, which was an edge and increase tbe angle of attack that
adverse characteristic in air combat. The results in a stall. Tt_e project engineer on
detailed improvements mentioned previously these tests was Maurice D. White, an engi-
were mainly directed at this aileron effective- neer who later movec: to the Ames Research
ness problem. Center and who was responsible for much
notable work in the :.tahility and control of
In addition to analyzing the flight test data, I airplanes. Perhaps )ecause of the large
did theoretical analyses showing the relation amount of data, it was impossible to draw
between the airplane design characteristics, any simple conclusions. In most NACA

26 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Brewster XSBA-1 Airplane

FIGURE 4.6. North

American BT-9B
airplane. This airplane
had displayed violent
roll-off at the stall with
no warning, a
unsatisfactory condition
in a trainer.

reports of that time, the desire was to present that led to this project, the objective was
a straightforward set of conclusions. As a probably to supply designers with systematic
result, this report was never published, but I design data for obtaining desirable flying
have always kept a copy of the rough draft. qualities on new Navy airplanes.
Through the years, when questions arose
concerning the stalling characteristics of air- The airplane was quite up-to-date at the time
planes, it was usually possible to find some it was designed in the 1930's, but rapidly
applicable information in the BT-9 data. became obsolete with the advance of military
aircraft design. As a result, the studies con-
ducted were of little interest for this particu-
lar airplane, but were considered to have
Brewster XSBA-1
general research interest. I was given the job
Airplane of performing research on the effects on fly-
ing qualities of the different configurations
A project using the Brewster XSBA-1 air-
(figure 4.1). The project obviously generated
plane was started some years before I came much less enthusiasm than tests of the latest
to work at Langley. In this designation,
fighter airplanes, but I dutifully ran tests on
X stands for experimental, SB for Scout
the original configuration and on the first tail
Bomber, and A for the Brewster Aircraft
modification and wrote reports on the results.
Company in Long Island, New York, which Later two other combinations were tested,
failed due to labor troubles and was taken one with a different vertical tail and one with
over by the Naval Aircraft Factory. This increased dihedral. To some extent, these
project, a joint effort by the Navy and the projects were used for education of some of
NACA, involved testing a series of horizon- the new engineers. The last report was not
tal and vertical tail surfaces of different sizes
published until 1946.
and different ratios of control surface chord
to main surface chord. Also, shims were pro- The entire project on the XSBA-I is an
vided to attach the wing panels to the fuse- excellent example of a study that is unsuited
lage at different values of dihedral angle. for flight research. The design and construc-
Though I am not familiar with the thinking tion of flight-worthy components is a lengthy

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 27

Short-Period Longitudinal Oscillations

process and the installation of these parts on plane on which the elevators were not mass
an airplane in the field is also very time- balanced (figure 4.7). This type of oscillation
consuming. The number of possible combi- could become quite violent. As a result, engi-
nations to be tested is excessive, and if a neers at Langley were working on methods
thorough study of each were made, the to predict these problems.
project would last a lifetime. The items cho-
sen for test were not the most critical for A study in progress when I arrived at
obtaining good handling qualities and the Langley was measurement of longitudinal
effect of the configuration changes could oscillations with free elevator control for a
have been predicted with sufficient accuracy Fairchild XR2K-1 a_rplane (Navy designa-
from available knowledge. Some effects, tion for the Fairchild 22, a parasol-wing light
such as the effect of dihedral, were known plane). Robert T. Jones and his coworkers in
even to the Wright brothers. Had complete the Stability Research Division had pub-
tests been conducted, the data would not nec- lished a series of reports on the effect of free
essarily be applicable to airplanes of differ- controls on the stability of an airplane,
ent design. thereby extending the existing theory about
the stability with controls fixed. To check
The method chosen by Gilruth of making this theory, William Gracey in the Flight
detailed studies on many available airplanes Research Division had modified the longitu-
of different designs proved to be much more dinal control system of the XR2K-1 to
effective in yielding new knowledge that is include a link with movable weights so that
applicable to future designs. This method has the inertia of the longitudinal control system
been used by all the NACA Research could be varied. This system was closely
Centers, the military services, and industrial related to the device that I had tried on the
organizations ever since his early work. Fly- towline glider at MIT. I discussed the analy-
ing qualities studies continued at Langley sis that I had made at MIT. As a result, I
until the start to the space program in 1958. became acquainted _ ith Robert T. Jones and
In chapter 16 of this volume, an account is was given some responsibility in running the
given of the last test made at Langley on tests. One highlight of the tests was to mea-
a high-speed fighter airplane, the Vought sure the damping of the elevator oscillations
F8U-1. In 1988, I wrote a summary report,
in the Langley Full-Scale Tunnel where the
Flying Qualities from Early Airplanes to the
airplane had been mounted for other studies.
Space Shuttle, that was presented as the Motion of the elevator was recorded when
Dryden Lecture of the American Institute of the control stick was deflected and released.
Aeronautics and Astronautics (ref. 4.6). It was a bitterly cold tay when the tests were
made with the tempe:'ature in the tunnel near
freezing. I climbed u_ a ladder into the open
shOrt-Period cockpit bundled up in an overcoat, which
provided little protection when the airspeed
Longitudinal was increased to 100 miles per hour. I served
Oscillations as the "pilot" to deflect the control stick
abruptly and record the motions of the eleva-
The short-period longitudinal oscillations of tor. In my entire e_perience at Langley, I
all airplanes of the period prior to WW II never had the opportunity to run another
were well damped. The only cases encoun- wind-tunnel test until after I had retired. The
tered in which this motion was considered Fairchild tests correLtted poorly with theory
unsatisfactory involved coupling between the because of excessive friction in the control
elevator motion and the airplane motion with system, but the resalts were nevertheless
the control free. This problem had been published as an Advance Restricted Report
observed on the Lockheed 14A transport air- (ref. 4.7).

28 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Spiral Instability

FIGURE 4.7. Lockheed

14H transportairplane,
a latermodel of one of
the original 16 airplanes
on which Gilruth's flying
qualities requirements
were based. This
airplane exhibited a
coupling between the
longitudinal oscillation of
the airplane and the
motion of the mass
unbalanced elevator,
which proved very
objectionable in rough
air. This experience
resulted in a requirement
to rule out such
oscillations, which may
result from a variety of

no solution had been obtained that did not

Spiral Instability
have other serious unacceptable conse-
Almost all full-scale airplanes exhibit an quences. Gilruth suggested a possible solu-
instability in which the airplane, when the tion consisting of a downward-deflected
controls are held fixed in level flight, gradu- fixed tab on each aileron. These tabs would
ally veers off into a diving turn or spiral. This tend to make the ailerons float up. Because
instability, called spiral instability, had been the ailerons are interconnected through the
known since the earliest days of aviation, but control system, they would not float up,
was generally ignored because instinctive except possibly a small amount due to flexi-
control inputs by the pilot correct the condi- bility of the control system. In a turn, how-
tion so readily that he is rarely aware of its ever, the outboard tip would be traveling
existence. To apply these control inputs, faster and the aileron on that tip would have
however, the pilot must have a reference with greater tendency to float up than the one on
which to judge the attitude, either from the the inboard tip. The result would be to cause
horizon or from cockpit instruments. With- the aileron system to deflect in a direction to
out such a reference, the pilot in an airplane oppose the turn.
is completely unable to determine the atti-
tude, because the familiar vertical reference I undertook the job of analyzing this arrange-
ment. At first, the idea seemed promising
supplied by gravity is combined with forces
because for the example chosen, the aileron
in other directions caused by the accelera-
motion would be sufficient to make the air-
tions of the airplane. Student pilots or other
plane spirally stable up to a lift coefficient
pilots, who are not sufficiently proficient in
value of 1.5, which corresponds to a low-
instrument flying, frequently get in to such a
speed flight condition in which the spiral sta-
spiral when they fly into a cloud or fog bank.
bility is normally most severe. In an actual
The maneuver has been called the "grave-
airplane, however, the aileron movement is
yard spiral" because so many pilots have
opposed by friction in the control system. I
been killed by it.
then calculated how steeply the airplane
Despite many efforts over the years to make would have to turn to cause the ailerons to
this motion of the airplane inherently stable, start to move against the control system

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 29

Snaking Oscillations

friction. Using the measured aileron friction While some of these explanations may have
in the Curtiss P-40 fighter airplane, which had some influence in rare instances, the true
amounted to 1.5 foot-pounds per aileron at explanation was first given by George
any reasonable value of airspeed, the results Schairer of the Boeing Company in an analy-
showed that the ailerons would never sis of this problem on the Boeing 314 flying
develop enough hinge moment to overcome boat, one of the China Clippers. He pointed
the static friction. This result is typical for all out that at small angles of sideslip, the rudder
such schemes that attempt to eliminate spiral had a tendency to float against the relative
stability. The motions are so slow that the wind, which caused the airplane to swing
aerodynamic hinge moments involved are around and yaw in :_he opposite direction.
small, and any friction moments are suffi- Friction in the rudder system, however, held
cient to overshadow them. Even on an air- the rudder in this position as the airplane
plane that is inherently spirally stable, the swung through zero sideslip. On reaching a
friction force in the controls has usually been sideslip in the opposite direction, the rudder
found to be sufficient to hold the airplane in a hinge moments would eventually break
banked turn that, without pilot intervention, through the friction force and the cycle
would lead to a spiral dive. A more sophisti- would be repeated in the opposite direction.
cated system, consisting of an autopilot with Thus, energy was fed into the oscillation by
sufficient force output to readily overcome the rudder, which caused the oscillation to
control system friction, is required to make build up to an amplitude where this energy
an airplane spirally stable in practice. This equaled that removed by the inherent damp-
example is typical of studies that I made, ing of the airplane.
occupying less than a week of work, that
On learning of this explanation, efforts were
revealed information of value over the years
made to verify it. A convenient test airplane
in discussions with manufacturers or with
was the Fairchild 22 on which an experimen-
other workers at Langley who were not
tal all-moveable vertical tail had been
always familiar with the practical aspects of
installed. This type of tail surface was an
airplane operation.
invention of Robert T. Jones and had the
advantage that hinge moments due to angle
of attack and due to deflection could be
adjusted separately with changes in the hinge
Snaking oScillations
location and tab gearing. The tests were
made covering a range of conditions and
Another stability problem that was quite friction values, and the validity of the theory
common in airplanes of the period around was established (ref. 4.8).
WW II was a tendency for a continuous
small-amplitude lateral oscillation in straight The question arises a_ to how such an appar-
and level flight. This problem was called ently obvious contn,I motion could have
"snaking" and its cause was quite mysteri- escaped detection, q'he explanation is that
ous. Among the explanations offered were because of the relatively low damping of the
response of the normal lateral oscillation of Dutch roll oscillatic.n, the rudder motion
the airplane to continuous small-amplitude required to sustain a constant-amplitude
turbulence, periodic flow separation from the oscillation is only a small fraction of the
wing root that affected the vertical tail, or amplitude of the yaw or sideslip. For exam-
nonlinear aerodynamic characteristics of the ple, in a typical snaki lg oscillation of plus or
wing or tail surfaces for small changes in minus two degrees ,)f sideslip, the rudder
angle of attack. One explanation, which will motion required mig it have been only plus
be discussed subsequently, was the unsteady or minus two-tenths )f a degree. This small
lift characteristics of the vertical tail at low motion was less than the sensitivity of con-
frequencies. trol position recorders used at that time, and

30 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Rudder Deflection Required for Trim

this motion could be absorbed by stretch in that encountered on even the highest pow-
the control cables without being felt at the ered fighters because the models had power
pilot's rudder pedals. to climb vertically, whereas the fighters did
not. The model builders, less inhibited by
Another little-known aspect was the ten-
dency of the rudder to float against the rela- their training than full-scale airplane design-
tive wind at small sideslip angles. Most con- ers, had found that the propeller torque could
trol surfaces float with the relative wind at be handled by offsetting the wing laterally. I
larger sideslip angles. In typical wind-tunnel therefore did not hesitate to suggest this
tests, measurements had been made at incre- solution for the problem on the full-scale
ments of angles of attack or sideslip of five aircraft. Tests were run on the Brewster
degrees, and as a result, the small changes in XSBA-1, which provided another useful
characteristics at very small values of angle research application of this airplane.
of sideslip were not detected.

In addition to flight tests, theoretical studies Though the wing could not be offset later-
were made to explain and quantitatively pre- ally, the same effect could be obtained by
dict the oscillation. These studies are dis- shifting the center of gravity laterally by
cussed in a subsequent chapter. unsymmetrical loading of the wing fuel
tanks. For a conventional propeller turning
clockwise when viewed from the rear, the

Rudder Deflection center of gravity should be offset to the right.

The method was found to reduce greatly the
Required for Trim
rudder deflection for trim in low-speed, high-
For many years, aeronautical engineers power conditions (ref. 4.9). This method is
instinctively built airplanes with a symmetri- highly effective because three separate
cal configuration. The aerodynamic symme- effects combine to reduce the rudder deflec-
try is destroyed, however, by the engine tion. The reduced aileron deflection reduces
torque and the rotation of the slipstream. On the adverse yaw, the gravity component
many early airplanes, the fin was offset to along the longitudinal axis provides a yaw-
attempt to align it with the rotating slip-
ing moment, and the left sideslip required to
stream and to reduce the rudder forces to
maintain straight flight is reduced. (This
maintain trim in takeoff and climb condi-
sideslip is required to offset a combination of
tions. With the greatly increased power and
the side force on the fuselage due to slip-
speed range of fighter airplanes in WW II,
stream rotation and the side force on the ver-
the ability of the rudder to trim the airplane
tical tail due to rudder deflection.) All these
became a serious problem. If the fin was off-
effects are in the direction to reduce the
set to attempt to trim the airplane at low
speed, the rudder pedal force to maintain an rudder deflection. In high-speed dives, use of
aim point on a target in a high-speed dive an offset fin or any other structural deflection
frequently exceeded 100 or even 200 pounds. gives a moment that increases as the square
The pilots could exert these forces, but the of the airspeed, which results in rapidly
aiming accuracy was found to be very increasing rudder forces to maintain trim,
adversely affected by the large and varying whereas the offset center of gravity provides
a moment that does not vary with speed.
These effects had been known to some
Model airplane builders bad long ago
encountered the problem of very high pro- airplane designers as early as WW I, but
peller torque on rubber-powered models. The the effectiveness of this technique on
effect of this torque relative to the aerody- high-powered airplanes was not widely
namic forces on the wings was larger than recognized.

Monographs in Aerospace History Number 12 Journey in Aeronautical Research 31

Progress in Flying Qualities Research

a rational way to tae measured handling

Progress in Flying
qualities. Gilruth's research had allowed
Qualities Research many of the handling-qualities requirements
to be stated in quantilative terms, such as the
My work on flying qualifies research contin- control-stick motions or forces required for
ued during and after the period of WW II. As specified maneuvers and flight conditions. I
head of the Stability and Control Section, I developed theories to predict these quantities
had the responsibility for analyzing the data based on the dimensions and design features
and writing reports on many of these tests. I of the airplane. As an example, I calculated
did not do this alone, of course. By the time I the variation of control force and control
had worked at Langley for 3 or 4 years, the position with airspeed in straight flight or
measurement of flying qualities had become with normal acceleration in maneuvers.
a somewhat routine procedure, and the mem- Included in these calculations were the
bers of the section were assigned as project effects of such features as springs or mass
engineers on the various airplanes under test. balance weights often incorporated in con-
The results of these studies were still of great trot systems. A flight study was made in the
interest, because something new about han- XSBA-1 airplane to verify these calcula-
dling qualities was learned from nearly every tions. This analysis required nothing more
airplane that was tested. A great deal of my than algebra, and though some engineers
time, however, was spent in reviewing and were no doubt familiar with these effects, it
correcting the various reports that were pro- is remarkable that little emphasis was given
duced and in bringing their quality up to the to these relations in aeronautical engineering
high standards that had become traditionally courses. Before Gilruth's research, however,
associated with NACA reports. the incentive for making such analyses did
not exist.
One aspect of work at NACA that I consid-
ered very desirable was that there were Starting in 1943, I taught evening courses to
always many challenging technical problems newer Langley emplcyees on flying qualities
to work on, but if at times I lacked the inspi- and on stability and control of airplanes.
ration or ambition to tackle such problems, These courses were part of the University of
there was always plenty of routine work to Virginia Extension Program that allowed
occupy the time. Reviewing the reports put new employees to study for advanced
out by the engineers was one such task. degrees. The notes for the flying qualities
Many of the engineers hired during the course were later published as a NASA
period of rapid expansion during WW II Report entitled Appreciation and Prediction
lacked the extensive technical background or of Flying Qualities, a name suggested by
report-writing ability to meet the standards Hartley Soul6 (ref. 4.10). This report was
established in earlier years. I spent a lot of later used by many o_her organizations such
time with the engineers with less training to as the Navy Test Piiot School at Patuxent
improve their skills. At the time, this seemed River, Maryland, and by colleges, including
a thankless task, but in later years many of my alma mater, MIT. Gilruth's report,
these engineers went on to high positions in Requirements for Satisfactory Flying Quali-
industry or in the space program, and I was ties, had become the basis for flying qualities
credited by some of the engineers with help- requirements of the Air Force and Navy. The
ing them to gain some essential knowledge military services i acorporated in their
required for their careers. requirements many findings of the NACA
studies of flying qtalities, in addition to
Meanwhile, I worked on many specialized some more stringen provisions based on
technical problems. One such activity was maneuvers required in air combat and other
relating the design features of the airplane in military operations. These requirements

32 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Static Longitudinal and Directional Stability

FIGURE 4.8. Grumman

XTBF-3 airplane, a
torpedo bomber used in
the war in the Pacific.
This airplane had near
neutral directional
stability at small sideslip
angles (0 to 5 degrees)
caused by blanketing of
the vertical tail by the
disturbed flow behind
the large greenhouse
canopy. The use of a
ventral fin restored the
directional stability at
larger sideslip angles so
that the characteristics
were not considered

were revised from time to time and I, along of attack. This quantity was used because it
with other Langley researchers, was required is measured in flight, whereas the variation
to confer with officials of the Air Force of pitching moment with angle of attack,
Flight Dynamics Laboratory and the Navy which could be obtained in a wind tunnel,
Bureau of Aeronautics to discuss the revi- was not available in the flight tests. Remark-
sions to the requirements and to write com- ably good agreement was obtained between
ments on the changes. the predicted and flight measured values.
This agreement resulted from careful inclu-
sion of all sources of pitching moment, such
as the forces on the propeller, the influence
Static Longitudinal and
of wing upwash on the propeller and fuse-
Directional Stability lage, and the flow field at the tail. The flow
field at the tail was based on the detailed
One of the more important considerations in studies made in the NACA Langley Full-
designing a new airplane is selecting the size Scale Tunnel by Katzoff and Silverstein, an
and configuration of the horizontal and verti- excellent example of basic research with
cal tails. Giiruth's requirements for satisfac- applications to airplanes long after the
tory flying qualities had given a rational research was conducted. In later years, many
basis for the design of these surfaces, but the other reports were written on prediction of
application of these requirement involved the longitudinal stability, but none provided any
calculation of aerodynamic forces on all improvement in the accuracy obtained by
parts of the airplane and the interference Gilruth and White.
effects of the airplane on the tall surfaces.
The design of the vertical tail appeared to be
Before I started work at Langley, Gilruth an even more critical problem than the
and Maurice D. White had written a report design of the horizontal tail. Many airplanes
in which the longitudinal stability was tested in flight had proved to have deficient
predicted on the airplanes that had been directional stability and as a result failed to
used in the handling-qualities measurements meet the flying qualities requirements (fig-
(ref. 4.11). The quantity used to compare the ure 4.8). As a result, I attempted to make an
predicted and measured longitudinal stability analysis of directional stability similar to that
was the variation of elevator angle with angle made by Gilruth and White for longitudinal

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 33

Static Longitudinal and Directional Stability

FIGURE 4.9. Portion of

tail area that must be
considered ineffective
to obtain agreement
between calculated
and measured variation
of rudder angle with
sideslip angle (low-wing

0 10 20
I 1 I

Scale, fi

34 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Static Longitudinal and Directional Stability

stability. The quantity used for the compari- variation of rudder angle with sideslip on
son with flight data was the variation of rud- most of the 19 airplanes that were used in the
der angle with sideslip in steady sideslips, a comparison.
quantity analogous to the variation of eleva-
tor angle with angle of attack in the longitu- I have often felt that the report should have
dinal case. This analysis was written as a been published, despite its use of arbitrary
proposed report and was reviewed by mem- methods, because the report included many
bers of the Stability and Control Division. factors that had previously been neglected in
Their conclusion was that the report should designing vertical tails. The report would
not be published because the analysis was also have encouraged wind-tunnel research-
too arbitrary and was not based on sound ers to investigate systematically some of the
theory or data. configuration features that led to the arbi-
An example of the arbitrary nature of the trary rules, thereby allowing a more rational
analysis is given in figure 4.9, taken from the analysis. No similar report based on wind-
report, which shows the amount of vertical tunnel data was ever published. Fortunately,
tail area that must be considered blanketed the designers soon found the advantages of
by the flow over the fuselage and canopy to much cleaner canopies, such as bubble cano-
give agreement with the flight tests. After pies, in reducing drag so that future airplanes
examining this figure (and a similar one for did not have such poor flow conditions at the
high-wing airplanes), the lines defining the vertical tail. In recent years, methods based
part of the tail considered ineffective were on computational fluid dynamics allow accu-
drawn according to a set of rules that consid- rate calculation of the flow field over the
ered the height of the canopy and whether it entire airplane. These methods are known as
had sharp or rounded comers. Obviously, the panel methods, as the entire surface of the
airplanes with large "greenhouse" canopies airplane is represented by a large number of
lost a lot of directional stability because of fiat panels that approximate the true surface
the poor flow over the vertical tail. Unfortu- contours. These methods neglect flow sepa-
nately, wind-tunnel data for the effects of the ration, but result in more accurate predic-
fuselage on the flow at the vertical tail were tions of longitudinal and directional stability
not available. As a result, accurate analysis because airplanes have become cleaner as a
of this effect could not be made. Despite the result of refined aerodynamic design and the
arbitrary nature of the rules defining the use of jet engines. The need for an analysis
effective vertical tail area, the analysis suc- similar to the one I attempted has therefore
ceeded quite well in predicting the measured decreased.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 35

CHAPTER5 Analytical Studies

As in most scientific and engineering work, Liebnitz, a contemporary of Newton. A

analytical studies are required in conjunc- branch of calculus called differential equa-
tion with experiments to understand the tions occupied a whole term at MIT. Many of
results of aeronautical research and to pre- the methods that were taught to solve differ-
dict the characteristics of new aircraft or ential equations, however, applied to special
aeronautical systems. Much of the progress forms of equations that did not occur in con-
in improving the performance and safety of nection with airplane stability and control.
airplanes relied heavily on analytical work, This subject requires the solution of simulta-
which in turn depended on the availability neous linear differential equations with con-
and understanding of analytical techniques. stant coefficients. The standard mathematics
This section outlines the status of analytical curriculum did not give this subject any spe-
techniques that 1 had encountered in my cial emphasis. As a result, it was left for the
studies and that were generally available to professors in the aeronautical courses to
aeronautical researchers at the time I started emphasize the importance of this particular
employment with the NACA in 1940. equation. These professors, in my courses at
MIT, were practical engineers without a
Studies of the stability and control of air- strong mathematical background. As a
planes, my field of specialization, relied result, graduates were provided with only a
almost completely on the application of Sir minimal introduction to the branch of mathe-
Isaac Newton's laws of dynamics, which in matics most useful for their subsequent work.
turn led to the need to solve differential I feel that this situation existed in most of the
equations. In my college courses in mathe- colleges in the United States at that time,
matics and physics, Newton's laws of motion with the result that engineers who wanted to
were among the first subjects studied, and go into this field had to spend a great deal of
the various means to apply them in different their time in reviewing work done largely by
scientific disciplines occupied most of electrical engineers or by some aeronautical
the subsequent curriculum. Mathematics researchers in Europe. Many chose to put
courses, particularly differential and integral their emphasis on other areas with which
calculus, were presented to students as a they were more familiar. The result was
general preparation for all the courses. undoubtedly a slowing down of progress in
These methods were originally developed this field.
largely by Newton to solve his own problems.
The present notation for differential and inte- The reason that simultaneous differential
gral calculus was originated by Jacob equations with constant coefficients arose in

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 37

aeronautical stability work may be explained considered to vary almost linearly with the
in somewhat more detail as follows. For each change in airspeed. By comparison with
variable that describes motion (the degrees many mechanical sys:'ems, the airplane as a
of freedom), a differential equation is set up. whole does not have any coulomb friction
In the case of longitudinal motion of a rigid tending to hold it in its trimmed condition.
airplane, for example, the variables would Coulomb friction is the force caused by rub-
be vertical displacement, horizontal dis- bing two solid surfaces over each other, and
placement, and pitch angle. Newton's third from an analytical standpoint, is usually con-
law states that force equals mass times sidered to be a constant force that is inde-
acceleration. Thus for each variable, the pendent of velocity and opposing the motion.
aerodynamic or gravity forces due to the The force caused by aerodynamic effects is in
motion, which may produce forces or marked contrast to that appearing in
moments proportional to the displacement or mechanical systems. This difference results
velocity, are equated to the mass times the in the validity of linear equations to describe
acceleration of that variable. Acceleration is the motion of an airplane, whereas linear
represented by the second derivative of the equations are usuall) a poor representation
variable with respect to time, velocity the of mechanical systems.
first derivative, and displacement does not
Solution of the system of linear differential
involve a derivative. As a result, a differen-
equations with constant coefficients can be
tial equation of the second order is written
accomplished by the classical method given
for each degree of freedom, which results in
in most books on differential equations.
three simultaneous differential equations.
These solutions go back to the work of math-
These equations, in the usual formulation,
ematicians in the earliest days of mathemat-
are linear for the following reasons.
ics and were perhaps, first summarized in a
paper by Leonhard Euler in 1739. This solu-
The airplane is first considered as flying in a
tion is considered elementary from a mathe-
trimmed condition. After a disturbance, the
matical standpoint, inasmuch as it involves
forces on the airplane change as a result of
elementary functions such as exponentials
the effects of changes in angle of attack or
and sine and cosine functions. From the
airspeed on the various components of the
standpoint of the practical engineer using
airplane. These changes are necessarily
the equipment available when 1 came to
small in the normal unstalled range of flight.
work at Langley, wt ich consisted of slide
The stall angle, at least on airplanes of
rules or mechanical calculators, the solution
WW H vintage or earlier, was usually about
is very time-consuming. Determining the
15 degrees, and most disturbances producing
characteristics of the various modes of
forces within the structural capability of the
motion, such as the phugoid oscillation or
airplane would be much less than this value.
short-period mode, requires solving for the
Ideal fluid theory shows that within this
roots of a fourth-deg,ee algebraic equation.
range, forces vary nearly linearly with angle
If the effects of a simple autopilot are
of attack. The only factor that would change
included, a sixth-de;ree equation results.
this condition would be the effect of the
This problem can be : olved only by trial and
boundary layer, but the boundary layer on
error or by methods of successive approxi-
full-scale airplanes designed for efficient
mations. Further let,gthy calculations are
flight is so thin that it causes little effect on
required to determine the constants giving
the forces. Likewise any change in airspeed
the amplitudes of eath mode for known ini-
caused by a disturbance is likely to be small
tial conditions.
compared to the initial large value of air-
speed. Therefore, though the forces vary as Aeronautical engim ers confronted with
the square of the airspeed, the variation in these problems mad_ efforts to devise sim-
force for a small change in airspeed may be pler methods for solving for the roots of

38 Monographs in Aerospace H;story Number 12 Journey in Aeronautical Research

equations, aswill bedescribed
in thesubse- A third approach for solving these systems of
quentmaterial.Atthesame time,andusually equations was the use of simulators, gener-
withouttheknowledge ofaeronautical engi- ally referred to at the time of their develop-
neers,electricalengineers haddeveloped ment as differential analyzers. Vannevar
approaches thatallowedpractical Bush developed a mechanical differential
solutionsof evenmorecomplex systems of analyzer at MIT that I saw about 1934 while
Suchsolutions I was still in high school. This machine took
wererequired in the design of electrical the equations in an integrated form so that
circuits, vacuum-tube feedback amplifiers, the various terms required integration rather

and transmission lines. One method, first than differentiation. The integration was

summarized in a book by Hendrik W. Bode of done by devices known as rolling-wheel inte-

the Bell Telephone Laboratories, entitled grators. The low torque output of these

Network Analysis and Feedback Amplifier devices was amplified by winch-type electro-

Design (refi 5.1) exploited the use of sinusoi- mechanical servos to drive a large array of

dal forcing functions of various frequencies shafting and gearing that allowed setting in

to the dynamic systems under consideration, the correct constants from the equations. The

from which the stability of the systems could whole equipment required a room about

be determined. This method was slow to be 25 by 60 feet. This machine had excellent

discovered by the aeronautical engineering accuracy and was used by an MIT student to

profession because of the unfamiliar nota- get some of the first solutions for the motion
of airplanes with automatic controls. Later,
tion and applications of the electrical engi-
such simulators were made with servo-
neers. Later, the method was called the fre-
driven potentiometers to enter the constants
quency-response method and was widely
and electronic operational amplifiers to per-
used. A second approach with the general
form the integrations. These machines had
title operational methods was also intro-
tremendously increased capabilities, but they
duced by the electrical engineers. A British
in turn became obsolete with the develop-
electrical engineer and mathematician
ment of electronic digital computers.
named Oliver Heaviside devised a system of
operational calculus about 1887. The Heavi-
Although the airplane as a whole can be
side operational calculus was publicized
described accurately by linear differential
in this country in a book by Vannevar
equations, many of the subsystems, such as
Bush, entitled Operational Circuit Analysis
control surfaces or autopilots, involve non-
(ref 5.2) and was introduced to the aeronau-
linear components. For example, control sur-
tical profession by Robert T Jones in NACA
faces have static friction, and electronic
report No. 560, A Simplified Application
devices often have switches that give a dis-
of the Method of Operators to the Calcula-
continuous output. In general, nonlinear sys-
tion of Disturbed Motions of an Airplane
tems include devices in which the output var-
(ref 5.3). Later, the book by Murray F.
ies in a nonlinear but continuous manner,
Gardner and John L. Barnes, Transients in
such as a crank, or in which the output var-
Linear Systems Studied by the Laplace ies discontinuousl); such as a bearing with
Transformation (ref 5.4), described the
coulomb friction or an on-off switch. Much
operational method based on Laplace trans- of the mathematical analysis of nonlinear
forms, which became generally accepted as systems when I was in college had been con-
easier to understand than the Heaviside
cerned with the continuous type of nonlinear
method. The advantage of operational meth- systems, though solutions were usually avail-
ods is that solutions for frequently encoun- able only for systems capable of representa-
tered equations and for special inputs such tion by special types of differential equa-
as steps and ramps can be obtained from tions. The only method of analysis for
tabulated precalculated formulas. discontinuous systems with which I was

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 39

Graphical Solution of the Quartic Equation

familiar was the phase plane method, a need for closed-form solutions was reduced.
graphical method described in the book Most calculations on the digital machines
by Nicolai Minorsky, Introduction to Non- are made by numerical methods. One such
Linear Mechanics (ref 5.5). I used this method is the step-by-step method in which
method on problems involving simple _.pes the response of a sys:em to a disturbance or
of autopilots. With the eventual development the trajectory of a vehicle is calculated a
of analog computers, such problems could be small increment at a time. Another numerical

solved readily. method is the Monte Carlo technique, in

which many trial _olutions are made to
A final point that should be appreciated by find the one with the best answer for the
the reader is the state of computational facil- problem. Many problems of structures or
ities at the time much of my work at Langley fluid mechanics are :.olved by finite-element
was conducted. Prior to about 1955, the only
methods, in which the equations relating
widely used computers were the slide rule each small element of a large structure or
and mechanical calculators such as the
flow field are solved. These methods involve
Marchant and Frieden that required several tremendous amounts of numerical computa-
seconds to multiply or divide two numbers. tion, which modem electronic computers can
To perform lengthy calculations, female handle in a very short time. Since I had none
employees called computers were employed of these methods in my college education, my
to calculate results with these machines by
facility with such techniques is much less
following sheets that had the necessary steps than that of students who grew up in the
listed in tabular form. Such calculation
computer age.
sheets are today called spreadsheets.
With this background I will describe some of
in making analytical studies of problems, a the analytical studie: that I conducted dur-
very desirable result was a closed-form solu- ing my early employment at Langley.
tion, which means a formula in which num-
bers can be substituted for any particular
case to determine the numerical value of the
desired quantity. The determination of these
Graphical Solution of
closed-form solutions had been the objective the Quartic Equation
of mathematicians and scientists for many
As pointed out in tl_e introductory section,
years. In the aeronautical field, for example,
the solution for the m 9tion of an airplane fol-
Ludwig Prandtl determined the formulas for
calculating induced drag of wings, Max lowing a disturban.'e was very tedious
because of the rumerical calculations
Munk derived formulas for moments acting
on ellipsoids in steady flow, and Theodore required. One of the problems encountered

Theodorsen derived formulas for the lift and was solving for the roots of higher degree

moments on oscillating wings. The formulas equations. A fourth-degree equation resulted

in the solution for th_ longitudinal motion of
are considered to be in closed form if they
give results in terms of known tabulated a rigid airplane. Fo' many problems, this
equation could be :educed to a second-
functions, such as trigonometric functions or
degree, or quadratic,,.'quation by considering
Bessel functions. Most of these formulas
the airspeed constanl, a valid assumption if
and their originators became very famous
short-period maneuv'.rs were being consid-
because airplane designers could calculate
ered. The lateral equations likewise result in
results accurately for a configuration that at
a fourth-degree equation if certain simplify-
least approximated the one in which they
were interested. ing assumptions are made. These fourth-
degree equations, alsc_ known as quartics, are
With the advent of analog computers and called the characterTstic equations for the
later of high-speed digital computers, the systems, and solving for the roots of these

40 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Solution of a Nonlinear Problem

equations is the first step in calculating the equations in a fraction of a second. As late as
response of an airplane to controls. 1962, however, there was still interest in
simplified methods for determining the roots
Quadratic, cubic, and quartic equations may
of algebraic equations, as shown by a report
be written as follows:
by James W. Moore and Rufus Oldenburger
quadratic ax + bx + c = 0 of which I have an unpublished copy.
3 This report presents a systematized proce-
cubic ax + bx 2 + cx + d = 0 dure similar to Lin's method and analyses
4 2 problems of convergence for cases, such as
quartic ax + bx3 + cx + dx + e = 0
unstable roots, in which convergence of
High school students of mathematics are the method may be slow. Oldenburger was
familiar with the formula for the solution of a well-known expert on servomechanisms
a quadratic equation. Formulas also exist for before the subject of automatic control
the solution of cubic and quartic equations, became a favorite subject for control
though they are considerably more compli- theorists.
cated than those for a quadratic. For this rea-
son, graphical methods or methods of suc- The airplane with fixed controls in the
cessive approximations have been sought for unstalled flight regime, fortunately, is beauti-
the solution of these equations. In a paper on fully linear. That is, all the aerodynamic
propeller governors, Herbert K. Weiss pre- forces and moments increase in proportion to
sented a set of graphs for the solution of the magnitude of the displacement or angular
cubic equations in terms of two parameters velocity, even down to very small magni-
calculated from the coefficients of the equa- tudes of motion. As a result, the measured
tion (ref. 5.6). Shih-Nge Lin, in appendix I of motion of airplanes had been found to be
his MIT thesis, described a method that he closely predicted by the theory. Another
developed for a solution of quartic and other favorable feature of linear systems is that a
higher order equations (ref. 5.7). This proce- given solution is applicable to all magnitudes
dure is a method of successive approxima- of motion. An increased initial disturbance
tions that uses repeated long division. This simply increases all quantities involved in
method, though it was considerably quicker proportion to the disturbance without chang-
than the classical techniques, was still quite ing their time dependence. This condition no
time-consuming without considerable prac- longer exists with nonlinear systems, and a
tice. I therefore attempted to develop a set of separate solution has to be obtained for each
charts for the quartic similar to those of magnitude of motion.
Weiss for the cubic. A brief summary of this
method and a sample copy of the charts are
presented as appendix IV. These charts were
never published. I used them to some extent Solution of a Nonlinear
in my work, but found that the time to solve Problem
for the parameters used in the charts and then
to return the solution to an expression involv- The problem of snaking oscillations, dis-
ing the original variables, required about as cussed in the previous chapter, is an example
much time as Lin's method. This method is of a nonlinear problem. The nonlinearity
presented mainly because it was my only arises because of friction in the rudder con-
excursion into pure mathematics. With the trol system. The rudder, instead of moving in
development of high-speed computers, of proportion to the motion of the airplane,
course, the engineer no longer has to be con- sticks until the aerodynamic hinge moments
cerned with these calculations because exceed the friction. Then the rudder starts to
readily available computer programs can move under the influence of both the aerody-
solve quartic or even much higher degree namic forces and the friction force.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 41

Solution of a Nonlinear Problem

By the time the snaking motion had been formulate, however, because of the simpler
explained, I was familiar with another equations governing the rolling motion. Even
method to analyze nonlinear systems. in this case, calculation of the aileron
Robert T. Jones, while analyzing the stability motions subject to the control friction and
of some of the first guided missiles devel- the aerodynamic moments on the aileron
oped by the Navy, had devised a technique proved to be quite difficult. After working on
assuming that the vehicle performs a sinusoi- the problem as a part time activity for several
dal oscillation. The actual control motion months, I succeeded in obtaining a closed-
resulting from this sinusoidal motion was form solution. A typical example of possible
then calculated or measured experimentally. steady oscillations fcr two cases is given in
He then calculated the response of the sys- figure 5.1. In the first case, the friction is
tem under the assumption that the work done enough to cause the aileron to stick during
on the vehicle per cycle must be the same as part of the cycle; in the second case, the aile-
that done by the action of the control, and the ron motion is continuous.
angular impulse imparted to the vehicle over
After going through all this work, I con-
a half cycle must equal the change in angular
cluded that it was hardly worthwhile to
momentum of the vehicle caused by the
attempt a similar solution for the more prac-
operation of the control. The work and
tical case of the rudder snaking oscillation.
momentum relations give two equations
The results on the aileron oscillations were
from which the frequency and amplitude of a
never published. Many years later, however,
constant-amplitude oscillation may be calcu-
a problem was encountered of control of a
lated. The expressions for the work and
missile in which the ailerons were operated
momentum imparted by the control may be
by a gyroscope sensing rolling velocity. This
shown to be related to the lowest order
problem was exactly equivalent to the prob-
cosine and sine components of the motion of
lem that I had solved. The only difference
the control when expressed as a Fourier
was that the hinge rroments proportional to
series. Jones' technique was therefore equiv-
rolling velocity were applied to the ailerons
alent to the frequency-response method,
by the gyroscope rather than by an aerody-
which was then generally unknown to aero-
namic floating tendency. It was possible to
nautical engineers. It was later found that it
use my analysis to obtain some design data
had been developed to an extensive degree
for the missile. As a result, my analysis was
with different notation by electrical engi-
not completely wasted after all.
neers for studying the stability of feedback
amplifiers. I was intrigued by this method, The frequency-response technique is based
because it appeared that the exact form of the on the assumption that although the control
control surface motion when subject to static motion is nonsinusoi:lal, the response of the
friction would be important. As a result, I controlled vehicle is very nearly sinusoidal
attempted to analyze the snaking problem by because the relatively, large inertia and slow
this method. I first tackled the problem of response of the velticle prevents it from
aileron snaking, a steady oscillation in roll responding appreciably to the higher fre-
caused by friction in the aileron control sys- quencies in the irregular control motion. This
tem. This problem had never been encoun- assumption justifies the calculation of the
tered in practice because of the much larger control motion that it based on a sinusoidal
damping of the roll subsidence mode of an vehicle motion. In ntost discussions of this
airplane when compared with the damping method, however, tht_ use of the lowest har-
of the lateral oscillation. A much larger ten- monic of the contro motion in calculating
dency for the controls to float against the rel- the vehicle response s not justified on physi-
ative wind would have been required to pro- cal grounds. Jones' independent approach,
duce an aileron oscillation than to produce a which first based the response of the vehicle
rudder oscillation. The problem was easier to on the momentum ard work relations over a

42 Monographs in Aerospace History Number 12 Journey in Aeronautical Research

Solution of a Nonlinear Problem

FIGURE 5.1. Time

histories of aileron
snaking oscillations as _,
calculated by my theory. o o

(a) Discontinuous
aileron motion (top).

(b) Continuous aileron

motion (bottom).

Total hinge mom_

_E _"---._._'___ rictio hinge moment, Hf

coT S toT A

Aileron sticks

coT S toT A
1 2 3 4 5 6

o o

Total hinge moment __

_°L_nge_e°mment'eHfM sin tot

toT I
0 1 2 3 4 5 6

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 43

Computation of Lateral Oscillation Characteristics

cycle and later showed that these relations Royal Aircraft Establishment (RAE) after
gave expressions corresponding to the lowest WW II. They called the technique the time
order harmonic of the control motion, gives a vector method and applied it to calculation of
clear physical interpretation of the applica- the Dutch roll mode. Later, W. O. Breuhaus
tion of the frequency-response method to of the Cornell Aeronautical Laboratory, on a
nonlinear systems and provides insight into visit to England, left-ned the method from
the accuracy of the method for this purpose. the RAE engineers and publicized it in the
In the ensuing years, methods based on United States (ref. 5.9). Prof. E. E. Larrabee
the frequency-response method were exten- of MIT, on a sabbatical leave at the NACA
sively used at the NACA and elsewhere for Langley laboratory, showed how the method
design of missile control systems, determina- could be used to calculate stability deriva-
tion of stability derivatives from flight data tives from measured lateral oscillation data
(now called parameter identification), and (ref. 5.10).
calculation of response to turbulence. The
The time vector me_od is based on the fact
many developments of the method, such as
that in a free oscillation of a linear system,
Bode plots, the Nyquist criterion, and root
the variables involved always maintain the
locus techniques gradually became part of
same ratios of amplitude and the same rela-
the aeronautical engineer's mathematical
tive phase angles. Fhe variables can be
depicted on a polar vector diagram in which
the amplitude of each variable is shown by
the length of an arrow and the relative phase
Computation of Lateral angles by the directions of the arrows. In a
damped oscillation, the diagram would rotate
and shrink, but always maintain the same rel-
Characteristics ative magnitude and angular separation of
the vectors. All inforraation about the motion
As pointed out previously, the calculation of
could therefore be obtained by taking a snap-
the airplane modes of motion prior to the
shot of the diagram, which showed it at a
introduction of high-speed computers was a
given instant of time. Fixed relations can also
tedious process. The solution for the lateral
be calculated between displacement, veloc-
modes of motion requires the solution of a
ity, and acceleration of any quantity based on
quartic equation. To simplify the process,
the frequency and damping of the motion.
approximate procedures were developed.
Simple expressions could be found for the The equations of motion involve a series of
characteristics of the spiral mode and the roll terms in each equati _n. In the lateral case,
subsidence, but accurate calculation of the the equations involve roll, yaw, and sideslip.
Dutch roll usually required solution of the Each term in the equations consists of one of
complete equations. A graphical method, these variables multiplied by a stability
first developed in 1937 by R. K. Mueller of derivative which gives the variation of the
MIT, proved to be applicable to the Dutch force or moment with the variable. For
roll mode. Mueller first used the method in example, a typical _.erm might be rolling
developing the world's first electric analog velocity multiplied lry the derivative, varia-
computer, which he used for the calculation tion of rolling mome it with rolling velocity.
of the longitudinal motion of airplanes All the rolling mom_:nts contributed by the
(ref. 5.8). He discovered, however, that once different variables ate added up to give the
he had perfected the graphical method, he no total rolling moment, which must equal the
longer needed the analog computer. Later the inertia in roll multiplied by the rolling accel-
method was discovered independently by eration. This term may be placed on the
K.H. Doetsch and W. J. G. Pinsker, two opposite side of the equation, which results
former German engineers working at the in a sum of terms eqt_al to zero.

44 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Computation of Lateral Oscillation Characteristics

Since the derivatives are constants, each term similar convergent procedure could be per-
varies with time in exactly the same way as formed analytically. After several trials of
the variable in that term. A polar vector dia- different methods, I discovered that the equa-
gram may therefore be drawn that includes tions could be solved for the ratios of the
each term in the equation with the tail of variables. Therefore, a similar iterative pro-
each vector coinciding with the head of the cedure could be set up starting with an
previous one. Since the terms add up to zero, assumed value of frequency and a value of
the diagram must close. A similar diagram damping of zero, to calculate values of the
may be drawn for each of the variables: roll, ratios of the variables. I discussed this
pitch, and yaw. method with Bernard B. Klawans, an engi-
neer in my branch. He worked out the details
In using the method, the quantities to be
of the procedure. As it turned out, with the
determined are the frequency, the damping,
initial values of frequency and damping, the
and the ratios of the variables. Usually a
ratio of roll to yaw could be calculated. Then
value of unity is assigned to one of the vari-
with this value and the values of frequency
ables, such as sideslip, and the ratios of roll
and damping, the ratio of sideslip to yaw
to sideslip and yaw to sideslip are to be
could be calculated. Finally, with these two
determined. All other quantities must be
ratios, a quadratic equation for the Dutch roll
known beforehand or estimated. Usually the
root could be obtained that gave new values
frequency is estimated first from the simple
for the frequency and damping. Mr. Klawans
relation of the frequency of the airplane
checked the results for a wide range of vari-
oscillating with a single degree of freedom in
ables. The calculations could be carried out
yaw. The damping is assumed to be zero.
readily on a slide rule or mechanical calcula-
From these assumptions, initial values of the
tor, and the solutions converged very rapidly,
ratios of the variables may be determined by
usually within two or three iterations.
the closure of the vector diagrams. Then,
Mr. Klawans published a Technical Note on
with these values, more accurate values for
the results (ref. 5.1 l).
the frequency and damping may be deter-
mined and these values used in a second
I thought this procedure was a worthwhile
attempt. The method is therefore an iterative
improvement over the graphical method, but
procedure. Usually the convergence is very
evidently it arrived on the scene too late. In
rapid and requires only two or three itera-
tions to reach a solution. that period, about 1956, analog computers
were already available, and some digital
The time vector method gained considerable computers had been introduced. Soon pro-
recognition because of the rapidity of solu- grams were provided to solve the equations
tion and because it gave a useful physical of motion in a matter of seconds. As a result,
picture of the relations between the variables the approximate procedures that had occu-
in an oscillation. It occurred to me, however, pied the efforts of stability and control engi-
that the graphical work might be avoided if a neers for many years fell into disuse.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 45

CHAPTER 6 Problems Encountered as a
Result of Waru'me

The period between World Wars I and H was The outbreak of hostilities in Europe showed
only 23 years. At the end of WW 1, the reac- immediately the need for increased arma-
tion of the public against military activities ment and other equipment that greatly
resulted in large cuts in expenditures for mil- increased the weight to be carried on air-
itary developments. The initial work of the planes. Prior to the war, a newly developed
NACA occurred in this period, which pursuit airplane like the Curtiss P-36 was
resulted in a substantial background of basic armed with two 30-caliber machine guns.
research information for the improvement of Fighter airplanes that had been tested in
airplanes. The airplanes themselves, how- combat soon were equipped with four to six
ever, started as fabric-covered, externally 50-caliber machine guns, armor plate, bul-
braced vehicles with relatively low power. letproof windshields, and self-sealing tanks.
Without the pressure of actual wartime activ- The result was an immediate increase in
ity, developments by the military services wing loading and a demand for higher
occurred rather slowly. power. Similar trends affected the design of
bomber airplanes.
The main incentive for increase in speed
The primary effect of these changes on the
occurred as a result of various air race
fying qualities of airplanes was greatly
competitions. In England, the development
increased control forces. The hinge moment
of the Schneider Cup seaplane racers,
(and hence the control force) to deflect a
engaged in international competition,
control surface of a given geometric shape to
resulted in high-power engines and metal
a given deflection varies as the square of the
construction that provided the background
airspeed and as the cube of the linear dimen-
for the Supermarine Spitfire fighter. In
sion of the surface. This relation occurs
the United States, a small number of sport
because the load on the surface varies as its
aviation enthusiasts like the Granville broth-
area, which increases as the linear dimen-
ers, who, working in a small garage in
sion squared. The hinge moment is caused by
Springfield, Massachusetts, produced the
the load multiplied by its moment arm from
fastest airplanes in the world. These racers
the hinge line, which varies directly with the
would fly almost twice as fast as the existing
linear dimension.
pursuit airplanes of the Army Air Corps.
With war threatening in Europe, these racing The result of this relation is that if the speed
planes provided the incentive for improved and linear dimension were both increased by
fighters and bombers that became available 10 percent, the hinge moment for a given
in small numbers at the start of WW II. deflection would be increased by 61 percent.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 47

The Quest for Reduced Control Forces

If the speed were doubled and the linear air stream were subject to icing that might
dimension increased by 50 percent, a reason- jam the controls. Balances that broke the
able change between prewar and wartime contour of the airfoil added drag. In addition
fighter airplanes, the hinge moment for a to such practical considerations, balances
given deflection would increase by a factor had to be selected or the basis of the hinge-
of 13.5. moment parameters such as the variations of
control surface hinge moment with angle of
Fortunately, Gilruth 's requirements had attack and with control deflection. These
shown quantitatively the satisfactory levels parameters had fundamental effects on the
of control forces required for different flight flying qualities. The effect on snaking oscil-
conditions and maneuvers. As a result, a lations of the variation of hinge moment with
great deal of research was conducted at the angle of attack has already been mentioned.
NACA and elsewhere to provide means for To obtain light control forces, both of these
obtaining the desired control forces. parameters had to be reduced.

Theoretically, the control forces could be

reduced to zero by reducing these hinge-
The Quest for Reduced moment parameters to zero, but in practice
Control Forces this goal could not be attained. One problem
was the nonlinearity of the hinge-moment
One of the most serious problems encoun- variations. For example, a control surface
tered by designers of military airplanes dur- that was properly balanced at low deflections
ing WW II was keeping control forces desir- might be overbalanced at large deflections. A
ably light while airplanes were being made second problem that limited the degree of
with greatly increased weight, size, and aerodynamic balance on large and high-
speed. Flying qualities research had shown speed airplanes was the effect of small
that maximum control forces should be kept changes in contour due to manufacturing dif-
below what a pilot could conveniently exert ferences. These differences might be almost
with one hand on the control stick or wheel. too small to detect, yet could cause quite
For ailerons, this force was about 30 pounds large changes in the control forces. The
on a control stick or 80 pounds on a control Germans, in an eff¢.rt to obtain very light
wheel. Increasing the mechanical advantage aileron forces on the Mel09 airplane, would
of the pilot's controls was impossible test fly the airplane and try different sets of
because of the limited size of the cockpit and ailerons until one was found that would give
the lag in deflecting a control wheel more forces in the desired range. The British, on
than plus or minus 90 degrees. Studies of testing the Spitfire, mentioned encountering
aerodynamic balancing devices to reduce the "rogue" airplanes that had different charac-
aerodynamic moments on control surfaces teristics from the standard airplanes, the rea-
became one of the main research objectives sons for which could not be detected.
of wind tunnels involved in stability
research. As a result of thest: problems, a practical
limit had to be set on the degree of aerody-
Aerodynamic balance on most airplanes namic balance, which was usually 25 to
designed prior to WW II was usually accom- 30 percent of the lorces produced by an
plished by locating some control surface area unbalanced control .,;urface. This degree of
ahead of the hinge line. Various arrange- balance, however, _ as nowhere near what
ments of these balances are shown in fig- was required to pro,ide desirable handling
ure 6. I. These balances had advantages and qualities on the large ;t or fastest airplanes. In
disadvantages from both mechanical and some cases, forces w Juld have to be reduced
aerodynamic standpoints. In general, bal- to about 2 to 4 percent of those of an unbal-
ances that were permanently located in the anced surface.

48 Monographs in Aerospace lv'istory Number 12--Journey in Aeronautical Research

The Quest for Reduced Control Forces

FIGURE 6.1. Types of

control surface
aerodynamic balance. C
Plain flap

Overhanging balance

Beveled trailing edge

Sealed internal balance

Balancing tab

Frise aileron

External airfoil balance

Horn balance

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 49

Flettner Tabs, Servo Tabs, Spring Tabs, and Whirlerons

An additional factor complicating the provi- Flettner Tabs, Servo

sion of light control forces was the universal
use at the start of WW II of fabric-covered Tabs, Spring Tabs, and
control surfaces. All control surfaces had to Whirlerons
be mass balanced about the hinge line to
avoid flutter. The fabric covering was used to The problems described previously with
reduce the weight of the surface behind the aerodynamically balanced control surfaces
hinge line so that the mass-balance weight could be overcome if it were possible to mul-
could also be reduced. This weight usually tiply the force exerted on the controls by the
exceeded the weight of the control surface pilot. Devices to provide such augmentation
itself because it had to be located quite close of force are called servomechanisms. A
to the hinge line. The fabric coveting bulged number of ways were devised to provide
in or out at high values of airspeed, depend- this force augmentation through aerody-
ing on the external pressure distribution and namic means. Sketches of some of these
on the way the internal volume of the control aerodynamic servomechanisms are shown in
surface was vented. Usually, the fabric near figure 6.2.
the trailing edge was sucked in, which
An arrangement called the Flettner tab had
resulted in a large increase in the variation of
been tried on large airplanes as far back as
hinge moment with deflection. To study
WW I. This device, which was invented by
these problems, it was necessary to know the
Anton Flettner, the same man who invented
tension in the fabric, a factor that varied
the Flettner rotor for propelling sailing ships,
widely with manufacturing techniques and
consisted of a small tab mounted at or behind
with age.
the trailing edge of the main control surface.
Inasmuch as no instrument for measuring The pilot's control was connected just to the
fabric tension was commercially available, I tab. When the tab was deflected, it moved the
made an instrument that served the purpose main control surface in the opposite direc-
quite well. I used a chrome-plated metal bell tion. Because the hinge moment to deflect a
that was left over from the fire alarm system control depends on the product--span times
in the office. A glass tube was attached to the chord squared it is apparent that very large
top. The bell could be sealed to the surface of reduction in the pilot's control effort could
the fabric with grease, and suction applied to be obtained.
the tube to cause the fabric to be sucked up
to the shape of a spherical segment inside the By the time I came to work at the NACA,
bell. A small stylus rested on the fabric, and this device, shown in figure 6.2(a), was usu-
its displacement was magnified by a lever ally called a servo tab, inasmuch as the tab
system inside the belt to move a rod up and acted as an amplifier to augment the pilot's
down in the glass tube. The formula for the force. It had the disa:lvantage that when the
deflection of a spherical diaphragm under pilot moved his control stick while at rest or
pressure then allowed the fabric tension to be while taxiing, the cc,ntrol surface appeared
calculated. floppy and did not respond as expected. To
overcome this problem, a spring was placed
Since wind-tunnel tests had never been made between the control linkage and the main
in high-speed tunnels to determine the effect control surface, so that the surface would
of contours simulating fabric deflection, move in the desired direction even at zero
some analytical work was done to determine airspeed. This arran gement, shown in fig-
the effect of the deflected fabric on hinge ure 6.2(b), was called a spring tab.
moments. Fortunately, this effort was short-
lived because manufacturers soon discarded Some analysis of spr ng tabs had been made
fabric covering and went to metal-covered by the British. I continued this analysis
control surfaces. to determine how various flying quality

50 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Flettner Tabs, Servo Tabs, Spring Tabs, and Whirlerons

FIGURE 6.2. Diagramatic

sketches of /-- Control stick
operated servo controls.

(a) Servo tab (top).

(b) Springtab(middle). _ / I
(c) Geared spring tab Free link

L Control surface

hinge line

Control stick

Spring -__ Free link


_- Control surface

hinge line

Control stick

Spring7 F Free link

II= /

parameters would be affected by this much reduced force at the pilot's control.
arrangement. Because the control surface The variation of the force reduction with
was moved by the spring at low airspeed, the speed could be regulated by the stiffness of
variation of hinge moment with the pilot's the spring. For aileron and rudder controls,
control deflection was equal to that of an this variation of control force with airspeed
ordinary control surface. At high speed, how- was not objectionable, but it might cause
ever, most of the force to deflect the control some difficulties on elevator controls for the
was supplied by the tab, which resulted in a following reasons. A conventional manually

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 51

Flettner Tabs, Servo Tabs, Spring Tabs, and Whirlerons

operated elevator control, in which the varia- thought will allow m appreciation of the
tion of control force with deflection or angle remarkable capabilities of spring-tab con-
of attack varies as the square of the airspeed, trois, which allow a human pilot, applying
may be shown to give a value of force per g comfortable forces to the control stick or
in maneuvers that is approximately indepen- wheel, to maneuver a vehicle weighing as
dent of airspeed. The spring tab, however, much as 300,000 pounds flying at an air-
gives a value of force per g that decreases speed of 300 miles per hour or more.
with increasing airspeed. Pilots had become
The spring-tab analy,_es were discussed with
accustomed to a constant value of force
various manufacturers, and spring tabs were
per g. Decreased force per g at high speed
widely used on large airplanes. Perhaps the
might lead to a tendency to over stress the
most notable example of their use was on all
airplane. This problem may be overcome by
the controls of the Douglas DC-6, which has
use of the geared spring tab shown in
had a remarkable record for reliability and
figure 6.2(c).
excellent handling qualities. Gilruth made
the remark that the DC-6 was "the last of the
A method of using a tab to reduce control airplanes," meaning it was the last very large
forces shown previously in figure 6.1 was airplane in which the pilot, moving the con-
called a geared tab or balancing tab. That is, trols directly with his own muscle power,
the tab was moved by a linkage to deflect in
could easily perform all the desired maneu-
proportion to the control deflection. This vers. This remark is not completely accurate
deflection was usually in the opposite direc-
because other large airplanes of the same
tion so as to reduce the control forces. This
period, such as the Boeing B-29, used
device would give a constant value of force
spring-tab controls. Nevertheless, it illus-
per g. However, this device suffered from the trates the desire of designers and pilots to
same disadvantage as all other types of aero-
have the feeling that they were directly con-
dynamic balance, that if the control had to nected to the controls with a reliable
be closely balanced, the forces would be
mechanical linkage and not dependent on
critically dependent on manufacturing toler-
more complex mechanisms such as hydraulic
ances. I devised the method shown in fig-
or electric systems.
ure 6.2(c) to combine the geared tab with the
spring tab so that at low speed the system Spring tabs had many desirable features, but
behaved as a geared tab, but at high speed the they had two disadvantages. One was a ten-
spring tab came into play to amplify the dency to flutter, which required careful anal-
pilot's control force, and therefore, avoid the ysis to avoid this dangerous condition.
sensitivity of the control forces to manufac- Another was that the tab, deflected in the
turing tolerances. This device was called a opposite direction fr,_m the control surface,
geared spring tab. With proper design, a con- reduced the maximum moment produced by
stant value of force per g could be obtained. I the control. In the ca._e of ailerons, for exam-
published reports on the spring tab and on ple, this effect woLld require ailerons of
the geared spring tab to show how they could increased span, which would leave less room
be designed to give the desired pilot forces. on the wing for high-lift flaps. To overcome
I also determined the limitations of the these disadvantages, ;t device called a whirle-
devices in controlling very large airplanes ron was proposed by Gilruth. This device
and showed that for airplanes as heavy as consisted of a small _¢indmill mounted on a
300,000 pounds (larger than any produced shaft projecting from the trailing edge of the
up to that time), the geared spring tab could wing. The windmill was connected to the
be used to provide desirable handling quali- aileron through gems and shafting so that
ties. These analyses are combined in an about two revolution:; of the windmill corre-
NACA Technical Report (ref. 6.1). A little sponded to full aileron deflection. Normally,

52 Monographs in Aerospace history Number 12--Journey in Aeronautical Research

Research on Closely Balanced Controls

the windmill blades were aligned with the

Research on Closely
wind, but they were connected to the pilot's
control stick so that the pilot could control Balanced Controls
the setting of the blades. The blade angle
The first really serious compressibility
actually was proportional to the difference
effects on airplanes were encountered in
between the pilot's input and the aileron
high-speed dives of fighter airplanes in
deflection so that when the aileron reached
WW II. When pilots put these airplanes in
the desired deflection, the blade angle would terminal velocity dives starting from about
be returned to neutral.
30,000 feet altitude, they were frequently
unable to recover. In cases where the air-
After analyzing this device, a wind-tunnel plane did recover, it often came back with
model was tested. The results appeared the wings bent or wrinkled due to excessive
favorable when the engineers tested it. How- accelerations in the pull out. The cause of
ever, when the Chief of Research, Floyd these problems was the formation of shock
Thompson, was invited to see it, he slammed waves on the upper surface of the relatively
the control stick hard over and the windmill thick wings, which caused flow separation.
hit the stop so hard that its blades and shaft- The separated flow, in turn, resulted in loss
ing were knocked loose. Another problem of downwash at the tail that caused a large
revealed by the tests was that the gyroscopic diving moment and a large rearward move-
moments on the blades, which tend to move ment of the center of lift on the airplane,
which required greatly increased elevator
them into the plane of rotation, caused unsta-
deflection to recover from the dive. The
ble stick forces. As a result, a second model
result was that the control force to recover
was built, with blades swept back and coun-
from the dive was greatly increased, some-
terbalanced to avoid the unstable stick
times beyond the strength of the pilot. As the
forces, and suitable shock absorbers put in to
air temperature increased with decreasing
avoid loads due to hitting the stops. This
altitude, the Mach number decreased, which
device worked well, but as will be discussed
sometimes resulted in rapid recovery of ele-
subsequently, the need for such devices was
vator effectiveness, resulting in an exces-
rapidly decreasing with further airplane sively violent pull out.
developments. Also, no thought had been
given to protection against icing. Gilruth proposed a type of elevator control to
overcome these problems. The elevator was
In running the whirleron tests, I learned to be replaced by an all-movable tail surface
(now sometimes called a slab tail). This sur-
something about the motivation of inventors.
face was hinged at its aerodynamic center, so
In connection with the project, I made a
that very little hinge moment would result
library search to see whether any similar
either from deflection or angle of attack.
work had been done in the past. I found that
Then, the tail was controlled by a servo tab,
the British had tested a similar device on a
which in itself would greatly reduce the
Handley-Page Bomber in the early 1920's.
pilot's control forces. The result was a tail
When I showed this report to Gilruth, he
that required essentially no control force.
commented that they had some really smart
Since flying qualities studies had shown that
engineers back in those days. After that, pilots required a control force in turns or pull
however, Gilruth never showed the slightest up maneuvers, a weight called a bob-weight
interest in further studies of the whirleron.
(a term coined by the British) was placed in
Perhaps the idea of being the originator of a the control system tending to pull the stick
device is an important factor in determining forward. This weight would supply a desir-
the inventor's enthusiasm for it. able value of force per g in maneuvers. This

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 53

Research on Closely Balanced Controls

FIGURE 6.3. All-movable

horizontal tail installed
on Curtiss XP-42

force per g would have the advantage of device had been used on previous airplanes
being independent of fore-and-aft location of to increase the hinge moment due to deflec-
the center of gravity of the airplane. With tion. After this change, the airplane was fly-
conventional elevators, the force per g ordi- able, but the pilots still considered the con-
narily increased with more forward center- trol extremely sensitive and unsatisfactory
of-gravity location, and vice versa, thereby for any kind of practical use.
limiting the center-of-gravity range in which
The cause of the difficulty was analyzed by
satisfactory control forces could be obtained.
comparing the stick force variation in a rapid
pull up on the XP-42 with that on a similar
An all-movable tail surface of this design
airplane, the Curtiss P-40, with a conven-
was installed on a Curtiss XP-42 airplane (a
tional elevator. Both airplanes required the
modification of the P-36) (figure 6.3). After
same stick force in a steady turn, but in the
making a high-speed taxi run, the pilot, Mel
abrupt short-duration pull up, the stick force
Gough, came back extremely perturbed and
on the XP-42 was very small, whereas on the
said that the airplane was completely uncon-
P-40, it was larger than that required for a
trollable. At first, no one could determine the
steady turn at the ,_ame value of normal
reason for this unexpected behavior. As a
acceleration (figure 6.4). These data are
result, I built a tail for a model glider that
somewhat incomplete because of the short
simulated the design features of the XP-42
length of the records, but they show quite
tail. When the model was flown, it performed
clearly the large diff_;rence in control forces
a continuous short-period longitudinal oscil-
to produce approximately the same value of
lation with the tail banging back and forth
acceleration in a rapid pull up. The rapid
between its stops. It was immediately appar-
buildup of stick force with the conventional
ent that the aerodynamic center of the tall
elevator gave the pilot a warning of what
was actually ahead of the hinge point, which
value of acceleration to expect if the maneu-
caused the tail to overbalance. Some rather
ver were continued, whereas this warning
subtle aerodynamic effects were then found
was not present with _.he XP-42.
to explain this condition, including the float-
ing tendency of the tab, the effect of aspect Later, to determine v hether any characteris-
ratio of the tail, and the effect of the trailing tics of the all-movable tail itself caused con-
edge angle of the airfoil used. This problem trol problems, the tat on the all-movable tall
was cured by attaching projecting strips was changed from a servo tab to a geared
along the trailing edges of the tail. This unbalancing tab. With this arrangement, the

54 Monographs in Aerospace history Number 12--Journey in Aeronautical Research

Research on Closely Balanced Controls

RGURE 6.4. Comparison

Curtiss P-40 Curtiss XP-42
of stick forces in rapid
conventional tail all-movable tail
pull ups with Curtiss P-
40 airplane and Curtiss 6
XP-42 airplane. Both
airplanes had the same
force per g in a steady
turn or pull up.
O 4

_J 3 3

% %
2 2

0 0
--.5 0 .5 1.0 .5 --.5 0 .5 1.0 1.5
Time, sec Time, sec

100 100

80 80

60 60

d d
40 E 40

20 _ 20

-20 -20
-.5 0 .5 1.0 1.5 -.5 0 .5 1.0 1.5
Time, sec Time, sec

control forces were similar to those on a con- geared unbalancing tab in the first stages of a
ventional airplane, and the handling qualities maneuver to a servo tab as the maneuver
with the all-movable tail were satisfactory. continued. This behavior could be obtained

Evidently what was needed to obtain the by use of a viscous damper in the linkage to
desired advantages of the servo tab con- allow the tail hinge moment to move the
trolled tail without the problem of inade- damping piston at a controlled rate, which in
quate forces in rapid pull ups was a mecha- turn moved the tab to a position that would
nism that would transfer the tab from a result in zero tail hinge moment. This system

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 55

A Control Surface Hinge-Moment Balancing Mechanism

was tried in the XP-42 with satisfactory always felt that the I_:ACA research on these
results. airplanes was one of the most important con-
tributions to knowledge of handling quali-
After the tests of the XP-42 were started, ties. As a result, I collected the reports that
studies were also made with closely bal- had been written on these projects in a sepa-
anced conventional elevators on a Bell P-63 rate folder and always kept them available
airplane. Again, the control forces were sup- for discussions or consultation with design-
plied by a bob-weight, and the same problem ers of new airplanes.
of lack of warning of the acceleration in a
maneuver was encountered. A similar vis-
cous damper installation (also called a dash-
pot) was installed in series with a spring con- A Control Su_tce-
nected to the pilot's control stick. This Hinge-Moment
system also reduced the undesired variation
Balancing Mechanism
of control forces with center-of-gravity posi-
tion and provided satisfactory handling char- In the previous discussions of control surface
acteristics in maneuvers.
hinge moments, the various approaches to
reducing the control forces to desirable val-
The WW II fighters encountered compress-
ues have been considered. The earliest tech-
ibility effects in high-speed dives, but they
nique was use of aerodynamic balance.
were still incapable of supersonic speeds
When airplane speeds and sizes became too
even in dives. With the rapid development of
large for this approach to work, various types
fighter airplane design after the war, how-
of tab controls such as spring tabs and geared
ever, fighters were developed that reached
spring tabs were used. When supersonic
supersonic speeds, first in dives and later in
speeds were reached, tab controls became
level flight. At supersonic speeds, the effec-
ineffective and hydraulic power controls
tiveness of a tab on a tail surface is very
were adopted.
small, and the hinge moments of the surface
itself undergo erratic variations in passing In the 1940's during the early stages of this
through the transonic region. As a result, the progression, I conceived a fundamentally
use of a servo tab to operate the tail as envi- different idea. I reasoned that if the position
sioned by Gilruth was not satisfactory. For of the control surface hinge line could be
these airplanes, hydraulic powered controls moved to the point that the lift on the surface
were developed that provided far more force multiplied by the distance to the hinge line
to actuate the control surfaces than the
just balanced the aerodynamic moment on
human pilot could supply. As a result, all the the surface, the surface would be in balance
previous emphasis on very closely balanced and the pilot's contr _1 force would be zero.
controls disappeared. The designer was still This idea may be lik reed to moving the bal-
faced with the problem of providing satisfac- ance point on a see-_aw so that it is in bal-
tory control forces for the pilot, however. In ance even though the weights on the two
most cases, the force input to the hydraulic ends are different.
actuator was close to zero, and all the pilot's
control force was supplied by a feel device. My first analyses considered that a motor or
servomechanism would be used to move the
The design of these feel devices proved to be hinge. Later, I did soJ ne work on the idea that
a difficult problem. In general, most design- the lift or control for, :e could move the hinge
ers first envisioned a bob-weight just as had line directly througlL some aeromechanical
been tried on the XP-42. Fortunately, the mechanism.
NACA research on this airplane and on the
P-63 showed how to design a satisfactory The logic for moving the hinge basically
feel device for longitudinal control. I have depends on the ide_ that if the lift on the

56 Monographs in Aerospace [fistory Number 12--Journey in Aeronautical Research

A Spoiler Operated Servo Control

surface is up, then a nose-up moment on the

A Spoiler Operated
surface should be balanced by moving the
hinge line forward, and a nose-down Servo Control
moment should be balanced by moving the
hinge line back. If the lift on the surface is The servo-tab and spring-tab controls
down, these motions should be reversed. described previously operate well at sub-
sonic speeds, but become ineffective at
A practical problem appears in that as the lift supersonic speeds because the ratio of con-
goes through zero, the hinge must be moved trol surface hinge moment to lift produced
from infinity in one direction to infinity in increases greatly under these conditions. The
the other direction. In many cases, the lift on reason for this change is primarily that the
a surface does reverse or oscillate in the pressures on the upper and lower surfaces of
course of a maneuver. As a result, the mecha- the wing caused by control deflection do not
nism for moving the hinge would have to be carry forward of the hinge line at supersonic
very fast. speeds. If a tab were used in an effort to
I first concluded that this solution would be defect the control, the tab would have to be
so large that the lift produced by the tab
impractical, but as devices for moving con-
would just about offset the lift due to the
trois became more and more complex, I kept
control surface. To overcome this problem, I
coming back to this idea to try different
conceived a device in which a spoiler ahead
approaches. My last attempt at analysis was
made about 1952. I never did reach a solu- of the control surface hinge line produced a
moment to deflect the control. This device
tion that appeared at all promising.
has also been referred to as a rotary servo tab
One problem that soon appears in attempts at or a vane-type servo control.
analysis is that the problem is nonlinear if a
reversal of lift is taken into account. I did not A sketch showing the operation of the device
have the necessary means to analyze such a is shown in figure 6.5. A vane or spoiler
problem. An analog computer would have operated by the pilot is mounted in a tube
made it possible to simulate different types ahead of the control surface hinge line. When
of mechanisms, but such computers were not the vane is deflected, its drag causes the tube
available at Langley at the time. to rotate. The rotation of the tube is linked to
that of the control surface, so that when the
Later, when analog computers did become
vane moves up the control surface also
available, it always appeared to me that moves up. This device should remain effec-
many of the nonlinear problems that engi-
tive at supersonic speeds because the spoiler
neers worried about in earlier years were no itself is known to produce a lift force in the
longer of interest to them and were never
correct direction, the reduced pressure
studied with analog computers. The excel-
behind the spoiler should aid in deflecting
lent analog computer complex at Langley the control surface, and the control surface
Research Center was discontinued in 1990
itself would produce lift in the same direc-
because of lack of use. Now, such problems tion. By use of a feedback linkage from the
can be studied by digital computers with pro- control surface to the tab input link, the sys-
grams that allow setup of the problem with tem could readily be arranged to operate in
block diagrams similar to those used for ana-
the same way as a servo tab or spring tab.
log computers. The speed of digital comput-
ers may not allow real-time solutions of A model of the device was built and tested
problems involving very high frequencies, by the transonic bump technique (fig. 6.6).
such as might have been encountered with The results showed that the control did retain
the balancing mechanism, though such prob- a reasonably large amount of effectiveness at
lems could be studied using an extended time least up to a Mach number of 1, the highest
scale. Mach number of the tests (ref. 6.2). A patent

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 57

The Introduction of Swept Wings

FIGURE 6.5. Diagram of

operation of vane-type
servo control.

(a) Control in neutral


(b) Vane deflected


(c) Vane and control

surface deflected

was obtained on the device, but it was never (ref. 1.1). Though there were initial doubts
used on an airplane. These tests were pub- about the validity of this concept, by the end
lished at a time when the use of hydraulic of the war experimental studies by the free-
controls was becoming common, so that the fall method and other techniques had shown
need for such a system no longer existed. that the concept was indeed correct. In addi-
This device, however, is probably the only tion, the developmer_t of the jet engine had
type of manual control ever tested that would made flight in the transonic range feasible.
retain enough effectiveness at supersonic
Prior to these developments, swept wings
speeds to control an airplane. The wind-
were in poor repute -'or use on subsonic air-
tunnel tests showed that there was consider-
planes. It was generally known that the stall-
able buffeting of the control when it was
ing characteristics of swept wings would be
deflected at high speed, which might be a
adversely affected by the flow of boundary
serious problem on a full-scale airplane.
layer to the tips as the stall was approached,
which causes a so-called tip stall that results
in rapid roll at the s all with no warning in
The Introduction of the form of buffetiz_g. In addition, swept
wings were poor frcm the structural stand-
Swept Wings
point and had lowe; maximum lift coeffi-
One of the most important developments in cients than unswept wings. As a result, very
the design of high-speed airplanes during the little experimental data were available on the
wartime years was the realization that a aerodynamic charact:;ristics of swept wings.
swept wing would have much less drag in the The result of this situation was that in practi-
transonic and supersonic speed range than an cally every wind tun[ el, the work in progress
unswept wing. The story of this discovery at was stopped and a program of research on
Langley is told in Engineer in Charge swept wings was init ated.

58 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

The Introduction of Swept Wings

FIGURE 6.6. Photograph

of test model of vane-
type servo control.
Semaphore-type vanes
deflected 12 degrees.

To get some idea of problems that might be not then known to most researchers. I pre-
encountered with swept wings, I made sim- sented these results at meetings of the
plified calculations to determine the magni- Langley Stability and Control committee.
tude of some effects that might result from
this configuration. The dihedral effect, or I also made a simplified analysis of the effect
rolling moment due to sideslip, of a swept of sweep on the weight and flexibility of a
wing was known to increase with lift coeffi- wing. The use of sweep was intended to pro-
cient. In addition to this problem, however, vide a high critical Mach number, but if the
was the effect of the airfoil section pitching wing had to be made thicker to prevent
moment on the stability derivatives. On an excessive weight or flexibility, this advantage
unswept wing, the section pitching moment might be offset. These results were circulated
affects only the longitudinal characteristics, in-house to members of the Structures
but on a swept wing the pitching moments Research Division, who acknowledged that
about axes swept with the wing panels have the results were sound and would have to be
components that affect the rolling moments. considered in selecting the wing span and
My analysis showed that a swept wing with thickness. Of course, much more detailed
cambered airfoils would have a reduced study would be required to actually design a
dihedral effect. Twist of the tips in the direc- swept wing for an airplane. The structural
tion to reduce the lift on the tips would also problems that I anticipated have been allevi-
reduce the dihedral effect. These effects were ated on most transports by use of a section

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 59

The Introduction of Swept Wings

FIGURE 6.7. Bell L-39

airplane incorporating
experimental swept wing
to study low-speed effects
of wing sweep.

near the wing root with a straight trailing ment of the slotted-throat wind tunnel made
edge and by the concentration of stiffness in transonic wind-tunnel testing possible.
a spar through the wing root as near as possi- To get full-scale data at an early date, the
ble to the trailing edge. Bell Aircraft Corporation also produced for
the NACA the L-39 Idrplane, a P-39 airplane
Experimental studies of the effects of sweep with experimental swept wings (figure 6.7).
were also made in the Flight Research This airplane was n)t capable of transonic
Division by methods devised by Robert R. speeds, but studies were made of the stability
Gilruth. These methods, to be discussed and control characteristics and of expected
later, were the wing-flow method and the low-speed problems such as poor stalling
free-fall method. The ready availability of characteristics. The use of wing slats and
these methods provided much-needed data to other stall-control devices was studied in
the aeronautical industry before the develop- flight.

60 Monographs in Aerospace t listory Number 12--Journey in Aeronautical Research

CHAPTER 7 Effects of WorM War H on
Research Activities

The research work performed by the NACA surprisingly even keel, probably because the
at Langley had experienced a slow but same director and many top management
stead), growth from its start in 1920 to the personnel continued in their positions
time I arrived at the center in 1940. The throughout this period. A brief summary of
number of personnel had increased from some of the effects of the war on my work in
15 to 739. The research facilities included the Flight Research Division is presented in
several wind tunnels and the Flight Research
this chapter.
Hangar, which were most all located in the
East Area, a section of Langley Field on the
side closest to the center of Hampton. 1 had
only a relatively short period of employment
Effect of the war on .....
before war was declared following the bomb-
ing of Pearl Harbor on December 7, 1941. Living and Working
As a result of the war, the aviation industry Conditions
in the United States experienced a period of
phenomenal growth and the research activi-
When I came to Langley in July 1940, I
ties at Langley were similarly expanded.
entered into what seemed like an ideal
The number of professionals at Langley
environment from the standpoints both of
increased from 277 in 1940 to 1158 in 1950,
living and working. I have described these
and the total number of employees increased
conditions in more detail in a journal article
from 739 to 3388. Two new NACA research
(ref. 3.1 ). This was my first experience living
centers, the Ames Aeronautical Laboratory
away from home and free from the rather rig-
in Mountain View, California, and the Lewis
Laboratory in Cleveland, Ohio, were estab- orous program of studies at MIT. I enjoyed

lished just before the war and were managed the opportunities for social life, the friendli-

by personnel transferred from Langley. The ness of the local people, and the many recre-

war naturally had a major effect on the ational activities. The model airplane club,

scope of the work conducted. New technical with members that included many expert
developments such as the jet engine and the model builders who had been hired by the
capability of flight at supersonic speeds Langley management from all over the coun-
required many new research facilities and try as technicians and model builders in the
created new technical fields of interest. The shops, gave me an opportunity to meet new
administration of the center, however, and people and to continue my hobby of model
the environment for research continued on a building and flying.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 61

Some Effects of the War on Research Activities

At the NACA hangar where I worked, some courses, under sponsorship of the University
evidence of military buildup was evident. of Virginia, were given at night at Hampton
Gilruth had already run flying qualities tests High School. Preparation of lectures for
on some of the new military airplanes, such these courses was a burden on me because I
as the Seversky P-35 pursuit plane, the am not a fluent speaker and had to know the
Boeing B-15 and B-17 bombers, and the course material in detail before I could
Vought F4U-1 Corsair Navy fighter. There present satisfactory lectures on it.
were still many active projects directed at
One source of concern during the war was
private and commercial flying. About five
induction into the armed services. Some of
light planes, including those built by Piper,
the Langley employees, of course, felt com-
Taylorcraft, Stinson, and Bellanca were
pelled by patriotism to join the military ser-
undergoing tests, and Gilruth had modified a
vices. Others had a hard time convincing
Piper Cub to make it stall- and spin-proof.
their local draft boards that aeronautical
Ongoing programs were being conducted by
research work was of sufficient military
the Aircraft Loads Section to make routine
value to merit deferment. Eventually, the
measurements of gust loads on airplanes in
NACA management in Washington, under
commercial operations, and the first pressur-
pressure from the industry to maintain the
ized commercial transport, the Lockheed
research and development capabilities at the
C-35, was used to explore turbulence at high
altitudes in thunderstorms. NACA laboratories, obtained approval from
Congress for a plan in which all employees
The impact of the war hit suddenly with doing essential work at the center would be
Pearl Harbor. Hangars were camouflaged inducted into the Army then immediately
and blacked out. For a month or so after the returned to their jobs at the center in civilian
declaration of war, Langley engineers were status. The details cf this arrangement are
assigned to guard duty at night in the described more fully in the book Engineer in
research facilities to warn of possible terror- Charge, a history of Langley by Dr. James
ist attacks. When I was on this duty, I tried to Hansen (ref. 1.1). I was affected by this plan
do a lot of report writing, but I found that the by having to go up to Richmond for my
reports written in the loneliness of night induction physical, but then immediately
tended to be rather rambling and usually returned to my regular work.
required rewriting by the light of day. Soon,
regular security guards were hired and engi-
neers were relieved of this responsibility.
Some Effects oj the War
The war inhibited social life for two reasons. on Research Activities
First, the work week was increased from
44 hours to 48 hours. With work all day Prior to the war, Langley had close relations
Saturdays, there was not much time even for with the military services through member-
necessary shopping. Also, gas rationing was ship on the main conmittee of the NACA of
in effect, which prevented any trips except in high-ranking military, officials. The projects
the local area. Everyone at Langley worked done by Gilruth on handling qualities, as
very hard during the war. This was a period well as studies by other sections of the Flight
of rapid expansion in which new engineers Research Division on loads and perfor-
were hired from engineering schools all over mance, were strongly appreciated by the mil-
the country. Many of the NACA engineers itary services. As a result, an arrangement
were sent on recruiting trips, but I did not was made to supply the NACA with the third
receive this kind of assignment. Later, start- airplane off the production line (later
ing in 1943, I taught courses to many of the changed to the fifth) of every new military
new employees in the subjects of stability airplane produced. This rule was followed
and control and handling qualities. These for many years. In view of the large number

62 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Some Effects of the War on Research Activities

of new military airplanes produced during beam of light on to a strip of photographic

the war years, this rule supplied Langley and film. The film was moved between a pair of
the other NACA centers with an excellent drums by an electric motor. A second mirror
opportunity to contribute to the war effort by was flicked up and down by a timer and
recommending improvements as well as by reflected another light beam onto the film to
doing research on new problems. give a time reference. A separate instrument
was usually required for each quantity being
One effect of the war effort was a marked
measured. The Instrument Research Division
increase in the number of visits by members supplied services for calibrating and install-
of the aeronautical industry to discuss the ing these instruments.
work in progress and a similar increase in the
number of trips made to companies and other
At the time of WW II, the aircraft companies
organizations to assist in the design and
lacked any such facilities for recording quan-
development of new airplanes. These visits
tities in flight. The usual technique was to
and trips, which continued after the war,
construct a photopanel, a light-tight box in
served to solve immediate problems as well
which a movie camera photographed an
as to stimulate the development of aviation
array of standard cockpit dial instruments
through interchange of ideas.
along with a clock to give a time reference.
One of the incentives for visits of the NACA Working up the data from these instruments,
personnel to companies was the availability frame by frame, was exceedingly tedious.
at Langley of recording instruments for flight
research. When the Langley Memorial Aero- One of my first trips was to the Republic
nautical Laboratory first opened at Langley Aircraft Corporation at Farmingdale, Long
Field in 1920, the director, or Chief Physicist Island, New York, accompanying Gilruth and
was Frederick H. Norton, a young doctoral an instrument man. The purpose of the
graduate straight from MIT. During his brief trip was to install instruments and make
tenure at Langley, 1920-1923, starling with a measurements of the rolling effectiveness of
staff of seven professionals, he constructed a the ailerons on a new airplane derived from
wind tunnel, started programs in aeronautical the P-47 Thunderbolt (figure 7.1). Later, a
and physical research, and established a Navy pilot flew a captured Japanese Zero to
flight research division. One of his objectives Langley for installation of flight research
was to design recording instruments to make instruments (figure 7.2). He returned
accurate measurements on airplanes in flight. with the airplane to Boiling Field, near
Dr. Norton went on to become a Professor of Washington, D.C., and made a single flight
Physics and later of Ceramics at MIT. in which many of the important maneuvers
Shortly before Norton's death at the age of to study flying qualities were performed. I
96, Dr. Hansen had a telephone conversation evaluated the film data and published a
with him, during which Dr. Norton stated report on these results (ref. 7.1). Since our
that the contribution that he made at Langley instrument technicians usually took several
in which he took most pride was the develop- weeks to install instruments in an airplane
ment of recording instruments to make scien- and the program for measuring flying quali-
tific measurements on airplanes in flight. ties often required about 20 flights, I was
impressed by the speed of the Navy pilot's
When I started work at Langley, the instru- investigation. For some years after that, I
ments used were still the same type as those tried to popularize the idea of building a self-
developed by Norton. They incorporated a contained set of instruments for a quick fly-
sensor to detect the physical quantity being ing qualities investigation, as well as a regi-
measured. The output of this unit was men of maneuvers that could be used to con-
mechanically linked to a small mirror, about duct the investigation in one or two flights.
one-quarter-inch square, that reflected a These ideas never caught on, however,

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 63

Some Effects of the War on Research Activities

FIGURE 7.1. Republic

P-47D-30. This
airplane and its later
developments were
influenced by work at

FIGURE7.2. Mitsubishi 00
airplane, popularly
known as the Japanese
Zero. This captured
airplane was instru-
mented at Langley and
tested by a Navy pilot.

because of the availability of the standard were unsuitable for flight recording because
techniques and the press of urgent work. they usually had no means to disconnect the
erection system, waich caused errors in
Though Langley was well ahead of the
accelerated maneuvers. Later developments
industry in the use of flight recording instru-
of magnetic tape recording and the use of
ments during the early 1940's, it was appar-
multichannel oscillographs to record many
ent to me, after having studied instrumenta-
tion under Dr. Draper at MIT, that the quantifies on one fihn were first introduced
instruments themselves were already well in telemetering syste_ns used by the Pilotless
behind the state of the art. The instruments Aircraft Research [_ivision, but were not
had inadequate damping and picked up large available for flight r( search. It was not until
amounts of engine vibration in flight. Also, many years later, wh,;n such instruments and
gyroscopic instruments to measure attitude sensors became com:nercially available, that

64 Monographs in Aerospace History Number 12 Journey in Aeronautical Research

Some Effects of the War on Research Activities

the old NACA instruments were finally Ames Research Center at Mountain View,
abandoned. California. Still, Douglas representatives
occasionally visited Langley to discuss han-
Returning to the subject of conferences and
dling-qualities work. Boeing, throughout its
trips to industry organizations, an interesting
existence, appeared to me to be extremely
point to mention is the large difference in the
independent. The manufacturer built its own
use of the NACA techniques and assistance
wind tunnel and rarely consulted with Lan-
by different companies. On the East Coast,
gley on new developments.
Republic and Vought seemed to have close
contact with the work in the Flight Research Many visits were also made to military
Division, whereas Grumman and Martin
installations such as the Naval Test Pilot
were much less involved. An example of the School at Patuxent River, Maryland, and the
dependence placed by Republic on the Flight Control Laboratory at Wright-Patter-
NACA capabilities occurred shortly before I son Field. I gave a series of talks to the test
came to work. Gilruth was asked to comment
pilots at Patuxent on compressibility effects
on the design of the airplane that later on fighters capable of transonic speeds.
became the P-47 Thunderbolt. Gilruth
advised increasing the size of the horizontal Several cross-country trips were made to
tail because of experience with inadequate conferences at the NACA Lewis Research
longitudinal stability on high-powered air- Center at Cleveland and to the NACA Ames
planes with large propellers and the resulting Research Center at Mountain View. In these
slipstream effects. This advice was taken by trips, a group of about 15 engineers would
Republic, which resulted in a design with get into the NACA DC-3, which was piloted
noticeably larger horizontal tail than most of by NACA test pilots. The trip to the west
the existing fighters. This feature proved coast usually required about two stops, one
valuable later because the airplane had the at Memphis, Tennessee, and one in such
turbosupercharger and other equipment places as China Lake, California; E1 Paso,
installed aft of the cockpit, which resulted in Texas; or Albuquerque, New Mexico. With-
more rearward center-of-gravity location out these trips, I would never have seen these
than originally planned. Without the larger cities firsthand with the beautiful mountain
tail, the airplane would never have had ade- scenery and the desert climate. I remember
quate longitudinal stability. in my first trip to the Ames Lab that we

Of course, I am not familiar with the work stayed in Rickey's Motel in Sunnyvale and
had a chance to see the sights of San Jose,
done in the wind tunnels. Most of the compa-
such as the Rosicrucian Museum and the
nies at that time did not have their own wind
Mystery House. I was tremendously
tunnels, so many of them probably employed
impressed by the beautiful city of San Jose,
Langley wind tunnels in tests of their new
with its gardens and palm trees, and the
designs. The Bell Aircraft Corporation, after
mountains rising in the distance both in the
the war, was strongly involved in the
East and the West. In those days, much of the
Research Airplane Program. I made several
area was still filled with orange groves. I
trips to Bell in connection with their pro-
thought it would be a wonderful place to live.
posed designs.
Evidently many other people felt the same
On the West Coast, many companies were way, as San Jose for many years after that
more closely associated with the NACA was the fastest growing city in the country.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 65

CHAPTER 8 New Ideas The Fruits of

One of the main objectives of a research invited to suggest new ideas in hope of com-
organization such as the NACA is to come up ing up with something useful. Poor ideas are
with new ideas that contribute in some way worse than useless, particularly if the pro-
to progress in aeronautics. It is surprising, poser is a person with enthusiasm and per-
however, how few engineers working at such sistence who is able to convince supervisors
a center ever conceive a new idea in their that his idea should be developed, often at
entire careers. Scientists noted for advance- considerable cost in manhours and facilities.
ments in their technical fields have stated
(and I believe it to be true) that to produce a In my position, 1 was fortunate in having
worthwhile new idea, the researcher must available the tools for experimental research
first have a complete knowledge of all that required development of new ideas.

advancements that have been made previ- Some of these ideas were of limited scope
ously. In addition, knowledge of related and were applicable to the problem at hand
fields is often helpful because the ability to or soon became obsolete with the rapid
recognize interrelations between disciplines progress of aeronautics. Others had broader
will reveal new concepts. A second important application. This chapter presents some
factor, based on my own experience, is to be examples of new ideas that I developed in my
involved in experimental work. Such work research work.

often requires new techniques, and the old

adage "necessity is the mother of invention"
certainly applies in this case. In addition,
The First Piloted
experimental work often produces unex-
pected results that stimulate thinking on the Simulator Used for
causes of these results.
Research at Langley

Of course, there are some engineers who One subject investigated in flying qualities
suggest new ideas that are not worthwhile tests was the ability of the pilot to control the
advancements. They may be ideas that have short-period yawing oscillations of the air-
already been tried or are contrary to plane. This mode of motion, called the Dutch
accepted physical principles. Such ideas may roll, was usually poorly damped, and with
result from a lack of sufficient background in the advent of swept-wing fighter airplanes
the field. The possibility of introducing bad flying at high altitude, the damping tended to
ideas is often the downfall of brainstorming become still less. On actual airplanes, how-
sessions in which all members of a group are ever, it was at that time difficult to vary the

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 67

The First Piloted Simulator Used for Research at Langley

FIGURE 8.1. Yaw chair

used for studying ability
of human pilots to
control lateral

damping to investigate the limiting condi- interesting result was that the pilot could
tions under which the pilot's control would damp out the oscillations just as well with
become marginal. To study a wider range of his eyes closed as with them open. This
conditions, a simulator called the yaw chair result showed the inportance of so-called
was built (figure 8.1). This device consisted kinesthetic cues, whi:h are obtained from the
of a chair mounted in a frame free to pivot on sensory mechanisms in the inner ear, in the
a vertical axis. The frame was restrained by control of airplanes by human pilots.
springs and was controlled by other springs
Though the yaw chair was a simple mecha-
operated by the rudder pedals. A hydraulic
nical device capable of studying only one
servomechanism was also installed that
type of control problem, it foreshadowed
could apply moments to the chair in phase
the development of much more complex
with its angular velocity, thereby allowing
general-purpose simulators that became pos-
the damping to be varied from stable to
sible with the develcpment of computer and
unstable. The availability of facilities in the
display technology. Operation of such simu-
flight research hangar to build devices of
lators later became a major research activity
angle iron and welded steel tubing was very
at the Langley Research Center. Before such
useful in constructing a simple device of this
developments, however, a simulator called
the NAP chair (for normal acceleration and
The studies made with the yaw chair showed pitch) was built that l_rovided the pilot with a
that the pilot could not damp out the oscilla- simulation of the short-period vertical
tions if the period was much less than one motion of an airphne applicable to tasks
second. Effects of the degree of damping or such as formation flight and in-flight refuel-
instability were also studied (ref. 8.1). One ing. In the 1950's, a _,imulator was developed

68 Monographs in Aerospace History Number 12 Journey in Aeronautical Research

Roll Coupling

to impart three-axis rotary motion to a cock- sensitivity and dynamic characteristics of

pit. This simulator had the objective of control force and visual and motion cues of
studying pilots' ability to control lateral the simulator. A human pilot can track a
oscillations involving large rolling motions. moving target with an accuracy of better than
At this time, however, the space age started 0.2 degree. Obviously this problem cannot
and the gimbal system ended up in applica- be studied with a simulator that projects
tions to study atmospheric entry or docking. the target with an accuracy of 0.5 degree. In
The rotary motion simulator was never used fact, the pointing accuracy of the simulator
for its intended purpose. projection system should be better than
0.02 degree, or one-tenth the pilot's tracking
Though these simulators had limited applica- error.
tion, they had the advantage of providing
very precise reproduction of the motion of
the actual vehicle together with visual cues
Roll Coupling
that came directly from the real visual scene.
Later simulators frequently used more com-
Roll coupling means the effect of rapid roll-
plex mechanisms to produce the motion of a
ing of an airplane on the motion in other
very heavy cockpit, but because of the mass
directions, that is, about the pitch and yaw
of the moving system and friction in the
axes. At the time I went to college, such cou-
drive mechanism they were unable to repro-
pling was not a subject discussed in the sta-
duce the actual motion precisely. Likewise,
bility and control courses. In general, air-
visual scenes that were produced by televi-
planes flying up to that time had not
sion or computer techniques frequently
experienced problems from this effect in nor-
lacked detail of the actual scene or involved
mal flight. It was recognized that in stalls or
lags inconsistent with the motion of the vehi-
spins, the interaction of motions about the
cle. Such simulators were often found to be
various axes had to be considered, but analy-
unsuitable for research on the pilot's
ses of these phenomena had rarely been
response characteristics and his interaction
attempted because of the complex and
with the vehicle dynamics. Simulator engi-
unknown aerodynamic effects caused by the
neers were slow to recognize these deficien-
stalled flow on the airplane.
cies, and when they did, much of the
research on the simulators was devoted to I would not have tried to analyze the effect of
improving the simulators rather than on rolling if it had not been for some experi-
solving real airplane control problems. In mental data that required explanation. Dur-
recent years, the importance of poorly ing WW II, as mentioned previously, air-
damped high-frequency airplane motions has planes had started to encounter serious
decreased because of the widespread use of compressibility effects in dives at transonic
automatic stability augmentation devices to speeds. Wind tunnels were unable to investi-
avoid such characteristics. As a result, many gate these problems because of a phenome-
simulator studies deal with problems such as non called choking. In a closed-throat tunnel,
development of instrument displays and as the airspeed is raised to the point at which
training of pilots in specific vehicles. Some shock waves start to form on the model, a
modern simulators are used for development small additional increase in speed will result
of flight control systems. For example, the in the shock waves spreading across the
ability of the human pilot to track a moving entire test section. Further attempts to
target with various types of control systems increase the Mach number in the test section
has been studied. In such applications, con- by applying more power to the tunnel simply
sideration must always be given to the sensi- result in greater pressure drop across the
tivity and speed of response of the human shock waves with no increase in Mach
pilot's sensory system as compared to the number.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 69

Roll Coupling

Gilruth, then head of the Stability and Con- A question arose as to the effectiveness of
trol Branch, was well aware of the compress- the elevator control in the transonic speed
ibility problems because of the tests being range. To study this problem, as well as to
made on high-speed fighter airplanes. He investigate the ability of the free-fall method
considered how the airplanes being flown in to obtain data in lifting conditions, it was
the Flight Research Division might be used decided to test a model of the XS- 1 airplane.
to obtain data at transonic speeds. He pro- A photograph of the model is shown in fig-
posed two new test methods, one called the ure 8.2. The model had a wing span of 7 feet
wing-flow method and the other the free-fall and weighed 1350 pounds. To maintain the
method. The wing-flow method will be dis- desired lifting condition in flight, the model
cussed subsequently. F. J. Bailey, head of the was equipped with a very simple type of
Performance Branch, was also responsible autopilot. An accelerometer sensed when the
for many of the ideas required to implement normal acceleration _that is, the acceleration
the free-fall method. The free-fall method normal to the plane of the wings and fuse-
consists of dropping heavily weighted test lage) went outside the range 0.4 to 0.8 g. If
models from an altitude of about 30,000 feet. the acceleration went outside these limits,
These models would reach Mach numbers of the elevator was moved at a slow rate to
1.0 to 1.2 before impacting the ground. New bring the acceleration back into this range.
developments in telemetering and radar dur- The model was in_ended to roll slowly
ing the war made it possible to obtain useful during the drop to maintain a predictable
aerodynamic data from these models during trajectory.
the course of the drop.
When I fly model airplanes, I customarily
The free-fall method was rapidly put into look at the model from the rear to check
operation, using first a B-29 bomber and whether the wing is twisted. As a matter of
later an F-82 fighter, to carry the test models. habit, I did this on the XS-1 model. Though
Actually this was a joint project with the the wing was supposed to be machined to an
British, in which the British were to develop accuracy of one thousandth of an inch, !
recoverable models that could be reused to could readily see that the wing was twisted
obtain more data, whereas the Americans about one-eighth of an inch at the trailing
under Gilruth were to develop expendable edge of each tip. Calculations showed that
models. After about two years of operation, this amount of twist would have caused the
Gilruth's group had successfully tested about model to roll at a rate of about 250 degrees
20 models, whereas the British had still to per second during the high-speed portion of
the dive. This rate was considered excessive
make a successful recovery of a model. The
validity of Gilruth's direct approach was thus (though we had no wiy to know, at that time,
demonstrated. what rate was reail_ excessive). Therefore
two small wedges, lit:e ailerons, were placed
Most of the free-fall testing was under the on the trailing edge to oppose the built-in
direction of Charles W. Mathews and Jim twist.
Rogers Thompson of the Flight Research
The model was dropped successfully. The
Division. The models tested were symmetri-
exact date of the drop is not known, but a
cal bomb-shaped streamline bodies with cru-
ciform tail fins and, in some cases, with cru- memorandum by C. W. Mathews for H. A.
Soul6 that gave the results of the drop is
ciform wings. These models fell in a zero-lift
dated September 3, 1_47. Data from the drop
trajectory, with measurements confined to
are shown in figure 8.3. The longitudinal
drag of various components.
control worked as i itended. At low Mach
About that time the Bell XS-1 (later called numbers, in low-den ;ity air at high altitude,
the X-I) rocket airplane was being devel- the full-up deflection of the elevator of
oped to attempt to reach supersonic speeds. 10 degrees is not enough to maintain 0.4 g.

70 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Roll Coupling

FIGURE 8.2. Drop model

of the XS-1 airplane.

FIGURE 8.3. Flight data

from drop test of XS-1
model. j_.--
.8 /
6 /
r. /
a:: .6


_=-_o .......... A,_M

...... _ vv
........ I M,v-- -_


_-_._ ...------'- v _. _,v.,_v ._r.._ _

o _---_-_ n__A/_ j

: o

o o 8 16 24 32 40 48 56 64
Time, sec

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 71

Roll Coupling

As the speed increases, the model is held in equations by considering small displace-
the desired limits of acceleration except for a ments in pitch and yaw and by assuming the
period of rather wild oscillations at a Mach rolling velocity to be constant. I had consid-
number around 0.75. At the highest Mach erable difficulty in deciding on the signs of
number, full-down deflection of the elevator the terms involving the coupling effect of the
of 4 degrees is not enough to hold 0.8g, and rolling velocity. Finally, I worked out the
the acceleration builds up to about 2g. equations both ways and arrived at the cor-
rect sign by considering the simplified limit-
The model was not equipped with any instru- ing case of a rotating rod, the solution for
ment to measure rolling velocity, but when it which I could visualize. The resulting equa-
was dropped, it was observed in the optical tions were also expressed in nondimensional
tracker to be rolling by the flashing of light form. Restoring moments were expressed in
from the wings. Near the end of the flight, terms of the natural frequencies in pitch and
however, the model stopped rolling and per- yaw of the nonrolling airplane, a technique
formed a pull out. It flew far from the desig- taught for oscillating systems by Professor
nated drop area on a former bombing range Draper. In addition, the frequencies were
called Plum Tree Island and crashed about a
nondimensionalized in terms of the rolling
mile away in the swamps around the town of frequency.
When the characteristics of the XS-1 model
Of course, dropping a heavy bomb in a popu- were substituted in the equations, the results
lated area is not considered very good prac- clearly showed the possibility of a divergent
tice, but evidently no one was aware of the motion even at relatively low values of roll-
incident. As a result, nothing was ever said ing velocity. The instability was likely to
about it, and the model, to this day, is proba- occur when the values of longitudinal stabil-
bly buried 15-feet deep in the mud some- ity and directional stability were markedly
where around Poquoson. different and when a large amount of the
weight was distributed along the fuselage. In
In examining the records further, the oscilla- the case of the XS-; model, the directional
tion mentioned previously was found to rep- stability was high, but the longitudinal stabil-
resent a violent pitching in angle of attack ity was low, partictdarly at high subsonic
between the positive and negative stall. The speeds.
motion below the stall was actually a diver-
gence, but when the stall angle was reached This analysis was published in a NACA
the resulting large increase in restoring Technical Note entitled Effect of Steady Roll-
moment pushed the model back in the other ing on Longitudinal and Directional Stability
direction. At this point in the trajectory, the (ref. 8.2). This type of instability had never
model was well below the Mach number been encountered on full-scale airplanes, and
for any serious transonic effects. Knowing indeed, it would not occur on the full-scale
the heavy weight of the model and its XS-1. The report predicted, however, that
observed rolling motion, I was led to analyze some of the new fighter airplanes then on the
the problem as some kind of gyroscopic drawing board could encounter this problem.
effect. In making the analysis, I used a set of In the case of the fighter airplanes, the condi-
equations called the Euler dynamical equa- tions favorable to instability involved long
tions, which had been taught at MIT by fuselages, short wings, and high values of
Professor Rauscher as well as in the graduate longitudinal stability together with low val-
course Introduction to Theoretical Physics. ues of directional stability, both of which
These equations are nonlinear, but because occur at low supersanic speeds. Later, the
the methods of solution that I had learned for report received quite a lot of attention when
the airplane stability equations dealt with lin- the problem was a:tually encountered on
ear equations, I attempted to linearize the airplanes like the _-3 and the F-100. The

72 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Roll Coupling

FIGURE8.4. Devices made

to illustrate the effect on
stability during rolling .375 in. dowel rod---_ z_[ .016 in. diameter music wire
motion of the equality or
inequality of stiffness
about the pitch and yaw

,//_.025 in. thick celluloid

i iI

existence of my theoretical analysis showed revealed a new type of instability that proved
the design changes required to correct the important in the design of future airplanes.
problem and may have saved some pilot's This experience shows that the important
lives by allowing quick action to correct the research results of a test are often quite dif-
designs. ferent from those intended when the tests are
Later, the phenomenon was called "roll cou-
pling" or "inertial coupling," and many tech- I have received more recognition from the
nical reports were written presenting more report on roll coupling than from any other
detailed analysis of this problem. The more work done in my career at Langley. I remem-
complete analyses took into account the non- ber giving a talk on the results at one of the
linear nature of the equations and required Langley Department Meetings, seminars that
solutions on analog or digital computers. As were held after working hours to allow dif-
a result, the solutions referred to individual ferent divisions to present their latest
cases. I was not sufficiently familiar with research results. I used some crude devices
these techniques at the time, and the neces- constructed of dowel rods coupled together
sary computing equipment was not then with either a piece of wire or a flat sheet of
available at Langley. My approach was to celluloid to illustrate the importance of the
use the methods that I had learned for the difference in stability in the two planes.
classical airplane stability equations, which These devices are sketched in figure 8.4.
were linear. This method revealed the basic When the rods were spun between the hands,
physical principles that caused the instabil- the rod connected with wire would remain
ity, a result that could not have been obtained stable, whereas the rod connected with the
by study of more exact solutions for individ- celluloid would fly out to a large deflection.
ual cases. I could see that my uncertainty The talk received a round of applause, some-
about the sign of the coupling terms could thing quite unusual in these meetings.
have been avoided if I had started with the
This work is also the only research that I
complete equations and then linearized them.
conducted that received recognition in the
When others had been able to analyze the
American Institute of Aeronautics and Astro-
complete equations, I felt that my knowledge
nautics magazine, Astronautics and Aero-
of this subject was somewhat deficient and
nautics, now called Aerospace America,
should have been treated more thoroughly in
which contains a monthly column entitled
the courses at MIT. I realize now, however,
"Out of the Past," presenting historical items
that the simpler approach gave a more basic
occurring 25 and 50 years previously.
understanding of the problem.
Twenty-five years after my report appeared,
The XS-I drop test showed that the elevator the magazine in June 1973 presented the fol-
control was adequate, but in addition lowing item:

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 73

A Method for Determining the Moments and Products of Inerzia of Full-Scale Airplanes

NASA research scientist William H. closely controlled, _aad in many cases the
Phillips of the Langley Memorial final drawings were not available until long
Aeronautics Laboratory Flight after the airplane was in production. Even
Research Division publishes Techni- today, in airplanes that are fitted with special
cal Note TN-1627 which contains a research equipment or airplanes produced by
theoretical prediction of the problem small manufacturers, weight and inertia
of inertial coupling. The phenomenon characteristics may not be accurately known.
later plagues aircraft having long fuse- For accurate correla:ion of flight data with
lages and short wings, where the mass theory, therefore, measurements of these
of the load is spread along the fuselage characteristics are often required.
with little spanwise distribution of
Determining weight and center-of-gravity
load. In particular, it forces a slow-
location might appea_- to be a simple physical
down in the operational introduction of
measurement if suitable scales are available.
the North American F-100A Super
Scales for weighing airplanes had been built
Sabre, and first manifests itself in
into the hangar floor, but in years after about
flights of the Douglas X-3 research
1950, accurate strai_ gauge load cells were
plane. Increasing the area of the verti-
usually more convenient to use. Despite the
cal tail and the span of the wings
basic nature of the measurement, new engi-
solves it.
neers, when assigned this task, often made
NACA Research Memorandum errors. As a result, a single engineer, William
H55AI3 by the NACA High-Speed C. Gracey, was assigned the job of oversee-
Flight Station, 4 Feb. 1955: NASA, ing the weighing of each new airplane as it
Aeronautics & Astronautics 1915- was received. Grac_:y had also written a
1960, pp. 60--61. report on measurement of the moments of
inertia of airplanes lay an oscillation tech-
nique. This method was occasionally used,
but to get moment of inertia in pitch, it was
A Method for usually more convenient to mount the air-
Determining the plane on jacks and oscillate it while it was
Moments and Products restrained by springs of known stiffness
attached under the ta 1.
of lnertia of Full-Scale
These methods still "ailed to supply data on
the tilt of the princigal longitudinal axis of
To compare flight measurements of airplane inertia. A report by Sternfield had shown that
motion with theoretical predictions, it is nec- the tilt of this axis was of critical importance
essary to know the weight, center-of-gravity in determining the damping of the lateral
location, and the moments and products of oscillations (ref. 8.3). At the Wright Air
inertia. In the case of lateral motion, the Development Center at Wright-Patterson
effect of product of inertia is more frequently Field, a proposal w_ s once under consider-
expressed in terms of the tilt of the principal ation to build a veq large facility that was
longitudinal axis of inertia from the refer- capable of holding h rge airplanes and oscil-
ence axis. At the present time, airplanes, at lating them to deternfine their inertia charac-
least those built by large airplane companies, teristics. So far as I know, this facility was
have known weight and inertia characteris- never built. I devised a technique, however,
tics because the companies have a weight that was at least capable of making accurate
control section that keeps up-to-date com- measurements of th,: moment of inertia in
puter records of the weight and exact loca- yaw and the tilt of t le principal axis of air-
tion of every item in the airplane. During planes as large as figaters with very inexpen-
WW II, however, the weights were less sive equipment (ref. &4).

74 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Towed Bodies

FIGURE 8.5. Test airplane

as set up for determining
moment of inertia and
tilt of the principal
longitudinal axis of
inertia. Spring brackets
retouched for clarity.

In this method, the airplane is suspended by

a single cable from a crane located in a han- _ = 1/2tan_l(21Zr tan 8I
gar. Most fighter airplanes are equipped with )
a hoisting sling to allow them to be moved
by cranes or helicopters, so this equipment is
already available. The airplane is restrained E tilt of principal axis from X body
in yaw by a pair of springs attached to brack- reference axis
ets ahead of and behind the suspension cable.
From the period of oscillation in yaw, the _5 spring angle for zero roll
moment of inertia in yaw can be determined.
Izr, IXr moments of inertia about Z and
A photograph of a fighter airplane set up for X body reference axes
these measurements is given in figure 8.5.
I have not seen this formula given in any ref-
The pitch angle of a line through the attach- erence book, though I am sure someone must
ment points of the springs can be varied. have worked it out previously. So many
When the airplane is oscillated in yaw under authors were involved in preparing the brief
restraint of the springs, it usually has a roll- report (ref. 8.4) describing this project that I
ing oscillation, the amplitude of which can did not include my name on it, but I always
be measured by a position measuring device thought that originating this technique and
at the wing tip. For some particular angle of deriving this simple formula were worth-
the spring attachment, however, the roll while contributions.
oscillation goes to zero. This point can be
found by interpolating between two of the
spring settings. The tilt of the principal axis Towed Bodies
is not equal to this angle at which the roll
is zero, but it is related to it by a simple One piece of apparatus required for most
formula flight tests was the trailing bomb or towed

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 75

Towed Bodies

airspeed head. This device was a heavily problem of transmission of waves down a
weighted body containing a static pressure string or cable, however, is a classical prob-
orifice that measured the static pressure suffi- lem in acoustics that was treated in the book,
ciently far from the airplane to avoid errors Theory of Sound by Lord Rayleigh (ref. 8.6).
due to the flow field of the airplane. The I studied Lord Rayleigh's analysis that
pressure was transmitted to the airplane showed how to divide the problem into ordi-
through a tube attached to the towline. This nary differential equations dealing with the
device was used to calibrate the airspeed time and position dependence of elements of
installation in the airplane, which usually the cable. My analysis showed that oscilla-
was in error by several miles per hour tions originating from unsteady flow in the
because of the influence of the flow about the region of attachment of the cable to the air-
airplane on the pitot-static head mounted on plane would be amplified exponentially as
the airplane. These towed airspeed heads had they traveled down the cable if the airspeed
a frequent tendency to become unstable or to was greater than the speed of wave propaga-
break loose and fall to the ground where they tion. The speed of wave propagation, in turn,
could conceivably cause damage or injury. I varied as the square root of the tension in the
made a study of the stability of these air- cable. With a weigh__ed body on the end of
speed heads with the same methods that I the cable, the tension did not increase very
had been taught at MIT to study the stability much with airspeed, and as a result, beyond
of airplanes. I also studied the previous liter- some value of airspeed, the oscillations
ature on the subject. A previous study by became unstable. I wrote a second report,
Herman Glauert, the famous British aerody-
which was published in 1949, treating the
namicist, had covered the longitudinal oscil- oscillations of a towed cable (ref. 8.7). After
lations. I treated the lateral oscillations. My
WW II, when much of the German wartime
analysis showed that the least stable region
research was made available, I discovered
was with the body close to the airplane,
that an almost identical report had been writ-
whereas Giauert showed the instability with
ten in Germany (ref. 8.8). These reports
the body far from the airplane. I discovered
revealed the true cause of the loss of the trail-
an error in Glauert's analysis in which he had
ing bombs. As a result, another method of
reversed the stable and unstable regions. In
static pressure measurement, called the trail-
view of his reputation, however, I did not
ing cone method, was developed and has
mention this discrepancy in my report, which
since been widely used. The drag of a cone
was published in 1944 (ref. 8.5). The Navy
on the end of the cable increases as the
was also interested in towed bodies for carry-
square of the airspe,_d, and therefore keeps
ing magnetometers to detect submarines. As
the cable stable as the airspeed is increased.
a result of their interest, I made tests of a
The static holes used to measure the air pres-
large towed body with a movie camera in its
nose to study its motion. Through the course sure are placed in the tube attached to the
of my career, I always kept an interest in towed cable, sufficiently far ahead of the
cone to avoid interference from this source.
towed bodies and kept a bibliography of
reports published on this subject.
I never received mltch recognition for my
Despite my analysis, towed bodies continued knowledge of towed bodies because so few
to break loose and get lost, and I finally real- people ever work in this area. Some discus-
ized that the problem was instability of sions were held wit a companies concerned
waves transmitted down the cable rather than with in-flight refueli:lg because instability of
instability of the body itself. Studying this the refueling hose was a matter of concern. I
type of instability got into the realm of par- was also selected as an anonymous reviewer
tial differential equations, a subject with of an AIAA paper dealing with the stability
which I never had much experience. The of towed cables.

76 Monographs in Aerospace ttistory Number 12 Journey in Aeronautical Research

CHAPTER 9 Unsteady Lift

By unsteady lift is meant the lift on a wing or calculate the gust magnitude producing the
airfoil on which the angle of attack is chang- response. A correction factor was used based
ing as contrasted to the lift under steady on unsteady lift theory of a two-dimensional
conditions, which exist in steady fight or in airfoil, which I considered to have question-
most static wind-tunnel tests. The primary able validity for a finite-span wing. As a
application of unsteady lift theory is to wing result of these problems, from time to time I
flutter, inasmuch as on airplanes, even in looked up references on unsteady lift theory
maneuvers, the angle of attack changes and became aware of the studies that had
slowly enough that the lift is nearly the same been made in this field.
as in steady flight. Other applications, how-
ever, arc frequently encountered, such as in At Langley, the recognized expert in the field
calculating the response of an airplane to was Dr. Theodore Theodorsen, head of the
gusts and in analyzing the flight of birds and Physical Research Division, who had written
a notable report entitled General Theory of
Aerodynamic Instability and the Mechanism
When 1 attended MIT, unsteady lift theory
of Flutter (ref 9.1). This report, published in
was considered a subject for specialists with
a knowledge of advanced mathematics and 1935, for the first time put the subject of flut-

was not taught in the regular courses. At ter analysis on a rigorous mathematical
basis and gave practical solutions to flutter
Langley, a similar situation existed. A few
problems. Incidents of flutter had caused air-
specialists in the Physical Research Division
had worked in this field and understood the plane crashes since the earliest days of avia-

analyses that had been made, but the aver- tion. For example, Lincoln Beachy, a dare-

age engineer in the wind tunnels or Flight devil stunt flyer who in 1913flew a strongly
braced biplane called the Lir Looper in air
Research Division had no knowledge of the
subject. shows around the country, was in a high-
speed dive in a show at San Diego when the
In my work at the Flight Research Division, 1 wings and ailerons fluttered, which caused
soon encountered problems involving the airplane to disintegrate and crash. Dur-
unsteady lift theory. For example, the all- ing WW 11, Anthony Fokker designed the
movable tail on the XP-42 airplane could not Fokker D-VIII, one of the first monoplanes
be flown without some kind of flutter check. with a thick unbraced cantilever wing. The
Spring tabs were known to be subject to flut- airplane had flown successfully, but a change
ter. In gust research, the acceleration on an was made that added weight and stiffness to
airplane encountering a gust was used to the rear spar. This change evidently reduced

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 77

Spring-Tab Flutter

the flutter speed, which resulted in the death The report by Walket which had been called
of several pilots. Since flutter usually occurs the greatest doctoral thesis ever written in
at high values of airspeed and involves the the field of aeronautics, was fascinating to
instability of a high-frequency mode of read and, through u.,'e of flow visualization,
oscillation, the wing extracts tremendous gave a clearer physical picture than any of
amounts of energy from the air stream that the mathematical reports. A point of interest
results in a practically explosive increase in is that Wagner and Walker studied the
amplitude and disintegration of the structure. response of the lift to a step change in angle
of attack, whereas Theodorsen and Glauert
Theodorsen's report made a very rigorous considered sinusoidal variations in angle of
attack. Mathematicians knew that the two
analysis of flutter theory. From the stand-
point of a beginner, however, this was a poor approaches should give equivalent results
report to study. Theodorsen outlined the for the linear systems considered, provided
mathematics in very elegant form, with many the analyses were correct. Later Garrick,
steps omitted, and gave no references to pre- who worked for Thecdorsen, put out a report
vious work. In the library, I found that only verifying that the two sets of results were in
about four reports had been written on the agreement. At that time, neither 1 nor most
subject of unsteady lift prior to Theodorsen's other engineers knew that the results were
report and that someone who completely equivalent, a fact that further confused the
understood these reports would know about complicated field of unsteady lift.
as much on this subject as any expert in
My own efforts in the field of flutter and
the field. (I have in mind the reports by
unsteady lift were primarily concerned with
Birnbaum (ref 9.2), Wagner (ref 9.3),
two projects, a devi :e to reduce the mass-
Glauert (ref 9.4), and Walker (ref 9.5)).
balance weight requ red to prevent flutter of
Unfortunately, 1 did not have the time or
spring tabs and an attempted clarification of
mathematical background to study these
the role of unsteady lift in explaining snaking
reports in detail, but I did find that the analy-
oscillations. The following sections will dis-
sis by Glauert, the British aerodynamicist,
cuss these projects.
published in 1929 was much easier to under-
stand than the report by Theodorsen. Why,
one may ask, was the work by Theodorsen so
much more widely acclaimed than that of spring-Tab Flu tter
Glauert? Theodorsen's report did, of course,
include the analysis of flutter as well as the The spring-tab problem arose because, as
theory of unsteady lift. Also, it arrived at the shown by Theodorsen, a reliable way to
much desired closed-form solution. That is, avoid control surface flutter was to mass bal-
the results were obtained in terms of Bessel ance the surface about its hinge line. If this
functions. (As shown later by L E. Garrick, rule was applied to a spring-tab system, both
they could also be put in terms of Hankel the tab and control surface would have to
functions.) Glauert, on the other hand, eval- be mass balanced. Then additional mass-
uated the corresponding functions by numer- balance weight woul t have to be used on the
ical integration. From the engineer's stand- control surface to 9alance the tab mass-
point, Glauert's results are just as useful, but balance weight. Thi; weight penalty seems
from the standpoint of the mathematicians bad enough, but so_e British reports had
who formed the main body of researchers in shown that the tab mass balance had to be
the field of unsteady lift, the closed-form located very close to the tab hinge line to be
solution was much more desirable. I did not effective, which mmns that this balance
fully agree with this reasoning, because weight had to be ccnsiderably greater than
Bessel functions themselves had been origi- the weight of the tab itself. The reason for
nally evaluated by numerical methods. this conclusion is shown in figure 9.1. When

78 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Spring-Tab Flutter

FIGURE 9.1. Diagram

stick (fixed)
showing optimal location
of mass-balance weight
Spring 7 /--- la_feoriS_Cetintnr_ifnte_b
to prevent flutter of
spring tab.
_ Control / /
"'x / / /_ Optimal location of

______-__- _- / ____tab balance weight

Flap j
• _e Tab j "" "

FIGURE 9.2. Schematic

diagram of installation of
tab balance weight ahead
of flap hinge line.


Flap hinge line Tab

-]_ Balance weight

the control surface is moved with the control way between these two points. The result of
stick held fixed, the tab ordinarily moves in the heavy tab balance weight and the added
the same direction as the control surface, but weight of the main control surface balance to
through a greater deflection. This movement balance the tab balance weight was an unde-
results from the kinematics of the tab link- sirably large weight penalty. I devised a link-
age, which are determined by the desire to age arrangement in which the tab balance
have the control follow a quick stick deflec- weight could be located ahead of the main
tion with very little overshoot. For preven- surface hinge line, where it did not add to the
tion of flutter, an angular acceleration of the weight behind the hinge line and yet would
control surface should cause the tab to jump cause the tab to jump ahead of its static posi-
ahead of the control surface. A little thought tion in a sudden control movement. A draw-
will show that the tab must jump ahead of the ing of the linkage is shown in figure 9.2. A
position where it would be under static con- memorandum was written for the Engineer
ditions. Referring to figure 9.1, it may be in Charge, Dr. H. J. E. Reid, proposing this
seen that a balance weight located at the device. As a result, a discussion was held
intersection of the tab and airfoil centedines with Dr. Theodorsen, who was familiar with
would do no good because it would not the problems of tab flutter, but who did not
move. One located at the tab hinge line appear to know of the recent British reports
would be ineffective because it would exert on the subject. He felt that the device would
no moment about the tab hinge. The best prevent low-frequency flutter, but might be
location for the mass-balance weight is half ineffective in preventing flutter of some

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 79

Spring-Tab Flutter

FIGURE 9.3. Drawing of

balance weight
installation in B-34
vertical taft.

high-frequency modes involving flexibility the balance weight moved oppositely from
of the structure or the linkage. He proposed its motion at low frequencies. Most of the
building an installation of the system in an flexibility was in tht_"thin-skinned structure
actual control surface to allow measurement of the tail, whereas he linkages themselves
of the high-frequency modes. were quite rigid. As a result of these tests, it
was concluded that the device might be sub-
With the endorsement of Dr. Theodorsen,
ject to high-frequency flutter, and no further
therefore, a vertical tail was obtained from
studies were made.
an Air Force B-34 bomber (similar to the
Lockheed Hudson) and the system was
I have heard recentl2y that the elevator on the
installed (figs. 9.3 and 9.4). This tail surface
Curtiss C-46 airplane did have a device simi-
was suitable for installation in a wind tunnel
lar to the one I tested to move the tab balance
for testing as part of an ongoing series of
weight ahead of tht control surface hinge
flutter tests.
line, though most actual spring-tab installa-
When Theodorsen was invited to the hanger tions on service ai_lanes did not have bal-
to see the installation, he hauled back with ance weights on the tab, and the main provi-
his fist and hit the tail a solid blow, which sion to avoid flutter was to make the tab as
caused it to shudder and shake. His comment light as possible. These installations were
was, "It will tooter!" Later vibration tests tested in flight at gra, lually increasing speeds
were made, which showed that there were to verify that they d_d not flutter within the
resonant modes at 14 and 25 hertz in which desired flight envelope. It therefore appears

80 Monographs in Aerospace History Number 12 Journey in Aeronautical Research

Effect of Unsteady Lift on Snaking Oscillations

FIGURE 9.4. Photograph

of balance weight
installation in B-34
vertical tail.

that the British analyses requiring the use the the force in phase with the velocity of the
balance weights for absolute prevention of motion is much larger than the force in phase
flutter were unduly conservative. Also, as with the displacement. If the surface were
mentioned previously, the need for spring- used as a vertical tail, for example, this the-
tab controls rapidly disappeared with the ory would indicate that in an oscillation of
introduction of hydraulic powered controls. low frequency, the tall itself might tend to
make the oscillation unstable.

The prevalence of snaking oscillations on

Effect of Unsteady Lift many airplanes, as mentioned previously,
made it tempting to look for some basic
on Snaking explanation of the oscillation based on aero-
Oscillations dynamic theory. Reports by Smilg (ref. 9.6),
a flutter expert at Wright-Patterson Field, and
A peculiarity of the two-dimensional by Pinsker (ref. 9.7), a well-known aero-
unsteady lift theory of Theodorsen and other dynamicist at the British RAE, made it clear
investigators is that the curve showing phase that serious thought was being given to this
angle of the lift as a function of frequency in explanation. Personally, I was well satisfied
an oscillation of angle of attack initially has with the explanation involving control sys-
an infinite negative slope. This result means tem friction that has been discussed previ-
that at very low frequencies of oscillation, ously, and I felt that it was incorrect to apply

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 81

Effect of Unsteady Lift on Snaking Oscillations

two-dimensional theory (that is, a theory that graphical integration. As a result, I did not
applies to a wing of infinite span) to the case realize my hope of obtaining a closed-form
of a low aspect ratio vertical tail. Proving this mathematical solution.
belief, however, was a difficult problem.
In 1952, the results were put in the form of a
report in which my analysis and the work of
A remarkable report had been published in
various investigators in the field of unsteady
1940 by Robert T. Jones on The Unsteady
lift of finite wings were compared, and the
Lift of a Wing of Finite Aspect Ratio effect of these results on lateral oscillations
(ref. 9.8). Jones had avoided the complex
of airplanes was demonstrated. The report
mathematics of previous investigators by
was reviewed by an editorial committee con-
considering the response to a step change in
sisting mainly of members of the Physical
angle of attack, then converting this result to
Research Division, including mathemati-
the response to sinusoidal oscillations. To cians who had devoted much of their careers
make this conversion in a simple way, Jones
to studying unsteady lift effects. These math-
fit the variation of lift following a step input
ematically inclined individuals were much
with an exponential curve. I was not sure more concerned with the mathematical
what effect this approximation would have at
development than with the application to air-
low frequencies; therefore, I attempted to
plane oscillations. They concluded that the
make an analysis similar to that of Jones, but
theory really did not contribute any new
working from the start with a sinusoidal
results in unsteady lift theory, and therefore
they recommended that the report should not
be published. I was disappointed that the
Today I would have less concern about the
paper was not published, mainly because I
accuracy of Jones' results, but at that time I
felt that the two-dimensional unsteady lift
was less familiar with the transformation
theory had been frt:quently applied incor-
between step and sinusoidai inputs. I hoped
rectly in the past. Also, I felt that my report
to be able to make an analysis in which the
would make the subject more understandable
mathematics, at least, was exact, thereby
to the average engineer.
obtaining a direct basis of comparison with
Theodorsen's two-dimensional results. An important point in the presentation of my
results was the method of expressing reduced
The basis of the analysis in Jones' theory frequency. In previous reports on unsteady
was to convert the vortex system of an infi- lift theory, the frequcncy had been expressed
nite wing, which consisted of a bound vortex in nondimensional fcrm by use of the param-
and a starting vortex, to that of a finite wing eter toc/V, where co is the frequency of the
by superimposing a pair of horseshoe vorti- oscillation in radians per second, c is the
ces that may be called "canceling vortices." wing chord, and V is the airspeed. I found
These vortices cancel the parts of the vortex that at low frequencies, the parameter oob/V,
system beyond the wing tips and add the where b is the wing span, was much more
trailing vortices of the finite wing. I used the suitable. This param ".ter would be meaning-
same technique, but allowed all these vorti- less in the two-dime_tsional case because the
ces to vary sinusoidally with time and posi- span b would be int nite. In the case of the
tion. The effects of each element of each vor- finite wing, however the results showing the
tex on the downwash on the wing was phase lag of the lift was nearly independent
obtained by integration. The resulting inte- of aspect ratio whet the reduced frequency
grals were evaluated with the help of Keith based on span was :lsed. This result shows
Harder, a mathematician in the Physical that at low frequencies, the trailing vortices,
Research Division. All the integrals could be which disappear ir the two-dimensional
found in tables of integrals except one term case, remain import_nt and, at low frequen-
of one integral. This term was evaluated by cies, are mainly re,,ponsible for the phase

82 Monographs in Aerospace ttistory Number 12--Journey in Aeronautical Research

Effect of Unsteady Lift on Snaking Oscillations

FIGURE 9.5. Amplitude

and phase angle of the
circulatory lift of
elliptical wings resulting
from a sinusoidal change
Aspect ratio
in angle of attack,
represented as the vector 2O

CL_, as functions 10
of ob/V,, where co is the 4
frequency in radians per
second, b is the 3
wingspan, and V is the 0



0 I I I I I I I I


Aspect ratio

-20 I I I I I I I I
0 2 4 6 8
Reduced frequency (ob/V

variations of the lift. The results indicate that on the three-dimensional theory is almost the
the slope of the phase angle curve remains same as that based on conventional stability
finite as the frequency approaches zero, theory, called the quasi-static case, which
though my analysis, because of the use of considers the angle-of-attack change at the
graphical integration, did not demonstrate tail due to yawing velocity, but then neglects
this point rigorously. unsteady lift considerations altogether. The
damping based on two-dimensional theory,
To illustrate these points, figure 9.5 shows
on the other hand, is considerably less over
the variation of the amplitude and phase
the whole range of frequencies shown.
angle of the circulatory lift with frequency
based on wing span. Note that the two- One beneficial result of this project was that
dimensional case can not even be shown I obtained a working knowledge of unsteady
when the frequency is based on wing span. lift theory. I prepared a talk on this subject
The damping of oscillations based on two- that minimized the mathematical aspects.
dimensional and three-dimensional theories This talk was presented on several occasions
are shown in figure 9.6. The damping based to engineers working in my division. Since

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 83

Effect of Unsteady Lift on Snaking Oscillations

FIGURE 9.6. Damping of

100 oscillations of a
weathervane computed
by finite-span, quasi-
70 static, and two-
dimensional theories.
N1/2 , the number of
cycles to damp to half
amplitude, is plotted as a
function of co_._b,
where co
30 I
is frequency in radians
per second, b is span of
x weather vane, and v is
x -I
velocity. Frequency
20 I

varied by changing the



10 _N_'N
"' ",, /-- Two-dimensional

'\ Finite-•,,

,,,\\ span /
, ,
x \

Quasi-static _ _, \ _x

.01 .02 .03 .04 .05 .07 .1 .2

this work was completed, the use of high- though some joint p'ograms have been con-
speed computers has greatly extended the ducted in recent ye_ s to combine the efforts
work on unsteady lift, which allows more of control theorists, structural dynamicists,
accurate calculations for wings with various and unsteady lift specialists to allow applica-
planforms and mode shapes at speeds tions of control theory to predict flutter of a
throughout the Mach number range. This flexible structure aJld to design automatic
field still remains a subject for specialists, flutter damping syst¢ ms.

84 Monographs in Aerospace I listory Number 12--Journey in Aeronautical Research

CHAPTER 10 An Automatic Aileron Trim
Device for Personal Airplanes

This subject is accorded the recognition of a give demonstrations of all the stability char-
separate chapter because of the depth of my acteristics discussed in his courses. Shortly
personal interest in the subject and because after I left MIT, Professor Koppen's sixteen-
of the long period of time over which the year-old daughter and her instructor, who
problem of spiral instability had repeatedly was giving her flying lessons, were both
come up for consideration. In a previous sec- killed when their small airplane crashed
tion of this paper, under the heading "Spiral after flying into a fog bank. Professor
Instability," the tendency of airplanes to Koppen, of course, was devastated and was
diverge from a straight path into a spiral unable to teach for two years after that.
dive was noted. This divergence was said to
be so slow that it is rarely noticed by the
pilot when he has visual reference to the
Development and Tests
horizon. If the airplane isflying under instru-
ment conditions, a pilot who is proficient in of the Trim Device
instrument flight likewise has no difficulty
I knew that the problem of spiral instability
with this type of instabilit3: Many crashes
have occurred, however, when novice pilots could be cured with a conventional autopilot,
flew into clouds or fog banks and were but the existing autopilots were too expen-
unable to keep the airplane flying level. This sive and heavy to be used in light airplanes.
Much of the complication and expense of
problem becomes more severe in light air-
conventional autopilots resulted from the
planes because they are usually equipped
need to operate the controls in proportion to
just with the basic instrumentation (turn and
various signals sensed by instruments in
bank indicator and airspeed) which require
the airplane. This process, in those days,
considerably more skill to fly blind than an
required the use of vacuum-tube amplifiers
airplane equipped with an artificial horizon.
and complex servomechanisms for convert-
ing small electrical signals to mechanical
1 was profoundly concerned with this prob-
outputs. I therefore gave much thought to a
lem because of the experience of one of my
simpler way to perform this function.
professors at M1T. The phenomenon of spiral
instability and its associated dangers had I had been interested in bang-bang controls
been discussed by Professor Otto Koppen in (also called discontinuous controls) in which
his courses on airplane stability and control. the signal to operate the control was
He was a pilot himself and he took me and switched either full on or full off. With this
all his other students up in a Fairchild 24 to type of mechanism, an on-off contact

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 85

Development and Tests of the Trim Device

replaces the complex electronic amplifier. airplane turning firs_ one way and then the
Such systems can be analyzed by the phase other.
plane method that was described in the book
Several methods of providing this anticipa-
by Minorsky on nonlinear systems (ref. 5.5).
tion (called a lead signal) were tried, all of
Preliminary studies of wing-leveler controls
which were more ccmplicated than desired.
were made by using this technique.
Finally, the idea was conceived of simply
tilting the spin axis of the gyroscope so that
At that time, about 1955, the electronic ana-
it would sense components of both rolling
log computer had recently become available.
velocity and yawing velocity. Since the roll-
This computer greatly aided the study of
ing velocity as the airplane rolls into a turn
nonlinear systems. One of the engineers in
leads the resulting yawing velocity, this
my branch, Helmut A. Kuenel, set up the
method provided the desired lead signal.
problem on an analog computer to study var-
ious ideas in more detail than could be done This method was so much simpler than any
of the other techniques that it was immedi-
using the phase plane method.
ately adopted.

As mentioned previously, most airplanes, A suitable gyroscopic device was provided

whether or not they are inherently spirally by H. Douglas Garner of the Instrument
stable, will diverge from straight flight Research Division. This device, taken from
because the friction in the control system is
other equipment, already incorporated a pair
large enough to hold the controls deflected of contacts on the gyroscope gimbal to
from the trim position. To try to overcome provide the desired operation of the trim
this problem, a previous flight study had motor. In addition, it had a means of electri-
been made in which preloaded centering cally torquing the gimbal to bias the rate of
springs could be connected to the aileron turn at which the contacts were centered.
control linkage to overcome the friction in Mr. Garner later provided valuable ideas on
the system. The spring device could be use of this feature in the system. The gyro-
adjusted to the proper trim position by the scope was mounted on a tilting platform and
pilot. This device prevented the airplane the system was installed in a Cessna 190 air-
from diverging when it was properly plane available at the NACA Flight Research
trimmed, but it could not take care of subse- Division.
quent changes in trim caused by asymmetric
fuel usage or by changes in power or air- The development of the system and results of
speed. Evidently, a device was needed to flight tests are reported in detail in a NACA
automatically move the centering spring unit report (ref. 10.1). A diagram of the system
to the correct trim position. taken from this report is given in figure 10.1.
The system provided excellent stabilization
A gyroscope was needed to sense yawing of the airplane in smooth or turbulent air. It
velocity (turning) of the airplane and move worked even better than expected as a result
the centering spring unit as required to can- of two fortuitous fi:atures. First, with the
cel this motion. A bang-bang control, in gyroscope set to the ,'orrect tilt angle, the air-
which a slow-moving electric motor con- plane, when released from a banked attitude,
nected to the spring unit would be switched returned to level flight in a practically dead-
to run in one direction or the other, appeared beat manner, with bttle or no overshoot. It
ideal for this purpose. In analytical and ana- would be expected that a different gyroscope
log computer studies, however, the device tilt angle would be required for each value of
was shown to need some anticipation of the airspeed, but it was f, mnd (and shown analyt-
return of the springs to the trim position, oth- ically) that the varial ion of angle of attack of
erwise the device would overshoot the trim the airplane as it flew at different values of
position and cause a slow oscillation with the airspeed was in th_ correct direction, and

86 Monographs in Aerospace tgistory Number 12--Journey in Aeronautical Research

Development and Tests of the Trim Device

FIGURE 10.1. Schematic

Force-sensitive switches
diagram of automatic (about 2 ° free play is provided Force-switch preloaded
aileron trim control between shaft and control
centering spring
device. wheel; wheel is spring preloaded

to the center of the free-play zone) Battery connection

Nose of airplane


actuator unit
Airplane -axis

Rate gyro Aileron preloaded

(tilted 35 ° from the centering springs

Jack screw

Magnetic restraint coil Indicator box
lights -

Sensitive axis J _

Precession axis j (_ Control box

(contains relays to switch
power from gyro contacts
Electrical contacts to actuator motor)

:tuator switch

Spin axis -_

Left aileron

approximately of the right magnitude, to pro- tem then behaved as a "dual mode control"
vide automatically the required variation with a discontinuous action at large displace-
with airspeed of the tilt angle from the flight ments and a linear action at small displace-
path. Second, any bang-bang, or discontinu- ments. This type of system had been advo-
ous, control would be expected to have a cated by some control theorists to take
small hunting oscillation about the trim posi- advantage of the best features of both types
tion. In the actual installation, however, the of control, but it had probably never been
engine vibration caused the gyroscope con- implemented in such a simple manner.
tacts to chatter as the airplane reached the
level attitude. As a result, the action of the A problem observed by the pilots was that if
control approximated a linear control and the the airplane was placed in a steady turn and
hunting oscillation was eliminated. This sys- then manually returned to level flight on a

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 87

Development and Tests of the Trim Device

desired heading, the control system, operat- I did not worry about it because I had no
ing more slowly than the human pilot, would interest in making money from the invention.
leave the ailerons deflected and would then
Later, when the report became known among
continue to stabilize the airplane on a head-
light airplane companies and control system
ing a few degrees displaced from that desired
manufacturers, about eight of these compa-
by the pilot. To avoid this problem,
nies sent representatives to Langley for fur-
H. Douglas Garner suggested that a pair of
ther discussion of the device. The first ques-
force switches be placed on the control
tion asked usually was, "Is the device
wheel between the pilot's grip and the wheel
covered by a valid patent?" Without a patent
itself, which would apply current to the
to give an exclusive right to produce the
torquing coils on the gyroscope gimbal. With
device, none of the manufacturers would
this system, as the human pilot held the air-
consider marketing it. As a result, I learned
plane on the desired heading, the airplane
the value of a patent even in cases when the
would be retrimmed on this heading by the
inventor was not interested in making
automatic system.
money. Several light 9lane autopilots were in
production at that time and continue to be
The force switches also allowed the pilot to
available today, but because they are expen-
make constant-rate turns of about three
sive and require considerable rebuilding of
degrees per second for instrument flight
the airplane for installation, they are rarely
maneuvers by holding a small force on the
used. These problems would have been
control wheel within the preload of the cen-
avoided by the system tested. Mr. Garner has
tering springs. The system therefore pro-
continued research in this field and has pro-
vided features of a more sophisticated auto-
duced designs for aulopilots suitable for con-
pilot in addition to the original objective of
struction and installetion by home builders.
acting simply as a wing leveler.
These devices further extend the capabilities
of the device tested by providing a heading
Because of the preload in the centering hold mode. These devices are highly suc-
springs, the human pilot had to apply a con-
cessful, but unfortunately many light air-
trol force exceeding this amount (about four
planes still are not equipped with autopilots
pounds on the control wheel) to maneuver or and crashes in ins_lment flight conditions
to hold a steady turn. The NACA test pilots continue to occur.
objected to this feature because of their expe-
rience with airplanes that did not oppose the The wing-leveler aut, _pilot described is nota-
pilot in holding a banked attitude. Beginners ble because, to my knowledge, it is one of
with little piloting experience, on the other the few successful applications of a discon-
hand, generally preferred the feeling that the tinuous control system for airplane control.
airplane would return to level flight if the Discontinuous controls have been widely
controls were released. investigated by control theorists. This type of
control is theoretical y more efficient in cor-
A patent disclosure for the system was filed. recting for disturban:es because it uses the
At that time, all the NACA patent applica- full power of the ;ervomechanism at all
tions were handled by one man at the NACA times. A control sys:em that uses its maxi-
Headquarters in Washington. About a year mum effort in correcting for a disturbance
later, the formal NACA report (ref. 10.1) was that is followed by maximum effort in the
published. I did not know at the time that a other direction to stop the motion at the
formal report placed the invention in the desired point has been called an optimal
public domain and that the patent application bang-bang control, t_ecanse it can bring the
was thereby voided. I was displeased that a system to the desirer point in a shorter time
year had passed without action by the Head- than any other system (for example, a linear
quarters office to file a patent application, but system) with the s trne maximum control

88 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Development and Tests of the Trim Device

effort. Despite this theoretical advantage, Koppen eventually went back to teaching at
these systems have found little use in aero- MIT, but his wife made him promise that he
nautical applications, though they have would never fly an airplane after that. As
sometimes been used for spacecraft attitude long as she was alive, he kept this promise,
control. In most cases, the tendency to hunt but after her death he took up flying again. A
back and forth in a limit cycle oscillation few years ago, I learned from someone in the
after correcting for the disturbance has been Langley Flying Club that Professor Koppen
a serious drawback. The availability of com- had been observed flying over Langley Field
puters makes the study of such systems on his way to Florida in an American
much easier. In recent years, a group of Yankee, a very small high-performance per-
researchers from a British university came to sonal plane. At the time, he must have been
Langley to discuss their work in this field. in his 80's. Later, I learned that Professor
They were very surprised to hear that a suc- Koppen was the oldest pilot in the country
cessful application of a discontinuous con- with a full instrument rating and that his
trol had been made forty years ago. plane was equipped with a wing-leveler
autopiiot and special navigational instrumen-
To conclude this account, the subsequent tation, to the extent that there was no room in
career of Professor Otto Koppen might be of the cockpit for anyone but the pilot. He died
interest. After the death of his daughter, in 1993 at the age of 96.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 89

CHAPTER 11 New Areas of Research

Many of the disciplines that I studied in col- Finally, the developments in control theory
lege were useful in my work at Langle3; as that occurred during the war allowed studies
has been shown in the preceding chapters. not only of automatic pilots, but also of the
Some other subjects of interest, however, control actions of human pilots in ways that
were completely new and resulted from the had not been considered previously. Because
rapid progress of aeronautics during the war of the strong dependence of airplane behav-
years. These topics included compressibility ior on the actions of the pilot, this subject

effects, that is, the effect of approach to the attracted a lot of attention and became a

speed of sound on the flow around aircraft. whole new field of research.

The word "compressibility" is derived from

the concept that at speeds attained by air-
craft prior to WW I1, the air flowing over the
The Wing-Flow "
airplane may usually be considered incom-
pressible because the local changes in pres- Technique
sure or density of the air caused by the pas-
sage of the airplane are very small compared The problem of compressibility effects dur-

to the value of these quantities in the ambient ing dive recoveries of high-speed fighter air-

atmosphere. As the airplane flies at a speed planes during WW II has already been men-
tioned. In chapter 8 in the section entitled
approaching the speed of sound, however,
"Roll Coupling," my branch head Robert R.
these quantities experience changes that are
Gilruth was said to have developed two tech-
no longer small compared to the ambient
niques to study compressibility problems,
conditions. As a result, large changes inflow
the free-fall method and the wing-flow
patterns and aerodynamic characteristics
method. This section describes the wing-flow
method and some of the research conducted
by this technique. I was not responsible for
Another development requiring new
the invention of this concept. It was typical
approaches was the necessity of hydrauli-
of Gilruth's approach to research, however,
cally powered controls in the primary control that he did not hesitate to assign capable
systems of airplanes. Such systems, as shown engineers to work on ideas that he consid-
previously, were required because the con- ered important, even though these studies
ventional means of operating the control sur- might be quite different from the traditional
faces were ineffective in the transonic and work of the group. As a result, my section,
supersonic speed ranges. the Stability and Control Section, was

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 91

The Wing-Flow Technique

FIGURE 11.1. Wing-flow

model and test apparatus
mounted on P-51 wing. A
flow-direction vane is
also mounted outboard
of the test model.

o jJ"

assigned the job of developing the wing-flow about one cycle per second to vary either
technique along with members of the Perfor- angle of attack or flap deflection. The forces
mance Section, and I was in charge of super- on the model are continuously recorded with
vising the design of equipment and conduct a strain gauge balance and a recording oscil-
of some of the tests. lograph. The dive lasts about 30 seconds and
in this period the Mach number at the model
In the wing-flow method, a small model is
increases from about 0.7 to 1.2. A drawing of
mounted on the upper surface of the wing of
an unswept semispan wing-flow model of a
a fighter airplane. When the airplane dives
rectangular wing with a full-span flap is
from a high altitude and a pull out is made at
shown in figure 11.2, and some of the data
a high but still safe Mach number, the flow
obtained are shown in figure 11.3. These
on the upper surface of the wing smoothly
results show the ability to study nonlinear
increases from subsonic to supersonic
aerodynamic characteristics by the wing-
speeds. A region of reasonably uniform flow flow method.
exists that may be used as the test region for
a small model. The choking phenomenon
The balances and other equipment for these
occurring in closed-throat tunnels does not
tests were largely designed by Harold I.
exist because of the lack of boundaries above
Johnson, an engineer in the Stability and
and at the sides of the test region.
Control Section. A vacuum-operated auto-
Figure 11.1 shows a photo of a small model mobile windshield-wiper motor was usually
mounted on the wing of a P-51 airplane. A used to oscillate the nodel or the flap. A sin-
special glove was built on the wing to give a gle 30-second dive covered the whole range
more uniform flow region, and the velocity of angle of attack an i Mach number, provid-
field in this region was measured in dives ing as much data as might normally be
before installing the model. As the airplane obtained in a couple 3f weeks of wind-tunnel
goes through its dive and pull out, the model testing. As a result, "he engineers were soon
is oscillated back and forth at a frequency of overwhelmed with data, but they worked

92 Monographs in Aerospace tlistory Number 12--Journey in Aeronautical Research

The Wing-Flow Technique

FIGURE 11.2. Drawing of

semispan model of
rectangular wing with
full-span flap used in
wing-flow tests.

NACA 65-009

4.87 in.


sheet rubber

Circular end

r__._____ Wing surface
t i
i i i
........ i


hard to evaluate it and to put out reports. As In conventional wind-tunnel testing, the air-
might be imagined, there was considerable speed and model configuration are carefully
criticism of the wing-flow method by some set to the desired conditions before taking a
wind-tunnel specialists because of the small reading on the balance. In an effort to obtain
size of the models and the nonuniformity great accuracy, control surface and flap
of the flow field. Nevertheless, the data deflections are sometimes set by building
obtained provided the only source of infor- separate inserts for each value of deflection.
As a result the tunnel has to be shut down
mation at transonic speeds. Some vindication
between tests at each control setting, which
for the development of the method by the
results in a very long time to complete the
engineers in the Flight Research Division
tests. To shorten the test time, the number of
was later received when the wind-tunnel
test points is reduced as much as possible.
engineers installed a transonic bump on the
For example, I have frequently used wind-
floor of the Langley 7-by 10-Foot High- tunnel test data in which readings were taken
Speed Tunnel. The flow field over the bump at increments of five degrees in angle of
was used to test small models by the same attack and sideslip and in which control set-
techniques as used for the wing-flow tings or flap settings were limited to a few
method. values through the deflection range. One

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 93

The Wing-Flow Technique

FIGURE 11.3. Typical data

/ from wing-flow test. Plot
of flap hinge moment as a
osE i i i .......
i ¸¸• i function of angle of
L Gap sealed : attack at various values
of Mach number, gap
.04 l- ..... -_--']7 _Ollc gap ....... _:. ! .... .i. ............... sealed and unsealed.

(a) Mach number from

.65 0 0.65 to 1.15 in steps
of 0.05.

.70 0

.75 0

.80 0

.85 0

.90 0

.95 0
M Ch
1.00 0

1.05 0

1.10 0

1.15 0



94 Monographs in Aerospace ['istory Number 12--Journey in Aeronautical Research

The Wing-Flow Technique

FIGURE 11.3. Concluded.

(b) Mach number from

0.9 to 1.0 in steps of .925 - - ..............
approximately 0.01.
_ Gap sealed

....... .01 lc gap


.925 .923




.956 ,957

.967 .966

.970 .972

.983 .978


great advantage revealed by the methods such features would be missed in conven-
developed for the wing-flow technique was tional wind-tunnel testing. In flight at high
that the data obtained were continuous speed, the variations of angle of attack and
through the range of variables. As a result, sideslip required to reach the limit loads on
small irregularities or nonlinearities in the the airplane structure are frequently only a
aerodynamic forces could be determined few degrees. Any small irregularities or
with as much accuracy as desired, whereas nonlinearities in this range are therefore

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 95

The Wing-Flow Technique

FIGURE 11.4. Wing-flow

model of Vought F7U-1
Cutlass. A haft model is
used, mounted on an
endplate and curved to
the contour of the P-51
wing section.

of extreme importance. The effects of non- Because of the lack of any other source of
linearities in the variation of rudder hinge transonic data, the results of the wing-flow
moment with angle of sideslip in causing tests were eagerly sought by the airplane
snaking oscillations has already been manufacturers during the period following
discussed. WW II. The tests of the 35-degree swept
wing in figure 11.1 _howed that the adverse
Another desirable feature of the wing-flow stability and control characteristics at tran-
technique was that it allowed study of the sonic speeds shown by most fighters with
effect of rate of change of variables as well unswept wings were avoided by this model.
As a result, the Chance Vought Company
as the effect of steady values. In actual flight,
laid out the design of the F7U Cutlass Navy
most flight maneuvers involve changing con-
ditions, but the effects of the rates of change fighter as a tailless airplane with the same
are not studied in conventional wind-tunnel sweep and aspect ra_o as the model. Some
tests. modifications were later made to incorporate
the control surfaces, vertical tails, and fuse-
lage, and a half model of the complete air-
As a result of these considerations, I and oth-
plane configuration was tested (figure 11.4).
ers have made suggestions to wind-tunnel
The completed airplane is shown under test
engineers, from time to time, to incorporate
at the NACA Langk y Flight Research Divi-
provisions in their mounting systems and
sion in figure 11.5. q his is probably the only
balances to allow tests under changing con-
case in history when the design of a new
ditions. Some progress has been made in use
airplane was inspir,;d by data taken in a
of models with motor-driven control sur-
30-second test run.
faces, but the desire to retain existing equip-
ment and traditional techniques has gener- The wing-flow testing continued for some
ally prevented any change in the methods of years, but by 1950 Langley aerodynamicists
mounting models or in the use of steady test developed the slottet_-throat wind tunnel that
conditions. enabled valid test results to be obtained

96 Monographs in Aerospace t!istory Number 12--Journey in Aeronautical Research

Power Controls and Pilot-lnduced Oscillations

FIGURE 11.5. Full-scale

Vought F7U-1 Cutlass.

through the transonic speed range. As a Arrow. This system was developed without
result, testing by the wing-flow and free-fall the aid of the theory of servomechanisms,
methods was discontinued. which did not become available until the
1940's. This inventor did not succeed in con-
vincing the automotive industry to use his
Power Controls and invention until WW II, when it was installed
in many Army trucks. After this demonstra-
tion of its practicality, power steering was
Oscillations soon adopted for automobiles and is now
almost universally used.
With the development of airplanes capable of
transonic and supersonic speeds, the use of Hydraulic controls were also used in large
manual control systems, even with the use of Navy gun turrets. These systems had excel-
the accumulated knowledge of closely aero- lent response and accuracy, but were much
dynamically balanced controls and spring too heavy for aeronautical use.
tabs, became impractical. Designers were
The first hydraulic control systems to be
forced to employ hydraulically actuated con-
installed in airplanes were called power
trois, even though the pilots and old-time
boost controls, which means that a certain
designers were suspicious of the reliability
of these devices. fraction of the force to operate the controls
was fed back to the control stick so that the
Hydraulically actuated controls actually had control forces would reflect the aerodynamic
a long background of experience, but not in hinge moments on the control surfaces just
the aeronautical industry. The first hydrauli- as in a manual control system. The first
cally actuated steering system for automo- installation of power boost controls was on
biles, later called power steering, was the ailerons of a Lockheed P-38. The test
invented by a Harvard professor who lived in pilot who made the initial tests of this system
my home town of Belmont, Massachusetts. said it gave him the feeling of "supreme
He installed the system in a 1923 Pierce domination of the air." Evidently the ability

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 97

Power Controls and Pilot-Induced Oscillations

to produce large values of rolling velocity in There was much interest in finding the cause
high-speed flight with light control forces of the pilot-induced oscillations and in deter-
was very desirable. mining a method to predict such problems on
a new airplane before it was flown. I felt that
Installations of power boost controls on large
the method used in analyzing the stability
airplanes, however, had less immediate suc-
of closed-loop systems known as the
cess. These systems installed in the Douglas
frequency-response method could be used
B- 19, a large four-engine bomber, resulted in
for this purpose. This method, by this time
a violent pitching oscillation in the landing
(about 1950), was available in textbooks.
approach that almost resulted in a crash.
Also, a convenient graphical method called
Similar systems in the Martin Mars, a large
the Nyquist criterion could be used to predict
four-engine flying boat, had many develop-
the presence of stability or instability. Some
ment problems.
new concepts were used in the application of
The first fighter airplane to require power this method, however. First, from handling-
controls was probably the Vought F7U qualities experience_ control forces rather
Cutlass. This airplane, a tailless design, than control displacements were believed to
required very wide chord elevons, which be the quantity of primary interest to the
could not conveniently employ aerodynamic pilot because the control displacements on
balance. To develop the hydraulic control the elevator of a fighter airplane at high
systems for this airplane, the Vought com- speed are very small. Second, the concept of
pany installed power controls in an F4U considering the human pilot as a black box
Corsair. This airplane was later made avail- in a closed-loop system was practically
able to the NACA for research work. unknown at that tiff:e, and reasonable esti-
mates had to be made of a mathematical rep-
This airplane, when flown by NACA test resentation of the pilot's response.
pilots, frequently encountered rather large-
amplitude, unexpected, longitudinal oscilla-
The analysis was made using experimental
tions when entering turns or pull ups. It was
data on the control system by oscillating the
soon realized that these oscillations repre-
control stick and measuring the resulting
sented an instability of the combined system
control forces and elevator deflections. These
consisting of the human pilot, the airplane,
tests were made at various values of fre-
and the control system. These oscillations
quency and amplitude. The normal accelera-
were called pilot-induced oscillations, or
tion of the airplane in response to the eleva-
PIt for short. After some practice, the pilots
tor motion was determined from well-
could fly without encountering these oscilla-
established airplane stability theory, and the
tions, and in fact, found it difficult to repro-
response of the pilot to airplane motion was
duce the condition for test purposes. They
estimated as a simple transfer function in
could, however, detect a tendency for the
which the pilot was assumed to apply a con-
these conditions to exist. The hydraulic sys-
trol force opposing jfitching velocity with a
tem could be turned off and the airplane
constant time lag of I).2 seconds. When these
flown with manual control, in which case
results were combim d and the Nyquist crite-
there was no tendency for oscillations.
rion plotted, conditions of instability were
When the controls were operated on the clearly shown. The source of the instability
ground with the hydraulic system turned on, was found to be friction in the hydraulic
everything felt perfectly normal, just like a valve that admitted Fluid to the servo cylin-
stationary automobile with power steering. der. These valves had tight-fitting rubber
In this case, however, there was no way for seals to prevent lea_:age of hydraulic fluid
the pilot to judge the response of the air- and required a force _f one to two pounds on
plane. The closed loop involving the air- the control stick to operate. Further study
plane, control system, and pilot was broken. showed that this val,,e friction resulted in a

98 Monographs in Aerospace t!istory Number 12--Journey in Aeronautical Research

Power Controls and Pilot-Induced Oscillations

FIGURE 11.6. Simple

simulator used to
-_ Screen
demonstrate to pilots the
possibility of pilot-
induced oscillations Slide
caused by characteristics
projector -_
of the power control

control force almost 180 degrees out of inches corresponding to 1 g change of accel-
phase with the stick displacement at small eration. Two lines were drawn on the screen
amplitudes of motion. and the pilot attempted to move the light spot
rapidly from one line to the other, simulating
I presented these results at a conference in
a lg pull up. The response with the basic
Washington that was called by the Bureau of
control system could be compared with that
Aeronautics of the Navy Department to dis-
with the power control system operating. The
cuss power control systems. The other con-
response with the power control system on
ferees were very interested in the data
frequently showed an oscillation, which the
because most previous opinions of the prob-
pilots considered very similar to that
lem had been based on speculation rather
obtained in flight.
than on a rational analysis.

As previously mentioned, there was no indi- In the design of modem airplanes, a mock-up
cation of any control problem when the con- of the control system and cockpit controls
trol stick was moved with the hydraulic sys- and instruments is often built to study possi-
tem turned on and the airplane on the ble control system problems. This type of
ground. A method was desired to demon- simulator, known as an iron bird, can be very
strate to the pilots that control difficulties useful in studying control feel and response
might be encountered in flight. A very rudi- characteristics. Frequently, however, normal
mentary simulator was built to provide this acceleration is displayed on a cockpit instru-
capability. As shown in figure 11.6, a slide ment that has low sensitivity in terms of
projector was mounted on flexure pivots and pointer motion per g units. In flight, the pilot
driven by a spring attached to the trailing detects changes in acceleration through sens-
edge of the elevator. The motion was damped ing the force acting on his body. Studies have
by a dashpot consisting of a piston with large shown that the sensitivity of the pilot to
clearance moving in a can full of oil. The changes or oscillations in acceleration is very
damping and spring constant could be large. To study pilot-induced oscillations,
adjusted so that the response closely approx- therefore, an instrument with large sensitiv-
imated the normal acceleration response of ity, such as the slide projector device used on
the airplane to elevator motion. The projector the F4U, should be used to give the pilot suf-
produced a spot of light on a screen along- ficient stimulus to represent small changes in
side the cockpit with a motion of several acceleration.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 99

Power Controls and Pilot-Induced Oscillations

FIGURE 11.7. Normal

.......... FgF-3 airplane
-- F4U-4B airplane acceleration during
formation flights of F4U-
4B airplane and F9F-3
3.0 r F9F-3 airplane leading

airplane with alternate
positions as lead airplane
and following airplane;
F4U-4B with and without
1.0 - -,-,-_z_ power controls.

0 I

3.0 l- F4U-4B airplane leading ,-


¢) 0 I

F4U-4B with power cont

z 2.0 - --t_"
1.0 .'- -- _ _ _ /

0 I I I I I 1 I I I I I I I I I I

3.0 F4U-4B airplane leading _., _


1.0 ....

0 2 4 6 8 10 12 14 16 18 20 22 ;4 26 28 30 32
Time, sec

A test to compare the characteristics of dif- pilot applies control to maintain a close for-
ferent control systems in flight to determine mation. Third, the F9F-3 is in the lead with
their susceptibility to pilot-induced oscilla- the F4U-4B with power control. The large-
tions would also be desirable. One technique amplitude oscillation_ of the F4U-4B as the
that was used in the F4U-4B tests is illus- pilot attempts to peJform this tight control
trated in figure 11.7. The F4U-4B was flown task are apparent. Fi aally, the F4U-4B with
in close formation with an F9F-3 airplane, power control is in the lead. The F4U-4B
which had a manual control system with flight path is somewt at unsteady. The F9F-3
is able to follow these oscillations with a
good control characteristics. First, the
records of normal acceleration are shown well-damped and rapid type of response.
with the F9F-3 in the lead and the F4U-4B These records show that a task requiring
tight control is required to bring out the defi-
with manual control. Second, the F4U with
ciencies in the control system.
manual control is leading. In general, the
records for the following airplane are Numerous studies of power controls and feel
expected to show some oscillations as the devices were made _y the Flight Research

100 Monographs in Aerospace history Number 12--Journey in Aeronautical Research

Human Response Characteristics

Division in the early 1950's (ref. 11.1). As by a black box if a suitable transfer function
aircraft manufacturers gained more experi- for his behavior could be found. Such a for-
ence with power control systems, many of mulation would allow prediction of the
the early problems disappeared as the impor- motion of an airplane under control of a
tance of valve friction was recognized and human pilot and might be of value in predict-
means to keep the friction low were devel- ing handling qualities.
oped. Pilot-induced oscillations with other
Some early attempts at representing the
causes, however, remain a persistent source
response of a controlled airplane were made
of trouble (ref. 11.2). Designers must try to
by Professor Otto Koppen and his students at
take advantage of accumulated experience
MIT. The stability of a controlled airplane
and modem analysis and simulation tech-
was first studied by Koppen in 1935 by
niques to try to avoid these problems.
assuming that the controls were moved in
proportion to the angular deviation of the air-
plane from the desired attitude (ref. 11.3).
Human Response Later this work was continued by two MIT
students, Herbert K. Weiss (ref. 11.4) and
Shih-Nge Lin (ref. 5.7), who took into
account lag in the action of the controls.
Autopilots for airplanes had been considered
These reports gave very useful information
by inventors even before the first flight of the
on the action of closed-loop controls, but lit-
Wright brothers. For example, Sir Hiram
tle attention was given to the question of
Maxim in 1893 made a working model of a
whether the controls were moved by a
steam-powered gyroscope and servo cylinder
human pilot or by an autopilot.
to maintain the longitudinal attitude of an
airplane such as his large steam-powered air- When I started work at Langley, Herbert K.
plane. In 1913 Glenn Curtiss demonstrated Weiss was working with the Coast Artillery
hands-off flight with one of his biplanes con- Board at Fort Monroe, also in the city of
trolled by an autopilot designed by Elmer Hampton, Virginia. He was interested in
Sperry. Later, in the 1930's, Sperry auto- human control of pointing of guns, or gun
pilots found practical use in transport air- laying, as the British called it. We had dis-
planes. Anyone working with autopilots cussions of the subject, and he described dif-
could not escape the realization that the ferent methods of control application by a
human pilot, in controlling an airplane, human pilot, such as control of position, rate,
worked on the same principle as an autopiiot. acceleration, and combinations of these
That is, the pilot could sense the disturbances quantities. These ideas were not immediately
from a desired path and apply corrections applicable to airplane control because the
through the control system to correct for the only mechanisms for performing these func-
disturbances. tions were heavy mechanical devices such as
differentials and bali-disk integrators. These
In the late 1930's and 1940's, theories
ideas, however, served to keep up my interest
became widely available to allow prediction
in the subject of human pilot characteristics.
of the controlled motion of an airplane when
the action of the autopilot could be stated An important advance in the analysis of the
mathematically by means of a differential human pilot in controlling a dynamic system
equation or by an expression known as a was made by A. Tustin, a British electrical
transfer function that related the input and engineer (ref. 11.5). His problem was to
output. With such techniques, the details of improve the tracking of a moving target
the mechanism could be neglected and the by an electrically rotated turret in a tank and
device replaced by a black box with the spec- controlled by a human operator. He was
ified transfer function. It was realized that able to characterize the human operator's
the human pilot could likewise be replaced response in terms of a simple control law

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 101

Landing Studies

involving a proportional and a lead term and taking part. I assigned one of my engineers,
a constant time lag caused by the combined James J. Adams, to follow this research and
effects of lags in the pilot's sensory and neu- to conduct studies applicable to our interests.
romuscular systems. In addition, he found This work continued over a period of many
that the human pilot response contained a years and extended into the period of the
superimposed random motion, which he space program.
called the remnant. He was able to use the
pilot transfer function to determine modifica-
tions to the turret control law that would
Landing Studies
greatly reduce the tracking error. He also
suggested methods for further analysis of During WW II, the wing loadings of air-
human pilot response that have occupied planes increased greatly over the values used
researchers for many years since that time. before the war, and ttigh-lift flaps and large
propellers resulted in a very steep glide. In
Inasmuch as military airplanes are used as landing these airplanes, pilots customarily
gun platforms, the tracking accuracy is kept the power on until just before touch-
important in determining the probability of down to reduce the sinking speed. There was
getting hits on the target. Tustin's success in considerable speculation and study, however,
improving the aiming accuracy of tank gun on the problems that might be encountered if
turrets gave hope that similar benefits could a power-off landing was required. One of the
be achieved by applying human response airplanes most extreme in terms of wing
theory to airplanes. Also, the handling- loading and sinking speed was the Martin
qualities requirements formulated as a result B-26 bomber. A flight investigation was
of the NACA research were all subjective made on this airplane around 1944 to investi-
measures, which were based on pilot opin- gate piloting techniques required. In particu-
ion. The ability to use a mathematical repre- lar, runs were made in which the pilot glided
sentation of the human pilot gave hope that a the airplane toward the ground with succes-
more quantitative method of specifying the sively increasing sinking speed to study the
requirements might be possible. These appli- altitude at which the flare should be started
cations gave an incentive for following and how accurately the pilot could judge the
closely the results of research in this field. flare when it was staaed at an altitude much
greater than had been used on most previous
Early research was generally performed by
airplanes (ref. 11.6).
recording the errors of the human operator in
tracking a moving dot on a cathode ray tube About 1947, I became interested in studying
and fitting the recorded response with a this problem analytically. Previous studies of
mathematical function that would match the the landing flare had assumed relatively
performance of the human as closely as pos- small values of the giide angle. This assump-
sible. Work of this kind was performed at tion allowed the trigmometric terms involv-
MIT and was sponsored by the Wright Air ing the glide slope £ to be simplified by the
Development Center at Wright-Patterson assumptions sin 0 = 0, cos 0 = 1. This
Field at several research organizations, assumption allowed calculation of the flare
including Franklin Institute, Systems Tech- paths to be made by .malytical methods. This
nology Incorporated, and the Cornell Aero- assumption was not very accurate with the
nautical Laboratory. At Langley, the equip- steep glide angles of the modern airplanes. I
ment for conducting such research was not felt that flare paths could be calculated quite
readily available until the development of accurately without tlds simplification by use
analog computers and simulators of various of the method of ut determined coefficients
kinds. By this time the study of human referred to previously. In this method, all
response had become an extensive field of terms in the equations are expressed as
research with many research organizations power series in the independent variable t

102 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Landing Studies

(time). Coefficients of like powers of t are published by Langley engineers Albert E.

collected, and the term involving each power Von Doenhoff and George W. Jones, Jr. a few
of t is multiplied by an undetermined coeffi- years later (ref. 11.7). They used numerical
cient. In using the method, the lowest order integration in calculating flare paths.
coefficient may be determined from the ini-
As an aside in connection with the paper by
tial conditions. Then the next higher order
von Doenhoff and Jones, the reader is
coefficient may be determined from this
reminded that von Doenhoff worked in the
coefficient, and so on until all coefficients
Low Turbulence Tunnel Branch and was
have been determined. In the work con-
mainly noted for his airfoil studies and for a
ducted, terms through the fifth order in t
book by Ira Abbott and von Doenhoff on air-
were used. Because the landing flare (in
foils. It might be considered surprising that
reverse time) is approximately a parabolic
an engineer with this background would
curve, the higher order terms rapidly
undertake a study of airplane landings. This
decrease in size and the method is highly
accurate. versatility and broad range of interests, how-
ever, is typical of many of the leading engi-
I set up the method in tables (now called neers at Langley during this period. It is this
spreadsheets) to be filled in with calculated versatility that allowed the engineers at
values by the female computers in the Langley to take a leading role in the space
branch. They used mechanical Frieden or program a few years later, even though this
Marchant calculators in the work. Many program involved disciplines very different
landing flares were calculated for various from their former specialties. Von Doenhoff
values of the parameters. It is well-known himself made an excellent analysis of space
that the shortest possible landing over an power systems comparing photovoltaic cells
obstacle may be obtained by making a flare and thermoelectric generators (ref. 11.8).
at maximum lift coefficient. The starting
Returning to the subject of landings, flare
point of the flare, however, must be at pre-
paths could be determined readily today by
cisely the right point to allow the wheels to
integrating the equations of motion on a digi-
touch the ground when the flight path is hori- tal computer. Also, if desired, simulated
zontal. My objective was to study flare paths
landings could be made by a human pilot in a
that would allow a margin for error and to
simulator with realistic displays of the land-
determine how much the landing distance ing area. Such studies have no doubt been
would be increased as a result.
made in connection with the development of
more recent airplanes.
Despite much study, I was unable to deter-
mine a satisfactory criterion for the margin One useful result that I did obtain from the
of error to be allowed. The error, of course, study was to show that an airplane such as a
depends on the ability of the human pilot to delta-wing airplane, which has relatively
judge altitude and sinking speed. Little work high lift-drag ratio at low angles of attack
had been done on the subject at that time. and rapidly increasing drag as the angle of
Also, I came to realize that there were attack is increased, could probably be landed
marked differences in the lift and drag char- quite easily with power off by a human pilot.
acteristics of different airplanes that would The fiat glide in the approach allows starting
affect the results. Airplanes like the B-26 the flare at a relatively low altitude. I gave a
with power off and flaps down would talk at a department meeting on this subject.
have high drag and a low value of lift-drag This conclusion was confirmed, many years
ratio throughout the landing maneuver, later, when an Air Force pilot safely landed
whereas delta-wing airplanes would have a an F-106, a delta-wing interceptor airplane,
decreasing lift-drag ratio throughout the on the short runway (5000 feet) at Langley
flare. As a result, the data were never pub- after his engine had failed at an altitude of
lished. A report with a similar objective was several thousand feet.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 103

CHAPTER 12 Measurement of Turbulence in
the Atmosphere

The study of turbulence has been of interest Dryden, later a director of the NACA and
to scientists and engineers for many years. by a famous British scientist, G. I. Taylor,
Because of the random nature of turbulence, as well as many other noted scientists
it is subject to the same types of analysis as (ref 12.2). I first became aware of these
random noise, such as is produced in elec- studies when 1 wrote my bachelor's thesis on
tronic circuits by the random discharge of boundary layers in 1939. Soon the subject of
individual electrons or in a rarefied gas by turbulence became a discipline of interest in
the random collisions of molecules. The ran- many fields and could easily become the
dom motions of small particles in a liquid focus of a lifetime career.
that result from bombardment of the parti-
The turbulence in the atmosphere is a ran-
cles by the molecules of the liquid is called
Brownian motion. This phenomenon was dom three-dimensional turbulence field that
can extend for large distances compared
analyzed by Albert Einstein in a famous
paper written in 1905 (ref 12.1). Turbulence to the size of an airplane penetrating it.
Early applications of turbulence theory
became of interest to aerodynamicists first
to atmospheric turbulence were made by
through studies of the boundary layer formed
T. yon Kdrmdn in 1934 (ref 12.3) and by
by air or liquid flowing over a solid bound-
G. L Taylor in 1935. These reports were writ-
ary. The boundary layer could be either lam-
ten before I started work, but I made an
inar or turbulent. The turbulent boundary
effort to review them after I had worked at
layer was found to produce skin friction drag
Langley for some years.
many times greater than that found in the
laminar boundary layer. The tendency for the
When an airplane flies through turbulence,
boundary layer to become turbulent depends
large loads are imposed on the structure that
strongly on the Reynolds number, a quantity
may be critical for the design strength of the
that increases with the size of the model
airplane or repeated smaller loads may lead
being tested, the speed, and the density of the
to fatigue failure of structural components.
test medium, and varies inversely with the
Also, the violent motions of the airplane may
viscosity of the medium. Early tests of small
cause airsickness or fatigue of the crew and
models in low-speed wind tunnels were
passengers. As a result of these effects, atmo-
found to give erroneous results because of
spheric turbulence was recognized as an
the greater tendency of the boundary layer to
important subject for research.
remain laminar on the model as compared to
that on a full-scale airplane. The boundary The type of turbulence most frequently
layer was studied extensively by Hugh L. studied in the theoretical analyses is

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 105

Analytical Studies of the Accuracy of Turbulence Data from Fiight Records

called homogeneous isotropic turbulence. this part of the spectrum the viscous sub-
Homogeneous means that the turbulence is range, indicating that viscosity effects in the
the same at any location, and isotropic air are important. When I became aware of
means that the turbulence field is the same in Heisenberg's report, 1 had it translated and
any direction. In such turbulence, an air- published as a NACA Technical Memoran-
plane would experience the same type of dis- dum. I thought that finis part of the spectrum
turbances flying in any direction. Not all might be of interest for airplanes, but it turns
atmospheric turbulence has these character- out that the waveler gth is so short in this
istics, but this theory appears to be useful in region that it has a t_egligible effect on full-
many practical applications. Two terms scale airplanes, though it may be of some
associated with this turbulence are the auto- interest in wind-tunnel studies of turbulence.

correlation function and the power spec-

The work on atmospheric turbulence made it
trum. The autocorrelation function shows
apparent that the accuracy of turbulence
how, on the average, the turbulent velocity at
measurements made in flight would be
some point is related to the turbulent velocity
important and that studies to show how well
at some other point. Thus, if a gust hits an
the assumption of tugmogeneous, isotropic
airplane, experience has shown that the
turbulence would apply in flight would be of
velocity is approximately the same at the
interest. As a result, studies made of these
wing tips as at the centerline. The velocity at
subjects are presented in the following
some point far away from the airplane
would, on the average, be unrelated to the
velocity at the airplane, which shows that the
correlation function approaches zero at large
distances. The power spectrum considers the
Analytical Studies of
turbulence in the atmosphere to be composed
the Accuracy of
of superimposed sinusoidal waves of differ-
ent wavelengths. The spectrum shows how Turbulence Data from
the amplitude of a wave, on the average, var- Flight Records
ies with the wavelength. It can be shown
mathematically that the correlation function In the time period of the early 1950's,
is related to the power spectrum, so that if numerous applications occurred in flight
one is known, the other can be calculated. research for the analysis of random data.
Records requiring this type of analysis were
The shape of the spectrum of vertical gusts obtained, for example, in studies of pilot
as predicted by von Kdrvn6n on a plot of the tracking error, in the response of airplanes to
gust power (that is, the square of the gust turbulence, and ii_ records of human

velocity in each increment of frequency) as a response to arbitrar.¢ inputs. The approach

function of frequency shows that the power used was to analyze the record by fitting it
is constant at very low frequencies. Then with a Fourier series, which showed the

at some frequency, the power starts to amplitude of the harmonics at various fre-
decrease, and falls off as the frequency to the quencies. The earliest attempts at this type of

-5/3 power. This region of the spectrum is analysis used a Cozadi rolling sphere ana-
called the inertial sabrange, indicating that lyzer, a delicate instr _ment that mechanically
the mass effects are important. This range is read the amplitude o' a component at a given
usually most important in influencing the frequency when tht. record was carefully
loads and motions of an airplane. At still traced by manually moving a stylus along the
higher frequencies, the spectrum was shown record. Later, when electronic computers
by W. Heisenberg, the famous physicist, to became available, the records were digitized
fall off very rapidly and vary as the fre- on punch cards and power spectral data were
quency to the -7power (ref 12.4). He called obtained, but the pr,_cess was quite lengthy

106 Monographs in Aerospace ttistory Number 12--Journey in Aeronautical Research

Analytical Studies of the Accuracy of Turbulence Data from Flight Records

FIGURE 12.1. Error in

["x..__ _ Wave length < record length
harmonic amplitude due
to finite length of record
for various phase angles
of harmonic. " "- - ....... - - -""" Wave length = record length

(a) Examples of waves

that just fit within a - -___ ............. .2_ _- - ---_"'_"_J Wave length 2x record length
rectangular band
(b) Plot showing ratio of
error to line width as 14
a function of ratio of
record length to
wavelength (bottom).


= 8
Maximum error

_ 6
Error averaged over
_ 4 all phase angles

Minimum error
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
Ratio of record length to wave length

with the capabilities of computers available considerably wider than that of the trace
at that time. Now, of course, the availability under smooth conditions. The width of the
of data recording in digitized form, the use of trace could be observed on a physical record
the fast Fourier transform, and the high or could be considered as indicative of an
speed of computation makes the analysis of error band that depended on the accuracy of
such data a routine procedure. the instrumentation involved. The actual

One question that came up, particularly in value of the recorded quantity at any time
the analysis of atmospheric turbulence, was could be anywhere within the width of the
how long a record was required to obtain suf- error band. A study was made to determine
ficiently accurate data on the low-frequency how large a variation in the amplitude and
components of turbulence. In the case of the phase of a given harmonic component could
early studies, a problem apparent to the occur without going outside the boundaries
researcher was that the time history record of the trace over the length of the recorded
obtained as a trace on oscillograph film had a data. In particular, it should be noted that
certain width. Frequently, engine vibration attempts were made to obtain data on
picked up by the instrument made this trace harmonics with wavelengths considerably

Monographs in Aerospace History Number 12 Journey in Aeronautical Research 107

Analytical Studies of the Accuracy of Turbulence Data from Flight Records

FIGURE 12.2. Plot of

140° 130° 120" 110O IO0 ° 90 ° 80O 70O 60 ° 50* 40 °
220 _ 230 ° 240O 250 ° 260* 270 ° 280 ° 290* 300 ° 310° 320 ° amplitude versus phase
angle for waves with
150 °
30° various ratios of wave-
330 °
210 °
length to record length.

20 °
160 °
2O3 °

170 °
350 °
190 °

180 °
0 2 4 6 8 10

_Ratio of error to line width

35O _
190 °
10 o
170 °

340 °
200 ° 20*
160 °

330 °
210 ° 30 °
150 °

220O 230 ° 240 ° 250 ° 260 _ 270 ° 280 ° 290 ° 300 _ 310" 320 °
140O 130 ° 120O 110 ° 100 _ 90O 80 ° 70 ° 60 ° 50 ° 40 °

longer than the length of the time history to line width as a fuaction of ratio of record
available. This analysis involved relatively length to wavelength for different conditions
straightforward trigonometric manipulations. of phase angle of the harmonic. The condi-
One reason for showing the results here is tions shown are those giving the maximum
that the resulting plot of amplitude versus error, those giving the minimum error, and
phase angle of the harmonics at various the error obtained by averaging over all
wavelengths has an unusually artistic form. phase angles. Inasm ach as the phase angles
in power spectral an dysis are assumed to be
First, in figure 12.1(a) are shown several
uniformly distribut(d, the latter curve is
examples of harmonic components of waves
probably more repr_ sentative of what could
that just fit within a rectangle with length
be obtained with a statistical analysis.
equal to the length of the record and width
equal to the error band. In practice, the Figure 12.2 shows _he polar plot of ampli-
record is many times longer than its width, tude versus phase an_le for harmonics of dif-
but the figure shows a much shorter and ferent ratios of wave length to record length.
wider rectangle for clarity. As shown, so long The nesting of the, arious lenticular curves
as the wavelength is less than the record presents an unusual artistic effect not often
length, the maximum error of the component seen in theoretical p_ots.
is the same as the width of the error band.
When the wavelength of the harmonic com- This analysis was never published, mainly
ponent is larger than the record length, how- because more sophisticated approaches were
ever, the error of the component exceeds the being developed be mathematicians with
width of the error band by an amount much more knowled ge of statistical analyses
depending on the phase angle of the compo- than I had. The next _ection describes a flight
nent. Figure 12.1(b) shows the ratio of error investigation to mak ,"accurate measurements

108 Monographs in Aerospace ltistory Number 12--Journey in Aeronautical Research

Measurements of the Spectrum of Atmospheric Turbulence Over a Large Range of Wavelengths

FIGURE 12.3. Photograph

of F9F-3 airplane
showing special
instrumentation for

high frequency
angle-of-attack vane

Very high frequencyJ

sideslip vane Airspeed

of the spectrum of atmospheric turbulence result of these considerations, a rule of

that uses the most advanced statistical tech- thumb is that approximately ten to twenty
niques available at the time. wavelengths of the lowest frequency under
consideration should be contained in the
record. Obviously, in considering atmo-
spheric turbulence that contains much energy
Measurements of the
at wavelengths of several thousand feet, a
Spectrum of very long record in uniform conditions of
Atmospheric turbulence is required.
Turbulence Over a
In my position as head of the Stability and
Large Range of Control Branch of the Flight Research
Division in the early 1950's, I was able to
plan programs of flight research and had
A much more important source of error than available test airplanes and instrumentation
the error introduced by the instrumentation is to perform research studies. One of my engi-
the variability of the power spectral data neers, Robert G. Chilton, a recent MIT
caused by the random nature of the process graduate, was familiar with the latest mathe-
being analyzed. Thus, various samples of a matical techniques to analyze records of tur-
given process taken at different times will bulence. The measurement of atmospheric
show different harmonic content, particularly turbulence was officially assigned to the
at the longer wavelengths. To obtain an accu- Aircraft Loads Division, but the results of
rate measure of the power spectrum, many such studies were of immediate use in my
records should be analyzed and the results branch to study response of airplanes to
averaged or one very long record taken under turbulence. Inasmuch as accurate measure-
uniform conditions should be analyzed. As a ments over a wide range of wavelengths

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 109

Measurements of the Spectrum of Atmospheric Turbulence Over a Large Range of Wavelengths

FIGURE 12.4. Power

10 6 spectrum of gust vertical
velocity for wavelengths
of 10 feet to 60,000 feet.
Dotted lines show 95
percent confidence
10 5

.j%_ ",

_, 104 ,x,"

x "

-d 103

_ 102

"_.. tl

xx x
_ 101

x 7-

10 0

10-1 I I I I IIII I I

10-5 10-4 10-3 10-2 10-1 I00

Inverse wavelength, cycles/ft

were not then available, I obtained approval The airplane and some of the instrumenta-
to conduct a flight test to obtain these data. tion are shown in fig)tre 12.3.
These tests used instrumentation with greater
accuracy than had been used previously, such With the aid of this _.quipment, a flight was
as a sun camera (rather than a gyroscope) to made at an altitude _,f 1700 feet over a dis-
determine airplane pitch angle and a high- tance of 170 miles between Norfolk and
frequency vane to measure angle of attack. Baltimore with a Na,,y F9F-3 Cougar fighter

110 Monographs in Aerospace Uistory Number 12--Journey in Aeronautical Research

Measurements of the Spectrum of Atmospheric Turbulence Over a Large Range of Wavelengths

airplane. The weather conditions were favor- ments at various altitudes and meteorological
able in that a cold front had just passed conditions. It is believed, however, that the
through the entire area, which gave a strong study presented in reference 12.5 is still one
ground wind and consistent clear-air turbu- of the most useful for comparing the spec-
lence over the entire path. The results of this trum of atmospheric turbulence with the the-
test are presented in a NACA Technical Note oretical predictions over a wide range of
(ref. 12.5). wavelengths.

The measured spectrum of turbulence is

shown in figure 12.4. To obtain maximum It is evident that my early analysis has little
accuracy in different frequency ranges, dif- relation to the more important problems of
ferent portions of the record were analyzed obtaining accurate power spectral data. The
with different spacing of the recorded points. artistic results of the analysis, therefore, may
Thus, for high frequencies, a shorter portion be considered more a mathematical oddity
of the record with more closely spaced than a useful analytical result. Subsequent
points was analyzed. As a result, the plotted developments of turbulence theory by
curve is broken into sections. Since these experts in this field, however, were followed
tests were made, many studies have been closely and allowed useful data to be
made by NASA, the Air Force, and other obtained on the spectrum of turbulence or in
organizations to obtain turbulence measure- other studies involving random data.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 111

CHAPTER 13 GustAlleviation

The response of airplanes to gusts has been a planes with reduced response to turbulence.
subject of concern to airplane designers All of these attempts were characterized by
since the earliest days of aviation. The an intuitive approach with no attempt at
Wright Brothers, flying in high winds at low analysis prior to flight tests, and all were
altitude on the seashore at Kill Devil Hills, notably unsuccessful.
North Carolina, purposely designed their
One of these airplanes (figure 13.1) was
gliders and their powered airplane with low
designed by Waldo Waterman. It had wings
or negative dihedral to avoid lateral upsets
attached to the fuselage with skewed hinges
due to side gusts. Pilots soon found the
and restrained by pneumatic struts that acted
necessity of seat belts to avoid being tossed
as springs. The effect of the skewed hinge
out of the seats of their machines when flying
was to reduce the angle of attack of the wing
through turbulence. One of the first women
panels when they deflected upward, and vice
pilots, Harriet Quimby, and her passenger,
versa. The response to gusts was not notice-
flying without seat belts in a Bleriot airplane
ably reduced from that of the airplane with
over Boston, Massachusetts, in 1911, were
the wings locked, probably because the
unfortunately thrown out of their machine
dynamic response of the system was not suit-
and fell 1500 feet to their deaths. In more
able. Also, the degree of flexibility of the
recent years, gusts have been recognized as
wings was limited because deflection of the
one of the sources of critical design loads on
ailerons would deflect the wings to oppose
airplanes, as well as a source of fatigue
the aileron rolling moment, which resulted in
loads due to repeated small loads. In addi-
reduced or reversed roll response.
tion to these design problems, many people
are susceptible to airsickness when flying The effect of wings with skewed, spring-
through rough air. loaded hinges is similar to the effect of bend-
ing of a swept wing. Airplanes with swept
Despite the long interest of airplane design-
wings do have smoother rides in certain
ers in the effects of turbulence, very few
frequency ranges and suffer from reduced
efforts have been made to design airplanes
aileron reversal speed when compared with
with reduced response to gusts. Even in
airplanes with unswept wings.
1995, none of the commonly used transport
airplanes or general aviation aircraft were A similar method that has been tried in flight
equipped with gust-alleviation systems. is to incorporate springs in the struts of a
conventional strut-braced high-wing mono-
In the past, several attempts have been plane. This method may be likened to the
made by airplane designers to build air- springs used in an automobile chassis to

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 113

FIGURE 13.1. Waterman's
airplane incorporating
wings attached to
fuselage with skewed
hinges. Pneumatic struts,
shown in extended and
compressed positions,
balanced lift forces in

reduce bumps. This method has also proved opposite from wha ° is normally obtained
ineffective, probably because of the slow with a downward-teflected landing flap.
dynamic response of the system. Also, the dynamic n sponse of the wing may
be too slow to provi_'e reduction of the accel-
Other schemes involving wing motion have
erations due to high. frequency gusts.
been proposed from time to time, and some
of them have been investigated in wind-tun- In England, shortl) after WW lI, a large
nel tests or in flight. One method that has commercial airplane called the Brabazon
been given considerable attention is the was designed. In th," design stage, a system
"free wing" concept. In this method, the was incorporated t.) reduce wing bending
wing is pivoted with respect to the fuselage due to gusts by ope'ating the ailerons sym-
about a spanwise hinge ahead of its aerody- metrically to oppos," bending due to gusts.
namic center, and its angle of attack is con- The ailerons were to be operated by a
trolled by a flap on the trailing edge. A seri- mechanical linkage .:onnected to the wing in
ous disadvantage of this method is that up- a way to be moved by wing bending. The
flap deflection must be used to trim the wing system was abandoned before the airplane
at a high-lift coefficient for landing. This flap was flown, and the airplane never went into
obviously reduces the maximum lift, just the production. Nevertheless, the project stimu-

114 Monographs in Aerospace llistory Number 12--Journey in Aeronautical Research

Background and Analysis of Gust Alleviation

lated interest in a flight project at the Royal

Background and
Aircraft Establishment (RAE) in which a sys-
tem of this type was tried in a Lancaster Analysis of Gust
bomber. This system used a vane ahead of Alleviation
the nose as a gust detector to operate the
ailerons symmetrically through a hydraulic I became head of the Stability and Control
servomechanism. The system was built with Section of the Flight Research Division in
little preliminary analysis, and when the 1943. During the wartime years, there was
pilot engaged the system in flight for the first little difficulty in deciding on the type of
time, the flight in rough air seemed notice- work to be done by the section. Most of the
ably more bumpy than without the system. By work was concerned with flying qualities or
reversing the sign of the gain constant relat- with improving the control systems of air-
ing aileron deflection to vane deflection, the planes. Some of this work has been
ride was made somewhat smoother. Later, described in the preceding sections or may
an analysis by an RAE engineer named be further seen from my list of reports
J. Zbrozek showed the reasons for the unex- (appendix II). By 1947, jet airplanes with
pected behavior. These reasons will be men- power control systems were being devel-
tioned later in the presentation. oped, and the use of automatic control to
improve the stability characteristics of air-
Another experimental program was con- planes was a rapidly developing field. Appli-
ducted on a C-47 airplane by the Air Force. cations of automatic control were therefore
This system was similar to that originally important subjects of research. The change
planned for the Brabazon. The ailerons were in emphasis was recognized in 1952 when I
arranged to deflect symmetrically upward was made head of the Guidance and Control
with upward wing bending, and vice versa, Branch of the Flight Research Division.
by means of a linkage which added a compo-
In the military services, the new technology
nent of this deflection to that of the conven-
of automatic control was largely applied
tional aileron linkage. Since the wing deflec-
in the development of guided missiles. At
tion provided a large driving force, no
Langley, the Pilotless Aircraft Research
servomechanism was required, and as a
Division, under Gilruth, had been established
result, a system of high reliability was
to use missiles for aerodynamic research.
expected The system suffered from the same
Some of the engineers in that division, how-
objections as the one tested on the Lancaster.
ever, became interested in guided missiles.
In addition, the inertia of the ailerons com-
Gilruth and William N. Gardiner started to
bined with flexibility of the operating linkage
perform tests on a missile with infrared guid-
caused the aileron deflection to lag behind
ance that used gyroscopic stabilization and
the wing deflection. Such a system is very
employed rocket propulsion techniques
conducive to flutter. To avoid flutter, the ratio
developed for aerodynamic testing. It later
between the aileron deflection and wing
turned out that this missile was very similar
bending had to be kept to a very low value.
in concept to the Sidewinder missile devel-
As a result, the system was unable to provide
oped by the Navy, which was one of the first
more than 9 percent reduction in wing bend-
air-to-air guided missiles widely employed
ing moments, a rather small improvement.
by the armed services.

Despite the discouraging results of these The Deputy Director of the NACA at that
experiments, the advantages of gust allevia- time was Dr. Hugh L. Dryden, a noted scien-
tion remained worthwhile. As a result, a tist, who was also a lay preacher in the
project was initiated at Langley to study this Methodist Church and a religious man. He
subject. These activities are described in the did not like to see the NACA centers unnec-
following section. essarily involved in military work. He issued

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 115

Background and Analysis of Gust Alleviation

a directive that no work with military appli- had devised a gust-alleviation system in
cations should be done at the NACA centers which the halves of the horizontal tail were
unless requested by the military services. attached by chordx_'ise hinges. These sur-
faces were connected by pushrods to flaps on
From the standpoint of a stability and control
the wing. On encountering an upward gust,
engineer, the study of missile guidance sys-
the tall halves would deflect up, moving the
tems was one of the most interesting and
flaps up and thereb)- offsetting the effect of
challenging fields available at that time.
the gust. Other features of the system made
Nevertheless, I had the same feelings as
the airplane insensitive to horizontal gusts
Dr. Dryden concerning the desire to avoid
and to rolling gust,,, all without adversely
emphasis on military projects. With this area
affecting the ability _f the pilot to control the
of research ruled out, I was faced with the
airplane. These many ingenious features are
question of what type of automatic control
too complex to discuss herein.
research with peacetime applications was of
most interest. At that time, airplanes with I was very impressed by the report for two
swept wings were starting to be fitted with reasons. First, Hirsch had performed an anal-
yaw dampers to improve the damping of lat- ysis to determine the relations between the
eral oscillations, particularly at high alti- tail, elevator, and flap hinge-moment charac-
tudes. Yaw dampers, however, did not teristics; linkage ratios; and other parameters
require much research. In most cases, com- so that the flaps moved exactly the right
panies were able to hook up a rate gyroscope amount to offset the effect of a gust. This
through an analog amplifier to the power analysis required consideration of so-called
control actuator on the rudder, and the sys- stability derivatives for all these components.
tem worked very well in damping out lateral At that time, stability derivatives were used
oscillations. to describe the stability characteristics of an
airplane, such as, the variation of pitching
After much thought, I concluded that one
field in which automatic control could be moment with angle of attack Cra and the
..... _t

variation of hft coefficient w_th angle of

applied and which had not previously been
studied to any great extent, was gust allevia- attack C L . Prior to, 1938, very few people
had ever _onsidered similar quantities for
tion. A gust-alleviation system is one that
control surfaces or flaps, such as variation of
provides the airplane with a smooth ride
flap hinge moment with angle of attack or
through rough air. At that time, all transport
variation of elevator hinge moment with
airplanes had piston engines and flew at alti-
deflection. Hirsch's analysis required a mul-
tudes below the top of storm clouds. Fear of
airsickness was a common problem and was titude of these quantities, as well as linkage
a deterrent to many people who wanted to ratios and other quantities describing the
travel on commercial airlines. mechanism, all tiex together by algebraic
equations that wert: solved to obtain the
I was already familiar with one study of gust desired design characteristics of the system.
alleviation that had been made. Philip
Donely, who was then a branch head in the The second important contribution by Hirsch
Loads Division, had formerly been designer was that he tested hi; system with a dynamic
of the gust tunnel at Langley. In this wind model. I have previcusly mentioned some of
tunnel, a model was catapulted at flying my experiments with a dynamic model. In
speed through a vertical jet of air to simulate Hirsch's case, the n _odel was mounted in a
a gust in the atmosphere. Evidently while wind tunnel so that i was free to pitch and to
doing this work, Donely had come across a slide up and down o:l a vertical rod. The tun-
report by Ren6 Hirsch in France on a study nel had an open tl-roat and was equipped
of gust alleviation done as a doctoral thesis with a series of slats ahead of the test section,
and published in 1938 (ref. 13.1). Donely similar to a veneti;_n blind, that could be
brought this report to my attention. Hirsch deflected to produce an abrupt change in the

116 Monographs in Aerospace llistory Number 12--Journey in Aeronautical Research

Background and Analysis of Gust Alleviation

flow direction. When Hirsch tested his model which the ailerons were moved symmetri-
in this artificial gust with the system locked, cally in response to gusts sensed by a vane
it immediately banged up against its stop at on the nose. All these attempts, made with-
the top of the rod. With the system working, out any theoretical analysis, had been unsuc-
however, it showed only a small disturbance cessful in that very little alleviation of
and settled back to stable flight. airplane accelerations was obtained. My
analysis showed why each of the attempts
Hirsch was so impressed by these results that
had failed. In general, the lack of attention to
he devoted most of his professional life to
pitching moments produced by the control
demonstrating the system in flight. I later
surfaces was responsible. In the case of the
corresponded with Hirsch, visited him in
Lancaster, for example, the symmetric
France, and saw his airplanes. This part of
deflection of the ailerons in the upward
the story will be mentioned later. At the time
direction proportional to a positive change in
I read the thesis, however, I simply realized
angle of attack produced a positive, or
that gust alleviation was a feasible idea and
upward, pitching moment. A positive varia-
that with automatic controls, it might be pos-
tion of pitching moment with angle of attack
sible to do the job more simply than Hirsch
represents a decrease in longitudinal stabil-
had done with his complex aeromechanical
ity. This decrease in stability increased the
response to low-frequency gusts, which resu-
I started, in 1948, to make analyses of the lated in an increase in the bumpy ride experi-
response of example airplanes to sinusoidal enced by the pilots. In addition, the aileron
gusts and the control motions that would be motion reduced the damping of the wing
required to reduce (alleviate) the response of bending oscillation, which increased the
the airplane or the accelerations that would effect of structural oscillations on the sensa-
be applied to the passengers. The method tion of the pilots. In considering the effect of
used was what has been described earlier as the ailerons on the pitching moments, it
the frequency-response method, which had should be realized that a symmetric upward
the advantage that the calculations were deflection of the ailerons produces a direct
greatly simplified when compared with cal- effect on the pitching moments of the wing.
culating response to discrete gust inputs. It also changes the wing lift distribution to
These early studies showed, as was known produce an increased downwash on the tall
from experience, that control by the elevators due to up aileron deflection, which further
alone was ineffective. Control by flaps on the increases the destabilizing variation of pitch-
wing was thought to be promising, but the ing moment with angle of attack. My analy-
analysis showed that in many cases, the flaps sis showed that the downwash effects on the
would produce excessive pitching response tail due to the deflection of flaps or ailerons
of the airplane. In general, control by a com- on the wing are very important in designing
bination of flap and elevators was required. a gust-alleviation system.
With flaps alone, successful results could be
After making these preliminary studies, I
obtained only by careful attention to the
analyzed a system in which a gust-sensing
pitching moments applied to the airplane by
vane mounted on a boom ahead of the nose
the gust and by the flaps.
was used to operate flaps on the wing
A survey was also made of previous attempts through a hydraulic servomechanism. Any
at gust alleviation on full-scale airplanes. gust-alleviation system working on this prin-
These included an airplane made by ciple reduces the lift produced by a change in
Waterman with wings pivoted on skewed angle of attack. For complete alleviation, the
hinges, a DC-3 modified by the Air Force so lift due to angle of attack is reduced to zero.
that the ailerons deflected symmetrically Since the pilot maneuvers the airplane by
in response to wing bending, and an Avro changing its angle of attack, this system
Lancaster bomber modified by the British in would prevent the pilot from making any

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 117

Background and Analysis of Gust Alleviation

longitudinal maneuvers. To restore this capa- system had the possibility of greatly exceed-
bility, the input from the control stick, nor- ing these limits, which made the airplane
mally used to move the elevators, was also either violently unstable or excessively sta-
fed to the flap servomechanism. With this ble. It was possible to solve the equations to
arrangement, when the pilot moves the stick determine the flap and elevator motion to
back to make a pull up, the flaps first go completely offset both the lift and pitching
down to produce upward lift. Then, as the moments applied to ;he airplane. This analy-
angle of attack increases in response to the sis showed that this objective could be
elevator motion, the flaps move back to neu- attained in two ways. In one method, both
tral and the pull up is continued with the air- the flap and elevator had to be moved in
plane at a higher angle of attack. The result is response to the gust, but the elevator motion
a faster response to control motion than was not in phase w_th the flap motion and
obtained with a conventional airplane. This generally had to lag behind the flap motion.
type of control, in later years, has been called In the other method, the elevator and flap
"direct lift control" and is advantageous for moved in phase, but the downwash from the
control situations requiring rapid response. flaps on the tail had to be of opposite sign
from that normally encountered. That is,
The analysis also showed that provision had
down flap deflection had to produce an
to be made for avoiding excessive pitching
upwash at the horizontal tail.
moments due to the gusts and flap deflection.
Since the flaps were moved in proportion to Having reached this stage in the analysis, the
angle of attack, any pitching moment from results appeared sufficiently promising for a
the flaps contributed directly to the variation report on the results and a flight program to
of pitching moment with angle of attack, demonstrate a gust-alleviation system. A job
which determines the longitudinal stability order request was submitted about June 1948
of the airplane. For satisfactory stability, the to obtain official approval for this work. This
pitching moment due to angle of attack must job order is presented here to illustrate the
be kept within prescribed limits. The addi- type of request required to get a research
tional contribution due to the gust-alleviation program approved.

Title: Theoretical and Experimental Study of Means to Increase the Smoothness of

Flight through Rough Air.

Est. Man Hrs: 3000 Cost: $6000

Description: A theoretical study has been made of various means to increase the
smoothness of flight through rough air. This job order is to cover placing this analy-
sis in a form suitable for a report, and preparation of a report on this study. In addi-
tion, measurements will be made in flight to verify the predicted r ".sponse character-
istics of airplanes, and bench tests will be made of servomechat isms intended for
use with the automatic control device.

Justification: The preliminary analysis has indicated that the successful operation of
a device to provide smooth flight in rough air requires careful selection of the design
parameters, but that with a suitable design very promising results may be obtained.
Such a device is not primarily intended to reduce stresses in tl'e airframe due to
severe abrupt gusts, but rather to reduce changes in acceleration _ hich are primarily
responsible for passenger discomfort and airsickness. The anal lsis indicates that
current types of servomechanisms are capable of providing the desired rapid
response of the controls. Inasmuch as a device of this type would _e of great interest
for airline operation, it is considered desirable to publish the resldts of the analysis
and to continue this work with a view to eventual flight demor, stration of such a

118 Monographs in Aerospace ttistory Number 12 Journey in Aeronautical Research

Background and Analysis of Gust Alleviation

Note the manhour estimate. The value of tems affect the controllability of the airplane,
3000, about 1.5 man years, was just a guess detailed studies were made of static and
based on the effort put into previous reports. dynamic longitudinal stability of airplanes
The cost of $6000 was based on a standard incorporating these systems.
rule of $2 per manhour. Little effort was
required for this phase of the job approval. In various studies of gust-alleviation systems
made by other researchers since the one
When the work described in this job order
described herein, the now popular approach
was undertaken, I appointed Christopher C.
of optimal control theory has been applied.
(Chris) Kraft, Jr. as project engineer. Kraft
In this method, some balance is sought
was then an engineer in my section with between the amount of control motion
experience in flying qualities and with work
required by the system and the amount of
using the free-fall and wing-flow methods. reduction of acceleration achieved. In the
Later in his career, he was a flight controller
study made by Phillips and Kraft, however,
during the Apollo missions and was made
an effort was made to achieve complete gust
director of the Johnson Space Flight Center
alleviation within the limits imposed by the
following Dr. Gilruth's retirement. Kraft and
assumptions of the theory. Complete allevia-
I made additional calculations to have a logi-
tion was found to be possible with the vane-
cal series of examples to place in the report, type sensor, but not with the accelerometer
which was later entitled Theoretical Study of
sensor. Inasmuch as complete alleviation was
Some Methods for Increasing the Smooth- found to be possible with reasonable control
ness of Flight through Rough Air (ref. 13.2).
motions, the use of optimal control theory
The report starts with a review of the avail-
when a vane sensor is used is really not opti-
able data on the causes of airsickness. The
mal and represents an inappropriate applica-
main source of data, which is still believed to
tion of this theory.
be the best available, was a series of tests
made during WW II at Wesleyan University, The theory used for the analysis is very simi-
in which subjects were tested with various lar to that taught by Professor Koppen in my
wave forms of vertical acceleration in an ele- courses at MIT. This theory, in turn, was
vator. The results showed that relatively low- based on the theory first presented by G. H.
frequency variations of acceleration that had Bryan in England in 1903 and later in the
periods of 1.4 seconds and greater were the textbook Stability in Aviation in 1911
most important causes of motion sickness. (ref. 4.1). This theory shows that the longitu-
Excellent control of the directional and roll- dinal and lateral motions of the airplane can
be considered separately and assumes small
ing disturbances of airplanes could be
disturbances so that all aerodynamic forces
obtained with conventional autopilots avail-
and moments can be assumed to vary lin-
able at the time the report was written. These
early with the magnitude of the disturbance.
devices, however, were relatively ineffective
This theory was extended to calculate the
in reducing the vertical accelerations of air-
planes. The report therefore concentrated on response of an airplane to gusts and reported
in NACA Report No. 1 and later reports in
the problems of longitudinal gust alleviation
and control. the period 1915-1918 by Edwin B. Wilson,
then a professor of Physics at MIT. To study
Following a section on the theoretical analy- the effect of a gust-alleviation system, it was
sis, examples were included in the report to necessary only to add to Wilson's theory the
show the use of elevator alone, flaps alone, additional forces and moments caused by the
and a combination of flap and elevator airplane's control surfaces as they were
motion to offset the effect of gusts. Next, two moved by the gust-alleviation system. Like
types of gust-sensing devices, a vane ahead Wilson, I considered that the gust was con-
of the nose and an accelerometer in the air- stant across the span, though this subject was
plane, were considered. Because these sys- studied in more detail later.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 119

Background and Analysis of Gust Alleviation

Two refinements were added to the theory, as interpretation. This interpretation is shown in
presented by Wilson, that were found to figure 13.2, in which an airplane with a vane-
be of importance for the study of gust- type gust-alleviation system is shown pene-
alleviation systems and that are believed to trating a region in the atmosphere where
make the results very close to what would be there is a change in vertical gust velocity
obtained with an exact computer simulation (called a step gust). First, the vane is
such as would be possible with modem elec- deflected by the gust. If the servomechanism
tronic computers. First, the penetration effect operating the flaps has a time lag equal to the
was considered, that is, the difference in the time for the gust to reach the wing, then the
time of penetrating the gust by a vane ahead flap moves just at the fight time and the fight
of the nose, the wing, and the tail. In amount to offset the lift change on the wing
Wilson's study, the gust was assumed to due to the gust. For the airplane response to
affect all parts of the airplane simulta- be zero, however, the pitching moment about
neously. Second, the time lead or lag effects the center of gravity caused by the flap
caused by gust penetration were approxi- deflection must be zero. This condition is not
mated by a iinearized representation to keep ordinarily obtained with conventional wing
the equations linear. This technique had been flaps, but may be obtained by moving the
used by Cowley and Glauert in a British elevators the correct amount in phase with
report published in 1921 to improve the cal- the flaps. A little later, the tail is affected by
culation of pitch damping of airplanes by the gust and by the downwash from the flaps.
taking into account the time for the down- The effect of the forces on the tail can be
wash leaving the wing to reach the tail eliminated in two ways. Either the flap
(ref. 13.3). A similar method was used in the downwash must be equal and opposite to the
present analysis to account for all the lead effect of the gust on the tail or the elevator
and lag effects, such as the lead of the vane must be given an additional movement to
in penetrating the gust, the lag in response of cancel the combined effect of the flap down-
the servomechanism operating the flaps, and wash and the gust.
the lag of downwash from the wing and flaps
in hitting the tail, in addition to the lag of the
The simple interpretation of the action of a
gust itself in reaching the tail. These lead and
perfect gust alleviation system has interest-
lag effects were found to be very important,
ing ramifications. First, the influence of the
particularly in affecting the damping of the
feedback of angle of attack from the vane
short-period longitudinal motion of the alle-
makes the alleviaticn system a closed-loop
viated airplane.
control system, but t le feedback decreases as
the system approaches the condition of com-
A useful advantage was found in this
plete gust alleviation. In this condition, the
approach in that all the effects of the allevia-
system behaves as an open-loop control,
tion systems studied could be considered as
because there is no motion of the airplane to
changes to stability derivatives of the basic
be sensed by the vare. Second, the consider-
airplane. Most of the effects of these deriva-
tives were known from experience, or at least ation of lag effects i_ seen to be exact in the

had simple physical interpretations. In addi- limiting case of p._rfect alleviation, even
tion, the order of the equations was not though these results were obtained from an
increased over that of the basic airplane. approximate lineariz _d theory. Third, the dis-
covery of this interpretation could presum-
These equations made it possible to solve for ably have been made a priori, without use of
the characteristics of the system that would any theory, but in my case, working through
produce complete gust alleviation, that is, the theory first and examining the resulting
that would produce zero response to a gust. formulas was necessary for me to realize that
In examining these formulas, I suddenly this interpretation existed. Many simple
realized that the results had a simple physical physical principles in the history of physics

120 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Design and Test of a Gust-Alleviated Airplane

FIGURE 13.2. Concept for

complete alleviation of
vertical gusts shown by
the theoretical analysis.
Note that the elevator
moves in phase with the
flaps, offsetting the flap
pitching moment. Flap
downwash, with opposite
direction from normal,
offsets the lift on the
Step gust
horizontal tail due to the
Lift due to gust-_ _Flap deflection /_/__ Elevation deflectio n

._ __--- __Cmsf=O

_ _ ___ _ _ _ ___ _ ___ ____ _ ___ _ _-- Lift due to flap

Lift on tail due

to gust
Flap deflection _-Elevation deflection

.___,__,__,__,__,__,__,__,_,__,_ o owowas

have been realized only after long periods of short-period vertical oscillation would result.
thought and analysis by their discoverers. As a result, the amount of alleviation poten-
tially available with this system was limited.
The vane-type system studied has some dis-
On the other hand, the use of the accelerome-
advantages when adjusted to give complete
ter sensor inside the airplane avoids the prob-
alleviation. The system results in an airplane
lem of having a delicate vane exposed to
with zero lift and zero pitching moment due
potential damage from handling. In practice,
to angle of attack. The airplane would
the system with an accelerometer sensor
respond to pilot's commands as provided by
would be much more likely to excite struc-
the direct lift control system, but would have
tural oscillations of the airplane, thereby fur-
no inherent tendency to stabilize in a new
ther limiting the gain and requiring a more
equilibrium condition. For this reason, sys-
tems were studied that retained a small detailed analysis to insure the safety of the
amount of longitudinal stability. Use of these
systems was found to be feasible and did not
seriously reduce the gust-alleviation proper-
ties of the perfect system. Design and Test of a ....
The studies of the use of an accelerometer to Gust-Alleviated
sense the gusts showed that the gain of the Airplane
system, that is, the amount of flap deflection
used for a given change in acceleration, had During the publication process of the report
to be limited otherwise a poorly damped on the analysis, work was started on a

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 121

Design and Test of a Gust-Alleviated Airplane

FIGURE 13.3. Two-view

477 drawing of the Beech B-
18 test airplane showing
the modified control
surfaces. Dimensions in

/-- Main elevator

Auxiliary elevator __--807" _

Auxiliary flap
Main flap


program to demonstrate gust alleviation in At the time the design was started, about
flight. The airplane chosen for the program 1950, electronic control systems had a poor
was a Beech B-18, a small twin-engine trans- reputation for reliability. These systems used
port. The airplane was obtained from the vacuum-tube amplifiers. Furthermore, the
use of techniques o! redundancy to improve
Navy and had the Navy designation C-45.
reliability had not tl en been developed. For
Because of the need for some major alter-
these reasons, man) of the design features
ations to the airplane control surfaces,
were governed by s;tfety considerations. All
Kenneth Bush and Edwin C. Kilgore of the
autopiiots in use at hat time were designed
Engineering Division were called into the so that the pilot coud readily overpower the
project to do the design work. Steve Rock of autopilot with his manual control system in
the Instrument Research Division was the event of a failure. This method could not
assigned to design the servomechanism to be used with the gust-alleviation system
actuate the control surfaces. because the wing flaps, which required large

122 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Design and Test of a Gust-Alleviated Airplane

FIGURE 13.4. Nose boom

and angle-of-attack vane
installation on Beech B-
18 test airplane.

operating forces, were not normally con- As stated previously, perfect gust alleviation
nected to the pilot's control stick. As a result, according to the theory could have been
the design features described in the follow- obtained either by driving the elevator sepa-
ing paragraphs were incorporated. rately from the flaps and with a different
A drawing of the airplane as modified for the phase relationship or the elevator could have
gust-alleviation project is shown in fig- been geared directly to the flaps and the
ure 13.3. A boom was built on the nose to downwash from the flaps altered to offset the
hold the angle-of-attack vane as shown in gust at the tail. The latter method was
figure 13.4. The wing flaps, which normally selected for the following reasons. With a
deflect only downward, were modified to direct mechanical linkage between the flaps
move up and down because both up and and the outboard elevators, the pitching
down gusts must be counteracted by the sys- moment due to flap deflection, a critical
tem. The elevator was split into three sec-
quantity for longitudinal stability, could be
tions, the two outboard segments being
finely adjusted as required and would hold
linked to the flaps for use with the gust-
its setting. If a separate servomechanism and
alleviation system and the inboard segment
electronic amplifier had been used to operate
being used in the normal manner for pitch
control (figure 13.5). Finally, small segments the elevators, the gain of the amplifier might
of the flaps near the fuselage were driven have drifted and caused the airplane to
separately from the rest of the flap system so become unstable. In fact, the gain of the
that they could be geared to move either in amplifier between the vane and the flaps
the same direction or in the opposite direc- often did vary in flight by amounts that could
tion from the rest of the flaps (figure 13.6). have caused violent instability if a similar

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 123

Design and Test of a Gust-Alleviated Airplane

FIGURE 13.5. Three-

quarter rear view of
Beech B-18 test airplane
showing elevator control
split into three segments.

FIGURE 13.6. Three-

quarter rear view of
Beech B-18 test airplane
showing modified flap
system with oppositely

!iii !ii ! deflected inboard


amplifier had been used to operate the electrical input variable-displacement pump
elevators. hydraulic servomechanism, which was taken
from a naval gun turret. Electrical signals
The small inboard segments of the flaps were
from the vane and the pilot's control column
used to reverse the direction of the flap
were combined in _ vacuum-tube amplifier
downwash at the tail, as required by the the-
and fed to the control valve of the servo-
ory if the elevators moved in phase with the
mechanism. For lal:ral control, the entire
flaps. This method of course, reduces the flap
effectiveness in producing lift. To regain suf- flap and aileron syst :m was deflected asym-
ficient flap effectiveness, the flaps and aile- metrically through i separate servomecha-
rons were geared together so that the flaps nism of the same type. The system was
and ailerons deflected symmetrically for gust designed so that the airplane could be flown
alleviation. These surfaces were driven by an through its original manual control system if

124 Monographs in Aerospace ltistory Number 12--Journey in Aeronautical Research

Design and Test of a Gust-Alleviated Airplane

the alleviation system failed or was switched remains one of the engineering problems of
off. In this configuration, the control wheel such systems that no one has yet tried to
on the pilot's side remained connected at all solve.
times to the inboard segment of the elevator
The airplane was instrumented with strain
and to the ailerons. In the gust-alleviation
gauges to measure wing shear and bending
mode, the ailerons were driven symmetri-
moments at two stations and tail shear and
cally through preloaded spring struts. This
bending moment at the root. The project
system remained connected in the manual
became a joint project with the Aircraft
mode, but the pilot could overpower the
Loads Branch. After the tests were com-
forces in the preloaded struts with his inputs
pleted, two reports had been published by
to the control wheel. With the system in the
each group: an initial and a final report on
gust-alleviation mode, the control wheel on
the gust-alleviation characteristics and an
the copilot's side was used to apply control
initial and a final report on the loads
inputs through the electronic control system.
(refs. 13.4, 13.5, 13.6, and 13.7).
In addition, it remained connected through
the mechanical linkage to the inboard por-
The tests occupied a long period of time.
tion of the elevator.
One of the main problems encountered was

When the pilot turned the alleviation system finding suitable rough air. The NACA test

off, the actuator driving the flaps was pilots were understandably conservative in

bypassed, and a separate hydraulic actuator flying experimental airplanes and usually

with its own accumulator forcibly drove the declined to do test flying in clouds or stormy
weather. Clear-air turbulence occurred on
flaps to neutral with a caliper-like linkage
that could capture the flaps in any position. occasions after passage of a cold front. These
conditions were used as often as possible in
Changes in angle of attack due to change in making the tests, but frequently the turbu-
airspeed or drift of the amplifier used to lence was of low intensity. For these reasons,
operate the flaps could have caused the trim some of the data were not as extensive as
position of the flaps to vary slowly in flight. might have been desired.
To maintain the trim position of the flaps at
zero deflection over long periods, a mechani- Despite these problems, results were
cal ball-disk integrator driven by the flap obtained with various sets of gearings to
linkage was used to feed an additional signal obtain varying degrees of longitudinal stabil-
into the flap servomechanism to slowly run ity, and cases with the inboard flaps in neu-
the flaps to the neutral position. This system tral and moving oppositely from the out-
had a time constant of 10 seconds, which board flaps were studied. A time history
was slow enough to avoid interference with comparing the airplane motions in flight
the gust-alleviation function or the control of through rough air with the system on and off

the airplane. To provide automatic control of is shown in figure 13.7. In these runs,
the lateral and directional motion of the air- obtained early in the test program, the

plane in the gust-alleviation mode, a Sperry inboard flap segments were locked.
A-12 autopilot was connected to the aileron
The results of later tests in which the inboard
and rudder systems. This autopilot was the
flap segments moved oppositely from the
most advanced type available at the time of
rest of the flap system are shown in fig-
the tests.
ure 13.8. These results are shown as power
Finally, the pilots landed the airplane with spectral densities of the normal acceleration
the wing flaps in neutral. This operation did and pitching moment plotted on log-log
not pose any problem on the long runways at scales. The results show that the system was
Langley Field, but the design of a high-lift effective in reducing the response in both
flap system that can also provide upward normal acceleration and pitching velocity at
deflection of the flap for gust alleviation frequencies below about 2 hertz. These plots

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 125

Design and Teat of a Gust-Alleviated Airplane

FIGURE 13.7. Comparison

_0 of flights in light
5 turbulence, basic
0 J I airplane, and gust-
> alleviated airplane.
Initial configuration with
center-section flaps in

(a) Basic airplane (top

1 three).
(b) Gust-alleviated
_J 0 t
airplane (bottom


-.2 I I t t t I I I I ] I I I I I I

0 2 4 6 8 10 12 14 16 18 20 22
Time, sec




0 J


-.2 t i i i i i i i i i 1 i i i i i i i i i i i



"_ l0
_'_ 0
_ 0 2 4 6 8 10 12 14 16 18 20 22
Time, sec

of power spectral density are the results of ate the differences, while plotting on log-log
the usual evaluation procedure for randomly paper tends to redu,:e the apparent differ-
varying quantities, but they do not give a ences. The question therefore arises as to
very clear comparison of the results obtained how the data could t e compared to give an
with the alleviation system on and off. The impression of the eff_ ctiveness of the system
power spectra present the square of the more meaningful to t_e user. For this reason,
recorded quantities, which tend to exagger- data on the normal acceleration responses

126 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Design and Test of a Gust-Alleviated Airplane

FIGURE 13.8. Comparison

Basic airplane
of power spectral
..... Alleviated airplane
densities of normal
x 1o-4
acceleration and pitching .10 -- lO--
velocity for basic
airplane and alleviated
airplane. Center-section
flaps deflected oppositely E
from outboard flaps. -_ .010 _ 1.0--

(a) Normal acceleration -_


(b) Pitching velocity "_ .0010 -- --= .10

(right). o .J=

J I i I i I _ i I t I i I
.01 .10 1.0 10 .01 .10 1.0 10

Frequency, Hz Frequency, Hz

FIGURE 13.9. Comparison

of normal acceleration of .22 :_ .22
basic airplane and Basic airplane
Alleviated airplane e
alleviated airplane .20
shown on linear scales on e¢ .20
two different types of oJ_ O

plots. -_ .16 _ .16

(a) Amplitude of
_ .12 _ .12
transfer function of
(n/c(g), or ratio of
_ .08 _ .08
normal acceleration o
to gust angle of /

.04 o .04
attack, as a function s
of frequency (left).
(b) P_ of normal
acceleration, or
2.0 )o I
Frequency, Hz Frequency, Hz
amplitude the
component of
normal acceleration
at each frequency, as
a function of
frequency (right).

are plotted in two different ways in fig- shows correctly the relative magnitude of the
ure 13.9. In part (a) of this figure, results are normal acceleration for the two cases at vari-
plotted on linear scales in the form of a trans- ous frequencies, but does not include the
fer function, that is, the ratio of normal variation of the actual gust forcing function
acceleration to gust angle of attack for sinu- with frequency. In part (b), the square root of
soidal inputs of various frequencies. This the power spectral density of the response is
plot is of interest to the control engineer and plotted as a function of frequency on linear

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 127

Analytical Studies of Additional Problems

scales. This plot attempts to show the actual ation was reduced jast when it was needed
magnitude of the normal acceleration at each most. A later unpuolished investigation of
frequency. This plot is believed to be more the effects of rate limiting demonstrated
meaningful in interpreting the passengers' these adverse effect_ and showed that with
impression of a ride in the airplane. sufficiently severe rate limiting, a continuous
sawtooth oscillation of the flap would be
As can be seen, the basic airplane has a large encountered in rough air, which resulted in
peak in the acceleration response at a fre- response greater than that of the basic
quency about 0.2 hertz. This low-frequency airplane.
peak is typical of the response of unallevi-
ated airplanes and occurs because of the A second problem was the nonlinear lift
larger amplitude of the turbulence input at characteristics of the flaps. The flaps
low frequencies. In the case of the alleviated deflected about plus or minus 25 degrees, but
airplane, the response is reduced to a fairly the slope of the curve of lift versus flap
constant, relatively low value in the fre- deflection fell off markedly before this
quency range between 0 and 1.6 hertz. For deflection was reached. As a result, a fixed
the degree of turbulence encountered, this gain between the vane and the flaps did not
reduction greatly improved the subjective give a uniform value of the ratio of lift to
impression of ride comfort. Though the gust angle of attack. This nonlinear response
results are not shown on the plots, the would be expected to introduce higher fre-
response at frequencies above 2 hertz with quency harmonics into the response and
the alleviation system operating were probably accounts for the increase in
slightly increased above those of the basic response beyond 2 hertz. This and other
airplane. Also, in turbulence of relatively characteristics were not known before the
large magnitude, the pilots noted a fore-and- tests were made. I concluded that on any
aft oscillation caused by drag of the flaps at future project of this type, wind-tunnel tests
large deflections. of the airplane to determine the control char-
acteristics would be advisable. Despite these
Though the results obtained were quite grati- deficiencies, enough was learned to show the
fying, the question arises as to why the per- feasibility of gust alleviation and to show
formance of the system was not better, inas- how the results could be improved in a future
much as the theory predicted perfect gust attempt.
alleviation. One of the main reasons was the
nonlinear characteristics of the servomecha-
nism used to drive the flap system. The
designer of the system in the Instrument Analytical Studies of
Research Division had been requested to Additional Problems
provide a rather sharp cutoff in the response
beyond 2 hertz, to avoid the possibility of The studies of gust alleviation introduced
exciting wing flutter. The C-45 was a very several problems th it had not been consid-
stiff airplane, with the wing primary bending ered previously in ai-plane design. One prob-
mode at 8 hertz. It was intended, therefore, lem was the effect ol variations in gust veloc-
that the output of the servomechanism ity across the wing ,,.pan. The use of a single
should be close to zero at a frequency of vane on the center line of the airplane to
8 hertz. It was not until after the device was sense the gusts would work perfectly if the
installed that it was found that the cutoff in gust velocity were constant across the wing
response had been obtained by rate limiting span, but would be ess effective if different
the output. This provided a steep cutoff, but values of gust velo_ ity were encountered at
the response was a function of amplitude. different stations alc,ng the span. This prob-
The result was that in a large amplitude gust, lem can be studied if the turbulence in the
rate limiting was encountered and the allevi- atmosphere is assu[red to be isotropic (or for

128 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Future Possibilities of Gust Alleviation

this application, axisymmetric); that is, it has about 98 percent alleviation was obtained by
the same characteristics regardless of the using a second-order linear filter with rea-
direction in which the airplane is flying. This sonable frequency and 0.7 critical damping
assumption appears reasonable for most (refs. 13.6 and 13.7).
types of turbulence. The theory of isotropic
Some time after completion of the tests, a
turbulence had been studied theoretically.
summary report was given as part of a lec-
The nature of the spectrum of turbulence,
ture series at Renssellaer Polytechnic Insti-
that is, the way the gust velocity varies with
gust wavelength, had been determined. The tute (ref. 13.10). This paper gives a more
complete discussion of the subject of gust
gust velocity has been found experimentally
alleviation than contained herein.
to vary approximately directly with the
wavelength even for wavelengths many
times larger than the wing span of the largest
existing airplanes. The most intense gusts, Future PosMbilities of
which cause the most disturbance to the air-
Gust Alleviation
plane, have wavelengths long compared to
the wing span and are therefore approxi- Some review of later efforts in the field of
mately constant across the span. Gusts with
gust alleviation may be of interest, inasmuch
wavelengths short compared to the span have
as the development of computers and auto-
low intensity and therefore do not disturb the
matic control technology would permit
airplane much. The use of a single vane on
approaches quite different from that used in
the center line is therefore quite effective.
the early NACA tests. It was a disappoint-
Gusts of wavelength short compared to the
ment to me that very little effort was made
span may cause up and down loads that aver-
by aircraft companies to incorporate provi-
age out across the span. For these gusts, the sion for gust alleviation even after the devel-
response of the vane at the center line would
opment of control technology would have
be too large. These gusts have such high fre- made it more feasible.
quency, however, that the vane response is
filtered by the lag in response of the flap ser- To my knowledge, the only airplane in
vomechanism. For a vane located ahead of service that incorporates a system perform-
the nose, the lag in the flap servomechansm ing some gust-alleviation function is the
is also beneficial in delaying the response Lockheed 1011. In some later models, the
sensed by the vane until the gust reaches the wing span was increased by extending the
wing. Considering these factors, the effec- wing tips to allow the airplane to carry
tiveness of a gust-alleviation system using a greater loads. To avoid changing the wing
vane on the centerline may be shown to be structure to withstand greater bending
about 98 percent as effective in axisymmetric moments, the ailerons were operated sym-
turbulence as it would be with gusts constant metrically by an automatic control system to
across the span. This problem was studied in reduce bending moments due to gusts.
more detail after the flight investigation was
completed (refs. 13.8 and 13.9). One reason for the lack of interest in gust
alleviation is that following the NACA tests,
An interesting optimization problem, which jet transports were introduced. As a result of
to my knowledge has not been solved, is higher wing loading, swept wings, flight at
to determine the optimal filter to place higher altitudes, and the use of weather radar
between the vane and the flap to obtain to avoid storms, these airplanes were much
the greatest amount of gust alleviation, con- less likely to encounter violent airplane
sidering the distance of the vane ahead of motions that would cause airsickness. In
the wing, the wing span, and the spectrum addition, the problem of gust alleviation
of atmospheric turbulence. This problem is became more difficult because the structural
only of academic importance, however, as flexibility of these airplanes placed their

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 129

Review of Work by Rend Hirsch

structural frequencies closer to the frequency about their chordwise hinges to cause the
range of interest for gust-alleviation. As a flaps to move in the direction to provide
result, structural response would have to be direct lift control. In this way, the loss of lon-
considered in designing the system. In recent gitudinal control due to the gust-alleviation
years, these reasons for avoiding the use of system was overcome. Hirsch's airplane
gust-alleviation systems have become less incorporated many other ingenious features,
significant. Extensive use is now made of including provisions for reducing rolling
commuter airplanes that fly at lower altitudes moments due to rolling gusts and lift due to
and frequently encounter rough air. In addi- horizontal gusts. His design also incorpo-
tion, methods have been developed to ana- rated large pneumatic servos operated by
lyze the structural response and to damp out dynamic pressure to restore damping in roll
the structural modes by use of automatic and to stabilize the rate of climb or descent.
control systems. The airplane had good handling qualities and
appeared to have been very successful in
providing a smooth ride, as shown by some
time histories in rough air with the system
Review of Work by turned on and off. After about 30 flights, the
Ren_ Hirsch airplane ran into a ditch at the end of the run-
way and was damaged.
In closing, a brief review is given of the work
of Rend Hirsch, whose thesis was mentioned No more was heard until 1967, when another
at the beginning of this chapter. Also, a few report appeared showing that the airplane
programs and studies applicable to gust alle- had been rebuilt and equipped with two
viation that have occurred since the NACA 180-horsepower motors (ref. 13.12). In this
program on the C-45 are reviewed. condition, the airplane made numerous addi-
tional flights with somewhat more complete
After reviewing Hirsch's thesis, no more was instrumentation. A photograph of the air-
heard of his activity until the early 1950's, plane in flight is shown in figure 13.10. From
during the course of the NACA program. At the data obtained, the results appeared very
this time, a French report was discovered similar to those obtained with the NACA
revealing that, following WW II, Hirsch had C-45 in that the accelerations due to gusts
made additional wind-tunnel tests in the were reduced by about 60 percent at frequen-
French large-scale tunnel at Chalais-Meudon cies below about 2 hertz, but were increased
on a model of a proposed airplane and had somewhat at higher frequencies.
built this small twin-engine airplane incorpo-
rating his system. The airplane was envi- All of Hirsch's work in designing and build-
sioned as a quarter-scale model of a piston- ing his airplanes was done with his own
engine transport of a class similar to the funds, though some help with instrumenta-
Douglas DC-6 or the Lockheed Constella- tion was obtained from ONERA, the French
tion, which were the largest transports in ser- equivalent of the N,_ CA. Outside of France,
vice in that period. Correspondence was Hirsch's work was 1 ttle known. His reports
established with Hirsch and additional described the work in general, but were not
reports and information were obtained sufficiently detailed ,o give engineering data
(ref. 13.11). Hirsch's airplane had a wing on all of the ingemous ideas and systems
span of 27 feet and had two 100-horsepower incorporated on his airplanes.
motors. It incorporated the same system
described in his thesis, in which the halves of In 1975 during a uip to France, I visited
the horizontal tail moved on chordwise Hirsch. He had foun( at that time that a twin-
hinges to operate flaps on the wings. The engine airplane was _oo expensive for him to
conventional elevators provided not only operate during the oil crisis, and he had
pitching moments, but moved the tail halves donated it to the Freach Air Museum. When

130 Monographs in Aerospace I4istory Number 12--Journey in Aeronautical Research

Review of Work by Rend Hirsch

FIGURE 13.10, Flight

photograph of Ren_
Hirsch's first airplane as
equipped with larger

FIGURE 13.11. Ren_

Hirsch in front of his
Aerospatiai Rallye light
airplane equipped with
gust-alleviation system.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 131

Later Studies by Other Investigators

I saw the airplane, it stood like a little jewel Air Force Base, Ohio, about 1968-1969
amid a group of dilapidated antique airplanes (ref. 13.13). In this project, a B-52 bomber
in an old WW I hangar at Villaroche. It is was equipped with an electronic analog-type
now on display at the new French Air flight control system to operate the existing
Museum at Le Bourguet. flight controls to damp out structural modes.
This work is important because consider-
At that time, Hirsch was starting modi- ation of damping of structural modes would
fication of a single-engine light plane, the be required in any a:tempt to install a gust-
Aerospatial Rallye, to incorporate his gust- alleviation system in a high-speed airplane.
alleviation system. This airplane was com- The report illustrates the success of modem
pleted and flying when I visited France again control analysis techniques (as they existed
in 1980. Hirsch is shown standing in front of at that time) in designing a modal damping
his Rallye airplane in figure 13.11. This air- system and in predicting the results obtained.
plane was never as successful, in Hirsch's
opinion, as the first one, probably because of The second contribution of note is the analyt-
its lower airspeed and lower frequency of ical work of Dr. Edmund G. Rynaski, for-
response of the flap systems. In recent years merly of Calspan and now at EGR Associ-
(1995), Hirsch modified a third airplane, the ates, in designing a system to alleviate both
Sobata Trinidad, with small canard surfaces the rigid-body motions and selected struc-
ahead of the wing root to operate the flaps. turai modes of an airplane (ref. 13.14).
The more forward position of these sensing Rynaski's work, based on matrix analysis
surfaces was intended to improve the techniques, shows how to provide essentially
response to high-frequency gusts. Hirsch open-loop control of the rigid-body modes,
died in August 1995 at the age of 87 without as was done on the ( -45 airplane, as well as
having had the opportunity to test his latest to provide open-loop cancellation of a
design. selected number of structural modes. This
method also makes p,)ssible improved damp-
Hirsch's dedication to the pursuit of gust ing of higher order structural modes by use
alleviation is a remarkable story in view of of closed-loop control.
the general disregard of this subject by the
rest of the aviation industry. Of course, the With the availability of digital flight comput-
aeromechanical systems used by Hirsch have ers and modem control actuators, different
been superseded by automatic controls using approaches should be considered for gust
computers and electrohydraulic actuators. alleviation. One approach would be to oper-
Hirsch readily admitted that he would prefer ate the flaps on the wing as a function of
such systems but was unable to afford them. angle of attack sensed by a vane or similar
device, but to operate the elevators by a mod-
em reliable pitch damper as part of a longitu-
dinal command control system to control the
Later Studies by Other pitching response ol the airplane. Another
Investigators approach would be tc calculate absolute gust
velocity on line by a method similar to that
Though a number of studies of gust allevia- referred to previously, in the work by Crane
tion have been made during the years since and Chilton in m_asuring gust velocity
the C-45 tests, most of them have not con- (ref. 12.5). This met!_od requires correcting
tributed any notable new developments. Only the angle of attack vleasured by a vane for
two are mentioned to bring the subject up to the inertial motions of the airplane at the
date. One is the so-called LAMS project, an vane location. This ,,ignal could be used as
acronym for Load Alleviation and Mode Sta- an input into a gust-; lleviation system with-
bilization, conducted at the Air Force Flight out the need to modily the normal control or
Dynamics Laboratory at Wright-Patterson handling qualities of _.he airplane.

132 Monographs in Aerospace History Number 12 Journey in Aeronautical Research

CHAPTER 14 Lateral Response to Random

The field of flight dynamics has been studied The method of calculating lateral response of
since the early days of aviation. As men- an airplane to discrete gusts was the subject
tioned previously, G. H. Bryan in England of a report by R. T. Jones (ref. 5.3). He
worked out the complete equations of air- showed that the conventional equations of
plane motion the same year as the Wright motion could be used to calculate the
brothers' first flight (ref 4.1). By the 1950's, response to a rolling gust by assuming that a
it was difficult to find any field of research in rotational motion of the atmosphere about
fight dynamics that had not been thoroughly the longitudinal axis acted on the airplane
explored. The gust-alleviation program, through the so-called stability derivatives
however, revealed one problem that had associated with rolling of the airplane, that
received very little attention. This problem is, the rolling moment due to rolling velocity
was the lateral response (that is, rolling and and the yawing moment due to rolling veloc-
yawing response) due to flight through atmo- ity. These derivatives were already used in
spheric turbulence. the standard lateral stability equations and
were well known as a result of previous theo-
retical and experimental research. The effect
of a rotational motion of the atmosphere
about a vertical axis could likewise be calcu-
Modeling the
lated with the aid of the yawing moment
derivatives. What was needed was a way to
determine the rolling and yawing motions of
As mentioned in the section on gust allevia- the atmosphere associated with random tur-
tion, the scale of turbulence in the atmo- bulence.
sphere is ordinarily so large that the gusts
may be considered constant across the wing The turbulence in the atmosphere was ordi-
span. This circumstance made it possible to narily characterized by what is known as the
use a single vane in the nose of the airplane point spectrum of turbulence. This spectrum
to sense the gusts with the possibility (in the- is what would be obtained by measuring the
try at least) of getting 98 percent complete velocity fluctuations along a straight line as
alleviation. In considering lateral response, it is traversed at a relatively high speed.
however, such an assumption is obviously Measurements of the point spectrum of verti-
incorrect, because if the gusts were constant cal velocity, for example, were made by
across the span, the airplane would receive recording the motion of a vane to measured
no rolling disturbance due to turbulence. angle of attack and correcting the readings

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 133

Calculation of Response

continuously for the inertial motion of the of the airplane response and added the
airplane at the location of the vane. This effects of the fuselage and tail to those of the
method was used in the measurements by wing by a method that I suggested
H. L. Crane and R. G. Chilton described pre- (ref. 14.5). Eggleston calculated the effects
viously (ref. 12.5). of the rolling moments, yawing moments,
and side forces separately and then added
To extend this information to obtain velocity
these contributions to get the response spec-
fluctuations across the wing span, the usual
trum of the resulting motion. An important
assumption made is that the turbulence is consideration in these calculations is that the
isotropic; that is, the spectrum of turbulence
lateral motion is only statistically, rather than
is the same no matter what direction the
uniquely, determined in flight through any
atmosphere is traversed. It seems reasonable,
given velocity pattern along the fuselage
for example, that an airplane flying through a
centerline because different velocity distribu-
region of turbulence from north to south
tions across the span may be encountered for
would experience the same amount of distur-
a given distributioq along the fuselage
bance as an airplane flying through the same centerline. The calculations can therefore
region from east to west. Using the assump-
produce only the power spectrum of the
tion of isotropy, H. S. Ribner presented an
response for a given point power spectrum of
analysis of response to turbulence using a turbulence.
two-dimensional Fourier series to represent
the disturbances (ref. 14.1). This approach I continued my analysis to determine the
was also described in a textbook by Bernard relation between the spectra of effective roll-
Etkin (ref. 14.2). Franklin W. Diederich, in a ing gusts and yawing gusts and the point
doctoral thesis at California Institute of spectrum of turbulence. Again, the results
Technology under R. W. Liepmann, calcu- of Eggleston and Diederich's calculations
lated the spectrum of lift on a wing flying (ref. 14.4) were used, but the effective atmo-
through turbulence by using a correlation spheric roiling and yawing motions were
function approach. This thesis was later pub- calculated and applied directly as inputs to
lished as an NACA Technical Report the lateral equations of motion. I worked
(ref. 14.3). Diederich also showed how to out some of the same examples used by
calculate rolling and yawing moments on the Eggleston. Much to my surprise, the results
wing, but the complete calculations were not turned out to be identical to those of
carried out. I considered the approach used Eggleston. These results were obtained with
by Diederich to be easier to comprehend much shorter calculations, and to me, gave a
than that used by Ribner, but both should clearer picture of the physical process
give the same answer if used correctly. involved (ref. 14.6). Later, a technical report
was published presenting both Eggleston's
matrix calculations _d my shorter method
(ref. 14.5) together with a simple 45-step
Calculation of
procedure for calculating the power spectra
Response of the lateral responses for specific cases.

I assigned an engineer in my division, The relation between the spectrum of rolling

John M. Eggleston, to work with Diederich gusts and the point _pectrum of turbulence
and complete the calculations of yawing and used in the analysis s shown in figure 14.1.
rolling moments (ref. 14.4). These calcula- Ordinarily, a power spectrum is presented as
tions were made for wings of rectangular, the square of the del ¢ndent variable plotted
elliptic, parabolic, and triangular planform against the independent variable on log-log
and involved rather complex mathematical paper. This type of l:lot is shown in part (a)
manipulations. Eggleston continued this of the figure. A squared variable is compati-
work to make a complete matrix formulation ble with the statisti:al significance of the

134 Monographs in Aerospace history Number 12--Journey in Aeronautical Research

Calculation of Response

FIGURE 14.1. Relations

between the power I I IIIIIII IIIIll
I I I,,,,,, I
, I, IIi
lllll i iiJii
spectrum of rolling gusts Iltltl
and the point spectrum
- Ratio
of span
I : :::::
to scale of turbulence, 13' = b/L [ i [ ii
[tlltl Ii
of turbulence for various I IIl[I I
-13= 1.0 ......
values of the ratio of I I I I I ill

scale of turbulence to f
wing span. 100
i i _iiiii I ]]I]I i ill _
n n i I IIII I]]'" III
(a) Ratio of power I ]]1]1 !
-- .25 ...... I III]1 i!j:
spectrum of rolling --, ! !!!!! I]]111
gusts to point IIIIII
spectrum of .125 ....... i:j
turbulence versus
10-1 I
reduced frequency I I I!!!! I lllII i I
i i ::::: IIII '_,,
as conventionally Z ::::: L IIIII
.0625 .....
plotted on log-log IIIIII
IIII1[ ! :
scales (top). i i , i i iii

(b) Square root of .03125; .....
10 -2
(D¢g/pg) 2as a
10-3 10-2 10 -1 10 0 101 102
function of
Reduced frequency, ¢o' = o)b/U
frequency plotted on
linear scales

2.0 (- DCg b
/ _g - 2g _-

1.6 3' = b/L ,'

_ ----.-.......__.__



.4 , I
4' .9_6?
J/ .03125

0 .2 .4 .6 .8 1.0 1.2 1.4 1.6 1.8 2.0

Ratio of span to wavelength, b/'L

quantity in which only the amplitude and not function of reduced frequency on linear
the phase relationship is obtained from the scales is shown in part (b) of the figure. This
analysis. Log-log paper is useful in showing plot shows the expected results that at very
a large range of the variables. This method of long wavelengths, the amplitude of the effec-
presentation, however, obscures a clear tive rolling gust is small because the gradient
impression of the actual shapes of the curves. of vertical velocity along the span is small.
The same quantities shown by taking the At high frequencies, where the wavelength is
square root of the ordinate and plotting it as a small compared to the span, the effective

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 135

Calculation of Response

rolling gust is again decreased because the high-speed computers were just becoming
fluctuations are averaged along the span. practical and when closed-form solutions of
Between, the effects reach a maximum. In a complex problems were becoming less nec-
region indicated by the sloping dotted line, essary. As a result, his work to produce
the disturbances approach an approximation widely useful charts and formulas appears to
obtained by assuming a uniform gradient be largely forgotten, even by engineers
across the span equal to the slope of the cen- working in the fields to which he made such
tral region of a sinusoidal gust with a wave- valuable contributions. Now, when similar
length corresponding to the frequency. The problems are encountered, a computer simu-
only rather puzzling aspect of the curve is the lation of the specific problem under consid-
slight tendency of the curves to turn up at eration is made with such techniques as
very low frequencies. Possibly this effect is finite-element methods or step-by-step solu-
caused by rare encounters with widely tions. The programs for making such calcu-
spaced large-amplitude rolling gusts, first in lations are generall) available to the engi-
one direction and later in the other, which neers involved. Though some familiarity
would introduce a harmonic of very low fre- with the physics of the problems is advis-
quency in the spectrum. able, a detailed knowledge of classical
theorems from aer¢_dynamic theory or of
At the time these studies were made, about
methods of manipulating complicated math-
1958, I wanted to conduct some flight tests to
ematical functions is usually not required.
verify the predicted relationship between the
measured point spectrum of turbulence and
The work on response to random turbulence,
the resulting lateral response of the airplane.
including the subject of gust alleviation, has
These methods might have also provided an
also received little attention from the aero-
alternate method to measure the spectrum of
nautical industry bec rose the main structural
atmospheric turbulence. Unfortunately, flight
design conditions on airplanes come from
testing of high-speed airplanes at Langley
rare encounters with discrete gusts in thun-
was discontinued at that time. To my knowl-
derstorms. Such events have been studied
edge, such flight tests have never been car-
through operational experience rather than
fled out, and they remain one of the few
by research programs designed to study the
uncompleted phases of flight dynamics.
details of response to turbulence. As a result,
The work of Franklin W. Diederich at much very interestin.; research work on this
Langley deserves special mention. His work, subject now lies bltried in old technical
which involved both aeroelasticity and reports that are occ tsionally unearthed by
response to turbulence, came at a time when scholars, but rarely u: ed in practice.

136 Monographs in Aerospace H story Number 12--Journey in Aeronautical Research

CHAPTER 15 Accident Investigations

The NACA and NASA were not routinely true airspeed, required for navigation pur-
called in to give advice on all accident inves- poses, is greater than the indicated airspeed
tigations, but in some cases in which the at altitudes above sea level because the den-
cause of a crash could not be readily deter- sity decreases with increasing altitude.
mined or when a new experimental design
The abbreviation MAC stands for mean aero-
was involved, the appropriate government
dynamic chord, a reference line based on the
organizations requested assistance. Acci-
wing plan form that is used in deriving cen-
dent investigations are important not only to
ter of gravity location and in computing sta-
correct defects in existing designs that might
bility derivatives.
cause further accidents, but also to provide
information for safer designs in the future. I
was directly involved in a number of these
investigations, some of which are sufficiently BAC-11 i Crash Near --
interesting to include in this autobiography.
Omaha, Nebraska, on
In this discussion, the term "dynamic pres- August 6, 1966
sure" is used. The dynamic pressure is
approximately the increase of pressure above From time to time, in investigations of
the surrounding atmospheric pressure crashes of civil aircraft, the NACA or NASA
caused by the flow impinging on a flat sur- were contacted by the Civil Aeronautics
face normal to the airstream. It is given by Board (CAB), later called the National
the formula one half times density times the Transportation Safety Board (NTSB), to
square of true airspeed. Such a value of pres- assist with accident investigations. One of
sure occurs at the nose of an open-ended these accidents that involved considerable
tube facing into the airstream, called a pitot study on my part was the crash of a
tube, and is used to measure airspeed as well BAC-111. This airplane was a small twin-
as to determine loads on the airplane caused engine jet transport, constructed by the
by the airflow. The term "indicated air- British Aircraft Corporation (BAC), that was
speed" is the airspeed shown by a standard used rather extensively on short-haul routes
airspeed meter when it is supplied with this in the United States (figure 15.1). In the
pressure and a reference pressure called the course of the study, I attended a meeting with
static pressure, which is the true ambient the British officials in England and other
pressure in undisturbed air. The instrument meetings in various locations in the United
is calibrated to read true airspeed for the States, including Omaha, Nebraska, and
density at standard sea level conditions. The Cleveland, Ohio.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 137

BAC-111 Crash Near Omaha, Nebraska, on August 6, 1966

FIGURE 15.1. The BAC-

111 Transport Airplane.

The crash occurred when the airplane, which broke up in normal flight without warning.
was flying at 5000 feet altitude (4000 feet Later, the CAB representative requested cop-
above the terrain) and carrying 38 passengers ies of the computer program to use in their
and 4 crew members on a scheduled Braniff analyses of other crashes.
Airways, Inc. flight, burst into flames in the
All possible causes of the crash were investi-
air and crashed in a field in a flat, upright
gated, both by BAC and by the CAB. These
attitude with no forward motion. As is usual
causes included such things as failure of the
in such crashes, there was little information
control system or feel devices, use of
available to determine the cause of the crash.
improper construction materials, use of
One of my first duties was to attempt to incorrect design procedures or requirements,
determine, from the locations of various and fire due to fuel or hydraulic leaks. A con-
pieces of wreckage around the crash site, the tributing factor involved in many studies was
approximate altitude and airspeed of the sec- that the airplane was flying through or above
tions at the time of breakup in the air. This a roll cloud when it burst into flames and was
kind of analysis requires calculating the tra- a few miles away from an approaching line
jectories of the various pieces as affected by of heavy thunderstor-ns.
their weight, drag, and surface area. In this
The cockpit voice lecorder was recovered,
case, the horizontal and vertical tail unit (a
but the only information on it was various
T-tail arrangement) and an outer wing panel
noises, such as a rushing sound of air follow-
were separated from the main wreckage. At
ing the breakup and a sounding of various
that time, digital computers were available,
warning horns durin:; the descent. There was
but I had little experience in using them. I no crash recorder su,'h as is now required on
used a program called the Mimic program, commercial airplanes to record airplane
which allowed the analysis to be conducted
motions or control p _sitions.
with the aid of a block diagram similar to
that which would be used in making an ana- In listening to the record on the voice
log computer solution and which had an recorder, it was no,ted that following the
unusually clear manual to describe its use. breakup, there was noticeable variation in
The solutions came out in good agreement pitch or "wow" as it is called by audio enthu-
with the expected altitude and speed of the siasts. I conceived the idea that the records
airplane, which indicated that the airplane obtained might be used to get some idea of

138 Monographs in Aerospace l tistory Number 12--Journey in Aeronautical Research

BAC-111 Crash Near Omaha, Nebraska, on August 6, 1966

FIGURE 15.2. Rolling

Accident sequence
acceleration and rolling ........ Three preceding cycles
velocity of BAC-I 11
airplane as determined "Rushing air" noise Start of word "Okay"
from voice recorder /

-6;=_10 I I i I I J I i i I I J I i I i I i i I
,.¢ 0.5 1.0 1.5 2.0
Time, sec


'_ 100 --


200 -

250 I I i I I I I I I I I I J I J i J i I
0 0.5 1.0 1.5 2,0
Time, sec

the motions of the airplane. The recorder had To simulate the maneuvers following the
a flywheel in the tape drive that had its axis breakup, engineers at the Langley 20-Foot
aligned with the longitudinal axis of the air- Vertical Spin Tunnel had a crude model of
plane. As a result, the speed of the flywheel the airplane built with the tail assembly and
would be affected by rolling accelerations of part of the left wing missing, as found on the
the airplane. A similar voice recorder was wreckage (figure 15.3). This model was
obtained for calibration tests in the NASA launched by a catapult from the top of
Instrument Research Division. The record the Langley Lunar Landing Facility (now
obtained from the accident had a faint Langley Impact Dynamics Research Facil-
800 hertz tone, probably a harmonic of the ity), a girder structure 250-feet high, which
aircraft power supply, that formed an excel- for the scale of the model used, corresponded
lent time base. I laboriously analyzed the closely to an altitude of 4000 feet. The model
data, one cycle at a time, to determine the would go into violent tumbling motion, but
movement of the tape and was able to plot a before hitting the ground would settle into a
record of rolling acceleration as a function of fiat descent, sometimes without rotating and
time. This record was then integrated to sometimes in a slow flat spin. The model
determine rolling velocity. The data showed always landed fight side up, probably
an abnormally rapid rolling velocity to the because of the wing dihedral.

left building up to over 100 degrees per sec- Another interesting phase of the investiga-
ond in less than 1 second following the start tion was a number of lectures by a well-
of the "rushing air" noise (figure 15.2). known meteorologist, Dr. Tetsuya Fujita of
Though this record was interesting, it could the University of Chicago, who discussed
not be relied on entirely because the tape mesoscale turbulence and who had coined
speed was also affected by large values of the term "downburst" to describe the down-
linear acceleration. ward rush of air cooled by heavy rain in a

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 139

Crash of Convair B-58 Bomber

FIGURE 15.3. Model of

damaged BAC-111 used
for catapult tests to
determine motion
following breakup.

thunderstorm. He pointed out that this air clusions were probably correct because no
would spread out on the ground and would similar crash of a BAC-111 ever occurred. A
advance at a high speed several miles ahead complete copy of the National Safety Board
of the main storm, which would result in roll report was published in Aviation Week
clouds or in heavy clear-air turbulence. (ref. 15.1).

The conclusion of the accident board was I proposed a wind-tunnel study of turbulence
that there was no fault in the design or con- of the type encountered in the accident, by
struction of the airplane and that the breakup using a large downward jet of air directed at
had been caused by an abrupt gust with the floor of a wind Lunnel operating at very
velocity well beyond the design require- low speed. This study was never made. I
ments. Members of the BAC calculated that have always wondeled whether the airplane
a gust of 144 feet per second angled at would have broken up flying anywhere into
45 degrees upward and 45 degrees from the the storm front at the altitude and vicinity of
right would have been required to break off the crash, or whether the gust was a single,
the T-tail in the manner that occurred in the isolated phenomenon with very small proba-
accident, whereas the design gust velocity at bility of being encountered by an airplane.
that time was 88 feet per second. Following
loss of the tail, the airplane would immedi-
ately pitch down to large negative values of
Crash of Conv,_ir B-58
acceleration, which would cause the outer
wing panel to break off. Bomber

The official accident investigation concluded The Air Force started a program about 1952
that the airplane met all design requirements to develop a supers,mic bomber. The result
and that no changes in design were required. was the Convair B-:;8 Hustler, a delta-wing
Warnings were issued concerning the possi- aircraft with four engines mounted in
bility of high turbulence at low altitudes nacelles under the Ning (figure 15.4). The
several miles from heavy thunderstorms. airplane was first flo_vn in 1956. A large pod,
Subsequent experience showed that the con- almost as large as t le fuselage itself, could

140 Monographs in Aerospace ltistory Number 12 Journey in Aeronautical Research

Crash of Convair B-58 Bomber

FIGURE 15.4. B-58

Hustler bomber airplane.

be carried under the fuselage. This pod was mittee, known as the SAE Aerospace Control
intended to be developed into a large nuclear and Guidance Systems Committee. This
weapon or a carrier for smaller munitions. committee was a remarkably informative
The B-58 was a remarkable airplane in many organization of the top control engineers of
ways in that it had much larger range, various companies, equipment manufactur-
installed power, and load-carrying ability ers, and universities. Meetings were held
than any supersonic airplane developed four times a year, usually at very attractive
before that time. By 1960, about 13 airplanes locations. The members had very few inhibi-
had been built. It is not surprising that, with tions on discussing their companies' latest
such an advanced design, problems would be projects because it was realized that a free
encountered in its development. By 1960, interchange of ideas would stimulate devel-
however, five airplanes had been destroyed in opment. Another feature that encouraged
accidents, and the Air Force was engaged in attendance was that no formal proceedings
a frantic effort to increase the safety of the were published. Instead, each speaker
vehicle and its operation. brought enough copies of his presentation to
a meeting so that each member could have
Of the accidents that occurred, the first four one. As a result, the attendees immediately
involved such unrelated causes as a ground had a stack of useful information on the lat-
fire during fuel transfer, pilot error, a crash est developments.
due to tire failure, and structural defects. The
fifth accident, however, occurred in flight Two of the men prominent in the SAE com-
while an experienced test pilot was testing mittee who assumed leadership roles in the
the engine control system. The airplane went accident investigation were Duane McRuer
and Irving Ashkenas, officials of a small
to a large angle of sideslip and disintegrated
in the air. This event might indicate a serious company called STI (Systems Technology
defect in the design. The Air Force set up Incorporated) whose business was studying
their own investigating board and also set up stability problems of airplanes. I was
a joint Air Force-industry board, of which I assigned the job of investigating the longitu-
was a member. dinal stability, and Irving Ashkenas headed
the study of lateral and directional stability.
Several of the board members were men I The Convair Corporation at Fort Worth,
had worked with previously on the Society where the board met, furnished computing
of Automotive Engineers (SAE) A-18 com- facilities to aid in the investigation.

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 141

Crash of Convair B-58 Bomber

I was very impressed by the effectiveness of As it turned out, the lateral-directional sys-
this small group of experts, who knew tem soon came under suspicion as the cause
exactly how to proceed with investigating the of the accident. All the quantities sensed by
problems encountered in the accident. I the autopilot were fed to the control system
thought how desirable it would be to have by a control law that required changing gains
such a capable group assigned as my own as a function of flight condition. These gains
engineers for conducting the projects in my were varied by servo-driven potentiometers,
division. Of course, such an arrangement or pots, in a manner similar to that used in
would not be possible because the group analog computers of the time. Investigation
consisted of the top men from each company of the wreckage showed that one of the pots
or organization, each one of whom normally in the autopilot was against its stop, which
supervised a group of less experienced indicated that the electric servo driving the
pot had probably experienced a hard-over
signal. Normally, it would be expected that
The proceedings of this board were classified
such a failure would change the handling of
secret. As a result, I was not allowed to take the airplane somewhat, but would not cause a
out any notes or documents. My account of catastrophic failure. In this case, however,
this investigation is therefore given entirely when the pot was driven against its stop, the
from memory and may be subject to errors. I
wiper on the pot ran off the end of the resis-
will attempt, however, to give a correct gen-
tance element and produced zero output. As
eral impression of the investigation.
a result, the gain of this particular feedback
signal was zero. Stability calculations
The emphasis, of course, was on the control
showed that with this feedback gain at zero
system of the airplane. This control system
and the others at their normal values, the air-
had been designed by the Eclipse-Pioneer
plane would have a highly unstable lateral, or
Division of the Bendix Corporation and
Dutch roll oscillation. Furthermore, with this
incorporated many novel features. An
particular combination of gains, the normal
attempt was made to place the various sub-
control actions of the pilot in attempting to
systems, such as the air data system, the
keep the wings level would further destabi-
gyroscope and accelerometer unit, the ampli-
lize the oscillation. It was therefore con-
fier and computer assembly, the engine con-
cluded that when tl_e failure occurred, the
trois, and the autopilot into an integrated sys-
airplane went into a divergent lateral oscilla-
tem. The various systems all came together
tion and reached sidtslip angles sufficient to
in the power control linkage assembly or
break off the tall.
PCLA. This complex mechanism, which fit
into a space about 2-feet wide by 3-feet long,
contained all the equipment for receiving the The findings of the board were undoubtedly
pilot's inputs and the autopilot signals and transmitted to the proper officials in the Air
provided control feel, ratio changing, com- Force, and correction1 of the problem might
mands to the control surfaces, and various involve simply limiti lg the travel of the ser-
safety and monitoring provisions. Usually, vos driving the pots to avoid exceeding the
such functions are distributed throughout the ranges required in flight. This accident, how-
airplane, and the complexity of this tightly ever, illustrates the problems that caused
packed unit must have caused a nightmare great reluctance of ai_lane manufacturers to
for maintenance. In fact, as I recall, the unit adopt electronic controls for airplanes and
was intended to be returned to the factory for that required 20 yea_ s or more before com-
maintenance. In accordance with the state of mercial airplane maaufacturers would use
the art in that time period, the system used all such devices in pri nary control systems,
hydraulic power controls and all analog despite the improved capabilities and lower
computation. cost and weight of tht:se systems.

142 Monographs in Aerospace History Number 12--Journey in Aeronautical Research

Crash of Convair B-58 Bomber

For many years, the design of each new air- occur in going through the transonic speed
plane had been supervised by a chief engi- range. Of course, this feature means that the
neer, who was sufficiently familiar with all airplane has no speed stability. Normally, a
elements of airplane design and construction statically stable airplane without an autopilot
that he could understand the operation of can be trimmed to fly at a given airspeed and,
these systems and could bring his long expe- because of the stable variation of pitching
rience to bear in avoiding dangerous fea- moment with angle of attack, it will tend to
tures. As soon as electronic control systems return to the angle of attack corresponding to
were introduced, the operation of these sys- the trim condition. The B-58 did not have
tems was beyond the knowledge of the chief this capability, but this problem was over-
engineer. In the case of the B-58, a company come by use of the autopilot system to auto-
skilled in autopilot design was given respon- matically hold any desired value of airspeed.
sibility for the design of the control system.
The electrical and electronic engineers, how- One of the requirements in Gilruth's require-
ever, did not have long experience with the ments for satisfactory handling qualities is
safety problems of airplanes or of the possi- that an airplane should require an increase in
ble catastrophic effects of a very minor fail- pull force on the control stick to reduce the
ure. In my experience in flight research, I speed to the stall. This feature is stated to
worked with airborne radar equipment that provide a valuable form of stall warning. In
employed several hundred vacuum tubes. the case of the B-58 in a flight in icing condi-
Usually two or more tubes would have to be tions, the airspeed head iced up so that the
replaced after a few flights, yet the designers pilot had no correct instrument indication of
of this system considered it to be highly reli- airspeed. Also, a delta-wing airplane like the
able. Later, more extensive studies of reli- B-58 has a gradual stall with no buffeting
ability requirements for control systems until large values of angle of attack are
were made. The basis of these studies were reached. The result was that the B-58 gradu-
that no failure should occur, not in the life- ally slowed down without the pilot's knowl-
time of an airplane, but in the lifetime of the edge until it stalled and fell out of control.
entire fleet of airplanes of a given design. This experience shows that the provisions of
This requirement amounts to a time between Gilruth's flying qualities requirements
failures of the order of 109 flight hours. The should be kept in mind when devising new
meeting of such requirements is still a major control systems. If the new systems fail to
problem, but with redundant systems and provide these features, some modifications
digital control equipment, these reliability should be made to provide an equal degree of
goals can be met. Just the realization on the safety.
part of the designers of the importance of
this degree of reliability represents a great Eventually, the problems of the B-58 were
advance in attaining the desired degree of worked out and pilot error was decreased
safety. with the introduction of training versions of
the airplane. In all, over 80 of these airplanes
Following the accident investigation, B-58 were manufactured. They never saw service
bombers continued to have crashes. Accident in war because the expense of operating
number 6 was a tail failure. Number 7 was an them was found to be very large when com-
interesting case of an unanticipated problem pared with that of subsonic bombers like the
resulting from the novel design of the flight B-52. The development of the weapons pod
control system. A feature of the longitudinal was dropped, but the airplanes were used for
control system was that the airplane was high-altitude reconnaissance and provided
automatically in trim at each value of air- valuable surveillance information in the
speed. This feature is desirable on supersonic Cuban missile crisis and in a volcanic erup-
airplanes because large trim changes often tion in Alaska.

Monographs in Aerospace Histo_. Number 12--Journey in Aeronautical Research 143

Accidents of the Republic F-105 Thunderchief Airplane

The airplane also had very light stick forces

Accidents of the
and low longitudinal stability in the condi-
Republic F-105 tions existing in dive pull outs. Such condi-
Thunderchief A irplane tions are known to be favorable to pilot-
induced oscillations. The control feel was
A number of cases had occurred in which the considered desirable by pilots, however, and
Republic F-105 airplane, following a pull out it was generally considered that the problem
from a dive-bombing run, experienced a of oscillations could be avoided by pilot
large amplitude pitching oscillation. In one training. The Republic Corporation had real-
case at Nellis Air Force Base, three airplanes ized that a pitch damper disconnect in the
were performing rocket missions in a race middle of a maneuver might trigger a pilot-
track pattern. One of the airplanes made a induced oscillation. The pitch damper sys-
pull up at 500 feet altitude, and a series of tem had therefore been modified to prevent
three longitudinal oscillations was observed pitch damper disconnect except in case of an
with accelerations of 7g and -lg. The fol-
actual failure of the pitch damper. Though
lowing airplane in the pattern, after pulling
details of this modi_cation are not available,
out at 300 feet altitude, experienced a similar
the change involved installing a duplicate
oscillation. At the end of the third cycle, the
pitch gyroscope and a monitor circuit.
airplane went to a large pitch attitude and
In examining the data from tests of the modi-
fied system, the NASA representatives noted
I made a trip with an instrumentation engi-
some inconsistencie_ between the measured
neer, R. H. Sproull, to visit the Republic
stick forces and the movement of the control
Aircraft Corporation at Farmingdale, Long
Island, New York, and then to the Aeronauti- stick. These data, if correct, could indicate a
cal Systems Division, Wright-Patterson Air malfunction of the control system. This dis-
Force Base, to discuss the accidents. Prior to covery caused some consternation among the
the trip, the Republic Corporation had Republic engineers, but it was soon realized
already made modifications to the control that the data might be incorrect if up and
system and had run flight tests at Eglin Field down forces on the control stick could affect
to study the problem. The main problems the strain gauge instrument that measured
facing the company were whether to stick force. Such forces might be applied by
unground the airplane and whether to place the pilot in pull-up and push-down maneu-
restrictions on maneuvers.
vers to prevent his head from bouncing
against the canopy. Tests were made that
The stabilizer control system incorporated a confirmed this source," of error.
pitch damper in series with the stabilizer
actuator. To protect against pitch damper
The final recommendation by the NASA rep-
failure, the pitch damper was originally
resentatives includeal removing the ground-
designed to cut off automatically if the accel-
ing restrictions on the airplane, but adding
eration exceeded 4g or -lg, followed by a
restrictions on dive angle and g value level
pitch damper actuator deflection of greater
during pull outs. This accident investigation
than 90 percent full travel in the direction to
extend the g value beyond these limits. The reemphasized the ko_owledge that has been

pitch damper could be caused to disengage, gained from numer, ms cases, namely, that
therefore, by a pull up exceeding 4g, fol- stabilizing devices should not be discon-
lowed by a rapid push down which caused nected as a safety :neasure, particularly in
the damper to oppose the negative pitching situations where the airplane is performing
velocity. violent maneuvers.

144 Monographs in Aerospace tlistory Number 12 Journey in Aeronautical Research

Analytical Study of the F-84 Following Accident in High-Speed Flyby

Beyond some value of dynamic pressure,

Analytical Study of the
however, the wing becomes unstable and
F-84 Following diverges until the structural strength is
Accident in exceeded.

High-Speed Flyby The dynamic pressure for divergence is

Several airplanes have broken apart in the air defined as the dynamic pressure at which the
during high-speed flybys. One of these was wing diverges when the wing root is rigidly
fixed. Airplanes are flown so that this
the Republic F-84 Thunderjet, an early
dynamic pressure is not exceeded in normal
unswept-wing jet airplane with tip tanks.
operations. The fuselage of an airplane in
Such an accident, which is tragic enough in
flight, however, does not provide a fixed
any case, is doubly embarrassing to the com-
restraint because of the change in angle of
pany because it occurs during a demonstra-
attack of the fuselage during a maneuver.
tion of the airplane to high officials of the
government or the Air Force. Movies of the Some effects on the motion of the airplane
might therefore be expected at speeds below
crash seem to show that the airplane
the divergence speed.
exploded in the air, but a frame-by-frame
analysis of the movie shows that the airplane
Equations for the longitudinal motion of the
actually pulled up very rapidly to a high
airplane were set up with the assumption that
value of normal acceleration, about 8g,
in addition to the usual degrees of freedom
before the breakup occurred.
(vertical motion and angle of attack) a third
I was curious about the cause of this and degree of freedom representing wing torsion
other similar crashes. Though I was not asso- was included. The twist angle, of course, var-
ciated with the accident investigation board, ies with the position along the wing span. To
I made a brief analytical study of the stability account for this effect, a technique known as
characteristics that might contribute to a the semirigid method, described in some
crash of this type. The study was never pub- British reports, was used (ref. 15.2). A varia-
lished, but it resulted in a memorandum for tion of twist angle along the span was
the Chief of Research dated October 11, assumed, as well as a distribution of tor-
1948, with the title "Effect of Approach to the sional moment of inertia. By use of a formu-
Wing Divergence Speed on the Longitudinal lation of the equations of motion called
Stability of an Airplane." Lagrange's equations, which integrates these
effects, the torsional mode may be specified
Wing divergence is a phenomenon caused by by a single variable representing, for exam-
the lift on a wing acting at a point ahead of ple, the twist angle at the tip.
the wing torsional axis. The torsional axis is
defined as the axis location at which a lift For a complete analysis of this problem,
force will produce no twist of the wing. wing bending should also be included, but it
Ordinarily, unswept wings have an aerody- was reasoned that wing bending would have
namic center, or point of application of the less effect on the motion because wing bend-
lift force due to change in angle of attack, at ing does not directly affect the pitching
about 25 percent of the chord, but this point moments acting on the airplane. Because of
would be still further forward with tip tanks. the lack of computational facilities at the
The torsional axis is usually at 35 to 45 per- time of this study, it was necessary to limit
cent of the chord. During a maneuver, the the number of variables. With the variables
wing therefore twists in a direction to assumed, the resulting motion consisted of
increase the lift and torsional moment above the short-period rigid-body motion and the
those of a rigid wing. If the structural restor- torsional mode of the wing. With the
ing moment exceeds the destabilizing effect assumption of constant airspeed, the analysis
of the air forces, the wing behaves normally. therefore required the solution of a fourth

Monographs in Aerospace History Number 12--Journey in Aeronautical Research 145

Accidents of the Beech Bonanza Involving In-Flight Structural Failures

degree equation, for which as mentioned pre- tively thick with ton-skinned monocoque
viously, solution techniques were available. construction. These wings showed adequate
torsional stiffness when tested on the ground.
The airplane characteristics assumed did not
When the wing was subject to high loads in
represent any given airplane, but were repre-
flight, however, the skin wrinkled exten-
sentative of a high-speed fighter airplane.
sively, which resulted in a loss of torsional
The wing divergence speed was assumed to
rigidity. In a high-g-value maneuver, the air-
correspond to Mach number of 0.9 at sea
plane could therefore be placed instantly in a
level. The center of gravity was assumed to
very unstable condition that was not evident
be at 31 percent of the mean aerodynamic
in flight preceding the maneuver. Fortu-
chord (MAC), which gave a stability margin
nately, these problems have largely disap-
in maneuvers at low speed of 2.5 percent
peared on modem airplanes that have thinner
MAC. The point of application of lift due to
wings and thicker w_ng skin.
wing torsion was at 25 percent MAC. The
results showed that when the wing torsion
was taken into account, the airplane became
longitudinally unstable in maneuvers above a Accidents of the Beech
Mach number at sea level of 0.61, and at a
Bonanza Involving
Mach number of 0.8 the time for the longitu-
dinal motion to double amplitude was less In-Flight Structural
than 1 second. Failures
A point of interest is that most flight tests of
Chapter 10 contained a discussion of the
high-speed airplanes are made at altitudes
problem of airplane crashes resulting from
greater than 15,000 feet. Often, a high-speed
spiral instability in flight under instrument
low-altitude pass to demonstrate the airplane
conditions. This type of crash usually
is the first time that conditions of maximum
resulted from a pilot with insufficient experi-
dynamic pressure have been encountered. ence in instrument flight becoming disori-
This circumstance may account for the num- ented and entering a spiral dive. With early
ber of accidents that have occurred in such
types of light airplanes, the airplane often
was destroyed by crashing into the ground. If
Problems of the type studied were encoun- the airplane emerged from the clouds with
tered on the Republic F-84 and the Northrop sufficient altitude, the pilot could usually
F-89 Scorpion, both of which had large tip regain his orientation and recover. In the case
tanks. The cure in the case of the F-84 was to of more modem personal airplanes with
put large triangular fins on the outside rear of higher wing loading and lower drag than ear-
the tanks, thereby adding a stabilizing effect tier types, excessive speed builds up rapidly
to the torsional moments acting on the wing, in a spiral dive, wh ch sometimes results in
and as an added benefit, reducing the structural failure in the air regardless of the
induced drag of the airplane. This modifica- visibility conditions.
tion was installed on the F-84E and later
Accidents of this type involving in-flight
models. The success of such empirical fixes structural failure have been encountered with
probably accounts for the lack of interest
all types of high-performance personal air-
in more detailed analytical studies of this
planes, but one type, the Beech Bonanza, had
an accident rate nuch higher than other
A nonlinear structural phenomenon also was similar airplanes. T fis airplane, with a dis-
evident on many airplanes of this period that tinctive V-tail, was first introduced in 1947
resulted in abrupt, unexpected incidents of (figure 15.5). The c( mpany stopped building
both wing flutter or airplane instability. The the V-tail model in 1982 after 10,405 had
wings of airplanes of this period where rela- been produced. B) 1986 there had been

146 Monographs in Aerospace l tistory Number 12--Journey in Aeronautical Research

Accidents of the Beech Bonanza Involving In-Flight Structural Failures

FIGURE 15.5. Beech B-35 !iili

Bonanza airplane as
tested at Langley. i!!ili?!!iiiiii!!iiii!!ii!ilili
_ii i! i

232 reported in-flight structural failures of rather rapidly if the airplane were nosed
these airplanes, which resulted in deaths of down.
over 500 people. Already in 1950 the high
Accident data showed that many of the in-
accident rate of this airplane had become
flight breakups were encountered at values of
apparent. As a result, the Civil Aeronautics
airspeed well beyond the placard limit. Any
Administration (CAA) requested that the
such accidents are not considered the fault of
NACA conduct flight tests on the Bonanza to
the designer, because the pilot is never sup-
try to determine any unusual characteristics
posed to exceed the placard limit. Further-