AGARD-AG-37 - Advanced Aero Engine Testing

n
uvc 533/6>*o-J
H * r
c,
0}
S.? : - CT^Tu/RBlrvJ^ f - EbSCilME^

AGARDograph No. 37
ADVANCED AERO ENGINE TESTING

ADVANCED AERO ENGINE
TESTING
Papers Presented at the Joint Meeting of the
AGARD Combustion and Propulsion
and
Wind Tunnel and Model Testing Panels
Copenhagen, Denmark
October, 1958
Editors
A. VV. MORLEV and JEAN FABRI
D. Napier & Son Ltd, ONERA
London, England Paris, France
Published for and on behalf of

ADVISORY GROUP FOR
AERONAUTICAL RESEARCH AND DEVELOPMENT
NORTH ATLANTIC TREATY ORGANIZATION
by
PERGAMON PRESS
London • New York • Paris • Los Angeles
1959
611 flHftl-37
PERGAMON PRESS LTD.
4 & 5 Filzroy Square, London W.l
PERGAMON PRESS INC.
122 East 55th Street, New York 22, N. V.
P.O. Box 47715, Los Angiles, California
PERGAMON PRESS S.A.R.L.
24 Rue des Scales, Paris V*
Copyright
©
1959
ADVISORY CROUP FOR
AERONAUTICAL RESEARCH AND DEVELOPS!! \ I
NORTH ATLANTIC TREATY ORGANIZATION
l.ihinry of Congress Card No. 59-11209
Printed in N o r t h e r n Ireland at T h e Univcrsilies Press, Belfast

CONTENTS
INTRODUCTION - - - - - - - vii
Theodore von Kdrmdn
PART I . — E N G I N E T E S T FACILITIES
PROPULSION SYSTEM PROBLEMS INVESTIGATED IN W I N D T U N N E L AND
ALTITUDE T E S T CHAMBER FACILITIES - - 3
E. J . Manganiello
Discussion - - 29
Author's reply - - - 32
SOME SELECTED PROBLEMS IN ENGINE ALTITUDE TESTING - 33

B. H. Goethert
F U L L - S C A L E PROPULSION TESTING IN W I N D TUNNELS - 79

R. IV. Hensel and H. K. Matt
Discussion - - 129
Authors' reply - 136
MESURES CONCERNANT LES ESSAIS DE MOTEURS EFFECTUES DANS LES

GRANDES SOUFFLERIES DE M O D A N E AVRIEUX (France) - 138
M . Pierre
Discussion - - 153
PART I I . — E N G I N E C O M P O N E N T TESTING
M O D E L TESTING OF COMPRESSORS AND TURBINES - - 159
S. F. Smith and C. E. Payne
Discussion - - 170
Authors' reply - -176
SOME EXPERIENCES IN COMBUSTION SCALING - 177

.1. / / . Lefebvre and G. A. Halls
Discussion - - 201
Authors' reply - - 204
LABORATORY AND F L I G H T TECHNIQUE FOR THE MEASUREMENT OF THE

TEMPERATURE OF T U R B I N E BLADING - 205
E. P. Cockshutt, G. G. Levy and C. R. Sharp
Discussion - - 227
Authors' reply - - 238
v
CONTENTS
APPLICABILITY OF SEPARATE COMPONENT T E S T RESULTS TO COMPLETE

PROPULSION SYSTEMS 239
D. S. Gabriel and L. E. Wallner
Discussion - - 257
Authors' reply - 259
T H E F U R T H E R APPLICATION OF G E N E R A L I Z E D PARAMETERS TO Tt RBO
JET PERFORMANCE RELATIONSHIPS — 260
/•,'. Macioce
PART III.—FLIGHT TEST PROBLEMS
F L I G H T T E S T DEVELOPMENT OF SUPERSONIC ENGINES 273

R. E, Pry or
Discussion - - 287
NAME INDEX - - 291
SUBJECT I N D E X - 295
1. PROGRESS in the development oi aTr'-'fJf?athing engines designed for flight
at high Mach numbers and thence at high altitudes became possible in the
last few years, due only to the parallel development of engine test facilities
in which complete propulsion systems and components can be investigated
under simulated flight conditions of pressure, temperature and velocity.
T h e experimental techniques utilized in these researches are quite
different from conventional wind tunnel techniques, even though the test
facilities do not differ basically from wind tunnels designed for experimen-
tation on aircraft and aircraft components.
In order to compare these two techniques and to check in-flight measure-
ments with the results obtained in ground testing facilities, two Panels of
the Advisory Group for Aeronautical Research and Development (AGARD)
—the Combustion and Propulsion Panel, represented by J . D u c a r m e
(Belgium) and A. M. Rothrock (U.S.), and the Wind Tunnel Panel,
represented by Maurice Roy (France) and J . M. Wild (U.S.)— organized a
two-day Meeting in Copenhagen, Denmark, at the time of A G A R D ' s
Eighth General Assembly (October 1958). T h e Flight Test Panel participated
in the proceedings of the Second Day which included flight test techniques
and correlation of test data.
2. T h e general problems in propulsion systems investigated in wind tunnels
and altitude test facilities were presented by E. J . Manganiello (NACA).
Typical of such problems are the determination of turbojet performances
at various altitudes, research on afterburner stability and efficiency, inlet
flow distortion problems, and transient phenomena related to inlet pressure
and temperature variations.
Ramjets are usually studied in free-jet facilities where the internal aero-
dynamics of the air intake can also be investigated, although in some cases
the ducted nozzle simulation techniques give adequate information concern-
ing internal aerodynamics. Interaction between internal and external flows
can be investigated in subsonic or supersonic wind tunnels also, the main
problems being exhaust nozzle performance, powerplant installation and
engine-airframc studies. B. H. Goethert of the Arnold Engineering Develop-
ment Center presented a review in which he described typical solutions of
these problems. The ven important problem of Bow distortion .it the
engine entrance section may be investigated by means of screens installed
in the cylindrical inlet duct to the engine or, in the case of free-jet installations,
by reproducing the actual angle of attack. Transient testing of propulsion
systems requires very quick and precise control of the inlet and exit conditions
in order to reproduce the rapid changes in thrust level suffered by an engine
in acceleration or in deceleration. Such rapid mass flow changes are
obtainable by means of automatic control systems which allow extensive
entrance or exit condition changes in very few seconds. T h e very high
altitudes of modern engine flights are simulated in test facilities by means
of auxiliary ejector systems which improve the pressure level provided by
conventional rotating exhaust machinery.
vii
INTRODUCTION
Several of the most up-to-date engine test facilities are described in a

paper by R. W. Hcnscl and H. K. Matt A R O : who consider the French
transonic tunnel at M o d a n e , the N A C A 8 X 6 ft and 10 X 10 ft propulsion
wind tunnels at Cleveland (Ohio) and the U.S. Air Force Propulsion wind
tunnel at the Arnold Engineering Development Center (Tullahoma,
Tennessee). A critical survey of the various techniques used for establishing
matched altitude conditions for M a c h number, pressure and temperature
show:s the latest developments in engine test facility design. Exhaust gas
removal, scavenging scoop and wall interference effects, special instrumenta-
tion, and unusual test techniques are important items in engine wind tunnel
testing, even for conventional airbreathing engines. Altitude rocket testing
involves new test requirements and, therefore, the development of new
facilities.
Typical engine tests performed at O N E R A ' s Modane Transonic Wind
T u n n e l are described by M . Pierre ( O N E R A ) . A huge wind tunnel,
originally designed for aerodynamic studies only, has been converted to a n
engine test facility; furthermore, several additional specialized high-speed
engine test facilities are under construction at the same location.
3. T h o u g h the ultimate purpose of aero-engine development is the testing
of the entire engine, the intermediate operations of engine component
testing are also very important in practical design.
Compressor testing can be m a d e either on full-scale models or on scaled
down models and these techniques are discussed by C. E. Payne. At Rolls-
Royce, scale model testing is always restricted to research investigations or to
the testing of new principles of compressor design. T h e development of
a compressor for an existing engine or for direct application to a project is
usually carried out on full scale units.
Similar problems arise also in combustion chamber researches for which
A. H . Lefebvre and G. N. Halls (Rolls-Royce) give an analysis of the
experimental difficulties encountered in scaling. Development tests are
often required on full-scale engines.
O n the other hand, in turbine testing, S. F. Smith of Rolls-Royce, reports
that it is usual to determine the fundamental aerodynamic design data from
model turbines. T h e main difference between compressor and turbine
testing comes from the fact that compressors are usually made u p of a number
of successive stages, whereas turbines have no more than one or two stages
and, therefore, the similarity laws may be applied more easily to turbines
t h a n to compressors.
T u r b i n e research work involves not only aerodynamic measurements.
E. P. Cockshutt, G. G. Levy and C. R. Sharp of the National Aeronautical
Establishment, C a n a d a , indicate means for measuring wall temperature on
ground test benches as well as during flight tests.
T o determine if test results obtained on separate engine components apply
to complete propulsion systems necessitates a n accurate analysis. D. S.
Gabriel and L. E. Wallner (NACA) show that many problems which must
be overcome in the development of each individual component of an engine
stem from an inadequate understanding of how the separate component
tests may be applied to the complete engine. T h e complex problem of
applicability of separate component performance tests to complete propulsion
viii
INTRODUCTION
systems involves considerations ranging from small details, such as the

proper arrangement and simulation of shaft seals, u p to such broad con-
siderations as the location of the engine on the airplane. It can be concluded
that, if all actual engine environmental conditions could be simulated
accurately on a test rig, separate component tests would be applicable to
the entire system. Thus the problem reduces to the question of what are the
really important environmental conditions for steady-state and for transient
performance determinations.
4. After very exhaustive ground tests have been conducted on a new
engine, it is time to begin all over again on a flying test bench before the
engine designed for high-speed propulsion is finally incorporated into an
airplane.
T h e description given by R. E. Pryor (General Electric Company) of the
successive steps between the reception of a new engine and the first supersonic
flight of a new airplane provides a most instructive picture of all the pre-
cautions that must be taken before an engine becomes airborne.
5. It was A G A R D ' s aim to bring together representatives from various
fields of activity to discuss common problems and to compare the relative
merits of the experimental techniques utilized in various test installations.
I believe that this aim has been achieved.
T H E O D O R E VON KARMAN
December 1958 Chairman, A G A R D
ix
LIBRARY. J3
PART I
ENGINE TEST FACILITIES

PROPULSION SYSTEM PROBLEMS INVESTIGATED
IN WIND TUNNEL AND ALTITUDE TEST
CHAMBER FACILITIES
By EUGENE J. MANGANIELLO
N a t i o n a l Advisory C o m m i t t e r lor Aeronautics
Lewis Flight Propulsion L a b o r a t o r y
The general classes oj propulsion system problems investigated in the wind tunnels
and altitude test chamber facilities of the NACA Lewis Flight Propulsion
Laboratory are outlined and some typical experimental results are given for
illustrative /imposes. Four categories of test facilities are considered—altitude
test chamber facilities, free-jet facilities, subsonic wind tunnels and supersonic
wind tunnels.
The types of turbojet investigations in altitude test chamber facilities that are
discussed include afterburner research problems, altitude performance calibrations,
altitude operating characteristics, inlet flow distortion effects, engine dynamics and
controls studies, and inlet pressure and temperature transients phenomena.
A brief summary of ramjet problems investigated in free-jet facilities is given
and a representative turbopropeller investigation conducted in a subsonic wind
tunnel is described.
Finally, supersonic wind tunnel investigations of problems involving interaction
between internal and external flow are discussed. Included in this category are
exhaust-nozzle studies, powerplant installation problems and engine-airframe
configuration studies.
INTRODUCTION
T H E INVESTIGATION of propulsion system problems in wind tunnel and
altitude test chamber facilities is an important part of the long and complex
process involved in the development of successful aircraft powerplants. Such
investigations provide the essential performance data and operating limits
of current propulsion systems at simulated flight conditions. Many diffi-
culties of altitude and high-speed flight operation may thus be solved that
would otherwise require extremely costly, hazardous, and time-consuming
flight tests.
In addition to these obvious development aspects relative to current
engines, investigations in wind tunnel and altitude test chamber facilities
also provide a wealth of research data and design information for future
engines. Various individual component improvements can be integrated
into a complete propulsion system, and this system and the interaction effects
of its different components can be evaluated under the actual engine condi-
tions corresponding to flight. During the course of these evaluations, many
unanticipated research problems are revealed and immediate solutions or
information leading to their definition and eventual solution may be
obtained.
E. J. MANGANIELLO
This paper outlines a n u m b e r of categories of propulsion system problems

investigated in the wind tunnels and altitude test chamber facilities of the
N A C A Lewis Flight Propulsion Laboratory. T h e problem areas considered,
although far from all-inclusive, are believed to be reasonably representative
of the principal type of engine tests conducted in the various propulsion
facilities that are in widespread use throughout the N A T O countries. Al-
though some consideration is given to individual component problems
principal emphasis is placed on investigations related to complete engines
or propulsion systems. Inasmuch as the companion papers being presented
at this J o i n t Meeting of the Combustion and Wind Tunnel Panels, 8th
General Assembly, discuss facilities, measurements, and test results in detail,
this paper provides onl) a fairly general or outline treatment of the subject.
Some typical experimental results are included, however, to illustrate the
various problem areas.
A L T I T U D E TEST CHAMBER FACILITIES

Altitude lest chamber facilities are herein construed as test chambers and
associated compressor and exhauster equipment that provide the engine
inlet and exit conditions corresponding to the stagnation and ambient
conditions, respectively, of high-speed, high-altitude flight. Such facilities
accurateK simulate the internal How conditions encountered l>\ .1 propulsion
system in flight b u t do not provide external flow simulation. A photograph
and sketch of a typical turbojet engine installation in an altitude test chamber
are shown respectively in Figs. 1(a) and (b). A discussion of the research
techniques utilized in altitude test facilities is given in FLEMING and WILSTED
(1953).
Component Research Problems
Many important individual engine component problems have been studied
in altitude test facilities. T h e turbojet afterburner is an outstanding example
of such component problems and is particularly appropriate because of the
severity of the afterburner problem at high altitude.
Afterburner problems involve both performance and operational aspects
such as internal aerodynamics, combustor performance, starting, cooling
and so forth. A sketch of an afterburner is shown in Fig. 2 with some of the
principal problem areas indicated.
T h e diffuser aerodynamics have an important effect on afterburner
performance. For example, steep velocity gradients at the diffuser exit
result in high local velocities with a t t e n d a n t low combustion efficiency,
high pressure losses and reduced operating limits. Furthermore, diffuser
flow separation accompanied by burning in the separated region along the
diffuser wall can produce low-frequency pressure oscillations.
Ignition system problems are those of achieving reliable ignition for
restarting u p to the altitude operating limits of the afterburner. T h e
combustion problem is a major one and probably accounts for the most
total effort in afterburner research. Some of the principal factors investigated
are llameholder geometry, fuel injector design, burner length and burner-
inlet velocity. T h e objective of investigation of these factors is, of course,
to obtain combinations giving high combustion efficiency over a range of
PROPULSION SYSTEM PROBLEMS
Fig. la.
Typical turbojet installation in altitude test chamber
ÊNGINE
COMPARTMENT
COMPARTMENT,
AIR AT RAM
PRESSURE
AND TEMP /ENGINE ALTITUDE
EXHAUST
+X=
•' THRUST
L SVST
Fig. lb.
Typical turbojet installation in altitude test chamber
r FUEL INJECTION EJCCTOR

IGNITION / COOLING-x PERFORMANCE|
OIFFUSER COMBUSTION CHARACTERISTICS

AERODYNAMICS
EFFICIENCY
FLAMEHOLDER OPERATING LIMITS
SCREEN
Fig.2.
.Afterburner research problems
E. J . M A N G A N I E L L O
altitudes while avoiding combustion screech or high-frequency oscillations.

Because of the high temperatures present in afterburners, cooling a n d
durability are important problems. Some cooling can be provided by passing
gas at the turbine-outlet temperature along the wall of the afterburner or
through a n internal liner. Some external airflow is also needed, particularly
for the nozzle. W h e n ejectors are used for pumping this ait. at tent ion must
be given to minimizing thrust losses.
By way of illustration, some representative experimental results from a n
afterburner investigation are presented in Fig. 3. Combustion efficiency
is plotted as a function of afterburner fuel-air ratio for a range of burner-
inlet total pressures. T h e corresponding altitudes for the particular engine
used at a flight M a t h number of 0-9 are also listed. T h e results show two
AFTERBURNER ALTITUDE
INLET TOTAL FT
PRESSURE,
L B / FT 2
O 3 130 23,000
a 2 IOO 35,000
100 O 1290 45,000
A 990 50,000
t> 770 55,000
Cl 590 60,000
90
COMBUSTION
EFFICIENCY.
PERCENT
80
70-
L _L
.2 .4 6 3 TO ill
AFTERBURNER FUEL-AIR RATIO, % OF STOICHIOMETRIC
Fig. 3.
Effect of altitude on afterburner performance
typical characteristics: decreasing efficiency with decreasing pressure, and

decreasing efficiency with both increase a n d decrease in fuel air ratio from
a narrow o p t i m u m range. Detailed treatment of a variety of afterburner
research problems is given in CONRAD el al. (1953), YOUNGER et al. (1952),
M O O R E a n d MASLEN (1954) a n d FLEMING et al. (1951).
Engine Performance Investigation

Altitude test chamber facilities have been used extensively to obtain
engine performance calibrations and operating limits over the ranges of
flight speeds and altitudes to be encountered in service. Such performance
investigations could involve a n excessive amount of testing if data were
obtained at all the combinations of flight altitudes and Mach numbers of
interest. Fortunately the a m o u n t of required testing can be considerably
reduced through use of generalization procedures involving pressure,
temperature and Reynolds number. I n practice it has been found convenient
to include Reynolds n u m b e r effects on engine performance through use of
a so-called Reynolds n u m b e r index defined as the ratio of Reynolds n u m b e r
of the compressor-inlet flow at flight altitude and Mach number to the
Reynolds n u m b e r of the compressor-inlet flow at standard sea-level condi-
tions. This index can b e reduced to the expression djcpy/O where D is the
PROPULSION SYSTEM P R O P I, P. M S
pressure ratio. 9 is the viscosity ratio and 0 is the temperature ratio of the
altitude to sea-level standard conditions. T h e manner in which this Reynolds
number index varies with altitude and Mach number is shown in Fig. 4.
If engine tests are conducted at a reasonable number of Reynolds number
indices, the data can be used to construct complete engine performance maps
for wide ranges of altitudes and Math numbers. Those performance variables
associated with fuel flow do not generalize with Reynolds number index
because of the dependence of combustion efficiency on chemical kinetics.
Reasonably good generalization of combustion efficiency has, however, been
obtained with the product of engine airflow a n d turbine-outlet temperature
which can be related to the fundamental velocity temperature-pressure
REYNOLDS NUMBER INDEX

0.025 .05 .10
100
TROPOPAUSE, 36.089 FT
4.0
H.O
I 2
FLIGHT MACH NO,
Fig. 4.
Variation of Reynolds number indices
relationship involved in combustion kinetics equations. Sufficient data to

establish this correlation arc obtained when testing over a range of Reynolds
n u m b e r indices; hence, additional tests are not required.
In using the Reynolds number index technique to generalize overall
performance, the engine pumping characteristics are plotted as, for example,
in Fig. 5. Here engine total-pressure and -temperature ratios are plotted
against corrected engine speed for a number of Reynolds number indices.
T h e corrected engine airflow is similarly plotted in Fig. 6, a n d combustion
efficiency is plotted against the product of engine airflow a n d turbine-outlet
temperature in Fig. 7. With these three figures (Figs. 5, 6 a n d 7), a suitable
combustion chart (TURNER a n d BOGART, 1949) that relates temperature
rise to fuel air ratio and the relation of Reynolds number index to altitude
and Mach number shown in Fig. 4, conventional engine performance
curves can be computed for the desired range of altitudes and flight speed.
A typical plot of engine performance thus obtained for one altitude is shown
in Fig. 8.
More detailed discussion and experimental results of engine performance
are given in N . D . SANDERS (1946), W A L L N E R a n d FLEMING (1949), CHILDS
ei al. (1957) and W A L K E R el al. (1952).
2.0- ALTITUDE, REYNOLDS
FT NO INDEX
18 0 5,000 88
Q 1 5,000 77
0 25,000 57
1.6 A 35.000 35
ENGINE T O T A L PRESSURE Q 45,000 22
RATIO a 50,000 17 ^
I 4
1 2
1.0
(A) ENGINE T O T A L P R E S S U R E RATIO
4 4
4 0
3.6
ENGINE TOTAL TEMP

RATIO 3.2
2,8
2,4
LJ_ I _L I 1. 1 J
40 50 60 70 80 90 100 10
CORRECTED ENGINE SPEED, PERCENT
(B) ENGINE TOTAL TEMP RATIO
Fig. 5.
Turbojet pumping characteristics for a range of Reynolds number indices
lOOr-
CORRECTEO AIR FLOW,

LB/SEC
ALTITUDE REYNOLDS
FT NO.
INDEX
5,000 08
1 5,000 .77
25,000 57
35,000 35
45,000 22
50,000 17
40 50 60 70 80 90 100 "0
CORRECTED ENGINE SPEED. PERCENT
Fig. 6.
Effect of altitude on generalized engine performance. Flight Mach No. 0-2.
. l/lilude Operating Characteristics

Two of the more important altitude operating characteristics that have
been the subject of extensive investigation are the operating limits as deter-
mined by compressor stall or combustion blowout and the ability to start an
engine at high altitude. Starting of a turbojet engine requires that (1)
ignition in the combustors containing sparkplugs or other ignition devices
100
90
COMBUSTION
EFFICIENCY
80
70 1 -
I
20 40 60 80 100 120
COMBUSTION PARAMETER, L8--R/SEC
Fig. 7.
Correlation of combustion efficiency with combustion parameter
FUEL
AIR FLOW,
FLOW, J LB/HR
.B/SEC^
"4^-7500
y E
NET THRUST. LB
200 4 0 0 600 800 1000 1200

TRUE AIRSPEED, KNOTS
Fig. 8.
Overall performance curves generated from typical engine pumping characteristics
be accomplished, (2) the flame be successfully propagated to the other

combustors in the engine, and (3) the engine accelerate from the starting
speed to maximum engine speed without encountering combustion blowout
or compressor stall and without exceeding the allowable temperature of
lite engine.
K- I- M A N G A N I E L L O
Some of the data obtained in turbojet altitude ignition investigations

are presented in WlLSTBD and ARMSTRONG (1951). T h r e e factors affect
ignition, fuel vaporization, sparkplug location and spark energy. R e p r e -
sentative data illustrating the effect of spark energy on ignition are shown
in Fig. 9. Spark energy in joules per spark is plotted against spark-repetition
2 -
EACH SYMBOL
O IGNITION 1 REPRESENTS
D NO IGNITION
1 J SEVERAL
ATTEMPTS
SPARK ENERGY,
JOULES/SPARK
NO IGNITION
40 80 120 160 200

SPARK-REPETITION RATE. SPARKS/SEC
Fig. 9.
Spark energy required for ignition. Ignition system, variable energy, 10.000 volts; fuel and
inlet air temp., 3 9 ' F ; fuel. I Ibjin.1 Reid vapor pressure; altitude. 50,000ft; Flight Mach
No. 0-6
rate per second for a particular set of operating conditions. As repetition

rate is increased, the energy per spark required for ignition is decreased.
M i n i m u m total energy occurred, however, at the low sparking rate of one
spark per second with about one joule per spark. Furthermore, for reliable
ignition at altitude the energy dissipated per spark is more critical than the
total energy per second.
Acceleration from starting or windmilling speed becomes more difficult
with increasing altitude for two reasons. T h e decreasing airflow through
the engine with increasing altitude reduces the possible accelerating energy.
At the same time, the reduction in stall margin with altitude further limits
the permissible accelerating energy. T o illustrate how the acceleration
limit of a turbojet engine affects the overall starting situation, the altitude
acceleration limits of a turbojet engine are compared in Fig. 10 with the
GNITION LIMIT
START REGION
ACCELERATION LIMIT
_l I I I I
25 50 75 100 125
FLIGHT MACH NO
Pig. 10.
Typical altitude starting limits
itl
P R O P U L S I O N SYSTEM PROBLEMS
altitude ignition limits. At low flight Mach numbers the acceleration charac-
teristics limit successful starting, whereas at high flight Mach numbers
ignition becomes the limiting factor.
Whereas combustion blowout was an important operating limit of early
turbojet engines, in modern engines compressor stall margin has become
a more critical limit. This problem is illustrated in Fig. 11 where stall lines
and steady-stale operating lines arc shown for a representative high-pressure-
ratio engine .11 two altitudes. Ai the highei altitude the operating line is
higher and the stall line lower, thus decreasing the stall margin. Below
about 80 per cent of rated engine speed the stall margin is almost non-existent.
COMPRESSOR
PRESSURE
RATIO
STEADY STATE
OPERATING LINES
60 70 80 90 100 110
Pig. I I .
Effect of altitude on compressor stall margin. Flight Mach No. 0-8
O f course, variable-geometry features are incorporated in modern turbojet

engines to overcome such limits.
Inlet Flow Distortion
In recent years, as a result of the trend towards high stage loading, inlet
flow distortion effects on compressor performance and compressor stall have
become a serious problem. T h e use of sharp-lipped inlets for high-speed
aircraft with concomitant increase in distortion of compressor-inlet airflow
during high angle of attack operation has aggravated the problem. M a n y
investigations of this phenomena have been conducted in altitude test
facilities, wherein various flow distortion patterns are simulated through the
use of various screen combinations in the engine inlet duct.
Some representative results of one of these investigations are presented
in Fig. 12 where the engine operating lines and stall limits are plotted for
conditions of (1) a uniform flow, (2) a circumferentially distorted flow, and
(3) a radially distorted flow. T h e reduction in stall margin for the distorted
flow conditions is appreciable, particularly at the part-speed condition.
Detail information on inlet flow distortion effects is contained in AI.FORD
(1955), ALFORD (1957) and CONRAD el al. (1955).
Inlet Pressure and Temperature Transients

Several years ago, several jet fighter aircraft experienced engine difficulties
during a r m a m e n t firings. These difficulties were eventually identified as
11
MANGANIEI.LO
J
conipii-ssot stall phenomena associated with temperature and pressure
transients resulting from ingestion of high-temperature or high-pressure
gases in the engine inlet. Extensive investigation in altitude test facilities
STALL LIMI
COMPRESSOR
PRESSURE
RATIO
INLET FLOW
CONFIGURATION
O UNIFORM
D CIRCUMFERENTIAL
DISTORTION
O RADIAL DISTORTION
60 70 80 90 100 110
Fig. 12.
Effect of inlet flow distortion on compressor stall margin. Altitude, 35,000 ft.
of this problem and closely related effects of diffuscr buzz and rapid maneuvers
have provided a fairly good understanding of the details of the compressor
flow dynamics and overall engine behavior.
By way of illustration, the effect of temperature transients or compressor
stall is presented in Fig. 13. T h e boundary between stall and stall-free
iGO
COMPRESSOR STALL
120
TEMP RISE
AT S T A L L 8 0
INCEPTION.
•F
40- NO COMPRESSOR STALL
0 1000 2000 3000 4000 5000 6000 7000

TEMP RISE R A T E . 'F/SEC
Pig. 13.
Compressor stall induced by inlet temperature transients. Engine speed, 93 per cent.
operation is plotted as a function of temperature rise at the compressor inlet

and of temperature-rise rate. T h e explanation for the greater tolerance
to inlet temperature rise at the higher temperature-rise rates lies in the fact
that, at the 93 per cent engine speed of these tests, stall occurred at the rear
stages of the compressor. Accordingly, for constant flow velocity or time
interval for the temperature transient to progress from the front to rear
stages, a given stall temperature at the rear stages predicates increasing inlet
temperature with increasing temperature-rise rate.
Discussion of inlet temperature and pressure transient effects is contained
in G A B R I E L et al. (1957).
I'J
Engine Dynamics and Controls

Full-scale altitude test chamber facilities have been a particularly powerful
research tool in the field of engine dynamics and controls. Testing in these
facilities, in contrast to flight testing, has permitted the use of the extensive
and specialized instrumentation required for adequate controls investigations
and has resulted in identification of problems a n d accumulation of data
necessary to the understanding of basic phenomena and development of
rational theory.
Among the many controls problems investigated are those related to
(a) engine starting and acceleration, (b) the avoidance of operating limits
such as combustor blowout and compressor stall, (c) protection of the
engine against damage by overtemperaturc and excessive stress, (d) mainten-
ance of optimum conditions of operation, and (e) achieving stable operation
and rapid response to changing load demands. An extensive amount of such
6r 40,000 FT
15,000 FT
!$
IS oo
^4
z ,•
UJ I - p
H .<
0
o o
<, n
u
5 6 7 8 9xl0 3
CORRECTED ENGINE SPEED, N/./5", RPM
Fig. 14.
Effect of altitude on corrected engine lime constant. Mach number, 0-624.
work has been carried out on both turbojet and ramjet engines (see e.g.,
.). C. SANDERS (1957), W Y A T T (1953), PERCHONOK and H E A R T H (1957),
Ki TCHUM and CRAIG (1952), O T T O and TAYLOR (1951), H E P P L E R ^ O / . (1952)
and SCHMIDT et al. (1957)); however, only a few representative examples of
this research will be cited.
Some of the problems of engine operation such as starting and acceleration
have already been mentioned. Basic to the problem of system stability
and response is a knowledge of the dynamic characteristics of the engine.
These engine dynamics are important because the control is constantly
responding to deviations in a control variable and acting to return the
variable back to its desired value. It has been found that a turbojet engine
is essentially a linear device and that its dynamic characteristics may be
conveniently and adequately described in terms of time constants. O n e
such time constant of importance is that which relates the transient engine
speed to a change in fuel flow; numerically, this time constant is the time
required for the speed to reach 63 per cent of its final value with a step change
in fuel flow imposed on the system.
T h e engine time constants obviously vary with both engine and flight
operating conditions. Theoretical considerations have indicated methods
of predicting the effects of flight conditions on engine dynamic characteristics
and one experimental verification is presented in Fig. 14. In this figure,
13
E. J. MANGANIELLO
the speed to fuel-flow time constant of an engine is plotted against the engine
speed. Both variables have been corrected to sea-level conditions as indicated
by theory and in a m a n n e r similar to that customarily used for steady-state
engine performance. Satisfactory generalization of the d a t a for both 15,000
and 40,000 ft is indicated. By means of this correlation, therefore, results
obtained at one altitude, say at sea level, may be used to predict the dynamic
characteristics at other altitudes.
From data such as that presented in Fig. 14, it may be determined that
the actual engine time constant will increase from about 2 sec at sea-level
to about 15 sec ai an altitude of 50.ooo li; i.e. the engine is very sluggish
at high altitude. As previously implied, inadequate recognition of this
characteristic can result in poor response or unstable operation of the
controlled system. An example of this problem of high-altitude instability
r^r* 1
TIME, SEC
Fig, 15.
Behavior of afterburner control
is presented in Fig. 15. Transient thrust, engine speed, and turbine-inlet

temperature are shown as a function of time. The transient imposed on the
engine is that of a throttle change from full non-augmented thrust to full
afterburning thrust—an increase in thrust of some 50 per cent. Although
the control provided rather satisfactory speed and temperature control, the
system broke into severe thrust oscillations after some 7 or 8 sec. As evident
in the top trace, thrust oscillation from the full dry to full afterburning level
occurred with a b o u t a 2 sec period despite the fact that this engine w:as very
stable at sea level and low altitudes. T h e source of the oscillatory behavior
was found to be in the changing dynamic characteristics of the engine with
increasing altitude and associated interaction effects among the various
control system loops. Based upon the knowledge obtained from such investi-
gations as that illustrated in Fig. 14, together with subsequent analog com-
puter analyses, the control system was appropriately redesigned and stable
operation at high altitude achieved.
FREE-JET F A C I L I T I E S
Altitude test facilities have been used for investigation of ramjet problems
as well as the turbojet problems discussed so far. Preliminary studies of
ramjet combustion-chamber and nozzle problems may be made with the
14
so-called connected-pipe technique wherein stagnation pressure and tempera-

lure conditions are provided at the combustor inlet.
The usefulness of such studies i ,m lie great!) extended l)\ resorting to
the free-jet method of testing wherein the ramjet inlet is submerged in a
supersonic jet of air in which the Mach number, pressure, temperature,
and anglc-of-attack conditions of actual flight are reproduced. I n this
manner, a good simulation of the internal flow conditions in the inlet diffuscr,
and hence throughout the engine, m,i\ be obtained. The external flow
conditions around the outside of the engine, however, are not reproduced.
A schematic diagram of a simple form of a free-jet facility is shown in Fig. 16
and a photograph of an installation is shown in Fig. 1 7. This is the same
JET
DIFFUSES
SUPERSONIC NOZZLE
IAI WITH JET DIFFUSER AT NOZZLE OUTLET

ADJUSTABLE OUTER WALL SPECIAL COWL
OF SECOND THROAT > SECOND / ON ENGINE
SUPERSONIC NOZZLE.
IB) WITH SECOND THROAT AROUND ENGINE COWL
Fig. 16.
Free-jet facilities with jet diffuser and second throat for pressure recovery
altitude test facility shown in Fig. 1 modified by addition of a supersonic

nozzle at the inlet pressure bulkhead. A comprehensive treatment of ramjet
facilities is given in LUNDIN (to be published). Additional information is
contained in SEARS and BAJEK (1957).
T h e three variables of principal importance to ramjet performance are
indicated in Fig. 18 along with their relative effect on theoretical aircraft
or missile range at a flight Mach number of 3-25. T h e much greater effect
of nozzle efficiency on range than either combustion efficiency or inlet
pressure recovery is not always appreciated.
These three variables and related characteristics of ramjets have been
the subject of many investigations in free-jet facilities. Inlet problems
include velocity distribution and dynamics, in addition to pressure recovery.
T h e combustion characteristics comprise efficiency, pressure loss, stability,
ignition and operating limits. T h e nozzle problem is primarily one of
efficiency.
In addition to these basic performance parameters, the control system
characteristics, both steady state and dynamic, have been and continue
to be a fruitful field for investigation. Finally, the cooling characteristics
and hardware durability introduce practical problems that are investigated
15
E. J . MANGANIELLO
Fig. 17.
Test section of free-jet facility with jet diffuser around engine inlel cowl
-PRESSURE RECOVERY-
100
PERCENT
0F
96
THEORETICAL
RANGE
0 2 4 6 8 10 12
PERFORMANCE LOSS. PERCENT
Fig. 18.
Ramjet performance variables. Flight Mach No. 3 25.
16
a n d solved in ramjet engine research. Included in most investigations of

these ramjet characteristics are a range of altitudes, M a c h numbers, fuel-air
ratios, and angles of attack.
By way of illustration Fig. 19 is presented showing the effect of angle of
attack on the blowout fuel-air ratio for an experimental ramjet engine.
T h e reduction in the fuel-air ratio at which blowout occurred with both
increase and decrease of angle of attack is attributed to the accompanying
distortion of air flow and resultant rich mixture in localized regions of the
combustion chamber.
rr.
COMBUSTOR
BLOW-OUT
FUEL-AIR RATIO
- 8 - 4 0 4 0
ANGLE OF A T T A C K , DEG
Fig. 19.
Effect of ramjet angle of attack on blow-out fuel-air ratio
SUBSONIC- WIND TUNNELS

Many of the problems of turboprop engines are, of course, identical to
those of the turbojet engine and can be investigated in altitude test facilities
through use of suitable dynamometer equipment. There are additional
problems, however, that are unique to the turboprop engine, namely those
associated with the propeller component. These problems can best be
studied in large subsonic wind tunnels of sufficient test-section diameter to
permit operation of the complete engine-propeller installation.
T h e Altitude Wind Tunnel of the NACA Lewis Laboratory, which has a
20-ft-diam. test section, has been effectively utilized for a number of turbo-
prop engine investigations (e.g. WAI.LNER and SAARI, 1950). T h e capability
of this tunnel to provide altitude test conditions is a desirable feature for
engine investigations. 1'hc conventional sea-level wind tunnels, however,
can and have provided valuable data relevant to turboprop propulsion
systems.
O n e of many interesting problems unique to the turboprop engine is that
of engine failure during takeoff, landing, or even during cruise. T h e re-
sultant high windmilling drag of the turbine engine can result in high
asymmetric thrust forces on a multiengine aircraft which, in some cases,
may be beyond the control capability of the airplane rudder or the structural
load limits of the tail assembly or both.
This problem was investigated for a particular turboprop aircraft at the
NACA Ames Laboratory. A photograph of the airplane installation is
presented in Fig. 20 and some of the representative test results are given in
Fig. 2 1 . T h e windmilling drag of the engine propeller combination is plotted
17
E. J . MANGANIELLO
Fig. 20.
Turbopropeller installation in Ames 40 X 80 f t wind tunnel
10,000
8000
6000
ENGINE
WINDMILLING
DRAG, LB
4000
2000
100 125" I!675 175 S53 225

VELOCITY, MPH
Fig. 2 1 .
Comparison of engine windmilling drag with allowable values for
multi-engine aircraft at sea level
18
for a range of flight speeds at sea-level conditions with the propeller in the
fixed-pitch condition. Superimposed are limits calculated from the airplane
d a t a which delineate areas of insufficient rudder control and excessive
vertical tail loads. At speeds below about 100 m.p.h. the asymmetric wind-
milling drag forces of the engine-propeller combination exceed the rudder
control forces, and above about 170 m.p.h. they exceed the safe load limits of
the vertical tail.
A possible solution to this problem is the incorporation of an automatic
decoupler between the propeller and engine. Another possibility is the use
of a propeller pitch control that automatically adjusts the blade angle to
minimize the windmilling drag of the engine propeller combination.
SUPERSONIC WIND TUNNELS

Many problems of supersonic propulsion systems and their components
involve interaction of internal and external flow. Investigation of these
problems and those associated with the external interaction between the
powerplant installation and the airframe require external flow simulation
such as is provided by supersonic wind tunnels.
Modestly sized tunnels are used successfully for studies of some of these
problems, but many of the problems require large tunnels capable of
accommodating full-scale engines and large-scale engine-airframe configura-
tions. T h e 8- by 6-ft and the 10- by 10-ft wind tunnels of the NACA Lewis
Flight Propulsion Laboratory, operating at Mach numbers of 0-8 to 2-1 and
2-0 to 3-5, respectively, are examples of large supersonic wind tunnels useful
for investigating such problems. Both tunnels are designed to handle
engine combustion products.
Detailed studies of inlet and induction system components (HERMANN,
1955; CONNORS and M E Y E R , 1956; MOF.CKEL el al., 1947; KANTROWITZ
a n d DONALDSON, 1945; EVVARD and BLAKEY, 1956; FRADENBURGH and
W Y A T T , 1954; SIBULKIN, 1954; W Y A T T and HUNCZAK, 1951; LLTDENS and
HUNCZAK, 1948; HUNCZAK and KREMZIER, 1950) comprise a large fraction
of the investigations conducted in the large supersonic tunnels. These are
similar to investigations conducted at smaller scale in small tunnels however
and, in the interest of brevity, will not be reviewed in this paper. Exhaust-
nozzle investigations, on the other hand, represent an area of component
research for which the large tunnels are particularly valuable and are,
accordingly, discussed as follows.
Component Research
Whereas inlet problems can be studied in combination with complete
propulsion systems without undue difficulty, even the large supersonic
tunnels mentioned previously are not large enough to test complete engine
installations without strong shock reflections from the tunnel walls on the
rear of the model. Consequently, nozzle research is conducted on models
that simulate only the afterbody of the installation.
A typical jet studies model is shown on Fig. 22. T h e model shell which
represents the aircraft external surface is mounted on a force balance so that
all body forces can be measured directly. High pressure air is brought
through the strut and simulates the engine air flow. Besides the usual nozzle
19
E. J . MANGANIELLO
problems which can be investigated, a variety of interaction problems are

also studied: interactions between airplane components (such as the tail)
and the j e t or between the free stream and the jet. T h e base pressures for
example have been found to be significantly different with external airflow.
As illustrated in the bottom rear half of the configuration, ejector nozzle
CONTROL SURFACE
BODY FORCES EXHAUST JET
MEASURED BY INTERACTION
FORCE BALANCE
TMI ^SECONDARY
AIR S U P P L Y
E J E C T O R NOZZLE '
CONFIGURATIONS -
AIR SUPPLY
Fig. 22.
Jet exit research problems
performance is also studied along with possible auxiliary inlet configurations
designed to supply secondary air to the ejector as efficiently as possible.
A representative result of exhaust nozzle investigations is illustrated on
Fig. 2 3 . T h e ratio of actual to ideal thrust predicted from static-test-stand
P R E D I C T E D FROM
QUIESCENT AIR D A T A -
WIND T U N N E L
IDEAL THRUST
DATA
FLIGHT M A C H NO.
Fig. 23.
Stream effects on nozzle thrust
nozzle tests is shown by the upper curve. These data would predict only a
nominal decrease in nozzle efficiency for off-design operation of a nozzle
designed for M a c h 3. T h e wind tunnel data in the lower curve show that
the depression of base pressure at the rear of the model due to external flow
expansion around the base has severely aggravated the nozzle internal flow
problem and has m a d e the nozzle very inefficient near sonic speed. T h e
wind tunnel tests also determine the additional afterbody drag that must be
considered in the nozzle evaluation.
In general, it has been observed that, whereas static test stands are very
useful in determining on-design nozzle performance values, the use of super-
sonic tunnel models such as those described is imperative for obtaining
realistic off-design data. A general discussion of exhaust-nozzle problems
is contained in E W A R D (1957).
20
E. J. MANGANIELLO
Turbojet Installations
O n e of the problem areas in which the large supersonic tunnels have a
demonstrated unique value is the general interaction between the induction
system and the turbojet engine. Acrodynamically, the turbojet is so compli-
cated that it has not been possible to predict its modifying influence on
induction system characteristics. In a like manner, the effects of the flow
distortion characteristics of induction system on engine performance cannot
be explicitly predicted.
Fig. 24.
Typical nacelle installation
A typical installation of a turbojet engine in a nacelle in a large tunnel

is illustrated by Fig. 24. A schematic arrangement of a nacelle test configura-
tion is shown in Fig. 25. In order to reduce the number of engine test-hours
at supersonic operating conditions and in order to obtain basic data relating
to the effect of the engine on the inlet characteristics, a large fraction of each
program is run with the engine removed, as shown by the bottom sketch.
I n this case, a duct liner is installed within the nacelle and a remotely
actuated exit plug is provided to simulate the engine back pressure.
An installation of the type illustrated is used to obtain several simultaneous
research answers. These arc indicated on the next several figures.
Although the factors influencing the onset of diffuscr flow oscillation,
or buzz, are not completely understood, it would be expected that the
reflection characteristics of the downstream passage would exert an effect.
21
! •
MANGANIELLO
This effect is illustrated by the data in Fig. 26 where the inlet mass-flow
ratio at which buzz was first encountered is plotted as a function of the
geometric position of the translating spike. For the tests without the engine
the inlet was found to have limited spillage capability over a wide range of
spike settings a n d , in fact, only became stable over a wide range of spillage
when the spike was retracted enough to cause the oblique shock to fall far
inside the lip. When the engine was installed behind the same inlet, however,
the stable spillage range of the inlet was generally increased and, in particular,
WITH ENGINE
PLUG CONTROL EXIT PLUG
Fig. 25.
Turbojet installation
unlimited spillage became possible for only moderate amounts of spike

retraction. Detailed consideration of the buzz problem is given in PEARCE
(1950), STERBENTZ a n d E W A R D (1955), STERBENTZ a n d DAVIDS (1955),
T R I M P I (1956 a a n d b ) , D A I L E Y (1955), FERRI and N u c c i (1951).
MASS
FLOW
RATIO
EXTEND SPIKE RETRACT SPIKE
DESIGN SPIKE POSITION
Fig. 26.
Effects of an engine on the stability limits of an inlet
As mentioned in a n earlier portion of the paper, much work is done in

altitude facilities on the effects of flow distortion on engine stall by using
screens to create particular distortion patterns. I n a n actual installation,
however, the distortion pattern m a y b e very complex a n d difficult to
22
simulate. Such a pattern is illustrated in the profile sketch of Fig. 27. This
pattern was obtained from a spike-type nacelle operating at a 6° angle
of attack. T h e flow into the compressor face is characterized by a concen-
tration of high-pressure cores in the upper right of the duct and completely
separated flow at the bottom. T h e profile al the compressor face changes
with variations in Mach number, angle of attack, and spike position. In
the wind tunnel installation it is possible to evaluate the effect of these
profiler on engine stall very rapidly through the whole range of operating
conditions anticipated from the inlet. Some of the general considerations
(if flow distortion arc found in WYATT I 1956).
LINES OF CONSTANT
PRESSURE RECOVERY
0.82
.85
90
.85
SEPARATION
DISTORTION AT THE
COMPRESSOR FACE. TOTAL PRESSURE PROFILES
AT THE COMPRESSOR
FACE.o- 6"
WITH ENGINE
WITHOUT
ENGINE
I J
110 120 130 140 ISO 160
CORRECTED ENGINE AIRFLOW. LB/SEC
Fig. 27.
Effect of an engine on distortion at the compressor face
T h e lower part of Fig. 27 shows that the distortion levels measured in inlet
component tests may be modified by the presence of the engine. T h e
distortion parameter plotted here is a measure of the maximum total-
pressure variation across the compressor face without regard to the detailed
profile. For the example shown, the presence of the engine tended to increase
the distortion from the inlet, probably because of prcrotation of the flow
induced by the compressor. Although the increase was small, it might be
significant if the engine were near the stall limit.
Another object of research on complete turbojet engine installations is
the determination of duct dynamics. Fig. 28 shows some aspects of such a
study. If a bypass port, used to control normal-shock position, is given a
step change in position, the normal shock will move. This movement will
not occur instantaneously, however, because of the presence of real fluids
in the duct. No motion will occur until an acoustic wave passes from the
bypass forward to the shock. This interval is indicated on the figure as the
dead time. T h e shock motion, and the concomitant pressure rise at a
static-pressure orifice just ahead of the shock, would then be gradual and
would be determined by the accumulated effect of pressure wave, velocity
wave, and entropy wave reflections between the shock and the engine. T h e
time required for the shock to travel to its final position is known as the lag
23
E. J . MANGANIELLO
time. A knowledge of the dead time a n d the lag time is essential for the
proper design of a n inlet control system.
T h e analytical prediction of the lag time for a given installation is uncertain
because of a lack of knowledge of the factors determining the effective
reflection coefficient of the turbojet engine and of the location of this effective
reflection plane in the engine. Experimental studies a r e used, therefore,
not only to determine the time constants for a particular installation, b u t
to accumulate d a t a that lead to the prediction of the engine reflection
CONTROL
SIGNAL
PRESSURE
DEAD -LAG TIME

"TIME'
BYPASS
POSITION
Fig. 28.
Inlet response to step change of bypass
properties. I n addition to duct dynamics many other aspects of the overall

propulsion system control problems are effectively investigated in supersonic
wind tunnels. Discussion of some of these investigations is contained in
VASU (1957) a n d SCHUELLKR (1957).
Ramjet Installations
Ramjet engine installations are investigated in the large supersonic
tunnels in the same m a n n e r a n d largely with the same objectives as the
turbojet installations. Fig. 29 shows a comparison of research goals for the
<**=
ALTITUDE CHAMBERS WIND TUNNELS
AND FREE JETS
I. COMBUSTOR I. SUBCRITICAL INLET OPERATION

DEVELOPMENT
2. DETERMINATION i . ACTUAL AIRFLOW DISTORTION
OF BURNER
CHARACTERISTICS
3. ENGINE FUEL CONTROL 3. INLET ENGINE CONTROL DYNAMICS
EVALUATION
U. ENDURANCE STUDIES 4. COMPLETE P O W E R P L A N T EVALUATION
Fig. 29.
Objectives of ramjet investigations
24
tunnel tests as compared with those in altitude and free-jet facilities. T h e

altitude facilities are predominantly useful in studies involving the combus-
tion process, although inlet and control information can also be obtained
for supercritical operation. In general, the wind tunnels are not particularly
advantageous for combustion studies per se, both because the external
airflow is not a prerequisite and because true temperature simulation is
difficult at high Mach numbers, which thereby penalizes the combustion
process. T h e tunnels are useful, however, for studying conditions in the
subcritical inlet flow regime, for evaluating the effects of flow distortion,
particularly under transient angle-of-attack conditions, and for determining
the dynamics and overall performance characteristics of the complete
configuration, including external drag. These points will not be elaborated
since they duplicate, in kind, the work just described for the turbojet engine.
Configuration Studies
Just as aerodynamic studies of isolated aircraft components must be
supplemented with studies of complete assemblies in order to evaluate
Fig. 30.
Configuration simulation
interference effects, so must aerodynamic configuration studies be supple-

mented with other studies in which the aerodynamic interferences of the
powerplant installation are evaluated. Large supersonic tunnels have proven
invaluable for these latter programs, both because Reynolds numbers
approaching flight values can be achieved and because the engine installation
can be represented by components sufficiently large to give realistic flow
simulation. Generally speaking, inlet flow simulation does not represent a
major problem with large tunnel configuration models. Simulation of the
exhaust flow field is, however, difficult.
Fig. 30 shows a method of simulating both the inlet and the exhaust-nozzle
flow of a turbojet engine installation that has been used with some success
at the NACA Lewis Laboratory. T h e lower part of the figure shows the
pressure field that might exist in flight around a turbojet installation. T h e
pressure field of interest from the exhaust nozzle is not that associated with
25
E. J . MANGANIELLO
the hot jet core, but rather the secondary far field associated with the shock
structure originating from the conflux of the internal and external flows.
T h e upper part of the figure illustrates a jet simulator technique in which
the pressure buildup in the turbojet engine cycle is replaced by a mass
addition just ahead of the nozzle. Several methods of injecting the auxiliary
mass flow have been found to be satisfactory, including the one illustrated.
By adjusting the mass addition and the plug fairing position it has been
found that both the inlet and exhaust-nozzle conditions can be adequately
simulated. This technique is discussed in greater detail in ENGI.ERT and
LUIDENS (1957).
A typical example of the effect of the exit jet on such a configuration is
presented in Fig. 31. T h e lift-drag ratio was measured for the body-wing
MAXIMUM
BODY-WING-ENGINES
LIFT-DRAG
I W I T H FLOW)
RATIO
BOOY-WlNG-ENGiNES
INO FLOWI
25
FLIGHT MACH NO
Fig. 31.
Effect of jet flow on configuration performance
alone and also with the addition of simulated nacelle engines for both an
exit jet and no jet. T h e addition of engines without simulating the flow
issuing from their exit nozzles gave pessimistic results. W h e n the exit jet
was provided by the simulation technique, improved and more realistic
performance was obtained.
With satisfactory jet-simulation techniques, there are a wide variety of
both general and specific configuration problems that can be investigated.
CONCLUDING REMARKS
T h e foregoing discussion has outlined some of the representative propulsion
system problems that have been investigated in wind tunnels and altitude
test c h a m b e r facilities. Discovery of these and other unanticipated major
problems has been shared between flight tests and ground facilities tests.
Irrespective of where they were first encountered, practically all of the
problems to date have been found amenable to successful investigation in
ground facilities of the type discussed. Indeed solutions to most of them
have been achieved in these facilities rather than in flight because of the
greater range and better control of test conditions, the more extensive
instrumentation and the more rapid accumulation of data.
T h e assistance provided by W. FLEMING, Assistant Chief, Propulsion
Systems Division, and D . D . W Y A T T , Associate Chief, Propulsion Aero-
dynamic Division, and other staff members of the NACA Lewis Flight
26
P r o p u l s i o n L a b o r a t o r y , in t h e p r e p a r a t i o n of this p a p e r is gratefully a c k n o w -
ledged.
SOMMAIRE
On passe en revue les differentes classes de problemes relatifs aux systemes de

propulsion et eludies dans les souffleries el installations d'essais en altitude
sirrmlit du NACA Lewis Flight Propulsion Laboratory. On considere quatre
types d'installations: les bancs d'essais en altitude simulee, les installations a j e t
libre, les souffleries subsoniques et les souffleries supersoniques.
Les recherches sur lurbocompresseurs conduiles sur les bancs d'essais d'altitude
simulee comprennenl les recherches sur la rechauffe, la determination des per-
formances el les caracteristiques de fonctionnement en altitude, V effel de la distorsion
de I'ecoulemenl dans les prises d'air, les eludes aux regimes variables du moteur
et sur la regulation, enfin les phenomenes Iransitoires {pression el temperature
d'admission).
On presente un bref aper(u des recherches effectuees en soufflerie a veine libre sur
les statoreacleurs et on decrit egalement une etude lypique de turbopropulseur
conduite en soufflerie subsonique.
On discule enfin les resuiiats, obtenus en soufflerie supersonique, d'etudes de
['interaction entre les ecoulements externe el interne. Cette categoric d'elude
comprend les recherches sur la tuyere d'ejeclion, I'installation du moteur, les re-
cherches sur la regulation de I'entree ainsi que sur la configuration moleur-cellule.
REFERENCES
ALFORD, J . S. Inlet Ducl-cnginc Flow Compatibility. Paper presented at Fifth
Int. Conf., J u n e 20-24, 1955.
ALFORD, J . S. Inlet Flow Distortion Index. Paper presented at Journces Int. de Sci.
Aero., Paris, France, M a y 27-29, 1957.
GUILDS, J . H O W A R D , REYNOLDS, T I I A I N E W . a n d G R A V E S , C H A R L E S C. " R e l a t i o n of
Turbojet a n d Ramjet Combustion F.lik ieney In Second-order Reaction Kinetics
and Fundamental Flame Speeds" NACA Rep. 1334, 1957.
CONNORS. ( \\n s F. a n d M E Y E R , R U D O L P H C. " D e s i g n Criteria for Axisymmetric a n d
Two-dimensional Supersonic Inlets a n d F-xits" NACA T N 3589, 1956.
C O N R A D , F-. W I L L I A M , H A N S O N , M O R G A N P. a n d M C A U L A Y , J O H N E. "Effects of
Inlet-air-flow Distortion on Steady-stale Altitude Performance of a n Axial-flow
Turbojet E n g i n e " NACA R M E 5 5 A 0 4 , 1955.
C O N R A D , E. W I L L I A M , S C H U L Z E , FREDERICK W . a n d U s o w , K A R L H . "Effect of
Diffuser Design, Diffuser-exit Velocity Profile, a n d Fuel Distribution on Altitude
Performance of Several Afterburner Configurations" NACA R M E53A30, 1953.
D A I L E Y , C H A R L E S L E E . J . Aero. Sci., Vol. 22, N o . 11, p p . 7 3 3 - 7 4 9 , Nov. 1955.
ENGLERT, GERALD W. and LUIDKNS, R O G E R W. "Wind-tunnel Technique for
Simultaneous Simulation of External Flow Field About Nacelle Inlet a n d Exit
Airslreams at Supersonic Speeds" NACA T4V388I, 1957.
EVVARD, J O H N C. Diffusers and Nozzles—Diffusers and Air Intakes. Vol. V I I of High
Speed Aerodynamics and Jet Propulsion—Aerodynamic Components of Aircraft al High Speeds,
eh. 1. sec. E, A. F. Donovan a n d H . R. Lawrence, eds., Princeton Univ. Press,
1957, p p . 586-638.
E V V A R D , J O H N G. a n d BLAKEY, J O H N W. " T h e Use of Perforated Inlets for Efficient
Supersonic Diffusion" NACA T N 3767, 1956. (Supersedes NACA R M F-51B10.)
1 1 KKI. ANTONIO and N c c c i , Louts M . " T h e Origin of Aerodynamic Instability of
Supersonic Inlets a t Subcrilical Conditions" NACA R M L50K30, 1951.
27
E. J . MANGANIELLO
FLEMING, W I L L I A M A . , C O N R A D , E. W I L L I A M a n d Y O U N G , A L F R E D W . "Experimental
Investigation of Tail-pipe-burner Design V a r i a b l e s " NACA R M E50K22, 1951.
FLEMING, WILLIAM A. a n d WILSTED, H . D E A N . " T u r b o j e t Research Techniques
Utilized in Altitude Facilities." Preprint No. 75, SAE, 1953.
F R A D E N B U R G H , E V A N A. a n d W Y A T T , DEMARQ_UIS D . "Theoretical Performance
Characteristics of Sharp-lip Inlets at Subsonic Speeds" NACA Rep. 1193, 1954.
(Supersedes NACA T N 3004.)
G A B R I E L , D A V I D S., W A L L N E R , L E W I S E . , LUBICK, R O B E R T J . a n d V A S U , G E O R G E .
Aero. Eng. Rev. Vol. 16, N o . 9, p p . 54-59, Sept. 1957.
HEPPLER, HERBERT, NOVIK, DAVID and DANDOIS, MARCEL. "Some Dynamic-
Characteristics of a Turbojet Engine for Large Accelerations" NACA R M F-52H04,
1952.
HERMANN, R U D O L F . Supersonic Inlet Diffusers and Introduction lo Internal Aerodynamics,
Minneapolis-Honeywell Regulator C o . , Minneapolis ( M i n n . ) , 1955.
HUNC.ZAK, H E N R Y R . a n d K R E M Z I E R , E M I L J . "Characteristics of Perforated Diifusers
at Free-stream M a c h N u m b e r 1-90" NACA R M E50BO2, 1950.
K A N T R O W I T Z , A R T H U R a n d DONALDSON, COLEMAN D u P . " P r e l i m i n a r y Investigation
of Supersonic Diffusers" JV44C44 WR L-713, 1945. (Supersedes NACA ACR L5D20.)
KETCHUM, J . R . a n d C R A I G , R . T . " S i m u l a t i o n of Linearized Dynamics of Gas-
turbine E n g i n e s " NACA T N 2826, 1952.
LUIDENS, R O G E R W . a n d H U N C Z A K , H E N R Y . " P r e l i m i n a r y Investigation of Cone-type
Diffusers Designed for M i n i m u m Spillage at I n l e t " NACA R M E 7 K I 9 , 1948.
LUNDIN, BRUCE T . Ramjet Test Facilities. Vol. X I I of High Speed Aerodynamics and Jet
Propulsion, ch. V , sec. E, Princeton Univ. Press. ( T o be published.)
M O E C K E L , W . E., CONNORS, J . F . a n d S C H R O E D E R , A. H . " I n v e s t i g a t i o n of Shock
Diffusers a t M a c h N u m b e r 1 -85. II—Projecting Double-shock C o n e s " NACA R M
E 6 L 1 3 , 1947.
MOORE, FRANKLIN K. and MASLEN, STEPHEN H. " T r a n s v e r s e Oscillations in a
Cylindrical Combustion C h a m b e r " NACA T N 3152, 1954.
O T T O , E-DWARD W . a n d T A Y L O R , BURT L . , I I I . " D y n a m i c s of a T u r b o j e t Engine
Considered as a Quasi-static System" NACA Rep. 1011, 1951. (Supersedes NACA
7W2091.)
PEARCE, R . B. Aviation Week, Vol. 52, N o . 3, p p . 2 1 - 2 5 , J a n . 16, 1950.
PERCHONOK, EUGENE and HEARTH, DONALD P. "Effect of Ambient-temperature
Variation on the M a t c h i n g Requirements of Inlet-engine Combinations a t Super-
sonic S p e e d s " ^ £ 4 4 T N 3834, 1957.
SANDERS, J O H N C. Control of Supersonic Propulsion Systems. Paper presented a t
Sixth Anglo-American Aero. Conf. ( E n g l a n d ) , Sept. 9 - 1 2 , 1957.
SANDERS, N E W E L L D . "Performance Parameters for Jet-propulsion Engines"
JV44C44 T N W Q o , 1946.
SCHMIDT, ROSS D . , V A S U , G E O R G E a n d M C G R A W , E D W A R D VV. " D e t e r m i n a t i o n of
Surge a n d Stall Limits of an Axial-flow Turbojet Engine for Control Applications"
NACA T N 3585, 1957. (Supersedes NACA R M E53B10.)
SCHUELLER, C A R L F". Inlet-engine Matching a n d Control Systems for Supersonic
Airplanes. Paper presented a t Journees I n t . de Sci. Aero., Paris, F'rance, M a y
2 7 - 2 9 , 1957.
SEARS, G . A . a n d BAJEK, J . J . Testing Air Breathing Supersonic Powerplants. Paper
presented a t A R S Meeting, San Francisco (Calif), J u n e 10-13, 1957.
SIBULKIN, M E R W I N . "Theoretical a n d Experimental Investigation of Additive D r a g "
4V44C44 Rep. 1187, 1954. (Supersedes #440.4 R M E51B13.)
S T E R B E N T Z , W I L L I A M H . a n d DAVIDS, J O S E P H . " A m p l i t u d e of Supersonic Diffuscr
Flow Pulsations" NACA T N 3572, 1955. (Supersedes NACA R M E52I24.)
STERDENTZ, WILLIAM H. and E W A R D , JOHN C. " C r i t e r i o n s for Prediction and
Control of Ram-jet Flow Pulsations" NACA T N 3506, 1955. (Supersedes NACA
A M E51C27).
28
63
un iinpojd |B.iatB| uoddns a[ :jaa.uoa sed isû JSBIUOUI ,»J 'adpuud U-J, •(££ •Sij) jo|na
ap siajja sap a p m ^ p JSEIUOUI 3| 'ssuuonuaiu 3 p iuu inb s.i|dui.ix.i s.i| inued ii'J.)A3pj »f
*833npiB sanbpnb u sjnoaaj aio.\E4p
.I|C(I:MI.KUI|IIII juop isa |[ saiuEsijjnsui luauuaiA.ip s.qp nn st.i s.>p s.mofiio) EJJAnoji is |i
•so[C|iuods|p saijaujnos sap a||iBj E| JIGS anb a|pnb :<)|piuESuEf\: anbqdxa 8,J ammo.) M|paj
a[pqa3 t | B SIBSS3 sap |os ns jamaap.i p 3|qiH0d sinofnoi srEd tsau |i iuaujasnajnai||Ej\ c
uois|ndojj E[ ap miii|Eij?ds
sa| ia anbuuBuApojayi ap S.IJSIJKI ).j(ls saj ajjua aqqEia is.is nil) .iini.n.i tnjd ua sn|d ap UOSIEIJ
E| red sanbqdx.i uns.mi agjE| aun suBp IUOS oipiUBSuEjAJ JEd sanbo.\a apmqimis ap sjnapBj
sap aua.\noaap B| ja sauauiouaqd sap anbisXqd as.<|EUB| suEp si|dui<i.i.n: s.uSiijd s.rj £
luEAins ji) np saaans a( aaJHSSE B lujns a|0||d ajqiuEqa E| ap aaiiEssind
E| ap inauiaajojuaj iir} |o.\ ua aaqaa jaiui.ud a| janbqdxa B iiBsujns inb aa 'jaqnaiuBd
KB3 33 SUBp apuoaas ap sauiaixjp sanbpnb ap aii|B3J ua JiBia •s.inan-Ajasqo XIIE ajpajo i
V a|i|B.)8i|83u njBd 11BAB 'siEssa sjaiuiajd B3| susp 'inb aSEuinuep iBpp a] anb luatuaiE
-ipaiuuji EJ3A3J auaujnos ua aj3[duioa sn(d as.\|EUB a||a.\nou aun ' J Q "|<>S HE siEssa sap supqEA
i!| .i|> siA-r-sj s amspiidaas uiri.M.i mi jaqasns ,i|i sril minimi ,>n M|iii.>nnii .iii.iiiiiinai|d .1; )
sajiEijixnE sjnas|ndojd sap .JSESJEI np juaiuoui 11c iii.nuaj3a.uo3 jaunqp: s ap Esnpj uiSuaj
•aiuBS|BjsiiBs J|BJ B inoi njsd JIBAB inb „ajqq iaf,, ai.i.i|||n<is 11.> |Uiod HE asiui aun ssjdy
jnajaEajojEjs un.p a p n p 1 ap sodojd E 'saauuE
sjna[sn[d Epp E A |i 'nasA a[duiaxa un njjajp af 'anbjBtuaj ajaiui.ud anaa jajisn||i j n o j
•sajqEsuadsipui
luaunuapiAa sap;uij CUOIJBOUUSA sa[ jnod anb iii.mi.inbnBjd |U3Ua3lAJ3]U1,U la-xna^ -JOA 11.>
t m n xiiB sjnoaaj a| SUBS UOIABJ E uoiiEidBpB uos ia jnajoui u n p aja]dmoa jmod nc isim iq
,1.1111 i.i[(ap a|q|ssod yfap isa p.nb .laiujujB mad uo 'ssssd 3| susp asuupi- IU.IMIOS DOimdoj
1; luamajiBJiuoQ qos nE jnaiour ap siBssa sap anbiuqaaj B( suEp saauuB sanbpnb sindap
sqdmoaaB sajScud spuBjg sa[ pjoqBp inoj isjauSqnos af | :('V"H"3'M'0) aiaraavj) ' J
NOissaosia
"SS6I 'iONISH IVH VDVN . ^ ^ H S

jap[oqauiE[.j snoue_\ |o sapsuajJEjeq^ ,«o[j-3>|E^\ [Emjaqiosj j o Apnit; jeiuaui
-uadx-j,, - y ivvmi/VV 'NasnaMoijAj p u e •% OIAVQ ^ a m n v f ) '-Q a o a o a g 'HaoNno^
'IS6I Y O ^ O S a t W VDVN .4C8I Jaqmnfj qoBjAj JB sjasnyifj iuaSj3Aip
-juaSjaAuoQ j o UOIIB3IIS3AUI u y „ --y AMNaj-i; 'MVZONinH p u c - rj smci>jvi\afj 'JU.VAJ\\
"9 C 6I ' J 3 Q ' ( 3 3 U E J J ) SIJBC[ ' p u B j n o t q n d o j j p u e uonsnqtiio;} ( J > [ \ ' ) \ 'iiunaaiv
ipti3A3[y -suiojqoj,! 3>[Biuj J i y aiuosjocln^; j o A\3iA3-y y •([ sinrj>ivi\a{T 'J-XVA^
'ES61 'ZlOi'.NJ.
VDVN ,.%u\l[:>''leVi ]3|ut-auiiiiia-iafoqjnj^ j o sisAp;n\ u \ ., -(j sinrjHvpvaQ 'XXVAW
"IS61 'OEHISa I\'H VDVN ..UOUBSIJSOAUI SupJEjs-apnjiqv
amSuo-isfoqjn [ j o sqns3^[ AjEuiuiqajj,, ;A •(' 'ONOaxntHV P " , ; CI H 'a3J.STiA,\
•0561 'OEHOSa IV}/ VDVN , , n m x m d m o i j SIJ p u c aatSng janodoid-autqanx j o
33UEIUJ0JJ3,! j o UOIJBSIJS3AUI o p n i p i v ^ • [ NU.HVJ / \ ' r a w g puB -g si.waq 'aaNTTV^\
'6F6I ' I I 0 6 3 W i l VDVN âauBuijqjjsj jossajduioo MO ( J
-[Bixy uo jaajj'j j a q u i n \ j sp]ouA3^jM -y wvmiAV 'ONIIV.I i.[ p u e r j si.u-iq 'aaNiav^w
"2S6I ' 8 0 a ? G a IVel VDVN . . s j a q u i n ^ sp|ouXa>j[ lapii-auiSug j o 33UB>J B JSAO
3UISU;.[ idfoqjnj A\o[(-|Bixy UB JO uopEn|BAg 3auEuijojja c j ||E-.ia.\() put; itianod
-U'°;),, \ \ s r r n \ \ "-i I.IV.WII I.IV>I;I put: ;) .V-IMIIS "A 1 I.IN.IJT ' " [ si.t.Mn;^ 'HaXTVy\\
Y C 6I '•'dy 'SSt-ISf' 'dd ' c -o_v % l - \ o j \ ' g w S V '™ m X O 'nsvA
(•ggg[ puB ggQl w x VDVN sapasjadng) -g^gl ' i m '<f>if VDVN ..""I'lPPN
)uan[i(| j o siaajjg a u i p n p m BjJBqg,, •CIIVNOQ ' x a v o o g p u c CIHVHOI>J[ - J 'aaNJsnj,
•«<')!'."I IVH VDVN sapas.iadn S ) -qgggl
' C 92I UJ. VDVN <t s J :>3 iia 3 I B 3 S S u i p n p u j uotjESpsaAuj iBiuauiuadx'j UE puE siaf
UIE^[ ui o p m q d u j y uoiiEspi,[ zznjj puE XiqiqEit; JOJ Ajoaqj y „ ' a iHaao-y ' w w i a x
(•8IVZS1IVH VDVN S 3 P J S
-jadns) -B9S6I 'qb9£ N J . VDVN i ^ i 0 3 m . S A B M XJBUOIJEIS-UOJSJ [Euoisuauiip-auQ
psyipojAi B Xq sjaf uiBAj oiuosjadns ui Suizzng j o sisAJEuy u y „ -'j jajattoy 'idwiaj
SWJiaOMd H31SAS MOISTIldOHd

E. J . MANGANIELLO
Fig- 32(a).
Montage classique; ecoulement de revolution
Fig. 32(b).
Montage adaple a line tuyere plane
Fig. 33(a).
Montage classique
Prolong a l e u r
Fig. 33(b).
Tuyire adaptee
30
ii(i|s|ii<l(ud aSjE| E ui ji isai p u s uonB||E)sui aqi j o p p o i u a[Eas | | n j B p|inq uBa noX J O 'XjqpEj
jnoX ui XqaBdEa MOUJIB a q i a.\Bq noXji UOUB||BISUI aqi jaf-aajj UBa no_\ -auin pE3| uopEJEd
-ajd puE i s o a j o s u u a i ui aAisuadxa ajE qaiqM j o ||B 's|x>qiaui p B x a ajoui j a q i o ajE ajaqj^
•auq ||Bis aqi U| a S u E i p E j o s u o n a i p a j d asBq 01 qaiqM uo sqnsaj aAUEjipmb
atuos ajE p a u i u u a i a p aq UBa |Bip ||B is.iq i y auo|B sua.uasjo asn aqi Xq auiSua U B J O 3 3 B J l u o j j
3qi IB sapjojd Moy [BiuaB 3 3 n p o j d 3 j oi a|qissoduii a q n b s; 11 lEqi UMOin| [\3M si i j •juEjjodtui
XJ.IA HI; a a u E u u o j j a d amSua uo suoiwoisip MOU ia|U| j o sjaajja aq L :UOIIJOISIQ AXO|J i.i|ii|
•suouipuoa jixa a[qEi.iE.\ .IAEIJ a.w p pip aaun a.w UEqi ajoiu
a p i q B A\ou>| oj a.\Eq 3_\\ 'sdEUi a a u E u u o j j a d XpiBjnaaE p q i . u d 01 papaau si XjiaiuoaS jixa ui
s.ijiin:qa jo aauatqpii .n[\ "xapin .ia((iunu s^ppuXa^vj puE paads auiiuia ui saSUEqa oj SB JJ.IW n
BajB jixa ui saSuBqa 01 p u o d s a j A\OJJJIB juBqnsaj puE saijsuajaEjEip 3uidujnd aqi aauis Xfddv
O) qnatjjqj ajoui a p i q B si p a q u a s a p a n b m q a a i aqi 'sajJEA E3JB jpca 3qj 3jaq.w auiSua UB Uj
•XjjauioaSJ jixa 3|qEiJE.\ ' j E p a p J E d ui—.ujauioaij .qqEiJEA j o laajja aqi p a y n u a p i X|ssajdxa
lou SEq puE m a i x a UE OJ p a / q E j a u a S SEq O [ P | U E S U B J \ ' j a d s d siq j o uopaas siqj u j : U O I I E 3
-IJS3AUJ aauBiujojjaj auriiug : ( a j j u a ; j j u a u i d o p i a f j HuuaauiHu'j p ( o u j y ) Mnag a q NJ >[
qEjajBj i j o d d n s jEd aiJEiuoui HE a j q s j s j s j d ajja 'ssa suiEuaa

suEp ' m a d inb p'm.nnu.idx.i .i.iqji.n; mi j a p d d t M ap sjiiaqiBp liBia apaaojd , u )UE|Eu2is UÎ
iiiiiin.ilin u u | \ •iuauia|<|BuaAuoa sail sjnojhnj aaiuasajd.u luaui.uiESsaaau sBd Bias au aiiujq
aqanoa ap j n a n i E q E| amaiu ,ttj -aiuaiap ap a u o / tq sutqi xiiEsja.\suEJi uoiss.ud a p siuaqiiuji
sap IIOSIBJ u.i aqBjjed SBd j p s au ajiuiq aqanoa B| a p a u p u i X s E| a n b 'lajja ua 'suiEja a f
'..uiEuiiuos saji u o i j E i u a u i u a d x a a u n t n b y^lTIMOd ¥ iqns ajoaua E 4 U anbipui aSEjuoui ,i\
JIM •inispp;uo([ jnd saasod s u o p s a n b xnap sap aunaiiE E apnjijJaa aa.\E a j p u o d a j xnad an .if
:(V>JHXO) a a a m a v o d
jsjsaj a|Bas-||nj ui sjnaao UBqi jaSjB] q a m u Sll|,iq XpEiujou OIIEJ siqj ' s a a n p o j d
uoiiEJiiSquoa s u p qaiq.w J-il-nuBip Xpoq 01 ssauxoiqi J.)XE[ X j s p u n o q |u OIIEJ .1 AiiBjuas.ud.uun
•up |o Xqnamq) aqi J.IAO Ijuinaii ui a a u a u a d x a Aire SBq .uaujB;} JI A\OU>( OJ a))q os(E p[noqs j
csIIE"
.iiiE(d aqi 01 s.qSuE iqSiJ IE puE p][EJBd sauE[d aqi in .HUES .up iou ajE qaiqA\ s.woy a a n p o j d
saop .iiibuiqi.il sup s.iauauadxa s ajaujEQ ui jaqiaq.w A\ou>( OI a5(i| p[no.v\ j p p o u j aqj punojB
siuaui.i.insE.mi a.iiiss.ud ,isi:q put; .ijnss.ud n u n s Xq u.wnqs aja.w .mini 'J.IA.IA\O[J saiJiaiuiuXsE
A\O[( XUB ajnsEaui jou p | n o a j os 'sqE.w auiqd aqi oj pqEJBd ' a u q ajjuaa p p o u j aqj q S n o j q i
auB|d E Uj ja.\E| XjEpunoq [apoui aqi qSnojqj sjuaujajnsEaui asjEUi X[UO p(noa ] "SHEAV pqBJEd
oMi puE s||EM p a d s q s o.wj qii.v\ | a u u m E UI O S E sjEaX a.\q auios anbiuqaai siqj pasn j
•jEOJqj p u u m pui.w aqi j o uiE.uisdn p a u o d d n s
s| |.ipnm .up q,i|i|A\ ui pauoiiuaiu aq sanbiuqaai aqi inoqB . u a u j E ^ -JAJ !)SB OI a>|q ppio.w j
: ( a i j p a ( g q s q 2 u g ) NOSfnvNOQ •§ ' j
•ajiEJiiqjE IIBJ E I'IOI loiu.iiq slid t;jin;jt:d ,iu sauqdiasqj x n a p

s.q aJiu.i a u u i q E| is j a p u B i u a p as .ip iin.ip u.i aiuaui isa U Q sassaii.uadXq sap auiBUiop a| suBp
suapuiiEuXpojaB sap a m o p SUES aatpuac) ni: Miu.iAt; | simp j a i i m a a a B s .nib mad au u o i s j n d o j j
tq |.i atlbiuiBuXpojayi .1.1111.1 aqqina isas nib asnanjJiuj is Elap uoiiBJoqB(pa Eq—f,
•aauapiaui
SUES siEssa sap j n o d ' a p p o u i n p a q a q a a j juaui3[q|suas jipuBJiJEjj j n i u a d ( a p E j s q o j ap
JUBAB ajuiod tq .i|> ajduioa uiEU.ii ua ,ia|n.qt;aj iuaui,i|nnaaj ap lUBjnoa ap aaEJJns aun juBAins
apjojd jna|BSuo|ojd u n 4 p (qcc "Sij) aoponofpB,'! 'ajaXnj B| a p saj.\a| sap anssi uopBqjnjJad
E| B a n p app<iui n p a | | a q a a | a p uoiiEiuuq E | ).i .inbisstq.i IHEIUOIU ,I\ a j i u o m Egg a j n S i j
•ajqq iaf ua a n b i u o s j a d n s iBSsa^ UOIIE|]EISUI
a u n ( p saiqiqissod s.q ajpuajp B jiAjas 'ISSOB in] 'jnad inb aaijiUB a j i n s un [BjapuSis af " u q u g
•aSsuajBa n p lUEpuodsajjoa j u a m
-aSuEqa u n apossB ajia j p p ajaXnj a p luamaSuEqa m o j B n p u a i u a uaig 'ajEiiiui ajaXnj E| a p
l u . n u . q n o a a j SBd a u o p a q j n u a d au .ISEIUOUI ,13 apaE) isa ua p i a p a a| : j . n p n j a p J I S B ^ p , n b
sdjoa-.uaijJEj] a p IUOUIB uopaas B| j n s a m d d s ^ juBjnoa a p a q m .1,1 MuiEisixa a | p u u o i s u , n u i p u [
ajsXnj a u n p |EIXB jUEJiioa a p a q m a( jasqBijaiBUi B 3jsisuoa p :apoiutuoa snjd a S s i u o u i
un ajaitSns q^g .un>i|.[ ' a n b i d o a s o u i s uopEAjasqo] qiaqjip a j p u a j 3p )3 3)_Biaads ajaXni
inn J 3 8 p s ( p jsa juapjsAuoaui u o $ ' j o p a nE xna/BS jaf a| j a u a u i B ^ j a u u a d ajpui[.\a ,i.y
ixtiEj-E-ajjod ua aiuoui 1.1 I U O U I B J ap )UEU3A '[EIXB ajpuqXa uti 4 p ajduioa juBuaj ua aa(nap3a
is.i .1.1.1X111 a n a ; ) - (EJ;C "X1 _.[; IIOI)II|OA,IJ a|i .1.1.1X111 ami J.isqnn i: .usisno.i Mud siqd ia '.nibissiqa
aiJEjuotu u f l -jopia nE uoissajd B|—SE3 suiEjjaa s u s p juauia[qEjapisuoa — a t | j n u a d nib ,ISB[|IS
SWaiflOHJ K3J.SAS NOISTilHOUd

E. J . HANOAMELLO
wind tunnel. Another method is the ducted nozzle technique which has been used quite
extensively al the Arnold Engineering Development Center.
Our combined and mutual objective in the testing of propulsion systems is to aid the
designer and the manufacturer lo solve as many of die problems that exist within a propulsion
system as it is possible to do before the engine itself is released for production and for Right.
The NACA, through its long experience and through hs early identification of problems which
may be encountered, has already forewarned the designer. The designer may use this
information to produce specific designs for installation in specific pieces of Hying hardware.
The AEDC then may assist the designer in the final development of the hardware by per-
forming an extensive development test program. Once an airplane begins its flight test
program, the pilots can ill afford to spend very much time worrying about the power plain.
The engine is naturally expected by the pilot to function properly and to produce the
necessary thrust al the right time. The pilot's basic objective is to lest the airplane as a flying
vehicle, not the engine. In the flight test of engines, it is usually best to use a proven airframe.
one that has shown itself to be stable and reliable. It is also important, however, thai lliis
chosen engine Might test airplane not be limited so as to deny exploration of die extension of
performance for the engine.
It is our firm conclusion that very thorough ground testing greatly improves the proba-
bility of real success of the combination engine-airframe and correspondingly, to a very great
degree, reduces the amount of flight testing required to verify airframe performance and
reliability. Testing in ground facilities also provides an opportunity to gather incidental and
important information not only on the overall engine performance but also on its mechanical
characteristics and the reliability or shortcomings of the engine and its several related
components. As an example, in one engine development program at Tullahoma. we tested
16 engines of varying configurations and logged more than 1,200 hr of test lime al alliindes
ranging up to 811.0(10 It and at Mach numbers up to 2. To have accumulated this amount of
time in actual flight would have required an estimated a1.000 lest Mights and ( in a years of
relatively hazardous operation.
During this testing, many incidents in the manner of mechanical and component failures
occurred. These failures and incidents were valuable to the development of the engine.
Every failure, every incident amounted to the gathering of flight experience without ever
leaving the ground. This incidental gathering of information can be and is used to render I he
product, the engine, a more valuable and a more reliable machine. In 100 hours of testing
early in this program on an original version of the limine, lid incidents occurred. In a later
version of the same engine tested later in the program, only 15 incidents occurred in the same
amount of test time, showing a marked improvement in the reliability of the engine.
This part of the value of ground testing was not stressed by Manganiello. 1 bring il up here
to show and to substantiate reasons for desiring extensive ground testing of propulsion
systems. It is important also to note that the testing accomplished during ihis lesi program,
for the most part, simulated exact pressure, temperature and Mach number conditions al the
entrance and at the exit of the engine.
AUTHOR'S REPLY
E. J . MANGANIF.LLO: Carriere's mention of some of his organization work on nozzle and
free-jet testing are interesting and welcome additions to the material 1 have discussed and
provide further illustrations of the many facets of propulsion system testing.
Le Beck has outlined the propulsion listing conducted by AEDC which is very similar
to and Complements that conducted by my organization. I agree wholeheartedly with his
premise that generalized research testing is not the complete story and that detailed testing of
specific components, engines and installations is required to obtain the final answer. Finally
I am happy thai he agrees with my conclusion that ground test facilities provide many-
advantages compared to Might lesls. However, we must not forget that, even aller exhaustive
testing in altitude test chambers and/or wind tunnels, some Might testing musi lie performed
to complete the long and complex process required for development of successful aircraft
propulsion systems.
32
SOME SELECTED PROBLEMS IN
ENGINE ALTITUDE TESTING
By BERNHARD H. GOETHERT
Engine Test Facility, ARO, Inc.
Engine Testing with Simulation of Air Inlet Distortion

Various schemes presently employed in the Engine Test Facility of the Arnold
Engineering Development Center for determination of engine performance as
influenced by inlet airflow distortions are described. The simplest method, dis-
tortion screens installed in the cylindrical inlet duct lo the engines, and more refined
schemes such as the ducted-nozzle simulation technique and the free-jet testing
method are briefly described, and their merits and special features are discussed.
Typical results obtained with these methods, as well as the limitations of each,
are presented by a selection of data from full-scale tests and model studies.
Transient Testing
For reliable transient testing nf propulsion systems—that is, for investigations
requiring quick changes in the thrust level of an engine—precise and rapid control
of the inlet and exit conditions of the Engine Test Facility are required. The
human operator is not capable of establishing such rapidly changing test conditions
with manual control, even in the case of the thrust level changes of relatively small
turbojet engines. Studies made by the Engine Test Facility staff to determine
theoretically and experimentally the response characteristics of the uncontrolled
plant are described, and typical experimental data are presented. By means of an
automatic control system similar to a conventional electronic analog computer,
it is possible lo improve the dynamic response of the plant control significantly al
all pressure levels by selecting the optimum values for the gain and time constant
in the employed "error-proportional" and "error-integral" control system. The
experimentally determined response characteristics of the test facility for a wide
frequency range of artificially produced disturbances are presented. The successful
application of this new automatic control system is demonstrated using data from
a transient test of a turbojet engine in which a large massfiow change occurred
within a few seconds.
Correction of Distorted Transient Test Records

In propulsion testing, particularly in experimental investigations of the thrust
buildup and decay of rocket engines, the process of thrust change occurs so rapidly
that the natural frequency of the measuring system is often not high enough to
yield distortion-free data. A specific case is used to demonstrate how, by means
of a system analysis, distorted records can be greatly improved when the response
of the measuring system to sinusoidal disturbances over a wide frequency band or
to step inputs can be experimentally determined.
Use of Ej'ectors in Engine Test Facilities for
Augmentation of Test Altitudes
The altitude range of engine test facilities having conventional rotating exhauster
machinery can be significantly improved by the addition of ejectors with very low
33
B, II. OOE T H E R T
secondary flow. Several ejector schemes are discussed, and experimental data are
presented which show that the steady-state altitude volumetric capacity of any
given exhaust lest plant is increased by a direct multiple of the pressure rise ratio.
For high altitude testing, ejectors perform as the natural first stage compressor-
device discharging to the inlet of a conventional exhauster. For transient testing—
that is, during thrust buildup or thrust decay- such ejectors dn not function properly
since sufficient exhaust energy is not available in the engine to drive the ejector. A
special arrangement nf auxiliary ejectors is discussed, and test results are shown
which demonstrate that smooth transition from auxiliary ejector action before
engine operation to main ejector operation after ignition of the main engines can
be obtained. It is more difficult to solve the reverse problem during throttling down
or slopping of an engine; however, for such cases encouraging results have also
been obtained.
INTRODUCTION
IN THE development cycle of a propulsion system, extensive ground facility
testing is required before the engine can be released for final checkout in
flight. T h e initial phase of the facility testing of new engines is usually
accomplished at the manufacturer's p l a n t ; development testing, which
i ltd in les simulation of high Vlach number and high all it tide High I cm id it ions,
is done in large engine test facilities usually owned and operated by specialized
testing agencies. Such a division of the testing mission is practical in view
of the high initial and operational costs of an engine ground-test facility
which tests at high altitudes and high M a t h numbers.
Since the last war, a n u m b e r of such high-altitude test facilities have begun
the testing of jet engines. All operate according to the same basic scheme.
T h e test engines are installed in altitude chambers, and air at the proper
pressure, temperature and humidity is provided to the engine inlets. After
the air passes through the engines, the exhaust gases are directed through
coolers and scrubber systems to a set of exhausters for final discharge to the
atmosphere. Such testing has become quite routine and the test facilities
involved differ mainly iu capacity; in some the air supply, in others the
exhaust machinery, is the limiting clement. In this basic engine test field
the possibility of many refinements still offers challenging problems and
significant rewards. Some problems are, for example, continuous or inter-
mittent lest operation, high temperature heaters and structures, simplification
of the complex test plant, the rapidly moving field of instrumentation.
I n recent years, the engine test facilities have shifted from the more
routine testing to a variety of highly specialized testing. In the case of the
Engine Test Facility of the Arnold Engineering Development Center,
for example, more engine operating hours arc presentK devoted to.specialized
tests ihan tn the straightforward direct-connect type of engine testing. Some
of the problems which require a growing part of the facility operating lime
a r e : engine controls problems arising from throttle bursts or rapid flight
nmucin ers, interference between the engine proper and the inlet particularly
at supersonic speeds), combustion stability, ignition and blowout limits at
transient conditions, extreme altitude testing with ambient pressures of less
than one per cent of sea-level pressure, engine operation with special fuels,
liquid and solid propellant rocket testing.
This paper will not attempt to cover comprehensively the wide field of
34
SOME SELECTED PROBLEMS IN E N G I N E A L T I T U D E TESTING
specialized engine testing. However, several typical test problems have been
selected for more detailed discussion. These problems, basically unrelated,
are among those which have been given extensive attention in the Engine
Test Facility in the recent years. They deal with:
Various methods of engine-inlet duct matching
Plant controls and instrumentation problems of transient testing
Augmentation of plant exhaust capacity for steady-state and transient
testing at extreme altitudes.
In the following sections, these selected problem areas, the test methods
employed at the E I T A E D C in the indicated fields, and some typical
test results are discussed.
I. S I M U L A T I O N OF AIR I N U T D I S T O R T I O N S FOR
AIR-BREATHING ENGINE TESTS
Definition of Problem
During the first phase of ground facility testing of a new engine, the
supply air is generally ducted to the engine in such a way as to maintain
uniform flow distribution over the inlet area. In actual installations of
the engine in airplanes or missiles, however, such favorable conditions
do not exist. Despite extensive efforts to develop inlets with uniform velocity
or pressure profiles, the most refined inlets still show significant variations
in total pressure over the inlet duct area. This is particularly true of inlets
designed for supersonic flight since external and internal shock waves
frequently produce boundary layer thickening or local separation.
T w o typical inlets and their Mach number distributions at the down-
stream end of the inlet d u d s are shown in fig, 1 *. These inlets, which were
designed for a moderate Mach number of 2, are of the side-inlet and annular-
rotationally-syimnctric-inlet type, respectively. The Mach number variation
ii\ci' the engine inlet area ranges between If) and 2H per cent of the mean
Mach number in the (light Mach number range from (Hi to approximately
2-0. Converted into total pressure terms, a Mach number non-uniformity
of 30 per cent corresponds to a 16 per cent variation in the total pressure
at a duct Mach number of 0-45. Although inlets arc known which, at the
design point, are considerably better than those shown here, at off-design
conditions flow non-uniformities quite frequently occur which are consider-
ably larger than those shown in Fig. 1.
It is obvious that the operation of an engine will be influenced significantly
by such inlet flow disturbances. For instance, reduction of the compressor
stall margin, instability of combustion, generally lower combustion efficien-
cies, local over-heating of the engine, etc., frequently occur as a result of
inlet flow distortions. Consequently, after completion of the uniform inlet
flow tests during the development cycle of an engine, additional tests must
be carried out with configurations consisting of the engine proper and the
inlet so that inlet flow distortions can be properly simulated.
In testing "engine-inlet" combinations il is necessary to provide engine
inlet air at the correct flight velocity, usually at supersonic values, as can
be done in propulsion wind tunnels or in free-jet test facilities having
supersonic nozzles. Such tests actually simulate the inlet duct conditions—
• Notation is given on page 78.
35
B. H. GOETHERT
shock-wave formations, boundary layer development, etc. This type of

testing, however, often becomes very expensive and time-consuming because
rather large facilities are required. For example, in free-jet test facilities
no more t h a n half the air provided by the facility flows through the engine;
the other half must bypass it in order to assure interference-free inlet condi-
tions at the engine inlet. Free-jet testing, therefore, approximately doubles
the air-handling requirement of the test facility over that required by simpler
testing methods.
Airflow
Airflow
Configuration A -Dual Fixed

D
Half-Cone Inlets
Configuration B -Translatlng-Cone
Nacelle Inlet
0.40 1
Ref. NACA Wind
* a Tunnel D a t a
o >
r-t a Configuration B
0.30
m~%
44 aa
w
\
o< ]
1
0.20 1
- - >
Configuration A
i
S S o.io
*> 71
9
u ai 6
o u
mi a,
«U
-I o a
0.4 0.8 1.2 1.6 2.0 2.4
Flight Mach Number
Fig. 1.
Variation of distortion at the compressor inlet as a function of flight Mach numbers
Frequently the necessary air-handling capacity for free-jet testing is

not available in a test facility. In such cases, the air can be ducted directly
and without appreciable spillage to the engine inlet so that approximately
half the capacity required for a free-jet facility is needed. I n such "direct-
connect" testing, the sensitivity of engines to air inlet distortions has been
evaluated by means of simple screens installed in the inlet duct. Such screens
allow simulation of various types of inlet distortions such as annular disturb-
ances occurring in rotationally-symmctrical inlets or sector-type disturbances
occurring in the side inlets or bifurcated inlets. T h e screen distortion method
has been used mainly to evaluate the sensitivity of a given propulsion system
with regard to an expected type of inlet flow disturbance. With standardized
screen assemblies it is possible to compare the disturbance sensitivity of
36
SOME SELECTED PROBLEMS IN ENGINE A L T I T U D E TESTING
different engines and to obtain valuable insight into means of developing

engines having reduced sensitivity to inlet flow distortion.
T h e disturbances produced by standardized screen assemblies are, in the
main, very unlike the disturbance profiles produced by actual inlets in
flight. A refinement of the disturbance screen method, the "ducted-nozzle"
technique*, has, therefore, been utilized at the A E D C for several years,
particularly for the testing of large ramjet engines. In ducted-nozzle
testing a large number of movable screens or disturbance flaps arranged in
the upstream portion of the inlet produce the desired disturbance profiles
when the inclination of these elements is adjusted. Furthermore, when the
duct is shaped internally to match the actual contours of the airplane or
missile, a reasonable degree of similarity between the ducted nozzle inlet
distortion profile and the inlet profile measured in supersonic wind tunnels
or in flight can be attained.
These three methods will be described in more detail in the following
sections.
Inlet Distorlinn Screen Method^

General arrangement.—A typical distortion screen assembly utilized in tests
in the Engine Test Facility is shown in Fig. 2. A support grid over which
screens of various densities and arrangements can be placed is located
upstream of the engine in the inlet ducting. By selection of the density and
the number of overlapping screens, the local total pressure disturbance
profile in the inlet can be controlled.
From among the many disturbance screen assemblies used for inlet flow
distortion tests at the E T F , two configurations (see Fig. 3), one producing
a "one-sector" and the second producing a "two-sector" disturbance
pattern in the inlet of a turbojet engine, have been selected for more detailed
discussion. T h e resulting Mach number non-uniformity at the inlet face of
the compressor is also shown in this figure. For both configurations the
amplitude of the Mach number disturbance is of the order of ^ 2 0 per cent
of the mean value. It is apparent that with the one-sector disturbance
configuration, the rotor blades are exposed much longer to the distorted
condition than in the case of the two-sector disturbance configuration. In
the two-sector type the local Mach number oscillates rather rapidly—in
fact, exactly twice as fast as in the one-sector type—between its limiting
values so that the rotor blades may not have time to adjust their flow pattern
completely to these rapidly varying conditions. T h e effect of such disturb-
ances can, therefore, be expected to be milder than the effect of disturbances
of the one-sector type.
T h e influence of inlet flow distortions can be demonstrated by the results
of experiments on the available stall margin of a turbojet engine. In fig. 1
the operating limits of the turbojet engine are represented for non-distorted,
tvvo-sector-type distorted, and onc-scclor-typc distorted inlet How. At the
high Reynolds number conditions (Reynolds number index = 0-8), the
* This method has been developed for AEDC testing in conjunction with Norih American
Aviation Corporation and Wrighl Aeronautical Division. Corliss-Wright Corporation.
+ I'liese tests were conducted at the Engine Test Facility of the Arnold Engineering
Development Center with my associate, R. E. Smith,Jr., serving as project engineer.
37
Pig. 2.
Typical screen imkdlatum
On* S e c t o r D i s t o r t e d
1.20
—a- _ - o .
J r*~ A
A\
/ J X
MC
0
&0) J 90 180 270 360 0 90 180 270 360
Annular Po.lllon, deg Angular Po.llloo, dag
Pig. 3.
Two types of inlet distortion simulated
38
Reynolds Number Index -
Vs
' ' 0.8 Reynolds Number Index
V / / / A 0.12 Reynolds Number Index
Z
i-
-
T7i 09
~
~
o
-.
Uniform pi
D
-
:
w
-
1, / / / n i I I I / / ' / / / / 1 K
M
tC
Two Sectors Distorted -
O
—
z
-
-
14 —
-i
C
=
One Sector Distorted
6000 6400 6800 7200 7600 8000 8400 8800
Corrected Engine Speed, rpm
Fig.4.
w o
Effects of inlet airflow distortion on surge-free operating range of compressor
B. H . GOETHERT
influence of the disturbances is relatively small. T h e stall margin in the

high speed range is practically unaffected; the stall margin in the low speed
range shows only a slight change.
At low Reynolds numbers, corresponding to high altitude flight conditions
(Reynolds n u m b e r indices of approximately 0-12 and below), the effect of
inlet flow distortions on the stall margin is extremely critical, particularly
in the high speed range. While the engine could be operated stall-free with
uniform inlet flow over a speed range from approximately 6500 to approxi-
mately 8000 r.p.m., the engine with the two-sector disturbance configuration
could be operated stall-free only between 6500 and 7800 r.p.m., and the one-
sector disturbance configuration could be operated only in the very narrow
10.04
±0.42 "^
1 "f- i=-
±0.39 \ *4^_
tM .
•f^.
±0.42
^3 %
• ±u.u 4 - f
X
M* avg.
;?
^
1 w
w*
„•
;i>1 1f
fy.
i
w'
1 J>
4 r^ T G
4 V
1 \
1
s
i
/
/
/
20 1
1 //
7"
i / Êstimated Engine
Operating Envelope
w /
1
1
t
0 0.4 0.8 1.2 1.6 2.0 2.4
Flight Mach Number
UJM Compressor Surge Boundary

With Unllorm Inlet Condition
LiUll Compressor Surge Boundary
With Two High and Two Low
Pressure Regions At Compressor
Inlet
LULL Compressor Surge Boundary
With One High and One Low
Pressure Region At Compressor
Inlet
Fig.5.
Influence of inlet-airflow distortion on the operating range of a
turbojet engine at military and maximum power
40
band between 6700 and 6800 r.p.m. T h e ability to determine the critical
influence of the inlet distortion pattern on the stall margin, especially when
considered in conjunction with the Reynolds number effects, gives such
testing great value. T h e designer is able to determine critical conditions
by means of simple testing and can direct the development of the engine
and the supersonic inlet toward an acceptable compromise.
T h e significance of the reduced stall margin which results from inlet
disturbances can be recognized by calculating the operating envelope of
the engine from the stall margin results shown in Fig. 4. Fig. 5 indicates the
operational limits for a distortion-free inlet at high altitude based on tests
made at the E T F . T h e high Mach number limit is imposed by stall at low
corrected engine speeds; the low Mach n u m b e r limit is imposed by high
corrected speed stall at the Reynolds number corresponding to an altitude
and Mach number. T h e curves for undistorted inlet flow indicate that the
engine has a larger potential for operation than is indicated by the operating
envelope (dashed lines) estimated by the manufacturer. However, when
inlet distortions are taken into consideration, the operating range shrinks
considerably. While the engine could still cover the guaranteed envelope
with the two-sector-type disturbance, the operating envelope for the one-
sector-type shrinks so drastically that at altitudes above 65,000 ft essentially-
only one Mach number is available for stall-free operation of the engine.
It would be desirable to establish standard distortion patterns for the
determination of the stall sensitivity of a large n u m b e r of air breathing
engines at identical distortion conditions. Much more benefit could be
obtained from disturbance testing of engines if the influence of the distortion
pattern could be studied on a broad basis in conjunction with rotor blade
depth, speed, stage loading, Reynolds number, etc. It might be suggested
that the A G A R D Combustion and Propulsion Panel select and standardize
such patterns for universal use in engine testing.
Propagation of disturbance patterns within the engine.—In the preceding para-
graphs the influence of inlet flow distortions has been discussed only as far as
compressor stall characteristics are concerned. Studies of the propagation
of inlet disturbances through the engine proper and the determination, for
example, of the extent to which inlet disturbances are still noticeable at the
exit of the compressor, at the inlet to the turbine, etc., are also of interest. A
typical set of data from one of the numerous tests of this type conducted at the
A E D C is shown in Fig. 6. T h e relative change of the inlet pressure to the
compressor for the one-sector-type disturbance pattern indicates that a
non-uniformity with a total spread of approximately 25 per cent of the mean
value exists. At the exit of the compressor the total pressure is almost
completely equalized; less than a one per cent total pressure variation
remains. As expected, the loading characteristics of the rotating stages
are such that the compressor automatically compensates for the pressure
disturbances. This result also verifies the effectiveness of windmilling stages
located ahead of the compressor as a means of smoothing out the total
pressure profile.
At the inlet of the compressor, the total temperature is practically uniform,
but at the exit (Fig. 6(b)) the temperature distribution is considerably dis-
torted as a result of the different loadings of the individual streamlines. A
41
li. II. GOETHERT
1.10
1.05 _A_
1.00
j\ J
I r\r -J * 1îsf "W
hL \v A pressure,
ii
•t 50% annulus ht.
,u -
, HO i
( a ) Compressor ] nlot
1.05
A Prescuro,
{ A JV .tr • — l 66?; annulus ht.
'2c -*- •4t*

Co) Compressor Outlet
A Tcr^p., 36%
anculus ht.
1.05
1.00
A Temp., 50%
.o:. ^ -n;mIt!H ht .
0 60 120 180 240 300 360

Circumferential Position, deg
(<) Tu rbln 3 In et
Fig. 6.
Typical effects of one sector distortion on temperature and pressure profiles:
Reynolds number index 0-8; rated corrected engine speed
Fig. 7.
Damaged area of first stage turbine diaphragm resulting from local over-temperature
operation (looking upstream)
42
distortion range of approximately 6 to 7 per cent of the mean total tempera-

ture exists. While such a distortion does not a p p e a r too significant at first
glance, the distortion is amplified in the engine combustor. Since approxi-
mately the same amount of fuel is added by the fuel nozzles to the airflow,
whether or not the airflow is uniform and has the same mass flow density
across the entire duct area, or is distorted and has local areas of a much lower
mass flow density, a significant non-uniform heating of the air occurs with
distorted inlet flow. Instead of the 6 or 7 per cent distortion of the tempera-
ture profile at the compressor outlet, a distortion of approximately 12 per
cent exists at the inlet to the turbine (Fig. 6(c)). T e m p e r a t u r e distortions
of such a magnitude are not insignificant. I n view of the high temperatures
at the entrance to the turbine for moderate loadings of approximately
1600°F (2000°R), temperature non-uniformities having a spread of approxi-
mately 250°F exist. Consequently, overheating of structural elements,
especially of turbine stator blades or tailpipe liners, occurs quite frequently,
leading to structural damage of these parts. A typical example of the
damaged turbine stator blades of an engine which was exposed to the
distortion profiles as shown in Fig. 6 may be seen in Fig. 7. T h e turbine
rotor blades do not show any structural damage bince they operate only
intermittently in the high-temperature regions.
Ducted-nozzle Distortion Methods

Description of technique.—As mentioned previously, it is frequently not
sufficient merely to determine the sensitivity of an engine to given standard
types of airflow distortion patterns; it is often necessary to closely simulate
the actual airflow distortion pattern occurring at the engine inlet at various
flight conditions. In many cases the ducted-nozzle method can be utilized
to provide such improved simulation of inlet flow distortion instead of the
more costly simulation method provided by either free-jet or propulsion
wind tunnel testing where the airflow and plant capacity requirements are
much greater.
In contrast to the straight cylindrical duct used in the simple disturbance
screen method, a major portion of the internal contour of the inlet duct is
duplicated in the ducted-nozzle technique. Simulation of the disturbances
which originate ahead of the inlet duct is then produced by adjustable vanes
which may be either solid or porous. Thus a wide range of distortion patterns
is possible.
A typical example of such a n application* for a large supersonic ramjet
installation is shown in Fig. 8. T h e actual missile configuration consisted of
a side inlet with a conical innerbody. In addition to the predominant
external compression of the flow, a small degree of internal compression
occurred between the cowl lip and the internal throat of the inlet. For the
ducted-nozzle test configuration, the duct contours were correctly duplicated
downstream from the inlet duct throat to the engine inlet. Upstream of the
inlet duct throat the contour was modified to provide a sonic ducted nozzle
* These tests were conducted at the Engine Test Facility at the AEDC with St. Ceorge A.
Brown, L. F. Webster, and their associates serving as project engineers in co-operation with
North American Aviation Corporation and Wright Aeronautical Division, Corliss-Wright
Corporation.
i;
B. H . G O E T H E R T
throat preceded by a n inlet bellmouth a n d plenum chamber for direct

connection to the facility air supply. T h e sonic ducted nozzle throat was
sized in such a manner that a t the desired flight conditions the correct
mean M a c h n u m b e r was simulated at t h e inlet duct throat. A system of
controllable vanes was located in the inlet bellmouth upstream of the sonic
ducted nozzle throat. These vanes could b e individually rotated by remote
control to provide local airflow distortions as required. A photograph
showing the detailed arrangements of the vanes, looking downstream from
the plenum chamber into the ducted nozzle, is shown in Fig. 9. A total of
13 vanes rotated around their vertical axes a n d 16 vanes rotated around
their horizontal axes.
DUCTED NOZZLE INSTALLATION
MISSILE INSTALLATION
Fig. H.
Schematic of ducted-nozzle system
Since wind tunnel tests must be conducted to establish the profiles to b e

simulated prior to testing with the ducted-nozzle technique, a complete
inlet model was investigated in a supersonic wind tunnel, a n d the total
pressure a n d M a c h n u m b e r profiles at the engine inlet were determined.
Because the combustion intensity a n d fixed engine exit nozzle normally
control the pressure level in the combustion chamber a n d inlet duct, the
wind tunnel tests were conducted with a cold flow model having a n adjust-
able exit nozzle in order to produce the proper combustion chamber
pressures. T h e ducted-nozzle vanes were then adjusted to closely approxi-
m a t e the distortion profiles obtained from the wind tunnel tests. A com-
parison of a typical wind tunnel profile a n d the corresponding full-scale
ducted-nozzle profile for a free-stream M a c h n u m b e r of 2 - 75 a n d a n angle
of attack of approximately 6° is shown in Fig. 10. T h e wind tunnel profile
is greatly distorted, the local Mach n u m b e r varying from 0-74 in the center
area of the duct to values of 0-40 a n d less in the area close to the duct wall.
T h e simulated profiles for both cold flow a n d hot flow full-scale ducted-nozzle
testing compare quite favorably with the profile obtained from wind tunnel
model tests. Certainly the main features of the distortion pattern are
correctly simulated.
44
Fig. 9.
Adjustable guide vanes.
WIND TUNNEL
T Y P I C A L P R O F I L E AT
TEST RESULTS DUCTED NOZZLE ENGINE OPERATION
(SPEC PROFILE) C A L I B R A T I O N DATA (/o*0495
VANE CONO 2* VANE COND * rt V I N E CONO 4 5

M0.2.75 " o - 2 " M „ . 2.75
a ' e.z Q ' 6. 2
V.
Fig. 10.
Typical flow distortion profiles at the engine face
45
B. H . GOETHERT
It should be noted that, in addition to the effects of free-stream Mach

n u m b e r and angle of attack, the distortion pattern depends primarily on the
pressure level in the combustion c h a m b e r — t h a t is, on the intensity of
combustion or on the fuel-air ratio. Hence, a strong inter-relationship
exists between the combustion intensity and the pressure recovery and air-
flow distortion in the inlet duct.
WIND TUNNEL PROFILES DUCTED NOZZLE PROFILES
FLIGHT CONDITION:
MACH 2 . 6 0 , i) d -.64S
DUCTED NOZZLE THROAT

ENLARGED 7 5 *
VANES REMOVED
FLIGHT CONDITION:
MACH Z.60, T/a'.GOZ
DUCTED NOZZLE
VANES AT 9 0 *
FLIGHT CONDITION:
MACH Z . 9 0 , i l d v 5 3 0
DUCTED NOZZLE(THROAT
REDUCED 7 I % 1
VANES AT 9 0 °
FLIGHT CONDITION:
MACH 2 9 0 , Y 4 ' 4
DUCTED NOZZLE
VANES AT 9 0 '
Fig. I I .
Mach number contour lines (M* = constant) from wind tunnel and ducted-nozzle tests
Ducted nozzle without vanes.—During the ducted-nozzle testing, it was
observed that the location of the terminating shock downstream of the inlet
duct throat section was of primary importance. This shock moves upstream
or downstream depending upon the pressure level in the combustion chamber.
As a result, a sensitive check on the simulation method can be obtained by
observation of shock location as a function of the pressure level in the
engine.
It was also found that for proper profile simulation very little artificial
distortion was required from the adjustable vanes as long as the M a c h
4fi
number at the inlet duct throat section was properly simulated. A systematic
study was therefore conducted without adjustable vanes to determine to
what extent the correct distortion profiles could be simulated by merely
adjusting the Mach number in the narrowest section lo the proper value
determined previously in wind tunnel tests.
In this study the effective area of the ducted-nozzle throat was varied by
insertion of wedges or by blowing in or sucking oil'air. Some typical results
are shown in Fig. 11. T h e comparison between the wind tunnel and the
ducted-nozzle profiles indicates that a high degree of simulation can be
obtained merely by proper adjustment of the Mach number in the inlet duct
throat section.
Transient testing.—The question arises: T o what extent can transient
conditions be simulated by the ducted-nozzle m e t h o d ? If it is assumed that
the flight Mach number and the angle of attack of the vehicle are not
changed during the transient and if only quick throttle movements, from
idle to maximum, for instance, are considered, the flow through the inlet
duct throat section and further downstream will not be influenced. Only
the location of the terminating shock will be changed. Consequently,
as long as the inlet continues to operate super-critically, transient conditions
can be satisfactorily simulated, since the resulting changes of distortion
pattern are produced by the terminating shock and its interaction with the
boundary layer. If transients of flight Mach number or angle of attack are
considered, the disturbances will travel downstream through the inlet duct
toward the engine. In this case, also, satisfactory simulation of the transient
conditions may be possible, since quick area changes at the narrowest
section of the duct, as well as changes of the controllable vane positions,
produce disturbances which travel downstream through the duct as in actual
flight. However, the different response rate of the boundary layer buildup
in the inlet ducting may introduce deviations which would require specific
investigation before the suitability of the ducted-nozzle technique for each
individual test could be assured.
Free-jet Test Methods

For both the distortion screen and the ducted-nozzle methods, the test
facility must supply air only to the extent directly required by the engine;
a very small additional amount of external airflow is needed for breathing
or cooling purposes. Thus small test facilities may conduct inlet simulation
tests on combinations of "inlet-plus-engine" assemblies by accepting the
compromise of artificial means of simulating the inlet flow distortion patterns.
These two techniques also require wind tunnel tests preceding the actual
hot engine tests for determination of the correct distortion pattern. If such
artificial means of producing the desired inlet distortion pattern are not
satisfactory and if test facilities of sufficient capacity are available, the so-
called "free-jet" test method which represents a significant improvement
in the natural production of inlet flow distortion may be employed. As a
matter of fact, if a large enough jet with the desired Mach n u m b e r can be
provided so that sufficient air may be directed externally around the inlet,
the inlet shock waves for external compression and the entire internal flow
pattern in the engine from the inlet lip to the exit can be as adequately
47
B. H . GOETHERT
simulated as if the engine were installed in a large propulsion wind tunnel

or were in actual flight. Naturally, the external flow conditions, even a
short distance behind the inlet lip, cannot be simulated in this m a n n e r since
the jet boundary will produce external interferences. Only when the correct
flow over the entire external test article must be determined simultaneously
with the internal flow are tests of the complete assembly in a large propulsion
wind tunnel required.
-NOZZLE
I TEST CELL
TEST INSTRUMENTATION.
Fig. 12.
Variable Mach number and variable angle-of-attack nozzle
Various free-jet assemblies have been in operation at the A E D C for
several years. These are either of the fixed-supersonic-nozzle type or of the
flexible-supcrsonic-nozzle type which permits quick changes of the test Mach
n u m b e r during engine operation. A typical installation for a simple super-
sonic flexible nozzle which has been in operation at the A E D C since 1954
for tests of full-scale ramjet propulsion systems is shown in Fig. 12. This
nozzle, with an exit cross section of 26 in by 26 in, is provided with a single
Fig. 13.
Free-jet installation showing angle of attack with a fixed Mach number nozzle
jack actuator for changing the M a c h n u m b e r quickly over wide ranges. T h e
m a x i m u m rate of change of M a c h n u m b e r is A M = 3^0-25 in one second.
T h e nozzle is also installed in such a way that it can be rotated quickly
between limits of ^ 1 5 ° in order to provide simulation of the flow inclination
of the vehicle. This rotation has a m a x i m u m rate of 15° per sec.
Another installation, set u p for one specific large supersonic inlet configura-
tion, is shown in Fig. 13. In this case, the supersonic nozzle is designed for
III
a specific Mach number slightly in excess of M = 3 ; again the angle of

attack may be adjusted quickly during testing.
O n e major requirement for free-jet testing is that disturbances originating
at the supersonic jet boundary must not propagate into the inlet flow to the
engine. Since at large angles of attack such a condition is sometimes difficult
to meet, critical inlet flow patterns are observed by means of the shadow-
graph method to determine whether or not an externally-originated shock
or disturbance is interfering with the inlet flow. Pictures of several typical
inlet shock wave configurations are shown in Fig. 14. It is evident that at
low chamber pressure, the shock waves do not interfere with the inlet flow;
at higher chamber pressures, interference-free inlet flow is doubtful or non-
existent. In such cases, a different arrangement of the inlet might be helpful
or, in extreme cases, a model must be utilized instead of the full-scale article.
T h e limitations thus imposed on the free-jet testing method have resulted
in a limitation of the maximum permissible test article size. In general,
50 per cent or more of the supersonic nozzle flow must be bypassed around
the engine. This value is somewhat larger at high angles of attack.
Pressure recovery devices for free-jet testing.—In free-jet testing, the air a p -
proaches the engine inlet at the low static pressure corresponding to the
ambient pressure of the altitude to be simulated. Hence, if no special
devices are employed, the exhaust gases of the engine and of the bypass air
flowing around the engine will be discharged at the same low pressure and
will impose severe requirements on the capacity of the facility exhaust
system. Since most exhaust plants are restricted in their capacity by the
volume flow, the mass flow capacity of a given exhaust system can be in-
creased if the exhaust gases are supplied to the exhausters at a higher
pressure level.
It is relatively easy to discharge the exhaust gases passing through the
engine at a higher pressure than the ambient altitude pressure. This can
be done, for example, by applying the choked-nozzle technique in which
the engine exhaust gases are not expanded to the ambient pressure but only
to a pressure slightly below that required to choke the exhaust nozzle. This
method permits the study of all internal conditions up to and including the
throat of the exhaust nozzle, but it does not provide simulation at the correct
conditions in the downstream portion of the exhaust nozzle. Even in this case,
however, methods are available (see Section I V below) by which the exhaust
gases can be expanded to low pressure levels corresponding to ambient
pressure and then recompressed by the use of supersonic diffusers or ejectors
before they enter the exhaust duct.
A method of using the kinetic energy of the bypass air to increase its
pressure level before it is discharged into the exhausters, based on the
principle of the supersonic second-throat diffuser, has been successfully
employed. This method (Fig. 15) utilizes a shroud which is fitted tightly
to the free-jet supersonic nozzle to avoid mixing of the bypass air with the
air in the surrounding chamber until after the bypass air has been raised to
a higher pressure level. Extensive studies have been conducted on models
and on full-scale installations to determine the optimum configuration for
such a supersonic diffuser. By careful contouring of the annular diffuser
duct, including splitting the duct into several separate ducts, a respectable
49
B. H. GOETHERT
CHAMBER PRESSURE LEVEL:

CORRECT
Fig. 14.
Shadowgraph pictures for free-jet installation at different chamber pressures
pressure recovery can be produced. Some typical results* are shown in

Fig. 16. These experimental d a t a indicate that for small angles of attack at
a M a c h n u m b e r of 3-1, for instance, a pressure recovery ratio of/» D //) c = 4-84
(corresponding to /) T //) D = 8-8) has been obtained by the use of a suitable
supersonic diffuser. This value compares with a ratio of 1 : 11 for normal
shock recovery at M a c h 3-1. T h e difference in these ratios indicates that
appreciable gains can still be expected from further refinements of the
(lill'user scheme.
T h e pressure recovery potential for the bypass flow in free-jet test assemb-
lies is, however, much poorer t h a n the pressure recovery potential of
conventional supersonic wind tunnels. This situation results from the fact
that the boundary layer along the internal walls of the supersonic nozzle
is a considerably larger portion of the mass flow through the supersonic
nozzle, because the high energy core of the supersonic nozzle flow is guided
through the engine. As a result, it is obviously exceedingly difficult to
recover the energy of the supersonic bypass flow. Experiments have there-
fore been carried out to reduce the b o u n d a r y layer thickness by means of
• These tests were conducted at the Kngine Test Facility nflhc AEDC with K. K. I.atvala,
P. M. Hood, and D. E. Anderson serving as project engineers.
5(1
CHAMBER PRESSURE LEVEL;
MARGINAL
CHAMBER PRESSURE LEVEL:

TOO HIGH
(c)
51
B. H . GOETHERT
suction. Such suction could be produced in actual configurations by

separate auxiliary ejectors, and extensive experiments using ejectors have
also been carried out. Some typical results are shown in Fig. 17. Suction is
obviously advantageous in improving the pressure recovery and, hence, the
-OUTER SHROUD
INNER SHROUD
Fig. 15.
Schematic of free-jet test configuration with second throat diffuser shrouds
operating range of a free-jet nozzle. It might even be advantageous in some

cases to separate the engine exhaust flow and the bypass flow before t l n \
discharge into the common exhaust duct. T h u s an auxiliary ejector could
be provided to raise the pressure level of the bypass air before it is discharged
Fig. 16.
Free-jet nozzle operating pressure ratio with approximately 3 per cent leakage
into the exhaust duct. Also a supersonic diffuscr behind the engine could
be employed to utilize the energy of the engine exhaust flow in order to
produce a pressure rise.
Statements on the Relative Merits of Inlet Flow Simulation Methods

I n the preceding sections, the advantages and disadvantages of the
methods utilized in the Engine Test Facility of the A E D C to simulate the
proper inlet flow conditions for testing of inlet-and-engine combinations
have been briefly discussed. Starting with the simple screen-distortion
method, proceeding to the ducted-nozzle method, and finally to the free-jet
test method, a progressively improved simulation of the flow conditions
52
SOME SELECTED PROBLEMS IN E N G I N E A L T I T U D E TESTING
can be obtained which finally climaxes in the most elaborate method the
placing of the entire system in a propulsion wind tunnel.
Each of these methods has its merits. M a n y basic questions concerning
engine sensitivity to inlet flow disturbances can be answered by the most
primitive—the screen distortion method. T h e advantages of the more
refined methods of extending the flow simulation range must be weighed
a
X
I \ LEAKAGE APPROX. 3 2 % -
! NO SECOND THROAT—j
I I 3.0
MACH NUMBER
Fig. 17.
Free-jet nozzle operating pressure ratio with and without suction
carefully against the greatly increased expenditure necessary and, most

significantly, against the longer test preparation time. In some cases the
most elaborate type of test facility is required to adequately simulate a
specific propulsion problem; in other cases the higher cost and longer time
required to prepare a full-scale propulsion test in a wind tunnel, or even in
a free-jet assembly, will point to the application of the simpler ducted nozzle
or distortion screen techniques. Each of these schemes has its own range of
useful application; the final selection must be made on the basis of the
merits and requirements of each individual case.
II. ENGINE TEST FACILITY PLANT PERFORMANCE

FOR T R A N S I E N T T E S T I N G
During the development testing of propulsion systems, not only must
the steady-state performance of the systems be investigated; the transient
conditions occurring during the maneuvers of a n airplane or missile or
during engine control point changes must also have extensive study. T h e
growing importance of such transient engine testing can best be demonstrated
53
B. H. GOETHERT
by considering the fact that during the past few years, more plant hours at
the A E D C have been devoted to the investigation of transient engine
performance than to engine steady-state performance.
T w o different types of engine transient testing occur. First, during
trajectory testing the engine is set at one operating point (for instance at
military or m a x i m u m power), and the environmental conditions such as
altitude and temperature are changed according to the desired trajectory
of the system. This type of testing simulates conditions similar to the rapid
climb or descent accompanied by large accelerations of the airplane or
POWER I
LEVER
POSITION
degrees 50
0
8000
ENGINE
SPEED 7000 \
RPM
6000 - ^ -
5000 \- - ~"T—-.
85
ENGINE
AIR rs
FLOW
lb/tec
EXHAUST
NOZZLE
AMBIENT 10 0
PRESS. 90 > f\ \
psia
'
80 r "-^
ENGINE
INLET
TOTAL
PRESS
psia
e 12 is
TIME-SECONDS
Fig. 18.
Variation of set conditions using manual facility control: deceleration of
small turbojet from rated power to idle power
missile. I n the second type of transient testing, environmental conditions

remain essentially unchanged and the loading of the engine is changed
rapidly. Examples of the second type of transient testing are quick throttle
bursts or throttle chops or rapid switchovers from non-afterburning to
afterburning conditions.
Special efforts are required to design a n engine test plant control system
in such a m a n n e r that the desired rapid transient response can b e obtained.
Fig. 18 shows a typical test trace for a relatively small engine. T h e purpose
of this test was to study the performance of the engine at constant altitude
a n d speed when the power lever of the engine was suddenly pulled back
from full non-afterburning thrust to near idle position. T h e trace shows that
the engine airflow dropped from 85 to approximately 60 lb/sec within 5 sec
a n d from 85 to approximately 45 lb/sec within 19 sec. During this maneuver,
the manually controlled plant could not maintain exhaust pressure constant
51
SOME S E L E C T E D PROBLEMS IN ENGINE ALTITUDE TESTING
within acceptable limits; the pressure changed during the control period
from 8-3 to 10-8 p.s.i. and dropped to 7-5 p.s.i. before it finally recovered. T h e
inlet total pressure changed even more drastically. Oscillation of the inlet
pressure between 11-5 and 13-6 p.s.i. occurred before the plant finally settled
down to steady-state condition. T h e performance characteristics of the
engine were therefore considerably distorted and even indicated a trend
opposite to the actual conditions of flight. When the throttle is suddenly
chopped in actual flight, the inlet total pressure is generally expected to
decrease because of the gradual decrease in the speed of the airplane. T h e
trace, however, shows that during the transient ground testing, the inlet
total pressure increased within the limits just mentioned.
It is obvious that for such rapid transients the h u m a n operator can not
maintain the steady-state conditions of the plant with the accuracy required.
Extensive theoretical and experimental studies* have therefore been carried
out in the Engine Test Facility of the A E D C to improve the transient response
of the plant.
Manual Control System Analysis

A schematic of the Engine Test Facility plant showing only one of the
several test cells is shown in Fig. 19. Only the pressure control system
-THROTTLE SYSTEM FROM ATMOSPHERE-
BLEED-IN VALVE-,
VERMER I I
-COMPRESSOR
EXHAUSTERS/A
~FR0M ATMOSPHERE TO ATMOSPHERE'
Fig. 19.
Schematic control system for one cell of the AEDC Engine Test Facility
portion of the plant is indicated; other controls such as temperature have

been omitted for the sake of clarity. T h e compressor plant provides air
through the main inlet duct to the test cell. T h e pressure and, consequently
the main flow on the air side, can be controlled by means of throttle valves
and a bypass line having a control valve. T h e exhaust pressure is maintained
by a system of exhausters which can also be controlled by throttle valves
and by a bleed-off valve to atmosphere. T h e basic problem is the control
of the pressures in the air supply and exhaust ducts so that the necessary
test conditions can be met.
In order to determine the basic response characteristics of the plant as a
• These suidirs wire ((inducted by the staff of the Kngine Tesl Facility of the AEDC with
G. V. Schwcnt, D. \V. Russell, W. K. McGregor, and L. F. Burns serving as project engineers.
55
H. GOETHERT
Fig. 20.
Exhaust plant response to sinusoidal disturbances
.2 .5 1.0 2.0
DISTURBANCE FREQUENCY
Fig. 21.
Phase angle of exhaust plant response to sinusoidal disturbances
preliminary to the investigation, sinusoidal disturbances of small amplitude

and varying frequencies were produced, and the resulting changes in the
inlet a n d outlet pressure in the test cell were observed. These disturbances
were produced by oscillatory movements of the valve settings in the inlet
and exit throttle systems and in the bypass line. T h e adjustments produced
changes in pressure in the inlet ducting to the test cell and in the test cell
itself, equivalent to the changes in air supply and exhaust imposed by a
test engine.
56
SOME SELECTED PROBLEMS IN ENGINE ALTITUDE TESTING
Some typical results of the experiments for a mean air supply pressure
of 20 p.s.i. and a mean exhaust pressure of 7-5 p.s.i. are presented in Figs. 20
and 2 1 . Fig. 20 shows the response of the exhaust plant to sinusoidal
disturbances. Note that only up to disturbance frequencies of 0-022 rad/sec
is the amplitude of the pressure disturbance within 90 per cent of the steady-
state disturbance expected from the movement of the control valve. When
the oscillatory frequency of the disturbance is increased, the amplitude
0.
z
<o.z o THROTTLE VALVES
• BY-PASS VALVE
I I
PRESSURE LEVEL 2 0 PSI
.1 .2 .5 1.0 240
Fig. 22.
Air supply plant response to sinusoidal disturbances
o.i 0.5 i.o

Fig. 23.
Phase angle lo air supply plant response lo sinusoidal disturbances
response is reduced rapidly. For instance, at a disturbance frequency of
3-0 rad/sec, the amplitude of the exhaust pressure is reduced to 3-3 per cent
of the steady-state disturbance. As the frequency of the disturbance is
increased, the excited pressure oscillation lags behind the input disturbance
by large phase angles. These were also measured (Fig. 21).
Similar data were obtained by producing disturbances in the air supply
pressure (Figs. 22 and 23). T w o types of disturbances—disturbances from
57
B. H. GOETHERT
throttling and disturbances from bypassing—gave essentially the same

results. Pressure amplitudes of 90 per cent of steady-state values were
obtained with disturbance frequencies of 0-045 and 0-080 rad/scc, respectively;
at disturbance frequencies of 1-0 rad/sec, a response of approximately 23 per
cent was recorded.
T h e following simplified equations for the weight flow balance of the
system can be utilized to compare the experimental data for the uncontrolled
plant with the theoretical calculations:
Weight flow into system :
All',,, ( \ . A A - C.,. A/'slKnaM
AA = control valve area change
Weight flow out of system:
A Wout = C 3 • AA>ut
Weight capacity change of system:
,,,. „ dAp o u t
AIFeap,=X ^
T h e n as long as the dynamic effect of the air in the ducts may be neglected,
the weight flow balance is:
Cm. A/)sig„ui = C 3 . A/)0ut + C 4 . — « - —

or in Laplace transform:
Apout(s) 1
A/)signal(.0 * i - ( l + Ti-s)
T h e values of A'j and TX can be calculated from the dimensions of the E T F
plant; they represent fixed plant constants. A comparison of the theoretical
data with the experimental curves (Figs. 20-23) shows that satisfactory
agreement has been obtained up to disturbance frequencies of 4 rad/sec
for the exhaust side and u p to 0-3 rad/sec for the air supply side of the plant.
At higher frequencies large deviations occur which are caused by the
dynamic effects of the air flowing in the ducts. Such effects require a more
refined theoretical treatment.
Automatic Control System Analysis

After the response of the plant to sinusoidal disturbances had been
determined, the output of control valves, of either the bypass or the throttle
control type, was investigated in the same manner. By adding a control
system which was similar to an electronic analog computer to the control
valve actuator system, it was possible to introduce into the control actuator
system a n artificial gain adjustment for an "error-proportional" control
term a n d a time constant adjustment for a n "error-integral" control term.
T h e control equation of the controllers can be represented in Laplace
transform in the following m a n n e r :
Signal out
2 v e i
Signal in '
58
By adding the uncontrolled plant to the automatic controller (as shown

schematically in Figs. 24 and 25), the final response of the plant can then
be presented in the following form
Apoui(s) = 1
A/'i„!.>) "" X O • ij.j)
i
" j : i . ( l + l/Ttr)
By adjusting the gain constant A'2 and the time constant T 2 , it is possible to
determine analytically or experimentally those values which represent the
most suitable compromise. By having the E T F automatic controllers built
so that the above two constants could be readily changed, it was possible
to tunc the plant for optimum response. T h e gain and the error-integral
time constants vary with the pressure level of the air in the plant since the
pressure level is equivalent to the spring constant of the system.
Plant With Valves
Input Signal ^ 1 A
Poui( s > - ^
Test Cell
A
P.ignol «
A
P o u l (*) I
A s
Ps„no|( )' K,«0+T,.tj
Fig. 24.
Manual plant control schematic
Controller Plant With Valves

A 1 A
Pin<») Pout<s>_
Test Cell
K,.(l+r,.s)
-APs„ro|W - A Pou.( S »
1f
r a " "Pjigno
A
Pou,< S )
A p o u , (s) I
A
Ps,gn 0 |( ) s K,«(I + T,.S)
I +•
-.**-)
Fig. 25.
Automatic plant control schematic
T h e equation just presented represents the influence of only one control
clement. In the actual case of the complete test facility control system,
several such elementary systems were integrated.
T h e response of the plant with the automatic control system is compared
with the response of the uncontrolled plant in Figs. 2 0 - 2 3 . A significant
improvement is obvious. While the uncontrolled exhaust plant follows an
input of sinusoidal disturbance within 90 per cent u p to frequencies of 0-022
rad/sec, the controlled system extends this limit to approximately 3-2 rad/scc,
59
B. H . GOETHERT
a n improvement by a factor of more t h a n 150. Similar improvements are

also noted for the air side control.
In addition to the investigation of the sinusoidal disturbance response
of the uncontrolled and automatically controlled plants, investigations were
carried out using step input disturbances of various magnitudes. T h e
approach was similar to that for the sinusoidal disturbances and the most
suitable combination of gain and error-integral time constants was selected.
T h e E T F staff has worked out a procedure by which the set values for the
automatic control can be easily determined and applied to a transient test.
Performance of the Automatically Controlled Plant

T h e automatic control system in the Engine Test Facility at the A E D C
has now been in operation for several years. It is used for steady-state
100
POWER
LEVER
POSITION JQ \
degrees
0
\
8000
SPEED 7000
RPM
6000
5000
ENGINE '70
AIR
FLOW
l 3
\
Ib/MC °
\
90
\
50
no
EXHAUST
NOZZLE tOO
AMBIENT 9 Q
PRESS
psia SO
/\
70
l7
ENGINE °
INLET 160
TOTAL
pila |4 a
S 12 16
TIME-SECONDS
Fig. 26.
Variation of set conditions using automatic facility control :
deceleration of medium-size turbojet from rated power to idle power
testing as well as for transient testing in order to assure that the test conditions
will be maintained constant at the desired level. As previously mentioned,
the system is not restricted to the control of the pressure level; similar devices
have been developed for the control of the temperature of the inlet air as
well.
A typical example of a transient investigation of a turbojet engine during
which the thrust level was quickly reduced from full speed non-afterburning
condition to near idle speed is shown in Fig. 26. This case can be compared
with curves showing the performance of the uncontrolled plant for a similar
60
test (see Fig. 18). It should be noted, however, that in the case of the
uncontrolled plant, the engine mass flow was only reduced from 85 to
45 lb/sec; in the case of the controlled plant, much more severe conditions
were selected, and an engine mass flow reduction from 160 to 80 lb/sec
occurred within 4 sec and finally to 55 lb/sec within 24 sec. T h e records for
both the exhaust pressure and the engine inlet total pressure indicate that
considerable improvement in this area also was accomplished by the addition
of the automatic control system. T h e remaining small disturbances in the
plant occur very early, during the first seconds of the transient when the
extremely rapid reduction of engine mass flow takes place. During the few
seconds of the transient, the overriding parameters for engine performance
are the reduction of mass flow, rotor speed, etc; the slight change in the
supply and exhaust pressures is of minor importance.
By manipulation of the input signal to the automatic control systems,
it is possible to simulate transient trajectory operation of engines. This
feature is particularly important in the testing of high powered airplanes
and missiles since very rapid acceleration or deceleration and extreme
altitude changes must often be simulated. Moreover, the E T F automatic
control systems are supplemented by devices which provide for a rapid
and controlled variation in inlet Mach number and flow inclination. Com-
plete trajectory simulation can therefore be provided.
III. C O R R E C T I O N OF VERY FAST T R A N S I E N T

MEASUREMENTS
It is frequently exceedingly difficult to devise mechanical measuring
equipment which possesses the necessary fast transient response to measure
such rapidly changing forces as, for example, the thrust buildup during
ignition of a rocket which occurs in a few milliseconds. Because of the high
thrust level, the support frame of the thrust stand must be relatively heavy,
but the spring constant of the entire system, including the thrust pickups,
cannot be increased satisfactorily. In addition, the mass of the system includes
the mass of the test rocket. Consequently, the natural frequency of the
thrust measuring equipment is not high enough to produce a correct record
of the thrust buildup during ignition. Although all possible efforts must be
made to devise systems having the necessary high response, it will be necessary
in many cases to record very fast transients with equipment having too low
a natural frequency.
An example of such a case is presented in Fig. 27(a) which shows the record
obtained during the thrust buildup of a solid propellant rocket firing. T h e
natural frequency was considerably lower than was desired; consequently
large oscillations of the thrust curve are indicated. T h e problem is, therefore,
the correction of this distorted thrust record and the separation of the true
thrust buildup of the rocket from the superimposed disturbances of the
thrust measuring system.
Derivation of Basic Equations

When the characteristics of the thrust measuring equipment are known—
that is, the moment of inertia, the spring constant, and the damping
61
B. H. GOETHERT
coefficient—the transient process for any thrust input can be calculated in a

m a n n e r similar to that discussed in the preceding paragraphs on the E T F
plant performance. T h e process is the reverse of most mathematical investi-
gations in which the disturbing force is known and the output is to be
calculated; in this case the oscillatory motion of the thrust stand is known,
and the disturbing force is to be calculated.
In many cases the dynamic equation of the thrust measuring system can
be simplified to include only one term for the inertia force, one term for
the spring force, and one term for the d a m p i n g force. Furthermore, when
it is assumed that the d a m p i n g force is proportional to the velocity of the
thrust stand deformation, the equation may be written as:
T{t)=I.X"{t)+C.X'{t)+K.X{t) (1)
T h e Laplace transform of the equation is t h e n :
T{s) = {I.s* 4- C.s + K) .X's) - I . X ' { t = 0)
where X'{t = 0) is the velocity of the system at the time / = 0. Since the
motion is assumed to start from rest, this term is zero.
With the function X(t) known from measurement, it is possible to represent
the experimental function X(t) by a Fourier-integral having known co-
efficients. However, only the initial part of this function is of interest during
the rapid thrust buildup, and it is generally feasible to replace the initial
2TT
part of the A'-function bv a Fourier-series for which the interval, At =
r
(O
can be selected in a suitable manner. T h e function X(t) may then be
represented b y :
X{t) = £«,,.sin (fico/) (3)
nio
X{s) = S a „ . ,2 -.—r.
" s 4- {nco)2
W h e n this series development is introduced into the dynamic equation

of the system, the nth term of the equation assumes the following form:
1 / . [ J » + (mo)*] + C . S + [K - I{mo) 2 \
c,fT"{s)= s* + JncoT2
nto.C.s n.(o[K — I{no>)*\

= n.a>I-\r „ —- . , 4 TTl—,, (4)
s 2 + (nu))2 s2-\- {nio) i
W h e n it is transformed back into real time, the nth-term then becomes:
. T J t ) = n . t o . C . c o s {mot) 4- \K — I{nco)2] .sin {mot) + ncoJ.U'{t) (5)

a
n
,•
with U'{1) defined as the unit impulse function and U'{t) A t = 1

Jo
62
SOME S E L E C T E D P R O B L E M S IN E N G I N E ALTITUDE TESTING
After some manipulations and the introduction of the dimensionlcss

constants of the system, the final equation reads:
T{t) = K . Z a H .2.Jj^.sin {nail + * . )
4- K . 2 a „ . ( « « J / W N ) . - — - (6)
io„
2.f(nw/w N )
where tan *„ = . _ j — ^ (7)
=
^ntf'f V{(2™XX2 4- [1 - XXX 2 ] 2 }
w N = \Z{K/I) = natural frequency
c = •-' • f = damping coefficient

Z A
In the above equation, the constants aH is known from the Fourier-series

analysis of the recorded thrust function.
Significance of the Unit Impulse Term U'(l)

In the preceding derivation of equations, it was assumed that the movement
of the thrust stand consists of a series of elementary sinusoidal motions which
start at the time / = 0 and have a definite velocity X ' n . O n the other hand,
the thrust stand was assumed to be at rest prior to the time ( = 0 since no
thrust was applied before rocket ignition. Hence the term U'{t) represents
the impulse which, within a very short time interval, accelerates the system
from rest to the initial velocity X ' n . This can be readily seen by considering
the impulse equation:
T„.dt = I.dv (8)
Av=-\ " .dt = Sfl„.n.w U'{l).dt = a n .n.co
.1 U'{t) .dt = 1 by definition
It can be shown that if the velocity change Av is exactly equal to the total
initial velocity of the sinusoidal elementary movements since Sfl n .n.co is
equal to the derivative of the series development for X{t) taken at t = 0 :
d
X'{t) = 3 - S a „ . sin {neat).
T h e unit impulse function U'{t) makes the acceleration at the time t = 0

infinitely large in order to produce the sudden velocity change. Since
infinitely large accelerations are not possible in a real system, the condition
that the sum of all unit impulse function terms disappears must be fulfilled
that is,
Sfl,.n = 0 (9)
63
B . H . COETHERT
This equation should be automatically satisfied when the series development

of the recorded thrust curve is sufficiently precise. If, however, this condition
is not satisfied in the numerical process, it is advisable to correct the higher
coefficient a n in a suitable manner.
Comparison with Indicated Spring Force

Since the indicated spring force T i n d , as well as the corrected thrust T can
be presented by a series of elementary sinusoidal functions, T m f = A\nti .
sin {noil) and T n = A T .sin {ncot -f- (f>n), respectively, it is feasible to
compare the amplitudes of the individual elementary functions as well as the
phase angles:
Amplitude R a t i o :
X „ / X d „ = y/{(He.tuolcos)* + fl - (nw/ftXT, (10)

Phase Angle:
• x 2c .noijoi-s
t a n <f>n = ; r= (11)
1 — (no)/o) N ) 2
For slow motions—that is, for io —*• 0—the amplitude ratio approaches
unity, which means that indicated and recorded thrust forces are equal. O n
the other h a n d , when the frequency is large compared with the natural
frequency of the system, the amplitude of the corrected thrust function
becomes exceedingly large compared with the amplitude of the indicated
thrust force. I n the latter case, the accuracy of the indicated correction
becomes relatively poor.
Without the above correction method, a sufficiently accurate picture of
the thrust function can be obtained only for transients with a frequency
spectrum which does not extend beyond 1/10 to 2/10 of the natural frequency
of the system. With the correction method described above, it appears
possible to extend the measuring range of a given system to transients for
which the significant frequencies may be as high as two times the natural
frequency of the system. T h e useful range of a given thrust system may
therefore be extended by another order of magnitude in comparison with the
uncorrected system.
Determination of the System Constants

Experimental determination.—Since to apply the indicated correction method
it is necessary only to know the amplitude ratio and the phase angle as a
function of the frequency (see Eqs. (10) and (11)), the system may be
excited by various frequencies and the amplitude ratio of output to input
and to the phase angle may be measured. T h e n by introducing these
experimental values into the equations, it is possible to arrive at the corrected
curve as previously described.
Such a correction method using experimentally determined amplitude-
ratio and phase-shift functions is also applicable to complex systems in which
the natural frequencies of more than one element of the system must be
considered. T h e indicated method can handle such complicated systems
without difficulty whenever experiments have been carried out with varying
frequency excitation.
64
SOME S E L E C T E D P R O B L E M S IN E N G I N E A L T I T U D E TESTING
Step input excitation.—In many cases it is not practical to apply excitation

over the wide frequency range required by the above method. It is usually
possible, however, to excite the system by a simple step input function. In
the rocket experiment previously discussed, a step input of the thrust force
was produced by loading the system with a weight attached to the stand by a
thin wire. When the wire was snapped, the desired step input was produced.
In this case, the natural frequency and the damping constant of the system
can be determined by mathematical analysis if the type of the dynamic
system is known. T h e disturbing force function of the system can be intro-
duced with the step input function U{t)—that is, T{t) = T 0 .U{t)—and in
Laplace transform: T{s) = T J s where T„ is the magnitude of the step input
force. T h e system equation may then be written:
T l l . l s = {I.s 2 + C.s + K).X{s) (12)
T h e solution for this equation can be readily established as before to obtain

finally:
K.X{t) I C \ I IV K /C\«-|
f. = I - « P ( - 2 7 x r v L x (2/)J-'
+ c
2 1. XX(rX') <13'
With T l n d (w x j) = K.X{coy,t) as the indicated thrust force of the system, in
dimensionless terms:
exp (—CMJ-1)
XiiMo = I - v. .,., c ( « X - \ •!.' '•-)./ <j>\ :I4I
ta
"^X(iX
Since the experimental record for the indicated thrust T inil is known, the
natural frequency to$ and the damping coefficient c of the system can be
determined.
Power Series Development of Thrust Curve

If the main emphasis of the correction method is directed to the initial part
of a n experimental curve, it may be more feasible to select another series
than the Fourier series as a representation of the measured curve. For
example, a power series development with increasing exponents is frequently
more practical. For this case the following results can be derived:
Measured Thrust Curve:

rind(WN/)=*.vaii.(WN./)" (is)
Corrected Thrust Curve:
T(a>j,t) = K . Z a n . { { c o } i . t ) " + 2.c.n.{(Ox.t)"-i
+ n.{n-\){a>s.tr-2\ (16)
65
B. H . GOKTHERT
THRUST BUILD-UP OF A ROCKET
A
i.o / \
3 '.» \
\
0.6
.0 6 .12
TIME,SEC
o. Uncorrected
b. Corrected
Fig. 27.
Typical traces of the Ihmst build-up of a full-scale rocket
lit.
when A' is the spring constant, OJN is the natural frequency, and c is the
damping ratio of the system as defined previously.
Application of the Correction Method

Correction of thrust-time records.—The correction method described above was
applied to the experimental uncorrected thrust record shown in Fig. 27(a).
As mentioned, the natural frequency and damping coefficient were deter-
mined from step input loading with a weight and a snapped wire. T h e
corrected curve is shown in Fig. 27(b). Il is evident that a considerable
improvement has been achieved. T h e experimental transient force function
was initially developed in a Fourier series. As a consequence of this choice,
the corrected thrust curve shows a number of wiggles which were caused by
the higher-order Fourier terms. T o remove these, the initial part of the
curve was developed in a power series, and the corresponding correction
method was applied. T h e initial portion of the corrected curve in Fig. 27(b)
shows the power series development; the Fourier-series development was
used for the remaining part.
Correction nf impulse integrals $ T . d t . — I n rocket propulsion testing, instru-
ments are frequently installed to measure the total impulse \ T . d t directly.
Since the thrust input to these instruments is subjected to the same dynamics
discussed before, an error in the indicated total impulse arises which can be
calculated with the equations previously derived.
If the most severe case of a step input is assumed, the magnitude of the
total impulse error can be readily calculated. In this situation it is practical
to express the error as a time interval Al during which the full step input
thrust Tit would act:
Jo T0
d<
= m/n _ C 2 ) - I e x p ( - c . w N l * ) . c o s k x V C ~ X ' W 07)
After some mathematical manipulations:

c o s . A t = 2.c (18)
With this equation the error times can be calculated for systems with
different damping coefficients, c, and natural frequencies, (Ojr. For a system
with 0>H = 150 c.p.s. and a damping coefficient c = 0-2, the impulse error
time of the recording systems is At = 0-003 sec. Thus, if such a system is
used for testing a rocket having a 10-sec burning time, a negligibly small
error of the total impulse measurement would occur.
T h e above method of determining the total impulse error could naturally
be refined by introducing a more representative thrust input function.
However, the error will be even smaller than in the case of the step input
function used in the above calculations.
Application of the Correction Method

It should be mentioned again that the correction method discussed may be
used profitably to extend the range of measuring equipment by an order of
67
B. H. GOETHERT
magnitude beyond the range over which it would normally record transient
data without appreciable distortion. Although the development of measuring
instrumentation and supporting equipment having the required high fre-
quency response would be preferable, the correction method provides a
means of obtaining accurate data in those many cases where a high frequency
response system is not available.
IV. H I G H A L T I T U D E T E S T I N G
Definition nf Problem
Engine operation at high altitude has numerous difficulties of a very
different nature—efficiency drop or stall of rotating machinery, deterioration
40,000
60,000 •
100,000
400 600
TEST CELL AIR FLOW, LB/
Fig. 28.
Exhaust requirements and test facility capacity for engines with inlet capture area of 10 f t '
of inlet diffuser performance, difficulties in maintaining stable combustion

at the low pressures. O n e serious effect of high-altitude operation has been
discussed previously—the narrowing stall-free operating range of a turbojet
engine (see Fig. 4). Combustion instability difficulties in the form of partial
or complete blowouts of the combustor or as periodic blowouts and subsequent
relights have been observed and require extensive investigation. With flame
periscopes, which permit observation of the combustion in a ramjet or in the
afterburner of a turbojet engine, a detailed optical record of the combustion
process can be obtained.
Since the operating range of air breathing and rocket engines is extending
rapidly to altitudes having very low ambient pressure, the problem of
handling the large volume flow required in extreme altitude testing with
conventional rotating machinery has become most difficult. A characteristic
68
performance diagram showing the exhaust capacity of a large engine test

facility is presented in Fig. 28. Even for a facility with such a large capacity,
the volume capacity of the system puts an unsatisfactory limit on plant
performance at altitudes of 80,000 to 100,000 ft. However, by means of
suitable diffusers placed behind the exhaust of operating engines, it is possible
to utilize the energy of the exhaust gases and, thus, to supplement the basic
DIFFUSER TEST ARRANGEMENT _
34
32 PRIMARY NOZZLE M= 2 . 8 9
30
28 332
289 1
26
24
1 \
1 \
2 ? 1 \
1 l
20
a I8 • \
I \
V- \
\
\
\
a" | 4 \
I2 \
IP \
X
8 *, " ^
v
6 s. - ^ , ~~~~-.
-J
4 __^, > "•"*•" • .
2
O 10 12 14 16
Fig. 29.
Performance of typical exhaust gas diffuser (cold flow)
capacity of the test plant for high altitude testing. In such diffusers, the
exhaust gases are compressed from the low ambient pressure in the test cell to
considerably higher pressure levels before they enter the basic exhaust plant.
Ejectors for Steady-stale Testing

A series of simple exhaust duct diffusers has been investigated* in E T F ;
some of the results are shown in Figs. 29 and 30. It is obvious that a com-
pression of the exhaust gases by ratios u p to 1 : 30 can be readily accomplished
with these simple devices. Such high compression ratios, however, require
that no bleed flow leak into the test chamber (see Fig. 31). In the case of a
* These studies were conducted at the Engine Test Facility of the AEDG in conjunction
with my associate, D. Taylor.
69
B . H . GOETHERT
diffuser pressure ratio of only 1 : 10, the capacity of the basic engine test
facility plant is increased by a factor of 10. With such an ejector the plant
exhaust capacity would be effectively extended, as is shown in Fig. 28.
It is of interest to note that the pressure in the engine test chamber can In-
reduced to values which are not only below the static pressure at the exit of
PRIMARY N O Z Z L E M-2.89
NO BLEED FLOW
32 1
O^yOm. -1.42 w - S.SOI
D,/o M -L25<M-5J2)
L / D , • 4.15
24
Mi 5.60
h »^ ^
l M l \\ s*. • • " .
24»9
\
\
\ I
12
1 -.
\\ \
V \
a v.
It
* " * - • * -
*"*•
4
B —
Fig. 30.
Effects of geometry variations on diffuser performance
the engine nozzle, b u t are also considerably below the values which are
obtained from one-dimensional flow theory assuming that the flow fills the
difluser without losses. It is evident, therefore, that the pressure in the engine
test c h a m b e r is controlled by the pressure in the boundary streamlines of the
exhaust jet. In the absence of viscosity effects, the flow in the outer boundary
region of the jet can expand to any arbitrary pressure within a large range
without violating any law of compressible flow. Consequently, the viscosity-
effect in the wall b o u n d a r y layer governs the pressure level to which the
exhaust jet expands and evacuates the test chamber. Hence, the problem of
70
SOME S E L E C T E D PROBLEMS IN ENGINE ALTITUDE TESTING
the lowest possible chamber pressures that can be produced by an exhaust

ejector is similar to the problem of the pressure level at the base of a blunt
body in parallel flow. T h e very great sensitivity of the pumping pressure
ratio to bleed-in flow is thus also explained, since the bleed-in flow thickens
the boundary layer and limits the maximum pressure rise which the difluser-
ejector is able to overcome. O n the other hand, thinning of the boundary
layer by means of auxiliary suction should enable the diffuser-ejector to
obtain even higher pressure ratios than those indicated.
Fig. 31.
Ejector performance with different amounts of bleed flow
T h e detrimental influence of bleed-in airflow has been documented in

many tests. In one case, with the test chamber completely sealed, a full-scale
diffuser-cjector established a pumping ratio of 21-5. Even with a small
bleed-in flow of only 4-5 per cent, the maximum p u m p i n g ratio dropped to
15. Obviously extraordinary care must be taken lo insure as small a leakage
flow as possible in order to produce good performance of the diffuser-ejector.
T h e tests described above utilized simple cylindrical diffusers with conical
subsonic diffusers. For more elaborate systems, particularly when the exhaust
Mach number of the engine is sufficiently high as in the case of rocket engines,
considerably higher pressure ratios are obtainable when the cylindrical
diffuscr is replaced by a diffuser with a contracted section in the form of a
second throat. Tests with diffusers of this type have been conducted and
71
B . H . G O F. T H E R T
have yielded performance curves such as are shown in Fig. 32. Pressure
ratios of 1 : 2")0 have been obtained.*
600 800
PRIMARY
Fig. 32.
Diffuser for high Mach number exhaust nozzle
V. E X H A U S T GAS E J E C T O R S FOR T R A N S I E N T T E S T I N G
T h e augmentation of the altitude at which the engine is operating can be
accomplished using ejectors as long as the engine is operating at a sufficiently
high power level. However, when the engine is not operating, the simulated
test altitudes in the test c h a m b e r cannot be maintained. This circumstance
is particularly critical in the case of starting and burn-out testing—that is,
when thrust buildup or decay is to be investigated (see Fig. 27). I n these
cases it is necessary to install an auxiliary ejector which is capable of reducing
the altitude pressure in the test chamber to a level which the operating engine
can maintain. Such an auxiliary ejector usually may have a smaller opera-
ting pressure ratio than the engine diffuser because, normally, the plant
characteristics are such that at low mass flow the plant can p u m p down the
test chamber to a lower pressure than it can at the high mass flows produced
(lining operation of an engine.
i lorwenlional Diffuse)-Ejector Arrangement

T h e auxiliary ejector is needed only during p u m p d o w n prior to the rocket
engine firing (see Fig. 33). As soon as the engine starts and builds up higher
pressure in the exhaust duct, the valve of the auxiliary ejector must be closed
• These tests were conducted at the Engine Test Facility nl' the A I'. DC by Stall' Engineer
L . J . David.
72
in order to prevent back flow. Moreover, the main diffuser behind the engine
must be closed during pumpdown operation before the engine is lighted, but
immediately after the engine ignites it must be opened very rapidly in order
to allow free passage of the exhaust gases through the diffuser duct. During
burn-out or closing down of engine operation, these same steps must be taken
in reverse order.
T o satisfy the various requirements described above, a quite complicated
valve system is necessary. T h e problem is further complicated by the fact
that the valves must operate at a very high speed; since the light-off of a
typical rocket engine occurs within a time period of less than 20 milliseconds,
only a fraction of this period is available for opening the exhaust duct valve.
AUXILIARY EJECTOR
Fig. 33.
Conventional diffuser-ejector arrangement
New Tandem Diffuser-Ejector Arrangement
M a n y of the difficulties just described arc avoided by the use of an auxilian
ejector and an engine exhaust diffuscr which are placed in tandem. This
scheme is demonstrated in Fig. 34 which shows a rocket engine with its
exhaust entrained by a diffuser duct so that the desired pressure recovery is
produced for altitude pressure augmentation. An auxiliary ejector with as
little blockage in the ducting as possible is placed behind the inlet of the
diffuser duct. This auxiliary ejector, operated by air or steam, produces Un-
necessary pumpdown action before the ignition of the rocket engine. .As soiiii
as the rocket engine ignites, the exhaust gases enter the diffuscr and establish
a shock system in the second throat supersonic diffuser which is formed by the
internal auxiliary ejector. When the energy of the rocket exhaust readies a
sufficiently high level, the pressure recovery of the rocket exhaust gases builds
up the pressure from the low test cell pressure to the high pressure in the
exhaust duct. Because of its larger mass, the rocket exhaust takes over
automatically from the auxiliary ejector without the action of special valves.
In fact, the auxiliary ejector acts as a quickly-responding pneumatic valve
73
B. H . GOETHERT
during start-up and close-down operation. After the rocket exhaust flow is
established, the air supply to the auxiliary ejector may either be closed down
completely or may continue to operate.
During shutdown of the rocket engine the sequence indicated above is
repeated in reverse order. As soon as the rocket power is no longer available
to provide the required pressure rise, the auxiliary exhaust ejector auto-
matically takes over and becomes the necessary barrier to prevent flow of the
high pressure air from the exhaust ducting back into the test chamber.
Various Configurations of the Tandem Diffuser-Ejector Arrangement

In the preceding paragraphs an auxiliary ejector arrangement was
discussed in which the auxiliary ejector is located in the center of the rocket
Fig. 34.
Tandem diffuser-ejector arrangement with axial ejector gas inlet
exhaust diffuser duct (Fig. 34). T h e air supply is provided through a long
central tube in order to prevent excessive blockage in the supersonic diffuser
duct. In this arrangement the auxiliary diffuser acts as a second throat
supersonic diffuscr to improve the pressure recovery of the hot supersonic
exhaust gases. Such an arrangement can easily be added to a n existing test
configuration or adapted to other pressure requirements by placing a simple
and compact auxiliary ejector assembly within the diffuser. T h e supports on
the auxiliary ejector are not welded or fastened to the internal wall of the
diffuser duct b u t simply ride on them. I n this way any difficulties with
differential expansion are circumvented.
In some cases, however, the m a x i m u m size of the auxiliary ejector that may-
be placed internally in the diffuser duct will b e too small to accommodate
the mass flow required to produce the desired ejector action. I n such cases
it may be advantageous to utilize a larger exhaust diffuser duct in the region
of the auxiliary ejector a n d to place a smaller diffuser inlet duct upstream.
Another method of providing a larger auxiliary ejector utilizes an angular
externally-mounted ejector instead of a central ejector (see Fig. 35). In this
case, a manifold must be placed externally around the exhaust ejector d u c t ;
74
the flow enters through openings in the duct wall. Difficulties with solid
blockage by the auxiliary ejector can thus be avoided and flow with high
energy which could accelerate the wall boundary layer is introduced along
the walls of the duct. Since it is known that the losses in the boundary layer
along the diffuser duct walls are the principal cause of limitation in diffuser
performance, even larger pressure ratios might be produced by such external
angular ejectors. However, such an external difluser inlet would require
more hardware modifications than the internal ejector and might not offer
the high degree of flexibility that the internal ejector does. T h e choice
DRIVER GAS
Fig. 35.
Tandem diffuser-ejector arrangement with annular ejector gas inlet
between arrangements will depend upon the special conditions prevailing in

each individual test case.
Typical Test Results with a Tandem Diffitser-Ejector

A tandem diffuser-ejector arrangement of the type described above was
recently employed in investigating a full-scale solid propellant rocket at
altitudes of approximately 100,000 ft. T h e test cell pressure time curve made
without the auxiliary ejector in operation is presented in Fig. 36(a); during
thrust buildup and decay, simulation was not satisfactory. When an auxiliary
ejector driven by air (Fig. 36(b)) was used, the test cell was evacuated to the
correct altitude pressure before rocket firing. After rocket firing, a smooth
transition from auxiliary ejector operation to operation of the main rocket
diffuser occurred; the thrust buildup period was, therefore, adequately
covered. In this case, however, the auxiliary ejector was not strong enough
to produce the high pressure ratio required during b u r n o u t ; consequently
the ejector flow broke down and the test cell pressure rose rapidly.
Subsequently, an auxiliary ejector driven by steam was used. Again
smooth transition was established during rocket ignition and the condition
at rocket burnout was somewhat improved (Fig. 36(c)). However, because
of the malfunction of a manual control valve, the ejector was not set to
operate at optimum conditions during rocket burnout. With proper ejector
setting, the transition at rocket burnout should be as smooth as at rocket
75
B. II. GOETHERT
(A) WITHOUT AUXILIARY EJECTOf

(
—
\ K:
1
1 x^ - ^ , 1
\
1
1
A
\
\ / i
r
X
-— i 1i
—_— —- >—- - .-
-IGNITION ^-BURNOUT
in WITH ÂNDE^ 1 DIFFUSER - EJECTOR A
I
TJ "< •.
If
--T4=^
4
czw v i
-k 1
•
1
- K NI no i >• 8 JRNOUT
n
C) WITH TANDEM 3IFFUSER - EJECTOR 1
/ itT-n *T£
^ -
K "^
j
— • - — 1—' 3
IG J I T ON «- BURN( ur
1
4 0 4 8 12 16 20 24 28 52 36 40 44 41
Time (sec)
Fig. 36.
Tandem diffuser-ejector performance
76
ignition. In other words, during burnout, the auxiliary ejector should take
over from the main rocket diffuser without significant variation of the
pressure in the altitude test cell.
CONCLUDING REMARKS
T h e preceding brief discussion of some problems of engine testing gives an
insight into certain specialized problem areas which face ground facilities in
the testing of propulsion systems at high flight Mach numbers and altitudes.
T h e modern development trends toward higher flight Mach numbers and
higher altitudes require more testing of the complete propulsion system the
engine proper, the inlet, the controls, etc.—in order to reduce expensive and
time-consuming flight testing and unwarranted risk of experimental airplanes
and missiles and their crews. Similar serious problems at high altitude flight
exist in the large field of combustion—which was not covered in this brief
report.
M u c h investigation and much flexibility is necessary to set up in existing
facilities test configurations which, on the one hand, simulate the desired
flight conditions correctly enough and, on the other hand, require only the
simplest test set-up and the smallest possible modification of the ground
facility. Such well selected testing schemes, however, will be beneficial not
only in terms of cost and effort, but will also, and sometimes more significantly,
save time in the test preparation and testing. T h e selected cases presented
in this report show how it has been possible to obtain proper matching
between the specialized testing problem and the characteristics of an existing
test facility.
SOMMAIRE
On decrit les differentes methodes utilisees en soufflerie pour la determination, en
altitude simulee, des performances d'engins, compte tenu de la distorsion des profils
des vitesses d l'entree: toiles metalliques de permeabilite variable, alimentation
forcee ou essais en jet libre. Ces essais ont ete conduits sur modeles de grandeur
reelle aussi bien que sur maquettes.
Le fonctionnement en regime transitoire et les rapides variations de poussee qui
lui correspondent necessitent un controle precis el rapide des conditions d'entree et
de sortie. Seals d;s centrales aulomatiques par servo-me'eanismes peuvent etre
utilises, mais necessitent un elalonnage prealable de leurs possibilites.
D'une facon generale, les instruments de mesure sont incapables d'enregistrer
correctemenl les phenomenes transitoir's trop rapides {cas de la montee de la poussee
d'un moteur fusee) mais au moyen de calculateurs appropries on arrive a relablir
les valeurs exactes deduites des enregistrernents deformes.
Ee domaine d'utilisation des installations en altitude simulee, dans lesquelles
la depression est realisee au moyen de turbn-enmpresseurs classiques, pent etre
largemenl etendu par I'addition d'ejecteurs a faible flux secondaire. On montre que,
pnur Iss essais de simulation d'altitude Ires eievee, les ejecteurs constituent naturelle-
ment le premier etage de compression le mieux adapte. En mouvement transitoire
il faut effectuer le passage d'ejecteurs auxiliaires a I'ejecleur principal pour
representer les conditions d'allumage en altitude, mais ['operation inverse, de
reduction des gaz, est plus difficile a simuler en soufflerie, bien que dans ce cas
egalement d'interessants resultats aient ete obtenus.
11
B. H. GOETHERT
NOTATION
A A r e a ; Fourier-series coefficient
av Fourier-series coefficient
C Damping constant
c Damping coefficient
/) Diameter
e Exponential, 2*718
f/a Fuel-air ratio
/ Mass inertia
K Constant
L Length
M Mach number
M* Ratio of local velocity to the speed of sound at conditions where M = 1
,V Rotative speed
n Index
p Pressure
s Laplace variable
T Total temperature, thrust
I Time
/" Unit step function
v Velocity
II Weight flow
X Displacement
X' Derivative of X with respect to time
X" Second derivative of .V with respect to time
a Angle of attack
r) Ratio of absolute total pressure to absolute total pressure of the A R D C
model sea-level atmosphere
7] d Diffuser efficiency
0 Ratio of absolute total temperature to absolute total temperature of
the A R D C model sea-level atmosphere
T T i m e constant
<f> Ratio of absolute viscosity to absolute viscosity of the A R D C model
sea-level atmosphere;
Phase angle
co Frequency
Subscripts
avg Average o Free-stream
C Test chamber P Primary
cap. Capacity PE Primary nozzle exit
D Duct PT Total pressure primary stream
EX Exhaust s Function of s
ind Indicated T T h r u s t ; total
n Index i Function of time
N Natural
7K
FULL-SCALE PROPULSION TESTING IN
WIND TUNNELS
By R. W . H E N S E L a n d H . K. MATT
AEDC Propulsion Wind Tunnel ARO. Inc.
The use of wind tunnels in the testing of propulsion installations is described.

Only major facilities capable of handling full-scale engines are considered. A
historical review of the development nf full-scale prnpulsion testing in wind tunnels
is presented, envering the range of early low subsonic speeds at clnse to sea level
ennditinns In high supersonic speeds at altitudes of over 30,000 m. The main
emphasis is on the higher velocity, continuous flow wind tunnels, frnm transnnic
upward. Specifically, this includes the French transnnic tunnel at Modane, the
NACA 8 X 6 ft and 10 X 10 ft propulsinn wind tunnels at Cleveland, Ohio,
and the I .S. Air Force Propulsion Wind Tunnel at the Arnnld Engineering
Development Center, Tullahnma, Tennessee. The various techniques for establish-
ing matched altitude conditions of Mach number, pressure, and temperature are
discussed. Detailed descriptions are given ofparticular items peculiar lo prnpulsion
testing in wind tunnels, namely, exhaust gas removal and contamination problems,
scavenging scoop and wall interference effects, special instrumentation and equipment
requirements, and unusual test techniques. Finally, future trends in propulsion
testing in wind tunnels are postulated, with emphasis on rocket installation testing
at high altitudes.
INTRODUCTION
T H E WIND tunnel, being an extremely flexible tool for all aerodynamic
phenomena, has been used very effectively in the testing of full-scale
installations with their propulsion systems in operation. This type of work
has been conducted for the purpose of attaining a number of objectives.
In the days of reciprocating engines, entire airplanes were placed in large,
low speed tunnels such as the XACA Full Scale Wind Tunnel at Langley
Field. A main purpose of such testing was the determination of the effect of
the power plant upon the airplane flight characteristics, particularly with
respect to the propeller slipstream. A second major objective was to resolve
the cooling and cowling problems of the propulsion installation. This work
was also performed in the largest tunnels with the complete airplane
installed, but more frequently only the airplane components adjacent to the
power plant were installed in wind tunnels having test section diameters in
the range of 4-9 to 7-3 m. Subsonic velocities up to those obtainable in flight
were available in some of the tunnels, but always at essentially ambient
atmospheric conditions. T h e first tunnel capable of truly simulating altitude
conditions over a reasonable range was the XACA 20-ft Altitude Wind
Tunnel at Cleveland.
T h e advent of turbojet engines, which are basically aerodynamic machines,
increased the requirements for test data at completely simulated altitude
conditions at progressively higher Mach numbers, which at the present time
79
R. W . HENSEL AND H. K. MATT
are well into the supersonic range. A n u m b e r of propulsion wind tunnels

were built in anticipation of the need for adequate ground test facilities for
both turbojet and ramjet installations. These facilities are also expected to
satisfy the most recent trend requiring live rocket data at altitude for purposes
of evaluating the effect of the rocket jet(s) on the flight article performance.
This paper consists mainly of a description of the methods and procedures
used to establish conditions for propulsion testing in the major wind tunnels
engaged in such activity. T h e emphasis is placed mi all problems, equipment
and techniques peculiar to propulsion work. Furthermore, only continuous
flow tunnels capable of handling full-scale air-breathing engines and large or
full-scale rocket motors are considered. Consequently, jet simulation
techniques, either hoi or cold, for jet investigations are not presented:
propeller testing has been excluded as well. Since wind tunnels serve a
unique purpose compared to engine altitude test cells only when used to test
the power plant in conjunction with the installation, including as much of
the neighboring airframe as possible, the scope of this paper is also limited
to cases of this type.
T h e final restriction is with respect to test velocities attainable. T h e
regions of major significance for military vehicles at this writing are from
transonic u p w a r d ; therefore, the propulsion test techniques presented cover
the transonic and supersonic M a c h n u m b e r ranges except for the historical
review. T h e specific facilities involved are the French transonic tunnel at
M o d a n e , the X A C A 8 X 6 ft and 10 X 10 ft propulsion wind tunnels at
Cleveland, O h i o , and the U . S . Air Force Propulsion Wind Tunnel at the
Arnold Engineering Development Center, Tullahoma, Tennessee.
H I S T O R I C A L REVIEW
Historically, wind tunnels have been used to perform tests which included
full-scale propulsion installations with various objectives in mind. T h e major
objectives may be summarized as follows:
1. Power plant effects on full-scale aircraft.
2. Reciprocating engine installation characteristics.
3. J e t propulsion engine installation characteristics.
T h e first two objectives could generally be attained by use of the large,
subsonic wind tunnels in existence during the time period of concern for
reciprocating engines. Most of the facilities so used are of the multi-purpose
type in that they were used for both aerodynamic and propulsion investiga-
tions. With the advent of jet propulsion, it became necessary to build large,
complex transonic and supersonic wind tunnels which included basic design
features specifically to perform propulsion testing. A more detailed review
of each objective follows.
Power Plant Effects on Full-scale Aircraft

I n the earlier days of military aviation, power plants were of the
reciprocating type using propellers to provide thrust. O n e of the main items
of interest requiring the use of wind tunnels was the measurement of power-on
effects on aircraft performance, particularly with regard to propeller slip-
stream effects. By their nature, these tests required facilities with large test
80
FULL-SCALE PROPULSION TESTING IN W I N D TUNNELS
sections; power considerations and the flight velocities attainable at the time
limited these full-scale tunnels to very low subsonic speeds.
T h e XACA Full-Scale Tunnel at Langley Field (DEFRANCE, 1933) was
the first to perform a test of the type described above. T h e YO-31A airplane
was tested in this tunnel in 1932 (see Fig. 1). T h e O X E R A Full-Scale Wind
Fig. 1.
.In YO-31 A airplane installed in the NACA full-scale wind tunnel at Langley Field
'Funnel at Chalais-Meudon ( L E M A R Z C , 1934) followed shortly thereafter in

1935 along similar lines. SWEBERO and DINGELDEIN (1945) and G O E T T and
PASS (1941) contain the results of representative power-on tests performed
in the XACA facility. Both tunnels have an open test section of elliptic
shape; the Langley Field tunnel is of 30 X 60 ft (9-15 x 18-3 m) size, a n d
the one at Chalais-Meudon is 8 X 16 m. M a x i m u m performance in each
case is roughly equivalent, namely approximately 50 m/sec. All testing is
performed at ground-level conditions, since the open test sections of both
tunnels are also vented to atmospheric pressure.
T h e use of wind tunnels in testing full-scale aircraft is climaxed in the
X A C A 40 X 80 ft (13-1 x 26-2 m) facility at Moffett Field, California.
Completed in 1944, this is undoubtedly the largest wind tunnel in existence.
It is further remarkable in its extremely high energy ratio of over 8, which is
outstanding for a tunnel with a maximum velocity of only 112 m/sec. T h e
40 X 80 ft tunnel has a closed test section with flat top and bottom walls
and semicircular sides of 40-ft diameter, and is also vented to the local
atmosphere. Its functions in propulsion testing are roughly analogous to
those of the earlier full-scale wind tunnels. A photograph of a representative
installation is shown in Fig. 2 ; ROGALLO el al. (1951) and KELLY and
TOLHURST (1955) describe typical tests with the propulsive systems in
operation.
81
W. HENSEL A N D H . MATT
T h e type of test referred to in the preceding paragraphs is basically of an

aerodynamic n a t u r e ; the power plant was operated mainly to determine its
effect upon the aircraft aerodynamics. However, as airplanes increased in
performance and grew in size and n u m b e r of engines, it became more
necessary to use scale models in smaller, but often higher velocity wind
tunnels. Power-on effects on the airplane aerodynamics were measured with
Pig. 2.
An A-26 airplane installed in the NACA 40 X 80fl wind tunnel
the aid of propellers driven by small, powerful electric motors installed inside
the model.
Very little use has been m a d e of the full-scale facilities for tests of jet-
propelled aircraft with the power plant in operation. Contributing factors
to this lack of interest have been the large disparity of the wind tunnel and
aircraft performance capabilities and the usual absence of large engine
airflow effects on the aircraft at low speeds.
Reciprocating Engine Installatinn Characteristics

T h e two problems which have received the greatest attention in tests of
reciprocating engine installations have been cowling or nacelle drag and
cooling. This type of work dates back to at least 1928, when the X A C A ran
tests utilizing a Whirlwind engine in the old Propeller Research Tunnel at
Langley Field (WEICK and W O O D , 1928) in connection with its basic work
on cowling drag reduction (WEICK, 1929a and b). T h e Langley Field and
Chalais-Meudon full-scale tunnels were also called upon to help solve cooling
and nacelle drag problems. An early example at Langley Field was the
nacelle of the U . S . Xavy's P3M-1 aircraft in 1932 ( D E F R A N C E , 1932).
82
As engine installation test requirements increased in the late thirties and

early forties, there arose a new family of tunnels which more or less specialized
in such problems. T h e maximum velocities of the later tunnels in this class
approached and, in some cases for a time, even exceeded the expected flight
speeds of the test articles.
T h e first facility to be used mainly for propulsion installation investigations
Fig, .'>.
The test section of the R A E 24 f t wind tunnel with a full-scale airplane installed
was the R A E 24-ft (7-3 m) tunnel at Farnborough (JENNINGS el al., 1936).

Unpublished reports of this facility, which are dated April and J u n e 1936,
present the results of radial engine cowling tests. Specific aircraft power-
plant installations, namely the Tiger VI and the Blenheim aircraft, were run
in the R A E tunnel in the period of 1937 38. This type of testing was
continued during the wartime period, with such well-known aircraft names
as Beaufighler, Buckingham and Spitfire as examples. Although most of the
propulsion test work of the R A E 24-ft tunnel has not been formally published,
it has been included in a complete monograph which summarizes the
aerodynamics of reciprocating engine cooling as of 1947 (HARTSHORN and
NICHOLSON, 1947). T h e 24-ft tunnel has a circular, open-jet test section
vented to the atmosphere. It is limited by a m a x i m u m velocity in the
vicinity of 51 m/sec, which is little different from the two earlier full-scale
tunnels. Fig. 3 is a photograph of the test section with a full-scale airplane
installed.
T h e French Hispano-Suiza tunnel (MASSUGER, 1941) which was placed in
operation in 1938, also specialized in engine installation tests, ll has an open
83
R. W . HENSEL AND H . K. MATT
test section of 5 m diameter a n d a m a x i m u m velocity of approximately

85 m/sec. A nacelle mounted in the test section is shown in Fig. 4.
Xext there appeared on the scene a group of three multi-purpose facilities
which were utilized to a large extent on reciprocating engine installation
cooling and/or cowling problems. T h e U . S . Air Force's 20-ft (6-1 m ) tunnel
at Wright Field, Ohio, began such testing in 1942; the nacelles of the B-36,
X P - 6 7 a n d C-74 aircraft ( M A R C U S , 1944; BUSSEY, 1945; M A R C U S , 1946)
were among those run in this facility. The 20-ft tunnel has a removable test
•••MB1
Fig.4.
Test installation in Hispano-Suiza tunnel
section insert so that either closed or open j e t testing could be performed

and, in fact, propulsion work was done using either configuration. T h e test
section is of circular cross section a n d the m a x i m u m attainable velocity for
the closed-jet case is 170 m/sec. A photograph of the C-74 installation is
presented in Fig. 5.
T h e XACA's 16-ft (4-9 m) high-speed tunnel at Langley Field was brought
into operation in early 1942 a n d was used for various power plant installation
investigations. T h e A-26 aircraft is a n early example. A n important
contribution of this facility was its development work o n the X A C A high-
speed cowling designs (see H A B E L a n d KORYCINSKI, 1944). As originally
built, the 16-ft tunnel was limited to a m a x i m u m velocity of 224 m/sec.
It was repowercd in 1950 in order to achieve transonic performance, but has
only done aerodynamic work since this date.
Another multi-purpose wind tunnel used for full-scale propulsion work is
that of the United Aircraft Corporation in Hartford, Connecticut. It is
evident that the multiplicity of purpose was so fundamental to this facilltx
that two interchangeable, closed test sections of 8-ft (2-4 m) a n d 18-ft (5-5 m)
size a n d octagonal shape were part of the original design. Xear sonic speed
is obtained with the 2-4 m test section e m p t y ; however, with the 5-5 m
throat, the m a x i m u m performance is only slightly over 90 m/sec. T h e
smaller test section is obviously intended for aerodynamic testing, but
84
FULL-SCALE P R O P U L S I O N T E S T I N G IN WIND TUNNELS
actually both were used for propulsion work. Examples of tests performed
in the 5-5 and 2-4 m sections are reported in GROSE (1946) and M C M A H A N
(1948) respectively; photographs of each installation are shown in Figs. 6
and 7. T h e one in the 2-4 m throat was a pulse-jet test for the U.S. Xavy.
Mach numbers of u p to 0-7 were obtained with propulsion systems in
operation in the smaller throat. Full-scale tests were discontinued after 1951.
All the wind tunnels previously described have one major characteristic
in common, namely that they are all vented to the atmosphere. In the more
Fig. 5.
Installation of the C-74 nacelle in the Wright Field 20 ft wind tunnel
highly powered tunnels, such as the Wright Field 20-ft a n d the Langley Field
16-ft, this venting takes the form of a positive air exchange system for cooling
purposes. Those which are vented in their test region obviously can be r u n
at only one altitude pressure level, that on which the wind tunnel is built.
Those open to the atmosphere elsewhere other than in the test section, could
effect a slight variation in altitude because of the expansion through the
nozzle. This expansion is limited by the maximum attainable Mach number
and only amounted to a maximum pressure altitude of 2700 m in the best
case, the Langley Field tunnel. I n any case, the pressure altitude a n d
velocity are related in one unique function except for the slight variations in
atmospheric conditions. Xor is positive temperature control available in any
of the facilities discussed thus far. Fortunately, PINKEL (1938), SCHEY et al
(1938) and PINKEL a n d ELLERBROCK (1940) of the XACA showed that it was
possible to correlate the cooling data from air-cooled reciprocating engines
in such a manner that the effects of altitude pressure and temperature
variations, as well as the engine variables, could be predicted. Thus, very
useful d a t a were obtained in these atmospheric wind tunnels even though the
operational altitudes of military aircraft had increased quite considerably.
85
R. VV. H E N S E L AND H. K. MATT
Fig. e.
Installation of R-1830-90-B experimental nacelle in the VAC 18ft wind tunnel
Fig. 7.
Installation of a U.S. Naty RTV-N-15 test capsule in the UAC 8ft wind tunnel
86
FULL-SCALE P R O P U L S I O N TESTING IN WIND TUNNELS
However, it became apparent to the XACA that the many problems

involved in the development of reciprocating power plants, of which cooling
was just one, did require a ground facility which could fully simulate altitude
conditions. T h e consequence of this requirement was the Altitude Wind
Tunnel at Cleveland, the first laboratory capable of matching both altitude
pressure and temperature conditions simultaneously over a reasonable range.
Obviously, this is a variable density tunnel having an internal refrigeration
cooling system. A scavenging system sized for the exhaust gases from large
Fig. 8.
A B-29 nacelle installed in the NACA 20fl Altitude wind tunnel
reciprocating engines was installed to keep the main tunnel airstream con-
tamination to a minimum. Altitude simulation u p to 16,000 m is obtained;
independent control of temperature and pressure assures a sizeable speed
range at any given altitude at the correct temperatures. T h e test section
diameter is 6-1 m and the maximum test section velocity is approximately
225 m/sec.
T h e Altitude Wind Tunnel was completed in late 1943 and was imme-
diately rushed into service in urgent tests of various propulsion installations.
O n e early example is the B-29 cowling and cooling studies ( M O N R O E and
SAARI, 1946, and Fig. 8). I n later years as the M a c h number capabilities of
the Altitude Wind Tunnel became more insufficient for the latest aircraft
propulsion installations, it was used more as an air supply in a fashion
similar to a test cell of a n engine test facility. Mach numbers well above 1-0
could be simulated in this manner.
Jet Propulsion Engine Installation Characteristics

T h e advent of the turbojet engine, which is basically a n aerodynamic
machine, considerably increased the requirements for ground tests, particu-
larly under simulated altitude conditions. T h e entire propulsive jet has to be
87
R. vv. HENSEL AND H. K. MA'n-
passed internally through the aircraft, making the optimum design of the
inlet ducting system a more critical problem than with reciprocating engines.
T h e inlet problems were further aggravated as flight M a c h numbers
approached sonic speed and as it became evident that the quality of the
airflow entering the engine compressor had an important effect upon
compressor, and thus engine performance. Hence, to insure proper matching
of the aircraft engine and induction system, it was extremely desirable to
match as much as possible of the flight spectrum in the wind tunnel.
SECTION A-B
V ontm*
POWER FOR
IGNITION
SCALE
Fig. 9.
V-l pulse-jet installed in the LFA 2 8 m wind tunnel
T h e first test of a jet propulsion installation was performed in Germany,

specifically in 1942 at the 2-8 m high-speed tunnel of the LFA at Braun-
schweig. T h e test article was the V-l pulse-jet engine; a sketch of the test
article in the test section is shown in Fig. 9. T h e tunnel was of the multi-
purpose category and was capable of near sonic performance. This facility,
which had its circuit erected in the vertical plane, included the interesting
feature of being able to retract the two corners most remote from the test
section into the long return leg, thus creating an open return wind tunnel
capable of testing propulsion installations without danger of using contami-
nated air (see Fig. 10). T h e increased severity of contamination in a wind
tunnel using just partial air exchange for jet propulsion engines as compared
to the reciprocating type is quite obvious. However, this type of open-return
tunnel suffers from the same lack of altitude performance as all atmospheric-
vented facilities, as discussed previously.
A similar multi-purpose, near-sonic tunnel of 3-m test section diameter was
completed near Munich in April 1945. In this case, the short return leg was
simply disconnected and removed for propulsion tests (see Fig. 11). This was
a variable density tunnel as well, but only for non-propulsion testing, of
88
68
Co ciS'
8 «
siaNNni Q N I M NI oNiisa.t. NOISKIJOUJ aivos-iinj

R. W . H E N S E L AND H. MATT
course. Xo productive work was accomplished in the Munich tunnel; both

it and the Braunschweig facility were dismantled in 1945.
T h e first wind tunnel test of a turbojet engine installation in the United
States was that of the Bell XP-59 aircraft in the X A C A Altitude Wind
T u n n e l in February 1944 (PRESTON et al., 1946). This test coincided with the
tunnel completion; in fact, only altitude pressure simulation was possible
because the refrigeration equipment was not completed in time. Although
the Altitude Wind Tunnel had not been designed for large jet propulsion
installations, particularly as regards exhaust gas removal, it proved to be a
SECTION B-B
.....,» SECTION A-A

2 80» C-)
camm
wot
Fig. 11.
The Munich 3 m wind tunnel
very useful tool in the testing of a n u m b e r of the early American jet aircraft
and engines. Moreover, it was for some years the only facility which could
simulate a large part of the flight spectrum of current aircraft. A variety of
turbojet and turboprop installations were tested; the P-80A airplane and the
General Electric TG-IOO and Armstrong Siddeley Python turboprop engines
in flight-type nacelles are examples.
During this same period of the middle forties, U.S. Air Force service-type
tests on j e t propulsion systems were conducted in the Wright field 20-ft
tunnel. T h e F-80 aircraft and the U.S. version of the V-l are examples; a
photograph of the latter in operation is shown in Fig. 12.
As the design performance of new aircraft entered the transonic and
supersonic ranges, it became rather apparent that the existing wind tunnels
were no longer adequate and that new facilities were needed. As in most
previous cases, the new equipment would also be of a multi-purpose nature,
in that non-propulsion work could also be handled. T h e r e are four labora-
tories in the Western World which fall into the category of interest, namely
the French 8-m transonic tunnel at M o d a n e , the X A C A 8 X 6 ft (2-4 X 1-8
90
F U L L - S C A L E P R O P U L S I O N TESTING IN W I N D TUNNELS
m) and 10 X 10 ft (3 X 3 m) propulsion tunnels at Cleveland, Ohio, and the

U.S. Air Force Propulsion Wind Tunnel at Tullahoma, Tennessee. T h e
Fig. 12.
dB-2 robot bomb in the Wright Field 20ft wind tunnel with pulse-jet engine operating
capabilities, techniques, equipment, and procedures for propulsion testing as

applied at these facilities is the main topic of this paper.
PROPULSION TESTING AT T R A N S O N I C AND

S U P E R S O N I C SPEEDS
As previously stated, the four propulsion test laboratories to be considered
in detail arc all of the multi-purpose type in that they also do other work not
involving engine operation, particularly tests of a n aerodynamic nature.
Basically, the tunnels are of two types: open to the atmosphere and variable
density. T h e Modane and the XACA 8 x 6 and 10 X 10 ft tunnels all fall
in the former category, while the P W T is in the latter. Each of the various
aspects of propulsion testing will be presented individually. However, a
brief outline of the basic characteristics of each facility would seem appro-
priate at this time as an introduction to the propulsion test features.
T h e Modane tunnel, which is operated by O X E R A (Office Xational
DT.uides ei de Ret herches Aeronaut iques . was relocated in the French Alps
in 1945-1951 after it had been partially completed at Oetztal in the Austrian
Alps. A diagram of the entire system is given in Fig. 13. T h e impressive size
of this facility is better indicated by a photograph (Fig. 14); this is very
probably the largest transonic tunnel in existence. Although of the closed-
return type, up to 70 per cent air exchange is possible. Discharge is annular
'•1
R. W. HENSEL AND H. K. MATT
HUHUP AREA
INSTALLATION
AREA
NE
TURBINE 1 , La-#Kf£aJL
FUEL
n • • • • ! ! ! i : ; ; ; ; ! ; ; ; i ! •
. iSiiiiiiiiilii;iiiiliiii:i:i:iiiiiii;lila
TURBINE 2 'UL
Fig. 13.
Diagram of the ONERA transonic wind tunnel al Modane
Fig. 14.
Photograph of the Modane transonic wind tunnel system
around the outside of the contraction section inlet; hence, the test section
total pressure is atmospheric. T h e drive system is directly hydraulic, using
two Pelton turbines, and is rated at a total of 110,000 h.p. T h e test section
walls are solid, limiting the tunnel-empty velocity to sonic; however, very-
close to Af = 1 may be achieved with the smaller installations, particularly
ramjet, when supported from a swept strut.
92
R. W . H E N S E L AND H . K. MATT
T h e XACA 8 X 6 ft tunnel at Cleveland was placed into productive

operation in 1949 with a design Mach number range of from 2-1 down to a
lower limit determined by model blockage or shock reflection. A noise
problem which developed immediately after initial operation, required that
an additional acoustic muffler be added downstream of the exhaust diffuser
in 1950 (see BERANEK et al., 1955). Because of this tunnel's proximity to
highly populated areas, the noise problem was not a function of the decibel
level-frequency spectrum alone. In 1956, the test section was modified from
solid walls to perforated walls in order to obtain transonic capability down to
a Mach number of 0-8. Fig. 15 shows this facility to be of the non-return type
TEST SECTION
PLENUM CMAMBtH
Fig. 15.
The NACA 8 x 6ft wind tunnel
with a large air dryer at the air inlet and a multi-stage compressor driven by
three electric motors totaling 87,000 h.p. A flexible nozzle precedes the test
section, which is enclosed by a large plenum chamber. Depending upon the
time of year, operating time is limited to the range of 1 1 0 hours before dryer
regeneration is necessary. T h e return circuit was closed and a cooler was
added in 1957 lo permit non-propulsion testing without this severe time
limitation, but with provision for retaining atmospheric discharge during
propulsion tests.
T h e XACA 10 X 10 ft tunnel has a Mach n u m b e r range of 2 0 to 3-5,
which complements the performance of the 8 X 6 ft very well. It is part of
the U.S. Unitary Plan for aeronautical research facilities and conducted its
first productive test in 1956. As may be seen in Fig. 16, the return circuit is
interchangeably open or closed through appropriate actuation of valves 1 and
13. In the open return case for propulsion work, the air dryer limits operating
periods between regeneration to a range of \ to 9 hours, depending on
ambient conditions. A very effective exhaust muffler installation is located
just downstream of valve 13 in the open return cycle (see HOOVER, 1958).
Two compressors are used, each with a separate electric motor drive system.
Compressor 1 is attached to 4 motors totaling 150,000 h.p., and compressor 2
is driven by 3 motors totaling 100,000 h.p. Compressor 2 is generally not used
until a Mach number of approximately 2-5 is reached. Two coolers are
93
included in the tunnel circuit, but they are basically for compressor inlet
temperature control during closed return operation. T h e solid-wall test
section is preceded by a flexible wall nozzle. A direct-fired gas heater has
recently been added just upstream of the contraction section.
T h e U . S . Air Force 16 X 16 ft (4-9 x 4-9 m) Propulsion Wind T u n n e l
(PWT) at the Arnold Engineering Development Center (AEDC), which is
also part of the U n i t a r y Plan and is operated by A R O , Inc. under contract
with the U.S.A.F., is in reality two separate wind tunnels. T h e transonic
circuit covers the M a c h n u m b e r range of 0-5 to 1-6 and the supersonic
circuit is designed for 1-5 to 5-0; a common drive system is used for both
SECONDARY-CttlVE
BLOC
COMPRESSOR 2
REENS AND
--IGHTENING VANES
FLEXIBLE-WALL NOZZLE
DAMPR ROOM
CONTROL ROOM
Fig. 16.
The NACA 10 X 10 fl wind tunnel
circuits. Fig. 17 shows a n outline of the transonic circuit, which performed

its first productive test in early 1957. T h e supersonic circuit is under
construction and is presently scheduled to begin test operations in 1960; it
will very probably be the largest supersonic wind tunnel in existence.
Fig. 18 shows the construction progress as of early 1958. Both circuits are of
variable density for all types of tests. A scavenging system is provided for
removal of products of combustion and is connected to the exhaust compressor
system of the A E D C Engine Test Facility ( E T F ) . T h e exhausted air is
replaced by dry air from the E T F air supply system. Pressure level control is
achieved by differential use of the exhaust and replacement air system.
Transonic circuit temperature control above the minimum dictated by the
cooling water temperature is achieved by variable water flow through the
cooler. T h e test sections are interchangeable in either circuit. Each circuit
has a flexible nozzle and a plenum chamber surrounding the test area. T h e
presently installed portion of the plenum evacuation compressor system is
powered by three electric motors rated at a total of 77,500 h.p. T h e m a i n
drive consists of four motors totaling 216,000 h.p. T h e m a i n drive is a
symmetrical system which can be split to half power for each circuit for
simultaneous operation. Xeither the M o d a n e tunnel nor the P W T have very
94
-
=
-r-
y.
S.
- -
Ln w
as
d
1 "BS1"
WLOMS
-
-
Fig. 17.
The transonic circuit of the AEDC propulsion wind tunnel
serious external noise problems; the decibel level is not particularly low,
but the neighboring population is.
Some of the material to be presented under this section has already been
published. T h e main references used are PIERRE el al. (1951) and R O Y (1956)
for the M o d a n e tunnel, X A C A M a n u a l (1956) for the X A C A facilities, and
Test Facilities Handbook, X A C A (1958), DELANO (1958) and TEMPELMEYER
et al. (1958) for the P W T .
Fig. 18.
The supersonic circuit of the A E D C propulsion wind tunnel, early 1958
Performance Capabilities
T h e two types of propulsion wind tunnels, open to the atmosphere and
variable density, show their difference quite clearly in a plot of the pressure
altitude attainable as a function of M a c h n u m b e r (Fig. 19). Because of the
fixed intake pressure and the upstream compressor characteristics in the case
of the XACA tunnels, there exists only a single function of pressure altitude
and M a c h n u m b e r for the atmospheric tunnels. T h e values given for the
P W T supersonic circuit are only estimates, of course, although transonic
tunnel experience has shown that they may be slightly conservative. T h e
m a x i m u m altitudes shown in both P W T circuits are largely dependent on
the leakage rates in the ducting and, therefore, are somewhat variable.
Full altitude simulation requires matching of both temperature and
pressure at the desired M a c h number. T h e pressure situation is quite
straight-forward, but temperature matching is a more stringent problem
from the facility standpoint and, therefore, some clarification of its need is in
order. First of all, it is readily apparent that a n incorrect stream temperature
•M,
F U L L - S C A L E P R O P U L S I O N TESTING IN WIND TUNNELS
combined with the desired M a c h number will lead to an incorrect velocity,

and hence volume flow, entering a turbojet engine compressor. However,
the engine rotor speed can generally be normalized to correct for temperature
mismatch by means of the parameter n / \ / 0 , where n is the measured rotor
speed and 0 the ratio of the actual to the reference stream temperature. A
problem does arise when the actual stream temperature exceeds the desired
value for matching. T h e n it is necessary that n be increased the same as \ / 0
in order to obtain data for the complete operating range of the installation.
In the region near the maximum rotor speed, which is usually an area of
major concern, this would require a physical overspecding of the engine.
Therefore, when the wind tunnel stream is too " h o t " , it is often not possible
to simulate a propulsion installation's peak performance at altitude.
10 s X 6 0
1 1 1
PWT SUPERSONIC
M-
R A N K NIC
CIRCU
\ -e
y*
// / iA\
/ P W T SUPERSONIC
/ CIRCUIT M I N I M U M
NACA 10 X 10
X \- N«( < • X t>

M0D««t
/ l \ .
y/ \» incur MINI) ftUM
— • -
/ ...
2 0 3 0
MACH NUMBER
Fig. 19.
Attainable pressure altitudes for the four propulsion facilities
Another item for which correct temperature matching is highly desirable

is the investigation of the response of the engine and its installation to
transients. Certain response characteristics depend very much on the
temperature level. T e m p e r a t u r e is also an important parameter when the
wind tunnel lest has for at least one of its purposes the gathering of service-
type data on the installation. T h e P W T is concerned with the latter problem
because- of its mission to test specific weapon systems. O t h e r but perhaps
more secondary reasons for matched altitude temperature operation include
the effect upon combustion characteristics, Reynolds number and test
article heat transfer characteristics, including their effect upon the external
flow boundary layer. Exit nozzle flow characteristics may also be somewhat
affected by temperature, particularly when a large amount of secondary flow-
is involved.
For a wind tunnel with a constant total pressure close to atmospheric,
there occurs the rather fortunate circumstance that close altitude temperature
97
matching occurs almost automatically through the transonic range. T h e

total temperatures which correspond to the pressure altitude performance of
the M o d a n c facility are plotted in Fig. 20. Obviously, the range of required
temperatures are quite close to what can be achieved in normal operation,
particularly in the wintertime. T h e temperature variation with increasing
M a c h n u m b e r agrees with the requirements for normal operation due to
increased power input and also the necessity of avoiding condensation.
T h e total temperatures for matching in the X A C A 8 X 6 ft tunnel, based
on the altitude characteristics shown in Fig. 19, are also given in Fig. 20.
SUMMERTI ME MAXIMUM
/
AVAIL AMI
/ I
uj-100 /
//
/
/' y
| X 6 —
X
s
14C D4.NE
s1
y
y x\
V AVAIL A8L
0.8 1.2 6
MACH NUMBER
Fig. 20.
Total temperatures required for matching in the Modane and NACA 8 X 6ft wind tunnels
T h e location of the tunnel compressor ahead of the test section has m a d e it

possible to obtain temperatures quite close to those required for the entire
M a c h n u m b e r range. T h e magnitude of mismatching depends on the
season of the year; the total annual variation for the 8 X 6 ft tunnel is also
included on Fig. 20.
Beyond a M a c h n u m b e r of 2-0, the necessary total temperatures increase
very rapidly, presenting a n u m b e r of difficult design problems for large
propulsion facilities. In the 10 X 10 ft tunnel, it was necessary to add a
heater upstream of the test section to obtain some temperature matching
capability since only compressor 2 contributes to the air heating process.
Complete matching is available to 41/ = 3-0, which corresponds to a total
temperature of 435°C, beyond which the stream temperature is too " c o l d " .
T h e P W T system has positive control of the temperature level by means of
variable flow (6000 l./sec maximum) through heat exchangers which use
reservoir water as a cooling fluid. T h e regions of complete altitude tempera-
ture matching are noted on the performance chart (Fig. 21) for both the
transonic and supersonic circuits. Since the minimum total temperature is

limited by the cooling water temperature, the low Mach number—high
altitude region is loo " h o t " . Some idea of the seasonal variation possible is
also given in Fig. 2 1 ; an 11°C drop in water temperature results in an
extension in the Mach number range for matched temperatures of 0-1. T h e
inverse problem occurs in the supersonic circuit, where the M a c h number for
matching is limited by the allowable total temperature. This limit was set
during the tunnel design at 340°C in order to be able to use the normal low
10 s x 6 0
50
/TOO. L
- T R * NSON 'COL0//////JS
cm UIT
^
i30
m/
O
u - Too^-r
A 340° C
ERATL RE
ml 2 0 "MOT" - p
1 CO 0 L I N G WATE
•c
sNN •c
SUPERSONIC
CIRCUIT
1
SL
Z.O 3.0
MACH NUMBER
Fig. 21.
P W T performance chart showing altitude temperature matching capabilities
temperature structural materials. Therefore, this tunnel will run too " c o l d "
beyond a Mach number of approximately 3-0. T e m p e r a t u r e control will be
achieved by means of two large heat exchangers, one ahead of the com-
pressor to keep compressor inlet temperatures below approximately 100°C
and the other downstream of the compressor for direct air temperature
control.
Exhaust Cas Removal

Scavenging techniques.—The problem of removing the products of combustion
during a propulsion test in a wind tunnel is generally the overriding factor in
facility design. Certainly a variable density tunnel is necessary for complete
altitude simulation. However, the complexity of a scavenging system with
Its requirement foi a sci of compressor* for exhausting the products ill
combustion to the atmosphere, and a complicated scavenging scoop structure
and mechanism constitute a real drawback. Furthermore, until just recently
the aerodynamic problems of scoop interference on a propulsion model, jet
spreading characteristics, scoop effects during non-propulsion tests, and
overall effect of the scavenging system on the wind tunnel performance were
largely unanswered. Although the exhaust compressor system required for
W
scavenging could also provide pressure level control and the need for a very
large dryer for the entire tunnel airflow could be eliminated and a smaller
unit substituted which just matches the exhaust flow capacity, the resulting
philosophy for new designs was to avoid scavenging systems.
Since the P W T had as perhaps its major design criterion that it be able to
conduct full-scale propulsion tests of specific weapon systems over as full a
range of their Mach n u m b e r - a l t i t u d e spectra as practicable, it is the only one
of the four laboratories under consideration which is of the variable density
type. T h e construction of a set of exhaust compressors was avoided by
connecting directly to the A E D C Engine Test Facility system. T h e E T F air
supply system is also used as a source of make-up air. A schematic sketch of
the P W T scavenging and make-up air systems is shown in Fig. 22.
TRANSONIC SUPERSONIC
CIRCUIT CIRCUIT
COMPRESSOR
| '^lt.tSraiii.1-17'1
h AFTER
\C0OLER
CONFIGURATION
II • 9 $ T A «
P
CONFIGURATION
ALSO AVAILABLE)
IP FLOW
J, ii CONVERTER
AjJY PASS I i
SYMBOLS
SCAVENGING
i
J ijl
XJI
TEST
SECTION
-I I I I—3 STAGE COOLER
* S
COMPRESSOR
^ 1 I
1,'
COOLER
SHUT OFF VALVE
CONTROL VALVE
Fig. 22.
Schematic diagram of the P W T scavenging and make-up air systems
As stated previously, the two N A C A facilities are provided with atmo-

spheric exhaust for propulsion testing, and therefore avoid airstream contami-
nation problems. T h e M o d a n c tunnel, by virtue of its ability to exchange
upwards of 20 per cent of its airflow and its tremendous volume, is able to
keep contamination within tolerable limits.
T h e P W T scavenging system is intended to capture all the exhaust gases
during normal operation and thus keep contamination to a negligible level.
Once the scoop axial location with respect to the nozzle exit is determined
from interference-free considerations, the m i n i m u m allowable scoop inlet
size is dependent upon the m a x i m u m rate of jet spreading expected for a
particular installation. T h e scoop inlet size fixes the scavenging and makeup
air requirements, which in turn limit the available operating range for a
given engine.
Jet spreading. ••- T h e jet spreading problem is a complex one involving many
variables, some of which are nozzle exit static pressure ratio, nozzle exit
M a c h n u m b e r , stream Mach number, afterbody shape and angle of pilch
or yaw. T h e A E D C has sponsored a series of model tests on this subject for
100
direct application to both circuits of the P W T . T h e final phase of the

program is currently in progress in the 1/16-scale model tunnels of the P W T .
T h e details of these model tests will not be discussed herein; however,
T A T R O (1955) contains some of the interim results and a list of additional
references on the jet spreading problem.
Airstream contamination.—In the P W T transonic circuit, airstream con-
tamination is monitored continuously; flow samples are usually taken at the
engine induction system inlet and adjacent to the top or bottom edge of the
scoop tip. A more systematic investigation of contamination was conducted
as part of a calibration program using a strut-mounted pod installation
containing a Westinghouse J-34 engine (Figs. 23 a n d 24). In this case, the
STATION DESIGNATION IN METERS

A U . OTHER DIMENSIONS IN CENTIMETERS
LONG SCOOP TIP

INTERMEDIATE SCOOP TIP
SHORT SCOOP TIP I
Fig. 23.
Diagram ofd-34 nacelle installation in the P W T transonic circuit
engine exit flow was always subsonic and therefore the tunnel airstream and
nozzle exit static pressures were equal. Three different scoop tip lengths of a
constant inlet diameter of 107 cm were used. There was no evidence of
contamination even with the shortest tip throughout the transonic Mach
n u m b e r range when the engine installation and scavenging scoop pitch
angles were both zero and the mass flow into the tip per unit area was
equivalent to that in the free stream. T h e effective edge of the jet was deter-
mined by pitching the scoop tip out of line with the engine exhaust. T h e
results of one such run at a Mach number of 0-95 (Fig. 25) indicate that a
vertical tip displacement of 16-5 cm was necessary before measurable
contamination was obtained. Therefore, for this particular case the total
included angle of the effective jet boundary is only 7-5 deg. This angle
represents about the minimum value which can be considered for this
particular installation geometry in the selection of the tip diameter. T h e
effects of angle of attack or yaw, exit static pressure ratios above unity and the
full range of relative velocities of the jet and the free stream, to name a few of
the more important parameters, would all tend to increase the required tip
diameter for a given axial position.
T h e transient effect on the oxygen concentration measurement at two
locations is plotted in Fig. 26 for the scoop tip displaced a distance of 48 cm,
which is more than sufficient to cause contamination. All conditions are the
101
Pig. 24.
Photograph ofd-34 nacelle installation in the P W T transonic circuit with short scoop tip
43 2 cm y» ,
-E ANALYZER PROBE LOCATION

SCOOP LIP
9
40 Q
UJ W
ym \ MACH NUMBER • 0 » 9
< Ul
Q- UJ
20 \ PRESSURE A L T I T U D E ' 1 0 , 7 0 0 METERS
SCOOP INLET MASS FLOW RATIO • 1.0
\ SHORT T I P INSTALLED, 107 c m . OIA
ENGINE SPEED t t i PERCENT OF RATED
0
9 20 1 22
0X1 GEN CON ;ENT RATI 3N, F ERC • N T
8F
•>- I -
20
Fig. 25.
Effect of scoop tip displacement on airstream oxygen content
same as for Fig. 25 except that the stream M a c h n u m b e r is 1-10 and the scoop
displacement is downward, so that the probe at the top of the scoop lip was
used. It is a p p a r e n t that the time lag between scoop displacement and a
correct reading of oxygen concentration is approximately 10 min in this
particular case; this is the time interval used in obtaining the data of Fig. 25.
It is also shown that the a m o u n t of contamination of the air entering the
102
FULL-SCALE P R O P U L S I O N TESTING IN WIND T U N N E L S
engine is almost negligible. This is not too surprising in view of the small
engine size; the weight flow was only 9-5 kg/sec at the conditions for which
the data are presented, which is a fraction of the scavenging and makeup air
capacity.
Maximum engine size.—The maximum engine size which can be tested in a
given wind tunnel is, among other things, a function of the exhaust gas
22
z 1, • - 4 8 . 5 CENTIMETERS
o 0 SCOOP LIP (TOPI
a ENGINE INLET
21 MACH NUMBER • 1.10

"-—,—-en
S5 \ X 1 HX-> o a
u
PRESSURE A L T I T U D E ' 1 0 , 7 0 0 METERS
u
z SCOOP INLET MASS FLOW RATIO • 1.0
Si 20
15
SHORT T I P INSTALLED, 107 c m OIA.
ENGINE SPEED • 9 2 PERCENT OF RATED
z
u
13
>-
0 10 20 30 40
5 T I M E , MINUTES
Fig. 26.
Transient effect of step displacement of scoop tip on airstream oxygen content
removal capabilities of the facility. Since these capabilities arc quite high for
the atmospheric exhaust tunnels, the NACA 8 X 6 ft a n d 10 X 10 ft, the
size limitation is determined by other factors. T h e M o d a n e facility is
somewhat affected by increasing engine size because of its less than complete
air exchange, although contamination is usually not serious because of the
MAXIMUM ALTITUDES
TUNNEL CURRENT ENGINES
/
r
/ ' w- /
X
ww- s
TUNNEL
rroiV1*
MINIMUM ALTITUDE
KMINIMUM ALTITUDE
'A 1
rX & f
FOR MAKE-UP AIRFLOW
SCOOP DIAMETER
A 2 METERS
JO 2 0 30
NUMBER MACH NUMBER
Fig. 27.
Effect of makeup air capacity on P W T performance capabilities for
scoop inlet diameters of 2 and 2j\/3m
dilution effect of the very large volume of the tunnel circuit. I n these three
facilities, the limiting installation sizes are determined by such considerations
as blockage, starting requirements a n d wall interference, particularly when
simulation of free stream exit conditions is desired.
The performance capabilities of the dry makeup air system available to
the P W T are presented in Fig. 27 for two scoop tip diameters, 2 and 2 / \ '3 m.
Kil
R. W . H E N S E L AND II. K. MATT
T h e larger value represents an estimate of the scoop size required for scoop-
interference-frec testing of large turbojet engines. T h e smaller size corre-
sponds to both an engine of intermediate size without scoop interference and
approximately to a large engine with the scoop tip connected directly to the
exhaust nozzle. T h e required scoop tip to engine exit area ratio to eliminate
scoop interference has been assumed to be a constant at 3, a value which
should be adequate for most normal turbojet installation test programs
according to initial jet spreading a n d interference investigations. Hence, a
large engine is defined as one having a nozzle exit diameter of 2 / y / 3 m, a n d
a n intermediate one of 2/3 m exit diameter.
A band of m a x i m u m altitude which can be expected from modern turbojet
engine installations is included in Fig. 27 to establish the upper limit of
possible operation in the P W T . T h e present dry make-up air limit is
70 kg/sec, which obviously would limit testing capability of large engines to
the direct connect case. However, there are plans for increasing the dry air
capacity in the near future u p to a value of 340 kg/sec, which will produce
the improvement in test capability shown in the figure. A separate drier
drawing in atmospheric air which is similar to those in use at the N A C A
propulsion tunnels has been designed.
T h e exhaust gas scavenging system mass flow available to the P W T more
than matches the make-up air capacity at all test conditions, even with the
contemplated increase to 340 kg/sec. Hence, even a further extension in the
altitude M a c h n u m b e r test spectrum is possible by further increases in dry
air capacity u p to approximately 500 kg/sec.
Interference Effects
I n full-scale propulsion testing, interference effects on the test article a r e of
three types, namely, that caused by the tunnel walls, the installation support
system and the scavenging scoop.
Wall interference.—The two types of wall interference of interest here are
those caused by solid wall a n d ventilated test sections. T h e M o d a n e and
N A C A 10 X 10 ft tunnels have solid walls; the other two facilities use
ventilated walls. Both the P W T a n d the N A C A 8 X 6 ft tunnel use the
"differential-resistance" perforated type developed at the A E D C (GOETHERT,
1956). Both slotted a n d perforated walls are quite successful in eliminating
the choking effect prevalent near sonic speed with solid walls. Therefore, it is
now possible to test a finite size installation in the vicinity of M a c h n u m b e r
1 -0 with the expectation of obtaining reasonably valid results. I n cases where
only the inlet flow is of primary interest, it is possible to obtain the desired
flow at sonic speed providing the engine inlet is not disturbed by more
upstream portions of the airframe. For those cases in which complete
simulation of the external airflow is also desired at transonic speeds, the wave
cancellation characteristics of the walls become a limiting factor. No type of
ventilation is completely effective just at and slightly above M a c h n u m b e r
1 0 . T h e perforated type used by the A E D C - P W T a n d the N A C A 8 X 6 ft
tunnel has been designed for a M a c h n u m b e r of approximately 1-3 and is
effective for the range of 1 • 15 to about 1 -6. Variable porosity could improve
the performance in the 1-0 to 1-15 M a c h n u m b e r range somewhat, but is
difficult to achieve in practice.
104
F U L L - S C A L E P R O P U L S I O N TESTING IN WIND T U N N E L S
T h e Modane tunnel is subject to the conventional subsonic wall inter-

ference and choking effects of all sonic solid wall wind tunnels. T h e large
test section size permits these effects to be minimized for reasonable test
article sizes. T h e test sections of the perforated-wall wind tunnels are
enclosed by large plenum chambers (see Figs. 15 and 17) and auxiliary
suction is provided to permit an increase in the size of installation which may
be tested successfully and to improve the overall tunnel efficiency. This
auxiliary system is particularly necessary for the N A C A 8 X 6 ft tunnel,
which was originally designed for the Mach number range of 1-4 to 2-1 with
solid walls and, consequently, has no flaps at the end of the test section.
Flaps are often used as a ready means of providing a certain amount of
plenum pumping, but at a considerable loss in tunnel operating efficiency.
T h e P W T has both flaps and a plenum evacuation system to go with its
16 X 16 ft (4-9 X 4-9 m) test section because of its mission of simulating the
external and exit flow characteristics of specific weapon systems propulsion
installations. Some of these installations are quite large in both frontal area
and length, making even the P W T test section, which also has a usable
length of over 12 m, too small to satisfy the external flow requirement at
transonic speeds in the largest cases.
At supersonic speeds, wall interference appears mainly in the form of wave
reflection. For a solid wall tunnel, such as the NACA 10 X 10 ft (see Fig.
28), the maximum length of wall interference-free installation is limited by
the Mach number, tunnel dimensions and the size and geometry of the most
upstream portions of the test article as well as its lateral location within the
test section. T h e length thus determined will be considerably shorter than
those of some test articles planned for the P W T supersonic circuit. T h e r e -
fore, a set of perforated wall liners has also been developed at the A E D C for
use in this tunnel; the design Mach number is 2-0 and the usable range is
expected to be from 1-5 to 3-0. T h e capacity of the plenum suction system is
also being increased to match the increased auxiliary flow requirements for
wave cancellation at supersonic speeds.
Support system interference.—Almost all test articles are installed in transonic
and supersonic wind tunnels by one of three m e a n s : (1) on one or more
stings through the base of the test article; (2) on a single or full-span strut
mounted to the test section walls, and (3) directly to the test section wall.
Only cases (2) and (3) are of interest for full-scale propulsion installation
testing.
T h e most common type of propulsion installation mounting is by means of
a strut. These struts can often become quite sizeable because of the various
demands for engine utility lines, controls and usually a considerable n u m b e r
of instrumentation leads and hence can present a n interference problem.
Fortunately, the problem is minimized for external pod arrangements where
the strut can, at least in part, represent the flight configuration. Otherwise,
the interference can only be minimized by the obvious techniques of reducing
the ratio of strut maximum thickness to test article mean diameter to a
minimum and by the use of sweep, preferably downstream from the installa-
tion to the wall. Also, for tests in which the base and nozzle exit flow is being
investigated, the strut trailing edge should be as far upstream with respect
to the base as possible. Another strut technique is to increase the strut chord
105
as far as practicable while keeping the cross-sectional area fixed. T h e natural

conclusion of this process is to approach zero thickness ratio, or a flat plate.
In this manner, at least the wave interference problem could be minimized.
Model tests on one particular test article at the A E D C have indicated that
chordwise strut extension downstream of the installation's base must be done
Fig. 28.
Test section of the NACA 10 X 10ft tunnel with 40 cm ramjet installed
carefully or the base pressure will be affected appreciably. T h e surface

pressures neat the base were less affected l)\ strut interference above M I •()
when the long strut was used. A practical limit is usually reached early in
the chordwise extension process due to structural limitations, particularly
when tunnel starting or stopping loads at supersonic speeds must be provided
for. In any case, the wake, always present off any support strut, makes it quite
106
desirable to locate any downstream external instrumentation in a peripheral

orientation out of line with the plane of the strut.
Mounting the test article directly to the test section wall offers, in principle,
a means of minimizing support interference, particularly when the test section
has a square or rectangular shape and the wall could represent a plane of
symmetry for the installation. Another advantage of this type of mounting is
that it increases the effective size of the wind tunnel since the test article size
can be cut by a factor of two. O n e of the first problems which arise from this
so-called "reflection-plane" technique is what to do with the tunnel boundary
layer. This could be rather critical with some inlet configurations placed
close to the aircraft fuselage. O n e solution which has had some success in
strictly aerodynamic model tests is simply to raise the plane of symmetry into
the airslream by at least the boundary layer displacement thickness by use of
a spacer block. A second procedure which appears promising, but has yet to
be attempted on full-scale propulsion installations, is to bleed off the boundary
layer just upstream of the test article. Those facilities which have auxiliary-
suction equipment, the A K D C - P W T and the NACA 8 X 6 ft tunnel, could
do this fairly readily. A third procedure, which has been used at various
times for aerodynamic tests, is to install the model on a solid plate located at
least one boundary layer thickness from the tunnel wall, permitting the tunnel
boundary layer lo pass underneath the plate. This technique has never been
too successful because of usually high flow inclination at the leading edge of
the plate, which causes cither an irregular M a c h number distribution or even
flow separation.
A second question which arises in the use of the reflection plane technique
in ventilated test sections is what to do with the tunnel wall to which the test
article is attached. The logical solution is to change the entire attachment
wall to the solid type since this configuration is aerodynamically correct for a
plane of symmetry. T h e other three walls would remain unchanged. Some
three-dimensional effects would be present because of the finite cross-stream
dimensions of the solid wall as compared to the requirement for infinite
extension, but these effects arc expected to be small. Some success has been
experienced in the use of this reflection-plane arrangemei'l al the AEDC
usiiIL« .MI aerodx uamic model and a small transonic w ind tunnel. T h e Abl K -
P W T is particularly suitable for this type of test because its test section walls
consist of a scries of readily interchangeable plates of roughly 2 X 2/3 m
dimensions (see Fig. 24) and both the perforated and solid type can be
combined in any desired pattern. A wall-mounted installation will usually
have its engine centerline off the tunnel centcrline. It is physically possible
to yaw the P W T scavenging scoop system at certain fixed angles toward
cither side wall and then pitch in unison with the test article as required for
exhaust gas removal.
Scavenging scoop interference.—The third major source of interference
encountered in full-scale propulsion tests is that which may be caused by the
interaction of a scavenging scoop with the flow at the rear of the lest article.
A corollary of this problem is the effect of the scoop flow and its proximity
during non-propulsion tests. Only the A K D C - P W T is concerned with these
problems, being the only one equipped with a scavenging system.
T h e installation using a J-34 engine in the P W T transonic circuit (Figs.
107
23 and 24) was also utilized to obtain a better understanding in full-scale of

the effects of scoop tip axial position and mass flow and of pitching the installa-
tion and scoop upon the flow at the rear of the installation. T h e cylindrical
scoop inlet design was obviously selected as the one which would keep
interference to a minimum, in the absence of streamline curvature. T h e
requirement for m i n i m u m streamline curvature at the scoop inlet specifies
that the mass flow per unit area at the inlet is approximately that in the
SVMB01
"o
0
TIP
INTERMEDIATE
l M • oao
a SHORT
I
1
o
n^
O -04 1 , F >
S
f? rt A M ' 1.10
°r— —0
r
• |
—H CD
. }
X BP
Cr
//
^'
42 43 44 45 46 47 4B
STATION, METERS
Fig. 29.
Effect of scavenging scoop tip length on afterbody pressures at unity inlet mass flow ratio
free stream. T h e short tip, which had its inlet 6-3 nozzle exit diameters
downstream of the end of the test article, was chosen as the reference for zero
interference. Axial static pressure distributions over the rear of the cylindrical
portion a n d the aft boat-tailed region of the test article's external surfaces
were used as a measure of the scoop effect.
T h e variation of these afterbody pressures with scoop length at a mass flow
ratio of unity is given in Fig. 29 for two M a c h numbers, 0-80 and 1-10.
Except for a slight deviation with the long tip at M = = 1-10, the results are
identical, indicating a general lack of scoop interference. T h e deviation at
1-10 disappears before M = 1-3, but does indicate that the short tip may be a
little too close for testing just at sonic speed. T h e high degree of sensitivity of
scoop mass flow upon base interference for the two longer scoop tips is
indicated in Fig. 30. Large differences even occur at A f = 1-45 and a flow-
ratio of 0-8. T h e assumption that the short scoop could be considered an
interference-free case is corroborated by the agreement of the data for all
mass flows.
T h e effect of scoop pitch angle with the long tip installed and the test
article held fixed was also determined and is plotted in Fig. 31 for Mach
108
INLET MASS M-OBO
SYMBOL TIP FLOW RATIO
0 LONG 0 5
O INTERMEDIATE
• SHORT OB
X ALL 1.0
CZ oo-
o <L^>- O -
T
^C
h
s 0 8
I
CA
« 08 INLET MASS ••Ul

SYMBOL TIP FLOW RATIO
o Lom O i
O INTERMEDIATE OS
O SHOUT OS
X ALL 1.0
•o
I —t
kf-
iX
< w-0.
-o—0
1
44?. 4 3 4 4 4 5 4 6 4.7 4.8
STATION. METERS
Fig. 30.
Effect of scavenging scoop lip length and inlet massfiow ratio on afterbody pressures
SYMBOL a,:.i.a) y.ccml

M. i BO
0 0 0
17 - 0 53 -83 1
1 <*"
SYMBOL a5ldesi yBlcml

M • 1 10
052
O 0 0
V - 0 BS -17 2
Bt
_J
¥**
J
*^
/
i(
44 45 46 47
STATION, METERS
Fig. 31.
Effect ofscavenging scoop pitch (with long tip) on afterbody pressures at unity inlet massflow ratio
109
R. W. HENS EL AND H. K. MATT
numbers of 0-80 and 1-10. T h e lack of sensitivity of the afterbody pressures

due to small scoop angles of attack is gratifying, particularly since it occurs
with the long tip. In practice, both the installation and the scoop would be
pitched together, but in opposite directions, in order to avoid contamination.
It can be concluded that scoop interference effects can be minimized for
configurations similar to those used in the preseni investigation, even with
the scoop tip within 1-4 exit diameters of the engine nozzle providing the
inlet mass flow ratio is near unity. However, the present ratio of scoop inlet
Fig. 32.
Relative location of model support strut and aerodynamic tip
to nozzle exit diameter, 2-5, is larger than would be used when a larger
engine is employed, which may have some effect upon interference.
T h e P W T ' s experience in the transonic circuit thus far with the use of a
scavenging system for propulsion systems testing has been satisfactory. Some
problems did occur due to the proximity of the scoop to the test section during
non-propulsion activities which are worthy of mention. During this type of
testing, a so-called "aerodynamic t i p " is installed at station 16-75 m in place
of a propulsion scoop tip (see Pig. 23). This aerodynamic tip has a length of
60 cm, which leaves only 90 cm between its leading edge and the end of the
removable lest section. T h e strut used to support sling-mounted aerodynamic
models is mounted in the rear of the test section; the relative location of the
rear of the support strut and sting boom and the lip is shown in Fig. 32.
T h e tip external shape is derived from a N A C A cowling series and the
inlet diameter is 122 cm. This large inlet diameter was chosen in order to
avoid a completely unreasonable area increase to the point of attachment of
110
the tip to the scoop structure, where the flow channel diameter is 187 cm, and
to allow for a planned increase in scavenging capacity. Rapidity of reducing
the tunnel pressure level improves directly with the inlet area. However, the
122 cm dimension is too large to choke the inlet and still maintain a constant
tunnel pressure level because of the present makeup air limitation.
Early in the operation of the P W T transonic circuit, a very low frequency
pulsation of rather large amplitude was experienced in the entire tunnel
duct and particularly in the rear of the test section. It was apparent that the
scavenging duct was acting as a huge organ pipe when a valve located in the
duct a relatively short distance from the tunnel shell was being used for
throttling purposes. T h e ducting length to this valve is in basic agreement
with that required for the measured frequencies, which varied with valve
position. T h e pulsation was mainly alleviated by keeping the valve in a fully-
open position during operation ; throttling is now done at the end of the long,
multi-cornered scavenging duct near the inlet to the Engine Test Facility
exhaust compressors.
At the present time, orifice plates of different diameters are installed in the
scavenging duct near the tip. T h e plates are used to choke the scoop; each
orifice diameter has an applicable tunnel pressure level range for which
choking is possible. T h e plates allow some use of the first valve in the sca-
venging line for throttling purposes. They also act as baffles to inhibit the
scoop pulsation even when unchoked. An interaction problem between the
scoop flow and the wake off the end of the sting boom on the support strut was
also alleviated by the use of choked orifice plates. T h e scoop is also frequently-
pitched in the upward direction to insure misalignment of the model and
sting boom wake and the inlet, particularly when it is not possible to choke
the inlet flow. T h e preceding experiences clearly indicate the need for some
type of remotely-controlled means for varying the scoop inlet area. T h e
P W T is planning to install such a mechanism, which will probably take the
form of a variable area inlet into the scavenging ducting located well down-
stream of the test section. In this case, the aerodynamic tip will be replaced
with a smoothly-shaped plug which would close and seal the front of the
scavenging scoop. T h e pressure recovery of the air inlet choking process will
also be improved in order to increase the efficiency of the scavenging
compression process.
Another non-propulsion test problem attributable to the scavenging
system duct concerns the effect of the latter on the transition of the test
section flow from supersonic to subsonic in the region of the sting support
strut. T h e scoop tip, because of the low inlet velocity ratio and rapid
diameter increase, represents a large blunt body. Therefore, there is a
strong, detached bow wave which extends upstream into the strut region at
low supersonic speeds. T h e bow wave tends to prevent downstream dis-
placement of transition during the process of increasing the tunnel pressure
ratio. Flow transition is frequently quite unsteady and not symmetrical in
the two separate channels formed by the support strut. T h e interaction of
the scoop bow wave tends to increase the unsteady flow regime in the region
of the support strut. It is readily apparent that unsymmetrical, unsteady-
flow constitutes a torsional forcing function for the strut which results in
appreciable lateral vibration at the upstream tip of a model support sting for
111
certain mechanical combinations of model and sting. T h e vibration mani-

fests itself in the dynamic output of the side force gages of an internal balance.
I n a few cases, it was necessary to use stiffer side force balance elements, to
add preloaded lateral cables attached to the sting or to reduce the tunnel
pressure level.
Since most of the problems caused by the scavenging scoop in the P W T
transonic circuit during non-propulsion testing are a result of the close
proximity of the tip lo the test section, it has been decided to modify the
upstream portion of the fixed scoop duct so that a sizeable length can be
removed. Tests in a 1/16-scale model of the transonic circuit have indicated
that this length should be at least 2-5 m in order to alleviate the sling
vibration.
In the P W T supersonic circuit, the scoop will be completely closed for
non-propulsion tests in a m a n n e r similar to the modification planned in the
transonic circuit. A tip shape is presently being developed; it will represent
a compromise between length, since a m i n i m u m weight is desired for handling
ease when switching to propulsion scavenging tips, and sacrifice in tunnel
pressure ratio requirements. T i p bow wave detachment is not expected
because the tunnel m i n i m u m M a c h n u m b e r is 1 -5. This tunnel will have a
long supersonic diffuser, so that flow transition will normally be well down-
stream of the test section. W h e n the scoop is closed, air exchange will be
achieved by means of a separate line attached to a point in the tunnel circuit
upstream of the nozzle and contraction section. Such a connection also has
the advantage of increasing the tunnel m a x i m u m altitude capabilities at the
higher M a c h numbers since the exhaust system inlet pressure would not be
dependent on the pressure recovery in the scavenging duct.
Test and Special Equipment

Model support and balances.—As discussed earlier, the most frequently used
support of test articles in the test section is by struts which also serve to route
supply lines a n d electrical leads to the engine. Since in most tests pitching
of the model is required, a remotely controlled angle of attack drive is
provided; Fig. 33 shows the schematic of a typical strut support of the
N A C A 8 X 6 ft tunnel. This arrangement has the center of rotation at the
tunnel wall, simplifying the wall seal problem.
As in aerodynamic testing, the electric strain gage technique is widely used
for force measurements in propulsion testing. Today all internal balances
and component force elements are strain gage devices. Their characteristics
are so familiar that no details need to be given. However, it might be men-
tioned that high temperature effects by the engine and undesirable force
interactions d u e to internal pressure differentials have to be carefully-
considered in the design of these balances. A typical rocket test installation
where forces on the airframe and on the motor can be measured independently
is shown in Fig. 34.
I n the early days of propulsion testing only external-type balances were
used (see Figs. 1 and 3) due to the lack of such small and simple sensing
elements as strain gages which could easily be inserted into the test article.
These balances were mostly of the "platform" type with suitable linkage to
separate the components. Force measuring was mainly done by beam
112
Hemote-controlled
anjjle-of-o.tta.ck drive
H E X . allowable loads of pl
Horizontal shear, ±2730 kg
Vertical shear, ±2730 kg
MoBtnt, ±2300 m kg Rote: All dimensions In meters
o
>
-
-
-
-
o
-
c
-
—
DB
2.44 m
v.
r
-.
~.
7
-
Pig. 33.
-
Strut mounted model. NACA 8 X 6ft wind tunnel
balances or hydraulic systems. However, in view of the extremely large

weights and forces involved in many propulsion tests, external balances are
still frequently utilized. In Fig. 35 an arrangement with diametrical
support used at the M o d a n e tunnel is shown; the six component balance
with a capacity u p to 20,000 kg in lift and 10,000 kg in drag has pneumatic
capsules as sensing elements. A platform balance, also with pneumatic
capsules, is used for strut-supported models.
Fig. 34.
Schematic of strain gage balance installation for rocket propulsion testing
As another example of a n external balance, Fig. 36 shows a schematic of

the vertical propulsion balance of the P W T . This balance has the center of
rotation in the tunnel centerline and a pitch range of ± 1 0 ° . It measures
thrust or drag force with two Baldwin strain gage load cells whose load range
can be selected. T h e load frame which carries pitching mechanism and
model is attached to the ground frame with flexures to eliminate friction.
BALANCE ELEMENTS
PLATFORM BALANCE
Fig. 35.
Balance installation in O N E R A transonic wind tunnel at Modane
T h e load range of this balance in the axial direction reaches 25,000 kg. In
Fig. 37 a photograph of this balance as installed in a test cart is shown. The
balance is completely enclosed for injection of inerting nitrogen gas since
fuel and oxidizer lines are routed through it; all electrical wiring and the
pitch drive motor are explosion proof.
Scavenging scoop.—Since the scavenging scoop is one unique feature of the
P W T , some of its characteristics shall be shortly described (sec Fig. 23). T o
accommodate test articles with engines of various sizes and numbers, the tip
114
SUPPORT FLEXURE
GROUND FRAME MODEL STRUT

MOUNTING TABLE
-
-
LOAD FRAME -
PITCH LEAD SCREWS SIDE FORCE I
-J
a
w
-
PJ
-
-
-
-
n
—
z
DESIGN LOADS DIAGRAM
LOADS MAY BE APPLIED SIMULTANEOUSLY -
DESIGN LOADS v.
COMSNED LOADS
LIFT ± 13000 Ha
THRUST ,. ± 16000 kg
see FORCE ± 7000 kg
FjrCHNG ± 86000 mkg
± 22000 mkg
MOMENT
YAWNG
-
CART FRAME MOMENT
± ZJOOOmkg -
MODEL STRUT MOUNTING
-
y
TABLE.
CENTER OF ROTATION ON z
TUNNEL CENTERLINE -
-
Fig. 36.
Schematic of vertical propulsion balance AEDC propulsion wind tunnel
W . HENSEL AND H . MATT
section is removable and thus permits tips of different lengths and cross
sections to be installed. T o follow the engine installation angle of attack, the
scoop is pitched proportionally. Since pitch speed and center of rotation of
the test article, axial location of the nozzle and length of scoop tip determine
the pitch rotation, the control system is designed that it can easily be preset
for any such configuration; the scoop then follows automatically any pitch
Fig. 37.
Vertical propulsion balance installed in the AEDC propulsion wind tunnel
change of the engine so that the jet centerlinc intercepts the scoop inlet
centerline to provide full capture of the jet. However, since airflow and jet
spreading effects generally deflect the jet, a second control system corrects
the scoop position accordingly. In this system, suitably arranged thermo-
couples in the scoop inlet sense the temperature of the exhaust jet and correct
the scoop position until the hot jet core is centered in the scoop inlet. Both
automatic control systems can be manually over ridden if required.
With exception of the removable tips, the front section of the scoop shell is
water jacketed. Approximately 12 m behind the inlet, a water spray system
cools the exhaust gases; 320 l.sec of water can be ejected under 7 atm
pressure. T h e water is removed through a barometric well.
Instrumentation and data processing.—In addition to the usual console
instrumentation needed to operate and monitor the engines such as tacho-
meters, pressure and temperature gages, vibration pickups a n d flow meters,
116
a very large number of pressure and temperature instrumentation and strain

gage elements is as a rule required for propulsion testing. Readouts of
numerous rakes and probes, of static orifices and thermocouples serve to
achieve the test objectives. T h e details of such instrumentation systems have
been covered in FREDERICK (1955), LUKASIEWICZ et al. (1956), D ' A N D R E A
(1956) DOYLE (1957) and KAISNER (1958), and can be omitted at this
time; however, a short description of a modern on-line system seems
appropriate. Again the system used at the P W T is discussed as typical.
Basically, it consists of the following elements:
Transducers to commute physical data into electrical signals.
Converters to bring analog signals into digital form (Sadie a n d Millisadic)
Scanners to share amplifier and computer channels with a greater n u m b e r
of transducers and to sequence the computing steps.
Recorders to record or tabulate the data (magnetic tape, punched tape or
cards, tabulators, oscillographs).
Computer (ERA 1102) to process raw data into final form.
X-Y plotters to automatically plot the data.
In Fig. 38 it is shown how these elements fit together to make u p the com-
plete automatic on-line d a t a recording and reduction system. At present,
there a r e 250 pressure transducers for J a t m or 1 a t m pressure differential, a
100 channel system for temperature measurements (Millisadic) with any
type of thermocouple material, and a 20 channel strain gage converter
system available for this automatic system. A complete on-line readout,
computing and tabulating cycle with all these components takes about 30 sec.
In addition to this automatic system, provisions for recording of transient
phenomena were made a n d a large number of sensing and readout instru-
ments such as dynamic pressure pickups, accelerometers, recorders, oscillo-
graphs, electronic counters, etc. are available for special cases.
Optical viewing systems.—In wind tunnels of the size discussed, the control
rooms are always remote from the test sections and provisions have to be made
for visual observation of the test article. In addition to periscopes, used for
instance at the NACA 8 X 6 ft tunnel, closed circuit television systems are
the convenient answer to this request. As a rule, several television cameras
cover the interesting areas and monitors, which can be switched to any
camera, are positioned at desired locations in the control room. Television is
also used to transmit the pictures of schlieren systems which are more or less
standard equipment of high speed wind tunnels. Fig. 39 shows the two
schlieren systems available at the NACA 10 X 10 ft tunnel.
Safety devices.—Various safety measures against fire and explosion which
are stipulated by the fact of the closed circuit operation of the P W T help to
minimize the hazards; some of them are shortly described as typical in the
following paragraphs:
• By electronically comparing the outputs of flow meters located in the feed
line and on the engine any detected difference in fuel weight flow between
these two stations indicates a fuel leak. If the difference exceeds one per cent
a warning signal is actuated and the fuel valves are automatically shut off.
Another method of leak detection uses the fact that all fuel lines inside the
tunnel are shrouded and the shroud lines charged with pressurized nitrogen
u p to 100 a t m and a flow rate from 1-7 to 26 m 3 / h r ; by monitoring the
117
CONTROL ROOM
«^HB£
-
PI
B!
-
>
r
>
H
•H*muj* • M«M.l .KMTKH
H
i£ce«
m
issr Ffc. 3*.
OFFICE BUILDING
Automatic data reduction system A E D C propulsion wind tunnel

nitrogen flow through the shrouds, leaks can be detected and protective
measures initiated.
All fuel line disconnects to the shrouded lines inside the tunnel which feed
the fuel to the engine are located in a water tank equipped with level indi-
cation; if due to a leaky disconnect, the water level increases, an alarm is
sounded.
Fig. 39.
Schlieren systems, NACA 10 X 10ft wind tunnel
Contamination of the tunnel circuit by combustible and explosive mixtures
is continually checked by a Perkin-Elmer detection and analysis system with
sensing elements located in all critical areas of the tunnel circuit and sca-
venging system. In this system, tunnel air samples are drawn into infra-red
analyzer units where the volume content of hydrocarbon molecules in the air
is measured by an optical-electronics system and an electrical signal produced
indicating its concentration in per cent of the lower explosive limit (LEL) of
the fuel used. If a preset lower limit (15 per cent of L E L ) is reached, a
warning is given. When the concentration exceeds a preselected higher
limit (40 per cent of L E L ) , fuel flow to the engine is automatically cut off and
the tunnel is shut down. At the same time, nitrogen is injected at a rate of
180 kg/sec into the balance housing, engine nacelle, scavenging duct and
plenum area to inert any combustible mixtures. T h e nitrogen system has a
capacity of 35 tons of gas at 160 atm and is large enough to allow several
shots before recharging. In addition to the survey of the whole tunnel
circuit for combustible mixtures, a second system is being used to check the
oxygen content of the air on specific test configurations to determine air-
stream contamination (sec discussion of results under Exhaust Gas Removal).
Air samples are taken at selected locations and pumped through an oxygen
analyzer (A. O. Beckman Inc.) that uses the paramagnetic characteristics of
oxygen to give a direct electrical readout of the oxygen percentage in the air;
two sensitivity ranges are available, one for 0 to 21 per cent and one for 16 to
21 per cent oxygen content.
119
A six ton carbon dioxide system under 20 atm pressure serves to extinguish
fires. Outlets of this system are in valve boxes, engine nacelle, instrument
disconnect panels, i.e. in areas where extensive damage to instrumentation
or test article could be done by fires or new fires started. An Edison Aircraft
Fire Detection System and thermal switches are utilized as sensing elements.
Fuel systems.—Several permanently installed fuel systems are available at
the P W T as well as at the N A C A tunnels; for special cases, mobile units
(tank trailers, bottle batteries) can also be used. As a n example, the P W T
liquid fuel supply systems may be described.
T w o independent fuel circuits for air-breathing engines (Fig. 40) are ready
to deliver metered and filtered (0-01 m m particles) fuel at any selective
pressure between 2 atm and 4-5 a t m and at a rate of u p to 10 kg/sec. An
electronic-hydraulic pressure control system holds the pressure constant
within ^ 2 per cent. T h e first circuit is equipped with dchydrator and heat
exchangers to maintain a constant fuel temperature (32°C i 1°). T h e
second circuit is provided with a proportioning unit to premix fuel combina-
tions such as water and alcohol. All this equipment is housed in a special
fuel building with fire protection systems similar to the ones described
previously. T h e metering instrumentation in these fuel systems has some
features interesting enough to be mentioned here. By means of a calibrated,
float-actuated potentiometer in the electronic measuring system, the turbine-
type flow meters (Potter Corp.) are compensated for different specific gravities
and therefore give correct direct readout of weight flow for any fuel with a
specific gravity between 0-68 and 1-0. Furthermore, in order to maintain an
accuracy of + 0 - 5 per cent for all flow rates, the fuel is automatically switched
to meters sized for best accuracy within the specific flow rate range; the
ranges for the various meters installed are 0-4-1-6-16 l.sec.
Propellant supply for liquid rocket testing is furnished by still another
system as shown in Fig. 41 ; this layout follows closely the system originally
used at the N A C A 8 X 6 ft wind tunnel. Fuel and oxidizer appear in two
separate circuits. Both propellants are fed to the engine by pressurized
helium or nitrogen (100 atm) which are also used to purge the lines. Tanks
(0-51 m 3 each), piping and valving are m a d e of stainless steel to prevent
corrosion. T h e principal supply lines are 50 mm in diameter while a 25 m m
diameter line which is connected to one tank may be used for small rockets
or to supply auxiliary propellants for pyrophoric ignition. An intricate
system of valves, flow meters and pressure regulators serves to control the
propellant flow. Vents, pressure relief valves and rupture disks are installed in
suitable line sections to minimize the safety hazards inherent in rocket firings.
T o give an idea of the capacity of this propellant system, it might be
mentioned that for a rocket with 2500 kg thrust and average performance
characteristics (spec, impulse 240 sec, oxidizer to fuel ratio 2-3) the total
burning time for one filling of the pressurized tanks is in the order of 4 min.
I n between firings, these tanks can be refilled from a liquid oxygen storage
tank whose capacity is sufficient for five fillings.
T h e 8 X 6 ft tunnel rocket fuel system has been used successfully in tests
involving scaled rocket motors in the vicinity of 500 kg of thrust. T h e P W T
system is nearing completion at this writing and is scheduled for initial use in
November, 1958.
120
DCyMCHAL'ZED
WATCH P L * NT
I
r
9 9 HZ. -
r.
i J -
-
-
-
z
-
-
-
._X~IL,
.Gr | SSj z
-
y
fUCL
HZh^ i
i i
y
-
~
i i Z
z
-
r
4flg. 40.
Test fuel system AEDC propulsion wind tunnel
HIGH VENT FROM NITROGEN
FROM MOBILE
NITROGEN SUPPLY A r - , * STORAGE TANKS
VENT
*XKE1
ENGNE
MOOEl
[}>#
SB
p.
y
CO
-
r
>
y
e
TO DRAIN
HIGH VENT
>
GASEOUS OXTTJEN
-;
CODING
^w-t CHECK VALVE FI FLOW RATE TRANSMITTER
fjwy HAND MU.VE LT LEVEL TRANSMITTER
iJwy PRESSURE RELIEF \ftLVE P. PRESSURE TRANSMITTER
PH PRESSURE REGULATOR
S H REMOTE CONTROLLED \ftLVE
P PRESSURE INDICATOR
453 FILTER
=£4= RUPTURE DISC m REMOTE SETTING
TC T»iCR CONTROLLED
Fig. 4 1 .
Rocket propellant supply system A E D C propulsion wind tunnel
FULL-SCALE PROPULSION TESTING IN WIND TUNNELS
Handling equipment for lest articles.—Full size propulsion units possess

remarkable weights and large dimensions. Handling of these test articles
therefore presents a problem which has to be considered. This is particularly
important for the installation into the test sections, since excessive time delays
due to handling difficulties mean a loss of productive tunnel operation time.
Careful attention has been paid to simplify and accelerate the installation
methods.
In the NACA 10 X 10 ft wind tunnel, the whole test section bottom can be
lowered to the next floor level by means of motor driven jacks (see Fig. 42),
Terminal pe.no! for
suspended modele~ Motor and drive
for lowering
tunnel floor
lat Mezzanine floor

al. 236
Mote: Work platforms not

shown are 2.SO below
tunnel floor.
Tunnel t e s t - s e c t i o n floor which

B»y be lowered to f i r s t floor
f — l a t Floor
/ e l . 331
r'. •.../J . M r a . *••'•'• I
All dlnenalona in metere

Fig. 42.
NACA 10 X 10ft wind tunnel lest section
permitting test articles to be rolled on with the help of special model trans-
port dollies. In the 8 X 6 ft tunnel, a large access hatch in the diffuscr floor
and use of cradle dollies accommodate model installation.
In both the Modane and the A E D C wind tunnels, which are sized to test
very large engines, a different approach has been taken. Here the whole test
section carrying the complete test installation is put on tracks and can be
exchanged with the help of motor driven transfer cars and self propulsion.
T h e weights which have to be moved are enormous, about 500 tons in the
case of Modane and 320 tons for the P W T . However, the tunnel time saved
123
K. VV. H F . N S E I . AND H. K. MATT
in the preparation and buildup of test articles in one test cart while another
cart is being tested is rewarding. In some cases, it is possible to make a
transfer without sacrificing one single operating shift.
Test Techniques
In conducting burning engine tests in wind tunnels, both tunnel and engine
operations have to be carefully programmed and matched. It is, for instance,
customary to start the tunnel and establish the Mach number at low air
densities to reduce starling loads on model and on drive motors. During this
startup cycle, the engine is generally wind-milling. If the time involved is too
long the engine fuel p u m p , which is normally cooled by the fuel, might r u n
hot and fail. U n d e r such conditions the engine will have to be started as soon
as the starting speed is reached and then left idling until tunnel conditions are
set, in order to maintain some cooling fuel flow through the p u m p . O n the
other hand, such a procedure would be Impossible for instance in the N A C A
10 X 10 ft tunnel which has to be started u p as a closed circuit tunnel
(aerodynamic cycle) and then switched over to the open, non-return cycle
after the supersonic flow is established in the test section. Since this starting
period requires more than one hour and the engine cannot be started before,
an engine type as assumed above could not be run without precautionary
measures.
O n e of the m o r e annoying problems of high altitude engine testing is the
startup of an engine with regular fuel at low air densities after an involuntary
or planned flame out. I n order to re-start, either the tunnel pressure has to be
increased to a practical level, a time consuming and expensive method, or the
normal ignition cycle substituted by some other scheme. Recently, the use of
very reactive chemical fuels such as ethylene oxide for re-start was successfully-
adapted. Unfortunately most of these fuels are very difficult to handle. As a
rule they are highly toxic, possess low boiling points (less than 10°C for
ethylene oxide), have extremely low flash points (less than —30°C I'm
ethylene oxide), and are very volatile, flammable and explosive. These
characteristics make wide and inconvenient precautionary measures
necessary. T h e fuels must be kept u n d e r refrigeration and pressure, all
equipment must be grounded and any source of unintentional ignition must
be avoided. Explosion-proof electrical equipment and use of non-sparking
tools are mandatory as well as some restrictions as to the use of certain metals
which could chemically react with the fuel to start a n explosion.
T h e conditions connected with the use of chemicals for engine starts are
described in some detail because they are equally true for rocket engine
firings in wind tunnels. Due to the limited life of a rocket motor and the
enormous fuel consumption, testing is done only in short bursts (1 to 2 min)
long enough to establish steady state conditions and to take the data.
Although in outside test stands, ignition of liquid and solid rockets is normally
accomplished by an electrical spark used directly or to ignite a squib and
after each firing these parts can easily be replaced, in a wind tunnel this
method is not feasible since it would d e m a n d pressurization and shutdown
of the tunnel after each firing. A device used at the N A C A 8 X 6 ft tunnel
consists of an electrical ignitor which extends from the tunnel wall to the
rocket nozzle and is automatically retracted again as soon as a propane pilot
124
flame is ignited and the temperature on the ignitor reaches 200°C; destruc-
tive impingement of the very hot jet on the ignitor is so prevented. Looking
for an easier method in the much larger 16 X 16 ft P W T , pyrophoric
chemical fuels drew attention again. Fig. 43 shows the schematic of a
planned rocket start system using triethyl-aluminum (TEA).
T h e hazards involved in dealing with liquid oxygen contributed to the
fact that rocket firings are generally controlled by automatic timers. All
steps including purging of lines, ignition of pilot flame, setting of starting flow
rates of fuel and liquid oxygen and finally of full propellant flow are part of
the automatic starting cycle after the "Start Rocket" button is pressed.
Similar is the "Stop Rocket" cycle. T h e operator has on the other hand the
possibility to " h o l d " some phases of the sequence or to stop the engine
operation in an emergency. Circuitry interlocks to stop the engine are also
provided for the case of failures, for instance when the engine does not fire
immediately, when the combustion chamber burns through, when loss of
cooling water occurs, etc.
F U T U R E TRENDS OF P R O P U L S I O N TESTING IN WIND TUNNELS

T h e A E D C - P W T is anticipating test requirements for full-scale installa-
tions of air-breathing propulsive devices u p to a Mach n u m b e r of nearly 5.
This tunnel is capable of attaining such a Mach n u m b e r in its 4-9 X 4-9 m
test section, but is presently limited in altitude temperature simulation
capability beyond Mach 3 (see Fig. 21). Studies are under way to investigate
means of increasing the supersonic circuit temperature level to at least that
required for M = 4. T h e core heating technique (ADAMS, 1958) appears
promising for this purpose since only the portion of air stream to which the
installation will be subjected is heated and thus some facility temperature
problems are alleviated.
Although the requirement for testing of full-scale air-breathing installations
is expected to remain high, a basic trend toward altitude testing of rocket
installations, both with and without external airflow, is now quite clear.
Preceding sections have touched upon the preparations which have been and
are being made at both the NACA 8 x 6 ft tunnel and the A E D C - P W T
transonic circuit for testing liquid propellant rocket systems. Solid propcllant
rockets require less facility equipment, but do present a potential hazard in
that a sizeable quantity of the propellant may be within an enclosed test area.
This type of rocket will also undoubtedly be tested as well, particularly in the
P W T when associated with a specific missile.
As before, the main contribution of a wind tunnel in the field of rocketry
is in the type of test which concerns the interaction of the propulsive unit and
its installation at simulated flight conditions. Both the power-off aero-
dynamic forces on the airframe and the rocket motor characteristics can be
obtained separately with reasonable accuracy. A burning test of the complete
installation in the wind tunnel establishes the interaction effects of the
exhaust jet or jets and the airflow. Of particular interest is the effect of the
jet on the flow in the vicinity of the base and on base pressure. Such items as
nozzle exit hinge moments, the effectiveness of jet vanes and the effects of
control or vernier jets can also be expected to be the subject of investigation.
T h e basic limitation on the size of rocket motors which may be tested at
125
CHECK VENT
VALVE
PURGE VENT
TO SUMP
17 a i m
I7alm
TO CONTROL f^Tf.
ROOM K V (PS;
-
' 0 - 1 8 olm
Ml TEA SHROUD =
-
y
vt
-
TO ENG
CHECK VALVES
-
rO >
CT) Z
KmJ KJ
3
7 0 aim
DRAIN
TO SUMP
>
-i
-
CODING
- U ^ r - CHECK VALVE PR PRESSURE REGULATOR

- c J o - HAND VALVE PT PRESSURE TRANSMITTER
- t £ o - PRESSURE RELIEF VALVE PS PRESSURE SWITCH
X REMOTE CONTROLLED VALVE
Fig. 4 3 .
Triethyl-aluminwn ( T E A ) engine start system
altitude in a wind tunnel is the spreading of the jet due to under-expansion

through the nozzle. In the P W T , the first problem is that of completely
scavenging the products of combustion. T h e rocket size limitations induced by
this requirement have been estimated (TEMPELMEYER etal., 1958). A second
problem is presented by the effective blockage the expanded jet, which could
even result in choking the tunnel under extreme conditions. Fortunately, a
rocket jet can be scaled fairly readily by the use of smaller motors having
the same nozzle shape and combustion chamber pressure characteristics
as the full-scale motor. This will undoubtedly be done for most of
the larger missiles. It also follows that the exact propellants to be used in
the flight article need not necessarily be used in the wind tunnel should they
be too corrosive or otherwise difficult to handle.
From the facility standpoint, it is desirable to have as wide a pressure
altitude Mach number spectrum as possible for rocket installation testing.
Altitude temperature matching may be considered of secondary importance.
Again, this points to the variable density type of tunnel. T h e P W T super-
sonic circuit maximum altitude performance is conservatively presented in
Fig. 2 1 . A recent improvement in exhaust system pressure ratio and diligent
attention to tunnel ducting leakage will result in a considerable rise in
maximum altitude, subject finally to the main tunnel compressor performance
deterioration at low pressure. Further improvement in both m a x i m u m
altitude and Mach number at the expense of test section size is possible by
using the tunnel main stream as an ejector and inserting one or more
secondary nozzles into the test section. A hot core heater would be very
useful to avoid condensation of the constituents of air at the higher Mach
numbers. However, present indications are that it may not be necessary to
simulate the exact flight M a c h number at high altitude. If this proves to be
so, the variable density supersonic propulsion wind tunnel will be a very
valuable tool for a wide range of flight conditions.
SOMMAIRE
Vauteur decrit le role des souffleries dans les essais de systemes de propulsion.
II etudie seulemenl les installations a ecoulement conlinu, transsoniques et super-
soniques, et pouvant contenir des moteurs a combustion en vraie grandeur: la
Soufflerie Transsonique francaise S1 MA de Modane, les grandes souffleries du
NACA au Lewis Flight Propulsion Laboratory de Cleveland, Ohio, et la Soufflerie
pnur essais de propulsinn de I' Arnold Engineering Development Center a Tullahnma,
Tennessee. U auteur decrit les performances de simulation d'altitude deces souffleries
par rapport aux nombre de Mach, pressions el temperatures. II etudie la technique,
les problemes et 1'experience acquise dans le domaine de ['evacuation des gaz
d'echappement par un systeme approprie dans une soufflerie a densite variable, el
donne quelques resultals detailles touchant les effets de contamination el d'interaction
causes par le dispositif de vidange et se faisant senlir sur la repartition des pressions
dans la region arriere du moteur. L'auteur donne enfin un apergu de I'infiuence du
systeme d'e'ehange d'air, en particulier du volume d'appoint d'air sec, sur la taille
maximum du moteur.
REFERENCES
ADAMS, RICHARD H. "A High Temperature Stream Tube for a Supersonic Wind
Tunnel" M I T Naval Supersonic Laboratory Tech. Rep. 303, May 1958.
127
R. W. HENSEL A N D H . K. MATT
BKKANEK, L E O L., L A H A T E , SAMUEL a n d INGARD, U N O . "Acoustical T r e a t m e n t for

the NAGA 8 x 6 Foot Supersonic Propulsion W i n d T u n n e l " NACA T N 3378,
J u n e 1955.
BUSSEY, LYNN E. "20-Foot W i n d T u n n e l Tests of Five Exhaust Stack Configurations
for the C-74 A i r p l a n e " A F T R 5237 (A77-21633), J u n e 1945.
I)'ANDREA, J . B. " A u t o m a t i c Indicating a n d Recording Data Systems of W i n d
T u n n e l s " AGARD Rep. 16, February 1956.
D E F R A N C E , SMITH J . "Results of Drag Tests on P3M-1 Nacelle" NACA ACR, April
1932.
D E F R A N C E , SMITH J . " T h e N A C A Full-Scale W i n d T u n n e l " NACA TR 459, 1933.
DELANO, JAMES B. "Full-Scale Propulsion Testing in the 16-Foot Transonic Circuit,
A E D C " 44£/X:-7W-58-31 (ASTIA /LO-157141), J u n e 1958.
DOYLE, J O S E P H P., J R . " I n s t r u m e n t a t i o n a n d Data H a n d l i n g Methods of Deter-
mining Load Distribution on W i n d T u n n e l Models by Pressure Distribution
M e a s u r e m e n t s " AGARD Rep. 114, M a y 1957.
FREDERICK, C. L. " A u t o m a t i c Pressure Measuring Systems Used with H i g h Speed
W i n d T u n n e l s " AGARDograph 10, J u n e 1955.
G O E T H E R T , B. H . "Shock W a v e Reflections in W i n d T u n n e l s " Proceedings of the
I X International Congress of Applied Mechanics, September 1956.
G O E T T , H A R R Y J . a n d PASS, H . R. "Effect of Propeller O p e r a t i o n on the Pitching
Moments of Single-Engine M o n o p l a n e s " NACA ACR (Wartime Rep. L-761), M a y
1941.
GROSE, R . M . "R-1830-90-B Experimental Nacelle T e s t s " VAC Research Dept. Rep.
R-4000-1, October 1946.
H A B E L , LOUIS W . a n d KORYCINSKI, P E T E R F . " A Comparison of T h r e e Spinner-
Diffuser Designs in a n N A C A D S Cowling for the Pratt & Whitney R-2800 E n g i n e "
NACA M R (Wartime Rep. L-579), J u n e 1944.
HARTSHORN, A. S. a n d NICHOLSON, L. F. The Aerodynamics of the Cooling of Aircraft
Reciprocating Engines R & M 2498 (A.R.C. M o n o g r a p h ) , 1947.
H O O V E R , R O B E R T M . "Acoustical Evaluation of the Acou-Stack T r e a t m e n t in the
N A C A 10 ft by 10 ft W i n d T u n n e l " WADC T N 57-393 (ASTIA A D 142164),
M a r c h 1958.
J E N N I N G S , W . G., T E R R Y , A. a n d PEARSALL, P . J . " P r e l i m i n a r y C a l i b r a t i o n of the
24-Foot W i n d 'Tunnel, R.A.E., with a Short Description of the T u n n e l " R A E
R & M 1720, J u l y 1936.
KAISNER, R . W . "Propulsion W i n d T u n n e l Digital Pressure System" AGARD Rep.
168, M a r c h 1958.
K E L L Y , M A R K W . a n d T O L H U R S T , WILLIAM H . , J R . "Full-Scale W i n d 'Tunnel Tests
of a 35° Sweptback W i n g Airplane with High-Velocity Blowing over the Trailing-
Edge F l a p s " 4V.4C44 R M A55109, November 1955.
L E M A R E C , G. " L a G r a n d e Soufflerie Aerodynamiquc de C h a l a i s - M e u d o n " Le
Genie Civil, p p . 401-406, November 1934.
LUKASIEWICZ, J . , V A N DER BLIEK, J . A. a n d SCOTT, J . G. " H i g h Speed Systems of
Wind T u n n e l Data H a n d l i n g " AGARD Rep. 17, February 1956.
MARCUS, M . J . "Full-Scale Powered Nacelle Tests of the McDonnell XP-67 Airplane
in the Wright Field 20-Foot W i n d T u n n e l " A F T R 5126 (44 77-9524), J u l y 1944.
MARCUS, M . J . "Full-Scale Powered Nacelle Tests of the Consolidated Vullee
Aircraft Corporation's XB-36 Airplane in the Wright Field 20-Foot W : ind T u n n e l "
A F T R 5430 (.4 77-9605), M a r c h 1946.
MASSUGER, M . LOUIS. Aircraft Production, April 1941. (Article based on translation
of a paper prepared by MASSUGER.)
M C M A H A N , D A V I D G. " R e p o r t on High Speed Tests of the R T V - N - 1 5 Test Capsule
in the U A C 8-Foot W i n d T u n n e l " If AC Research Depl. Rep. R - l 4055-3, December
1948.
128
M O N R O E , LOUIS L. a n d SAARI, M A R T I N J . " A n Investigation of Cowl Inlet for the

B-29 Power-Plant Installation" 4V44C44 M R E5K30 (Wartime Rep. E-205),
J a n u a r y 1946.
NACA. Manual for Users of the Unitary Plan Wind Tunnel Facilities of the National
Advisory Committee for Aeronautics NACA (Washington), 1956.
P I E R R E , M . , FASSO, G . a n d C H R I S T O P H E , J . " L a Soufllerie Sonique d e 1'ONERA a
M o d a n e " Atotnes, J u l y 1951.
PINKEL, BENJAMIN. " H e a t Transfer Processes in Air-Cooled Engine Cylinders"
NACA 77*612, 1938.
PINKEL, BENJAMIN a n d ELLERBROCK, HERMAN H . , J R . " C o r r e l a t i o n of Cooling Data
from a n Air-Cooled Cylinder a n d Several Multicylinder Engines" NACA TR 683,
1940.
PRESTON, G . M E R R I T T , BLACK, F R E D O . , J R . a n d J AGGER, JAMES M . "Altitudc-Wind-
T u n n c l Tests of Power-Plant Installation in Jet-Propelled Fighter" NACA M R
E5L17 (Wartime Rep. E-274), February 1946.
R O G A L L O , V E R N O N L., R O B E R T S , J O H N C , a n d O I . D A K E R , M E R R I T T R . "Vibratory
Stresses in Propellers O p e r a t i n g in the Flow Field of a Wing-Nacelle-Fusclagc
C o m b i n a t i o n " NACA T N 2308, M a r c h 1951.
R O Y , MAURICE. "Caracteristiques et Possibilites d u Centre d'Essais d e T O N E R A a
Modane-Avrieux ( F r a n c e ) " ONERA Publication No. 8 5 , 1956.
S C H E Y , OSCAR W . , PINKEL, BENJAMIN a n d ELLERBROCK, H E R M A N H . , J R . "Correction
of Temperatures of Air-Cooled Engine Cylinders for Variation in Engine a n d
Cooling Conditions" 4V44C44 TR 645, 1938.
SWKBERG, H A R O L D H . a n d DINGELDEIN, R I C H A R D C. " S u m m a r y of M e a s u r e m e n t s in
Langley Full-Scale T u n n e l of M a x i m u m Lift Coefficients a n d Stalling Charac-
teristics of Airplanes" NACA TR 829, 1945.
T A T R O , ROBERT E. " T h e Spreading Characteristics of Choked Jets Exhausting into a
Supersonic S t r e a m " AEDC-TR-55-2, October 1955.
T E M P E L M E Y E R , K . E., W I N D M U E L L E R . A. K... ESTABROOKS, B. B. a n d PINDZOLA, M .
"Rocket Engine Testing Capabilities of the Propulsion W i n d T u n n e l Facility"
44£7)C-7W-58-17 (ASTIA AD-\b2Q$b], April 1958.
Test Facilities Handbook. "Propulsion Wind Tunnel Facility", Vol. I l l Arnold Engi-
neering Development Center, Tullahoma, Tennessee, January 1958.
WEICK, FRED E. " D r a g a n d Cooling with Various Forms of Cowling for a Whirlwind
Radial Air-Cooled E n g i n e — I " 4V4C44 TR 313, 1929a.
W E I C K , F R E D E. " D r a g a n d Cooling with Various Forms of Cowling for a Whirlwind
Radial Air-Coolcd E n g i n e - I I " .V.4C4 TR 314, 1929b.
W E I C K , FRED E. a n d W O O D , D O N A L D H . " T h e T w e n t y - F o o t Propeller Research
Tunnel of the National Advisory Committee for Aeronautics" NACA TR 300, 1928.
DISCUSSION
E. W. WASIELEWSKI (Curtiss-W'right Corporation):
The disturbance screen method has been very valuable in early demonstrations of the
serious effects of inlet distortions on turbojet engine performance and durability. In view of
the results obtained with ducted nozzle distortion testing, however, turbojet inlet problems
would appear to be simulated mosl effectively with combinations of vanes varied in accor-
dance with the results of inlet tests carried out at the required values of Mach and Reynolds
numbers. Because of the great variety of inlet configurations which are possible, an early
determination ol the inlel geometry is essential so that proper adjustments can be made in the
engine design. General information obtained from the study suggested by Goethert would be
a valuable aid to the engine designer.
Much testing employing the ducted nozzle distortion technique has been performed at the
Wright Aeronautical Division. As the result of this experience, our conclusions vary somewhat
from those indicated by Goethert. The difference between the ducted nozzle calibration data
and the results obtained with the engine operating at the proper fuel-air ratio have been
129
R. W. HENS EL AND H. K. MATT
negligible in our tests and smaller, in any case, than those obtained by ARO. Il should In-
pointed mil in ihis regard that the distortion pattern is also affected by burner geometry (as
well as inlet conditions and fuel-air ratio); we have been especially careful about using the
same burner geometry in both our calibration and burning tests.
Typical of our results is the profile comparison between wind-tunnel and ducted nozzle
data shown in Fig. 44. I believe that the simulation factor referred to there is the same as the
WIND TUNNEL DUCTED NOZZLE
SPECIFICATION ACTUAL
PROFILE TEST PROFILE
SIMULATION
FACTOR . 0 8 2
VANE ANCLE
SETTINGS FROM
54* TO 170*
MACH 2.75
<>6.2*
SIMULATION
FACTOR , 0 6 8 4
VANE ANGLE
SETTINGS FROM
4 4 ' TO 162°
MACH 3.05
«»6.2*
Fig. 44.
Profile comparison mind tunnel vs. ducted nozzle data
"average error" mentioned by Goethert, It will be observed that the profiles are generally
of the same shape for both types of testing. Visual inspection of many test results indicates
that deviation from the specified Mach number rarely exceeds 10 per cent. Close simulation
was found to be especially important near the boundaries, where small differences can
appreciably affect the durability of a cooled engine. Although the position of the terminal
shock is significant, the proper setting of the vanes is essential to a good profile. The large-
range of angular positions required is indicated in Fig. 44. Even small deviations from these
sellings produced significant changes in the profile. While a partial simulation can be
obtained by merely positioning the shock (with a specified Mach number at the minimum
section), it is not at all adequate to meet the accuracy required.
Goethert has shown the benefits which can be realized in the form of low pressure ratio by
attention to the detail design of free-jet installations. Further demonstration of the low
pressure ratios obtainable was given by a series of model tests.* The arrangement used in tin-
tests is shown in Fig. 45. Although no boundary-layer control was employed, there arc
diffusers at both the nozzle exit and the engine exit. The best results obtained (after studying
several configurations) arc shown in Fig. 46. During the tests engine burning was simulated by-
injecting air into the combustion chamber from an external source. This procedure makes it
possible to produce conditions in the exhaust nuzzle close lo lliiisc encountered in the actual
engine. An important element in obtaining low pressure ratios is the engine exit diffuser. As
shown in Fig, 47, the convergent-divergent diffuser (as compared with the cylindrical diffuser)
not only reduces the tunnel pressure ratio, but also is less sensitive to engine inlet performance.
These tests offer an improvement over the results in Fig. 46, which were obtained with a
cylindrical diffusi-r.
A thorough job has certainly been done at Tullahoma in the investigation of facility
characteristics to make possible the study of turbojet engines. The application of these
mathematical techniques to the measurement of rocket thrust transients and the ingenious use
of ejectors to provide a constant altitude pressure during a rocket test are good examples of
what can be accomplished with facilities of this type. While much progress in the rocket field
• These tests were carried out for Curtiss-Wright by the KluiDyne Engineering Corporation.
130
ENGINE MODEL W I T H
SIMULATED SUPPORT
NOZZLE EXIT
DIFFUSER ( N E D )
ENGINE MODEL
EXHAUST NOZZLE
AND DIFFUSE*
TEST C E L L
O S S INLET MODEL
- FACILITY MODEL
SUPERSONIC NOZZLE
Kg. 45.
General facility model layout
RAM PRESSURE
* EXHAUST PRESS'JUE
NOZZLE STARTING -
SIMULATED ENGINE
NOT BURNING
L
•-NOZZLE RUNNING
SIMULATED ENGINE
BURNING
24 Ze it JO SZ 3< 36 38 «0
TEST SECTION MACH NUMBER
Fig. 46.
Required supersonic nozzle pressure ratios
80
75
I
o
1-
S 7.0
TUNNEL REMAINS STARTED AND ENGINE

JEXHAUST IS FULLY EXPANDED
I f I
40 42 44 46 ,48 50 52 54 56 58 60
1[ - INLET RECOVERY
Fig. 47.
The effect of an engine nozzle exhaust diffuser on exhaust altitude simulation
131
R. W . HENSEL AND H . K. MATT
has been made withoul die use of altitude facilities, Goethert'" work has shown that il is
possible to carry out significant programs directed towards the improvement of this important
engine type with facilities originally designed for die development of air-breathing power
plants. We have been giving close attention to this problem at the Wright .\eronautical
Division. Since the Wood-Ridge Altitude Facility employed Steam ejectors in the original
design, its equipment is especially suitable for extension into the field of rocket testing along
the lines suggested by Goethert.
V. W. GREENHOUGH (Bristol Engines Ltd.):
The techniques oi connected testing has been deall with in some detail by Goethert. I would
endorse all that has been saiil about the advantages of this type of testing in reducing the size
of test plant to be used and consequently the cost of testing; provided that it is fully appre-
ciated that perfect simulation is never obtained and that the major value of this technique is lo
establish quickly and cheaply the main trends followed by more representative tree-jet
testing to confirm the results.
Willi free-jet testing an attempt is made to establish accurate aerodynamics at the intake.
I would like lo ask Dr. Goethcrt if he has ever met cases where the adequacy of the method
could be questioned because of differences between test-bed and flight results. What I have
in mind is the difference due to turbulence in air from a blown nozzle which is obviously not
the same as that in a free atmosphere. This could consequently affect the airflow through the
engine. Although we have at Bristol sometimes suspected that such effects exist, we have not
succeeded in proving or disproving that the different turbulence mndiiiiuis win- in fact
responsible lor different test-bed and flight results. Had Dr. Goethert any evidence or
suspicions that test results could be invalidated in this way?
Our experience of free-jet testing is very similar to that at AEDC, having the same diffi-
culties with test-bed installations where the spill factor is in the order of 50 per cent, particu-
larly when the engine is operating near critical. At this point wc have found in the worst
cases that the apparent combustion rich-extinction limits were in fact a function of the engine
intake-nozzle combination only and not a combustion result.
The pressure recovery figures quoted for spill diffuscrs arc of fundamental importance in
the design of the test plant and speaking from memory they are of the same order as our
Bristol results. The amount of boundary layer bleed required (5 per cent) appears to be
higher than might be expected.
Other advantages of the boundary layer bleed which I think should be stressed are:
firstly, the engine intake may be positioned further from the nozzle and a wider view obtained
<il the shock configuration upstream of the intake lip and, secondly, incidence testing is much
easier. Both these advantages arise from the absence of shock reflections at the jet boundaries
when tIn- bleed air is exhausted at the nozzle static pressure.
The section of the paper on plant performance for transient testing is particularly interesting
both from the point of view of establishing engine performance and also in making testing
quicker and therefore cheaper. Apart from (he purely technical problems in designing and
constructing this type of equipment, reliability and safety considerations must be rather a
worry.
We havi not been successful with attempts to improve instrumentation accuracy by correct-
ing for lime lags. In cases where we have tried to do this non-linearities in some of the instru-
mentation equipment has made the problem so difficult mathematically, or in other cases
completely impossible, that attempts have been abandoned and faster instrumentation
substituted or the test rate slowed down to suit.
Being associated with the steam ejector driven high-altitude test plant at Bristol, the section
in the paper on devices to convert engine exhaust gas kinetic energy back to pressure is of
direct practical interest to me and obviously to all others associated with high altitude
testing. Would the author say whether direct measurement of engine thrust was possible
with these devices in use or did difficulties in establishing the static pressure distribution
adversely affect test results?
J. SURUOUE (ONERA):
Les dispositifs d'extraction par ejecleiir a llus moteur supersonique d'air ou de vapeur
il'iau ou d'un melange dc ces deux fluides sont particulierement appropries aux essais en
altitude simulex- de foyers et d'engins complets et peuvent meme presenter sur plusieurs points
une superiority sur les dispositifs comprenant des machines tournantes.
132
lis constituent des installations simples et cxonomiqucs particulieremcnt intercssantes par

leur faible prix de premier etablissemenl et d'enlretien.
L'evacuation des gaz bruits chauds ne pose aurun problemc puisqu'ils sont tres fortemenl
dilues par Ic fluidc inducteur. Au contraire, avant d'etre aspires par des comprcsscurs, ils
doivent etre pr£alablement refroidis et evenluellemenl 6pur6s chimiquement s'ils sont
agressifs commc dans le cas de lulilisation de combustibles speciaux ou de combinaisons
genre statofusee.
La rusticite de ce genre d'iiistallations leur confere une extreme facility d'emploi en meme
temps qu'une grande sexurin*.
Enfin, ils presentent aux plus faibles pressions d'exlrartion une capacile trcs eievee ( L E
iiKi\kset al, 1956).
Leurs caracteristiques debit masse entraind—pression residuelle restcnt en effet sensiblemenl
iineaires.
£changeur
0-51<g/cr
Fig. 48.
Au contraire. ime pompe rotative, necessairement a capacite volumetrique constante, ne

peut cxtraire a faible pri-ssimi i|u'un debit-masse restreint, La simulalion des altitudes
elevdes au moyen de t-iimpressenrs roiaiifs exige done la multiplication des etages de com-
pression, ainsi que I'ont fait ressortir plusieurs auteurs (BVKOWSKI, 1957 ct BRUCE and
SCHUBERT, 1956). (Via conduii a des installations importantes d'etablisscment et d'enlretien
routeux.
A cette prupriete est lice la possibilite inleressante d'une variation Ires rapide du niveau dc
pression realise, c'est-a-dire di- la simulation d'une ascension a tres grande vitcsse, pcut-etrc
plus facilemeni qu'avec une association ejecteurs-compresseurs.
A titre d'illustratimi. et pour mnntrer les possibilites d'ejecteurs seuls. je donncrai quclques
indications sur une installation de capacity moyenne realisee a TONER A
Elle est alimentee en air sous ime pression de 7 aim par une station desservanl un ensemble
d'installalions d'essais.
Ellc permet d'etudier la man be en veine forcee des foyers de staloreacteurs de diametre
250 mm a une pression un peu infeiicure a 0-5 atm.
I.e schema general en est indique sur la Fig. 48.
Une partie du debit est derivee vers Ic foyer cn essai par I'inlermediaire d'un echangeur,
donl le circuit primaire absorbe lui-meme unc fraction du debit global, et a la sortie duquel le
llux seenndaire se trouve porle m 300 C environ. Cetle temperature correspond aux condi-
tions d'entree du foyer resultant d'une captation a un nombre de Mach voisin de 3 sans
alteration de la teneur de 1'air en oxygene. Un rediauffage par combustion interne permet
evemuellement d'elever la temperature a Ifill ('.. i e <|iii correspond a M = 3-4.
Le debit d'air qui resle disponible pour le llux moteur de I'ejii i i n r est lui-meme chauffe a
350 C. par combustion interne, puis recoil une injection d'eau qui ramenc la temperature a
350 C. Cctte preparation du flux moleur assure une economic de 35 pour cent par rapport
a la quantite d'air qui serait necessairc sans elle.
m
La pression dans le caisson de reprise de I'ejecteur est regiee par 1'admission d'un debit d'air
addilionnel aspire a 1'atmosphcre et on peut realiser une simulation d'augmcntation rapide de
I'altitude par simple fermeture de cette admission.
Un calcul simple montrc que la vitesse dc variation de la pression est uniquement comman-
dee par la vitesse de manoeuvre de la vanne. Si celle-ci etait fermee instantanement a partir
d'une position donnant unc pression/), la prcssion diminucrait suivant unc loi sensiblemcnt
cxponenliclle definie par la pente de la caracteristique de 1'ejecteur.
Dans le cas actuel, on trouve que 1'on passcrait de 0-6 atm a 0-31 atm, soit a 0 0 1 pres la
pression d'adaptation, dans un delai de 0-27 sec/m' dc capacite secondairr.
REFERENCES
BRUCE, W. L.. and SCHUBERT, W. A., Chemical Engineering, 6 (1956).
BYKOWSKI, R. M. Environmental Quarterly, 1 (1957).
LE GRIVES, E., FABRI, J . and PAULON, J . "Diagrammes pour le Calcul des fijecteurs i Flux
Moteur Supcrsonique," N.T. ONERA No. 35 (1956).
P. PH. CHAFPIOTTE (Societe d'Exploitation des Materiels Hispano Suiza):

Utilisation d'ejecteurs pour certains essais: Ayant eu, il y a quclques annees, a resoudre un
urgent probleme de developpement d'un propulseur destine au programme d'intercepteurs
legers, il nous est apparu qu'en 1'absence de toute autre installation conventionnelle assez
largement dimensionnee pour les essais d'altitude la voie la plus rapide residait dans 1'utilisa-
lion d'un cjcctcur supersonique utilisant un lurbo-reacteur de servitude corame puissance
motrice.
Fig. 49.
Banc d'essai de chambre de combustion en depression (ijecleur supersonique, Banc 16).
1. Sortie TR H.S. "Nene"
2. Vanne de riglage dibit de by-pass
3. Vanne de riglage pression
4. Vanne enlrie d'air froid
5. Vanne d'air chaud
6. Robinet de riglage fin
7. Chambre de combustion
8. Calorifugeage
Bien que ce procede posse.de en lui-meme un mauvais rendement ct cntraine des frais
d'exploitation assez eiev-es a cause de la consommation du turbo-reacteur, il s'est reveie a
1'usage parfaiternent rentable si 1'on fait intervenir dans le cout global de la mise au point lis
frais d'investissement initiaux.
En effet, 1'cmploi d'un ejecleur pour extraire les gaz d'un foyer dispense non seulement de
machines mecaniques coutcuses, mais egalement des appareils sans lesqucls cites ne pourraient
fonctionncr (laveurs et rcfroidisseur). II est v-rai que, bien controiees, les operations de lavage
et de refroidissement peuvent augmenter I'cfficaciie du pompage.
L'installation realisee est schematisee a la Fig. 49. On notcra la position des organes qui
permettcnt de regler Tejccteur & son point de fonctionnement optimum. On remarquera
egalcmcnt que le prechauffage dc 1'air amene dans Ic foyer est opere dc fac,on tres economique
134
en utilisant les circuits de refroidissement de 1'ejccteur lui-meme. C'est en raison de ce

rechauffage qu'on a rcnonce a augmentcr 1'efficacite du pompage en refroidissant par
injection d'eau. On n'obticnt pas lout a fait la temperature rccherchee, mais on s'en approche
suffisammcnt pres pour ne pas introduire d'errcur grossierc dans les essais du foyer.
Le flux inducteur est produit par I'echappcmcnt d'un turbo-reacteur "Nene" standard
declasse dont on a simplemcnt remplace le convergent d'echappement par un convergent-
divergent adapte a ce travail particulier (la section au col est de J> 450 au lieu de la section dc
sortie standard dc rf> 476-2).
La Fig. 50 donne les courbes debit/pression oblenues, et la Fig. 51 les possibilites resumês
d'un tel dispositif.
(j.8 s '
Temperature d u f l u x Induit - 4 0 0 °K
{goz a 1100 °K r e f r o i d l s par injection cfeau,
m o x i ne donnant pas de c o n d e n s a t i o n , / ^
dons la t r o m p i• )
>
0-6
w ^ '
Tempe •otune du flux induit
100 %
^
0*4 ^y y^'
X y ^ T e m p e r a t u r e du f l u x induit
—"^ y
2 8 8 °K
0-3 Jet Inducteur: Debit 4 0 k g / s e c
Pression; l - 8 8 o t m .
Temperature 9 7 3 °K
0 1 2 3 4 5 6 7 8 10 12
Debit Induit, kg/sec
Fig. 50.
Trompes supersoniques—(cas du jel inducteur produit par un T.R. "Nene") Caractiristique
dibit pression d'un flux induit a une tempirature de 1100°K et 288'K
,a
6 _
0 10 20 30 40 50 60 70 80 90 100 no
0, kg/sec
Fig. 51.
Essai de chambre de combustion en dipression (utilisation d'un "Nene" avec une trompe)
Z altitude, Q dibit nominal du riacteur au sol au point fixe, M nombre de Mach de vol de
Vavion.
Hypotheses: Rendement du manche — 0,94
Rendement adiahatique du compresseur = 0,84
Nombre de Mach constant it I'entrie de la chambre
in 135
R. VV. HENSEL AND II. K. MATT
C'est sur cette installation qu'a ete faite I'etude systemalique en altitude du systeme
d'injection tournanle qui a fait 1'obji-i d'un precedent commemaire ( TIAI ITOTTK, 1958).
REIKRENCE
CIIAFFIOTTE, P. Pll. Combustion £? Propulsion. Third AGARD Colloquium, p. 359 363.
Pergamon Press. London, 1958.
AUTHOR'S REPLY
B. H. GOETIIERT:
Regarding the numerous questions on free-jet testing raised during the discussion ol my
paper, I wish to point out that the pressure ratios required for operation of a free-jet nozzle
with second throat diffuser are given on Fig. 17 as typical values for a good installation ol
this type. It cannot be inferred, however, that these values represent the upper limits. I am
happy to hear that Wasielewski's data for the operating pressure ratios for the free-jet
installations discussed by him are somewhat lower than those presented by me. It should be
kept in mind, however, that the operating pressure ratios depend very much upon the specific
conditions of the free-jel installation, such as overall nozzle and diffuser geometry, angle ol
attack range, leakage flow, and many more. As a matter of fact, several installations ol this
type in the Tullahoma facility led lo considerably poorer pressure recovery ratios than those
presented in my paper, while the test data of Fig. 17 are among the best we have
investigated in a small pilot facility. Il is in lie noted that the data of Figs. 16 and 17 have
been obtained for starting conditions.
Pressure pick-ups
Fig. 52.
Greenhough commented that the quoted value of 5 per cent suction (Fig. 17) is rather
high. I agree with him in principle. It is to be kept in mind, however, that the amount of
suction must be judged in relationship to the high leakage flow (approximately 3 per cent)
for the free-jet installation used in the model tests. The special nozzle for these tests was of the
flexible-wall type which, particularly in the case of fast transient testing, exhibits detrimental
leakage condition. We have indications, however, that the full-scale nozzle installation at
Tullahoma exhibits less than 3 per cent leakage flow.
Greenhough asked also in his comments whether ejector-diffusors located in the immediate
vicinity of the engine exit nozzle might cause interference with the engine and, therefore,
make force measurements impossible. In answer to this question I can mention that in
practically all our rocket tests in the Tullahoma facilities the thrust of the engine was
measured directly with the high precision required. We arc fully aware of the difficulties
which arise in connection with the possibility of interference around the exit nozzle. In order
to check experimentally whether an appreciable pressure field is produced by the diffuser
arrangement, we have conducted several tests in which we measured the static pressures al
the rocket outer surface at several critical points, significantly also on both sides of a tail wing
shown on Fig. 52.
We found that the pressures on both sides of the tail wing were essentially identical as long
as no significant leakage flow into die dilfusi-r inlet existed. However, when leakage flow of
even relatively small quantity occurred, a pressure distribution was produced which led to
significant pressure distribution changes. It is necessary, therefore, to reduce the leakage flow
to unusually small values in order to avoid the indicated difficulty.
Wasiclewski commented that in tests at the Wright Aeronautical plant no change in flow
distortion profile of a ducted nozzle installation was detected when hot-engine or cold-flow
operation of the ducted nozzle was chosen at equal total pressure recovery ratio of the inlet.
In our Tullahoma facility, however, we conducted some tests in which we found some
difference between hot and cold operation. It is my opinion that Wasielewski's statement is
correct that no difference between hot and cold operation will occur when the measuring
136
station for determining the flow distortion profile is located far upstream of the combustor.
When the measuring station is moved further downstream, however, the How disturbance
pattern could be influenced by the locally different combustion in the combiislor while cold
flow operation simulates only the overall pressure in the combuslor chamber by means of a
throttle valve.
137
MF4SURES CONCERNANT LES ESSAIS DE MOTEURS
EFFECTUES DANS LES GRANDES SOUFFLERIES
DE MODANE-AVRIEUX (France)
par MARCEL PIERRE
Grandes Souffleries de Modane-Avrieux
Office National d'Etudes et de Recherches Aeronautiques
L'auleur presente les essais de moteurs executes dans la soufflerie S I MA du Centre

d'Essais de Modane-Avrieux, exploitee par VOffice National d'Eludes et de
Recherches Aeronautiques {France). Ces essais entrent dans le cadre de I'activile
de ce Centre telle que Va presentee M . Maurice Roy, Directeur General de
I'O.N.E.R.A. dans sa communication au 2eme Congres Aeronautique Europeen
sur les "Caracteristiques el Possibilites du Centre d'Essais de Modane-Avrieux"—
{La Haye—Septembre 1956).
Apres avoir enumere les quelques parlicularites se rapportant plus specialemenl
aux essais moteurs, {echange d'air, stockage du carburant, banc d'essais mecaniques)
Vauteur donne la description du chariot de mesures dans lequel s'executent les essais
((tl = 8 m — L = 14 m — poids 500 T ) .
La chambre de distribution des carburants, la balance a six composantes et la
chambre de mesures font robjet de quelques precisions.
Les essais qui ont deja ete realises des la fin de I'annee 1952 concement: des
slato-reacteurs {q> 600 m m ) , un deviateur de jet sur avion Vampire, un engin
cible complet, des combinaisons turbo-reacteurs et de slato-reacteurs de moyenne et
grande puissance, un groupe mnto-propulseur a helice {moteur Pratt-Whitney de
600 cv), des engins speciaux complets.
Pour terminer, I'auteur donne une description succincte d'une soufflerie supersonique
{<f> 600 m m — M = 4,5) pour essais de stato-reacteur, en cours d'achevement, et
un apercu du developpement acluel du Centre d'Essais airothermodynamique de
Modane-Avrieux: soufflerie supersonique continue 1,85 m X 1,75 m — M = 3,
et soufflerie supersonique intermittente 0,80 m X 0,76 m — M = 5,5 dont la
mise en route est imminente.
F4N 1956, a Scheveningen, lors du deuxieme Congres Aeronautique Europeen,

M . Maurice Roy, Directeur General de 1'ONERA, a present^ une
description du Centre d'Essais Aerothermodynamiquc que I'Office possede a
Modane-Avrieux, et donnc; de n o m b r e u x exemples des genres d'essais deja
efiectues ou possibles.*
L a prdsente communication vise specialement a faire le point de la
situation actuelle en ce qui concerne les essais de moteurs dans ce Centre.
Les essais qui vont etre presented ont ete executes dans la soufflerie S1 M A
(Fig. 1) dont la veine d'experiences mesure 8 m de diametre, 14 m de long et
dans laquelle le n o m b r e de M a c h 1,03 a ete obtenu dans certains cas d'essais
aeVodynamiques.
• Publication ONERA No. 85, Septembre 1956.
138
MESURES C O N C E R N A N T LES ESSAIS DE MOTEURS
1. LA S O U F F L E R I E SI MA
Sans entrer dans les details descriptifs de fonctionnement de cette soufflcrie
maintenant bien connue, il est utile d'en rappeler les caracteristiques se
rapportant particulierement aux essais moteurs:
—1'equilibre thermique est obtenu par echange d'air avec 1'atmosphcrc.
Le dispositif principal permettant 1'introduction d'air frais dans le circuit est
constitue par deux cheminees verticales situees immediatement en aval du
coude I I . Dans le but de faire varier le taux d'echange, ces cheminees sont
munies de volets de bord de fuite reglables telecommandes depuis la chambre
de mesures. L'evacuation de 1'air chaud s'effectue par la fente annulaire
situee entre la paroi de la chambre de tranquillisation et le convergent.
•*^r
Fig. 1.
i
—1'echange d'air permet egalement de maintenir dans la veine fluide unc
leneur en gaz brulcs compatible avec le bon fonctionnement des moteurs en
essais.
—I'alimentation en c a r b u r a n t des moteurs s'effectue a partir d'une
installation dc stockage c o m m u n e a toutes les souffleries. Cette installation
comprend 6 citernes de 25 m 3 chacune et une station de pompage capable
d ' u n debit maximum de 30 T / h sous une pression de 6 kg/cm 2 .—Dans le cas
de pression d'alimentation plus eievee (stato-reacteur) les pompes sont
fournies par le Constructeur. Le carburant arrive jusqu'a la veine d'experi-
ences oil sa distribution est assuree par un equipement installe sur le chariot
d e mesures.
139
M. PIERRE
De menu- que pour chacune des deux autres veines d'experiences, ce

chariot peut occuper dans le bailment de mesures plusieurs positions
reservejes: au montage des moteurs ou engins, a la verification du fonctionne-
ment mecanique de ceux-ci, aux essais proprement dits.
—le banc d'essais servant a la verification mecanique prealable des moteurs
est constitue par deux tours verticales accolees au corps principal du
bailment, ces deux tours formant les deux branches d ' u n U dont la partie
horizontale est constitute par la veine d'experiences portee par le chariot
de mesurcs.
Ce banc d'essais est equipc d ' u n tube telescopique refroidi it 1'eau, pcr-
mcttant dc collecter les gaz des reacteurs afin d'eviter q u e ceux-ci ne
deteriorent les parois des cheminees.
2. LE C H A R I O T DE MESURES
Le chariot de mesures (Fig. 2) est constitue de la veine d'experiences
c
j t ¥eiae ¥
Inbre des
H-burants
t l d'experiences M
L
Fig. 2.
•
proprement dite portee par u n chariot pcrmettant les deplacements a
1'interieur du bailment dc mesures.
Lors des essais, la veine d'experiences est verrouillee a 1'amont et a 1'aval
sur le rcstc du circuit acrodyuamique dont elle cn constilue une partie.
U n e liberte' de deplacement entre la partie superieure (section d'essai) et
la partie inferieure (chariot) permet les libres dilatations de ces ensembles.
Le poids total du chariot et de ses equipements est d'environ 500 tonnes, sa
longueur et sa largeur sont de 14 m, sa hauteur de 15 tn.
P e n d a n t les essais, la pression regnant a 1'inteVieur de ce chariot est
d'environ une demi-atmosphere, 1'etancheite doit done etre assuree. En
particulier, un joint vient se plaquer sur le sol pour isoler la partie inferieure
de 1'atmosphcre, ce joint etant releve pendant les deplacements.
Ce chariot comporte trois equipements principaux.
la chambre de distribution et de mesure de debit du carburant,
la balance destinee a la mesure des efforts,
140
MESURES CONCERNANT LES ESSAIS DE MOTEURS
la chambre de mesurcs insonorisee on sont concentres les apparcils de

controle et de mesure servis par Ic personnel experimentatcur.
2.1.—La chambre de distribution des carburants comprend le circuit
d'alimentation equipe de 1'apparcillage necessaire a la regulation de la
pression et a la mesure des debits.
Le circuit permct d'alimcnter simultanemcnt deux moteurs avec un meme
carburant ou un moteur avec deux carburants differents.
La regulation de la pression regnant dans le circuit situe en amont des
appareils de mesures, est obtenue au moyen d ' u n devcrseur prealablement
regie a la valeur desiree de la pression, deverseur assurant cgalcment le rctour
dans 1'installation de stockage du carburant non consomme.
La mesure du debit volume est basee sur Ic principe consistant a trans-
former ce debit en un nombre d'impulsions electriques qui lui soit propor-
tionnel. Ces impulsions sont recues par un compteur electronique.
Les compteurs volumetriques utilises sont, soit des compteurs a piston
auxquels un emetteur d'impulsions (reluctance variable) a ete adapte, soit
des compteurs a turbine dans lesquels le dispositif a reluctance variable est
inclus.
La mesure du debit masse s'cffcctuc en pesant, pendant un temps deter-
mine ( 1 - 2 - 6 - 3 0 ou 60 sec) la quantite de carburant alimentant le moteur.
Cette mesure s'execute entre deux tops declanches en regime continu
d'alimcntation; de ce fait, 1'imprecision resultant d'ouverturc ou dc fermeture
de vanne est eliminee.
La bascule servant a la pesee du carburant comporte unc aiguille qui
entrainc un disque par I'intermediaire d'un accouplement elastique. Le top
de depart est declanchc par un petit aimant portc par 1'aiguillc. Au top
d'arret le disque est immobilise alors que 1'aiguille poursuit son mouvement.
La rotation de ce disque est proportionnellc a la quantite de carburant
consomme pendant le temps de reference separant les deux tops.
U n autre disque, perce de 3000 fentes, et une cellule photo-electrique
permettent de definir avec une grandc precision la rotation du premier
disque.
La mesure du debit s'effectue a ;f 1/2000 pres.
Dans les deux cas de mesures de debits, les impulsions sont envoyes a un
compteur electronique servant egalement a la mesure des regimes de
rotation.
2.2.- La balance a 6 composantes qui equipe le chariot " m o t e u r s " permet
deux sortes de montage (Fig. 3) diametral et sur plateforme avec voilure et
sur mat. La structure inferieure de la balance repose sur des fondations
independantes de celles du batiment.
Dans le premier cas de montage, 1'aile supportant le moteur est fixee a
deux caissons dynamometriques coulissant sur les poutres verticales de la
structure superieure de la balance. Ces caissons peuvent se regler en hauteur
sur une course de 1 m afin de placer 1'engin Ic plus prcs possible du centre de
la veine. Ces caissons c o m p o n e n t a leurs parties superieures des platincs
orientables sur lesquelles 1'aile se fixe directement, platine permettant la mise
en incidence du moteur de —5° a -)-15°.
Dans le second cas de montage, le mat supportant le moteur se fixe sur la
plateforme orientable reposant sur la structure inferieure de la balance.
141
M. PIERRE
Entre ces deux parties des elements dynamometriques permettent la mesure

des 6 composantes (portance maxi de 8 T ) . Eventuellement, cette balance
plateforme peut servir a 1'execution d'essais aerodynamiques sur une demi-
aile de grandes dimensions.
L'ensemble de la structure de la balance est tres rigide, les deformations
sous les charges maximums sont faibles (13/10 m m au maximum) bien que
les efforts appliques soient importants (10 tonnes en trainee, 20 tonnes en
portance dans le cas du m o n t a g e diametral).
Les dynamometres pneumatiques utilises transforment 1'effort a mesurer
en une difference de pression d'air; la course de ces dynamometres est tres
faible, de I'ordre de 15 X 10 6 m m . Cette difference de pression est appliquee
Ac
avec voilure
Fig.3.
a une capsule rdceptrice equipant une balance automatique de haute

precision (1/10.000). C h a q u e balance automatique (il en existe une p a r
composante a mesurer) entraine u n compteur place dans la chambre de
mesures. Les experimentateurs peuvent done lire directement les efforts qui,
p a r ailleurs, sont enregistres sur une machine a ecrire.
U n e perforatrice de b a n d e complete 1'equipement, la bande perforce
servant a 1'envoi (par teletype) des resultats bruts au centre d e calcul de
1'ONERA a Paris, pour permettre les depouillements.
Le tarage des balances diametrale et plateforme ont confirme leur bonne
precision (superieure a ^ 3/1000). Des reglages effectues pendant le tarage
ont permis d'annulcr les interactions entre les differentes composantes.
La precision de cette balance concue pour la mesure des fortes charges
(maximum 20 tonnes) s'etend a la mesure de faibles efforts puisqu'au cours
d ' u n tarage complementaire, des forces de 750 kg ont ete mesurees a ^ 1/100
pres.
2.3. Chambre de Mesure.—La c h a m b r e de mesure, insonorisee et climatisee,
permet de soutenir u n travail continu, dans des conditions excellentes, malgre
les bruits emis par les moteurs.
142
L'installation de raesures proprement dite comprend:

u n pupitre "Chef d'essais",
deux pupitres " C o m m a n d e et controle du ou des moteurs"
un pupitre "Mesure des forces",
u n tableau "Mesures des parametres moteurs"
Le pupitre du " C h e f d'Essais" permet de preparer, de diriger ou d'inter-
rompre 1'essai grace a u n ensemble d'appareils de controle et de liaisons
cleclriques et telephoniques.
Les deux pupitres " C o m m a n d e et controle du ou des moteurs" permettent
la commande des moteurs (a piston, reacteurs, stato-reacteurs, composite)
ainsi que le controle global du fonclionnement de ces moteurs: nombre de
tours, temperatures, pressions, etc. . . . Ces pupitres comportent egalement
la commande du dispositif de lutte contre 1'incendie (au gaz carbonique).
Le pupitre "Mesure des forces" permet la mise en oeuvre de la balance
ainsi que le controle de son propre fonctionnement, la lecture des efforts sur
des compteurs visiblcs, 1'imprcssion des resultats sur machine a ecrire
electrique et la perforation des bandes.
Le pupitre "Mesures des parametres moteurs" comporte essentiellement
1'appareillage destine aux mesures (en dehors de celles deja controlees en vue
de surveiller le bon fonctionnement des moteurs): regime de rotation,
temperatures, debits de carburant, prcssions.
Les regimes de rotation sont mesures automatiquement; ils peuvent, soit
etre lus sur un compteur electronique, soit enregistres sur une bande, soit
imprimes sous forme de tableaux de chiffres par une machine a ecrire.
Les mesures des temperatures sont executees de facon automatique a I'aide
d'une chaine permettant 1'impression des resultats et la perforation d'une
bande.
L'equipement de base de la chambre de mesures peut etre complete a
la demande en fonction du programme des essais.
3. ESSAIS REALISES
Les premiers essais executes dans la soufflerie SI M A ont debutd en 1952
p a r des essais avec combustion.
3.1.—Un stato-reacteur de 0,60 m de diametre et de 4,30 m de long, a
fait 1'objet de trois series d'essais en Octobre 1952, Decembre 1953 et Janvier
1954 (Fig. 4). Ces experiences etaient destinees a 1'etude de la combustion.
Ce stato etait suspendu a un m a t vertical h a u b a n n e lateralement, equipe
a sa base d'un dynamometre destine a mesurer la poussee du stato-reacteur.
Ce m a t droit, construit en alliage leger, a une epaisseur relative de 10 pout-
cent et une corde de 1,30 m .
O u t r e les mesures normales (debits, pressions de carburant, temperatures
et debit d'air) des pressions etaient mesurees le long des parois du stato-
reacteur et de la veine d'experiences.
D u r a n t les essais, trois types de diffuseur et 41 configurations de combustion
(grille d'injection, injecteur, stabilisateurs de flammes) ont ete etudies pour
la g a m m e complete des richesses et pour des nombres de M a c h atteignant 0-8.
La comparaison des performances et de la stabilite de combustion pour
les diverses configurations essayees a permis de retenir les meilleures
dispositions compatibles avec les conditions d'emploi prevues.
143
M. PIERRE
3.2.—Un essai de deviateur de jet a ete realise en deux series (Novembre

1952 et Mai 1953) dans des conditions reelles de vol (Fig. 5).
Fig.4.
Fig. 5.
Dans ce but, un avion Vampire, cquipe d ' u n turbo-reacteur Goblin I I ,

a ete monte diametralement dans la veine apres que 1'envergure des ailcs
ait ete reduite a 8 m.
La suspension de 1'avion en veine, en trois points, comportait un tourillon
a chaque extremite d'aile et un m a t vertical sous 1'avant du fuselage. Des
144
dynamomctres places cntrc I'avion et le chariot de mesures permettaient la

mesure de la portance, du moment de tangage ainsi que de la trainee.
L'appareillage usuel (tachymctre, debitmetre) caracterisait Ic fonctionne-
ment du moteur. Des enregistreurs de temperatures a voies multiples et des
multimanomctres pcrmettaient d'apprecier Taction du deviateur de jet.
U n e visualisation simple du jet a ete realise au moyen de photographies
prises a 1'aidc de lampes a eclairs.
Pendant les deux scries d'essais, differents obstacles acrodynamiques et
mecaniques ont ete essayes. Des configurations ont ete rcalisees pour divers
regimes de turbo-reacteur et pour des vitesses de la soufflerie allant jusqu'a
Af = 0,65.
3.3.—Les essais d'un second stato-reacteur de 0,65 m de diametre et de
2,80 m de long ont eu lieu en Decembrc 1952 et Novembre 1953. La Fig. 6
Fig. 6.
represente ce stato-reacteur equipe d'une canne de prises de pression destinee

a la connaissance preliminaire du c h a m p aerodynamique.
Ces essais concernaient 1'etude de la combustion. Le montage etait
identique a celui utilise pour le premier stato-reacteur.
Les essais de deux diffuseurs et de onze dispositifs de combustion (grilles,
injecteurs, accroche-flamme . . .) eurent lieu a diverses richesses et pour des
vitesses de la soufflerie allant jusqu'a M = 0,85.
3.4.—Les essais d'un propulseur composite: turbo-stato-reacteur furent
entrepris en Juillct 1954. Ces essais consistaient essentiellement a verifier la
possibilite de bon fonctionnement de cet ensemble et a etudier la combustion
du stato-reacteur. Le stato-reacteur comprenait un corps central contenant
un turbo-reacteur "Palas." L'engin mesure 1 m de diametre et 5,60 m de
longueur; il etait monte a la base du mat haubanne. L'appareillage de
mesures comportait, en outre, deux dynamomctres de poussee: 1'un pour
le turbo-reacteur seul, 1'autre pour le propulseur complet. Cet essai necessitait
de nombreuses mesures effectuees par une equipe que la Fig. 7 represente
au travail, alors que la c h a m b r e de mesures n'etait pas encore completement
installee.
Deux dispositifs de combustion ont etc essayes en fonction du regime du
145
M. PIERRE
turbo-reacteur et de la richesse du stato-reacteur, ceci pour des vitesses de

la soufflerie allant j u s q u ' a M = 0,80.
Ces essais ont confirme la possibilite d ' u n fonctionnement satisfaisant d ' u n
tel propulscur. Ils ont permis, en outre, de selectionner, le meilleur dispositif
de combustion.
3.5.—Les essais d ' u n engin cible ont eu lieu au mois de Decembre 1954
(Fig. 8). Le but de ces essais concernait 1'etude du comportement general
a e r o t h e n n o d y n a m i q u e de 1'engin. Celui-ci, d'unc longueur de 6,10 m et
d'une envcrgure de 3,60 m est muni d ' u n empennage papillon et propulse
par u n turbo-reacteur " M a r b o r e " .
Fig. 7.
L'appareillage de mesures propre a 1'essai comportait un dynamometre

de poussee, u n enregistreur de position des gouvernes, u n enregistreur de
frequence des battements de spoilers.
Pour des raisons de securite, compte tenu de la facon inhabituelle dont
les efforts se transmettaient dans la structure de 1'engin par suite du mode de
sa fixation dans la veine d'experiences, des strain-gages temoins mesuraient
le m o m e n t d'encastrement de la voilure.
Les parametres etudies etaient: 1'incidence de 1'engin, le braquage des
gouvernes de profondeur, 1'amplitude et la frequence de battement des
spoilers, le regime du turbo-reacteur.
Le n o m b r e de M a c h m a x i m u m obtenu a ete de 0,79.
Ces essais ont permis d'etablir le bilan trainee—poussee de 1'engin et
d'apprecier son comportement general dans les conditions de vol.
3.6.—Trois series d'essais de reservoirs pendulaires places sous un stato-
reacteur se sont deroulees en Avril 1955, Juillet 1955 et M a i 1957 (Fig. 9).
lis consistaicnt a etudier d ' u n e part, les interactions entre plusieurs reservoirs
146
pendulaires et le corps du stato-reacteur et d'autre part la combustion et la

regulation de ce stato-reacteur.
L'engin reel, suspendu au mat, mesurait 0,65 m de diametre et 2,80 m
de longueur.
Fig. 8.
Fig. 9.
147
M. PIERRE
Les paramctres etudies jusqu'a AI = 0,87 etaient les suivants:

formes et dimensions de reservoirs pendulaires,
dispositif d'allumage et de combustion,
mode d'injection (continue ou pulsatoire),
nature du carburant.
Ces essais ont permis de selectionner un reservoir provoquant la plus faible
variation de trainee globale et de choisir la meilleure solution concernant
la combustion et la regulation du stato-reacteur.
3.7.—Les essais d'un important propulseur composite turbo-stato-reactcur
ont eu lieu en deux series, Avril et Decembre 1956 (Fig. 10). Ils avaicnl
pour but 1'ctude de la combustion de ce propulseur composite.
Fig. 10.
Le fuselage reel d'un avion, de 13,50 m de long et de 1,50 m de diametre

etait place dans 1'axe de la veine d'experiences. U n turbo-reactcur ATAR
est monte a Tinterieur et au centre de ce fuselage. La partie annulaire
comprise entre le fuselage et le corps central renfermant le reacteur constitue
le stato-reacteur.
Afin d'etre monte diametralcment dans la veine et place sur la balance
molcur, cc fuselage avail ete muni d'ailes droites fixees aux elements
dynamometriques de la balance moteur.
Etant donne 1'importance exceptionnelle dc 1'effet bouchon du a la
presence de cet avion, des mesures de pressions a la paroi controlaient la
validite des ecoulemcnts amont et aval.
Plusieurs tuyeres de turbo-reacteur, chambres de stato-reacteurs et
dispositifs de combustion ont ete essayes pour des vitesses de la soufflerie
allant jusqu'a M = 0,75.
Ces essais ont permis de selectionner la meilleure configuration d'ensemble,
d'apprecier la tenue de la chambre avionnable et le bon fonctionnement du
dispositif de regulation.
En outre, au cours de ces experiences, le pilote d'essais qui a fait effectuer
les premiers vols a cet avion et en a poursuivi ensuite la mise au point en vol,
148
est venu sc familiariser avec les commandes de cet important composite en

les manoeuvrant depuis la chambre dc mesures et a pu ainsi en etudier a
1'avance et pratiquement les "reactions".
3.8.—Les essais d'un groupe moto-propulseur a helice ont ete executes en
Fevrier 1957 (Fig. 1 1 ) /
Leur but etait d'etudier Ic refroidisscnu-nl du groupe molo-propulscur et
de comparer les performances obtenues avec deux helices de type different.
La maquette consistait en un fuseau moteur d'avion reel monte sur un
element de voilure. Le fuscau, d'un diametre de 1,35 m est muni d'un
moteur Pratt & Whitney de 600 cv.—Le diametre des helices est dc 2,60 m
et la profondcur de 1'aile dc 3,30 m.
Fig. 11.
La maquette etait montee diametralement sur la balance.

L'appareillage de mesures particulier a cet essai comprcnait un bati moteur
dynamometrique permettant la mesure du couple moteur.
Les paramctres etudics etaient: 1'incidcnce de la maquette, le braquage
des volets hypersustentateurs, 1'ouverture des volets de refroidissement, la
position de la jambe dc train d'atterrissage, 1'ouverture des volets de radiateur
d'huile.
Les vitesses de vent ont varie de 100 a 550 km/h.
Ces essais ont permis de determiner la position optimum des volets de
refroidissement et de constater la quasi egalite des performances du G M P
avec 1'une ou 1'autre des deux helices essayees.
3.9.—Les essais d'un engin special complet ont eu lieu de Juillet a Octobre
1957.
La Fig. 12 represente 1'engin reel dans le chariot de mesures celui-ci est en
position de montage dans le hall. Les remplissages correctifs de la loi de
section longitudinale apparaisscnt nettement. A 1'arricre plan un autre
chariot de mesures reserve a des essais aerodynamiques est visible.
Les essais executes sur cet engin reel avaient pour but d'etudier le
149
M. PIERRE
comportement aerodynamique d e 1'engln complet et la combustion du stato-

reacteur.
L'engin a une envergure de 4 m et une longueur de 3,60 m . II £tait m u n i
d ' u n reservoir pendulaire.
L'appareillage de mesures, particulier a cet essai, comportait une balance
situee a la base du m a t et destinee a mesurer les trois composantes X, Z et M ,
des dynamomctres des moments de charniere, des appareils photographiques
pour mesurer les deformations de la voilure.
Les parametres etudies j u s q u ' a des nombres de M a c h de 0,8 etaient:
la grille d'injection, la c h a m b r e de combustion, Tangle de calage du reservoir
Is ^FjP* _âaViBBBw^ ap4aBBVBBP^BBiâBBBBi\aa. ^^â.

aaaal
i j f ^ ^ y f ? ' r 9 ~ 3 ^ a a W âaaaiaaaaaaaaaLaaal
riLlxif il
\\W |Z S^ //I Jri/' * U 1
am WW F^MlBlIC
WmM^L ~" x^PB m m
* ^^S ' 4aaa6
BBBBBa. * wwWm\
Fig. 12.
pendulaire, le calage des volets, le braquage des elevens, Tincidencc de

Tengin complet.
Ces essais ont fourni les caracteristiques aerodynamiques de Tengin et ont
permis de selectionner la meilleure configuration de Tensemble "combustion".
Le m a t support dont il vient d'etre question a plusieurs reprises sera
remplace incessamment p a r u n m a t plus fin, legerement en fleche (18°).
L'epaisseur relative est de 6 pour cent la corde (cote moteur) est de 2 m.
La rigidite de ce mat, construit en acier, a ete definie telle que la majorite
des essais puisse s'executer sans haubans lateraux. Cependant, ceux-ci
pourront etre monies lorsque les conditions de Tessai Texigeront: risques
d'efforts transversaux importants ou risques de vibrations.
4. Tous les essais qui viennent d'etre presented ont ete executes en subsonique
dans la soufflerie SI M A .
Certains d'entr'eux, comme les stato-reacteurs par exemple, pourront etre
poursuivis en supersonique dans une soufflerie a rafales, notee S5 MA, en
cours d'achevement.
Dans cette soufflerie, u n jet supersonique debouche a M = 1,5-2,5-3,5-4,5
150
dans un caisson d'essais de 2,50 m de diamctre et 11 m de long (Fig. 13).

Cette soufflerie est alimentcc p a r des reservoirs spheViques d'un volume
global de 5500 m 3 dans lesquels de Tair sec est comprime a 10 atm. Selon
le nombre de Mach, la duree des rafales varie de 50 a 80 sec. U n e vanne
a ouverture rapide (2 sec) permct Ic dcclenchement de la rafale et u n
rechauffeur a combustion de kerosene portc la temperature de Tair a
300°C maximum.
Trois cas de fonctionnement sont prevus:
(a)—le caisson est relie directement a Tatmospherc: Ic nombre de M a c h
maximum obtenu est de 2,5 et Taltitude de retablissement est au
maximum de 8 km (cas I de la Fig. 13).
cat 1
: *4)BBB Vonne rapid*
^^'DatandW
Vonoe (fiiolamao*
• to. 7
cas 3
Pig, 13.
(b)—une premiere trompe a air extrait Tair dans le caisson d'experiences

pour le rejeter a I'atmosphere: le nombre de Mach maximum obtenu
est de 3,5 et Taltitude de retablissement est de Tordre de 20 km (cas
2 de la Fig. 13).
(c)—une seconde trompe, placee en serie avec la premiere, contribue a
diminuer encore le niveau de pression dans le caisson, permettant
ainsi Tobtention de M = 4,5 et d'une altitude de r^tablissement de
30 km (cas 3 de la Fig. 13).
5. Cette soufflerie S5 M A fait partie de Textension, en cours d'achevement,

du Centre d'Essais de Modane-Avricux ( O N E R A ) (Fig. 14).
Cette extension comprend en outre:
—une soufflerie supersonique a fonctionnement continu destined a Texecution
d'essais aeVodynamiques, notee S2 MA. Des nombrcs de Mach compris
entre 0,9 et 3 doivent etre obtenus dans une veine d'experiences de
1,85 m X 1,75 m. La pression gcneratricc peut varier de 0,2 a 2
atmospheres.
—une autre soufflerie supersonique mais a fonctionnement intermittent,
ii
151
M. P I E R R E
notee S3 M A , est dotee d'une veine d'experiences de 0,80 m X 0,76 m

dans laquelle le nombre de M a c h 5,5 doit clre obtenu.
Reservee a des essais aerodynamiqtics, de flottement et d'echange de
chaleur, cetle souffleric pcrmet d'obtenir des rafales variant de 30 a 60 sec
en fonction du nombre de Mach obtenu.
Fig. 14.
Fig. 15.
La Fig. 15 illustre Teffort particulierement soutenu par T O N E R A

concernant Tedification de ces nouvclles installations dont cette derniere
photographic vous montre, a Tarriere plan, Tctat d'avancement actuel.
152
SUMMARY
The author describes the engine tests carried out in the S I MA wind-tunnel of the
Modane-Avrieux Test Center run by the Office National d'Etudes et de Recherches
Aeronautiques (France).
After reviewing a few characteristics more specifically related to engine tests
{air exchange, fuel provision, mechanical lest beds) the author describes the
measurement cart in which the tests are carried nut (<£ = 8 m — L = 1 4 m —
weight 500 T ) . The fuel distribution chamber, the six-component balance and
the lest section are stressed.
The tests cnnducted since late 1952 have concerned: ramjets {qj 600 m m ) , a
reverse thrust on a Vampire aircraft, a complete target missile, medium and high
power turbnjets and ramjet systems, a propellor power plant (600 c v Pratt-11 'hitney
engine), and complete special engines.
In conclusion, the author gives a brief description of a supersonic wind-tunnel
{<f> 600 m m — M = 4-5 )for ramjet tests, which is in process of completion, and
outlines the plans for further developing the aernlhermodynamic Test Center at
Modane-Avrieux: a continuous supersonic wind-tunnel l - 8 5 m x l ' 7 5 m —
M = 3, and an intermittent supersonic wind-tunnel 0-80 m X 0-76 m — M =
5-5 the construction of which will be started in a near future.
DISCISSION
O. FRENZL (S.N.E.C.M.A.):
II ressort nettement de plusieurs conferences que les efforts actuels tendent a 1'utilisation
frequente des trompes a induction en remplacenicnt des installations avee comprcsscur
coutcuses et sensibles. Goethcrt evoqua, cntrc aulrcs, un entrainement auxiliaire par jet de
Accumulateur
d ' e a u choude
Trompe
Recupereteur
d'eau
Fig. 16
Trompe a eau chaude monoilage
vapeur pour un banc d'essais-fusees et, Pierre, one trompe a air comprint fonctionnant avec
un ou deux etages, en vue de Taugmcntation du nombre de Mach ou de la simulation des
conditions d'altitude dans une soufflerie. A ce sujet, il semblcrait interessanl de signaler qu'en
France, a la SNECMA, a ete developpec unc trompc qui utilise comme agent-moteur de 1'cau
chaude a la pression dc saturation et qui presenle dc nombreux avantages par rapport aux
trompes connues.
.\insi qu'il ressort de la Fig. 16. 1'cau chaudc provenanl d'un accumulaleur suns pression
passe a travers une tuyere dans laquelle elle s'evapore particllemcnt (FRENZI., 1956), mais rcstc
en grande panic liquidc et est puherisec finement. Le jet, a densite specifique rclativement
153
M. P I E R R E
eievee, sortant dc la tuyere, refoulc Fair dans la chambre de melange et dans le diffuseur,
contre la pression atmospherique. Par 1'intermediaire de 1'incrtie de masse, des quantites
d'eau chaude tr£s imporlantes peuvent ctrc separees du melange, a 1'cxtremite du dilTuscur,
au benefice de la recharge d'accumulateur.
20(-Debit d'air aux conditions

I ijf optimum (kg/s)
h
Debit d'air
15 pour /?=52-5 •
atm. abs.
0-52,,
0-52kg/s
.b 10
1-30'
oX-^—-X~pX-
X~-~ S-SQ 5 ' 3-89
0-2 0-4 0-6 0-6

Pression totole oe I'oir en amont de lo trompe
Pression atmospherique
Fig. 17
Caractiristiques d'une trompe a eau chaude monoitage (air aspiri a une tempirature 18°C)
Fig. 18
Trompe a eau chaude d plusieurs itages
Les essais effectues sur des trompes a eau chaude monoetages prouvent que la quantite
neccssaire en agent-moteur (Fig. 17) est seulcment d'environ la moitie de celle d'une trompc
a air c o m p r i n t correspondante, et environ egale a la quantite de vapeur motrice d'une
154
MESURES CONGERNANT LES ESSAIS DE MOTEURS
trompe a vapeur a des pressions habituelles. Ceci presente des avantages considerables du
point de vue du fonctionncmcnt economiquc, etant donne que 1'cau chaude est trois fois
moins chere que la meme quantite de vapeur ou d'air comprime. En outre, 1'accumulation
d'eau chaudc etant nettement moins couteuse que cclle de la vapeur ou meme de Fair
comprime (FRENZL, 1955) il est egalement possible, avec ce nouvel entraincment, de cons-
truire de grandes souffleries supersoniqucs capables dc longues durees de fonctionnement.
40 60
Py, k g / c m 2
Fig. 19
Trompe a eau chaude a dibit d'air ziro
100
Rendement thermique
A titre d'exemplc, signalons qu'unc soufflerie de 85 X 85 cm, Mach = 3, enlrainee au
moyen d'une trompc a cau chaudc, soufflerie dont la duree dc fonctionnement est de 6 a 15
min, se trouve en cours dc montage a St. Cyr; il est projete une soufflerie analogue ayant unc
section dc la vcinc de mesure egale a 3 x 3 m (BROCARD, 1956).
En ce qui concerne la trompe a eau chaudc mullietage (Fig. 18) il est permis dc s'altcndre a
des avantages particuliers, du fait qu'il est facile de separer une quantite importante dc
1 'agent-motcur cntre les etages.
155
M. PIERRE
Si 1'on considerc Ic simple cas d'unc trompe avec un debit d'air egal a zero, il est possible de
calculer les relations representees sur la Fig. 19 en se basant uniquement sur 1'equation
d'energie. La teneur cn vapeur A"2, k la sortie du diffusenr, augmente lorsque la pression de
1'eau chaude P t croit ct que la pression a la sortie dc la trompe P 2 diminue. Dans la pratique
ccile leneur sera d'environ 0,33, etant donne qu'il est possible, pour des faibles pressions de
sortie P., d'utiliscr egalement des faibles pressions d'eau chaude P , d'environ 5 it 15 atm, sans
pertes remarquables de rendement, ainsi que le rcpresenlc la Fig. 20. 63 pour cent de I'agcnl-
moteur pcuvent etrc separes sous forme d'eau, sans refroidissement.
D'apres la Fig. 19, il est en outre possible de constater que la temperature dc sortie /,
diminue, lorsque la pression de sortie decroit et qu'a P . — 0,006 atm elle atteint Ic point de
congelation. Lors de pressions encore plus basses, il s'echappc de la vapeur ct de la glace du
difTuseur. Ces basses temperatures sont tres soubaitables du point de vue travail dt- com-
pression el evitenl le refroidissemenl enlre les preinii-rs etages. I.urs(|ue la trompc refoule de
Fair ou nn melange vapeur-air, les valeurs X% ct t% valient par rapport a celles de la Fig. 19 et
ceci d'autant moins que le rapport des quantites dc 1'agent-moleur cn fonction de I'agent
entraine, est grand; le comportcmenl de base ne s'en trouve toutefois pas modifie.
II est indique d'injecter de 1'eau froide en aval du dernier etage en vue de recuperer
davantage de chaleur, grace a la condensation. L'injection d'eau froide entrc les derniers
etages rc-duit la quantite de vapeur a comprimer. Mais, si l t est inferieure a la temperature
de 1'eau dc rcfroidissement, 1'injection devient nocive.
Hien que la trompe a eau chaude monoetage, qui a ete 1'objcctif de developpemrni dc la
SNECMA (FRENZL, 1958a et b), soit prochainement utilisec cn Europe, il semble necessaiie,
cu egard aux nombres de Mach et aux altitudes de vol sans cesse croissants, de deVclopprr des
trompes a eau chaude multietage.
REFERENCES
BROCARU, J. Les moyens industriels d'experimentation aerodynamique dc 1'AICMA.
Deuxieme Congres Aironautique Europien 1956—Scheveningen.
F'RENZL, O. Uber die Entuicklung von intermittierend arbeitenden Windkanalen mit
Strahlantricbcn, dahrbuch der W.G.L., 1955.
FRENZL, O. Stomung verdampfenden Wassers in Diisen, Maschinmbau u. Warmewirtschaft.
Wii-n, Heft 1 und 2 1956.
FRENZI., O. Der Heisswasserstrahlapparat, seine Bedeulung fur Forschungs -und Versuchs-
anlagen. I.iiftfahrttechuik, Februar 1958a.
FRENZL, O. Soufflerics intermittentei a trompe d'induction par eau chaude. DOCAERO
No. 51, Juillct 1958b.
156
PART II
ENGINE COMPONENT T E S T I N G
MODEL TESTING OF TURBINES
AND COMPRESSORS
By S. F. S M I T H a n d C. E. P A Y N E
Rolls-Royce Ltd., Derby, F.ngland
The merits nf model testing as a means of evaluating the aerodynamic characteristics

of turbines and compressors are discussed in this paper and some consideration is
given lo the correlation of model and full scale results. The paper also includes a
brief account of the mechanical construction nfthe models, the layout of the test plant
and the instrumentation techniques which have been evolved at Rolls-Royce.
It is concluded that model compressors which are used primarily for research wnrk
form a very valuable adjunct to the development work carried out on full scale
compressors. Model turbines have such great advantages over engine turbine tests
that we believe this to be the only economic means of carrying out a large scale
programme of turbine aerodynamic research and develnpment.
PART I: MODEL TESTINC OF TURBINES

By S. F. SMITH
I T HAS for several years now been the practice at Rolls-Royce to determine
the basic aerodynamic characteristics of the turbines of our engines from
tests of model turbines. I n addition the model turbines are used as the
primary means of obtaining information for the aerodynamic research and
development which is carried out to improve turbine performance. T h e
first model of the Denvent I turbine was tested in 1945 and since then the
programme of tests has been fairly continuous at an average rate of just less
than one test a week; u p to the present some 550 different turbines
configurations have been evaluated.
In this paper it is intended to outline the advantages of this type of model
test work, and to give some details of the construction of our model turbines
and the layout of the test plant.
Advantages of Model Turbine Tests
T h e basic reason for all model or separate component testing on jet engines,
as opposed to full scale engine tests, is to obtain accurate and consistent test
data at the least cost and in the shortest time. It is important that the model
test data should be available quickly, as at least part of its value is lost if the
results are not available in advance of the engine.
I n the case of model turbines there are three major factors which enable
one to achieve these objectives: firstly, the turbine inlet temperature, for the
purposes of the test, can be reduced to a value of say 100-200°C (200-400°F),
secondly, the turbine can be scaled to any suitable size and thirdly, the
turbine is divorced from the other components of the engine and from the
engine mechanical construction.
T h e reduction in the turbine inlet temperature is most important since
this allows the correct air M a c h numbers within the blading to be reproduced
159
S. F . SMITH AND C E. PAYNE
at lower values of actual blade speed and, therefore, low stress levels; this
together with the low temperature makes it possible to use inexpensive and
easily machined materials l<n the construction of the model turbine. Testing
at relatively low temperature is also a notable factor in helping to obtain
accurate test data easily and with relatively inexpensive equipment. At a
turbine inlet temperature of 100 to 200°C it is not difficult to obtain a uniform
level of temperature which is easy to measure accurately. In contrast turbine
inlet temperatures of the order of 1000°C, are only practicable if the air is
heated in combustion chambers close to the turbine. This means that high
temperatures of this order will be non-uniform and the assessment of a mean
turbine on which to base the turbine efficiency is a matter of some difficulty.
Furthermore at low temperatures, subsidiary airflows for cooling purposes
are unnecessary and the absence of extraneous air flows obviously simplifies
the interpretation of the test results. A further advantage of testing cold is
that thermal expansions are negligible so that the various turbine clearances
which are important from the performance aspect can be measured accurately
without difficulty.
T h e advantages of scaling a turbine to a suitable size are obvious, since
this enables one to choose a size which is the most economical to manufacture.
T h e combination of a modest turbine size together with low temperatures
and pressures makes the size and expense of the lest plant small compared
with that which would be necessary to test a full size turbine at full tempera-
ture and pressure. I n our case the model turbines arc between 15 and 20 in.
tip diameter, the smaller diameter turbines are those which pass a large
flow per unit frontal a r e a .
O n e of the prime requirements for a modern aero gas turbine engine is
that it shall be small and light in relation to the thrust or power it produces.
This requirement inevitably means that in an engine turbine there is very
little space for detailed instrumentation. A model turbine, in contrast, can
be designed specifically to suit any required degree of instrumentation and
to simplify the instrumentation necessary. Furthermore, since a model
turbine is not tied to the mechanical design of an engine, it can be built and
tested at the earliest stages in the design of a new engine. It has proved most
useful to have a complete aerodynamic calibration of the turbine well in
advance of the first test of a new engine.
Mechanical Construction of Model Turbines

T h e general layout and the mechanical construction of a typical model
turbine is shown in Fig. 1.
T h e mechanical parts of the turbine, that is the shaft, bearings, oil system
and main casing, shown in this diagram are common to all our model
turbines and are simple, robust and extremely reliable. T h e r e are four
identical such "milk c h u r n s " and we normally aim to have one turbine on
test, a second awaiting test and the remaining two being rebuilt for further
tests.
T h e turbine blades a n d the shroud rings and exhaust unit are individual
to one particular turbine configuration and, with exception of the sheet
metal inlet flare and the exhaust cone, all of these parts are m a d e in
aluminium. T h e blades are normally die cast, Fig. 2 shows a typical turbine
160
MODEL TKJTINIi ()!• T U R H I N E S AND COMPRESSORS
Fig. I.
Three-stage modil turbine
blade in the " a s cast" condition a n d as a finished piece with machined root
and tip shrouds. This method of manufacture and the " T " type root fixing
were chosen to facilitate tests of restaggered or repitched blading.
Test Equipment
T h e layout of the rig compressors and control valves are shown in Fig. 3.
T h e 6000 h.p. plant has an open circuit, blow-suck cycle, the air being
compressed by two 1750 h . p . compressors a n d supplied at the turbine at
pressures u p to 4-0 a t m a n d at a temperature which can be varied between
30 and 200°C. T h e turbine exhausts to a sub-atmospheric pressure of \ a t m
and from this pressure the air is recomprcsscd by two 1250 h.p. compressors.
T h e power for the rig compressors is supplied by synchronous electric
161
S. F . SMITH AND C. E. PAYNE
motors and the control of the turbine pressures are achieved by the use of
blow-off and throttle valves.
T h e compression ratio of the two blowing compressors was chosen to give
a m a x i m u m turbine inlet temperature of 200°C which is just sufficient to
prevent icing of the turbine outlet at the highest expansion ratios. T h e air
for the turbine tests is filtered through two banks of filters; this is essential
FILTER HOUSE,
BLEED OUT
AIRMETEH
11 np y^—
- » B L E E O IN » - FROM TURBINE
Pig. 3.
Compressor house
as the performance of model turbines deteriorates quite markedly if dirt is
allowed to accrete on the blades. A loss of 3 per cent of turbine efficiency
has been recorded after less than 2 hr running with unfiltered air. A similar
loss has been observed if the turbine outlet temperature falls below freezing
a n d ice forms on the turbine blading.
Turbine Test Cells

Figure 4 shows the layout of the turbine test cells and control room. This
layout is fairly typical of the latest test bed practice, and the controls and
instrument reading points are all within the central sound proof control room.
Instrumentation
T h e instrumentation of the turbine is deliberately simple and straight-
forward and calls for little comment. I n general we prefer to calculate our
total pressures from static pressures measured in the jet pipe or in a parallel
annulus at the turbine outlet, as our experience of direct measurement of the
turbine outlet total pressure close to the turbine shows this leads to a rather
optimistic estimate of turbine efficiency.
T h e turbine temperature d r o p is derived from the turbine shaft power
output which is measured both by a n optical torquemeter a n d from a
measurement of the brake torque. These readings give values of power
normally within £ per cent, the torquemeter reading being slightly the
higher. Although the turbine outlet temperature is measured this is done to
162
MODEL TESTING OF TURBINES AND COMPRESSORS
check quickly that the turbine is not icing; the efficiency derived from the
turbine outlet temperature measurement is discounted as it is almost always
higher and less consistent than the value given by the power output. This
may be due to "wet b u l b " effects caused by condensation or the presence of
traces of oil in the turbine air.
S ' "»4*.
Fig. 4.
Turbine rig
Correlation of Model and Full Scale Results
There is an appreciable change in the physical properties of air between
the relatively low temperatures at which model turbines are tested and the
high temperature at which the full scale engine turbines operate. If the
model results are to read across satisfactorily to the engine we must have
some way of allowing for this, as well as for the effects of Reynolds number
and the differences between model and engine bearing and windage losses.
Reynolds number.—The size of our model turbines and the pressures a n d
temperatures at which they are tested give Reynolds numbers which lie
within the range through which an engine turbine operates from sea level
to its cruising altitude. The lowesl tin bine Reynolds number at which any
of our engines are normally operated is about 40,000 and we have tested a
model turbine down to this value of Reynolds number without detecting any
change at all in the turbine performance. It does, therefore, appear that,
within the range in which we are interested, the turbine Reynolds number is
not an important parameter.
Disc windage and bearing loss.—The air surrounding the discs of a model
turbine is at substantially the same density as the main gas stream, whereas
in an engine turbine the disc air is at a very much lower temperature and
higher density than the main gas stream. This will mean that in the model
the disc windage losses will be a smaller proportion of the total power output.
For a given disc we have
Windage power oc = - U 3
1
n
and Turbine power cc p s j T T
so that Windage power oc
163
S. F . SMITH AND C. E . PAYNE
when T T is the temperature of the turbine gas

T D is the temperature of the turbine air surrounding the disc
U is the rim speed.
Now the value of U j \ / T will be the same for the model and engine turbines
so that the engine turbine windage will be between 2 and 3 times greater
than that of the model. We have measured the windage power of model
turbine discs and for a typical turbine the windage power is of the order of
j\j per cent of the turbine power output. For engine turbines then the windage
power is of the order of f-g- ^ per cent of the turbine power.
T h e bearings of a model turbine are deliberately lightly loaded and, for
the same size of turbine, the model turbine bearings have roughly the same
dimensions as an engine turbine, even though the rotational speed and
power output of the model turbine are much less. For identical model and
engine turbines and bearings we have
Bearing power input oc N 2
Turbine power o u t p u t oc P y / T
Bearing power i N \2 WT
so that bearing loss = =—r-r- - QC {—7=,} X ~ 5 ~
l u r b m e power \WT] P
and for the same value of Njyf T for model and engine
Model bearing loss T model P engine
Engine bearing loss T engine P model
T h e air pressure in a model turbine will be approximately one quarter
of that in an engine and the temperature one third, so that the model
bearing loss will be two or three times that of an engine turbine. From our
measurements an average value of model turbine bearing loss is ^ per cent
and a comparative engine value would be about ^ per cent.
Taking the bearing and windage loss together we find that rather
conveniently for most of our turbines the sum of these losses is the same for
both engine and model turbine and there is no correction necessary to the
model results.
Specific Heat Ratio

T o obtain the same values of incidence and M a c h number in the model
turbine we test at the engine values of u j \ ' T and {Cp A T ) / T , where a is the
blade speed, T t h c turbine inlet total temperature, A T the temperature drop
and Cp the specific heat. T h e model turbines are also made geometrically
similar to the hot engine turbines.
This method is acceptably simple to use and gives very nearly identical
values of incidence and M a c h n u m b e r for the engine and model turbines.
T o illustrate this Fig. 5 shows the velocity triangles computed for two
identical turbines, one using gas of specific heat ratio y = 1-33 and the
other for cold air y = 1-4. As will be seen the incidences and M a c h numbers
are very close indeed and the small difference will have no detectable effect
on the turbine efficiency. Also important is the ratio of the stage inlet and
outlet capacity, since in a multi-stage turbine this must be correct to give the
tight work distribution on each stage. Here again the same u f \ / T and
164
{Cp A T ) j T method of comparison is very good: in the particular case

considered the error is only j 1 ,, per cent.
In order to confirm that the performance of an engine turbine can be
accurately predicted in this way from model turbine test data, we have, for
several different engine types, carried out detailed analyses of engine lest
data to determine the performance of the turbine. In general we have found
that there is very good agreement between the model and engine data.
\~rix Tuaaiwe
OUTLET j^yj
J O
INLET 5i^ ' '
COLD TURBINE
A-- OUTLCT M/T
Pig. 5.
Comparison of velocity triangles al the same uj \ T and (Cp \ T ) f T
Conclusions
T h e general conclusion that we have reached after 14 years' experience of
model turbine testing is that this is probably the only satisfactory way in
which it is practicable to carry out a comprehensive investigation of the
aerodynamic characteristics of turbines. T h e large body of information,
which has been built up at Rolls-Royce, has been acquired at a cost which
is very small indeed in comparison with the total cost of engine development.
PART I I : SCALE MODEL COMPRESSOR TESTING

By C. E. PAVNF.
.As a general principle, scale model compressor testing has been restricted
at Rolls-Royce to research investigations and the testing of new principles of
compressor design. With few exceptions the development of the compressor
for an existing engine or for direct application to a project has been carried
out on a compressor of engine or near-engine size.
T h e advantages of carrying out a research programme on small-scale
compressors of similar size fall into two distinct categories. Firstly there are
the obvious advantages associated with small size. Apart from the reduced
cost of manufacture of many of the compressor parts, these include the
economies accruing from the correspondingly reduced demands on the size
and power consumption of the plant used to drive the compressor. Probably
of greater significance in the long run, however, are the advantages which
165
S. F. SMITH AND C. E. PAYNE
derive from the similarity in size of the compressors tested. T h e use of a

more or less constant compressor diameter enables many major components
to be m a d e common to a series of compressors, so effecting a considerable
economy in manufacture. T h e r e remains a further advantage, of particular
importance in the carrying out of a research programme. This is that
constancy of size, method of construction and proportional effect of m a n u -
facturing tolerances make for validity of comparison of the aerodynamic
performance of one compressor with another.
T h e rig upon which the majority of model compressor tests take place is
illustrated diagrammatically in Fig. 6. Power is derived from 5000 h . p . d.c.
motors through step-up gearboxes and the arrangement is too straight-
forward and familiar to w a r r a n t detailed description. Fig. 6 shows the
A^W-'K'AV.WA'A'A'A'AT
Fig. 6.
Arrangement of rig for testing scale model compressors
stations at which measurements are m a d e externally to the compressor itself.

These include air inlet temperature and mass flow measured at the airmeter,
and inlet pressure measured in the settling tank downstream of the inlet
throttle valve. I n p u t torque is measured by means of an optical torquemeter
which forms the final drive m e m b e r to the compressor. T h e instruments
used for measuring air outlet a n d inter-stage conditions are considered as
p a r t of the compressor itself and the type of instrument and the position it
occupies vary with the compressor and nature of test. Fig. 7, which shows
the high pressure stage and outlet section of a 5-stage model compressor,
includes instrumentation typical of that used for measurement of compressor
delivery conditions in a general test. Outlet static pressure is measured at
tappings on the inner a n d outer walls of a parallel annular extension to the
compressor outlet. T h e values of compressor delivery total pressure normally
used in the analysis and presentation of our results are those estimated from
the static pressures in conjunction with the measured mass flows and
temperatures. Downstream of the plane of static pressure measurement are
located forward-facing stagnation pockets, usually 8 in number. These are
arranged in the same axial plane but are disposed radially so as to span the
166
MODEL TESTING OF T U R B I N E S AND COMPRESSORS
compressor exit annulus. In each of them is fitted a resistance thermometer

and total-head tube. As well as providing average values of the compressor
delivery temperature and pressure (the latter used as a check), these
instruments provide an approximate idea of the radial variation of compressor
outlet conditions. A more detailed survey of the profile of compressor delivery
total-head, if required, is obtained by radial and circumferential traverses
OUTLET TOTAL 4 STATIC

, PRESSURE t FLOW
' DIRECTION
tTRAVIRS»4G YAWMETER) ;
Fig. 7.
Model compressor, measurement of compressor delivery condition
with a cylindrical total-head yawmeter also calibrated for the measurement

of static pressure. Inter-stage pressures are measured at static pressure
tappings on the compressor outer annulus in planes just downstream of the
stator trailing edges.
Turning now to the construction of the compressor itself, Fig. 8 shows a
^sssffli:'...^
Fig. 8.
Five-stage model compressor
sectioned view of a typical 5-stage high capacity compressor of about 3-5

compression ratio. T h e main compressor casing, the shaft, and the front and
rear bearing housings are all " p e r m a n e n t " components of the rig—that is,
they are used for testing all compressors for which their length is appropriate.
A shorter casing and shaft are available for use when the overall compressor
length is less that 12 in.—low pressure ratio blowers and high pressure units
167
S. P. SMITH AND C E. PAYNE
of high hub/tip ratio blading with correspondingly short chords. High

pressure ratio compressors of a length exceeding 18 in. are tested in a rig
comprising a longer casing and a modified shaft whose rear bearing is
supported within the rotating assembly. This arrangement avoids the
possible whirling difficulties associated with an excessively long compressor
shaft. T h e compressor blades are machined in aluminium or a l u m i n i u m -
bronze alloy in the case of high temperature stages and their usual method
of attachment can be seen clearly on Fig. 8. T h e rotor blades have a T-shaped
root form and are clamped between discs whose rims are soft-plated in copper
to allow for slight variations in blade root thickness. Each rotor stage is
individually assembled and splined onto the compressor shaft. .As well as
allowing easy variation of the n u m b e r of blades in any row, by the inter-
position of packing pieces between blades, this arrangement means that each
disc can accommodate blades over a range of root diameter and so a limited
n u m b e r of disc sizes are required to cover all designs of compressor. T h e only-
components of the rotating assembly which are peculiar to any one design
are the spacers between the discs which, together with the blade platforms,
form the inner annulus of the compressor. Stator blades are fitted to half-
rings which are keyed to the compressor casing. These rings, like the disc
spacers, form the annulus of the compressor and so require replacement for
any design change which involves modification of the compressor annulus
dimensions.
Not unreasonably, one of the earliest experiments carried out on the model
compressor rig was an investigation of the effect of size on the measured
performance of an axial compressor. An approximately J-scale model was
m a d e of a compressor which was in use for compressor development for an
engine. Considerable care was taken to scale all axial and radial clearances
and to retain all the features of mechanical design which might possibly
affect the performance—as for example the method of stator blade shrouding.
T h e scaled compressor was then tested at the same corrected speeds as the
larger compressor and at the same inlet pressure, thus halving the Reynolds
number. T h e results are plotted in Fig. 9, the mass flow function of the large
compressor being scaled lo the model compressor size. At the compressor
design speed, the scale compressor shows a loss of efficiency of 2 per cent
together with a reduction in capacity. T h e scale compressor was then
re-tested at a higher inlet pressure and Reynolds number. 11 was unfortunately
not possible to achieve the same value of inlet Reynolds n u m b e r as that at
which the large compressor had been tested. However, by extrapolating the
results at the two values of R n it is clear that, at the same Reynolds number,
the efficiency of the two compressors is sensibly the same. From the point
of view of the engine user, a 50 per cent reduction in size (from 30 to 15 in.
diameter) is accompanied at the same inlet conditions by a 2 per cent loss in
efficiency. Similarly extrapolating the values of compressor capacity in
terms of mass flow/frontal area it is seen that the capacity of the scale
compressor falls short of that of the larger compressor at the same Reynolds
number.
O t h e r early work for which the model compressor rig was used included
a comparison of the performance of compressors of 4, 5 and 7 stages, all being
designed for the same pressure ratio, to accept the same air mass flow and
168
MODEL TESTING OP T U R B I N E S AND COMPRESSORS
with the same inlet and outlet annulus dimensions and rotational speed.
T h e results of these tests indicated the existence of an optimum compressor
stage loading—the 5-stage compressor being more efficient than the more
heavily-loaded 4-stage or the lighter-loaded 7-stage unit.
Subsequently an investigation was made of the effects of variation of blade
camber and space/chord ratio and of the extent to which high blade camber
could be exchanged for low pitch/chord ratio—as indicated by cascade tests.
These tests proved the adequacy of the correlated cascade data in use for
design purposes, al least as a means of determining the effects of changes in
blade space/chord and camber. T h e results of another series of tests enabled
comparison of the effects on compressor matching of change of blade
geometry with change of annulus area.
V \
«s-
-S
IAAU. F i n . FUNCTION M f ***eM.tvm*
7-*
TV H A M - M t t M
*-"
TV OM4- l&OOM (A
A3
——— •it • V
C4 O 1s aO 2a "
g NO M
Fig. 9.
Effect of size on compressor performance, I lOOfljsec lip speed
More recently, basically different types of compressor design have been

manufactured and tested on the model compressor rig. Included in these
was a design of high pressure compressor in which constant annulus height
was maintained through the rear half. T h e object was to attempt to improve
efficiency by maintaining a reasonably low hub/tip ratio in the rear stages
thereby keeping lip clearances small in relation to the blade height. We were
concerned to measure the effect on performance of designing for the very
low and falling axial velocities which result from the annulus configuration.
Additionally we have tested and are testing different designs of low pressure
stage to increase the flow for a given compressor frontal area. T h e distribution
of aspect ratio and stage loading required to secure the maximum pressure
ratio from a compressor of given length and weight has also been the subject
of a number of tests.
In addition to research on subsonic axial compressors, some supersonic
compressor testing has also been carried out on the model compressor rig
although this programme has now been concluded.
169
S. F . SMITH AND C E . PAYNE
In all this work it would be hard to over-stress the advantage of not having
comparison between the various compressors complicated b y changes of size,
mechanical design, a n d method of testing.
SOMMAIRE
Les auteurs passent en revue les differents points d'interet des essais sur modele
comme moyen d'evaluer les caracteristiques aerodynamiques des turbines et des
compresseurs et eludient la correlation entre les resultats oblenus sur maquettes et
sur les machines de vraie grandeur. L'expose comprend aussi une description de la
construction mecanique des modeles, de la disposition de Vinstallation d'essai el de
la technique experimenlale developpee par Rolls-Royce.
En conclusion, les maquettes a echelle reduite de compresseurs utilisees pour les
recherches initiates representent un complement tres utile pour V etude des compresseurs
de vraie grandeur. Les essais des modeles de turbine ont un si grand avantage sur
les essais fails avec des turbines du moteur complet que nnus pensons que c'est le seul
moyen economique d'entreprendre un programme de recherche et de developpement
aerodynamique des turbines.
DISCUSSION
H. SCHLICIITIM; and K. GERSTEN: (Aerodynamische Versuchsanstalt Gottingen und
Technische Hochschulc Braunschweig). The model tests of compressors and turbines as
described by Smith are carried out in such a way that the Maeh number has the same value in
the model and the full scale machine. This means that the Reynolds number is different in
both cases. With regard to the influence of Reynolds number the following results were
obtained by Smith:
1. for turbines Smith slated, that there is not much influence of the Reynolds number, if it
does not go below 40,000.
2. for compressors Payne stated, that there is a loss of maximum efficiency of about 2 per
cent, if the Reynolds number gets down to half its value.
The following remarks on the influence of Reynolds number are based on theoretical and
experimental research of cascades and also on tests of a one-stage axial-flow compressor,
carried out at the Technical University Braunschweig and the Aerodynamischc Versuch-
sanstalt (AVA) Gotlingen in recent years.
Reynolds-Number f\g Reynolds-Number f>g=S^.

V
supercritical
Fig. 10.
Influence of Reynolds number on lift and drag coefficient of a single airfoil.
Profile Go 797, thickness ratio t)c = 0 1 6 , angle of attack a = 8 °
170
MODEL TESTING OF T U R B I N E S AND COMPRESSORS
1. Reynolds Number Effect

For a single airfoil and also for cascades of turbomacbincs it is a well-known fact that the
flow becomes very bad if the Reynolds number is decreased below a certain critical Reynolds
number (CHARWAT, 1957; DAVIS et al, 1951; RIIODEN, 1956; SCHMITZ, 1953). This is
demonstrated in Fig. 10 for a single aerofoil. 'Flic lift coefficient drops very rapidly and the
drag coefficient increases very suddenly, if the Reynolds number gets below the critical
Reynolds number, which in this case is about Re * 150,000. In the subcritical range, e.g. at
low Reynolds numbers (Fig. 10a), the boundary- layer is laminar and separates immediately
behind the point of minimum pressure. This gives a large separated area with strong increase
oi drag and decrease of lilt. In the supercritical range (l"ig. 10b) there is transition in the
boundary layer from laminar to turbulent flow, and thus separation is avoided.
This occurs also on blades of cascades.
2. Calculation of Reynolds Number Effect in the Supercritical Range

This Reynolds number effect on the aerodynamic parameters of a cascade is, of course, of
fundamental importance, if the results of experimental investigations on a model are trans-
ferred to the full scale machine. By theoretical methods this transfer is only possible, if for
2
d/c • 10
p B .so"
P,m05*
Theory (with separation)
° Measurements
Fig. I I .
Loss coefficient £ K2 = A i j / i / m of a turbine cascade against Reynolds number Re w^c/v.
Solidity ratio cjd = 1-0, blade angle fi B = 50°, angle of inflow p", = 1 3 5 "
theory (taking into account small trailing edge separation)
measurements
both cases (full scale machine mid model) the Mow is supercritical. In this case the influence
of Reynolds number on the aerodynamic parameters, especially on the loss coefficients, can
be calculated theoretically with good accuracy (GERSTEN, 1959). In recent years my colla-
borators SPEIDEL and SCHOLZ (1957) have worked out a method to calculate all aero-
dynamic parameters ofa two-dimensional cascade from boundary layer theory, and thus also
to calculate the influence of Reynolds number on the two-dimensional cascade flow.
Good agreemenl of lliis theory with experimental results was obtained.
In Fig. 11 the influence of Reynolds number on the loss coefficient ofa turbine cascade in
two-dimensional flow is given.* The theoretical curve includes losses due to friction in the
boundary layer, losses in the wake and also some losses due to separation near the trailing
edge. The agreemenl between theory and measurements, which cover a rather wide range of
Reynolds numbers, is very good. But these two-dimensional losses are only one part of the
total losses in a turbomachinc.
* In these comments the Reynolds numbers are based on the blade chord c and on the exit
velocity ui. of the cascade. In the paper by Smith the Reynolds number is based on the axial
chord c . sin ji B and the exit velocity «',. But this difference in defining the Reynolds number
does not affect the conclusion of this paper.
171
S. F. SMITH AND C. E. PAYNE
There are always considerable additional losses at the side walls and at the clearances,
which are caused by secondary flow effects. We have recently investigated the influence of
Reynolds number on these secondary flow losses and Fig. 12 gives some results on this. Here
102<y.
Fig. 12.
Influence of Reynolds number on secondary flow losses of a turbine cascade. Solidity ratio
cjd = 1-33, blade angle fiB = 55", angle of inflow /j", = 90", h = blade length,
s = gap width.
1 h - oo
2 hjc = 2-5
3 h\c = 2-5; sjh = 0 0 2 .
the two-dimensional loss coefficient " 1 " and the total loss coefficient is plotted against the
Reynolds number for a turbine cascade for both cases, cascade with sidewall " 2 , " and cascade
with clearances " 3 . " ll is very remarkable that both additional losses due in side-wall and due
to clearance, are independent of the Reynolds number.
3. Influence of Rotation on Behaviour of Boundary Layer and on Critical

Reynolds number
When using the results of cascade experiments for the design ofa rotor ofa turbomachine a
fundamental question, which has not been investigated very much, is how the results of the
cascade experiments are affected by the influence of centrifugal and Coriolis fones. A
contribution to this problem has recently been given by MUESMANN (1958) in the Aerodyna-
mische Versuchsanstalt Gdttingen. Using extensive lift and drag measurements on single
airfoils in a very large range of Reynolds numbers, Muesmann calculated the characteristic
curves of a one-stage axial flow compressor, that means the efficiency coefficient and the
coefficient of total head against mass flow coefficient, taking into account the considerable
variation of Reynolds number along die radius of the rotor. In Fig. 13 these calculated curves
are compared with measurements which were also done by Mucsmann. The two diagram!
on the left-hand side refer to the high Reynolds number Re = R'ojjv — (i4.(U)(Cbnd the two
diagrams on the right-hand side refer to the low Reynolds number Re - 21,000(3 For both
Reynolds numbers local values of efficiency and pressure coefficient are given in the upper
diagrams and total values in the lower diagrams. On comparing the calculated curves and
the measurements, the following important statement can be made:
For the high Reynolds number the agreement between calculation and measurement is
very good. In this case the flow is supercritical on the rotor as well as on the single airfoil.
For the low Reynolds number, however, there arc considerable deviations between calculated
and measured values, especially for the local values near to the hub. From detailed investi-
gations it was found that these deviations are due to the fact that the transition is considerably-
affected by centrifugal and Coriolis forces (GREGORY et al, 1955).
172
O n t h e r o t a t i n g blade t h e critical R e y n o l d s n u m b e r is always considerably smaller t h a n o n

s t a t i o n a r y cascades. As a c o n s e q u e n c e of this, there exists a c e r t a i n r a n g e of low R e y n o l d s
n u m b e r s , in which in t h e stationary cascade t h e flow is subcrilical a n d thus has a b a d effi-
ciency d u e to l a m i n a r s e p a r a t i o n . But in t h e rotor t h e flow is still supercritical for this r a n g e
of R e y n o l d s n u m b e r s a n d thus has good efficiency, because separation does n o t occur d u e to
p r e m a t u r e transition caused by centrifugal a n d Goriolis forces.
rre-^rg^ftOOO
n-SOO U/min
Jmf%
1.0
r-*-
T
calculation Z
1~* measurements
I : subcrit/cat 0.2 - E ^
I : supercritical
I .1 0.2 0.3 0 q«
Fig. 13.
Pressure coefficient and efficiency of an axial flow compressor;
measurements of Muesmann (1958).
y> = pressure coefficient
>j = efficiency
rp • • massflow coefficient
Vioc = tocal pressure coefficient
»?loc = local efficiency
n = number of revolutions
g = total head
Q = Rate of flow
N = power
T h i s result c a n also be p u t this w a y :

T h e investigations on stationary cascades in the supercritical r a n g e of Reynolds n u m b e r s c a n ,
to a c e r t a i n extent, b e a p p l i e d in the r o t a t i n g m a c h i n e for such low R e y n o l d s n u m b e r s , w h e r e
the flow would be subcritical in t h e stationary cascade.
Thus I c o m e to the following conclusions:
W i t h regard to t h e influence of R e y n o l d s n u m b e r o n t h e m o d e l tests of compressors a n d
turbines t h e following statements c a n be m a d e :
1. F o r supercritical flow t h e influence of Reynolds n u m b e r o n t h e loss coefficient c a n be
calculated with good a c c u r a c y from b o u n d a r y layer theory.
2. Centrifugal forces a n d Coriolis forces to a certain extent h a v e a favourable effect, e.g.
for a certain r a n g e of R e y n o l d s n u m b e r s they give supercritical flow for t h e rotor,
whereas t h e flow on t h e single aerofoil a n d t h e stationary cascade is subcritical.
REFERENCES
C H A R W A T , A . F . d . Aero. Sci. 2 4 , p p . 3 8 6 - 3 8 8 , 1957.

DAVIS, H . , KOTTAS, H . and M O O D Y , A. M . G. " T h e I n f l u e n c e of R e y n o l d s N u m b e r o n t h e
Performance of T u r b o m a c h i n e r y " Trans. A . S . M . E . 73, p p . 4 9 9 - 5 0 9 , 1951.
G E R S T E N , K . " D e r F.influss d e r Rcynoldsschen Z a h l a u f d i e Stromungsverluste in e b e n e n
Schaufelgittern" T o b e published in Abhandlungen der Braunschweigischen Wissenschqftlichen
Gesellschafl, 1959.
G R E G O R Y , N . , S T U A R T , J . T . , a n d W A L K E R , W . S. " O n t h e S t a b i l i t y of T h r e e - d i m e n s i o n a l
B o u n d a r y Layers with Application t o t h e Flow d u e to a R o t a t i n g Disk" Phil. Trans. Roy.
Soc., London A 248. p p . 155-199, 1955.
173
S. F . SMITH AND C E. PAYNE
MUESMANN, G. "Zusammenhang der Stromungseigenschaften des Laufradcs d n a Axial-

gebliises mil i bmii eines Einzrllliigels" Z. f. Flugwissensch. C (1958), pp. 345-362.
RIIODF.N, H. G. "Effects of Reynolds Number on the Flow of Air Through a Cascade of
Compressor Blades" ARC-Rep. 2919, 1956.
ScHMITZ, F'. W. "Zur Aerodynamik der kleincn Reynolds-Zahlen" Jahrbuch der WGL,
pp. 149-166. 1953.
SPEIDEL, L. and SCHOLZ, N. "Untersuchungen iiber die Stromungsverluste in ebenen
Schaufelgittern" VDI-Forschungsheft 464, 1957.
P. P H . CHAFFIOTTE (Societe d'Exploitation des Materiels Hispano Suiza): Essais de

turbines La communication de Smith apporte des renseigncments precicux sur une technique
d'adoption relativement recente en F'rance.
En effet, au moins en ce qui concerne les gros moteurs, pendant longtemps les programmes
fran;ais n'ont ported que sur des turbines k un seul etage. On a juge. peut-etrc a tort d'ailleurs,
que FexpeHmentation directe etait preferable a cause de 1'impossibilite ou Ton sc trouvait de
realiscr sur un modele les memes distributions de vitesses et de temperatures, a un facleur de
similitude pres, que sur la turbine reelle.
Les essais sur turbine reelle visaient a donner les cartes de temperature, de pression totale
et de vitesse a la sortie des roues, afin de verifier que les hypotheses de calcul etaient
confirmees.
Au contraire, lorsque le programme de developpemcnt des turbines a plusieurs etages prit
de 1'extension en F'rancc, il devint Evident que la methodc directc nc suffisait plus. On cut
alors recours a des bancs d'essais de "turbines froides". Les machines reelles etant petiles,
on n'a pas change 1'echelle. On s'est borne k rcmplacer le metal refractaire par de 1'alu-
minium. Les aubes sont taillees directement dans la masse et font corps avec les roues,
comme les turbines "chaudes" dont elles utiliscnt I'otitillage de fabrication.
La Fig. 14 donnc 1'excmple d'une de ces turbines d'essais essayed montee sur une am ii-une
installation d'atmospherisation de motcurs a pistons.
Les machines d'exlraction ont une puissance totale de 900 ch. seulement et donncnt un
rapport dc detente voisin de 6 pour un debit dc 1,5 kg/sec suffisant pour essayer avec aspiration
a la pression atmospherique des turbines dont le diametre du premier etage est de 250 mm au
sommet d'aube. L'air d'admission petit subir plusieurs traitements illustres par la Fig. 15 et
qui, tous, tendent a eviter le givragc:
(1) Utilisation des dessiccateurs d'air d'admission ct admission a faible temperature.
(2) Admission dirccte, apres detente par vannage, de Fair provenant des compresseurs
du groupe frigorifique, de maniere k profiler de Fechauffement du a la compression.
(3) Rexhauffagc par un echangeur edectrique d'une centainc de kW.
Cette troisiemc solution est la plus souple et aussi cclle que nous utilisons de preference.
Mais une combinaison 1 et 3 serait probablcmcnt la plus intenfssante car elle elimiiu- preaqUC
totalemenl 1'eau et rend plus sures les mesurcs de temperalure. L'interet d'utiliser de Fair
filtre nuns ;t echappe, et ['experience de Smith sur 1'influence des impuretes est originale el a
retcnir.
En 1'absence d'un couplemetre cale directement sur 1'arbre de la turbine, on calcule le
rendemcnl a parlir du rapport de detente des pressions totalcs amont et aval, la pression
totale aval etant deduitc de la prcssion statique, du debit mesure, dc la temperature et de la
section. On attache une certaine importance a Fangle de sortie des roues, et c'est pourquoi
on fait une exploration directionnelle etagc par etage en essayant successivement et separe-
ment chacun des etages.
Essais de compresseurs.—Contrairement k la politique mcnlionnee ci-dessus pour Irs essais
de turbines, on a travailie tres tot sur des maquettes de compresscurs et construit des machines
spedalcs d'essai a un ou plusieurs etages; 1'une de ces machines comportait 4 etages dont
toutes les aubes mobiles ou fixes pouvaient etre decalees a Farret.
Ce comprcsscur de petites dimensions a ete d'une grande utilite il y a plusieurs annexes pour
la determination des coefficients qui permettent de passer des essais de grilles d'aubes a la
couronne d'aubes reelles; mais on n'a pas con$u a proprcment parler de maquctte a echclle
reduitc de compresseur red.
On a prefere, par exemple, realiscr un compresseur d'etudes identique k celui de la
machine rcelle en projet, mais avee des aubes de stator orientables a 1'arrct.
Pour faciliter 1'etude de compresscurs supersoniques, on a mis recemment en exploitation
174
Plan d'exploration
pression totale
d'rection, temperature
Prises pression
statique
Thermometre
Fig. 14.
Hunt de turbines froides. Mm/age de In turbine "Bastan'
Filtr-e
1 Dispositif Extrocteur
Compresses Antipompoge
du cornpresseur
XrTK fcX : • :
Vonrw de regiage
fngerant j ' I L ^ y P " Jyj^-| du taux de detente
P
1 Refrigerant *&.n..
Purge f» ^ J Turbine _
Vannc de regiage
du toux de detente
Vanne de regloge
de taux de detente
Chauffage par T-l.Y ia j\ ar Turbine

resistances
Fig. 15.
Schemas de principe du banc de turbines froides
175
9LI
•Suiiqaqqai,- SE u o j s n p u o a auiBS aqj X|pBojq 01 sn

p.>| a.vsq s q n s a j j q x sapEjq .iqi j o uiE3Jisdn jsnfaznES E 8uiAEq Xq assajaui SSO| apEjq .ipEJSEj
oq) iu.)A,)jd 01 3|qE u a a q aAEq 3fr\ aauEUiJojjad ui sso] ou SMoqs a u i q j n j p p o i u 8ui)Ei<u
E JOJ ElEp jssj aqj 'j3A.)Moq '((OO'Ot UEqj j a q S i q j a q t u n u sp|OuXay j o san|EA JB SSO| ,>pB|q
a q j ui sasEdjoui 3|qEp.)jddE pa.viasqo aAEq w apsasBa a u i q j n j E U ] - a a X o y - s u o y JE p.iurejqo
uaaq SBq qaiqM EjEp qipv 1u.uuaa.18B |EjauaS ui si S3pE|q jossajduioa puE a u i q j n j j o j a q u i n u
s p p u X a y (Baijija aqj u o UOIJBJOJ J O a a u a n g u t aqj u o JJJOM Supsajajui jsoui s ^ u i j t p q q a s
s u p iii.ii|iiii.i oj spuaj aauBiujojjad a u j q j n j [apoui puE aui8ua uaaMjaq
in.mi.i,ii^r p-j.iu.iS p o o 8 aqj_ "aauBuiJojjad a u i q j n i j o ssoj o u a.\i3 oj qjooius Xpuaiaujns OSJE
SI ajq apE|q a|qEidaaaE UE sa.\i8 qaiqM a]qojd ajnjBjaduuaj B jEqj pajEaipui siqj puE ' a | y o j d
,i.mii!.i.idui.ij j o o d E j o jaajja aqj u o EjEp auiSua l E j u a i u u a d x a JO JUIIOUIE []BUJS E 'jaAaMoq
•a.\Eq a \ \ -aJUBUijojjad a u i q j n j u o sapjojd X i p o p A puB ajnjEjaduiaj j o jaajya aqi a u i u u . n . i p
oj siu.nuij.idx.i a u i q j n j p p o t u XUE j n o p a u j s a 'jaX SE 'JOU aABq ,)M aaXo-sj-s||oy i y -aujqanj
.niitiu.i UE Xq p a a u a u a d x a ,JJE qaiqM sajqojd X i p o p A puE ajnjEjaduiaj jajui aqj ajE|nuiis
jou p i p a a X o y - s | i o y JB pasn s a u i q j n j p p o u i aqi JBqi j n o pajuiod a i j o y j E q g ni.i.iwt; -a; •$
ATday s.MOHxnv
Y561 VEJAJ '6/; - \ o \ ' - g ' W S ' V s u m X . . J - > t | l o u V °* S B O - > u O UJOJ .l a a u E u u o j j a j jotaaidoiOQ
j o suoijE|SUEJJ_ aqj ui p a j a j u n o a u g s u i a | q o j , j , . • \ \ - \ \ ' x o o i i . \ \ puB f -TAJ ' N N V H U M V H
aDNauaaa^
•jnassajdiuoa a[ sajdB ja JUEAB j u a u i a i n o a a j a p sjijEjuasajdaj
uaiq spjojd sap j i u a j q o E UO6BJ a p sjnajaas saSiB) a p j n s saa.vapj 'sajnjBjadiuaj ja suoissajd
sap aauEssiBuuoa B] jEd s n u a j q o j u a j n j s j u a u i a p u a j s a g -saadnoaaj a p sBd j u o ( u saauBssind
s.i| 'UOIJEIIBJSUIJ j n s ajjaiuajdnoa a p aauasqB,] u g "sajnsaui a p s j n a j j a sap .inai.i.um i
E juos siuauiapuaa sa] ja sjiqap sa[ 'uoissajd a p s j j o d d E j saj j n s s n u a j q o sjJEaa s a g
Z E 8 sap aiuBisuoa E| S J —
s a n b q p a d s sjnajEqa sap j j o d d s j a| / . —
uoijEJidsB.i E ajnjEjaduiaj ja uoissajd a p suoijipuoa sa| j _ ja d
assEUi i i q a p uos £)—
.i.l.ipIMl. i.i ZES 3 | SUEp UOS n p .ill.I.i|.l,l E[ E
.i|iqoui n p s a n b u a q d u a d sassaiiA S3] f) —
•ZE8 aj juEuSisap saaipui sap juos £ ja ] — : s . » | p n b s 3 | suBp
(!¥•
X)„X) X)
•(ITS) 'X 'X
:suoiiB|3J sa| ju3ui3|diuis jUEAJasqo
ua s n s s a p - p ZES x n a p sa| M M : jnassajduioa amain u n j n s s n u 3 j q o sjE]|nsaj S3p u o u B p j J o a
a i u a | | a a x a a u n jsrjua.v n d E U Q " ( i g g l ) x o o i r \ \ ja uNVKLtivp, JBd ajiEj aja E i B s s a ^ Z E 8
a[ 33AB s n u a i q o sjBimsaj sap |aaj ZBS ire uoinsodsuBJi a p suoijipuoa sap apnja a u u o q a u r \
'sa[qiBj saauBssind sap .I.IAI; l.l sa|qEuuosiEj sanbiuEaam sassajiA sap
B 'aipEj a u u o q a p s a u u p E u i sap JUS J,)(]IEAEJJ a u o p j n a d u g sjaij n p snjd n a d u n ; p ' p u o a a s
a| s u E p — J I E ^ a p a | p a a p aijiouj E| a p SUIOUJ n a d u n ^ j jsa j a i u i a j d .>\ s u s p uos n p assajiA B^]
' * J 5 I O z O auEqjaojon[)EJiaj
-ojo[qaip n p n o E J * I 0 3 auEqjauiojon[jqx)jo|qaip n p s p i a j p 3 | p n b E ] suEp UOIJE||EJSUI a u n
-
3NAVd a -o QNV H1IWS "J 'S
SOME EXPERIENCES IN COMBUSTION SCALING
By ARTHUR H. LEFEBVRE and GORDON A. HALLS
Rolls-Royce Ltd. Derby, England
The primary object of this paper is to survey the state of development of combustion
scaling in the gas turbine fields. It is written entirelyfrom an engineering viewpoint
with special reference to the problems which have emerged during the past six years
in the practical application of scaling principles.
It is suggested that scaling of combustion chambers is not a worthwhile
proposition, mainly because the potential economies in operating cost are more than
offset by the difficulties and expense of manufacturing small components to the
required order of accuracy.
An equation is presented which relates cnmbustinn performance to the operating
variables of pressure, temperature and velocity and to the dimensions and specific
pressure loss of the system. It is considered that this equation provides a basis for
an appreciable relaxation of the scaling rules proposed by Stewart and should widen
the scope of their application. It is further suggested that this equalinn can be
applied to afterburners and combustion chambers in assessing the relative merits of
various designs, even when the available performance data relate to systems
differing in size and operating conditions.
In their application tn afterburner systems the scaling procedures recommended
by .Stewart proved almost cnmpletely successful right from the outset. The main
problem which has arisen has been the inability of the small scale models to
reproduce the high metal temperatures experienced on actual engines. In this paper
calculations of metal temperatures for both combustion chambers and afterburners
are included tn illustrate the order of temperature difference as a function nf the
scaling factor. Two possible methods of obtaining engine values of metal
temperature on the model are suggested.
INTRODUCTION
T H E BASIC requirements for the scaling of gas turbine combustion systems
were formulated by STEWART (1955) following a thorough and comprehensive
analysis of the fundamental component processes occurring in a combustion
chamber. T h e main conclusions of this work, which was carried out at the
X.G.T.E. between 1950 a n d 1952, were eventually published in the form of
a paper to the A G A R D Combustion Colloquium held in Liege in 1955.
During the same period active interest in the problems and potentialities of
scaling was also being shown in the U.S.A. and this interest is reflected in
the papers presented at the same Colloquium by W E L L E R (1955) and W A Y
(1955). I t might reasonably have been expected that these studies would
result in the widespread application of combustion scaling throughout the
gas turbine industry. In the event, however, this anticipated general
acceptance of scaling did not materialize.
At first sight this seems rather surprising in view of the tremendous
potentialities of scaling in effecting economies in power a n d in elucidating
177
A. H . L E F E B V R E AND G. A. HALLS
combustion phenomena. However, there is a major obstacle a n d this stems

from the primary role of scaling which is the simulation of very low com-
bustion pressures in systems which are actually operating at much more
convenient levels of pressure, i.e. at or above atmospheric. I n combustion
chambers the lowest operating pressures correspond to a few p.s.i.a. and
thus for scaling to b e fully effective it is necessary to have models which
are not larger than a q u a r t e r full-size. Unfortunately this requirement is
virtually prohibitive because the sizes of combustion chamber currently-
employed in gas turbines a r e such that the manufacture of any model
smaller than about half full-size is impractical, owing to the accuracy-
required to ensure exact geometric similarity. T h e scaling of fuel atomizers
presents the greatest difficulty, since the fine machining of even full-size
atomizers calls for great skill. T h e situation is further aggravated by
PROBERT'S evidence (1955) that atomizers should be scaled down relatively
more than the combustion c h a m b e r itself, a n d that several atomizers are
needed for any given combustion c h a m b e r to ensure accurate scaling over
the entire operating r a n g e .
Another contributory factor to t h e decline in interest in combustion
chamber scaling has been the development of easier a n d cheaper methods
of simulating low combustion pressures. O n e such method has been
described by LEFEBVRE a n d HALLS (1958) in which low pressures are
simulated by the addition of water to the combustion zone. This technique
has useful advantages in that no special models are needed a n d the pressure
ratio can be infinitely varied.
Some of the obstacles to the application of scaling principles, particularly
those relating to the scaling of the fuel injection processes, were anticipated
to a large extent by the early workers. They expressed the hope that certain
fuel spray factors would prove less important than a t first envisaged a n d that
the remaining processes might prove more amenable to scaling procedures
if only over a restricted range of operating conditions. It is suggested below
that in fact these hopes have largely materialized.
According to Stewart a small-scale model can be used to predict the
performance of a full-size c h a m b e r provided that the fuel, fuel-air ratio,
velocity a n d temperature are the same for both systems a n d that gas pressure
is inversely proportional to linear scale. I t is suggested here that some
relaxation of these rules is permissible. Specifically it is argued that
performance d a t a obtained from one combustion c h a m b e r can be used to
predict the performance of another c h a m b e r operating at different conditions
of temperature, pressure a n d velocity. T h e only reservations which need be
m a d e are that the two chambers be similar in geometry a n d airflow
distribution a n d that atomization be always adequate. It is further
suggested that the relative merits of various designs of combustion chamber
can be assessed although the available performance d a t a m a y relate to
systems differing in size a n d in operating conditions. This is an ambitious
proposal since it essentially demands a known relationship between pressure,
temperature, velocity a n d size. However, it is hoped to demonstrate that
such a relationship does exist which holds with sufficient accuracy over the
range of velocities normally encountered in gas turbine combustion chambers.
In their application to afterburners the principles of pressure scaling as
178
SOME E X P E R I E N C E S ON COMBUSTION SCALING
advocated by Stewart have been considerably more successful. T h e reasons

for this success are not difficult to seek. Firstly, afterburner systems are
mechanically very much simpler than combustion chambers a n d are,
therefore, considerably easier a n d cheaper to manufacture on a small scale.
This mechanical simplicity combined with their comparatively large size
permits the use of scaling factors as large as four or five to one. Thus scale
afterburner rigs having small power requirements can be operated at low
mass flows a n d a t convenient levels of pressure a n d yet be simulating
conditions corresponding to those experienced by the full-scale system at
extreme altitudes. Scale rigs of this type have been continuously employed
in afterburner development at Rolls-Royce since 1951.
THEORETICAL CONSIDERATIONS
Stewart's conclusions are that, provided the processes of fuel distribution and
atomization can be adequately scaled, then the best method of obtaining
similar performances from geometrically similar chambers is to operate each
unit a t the same fuel-air ratio, temperature a n d velocity a n d a t a pressure
which is inversely proportional to linear dimension. This type of scaling in
which the product P D is kept constant is now usually referred to as " P D
scaling" or "pressure scaling".*
Way has also studied the general problem of similitude between model
and prototype assuming an /ith power pressure dependence for the reaction
rate. H e concluded that the essential requirements are fulfilled when
î W ' vx \ D J
It is apparent from these equations that the scaling procedure to be adopted
in any given circumstance is largely dependent on the value chosen for n.
O n e of the principal deficiencies in scaling theory has been the difficulty
of reconciling the successful application of scaling rules in which it is assumed
that n = 2, to practical combustion systems whose performance invariably
correlates with pressure exponents demonstrably less than 2. This difficulty
was discussed by GREENHOUGH a n d LEFEBVRE (1956) in terms o f a proposed
burning velocity "theory" in which combustion efficiency was assumed to
depend on the ratio of turbulent burning velocity to gas velocity a n d also
on the Reynolds number. This gave a pressure exponent of 1 -75 for practical
systems which, however, were known to scale satisfactorily on the basis of
n equal to 2.
T h e significance of Reynolds number in the combustion process has also
been emphasized by W a y , ZUKOSKI (1956), SPALDING a n d T A L L (1954),
BRAGG and HOLLIDAY (1955) a n d PUTNAM and JENSEN (1949). According
to Way similarity of behaviour depends primarily on the Reynolds number
and a time ratio -n-,. "Reynolds number will affect the speed of mixing and
dilution while TT, fixes the residence time in relation to the reaction time.
At low Reynolds numbers, a larger value of TTX m a y be necessary to
compensate for poorer mixing in order to obtain a given degree of complete-
ness of combustion. At high Reynolds numbers the mixing patterns probably
tend to become independent of Re, so that -n, would be of chief importance."
• See page 199 for list of symbols.
179
A. H. LEFEBVRE AND G. A . HALLS
Bragg and Holliday's views on the role of Reynolds n u m b e r show

substantial agreement with those of Way. According to these authors the
pattern and performance of a burning flow system are specified for any
given fuel-air ratio by the Reynolds n u m b e r and the "reaction rate n u m b e r " .
This inter-relation of Reynolds number and reaction rate has been demon-
strated graphically by Bragg and Holliday using stabilizer blow-off data
previously correlated by Spalding a n d Tall in terms of Peclet numbers,
based on flame speed, and gas velocity. T h e y conclude that blow-off
velocity at first depends on both the "reaction rate n u m b e r " and the
Reynolds n u m b e r but at high values of the latter depends only on the
reaction rate.
T h e views of Bragg and Holliday and to some extent those of Way have
been formulated by H E R B E R T (1957) in a very simple manner. According
to Herbert, combustion performance is determined by a second order
reaction modified by the influence of mixing factors which can be related
to the Reynolds number. This he expresses quantitatively in the form of a
chemical air loading group which, for a constant temperature system,
becomes
p^)-3 ( P D Y (1)
This group can be expressed in a form more suitable for scaling as
(PD)1-' m
Thus, according to this equation, the exponents of P and D are essentially
the same and P D scaling is always possible. This conclusion stems, of course,
directly from the initial assumption of a second order reaction. However,
the most interesting feature of Eq. (2) is that it provides a reasonable
explanation for those instances where systems have scaled satisfactorily with
reaction orders apparently less than 2. Thus, for example, the reaction order
of 1-75 reported by Greenhough a n d Lefebvre is explained by Herbert as a
basic value of « = 2 modified by a value for c of 0-25. Using the same
arguments the pressure exponent of 1 -8 obtained with the stirred reactor of
LONGWELL and WEISS (1955) is explained on the basis that c = 0-2.
At this point it is of interest to speculate why the expression derived by
Greenhough and Lefebvre should differ from Herbert's since both are
basically the product of reaction kinetic and Reynolds n u m b e r groups.
It will be shown below that both theories do, in fact, lead to very similar
quantitative relationships between combustor dimensions and operating
conditions and that at least one apparent difference between the two is due
largely to differences in interpretation. However, it should be m a d e clear
that although both theories make use of Reynolds number, differing assump-
tions arc m a d e for its influence on the combustion process. Herbert introduces
Reynolds number to explain the influence on performance of departures from
homogeneity in the combustion zone; c is a measure of this departure and is
fixed for any given system regardless of operating conditions. I n the burning
velocity theory Reynolds n u m b e r plays a much more direct role in the
combustion process by virtue of its influence on flame speed.
180
SOME E X P E R I E N C E S ON C O M B U S T I O N SCALING
T h e burning velocity theory is based on two main assumptions, firstly that

combustion performance is determined by the ratio of burning velocity to
gas velocity and, secondly, that normal burning velocity is considerably
influenced by both small and large scale turbulence. T h e influence of small
scale turbulence has been studied by DAMKOHLER (1940) who showed that
its primary effect is to increase the transport processes within the burning
zone. His equation relating turbulent burning velocity with laminar flame
speed can be expressed as
S t = K.SL.Re0& (3)
where Re — Reynolds number and K is a constant. Although Damkohler's
limited experimental data were in general agreement with the above
expression, the Bunsen flame experiments of WILLIAMS and BOLLINGER (1949)
yielded an empirical relationship in which the ratio of turbulent to normal
burning velocity is proportional to Z ) 0 2 " ^ 0 ' 2 4 .
In view of this and other apparent inconsistencies, the equation for
turbulent burning velocity is expressed, for the time being, as
S t cc S L Re a (4)
T h e influence on burning velocity of large-scale turbulence is explained

by KARLOVITZ (1952) as being due to the fluctuating turbulent motion
causing small volumina of gas, containing small portions of the instantaneous
flame front, to diffuse away from their original positions. Simultaneously
the small portions of the instantaneous flame fronts move with their normal
burning velocity from one small volume of gas to the next. This effect he
expressed quantitatively in the following relation
S T = Vi2SLS) (5)
where u = r.m.s. value of fluctuating velocity. In the present theory the
concept of Karlovitz is amended in the following manner. Instead of the
flame fronts propagating from the small volumina at normal burning
velocity it is considered more probable that they will progress at a speed
which is governed by the small-scale turbulence of the approach stream.
In the following analysis, therefore, the combustion zone is visualized as a
region in which large-scale turbulence produces small volumina of gas which
propagate with a burning velocity governed by the small-scale turbulence
present.
Eq. (5) then becomes
ST = v W ) (6)
It is assumed that the fuel which burns does so completely. Combustion
inefficiency arises when the rate of combustion in the burning zone is
insufficient to cope with the throughput of fresh mixture.
T h e heat supplied to the primary zone = r M H Chu/sec.
Heat released by combustion = pAfS T C v A T .
pASjwC. A T
T h e n combustion efficiency i/ - . .„
but r H = C p A T by definition.
181
"=w
pA f S T
(')
s.
,T
- v ^
Substitution of (4) and (6) in (7) gives
r f o - * ^ «
Now the large scale turbulence should be related to the pressure drop A P
across the flame-tube by an expression of the form
K, A P I 0 5
u oc (10)
in which K 2 is the fraction of the total pressure drop promoting useful

turbulence.
Also, the dynamic head, d, in the combustion zone is given by
pF2
dec^- (II)
By substituting Eqs. (10) a n d (11) in (9) a n d assuming that A f is

proportional to A we have
t fcc^R t '\-jj (12)
T h e above equation relates combustion efficiency t o a " b u r n i n g velocity

n u m b e r " (S L jV), Reynolds n u m b e r and the specific pressure loss factor of
the system.
I n order to express Eq. (12) in terms of combustor dimensions and
operating conditions, Reynolds n u m b e r must be written as p V D p r 1 , t h e
viscosity term can then be eliminated by putting pi oc T 0 ' . L a m i n a r flame
speed can be related to the inlet conditions of pressure a n d temperature by
a n expression of the form
^ o c p^-mf'T)
where n is the reaction order a n d / ( T ) varies with the fuel and the fuel-air
ratio. Eq. (12) then becomes
0-5
n« -- 2
2 " /AP\1-°
/«(i-.)(PZ))i-..«-rJ f{T)
X " cc -
VTi-»
Some simplification of the temperature terms in the above equation is

possible because, in practice, the variation of temperature is small compared
with that of the other operating variables of mass flow and pressure. I t has
182
been shown by BRAGG (1953) that the m a x i m u m air loading for a system
operating at stoichiometric fuel-air ratio varies exponentially with the inlet
temperature. This exponential relationship was confirmed by LEFEBVRE
(1955) for other fuel-air ratios. It was found that in general,
j-^QC exp(r/A).
T h e variation of b with fuel-air ratio is given in ZUKOSKI (1956).

Eq. (13) can thus be rewritten as
0-5
" - " I AP\•~a
^ pm-a)(PDy-ai—A Texp(Tfb)
tf " oc X (14)
An alternative form of the above expression in which V is replaced hv

M T / P A is
0-5
»-2 -2_/AP\1-a
_2_ AP P * l - ' > { P D ) l - a \ — r ) exp{T/b)
X " * j j — (15)
DISCUSSION OF T H E O R Y
If in Eq. (15) values of 1-75 and 0-5 are attributed to n and a respectively
then the exponent of pressure becomes 1-75, which has been found experi-
mentally to be the best value for correlating data on combustion efficiency,
and the equation becomes identical to that previously derived by G R E E N -
HOUGH and LEFEBVRE (1956). However these same values of n and a when
applied to Eq. (14) lead to a scaling group of P°' 7 -'D which is clearly incon-
sistent with the known ability of combustion systems to scale on a P D basis.
These difficulties, arising from different exponents of P and D in Eq. (14),
disappear when using Herbert's equation in which the exponents of P and
D are essentially the same. T h e apparent inability of the burning velocity
equations to scale is, however, not necessarily due to weakness in the theory
but to a failure to observe Stewart's rule that scaling demands a value for
n of 2. For example, if n and a are given values of 2 and 0-43 respectively,
the exponent of pressure remains at 1-75 in Eq. (15) b u t pressure and
dimension now appear as (PZ)) 0 7 5 in Eq. (14). In this equation the absolute
value of the D exponent depends only on the value attributed to a but it will
always be the same as the exponent of P provided that n = 2, and thus P D
scaling will always be possible.
Influence of Reynolds Number

T h e r e is evidence to suggest that the exponent of Reynolds number varies
slightly with combustion chamber operating conditions. For example, where
a n increase in Reynolds number is brought about by an increase in P D the
value of a is diminished. If, however, Reynolds number is increased by a n
increase in velocity the value of a is increased. Evidence on the effect of
pressure is provided by CHILDS and GRAVES (1956) who conducted a series
of tests using propane fuel and plotted, for various fuel air ratios, combustor
183
A. H. LEFEBVRE AND G. A. HALLS
air-flow rate against pressure with parametric lines of combustion efficiency.

At low values of pressure the slope of the curve is approximately 2. As the
pressure increases the slope is reduced a n d at high pressure is of the order of
1-3. Now the pressure exponent given by Eq. (15) is 0-5n/(l — a). This
represents the " a p p a r e n t reaction o r d e r " in which n is the true reaction order.
O n this basis, if the value of n is 2, then values o f a of 0-5 and 0-23 apply at
low and high pressures respectively.
Experimental evidence to support the view that a increases with increasing
velocity is provided in Zukoski's results for baffle stabilisers. Zukoski found
in the turbulent region that an increase in Reynolds number, brought about
by a n increase in velocity at constant pressure, changed the correlating group
from l j { P D y ) : ' to V/[PD). Now according to Eq. (14) the correlating group
for a value of n of 2 is
V
{PD)'*1-')
T h u s Zukoski's results on the effect of increasing Reynolds n u m b e r is

explained as an increase in a from 0-33 to 0-5.
General Stability Correlating Group

This is provided by Eq. (14). If in this equation the temperature and
pressure loss terms are omitted a correlating group is obtained in the form
V
pZfj, (16)
in which
0-5n - 1
X - a
and y =
It is apparent that x = y only when n = 2 and thus this becomes an

essential stipulation for P D scaling. This confirms the original postulations
of Stewart and Way. Numerous examples are available in the literature
(ZUKOSKI, 1956; SPALDING and T A L L , 1954; PUTNAM and JENSEN, 1949) of
experimentally determined correlating groups in which x and j are different.
In the above group x is greater or less than y depending on whether n is
greater or less than 2. Specifically, for any given fuel, x is greater or less
than y depending on whether the burning velocity increases or decreases
with increasing pressure. Observed differences between x and y are, of
course, inexplicable on the basis of Herbert's equation (2).
Il is of interest to compare expression (16) with the correlating group for
stability data proposed by Spalding and Tall which is of the form
PDSf,
This also predicts possible differences between the exponents of P and D
due to the effect of pressure on burning velocity.
184
T o summarize, expression (16) provides a correlating group for stability

data in which the exponent of D is dependent on the influence of Reynolds
number on burning velocity; any observed differences between the exponents
of P and D are attributed to the influence of pressure on burning velocity.
T h e complete stability parameter which includes additional terms to account
for the dependence on inlet temperature and stabiliser drag coefficient is
given by Eq. (14).
APPLICATION OF T H E O R Y
Stewart's analysis of the fundamental component processes occurring in a
combustion chamber led to the conclusion that, provided certain simple
rules were adhered to, similarity in combustion performance would be
obtained from chambers of different size. Additional and more complicated
scaling requirements emerged following a series of tests carried out by Probert
in which it was demonstrated that a reduction in atomizer size led to
increasing losses within the atomizer itself. In consequence it is necessary
to scale the atomizer to a smaller size than the remaining combustor
components.
In the few reported instances in which aerodynamic, chemical and fuel
spray scaling requirements have all been strictly observed, the degree of
similarity in combustion performance between model and prototype has been
excellent. Nevertheless it is certain that combustion chamber scaling as
envisaged by Stewart is only of limited practical application. T h e expenditure
in time and money required to produce accurately scaled components, plus
the probability that a scries of model atomizers are needed to cover the full
operating range, arc formidable objections far outweighing any economies
in air power requirements. Some relaxation of the present rigid scaling rules,
thereby widening considerably the scope of their application is clearly
required. Since present-day chambers are designed to give adequate
atorni/.ation and distribution of fuel over the full range of operating
conditions, it is considered that some relaxation in the rigid scaling rules for
atomizers is justified. It is also suggested that a limited variation between
model and prototype in V, T and P D is permissible on the basis of Eqs. (14)
or (15). In the following sections the relationship derived in these equations
between combustion chamber operating variables, dimensions and pressure
loss factor arc examined in the light of the available experimental evidence.
Relationship Between Combustion Chamber Operating Variables

In applying Eq. (15) it has been found that experimental results on
combustion efficiency correlate best on an " a p p a r e n t " reaction order of 1-75.
This value not only correlates satisfactorily results obtained over a wide range
of operating conditions but also those obtained from various designs of
combustion chamber. This exponent of 1-75 is consistent with values for n
and a of 2 and 0-43 respectively and hence Eq. (15) becomes,
p\-::>.\n»-.:< cxp (Tjb)
M
where A is dependent on fuel-air ratio. Appropriate values of A for combustion
chambers are 300 at 60 air-fuel ratio and 150 at 100 air fuel ratio.
185
A. H . L E F E B V R E AND G. A. HALLS
Fig.l.
Combustion efficiency correlation of an aircraft gas turbine combustion chamber
at 60 air-fuel ratio
lOOr
Fig. 2.
Combustion efficiency correlation of an aircraft gas turbine combustion chamber
al 100 air fuel ratio
186
T h e application of this equation in correlating experimental data on

combustion efficiencies obtained from a single combustor is illustrated in
Figs. 1 and 2. The degree of correlation achieved must be considered very
satisfactory having regard to the fact that the data were obtained from tests
carried out at three different establishments and that the tests covered a
range of pressure of 20 : 1, a range of velocities of 3 : 1 and a range of inlet
temperatures of 2 : 1. Similar correlations have been obtained for this and
many other combustion chambers at several values of air-fuel ratio. T h e
general conclusion is that pressure, temperature and air mass flow can be
interchanged without affecting combustion efficiency, using the parameter
expressed above. This relationship breaks down at conditions where the
effects of atomization and heat losses become significant; fortunately such
conditions usually lie outside the range of engine experience.
Influence nf Combustion Chamber Dimensions

T h e interchangcability between combustion chamber dimensions and
operating pressure as expressed in the P D relationship has been confirmed
experimentally by Stewart and Probert and also at Rolls-Royce from
/ f
// /
| OJ
y< X
-^ '¥*/
<^ ...,—•••1. Ml*
t ••' - ^ K )
c x^
T
* ..
Fig. 3.
Correlation of the ignition peiformance of an aircraft gas turbine combustion chamber.
Inlet temperature constant at 288"K
observations of the effect of pressure scaling on combustion efficiency. In

order to illustrate that ignition performance can also be scaled, an example
is taken from ignition data obtained on two combustion chambers which
were similar in design but differing in size by a factor of two. T h e first step
in the analysis of such data is to determine statistically the value of the
apparent reaction order and hence the appropriate value of a. In this
instance an apparent reaction order of 1-3 was obtained and the actual
correlation achieved on the full scale chamber is shown in Fig. 3. T h e
exponent of P i n Eq. (14) thus becomes 0-3. T h e next step is to determine the
exponent of D, bearing in mind that for pressure scaling it is essential that
the exponents of P and D be the same. Unfortunately it is virtually impossible
to determine this exponent by analytical methods. However, Fig. 4 shows a
satisfactory correlation of the two ignition loops when plotted against the
group (PD) 0 ' 3 . At their peaks the difference between the two loops is only
11 per cent of the abscissae dimensions. For comparison the same data are
187
H . L E F E B V R E AND G. A. HALLS
03
_02
<
cr
<
Fig. 4.
Correlation of the ignition performance of prototype and half-scale combustion chambers.
Inlet temperature constant at 288 K
Fig. 5.
Correlation of the ignition performance of prototype and half-scale combustion chambers.
Inlet temperature constant at 288"K
shown in Fig. 5 plotted against a non-scaling group of the form P^'^D. Here
the correlation is nothing like as good, the difference between the two peaks
being 41 per cent on the same basis.
Although the experimental evidence so far obtained cannot be regarded
as conclusive, no evidence has been found with kerosine fuels to conflict
188
with the view that the exponents of P and D in Eq. (14) are the same and
that pressure scaling can therefore be applied.
Influence nf Specific Pressure Loss

In practical combustion systems the influence of pressure loss on combustion
performance is largely dependent on the absolute value of operating pressure,
its effect being greatest at the lowest pressure. As the pressure increases its
influence begins to diminish a n d eventually a pressure level is reached at
|
—*
j "^
8
o"0
1 '•
B
w.,n T.
r
P esfe
Fig. 6.
Combustion efficiency correlation of a combustion chamber operating al three different levels
of pressure loss. Pressure loss term omitted from parameter
which performance becomes independent of specific pressure loss. Al these

conditions any increase in pressure loss can actually reduce combustion
efficiency by virtue of the chilling, through too rapid mixing, of partially
burned combustion products in the dilution zone.
T h e reduced significance of specific pressure loss at high pressures is
predicted by Eq. (15) in which a decreases with increasing pressure.
Experimental results on the influence of pressure loss on performance a r e
relatively few. MESTRE (1955) found, for gutters at atmospheric pressure,
that blow-off velocity was proportional to the drag coefficient, i.e. propor-
tional to AP/rf. This corresponds to a value o f a of0-5. Mestre's results for
circular baffles showed a reduced dependence of blow-off velocity on AP/rf.
T h e actual relationship obtained was V B O oc (AP/a*) 0 ' 5 . This value of
pressure loss exponent was also obtained by C L A R K E , HARRISON a n d O D G E R S
(1958) using a Longwell-type stirred reactor over a pressure range of 0-2 to
1 atmosphere.
Evidence on the influence of pressure loss on combustion chamber
performance has been o b t a i n e d at Rolls-Royce (LEFEBVRE a n d MURRAY,
1957) using a special design of chamber in which t h e specific pressure loss
factor could be varied without causing any material change in flow distribu-
tion. Rig tests were carried out at atmospheric pressure and the combustion
efficiency varied by altering the rate of air mass flow. T h e results obtained
at three different levels of APjd are correlated in Fig. 6 using a parameter
in which the AP/</ term is omitted. Three separate curves are shown
corresponding to the three values of AP/rf. Fig. 7 shows the same data when
189
H. LEFEBVRE AiND G. HALLS
plotted against a correlating parameter which now includes a pressure loss

term in the form [APjd)°' : '. T h e effect is to bring the data onto a single curve.
Because of the general lack of information on the effect of pressure loss on
performance it is not yet possible to ascribe with certainty a definite value
for the APjd exponent. T h e experimental values reported above range from
0-5 to 1, If the apparent reaction orders obtained from correlating combustion
ao
1O
0
Io ] o •to so
Fig. 7.
Combustion efficiency correlation of a combustion chamber operating at three different levels of
pressure loss. Pressure loss term included in parameter
efficiencies are used to determine a, then this value ofa may be used (Eq. 15)
to calculate the exponent of APjd. T h e value of pressure exponent of 1-75
used in Figs. 1 and 2 and the values of 2 and 1-3 reported by Childs and
Graves give APjd exponents of 0-88, I and 0-65 respectively. With present
day combustion chambers the effect of pressure loss on performance is
mainly confined to sub-atmospheric pressures where combustion efficiencies
can be significantly less than 100 per cent. For correlating data obtained
within this range a constant value of 0-88 for the APjd exponents is recom-
mended. This is consistent with an apparent reaction order of 1-75.
T h e above discussion may be summarized into two arguments.
(a) Provided that the test range excludes conditions of very low pressure and
velocity, where the effects of atomization and heat losses cannot be ignored,
the P D scaling rules of Stewart can be extended to include variations in
velocity and temperature. Geometrically similar systems will exhibit similar
combustion performance provided that the following group is maintained
constant.
a
( P D ) 1 ^ T e x p (Tjb)
where values of a applicable to combustion chambers are 0-43 and 0-23 for
the performance characteristics of combustion efficiency and ignitability
respectively.
(b) Performance data from systems of different geometry should be correlated
using Eq. (15) in which n = 2 and a = 0-43 or 0-23 as quoted above.
Without this equation the combustion performances of individual systems
can only be compared provided they are of the same size and have been
190
SOME E X P E R I E N C E S ON COMBUSTION SCALIiMG
tested at identical operating conditions. Such instances only occur in

practice in respect to the same chamber before and after some design
modification. T h e main advantage of Eq. (15) is that it provides a method
of scaling combustor dimensions and operating conditions to common values
so that any differences in performance which remain can be attributed
directly to design. This is a tremendous asset when attempting to select a
design for a new combustion chamber from a choice of several existing
designs, none of which is of the required size or has been tested at the
required operating conditions.
Fig. 8.
Combustion efficiency correlation of three different designs of combustion chamber
T h e manner in which the parameter is used can be demonstrated by-

reference to Fig. 8. This figure illustrates correlated performance curves for
three different combustor designs. Clearly design A is superior to design C
because for any given value of combustion efficiency the parameter has a
lower value. This means that at any given operating conditions of AI, P , and
T, design A can equal the performance of design C and yet be made smaller
in size. In some instances the optimum design is not so clearly defined.
Comparison of the curves in Fig. 8 for chambers A and B shows that, although
chamber A has the better performance over part of the range, chamber B is
superior at conditions which on an actual engine would correspond to
operation at very high altitudes.
SCALING OF AFTERBURNER SYSTEMS

Tests on small-scale afterburners commenced at Rolls-Royce in 1951 and are
now carried out as a normal development procedure. T h e scaling technique
employed follows exactly the stipulations laid down by Stewart for a reaction
order of two. T h e original tests were carried out on a third-scale model at
engine values of velocity and temperature and three times the engine
191
A . H . L E F E B V R E AND G. A. HALLS
pressure. These tests were not completely successful, largely because the
inlet velocity profile of the prototype was not exactly reproduced and also
because of slight manufacturing errors on the model (JOHNSON and G R E E N -
HOUGH, 1952).
More attention to detail, for example, ensuring that dimensions and
metal thicknesses of individual components were accurately scaled, led to
considerable improvements. Surprisingly, the problems anticipated in the
M —
<*
\ \ .,-
\\
\\
4^
^
vs" 3
fe
X
V
20 30 40 SO 60
attitude x » - ' ( « l .
Fig. 9.
Comparison of engine and quarter-scale rig for five different afterburner systems
scaling of fuel injection processes never materialized. In the course of the

early rig tests it was observed that, within limits, combustion efficiency was
largely independent of the size of the fuel jet. It was therefore concluded that
the diffusion of fuel was sufficiently rapid to mask any effects arising from
droplet size and penetration. These views have been confirmed by sub-
sequent experience, although it is customary, when introducing a new
design of fuel injector, to carry out traverses to ensure that adequate scaling
of the fuel distribution pattern is in fact being achieved.
T h e success obtained in pressure scaling afterburner systems may be
illustrated by reference to Fig. 9, reproduced from GREENHOUGH and
LEFEBVRE (1956), which shows results of tests carried out over a range of
(light conditions for five different stabiliser configurations. T h e full and
broken lines refer to quarter-scale rig and full-scale flight results respectively.
In comparing the performances of various afterburner designs the same
difficulty is encountered as with combustion chambers, in that experimental
data usually relate lo systems differing both in size and operating conditions.
This difficulty can again be resolved with the aid of Eqs. (14) or (15). With
afterburners, however, there is an additional complication in that part of
the fuel is consumed in the stabiliser recirculation zone and the remainder in
192
the "propagation" zone downstream. T h e difficulty arises because the

value of a is not the same for the two zones.
This point can be demonstrated by reference to Figs. 10, 11 and 12 in
which NACA performance data (KING, 1957) are shown plotted against a
correlating parameter based on Eq. (15). T h e same stabilizer configuration
applies to all figures and the only differences arise in the length of burning
m
lm <
•o
~J '• L ,:
\ >1
— " <
(*_> ^____^_
.*
I
l«
9
E
UJ
Z
S so // X
S /y •
a
PRESSURE
-
s-2
•••
w •
-
ia
I
• '
• . • l!S
10 L • DISTANCE WT ml EN STABILISERS AND MEASURING PLANE j s r
10
0 ) • 1 a lO 12 n lo
M
Fig. 10.
Correlation of N.A.C.A. data on afterburner combustion efficiency. Inlet air temperature
sensibly constant. Burning length 2-5ft. Pressure exponent 1-85
100
' wem<^
> f*
/
• - • • '
4 - B-S -
TO
U" SS FT.
*n 1 1 1
p IK
M
Fig. I I .
zone downstream. Fig. 10 shows the correlation obtained with a short jet
pipe. T h e results correlate best on a pressure exponent of 1 -85 corresponding,
say, to values of n and a of 2 and 0-46 respectively. When, however, the
same correlating parameter is applied to the results obtained with a long
jet pipe the correlation is quite poor; Fig. 11 shows that the data fall on
three separate curves corresponding to the three levels of pressure at which
the tests were carried out. In order to produce a unique curve the pressure
193
A. I I . L E F E B V R E AND G. A. HALLS
exponent has to be reduced to 1-25 as shown in Fig. 12. This corresponds to

values for n and a of 2 a n d 0-2 respectively.
O n e possible explanation for this difference in " a p p a r e n t reaction o r d e r "
between the two zones, which is consistent with the views already expressed,
is that the value of a in Eq. (15) decreases with increase in P D . In an
afterburner system P is sensibly constant but D differs considerably between
J? •
>
u
•
Z 90
Ul
u ^»
C • X
i • • t a
• IM
*
u TO
L • 5 5 FT
i
I 1
p i wS
M
Fig. 12.
the two zones. T h e dimension of the first zone is characterized by the width
of the stabilizer whereas the characteristic dimension for the second zone is
clearly the diameter of the j e t pipe. T h u s the overall value of a and, hence,
the value of the pressure exponent to be used in correlating afterburner
performance data must be somewhere between the values for the two
separate zones. It will, of course, vary between different afterburner designs.
T h e fact that different designs of afterburner require different pressure
exponents for the o p t i m u m correlation of their individual performance data
might, at first sight, a p p e a r to present a major obstacle to the assessment of
various designs. Fortunately, most practical designs of afterburner vary
much less in length/diameter ratio than in the example quoted and thus it is
permissible to correlate the test results of all designs against a common
pressure exponent of, say, 1-4. This procedure follows closely that used in
comparing combustion chambers where a mean value of 1-75 is used
throughout the entire operating range.
Perhaps the most serious problem which has emerged from the use of
model afterburner systems is their inability to reproduce the very high
metal temperatures experienced on the actual engine. This can be a
considerable handicap because there is clearly little point in developing
small-scale systems to give high performance if their full-scale application
entails serious re-design to withstand the more arduous operating con-
ditions. This problem is also present, though to a lesser degree, in the
scaling of combustion chambers as the following rough analysis will show.
THE EFFECT OF SCALING ON AFTERBURNER

METAL T E M P E R A T U R E S
An idealized distribution of gas temperature across an afterburner jet pipe
is illustrated in Fig. 13. For the purpose of calculation it is assumed to
comprise a central core of gas at stoichiometric flame temperature, T f ,
194
surrounded by an annulus of exhaust gas which is not involved in the

combustion process and is at the normal jet pipe temperature, T 4 . T h e
mean gas temperature at the exit from the jet pipe is T s .
Heat is transferred to the jet pipe walls by radiation from the central core
of flame, and is removed by convection to the annulus of exhaust gas. If the
wall is assumed to be perfectly lagged then, when it attains equilibrium
temperature, Tu., these two quantities must just balance.
C= 7
tR i C X\
<:;
T, <X Tf _, TN
<:
•777r777->V777777777T>^7Tr77V77r777 V
Fig. 13.
Diagram illustrating temperature distribution and heat flows in an afterburner system
Radiation
T h e net radiation interchange between the flame and the walls is given by
the expression
R = s(^j{e,T}-a w T*) (I?)
Both ef and a w may be obtained from standard emissivily charts at the

appropriate temperatures and using calculated values for the partial
pressures of water vapour and carbon dioxide present in the combustion
products. Also required in these calculations is a value for the beam length;
for a typical reheat system this is estimated to be 0-12L, where /. is the
diameter of the central core of flame.
This method of determining emissivities is a tedious process, particularly
for pressures other than atmospheric. However, a convenient simplification
may be achieved by the use of empirical formulae which, for most practical
purposes, yield values which approximate closely to those obtained from the
standard charts. For example it can be shown that, over the range of
operating conditions and combustor dimensions normally encountered in a
gas turbine, the standard charts give values of emissivity which, when
corrected for the effect of total pressure, may be expressed as follows:
e = 1-18. 1 0 6 . P ° ' 8 . (r/)°-° r - 2
For kerosine the total emissivity is approximately twice the non-luminous

value. Hence ef and a w become
e, = 2-36 . 106 . P ° s . (r/) 0 ' 6 T f 2 (18)

08 0fl 2
aK = 2-36 . 10* . P . (r/) T~ (19)
195
A. H. L E F E B V R E A N D G. A. HALLS
If D is the jet pipe diameter, the ratio LjD is calculable at any engine
condition from the equation
(20)
W^=0-067XX
where
r a = overall afterburner fuel-air ratio
re = overall engine fuel-air ratio.
Inserting typical afterburner test bed operating conditions into the above
equation, i.e. T 4 = 1000; 7\ 2000; T f = 2270; r a = 0-0525;
re = 0-0181; gives LjD = 0-6.
T h u s / = 0-72 X 0-6D = 0-43Z)
Substituting Eqs. (18) a n d (19) into Eq. (17) gives
R = 0-00247P OH 4O"- ,, (r / 2 - T 2 ) Chu/ft 2 hr.
Convection
T h e heat removed from the walls is calculated using the classical expression
for convective heat transfer for the turbulent flow of gas in pipes.
Nu = 0-02Pe 0 8
It is assumed that the convective heat transfer process at the walls is

unaffected by the combustion within the pipe. T h e rate at which heat is
removed from the walls by convection is then equal to
I. . P"' 8 , D - ° ' 2 . { T w — T 4 ) Chu/ft 2 hr.
B depends only on the temperature and velocity of the gas stream and so
for pressure scaling purposes can be regarded as a constant. For a jet pipe
velocity and temperature of 600 ft/sec. and 1000 respectively the value of B is
26-3.
Application
U n d e r equilibrium conditions R = C and hence
0-00247P 08 Z) 0 - 6 (7> 2 - T 2 ) = 26-3P°' 8 Z)- 0 - 2 (r„, - Tt)

or T l + \0,650D-°*TW - ( 1 0 , 6 5 0 / ) - ° 8 J T 4 -)- Tf) = 0
For a pipe 3 ft in diameter in which T 4 = 1000 and T f = 2270, T w is

derived from the above equation as 1590. T h e variation of T w with scale is
illustrated in the upper curve of Fig. 14 and demonstrates that a quarter-scale
model [D = 0-75 ft) experiences a wall temperature some 300°C lower
than its full-size equivalent.
If the absolute values of T w in this Fig. appear abnormally high this is
mainly because 100 per cent combustion efficiency has been assumed
throughout the burning zone. Jet pipe wall temperatures have been calcu-
lated for more practical afterburning installations than the simple example
considered above; the results only confirm that a reduction in size of jet
pipe, for a constant value of P D , is accompanied by a reduction in wall
temperature.
196
THE EFFECT OF SCALING ON FLAME-TUBE

WALL T E M P E R A T U R E S
T h e effect of PZ) scaling on flame-tube wall temperatures has been discussed
by LEFEBVRE (1954). In general, a flame-tube receives heat by radiation and
convection from the burning gases, and loses heat by radiation to the outer
casing and by convection to the air flowing in the annulus formed between the
flame-tube and outer casing. U n d e r equilibrium conditions the flame-tube
temperature is such that the heat flows to and from the wall are just equal,
i.e. R. + C. = R z + C2
••
/ •
< /
-J
' 0#J \t*~

^ $2^
S
i 1 • r i •
!
Fig. 14.
Variation of wall temperatures with scaling factor for an afteibuinn and
combustion chamber
Substituting some actual combustion chamber operating conditions and

dimensions into the equations for the individual components of P , , C,, R 2
and C'„ gives
/?! = 0-003P°84D°-,!(22702 - r 2 ) Chu/ft 2 hr
C. = 4-8P°-84O-°-2(2270 - T J Chu/ft 2 hr
R, = 0-4. 10~ 8 (T* - 5004) Chu/ft 2 hr
where 0-4 represents a combined emissivity for the flame-tube and outer
casing. It is assumed that the temperature of the outer casing is equal to the
annulus air temperature T 2 .
Cm = 16-43P 0 ' 8 /) ° ' 2 ( X - 500) Chu/ft" hr.
T h e above equations have been used to calculate T w for various values of
D up to a maximum of 1 ft whilst maintaining P D constant at a value of
2 ft atm. T h e variation of T u with D is shown in Fig. 14 which illustrates
197
how a reduction in c h a m b e r size, for a constant value of P D , is accompanied

by a reduction in flame-tube wall temperature.
T h e general conclusion from the foregoing is that, where P4D scaling is
carried out on combustion chambers or afterburners at constant inlet
temperatures and velocities, the small-scale system will experience lower
metal temperatures. If an exact reproduction of metal temperatures is
required, it may be accomplished by various methods such as increasing
the inlet temperature to the smaller system or by the use of fuels of higher
emissivity. Some estimate of the increase in metal temperature afforded by
these methods may be gained using the equations quoted above.
CONCLUSIONS
(1) T h e difficulty and expense of manufacturing small-scale components
to the order of accuracy required prohibits the general application of P D
scaling to the development of aircraft combustion chambers.
(2) T o permit their wider use, two modifications to the existing scaling
rules are proposed. Firstly, a limited variation in velocity and temperature
is recommended on the basis of Eq. (14). Secondly, because present-day
combustion chambers enjoy adequate atomization and distribution of fuel
over their normal operating range, it is unnecessary to adhere strictly to the
stringent atomizer scaling rules as stipulated by Stewart and Probert.
(3) In its relaxed form, as expressed in Eqs. (14) and (15) scaling theory
can be used in the following ways:
(a) to correlate d a t a on combustion efficiency, ignition or stability
obtained over a wide range of operating conditions on any given
c h a m b e r . In this way performance d a t a obtained at convenient
operating conditions can be used to predict performance at other
operating conditions.
(b) to compare the performances of combustion systems differing in
geometry a n d size.
(4) T h e value of the pressure exponent to be used in correlation is obtained
from Eq. (15). This pressure exponent, termed " a p p a r e n t reaction order,"
comprises the true reaction order n modified by an a m o u n t depending on the
influence of Reynolds n u m b e r on burning velocity.
(5) Eq. (14) is suitable for correlating stabilizer blow-off velocities in
terms of pressure, temperature, dimensions and pressure loss. T h e exponents
of pressure and dimension are determined by the influence of Reynolds
n u m b e r on burning velocity, any difference between these exponents being
attributed to the effect of pressure on burning velocity.
(6) In their application to afterburner and ramjet systems the P4D scaling
rules have proved highly successful.
(7) W i t h P D scaling, metal temperatures are always lower on the model
than on the full-size unit.
SOMMAIRE
L'objet essentiel de ce memoire est de passer en revue Vela! acluel des recherches sur
V effet d'echelle dans les foyers de turbo-reacleurs. Le point de vue adopte est celui
de Tingenieur et on insiste specialement sur les problemes qui sont apparus pendant
ces six dernieres annees dans les applications pratiques du principe de similitude.
On arrive a la conclusion que la reduction d''echelle dans les foyers de reacteurs
198
n'est pas une operation rentable, car Vecnnnmie possible par fonclinnnement a petite
echelle est loin de compenser les difficultes el les depens?s consecutives a la realisation
de la maquette avec une precision suffisante.
On presente une relation entre les performances du foyer el les parametres
essentiels tels que pressinn, temperature et vitesse des gaz en cnurs defonctionnemenl,
ainsi que les dimensions du foyer et la perte de charge. Cette equation dnit fournir
la base pour une extrapolation des regies de similitude proposees par Stewart.
Cette equation pourrait encore s'appliquer aux foyers de rechauffe, a condition
d'eslimer convenablement les meriles relalifs des differents types de realisations,
mime si les donnees actuelles ne portent que sur des engins de tallies ou de conditions
de fonctionnement differentes.
Les lois de similitude recommandees par Stewart s'avererent des le depart tnut d
fail correcles pour les foyers de rechauffe. La plus grande difficulte reside dans
Timpossibiliie de reprnduire a petite echelle les temperatures de parois elevees des
foyers reels. On monlre ici les valeurs calculees des temperatures de parois de
foyers de reacleurs ou de foyers de rechauffe afin de souligner Vinfluence de V echelle
sur cette temperature. On presente, enfin, deux methodes pour reproduire sur les
maquetles les vraies valeurs des temperatures de parois.
SYMBOLS
S/ normal burning velocity, ft/sec
St turbulent burning velocity (small scale), ft/sec
ST turbulent burning velocity (small and large scale), ft/sec
M air mass flow, lb/sec
P static pressure
T air inlet temperature, °K
A cross-sectional flow area, ft2
D characteristic dimension of combustion system, ft
V gas velocity, ft/sec
A t flame area, ft2
AT" theoretical temperature rise due to combustion
AP pressure drop across combustion system, cold
d dynamic head based on total flow area (Same units as AP)
H lower calorific value of fuel, Chu/lb
r fuel air ratio by weight
Cp mean specific heat at constant pressure
p density, lb/ft 3
pt viscosity
rj combustion efficiency
Re Reynolds number
Nu Nusselt number
/ beam length for radiation
S Stefan-Boltzmann constant = 1-01 Chu/ft 2 /hr
ew wall emissivity (average value for Nimonic is 0-7)
ef flame emissivity at temperature T f
a w flame absorptivity to radiation from a source at T w
T4 afterburner inlet temperature, °K
Ts afterburner exit temperature, °K
Tf stoichiometric flame temperature, °K
it
199
A. H. LE F E B VRF, A N D G. A . HALLS
Tw wall temperature, ° K
R radiation heat transfer rate, Chu/ft 2 hr
C convective heat transfer rate, Chu/ft 2 h r
REFERENCES
BRAOO, S. L . Rolls-Royce Int. Rep., J u l y 1953.
BRAGG, S. L. a n d HOLLIDAY, J . B. " T h e Influence of Altitude O p e r a t i n g Conditions
on Combustion C h a m b e r Design" A G A R D . Selected Combustion Problems II, p . 270.
Butlcrworths, London, 1955.
CHILDS, J . H . a n d GRAVES, C . G. "Correlation of T u r b i n e Engine Combustion
Efficiency with Theoretical E q u a t i o n s " Sixth Symposium on Combustion, p . 869.
Reinhold, New York, 1956.
C L A R K E , A. E., HARRISON, A . J . a n d OIKIERS, J . " C o m b u s t i o n Stability in a Spherical
C o m b u s t o r " Seventh Symposium on Combustion, 1958. Butterworths, London, 1959.
DAMKOHLER, G. Z. Z . Eleclrochem. 46, N o . 11, p . 6 0 1 , Nov. 1940.
G R E E N H O U G H , V. W . a n d LEFEBVRE, A. H . " S o m e Application of Combustion Theory
to Gas T u r b i n e D e v e l o p m e n t " Sixth Symposium on Combustion, p . 858. Reinhold,
New York, 1956.
H E R B E R T , M . V. "Aerodynamic Influences on Combustion—A Survey of Flame
Stability D a t a " Unpublished M.o.S. Rep., J u l y 1957.
JOHNSON, F. M . a n d G R E E N H O U G H , V. W . Unpublished A.R.C. Rep., J u n e 1952.
KARLOVITZ, B. " O p e n T u r b u l e n t F l a m e s " Fourth Symposium on Combustion, p . 60.
Williams & Wilkins, Baltimore, 1952.
K I N G , C. R . Unpublished N A C A d a t a , 1957.
LEFEDVRE, A. H . "Scale Effect on Flame T u b e T e m p e r a t u r e s " Rolls-Royce Int. Rep.,
Nov. 1954.
LEKEBVRE, A. H . Rolls-Royce Int. Rep., October 1955.
LEFEBVRE, A. H . a n d HALLS, G. A. " T h e Simulation of Low Combustion Pressures by
W a t e r Injection" Seventh Symposium on Combustion 1958. Buttcrworths, London,
1959.
L E F E B V R E , A. H . a n d M U R R A Y , F . R. Rolls-Royce Int. Rep., M a y 1957.
L O N G W E L L , J. P. a n d WEISS, M . A. bid. and Engng. Chem., p . 1634, Vol. 47, No. 8,
1955.
M E S T R E , A. " E t u d e des Limitcs d e Slabilitc cn Relation avec la Resistance des
Obstacles a 1'Ecoulement" A G A R D . Combustion Researches and Reviews, p . 72, 1955.
PROBERT, R . P . "Application of Research lo Gas T u r b i n e Combustion P r o b l e m s "
Proc. Joint Conf. on Combustion, p . 400, Institution of Mechanical Engineers, 1955.
PUTNAM, A. A. a n d JENSEN, R. A. "Application of Dimcnsionless N u m b e r s to Flash-
back a n d other Combustion P h e n o m e n a " Third Symposium on Combustion and Flame.
p. 89. Williams & Wilkins, Baltimore, 1949.
SPALDING, D . B. a n d T A L L , B. S. Aero. Quart. Vol. V , p . 195, Sept. 1954.
STEWART, D . G. "Scaling of G a s T u r b i n e Combustion Systems" A G A R D . Selected
Combustion Problems I I , p . 384. Butterworths, London, 1955.
W A Y , S. "Combustion in the Turbojet E n g i n e " A G A R D . Selected Combustion Problems
II, p . 296. Bulterworths, L o n d o n , 1955.
W E L L E R , As. E. "Similarities in Combustion, A R e v i e w " A G A R D . Selected Combustion
Problems II, p . 371. Buttcrworlhs, London, 1955.
WILLIAMS, D . T , a n d BOLLINGER, L. M . " T h e Effect of T u r b u l e n c e on Flame Speeds
of Bunscn-type F l a m e s " Third Symposium on Combustion and Flame, p . 176. Williams
& Wilkins, Baltimore, 1949.
ZUKOSKI, E . E. "F'lame Stabilization on Bluff Bodies a t Low a n d Intermediate
Reynolds N u m b e r s " Rep. No. 20-27 Jet Propulsion Laboratory, California Institute of
Technology. Also Sixth Symposium on Combustion, p . 942, 1956.
200
DISCUSSION
S. S. PENNER (California Institute of Technology): Lefebvre and Halls have prepared a
comprehensive review of procedure! for scaling gas turbine combustion chambers and
afterburners. The practical value of their correlation equations is well demonstrated, ;ii leasi
for the ranges of design variables for which test data are available.
The derivation of scaling procedures for all combustion systems begins ultimately from the
dimensionless groups which arise when the appropriate set of conservation equations is
reduced to dimensionless form (DAMKOHLER, 1936; PENNER, 1957). The artistry of practical
scaling involves the judicious selection of the particular dimensionless groups which turn out
to be of primary, if not of overriding, importance. Study of the methods used by Lefebvre
and Halls and by oilier workers dealing with gas turbine conibiistors (WAY, 1955; STEWART,
1955) or rocket engines (PENNER, 1957; CROCCO, 1955; BARRERE, 1956) suggests almost
unanimous agreement that the Reynolds number(s) and Damkohler's first similarity para-
meter D, (which measures the ratio of an overall reaction lime lo a residence time) are of
particular importance. The detailed procedures used for applying these dimensionless groups
to the derivation of correlation parameters for engine scaling vary between different investi-
gators. An ambiguous selection is impossible because we are generally unable to specify a
valid procedure for estimating the effective overall reaction lime. Lefebvre and Halls use
heuristic arguments for modifying relations obtained by the use of turbulent burning velocity
iheories. The practical success of their procedure depends on the inlroduclion ofa Reynolds
number term with an adjustable exponent into the expression for the turbulent burning
velocity for premixed gases.
It is interesting to observe that the authors qualify the applicability of their results by
phrases suggesting that 'the effects of atomizalion" should not "become significant". It is
rather remarkable that combustion systems, in which the initial conditions correspond to
liquid sprays in a gaseous medium, can be correlated at all by using experimental and
theoretical results which really apply to premixed gaseous reactants (e.g. stirred reactor data,
nearly second-order chemical reaction rales, etc.). On the basis of results derived from r n r n i
experimental and theoretical studies concerned with diverging rocket chambers, (On i. el nl.,
1958; WILLIAMS et al., 1958) we are inc lined lo lake a rather cynical view with regard to the
fundamental significance, but not with regard to the practical value, of semi-empirical
correlations of experimental data. Thus we have obtained equally acceptable correlations
for all of the measured parameters, and equally acceptable estimates for the "fundamental"
constants (e.g. activation energy, mean drop diameters, drop-size distributions, etc.), by
using any of the following four radically different combustion models: (1) premixed gaseous
reaciants, (OiLLf/fl/., 1958) (2) liquidfuel sprays burning in gaseous oxidizer, i WILLIAMS,/«/,,
1958) (3) liquid oxidizer sprays burning in gaseous fuel, (WILLIAMS el al.. 1958) and (4) liquid
oxidizer sprays burning with liquid-fuel sprays (WILLIAMS et al.. 195)!i. The only requirement
which seems to be really essential for the practical correlation of results is the use of an
adequaii- number of variables with adjustable exponents. This slate of affairs is not likely lo
be improved until the combustion processes in practical systems are understood in much
greater detail than is the case al the present lime. An important step in this direction has
been made recently by the development o f a theory of spray burning. (WILLIAMS, 1958a)
In particular, the meaning of reaction order and of its relation lo the spray-size distribution
and mean drop diameter has been clarified. (WILLIAMS. 1958c) Also the difficult problem of
"flame front propagation" through a combustible gas-liquid mixture has been solved
(WlLUAMS, 1958b), with results which are in very good agreement with measured (Bl RGOYNE
and COHEN, 1954) propagation velocities. Full utilization of the methods for treating
combustion processes in heterogeneous systems should lend to the logical derivation of really
universal scaling parameters for rocket engines and gas turbine combiislors, in which essential
restrictions relating to geometric similarity, injector scaling, etc. are no longer imposed.
REFERENCES
BARRERE, M. MOUTET, A. and SARRAT, P. "Inslabilite de la Combustion dans les Moleurs
Fusses, Etude Experimental ONERA Pub. 82, Paris. 1956.
BURGOYNE, J. H. and COHEN, L. Proc. Roy. Soc. London A 225, p. 375. 195-1.
CROCCO, L. C AGARD Selected Combustion Problems I I p. 397 et seq. Butterworlhs, London,
1955.
DAMOKHLER, G. Z. Elektrochem. 42, p. 846, 1936.
201
A, H . L E F E B V R E AND G. A. HALLS
GILL, G., ECKEL, E. P., WILLIAMS, F. A. and PENNER, S. S. Seventh International Combustion
Symposium, Oxford, 1958.
PENNER, S. S. Chemistry Problems in Jet Propulsion, Chapters XXV and X X V L Pergamon
Press, London, 1957.
STEWART, D. G. AGARD Selected Combustion Problems I I p. 384 et seq. Butterworlhs, London,
1955.
WAY, S. AGARD Selected Combustion Problems 11, p. 296 el seq., Uulterworths, London. 1955.
WILLIAMS, F. A. Ph.D. Thesis, California Institute of Technology, Pasadena, June 1958a.
WILLIAMS, F. A. "Spray Combustion Theory" Tech. Rep. 26, Contract DA 04-495-Ord-446,
California Institute of Technology, Pasadena, August 1958b.
WILLIAMS, F. A. " T h e Apparent Order of Heterogeneous Spray Reactions" Tech. Rep. 27,
Contract DA 04-495-Ord-446, California Institute of Technology. Pasadena, September
1958c.
WILLIAMS, F. A., PENNER, S. S., GILL, G. and ECKEL, E. F. "Heterogeneous Burning in
a Diverging Reactor" North American Aviation Corp., Canoga Park, Calif., and Gruen
Applied Science Labs., Pasadena, Calif., October 1958.
E. E. ZUKOSKI (California Institute of Technology): I have two comments to make on the

interesting paper of Lcfebv-re and Halls. The firs! concerns the interpretation of data from my
experiments (ZUKOSKI and MARBLE, 1955; ZUKOSKI, 1956) which the authors use to relate
combustion rate and Reynolds number. A great deal of confusion has been caused in the
study of the flame stabilization process because the important dimension in die problem was
assumed to be the diameter of the flame holder, D, whereas it is actually the length of the
recirculation zone, /. (ZUKOSKI and MARBLE, 1955; /IKOSKI, 1956). Since the ratio LID
depends in detail on the fluid-dynamic aspects of the flow, different observers have found
radically different values of blow-off velocities for the same diameter holder and they have
had to use different exponents, n, to obtain a correlation of the form, (VjD n ). The correct
group is (VjL) and the exponent on L is unity for all flame holder types so long as the flow is
turbulent.
'The results given in ZUKOSKI and MARBLE, 1955; ZUKOSKI, 1956 cannot lie interpreted
as indicating that the exponent a changes from 0-5 lo TO. Rather this work indicates that.
when the correct characteristic dimension is used, the combustion process is seen to be
independent of the Reynolds number so long as the (low is turbulent and not laminar.
My second comment concerns the afterburner discussion given under the heading 'scaling
of afterburner systems". The authors assume that a difference in combustion rates exists
between the stabilization region and the flame spreading region further downstream, and ihey
then use ihis assumption to explain the large change in pressure exponent a required to
correlate data, cf. Figs. 10, 11 ami 12, obtained from burners which differ only in tail pipe
length.
Recent experimental results obtained at Jet Propulsion Laboratory, California Institute of
Technology (THURSTON, 1958). indicate that material consumed in the stabilization region
is small so long as the blockage is moderate and, hence, even if the combustion rate in this
zone is different from that further downstream, the difference could not explain so large a
change in exponent. Further, the experimental work indicates that the spreading process is
very sensitive to local flow conditions such as velocity fluctuations. Thus, it seems more
reasonable to us to assume that the increased tail pipe length made a difference in these
conditions which, in turn, changed the dependence of combustion rate on pressure.
The large change required in the correlation when a simple change of tail pipe length is
made is indicative of the almost complete lack of fundamental knowledge concerning I In-
spreading process. Without a better understanding of the flame spreading mechanism, the
development of scaling laws will continue to be a numerical exercise, only useful for limited
extrapolation of existing hardware.
REFERENCES
'THURSTON, D. VV. "An Experimental Investigation of Flame Spreading from Bluff Body
Flame Holders" USN, Ae Thesis. California Institute of Technology, Pasadena, California.
June 1958.
ZUKOSKI, E. E. "Flame Stabilization on Bluff Bodies at Low and Intermediate Reynolds
Numbers" Rep. 20-27 Jet Propulsion laboratory, California Institute of Technology, Pasadena,
California. Also Sixth Symposium on Combustion, p. 942. 1956.
202
ZUKOSKI, E. E. and MARBLE, F. E. "Experiments Concerning the Mechanism of Flame

BlowofT from Bluff Bodies" Proc. Gas Dynamics Symposium on Aerothermochemistry (held al
Evanston, Illinois, August 22-24 1955), pp. 205-210. North-western University Press,
1955.
ERNST SCHMIDT (Technische Mochschule Miinchen): In adding some comments to the
excellent paper of Lefebvre and Halls it is perhaps of interest to discuss the problem from a
more general point of view.
If you have a problem which cannot be solved with the help of mathematics, since the
differential equations are either too complicated, or not even known, then you can rely on
similarity theory or on dimensional analysis to find, not the complete solution, but at least
the main trends of it, as we learned in the field of fluid dynamics from Osborne Reynolds.
He taught us that the problem depends upon one dimensionless quanlily, later called
Reynolds number if the boundary conditions are similar so that they can be determined by one
characteristic length only. Reynolds number, as is well known, can be understood as the
ratio of the forces of inertia to the forces of viscosity acting upon an element of the fluid.
In proceeding to higher velocities of flow we learned later lhal die behaviour of the fluid
changes essentially if we near the velocity of sound, and another non-dimensional quantity —
the ratio of the velocity of flows lo the velocity of sound, the Mach number—comes in, together
with the ratio of the two specific heats C„ and C„.
If gradients of temperature increase so that the conduction of heat cannot be neglected, the
Prandtl number—the ratio of the transport of momentum by viscosity in a gradient of velocity
to the transport of heat per gradient of temperature—has to be considered.
In chemical processes e.g. in combustion, the substances before and after combustion are
transported to and from the flame front, not only by flow motion, but also by diffusion in a
field of partial pressures. Therefore the ratio of the transport of momentum by viscosity to the
transport of mass by diffusion, the so called Schmidt number, has to be noticed. This number
was given my name by American investigators and later internationally agreed because I
introduced this quantity in a paper published in 1929, but baptizing this quantity is beyond
my responsibility.
Up to now. we have not covered the chemical process of combustion itself. For this purpose
two non-dimensional quantifies were introduced by Damkohler, a collaborator of mine in die
Luflfahrl-Forschungs-Anstalt Volkenrode.
The first Damkohler number can be understood as the ratio of two periods of time; namely
the time necessary for the characteristic velocity of flow, which also enters into Reynolds
number, to cover a distance equal lo the characteristic length of tin- combustion chamber,
and ofa period of lime which is a characteristic for the velocity of the chemical reaction, for
example, the time in which half of the number of molecules of the react ants combine to the
reaction products.
The second Damkohler number can be understood as the ratio of the heal or enthalpy (to be
more precise) of reaction to the enthalpy of the reactanls before combustion.
Now we already have seven non-dimensional groups dealing with what happens in a
combustion chamber. Two of them—Reynolds number and Damkohler's first number—
contain the characteristic length of the combustion chamber.
However, this does not mean that the scaling ofa combustion chamber is only affected by
both these quantities. Even if they are the most important ones, the whole problem of
combustion depends also upon the other five quantities.
If the combustion takes place in a gravitational field or in a field of centrifugal force two
more dimensionless quantities are to be considered: the Froude number, the ratio of the
stagnation pressure of die characteristic velocity to the potential of gravity, and the Grashof
number dealing with a buoyancy of the hot gases in a colder environment. Both these
quantities contain the characteristic length. The shape of the Bunsen burner flame and also
the behaviour of industrial furnaces, (like Siemens Marlin furnace for steel), cannot be
understood without considering these quantities. However they can be neglected in aircraft
combustion chambers with their high velocities of flow. But even if we do so. we have to deal,
in a combustion problem, with seven independent variables, which are represented in the
most general way by the dimensionless quantities introduced above.
But as our imagination is confined to three dimensions and we are accustomed to functions
of one co-ordinate only, we usually simplify the problem by keeping constant all variables
except one and studying the influence ofa change of this co-ordinate only.
203
A . H . L E F E B V R E AND G. A. HALLS
In the case of scaling combustion chambers wc are mainly interested in the influence of an
alteration to the characteristic linear dimension of die chamber compared with a chamber
similar in shape and of standard size. But we have to take care that the influence of any
alteration of scale is not the same if the other diinensioiilrss quantities also undergo alterations.
Only a limited range of extrapolation may be allowed. However, we may find relations
between the other non-dimensional variables for which the gradient of the influence of the
one variable concerned is the same.
Such results can only be found by experiment and in this way the authors stated, for
instance, that the combustion efficiency under quite a number of different conditions is given
by the same curve if it is plotted against a certain function ofa non-dimensional combination
of the variables encountered. Another important result is the experimental confirmation of
the statement that a combustion chamber running at the low pressure of high altitudes can be
simulated by a chamber of smaller size running with higher pressure, so long as the product of
linear dimension and pressure has the same value.
If I am allowed to give some advice. I propose that all relations in this field should lie given
as functions of non-dimensional variables. In this way the results are more general and we
get rid of all the difficulties connected with the conversion of units.
REFERENCE
SCHMIDT, E. Gesundheits-Ingenieur, 52, pp. 525-529, 1929.
A U T H O R S ' REPLY
A. H. LEFEBVRK: We share Pinner's scepticism concerning the fundamental significance of
the formulae we have derived. We hope that perhaps the end has justified the means and that
these formulae will prove as useful to other combustion engineers as they have to ourselves.
Penner remarks that we qualify our results by phrases suggesting that "the effects of
atomization should not become significant". 'This is hardly correct. What in fact wc do
suggest is that in present day combustion systems aloinization is not limiting to performance
over the normal range of engine operation. This is proved by the fact that formulae derived
on the assumption of gaseous reaclants give very satisfactory correlations of performance data
obtained from combustion systems featuring spray injection. This may seem "remarkable"
but it is nevertheless true.
Zukoski rightly emphasizes our lack of knowledge concerning certain combustion processes.
There appears to be few immediate prospects of any appreciable improvement in this
situation; in the meantime the engineer has of necessity to make gross assumptions concerning
the burning process in order to derive working formulae. Such assumptions can onlv In-
justified by the practical value of the results obtained.
E. Schmidt discusses the nine important groups of significance to scaling. Way has
indicated thai for many combustion systems these are effectively reduced to two, namely
Reynolds number and Damkohler's first number. Wax's views are very similar to ours and
their development would lead In almost identical formulae.
Regarding units, all the formulae wc quote are essentially non-dimensional since they sicin
from the primary assumption lhat combustion performance is determined by the ratio of
turbulent burning velocity to approach stream velocity. However, viscosity, for example, is
expressed as a constant multiplied by a function of tempera! lire. With use such constants tend
lo be omitted and in this way the equations acquire dimensions.
204
LABORATORY AND FLIGHT TECHNIQUE FOR
THE MEASUREMENT OF THE TEMPERATURE
OF TURBINE BLADING
By E. P. COCKSHUTT, G. G. L E V Y and C. R. S H A R P
National Aeronautical Establishment, Canada
SUMMARY
The evolution ofa flight-worthy instrumentation system for measuring turbine rotor
blade temperatures in an Orenda 14 turbojet engine is described. The system
comprises sheathed chromel-alumel thermocouples installed in drilled turbine blades
and used in conjunction with mercury sliprings mounted in the exhaust bullet i f the
engine.
The instrumentation has been used lo measure, at sea level static conditions, the
turbine blade cooling produced by the pre-turbine injection of reheat fuel. An
instrumented engine is currently being flown in a Sabre 6 aircraft lo obtain similar
data at altitude.
1.0. I N T R O D U C T I O N
W H E N it becomes necessary to examine the altitude performance of an
aircraft engine, three types of facilities may be used—altitude wind tunnels,
altitude test beds and flying test beds. In a recent investigation involving a
medium sized turbojet engine, no altitude tunnel was available and the
capacity of the altitude test bed was inadequate to accommodate the required
engine and therefore a flying test bed was indicated. Since the development
of suitable instrumentation involved some difficulties and since the perfor-
mance finally obtained was reasonably gratifying, it seemed that a description
of the system and techniques might be useful to other laboratories.
During the past five years, an experimental investigation of turbojet
thrust augmentation by means of exhaust reheat or afterburning has been
carried out at the Engine Laboratory and the Flight Research Section of the
National Aeronautical Establishment, Canada. O n e special feature has
characterized the N.A.E. investigations: this is the injection of the greater
part of the reheat fuel flow into the gas stream upstream of the turbine rotor
blades. This pre-turbine injection system has several important advantages
over the more conventional tailpipe injection. First, the turbine rotor atomizes
and distributes the fuel so that only a simple fuel injection system is required.
Secondly, a maximum distance is available for evaporation of the fuel and
for mixing of the fuel with the exhaust gases before they reach the flame
holder, so that combustion is more readily initiated and propagated.
Thirdly, the liquid impinging on the turbine rotor blades cools them,
permitting an uprating of the basic engine in addition to the augmentation
produced by the reheat.
In parallel with the development of the reheat system proper, a separate
programme has been conducted to permit the measurement of turbine rotor
205
E . P . C O C K S H U T T , G. G. LEVY AND C. R. SHARP
blade temperatures, in order that the cooling effect of the pre-turbine fuel
injection may be assessed. A preliminary study was performed on a Derwent
engine (with a single-stage turbine) in which blade temperatures were
established by two methods—by measurement of the softening of hardened
steel inserts and by the use ofa thermocouple-slipring instrumentation system.
This paper reports on the extension of this work to the measurement of blade
temperatures in an O R E N D A 14 engine (with a two-stage turbine) with a
flight-worthy thermocouple-slipring system.
2.0. D E S C R I P T I O N OF ENGINE AND REHEAT SYSTEM

T h e running reported herein has been conducted with a standard O R E N D A
14 engine as used in the Sabre 6 aircraft. T h e engine was fitted with a reheat
system employing pre-turbine injection, the salient features of this system
being indicated in Fig. 1. A Sabre 6 aircraft has been used as a test vehicle
for the reheated engine to study its performance at high altitude.
T h e geometry of the aircraft imposed an upper limit on the size of (fixed)
propelling nozzle which could be fitted to the engine, for this reason the
thrust boost of the reheat system was limited to the rather modest value of 20
per cent. T h e reheat fuel flow to produce this boost is about 90 per cent of
the rated engine fuel flow, and for stability reasons about 40 per cent of the
reheat fuel is injected at the flame holder; as a result, the pre-turbine reheat
fuel flow amounts to some 52 per cent of the rated engine fuel flow at the
reheat design point. All testing was done with wide-cut aviation gasoline
(J P-4 or N A T O F-40).
3.0. D E S C R I P T I O N OF I N S T R U M E N T A T I O N SYSTEM
T h e main features of the instrumentation system m a y be seen in Fig. 2.
Metal-sheathed thermocouple assemblies are installed in drilled turbine
rotor blades; a total of eight blades, either first or second stage, or distributed
between the two stages, may be instrumented during any test. T h e sheathed
thermocouples are clamped to the turbine discs and terminate at the h u b of
the second stage turbine disc, where a transition is made to flexible extension
leads. T h e extension leads are carried back to the 12-channel mercury
slipring unit which is located in the exhaust bullet; the sliprings arc driven
by a coil spring flexible drive. Large quantities of cooling air are brought
into the bullet to cool the sliprings and one of the air ducts is used to bring
the copper leads from the .sliprings clear of the exhaust unit. Stationary
thermocouples are used to establish the " c o l d " junction temperature at the
mercury sliprings, since the rotating blade thermocouples indicate only the
difference between blade and slipring temperatures.
T h e following paragraphs describe in detail each of the main components
of this instrumentation system.
3.1. Blade Thermocouples

All the blade temperature measurement work described herein has been
carried out with sheathed thermocouple assemblies 0-062 in. in diameter,
produced bv Aero-Research Inc. of (thicago. These thermocouples i iinsist of
chromel and alumel thermocouple wires embedded in magnesium oxide
206
Flame tube
-
-
Enlarged view 5 -
1 st stage stator
blade Location of igniter jet Z
w
-
-
>
Enlarged view Ar a
Location of main fuel nozzle D-r -
Section thro' 'DD Section thro , r CC
Reheat throttle flame holder
Reheat shut off cock . Mam reheat fuel manifold
Propelling nozzle >o
-*C]-.^--. f^ ThermocoupleI ** C w
•ml Insulating shroud Ejector nozzle -
>
H
d
-
Fuel s
thermometer - {"" -
>
Engine fuel flow Reheat fuel flow Non-return d
valve L 19" — mi. 5'i -
sensing element sensing element -
Pilot fuel flow
Tailcone Diffuser Combustor section -
sensing element 2
-
Port view
Fig.l.
Layout of Orenda 14 with N.A.E, reheat system
E. V. C O C K S H U T T , G. G. LEVY AND C. R. S H A R P
COOLING Ai« SUPPt»
Fig. 2.
Cross section of instrumented engine
insulation and enclosed in type 304 stainless steel tubing; the whole assembly
is then swaged down to the 0-062 in. diameter.
During the previous programme, 0-040 in. diameter thermocouples were
used, but the experience gained then and subsequent correspondence with
the manufacturer suggested that better life would be obtained with the
larger (0-062 in.) thermpcouples.
T h e hot thermocouple junction is produced by the manufacturer and is
protected by spinning over the end of the stainless steel sheath. T w o types of
hot junction are available, in that the junction may be electrically insulated
from or grounded to the sheath. In order to bring out a m a x i m u m of
information through a limited n u m b e r of slipring channels, it was recognized
that insulated junctions were preferable. An admittedly limited scries of
208
TURBINE BLADE TEMPERATURE MEASUREMENT
tests revealed no difference in thermal performance of the two types of

junction, so that insulated junctions were used for all but the preliminary
work.
T h e life of the blade thermocouples has been adequate and has not,
contrary to expectation, proved the limiting factor on the time between
instrumentation system overhauls. Three thermocouples were used satis-
factorily for 70 hr in one engine and then were replaced only because the
blades were to be inspected. T h e failures that have occurred have most
often been caused by resoldering the flexible extension leads rather than by
normal operation of the engine.
3.2. Blade Modifications

In order lo produce reasonably long thermocouple life and to prevent
disruption of the gas flow, it appeared that the thermocouples should be
embedded in the turbine blades rather than attached to them externally.
Three methods for producing holes in the blades to receive the sheathed
thermocouple assemblies were investigated :
(1) a welding technique used in a previous programme,
(2) electric discharge machining, and
(3) conventional drilling.
T h e welding technique comprised the following steps:
(1) milling a slot about x \ in. wide up the blade,
(2) laying a -,', in. diameter carbon filler rod in the base of the slot,
(3) filling the rest of the slot with weld metal,
(4) removing the surplus weld material to restore the blade aerofoil, and
(5) drilling out the carbon filler rod.
This technique was the subject o f a most helpful investigation by the Physical
Metallurgy Division of the Department of Mines and Technical Surveys,
Ottawa, Canada. These studies showed that the welded blades used during
the Derwent tests, although they satisfactorily completed the programme,
actually had small internal cracks in the weld material. Subsequent testing
suggested that the cracking was primarily due to the physical presence rather
than the chemical composition of the carbon filler rod, and this discouraged
further work with this technique.
Electric discharge machining superficially appeared a very attractive
method for producing thermocouple holes in the turbine blades, but in
actual fact was most disappointing. O n a brief investigation, the electrode
wandered badly ( | in. in a 4 in. run), the hole was badly out of round
(0-050 in. by 0-080 in.), the time required was very long (12 hr), and the
expenditure of electrode was high (17 ft). Fortunately, other methods were
successful about the time of this test and electric discharge machining was
not investigated further.
Conventional drilling, rather surprisingly, was successful in producing
a completely adequate number of thermocouple locations, and accordingly
has been used throughout this test programme. T h e only tricks used were
(a) to counter-rotate the work at a low speed to discourage wandering of the
drill, and (b) to employ extremely frequent drill renewal. Because of the
cambered aerodynamic profile of the blades, only two chordwise positions
on the blades could be reached by drilling up from the root, but these seemed
209
E. P. COCKSHUTT, G. G. LEVY AND U. SHARP
quite a d e q u a t e to give a good picture of the blade temperature distribution;

T h e schematic layout of Fig. 3 indicates the twelve first-stage and six second-
stage thermocouple locations available during the programme, although
only eight instrumented blades with one thermocouple per blade were used
at any one time. Besides the reduction in blade cross-sectional area produced
by the 0-062 in. drilled hole, the only further weakening of the blade was the
MODIFIED
F I S THEE
SERRATION
SECOND STAGE
ROTOR BLADE
Fig.3.
Turbine blade thermocouple locations
removal of part of the lowest serration of the fir-tree fixing to bring the
sheathed thermocouple clear of the base of the blade (Fig. 3). No deterior-
ation of the turbine blades due to these modifications for receiving the
thermocouples was detected at any time during the p r o g r a m m e ; visual and
fluorescent penetrant inspections for cracking were carried out after a 25 hr
pre-flight rating test and a 3 hr acceptance test, and tip clearance checks were
made to detect excessive spanwise growth.
3.3. Disc Modifications

Modifications to the turbine disc assembly were needed to support the
thermocouples on their route from the drilled turbine blades to the terminal
210
T U R B I N E BLADE T E M P E R A T U R E MEASUREMENT
block in the h u b of the second-stage turbine disc. Referring to Fig. 2, it will

be seen that thermocouples from the first-stage turbine blades are carried
down the front face of the first-stage turbine disc, aft through specially
drilled turbine bolts, and then down the rear face of the second-stage disc to
the terminal hub. Thermocouples from second-stage turbine blades are
carried down the rear face of the second-stage disc directly to the terminal
hub. T h e thermocouples are secured to the discs by small clamping plates
Fig. 4.
Instrumented turbine tlisc assembly
spaced about two in. a p a r t ; tapped holes in the discs receive the screws which
secure these plates. Fig. 4 shows the front of an instrumented disc assembly.
Early test results with the turbine blade instrumentation suggested that
only the first-stage blades were receiving appreciable amounts of blade
cooling, and hence that first-stage blade temperature measurement should
receive priority. Accordingly, the installation was designed so that eight
first-stage blades could be instrumented—this maximum being established
by the terminal hub and slipring design. T h e n if some of the first-stage
thermocouples dropped out during a scries of tests, instrumented second-
stage blades could be installed in their stead without requiring removal of
the turbine from the engine.
211
E . P . C O C K S H U T T , G. G. LEVY AND C. R. S H A R P
3.4. Terminal Hub

T h e terminal h u b installed in the centre of the second-stage disc fulfils
a two-fold purpose—to provide a suitable electrical connection between the
sheathed thermocouple assemblies a n d the flexible extension wires, a n d to
provide a suitable point from which to drive the sliprings. T h e chief
problems in designing the electrical connections were the high temperature
(200 to 300°C) a n d strong centrifugal force (4000 g's) which were present.
T h e high temperature dictated the use of silver solder rather than soft solder,
SECOND STAGE
TURBINE DISC
SHEATHED THERMOCOUPLE
TERMINAL H U B -
S0LDERING PLATE -
-CLAMPING PLATES
,C0VER PLATE ASSEMBLY
FLEXIBLE EXTENSION LEADS.

SPACER RING
• D BOLT
SLIPRING DRIVING SPLINE
- ENGINE CENTRE
LINE
g g g
i",^^"7^7^^^^]
^W2322^^
Fig. 5.
Terminal hub
212
TURBINE lil.ADE TEMPERATURE MEASUREMENT
stainless steel for all metallic components, and mica and ceramic for all
insulators. T h e strong forces meant keeping the length of unsupported
thermocouple wires to an absolute minimum. T h e details of the terminal
h u b may be seen in Fig. 5, which has been enlarged from Fig. 2.
T h e terminal hub was originally intended to be dowelled and screwed to
the second-stage disc, but on the advice of the engine manufacturer these
fastenings were omitted on the flight engine to eliminate stress raisers in this
highly loaded portion of the disc. T h e terminal h u b is thus actually secured
only by the friction produced when the cover plate is pulled down on to the
terminal hub proper.
T h e element used to drive the sliprings is a female spline hung from the
cover plate, also visible in Fig. 5.
In order to minimize vibration of the instrumented engines, the turbine
disc and shaft assemblies, complete with instrumented blades and terminal
hub, were dynamically balanced. After balancing, only diametrically
opposite pairs of matched blades were permitted to be interchanged.
3.5. Flexible Drive

T h e decision to install the sliprings in the exhaust bullet of the engine,
rather than at the cooler and more accessible front end, was dictated by the
requirement that the instrumented engine be suitable for installation in an
F86 Sabre aircraft. T h e problems of slipring access and installation in this
aircraft, plus the problems of modifying the engine to carry thermocouple
leads through to the front end, seemed more formidable than the problems
of locating the sliprings in the exhaust bullet. T h e two chief problems in
installing the sliprings in the bullet seemed to be the temperature problem
and the drive problem.
T h e drive problem stemmed from the fact that the sliprings were supported
by the relatively flexible sheet metal structure of the exhaust bullet, and yet
were to be driven from the turbine centre line. This indicated the need for
a flexible coupling element between the terminal h u b and the sliprings and
was the occasion for an investigation into flexible drives. T h e problem could
be stated thus: to produce a flexible coupling capable of transmitting
fractional horsepowers, at moderately high temperature (200 — °C), high
rotational speed (8000 r.p.m.), and with misalignments of up to J in. T h e
drive which evolved consisted of a simple helical spring made of stainless
steel, working on supporting adapters of aluminum alloy; this may be seen
in Fig. 6.
A rather simple analysis showed that the whirling speed of the helical
spring drive could be expressed as follows:
ml
where f = whirling speed
d = spring wire diameter
r — spring coil radius (mean)
n number of active turns
/ = active spring length
K. = dimensional constant
213
SPLINE
ADAPTER
n
:
n
T.
s.
-
H
H
CENTRE
LINE
-
<
>
Z
o
-
SHAFT
SEALING
PLATE
-
HIJ. 6". Flexible drive

T h e stress in the spring wire could likewise be expressed as:
IT K. 8 —-.
1
ml
where a —• spring stress
d = shaft misalignment
K 2 = dimensional constant
Unfortunately any change in the geometry of the spring which raises its
whirling speed also raises its stress level and the best drive must be established
empirically as that one just stiff enough to avoid whirling at the rated speed.
T h e geometry finally established for these tests was as follows:
d= 0-080 in.
r= 0-335 in.
n= 5 i active turns
/= If in. (total length = 2j| in.)
T h e spline unit just forward of the flexible drive was introduced solely to
facilitate the marry-up of the bullet and exhaust unit to the back of the
engine. T h e assembly procedure was to build the drive, starting at the
slipring shaft, up to the male spline on the exhaust unit, and to offer this up
to the engine, guiding the spline in with the flexible thermocouple leads.
3.6. Mercury Sliprings

In the programme conducted on the Derwent engine, the electrical signal
from the rotating blade thermocouples was brought out through mercury-
type sliprings. These sliprings had been designed and developed by Rolls-
Royce Limited, Derby, and had been modified only to the extent of specifying
local hardware.
T h e success of the mercury sliprings in that project led to their adoption
for the O R E N D A programme described here. It was, however, necessary to
redesign the sliprings completely, in order to increase their capacity (from
6 to 12 channels) and to modify their external geometry (for installation in
the exhaust bullet).
T h e principle of operation may be seen from the enlarged detail of Fig. 7.
Each channel consists of a rotating copper ring, which turns with the shaft,
and a concentric stationary copper ring, which is clamped in the housing.
T h e thin annulus (0-005 in.) between the rotating and stationary rings is
filled with mercury (about 0-1 cm 3 ) to provide the electrical contact between
the two rings. T h e mercury is retained in the annulus by the side rings on
either side, which provide a labyrinth seal about the rotating shilling; the
mercury is added through the screwed filler plug at the top. T h e rotating
slipring is insulated from the shaft by a laminated plastic bushing.
T h e complete slipring assembly is shown in Fig. 8 and consists of twelve
channels identical with that described above, built u p on a common shaft.
T h e channels are insulated from each other by thin (0-003 in.) Mylar
polyester washers. T h e shaft is driven from the toothed drive at the left-hand
end, and wires from the engine are brought through the hollow shaft lo the
rotating terminal block at the right-hand end. Copper wires are carried back
215
from the rotating terminals to the individual sliprings along grooves in the
outside of the shaft.
T h r e e sets of sliprings were used for most of the system development
running and three more sets were built, incorporating various minor improve-
ments for the flight programme. Slipring life has been somewhat uncertain.
FILLER PLUG
"SPLINE" CHANNEL"
FOR INPUT WIRE
\\\V\V.VX,-
CENTRE LINE
SHAFT
ROTATING
INSULATING
WASHER
STATIONARY
COPPER
RING
Fig. 7.
Typical channel, 12-channel mercury slipring unit
varying from about 5 to 30 hr of engine running time. Useful slipring life,

except for accidents such as overheating or rough handling has generally
216
HOUSING TEFLON-COVERED -i
£=( COPPER WIRES -
-
FLEXIBLE
THERMOCOUPLE
—
EXTENSION
Z
LEADS
r.
~
-
>
a
-
-
-
•0 -
>
-
-
-
R
-
ROTATING
TERMII -
-
BLOCK -
-
-
Fig. 8.
I2-Channel mercury slipring unit
been terminated by loss of steady continuity in one or more channels; this

has been attributed to loss of liquid mercury through amalgamation into
the copper of the rotating and stationary sliprings. Refilling with mercury
is, however, usually only a temporary palliative and stripping, cleaning,
and rebuilding are generally necessary to restore the unit to satisfactory
performance.
3.7. Thermocouple Circuitry

As is indicated in the above section, the mercury sliprings are wired
internally with copper, so that the cold junction for each of the blade
thermocouple circuits occurs at the rotating terminal block. T h e temperature
al this point is established by two stationary thermocouples located within
half an inch of the rotating terminals; for the test bed arrangement these
thermocouples terminate at the measuring instrument in the control room.
In order to reduce the labour involved in data reduction, the voltage from
one of the stationary cold-junction thermocouples was added to the potential
of the blade selected on the multi-point selector switch, as was also a manually
adjusted voltage corresponding to the room temperature level; the sum of
these three voltages was then measured on a self-balancing potentiometer.
T h e circuit is indicated schematically in Fig. 9.
T h e figure also indicates how the signals from two thermocouples are
brought out through three slipring channels, rather than through the four
which would be used conventionally, by sharing a common negative slipring.
This method would obviously be unsatisfactory if the blade thermocouples
were grounded rather than insulated, since two parallel circuits would in
that case exist across the selector switch terminals and an erroneous result
would be obtained.
3.8. Bullet Modifications

Although the location of the sliprings in the exhaust bullet was attractive
from t h e p o i n t o f view of accessibility, the high temperature of the surrounding
gases (700°C) presented a formidable cooling problem. T h e m a x i m u m
operating temperature for the sliprings appeared to be about 150°C, being
determined by the Mylar and laminated plastic insulating materials and by
the bearings; for design purposes, however, a m a x i m u m safe temperature for
continuous operation of 100°C was established. Air cooling was selected in
achieve this temperature, and the cooling circuit which evolved is clearly
visible on the assembly drawing (Fig. 2).
T h e sliprings are supported in the centre of a structure formed by three
concentric tubes and are cooled by air led in through the bottom bullet
support strut. T h e cooling air flows back across the drive shaft rear bearing
and the sliprings and passes out into the annulus between the inner and
intermediate tubes. T h e air then flows forward in this annulus and then
back through the annulus between the intermediate and outer tubes.
Finally the cooling air is collected at the aft end of the outer tube and is
ducted out to the atmosphere through a second bullet support strut, shown
at the top of the figure.
Seizures of the forward bearing on the slipring drive shaft, during the
early phases of the programme, indicated the need for some additional cooling
218
INSULATED BLADE
THERMOCOUPLES
H
C
-—
y.
H
-
r
>
-
-
-
-
-
-
- •
FLEXIBLE -
EXTENSION LEADS -
COLD JUNCTION
COMPENSATING THERMOCOUPLE -
>
PB
c
-
-
-
WIRING z
Regions enclosed ttart
ore assumed lo be at
•
Criromel
Alumel
-
umlornr temperature
Copper V O L T A G E D I V I D E R FOR
ROOM T E M P E R A T U R E COMPENSATION
Fig. 9.
Schematic thermocouple circuit
and the extended shaft and sealing plate (Fig. 6) were added. Cooling air
now flows forward over this bearing and is ducted out to the atmosphere
through a third bullet support strut which is not shown in the general section.
This modification not only cooled the bearing but also eliminated the pressure
differential across it, transferring it forward to the sealing plate; no further
bearing seizures have been encountered since the modification was
incorporated.
3.9. Cooling Air Supply

For the early development work, cooling air was supplied to the bullet
installation described above from the plant compressed air supply in the
laboratory. For flight work, however, it was obvious that a portable source
would be required and, after an unsuccessful search for an electric blower
with sufficient pressure and flow capacity, a cockpit air-conditioning unit
from a Sabre aircraft was used instead.
This air-conditioning unit takes eighth-stage bleed air from the engine
compressor and cools it in a heat exchanger. T h e bleed air is then expanded
down almost to ambient pressure in an expansion turbine, simultaneously
dropping its temperature. T h e work extracted by the expansion turbine is
used to drive a fan, which in turn draws air across the heat exchanger
mentioned above. T h e cooling air delivered is at roughly ambient air
t e m p e r a t u r e ; at full engine speed a mass flow of 0-15 lb/sec is supplied to
the bullet cooling system at a pressure of 9 in. H g . Although this is an
appreciable air flow rate, it is a small fraction (0-15 per cent) of the engine
mass flow and no effect of this air bleed on the engine performance has been
detected.
T h e air-conditioning unit does not supply any cooling when the engine is
shut d o w n ; hence plant air is supplied for about half an hour after engine
shut-down to prevent heat from the hot engine from soaking the sliprings u p
to a dangerous temperature.
4.0. lil.ADE T E M P E R A T U R E MEASUREMENTS

T h e following paragraphs describe the use of the thermocouple-slipring
instrumentation system, described above, to determine the blade cooling
produced by the N.A.E. reheat system under sea level static conditions. An
instrumented engine is currently being test flown to obtain similar d a t a at
altitude.
4.1. Method of Analysis
T h e determination of the net blade cooling is necessarily accomplished in
two stages:
(a) determination of blade temperatures relative to the turbine outlet
(gas) temperature for the dry (unreheated) engine; and
(b) determination of blade temperatures relative to the turbine outlet
temperature at the reheat design point.
Relative blade temperatures are used in preference to actual temperatures
because the actual blade temperatures are strongly affected by the perfor-
mance parameters of the engine—in the case of the dry engine, the rotational
speed and the final nozzle size. O n the other hand the relative blade
220
temperatures (referred to turbine outlet temperature) are insensitive to these

parameters, so that data obtained on runs with slightly incorrect nozzle size
or speed settings may be included in an average. This point is illustrated in
Fig. 10; the spread of the actual temperatures in the upper portion is
contrasted with the lack of spread of the relative blade temperatures in the
lower portion of the figure (of the order of 2 or 3 degrees).
7500 7600 7700 78 0 0

CORRECTED ENGINE SPEED (RPM)
X -•-—
IM X
9
Vll -3- nUM
49 i
46 1
4T
41 o
1
7400 7500 7600 77 00 78 00
CORRECTED ENGINE SPEED (RP.M.I
Pig. 10.
Blade temperature correlation
T o compensate for variations in the ambient temperature, all the observed

temperatures (both blade and gas) have been corrected using the conven-
tional thermodynamic relations. Strictly speaking, these corrections are valid
only for gas temperatures, since they are derived from considerations of
thermodynamics and not from those of heat transfer; the only justification
for using them for blade temperatures as well is that no consistent spread of
data attributable to ambient temperature effects has been detected using
this scheme. Two circumstances help to decrease the effects of ambient
temperature variation:
(a) both dry and reheated runs have generally been made at about the
same ambient temperature conditions, and
221
E. P . C O C K S H U T T , G. G. LEVY AND C. R. S H A R P
(b) the use of blade temperatures relative to turbine outlet means that
only these differences are subject to inaccuracies in the correction
procedure.
4.2. Measured Cooling

4.2.1. Dry blade temperatures.—Fig. 11 shows the relative temperatures of
eight instrumented first stage blades as determined during a series of dry
calibrations. It will be seen that the m a x i m u m blade temperatures arc
• 157 * na
» 151 » 135
• 156 a 145
* •
• 148 •» 136
N «
Engine speed 7600 r.p.m.

Turbine outlet temperature = 660 C
Engine fuel flow 6300 Ib/hr
Thrust 6600 Ib
Fig. 11.
Blade temperatures dry engine: blade temperature less turbine outlet temperature (DC)
slightly in excess of 150°C above turbine outlet temperature. T h e maximum

leading edge blade temperature occurs near mid span of the blade and drops
off towards both root and t i p ; this temperature profile is believed to be a
good reflection of the turbine inlet gas temperature profile. T h e trailing edge
blade temperatures are some 10 to 20 degrees lower than the leading edge
temperatures, with the difference increasing towards the blade tip.
222
4.2.2. Reheated blade temperatures.—Fig. 12 shows the relative temperatures

of the same eight blades obtained during a series of reheat calibrations. It
will be seen that the leading edge thermocouple closest in spanwisc location
to the fuel injector nozzles is cooled to slightly below the turbine outlet
Engine speed 7600 r.p.m.

Turbine outlet temperature 661 C
Engine fuel flow 6490 Ib/hr
Thrust 79901b
Reheat fuel flow 5860 Ib/hr
Pre-turbine fuel flow 3290 Ib/hr
Fig.
Fig. 12.
12.
lil ad r temperatures reheat design point: blade temperature less turbine outlet temperature (°C)
temperature. It is clear that the cooling is rather localized, since the stations
only I in. outside and inside this location are at least 100°C above turbine
outlet temperature. T h e trailing edge of the blade is considerably less well
cooled than the leading edge; this stems from the fact that the liquid drops
all impinge on the leading portion of the blade aerofoil.
223
'
«?
K 25 X 30
II 108 » 85
» 161 N 105
u » * 0 1181 It 1151
» 13 - 55
• 1101 « <S I
4 K
Pre-turbine fuel flow 5 2 % of rated engine fuel flow
Fig. 13.
Blade temperatures—net cooling: dry blade temperature less reheated blade temperature (°C)
-15 -5 -»
ft ft 4
93 57 35
Pre-turbine fuel flow = 7 9 % N •
of rated engine fuel flow «
194 150 66
Fig. 14. K ft M
Blade temperatures—net cooling
with Derwent engine: dry blade
137 til 67
temperature less reheated blade ft ft It
temperature ( J C)
15 -* 54
ft ft II
fL
224
4.2.3. Net blade cooling.—In Fig. 13, the reheated blade temperatures are
shown relative to the dry blade temperatures, thus indicating the net cooling
obtained. T h e numbers for the first-stage blades follow from those shown on
the previous two figures, taking account of the one-degree difference in
turbine outlet temperatures. T h e cooling data for the second-stage blades
actually come from an earlier series of tests and are put in parentheses to
draw attention to this fact; regardless of the validity of this interpolation,
it is clear that the second-stage blades do not receive any useful degree of
cooling with this configuration.
If one averages the cooling data over the instrumented half of the first-stage
blades, there is a net cooling of about 70°C with a pre-turbine fuel flow of
52 per cent of rated engine fuel flow. In the earlier Derwent study (Fig. 14),
there was a net cooling of about 100°C over a comparable blade area, for a
pie-turbine fuel flow of 79 per cent of rated engine fuel flow. It thus appears
that there is no marked difference in cooling effectiveness between the two
applications of the system.
4.3. Transient Temperatures

T h e response of the turbine blade temperatures to engine transients was
briefly studied using an Offhcr pen recorder. A typical sequence is shown in
Fig. 15, which shows parallel traces from a jet pipe thermocouple and a
turbine blade during a sequence of starting, idling, acceleration, reheat
light-up and full reheat operation.
During light-up, the jet pipe thermocouple sees a maximum temperature
of about 785°C some 12 sec after light-up; the blade on the other hand,
reaches a maximum temperature of only about 600°C some 20 sec after
light-up. Thus, during a normal light-up there is a blade temperature
change of about 30 degrees/sec.
T h e reheat light-up is effected by opening the reheat solenoid, advancing
the reheat throttle to the idling position and actuating the hot streak igniter.
Although actuation of the solenoid and igniter produce appreciable fluctua-
tions in jet pipe temperature, no significant reflections in blade temperature
are visible. At no point do the blade temperature change rates during reheat
light-up approach those during engine light-up; admittedly, this might
change if one had a fast-acting fully-automatic reheat control system with a
variable-area nozzle which would permit light-up at full reheat fuel flow.
5.0. CONCLUSIONS
(1) A flight-worthy thermocouple-slipring instrumentation system for
measuring turbine rotor blade temperatures has been successfully used for
determining the cooling produced by pre-turbinc fuel injection.
(2) Test bed measurements show that the first-stage turbine blades of an
O R E N D A 14 engine receive a useful cooling of about 70°C with a pre-
turbine fuel injection rate of 52 per cent of engine fuel flow.
(3) During 120 hours of test bed running with the instrumentation system,
no failures detrimental to the engine operation or performance occurred.
225
rr^n-^nssnsr aAgto -> m ' -grTj-M I; , r ' M4-"-' ~mtm
4rr4:~
. - ^ g X _ U w i Jf4^4-4^^44"î -1-1 i i !•'• -X-M-r i—:= i X t T"U---X—
SHot»w: a /tag p T y i I i s j l ^ -| -.- i - i ) I i i i i
-X
IC
a
-
X:> CXX
Fig. 15.
Transient blade temperatures
SOMMAIRE
On decrit le systeme d'instrumentation utilise pour mesurer en vol la temperature
des aubes d'une turbine Orenda 14. Ce systeme comprend des thermocouples en
chromel-alumel, proteges et places dans des Irons perch dans les aubes de la turbine,
les indications etant Iransmises par contact tournant month sous le carenage aval
de la tuyere a"ejection.
Ce montage experimental a ete utilise au sol pom determiner dans les conditions
atmospheriques standard le refroidissement des aubes par le carburant de rechauffe
injecte en amont de la turbine. I 'n moteur equipe de la meme fagon est actuellement
essaye en vol pour obtenir les donnees analogues en altitude.
DISCUSSION
K. LEIST (Technische Hochschule, Aachen):
Measurements of the thermoshock effect al rotating turbine blades. Cockshult referred in his
excellent paper lo temperature measurements on the operating engine. Allow me lo make
some remarks about tests ofa certain similarity.
At my Institute for Turbomachines al the Technical University of Aachen an investigation
of a more fundamental type has been started, dealing with the respective limitations by
thermoshock of heating the blades by increasing the speed of turbojets or, cooling them by
slopping the engine. These tests have attempted lo ascertain experimentally the gradients
and differences of temperature actually occurring in rotating blades.
As is known, high additional ihermoslresses occur in cases where the temperature of entry
changes by reason of the fact that the leading edge and. in particular, the very thin trailing
edge of the blade profile shows considerable temporary differences compared with the
temperature of the blade centre, due to the smaller mass to be heated or cooled, and due to
the smaller emission and absorption of heal respectively. In being healed considerable
pressure slresses arise in the edges, because the profile-edges are getting heat faster than the
profile centres; in being cooled tension-stresses arise by the air (low streaming further out of the
slopping compressor, even aller stopping the fuel. These stresses easily lead to cracks and
fractures.
Although these tbermoshock effects in guiding blades wilh their higher temperatures are
lo some extent stronger than wilh the rotating blades, nevertheless with the latter dangerous
situations will arise, due lo the fact that I he fast rotating blades are already highly loaded
mechanically by centrifugal forces as well as thermally.
In the course of the thesis of my assistant. H. Seheele, a lesl turbine and a measuring se! was
used which had been developed lor discovering the distribution of temperature in rotating
blades when stationary and for determining the cooling effect of differenl cooling methods.
This list engine was made for measuring even rapidly varying temperatures with the aid of a
highly sensitive light spot recorder, i.e. a galvanometer with a turning mirror and a lighl
indicator.
Fig. 16 is a photograph of the test turbine with the combustion chamber, gas pipes, turbine
casings and a waterbrake. The air is supplied by .1 compressor not connected wilh the
turbine.
Fig. 17 is an axial section of the turbine, ll shows the gas-entry casing, the exhaust gas
casing, the guide vanes and rotating blades, the rotor running in two bearings and carrying
on the left side the clutch as communication to the walerbrake consuming the power
produced. The axial force caused by reaction effects is accepted pneumatically by a rotating
piston loaded with a pressure high enough for the axial force of the rotor lo be balanced.
The measurement shaft connected with the turbine wheel on which the contact discs for
deriving the deflection currents are fixed can be seen. Temperatures of 32 measuring points
of the rotor can be taken, eight of which are fixed in the wheel and in the shaft; the other 24,
however, will be used to measure temperatures of the rotating blades at various points.
Fig. 18 shows Ihc three investigated cross sections, namely at the root in the pitch circle
near the tip of each blade, on which at every eight points nickel, iiii kel-chrome thermocouples
are fixed. Interesting values may be expected from the trailing edge, the profile centre and
Ihc leading edge.
227
The therinocurrenls (as Fig. 19 shows) arc transferred to sliprings, from which they arc
picked up in this case by sliding contacts and directed to the measuring instrument. When
in motion an interior switch can be pushed in four different axial positions so that, by aid of
eight contact discs, 4 X 8 = 32 measuring points can be controlled. Precise checks con-
firmed the accuracy of picking up the current via the sliding contacts. Lack of time prevents
me showing further details of the equipment.
Fig. 16.
Fig. 20 shows a test wheel provided with many isolated thermowires radially directed to
the centre. Since the procedure of heating and accelerating in starting and cooling and
decelerating in stopping is not exactly defined, to a certain extent it may even depend on the
initiative of the pilot, reproduceable results were required to increase and decrease the speed
of the test turbine according to a definite program scheme in which pressure, speed and gas
228
-
-
—
y
-
-
>
a
-
-.
rc w
ic -
- -
-
>
-
-
-
-
w
>
DA
c
r.
M
s
2!
H
«*. 17.
7V»« ijaj turbine for measuring blade temperatures
t t = 750°C, Umax = 15,000 r.p.m.
I.. P. COCKSHUTT, G. G. LEVY AND C. R. SHARP
1 |
1 1
1
1-
J
1
BaaKk.
^Qlsv
Hi.
âk
• " ^ k —
^ m rm
IsflsW
^t
•^ kA
^ \ %
•
•HA*
~T% %
r^^
>
-x
Root section p rofile 2 4 0 / I itei-mediore section Heod section profile 320 &
/= 15-70 P rofile aSOjii '=20-94
*=48 ( =18-32
Fig. 18.
Fig. 19.
230
temperature changed in a simple and defined way. So at first a constant increase of speed
proportional to time was chosen and an adequate resulting increase of pressure as well as gas
temperature in front of the nuzzles, increasing at equal intervals from the point of ignition to
the operating temperature, but DO higher.
I am sure we will find better and more exact approaches to the real starling and stopping
procedures which, with some practice, the lest turbine should Ix- able to reproduce rather
well.
Fig. 20.
Furthermore, in order to preserve the test equipment, the risk of preliminary damage to
the blades by thermoshock effects was reduced by using low maximal speeds, arising to only
8000 r.p.m., as well as gas temperatures of only 5Q0' C. Nevertheless these tests already show
sunn- informative results and tendencies,
As an example. Fig. 21 shows a scries of measurements on the temperatures of the inner
cross section, that is the profile close lo the blade-root, with 3 different durations of the
an ileraling period. The gas temperature was increased once in 3, once in 2, once in 1 and
once in about hall a mini Mr up to ils maximal value. The graph of pressures in front of the
nozzles and of the speed is plotted in the diagram scries in the lower part of the figure. Tin-
uppermost row shows tin- increasing temperatures. Below the curve of gas temperature
which, according to the lest program, is increased al equal intervals from the point of ignition.
at each lime, the graphs of the temperatures of the trailing edge, the leading edge and the
profile centre are plotted. The irregularities within the rising temperatures resulted from the
fact that the fuel supply, as well as the air flow, to the turbine had to be controlled by band
after the mentioned program. Most interesting arc the temperature differences, firstly,
between the trailing edge and, secondly, between the leading edge, and the centre of tin-
blade.
Ii;
231
Heating in 3 min Heating in 2 min Heoting in I min Heating in'S min
500 10-
L-»s"f o
Lrp-
400 U4* c
•d a
300 , a - Gos before nozzles
b - Trailing edge
200 o
c = Leading edge 0
o
r-
100 Jl d - Profile centre -J.
X
v\
f
~.
H
si
ignition
.
Ignition Ignition Ignition
-.
120 O
e- At trailing edge-profile
IC
centre a
OJ 80 f * A t ' leading edgeurofile
K3 8 centre
£ 40
/^"-*XX e X—d^— e
— <
jy^-J f
f- f
>
y
8000 tVM -ll-i 1
Pn A-~ z
J - - l l M T M ']
X
V-
r>r "r
d nT
"
ii6000
4000
A
S/
; _
it i 11 120
80
Pressure before
turbine,
.-. V ' p 0 , mmHg
\T
| ^ 2000 f 40
0 > i 1\ 5 0 > ;S 4 D 0 1 2 3 4 's 0 2 P

:
Time, Time, Time, min Time min
Fig. 2 1 .
Increase of temperatures within rotating turbine blades, measured in the inner section
These arc shown in the next series of diagrams, where the fluctuations of the upper curves
can be recognized too. Il may be seen thai a retardation of heating in the upper diagram is
followed by a decrease in the temperature difference, that is to say the thermoshock effect.
The higher curves show the higher temperatures of the trailing edge. The considerably lower
temperature differences between tin- leading edge and the profile centre arc drawn below.
Since the fluctuations of the temperature differences,—of, that is to say, only a very few
degrees,—are caused as mentioned by not exactly following the test program, and differing
slightly from the exact planned course, they can be equalized very well by medium curves,
shown by the dashed lines.
Troiling edge-profile centre
12D m
2 mm
100
Îmm
BO Ifv- 2 min
Z V 60 - 3 min ti
Z "si
f//x\5 < X \ .
4*0 I 2
Ignition Time, min
Leodmg edge-profile centre
Fig. 22.
Differences of temperature
Fig. 22 shows a comparison of these medium curves of thermoshock effect, by drawing

together the overtemperatures over the temperature of profile centre of the trailing and the
leading edge for all the four accelerating periods. It is clearly to be seen, how the thcrmo-
shock increases considerably with a decreasing duration of acceleration. The dash-dotted
curve indicates how the shock will enlarge furthermore, if the heating were still quicker, as it
might be according to the actual starting conditions.
Fig. 23 shows a further similar diagram for the slopping period, the time not being altered,
as the retardation was caused by stopping the fuel supply. In this case, of course, the
temperatures of the leading and trailing edges decrease more quickly and have lower values
than those of the profile centre. Therefore considerable additional tensile stresses arise in the
edges, which can easily lead to crack effects in them.
The lowering of the temperature at entry down to the temperature of compressed air look
place within 3-4 min, the overlemperature of the blade centre had vanished in about the
same lime, whilst the difference of temperature in this case reached its maximum of 90 "C very
much earlier, namely in about 1/2(1 of the time in which the gas loses its temperature.
Meantime some more investigations were made with gas temperatures of 640°C. Such a
test is shown in F'ig. 24. The biggest temperature difference with an acceleration of one
minute reached a value of I40°C.
The calculation of the additional stresses resulting from the unequal temperature field in
the blade cross section is not easy, as it depends both on the individual profile shape and the
233
P. COCKSHUTT, G. G. LEVY AND C.
distribution of temperature over the entire cross section. According to the simplified
assumption thai the blade in its thick middle part does not alter its length at all, a temperature
difference of, for instance, 100°C corresponds with a thermostress of about 25 kg/mm 2 .
Further improvements may be sought in respect of higher gas temperatures, higher
circumferential and gas velocities, quicker heating and cooling the blades and generally with
regard to connecting the test procedure with the real course of operating, starting andstopping
of jet engines.
400
300
5 "
t: a 200
100
'Gas b
0 1 2 3
Gas shut Time, min
100
A / , " (profile centre-trailing edge
^ 1 / , ' , (profile centre-leading edge)
7000 140
6000 120 a>
v c 5000 i 100
c
lo
4000 \s 80 a> E
o E
3000 "x "• . - w , - jooooc
n
T 60
2000 .JS Po
40
a> cj
1000 20
0 0
Time. min
Fig. 2 3 .
Decrease of temperatures within rotating turbine blades,
measured in the inner section (slopping phase)
J. H . J O H N S T O N ( N G T E ) :
This paper provides an excellent description of instrumentation techniques associated with
blade temperature measurements, and the conclusions as to the general reliability of the
system developed at NAE are of considerable interest.
In general our experience at IN'GTF. regarding the problems of instrumenting turbine
rigs for blade cooling experiments has been similar to that described in the paper. For
234
example we would endorse all the comments regarding the welding, spark machining and
drilling methods for installing thermocouples with the additional qualification that a further
disadvantage of the welding technique may be the possibility of either blade distortion or
stress eoncenlralion occasioned by the welding process. An alternative method which has
been used with success in this country (for installing blade thermocouples in air cooled blades I
is for the sheathed wires to Ix- laid in a shallow slot milled into the blade surface, the thermo-
couple being keyed into plan- by local sand blasting followed by metal spraying.
Wc note that the wiring system advocated in the paper features the use of a common
negative slipring for two thermocouples and therefore requires that all thermocouples be
insulated. Although this has the advantage of minimising slipring requirements we would
700
'At"trailing edge-profile centre
efore nozzles -s
600
500
f leading edge-profile centre
400
300
I
200 F
E
30 r
Ignition Time,
Fig, 24.
Increase of temperatures within rotating turbine blades, measured in
the inner section. Healing in - > " " " ('g., % 650'C)
favour the use of separate lines for each thermocouple. This removes the necessity for
insulating the instrument heads which both simplifies installation and renders the system less
susceptible lo failure. Insulated thermocouples might also prove unsatisfactory for the
measurement of transient conditions, as any thermal lag associated with the insulation may-
mask the actual rate of change of blade temperature,
P. PH. CIIAFFIOTTE (Society d'Exploitation des Matcriels Hispano Suiza):
Mesure de tempiralure d'aubes de turbine.—La mesure directe de cctte temperature a toujours
ti-nte les cxperimeiitaieurs, et la premiere idee qui vient a I'esprit conduit a placer des
thermocouples sur les aubes ou les parties tournantes dont on vent mesurer la temperature et
etablir entre ITS llierino. ouples el les organes dc lecture ou d'enregistrement une liaison par
collei leurs lournants.
Les difficulles soulrvees par rctte melhodc resident preciscment dans la realisation de ccs
collectcurs ct leur securite de fonclionnement.
L'emploi de jauges de contraintes deslinees ;i 1'etude des contraintes et des vibrations
nexessite, d'ailleurs, les memes organcs delicats.
235
B. P . COCKSHUTT, G. G. LEVY AND C. R. SHARP
Sur la Fig. 25, on montrc le schema dc ce qui a etc realise pour les mesures faites au banc
d'essai sur un petit turbo-reactcur.
La mesure directc par thermocouples a etc jugee trop delicate pour etre montec sur avion.
On avail cependant a repondre aux questions suivantes: quelle est la difference entre la
temperature reellc du metal des ailettes et la reference de temperature des gaz utilisec par le
pilote. et comment evolue cette difference en fonction de 1'altitude ? Autrcment dit, compte
tcnu des alterations subies par la distribution des temperatures en fonction des conditions de
vol, dans quelle mesure la temperature de reference pilote rend-t-clle comptc de la tempera-
ture reelle supporter; par le mental des turbines?
Pour repondrc a ccs questions, il n'etait pas absolument necessaire de fairc des mesurcs
poncluellcs et des discriminations entre le bord d'attaquc et le bord de fuite. Aussi a-t-on
6t6 d'emblex seduit par le vicux principe du pyrometre optique.
Extensometre a haute Chemise d eau de

temperature refroldlssement
E x t e n s o m e t r e a resistance Eau '
S o r t i e de f ils
Arrivee
d'air c o m p e i m *
Chemise d'air de
refroidlssement
Col l e c t e u r t o u m o n t
Fig. 25.
Mesures de contraintes sur compresseur et turbine en fonctionnement.
Montage du collecteur dans le cone d'ichappement
Principe de I'appareil realise.—On utilise une cellule photodectriquc dont la sensibilite

spectrale est pratiquement comprise entre 7000 et 9000 A. C'est une cellule a vide a
cathode caesium-argent. Gc choix permet essentiellement d eViter 1'absorption parlielle du
rayonnement par C 0 2 et H 2 0 .
La faible largeur spectrale de la cellule conduit a admcttre que le rayonncment mesure est
monochromatique.
Une surface d'environ I cm 2 est visee, a travcrs une optique en silice fondue, dans la
region du bord dc fuite des ailettes qui est prexisdment la plus critique. Cettc definition est
assez fine pour permettre unc bonne exploration de la hauteur des aubes.
Naturellement, 1'apparcil ne donne pas la temperature d'une aube delerrninee mais opcrc
la moyennc de ces temperatures sur un rayon selectionne: par le calagc angulaire du pyro-
metre. Ce mode dc mesure est homogene au mode de chauffe des aubes, pour le systcmc de
combustion a injecteurs fixes tout au moins.
La haute sensibilite de la cellule donne un signal suffisant a parlir dc 500°C. Enfin, on a
une bonne stability, le point de saturation de la cellule etant bas.
Description (Fig. 26).—La cellule photoelectrique recoil le flux lumineux issu d'un point des
aubes par I'intermediaire d'une lentille et d'un obturateiir toumant.
La cellule fournit un courant proportionncl a 1'eclaircment recu, module a la frequence F
dc 1'obturateur tournant (200 Hz).
Un amplificateur sclectif donne unc puissance suffisante pour actionner un galvanometre
enrcgistreur standard.
La zone de visex est deffnie par un diaphragme. La distance de visee est ajustable de 25
cm a i m .
Le refroidissement de la cellule en-dessous de 50° est assure par une prise d'air speciale et un
ventilateur entraine par un moteur triphas£ 400 Hz absorbant 200 VV. L'optiquc est
refroidie par 1'air pr£lcv6 sur le rcfoulement compresseur.
236
ObturoteurDiaPhrCT3me
t o u m a n t • Cellule
. photo - e l e c t rique
Fig. 26.
Pyrometre photoelectrique
Lentille Dlaphnogme
Ailette
4^JH^P^
J
\ pbturoteur elect J
[MoteurJ-4-
Arnplificateur
Alimentation
Indicateur o u
[24V10A \ enregistreur
Convertlsseur
|24V/115V4Q0HZ
Pig. 27.
Pyrometre pholoelectrique: principe du circuit
Caracteristiques de 1'appareil utilise'.—

Surface observer: 1 cm'.
Etendue de mesure: 600 a SOO'C.
Alimentation: 24 V 10 A ou 115 V 400 H 100 VA.
Signal de sortie: 0 a 2 V sur une impedance de 10.000 ft.
Precision. La precision mesuree a £le de ± 2 ° a 800° el la fidelity meilleurc que 1°C pour
un pi mvoir emissif constant de la surface dc 1'aubc. On a, par ailleurs, pu mcttrc en evidence
que celui-ci variait trcs peu (0,90 a 0,93) de 500 a 900°C.
La Fig. 27 donne le schema electriquc realist.
237
8£S
•so|dnoooui.ioqi p o i e m s u ; qii.w Joiduiis

\|<|cmsuouiuii S u n o o q s oiqnoji puE S u n p o q o I I I U J U p u n o j o.\Eq sjoquiE oqi -.w.ii \ (o juiod
|BOI]OBjd E UIOJJ IUE1SUO0 0UJ11 0pB[q Qqi Ol JAIIB'.U IJOL[S SI IUE1SUOJ JUJI1 ,l|dllO.)0111.I,1ip
oqi joqioq.w si u o u s j n b lUBiJocluii .MIi 'jOAO.woq siii.iisue.il opB|q SjU|Apii)S U Q SJUEJSUOO ouiii
jJUOqs iBq.wouios OABq fjlM s,)|dno.)ouuoq] p.ipuiiojH i c q i piupE isnui ouo ' s o | d n o j o u u o q i
p o p u n o j g p u e poiEjnsui j o siu.nu .i.uiEpj oqj uo siuouimoo s u o j s u q o f 01 joodsoj q i i , \ \
poiBJisuouiop SEq oq c i j q i | i n b o s i p ojniej.icliuii
oqi ,<q |i.i.inpii.nl SOSSOJJS oqi SunEn'EAO j o uj.»|qojd i p u u j i p ojoiu UOAO oqi 01 popuoixo
oq ppio.w suo.w siq i s q i odoq ppioM ouo '5poqs [Buuoqi u o soipnis s isio-j 01 pjESoj qit \ \
: litiHsM-jo;} - j - 3
AldHtt S.ÔHinv
•]) - E opi'ioo E| op j n o n i E oouBiquiB] op .uniBJoduioi B| j.ipiSioj op O'qesuodsipu;
is.i |i -51111.11110 | -onbiuojioojo^p JOIISSOOOU sEd ou op IOA UO SIESSO SOJ j n o d o'qBiooiddt!
OSEIUEAB,! OJJJO 10 OiqiqElS JUUnl| S.U1 .11111 B || ; .ip:)ill UO|lE|pBJ E ISO |l,)JEddE.'I -OUOISIE.iq \ \
op m o d u u smrp .loiuoui qmofd op ojnjjns nE oiui;is|soj-o)oqd .)|ii||Oj oun ossq .mod E JI
•VVH'-l.VOd J E C ' ' O O U E J J u.i IU.IUI.I|BSO 'oddo|OAop oio B o n b p d o pojEddB .uiiit! U Q
dJiVHS 'M -O QNV A A 3 1 'O "O 'iXTHSMOOO 'd "3

APPLICABILITY OF SEPARATE COMPONENT TEST
RESULTS TO COMPLETE PROPULSION SYSTEMS
By DAVID S. GABRIEL and LEWIS E. WALLNER
National Advisory Committee for Aeronautics
Lewis Flight Propulsion Laboratory
Cleveland, Ohio
SUMMARY
Bnth steady-slate and transient performance characteristics of complete propulsion
systems will be discussed. Although drawing upon a decade of research experience,
attention will be concentrated on modem high flow engines. Components considered
will be inlets, compressors, engine combuslors and afterburners, turbines and exhaust
nozzles.
In steady state, the differences between isolated component performance and
component performance as a part of the complete propulsion system will he shown
In he small provided all pertinent environmental conditions are properly simulated.
In many cases, however, Ihc true component environment is extremely complex and it
becomes impractical In provide an adequate simulation. Under these conditions,
large performance differences may occur between isolated component and complete
engine results.
The transient behavior oframjet ami I in bojet engine components may be critically
affected by the presence of related ducts nr nther components. Il will be shown that
isolated component tests may be so misleading that transient engine nperation based
upon these results alnne may lead to dangerous engine operation or even to engine
failure.
INTRODUCTION
BEFORE an engine becomes a useful propulsion system it must undergo a
process of years of development. Invariably, this process proceeds from
separate, isolated component tests of the engine elements such as inlet,
compressor and combustor lo sea-level stand tests of the assembled main
engine components, perhaps to wind-tunnel or altitude-tank tests of the
complete propulsion system and finally to flight tests. Experience has shown
that many of the problems that must be met and overcome in this develop-
ment stem from an inadequate understanding of how the early separate
component tests may be applied to the complete engine. .As a contribution
to a better understanding in this important field, this paper summarizes some
experiences gained at the NACA Lewis laboratory in the applicability of
separate component performance tests to complete propulsion systems.
T h e problem is a complex one involving considerations ranging from small
details such as the proper arrangement and simulation of shaft seals all the
way u p to such broad considerations as the location of the engine on the
airplane. T h e discussion will deal primarily with those aspects of the problem
that have been found lo be fairly general and will consider only briefly many
of the smaller but important details that may be peculiar to one particular
design.
239
D. S. GABRIEL AND L. E. WALLNER
It may be taken as axiomatic that, if all actual engine environmental

conditions could be accurately simulated on a test rig, separate component
tests would be completely applicable to the entire system. T h e problem,
therefore, resolves into the question of what are the really important environ-
mental conditions. T h e r e may be one set of these conditions for steady-state
and another for transient performance; both will be explored. T h e procedure
followed is to compare separate component and full-scale system tests by
drawing upon over 10 years of N A C A experience with dozens of components
and engines. Of course, even with an extensive experience, it is never certain
that all important problems are covered. T h e future will probably turn u p
additional unexpected problems.
STEADY-STATF. RELATIONS
Steady-state performance is defined as the performance under equilibrium
conditions or the performance obtained when engine operating or flight
conditions change very slowly. By this definition, engine stall or blowout
limits encountered in normal flight are included, whereas the engine behavior
after stall, during engine acceleration or other rapid disturbances, is not
included.
A review of the experience and literature available indicates that the
factors listed in Fig. 1 have been found, in general, to be the important
environmental conditions that must be accurately simulated to obtain
steady-state component test results applicable to complete propulsion systems.
1. Average values of inlet temperature and pressure
2. Distribution of pressure, velocity and temperature at component inlet
3. Heat losses to surroundings
4. Mechanical factors (a) Leakage or bleed
(b) Seals
(c) Clearances
Fig. 1.
Factors influencing applicability of steady-state component performance to complete engines
T h e average values of inlet temperature and pressure, to be actually

encountered by the component when it operates as a part of the complete
propulsion system, must be accurately simulated in component tests if results
applicable to the complete engine are to be obtained. T h e effects of the
average inlet pressure and t e m p e r a t u r e level on compressor a n d turbine are
primarily Reynolds n u m b e r effects and may be large at high-altitude
conditions. Well known are the large effects of these variables on the
performance of combustors. Because extensive information is available in
other publications on the effects of inlet pressure a n d temperature on
component performance, further discussion will not be undertaken herein.
Of the factors listed in Fig. 1, the inlet distribution effects are the most
complex and difficult to simulate. Nonuniform distribution at the outlet of
any component m a y have a large effect on the next downstream component.
These inter-relationships exist between inlet and compressor, compressor and
combustor, combustor and turbine, and turbine and afterburner. As the
d a t a will show, some of these effects can be large.
I t is common practice in compressor rig tests to insulate the case of the
test compressor for simplicity of d a t a analysis and improved reproducibility
240
A P P L I C A B I L I T Y OF S E P A R A T E C O M P O N E N T TEST RESULTS
of results. Of course, in application the compressor may be uninsulated and

exposed to a variety of cooling atmospheres. T h e potential for error is fairly
high from improper simulation of these surroundings.
Combined under mechanical factors is a multitude of detailed considera-
tions, each of which, although possibly small in itself, may contribute to a
fairly large error. Factors such as listed in Fig. 1, and others, must be made
to conform to actual practical applications. Although no extensive discussion
of these factors will be attempted because of their great dependence on the
details of specific designs, it is not intended to minimize their importance.
Distribution Effects
T h e effects of nonuniform distribution of pressure, velocity and temperature
on each of the turbojet engine components and some consideration of ramjet
engines are briefly discussed in the subsequent paragraphs. Although the
literature is extensive on these subjects, only representative examples are
given herein.
Effect of inlet on compressor and turbine.—Most compressors are developed on
tests rigs, and engines arc designed by assuming that a uniform pressure and
AP/P AV -II%
Fig. 2.
Effect of angle of' attack foi
velocity distribution exists at the compressor inlet. Experience has shown,

however, that under normal operating conditions the distribution of pressure
and velocity may be far from uniform. For example, pressure distributions
such as shown in Fig. 2 may be encountered. As shown in the figure, large
regions of lower than average pressure may exist around the circumference
of the compressor inlet. Similar variations in pressure may exist in the
radial direction. The maldistribution of pressure may result from local
separation or shock losses in the inlet and may produce pressure variations
(air distortion) as high as ± 2 0 per cent. T h e magnitude and extent of these
air distortions vary with engine speed, llight speed, type of inlet, angle of
241
D. S. GABRIEL AND I.. W ALLN ER
attack or yaw, presence of auxiliary inlets, bleed doors and other similar
factors. T h e actual in-flight distribution may be very complex.
Relative to a uniform inlet pressure, the distorted or actual profiles may
have a rather profound effect on engine performance. Large low-pressure
regions may decrease the engine stall margin, lower airflow and component
efficiencies, and require a reduction in average turbine-inlet temperature.
T h e effect of nonuniform inlet distribution on compressor pressure ratio
and turbine-outlet temperature is illustrated in Fig. 3 for a simple but
COMPONENT
PRESSURE .8
AND 1 l l l l l l
I f MP Ul COMPRESSOR INLET PRESSURE
PROF'LE 10b
" " " ^ • ^ ^
RAT OS, 99
I I I I I I
LOCAL
IBI COMPRESSOR OUTLET PRESSURE
AVERAGE 112
00
ae
60 120 180 240 300 360
CIRCUMFERENTIAL POSITION. DEGREES
ICI TURBINE OUTLET
Fig. 3.
Effect of inlet pressure distortion on engine pressure and temperature profiles
UNDiSTORTEO
DISTORTED
80 p
70 h
COMPONENT
EFFICIENCY.
PERCENT g 0 L I
1A1 COMPRESSOR
85r-
n CORRECTED EXHAUST OAS TEMP

IBI T U R B I N E
Fig. 4.
Effect of inlet flow distortion on compressor and turbine efficiencies
representative case. A compressor-inlet circumferential pressure variation

of ± 1 5 per cent from the average was imposed. T h e compressor-outlet
pressure is nearly constant around the circumference. T h e compressor
pressure ratio, as indicated by comparison of inlet a n d outlet pressure
profiles, varies widely around the circumference. T h e deficiency in airflow
in the low-pressure region between about 130° and 300° causes a local
enrichment in the engine combustor and a resultant severe temperature
242
A P P L I C A B I L I T Y OF S E P A R A T E COMPONENT TEST RESULTS
gradient around the turbine (Fig. 3(c)). Although the figure shows turbine-
outlet temperature, a similar gradient in turbine-inlet temperature exists.
T o avoid overtemperature and failure of the turbine stators in this region,
it is necessary to reduce the average turbine-inlet temperature, thereby
reducing thrust. These thrust losses may be very large if the inlet flow-
distortion is in the order of 15 to 20 per cent.
As shown in Fig. 4, a change in compressor and turbine efficiency may
occur as a result of the large circumferential pressure ratio variation.
Although it is not shown on the plot, the average airflow through the engine
may decrease by a similar amount. These changes in efficiency and flow
are not great and are caused by a slight shift in the operating conditions of
the components due to the variation In iniii conditions. In snmc cases tins
shift can actually be favorable so that component efficiency may increase
slightly or may not change.
Probably the most important effect of air distortion is its reduction of the
stall margin. The stall margin is defined as the difference in compressor
UNIFORM
FLOW
DISTORTED
FLOW
COMPRESSOR
PRESSURE
RATIO
STEADY-STATE
OPERATING LINE
COMPRESSOR SPEED — -
Fig. 5(a).
Effect of inlet flow distortion on engine operating limits,
stall limits
pressure ratio between the steady-state equilibrium operating line and the
stall-limited pressure ratio. A typical case is illustrated in Fig. 5. With a
uniform inlet distribution, ample steady-state margin between compressor
stall and operating line exists over the normal operating speed range of the
engine. However, with distortion, the stall line is considerably lowered and
may intersect the operating line at high engine speeds as indicated. Thus,
an engine designed on the basis of compressor component tests with uniform
inlet flow conditions may be inoperable in actual flight at high corrected
engine speeds such as are encountered at high altitudes. T h e reduction in
stall-limited altitude may be as much as 15,000 to 20,000 ft, as shown in
Fig. 5(b).
Effect nf compressor on combustor.—In many engines the distribution of
velocity and pressure at the compressor outlet or combustor inlet is far from
uniform. T h e gradients may vary considerably with engine operating
conditions as shown in Fig. 6. Obviously, such a large variation will have
243
D. S. G A B R I E L AND L. E. WALLNER
appreciable effect on combustor-outlet and, in turn, on turbine-outlet

temperature profiles. These p h e n o m e n a and their effect on combustor-outlet
temperature profiles have been discussed in an A G A R D paper, CHILDS (1957).
No extensive discussion will be given herein. It is perhaps of greatest
significance to note that these distribution effects in combustors are well
60
ALTITUDE.
THOUSANDS OF FT
nc
/ SURGE FREE
/ OPERATION- |
DISTORTED FLOW
L J_ I I L
80 85 90 95 100 105
COMPRESSOR SPEED, PERCENT
Fig. 5(b).
Effect of inlet flow distortion on engine operating limits,
altitude limits
PERCENT
SPEED
88
l.lp
TURBINE-OUTLET
TEMP PROFILE
RATIO, 1.0 -
LOCAL/AVERAGE
,9
L
COMPRESSOR-
OUTLET
VELOCITY.
FT/SEC
0 20 40 60 80 100
HUB TIP
RADIAL DISTANCE. % PASSAGE HEIGHT
Fig. 6.
Effect of compressor-outlet velocity profile on turbine-outlet temperature profile
known and may be simulated with reasonable ease and accuracy on separate
combustor test rigs. Even so, some final tailoring of the combustor operating
as an integral part of the engine is usually required for achievement of a
completely satisfactory combustor profile.
Effect of turbine on afterburner.—The effect of nonuniform turbine-discharge
conditions on afterburner performance was also discussed by CHILDS (1957).
Typical results are reviewed in Fig. 7. T h e distribution of velocity at the
afterburner inlet is nonuniform at best. As indicated in the figure, the
244
A P P L I C A B I L I T Y OF S E P A R A T E COMPONENT TEST RESULTS
variations from test rig to engine can be large if special efforts are not m a d e
for proper simulation. Usually the conditions to be simulated are not well
known until after a complete engine test is made. In Fig. 7 two rather
extreme cases are shown and it is evident that very large differences in
afterburner performance may result. In separate rig test of the afterburner,
BOO
600 CONNECTED P I P E ^ /
INLET
VELOCITY. 400
FT/SEC
/ ^ "
/-— TURBOJET
?00 AFTERBURNER
/
0
IDUCT RADIUSI
Fig. 7(a).
Comparison of afterburner performance in engine and connected pipe tests,
velocity profiles
CONNECTED PIPE
PROFILE
ro
COMBUSTION
EFFICIENCY. 6 o
PERCENT
50
40
.03 .04 .05 06

FUEL-AIR RATIO
Fig. 7(b).
Comparison of afterburner performance in engine and connected pipe tests,
combustion efficiency
therefore, it is essential that a reasonable simulation of turbine-discharge or

afterburner-inlet profile be achieved if significant test results are to be
obtained.
Distribution effects in ramjet engines.—A ramjet engine consists largely of an
inlet and combustion chamber and, by analogy to the afterburner, it would
be expected that the distribution of velocity and pressure at the discharge of
the inlet would have an effect on the performance of the combustor. A
representative case of this interaction is shown in Fig. 8. T w o different inlets
connected to a ramjet engine produced the differences in inlet velocity profile
shown in Fig. 8(a). In one case, the inlet was typical of those used in
laboratory tests of combustors and produced a fairly uniform velocity profile.
In the other case, the inlet was a flight test type tested in a free jet with a
245
D. S. G A B R I E L AND L. F. . W A L L N E R !
resultant nonuniform profile. Large differences in combustor performance

resulted as shown in Fig. 8(b). Subsequent tests in which the inlet profile
from the free-jet tests was reproduced in a laboratory rig by using various
screens or blockages produced combustor performance closely corresponding
to that measured with flight-type inlets. If the combustor-inlct profiles are
FLIGHT-TYPE INLET
CONNECTED PIPE
l.6r-
1.2
LOCAL
MACH NO.
AVERAGE
MACH NO.
CENTER
IDUCT R A D I U S ) 2
Fig. 8(a).
Effect of velocity profile on ramjet engine performance,
velocity profile
I00r
FLIOHT-TYPE
INLET
CONNECTED
PIPE
9 0 r-
COMBUSTION
EFFICIENCY.
PERCENT
7 0 "
.02 .03 .04

FUEL-AIR RATIO
Fig. 8(b).
Effect of velocity profile on ramjet engine performance,
combustion efficiency
known a n d properly reproduced in separate component test rigs, flight

combuslor performance may be accurately obtained.
Summary of distribution effects.—Distortion or distribution effects arc of prime
importance in achieving component performance tests directly applicable to
the complete propulsion system. Within the normal operating range of an
engine, its component efficiencies may vary as much as 5 to 50 per cent
between the actual environment and the ideal uniform conditions usually
used in component test rigs. Similarly, the stable operating range of the
engine may be lower by as m u c h as 15,000 to 20,000 ft between the actual
flight environment and the uniform inlet environment usually provided on a
246
APPLICABILITY OF S E P A R A T E COMPONENT TEST RESULTS
ground test stand. Fortunately, the important elements of these problems

are known and may be accounted for in component test rigs by proper
artificial simulation. Many ingenious devices have been developed for
application of appropriate inlet gradients in component test rigs. Usually it
is found that, to investigate the major effects of maldistribution, the complex
flight environment need not be exactly duplicated. Even though the detailed
environment may not be known until flight tests are obtained, the component
tests, if conducted under the inlet conditions revealed by component tests of
the next upstream clement of the propulsion system, will reveal the major
problems to be accounted for. Final tailoring required in the complete
engine should then be small.
Case Heat Losses

Large axial temperature gradients along the outside wall or case of engines
are encountered in flight. These gradients may occur from compressor inlet
UNINSULATED
INSULATED
EQUIVALENT
WEIGHT
FLOW
58
• • •
COMPRESSOR ~"t
EFFICIENCY,
PERCENT
roCOMPRESSOR PRESSURE RATIO —
Fig. 9.
Effect of insulating compressor casing on compressor weight flow and efficiency
to outlet, along the engine combuslor, or along the afterburner. T o achieve

the maximum in data resolution the case of test components on separate test
rigs are often insulated. Of course, in flight, the ambient environment seen
by the engine case may be highly variable, ranging from very cool high-
altitude ambient conditions at low flight speeds to high-temperature ram air
at supersonic speeds.
In Fig. 9 an example is shown of the error that may be introduced by
direct application of compressor performance with an insulated case to
compressor performance in flight as a complete engine component. In this
case, the ambient temperature surrounding the compressor corresponded
approximately to the ram-air conditions in high-altitude subsonic flight or,
in other words, simulated the condition in which the engine compartment
was ventilated by ram air. T h e heat losses or energy extraction from the
compressor for the uninsulated case is sufficient to raise the adiabatic
efficiency about 4 per cent and the airflow by about 7 per cent. It is apparent
that these differences are large enough to require correction of insulated
component results for accurate application to the actual flight environment.
247
S. G A B R I E L AND L. E . WALLNER
Similar conditions apply to engine combustor and afterburner, although the

performance differences are usually smaller for these components.
Mechanical Effects
T h e many mechanical details that must be properly simulated may have
effects on the applicability of component tests to the complete engine ranging
from very large to small. Because the magnitude of these effects is usually
difficult to predict, the only resort is to adhere rigidly to realistic simulation.
Fig. 10 shows an example of the error that can accumulate in improperly
95 r
COMPONENT RIG
TURBINE
EFFICIENCY. 9Q _
PERCENT
TURBINE WORK OUTPUT — •
Pig. 10.
Effects of some combined factors on simulation accuracy of turbine performance
conducted component tests. I n the turbine rig tests the following factors
were simulated incorrectly by substantial a m o u n t s : rotor tip clearances were
one-half of those actually used in the engine, a uniform temperature profile
was used in the rig tests, and actual turbine-outlet whirl angles were ignored
— COMPLETE
ENGINE
o SEPARATE
1AI COMPRESSOR COMPONENT
TESTS
^ S . n .
I
PERCENT, SPEED
IB) TURBINE_
O
EFFICIENCY.
PERCENT
WORK OUTPUT —
l00_ IC) C O M B U S T O R
90-
80-^
70
PERCENT SPEED
Fig. I I .
Comparison of component rig tests and a complete engine lest
in computing turbine efficiency. T h e largest effect is believed to be due to

the differences in tip clearance. T h e accumulated error in comparing
separate component and complete engine tests was about 5 per cent. This,
of course, is a highly significant error and makes the separate component
tests inapplicable to the complete engine for most purposes. I n contrast to
Fig. 10, many tests have been made, such as those shown in Fig. 11, in which
highly applicable results were obtained if proper attention was paid to the
2-lii
many important details of mechanical simulation. As shown in Fig. 11, very

good agreement between component and complete engine tests may be
obtained in compressor, combustor and turbine if the many details of
simulation are rigidly adhered to.
T R A N S I E N T PERFORMANCE
T h e behavior of turbine engines in transient operation involves additional
considerations to the usual steady-state parameters. These factors include
stall limits, inertia of the rotor, transient turbine temperature limits as
affecting life or damage limits, type of engine control, speed of response of
variable engine geometry features, and others. Dominating the transient
operational characteristics is the compressor-stall limit. T w o types of
compressor stall are well known. These are a partial or local breakdown of
flow resulting in stalled regions occurring in the compressor and rotating
around it (rotating stall) and a more complete flow breakdown that
commonly affects all of the flow in the compressor (usually called stall).
In this paper only the more complete stall is considered. Many of the general
concepts discussed will apply also to rotating stall but would necessarily be
applied in a different manner. Stall is usually followed by surge, which is
defined as large-amplitude pressure, temperature and flow oscillations
throughout the engine. Eitherstall orsurge may result in combustor blowout.
T h e average turbine-inlet temperature after stall may reach damaging
heights. T h e stall limit, therefore, represents the useful operating limit. In
contrast to stall, which is an aerodynamic phenomenon similar to wing stall
and occurring in the compressor only, surge is a resonant unstable operation
involving the complete propulsion system. Compressor stall is the triggering
event for surge, but surge behavior depends on the entire propulsion system.
Propulsion systems are usually designed to avoid stall or surge in the normal
flight regime. However, experience has shown that in most military applica-
tions, wherein the engine is designed to operate as close as possible to its
operating limits, compressor stall unavoidably occurs at some extreme flight
condition. In these cases it is imperative to know how the engine will behave
after stall occurs, in order that the pilot may take the proper corrective action
to get out of stall without damaging the engine or with minimum interference
to his flight maneuver.
In considering the applicability of component tests to complete propulsion
systems, we are therefore interested in determining the predictability of the
acceleration rates, stall limits and behavior after stall (i.e., surge, blowout,
or damage limits) from separate component tests. When reviewing the
experience available on these questions, the factors listed in Fig. 12 are found
to be of primary importance.
1. Distribution or distortion effects
2. Resonance characteristics
3. Rate of change of operating conditions
Fig. 12.
Factors influencing applicability of transient component performance to complete system
.As has been discussed in connection with steady-state performance, non-

uniform distribution of pressure, velocity and temperature has a great effect
on stall limit and consequently on transient performance.
249
D. S. G A B R I E L AND L. E . WALLNER
Behavior of turbine engines after compressor stall is dependent on the

resonance or damping and resistance properties of the entire propulsion
system. In some operational situations, knowledge of these characteristics is
essential to assure recovery from stall without d a m a g e to the engine.
T h e rate of disturbance of the equilibrium operating conditions may have
a profound effect on the stall limit. T o predict transient behavior it is
necessary lo understand how rate of disturbance affects stall limits and to
know the rapidity of the disturbances to be encountered.
Distortion Effects On Acceleration Time

T h e reduction in margin between the steady-state operating line a n d the
stall limit (shown in Fig. 5), of course, reduces the excess torque available
for engine acceleration. An example of the magnitude of the resultant
increased acceleration time is shown in Fig. 13. T o accelerate from 85 to
20 -
INCREASE IN 15
ACCELERATION
TIME.
PERCENT 10
5 -
0 2 4 6 8 10
INLET PRESSURE DISTORTION, PERCENT
Fig. 13.
Effect of inlet flow distortion on required acceleration time from 85 to 100 per cent speed
100 per cent engine speed, an increase in acceleration time of as much as

20 or 25 per cent may be required for fairly large inlet pressure distortions
at moderate altitudes. At higher altitudes, of course, the increase in
acceleration time may become much larger. These effects are predictable
from component tests with properly simulated inlet distributions.
Behavior After Stall

Occurrence of surge.—As previously mentioned, whether or not a n engine
enters into the violent oscillations of surge after the compressor stalls depends
on the resonance characteristics of the entire propulsion system. T o illustrate
this dependence, Fig. 14 shows a time history of discharge pressure for a
compressor on a component test rig. In one case, the discharge of the
compressor is connected to a large volume pressure t a n k ; and, in the other
case, the compressor discharge is connected to a m i n i m u m volume receiver.
As is evident from these traces, after the inception of stall for the large
receiver volume the characteristic oscillations of surge are encountered; but,
in the case of the small receiver volume, the stalled compressor found a new,
lower pressure, stable operating condition. In both cases, the conditions and
250
character of the initial stall were identical, but differences in resonance

characteristics of the two systems produced entirely different behavior after
stall.
Analogous results have been found in turbojet propulsion systems.
.-\lthough it might be argued that all turbine engines look about the same
IAI LARGE RECEIVER VOLUME

COMPRESSOR
DISCHARGE
PRESSURE
IBI SMALL RECEIVER VOLUME
TIME —•
Fig. 14.
Effect of receiver volume on compressor stall characteristics
TURBINE
FAILURE
101 T U R B I N E - D I S C H A R G E GAS TEMP
ICI FUEL FLOW
IBI ENGINE SPEED
(•STALL
Fig. 15.
Time history of a compressor stall without surge in a complete engine
and would be expected to have the same resonance or surge characteristics,

experience has shown that this is not the case. In fact, the critical or
discontinuous nature of resonance phenomenon in general leads one to expect
that small changes in damping or resistance may eliminate engine surge, and
this has actually been found to be true at some operating conditions in a
small number of engines. T h e absence of surge with its attendant turbine
overtemperature, thrust losses, high stresses and other highly objectionable
features might seem at first to be a pure blessing. However, as illustrated
in Fig. 15, if the absence of surge cannot be predicted before flight, the
consequences may be even more drastic than if surge occurs.
Shown in Fig. 15 is a time history o f a compressor stall in a complete
251
GABRIEL A N D L. E. WALLNER
propulsion system in which surge, unexpectedly, did not occur. At the

initiation of stall, compressor efficiency, airflow and pressure fell off sharply.
T h e engine found a new stable operating point, but the inefficient system
caused a drop off in engine speed. Installed on the engine was a control
which had the principal function of maintaining the engine speed at the
value set by the pilot. W h e n the speed fell off, the control recognized the
error and introduced a higher fuel flow in an attempt to compensate.
Because of the low performance of the components the engine failed to
respond, with the result that the turbine-inlet temperature was driven to an
excessively high value and failure of the turbine blades resulted. As can be
seen, the whole event takes only a few seconds and, because of the absence
of the characteristic pounding usually accompanying surge, the pilot may
be completely unaware that anything serious is going on until it is too late.
As this example shows, a knowledge of the engine behavior after stall is
essential. At the present time, the fundamental differences between stall and
surge behavior of engines, which is apparently a property of the complete
system, may not be predicted from component tests.
Combustor blowout.—.As was pointed out by CHILDS (1957) engine surge may
have a great effect on combustor blowout limits. This effect is shown in
Fig. 16 taken from GUILDS (1957). During normal steady-state operation the
ALTITUDE DECREASE
FOR STEADY-STATE
DECREASE IN COMBUSTOR OPERATION
ALTITUDE.
THOUSANDS
OF FT
40
DECREASE IN ALTITUDE
FOR COMBUSTION WHEN
COMPRESSOR SURGES
4,
eoL 50 60 70 80 90 100
PERCENT RATED ENGINE SPEED
Fig. 16.
Effect of engine surge on turbojet combustor operating limits
combustor blowout limits are in a very high altitude region. If the engine
surges or stalls, however, blowout limits may drop as much as 20,000 ft.
Prediction of these reduced blowout limits from separate component tests is
not possible at the present time.
Diffuser instability.—Another propulsion system component affected by the
resonance characteristics of the entire system is the inlet. It is well known
thai supersonic Inlrls operating with the terminal slunk expelled limn the
inlet (subcritical operation) may experience a violent flow instability similar
in its outward effects to compressor surge. This diffuser buzz, if encountered
in flight, may cause excessive structural loads on the engine, uncontrollable
thrust variations, or combustor blowout. It has been shown (STERBENTZ and
EVVARD, 1955) that the magnitude and frequency of the flow oscillations
depend on the resonant characteristics of the diffuser ducting and the ducting
252
or chamber following the diffuser. Partial success in predicting the diffuscr

operating conditions for the onset of buzz has been reported.
As shown in Fig. 17, the presence of an engine downstream of the diffuscr
may have a large unexplained effect on the unstable operating region.
Comparing wind-tunnel results from a simple test rig in which an empty
duct was connected to the inlet discharge wilh tests in which the duct was
STABLE
(SUPERCRITICAL „-—'—
O P E R A T I O N ) -y ,
^ w ^ — - " " ~ ~y
m. ^ ^ s s
^.w-——- m
S s STABLE
S wS~\
MASS-FLOW STABLE ^y^y \
RATIO ^y^ s \
'Jr / /
1
DIFFUSER
s / y \ / / B U Z Z RIO
/ TEST
^DIFFUSER
BUZZ WITH /
ENGINE
SPIKE POSITION —
Fig. 17.
Effect of engine on diffuser buzz characteristics
replaced with an engine shows very little correlation. Although qualitative

explanations for the differences exist, quantitative prediction of the buzz
limits from separate component tests is not presently possible.
Disturbance Rate Effects

T h e final category of factors affecting the applicability of component tests
to complete propulsion systems to be discussed is that related to the rate of
change of operating or ambient conditions in the propulsion system. In
actual flight an engine may be exposed to numerous types of disturbances
ranging from slow climbs or accelerations to very rapid disturbances caused
by local air turbulence, ingestion of armament exhaust, and others. T h e rate
of disturbance may have a large effect on the engine operating limits.
Rate effects on compressor.—Disturbance rate effects on compressor stall limits
are illustrated in Fig. 18. T h e amplitude of inlet pressure disturbance that
can be tolerated by an engine before stall occurs is indicated by the curve.
This limiting amplitude is greatly reduced as the rapidity of the disturbance
increases. For slow disturbances a large amplitude of inlet pressure change
can be tolerated. Disturbances in this slow category might correspond to an
airplane diving or to slowly opening an air bypass door in an inlet. More
rapid disturbances might be introduced by diffuser buzz, cannon firing, or
atmospheric turbulence, or by flying through shock waves.
T h e simulation of these rapid disturbances in a component test rig may be
very difficult, although with complicated equipment it has been done. In
contrast to many other transient effects, the decline in stall limits may be
accurately predicted from a knowledge of the steady-state performance of
the compressor from an isolated component test rig, which makes it unneces-
sary to conduct tests with the actual rapid transients. T h e accuracy of the
253
prediction is indicated by the comparison between the line in Fig. 18, which
is the stall limit computed from steady-state data, and the test points, which
are the measured stall limits in simulated flight. T h e method used for the
prediction is discussed in GABRIEL et al. (1957) and is described in the
following two figures.
i.Or TESt DATA
CALCULATED LIMIT
MAGNITUDE
OF
STALL
LIMIT NG
DISTURBANCE
DISTURBANCE RATE
Fig. 18.
Effect of disturbance rate on stall limiting disturbance
In Fig. 19 the lines shown are the operating line and the stall limit for
steady-state or very slow transient operation. In Fig. 19 the ordinate is
overall compressor pressure ratio instead of limiting inlet disturbance as used
in Fig. 18. These lines, which are obtained by steady-state component or
engine tests together with the component efficiencies, are sufficient informa-
tion to compute accurately the response and operating limits of an engine
NLET
TRANSIENTS
o PRESSURE TIME
FOR
D TEMP
STALL.
SEC
0 02O
COMPRESSOR
PRESSURE .02 n 2
RATIO ^ ^ 05
w*
QUASI-STEADY y
STALL LIMIT - ^ '
S
s STEADY- STATE
/ Ww»
O P E R A T ING L I N E
i
100
EQUIVALENT ENGINE SPEED. PERCENT
Fig. 19.
Effect of rapid inlet-pressure oscillation on compressor stall limits
during a slow transient such as a climb or engine acceleration. T h e d a t a

points shown on the plot are the results of tests in which the compressor was
subjected to rapid sinusoidal variations in inlet pressure and ramps or
constant rate disturbances in inlet temperature. These fast disturbances
caused stall at the conditions indicated by the symbols and in the times after
254
APPLICABILITY OF S E P A R A T E COMPONENT TEST RESULTS
the initiation of the disturbance, as noted beside each symbol. .As these d a t a
show, for rapid transients the conditions for compressor stall are more widely
different than the steady-state lines would indicate. T o predict the stall for
rapid transients it is therefore inadequate to consider the compressor as a
simple continuous flow device. We must, instead, recognize that, within the
compressor itself, dynamic effects are taking place which permit the matching
of the individual stages to depart from their normal steady-state relationships.
T h e shift in stage matching comes about as a result of the time required
for a flow disturbance to propagate from one stage to another through the
compressor. T o analyze and predict the change in matching therefore
requires a method of computing the lag or propagation time o f a disturbance
from the basic geometry and known steady-state performance of the
compressor.
A method of handling these effects is illustrated in Fig. 20. Fig. 20(uppcr)
STAGES 1-4
COMPRESSOR
Fig. 20.
Dynamical engine model
shows the normal engine cross section which is schematically the geometry
to be considered. In Fig. 20(lower) is shown the imaginary dynamic model of
the engine that must be used to predict the stall limits in the presence of rapid
transients. Briefly, the method is to consider the compressor as a series of
small groups of stages within which the steady-state performance and stall
limits hold. Each group of stages is imagined to be followed by an empty
volume equal to the actual flow volume of the stage group. In this manner, by
appropriate mathematical methods the time required for a flow disturbance
to propagate from the stage group inlet to the stage group outlet can be
computed. By the use of this model, with proper consideration of combustor
volume or lag effects, the change in stall limits with rate of change of the
disturbance can be computed from a knowledge of the steady-state
performance of the various compressor stages and other engine components.
Rate effects on combustor.—The rate of change of the disturbance in the engine
may have a rather mysterious effect on combustor blowout limits as indicated
in Fig. 21. The altitude at which the combustor blows out following a
compressor stall decreases appreciably as the rate of disturbance increases.
255
D. S. GABRIEL AND L. E. WALLNER
This effect, which has been noted in several engines, is not well understood
a n d cannot be predicted from individual component tests.
COMPRESSOR STALL
F O L L O W E D BY
COMBUSTOR BLOWOUT
THOUSANDS
OF FT
STABLE
OPERATION
COMPRESSOR STAl ;
SLOW FAST
DISTURBANCE RATE
Fig. 21.
Effect of compressor flow disturbances on turbojet combustion
CONCLUSIONS
A review of over 10 years of experience with many individual component and
complete engine tests at the N A C A Lewis laboratory has revealed a n u m b e r
of environmental factors that must be properly simulated if component test
results applicable to complete propulsion systems are to be obtained.
Application of component test results to complete engine transient perfor-
mance prediction depends upon several factors in addition lo those that are
important for steady-state performance.
In the steady-slate case, it is shown that the testing and development of
the individual engine components must be carried out in the presence of
nearly actual average conditions of inlet pressure, velocity and temperature,
as well as distributions of these variables to be encountered in the complete
engine in flight, if applicable results are to be obtained. Heat losses must be
properly simulated as well as numerous detailed mechanical arrangements.
T h e flight environment m a y not be known in detail until after flight tests
are m a d e and the distribution effects in this environment may be so complex
as to preclude exact simulation. However, the important environmental
conditions for steady-state operation are known and component tests made
with the approximate environment to be expected will reveal the main
features of the performance of the component. Only small development and
matching problems will then remain for complete engine tests. I n this sense,
for the steady-state case, component tests are highly applicable to complete
engines.
For transient performance, applicability of component test results to
complete engines depends on the rate of the transient and the resonance
characteristics of the engine as well as upon the effects of nonuniform inlet
distributions. In contrast to the steady-state effects, most of the important
transient performance effects (with the exception of rate effects on compressor
stall and distribution effects on acceleration time) cannot be predicted from
component performance tests. T h e reason for this lack of applicability is
that the important elements of the environment which influence the transient
256
behavior are not understood in sufficient detail to permit proper simulation

in component tests. Even if these environmental conditions were understood,
the probable strong interaction effects between components, the complexity
of the problem and the probable reluctance to appreciably compromise
design to favor these transient problems, make it unlikely that it would be
profitable to attempt simulation on component rigs.
It therefore appears that, if component testing and development are
undertaken with the application of the principle that the component cannot
stand alone but is an inseparable part of the propulsion system, the results
are highly applicable to steady-state performance of complete engines but
only partially applicable to the transient performance of complete engines.
SOMMAIRE
On discute dans cet expose les performances de systemes propulsifs en regime elabli
et en regime transitoire. Bien que s'appuyanl sur Vexperience acquise en dix ans
cle recherches, Tauleur etudie essentiellement les moteurs modernes d grand debit
speciflque. Iês elements eludie's sont I'entree d'air, le compresseur, le foyer principal
et le foyer de rechauffe, la turbine et la tuyere d'ejection.
En regime etabli, on montre que les differences de performances entre un element
isole et I'element monte sur la machine sont minimes a condition de simuler
convenablement les aifferents parametres definissanl Tinteraclion de ('element avec
son mvironnement. Dan. beaucoup de cos il est bien difficile de reproduire ces
conditions et on obtient alnrs des differences considerables entre les performances de
['element isole et celles du moteur.
Le compnrlemenl d'un element de slalnreacteur ou de turboreacteur en regime
transitoire pent etre profondement affecle par la presence d'une canalisation
d'alimentation ou d'autres elements du banc. On monlre que dans ce cas les
performances obtenues peuvenl etre tellemenl errone'es c/u'une prevision de fnnctinnne-
menl d'engins en regime transilnire, fondee uniquement sur ces essais, peut conduire
a des risques serieux allant jusqu'a la rupture de la machine.
REFERENCES
CHILDS, J . H O W A R D , "Effects of O t h e r Engine Components on Performance of T u r b o -
jet Combustors and Afterburners." Combustion and Propulsion T h i r d A G A R D
Colloquium Pergamon Press, London, l!)f)8.
G A B R I E L , D A V I D S., W A L L N E R , L E W I S E., LUDICK, R O B E R T J . and VASU, G E O R G E .
Aero. Eng. Rev. Vol. 16, No. 9, p p . 54-59, Sept. 1957.
STERBENTZ, WILLIAM H . and R W A R D , J O H N C. "Critcrions for Prediction a n d
Control of Ram-Jet Flow Pulsations" NACA T N 3506, 1955. (Supersedes NACA
R M E51C27.)
DISCUSSION
P. H. CALDER (DC Havilland Ltd., London): The author of this paper has discussed the
factors which make component rig test results, obtained under uniform entry and exit
conditions, inaccurate when applied to a complete engine where the component conditions
are no longer ideal. In the following comments I propose lo mention a few points based on
my own experience of component and engine testing and lo ask llie author some questions
which may be of general interest.
The compressor performance ofa turbojet, as one might expect, is considerably affected by
the flow distribution presented to it by the intake. The paper q u o t a as an example the case
of a conical centrcbody at Mach 2-0, which gives a one-sided distribution of pressure and
hence of velocity and mass flow wilh increasing incidence. Intakes designed for high speeds
tend to have very sharp front ends and frequently give distribution troubles at take-off speeds
257
and during ground running, as well as at their design speeds. With a symmetrical intake the
low speed flow distribution can give a severe radial distortion which is, however, symmetrical
in a circumferential sense. The use of spill doors, for control of intake to engine matching at
supersonic speeds, can also cause severe radial bias having circumferential symmetry. Thus
the engine designer is likely to be faced with a mixture of two basic kinds of flow distortion
approaching the compressor: the radial bias with symmetry and the side-to-side bias, caused
by incidence effects or by intakes with cranked shapes. Experience suggests that for a given
degree of flow distortion, the side-to-side bias, whether this is in the form of a region of low
velocity or a region of high velocity, has a more severe effect on compressor stall and surge
characteristics than the circumferentially symmetrical radial bias. Clearly this may vary
from engine to engine, but I would like to ask the authors if they have come to any generalized
conclusions on this point from their experience with large numbers of engines and if they can
quote any comparative figures.
Compressors designed with low axial flow velocities tend to accept flow distortions al entry
more easily than units designed for very high axial velocities. There is, clearly, a limit to the
amount of distortion, of either type, that a good compressor can be developed to accept. It
would be most interesting if the authors could quote some figures to show the level of flow
distortion beyond which the intake must not be allowed to go, if surge is to be avoided.
In considering the effect of compressor outlet conditions on combustion chamber
performance the paper mentions the effect of radial velocity profile on turbine inlet tempera-
ture distribution. Fig. G shows a profile having a low velocity at the hub and a high velocity
at the tip. Most of the velocity profiles I have come across are of this type or arc symmetrical,
with varying amounts of reduction at the lip. I would like lo ask if the authors have encoun-
tered the reverse of this with high hub velocities on any of the engines tested at NACA .'
Fig. 3 shows that the effect of circumferential flow bias at compressor entry continues
through the compressor and causes local overheating of the turbine. There docs not, however,
seem to be much change in compressor delivery circumferential pressure distribution, so
presumably the unit described has a low mean outlet velocity where large velocity changes do
not greatly affect the total pressure. No mention has been made of circumferential velocity
changes caused by the compressor itself when intake conditions are uniform. It is possible to
get conditions of compressor operation where, due to rotating stall cells, considerable circum-
ferential velocity profiles can exist with complete reversals in profile shape occurring more
than once round the combustion entry.
It has been found on numerous occasions that, although very careful attempts were made
to simulate compressor outlet conditions on a test rig, the correlation between combustion
temperature profile on the rig and on the engine was poor. This suggests that the data, on
which compressor delivery mean conditions were based, for obtaining one dimensional
combustor rig entry radial flow biasing, was inadequate, and implies in some cases the need
for a detailed circumferential simulation, as well as a radial one. Normally this sort of thing
is not lightly undertaken, as the amount of work involved is tremendous.
Another factor, which has not been mentioned, is the possibility of the turbine having an
effect on the combustor outlet temperature pattern with the compressor conditions fixed.
This effect has been experienced on an engine, which was tested with the identical compressor
and combustor, on two successive occasions, but with different turbines. Investigation is
still in progress and it would be interesting to know whether the author has experienced tin-
need for two-dimensional simulation at the combustor entry or the need for simulating
exhaust conditions on test rigs.
The authors have mentioned that the effect of pressure and temperature level on combustor
efficiency and stability limits is well known. The effect of these two factors on outlet tem-
perature distribution, however, is not so clear. The effect of inlet velocity is understood and.
as 1 have mentioned previously, there may be, in some cases, an outlet effect connected with
the turbine, but I would like the authors' opinion regarding the effect of pressure level on
outlet temperature profile. Il is often very convenient to do an appreciable amount of
combustion chamber development al atmospheric pressure and it would be valuable lo know
whether the temperature pattern can be relied upon, if the upstream and downstream
conditions are correct as regards velocity.
In connection with measuring turbine efficiency and performance on the lest rig, the
authors mention the importance of maintaining correct mechanical clearances and inlet
temperature profiles. It is sometimes very convenient lo test turbines under cold flow
conditions and at reduced pressure levels, because of power supply and absorption difficulties.
258
6CS
•suoiiBoqddE isom JOJ OSUEJ [ripen oqi JOAO
XiqiqEis oioiuojd qoiqM o]qB|iEAE OJE sooiAop 'qooojos j o u j n q j O I J E JO OSBO oqi ui X|jB|no'iJEj
•pojnssB si i s j j J i u i j o OSUBJ oqi ui XiqiqBis JBqi sisoi OA>SII|OUOO UIOJJ s p u n jodo|OAop oqi
p i u n o n u i i u o j pjnoqs )U.nudo|.).\.ip lu.iuodiiio.) .ii|.], -.mo XjoioEjsuBsun UE SI 'oouEqjinsip
AUB jsoiu|B Xq XiqiqEisui oiui p o d d u i .)() UE.) II lEqi os 'jop|E;) Xq p.issn isjp joujnqjoijB oqi
SE ( E J I I U J SB si Xi||iqBis JSOI|A\ luouoduioo B iBip u o i u i d o s.joqjnB oqi si 11 Xqdosopqd ]BJouoij
)o joiiBui E s y -J.IUJII(IJ.II|E .up -jop'B;) Xq pouotiuom SE ' p u s jossojduio.) Oljl '10)111 01)1 .UlT
siqi j o s,)|dmEX3 -XiqiqEis |BUi4iJEUi OAEq UE.) 'siiioiiodiiioo j o q i o JO oouosojd oqi ui 'qoiqM
siuouoduioo j o j o q i u n u E OJB o j o q x -XiqiqEis j o s m o | q o j d sujoouoo luiod puooos o q x
•ouiltu.i oiojdmoo oqi
JOJ /jumiEuioj suio|qojd iuomdo|OAop ||Ems oq SXEM)B | ) I M ojoqi jodBd oqi ui p o u o n u o u i s y
•ouiHuo oio|diiioj oqi j o .i,\i|i;|ii.is.ij<l.u X|.i|o|diiioo oq lou )|IM sqnsoj oqi -OSJUOO J Q pojOAOO
-sip oq uoos | | I M uo|iE|nmis j o oojjiop XJESSOOOU oqi |||>)s puE oououodxo qii.w suoiiipuoo
osoqj j o p u n UIBSE SIS.)) m o u o d m o . i u u o j j o d pin; iii.iiiniliiiii.i uiEOJisuMop )xou oqi j o lojui
oqi IE s u o i q p u o o osoqi osodiui ol si dojs puooos o q j ' j o p n o sp I B poonpojd suoiiipuoo MOU j o
lii.Himii.MAU.i oqi ,<||ii|.i.nM .ijnsiMiii pm; in.modulo.> q.nr.i )soi o | X|OIEJE(1.IS ')SJI( SI ojnpooojd
siq J , • ojnpooojd j o d o j d oqi MO||OJ qoiqM soiuujEjSojd iuouido|o.\o[) luouoduioo ui pojo.visuE
ll.i.K) o.\i!i| i m p -ill M o q i i n : .i.i|qir; ) M| p.»)SE sin 11 (S.)llll . u p |n | | B A | . I I M \ ; I Ilil-.IV" ) ' g •( |
Aid:i>i s.aoHinv
•uiEOJisdn oqi SB iiui||oJ|Uoo SB )<m uoijo s| loo)).i uiE.uisu.v\op oqi
q S n o q q E 'uiEoJisdn oqi SE ||OA\ SE pojopisuoo oq isnm l u o u o d m o o uiBojisuMop oqi suoiiipuoo
OUlifUO JO UI111B|11U1IS Jjl.1 l.l.1.1.ill.) A|.llH]OS()B Jo) 1EI|1 p.)pil|0UO0 oq UBO )l JjUlllHoJOJ .11)1 11111.1 .|
•sioojj.) J o q i o J O o|iiojd ojniEJoduioi oi 0 | q E i n q u i i B
'p,H| U1EJ E UO SJSOl 0|EOS ||I1J |Ol| pilE S)S.1| (Opolll A\o|| |)[D.) U.)OA\|.)<| O0IIBUI.il IJ.l.ld .)|ZZI111 lllIOS
-jodns u o oouojojjip XUE p u n o j OABI) sjoqmE o q i j i JEOI( cq Hunsojoiui j q pjnoM IT -poq isoj
(BuoiiuoAtioo E u o o|E.)s | | n | (i.iuijiij.i.id ,K| HIM is,i| s]i|i -J.IAOAM)!) 'X[o)BimiJoj -soiiBJ ojnssojd
MO| IB ui soiuoo Xii;iqEisuj OJOI|M luiod oqi oiBoipui oj opEUi oq XpsEO IOUUEO sisoi | o p o j \
-suo;|Biiuii| |UB(d isoi <>i o n p o p u s ||nj uiBiqo o | l|iioij)tp si i|.)|i|M ElEp oouBiiuojjod [njosn XJOA
OAlJj UB,) odXl A\ll|l ,l|<|llll|) Jll -lUoilJOAip-lUoSjOAUOO JO S.qZZOU 4-jUq(.)dllJ<l |,1))11UI Ul) SlSOl AU)U
p|oo iBqi p u n o j uooq si;q i| •s.iuiiiii.i uo UOIIEJ.HIO opios [|nj oi so|zzou iiui||.i(liu<l AJIOIUOOS
o|qEUE.\ puB BOJB O|(|EIJBA u o )|JOM isoi S u j o XiqiqBOiiddE oqi III II ill.mi iou soop jodBd o q j
•suoinpuoo
uiBOJisUMop ()UB uiE.uisdn i|lo<| Xq poiooj'B si joujnqjoijE oqi snq [ "UiBOJisuMop poAOiu
SEA\ qinoui||o<| oqi uoi|A\ pojEoddssip qoii)A\ ojzzou 3 u q | o d o j d oqi qiiA\ u o p i p u o o IUBUOSOJ B
SuisnEO SEM poq isai oqi u o q i n o m | | o q joioofo isnBqxo o q i j o uoiiisod oqi IEI|1 p u n o j uo.jq SBI) i|
SUO'SEOOOJO j.x|iiinii E U Q -qooojos u o u s n q u i o o OSIIBO UEO qoiqM j u o u o d m o o Xjuo oqi l o u si
o|zzou JEuq o q x "XBM JEpmis B ui joujn<)joijB oqi oi poniuisiiEJi oq UBO O|ZZOU JO odXl o u i u o o
-uoo MO)J . q q n o p oqi UIOJJ AUHqBisuj -JOXE| XjBpunoq oqi qiinojqi lUEoJisdn p o u u u s u E J i
oq ismu IOOJJO oqi XjqEiunsojd -oiuos si leojqi , i | / / i m oqi in XIIOO|.IA oqi s y qooojos joujnc)
-JOIJB JJO SJOSSUI uoijo ' O I I E J ojnssojd u8;sop si; MO|oq hiiiiiun.i ' q u n luoirJOAip-iuoSjoAuoo
B ui XiipqEisui SE mojsXs iBoqoj B j o uoUBj.ido qiooius oqi pojjB OS]E UBO suoiiipuoo 0|ZZI)\
•o|qnoji oqi pojno X|.>i.i[duioo sm.minjisui oqi j o (BAomoAj -ouE|d j o z q i q s i s oqi UIOJJ
XBME IJ i) .nuns .iiiiqjni oqi j o uiBOJisiiAvop lonp oqj oiui po]josu| OJOM SOSBJ .qdnoooujjoqi
xis uoq.w XiqiqEisui OJOAOS OIUI UEJ inq -A||JOIOEJSIIES 3ui|BJodo uooq pBq qoiqM joujnqjoijE
UE UI pojjnooo s;qi j o o|duiBXo u y -qooojos j o losuo oqi osnEO UEO soouEqjnisip mBOJisdn
j o u i m 'joAOMoq 'OSBO siqi ui uoAg -X|ddns J I B uijojiun E Sjujsu oSJEis poouBAps UB qoBOJ
nil.) m,)uidii|.iA.i|> puE 'pBq XJOAISSOOXO OJE osoqi sso|un ' s u n i p p u o o ouiqjni Xq X||UB0Uiu8is
poioojjB l o u X]|Bnsn si XEJJB jozqiqEis oqi ]B opjojd Xi|.>o|.).\ oqi 'IUOUIOSUBJJB J.)1)B| siqi q i | \ \
odid u o n i s u B J l J o qiSuo) E Xq ouijiuo oqi UIOJJ poiBJEdos si JoujnqjoijE oqi ss,i|un 'oiisqEOJun
oiinb s| o|i)ojd u u o j i u n E qiiM iiuuso) ii|N| -pouiBKio uoiiBJodo XJOIJBJSIIES puB popio.XE
oq oi si qooojos jl 'ujonBd MO[|JIE oqi 01 poisnfpi; A||njojEO oq oi OABq uoimqiJisq) jonj puE
uSisop jozi|;qB|S oqi 'oui()jni oqi j o j n o ujoiiBd MO[j u u o j i u n - u o u E puB ouiiiuo u p m po|dnoo
-.isd| > uioisXs iB.iq.u E i|l|,\\ -qoo.uos JO XiqiqEisui uoi|snqiuoo oi oso|.) AJJIBJ SAEA\|« .LIB
'qiSuoJis ojiuxiui ||EJOAO ii.ii.iiiKiiqoiois E oi d n oiBJodo o] pouSisop 'suioisXs j o u j n q j o y y
•isoi ioq ojqojd
ojniEJoduioi posEiq Xpoojjoo B 1|IIA\ uosiJEtfiuoo ui "X|puooos 'puB ojnss.ud | | n j JE isoi i o q
o|i)ojd ojiiiBJodmoi iiuojiun E qiiM uosuBduioo ui 'Xpsjij ' p o j j n o u i oq o] X|oi|q si JOJJO |BUM
A\OU>) oi pojsojoiui XJOA oq p|tioM | 'siiopipuoo |EUoisuoujip-uou IOOJJOO oqi IB poionpuoo OJB
sjsoi oqi J I pin; po||ojuioo X||njo.iEo OJB 'OOUBJEOJO d[] SB qons 'suoilEJopisuoo [EOiUEqootu |[
sxinsaii i s a i maNOdwon axvuvtias JO A i n i a v u n d d v

THE FURTHER APPLICATION OF GENERALIZED
PARAMETERS TO TURBOJET PERFORMANCE
RELATIONSHIPS
By E. MACIOCE
Italian Air Force
// is well known that a non-dimensional presentation of the variation ofgross thrust,

fuel consumption and airflnw nf a turbnjet in terms of the basic independent non-
dimensional parameters for flight speed and r.p.m. allows the ready calculatinn nflhe
performance nf the engine over a wide range of operating conditions.
In this paper the non-dimensional treatment is extended to quantities which may
be deduced from elementary turbojet performance data and which can be presented
as functions of the same basic independent parameters. Further aspects of the
performance of the engine can be examined by the method of generalized parameters,
which opens a wide field of analysis of interest to propulsion engineers.
1. G E N E R A L I Z E D T U R B O J E T PERFORMANCE
W E CONSIDER a family of similar turbojets with fixed propelling nozzles. It is
permissible to neglect viscosity and heat conduction effects. T h e n assuming
constant loss factors and constant specific heats for the working fluid, wc know-
that the gross thrust S* of an engine depends on eight quantities, viz. the
atmospheric pressure />„, the total pressure and total temperature at the
compressor intake P t and T x , the r.p.m. N., gas constant gR, the air mass
flow Q, the rate of supply of chemical energy in the fuel W c , and finally on
iln- size of the engine, as represented by a typical dimension D, conveniently
taken as the intake diameter.* From the theory of dimensions, with nine
independent quantities all measured in the same four fundamental units,
here taken to be mass, length, time and temperature, there exist five related
non-dimensional groups of these quantities. In the case of the turbojet it is
convenient to express terms containing the gross thrust, air mass flow and
fuel consumption as non-dimensional functions of intake pressure and engine
speed. T h e usual five groups thus functionally related are shown in the
following equations
S* Q{gRT.)i We ll\ ND \
PtD*' l\D- ' P.LVigRTm). J
\p0'{gRT1)ij (1)
T h e proper non-dimensional form for the fuel weight flow rate (F), as a
function of the same two independent variables, when the heating value per
lb of the fuel is a constant, is F j [ P l D { g R T 1 ) i ] . A correction factor is some-
times introduced to take care of altitude effects on non-dimensional fuel
consumption. It is also to be noted that in place of the variable Pilp 0 we
may use the flight M a c h n u m b e r M or its equivalent V/{gRT 0 )* where V is
flight velocity, so long as we can fix a value to the isentropic efficiency of the
air intake (see Eq. 3) Pressures and temperatures are in absolute values.
* See page 268 for list of symbols.
260
THE F U R T H E R A P P L I C A T I O N OF G E N E R A L I Z E D PARAMETERS
When the performance o f a particular engine is to be examined, D and gR

become constants and may be dropped from Eq. (1). Again it is usual to
relate atmospheric pressure and temperature to the standard sea level
conditions p o l and T o l . Eq. (I) is then simplified and becomes
where d a = PJpw,. and 0 n = T J T n .

It must be noted that Eq. (2) is not "non-dimensional" but so far as the
study of a particular engine is concerned this lack of generality is of no
significance.
2. EXTENSION OF T U R B O J E T ANALYSIS
TO DEDUCED Q U A N T I T I E S
2.1. A number of other useful performance parameters and characteristic
quantities can be deduced from the commonly used non-dimensional groups
mentioned above and these may be presented graphically as functions of the
same two independent variables P^pQ and NDj{gRT-.)i.
Any such quantity which involves flight speed will however need a
knowledge of the intake efficiency tj\ before it can be so represented. In such
cases it is convenient to employ the intake equations in the form:
TJT 0 = (l + ^ - M ^ j
(3
P.lPo= (l + r ) J ~ M ^
V=M{ygRTm)*
2.2. Air and Fuel Flow
If we introduce C to represent the mass flow rate of the fuel (C = Fjg),
the non-dimensional parameters for air mass flow and fuel mass flow become
respectively { Q / P ^ i R T ^ g ) - and (C//° 1 /))(/i > r ] / 1 ? )-».
Dropping constants we can express Q and C in terms of our independent
variables as follows:
«Z?. '; -f( P l N
\ (4)
We cannot obtain the total turbine mass flow K = Q -f- C from the simple
addition of the non-dimensional forms of Q and C, due to the difference
caused by the factor 7\ appearing between the two.
And because we have
QTi CT. {Q+C)Ti
+
P. PlT*~ I\
the total mass flow K = Q 4- C c a n be obtained by means of generalized form
r1
KTl IP, N\
for each fixed value of T v
261
E. MACIOCE
T h e values
QT* J ^
"P. 'P.Ti
are obtained from generalized curves Eq. (4).

Since the non-dimensional form of the air mass flow is (p/f > 1 /) 2 )(/i?T , ,5)*,
we assume as the non-dimensional form of A':
KjjgRTtf
I\D*
From now on, for the deduced quantities containing K, the simplified
hypothesis 7 \ = cost will be always considered. If we consider T 1 variable
the analysis will be more difficult. T h e non-adimensionalization in such
cases involves breaking u p the deduced quantities into several terms.
2.3. Deduced Performance Data

I m p o r t a n t , easily deduced, performance d a t a are the net thrust, useful
flight power, engine gross a n d engine net power.
The net thrust, S = S* — Q Vx in non-dimensional terms becomes:
S S* QjgRT.)- V
U )
P ^ PJfr- P X D* (gRT.)-
T h e form of the thrust term is similar to that of the non-dimensional

gross thrust. T o deduce it the intake efficiency rj ih is required. T h e graphical
presentation will follow the law:
X£Xi)
T h e useful flight power is given by
WU = S V = (S* - QV)V
which is easily rendered non-dimensional by dividing through by

P 1 D i { g R T , ) i . It is noted that the same factor is used for reducing the rate
of supply of fuel energy to non-dimensional form (Eq. (1)).
W e then have
For plotting the non-dimensional parameters of net thrust and useful flight
power wc thus need only to have available the non-dimensional plots of S*,
Q and V.
2.4. Non-dimensional Gross and Net Power

T h e gross power is W* = JAW2, and net power W m = ^Kw 2 — \QV 2 ,
where w is the jet exit velocity.
262
THE FURTHER APPLICATION OF G E N E R A L I Z E D PARAMETERS
T h e gross power developed by the engine alternatively may be written

1 .V*2
T o generalize Eq. (10) it is noted that the non-dimensional forms of S*

and A' are respectively S*/7 , 1 /) 2 and K { g R T l ) i j P l D 2 , so to render the right-
hand side of this equation non-dimensional it is necessary to divide through
W { g R T . ^ D K T h e non-dimensional form of W* is thus W + K g R T ^ P ^ D 2
or dropping constants I1 T *// > I 7'J.
II* 1/.S'*\2 P.
(11
i\r\ 2 \ P J • KT\ )
We already have S*/P l as a graph of P l / p 0 and N j T \ , and have A' (at any
particular intake temperature) also as a graph of the same two quantities.
IT* can then be found from Eq. (11).
T h e net power developed by the engine, W m = W* — \QV 2 , can be found
from
Wn II * 1 QV~
(12
PXT\ ~ 1\T[ ~ 2 P j \ ^
For the last term in Eq. (12) we need the graph
QV 2 I P. N\
p.n
which involves the intake efficiency
3. Q U A L I T A T I V E C H A R A C T E R I S T I C QUANTITIES
3.1. T h e three principal quantities which reflect a qualitative aspect of the
fluid-dynamic phenomena taking place inside the engine, are the fuel flow
per unit thrust (C/S), the air-fuel ratio (Q/C) the thrust per unit air mass
flow S/Q.
Of these quantities the non-dimensional forms are derived directly,
CD
r
:i> Ii gg \ yi Q R T , S 1
S \ R T . I ' CD ', ~and
" " Q (gRT.)*
respectively. These three quantities arc all functions of the two independent
variables P,//*,, and fffTi. O n removing the constants the three following
graphs are realized
C QT
' *' ~ ' ~ fJ i P l - \ (14)
ST[ ' C ' QT[' \ p 0 ' T\]
Note that the air-fuel ratio parameter contains T-, as a factor although
QjC is a pure ratio.
3.2. Power per Unit Air Flow and Specific Fuel Consumption
T h e gross power per unit mass air flow is
W7* S* 2
m = (15)
<? 2AQ
18—120 PP.) 263

E. MACIOCE
For non-dimensional grouping it is necessary to write:

W* 1 1 / S* \ 2 PxD2 /',/->'-
(16)
2
Q g'RT. ~ 2 [ p . D ) • K j g R T ^ ' Q i g R T J -
From (16) the non-dimensional form for W f J Q is:
W* 1 II*
0 " ' RT ' °r' wltnout constants
. QY
Eq. (16) shows that the gross power per unit mass air flow may be obtained
by means of the three generalized curves already obtained for S*, A and Q.
T h e net power per unit air mass flow is lV m /Q and is given by
W W*
T h e non-dimensional form of W m jQ is W m j Q g R T v
For a given intake efficiency V2j T t is a known function of P x /p 0 and for
given operating conditions W*JQ is already known. T h u s Eq. (17) will give
W J Q in terms of P J p 0 , N j T \ and T v
For the specific fuel consumption wc shall calculate the fuel weight flow rate
per gross h.p. and the fuel weight flow rate per net h.p.
First, per unit of gross power:
F 2FK2FQ 2PC
=
iy* 5*2 y*2 'r y*2 (.'")
2FQ , 2
g if.)'
T h e non-dimensional forms of FQjS* and (FjS*) 2 are respectively FQD/S* 2
2
and F 2 D 2 1 S * 2 g R T v
Hence the relationships
FQ_ F
g*2 ' 5*2 T" •,; = X : ^ ) <w
and from the graphs of these partial quantities, given the intake conditions,
the sum F/lV* t is found.
Alternatively we m a y use the expression 2FK/S* in Eq. (18). This
requires separate reference to the value of T x before A can be evaluated.
T h e non-dimensional form is FKDjS* 2 .
O n the net power basis the specific fuel flow is F/[V m . This quantity is
found conveniently by means of its reciprocal:
W W* OV 2
F m =^f~% ^
2
Here the term W*JF is already known. T h e term Q V j2F may be made
non-dimensional by dividing by D and then in the form QV 2 jFD, can be
plotted as a function of {PJpQ, NjT^) given the intake efficiency. T h e value
of W m jF as a function of the two given independent variables is thus known.
264
THE FURTHER APPLICATION OF G E N E R A L I Z E D PARAMETERS
Note that, apart from constants, the ratio W m jF is the thermal efficiency of
the engine.
3.3. Thrust per Unit Power

The gross thrust per unit gross power is
S* 2S*{Q+C) 20 2C
+ (22)
IV* s*- s* s*
It will be found that the non-dimensional forms for the right-hand terms
are respectively 2Q(gRT l )ljS* and 2CD(gjRT,)*jS*. It is possible therefore,
eliminating the constants, to establish the following pair of relationships:
QT\ C IP, N\
; (23)
S* ' S*T[ ' \ p „ ' T\)
T h e quantity S*jW* l can thus be obtained from graphs (23) using Eq. (22).
For the net thrust per unit net power we have
_S_ S* - QV
Wm IT*, \QV 2 {
^>
It is convenient here to obtain S/P 1 as in Eq. (8) and IV„JP 1 T[ at any
particular intake efficiency and temperature, as a function of the same
independent parameters from Eqs. (11), (12) and (13).
3.4. The Efficiencies

Typical efficiencies, thermal, propulsive, overall, are basically non-
dimensional since they express ratios of homogeneous quantities, in this case
powers.
T h e thermal efficiency is the ratio of the net power to the energy supplied in
the fuel
1Hx=lVJW. (25)
Here W m / P 1 T l is already established (Eq. (12)) as a graph in terms of
independent parameters P x jp Q , N j T \ and intake efficiency. T h e non-
dimensional group form for Wc, leaving out constants, W J P X T \ , appeared
in Eq. (1). These two graphs give T^H as a direct ratio. Alternatively tjih
may be written W m jFH where H is the calorific value of the fuel; W m \ F
may be found as described in Section 3.2.
T h e propulsive efficiency r/r, is the ratio of the useful flight power to the net
power developed, i.e.
SV
'/P ,,- (26)
T h e non-dimensional form of SV is S V j P l D 2 ( g R T x ) i ; or leaving out

constants, SVjP^ T\ which may be plotted. T h e denominator of (26) is
known in the form W f J P ^ ^ . Hence by direct ratio r/p is found (for a given
intake efficiency) as a function of P,jp 0 and N j T \ .
Finally, the overall efficiency i] g , the ratio of the flight useful power to the
energy applied in the fuel, is
SV
"' it;
265
Table I. Turbojet Parameters—Basic and Derived Quantities
Factor
N - D Form with
r, • Equivalent which m a k e s Dimensions
Quantity Symbol Dimensions w?. Non-dimensional Form Constant Factors
Expression Quantity of Factor
Removed
Non-dimensional
S* £• 1
Gross thrust S* Iml-' l^m-U 1
P,D* P,40«
P.
Q(gRT l )i (gRT,).
Air mass flow rate Q mt~l QT.\ m-'t l
P,D* P,D'
Pi
3 F F 1
Fuel weight flow rate F iml l~lm H3
P l D{gRT l ). P.T.- P-DigRtT).
F C
Fuel mass flow rate C ml ' m-H1
g PXD \ R T j P.T X \ i\i)\ia\)
K(gRT x ). AT,1
Total jet pipe mass flow K ml1 mltl
to Q + C PVD* Pv P t D*
a S* QjgRT,). V £* QT. 1 1
Cn Net thrust S Imt-* I hn 'I-
S* - QV PJP P X D' ' (gRT,). Pi P\ T.\
Useful flight power w, I'-ml 3

SV
S V i. JL 1
I'm1!'
P.D-- IgRTJi P. ' T.i P l D'(gRT l )i
Engine gross power »„* l-ml '

1 S" us*y />, 1
I'm-W
2 " K APiDV
2\ KURTjk •APJ AT,i P.D^gRT^
II™* k 1
Engine net power Wm Pm. ' 11" .QV' 19Jj Yl lrhr l P
Pt&igRTJi PJ\. 2 p.' • r, P,D 3 (gRT,)\
1 Q(gRT l )i P_
2 P,D> ' gRT,
c
Fuel mass flow rate per
unit net thrust s
I 'i'
P.DXRTJ ' S
JL £
PyTJ ' S X)' IH-i
Q QlgRT t ). P j D f R T M QT,i P J \ . RT,

Air-fuel ratio
l\D* ' C \ g } I)
c Pi ' c
Table I (continued)
Factor
N - D Form with
Equivalent which makes Dimensions
Quantity Symbol Dimensions Non-dimensional Form Constant Factors
Expression Quantity- of Factor
Removed
Non-dimensional
Net thrust per unit air S P,D' I ,-i,i

mass flow Q P,D' ' Q(gRT,)i Pi ' QT.i (gRT,).
Gross p o w e r per unit air II m* '2[ 2 lira* PlD* WJ> _£._ 1
TH-
mass flow Q P,D3(gRT,). ' Q(gRT,)i P,T,i ' QT,i gRT,
N e t p o w e r per unit air I* m If * WS J 1 V* Wm* 1 V 1

mass flow Q -iv Q • gRT, 2 gRT, QT, 2 T, gRT,
T'P
Fuel weight flow r a t e per F F P,D'(gRT,)\ F P.Tti

D P
unit gross p o w e r Wm* P,D(gRT,). ' \V m * P , T , i • Wm*
a
^J
Reciprocal of
Fuel weight flow r a t e per F Wr,* QV* F P,D'(gRT,). F P.TX\
D P
unit net p o w e r \V m F 2F P,D(gRT,)i If. P,T,\ " lf»
Gross thrust per unit S* 2JC S* P,D*(gRT,). S* P,T,i
V (gRT,)i Pt-3
gross p o w e r It r * S* P,D* " Wm* wy
l\'
Net thrust per unit net S S ' - Q V S P, I P (gRT,). S^ P,T,k /i--i
power Wm (gRT,).
W„* ~ t\QV P,D* ' It •„ Pl' If,,,
Wm P,DHgRT,)i Wr. P,T.\
T h e r m a l efficiency tjtb 1
w, P,D*(gRT,)i ' W, P,T,k W.
Propulsive efficiency rjp JX \\\ P,IT- g R T , i lf„ P.TJk

I
Wr. P,D*(gRT,).' 11,,, P.T.f
O v e r a l l efficiency i/g
w. II. P,D'(gRT,). W, PxT.i
w. P,D'(gRT,). If P,T,\ if.
E . M A CIO C E
T h e generalized form of SV again is SVjP x T\ and that of W. (Eq. (1)) is

W t j P x T \ . T h u s the value of rj g is obtained by the simple ratio of the
quantities S V j P x T \ and W C IP X T\. For a given intake temperature and
efficiency there will again be relationships of the form
I-1 - ]
[po'Tif
3.5. T h e quantities deduced in the above paragraphs are collected together
in T a b l e 1. This shows the corresponding non-dimensional quantity and the
quantity with constants removed. T h e last two columns give the appropriate
factor which makes the quantity non-dimensional.
4. C O N C L U S I O N
By extending the non-dimensional approach to the performance of the
turbojet engine a n u m b e r of quantities have been derived in the form of
functional relationships of two independent parameters.
W e have considered in this category, the total jet pipe flow, the net thrust;
useful flight power, the gross and net power.
Following on from these derived quantities we have considered a n u m b e r
of characteristic performance quantities, viz. fuel flow per unit thrust,
air-fuel ratio, thrust per unit air flow, power per unit air flow, fuel flow
per unit power and the thermal, propulsive and overall efficiencies.
SOMMAIRE
La presentation sous forme adimensionnelle des variations de la poussee brute, de la
consommalion de combustible et du debit d'air d'un lurbnreacteur en fonctinn des
parametres adimensionnels de base (vitesse de vol et vitesse de rotation) permet le
calcul des performances sur moteur dans un grand domaine de variation des conditions
de fonclionnement.
On exlrapole ici la methode non-dimensionndle a des grandeurs que Ton peut
deduire des performances elemenlaires de la machine et qui s'expriment au moyen
des memes parametres independanls de base. On pent ainsi examiner de nouveaux
aspects des performances de la machine au moyen de ces parametres generalises,
nuvrant ainsi un nouveau champ d'investigation aux ingenieurs charges des etudes de
la prnpulsion.
REFERENCES
CAPETTI, A. Les performances des moteurs d'avions, Selected Combustion Problems II,
A.G.A.R.D. Bulterworths Scientific Publications, London, 1956.
DURHAM, F'RANKLIN P. Aircraft Jet Power Plants; Prentice-Hall, Inc., New York, 1951.
MACIOCE, E. "Rappresentazionc adimensionale per il calcolo delle prestazioni c
dclle grandezze caratteristiche dei turbomotori acrci," L'Aerotecnica, Ottobre 1952.
SYMBOLS „.
Dimensions
g Acceleration of gravity l+lt~2
y Ratio of specific heats —
pn Air static pressure /"'m1.-2
/>„! S t a n d a r d air pressure at sea level /"'m1/-2
T0 Air static temperature T
268
THE F U R T H E R A P P L I C A T I O N OF G E N E R A L I Z E D P A R A M E T E R S
T01 Standard air temperature at sea level T

T, Air total temperature at compressor inlet T
w J e t velocity ll-1
D Turbojet linear dimension I1
F Fuel weight flow rate /'m'r3
C Fuel mass flow rate w1/-1
Q Air mass flow rate m}t~x
A J e t pipe mass flow rate (Q 4- C) m 1 !^ 1
M Mach number
N Engine r.p.m. t^ 1
Px Air total pressure at compressor inlet l~ m l~ 2
l l
V Flight speed Pr1

R Gas constant for air I- T~ x
S* Gross thrust Pm1!-'
S Net thrust /'m1/-2
II r Fuel energy supply rate Pm^r3

If7* Engine gross power An 1 . 3
Wm Engine net power / 2 ;« ] /~ 3
Wu Useful flight power I'-nPl :!
rp. Intake efficiency
rju, Thermal efficiency
>7p Propulsive efficiency
7] g Overall efficiency
dtl Intake total pressure relative to standard sea-level pressure
6tl Intake total temperature relative to standard sea-level temperature
f{ ) Continuous function(s) of ( ).
269
PART III
FLIGHT TEST PROBLEMS
FLIGHT TEST DEVELOPMENT OF
SUPERSONIC ENGINES
By R O Y E . P R Y O R
Flight Test Operations, General Electric Company, Evendale, Ohio, U.S.A.
A brief review is presented of the procedures, organization, training and method of

operating necessary to conduct a flight test program on a supersonic engine. The
reasons for engine manufacturer flight programs are discussed, along with an
analysis of concurrent engine and airplane development. Snme typical prnblems are
presented as a portion of a typical and representative development test program.
Finally, the contributions of this type prngram are presented as conclusions to the
paper.
INTRODUCTION
T H E FLIGHT test development of supersonic engines is a test program tailored
to meet the requirements of simultaneous development ofa complete weapon
system. A few years ago, when airframe designers and manufacturers were
struggling to reach Mach 1, they could research the industry and take their
choice of several proven engines capable of providing the thrust required
over the planned flight envelope and design their aircraft around one of
these engines. Thrust to weight ratio between these engines was approxi-
mately equal and, in general, performance capabilities were similar. As the
state of the art advanced through research and development and the urgent
need for high performance weapons systems was stressed by the using
commands, designers started thinking in terms of aircraft that would be
capable of flying at Mach 2 or more and at altitudes far in excess of 50,000 ft.
As design studies progressed it was found that several airfoil shapes could be
used to satisfy the new performance criteria. However, the choice of engines
to meet these new requirements of Mach number and altitude was extremely
limited or non-existent. T o develop a weapon system with major advance-
ment in performance capabilities, the engine manufacturer was required to
assume a major part of the responsibility in designing and developing a new
engine with high thrust to weight ratio capable of operating over a flight
envelope which was more than double that of previous applications. T h e
flight test program covered in this paper is designed lo meet the requirements
of development testing lo provide a fully qualified engine to mate with a
weapon system program of this type.
WEAPON SYSTEM DEVELOPMENT T I M E TABLE

Let us assume for the purposes of discussion that, at some point in lime, the
requirements for a weapon system are made in the form of a mission
capability including specific altitudes and Mach numbers. Perhaps several
aircraft manufacturers and engine manufacturers selected to make design
studies will begin to lay down the basic design requirements for their
respective products. Once the aircraft manufacturer and engine manu-
facturer have been selected, these two companies will combine their efforts
273
R . E . P RYO R
toward producing that weapon system which is capable of meeting the basic
requirements. T h e aircraft manufacturer will outline the time table for
development of the aircraft itself and coordinate this schedule with the
engine manufacturer. T h e engine development cycle, in addition to factory
test and altitude tank testing, will require advanced flight testing in existing
aircraft in order to have a fully qualified engine for the prime application.
A typical time table might look like thr.t shown in Fig. 1. It can be seen
A R JWFT APPLICATION 'AIRCRAFT* 1 -
ENGIKEERIHG-I
[MANFT
FACTORY TEST
IFLYIMQ TEST eUD'
JTEST A / : -A-';
XT/VC/S'
I
1
1
Pig. 1.
Aircraft and engine development cycles
that the development cycles to be completed on the engine are planned to

meet the overall weapon system program requirements.
D u r i n g the engine design phase m a n y component tests are conducted to
insure m a x i m u m reliability and operational suitability. Soon after testing of
the power plant as a complete unit in factory test cells, it will start simulated
altitude testing in altitude wind tunnels and altitude tanks. This testing,
although completely idealistic from an altitude and Mach number point of
view, does not produce completely satisfactory results as far as true flight
operation is concerned. Flight test programs are planned to cover the entire
(light m a p and verify the results obtained from other tests.
It should be noted that the engine development cycle calls for flight test
much earlier than the completion of the first aircraft. This means that flight
testing must have begun and been completed prior to the existence of the
true application aircraft. For this reason, the flight test program is planned
to cover all possible areas of operation of the engine prior to its actual
installation in its prime application.
F L I G H T T E S T PLAN AND PROGRAM

A minute examination of the mission requirements of the new weapon
system indicates immediately that the engine is required to operate satisfac-
torily over a wide range of speeds and altitudes. These may be shown by
reference to Fig. 2. Here it is seen that the airspeed range from zero to high
M a c h n u m b e r is covered as well as a complete altitude range from sea level
to high altitude. In addition to the normal flight envelope as presented here,
the engine must be capable of delivering high take-off thrust and good
performance at low altitudes and low Mach numbers. T h e cruise condition,
which in this case we will assume to be subsonic, must also be economical
274
F L I G H T TEST D E V E L O P M E N T OF SUPERSONIC ENGINES
a n d adequate for the aircraft mission. And finally, good high speed perfor-
mance is required to consummate the complete mission profile. Complete
thrust modulation including afterburner range of operation is required
throughout the flight envelope.
Fig. 2.
Engine flight envelope
T o test this engine fully throughout its entire operational area, flight
testing is started as soon as adequate experience is gained in ground and
simulated altitude testing. A common and accepted practice in the aircraft
Fig. 3.
Flying test bed in flight
industry is to utilize flying test beds as the first flight medium for any new
development engine. Typical of such test beds used in the United States are
aircraft similar to that shown in Fig. 3. This is a B-45 and incorporates a pod
in which the engine is suspended under the aircraft (Fig. 4). This pod is
retractable for landing and take-off purposes into the bomb-bay of this
particular aircraft and makes an extremely good test bed due to its relatively
275
E. PRYOR
m g . 4.
Nacelle installation on flying lest bed
MACH
Fig. 5.
Flying test bed flight envelope
high altitude, its reasonable M a c h numbers and its instrument carrying

capability.
Flying test beds are selected as early flight aircraft for a n u m b e r of reasons.
It is possible to obtain flight testing prior to the preliminary qualification of
an engine. Qualification is not a requirement for flight in a test bed, although
the preliminary qualification is a prerequisite to flight utilizing a new engine
as a prime power plant. This type of flying can begin during the course of
simulated altitude testing as well as additional factory testing. T h e reliability
a n d acceptability of the product is completely dependent on adequate test
276
FLIGHT TEST DEVELOPMENT OF S U P E R S O N I C ENGINES
programs. In our assumed case, we may say that the flying test bed selected
would utilize or be capable of utilizing an envelope approximately equivalent
to that shown here (Fig. 5). From this chart it can be seen that approximately
40 per cent of the Mach number range can be covered throughout approxi-
mately 6fJ^70 per cent of the altitude range. Arguments can be advanced
that this is of small value due to the fact that the extreme high speed and high
altitude characteristics of the engine cannot be evaluated with this type o f a
vehicle. However, it must be remembered that the aircraft capable of flying
at these altitudes and Mach numbers does not exist at this point in time.
Fig. 6.
Test aircraft A
Also, by utilization o f a relatively low speed Hying test bed, flying and flight
testing can be gotten under way much in advance of the availability of flight
test information if qualification had to be completed prior to first flight.
T h e types of test that can be accomplished in this type o f a test bed are
numerous. Obviously basic performance, both thrust and fuel consumption,
can be immediately evaluated on an extremely detailed basis. In addition,
one of the greatest assets of the flight test program at this point in time is the
evaluation of the control system as it operates under transient altitude and
Mach number conditions. Windmilling data relating to windmilling speeds
at various altitudes and Mach numbers and general compressor information
can be obtained and finally, and possibly most important, is the fact that ait-
starts and the air startability of the engine can be evaluated and developed
further, if required.
T h e relatively small flight envelope of the lesl bed can be evaluated in a
short period of time to provide valuable information in the early develop-
ment program. After completion of the preliminary qualification, flight test
vehicles utilizing the new power plant as a prime source of power are used to
further expand the flight envelope. A number of such aircraft are studied
and ultimately two prospects stand out as being the most desirable from an
installation point of view as well as from the performance point of view of the
basic airframe itself. T h e first of these selected again is a relatively slow speed
aircraft but gives us an opportunity to evaluate a slightly expanded envelope
in a true flight article. A typical such flight test vehicle utilized is that
pictured in Fig. 6. This aircraft might have a flight envelope in reference to
our new engine envelope approximately as pictured in Fig. 7. You will note
that we have now slightly enhanced the altitude testability with this vehicle;
however, ability to test transient conditions is greatly increased.
277
E. PRYOR
T h e second aircraft selected would probably be a higher speed aircraft

more closely approximating the full Mach number capability of the new
engine. It would be capable of slightly higher altitudes, again more closely
approximating the final altitude at which the engine will be required to
operate. A typical type aircraft of the twin engine variety which is ideal for
this type of testing is shown in Fig. 8. This aircraft would have a flight
4 ENGINE
** ENVELOPE
A
fi\
5
t- // i
i
-,
•i i
. h
li.
FLYING TEST
BED
' '
Fig. 7.
Test aircraft A flight envelope
Fig. 8.
Test aircraft B
envelope in reference to our new engine envelope approximately as shown in

this graph (Fig. 9). I n a program such as we are interested in, these two
aircraft would be started in the flight test program as early as possible
following preliminary qualification of the engine. Considerable modification
would be involved in both aircraft since they were not originally intended
for flight with this particular power plant. Some of the items that could be
278
tested would of course be the relatively low speed characteristics of the engine
in the second vehicle. In addition, maneuver effects upon the engine in
operation could be studied. Effects of angle of attack, distortion and ability
to make wave-off's could be studied, as well as air startability and evaluation
of specific types of ducts. T h e evaluation at higher flight speeds of wind-
milling characteristics in the twin engine aircraft as well as air startability at
the higher Mach numbers and higher altitudes could be evaluated. Further
information can be gained in regard to controls evaluation, performance in
— -TEST AIC A
FLYING TEST
BED
Pig. 9.
Test aircraft B flight envelope
the thrust and specific fuel consumption areas. In our example, these aircraft
could and would be utilized to study every aspect of engine operation from
the very low speed to the very high speed and from the very low altitude to
the very high altitude type of operation.
Ultimately, it would be desirable of course, to evaluate the engine
completely in its approved and chosen application. However, in our case
for the moment we might consider the fact that once the original requirements
and specifications had been completed for the engine, additional aircraft
might be designed around the same engine even though its original intent
was for an entirely different aircraft.
A second aircraft with less complexity could be made available at a point
in time earlier than the original application. In this case, additional test
programs would be started utilizing the new application aircraft. We might
also assume that its envelope might be very similar to the one noted here
(Fig. 10).
Here is, indeed, a rare opportunity to evaluate our new engine almost
throughout its complete flight envelope. With this application aircraft
available it would be completely possible to again evaluate controls operation,
afterburner operation, maneuver effects, angle of attack, distortion and many
other aspects not investigated prior to utilization of this aircraft. U n d e r these
test conditions, information can be gained to the extent that the engines can
279
R. E . PRYOR
be fully evaluated and those modifications required can be designed and

incorporated into production articles, prior to the time they meet large scale
production.
With the advent of the planning indicated we might re-examine now the
weapons system development time table specifically, and in detail, the time
table associated with the engine—here it can be seen that all portions of the
testing now intermesh to produce a usable engine at the same point in time
TEST AjC A
TEST A/C 8
FLYING TEST
BED
Fig. 10.
Application aircraft # 2 flight envelope
AIRCRAFT APPLICATION AIRCRAFT " 1

' r rTT-ir^.afF.-u
L
r-.^~-Ba« REDESIGN
ENGROg
ENGINE ENGINEERING
MANF;
' FACTORY TEST

| FLY TEST BED |
BASIC DESIGN
| TEST A/C 'A-
I |TEST AJC'a-
m
APR AfC n 2'
Fig. 11.
Aircraft and engine development cycles
when the aircraft becomes available. You will also note on this chart (Fig. 11)
that the additional application aircraft has been introduced and is shown in
fact to be actually available prior to complete availability of the first
application aircraft.
W h e n flight test of the application aircraft starts at the aircraft manu-
facturers facility, engineering liaison will be maintained with the aircraft
280
manufacturers to obtain the greatest amount of information a n d experience

possible from the testing they perform. T h e greater number of aircraft
available to them increases the total number of test hours flown considerably
over what can be flown wilh the aircraft available to the engine manufacturer.
For this reason, plans must be put into effect early in the program to enhance
the engineering availability to those people and those organizations at the
aircraft manufacturers plant, as well as the Air Force and Naval installations
where these aircraft will be evaluated. This program serves a two-fold
purpose. First of all, it permits the greatest degree of intelligence gathering
on the engine. Secondly, it makes available to the aircraft manufacturer
competent engineering personnel to aid them in the solution of problems,
encountered as associated with the engine manufacturer's flight test programs.
O R G A N I Z A T I O N A L PLANNING
T h e organization necessary to accomplish these flight programs consists of
two groups, one at the factory to furnish engineering a n d other technical
support and the operating unit at the remote site, capable of operating the
aircraft themselves to produce the necessary data, evaluation and tests to
,
X
QUALITY CHIEF TEST INSTRUMEN PROOjfiEMENT]
ENGHaEERNG MAINTENANCE ANO
CONTROL PILOT TATlCN CONTRACTS i
BAILMENT ENGINE SHOP TEST DATA

AND
FLT. TEST
" TEST CELL
PIOTS REDUCTIONS
CUSTOMER \ A/C A/C INSTRU-

L - j P t RAT ,ONS
FLT TEST IMAINTENANCEJ MENTATION
INSTALLATION
DESIGN
Fig. 12.
Flight test organization chart
meet the program schedules. In the case of our test programs our test flying is
accomplished at Edwards Air Force Base in the Mojave Desert of California.
This organization consists of a director of the flight test operation plus the
engineering operations, maintenance, data analysis, instrumentation and all
other contributing organizations necessary to operate a major flight facility
(Fig. 12). T h e organization we use consists o f a project engineer assigned to
a specific aircraft and made acquainted with every portion of the detailed
program being flown on that aircraft. Supporting these engineers will be the
necessary maintenance people to keep the aircraft and engine in top flying
condition, as well as those specialists necessary to install and maintain
instrumentation equipment for the best possible return of information for
281
R. E . PRYOR
hours flown. T h e engineering test pilots and the operations people maintain
the flight schedules in liaison with other activities at the Air Force Base.
T h e factory organization (Fig. 13) is established in such a manner that we
have full engineering coverage to maintain counterparts of each of the project
engineers at the remote location. These men operate effectively as two-man
teams with the remotely located project engineers and are assigned specifi-
cally to a given aircraft. I n addition, a data analysis and instrumentation
organization is maintained at the factory to enable engineering personnel at
MANAGER
1
DEVEL
- 1
[CUSTOMER i CCXMMEROAL REL l ABUT Y 1
REVIEW AND
OPERATION
(CONTROL AND
ENGINE
ENG'R - G ENG'R'G EVALUATION | SUPPORT
Fig. 13.
Factory organization chart
that location lo get the greatest utilization of the data transmitted from the
remote site. A second responsibility of the factory flight test group is the
customer flight test engineering organization and is associated with that
liaison mentioned previously with the aircraft manufacturer which takes
place once the prime weapons system goes into flight status. A third group
is responsible for reliability review and coordination. This organization is
primarily devoted to evaluating all information relating to the engine and its
operation which makes or determines the engine's functional reliability or
operational suitability. In addition they analyze maintenance and opera-
tional procedures to insure complete and adecpiatc instructions for the future
using commands. They also evaluate the tools of the maintenance operation
to determine if they arc suliicient and adequate to do the j o b .
This entire organization, both remote and factory, is designed primarily
to enhance the flight testability of the engines and to produce the greatest
amount of information commensurate with good practice from the smallest
amount of test flying time possible.
Training of personnel for both the remote and the local organizations is
of vita I necessity In any testing program on any type of engine. This is much
more important when it comes to testing supersonic type power plants.
Engineers trained to do their jobs in the formulation of test programs and in
the evaluation thereof begin their training early in the program during the
time when the engines are still in the design phases. M a n y of the engineers
are trained in the testing organizations where the engine is evaluated within
the factory or within the simulated altitude facility. T h e training program
is generally one of association with the engine itself and the type of operation
accompanying it. Usually the better flight test engineer and project engineer
are those who have been associated with flight test programs previously and
who are now only required to learn about a new engine design. Their close
association with the design engineers and the development engineers produces
282
a broad understanding of the overall engine and enables them to make a

better evaluation of the problems as they arise.
Pilot training is a completely different phase of the operation. It is
associated with, and very similar to, the type of training the engineers
undertake in relation to the technical details and operational characteristics
of the engine. Minimum requirements for pilots include 1000 hr ofjet fighter
type aircraft and a minimum of 1500 hr total time. Selection of pilots is
based upon past test programs. T h e pilots selected for testing a supersonic
engine must be completely current in the aircraft they arc to fly, including all
test aircraft and timely check-outs in the new application aircraft. They
should be familiar with the instrumentation systems, engine operation and
performance and aircraft capability.
Crew training is not quite as detailed in a technical nature, but high
proficiency is required to get the most out of available test hours and to
assure ready availability of the aircraft for flight tests as required. T h e crews
should be familiar with engine assembly and disassembly procedures and
technical aspects of the engine as required for maintenance and operation.
In regard to data analysis no additional training is necessary except in
those methods of data collection and data reduction that arc peculiar to
supersonic vehicles and advanced d a t a systems.
METHODS AND O P E R A T I O N
All flight test programs are planned in a broad sense in the factory. From
requirements imposed on flight test by the various design sections, test
programs are evolved in detail, including the points to be flown and type of
operation to be used. This program is then transmitted to the remote facility.
At the remote facility the program is reviewed, modified to make it
applicable to the specific aircraft and put into operation as quickly as feasible.
Necessary instrumentation is installed, recording equipment is calibrated
after installation, and is set u p to extract the parameters deemed to be
advisable to the particular test in question. T h e test is then flown, utilizing
the pilots and the crew people previously trained for this purpose, and data
is generated which is reviewed on the spot for consistency and reliability.
T h e data generated is ultimately copied, tabulated, plotted and transmitted
back to the factory where it is analyzed either by the flight test engineers
or the project engineers. This data, or the reports generated therefrom,
represent the input of information to the design groups. This input is used
for two purposes primarily. First, to evaluate the new design or secondly,
to determine what is required to precipitate a new design with the ultimate
objective to produce that product which meets all requirements for the entire
weapon system.
DATA AND I N S T R U M E N T A T I O N SYSTEMS

T h e data collecting methods have improved rapidly in the past few years
from the days when aircraft were tested purely on the basis of pilot comment
from the observation of aircraft and engine characteristics and a few
instruments made available in the cockpit. While the pilot comments and
observed information are still very important parts of the data gathering
system, other methods are now used, such as photopanel, oscillograph,
283
R. E. PRYOR
airborne tape recorder or telemetered information to ground stations. This

latter system is gaining a great deal of prominence and is being expanded to
the point where data is not only collected automatically on magnetic tape
but is reduced through standard procedures by the use of electronic com-
puters. T h e latest systems available for collection of d a t a by magnetic tape
possess a tremendous degree of flexibility in that, once the data is recorded,
the engineer still has the option to effectively rerun the test by simply
rerunning the tape at the speed most advantageous for his analysis of the
program. T h e test can be rerun faster, slower or at the original speed to
permit analysis and investigation of each minute difference in the record.
F L I G H T TEST C O N T R I B U T I O N
T h e contributions of the flight test operation to the development of a new
engine are many. A few have been selected here to demonstrate the
contributions to be realized and to point out some of the areas where flight
testing has assisted in the overall evaluation or improvement of an engine
program.
.ESTIMATED COMPRESSOR
', SURGE S
X V *w*
\ ^ 100% ACTUAL
-^CORRECTED
EST COMPRESSOR SURGE' 1

COMPRESSOR INLET TEMPERATURE
Fig. 14.
Expected surge regions
A typical example of the type of problem that can be solved is represented

by Fig. 14. It will be noted here that the horizontal line represents 100 per
cent r.p.m. It is a fixed r.p.m. position for military power and is plotted vs.
compressor inlet temperature. T h e diagonal line represents corrected engine
speed. The two parabolic sections al either end of the inlet temperature scale
represent those areas that were calculated to be surge regions. This is typical
of the high and low corrected engine speed stall areas that might be exhibited
in a supersonic engine. 11 should be noted that the corrected engine speed
curve falls outside the stall region when the physical engine speed is
maintained at a constant 100 per cent.
In flight test (Fig. 15) it was discovered that the stall margin at the high
and low corrected engine speed areas was inadequate over the entire flight
ma]), due to inlet duct matching with a particular application and production
tolerances. A review of the overall problem indicated that minor improve-
ment could be m a d e by changes to the inlet duct. However, a major
284
FLIGHT TEST DEVELOPMENT OF S U P E R S O N I C ENGINES
improvement was accomplished by minor changes in the engine control

system (Fig. 16). T h e changes incorporated schedule physical engine speed
as a function of compressor inlet temperature to increase the stall margin in
the high and low corrected engine speed areas.
,EST COMPRESSOR
(SURGE
ÂCTUAL COMPRESSOR SURGE
!00% ACTUAL
ACTUAL COMPRESSOR SURGEj
EST COMPRESSOR SURGE \ '

Fig. /.;.
Actual surge regions
A second program that can be evaluated and solved in flight test is that of
afterburner operation and its effect on the main engine. In this example it
should be pointed out that the idealistic afterburner would be one such that
the thrust increase would be a direct extension of military power as the
.EST. COMPRESSOR
V \SURGE
/
ACTUAL COMPRESSOR SURGE
^ 100% ACTUAL /
^•^CORRECTED
ACTUAL COMPRESSOR S U R G E ^ S ^ \
EST. COMPRESSOR SURGE " \ \
Pig. Ki.
Corrective action to avert surge
throttle is advanced. This cannot be completely achieved in practice but it

can certainly be approximated. Complete modulation of the afterburner
throughout the entire envelope is a necessity as pointed out earlier, due to
the fact that it is necessary to operate in a very wide regime of Mach numbers
and altitudes with the use of afterburner. In addition, wc must have an
optimum afterburner cruise which can only be achieved through use of a
285
R. E . PRYOR
constant turbine discharge temperature. This requires a variable exhaust

nozzle and nozzle control system, as a part of the afterburner. Also, the
effect of afterburner operation on the main engine in the high and low
corrected speed areas as indicated in the previous figures is of great importance.
If speed drivedowns are excessive during afterburner lights at high Mach
numbers, or if the engine overspeeds when coming out of afterburner at low
M a c h number, a stall can be encountered. T h e afterburner flight test
program on any supersonic engine includes not only the development testing
of the afterburner but, equally important, the evaluation of its effect on
main engine operation.
O n e of the greatest assets of a flight program becomes available when it is
necessary to make component design changes either for product improvement
or for the solution of service generated problems. It is on the basis of
evaluation in flight of these component changes, prior to their production
introduction, that flight test has one of its greatest attractions. It is our
practice to fly a complete evaluation on any component change prior to the
design change proposal for production.
After completion of initial development testing it is highly desirable to run
an accelerated service test program to test and demonstrate servicability and
reliability of the engine in the application aircraft. A program of this type
points out many areas where minor changes would be desirable to improve
the operational suitability.
Flight Test Operations is also responsible for the development of opera-
tional and maintenance procedures as applied to the new variety of engine.
T h e operational procedures to be utilized in service on this engine should be
developed at this point in time. T h e proper maintenance procedures, use
and evaluation of the tooling, and all other facets relating to ground and
flight operation of the new engine are checked by flight test.
CONCLUSION
I n conclusion, I should like to review briefly the major accomplishments to be
performed by flight test of supersonic engines. In the first place, there is
definitely a "state-of-the-art" advancement to be achieved by this type of
testing prior to actual flight in an application aircraft. Secondly, advanced
engine availability for aircraft of similar capabilities is possible and the engine
can be made available for new applications in the same M a c h n u m b e r and
altitude capability. And thirdly, a proven flight test method and system is
now available for further "state-of-the-art" advancement.
SOMMAIRE
Uauteur passe brievcment en revue les precedes, V organisation, la formation et les

methodes pratiques necessaires a la poursuite d'un programme d'essais en vol de
moteurs supersoniques.
II discute les raisons qui incitent les cnnstrucleurs de moteurs a etablir des
programmes de vol, et analyse les phases du developpement parallele du moteur el
de ['avion. II presente ensuite quelques-uns des problemes typiques qui se posent a
['occasion d'un programme classique d'essais en cours de realisation. L'autetir
conclul en exposanl la valeur de ce programme type.
286
REFERENCES
Flight Test Engineering M a n u a l — M a y , 1951, A F Tech. Rep. N o . 6273, Revised
J a n u a r y 1953.
Flight Test Engineering M a n u a l , General Electric C o m p a n y GEI-24277.
Installation H a n d b o o k for T u r b o - j e t Engines, General Electric C o m p a n y GEI-41450.
PERKINS, COURTLAND D . a n d D U R B I N , E N O C H J . Vol. I — P e r f o r m a n c e , Vol. I I —
Stability a n d Control, Vol. I l l — I n s t r u m e n t a t i o n Catalog, Vol. I V — I n s t r u m e n -
tation Systems, AGARD Flight Test Manual.
DISCUSSION
W. A. FLEMING (N.A.C.A. Cleveland): In this paper Pryor has very comprehensively and
effectively described the purpose, the organization, the methods of conduct and the contri-
butions of flight test development of supersonic turbojet engines. In his discussion he has
credited the flying test bed portion of (light test development with a more important role in
the engine development programme than I believe it really holds.
Flight testing might be considered as the ultimate point in an engine development pro-
gramme, for it is here that the engine is exposed to its natural (light environment and is
finally installed and operated in the confines of the aircraft that was designed for its use.
Important as it is, flight testing is but one of several equally important components of an
engine development programme, the other principal parts being component testing, sea-level
testing and altitude facility testing. Each type of development testing has specific areas in
which it is most effective and consequently there is a particular time period during the engine
development schedule when each can be applied mosl efficiently.
Modern turbojets are as much a component of present day supersonic aircraft as a com-
bustor and a compressor arc components o f a turbojet engine. Limitations often exist when
attempting to apply engine component test results to predict full-scale engine performance
and operation. We might likewise expect the existence of similar limitations when using
(lying test bed data to predict the manner in which an engine will perform and operate when
installed in the final aircraft.
The weakness of the (lying lesl bed in providing the conditions of natural environment that
(light testing is intended to simulate, stems almost wholly from the substantial effect that inlet
flow distortion has on the performance and, particularly, the operating characteristics of a
turbojet engine. Both the magnitude and the extent of the flow distortion can vary substan-
tially from one airplane to another as the inlet duct location and geometry change. This
sensitivity of inlet flow distribution to inlet geometry is illustrated in Fig. 17 for three inlet
duct arrangements. From these pressure distributions it can be seen how both the magnitude
and the extent of radial and circumferential flow distribution differ substantially from one
inlet to another. Furthermore, although not shown here, it has been found that the effect of
angle-of-attack on the flow distribution at the engine inlet differs measurably from one type
inlet to another.
It is obvious that in the design of any aircraft, best inlet design practice should be exercised
to minimize the inlet flow distortion. Clever as the inlet designer may be, the fact remains
that the engine must be able to digest some measure of inlet flow distortion.
Two of the major contributions credited to the flying test bed portion of engine flight test
development are the definition of engine operating limits and the development of the control
system to permit stall-free operation and reliable restart capability. Let us quickly examine
the effect that inlet flow distortions such as those shown in Fig. 17 can have on these major
operating characteristics.
A typical example of the inlet flow distortion effect on the altitude operating limits of a
turbojet engine is shown in Fig. 18. It can be readily seen that inlet flow distortions can
lower the altitude limits of an engine by as much as 15,000 to 20,000 ft. To obtain a repre-
sentative indication during development testing as to what operating limits the engine can be
expected to have in the final aircraft, the engine operating limits must be obtained with an
inlet flow distribution simulating that of the final aircraft.
To illustrate how inlet flow distortion can effect the development of the engine control,
some typical fuel flow stall limits are shown in Fig. 19. Presence of the inlet flow distortion
reduced the stall margin by 20 to 25 per cent. Consequently, a control developed with
287
R. E. PRYOR
NOSE INLET BOTTOM SCOOP INLET SIDE SCOOP INLET
flP/PAV • 10% <1P/P1V • 13% A P / P . „ -16%
Fig. 17.
Typical flow distortions with various types of supersonic inlets
60
ALTITUDE.
THOUSANDS OF FT
/
/ SURGE FREE
/ OPERATION-
DISTORTED FLOW
L JL
80
J
85 90
I
95
I
100
L
105
ENGINE SPEED. PERCENT
Fig. 18.
Effect of inlet flow distortion on engine operating limits
120
100
80
FUEL FLOW.
PERCENT
60
TYPICAL
ACCELERATI
SCHEDULE
20
L. _1_
80 85 90 95 100
ENGINE SPEED. PERCENT
Fig. 19.
Effect of inlet flow distortion on fuel flow stall margin
288
nearly uniform inlet flow, thus having an acceleration schedule similar to the one indicated on
the figure, would be most unsatisfactory with distorted inlet flow. Conversely, a control
designed to be compatible with large flow distortions would react much more sluggishly than
necessary in the presence of small flow distortions. Thus, as with the operating limits, proper
simulation of the inlet flow distribution of the final aircraft is necessary to effectively develop
ihe engine control.
The type of inlet flow distortions that will exist in the final aircraft can be determined
readily from model tests in a supersonic tunnel. These distortions can then be simulated by
the use of a complex flow diverting device installed ahead of the engine. Such a flow diverting
device can be installed in an altitude facility wilh relative ease thus making it possible to
simulate not only altitude and flight spied but also the inlet flow distribution of the final
aircraft, or any others fur that mailer. On the oilier hand, installation ofa complex flow-
diverting device in a flying lest bed would be difficult .mil in many rases impossible. In view
of the comparative ease with which the pertinent engine environmental conditions can be
simulated in an altitude lest facility, it appears to be an obvious choice over the flying test bed
for most engine development testing al simulated flight conditions.
The altitude test facility also holds an advantage over the flying test bed with regard to the
rate and the accuracy with which lest results can be obtained. To obtain engine performance
and operational data accurately requires extensive and highly sensitive instrumentation,
carefully controlled conditions and a great deal of engine operating time. For example, an
engine properly instrumented for a detailed performance evaluation contains as many as 150
pressure tubes and 50 thermocouples. Similarly, an engine instrumented to obtain detailed
transient measurements throughout the compressor to record its response to armament firing
would require no less than 20 lo M) channels of extremely high response rale transient
pressure and temperature instrumentation. Flight installations have their limitations both
as to the amount of instrumentation that can be handled as well as the refinement and
accuracy of such instrumentation.
Comparing the rale al which engine development can proceed with these two testing
techniques, il is found that the pace of engine development by flight testing is relatively slow-
compared to that in an altitude facility. Fur example, as many as 1600 engine hours have
been logged in a single altitude facility during a one-year period. 11 would be impossible to
equal this effort without a fleet of flying lest beds.
To conclude, the justification for engine lliidil lest development is that it provides the
natural environment of speed and altitude. To be fully representative it must also provide the
installation environment of the aircraft in which the engine is to be used. Because of the
relative ease with which the proper inlet flow environment can be simulated in an altitude
facility and llie rate al which engine time can be lugged, the altitude facility is an obvious
choice over the flying test bed for most detailed evaluations of engine performance and
operating characteristics during engine development testing. The final slip in the engine
development programme must be to make spot checks ollhe engine performance and opera-
ting characteristics by flight tests, with the engine installed in the aircraft designed for its use.
C.H.LEE(Handley-Page Ltd.,London): In this paper,we have heard much about the bad
effects that intake flow distortion has on engine behaviour: both operational behaviour
(surging, etc.) and also reliability (e.g. hot spots causing trouble) being adversely affected.
From this, the speakers have drawn the moral that the airframe designers ought to design
better intakes.
While I agree that such a conclusion is fair, it is not the only conclusion that may be drawn.
As one concerned with airframe design, 1 would say lhat there is now emerging a new
desideratum for engine design, namely engines must be as tolerant as possible of intake
velocity maldistribution. It seems to me that engines are nowadays lu-c inning very tempera-
mental and I should like to know how much performance would be losl if some tolerance were
designed in deliberately. I do not imagine one could draw up a precise "balance sheet",
but some indication might be obtainable.
I can think of two English jet engines, very similar in size and performance, one is tolerant
and the other much less so; there docs not seem to be anything to choose in the overall
efficiencies of these two engines. This seems to suggest that tolerance can be obtained fairly
cheaply.
For some aircraft, unsymmelrical intakes may be very advantageous to the airframe, hence
the emergence of "tolerance" as a desirable engine characteristic. The airframe designer
289
R. E . PRYOR
will, clearly, do all he can, but, inevitably, there will Ix- some flight cases when some flow
distortion cannot be avoided. A designer might, then-fine, find that he preferred a tolerant
engine that wasjust a little less efficient than a rather "touchy" engine right at the peak of its
efficiency. It is for the aircraft designer to decide what combination of airframe and engine
gives the best overall aircraft efficiency.
J . E. P. DUNNING (Rocket Propulsion Establishment, Wcstcott): Lee has commented on
the sensitivity of engines to intake conditions and asks why this should be so. The more that is
demanded of an engine in the way of performance the more sensitive will it become. As
Pryor showed and as was confirmed by Fleming, an engine which has to take an aircraft off
and then propel it at Mach No. 2 0 operates over a wide range of corrected speed conditions
( N j \ / T ) and it must have the best intake. This will surely enforce the general acceptance of
podded engines—whether for subsonic or supersonic aircraft. The Boeing 707 is one example
anil the French, in their logical way, have produced the Caravelle. In case the aircraft people
start lo object on the score of drag, I would quote the Bristol Bloodhound which is supersonic.
'I'he engines, in this case ramjets, are podded.
Only by podding can the engine designer exercise his proper control over intake design
free from the machinations of the aircraft designer.
G. H. LEE: In answer to Dunning's suggestion that designers ought to install all engines in
pods, it is not possible to generalize. The pods or buried engine argument is a big one that
cannot be settled in general; one can only consider each case on its merits and make the
appropriate choice then. The points made above about the efficiency of the complete
aircraft are relevant.
Even podded engines would have distorted intake (lows when the pods are at incidence or
arc yawed.
E. W. WASIELEWSKI (Curtiss-Wrighl Corp., U.S.A.): Although considerable effort is being
made to provide engines less sensitive to distortion, equal effort needs to be expended to
provide inlets which allow a reasonably designed engine to operate without danger. The
final answer is, of course, a proper compromise between the requirements of the inlet and the
engine.
B. H. GOETHERT (ARO Inc., Tullahoma): In the paper of Pryor, the need for flight
tests and their proper scheduling was discussed in a comprehensive manner. The numerous
superior characteristics of ground facility testing are stressed in some of the preceding papers.
I would like to draw attention to the fact that a flight test programme, closely co-ordinated
with ground facility test programme, is needed to bring the individual possibilities of the two
lest methods to their fullest utilization. The flight testing is naturally required as a final
check of the complete airframe engine system. However, the extent of flight testing wilh ils
much larger costs, more serious hazards to airplane and crew, and its limited data recording
capacity, can be considerably reduced, and the development time from the prototype to the
finally acceptable airplane may be appreciably shortened. Also, difficulties discovered during
flight testing can be more readily corrected when the cause of the difficulties and various
means of correction can be investigated in ground facility testing. I think, that Pryor will
agree with me, that during flight testing of some engines of his Company, many good examples
of supplementary flight and facility tests could be cited which proved extremely Ix-neficial
for the solution of critical engine problems.
Regarding the question: "Which problem should be scheduled for ground lesl facilities
and which for flight tests?" I advocate that problems which can be investigated in ground
facilities, should be scheduled for ground testing at as early a date as possible and as thoroughly
as is feasible. Naturally, the huge test facilities existing today have their limitations,
consequently not all propulsion system problems can be examined in ground test facilities
with the completeness desired. However in such cases, test methods, less comprehensive and
less elaborate than the testing in a large propulsion wind tunnel, can give reliable results in
support of a well co-ordinated flight test programme.
290
NAME INDF4X
ADAMS, R. H., 125, 127 DONALDSON, C. D., 19, 28
A L F O R D , J . S., 11, 127 D O N A L D S O N , I. S., 31
A N D E R S O N , D . E., 50 Dovi 1 , J, P.. Jr.. I 17. 128
ARMSTRONG, J. C , 10, 29 DI'CAIME, J., vii
DUNNING, J. E . P., 290
DURBIN, E.J., 287
BAJEK.J.J., 15,28 D U R H A M , F. P., 268
BARRERE, M., 201
B E R A N E K , L. L., 93, 128
EASTABROOKS, B. B., 96, 127, 129
BLACK, F . G . , J r . , 90, 129 ECKEL, E. F., 2 0 1 , 2 0 2
B L A K E Y , J . W., 19,27
ELLERBROCK, H. H., Jr., 85, 129
VAN DER B U C K , J . A., 117, 128
ENGLERT, G. W., 26, 27
BOGART, D., 7, 29
EVVARD, J. C, 19, 2 2 , 27, 2 8 , 252, 257
BOLLINGER, L. M., 181, 200
B R A G G , S. L., 179, 180, 183, 200
BRAITIIWAITE, W. M., 7, 29 FABRI.J., 133, 134
BROCARD, J., 155, 156 FASSO, G . , 96, 129
B R O W N , S T . G . A., 43 F E R R I , A., 22, 27
BRL-CE, W . L., 133, 134 FLEMING, W . A., 4, 6, 7, 26, 2 8 , 29, 2 8 7 - 2 8 9
B U R G O Y N E . J . H., 201 F R A D E N B U R G H , E. A., 19, 28
B U R N S , L. F . , 55 F R E D E R I C K , C . L., 117, 128
BUSSEY, L . E., 84, 128 F R E N G L , O . , 153-156
B Y K O W S K I , R. M., 133, 134
G A B R I E L . 1). S,, viii, 6, 12, 2 8 , 29, 2 3 9 - 2 5 9 ,

254
CALDER, P. H., 257-259
GEISTEN, K., 170-174
C A P E T T I , A., 268
G I L L . G . , 201,202
CARRIERE, P., 29-32
l,..i 1111 111. I!. II., vii, 33 78. 104, 128,
C H A F F I O T T E P. P H . , 134-136, 1 7 4 - 1 7 6 , 2 3 5 -
129-137,290
238
GOETT, H.J., 81, 128
C H A R W A T , A. F . , 171, 173
GRAVES, C. C , 7, 2 7 , 183, 190, 200
GUILDS, J. H., 7, 27, 183, 190, 200, 244,
GREENIIOUGH, V. W., 132, 136, 179, 183
252, 257
192
C H R I S T O P H E . J . , 96, 129
G R E G O R Y , N., 172, 173
G L A R K E , A. E., 189, 200
G R O S E , R. M., 85, 128
COCKSHUTT, E. P., V I I I , 2 0 5 - 2 3 8
C O H E N , L., 201
C O N N O R S , ] . F . , 19, 2 7 , 2 8 H A H E I . , L. W . , 84, 128
C O N R A D , E. W . , 6, 11, 27, 28 H A L L S , G . N . , viii, 177-204, 178
C R A I G , R. T., 13, 28 HANSON, M. P., II, 27
G R O C C O , L . G., 201 H A R R I S O N , A. J . , 189,200
HARTMANN, M. J., 176
H A R T S H O R N , A. S., 83, 128
D A I L E Y , C . L., 22, 27 H E A R T H , D. P., 13, 28
D A M K 6 I I I . F R , (J. Z., 181, 200, 201 II...MM 1,. R. W . . viii, 79 137
DANDOIS, M., 13,28 H E P P L E R , H., 13, 28
D ' A N D R E A , J . B., 117, 128 HERBERT, M. V., 180, 183, 184
DAVID, L.J., 72 H E R M A N N , R., 19, 28
DAVIDS, J., 2 2 , 2 8 HOLLIDAY, J. B., 179, 180, 200
DAVIS, H., 171, 173 H O O D , P. M . , 50
D E F R A N C E , S. J . , 8 1 , 8 2 , 128 H O O V E R , R. M., 93, 128
D E L A N O , J . B., 117, 128 H U N C Z K , H, R., 19, 2 8 , 29
DINGELDEIN, R. C, 81. 129 H U N T L E Y , S. C , 7, 29
291
363
ZSZ 'Z5Z '8?; 'ZZ " H M ' z x N a a a a x S 00Z '681 '"& A 'AvaaniAi
f i l Mil '". 'laaiaas f i l ' Z i l ' ' O 'NNVKSailJAJ
00Z f 8 l "081 ' 6 i l " a (J ' o N i a i v d s 102 '"V 'xainoyyi
9 i l ' O i l ' 5 9 1 - 6 6 1 '!!! A "A S ' H J m S 8Z '9 ' "X A 'aaooiAj
i £ " x f " 3 >I'"J->ws Eil ' l i l ' O T V v'AaooiAi
8Z '61 'TV ' t a x t t u a s 6ZI 7 8 ' 3 " 3 'aoaNOiAi
sez-eoz '!nA '"& D JIVHS 8Z '61 ' " 3 A\ 'aaMoaoiAi
S Z ' S I " V "O ' s u v a g 6Z ' 9 " H A\ 'Nas-ia5ioij\
SZ'ill ' , 0 , f ' " < » S i Z '61 ' 0 Ml 'aaAatAj
gg ' - / \ -Q '1N3MH3S; 00Z '681 ' V ' a a x s a j v
i Z ' 9 ' M A 'az-inHDS 8ZI '58 'O G 'NVHVWOTAI
8Z 'fZ "A O ' a a i - i a r m o s 55 '"H "M 'aooaa;)DiAi
f £ I '££1 ' V M ' x a a a n H o g 8Z '£1 ' M 3 ' M v a o o j ^
8Z '61 ' H V ' a a a a o a i i o s i Z ' H ''3 f'Av-mvoN
til Mil "K'ZIOHOS i £ I - 6 i 'F!! A '"M H ' X X V I A I
f i l Mil " A V 3 'ziWHas 8ZI ' £ 8 ' 3 TV 'aaonssviAt
8Z '£1 ' a Ml 'J-aiwiios 8Z '9 ' H S ' w a i s v ^
FOZ-EOZ ' 3 ' x a i w H o s 8ZI ' f 8 ' f TV 'svjaviAi
f i l - O i l ' H 'OKUHOnHOS £0Z 'ZOZ ' ' 3 ' 3 ' a i n a v i A i
6ZI '58 'AA O 'AHHOS 6Z 'Z£-£ '!! A ' T ' 3 'OTiaiNVONVjAj
I0Z "A ' x v a a v f ; 69Z-09Z ' - 3 'aooiovjAT
8Z ' i " a "N 'saatiNVs
£1 ' 3 f ' s a a a N V S
i 8 '6Z ' i l ' f TV ' t a w s 8 Z / 5 I ' M *S ' m a n o r ]
8ZI ' i l l ' T ' o i M a i s v s n T
8Z ' i Z '9Z '61 '"AA MI 'sNacimT
55 ' A \ G ' l a s s n y iSZ 'f5Z '8Z 'ZI "["H '>IDIIJI.T
8 £ l '6ZI ' 9 6 ' ! ! A " y i ' A o y 00Z '081 "A T ' T i a M O N o n
|IA ' - j ^ - y ' s i o o a n x o y 8EZ-50Z '!! A ' O O ' A A 3 T
6ZI ' 1 8 ' 1 A ' o - n v o o > i 8ZI Ms ' o ' 3 3 « v F \ n
6ZI ' 1 8 ' O f ' s x a a n o H 8£Z ' f £ Z - i Z Z ' " a ' - " ' 3 3
f i l Mil ' O H 'Naaoiiy f £ I '££1 ' 3 ' s a A i a o a q
i Z ' i 'AA ' X ' s a a o N A a y i 6 I 'Z6I '681 '£81
'6il 8il ' 9 0 Z " i i I '!!! A ' H V ' a a A a a a a - r
06Z-68Z ' H O ' 3 ^ 3
-
OOZ'fSI ' 6 i l ' " V V KvNx:ia Z £ - I £ '"X Ml ' M o a g a i
0 6 Z " £ i Z 'xi ' - 3 Ml ' a o . \ a , i 05 '"A 3 'V1VAJ.VT
O O Z ' i S I ' 8 i l " d "H x a a a o a j 8ZI ' ' S ' a x v a v T
6ZI ' 0 6 'TV O ' N O i s a a j
6ZI ' 5 8 ' a 'naMNij
6ZI ' i Z I ' 9 6 " V i ' v - i o z d N y 8Z'6I ' f 3 'aaifwaa^
9 5 I - 8 E I '6ZI ' 9 6 'l!! A " V i ' a a a a y Eil Mil 'H 'SV-LLO>I
i 8 Z '"G O s s i i a a a j 8ZI ' f 8 "A d ' i a s N i o A a o »
8 Z ' £ I ' ' 3 'lONOiwaa.i 00Z 'Z6I ' ' H "O ' O N ' N
fOZ 'ZOZ MOZ ' ' S 'S 'aaNs-i.i HZ 'EI ">I T ' K H i L ' i - i N
8Z1 ' £ 8 ' T ' J ' n v s a v a j 8ZI M8 ' M Vi ' A - n a ^
8Z 'ZZ ' 3 H ' a o a v a j xi 'mvwavN{ KOA
9 i I - 5 9 I '651 '!!! A " 3 ' D ' w u . v i 00Z '181 ' 3 ' z x i A o i a v > i
f £ I '££1 ' f N o a a v d 8Z '61 '"V 'zxiMoaxNV>i
8ZI ' 1 8 ' H Hssvd 8ZI ' i l l " M H 'aaNsiv-M
8Z '£1 " , \ \ 3 ' o x x o 8EZ 'SEZ 'fEZ ' H f 'NoxsNiiof

6ZI ' 1 8 ' " S TV a a M v a i Q 00Z 'Z6I " V i A 'NOSNiiof
O0Z'68I " f s a a a a o 00Z ' f 8 I ' 6 i l ' V "H ' N a s u a f
8ZI 'E8 " D ' M 'soNiNNaf
6ZI ' 0 6 'TV - f ' a a o o v f
i Z 'ZZ " V i 1 ' L i ' i \
8 Z ' £ I " Q 'X>AO\-
8ZI '£8 "A 3 KOS-IOH;>I\ 8ZI '£6 ' ' A ' a a v o N i
xaaNi awvn
NAME INDEX
STEWART. 1). (.., 177, 183, 185. 187 191, WALLNER, L. E.. viii, 7. 12, 17, 28, 29
200,201,202 239-259, 254
STUART, J . T., 172, 173 WASIELEWSKI, E. W.. 129-132, 136. 290
SURUGUE,!., 132-134 W A Y , S., 177, 179, 180, 200, 201, 202
SWEBEIG, H. H., 81, 129 WEBSTER, L. F., 43
WEICK, F. E., 82. 129
T A L L , B. S.. 179, 180, 184, 200 Wnm, M. A.. 180,200
TATRO, R. E., 101, 129 WEI.IER, A. E., 177,200
TAYLOR, D., 69 WILCOX. W. W.. 176
TAYLOR, B. L., 13, 28 WILD, J . M., vii
TEMPELMEYER, K. E., 96, 127, 129 WILLIAMS, F. A., 201,202
FERRY, A., 83, 128 WILLIAMS, D. T., 181,200
THURSTON, D. W., 202 WILSTED, H . D., 4, 10, 28, 29
TOLHURST, W. A., Jr., 81, 128 \ \ INDMUELLER, A. K.., 96, 127, 129
IKIMI-I. R. L., 22,29 WOOD, D. H., 82. 129
TURNER, L. R., 7, 29 WYATT. D. D., 13, 19, 23, 26, 28, 29
Usow, K. H., 11 YOUNG, A. W., 6, 28

YOUNGER, G. G., 6, 29
VASU, G., 12, 13, 24, 29, 254, 257
WALKER, C. L., 7, 29 ZUKOSKI, E. E., 179, 183, 184, 200, 202,

WALKER, W. S., 172, 173 203, 204
29:.
563
£f ' i £ 'jnb;ui|j.u 681 'sso| 3jnss.)jd j o
Ogi ' u o s u E d u i o o .ipjiud JJ 281 M-inssjjdjo
j_f '[0.HUO3 B3JB 281 ' J 3 q u m u X i u o p . v S u m j n c i j o P . I J J J
M'ZZOU p . 11)11(1 'X3U313IJJ3
\ f 'OUlSu.) 18( ' X j o a q i X i p o p A S u i i u n q
uiiniM uoiiBSEdoxd 'liuiisJl 33iiBC|.m|sifi JJ 6 ' i n o M O | q
'uonsnquio;)
JJ 9£ ta" '-
9J, 'sUOSUBdlllO.) 8 9 ' X i q p E J 1S31 j o A i p E d E Q
iqSiii i-iiiinu puiA\ 'sujoiicd 'uoiijoisj0
<zzn
II 'Mop IJIUI p.iuoisifj zsz 'zz "vm a
O i '3DUEUi.ioj.ijd j . i s i u j i f r Z8I M a q u i n u
Si ' £ i 'uj-'pun Z81 " 1 8 1 'A3U3I3IJJ3 u o i i s n q i u o j
2/_ ' r e u o i i t i . i M i u . i 'Xipop.v auiujii}[
f l 'JE[nuuB l \ 'lafuiEj
•JO|.),l|',)-,l.)SllUI([ Z5Z ' J o i s n q m o j
ill '911 '«ptnmj j j 6 'uoiisnquKU
uotspidojd snjos-imj 'Jjuissjjojd BIE(T ')noMo|g
Of I ' i l Z I I ' p U U n i pUIM 'S33UE1EH
96[ 'jjujnqj.n|K 'uoiu.iAuoQ | 9 'uoiiE|nmis XJOI.I.IIEJI

l
f q ' u o i i u a . x l o 1U.1ISUBJ1 09 33UBmJojj3d
g g 'sisAiEUE U U I S A S - I E I I U E U I 'iui!|d p.)||oJiui)3 XnB.JiiBmomv
g g ' j S U o d s j J 3J1EUIOII1B 85
' A l ' l p B J IS31 'S|SX|BUB ' X i q i J E J , i p n i | l | E | O J ] U 0 3 311EUIOJI1Y
e.l - i . i f o q j n ) j o s J i u i B u X p ^ 8 1 "J3pJO UOI13E3J l U . i J E c l d y
•[OJIUOQ 8 6 ' i j U l i p l E U I JOJ S3JlllEJ.Klui.il
g2 's.iipnis uoiiBjuaijuo;^ j_ 1 | 'S.JJIA.ip Xl.lJBS
55Z '|-'poui 3 i i u B u X p j o a s n 20| '1U31U03 U J S X X O
| g £ ' . O o i s i q .iiuii J O I '3ZJS 3 U | 3 U 3 lllllllllM-lll
58Z ' O f ' Z I 'uii'JBui £ 6 'S3iii|iqBdE3
f 5 Z '£5Z 'DIBJ oauBcjjnisip lujiqiui! j o •|.)uuiii p u i M a p r q p r v
] ] ' o p n i i i | B JO ID.IJJ.) JJ 9 6 " 8 9 ' J J U H i x u o j a j d s . ) i | i | p e j i s 3 i
6 f Z 'uo!iiuii.>p J J f ' S 3 n b i u q 3 . ) i i| U I M V H
g 8 2 ' u o i i . n ; .i.\i|.).ujo.) '||Bis ao
JJ £ £ ' " • - " J sui3|qojd suiSuj
JJ g g | 'sisai | j p o m ' j o s s - u d m o ; } 'jjqiuBip 3pniii|v
9il 'p!nllau!1JOMJo 101
ZfZ ' 9 £ ' u o p j o i s i p J J I U I j o
•|.Hllllll 3 p i l | l l | E UI UOIIEUJUIE1U03 U I B 3 J | S J I \
L f l 'SJSSO| 1EJII, j o 1 M J J J
JJ f 6 I ' 3 . m i E . I 3 d u i 3 l (JBA\
Uj\ ' s p p O U l J O UOI1311JISU03 'JOSS.ijdlUOQ j ) g 'q3JE3S3j
691 'i(uipEo| JSBIS 961 '561' " ! uouEipEJ
8 9 1 'l3->JJ-> ^ l E a s 65Z 'Xi'iiqEism
891 'siijjj.) j . i q i m i u sp|ouX,)}i 58Z 'lU3Uldo|3A3p iqSiij
•|EIXB -joss3jduio;} f f Z ' s u i q j n i j o I33jj3
95Z 'S5Z 'J1KJ J J U B q j n i s i p j o 9 6 1 'Ul UOI133AU03
£ £ 2 -JOSS.JjdlUOO JO 133JJ3 Jj,| ' J 0 J I U 0 3
2g2 'mo.wo|q 'Joisnqmo;-) 9 SI.I.IJJ.) .ipnjii|E
f 8 I 'AHliqeis 'j3Ujnqj3ij\-
JJ U \ -SuiiEOS 02 'q3JB3ssj Xpoqjsijy
8 8 1 ' i 8 I "8UI1B3SJO 133JJ3 'S11UIII l\\ 'sm.iisXs u o i i E ] U 3 m n j i s u i
JJ 281 'JJiilBjadmoi j o 9 1 1 - f l l *33UBlBq
£81 'Z81 'J-iqumu s p p u A o y j o T 6 ' p u u m puiA\ u o i s | n d o j d
JJ J 8 1 'J->PJO U 0 1 1 3 B 3 J J 0 133JJ3 'OQ3V

( p m u i ) u o 3 ) A3u,)i3ijjo u o i i s n q u j o ; ) 0 1 'S|S3l UOIlEJ3]333V
xaaNi xoHfaas
SUBJECT INDEX
Ducted nozzle, location of shock, 46 Inlet flow distortion (continued)
transient testing, 47 effect on turbine outlet temperature, 242
on turbojet acceleration, 250
Ejector, bleed flow effect, 71 on turbojet performance, 287
rig combustion chamber tests, 134 Inlet flow distortion radial and circum-
small engine tunnel 1'ONERA, 133 ferential, 11, 258
steady-state testing, 69 screen assemblies, 37
transient testing, 72 surge margins, 37
Engine, altitude testing 33 ff, 68 ff typical profiles, 45, 46, 241, 288
component problems. 1 wind tunnel, ducted-nozzle comparisons,
dynamics, control problem, 13 46
flight envelope, 275 ff Inlet Sow,
performance, altitude chamber tests, 6 duct dynamics, 23
size limit, 103 simulation, 23, 25, 35 ff
stall sensitivity, 41 simulation methods, relative merits, 52
test facility, transient testing, 53 Inlet pressure, transients, 12
Kxhaust, flow field simulation, 25 Inlel temperature effects, transients, 12
gas diffusers, 69, 130 Instrumentation, model tests, 162
gas removed, 99 full-scale propulsion tunnels, 116
nozzle problems, 20 turbine blades, 206 ff, 234 ff
Exhauster plant, Interference, transonic tunnels, 104 II
pressure recovery devices. 49. 130
requirements, 68 Jet-diffuscr, free jet facility, 16
response to sinusoidal disturbances, 56 Jet spreading, AEDC tests, 100
Fast transient corrections,
application, 67 Laplace transform, fast transient testing. ii_'ll
basic equations, 62 LFA tunnel, Braunschweig, 88
determination of system constants, 64 Mercury sliprings, 215 ff
power series thrust curve. Ii."> Munich 3m. tunnel, 88
step input excitation, 65
useful frequency range, 64 NACA tunnels, Cleveland, 87, 93 ff
Free jet, assemblies Langlev Field. 81, 84
change of angle of attack, 48 Moffett Field, 81
change of Mach number, 48 Nacelle installation turbojet, 21
Free jet, facilities iisiiug, 14 ff, 47 ff. 132 Noise, tunnel problems, 93
critical inlet shock patterns, 49 Non-dimensional parameters, turbojet, _'li<l ff
pressure recovery devices. 49 Normal shock position control, 23
second throat diffuser, 49, 52 Nozzle problems, 20
supersonic inlet arrangements, 31, 48
Fuclsystem, propulsion wind tunnel, 120 ONERA supersonic tunnel, Modane. 151 If
Full-scale propulsion wind tunnels, 79 ff Test Centre, Modane, 90 ff, 138 ff
Future trends, propulsion wind tunnel transonic tunnel, Modane,
testing, 125 balance, 114, 141
control cabin, 142
High altitude testing, <iB If fuel system, 139
Hispano-Suiza tunnel, 83 trsis completed, 143
Historical review, propulsion wind tunnel ONERA tunnel Ghalais-Meudon, 81
testing, 80 ff
Hot water ejector SNECMA, 153-156 Propulsion wind tunnel.
AEDC, 94
Inlet buzz, 22 altitude start-up with highlv reactive fuel,
effect of engine on, 253 124
Inlet flow distortion, automatic data reduction, 117
definition of problem, 35 contamination detection, 119
effect on component efficiencies, 242 fuel leak detection, 117
on compressor pressure ratio, 242 fuel system, 120
on distortion caused by engine, 23 future trends, 125
on ramjet engine, 43, 245 handling equipment, 123
on Reynolds number, 37 ff interference effects, 104
on surge margins, 12, 243, 288 optical viewing system, 117
296
L6Z
29 ' u o n o u n j s s m d m i pufr 56Z ' p p o m 3 i m B u X p j o 3sn
^ 8 'pJOJiaEjj ' ( . m u n i ijBJ3Jiy pJimri | g g 'Xjoisiq .mill
f SZ ' £ 5 Z '31EJ 3 3 U E q j n i S i p ! U 3 i q m B J O 133JJ3
/ [ -Suiis.ii do.idoi(.m | 6^2 ' u o i i m i p p
JJ \ f i ' u o p a o i s i p i,)|ui g 8 2 'UOI13E 3AI1.X1.1JU.)
JJ IfZ ' S P 3 J P IBnjnui u o i m q u i s i p MOJJ ' j O S S 3 j d u i l ) 3 -\\V.\tm]
'siu3uodtuo3 'isfoqjiiT^ £ 2 ' X i p B d E 3 -34lB[|lds;
i ) 50Z 's3jniEJ3dui3i 3pE|q suiqjni f!S\ "S|S31 p p o i l l UO 13JJJ3 'OI1E4 1B3I] 3 l j p 3 d t , '
£5Z ' 6 f Z '33UBuuojaad IU3ISUB.II £ 1 2 '.JAUp 3iqiX3U
f 'UOllE[|ElSUI 1S31 jj gi2 'Xjii3j3iu s8audi|s
£ 'S3IlSU313BJEq3 t l ' u i c l u m d £ 2 '(0J1UO3 u o i i i s o d > p o q t ;
OZ ' • s - ' l 3]ZZOU /_£ ' U J 3 H E d AU)(J J3JU1 ,>]E|nuIlS Ol ' S U 3 3 J 3 g
JJ ()92 ' S J 3 l 3 l U E J E d |BUI)lSU3lUip-UOU 65Z 'MM3J3S
£ 0 1 ' S 3 p n i l ) | E UiniUIXEUl Z l l i O l '£01
£ 8 2 '33UEU1.10JJ3CI •SlS.ll |,)UUI1| l.)l'o[(jni '33UEUUOJJ3d d o o s ^
0 g 2 '3UIII U O I l E J 3 p 3 3 B 'UOll.llllSip .«<)[) 13]UI l_\\ -sui.iisXs a a j s i r r p g
Q f '3dl>(3AU3 S u i l B J s d o 3UlSll.) fll TOI ' d o o a s 8ui3u3AB3$
f5 ' A 1F £91 ' a t n q j r u
- m i B U l p 3 J ] 0 J ! U 0 3 X l | | p E J '1S31 U O I ) B J 3 p 3 3 p 861 ' i 6 I 's3jniBj3diu.)i s q n j j u r e y
£ [ 'S3IUlEUAp [OJIUO.) Jl i i l 'uoiisnquio.j
']3foqjn ] 89[ 'joss3jdmi)3 ]EIXB
£ 9 1 -Ul SSO| ,)SBpUIA\ J) 161 '43ium|.i.)i|B
JJ 9QZ ' S 1S31 SpEJCJ JOJ S 3 ] d n 0 3 0 l U J 3 i p 'iiui|EJs;
£9| 'UO 133JJ3 3JE3S
JJ 651 ' n s a , P P o r a £ 1 1 'U011E[|BISU1 a S t l B S UIBJ1S
JJ OZZ 'SOZ ' j ° uiE3Jisdn uoii33fui pnj 221~OZI 'msisXs X[ddns i u E | p d o j d
9 £ ] 'S33U3J.)JJ3JU1
Jl Z£Z 'ZZZ
£ 9 'U0I133JJO3 ' S J B J S S I U I 3 S | I l d m |
'JJ gOZ ' s J U 3 U I 3 j n s E 3 U I 3 J t l l E J 3 d u i 3 l 3pE|q
g21 '2u;is3i p u u n i pui.w ui p u j J i 3jninj
'S3uiq.ni]
19 ' l U 3 l U U 3 d x 3 111.11-, 1111.11 JSEJ
501 6Z "•13U.U3JJ31UI l a o d d n s
\l_ 'jOl3.)f.)-J3Snj)Ip
JJ ofi 'Su;is3i a o i t m d o i d
'JS31 ! 3 > p o y
' s p u u n i amosuEj j
JJ IEZ '5ZZ ' s a p E [ q . m i q j n i 's.ijniEj.idiusi £91 '3Uiqjni
19 'UOI133XI03 ' l u s u i d m b o . \ 3 u 3 n b 3 j j A\O| i £ 'l- 1 !"!
2 i ' U O l l E J s d o JO]33l'3 )SUBl|X3 7.l_\ - l l i p i A [E311IJ3 UO UOIlB]OJJO 33U3n[)UI
£g ' X i q p s j iS3) 3u;8u3 g 'xapui
I f ' S u i i s s i 3|zzou p.iionp 9 '33UEUI.IOJ.I3d 3 U l 3 U 3
21 ' s S u u i j I U S U I E U U E JJ 6 i l 'no[isnqmo3
'S133JJ3 JU3ISUEJ [ JJ O i I 'S3PE3SE3
19 'is3i XJOISSTEJX 89] 'Joss,jjdmo3 |EIXE
f Z '£1 ' 3 u ; 2 u 3 i3foqjni 'IUBISUOS 3iuij_ ' J 3 q m n u sp|ouX.)}|
JJ iZZ ' i s o q s o i u a s q j , f ' j s q u i E i p 1S3] . i p m i i i B ' S 3 n b m i | 3 3 i q 3 J B 3 S 3 Jl
JJ 90Z 'sspEjq 3Uiqjni ' S 3 | d n 0 3 0 u u 3 i ) x (j3Ujnqj3ijv 33s) i E 3 q s y
i Z Z ' S 3 p E ] q ,illll{.lll | 'SS3JIS | B l U J 3 q x i O I ' 3 n b l U i p 3 1 3UEld-U01l33U3>l
gg 'IUSISXS [0JIU03 [Eiiucm
£ f 'IS31 3[ZZOU p 3 1 3 n p '3[ISSIU1 31UOSJ3dllS
0 9 'UI31SAS p 3 | [ O J I U 0 3 X | i B 3 i i E u i o i n E
i l _ 5 1 'SIS3) .)3UEUUOJJ3d
'XlI|I.)EJ JS.. I
£ f ' g i 'U011E|]E1SUI
JJ | 9 I ' s s u i q j n i j o Suiissj ppom
g ^ 2 ' u o i u o i s i p J3[UI
JJ 991 'sjoss3jduio3 j o Supra) ppom
JJ g i 'JS31 XlI|13BJ ]jf.).).!(
' l i i . n u d i i i b . i is.) |
'13|UIE,VJ
£8 ' q S n o j o q u j B j ' p u u m g y y
2 g 2 ' J O J S n C | m 0 3 UO 133JJ3
961 ' g 6 I 'J.iUjnqj.njE ui u o i i E i p E y
6-J.2 ' u o i i i u y s p '33jnt;
go I "3ju3J3jj3iuj p o d d n g i £ Z '9EZ '3ui3.)poioi]d ' j 3 i 3 m o a X j
[ £ ' S l U S m s S u B J J E 13]UI 16 ' 8 8 'S»S3» i s r ^ i n j
J J 6 1 ' s p u u n j pTJIM 3 i u o s j . ) d n v ; ^ 2 ( 'S3nbiuq33j IS31
gg ' u o i i E i p x 3 i n d u i d3ig l_ J1 'm3isXs U3J3'iq3S
01 'uoqiuSi 11 I 'S33lA3p Xl3JES
6 -S133JJJ 3 p n i i i ] B q S i q ' S U I ) . I E I < , - 6 6 ~ i 6 'XiqiqBdBS 33UBUiJoja.)d
\ f ' 3 U | S U 3 JO XlIAIlISUSS 1[B)I^ (pmuipio}) O Q 3 V ' p u u n i pui.w uois|ndo.i,j
xaaxT xoafans
863
ZI I '501 'sixoddns ppom p u u n i p u i \ \ f6 'smoqEiinx 0 Q 3 V

£8 'qSnoJoqiuEj av>I :sis3j uois|ndojd
JJ 8£1 '06 'auEpojAf V H 3 N O Of I
18 'PP!3 WJJOJAJ VOVN 'JJ 2II 'iu.)uidmb.) 33UE[Eq 'spuuni pui/yy
f8 '18 'PF!3 XaiSuEq VOVN 81 'dojdoqjni SEjp Smiiiuipui/w
JJ £6 ' i 8 'puBp.\3io VOVN £9] 'S3uiqjnj 'sso) aSEpui^
88 'qîuniAT
88 '2pA\q3sm,Ej{i v.43 f8 ' P R 3 HtSiJAA
(pmui)uOw) sisal uoisindoad 'spiuuni puiAA, JJ 16 'BUJoqEflnx ' p u u n i HVST1
xaa.Mi xoafaas

AGARD-AG-37 - Advanced Aero Engine Testing

Uploaded by

Copyright:

Available Formats

AGARD-AG-37 - Advanced Aero Engine Testing

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AGARD-AG-37 - Advanced Aero Engine Testing

Uploaded by

Copyright:

Available Formats

n

S.? : - CT^Tu/RBlrvJ^ f - EbSCilME^

ADVANCED AERO ENGINE TESTING

Published for and on behalf of

l.ihinry of Congress Card No. 59-11209

Printed in N o r t h e r n Ireland at T h e Univcrsilies Press, Belfast

SOME SELECTED PROBLEMS IN ENGINE ALTITUDE TESTING - 33

F U L L - S C A L E PROPULSION TESTING IN W I N D TUNNELS - 79

MESURES CONCERNANT LES ESSAIS DE MOTEURS EFFECTUES DANS LES

SOME EXPERIENCES IN COMBUSTION SCALING - 177

LABORATORY AND F L I G H T TECHNIQUE FOR THE MEASUREMENT OF THE

APPLICABILITY OF SEPARATE COMPONENT T E S T RESULTS TO COMPLETE

PART III.—FLIGHT TEST PROBLEMS

F L I G H T T E S T DEVELOPMENT OF SUPERSONIC ENGINES 273

Several of the most up-to-date engine test facilities are described in a

systems involves considerations ranging from small details, such as the

ENGINE TEST FACILITIES

This paper outlines a n u m b e r of categories of propulsion system problems

A L T I T U D E TEST CHAMBER FACILITIES

r FUEL INJECTION EJCCTOR

OIFFUSER COMBUSTION CHARACTERISTICS

altitudes while avoiding combustion screech or high-frequency oscillations.

typical characteristics: decreasing efficiency with decreasing pressure, and

Engine Performance Investigation

REYNOLDS NUMBER INDEX

relationship involved in combustion kinetics equations. Sufficient data to

ENGINE TOTAL TEMP

CORRECTEO AIR FLOW,

. l/lilude Operating Characteristics

200 4 0 0 600 800 1000 1200

be accomplished, (2) the flame be successfully propagated to the other

Some of the data obtained in turbojet altitude ignition investigations

40 80 120 160 200

rate per second for a particular set of operating conditions. As repetition

O f course, variable-geometry features are incorporated in modern turbojet

Inlet Pressure and Temperature Transients

40- NO COMPRESSOR STALL

0 1000 2000 3000 4000 5000 6000 7000

operation is plotted as a function of temperature rise at the compressor inlet

Engine Dynamics and Controls

is presented in Fig. 15. Transient thrust, engine speed, and turbine-inlet

so-called connected-pipe technique wherein stagnation pressure and tempera-

IAI WITH JET DIFFUSER AT NOZZLE OUTLET

IB) WITH SECOND THROAT AROUND ENGINE COWL

altitude test facility shown in Fig. 1 modified by addition of a supersonic

a n d solved in ramjet engine research. Included in most investigations of

SUBSONIC- WIND TUNNELS

100 125" I!675 175 S53 225

SUPERSONIC WIND TUNNELS

problems which can be investigated, a variety of interaction problems are

A typical installation of a turbojet engine in a nacelle in a large tunnel

PLUG CONTROL EXIT PLUG

unlimited spillage became possible for only moderate amounts of spike

EXTEND SPIKE RETRACT SPIKE

DESIGN SPIKE POSITION

As mentioned in a n earlier portion of the paper, much work is done in

DEAD -LAG TIME

properties. I n addition to duct dynamics many other aspects of the overall

I. COMBUSTOR I. SUBCRITICAL INLET OPERATION

tunnel tests as compared with those in altitude and free-jet facilities. T h e

'2c -- •4t