-- N65-23345

NASA SP-36

AERODYNA IC DESIGN OF
AXIAL-FLOW COMPRESSORS
REVISED

Prepared- bN members of the staff of Lewis Research Center, Cleveland, Ohio.

Edited bg h m . A.~ Jo- and ROBERT This publication supersedes
0. BULLOCK.
decassified NACA Memorandum E56B03, E56BOa, and E56BO3b, 1956

S&i@ d Tecbnuul Information Division 1965

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION
- warbington, D.C.
“r
i L :
-e
i
 CONTENTS
CHAPTER
I. OBJECTIVES AND SCOPE________________________________________---
IRVING A. JOHNSEN AND ROBERT 0. BULLOCK
'^"T J
II. COMPRESSOR DESIGN REQUIREMENTS_________ - - - _ _ ___ _ __ __ -- - ---- 9
ROBERT 0. BULLOCK AND ERNST I. PRASSE
III. COMPBESSOR DESIGN SYSTEM____________________________________53 -*/
ROBERT 0. BULLOCK AND IRVING A. JOHNSEN
IV. POTENTIAL FLOW IN TWO-DIMENSIONAL CASCADES:- - - - - - - - - - __ __ J 101
WILLIAMH. ROUDEBUSH
V. VISCOUS FLOW IN TWO-DIMENSIONAL CASCADES_ _ _ _ _ _ - _ - _ _ _ - _15 Ir' _ 1_ - -
WILLIAMH. ROUDEBUSE A N D SEYMOUR LIEBLEIN
VI. EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES----------- J 183
SEYMOUR LIEBLEIN
VII. BLADE-ELEMENT FLOW IN ANNULAR CASCADES--- - - - - - - - - - - - - - - - - - I/ 227
WILLIAMH. ROBBINS,ROBEFJT J. JACKSON, AND SEYMOUR LIEBLEIN
VIII. DESIGN VELOCITY DISTRIBUTION IN MERIDIONAL PLANE_----_- -- -
CHARLES C. GIAMATI,JR.,AND HAROLD B. FINGER
J 255

IX. CHART PROCEDURES FOR DESIGN VELOCITY DISTRIBUTION---_ - - - -377

J
ARTHUR A. MEDEIROS AND BETTYJANE HOOD
X. PREDICTION OF OFF-DESIGN PERFORMANCE OF MULTISTAGE
If
C O M P R E S S O R S _ _ _ _ _ _ _ - _ _ _ - - - - - - - _ - - - _ - _ _ _ _ _ - _ - _ - - - - _ _ - - - - 297
--__
WILLIAM H. ROBBINSA N D JAMES F. DUGAN, JR.
XI. COMPRESSOR STALL AND BLADE VIBRATION_______________________
ROBERT W. GRAHAM AND ELEANOP COSTILOW GUENTERT
J 311

XII. COMPRESSOR SURGE________________________________________---_--- 331 4/

MERLEC. HUPPERT
XIII. COMPRESSOR OPERATION WITH ONE OR MORE BLADE ROWS
STALLED________________________________________--___--__--_-_ i/ 341
WILLIAMA. B~NSER
XIV. THREE-DIMENSIONAL COMPRESSOR FLOW THEORY AND REAL
FLOW EFFECTS________________________________________-----_- 64' 365
HOWAED2.HEEZIGAND ARTHUR G. HANSEN
XV. SECONDARY FLOWS AND THREE-DIMENSIONAL BOUNDARY-
LAYER EFFECTS________________________________________-----_-
ARTHURG. HANSENAND HOWARD 2. HERZIG
J 385

XVI. EFFECTS OF DESIGN AND MEASUREMENT ERRORS ON COMPRES-

SOR PERFORMANCE________________________________________- -
ROBEETJ. JACKSON AND PEGGY L. YOHNEII
XVII. COMPRESSOR AND TURBINE MATCHING _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . ____________ 469
JAMES F. DUBAN, JE.
REFERENCES ______ __ __ __ _______ __ ___ _________ __ ____ ___________ __ __ _____ _ - 496

Precediog page blank

 ,

CHAPTER I
OBJECTIVES A N D SCOPE
By IRVING and ROBERT0.BULLOCK
A. JOHNSEN -
This$rst chapter of a report on the aerodynamic a design system, and stimulated by the urgent need
design of axid$m compressors presents the general for improving gas-turbine engines, research on
objectives and scope of the Over-aU report. The basic axial-flow compressors has been accelerated both
problem of compressor M g n is outlined, and the /Gn this country and abroad. The results of this
approach generally taken to accomplish its solution I
research have been presented in numerous publi-
is pointed out. l7w &een succeeding cations. In the majority of instances, each of
the report are summ.arized. these reportg presents only a fragmentary bit of
information which taken by itself may appear to
INTRODUCTION
have inconsequential value. Taken altogether
Currently, the principal type of compressor and properly correlated, however, this information
being used in aircraft gas-turbine powerplants is represents significant advances in that science of
the axial-flow compressor. Although some of the fluid mechanics which is pertinent to axial-flow
early turbojet engines incorporated the centrifugal compressors. It was the opinion of the NACA
compressor, the recent trend, particularly for high- Subcommittee on Compressors and Turbines and
speed and long-range applications, has been to the others in the field that it would be appropriate to
axial-flow type. This dominance is a result of assimilate and correlate this information, and to
the ability of the axial-flow compressor to satisfy present the results in a single report. Such a
the basic requirements of the aircraft gas turbine. compilation should be of value to both neophytes
These basic requirements of compressors for and experienced designers of axial-flow compres-
aircraft gas-turbine application are well-known. sors. Realizing the necessity and importance of
In general, they include high efficiency, high air- a publication of this type, the NACA Lewis
flow capacity per unit frontal area, and high laboratory began reviewing and digesting existing
pressure ratio per stage. Because of the demand data. This report represents the current status
for rapid engine acceleration and for operation of this effort.
over a wide range of flight conditions, this high This chapter outlines the general objectives and
level of aerodynamic performance must be main- the scope of the design report and indicates the
tained over a wide range of speeds and flows. chapters in which each specific phase of compressor
Physically, the compressor should have a minimum design information is discussed. The general com-
length and weight. The mechanical design should pressor design problem and the approach usually
be simple, so as to reduce manufacturing time and taken to accomplish its solution are indicated.
cost. The resulting structure should be mechani- The various aspects of compressor design to be
cally rugged and reliable. treated in the over-aU compendium are outlined,
It is the function of the compressor design as well 85 the specific sequence in which they will
system to provide compressors that will meet be presented.
these requirements (in any given aircraft engine Because axial-flow compressors are most ex-
application). This design system should be accu- tensively used in the field of aircraft propulsion,
rate in order to minimize costly and time-consum- and because this field requires the highest degree
ing development. However, it should also be as of excellence in comprwsor design and perform-
straightforward and simple as possible, consistent ance, the attention in this over-all report has
with completeness and accuracy. been focused primarily on the problems pertinent
In an effort to provide the basic data for such to the axial-flow compressor of turbojet or turbo-
1
691-564 0 4 3 - 2
2 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

prop engines. The results, presented, however, the annular flow area is reduced to correspond to
should be applicable to any class of axial-flow the decreasing volume. This change in area may
compressors. be accomplished by means of varying tip or hub
diameter or both.
DESCRIPTION OF AXIAL-FLOW COMPRESSOR In this compression process certain losses are
The basic function of a compressor is to utilize incurred that result in an increase in the entropy
shaft work to increase the total or stagnation of the air. Thus, in passing through a compressor,
pressure of the air. A schematic drawing of an the velocity, the pressure, the temperature, the
axial-flow compressor as installed in a turbojet density, the entropy, and the radius of a given
engine is shown in figure 1. In the general config- particle of air are changed across each of the blade
uration, the first row of blades (inlet guide vanes) rows. The compressor design system must pro-
imparts a rotation to the air to establish a specified vide an adequate description of this flow process.
velocity distribution ahead of the first rotor. The HISTORICAL BACKGROUND
rotation of the air is then changed in the first rotor,
and energy is thereby added in accordance with The basic concepts of multistage axial-flow-
Euler’s turbine equation. This energy is mani- compressor operation have been known for ap-
fested as increases in total temperature and total proximately 100 years, being presented to the
pressure of air leaving the rotor. Usually ac- French Academie des Sciences in 1853 by Tour-
companying these increases are increases in static naire (ref. 1). One of the earliest experimental
pressure and in absolute velocity of the air. A axial-flow compressors (1884) was obtained by
part, or all, of the rotation is then removed in C. A. Parsons by running a multistage reaction-
the following stator, thus converting velocity type turbine in reverse (ref. 2). Efficiencies for
head to static pressure. This stator also sets this type of unit were very low, primarily because
up the distribution of airflow for the subsequent the blading was not designed for the condition of
rotor. The air passes successively through rotors a pressure rise in the direction of flow. Beginning
and stators in this manner to increase the total a t the turn of the century, a number of axial-flow
pressure of the air to the degree required in the compressors were built, in some cases with the
gas-turbine engine cycle. As the air is com- blade design based on propeller theory. However,
pressed, the density of the air is increased and the efficiency of these units was still low (50 to 60
r----lnlet guide vane
I
I r-- Rotor
I I
1 I ,--Stator
I 1 I

RQURE
1.-Axial-%ow compressor in turbojet engine.
 OBJECTIVEB AND SCOPE 3
percent). Further development of the axid-flow 7). Since that time, considerable research has
compressor was retarded by the lack of knowledge been directed at extending aerodynamic limits in
of the underlying principles of fluid mechanics. an attempt to maximize compressor and gas-
The advances in aviation during the period of turbine performance. One of the major develop-
World War I and the rapidly developing back- ments in this direction has been the successful
ground in fluid mechanics and aerodynamics gave extension of allowable relative inlet Ma
new impetus to research on compressors. The without accompanying sacrifices
performance of axial-flow compressors was con- (ref. 8 ) . The subject of allowable bbde loading,
siderably improved by the use of isolated-airfoil or blade surface diffusion, has also been attacked
theory. As long as moderate pressure ratios per with a degree of success (ref. 9). Accompanying
stage were desired, isolated-airfoiltheory was quite improvements such as these have been an increas-
capable of producing compressors with high e& ing understanding of the physics of flow through
ciency (ref. 3, e.g.). Compressors of this class axial-flow compressor blading, and corresponding
were used in such machinery as ventilating fans, improvements in techniques of aerodynamic
air-conditioning units, and steam-generator fans. design. Therefore, in view of the rapid advances
Beginning in the middle 1 9 3 0 ’ ~interest
~ in the in recent years, it appears appropriate to sum-
axial-flow compressor was greatly increased as marize the present state of the art of compressor
the result of the quest for air superiority. Effi- design.
cient superchargers were necessary for recipro-
COMPRESSOR DESIGN APPROACH
cating engines in order to increase engine power
output and obtain improved high-altitude aircraft The flow through the blading of an axial-flow
performance. With the development of efficient compressor is an extremely complicated three-
compressor and turbine components, turbojet en- dimensional phenomenon. The flow in the com-
gines for aircraft also began receiving attention. pressor has strong gradients in the three physical
In 1936 the Royal Aircraft Establishment in dimensions (axial, radial, and circumferential),
England began the development of axial-flow com- as well as time. Viscosity effects in compressors
pressors for jet propulsion. A series of high-per- are significant and must be accounted for. In
formance compressions was developed, culminating general, the design control problem becomes
in the F.2 engine in 1941 (ref. 4). In Germany, more critical as the level of compressor per-
research such as that reported in reference 5 ulti- formance is increased. In order to provide
mately resulted in the use of axial-flow compressors ease of application, the compressor design system
in the Jumo 004 and the B.M.W. 003 turbojet en- must reduce these complications and establish
gines. In the United States, aerodynamic research rational and usable procedures.
results were applied to obtain high-performance Because of the complexity of the problem, no
axial-flow units such as that reported in reference complete solution is currently available for the
6. In the development of all these units, in- three-dimensional, time-unsteady, viscous flow
creased stage pressure ratios were sought by through an axial-flow compressor. In the main,
utilizing high blade cambers and closer blade designers have resolved these diEculties by
spacings. Under these conditions the flow pat- making approximations that permit the use of
terns about the blades began to affect each other, two-dimensional techniques. These approxima-
and it became apparent that the isolated-airfoil tions are usually based on the assumptions of (1)
approach was inadequate. Aerodynamic theory blade-element flow and (2) axial symmetry.
was therefore developed specifically for the case The blade-element approach assumes that flow
of a lattice or cascade of airfoils. In addition to in the blade-to-blade or circumferential plane
theoretical studies, systematic experimental in- can be described by considering the flow about
vestigations of the performance of airfoils in cas- blade profiles formed by the intersection of a
cade were conducted to provide the required flow surface of revolution and the compressor
design information. blading (fig. 2).
By 1945, compressors of high efficiency could be Axid symmetry assumes that an average value
attained through the employment of certain can be utilized to represent the state of the air
principles in design and development (refs. 2 and in the blade-to-blade plane. Equations describing
4 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

fi rBlade element

,-Flow surface

FIGURE
2.-Flow in circumferential plane.

radial variations of these average values may

then be written for continuity, energy addition,
and equilibrium in the hub-to-tip or meridional
plane (fig. 3). In essence, then, a combination of FIGURE
3.-Flow in meridional plane.
two-dimensional solutions in the two principal
planes (circumferential and meridional) is used
to approximate the complete three-dimensional sional solution to the design problem, this quasi-
flow. three-dimensional approach has achieved general
In applying this approach to compressor design, acceptance in the field.
ccecond-order corrections are used to account for In practice, the aerodynamic design of a multi-
three-dimensional variations from this simpliiied stage axial-flow compressor may be considered to
flow picture. Experimentally obtained data are consist of three principal phases:
utilized to account for effects such as those (1) Determination of stage-velocity diagrams
arising from viscosity, time-unsteady flow, and for design-point operation
blade-row interactions. Empirical limits are (2) Selection of stage blading
established for such aerodynamic factors as (3) Determination of off-design performance
maximum permissible Mach number and blade The first part of the design involves the determi-
loading. nation of the various air velocities and flow angles
No rigorous theoretical justification of this from hub to tip at the inlet and outlet of each
simplified design approach can be made. It blade row, to best achieve the design-point require-
appears sufficient to state that comparatively ments of the compressor (i.e., pressure ratio and
excellent compressors can be and have been weight flow). The annular configuration (varia-
designed by simplified approaches such as these. tion of hub and tip contours through the compres-
In the absence of a more complete threedimen- sor) is determined. Next, the blading is selected
 OBJECTNES AND SCOPE 5
to satisfy the design-point velocity diagrams and SCOPE OF DESIGN REPORT
to obtain high efficiency. Basicahy, this selection Because of the complexity of the compressor de-
requires knowledge of loss and turning character- sign problem,'even the simplest design system
istics of compressor blade elements. With the necessarily includes many dift'erent phases. In
compressor geometry establfshed, the h a l step is order to summarize existing compressor infor-
the estimation of the performance characteristics mation as clearly and logicdy as possible, this
of the compressor over a range of speeds and flows. over-all compendium is'divided into chapters, each
In view of the importance of offdesign operation, concerning a separate aspect of compressor de-
this procedure may be iterated so as to properly sign. The degree of completeness of these chap-
compromise design-point operation and the range ters varies greatly. In some cases, rather com-
requirements of the engine. plete information is available and specific data are
A more complete discussion of the compressor given that can be fitted into detailed compressor
design system adopted for this over-all report is design procedures. In other cases, the informa-
given in chapter 111. The generalities of the tion is not yet usable in design. The chapters may
concepts involved have been given merely to give only a qualitative picture of the problem, or
clarify the general approach to the problem. they may merely indicate the direction of future
research. Those aspects of the compressor prob-
OBJECTIVES OF DESIGN REPORT
lem which are considered pertinent are included,
The desire to provide a sound compressor design however, regardless of the present applicability of
system has formed the basis for most research on the information.
axial-flow compressors. As a result, in this coun- The following discussion provides an over-all
try and abroad, design concepts and design tech- perspective of the material covered in this compres-
niques have been established that Wiw provide sor design compendium. Each chapter is sum-
high-performance compressors. In general, these marized briefly, and the relation of each to the
various design systems, although they may differ over-all report is indicated.
in the manner of handling details, utilize the same In order to provide proper emphasis in the de-
basic approach to the problem. This over-all sign summarization, it is desirable to establish and
report is therefore dedicated to summapizing and evaluate the essential characteristics of compres-
consolidating this existing design information. sors. Chapter I1 accomplishes this objective by
This effort may be considered to have three general first evaluating engine requirements with respect
objectives: to airplane performance. These required engine
(1) To provide a single source of compressor characteristica are then used to identify essential
design information, within which the major requirements of the compressor. Characteristics
(representative) contributions in the litera- of the compressor that are directly related to en-
ture are summarized gine performance, such as compressor pressure
(2) To correlate and generalize compressor ratio, efficiency, airflow capacity, diameter, length,
design data that are presently available and weight, are discussed. Other considerations
only in many different forms and in widely in compressor design, including offdesign require-
scattered reports ments and the relation of the compressor to the
(3) To indicate the most essential avenues inlet diffuser, combustor, turbine, and jet nozzle,
for future research, since, in a summariza- are discussed. Compressor design objectives,
tion of this type, the missing elements based on these considerations, are summarized;
(and their importance to the design system) these objectives indicate the direction in which
compressor designs should proceed.
become readily apparent
Chapter I11provides a general description of the
In this compressor report, an effort is made to compressor design system that has been adopted
present the data in a fundamental form. To for this report on the aerodynamic design of axial-
illustrate the use of these data, a representative flow compressors. The basic thermodynamic
design procedure is utilized. However, since the equations are given, and the simplifications com-
design information is reduced to basic concepts, monly introduced to permit the solution of these
it can be fitted into any detailed design procedure equations are summarized. Representative ex-
6 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

perimental data are presented to justify these the correlations of the available data, rules and
simplifications. This chapter thus provides a relations are evolved for the prediction of blade-
valid simplified model of the flow, which is the real profile performance. These relations are devel-
basis of a design system. The elements of the oped in simplified form readily applicable to
resulting design system are then individually sum- compressor design procedures
marized ;basic equations and techniques are given. Because of modifying effects (wall boundary
Finally, the limitations of layers, three-dimensional flows, etc.), blade-
pointed out, and promising element characteristica in an annular cascade
research are indicated. can be expected to differ from those obtained
The literature on plane potentid flow in cascades in two-dimensional cascades. Chapter VI1 at-
is next reviewed (ch. IV). Many of the methods tempts to correlate and summarize available
are evaluated within the bounds of limited avail- blade-element data as obtained from experimental
able information on actual use. Some of the tests in three-dimensional annular cascades (pri-
methods that have beeh used successfully are marily rotors and stators of single-stage compres-
presented in detail to illustrate the mathematical sors). Data correlations at minimum loss are
techniques and to indicate the nature of the obtained for blade elements at various radial
actual computation. The potential-flow theories positions along the blade span. The correlations
discussed include both the design and analysis are compared with those obtained from two-
problems and consider both high-solidity and dimensional cascades (ch. VI). Design rules and
low-solidity applications. Compressibility is con- procedures are recommended, and sample calcula-
sidered, but effects of viscosity are ignored. tion procedures are included to illustrate their
A necessary adjunct to this subject of two- use.
dimensional potential flow is the consideration As discussed in the preceding paragraphs,
of two-dimensional viscous effects, presented in chapters IV to VI1 deal with the two-dimensional
chapter V. In this chapter, the problem of blade-element aspect of design. The design
boundary-layer growth in the calculation of problem in the meridional or hub-to-tip plane is
twodimensional flow about compressor blade introduced and summarized in chapter VIII.
profiles is reviewed. A qualitative picture of This meridional-plane solution presumes the
boundary-layer behavior under various conditions existence of the required blade-element data to
of pressure gradient, Reynolds number, and satisfy the velocity diagrams that are established.
turbulence normally encountered in two- The general flow equations are presented, together
dimensional blade-element flow is presented. with the simplifying assumptions used to deter-
Some typical methods for computing the growth mine the design velocity distribution and flow-
and separation of laminar and turbulent boundary passage configuration. Techniques for accounting
layers are presented. Analyses for determining for effects of viscosity (particularly for wall
the total-pressure loss and the defect in circulation boundary layers) are described. The application
are discussed. of these design techniques is clarified by a sample
Because of recognized limitations of theoretical stage design calculation.
calculations such as those presented in chapters Since procedures for determining the design
IV and V, experimental blade-element data are velocity distribution and flow-passage configura-
generally required by the designer. The available tions in the meridional plane are usually iterative,
experimental data obtained in twodimensional it is desirable to have approximate techniques
cascade are surveyed and evaluated in chapter available to expedite this process of stage design.
VI. These data (for conventional compressor The equations for radial equilibrium, continuity,
blade sections) are presented in terms of sign%- energy addition, efficiency, and diffusion factor,
cant parametere and are correlated at a reference as well as vector relations, are presented in chart
incidence angle in the region of minimum loss. form in chapter IX. An example of the applica-
Variations of reference incidence angle, total- tion of the chart technique to stage design is
pressure loss, and deviation angle with cascade included.
geometry, inlet Mach number, and Reynolds In addition to the design-point problem, the
number are investigated. From the analysis and compressor designer is vitally concerned with
 OBJEemvES AND SCOPE 7
the prediction of compressor performance over a solution of this off-design problem, however, is a
range of flow conditions and speeds. Three tach- qualitative underst
niques for estimating compressor off-design per- volved. An an
formance are presented in chapter X. The fmt problem in high-press
method establishes the blade-row and over-all flow compressors is p
performance by means The principal
element characteristics. efficiency, multip
lizes generalized stage cteristics a t interme
stage-by-stage calcula intermediate-speed surge or stall-limit character-
which is based on istics. The effects of compromising stage match-
characteristics of existing compressors, may be ing to favor part-speed operation are studied.
used to estimate the complete performance map Variable-geometry methods for improving part-
of a new compressor if the compressor design con- speed performance are discussed.
ditions are specified. The advantages and limita- The design approach adopted for this series of
tions of each of these three offdesign analysis reports is based essentially on twodimensional
techniques are discussed. concepts, assuming axial symmetry and blade-
Chapter XI is the first of a group of three element flow. With the continuing trend toward
concerning the unsteady compressor operation increasing requirements in compressors, however,
that arises when compressor blade elements stall. a condition may be reached where this simplified
The field of compressor stall (rotating stall, indi- approach may no longer be adequate. Therefore,
vidual blade stall, and stall flutter) is reviewed. chapter XIV is devoted to a summarization of
The phenomenon of rotating stall is particularly those existing design methods and theories that
emphasized. Rotating-stall theories proposed in extend beyond the simplified-radial-equilibrium
the literature are reviewed. Experimental data axisymmetric design approach. Design proce-
obtained in both single-stage and multistage com- dures that attempt to remove the bwodimensionaliz-
pressors are presented. The effects of this stalled ing restrictions are presented. Various phases
operation on both aerodynamic performance and of three-dimensional flow behavior that assume
the associated problem of resonant blade vibra- importance in design are discussed, including
*

tions are considered. Methods that might be radial flows, the over-all aspects of secondary
used to alleviate the adverse blade vibrations due flows, and time-unsteady effects.
to rotating stall are discussed. As pointed out in chapter XIV, secondary
Another unsteady-flow phenomenon resulting flows represent one of the most critical aspects of
from the stalling of compressor blade elements is the three-dimensional design problems. In view
compressor surge. It may be distinguished from of the growing importance of this subject, exist-
the condition of rotating stall in that the net flow ing literature on secondary flows and three-
through the compressor and the compressor dimensional boundary-layer behavior is summa-
torque become time-unsteady. Some theoretical rized in chapter XV. The material is discussed
aspects of compressor surge are reviewed in from two aspects: (1) the principal results ob-
chapter XII. A distinction is made between tained from experimental studies, and (2) the
surge due to abrupt stall and surge due to pro- theoretical treatment of the problem. The ex-
gressive stall. Experimental observations of surge perimental phase is directed at providing a
in compressor test facilities and in jet engines are qualitative insight into the origin and nature of
summarized. the observed secondary-flow phenomena. The
The blade-element approach to the prediction theoretical results include a summary and evalu-
of off-design performance (as presented in ch. X) ation of both the nonviscous dnd the boundary-
is essentially limited to the unstalled range of layer approaches.
operation. Because of the complexity of the flow Errors in blade-element design can seriously
phenomenon when elements stall, no quantitative affect over-all compressor performance, since these
data are available to permit a precise and accurate errors not only cause deviations from desired
synthesis of over-all cornpressor performance in blade-row performance, but also alter the inlet
this range. A prerequisite to the complete conditions to the next blade row. The effects of
8 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

errors in the three basic blade-element design and more approximate method. In addition, a
parameters (turning angle, total-pressure loss, and simple technique for establishing an engine
local specEc mass flow) on compressor perform- operating line on a compressor map is reviewed.
ance are analyzed in chapter XVI. The results An available technique for matching during
are presented in the form of formulas and charts. transient operation is also discussed. The use of
These charts may be used to indicate those design this method permits engine acceleration charac-
types for which the design control problem is most teristics and acceleration time to be approximated
critical and to estimate the limits in performance for either single-spool or two-spool engines.
that can be anticipated for design data of a given
CONCLUDING REMARKS
accuracy. Typical design cases are considered,
and signiscant trends are discussed. A second The subsequent chapters in this report sum-
phase of this chapter concerns accuracy of experi- marize available information on the aerodynamic
mental measurements. Proper interpretation and design of axial-flow compressors. It is recog-
analysis of experimental data require that meas- nized that many techniques have been proposed
urements be precise. This chapter presents a for describing the flow in an axial-flow compressor
systematic evaluation of the effect of measure- and for accounting for the complex flow phenomena
ment errors on the measured compressor perform- that are encountered. Obviously, consideration
ance. These results, which are also presented in of all of these techniques is impossible. However,
chart form, can be used to estimate the required the available literature in t.he field is reviewed
accuracy of instrumentation. extensively, and the material presented is con-
One of the most important aspects of gas- sidered to be representative and pertinent. In
turbine engine design, particularly for applications general, the attempt is made to present the in-
where high power output and wide operating formation in its most basic form, so that it may
range are required, is that of compressor and tur- be fitted into any generalized design system.
bine matching. The existing literature on com- Because of the many diBcult and involved
pressor and turbine matching techniques, which problems associated with compressor design, very
can be used to compromise properly the aero-
few of these underlying problems are treated with
dynamic design of the compressor and turbine to
achieve the best over-all engine, is summarized finality. In some cases, the problem is only
in chapter XVII. Both single-spool and two- partly defined. Nevertheless, many successful
spool engines are considered. For equilibrium designs (by present standards, a t least) have been
operation, the basic matching technique, which made with the use of this information. The voids
involves the superposition of compressor and in the information clearly indicate the research
turbine maps, is presented, as well as a simplified problems for the future.
 CHAPTER VI
EXPERIMENTAL FLOW I N TWO-DIMENSIONAL CASCADES
t
By SEYMOUB
LIEBLDIN \

Available e x p e r i m d two-dimensional-cascade the correlation of isolated data was very difEcult.

data for conventional cornpressor blade sections are Some efforts were made, however, to correlate
correlated. The two-dimensional cascade and some limited experimental data for use in compressor
of the principal aerodynamic factors involved in its design (e.g., ref. 187). The British, in particular,
operation are first briefly described. Then the data through the efforts primarily of Carter and Howell,
are analyzed by examining the variation of cascade appear to have made effective use of their early
performance at a reference incidence angle in the cascade investigations (refs. 31 (pt. I) and 188
region of minimum loss. Variations of reference to 190).
incidence angle, total-pressure loss, and deviution In recent years, the introduction of effective
angle with cascade geometry, inlet Mach number, tunnel-wall boundary-layer removal for the estab-
and Reynolds number are investigated. lishment of true two-dimensional flow gave a
From the anulysis and the correlations of the substantial impetus to cascade analysis. In
available datu, rules and relations are evolved for the particular, the porous-wall technique of boundary-
prediction of the magnitude of the reference totat- layer removal developed by the NACA (ref. 191)
pressure loss and the reference deviation and inci- was a notable contribution. The use of effective
dence angles for conventional 6 W e proJles. These/\
relations are developed in simplged forms readily ' tunnel boundary-layer control has resulted in more
consistent systematic test data (refs. 39, 54, 123,
applicable to compressor design procedures.
INTRODUCTION
P@
Because of the complexity and three-dimensional
and 192 (pt. 11))and in more significant two-
dimension2 comparisons between Georeticd and
experimental performance (refs. 98, 167 (pt. I),
and 193). With the availability of a considerable
character of the flow in multistage axial-flow amount of consistent data, it has become feasible
compressors, various simplified approaches have to investigate the existence of general relations
been adopted in the quest for accurate blade- among the various cascade flow parameters.
design data. The prevailing approach has been Such relations curtail the amount of future
to treat the flow across individual 'compressor experimental data needed and also result in more
blade sections as a two-dimensional flow. The effective use of the data currently available.
use of twodimensionally derived flow characteris- Since the primary function of cascade informa-
tics in compressor design has generdy been satis- tion is to aid in the design of compressors, the
factory for conservative units (ch. 111). present W t e r expresses the existing cascade data
In view of the limitations involved in the in terms of parameters applicable to compressor
theoretical calculation of the flow about two- design. Such expression not only facilitates the
dimensional blade sections (chs. IV and V), design of moderate compressors but also makes
experimental investigations of two-dimensional possible a rapid comparison of cascade data with
cascades of blade sections were adopted as the data obtained from advanced 'high-speed com-
principal source of bladedesign data. Early pressor configurations. Since the bulk of the
experimental cascade results (e.g., refs. 184 to available cascade data has been obtained at low
186), however, were marked by a sensitivity to speed (Mach numbers of the order of O.l), the
individual tunnel design and operation. This was question of applicability to such high-speed units
largely a result of the failure to obtain true two- is very significant. It is necessary to determine
dimensional flow. Under these circumstances, which flow parameters can or cannot be applied,
183

1 Preceding Page blank '

184 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

to what extent the low-speed data are directly compressibility correction factor in loss
usable, and whether corrections can be developed equation
in those areas where the low-speed data cannot be Kf correction factor in incidence-angle rela-
used directly. tion
In this chapter. the available cascade data Ka correction factor in deviation-angle rela-
obtained from a large number of tunnels are tion
reworked in terms of what are believed to be M Mach number
significant parameters and are correlated in m,mc factors in deviation-angle relation
generalized forms wherever possible. The per- n slope factor in incidence-angle relation
formance parameters considered in the correlation P total or stagnation pressure
are the outlet-air deviation angle and the cascade P static or stream pressure
losses expressed in terms of blade-wake momentum Re, Reynolds number based on chord length
thickness. The correlations are based on the s blade spacing
variations of the performance parameters with t blade maximum thickness
cascade geometry (blade profile shape, solidity, V air velocity
chord angle’l and inlet flow conditions. In view of Y coordinate normal to axis
the dif3Frculties involved in establishing correla- 2 coordinate along axis
tions over the complete range of operation of the a- angle of attack, angle between inlet-air
cascade at various Mach number levels, the direction and blade chord, deg
analysis is restricted to an examination of cascade B air angle, angle between air velocity and
performance a t a reference incidence-angle loca- axial direction, deg
tion in the region of minimum loss. AB air-turning angle, pI-p2, deg
The chapter is divided into four main sections: To blade-chord angle, angle between blade
(1) a brief description of the two-dimensional chord and axial direction, deg
cascade and of the parameters, concepts, and data 6 wake full thickness
involved in the analysis; (2) an analysis of the 6* wake displacement thickness
variation of the reference incidence angle with 6O deviation angle, angle between outlet-air
cascade geometry and flow conditions; (3) an direction and tangent to blade mean
analysis of the variation of total-pressure loss a t camber line a t trailing edge, deg
the reference incidence angle; and (4) an analysis 6; deviation angle of uncambered blade
of the variation of deviation angle a t the refer-
section, deg
ence incidence angle.
e* wake momentum-defect thickness
SYMBOLS K blade angle, angle between tangent to
The following symbols are used in this chapter: blade mean camber line and axial
direction, deg
A flow area
b exponent in deviation-angle relation P density
e chord length U solidity, ratio of chord to spacing
D diffusion factor (based on over-all veloc- (P blade camber angle, difference between
ities) blade angles a t leading and trailing
DlOC local diffusion factor (based on local
velocities)
-w edges, K ~ - K ~ , deg
total-pressure-loss coefficient
d exponent in wake velocity-distribution Subscripts:
relations
av average
PH function
wake form factor, 6*p* ;.e. incompressible equation
i incidence angle, angle between inlet-air inc incompressible
direction and tangent to blade mean 1 lower surface
camber line a t leading edge, deg mux maximum
i0 incidence angle of uncambered blade min minimum
section, deg ref reference
 EXPERLMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 185

FIGUBE
l23.-Layout of conventional low-speed cascade tunnel (ref. 168).

sh blade shape
t blade maximum thickness
U upper surface
z axial direction
e tangential direction
0 free stream
1 station a t cascade inlet
2 station at cascade exit (measuring station)
10 10 percent thick
PRELIMINARY CONSIDERATIONS
DESCRIPTION OF CASCADE

A schematic diagram of a low-speed two-

dimensional-cascade tunnel is shown in-figure 123
to illustrate the general tunnel layout. The
principal components of the conventional tunnel I ine I
I Measuring
are a blower, a diffuser section, a large settling I plane
chamber with honeycomb and screens to remove
any swirl and to ensure a uniform velocity dis-
tribution, a contracting section to accelerate the
flow, the cascade test section, and some form of
outlet-air guidance. The test section contains
a row or cascade of blades set in a mounting device FIGURE124.-Nomenclature for cascade blade.
that can be altered to obtain a range of air inlet
angles (angle p1 in figs. 123 and 124). Variations of suction through slots or porous-wall surfaces.
in blade angle of attack are obtained either by Examples of different tunnel designs or detailed
rotating the blades on their individual mounting information concerning design, construction, and
axes (i.e., by varying the blade-chord angle -yo) operation of the two-dimensional-cascade tunnel
while maintaining a fixed air angle or by keeping can be obtained from references 39, 122, 168, 191,
the blade-chord angle fixed and varying the air and 194.
inlet angle by rotating the entire cascade. Outlet ignating cascade '

flow measurements are obtained from a traverse

layer control in the cascade is provided by mea& related mean lines (refs. 39 and 123), the circular-
186 AERODyNABdIC DESIGN OF AXIAL-FLOW COMPRESSORS

arc mean line (ref. 3 1, p t. I), and the parabolic-arc dicated in figure 124, these angles are based on
mean line (ref. 192, pt. 11). Two popular basic the tangents to the blade mean camber line at the
thickness distributions are the NACA 65-series leading and trailing edges. The use of the devia-
thickness distribution (ref. 39) and the British tion angle, rather the turning angle, as a
C.4 thickness distribution (ref. 31, pt. I). A measure of the air direction has the advan-
high-speed profile has also been obtained from the tage, for coyelation purposes, of a generally small
construction of arc upper and lower variation with incidence angle. Air-turning angle
surface (ref. 40); is referred to as the is related to the
double-circular-arc blade. angles by
PEBFOBMANCE PARAMETEBB Ag=p+i--s” (57)
The performance of cascade blade sections has Incidence angle is considered positive when it
generally been presented as plots of the variation tends to increase the air-turning angle, and devia-
of air-turning angle, lift coefficient, and flow losses tion angle is considered positive when it tends to
against blade angle of attack (or incidence angle) decrease the air-turning angle (fig. 124).
for a given cascade solidity and blade orientation. The use of incidence and deviation angles re-
Blade orientation is expressed in terms of either quires a unique and reasonable definition of the
fixed air inlet angle or fixed blade-chord angle. blade mean-line angle at the leading and trailing
Flow losses have been expressed in terms of 60- edges, which may not be possible for some blade
efficients of the drag force and the defects in shapes. The principal difiiculty in this respect is
outlet total pressure or momentum. A recent in- in the 65-(Alo)-seriesblades (ref. 39), whose mean-
vestigation (ref. 156) demonstrates the significance line slope is theoretically infinite at the leading
of presenting cascade losses in terms of the thick- and trailing edges. However, it is still possible to
ness and form characteristics of the blade wakes. render these sections usable in the analysk by
In this analysis, the cascade loss parameters arbitrarily establishing an equivalent circulai -arc
considered are the wake momentum-thickness mean camber line. As shown in figure 125, the
ratio O*/c (ref. 156) and the total-pressure-loss equivalent circular-arc mean line is obtained by
coefficient Wl, defined as the ratio of the average drawing a circular arc through the leading- and
loss in total pressure across the blade to the inlet trailing-edge points and the point of maximum
dynamic head. Cascade losses are considered in camber at the midchord position. Equivalent
terms of Ul, since this parameter can be conven- incidence, deviation, and camber angles can then
iently used for the determination of compressor be established from the equivalent circular-arc
blade-row efficiency and entropy gradients. The mean line as indicated in the figure. The rela-
parameter e*/c represents the basic wake develop- tion between equivalent camber angle and isolated-
ment of the blade profile and as such constitutes airfoil lift coefficient of the NACA 65-(Alo)-series
a significant parameter for correlation purposes. mean line is shown in figure 126.
Values of e*/c were computed from the cascade A typical plot of the cascade performance pa-
loss data according to methods similar to those rameters used in the analysis is shown in figure
presented in reference 156. The diffusion factor 127 for a conventional blade section at fixed
D of reference 9 was used as a measure of the solidity and air inlet angle.
blade loading in the region of minimum loss.
DATA SELECTION
In the present analysis, it was necessary to use
a uniform nomenclature and consistent correlation In selecting data sources for use in the cascade
technique for the various blade shapes considered. performance correlations, it is necessary to con-
It was believed that this could best be accom- sider the degree of twodimensionality obtained
plished by considering the approach characteris- in the tunnel and the magnitude of the test
tics in terms of air incidence angle i, Reynolds number and turbulence level.
the acteristics in terms of the camber Two-dimensionality.-As indicated previously,
angle 9, and the air-turning characteristics in test results for a given cascade geometry obtained
terms of the deviation angle 13’ (fig. 124). As in- from dif€erent tunnels may vary because of a fail-
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 187
/Point of
/ maximum
fixed incidence angle are presented in figure 128.
65-(Ala) -series I camber S i a r pronounced effects are observed on the
deviation angb. AB discussed in chapter V, the
loss variation with Reynolds number is associated
primarily with a local or complet
the laminar boundary layer on the
The data used in the correlation are restricted to
values of blade-chord Reynolds number from about
2.OX1O6 to 2.5X1OS in order to minimize the
effects of different Reynolds numbers. Free-
stream turbulence level was not generally deter-
mined in the various cascade tunnels.
In some cases (refs. 39 and 195, e.g.), in tunnels
with low turbulence levels, marked local laminar-
FIGURE 125.-Equivalent circular-arc mean line for separation effects were observed in the range of
NACA 65-(Alo)-series blades. Reynolds number selected for the correlation.
Illustrative plots of the variation of total-pressure-
ure to achieve true two-dimensional flow across loss coefficient with angle of attack for a cascade
the cascade. Distortions of the true two-dimen- with local laminar separation are shown in figure
sional flow are caused by the tunnel-wall bound- 129. In such inbtances, it was necessary to esti-
ary-layer growth and by nonuniform inlet and mate the probable variation of loss (and deviation
outlet flow distributions (refs. 191 and 168). In angle) in the absence of the local separation
modern cascade practice, good flow twodimen- (as indicated in the figure) and use values
sionality is obtained by the use of wall-boundary- obtained from the faired curves for the correlac
layer control or large tunnel size in conjunction tions.
with a large number of blades, or both. Ex- The specific sources of data used in the analysis
amples of cascade tunnels with good twodimen- are indicated by the references listed for the vari-
sionality are given by references 39 and 194. ous performance correlations. Details of the tun-
The lack of good two-dimensionality in cascade nel construction and operation and other pertinent
testing affects primarily the air-turning angles and information are given in the individual references.
blade surface pressure distributions. Therefore, APPROACH
deviation-angle data were rejected when the two-
dimensionality of the tunnel appeared questionable In a correlation of two-dimensional-cascade data
(usually the older and smaller tunnels). Practi- that is intended ultimately for use in compressor
cally all the cascade loss data were usable, however, blade-element design, the variations of perform-
since variations in the measured loss obtained from ance parameters should be established over a wide
a given cascade geometry in different tunnels will range of incidence angles. Experience shows (fig.
generally be consistent with the measured diffusion 130) that the variation of loss with incidence angle
levels (unless the blade span is less than about 1 for a given blade section changes markedly as the
or 2 inches and there is no extensive boundary- inlet Mach number is increased. Consequently,
layer removal). correlated low-speed blade performance at high
Reynolds number and turbulence.-For the and low incidence angles is not applicable at high
same conditions of two-dimensionality and test- Mach numbers. The low-speed-cascade perform-
section Mach number, test results obtained from ance is therefore considered at some reference
cascades of the same geometry may vary because point on the general loss-against-incidence-angle
of large differences in the magnitude of the curve that exhibits the least variation in location
bladechord Reynolds number and the and in magnitude of performance parameters as
free-stream turbulence. Examples of the effect Mach number is increased.
of Reynolds number and turbulence on the The reference location herein is selected as the
losses obtained from a given blade section at point of minimum loss on the curve of total-
188 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS
 EXPERIMENTAL FLOW IN ~ O - D l A I E N d l f o N A L CASCADES 189
14
0
al
0
9 IO
m

6
.55
a
$5
.I 2

8
Q .08
2
'3 .04

Q12 -8 -4 0 4 8 12 16
Incidence angle, i ,deg

FIGURE 127.-Illustration of basic performance parameters

for cascade analysis. Data obtained from conventional
blade geometry in low-speed two-dimensional tunnel.

.04

.02 (a) NACA 65-810 blade. Inlet-air angle, 30'.

(b) NACA 65-(12)lO blade. Inlet-air angle, 45O.
FIGUBE129.-Loss characteristics of cascade blade with
local laminar separation. Solidity, 1.5; blade-chord
e Reynolds number, 2.45X 106 (ref. 39).
.- . I C
-
.-0
c
W
u
0
u) .OE
pressure loss against incidence angle. For con-
u)
-
0 ventional low-speed-cascade sections, the region
of low-loss operation is generally flat, and it is
:
I

!2 .OE difEcult to establish precisely the value of incidence

2 angle that corresponds to the minimw loss. For
-
F
0
e practical purposes, therefore, since the curves of
I-" .04 loss coefficient against incidence angle are gener-
ally symmetrical, the reference minimum-loss loca-
tion was established at the middle of the low-loss
.02 range of operation. SpecZcally, as shown in figure
131, the reference location is selected as the inci-
dence angle at the midpoint of the range, where
0 range is defined as the change in incidence angle
corresponding to a rise in loss coefficient equal to
(a) NACA 65-(12) 10 blade. Inlet-air angle, 45O; solidity, the minimum value. Thus, for conventional cas-
1.5 (ref.39) : cade sections, the midrange reference location is
(b) Lighthill blade, 50 percent laminar flow. Inlet-air considered coincident with the point of minimum
angle, 45.5'; solidity, 1.0 (ref. 167, pt. I).
loss. In addition to meeting the abovementioned
FIGURE 128.-Effect of blade-chord Reynolds number and
free-stream turbulence on minimum-loss coefficient of requirement of small variation with inlet Mach
cascade blade section in two-dimensional tunnel. number, the reference minimum-loss incidence
' 190 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

Incidence angle, i, deg

(a) C.4 Circular-arc blade. Camber angle, 25'; maximum- (c) Double-circular-arc blade. Camber angle, 25'; maxi-
thickness ratio, 0.10; solidity, 1.333; blade-chord angle, mum-thickness ratio, 0.105; solidity, 1.333; blade-chord
42.5' (ref. 40). (b) C.4 Parabolic-ard blade. Camber angle, 42.5' (ref- 40). (d) Sharp-nose blade. Camber
angle, 25' ; maximum-thickness ratio, 0.10; solidity, angle, 27.5'; maximum-thickness ratio, 0.08; solidity,
1.333; blade-chord angle, 37.6' (ref. 40). 1.15; blade-chord angle, 30' (ref. 205).
FIGWRB
130.-Effect of inlet Mach number on loss characteristics of cascade blade sections.

angle (as compared with the optimum or nominal

incidence settings of ref. 196 or the design incidence
setting of ref. 39) requires the use of only the loss
variation and also permits the use of tke diffusion
factor (applicable in region of minimum loss) as
a measure of the blade loading.
At this point, it should be kept in mind that
the reference minimum-loss incidence angle is
not necessarily to be considered as a recommended incidence ongle, I' ,deg
design point for 'aompressor application. The
selection of the be& incidence angle for a par- F'IGURE 131.-Definition of reference minimum-loss iaci-
dence angle.
ticular blade element in a multistage-compressor
design is a function iof many considerations, such reference location is established primarily for
as the location of the blade row, the design Mach purposes of analysis.
number, and the type and application of the design. Of the many blade shapes currently in use
In general, there is no one universal definition in compressor design practice (i.e., NACA 65-
of design or best incidence angle. The cascade series, C-series circular arc, parabolic arc, double
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 191
circular arc), data sufficient to permit a reasonably in the use of reference-incidence-angle data from
complete and significant correlation have been these two methods.
published only for the 65-(Alo)-series blades of With the definition of reference incidence angle,
reference 39. Therefore, a basic correlation of performance parameters, and analytical approach
the 65-(Al0)-series data had to be established established, the procedure is first to de
first and the results used as a guide or foundation how the value of the reference min
for determining the corresponding performance incidence angle varies with cascade geometry and
trends for the other blade shapes for which only flow conditions for the available blade profiles.
limited data exist. Then the variation of the performance parameters
Since the ultimate objective of cascade tests is determined at the reference location (asindicated
is to provide information for designing com- in fig. 127) as geometry and flow are changed.
pressors, it is desirable, of come, that the structure Thus, the various factors involved can be ap-
of the data correlations represent the compressor praised, and correlation curves and charts can
situation as closely as possible. Actually, a be established for the available data. The
blade element in a compressor represents a blade analysis and correlation of cascade reference-point
section of fixed geometry (Le., fixed prosle form, characteristics are presented in the following
solidity, and chord angle) with varying inlet-air sections.
angle. In two-dimensional-cascade practice, how- INCIDENCE-ANGLE ANALYSIS
ever, variations in incidence angle have been
PBELIMINAEY ANALYSIS
obtained by varying either the inlet-air angle or
the blade-chord angle. The available systematic In an effort to obtain a general empirical rule
data for the NACA 65-(A,,)-series blades (ref. 39) for the location of the reference minimum-loss
have been obtained under conditions of fixed incidence angle, it is first necessary to examine the
inlet-air angle and varying blade-chord angle. principal influencing factors.
Since these data form the foundation of the It is generally recognized that the low-loss
analysis, it was necessary to establish the cascade region of incidence angle is identified with the
absence of large velocity peaks (and subsequent
performance correlations on the basis of fixed
decelerations) on either blade surface. For infi-
inlet-air angle. Examination of limited unpub- nitely thin sections, steep velocity gradients are
lished low-speed data indicate that, as illustrated avoided when the front stagnation point is located
in figure 132, the loss curve for constant air inlet at the leading edge. This condition has fre-
angle generally falls somewhat to the right of quently been referred to as the condition of
the constant-chord-angle curve for fixed values “impact-free entry.” Weinig (ref. 80) used the
of Dl and yo in the low-loss region of the curve. criterion of stagnation-point location to establish
Values of minimum-loss incidence angle for fixed the variation of impact-free-entry incidence angle
81 operation are indicated to be of the order of lo for infinitely thin circular-arc sections from
or 2’ greater than for fixed yo operation. An potential-flow theory. Results deduced from ref-
approximate allowance for this difference is made erence 80 are presented in figure 133(a). The
minimum-loss incidence angle is negative for
infinitely thin blades and decreases linearly with
camber for fixed solidity and blade-chord angle.
While there is no definite corresponding i-inc
dence-angle theory for thick-nose blades with
rounded leading edges, some equivalent results
have been obtained based on the criterion that the
0
location of the stagnation point in the leading-
Incidence angle, I‘, deg edge region of a thick blade is the controlling
FIGURE132.-Qualitative comparison of cascade range
factor in the determination of the surface velocity
characteristics at constant blade-chord angle and con- distributions. Carter, in reference 190, showed
stant inlet-air angle (for same value of & in region of semitheoreticdy on this basis that optimum
minimum loss). incidence angle (angle at maximum lift-drag ratio)
 AERODYNAMIC DESIGN OF AXIAL-F‘LOW COMPRESSORS

I I I I I 1
20 40 60
Comber angle,

(a) “Impact-free-entry” incidence angle for infinitely thin C-series profiles according to semitheoretical develop-
blades accordingto potential theory of Weinig (ref. 80). menta of Carter et a2. (refs. 190 and 196). Outleeair
(b) “Optimum” incidence angle for 10-percent-thick angle, 20°.
FIQURE 133.-Variation of reference incidence angle for circular-air-mean-fine blades obtained‘from theoretical or semi-
theoretical investigations.

for a conventional 10-percent-thick circular-arc A preliminary examination of experimental

blade decreases with increasing camber angle. cascade data showed that the minimum-loss
The results of reference 190 were followed by incidence angles of uncambered sections @ = O )
generalized plots of optimum incidence angle in of conventional thicknesses were not zero, as
reference 196, which showed, as in figure 133(a), indicated by theory for infinitely thin blades (fig.
that optimum incidence angle for a 10-percent- 133(a)), but always positive in value. The
thick C-series blade varies with camber angle, appearance of positive values of incidence angle
solidity, and blade orientation. (In these ref- for thick blades is attributed to the existence of
erences, blade orientation was expressed in terms velocity distributions at zero incidence angle that
of air outlet angle rather than blade-chord angle,) are not symmtrical on the two surfaces. Typical
The plot for an outlet- ngle of 20° is shown in plots illustrating the high velocities generally
figure 133(b). Apparently, the greater the blade observed in the inlet region of the lower (pressure)
circulation, the lower in magnitude the urn- surface of thick uncambered blades at zero inci-
e to dence angle are shown in fime 134. Apparently,
tion of an increase in incidence angle from the zero value
minimUm-loss incidence angle for conventional is necessary in order to reduce the lower-surface
city to a more equitable distribution that
by thin-airfoil theory. in a minimum of the over-all loss. This
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 193
zero-camber thickness effect will appear only for
blade-chord angles between Oo and goo) since, as
indicated by the highly simplified one-dimensional
model of the blade passage flow in figure 135, the
velocity distributions at these limit angles are
symmetrical.
The effect of blade thickness blockage on
impact-free-entry incidence angle
(uncambered) blades of constant chordwise thick-
ness in incompressible twodimensional flow is
investigated in reference 34. The results of
reference 34 are plotted in terms of the parameters
used in this analysis in figure 136. It is reasonable
to expect that similar trends of variations of
zero-camber reference. minimum-loss incidence
angle will be obtained for compressor blade
profiles.
On the basis of the preceding analysis, therefore,
it is expected that, for low-speed-cascade flow,
reference minimum-loss incidence angle will gen-
Percent chord
erally be positive at zero camber and decrease with
increasing camber, depending on solidity and
(a) Inlet-air angle, 60'; solidity, 1.5. blade-chord angle. The available theory also
(b) Inlet-air angle, 30'; solidity, 1.0. indicates that the variation of reference incidence
FIGURE134.-Illustration of velocity distribution for angle with camber at fixed solidity and chord
uncambered blade of conventional thickness at zero
incidence angle. Data for 65-(0)lO blade of reference angle might be essentially linear. If so, the
39. variations could be expressed in terms of slope

-0Q.
v,

-<--=- tlf
Yt
(a) (b) (C)

(a) yo=Oo. (b) yo=9O0. (c) O'<y0<90'.

FIQURE
135.-Effect of blade thickness of surface velocity at zero incidence angle for uncambered airfoil section according
to simplifted one-dimensional model.
691-564 0-65-14
194 AERODYNAMIC DESIGN OF AXUL-FLOW COMPRESSORS

blades of reference 39 are to be used as the basis

for a generalized corr
blade sylapes, it is pro
reference incidence
form
zo =

where (io)1o represents the va

camber incidence angle for the
65-series thickness distribution, (K1) represents
any correction necessary for maximum blade
thicknesses other than 10 percent, and
represents any correction necessary for a blade
shape with a thickness distribution different from
that of the 65-series blades. (For a 10-percent-
thick 65-series blade, ( K J1= 1 and (Ki)sn= 1.)
FIGURE136.-Theoretical variation of “impact-free-entry” The problem, therefore, is reduced to finding the
incidence angle for constant-thickness uncambered values of n and io (through eq. (262)) as functions
sections according to developments of reference 34. of the pertinent variables involved for the various
blade profiles considered.
and intercept values, where the intercept value
represents the magnitude of the incidence angle NACA 65- (A1,)-series blades.-From the exten-
for the uncambered section (function of blade sive low-speed-cascade data for the 65-(Alo)-series
thickness, solidity, and blade-chord angle). Ref- blades (ref. 39), when expressed in terms of
erence minimum-loss incidence angle may also equivalent incidence and camber angles
vary with inlet Mach number and possibly with (figs. 125 and 126), plots of io and n can be
Reynolds number. deduced that adequately represent the minimum-
loss-incidence-angle variations of the data. The
DATA CORRELATIONS deduced values of io and n as functions of solidity
Form of correlation.-Although preliminary and inlet-air angle are given for these blades in
theory indicates that blade-chord angle is the figures 137 and 138. The subscript 10 in figure
significant blade orientation parameter, it was 137 indicates that the io values are for 10-percent
necessary to establish the data correlations in maximum-thickness ratio. Values of intercept io
terms of inlet-air angle, as mentioned previously. and slope n were obtained by fitting a straight line
The observed cascade data were found to be to each data plot of reference incidence angle
represented satisfactorily by a linear variation against camber angle for a fixed solidity and air
of reference incidence angle with camber angle inlet angle. The straight l i e s were selected so
for fixed solidity and inlet-air angle. The varia- that both a satisfactory representation of the
tion of reference minimum-loss incidence angle variation of the data points and a consistent
can then be described in equation form as variation of the resulting n and io values were
obtained.
i=i,+ncp (261) The deduced rule values and the observed
where io is the incidence angle for zero camber, data points compared in -re 139 indicate the
and n is the slope of the incidence-angle variation effectiveness of the deduced representation. In
with camber (i-io)/(p. several configurations, particularly for low cam-
Since the existence of a finite blade thickness bers, the range of equivalent incidence angle
is apparently the cause of the positive values of covered in the tests was insuf5cient to permit
io, it is reasonable to assume that both the an accurate determination of a minimum-loss
magnitude of the maximum thickness and the value. Some of the scatter of the data may be
thickness distribution contribute to the effect. due to the effects of local laminar separation in
Therefore, since the 10-percent-thick 65-series ge characteristics of the sections.
 EXPERIMENTAL FLOW IN TWO-DlMENSIONAL CASCADES 195

.
W
V

W
u
.-
c

e
W
N

Inlet-air angle, p, , deg

FIGURE
137.-Reference minimum-loss incidence angle for zero camber deduced from low-speed-cascade data of 10-per-
cent-thick NACA 65-(A13-series blades (ref. 39).

Although the cascade data in reference 39 and turbine nozzles), a wide low-loss range of
include values of inlet-air angle from 30" to 70" operation is usually obtained. The solidity
and values of solidity from 0.5 to 1.5, the deduced extrapolations were attempted because of the
variations in figures 137 and 138 are extrapolated uniform variations of the data with solidity.
to cover wider ranges of fll and u. The extrapo- However, caution should be exercised in any
lation of io to zero a t &=O is obvious. Accord- further extrapolation of the deduced variations.
ing to theory {fig. 133), the value of the slope C-Series circular-arc blades.-The various
term does not vanish a t &=O. In figure 138, thickness distributions used in combination with
therefore, an arbitrary fairing of the curves down the circular-arc mean line have been designated
to nonzero values of n was adopted as indicated. C.l, C.2, C.3, and so forth (refs. 196 to 198). In
Actually, it is not particularly critical to deter- general, the various C-series thickness distribu-
mine the exact value of the slope term a t @,=O tions are fairly similar, having their maximum
necessary to locate the reference incidence angle thickness located at between 30 and 40 percent
precisely, since, for such cases (inlet guide vanes of the chord length. The 65-series and two of
196 AERODYNAMIC DESIGN OF AXIAL-F'LOW COMPRESSORS

C
L
0
e
0
0
(c

0)
a
-
0
cn

Inlet -air angle, p, ,deg

FIGURE138.-Reference minimum-loss-incidence-angle slope factor deduced from low-speed-cascade data for NACA
65-(A1&series blades as equivalent circular arcs.

the more popular 6-series thickness distributions series mean line and a true circular arc (fig. 125),
((2.1 and (2.4) are compared on an exaggerated the applicability of the slope values in figure 138
scale in figure 140. (The 65-series profile shown to the circular-arc mean line was investigated.
is usually thickened near the trailing edge in For the recent cadcade data obtained from tunnels
actual blade construction.) having good boundary-layer control (refs. 167,
In view of the somewhat greater 'thickness (pt. I) and 199), a check calculation for the 10-
blockage in the forward portions of the C-series percent-thick C.4 circular-arc blades using figures
blades (fig. 140), it may be that the minimum-loss 137 and 138 with (Kf)8b=l.lrevealed good
incidence angles for zero camber for the C-series results. For the three configurations in reference
blades are somewhat greater than those for the 65- 199 tested at constant /31(p=30"), the agreement
series profiles; that is, (KJ&l. In the absence of between observed and predicted minimum-loss
any definitive cascade data, the value of (Kf)sh for incidence angles was within lo. For the one
the C-series profiles was arbitrarily taken to be configuration in reference 167 (pt. I) tested at
1.1. Observed minimum-loss incidence angles constant y0((p=31"), the predicted value of
for an uncambered 10-percent-thick C.4 profile minimum-loss incidence angle was 1.7' greater
(obtained from ref. 192, pt. I) are compared in than the observed value. However, in view of the
figure 141 with values predicted from the deduced general 1' to 2' difference between fixed B1 and
values for the 65-series blade (fig. 137 and fixed 7' operation (fig. 132), such a discrepancy
eq. (262)) with an assumed value of (KJab=1.l. is to be expected. On the basis of these limited
(For 10-percent thickness, (Kf)I=1.) data, it appears that the low-speed minimum-loss
In view of the similarity between the 65-(A1,)- incidence angles for the C-series circular-arc
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 197

FIGURE139.--Comparison of data values and deduced rule values of reference minimum-loss incidence angle for 65-(A10) 10
blades as equivalent circular arc (ref. 39).

blade can be obtained from the io and n values of cidence angles for the double-circular-arc blade
the 65-series blade with UG)8h=l.l. should be somewhat different from those of the
Double-circular-arc blades.-The double-circu- 65-series section, with perhaps (KJsnS 1. It
lar-arc blade is composed of circular-arc upper can also be assumed, as before, that the slope-
and lower surfaces. The arc for each surface term values of figure 138 are valid for the double-
is drawn between the point of maximum thick- circular-arc blade. From an examination of the
ness at midchord and the tangent to the circles available cascade data for the double-circular-arc
of the leading- and trailing-edge radii. The blade (9=25O, u=1.333, ref. 40; and (p=4Oo, u=
chordwise thickness distribution for the double- 1.064, ref. 197), it appears that the use of figures
circular-arc profile with 1-percent leading- and 137 and 138 with a value of (Kt),h=0.7 in equa-
trailing-edge radius is shown in figure 140. Lack tions (261) and (262) results in a satisfactory
of cascade data again prevents an accurate comparison between predicted and observed values
determination of a reference-incidencegle rule of reference incidence angle.
for the double circular arc. Since the double- Other blades.-Similar procedures can be ap-
circular-arc blade is thinner than the 65-series plied to establish reference-incidence-angle cor-
blade in the inlet region, the zero-camber in- relations for other blade shapes. Cascade data
198 ABRODWAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

Percent chord
1400.-Comparison of basic thickness distributions for conventional compressor blade sections.
FIGURE

?6
-0
speed data obtained from blades of variable
.$ thickness ratio (refs. 202 and 203) are not com-
pletely definitive, it was possible to establish a
-G 4
w preliminary thickness-correction factor for ref-
K
0 erence zero-camber incidence angle as indicated
z 2 in figure 142 for use in conjunction with equation
C
W (262).
.-u
-0
Effect of inlet Mach number.--The previous
-c correlations of reference minimum-loss incidence
0 IO 20 30 40 50 60
inlet-air angle, PI, deg angle have all been based on low-speed-cascade
data. It appears from limited highapeed data,
FIQURE 141.-Zero-camber minimum-loss incidence angle however, that minimum-loss incidence angle will
angle for 10-percent-thick C.4 profile. Solidity, 1.0
(ref. 192, pt. I). vary with increasing inlet Mach number for
certain blade shapes.
are also available for the C-series parabolic-arc The variations of minimum-loss incidence angle
blades (refs. 40, 192, 200, and 201) and the with inlet Mach number are plotted for several
NACA 65-(AI)-series blade (ref. 123); but, in blade shapes in figures 143 and 144. The extension
view of the limited use of these forms in current of the test data points to lower values of inlet Mach
practice, no attempt was made at this time to number could not generally be made because of
deduce corresponding incidenceangle rules for reduced Reynolds numbers or insufficient points
these blades. to establish the reference location at the lower
Effect of blade maximum thickness.-As indi- Mach numbers. In some instances, however, it
cated previously, some correction (expressed was possible to obtain low-apeed values of inci-
here in terms of (KJt, eq. (262)) of the base dence angle from other sources.
values of (io)lo obtained from the 10-percent- The blades of Sgure 143 show essentially no
thick 66series blades in figure 137 should exist variation of minimum-loss incidence angle with
for other values of blade maximum-thickness inlet Mach number, at least - p to a Mach number
ratio. According to the theory of the zero- of about 0.8. The blades ol: figure 144, however,
camber effect, (&)# should be zero for zero evidence a marked increase in incidence angle with
thickness and increase as maximum blade thick- Mach number. The difference in th8 variation of
ness is increased, with a value of 1.O for a thickness minimum-loss incidence angle with Mach number
ratio of 0.10. Although the very limited low- in figures 143 and 144 is associated with the
 EXPERIMFXNTAL FLOW l
X TWO-DIMENSIONAL CASCADES 199

L
0
c
U
0
*
C
0
.-
c
U
W
L
0
0

Maximum-thickness r a t i o , t / c

FIGURE142.-Deduced blade maximum-thickness correction for zero-camber reference minimum-loss incidence angle
(es. (262)).

different way the general pattern of the loss varia- 130 (c) and (d) ,the increase in loss occurs primarily
tion chmges with increasing Mach number for the on the low-incidencewgle side; md a positive
two types of blades. For the thick-nose blades, shifting of the ~ u m - l o s incidence
s angle
as illustrated in 130 (d and fi),the loss res,&. Data for other thickmnose sections in
coefficient increases with Math number at both
the high and lowincidence angles,thus tending to reference 201 Show the h1OSS to OCCW at both
m h t a j n the 8-8 point of minimum loss. For ends of the curve, but plots of reference incidence
the sharp-nose blade, as illustrated by figures angle against Mach number could not validly be
200 AERODYNAMIC DESIGN OF AXLAL-FLOW COMPRESSORS

CI,
aJ
.. I I I I I I I I l I l l l

$8
W
0
c
w 4
.-u E
0
-c -C

0
0 I I I I I A 1 - 4 1 l I I I

-4 Inlet Mach number, MI

.I .2 .3 .4 .5 .6 .7 .a
inlet Mach n u m b e r , M I
(a) Double-circular-arc blade. Camber angle, 25" ; man-
(a) C.4 Circular-arc blade. Camber angle, 254; solidity mum-thickness ratio, 0.105; solidity, 1.333; blade-chord
1.333; blade-chord angle, 42.5' (ref. 40). angle, 42.5' (ref. 40).
(b) (2.4 Parabolic-arc blade. Camber angle, 25'; solidity, (b) Blade section of reference 205. Camber angle, 27.5';
1.333; blade-chord angle, 37.5'; maximum camber at maximum-thickness ratio, 0.08; solidity, 1.15; blade-
40-percent chord (ref. 40). chord angle, 30'; maximum thickness and camber at
(c) C.7 Parabolic-arc blade. Camber angle, 40'; solidity, 50-percent chord.
1.0; blade-chord angle, 24.6O; maximum camber at
45-percent chord (ref. 216). FIGURE144.-Variation of reference minimum-loss inci-
dence angle with inlet Mach number for sharp-nose
FIGURE143.-Variation of reference minimum-loss inci- sections.
dence angle with inlet Mach number for thick-nose
sections. Maximum-thickness ratio, 0.10.
slope value n as given by equation (261). De-
made for these blades because of evidence of strong duced values of i, and n were obtained as a function
local laminar-separation effects. of B1 and u from the data for the 10-percent-thick
Since the most obvious difference between the 65-(Alo)-seriesblades of reference 39 as equivalent
blades in figures 143 and 144 is the construction circular-arc sections (figs. 137 and 138). I t was
of the leading-edge region, the data suggest that then shown that, as a first approach, the deduced
blades with thick-nose inlet regions tend to show, values of (io)lo and n in figures 137 and 138 could
for the range of inlet Mach number covered, also be used to predict the reference incidence
essentially no Mach number effect on minimum- angles of the C-series and double-circular-arc
loss incidence angle, while blades with sharp lead- blades by means of a correction to the (io>lo
ing edges will have a significant Mach number values of figure 137 (eq. (262)).
effect. The available data, however, are too The procedure involved in estimating the low-
limited to confirm this observation conclusively a t speed reference minimum-loss incidence angle of
this time. Furthermore, for the blades that do a blade section is as follows: From known values
show a Mach number effect, the magnitude of of B1 and u, (io)lo and n are selected from figures
the variation of reference incidence angle with 137 and 138. The value of (&), for the blade
Mach number is not currently predictable. maximum-thickness ratio is obtained from figure
142, and the appropriate value of ( K J Sish selected
SUMMARY
for the type of thickness distribution. For NACA
The analysis of blade-section reference mini- 65-series blades, (Kt)sh=l.O; and it is proposed
mum-loss incidence angle shows that the variation that (KJsh be taken as 1.1for the C-series circular-
of the reference incidence angle with cascade arc blade and as 0.7 for the double-circular-arc
geometry a t low speed can be established satis- blade. The value of io is then computed from
factorily in terms of an intercept value io and a equation (262); and finally i is determined from
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCAFES 201
the blade camber angle according to equation (261). Velocity variation V2 l /
across blade spacing-,,
It should be noted that the values of (&),a
given for the circular-arc blades are rather tenuous
values obtained from very limited data. The use
of the proposed values is not critical for good
accuracy; the values were included primarily for
completeness as a reflection of the anticipated
differences in the blade thickness blockage effects.
Further experimental data will be necessary to
establish the significance of such a correction.
Also, a marked increase in reference minimum-loss
incidence angle with Mach number is to be ex-
pected for sharp-nose blade sections. The magni-
tude of the Mach number correction for these
blades is currently unpredictable.
LOSS ANALYSIS
With the location of the low-speed reference
minimum-loss incidence angle established for
several conventional blade sections, the magnitude
of the losses occurring at this reference position
(fig. 127) will now be investigated. Accordingly, F’IGURE 145.4chematic representation of development of
the nature of the loss phenomena and the various surface boundary layers and wake in flow about cascade
factors Muencing the magnitude of the loss over blade sections.
a range of blade c6nfigurations and flow conditions
are first analyzed. The available experimental For incompressible flow, Po-po is equal to the
loss data are then examined to establish funda- conventional free-stream dynamic pressure poV72.
mental loss correlations in terms indicated by The total-pressure-loss coefficient is usually deter-
the analysis. mined from consideration of the total-pressure
variation across a blade spacing s (fig. 145).
PRELIMINABY ANALYSIS
A theoretical analysis of incompressible two-
Two-dimensional-cascade losses arise primarily dimensional-cascade losses in reference 156 shows
from the growth of boundary layer on the suction that the total-pressure-loss coefficient at the cas-
and pressure surfaces of the blades. These sur- cade-outlet, measuring station (where the static
face boundary layers come together at the blade pressure is essentially uniform across the blade
trailing edge, where they combine to form the spacing) is given by
blade wake, as shown in figure 145. As a result
of the formation of the surface boundary layers,
a local defect in total pressure is created, and a
certain mass-averaged loss in total pressure is
determined in the wake of the section. The loss
in total pressure is measured in terms of the total-
pressure-loss coefficient,; defined generally as the
-
ratio of the mass-averaged loss in total pressure where is the loss coefficient based on inlet
A P across the blade row from inlet to outlet sta- dynamic head, O*/c is the ratio of wake momentum
tions to some reference free-stream dynamic pres- thickness to blade-chord length, u is cascade
sure (Po-j$relt or solidity, b2 is the air outlet angle, and H2 is the
wake form factor (displacement thickness divided
by momentum thickness). The wake character-
istics in equation (264) are expressed in terms of
202 AERODYNAMIC DESIGN OF AXIATJ-FLOW COMPRESSORS

conventional thickness in a plane normal to the Caseade-inlet Mach number also influences the
wake (i.e., normal to the outlet flow) at the meas- magnitude of the subsonic diffusion for a fixed
uring station. Definitions of wake characteristics cascade. This Mach number effect is the con-
and variations in velocity and pressure assumed ventional effect of compressibility on the blade
by the analysis are given in reference 156. The velocity distributions in subsonic flow. Com-
analysis further indicates that the collection of pressibility causes the maximum local velocity on
terms within the braces is essentially secondary the blade surface to increase a t a faster rate than
(since H2is generally 5 about 1.2 a t the measuring the inlet and outlet velocities. Accordingly, the
station), with a magnitude of nearly 1 for conven- magnitude of the surface diffusion from maximum
tional unstalled configurations. The principal velocity to outlet velocity becomes greater as
determinants of the loss in total pressure at the inlet Mach number is increased. A further
cascade measuring station are, therefore, the cas- secondary influence of Mach number on losses is
cade geometry factors of solidity, air outlet and obtained because of an increase in losses associated
air inlet angles, and the aerodynamic factor of with the eventual mixing of the wake with the
wake momentum-thickness ratio. surrounding free-stream flow (ref. 37).
Since the wake is formed from a coalescing of On the basis of the foregoing considerations,
the pressure- and suction-surface boundary layers, therefore, it is expected that the principal factors
the wake momentum thickness naturally depends upon which to base empirical cascade-wake-
on the development of the blade surface boundary thickness correlations should be velocity diffusion,
layers and also on the magnitude of the blade inlet Mach number, blade-chord Reynolds number,
trailing-edge thickness. The results of references and, if possible, turbulence level.
156, 202, and 204 indicate, however, that the DATA CORBELATIONS
contribution of conventional blade trailing-edge
thickness to the total loss is not generally large Velocity diffusion based on local velocities.-
for compressor sections; the preliminary factor Recently, several investigations have been re-
in the wake development is the blade surface ported on the establishment of simplified diffu-
boundary-layer growth. In general, it is known sion parameters and the correlation of cascade
(ch. V, e.g.) that the boundary-layer growth on losses in terms of these parameters (refs. 9,38, and
the surfaces of the blade is a function primarily 156). The general hypothesis of these diffusion
of the following factors: (1) the surface velocity correlations states that the wake thickness, and
gradients (in both subsonic and supersonic flow), consequently the magnitude of the loss in total
(2) the blade-chord Reynolds number, and (3) pressure, is proportional to the diffusion in
the free-stream turbulence level. velocity on the suction surface of the blade in
Experience has shown that blade surface velocity the region of the minimum loss. This hypothesis
distributions that result in large amounts of diffu- is based on the consideration that the boundary
sion in velocity tend to produce relatively thick layer on the suction surface of conventional
blade boundry layers. The magnitude of the compressor blade sections contributes the major
velocity diffusion in low-speed flow generally de- share of the wake in these regions, and therefore
pends on the geometry of the blade section and the suction-surface velocity distribution becomes
its incidence angle. As Mach number is increased, the governing factor in the determination of the
however, compressibility exerts a further influence loss. It was further established in these correla-
on the velocity diffusion of a given cascade tions that, for conventional velocity distributions,
geometry and orientation. If local supersonic the diffusion in velocity can be expressed signifi-
velocities develop a t high inlet Mach numbers, cantly as a parameter involving the difference
the velocity difFusion is altered by the formation between some function of the measured maximum
of shock waves and the interaction of these shock suction-surface velocity V,,, and the outlet
waves with the blade surface boundary layers. velocity Vz.
The losses associated with local supersonic flow Reference 38 presents an analysis of blade-
in a cascade are generally greater than for subsonic loading limits for the 65-(Alo)10 blade section in
flow in the same cascade. The increases in loss terms of drag coefficient and a diffusion parameter
are frequently referred to as shock losses. given for incompressible flow by (va,,,-V:)/vz,,,.
 EXPERIMENTAL F'LOW DN TWO-DIMENSIONAL CASCADES 203
Results of an unpublished analysis of cascade the basic friction loss (surface shear stress) of the
losses in terms of the momentum thickness of the flow and also, to a smaller extent, the effect of
blade wake (as suggested in ref. 156) indicate the finite trailing-e thickness. The correla-
that a local diffus ammeter in the form given tion of figure 146 indicates that wake
previously or in momentum-thickness ratio at reference incidence
satisfactorily correlate expe angle can be estimated from the computed local
data.' The term 'local diffusion parameter" is diffusion factor for a wide range of solidities,
used to indicate that a knowledge of the maximum cambers, and inlet-air angles. The loss relations
local surface velocity is required. The correla- of equation (264) and reference 156 can then be
tion obtained be tween calculated wake momen- used to compute the resulting loss in the total
tum-thickness ratio O*/c and local diffusion pressure.
factor given by Velocity diffusion based on over-all velocities,-
vmU2-v' In order to include the cases of blade shapes for
DlOC=
V- (265) which velocity-distribution data are not available,
a diffusion parameter has been established in
obtained for the NACA 65-(Al,)-series cascade reference 9 that does not require a specific knowl-
sections of reference 39 at reference incidence edge of the peak local suction-surface velocity.
angle is shown in figure 146. Values of wake Although originally derived for use in compressor
momentum-thickness ratio for these data were design and analysis, the diffusion factor of refer-
computed from the reported wake coefficient ence 9 can also be applied in the analysis of cascade
values according to methods similar to those losses. The diffusion factor of reference 9 at-
discussed in reference 156. Unfortunately, blade tempts, through several simplifying approxima-
surface velocity-distribution data are not available tions, to express the local diffusion on the blade
for the determination of the diffusion factor for suction surface in terms of over-all (inlet or outlet)
other conventional blade shapes. velocities or angles, quantities that are readily
determined. The basis for the development of
the over-all diffusion factor is presented in detail
in reference 9 and is indicated briefly in figure 147.
The diffasion factor is given by

which, for incompressible two-dimensional-cascade

flow, becomes
Local diffusion facto;,

FIQUBE146.-Variation of computed wake momentum-

thickness ratio with local diffusion factor at reference
incidence angle for low-speed-cascade data of NACA As in the case of the local diffusion factor, the
65-(A,0) 10 blades (ref. 39). diffusion factor of equation (266) is restricted to
the region of minimum loss.
The correlation of figure 146 indicates the Cascade total-pressure losses at reference mini-
general validity of the basic diffusion hypothesis. mum-loss incidence angle are presented in refer-
At high values of diffusion (greater than about ence 9 as a function of diffusion factor for the
0.5), a separation of the suction-surface boundary blades of reference 39. In a further unpublished
layer is suggested by the rapid rise in the momen- analysis, a composite plot of the variation of
tum thickness. The indicated nonzero value of computed wake momentum-thickness ratio with
momentum thickness at zero diffusion represents D at reference minimum-loss incidence angle was
obtained from the available systematic cascade
aA later analysis of cascade totfd-pressure losses is given in Andy& of
Expe-mtal Low-Speed IAXS a d Stall CharaoteristieOf W0-D
data (refs. 39 and 192) as shown in figure 148.'
CompressorBlade Caseadesby sepplour Liebein. NACA R M E67A28.1957. Blade maximum thickness was 10 percent in all
204 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS
v) .06
In
W
e
1
32.04
Ta3
s 0.02
c-
e
.P
W L

E
r:
0 .I .2 3 .4 .5 .6 .7 .8
Diffusion factor,D
-
.-V
FIQUBE148.-Variation of computed wake momentum-
-
0
W thickness ratio with overall diffusion factor at reference
> incidence angle for low-speed systematic cascade data
of references 39 and 192. Blade maximum-thickness
Pressure ratio, 0.10; Reynolds number, =2.5X 1W.
surface
by the increased rise in the wake momentum
thickness for values of diffusion factor greater
than about 0.6.
For situations in which the determination of
a wake momentum-thickness ratio cannot be made,
a significant loss analysis may be obtained if a
simplified total-pressure-loss parameter is used
that closely approximates the wake thickness.
Since the terms within the braces of equation

2(
sy
(264) are generally secondary factors, a loss pa-
rameter of the form Ul should con-
stitute a more fundamental expression of the basic
Ioss across a blade element than the loss coefficient
alone. The effectiveness of this substitute loss
parameter in correlating two-dimensional-cascade
losses is illustrated in figure 149(a) for all the data
for the NACA 65- (A,,)-series blades of reference
39. (Total-pressure-loss coefficients were com-
puted for the data from relations given in ref. 9.)
A generalized correlation can also be obtained in
terms of ;J1 ~ 2uB Z aa, shown in figure 149@), but
its effectiveness as a separation indicator does not
appear to be as good. Such generalized loss
parameters are most effective if the wake form
does not vary appreciably among the various
data considered.
Effect of blade maximum thickness.-Since an
increase in blade maximum-thickness ratio in-
FIQURE147.-Basis of development of diffusion factor creases the magnitude of the surface velocities
for cascade flow from reference 9. D=v"'az-v2 (and therefore the diffusion), higher values of
Vas wake momentum-thickness ratio would be expected
= vmaz- vs;V,,, = V,+f
Vl
('9); thus, equations (54) for thicker blades. From an analysis of limited
and (266). available data on varying blade maximum-
thickness ratio (refs. 202 and 203), it appears that
cases. A separation of the suction-surface the effect of blade thickness on wake momentum-
boundary layer at high blade loading is indicated thickness ratio is not large for conventional
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 205
cascade configurations. For example, for an in-
crease in blade maximum-thickness ratio from
0.05 to 0.10, an increase in O*/c of about 0.003 at
.04
D of about 0.55 and an increase of about 0.002 at
D of about 0.35 are indicated. The greater in-
.02
crease in wake O*/c at the higher diffusion level is
understandable, since the rate of change of O*/c
with DIocincreases with increasing diffusion (see 0
fig. 146).
If blade surface velocity dBtributions can be
determined, then the thickness effect will auto-
matically be included in the evaluation of the Diffusion factor, D
resulting local diffusion factor. When an over-
all diffusion factor such as equation (54) is used,
variations in blade thickness are not reflected in
(a) BasedonGI
(z
-
(b) Baaed on Z1.
;)2*

the corresponding loss prediction. However, in FIGURE149.-Variation of loss parameter with diEusion
view of the small observed effect and the scatter factor at reference minimum-loss incidence angle
of the original P / c against D correlation of figure computed from low-speed-cascade data of NACA
148, it is believed that a thickness correction is 65-(A,0)10 crrscade blades (ref. 39).
unwarranted for conventional thickness ranges.
150. Loss variations with Reynolds number
However, the analysis does indicate that, for high
diffusion and high solidity levels, it may be ad- over a range of incidence angles for a given blade
visable to maintain blade thickness as small as shape are shown in figure 151. A composite plot
practicable in order to obtain the lowest loss at of the variation of total-pressure-loss coefficient
the reference condition. .I5
Thus, the plots of figures 146, 148, and 149
show that, when diffusion factor and wake
mom entum -thickness ratio (or total-pressure-loss .IO
parameter) are used as the basic blade-loading
and loss parameters, respectively, a generalized c
c- .05
correlation of two-dimensional-cascade loss data c
.-
a3
is obtained. Although several assumptions and 0

restrictions are involved in the use and calculation

;E
u-
W I I I I I(a)I
of these parameters, the basic diffusion approach :: 0' 8 16 24 32 40
Angle of attack, a ,deg
constitutes a useful tool in cascade loss analysis.
In particular, the diffusion analysis should be
investigated over the complete range of incidence
angle in an effort to determine generalized off-
design loss information. .
Effect of Reynolds number and turbulence,-
The effect of blade-chord Reynolds number and
turbulence level on the measured losses of cascade
sections is discussed in the section on Data Selec-
tion, in chapter V, and in references 39, 167
(pt. I), and 183. In all cases, the data reveal an Incidence angle, i,deg
increasing trend of loss coefficient with decreasing
(a) 65-Series blade 6&(12) 10. Solidity, 1.5; inlet-air
Reynolds number and turbulence. Examples oi angle, 45O (ref. 39).
the variation of the total-pressure-loss coefficient (b) Circular-arc blade lOC4/25C50. Solidity, 1.333; blade-
with incidence angle for conventional com- chord angle, 42.5' (ref. 40).
pressor blade sections at two different values 150.-Effect of Reynolds number on variation of
FIGUFUG
of Reynolds number are illustrated in Sgure loss with incidence angle.
206 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

value is in effective agreement with the value of

limiting Reynolds number deduced herein.
The desirability of conducting cascade investi-
gations in the essentially flat range of the curve of
loss coefficient ag nolds number in order
to enhance the corr of data from various
tunnels, as well as from the configurations
of a given tunnel, is indicate cade operation
in the flat range of Reynolds number may also
yield a more significant comparison between
observed and theoretically computed loss. Reyn-
olds number and turbulence level should always
be defined in cascade investigations. Furthermore,
the development of some effective Reynolds
number (ch. V) that attempfs to combine the
effects of both blade-chord Reynolds number and
turbulence should be considered for use as the
independent variable.
Effect of inlet Mach number.-In the previous
correlations, attention was centered on the various
FIGURE151.- Variation of total-pressure-loss coefficient
with blade-chord Reynolds number for parabolic-arc
factors affecting the loss of cascade blades for
blade 10C4/40 P40. Inlet-air angle, 28" to 40";
solidity, 1.333 (ref. 183).

at minimum loss with blade-chord Reynolds

number for a large number of blade shapes is
shown in figure 152. Identification data for the
various blades included in the Sgure are given in
the references. For the blades whose loss data are
reported in terms of drag coefficient, conversion to
total-pressure-loss coefficient was accomplished
according to the cascade relations presented in
reference 9. The effect of change in tunnel
turbulence level through the introduction of
screens is indicated for some of the blades.
It is apparent from the curves in figure 152 that
it is currently impossible to establish any one value
of limiting Reynolds number that will hold for all
blade shapes. (The term limiting Reynolds
number refers to the value of Reynolds number at
which a large rise in loss is obtained.) On the
basis of the available cascade data presented in
figure 152, however, it appears that serious trouble
in the minium-loss region may be encountered
at Reynolds numbers below about 2.5X105.
Carter in reference 190 places the limiting blade-
ber based on outlet velocity Blode-chord Reynolds number, Re,
at 1.5 to 2.0X105. Considering that outlet FIGURE152.-Composite plot of loss coefficient against
Reynolds number is less than inlet Reynolds er in region of minimum l o s ~
number for decelerating cascades, this quoted blade sections at low speed.
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 207
essentially incompressible or low Mach number
flow. Tests of cascade sections at higher Mach
number levels have been relatively few, primarily
because of the large power requirements and
operational difliculties of high-veloci
As a consequence, it has not been
establish any empirical correlations that will permit
the estimation of Mach number effects for con-
ventional blade sections. The limited available
dat.a indicate, however, that a marked rise in loss
is eventually obtained as Mach number is in-
creased.
A typical example of the variation of total-
pressure-loss coefficient with inlet Mach liwnber
for a conventional cascade section at fixed inci-
dence angle in the region of minimum loss is
presented in figure 153(a). The inlet Mach
number at which the sharp rise in loss occurs is
referred to as the limiting Mach number. The
variation of the wake profile downstream of the
blade as Mach number is increased is shown in
figure 153(b) to illustrate the general deterioration
of the suction-surface flow. The flow deteriora- Percent blade -spacing, s
tion is the result of a separation of the suction-
surface boundary layer induced by shock-wave (a) Total-pressure-loss coefficient.
(b) Blade wake.
and bound ary-layer interactions.
Inview of the complex nature of the shock- FIGURE153.-Variation of cascade blade loss with inlet
Mach number for NACA 6 5 4 12AlO)10 blade in region of
wave development and its interaction effects, the minimum loss (ref. 122).
estimation of the variation of minimum total-
pressure loss with inlet Mach number for a given camber distribution on high Mach number per-
blade is currently impossible. At the moment, formance are discussed extensively in the litera-
this pursuit must be primarily an experimental ture (refs. 123, 200, and 201). Results indicate
one. Schlieren photographs showing the forma- that, for the range of blade shapes and Mach
tion of shocks in a cascade are presented in ref- numbers normally covered, camber distribution
erences 41, 205, and 206, and detailed discussions does not have a large effect on maximum Mach.
of shock formations and high-speed performance number performanee as obtained in the two-
of two-dimensional-cascade sections are treated dimensional cascade.
in references 41, 205, and 207 to 209. Cascade
SUMMABY
experience (refs. 40 and 205) and theory (refs. 41,
88, and 209) indicate that a location of the point From the foregoing correlations and considera-
of maximum thickness at about the 50-percent- tions, the low-speed loss in total pressure of con-
chord position and a thinning of the blade leading ventional two-dimensional-cascade sections can
and trailing edges are favorable for good high readily be estimated. If blade surface velocity
Mach number performance. The avoidance of a distributions are available, the suction-surface
throat area within the blade passage is also local difFusion factor Dlacis determined according
indicated in order to minimize the effects of flow to equation (265) and a value of 6*/c is then
choking. Discussions of the choking problem selected from figure 146. In the absence of blade
are presented in references 203 and 208, and blade surface velocity data, the diffusion factor D is
throat areas ar . for several blade shapes in computed from over-all conditions by means of
references 123 0 to 212. The effects of equation (54)and 6*/c is selected from Sgure 148.
208 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

With 6*/c determined, the total-pressure-loss constant, then variations of H and e* and of the
coefficient is computed according to equation (264) relation between e*/c and 0' with outlet free-stream
from the cascade geometry and a pertinent value Mach number can be established analytically to
of wake form factor H. illustrate the nature of the compressibility effects.
According to reference 156, for cascade meas- Curves of the variation of the ratios of com-
uring stations located more than about % chord pressible to incompressible form factor H/Hi,,
length downstream of the blade trailing edge, the and momentum thickness e*/eX, with outlet
value of H will generally be less than about 1.2. Mach number for various d values obtained from
For practical purposes, it was indicated that a numerical integration of the wake parameters
constant value of H of about 1.1 can be used over inTolved are shown in figures 154 and 155. Re-
a wide range of cascade configurations and in- cently, the increasing trend of H with M, was
cidence angles for measuring stations located be- substantiated experimentally at the NACA Lewis
tween 36 and 1%chord lengths behind the trailing laboratory in an investigation of the wake char-
edge. Loss coefficients based on inlet dynamic acteristic of a turbine nozzle (unpublished data).
head can then be determined, if desired, from Curves of the ratio of the integrated value of
equation (266). The estimation of losses based on 0' obtained from a given value of e*/c in a com-
the diffusion factor D can, for example, produce a pressible flow to the value of 0' computed from the
value of solidity that results in the least computed same value of e*/c according to the incompressible
loss coefficient for a given velocity diagram. relation of equation (264) are shown in figure 156.
The accuracy of the results obtained from the It should be noted that for compressible flow the
prediction procedure outlined is subject to the denominator in the loss-coefficient definition
limitations and approximations involved in the (eq. (263)) is now given by P-p.
diffusion analysis and wake momentum-thickness In summary, therefore, an accurate prediction
correlations. Strictly speaking, the procedure of the variation of reference total-pressure loss
gives essentially a band of probable loss values at with inlet Mach number for a given cascade
the cascade measuring station about % to 1%
chord lengths downstream of the blade trailing
edge for the reference-incidence-angle setting and
Reynolds numbers of about 2.5X106 and greater
at low speed (up to about 0.3 inlet Mach number).
It should also be noted at this point that the loss
values obtained in this manner represent the low-
speed profile loss of the cascade section. Such
loss values are not generally representative of the
losses of the section in a compressor blade row or in
a high-speed cascade.
A corresponding loss-estimation technique for Outlet Mach number, M2
high Mach number flow is currently unavailable
FIGURE154.-%tio of compressible to incompressible
because of the unknown magnitude of the com- form factor for constant value of exponent in power
pressibility effect on the wake momentum-thick- velocity distribution. db.
ness ratio of a given cascade geometry. Further-
more, both the wake form factor H and the relation
between e*/c and J (given for incompressible flow
by eq. (264)) vary with Mach number. For
example, if the velocity variation in each leg of the
wake is assumed to vary according to the power
relation

FIGURE
155.-Ratio of compressible to incompressible
momentum thickness for constant full thickness and
where 6 is the thickness of the wake and d is some exponent for power velocity distribution.
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 209
at zero incidence.

mum loss.

angle is zero a t zero camber angle. However,

analysis indicates that this is not the case for
Outlet Mach number, M2- blades of conventional thicknesses. A recent
Fxau~m 156.--Correction factor Kc for calculation of
theoretical demonstration of the existence of a
total-pressure-loss coefficient for compressible flow on positive value of zero-camber deviation angle ac-
basts of incompressible equation (264) as determined cording to potential-flow calculations is given by
from model wake form with power velocity profile. Schlichting in reference 193. The computed var-
iation of zero-camber deviation angle for a con-
blade is currently impossible. At the moment, ventional 10-percent-thick profile at zero incidence
this pursuit is primarily an experimentalone. Fam- angle as obtained in the reference is shown in
ilies of curves of wake momentum thickness and figure 158.
form factor against diffusion factor are required It will be recalled from the discussion of the
(with appropriate definitions for subsonic or zero-camber minimum-loss incidence angle that,
supersonic flow) as in figure 146 or 148 for a wide for the conventional staggered cascade (Oo<ro
range of inlet Mach number. Analytically, a <goo) with finite blade thickness set a t zero in-
simple compressible relation is needed between cidence angle, a greater magnitude of velocity
O*/c and Z as a function of Mach number.

DEVIATION-ANGLE ANALYSIS
PRELIMINARY mfirsrs
The correct determination of the outlet flow
direction of a cascade blade element presents a
problem, because the air is not discharged at the
angle of the blade mean lime at the trailing edge,
but at some angle 6" to it (fig. 124). Since the
flow deviation is an expression of the guidance
capacity of the passage formed by adjacent blades,
it is expected that the cascade geometry (camber,
thickness, solidity, and chord angle) will be the
principal influencing factor involved.
From cascade potential-flow theory (ref. 80,
e.g.), it is found that the deviation angle increases
with blade camber and chord angle and decreases
with solidity. Weinig in reference 80 shows that Franx~ 157.-Theoretical variation of deviation-angle
ratio for infinitely thin circular-arc sections at "impact-
the deviation angle varies linearly with camber for free-entry"
a given value of solidity and chord angle for theory of re

691464 0-65-1s
210 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

inlet angle. The variation of reference deviation

angle can then be expressed in equation form as

(268)
where 6: is the reference deviation angle for zero
camber, m is the slope of the deviation-angle
Blade-chord angle, y o , deg variation with camber ( S " - ~ ~ ) / Q , and (p is the
camber angle. As in the case of the analogous
FIGUBE 158.-Theoretical variation d deviation angle for
conventional uncambered 10-percenbthick blade section terms in the reference-incidence-angle relation
at zero incidence angle as presented in reference 193. (eq. (261)), 8; and m are functions of inlet-air
angle and solidity.
occurs on the blade lower (concave) surface than The influence of solidity on the magnitude of
on the upper (convex) surface (fig. 134). Such the slope term m could also be directly included
velocity distributions result in a negative blade as a functional relation in equation (268), so that
circulation and consequently (as in'dicated by the equation (268) could be expressed as
solid vectors in fig. 159) in a positive deviation
angle. Furthermore, since the deviation angle
increases slightly with increasing incidence angle
(dso/di is positive in potential cascade flow),
positive values of deviation aage will likewise be where ma,,represents the value of m (i.e., (6'-
obtained at the condition of minimum-loss inci- S:>/Q) at a solidity of 1, b is the solidity exponent
dence angle (as illustrated by the dashed vectors (variable with air inlet angle), and the other terms
in fig. 159). Since the zero-camber deviation are as before. It will be noted that equation
angle arises from essentially a thickness blockage (269) is similar in form to the frequently used
effect, the characteristics of the variation of deviation-angle rule for circular-arc blades origi-
minimum-loss zero-camber deviation angle with nally established by Constant in reference 186
cascade geometry would be expected roughly to and later modified by Carter in reference 88.
parallel the variation of the minimum-loss zero- Carter's rule for the condition of nominal inci-
camber incidence angle in figwe 137. The low- dence angle is given by
speed reference-deviation-angle correlations may,
therefore, involve intercept values as in the case
of the reference-incidence-angle correlations.
In addition to the cascade-geometry factors in which m, is a function of blade-chord angle.
mentioned, the low-speed deviation angles can Values of m, determined from theoretical con-
also be affected by Reynolds number, turbulence, siderations for circular-arc and parabolic-arc mean
and Mach number. The thickened surface bound- lines (ref. 88) are shown in figure 160. In the
ary layers resulting from low levels of Reynolds
number and turbulence tend to increase the devia-
tion angle. Variations in inlet Mach number can
affect the deviation angle of a fixed two-dimen-
sional-cascade geometry because of the associated '0
-Zero
changes in blade circulation, boundary-layer de- ---Minimum-loss
velopment, and outlet to inlet axial velocity ratio
(compressibility effect on pV,).
DATA COBBELATION8

Form of correlation.-Examination of deviation-

angle data at reference incidence angle reveals ,
/
- - 1 --- Axis
that the observed data can be satisfactorily rep-
resented by a linear variation of reference deviation IGURE
159.-Outlet flow direction for cascade of staggered
angle with camber angle for fixed solidity and air uncambered blades.
 EXPERIMENTAL FLOW IN TWO-DIMENSION& CASCADES 211
thickness ratio. Values of the intercept term 6;
and the slope term m were obtained by fitting a
straight line to each data plot of reference equiva-
lent deviation angle against equivalent camber
angle for a fixed solidity and air inlet angle. The
straight lines were selected so that both a satis-
factory representation of the variation of the data
points and a consistent variation for the resulting
10 20 30 40 50 60 6: and m values were obtained. The extrap-
Blade -chord angle, yo,deg olation of the values of m to &=O was guided
by the data for the 65-(12Alo)10 blade a t solidities
FIGUBE
16O.---Variation of faetor m, in Carter's deviation-
angle rule (ref. 88).
of and 1.5 reported in the cascade ,-,uidevane
investigation of reference 213 (for an aspect ratio
ensuing correlations, both forms of the deviation- of 1, as in ref. 39).
angle relation (eqs. (268) and (269)) are used, For the deviation-angle rule as given by equa-
since each has a particular advantage. Equation tion (269), deduced values of ms=l and exponent
(268), with m plotted as a function of #?, and u, b as functions of inlet-air angle are presented in
is easier to use for prediction, especially if the figures 163 and 164. The deduced rule values
calculation of a required camber angle is involved. (eq. (268) or (269)) and the observed data points
Equation (269) may be better for extrapolation are compared in figure 165 to indicate the effective-
and for comparison with Carter's rule. ness of the deduced representations. The flagged
As in the case for the zero-camber reference symbols in the high-camber range in the figure
minimum-loss incidence angle, the zero-camber represent blade configurations for which boundary-
deviation angle can be represented as a function layer separation is indicated ( D greater than about
of blade thickness as 0.62). In view of the higher loss levels for this
condition, an increase in the magnitude of the
6:= (Ka)a(Kd)t(8:)lo (271) deviation angle is to be expected compared with
the values extrapolated from the smaller cambers
where (Sz)lo represents the basic variation for the for which a lower loss level existed.
10-percent-thick 65-series thickness distribution, C-Series circular-arc blades.-In view of the
(Kdsnrepresents any correction necessary for a absence of systematic cascade data for the C-
blade shape with a thickness distribution different circular-arc blade, an accurate determination
from that of the 65-series blade, and (Kd repre- of the rule constants cannot be made for this blade
sents any correction necessary for maximum blade shape. However, a preliminary relation can be
thicknesses other than 10 percent. (For 8 10- deduced on the basis of limited data. It appears
percent-thick 65-series blade, (Kdt and ( K a > s n are that, for the uncambered C.4 section (ref. 192), if
equal to 1.1 The Problem, therefore, is reduced a value of (&)rb equal to 1.1 (as for the determi-
to finding the ValUes of m, b, and 6: (though eq. nation of &) is used, a satisfactory comparison
(271)) as fUIlCtiOnS Of the pertinent variables in- between predicted and observed 6; values is
volved for the various blade shapes considered. obtained.
NACA 65- (Alo)-.series blades.-From an exam- The characteristic number mu-l in the devia-
ination of the plots of equivalent deviation angle tion-angle design rule of equation (269) for a
against equivalent camber angle a t reference given blade mean line corresponds to the value of
minimum-loss incidence angle obtained from the (a0-66o,)/(p a t a solidity of unity. Cascade data
cascade data, values of zero-camber deviation for a C.4 circular-arc profile obtained from tunnels
angle can be determined by extrapolation. The with good boundary-layer control are presented
deduced plots of zero-camber deviation angle in references 167 (pt. I) and 199 for a solidity of
and slope term m as functions of solidity 1.0 for &=30°, 42.5O, 45O, and 60°. Values of
and air-inlet angle are presented in figures 161 and (60-660,)/(p were computed for these blades accord-
162 for these blades. The subscript 10 indicates ing to the s", variations of figure 161. A value of
that the values are for 10-percent maximum- mu-lfor &=Oo was obtained from the per-
212 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

Inlet-air angle, p, , deg

FIGUEE161.--Zero-camber deviation angle at reference minimum-loss incidence angle deduced from low-speed-crtscade
data for 10-percent-thick NACA 65-(A1,+seriea blades (ref. 39).

formance data of a free-stream circular-arc inlet variation based on values of b and mn=l obtained
guide vane presented in reference 214. These from figures 164 and 166, respectively, is shown
values of m are plotted in figure 166 against inlet- in the figure by the solid line. A satisfactory
air angle, and the proposed variation of mopl agreement with these circular-arc data is thus
for the circular-arc mean line is shown by the demonstrated for the vdue of b obtained from the
solid line. 65-series data. On the basis of these results,
In the absence of data covering a range of deduced curves of m against B1 for a range of
solidities, it is assumed that the solidity exponent solidities (for use in conjunction with eq. (268))
b in the deviation-angle rule of equation (269) is were computed for the C-series circular-arc blade
independent of the profle shape and will therefore as indicated in figure 168.
also be applicable for the circular-arc mean line. Double-circular-arc blades.-Although limited
This assumption agrees with limited experimental data are available for the double-circular-arc blade
data. The variation of ratio of deviation angle to (refs. 40 and 197), it was felt that these data could
camber angle obtained from constant-thickness not be reliably utilized in the construction of a
circular-arc guidevane sections of reference 2 15 deviation-angle rule because of the questionable
(6z=Oo for guide vanes) over a wide range of two-dimensionality of the respective test tunnels.
solidities is shown in figure 167. A computed However, since the Cseries and the double-
 EXPERIMENTAL FllrOW IN CASCADES 213

F
L
0
c
U
.&-
0

Inlet-air angle, p, ,deg

FIGURE162.-Deduced variation of slope factor m in deviation-angle rule for NACA 65-(A+~ieriesblades aa equivalent
circular arc.

circular-arc blades differ only in thickness distribu- predicting the deviation angle of circular-arc-
tion, it is reasonable to expect that, as in the case mean-line blades, some results obtained from the
of the reference-incidence-angle correlations, only use of Carter’s rule were compared with the de-
the zero-camber deviation angles will be materially duced rule of equation (269) with figures 161, 164,
affected. Therefore, the slope-term value m and 166. The principal difference between the
deduced for the C-series circular-arc blade (fig. two rules occurs in the blade orientation parameter
168) might also be used for the double-circular-arc used for the m variation and in the 6; and b varia-
blade, but the 6; v tions. The value of the solidity exponent of % in
arbitrarily selected tion (270) was originally obtained lim-
a. Carter, in a later work pro-
cates
circular-arc blade. values close to 1for accelerating cascades and close
Comparison of rules.-In view of the widespread to X for decelerating cascades. The variation of
use of Carter’s rule (eq. (270) with fig. 160) for b obtained from the NACA 65-(Alo)-series blades
214 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

Inlet-air angle, PI , deg

FIGUBZ163.-Value of m,,=l in deviation-angle rule for 65-(Ajo)-series blades aa equivalent circular arc (deduced from
data of ref. 39).
I

c
c
a,
c
Q
0
x
W

Inlet-air angle, P , , deg

FIGURE
164.-Value of solidity exponent b in deviation-angle rule (eq. (269)) (deduced from data for 65-(Alo)-series blades
in ref. 39).
 EXPERIMENTAL FLOW IN TWO-DIMENSION& CASCADES 215

Equivolent

(a) Solidity, 0.5 and 0.75. (b) Solidity, 1.0.

(c) Solidity, 1.25. (d) Solidity, 1.5.
FIGURE165.-Comparison between data values and deduced rule values of reference minimum-lo= deviation angle for
NACA 65-(A3lO-i3eries blades as equivalent circular arc (data from ref. 39).

as equivalent ckwlar arcs in figure 164 essentially Blade-chord angle for Carter’s rule was computed
confirms this trend. Actually, the deviation-angle from the equation
rule in the form of equation (269) constitutes a
y0=p1-2-- * Q
modification of Carter’s rule. 2 (272)
In addition to the basic differences between the
rules in the magnitudes of the m, b, and 6; values, Reference incidence angle was determined from
it is noted that Carter’s rule was originally devel- equations (261) and (262) and figures 137 and 138.
oped for the condition of nominal incidence angle, The plots of figure 169 show that, in practically
whereas the modified rule pertains to the reference all cases, the deviation angles given by the modi-
minimum-loss incidence angle. However, since fied rule are somewhat greater in magnitude than
Carter’s rule has frequently been used over a wide those predicted by Carter’s rule for the 10-percent-
range of reference angle in its application, both thick blade. This is particularly true for the high
rules were evaluated, for simplicity, for the refer- inlet-air angles. Thus, greater camber angles are
ence minimum-loss incidence angle. required for a given turning angle amrding to
An illustrative comparison of predicted reference the modified rule. Differences are even less for
deviation angle as obtained from CarterIs rule and the double-circular, arc blade, as indicated in figure
the modified rule for a -lO-percent-thick, thick- 170, since the 60,values are smaller for these blades.
nosed circular-arc blade is shown by the calculated However, it should be kept in mind that the
results in figure 169 for ranges of camber angle, magnitude of the factors in the modified rule are
solidity, and inlet-air angle. Deviation angles in proposed values based on limited data. Further
figure 169 were restricted to cascade configurations research is required to establish the modified rule
producing values of diifusion factor less than 0.6. on a firmer foundation.
216 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

Inlet-air angle, pi , d e g

FIGURE166.-Deduced values of m,-l for circular-arc mean line.

Effect of blade maximum thickness.-Available .4

data on the variation of reference deviation angle
9.
with blade maximum-thickness ratio obtained 6
co
from cascade investigations of the 65-(12AI0) .3
blade of reference 202 are shown in figure 171. g
z
c

The solid symbols representing the values of

deviation angle at zero thickness were determined 0,
6 .2
by subtracting the values of (60,),, obtained from c
0
figure 161 from the measured value of deviation c
I

angle at 10-percent maximum thickness obtained .-

+
0
.I
from the data in figure 171. A very reasonable >
a,
variation with thickness ratio, as indicated by the n
faired curves, is thus obtained for all three con-
figurations. The increasing slope of the deviation- 0 .8 1.6 2.4 3.2 L 3
angle variation with increasing thickness ratio is Solidity, CT
believed due to some extent to the accompanying
FIGURE 167.-Comparison of experimental deviation-
increase in wake losses. angle ratio and rule values using solidity exponent given
Preliminary values of a correction factor for by figure 164. Data for cmular-arc inlet guide vanes
maximum-thickness ratio (IQ1 deduced from the in annular cascade (ref. 215).
data of figure 171 are shown in figure 172. In the
absence of further data, it is proposed that this the variation of the loss. Therefore, factors
correction curve is also applicable to other con- involved in the deviation-angle variation are the
ventional blade shapes. same as those for the loss behavior. Correspond-
Effect of Reynolds number.-In view of the ingly, no Reynolds number correction factors that
large rise in loss as blade-chord Reynolds number will be applicable for all blade configurations have
is reduced (fig. 152), a corresponding rise in devia- been established. The deduced deviation-angle
tion angle (or decrease in turning angle) is to be rule developed herein is applicable at Reynolds
expected. Experimental confirmation of the numbers of about 2.5X lo6 and greater.
marked effect of Reynolds number on blade deviation Effect of inlet Mach number.-Experimental
angle at fixed incidence angle is illustrated in variations of minimum-loss deviation angle with
figure 173 for several compressor blade shapes. inlet Mach number are presented in figure 175 for
The variation of deviation angle with Reynolds two circular-arc blades. Further cascade data in
number over a range of incidence angle is demon- terms of air-turning angle at fixed angle of attack
strated in figure 174. In all cases the variation are shown in figure 176 for two other compressor
of the deviation or turning angle closely parallels blade shapes. (Since the data in fig. 176 were
 EXPERIMENTAL -D 217

inlet -air angle I p, ,deg

F~GURE
168.-Deduced variation of slope factor m in deviation-angle rule (eq. (268)) for circular-arc-mean-line blades.

obtained at constant angle of attack, the varia- angle in the data is always associated with the
tion of turning angle is an inverse reflection of sharp rise in loss.)
the variation of deviation angle.) The data of Variation with incidence angle.-Thus far, of
figures 175 and 176 indicate that deviation angle necessity, the analysis has been conducted for
varies little with inlet Mach number up t o the flow conditio
limiting value. As indicated in the Preliminary the general
Analysis section, the resultant Mach number Ultimately, of course, it is desired to predict flow

on the relative

Large increases in
however, when the
21 8 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS
12

m
0)
U

0
60
- 0

Comber angle, ?,de9

Comber angle, ?,deg (a) Solidity, 1.5; inlet-air angle, 40°.

(b) Solidity, 1.0; inlet-air angle, 60".
(a) Solidity, 1.5. FIQURE 17O.-Comparison of calculated reference devia-
(b) Solidity, 1.0. tion angles according to Carter's rule and deduced
modified rule for circular-arc blades of different thickness.
FIGURE169.-Comparison of calculated reference devia-
tion angles according to Carter's rule and deduced
modified rule for 10-percent-thick, thick-nose circular-arc
blades.
I
is currently available on the effect of losses, at- m
0
tention is centered on deviation-angle variations
in the region of low loss, where the trend of I
M
variation approaches that of the potential flow. -acI
Examination of .potential-flow theory (Weinig, 0
0
ref. 80, e.g.) shows that a positive slope of devia- t
0
0
'F
tion angle against incidence angle exists (i.e., .-
deviation angle increases with incidence angle). n
s
Calculations based on the theory of Weinig reveal
that the magnitude of the slope varies with
solidity and blade-chord angle. The deviation- Moxirnum-thickness rotio, f IC
angle slope approaches zero for infinite solidity
FIGURE171.-Variation of deviation angle with blade
(deviation angle is essentially constant at high maximum-thickness ratio for NACA 6 5 412Al0) blade
solidity) and increases as solidity is reduced. At in region of minimum loss (data from ref. 202).
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 219

-.
5."
-
L
0
V
0
'c

c
.-
0
L
V
0)
L
L

0
V

0 .02 .04 .06 .08 .IO .I2

Moximurn- thjckness ro tio, t / c

FIGURE172.-Deduced maximum-thicknesscorrection for zero-camber reference minimum-loss deviation angle (eq. (271)).
220 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS
8

FIGUBE175.-Variation of reference deviation angle with

inlet Mach number for circular-arc blades. Solidity,
1.333; blade-chord angle, 42.5' (ref. 40).

Blade-chord Reynolds number, Re,

(a) lOC4/25C50 blade. Solidity, 1.33; blade-chord angle,

42.5' (ref. 40).
(b) 10.5 2A/25C50 blade. Solidity, 1.33; blade-chord
angle, 42.5' (ref. 40).
(c) NACA 65-(12) 10 blade. Solidity, 1.5; inlet-air angle,
45' (ref. 39).
FIGURE 173.-Illustrative variations of reference deviation
angle with Reynolds number.

Inlet Mach number, MI

(a) T1(18&La)08 blade. Solidity, 1.6 (ref. 207).

(b) 65-(12Alo)10 blade. Solidity, 1.0; inlet-air angle, 45';
angle of attack, 16.5' (ref. 122).
FIQUEE176.-Vsriation of air-turning angle with inlet
Mach number in region of minimum loss.

(high solidity and low blade angle), the less

sensitive the deviation angle is to changes in
incidence angle.
For analysis purposes, since the region of low
Blade-chord Reynolds number, Re,
loss is generally small, the variation of deviation
angle with incidence angle for a given cascade
F I Q U R174.-Variation
~ of deviation angle with Reynolds geometry in the region of minimum loss can be
number for 10C4/40 P40 blade. Solidity, 1.33 (ref.
183). represented as

constant solidity, the slope of deviation angle

against incidence angle increases as the chord
angle is increased. These trends indicate physi- where (dSo/di),,, represents the slope of the
cally that the greater the initial guidance effect deviation-angle variation at the reference incidence
 EXPERIMENTAL 0-DIMENS10 CADE 221
I

al
C
U
al
'p
U
c
V
Q)

C
al
L
a,
L
a,
L

Solidity, u

FIGURE177.-Deviation-angle slope dP/di at reference incidence angle deduced from low-speed data for NACA SS-(A10) 10
blades (ref. 39).

angle. An empirical determination of the magni- angle with cascade geometry at low speed can
tude of the slope of the variation of deviation be satisfactorily established in terms of an inter-
angle with incidence angle was obtained from cept value 6: and a slope value m as given by
an analysis of the low-speed experimental data equation (268). The experimental data could
for the 65-(Alo)10 blades of reference 39. From also be expressed in terms of a rule similar in form
the plot of deviation angle against incidence to Carter's rule, as indicated by equation (269).
angle for each configuration (as in fig. 127, e.g.), Deduced values of and m were obtained as a
the slope of the curve at the minimum-loss inci- function of and u from the data for the 10-
dence angle was evaluated graphically. The percent-thick 65-(Alo)-series blades of reference
deduced variation of reference slope magnitude 39 as equivalent circular arc (figs. 161 and 162).
d6"ldi obtained from fairings of these values is Rules for predicting the reference devi
presented in figure 177 as a function of solidity of the C-series and double-circul
and inlet-air angl Qualitative agreement with were also deduced based on the correlations for
theory is strong1 dicated by the data. Since the 65-(Alo)-seriesblades and on limited data for
the phenomenon is essentially a guidance or the circular-arc blade (figs. 161 and 168).
channel effect, it is anticipated that the slope The procedure involved in estimating the

the low-loss range of o

equation (273) and figure 177. blades. The value of (Ka)i for the blade
SUMMARY maximum-thickness ratio is obtained from figure
The analysis of blade-section deviation angle 172, and the approximate value of (Kd, is
shows that the variation of reference deviation selected for the type of thickness distribution.
222 AERODYNAMIC! DESIGN OF AXIAL-FLOW COMPRESSORS

For the 65-series blades, (Ka),n=l.O, and it is

proposed that be taken as 1.1 for the C-
series blades and as 0.7 for the double-circular-arc
blade. The Val

compressor design.
The camber angle required to produce a given a summary of the
turning angle at the reference research with re
speed can readily be calculated b
preceding incidence-angle and deviation-angle
correlations when the inlet-air angle and blade
solidity are known. From equations (57), (261), and the referenceincidence and deviation angles in
and (268), the camber angle as a function of the satisfactory agreement with existing cascade data.
turning, deviation, and incidence angle is The rules may also be of help in reducing the
necessary experimental effort in the accumulation
AS-(io-62) of further cascade data.
l-m+n (274)
However, the present analysis is incomplete.
Many areas, such as the deviation-angle rule for
or, in terms of the thickness corrections (eqs. the double-circular-arc blade, require further data
(262) and (271)), to substantiate the correlations. Furthermore,
additional information concerning the influence of
high Mach number and off-designincidence angles
of cascade performance is needed.
Finally, it is recognized that the performance of
For simplicity, since (Kf>*,,=(K&h=K6n, equa- a given blade geometry in the compressor con-
tion (275) can be expressed in the form figuration will differ from the performance es-
tablished in the two-dimensional cascade. These
differences result from the effects of the various
three-dimensional phenomena that occur in com-
where represents some correction factor for pressor blade rows. It is believed, however, that
blade thickness, such that a firm.foundation in two-dimensional-cascade flow
constitutes an important step toward the complete
understanding of the compressor flow. The extent
to which cascade-flow performance can be suc-
Curves of the values of (&So- as a function of cessfully utilized in compressor design can only be
and u are given in figure 178; curves of the values established from further comparative evaluations.
of 1-m+n as a function of Dl and u are given in Such comparisons between observed compressor
figure 179(a) for the 65-(Alo)-series mean line and performance and predicted two-dimensional-cas-
in figure 179(b) for the circular-arc mean line; cade performance on the basis of the rules derived
and values of Et are plotted as a function of herein are presented in chapter VII.
EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 223
224 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

C
+
F
-I

Inlet-air angle, ,B, , deg

(a) NACA 65-(A&series blades aa equivalent circular arc (ea. (276))

FIGURE179.-Variation of 1-m+n.
 EXPERIMENTAL FLOW IN TWO-DIMENSIONAL CASCADES 225

C
+
F
-I

Inlet-oir ongle, /3, ,deg

(b) Circular-am-mean-line blades (eq. (276))

Iifanm 179.-Concluded. Variation of 1 -m+n.

691-564 0-65-16
226 AERODYNAMIC DESIGN OF A x w r F L O W COMPRESSORS

I.

Maximum-thickness ratio, f/c

WGUBE18O.-Variation
-
of thicknewcorrectionfactor Rt for camber calculation (eq. (276)).
 CHAPTER VZZ -$Lbd
BLADE-ELEMENT FLOW I N ANNULAR CASCADES
H. ROBBINS,
By WILLIAM J. JACKSON,
ROBERT LIEBLEIN
and SEYMOUR

A- blade-element analysis is made of anndar- compressor design procedure is given in chapter

cascade data obtained primarily from single-stage- VIII. Secondly, proper blade sections are selected
compressor test installations. The parameters that at each radial position and stacked in proper
describe blade-element $ o w (total-pressure loss, relation to each other to establish the design
inqidence angle, and deuiutwn angle) are discussed velocity diagrams at each radius. In order to
with reference to the many vurhbles ageding these satisfy the design requirements successfully, accu-
parameters. The blade-element data are correlated rate blade-row design data are needed. Success-
over a fairly wide range of inlet Mach number and ful analysis of a compressor (the analysis problem)
cascade geometry. Two blade shapes are considered also depends upon accurate blade-row data, not
in detail, the 66-(Al0)-seriesprofile and the double- only at the design point but also over a wide
circular-arc airfoil. Compressor data at three range of flow conditions (ch. X).
radial positions near the tip, mean, and hub are In general, compressor designers have relied
correlated at minimum-loss incidence angle. Curves primarily on three sources of blading information:
of loss, incidence angle, and dewiation angle are (1) theoretical (potential-flow) solutions of the
presented for rotor and stator blade elements. These flow past airfoil cascades, (2) low-speed two-
correlation curves are presented in such a manner dimensional-cascde data, and (3) threedimen-
that they are directly related to the low-speed two- sional annular-cascade data. Potential-flow solu-
dimensional-cascade results. A s f a r rn possible, tions have been used to a limited extent. In
physical explanations of the $ow phenomena are order to handle the complex mathematics in-
presented. In adddim, a edcdation procedure volved in the theoretical solutions, it is necessary
is given to iuustrate how the correlathn cukes could A to make simplifying assumptions concerning the
be utilized in compressor design.- flow field. Among the most important of these
is the assumption of a two-dimensional flow
INTRODUCTION
field with no losses. Unfortunately, in some cases
Axial-flow-compressor research has generally been these assumptions lead to invalid results unless
directed toward the solution of either compressor experimental correction factors are applied to
design or compressor analysis problems. In the the computed results. These solutions are re-
design problem, the compressor-inlet and -outlet viewed in chapter IV.
conditions are given, and the compressor geometry A considerable amount of blade design data
must be determined to satisfy these conditions. has been obtained from low Mach number
In contrast, for the analysis problem the inlet experimental two-dimensional cascades. A rather
conditions and compressor are specified, and the complete study of the cascade work that has
outlet conditions are desired. (The analysis been done to date is presented in chapter VI,
problem is sometimes referred to as the “direct which correlates cascade data at minimum-loss
compressor problem.”) incidence angle for a wide range of inlet conditions
There are two phases of the axial-flow-com- and blade loadings. Low-speed twodimensional-
pressor design problem. In the first phase it is cascade data have been applied successfully in
necessary to prescribe desirable velocity distri- many cornpressor designs. However, with the
butions at each radius of the compressor that will design trends toward higher Mach numbers
ultimately fulfill the design requirements. A and higher blade loadings, these cascade results
discussion of the velocity-diagram phase of the have not always been completely adequate for
227
228 . k k R O D 3 W M C DESIGN OF AXIAL-FLOW COMPRESSORS

describing the compressor flow conditions, par- Ka correction factor in deviation-angle relation,
ticularly in regions of the compressor where function of blade maximum-thickness
threedimensional-flow effects predominate. ' ratio and thickness distribution .
Because of such effects, it becomes essential M Mach number
that blade-element data be obtained in a three- m factor in deviation-angle relation at u= 1
dimensional-compressor environment. These (ea. (280)), function of inlet-air angle
threedimensional-cascade data (obtained pri- m, factor in deviation-angle relation (eq. (282)),
marily from single-stage compressors) may then function of blade-chord angle.
be used to supplement and correct the theoretical n slope factor in incidence-angle relation (eq.
solutions and the two-dimensional-cascade in- 279)), function of inlet-air angle and solidity
formation. Some success has been obtained in P total or stagnation pressure, lb/sq f t
correlating annular-cascade data with the theory P static or stream pressure, lb/sq f t
and the twodimensid-cascade results (refs. T radius
32, 214, and 218 to 220); however, the range s blade spacing, in.
of variables covered in these investigations is T total or stagnation temperature
not nearly complete. t blade maximum thickness, in.
The purpose of this chapter is to correlate and V air velocity, ft/sec
summarize the available compressor data on a B air angle, angle between air velocity and
blade-element basis for comparison with the axial direction, deg
two-dimensional-cascade data of chapter VI. An AB air-turning angle, pl-pz, deg
attempt is made to indicate the regions of a Y ratio of specific heats
compressor where low-speed two-dimensional- YO blade-chord angle, angle between blade
cascade data can be applied to compressors and chord and axial direction, deg
also to indicate the regions where cascade results 6O deviation angle, angle between outlet-air
must be modified for successful application to direction and tangent to blade mean
compressor design. Two blade sections are con- camber line at trailing edge, deg
sidered in detail, the NACA 65- (A,,)-series blade 1 efficiency
and the double-circular-arc airfoil section. Par- K blade angle, angle between tangent to blade
ticular emphasis is placed on obtaining incidence- mean camber line and axial direction, deg
angle, deviation-angle, and loss correlations a t Q solidity, ratio of chord to spacing
minimum loss for blade elements near the hub, (P blade camber angle, difference between
mean, and tip radii of both rotor and stator blades. blade angles at leading and trailing edges,
Empirical correction factors that can be applied K i - K a , deg
to the two-dimensional-cascade design rules are
given, and application of the design rules and -
W
0
angular velocity of rotor, radians/sec
total-pressure-loss coefficient
correction factors to compresor design is illustrated.
Subscripts:
SYMBOLS
ad adiabatic
The following symbols are used in this chapter: C compressor
a, speed of sound based on stagnation condi- GrV inlet guide vanes
tions, ft/sec h hub
b exponent in deviation-angle relation (eq. id ideal
(280)), function of inlet-air angle m mean
G chord length, in. min minimum
D diffusion factor o zerocamber
i incidence angle, angle between inlet-air R rotor
direction and tangent to blade mean S stator
camber line a t leading edge, deg ST stage
Ki correction factor in incidence-angle relation, t tip
function of blade maximum-thickness z axial direction
ratio and thickness distribution e tangential direction
 BLADE-ELEE4ENT FLOW IN ANNULAR CASCADES 229
1 station a t inlet to blade row or stage design velocity diagram of the blade row. The
2 station a t exit of blade row or stage basic parameters defining the flow about a blade
2-0 low-speed two-dimensional cascade element are indicated in figure 182. Stated
10 blade maximum-thickness-to-chord ratio of simply, blade-element flow is des
10 percent variations of the loss in total press
blade row and of the air-turning
Superscript: incidence angle (or angle of attack).
relative to rotor FACTORS AFFECTING BLADE-ELEMENT PERFORMANCE
PRELIMINARY CONSIDERATIONS
The flow about a given blade element in a
BLADE-ELEMENT CONCEPT compressor configuration is different from that
In current design practice, the flow distribution in a two-dimensional cascade because of three-
at the outlet of compressor blade rows is deter- dimensional effects in compressor blade rows.
mined from the flow characteristics of the individ- These three-dimensional effects influence th
ual blade sections or elements. The blade- magnitude of the design incidence angle, the 10s
element approach to compressor design is discussed in total pressure, and the deviation angle.
in detail in chapter I11 and in reference 221. To Incidence angle.-In the low-speed two-dimen-
review briefly, axial-flow-compressor blades are sional cascade, the minimum-loss incidence angle
evolved from a process of radial stacking of depends on the blade geometry (camber, solidity,
individual airfoil shapes called blade elements. and blade thickness), the inlet-air angle, and inlet
The blade elements are assumed to be along sur- Mach number (ch. VI). In compressor opera-
faces of revolution generated by rotating a stream- tion, several additional factors can alter the
line about the compressor arris; this stream surface minimum-loss incidence angle for a given element
of revolution may be approximated by an equiv- geometry-for example, differences in testing
alent cone (fig. 181). Each element along the procedure. I n compressor operation, incidence
height of the blade is designed to direct the flow angle, inlet-air angle, and inlet Mach number
of air in a certain direction as required by the vary simultaneously; in contrast, cascades are

Cornpre ssor
blade elements3

/
FIGURE18l.-Compressor blade elements shown along conical surface of revolution about compressor axis.
230 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

motions and disturbances generated by the casing

wall boundary layers, from blade tip clearance,
from radial gradients of total energy, and from
interactions of adjacent blade rows. The com-
pressor loss picture is further complicated by the
tendency of boundary-layer fluid on the com-
blade surfaces and in the blade wake to be
radially. As a consequence of this
phenomenon, the loss measured do
given blade element may not nec
the actual loss generated at that element, but
something more or less, depending on the radial
location of the element.
It is expected, therefore, that the factors
influencing the magnitude of the blade-element
loss in the compressor will include the factors
affecting the profile loss (blade surface velocity
distribution, inlet Mach number, blade-chord
Reynolds number, free-stream turbulence, and
blade surface finish) and the factors affecting the
three-dimensional losses. Investigations of com-
pressor blade-element losses based on surface
FIGURE182.-Rotor blade-element properties. velocity distribution, as expressed in terms of
diffusion factors, are presented in references 9
often operated with fixed inlet-air angle and inlet and 35. The essentially secondary effects of
Mach number. Some net difference may be ob- blade surface finish and trailing-edge thickness
tained in the range characteristics and therefore on compressor loss are investigated in references
in the location of the point of minimum loss be- 223 and 224. Results of tests of blade-element
tween cascade operation at constant inlet-air performance (ref. 225 and ch. V) and over-all
angle and compressor test operation (with varying performance (refs. 226 and 227) at varying
inlet-a~angle). Reynolds numbers indicate that there is no
In addition to these blade-element considera- signifkant variation in loss for Reynolds numbers
tions, of course, there are sources of difference above approximately 2.5X105. (Since most of
arising from compressor three-dimensional effects. the compressor data used in this analysis are
For example, radial variations of minimum-loss for Reynolds numbers greater than 2.5X106, no
incidence angle that are not consistent with the Reynolds number effects are believed to exist
trends predicted from cascade blade-element for the data.) Some variations of compressor
considerations have been observed in compressor loss with inlet Mach number have been estab-
rotors (refs. 56 and 222). Apparently, radial lished in references 52, 56, and 228. These
position may also be a factor in determining results, however, are not complete indications
compressor minimum-loss incidence angle. of Mach number effects (shock losses), since
Total-pressure loss.-In the two-dimensional the corresponding variations of blade diffusion
cascade, the magnitude of the loss in total pres- with Mach number are not identified. An
sure across the blade element is determined from attempt to separate the variation of diffusion
the growth of the blade surface boundary layers and shock losses with Mach number by means
(profile loss). In the actual compreasor, the loss of an analysis based on the diffusion factor of
in total pressure is determined not only by the reference 9 is presented in references 222 and 229.
profile loss, but also by the losses induced by the Although some aspects of the compressor
three-dimensional nature of the flow. These three-dimensional-flow phenomena are known
threedimensional losses result from secondary (chs. XIV and XV), the specific factors or
 BLADE-ELEMENT FLOW IN ANNULAR CASCADES 231
parameter affecting compressor three-dimensional and Reynolds number are small for the range of
losses have not been established for analysis data considered in this survey.
purposes. At present, the three-dimensional loss CORRELATION APPROACH
can be treated only on a gross basis as a difference
between the total measured loss and the In this chapter, annular-cascade data are com-
profile loss. pared with the two-dimensional-cascade correla-
tions of minimum-loss incidence angle, total-pres-
Deviation angle.-In the two-dmensional cas- sure loss, and deviation angle of chapter VI. In
cade the minimum-loss deviation angle. varies this way, compressor investigations serve as both
primarily with the blade geometry and the inlet- a verification and an extension of the two-dimen-
air angle. Experience with compressor operation
sional-cascade data. Two-dimensional-cascade
indicates that blade-element minimum-loss devia- data correlations and rules, in conjunction with
tion angle is also sensitive to three-dimensional correction factors deduced from the three-dimen-
effects. The two principal compressor effects sional data, can then be used for compressor design
are secondary flows and changes in axial velocity and analysis.
across the blade element. Secondary flows are
With this approach in mind, all available single-
treated in chapter X V and in reference 43. stage data were collected, computed, and plotted
Corrections are established in reference 43 for in a form considered convenient for correlation.
the effect of secondary flows on the outlet angles The blade and performance parameters used in
of compressor inlet guide vanes. At present,
the analysis are similar to those used in the two-
however, rotor and stator secondary-flow effects
dimensional-cascade correlations of chapter VI.
can be treated only on a gross basis.
Camber angle, incidence angle, and deviation angle
The effects of changes in axial velocity ratio on (fig. 182) are used to define the blade camber, air
the turning angles of a k e d blade-element geom- approach, and air leaving directions, respectively.
etry are conclusively demonstrated in the rotor These angles are based on tangents to blade mean
investigations of reference 218. There are several camber line at the leading and trailing edges. As
origins of varying axial velocity ratio across a in chapter VI, the NACA 65-(Alo)-series blades
compressor blade element: (1) contraction of the are considered in terms of the equivalent circular-
annulus area across the blade row, (2) compres- arc camber line (figs. 125 and 126, ch. VI).
sibility, which varies axial velocity ratio for a Loss in total pressure across the blade element
fixed annulus area, and (3) differences in the is expressed in terms of the loss parameter J cos
radial gradient of axial velocity at blade-row inlet where the relative total-pressure-loss coeffi-
and outlet, which can arise from the effects of cient i;' is defined as the mass-averaged defect in
radial-pressure equilibrium (ch. VIII). Although relative total pressure divided by the pressure
several attempts have been made to establish equivalent of the inlet velocity head:
corrections for the effect of change in axial veloc-
ity ratio on deviation angle (refs. 218 and 191),
these proposed correction techniques have not
been universally successful. The principd diffi-
culty involved in the axial velocity corrections is For stationary blade rows, or no change in stream-
the inability to' determine the corresponding line radius across the rotor, the numerator of
changes in blade circulation (i.e., tangential ve- equation (58) becomes the decrease in relative
locity). Values of axial velocity ratio were identified total pressure across the blade row from inlet to
for the deviation-angle data presented, although no outlet. The relative total-pressure-loss coeEcient
attempt was made to apply any corrections. was computed from stationary measurements of
total pressure and total temperature and from the
Some of the secondary factors influencing de- computed relative inlet Mach number according
viation angle, such as inlet Mach number and to reference 9. The total-pressure-loss parameter
Reynolds number, are investigated in references 7 cos p:/2u, as indicated in chapter VI, can be
52, 56, and 218. These results indicate that the used as a signifmint parameter for correlating
variations of deviation angle with Mach number blade losses.
232 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSOBS

i3

~ Incidt
e angle, i, deg

FIGUBE183.-Example of typical variation of blade-element performance parameters with incidence angle. Transonic
rotor with double-circular-arc blade sections at tip speed of 800 feet per second; data for blade row 17 (table 11) at
tip position (ref. 55).

The diffusion factor, which is used as a blade- 183. The data represent the variations of the
loading parameter, is defined in reference 9 for no flow at fixed rotational speed. Plots for stator
change in radius as follows: blade rows show similar trends of variation. As
in chapter VI, a reference point was established
D=(l-g)+ vi, -vi*
1 a as the incidence angle for minimum loss (fig.
2av; (278) 184(a)), and the bladeelement flow was analyzed
at this reference point. I n cases where minimum-
A typical example of the plotted performance loss incidence was not clearly defined, the refer-
parameters for a rotor blade row is shown in figure ence point was taken as the mean incidence of the
 BLADE-ELEMENT FLOW IN ANNUL& CASCADES 233
incidence-angle range for which values of 5 at the
end points are twice the minimum value (fig.
184(b)). In some instances, near the compressor
tip the loss-against-incidence-angle curve increased
continuously from a minimum value of loss param-
eter at the open-throttle point. In presenting data
for these cases several points near the minimum-
loss value are plotted.
One of the primary objectives of thir, analysis
is to determine Werences in blade-element per-
formance with compressor radial position. There-
fore, three radial positions along the blade span
(near the hub, mean radius, and tip) of each blade
row are considered. The radial positions at the
hub and tip are approximately 10 to 15 percent
of the passage height away from the inner and
outer walls, respectively, which are outside the z3
In
wall boundary-layer region in all cases. The anal-
ysis is directed toward correlating the loss and %
-0
deviation-angle data at reference incidence angle c
0
and determining the variation of reference inci- I-
dence angle with blade geometry and Mach num-
ber at the three radial positions.
EXPERIMENTAL DATA SOURCE9

There are three sources of three-dimensional- t

Gin
‘-Reference
incidence
cascade blade-element data : stationary annular- (b) I engle ,
cascade tunnel investigations, multistage-compres- Incidence angle, i ,deg
sor investigations, and single-stage or single-blade-
row compressor investigations. A relatively small (a) Minimum loa.
amount of data has been accumulated from blade- (b) Mid-range.
row investigations conducted in stationary annu- FIQTJBE
184.-Definitiona of reference incidence angle.
lar-cascade tunnels. Tunnels of this type have
been used primarily for inlet-guide-vane investi- the outlet system to vary the compressor back
gations. Typical examples of annular-cascade pressure. In this manner, the compressor mass-
tunnel investigations are reported in references flow rate can be controlled. In an installation
215 and 225. Numerous multistage-compressor such as this, compressor performance over a range
investigations have been conducted both in this of speeds and mass flows can be obtained simply.
country and abroad. Unfortunately, the data In many cases, test rigs similar to figure 185 have
obtained from these investigations are too limited been operated with only guide vanes and rotors
to permit the construction of individual blade- or with rotors alone.
row-element performance curves similar to those Many phases of compressor research have been
illustrated in figure 183. conducted in single-stage-compressor test rig5; and,
The data used in this investigation were obtained in reporting these phases, complete blade-element
primarily from investigations of single rotor rows results are not usually presented. Therefore, it
or of single-stage compressors. A typical single- was necessary to collect availabl nal data
and rework them in terms of the parameters of
the analysis. Since only NACA original data
of a row of inlet guide vanes, a rotor blade row were available in blade-element form, the data
driven by a variable-speed motor, and a stator analysis is based maidy on single-stage-compressor
blade row. A &char& throttle is installed in investigations conducted at the Lewis laboratory.
234 AERODYN-C DESIGN OF AXIAL-FLOW COIMPRESSORS

dynamometer

Filter, SCI

FIQUBE185.4chematic diagram of single-stage-compressor test installation.

The measurements taken and the instrumentation INCIDENCE-ANGLE ANALYSIS

used vary somewhat from compressor to compres- METHOD OF COBBELATION
sor; in most cases, however, it is possible from the In correlating blade-element reference-incidence-
available data to reconstruct complete experi- angle data, measured values of incidence angle are
mental velocity diagrams and to determine the cornpared with values of reference incidence angle
bladeelement performance. Radial survey meas- predicted for the geometry of the blade element
urements were made after each blade row. according to the low-speed two-dimensional-
Normally, total pressure, static pressure, total cascade correlations of chapter VI. In chapter
temperature, and air direction were measured. VI, the low-speed two-dimensional reference in-
The pressure- and temperature-measuring devices cidence angle is expressed in terms of the blade
were calibrated for the effect of Mach number. geometry as
Most of the compressor investigations that were
adaptable to this analysis were conducted on ;2-D=Ki(io)IO+np (279)
NACA -65(Alo)-series airfoil shapes and double-
circular-arc airfoils. Therefore, the analysis is where K r is a function of blade thickness distri-
concerned solely with these airfoils. The 65-(A10)- bution and maximum-thickness ratio, (io)lo is the
series airfoil has been used extensively in subsonic zero-camber incidence angle for the 10-percent-
compressors; and the double-circular-arc airfoil, thick airfoil section (function of air-inlet angle 8:
which is a relatively simple airfoil shape, has and solidity u), and n is equal to [ ( i - i o ) / p 1 2 - ~
been used effectively in transonic compressors. (also a function of 8: and u). Values of K,,(&)lo,
Details of the characteristics of the various blade and n for the circular-arc and 65-(A;o)-series blade
rows used in this analysis are summarized in table are repeated in figures 186 to 188 for convenience.
11, and details of the instrumentation, calculation The comparisons between measured blade-
procedure, and accuracy of measurement me element reference incidence angle & and predicted
given in the listed references. two-dimensional incidence angle &-D are expressed
 BLADE-ELEMENT FLOW IN ANJWLAR CASCADES 235
TABLE ,II.-DETAILS OF SINGLE-STAGE ROTORS AND STATORS

Blade Descrip Refer-

row tion ellm

6bSeries blade section

-
1 Rotor 14 0.6 &2,828.994,1104,
1214
0.30-0.76 I I I
19.9 16.3 1.31 1.31 1.010 0.690 30.8 230
2 Btator 14 .66 662,1104 .26- .73 30.1 30.1 1.31 1.31 .996 .62019.6 231
3 Rotor 30 .80 604,672, M O .36- .70 30.1 30.1 2.90 2.90 1.m -808 28.8 44
4 Rotor 14 .bo 1104,1214 .Bo- .80 30.1 30.1 1.31 1.31 1.010 .690 21.1 231
6 Rotor 14 .60 667,743 .39- .72 30.1 30.1 1.31 1.31 .962 .608 16.2 231
6 Stator 14 .62 371,667,743 .!2% .66 30.1 30.1 1.8 1.31 .993 .631 16.2 231
7 Rotor 14 .60 646 .35- .bt? 40.0 23.9 1.31 1.31 .966 .m 2f.o 230
8 Rotor 14 .60 662 828 1104 .30- .a6 40.0 23.9 1.31 1.31 .966 .m 21.0 230
9 Rotor 14 .M) 652' 828' 1104,1214 .30-.76 40.0 23.9 1.31 1.31 .966 .m 21.0 230
10 Stator 14 .63 412: 617: 823 .26- .74 30.1 30.1 1.31 1.31 .870 .687 16.1 231
11 Rotor l4 .80 669 763 836 .6Z .76 46.2 34.1 1.35 1.35 .823 .692 21.4 35.4 232
12 Rotor .80 669' 763' 836 .49- .76 46.2 34.1 1.36 1.36 1.12 .943 21.4 36.4 232
13 Rotor 14
l4 .80 669' 763' 836 .49- .76 46.2 34.1 1.36 1.36 1.69 1.35 21.4 36.4 m
14 Rotor 14 m:736: 874 30.3 19.4 1.46 1.82 1.20 1.20 42.6 224
-80 .KO- .ga
I I I -
Circulsr-arc blade seetion
-
16 Rotor 14 0.4 600,m,1o00, low 0.33-1.08 40.3 11.4 2.00 2.00 1.778 233
16 Rotor 18 .6 600,700,m,eao, .3s-l.07 28.3 7 1.6 1.6 1.63 234
lo00
17 Rotor 14 .6 600, SOo, e00, lo00 .37-1.17 29.4 13.7 2.09 2.32 1.28 66
18
19
20
Rotor
Rotor
Rotor
14
14
14
.6
.6
.6
3 E,lo00,1120
s0o,eao,lo00,llzo
.&1.12
.60-1.22
.4-.82
29.4
23.1
23.1
13.7
4.3
4.3
2.09
1.60
1.60
2.32
1.60
1.60
.86
1.40
1.40
66
229
229
2f Stator 17.36 .62 600,800.900 .4l- .63 62.0 62.0 3.26 3.26 1.M 236
22 Stator 17.36 .60 m,1 m .&.ti6 20.6 20.0 2.66 3.23 1.46 222
-
in terms of the difference (&-&J. Thus, a curacy generally contributes h e a d y to the data
value of zero of the difference parameter corre- scatter, especially at hub and tip.) Nevertheless,
sponds to an equivalence of the two incidence the results of the comparison are indicative of the
angles. In view of the established tendency of trends involved, and it is possible to make some
the reference incidence angle to increase somewhat general observations.
with inlet Mach number (ch. VI), it was thought For the rotor mean-radius region, where three-
desirable to plot the variation of the difference dimensional disturbances are most likely a mini-
parameter (&-&-D) against relative inlet Mach mum, the rotor minimum-loss incidence angles
number for the three radial positions at hub, mean, are, on the average, about '1 smaller than the
and tip. corresponding cascade-predicted values. This dif-
NACA 65- (Al0)-series blades.-The results of ference may be a reflection of some of the compres-
the comparison between compressor and two- sor influences discussed previously. The data
dimensional-cascade reference incidence angles for also indicate that no essential variation of refer-
the 65-(Alo)-seriesblades are presented in figure ence incidence angle with relative inlet Mach
189 for hub-, mean-, and tip-radius regions. Both number exists up to values of M,' of about 0.8.
rotor and stator data are presented; the stator The 65-(A10)-series blade, having a thick-nose
data being represented by the solid points. Dif- profile, apparently exhibits the same approximate
ferent values of incidence angle for a given symbol constancy of minimum-loss incidence angle with
represent different compressor tip speeds. As Mach number as indicated for the British thick-
might be expected in a correlation of this type nose C-series profile in the cascade comparisons
involving data from different test installations and of chapter VI.
instrumentations, the data are somewhat scattered, At the rotor tip, the compressor reference inci-
particularly in the hub and tip regions. It has dence angles are from 0' to 4' less than the pre-
not been possible in these instances to evaluate dicted cascade values. As in the case of the
the significance or origin of the scatter. (In rotor mean radius, no essential variation with
compressor investigations, instrumentation inac- inlet Mach number is observed in the range of
236 AERODYNAMIC DE&IGN OF AXIAL-FLOW COMPRESSORS

C
0
._
c
V
E
0
V

Blade moximum- thickness rotio, t/c ,

FIGIJIZE
186.-Thickness correction for zero-camber reference incidence angle (ch. VI).

data covered. The lower values of rotor reference confused by the wide range of data. A tendency
incidence angle were generally the result of a of the compressor incidence angles to be some-
change in the form of the variations of loss against what larger than the corresponding cascade
incidence angle in the rotor, as illustrated in values, with an average difference of about 1’ or
*e 190. The change in form may be explained 2O, is indicated.
on the basis of a probable increase in rotor tip B’or the stator mean-radius and hub regions,
three-dimensional losses (centrifuging of blade close agreement between compressor and cascade
boundary layer, tip-clearance disturbances, etc.) incidence angles is indicated for the range of
with increasing incidence angle. Mach numbers covered (to about 0.7). Con-
At the rotor hub, the situation is somewhat siderable scatter exists in the stator data at the
 BLADE-ELEMENT FLOW fN

a
Inlet-air angle, p ; , deg
4
FIGURE187.-Zero-camber reference incidence angle for
NACA 65-(Alo)-series and true circular-arc blades of
10-percent maximum-thickness ratio (see fig. 137, 0
ch. VI, for larger print).
-4
-.5
-82 .3 .4 .5 .6 .7 .8 .9 1.0
-.4 Relative inlet Mach number, Mi
C
L FIGURE189.-Variation of compressor reference incidence
2
U
-.3 angle minus two-dimensional-cascade-rule incidence
0
c angle with relative inlet Mach number for NACA
al
a -.2 65-(Alo)-series blade section.
-
0
m
It is immediately apparent that rotor reference
-.I
incidence angle at all radial positions increases with
increasing Mach number. The data indicate that
0 10 20 30 40 50 60 70 the magnitude of the increase in reference incidence
inlet-air angle, p,', deg angle with Mach number is larger a t the hub than
at the tip. The hub data points in figure 191 were
FIGURE188.-Reference-incidence-angle slope factor for obtained from blade elements of relatively high
NACA 65-(Alo)-series blades as equivalent circular arc
and for true circular-arc blades (see fig. 138, ch. VI, camber. Both potential-flow and low-speed-
for larger print). cascade results indicate that this type of con-
figuration is associated with a negative value of
compressor tip ; therefore, no definite conclusions reference incidence angle. As inlet Mach number
can be made concerning the variations of incidence is increased, the increase in incidence angle in the
angle. positive direction must be fairly large in order to
Double-circular-arc blade.-The results of the avoid high losses associated with blade-row
double-circular-arc airfoil correlation are pre- choking. In contrast, a t the compressor tip,
sented in figure 191, where compressor reference since the blade cambers are generally lower (see
incidence angle minus low-speed-cascade-rule table 11), the low-speed incidence angle is higher
incidence angle (eq. (279)) is plotted against and the required rate of change of incidence angle
relative inlet Mach number for the hub, mean, with increasing Mach number is not as large.
and tip radial positions for both rotors and stators. Unfortunately, low Mach number data were not
The dashed curve represents the variation ob- available to permit extrapolation of the rotor
tained with a 25O-camber double-circular-arc blade incidence-angle variations to zero Mach number
in high-speed two-dimensional cascade (ch. VI). (level of cascade correlation). However, it is
238 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

numbers considered. In contrast, the double-

circular-arc blade sections eihibit a pronounced
variation of reference incidence angle over the
range of Mach number investigated. Significant
difTerencea between the twodimensional-cascade
data and the rotor data were observed a t the com-
pressor tip. In contrast, at the mean radius and
hub, the differences in two-dimensional-cascade
data and rotor data were relatively small, even
though the flow field was three dimensional,
Additional data are required to determine the
Incidence angle, i, deg variation of stator reference incidence angle,
FIGURE190.-Typical variation of loss with incidence particularly for the double-circular-arc airfoil
angle for rotor blade element near tip and in two- sections. Also, no information has been presented
dimensional cascade for same blade geometry and inlet concerning the allowable incidence-angle range for
Mach number.
efficient (low-loss) operation and the variation of
this range with inlet Mach number. Investi-
believed that there will be very little change in the
rotor incidence angle for values of Mach number gations of these phases of compressor research are
below about 0.4 to 0.5. Extrapolated values of very essential to fill gaps in the compressor design
rotor reference incidence angle at zero Mach and analysis procedures.
number appear to be of the order of 0 . 5 O at the
hub, 1.5' a t the mean radius, and 2.5' at the tip
below cascade-rule values.
The double-circular-arc blade element in the 12
compressor rotor exhibits the same general inci-
dence-angle characteristic with Mach number that 8
was observed for sharp-nosed blade sections in the
4
high-speed two-dimensional cascade (ch. VI).
As indicated in chaper VI, the increase in refer- 0
ence incidence angle with Mach number is asso-
ciated with the tendency of the range of the blade
to be reduced only on the low-incidence side of the
loss curve as the Mach number is increased.
The rotor data for the double-circular-arc
section, like those for the 65-(Alo)-series blades,
are comparable with the cascade variations at the
mean radius, somewhat higher at the hub at the
higher Mach numbers, and noticeably lower at the
tip. Apparently, the same type of three-dimen-
sional phenomenon occurs at the tip for both
blade shapes.
The available double-circular-arc stator data
are too meager for any conclusions.
SUMMARY REMARKS

The variation of reference incidence angle for Relative inlet Mach number, 4'
65-(AlO)-seriesand double-circular-arc blade sec-
FIGWEE191.-Variation of compressor reference incidence
tions has been presented. No Mach number angle minus two-dimensional-caacade-rule incidence
effect on reference incidence angle was observed for angle with relative inlet Mach number for double-
the 65-(Alo)-seriesblades for the range of Mach circular-arc blade section.
 n Z .
.
.
n -7 n.m--
! LUW
ISLAJJL-LLbMBNT l
m-I

ANNULAR CASCADES 239

TOTAL-PRESSURE-LOSS ANALYSIS are infIuencing the tip loss. The specific three-
CORRELATION OF DATA dimensional factors or origins involved in the loss
For two-dimensional-cascade data obtained at rise at the tip are not currently known. The
low Mach numbers, the values of total-pressure- principal conclusion reached from the plot is that
loss parameter G cos /32/2cr plotted against diffusion the likelihood of a rising loss trend on the rotor
factor (eq. (278)) form essentially a single curve tip exists for values of diffusion factor greater than
for all cascade configurations. The diffusion- about 0.35.
factor correlation of loss parameter was applied to The stator losses at all radial positions in
data obtained over a range of Mach numbers from figure 192 appear to be somewhat higher than
single-stage axial-flow compressors of various those of the two-dimensional cascade, particularly
geometries and design Mach numbers. Values of at the higher values of diffusion factor.
total-pressure-loss parameter calculated from SUMMARY REMARKS
single-stage-compressor data are plotted against
diffusion factor for the hub, mean, and tip meas- Rotor and stator blade-element loss data were
uring stations in figure 192. Each symbol repre- correlated by means of the diffusion factor. The
sents the value of diffusion factor and loss param- losses for stator and rotor blade elements at hub
eter at reference incidence angle at a given tip and mean radii were somewhat higher than those
speed. Also plotted as a dashed curve is the cor- for the two-dimensional cascade over the range
responding correlation presented in chapter VI of diffusion factor investigated. At the rotor tip,
for the low-speed two-dimensional-cascade data. the losses were considerably higher at values of
The data of figure 192, which were obtained from diffusion factor above approximately 0.35.
the rotor and stator configurations summarized in The foregoing blade-element loss analysis is
table 11,represent both 65-(A,,,)-series and circular- clearly incomplete. The need for additional
arc blade sections. The plots of figure 192 work is indicated for such purposes as evaluating
essentially represent an elaboration of the loss- the origin and magnitude of the tip-region losses.
diffusion correlations of reference 9. The loading limits for rotors at other than the tip
The most important impression obtained from region and for stators at all blade elements have
the rotor data plots is the wide scatter and increas- not been determined, because, for the available
ing loss trend with diffusion factor at the rotor tip, data, the diffusion factors at reference incidence
while no discernible trend of variation is obtained do not extend to sufliciently high values. Single-
at the rotor hub and mean radii. For the rotor stage investigations are needed over the critical
hub and mean radii, it can be assumed that the range of Reynolds number to determine the effect
rotor blade-element loss parameter follows the of Reynolds number on the blade-element loss.
cascade variation but at a higher average magni- It is desirable to isolate the effects of velocity
tude. Unfortunately, the range of diffusionfactor diffusion and shock waves on the loss at high Mach
that could be covered in the compressor tests was number operation. The loss correlations pre-
not sufficient to determine whether a marked rise sented should also be extended so that the data are
in loss is obtained for values of diffusion factor applicable over a range of incidence angle. This
greater than about 0.6 (as in the cascade). would be of extreme value in the compressor
It is apparent from the loss trend and data analysis problem.
scatter at the rotm tip that a different loss phe- DEVIATION-ANGLE ANALYSIS
nomenon is occurring in the tip region. It is
recognized that a part of the scatter is due to the In addition to design information concerning
general instrumentation inaccuracy in the highly blade-element losses and incidence angle, it is
turbulent tip regions. In view of the usually large desirable to have a rather complete picture of the
radial gradients of loss existing in the blade tip air deviation-angle characteristics of axial-flow-
region, small variations in positioning radial compressor blade elements. Therefore, the two-
survey probes can cause noticeable differences in dimensional-cascade correlation results are re-
the computed results. Nevertheless, it is obvious viewed and supplemented with annular-cascade
that factors other than the blade-element dif€usion data in this section.
240 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

-05

'
10
rn
Blade row
(toble II)
I
2
0

.
0 3
b 4
.IO I I
0 5
6
Rotor meon 0 7
0 8
n 9
4 10
.05 - n h I I
a 12
4 13
V 14
0
0
U 17
18 I

Stotor tip
a
t
I. .= 4 .
----Ic _ _ _ _ _ _ _ _ . _ -_-.-- - - -
- /

J----'- J
 BLADE-ELEMENT FLOW IN ANNULAR CASCADES 241

Blade maximum-thickness ratio, t / c

FIGURE
193.-Thickness correction for zero-camber deviation angle (ch. VI).

METHOD OF CORRELATION
As was shown previously, the reference incidence
As in the analysis of reference incidence angle, angle of the compressor blade element may differ
the correlation of blade-element deviation angle somewhat from the corresponding two-dimensional
at reference incidence is presented in terms of a reference incidence angle. Since deviation angle
comparison between measured blade-element devi- will vary with changing reference incidence angle
ation angle and deviation angle predicted for the for a given blade geometry (depending on solidity),
element accordirig to the low-speed two-dimen- the two-dimensionaldeviation angles were corrected to
sional-cascade correlations of chapter VI. In the reference incidenceangles of the compressor blade
chapter VI, the low-speed two-dimensional-cascade elements. The corrected deviation angle, as
deviation angle at reference incidence angle is suggested in chapter VI, is given by
expressed in terms of blade geometry as

where (dS0/di),-, is the slope of the two-dmen-

where Kais a function of maximum-thickness-to- of deviation angle with incidence
chord ratio and thickness distribution, (6z)10is the Values of Ka,
zero-camber deviation angle for the lo-perce cular-arc and 65-
thick airfoil section (function of a: and u), m is a figures 193 to 197
function of pi for the different basic camber distri- for convenience.
butions, and b is an exponent that is also a function Deviation-angle comparisons for the double-
of a:. circular-arc blade were also made on the basis of

691-561 0-65-17
242 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

oa
M,

tion angle (eq. (281)) for the 65-(Alo)-serieS air-

= and stators are
ch number for
tip radial positions in fi
es of deviation angle corr
FIQIJBE
194.-Zero-camber deviation angle for NACA compressor reference incidence angle. As in the
65-(A&series and true circular-arc blades of 10-per- cam of the incidence-angle and loss correlations,
cent maximum-thickness ratio (see fig. 161, ch. VI, for
larger print). there is considerable scatter of data, particularly
in the hub and tip regions. Some of the scatter
Carter's rule for cascade blades (ref. 88): is believed due to the effects of three-dimensional
flows and changes in axial velocity ratio across
the element, but perhaps the most important
factors are instrumentation differences and errors.
It is generally recognized that it is difficult to
where m, is a factor that is a function of blade- measure compressor air angles with an accuracy
chord angle (fig. 198). better than about f1' to 1.5'. The correlations
Carter's rule, which has been used extensively must therefore be evaluated on an average or
in the design of circular-arc blades, was used as trend basis.
the basis for the more elaborate rule of equation The correlation of rotor data in the mean-radius
(280). In the calculations, Carter's rule was region is fairly good; axial velocity ratio varied

Inlet -air angle, p,', deg

FIOWE195.-Factor m in deviation-angle rule (see figs. 163 and 166).
 BLADE-ELEMENT FLOW IN ANNULAR CASCADES 243
ity ratio at essentially constant circulation for the
stator tends to decrease deviation angle.) As in
the caae of h e rotor, no essential variation of
deviation angle ach number is detected
for the stator wit range of Mach numbers
investigated.
Double-circular-arc blade.-Blade-element
two-dimensional-cascade deviation angles (eq.
(281)) obtained for the double-circular-arc blade
are compared in figure 199(b). The scatter of
data is generally less than for the 65-(A10)-series
blades, partly because of the generally more ac-
curate measurements taken in these investigations
Inlet - air angle, p,' , deg
(all are more recent than the data of fig. 199(a)).
FIGURE196.-Solidity exponent b in deviation-angle rule On the average, at the lower Mach numbers
(see fig. 164, oh. VI, for larger print). the blade-element deviation angles were about
1.5' less than the two-dimensional values at the
from about 0.9 to 1.10. On the average, the tip, 1.0' greater at the hub, and equal to the
rotor mean-radius deviation angles are about 0.5' two-dimensional values at the mean region.
less than the cascade values. These results agree Ranges of axial velocity ratio covered for the
with previous experience (refs. 218 and 219), data were 0.85 to 1.05 at the tip, 0.95 to 1.5 at
which indicated rotor turning angles approxi- the hub, and 0.90 to 1.15 at the mean radius. A
mately 1' greater (i.e., deviation angles 1' less) slightly increasing trend of variation with inlet
than the two-dimensional-cascade results. If data Mach number may be indicated at the mean
points for the rotor tip having axial velocity ratios radius and possibly also at the hub.
less than 0.8 are neglected, the average deviation The double-circular-arc stator data available
angle is about 0.5' less than the cascade value. (solid symbols) are too limited to permit any re-
Axial velocity ratio for the tip-region unflagged liable conclusions to be drawn. It appears,
data varied between 0.8 and 1.05. For the hub, however, that at the stator mean radius, the
on the average, the blade-element deviation angles blade-element deviation angles may be about 0.5'
were about 1.Oo greater than the corresponding less than the two-dimensional-cascade values.
two-dimensional values. Hub axial velocity ratios This is essentially the same trend observed for the
vaned between 1.0 and 1.3. As in the two- 65-(Alo)-series stators at mean radius in k u r e
dimensional cascade (ch. VI), no Mach number 199(a). Blade-element deviation angles appear
effect on deviation angle is indicated over the to be greater at the tip and smaller at the hub
range of Mach number investigated for all three
regions. I .o
For the stator mean-radius (Vz,z/Vz,l=l.O to a
u
.0
1.1) and hub-radius (Vz,z/Vz,1=0.85to 1.05) U
regions, the average deviation angles are both u ?
about 1.0' lower than the corresponding two- -73
.
C 2 .6
\
F o
dimensional values. A t the stator tip, the ,M .4
average blade-element value is indicated to be oz.
a,
about 4' less than the two-dimensional value. 8
- 2
cn
However, these data all have high axial velocity
ratios (from 1.1 to 1.5). It is expected that, on 0 .2 .4 .6 .8 1.0 1.2 1.4 1.6 1.8
the basis of constant axial velocity, the probable Solidity, u

average blade-element deviation angles at the FIGURE197.-Deviation-angle slope (dSo/dz>ap at refer-

stator tip might be several degrees closer to the ence incidence angle (see fig. 177, ch. VI, for larger
two-dimensional values. (Increasing axial veloc- print).
244 AERODPNAMIC DESIGN OF AXIATJ-FLOW COMPRESSORS

F"
L
0
c
u
0
LL

Blade-chord angle, ,o ,deg

FIQURE
198.-Variation of m, for circular-arc compressor cascades (ref. 88).

than the two-dimensional values. Ranges of SUMMARY REMARKS

axial velocity ratio were 1.0 to 1.25 at the tip, From the comparisons of measured and pre-
0.95 to 1.27 at the mean radius, and 0.9 to 1.30 dicted reference deviation angles for the NACA
a t the hub. 65-(AI0)-series and double-circular-arc blades, it
was found that the rules derived from two-
dimensional-cascade data can satisfactorily pre-
dict the com pressor refereme blade-element
deviation angle in the rotor and stator mean-
radius regions for the blade configurations pre-
obtained from the modified rule of equation (281) sented. Larger differences between rule and
for the range of blade-element measured values were observed in the hub and
in the data, the agreement wi tip regions. These differences can be attributed
data remains quite good. to the effects of three-dimensional flow, differences
 ES 245

C
.-
c
0

.-0
W
5
- 0 -
 246 AERODYNAMIC DESIGN OF AXWI-FLOW COMPRESSORS

4 curves at reference incidence angle is a dii3idt

task because of the scatter of the experimental
0

-4

8
$ 4
ON
m
' 0
mU
O

-4

0
desired blade-element turning angle Ap' and rela-
-4' ' ' ' ' ' ' ' ' ' ' ' ' ' '
.4 .5 .6 .7 .8 .9 1.9 1.1 ' '
1.2 tive inlet Mach number Mi are obtained from the
Relative inlet Mach number, M , design velocity diagram. Camber and turning
angles are related by the equation
FIGURE
2OO.-Variation of compressor deviation angle
minus deviation angle predicted by Carter's rule at p=p:-p;+sO-i (285)
reference incidence angle with relative inlet Mach
number for double-circular-are blade section. Compressor blade-element incidence angles (eqs.
(279) and (283)) and deviation angles (eqs. (281)
tions of relative inlet Mach number for several and (284)) are given by
radial positions along the blade height in figures
201 and 202. The curves in figures 201 and 202 i,=K(io)10+w+ (ic--iz-,) (286)
are faired average values of the data spread and,
strictly speaking, represent bands of values. In
view of the very limited data available, com-
pressor correction curves could not reliably be
established for the stator deviation and incidence
angles. Substituting equations (286) and (287) into equa-
Establishing single deduced blade-element loss tion (285) and rearranging terms yield

All terms on the right side of equation (288) rotor and complete stage can be computed by
' can be determined from the velocitydiagram means of the techniques and equations presented
properties, the specified blade shape and thickness, in the appendix to this chapter. If the change in
and the specified solidity. After the camber radius across the blade row can be assumed small,
angle is determined, the incidence and deviation blade-element efficiency can be determined through
angles'can be calculated from equations (286) the use of figures 204 to 206 from the selected
and (287). Rotor blade-element loss parameter values of Z' and the values of Mi and absolute
is estimated from the velocitydi Busion total-preasure ratio or total-temperature ratio
factor and the curves of figure 2 e total- obtained from the velocity diagram.
ent w' is then r cedure can best be illustrated
from the blade-element solidity and relative air ple. Suppose the following
outlet angle. Blade-element efficiencies for the specified rotor design values represent typical
 BLADE-ELEMENT FLOW IN ANNoTrAR CASCADES 247

Relotive inlet Moch number, M i

(a) NACA 65-(A10)-series blades.

(b) Double-circular-arc blades.
FIGURE201.-Deduced variation of average rotor reference incidence angle minus low-speed two-dimensional-cascade-
rule reference incidence angle with relative inlet Mach number.

0
0)
'p
-2
4
OtQN
I
2
a00
0

-2
.3 .4 .5 .6 .7 .8 .9 1.0 1.2
Relotive inlet Moch number, M,'

(a) NACA 66(A10)-seriesblades.

(b) Double-circular-arc blades.
FIGURE202.-Deduced variation of average rotor deviation angle minus low-speed two-dimensional-cascade-rule devia-
tion angle at compressor reference incidence angle with relative inlet Mach number.
248 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

-N
@a

13

I
W
c

E
xF
m
m
-
0
I
W
1
3
m
m
a,
I

-e
c
0 .O
0
I-

0
D i f f u s i o n factor, D

(a) Rotor.
(b) Stator.
FIGURE203.-Deduced variation of total-pressure-loss parameter with diffusion factor at reference incidence angle for
NACA 65-(Alo) -series and double-circular-afcblades.

values at 10 percent of the passage height from (1) From the value of Mi and figures 201 (b)
the compressor tip: and 202(b),
ic-&-D=4.O0 -1.5'
1
=56.9' S&&=
@'=10.9O (2) From the values of &, U , and t/c and figures
186 to 188 and 193 to 197,
obtained from velocit -diagram
calculati'ons (ch. $111) K,=0.54 (&)10=4.40 n=-0.22 Ka=0.37
D=0.35

T"=
Tl
1.091
J (S~)lO=l.6' m=0.305 b=0.714
dP
(~ )~-~=0.095

u= 1.o (3) When the values of steps (1) and (2) are
assumed values
t/c=O.OS substituted in equation (288), the value of blade
camber p=8.4'.
The problem is to find the camber, incidence, (4) From equations (286) and (287), &=4.5
and deviation angles and the total-pressure-loss and 6;=2.0.
coefficient for a double-circular-arc airfoil section (5) For calculation of the total-pressure-loss
that w i l l establish the velocity-diagram values. coefficient, the diffusion factor (0.35) and figure
 BLADE-ELEMENT FLOW IN ANNULAR CASCADES 249

E
v)

E
Q
-
-
0
c
0

Total-pressure-loss coefficient , W’
FIQURI ient and inlet Mach number.
250 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

9
VI
9
e
-
0
c
0
c

FXGUIZE
205.-Variation of relative total-pressure ratio with absolute total-temperature ratio and efficiency for rotor
 BLADE-ELF*MENT FLOW IN’ ANNULAR CMCADES 251

L
0
c
2
0
c

x
V
C
.-
a,
2
L

a,

0
.-c
0
U

Rotor total-pressure ratio, ( $ 1

FIGURE
206.-Variation of ratio of stage to rotor efficiencywith rotor absolute total-pressure ratio aa function of stator
recovery factor.
203(a) yield a value of 0.025 for the loss param- At any rate, the establishment of flow angles
eter (2 cos /3:)!2,7, and and blade geometry at the reference incidence
angle can serve as an anchor point for the deter-
/3; =/3; -A/3' =56.9 -10.9=46 .O O
mination of conditions at other incidence-angle
COS &=0.6947 settings. For deviation-angle and loss variations
over the complete range of incidence angles,
Theref ore, reference can be made to available cascade data.
Such low-speed cascade data exist for the NACA
- w'cosf9;
2u )-=2u 0'025x 2=o.072
0.6947 65-(Al,)-series blades (ref. 54).
It is recognized that many qualifications and
(6) For a negligible change in radius across the limitations exist in the use of the foregoing design
blade element, the following values can be found procedure and correlation data. For best results,
from figures 204 and 205: the application of the deduced variations should
be restricted to the range of blade geometries
(camber, solidity, etc.) and flow conditions (inlet
Mach number, Reynolds number, axial velocity
ratio, etc.) considered in the analysis. In some
The preceding example has been carried out for cases for compressor designs with very low turning
a typical transonic rotor blade section. A similar angle, the calculated camber angle may be nega-
procedure can be used for stator blade sections when tive. For these cases it is recommended that a
adequate blade-element data become available. zero-camber blade section be chosen and the in-
SUMMARY REMABKS cidence angle selected to satisfy the turning-angle
requirements. The data used in the analysis were
The foregoing procedures and data apply only
obtained for the most part from typical experimen-
to the reference point (i.e., the point of minimum
tal inlet stages with essentially uniform inlet flow.
loss) on the general loss-against-incidence-angle
Nevertheless, such data have been used success-
variation for a given blade element. The ref-
fully in the design of the latter stages of multi-
erence minimum-loss incidence angle, which was
stage compressors. It should also be remembered
established primarily for purposes of analysis, is
that the single curves appearing in the deduced
not necessarily to be considered as a recommended
variations represent essentially average or repre-
design point for compressor application. The
sentative values of the experimental data spread.
selection of the best incidence angle for a particular
blade element in a multistage-compressor design Also, in some cases, particularly for the stator, the
available data are too limited to establish reliable
is a function of many considerations, such as the
correlations. Considerable work must yet be
location of the blade row, the design Mach number, done to place the design curves on a firmer and
and the type and application of the design. How- wider basis. The design procedures established
ever, at transonic inlet Mach number levels,' the and trends of variation determined from the data,
point of minimum loss may very well constitute a however, should prove useful in compressor blade-
desired design setting. element design.
 APPENDIX
EQUATIONS FOR BLADE-ELEMENT EFFICIENCY

By definition, for a complete atage consisting For the rotor alone, the blade-element efficiency
of inlet guide vanes, rotor, and stator, the adia- is given by
batic temperature-rise efficiency of the flow along
a stream surface is given by

.-I, a

From equation (B3) of reference 9, the loss

From the developments of reference 7 (eq. (B8)in coefficient of the rotating blade row (based on
the reference), the absolute total-prasure ratio inlet dynamic pressure) is given by
across a blade row P2/P1 can be related to the
relative total-pressure ratio across the blade row
PL/P;according to the relation
.f=(5) id

where (PL/P;)M is the ideal (no loss) relative total-

pressure ratio. The relative total-pressure ratio is
also referred to as the blade-row recovery factor. For any blade element, then, from equation (58),
For stationary blade rows &e., inlet guide vanes
and stators), (Pi/P;)a is equal to 1.0. For rotors,
the ideal relative total-pressure ratio (eq. (B4) of
ref. 9) is given by
Y

The relations presented in equations (A4),

(A5), and (A6) indicate that four quantities are
in which MT is equal to the ratio of the outlet required for the determination of the blade-
element wheel speed to the inlet relative stagnation element efficiency across the rotor or stage: the
, ~ ) rl/r2
velocity of sound ( ~ r ~ / a :and , is the ratio rotor absolute total-temperature ratio, the relative
of inlet to outlet radius of the streamline across total-pressure-loss coefficient (based on inlet
the blade element. (For a flow at constant dynamic pressure), the relative inlet Mach
radius (cylindrical flow), (PL/P;)ta is equal to 1.0.) number, and the ideal relative total-pressure
Thus,from equations (Al) and (B), ratio. Thus, the blade-element efficiencies for a
given stage velocity diagram can be calculated if
the loss coefficients of the blade elements in the
various blade rows can be estimated.
For simplicity in the efficiency-estimation pro-
cedure, effects of changes in radius across the
blade row can be assumed small @e., rl=rz), so
that the ideal relative pressure ratio is equal to
263
254 AERODYNAMIC DESIGN OF AXIAL-FLOW COMPRESSORS

unity. Then, equations (A4), (A5), and (A6) total-pressure ratio as

.W-1
become, respectively,

The charts are used as follows: For known or

estimated values of rotor total-pressure-loss co-
efficient Z' and relative inlet Mach number
Mi of the element, the corresponding value of
relative recovery factor PL/Pi is determined
from figure 204. From the value of rotor-
element absolute total-temperature ratio TJT,
and (obtained from calculations of the design velocity
diagram) and the value of (Pi/Pi) obtained from
figure 204, the rotor-element efficiency is deter-
mined from figure 205. Rotor absolute total-
pressure ratio can also be determined from the
dashed lines in figure 205.
For purposes of rapid calculation and pre- If inlet guide vanes and stators are present,
liminary estimates, the efficiency relations are the respective recovery factors of each blade
expressed in chart form in figures 204 to 206. row are first obtained from figure 204. The
The relation among relative recovery factor, product of the two recovery factors is then cal-
blade-element loss coefficient, and inlet Mach culated and used in conjunction with the rotor
number (ea. (A9)) is presented in figure 24. absolute total-pressure ratio in figure 206 to
A chart for determining rotor blade-element determine the ratio of stage efficiency to rotor
efficiency from relative recovery factor and ab- efficiency. A simple multiplication then yields
solute total-temperature ratio (eq. (A8)) is given the magnitude of the stage efficiency along the
in figure 205. Lines of constant rotor absolute element stream surface.
The charts can also be used to determine gross
total-pressure ratio are also included in the
or mass-averaged efficiencies through the use of
figure. Figure 206 presents the ratio of stage over-all loss terms. Furthermore, the charts can
efficiency to rotor efficiency for various stator be used for the rapid determination of relative
or guide-vane recovery factors. The ratio of total-pressure-loss coefficient from known values
stage efEciency to rotor efficiency is obtained of efficiency, pressure ratio, and inlet Mach
from equation (Al) in terms of rotor absolute number on an element or gross basis.