Revised Fortran Program For Calculating Velocities and Streamlines I PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 112

w_

j •

NASA TECHNICAL NOTE NASA TN D-8430

t477- 18992
Z
I-- He,. _ o_
MF Ao I
UncJ.a-_
_I/01 21299
Z

\_ "_ _.- _I .

REVISED FORTRAN PROGRAM FOR


CALCULATING VELOCITIES AND STREAMLINES
ON THE HUB-SHROUD MIDCHANNEL STREAM
SURFACE OF AN AXIAL-, RADIAL-, OR
MIXED-FLOW TURBOMACHINE OR ANNULAR DUCT

I - User's Manual

Theodore Katsanis and William D. McNally


Lewis Research Center
Cleveland, Ohio 44135

NATIONAL AERONAUTICSAND SPACE ADMINISTRATION . WASHINGTON, D. C. • MARCH 1977


1, Report No. 2. Government Accession No. 3. Recipient's Catalog No.
NASA TN D-8430

"' 4. Title and Subtitle REVISED FORTRAN PROGRAM FOR CALCULATING 5. Report Date
VELOCITIES AND STREAMLINES ON THE HUB-SHROUD MIDCHANNEL Narch 1977
STREAM SURFACE OF AN AXIAL-, RADIAL-, OR MIXED-FLOW
6, Performing Organization Code
TURBOMACHINE OR ANNULAR DUCT
I- USER'S MANUAL

7. Author(s) 8. Performing Organization Report No

E -8734
Theodore Katsanis and William D. McNally
10. Work Unit No.

9, Performing Organization Name and Address


505 -04
Lewis Research Center
11. Contract or Grant No,

National Aeronautics and Space Administration

Cleveland, Ohio 44135 13. Type of Report and Period Covered

12. Sponsoring Agency Name and Address Technical Note


National Aeronautics and Space Administration
14. Sponsoring Agency Code
Washington, D. C. 20546

15, Supplementary Notes

Supersedes NASA TN D-7343 (FORTRAN Program for Calculating Velocities and Streamlines
on the Hub-Shroud Mid-Channel Flow Surface of an Axial- or Mixed-Flow Turbomachine, 1973)

16. Abstract

A FORTRAN-I V" computer program has been developed that obtains a detailed subsonic or shock-
free transonic flow solution on the hub-shroud midchannel stream surface of a turbomachine.

The blade row may be fixed or rotating, and the blades may be twisted and leaned. Flow may be

axial, mixed, or radial. This program is a revision of a previous program and this report
supersedes NASA TN D-7343. Upstream and downstream flow variables may vary from hub to
shroud, and provision is made to correct for loss of stagnation pressure. The results include

velocities, streamlines, and flow angles on the stream surface as well as approximate blade
surface velocities. Subsonic solutions are obtained by a finite-difference, stream-function so-

lution. Transonic solutions are obtained by a velocity-gradient method that uses information

from a finite-difference, stream-function solution at a reduced mass flow.

17. Key Words (Suggested by Author(s)) 18. Distribution Statement

Meridional plane; Turbomachine flow; Midchannel Unclassified - unlimited


stream surface, Axial-flow turbomachine; Mixed- STAR Category 01
flow turbomachine; Transonic flow; Radial-inflow

turbine; Centrifugal- compressor

19. Security Classif, (of this report) 20. Security Classif, (of this page) 21. No, of Pages 22, Price"

Unclassified Unclassified 105 A06

For sale by the National Technical Information Service, Springfield, Virginia 22161
CONTENTS

Page
SUMMARY ....................................... 1

_TRODUCTION ....................................

METHOD OF ANALYSIS ...............................


BASIC ASSUMPTIONS ...............................
SOLUTION BY COMBINATION OF METHODS ...................
SUBSONIC STREAM-FUNCTION SOLUTION ....................
TRANSONIC VELOCITY-GRADIENT APPROXIMATE SOLUTION .........
BLADE SURFACE VELOCITIES ..........................

APPLICATION OF PROGRAM ............................

DESCRIPTION OF INPUT AND OUTPUT ....................... 10


INPUT ........................................ 11
Input Dictionary .................................. 17
Special Instructions for Preparing Input ..................... 25
(a) Units of measurement ........................... 26
(b) Damping factors FNEW and DNEW .................... 26

(c) Hub and shroud flow channel geometry .................. 26

(d) Orthogonal mesh .............................. 27

(e) Upstream and downstream flow conditions ................. 27


(f) Mean blade surface and thickness coordinates ............... 28

(g) How to specify points for spline curves .................. 29

(h) Requests for output data .......................... 29

(i) Incompressible flow ............................ 31


(j) Straight infinite cascade .......................... 31
(k) Choosing a value for REDFAC ....................... 31
(l) Choosing a value for ANGROT ....................... 31
OUTPUT ....................................... 31

Printed Output .................................. 32


Plotted Output .................................. 52

Error Messages ................................. 66

NUMERICAL EXAMPLE ............................... 7O

iii

PRECEDING PlaiCE BLANK NOT FILMED


APPENDIXES

A- GOVERNING EQUATIONS ........................... 73

B - DERIVATION OF STREAM-FUNCTION EQUATION .............. 77

C o DERIVATION OF VELOCITY-GRADIENT EQUATION ............ 85


D- LOSS CORRECTIONS .............................. 90

E - DEFINING REDUCED--MASS- FLOW PROBLEM ................ 92

F - INCIDENCE AND DEVIATION CORRECTIONS ................. 93

G- BLADE SURFACE VELOCITIES AND BLADE-TO--BLADE

AVERAGE DENSITIES ............................. 95

H- SYMBOLS .................................... 98

REFERENCES .................................... 101

iv
REVISED FORTRAN PROGRAM FOR CALCULATING VELOCITIES AND

STREAMLINES ON THE HUB-SHROUD MIDCHANNEL STREAM

SURFACE OF AN AXIAL-, RADIAL-, OR MIXED-FLOW

TURBOMACHINE OR ANNULAR DUCT

I - USER'S MANUAL*

by Theodore Katsanis and William D. McNally

Lewis Research Center

SUMMARY

A FORTRA N-IV computer program has been developed that obtains a detailed sub-
sonic or transonic flow solution on the hub-shroud mldchannel stream surface of a
single blade row of a turbomachine. A solution can also be obtained for an annular
duct without blades. The flow must be essentially subsonic, but there may be locally
supersonic flow. The solution is for two-dimensional, adiabatic shock-free flow. The
blade row may be fixed or rotating, and the blades may be twisted and leaned. The
flow may be axial, mixed, or radial. Upstream and downstream flow conditions can
vary from hub to shroud, and provision is made for an approximate correction for loss
of stagnation pressure. Viscous forces are neglected along solution mesh lines running
from hub to tip.
The present program is a revision of a previous program and this report super-
sedes NASA TN D-7343. The primary revisions are to extend the program to handle
nonaxial flows without restriction, to handle annular ducts without blades, to allow for
any specified streamwise loss distribution, and to make numerous detailed improve-
ments for more accurate and efficient calculations.

The basic analysis is based on the stream function and consists of the solution of
the simultaneous, nonlinear, finite-difference equations of the stream function. This
basic solution, however, is limited to strictly subsonic flow. When the re is locally
supersonic flow, a transonic solution must be obtained. The transonic solution is ob-
tained by a combination of a finite-difference, stream-function solution and a velocity-
gradient solution. The finite-difference solution at a reduced mass flow provides infor-
mation that is used to obtain a velocity-gradient solution at the full mass flow.
The program is reported in two volumes, with part I as the user's manual and
part II as the programmer's manual. This report, part I, contains all the information
necessary to use the program as it is. It explains the equations involved and the method
of solution and gives a numerical example to illustrate the use of the program. Part II
includes the complete program listing and a detailed program procedure.
,
Supersedes NASA TN D-7343.
INTRODUCTION

The design of blades for compressors and turbines ideally requires methods for
analyzing unsteady, three-dimensional, turbulent viscous flow through a turbomachine.
Clearly, such solutions are impossible at the present time, even on the largest and fast-
est computers. The usual approach at present is to analyze only steady flows and to
separate inviscid solutions from viscous solutions. Three-dimensional inviscid solu-
tions are just beginning to be obtained with the present generation of computers, but they
require excessive computer time. So, at present, inviscid analyses usually involve a
combination of several two-dimensional solutions on intersecting families of stream
surfaces to obtain what is called a quasi-three-dimensional solution.
Since there are several choices of two-dimensional surfaces to analyze and many
ways of combining them, there are many approaches to obtaining a quasi-three-
dimensional solution. Most two-dimensional solutions are either on a blade-to-blade

surface of revolution (Wu's S1 surface, ref. 1) or on the meridional or midchannel


stream surface between two blades (Wu's S2 surface). However, when three-
dimensional effects are most important, significant information can often be obtained
from a solution on a passage cross-sectional surface (normal to the flow). This is
called a channel solution (fig. 1).

In this report a solution to the equations of flow on the meridional S2 surface is


carried out. This solution surface is chosen when the turbomachine under consideration
has significant variation in flow properties in the hub-shroud direction, especially when
input is needed for use in blade-to-blade calculations. The solution can be obtained
either by the quasi-orthogonal method, which solves the velocity-gradient equation from
hub to shroud on the meridional stream surface (ref. 2), or by a finite-difference me-
ihod, which solves a finite-difference equation for stream function on the same stream
surface. The quasi-orthogonal method is efficient in many cases and can obtain solu-
tions into the transonic regime. However, there is difficulty in obtaining a solution
when blade aspect ratios are above 1. Difficulties are also encountered with curved
passages and low-hub-tip-ratio blades. For such cases, the most promising method is
the finite-difference solution, but this solution is limited to completely subsonic flows.
Finite-difference programs for flowon the midchannel surface of a turbomachine
have been reported in the literature. However, many of these programs are proprie-
tary or are of limited generality. The program reported herein is very general and has
been thoroughly tested and refined as the result of extensive usage at the NASA Lewis
Research Center.
The program described in this report uses both the finite-difference and the quasi-
orthogonal (velocity gradient) methods, combined in a way that takes maximum advan-
tage of both. The finite-difference method is used to obtain a subsonic-flow solution.

2
Blade-,

Flow
\ ,- Midchannel
] surface S2
Orthogonal
Blade channel
surface -,

L Blade-to-blade
surface S1

CD-I1362-01

Figure 1. -Two-dimensionalanalysissurfaces in a turbomachine.

The velocity-gradient method is then used if necessary to extend the range of solutions
into the transonic regime.
A computer program called MERIDL has been written to perform these calcula-
tions. This program was written for an axial-, mixed-, or radial-flow turbomachine
blade row, either a compressor or turbine, or for an annular duct. Upstream and down-
stream flow conditions can vary from hub to shroud. The solution is for compressible,
shock-free flow or for incompressible flow. An approximate correction for loss of
stagnation pressure through the blade row is provided. The blade row may be either
fixed or rotating. The blades may be twisted and leaned and can have high aspect ratio
and arbitrary thickness distribution.
The solution obtained by this program also provides the information necessary for
a more detailed blade-shape analysis on blade-to-blade surfaces (fig. 1). A useful pro-
gram for this purpose is TSONIC (ref. 3). Information needed to prepare all the input
for TSONIC is calculated and printed by MERIDL.
The MERIDL program reported herein is a revision of the program described in
references 4 and 5. Two types of changes were made: first, extension of the program's
capability to handle cases beyond those originally offered; and second, revisions to im-
prove the accuracy and efficiency of the program. Although the input form has been ex-
tended to handle additional input where required, any input that was satisfactory for the
original MERIDL program is still satisfactory for the revised MERIDL program. The
following major extensions and revisions were made to the program (additional internal
changes are documented in part II):
(1) The program has been extended to handle nonaxial flow configurations, including
both radial turbines and centrifugal compressors.
(2) The program has been extended to handle an annular duct without blades.
(3) The program has been extended to permit the user to specify an arbitrary
streamwise distribution of loss within the blades. This is in addition to the original
provision for hub-shroud loss distribution.
(4) The program has been modified so that the blade thickness can be specified pre-
cisely by a set of tangential (circumferential) thickness coordinates. The original pro-
gram required specification of thickness normal to the mean camber line on an input
blade section. This normal thickness was influenced by blade lean, camber, and non-
parallel blade surfaces and was difficult to specify accurately for some blade shapes.
(5) If desired the leading- and trailing-edge, mean-camber-line tangent angles can
be specified as input. This simplifies the specification of some blade shapes.
(6) Output quantities have been added to station-line output to give absolute velocity
components and to give static as well as absolute and relative total temperature, den-
sity, and pressure.
(7) Several informational messages have been added to the output.
(8) Additional error messages have been provided.
(9) Upstream and downstream boundary conditions have been changed to improve
both the convergence and the quality of the solution near these boundaries.
(10) Interpolation and calculation procedures near the leading and trailing edges have
been improved to give better convergence and smoother solutions in these regions.
(11) Numerous small changes have been made to improve the accuracy and efficiency
of the program.
The MERIDL program has been implemented on the NASA Lewis time-sharing
IBM- TSS/360- 67 computer. For the numerical example of this report, the storage of
variables required 60 000 words for a 21 × 41 grid of 861 points. The amount of varia-
ble storage could be easily reduced by equivalencing of variables or by using a coarser
mesh. Storage requirements for the program code depend on the computer system and
compiler being used. Run times for the program range from 3 to 15 minutes on
IBM-TSS/360- 67 equipment.
The MERIDL program is reported in two volumes, with part I as the user's manual

4
and part II (ref. 6) as the programmer's manual. This report, part I, contains all the
information necessary to use the program. It explains the method of solution, de-
scribes the input and output, gives a numerical example to illustrate the use of the pro-
gram, and derives the equationsused (in appendixes A to G.) Symbols are defined in
appendix H. Part II includes a complete program listing, detailed program procedure,
andappendixesthat derive the special numerical techniques used.

METHOD OF ANALYSIS

BA SIC A SSUMP TIONS

It is desired to determine the flow distribution through a stationary or rotating


cascade of blades on a midchannel hub-shroud stream surface. The following simpli-
fying assumptions are used in deriving the equations and in obtaining a solution:
(1) The flow relative to the blade is steady.

(2) The fluid is a perfect gas with constant specific heat Cp.
(3) The only forces along a hub-shroud orthogonal mesh line are those due to
momentum and pressure gradient.
(4) There is no heat transfer.
(5) The midchannel surface is a stream surface that has the same shape as the
blade mean-camber surface, except near the leading and trailing edges, where an arbi-
trary correction is made to match the free-stream flow.
(6) The velocity varies linearly between blade surfaces.
(7) The relative stagnation pressure loss is known through the blade row.
(8) The upstream and downstream boundaries of the solution region are orthogonal
to the streamlines.

The flow may be axial, mixed, or radial. Whirl, stagnation pressure, and stagna-
tion temperature may vary from hub to shroud, both upstream and downstream of the
blade row. The blade row may be either fixed or rotating, with leaned and twisted
blades; or there may be no blades at all. Within the given assumptions, no terms are
omitted from the equations.
In connection with assumption 3, the viscous forces along a hub-shroud orthogonal
mesh line are neglected, since these forces are usually very small. The viscous for-
ces in the streamwise direction are much larger. The effects of these streamwise
viscous forces are considered indirectly by specifying a streamwise total-pressure-
lo s s dis tribution.

5
SOLUTIONBY COMBINATIONOF METHODS

A flow analysis on the meridional stream surface can be obtained either by the
velocity-gradient methodor by the finite-difference method. The finite-difference
method is limited to subsonic flow; the velocity-gradient method is limited to relatively
low-aspect-ratio blades. The most accurate solution is obtainedby the finite-
difference technique, so that this method is usedwhere possible (i. e., for subsonic
flow). With locally supersonic flow, the finite-difference solution is first obtained at
a reduced mass flow (appendix E) for which the flow field is completely subsonic. The
streamline curvatures andflow angles throughout the passagethat are obtained from
this solution provide the information necessary to obtain an approximate velocity-
gradient solution at full mass flow, regardless of aspect ratio.

SUBSONIC S TREAM- F UNCTION SOLUTION

The stream-function equation (eq. (A1), appendix A) is a partial differential equa-


tion on a midchannel hub-shroud stream surface (assumption 5, p. 5). It is in one un-
known (the stream function) as a function of two variables and is derived from Wu's

momentum equation (eq. (96), ref. 1) on what he calls an S2 surface. Equation (A1) is

_on region

_" is on this surface

Section A-A

c-Solution region
Sh roud-_, /

Flow _ A

,- Downstream
boundary
U

boun
"_ Hub

Figure 2. - Solution region.


nonlinear but can be solved iteratively by the finite-difference method when the flow is
completely subsonic.
A finite region (as indicated in fig. 2) is considered for the solution of equa-
tion (A1). It is assumed that the upstream and downstream boundaries are sufficiently
far from the blade so as to have a negligible effect on the solution. Equation (A1) is
elliptic for subsonic flow. Therefore, when the flow is entirely subsonic, equation (A1)
can be solved when proper boundary conditions are specified on the entire boundary of
the region. These conditions are the values of the stream function or its normal de-
rivative on all four boundaries. The stream function is zero on the hub and 1 at the

shroud. On the upstream and downstream boundaries, the derivative of the stream
function normal to the boundary is assumed to be zero. This is equivalent to assuming
that the upstream and downstream boundaries are orthogonal to the streamlines (as-
sumption 8, p. 5).
The numerical solution of equation (A1) is obtained by the finite-difference method.
A grid must be used for the finite-difference equations. The type of grid that was
chosen is an orthogonal mesh that is generated by the program. The method used to
generate the grid follows that reported in reference 7. The space between the hub and
the shroud is divided into equal increments along several hub-shroud lines. Spline
curves are then fit through the resulting points to obtain the streamwise orthogonals
(fig. 3). The normal orthogonals are obtained by a predictor-corrector technique.
This technique is analogous to the second-order Rungo-Kutta method for solving ordi-
nary differential equations, which is known also as the improved Euler method or
Heun's method (ref. 8). The distances along orthogonal mesh lines are s and t. The
s-distance is in the streamwise direction and the t-distance is normal to this, as indi-
cated in figure 3. With the mesh determined, the finite--difference equations on the
orthogonal mesh are given in part II of this report (ref. 6). Note that the orthogonal
mesh is not used as a coordinate system.

_- Streamwise or "horizontal"
orthogonals

Fiqu re 3. - Orthogonal fi nite-difference mesh on solution region.


The finite-difference equations are nonlinear since the original equation (A1) is non-
linear. These equations can be solved iteratively. On the first iteration an initial den-
sity is assumed; this linearizes some of the terms. The remaining nonlinear terms are
omitted for the first iteration so that the finite--difference equations are entirely linear-
ized. These linearized equations are then solved to obtain the first approximate solu-
tion for the stream function. This solution provides information that is used to obtain a
better estimate of the density and an estimate of the other nonlinear terms. The equa-
tions are then solved again to obtain an improved solution. This process is repeated,
and by iteration a final converged solution can be obtained if the flow is subsonic.
For each step of this iteration the linearized finite--difference equations must be
solved. The method used to solve the equations is successive overrelaxation (ref. 9)
with an optimum overrelaxation factor. Since this is also an iterative method, we have
two levels of iteration. The overrelaxation is performed in the "inner iteration," and
the corrections to the nonlinear terms are made in the "outer iteration." The inner
iteration is internal to the program, so the only iteration apparent to the user is the
outer iteration. When an iteration is mentioned in this report, it is always the outer
iteration.

After the stream function is obtained, the velocity distribution is found by numeri-
cal partial differentiation of the stream function and by using equations (A5) and (A6).
The details of the numerical procedure and programming technique are described in
part II (ref. 6).

TRANSONIC VELOCITY-GRADIENT APPROXIMATE SOLUTION

For the case where there is locally supersonic flow, equation (A1) is no longer
elliptic in the entire region but is hyperbolic in the region of supersonic flow (ref. 10).
This changes the boundary conditions and means that there will probably be shock
losses in going from supersonic to subsonic flow. The finite-difference method cannot
be used with locally supersonic flow. However, an approximate solution can be ob-
tained by getting a reduced-flow solution with the finite-difference method and extending
this to the full flow by using the velocity-gradient method. This technique is described
in reference 3.

The velocity-gradient equations are equations (A7) to (All). Equation (A7) is


solved as an initial-value problem, where the velocity W is specified at the hub for
any given vertical mesh line running from hub to tip. By finding several solutions for
varying values of W at the hub, a solution satisfying the specified mass flow
(eq. (A12)) will be found. When equation (A7) has been solved, subject to giving the

8
correct mass flow, for every hub-shroud mesh line in the region, the entire velocity
distribution at full mass flow has been obtained.

BLADE SURFACEVELOCITIES

The solution that is obtained by either the finite--difference or velocity-gradient


method is for the midchannel surface betweenthe blades. Whenthere are blades, the
blade surface velocities are of greater interest. These can be estimated since the
blade loading is dependenton the rate of changeof whirl. By assuming a linear varia-
tion of velocity betweenblade surfaces, equation (A13) can be derived for calculating
the blade surface velocities.

APPLICATION OF PROGRAM

The program can be used both for analysis and as a design tool. When it is used
for design, other programs should be used with this program. For axial compressors,
reference 11 describes a program that will give blade mean-camber-line coordinates
and thicknesses for an axial compressor blade. This blade design can be checked by
using the MERIDL program to analyze the flow distribution in detail. Usually, changes
must be made to the blade design to achieve a desirable flow distribution. These
changes may involve more than just the blade shape; for example, hub and shroud pro-
file, inlet and outlet whirl distribution, and loss distribution may have to be changed.
Of course, the accuracy of the MERIDL solution depends on the accuracy of the bound-
ary conditions used.
When a reasonable flow pattern is achieved by the MERIDL flow analysis on the
midcharmel stream surface, more-detailed blade surface velocities can be obtained by
flow analyses on various blade--to-blade stream surfaces. A useful program for this
purpose is TSONIC (ref. 3). Most of the information required to compute input for
TSONIC is calculated and printed directly by MERIDL. Item 12 of the section Printed
Output (p. 50) explains how to obtain and use this output from MERIDL. Further
changes in blade shape or whirl distribution may be considered at this time. Refer-
ence 12 (ch. VII) gives information on incidence and deviation for good designs and for
off-design conditions.
For cases where the flow is well guided in the channel but has large variations,
both from blade to blade and from hub to shroud, the CHANEL program (ref. 13) is
useful. The CHANEL program obtains a solution on a channel cross-section surface.

9
The CHANEL program is particularly useful for calculating choking mass flow through
a blade row.

DESCRIPTION OF INPUT AND OUTPUT

The principal block of input required for the program is a geometric description of
the blade row to be analyzed. The geometry is given in cylindrical coordinates, as
shown in figure 4. The blade shape is given as a series of blade sections from hub to
shroud. Each blade section is described by a set of z and r mean-camber-line
coordinates and two sets of blade-shape coordinates on any general, smooth surface of

w;
w2.W m+

Figure 4. - Cylindrical coordinate system and velocity components.

10
revolution. Other inputs include upstream and downstream flow conditions, a geo-
metric description of hub and shroud, appropriate gas constants, operating conditions
such as mass flow and rotational speed, and a description of the finite-difference solu-
tion mesh. There are also input variables to indicate the desired blocks of output data.
Ouptut is given at any or all of three principal geometric locations: (1) on all mesh
points of the orthogonal solution mesh, (2) along user-designated streamlines through
the blade row, or (3) along user-designated lines from hub to shroud (station lines}.
Output consists mainly of z- and r-coordinates and velocity components and angles.
At station lines more output is given, including density, temperature, and pressure.

INPUT

Figure 5 shows the input variables as they are punched on the data cards. The first
input data card is for a title, which serves for problem identification. Any information
may be put in the 80 columns of this card.
All the numbers on the three input cards beginning with MBI, LSFR, and IMESH are
integers (no decimal point) in a five-column field (fig. 5}. These must be all right ad-
justed. The input variables on all other data cards are real numbers (punch decimal
point} in 10-column fields.
Figure 5 indicates that several options exist for specifying upstream and down-
stream flow conditions. First, the user can specify these conditions either as a func-
tion of the stream function {SFIN and SFOUT) or as a function of radius (RADIN and
RADOUT). Next, the user can specify either whirl (LAMIN and LAMOUT} or absolute
tangential velocity (VTHIN and VTHOUT}. At the downstream input station either abso-
lute total pressure (PROP} or fractional loss of total pressure (LOSOUT} may be given.
Finally, a loss distribution (PERLOS} within the blade may be specified. If the loss
distribution through the blade is not given, the loss is assumed to vary linearly between
the leading and trailing edges.
Input variables are both geometric and nongeometric. The geometric input varia-
bles are shown in figures 6 to 10. Further information concerning the input variables
is given in the section Special Instructions for Preparing Input (pp. 25 to 31).

11
12
REPRODUCIBILITY OF THE
ORIGINAL PAGE IS POOR
Z

o
._="

m m

o 8
A

.-J
J

Z
Z
0
o

Q_

Z
Z

A
o
o Z
v
._J i
0
Z

I.--

o
o

.-.I
.--J
Z Z
Z

o o
A
A
m m _ m _ m
/ _ m
I_ m I---
0
0
._J
Z
0
z Z
o 0 0
Z 0
Z Z Z
Z
Q _ _ Q
I--- A wA 1 _A

0 0
....1 I-- I-- ._J
I:_ C.)
V'I I_

N _ _ _ N
I

saP_lq 0u aJe a JeLl} _! spJe:_ asaq} l!tuO

13
[i] Hub or shroud spline point
0 Hub point where orthogonal
NBLPL = 6 -_ mesh begins, ends, or

E3 _ changes spacing
i - _ ONTIP=7

L I \ ,' section
I I.... -1' input

RI"_P L _Jz_ t 'see fig" lO) _Zo#oHuUTB=ll

Figure 6. - Input variables -'hub, shroud, and blade sections.

..,- ZTOUT 1

OUT = 5

-, T I 1 _FSFOUT or RADOUT |

SFo[_RADINI ? I / -_-(_ _PROP or LOSOUT


T';' _u,m[ _ I / ' _LLAMOUT or VTHOUT

LAMIN I 71 I _-'-zoMom-
or l] I
" I J / _ RTOUT

) I_ RADOUTI
ZOMIN-,,./
,.OUT
"-_ZHIN _ RADIN -_
= R_IN I ZHOUT-- , '
Z

Figure 7. - Input variables - upstream and downstream flow variables.

14
0 Hub point where orthogonal mesh begins.
ends. or changes spacing

MHT = 9
J f

8
,- Downstream
7 boundary of
-4-
orthogonal
mesh

Upstream boundary I
of ortho, )nal mesh--, MM = 20

IMBI = 5 ROMOUT
ROMBO
-_--- 70MI N-_
-.,, -_ZOMB
,._ ÷ ROMB! ZOMBO
t ZOMOUT_
I=Z
ROMIN_ _
Figure 8. - Input variables - orthogonal mesh.

= ZTST
NSL = 8

FLFR(8I = 1.0 I I

FLFR(T) II I
I I I
FLFR(6) 1 1 I
L E
I I I
I t _--I-_
I I

FLFR 131 I I

RTST
FLFR(2)_ I I

FLFR(1) - O. "__RHST

= ZHST =

Figure 9. - Input variables - locating streamlines and station lines for output.

15
E
u

A
E

|
m

A
u

c:l_
ql
.g
_4

16
Input Dictionary

The input variables are described in terms of a consistent set of SI units: newtons,
kilograms, meters, joules, kelvins, and seconds. The program, however, will run
with input in any consistent set of units.
The input variables, in the order that they appear in figure 5, are the following:
GAM Specific-heat ratio, y
AR Gas constant, J/(kg) (K)
MSFL Total mass flow through entire circumferential annulusof blade
row, kg/sec
OMEGA Rotational speed, w, rad/sec. Note that w is negative if rotation
is in opposite direction of that shown in fig. 4.

REDFAC Factor by which mass flow (MSFL) must be reduced in order to


assure subsonic flow throughout flow passage. REDFAC may be
left blank, in which case a value of 1.0 will be used. See sec-
tion (k), p. 31.

VELTOL Convergence tolerance on maximum velocity change in each iter-


ation, over all mesh points, for reduced-mass-flow solution.
VELTOL may be left blank, and a value of 0.01 will be used by
the program. Whatever value is given is multiplied by the mini-
mum of FNEW and DNEW before it is printed and subsequently
used by the program. A value of 0. 001 for VELTOL is a tight
tolerance, 0.01 is a medium tolerance, and 0.1 is a loose toler-
ance.

FNEW Damping factor on calculation of Ft from iteration to iteration.


A value of 0.5 is suggested for FNEW. FNEW may be left blank,
in which case the program will use a value of 0.5. See sec-
tion (b), p. 26.

DNEW Damping factor on calculation of a(rV0)/at within blade row from


iteration to iteration. DNEW may be left blank, and the program
will use a value of 0.5. DNEW does not have to be equal to

FNEW. See section (b), p. 26.


MBI Number of vertical mesh lines from left boundary of orthogonal
mesh (ZOMIN) to point of first mesh-size change (ZOMBI). Use
MBI = 0 if there are no blades. See fig. 8 and section (d), p. 27.

17
MBO Total number of vertical mesh lines from left boundary of orthogo-

nal mesh (ZOMIN) to point of second mesh-size change (ZOMBO).

Use MBO = 0 if there are no blades. See fig. 8 and section (d),

p. 27.

MM Total number of vertical mesh lines from left to right boundaries

of orthogonal mesh (ZOMIN to ZOMOUT), maximum of I00. See

fig. 8 and section (d), p. 27.

MHT Total number of horizontal mesh spaces from hub to shroud of or-

thogonal mesh, maximum of 100. See fig. 8 and section (d), p. 27.

NBL Number of blades in total circumference of blade row. Use

NBL = 1 if there are no blades.

N-HUB Number of spline points given in ZHUB and RIqUB arrays, maxi-

mum of 50. See fig. 6 and section (c), p. 26.

NTIP Number of spline points given in ZTIP and RTIP arrays, maximum

of 50. See fig. 6 and section (c), p. 26.

NIN Number of data points given in upstream arrays of flow properties

(SFIN, RADIN, TIP, PRIP, LAMIN, VTHIN), maximum of 50.

See fig. 7 and section (e), p. 27.

NOUT Number of data points given in downstream arrays of flow proper-

ties (SFOUT, RADOUT, PROP, LOSOUT, LAMOUT, VTHOUT),

maximum of 50. See fig. 7 and section (e), p. 27.

NBLPL Number of blade planes or blade sections on which data (ZBL,

RBL, etc. ) are given to describe mean blade shape and blade

thickness, maximum of 50. See fig. 6 and section (f),p. 28.

NPPP Number of data points per blade section or blade plane in ZBL,

RBL, etc., arrays, maximum of 50. See fig. 10 and sec-

tion (f),p. 28.

NOS TA T Number of stations from hub to shroud (located by coordinates in

ZHST and ZTST) at which output is desired, maximum of 50.

See fig. 9 and section (h), p. 29. NOSTAT may be left blank, in
which case no input cards should be included for ZHST, ZTST,

RI-IST, and RTST arrays.

NSL Number of streamlines from hub to shroud (designated by values

in FLFR) at which output is desired, maximum of 50. See fig. 9.

NSL may be left blank, in which case no cards should be included

18
for FLFR array. If NSL is left blank, theprogram will set it
equal to 11 and print requested streamline output on 11 stream-
lines that vary by 10 percent of total flow (i. e., 0, 10, 20, ...,
100 percent).

NLOSS Number of points at which fractional loss (PERLOS) is specified,


maximum of 50. If NLOSS is zero or left blank, the program will
use a linear loss distribution in the streamwise direction. In this

case, no input cards should be included for PERCRD or PERLOS


arrays.

LSFR Integer (0 or 1) indicating whether upstream and downstream flow con-


ditions are given as a function of stream function (0) or radius (1).

LTPL Integer (0 or 1) indicating whether downstream total pressure (0)


or fractional loss of stagnation pressure (1) is given as input.

LAMVT Integer (0 or 1) Indicating whether upstream and downstream whirl


(0) or tangential velocity (1) is given as input.

LROT Integer (0 or 1) indicating whether coordinate rotation (ANGROT)


is necessary (1) for hub and shroud spline fit curves. See sec-
tion (/), p. 31.

LBLAD Integer (0, 1, or 2) indicating which two blade--shape coordinates


are given as input. If LBLAD = 0, mean blade shape O (THBL)

and normal thickness t n (TNBL) are given. If LBLAD = 1,


mean blade shape 0 (THBL) and tangential 0-thickness t0/r
(TTBL) are given. If LBLAD = 2, upper blade surface 01
(TH1BL) and lower blade surface 02 (TH2BL) are given. See
fig. 10 and section (f), p. 28.

LETEAN Integer (0 or 1) indicating whether leading- and trailing-edge mean

blade--shape angles file and /_te (BETALE and BETATE) are


specified (1). See fig. 10 and section (f), p. 28.

A NGROT Rotation angle of axis used for all streamwise spline fit curves,
deg. Omit this card if LROT is (0). See section (l), P. 31.

ZOMIN z-Coordinate of intersection of left boundary of orthogonal mesh


with hub profile, m. See figs. 6 and 8 and section (d), p. 27.

ZOMBI z-Coordinate of intersection of vertical mesh line with hub profile


where first change in mesh spacing occurs (MBI), m. See figs.
6 to 8 and section (d), p. 27. Set ZOMBI and ZOMBO to zero
for no blades case.

19
ZOMBO z-Coordinate of intersection of vertical mesh line with hubpro-
file where secondchangein mesh spacing occurs (MBO), m.
Seefigs. 6 and8 and section (d), p. 27.
ZOMOUT z-Coordinate of intersection of right boundary of orthogonal mesh
(MM)with hub profile, m. Seefigs. 6 and8 and section(d), p. 27.
ItOMIN, ROMBI, r-Coordinates corresponding to ZOMIN, ZOMBI, ZOMBO, and
IIOMBO, and ZOMOUT, m. Leave these spacesblank if LROT = 0.
ROMOUT
ZI[ [,'13 Array of z-coordinates of input points defining hub or bottom
boundary of flow channel, m. See fig. 6 and section (c), p. 26.

Illt UB Array of r-coordinates of input points defining hub or bottom


boundary of flow channel, m. See fig. 6 and section (c), p. 26.

ZTIP Array of z-coordinates of input points defining shroud or top


boundary of flow channel, m. See fig. 6 and section (c), p. 26.

tlTIP Array of r-coordinates of input points defining shroud or top


boundary of flow channel, m. See fig. 6 and section (c), p. 26.

ZttIN z-Coordinate of intersection with hub profile of line on which up-


stream flow conditions are given, m. Leave this entire card
blank if LSFR=0, LAMVT=0, andMBI¢0. See fig. 7 and
section (e), p. 27.

ZTIN z-Coordinate of intersection with shroud profile of line on which


upstream flow conditions are given, m. See fig. 7 and sec-
tion (e), p. 27.

IttlI N, RTIN r-Coordinates corresponding to ZHIN and ZTIN, m. Leave these


spaces blank if LROT = 0. If RADIN is given as input
(LSFR = 1), RHIN and RTIN cannot be equal to each other.

SFIN Array of values of stream function for input points from hub to
shroud along line on which upstream flow conditions are given.
SFIN is given when LSFR = 0. See fig. 7 and section (e), p. 27.

RADIN Array of r-coordinates of input points along line from hub to


shroud on which upstream flow conditions are given, m. RADIN
is given when LSFR = 1. See fig. 7 and section (e), p. 27.

TIP Array of absolute total temperatures T!1 at input points along line
from hub to shroud on which upstream flow conditions are
given, K. See fig. 7 and section (e), p. 27

2O
PRIP Array of absolute total pressures p_ at input points along line
from hub to shroud on which upstream flow conditions are given,
N/m 2. See fig. 7 and section (e), p. 27.

LAMIN Array of values of absolute whirl (rV,_ at input points along line
\ v/ i
from hub to shroud on which upstream flow conditions are given,
m2/sec. LAMIN is given when LAMVT = 0. See fig. 7 and sec-
tion (e), p. 27.

VTHIN Array of values of absolute tangential velocity (V_] at input


\t,/ i
points along line from hub to shroud on which upstream flow con-
ditions are given, m/sec. VTHIN is given when LAMVT = 1.
See fig. 7 and section (e), p. 27.

ZHOUT z-Coordinate of intersection with hub profile of line on which


downstream flow conditions are given, m. Leave this entire
card blank if LSFR=0, LAMVT=0, andMBI_0. See fig. 7
and section (e), p. 27.

ZTOUT z-Coordinate of intersection with shroud profile of line on which


downstream flow conditions are given, m. See fig. 7 and sec-
tion (e), p. 27.

RHOUT, RTOUT r-Coordinates corresponding to ZHOUT and ZTOUT, m. Leave


these spaces blank if LROT = 0. If RADOUT is given as input
(LSFR = 1), RHOUT and RTOUT cannot be equal to each other.

SFOUT Array of values of stream function for input points from hub to
shroud along line on which downstream flow conditions are given.
SFOUT is given when LSFR = 0. See fig. 7 and section (e), p. 27.

RADOUT Array of r-coordinates of input points along line from hub to


shroud on which downstream flow conditions are given, m.
RADOUT is given when LSFR= 1. See fig. 7 and section (e), p. 27.

PROP Array of absolute total pressures p_ at input points along line


from hub to shroud on which downstream flow conditions are
given, N/m 2. PROP is given when LTPL = 0. See fig. 7 and

section (e), p. 27.

LOSOUT Array of fraction of absolute totalpressure loss (p'


o, id - p_)/
p'
o, id at input points along line from hub to shroud on which
downstream flow conditions are given. LOSOUT is given when
LTPL = 1. See fig. 7 and section (e), p. 27.

21
For the case without blades, omit input cards for the variables LAMOUT through
BETATE, inclusive.

LAMOUT Array of values of absolute whirl (r)V0 o atinputpointsalongline


from hub to shroud on which downstream flow conditions are
given, m2/sec. LAMOUT is given when LAMVT = 0. See fig. 7
and section (e), p. 27.

VTHOUT Array of values of absolute tangential velocity (V0),, at input

points along line from hub to shroud on which downstream flow


conditions are given, m/sec. VTHOUT is given when
LAMVT = 1. See fig. 7 and section (e), p. 27.

ZBL Two-dimensional array of z-coordinates of points describing


mean blade surface, m. See figs. 6 and 10 and section (f),
p. 28. This surface is described by a series (from 2 to 50) of
blade sections from hub to shroud. The hub section is given
first, followed by successive sections up to the shroud.

RBL Two-dimensional array of r-coordinates, corresponding to ZBL,


of points describing mean blade surface, m. See fig. 6 and sec-
tion (f), p. 28.

THBL Two-dimensional array of 0-coordinates, corresponding to ZBL,


of points describing mean blade surface, rad. See fig. 10 and
section (f), p. 28. The tangential coordinate 0 is positive in
direction of positive rotation (fig. 4). The origin of
0-coordinates can be anywhere around the circumference.
THBL is given only when LBLAD is 0 or 1.

TNBL Two-dimensional array of blade thicknesses normal to mean cam-


ber line, corresponding to ZBL, RBL coordinates, m. When
there is little blade lean, small blade surface curvatures, and
near-parallel suction and pressure blade surfaces, it is feasible
to use the blade normal thicknesses for TNBL array (LBLAD=0).
Otherwise it is recommended to give either the tangential thick-
ness, TTBL (LBLAD = 1), or the suction and pressure blade sur-
face coordinates, THIBL and TH2BL (LBLAD = 2). See fig. 10(a)
and section (f), p. 28.

TTBL Two-dimensional array of blade tangential (circumferential)


thicknesses, corresponding to ZBL, RBL coordinates, rad.
TTBL is the blade tangential thickness in meters, divided by

22
RBL. TTBL is given only whe_ LBLAD = 1. See fig. 10(b) and
section (f), p. 28.

THIBL Two-dimensional array of upper blade surface coordinates (fig.


(10(c)) corresponding to ZBL, RBL coordinates, rad. TH1BL is
given only when LBLAD = 2. See fig. 10(c) and section (f), p. 28.

TH2BL Two-dimensional array of lower blade surface coordinates (fig.


(10(c)) corresponding to ZBL, RBL coordinates, rad. TH2BL is
given only when LBLAD = 2. See fig. 10(c) and section (f), p. 28.

BETALE Array of leading-edge, mean-camber-line tangency angle, deg.


BETALE is given only when LETEAN = 1; omit this card if
LETEAN= 0. See fig. 10(d) and section (f), p. 28.

BETATE Array of trailing-edge, mean-camber-line tangency angles, deg.


BETATE is given only when LETEAN = 1; omit this card if
LETEAN= 0. See fig. 10(d) and section (f), p. 28.

ZHST Array of z-coordinates of intersections of hub-shroud output sta-


tion lines with hub profile, m. Omit this card if NOSTAT = 0.
See fig. 9 and section (h), p. 29.

RttST Array of r-coordinates corresponding to ZHST array. Omit this


cardif NOSTAT=0or if LROT=0. See fig. 9and section (h), p. 29.

ZTST Array of z-coordinates of intersections of hub-shroud output sta-


tion lines with shroud profile, m. Omit this card if NOSTAT = 0.
See fig. 9 and section (h), p. 29.

RTST Array of r-coordinates corresponding to ZTST array. Omit this


cardif NOSTAT= 0or if LROT=0. Seefig. 9and section (h), p. 29.

FLFR Array of values of stream function designating streamlines along


which output is to be printed. Omit this card if NSL = 0. See
fig. 9 and section (h), p. 29. If no cards are given for FLFR
array (i. e., NSL = 0), the program will automatically assign the
followingll values to FLFR: 0.0, 0.1, 0.2, ..., 0.9, 1.0.

PERCRD Array of streamwise fractional distances corresponding to the


PERLOS array. Omit this card if NLOSS = 0. PERCRD should
have 0. and 1. for the first and last values, respectively. When
there are blades, PERCRD is the fractional distance between the
leading and trailing edges. When there are no blades, PERCRD
is the fractional distance between the upstream and downstream
input stations.

23
PERLOS Array of fractional loss distribution in the streamwise direction
within the blade, or between upstream and downstream input sta-
tions when there are no blades. Omit this card if NLOSS= 0.
PERLOS should have increasing values from 0.1 to 1., although
the final value need not reach 1.

The remaining seven integer variables, beginning with IMESH, are used to indicate
what output is desired. The program reduces the input mass flow by REDFAC and
solves the resulting problem iteratively with a stream-function analysis. The results
of this analysis are then used with full mass flow to obtain a transonic analysis by
velocity-gradient methods. Output can be obtained after each reduced-mass-flow iter-
ation and also after the final transonic solution.

For all these variables except ISUPER, the integer given indicates the multiple of
iterations at which the user wishes the output associated with the variables to be
printed or plotted. A zero in any of these indicates that the output associated with that
variable is to be omitted. A 1 will print or plot the output on every iteration. A 3, for
example, would give the output on every third iteration. Any nonzero integer will cause
output to be given on the first and last iterations and after the transonic solution, in
addition to the output called for at other iterations. ITSON, however, does not give
output on the first iteration. A large integer will obviously give only the first iteration
and the final converged iteration. Care should be used not to call for more output than
is really useful. The following list gives the output associated with each of these
variables:

IMESH Major output at every mesh point of the orthogonal mesh, that is, mesh
point indices and coordinates, stream-function value, relative velocity
components and magnitude, critical velocity ratio, and flow angles. This
output is usually not requested, except for debugging purposes, since
output called for by ISLINE or ISTATL is preferable in most cases.

ISLINE Major output along streamlines (indicated by values in FLFR) at each point
where streamlines are crossed by vertical mesh lines of the orthogonal
mesh. Output includes z, r, and m streamline coordinates, relative
velocity components and magnitude, critical velocity ratio, flow angles,
and streamline curvature. Where the streamline passes within the blade
region, an estimate of suction- and pressure--surface velocities is also
printed.

ISTATL Major output along station lines from hub to shroud, at locations specified
by ZHST and ZTST arrays and at values of stream function specified by
FLFR array. Output corresponds to that given for ISLINE with the addi-
tion that absolute velocity components, as well as density, pressure, and

24
temperature for static conditions and for absolute and relative stagnation
conditions are printed. ISTATL should be zero if NOSTAT is zero.
1-PLOT Plotting indicator requesting output to be plotted on microfilm. Any non-
zero value in IPLOT will causethe input data and generatedorthogonal
mesh to be plotted. Also, at each iteration that is a multiple of IPLOT,
streamlines will be plotted and meridional and surface velocities will be
plotted for each streamline value. These will also be plotted after the
final transonic solution.
ISUPER Integer (0 or 1) indicating which solution (subsonicor supersonic) of the
velocity-gradient equation is desired. If ISUPER= 0, only the subsonic
solution will be printed. If ISUPE R = 1, both subsonic and supersonic
solutions will be printed.

ITSON Integer indicating when information is desired for use in calculating input
for the TSONIC program (ref. 3). If ITSON = 0, no information will be
given for TSONIC. Otherwise. information will be listed with other out--
put. Usually, the user only wants TSONIC data after subsonic conver-
gence is reached. Using a high value for ITSON (i. e., > 20) will achieve
this result.

IDEBUG Integer indicating whether additional debug output is desired. If


IDEBUG= 0, no extra output is printed. If IDEBUG > 0, blade geometry
information on the alternate mesh and the coefficients of the finite-
difference equations, as well as 21 arrays of debug output on the or-
thogonal mesh, are printed. Fifteen of these arrays change and are re--
printed after each iteration of the reduced-mass--flow solution that is a
multiple of IDEBUG.

Special Instructions for Preparing Input

It is unusual to have no errors in input to MERIDL the first time any new data set is
run. Therefore, input should be checked thoroughly before it is submitted. Errors are

commonly made for the following reasons: inconsistent units; improper sign on w, V 0,
or whirl; input for arrays not agreeing with the input bounds for those arrays; and up-
stream and downstream input not being of the form specified by LSFR, LTPL, and
LAMVT. Also geometric input into the hub and shroud arrays and the blade-geometry
arrays should be smooth enough that the hub, shroud, and blade sections will be fit well
with cubic spline curves (see section (g), p. 29). These input geometric arrays are all
plotted on microfilm by the program after spline fitting is completed; the microfilm

25
output will indicate whether or not the input was smooth. All output should be checked,
especially that from a new input data set, to see if it is reasonable.
(a) Units of measurement. - The International System of Units (ref. 14) is used
throughout this report. However, the program does not use any constants that depend
on the system of units being used. Therefore, any consistent set of units may be used
in preparing input for the program. For example, if force, length, temperature, and
time are chosen independently, mass units are obtained from Force = Mass × Acceler-
ation. The gas constant R must then have the units of (Force x Length)/(Mass x
Temperature). Density is mass per unit volume, and mass flow is mass per unit time.
Output then gives velocity in the chosen units of length per unit time. Since any consis-
tent set of units can be employed, the output is not labeled with any units.
(b) Dampin_ factors FNEW and DNEW. - The input variables FNEW and DNEW are
used as damping factors on the Ft and 0(rV0)/0t terms of equation (A1). During each
iteration of the subsonic or finite-difference portion of the program, new values for Ft
and a(rV0)/0t are calculated. The calculated changes to these terms are often so
large that, if the full change was accepted on each iteration, the solution would diverge.
Using values of FNEW and DNEW less than 1.0 allows only a portion of these changes to

be used. The value of FNEW or DNEW is the fraction of the predicted change in F t or
O(rV0)/_t to be added to the previous values of these variables. Therefore, if Ft or
(rV0)/0t does not converge, FNEW or DNEW should be reduced. However, this will
reduce the rate of convergence so that caution should be used that FNEW and DNEW are
not made too small. Because of the reduced rate of convergence, VELTOL is auto-
matically reduced to obtain the same accuracy as when FNEW and DNEW are 1.0.
Limited experience has shown that for subsonic flows, where the gradients of flow
variables are not large, converged solutions can be obtained with FNEW and DNEW
equal to 1.0. For most practical flows, however, this is not the case and some damp-
ing is required. Values of FNEW and DNEW in the range of 0.5 are most commonly
used, although at times values as low as 0.05 have been used. (Note that FNEW and
DNEW do not have to be equal to each other.)
The user will have to gain experience with FNEW and DNEW before he can use
them effectively to maximize rates of convergence. Maximum and minimum values and

the maximum change for Ft and 0(rV0)/at (DVTHDR) are printed with theoutput on
each iteration. The maximum relative change in velocity at any point is also printed,
it should gradually approach VELTOL for convergence. By observing these values, the
user can learn the effects of changes in FNEW and ]:)NEW for different runs of the pro-
gram.
tc) Hub and shroud flow-channel _eometry. - The hub and shroud flow-channel
geometry is specified in the ZHUB, RHUB and ZTIP, RTIP arrays. Both of these curves
must have the same z-origin (typically the blade leading edge at the hub). These two

26
arrays must extendfar enoughupstream and downstream to cover the upstream and
downstream boundaries of the orthogonal mesh, as well as the upstream and down-
stream input stations where streamflow data are given. If they do not extend this far,
they will be linearly extrapolated and an incorrect flow channel may result. Relatively
few points are needed to describe these smooth surfaces (2 to 10 is a typical range for
N-HUB and NTIP) in order to have the program calculate smooth, accurate spline fits of
these surfaces (fig. 6).
If the user knows the amount of the boundary-layer blockage along the hub and
shroud profiles, he should revise the hub and shroud arrays to include this blockage
effect. In this situation the output will represent a more realistic flow condition.
/d) Ortho_onal mesh. - The number of orthogonal mesh lines is specified by MBI,
MBO, MM, and MHT; and the positioning of the mesh is specified by ZOMIN, ZOMBI,
ZOMBO, and ZOMOUT. When ANGROT is specified (LROT = 1), ROMIN, ROMBI,
ROMBO, and ROMOUT must also be specified. The four z-coordinates, all located on
the hub, must use the same z-origin as all other geometric input. The ZOMBI and
ZOMBO locations are usually close to the leading and trailing edges of the blade at the
hub, although they do not have to correspond to these locations (fig. 8). Mesh-size
spacing in the horizontal direction is established by the relation between the four
z-coordinates (ZOMIN, etc.) and the number of mesh spaces requested. For instance,
MBO - MBI evenly spaced mesh spaces along the hub will be located in the distance
given by ZOMBO - ZOMBI. Mesh spacing in the vertical direction is determined by
locating MHT spaces in the hub--shroud distance. Usually, none of the vertical mesh
lines follow the blade leading- or trailing-edge lines since the blade edges are not
usually orthogonal to the hub or the shroud (fig. 8). So some vertical mesh lines will
pass in or out of the blade region as they progress from hub to shroud. The suggested
number of mesh lines between ZOMBI and ZOMBO that are required to cover the blade
adequately is 15 to 30, depending on blade geometry.
(e) Upstream and downstream flow conditions. - Upstream flow conditions are
given in the SFIN (or RADIN), TIP, PRIP, and LAMIN (or VTHIN) arrays, which are all
of length NIN. Downstream conditions are given in the SFOUT (or RADOUT), PROP
(or LOSOUT), and LAMOUT (or VTHOUT) arrays, which are all of length NOUT
(fig. 7). Upstream and downstream flow conditions are used, along with the assump-
tion of conservation of angular momentum along streamlines, to establish the flow field
in the upstream and downstream portions of the solution region:
Ordinarily, the upstream and downstream flow conditions are given as a function of
stream function (SFIN and SFOUT). In this case these input values apply at all points
along streamlines upstream or downstream of the blade. If, in addition, the whirl is
specified (LAMVT = 0) and there are blades (MBI _ 0), the ZHIN, ZTIN and
ZHOUT, ZTOUT inputs are superfluous, so that these variables need not be specified.

27
However, in most cases, legitimate values must be supplied for ZHIN, ZTIN and
ZHOUT,ZTOUT. In this case the upstream conditions are given on a straight line that
passes through the two points given by ZHIN on the hub and ZTIN on the shroud. Down--
stream conditions are given on a straight line that passes through ZHOUTon the hub
and ZTOUT on the shroud. Theselines may lie anywhere in the regions from the blade
edges upstream and downstream to the boundaries of the orthogonal mesh (figs. 7 and8).
If LROT is 1, values for RHIN, RTIN, RHOUT, and RTOUT must also be given as input.
The arrays of upstream and downstream input do not necessarily have to extend all
the way from the hub to the shroud or lie on radial lines. They will be linearly extrap-
olated to the hub and the shroud, if necessary, by the program, should the user only
give data in a portion of the flow channel.
(t) Mean blade surface and thickness coordinates. - The blade shape is described
from hub to shroud by four arrays, all of which are two dimensional (figs. 6 and 10).
Each of these arrays has NBLPL blade sections or planes, with NPPP points in each of
these sections. When giving data for each of these four arrays, start each new section
of data (NPPP points) at the beginning of a new card. There does not have to be any
geometric relation between analogous points on adjacent blade sections. All the ZBL
data for all the sections are given, followed by all the RBL data, etc. The origin for
the z-coordinates of ZBL should be the same as that used for all other z-coordinate in-

put arrays. Of the four arrays necessary to describe the blade shape, ZBL and RBL
are always given. There are three options, controlled by LBLAD, for the other two
arrays (see LBLAD definition in the input dictionary).
When LBLAD is 0 or 1, the THBL array must be given (figs. 10(a) and (b)). The
THBL array is for input 0-coordinates of the blade mean camber surface. Alterna-
tively, the midchannel stream surface, obtained from a blade--to-blade solution (ref. 3),
may be used for the THBL array. The program provides a solution on the midchannel
stream surface. It obtains that stream surface by smoothly fairing the inlet and out-
let flow angles onto the input blade surface at an appropriate distance back on the blades
from the leading and trailing edges. This distance is a function of solidity (appendix F).
The TNBL array is given when LBLAD = 0 (fig. 10(a)). The TNBL array is for in-
put blade thicknesses normal to the blade--section mean camber line and lies on a sur-
face of revolution cutting through the blade (figs. 6 and 10). So, in general, the thick-
nesses lie on a curved line whose ends may be at different radii and may or may not be
normal to the blade surfaces (fig. 10(a)). Because it is difficult to obtain proper block-
age or proper TSONIC input, caution should be exercised when giving TNBL input. Al-
so, when TNBL is given as input, the THBL coordinates must be for the blade mean
camber surface and not the midchannel stream surface.
The TTBL array is used for input tangential (circumferential) blade thicknesses
(when LBLAD = 1) and is given in radians (i. e., t0/r) (fig. 10(b)). Thick blades,

28
blades with high curvature, or blades with significant lean should use TTBL input in
preference to TNBL input.
The TH1BL and TH2BL arrays are used to give blade surface 0-coordinates (when
LBLAD = 2) (fig. 10(c)). This input is similar to that used in the TSONIC program
(ref. 3).
To obtain a smooth solution near the leading and trailing edges, it is recommended
to use nonzero values of leading-edge and trailing-edge thicknesses, as figure 10 indi-
cates.
In addition to the four arrays necessary to describe the blade, there are two one-
dimensional arrays (BETALE and BETATE) that are optional. These arrays specify
the leading-edge and trailing-edge mean-camber-line tangency angle in degrees. The
angles are measured on the input blade-section surfaces (fig. 1 0(d)). Use LETEAN = 1
when BETALE and BETATE are given, and LETEAN = 0 when they are not given.
The first blade section given at the hub or the last one at the blade tip does not
necessarily have to conform to the hub or shroud profile. It can be given within the
flow region, crossing the boundary, or completely outside the boundary (fig. 6). Ex-
trapolation or interpolation will be used when necessary to obtain blade data where the
blades meet the hub and shroud profiles.
The user should attempt to give smooth data for these four blade-shape arrays.
The program makes use of spline fits of these curves to compute gradients that are
used throughout the program. The mean-camber-line first and second derivatives, as
well as microfilm plots of the blade sections, given as output will indicate whether input
data for these arrays were acceptable.
(g) How to specify points for spline curves. - All the input arrays are fit with cubic
spline curves for the purposes of interpolating, calculating derivatives, or any other
required calculation. A cubic spline curve is a piecewise cubic polynomial that ex-
presses mathematically the shape taken by an idealized spline passing through the given
points. Reference 15 describes the method used for determining the equation of the
spline curve. By this method, smooth curves can be specified accurately with a few
points, usually not more than four or five. Curves with uneven places, dips, or highly
variable curvatures require more points and are more difficult to fit properly. As a
guide, enough points should be specified so that a physical spline passing through these
points would accurately follow the curve. The minimum number of points to follow the
curve should be used, since the closer the spline points are, the greater the effect of an
inaccuracy in a coordinate.
(11) Requests for output data. - The seven variables IMESH, ISLINE, ISTATL,
IPLOT, ISUPER, ITSON, and IDEBUG all request different portions of output from the
program. The optional arrays ZHST, PJ-IST, ZTST, RTST, and FLFR are also used in
this regard. The user should be careful to request only the output needed. Usually,

29
Shroud

Horizon_ I J

orthogonals-",_L__z_
-- _-_'-_-
.... I....
- --I-_7_.]-.J
] - -l]< -i:
- t -I -L- ,_-Station
I j, ["-_._:'"_I II ['.'_ lines

! _ I,._,,,oo
_ I I.'I 'l--
Vertical_ i ' _,_1_,_...._ l_j_ ._]..-_ .__
o.hogona,s. ,__ ,

Streaml

A MenShh_nr t output - at intersections of


Hub orthogonal mesh
0 Streamline output - where streamlines
cross vertical orthogonats
[] Station-line output - where streamlines
cross user-designated station lines
Streamlines and station lines
..... Orthogonal mesh lines

Figure 11. - Location of three major types of output.

only one of the three major types of output (at the mesh points (IMESH), along stream-
lines (ISLINE), or along hub-shroud stations (ISTATL)) is needed, since the ISLINE and
ISTATL outputs are obtained from calculated mesh-point data by interpolation (fig. 11).
Most likely, streamline output (ISLINE) or station-line output (ISTATL)would be chosen.
The frequency at which output is requested is also significant; this is controlled by the
values given to IMESH, ISLINE, and ISTATL.
ISTATL output should be requested only if NOSTAT is greater than zero (and the
ZHST, RHST, ZTST, and RTST arrays are given). If LROT = 0, omit the RHST and
RTST cards. Output will be given along straight lines from hub to shroud that connect
corresponding points in these arrays. Through the values in these arrays, the user
can control exactly the locations at which his output is given. These lines do not have
to be radial. Output can be requested at several upstream and downstream locations,
at the blade leading and trailing edges, and at several stations within the blade at per-
centages of chord from the leading to the trailing edge. To obtain output at the leading
and trailing edges, the values in ZHST, RItST, ZTST, and RTST should correspond
(within a tolerance of 1 percent of local chord) to the _tersection points of the leading
and trailing edges with the hub and shroud.
The array FLFR need be given only if the user wishes to specify the values of
streamline along which he wishes output (when ISLINE or ISTATL are used). If the
user does not specify otherwise (using FLFR), the program will automatically give out-
put at 10-percent streamlines.

3O
(i) Incompressible flow. - Incompressible flow cases can be executed on MERIDL
by using special input for GAM and AR. Use GAM = 0. and set AR equal to the fluid
density (RHO). In this case, TIP, PRIP, and PROP are not used by the program, but
blank cards should be furnished with the input. No other special input is required for
an incompressible case, except that REDFAC should be set to 1.0 to avoid the transonic
solution, which would be redundant and less accurate.
(i) Straight infinite cascade. - The program is primarily designed for circular,
stationary, or rotating blade rows; but the input can be adapted to apply to a straight
infinite cascade as well. Since the radius for such a cascade would be infinite, an arti-
ficial convention must be adopted. The user should pick a large average radius for his
cascade, in the neighborhood of r = 1000. Then, since the blade pitch P is known, the
approximate number of blades to use can be calculated from NBL = (2vr)/P. This value
must be rounded to an integer for use in the input. With this value of NBL and a large
mean radius, all the r-coordinate arrays of input (RHUB, RTIP, RADIN, RADOUT, and
RBL) can be established about the mean radius, and a value of MSFL can be calculated.
Using this artificial input at large radius, the program will obtain a solution for a pitch
that varies very slightly from hub to shroud and, therefore, simulates almost exactly a
straight infinite cascade.
(k) Choosing a value for REDFAC. - If possible, REDFAC should be 1.0. However,
if there is locally supersonic flow, REDFAC must be less than 1.0, usually between 0.5
and 0.95. For the best accuracy the largest possible value of REDFAC should be used
(see the section NUMERICAL EXAMPLE, p. 70). When REDFAC is 1.0, the full mass
flow will be used in the finite--difference solution for the stream function, and no tran-
sonic velocity-gradient calculation will be made.
(l) Choosing a value for ANGROT. - If the flow is close to axial (within about 45 °
from axial), ANGROT need not be given (use LROT = 0). If the flow deviates more than
45 ° from axial, use LROT = 1 and specify ANGROT in degrees so as to minimize the
maximum slope of the hub or shroud from the rotated axis. For example, an impeller
with axial inlet and radial discharge should have ANGROT = 45 °. A radial inflow tur-
bine rotor with axial discharge should have ANGROT = -45 °.

OUTPUT

There are four different types of output generated by the MERIDL program:
(1) Main output - controlled by the variables IMESH, ISLINE, ISTATL, and ISUPER
(2) Debug output - controlled by IDEBUG
(3) Information for the TSONIC program - controlled by ITSON
(4) Plotted output - controlled by IPLOT

31
Most of this output is optional and is controlled by the final input card, as already de-
scribed.
The output controlled by each of the variables IMESH, ISLINE, and ISTATL is es-
sentially the same but is given at different locations for the convenienceof the user.
The IMESH output is given at the orthogonal mesh points along horizontal mesh lines,
as indicated in figure 11. The ISLINE output is given along streamlines where the
streamlines are intersected by the vertical orthogonal mesh lines (fig. 11). The ISTATL
output is given from hub to shroud along station lines (fig. 11) where these lines are
intersected by the streamlines. Also, additional ISTATL output is given beyondthat
given with IMESH or ISLINE output.
In the following sections, output is presented from the problem solved in the section
NUMERICAL EXAMPLE (p. 70). Since the complete output would be lengthy, only the
first few lines of each section of output are reproduced here. For debugoutput
(IDEBUG_ 0), output labels are simply internal variable names, which are defined in
the main dictionary of part II (ref. 6}.
The following three sections discuss the different sections of printed output, the
plotted output, and all possible error messages.

Printed Output

Table I presents the printed output from the numerical example. Each section of
this output has been numbered to correspond to the following description:
il) The first output is a listing of the input data. Variable namesare used as
labels, and the output corresponds to the input form (fig. 5). This output is listed for
every run, regardless of the values given to IMESH, etc.
(2) The second output gives information to assist the user in checking input blade
geometry. This output is listed for every run. For each input blade section from hub
to shroud the following is given: ZBL, RBL, THBL, and TTBL, all defined as in the
input dictionary; blockage at each input point (tangentialblade thickness TTBL divided
by blade--to--bladepitch in radians); SZRBL, the meridional arc length of the input blade
section; and DTHDSPand DTHDSP2, first and secondderivatives of TttBL against
SZRBL along the input blade sections. The smoothnessof DTHDSPand DTHDSP2indi-
cates where inaccurate data or errors may have occurred in input blade geometry.
(3) The third output corresponds to IDEBUG. It has four principal sections, as the
printed output indicates. The first section gives blade geometry where the blade is in-
tersected by the 21 × 21 alternate mesh usedfor calculating gradients of the mean
camber surface. The z, r, 0, and tangential thickness coordinates are given, followed
by the gradients of 0, calculated by spline curves, along the two directions of the alter-

32
i,

_2
C o

I
i

c _

Oo
z_
c_

c
c o

o o

_ _o
m 0-, u- _ o o

• .,m_
o o c c- o
z I

o_ _ _ _

o
• . • . •
c ?
m

o o o_ t_ o o
e,,M
m * o cc o_ :_
i t, _1.
o _.
_o °
z ° = _ _
mo .°.°0 oo
t6 ,i. _u
ooooo _o o _ _o
I I

o o

o_ _
o_o_ oo oo _ _......
.... F0_0
_omo_
_o _
°.... _o ........ °
OD r-- M o
? oo ?_
..-I

o
I

o_
o_ oo 0_o

_ o_
_oo _ _ _
_ _ _ _- _..
,...° , ,
ooooo
oo I I _ o oo

_r,_

o_
oO

_d_dd_ _ d od
II I I

33
ur,
u-

q ,-
,.-:

0"o0"oOoo o r_ o o o o _' c, c, o _ _ <_, o c o, o ,_ o c o o o ,.,> _, c,. c _, c c. 0- o o o o =, c o_ o _- .--, _- _ o_ r- c

.... . • . , • . • . , • o • • . * ......... . .... , ............

r'_ _"

............ . ..... o .......... . ..........................

............. ?? .... ,o o
_o o_ ,.', ....
................... o ...... o .............. _ ...........

o= _ 7 ........
ooo_ojoo_ojoooo_ojooo
, ...... ,
ooo_o,:;oj_-jo_:,=,_'oo
. , , , , .
.............. • " • •
c_o oc'o o oooooooooooc

...... 0i I_) c" cI .... l °i ....i i °l 0i

['_ ,,_ 0
o, o-_

• • . .... , • • 0 • • • • . • • . * • • . • 0 ..... . • • • 0
t

oooooooooooooooooooooo oooc ooooc ooooooooo oooo oooooooooooo I

.............. oo V ..... o ooo o_ o _ o


_'_'_'_'''_,_,_' _ ,.',_,-',_ , ......

r,l_ Lr_ ,::t

=:= , . • • 0 • . 0 • 0 • 0 • • 0 ..................
i i
oooooooooooooooooooooo oc_;ooo_ojo_o_o,_o_ooooo .............. _oooooo_o_oo

............ ,,, ,o _, _, o
o........
oo _0 o _ o _ _ 00 ',0
ooo..........
e' c0 e0 o u'_ o o o _ o _n o u_ 0_o o _, o o 0_ o_ oo o ooo o o o o o o o o o_ooooC_ooooo o

_ oo

• o • • • o • • • o • • . • • • o • • • o • • o o • , • . • o • • • o • o . • • o o
it
I_ ooooooooo0oooooooo0ooo oooooooooooooooooooooo oooooooo0ooo
I I t I o

_oo

Ii
_ddd_d_dd_d_d_ d_dgdd6dddddd_X_Z_dd dd_,d,JgZjgZEd
I ! I I _ I

34
REPRODUCIBILITY OF Ti=t]_
_,IGINAI_ PaGE IS POOR
c ,+ c, • r'., o c' o c o
c _ cj

c c ¸ o'_ _ _ o_ o" o', o, o_ o_

c _ c _ o o o c c_, o o c_

+ , t i i / i I _ I

c,c _ 4"c :_c c m _- _ o e" c_ _ c, _ c, _, oo c _' c _, o o c_

o o

occ c,c Cc+c,c c oo_c o_ooc ooooooooooooo o o

.111_

g _
o _

........ • -.. ......0..o.,..


o o

+_ o _ c, c_ o c, oc_o _o o oo o o o oo oo o oOo _
p

o c ¸ o o
o o c_ c
o o

_- r_
o _

oco,_oooccc_ _ o_ c;_oo oooooc oc oooc-_ o o c o

H
_J _o_c ooooo_oor_ooooooo

' I g ' ''


oo oo oo
o _o
u_ ,_ oo oo oo
oo oo
m _ cc
oo oo oo
oo coo
o oo u_o
r_ . _ •
• , m,
o o o c
I , I I I

e_ o o o
a. i

_ _ o _o _o
oo _,o
g_
o _o° _
.................... . ..... , , ° ° , ° ,-I , ° M . ° • . ° • _o
o o
c-c'oo©_o _cc o oc,_ r_ o o oo o ooo o_ oo o o o Oo _ oo oo oo
o-
O,e
re
o <_ ,_ _ o _ooooo oooooooooooooooo

, , , - _ . . .
m oo
_ _ oooo _ oo o ,o o _ o "o _ _ o 00 o _ o _ o _0 o oo _ o _ _ _ oo oo
_oo_oo
_oo0oo
_o_o
_oo_oo _g
,-1oo
e._oo

"_oooooo _o_ c_oc ooooooooooooooooooo g ds oo oo


I I

35
I

_, °_,

_o_o_,_oo_"
_ t

_ ro

t w-

; _._ _.___._ ;_-_._.

_oo_oo_ _o_O_
_ _oo_oo_

,_ .....c_. o °oc_oc_oooooooo
° ....... °

o o_"
" _o_ ""o
o "o_ L_,_
__''_
....

_c0 ° °o_c_oooooc,
° ° ° ° ° " " " ° c_,_oc>
" ° "

7- ......... o_r_ _

3_
c
t

°°°°..°

iiiiiii

.......... • °
o_oooooooooo

oooo

_°°.°.°...._.
oooooo0ooooo

ooooooo
* O*

oOoOoooooo_m

mmmm_m
°,°°°,,
ooooooo

:m*
_e
01 lilt

oooooooooo_

0°°°.°°
..,....°°...
oooooo0ooooo II

oooooooooooo
oooooo_oOoOo

0ooOoOo_oOoOo
_oooooooooooo
• ... ... °°

ooooooo

,,??
°° ??_ oo?_
Ii .......
IIit111

E _0ooooo0oooo

,°°_.°_
ooooooo

y
o0

37
g_I_oDUCIBILIT¥ OF Yl_
6_IGIN_'Ju pAGE IS pOOR
i

_ ..... _c_ _

c_ ............

Iii1_11111oo

_ooooo_oC_oa_¢

_ooOo_oC, o_c_r ._

,._ oooo ooooo_ e-i _


:': _ • ° . ° . ° . ° . ° . °
_.o_ooocooooo
i i

o, ° , . . ° ......

:o:
n n

.J
r-
_oOoOoOoOoO
_0.... ..... •
.q ii
o

_m

ii

.:o."
0..°..° .... _

_111111_1111

II II II

= °_

ooooooooooo_
_ooooooooooo
ooooooooooo_
_ooooooooooo
0..°°,.° ....
_Mm oOooOooooooo
a

H N W

m...,0....,,.
_N

Mu

N*
U

38
_=*,..,,,,,,,.

oocoooooooo_
oocoooooooo

ooooooooooo
_oooooooooooo
m,,°°°00°0,0.
OoooOoOooooo

*0_*.,00.,.0.,.

ooooooooooo_
ooooOoOoooo
_ooooooooooo

_ooooooooooo_
.,.oo.°o.°o°
_0_,,,,0,,.,,0,

* _
_0_o_o_

0.0,000°0..0 :o_: _0..0°°..0,,°


_oooooooooooo

:o_8: **:*:
_,o,.,.o....o

: o*
_oooo_oooo me O*

*_o:
*
:_:o. _*
_*o*.*o*.O*.O
_ ..... 000 ....
oo_ooocooooo
iiii ii
O*

:.:.:
:_ ............
oooooooooomm
oooooooooo_o
_oooooocooo_,

_ooooooooo_
_ooooooooo
._ 0 *
_000.*000°0_
oooooo_ooo
_oooo00oooooo
mDcooooooooooo
. oooooooooooo
* m _oooooooooooo

_ °,0000000.0,
oo0oO0oooooo
o

oooooooooo_ * m
oooooooooo_

_ooooOoOooo_
°°00°°0°°._. * t

ooooooocool_

,°0°0.°°.,0.
oooo0ooooooo

ooooooooo_oo
ooooooOoooo_ IIIIIIOIIIII

_oooooooooo_
_oooooooooo_
=oooooooooo_ _ o_o_oo_o
oooooooooo°_

oooooooooo
,.0,°.,,°°..

oo0oo0ooo0_o
iIiII1|111

39
ooooooooooo
._o_oooooooo

_mo_coc_ooooo

_oococccooo_
, ..... ..,..,
_moooooocoooom

._,,.,,......

m_ mcoooooooooo
°

_ooooooooooo_
_.,.°,.0,.0,,

t_Z .... o ....... • _ _

• C
E °m
ii

:_=,.
_U°.°°°°°°°°°°.

,c :777777 ......
lilJll •_ .........
• J.

,-1.
ram*

_ "_'°'°'°°'°°°°

: .=:

_°°° .... °°.°0

* mz.

:*:*: _* :*:*:

!i:: O
_
=
°°°.°°°. °°
m* _ O*_

=*

:- _.°°°0.°°.°°0

* O *
* U*.CD :o :
D- O
_ d
_°°°°°.°°°.°°

• U

:= _,.,o,,,,,,,.
_o°oooooooooo
* O _oooooc_o_
_ _mooooooooooo
oooooooooo

:= :=o: . 0 ° °
ooOo_ooooo_ . • • 0 • , • °

ooooooooooo
IIIIIIIiii
0 / I , I I _ I I I _ ,

I=u _m_mmm_pm
.°0.°°°.°.°° • • • . • • • ° 0 • •
:e ooooOoocoC_o

_oooooooooo
I I I I I I I I I i

o
* m

• • • • ° • • ° ° ° •
ddddddddgdgd ooocoooo_o_
I I I , , I i I I I i

7?77777777 °°
II

M_ Om_o_oo_o

°.°°, °° °° • °, ........
ooooodoogood
IIIIIIiiiii

4O
c 0-u

=, , .

e.

2
_ _; . .

u_

<

_'-4 ,.4 == m 0 m
_:L_#.. • • •

_ ... ........ ° ....

m _
u
iii I

_ e-,i _ e,.l

0.. .... ...°°°.°°°

_ d _z
• _u"

ul

.°.,..°°0.°..,..

L_
.****.
*z*

_ _ooo ooo
_ooo

OoO
_ .°....°.°..°°.°.

ooo
iii

_gg o ...... .° ..... ..°

U_

ooo
IIi

z_

U°._°.,°.°.....,°
• ° °
ooo
I

o_nec

_ 0 _o_ _or_ _ ecrmr_o_,_

• ° . ° 0 • • , • ° ° • • .
OOGOE_O_OOQO ooo

41
:a :

m_

c.
i

•, mmm_

• M M

u,"
o. _ o:
:'=:
0_
_....
mt a_
o0,00

7
i
null-

oun _-
I
[] _ '
!,1

g _

m m_

°....

g - n

,-_: ....

_eoo

cc

42
:- T,FZZ

J_ ....................

i i

u" u- _ ,i-_; ~- _:

_r e ¸

r,_ d, • ,_ c u-, _,_ ._ i_- :_ _ e-- ,_ o, _-


.... ° . , !......................
p-

u: u_

m ,n

t i q i i i
p

IIl , ii'l lii i , l , i ; IIi i


a. _,J :,1

_-___ ........ ......


......................... _, -" _' 2
.... ~ ....

............... _,Z Z2
:I: Z _',,L

--: ,r _ _ ,c r-- ,_,, -, _ .,- _ m e, :_ ,, c-4

..... ;2;
k) ............

43
_=...°.0..

:m

* U

_._

_ _?_TfTTT?
z *** e e
* m
* m . . : g
: :o.

J
_Ummmm_m

i .....
* _* o*

: :5
d m _
* * m.

_ A

g: ,.xo
_0 :
: . m I_*0,

-;;;2;;;;
>B: _ °

g
_m_ m

* 0-_* _ J ........
, u_ _,

• >•_:
_* t_*

* * * * _.*
_--oooooc, oo m
* _ *.J* • ...1
ococoooo * M*
* r e -'_* * z
* o_ocooo
e iIiIill
ell
.* _ * **
* m :e

_0_ _ _
..° .....

°°°.....

°. ......
ocoooooo
cooooooo

_&'Z' '_ _

Ow
Iiiiiiii

)
J

44
:F-

:==:
: .=:
* _L_
,JO • O*

* .'n

C_
[.-0• _•

o:
:,-=_ :=: _e

M. _a* •

M*

M_

• U*

• UQ* o-m*

_u• O*
_. ooooocoooo_
U_oooooooooo

_ ,0°°,,°°°°*
• b •

iiiiiiiiiii

• z,-l*

O* _U mm_m_m_m_m
,0 .... *°°,°

• _,.1*

??? ........
• m

_o_mmmmm_mmmm

.u 22%%%_2%222
ooooooooooo
lllllllllll

°°,0°,,,,0.
ooooooooo_o

,J
%

45
REPRODUCIBILITY OF THE
ORIGINAL PAGE IS POOR

* * •**

I¢1,

m•

5
**•*******

• _Om*

: • _lo.
....... °

• o ::m_

B o* _*
O
:_,:. o


o
[-.

O* tn _.• _ u_
i 0,..,.*°,
•_• ¢.)•

a_ • :=, _:=•
O* * • _-I _1 * =,
i
b=_ e * .c _.*

_•0 •

• z* : • m _rn
• z* * _J,m, e
:=:
*0,
•**••**•**

_*
m•
eL*
_e
O*
=* , _,=N:

I * _[--

_ °°..°°.°
M•
_• 0

•4•
oa•

* Ue
o.°.°..°°
*z*
_1_"

...1•

u°°...°.o

• • 0 •

".:
*•••**•*•*

:_1:2 =•
: _* =:
Iz_)

** •**•*

/ k

co

46
o !

r_
I

u_ N .... _
_u_

_ _°.°°

UIII_IIIIIIIIIIIIIIII o _411

_1111

o r_
o 0

_m

_c
_ r_

Jill

,)
I

,,??
° r _ _

_ dggd
_ o

, , -- 2

47
nate mesh. This first section of debug output is given only once. The second section
of debug output lists orthogonal mesh quantities that are held constant. These quantities
are also listed only once. The third and fourth sections of debug output list the coeffi-
cients of the matrix equation that is solved for the stream function and the orthogonal
mesh quantities that change with each iteration. These two sections are listed follow-
Lag an iteration of the solution, as called for by the value of IDEBUG. The output
giving the changing orthogonal mesh quantities contains two different sets of variables.
(4) The fourth output indicates the calculated value of overrelaxation factor ORF to
be used in the successive overrelaxation solution of the finite-difference equations.
This matrix equation for the stream function u is solved iteratively (this is referred to
as the "inner iteration") during each of the "outer iterations" of the reduced-mass-flow
solution. The ORF output is given automatically on each run of the program.
(5) The fifth output lists the calculated LOSOUT array at the points along the line
where input downstream flow conditions are given. This array is printed if PROP is
given as input.
(6) The sixth output lists the ILE and ITE arrays. For the jth horizontal mesh line
from the hub, ILE(J) and ITE(J) are the vertical mesh line numbers of the mesh points
that are within the blades and closest to the leading and trailing edges.
(7) The seventh output gives convergence information about DVTHDT, FT, and the
relative velocity. Convergence is determined by the maximum relative change in veloc-
ity. When the maximum change becomes less than the input value of VELTOL (reduced
by the minimum of FNEW or DNEW), the stream-function (either reduced or full mass
flow) solution is considered converged. Also given are the indices where the maximum
change occurred, the average relative change in velocity, and the number of uncon-
verged points.
If there is a convergence problem, the behavior of DVTHDT and FT are of interest.
Therefore, the maximum and minimum values and the maximum change in DVTHDT
(2(rV0)/0t) and FT(F_ are printed for each iteration. The input variables DNEW and
FNEW are used to control the fraction of the calculated changes of these variables to be
used for the next following iteration. If the maximum change in either DVTHDT or FT
is increasing with each iteration, it is suggested that either DNEW or FNEW be de-
creased.

(8) The eighth output corresponds to IMESH for one of the iterations of the stream-
function solution. The output is given along each horizontal mesh line at each mesh
point. The output given at each mesh point includes the following: z- and r-coordinates

of the point, stream function u, meridional velocity Wm, relative tangential velocity
W0, relative velocity W, critical velocity ratio W/Wcr , meridional flow angle _,
relative flow angle fl, and mesh angle _.

48
(9) The ninth output corresponds to ISLINE for one of the iterations of the stream-
function solution. This output is given along each streamline, corresponding to a given
stream-function value. The points along the streamline correspond to where it is inter-
sected by the vertical mesh lines. The origin for the m-coordinate is located where
each streamline begins at the upstream boundary of the orthogonal mesh. The output
given at each streamline point includes the following: z-., r-, andm-coordinates of the
point; meridional velocity Wm; relative tangential velocity W0; relative velocity W;
critical velocity ratio W/Wcr: meridional flow angle a; relative flow angle /3;
streamline curvature 1/rc; and within the blade-to-blade passage, the estimated
suction-surface and pressure--surface velocities.
(10) The 10th output corresponds to ISTATL for one of the iterations of the reduced-
mass-flow solution. Instead of output being given along horizontal mesh lines (IMESI-1)
or streamlines (ISLINE), it is now given in the other direction, along lines from hub to
shroud. This output is given along each of the station lines specified by the input
ZHST, ZTST arrays. Output is given at each point where these station lines are crossed
by the streamlines (the values in the FLFR array). The output given at each point is
identical to the output given for ISLINE, with the addition of the stream function u. The
second row of output includes the following additional variables: a repeat of z- and

_Nnl )fs-- -' I


(Wm Jbf

Figure 12. - Definition of incidence angles.

49
r-coordinates of the point, static pressure p, static temperature T, static density p,

absolute tangential velocity V0, absolute velocity V, absolute total pressure p', abso-

lute total temperature T', absolute flow angle flabs' relative total pressure p", and
relative total temperature T".

(11) The 11th output consists of blocked and unblocked incidence and deviation angles

and blade angles at the blade leading and trailing edges where each of the horizontal

mesh lines intersects the blade. The blocked incidence ibf or deviation is based on
the velocity diagram within the blade (subscript bf in fig. 12); the unblocked inci-

dence ifs is based on the free-stream velocity diagrams (subscript fs in fig. 12).
This output is printed after each iteration in which output is given for either IMESH,

ISLINE, or ISTATL. These angles are defined as shown in figure 12.

(12) The 12th output corresponds to ITSON. It is a listing of the information re-

quired to prepare input for the TSONIC blade-to-blade analysis program of reference 3.

(The TSONIC program is being updated and will be reported in the future. The defini-

tions of all the input variables for TSONIC will be given in the updated version of

TSONIC.) The TSONIC program was run for both the hub and tip blade sections of the

compressor blade used in the numerical example of this report. The input numbers
shown in the input form in figure 13 are those computed for use with the TSONIC run at
the hub.

The information printed by MERIDL can in some cases serve as input for TSONIC

without revision. In other cases, some of the input may need revision. The informa-

tion from spline fits (derivative, second derivative, and curvature) of the THSP-

against-ZMSP arrays for each blade surface is printed alongside these arrays. The

user may wish to make small changes in the BETI and BETO tangency angles, which

serve as end conditions for these spline fits, in order to smooth the resultant first and
second derivatives.

The final five arrays of stream-channel data - ZMRSP, RMSP, BESP, WOWCR,

and PLOSS - can in general be used as they are printed, unless the user wishes to re-

duce the number of points (NRSP) in them. The BESP array is calculated to correspond

to the given mass flow ZMSFL. Any value could be used for ZMSFL if BESP is chosen

accordingly. The value of ZMSFL calculated by MERIDL is 1 percent of the mass flow

between any two blades. The BESP array is sometimes bumpy and may also have to be

smoothed before it can be used as input to TSONIC. The derivatives of BESP against

ZMRSP are printed and will indicate if smoothing is necessary. The WOWCR array is

only needed as TSONIC input if REDFAC < 1.0 is used for TSONIC. The printed PLOSS

array (NLOSS > I) can be used for TSONIC rather than a single value for PLOSS if a
nonlinear loss distribution through the blade is used in MERIDL.

(13) The 13th output is analogous to outputs 8, 9, 10, 11, and 12 for the variables

IMESH, ISLINE, ISTATL, and ITSON but is given after the transonic velocity-gradient

5O
5]6 loin 15116 20[21 25]26 30131 35 36 40141 45146 5o151 60 61 70 71 80

TITLE
DATA12 - AXIAL COMPRESSOR ROTOR - HUB - STREAMLINE NO. I - U = 0.00

GAM AR
1.4 287.053 TIP 15
288. RHOIP
I. 22534 [ OMEGA
-826.55 ZMSFL
O.025194

REDFAC VELTOL
1.0 0.001

MBI MBO MM MBBI NBL]NSPLi]NSLP2NRSP


10 50 60 20 NSL 0

LRVB LOSS LWCR LIPS 171 20/ _._


0 I' 0 0

RVTHI RV1-HO FSMI FSMO SSM1 SSM2

-9.88533 -27.6253

CHORD STGR

O. 113429 O.0.54930
RII BETII ROI BETO1
O. 001346 29.516 0. 0009226 -23. 859

ZMSP1 ARRAY
0. 00068 O. 00629 O. 01247 0. 01863 O. 02479 O. 03093 O. 03707 0. 04320

0. 04933 .0. 05545 0. 06156 0. 06767 0. 07378 0. 07989 0. 08599 0. 09209


O. 09819 0. 10429 0. 11039 0. 11288

THSPI ARRAY
O. 00762 O. 02745 0. 04682 0. 06365 0. 07815 0.09046 0. 10067 0. 10889

0.11517 0.11954 0.12199 0.12249 0.12103 0.11747 0.11181 0. 104(X)

0. 09397 0.08159 0.06672 O.06009

RI2 BETI2 RO2 BETO2

0.001346 17.593 0.0009226 -9.602

ZMSP2 ARRAY
0.00175 O.00629 0.01247 0. 01863 O. 02479 0. 03093 0. 03707 0. 04320
0. 04933 0. 05545 0. 06156 0. 06767 0. 07378 0. 07989 0. 08599 0. 09299

0. 09819 0. 10429 0. 11039 O. 11235


THSP2 ARRAY
-9. 00835 0.00077 0.01221 O.02263 0. 03198 0.04027 0. 04751 0. 05369

0.05879 0.06283 0. 06581 0. 06775 0. 06866 0. 06847 0. 06720 0. 06487

0.06145 O.05695 0. 05137 0. 04937


ZMRSP ARRAY

-0. 04741 I -0. 03285 -9. 01860 -9. 00455 ] 0. 00629 0. 02479 0. 04320 0. 06156

0.07989 [ 0. 09819 0.11648 0. 13173 ] 0. 14544 0. 15916


RMSP ARRAY

0.14107 ] O.14594 0.14979 0.15283 0.15478 O.15753 0.15968 O.16130


0. 16245 [ 0.16322 0. 16372 0. 16399 0. 16417 0. 16431
BESP ARRAY

O. 002268 O. 002233 0.002205 O.002208 O.002228 0. 002199 O.002163 O. 002110


0.002049 0. 001989 0. 001926 0. 001902 0. 001892 0. 001885

PLOSS ARRAY
0.00 0.00 O.00065 0.00249 0.00432 O. 00634

0. 00851 0.01119 0.00


0. 01447 ] 0.00
0. 01447 0.01447 0.01447

IBSURF IPLOT
IMESH5 I ISLINE5 i I

Figure 13. - TSONIC input form. (Data shown are for the numerical example at the hub. )

51
soluUon. Two solutions can be obtained by the velocity-gradient method: the larger or
"supersonic" and the smaller or 'tsubsonic." The input variable ISUPER controls which
of these is obtained. The solution listed here is the smaller (i. e., the subsonic solu-
tion). If REDFAC = 1.0, no velocity-gradient solution will be obtained or printed be-
cause the final iteration of the stream-function solution is the correct solution to the
problem.

Plotted Output

Since the printed output from a MERIDL run can be quite large, microfilm plots
are made to enable the user to quickly check the quality of both his input and output
data. The coding that generates these plots uses NASA Lewis in-house plot routines
and would have to be recoded for operation on another system. The principal sections
of input data plotted are the upstream and downstream distributions of flow properties;
the input blade sections; and the hub, shroud, and blade leading- and trailing-edge
geometry. The frequency at which output data are plotted is controlled by the variable
IPLOT. The plotted output data begin with the generated orthogonal mesh. Then for
each iteration of the solution (indicated by IPLOT) and for the transonic solutions,
streamlines are plotted, as well as midchannel and blade surface velocities along each
streamline from hub to shroud. The user should carefully check the plots of his input
data to ensure that the program's spline fits of these data are smooth.
Selected examples of some of the microfilm plots generated by the numerical ex-
ample are presented in figure 14. The description of these plots is as follows:
(1) Figures 14(a) to (e) present the input upstream and downstream flow conditions.

These quantities t F'I' Pi" k or (V)0i, Po' ortotalpressureloss, (rV0) ° or(V0)o)


are plotted against either stream function or radius, whichever was used as input.
(2) Figure 14(f) shows the optional input of total-pressure-loss distribution against
fraction of chord.

(3) Figures 14(g} and (h} indicate hub and shroud blade sections plotted from input

data supplied by the user. All blade sections given as input are plotted.
(4) Figure 14(i} shows the hub and shroud channel geometry and the blade leading
and trailing edges.
(5) Figure 140) shows the generated orthogonal mesh.
(6) Figure 14(k) shows the streamline pattern for the final iteration of the reduced-
mass-flow solution.

52
.50

.28

.26

.24

,,et

.22

.2O

.18

.16

.14
25O 260 270 280 2go 300 310 32O 330 340 350

INPUT ARRAY - TIP

(a) Inlet absolute total temperatu[e.

Figu[e 14. - Microfilm plots of input and output..

53
,54

•28

.24
N

"" .22

.20

.18

.16

.14
e.N|10| o.lleeoe e.lo_eoe e. 11elllle
10 06 I ii 06 ! Io N 11 le I1(I

(b) IntetabsoJutetotalpressure

Figure14.- Continued.

54
,2B
\
.26
\
•2,4

.2.2
\
.20

.1B

.IB

-40 -2O -15 -!0 -5 0

- V+_IN

(c) Inlet absolute tangential velocity.

Figure 14. - Continued.

55
,3O

,28

.26

.2.4

ql

.22

.2O

.18

.t6

.14
l.t t_fltt O.o_ldlOO |, oal4el
IIi N u IO Ol s _| 04

IId%PT ARRAY - PROP

(d) Outlet absolute total pressure.

Figure 14. + Continued.

56
.5O

.28

.26

.24
/
¢::b

.22
B"
/
.2O

.18

.16

.14
-170 -!(;5 -160 -155 -150 -145 -140 -155 -I]0 -1 5 -_20

INPUT _RAV - V'rI.tOUT

(e) Outletabsolutetangentialvelocity.

Figure14.- Continued.

57
1.0

.9
/
/
.8

.7
/
.6 /
U'3
.5
/
r
c_')
. J
L_

Jl
.4

,3 ,,,

.2

/
• OI
.0 ,I •2 .3 ._, .5 .6 .7 .8 .9 !.0

.:'ERCRD

(f) Pressure loss distribution.

Figure 14.- Continued.

58
,lq

.16

,14
.q[
I..--

0,"- .12

.4[ .I0
!

I,.J

-4[
•O8
Z
N

0
r
0

..4

.04
p--

l.J
0

-. 02
• O0 .02 .04 .06 .08 .10 .12 .14 .16 • 18 .20

BLADE SECTION tqERIDIONAL COORDINATE

(g)inputbladesectionsfrom
ZBL, RBL,THBL,andTNBL- blade
section1.

Figure ]4.- Continued.

59
,111

I
,1G

.14

i'-
m
.12

mr
.10
!

W
I-- • 08
Z

o . OG
o

-.,I

• 04

• 02

.00

-. 02
.00 • 02 .Oa .06 .08 .10 .12 .14 .IG .18 .20

BLAi_ _CTION NERIDIOMAL COORDINAT[

(h) Input blade sections from ZBL, RBL, THBL, and TNBL - blade section 11.

fisu[e 14. - Continued.

6O
Z
o

Z DIRECTION

(i) Hub, shroud, and blade boundaries in meridional plane.

Figu[e 14.- Continued.

61
I i I ! ! _ i l_H-H-H--i

o
N

u
w

Z D IRECT I01_

(j) 0rthogonal mesh in meridional plane.

Figure 14.- Continued.

62
Z
o
N

U
w

Z DIRECTION

(k) Streamline plot in meridional plane.

Figure]4. - Continued.

63
340

520

300

2_0

220 _-

2O0

160

.025 .050 .0_ .tO0 .125 .150 .I_ :200 .225 .25O

I'IL"R
ID I0_I_ C_ INJI,
T[

(I) Meridional and blade surface relative velocities-streamline 1 Normalized stream function, 0

Figure ]4 Continued

64
340

32O

3OO

290

2,40
u
0

22O

200

_GO

140 i
.000 ,025 .050 .075 .I00 ,125 .150 ,175 .200 ._5 .250

HL"RIDION_. COORDINATE

(m) Meridionaland blade surface relative velocities- streamline ll. Normalized stream function, 1.0.

Figure 14.- Concluded.

65
(7) Figures 14(/) and (m) show meridional velocities and blade surface velocities
for the hub and shroud blade sections for the same iteration. A similar plot is made
along each of the streamlines from hub to shroud.
Streamline and velocity plots similar to figures 14(k), (l), and (m) are made after
each iteration (indicated by IPLOT) of the finite-difference, stream-function solution
and also after the velocity-gradient solution.

Error Messages

A number of error messages have been incorporated into the program. These
error messages are listed here, followed by suggestions for finding and correcting the
cause of the error:

M_eMHT_NHUB#NTIPwNIN*N_UT*N_LPLrNPP_*N_STATeNSL*NL_SSeLSF_wLTPLrLAMVTwLR_TwLBL_D*_ LETFAN
IS TCC LARGE OR TOO SMALL

The input dictionary gives the maximumandminimum values for all these input varia-
bles.

WHEN UPSTREAM AND DOWNSTREAM INPUT DATA ARE GIVEN AS A FUNCTION OF FADIUS,
THE_E MUST BE A CHANGE IN VALUE BETWEEN RHIN AND NTIN AND ALSO BETWEEN FHOUT AND RTOUT
AND A CORRESPONDING CHANGE IN THE RADIN AND RADOUT ARRAYS

RHIN must differ from RTIN (and RHOUT from RTOUT) so that there will be a variation
in the RADIN (and RADOUT) arrays, which are used to identify points along the hub--
shroud lines at the upstream and downstream input stations.

LININT CANNOT FIND INTERPOLATED VALUE


I = I J = 2

This message should occur only with erroneous geometry input.

INPUT PROP VALUES ARE LARGER THAN IDEAL TOTAL PRESSURE, RESULTING IN NEGATrVE LOSS

This message is for information only, the program will continue normally.

PROGRAM STOPPED IN NEWRHO DUE TO EXCESSIVE STREAM FUNCTION GRADIENT

This is usually caused by having supersonic flow in some region of the reduced-mass-
flow solution. Try a smaller value for REDFAC.

66
THE UPSTREAM INPUT WHIRL OR TANGENTIAL VELOCITY IS TOO LARGE

This message is printed by subroutine NEWREO or TVELCY if the upstream whirl gives
a tangential velocity so large as to result in a negative relative stagnation temperature.
The upstream whirl or tangential velocity given as input should be corrected.

BEWARE. THERE ARE I VERTICAL ORTHOGONAL MESH LINES WHICH ARE CHOKED.
LOCATIONS OF THESE LINES ARE GIVEN ABOVE AT THE BEGINNING OF THE TRANSONIC OUTPUT.
OUTPUT ON ANY STATION LINES LOCATED NEAR THESE CHOKED ORTHOGONAL LINES MAY BE IN EEBO_.

When this message is printed, the user should be extremely cautious in using station
line output after the transonic velocity solution. However, there may be little problem
if the choking mass flow is close to the full mass flow or if the station line is not near
one of the choked orthogonals.

A VELOCITY GRADIENT SOLUTION CANNOT BE OBTAINED FOE VERTICAL ORTHOGONAL MESH LINE I : I
AMY SUBSEQUENT OUTPUT FOR THAT MESH LINE _AY BE IN ERROR

If the approximate transonic solution for any particular vertical mesh line cannot be
found for some reason, this message will be printed. Some of the following error mes-
sages will be printed to help determine the problem. The program will proceed to the
next vertical mesh line.

CONTIN COULD NOT FIND A SOLUTION IN 100 ITERATIONS.

This message is printed when the transonic flow solution for some vertical mesh line
cannot be obtained by subroutine CONTIN in 100 iterations. Possible causes for this
error message could be unusual or impossible upstream or downstream input quantities
(TIP, PRIP, LAMIN, VTHIN, PROP, LOSOUT, LAMOUT, or VTHOUT arrays).

ITERATION PROCEDURE HAD TO BE RESTARTED TO AVOID NEGATIVE TEMPERATURE OR VELOCITT


MAGNITUDE LESS THAN TANGENTIAL VELOCITY, OR AFTER ADJUSTMENT OF STAGNATION TEMPERATURE,
STAGWATION DENSITY, OR WHIBL.
RZSTART OF ITERATION PROCEDDRE (LOOP TO STATEMENT 90) HAS ABORTED AFTER 1000 OR MORE
TOT&L I_ERATIONS (LOOP TO STATEMENT 150).

This message is printed when the transonic velocity-gradient solution cannot be found
by subroutine TVELCY in 1000 iterations for some vertical mesh line. Possible causes
are high input whirl, impossible upstream or downstream flow condition, and complex
or erroneous blade or wall geometry.

67
THE MAXIMUM MASSFLOW FOR WHICH A SOLUTION COULD BE OBTAINED WAS 0.120C0COE-02
THE MINIMUM MASSFLOW FOR WHICH A SOLUTION COULD BE OBTAINED WAS 0.I01000_E-02
THE MAXIMUM VALUE OF W AT THE HUB FOR WHICH A SOLUTION COULD BE OBTAINED WAS lO1.1_00
THE MINIMUM VALUE OF W AT THE HUB FOR WHICH A SOLUTION COULD BE OBTAINED WAS 99.7999g

This information is printed whenever the transonic, velocity-gradient solution cannot


be found for some vertical mesh line. This information has debug value if there is no
apparent reason for difficulty.

THE ITERATION PROCEDURE WAS RESTARTED 380 TIMES


WHUB WAS INCREASED 112 TIMES
NHUB WAS DECREASED 56 TIMES
BOUNDARY VALUES (TIPBD¥, RHOIP, LAMBDA) WERE ADJUSTED 34 TIMES

This information is printed whenever the transonic solution cannot be found after re-
starting the iteration procedure. This information has debug value if there is no
apparent reason for difficulty.

MSFL EXCEEDS CHOKING MASS FLOW FOR VERTICAL ORTHOGONAL MESH LINE I = I
C_OEING MASS FLOW = 0.950000E-03

This message is printed if the choking mass flow calculated by subroutine TVELCY is
less than the input mass flow MSFL. The following message will be printed after all
vertical lines have been checked:

CHOEING MASSFLOW IS 0.96000 OF THE INPUT MASSFLOW


INPUT MASSFLOW = 0.15000E-02
MINIMUM CHOKING MASSFLOW = 0.14_00E-02

If the previous message was printed for any vertical mesh line, this message will be
printed at the end to give the minimum choking mass flow, which is the choking mass
flow for the blade row estimated by MERIDL. Use of the CHANEL program (ref. 13)
is recommended to obtain a more accurate estimate of the choking mass flow.

SUPERSONIC CORRECTION - LEADING EDGE FOR J = 2

For the full-mass-flow solution when REDFAC < 1, the blockage correction at the lea&
ing edge requires calculating the density by an iterative procedure. Since the iterative
procedure depends on whether the flow is subsonic or supersonic, the program first
checks the subsonic procedure; if this fails, it is usually because the flow is super-
sonic. I£ the supersonic solution is found, the preceding message is printed for infor-
mation only.

68
SUPERSONIC CORRECTION - TRAILING EDGE FOR J = 2

This is the same as the previous message but at the trailing edge.

NO DENSITY CORRECTION - LEADING EDGE FOR J = 2

If after the subsonic procedure fails the supersonic procedure is tried and also fails,
that procedure is aborted and the density within the blade is taken to be the free-stream
density. This problem would probably occur because of numerical problems associated
with near-sonic flows, high blockage, and complex or erroneous wall or blade geometry.
The program will continue after this message, although output near the leading edge
may be poor.

NO DENSITY CORRECTION - TRAILING EDGE FOR J = 2

This is the same as the previous message but at the trailing edge.

INRSCT HAS FAILED TO CONVERGE IN 20 ITERATIONS


TOLERANCE = 0.ICI000E-05
DISTANCE BETWEEN LAST two INTERSECTION POINTS = 0.678900E-05

Subroutine INRSCT finds the intersection coordinates of input station lines, mesh lines,
and streamlines with the blade leading or trailing edge by an iterative method. If the
tolerance cannot be met after 20 iterations, the message is printed. If the distance be-
tween the last two intersection points is only slightly larger than the tolerance, a satis-
factory solution will be obtained, with some loss of accuracy. If the distance is exces-
sive, there is probably some error in the geometry input.

SPLINE ERROR -- ONE OF THREE POSSIBLE CAUSES


I. ADJACENT X POINTS ARE DUPLICATES OF EACH OTHER.
2. SOME % POINTS ARE OUT OF SEQUENCE.
3. SONE X POINTS ARE UNDEFINED.

NUMBER OF POINTS = 3

ARRAY Y ARRAY
0.00000 10.00¢
0.00000 12.000
1.0000 15.000

The spline points for a spline curve must be distinct and given in sequence. If not, the
spline fit subroutine (either SPLINE, SPLINT, SPLISL, or SPINSL) will print this mes-
sage, and the program will terminate. Since spline curves are used so extensively in

69
the program, it is difficult to state the possible cause. However, the printout of the
spline points should assist in pinpointing the causeof the error.

SPLINE ERROR -- NUMBER OF SPLINE POINTS GIVEN IS LESS THAN TWO

NUMBER OF POINTS = -3

X ARRAY ¥ ARRAY
0.00009 I0.000

At least two points must be given to determine a spline curve. If not, the spline fit sub-
routine (SPLINE, SPLINT, SPLISL, or SPINSL) will print this message, and the pro-
gram will terminate° The one printed spline point may give a clue as to the error.

NUMERICAL EXAMPLE

An example of an axial-flow compressor rotor is used to illustrate the use of the


program and to show the type of results that can be obtained. Flow is analyzed on the
midchannel stream surface of an axial-fiow rotor designed with the computer program
of reference 11. The design pressure ratio is 1. 275, the inlet hub-tip radius ratio,
0.5; the aspect ratio, 1.5, the tip solidity, 1.0; and the tip relative Mach number at
the inlet, 0.9. Although tip relative Mach number is near sonic, there are no locally
supersonic regions on the midchannel stream surface.
The printed output from this example is given in table I. The input for this exam-
ple is item 1 in that table. The number of mesh points used was 861:41 in the axial
direction, and 21 in the radial direction.
This rotor was run twice, with reduction factors (REDFAC) of 0. 99999 and 0.70.
This was done in order to test the effect of reduction factor and also the accuracy of the
velocity-gradient method in comparison with the finite--difference method. Since both
cases have REDFAC < 1.0, both the subsonic (finite difference) and the transonic
(velocity gradient) solutions will be obtained in each of them. (If REDFAC = 1.0, only
the finite-difference solution is obtained.)
The first case (REDFAC = 0. 99999) permits a comparison to be made between the
approximate velocity-gradient method (used to obtain transonic solutions) and the more
exact finite-difference, stream-function solution (used for subsonic solutions). Since
the reduction factor is 0. 99999 (<1.0), both solutions will be obtained, but the answers
should be very close if the methods compare well.

7O
-- Finite difference (REDFAC• O.99999)
26C
----- Velocity gradient (REDFAC= O.99999)
...... Velocity gradient (REDFAC= 0.70)

22C

Blade i _
_ edee' l I "_gllade
__edge trailintleadine
18(]
(a)Hub section.
28O

200

160 I 'ng edle t I Bladetrailingledge

(b) Mean section.


280--

2111--

200 -- __
Blade leading edge
IBlade-t_ n]'_-e_e
16o l II I t I I I ]
-.08 -.04 0 .04 .08 .12 .]6 •20
z-Coordinate,
meters

(c)Tipsection.

Figure15.- Midchannelvelocities
foraxial-flow
compressor
example.

The second case (REDFAC = 0.70) permits a comparison of reduction-factor effect.

Since the finite-difference solution is now obtained at only 70 percent of the mass flow

and rotational speed (appendix E), the results for the full-mass-flow, velocity-gradient

solution will be less accurate than when a higher reduction factor is used.

Figure 15 shows midchannel flow surface velocities for both reduction factors and

both types of solution (the finite difference and the velocity gradient). Results are pre-

sented for three values of stream function (0.0, 0.5, and 1.0) at hub, mean, and tip.

On all parts of this figure, the solid line represents velocities from the final finite-

difference iteration of the REDFAC = 0. 99999 solution. This solution is mathematically

most accurate. The results presented here show a much better comparison of velocity-

71
MERIDL (REDFAC• O.99999)
440-- ..... TSONIC

400

35O

P__.
eL,

E
320 t
|

280 -- \\\
\ ___s_ S_

\\
240 _ \

200

160 I i J
.04 .08 .12 0 .04 .08 .12
m-Coordinate, meters

la) Hub section. (b) rip section.

Figure 16. - Blade surface velocities for axial-flow compressor example.

gradient to finite-difference results than that presented previously for the MERIDL pro-
gram in reference 4. The two different velocity-gradient solutions are likewise now in
much better agreement.
Figure 16 shows a comparison of blade surface velocities estimated from the
MERIDL program (appendix G) and calculated from the TSONIC program (ref. 3). The
TSONIC velocities are more accurate for the design blade shape.

Lewis Research Center,


National Aeronautics and Space Administration,
Cleveland, Ohio, July 28, 1976,
505-04.

72
APPEND_ A

GOVERNING EQUATIONS

The cylindrical coordinate system is shown in figure 4. However, the solution is


obtained on an orthogonal mesh, as illustrated in figure 3. The variables s and t are
used to denote distance along the streamwise and normal orthogonals, respectively.
For subsonic solution, the stream function is used. In appendix B we obtain the
stream-function differential equation derived from the assumptions given in the section
METHOD OF ANALYSIS:

32u +02u 8u{slncp +i 8_BB


+l __ _ +i _B +l _ +
8s 2 Ot 2 0s B 8s p 8s 8t 8t\ r B 8t p 8t

+4 +_ +F =0 (A 1)
+wW s L r O(rVo)
ot
w2 t

where

(A2)

2 \Cpp" 8t T" 8t /

2 RT" Op_
= co r cos _o (A3)
p" Ot

gt- 80 1 _p (A4)
8t p 80

Equations (A1) to (A4) are derived from the hub-shroud momentum equation in
appendix B. Note that all the partials in equations (A1) to (A4) are on the stream sur-
face except for 3p/0 0 in equation (A4), which is at constant z and r.
The derivatives of the stream function satisfy the equations

8u _ rBpWt
(A5)
0S W

73
0u_ rBpWs
(A6)
0t w

For the final transonic solution the velocity-gradient method is used. The stream-
line curvatures and flow angles needed for the velocity-gradient equation are obtained
first from a reduced-flow, subsonic-stream-function solution.
The velocity-gradient equation is

/ \
dW = _aW+b +_+dcosHIdt+_+Wf (AT)
\ W ] W

The coefficients, a, b, c, and d, are given by different expressions in the blade re-
gion, in the upstream region, and in the downstream region. These coefficients are
given as follows:

Blade- region coefficients:

a = c°s2_ cos(_ - _) _ sin2_ cos _p+ sin _ sin fl cos/3 v__=


r r 0t
c

(AS)
b=cos/3
dWdmm sin(a- go)- 2w sin/3 cos _o +r
('dWo
cosfl_-_m +2w sin
a/ 3 0

c=O

d=O

Upstream-region coefficients:

74
a=C°S(_- _)
r
c

b=O

(A 9)
C cos(_ - _) + _ + wr2
r

dW
d =---_msin(_ - ¢)
dm

Downs tream- region co efficients:

a = cos(a - _)
r
e

b=0

C _ m
(rV°+wr) (A10)
cos(c_- (p) +
I!rVot_:°zr21IrVoIolc w2 r

dW
d D
m sin(c_- _)
dm

Finally, in all three regions, we have

RT r,
e = CpT}- w dh- CpdT" +-- dp"
p"
(A11)

f_ dT" Rdp"

2T " 2Cpp" J

75
Equations {A7) to (All) are derived in appendix C. Equation (A7) is solved as an
initial-value problem, where the initial value of W is specified at the hub for any given
t-line running from hub to tip. By finding several solutions for varying values of W at
the hub, a solution satisfying continuity will be found; that is, the solution will satisfy

_0 ttip pWrB cos(a - (p)cos _ dt = w (A12)

When equation (A7) has been solved, subject to satisfying equation (A12), for every hub-
to-tip mesh line in the region, the entire velocity distribution is obtained.
The solution obtained by either the finite-difference or velocity-gradient method is
for the midchannel surface between the blades. Of greater interest are the blade sur-
face velocities. These can be estimated since the blade loading depends on the rate of
change of whirl. By assuming a linear variation of velocity between blade surfaces,
velocities can be calculated. In appendix G the following equation for calculating blade
surface velocities is derived:

d(rV 0)
-} cos --
Wl = Wmid 2 dm

(A13)

Wt r = Wmid + B cos// d(rV0)


2 dm

76
APPEND_ B

DERIVATION OF STREAM- FUNCTION EQUATION

Wu derives the following radial- and axial-momentum equations (eq. (96), ref. 1)
on a meridional stream surface:

Wo 3(rVo) [aWr OWz/_ _3I +TSS+ F


r Or + Wzk _z _r] Or 8r r

(B1)

_OWr OZz._ WO 3(rVo)_ 3I +T 3s +F z


- Wr k az Or / r az az 3z

The partial derivatives here all refer to the rate of change on the midchannel stream
surface. Wu used a bold derivative symbol to indicate this, but it is not necessary here
since we consider the equation only on the stream surface. The first equation of (B1) is
the radial momentum equation and the second is the axial momentum equation.
Our flow solution is obtained by using finite-difference methods on an orthogonal
mesh. The orthogonal mesh distances are s in the throughflow direction and t in the
hub-tip direction. Figure 3 illustrates the s and t distances and also shows the angle
_p between the s-direction and the axial direction. We want the momentum equation in
the t-direction. This t-momentum equation will be used to derive the desired stream-
function equation. The t--momentum equation is obtained by adding the t--components of
the radial and the axial momentum; that is, by multiplying the first equation of (B1) by
cos ¢p and subtracting the second equation of (B1) multiplied by sin q_. Also, the chain
rule for a partial derivative is

a 3 3
--= cos _--- sin q_--
3t Or Oz

Therefore, the momentum equation in the t-direction can be written

Wo
r
a(rVo)
at
+ (W z cos _o + W
r
sin _o)
l,w r
_r /
_ aI as
a'_ + T_-_ + Ft (B2)

77
where Ft is the t-component of the vector F, which is given by equation (97) of ref-
erence 1 as

nt ap (B3)
Ft -
n0r p a 0

Here n t and n o are the t- and 0-components of the unit normal to the stream surface.
Note that ap/a0 is in the blade-to-blade direction at constant r and z and not on the
stream surface as theother partials are. In the t-direction, the stream surface has an
angle to the meridional plane. The tangent of this angle is r(a0/at). Hence, the com-
ponents of the normal to the stream surface must satisfy

no_ 1
(B4)
nt r(a0_
\at/

We can substitute this into equation (B3) to get

Ft_a0 1 ap (B5)
at p aO

It is desired to express the t-momentum equation (B2) entirely in s- and


t-components and their derivatives. First, we will use

W s = W z cos cp + W r sin _p (B6)

This leaves only

aw aw
r z

az ar

in equation (B2) to be expressed in s- and t--components. For this we use

W r=w

W z=w
tcos

s cos
e+W

_p-
s sin

W tsin
) (B7)

78
.nd the chain rule (noting that Os/Or= sin q_, etc.)

O_.. = sin cp O__ + cos ¢pO__


Or _s Ot
(B8)

_.a = cos ¢_---- sin ¢0_.


0z Os 0t

3y using equations (B7) and (B8) we can express, after combining and simplifying,

0Wr 0Wz- 0Wt 8Ws + W _ + W t 0(p (B9)


0z Or 0s 0t s _s 0t

the desired t-momentum equation in s and t could now be given by substituting equa-
;ions (B6) and (B9) into equation (B2).
It is desired to obtain the stream-function equation expressing the velocity compo-
nents as partials of a stream function. This will reduce the number of unknowns in the

momentum equation from two (W s and W t) to one (the stream function). But it changes
the equation to a second-order partial differential equation. A normalized stream func-
tion u will be used. That is, u is zero at the hub and 1 at the shroud. The value of
the stream function u can be obtained by integrating along a vertical mesh line.

,t

u=lw ]0 pwsrBdt

By differentiating this expression,

_u r__ w
s
Ot w

Similarly, the stream function u can be obtained by integrating along a vertical mesh

line to any given mesh point s 1, t I and then integrating along a horizontal mesh line:

u = 1--
w
/0 tl
pW s rB
1
dt - --
W
/s Sl
PWtrB ds

We can differentiate this expression with respect to s to obtain

79
ORIGINAL PP,CE _,'_
OF POCR QUALITY
au _ rB 0 Wt
as w

Hence, the derivatives of the normalized stream function in the s- and t-directions are
related to the velocity components by

au _ rBp Wt
as w
(B10)

au_ rBp W
S
at w

The quantity B is the stream-surface thickness corresponding to the overall mass


flow w. However, the variation in local stream sheet thickness is not known, so that
the overall blade-to-blade spacing is used, and w is the mass flow through one blade
channel. B is a function of z and r, so that we can use

B(z, r) = 0tr(Z, r) - 0 l(z, r) (BII)

re-
where 0tr and 0l are the 0-coordinates of the trailing and leading blade surfaces,
spectively.

By solving equation (B10) for Wt and Ws and differentiating, we obtain

V
aWt _ w 102u _U +± a_BB
+1
_S B as p 0s
as rBp Las 2

(B12)

aWs_ w [_a2u
ot rBp la_2 at B at P at

Equations (B10) and (B12) are now substituted into equation (B9) to eliminate W and
S

Wt in favor of the stream function u:

8O
0W r

Oz
0W z _

ar
w

rBPL
[_ 02u

Os2
02 u + 0u (sin_

Ot2 -_s\ r
+1
B 8s _
8B +_i Op_
p Os 0t

(B13)

Now substitute equations (B6) and (B13) into equation (B2) and rearrange to obtain

u .... 0u +1 0B +1 Op_ 0___ _ 8u +i__ __


0B +i Op +0__
0s 2 Ot 2 Os B 0s p 0s 8t _ B 0t p 0t 0s

+ rBp_ OI +T__ +F = 0 (B14)


IW0 O(rV0) 0s tl
WWs/ t. at at at

The entropy change dS can be calculated from

dS= C dT R d]2
P T p

Since the entropy of a particle at actual flow velocity is the same as the entropy of that
particle at stagnation conditions (either absolute or relative), we can also use

dT" R dp"
dS = Cp _-]; P" (BI 5)

This takes less calculation than using static values.


The static temperature T is calculated from T = T" - W2/2C . Using this with
P
equation (B15), since I = CpT!-I wh, yields

___aI +T__Os_ a(CpTI - wX- Cp T") RT" _ + K__ Rap"

at at at p" at 2 _Cpp" at

81
But

T!- T"-2w_'- __wrp2


1 2C
P

and

Or
--=COS (p
0t

so that

(B16)
_ aI +Ta__2_s =w 2 rcos (p _ RT'_____'
p" at 2
_p"+W2_cR_____ 8p"
at 0t pp" 0t

Now, substitute equation (B16) into equation (B14) to obtain

__02u+_____82u
_u_ +___10B p 0s Ot - _t
8s 2 at2 3s B as B 0t p 0t as]

+ rBpI!0 a(rVo) =0 (BIT)


+}W 2 +_ +F 1
WWs[ r at

where

(B18)
_ =I_cR 8p" 1 OT"/
2 pp" 8t T" Ot

= w 2 r cos ¢ RT" 0p" (BI9)


p" Ot

and F t is given by equation (B5).


In the program, sin _ and cos _0 are stored, so that it is more convenient to cal-
culate the partials of _p in equation (B17) by

82
O__ _ 1 a (sin _o) _ 1 O (cos, (p)
as cos _o as sin ¢ 0s
(B 20)

O__ = 1 a(sin _p) _ 1 O(cos _p)


at cos ¢ at sin (p at J

Either the middle term or the right term is used, depending on whether cos _p or
sin q_ is near zero.
In equation (B17), Ft and a(rV0)/at must be calculated differently outside the
blade and within the blade. Within the blade, atrv0)/at is calculated by a finite-.
difference approximation using the previous iteration. For the first iteration, the par-
rials are assumed to be zero. Outside the blade, the partials are calculated from the
input whirl distribution. From the input, the whirl is specified (or earl be estimated)
as a function of the stream function u. Hence, upstream of the blade,

O(rVo) _Ok_dR au_dX rBpWs


(B21)
Ot at du at du w

Similarly, dmwnstream of the blade,

o (rV o) d (rV0)o rBpW s


(B22)
at du w

The Ft is caused by the hub-shroud pressure gradient induced by the blade lean.
We can calculate F t from equation (B5). The blade-to-blade pressure gradient ap/aO
can be calculated from the blade loading by assuming constant entropy blade to blade,
so that

dp = p Cp dT (B23)

Using the fact that T = T" - W2/2Cp, we get

-ow aw (B24)
O0 DO

since T" is constant from blade to blade.


Now, substitute this in equation (B5) to get

83
Ft = _ ___00
W 0W (B2 5)
at 00

The blade-to-blade velocity gradient is calculated from d(rV0)/dm, as explained in ap-


pendix G, by using equation (G2). Outside the blade, there is no blade-to-blade pres-
sure gradient. Hence,

Ft = 0 (B26)

outside the blade.

84
APPENDIX C

DERIVATION OF VELOCITY-GRADIENT EQUATION

The general velocity-gradient equation is an expression for the value of the direc-
tional derivative of the relative velocity. The velocity-gradient equation is derived
from Newton's three-dimensional force equation. In this program the velocity-gradient
equation is used to determine the approximate velocity distribution when there is locally
supersonic velocity on the meridional midchannel stream surface.
The general velocity-gradient equation can be written

(C1)

where

cos 2 sin 2 dW
W cos a' _ W /3 + sin _ cos _
r-
m - 2w sin/3
r r dm
C

dW
b - W cos2/j- sin a + cos a cos/3 m (C2)
r dm
C

C = W sin o_ sin/3 cos/3 + r cos/3 (dw0 + 2w sin _


7
Equation (C1) is the same as equations (B13) and (B14) of reference 2 but with the addi-
tional term T(dS/dq), to allow for variation in entropy. The entropy variation is known
from the input specification of relative total-pressure-loss distribution. Entropy varia-
tion along streamlines for adiabatic flow is possible only with viscous forces. However,
these forces are primarily in Me streamwise direction so that it is reasonable and con-
sistent to neglect viscous forces along the hub-shroud mesh lines.
Outside the blade row, c = 0. This follows from the fact that d(rV0)/dm = 0 in this
region. Using this identity and the relation W 0 = V 0 - cvr, we can derive the expres-
sion

85
dW0 _ W 0 + 2wr
sin o_
dm r

When this is substituted into the expression for c, we find that c = 0.


The general velocity-gradient equations (C1) and (C2) will be applied along vertical
mesh lines in the t-direction, so that q = t. (See appendix B and fig. 3 for a descrip-
tion of the orthogonal mesh.) Then

dr
- cos (¢
dq
(c3)

dz_
-- sin q_
dq

Note that, by using equation (C3),

sin(a - (p) = sin c_ dr + cos adz "_


dq dq
(C4)

dr dz
cos(_ - _) = cos _- sina--
dq dq J

Using equations (C3) and (C4) in equations (CI) and (C2), we can get

_ In sin(_ - (¢) - + 2w sin


dW W COS2fl cos(c_ - 9o) + cos/3 dm
dW r
/ w sin2/3 fit
dt rc

cos W
+cd0+l(
at° (C5)

We can express T dS in terms of the relative velocity W and relative stagnation


conditions. We use

86
W2
W _ W Tt mm (C6)
2C
P

By using equations 0315) and (C6), we get

dT" RT" dp" +W2_ R--_-_ C dT "_


-- (C7)
TdS= Cp P" Cp \2p" - P2T"/

Now we can write the velocity-gradient equation (C5) as

dW = (aW + b)dt +e + Wf (C8)


W

where

2 . 2
a = cos /3 cos(a, - _) sm /3 cos _ + sin c_ sin/3 cos fi dO
rc r dt

b=cosl?-_mm sin(o_-_o)-2a_ sin_ cos e)+reos/3/--_m


/dw0 +2co sin
(C9)

e = Cp dT_ - co dk- Cp dT" + _ dp"

f = dT" _ _
T" _C P"
2 2 P

Equations (C8) and (C9) can be used directly within the blade row. However. out-
side the blade row. the angle /3 is not known directly so that we desire to eliminate
in favor of _, r, and W. This can easily be done since

k
W0 - o0r
r

and

87
Wo
sin _ =
W

Then cos 3 is obtained from

cos2fi = 1 - sin2_

Also, c = 0 in equation (C2) outside the blade as was shown previously. With these
substitutions, equation (C5) becomes

/ \
dW=_aW+ c +d COS /_|dt + e +Wf (C10)
\ W / W

where

a-
cos(a - _)
r
c

e _-
cos(a - _) +
r2 r2_/IX - r cor
C
2

dW
d __ m sin(a - _o)
dm

(C_I)
RT"
e = C dT!- co dk- C dT" +-- dp"
p 1 p p,,

f_ dT" R dp"

2T" 2 Cpp"

cos _ =

88
Downstream of the blade, outlet whirl should be used. Hence, in equation (Cll),
Equations (A7) to (All) are obtained directly from equa-
(rVo) isusedinsteadof _-.
0
tions (C8) to (Cll).

89
APPENDIX D

LOSS CORRECTIONS

An approximate loss correction is made by reducing the relative ideal stagnation


pressure. This loss in stagnation pressure is specified by the input at the downstream
input station by specifying either the loss fraction or the stagnation pressure. These
quantities may vary from hub to tip. If the outlet stagnation pressure is given as input,
the fractional loss of stagnation pressure is calculated. So, in all cases, the fractional
loss of stagnation pressure at the outlet input station is known from hub to tip.
For the case where the downstream stagnation pressure is given as input, the frac-
tional loss of stagnation pressure is calculated. First, the outlet total temperature T'
o
is calculated from the change in whirl along a streamline,

T' = T: + (DI)
o I C
P

Then the ratio of actual to ideal stagnation pressure is calculated from

0o ('y- l)
(D2)
(";,)
ideal V;4
Within the blade, the loss is distributed linearly or as specified by the input, from zero
at the leading edge to the fraction given or calculated at the blade trailing edge. The
loss fraction is the same whether it is expressed in terms of relative or absolute stag-
nation pressures. Thatis,

(D3)
vt !

Pideal Pideal

When this ratio is known, the density can be calculated as a function of the velocity W.
as follows:

90
, T p"
P = Pi ,'L'T (D4)
V{ I P_.tdeal

where

T = T! - w2 + 2w)_ - (c0r) 2
(D5)
I 2C
P

Equation (D4) applies the loss correction by calculating the density from the ideal den-
sity. Equation (D4) is derived as equation (B6) in reference 13. Equation (D5) is equa-

tion (3) of reference 2. The value of _,_"/_":_ideal varies at each point of the region and
is recalculated with each iteration. The value of p is used in checking continuity by
means of equation (A12).

91
APPENDIX E

DEFINING REDUCED-MASS-FLOW PROBLEM

When the midchannel stream surface has locally supersonic flow, the solution can-
not be obtained directly by solving the stream-function equation (eq. tA1)). However,
an approximate solution can be obtained by getting a reduced-flow solution and extending
this to the full-flow solution by the velocity-gradient method. This full-flow solution
depends strongly on the reduced-flow solution, so it is important to establish the condi-
tions that will give the most statable stream-function solution. With a reduced mass
flow, if both the inlet and outlet whirl and the rotational speed are reduced in the same
ratio as the mass flow, a similar flow will result. That is, the reduced-flow stream-
lines will have approximately the same angles and curvatures as the full-flow stream-
lines throughout the flow field.

Another consideration is the boundary conditions at the upstream and downstream


boundaries. Again, the flow distribution will be similar to the reduced mass flow if the
mass flow, inlet and outlet whirl, and rotational speed are all reduced by the same
ratio.

In summary, then, for the reduced-mass-flow solution, the mass flow, inlet and
outlet whirl, and rotational speed are all reduced in the same ratio. This reduction
factor is specified in the input by REDFAC, as explained inthe section INPUT.

92
APPEND_ F

INCIDENCE AND DEVIATION CORRECTIONS

The solution region is divided into three subregions: upstream, within the blade,
and downstream. These three regions should match each other; that is, there should
be no difference in the tangential momentum of a particle on either side of the lines at
the leading and trailing edges. If there were zero incidence and deviation angles, this
condition would be satisfied automatically by using the actual blade shape in the blade
region. Since there are almost always nonzero incidence and deviation angles, the
mean flow will not follow the blade surface near the leading and trailing edges. Be-
cause of this, an empirical correction is made to the blade shape near the leading and
trailing edges.
The blade-shape correction is made for a distance from the leading or trailing
edge. This distance varies between 1/6 and 1/2 of the blade chord, depending on solid-
ity. The solidity is defined to be the ratio of the true blade chord along a streamline to
the blade-to-blade distance. When there is a change in radius, the solidity differs at
the leading and trailing edges. For calculation purposes in the program the true blade
chord along an s-coordimte line is used instead of that along a streamline. For a solid-
ity of 2 or more, the distance is 1/6 of the chord; and for a solidity of 1/2 or less, the
distance is 1/2 of the blade chord. The distance varies linearly with solidity between
these limits. The correction is made to the flow angle so that tangential momentum is
continuous at the blade leading or trailing edge, and then the angle correction is lin-
early decreased over the prescribed distance within the blade. This correction must be
made to the flow angle t3. However, the blade shape in the stream-function equation is
expressed by the partial derivatives, a0/_s and _0/0t. Therefore, with the corrected
fl, the appropriate correction is made to _0/_s. It is assumed that the s-coordinate is
close to a streamline. Therefore, only _0/_s is changed, and _0/_t is not changed.
The free--stream flow angle is not known in advance at the blade leading edge.
Therefore, an iterative procedure is used to make the blade--shape correction. On the
first iteration, no blade-shape correction is made. After the first iteration, an approx-
imate solution is obtained, and the flow angle f3 is calculated throughout the region.
From the requirement of continuous tangential momentum,

From continuity,

93
w4s 0WmB
' b
where NBL is the number of blades. Hence, since tanfl = W0/Wm,

B × NBL Pbf
tan flbf = tan fifs (F1)
2_ Pfs

Equation (F1) is used to calculate flbf" The difference fibf - fib (see fig. 12) is then
varied linearly within the blade for the specified distance. We can express fl in terms
of the partials of O by

DR ]
tan_bf=r--=r
do
dm f
cos(oz- ¢) + "--=
at
sin(a- q_)]
J

When this is solved for (a0/aS)bf, we obtain

tan _bf a0
sin(_ - _)
r at
(F2)
f cos (_- (p)

Equation (F2) is used to calculate (a0/aS)b f from the linearly varying _ at each mesh
point. The calculations for these blade-shape corrections are done by subroutine
INDEV for each iteration. When REDFAC is less than 1, the final transonic calcula-
tion uses the same method to correct fl within the blade (subroutine LINDV), but
00/as is not needed.
APPENDIX G

BLADE SURFACE VELOCITIES AND BLADE-TO-BLADE AVERAGE DENSITIES

The blade surface velocities can be calculated once a meridional midsurface solu-

tion is obtained. The blade--to--blade velocity gradient _W/a0 depends on dW0/dm.


To obtain the desired relation, we use the velocity-gradient equations (C1) and (C2). In

this case, q = 0 and dr/d0 = dz/d0 = 0. Also, it is assumed that TI, X, and the en-
tropy S are constant in the blade--to-blade direction. With this, equations (C1) and (C2)
become

0W
m=Wsin _ sin_ cosfl +r cosfl (GI)
D0

Using the fact that V 0 = W 0 + wr and W 0 = W sin_, we can rewrite equation (G1) as

DW d(rV0)
- cos fl _ (G2)
00 dm

Since only a midchannel solution is obtained, we assume that cos _ and d(rV0)/dm
are constant from blade to blade. This means that W varies linearly from blade to
blade. (This may be in considerable error near the leading or trailing edge.) With this
assumption, then

_ Wl + Wtr
(G3)
Wmid 2

Integrating (G2) from blade to blade (from 0 = Ol to 0 = 0tr ) and using (G3), we obtain

B cos fl d(rVo)
Wtr = Wmi d +
2 dm
(64)

B cos _ d(rV0)
Wl = Wmid - 2 dm

Note that equation (G4) is very close to that developed by Stanitz (eqs. (16) and (17),
ref. 16).

95
Equation (G4)is used in subroutine BLDVEL to obtain blade surface velocities from
the midchannel meridional solution.
It is desirable to consider the blade-to-blade variation in density to satisfy continu-
ity for the blade passage. Equation (G4) gives irfformation we need to do this. Since the
midchannel solution is considered to be representative of all meridional-plane stream
surfaces, we can consider the solution to be based on average blade-to-blade conditions.
This means that in equation (B10) average blade-to-blade velocity anddensity shouldbe
used. Equation (B10) becomes

0u_ rB @Wt)
0s w av
(G5)

0t w v

If we use Simpson's rule to calculate the average value, we obtain

PWs )a v - OlWs_ l +4PmidWs_mid


6 +OtrWs_ tr (G6)

The velocity component WS can be expressed in terms o£ the velocity and flow angles
by

W = W cos ¢ cos((_ - _) (G7)


s

This relation holds also on the blade surfaces, since the flow angles are assumed to be
constant in the blade-to-blade (0) direction, so that

W s,l = W l cos _ eos(e- _)


(G8)

Ws, tr = Wtr cos ¢_ cos(or - _o)

By using equations (G3), (G4), and (GS) in equation (G6), we can express (pW s v as

oW s) = OavWs + Pl - Ptr cos/3 cos(s- q)){W l - Wtr) (G9)


av _-2

96
where the subscript mid is omitted from Ws, and

- Pl + 4Pmid + Ptr
Pay

\_en a solution of equation (A1) is obtained, all quantities in equation (G9) can be esti-

mated from the previous iteration, except for W s. So we solve for W S in equa-
tion (Gg) and use equation (G5) to obtain

_)u
VV_

W - 3t (Ol - 0t r) cos _ cos(o ' - _o)(_) - Wtr) (GI0)


s rBp av 12p av

This is the equation used in sub_x)utine NEWRHO to calculate W from 0u/0t. On the
s
first iteration, Pay = O'i'deal is used, and the second term is omitted. After this, the
values of Omid, Ptr' Ol' Wl' Wtr' cos 5, and eos(a - _0) from the previous iteration
are used.

In a parallel manner, we can derive the equation for Wt:

_u

w w (p/ - Ptr) cos 3 sin(a - ¢)(_ - Wtr)


Wt _ 0s (Gll)
rBp av 120 av

97
APPENDIX H

SYMBOLS

a coefficient in velocity-gradient equations (A7) and (C1)

B tangential space between blades, rad

b coefficient in velocity-gradient equations (A7} and (C1)

C specific heat at constant pressure, J/(kg)(I4)


P
C coefficient in velocity-gradient equations (A7) and (C1)

d coefficient in velocity-gradient equation (A7)

e coefficient in velocity-gradient equation (A7)

F vector normal to midchannel stream surface and proportional to tangential pres-


sure gradient, N/kg

f coefficient in velocity-gradient equation (A7)

I rothalpy, C T!- wk, meters2/sec 2


p 1
i incidence angle, rad

m meridional streamline distance, meters

n unit vector normal to midchannel stream surface

P pressure, N/meter 2

q distance along an arbitrary space curve, meters

R gas constant, J/(kg)(K)

r radius from axis of rotation, meters

r radius of curvature of meridional streamline, meters


c

S entropy, J/(kg) (K)

S distance along orthogonal mesh lines in throughflow direction, meters

T temperature, K

t distance along orthogonal mesh lines in direction across flow

U normalized stream function

V absolute fluid velocity, meters/sec

W fluid velocity relative to blade, meters/sec

98
w mass flow, kg/sec
z axial coordinate, meters
angle betweenmeridional streamline and axis of rotation, rad, see fig. 4
fl angle between relative velocity vector and meridional plane, rad; see fig. 4

_/ specific-heat ratio

coefficient in stream-function equation, defined in eq. (A3)

0 relative angular coordinate, rad; see fig. 4

prerotation, (rV0) i, meters2/sec

coefficient in stream-function equation, defined in eq. {A2)

p density, kg/meter 3

(p angle between s-distance line and axis of rotation, rad, see fig. 3

stream function, kg/sec

w rotational speed, rad/sec_ see fig. 4

Subscripts:

av average blade-- to-blade

b blade

bf blade flow

cr critical

fs free stream

h hub

i inlet

l blade surface facing direction of positive rotation

m component in direction of meridional streamline

mid midchannel blade-- to-blade

O outlet

r component in radial direction

S component in s- direction

t tip, or component in t-direction

tr blade surface facing direction of negative rotation

99
z componentin axial direction
0 component in tangential direction

Superscripts:

' absolute stagnation condition

" relative stagnation condition

100
REFERENCES

1. Wu, Chung-Hua: A General Theory of Three-Dimensional Flow in Subsonic and


Supersonic Turbomachines of Axial-, Radial-, and Mixed-Flow Types. NACA TN
26O4, 1952.

2. Katsanis, Theodore: Use of Arbitrary Quasi-Orthogonals for Calculating Flow Dis-


tribution in the Meridional Plane of a Turbomachine. NASA TN D-2546, 1964.

3. Katsanis, Theodore: FORTRAN Program for Calculating Transonic Velocities on a


Blade-to-Blade Stream Surface of a Turbomachine. NASA TN D-5427, 1969.

4. Katsanis, Theodore; and McNally, William D. : FORTRAN Program for Calculating


Velocities and Streamlines on the Hub-Shroud Mid-Channel Flow Surface of an
Axial- or Mixed-Flow Turbomachine. I- User's Manual. NASA TN D-7343.
1973.

5. Katsanis, Theodore; and McNally, William D. : FORTRAN Program for Calculating


Velocities and Streamlines on the Hub-Shroud Mid-Channel Flow Surface of an
Axial- or Mixed-Flow Turbomachine. II- Programmer's Manual. NASA TN
D-- 7344, 1974.

6. Katsanis, Theodore; and McNally, William D. : Revised FORTRAN Program for


Calculating Velocities and Streamlines on the Hub--Shroud Midchannel Stream
Surface of Axial-, Radial-, or Mixed- Flow Turbomachine or Annular Duct.
II- Programmer's Manual. NASA TN D-8431, 1977.

7. McNally, William D. : FORTRAN Program for Generating a Two--Dimensional


Orthogonal Mesh Between Two Arbitrary Boundaries. NASA TN D-6766, 1972.

8. McCracken, Daniel D. ; and Dorn, William S. : Numerical Methods and FORTRAN


Programming. John Wiley & Sons, Inc., 1964.

9. Varga, Richard S. : Matrix Iterative Analysis. Prentice-Hall, Inc., 1962.

10. Shapiro, Ascher H. : The Dynamics and Thermodynamics of Compressible Fluid


Flow. Vol. I, Ronald Press Co., 1953.

11. Crouse, James: Computer Program for the Definition of Transonic Axial-Flow
Compressor Blade Rows. NASA TN D-7345, 1973.

12. Johnsen, Irving A. ; and Bullock, Robert O., eds. : Aerodynamic Design of Axial-
Flow Compressors. NASA SP-36, 1965.

13. Katsanis, Theodore: FORTRAN Program for Quasi-Three-Dimensional Calculation


of Surface Velocities and Choking Flow for Turbomachine Blade Rows. NASA TN
D-6177, 1971.

i01
14. Mechtly, E. A. : The International System of Units: Physical Constants and Con-
version Factors (Second Revision}. NASA SP-7012, 1973.

15. Walsh, J. L.. Ahlberg, J. H. ; and Nilson, E. N.: Best Approximation Properties
of the Spline Fit. J. Math. Mech., vol. 11, no. 2, 1962, pp. 225-234.

16. Stanitz, John D. _ and Prian, Vasily D. : A Rapid Approximate Method for Deter-
mining Velocity Distribution on Impeller Blades of Centrifugal Compressors.
NACA TN2421, 1951.

102 NASA-Langley, 1977 E-8734

You might also like