Jiaa TR - 111: Joint Institute FOR A R - R - Nautics AND Acoustics

https://ntrs.nasa.gov/search.jsp?
R=19940024221 2019-06-13T17:04:29+00:00Z
NASA-CR-195699
JOINT INSTITUTE FOR A R__R__NAUTICS AND ACOUSTICS
National Aeronautics and _'7_/e"=t

• 2_f.,d_
-I- ..'__
Space Administration
JIAA TR - 111
Ames Research Center Stanford University
A Method for the Modelling of Porous and Solid Wind

Tunnel Walls in Computational Fluid Dynamics Codes
By
Thomas John Beutner
Stanford University
Department of Aeronautics and Astronautics
Stanford_ CA 94305
December 1993
(NASA-CR-_9_699) A _eTHOD FOR THE N94-28724
MODELLING OF POROUS ANO SOLID WIND
TUNNEL WALLS IN COMPUTATIONAL FLUID
DYNAMICS CODES (Stanford Univ.) Unclas
124 p
G3/34 0000876
Acknowledgements
This research has been supported by a grant (NCC 255) from NASA Ames Research
Center. Support has also been provided by the U.S. Air Force Palace Knight Program.
Abstract
Porous wall wind tunnels have been used for several decades and have proven
effective in reducing wall interference effects in both low speed and transonic testing. ,
They allow for testing through Mach 1, reduce blockage effects and reduce shock wave
reflections in the test section. Their usefulness in developing computational fluid
dynamics (CFD) codes has been limited, however, by the difficulties associated with
modelling the effect of a porous wall in CFD codes. Previous approaches to modelling
porous wall effects have depended either upon a simplified linear boundary condition,
which has proven inadequate, or upon detailed measurements of the normal velocity near
the wall, which require extensive wind tunnel time.
The current work was initiated in an effort to find a simple, accurate method of
modelling a porous wall boundary condition in CFD codes. The development of such a
method would allow data from porous wall wind tunnels to be used more readily in
validating CFD codes. This would be beneficial when transonic validations are desired,
or when large models are used to achieve high Reynolds numbers in testing.
A computational and experimental study was undertaken to investigate a new
method of modelling solid and porous wall boundary conditions in CFD codes. The
method utilized experimental measurements at the walls to develop a flow field solution
based on the method of singularities. This flow field solution was then imposed as a
pressure boundary condition in a CFD simulation of the internal flow field. The
effectiveness of this method in describing the effect of porosity changes on the wall was
investigated. Also, the effectiveness of this method when only sparse experimental
measurements were available has been investigated. The current work demonstrated this
approach for low speed flows and compared the results with experimental data obtained
from a heavily instrumented variable porosity test section.
ii
The approach developed was simple, computationally inexpensive, and did not
require extensive or intrusive measurements of the boundary conditions during the wind
tunnel test. It may be applied to both solid and porous wall wind tunnel tesSs.
oo°
III
Table of Contents
Acknowledgements .................................................................................................. i
Abstract ................................................................................................................... ii
Table of Contents ..................................................................................................... iv
List of Figures .......................................................................................................... vi
List of Tables ............................................................................................................ xi
Nomenclature ........................................................................................................... xii
Chapter 1 Introduction ............................................................................................. 1
1.1 Background ................................ ................................................................. 1
1.2 Motivation for Modelling the Porous Wall Boundary Condition ............... 2
1.3 Previous Approaches to Modelling Porous Walls ....................................... 4
1.4 Current Approach ........................................................................................ 5
Chapter 2 Porous Wall Theory ................................................................................ 8
2.1 Introduction ................................................................................................. 8
2.2 Development of Classical Porous Wall Boundary Condition ..................... 8
2.3 Methods of Describing the Effect of a Porous Wall in CFD Codes ............ 13
2.4 Theoretical Approach of this Work .......................................................... 14
2.5 Extension to Three Dimensions .................................................................. 15
Chapter 3 Two-Dimensional Governing Equations and Navier-Stokes Solver ......
3.1 Introduction ................................................................................................. 1_
3.2 Governing Equations ................................................................................... 14"
3.3 Numerical Solver ........ ................................................................................. 20
3.4 Boundary Conditions ................................................................................... 21
3.5 Generation of Computational Grids ............................................................ 23
Chapter 4 Experimental Apparatus ......................................................................... 25
4.1 Wind Tunnel Test Section ........................................................................... 25
iv
4.2 Wind Tunnel Models ................................................................................... 26
4.3 Test Section Instrumentation ....................................................................... 27
4.4 Wind Tunnel Flow Quality .......................................................................... 27
4.5 Error Analysis .............................................................................................. 28
Chapter 5 Results ..................................................................................................... 33
5.1 Two-Dimensional Case ............................................................................... 33
5.1.1 Preliminary Computational Studies .................................................... 33
5.1.2 CFD Simulations of Experiments..i .................................................... 35
5.1.3 Effect of Changes in Porosity Parameters .......................................... 36
5.1.4 Effect of Changes in Singularity Placement ....................................... 38
5.1.5 Comparison of Alternate Boundary Conditions for Solid Wall

Simulations ......................................................................................... 38
5.1.6 Grid Refinement Effects ..................................................................... 39
5.1.7 Effects of Using Sparse Data .............................................................. 40
5.1.8 Flow Inclinations Near the Boundaries .............................................. 42
5.2 Three-Dimensional Case ............................................................................. 42
5.2.1 Method of Singularities Comparison to Experimental Data .............. 43
5.2.2 Effects of Sparse Data and Reduced Numbers of Singularities ......... 43
5.2.3 Effects of Uncertainties in Porosity Parameter .................................. 46
Chapter 6 Conclusion .............................................................................................. 102
References ................................................................................................................ 104
v
List of Figures
1 Arrangement of Panels for Panelling Code ..................................................... 17
2 Two-Dimensional, 250 x 70 CFD Grid ........................................................... 24
3 Exploded View of Variable Porosity Wall ...................................................... 29
4 Schematic Diagram of Wind Tunnel ............................................................... 30
5 Photograph of Wind Tunnel Test Section ....................................................... 30
6 Boeing Advanced Transport Airfoil Cross Section ......................................... 31
7 Empty Tunnel Flow Survey, Solid Wall Test Section .................................... 31
8 Empty Tunnel Flow Survey, Porous Wall Test Section .................................. 32
9 Velocity Perturbations on Walls predicted by Navier-Stokes Solver ............. 47
10 Errors in Matching Pressure Profiles on Walls ............................................... 47
11a Method of Singularities Matches to Experimental Data, o_ = 5 degrees ......... 48
11b Comparison of Experimental and CFD Pressure Profiles on Airfoil,

(x = 5 degrees ................................................................................................... 48
12a Method of Singularities Matches to Experimental Data, ¢x = 6 degrees ......... 49

¢x = 6 degrees ................................................................................................... 49
13a Method of Singularities Matches to Experimental Data, (x = 7 degrees ......... 50

o_ = 7 degrees ................................................................................................... 50
14a Method of Singularities Matches to Experimental Data, ¢x = 8 degrees ......... 51

c_ = 8 degrees ................................................................................................... 51
15a Method of Singularities Matches to Experimental Data, ot = 9 degrees ......... 52

¢x = 9 degrees ................................................................................................... 52
vi
16a Method of Singularities Matches to Experimental Data, a = 10 degrees ....... 53

o_ = 10 degrees ................................................................................................. 53
17a Method of Singularities Matches to Experimental Data, o_ = 11 degrees ....... 54

= 11 degrees ................................................................................................. 54
18 Root Mean Square Errors as a Function of Porosity Parameter ...................... 55
19 Method of Singularities Fit to Experimental Data for P = 0. and P = 1 .......... 55
20 Airfoil Pressure Profiles Resulting from Boundary Conditions

Using P = 0. and P = 1............. ,....................................................................... 56
21 Effect of Porosity Parameter Changes on Singularity Strengths ..................... 56
22 Effect of Singularity Location on Errors in Matching Wall Pressures ............ 57
23 Grid Used for Viscous Wall CFD Simulations ............................................... 57
24 Grid Used for Free-Air CFD Simulations ....................................................... 58
25 Effect of Using Different Boundary Conditions to Simulate the

Solid Wall o_ = 8 degrees Wind Tunnel Test ................................................... 59
26 Refined, 350 x 105 CFD Grid ......................................................................... 59
27 Comparison of Airfoil Pressure Profiles using Standard and Refined Grid .... 60
28a Comparison of Method of Singularities Fit to Sparse and Fine Data on

Walls, o_ = 5 degrees, Solid Wall .................................................................... 61
28b Comparison of Airfoil Pressure Profiles Using Boundary Conditions based

on Sparse and Fine Data, _ = 5 degrees, Solid Wall ....................................... 61

Walls, cx = 6 degrees, Solid Wall .................................................................... 62

on Sparse and Fine Data, _ = 6 degrees, Solid Wall ....................................... 62

Walls, o_ = 7 degrees, Solid Wall .................................................................... 63

on Sparse and Fine Data, o_ = 7 degrees, Solid Wall ....................................... 63
vii
31a Comparison of Method of Singularities Fit to SparseandFine Data on
Walls, tx = 8 degrees, Solid Wall .................................................................... 64

on Sparse and Fine Data, o_ = 8 degrees, Solid Wall ....................................... 64

Walls, tx = 9 degrees, Solid Wall .................................................................... 65

on Sparse and Fine Data, tx = 9 degrees, Solid Wall ....................................... 65

Walls, tx = 10 degrees, Solid Wall .................................................................. 66

on Sparse and Fine Data, tx = 10 degrees, Solid Wall ..................................... 66

Walls, tx = 11 degrees, Solid Wall .................................................................. 67

on Sparse and Fine Data, o_ = 11 degrees, Solid Wall ..................................... 67

Walls, tx = 5 degrees, Porous Wall .................................................................. 68

on Sparse and Fine Data, tx = 5 degrees, Porous Wall .................................... 68



Walls, t_ = 7 degrees, Porous Wall .................................................................. 70


,o°
VIII
on Sparse and Fine Data, c_ = 8 degrees, Porous Wall .................................... 71

Walls, c_ = 9 degrees, Porous Wall .................................................................. 72 '

on Sparse and Fine Data, _ = 9 degrees, Porous Wall .................................... 72

Walls, o_ = 10 degrees, Porous Wall ................................................................ 73

on Sparse and Fine Data, ot = 10 degrees, Porous Wall .................................. 73

Walls, ct = 11 degrees, Porous Wall ................................................................ 74

on Sparse and Fine Data, o_ = 11 degrees, Porous Wall .................................. 74
42 Normal Velocities Near the Walls, tx = 5 degrees ........................................... 75
43 Normal Velocities Near the Walls, tx = 8 degrees ........................................... 75
44 Three-Dimensional Pressure Profiles and Method of Singularities Fit to

Pressure Profile, Upper Wall, Solid Wall Test, (x = 20 degrees ...................... 76-77

Pressure Profile, Lower Wall, Solid Wall Test, o_ = 20 degrees ..................... 78-79

Pressure Profile, Side Wall, Solid Wall Test, o_ = 20 degrees ......................... 80

Pressure Profile, Upper Wall, Porous Wall Test, _ = 20 degrees ................... 81-82

Pressure Profile, Lower Wall, Porous Wall Test, cc = 20 degrees ................... 83-84

Pressure Profile, Side Wall, Porous Wall Test, o_ = 20 degrees ...................... 85
50 Locations of 342 Measurements Used in Developing Method of

Singularities Solutions ..................................................................................... 86
51 Locations of 223 Measurements Used in Developing Method of
ix
Singularities Solutions.....................................................................................
87
52 Locations of 136MeasurementsUsedin Developing Method of
88
53 Locations of 38 MeasurementsUsed in Developing Method of
89
54 Root-Mean-SquareErrors in Matching Three-DimensionalWall Pressure
Profiles, Solid Wall Data.................................................................................
90
55 Root-Mean-SquareErrors in Matching Three-Dimensional Wall Pressure

Profiles, PorousWall Data..............................................................................90
56 Three-Dimensional PressureProfiles and Method of Singularities Fit to
PressureProfile Basedon SparseData, Upper Wall .......................................91-92
PressureProfile Basedon SparseData, Lower Wall ...................................... 93-94
58 Three-DimensionalPressureProfiles andMethod of Singularities Fit to
PressureProfile Basedon SparseData, SideWall ..........................................95
59 Three-Dimensional PressureProfiles andMethod of Singularities Fit to
PressureProfile Basedon SparseData, Upper Wall .......................................96-97
PressureProfile Basedon SparseData, Lower Wall ...................................... 98-99
61 Three-DimensionalPressureProfiles andMethod of Singularities Fit to
PressureProfile Basedon SparseData, SideWall ..........................................100
62 Root-Mean-SquareErrors in Matching Pressureon Walls as a Function of

Porosity Specified in Method of Singularities................................................ 101
x
List of Tables
Root-Mean-Square Errors in Matching Pressure Coefficients on Walls

and Model ........................................................................................................ 36
2 Root-Mean-Square Errors in Matching Pressure Coefficients on Walls

and Model Using Sparse Data to Determine Singularity Strengths ................ 41
3 Singularities Retained in Three-Dimensional Method of Singularities .......... 45
xi
Nomenclature
C airfoil chord
Cp pressure coefficient
CpPlenum plenum pressure coefficient
h wind tunnel test section height
m line source strength
M Mach number
n outward normal coordinate direction
P pressure
poo freestream pressure
P porosity parameter
PL porosity of lower wall
PS porosity of side wall
PU porosity of upper wall
Uoo freestream velocity
U, V two-dimensional Cartesian velocity components; u is streamwise
component, v is vertical component
U,V,W three-dimensional Cartesian velocity components; u is streamwise
component, v is spanwise component, w is vertical component
W tunnel width
complex velocity function
Vo empty tunnel normal velocity at walls
two-dimensional Cartesian coordinates, origin at airfoil leading edge

X, y
(a=O), x is the streamwise coordinate, y is the vertical coordinate
three-dimensional Cartesian coordinates, origin at wing root leading edge

x, y, z
(_----0), x is the streamwise coordinate, y is the spanwise coordinate, z is
the vertical coordinate
xii
Xo, YO origin of Cartesian coordinates
Yo semispan of three-dimensional singularity
angle of attack
I_M 2
artificial compressibility parameter
perturbation velocity potential
potential due to a line source
potential due to a horseshoe vortex
potential due to a point vortex
potential due to a point x-doublet
potential due to a point source
potential due to a point y-doublet
two-dimensional vortex singularity strength
F horseshoe vortex strength
two-dimensional x-doublet singularity strength
_T two-dimensional source singularity strength
CO two-dimensional y-doublet singularity strength
xiii
Chapter I
Introduction
1.1 Background
Ventilated wall wind tunnels have been in use for several decades and have been
useful in reducing wall interference effects at subsonic and transonic speeds and
allowing for testing through Mach 1. A series of improvements have been made to the
earliest ventilated wind tunnels leading to modem porous wall test sections. The
usefulness of porous wall wind tunnels for computational fluid dynamics (CFD)
validation efforts has been limited, however, by difficulties associated with modelling
the porous wall boundary condition.
It has long been recognized that the corrections to wind tunnel data for open and
closed test sections were of opposite signs [1]. Furthermore, in transonic testing, shock
waves which impinge on a solid boundary are reflected as shock waves, whereas shocks
which impinge on a free air boundary in an open jet test section are reflected as
expansion waves. Theodorsen suggested that a wind tunnel might be constructed which
would reduce wind tunnel wall interference by using a partially open boundary condition
in which one, two or three walls would be removed from a solid wall test section [2]. •
Wright and Ward tested one of the first successful ventilated wall wind tunnels which
used several streamwise slots in the tunnel walls [3]. They found that blockage
interference was reduced with this wind tunnel. In addition, they found that ventilated
walls alleviated choking problems at transonic speeds and permitted testing through
Mach 1. These two effects haveled to ventilated wall test sections being used for both
low speedand transonicwork [4]. The work by Wright andWard led to other ventilated
wind tunnels, alsousing streamwiseslots to ventilate the test section walls. It was soon :!
realized, however, that streamwise slots allowed for the reflection of shock and
expansion waves from the walls. These reflected waves could impinge on the model and
result in data which was uncorrectable [5].
Porous wall tunnels alleviated this problem by significantly reducing the shock
reflections from the walls [6]. The porous wall was constructed with a pattern of small,
discrete holes in the wall. The differential resistance wall was a further refinement to the
porous wall concept, using holes with axes inclined to the normal. The differential
resistance wall was found to provide cancellation of both shock and expansion waves
and allowed for inflow and outflow resistance of the wall to be balanced [1].
1.2 Motivation for Modelling the Porous Wall Boundary Condition
As high speed computers have increased the flexibility and capability of
computational modelling of flow fields, a new emphasis has been placed on obtaining
wind tunnel data which may be used to calibrate and validate computational fluid
dynamics (CFD) codes [7]. Increasingly, there has been a trend toward modelling entire
wind tunnel flow fields, including support struts and wind tunnel walls [8, 9]. This has
led to an increased use of solid wall wind tunnels in transonic testing. The simplicity of
modelling a solid wall boundary condition has made their use attractive despite the
disadvantages of substantial wall interference [10].
Porous wall wind tunnels have long been used for both low speed and transonic
wind tunnel testing because of their desirable effect of reducing wind tunnel wall
interference. With mounting evidence of the importance of Reynolds number on flap
gap sizing and maximum lift predictions [11, 12], the need for performing high
Reynolds number tests to validate CFD codes which will be used in the design process is
2
of growing importance. While a solid wall wind tunnel may be used for such tests, wall
interference can become a limiting factor on the size of models.
In general, for test sections of identical dimensions, a porous wall wind tunnel
will allow testing of larger models at transonic speeds without shock reflections from the
tunnel walls impinging on the model. In addition, porous walls may eliminate shock
boundary layer interactions on the walls and reduce such interactions on the model by
eliminating reflected shocks from the walls. Thus, the use of porous wall data in CFD
validation may reduce the grid refinement required near the wind tunnel walls while
providing a more realistic assessment of the applicability of a turbulence model to free
air calculations. Ultimately, the goal of many CFD validation efforts is to develop CFD
codes which are capable of predicting free air performance. These codes may be useful
for design purposes even if they are not sophisticated enough to resolve multiple shock
boundary layer interactions which may be associated with shock reflections from solid
wall wind tunnels.
Some attempts have been made to model discrete slots in CFD codes either by
modelling the slots in an approximate manner or by solving coupled equations to
describe the boundary condition [13, 14]. However, porous wall wind tunnels have been
shown to be superior to slotted walls at cancelling shock wave reflections [6]. Porous
walls pose a problem in CFD validations, however, since the modelling of the wall
boundary is more difficult for a porous wall wind tunnel.
The intricacies of the porous wall make it prohibitive to depict the actual wall
geometry in a viscous CFD grid. Since the porous wall wind tunnel may have several
thousand holes on the walls, modelling the individual holes and the viscous effects
associated with each hole in a CFD grid is not possible given the current limitations on
computer speed and memory. Thus, the effect of the porous walls must be dealt with
either by correcting the test data to free air conditions, or by modelling the porous wall
by appropriate means in the CFD code.
3
1.3 Previous Approaches to Modelling Porous Walls
Numerous approaches for correcting porous wall data to free air conditions have
been proposed. These methods use a variety of approaches, based on model pressure
and force measurements [15], wall boundary pressure or velocity measurements, or
pressure rail measurements [4, 16]. These methods generally produce a global
correction to the velocity and angle of attack, based on classical reflection techniques
[1]. The test data is then taken to be equivalent to data from a test in free air at the
corrected angle of attack and velocity. Additional corrections to drag and moment
coefficients and Mach number are sometimes included [1]. This results in a useful
comparison of bulk flow measurements, such as lift and drag coefficients. However,
these methods are of limited use in performing CFD validations and calibrations because
they have the effect of altering the entire flow field. When performing CFD validations,
it is desirable to compare as much of the flow field as possible [7].
Other approaches have sought to develop a boundary condition which may be
used at the porous wall boundary in the CFD code. These methods have depended on
either a simple, universal boundary condition, [1, 6] or on detailed measurements of the
boundary conditions during the wind tunnel test which are then imposed as boundary
conditions in the CFD code [17-21].
A universal boundary condition has proven difficult to determine experimentally,
and to implement computationally. The porosity parameter contained in such a
boundary condition can only be determined by extensive dedicated testing. In addition,
this parameter may be different for otherwise identical top and bottom walls, and it may
depend on Mach number, stagnation pressure, model size and orientation [22].
Additionally, such a boundary condition may be destabilizing in CFD codes and can
actually prevent convergence [23, 24].
Most current efforts in ventilated wall interference research have been directed
toward making detailed measurements of velocity or pressure in the wind tunnel and
4
using theseto develop a boundarycondition in CFD codes. The approachusedby King
and Johnson in modelling a slotted wall boundarycondition was to use a rail suspended
midway between the model and the tunnel wall to make pressuremeasurementsatevery
point in the flow corresponding to a boundary point in the CFD code [17]. Later, to
allow for additional grid refinement, a spline fit was made to this data as a meansof
interpolating pressure measurements [18]. This approach required extensive

measurementsof the pressureon planes away from the wall and has not been extended
to three dimensions.
Jacocksuseddetailed measurementsof smile pressure,normal velocity adjacent
to the wall and boundary layer thickness at the wall to develop an equivalent inviscid
normal velocity profile basedon boundary layer integration along the wall [19]. Crites
and Rueger extended this method for limited three-dimensional problems by using
extensive pressure measurementson the walls and extensive calibrations of the tunnel
walls to estimate the normal velocity through the porous wall and, again, develop an
equivalent inviscid normal velocity profile. This profile was then usedas a boundary
condition in numerical solvers for the purposeof developing incremental corrections to
wind tunnel data [20, 21]. These approaches required detailed measurements of the
wall boundary conditions and large amounts of dedicated test lime for calibration at each
plenum pressure, Mach number and Reynolds number, making them costly in practice.
In most practical cases, the extensive measurements and calibrations required at each
test condition have limited these techniques to two-dimensional flows.
1.4 Current Approach
The approach presented here represents an attempt to describe the effects of a
porous wall in a CFD code based on sparse measurements of the flow field in the wind
tunnel. The usefulness of this theory has been demonstrated by carrying out wind tunnel
experiments in a heavily instrumented, low speed (M = .07) research wind tunnel and
performing CFD simulations of the experiments. This test section utilized variable
5
porosity walls and allowed for extensivepressureandvelocity measurementsto be made
near the boundariesof the test section.
The theoretical approach to describing the porous wall boundary utilized the
classical porous wall boundary condition as a meansof interpolating and extrapolating
boundary conditions measuredduring wind tunnel tests. A solution for the flow nearthe
walls, basedon the method of singularities, was developedusing pressuremeasurements
made on the walls during wind tunnel tests. The singularity expressionsused in this
method satisfied the classical porous wall boundary condition. The method of
singularities solution was then used to specify the pressure at CFD boundary point
locations.
The current work extended the classical method of singularities by including
higher order singularities. The current work also differs from classical approachesin
that it attempted to model the boundary condition in the CFD code, rather than applying
corrections to the databasedon measurementsof the boundarycondition. Furthermore,
unlike some limited previous attemptsat modelling the boundarycondition, the current
work utilized experimental pressure measurementson the walls and used classical
porous wall theory to extend the measurements.This approachdid not require extensive
calibrations of the wind tunnel walls, and used static pressure measurementson the
walls, which may be obtainedrapidly during wind tunnel tests.
In two dimensions, this approachwas developedusing analytic expressionsfor
the singularity strengths. In three dimensions, a numerical solution was required to
determine the pressure profiles induced by singularities in the presenceof the porous
wall boundary condition. Once obtained, however, these numerical solutions may be
usedfor multiple tests.
The current work consisted primarily of two-dimensional, low speed studies.
Experimental work was conducted in the Stanford Low Speed Wind Tunnel and
computational studies were performed using an incompressibleflow solver. Although
6
the Mach number range of the research tunnel used for this work was very low, some
consideration was given to the implementation of this method for higher speed flows.
Limited consideration was also given to the extension of this method to three-
dimensional flows.
7
Chapter 2
Porous Wall Theory
2.1 Introduction
One of the major advantages of porous wall wind tunnels is their ability to reduce
or eliminate shock wave reflections in transonic testing. In addition, they have the effect
of reducing wall interference for low speed flows, making them useful for cases where
substantial wall interference is expected in low speed testing [25]. Classical approaches
to correcting data from porous wall wind tunnels are slowly being replaced by efforts
aimed at modelling the porous wall boundary condition in CFD codes.
2.2 Development of Classical Porous Wall Boundary Condition in Two Dimensions
It has widely been assumed that the flow through the porous walls is basically
viscous in nature. This has led to a simple theory, based on an analogy to pipe flow, for
describing the porous wall boundary condition [1, 6]. This boundary condition assumes
that the airfoil perturbations are sufficiently small near the walls so that linearization
applies up to high subsonic Mach numbers near the walls. An implicit restriction of this
theory is that the flow at the walls must be subsonic. In practice, however, flow near the
walls has been found to be subcritical at freestream Mach numbers up to .9, lift
coefficients as high as 1.5 and a height to chord ratio of 4 [26].
Classical porous wall theory [1] typically uses a point source, point vortex, and
point source and vortex doublets to describe the free-air flow past a lifting airfoil, as
given in Equation 2.1.
8
qbF=Oo + _' + Ott + Oco (2.1)
where:
13 In4x 2 + (l_y)2
7 a tan I_Y (2.3)

q_' = 2_ x
Ix x (2.4)
*tt = 2_13 x 2 +(l_y)2
to 13y (2.5)
*t_= 2"--'_ x2+(_y) 2
Note that Equations 2.4 and 2.5 may be interpreted as first derivatives with respect to x
of Equations 2.2 and 2.3, respectively.
The potential _ is a solution of the linearized potential equation:
2
[_2 _)2(_F + 0 (_F _ (2.6)
3"-7- _ - 0
except at the origin (x = y = 0), where _ is singular. In addition,
3*F 0*F --*0 as r=4x 2 +(_y)2 ----)oo (2.7)

3x ' 3y
An interference potential, _¢, which accounts for the interference caused by the
presence of the walls must be solved for. Again, in the region between the walls,
-2 < y < , _V¢ must also satisfy the linearized potential equation:
_2 32t_w 32_w (2.8)

_x 2 + _d2 = 0
The free air potential and interference potential together form the potential
function 0:
t_ = 0tlF + _W (2.9)
9
which is requiredto satisfy the following boundaryconditions:
_ h
Pu _)_+ m = 0 at y=--
_x _t 2 (2.10)
h
PL _)(_ _)(_ - 0 at y = ---
_ 2 (2.11)
The basis for these boundary conditions is easily seen. The x-axis is aligned with the
freestream flow in the wind tunnel so that _y = v, and = u. According to linearized
small perturbation theory, u should be proportional to the difference between free stream
and local pressure. If the plenum is maintained at freestream static pressure, this will
also be the pressure difference across the wall. Based on an analogy to pipe flow, the
velocity normal to the wall, v, should be proportional to the pressure difference existing
across the wall. The constant of proportionality is the porosity parameter, PU for the
upper wall and Pt. for the lower wall. Different porosity parameters may be used for the
upper and lower walls since the flow through the two walls is different in character. In
the case of the upper wall, there is a suction over most of the wall resulting in flow into
the test section. This flow comes from the plenum where the flow is essentially at rest.
For the lower wall, there is an increased pressure, resulting in flow out of the test
section. This flow has considerable streamwise momentum. The result is that flow
through the lower wall often separates within the holes in the wall [1, 20]. Thus,
identical upper and lower walls may offer differing resistance to flow through the walls.
In part, this may be compensated for by inclining the holes in the wall at an angle with
respect to the normal direction. The different porosity parameters specified on each wall
offer another means of accounting for this difference.
Analytic expressions for the velocities induced by simple singularities in the
presence of the boundary conditions of Equations 2.10 and 2.11 have been developed [ 1,
27] and are summarized here.
10
The wind tunnelcoordinatesarenondimensionalizedas,
X=I x - Y
y_-- (2.12)
h
and the complex position function Z is def'med as,
Z =X+iY (2.13)
and the complex velocity notation is def'med as,
WF = /g(_----E i 0(_F (2.14)

0X _)Y
O*w i._w (2.15)

W W =_-
_)X OY
W = W w + WF - 0t_ "0_ (2.16)

0X 1 by
With the form of the free-air potential due to a singularity known, it remains to
determine the form of the complex velocity function W w.
The complex velocity funtion W w may be found by the method of images. The
approach of Reference 1, which is followed here, determines the interference potential
arising from a singularity placed between two infinite boundaries described by
Equations 2.10 and 2.11. As this approach may be found in detail in the literature [1,
27], only the result will be shown here.
Considering the potential flow due to singularities in the presence of an infinite
porous wall boundary condition as described by Equations 2.10 and 2.11, it may be
shown that the complex velocity functions for a source, Wo, and a vortex, W, t, are:
(y
Wo= _ [ (B+E) + X(Pu)X(P L) ] (2.17)
W. t = iy(B - E) (2.18)
11
where:
1 e_ttU
2t----&L
(z-z°)
B = -- 0 < tU + tL < 1 (2.19)'
2 e ' 'J - 1 2
i_tu-tL
E = 1 e_ z "Z-z°) tU + tL
---- e 2 0 < < 1 (2.20)
2 +1 2
(2.21)
z(Pu)Z(Pr ) = 10 Pu = PL =0
Otherwise
Velocity Functions for source and vortex doublets in the presence of infinite porous
walls may be obtained by differentiation of the above formulas:
H d
W_t = 13fh dZ (B+E) (2.23)
c0i d
Woo - (B - E) (2.24)
dZ
Higher derivativesof the source and vortex may also be taken if additional singularities
are desired. In this work, up to the fourth derivativesof the source and vortex were
retained in the singularityexpansions.
It is noteworthy that the velocitiesinduced by a source in this formulation are
discontinuous at P=0 (see Equations 2.17 and 2.21). Classical wind tunnel interference
corrections model the wall interference using infinitestrip theory. This approach
develops corrections based on linear theory solutions for simple singularities in the
presence of two inf'mite walls. When these walls are solid, the mass introduced by the
12
Sourcemust exit either upstream(x = -_) or downstream(x = +_.) Generally, a bias is
introduced into the velocity functions in order to force the velocity disturbanceto go to
zero at x = -_ (seeEquation 2.21), but the influence of the source far downstream
remains non-zero. However, for a porous wall, no matter how small the porosity, the
mass contribution from the sourcemay leak out through the walls. Thus, for an infinite
porous wall, the velocity contribution from a sourcewill go to zero at both x = -** and
x = +_. This discontinuity is well known [28]. The use of infinite strip theory
introduces finite test section errors [29], but it is still common in practice as an
engineering approximation. The effects of the discontinuity at P---Ocaused by the
infinite strip theory will be considered further in Chapter5. The velocity profile on the
wind tunnel inflow plane and the presenceof a breathersectionat the outflow plane also
introduce finite test section effects. In this work, singularity solutions were developed
using infinite strip theory. Finite test section effects associatedwith the inflow and
outflow plane wereminimized by specifying boundaryconditionsbasedon experimental

measurements on these planes. A velocity profile, based on five-hole probe
measurementswas specified on the inflow plane anda constantpressurewas specified at
the breatherlocation.
2.3 Methods of Describing the Effect of a Porous Wall in CFD Codes
Classical porous wall theory has been widely used to apply corrections to the
measured wind tunnel data based on the interference potential _W- The use of such
corrections is undesirable for CFD validations, however, because they alter the entire
flow field in order to impose a correction which is valid at only one point. Recent
efforts in CFD validations have attempted to model wall boundary conditions directly in
the computations. For a solid wall wind tunnel test, this may be accomplished by
specifying either a slip (inviscid) boundary condition at the edge of the boundary layer,
or a no-slip (viscous) boundary condition at the location of the walls. For a porous wall,
13
however, the complexity of the wall geometrydoesnot allow for the wall to be modelled
exactly given the currentlimitations on computermemory and speed.
Some attemptshavebeenmadeto use the boundary conditions of Equations2.10
and 2.11 directly in CFD calculations. This approachis questionablefor three reasons.
First, there is not a good theoreticalapproachfor determining the value of the porosity
parameter contained in these equations and so the porosity parameter must be
determined from experimentaldata[1]. Second,there is experimental evidencethat the
linear relationship betweenperturbationvelocities implied by Equations 2.10 and 2.11
do not apply over the entire wall [19]. Third, and most important, the application of
Equations 2.10 and 2.11 as boundaryconditions in CFD codes is destabilizing and can
actually prevent convergence[23, 24]. Still, classical porous wall theory doesprovide
some insight into the natureof the flow neara porous wall.
Some attemptshavealsobeen madeto model a porous wall boundarycondition
by specifying a normal velocity profile on the walls. The normal velocity profile
developed in theseapproachesis an equivalent inviscid normal velocity profile basedon
an integral boundarylayer analysis. This analysis requires a knowledge of the normal
velocity at the wall surface. Theseapproacheshave been successfulin describing the
effects of a porous wall, but their application has been limited by the extreme
requirements for wind tunnel calibrations and measurements during tests. These
approachesrequire either detailednormal velocity measurementsnearthe wall during a
test or detailed pressuremeasurements

on the wall coupled with extensivecalibrations of
the wind tunnel walls in orderto determinethe relationship betweenpressuredifferences
across the wall and velocities inducedthrough the wall [19, 20, 21]. The calibrations
must be performed at every Mach number and Reynolds number which will be used
during testing andthey requirean adjustablepressureplenum. Either implementation of
this approach requires extensive wind tunnel tests for the purpose of determining
boundary conditions. Furthermore,the integral methodsuseddependupon very detailed
14
measurementsduring the test, limiting the facilities in which they may be used. While
some attempts have been made to apply theseproceduresto simple three-dimensional
cases,their usefulnessis limited by testfacility requirements.
2.4 Theoretical Approach of the Present Work
The approach used for modelling the porous wall boundary condition in this
research was to make use of classical porous wall theory as a means of interpolating and
extrapolating limited experimental measurements of the pressure on the porous walls.
In this work, the classical porous wall boundary condition of Equations 2.10 and
2.11 was used to develop pressure profiles on the walls using the method of singularities.
The singularity solutions were developed as discussed in Section 2.2. Singularity
strengths were determined based on least squares matching of velocity profiles inferred
from pressure measurements made on the walls. The resulting method of singularities
solution was then used to specify a pressure boundary condition in the CFD code. Thus,
flow field corrections were eliminated in preference to a means of specifying
experimental boundary conditions in the CFD simulation. Furthermore, by using
singularity solutions based on the classical porous wall boundary condition,
experimental data could be interpolated and extrapolated over the length of the walls.
Classical two-dimensional porous wall theory typically uses four singularities (a
source, a vortex, a source doublet and a vortex doublet) to develop corrections to the
flow field. The current approach extended classical theory by using as many as ten
singularities. These singularities included a source, a vortex, and ftrst, second, third and
fourth derivatives of the source and vortex in the streamwise direction. Furthermore, in
the current apporach, no corrections were applied to the wind tunnel data. Instead, the
pressure profile existing on the walls during the experiments was modelled in the CFD
code. In addition, classical porous wall boundary conditions were used only to develop
the singularity solutions, and were not applied directly in the CFD code. In this work,
15
the porosity parameter was found by minimizing the root-mean-square errors in
matching pressurecoefficient measurementson the walls.

2.5 Extension to Three Dimensions
Extension of this approach to three dimensions requires a panel method, or other
numerical solver, to determine three-dimensional singularity solutions in the presence of
a porous wall boundary condition. Again, as in the two-dimensional case, simple
singularities and their derivatives may be used. In this work, a line source of span 2y 0,
(2.25)
= rnln[ [_(y- YO)+4 x2 +132(y- yo) 2 + (13z) 2 ]
qbS 4_ t_(y+yo)+4x 2 +_2(y+yo)2 +(_z) 2
and horseshoe vortex of span 2y O,
0v z )
+Atan z(x2
xyy0 x /1
+ 132(y + y0)2 + [_2z2) _ z(x2 + [32(y- y0)2 + [32z2)1_ (2.26)
along with the first, second and third derivatives with respect to x of the line source and
horseshoe vortex, the fin'st, second and third derivatives with respect to Y0 of the line
source and horseshoe vortex, and a first derivative with respect to z of the horseshoe
vortex were used to develop the method of singularities solution. A uniform freestream
term was also included in each case as a means of biasing the source term. In Equations
2.25 and 2.26 above, x is the Cartesian coordinate in the streamwise direction, y is the
Cartesian coordinate in the spanwise direction, and z is the Cartesian coordinate in the
vertical direction. The singularities are located at x = 0, z = 0, and centered about y = 0.
They have a span of 2y 0.
16
The panel methodusedconstantstrengthdoublet panelsto model the wind tunnel
walls [30]. The walls weremodelled over 24 chord lengths. Geometric stretching of the
panels in the streamwisedirection wasusedto allow for better resolution of the solution
near the location of the model. A total of 384 panels were used to model each of the
four walls. Inflow andoutflow planeswere not panelled. This approach,which models
only a finite portion of the walls rather than treating them as infinite, results in
continuous behaviour at P = 0. Figure 1 showsthe arrangementof panels on the walls.
In Figure 1, as in Equations2.25 and 2.26, the singularities are located at x = 0, z = 0
and centered at y = 0. The span of the singularities was equal to the wing span. A
simple shift in the coordinate system may be used to place the singularities at any
streamwise location desired. For the calculation of singularity strengths basedon least
squaresmatches to experimentaldata,the singularities were shifted to the quarter-chord
location on the wing.
For three-dimensionaltesting, the boundary conditions analogousto Equations

2.10 and 2.11 are:
PtJ m
_ __
D0 = 0 at z=--h (2.27)
_x + Oz 2
h
PL _ 0_ = 0 at z=---- (2.28)
_x 0z 2
8_ _ = 0 at y=W (2.29)
Ps_+_ "
In the case of full-span testing, an additional boundary condition of the form of Equation
2.29 may be imposed on the remaining wall. In the case of half-span testing, as in this
work, the remaining side wall is a plane of symmetry. The boundary conditions of •
Equations 2.27'2.29 were enforced at control points located at the center of each panel.
Bicubic spline fits were used to interpolate solutions between control points on the
panels.
.... ..... 17
Chapter 3
Two-Dimensional Governing Equations and Navier-

Stokes Solver
3.1 Introduction
For numerical simulation of the wind tunnel tests performed in the Stanford Low
Speed Wind Tunnel, an incompressible Navier-Stokes solver (INS2D) was used [31].
Although methods of describing porous walls in CFD codes are of primary interest for
transonic codes, an incompressible code was used in this case because of the very low
Math number of the tests conducted in this research tunnel. The method described in
this and the previous chapter may also be used with transonic flow solvers [32].
3.2 Governing Equations
The equations presented here have been nondimensionalized as follows:
Ui = ui Xi = xi i = tUref p = p -- Pref
Uref Xref Xref pu2f
{ij= p'Ci2J
fure _= XrefUrefV = Re-1_: (3.1)
where the - superscript denotes nondimensional quantities. Hereafter, the ~ superscript
will be dropped for convenience.
In the above equations, ui = u or v, the Cartesian velocity components for i = 1,
2, respectively and xi = x or y, the Cartesian spatial coordinates for i = 1, 2,
respectively. The pressure, density and kinematic viscosity are denoted by p, p, and v,
19 PI_IBCIOING PA_ _LANK NOT F_ll/t_13

respectively. The subscript ref denotes reference quantities; Uref is the measured
freestream velocity upstream of the model, Xref is taken as the airfoil chord.
The Navier-Stokes equations for incompressible, two-dimensional, constant
density flow may then be written in conservative form as follows:
_u _v
(3.2
_---i-+ (e-e,)+ (f-f,,)=0

(3.3)
where:
:iul (3.4)
eE::+P
1 [vo] f =
v2 +p
(3.5)
I]
e v = ,Cxy
fv -
"Cyy
(3.6)
For turbulent flows, these equations represent Reynolds averaged quantities. The
Boussinesq approximation is used for the Reynolds stress. The viscous stress tensor
may then be written as:
xij = (v+vt) + (3.7)
where v t is the turbulent eddy viscosity.
As the thrust of this work is the development of a methodology for describing
porous wall boundary conditions in CFD codes, the details of the CFD code and
turbulence modelling will not be discussed here. A brief description of the CFD code is
included in the following sections. Additional details about the CFD code may be seen
in References 31 and 33. The results shown from the INS2D code used a Baldwin-Barth
turbulence model. The Baldwin-Barth turbulence model is in wide use in a variety of
20
CFD codes currently. Additional details about this turbulence model may be seen in
Reference 34.
3.3 Numerical Solver
The INS2D code uses a primitive variable formulation of the incompressible
Navier-Stokes equations. In the primitive variable formulation, the equations are written
with pressure and velocity as the dependent variables.
The equations are transformed into generalized coordinates using the
transformations:
_=_(x,y, 0
rl = Tl(x, y, 0 (3.8)
The system of equations then becomes:
_9 U _9 v
_(-j-) + _(-_-) = 0 (3.9)
(3.10)
_t _(e- ev) - f-e v
where J is the Jacobian of the transformation and:
0_x[u]. (3.11)
lr_:,,
= 7[_yp + vU + _tvJ
+'.'<-'
+ _= lr,-,:,
7Lrlyp
+uv
+
+ .,.i,,.,
vV
1
+ rltv.] (3.12)
The contravariant velocity components, U and V are given by:
(3.13)
U = _xU + _yV V = rlxU + rlyV
The metrics of the transformation are represented with subscript notation as, for
example:
_x - _
_x
21
The INS2D code uses an artificial compressibility approach for solving the
incompressible Navier-Stokes Equations[31, 35]. In this approach,a time derivative of
pressure is added to the continuity equation (Equation 3.9) and an artificial

compressibility factor, _, is introduced:
_P =- _(V-6) (3.14)
3x
This results in a hyperbolic systemof equations. This systemmay then be marchedto a
steady-state solution in pseudo-time. The right hand side of Equation 3.10 and the
divergence of the velocity field approach zero as the solution approaches steady-state.
This hyperbolic system of equations also has artificial pressure waves which are
finite in speed. Therefore, compressible flow algorithms may be used for solving the
system of equations. The INS2D code uses third order accurate, locally upwind
differencing of the convective fluxes and second order accurate central differencing of
the viscous terms. The upwind differencing is biased by the signs of the eigenvalues of
the local flux Jacobian, following the method of Roe [36]. The set of numerical
equations is solved using a Gauss-Seidel type line relaxation scheme.
All computations were performed on the CRAY Y-MP computer at NASA Ames
Research Center. Solutions typically converged in approximately 1500 iterations,
requiring approximately 30 minutes of CPU time on the CRAY Y-MP computer.
Convergence histories for cases using slip wall boundary conditions were nearly
identical to those for cases using a specified pressure boundary condition.
Additional details concerning this algorithm and its numerical implementation
may be found in Reference 31.
3.4 Boundary Conditions
The method of artificial compressibility used in the INS2D code allows for
boundary conditions to be specified using the method of characteristics, similar to
compressible flow solvers. In the case of the artificial compressibility scheme, however,
22
the method of characteristics is merely a mathematical construct which allows for
wavesto exit at the boundariesof the computational domain without reflection.
There are three characteristic lines associatedwith three eigenvalues of the
systemof equations: u, u+c, u-c, where c is the artificial speedof sound in this method.
At an inflow boundary, there is one characteristic line carrying information from the
interior of the solution to the boundary. At an outflow boundary there are two
characteristic lines carrying information from the interior of the solution to the
boundary.
The approach used in this work was to treat both the outflow plane and top and
bottom walls as outflow boundaries. Although not strictly correct for a wall boundary
which allows both inflow and outflow, this approach is consistent with that used by
other researchers [17-21]. The problems associated with making velocity measurements
near a wind tunnel wall usually result in only one flow variable being specified as a wall
boundary condition in wind tunnel simulations. Additionally, an abrupt change from
inflow to outflow boundary conditions on the wall usually results in discontinuities in
the solution.
The outflow boundary condition, as implemented in the INS2D code, can most
easily be understood by consideration of a limiting case in which the method is applied
to a uniform, orthogonal grid. In this case, for an outflow boundary, the method reduces
to:
PlI
Vn = (3.14)
(3.15)
Un = Pn 1- (v+ --
.,iv2 + _ V2 -+ _
23
where the n-subscript denotes differentiation in the direction normal to the boundary.
For other cases, the metrics of the transformation and local rotations also affect the
equations.
With p specified at the boundaries, and for cases where the artificial
compressibility parameter _ >> u, v, as was the case in this work, this method results in
an extrapolation of interior velocities to the boundary. It does, however, still allow for
small variations in pressure and velocity normal to the surface to avoid wave reflections.
It therefore allows for faster convergence than a simple extrapolation scheme.
3.5 Two-Dimensional Grid Generation
Grids were developed which conformed to boundaries on which experimental
data was available. The inflow plane of the grid corresponded to the inflow plane on
which five-hole probe data was taken. Top and bottom boundaries of the grid
corresponded to the nominal (0-degree inclination) top and bottom wall locations. The
outflow plane corresponded to the breather section location. These grids were used for
all slip wall computations and for all cases in which pressure was specified as a
boundary condition. A special grid was developed for the viscous solid wall case. For
that case, the top and bottom boundaries of the grid corresponded to the inclined wall
locations.
All grids used with the INS2D code were developed using GRIDGEN software
[37]. Surface definitions were based on spline fits to approximately 2000 measurements
of the model geometry. The very high resolution of measured model geometry allowed
for excellent geometry definition in the grids as well as resolution of model defects
associated with the attachment of the model flap and spoiler. Orthogonality was
maintained at the airfoil surface by using exponential blending of grid lines near the
surface. The Thompson Middlecoff elliptical solver was used within the GRIDGEN
software to develop the grids. This solver maintains boundary distributions of grid
points in the interior of the grid and allows for easy refinement of the grid in boundary
24
layers over the airfoil. Spacing was controlled so that approximately 15 to 20 grid points
were located within the boundary layer on the airfoil. All of the grids developed were
C-grids. Normal grid dimensions were 250 x 70, although a test case was run using a
350 x 105 grid to examine the effect of grid refinement. An example of these grids may
be seen in Figure 2.
Figure 2. Grid used for two-dimensional CFD simulations for ot = 5 degrees.
25
Chapter 4
Experimental Apparatus
4.1 Wind Tunnel Test Section
A specially constructed test section was built for use in the Stanford Low Speed
Wind Tunnel. This test section was heavily instrumented with static pressure tappings
on the walls and with five hole probe access on the inflow and outflow planes and on
planes near the top and bottom walls. The test section was designed to allow for two-
dimensional testing and three-dimensional half-span testing. For two-dimensional
testing the top and bottom walls were porous and both side walls were solid. For three-
dimensional half span testing, the model was mounted to a solid wall and the remaining
three walls were porous.
The test section was .457 m x .457 m and the porous portion of the test section
was .76 m long. The porous walls were constructed using a layered design to allow the
open area ratio to be varied. The layer of the wall nearest the flow was a slotted layer
with .00254 m streamwise slots located at .0254 m intervals. The middle layer of the
wall consisted of inserts which had .0229 m by .00254 m rectangular holes located at
.0254 m intervals. These holes were inclined to the normal at 15 degrees, giving the
porous wall a differential resistance to inflow and outflow. The outer layer of the wall
was also slotted similar to the inner layer and could be slid laterally to vary the open area
ratio of the wall. Open area ratios of the walls could be varied from a solid wall
condition to an open area ratio of 9-percent. An atmospheric plenum was used for all
porous wall tests. An exploded view of this wall design may be seen in Figure 3.
26
An atmospheric breather section was also used in all tests and was located at the
end of the test section. The nominal freestream velocity for all tests was 24. m/see and
the nominal Mach number for all tests was M = .07. The nominal Reynolds number
based on chord was Re = 3.2 x 105 for the airfoil tests and 2.4 x 105 for the three-
dimensional wing tests. The test section top and bottom walls were adjustable and were
inclined normal to the freestream to compensate for boundary displacement layer growth
through the test section. Displacement layer thickness was measured by a boundary
layer rake located near the end of the porous portion of the test section. Displacement
layer thicknesses were measured on both the top and bottom walls with the model in
place. The rake was removed during normal tunnel operation. A schematic of the wind
tunnel may be seen in Figure 4, and a photograph of the test section may be seen in
Figure 5.
4.2 Wind Tunnel Models
The two-dimensional model used in these tests was a Boeing Advanced
Transport Airfoil. A cross section of this airfoil may be seen in Figure 6. The model
had a flap and a spoiler which were in the retracted position for these tests. The airfoil
chord was .203 m and the maximum thickness ratio was 11.3 percent. Measurements of
the model were made with a Leitz Precision Measuring Machine with an accuracy of
2.54 x 10 -6 m, or approximately .00125 percent of the airfoil chord. Approximately
2000 measurements were made to define the airfoil cross section. Grit was applied to
the airfoil between 2.5 and 5.0 percent of the chord [38, 39]. Nominal grit diameter was
.03175 cm. Oil flow studies were undertaken to ensure the two-dimensionality of the
flow field and to ensure that the grit applied to the model successfully triggered
transition on the airfoil.
The three-dimensional model used in these tests was an unswept half-span wing
with an aspect ratio of 1.5, a span of .229 m and a chord of .152 m. The wing had a
constant NACA 0018 airfoil cross section. The wing tip was a half body of revolution
27
with diameter equal to the local thicknessof the wing. Grit was applied to the wing
between 2.5 and 5.0 percent of the chord. Nominal grit diameter was .03175 cm. Oil
flow studies were undertaken to ensurethat the grit successfully triggered transition on
the wing.
4.3 Test Section Instrumentation
The test section was instrumented with static pressure tappings on the porous
walls and five-hole probe accessnear the top and bottom walls and on the inflow and
outflow planes. The tunnel was instrumented with 78 static pressure tappings on the
centerlines of the upper and lower walls and a total of 482 static pressure tappings on
the upper and lower walls. An additional 121 pressuretappings were located on the
porous side wall. Pressuretappings were also located on both models. The five-hole
probe was usedto determine flow inclinations and velocity perturbations on the inflow
and outflow planesas well as planesnearthe top andbottom walls.
Static pressuredata was acquiredby three Scanivalveswhich were calibrated on
each run. Together, these Scanivalves were capable of making 576 pressure
mesurements on each run. Nominal freestream velocity was determined by static
pressure measurementstaken in the inlet contraction bell. These measurementswere
sufficiently far upstream so that interference from the model was negligible. The
velocity measurementswere calibratedin empty tunnel runs.
4.4 Wind Tunnel Flow Quality
Empty tunnel flow quality surveys were conducted and may be seen for solid
wall and porous wall cases in Figures 7 and 8, respectively. A velocity variation of
approximately 4 percent may be seen in these surveys at the inflow plane location. This
effect appears to be local to the inflow plane of the test section, and is much less
pronounced near the model location. Instrumentation access holes restricted the range of
positions at which measurements could be made. In these cases, as in the data which is
28
shown later, the nominal free stream velocity is basedon static pressure measurements
made in the contraction section of the wind tunnel.
4.5 Error Analysis
All pressure measurements were obtained with instrumentation calibrated against
a secondary standard traceable to the National Bureau of Standards. Pressure
measurements on the walls and airfoil were made with three Scanivalves, each having
four 48-port barrels. Two ports on each barrel were used for a reference and calibration
pressure. This allowed the pressure transducer in each barrel to be calibrated on each
run. A settling time of approximately 1 second was allowed after each advancement of
the Scanivalve barrel. A low pass (10 Hz) filter was used to filter data from the pressure
transducers. Pressure readings were based on an average of 200 measurements taken at
1 kHz. Standard deviations were computed for these measurements and were typically
0.01-percent or less of the mean data values. A calibration pressure of 0.100 psi (.689 x
103 N/m 2) was used for wall pressure measurements and a calibration pressure of 0.200
psi (1.378 x 103 N/m 2) was used for airfoil pressure measurements. Uncertainties in the
calibration pressure measurements were + 0.6 x 10 -3 psi (4.13 N/m2), based on
manufacturer supplied data. Additionally, discretization by the data acquisition board
resulted in a minimum resolution error of .1 x 10 -3 psi (.69 N/m 2) for the airfoil
measurements and .05 x 10 -3 psi (.34 N/m 2) for the wall measurements. Non-linearity
of the pressure transducers was estimated at less than .2-percent based on a comparison
of known pressure differences and transducer outputs. These errors combined to give an
overall error of approximately .9-percent in Cp measurements on the airfoil and 1.2-
percent in Cp measurements on the walls.
Five-hole probe pressure data was obtained by dedicated 1 psi Druck pressure
transducers. These transducers were calibrated on each run using the secondary standard
discussed above. Calculation of flow angularity without direct measurements of velocity
perturbations led to results of unknown accuracy. The relationship between pressure
29
coefficient measurementswith a five-hole probe and flow anglesis complex. Adding to
uncertainties in pressure mesaurements were uncertainties regarding probe alignment. A
qualitative assessment of the probe data could be made by comparing data taken from
J
identical runs, or by observing the variations in repeated probe measurements made at
the same point in the flow field. This qualitative assessment suggested that typical probe
error was on the order of 0.5 degree.
.0229 m
.00254 m
.00318 m
Top Layer is
Moveable
.0229m
.U2
.0127 / " _ 254m
I m / _-.uz 29 m//, ./00
15 deg.
.00318 m
FLOW
/
Figure 3. Exploded view of variable porosity wall.
30
Variable Pitch Fan
Turning Vanes
Variable Speed Motor
I Breather Vent ,,_
Test Section
Turbulence Damping Screens
Figure 4. Schematic diagram of Stanford Low Speed Wind Tunnel.
Figure 5. Variable porosity test section. Airfoil model is shown installed.
31
0.05
O,
-0.05-
-0.1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 I
X/C
Figure 6. Boeing Advanced Transport airfoil Cross section.
1.10
----O---
---e-- InflowLeading
Wing Plane ofEdge
Test Location
Section
1.08
1.06
1.04
1.02
8
=)
1.00
::3
.98
.96
.94
.92
I I I I I I I I I
.90
-.5 -.4 -.3 -.2 -.1 0.0 .1 .2 .3 .4 .5
z/h
Figure 7. Empty tunnel, solid wall flow quality surveys at test section entrance and
model location.
32
1.10
1.08 _ InflowLeading
Wing Plane ofEdge
Test Looation
Section
1.06
1.04
1.02
8
1.00
.98
.96
.94
.92
I I I I I I I I I
.90
-.5 -.4 -.3 -.2 -.1 0.0 .1 .2 .3 .4 .5
z/h
Figure 8. Empty tunnel, porous wall flow quality surveys at test section entrance and
model location.
33
Chapter 5
Results
5.1 Two-Dimensional Case
Both computational and experimental studies were undertaken to evaluate the
method of singularities for application to two-dimensional flows. Preliminary
computational studies used a CFD solution to determine the number of measurements
and the number of singularities required to develop an adequate description of the flow
near the walls using the method of singularities. The experimental data was used to
determine the suitability of this method to actual flow fields. Effects considered were:
1) the number of singularities and flow field measurements required to describe the flow
near the walls, 2) the ability of this method to capture the effect of changes in the open
area ratio of the walls, 3) the effects of uncertainties in the porosity parameter specified
in the method of singularities, 4) the effects of singularity placement, 5) comparisons of
alternate methods of describing a solid wall boundary, 6) the effects of grid refinement
in the CFD code, 7) the effects of using sparse data to develop a wall pressure prof'fle,
and 8) flow inclinations near the boundary, as computed by the CFD code and measured
by the five-hole probe.
5.1.1 Preliminary Computational Studies
As a means of evaluating the method of singularities, a CFD simulation using
slip wall boundary conditions was developed. This simulation was then treated as a
psuedo-experiment and data from the wall boundaries was used to develop a method of
singularities solution. The use of a computational result instead of experimental data for
34
this purpose allowed for a detailed comparisonof pressureprofiles on the walls and a
comparison of the effects of using sparsedata to develop the method of singularities
solution. Figure 9 shows the pressureprofiles on the walls as predicted by the CFD,
simulation using slip wall boundary conditionsfor an angleof attack of 5 degrees.
Figure 10 showsthe root-mean-squareerrors resulting when various numbers of
singularities and datapoints were usedto developthe potential flow solutions. The data
points were distributed as uniformly as grid spacing would allow over the top and
bottom walls. Up to four derivatives of a sourceandvortex were retained in the method
of singularities solutions. The root-mean-squareerrors shown in Figure 10 were based
on a point-by-point comparison of the pressurepredictedby the CFD simulation and the
pressure predicted by the method of singularities at each grid point on the upper and
lower walls. The root-mean-squareerrors havebeen normalized by the peak pressure

coefficient on the walls. As may be seenin Figure 10, a relatively small number of
singularities was adequateto describe the far field flow. In addition, a small number of
measurementlocations was sufficient to developa potential flow solution. Additional
data from the wails allowed for a better description of the pressure prof'de in the method
of singularities, but beyond 25 data points the improvement was very small.
The use of additional singularities resulted in the need for additional data points
in order to produce acceptable fits to the velocity prof'des. It was found that the number
of data points required to produce a good fit to the velocity profiles was approximately
twice the number of singularities retained in the potential flow solution.
While these trends were observed for a variety of airfoil simulations [32, 40, 41],
caution should be exercised when applying this method to different geometries. Some
computational experimentation, of the type just described, may be required in order to
develop confidence in this method for different geometries such as multi-element
airfoils, blunt bodies, or configurations with very low height-to-chord ratios.
35
5.1.2 CFD Simulationsof Experiments
The next stepin evaluating this method was to use it to perform simulations of a
series of wind tunnel tests. The tests were conducted in the variable porosity test section
described in Chapter 4. The pressure profiles measured on the walls during the 5 degree
angle of attack tests may be seen in Figure 1 la. This figure shows data from both the
solid wall tests and the porous wall tests with a 9-percent open area ratio. Also shown in
Figure 1 l a are the potential flow solutions based on this data. These potential flow
solutions were developed using the 77 measurements on the wind tunnel walls and a
total of 10 singularities (a source, a vortex, and the first four derivatives in the
streamwise direction of a source and vortex.) Figures 12a-17a show similar results for
angles of attack of 6, 7, 8, 9, 10, and 11 degrees. As can be seen in these figures, the
potential flow solutions showed excellent agreement with the experimental data.
The potential flow solutions shown in Figures l la-17a were imposed as
boundary conditions in the Navier-Stokes solver and the resulting pressure profiles
obtained on the airfoil are shown in Figures l lb-17b. Also shown in Figures 1 lb-17b
are the airfoil pressure profiles obtained in the wind tunnel tests. Experimental and
computational results compared very well. The largest discrepancies were
approximately 6-percent of the peak values. These discrepancies occurred at the
location of the grit transition strip on the model. At locations away from the grit strip,
however, agreement was much better with typical discrepancies of approximately 1-
percent of peak values. A comparison of airfoil pressure profiles from the solid wall and
porous wall tests showed that this method simulated the effect of a porous wall very
well. As the porosity of the wall was varied, the trend in pressure profiles on the airfoil
was duplicated in the CFD simulations by imposing the potential flow solution as a
pressure boundary condition in the CFD solver.
Also of interest in Figures l lb-17b are the slight deviations seen in the pressure
profiles on the airfoil between x/c = 0.75 and x/c = 0.80. These deviations may be seen
36
on both the upper andlower surfacesand arepresent in both the experimental andCFD
data. They are the result of small ridges on the airfoil model between the main airfoil
and the retracted flap and spoiler.
The root-mean-square errors in matching experimental pressure coefficients on
the wails using the method of singularities are summarized in Table 1. Also shown in
Table 1 are the root-mean-square errors between measured pressures on the airfoil and
pressures predicted by the CFD code with the method of singularities boundary
condition imposed. In both cases, these root-mean-square errors are based on a point-
by-point comparison at all locations where experimental data was available. The root-
mean-square errors shown for wall and airfoil cases have been normalized by peak
pressure coeffiecients measured on the wall and airfoil, respectively, in each test.
Wall Measurements Airfoil Measurements
Angle of Solid Wall Porous Wall Solid Wail Porous Wall

Attack (deg)
5 0.00777 0.01485 .02669 .03023
6 0.00732 0.01435 .02368 .02703
0.00743 0.01420 .02205 .02023
0.00722 0.01357 .01995 .01753
9 0.00705 0.01300 .02045 .01525
10 0.00739 0.01350 .02331 .01819
11 0.00756 0.01327 .02635 .02234
Table 1. Root-mean-square errors in matching pressure coefficients on the walls using

the method of singularities, and in matching experimental pressure coefficients on the
airfoil with the INS2D CFD code.
5.1.3 Effect of Changes in Porosity Parameter
In the results shown above, the porosity parameter used in the method of
singularities was determined by minimizing the root-mean-square errors in matching the
wall pressure profile. As a means of evaluating the sensitivity of this method to
uncertainties in the porosity parameter specified in the method of singularities, a variety
of porosity parameters were used to develop potential flow solutions in the method of
37
singularities. Figure 18 shows root-mean-squareerrors in matching pressure coefficients
on the walls as a function of the porosity parameter specified in the method of
singularities for the 9-percent porous wall, 8-degree angle of attack case. The root-
mean-square errors shown in this figure have been normalized by the peak pressure
coefficient measured on the walls. These results are typical of the results at other angles
of attack. It is noticeable that the curves shown are discontinuous at P = 0. The reasons
for this discontinuity have already been discussed in Chapter 2. In all the cases
examined, the optimum porosity parameter was found to be P = 0. For non-zero
porosities, the optimum porosity depends on the number of singularities retained and the
optimum tends to be very shallow. This again points out the danger of using Equations
2.10 and 2.11 directly as a boundary condition in CFD codes.
Figures 19 and 20 show a comparison of results obtained by specifying a
porosity parameter of 0. and 1. in the method of singularities solution. These results are
for the porous wall tests at an angle of attack of 8 degrees, and using 10 singularities in
the method of singularities. Figure 19 shows the resulting pressure coefficient prof'tles
on the walls, and Figure 20 shows the resulting pressure coefficient profiles predicted by
the CFD code on the airfoil. As can be seen in Figure 19, the finite test section effects
associated with the change in porosity are most noticeable near the inflow and outflow
planes of the test section. The trend in pressure profiles was captured with either
specified porosity. As can be seen in Figure 20, the discrepancies in the pressure
profiles of Figure 19 have little effect on the airfoil pressure profiles. Overall, this
method is relatively insensitive to uncertainties in the porosity parameter. By using a
least-squares approach to determine singularity strengths, this method allows singularity
strengths to vary as needed in order to provide an optimum fit to the experimental data
for any specified value of porosity. This effect may be seen in Figure 21, which shows
the singularity strengths in the method of singularities as a function of the porosity
38
parameter specified on the walls. For clarity, only 6 singularity strengths are shown in
this figure.
5.1.4 Effect of Changes in Singularity Placement
In the results discussed above, all singularities were placed at the quarter-chord,
tunnel centerline location (x/c = 0.25, y = 0.0). Figure 22 shows the effect of varying the
location of the singularities. Singularity locations were varied from x/c = 0.0 to x/c =
1.0. The resulting root-mean-square errors in matching the pressure profile on the walls
for the solid wall, 8-degree angle of attack case are shown. Again, the root-mean-square
errors have been normalized by the peak pressure coefficient on the wall. As can be
seen, this method is relatively insensitive to the position of the singularities, provided
that the singularities are placed ahead of the mid-chord point. An increase in the root-
mean-square errors occurs as the singularities are placed further downstream, although
even these errors are relatively small. As noted earlier, some caution should be
exercised when applying these results. When using this method with geometries
significantly different from those used here, such as multi-element models or extremely
low height-to-chord ratios, it may be necessary to distribute singularities in order to
adequately model the pressure profile on the walls.
5.1.5 Comparison of Alternate Boundary Conditions for Solid Wall Simulations
As a means of comparing this approach to other methods of specifying boundary
conditions in CFD codes, CFD simulations were performed of the solid wall 8-degree
angle of attack test using 3 alternate boundary conditions. These alternate boundary
conditions were: 1) A slip-wall boundary condition (v = 0) applied at the nominal wall
location, 2) A no-slip boundary condition (u = v = 0) applied at the physical wall
location, and, 3) A free-air boundary condition (u = u_,, v = vow, p = p_) applied at a
distant boundary with corrections applied to the airfoil angle of attack and the freestream
velocity.
39
The grid used for the slip wall simulation was the same grid used to specify the
pressure boundary condition. The outer boundary of this grid corresponded to the
nominal wall location. As discussed earlier, the walls were inclined to adjust for the
growth of the displacement layer through the test section.
Figure 23 shows the grid used for the no-slip wall simulation. This grid models
the wall inclinations through the test section and uses additional refinement near the
walls to resolve the wall boundary layer.
Figure 24 shows the grid used in the free-air simulations. For the free-air
boundary condition, classical incidence and velocity corrections were applied to the
wind tunnel data [42]. The CFD simulation was performed with the airfoil at the
corrected angle of attack and the velocity corrections were applied in non-
dimensionalizing the data. The outer boundary was located 30 chord lengths from the
airfoil.
Figure 25 shows the pressure profiles on the airfoil resulting from these
simulations, as well as the simulations using the pressure boundary condition from the
method of singularities. Experimental data is also shown for reference. The specified
pressure boundary condition, based on the method of singularities, produced the closest
match to the experimental data. The slip wall and no-slip wall showed similar results
which were also in close agreement with the experimental data. The free-air simulation
showed significant discrepancies (approximately 18-percent of peak values) with
experimental data at the pressure peak, although overall agreement was much closer.
Overall agreement between these methods was close, although differences may clearly
be seen in the area near the pressure peak on the airfoil. The overall agreement of these
approaches demonstrated the consistency of this method with other means of specifying
boundary conditions for solid wall tests. This agreement gave added confidence in this
method when it was used for simulating porous wall boundary conditions.
40
5.1.6 Grid Refinement Effects
The effects of grid refinement were also consideredfor the solid wall data from
the 8-degreeangle of attack case. The grids used in the aboveresults had already been
refined to the point where additional grid refinement appearedto havelittle effect on the
solution. For comparison purposes,however, an additional result was computed on a
further refined grid using approximately twice the numberof grid points as were usedin
the standard grids typical of Figure 2. Figure 26 shows a 350 x 105 grid which was
used to simulate the 8-degree angle of attack, solid wall case. Figure 27 shows a
comparison of the results obtainedfor this caseon the 350 x 105grid and the 250 x 70
grid which was used in the other simulations. As can be seenin this figure, differences
between these results are minor. Some additional ref'mementwas evidenced near the
flap and spoiler ridges on the aft portion of the airfoil, but overall results indicated that
the 250 x 70 grid was sufficiently ref'medso that remaining errors associatedwith grid
ref'mementwere minor.
5.1.7 Effects of Using SparseData
The effect of using sparsedata to develop potential flow solutions was also
investigated. Figure 28a shows 8 data points selectedfrom the complete set of wind
tunnel data for the solid wall test at 5 degreesangle of attack. A potential flow solution
basedon these 8 data points is also shown. In addition, the complete set of wind tunnel
data and the potential flow solution based on the complete set of data is also shown for
reference. For the sparse data (8 data point) solutions, only first derivatives of the
source and vortex were retained in the method of singularities, resulting in a total of 4
singularities being used to develop the potential flow solution. Figures 29a-34a show
similar results for solid wall tests at 6-11 degrees angle of attack and Figures 35a-41a
show similar results using data from the 9-percent open area ratio porous wall tests. The
potential flow solutions based on 8 data points show close agreement with the potential
flow solutions based on 77 data points. Root-mean-square errors between experimental
41
measurementsand method of singularities solutions based on 8 data points are shown in
Table 2. These root-mean-square errors axe based on a point-by-point comparison of the
entire set of 77 experimental measurements with the method of singularities solutign
based on 8 measurements. The root-mean-square errors are normalized by the peak
pressure measured on the walls for each case. As expected, these errors were higher
than the cases in which 77 data points were used to develop the solution (see Table 1),
however the trend of the pressure profdes is still captured using very sparse data. This
agreement indicates that this method was robust enough to allow for a good description
of the boundary conditions on the walls even when only very sparse data was available
from the wind tunnel walls.
Wall Measurements Airfoil Measurements
Angle of Solid Wall Porous Wall Solid Wall Porous Wall

Attack (deg)
5 0.01847 0.03960 .02576 .02978
6 0.01720 0.04033 .02294 .03163
7 0.01786 0.03706 .02077 .02341
8 0.01628 0.03369 .01994 .01986
9 0.01689 0.03420 .01894 .01397
10 0.01678 0.03330 .02136 .01478
11 0.01703 0.03303 .02485 .01773
Table 2. Root-mean-square errors in matching pressure coefficients on the walls using

the method of singularities based on 8 experimental measurements, and in matching
experimental pressure coefficients on the airfoil with the INS2D CFD code.
The pressure profiles shown in Figures 28a-41a were imposed as boundary
conditions in the CFD code and the resulting pressure profiles on the airfoil are shown in
Figures 28b-41b. For reference, the airfoil pressure profiles which resulted by imposing
the boundary conditions of Figures 1 la-17a in the CFD code are also shown in Figures
28b-41b. This allows a comparison of the effect on the airfoil of using sparse data to
develop the wall boundary conditions. As can be seen in these figures, the method of
singularities provided a means of describing boundary conditions which was sufficiently
42
robust so that little effect was seen in the airfoil pressure coefficient profiles from
significantly reducing the number of measurements used to develop the boundary
conditions. Root-mean-square errors between experimental data on the airfoil and the
CFD results using boundary conditions based on 8 data points are also summarized in
Table 2. Again, these errors have been normalized by the peak pressure measured on the
airfoil in each case. These errors are comparable to those for the cases in which 77 data
points were used to develop the boundary condition (see Table 1), and indicate that this
method may produce close agreement between computed and experimental
measurements on the airfoil even when only very sparse data is available to develop
boundary conditions.
5.1.8 Flow Inclinations Near the Boundaries
Five-hole probe measurements were made on a plane near the boundaries, as
described in Chapter 4. These measurments were used to develop a qualitative
assessment of the flow inclinations near the walls.
Figures 42 and 43 show the normal velocities measured near the walls during
porous wall tests of the airfoil at angles of attack of 5 and 8 degrees, respectively. In
these figures, the normal velocity is plotted against the streamwise location. Velocity
profiles are shown for both the top and bottom walls, and all velocity measurements are
normalized by freestream velocity. Also shown in these figures are the computed
normal velocities from the CFD code. These normal velocities are taken from a plane in
the CFD result corresponding to the five-hole probe measurement plane. These CFD
results were obtained by specifying the pressure boundary conditions of Figures 11 and
14 in the CFD code. The difficulties of making experimental flow inclination
measurements with the five-hole probe may be seen in Figure 43 which shows
discrepencies between data obtained from two consecutive runs. These uncertainties
make any quantitative comparison of computed and measured results questionable. A
qualitative comparison of computed and measured results shows that the CFD code with
43
a pressureboundary condition is capturing the trend of the normal velocities near the
walls over most of the boundary.
5.2 Three-Dimensional Case
Limited consideration was also given to three-dimensional applications of this
method. No CFD simulations were performed for the three-dimensional data. However,
the method of singularities was used to develop potential flow solutions based on the
data measured on the walls. The effects of using different numbers and arrangements of
measurements were considered, as well as the effects of using different numbers of
singularities.
5.2.1 Method of Singularities Comparison to Experimental Data
Figures 44a, 44b, 45a, 45b, and 46 show the pressure profiles on the upper, lower
and side walls from the three-dimensional solid wall tests of the half-span wing at an
angle of attack of 20 degrees. Also shown in these figures are the pressure profiles
developed by the method of singularities using 342 data points and 16 singularities to
match the data. Figures 47°49 show similar results for the 9-percent open area ratio
porous wall tests.
As can be seen in these figures, the agreement of the method of singularities fit to
the experimental data is better in two dimensions than in three dimensions. This is in
part due to the finite test section effects near the inflow plane and outflow plane. The
abrupt change in boundary conditions associated with the breather section and inlet of
the wind tunnel test section caused flow anomalies which the method of singularities
could not adequately model. Only data from the central region of the test section is
shown in Figures 44-49. The root-mean-square errors based on a comparison of
experimental pressure profiles and those predicted from the method of singularities are
5.1-percent of the peak pressure coefficient for the solid wall case and 7.6-percent of the
peak pressure coefficient for the porous wall case. Over the central region of the test
44
section, the method of singularities capturedthe generaltrend in pressureprofiles on all
three walls.
5.2.2 Effects of SparseData andReducedNumbersof Singularities
The effects of using reducednumbersof singularities and sparsedatain the three

dimensional method of singularities were also investigated. Figures 50a-c show the
position of the 342 pressure tappings on the walls which were used to develop the
method of singularities solutions shown in Figures 44-49. Figures 51a-c show 223
measurementsarrangedalong the centerlinesof eachwall and three vertical or spanwise
lines on each wall. These data points were also used to develop a method of
singularities solution. The resulting solution was then evaluated based on a point-by-
point comparison to the complete setof 342 experimental measurements. Figures52a-c
shows 136 measurements arranged along the centerlines of each wall and a single
vertical or spanwise line on each wall. Figures 53a-c shows a total of 38 sparsedata
locations taken from the data in Figure 51a-c. The measurementsindicated in Figures
51, 52 and 53 were also usedto developmethod of singularities solutions which were
evaluated based on point-by-point comparisons to the complete set of 342

measurements.
Figures 54 and 55 show the root-mean-squareerrors resulting from a point-by-
point comparison of the entire datasetandthe methodof singularities solutionsbasedon
the data at the positions shown in Figures50-53. Resultsare shown in thesetwo figures
for the solid and porous wall, respectively. Figures 54 and 55 also show the effects of
using various numbers of singularities to developthesesolutions. The resultsshown are
basedon 3, 7, 11, and 16singularities being retainedin the solution. Table 3 showsthe
order in which terms were included in the singularity solutions. All root-mean-square
errors shown in Figures 54 and 55 have been normalized by the peak pressure
coefficients measuredon the walls.
45
Number of Singularities Used
Singularities
3 Singularity Case Uniform freestream, Line Source, Horseshoe Vortex
Uniform freestream, Line Source, Horseshoe Vortex

7 Singularity Case First Derivatives of Source and Vortex in x-direction
First Derivatives of Source and Vortex with respect to Y0

First Derivatives of Source and Vortex in x-direction

11 Singularity Case First Derivatives of Source and Vortex with respect to Y0
Second Derivatives of Source and Vortex in x-direction
Second Derivatives of Source and Vortex with respect to y0

First Derivatives of Source and Vortex in x-direction
First Derivatives of Source and Vortex with respect to YO

16 Singularity Case Second Derivatives of Source and Vortex in x-direction
Second Derivatives of Source and Vortex with respect to Y0

Third Derivatives of Source and Vortex in x-direction
Third Derivatives of Source and Vortex with respect to Y0

First Derivative of Vortex in z-direction
Table 3. Singularities included in three dimensional method of singularities solutions.
The span of the line source, horseshoe vortex and their derivatives was equal to
the span of the wing. Although not actually a singularity, the freestream term was
retained in all cases to allow for the upstream contribution of the source term to be
cancelled. As can be seen in Figures 54 and 55, the effect of adding additional
measurements was to produce a better fit to the overall pressure profile on the walls.
Additionally, as seen in the two-dimensional results, the addition of extra singularities in
the potential flow solution required additional data points in order to provide an
improved fit to the data. Overall lower root-mean-square errors were obtained for the
cases of sparse data by using only 7 singularities instead of 11 or 16 singularities. As
more data was used, however, the root-mean-square errors were reduced by using
additional singularities to describe the flow field. These results suggest that
46
approximately 10 datapoints should be usedfor eachsingularity retained in the method
of singularities. As with the two dimensional results, however, caution should be
exercised when applying these trends to different geometries. Some experimentation
with numbers of singularities and data points may be required in order to develop
confidence in this method for geometries and test conditions which are significantly
different from thosein this research.
Examplesof the wall pressureprof'tles obtainedby using reduceddata to develop

the method of singularities solution may be seenin Figures56-61. Figures 56-58 show
the pressureprofiles on the top, bottom and side walls from the solid wall 20 degree
angle of attack test and from the method of singularities using 7 singularities and 38
measurementsat the positions shown in Figure 53. Figures 59-61 show similar results
for the 9-percent open arearatio tests. As can be seen in these figures, even when only
sparse data was available, the method of singularities captured the general trend in the
pressure profiles over the central region of the test section.
5.2.3 Effects of Uncertainties in Porosity Parameter
The effect of the porosity parameter specified in this method was insignificant.
As discussed earlier, the three-dimensional formulation of the method of singularities is
continuous in its behavior at P = 0. Figure 62 shows the effect on root mean square
errors of varying the porosity parameter from 0.0 to 1.0. These root-mean-square errors
were based on a point by point comparison of the experimental measurements from the
20 degree angle of attack porous wall tests and the method of singularities solutions
developed using porosities of 0.0, 0.5 and 1.0 in the panelling solutions. The complete
set of 342 experimental measurements were used for these comparisons, and 16
singularities (see Table 3) were used to develop the method of singularities solutions.
All root mean square errors were normalized by peak pressure coefficients measured
during the test. Virtually no change was observed in the root-mean-square errors for
these changes in the porosity parameter.
47
-.30
-.25
-.20
-.15
r-J
O -.10
-.05
0.00
atl
.05
.10 I I I I
-2. -1. O. 1. 2. 3.
)de
Figure 9. Velocity perturbation prof'fles on the top and bottom walls predicted by
Navier-Stokes solver using a slip wall boundary condition.
.030
¢/1
.028
g
.026
.024
Z
.022 Source, Vortex and 1 Derivative of Each

Source, Vortex and 2 Derivatives of Each
.c_
Source, Vortex and 3 Derivatives of Each
g .o2o Soume, Vortex and 4 Derivatives of Each
LU
crj
.018 ._.. ¢C O ¢ ¢ 0 O 0
n-
7 '7. "7 _' _7 V
.016 I I I
O. 50. 100. 150. 200.

Number of Data Points Used
Figure 10. Root-mean-square errors as a function of the number of data points used to
develop the method of singularities solution. Errors are normalized by peak pressure
measured.
48
-.4
Sdid Wal Wind Tunnel Data
Method of Singula_ies Fit to Solid Wal Data
Porous Wall Wind Tunnel Data
-.3 Method of Sin_lularities Fit to Po¢ous Wall Data
-.2 Upper Wall
-.1
0.0
.1 Lower Wall
.2 = I I I
-2. -1. O. 1. 2. 3.
X/C
Figure lla. Method of Singularities matches to pressure profiles on the wind tunnel
walls for airfoil tests at (x = 5 degrees. U_ = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
-1.5
CFO Solution using Sold Wall Procure Boundary
Condition from the Me_hod o_ Singularities
Pocous Wall Wir_l Tunn_ Data
J.._. Solid Wal Wind Tunnel Test Data
-1.0 CFD Solution using Porous Wall Pressure Bounda_,
Condition from the Method o( Si_utarities
-.5 Upper Surface
.0
Lower Surface
.5
1.0 ! I I
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
Figure llb. Comparison of experimental and CFD pressure profiles on the airfoil for
(x = 5 degrees. Uoo = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
49
-.4
Method d Singularities Fit to Solid Wal Data
Porous Wa_ W'_I Tunnel Data
i _ Solid W -,I W'm<l Tunnel Dart
Me/hod _ $in_ulan'Ses Fit to Porous Wall Dat_
-.3
-.2
Q.
-.1
0.0
.1
.2 = = i j
-2. -1. 0. 1. 2. 3.
x/c
Figure 12a. Method of Singularities matches to pressure profiles on the wind tunnel
walls for airfoil tests at o_ = 6 degrees..U,_ = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
-2.0
CF0 _ using Solid Wall Pressure Bounda_
C_d'_o_ from the Melhod ol Singularities
Por¢_ Wall Y_nd Tunnel 13a_a
- 1.5 CFD Soh_Jon us_g Porous Wall Pressure Bounda_
co_rdioq from the Med'x_d of Sin_lularitie=
-1.0
".5
.0 I
.5
1.0 t ! I I t , t
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x/c
Figure 12b. Comparison of experimental and CFD pressure prof'des on the airfoil for
o_= 6 degrees. U_ = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
50
-.4
,Solid Wel Wind Tunnel Data
Method of S_ularities Fit to Solid Wd Data
-.3 Melhod of Sir_ularflie= Fit to Porous Wall Oat_
-.2
t-_
¢.2 -.1
0.0
.1
.2 = I f w
-2. -1. O. 1. 2. 3.
X]C
wails for airfoil tests at tx = 7 degrees. U_ = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
-2.0
CFD Solution using Sold Wall Pressure Bounda_
Condition from the Method of Singularities
-1.5 J._ Solid Wall Wind Tunnel Test Data
CFD Solution us_g Porous Wall Pressure Bout, daft
Condition lrom the Method of Sin_luku'ities
-1.0
O -.5
.0
.5
1.0
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
Figure 13b. Comparison of experimental and CFD pressure profiles on the airfoil for
o_ = 7 degrees. U_ = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
51
-.4
Solid Wal Wind Tunn_ Data
Method ol SingulwJnties F'd to Solid Wall Data
Porous Wall Wind Tunnel Oata
-.3 Method of Sin_lulanlies Fit to Porous Wall O_a
-.2
CL
O -.1
0.0
.1
.2 I = , =
-2. -1. O. 1. 2. 3.
x/c
walls for airfoil tests at (x = 8 degrees. U,o = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
-2.5
4, Solid Wal Wind Tunne_ Test Data
CFD Solution using Sold Wall Pressure Bouoda_
Condition ITom the _ of Singularities
-2.0 <> Porous Wall Wend Tunne_l.Data
"" - CFD Solution u_ing Poror._ Wall Pressure Boundaly
-1.5
-1.0
O.
c_
-.5
.0
.5
1.0 I I I I I
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x/c
(x = 8 degrees. U_ = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
52
-.4
Method a( 8;ngukattiu Fit to Sold W=II Data
O Porous Wall W'md Tunnel Da_
I • Solid Wd _ Tunnel Data
- - - Method oi Sin_ukulti_ Fit to Poroul wan
-.3
-.2
cO -.1
0.0
.1
.2 i = = =
-2. -1. O. 1. 2. 3.
X/C
Figure 15a. Method of Singularities matches to pressure prof'des on the wind tunnel
walls for airfoil tests at o_= 9 degrees. Uoo = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
-3.
CFO Solution using Solid Wall Pressure Boundary

i._O Solid Wall Wind using
Turmel Test Wall
Data Pressure
CFD Solution Porous Boundmy
Condition from the Method of Sin_lularities
-2.
-1.
1. I I I I I I I
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
ot = 9 degrees. Uoo = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
53
-.4
t Solid Wal Wind Tunnel Data
-- Method d Singularilies Fit to Solid WaR Dala
0 Porous Wall Wind Tunnel Data
-.3 " - - Method of Sin_luladfies Fit to Porous WaR Data
-.2
CL
C3 -.1
<>
0.0
.1
.2 ! i i j
-2. -1. 0. 1. 2. 3.
x/c
walls for airfoil tests at o_ = 10 degrees. Uo_ = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
-4.
I • Solid Wal Wind Tunnel Test Data
I _ CFD Solution using Solid Walt Pressure Boundmy
• | _ Condition from the Me_hod of Singularities
I O Porous Wall W_nd Tunnel Data
-3. ._ I" " " CFD Solution using Porous Wall Pressure Bounds,
| Condition from the Method of Sin_/ularities
-2.
t-_
¢..)
-1.
,,_ __,.... _ --- ¢-_¢.- ......

° I I I I I I
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x/c
ct = 10 degrees. U_o = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
54
-.4
Method c_ Singuladli_ Fit to S_id Wall Da_
I _ Solid Wal
Porous WallWird
W'md
Tunnel
Tunnel Data
-.3 Method o_ Singularities Fit Io Porous Wall Data
-.2
tel
-.1
0.0
.1
.2 I ! I I
-2.
-1. O. 1. 2. 3.
X/C
walls for airfoil tests at (x = 11 degrees. U_ = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
-5.
• Solid Wal Wind Tunnel Test Data
CFD Solution using So&:l Wall Pressure Boundaly
Cot_dJon from the MeOzod o_ Singularities
-4. 0 Poro_J= Wall Wind Tunnel Data
" " " CFD Soluti<m usktg Poro_ Wall Pressure Boundary
-3.
O.
-2.
-1.
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
(x = 11 degrees. Uoo = 24. m/s, Re = 3.2 x 105, h/c = 2.25.
55
.07
.06
t-
O
.05
.04
.03
c-
.02 1st derivatives retained

LU 2nd derivatives retained
CO 3rd derivatives retained
.01
4th derivatives retained
cr
.00 , = I , I t =
-1. O. 1. 2. 3. 4. 5. 6.
Porosity Parameter (P) Specified on Walls
Figure 18. Root-mean-square errors as a function of porosity parameter specified in

method of singularities solution. Experimental data from _ = 8 degrees, porous wall
wind tunnel test.
-.3
• Experirnental Data
Method of Singularities Fit
Using P = O.
-.2 Method of Singularities F'_
Using P = 1.
-.1
¢-_
0.0
.1
.2 = I = I
-2. -1. O. 1. 2. 3.
x/c
Figure 19. Method of singularities fits to experimental data using a porosity parameter
of P = 0. and P = 1. Experimental data from the porous wall wind tunnel test at o_ = 8
degrees.
56
-2.0
Experimental Data
CFD Result using Boundary Condition based on P--0

-1.5 CFD Result using Boundary Condition based on P=I
-1.0
Upper Surface
Q.
-.5
.0
.5
1.0 _m
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
Figure 20. Effect of using boundary conditions based on porosity parameters of P = 0.
and P = 1. on computed airfoil pressure profiles for (x = 8 degrees. Boundary conditions
used are shown in Figure 19.
.
t/)
e- Source
l20 Vortex
t--
o .
First Derivative of Source
tJ_ ----I--- First Derivative of Vortex
---e---- Second Derivative of Source
-----e--- Second Derivative of Vortex
=
t2o
t--
°m
¢a9
°
-1. I I I I I
o0 .2 .4 .6 .8 1 o0
Figure 21. Effect on singularity strengths of changes in porosity parameter specified in

method of singularities. Method of singularities solutions were based on ot = 8 degrees,
porous wall wind tunnel test.
57
.020
ta
t-
O
.015
tJ
.c:
¢.-
.010
ta
.t-
IL_
o .005
III
CO
tr
.000 v t v t = t n n n
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x/c Placement of Singularities
Figure 22. Effect of changes in the position of singularities on the root-mean-square

errors in matching pressure prof'des for the o_ = 8 degrees, solid wall wind tunnel test.
Figure 23. Grid used in the o_ = 8 degrees simulation using viscous wall boundary
conditions.
58
Figure 24. Grid used in the free-air simulations.
59
-3.
Experimental Data
Method of Singularities Boundary Condition
Slip Wall Boundary Condition
No-Slip Boundary Condition
Free-Air Calculation with Corrections Applied
-2.
Q.
rO -1.
1. _____1 I .!__
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
Figure 25. Pressure profiles on the airfoil from CFD calculations using specified
pressure, no slip wall, viscous wall and free-air boundary conditions.
Figure 26. Refined grid for _ = 8 degrees test case. Grid dimensions are 350 x 105.
60
-2.5
...... 250 x 70
375 105Grid
Grid
-2.0
-1.5
-1.0
c.)
-.5
.0
.5
1.0 = = w I = =
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x/c
Figure 27. Comparison of results from 250 x 70 grid and 350 x 105 grid. Results are
for solid wall test at ot = 8 degrees.
61
-.4 O 77 Data Points Measured on Wall=
Method of Singularities F'd to 77 Oat_ Points
• 8 Data Points Measured on Walls
- - - MeOtod of Sin_=uladties F'd to 8 Data Poir_

-.3
-.2
Q.
¢.9 -.1
0.0
Lower Wall
.1
.2
-2. -1. O. 1. 2. .
X/C
Figure 2,8a. Comparison of method of singularities fit to complete set of data and sparse
data on the walls for solid wall test at tx = 5 degrees.
-1.5
_ OFt)
CFD _ SoluSoct U=ing
Using the
the Method
Method o_
o( Singulars
Sk_gula'ities
Fit to 8 Data Point= a= a Bounda_f
-1.0
Fit to 77 Data Points as a Boundary C_',dition

Upper Surface
-.5
t'_
¢.9
.0
Lower Surface
.5
1.0 I I I ! I l I I !
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
Figure 28b. Comparison of CFD airfoil pressure profiles obtained using the boundary
conditions developed from sparse and fine data, shown in Figure 28a.
62
-.4 77 Data Points _ed on Wafl¢
m Method o_ Singularities F'd to 77 Data points
8 Oats Points Measured on WId_
i O
Method o_ Si_ularilies F'd to 8 Data pob/s
-.3
-.2
CL
c) -.1
0.0
.1
.2 i , , i
-2. -1. O. 1. 2. 3.
x/c
Figure 29a. Comparison of method of singularities fit to complete set of data and sparse
data on the walls for solid wall test at (x = 6 degrees.
-2.0 CFD Solutkm Using the Method of Singuk._m_

F'd to 77 Data Points as a Boundazy Condib<_
- - - CFD Solution U_dng the Method o( Singularities
=ul • Boundar
-1.5
-1.0
CL
-.5
.0
.5
I I I I I I I I
1.0
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
63
-.4 77 Data Points Meas_ed on Wall=
Method o( Singuladbe$ Fit to 77 Data Points
¢ 8 Data Points Measured on Walb
Method ol _i_ Fit to 8 Data Pointl

-.3
-.2
o -.1
0.0
.1
.2 t , ...L_ ..j_
-2. -1. O. 1. 2. 3.
X/C
data on the walls for solid wall test at tz = 7 degrees.
-2.0
F'd to 77 Data Points as a Bound==ry Condition
CFD Solution Usir_ the Method d Singularities
!.._ CFD Solution Using the Method ell Singutatities
Fit to 8 Data Point= as • Bounda_ Condition
-1.5
-1.0
-.5
.0
.5
1.0 I I , I I I I I !
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
64
-.4 77 Data Points Measured on Walls
m Method ol Singularities Fit to 77 Data Points
II 8 Data P_ Measured on Wall=
I O
of Sin_ula_lie_ Fit to 8 Data Points
-.3
-.2
-.1
0.0
.1
.2 , i I i
-2. -1. O. 1. 2. 3.
x/c
data on the walls for solid wall test at c_ = 8 degrees.
-2.5 CFO Solutk_ Using the Method o( Singularities

Fit to 77 Data Po_t$ as a Boundary Condition
CFO Solution Using the Method o_ Singularities
Fit to B Data Points as a Bounda_ Condition
-2.0
-1.5
-1.0
Q.
C)
-.5
.5
J 1 L.............__J._
1.0 ___
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x/c
Figure 3lb. Comparison of CFD airfoil pressure profiles obtained using the boundary
65
-.4
Idelhod ol Singula.,ilie¢Fa to 77 Data Points
•o 8 Data
I 77 Oatapoint=
PointsMeasured
Measuredon onWall,,
WeJll
Method ol Sin_ul_ritie_ Fit to8 Data Points
-.3
-.2
O -.1
0.0
.1
.2 , D = !
-2. -1. O. 1. 2. 3.
x/c
data on the walls for solid wall test at tx = 9 degrees.
-3. I _ CFD Solution Uldng the Melhod cd Sk_gularitil=
I_ I F=t to 77 Data Points as a Boundary Condition
-2.
Q.
-1.
0•
! I I I ! I
•
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
66
-.4 I o 77DataPoin_Mea_redonWalb
-- Method of Singulsrities F_ Io 77 Data Point=
• 8 De_ Points I_kmsumdon Walb
-.3
-.2
--- I_U'_od of Si_ut_im F'dto 8 im Points
-.1
0.0
.1
.2 , , i ,
-2. -1. O. 1. 2. 3.
x/c
data on the walls for solid wall test at tz = 10 degrees.
-4. CFD _ U=Jngthe Method of S_u_

Fit 1o 77 DeSa Points as a Boundmy
CFO Solution Using the Method d S_r_ufadtJes
FI to 8 De_ Points m e
-3.
-2.
Q.
o
-1.
I I I I I I I
°
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
67
-.4 l o 77 Data Points Measured on W_
| -- Method of Singularities Fit to 77 Data Points
/ • 8 Data Points Measured on Walls
-.3
-.2
Q.
-.1
0.0 j_
.1
I I I I
.2
-2. -1. O. 1. 2. 3.
x/c
data on the walls for solid wall test at ot = 11 degrees.
-5.
F'_to 77 Deta Points as a Boundmy Co_lli6oq
CFD
CFD Solution
Solution Using
UJ_g the
the Me_od of Singu_
-- _ _ Singularities
FI to 8 Data Points as • Bo_
-4.
-3.
O-
¢9 -2.
-1.
Oo
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
conditions developed from sparse and free data, shown in Figure 34a.
68
-.4
Method of Singularities F'd to 77 Data Points
• 8 Data Points Me_ured on Walb
I ¢ 77 Data Points Measured on Wafts
Method (_ Si_ularlles F_ to 8 Data Points
-.3
-.2
c_ -.1
0.0
.1
.2 = , I ,
-2. -1. 0 1 2 3
x/c
data on the wails for porous wall test at o_ = 5 degrees.
-1.5
Fit to 77 Data Points ,,- a Boundaff Condition
-- CFD
CFD Solution
Solution Using
Using the
the Method
Method d Singularities
o( S_gulsriSes
Fit to 8 Data Points u • Bo Condition
-1.0
-.5
.0
.5
1.0
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x/c
69
-.4
Method _[ Singulacities Fit to 77 Data Points
• 8 I_t,a Po_l_ Mea=ured on Wall=
I 0 77 Deta Poinl= Meamured on Wall=
Method oil Sin_pJ_ie_ F=t to 8 Data Points
-.3
-.2
Q.
o -.1
0.0
.1
.2 i i , ,
-2. -1. O. 1. 2. 3.
x/c
data on the walls for porous wall test at o_= 6 degrees.
-2.0 / I -- c_ sou= u=.=_ _._ o,Sin===_
-1.5
I F=t to 77 Data Points as a Boundan/
CFD Solut_a U_ng me Method
Fit to 8 Data Points
Condition
o( Singularities
as a Bounda_, Conditioct
-1.0
¢,.) -.5
.0
.5
1.0
'0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
70
-.4 77 Data Points Measured on Walk
Me_Kxl ol Slngularltin Fd to 77 Data Point=
8 Data Points Measured on Wa_b
I O
MeOmd o( Si_uladtie= Fd to 8 Data Point=
-.3
-.2
¢-_
¢D -.1
0.0
.1
.2 = I I =
-2. -1. O. 1. 2. 3.
X/C
data on the walls for porous wall test at ot = 7 degrees.
-2.0
Fd to 77 Data Points a= a Boundary Condition
J _ CFO Solution Using the Method o( Singularities
- - - CFD Solution Using the Method o( Singularities
• o.
-1.5
-1.0
cO -.5
.0
1.0 i ! i , i !
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
conditions developed from sparse and f'me data, shown in Figure 37a.
71
-.4
Method of S_,n_ularities F'd to 77 Data Po_,s
• 8 _ Points Measured on W_II
I o 77 Data Po_s Measured on Wal_
Method of Singul_tlies Fd to 8 O_ Points
-.3
-.2
¢-_
¢,) -.1
0.0
.1
.2 i i i i
-2. -1. O. 1. 2. 3.
x/c
-2.5
F'd to 77 Data Points as 8 Boundary Condition
CFO So_'ion U_ng the Method o( SingulariSes
L I._---_ o_FI s_._
to 80ata u_ _s_._
Points as a Bounda_f Condition
-2.0
-1.5
-1.0
O.
o
-.5
.0
.5
1.0 I I I I I I I I I
•0 .1 .2 13 .4 .5 .6 .7 .8 .9 1.0
X/C
72
-.4 77 Data Points _ oct Wahl
m
Method ol Singularib_ Fit to 77 Data Points
8 Data Points MeastmKI on Walb

i O
-.3 Method of Sin_luladliu Fit to 8 Data Points
-.2
o -.1
0.0 .. O
.1
s s
.2 , , I ,
-2. -1. 0. 1. 2. 3.
x/c
data on the walls for porous wall test at ct = 9 degrees.
-3.
Fit to 77 Data Points as a Bo_tdmy Condition
CFD Solutio¢_ Using the

the Me_od ol Sk_gularities
-- CFD Solution Using Me_xl of Singularities
Fit to 8 Data Point= as = Bounda_¢ Condition
-2.
O -1.
Oo
o I I I I I I
o0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
73
-.4 0 77 Oats Point= Measured on Waits
-- Method ol Singularities Fit to 77 Data Po_ntl
• 8 Data Points Measmed on Wafts
" - - Method ot Sil_ulariliel F'd to 8 Data Pointt
-.3
-.2
O.
-.1
0.0
.1
.2 : = , I
-2. -1. O. 1. 2. 3.
x/c
-4. Sok_=. Usingthe Method¢_S_u_
L I-
I='d to 77 Data Points as a Boundaly Cor<litkm
CFD Solul_ Using the _ of Singularibes
Fil to 8 Data Points as • Bounda_
-3.
-2.
O.
c_
-1.
1. I I I I ! I
.0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x/c
74
-.4 ¢, 77 Data Point,, Measured o_ Wa_
Method o( Singularities Rt to 77 Data Points
• 8 Data Points Measured on W_It
- - - Method of Sln_lularlties Fit to 8 Data Points

-.3
-.2
¢D -.1
0.0
.1
.2 l l i l
-2. -1. O. 1. 2. 3.
x/c
data on the walls for porous wall test at tz= 11 degrees.
-5.
CFD Solution Using the Method of Singularities
Fit to 77 Data Points as a Boundaly Coalition
CFD Solution Using the Method of Singutarities
Fit to 8 Data Points as a
-4.
-3.
t'>
O -2.
-1.
•0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
X/C
Figure 4lb. Comparison of CFD airfoil pressure profiles obtained using the boundary
75
.O2
.00
-.02
8
-.04
-.06
V Bottom Wall, 5-Hole Probe Data

-.08 A Top Wall, 5-Hole Probe Data
Bottom Wall, CFD Result
II
Top Wall, CFD Result
-.10 =
-2. -1. O. 1. 2. 3.
X/C
Figure 42. Normal velocities near the walls, as measured by five-hole probe and
computed by CFD code. Porous wall tests, o_= 5 degrees.
.O2
-.04
V Bottom Wall, 5 Hole Data

A Top Wall, 5 Hole Probe Data
CFD Result Bottom Wall
CFD Result Top Wall
X/C
Figure 43. Normal velocities near the walls, as measured by five-hole probe and
computed by CFD code. Porous wall tests, o_= 8 degrees.
76
-.25
y/W
_.oooooeoOooooOOOooo • 0 0
oO0 Ow''
Experimental Data
-.25
• •, o • • o Method of Singularities Fit to D_
y/W = .4167 •
-.25 g
y/W = .3611 • •
• •
• 0
0
o
9
-.25
y/W = .3056 • •
• •
• 0
0
-.25 g
y/W = .2500
¢-_ 0 O°°°°°°°° O
0
-.25 0
y/W =. 1944 • •
-.25
y/W = .1389
g 0
g
0
-.25
y/W=.0833 g •
g
8
-.25
y/W=.0289 g
g
O0 , ,v, ,
• I ' ' ' ' I' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' , , , i , , , ,
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

x/c
Figure 44a. Upper wall experimental data and method of singularities fit to the
experimental data, using 16 singularities and 342 measurements to determine method of
singularities solution. U.. = 24. m/s, Re = 2.4 x 105, h/c = 3.0, (_ = 20 degrees. Solid
wall boundary condition.
77
-.25
y/W = .9711
0
-.25
y/W = .9167
•
•o 8 8 8 8 J
-.25"
y/W = .8611
8 e
e
-.25
y/W = .8056
8 8 8 • • • •
-.25 •
y/W = .7500
S
8
-.25
y/W = .6944
8 • • • • •
8
0
-.25
2
y/W = .6389
8
2
-.25
y/W = .5833
• • 0 •
• 0
0
0
-.25
y/W = .5278
• 0
!oeEXp erimental Data

0.0 ' '- ' I ' '' ' I ' ' ' ' I ' ' ' ' i , , , , i ,
Method of Singularities
' I' ' ' ' I ' ' , , i
Fit to D,ata,
,
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
Figure 44b. Upper wall experimental data and method of singularities fit to the
singularities solution. Uo. = 24. m/s, Re = 2.4 x 105, h/c = 3.0, _ = 20 degrees. Solid
78
• -.16
y/W = .5278
0
8 I•0 Experimental
Method Data
of Singularities Fit to Data
-.16
y/W = .4722
0
• • • • 8 • • ooooOooeooooOO°°°°°°eeo°
-.16
y/W = .4167
0
-.16
y/W = .3611
8
0
-.16
y/W = .3056
¢..)
-.16
y/W = .2500
oOOO 0
O o
oO
• e _3_°
-.16
y/W =. 1944
0
0 8
-.16
y/W =. 1389
0 • 0
0
0
e 8 = o
-.16
y/W = .0833
8
8
8 it
.04 .... I .... I .... U .... I ' ' ' ' I ' ' ' ' I .... I ' ' ' ' I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
Figure 45a. Lower wall experimental data and method of singularities fit to the
singularities solution. U,_ = 24. m/s, Re = 2.4 x 105, h/c = 3.0, (z = 20 degrees. Solid
79
-.16
y/W = .9711
It it •
JOeExperimental Data
Method of Singularities Rt to Data,
-.16
y/W = .9167
-.16
y/W = .8611
il
-.16
=.8056
-.16
y/W = .7500
0
o 0
-.16
y/W=.6944
• • • •
0
8
8
-.16
y/W=.6389
Q
8
-.16
y/W = .5833 •
8
8
.04 .... I .... i .... I .... I .... I .... I .... I ' ' ' i I i i i i
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

x/c
Figure 45b. Lower wall experimental data and method of singularities fit to the
singularities solution. U_, = 24. m/s, Re = 2.4 x 105, h/c = 3.0, _ = 20 degrees. Solid
80
•-.14
z/h = .3056 •
• •
0 I_) Method
E'xpelimentalData
of Singula_ties Fit to Data
-.14
z/h = .2500 •
o • 8
-.14
_.1_ - - e _ • •
o e 8
o o
-.14.
_---J=
_ ..... i
e 8
-.14
Wh = .0833 e 8 o
-.14
z/h = .0278 8 8 8
0
r,_-.14
z/h = -.0278 88eeeeeSe888e8888ee
-.14 00303 8 e 8
z/h= 8
8
8
0
-.14 8
z/h = -.1389 8
8
-.14
z/h = -.1944 8 8 8 8 •
8 8
e
-.14
zJh= -.2500 8
0
-.14
_h =-.3056 •
• O
$ o
0.0 .... I .... I .... I' ' ' ' I .... I .... I .... I .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

X/C
Figure 46. Side wall experimental data and method of singularities fit to the
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, h/c = 3.0, o_ = 20 degrees. Solid
81
-.14
y/W = .4722 ooooO00OoOOOO00oooO0000o • • •
Method of Singularities Fit to Data

-.14
y/W = .4167 •
• •
-.14
y/W = .3611 0 • •
0 •
0
-.14
y/W = .3056 • • •
g
0
• •
-.14
y/W = .2500 •
oOOO Oo°°
oO
8 • •
-.14
y/W = .1944 0
0
8
-.14
y/W =. 1389 8 •
o • •
-.14
y/W = .0833
-.14
y/W = .0289 0
0 0
.O2 .... i ' ' ' ' I ' ' ' ' I .... I .... I ' ' ' ' I ' ' ' ' I .... I ' ' ' '
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. U_ = 24. m/s, Re = 2.4 x 105, h/c = 3.0, o_ = 20 degrees. Porous
82
-.14
y/W = .9711
8 8
-.14
y/W = .9167
0 0 • l
8 8 0
-.14
y/W = .8611
0
8 0
-.14
y/W = .8056
0 • 0
0
-.14
y/W = .7500
Q. 0
tO
-.14
y/W = .6944
0
0
-.14
y/W = .6389
0
-.14
y/W = .5833
$ 0 8 •
0 0
-.14
y/W = .5278
0
!_ Experimental
Method of Singularities
Data Rt to Data
.02 .... I .... I ..... n 'l .... I ' ' I ' ' ' ' I .... I ' ' ' '
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
_C
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, h/c = 3.0, cz = 20 degrees. Porous
83
-.06
=.5278
0
-.06 G
, y/W = .4722 O0000oooooooOOOOOOooOO o°

_Ooeee
-.06
y/W = .4167
0
0
O
-.06
y/W = .3611
-.06
y/W=.3056
8
0
¢O
0
-.06
y/W = .2500
• o o°88°0
• oO O0
-.06
y/W = .1944 8
8 •
-.06
y/W=.1389
0
0
• 0 0
-.06
y/W=.0833
0
0
,_ 0 ! O Experimental Data
Method of Singularities Rt to Da_
.O6 ' ' ' ' I .... I' ' ' ' I ' ' ' I ' ' ' ' I ' ' I' ' ' ' I ' ' ' '1 '
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2. 2.5
X/C
singularities solution. U_ = 24. m/s, Re = 2.4 x 105, h/c = 3.0, _ = 20 degrees. Porous
-.06
y_=.9711 O
O g
I
I_ Experimental
Method Data
of Singularities Fit to Data
-.06
y_N = .9167 8
8 O 8
I
-.06
y_N = .8611 8
-.06
y_V=.8056
tl
-.06
y/W=.7500
¢._ II
-.06
y_V=.6944 i
II 8
t t t
t
-.06
i y/W=.6389 I
8 8
-.06
y/W=.5833 i
0
I
.O6 .... I .... I .... I .... I .... I .... I .... I .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, h/c = 3.0, o_= 20 degrees. Porous
85
-.08
z/h = .3056
8
-.08
z/h = .2500
o
8 0
-.08
z/h = .1944
Q
• 0
-.08 o t • •
= .1389
0 0
-.08"
= .0833
8
-.08
z/h = .0278
8
0
- -.0278
0 o _ • eoOeeeseeeeeeeeee°e
-.08
z/h = -.083,3 •
O
0 $
-.08
z/h = -.1389 0
0
8
-.08 D
z/h = -.1944 8 o
8 8 8 e e
-.08
z/h = -.2500
0
0
8
-.08
z/h = -.3056
0
.8
• •
tO,Experimental Data
Method of Singularities Fit toData
.01 .... I .... I .... I ' '''l .... ,' ', .... i .... ,' '
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
X/C
singularities solution. Uo. = 24. m/s, Re = 2.4 x 105, h/c = 3.0, ct = 20 degrees. Porous
86
Z •
0.9-"
0.8--
0.7-.
0.6-.
• • • • • • • ••eooooooooeeooeoeoeooo• e • •
0.4- "
0.3- • • • • • • * • •
• ••••••eoee• •
0.2-.
0.1_ •
i
Upper Wall
0 " • • • e,
.... I .... I .... I .... I .... I .... I ' ' " i .... ! ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

X/C
1 •
°-71:
0-61.
o5 :
i
°,1:
03-]:
°21:
°1ol-
, ................ I .... I .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

X/C
0.5
0.4-
0.3-
0.2-
0.1
_ 0, • • • • • oeoeoeooooooooeoo•o
-0.1 -
-0.2'
-0.3-
-0.4
Side Wall
-0.5 ........ I ........ I ........ ] ........ [ I , ;
"2 "1'.5 -1 _'0'.5 0 015 1 115 2 2.5

X/C
Figure 50a-c Locations of the complete set of 342 measurements on the top, bottom
and side walls.
8?
1
0.9
0.8-
0.7•
0•6-
_>0.5- • • • • • • • • oeoeooeoeoeo6oooeooeoeoo • • •
0.4.
0•3.
0•2.
0.1.
02 . . . Upper Wall
-2 ........ -1'.s- ""d.; _ ' " o
d ..... 'ols 1 I ..... '1'.s 2 ! " " ' '2•s
x/c
1
0.8"
0.7-
0.6-
_>, 0.5- • • • • • • • • •ooooo oeooe e•o eo•• ••0 • •
0.4-
0.3-
0.2.
0.1
Lower Wall
0 .... I .... I .... I .... I .... I .... I .... I .... I ....
-2 -1.5 -1 -0.5 0 0•5 1 1.5 2 2•5

x/c
0.5
0.4-
0.3 ....
0.2 ....
u,1 • ° •
0
• • • • • ••eoo•oooooo•o•oo••
- u. -
"1: " " *
-0.2: • " "
-0.3 ....
-0.4.
Side Wall
-0.5 .... _ .... _ .... _ .... _ .... _ .... _ .... _ ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
_C.
Figure 51a-c. Locations of the reduced set of 223 measurements on the top, bottom and
side walls. Measurements are distributed along the centerlines of all three walls and 3
cross-planes.
88
0.9"
0.8-
0.7-
0.6-
_>,0.5- • • • ooooeoeoeeoooee6eooeoooo • • •
0.4.
0.3-
0.2-
0.1.
• Upper Wall
02 .... I .... I .... I .... I .... I .... I .... ! .... 'l ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

x/c
1-
0.9"
0.8.
0.7.
0°6"
0.5.
• • • •oeoeeoeooooeeooeeeeeee• •
0.4-
o.3-]
0.2.
0.1•
Lower Wall
.... I .... I .... I .... t .... I .... ! .... I .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

_C
0.5
0.3
0.2
0"41
0.1
o:
• ••••••ooeooooooooo•
-o._:
-0.2.
-0.3.
-0.4.
Side Wall
-0.5
-2 -;s -1 -o.s o o'.s _ _'s 2 zs
........ I .... I .... I ........ [ ........ I I ' I
x/c
Figure 52a-c. Locations of the reduced set of 136 measurements on the top, bottom and
side walls. Measurements are distributed along the centerlines of all three wails and 1
cross-plane.
89
1
0.9-
0.8-
0.7-
0.6-
_>0.5-
0.4.
0.3.
0.2.
0.1 " •
0
" .... I .... I .... I .... I .... i .... ( .... i
Upper
........
Wall
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
X/C
0.92
0.8-
0.7-
0.6-
_>05.
0.4.
0.3. •
0.2.
0.1. •
0 : Lower Wall
.... I .... I .... I .... I .... I .... I .... I .... I ....
°2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5

X/C
0.5
0.4-
0.3-
0.2-
0.1 o
_ o
.0.1
-0.2 •
-0.3- o
-0.4-
Side Wall
-0.5
-2 ........ -l'.s - 1 .... -0.5 I .... o I ........ ols 1 I ........ 1'.5 2 I ' ' ' zs
x/e
Figure 53a-e. Locations of the reduced set of 38 measurements on the top, bottom and
side walls. Measurements are distributed along the centerlines of all three walls and 1
cross-plane.
90
.13 3 Slngularltte4
1/) 7 Singularitlas
.12 11 Slngular_es
16 Singularities
.11
8
.10
.09
J .08
.c
Ik,,, .07
UJ .06
rr .05
I I I I I I
.04
0. 50. 100. 150. 200. 250. 300. 350.
Number of Points Used to Determine Singularity Strengths
Figure 54. Effect on root-mean-square errors of using various numbers of singularities

and various numbers of data points to develop method of singularities solutions. Results
are shown for solid wall test at (_ = 20 degree.s.
.13
3 Singularities
ct)
7 Singularities
.12 ---4)-- 11 Singular'tie4
16 Singularities
A A A
.11
.10
L- .O9
2
L-
uJ
.O8
mr-
.07 I I I I I I
0. 50. 100. 150. 200. 250. 300. 350.

Number of Points Used to Determine Singularity Strengths
Figure 55. Effect on root-mean-square errors of using various numbers of singularities

and various numbers of data points to develop method of singularities solutions. Results
are shown for porous wall test at (x = 20 degrees.
91
-.25
i y/W = .4722 ooooOOOoOoOOOooo 0 O 0

doO0 O°° _ _.
0 0 • • • • _ 0 "- • Experimental Data

0 Method of Singularities Fit to Data_
-.25
y/W = .4167 • •
0
-.25
y/W = .3611 • • 8 •
o 8
-.25
y/W = .3056 O • 8 O
8
0
0 •
6
-.25
y/W = .2500
I:1. °°eeSo
-.25
y/W = .1944 8 8
0
-.25
y/W = .1389 _ 8
0
$
-.25
y/W = .0833 _ 8
0
-.25
y/W = .0289 _ 8
0
0.0 .... I .... I .... I .... I .... I .... ! .... I ' ' ' ' I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. U_ = 24. m/s, Re = 2.4 x 105, h/c = 3.0, _ = 20 degrees. Solid
92
• -.25
= .9711
y/W 0 8 8
I
-.25
yNV = .9167
• 8 8 • 8 8 o
O @
-.25
y_V = .8611
8 @
O
-.25
y/W=.8056
@ • @ @ •
@
0
@
-.25
y_N=.7500
@
O @
O
-.25
y/W=.6944
0 @ • •
@ 0
@
O •
-.25
y/W=.6389
@
@
0
-.25
y/W=.5833
@ 0
0
o •
-.25
y/W = .5278
@
0 IOf Experimental Data
0.0 ' 'i' ' I .... I .... I .... I ' i L ! m ¢ M_._o.
, , ,i,o, _,_o,D
,=_,
S,r_u,,,..o.
0 , 0 i , , , ,
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, h/c = 3.0, _ = 20 degrees. Solid
93
-.16
y/W = .5278 • •
-.16
S
I"
Expedment_
°"
Method of Singularities Rt to Data
y/W = .4722 ilttllltllllililHNNIO I_INB t

o 8ooeeee
-.16
y/W = .4167
$
0 •
-.16
y/W = .3611
0
-.16
y/W =.3056
0
C_ 0
-.16
y/W =.2500
0
O0000 o°•OO•
8 8
-.16
y/W =.1944
8
-.16
y/W =.1389
8 8 0
0
0
0 8 0
-.16
y/W=.0833
8
0
0
.
04 .... | ! i I ' I ' ' ' ' I' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' '
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, bJc = 3.0, (x = 20 degrees. Solid
94
-.16 .
y/W = .9711
Q
0
e° Method
Experimental Data
of Singularities Rt to Data
-.16
y/W = .9167
8
0
-.16
y/W = .8611
tD
0
-.16
y/W =.8056
-.16
y/W =.7500
C:L
rO 0
-.16
y/W =.6944
• • • •
-.16
y/W =.6389
0 Q
-.16
y/W=.5833
0 8
.04 .... I .... I .... I .... I .... I .... I .... I .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, h/c = 3.0, _ = 20 degrees. Solid
95
-.14
z/h = .3056 8 8
I Experimental Data
-.14 z/h = .2500
o • e e
-.14 : z/h = ,1944 0 •
• • • 0 O
-.14 : z/h = .1389 • 0

i o •
-.14-'. -____ - -- -- ; - ,
i o 0
-.14 : z/h = .0278 8 •
"" " z/h =-.0278 __888880000000

" O
__._I t,, = = e
"-" _ z/h =-10833 8
8
t
- 14 /z/h _ 1389 •
O
O
0
-.14
z/h =-.1944 • o o
• O
i O O
-.14
z/h=-.2500 • •O
• O
0 o
-.14 g
z/h =-.3056 • 0
• O
o
0.0 .... I ' ' '" I .... I .... I .... I ' ' ' ' I .... I .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, h/c = 3.0, o_ = 20 degrees. Solid
96
-.14
y/W = .4722 oOO0OOooooo O 8 8
0880_80000_0 Experimental Data
0 • • • • • • 8 I O Method of Singularitios Rt to Data
-.14
y/W = .4167
8 8
-.14"
y/W = .3611 • •
e 8
g
8 o e
-.14
y/W = .3056 • •
O 8
O
8 8 e
-.14
y/W =.2500 oooe 8
I:l. og
• O O0
o g
-.14
y/W =. 1944 •
0
0
-.14
y/W =.1389 8 8
0
O
-.14
y/W =.0833 •
O
-.14
y/W = .0289 0
O
0
.02 I I I I I I I I
' I ' '' ' I ''' '1' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' , , ,
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, h/c = 3.0, _ = 20 degrees. Porous
97
-.14
yNV = .9711
O 8 0 0
-.14
y_N = .9167
I 0 • 8
8 8 • 8
-14
y_N = .8611
8 0
-14
y_N = .8056
• • • • O
8 •
-14
y/W = .7500
0
¢O
0 •
-14
y/W = .6944
• 0 0 • •
-14
yAN = .6389
0
-14
y/W = .5833
0 0 • 8
0
0
• I
-14
y/W = .5278
• • I_Exp e_mentalData
.02 .... , .... , .... , .... , .... , .... , .... , .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
_C
singularities solution. U_ = 24. m/s, Re = 2.4 x 105, h/c = 3.0, o_ = 20 degrees. Porous
98
-.06
yAN = .5278
-.06 i
y/W@='_@Oooo4722 o••OoooooooOO00@@@_o°_ @'@
-.06
y/W = .4167 @
@ •
-.06
y/W = .3611 @
@ •
-.06
y/W=.3056 @
¢-_
o
-.06 @
yAN = .2500 @ @ o o
@ @ ooO @•°@
-.06
y/W = .1944 @
8 •
-.06 O
y/W = .1389 @ •
• @
o •
-.06
y/W = .0833
@
O
I00 Experimental Data

.06 ' ' ' ' I' ' ' ' I' ' ' ' I' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' '
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. U_, = 24. m/s, Re = 2.4 x 105, h/c = 3.0, _ = 20 degrees. Porous
99
-.06
6
y/V_= .9711 •°
-.06 I" Method

o,mor,ta, 0=,, Fit to Data,,
of SingulariSes
y/W = .9167
8 $ •
-.06
y/W = .8611
o 8 8
-.06
y/W = .8056
tt
• it
-.06
y/W = .7500 8
¢D
-.06
y/W = .6944
8 8
• t
-.06
y/W = .6389 il
IP
-.06
y/W = .5833 8
0
.06 .... I .... I .... I .... 1 .... I .... i .... I .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
x/c
singularities solution. Uoo = 24. m/s, Re = 2.4 x 105, h/c = 3.0, o_ = 20 degrees. Porous
100
-.08
z/h = .3056
tl
= g g
-.08 O Expeximental Data
z]h .2500
g
8
-.08
z/h = .1944 0
e g g
g o
= g g g
-.08
z/h = .1389
0
0
-.08
z/h = .0833
0
-.08-
z/h=.0278
g
e
z/h = -.0278
g o 2 g • oeee__8eeeeee°
-.08
z/h = -.0833
g
• 8 o
-.08
z/h = -.1389
• 8
• 8 o
-.08 g
z/h=-.1944
0 0
• 0
• 0
• 0 0
-.08
z/h = -.2500
• 8 o
-.08
z/h =-.3056
• 8 o
.01 .... I .... I ' ' ' ' I .... I .... I .... I .... I .... I ....
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
X/C
singularities solution. U_ = 24. m/s, Re = 2.4 x 105, h/c = 3.0, o_= 20 degrees. Porous
101
.12
c- .11 A A A
O
O 3 Singularities
-----13--- 7 Singularities
.c: .10 11Singular'_ies
r--
o 16 Singularities
.09
c [] [] []
lb.
o O 0 0
L--
LU .08
O_ V _7 V
rr !
.07 w = , w =
-.2 .0 .2 .4 .6 .8 1.0 1.2
Figure 62 Root-mean-square errors as a function of the porosity specified in the method

of singularities. Data shown is for 0_ = 20 degrees, porous wall test.
102
Chapter 6
Conclusion
A method for describing porous wall boundary conditions based on sparse,
nonintrusive measurements of flow quantities at the wall boundary has been developed.
This method utilized a potential flow solution based on least squares matching of
singularity strengths to measured experimental data. The consistency of imposing a
pressure boundary condition based on this method with other means of describing a wall
boundary condition has been demonstrated in computational studies. This method has
been shown to provide a good description of the entire wall boundary condition even
when as few as 8 measurements were used to develop the method of singularities
solution. The ability of this method to predict the effects of changing porosity in a wind
tunnel test has been demonstrated by simulation of experiments performed in a variable
porosity test section.
This method has been extended to three-dimensional porous wall testing. In
three dimensions, the method of singularities allowed the porous wall boundary
condition to be modelled without the need for normal velocity perturbation
measurements and without the need for extensive calibrations of the wall. An
experimental study has shown that the method was capable of capturing the trends in
pressure profiles existing on the walls in three-dimensional porous and solid wall tests.
In low speed tests, the method of singularities has been found to be rather
insensitive to the value of the porosity parameter specified. This allowed for the
porosity parameter to be found by means of least squares matching. While changes in
the porosity parameter did have a strong effect on the singularity strengths, the effect on
the overall match to the pressure profile was minimal. Since the porosity parameter was
103
not specified as a boundarycondition in the CFD code, andno corrections were madeto
the wind tunnel data, this method eliminated many of the concerns associatedwith the
classical linear porous wall boundary condition. This method also allowed foLa
simulation of the entire flow field and direct comparisonof the flow field to wind tunnel
data without the needfor corrections to the experimentaldata.
104
References
I) Mokry, M., Chan, Y.Y., and Jones, D.L, "Two-Dimensional Wind Tunnel Wall
Interference,"AGARD Report AG-281, 1983.
2) Theodorsen, T., "The Theory of Wind-Tunnel Wall Interference," NACA Technical
Report 410, 1931.
3) Wright, R.H. and Ward, V.G., "NACA Transonic Wind-Tunnel Test Sections,"
NACA Technical Report 1231, 1956.
4) Shujie, W., Yuan, R. and Ruiqin, C., "Investigation of Wall Interference at High
Angle of Attack in a Low Speed Wind Tunnel with Slotted Wall," AIAA Paper 87-2611,
1987.
5) Ritchie, V.S. and Pearson, A.O., "Calibration of the Slotted Tests Section of the
Langley 8-Foot Transonic Tunnel and Preliminary Experimental Investigation of
Boundary Reflected Disturbances," NACA RM L51K14, July 7, 1952.
6) Baldwin, B.S., Turner, J.B., and Knechtel, E.D., "Wall Interference in Wind Tunnels
with Slotted and Porous Boundaries at Subsonic Speeds," NACA TN 3176 May 1954.
7) Marvin, J.G., "Accuracy Requirements and Benchmark Experiments for CFD
Validation," AGARD CP 437, May 1988.
105
8) Lan, C.E. and Thomas, J.P., "Application of Computational Aerodynamics to
Assessment/Correction in Wind Tunnel Testing," KU-FRL-887-1, The University of
Kansas Center for Research, Inc., Sept. 28, 1990.
9) Levy, L.L., "Experimental and Computational Steady and Unsteady Transonic Flows
about a Thick Airfoil," AIAA Journal, Vol. 16, No. 6, June 1978.
10) Potsdam, M. and Roberts, L., "A Numerical Study of the Effects of Wind Tunnel
Wall Proximity on an Airfoil Model," JIAA TR-99, Stanford University, Department of
Aeronautics and Astronautics, Sept. 1990.
11) Lynch, F.T., "Experimental Necessities for Subsonic Transport Configuration
Development," AIAA Paper 92-0158, January 1992.
12) Olson, L.E., "Future Experimental Needs in Low Speed Aerodynamics," AIAA
Paper 92-0157, January 1992.
13) Agrell, N. and Pettersson, B., "Numerical Computations and Measurements of
Transonic Flow in a Slotted-Wall Wind Tunnel," AIAA Paper 87-2610, 1987.
14) A1-Saadi, J.A., and DeJarnette, F.R., "Wall Interference Calculation in a Transonic
Test Section Including Simulation of Discrete Slots," AIAA Paper 92-0032, Jan. 1992.
15) Rizk, M.H. and Lovell, D.R., "Euler Procedure for Correcting Two-Dimensional
Transonic Wind-Tunnel Wall Interference," AIAA Journal, Vol. 26, No. 12, December
1988.
106
16) Gopinath, R., "Wall Interference Evaluation from Pressure Measurements on
Control Surfaces,"AIAA 82-4301, Journal of Aircraft, Vol 19, No. 12, pp. 1097-1098,
December, 1982.
17) King, L.S. andJohnson,D.A., "Calculations of Transonic Flow About an Airfoil in a
Wind Tunnel," AIAA Paper80-1366, 1980.
18) King, L.S. and Johnson, D.A., nTransonic Airfoil Calculations Including Wind
Tunnel Wall-Interference Effects," A/AA Journal,Vol 24, No. 8, August 1986.
19) Jacocks, J.L., "An Investigation of the Aerodynamic Characteristics of Ventilated
Test SectionWalls for Transonic Wind Tunnels," PhD Dissertation, The University of
Tennessee,December1976.
20) Crites, R. and Rueger, M., "Modelling the Ventilated Wind Tunnel Wall", AIAA
Paper92-0035,January 1992.
21) Rueger, M. and Crites, R., "Wind Tunnel Boundary Interference Prediction and
Correction," AIAA Paper92-0036, Jan. 1992.
22) Smith, J. "A Method for Determining 2D Wall Interference on an Aerofoil from
MeasuredPressureDistributions Near the Walls and on the Model," NLR TR 81016U,
January 1981.
23) Cooper, G.K. and Sirbaugh, J.R., "PARC Code: Theory and Usage," Arnold
EngineeringDevelopment Center, AEDC-TR-89-15, December 1989.
107
24) Potsdam,M., andBeutner, T.J., Unpublished resultsusing unstructuredgrid solver,
1990.
25) Shujie,W., Rongxi, Y., and Ruiqin, C., "Investigation of Wail Interference at High
Angle of Attack in a Low Speed Wind Tunnel with Slotted Wall, N AIAA Paper 87-2611,
1987.
26) Murman, E. M., "Computation of Wall Effects In Ventilated Transonic Wind
Tunnels," AIAA Paper 72-1007, September, 1972.
27) Mokry, M., Peake, D.J. and Bowker, A.J., "Wall Interference on Two-Dimensional
Supercritical Airfoils Using Wall Pressure Measurements to Determine the Porosity
Factors for Tunnel Floor and Ceiling," LR-575, National Research Council Canada,
Feb. 1974.
28) Mokry, M., "A Wake-Blockage Paradox in a Perforated Wall Wind Tunnel," AIAA
Journal, Vol. 9, No. 12, pp 2462-2464, December 1971.
29) Slooff, J.W. and Piers, W.J., "The Effect of Finite Test Section Length on Wall
Interference in 2-D Ventilated Windtunnels," AGARD CP 174, Paper 14, October 1975.
30) Katz, J., "Low Speed Aerodynamics From Wing Theory to Panel Methods, AE 620
class notes, San Diego State University, September, 1989.
31) Rogers, S.E., "Numerical Solution fo the Incompressible Navier-Stokes Equations,"
NASA TM 102199, Nov. 1990.
108
32) Beutner, T., Celik, Z., and Roberts,L., "Modelling of Solid/PorousWall Boundary
Conditions for the Validation of Computational Fluid Dynamics Codes," AIAA Paper
92-0033, January 1992.
33) Rogers, S.E., Wiltberger, N.L. and Kwak, D., "Efficient Simulation of
Incompressible Viscous Flow Over Single andMulti-Element Airfoils," AIAA Paper92-
0405, January,1992.
34) Baldwin, B. and Barth, T., "A One-EquationTurbulence Transport Model for High
Reynolds Number Wall-Bounded Flows," NASA TM 102847,Aug. 1990.].
35) Peyret, R. and Taylor, T.D., "Computational Methods for Fluid Flow," © 1983,
Springer-Verlag New York, Inc.
36) Roe, P.L., "Approximate Riemann Solvers, Parameter Vectors and Difference
Schemes,"Journalof ComputationalPhysics,Vol. 43, pp357-372, 1981.
37) "The GRIDGEN 3D Multiple Block Grid Generation System, Vol. 1 and 2,"
WRDC-TR-90-3022, 1990.
38) Braslow, A.L. and Knox, E.C., "Simplified Method for Determination of Critical
Height of Distributed Roughness Particles for Boundary-Layer Transition at Mach

Numbers from 0 to 5," NACA TN 4363, Langley Aeronautical Laboratory, September
1958.
109
39) Braslow, A.L., Hicks, R.M., and Harris, R.V., "Use of Grit-Type Boundary-Layer-
Transition Trips on Wind Tunnel Models," NASA TN D-3579, Washington, D.C.,
September1966.
40) Beutner, T.J, Celik, Z.Z., and Roberts, L., " Determination of Solid/Porous Wall
Boundary Conditions from Wind Tunnel Data for Computational Fluid Dynamics
Codes," AIAA Paper93-0530,January,1993.
41) Beutner, T.J, Celik, Z.Z., and Roberts, L., "Determination of Solid/Porous Wall
Boundary Conditions from Wind Tunnel Data for Computational Fluid Dynamics
Codes," AGARD CP 535, Paper16, October, 1993.
42) Rae, W.H., and Pope,A., "Low-SpeedWind Tunnel Testing, SecondEdition," John
Wiley & Sons,New York, 1984,pp 344-444.
110

Jiaa TR - 111: Joint Institute FOR A R - R - Nautics AND Acoustics

Uploaded by

Copyright:

Available Formats

Jiaa TR - 111: Joint Institute FOR A R - R - Nautics AND Acoustics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Jiaa TR - 111: Joint Institute FOR A R - R - Nautics AND Acoustics

Uploaded by

Copyright:

Available Formats

https://ntrs.nasa.gov/search.jsp?

JOINT INSTITUTE FOR A R__R__NAUTICS AND ACOUSTICS

National Aeronautics and _'7_/e"=t

A Method for the Modelling of Porous and Solid Wind

Thomas John Beutner

reflections in the test section. Their usefulness in developing computational fluid

the wall, which require extensive wind tunnel time.

A computational and experimental study was undertaken to investigate a new

from a heavily instrumented variable porosity test section.

Table of Contents ..................................................................................................... iv

List of Figures .......................................................................................................... vi

List of Tables ............................................................................................................ xi

Nomenclature ........................................................................................................... xii

Chapter 1 Introduction ............................................................................................. 1

1.1 Background ................................ ................................................................. 1

1.3 Previous Approaches to Modelling Porous Walls ....................................... 4

1.4 Current Approach ........................................................................................ 5

Chapter 2 Porous Wall Theory ................................................................................ 8

2.1 Introduction ................................................................................................. 8

2.2 Development of Classical Porous Wall Boundary Condition ..................... 8

2.4 Theoretical Approach of this Work .......................................................... 14

2.5 Extension to Three Dimensions .................................................................. 15

Chapter 3 Two-Dimensional Governing Equations and Navier-Stokes Solver ......

3.1 Introduction ................................................................................................. 1_

3.2 Governing Equations ................................................................................... 14"

3.3 Numerical Solver ........ ................................................................................. 20

3.4 Boundary Conditions ................................................................................... 21

3.5 Generation of Computational Grids ............................................................ 23

Chapter 4 Experimental Apparatus ......................................................................... 25

4.1 Wind Tunnel Test Section ........................................................................... 25

4.3 Test Section Instrumentation ....................................................................... 27

4.4 Wind Tunnel Flow Quality .......................................................................... 27

4.5 Error Analysis .............................................................................................. 28

Chapter 5 Results ..................................................................................................... 33

5.1 Two-Dimensional Case ............................................................................... 33

5.1.1 Preliminary Computational Studies .................................................... 33

5.1.2 CFD Simulations of Experiments..i .................................................... 35

5.1.3 Effect of Changes in Porosity Parameters .......................................... 36

5.1.4 Effect of Changes in Singularity Placement ....................................... 38

5.1.5 Comparison of Alternate Boundary Conditions for Solid Wall

5.1.6 Grid Refinement Effects ..................................................................... 39

5.1.7 Effects of Using Sparse Data .............................................................. 40

5.1.8 Flow Inclinations Near the Boundaries .............................................. 42

5.2 Three-Dimensional Case ............................................................................. 42

5.2.1 Method of Singularities Comparison to Experimental Data .............. 43

5.2.2 Effects of Sparse Data and Reduced Numbers of Singularities ......... 43

5.2.3 Effects of Uncertainties in Porosity Parameter .................................. 46

Chapter 6 Conclusion .............................................................................................. 102

References ................................................................................................................ 104

1 Arrangement of Panels for Panelling Code ..................................................... 17

2 Two-Dimensional, 250 x 70 CFD Grid ........................................................... 24

3 Exploded View of Variable Porosity Wall ...................................................... 29

4 Schematic Diagram of Wind Tunnel ............................................................... 30

5 Photograph of Wind Tunnel Test Section ....................................................... 30

6 Boeing Advanced Transport Airfoil Cross Section ......................................... 31

7 Empty Tunnel Flow Survey, Solid Wall Test Section .................................... 31

8 Empty Tunnel Flow Survey, Porous Wall Test Section .................................. 32

9 Velocity Perturbations on Walls predicted by Navier-Stokes Solver ............. 47

10 Errors in Matching Pressure Profiles on Walls ............................................... 47

JOINT INSTITUTE FOR A RRNAUTICS AND ACOUSTICS