Sie sind auf Seite 1von 8

78-156

i
/

A Solution to the 2-D Separated Wake Modeling


of
Problem and Its Use to Predict Gmax
Arbitrary Airfoil Sections
M.L. Henderson,
The Boeing Co., Seattle, Wash.

AIAA i6TH AEROSPACE


SCIENCES MEETING
Huntsville, AlabamdJanuary 16-18, 1978
For permission to copy or republish, contact the American institute of Aeronautics and Astronautics,
1290 Avenue of the Americas, New York, N.Y. 10019.
Cupyrlphl @ Amerlrsn L ~ P I I Iof
Y IAeromulle'md
~
AsImna~IIc&Inr.. 1978. Allrlpblirrurrrll.

TWO-DIMENSIONAL SEPARATED WAKE MODELING AND ITS USE TO PREDICT MAXIMUM SECTION LIFT COEFFICIENT

,v,

M. L. Henderson*
Low Speed Aerodynamics Rescarch Unit
Baeing Commercial Airplane Company
Seattle, Washington

Nomcnclature

Static prcssurc

Radius of curvature of wakc or jet

Vclocity vector

;1

~ u r f a c enormal unit vcctoi

surface tangent unit vector

A@ Surface correction snglc


p

Ucndty

to verify the results a t the Reynolds numbers where testing is


fcdsible. This improvcd capability allows the dcsigncr the frecdoni
to design to a spccific application rather than t o the testing conditions in the tunnel.
Full computational analysis of airfoil sections has been
limited by our inability to compute the cffects of separation o n
section forces and momenta. Figure I indicates the relative importancc of separation effccts. Figurc I(a) compares wind tunnel
data with a potential flow analysis of the airfoil gcometiy with
no attcinpt to account for the effccts of eithcr boundary layer
or scparation. A t low angles of attack where the boundary layer
is thin and there is little, if any, separation, potcntial flow analysis
of the surface alone is a fair approximation to the data; but as
the angle is increased, thc lift predicted is, of course, poor.
(81

RNALYSIS O f GEOMETRY ALONE

Abstract
__

. .

A technique for computing the lift of separating <multielement airfoils in incompressible flow is presented. The procedure
cmploys repeatcd application of a panel mcthod to solvc for the
separated wake displacenicnt surface using entirely inviscid boundary conditions. Rcsults arc prcscnted that compare computcd
prcssurc distributions with those measured in thc wind tunnel for
airfoils with one, two, and four elements w i t h separation on each
clcment. A method employing this technique is prescnted which
shows promise in predicting airfoil section lift through stall.

DATA

Introduction

I n most rcalistic applications, knowledge of maximum section cocfficient (Cl,,,ax) i s important, if not crucial, to airfoil
design. Since heretofore Clmax could not be predicted using
existing computer tools, heavy emphasis was placed on expensivc
wind tunnel tests, cvcn though the tests could not bc performed
at the proper Rcynolds number for flight. This has traditionally
Icd to risk in airfoil application and to considerable conscrvatisin
in design to reduce the risk. Our intention in this work has becn
to devclop the capability to compute airfoil lift and pitching
moment chariicteristics over a wide range of flow conditions and

ACKNOWLEDGEMENTS
The test cases in figurer 11-14 werc funded by NASA contract No.
NASI-14742, "Selemed Advanced Aerodynamic and Active Control Concepts Development", which commenced on August 12, 1917.
The author wishes 10 thank Drs. Forrester T. Johnson and Paul E. Rubbort
for their invaluable assistance in the use o f Dr. Johnson's powcrful panet
method teclmalogy to solve the boundary d u e problem of figure 3 .
* Specialist Engineer

&NGLE

16

24

OF ATTACK. DEGREES

Figure 1. Relative importance of Separation Effects

Rarcly will thc boundary laycrs be thin enough that potcntiill flow analysis of the bare geometry will be sufficicntly accu-

rate. Figurc I(b) displays the effect of modeling thc boundary


layer, but not scparation. A t low anglcs of attack, where there
arc no separations, thc agreement with the wind tunncl data is
good. When thc angle of attack is incrcased and separations begin
to OCCUT, the predicted and measured lift diverge. In figure I(c),
tlic boundary layer and separation have been modeled. Obviously,
separation cffccts must be accounted for to predict maximum
lift. In fact, most of the lift curve shows thc effects of separation so that ;accurate results were obtained, in this example, only
when both boundary layer and separation werc accountcd for.
This will bc the cdSe for any gradually scparating section, such as
the CAW-I used in thc example. It i s true that sharply stalling
sections haye little separation until catastrophic separation occurs
at Clmax. When analyzing such sections, accounting for the
boundary laycr alone i s accurate t o lifts very near stall. However,

combination would be an accuratc simulation of a separating


section and a potential flow analysis of it will produce the proper
pressure distribution. This concept, as applied to airfoil anal sis,
is not new. It w d S eniployed as early as 1940 by A. Walz to

2.2

:
u

1.61

compute tlic effect of the wake behind a split f l a p Walz, however,


determined the shapc of the displacement body by trial and error.

Ly

Y
Y

YI

1.4-

0
0

t
3

1.0-

.6

I/
I

-.
..
i_

this displacement body using B panel method. Such a tcchnique


is useful bccaiise tlie shape and position of the displacement
body, once the points of separation are known, may be determined by inviscid conditions which result directly from the
physics of the problem.

0
ANGLE

Thc basic problem is one of cornpotins, in general, tlic shape


of the scparation displacemcnt body. This is accomplislied by
iteratively solving a boundary value problem on the surface of

OF

18

24

ATTACK. DEGREES

The wake trailing tlie separating airfoil elenicnt is a free wake


and as such cannot support any force across its displacement
thickness. Neglecting gravity effects, this implies that the pressure difference across tlie wake displacement surface [nust balance
the acceleration duc t o the curvature of tlic wakc. This rcsults in
tlic requirement that:

Id ANALYSIS WITH BOUNDARY LAVER AND SEPARATED WAKE MODELED


Z.ZY

-2

ai'jao = - P V , , ~ ~ / R

60

where Vwake is the average ordered velocity parallel to the wake


centerline inside tlic wakc displacement thickncss, and q is a dis-

2 1.0

.6

placement perpendicular to tlic wakc centerlino. In incompressiblc flow, this results in the following velocity relationship:
0

16

-1
24

ANGLE OF ATTACK, DEGREES

Since the Vw,ke is small, tlie above cquationmay beapproximated by:

Figure 1. Concluded

aviaq

v'
sharply stalling sections are cxceptions in practical application
and any gcncral lcclinique for predicting airfoil lift and pitching
monxnt must account for thc possible effects of gradual scparation.

Further, it is well known that in two dimcnsional flow the


prcssurc is constant in thc wake over the separating airfoil.

In incompressible flow, tliesc conditions lead t o simple


velocity boundary conditions on the displacement body. Thcse
are listcd in table I and illustrated in figure 2.

Approach

Thc tcchniquc t o be discussed is based on the premise that

The conditions shown in table 1 arc not sufficient t o obtain


a u n i w c solution as t l w stand, because the value of the constant pressitre in the separated wakc has not bccn spccified. I n
fact, tlicre is a solution for cvcry prcssurc one might cliouse. The
pressure in this region iniist be an inherent part of the solution
to thc boundary value problem in order for the solution to be
gcneral.

the cffccts of a scparatcd wakc on a n airfoil prcssurc distribution


BTc known if t,lC
strca,nlinc ~lirplacemcnt caused by tl,e wake is

known. G~~~~ this


the
pretlicting lift af a n y
parts+
angle of attack ,nay bc conveniently scparatcd illto
dcter,ninatio,l of tlie separation poi,lt and deter,nination of the
streamline clisl,lacenient caused by a separated wake attached to
thc airfoil (and boundary laycr) a t the scparntion point.

We have found that if the pressure in the constant prcssure


region of t h c separated wake is set equal to the pmssurc a t the
airfoil trailing odgc and if tlie scparatcd wake displaccment surf~ce

Prediction of Airfoil Lift and Pitching Moment


Separation Point Known

Table 1. Boundary Conditions for an Airfoil Having Separated

In this section, the techniquc for computing the streamline


displacement resulting from separation will be discussed in tlctail.
I t will be assumed t h a t the separation points (i,e., the positions
on the airfoil surface where the flow dctaclis and the boundary
layer assumptions no longer hold) m e k n o w n

\./

Flow

Tlic primary artificc used in this technique is the wake


displaccmcnt body. If tlic displacement of the streamline downstream of tlic separation point on the upper surface and extcnding
beyond the trailing edge, is known, then a displacwnent body
wliose boundaries are defincd by this streamline and the strcamlinc entcnding from the trailing,edgc of the lowcr surface may bc
addcrl t o the forward unscparated portion of the airfoil. This

the separated element. It can be shown that a jet will produce


different streamlines aft of an airfoil than will a wake; and that
tlic difference will become important in regions of curved flow,
such as near the lcading edge of a wing or over trailing edge
flaps. Whilc a dctailed discussion of this effect is beyond the
scope of this papcr, the crux of the difficulty can be seen by
comparing the velocity gradient nomlal to the wake ccnter produced by a jet and a wake. Since
aV/aq = V,,,,ke orjet/R and V,,ke << Vjet, i t can be seen that
in a curveti flow field thc velocities across the displacement surface
of a wake and j c t ,nust be different, and thereforc, the shape of
the jet must necessarily be increasingly different from that of a
wake as tlie radius of curvature decreases and tlie jct velocity
incrcases. In this author's cnpcrience, the net effect of this diffeercnce is that thc jet simulation of the free wake is a sufficiently
inaccurate model to cause significant errors in the pressure distributions of airfoil elcmcnts ovcr which the wake flows. Since
prediction of the effect of t h c separated wakes on trailingeleinents
is neccssary t o properly analyrr multielcmcnt airfoil systems,
this type of solution was not employed.

TRAILING WAKE
SEPARATED WAKE

v.n-0

"."=a
V . T lower

j-V.T=CONST--

v.n-0

AIRFOIL SURFACE AND BOUNDARY LAYER


- n

"-"in
Figure 2. Boundary Conditions

is reqitircd to dcpait thc airfoil (plus boundary laycr) tangcntjally


a t the point of reparation, a unique, accurate modeling of the
separatcd wake results.

Tlic procedure for rcsolving thc shapc of thc displacement


body that m w t s all tliese boundary conditions simriltancously
is as follows: A p a n 4 method is used t o solve the boundary
value problem shown on figiirc 3 on ;in initial guess of the wake
gcometty. Thc dirplsccmcnt surface tangency requiremcnt is
approximated by the specification that the normal velocity bc
zero a t the front edge of the sepilratd wake. This rcsults in n
solution in which all boundary conditions are met, except the
streamline condition on the wakes. Since tlie initini guess at the
wake sbapc is not generally correct, the wake gemnctiy is nut a
Streamline a n d the normill vclocity on the wake surfaces will not
bc ?.era A local surfacc angle correction may be dciined using
tile local value of normal vclucity and tlic spccified tJngCntial
velocity as follows:

L/

Resuits With Separation Point Known


Figures 4 and 5 illustrate f h c pressure distribution and wake
shapes generated using the previously described technique to
analyze the CAW-I airfoil a t 1 8 O angle of attack. T h e starting
wake in this case was quite crudc; yct convergence was obtained
in three iterations. As the figure shows, the prcdicted lift is very
close IO the measurcd vilur. Figure 6 displays thc expcrimental
and calculatcd pressures obtained for thc same airfoil a t lzO,
16', and 20' angles of attack. The theoretical lift shown in
figure I is derived frvm these calculated ~)ressuredistributions.
THEORY

- GEOMETRY ALONE

THEORY
SEPARATED WAKE
MODELED 13 fTERATtONSI
-EXPERIMENTIREF
61
~

-I

-5

Y.n-0
,%T EDGE

Figure 3. Boundary Value Problem for Calculation o f Wake Shape


Figure 4. Comparison of Experimental and Theoretical Pressure
Distributions

Each panel represcnting the separated wake is then rotatcd


by this angle to force tlic wakc geometry to more closely approximate a streamlinc. The boundary v d u c problem is thcn solvcd
with the improvcd wake geometry and the cycle iepcated until
convergence is obtained. The function f(s) in figure 3 is used to
assure closure of the trailing wake and has little effcct on the s o b
tion. Wlicn i%stable wakcshape is found, the wake andunseparated
airfoil gcornctry (plus boundary layer displaccmcnf thickness)
. i? = 0 everywlicre). The
arc analyzed as a solid body (i.e.,
resulting pressure distribution is then integrated to obtain lift
and pitching moment.

SEPARATION POINT
STARTING WAKE

FlRST ITERATION

SECOND ITERATION

I:icob3, in his cxccllcnt work on the problem, took a diffcrent


approach in modeling tlw separated wake. To obtain a constant
pressurc region over the separated airfoil clement, Jacob employed
a tailored source distribution in the separated region to achieve
the rcquircd strcanilinc deflcction from the airfoil geometry. 'This
modeling results in a jet wake issuing from the trailing edge of

--_

-..

.
.
.
,

Figure 5. Wake Displacement Sody Development-GAW-1,

a = l8.29

GAW-l AIRFOIL
BEYMOLDS NUMBER = 6 x lo0

--- - - -.- :---=-

-5

c'

E0

-4

-3

Y
w
Y

0
0
w

agreement is still quite good. It is especially encouraging to note


that thc effect of the rather massive airfoil wake on the flap
pressure has been accurately calculated.

M h C H NUMBER = 0.15

E X P E R l M E N T ( R E F . 61

-2

3
Y)

-2.0

-1.5-

-l.O-

-3

tLL

.1

Le

-2.5

HEORV-SEPARATED WAKE MODELED

ANGLE OF ATrhCK = 8.48'


SPOILER ANGLE. *SP = 16'

ALCULATED

- SEPARATED WAKE,

NO BOUNDARY LAYER

-.6

GAW.1 AIRFOIL
REYNOLDS NUMBER = 6 x

toe

0.15

Y)

YLL

0 -

.6

THEORY-SEPARATED

PLUS BOUNDARY LAVER


EXPERIMENT

MACH N U M B E R

&

-4

1.o

WAKE MODELED

1.d

160

(I i

-3

Figure 7. Comparison o f Experimental and Calculated Pressure


Distributions on an Airfoil With Deflected Spoiler

-2
.l
0

HORD FRACTION, XIC


GAW-l AIRFOIL
REVNOLDSNUMBER=BI~O~

-8

E0
Y
Y

0
0

-6
-5

0 -6

..- _- -_____ - - ._
MACHNUMBER=O.lS

t
E

POTENTIAL FLOW

$! -5

ANALYSIS OF
SEPARATED FLOW

k
Y

EXPERIMENTIREF. 61

NASA OAW-l AIRFOIL


30% FOWLER FLAP
ANGLE OF ATTACK = oo
FLAP ANGLE = aOa

-4

EXPERIMENT

E -3
YI

LL
2

THEORY-SEPARATED WAKE MODELED

-2

= 20'

(I

-3

-1
I

c'

-7

-7k
&

-2

-1

CHORD FRACTION. X l C

0
1

Figure 8. Comparison o f Experimental and Calculated Pressure


Distributions on a Two-Element Airfoil With Separatinn
on One Surface

Figure 6. Comparison o f Experimental and CalculatedPressure


Distributions on a Partially Separated Single Element
Airfoil

N&SA GAW-1 A\IRFOll


30% FOWLER FLAP

Figure 7 shows the results of applying the technique to the


spoiler separation problem. The separation point is fixed at the
spoilcr trailing edge. It is' interesting t o note the effect o f the
modeling of the boundary layer displacement thickness in this
case. Thc boundary layer is thick at the spoiler hinge and thin
at its lip due to the favorable pressure gradients OVCI the spoiler.
This change in thickness, when accounted for in the modeling,
modifies thc effective spoiler deflection angle and has a significant
effect on the calculated pressure distribution, even though the
separation resion is quite large relative t o the boundary layer
thickness.

0"
c'

ANGLE OF ATTACU = 12.5'


FLAP ANGLE = 40

-'
.4

$
-z
Y
LL

Figures 8 and 9 show tlic results of applying the technique


t o separations on a t w o d e m e n t airfoil system. The configuration
is the GAW-I with 30% chord Fowler flap deflected to 40 degrees.
The wind tunnel data shown arc those of reference 5 . In figure 8,
there is separation on the flap only and even though n o boundary
layer has been rnodelcd in this calculation, the results are encouraging. In figure 9, both airfoil elements separate and thc

CHORD FRACTION, X / C

Figure 9. Comparison o f Experimental and Calculated Pressure


Distributions on a Two-Element Airfoil With Separation
on Both Surfaces
4

Results With Separation Point IJnknown

Prediction of Airfoil Lift and Pitching Moment


Scparation Point Unknown
Once the effect of a separation and its trailing wake on a
section prcssure distribution can be accounted for, a tilethod can
be devised to predict lift and pitching moment througll stall if
the onscf of separation can bc predicted. The simplest method
would be to compute the pressurc distribution bdSCd on tlie section geometry alone (c.e., no wake, no boundary layers modeled),
compute t h c boundary layer thickness and scparation points
from this pressure distribotion, and model thc separations as
discusscd previously to dctermine lift and pitching momcnt. The
problcm here i s t h a t the pressurc distribution uscd t o comixitc
t l x scparation points analysis of t h e bare gcoinctry is signifiT C
(wiiicli has
cantly different tliaii the final ~ ~ ~ S S U distribution
sepnrations and boundary layers modelcd). A bounil;iry laycr
analysis of the final distribution will not produce the Same boundary laycr thickness 3s previously computed, and more importantly,
may not produce the same separation point.

Obviously. sonic tcchnique for repcating t h c boundary


layer analysis a n d separation modeling scquencc tllat will scttlc
on a stable separation point is needcd. The tcclmiqur used for
single clement ;airfoil sections is:
I,

Figurc 10 displays the rcsults of applying tlic single clement


technique t o a wing with ii s m a l l leading edge flap. The leading
cdgc flap did not separate hecausc of the unloading cffect of the
wing scparations, so that the single clement tcchniqiie would bc
cxpected to apply. The comparison with tlie wind tunnel data is
good. The error in maximum lift is approximately 1% and the
theorctical pressure distributions show remarkable agreemcnt
w i t l i the experimcntal ones. Of particular i n t e m t is tlic shapc
of thc lift cwvc with boundary layer only. Evcn a t low angles of
attack, modeling of tlie boundary layer alone does not account
for all t h e difference between tlic data and the potential flow
analysis of thc geomctry alonc. Small, scemingly insignificant,
separations (2% to 5% from t h e trailing edge) can be shown to
nccoiint for thc remainder of the lift loss. This illustrates t h e
nccd for accounting for separation in the prediction of tlie cntire
had a n analysis modeling the boundary
lift ciirvc not just C;,,I
layer only bccn cmploycd in this C ~ S C ,predictions of lift would
have bccn optimistic at angles of attack as low as 8 dcgrees.

L/

In figures 11-14. the rcsiilts of applying the multielcmcnt


tcchnique t o a much morc complex airfoil system are compared
witli wind tunncl data. Again the agrecmcnt is reasonably good
with maximum lift prcrlicted about 2% high and the pressure
distributions in good a g r e m e n t . Thesc icsults arc particularly

Compute t i l e Ipotcatial flow pressrirc distribution O S tlie

baic geometry a t :tic anqlc of attack of intcrest.

2.

Compute the displaccmcnt thickncss and separation point


based on tlic p r ~ s s u r edistribution from step I

3.

Displnce tlie suifacc by the ilisp1;icen~cnt tlrickness 111) tu


tlie separation point a n d a d d tlie starting wakc aft O S tlie
rcpamtion point.

4.

I t c r a t c thc boundary value problem of figrirc 3 to rcliix tlic

scpaintcd wake shapc.

5.

Compiitc flit tlispliiccnienf thickncss and scpeiation point


boscd on tlic final prcssurc distribution from stcp 4.

6.

Cyclc steps 3 tu 5 until tlic scp;mtion point Iias canvcrged.

A modificd tccliniqiic )nust bc applicd whcn malyzing


iniilticlcment airfoil systems because of thc strong leading cdgctrailing edge coupling betwcen the elemcnts. Thc Imic problcni
is that thc l u a r l i n q cdgc pressure of n tmiling d ~ n i c n tcvntrois,
to .r large extent, t h e trailing edge prcssurc of the i n a t c l c n i c n t
forward. Thc boundary lnycr of t h c forward clciiiciit is very
scnsitive to tlie prcssrire gradicnt near its !miling edgc. su that a n
error in thr tniilin. edge prcssiirc caoscd by a n error iii iiic t r d i n g
clenicnt lcading urlge prcssurc will cause inaccurate prcdiction of
separation on tlie forward elcmcnt. If handled ineoriectly, the
separation point errors will incrcase for each wccccding forward
clement and will cause divcigence in the all solution. For
this reason, t l i e single clcmcnt technique is applied to thc trailing
edge elenient first a n d , having rclaxcd tlie separation point on i t ,
t h e bound;iiy laycr a n d sep;iration arc rnodelcd on tlic ncxt
clcment anti the wzkcr a n d scparations on both clcments arc
iclaxcil. This process is cuntinucd forward until all scpal-ations
ond boundary layers arc relaxed simoltaneously. I n tlic cases run
to datc, t h i s ~procedurclhas produccd well-convergcd solutions.

Thcsc techniques tend to produce steady forward IIIOVCnient of the separation point and so arc probably over-damped.
Iloweuer. this type of convergcncc has, in our expcricnce, lead to
a higli succcss rate in obt;iining rclaxed solutions, tlioiigir possibly
a t t h c cxpense of solution execution time.

XWING/CWING

Figure 10.

Comparison o f Experimental and Calculated Pressures


and Lift on an Airfoil, Plus Leading Edge Flap With
Separation on the Main Airfoil Surface

- THEORY - GEOMETRY ONLY


x THEORY - BOUNOARY LAVER MODELED

1s

ANGLE OF ATTACK.

Figure I ? .

24 0
o/

a 2 -0.4 -0.6
PITCHING M O M E N T
COEFFICIENT

Comparison of Theoretical and Experimental L i f t


and Pitching Moment Curves

i
,

__.,

Separation Point Prediction

cncouraging since, at stall, every element in the airfoil system was


separated to some degree. It is also interesting to notc that the
aft flap was partially separated at all angles of attack. Consequently, even at four degrees angle of attack, lift loss predicted
by modcling of the boundary layer alone was only about B third
of thc total lift difference between analysis of the base geometry
and thc data. As figure 1 I shows, modeling of the separation
accountcd for thc rcmainder of the difference.

In the examples of figures 10-14, a straight-forward momcnturn integral boundary layer approach was used t o solve for the
displacement thickness and for prediction of separation. While a
full discussion of the details of the particular techniquc used goes
beyond the scope of this paper, a summary of its thearctical
components is given below.
Theory
-

Component
E WITH SEPARATION

Figure 12.

211

SEPARATED
WAKES

&-->

&.NO BOUNDARY LAYER


WINO TUNNEL DATA

R N = Z 106
M.O.16

ANGLE OF &lTACK = 21.7 DEGREES

SEPARATED WAKE

WAKES

XIC

Figure 14.

Poulhausenl-Garadia

Laminar bubblc

Henderson (empirical)

Transition

Granvillcl

Turbulent boundary layer

Momentum integral', power law


velocity profilel, Garner's
equation for form parameter',
Ludwig-Tillman equation for
shear-stress at walll.

'Turbulent separation

H>3.0

In our experience, simple integral approaches, such as tlie one


outlined above, will do a creditable job of computing boundary
laycr characteristics up to separation, if the boundary layer
velocity profilc assumption (in this case a power law profile) is
not violated by the arrangement of wakes from previous airfoil
elements. That is, the boundary layers are nonconfluent. Further,
since the separated wake is computed using entirely invisid considerations, tlie boundary laycr computation is only required to
operate where similar velocity profile assumptions are valid. The
fact that the simple nonconfluent boundary layer technique
performed well for the configuration of figure 1 1 indicates that
thc boundary layers involvcd were conconfluent. Boundary layer
velocity profiles takcn during the wind tunnel test of this configuration show that the flap boundary laycrs are indeed nonconflucnt, but that the wing boundary layer is not due to the
influence of the leading edge flap wakc. While it is difficult t o
draw a positive conclusion from the few complex cases that have
the results shown here may
been run, (such as that o f figure I I),
indicate that boundary layer-wake confluence may not be as
strong an influence on separation as is generally believed. Furthermore, the problem of gap optimization for maximum lift may be
salved without resorting to the computation of the effects o f
strongly confluent boundary layers at all. Since the positions o f
the wakes from each element are known, the designer may easily
avoid any gaps that might produce strong confluence; experience
has shown that such gaps would be inferior t o those that d o not
produce confluence. The problem then becomes one o f optimizing
those gaps that do not produce strong confluence. This is not t o
say that the ability to compute thc effects o f confluence is not
desirable; only that in the interest o f reliability and economy,
simple boundary layer techniques may sewe well when used with a
realistic model of the inviscid effects of the boundary layers
and separations.

Comparison of Theoretical and Experimental Pressure


Distribution on Four-Element Airfoil, a = 16.60

ANALYSIS OF BARE GEOMETRY

Poulhauscnl

Laminar scparation

In practice, this boundary layer solution gives results, both in


computed displacement thickness and in separation point, which
arc w r y similar t o those of the Nash-Hicks turbulent analysis4.
Readers familiar with boundary layer computation techniques
will recognize the approach described above as rather basic, one
might even say primitive, especially when compared t o tlie multilevel integral solutions and finite difference solutions that h a w
been developed in recent years. How can the results in figures
10-14 then be explained?

Comparison of Theoretical and Experimental Pressure


Distribution on Four-Element Airfoil, a = 4.4"

.... ANALYSIS OF BARE GEOMETRY


-ANALYSIS
WITH SEPARATION A N 0 BOUNDARY LAYER

Figure 13.

Laminar boundary layer

Comparison of Theoretical and Experimental Pressure


Distributions, a = 21.7'

Refcrences

p&vp

I.

Schlichting, H.: Boundary Layer Theory. McGiaw-llill Book


Co., Inc.,pp. 238-251,391.406, 566.519.

2.

Riegels, F. W.: Airfoil Sections. Butterworth & Co. Publishers Ltd., 1961,pp. 226.

3.

Smith, A.M.O.: Remarks on the Fluid Mechanics of the Stall,


AGARD~LS-74,21974, pp. 2-1 to 2-33.

4.

H a h n , M.; Rubbert, P. E . ; Mahal, A.. Evoluotion ofSeporolio,z Criterion and Their Armlicolion to Seoaroted 1 7 1 0 ~
Anolysis Air Force Flight Dynamics Laboratory, AFFDLTR-72.145, 1973.

5.

Wcntz, N . H.: Scetharam, H . C.: Dcvelopment of Fowler


Flup System for n High Performance General Aviotion Airfoil. NASA CR-2443, 1974.

6.

McGhee, R . J.; Bcasley, W . D.:Low-Speed Aerodynamic


Clzarocleristics of a 17-Percenl-Thicl; Airfoil Section Designed f o r General Aviation Applications. NASA TN D-7428,
1973.

..

Das könnte Ihnen auch gefallen