..
NASA Contractor Report 2962
A Geometry Package for Generation of InputData for a ThreeDimensional
PotentialFlow Program
N. Douglas Halseyand John L. Hess
CONTRACT NAS114402 J U N E 1978
TECH LIBRARY KAFB, NU
NASA Contractor Report 2962
A GeometryPackage for Generation of Input Data for aThreeDimensional
PotentialFlow Program
N. Douglas Halsey and John L Hess . McDonneZZ DongZas Corporation Long Beach, California
Prepared for Langley Research Center under Contract NAS 1 14402
National Aeronautics and SpaceAdministration
Scientific and Technical Information Office
1978
.
. 1 .0 Table Contents .......................... of 2 .0 Index of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . 3.0 Sumnary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.0 Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.0 Nomenclature and Arrangement of Input Points . . . . . . . . . . . . 7.0 Paneling of Isolated Components . . . . . . . . . . . . . . . . . .
1 0 TABLE OF CONTENTS
iii
V
1
2
7
9
13 13 17
17 18
General Features of the Paneling Method . . . . . . . . . . . . Distribution of Points Along Nlines . . . . . . . . . . . . . . .............. 7.2.1 Distribution. Input Unaltered 7.2.2 Distribution. Input Augmented i n Number . . . . . . . . . 7.2.3 Constant Increments in Arc Length . . . . . . . . . . . . 7.2.4 Cosine Spacing . . . . . . . . . . . . . . . . . . . . . . 7.2.5 CurvatureDependent Distribution . . . . . . . . . . . . 7.2.6 UserSpecified Distribution ............... 7.3 Distribution of Nlines .................... 7.3.1 The PlanarSection Mode . . . . . . . . . . . . . . . . . 7.3.2 The ArcLength Mode . . . . . . . . . . . . . . . . . . . 7.1
7.2
20 20
22
24 24 25 31 32 32 32 33 34 35 37 42 43 47 47 48 51 54
8.0 Calculation of Intersection
................. 8.1 General Features of the Method . . . . . . . . . . . . . . . . . . 8 . 2 Restrictions and Limitations o f the Method . . . . . . . . . . . o f the Method of Solution ............... 8.3 Details 8.3.1 The I n i t i a l Search f o r "Line Segments and Surface Elements Containing Intersection Points .........
Curves
8.3.2
f o r "line Segments and Surface Elements ContainingIntersectionPoints and the Approximate Determination o f the Intersection Points . 8.3.3Derivation of a Mathematical Representation of the Surface of an Element .................. 8.3.4 Computation of More Precise Values o f the Coordinates of the Intersection Points ............... 8.3.5 Test Cases f o r the Intersection Method . . . . . . . . .
The Final Search
.
9.0
Final Repaneling of Components . 9.1 9.2 9.3 9.4
................... General Considerations . . . . . . . . . . . . . . . . . . . . . Intersecting Components . . . . . . . . . . . . . . . . . . . . Nonl ifting Intersected Components . . . . . . . . . . . . . . . L i f t i n g Intersected Components . . . . . . . . . . . . . . . . .
iii
I
.
10.0Conclusions 11 0 References
.
.............................. ..............................
61
62
iv
2.0
INDEX OF FIGURES
1. 2 .
3.
Typical element distribution for a trapezoidal wing Typical element distribution for a wingfuselage
.......... case . . . . . . . . .
3
3
4.
5.
6.
..................... 4 Definition of frequently used terms . . . . . . . . . . . . . . . . . . 9 Illustration of numerical differentiation procedure . . . . . . . . . . 15 Comparison of curvefit methods . . . . . . . . . . . . . . . . . . . . 16
Method of distributing points on an Nline  input distribution, . . . . . augmented Point distribution on a supercritical wing section . . augmented
Surface elements used by NLR, Amsterdam, for an external store configuration (1780 elements)
7.
8. 9. 10.
. . . . . . . . . . . . . . . . . . . . . . . . . . 19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 .. . . ... . . . . . . . . . . 20 Nline  cosine distribution . . . 21 . . . . . . . . . 22 . . . . . . . 24 . . . . . . . . 24 . . . . 28
 constant  input
distribution,
Point distribution on a cylindrical fuselage section . . . ... increments in arclength Method o f distributing points on an
11.
Point distribution on a supercritical wing section  cosine . . . ... . . . distribution
.... ..... . . . .
12. 13.
14.
Point distribution on a supercritical wing section  curvature.... . . . dependent distribution . . . . .
. . . ..
Point distribution on a section of a nonlifting component curvaturedependent distribution .. ...... .
Redistribution o f elements on a trapezoidal wing (cosine spacing chordwise, constant increments spanwise, planarsection mode) .
15.
Redistribution of elements on a supercritical wing (cosine spacing chordwise , constant increments spanwi se, planarsection mode , mu1 tisection component option) . . .
. . . . . . . . . . . . . . . . . 29 ... . . . . . . . . 30
16.
Redistribution of elements on a fuselage (constant increments around circumference and in axial direction, planarsection mode, multi... .. . . . . secti oncomponent option)
...
17.
18.
Comparison of planarsection and arclength modes of distributing N1 ines  strut on a thick wing . . . . Intersection method test case  fntersection of two circular cylinders . . . .
V
. . . . . . . . . .’. . . . . . 31
. . . . . . . . . . . . . . . . . . . . . . . . . . . 44
19.
20.
Intersection method test case intersection of two spheres . Intersection method test case  intersection of two
. . . . . 45 ellipsoids . . . . 45
21.
22.
Illustration of intersection method  thick wing intersecting a circular cylinder. . . . . . . . . . . . . . . . . . . . . . . . . . Final repaneling o f intersecting components wingfuselage case
..
46
. . . 49
50
23.
Final repaneling of intersecting components  wingfuselagetiptank case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24. Final repaneling of nonlifting intersection components wingfuselage case (zeroincidencewing, Nlines through every point on intersection curve) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55
25.
Final repaneling of nonlifting intersected components  wingfuselage case (wing with10" incidence, Nlines through every point on inter55 section curve) . . . . . . . . . . . . . . . . . . . . . . . . . . . . Final repaneling of nonlifting intersected components wingfuselage case (zeroincidence wing, Nlines through every other point on intersectioncurve) ......................... Final repaneling of lifting intersected components wingpylon case. Final repaneling of lifting intersected components wingpylon case (sparse element distribution on pylon) . . . . . . . . . . . . . . . Final repaneling of lifting intersected components  wingpylon case wing) . . . . . . . (more points on pylon than surrounding region on Final element distribution on a wingfuselagepylon case
26.
56
27. 28.
. 57
. 58
. 59
29. 30.
. . . . . . . 60
vi
3.0
The preparationofgeometricdata
SUMMARY
f o r i n p u t t o threedimensionalpotentialflow programs i s a very tedious, timeconsuming (and therefore expensive) task. T h i s report describes a geometry package t h a t automates and simplifies this task to a large degree. Inputto the computer program f o r the geometry package consists of a very sparse set of coordinate data, often w i t h an orderof magnitude fewer pointsthan required f o r the actualpotentialflowcalculations. Isolated components, such as wings, fuselages, etc. are paneled automatically, using oneof several possible element distribution algori thms . Curves of intersection between components are calculated, u s i n g a hybrid curvefit/rurfacefit approach. Finally,intersecting components arerepaneled so thatadjacent elements on either side of the intersection curves line up i n a satisfactory manner forthepotentialflowcalculations. The geometry package has been incorporated into the NASA Langley version of the 3D l i f t i n g potentialflow program and i t i s possible to r u n many casescompletely (from i n p u t , through the geometry package, and through the flow calculations) without interruption. Use of this geometry package can significantly reduce the time and expense involved i n making threedimensionalpotentialflowcalculations.
4.0 W i t h theadvent
INTRODUCTION
of modern highspeed computers,theaerodynamicist gained the ability t o study the effects of configuration changes i n great detail. Whereas before he m i g h t have been able t o estimatetheeffects of changes i n such grossgeometricparameters as sweep and aspect r a t i o on the t o t a l l i f t of an isolated wing, he now can accurately calculate the effects of changes of quite small details of the shape of very complex configurations (wingfuselagenacellepyloncases, for example) on n o t only l i f t b u t also on spanwise and chordwise load distributions, local pressures, flow angles,etc. This g a i n i n computational a b i l i t y has made thepreparation of i n p u t d a t a for describingconfiguration geometry a very tedious, timeconsuming task. General threedimensional potentialflow programs, such asthe Hess program, (references 1 and 2 ) requirethe geometry t o be i n p u t a s thecoordinates of sets of points, which are grouped t o form quadrilateral surface elements. Many more elements are generally required t o o b t a i n accurate flow solutions than the number required for adequate pictorialrepresentation. For example, thesimpletrapezoidal wing of figure 1 , which can be represented f a i r l y accurately by twenty or so elements,requires on theorder of two hundred elements t o obtain a reasonable flow solution. More complex configurations, such asthewingfuselagecase of figure 2 , m i g h t use nearly a thousand. A case which was recently run by personnel of NLR, Amsterdam, using 1780 elements t o represent an externalstore configuration is shown i n figure 3 (reference 3 ) .
There i s a clear need t o automateas much as possible the preparation of the i n p u t coordinates, both i n order t o reduce the number o f pointsinput t o the programs and inorder t o relax some o f the restrictions on how thepoints must be distributed. The main d i f f i c u l t y i n doing t h i s i s t h a t thereare so many l o g i c a l l y differentcases t o consider. For example, a wing has different spacingrequirements t h a n a fuselage; a fuselage has differentrequirements t h a n a nacelle,or a pylon, etc. Moreover, bodies o f similar type may require different numbers and distributions of points due t o theproximity of other bodies. I f there are intersections between bodies,additionalrequirements are imposed on the p o i n t distributions. A geometry package must be very
2
Figure 1 .
Typical element distribution for a trapezoidal wing.
Figure 2.
Typical element
distributionfor
a wingfuselage case.
3
,Figure 3.
Surfaceelements used by NLR, Amsterdam, f o r an e x t e r n a ls t o r e c o n f i g u r a t i o n (1780 elements). so many d i f f e r e n t cases, b u t i t should not be so t o use. f o r use w i t h t h e method
very flexible to apply to
f l e x i b l e t h a t i t becomes cumbersome This report describes
a geometrypackagedeveloped
p o t e n t i a l  f l o w program o f r e f e r e n c e s 1 and 2 and f o r e x t e n s i o n s o f t h e
which may replace i t i nt h ef u t u r e .S i n c et h er e q u i r e m e n t s o f f u t u r ep r o grams a r e m o s t l y s p e c u l a t i v e a t t h e p r e s e n t , c o m p a t i b i l i t y w i t h t h e p r e s e n t program i s emphasized. The geometry package has been of t h ep o t e n t i a l  f l o w program. incorporatedintothe NASALangley version
This program accepts
inputeitherinthe
4 (withminormodifications)or i nt h ef o r m a to ft h e programdescribed i n reference 5. With a small amount ofadditionalinputtocontrolthe geometry package, i i s p o s s i b l e t o r u n t a number o f f r e q u e n t l y o c c u r r i n g C ~ J ~ S completely without human i n t e r v e n t i o n .
o r i g i n a li n p u tf o r m a td e s c r i b e di nr e f e r e n c e For example, i s o l a t e d wings or bodies may be i n p u t t o t h e program w i t h a i l minimumnumber o f p o i n t s and t h e geometry package w augment and r e d i s t r i b u t e thepointstothe number and d i s t r i b u t i o n s p e c i f i e d bytheuser. These coordinates can then be punched on cards f o r use w i t h a u t o m a t i c p l o t t i n g programs t o i n s p e c t t h e r e s u l t s b e f o r e p r o c e e d i n g , o r t h e p o t e n t i a l  f l o w
4
_ I
program can proceed immediately to analyze the flow. Options are provided to allow theuser to tailor the distributions to needs of his particular the problem, so it should also be possible to run cases having multiple nonintersecting wings orbodies with no human intervention. More difficult cases, involving intersecting components, also be treated by the geometry can as package. Some of the simpler cases involving only two components, such a wing/fuselage case, can also run completely without intervention, more be but complex cases almost certainly should be checked before proceeding with the flow calculation. Some cases cannot be completely handled by the geometry package, but in these cases, use of the geometry package to augment point the distributions and to calculate intersection curves still results in a significant reduction in the effort required to prepare the coordinate data. A n outline of the major features of the geometry package is given below.
o
Significant reduction in effort required to input a case to the potential flow program. Two separate modes of program input available. Complete compatibility with the potentialflow program allowing many cases to be run completely without interruption between the geometry package calculations and the potentialflow calculations, Paneling of isolated components o Allows very sparse input coordinate data. o Provides output coordinate data suitable for potentialflow analysis. o Uses independent cubic curvefits for interpolation in two directions on surfaces (Nlines and "lines). o Provides six options for the point distributions on Nlines and four options on "lines. o Allows all Nlines on a component to lie in parallel planes, if desired. Calculation of intersection curves o Requires a distinction between intersecting and intersected components.
5
o
o
o
o
o o
o
Intersecting components represented by their "lines only. Intersected components represented by threedimensional surface fits. Intersection curves defined by arrays of intersection points between the "lines of the intersecting components and the fitted surfaces of theintersected components.
o
Repaneling of intersecting and intersected components o Intersection curve made an Nline of the intersecting component. d Other Nlines on intersecting component shifted to restore a smooth distribution. o Extra strip inside intersected component automatically generated, if desired. o Intersected components repaneled to insure that elements on adjacent components line up along the intersection curves. o Simpler repaneling options also provided.
The remainder of this report documentsthe theory and operation of the geometry package. Section 6 defines the geometric terms used i n the later sections and discusses the basic philosophy o f the geometric input to the method. Section 7 describes the paneling of isolated, nonintersecting components, including the options available, the applicability and limitations of the options, user requirements, methods used, and sample results. Section 8 covers themethod o f calculating curves o f intersection between components, including the theory, restrictions, and verification cases. Finally, section 9 describes the methods that have been provided for repaneling components after having calculated the curves ofintersection between them.
6
5.0
A A,B,C,D
A,B,C,D, EsFsGsH,
SYMBOLS
M a t rs x r f a c e  f i o e f f i c i e nan g e b r a i c ou if ct it s l C o e f f i c i e n t so ft h ee q u a t i o no f A l g e b r a i cs u r f a c e  f i tc o e f f i c i e n t s .
~
form.
a plane.
I,J,K,L,
. .
M,N,O,P
C
Local value of the chord of
acomponent.
d
F
f
G
S t r a i g h t  l i n e d i s t a n c e between adjacent points onacurve. Also used f o r t h e d i s t a n c e f r o m a p o i n t t o a plane. Functions used t o express a c u b i c c u r v e i n terms o f t h e f u n c t i o n a l values and f i r s t d e r i v a t i v e s a t t h e ends o f t h e c u r v e o n l y . Dummy v a r i a b l e used t o expressthegeneralform w i t hs e v e r a ld i f f e r e n tv a r i a b l e s . Matrixofsurfacefitcoefficientsingeometric General s u b s c r i p t used i n a v a r i e t y o f ways. Parameter used i n s p e c i f y i n g t h e d i s t r i b u t i o n o f N  l i n e s component. Curvature a t a p o i n t ona curve. a l i n e o r curve. ona o f a f u n c t i o n used
form.
i
K
k
R
Subscript denoting variables associated with
M
M a t r i x of constants used i n c o n v e r t i n g s u r f a c e  f i t c o e f f i c i e n t s fromgeometric t o a l g e b r a i c form. Curve onacomponent, generally running spanwise on l i f t i n g componentsand i n t h e a x i a l d i r e c t i o n on n o n l i f t i n g components. T o t a l number o f d e f i n i n g p o i n t s on a curve.
"line
N N1 i ne
Curve on acomponent, generally running chordwise on l i f t i n g componentsand i n t h e c i r c u m f e r e n t i a l d i r e c t i o n on n o n l i f t i n g components. Totalarclengthof a curve.
P
Normalized point number o f the point having index Arc l e n g t h a l o n g a curve.
i.
7
so’sl
S
Arc lengths at beginning and end of acurve. Subscript denoting a surface. Superscript denoting the transposeof amatrix. Parameters used in calculating surface fits. Coordinates of a point in a Cartesian coordinate system. Also subscripts referring to these coordinates.
T
8
Angle around a circle circumscribed about an airfoil section, used in determining the cosine point spacing distribution. Angle of a curve at the defining point having index i. Angle of a straightline segment of a curve beginning at the defining point having index i.
a
I
6.0
NOMENCLATURE ARRANGEMENT AND
OF I N P U T O I N T S P
Before describing the
geometry package i t s e l f , i i s necessary t o t
discussthegeneral scheme f o r i n p u t t i n g p o i n t s and f o r o r d e r i n g t h e elements t h a t t h e y f o r m and t o d e f i n e some o f t h e termswhich a r e used f r e q u e n t l y throughouttheremainder ofthereport. Mostprocedures and d e f i n i t i o n s a r e identicaltothosedescribed
i n references 1 and 2, b u t some ( f o r example,
t h ed e f i n i t i o no f" i g n o r e de l e m e n t s " ) have been changed s l i g h t l y . Reference should be made t o f i g u r e 4 t o c l a r i f y t h e d i s c u s s i o n which follows.
S T R I P ON N  L I N E S ON
MLINES ON INTERSECTED
ON
ELEMENT INTERSECTED INTERSECTING COMPONENT INTERSECTING
CQYPONENT
V

Figure 4.
D e f i n i t i o no ff r e q u e n t l y
used terms.
9
A complete configuration (such as a wingfuselagenacellepylon case)
is assumed to be constructed of a number of components, each of which is a set of associated points. Normally a single component is used to represent a complete body (such as a wing or a fuselage), but any number of components per body is allowed. There are twotypes of components  nonlifting and lifting. Nonlifting components, such as fuselages or other bluntended bodies, are represented by source distributions over their surfaces and hence have no circulation. Lifting components, such as wings or other bodies with sharp trai 1 ing edges , are represented by both surface sourceand dipole distributions. Circulation about any section of a lifting component is adjusted in such a way as to satisfy the trailingedge Kutta condition. Lifting components also have associated dipole sheets which represent trailingvortex wakes. Points on the wakes must be input to the program, as well as points on the bodies, and are considered to belong to the same components as the associated body points. Each component consists of a set of points which can be connected in such a wayas to form a network of intersecting lines called Nlines and "lines. Nlines on lifting components are the lines running approximately in the chordwise direction. They divide a wing, for example, into a number of distinct sections. "lines are the lines connecting corresponding points on the Nlines. On nonl ifting components Nlines are generally those surrounding the major axis of the body (if it is possible to define an axis at all). In principle, the roles of the Nlines and "lines on nonlifting components can be reversed without adversely affectingthe execution o f the program, but in most portions of the geometry package, it is assumed that the Nlines play the role described above. The potentialflow program does not require that all Nlines on nonlifting components have the same number of points. However since having variable numbers of points per Nline makes it impossible to define "lines, most applications o f the geometry package require each Nline to have the same number of points as each other Nline and each "line to have the same number of points as each other "line. Nlines may not cross other Nlines (though they may touch at apoint); "lines also may not cross other Nlines (though they may also touch at apoint).
10
Points mustbe
ordered so t h a t a l l p o i n t s on t h e f i r s t N  l i n e ( i n c l . y d i n g
wake p o i n t s a f t e r t h e body p o i n t s f o r l i f t i n g components) a r e i n p u t consecutively,followed by a l l p o i n t s on t h e second N  l i n e and so on. The f i r s t N  l i n e i n p u t may be t h e one a t e i t h e r extreme o f t h e component, b u t t h e choice determines the order of input of points along the Nlines. In general, the order must be such t h a t t h e n e g a t i v e o f t h e c r o s s p r o d u c t o f t h e v e c t o r from one p o i n t on an N  l i n e t o t h e n e x t p o i n t on t h e N  l i n e w i t h t h e v e c t o r f r o m a p o i n t on the Nline to the corresponding point on t h e n e x t N  l i n e r e s u l t s i n a vector which is directed to the exterior of the component. Thisrequirement may be s a t i s f i e d on a wing, f o r example,by o r d e r i n g t h e N  l i n e s f r o m t i p t o r o o t and o r d e r i n g p o i n t s on the Nlines f r o m t h e t r a i l i n g edge along the lower surface to the leading t r a i l i n g edge. edgeandback along the upper surface to the i s s a t i s f i e d by arrangingthe On a fuselagetherequirement
Nlines from f r o n t t o backand t h e p o i n t s on each Nline increasing counterc l o c k w i s e( l o o k i n ga f t ) . The requirement i s a l s o s a t i s f i e d by reversingboth theorderoftheNlines and t h e o r d e r o f t h e p o i n t s p a r t s o f t h e geometry package r e q u i r e t h a t N  l i n e s front of the body, o r more generally, that Nlines f i r s t "line s t a r t a t the end f a r t h e s t f r o m t h e i n t e r s e c t t h e component. The area between adjacent Nlines on a component i s designated a s t r i p . Each s t r i p on a l i f t i n g component has one c h a r a c t e r i s t i c v a l u e o f t h e d i p o l e d i s t r i b u t i o n and one l o c a t i o n where theKuttacondition issatisfied. On l i f t i n g components, i i s p o s s i b l e t o s p e c i f y t h a t t distribution but no s o u r c e d i s t r i b u t i o n s and no a s t r i p havea dipole boundary c o n d i t i o n s on any on theNlines. However on fuselages s t a r t a t t h e on n o n l i f t i n g components of any o t h e r components which
o f i t s elements. Such a s t r i p , c a l l e d an e x t r a s t r i p , i s u s e f u l f o r a v o i d i n g theabruptendingof a dipole sheet (which would result i n a concentrated vortex) along the curve of intersection
o f two componentsand
for controlling
the behavior of the dipole sheet near wingtips v o r t i c i t y o p t i o n i s used. R e f e rt or e f e r e n c e
when the piecewise linear 1 f o r more d e t a i l s .
The a r e a ( g e n e r a l l y q u a d r i l a t e r a l ) between adjacent "lines on a s t r i p i s termed an element. Each element has one c o n t r o l p o i n t where t h e boundary conditions are satisfied and one characteristic value of the source
11
distribution. It is possible to designate some elementsto be ignored elemen'ts. These elements do not have source distributions and no boundary conditions are applied tothem. If the component is lifting, ignored elements do have dipole distributions, however. References 1 and 2 allow for ignored elements to be defined only for lifting components, this restriction has been but lifted in the present program.
7.0
PANELING OF ISOLATED COMPONENTS o ft h eP a n e l i n g Method package i s t o panel
7.1
General Features
The f i r s t o p e r a t i o n performedbythegeometry
(i.e. , d i s t r i b u t e t h e elements) the components as isolated bodies, whether or n o t any o f them i n t e r s e c t . A l t h o u g h t h e r e s u l t i n g e l e m e n t d i s t r i b u t i o n s on i n t e r s e c t i n g components may n o t be useful for analyzing complete configurations, they serve asa starting point for determining the final distributions as w e l l as a l l o w i n g a c o n f i g u r a t i o n " b u i l d  u p " t o be performed(i.e.¶the s u c c e s s i v e a d d i t i o n o f components can be performed, i n order to determine i n t e r f e r e n c ee f f e c t s ) .
I i s assumed t h a t each t component
i s completely o f more than
independent o f a l l o t h e r s .
Since a s i n g l e body may be composed
onecomponent,however,and since the close proximity of another body may modifytheelementspacingrequirements o f acomponent, i i s n o t always t r u e t t h a ti n d i v i d u a l components a r ec o m p l e t e l yi n d e p e n d e n t .S u f f i c i e n tf l e x i b i l i t y hasbeen designed i n t o t h e p a n e l i n g method t o a l l o w p o i n t s t o bematchedwhere componentsmeet(making their "lines continuous) and t o a l l o w t h e u s e r t o s p e c i f y whatever d i s t r i b u t i o n he deems p r o x i m i t y o f o t h e r components. The panelingof an i n d i v i d u a l component i s accomplished i n twosteps. F i r s t t h e p o i n t s on t h e i n i t i a l N  l i n e s a r e augmented i n number and r e d i s t r i butedaccording t o t h e number and the spacing algorithm specified by the user. For this calculation, the Nlines must be roughly chordwise on l i f t i n g components i s required, componentsand, ift h e l a t e r r e p a n e l i n g o f i n t e r s e c t e d the Nlines must be roughly circumferential about an a x i s on n o n l i f t i n g components. Pointsinthe wake o f an Nline o f a l i f t i n g component are body p o i n t s .P r o v i s i o n has been made t o d i s t r i b u t e di n d e p e n d e n t l yo ft h e appropriate to account for the
allow the user to input different numbers o f p o i n t s on t h e i n i t i a l N  l i n e s of a component. The process o f r e d i s t r i b u t i n g p o i n t s a l o n g t h e N  l i n e s makes t h e numbers o f p o i n t s equal , t h u s a l l o w i n g M  l i n e s t o be formed. The second step i s t o augment and r e d i s t r i b u t e t h e N  l i n e s a c c o r d i n g t o t h e number and a l g o r i t h ms p e c i f i e d by theuser.This i s done byaugmenting and r e d i s t r i b u t i n g points along each "line. 13
The process o f r e d i s t r i b u t i n g p o i n t s a l o n g e i t h e r N  l i n e s o r " l i n e s requires a method o fi n t e r p o l a t i n ga l o n gg e n e r a lc u r v e si n these curves (particularly the Nlines) are not usually monotonic x,y, or
Z
space.
Since
i n either
the independent between a point
coordinates, some otherparametermust
be usedas
variableofinterpolation.
I n thepresent method, theparameter chosen i s
thearclengthalongthepolygonformedbyconnectingstraightlines adjacentpoints. I n t h e remainder o ft h i sr e p o r tt h et e r m" a r cl e n g t h " a l w a y sr e f e r st ot h i ss t r a i g h t  l i n ea p p r o x i m a t i o n .I n t e r p o l a t i o nf o r on a cuPve r e q u i r e s t h r e e s e p a r a t e i n t e r p o l a t i o n s , Each s e p a r a t ei n t e r p o l a t i o n numerical differentiation of the
one f o r each coordinate,
i s accomplished i n twosteps.
In thefirst
step,
dependent v a r i a b l e w i t h r e s p e c t t o t h e a r c I n t h e second step,the
length i s performed a t each definingpointonthecurve.
values o f t h e f u n c t i o n and i t s d e r i v a t i v e s a t t h e ends of the segments o f t h e curve are used t o d e r i v e t h e c o e f f i c i e n t s o f c u b i c i n t e r p o l a t i n g p o l y n o m i a l s . The n u m e r i c a l d i f f e r e n t i a t i o n i s done using a "weightedangle"approach
(emi) of s t r a i g h t  l i n e a p p r o x i m a t i o n s t o segments o f t h e c u r v e a r e f i r s t c a l c u l a t e d . The angle a t t h e m i d p o i n t o f each segment i s assumed t o equaltheangle of t h es t r a i g h t  l i n ea p p r o x i m a t i o nt ot h e segment. The angle a t any o f t h e g i v e n p o i n t s on thecurve ( e i ) i s thendeterminedbytaking an average o f t h e angles o f a d j a c e n t segments, weighted by the distances t o t h e m i d p o i n t s o f the segments,
(see f i g u r e 5). I nt h i s approach, t h ee n g t h sd i ) l ( and angles (7.1 .l) Angles o f t h e f i r s t and l a s t p o i n t s on a curve are determined by extrapolation, between t h e f i r s t
assuming constantcurvature calculated angles.
two a n d ' l a s t two p o i n t s .
The
derivativesatthegivenpointsarethen
found by taking the tangents of the
In order to facilitate the treatment of bounded by s t r a i g h t l i n e s , o r a fuselagewith a c y l i n d r i c a lm i d s e c t i o n ,
components having sections a k i n k i n i t s t r a i l i n g edge a m u l t i s e c t i o n component o p t i o n
such as a wing with
i s provided i nt h en u m e r i c a ld i f f e r e n t i a t i o np r o c e d u r e .W i t ht h i so p t i o n ,
14
S Figure 5. Illustration of numerical differentiation procedure.
thecurve i s divided i n t o severalsections. A t the ends of thesections, extrapolations are used t o determine the derivatives, just as they are a t the ends of thecompletecurve.Sectionsconsisting of j u s t two pointsare represented by straight lines. The slopes of thesestraightlines then determine the derivatives a t the ends of adjacent sections. The values o f the function and its derivative a t each end of a curve segment constitute four pieces of information which can be used t o determine thecoefficients of a cubic polynomial approximating thecurve. The form of the polynomial is
(7.1.2)
where S is the independent variable length (arc ranging from values So t o SI, f representsthe dependent variable (x, y, o r z coordinates), and primes denote differentiation. The form of i t s derivative is
f'
(5) =
f ' ( S o ) + (S
 So)
f"(S,)
f
;(s
SO)* f ' I ' ( S O )
(7.1.3)
15
Given the values of f and f ' a t So and SI, the simultaneous solution of f"(So) and f " ' (So). of equations (7.1.2) and (7.1.3)yields the values Equation (7.1.2) can then be used t o determine the value of thefunctionf(S) a t any valueof S w i t h i n the given curve segment.
T h i s curvefit method does not insure continuity of the second derivative of the function and thus i s not a cubic spline f i t i n the usual sense (reference 6 ) . I t was chosen rather than a spline method because of 'its consistentlysuperiorresults i n severaltestcases. For example, figure 6
shows a comparison between t h i s method and a truecubicspline method. The interpolated coordinates found by thepresent method areconsiderably method. Both methods less wavy t h a n thosecalculatedusingthespline fail to represent the shape accurately i n the after region of the body, and theproximityofthe ends ofthe because of thehigherlocalcurvature curve. Other comparisons involving a i r f o i l s and more generalshapesalso showed smoother results for the present method thanthespline method, i n
+
e
INFL'T POINTS INTERPOLATED POINTS (PRESENT METHOD) INTERPOLATED(SPLINE POINTS METHOD)
Figure 6.
16
Comparison of curvefit methods.
i n reference 6 t h a t , i n general , a s p l i n e produces t h e smoothest o f a l l p o s s i b l e c u r v e f i t s . The c o n t r a d i c t i o n o f t h i s p r o o f may p o s s i b l y be. due t o t h e unusual character of the independent variable (t i being defined as a quantity which varies smoothly along a nonsmooth curve) o r i t may be due t o t h e p r e s e n t a t i o n of t h e r e s u l t s b y a graph o f one
spite of the alleged proof interpolated result (the zcoordinates) results(thexcoordinates). as a f u n c t i o n o f o t h e r i n t e r p o l a t e d
7.2
D i s t r i b u t i o no fP o i n t sA l o n g
NLines
Each o f t h e o p t i o n s f o r d i s t r i b u t i n g p o i n t s a l o n g N  l i n e s ( e x c e p t t h e trivial option of leaving the initial distribution unchanged) r e q u i r e s an comarray of normalized arc lengths which applies to every Nline of the
ponentunderconsideration. The formulas f o r t h e s ea r c  l e n g t hd i s t r i b u t i o n s aregiven below. Given t h es p e c i f i e dd i s t r i b u t i o n s ,t h e method c a l c u l a t e s t h e d i s t r i b u t i o n oneach i n i t i a l N  l i n e and i n t e r p o l a t e s each coordinate independently t o determinethevalues a tt h ed e s i r e dl o c a t i o n s .I n some cases a l l p o i n t s onan Nlinecoincide(asatthe end o f a pointed body, f o r example). Then t h e method does notattempt t o i n t e r p o l a t e , b u t s i m p l y p r o v i d e st h es p e c i f i e d number o f p o i n t s t o t h a t N  l i n e . The f o l l o w i n go p t i o n s are available for distributing points 1.
2.
3.
on Nlines:
I n p u ti s t r i b u t i o n ,n a l t e r e d d u I n p ud i s t r i b u t i o n , t Constant increments Constant increments augmented i n number i n r ce n g t h a l on the superscribed circle (cosine spacing)
4. 5.
Curvaturedependent d i s t r i b u t i o n U s e r  s p e c i f i e ds t r i b u t i o n di InpuDistributionUnaltered t ,
6. 7.2.1
With t h i s o p t i o n t h e method does no i n t e r p o l a t i o n . I should be used t whenever t h e i n i t i a l d i s t r i b u t i o n a l r e a d y c o n t a i n s a s u f f i c i e n t number o f f component underconsideration p r o p e r l y spaced p o i n t s . I t h e N  l i n e s o f t h e a r e t o be r e d i s t r i b u t e d , o r i t h e component i s involved i n an i n t e r s e c t i o n w i t h f another component, then the number of p o i n t s i n p u t mustbe t h e same on each I n a d d i t i o n ,t h ed i s t r i b u t i o no fp o i n t s on adjacent component o f t h e N  l i n e . 17
Nlines should
be f a i r l y s i m i l a r so t h a t when corresponding points on Nlines make smooth curves.This istheonly.
areconnected,theresulting"lines distributionsofpoints
option which may r e s u l t i n d i f f e r e n t numbers o f p o i n t s and d i f f e r e n t among theNlines
of a component.
Although no necessary t o d e t e r 
interpolationsarerequired,
some c a l c u l a t i o n s a r e s t i l l
mine the derivatives with respect to arc length of the coordinates at each p o i n t on each Nline. These a r e needed i n l a t e r c a l c u l a t i o n s t o d e t e r m i n e intersection curves. 7.2.2 I n p u tD i s t r i b u t i o n , Thisoptionresultsin distribution,butcontains Augmented i n Number a distribution which is similar to the initial a d i f f e r e n t number o f p o i n t s . The i n i t i a l d i s t r i on case defining
b u t i o n used i n t h e c a l c u l a t i o n i s t h e d i s t r i b u t i o n f o r t h e f i r s t N  l i n e t h e component, unless a l l p o i n t s c o i n c i d e a "normalized point number", pi on t h e f i r s t N  l i n e . I n t h a t The method worksby one, t h e d i s t r i b u t i o n on the second N  l i n e i s used.
, ranging from zero to
pi = (i l ) / ( N

 1)
(7.2.1)
where
N i s t h e t o t a l number of points. Arrays o f normalized point number a r e formed f o r b o t h i n p u t and o u t p u t d i s and tributions. The o u t p u ta r c  l e n g t hd i s t r i b u t i o ni sd e t e r m i n e db yi n t e r p o l a t i n g number t o t h e the curve of input arc length versus input normalized point
i i s t h ei n d e xo ft h ep o i n t
outputvaluesofthenormalizedpoint number. Thisprocedure i s i l l u s t r a t e d i n f i g u r e 7. S u i t a b l e cases f o r t h i s o p t i o n i n c l u d e b o t h l i f t i n g and nonl i f t i n g t y p e components. without having to determine shown i n f i g u r e 8.
I is especially useful for precisely control t
1i n g
the desired distribution without having to load i n advancewhat arclengthsare.Typicalresultsfor
a l a r g e number o f p o i n t s and the numerical values of the
a s e c t i o n o f a s u p e r c r i t i c a lw i n ga r e
18
.+
+
+
.
+
.+
. + .
+’
+
+
0.0
0.0
. +’
0.2 0.4 NORMALIZED POINT NUMBER
0.6
0.8
 [(i 1 ) / ( N 
 1)]
Figure 7 .
Method of distributing points on an Nline  input distribution augmented.
+
INPUT (N POINTS
= 15)
o OUTPUT POINTS (N
= 25)
Figure 8 .
Point distribution on a supercritical wing section tion, augmented.
+
INPUT *WTE:POINTS POINTS ALL ARE LOCATED AT EQUAL INCREMENTS OUTPUT POINTS I N NORMALIZED POINT NWBER
1 .o
 input
distribu19
7.2.3
Constant Increments in Arc Length
This option results in points distributed uniformly according to the formula Si/P = (i

l)/(N
 1)
(7.2.2)
where S is the arc length measured from the first point and P is the total arc length around the perimeter of the Nline. It should be used for smooth bodies which do not have large variations in curvature (for example, the cylindrical fuselage section illustrated in figure 9. It should usually not be used for wing sections, since, for a reasonable number of points, there would be too few points to accurately define the shape near the leading edge or other highcurvature regions and too coarse spacing at the trailing edge for the Kutta condition to be accurately applied.
Figure 9. Point distribution on a cylindrical fuselage section  constant increments in arc length. Cosine Spacing Cosine spacing is a commonlyused distribution, datingfrom some of the classical twodimensional theories (reference 7). Its name derives in an obvious way from the usual formula for the spacing 20 7.2.4
Xi/C

1 2 C 1 + cos[(i

l)a/(N
 1)]>
(7.2.3)
where c i s t h e chord, which i s assumed t o be p a r a l l e lt ot h ex  a x i s . The s i g n i f i c a n c e o f t h i s method,as i l l u s t r a t e d i n f i g u r e 10, i s t h a t p o i n t s a r e chosen havingthexcoordinatescorrespondingtoequalincrements i n angle around a c i r c l e c i r c u m s c r i b e d a b o u t t h e N  l i n e w i t h l e a d i n g and t r a i l i n g edges
i n veryfinespacingnearleading and t r a i l i n g edges and coarsespacing i n regions away fromthese edges. I i s a t useful distribution option for wing sections but would probably not be comnonlyused f o r n o n l i f t i n g components.
t o u c h i n gt h ec i r c l e .T h i sr e s u l t s
I
EOUAL INCREMENTS OF
Figure 10.
Method o f d i s t r i b u t i n g p o i n t s
on an N  l i n e
 cosinedistribution.
Equation (7.2.3)
isnotdirectlyusefulfordeterminingthespecified method o f t h i s geometry package requires.
arclength distributions which the
To do t h i s , t h e w i n g s e c t i o n i s f i r s t scaled t o u n i t chord, t r a n s l a t e d t o put the origin of the coordinate system a t t h e l e a d i n g edge, and r o t a t e d t o make thechordpara1le1tothexaxis. c i r c l e i s then calculated for The angle (6) about thesuperscribed each p o i n t on the Nline by solving the equation
Xi/C
= 7 (1 + cos B )
1
(7.2.4)
21
using the transformed values of the xcoordinates. The arclength distribution corresponding to the input points is also calculated. For both the angle and arclength calculations, only the first Nline on a component is used (unless its length is zero, in which case the second one is used). Given the values of thearclength distribution and the angle distribution for the input coordinates, and the desired values of the angle distribution for the output coordinates (uniform increments), the arclength distribution for the output coordinates is determined by interpolation. Results of this option for the same supercritical wing section used previously are shown in figure 1 1
.
Figure 11.
Point distribution on a supercritical wing section cosine distribution.
7.2.5
Curvature  Dependent Distribution
The accuracy and computational efficiency of thepotentialflow method are both improved by using fine spacing in regions of high curvature and coarser spacing in regions of low curvature. This option produces a point distribution which depends on the curvaturedistribution along the first Nline of the component (or the second if the length of the first is zero). If the component is of lifting type, the point distribution is also made to be a function of the proximity to the trailing and leading edges in order to allow the trailingedge Kutta condition to be accurately satisfied and to keep the variation o f the spacing smooth in the leadingedge region where the curvature varies rapidly.
22
The following relationship between the curvature the spacing increments and is employed : Asi = (1  A
S ~ ~ ~ / A S ~ki/kmax) ( + AS mi n/Asmax'Asmax(7.2.5) ~ ~ ) ~

where Asi is the ith increment in normalized arc length between adjacent points, ki is the absolute value of the curvature the center of the at ith segment, kmax is the maximum absolute value of the curvature the Nline, on and Asmin and Asmax are, respectively, the minimum and maximum allowable increments in arc length. In this application, the ratio Asmin/~smax is specified to be 0.25. To implement this relationship requiresan iterative procedure. First the arclength and curvature distributions are calculated. Then, if the component is of lifting type, the variable which plays the role of the curvature in equation (7.2.5) is modified to makeits value at the trailing edge equal to its leadingedge value and to make it vary linearly from the leading and trailingedge values to the values at 7.5 percent o f the Nline's perimeter on either side of the leading trailingedge points. This and modification insures thatpoints will be closely spaced near the trailing edge and that the spacingwill not vary abruptly near theleading and trailing edges. To start the iterations, estimates of the values of bsmax and all the values of AS^ are required. These are a1 1 taken to be equal to the total length o f the curve divided by the number of segments it contains. Given the estimated values o f Asmax and AS^, the curvature at the center o f each segment (modified as described above)is determined by interpolation of the curvature versus arclength relationship and used in equation (7.2.5) to update the estimatedvalues of AS^. These values are then scaled to make their sum equal to the total length of the Nline and searched to determine the value of AS,,,^^. These updated values of AS^ and ~s~~~ are then used in the next iteration and the process is repeated a maximumof five times or until the required scale factor for the lengths of the increments is sufficiently close to unity. ion for the supercritical wing section used previously Results of this opt are are shown in figure 12. Results for a section of a nonlifting component shown in figure 13. 23
Figure 12.
Point distribution on a supercritical wing sectioncurvaturedependent distribution.
Figure 13. Point distribution on a section of a nonlifting componentcurvaturedependent distribution.
7.2.6
UserSpecified Distribution
For Nlines o f unusual shape or when components are located very close together, it is possible that none of the options described above will produce an adequate point distribution. To account for this possibility, provision has been made for the user to specify (as input data) values o f the normalized arc lengths, which are used directly to determinethe augmented coordinates.
7.3
Distribution of Nlines
After the points on the initial Nlines have been redistributed, each Nline on a component has the same numberof points distributed in a similar manner. Connecting corresponding points on a1 1 N1 ines generates a set o f 1 ines designated "lines. Since the distribution of points is the same on each Nline, the "lines are smooth and have fairly small curvature. The process o f
24
augmenting and redistributing the Nlines is accompllshed by augmenting and redistributing pointsalong these "lines. For thesecalculations, i t is required that each "line have a total length greater than zero.
There arefouroptionsavailablefor
1.
2.
3.
4.
the distribution
Of
"lines.
They are:
Input d i s t r i b u t i o n , una1 tered Input distribution, augmented i n number Constant increments Userspecified distribution
t o the descripThe descriptions of a1 1 four options are completely analogous o f points on Nlines tions of the corresponding options for distribution described i n section 7.2. There are two modes of operationof this portion of the geometry package. The f i r s t (and most common) mode i s designated the planarsecti on mode and the second i s termed thearclength mode i n thediscussions which follow.
7.3.1
The P1 anarSection Mode
The usual way of i n p u t t i n g data to the potentialflow program is t o make theNlines on a component planarcutsperpendicularto some axis of the component. For nonlifting components, t h i s convention is just a convenience t o theuser. For l i f t i n g components, however, theelementsare approximated by trapezoids, so that i f the N1 ines are notpara1 le1 , the accuracy of representation of the component by i t s elements is notas good. The planar section mode of operation redistributes the points on the "lines i n such a way that a l l Nlines, exceptperhaps the first and l a s t ones T h e f i r s t and l a s t Nlines generally define the on a component, are parallel. planform view ofthe component, which is not required to be composed onlyof straight lines. The method does not a l t e r these Nlines d u r i n g this portion of the calculations. The distributionsdescribed above generally refer to the distances between the parallel planes, normalized by the distances between the f i r s t and l a s t Nlines. A t the edqesof the component, when the f i r s t and l a s t Nlinesarenotplanar, the distributions refer to distances between the f i r s t
25
p o i n t on t h e N  l i n e under consideration
and t h e f i r s t p o i n t on t h e f i r s t N  l i n e
o f t h e component, r e s o l v e d i n t h e d i r e c t i o n p e r p e n d i c u l a r t o t h e p a r a l l e l p l a n e s and normalized by the distance between t h e f i r s t p o i n t s on t h e f i r s t and l a s t N  l i n e so ft h e component. The o r i e n t a t i o n o f t h e p l a n e s i s d e f i n e d b y s p e c i f y i n g (as i n p u t d a t a ) t h e d i r e c t i o n c o s i n e s o f a vectorperpendiculartotheplanes. For t h i s c a l c u l a t i o n , t h e sense o f t h e normal v e c t o r i s n o t i m p o r t a n t . However, i f intersection curves are calculated and the components aresubsequentlyrepaneled,thenthevector must p o i n t i n t h e d i r e c t i o n o f i n c r e a s i n gN  l i n e s . I the f directioncosinesarenotspecified and t h e p l a n a r s e c t i o n mode i s used, d e f a u l t
1 .O, 0.0) f o r l i f t i n g components are assumed. When thedefaultvaluesare used, Nlines must be i n p u t from f r o n t t o back on n o n l i f t i n g components and from t i p t o root o f l i f t i n g components.
values o f (1.0, 0.0,
0.0) f o r n o n l i f t i n g
components and
(0.0,
i s t o determine theequationoftheplanesrepresentingtheNlines and t h e i n t e r p o l a t i n g p o l y nomialsrepresentingthe"line segments. The equations o f t h e planes o f the N1 ine2 are o f form
The f i r s t step i n d i s t r i b u t i n g t h e p o i n t s on t h e " l i n e s
Ax
where
A,
+
B t Cz t Di y
= 0
(7.3.1)
B, and C are equal
normal to the planes The values o f
Di
t oh e t
thevector component).
x, y, and z d i r e c t i o n o s i n e s f c o (andhence are the same f o r a1 1 planes on t h e ( d i f f e r e n tf o r each plane) given are by
Di = Dl + Ki (D2
 Dl )
(7.3.2)
where
Dl
and
D2
are coefficients the the of planes passing through the component and underconsider
f i r s t and l a s t p o i n t s , r e s p e c t i v e l y , on t h e f i r s t " l i n e o f t h e Ki i s t h ev a l u eo ft h ed i s t r i b u t i o np a r a m e t e rf o rt h eN  l i n e
ation.
Dl
i s determined the from relationship
(7.3.3)
where "line.
xl,
De
y,
,
and
z1
are coordinates the
o f t h ef i r s tp o i n t
on t h e f i r s t on the
i s determined i n a s i m i l a r manner, u s i n gt h el a s tp o i n t The i n t e r p o l a t i n gp o l y n o m i a l sf o rt h e" l i n e i n s e c t i o n 7.1.
first"line.
segments a r e
determinedbythecurvefitproceduredescribed 26
The second step i n d i s t r i b u t i n g points on the "lines i s t o determine which "line segments intersect which planes. For this step, i t i s assumed that"line
curvature i s small enough so that straightline ments may be used.
approximations to the "line
seg
Then, knowingwhich "line segments intersect which planes andknowing the ofeach interequations of the "line segments and theplanes,thecoordinates sectionpoint can be computed. To do this for any point,substitutetheexpressionsforx,y,zas a function of arclengthalongthe"line segment (each This having the form of equation(7.1.2))intotheequationoftheplane(7.3.1). gives a singlecubicequation w i t h S i , thearclengthalongthe"line segment totheintersectionpoint,astheonly unknown. This i s solved by Newton's method. Using the intersectionpoint of the plane and thestraightline approximation to the "line segment t o determine the starting point, the method typically requires only three or four iterations to converge t o an adequatesolution. Figures 14 and 15 show thetop views of two typical wings both before and after redistributing the Nlines. Since the wing i n figure 14 is trapezoidal i n planform and has a linear twist distribution, only the Nlines at the t i p and rootare needed i n the initial geometricrepresentation.Since the initial representations of the t i p and root sections are already planar and all "lines are s t r a i g h t l i n e s , the planarsection mode and thearclength mode (described i n thenextsection) produce identicalresults. A more general case, for which the two modes ofoperationshouldgivevery different In this case,the"lines on the entireoutresults is shown i n figure15. board half of the span are straight lines, b u t they are curved on theinboard portion.Therefore,themultisection component optiondescribed i n section 7.1 i s used,allowingthelinearportionofthe geometry t o be represented using only two Nlines. Figure 16 shows thetop view of a fuselagebefore and a f t e r redistributing the Nlines. This fuselage has a cylindricalmidsection w i t h semiellipsoidalsectionsfore and a f t . Again, themultisection component optionallowsthelinearportionto be represented using only two Nlines. In all three of these cases, the redistributed N1 ines are equally spaced.
27
~
" "
..
. ._ . 
__
" . . .
"
. . ..
__
___ _..
.. .
..
. . _. . .
. .  ._
INITIAL ELEMENT DISTRIBUTION
ELEMENT DISTRIBUTION AFTER REPANELING
Figure 14.
Redistribution o f elements on a trapezoidal wing {cosine spacing chordwise, constant increments spanwi se, planar secti'on mode).
28
INITIAL ELEMENT DISTRIBUTION
ELEMENT DISTRIBUTION AFTER REPANELING
Figure 15. Redistribution o f elements on a supercritical wing (cosine spacing chordwise, constant increments spanwise, planarsection mode, mu1 tisection componentoption).
29
0
w
I N I T I A L ELEMENT DISTRIBUTION
ELEMENTDISTRIBUTIONAFTERREPANELING
Figure 16. Redistribution o f elements on a fuselage (constant increments around circumference and in axial direction, planarsection mode, multisection component option).
7.3.2
The ArcLength Mode
In some cases the Nline a t the edge of a component may be so highlynonplanar t h a t a strip bounded by this Nline andan adjacent planar Nline would vary strongly i n w i d t h , resulting i n areastoosparsely covered w i t h elements. An example of this i s the underwing pylon shown i n figure17(a). In this figure thethicknessofthe wing has been exaggerated tohelpillustratethepoint. In such cases the arclength mode of d i s t r i b u t i n g Nlines on the component should be t o fractionsof used. In this mode thespecifieddistributionparametersrefer the total arc lengths of the “lines, rather than to normalized distances between planes. The points on all “lines o f a component are d i s t r i b u t e d i n a similar manner, so that the Nlines i n the vicinity o f thenonplanar edge of the component curve smoothly around to create a more uniform distribution of the elements, asillustrated i n figure17(b). In this mode of operation,the method of redist r i b u t i n g pointsalong Mlines i s completelyanalogous to the method ofredist r i b u t i n g pointsalongNlines.
PlanarSection (a) Figure 17.
Mode.
( b ) ArcLength Mode.
Comparison of planarsection and arclength modes o f distribution of Nlines  strut on a thick wing.
31
I
8.0 8.1
CALCULATION OF INTERSECTION CURVES General Features of the Method
The second major operation performed by the geometry package i s the cal culationofthecurves of intersection between body components. For this calcuof one of the components (designatedthe lation i t i s assumed that the "lines intersecting component) pierce the surface of the other component (designated theintersected component). The finalsolution is a set of intersectionpoints, one for each augmented Mline on the intersecting component. In general , the o f a curve and a surface and hence method calculates the intersection point requires a surfacefit method as well as a curvefit method. The surfacefit method usesthetheory of parametriccubicsurfacepatchesoriginated by Coons (reference 8 ) and extensively developed and applied by a number of other investigators(reference 9 , f o r example). The curvefi t method i s the same one used i n thepaneling of isolated componentsand described i n section 7.1.
8.2
Restrictions and Limitationsof
the Method
I
The intersection method i s designed t o handlecases which occurfrequently i n aircraft applications, such as wingfuselage intersections, wingpylon intersections,etc.It i s notdesigned for complex cases i n which the interf i l l e t s o r smooth transisectioncurve is discontinuousorforcasesinvolving tions from component one tothenext. Use of the method i s limitedtocases i n which thefollowingrestrictionsapply:
1.
A distinction can be made between intersecting and intersected components. A component can intersect only one other componentand can be intersected by only one other component. If a body intersects or i s intersected by severalbodies, i t must be divided up intoseveral can intersect one component be and components. A single component intersected by another component, however.
2.
3.
The "lines on the intersecting component pierce the surface of the intersected component a t a sharpangle.Surfacesarenottangent where they meet.
32
4.
The “fines on the intersecting component extend sufficiently far into the interior of the intersected component to allow a planar representation of the elements on the intersected component to be used in the process of searching for the elements whichare intersected by the M1 ines. The intersecting componenthas at least one Nline which lies entirely in the interior of the intersected component. No ”line on the intersecting component intersectsthe intersected component more thanonce.
8.3 Details of the Method of Solution
5.
6.
The method of calculation of the intersection curve between two components that requi res one component be assigned the role of the intersecting component and the other component be assigned the role of the intersected component. This is done by the user in the input data to the geometry package. The intersection curve is then defined by the set of intersection points between the “lines on the intersecting component and the elements on the intersected component. In order to define this curve at a sufficient number of points, every “line on the intersecting component (after redistributing and augmenting the points on the Nlines) is used. On the intersected component, in principle either the original input elements or the elements after augmenting points on just the Nlines, or the elements after augmenting points on both Nlines and “lines could be used for defining surface patches which are intersected by the ”lines of the intersecting component. Since parametric cubic surface patches are used, however, it is not necessary to use the large numberof elements that exist after augmenting points on both Nlines and “lines. Since the input points may vary in number from one Nline to the next, and since the distributions may also be dissimilar, use of the original input elements could result in surface patches having much more extreme curvature(and hence less accuracy) than use of either of the other two sets of elements. Therefore, the elements after augmenting points on the Nlines but before augmenting points on the Mlines are used for the formation of surface patches.
I !
One possible method for calculating intersections of an M1 ine and a surface might operate as follows: Derive mathematical formulas to represent
33
each segment of the "line and each element of the surface and f i n d the roots o f theequationsobtainedby.equatingcoordinatevalues on the segments and the elei f i t does have real roots ments. If a particularequation has no real roots, or b u t the points they represent fall outside the bounds of theelement or the segment, thentheelement and the segment have no intersection points. Ifthe equation does have real roots and thepointstheyrepresent do f a l l w i t h i n the bounds of both theelement and the surface, then the roots are the intersection points. When a l l such equations have been solved,theintersectionpointsaregrouped, somehow, to determinetheintersectioncurves. Such a method would have tosolve a large number of equations(a number equal to the number of "line segments multiplied by the number of surface elements) and many of these equations would f a i l t o have solutions. I t would, therefore, be very uneconomical t o use.
i n the present method, f i r s t compares maximum and minimum coordinate values on the "line segments and the surface which obviously elements and eliminates from considerationthosecombinations
A more economical approach,adopted
contain no intersectionpoints. Then .an approximateintersection method, w i t h planarelementrepresentations, i s used to determine which "line segments and surfaceelementscontainintersectionpoints and to get approximatevalues of theintersectionpointcoordinates. Then, mathematical representationsare obtainedonlyforthoseelementscontainingintersectionpoints, and these are M1 ine segments, to calculate used,together w i t h the representations of the more accurately.Details of this approach are given theintersectionpoints bel ow.
8.3.1
The I n i t i a l Search for "Line Segments and Intersection Points
Surface Elements Containing
The basicinformation needed to conduct the i n i t i a l stages of thesearch incl udes the maximum and minimum values of each coordinate on each element of the intersected component, on each strip of the intersected component, on each segment of the intersecting "line, and on the entire intersecting "line. extend some distance into Because i t i s assumed that the intersecting "lines the body and a planarelementrepresentation canbe used, i t i s legitimate to an element must occur a t one assume that the maximum and minimum values on of i t s corners. A quick comparison of thecoordinatesofthecornerpoints
34
t h u s determines the extrema. Extreme values on a strip aredetermined'by comparingtheextremevalues o f each of i t s elements. Values on the"line segments are determined more accurately, assuming cubic relationships between the coordinatevalues and thestraightlinearclengthsalong the "line (equations(7.1.2) and (7.1.3)). Equation equation (7.1.2) tozero(separatelyforx,y, and z) and solvinggives two valuesofarclengthalongthe"line(foreachcoordinate). The real solutions which fa1 1 w i t h i n the bounds of the segment are used to cal culate coordinate values which are compared, along w i t h thevalues a t t h e ends o f the segment, to'determinethe extreme values. Maximum and minimum values on the entire intersecting "line are determined by comparing the extreme values of each of i t s segments.
Having determined the extreme values of thecoopdinates on s t r i p s and elements of the intersected componentand on segments and the entirety of the intersecting "line, the searchingprocedurebegins by comparing the extreme coordinatevalues on the f i r s t strip of the intersected component w i t h the If the rangeof any of the coordinatevalues valuesfor the entire "line. on the strip f a i l s t o include a t l e a s t p a r t of therange of the values of the same coordinate on the "line, then no intersection can possiblyoccur on that strip. Successive strips are compared w i t h the"line u n t i l one i s found which may possiblycontaintheintersectionpoint. Then each element on the strip i s compared w i t h the"line u n t i l a possibleintersected element i s found. Then each segment on theintersecting"line is compared w i t h theelement u n t i l a possibleintersecting segment i s found.
8.3.2
The FinalSearch f o r "Line Segments and Surface Elements Containing Intersection Points and the Approximate Determination o f the Intersection Points
Having found an element on the intersected componentand a segment of the intersecting "line which share, a t l e a s t p a r t i a l l y , the ranges of t h e i r x, y , and zcoordinates, a more careful check on whether or not they intersect i s made. If the element i s notalreadytriangular, i t is dividedinto two triangles by drawing one of the diagonalsacross the quadrilateral region. Each element i s t h e n represented by two planar,triangular subelements. The planar coefficients of each subelement are determined, starting from the general form of the equation o f a plane 35
Ax+ B y by f i r s t d i v i d i n g t h r o u g h b y
+
Cz + D = 0
(8.3.1)
D ( i l l u s t r a t i n gt h a tt h e r ea r eo n l yt h r e e independent unknowns i n equation(8.3.1)) and r e a r r a n g i n g t o g e t A/Dx
+
B/Dy
+ C/Dz = 1
(8.3.2) subelement
and t h e n s u b s t i t u t i n g t h e c o o r d i n a t e s o f t h e c o r n e r p o i n t s o f t h e i n t o e q u a t i o n (8.3.2) i se a s i l ys o l v e df o r t o o b t a i n a thirdorder system
of linear equations which are then found by
D.
A/D,
B/D, and C/D.
A, B, and C
m u l t i p l y i n g b y any a r b i t r a r i l y chosen ( f i n i t e , n o n z e r o ) v a l u e o f
A check i s then made t o see if the ends o f t h e " l i n e segment l i e on oppositesidesoftheplane.This i s done bycomputingthedirecteddistance from each p o i n t t o t h e p l a n e , u s i n g the.formula
(8.3.3)
d 2+ B2 A
where
+
C2
d i st h ed i s t a n c e and (xl , y1 , z1 ) a r et h ec o o r d i n a t e so ft h ep o i n t . I t h ev a l u eo f f d has t h e same s i g n f o r b o t h end points,the segment does n o ti n t e r s e c tt h ep l a n e o f t h e t r i a n g u l a r subelement. I f t h i s i s t r u e f o r bothsubelements on theelement,theelement and t h e " l i n e segment do n o t i n t e r s e c t and the next segment i s considered.
When an " l i n e segment which does cross the plane of oneof the subelements i s foundusing seg
i s found, t h e p o i n t o f i n t e r s e c t i o n of the segment and the plane the method described i n section 7.3.1. ment i s againrepresentedby r e g i o no ft h e subelement.
i i s next necessary t o checkwhether t
F o rt h i sc a l c u l a t i o n ,t h e" l i n e
a cubicpolynomial.Giventheintersectionpoint, o r n o t i fa1 1s w i t h i n t h e t r i a n g u l a r t method checks two points, t h a t each s i d e and the To do t h i s ,t h ep r e s e n t To determine that
of the triangle lies
on the same s i d e o f b o t h t h e i n t e r s e c t i o n p o i n t
A and B y l i e on the same s i d e o f a l i n e f r o m p o i n t s 1 t o 2, the method takesthecross p r o d u cotfh e e c t o r sr o m t v f A t o 1 and from A t o 2 and the cross product o f the vectors from B t o 1 and from B t o 2 and checks whether or not the dot product of the two r e s u l t i n g v e c t o r s i s p o s i t i v e .
opposite apex o f t h e r i a n g l e . t 36
If the p o i n t does not f a l l w i t h i n the bounds of either t r i a n g u l a r subelement, thenext segment of the "line is checked. I f none of the segments intersectstheelement,thenext element i s checked i n the above manner. If no element on the current strip contains the intersection p o i n t , the entire procedure,includingthe i n i t i a l searchdescribed i n section 8.3.1 , i s repeated for the next strip, and so on, u n t i 1 e i t h e r an element of the intersected body and a segment of the intersecting "line which contain the intersection p o i n t are found, o r all possible combinations of elements and Mline segments areexhausted.If an approximate intersection point is found i n this manner, i t i s used as a s t a r t i n g p o i n t for the iterative method described below, t o determinea more preciseintersection p o i n t . I f the method f a i l s t o f i n d an approximate intersection p o i n t , calculations continue, s t a r t i n g w i t h thenext"line on theintersecting component. In thiscase, execution of the computer program i s terminated after the intersection calculations since the final repaneling method requires the intersection curve t o be u n i n terrupted and compl etel y def i ned .
8.3.3
Derivation of a Mathematical Representation of theSurface Element
In order t o compute a more accurate intersection point,
of an
i t is necessary t o obtain an equation f o r thesurface of the element.This i s done u s i n g the theory of parametric cubic surfaces described i n references 8 and 9. The following discussion i s a sumnary of relevantportions of thetheory.
Requiredgeometric quantities a t each corner p o i n t of theelement under consideration include following the (as well as corresponding the y and z Val ues ) :
where thesubscriptsindicatedifferentiation and u and w are the parame t e r s upon which thesurface f i t is based. On the boundary curves of an element, u represents a fraction of thestraightlinearclength between the two M1 ine boundary curves and w represents a fraction of the straight1ine arc 1ength between the two N1 ine boundary curves. On the interior of an element, u and w representquantities analogous to those defined abov'e
37
~
but, for
elements w i t h compound c u r v a t u r e , t h e p h y s i c a l i n t e r p r e t a t i o n a t a c o r n e ri n d i c a t e st h ed e r i v a t i v eo f
parameters i s l e s s obvious. xu
o f the x
with respect to straightline arc lengthalongtheNlinepassingthroughthe corner, normalized by the straightline distance between adjacent points on t h e N1 ine. xw i n d i c a t e s t h e d e r i v a t i v e w i t h r e s p e c t t o s t r a i g h t  l i n e arc length along the "line passing through the corner, normalized by the s t r a i g h t  l i n ed i s t a n c e i n d i c a t ec r o s sd e r i v a t i v e between adjacent points on t h e l i n e . " xuw and xwu and Nline. terms i nt h ed i r e c t i o n so ft h e" l i n e and augment p o i n t s a l o n g t h e " l i n e s xuw and
The f i r s t d e r i v a t i v e terms (not normalized) were previously required for the calculations to redistribute able here. and Nlines. be a p p l i c numerThey need o n l y be normalized by the proper element side length to The cross derivative terms ically differentiating the along the "lines unnormalized f i r s t d e r i v a t i v e terms and thennormalizingbythe The two c r o s s  d e r i v a t i v e used t o r e p r e s e n t
xwu are obtained by
and Nlines, respectively,
productofthelengthsoftheadjacentelementsides. values a t each cornerarethenaveraged,sincetheequation the surface implies that they are equal. These geometricquantities,for geometric form." (shown o n l y o r h e f t x, y , and
z
and f o r each c o r n e ro ft h e
element, c o n s t i t u t e t h e s o  c a l l e d " p a r a m e t r i c c u b i c p a t c h c o e f f i c i e n t s The c o e f f i c i e n t s can be x coordinates): arranged i n m a t r i x f o r m
in
as f o l l o w s
I
CGXl =
x10
xll xwlo
X X
xl w
"
x
uoo
uol
X
uwoo u o w l
X
X
(8.3.5)
I xulo L
I nt h i sm a t r i xt h e The "line.
y
xull
uwl 1 0 uwl
as above.
u and w
subscriptsagainrepresentderivatives
0
and The
1 subscripts ogetherndicate he orner f he t i t c ot
first indicating the
element being and the 1
considered, the
N1 i n e and t h e second i n d i c a t i n g t h e
0
indicates the
lowernumbered l i n e on the element
indicatesthehighernumberedline.Similarmatricesarealsoformedforthe and zcoordinate terms.
38
The matrixof coefficients i n geometric form, [,, G] and the corresponding y and z matrices contain a l l the information needed t o derive algebraic expressions for the coordinates a t any point on the surface of the element i n terms of the parametricvariables. These expressionsare of the following form (again shown only f o r the x coordinates):
x(u,w) = w 3[Axu 3 + Bxu 2 + CXu +
DX]
+ w 2[Exu 3 + Fxu 2 + G,u + Hxl
+ w [Ixu 3 + Jxu2 + Kxu +
+
(8.3.6)
Lx]
[Mxu 3 + Nxu 2 + PXu + Px]
The coefficients of equation(8.3.6) can be grouped t o form the socalled "matrix of coefficients i n algebraic form," [A,] as follows:
p x Bx cx
CAX1
=
Ex Fx Gx
I J x Kx X
LMX
Nx
b, L X pX
of coefficients i n geometric form may be converted i n t o thematrix o f coefficients i n algebraic form, i t i s f i r s t necessary t o study theproperties of equation(8.3.6)further. On the boundaries of the element and the other one of the parametric variables i s constant, either zero or one, varies from zero t o one. For definiteness, assume thatthevariable w is equal tozero. Equation (8.3.6) then reduces t o a cubic equation w i t h u as the independent only variable. On the oppositeside of the element, w i s equal t o one and equation (8.3.6) reduces toanothercubicequation,again w i t h u as the only independent variable. The entiresurface of the element can be considered t o be a collection o f cubiccurves w i t h w constant and u variable. Each of thesecurves has the form ofequation(7.1.2), which be can converted to the form
To show how the matrix
I
(8.3.7)
f ( u ) = F1(u)f(0)
+ F 2 ( u ) f ( l ) + F3(u)f'(0) + F 4 ( u ) f ' ( l )
(8.3.8)
39
by solving for the secondand t h i r d d e r i v a t i v e terms i n (7.1.2) s e c t i o n 7.1. When t h i s i s done, theterms F1 (u),F2(u),F3(u),. given by 2 F1 (u)= 2~3
as described i n and F4(u)are
 3~ +
+
3u
1
2 F2(u) = 02u3 F3(u) = u3 F4(u) = u 3
 2u2 + u
(8.3.9)
 u2
(8.3.10) where 2 +1 +1
CMI =
t 3 1 2 0 + 1
0
(8.3.11 )
0
I nm a t r i xn o t a t i o n ,e q u a t i o n (8.3.8)
0
0
becomes (8.3.12)
F(u) = [F]
= [u3

[f (0) f ( 1 ) f ' ( 0 ) f'(l)lT
u2 u 1 1 M [I
[f(O)
f(1) f'(0) f'(1)IT
where T (8.3.12), and w = 1
indicatesthatthetranspose o f t h em a t r i xi s t o be taken. Now, using the equations for x(u) and xw(u) on the boundary curves w = 0 can be w r i t t e n as 3 2 X(U,O) = [u u u 1 1
[MI
. [xoo
xl0
xu IT 00 x u l o
Xw(U,l)
=
[U
3 2 u u 13
[MI
9
[, X
01 wll w l u xuwll o
X
X
IT
40
o r more compactly (8.3.14) Equation (8.3.12) can also be used t o derive the expression
x(u,w) = [w 3 w2
or equivalently
W
11
[MI
[X(U,O)
X(u,1)
Xw(U,o)
xw(u,l)lT
(8.3.15)
Combining (8.3.1.4) and (8.3.16)gives
x(U,w)
=
[u?
U2
u 11
[MI
[G,]
[MIT
3 2 T w w 1 I
[w w w 13
3 2
T
(8.3.17)
Sinceequation
X(U,W)
(8.3.6) can be
rearranged and written as
=
[u3 u2 u !I
*
LAXI
[W
(8.3.18)
coefficients i n geometric form i s converted t o thematrixof coefficients i n algebraic form by the operation
(8.3.19)
the matrixof
Expressions f o r the y and z coordinates functions as of the parameters and w areobtained i n asimilar manner, u s i n g thematrices of geometric q u a n t i t i e s [G 1 and [G,].
Y
u
Including coefficients for all three coordinates, there are fortyeight parametriccubicquantitiesassociated w i t h each element. Except f o r the coordinatesthemselves (twelve of the fortyeight quantities), these quantities aregenerallynotshared by adjacent elements. However, these coefficients a r e a l l derived from geometricdata which i s continuous from element to elementand, therefore, i s shared by adjoiningelements.Therefore,alarge reduction i n the computer program storagerequirements can be made, a t a small expense of additional computatjon time, by storing all data i n the (unnormal ized) geometric formand converting to the algebraic form of equation (8.3.6) only when actually needed. Therefore, the coordinates, their derivatives along the Nlines, and "lines, their crossderivatives, and the arc
41
lengths between adjacent points are method.
8.3.4
the onlygeometricdata
used i n the present
Computation of More Precise Values of the CoordinatesoftheIntersection P o i n t
A t this point, analytic expressions have been obtainedforthecoordinatesalong the intersecting Mline segment i n terms of s , thearclength alongthe segment (equation(7.1.2)) and forthecoordinates on thesurface of the element i n terms of the parametric variables u and w (equation subscript s and points (8.3.6)). Designating points on the surface by the E, expressions have been obtained on the "line segment by the subscript intersecf o r xQ(s),y,(s), z Q ( s ) . x S ( u , w ) , ys(uyw), and zs(u,w). A t the tionpoint
(8.3.20)
Equation (8.3.20) represents a system of threesimultaneousnonlinear tions i n the three unknowns (u,w,s).
equa
The s o l u t i o n of this systemofnonlinearequationsrequires an i t e r a t i v e procedure which must s t a r t from some estimate of thesolution. An estimate o f the solution, i n terms of thecoordinates of the intersection point (xi , yi , z i ) was obtained d u r i n g thesearchingproceduretodetermine which "line segment intersects which element of theintersected component. An estimate of s i , thearclengthalongthe Mline segment totheintersection point, was alsoobtained. The coordinatedata must be used t o determine estimates of the values o f the u and w variablesattheintersectionpoint ( u i , wi ) . In the present method , the two pl anar subelements are converted planar element and the values of u and w i n the planar into a single element arethendetermined. The planarelement i s formed from thethree cornerpointsdefiningthesubelement i n which the approximation to the intersection point i s located and a fourthpointobtained by rotating the other the two subcornerpoint of the original elementaboutthelineseparating elements u n t i l i t lies i n theplane of theotherthreepoints. 42
The parametric coefficients of a point i n a planarelement bounded by straight lines can be determined most easily be f i r s t noting that all coefficients of second and thirdorder terms i n equation (8.3.6) must vanish and then by f i n d i n g the other four coefficients directly from the properties of the sides of the element. The result i s thatequation (8.3.6) reduces t o x(u,w) = x00
+
(x10  x00)u
+
(x01 xoo)w
+
(x11  x10  x01 +
XO&W
(8.3.21) Equations f o r the y and z coordinates have similar form. Given the values of x,y and z a t a point, equation (8.3.21) and the corresponding y and z equationsconstitute a system of threeequationsforthe two unknowns ( u and w ) . The nonlinear term uw i s eliminated from each equation and then one equation i s added t o eachof the other two equations to result, finally, i n a systemof two independent linear equations for the two unknowns. (si , u i , w i ) a t theintersection Given the estimate of thevariables system (8.3.20). The point, Newton’s method i s used tosolvethenonlinear process generally takes only four or five iterations to converge to an error i n thesquarerootofthe sum of thesquares o f the variables of less t h a n f o r a typical case. When solved, si can be used i n equation (7.1.2) and the corresponding equations for y and z to obtain the coordinates of the intersection point. Alternatively, ui and wi could be used i n (8.3.6) and the corresponding y and z equations.
8.3.5
Test Cases fortheIntersection
Method
In order to verify the accuracy of the present method, a number of t e s t cases were run. These arenot examples of r e a l i s t i c a i r c r a f t components, b u t they were chosen because their intersection curves can be analytically determined. Figure18 shows a relatively simplecase,theintersectionof two circular cy\ inders, Using onlyfourelementstorepresent each quadrant of each cylinder produces remarkably good results. There i s no discernable d i f ference between thetheoretical and the calculatedresults(tothescale plotted). This i s an especiallyaccuratecasefor any method, sincethe intersecting “lines are straight lines and theelements on the intersected
43
A more d i f f i c u l t case, t h e i n t e r s e c t i o n o f two spheres, i s shown i n f i g u r e 19. This case t e s t s t h e f u l l c a p a b i l i t yo ft h ei n t e r s e c t i o n method. The i n t e r s e c t i n g " l i n e segments a r e c i r c u l a ra r c s . The i n t e r s e c t e d elements have curvature i n b o t hd i r e c t i o n s and some of the intersected elements have zerolength sides (i.e. , some elements aretriangular).Nevertheless,thepresentmethod'sresults,using 250 e l e ments on each sphere, a r e v e r y a c c u r a t e ; t h e c a l c u l a t e d i n t e r s e c t i o n . c u r v e i s o n l yv e r ys l i g h t l yd i f f e r e n tf r o mt h et h e o r e t i c a l one. Figure 20 shows the f i n a l t e s t case, t h e i n t e r s e c t i o n o f two e l l i p s o i d s . L i k e t h e case o f t h e two spheres, t h i s case shows good r e s u l t s ,u s i n g 250 elements body. per The method f a i l e d t o f i n d i n t e r s e c t i o n p o i n t s f o r t h e " l i n e s p a r a l l e l t o t h e xaxis i n f i g u r e 20, sincethese"lines meet t h e i n t e r s e c t e d component tangential ly.
component have curvatureinonly one d i r e c t i o n . In order to illustrate the a i r c r a f tc o n f i g u r a t i o n s ,f i g u r e givetheintersectioncurve greatly exaggerated. method f o r a c o n f i g u r a t i o n more t y p i c a l of
21 i s included.This
shows a wingintersecta typicalfuselage. To has been
i n g a c y l i n d r i c a ls e c t i o ns i m i l a rt ot h em i d s e c t i o no f
more character, the thickness of the wing
7
THEORETICAL INTERSECTION
CURVE
CALCULATED INTERSECTION POINTS
J
Figure 18.
I n t e r s e c t i o n method t e s t case cylinders(sideview).
 intersectionof
two c i r c u l a r
44
SIDE VIEW
END VIEW
THEORETICAL INTERSECTION
0
CURVE
" " "
THEORETICAL INTERSECTION
CURVE
CALCULATED INTERSECTION POINTS
CALCULATED INTERSECTION POINTS
Figure 19.
Intersection method test case  intersection o f two spheres.
Y
Figure 20.
Intersection method test case intersection o f two ellipsoids (sideview).
45
9.0
FINAL REPANELING OF COMPONENTS 9.1 General Considerations
The final major operation performed by the geometry package is the redistribution of points on a l l components which intersect other components. The pointspacingrequirements of the potentialflow method for intersecting bodies arenotas wellunderstood asthoseforisolatedbodies. The accuracy of surfacesingularity type methods i n regionsnear concave corners is a matterof dispute i n twodimensions (reference 10) and has not been extensively.studied i n three dimensions. Some obviousrequirements can be identified, however. One i s thatportionsof components which fallinsideother components should be eliminated,or i n certaincases,designatedasextra s t r i p s o r ignoredelements. Another i s thattherepanel i n g shouldnotcause abrupt changes i n the elementspacing.Therefore, the entire component should be repaneledto produce a smooth transition from thepaneling i n theregion of the intersection curve to the paneling i n distant regions, rather than just I t i s alsopossible theregion o f theintersection curvebeingrepaneled. that future potentialflow methods (suchas the method of reference 5) will require the matching ofthecornersoftheelementsofadjacent components t o eliminate or at least reduce the size of any gaps between elements.
I t i s possibleto envision innumerable different types of intersections between bodies, each of which would have i t s own specialpanelingrequirements. I t i s not possibleto develop a method sufficiently general to deal w i t h them a l l . However, plausiblepaneling schemes can be developed forcertain frequently occurring configurations, such as wingfuselage or wingpylon cases,sothatthesecases can be handled routinely. The present method divides all components which areinvolved i n intersections into three distinct general categories and providesseparate means ofdealing w i t h each o f them. A 1 intersecting components (those wh M1 ines which pierce other components) 1 it are repaneled one way, w i t h only minor variations (whether they are l i f t i n g ornonlifting).Nonliftingintersected components are repaneled differently, and l i f t i n g intersected components are repaneled i n a different way s t i l l . The methods used for these three categories are described i n thefollowing sections.
47
9.2
Intersecting Components
Typical examples of intersecting components includethe wing i n a wingfuselage case, the pylon i n a wingpylon case, the winglet i n a wingwinalet case, etc. A though a1 1 these examples are of 1ifting intersecting components, 1 they can also be n o n l i f t i n g (struts, for example). L i f t i n g and n o n l i f t i n g intersecting components arerepaneled i n essentially the same manner. There i s no redistribution o f points alongNlines,only a redistribution of Nlines. In every case, a new Nlinealongthe intersection curve i s added. Ifthe component i s of l i f t i n g type and i f the intersecting end of the component i s designatedas having an extra strip, thentheentireareainsidetheintersected component (from the intersection curve to the end Nline) i s made i n t o a single strip by eliminatingallinterveningNlines.Ifthe component i s o f n o n l i f t i n g type, or i t does not have an extra strip at the intersecting end, then the portion insidetheintersected component i s simply eliminated. In the simplestcases, no other redistribution of points or Nlines i s necessary. on the exterior of the component In most cases, however, theremainingNlines must be moved i n order t o avoid i r r e g u l a r i t i e s i n thewidths o f the strips. between Nlines by the This i s done by simply scaling the specified distances fraction of the span of the componentwhich lies outside the intersected component. To be more precise,thespecifieddistances mentioned above are either the normalized distances between theplanes of the N1 ines (for the planarsection mode of operation) or the fraction of the arc lengths along the"lines (for thearclength mode o f operation). The fraction of the span of the componentwhich lies outside the intersected component is the fraction of the arc length along the "line under consideration (for the arclength mode of operation) or the fraction of the arc length along the f i r s t M1 ine of the component (for the pl anarsection mode o f operation). The numerical techniquesrequired t o move theNlines i n t h i s manner involveonlyinterpolat i o n procedures or cubiccurveplaneintersectionprocedures which were described i n section 7 above.
An example o f the program capabilities for repaneling intersecting components i s shown i n figure 22. Figure 22(a) shows a wingfuselagecase w i t h the elements required fortheinitial geometricrepresentation. In have repaneled,theintersection figure 22(b), theisolated components been
( a ) I N I T I A L ELEMENT DISTRIBUTION
.(b) ELEMENT DISTRIBUTION
AFTER FINAL REPANELING OF WING
Figure 22.
Final repaneling
of i n t e r s e c t i n g components wingfuselage case.
49
curve has been made an N  l i n e and a l l e x t e r i o r N  l i n e s t o produce a smooth componenthasbeen the extra strip
havebeen
redistributed
d i s t r i b u t i o n .I na d d i t i o n ,t h ea r e ai n s i d et h ei n t e r s e c t e d made i n t o an e x t r a s t r i p , b u t f o r c l a r i t y o f p r e s e n t a t i o n ,
has n o t been shown i n t h e f i g u r e . i n which abody i n t e r s e c t s more than one o t h e r However, i n thesecases For shown i n f i g u r e 23, outboard component
More complicatedcases, t h ei n t e r s e c t i n g thewing body must
body , can a l s o be t r e a t e d by t h i s geometrypackage. example, i n t h e case o f t h e w i n g  f u s e l a g e w i t h t i p  t a n k
,
be d i v i d e d i n t o more than one component. semispan) i n t o two components, an andan
i s divided(atfiftypercent
inboard component which intersects the fuselage whichintersectsthetiptank.Sincethe b u t i n gp o i n t s on the Nlines
same a l g o r i t h m was used f o r d i s t r i 
o f each component, and s i n c e p o i n t s on theNlines o f i n t e r s e c t i n g components do n o t g e t r e d i s t r i b u t e d , t h e r e i s no mismatch i n theelements a t t h ej u n c t i o n between t h e components. Because of t h e r e d i s t r i b u t i o n o f theNlines, however, t h e r e i s a smalldifference i n t h e w i d t h of t h es t r i p si nt h ei n b o a r d and outboardportions o f thewing. Cases i n which a body completelypiercesseveralotherbodies can be handledby d i v i d i n g t h e i n t e r s e c t i n g body i n t o components i n an obvious way.
Figure 23.
F i n a lr e p a n e l i n go fi n t e r s e c t i n g case.
components wingfuselagetiptank
50
9.3 Nonlifting Intersected
Components
Typical examples of nonl i f t i n g intersected components include the fuselage i n awingfuselagecase,thetank i n awingtiptankcase,etc. The most common i n g method i n application is tothewingfuselagecase.Therefore,therepanel this geometry package was conceivedprimarily w i t h fuselages i n mind, b u t i t should also be useful for other types of cases.
Three options are provided for the final repaneling of nonlifting intersected components. The simplestoption i s t o do no repaneling a t a l l , i n which casethefinal element distributionsareas shown i n figures 22 and 23. If this option i s used, aportion of the intersected component falls inside the interan adequate secting component. If the intersecting component is notverythick, potentialflowsolution may s t i l l be obtained i n spite of this. If the intersecting component covers a significant area on the intersected component, then the elements inside the intersecting component should be designatedignoredelements (elements w i t h no singularities and no boundary conditions).Sincethere i s no repaneling, however, some elements are only partially covered and the decision whether ornottheelementsshould be ignored requires some user judgment. For this reason, no mechanism has been provided for automatically designating elements t o be ignoredelements on nonliftingintersected components. If the using the userdesiresto use ignoredelements, he must repaneltheconfiguration geometry package, punch theresultingcoordinates on cards, and terminate execution of the program. Then he must executethe program a second time, u s i n g the i n p u t data,designatingtheignored punched output from theprecedingcaseas elementshimself w i t h the appropriate flags, s k i p p i n g the geometry package, and proceeding straight to the potential flow calculations.
on the intersected component is notinfluenced by the element distribution on the intersecting component. Experience w i t h the method ofreferences1 and 2 indicates that there i s no strong need to repanel fuselages i s most wingfuselagecases;accuracy of the potentialflow solution is usuallyadequate u s i n g a distribution of elements appropriate for an isolatedfuselagecase. However, i n some cases even better accuracy i s desired. Also the newer potentialflow methods, using surfacedoublet d i s t r i b u tions, may require the elimination of the smallgaps which result when the edges
51
W i t h the f i r s t option, the element distribution
of adjacent elements are not carefully aligned. The second and third options (both very similar) repanel nonlifting intersected components to produce element distributions which depend strongly on the element distributions on the intersecting components. The calculations in both options start by defining a leadingedgeand a trailingedge point on the intersection curve. The trailingedge point is assumed to be the first or last point on the intersection curve, since this curve is an Nline of the intersecting component and points on Nlines are input starting at the trailing edge, working around the perimeter, and ending back at the trailing edge. The leadingedge point is found by searching for the point on the intersection curve which has the smallest projected distance in the axial direction from the front of the interesected component. Because of this, nonlifting intersected components whichare repaneled with one of these two options must be input starting at the upstream end, must have roughly streamwise "lines, and must be paneled using the planarsection mode of operation. The axial direction is defined by the direction cosinesof the vector perpendicular to the planes of the Nlines. Planar cuts are then made through the leading and trailingedge points on the intersection curve, perpendicular to the axis of the intersected component. The points of intersection of these planes and the "lines of the intersected component define new Nlines. All Nlines forward o f the leadingedge point and all those aft of the trailingedge point are then redistributed in a manner similar to the redistribution of the Nlines on intersecting components. Forward of the leadingedge point, this is done by scaling the projected distances in the axial direction from thefront o f the component by a factor equal to the ratio of the distance from the front of the component to the leadingedge point and the distance from the front of the component to the first Nline aft of the leadingedge point. Aft o f the trailingedge point, the redistribution is done the same way, except that the back of thecomponent, the distance to the trailingedge point and the distance to the first Nline in front of the trai 1 ingedge poi nt are used. The two options differ only in the distribution of Nlines in the region between the leading and trailingedge points on the intersection curve. One of the two options passes planar sections through each point on the intersection curve and uses the points of intersection o f these planes and the
52
Mlines of the intersected component todefine new Nlines. T h i s option can be used when an element distribution w i t h no gaps between adjacent elements is desired. The otheroptionpassesplanarcutsthroughevery second point on the intersection curve, starting w i t h theleadingedgepoint and working a f t on both upper and lower surfaces of the intersectioncurve.Since each surface may contain either an even or odd number of points, i t i s possible that planes will be passedthrough two consecutivepoints a t the back ofthe intersectioncurve. T h i s option can be used when gaps between adjacent elements can be tolerated, b u t i t i s desired to keep their size fairly small.
Having redistributed the N1 ines , i t is next necessary to redistribute the points on each Nline. The method f o r doing this is the same for both availableoptions. First, a search of the Nlinepassingthroughtheleadingedge point on the intersection curve i s conducted to f i n d the point closest ( i n arclengthalongtheNline)totheleadingedgepoint. T h i s point i s moved tocoincide w i t h theleadingedge p o i n t . The otherpoints on this Nline are redistributed i n a manner which makes the resulting arclength For distributionreasonably smooth and similar tothepreviousdistribution. points on the Nlinebeforethepointclosesttotheleadingedgepoint on theintersectioncurve, this i s done by scaling the arc l e n g t h alongthe Nline by a factor equal to the ratio of the arclengthtotheleadingedge point and thearclengthtothepointclosesttotheleadingedgepoint. For points on theNline after the point. closest to the leadingedgepoint,arc lengths are scaled the same way, b u t using arc lengths from the other end of the Nline. Having determined thearclengthdistribution of theredistributed points, the coordinatesare found by the same interpolationproceduredescribed i n section 7.1. Pointsareredistributed on Nlinesforward of the leadingedge point using the same arclength distribution as on the Nline which passes through this point. Between leadingedge and trailingedgepoints, points on Nlines are redistributed i n a similar manner by scaling the arc lengths by a factor determined from the ratio o f the distances to the intersection point and to the point on the same M1 i ne as the point cl osest t o the leadingedgepoint. Aft o f the trailingedgepoint,points on a l l Nlines are redistributed using the same arclength distribution as the redistributed points on the N1 inepassingthrough the t r a i 1ingedge point. Because the
53
Nlines t h r o u g h the intersection points between theleading and trailingedge i t i s necessary t o break the points are broken by the intersection curve, component i n t o more than one component. Presently,the program breaks the component i n t o four smaller components , one forward o f the leadingedge p o i n t , onebelow the intersection curve, oneabove the intersection curve, and one a f t of the intersection curve.
of this repaneling method i s shown i n figure 24. Thiscase shows awingfuselage w i t h the wing a t zeroincidence, locatedslightly above the axis of thefuselage. In thiscase, use has been made of the option which passes an Nline through every p o i n t on the inter25 shows a similarcase,.usingthe same option. In sectioncurve.Figure this case, however, the wing has tendegrees of incidence, illustrating the p o i n t t h a t , for such cases, the present method bunches points below the In intersection curve and spreads them out above theintersectioncurve. extreme cases, this may r e s u l t i n an unacceptable.element distribution w i t h many elements bunched t i g h t l y near the bottom of the component and very few elementsnearthe top. However, most caseswill probably be much less shown, for which theelement distribution is adequate. extreme than the one Figure 26 shows a wingfuselagecasesimilar t o the one o f figure 24, b u t w i t h the wing located a t the same levelastheaxis of thefuselage. In this case,the option which passesNlines through every second p o i n t on the intersection curve has been used.
A typical example o f the results
9.4
L i f t i n g Intersected Components
The most common example of a lifting intersected component i s the wing i n a wingpylon o r wingwithstrut case.
of nonlifting intersected components, l i f t i n g components are treated three different ways. Again, thesimplest option i s t o do no repaneling a t a1 1. In this case, p a r t of theintersecting component covers a portion of thesurface of the intersected component and the remarks of sect i o n 9.3 concerningaccuracy of the solutions, the desirability of using ignoredelements, and the mechanism for generating them again apply.
As i n thecase
54
TOP VIEW
SIDE VIEW
(fuselage only)
Figure 24. Final repaneling of nonlifting intersected components  wingfuselage case (zeroincidence wing, Nlines through every point o f intersection curve).
SIDE VIEW (fuselage only)
Figure 25. Final repaneling of nonlifting intersected components wingfuselage case (wing with 10" incidence, N1 ines through every point on intersection curve).
55
TOP VIEW
SIDE VIEW
Figure 26.
Finalrepaneling of nonliftingintersection components wingfuselag: case(zeroincidence wing, Nlinesthroughevery otherpoint on intersectioncurve).
In some cases the intersecting component covers a significant portion of the surface of theintersected component, indicating a need for the use o f ignored elements, b u t thepaneling of the intersected component i s so coarsethatlarge gaps would be created by t h e i r use. The geometry package provides an option whic; greatlyreducesthesize of these gaps. In this option a search i s conducted t o f i n d the points on the intersection curve having the minimum and maximum projectel on theintersected component from the f i r s t point distance i n theaxialdirection on the f i r s t Nline of the component. The intersected component be must paneled usingtheplanarsection mode of operation for this calculation t o be performed. The axial direction is defined by thedirectioncosines of thevectorperpendiculartotheplanes of theNlines. Nw Nlinespassingthroughthe e minimum and maximum points on the intersection curve and lying i n planesperpendiculartothe
56
axial direction are t h e n created. The Nlines on either side of the intersection manner using the same procedure as the rediscurve are redistributed i n a smooth tribution of Nlines upstream anddownstream of the intersection curve on a nonl i f t i n g intersected component. No redistribution o f points on theNlines i s performed. The elementscovered by theintersecting component areautomatically designatedasignoredelements by the geometry package and i t is possible t o proceed directly to the potentialflowsolutionwithoutcheckingor changing the results of the geometry package. Figure 27 shows the results of this repaneling option for the wing i n a wingpylon case. When more accuracy i n theregion of the intersection curve is desired, a t h i r d option can be used, resulting i n no gaps between elements. T h i s option first repanels the intersected component i n a manner nearly identical imneditothe one described above. The onlydifference is thattheNlines ately on either side of the intersection curve do not actually pass through each is offset a any o f thepoints on theintersectioncurve.Instead, distance equal t o the thickness of the intersecting component (projected i n theaxialdirection). The elements covered by theintersecting component are againdesignatedto be ignoredelements. Nw the method adds a new nonlifting o
Figure 27.
Finalrepaneling
o f lifting intersected
components wingpylon
case.
57
component w i t h elementsdesigned t o f i l l i n the gapsaround the intersection curve. The new component has only one strip (two Nlines). One of the The other N1 inefollowsthe Nlinescoincides w i t h theintersectioncurve. boundaries of thequadrilateralregioncontainingtheignoredelements. Figure 28 shows the resulting element distribution for a case w i t h very sparse p o i n t spacing on thepylon. Because o f thecorrespondence o f the numbers of points on the pylon and on theboundaries o f theregioncontainingtheignored a t theleadelements,thepaneling produced onlyquadrilateralelements(except If the numbers do not correspondas i n ing and t r a i l i n g edges of thepylon). figure 28, then one or the other of theNlines on the new component contains in repeatedpoints and t r i a n g u l a r elements are produced. Pointsarerepeated such a way t h a t thelength of each "line on the new component is minimized. Figure 29 shows a case having more points on the pylon than on theboundaries of theregioncontainingtheignoredelements, showing therepeatedpoints on one Nline and the resulting triangular elements. Nlines which bound extra strips falling inside another component are n o t moved d u r i n g therepaneling of l i f t i n g intersected components. Because o f t h i s and because none of theoptionsdescribed above in thissection changes the
BOlTOM
VIEW
REGION CONTAINING
Figure 28.
58
Final repaneling of lifting intersected components  wingpylon case (sparse element distribution on pylon).
BOTTOM VIEW
REGION CONTAINING IGNORED ELEMENTS AND EXTRA NONLIFTING ELEMENTS
Figure 29.
Finalrepaneling of l i f t i n g intersected components wingpylon (more points on pylon thansurroundingregion on wing).
case
point distributions on Nlines, therepaneling of a l i f t i n g intersected component does nothing to destroy anymatchup of the elements along the curve of interwhich i t pierces.Therefore, one sectionofthe component w i t h another component component can playtheroleof both an intersected and an intersecting component (although one component cannot intersect or be intersected by more than one other component). Figure 30 shows thefinal element distribution for a wingfuselagepyloncase. In this case,the wing and pylon were f i r s t repaneledas intersecting components; t h e n the fuselage and the wing were repaneledas intersected components.
59
I
" "
. ....

.
._ .
,
Figure 30.
F i n a le l e m e n td i s t r i b u t i o n
on a wingfuselagepylon case.
60
10.0
CONCLUSIONS
The geometry package described above provides a m a n s for significantly reducing the effort required to prepare the input data for threedimensional potentialflow calculations. Data may be input to the program using either the original input format of reference 4 or the format of reference 5. Geometric input data generally consists of thecoordinates of sets of sparsely defined points. In most cases, the nunJ>er of points input can be at least an order ofmagnitude less than the number of points required for the potentialflow calculations. Each component of the configuration is automatically paneled using one of several a1 gori thms provided by the method. The number of algorithms provided and the generality of some of them provide the user with a great deal of flexibility in determining the character of the resulting element distributions. Curves of intersection of components are automatically calculated and all intersecting components are repaneled in a suitable manner for the potentialflow calculations. In many cases the potentialflow calculations can be performed in the same computer run as the geometry package calculations, without interruption. To allow for intermediate checking, however, and to provide for those cases which cannot be run completely without interruption, provision has been made for the punched output of the defining coordinate data at each stage of the geometry calculation in a format suitablefor reinput to the program.
61
11 1.
.o
REFERENCES
Hess, J.L.: C a l c u l a t i o no fP o t e n t i a l Flowabout A r b i t r a r y ThreeDimensional L i f t i n g Bodies. F i n a l Tech. Rept. McDonnell Douglas Report No. MDC J567901, October 1972. ( A v a i l a b l e f r o m DDC as AD 755 480. ) Hess, J.L.: The Problem o f ThreeDimensional L i f t i n g P o t e n t i a l Flow and I t s S o l u t i o n b y Means o f S u r f a c e S i n g u l a r i t y Dhtribution. Computer 4, No. 3, November 1974. Methods i n Applied Mechanics and Engineering,Vol. Loeve, W. and Sloof, J.W.: On the Use o f "Panel Methods" f o rP r e d i c t i n g Subsonic Flow About A e r o f o i l s and A i r c r a f t C o n f i g u r a t i o n s . NLR MP71018U, October 1971. Mack, D.P.: CalculationofPotentialFlowaboutArbitraryThreeDimensional L i f t i n g Bodies. Users Manual. McDonnel1 Douglas Report No. MDC 5567902, October 1972. T u l i n i u s , J., e ta l . :T h e o r e t i c a l P r e d i c t i o n of A i r p l a n e S t a b i l i t y D e r i v a t i v e sa tS u b c r i t i c a l Speeds. NASA CR132681 , 1975. G r e v i l l e , T.N.E.: Theory and ApplicationsofSplineFunctions. Press, New York , 1969. Academic
2.
3.
4.
5. 6.
7.
Weber, J.: The C a l c u l a t i o n o f t h e P r e s s u r e D i s t r i b u t i o n Over theSurface o f TwoDimensional and Swept Wings withSymmetricalAerofoilSections. A.R.C. Tech. Report R&M No. 2918, 1956. Coons, S.A. : Surfaces f o r ComputerAided Design o f Space Figures. Massachusetts I n s t i t u t e o f Techno1ogy , ESLMemorandum9442"139, January 1964. Peters, G.J.: I n t e r a c t i v e Computer GraphicsApplicationoftheBiCubic ParametricSurfacetoEngineering Design Problems. Paper presented t o 1973 National Meeting, Society of .Industrial and Applied Mathematics, June 1973. Hess, J.L. : Review o f I n t e g r a l  E q u a t i o n Techniques f o r S o l v i n g P o t e n t i a i Flow Problems w i t h Emphasis on theSurfaceSource Method. Computer Methods i n Applied Mechanics and Engineering,Vol 5, J u l y 1974.
8.
9.
10.
.
62
I
1 I
~
~___
"_
No. Report 1.
2. Government Accession No.
NASA CR2962
4. T i leandSub
.
~ ~ ~
3. Recipient's Catalog No.
A keometl; Package forGeneration o f Input Data . ; for a ThreeDimensional PotentialFlow Program
5. Report Date
June 1978
____________
8. Performin Or nization Report
No.
7. Authorb)
N. Douglas HalseyandJohn
9. Performing Organization Name and Address
L. Hess
McDonnell Douglas Corporation Douglas Aircraft Company Beach, Ca. 90846 3855 Lakewood Blvd., Long
12. Sponsoring Agency Name and Address
%
MDC J j 7 f i
Work Unit
._10.
No.
11. Contract
NASl 1 4402 or Grant No.
13. Tvpe of Rep rt and Period Covered
N S LangleyResearchCenter AA Hampton , Va. 23665
15. Supplementary Notes
I
James L. Thomas
.
Contracfor Report 5 7 76 8/22/77
14. Sponsoring Agency Code
. ~ . "

~
Langley Technical Monitor: Final Report
16. Abstract

.
  .. "
.
~
.
. . . ..

~

The preparationofgeometricdatafor i n p u t tothreedimensionalpotentialexpensive) flow programs i s a very tedious, timeconsuing(andtherefore task. This reportdescribes a geometry package t h a t automates and simplif i e st h i st a s kt o a largedegree.Inputtothe computer program f o r the geometrypackage consists o f a verysparse set o f coordinate data, often w i t h an order o f magnitude fewer pointsthan required for the actual potent i a l flowcalculations.Isolated components, such a s wings, fuselages, etc. a r e paneledautomatically,using one ofseveralpossibleelement distributionalgorithms. Curves ofintersection between components a r ec a l c u l a t e d , using a hybridcurvefit/surfacefitapproach.Finally,intersecting on e i t h e r side of the components are repaneled so thatadjacentelements intersection curves line u p i n a s a t i s f a c t o r y manner for the potentialflow calculations. The geometry package has been incorporatedinto the NASA Langley version of the 3D l i f t i n g p o t e n t i a l f l o w program and i t i s possible t o run many casescompletely(frominput, t h r o u g h the geometrypackage, and throughtheflowcalculations)withoutinterruption. Use of t h i s geometry package can significantly reduce the time and expense involvedin making threedimensionalpotentialflowcalculations.
17.Key
Words (Suggested by Author(s))
Aerodynamics Geometry Preparation ThreeDimensional Flow
19. Securitv Classif. lof'thisreport)
I APPROVED FOR PUBLIC RELEASE;
18. Distribution Statement
DISTRIBUTION UNLIMITED
I
20. Security Classif. [of this p a g e )
Unclassified Unclassified
'For sale by the National Technicd Information
Subject Category 02 I 21. of Pages J 22. Price' 66 $5.25
NO.
J
1
~~
Service. Springfield, Virginia 22161
NASALangley, 1978