ad-a112 371 oo state univ coj.umbus electroscience lab … · 2014. 9. 27. · ad-a112 371 oo state...
TRANSCRIPT
AD-A112 371 OO STATE UNIV COJ.UMBUS ELECTROSCIENCE LAB F/6 20/3HIGH FREQUENCY SCATTERING BY CURVED SURFACES.(U)PU 74 P H PATHAK, R J MARHEFKA. W D BURNSIDE N6226--72-C-0354
UNCLASSIFIED EM-3390-5 ,
7'IEEEllEE-lEEiiEiiEEE-EEEEEEEIIEEIIIIIIIIIImI
UIINl I- ~iI'l' _ 11111221111 1111 2 .
111111L25 LA1 4 _____
MICROCOPY RESOLUTION TEST CHART
PHOTOGRAPH THIS SHEET
-I LEVEL ow~ eJ.L
-l Erft rot ience LaL.
i-hb regUet~cT scatoeiwgv% lay Cwvvej £utl~lrr
DOCUMENT IDENTIFICATION
Approved for public release;Distribution Unlimited
DISTRIBUTION STATEMENT
ACCESSION FORNTIS GRA&Iuric TAB DTICUNANNOUNCED ELECTJUSTIFICATION MAR 2 4 1982
S Di, DISTRIBUTIONI
AVAIL'ABILITY CODES
DIST AVAIL AND/OR SPECIAL DATE ACCESSIONED
DISTRIBUTION STAMP
. 03 06 IV7DATE RECEIVED IN DTIC
PHOTOGRAPH THIS SHEET AND RETURN TO DTIC-DDA-2
DTIC FO DOCUMENT PROCESSING SHEETCToc , 70A
I
4
VI~,i
9
4
4
j
-t
t
I
HIGH FREQUENCY SCATTERING BY CURVED SURFACES
P. H. PathakR. J. MarhefkaW. D. Burnside
TECHNICAL REPORT 3390-5
June 1974
Contract N62269-72-C-0354
1 Naval Air Development CenterWarminster, Pa. 18974
I
F7 1
| .. NCI ASS IE I EDSECURITY CLASSIFICATION OF THIS PAGL (When Data Entered)
READ :NSTkUCTT-_'.:'REPORT DOCUMENTATION PAGE I3RE CILTIN CT I
R1. REPORT NUMBER 2. GOVT ACCESSION NO.1 3. RECIPIENT'S CATALOG NUMBER
4, TITLE (end Subtitle) 5. TYPE OF REPORT & PERIOD COVERED
HIGH FREQUENCY SCATTERING BY CURVED SURFACES Technical Report
6. PERFORMING ORG. REPORT NUMBER
-ESL 3390-57. AUTHOR(e) S. CONTRACT OR GRANT NUMBER(&)
P. H. Pathak Contract N62269-72-C-0354R. J. Marhefka
* W. D. Burnside9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT. TASK
The Ohio State University ElectroScience Lab AREA& WORK UNIT NUMBERS
Department of Electrical EngineeringColumbus, Ohio 43212
11. CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATE
Naval Air Development Center June 1974Warminster, Pa. 18974 Ia. NUMBER OF PAGES
6114. MONITORING AGENCY NAME & ADDRESS(Ji dilletent from Controlling Office) 15. SECURITY CLASS. (of this report)
Uncl ass i fiedISO. DECLASSI FICATION/DOWNGRADING
SCHEDULE
IS. DISTRIBUTION STATEMENT (of this Report)
DUTMIDUTJON STATEMENT A
Appioved for public releaseiDistribution Unlimited
i7. DISTRIBUTION STATEMENT (of the ebstract entered In Block 20, il different from Report)
IS. SUPPLEMENTARY NOTES
19. KEY WORDS (Continue on reverse side it necessary and identfy by block number)
Curved surface diffraction Wing mounted antennaHigh frequency analysis Fuselage scatteringTransition regionNear zone scattering
~ 20. ABSTRACT (Continue on reverese ide It necessary and Identify by block number)
This report deals with the geometrical theory of diffraction (GTD)solution to the problem of plane wave high frequency (h.f.) scattering bya smooth, perfectly-conducting cylindrical surface. According to the GTD,
1 the surface diffracted rays accurately describe the field behavior deep inthe shadow region behind the obstacle; whereas, the incident and reflectedrays adequately describe the field in the lit region sufficiently far fromthe optical shadow boundary. These ray solutions fail within the trans-ition region adjacent to the shadow boundary; hence, an approximate h.f.
DD FANm,, 1473 EDITION OF I NOV S Is OBSOLETE UNCLASSIFlF
I SECURITY CLASSIFICATION OF "tIS PAGE (Wh.n Date entered)
VIP - *
UNCI ASTIFTFDSECURITY CLASSIFICATION OF THIS PAGE(1('W4 D.%o Enteted)
I20. solution is developed for describing the field within the transition region.The usefulness and the accuracy of this transition region solution isillustrated by analyzing the h.f. plane wave scattering by a 2-0 cylindri-cally capped thick half plane, and a 2-D circular cylinder (for which anexact solution is available for comparison), respectively. The presenttransition region solution is in a form which facilitates its generalizationto treat 2-D, and 3-D convex surfaces of non-constant curvature. A simpleexample of this generalization is illustrated by analyzing the h.f. planewave scattering by a 3-0, hemispherically capped semi-infinite circular
tpcylinder. The work described in this report is important to the analysis ofthe problem of h.f. scattering by a finite elliptic cylinder excited by afinite source; this problem is directly relevant to a study of the effectof an aircraft fuselage on the performance of wing mounted antennas.
UNCLASSIFIEDSECURITY CLASSIFICATION Of THIS PAGE(befn Date Snhetd)
I7
ABSTRACT
This report deals with the geometrical theory of diffraction (GTD)solution to the problem of plane wave high frequency (h.f.) scattering bya smooth, perfectly-conducting cylindrical surface. According to the GTD,the surface diffracted rays accurately describe the field behavior deep inthe shadow region behind the obstacle; whereas, the incident and reflectedrays adequately describe the field in the lit region sufficiently far fromthe optical shadow boundary. These ray solutions fail within the trans-ition region adjacent to the shadow boundary; hence, an approximate h.f.solution is developed for describing the field within the transition region.The usefulness and the accuracy of this transition region solution isillustrated by analyzing the h.f. plane wave scattering by a 2-D cylindricallycapped thick half plane, and a 2-D circular cylinder (for which an exactsolution is available for comparison), respectively. The present transitionregion solution is in a form which facilitates its generalization to treat2-0, and 3-D convex surfaces of non-constant curvature. A simple exampleof this generalization is illustrated by analyzing the h.f. plane wavescattering by a 3-D, hemispherically capped semi-infinite circular cylinder.The work described in this report is important to the analysis of the problemof h.f. scattering by a finite elliptic cylinder excited by a finite source;this problem is directly relevant to a study of the effect of an aircraftfuselage on the performance of wing mounted antennas.
I
1| ii
I
1I
TABLE OF CONTENTS
Page
I. INTRODUCTION 1
If. HIGH FREQUENCY SCATTERING BY SOME SPECIAL2-D, SMOOTH, CYLINDRICAL SURFACES 2
A. Transition Region Solution for the H.F. PlaneWave Scattering by a Circular Cylinder 6
B. H.F. Scattering by a Perfectly-Conducting2-D, Rounded-End Thick-Screen 23
III. HIGH FREQUENCY SCATTERING BY A 3-D,
SMOOTH CONVEX SURFACE 39
IV. CONCLUSIONS 47
Appendix
I EVALUATION OF THE INTEGRALS FOR THEFIELD IN THE TRANSITION REGION 49
II KIRCHHOFF DIFFRACTION BY A HALF-PLANE 55
REFERENCES 60
iii
h A .. . .....
II
j I. INTRODUCTION
On-aircraft antenna pattern performance is very much dependent on
I ithe scattered field from the fuselage. This point was emphasized in the
analysis of fuselage mounted antennas[l]. For certain sectors of the
Ivolumetric pattern, the fuselage may even play a more dominant role forantennas mounted off the fuselage. This is especially true in the
' Jazimuth plane for wing mounted antennas. For many years scale model
measurements have been used to study these effects; however, such
measurements are both time consuming and expensive. On the other hand,41 numerical solutions have been successfully applied for the past few
years to study designs and locations for fuselage-mounted antennas.
i Fundamentally, the analyses developed for fuselage-mounted antennas can
* ibe extended to handle off-fuselage antennas.
Based on the numerical techniques previously developed[l,2,3], it
is proposed that similar solutions be investigated for wing-mounted
antennas. Specifically, the near-zone scattered field solutions for a
I finite flat plate and finite circular cylinder have already been
developed[l]. However, these models are not as versatile as desired.
Further the finite cylinder is limited to electrically small radii due
to use of a modal solution. Consequently, it is proposed that a high
frequency solution for sources in the near zone of a finite elliptic
I cylinder be employed. This model allows for variations in the cross-
section of the theoretical model and is not limited to lower frequencies,
i.e., to electrically small bodies.
1 iThe theoretical solutions necessary to analyze this problem are
developed in this report in terms of the solution to the scattering by a
two-dimensional circular cylinder. The geometric optics solution
satisfactorily handles the lit zone; whereas, the geometrical theory of
diffraction can be used in the deep shadow, However, both solutions fail
on and near the shadow boundary; hence, an approximate asymptotic high
II
frequency result is developed for analyzing the field within the transitionregion adjacent to the shadow boundary. Further, the solutions developed
herein can be extended to handle arbitrary curved surfaces including the
three-dimensional case. These solutions will provide the necessary
theoretical background to complete the study of a finite elliptic cylinderwhich will be treated in a future report.
The canonical problem that is treated involves the calculation of
the total high frequency field which is observed at a finite distance
from a convex, cylindrically shaped diffracting obstacle when it is
illuminated by an incident plane wave. This problem is directlyequivalent to the relevant problem of calculating the radiation pattern
of a source located at a finite distance from the same smooth, convexdiffracting obstacle via the reciprocity theorem for electromagnetic
fields.
II. HIGH FREQUENCY SCATTERING BY SOME SPECIAL2-D, SMOOTH, CYLINDRICAL SURFACES
This section deals with the high frequency scattering of electro-
magnetic waves by a smooth, two-dimensional perfectly-conducting cylindrical
surface. In this study, special emphasis is given to the region near theshadow boundary. A solution is desired which can be generalized (based on
high frequency approximations) to treat the scattering by three-dimensional
surfaces without having to solve additional canonical problems. The general
problem of the diffraction of waves by smooth, convex surfaces for grazingangles, (i.e., in the transition region adjacent to a shadow boundary) is
difficult to solve and it has been the subject of investigation by variousauthors[4] who considered some general cases; however, owing to the mathe-
matical complexity of their solutions, their results do not appear to be in
a form tractable for application to engineering problems of practical
interest. For the special case of high frequency plane wave diffraction by
a perfectly-conducting circular cylinder, Goriainov[5] has obtained results ii'
ii
I!
for grazing angles which are in a form suitable for numerical calculations.
Wait and Conda[6] extended Goriainov's high frequency results to treat the
plane wave diffraction by a circular cylinder with an impedance type
boundary condition; they also included the case of spherical wave incidence.
Most of the previous works on curved surface diffraction for grazing angles
Iincluding the work in References [5] and [6] make use of Fock integrals
introduced by Fock in his fundamental work on diffraction by smooth surfaces
I(many of Fock's papers are listed in Reference [4]). Wait and Conda[6]
indicate that their high frequency (h.f.) solution for the plane wave
I diffraction (for grazing angles) by a circular cylinder can be interpreted
as being the superposition of the result based on Kirchhoff diffraction
theory, and a second term which is viewed as a correction to the Kirchhoff
theory. This observation by Wait and Conda[6] constitutes the physical
basis for the development pursued in this report to arrive at a result
*, which could be simply extended to treat the h.f. scattering (in the
transition zone) by arbitrary 2-D and 3-D smooth, convex, conducting surfaces.
* IThe decomposition of the field scattered by the 2-0 cylinder into a sum of
a Kirchhoff diffraction term and a surface diffracted term (viewed as a
correction to Kirchhoff theory) as observed by Wait and Conda[6] was also
indicated by Fock (see pp. 194-196 of Reference [6]). In his work on the
3-D scattering problem of the Fresnel diffraction by a sphere for the special
case of source and observer raised to small heights above the sphere, Fock
expressed the scattered field in terms of a superposition of two integrals;
one of these integrals happened to be independent of the electrical properties
of the surface, and the other explicitly contained the surface properties
, Iand its associated electrical properties. Fock then showed that the integral
for the scattered field which is independent of the electrical properties
* I reduced to a Fresnel edge diffraction term; this term corresponds toI Kirchhoff diffraction. On the other hand, the integral which depended on
" the electrical properties of the surface was shown by Fock to yield the
_| surface diffracted term (which is viewed as a correction to Kirchhoff
diffraction). Consequently, it is believed that the basis of our analytical
j development given in this report is justified in view of the above remarks.
!3
Since the circular cylinder has two penumbral regions on its surface
as shown in Fig. 1, one needs to isolate the contribution to the field
y /"P (p,4)TRANSITION FIELDSREGION LIT / POINT
. (REGION
SHADOW __Pu
BOUNDARY
SHADOW REGION 0i
L . PENUMBRA
SSHADOW
BOUNDARY p
TRANSITIONREGION
Fig. 1. Geometry for the plane wave scatteringby a circular cylinder.
from each of these penumbral regions in order to allow an extension of
the circular cylinder analysis to treat the h.f. waves diffracted from
a single penumbral region, such as that which occurs on a smooth convex
surface of the nose of an aircraft when it is illuminated by a wing
mounted antenna. One notes that the angular extent of the penumbral
regions and the transition regions (which lie adjacent to the shadow
boundaries) is approximately 2(ka/2) 1 / 3 in radians (ka = circumference
of the cylinder in wavelengths). In Section II-A below, it is shown that
the h.f. field contributions arising from each of the two penumbral regions
on a circular cylinder can be identified, and separated in a manner whichenables one to calculate the h.f. field within a transition region which is
4
j . .... .
I
'~I
I associated with a single penumbral region on the cylinder. Such a decompo-
sition of the circular cylinder solution facilitates the analysis of the
problem of h.f. plane wave diffraction by the convex portion of the 2-0,
perfectly-conducting, rounded-end, thick-screen configuration of Fig. 2.
-: j TRANSITIONREGION
LIT REGION
SHADOW Iu
BOUNDARYP\ PENUMBRA
J2 0 d1
S HADO WI REGION
G O 00
Fig. 2. Geometry associated with the plane wave scatteringby a cylindrically tipped thick half-plane.
D One notes that the configuration of Fig. 2 consists of a half cylinder of
radius (a) which is smoothly joined to a thick half plane whose thickness
is 2a. The curve defining the boundary of the configuration in Fig. 2 is
not closed, and therefore there exists only one penumbral region on the
. j convex portion of this scattering geometry as illustrated in Fig. 2. The
results for the h.f. scattering by the canonical configuration of Fig. 2
j are given in Section II-B; it is conjectured that these results can be
extended in a systematic fashion to treat the h.f. scattering by arbitrary
j convex conducting surfaces if one employs the hypothesis valid at high
I5
I
II
frequencies, namely that h.f. diffraction and reflection are localphenomena (i.e., h.f. diffraction and reflection mechanisms are localized
to the immediate vicinity of certain points on the surface of the scatterer,
which correspond to points of diffraction and reflection, respectively). Anextension of the results in Section II-B for the 2-D canonical configuration
2 of Fig. 2 to analyze the problem of h.f. scattering by a 3-D, perfectly-
conducting hemi spherically capped, semi-infinite circular cylinder geometry
is indicated in Section III. The more general extension of the results of
Section II-B to handle arbitrary curved surfaces will be the topic of a
future investigation.
A. Transition Region Solution for the H.F. Plane WaveScattering by a Circular Cylinder
Let the total field, u in a region external to the 2-D perfectly-
conducting circular cylinder of Fig. 1 be the superposition of the incident
field (ui ) and the scattered field (us ). In the case of TMz excitation,the electric field associated with the incident plane wave (ul) is entirely
parallel to the axis of the cylinder (i.e., along z); furthermore, thetotal electric field denoted by u is also entirely z-directed. Similarly,
ifor the TEz excitation of the cylinder, u and u represent the incident
and total magnetic field intensities, respectively, which are entirely
z-directed. Let us denote the value of u for the TMz case, and let uh
denote the value of u for the TE case. Then, us = ui+s satisfies the 2-Dh h
homogeneous, reduced wave equation in the region exterior to the cylinder,
i.e.,
(1) (V2 + k2) us = O,h
subject to the boundary conditions;
(2a) us =0 on the cylinder,
or
6
I I73
I (2b) h - 0 on the cylinder.
In addition,
(3) us must satisfy the Sommerfeld radiation condition as p-p for anh
I eiWt time convention which is assumed and suppressed.
I The incident plane wave field is given by:
i =ikx(4) us e
h
The quantity k present in the above equations refers to the free space
S I wave number. Since Eq. (4) satisfies the reduced wave equation, it
follows from Eq. (1) that us must also satisfy the reduced wave equation.
IAn application of the 2-D Green's theorem to the region surrounding the
cylinder of Fig. 1 yields: ui 127T G 0o(PIP-) GoP p ) uP )
(5) u(P) = u (P) + joiG (PI'IP-) p =a ado'
where G (PIP') is the 2-D free-space Green's function:
1 (6) Go(P)P') - LH(2 ) (k- o').
0 ~ 4- o
' The parameter (-) denotes the field point position vector of P = P(po);
and P' denotes the cylindrical surface position vector of P' = P' ')
with p' = a on the cylinder. Clearly, the integral term on the RHS of-'Eq. (5) is the scattered field us . Then, from Eqs. (2), (3) and (5)
(7a) uS(p) = _ (P IP ') au ' 5 =aS 0 p
1 and
I (7b) uS(p) = 27r (P') .p,-ado',h o [hp=a
1 7
I
with the understanding that
i S(8a) us= u + u for the TMz case,) s u
and
(8b) uh = ui + us , for the TEz case.
It is well known[4,5] that a valid asymptotic h.f. approximation
for the surface field which is induced within the penumbral regions on
the surface of a circular cylinder, when it is illuminated by a plane wave,
is given by the following set of equations:
aua 2 -ikaio ,~ -ikaS](9a) ka - ( k e + J(4) e 2!L 'ii p'=a )=0 k L 2
and
SFA -ikal -ika2
(9b) FUhiP1 9 g(Cl) e I + g(92 ) e LL bhp'=a LOL
The functions which appear above are the universal Fock functions defined
by ;A(10a),(10b) e- d= gKz er1 i2 t 4-'r'J1 ~ d-r
in which the Fock-type Airy function w2(T), and its derivative w2(T) are
defined by:
_iaIb 1-E e" dz ; w; (.) -z e 3dz.(lla),(llb) w2 (i) - J r2 -~T r2 ~e z
The Fock parameters 41 and t2 are given by:
(12a),(12b) *1 = aZ/ 3 ; ; 2 = _ 2
8
II
with = - Tr/2 + 2i7r, and 2 = 37T/2 - + 2iri. The contours of
integration rI and r 2 for the integrals in Eqs. (lOa,lOb) and (lla,llb)
I are shown in Fig. 3.
V I- T OR ImZ
rI12 Rer OR ReZ
I
I Fig. 3. Contours of integration for the Fock and Airy functions.
The two terms on the RHS of Eq. (9a) and Eq. (9b) are commonly viewed
as two oppositely traveling creeping waves (associated with the surface
current) on the cylinder which have encircled the cylinder k times. If
I (ka)I/31r >> 1, the effect of the multiply encircling terms is negligible
and the inclusion of only the I=0 term is sufficient. In this report, one
is concerned with eventually applying the appropriate circular cylinder
results to the semi-infinite geometrical configuration of Fig. 2; hence,
it is meaningful to retain only the Z=0 term in the present analysis.
I Employing the far-zone approximation to G (PIP') leads to the3aG 0(PIP,) 0following results for G (PIP') and 5p,
[G (PP) i e- (kp -T/41 e ikacos( -O')(13) [Go(P P')]p':a 2 -
I9II
(14) [DG O(PIP'fl cosk 2 ' e- i(kp-ir/4) e ikacos(o-o')
The far-zone approximation implies that kip - pl' >> 1 and p >> (P'=a) in
Eqs. (7a,7b).
It is now possible to obtain asymptotic approximations to theintegrals in Eq. (7a) and Eq. (7b) for the scattered fields subject tothe restrictions indicated in Eqs. (13) and (14). Employing Eqs. (9a)and (13) in Eq. (7a), and expanding the exponential part of the integrandabout a stationary point allows one to write the integral in terms offunctions which are tabulated; thus, one is led to the following result for
< < (ka)-1/3.
(15) u5(P) e -icf2 ei~r/4 sik~-l +
2w/ ~ ka~l3 ~ -ika p -ika ijj
A similar asymptotic solution for the integral in Eq. (7b) is obtained byincorporating Eq. (9b) and Eq. (14) in it's integrand; thus,
(16) us(P) -eikp LF e' i/ 4 sinka(o-r) 1 +
+ e iwf/4 2 ( k a)1/3 *( ) e -ika * I + q* ( 92 e - i 1
The results indicated in Eqs. (15) and (16) were obtained earlier byGoriainov[5], and Wait and Conda[6). The universal functions p*(C) andq*(4) are the compl.ex conjugates of the p(4) and q(4) functions defined
by Logan[4,5,7) (p*(t) = G(t,q+-) e in/4 and q*(t) = G(4,q=O) eifl/ 4,
where G(4,q) has been introduced by Wait[6]). The functions e- -i n/4
* 10
and e- i / 4 q*(4) are plotted in Figs. 4a and 4b, respectively. The
I arguments 41 of the p*(41) and q*(41) functions are given by:2 2 2
* 0.8
0.6
1 0.4-
1 0.2-
1 §0
CL
1 -0.2-
* 1 -0.4
1 -0.6
-0.8 I-3 -2 1t 0 1 2
Fig. 4a. Plot of e ~/ *4 s4bsdoLogan's tabulated data[7] for p(t).
0.6
0.4
0.2
0 -
-1.01-0.4
-0.6 .1
-0.8 1
-3 -2 -I 0 I 2
-ilT/4Fig. 4b. Plot of ei q*(C) vs 4 based onLogan's tabulated data[7] for q(&).
(17a,17b) tl : ( a )1/3 1 ; = (_a)l/3 '2
with Tr1 - , and '2= -
The details of the asymptotic approximation to the integrals in Eqs. (7a)
and (7b) do not appear to be readily accessible in the literature, and are
therefore given in Appendix I for the sake of completeness.
It has been pointed out by Wait and Conda[6] that the first term on j
the RHS of Eqs. (15) and (16) corresponds to Kirchhoff diffraction;
whereas, the remaining terms may be viewed as a correction to the Kirchhoff
12
I
Itheory. The Kirchhoff result is independent of the electrical properties
of the surface; it is obtained by assuming that the field scattered
by the cylinder of Fig. I is equal to that scattered by a strip of the
same geometrical cross-section (i.e., strip width = diameter of the
cylinder) which is located at x=O as in Fig. 5a. It is possible to provide
j Ia ray interpretation for the scattered fields given in Eqs. (15) and (16);
the Kirchhoff term is composed of the field associated with the Kirchhoff
3'' Idiffracted rays which emanate from the edges P1 and P2 (of the strip) as
illustrated in Fig. 5a; whereas, the terms involving p*( I) and q*( l )
are the fields associated with the ray which is launched by the incident
I field at P1 and shed along the forwiard tangent at P3 after creeping adistance PIP on the cylinder (PIP 3 = a lPi ) as illustrated in Fig. 5b. A
I pseudo-ray interpretation[8] is possible for the terms p*(4 2 ) and q*( 2)
as illustrated in Fig. 5b, wherein the ray associated with these terms is
I launched at P2 ' and is shed along the backward tangent at P4 after creeping
a distance P2 P4 on the cylinder (P2P4 = aj 4 2 1 = aj-,); this ray system
unlike the ones described above, does not satisfy the generalized Fermat's* U principle which is employed in Keller's geometrical theory of diffraction[9],
*but it is nevertheless useful in visualizing the geometrical coordinatesI ikaV 2 -ikaV 2which are associated with tne terms p*(g 2 ) e and q*(2 ) e . In
Fig. 5a, u k(P) denotes the Kirchhoff edge diffracted field emanating from1I Pl; similarly, u4P) denotes the field arising from the edge P2 so that
SI (18) u (P) + u (P) -sinka(o-w) F2 iff/4 e - i k p((_8) U I (P + '(p for o Tr
I corresponds to the Kirchhoff result which is contained in Eq. (15) and Eq.
(16), respectively,
One is now able to identify the scattered fields which arise from
* 3 the two isolated penumbral regions shown in Fig. 1. Specifically, the
scattered field at P (in the close vicinity of *=r) arising from the
I penumbral region associated with P1 is:
1 13
_7 7
,-,-STRIP OF WIDTH 20
2 P2 (0,-0
(a)
Y
q*(Ca)
(b)
Fig. 5. Physical interpretation for the results in Eqs. (15) and (16).
14 71
di
IP' I
(Iga) u (P,P 1 ) u,(P) - e 1 4 2 (-a)1/ P_(,) e Ia e
I for the TMz case and o ;,
s( ki T -ikIa, ikp(19b) Uh(Pl) Ul(P) e k (-)1/3 q*(4i) e I e
Ji for the TEz case and o I
Similarly, the scattered field at P emanating from the penumbral region
I associated with P2 is:
-ika ikp! (20a) Us(P' ,(P 2) - * -- p 2 ) e ,
for the TMz case and 0
k i~r/4 J-2-ik ikP(20b) us(P,P 2 ) -k e-i/ 4 F(a)l1/3 q 2)e 2 e
for the TEz case and •
Re-writing the RHS of Eq. (18) for 0-", one obtains the following:
(21) sinka(4-l) e iTr/ 4 e-ikp e - ika(0--n)
e ika(0-iT) 1 e -i7/4 e- i kp
(0 (-7t) J Y_ VlPk
One is now able to associate the first term on the RHS of Eq. (21) with
u ; likewise, the second term on the RHS of Eq. (21) corresponds to U2 .1 e ika( -) ot
3 However, the individual terms ± (0_7), on the RHS of Eq. (21) are
unbounded at O=w (even though the LHS of Eq. (21) remains finite);
i consequently, they must be replaced by the bounded Kirchhoff result for
15I
.. . .--- II . ., = l_ ". , -_..:L-- .. . . ._ I" ,
.14
ku which is deduced from the result for the Kirchhoff diffraction by a
conducting half-plane. The u term is now viewed as being the h.f. field
diffracted from the edge P1 of a half-plane located at x=O for --<y<a.
Similarly, the u2 term is viewed as being the diffracted field from the
edge P2 of a half plane located at x=O for -ay<a. In doing so, one assumes
that the h.f. fields diffracted by the strip in Fig. 5a is the superposition
of the fields associated with rays diffracted from the edges P and P2 of
two overlapping half planes (with their common domain of overlap = width
of the strip); this assumption is reasonable since ka is assumed large. The
Kirchhoff result for the field diffracted by the edge P1 of a half-planewhen a plane wave is normally incident on it is given in terms of a KirchhotffJ
edge diffraction coefficient, Dk as
e-ik (p-asi n¢)k i e..kPs ~(22) u, (P) r ui(P 1) DR()
where ui(P) is unity, and
(23) DR(q) - e f 7 (- 2-) F(X)
with
(24a) F(X) = 2iJ ]vX-,e e-XT dT
and
(24b) X = 2kpcos .
ul(P) of Eq. (22) with Dk(¢) as in Eq. (23) approaches the following
result as O-f:
(25) Uk (P) e'ika(1") \F e-iir/4 e ikp F(X)
16
I
<Iwhich is identical to the first term on the RHS of Eq. (21) except for
3 the additional factor F(X).
I F(X) - (VX - 2Xe i T/ 4) ei(Tr/4 + X
as p-*1, thereby ensuring a bounded result for ul(P) of Eq. (22) at the shadow
I boundary corresponding to O=iT. The derivation of the result in Eq. (22)
is summarized in Appendix II. The function F(X) was introduced by
SHutchins and Kouyoumjian[lO] to describe the h.f. diffracted fields in
the transition regions of the shadow and reflection boundaries associated
with a perfectly conducting wedge illuminated by a plane wave. A plotkindicating the behavior of F(X) is given in Fig. 6. The field uk(P) is
similarly expressed in terms of D k as
(26) uk () D-ik(p+asin4)2(P)_ u'(2 Dk(2k )--
w " h " ±ikasin >which is also finite at *=%. The exponential phase factors e in
Eqs. (22) and (26) arise from choosing the origin as the phase reference.
The results in Eqs. (22) and (26) facilitate the calculation of the
Kirchhoff diffracted fields uk and uk over a greater angular region about
the shadow boundary than just within the immediate neighborhood of the
shadow boundary at 0=w. It can be shown by a careful limiting operation
that a superposition of Eqs. (22) and (26) also leads to the result on the
LHS of Eq. (21) for € ir. In this limiting operation, the fields in Eq. (22)
| and Eq. (26) are initially assumed to emanate from P1 and P2 ' respectively
to the far-zone field point P (in the vicinity of o=w) along nearly parallel
ray paths which converge at P. Next, the field corresponding to this nearly
parallel convergent ray system is allowed to approach the parallel ray
system (of Fig. 5a) in the mathematical limit. One may thus employ the
field representations of Eqs. (22) and (26) for the uk and uk terms present
in Eqs. (19) and (20), respectively, to calculate the Kirchhoff diffraction
contribution to the scattered field us over a larger ingular domain aboutShs
0=7r. However, before the results of Eqs. (19) and (20) for us can beh
1 17
-t (S3380930) 3SVHdiAO a 0 in 0 o
Ln * W ) V) N N -n -0 0
100
t
01
.18
~I
extended to calculate the scattered field over this larger angular domain
about =ir, one must modify the terms involving p*(4) and q*(4) (which are
viewed as corrections to Kirchhoff diffraction) in a manner corresponding
I to that employed for ul in Eqs. (22) and (26). From Eqs. (A-10) and (A-24)
of Appendix I: 2
(27a) e() = L e-r V(t) d + 1
I v' r ~ QW2 ) . 2V
where
(27b) p*( if the operator Q = I
2q*(t) if the operator Q =
and 4 = 1 or 42" The function QV(T) is directly related to the Miller
type Airy function as described in Appendix I. Clearly, the Krchhoff
correction terms in Eqs. (19) and (20) may be expressed in terms of
I '( ) in Eqs. (27a) and (27b) as:
(2a)-ill/4~ , (2IkadI/3 i ^-ika p e -
e -in4 e- (ka)/3 e-ika e -ikp e i w dr -
(28al- L T4b),
- (unbounded Kirchhoff result of Eq. (21) for
with t =l corresponding to = l as before. The unbounded Kirchhoff
I result for 2 Oi is seen from Eq. (21) to be:
I )eika(o-w) e- i/ 4 e-ikp(28 b ) t - ') p
I The integral over the contour r1 in Eq. (27a) is unbounded at 4=0 (which
corresponds to the shadow boundary at *=f), but (4 ) in Eqs. (27a) and
I (27b) (and therefore the result in Eq. (28a)) is bounded[4,7] because the
I 19
I
singularity in the 1-term of Eq. (27a) exactly cancels the singularity
in the integral (over r,) when 4-0. It is then obvious that the Kirchhoffcorrection terms in Eqs. (19) and (20) can be appropriately modified to be
useful over a larger angular domain about the shadow boundary (=r), if the
unbounded Kirchhoff result for * in the close vicinity of 7 in Eq. (28a) is
replaced by the unbounded Kirchhoff result valid over a larger angular
domain about 0=7. This is accomplished by noting from Eq. (22) that the
-!factor F(X) keeps D(4) bounded at *=ff, and:
Cbounded Kirchhoff unbounded Kirchhoff xdiffraction field) (diffraction field x F(X).
Thus, Eq. (28a) which is valid for *% is now modified by replacing
-ikal 2 e-i kp
e- iw/4 2 (ka)l/3 () e ek2
VP-
with
-i kaW1
(280 e- i1/4 F ka)l/3 2 e-ikp
rp2 2rff t2
+ e-/' 4 s±l e -ik(masinf)sine ~p
Combining the Kirchhoff diffraction results of Eq. (22) and Eq. (26)with the Kirchhoff correction terms of Eq. (28c) yield an expression forus(P) which is useful over a larger angular domain about *=i (in contrasthto Eq. (19) and Eq. (20) which are good only in the close vicinity of O=w):
20
I (29a) us (P ,P1 E .i ±r F(X1)-l] eik(asin9))2 2ikV
I ~-ika*i
*~ 2 e e1 I___L 2 - 2 -J v~
Iwith X1 2kp cos 4/2, and X = 2kp co (21-ff4)/2.
fu 5(PP1) corresponds to the scattered field at P arising from the* 2
penumbral region associated with Pior P2. Also
us(PP 1) when P*( ,), for the TM~ case
I(29b) u5(PP 1) = 222 u h(PP), when (1 q*( ,), for the TE z case.
* 12 2 2The total hM. field at a point P arising from the vicinity of P1 on the
I cylinder may be written as:
U) + us(PP 1) on the lit side (O<ff)(30a) u5 ( P) u5(PP ) on the shadow side ( >n)
*for the TM zcase,
and
(3b) Uh(PP ) rWul(P) + uS(PP) on the lit side (O<w)(30) h(,P) ~'~IuS(P'P) on the shadow side (0>7)
for the TE z case.
The total M. field at P arising from the vicinity of P2 on the cylinderis similarly expressed as.
21
U + uS(P,P2) on the lit side (0<7)" :i (30c) Us(PP2) U
2) us (P 'p on the shadow side (>i)
for the TMz case,
and
su(P) + u(PP 2) on the lit side (0<7)(30d) u (P ) + h Uh(P
h ,2 u (PP2) on the shadow side (¢>r).
for the TE case.
The total h.f. field at P in the transition region is the sum of the
fields arising from the vicinity of P1 and P2 on the cylinder, thus:
(31a) u U(P)=us(P,Pl) + Us(PP 2) for the TM case
and
(31b) Uh(P) = Uh(P,Pl) + Uh(P,P2) for the TEz case.
The results of Eqs. (31a) and (31b) will be employed in part B which
follows, to calculate the fields in the transition region associated
with the canonical problem of Fig. 2. One notes that the results in
Eqs. (29a) and (29b) are based on a direct generalization of the results
in Eqs. (19) and (20). The accuracy of these results will be indicated in
the next section wherein they are applied to specific problems. Possibly,
a more accurate asymptotic h.f. result for uS(P,Pl) (when P is not2
necessarily in the close vicinity of 0=ir shadow boundary) obtained from
first principles might involve terms that are not a simple superposition
of the effective edge diffracted field term, and a surface dependent field
term as in Eq. (29a); such an asymptotic solution based on mathematical
rigor rather than on simplifying physical assumptions to bypass mathematical
complexities appears to be a difficult task. However, judging from the
22
numerical results obtained from the use of Eq. (29a) in the following
sections, it appears that the present solution yields sufficiently accurate
values for engineering applications.
B. H.F. Scattering b a Perfectly-Conducting2-D, Rounded-End hick-Screen
The canonical configuration of interest is illustrated in Fig. 2.
I It is desired to calculate the total h.f. field surrounding this 2-D
rounded-end, thick-screen configuration when it is excited by a normally
I incident electromagnetic plane wave of the TM or the TE type. Thesolution to this canonical problem is of relevance to a study of the
I effects of the aircraft fuselage on the radiation from wing mounted antennas,
U as indicated previously. Figure 2 illustrates the junctions J and J2
resulting from the smooth join of the half-cylinder onto the thick half
plane. In the present analysis, the diffraction effects arising from the
junctions J and J2 are not included for the following reasons:11) The actual fuselage of an aircraft will be approximated by
I this model for which these junctions will only be artificially
added to our model in that they are not present in the actual
* I aircraft structure.
2) The junction effects are important only if the penumbral
I region lies in the neighborhood of the junction, or if the field
point lies in the neighborhood of the reflection boundary associated
I with the junction. It is clear therefore, that in the present
problem, which deals with normal incidence and with aspects within
I and near the transition region, the junction effects are negligible.
" IThe total h.f. field in the lit region for the problem in Fig. 2 is*adequately described by the fields associated with the incident and
reflected rays of geometrical optics, whereas, the h.f. field in the shadow
123
I
region is accurately described by the surface diffracted rays of Keller's
geometrical theory of diffraction[9] (hence by the GTD). However, the
geometrical optics field description fails near and on the reflection
boundary (which is also the shadow boundary in this case). On the other
hand the GTD field description requires an increasing number of surface ray
modes to accurately describe the field near the shadow boundary thereby
making the GTD result cumbersome to employ. Thus, within and near the
transition region adjacent to the shadow boundary, one may employ the
result developed in part A to calculate the total h.f. fields there. The
specific results for the lit, transition, and shadow regions associated
with the canonical configuration of Fig. 2 are indicated below.
Field Description in the Lit Region - The total field in the lit
region is a superposition of the incident field u and the reflected field
ur predicted by geometrical optics, in which
(32a) ui = us eikX, for TMz case,h TEz
and,
(32b) u r = Ur = u (PR) R Pr e-iks
h h hr
where the subscript s is used for the TMz case, and the subscript h is
used for the TE case as before. The result in Eq. (32b) is in a ray-
optical form[ll] which is valid in the near zone. PR denotes the pointiTof reflection in Fig. 7a, and u (PR) is the value of the incident (TZ)
hzfield at PR* The reflection coefficient, Rs is defined by:
h(32c) Rs = - 1, for the TMz case,
(32d) Rh = + 1, for the TEz case.
24
_________________________________________ I
-I
II
0 U'
I
iS I Fig. 7a. Reflected ray geometry.
- s denotes the distance from R to the field point P in the lit zone. The*I quantity Pr within the radical in Eq. (32b) refers to the caustic distance(in the plane of reflection) associated with the reflected wavefront; r
• I is found to be:
a cos 0 0I (32e) Pr - 2The total field at a point P in the lit zone is thus given by:
u3(p) = a cos° -iks TMz
(32f) hsP u~)s us(PR) ( al) .acos o + 2s ; for TEz case.h h h 0I
25
I
If far zone results are desired, one must enforce the condition s >> P
in Eq. (32b). The result in Eq. (32f) is valid in the lit zone for aspects
sufficiently far from the reflection boundary (which occurs when PR coincides
with P) "
:* Field Description in the Transition Region - The total field in the
transition region may be immediately written down from Eqs. (30a) and (30b)
given in Part A. It is conjectured here that the fields in the transition
region of Fig. 2 are identical to those in Eqs. (30a) and (30b) which
arise from the penumbral region associated with P1 on the circular cylinder
of Fig. 1. This conjecture is based on the h.f. approximation which allows
one to consider h.f. diffraction as being a local phenomenon. The
results in Eqs. (30a) and (30b) for the TMz and the TEz cases, respectively,
are valid in the far zone; however, since h.f. diffraction is a local
phenomenon, one may also directly modify the far-zone results of Eqs.
(30a) and (30b) to obtain the following near zone transition fields at P:
(33) Us (P)--us (P-( )e sina 2:_ F •
h h "d
e-ika' e ie/4 ] Liks)l
with the understanding that '( ) P*(R) for u s(P) and (t) = q*( ) for
Uh(P); also H(ir-a) is the Heaviside step function whose value is unity if
c<ir and zero if a>w. The near zone result of Eq. (33) is obtained by
simply replacing the far zone ray distances by their corresponding near
zone ray distances. The quantities s, , and in Eq. (33) are given in
terms of the near field coordinates by:
)so in the lit portion of the transition region(34a) s in the shadowed portion of the transition region
26 I
. . . . ... . . . . . . . . . . . { , , -' -i I ...... .. . ..... . . ...L " ( ll l I]
iI
I (ka I/3 l-pp , in the lit portion of
(34b),(34c) - 2 ' the transition region.2 i s , in the shadowed part of
V the transition region.
where the angles @, and s and the distances s. and ss are indicated in
l jFig. 7b below. The F(X) function appearing in Eq. (33) is defined earlier
J P(LIT)
tI
a j
Fig. 7b. Equivalent ray geometries for the fields in the lit andl shadowed portions of the transition region associated :
IU
with the shadow boundary at P1.
l in Eq. (24a). The new argument (for near zone calculations) of F(X)
g is clearly:
(35a) X = 2kd cos2 ~
g 27
I
where a is the angle (measured in the cw sense) subtended by i- Kirchhoff
diffracted ray (directed along PP) and the negative of the incident ray
direction (i.e., along + x).
ig in the lit part(35b) = in the shadowed part.
The distance d is measured from P1 to P and
d in the lit partL4(35c) d
)d = in the shadowed part.
The quantities a., as, d and ds are illustrated in Fig. 7b.
The transition region solution discussed above must serve a two-
fold objective, namely that it should provide a sufficiently accurate h.f.
field description within the transition region; secondly, it should blend
smoothly with the geometrical optics solution in the lit region, and also
blend smoothly with the GTD solution in the shadow region. Some numerical
calculations which indicate the accuracy of the present results are provided
later.
Field Description in the Shadow Region - For aspects in the deep
shadow region, the scattered field (the incident field is zero within the
shadow region) is given a ray optical representation valid in the near
zone. The result based on GTD is given as[ll]:
Sh
h s .pa.i ka s s iks(36) u(P) (P) u(P ( e p Dh(P e
h h = h
~sS h a -iks TMi. S -pks'iasDh P
)e ,fr z
= h (Pl ) e P Ss) Dh(P , for TE case,
P=D z
28
I
since us(Pl) = 1. The geometrical details are given in Fig. 7c for the
3 h
, I ui
P, u
1 'P
I
Fig. 7c. Ray geometry for the GTD fields.
I surface ray modes which traverse a path PIPDP to the field point. Even
though the inclusion of an infinite number of GTD surface ray modes is
indicated in Eq. (36), it is generally found that sufficiently accurate
results for the field are obtained by summing only the first five modes
(corresponding to p=l to p=5), provided the field point P is far-removed
from the transition region. D and h are the cylinder diffraction coef-
ficient and attenuation constant for the p-th surface ray mode, respectively.
I D.R. Voltmer (in his Ph.D dissertation, Ohio State University, 1970) has
obtained diffraction coefficients and attenuation constants to orderI 2 ) 2/3 (here Pg = local radius of curvature of the surface along the ray
path) which are valid for smooth convex surfaces of non-constant curvature
j (this includes the circular cylinder, and the sphere as special cases); a
1 29
I
table of his diffraction coefficients and attenuation constants is, also,
available in Reference [8]. The superscripts s and h on the diffraction
coefficient and attenuation constant denote the TM and the TE cases,z Z
respectively.
Numerical Results - The numerical results for the near field patterns
of a circular cylindrically tipped thick half plane (as shown in Fig. 2)
illuminated by a TMz (acoustic soft) or TE (acoustic hard) plane wave have
been calculated via the formulas developed above. These results are
illustrated in Figs. 8 through 10. The geometrical optics field shown
as a dot-dashed line, consists of the incident and reflected fields up to
the shadow boundary. The transition field, shown as a solid line, starts
at 0=900 and is shown into the deep shadow region. The surface ray mode
field (obtained by summing over only the first 5 modes) is shown as a
dotted line and is valid in the deep shadow region. Note that the
geometrical optics field and the transition field blend well in the angular
range 900<o<1200 . The transition field and the surface ray mode field for
the hard cases blend in the neighborhood of 0=180°. The soft cases do not
blend as quickly. The phase for the above near field patterns also blended
well; however, it is difficult to illustrate and will not be shown here.
The validity of the above high frequency solutions is demonstrated
by applying them to the problem of plane wave scattering by a circular
cylinder (as shown in Fig. 1). The results are then compared with the
exact solution for the problem which is found from the eigenfunction
I If expansion in Figs. lla and llb. The high frequency solution is composed
of the superposition of fields from the two penumbral regions associated
with P1 and P2 using the appropriate results (of part B) in their
respective regions of validity. The good agreement lends confidence to
the present high frequency solutions.
30
GEOM. OPTICSK I-TRANSITION
000'SURFACE RAY MODES
>110
13
--- GEOM. OPTICS
-TRANSIT IONo.e.eSURFACE RAY MODES
.......
fa.
* Fig. 8b. Near field pattern for cylindrically tippedthick half-plane, soft case, ka=lO, kp=1OO.
32
7I7
--- GEOM. OPTICSI - TRANSITION
o o 9 SURFACE RAY MODES
V 10
Iw* 0
0- I2Iu
'I>_j-3
I Fig. 9a. Near field pattern for cylindrically tippedthick half-plane, hard case, ka=2O, kp=5Q.
*33
--- EOM. OPTICS-TRANSITION
s .. SURFACE RAY MODES
-1
Iw40
-t3
Fig. 9b. Near field pattern for cylindrically tippedthick half-plane, soft case, ka=20, kp=50.
34
--- GEOM. OPTICSI~- TRANSITION0000*SURFACE RAY MODES
V IV
.1101
-2
-3iI3
thick half-plane, hard case, ka=40, kp=100.
1 35
--- GEOM. OPTICS-TRANSITION
e SURFACE RAY MODES
Fig. l0b. Near field pattern for cylindrically tippedthick half-plane, soft case, ka=40, kp=1OO.
36
*900 EXACT SOLUTION
- - GTO SOLUTION
-1
I18000I 5 X FIELD POINT
IY0 TE INCIDENT
2 tx PLANE WAVE
Fig. Ila. Comparison of high frequency solution with* exact solution for near field pattern ofI circular cylinder, hard case, ka=20, kp=5O.
I3
90 EXACT SOLUTION-- -GTD SOLUTION
CID
50X
TM INCIDENTL'JAY4/71 X PLANE WAVE
Fig. llb. Comparison of high frequency solution withexact solution for near field pattern ofcircular cylinder, soft case, ka=20, kp=50.
38
III. HIGH FREQUENCY SCATTERING BY A 3-D,I SMOOTH CONVEX SURFACE
As developed in the previous section the transition solution can be
divided into two terms. These are the Kirchhoff term, which is independent
of the polarization and curved surface properties, and a second term which
is dependent on the surface curvature and associated electrical properties.
The usefullness of this solution is that each of these terms can be system-
I atically modified to three-dimensional surfaces (based on the high frequency
assumption that diffraction is a local phenomenon). One notes that the
geometrical optics and geometrical theory of diffraction solutions are
well-known for the three-dimensional case as presented in Reference [11].
S I In this section, the near zone scattering by a spherically capped
semi-infinite circular cylinder when excited by a normally incident plane
wave is examined as shown in Fig. 12. The object of this study is to
illustrate the three-dimensional nature of our high frequency solution and
to verify that it blends nicely with the geometrical optics field. The
field surrounding the configuration in Fig. 12 is given below in the lit,
I transition, and shadow regions in a manner similar to that indicated for
the canonical configuration of Fig. 2, in part B of the previous section.
3 Once again, the diffraction effects arising from the junction of the hemi-
1spherical cap and the cylinder are ignored for reasons indicated previously.
-I Field Description in the Lit Region - The geometrical optics solution
composed of the incident field (ul) and the reflected field (ur) observed
at a near zone field point P in the lit zone is explicitly given by:
(37) (P) i +u(PRs a2cSo e- iks TMz
(37) Us( R) (acoso +2s)(a+2scOSo forTE case.h h h h 0 0 z
The point of reflection PR and other geometrical parameters (such as a, s,
c'0 etc.) are the same as in Fig. 7a. The reflection coefficient Rs wash
39 1.II
! P( LIT)
o S*
_W U'
P (SHADOW)
Fig. 12. Geometrical configuration for the hemisphericallycapped semi-infinite circular cylinder.
previously defined in Eqs. (32c) and (32d). us(PR) denotes the value of
u at the point of reflection PR on the hemispherical cap.h
Field Description in the Transition Region - The total field in the
transition region for the 3-D geometry of Fig. 12 can be obtained by
generalizing the 2-D results for the geometry in Fig. 2 which are indi-
cated in Eq. (33). This generalization from 2-0 to 3-D is based on the
ray-optic description for the fields; analogous generalizations have been
previously indicated elsewhere[8,9,11,12] for certain antenna and scattering
problems. We obtain
40 .]
II
I (38a) u5 P%~PHi ~ + eiir/4 [F(X)-i] a e- ikd)
(F ('42(ka) 1/3 eika e- -ff4FW - I -iks f(s))
I where H(-c), '( ), F(X), a, s, d and i have the same meaning as in Eq.
(33). The quantity f(s) is given by:
b - in the illuminated region (a<)I (38b) f(s): . -
FdLn C 0 in the shadow region (ci'r).
The geometrical details are the same as those indicated in Fig. 7b. The
quantity "o is the spatial spread of the field on the surface of the
hemispherical cap along the ray path P1Ps of Fig. 7b. In particular, dnois the width of two adjacent rays at Pl; whereas, dn i the width of the
same two adjacent rays evaluated at P. The quantity is given by:
(38c) sinIe in the lit penumbra region (=i/2 - 14,,J) i.e., for 6<1
where 0 l{1tit h shdwpnmrwr e h in the shadow penumbra region (=-/2 + J*s1) i.e., for 6>1
Also, the caustic distance p appearing in Eq. (38a) is the radius of
curvature of the diffracted wavefront in the plane_L x-y plane, but contain-
ing the ray shed from the surface; it is given by
(38d) p = atane.I_Incorporating Eqs. (38c) and (38d) in of Eq. (38b) leads
to '~dn s P+s| to
1 41
I
1
(38e) s (as - ne+scosOe
Thus, p>O when e=elit, whereas p<O for e=esh. If
-F+s) 0
then
sP+s) ei/
because a caustic has been crossed. One notes that FLdn has been replacedby unity in the lit part of the transition region (see Eq. (38b)) because
it is conjectured that the fields scattered into the lit portion should
be independent of the spatial spread factor of the rays along PIPL on the
surface (this fact is directly evident in the reflected ray description for
the lit region).
In order to keep the Kirchhoff diffracted term bounded one notes
that the X in F(X) of Eq. (38a) should be
=2k [~P+d)](38f) (e2d) cos 2L/2.
L ~e J
The quantity pe is the caustic distance for the Kirchhoff diffracted ray
in the plane containing the diffracted ray and the edge of the curved
screen at Pl; it is computed easily via the formula for caustic distances
for curved edges given by Kouyoumjian[12].
Field Description in the Shadow Region - The iffracted ray field with-
in the shadow region at a field poin + P in the near zone is obtained by
extending the 2-D result of Eq. (36) to the 3-D result in the usual manner[ll].
42
6L
Iss haS k S d
h S- d 0A5 (39) us(P),,tj(P) 0 ~P)hP)P j ijn is .oPeis3 h h ~~h p n
H TMzIfor TE case
Sz
s h a' s-h - s ha -iks
where u I(P I,
h
and dn r- a
dln -ST-P- ') s (asi ne+scose)
as noted previously in Eq. (38e). The diffraction coefficient Dpand theh
ateuto constant ah are indicated in Reference [8] for the sphericalp
surface. The geometrical details pertaining to the GTD ray description ofI Eq. (39) for us(P) are the same as those in Fig. 7c. If
1 h
sP~s < 0,STTS
I then 0 p
1must be interpreted as
due to the crossing of a caustic.
43
It is evident from the results in Eqs. (38a), (38b), and (39) that
a caustic of the diffracted rays is present for =Tr (and a=-sin - ain
Eq. (38a)). One notes that on crossing the caustic which occurs at O=w
(i.e., along the negative x-axis),two additional diffracted rays which
creep around either side of the cylindrical surface also contribute to the
field at P; the field of these rays is easily calculated by GTD. The
caustic field may be assumed to result from the radiation by equivalent
half-ring electric and magnetic currents in which the half-ring is formed
by the locus of the points of diffraction of the rays shed from the
hemispherical cap to the field point P for which O=w. The equivalent ring
current concept[8,14,15,16] proves to be a useful tool for evaluatingthe proper field behavior at and in the vicinity of a caustic.* Such an
alternative field representation for the caustic region is necessary
because the simple field description of Eqs. (38ab) and (39) fails in
this region (the results of Eqs. (38a) and (39) become infinite at the
caustic). The transition from the single diffracted ray arriving at P
in the shadow region (whose field is given by Eq. (39)) before crossing the
caustic to the three diffracted rays which arrive at P in the shadow region
after crossing the caustic (the three rays result from the ray corresponding
to Eq. (39) shed from the hemispherical cap to P plus the two additional
rays which creep around either side of the cylindrical surface before
being shed towards P) will not be treated in the present report. This
transition is described by the solution obtained from the equivalent
current concept for the caustic region. The caustic field analysis will
be completed in a future report.
Numerical Results - The near field patterns for the hemispherically
capped cylinder of Fig. 12 for a radius of ka-lO and observation distance
of kp=lO0 are presented in Figs. 13a and 13b, when this structure is
illuminated by a TEz or TM type plane wave, respectively. The variousz zfields are represented as done previously for the 2-D problem. As can be
seen the transition solution blends well with the geometrical optics fieldfor both the hard and soft cases. The presence of the caustic at *=w,
*A study of the caustic due to the curved screen Kirchhoff diffraction term(i.e., for a=sin-1 a/d in Eq. (38a) when O=w) will also be completed in afuture report.
44
GEOM. OPTICS
-TRANSITION
e e *.. SURFACE RAY MODES
510
I Fig. 13a. Near field pattern of hemispherically cappedcylinder, hard case, ka=lO, kp=1OO.
1 45
__GEOM. OPTICSTRANJSIT ION
**@SURFACE RAY MODES
t-t
IC
iw
Fig.13b Nea fild ptten ofhemspheicaly cppecyidesotcae k~Ok0lQ
46
20U
however, interferes with the blending of the transition solution with the
I field of the surface ray modes. As stated above, the singularity due to
the caustic can be easily corrected; however, this has not been done here
I but will be completed in a future report. When this caustic correction is
incorporated, it is expected that the blending of the transition solution
with the surface ray GTD solution will be as good as that indicated earlier
for the 2-D problem.
Ii IV. CONCLUSIONS
The object of this report has been to develop the complete solutions
I for the fields scattered by curved surfaces illuminated by near zone
antennas. Specifically the high frequency solutions for curved surfaces
*have been completed for several years except for a transition solution
which is needed around the shadow boundaries. An approximate transition
g region asymptotic solution is presented and developed in such a form that
| it can be extended to arbitrary curved surfaces. Combining this solution
with the geometrical optics term in the lit region, and the geometrical
1theory of diffraction term in the shadow region, should provide a completehigh frequency solution for the scattered fields of antennas illuminating
curved surfaces in the near zone.
j The solution is presented in terms of the thick half plane with
a cylindrical cap showing how the various terms blend together in each
region. The transition region solution developed here (Eqs. (33) and
(38a)) can be shown to analytically reduce to the GTD surface ray mode
solution as F1X)iI in th6 shadow region sufficiently far from the shadow
I boundary. de have not yet shown that this transition solution can
analytically reduce to the geometrical optics reflected field as F(X)4I
j in the illuminated region sufficiently far from the shadow boundary.
However, the numerical results presented indicate that the transition
j region solution does indeed blend very nicely with the GTD and the
I47
geometrical optics solutions. Although the present transition region
solution appears to be fairly accurate, it would be desirable to obtain
a transition region solution from a more rigorous asymptotic analysis of
the canonical circular cylinder scattering problem.
Even though the solution presented describes the one term as being
a Kirchhoff solution, one can, also, show that this term can result
using the GTD edge diffraction coefficient instead of the Kirchhoff edge
diffraction coefficient. Using this approach the values of the transition
solution do not change much near the shadow region, but they might provide
improvements away from the shadow boundary. This approach will be investi-
gated for the spherically capped cylinder in the caustic region in that
in this sector the Kirchhoff and GTD solutions are observably different.
While the transition solution is being completed, the solution for
a finite elliptic cylinder will be developed. This solution will provide
a versatile model for the fuselage when the antenna is not mounted directly
on it, i.e., such as wing or pod mounted antennas. It can, also, be used
to examine the scattering effects of stores, etc. With the completion of
this solution, one should have all the necessary theoretical tools to
analyze any high frequency on-aircraft antenna pattern using a reasonable
model representation of the actual aircraft.
48
-- i
S I
IAPPENDIX I
EVALUATION OF THE INTEGRALS FOR THEFIELD IN THE TRANSITION REGIONl
* A summary of the details concerning the asymptotic evaluation of
l the integrals in Eqs. (7a) and (7b) of Section II is provided in this
appendix. The TM case (correspondinq to Eq. (7a)) is considered first;* z
it is followed by the TE case (of Eq. (7b)).
TM case: Employing Eqs. (9a) and (13) in Eq. (7a) leads to:
1~~2ti F2 -i(kp- kao(-'
l S~) 44 4)ikacsL-
I (A-1) us(P) e 0 L- ka 2 =
2 -a/3i4 -ika- k1 e + )e ado'.
Let
) Go r ) -ika~l+ikacos(o-o')- (A-2) II - J og{led'
=0
(A-3) 2 e ika 2+ikacos(o-o '). (a-3) 12 -- " (92)e d'
then Eq. (A-l) may be written as:
1i 2 -i(kp - ) (2 -/ 2
- (A-4) u5(P) - e- NPe 2) [1+ I 2
with the understanding that only the t=O term is of interest for the reason
{ I mentioned in Section II. The integral I,, of Eq. (A-2) is evaluated first.
-One begins by incorporating Eq. (lOa) in Eq. (A-2); thus:
I_ 49
~I
I
-. I
27r -ilT -ika(o'- 2-)+ikacos(o-o')(A-5) 1 ido' di(T ) e- 1 d =0 0 r 2
with C (r )a . The sum on X can be incorporated in the 0'
integration if one changes the limits of integration to --xo'<-, since the
stationary points occur at *'=€-(4L-l)i/2-z'. Expanding the exponentialeika[(4'-w/2)-cos( ¢- € ')] in Eq. (A-5), about the stationary point 4'ecorresponding to Z=O only (since multiple encirclements are neglected), and
retaining terms only to the third power in (0' ) one obtains:
_ka/3( -T/2).r -ika 3
11 L d Jr di wF eika(o-'r)e T6 -Os2 Ior
ka )1/3
(A-6) Il dT W
2
after re-arranging terms, and interchanging the orders of integration.
Introducing a change of variables:
(A-7) (T )I13 ,- s ) y,
allows one to simplify Eq. (A-6) to:
ka 133
(A-8) 11 eika(') Jrd () e ka-1/3 3- +3
Sw2 dy e
50
i
One notes that the integral on y yields the Miller type Airy function[l3]
I denoted by AiCTr); specifically:
3
(A--- )5 dy e (r+ 2VITAi (T 2 V (-)
I Defining the function p*(&) introduced in Sectirn II, namely[4]:
( A-10) P* (4) = 1 f~ ~ T 1r2
one ay ow ewrie E. (-8)in terms of Eq. (A-9), and Eq. (A-la) as:
I(A-11) I~ 2e 2/ir
Iwhere P2
I (A-12) 2l-(al3~
A similar analysis for 1 2 in Eq. (A-3) leads to:
(A-13) 1 2 ' 2eik(P~~ Tr )/ [P*('2)-I 2V r 2
where
I (A-14) C a
IIncorporating Eqs. (A-11) and (A-13) in Eq. (A-4) leads finally to:
* I (A-15) u s(P) k ~ e~/ ikC +
+ elW ?Z ~f(La)l/3p*( l + p*(C)
I 51
where I and *2 are defined in Eqs. (17a) and (17b). The result inEq. (A-15) corresponds to that indicated in Eq. (15).
TEz case: Employing Eq. (9b) and Eq. (14) in Eq. (7b) leads to:
27r k -i (kp- 14- ekaos4,-o,(A-16) uS(P) % a d*' \ cos(o-+') e e
0
(4 1) e + g(i2) ek=0 J
Let:
r r2; -ika~ip+ikacos( -')(A-17) 2 - = o g(4, ) e cos(O-O')dO';
(A-8)2 e -ky kcsO )cos (O-O')dO',
then Eq. (A-16) may be written as:
(A-19) uo(P) 7 (e) e ) + 2.
Following the procedure for the TMz case, one may write 1l in terms ofthe stationary point at *s=*-w/2 (corresponding to the L=O term) as:
(A-20) ?l g(Cl )W€"s)eik ( do-i - € € )
1 R i
where cos(€-€') in Eq. (A-17) has been approximated by ( '-,s) in theneighborhood of the stationary point, and the limits of integration have been
changed to #' I<" as done for the TMz case. Incorporating Eq. (lOb) intoEq. (A-20) gives:
52
• Jli
t II
(ka)1/3 iiT )
-i -) 3_)i ka (Or)_i a 0 )3(A-21) J, l_ " do'f dr e ,, • (o' 's)e-s
I e-ika(o-'r 1) r e
f' dT ' W
k, .3
" 3 Introducing Eq. (A-7) into Eq. (A-21) yields:
_ ka 1/3 ]3
(A-22) d, w (e)-2/3_ dyye
I One recognizes the integral on y to be related to the derivative of theAiry function Ai(r); namely[13]:
(A-23 1 dy~ - i ( - Y+ Y )
(A-23) -dyye 3 a 2iVT Ai '(t) = 2i V'(r).
j One also defines the function q*(C) introduced in Section II; namely[4]:
(A-24) q*( ) =-L ei4- dlw T)_ + 1
Il / 2- 24
j Incorporating Eqs. (A-23) and (A-24) in Eq. (A-22) finally gives:
i (A-25) ?l eika(- ) 2 i (.a- 2/3 E*(1) - 1
I with tl defined in Eq. (A-12). A similar analysis for 12 leads to:
I (A-26) 1'2 eik( .~v2i (ka)-2/33[*(42)- __
1 53I
with t defined in Eq. (A-14). Employing Eqs. (A-25) and (A-26) in Eq.
(A-19) gives
e' C i{r/4 sinka(O-ir)(A-27) us(P) --ep L{ e( i+
+ h ) -ikaip -ikai 2 }l+ e-i,/4 F k ()"'3 *(, e 1+ q*(C2) e
2 2'
with *1 defined in Eq. (17a) and Eq. (17b). The result in Eq. (A-27)2
corresponds to the desired result indicated in Eq. (16).
54 li
APPENDIX II
I KIRCHHOFF DIFFRACTION BY A HALF-PLANE
A plane wave field u is incident on the half-plane at y=O for
* 1x>O as indicated in Fig. A-I. In this appendix, the expression for the
I';'i( p, 4
.I o y
I HALF -PLANEU K,
IFig. A-i. Half-plane geometry.
: " Ifield diffracted by the edge (at 0) of the half-plane is derived by using
a Kirchhoff approximation. Only the significant details of the analysis
are indicated. It may be verified that an application of the 2-D Green's
theorem to the region surrounding the half-plane leads to the followingj integral equation for the electric surface current density, Js induced on
the half plane when a plane wave consisting of a z-directed electric field
I (TMz case) is incident on it[17l:U
(A-28) u(P)-- - kj G(x,y0x',O) Js (x',o) dx'
" 55
l I
where G is the 2-D free space Green's function, and
0+
i wjs(Xt ,O) au (x',O) au (x',0)
(Note: surface current density Js is z-directed.) u(xy) represents thescattered z-directed electric field surrounding the half-plane. The geo-metrical optics approximation allows one to let
-1, sine eikx'cos¢'1 W1
and
-1 au (x',0) .
Employing this approximation in Eq. (A-28) gives:
(A-29) u(P) % - iw Jo G°(x'yjx"O0) oe i x c s dx'
G 2iksin¢ J G(x,yx',O) eikx 'c os¢ ' dx'.00
(Note: k = wv free space wave number, and Z° = free space wave
impedance = iVp-/.) The representation for G chosen here is differentfrom that used in Reference [17] for solving the same problem (however,
the various alternative representations for G0 may be obtained via the
use of the techniques for construction of Green's functions which are also
given in Reference [17]); hence, the method for solving Eq. (A-29) differs
from that in Reference [17]. In the present analysis, the followingexpression is employed for Go:
ikxXX' 1-ik k2_k2 ly-y' I(A-30) Go e dk ewa 0
oU W1-K 2 2x
56 i
I
Incorporating Eq. (A-30) in Eq. (A-29) gives:
e x ik x'+i kx'cos$'
(A-31) u(P) 27r dkx dx' e-k2_k 2 o
The integral
Sdx' ik x'+ikx'cos4'Sdx' ex
O
in Eq. (A-31) gives the result
| ik +kcosW'
xV (Here k = I - ik2 where the loss k2 << kI and k, > 0. The loss is added2
for ease of analysis; k2 may be set equal to zero in the final solution.)
# 1 Thus, Eq. (A-31) becomes:
n -ik2k 2ylj-ik x
(A-32) u(P) -iksin dk e x
2r k~ Fk k2 (k x+kcoso')dI
3The integral in Eq. (A-32) may be evaluated by the steepest descent method
which includes the Pauli-Clemmow modification[10] (for a pole near a saddle
point) after employing the polar transformations kx = kcosa, and ky = +ksina.
The integral in Eq. (A-32) is thus transformed into a contour integral in
the complex c-plane as:
isino' I e - i k pc os (o,; o ) O^
(A-33) u(P) C coscs da
jC cos+coso'<2
The contour of integration C for Eq. (A-33) is shown in Fig. A-2. Since
' : nv12 in the present problem, Eq. (A-33) becomes for 0<0<n (the results
for n<<27c are symmetric about ;=n):
1 i57
ir e -ikpcos(-)
(A-34) U(P) dICosaThe integration contour C may be deformed onto the steepest descent
path (SOP) through the saddle point at a=;--Q as indicated in Fig. A-2.
IMO 01 CONTOUR C
SADDLE POINT
AT G, s
REFLECTED WAVE POLESDP AT C1- IF
Fig. A-2. Contour of integration in the complex a-plane.
The reflected wave pole is crossed if as <7r/2 and its contribution (residue)
must be included. The saddle point contribution to SOP gives the desired
diffracted field; without giving details of the SOP evaluation (which are
available for the wedge diffraction problem in Reference [10)), the result
for the diffracted field ud is:
* (A-35) ud, T2. i(k- F(X)
coso > '
where F(X) is defined in Eq. (23) of Section II; here X is given by:
2i(A-36) X = 2kpcos -
58
Converting the result of Eq. (A-35) (which is valid for the coordinate
I system of Fig. A-l) to a result that is valid in the coordinate system of
Fig. A-3 in which *=*- jand *'=€'- j= 0 (for 0' = ), one obtains:
I : ui
I
I HALF- PLANE
Fig. A-3. Transformed coordinate system for the half-plane problem.
-A-37) udp 2 F 2 I "/4 2 eikP(A3) P in.' _0 T ef F(2kpcos e -) P
i( eikpD" Opk() e r
in which, the Kirchhoff diffraction coefficient, Dk() is
(A38a) 00€) = -i-n (X
l where
| (A-38b) X 2kpcos2 0.
The result in Eq. (A-38a) checks with that given in Eq. (22) of Section II.
I The ud(p) for the TEz case is assumed to be the same as in (A-37); this
assumption is reasonable within the transition region near 4=r, where the
I Kirchhoff result is relatively insensitive to polarization effects.
I59
REFERENCES
1. Burnside, W.D., "Analysis of On-Aircraft Antenna Patterns," Report3390-1, August 1972, ElectroScience Laboratory, Department ofElectrical Engineering, The Ohio State University; prepared underContract N62269-72-C-0354 for Naval Air Development Center.
2. Marhefka, R.J., "Roll Plane Analysis of On-Aircraft Antennas," Report3188-1, December 1971, ElectroScience Laboratory, Department ofElectrical Engineering, The Ohio State University; prepared underContract N62269-71-C-0296 for Naval Air Development Center.
3. Burnside, W.D., "Analysis of On-Aircraft Antenna Patterns," Report3390-2, May 1973, ElectroScience Laboratory, Department of ElectricalEngineering, The Ohio State University; prepared under ContractN62269-72-C-0354 for Naval Air Development Center.
4. Logan, Nelson A., "General Research in Diffraction Theory," Vol. 1,Report LMSD-288087, Missiles and Space Division, Lockheed AircraftCorp., December 1959. (This reference contains an extensivebibliography dealing with previous work on the diffraction of wavesby smooth convex surfaces).
5. Bowman, J.J., Senior, T.B.A. and Uslenghi, P.L.E. (Eds.), Electro-magnetic and Acoustic Scattering by Simple Shapes, North HolandPublishing Co., 1969, (p. 103 and p. 112 list Goriainov's results).
6. Wait, J.R. and Conda, A.M., "Diffraction of Electromagnetic Wavesby Smooth Obstacles for Grazing Angles," Jour. of Res., N.B.S.,Vol. 63D, No. 2, Sept.-Oct., 1959, pp. 181-197.
7. Logan, Nelson A., "General Research in Diffraction Theory," Vol. 2,Report LMSD-288088, Missiles and Space Division, Lockheed AircraftCorp., December 1959. (This reference contains tabulated valuesof the p(t) and q(4) functions).
8. Pathak, P.H. and Kouyoumjian, R.G., "The Radiation from Aperturesin Curved Surfaces," Report 3001-2, December 1972, ElectroScienceLaboratory, Department of Electrical Engineering, The Ohio StateUniversity; prepared under Grant NGR 36-008-144 For NASA LangleyResearch Center, Hampton, Virginia. (pp. 22-23 contain a descriptionof the pseudo-ray system introduced by Kouyoumjian for the Fock currentin the lit part of the penumbra region on a smooth convex surfaceexcited by a plane wave.)
9. Keller, J.B., "Geometrical Theory of Diffraction," J. Opt. Soc. Am.,Vol. 52, No. 2, February 1962, pp. 116-130.
60 7
II
10. Hutchins, D.L. and Kouyoumjian, R.G., "Asymptotic Series Describingthe Diffraction of a Plane Wave by a Wedge," Report 2183-3, December1959, ElectroScience Laboratory, Department of Electrical Engineering,The Ohio State University; prepared under Contract AF19(628)-5929 for
Air Force Cambridge Research Laboratories. (AD 699 228).
1f. Kouyoumjian, R.G., "Asymptotic High Frequency Methods," Proc. IEEE,
53, 1965, pp. 864-876.
12. Kouyoumjian, R.G. and Pathak, P.H., "The Dyadic Diffraction Coefficientfor a Curved Edge," Report 3001-3, August 1973, ElectroScience Laboratory,Department of Electrical Engineering, The Ohio State University;prepared under Grant NGR 36-008-144 for NASA Langley Research Center.(A paper on this subject is currently in preparation for publication.)
13. Abramowitz and Stegum (Eds.), Handbook of Mathematical Functions,National Bureau of Standards Publication, June 1964, p. 478.
14. Ryan, C.E. and Peters, L., Jr., "Evaluation of Edge-DiffractedFields Including Equivalent Currents for the Caustic Regions," IEEETrans. Ant. and Prop., Vol. AP-17, No. 3, May 1969. (See correctionsto this paper in March 1970 issue of IEEE Trans. AP.)
15. Ratnasiri, P.A.J., Kouyoumjian, R.G. and Pathak, P.H., "The WideAngle Side Lobes of Reflector Antennas," Report 2183-1, March 1970,
ElectroScience Laboratory, Department of Electrical Engineering,The Ohio State University; prepared under Contract AF 19(628)-5929for Air Force Systems Command. (AFCRL-69-0413) (AD 707 105)
16. Burnside, W.D. and Peters, L., Jr., "Axial Radar Cross Sectionof Finite Cones by Equivalent Current Concept with Higher OrderDiffraction, Radio Science, Vol. 7, No. 10, October 1972, pp. 943-948.
17. Felsen, L.B. and Marcuvitz, Radiation and Scattering of Waves,Prentice-Hall, Inc., 1973, pp. 721-724.
1.
I.61
* *1