The Supersonic Blunt-Body Problem—Review

and Extension
MILTON D. VAN D Y K E *
Ames Aeronautical Laboratory, NACA

SUMMARY mainly analytical, though within t h e last year elec-

A survey of existing analytical treatments of the supersonic or
hypersonic blunt-body problem indicates that none is adequate
for predicting the details of the flow field. Reasons are given for
the failure of various plausible approximations. A numerical
method, which is simpler than others proposed, is set forth for
solving the full inviscid equations using a medium-sized electronic
computer. Results are shown from a number of solutions for
bodies that support detached shock waves described by conic
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744
tronic computers have been brought to bear.
T h e present paper aims, first, to evaluate t h e mass
of existing analytical t r e a t m e n t s and, second, to p u t
forth a new and relatively simple numerical procedure.!
T h e survey will show t h a t existing analytical methods
are generally inadequate for predicting t h e details of
flow near a blunt nose. W i t h the numerical method,

sections. accurate solutions have been carried out for members

of a one-parameter family of plane and axisymmetric
SYMBOLS
bodies.
B = bluntness of conic section [see Eq. (1)]
C = 1 - Bs (1) R E V I E W OF E X I S T I N G ANALYTICAL SOLUTIONS
L-p = pressure coefficient referred to free-stream condi-
tions Existing analytical attacks on the supersonic blunt-
f = P/Py body problem fall mainly into four categories: (1)
M = free-stream Mach Number
Potential flow a p p r o x i m a t i o n s . 1 - 4 (2) Taylor series
P = pressure referred to p ra Fco 2
R nose radius expansions from shock. 5 - 1 1 (3) Incompressible ap-
=
r = coordinate normal to free stream proximations. 1 2 - 1 5 (4) Newtonian approximation and
u, v = components of velocity in £, rj directions improvements thereon. 1 6 - 2 5
V = velocity vector referred to free-stream speed T h e only other noteworthy approach is t h a t of ex-
X = coordinate in free-stream direction
plosion and similarity solutions 26, 27 which gives the
y = adiabatic exponent
asymptotic flow field downstream on a slender blunt-
8 = (7 - D/(y + 1)
A = standoff distance of shock from body nose nosed body. This elegant and useful theory is not in-
Z,V = orthogonal curvilinear coordinates [see Eq. (2)] tended to treat t h e vicinity of t h e nose and will, accord-
A£, A77 = mesh widths in numerical computation ingly, not be considered here. T h e four methods listed
V = 0 for plane flow, 1 for axisymmetric flow above will now be discussed in succession, drawing
P = density referred to free-stream value where necessary on the accurate numerical solutions
V = stream function
CO = modified stream function [see Eq. (11)] described in section (2) as a standard of comparison.
e = polar angle Other numerical methods 2 8 " 3 2 will be discussed in
( ), = value on shock wave section (3).
( )b = value on body
( ) . = free-stream value (1.1) Potential Flow Approximations
( )st = value at stagnation point
T h e oldest attack on the blunt-body problem con-
INTRODUCTION sists in attempting somehow to relate the actual flow
behind the detached shock wave to the subsonic (often

P HOTOGRAPHS of the detached shock wave t h a t forms

ahead of a blunt body in supersonic flight show a
beautiful simplicity t h a t has in t h e last decade lured a
number of aerodynamicists into trying to predict its
shape. Recently there have arisen more urgent rea-
sons for finding the complete flow field (particularly in
hypersonic flow), and a considerable fraction of the
theoretical aerodynamicists in t h e United States,
England, and (apparently) Russia have become in-
volved with the problem. T h e attacks have been
incompressible) potential flow past the same body.
T h e standoff distance for spheres in air (7 = 7/5) pre-
dicted b y several such theories is compared in Fig. 1
with the results of the numerical solutions described
later. T h e standoff distance is a convenient test be-
cause it is easily measured from schlieren photographs
and should be given accurately b y any theory t h a t is
to predict other more i m p o r t a n t properties of the flow
field. Moreover, its prediction is actually the main
objective of most of these approximations.

Presented at the Hypersonic Aerodynamics Session, Twenty- f Another survey of the blunt-body problem in the special case
Sixth Annual Meeting, IAS, New York, January 27-30, 1958. of hypersonic flow will appear in the monograph Hypersonic Flow
Revised and received April 22, 1958. Theory by Wallace D. Hayes and Ronald F. Probstein, to be
* Aeronautical Research Scientist. published soon.
485
 486 JOURNAL OF THE A E R O / S P A C E S C I E N C E S —AUGUST, 1958

tional terms in the special case M = 2 (reference 11).

F I G . 1. Potential flow approximations for standoff distance of
shock from sphere with y = 7/5.
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744
T h e variation of Stokes' stream function along the axis
of revolution behind a paraboloidal shock wave is com-
pared in Fig. 2(a) with the accurate numerical solution.
Successive terms of the Taylor series are seen to
meander with no clear indication of either convergence
or divergence.
T h e expansion procedure can be simplified b y intro-
ducing the natural curvilinear coordinates associated
with the shock wave—parabolic coordinates for a parab-
oloidal shock wave, etc. T h e Taylor series have then
the advantage of being not double b u t only single ex-
pansions (in t h e coordinate leading away from [the
shock). I t m a y also be anticipated t h a t the conver-
gence properties of the series will show u p more clearly
in the natural coordinates. I t is easy to recast Ca-
bannes' series in parabolic coordinates. T h e result

T h e accuracy is seen to be mediocre, particularly in

.50
hypersonic flow (and is still worse for plane flow). V r ^ SHOCK
Closest agreement is given b y the theory of K a w a m u r a , 3 x^
N>
•
!
y^
A
WAVE

who makes a straightforward m a t c h of streamline \ \ ^N. f /

slopes behind t h e vertex of t h e shock wave with those
of the incompressible potential flow. His results have
.25
\ \\\ V ^
\\
V \ .

been reproduced b y Heybey 4 who also a t t e m p t s to esti-

\\\ \ X
V\ \%
72 \ \ ^2 TERMS
m a t e a correction for compressibility. However, it V NN
will be seen in section (1.3) t h a t vorticity rather t h a n ! V\ \ \ l 1

compressibility is the chief factor neglected at high \\\

\\ \
M a c h Numbers and t h a t Heybey's estimate of the com- V\
V N
pressibility effect is too large. In any case, these
theories are of no value for predicting such quantities -.25 h
as surface pressures (which b y assumption have the
form corresponding to incompressible potential flow). V\\\4
Y&6
\V5
^NUMERICAL
(/ .2) Taylor Series Expansion From Shock -.50
D .10 .20 w .30

If t h e form of a detached shock wave is known or

assumed, t h e flow variables (pressure, density, velocity) (a) Cartesian coordinates.
just behind it are given b y the oblique shock relations,
and their first derivatives can be found b y substituting
into the equations of motion. Higher derivatives can
be found b y first differentiating t h e equations of motion.
According to t h e Cauchy-Kowalewski theorem (see, e.g.,
reference 33) t h e flow field is analytic somewhat down-
stream of t h e shock wave, and there is physically no
reason to doubt t h a t the region of analyticity extends
to the body. One m a y therefore a t t e m p t to represent
the flow layer between t h e shock and b o d y b y a few
terms of a Taylor series expansion starting from the
shock wave. This idea, introduced b y Lin and Rubi-
nov, 5 has been applied to plane, 8 axisymmetric, 6 , 7 - 10' n
and three-dimensional 9 flows.
T h e expansion has been pursued furthest by
Cabannes for axisymmetric flow. T a k i n g the shock
wave to be described in Cartesian coordinates b y x =
2anrn, he gives t h e coefficients in t h e double Taylor
series for the various flow variables out to x 4 , x2r2, r 4
(and of Stokes' stream function to x 6 ) for (t>) Parabolic coordinates.
7 = 7/5 and general free-stream M a c h N u m b e r M
FIG. 2. Variation of stream function along axis behind para-
(reference 10), and the stream function to two addi- boloidal shock wave at I f = 2 with y = 7/5.
 S U P E R S O N I C B L U N T - B O D Y
corresponding to Fig. 2(a) is shown in Fig. 2(b), and
it is now clear t h a t a t the nose of the body the series
diverges.
T h e probable source of the divergence has been sug-
gested in reference 34. T h e Taylor series expansion
describes not only the flow downstream of t h e shock
wave b u t also its analytical continuation upstream.
As indicated in Fig. 3, this fictitious flow will contain
a limiting line which intersects t h e axis at the sonic
point and in t h e supersonic region is t h e envelope of
outgoing characteristics. (Such a line exists also in t h e
well-known Taylor-McColl flow past a circular cone;
its location is, in fact, tabulated b y Kopal. 35 ) T h e
flow variables are nonanalytic a t the limiting line where
they vary as half powers of the distance. Hence if the
shock wave is closer to the limiting line t h a n to the
body, the Taylor series will not include the body in its
radius of convergence. This suggestion has been veri-
fied b y running the numerical procedure backward to
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744
calculate the fictitious flow in t h e example of Fig. 2.
On the axis, the limiting line was found to lie only
three-fourths as far from the shock wave as does the
body nose.
T h e situation is comparably bad for other shapes and
Mach Numbers and worse away from the axis or in
plane flow. Hence t h e method of Taylor series expan-
sion from the shock wave cannot be expected to suc-
ceed. Nevertheless, the first three terms of the series
(in natural coordinates) can serve as a useful qualitative
model in the case of axisymmetric flow.34 For a parab-
oloidal shock wave at M = °° (and 7 = 7/5) the
model predicts the ratio of standoff distance to body
nose radius correct to within 0.8 per cent, so t h a t it
m a y also have some quantitative value. However, it
deteriorates away from the axis and does not predict a
real body out to the sonic point.
(1.3) Incompressible Approximation
At high supersonic M a c h Numbers the density varies
only slightly between the shock wave and the body in
the vicinity of the stagnation point. I t therefore seems
appropriate to solve t h e equations of rotational incom-
pressible flow together with the exact conditions just
behind t h e shock wave. T h u s Lighthill 1 3 has obtained
the solution in closed form for t h e special case of a
spherical shock wave a t infinite M a c h N u m b e r (that
is, in t h e ''strong shock" approximation), the body
being found as a concentric sphere. Similarly, Whit-
ham 1 4 and Hayes 1 5 have shown t h a t in plane flow a
circular shock leads to a concentric circular body.
T o assess this approximation, numerical solutions
have been carried out for a spherical and a circular
shock wave in a stream of infinite M a c h N u m b e r (all
solutions for M = °o were actually carried out at M =
10,000) with 7 = 7 / 5 . In axisymmetric flow t h e re-
sulting body shape, shown in Fig. 4(a), lies close to the
sphere of the incompressible approximation out
roughly halfway to the sonic point. In plane flow, as
might be expected, the agreement is poorer (though it
P R O B L E M 487
would improve in either case as 7 —*- 1). T h e body is
roughly the front portion of a 4 : 1 ellipse standing on
end rather t h a n a circle.
For t h e flow near the axis, these results were given
earlier b y Hida. 1 2 His analysis is more general in t h a t
all supersonic M a c h N u m b e r s are considered. Fig.
4(b) shows t h a t his results for standoff distance are
fairly accurate throughout the M a c h N u m b e r range,
the remaining discrepancy being presumably a t t r i b u t -
able to compressibility. T h u s a t infinite M a c h N u m -
ber the approximation of irrotational incompressible
flow3, 4 gives 0.094 as the standoff distance for a sphere
in air; the increment due to rotation is 0.026 according
to t h e Hida-Lighthill theory and the remaining differ-
ence of 0.008 from t h e value 0.128 of t h e numerical
solution m u s t represent the effect of compressibility.
(Heybey's estimate of t h e compressibility effect4 is
0.024 in this case.) I t might be possible to include
compressibility b y iteration in t h e style of the Janzen-
Rayleigh approximation for subsonic flow. If so, this
would seem the most promising of the analytic theories.
(1.4) Newtonian Approximation and Improvements
Thereon
In simple Newtonian theory, fluid particles are im-
agined to lose their normal component of m o m e n t u m
upon striking the surface of t h e body. T h e local pres-
sure coefficient is then given b y 2 sin 2 6, where 6 is the
angle t h a t t h e surface makes with t h e stream direction.
Lees 36 has proposed a modified Newtonian theory which
consists merely in scaling down this result so as to be
exact at the stagnation point where the correct value is
known. H e suggests t h a t because of compensating
effects, this will be more accurate t h a n the shock-layer
or " Newtonian-plus-centrifugal" approximation 1 6 , 1?
which is the actual limit of the solution f or M —• °° and
7—1.
T h e surface pressures on a sphere and circular cylin-
der a t M = °° with 7 = 7/5 according to these three
approximations are compared in Fig. 5 with numerical
solutions. T h e modified Newtonian result is seen to
fall either above or below the accurate solution, depend-
ing on M a c h N u m b e r and body shape, b u t to be re-
markably accurate when 7 = 7 / 5 .
Chester 2 3 and Freeman 2 4 have taken t h e shock-
layer or Newtonian-plus-centrifugal solution as a first
step and sought to improve it systematically b y suc-
cessive approximations. T h e result is a double series
expansion of the flow quantities for small 5 = (7 — 1) -f-
(7 + I) and M~2 (a single series expansion in t h e work
of Freeman, who considers only the "strong shock"
case, corresponding to M = °°).
Unfortunately, t h e series appears to converge slowly
for practical values of 7, and worse in axisymmetric
t h a n plane flow. This is illustrated in Fig. 6 for the
standoff distance in Chester's example of a parabo-
loidal shock wave. B y appeal to t h e model mentioned
a t the end of section (1.2), t h e present writer has sug-
gested 3 4 t h a t Chester's series actually converges for
7 > 2.2 (corresponding to 8 < 3/8), whereas for real
 JOURNAL OF THE A E R O / S P A C E S C I E N C E S — AUGUST, 19 5 J

LIMITING LINE 2.0 r

OUTGOING / '
CHARACTERISTIC / / ,

/ ' / / SHOCK WAVE 1.5

SONIC LINE
1.0
N
\
\
.5 NUMERICAL SOLUTION \
NEWTONIAN \
NEWTONIAN-PLUS-CENTRIFUGAL
MODIFIED NEWTONIAN
i i i i i
10 20 30 40 50
FIG. 3. Schematic sketch of analytical continuation of flow 0,DEG
field upstream through shock wave, showing limiting line (from
reference 34).
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744

(a) C i r c u l a r c y l i n d e r .

2.0

.6r
'CIRCULAR OR SPHERICAL SHOCK
/ INCOMPRESSIBLE APPROX
1.5
AXISYMMETRIC BODY

SONIC POINT
1.0

.5 NUMERICAL SOLUTION
NEWTONIAN
NEWTONIAN-PLUS-CENTRIFUGAL
MODIFIED NEWTONIAN
1 l l 1 1
0 10 20 30 40 50
0,DEG

("b) S p h e r e .
ra
FIG. 5. Surface pressure distributions at M = , 7 = 7/5.

.24
/ /
I TERM OF / /
.20 CHESTER'S SERIES / / 3 TERMS
/ /
/ /
.16

lb
.l2

.08

.04

1.0 I. 1.2 1.3 1.4 1.5 1.6 1.7

FIG. 4. Accuracy of incompressible approximation, (a) Plane ADIABATIC EXPONENT/
and axisymmetric bodies that support circular and spherical
shock at M = °°, 7 = 7/5. (b) Standoff distance of shock from FIG. 6. Variation with 7 of standoff distance for bodies that
sphere with 7 = 7/5. support paraboloidal shock at M = °o.
 S U P E R S O N I C B L U N T - B O D Y P R O B L E M 489

gases 7 does not exceed 1.67; b u t it is clearly not useful b y £ = 0 (Fig. 7). Special cases are
for 7 above 1.1. I n the same note it was shown how,
x/R = (1/2) (1 + i2 - v2), for parabola (Bs = 0)
by comparison with the model, the convergence of
Chester's series for the standoff distance can appar- x/R = 1 - V l - J2??, for circle (Bs = 1)
ently be accelerated. T h u s at y = 1.667 in the ex-
ample of Fig. 6, five terms of t h e modified series give For ellipses, Eq. (2) gives only the left half which is the
A/Rb = 0.198, which differs from the accurate numeri- part t h a t m a y form a shock wave. T h e right half of
cal result b y only 4 per cent. T h e convergence of the t h e ellipse is obtained b y taking a plus instead of a
series is even worse for surface pressure, and there no minus sign ahead of the radical in Eq. (2).
useful modification has been found. In this coordinate system the line element is given by
ds2 = hW + h22drj2 + phz2dtp2 (3a)
(2) D E S C R I P T I O N OF N U M E R I C A L M E T H O D
where v = 0 for plane flow and v = 1 for axisymmetric
T h e preceding survey shows t h a t none of the existing flow, so t h a t the last t e r m (involving the azimuthal
analytic treatments is adequate for calculating t h e de- angle cp) appears only in the latter case, and
tails of flow near a blunt nose. One must t u r n instead
h2 = ( Q 2 + u 2 ) / ( l - Bse),
to numerical solution of the full equations. A rela-
tively simple numerical scheme will be described h22 = ( Q 2 + v2)/(C + BsV2), h2 = £ V (3b)
which has been programed for a medium-sized elec-
where C = 1 — Bs.
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744

tronic computer.
(2.2) Equations of Motion
(2.1) Coordinate System
Let V be the velocity referred to the free-stream
It is convenient to assume a family of detached shock
speed Fco, p the density referred to its free-stream value
waves of known shape. T h e resulting simplification
Poo, and p the pressure referred to pooVo,2. Then the
more t h a n offsets t h e disadvantage of having to accept
whatever body shapes result. Another possible objec-
tion will be dealt with in section (3.4). Schlieren pictures
show t h a t the shocks produced b y simple shapes such r 2 =2RX-BX 2
as spheres, paraboloids, and ellipsoids (or their plane 77 = 1
counterparts) are themselves nearly conic sections.
Accordingly, the present method has been applied to
the family of plane and axisymmetric bodies t h a t
support detached shock waves described b y conic
sections—hyperbola, parabola, prolate ellipse, circle,
or oblate ellipse. I t could, however, be extended to
other analytic shock shapes.
In Cartesian coordinates originating from its vertex
(Fig. 7), any conic section is described b y
2Rx - Bx2 (1)
Here R is the nose radius, and B the bluntness, which
is a convenient parameter t h a t characterizes the eccen-
tricity of the conic section. (The bluntness B is b2/a2
for an ellipse b u t —b2/a2 for a hyperbola.) As shown
in Fig. 8, the bluntness is zero for a parabola, negative FIG. 7. Coordinates for conic section.
for hyperbolas, positive for ellipses, and unity for a
circle.
I t is advantageous to choose a natural coordinate
system t h a t contains the shock wave as one of its sur-
faces. T h e conventional hyperbolic, parabolic, elliptic,
and polar coordinates are unsatisfactory because their
definition changes at B = 0 and again a t B = 1 whereas
conic sections clearly form a single continuously vary-
ing family. A suitable unified orthogonal system t h a t
covers the entire range is introduced b y setting

x = (R/Bs) [1 - V(T Bse) (i -B8 + B8^)]\ (2)

r = Rfr
T h e shock wave of bluntness Bs is described b y t] = 1; B=l CIRCLE

the downstream axis b y 77 = 0; and the upstream axis FIG. 8. Conic sections of varying bluntness.
 490 JOURNAL OF THE A E R O / S P A C E S C I E N C E S — AUGUST, 1958

my V(ce + v2)/(c + B^)PU}( +

my V(ce + v2)/(i - Be)pv]n = o
2
P ( « « { - [c&*/(ce + v )} +
V(C + Br,*)/(I - 3^) »{«, +
lnu/(Ce + r,2)l})+P( =0
(5)
p(ro, - MV(Cfs + us)] +
V ( l - 5J 2 )/(C + Br,*) u{v( +
lQv/(Ce + v2)}})+P„ = 0
tVi - se (.p/p7)i +
vVc + S172 (M> 7 ), = 0
FIG. 9. Variation of density along axis behind parabolic shock
wave at M = °°, 7 = 7/5. where u, v are t h e components of F i n the £, rj directions.
T h e first (continuity) equation is satisfied b y intro-
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744

equations of continuity, motion, and energy are, in ducing the stream function SP according to
vector form,
* , = {ivY V(ce + n2)/(C + Bj)pu ]
d i v ( p F ) = 0, p(F-grad)~V + grad p = 0, (6)
*f = - (^)V(a 2 + V
2
)/Ti - Be)Pv\
V.grad(£/V) = 0 (4)
T h e n the last (energy) equation simply states t h a t
where 7 is t h e adiabatic exponent. Transforming
these to t h e (£, 77) coordinate system (using Eq. (3), and P = PVW (?)
dropping t h e subscript on Bs for simplicity) gives Using this to eliminate t h e pressure from the equa-
tions of motion gives

****, - %%t *t{*«[(p,/p) + Wv)] - *,[(p*/p) + ("/*)]} - iv/(ce + v2)} ( v +

[(C + Br,*)/(l - £| 2 )R, 2 } + [51/(1 - 3£ 2 )]M^ +
(IV)' [(Q 2 + i?s)/(i - 5?2)] (YP7P, + P T + T O = 0
(8)
*Jr.
v £y SEyJr,, + * , { * { [ ( P , / P ) + ( v A ) ] - ¥ , [ ( p t ' p ) + (v/f)]} - [Q/(Ce + v2)l X
{[(1 - Bf*)/(C + 5 i j * ) ] V + *,»} - [S„/(C + B„*)]*{*, +
2 2
iev y nee + n )/(c + M) (TP7>£ + P7+TO = 0
(2.3 ) Initial Conditions

Values of u, v, p, p just behind t h e shock wave (at

rj = 1) are found from t h e oblique shock relations— P = (7 + i)M2(i - Be)/mi + ce) +
e.g., reference 37—in terms of t h e slope of t h e bow (T - 1)M 2 (1 - Be)], (9)
wave, — B£2/%. Expressed in terms of t h e stream
M> == j i + 7 ( i + „), * = p^+"t a t r, = 1
function, these give t h e initial conditions

and for t h e function /(SF)

, , 2 T M 2 ( 1 - Bs2) - (7 - 1)(1 + Cs2)

7 ( 7 + 1 ) ^ P (1 + Cs2)
[2(1 + Cs2) + (7 ~ 1)-M2(1 - Bs2)'
, s2 = [(1 + „ ) ¥ ]2 / d + i') (10)
, (7 + l)Af 2 (l - Bs2) .

(2.4) Form of Problem for Numerical Computation

For numerical work it is advantageous to use (1 + v)^/il+v, which is constant on the shock and elsewhere more
nearly independent of £ t h a n is ^ and vanishes only on the b o d y rather t h a n also on the axis of symmetry. Hence set

*&i?) = [£1+7(i + *0M&f) (ii)

Then the initial conditions become
 SUPERSONIC BLUNT-BODY PROBLEM 491

p = (7 + l)M\l - Be)/[2{1 + Q 2 ) + ( 7 - 1)M 2 (1 - B ^ l co = 1, co, = (1 + y)p, a t 17 = 1 (12)

a n d t h e equations of motion

fog 2 + v ( f co€ ££2 SPA , S 2 ^^

7 1 ^ ^ f ^ 1 + *. P 1 + vll + ?
fco^ — I co + -
1 + y / \ l - ££ 2
1

+ + w+ + V H V , c+Bv*/2 &,
co +
i^)("' r+^) f ( r+^) cv+7* 1 + v. + 1 - 5 { \ 1 + f/ J

A ^ - l f r ' - l ^ r (13a)
£<•>! p, a>, + fro;, Br, £cOf \ _ ft), |p {
co +
l +v L 1+ v V C + Br,'- 1+ J 1+^ J
2
(1 + v)C 1 - 5 f Y £cOj &>,
w H + ( l + vjp 7 y?7 7 fe (13b)
Q 2
+ r, 2
LC + 5 W ' 1 + vl ' \1 + v C + Br,2 £p ^ A 1 +»
2 T M 2 ( 1 - Bs2) - ( 7 - 1) (1 + Cs2) 2(1 + a 2 ) + (7 - 1)M2(1 - Bs2)
where f= , 5 2 = J 2 co 2/(1+ " ) (13c)
7 ( 7 + l)M2 (1 + C52) (7 + 1)M2(1 - Bs2)
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744

(2.5) Numerical Integration

The initial value problem of E q s . (12) and (13) has
been solved numerically b y forward integration from
the shock wave (77 = 1) toward smaller values of 77.
Over each interval (A77), p and co, are extrapolated
linearly. A t each new value of 77 the ^-derivatives are
evaluated b y 11-point numerical differentiation. 38 T h e
procedure m a y be summarized as follows:
(0) Calculate initial values a t 77 = 1 from E q . (12).
(1) Calculate ^-derivatives p^, co^, co^, co^, b y 11-point
numerical differentiation.
(2a) Calculate p, from E q . (13a).
(2b) Calculate co,, from E q . (13b).
(3a) Extrapolate p and co, linearly t o next smaller
value of 77:
p (m + l) = p(m) _ (A77)p,W, c o , ( m + 1 > = cor (m)
dimensions A£ and A77 are arbitrary (but equal for all
steps). One prescribes t h e parameters
M = free-stream M a c h N u m b e r
Bs = bluntness of conic section describing shock
v = 0 for plane, 1 for axisymmetric flow
7 = adiabatic exponent
A£, A77 = mesh dimensions
Flow variables (and their 77-derivatives) are printed for
each mesh point. T h e body shape and surface pres-
sures can then b e found b y interpolation. Machine
computing time is \1/\ minutes per value of 77. I t will
be seen t h a t 4 to 8 steps in 77 yield ample accuracy so
t h a t a typical case requires 5 to 10 minutes computing
(AT?)* (m) time.

(3b) Extrapolate co using averaged value of (2.6) Instability

Am + l) „W - (1/2) (AT?) (co/™) + < » + »)
(4) Repeat steps (1) to (4) a t new value of 77 a n d con-
tinue until co is negative.
After completion of t h e integration, t h e accuracy of
p (and hence of p) can be greatly increased b y recom-
puting it using t h e averaged value of p, over each in-
terval A77, so t h a t t h e value after t h e rath step is given
by
p (») = p«) - ( A , ) [(l/2)p,<») + p,(» + p/2> +
. . . + p„c—» + (l/2)p,<»>] (14)
Fig. 9 shows in a typical case t h e improvement obtained
from this "external iteration."
This procedure has been coded for machine compu-
tation on t h e I B M 650 electronic digital computer with
floating decimal a t t a c h m e n t . F o r each value of 77
the flow quantities are calculated a t t h e 20 equally
spaced values £ = {n + 1/2) (A£), 0 < n < 19, which
straddle the axis of symmetry £ = 0 in order to avoid
the indeterminate form pj^p in E q . (13). T h e mesh
In t h e subsonic region t h e initial value problem is
unstable. This means t h a t small errors (such as arise
from rounding t o the 8 significant figures used) grow
in geometric progression. T h e instability has been
tested in a typical case b y perturbing t h e initial data,
changing co from 1.000 0000 to 1.000 0010 a t t h e point on
the shock wave nearest to t h e axis (where t h e equation
is most elliptic and hence most unstable). Fig. 10
shows t h a t in 11 steps (last step interpolated) from
shock t o body a unit error grows b y a factor of 160.
This means t h a t rounding errors of 1/2 in t h e last
place can amplify t o invalidate no more t h a n 2 of t h e
8 significant figures used. Near t h e stagnation point
the error grows b y a factor of about 3 in each step.
(This m a y be compared with t h e maximum factor of
5.8284 for Laplace's equation in a semi-infinite
domain. 39 ) Hence t h e error would swamp all 8 signifi-
cant figures if i t were necessary to take as m a n y as 24
steps from shock t o body. Fortunately, m a n y fewer
steps t h a n even 11 suffice, so t h a t instability is of no
practical concern.
 492 J O U R N A L OF THE A E R O / S P A C E
Another type of instability peculiar to t h e particular
differentiation scheme used is more serious and, in
practice, limits the downstream extent of flow field t h a t
can be calculated. T h e 11-point differentiation scheme
uses central differences where possible, b u t t h e largest
5 values of £ must use progressively more noncentral
schemes. These involve large differences of small num-
bers and so produce instability which is more severe in
the hyperbolic t h a n the elliptic region. I t eventually
causes oscillations of flow variables for t h e highest few
values of £, and stops t h e computation b y producing
negative values of p. This difficulty has been allevi-
ated b y using 7- and 5-point differentiation schemes at
the highest two values of £. I t is to be entirely elimi-
nated b y recoding the program to use only central-dif-
ference schemes, starting with more values of £ and
dropping t h e highest 5 in each step in rj.
A second imperfection in the present procedure ap-
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744
pears when t h e factor {co + [£cos/(l + v)]} in Eq. (13)
vanishes, leading to a singularity in co^. This can hap-
pen only inside the body (for co < 0) but, because of the
instability illustrated in Fig. 10, its effect occasionally
spreads so rapidly as to affect the actual flow field.
(2.7 ) Accuracy—Comparison With Garabedian-
Lieberstein Method
T h e accuracy of t h e numerical method has been
tested b o t h internally, b y refining the mesh size, and
externally, b y comparing with experiment and with the
numerical procedure of Garabedian and Lieberstein. 29
T h e Garabedian-Lieberstein method is like t h a t pro-
posed here in starting from a given analytic shock shape
(and conic sections were actually used) and solving the
initial-value problem with electronic computers. I t
differs in t h a t it avoids the instability in the subsonic
region b y continuing the initial d a t a analytically into a
fictitious third dimension where the equation is hyper-
bolic rather t h a n elliptic. T h e solution is carried out
b y the numerical method of characteristics on a num-
ber of planes in the fictitious three-dimensional space
t h a t intersect the real physical plane along curved
lines. T h a n k s to known uniqueness proofs for the nu-
merical method of characteristics, the Garabedian-
Lieberstein method possesses a degree of mathematical
rigor seldom attained in aerodynamic research.
T h e price paid for eliminating instability is a large
increase in the computing time (and cost) compared
with the present method. Also the solution is arti-
ficially restricted to the subsonic region (though an-
other solution can be carried out for the supersonic
region).
I t can be shown t h a t the present procedure converges
as the mesh size is refined, despite its instability. In-
deed, it differs only in degree from a n y stable numerical
procedure, in which one m u s t strike a balance between
the truncation error resulting from too few steps and
the round-off error resulting from too m a n y ; here the
accumulated error simply grows in geometric rather
t h a n arithmetic progression. W h a t is lacking is a proof
t h a t the solution to which t h e present method con-
S C I E N C E S —AUGUST, 1958
verges is the correct one. (The question of whether
a real body exists for any given shock wave is an open
question in either method.) However, it reproduces
the results of t h e Garabedian-Lieberstein method to
four significant figures, as shown for one case in Figs.
11 and 12. Fig. 12 indicates t h a t the present method
is t h e more accurate near the sonic line.
In general, as in t h e example of Fig. 11, four to six
steps in rj along the axis are found to yield t h e standoff
distance correct to within 1 per cent, with comparable
accuracy in the other flow quantities. I t was thought
worthwhile to carry out the solutions discussed below
only t o this standard of accuracy.
(3) E X A M P L E S AND D I S C U S S I O N
Some 50 examples have been calculated, covering a
wide range of shock bluntness, M a c h Number, and y
in b o t h plane and axisymmetric flow. Only axisym-
metric flow will be considered henceforth as being of
more practical interest. A complete compilation of both
plane and axisymmetric solutions will be issued later
as an N A C A publication b y t h e writer and Helen
Gordon, who carried out the coding and machine com-
putation.*
(3.1) The Family of Bodies
At each M a c h N u m b e r (and 7) the family of conic-
section shock waves leads to a one-parameter family of
bodies. Fig. 13 shows various members of the family
for M = °° . T h e downstream extent of the body was
limited in each case b y the end instability discussed in
section (2.6), which could be eliminated.
Except possibly for the very bluntest, t h e bodies can
all be accurately described b y conic sections back at
least to t h e sonic point or limiting characteristic.
Thereafter the solution could be continued, or modi-
fied, b y t h e method of characteristics. However, it is
a virtue of t h e present method t h a t it gives t h e sub-
sequent shape. T h e sonic line and the change of type
from elliptic to hyperbolic play no significant role.
F r o m this point of view it m a y be said t h a t the blunt-
body problem is not a transonic problem.
(3.2 ) Comparison With Experiments on Spheres
Because t h e bodies are found all to be represented
closely b y conic sections, it is possible to consider a
fixed shape over a range of M a c h Numbers. Enough
experience has been accumulated a t 7 = 7/5 t h a t a de-
sired body can be produced in one or two a t t e m p t s .
T h u s the sphere has been solved for a number of M a c h
Numbers from 1.30 to 00 (and values closer to 1 would
offer even less difficulty). T h e standoff distance is
compared in Fig. 14 with experimental results from
references 40 to 44. T h e surface pressure distribution
is compared in Fig. 15 with experiments a t I f = 1.30
(reference 45) and M = 5.8 (reference 41).
* We are indebted to Marcelline Chartz for continued guidance
through the mysteries of machine computation.
 S U P E R S O N I C B L U N T - B O D Y P R O B L E M 493

.16 1.8 STOP

-.05 .27 -1.9
.01 0 -.55
-.01 0 .47
.01 .01 -.20
0 -.01 .21
0 .03 -.50
-.01 -.15 .15
-.04 .17 1.8
-.01 .10 -.97 -3.3
o HEBERLE, WOOD, GOODERUM (REF40)
t .02 .02 -.55 1.5
• OLIVER (REF4I)
c .02 0 -.1 1 .30 I.I O CRAWFORD, MCCAULEY(REF42)
i
, .01 0 -.01 .09 -.07 -1.8 A CHARTERS, UNPUBLISHED BRL PHOTOS
-.01 0 .03 -.03 . 2 2 3.8 ^SUGIMOTO (SEE REF 3)
0 0 -.04 .05 -.13 - 1 0 a LADENBURG, WlNCKLER, VAN VOORHIS (REF43)
1 -.01 -.03 -.02 -.35 -1.8 9.0 QRAINEY(SEE REF 44)
.01 .02 .02 .02 .05 .16 .25 1.0 5.9 2 9 -NUMERICAL SOLUTION ^
0 -.01 -.01 . 0 4 .14 .37 1.3 24 1 1 20
.01 .02 .03 -.1 1 -60 -1.7 -4.5 -12 - 3 3 -141
1 i 1 i.o .97 .93 .89 .26 -1.4 -5.8 -18 -46 -120 - 3 2 0
— •
A AXIS OF SYMMETRY A
SHOCK WAVE NOSE OF BODY

FIG. 10. Growth of unit error along axis behind parabolic

detached shock wave at M = oo, y = 7/5.
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744

_1 1 1 1 I L _ l
GARABEDIAN-LIEBERSTEIN 0.10982 8 10

FIG. 14. Standoff distance for sphere in air.

I I
rOI .02 .03 04 .05
MESH WIDTH, AT?

FIG. 11. Variation with mesh size of standoff distance'for^body NUMERICAL SOLUTION
that supports hyperboloidal shock wave asymptotic to Mach cone
at M = 5.8, T = 7/5.
Pst EXPERIMENT
(FROM DENSITY MEASURED
IN REF 45)

45 90
PRESENT SOLUTION 0.DEG
COARSE MESH(A77=.04) '
= .025)
FINE MESH ( A T ; = .025
(a) M = 1.30
o GARABEDIAN-LIEBERSTEIN SOLUTION

.4
r/R.

FIG. 12. Surface pressure distribution over body that supports

hyperboloidal shock wave as}^mptotic to Mach cone at M —
5.8, 7 = 7/5.
NUMERICAL SOLUTION
SHOCK BLUNTNESS B-

SONIC
POINT

FIG. 13. Family of axisymmetric bodies t h a t support detached ("b) M = 5.8

shock waves of conic-section shape at M = °°, 7 = 7/5 (drawn
for equal shock radii Rs). FIG. 15. Pressure distribution on sphere in air.
 494 JOURNAL OF THE A E R O / S P A C E S C I E N C E S —AUGUST, 1958

the various strips yields a system of ordinary differ-

3 - ential equations t h a t are integrated on an electronic
digital computer. As in the Garabedian-Lieberstein
method, t h e sonic line assumes a spurious significance
which necessitates special t r e a t m e n t in its vicinity.
Belotserkovsky has published results for a circular
cylinder a t M = 3, 4, 5 (with 7 = 7/5) which agree
well with those from t h e present m e t h o d ; and it would
seem reasonable to assume t h a t he has also considered
more practical axisymmetric bodies.
Uchida and Yasuhara have given a very tedious
method of successive approximations which does not
seem suited to machine computation. Their one pub-
lished example of a circular cylinder at M = 2 (7 = 7/5)
agrees fairly well with results from the present method.

(3.4) Shock Insensitivity

I
0 I 2 An objection sometimes raised against the inverse
SHOCK BLUNTNESS, Bs method of starting from an assumed shock wave is t h a t
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744

its shape is insensitive to changes in the body shape.

FIG. 16. Variation of body bluntness with shock wave blunt-
ness at M = oo, T = 7/5. A near nonuniqueness is implied according to which
essentially the same shock wave would correspond to
various bodies. However, no such difficulty exists
(3.3) Comparison With Other Numerical Methods
within the family of shapes considered here. Fig. 16
After developing the present numerical procedure, shows t h a t for shapes sharper t h a n a parabola the
the author learned of two other independent t r e a t m e n t s
b y essentially the same method. Priority clearly goes
to K. Mangier, of the Royal Aircraft Establishment, HOCKS,
although his work is not yet published.* At a meeting
in Holland in J u n e 1957, he reported a h a n d computa-
SONIC
tion of the plane flow behind a parabolic shock wave
POINTS
at M = 7 (with 7 = 7/5). T h e body was found to be
circular within half a per cent to behind the sonic line
(in conformity with results of the present method)
and was continued to a semicircle b y the numerical
method of characteristics.
More recently, Zlotnick and Newman 3 0 have applied
a similar procedure to t h e spherical shock wave a t
several high M a c h Numbers. T h e accuracy of their
examples is low (the pressure at the stagnation point
being off b y some 3 per cent and the standoff distance FIG. 17. Axisymmetric bodies at ¥ = co , 7 = 7/5 matched
with same nose radius and with same sonic point.
low b y 10 per cent), a t least partly because the thick-
ness of the layer between shock and body is neglected
a t one point. T h e y apparently a t t e m p t to suppress
t h e instability b y filtering out high harmonics (which
are the most unstable). T h e present author believes
such a t t e m p t s to be dangerous as well as unnecessary.
In contrast with these indirect methods (that start
from an assumed shock wave), Belotserkovsky 3 1 and
Uchida and Yasuhara 3 2 have described methods for
solving the direct problem of a given body. Following
the so-called integral method of Dorodnitsyn, Belot-
serkovsky divides the layer between shock and body
into n strips (2 or 3 in practice) across each of which
the flow variables are approximated b y polynomials.
Integrating the equations of motion analytically across
* ADDENDUM: Mangler's work is now available in: Mangier,
K. W., and Evans, M. E., The Calculation of the Inviscid Flow
Between a Detached Bow Wave and a Body, R.A.E. Tech. Note FIG. 18. Result of trying constant effective 7 to simulate real
Aero 2536, October, 1957. gas effects, M = 14.2.
 S U P E R S O N I C B L U N T - B O D Y
bluntness of shock a n d b o d y change a t t h e same r a t e ;
and although t h e shock insensitivity increases for
blunter shapes, it b y no means approaches infinite slope
which would imply complete insensitivity.
(3.5) Shoulder Choking
Increasing shock insensitivity for b l u n t bodies is
related to t h e phenomenon of shoulder choking pointed
out b y Busemann 4 6 a n d Hayes. 1 8 According to this,
the standoff distance a n d t h e shock shape in t h e sub-
sonic region are determined mainly b y t h e shape
of t h e b o d y near t h e sonic point. T h i s is illustrated in
Fig. 17 b y superposing first t h e vertices and then t h e
sonic points of two different bodies. T h e effect would
become even more pronounced for b l u n t e r shapes (and
in plane flow).
(3.6) Possibility of Treating Real Gases
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744
T h e present numerical m e t h o d is suitable for t h e in-
clusion of real gas effects assuming equilibrium t h e r m o -
dynamics. H a n d calculation would probably be neces-
sary if t h e s t a n d a r d tables were t o be used, b u t t h e y
could be approximated analytically for purposes of
machine c o m p u t a t i o n .
Several investigators have suggested t h e use of a con-
s t a n t effective y t o account roughly for real gas effects.
This idea was tested b y s t a r t i n g from t h e observed
shock for a hemisphere flying a t M = 14.2 (reference
47), which clearly involves significant gas imperfec-
tions, and a t t e m p t i n g to reproduce t h e body. The
observed standoff distance is obtained b y choosing 7 =
1.35 which according to section (3.5) m u s t be the ef-
fective value in t h e vicinity of t h e sonic line. Fig. 18
shows t h a t t h e b o d y shape is t h e n closely reproduced.
A t least in this case of mild gas imperfections, t h e idea
shows promise.
REFERENCES
Potential Flow Approximations
1
Laitone, Edmund V., and Pardee, Otway O'M., Location of
Detached Shock Wave in Front of a Body Moving at Supersonic
Speeds, NACA R M A7B10, 1947.
2 24
Nagamatsu, Henry T., Theoretical Investigation of Detached
Shock Waves, GALCIT Pub., 1949.
3
Kawamura, T., On the Detached Shock Wave in Front of a Body
Moving at Speeds Greater Than That of Sound, Univ. of Kyoto,
College of Science, Memoirs, Ser. A, Vol. X X V I , No. 3, pp. 207-
232,1950.
4
Heybey, W. H., Shock Distances in Front of Symmetrical
Bodies, NAVORD Rep. 3594, December 24, 1953.
Taylor Series Expansions From Shock
5
Lin, C. C , and Rubinov, S. I., On the Flow Behind Curved
Shocks, J. Math, and Phys., Vol. 27, No. 2, pp. 105-129, July,
1948.
6
Dugundji, John, An Investigation of the Detached Shock in
Front of a Body of Revolution, Journal of the Aeronautical Sciences,
Vol. 15, No. 12, pp. 699-705, December, 1948.
7 28
Lin, C. C , and Shen, S. F., An Analytic Determination of the
Flow Behind a Symmetrical Curved Shock in a Uniform Stream,
NACA T N 2506, 1951.
8
Melkus, H., Uber den abgelosten Verdichtungsstoss, Ingenieur-
Archiv, Vol. 19, No. 3, pp. 208-227, 1951.
P R O B L E M 495
9
Cabannes, M. Henri, Contribution a Vetude theorique des
fluides compressibles, Ecoulements transsoniques, Ondes de choc.
Chapitre I I I . Etude de l'onde de choc detachee au voisinage de
son sommet. Ecole Normale Superieure, Annales Scientifique,
Ser. 3, Vol. 69, pp. 31-46, 1952.
10
Cabannes, Henri, Determination theorique de V ecoulement d'un
fluide derriere une onde de choc detachee, ONERA note technique
no. 5, 1951.
11
Cabannes, H., Tables pour la determination des ondes de choc
detachees, La Recherche Aeronautique, No. 49, pp. 11-15, Jan.-
Feb., 1956.
Incompressible Approximations
12
Hida, Kinzo, An Approximate Study on the Detached Shock
Wave in Front of a Circular Cylinder and a Sphere, J. Phys. Soc. of
Japan, Vol. 8, No. 6, pp. 740-745, Nov.-Dec, 1953. (Also Vol. 10,
No. 1, pp. 79-81, January, 1955.)
13
Lighthill, M. J., Dynamics of a Dissociating Gas, Parti, Equi-
librium Flow, J. Fluid Mech., Vol. 2, Pt. 1, pp. 1-32, January,
1957.
14
Whitham, G. B., A Note on the Stand-Off Distance of the
Shock in High Speed Flow Past a Circular Cylinder, Comm. on
Pure and Applied Math., Vol. 10, No. 4, pp. 531-535, November,
1957.
15
Hayes, W. D., Constant-Density Solutions. Circular Cylin-
der. Ch. 4 of Hypersonic Flow Theory, by Hayes, Wallace D.,
and Probstein, Ronald F . ; Academic Press. (To be published).
Newtonian Approximations and Improvements Thereon
16
Busemann, A., Fliissigkeits und Gasbewegung, Second Ed.,
pp. 275-277; Handworterbuch der Naturwissenschaften, Gustav
Fischer, Jena, 1933.
17
Ivey, H. Reese, Klunker, E. Bernard, and Bowen, Edward
N., A Method for Determining the Aerodynamic Characteristics of
Two- and Three-Dimensional Shapes at Hypersonic Speeds, NACA
T N 1613, 1948.
18
Hayes, Wallace D., Some Aspects of Hypersonic Flow, Ramo-
Wooldridge Corp., January 4, 1955.
19
Hayes, Wallace D., Hypersonic Flow Fields at Small Density
Ratio, Ramo-Wooldridge Corp., M a y 12, 1955.
20
Serbin, H., Hypersonic, Non-Viscous Flow Around A Circular
Disk Normal to the Stream, RM-1713, Rand Corp., M a y 3, 1956.
21
Serbin, H., Hypersonic, Non-Viscous Flow Around a Sphere,
RM-1772, Rand Corp., August 13, 1956.
22
Probstein, Ronald F., Inviscid Flow in the Stagnation Point
Region of Very Blunt-Nosed Bodies at Hypersonic Flight Speeds,
Brown Univ., Div. of Eng., September, 1956. (Also pub. as
W A D C T N 56-395, 1956.)
23
Chester, W., Supersonic Flow Past a Bluff Body With a De-
tached Shock, Part I: Two-Dimensional Body, Part II: Axisym-
metrical Body, J. Fluid Mech., Vol. 1, Pt. 4, pp. 353-365, October,
1956; Vol. 1, Pt. 5, pp. 490-496, November, 1956.
Freeman, N. C , On the Theory of Hypersonic Flow Past
Plane and Axially Symmetric Bluff Bodies, J. Fluid Mech., Vol.
1, Pt. 4, pp. 366-387, October, 1956.
25
Li, Ting-Yi, and Geiger, Richard E., Stagnation Point of a
Blunt Body in Hypersonic Flow, Journal of the Aeronautical
Sciences, Vol. 24, No. 1, p p . 25-32, January, 1957.
Explosion and Similarity Approximation
26
Lees, Lester, and Kubota, Toshi, Inviscid Hypersonic Flow
Over Blunt-Nosed Slender Bodies, Journal of the Aeronautical
Sciences, Vol. 24, No. 3, pp. 195-202, March, 1957.
27
Cherny, G. G., Hypersonic Flow Past an Aerofoil With a
Slightly Blunted Leading Edge, Doklady, Akad. Nauk SSSR, Vol.
114, No. 4, June, 1957 (Russian)
Numerical Methods
Garabedian, P. R., Numerical Construction of Detached
Shock Waves, J. Math, and Phys., Vol. 36, No. 3, pp. 192-205,
October, 1957. See also: Lin, C. C , Note on Garabedian1 s
Paper (<Numerical Construction of Detached Shock Waves," pp.
206-209.
 496 JOURNAL OF THE A E R O / S P A C E
29
Garabedian, P. R., and Lieberstein, H. M., On the Numerical
Calculation of Detached Bow Shock Waves in Hypersonic Flow,
Journal of the Aeronautical Sciences, Vol. 25, No. 2, pp. 109-118,
February, 1958.
30
Zlotnick, Martin, and Newman, Donald J., Theoretical Cal-
culation of the Flow on Blunt-Nosed Axisymmetric Bodies in a
Hypersonic Stream, AVCO Mfg. Co., Rep. RAD-TR-2-57-29,
September 19, 1957.
31
Belotserkovsky, O. M., Flow Past a Circular Cylinder With a
Detached Shock Wave, Doklady, Akad. Nauk SSSR, Vol. 113, No.
3, pp. 509-512, 1957 (Russian).
32
Uchida, Shigeo, and Yasuhara, Michiru, The Rotational Field
Behind a Curved Shock Wave Calculated by the Method of Flux
Analysis, Journal of the Aeronautical Sciences, Vol. 23, No. 9,
pp. 830-845, September, 1953.
Other References
33
Courant, R., and Hilbert, D., Methoden der Mathematischen
Physik, Vol. 2, p. 39; Julius Springer, Berlin, 1937. (Also pub.
by N. Y. Univ. Inst, for Math, and Mech., 1950-1951; and pub.
by Interscience Publishers, 1953. English text.)
34
Van Dyke, M. D., A Model of Supersonic Flow Past Blunt
Downloaded by 123.2.138.201 on July 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/8.7744
Axisymmetric Bodies, With Application to Chester's Solution, J.
Fluid Mech, Vol. 3, Pt. 5, pp. 515-522, February, 1958.
35
Staff of the Computing Section (Under the direction of
Zdenek Kopal), Tables of Supersonic Flow Around Cones, Tech.
Rep. No. 1, Center of Analysis, M.I.T., 1947.
36
Lees, Lester, Hypersonic Flow, IAS Preprint No. 554, 1955.
(Also C.I.T. pub. 404.)
37
Ames Research Staff, Equations, Tables, and Charts for Com-
pressible Flow, NACA Rep. 1135, 1953.
38
Bickley, W. G., Formulae for Numerical Differentiation,
Math. Gazette (London), Vol. 25, pp. 19-27, 1941.
On Optimum Thin Lifting Surfaces at Supersonic Speeds
{Continued from page 479)
20
Graham, M. E., Examples of Calculation of Minimum Super-
sonic Drag due to Lift by Solution of Two-Dimensional Potential
Problem, Douglas Report No. SM-22754, March, 1957.
21
Heaslet, M. A., and Fuller, F. B., Drag Minimization for
Wings in Supersonic Flow with Various Constraints, NACA T N
4227.
22
Tucker, W. A., A Method for the Design of Sweptback Wings
Warped to Produce Specified Flight Characteristics at Supersonic
Speeds, N A C A R M L51F08, 1951.
23
Graham, E. W., Approximation of Optimum Lift Distributions
from their Spanwise Moments, Douglas Aircraft Report No. SM-
23021, December, 1957.
24
Beane, B. J., Calculation of Optimum Lift Distribution for the
Sonic Edge Diamond Planform Wing from Spanwise Moments
Obtained from the Corresponding Artificial Singularity Distribu-
S C I E N C E S —AUGUST, 1958
39
Hyman, Morton A., Noniterative Numerical Solutions of
Boundary-Value Problems, Appl. Sci. Research, Vol. B2, No. 5,
pp. 325-351, 1952. (Also pub. as NAVORD Rep. 1813, Aero-
ballistic Res. Rep. 26.)
40
Heberle, Juergen W., Wood, George P., and Gooderum,
Paul B., Data on Shape and Location of Detached Shock Waves on
Cones and Spheres, NACA T N 2000, 1950.
41
Oliver, Robert Earl, An Experimental Investigation of Flow
About Simple Blunt Bodies at a Nominal Mach Number of 5.8%
Journal of the Aeronautical Sciences, Vol. 23, No. 2, pp. 177-179,
February, 1956. (Also available as GALCIT Pub. 386 and
Memo 26, June 1955.)
42
Crawford, Davis H., and McCauley, William D., Investiga-
tion of the Laminar Aerodynamic Heat-Transfer Characteristics of
a Hemisphere-Cylinder in the Langley 11-Inch Hypersonic Tunnel
at a Mach Number of 6.8, NACA T N 3706, 1956.
43
Ladenburg, R., Winckler, J., and Van Voorhis, C. C , Inter -
ferometric Studies of Faster Than Sound Phenomena, Part I. The
Gas Flow Around Various Objects in a Free, Homogeneous, Super-
sonic Air Stream, Physical Review, Vol. 73, No. 11, pp. 1359-
1377, June 1, 1948.
44
Love, Eugene S., A Re-examination of the Use of Simple Con-
cepts for Predicting the Shape and Location of Detached Shock
Waves, NACA T N 4170, 1957.
45
Gooderum, Paul B., and Wood, George P., Density Fields
Around a Sphere at Mach Numbers 1.30 and 1.62, NACA T N
2173, 1950.
46
Busemann, Adolf, A Review of Analytical Methods for the
Treatment of Flows With Detached Shocks, NACA T N 1858, 1949.
47
Seiff, Alvin, The Use of Gun-Launched Models for Experi-
mental Research at Hypersonic Speeds, AGARD Report 138, July,
1957.
Hon, Douglas Aircraft Report No. SM23017, December, 1957.
25
Germain, P., Sur La Determination Locale D'une Aile Opti-
mum En Regime Supersonique, La Recherche Aeronautique, No.
60, September-October, 1957.
26
Yoshihara, H., and Strand, T., On Jones' Criterion for Mini-
mum Drag Wings with Subsonic Edges, Theo. Aero. Note No. 9,
Convair (San Diego) (to be published in the Readers' Forum of
the Journal of the Aeronautical Sciences).
27
Zhilin, Yu L., Minimum Drag Wings, Prikl. Mat. i Mekh.,
Vol. 21, No. 2, pp. 213-220, March-April, 1957.
28
Germain, P., and Gibault, R., Quelques Resultats sur les
Ailes Delta Portantes a Trainee Minimum en Regime Superson-
ique, Breves Informations, La Recherche Aeronautique, No. 61,
pp. 54-56, November-December, 1957.