Beruflich Dokumente
Kultur Dokumente
INSTITUTEOF TECHNOLOGY
CRANFIELD
Schoolof MechanicalEngineering
PhDThesis
Year 1987-88
Academic
A, HADDAD
SUPERSONIC DESIGN
NOZZLE OFARBITRARY
CROSS-SECTION
ABS7RAGr
Acknowledgments
LIST OF CONTENTS:
La gt
.
Abstract i
..................................................
Acknowledgements................................ ii
La
Section 6: Experimental 81
apparatus and procedure ..........
6.1 General ....................... 81
0*0000000000000.. 0.000
6.2 Experimental apparatus 81
... oo. o ..... oo.. ... o.. **o. **
6.3 Experimental procedure. 87
-o ....... o...... o.. oo. o. o-o-o
Pape
supersonic flow.
Appendix B: Modified Euler predictor-corrector algorithm
based on the average property method.
Appendix C: Subroutine 'WALCON'.
Appendix D: Boundary layer computation.
Appendix E: Starting velocity profile, starting enthalpy
profile and initial pressure gradient.
Appendix F: Mass flow calculation methods.
Appendix G: I-Deas.
Appendix H: Phoenics.
Appendix I: Thrust and thrust coefficient.
Appendix J: Subsonic section design of the elliptical
nozzle and supersonic section design of
the two-dimensional wedge nozzle.
vi
LIST OF FIGURES:
PaRe
NOTATION :
a speed of sound
A* area
A throat area
A Attachment angle
Aa exit angle
Ae coefficient of the wall contour parameter
w
B: coefficient in the law-of-the-wall equation chapter 2.4.4
Bw: coefficient of the wall contour polynomial
r radius
R specific gas constant ( except in equation D. 12 where it
represents the parameter defined in equation D. 16
R nozzle throat downstream radius of curvature
td
R nozzle throat upstream radius of curvature
tu 6 ue /\j
R6* Reynolds number based on displacement thickness,
T: temperature
T(+); T(_) : coefficients in finite-difference equations ( Appendix
A ).
Greek notation :
,
ot : molecular diffusivity or angle
(I effective heat conductivity
e:
y: specific heat ratio
CF; CF
11 2 : coefficients in the compatibility and characteristic
equations ( Appendix A ).
*
n: non-dimensional coordinate normal to the wall, y/6
lp : stream function
density
T: shear stress
T: non-dimensional effective viscosity, VAu
T9: non-dimensional effective conductivity, Ke /T Ue
: ambient state
)a : state of mercury
)tg : total state.
)t, : total state at the edge of the boundary layer
)th e : state at throat
: wall state
w : known or calculated point
)2 : known or calculated point
: solution point
4 : predicted value at solution point
T : corrected value at solution point
: along left-kand characteristic
: along right-hand characteristic
1
SECTION1: Introduction
1.1. General
1.2.1 Applications
1.2.2 Scope
three-dimensional nozzles.
This thesis has been divided into nine main sections. While the
first briefly introduces the method investigated and its main
applications, the second section reviews some of the literature which
has been used in the development of the research. This includes
analytical detail as with the method of characteristics ( for main
flow computations ) and boundary layer calculation methods, as well as
a broader view of the concept of integrating nonaxisymmetric nozzles
into fighter aircraft and its applicability.
Section 3 describes the specific implementation
of the method of
characteristics as well as the computations of the inviscid flows
within the two axisymmetric contours. The subsequent section deals
with the calculation of the viscous boundary layer, under the
assumption that the viscosity is small enough for its influence to be
confined in the immediate neighbourhood of the walls. Its implications
for the inviscid flowfield are assessed.
Section 5 shows how, using the method investigated, the
axisymmetric nozzles are transformed into nonaxisymmetric ones.
Section 6 and 7 introduce the studies undertaken to validate this
design procedure. Section 6 will deal with the experimental programmeý
involving design, manufacture and testing of a supersonic nozzle of
elliptical cross-section whilst that which follows investigates the
three-dimensional simulation using a grid generation code "I-Deas" and
a generalized flow analysis code "Phoenics".
Results,. comparisons and conclusions are presented in the last two
sectims.
5
. I. I. a
v V1
PI
pp
Fig. I. I. b
vv2 V3
Fig. I. I. c
2.1. General
As can be seen from Figure 2.1 which shows the comparative length
as a function of the area ratio for various types of nozzles, the
ideal nozzle that gives maximum thrust performance is particularly
long and consequently heavy.
44)
so
INAL NOZE169
LIN, to
is c""'co" MEW N
ovfww ceftwoun
00
Fig. 2.1: Length compaWon
o
oLus so '9-r *61IL9
o6 va4iou.6 type.6
o6 nozztu.
"I 10 to so 40
ý Ve + Pe Ae
Cfv, 2.2
conical Pcc Ath
In the case of small rocket engines fitted with small area ratio
nozzles, simple fabrication methods are prefered and it is common
practice to use conical nozzles. Usually, a half cone of 15* is used
for the divergent section ( Figure 2.2 ).
Figure 2.1 shows that the length of a 15* conical nozzle is less
than that of an ideal nozzle. 7he decrement in thrust coefficient of
the latter, when compared to the uniform exit nozzle, is only about
1.7%.
'ce R
Rt
Re
The throat
of such nozzle, as shown in figure 2.6, is located as
an annulus at the outer diameter. At the outer edge of the annulus,
the exhaust gases expand abruptly to the ambient pressure. The plug
surface may be designed in order to produce uniform exit flow parallel
to the nozzle axis. The cowl-lip diameter would be the same as the
exit diameter of a uniform flow of a Con-di nozzle expanding the gases
to the same ambient pressure. However, the plug-type would be much
i
13
1. ____
-T
R -
- PLUG OZZLE
I-IT - 1-
%: ..
,
t4- 1111k
00 6
1 -ý4:; oo-, ý .
PLUG
NOZZLI
EXPERIMEMIAL
,
POINIS
1.2 , WITNSEPAIMON 6-1.40
"* +
'
IWITNDUTSEPARVIOl
goo
are the cause of the thrust increment of plug nozzles over that of a
Con-di one.
Re
y(x) -Aw+Bwx+Cwx2
where the coefficients Piw, Bw, and Qw are determined by specifying the
attachment angle, the exit radius and by requiring that the polynomial
contour attach continuously to the circular-arc initial expansion
contour.
In order to assess whether such a simulated shape pan develop
thrusts comparable to the ones developed by the calculus of variations
contours, the authors compared thrusts developed by the latter
contours ( Rao's method ) with the thrusts developed by polynomial
contours. It was found that both methods predict essentially the same
maximum thrust ( i. e. for zero ambient pressure, agreement was within
0.2% ) justifying the new proposed approach.
7his approach was used to simulate the contoured wall of the
axisymmetric nozzles investigated in this study.
2.4.1 General
non turbulent fluids at the outer edge of the layer into the turbulent
region. Green et al, 1973 ) used the entrainment relation as an
additional equation to close the boundary layer set of governing
equations.
u+-A Ln y++B
where :
UT Tw
U+y+y U. )k
U-T ,vw T Tw- and
following relation :
"1
.8 M-corm
ISDUUM
1.35
T-8 A0 ýML
M-Al
Am
lz- *a" .7 -
I
.4
'y 0.1
35 3L 3
Npol NPR
24
Phase I Static Test JNASA-L&RC) Phan 2 Wind Tunnel Test (AEDC I ST)
0 Nozzle Design/Fabrication 9 Nozzle Configuration Effects on
and Static Performance Jet/Astframe Interactions
Characteristics
Metric Brooks
Aft Fwd
Test PcW
GE ALIBEN GE 2-D 00
0 Metric Nozzle/Attbody +Wing
0 Non-Metric Fo(sbody
P&WA/NASA
Plug
P&WA
2-D OD
Reverser
Mode Shown)
Air-to-Air Air-to-Surface
Air 11 rf
P&WA/MCAIR
Variable IncidencePlug (VIP1
I
-7 VARIABLE CENTERBODY
WEDGENOZZLE
44PANELS)
DLtMLPtLl
DUCT
SIDEWALL
AFTBODY
EXTENSION
-4zol .. 111.1
-*111, ýI
DUCT
TRANSITION
VARIAB E
'THROATPLANE WEDGE
CONSTANT CLOSURE ý
SOUIAREDUCT 12PANELS)-,,
EXIT PLANEO'SICTION
BEFORE
Cl AFTBODY AFTER
STRuCTURE
MODIFIED
2.6.1 General
Static
pressure is the pressure indicated by a measuring device
MM Poo
-
Boundary layer
M/ý/, /7777,
T"T"MIM-111111111
0 Pressuregauge
To manometer
or gatige
Fig. 2.14. a
P
Af., >I
Lg. 2.14. b
Pi.
p=
M.
12 =--0- To manonieter
To gatige or manomeler
dr
3.1 General
Rtu Rt Aa
Ye
Yt Ya
x
Xe ol
Equations governing the flow described above are : the gas dynamic
equation ( Equation 3.1 ), the irrotationality condition ( Equation
31
2) 2) 3v a2v
u2 -a _Lu +( v2 -a +2 uv -Lu -6 -03.1
ax ay ay y
au av ( 3.2 )
ay - 5x M0 oet
u, v ) ( 3.3 )
ooo
ox
. 40 0
x P: u
ell
aý m 0
u tn
b3
In po
P
Id VA
0) 64 14
0
Z
........... V. 6 '6
4; C;
A snlovd 37ZZON
1%E
0
U.
0
0.-
0
ix
:9
j u
1%
14
04 0
u F4
u
------- ------ ....... ...... .........
IH
u
sc
$4 0
I co od
94 14
C3 u 04
M
V)
IH
V 0
1.3
tc
14
bd 0:
od C)
0 0
I co
0
u
C14
-------- ---
(4 44 0
3.4.1 General
LL
CP
w j
W
0
z
0
u
cc
.......... to
u
6ý
14
I I X: to
.1 04 u
z
co
U) 0
C3 u sn
04 W
I " to
63
U
04 64
t4
0
rd
14
14
0
fn
0 C. 0 ci ci 66 C,
69 1ý C! NN--
1ý ej 't .9
m 40 C,
> W w
1.- 39 LL
LL U. U.
F5
I.:
Q
dc -C
it
-K
ac J
ki 00
;z tu
I , I. I,
I,
II I " I I .
. ld
ca
.7, j
64
b3
j.
j I
IC U
eq
t -&/I III, tI *
-*.. * .. 14
. IIII.. ... ... ..., tq
63 mg
as
Y!
In
z0 ý3 000000 0-
1ý 9 llý Ci Cý lq: 9
(n (n C4 N C4 --a00
READlN?L'r DATA
0-
ALMN
IMITE DAII ,
SIARLI%t II
II INTPOIK II
11
A.'klnl\l II
No W. Wm Pull.!M 01,
I%ITIAL LlbZ REACHEI)
.
yes 6RCULAR-; WC
FLO; IELV FROM
THROATCONrOUR.
I 6all
IS WMM PRESPECIFIM\
NO
K.ANITS HM ?
S REACHM
REAC.
rL
FLOWFIE0 FROMTNF
T'll 31
SIC -M DEIt
SECOND-ORPER
Yes
PýO: WALL
LNMOOXZALý
m OR AXIS OR
INIIERIOR,
AMPOINT Ax Interior
"t
6ýALLPOIN7 ? IN17PPOINTr
I
6all
.Vpol..
IS EXIT LIP
POINTSEACHED?
ýIMITE
TS
ývmý
determined.
Investigations carried out for the last ten yearst involving the
integration of the propulsion system with the airframe in a fighter
aircraft ( see chapter 2.5 ), 1ýave shown it is possible to improve
fighter performance by incorporating non-axisymmetricýnozzles. Because
of this particular application for which the two-dimensional wedge
configuration was designed, a different approach had to be followed in
the determination of the starting parameters, which permitted the
computation of the axisymmetric nozzle to be transformed into a
two-dimensional wedge nozzle subsequently.
0/
7-14040
Saw
=
K16.981
0 Pwkýfs co lack. "A
'Rafmomn 22and 231
Fig. 3.7: Compa4i. 6on oj the
dij6e-tent tAanzonic
itow ptediction method.6
in te4mz oj the th4oat -
concenUation jactoA. 0.00
0a3 #I ly,
sections. The first one permits a sharp expansion and is situated just
after the throat. Ihe second part, where the flow is still
accelerating, tends to straighten the flow to achieve a near axial
direction at the exit. Ihe maximum angle of divergence is the contour
angle at the attachment point, joining the downstream circular-arc
contour to the quadratic wall. A high value for the contour angle
increases the risks of flow separation from the walls. A small angle,
however, may not permit the complete expansion of the gases within the
desired length and would result in a longer nozzle and hence a weight
penalty. 7he contour angle was chosen equal to 15%
Tt 4030K 11009K
5 52
Pt 15 10 NIM2 3.26 10 N/m
2
Pa 1.013 105 N/m 1.013 105 N/m2
Yt 0.020 m 0.325 m
Rtu 0.040 m 0.650 m
Rtd 0.010 m 0.162 m
Aa 15" 2"
Ae 5* 2*
Xe 0.060 m -
Ye - 0.345 m
Solutions for the two axisymmetric nozzles ( for both elliptic and
wedge designs ) are represented in figures 3.8 and 3.9 respectively.
Both the nozzle wall and centreline pressures are plotted in each
case.
Due to the higher value of the attachment angle of the f irst
configuration, a sharper expansion took place near the throat ( along
both the centreline and wall ). Within the axisymmetric-for-. iedge
nozzle, the flow tends to expand gradually from the throat to the
exit.
45
F
-*U
I
'.
47
4.1 General
6* -f0(1- u) dy
(» -0
Oeue
a ( m )+ L ( rv )-0 ( 4.1 )
7x 999
ay
ap 0 ( 4.2
VY . ...
49
;T
ay
T-V Tx + *so ( 4.3 )
Dht 3ht
+ ý; 7 + U-f ( 4.4 )
ax ay ay
uv ( 4.5
p see
K ah 4.6
ay - vh'
ýu
- ulv, - il t ay ( 4.7 )
ooe
3h
- v'h' mK t DY o9o ( 4.8 )
uu
vý... ( 4.9
ay e ay
50
(+K) ah- K gh
-9- *es ( 4.10 )
ay ay
1a( 0 ( 4.11
rx r pu +L( pv
w ay so*
rw
LLI ;u 2ue
PU + pv %U +L(r T) ( 4.12 )
ax ay e ax
;y rw so*
aht a ( r ; ht
Pu Tx- + pv ýY- soTy (q+ UT 4.13
rw
where : au
r ve ay 4.14
rw
rK ah 4.15
p rw e ay *so
4.4.1 General
4
-m
=M _r
x 1-
CL) rj
eine
SA CL,
A
1 Z -122,
P.:
ch.f..r
cq
41 aj
Co
N
Q
Ike
w
LL.
Cm 3:
C
IA
-ALI Cy-
0
m
ý- ",-I r= E
:' j
' ýE
Ew
ý-m-. =
Q,
=u .-0
aj. ý Mu
je lik. 0
cn 0
1 to
t:
.to C,
06
4A CL
IA
4)
Ca.
CP e V) --112
2E
54
Results
of the experimental investigation are shown in tables
4.2. a, 4.2. b, and 4.2. c. 7hey, respectively, represent the three
stations at which the measurements were performed. 7he theoretical
predictions are presented in tables 4.3. a, 4.3. b, and 4.3. c.
7he initial velocity profile, the 1/7-power law along
representing
with the experimental profile are shown in figure 4.4.7he
experimental and predicted profiles at the two other stations are
shown in Figure 4.5 and 4.6. Good agreement is found between the
predicted and measured profiles within the boundary layer which
developed along the nozzle.
Figure 4.7 shows the distribution of the boundary layer
displacement thickness stations and those predicted by
at the three
the program. 7he experimental value of the displacement thickness at
station Ill has been taken as an initial condition in order to carry
out the computations.
7he development of the boundary layer seem to have been well
predicted by the program. Good agreement between theoretical
predictions and measured values is found. Calculated errors were
within 1.5%.
uz
0- w
c -C 0
6.1
U3 W
14
[a 0
1
R too I&
04 u
co a: m
4d M
b ............. 0) 8
to- 0 sw
go
m
0
z
14 L64
(D b. 14
.A.
C; 0 a
- co a. w
U
0 Ir
64
--------C6--------iC6 .. ...... 60
19.
1ý li le Iq 9 V! 9 V! C!
-r V. en fn
*-f/J 31 VNIO&OOOWrOVO
56
57
jo
ii
0.0%
uv
61
W
123
w
04
ld
0
0 P%
le
C4
C)
C4
14
OC b3
a ta
Mc, 0 0
0. v:
w 'm
a
t4
llý 9 llý 19 C;
; co (D in 1; 1; N-a 0
SS3NXOIHJ JN3U33V'ldSlO
64
3 69
0.0 0.0710 0.1421 0.2132 0.2843 0.3554 0.4265 0.4976 0.5687 0.6398
0.7109 0.7320 0.9531 0.9242 0.9953 1.0664 1.1374 1.2085 1.2793 1.3501
1.4218 1.5640 1.7062 1.8484 1.9906 2.1328 2.2749 2.4171 2.5593 2.7015
2.3437 2.9859 3.1281 3.2703 3.4124 3.5546 3.6968 3.8390 3.9812 4.1234
4.2656 4.4078 4.5499 4.6921 4.8343 4.8400 4.8450 4.8500 4.855 4.86
4.8650 4.8700 4.875 4.88 4.885 4.8900 4.8950 4.9000 4.905 4.91
4.915 4.9200 4.925 4.93 4.9350 4.9400 4.9450 4.9500 4.955
45
1.0000 0.5974 0.5449 0.5096 0.4821 0.4592 0.4394 0.4217 0.4057 0.3910
0.3773 0.3646 0.3526 0.3412 0.3304 0.3201 0.3102 0.3007 0.2915 0.2826
0.2741 0.2577 0.2422 0.2274 0.2133 0.1998 0.1868 0.1742 0.1621 0.1503
0.1389 0.1278 0.1169 0.1063 0.0960 0.0359 0.0759 0.0662 0.0567 0.0473
0.0381 0.0290 0.0201 0.0113 0.0027
6
0.1813 0.0672 0.0442 0.0290 0.01915 0.00152
63 -2 -1 2.1 0.086 9496.0 1.05 535.0 1.0
31.01 2.1 1.0 0.0 26.0 0.0 0.0 0.0
138.89 3.7 1.0 0.0 55.0 0.0 0.0 0.0
241.81 4.4 1.0 0.0 82.5 0.0 0.0 0.0
65
41
Ir qj tr u0
1-4 cd
; ; Z
CC,C WC,0 C,C, C, C,J, C,a0a
Ei d
"d
f"
"&P.
OD
ý .4P. .., lo al . .0, fl- r, n. P. ry
10 'a F" Ir in Ill. 410 01 0 rm Ir 0 ry Ir It al
C: 17 " 1ý 1ý 9 a-
%
.Ii .
tu Co LM le lo 10 10 10 1- f. - r-
x: M
e
-r It Fft V- V- co co In W&&c
cr)
9 ci
14 .b
1093 ;8 ; ;
C;C;c;f; l , , l;44z4 ,; zz gC;
01 Wa r4 'o 't ::,:, l r r, &0
mWW If M% 0 p- 00
'a 00
&1 0.0 0
Co
9-
L) mf :1
-1 1ý11-% ID * 1
ý 0.
V: 1ý1ýicý C', , V.
0:0:0: ý1ý99
La
.. .. .
Z >
i
tu
3 9 2 "- lllt ""ý lll, W -4 lll s"- . 4 ll, # N t v ry 4 NNf yf -4 r y t y
cb
0,
Q .
-t "" -f UN 4P IP -IP W4 10 0
V 9
Olt 41,
41, ir kn In a dD .4 10 10 qj
a
a 4D Q Iz .0 .010
0,
Ic p- r- p-
4 10
ci 10 0
A
EI % 4)
4) 9 02
e* cj
;0 13
(L) -
-C -TUNIOP-00, UNWO. a Ok It NW%Ok lowly,
w lo lllo fn 4D V- Ill, llll w 0. , 04 W ", llll
ryw%wfow
a, 01 C, "" -0
r.
0 u -e ry 1 ell r,, tv ry ry In M f"
l> Co 1. f"1
4 f-4 ev 14 FM FY ev r yN
bo
0 4)
ý144 41 fm CO ty D 't g. ml If- f. - Q. 'r 0 PY 4D a, 0 -0 a -d
g C- 4; bd "I 1- 31 ID " 1
Ir f .1tr
0000
11 UN tr J% kf% . ..,l" ...
1
If% .
Nry NVY Ol fn 'r
Y % UN kn um um -A% V,
ill
br.
cu ..
1 .
cz CD, 41
e ß' f", o JC
C, 0c, lop. 4p 09-t
I'll N ry I'm f4 ry r" f.
'Oolllle-o-Ofn
m o%, r r W%-M o
cg
0,10
000000000000CO0000000. "
lob
a, le -r r -t 't 'r v.
fV On W0 OD W Ir CP r Ift :! 00 000 IP CP
ir cc
". 0
9 9
0 0 *00 9 9 999
990 00
66
40 0 0% td%p. 0- m0 -0 0 fm 0 p- o cp Z
1.4,8.4,
l«, ", c2
4.1 1. l" 10 i-. 0: lý lý W, c: 9 c: lý je lý lý W. ll:
j000 .. . .
0 90 00 C) c; C) 0u C) 00 gý ý-------
r cryfme t Omen
CUM- 4 w%c- m0ee
c2, C, ein
Co0%f. ;
0 p.
o0
0N
UNcy.'0
enr
m
e 04 cr
e
c3.
c; 0 4;c;c;c;4;c;L,Z000c; o c;,ý 0J ty4
, 0
0emr
e ty i- e0Nr, Z ic f-. ry 0 fn e F- rw0 rm
6 .. --. C, 01 fn UN 0 :D e e t- u- e& (> 00
Z: 10 -r
. fft mAm
r- 43 IM ,5
a-- CD, Z
690 u% en 0o r- ic e 0. cp Co cp ee cp & 0. o c)
-0
. r4 P-4 -r
E., C6 c4
.. c; -.
jZc;
r;
ý
:; 0jý (ý
ji0c;
0090Z cý
jZýZ
-* -,
0
r0C. 0 0 :: e00 f4 :: 0& 94 e &^ C. fy e F-*w% e F. Co e 0,0 o0o
Z:
2CP
c72 0, ty , tr «%. n f- 90 ein 0% * f. 0-eýe0
e
in &&0, CM v. 0000
Co Ai . 00C000000----
41
tu oc.. ci
m UN ee CM C) N p- e r. c ce ce 0,
Co
-f-4
12.0,
0 90
;ý
(V. a ZUl» 9
>% Co -M *
r
0ee0N00 NNNNNNN
10 p- 40 Co ee c- e
we 0,0
rre c,w% w
w%44 f"
000r. * N,-l0mg-!! C) NA en.
r ',' P l, 0,
p P-e. cum1 0w cr
-r -r Ir e Ir -r Ir , I , IM td%tel% fl, «lo
" u% , «,
. 11 . ,
ul, ',I A, ir"
ELA"',
w% ý ip qm r. - tu "% 0 gý - f-
=% b4 bm r Kp ý cm rw F- f- (P- wr fý rt m fn fm M on M
A -Q NP Kp Co lu f. ýý v% w% r
44 Ilb U
ty f4 N ewty ri ei r4 ry r4 e4 r4 ew ZZZ- Zo Z
0%> F- 2 u%Z; 9 e r- rw UN:; -4 fn r- 0- 1- 40 c2 0
10 fm g4
u% ne
CD
j% -
Co C UN r
-r rereere
r4 ý oý bfl f4 Co 0 10 w%b%bINAe W%e %M
«%«t -t «t -4 ry ty r4 ty -4 ew ewly -w -i
cb 0 @. ec -0 0% m%CCDCo o 0% e Co eoe
10 l" 'm j% f. 00 ein UN p.
Q0
1. " ,p ýll lý
22 -- 0 fm
LZ %4 tt 9999 .0
c) o0000C,: o. 0 e4 0'
it
0000,000000
e
G
".0 --'. C
c- E , rl re
co
40 ý
.wIII
Uý V. ý le
lZ
ro -0 C). %P cy. W* CU%., # 41 w 4-1
Q. Cc
l C'
c
CP wý P- cc ýo 0 P- 1- 10 40 I- c
ti , tr ftj 00--w fli 4. in
(0 t-
G,
O z
o
f
Of- mu%o onkf%;: Lý Uý C' Nil' C I- aDfOWW WO
-6 41 .9
Q. al
64
Z Z. UN 4NN 1- 0N 47ý 0& In %t 64%0 1- 0
Na , IM WN CD ;: t %0.0 Pft ; fn 10 P. a 10 o l 0ý V. (), a , 0
w v
"4 P-4
94
Cb tn
Z; ;ý Iz.
0ý .w aI
E4
:0 r -
iý 1: ".ý%ýýý 0.1iýýýtt
.
B C'3 C;
U.
ý fý 9 In on r P.
0 V. ul %A N %ft CY 30 41 ID 6P u0P. N0N
ca
"4
4)
e I- U*%
W
ýo 0 Fn .0 -Z 0 -r c
4A 4% -0 .00 t-
ýr ý (), a :;
CO QJ OD CP
N ey NN Py N
LZ
9w Li Cd 0) -T %? - LO%a %r 04 Ir co -f rý zV aD ul 1-1 0 %&)I% ;:
# .0 In :
.0(13 ým P.- aD Iý 0 In 4 P. ull ty. NN FM
0 ;ý N4 Ic M a, ccc0
14 N ry N In Irl en W%
co
co
v-4 1-4
0 I I "1%0"20 -A) P% ()ý t- N It ty -ý W, N V, :9 In w 11%Lrs UN 10 0.
t, 44 tia E C w co00 -n % Ir Ir 0 .0 ?ý 0- 0 -, NN
.1 -r If
In '" "' ; f. In In
r IT W %T V% IF %M V% u l%V, V% 0 , UN tr 0
5
cc ch .
ll, C,4 CO C)
E a. w rý 4r fý 0N PIN on, It -J%0
0
Wv0W, GLI -N %r In V, U, 41 4J 4) 4, Wf ý
4-4 G
f- OD no mm go 10% CP a. 0 , 0. V , CP Oý
09 0
qi 0
0. 0 0
1%,tp In 0 In Nu-, IT lu v0 cu cu a, .0 Ir ry L;
10 $4 %P a, rý W, T P, #No- tr v JN In
ci In In
0) V
C4 I_,
0)
CL
S
I..
-4 Vý in fý a, 10 In co -T Nm
-I a- =0 'r Vý u% N ,; I v. a. w7
Cd It
W, crIm -1 cc w In -&* UN 0aZmwQ00 47 a , C,
N
0
10 Wý QPA J% -T Z 't Wot*"
62;
N0 In M -T V% aD -o co m UN w U,
03 %r 30 no 0 r-
coo "t Lo. 'D .4,0 'o co oN 't I"%
0-N - r ý In 0: r :
.. .. ;
CIS u Lo. L. 1ý70L f-11U
.b e-
C» Q
Co
4,
0
4L) t-b
o-
C-1N f-4 ry NNN
1%1
co
4j
N
N
INJNN ry fy NN Fý NNNNNN
Z; In rý %r I- V f. P. mw:; In ') Ir ;ý fr cv, P- On CU In f. Mý
00000-40SIN locoo Ir U% 10 ' Fft 2D 01 W
99999999991
68
5.1 General
at L* 5.1
- PVY and . PUY
9z ay
72
;J1
39
w
.j
N w 9-3
N p Ix
u
l 0
..........
19 0
or) w is
VII le 95 m
14
.......
-**. ,-,.., Z 14
I,, 6 N
--. 0 Z Z
........ -
in
wn) lk
rn M cla N-
Ip
c syc)
M(
P,
IPYN)
lj., Yj)
ll'yl)
0 (ZO,YO)
The inlet is the part ofthe nozzle which links the convergent
section to the test section - It must match the dimensions of the
latter's exit section ( whose radius equals 44 mm ) and was designed
to do so. The inlet is shown in Figure 5.4 along with the convergent
and divergent sections.
N
7i6
N
N
0
u
-r4 m
S:
.q
41 41
u
L4
U.
9
41
Aj
8
41
-r4
Ln
77
Fig. 5.5: Ax
2-
CA
LU
IH
u N
.......... ..........
V) 9
30
II ;ý 1 -
a 94
0 li ad
0
V, 'It
II 14 w
Z Z
-4 lot
ra Ul
14 m
VA
...
*-
0,04-
(D liý
m
w N ci CD w ...
coo
Table 5.1
Table 5.2
Table 5.3
Table 5.4
7ION 6:
SEC, Experimental apparatus and procedure
6.1 General
6.2.2.1 General
The internal
probe ( Figure 6.2 and plate 2) was designed in
order to measure the static pressure along the centreline of the
supersonic section of the elliptical nozzle. It mainly consists of a
hypodermic tube of 300 mmlength and 2 mmdiameter. Four pressure
83
flow meter
ff valve
e bed heater
section
taps, 0.25 mmin diameter and drilled at 90* intervals, are used to
measure the static pressure. 7hey are situated at 12 probe diameters
distance from the tube nose ( Figure 6.2 ).
7he rear of the hypodermic probe is strenghtened by a larger tube,
50 mm in length and 6 mmin diameter, which is soldered to a 100 mm
long bar of losenge cross-section. The bar's upper end is designed to
adapt to the existing traverse plate 1) moving the probe to the
required position. The bar is drilled inside ( Figure 6.2 ) providing
an air passag-ýconnected to a tube of 15 mmlength ( plate 2 ). 7he
latter tube may be connected to the pressure gauge or manometerby a
flexible pipe.
Fiý
nozzle ( Figure 6.3 and plate 3 ). 7he probe holder consists of a main
tube of 70 mmlength, 5 mm outer diameter and 2 mm bore. It is
positioned upstream of the nozzle in a flanged section and is
maintained on the central axis by means of three legs positioned at
120* intervals ( Figure 6.3 and plate 3 ). 7he upstream end of the
probe holder's bore is closed and represent the sliding end of the
internal probe. At that position, the static pressure tappings will be
positioned at the throat. By moving the probe downstream, the tappings
are moved along the nozzle centreline. When they reach the exit
section, the forward end of the probe would still be retained inside
the probe holder ( length of the retained part is 5 mm ).
In order to measure the exit static pressure along the minor and
major axes of the elliptical nozzle) an external probe is provided
( Figure 6.4 and plate 4 ). It consists of a hypodermic tube, 114 mm
in length and 2 mmin diameters provided with a conical nose of 4
half-angle. Me tube is strenghtened in the same way as for the
86
The test
articles are principally represented by the elliptical
nozzle necessary for the study ( Figure 5.4 of section 5 and plate
5 ). Designed from an axisymmetric shape using the method investigated
in this the nozzle was manufactured at RAE Bedford.
thesis,
In order to achieve the necessary finish ( N7 or better ) of the
bore any protuberance in the surface may disturb the supersonic
flow the internal elliptical shape was cast in an epoxy resin
( Araldite MY 750 ) filled with aluminium. The hardener used is HT 972
and the maximum working temperature 130*C.
The elliptical shape of the divergent section of the nozzle was
describedgat each cross-section, by the stations defined at the exit
and traced back to the throat ( section 5 ). The subsonic section,
effecting a smooth transition from the elliptical throat to the
axisymmetric supply pipe, was defined
in the same way ( Appendix J ).
Fourty cross-sections were used, each cross-section being defined by
fifty two stations. Tolerance was of the order of the micron.
The external body, used to support the internal resin shape and
designed to fit the pebble-bed test section to which it is bolted, was
all machined in an aluminium alloy. Tolerance was of the order of
0.1 mm.
The two part-, ( ell"Ptical internal shape and external body ) were
assembled and received the necessary heat treatments.
Static pressure tappings along the nozzle wall were meant to be
provided. The usual method for measuring the pressure along a wall is
to drill a small hole normal to the surface of the wall and connect it
to a manometergpressure gauge of similar other device ( see chapter
2.6 ). Drilling a hole, in this case where the internal shape is
elliptical ( difficulty in drilling satisfactory static pressure
tappings normal to the elliptical internal surface ) and constituted
by a resin, may give rise to imperfections ( such as burrs ) around
the tappings which would probably disurb the flow locally. Because of
this, wall tapping of the elliptical nozzle was avoided and only
static pressure measurement along the centreline and at the exit were
performed.
6.3.1 Calibration
The internal probe runs along the complete length of the nozzle
from the entry section of the convergent to the exit. Its forward end
is blopked off inside the probe holder situated upstream of the nozzle
( chapter 6.2 ) and consequently the measurementsit performs give the
true value of the pressure.
7he external probe, being inserted into a supersonic flow, will
have an attached shock at the nose. Due to the compression caused by
the shockg the static pressure readings will be higher than the true
pre ssure in the absence of the probe. In order to demonstrate this,
the internal and external probes were used simultaneously to measure
the static pressure at the centre of the exit cross-section. 7he
corections needed were of the order of 1.5%.
7he correction was carried out by adopting the following
procedure :
1. Using the isentropic relationships, the Mach number ( M, in figure
6.5 ) just before the shock is calculated,
2.7he cone half-angle of the forward part of the extenal probe
C
in figure 6.5 ) is designed equal to 4%
3. From figure 6.5. for a given value of M, and a cone half-angle 6cs
the ratio of the measured pressure to the real pressure ( PC/Pj in
figure 6.5 ) may be determined.
The ratio of the measuredpressure to the real one is found close
to 1( PC/Pi = 1.1 ). It can be seen, in figure 6.5 that for Mach
numbers comprised between 1 and 2.5 ( which is the range of change of
the Mach number within the divergent section of the elliptical
nozzle where the measurements are taking place ), the curve
representing the cone half-angle of 4* is very close to the Mach
line. Hence, the close-to-1 ratio of the measured to the real static
pressures.
Be - 50 deg
30
45
25
40
20
35
5.0 -,000
-. 00
00
---- ---. 000 1
15
4.0
3.0
PO//, 00or 00or 10
2.0 0
5
Mach line
1.0 I>
2345 6
Mach number before shock wave, M,
- Jet centerline
. F%IL VW
), Jet centreline
)undary
Figure 6.7 traces the two waves back into the supersonic section
of the elliptical nozzle where they are found to interssect the nozzle
wall at a distance z, = z2 - 19.8 mm from the throat. On that basis
and knowing that the attachment point is situated 2.6 mm far from the
throatt it seems unlikely that the internal oblique shock wave is
generated by the discontinuity in the wall curvature.
92
7.1 General
Surfaces are needed to close the contour and define the volume.
Figure 7.3 shows the different surfaces ( denoted SF1 to SF6 ) and
Figure 7.4 the volume they enclose.
The mesh, shown in Figure 7.5, was generated by prescribing the
number of elements required along the external edges (5 along x, 5
along y and 15 along z ). 7he degree of concentration of the elements
may clearly be chosen to match the requirements of particular flow
problems. Since it was expected that the flow property gradients
would be large in the throat region, a concentration of degree 2
( implicating that the distance between the sections along which the
flow is computed, will be closer was generated immediately
downstream of that particular section.
The last step in the procedure is the division of the volume
describing the one quadrant of the elliptical nozzle into elements
( or cells ). Since the edges along the radial and azimutal directions
were divided into 5 stations each, the total number of elements
describing a cross-section is 25. The number of sections in the axial
direction was selected equal to 15 making the total number of elements
375. Figure 7.6 shows the different nodes ( or grid points ) and
figure 7.7 the different elements constituting one quater of the
elliptical nozzle.
Results from thissimulation represent the necessary geometrical
data. Stored in a special file, these data along with the control data
file ( comprising the thermodynamic boundary conditions, i. e.
pressure, density, mass flow ) form the input to 'Phoenics' enabling
the use of its body fitted coordinate version.
7,,
4>\
'n
CV
11+
0+0+-O+.It
L(l)
(D rý M
c4z-
CN 04
0+ 11+
C%4
0+
co
C14
0+
04
C'4 0+
11+
C14
;ý+
N
Q
C14 11
+
ro
0+
ul r)
C14 0+
Q+
11+
K) ce
LO
n r-
co =0
PO (71 0 a
pf) z
qr t:
0+0+
-t3 C)
98
NA
C) CJ
V No
Z aq
;3
-r-T-
'o Q, Q
S:
ts %42 V.
99
>-7N>\
*q w
2: tj
%40
U-
100
NA
-cc)
o -'
4)
U.
101
>-7t,
4>\
lw
lw
cm
lu C)
ý
-2:
2 ci
102
A(r
1 0.0
S:
AL*711
%i %=
o 460
3irv
. ts
*^ *1
6 ** g. Sx
->*, Iti qw
S:
W S:
44 w
ts Q)
103
104
.
4
CW
%.= W
CW
0/ 07 Z\
07 0// ZZZ
10
87 87
07 07 07
t7
0 lw
Z ýý jý
t7 >/ oý/
00
tz
p7 07 o/ 0/
p7 o7 jýi w
OD
0/
iý/
u
Ln
L
0
)
.0..
u
105
QL
C) 42 . , -nl
.. "%)
.C)
"%) E lw
ei
, IL Uli
vý gw
2:
EwZZZZ
w/1 vi
cj
tý wW
P%
7
P/ p7 /ZZZ0 Co
LL
p7
p7 W/
077.7 1 @0
vi
L
0
Ppt. )
106
-w ý
2: Obi
ei N
cm w
die 0
ei =
l'Ne ei
«ti CM
%= tg
Q st
AA
vvv
vv
cn /V Uxy
;sZ:
Iti 0
Z -9 Y
g=N
/
107
ß4
ei Ici
,Ze
*-90 *hý
dc>
19
* lle
:ý
lIxi cn
C) de
0
w
cr)
(D
--9
ci
u
0
.tiu
7\,
4x
108
lu
uE
43 `Q
«ti r.
ei E (ZN)
*e E. E
%42r-. ly
-xi - 'u /1/ \
/1
/ 1/
CD
/1/
--e cn
ci m de
1/ //
1/11/ w
cr)
(D
/1/ /
// /1
/ vi
/// L
0
/1 // 0)
N /1/ /
NTN,
109
=.r u
110
y y
xx
Fig. 7.14: G-tid uzed to dezc4ibe Fig. 7.15: Pupozed g4id dezigned
the ettipticat nozzte. to etiminate the tou o6 o4thogonatity.
3. In most cases, the boundary layer which develops along the wall
of supersonic nozzles is very thin. Fo the present simulation to
inside boundary 'r layer, be
achieve computations the the grid must
greatly refined allowing nodes ( at which the computation are carried
out ) to be situated inside the boundary layer.
In the case of the axisymmetric nozzle from which the elliptical
nozzle was designed using the method investigated in this thesis, the
boundary layer thickness was found to be of the order of 1/10 mm. For
the three-dimensional simulation to achieve computation inside the
boundary layer, a grid comprising an excess of one million cells must
then be used. With a view towards the high computer time and large
storage required to achieve a similar design, such a simulation
could not be realistically considered.
. 1ý
0
ill
8.1 General
expansion wave from the wall reaches the axis is such a point and is
labelled by 'A' in Figure 8.1.7he same change in slope is shown by
the method of characteristics computations which are performed along
characteristics and thus do not require interpolation, giving the true
character of the solution. The three-dimensional calculations smooth
out this point. Ihis diffusive characteristic is inherent to any
three-dimensional calculation scheme because of the necessity to
interpolate and because of the need to satisfy the stability criterion
which requires that numerical disturbances be propagated at a velocity
greater than the infinitesimal propagation speed i. e. the speed of
sound.
In Figure 8.1. the two-dimensional solution of the flowfield
y
, ký
L, \\>,
. .,
x
Fig. 8.2: Th4oat c4ou-zection.
computation ( section 3 ).
The finite-difference calculations also predict that the expansion
along the walls ( Figure 8.3 ) proceeds smoothly along both the minor
and major axes of the elliptical nozzle. Within the axisymmetric
nozzle, however, the method of characteristics calculations show that
the pressure drops sharply in the immediate vicinity of the throat
tending to stabilize further downstream. This discrepancy between the
two solutions may be attributed to the way the elliptical
configuration has been generated by 'I-Deas' ( gridding ) and its
implication on the three-dimensional calculations carried out by
Thoenics'. The results of the axisymmetric computations carried out
by the method of characteristics, which show the true character of the
solutioný are characteristic of an axisymmetric contoured supersonic
nozzle with a relatively large contour angle which allows a large part
of the expansion to occur near the throat, the remaining contoured
walls straightening the flow. The advantageous result is a
substantially shorter nozzle.
In the time available to the study we were unable to add pressure
tappings to the elliptical nozzle; the highly curved surfaces and
small scale making this an awkward task ( section 6.2.4 ). However,
the above compare
results well with those of ( Ransom et al, 1972 )s
whose value reaches 35* in Ransom's case and 159 in our case.
RCI 1ý.ý13*
0.5
35'
OA -A
.--
pFtESSURE
ftATIO 03,
WALL
(P/PT)
02
A CENTERLINE
& 2-D
0.1 9 3-D Fig. 8.4: CompatiAon oj the
watt and centAetine p4e,6zu4e
0 1.0 2D 30 4.0 5.0 6.0 7D
4atio. 6 a-6 a junction o6 axiat
(L/RT)
NOZZLE LENGTH po.6ition in a contomed nozzte.
(Ran.6om et at, 1972)
116
Figure8.3 also suggests that the flow expands more rapidly along
the minor axis than the major axis. The figure also reflects the
azimutal pressure gradient round the contour ( since the data shown
represent the predicted pressures at the same cross-sections ) which
is typical of three-dimensional configurations causing cross flows and
adding to the build up of the boundary layer. However, experimental
results ( illustrated by Figures 8.1 for the pressure distribution
along the centreline, Figure 8.8 for the pressure distribution at the
exit cross-section and Plate 6 for the schlieren picture of the flow
exiting from the elliptical nozzle ) show that the flow behaved in a
satisfactory manner with no evidence of separation. The results
illustrated by Figure 8.3 fall close to the predictions of Ransom,
1972 ) performed along the so-called super-elliptic nozzle Figure
8.5 ) and illustrated in Figure 8.6.
35ý
30
25
4
r, 20
2-D; AXISTMMETRIC.
35
30 30
25 Is
?
20 *I_1. 20
_. _ 0t
is is
10
10
0 f
-5 A
-10 0 ,"
-20 . 20 '%
1
-25 -25 .
-30 -30
. 35 F - ýo ts io . 35
50 13 20 25 30 35 40 qs 0a4 6s to 02 1`1 Is to 20 22 2`1 26 !$ 30
Z to 2
X10-2
Z. fool" PýPtotdf Jr. 3-0.6 P00040964
35
30
/
25 / is
I
10 20
20 10 /
. .
is -`0 13
10
10
-5
. 20
ItI
.
. 30,
-30 1
. 35
-35 12 14 16 to 20 02 461 so 12 11 16
9 10
42.14- 52.
considering the major axis, the pressure gradient along the same
cross-section equals 10% of the total pressure while for the minor
axis, it decreases to 2% of the same pressure. However, these
differences diminish in an appreciable way when we travel further
downstream into the supersonic section ( z= 20.69mm, 42.14mm & 52.05mm
in Figure 8.8 ) until the exit ( illustrated by Figure 8.8 ) where the
two curves are very similar. Differences may be observed between the
three-dimensional solution performed by 'Phoenics' for the elliptical
nozzle and the method of characteristics solution carried out for its
ratio T/M, the axisymmetric nozzle exhibits a 20% advantage over its
st ?
6; %
3-C
IL
!2
Xc
j
Ic
0-
z
W
au
mi
u
x
-r4
w Aj
A R C RT 21,119.1 2
0-4
z
s
Z
0
In
ow
ul
4h.
Vý) A
120
3-0 ; WEDGE,
2-0 ; AXISYMMETRIC.
........
Xfol
35
Xfol
30 30
25 25
20 20
10 10
-10 -10
-15 -13
-20 -20
. 25 -25
-30
0 367
67
X10-1 x1r,
p"ptotel
Xfol Xfol
35
30 30
25 25
20 20
is
10 to
5
-10 -10
-13 -13
-20 -20
. 25 -25
-30 . 30
TABLE 8.1
TABLE 8.2
SEMON 9: Conclusion
LIST OF REFERENCES:
BERMANyK.
'The plug nozzle. A new approach to engine design'
Astronautics, Vol. 5, Nb 4,1960
BOUSSINESQ,J-
'Theorie de Vecoulement tourbillonant'
mem. Pre. par. div. Sav. 23, Paris (France)
BUSEMAN9A.
'handbush der experimental- physik'
Wien-Harnus, Vol. 4, pp. 407-422,1931
( translated as ARC 8224, F. M. 757 )
COOKSON,R. A.
'Supersonic combustion studies- Part I: Construction and
peformance of a high-enthalpy facility'
CoA report 200, Cranfield Inst. of Tech., 1967
ECKERT, E. R. G
'Engineering relations for friction and heat transfer to
surfaces in high velocity flow'
J. Aeronaut. Sci., Vol. 22, Nb 8, pp. 585-5879 1955
FLAHERTY9 R. J.
'A method for estimating turbulent boundary layer and heat
transfer in an arbitrary pressure gradient'
Rep. UAR-G51, United Aircraft Corp., 1968
GAL-OR, B.
'New fighter engines, a review', Part I
International journal of Turbo and jet Engines, Vol. I, Nb 3,
pp. 183-194,1984
HAMA, F. R.
'Boundary layer characteristics for smooth and rough surfaces'
Trans. Soc. Nav. Architects and Mar. Eng., Vol 62,
pp. 333-358,1954
HALL, I-M-
'Transonic flow in two-dimensional and axially-symmetric nozzles'
Quat. Journal of Mech. & Applied Math., Vol. 15, Part 4,
pp. 487-508,1962
HOFFMAN, J. D.
'Accuracies studies of the numerical method of characteristics
for axisymmetric, steady supersonic flows'
Journal of computational physics, Vol. II, Nb 2, pp. 210-239,1973
I-Deas Supetarb
Engineering analysis, Pre and post-processing
Published by Structural Dynamics Research Corporation, 1986
JOHN9 J. E. A.
'Gas Dynamics#, Second Edition
Allyn & Bacon Inc., 1984
KHALIL, R. A.
'Aerodynamic studies of rocket motors and nozzles'
MSc. thesisp School of Mech. Eng., Cranfield Inst. of Tech., 1987
KUHN, G. D.
'Calculation of compressible, nonadiabatic boundary layers in
laminar, transitional, and turbulent flow by the method of integral
relations' ; NASA CR-1797 1970
NASACR-1797,1970
MALINA, F. J.
'Characteristics of the rocket motor based on the theory
of perfect gases'
J. of Franklin Institutel Vol. 230, pp. 433-454,1940
MASSAUs J.
'Memoire sur Vintegration graphique des equations aux derivees
partielles'
F. Meyer-Vaulon, 1899
Mc DONNELLAIRCRAFT Co.
'Experimental evaluation of non axisymmetric exhaust nozzles'
Final report, AFFDL-TR-78-185, Vol. I& 11,1978
Mc KENNY, J. D.
'An investigation of flow separation in an overexpanded
supersonic nozzle'
mSc. Thesis, California Inst. of Tech., Pasadena, California, 1949
MELLOR, G. L.
'Turbulent boundary layers with arbitrary pressure gradients and
divergent or convergent cross-flows'
AIAA journal, Vol. 5s pp. 1570-1579,1967
MITCHELL9 A. R.
'Solution of the Von-Mises boundary layer equations using
a high speed computer'
Math. of Computations, Vol. 15, Nb 759 pp. 238-242, July 1961
PALMER, J. R.
'Course Notes'
School of Mech. Eng, Cranfield Inst. of Tech., 1985
133
PASKONOV, V. M.
'A standard program for the solution of boundary layers,
Numerical methods in gas dynamics, Moscow, 1963
PICKNEYO S. Z.
'Method of predicting compressible turbulent boundary layer in
adverse pressure gradients'
NASA TM X-2302,1971
RAO, G. V-R.
'Exhaust nozzle contour for optimum thrust'
Jet Propulsion, Vol. 28, Nb 6, pp. 377-382,1958
RAOv G. V. R.
'The E-D nozzle'
Aeronauticss Vol. 5, N 9,1960
RAO, G. V. R.
'Recent developments in rocket nozzle configurations'
ARS journal, Vol. 31, pp. 1488-1494,1961
ROSENHEAD9L.
'Laminar boundary layers'
Clarendon Press, Oxford, 1963
SAUER, R.
'General characteristics of the flow through nozzles at near
critical speeds'
NACA TM-1147
134
STEWARTSON9K.
'Correlated incompressible and compressible boundary layers'
Proc. Royal Society, Ser. A, Vol. 200, Nb A-1060, pp. 84-100,1949
TOWNEND9L. H.
'Private communication', 1985
VAN DRIEST9 E. R.
'Turbulent boundary layer in compressible fluids'
J. Aeronaut. Sci., Vol. 18, N 3, pp. 145-160,1951
VAN DRIEST, E. R.
'The problem of aerodynamic heating'
Aeronaut. Eng. Review, Vol. 15, Nb 10, pp. 26-41,1956
WINFRESS, R. N.
'Internal ballistics of solid fuel rockets'
Mc Graw Hill Book Co. Inc., New York, 1950
A. 1 Governing equations
2-a2 au LE au_
(u ) ,( v2-a2 ) +2 uv ,a2v. 0 ( A. 1 )
ay Dy oeu
y
au Dv ( A. 2 )
*es
ay -.57
A. 2 Characteristic equation
"i )+ CY2. aU
2_ 2_+ au ( 2uv ]+( 2v
a(ua)(
1
_a )(
ax 2. ax
cY, ( u2-a2 ) ay
du LU ýu
li ,U +X ay
( A. 5 )
so*
dy Dy ýv
.
uy- rx +X ay
2uv + C72
22
al u -a
oee ( A. 6 )
( v2-a2
-cr 2
2A-2 I+
a1[(u2 -a uv 02 ( -1 )-
( A. 7 )
ooo
al ( v2-a2 )+ CY
2(X)-0
A-3
For this system to have any solution other than the trivial one of
zero, the determinant of the coefficient matrix must vanisho leading
to :
dy uv :ta2( M2_1 )k ( A. 8
).
ax (+ ...
2 u2-a2
V
Streamline-_,,.
1
V
%rl -JL j
Fig. A. I Fig. A. 2
ct
u-V cos
» sij, (, )
( A. 9 ) ( A. 10 )
v-V sin 0 so*
sin a
tg-, cotg
A. 9 and A. 10 into
Substituting A. 8s rearranging and applying the
standard trigonometric identities yields :
tg +a ( A. 11 )
dx (+
A-4
x
Fig. A. 3 : Chatacte4i. 6tic. 6 (Mach tinu) in a Zteady, 2-D
i4-totationat itow.
A. 3 Compatibility equation
2) du(-) )X 1 dv(
u2-a +[2 uv +( u2-a2 (+q_)
) ( A. 12 )
ooo
A. 4 Finite-difference equations
A-5
( AX(+S-)
2 I AV(+,
[2 u(+, v(+, -( Jý (+, ) X(+S-) -
-) -) _)-a (+, -) -)
6 a2(+. V(+. Ax
-) -) 0 ( A. 14 )
(+ ) '. oee
), (+s_) - tg ( eo*(A. 15 )
y,
Right-running
characteristic C_
Solution
Yj
point
Y4
Y2
Left-running
characteristic C+
1x2 X4
Y4 -1 (-) x4 ý yl -ý (-) X1 ( A. 17 )
(u ) X(+) I v4 = ( A. 18 )
(+) -a (+) e*o
2(_) )
( tý(-) -a u4 +[2 u(_) v(_) -
where :
a2(+) V(+) 2
( X4 - X2 +( uý(+) -a u2
Y(+)
[2 u(+) v(+) -( U2(+) - a2(+) ) X(+) ) v2 ( A. 20 )
...
a (-) v (-)
( x4 - xl )+(u2 )u1
(_) - a2 (-)
u U2 V(+) V2 y Y2
eoo(A. 22 )
u Ul V(-) Vi y Yi
U2 + 134 V2 + V4 Y2 + Y4
U(+) v2 y2
2
oee ( A. 23 )
Ul + U4_ V1 + V4 Yl + Y4
U(-) V(-) y2
2 2
*ý-Right-running
characteristic C-
Solution
point
4 F-
):
MIRROR e ,,,
IMAGE ":
-'-Left-running
2 ooýý'
characteristic C+
y
1 Aýý
Solid
boundary --
ion
/ point
I
Left-running
characteristic C+
known interior
2
point
y- y( x ( A. 24 )
LY ( A. 25 )
- tg 0
dx
y
Solid -,,, 4
boundary
%INPrespecified
solution point
i'%k1ccharacteristic
I ýl t-running
2 AkLef
001 C+
Right-running 1
characteristic C_
x
Fig. A. 7 : Inve4ze mtt point.
may result in too large spacing of the solution points along the wall.
In that case, the inverse method may be employed. It involves
prespecifying the position of the solution point and using the method
of characteristics to determine the flow properties at these points.
Figure A. 7 illustrates schematically the situation where point 4 is
the prespecified points point 2 being the intersection of the
rearward left-running characteristic C(+) through
the iolution point 4
and the right-running characteristic C Because the solution point
(-)*
is prespecified, x4 and y4 are known and only the flow properties at
point 4 have to be determined, involving the calculation of the
location and properties at point 2. The location is established by
determining the intersection of the rearward left-running
characteristic 24 ( Equation A-16 ) with the
right-running
characteristic 13 ( Equation A. 17 ). Flow properties at this point are
determined by linear interpolation between points 1 and 3.7he
solution of equations A. 16 and A. 25 will determine the properties at
point 4.7he procedure, for an inverse wall point, is incorporated in
subroutine 'Invpoint' ( section 3 ).
y y
1
-Right-running
-Right-running characteristic
characteristic Cý) q-)
(k-)- -) S.
4 Left-running
,
characteristic
Left-runninp (X W)
0)
cc
characteristiý
2
X4=xl x x4'cx2
Fig --A. 8 ' Caze whe4e X, a 00 J04 Fig. A. 9 : Caze whe4e X, jo)t
%int. ira.
an tnte4io4 inte4iot +)
an point.
properties at an interior point. 7he f irst one occurs when the slope
of the right-running characteristic C(-) is infinity ( Figure A. 8 ).
In that cases x4 is taken equal to x, and y4 obtained from equation
A. 13 when applied to the right-running characteristic C(+)e
The second case occurs when the slope of the left-running
characteristic C(+) is infinity ( Figure A. 9 ). In that case, x4 is
taken equal to x2 and y4 obtained from equation A-13 when applied to
the right-running characteristic CH"
The two possibilities have been taken into account and
the solutions included into the computer program ( subroutine
'Intpoint' ).
Solid
- Left-running
characteristic
(,Xor -)
As in the internal
point procedure, a technique has been
incorporated within the computations of a direct wall point to handle
the case of X(+) - -( Figure A-10 ). In that case, x4 is taken equal
to x2 and y4 determined from the equation of the wall contour.
B-1
NO) [ NO) I
y (xO+h) -Y +h y' xo, Y
(n) [
y xo, y(xo) B. 1
( 1 ( B. 2 )
y(x 0+h) - y(xo) +h y' xo, y(xo) 9*9
dy -, f (x, y ) dx oto ( BA )
Ax ( B. 6 )
+ eee
2
and replacing f(
xi, yi ) in equation B. 5 by the value of f( x, y
determined at the mid-point of the interval. 7hus :
&X Yj + yoi+i
Y, +f(x, + ) &X ( B. 7 )
22I oet
ni+l AX ) Ax ( B. 8
Y: = yi +f( xi '21 yi
+2 ...
yn
where i+1 is the value of y after n applications of the corrector.
C-1
CA General
Dive,%gent .6ection o6 a
contou4ed nozzte.
expressed as :
2 ( C. 1 )
/otxy x2 +m Rtd ob*
2 ( C. 2 )
/Oxy x2 +(y- Rtd - yt )2-R ...
td
C. 3 Attachment point
Ya ýyt+R 1- cos ( Aa C. 4
td
y=AW+Bwx+ C X2 *so ( C. 5 )
w
2ý
x-xa )' Ya AW +Bw xa + Cý xa Ya
y(
dylx ) B+2Cx- tg (A
m tg ( Aa C-6
wwaa
.x a
C-3
dyl
- tg ( A42 ) Eý +2% xe - tg (Ae)
dx
x-xe
where the unknowns are the coefficients of the wall polynomial Aws Bw
and Cw.
The second method involves the use of the supersonic tables.
Having specified the expected exit Mach number, the tables are used to
determine the corresponding area ratio permitting a complete
expansion. In this case, the equations involved are :
2
y( x=xa )= Ya AW+ Bw Xa + ýw xa . Ya
*oo ( C.7 )
9y I- (Ae)
. tg (Ae Bw+2Cwxe 'r tg
dx
"xe
where the unknowns are the coefficients of the polynomial wall Aws ks
and Cw and the nozzle length xe.
D-1
D. 1 Governing equations
u(X, o )- a9e(D.
1 )
11 .6* ooo ( D. 6 )
Pe ue(x) -P u(x, y)
so* ( D. 7 )
Pe u
e(x)
D-2
-ht (X, Y)
e D. 8
9 I=ht, hr
ht, -
e
D. 9
p (x, y)
ve
T =. D. 10
-ue
K
e* ( D. 11 )
g oet
ue 6
PWVW af I
(1+ c n)-i [d (1-f I) IIII+( (Q+R) (n-f - 6* FX df
ad Peue
PWVW af I ad
(Q+R) (n-f) - -6* d' -(Pd+6* ) (2-f')
Peue ax ax
( D. 12 )
es*
Y-1 M2
2e
(J+C -1 d2(, _f, )2
a9 'il 12me) Ke
H( 1+
af Pwvw ago
(Q+R) (,n-f I got 1) 6* D. 13
Tx - ý Te ax
D-3
1 aue
D. 14
where :
Ile ax
Pe ue )
( Dol5 )
ax ose
Peue
a(rw)
R'* D. 16
rw ax
Ca 2( ýý ) cos a D. 17
rw
6*9(X R6*
T =. IK+XXD. 18
__K R6*
6*
K
v1-4. [T-vIID. 19
-L
ve R6 ve R6
Krp rt
The above set of equations along with the boundary conditions are
to calculate the development of the boundary layer.
sufficient
However, small inaccuracies in the numerical solution which are
one step in x, are frequently cumulative. Creater
negligible after
be achieved by correcting the integral parameters
accuracy can
obtained from the numerical solution of equation D. 11 by comparing
them to those obtained from the Von-Karman integral momentumequation
be obtained by integrating equation 4.15 of section 4
which can
the layer. The result is :
accross
due
ue2 + rw 6* Pe ue U- m rw ( *rw + ow vw ue
dx we
o99(D. 20 )
D-4
puu
where :0-r(1--) dy D. 21
0 Peue ue
D. 2 Numerical method
Ve_ K2 v2 au )
.( ( D. 22 )
...
6*6*
K ue K ue ý7
D-5
ve au
(K2y2 ) oee ( D. 23 )
v v ay
-k 22 au
ve ý výp- ý', ue Ky-D. 24
ay
Equivalent to
above the formulation but offering some
computational advantage is the functional form suggested by ( Mellor,
1967 ) which may be expressed as :
Ve
X) with X-KyD. 25
6K Ue ue
ve y(
. X with -r/p )k D. 26
vv
1v ve -v 28 )
*oe(D.
Prt
er
D-7
V,
6.9)3
0.41
y( YM )ý
4ý
y( Tla
0.41
Ue 6K*
a-. (1E. 1
u6
e
Tt - Tw 1/4
.(Y) ( E. 2
...
Tt, Tw A
e-
Pe ue -pum-p(y) 1/7
1- -g ( E. 3 )
ooo
Ue Pe
-Pe
Ue
2
Te Tt, 2 Cp
e E.4
.
Pe T Tt
_ U2
2 Cp
E-2
2
u2 ue!ý.
(Y )2/7 ooo ( E. 5 )
ý=
2 Cp 2 Cp 6
Pe Ue -Pu.
Pe Ue 2
ue
Tt, e 2 Cp
, «: (Z )1/7 ( E. 7 )
2 so*
1/4 ue 2/7
(Z) ( Ttge - Tw + Tw _ (y)
A2 Cp 6
h - ht ht Tt
t, e 1--. 1- ( E. 8 )
-. *eo
ht, ht, Tt,
e e e
TW TW
Tt, e 1/4
(Y ) eue ( E. 9 )
T
1- w
Tt,
e
Simplification leads to
E-3
Tt Tw Tw
+ ... ( C. 10
Ttse Tt, Tt,
e e
Substituting into equation E. 8 leads to the expression of the
initial enthalpy profile used to start the flow computations :
ht, ht 1/4 Tw Tw
e- + Cal
ht 'e Tt, Tt,
e e
Tt ut
... ( E. 12
AP Tt T
ý RT
E. 13
Au
Tt )1/2 (YE. 14
Tt
T
Tt
'Y R
1+ M2
p. (E. 15
Amy
P-P( AsM )
Therefore :
dP BP dA BP dM
eso ( E. 16 )
dx aA dx 3M dx
From equation E. 13 :
R -Tt
ap 1+2m2
ooo ( E. 17 )
aA A2ym
(Y-1) R It M 'y R Tt
ffi m[Y
21+ Y-1 M2
ap
M2 )21+y21
2
M2
*to ( E. 18 )
dM dM dA
*to ( E. 19 )
dx dA dx
m
dA
dM
20 )
**o(E.
dA Ad (A/
dA*d (M/Aý-
Thus :
dA
dM dx ( E. 21 )
dx d (Ale)
d(Mle
Y+l
. Y-l
ooo ( E. 22 )
Tt
AP
YR
Y-1 (mm e2 2-Jy-r
e2
2 Aý7)2 +Y21
( E. 23 )
m
dx 1+ -1 M2 1+ 'Y, M2
dx
22
-1Y-1 2-Y
m ly-l
;
-Y-1 2-) MYR
*Y-1 M2)2 Y-1
+ 222 1+ M2
Tt E. 24
(Y R1+ Y-1
2 M2
dA
+
dx A2 Ym dx
(Y-1) R Tt MYR Tt
M,
2 1+ 22 M2 )2 Y-1
1+ M2 dM
Ay M2 dx
*so ( E. 25 )
7he mass flow can be calculated from measured data in two ways.
Method I uses an orifice flow meter and is based on the pressure drop
across the orifice. Method II uses the stagnation pressure and
temperature along with isentropic flow theory to expand the gas within
the nozzle.
2d2K[2P (AP
4
p
RT
AP . pHg gAH
where :
d- orifice diameter
P- pressure prior to orifice
z- compressibility factor ( 0.99 for P- 10-25 atm
where Pth
"th -RT
th
Wth - Mth ath = ath -( 'Y R Tth
Substituting leads to :
I
Pth
(YR Tth )k Ath
At the sonic Mach number and for a specific heat ratio of 1.4, the
isentropic flow tables ( Keenan& Kaye, 1948 ) give :
Pth w 0.52828 Pt
Tth - 0.83333 Tt
Final 1Y :
0.52828 Pt
il ---(YR0.83333 Tt ) ( F. 3 )
th so*
R 0.83333 Tt
G-1
APPEJOIXG: I-Deas
> Model preparation and checking: This task enables the generation
of the necessary nodes, elements and physical and material tables. It
also checks the validity of the model before analysis.
APPENDIXH: Phoenics
H. 1 General
H. 2 Geometry definition
very much curved ( especially the elliptical nozzle ) and thus the
grid option ( often called boundary fitted coordinates )
curvilinear
had to be used. 7he grid and the nodes at which the equations
by the I-Deas supertab
are solved was produced software module
( Appendix G ).
L( ( ri
ri Pi oi + div Pi v ei - r, ri grad cpi ri Si
axi ...
( H. 1 )
P where: P- density
P- pressure
R- gas constant
T- temperature
Y-1
where Ptotal and Ttotal are the total pressure and temperature whose
values are prescribed as initial conditions.
I-1
I. 1 7hrust
i0i )
Tc-ý Ve +( Pe - Pa ) Ae a00(
tc
syc)
M(
FNIYN)
Ij 9Yj)
Ii syl)
0
(ZO,YO)
Point i (zi, yi) lies on the circle of centre C and radius Ri, thus:
where:
zoc m0 is the axial distance between the centre of the circle
and the origin of the coordinate system,
ziM0
ZM,YM are prescribed by the inlet section which must match
the exit of the supplying pipe.
2+ 2- 2
Ybi Im Yi
JOC ( J. 3 )
2( Ym - Yi)
Ri -I Yj - Yoc I ooo ( J. 4 )
where:
z 0
oc
yoc and Ri are defined by equations J. 3 and J. 4.
2+
YM ZM2 y12 ]2
z2 +[y __ .( yi _ Yoc )2; i- 19N
2( yM - Yi )
006 ( j65 )
Ye
xB3 ý Ye (0.30 and
-'BJ (2)
oeo ( J. 7 )
Ye
xc -0 and YC -
(2)k
.j
I
0
Li
C,4 c)
I
OVA
0
0)
.L.j
0
., A
4..1 (
U
0) r-
0
0m
up
II
-L- v iange.
2- Probe holder.
3- 120' Positioned legs.
K 7
V)
V)
0
0N
N
-A
u
-4
(2) Aj C)
CL
_C:
4-) -4 4-)
-4 U
(U