i

FTD-HT-23-1483-67

(M FOREIGN TECHNOLOGY DIVISION

CO
00
CO

«^5

MODEL SCALING RULES FOR TURBOJET AND GAS

TURBINE COMBUSTION CHAMBERS

by
Tan-shih Ch'en

Hi
0 KAR X11969 !'.

Distribution of this document is

unlimited. It may be released to
tht CSearinghot/s«, Dcp-irtment of
Comrne.ce, f« Mt« lo trie gutoul
R«produc«d by »he
public.
CLEARINGHOUSE
for federal Scientific 8 Technical
Iniormotior Springfield Va 22151

Si
-
 ..
,,,

THIS DOCUMENT IS BEST

QUALITY AVAILABLE. THE COPY
FURNISHED TO DTIC CONTAINED
A SIGNIFICANT NUMBER OF
PAGES WHICH DO NOT
REPRODUCE LEGIBLYo
 FTD-HT- 23-1483-67

EDITED TRANSLATION

MODEL SCALING RULES FOR TURBOJET AND GAS TURBINE COMBUSTION

CHAMBERS

By: Tan-shih Ch'en

CH/OOI7-63-011-004
i
THIS TRANSLATION IS A RENDITION Of TNI ORICI-
NAL POtllON TEXT WITHOUT ANT ANALYTIC AL OR
EDITORIAL COMMENT. STATEMENTS OR TNtORIIS PREPARED ETi
ADVOCATEDOR IMPLIED ARE THOSE OE TNI SOURCE
AND DO NOT NECESSARILY REFLECT THE POSITION TRANSLATION DIVISION
OK OPINION OP THE POREION TBCNNOLOOT Of- E0REI6H TBCNNOLOOT DIVISION
VWON, WP.APR. OHIO.

FTD~ HT - 23-14&3-67 Dlte 2 Aug 1968

 DATA HANDLING PAGE
OV ACCESSION NO. 9&-DOCUMENT LOC "OPIC TAGS

TP8001429 combustion chamber, gas turbine,

jet flow, turbojet engine
MODUL SCALING RULES
O»-T.TLE
FOR TURBOJET AND GAS
TURBINE COMBUSTION
CHAMBERS
47-SURJECT AREA
21

«.AUTHO^flMUnOB 10-CATE OF INFO

CH'EN, T.
43. SOURCE »■DOCUMENT NO.
CHI HSIEH KUNG CH»ENG HSUEH PAO
(CHINESE) FTQ-HT-23-1483-67
»•PROJECT NO.

723>01-70
6»-SECURITV AND DOWNGRADING INFORMATION 64>CONTROL MARKINGS 9>HEADER CLASN

UNCL, 0 NONE UNCL

7»-REEL/FRAME NO. 77-SUPERSEOEG TtVCHANGES 40-GEOGRAPHICAL NO. OF PACES

AREA

1886 0495 CH 35
CONTRACT NO. •CC. ». PUBLISHING DATE TYPE PRODUCT REVISION FREQ

65-BX7OH387 94-00 Translation NONE

iwnrr ACCESSION NO.
02-^/0017/63/011/004/00^1/0063
ABSTRACT
~—>Many rules for scaling turbojet and gas turbine combustion
chambers have been developed in the past ten years, but none has been
accepted as a general rule. This paper describes various methods..in .
order to find possible scaling rules for such a condxrtton system.
A scaling theory with respect to nonhomogeneous combustion of droplets
under jet flow is suggested primarily to check various existing model
scaling methods. On the basis of the suggested theory, the author
finally proposes some new scaling rules and also offers conclusions
for reference use in future studies. (y

AFSC XgfL * dor FO»V no oemm) AF«C (»A*»,

■■""—■' ■■ HUI i.iiiiaittfntfiifi[t'iijg

 MODEL SCALING RULES FOR TURBOJET AND
GAS TURBINE COMBUSTION CHAMBERS

Tan-shih Ch'en

ABSTRACT

Maiqr rules for scaling turbojet and gas turbine combustion chambers

have been developed in the past ten years» but none has been accepted

as a general rule. This paper describes various methods in order to

find possible scaling rules for such a combustion system» A scaling

theory with respect to nonhomogeneous combustion of droplets under jet

flow is suggested primarily to check various existing model scaling

methods. On the basis of the suggested theory, the author finally

proposes some new scaling rules and also offers conclusions for reference

use in future studies.

I. Introduction

In order to keep abreast of rapid developments in jet engines and

gas turbines in civilian use, design and testing of combustion chambers

have obviously become very important problems. Since combustion is a

complex process, most combustion chambers designed ir. the past were

based on experience alone. Construction of a high-performance prototype

generally requires a series of e-xperiments and adjustments and wastes

lnrg3 amounts of manpower and materials. This it? considered uneconomical

as far as 3aving3 in production and promotion of design technology are

FT0-HT-23-1UB3-67 - 1 -

"^MaM,-"mam STOlSgJB
concerned. Thus, the search for an applicable rule for combustion

chamber scaling has become a practical problem.

Progress has been made in studies on combustion chamber scaling

during the past ten years, with scaling theories and rul«3 continually

published•>■*"' ~* *J. The existing rules can be classified fundaroentally

into two categories. In the first category, actual combustion chambers

are treated as a homogeneous gas reaction system; as for example the

model scaling rules developed on the basis of turbulent combustion

velocity theories *<** *°J and those derived from G. Damkohler's five

similarity criterial^iI41WJ. Combustion characteristics of liquid

fuels are disregarded under this category, and only empirical correction

indices such aa experimental Indicesl5Jlcj related to the assumed order

of reactior and mixing factors are introduced in the final analytical

;
I
results. The second category is based principally on combustion of
i .

droplets in jet flow, with careful attention to combustion characteristics i \

of liquid fuel and to scaling requirements for gas reaction, as for

example the scaling rulca described by Stewart »>> and by Herbert*- '■*,

thosel^i falling between Stewart's and Herbert's, and those established

purely by supposition through experimentation^. To »urn up, there are

numerous foraa; of existing scaling rule3, which differ greatly from one

another. It has not yet been possible tc reach a unanimous viewpoint and

understand their inherent connections. Thus, how to classify the

existing rules, whether or not there aro other possible schemes, and how

to investigate these «ehernes are worth further investigation.

FTD-KT-23-1U&3-*? - 2 -

- - —:
 t

The key approach used In this paper to investigate model scaling

laws controlled by droplet combustion under jet flow is through the

similarity and combustion theories, because the processes In the

combustion chambers of moat jet engines and fuel-burning gas turbines,

with the exception of those in a few vaporization-type and gas-burning

combustion chambers, are controlled by thermal diffusion in droplet

jet flow. Their working characteristics are closely related and very

sensitive to atomizaticn, mixing, vaporization, and combustion of liquid

fuel. Much of the literature, including17]1SJ, reports that various

working conditions of an actual nozale can be simulated by using a

series of small nozzle models, and this is an obvious example,

II. Theoretical Data for Model Scaling

Combustion of liquid fuel is characterized by the following

component processes: atomisation, mixing, vaporisation, and eombu3tion„

In order to make the actual combustion prcee?s and that is a model

similar, in addition to the necessary geometric similarity in combustion

chambers and their major components (including sprayers and flame

stabilizers), similarities must be obtained to satisfy as far as possible

th&se component processes and also the nsonodromy conditions of combustion

systems. As for the Internal heat-transfer and combustion gas cooling

problems, generally speaking they are relatively easy to solve, and

thus further discussion is unnecessary.

FTD-HT-23-lu8>6? - 3 -
 1. Atomization

To sake the atomization processes in two similar combustion

chambers identical, the same lav governing the size and distribution

of droplets in both combustion spaces must be applied. Fineness of

atomization in jet flow can be designated by the mean droplet diameter

ds. According to dimensional analysisU-0JL11J, th» "imensionless mean

droplet diameter can be expressed by the following dimensionless func-

tional relation«

J-fCtaj,^, 4,51,1, Akt

d J 3 f v d d d
>. CD
j g 4g g j j j

where dj is the fuel nozzle orifice diameter; 'x> ^2» ^3*eto» are

*lv'I'
geometric nozsle dimensions; Rej = 'f ~ _gv2
l' I and We* = id1
:i: f.~A,. are

Reynolds number and Weber number, respectively; \,JJ, v, and <J~

represent density, viscosity, velocity, and surface tension coefficients,

respectively, with subscripts / and g to denote liquid and gas; and 9

is the injection angle of Jet flow. The injection angle is related

primarily to the structural parameters of the nozzleU^J# Although Q

is influenced slightly by the combustion pressure and fuel injection

pressure, these effects can be disregarded in approximation. Equation

(1) for two geometrically similar nozzles can be written as

The applicable function form of the above equation has not yet been fully

developed. If the same fuel and the same inlet temperature are taken,

then the quantitative relationship comparatively suitable for a centrifugal

- U -
nozzle is

jä= (a+ Mtojjltej-Jtp.)?, (3)

where a and b are coefficients. According to reference

fl2lJ
L
,a = 23.5 and

b = 0.0004* This shows that the Reynolds number for jet flow has only a

slight effect on -i£. Since the effect of velocity ratio ^L is not

J ©
involved in the equation, it is desirable to maintain the same ratio

value in model scaling.

The penetration depth of droplets in .Jet flow is also an important

index of atomization. Theoretical analysis shows that penetration depth

S of droplets is directly proportional to the square of diameter dg and

injection velocity v.. Thus the approximate quantitative equation*-1'*• 'J

is

Sccd^f-^y-V0'055. (4)

2. Mixing

Mixing of fuel and air in the combustion chamber of a gas turbine

is through turbulent diffusion. Using a combustion system wiT.h the same

kind of fuel and same temperature, similarity in mixing is obtained under

the following necessary condition:

Reg = constant, (5)

P v L
where Reg = -j? K is the Reynolds number of airflow. L is the scaling

dimension of the combustion chamber.

The combustible mixture system fcreed by droplets existing in both

liquid and gas phases satisfies only the equivalent Reg condition, and

- 5 -
is inadequate to warrant mixing similarly, which includes similarity

in final mixing results, i.e., similarity in the concentration fields.

Thus, the supplementary condition is

2
f = constant • 2X£i!l ■ constant, (6)

and
f' = constant, (6a)

where f is the overall fuel-to-air ratio and f' is the fuel-to-air ratio

at an arbitrary point in the combustion apace. Notation C in equation

(6) denotes the flow coefficient of a fuel nozzle. The flow coefficient

obtained by dimensional analysis is

C^f(Rej). (7)

The applicable quantitative relationship*-1^ may be taken as

C = otRej"9 . (7a)

To satisfy the conditions in equation (6a), in addition to similarity

in configurations and arrangements, the ratio of the nozzle design

dimension to the combustion chamber scaling length must be cons tarx-,

*1
— — constant. (8)
L

3. Vaporization

Vaporization of droplets in a combustion chamber is a very complex

process, because droplets change speed and temperature constantly after

injection from the nozzle. At the same time, external condiliuna and

influencing factors also vary constantly with droplet movements. In

- 6 -
 :
" ~"~'v#plJil?iÄ '

approximation treatment, however, droplet vaporization can be taken

as a process consisting of a nonsteady stage and a stead/ stage.

Droplets move with constantly declining speeds and rising

temperatures during the nonsteady period, as full entry into the main

stream has not yet been attained and flame has not yet started nearby.

Thus, droplets undergo pure vaporization during this period under a

forced convective condition.

When droplets enter the main stream, their velocity relative to

that of the main stream io taken as approximately zero. After passing

through the nonsteady period, droplets have gained a surface tempera-

ture to reach thermal equilibrium for steady vaporization. Entry into

the hot combustion region with flame nearby also enables the droplets

to reach a state for steady vaporization as well as for combustion.

As for temperature field similarity, the principal similarity

criterion governing nonsteady vaporization in combustion systems

producing reactants of the same composition and characteristics should

be the droplets' own Reynolds number Re^. In order to make nonsteady

processes similar, the same Reynolds number must be used for average

droplets in two similar jet flows, or

Red = Lglg^g = constant. (9)

It is generally believed that the nonsteady vaporization period

is much shorter than the steady combustion period, and this is

especially true for heavy fuels. Nonsteady vaporization becomes

relatively more important v&) in combustion of droplets under high

pressure.

- 7 -

1
■ ' ' " --Wr-»- -— -
 U. Steady Combustion

Vaporization and oombustion occur simultaneously under a steady

combustion condition. In order to make combustion processes similar,

the ratio of combustion time tD to the retardation time tr of droplets

in the oombustion chambers must be kept constant, or

T— - constant. (10)

According to a series of single-drop combustion tests and theoret-

ical analyses v-'* 1*0), the law governing combustion of droplets is

d2 ■ d£ - Kt, (11)
where d is the droplet diameter and K is the vaporisation constant.

Recent l^cerature lr*'l**J reports that the above equation is also ap-

plicable to various droplet sizes and comparatively homogeneous distri-

bution of dropleis in jet flow and to combustion with multiple dripping,

under the last condition, d should be the mean droplet diameter and K

the mean vaporization constant. According to combustion theories(l?' l1^',

vaporization constant
r«c-^_ln(i+B) , (12)

where \ and C» denote the coefficient of thermal conductivity and

specific heat of gas, respectively, and B is a dlmensionless transfer

parameter. K approaches the same value -ander identical fuel and tempera-

ture conditions. The effect of worldL-ag pressure p is. a combustion

chamber on vaporization constant 2 id atiZV \-. i^t, ruinate. According

Km** (13)

- 3 -'
 J
According to the nonsteady combustion theory' for combustion of

droplets in a high-pressure combustion chamber,

JCce> » (U)

Since the pressure effect on combustion velocity or combustion time is

still unclear, it is desirable to disregard this factor in model scaling.

If equation (11) and tp = -L are substituted in equation (10),

'g
then

_£ = _£_£. = constan^. (15)

KL

5. Stability of Combustion

Flame blowoff characteristics of i combustion chamber often serve

as an important index of simulation, and are considered in scaling scheme

selection. A flame stabilizer is generally installed in the combustion

chamber of a gas turbine. Since the temperature in the combustion gas

circulation region in the rear section of a flame stabilizer is rather

high, it 13 believed that complete atomiz&tion of droplets occurs in

this region and the following results can be obtained from a homogeneous

gas mixture system '3'°' t

VW*?11-1!., (16)

where V^0 is the critical flame blowoff velocity and n is the assumed

order of reaction. Equation (16) shows that the blowoff velocity can be

lowered in a reduced scale model with the working pressure remaining

unchanged. Under such a condition, it is difficult to obtain an airflow

velocity in a model higher than that in a real object without the aid of

- 9

HMH ~'
a flame stabilizer. In addition, the model has a very narrow working

simulation range. To make the blowoff characteristics similar, it is

necossary to satisfy

-H2 = constant. (17)

V
g

Since combustion is a very complex process, it is still impossible

at present to derive an appropriate similarity criterion strictly by

resorting to a system of equations describing a process. Similarity in

the actual performance of important component processes can be obtained

only on the basis of scaling requirements.

III. Analysis of Existing Model Scaling

Schemes I

The writer analyzed and calculated the existing scaling schemes on

the basis of the theory suggested above. Analysis was carried out in
i

two major steps. Step I was checking of the necessary conditions of

dimensionless similarity parameters in the processes, and Step II was

comparison of actual principal working parameters or important technical

Indices with those of modele Calculation results are tabulated in

Table 1.

Scheme I in Table I is actually the so-called Stewart scaling rule.

Calculation rosults in the table show that various .'ndividual dinKmsionleas

similarity parameters of that scheme in atomiz^tian are rather unsatis-

factory, as We* and IL are not identical, particularly the severe effect

of Weber number Wej for jet flow on the breakdown of liquid in the

- 10 -

!
 :

later period. The mean droplet diameter increases with the declining

We* value as the actual dimensions are reduced in scale. These two

opposing factors, however, nearly cancel each other as combustion

pressure increases. Thus, the final results in atomization, i.e.,

-SM/lsS ■ m,0,02 and jjj/jj1* = m^0'13 are still able to attain approxima-

tion. Mixing in Scheme I is comparatively satisfactory. First, the

overall fuel-to-air ratios are identical and the Reynolds numbers for

air flow are also identical. Secondly, although the concentration fields

cannot be kept strictly similar due to nonconformity in geometric scaling

of the nozzle and the combustion chamber, the condition is not considered

' serious. Moreover, the nozzle orifice diameter has not been reduced to

scale, and this practice makes construction even easier. The most serious

drawback of Scheme I is that dimensionless combustion time }h in the model

and in a real object cannot be identical, as the model has a much longer

combustion time. As a result, similarity in combustion of droplets cannot

be obtained, for combustion efficiency in the model may be higher. Another

drawback of Scheme I is that the blowoff velocity in the model is lower

than that in the real object, because the effect induced by a reduction

in flame stabilizer geometric dimensions is greater than the effect induced

by an increase in pressure. Thus, the blowcff characteristics are also

dissimilar. This scheme, howevor, exhibits superior simulation quality

as compared with th<* other achenes.

Scheme I has been supplemented by Keroert and Bamford*■'-», who

believe that similarities in atomization and in •rcubustion can be obtained

- 11 -

-__ ■ ■ f»aMtiiM *. i■ i■■■ MB ima+ataim&wm^m'mamwH'^jäKt-

■■■-
£=:. yjyjj^gbt^ygy^-^i^B^S^

•
m
M
40 « >>
>4?
82
a o <*>
•H 1-1
4» ü O
cd o -H
rlrl+»
rö O (D (0 aj
. J3
M 4»
P
<H a c u
«H o o
a •H 0) «H »
a n 4» > o
O (4 © •rl . »4 I-»
«H V N 43 M © t-,

I
4* 4» (V cd © ,0
parame
condl

IH* a 4»
© © 3 ©
.
u
o .
© «0 t-4 >
3 w > *o O
» © 1-1 H fc H
43 o © <M
0) ß fH © Ö£
85
SB n
© © B ©
« « a •>-!

•m ••s
••* •*K »»V m m
0) ••« 0J 4* ß © ©
ü 43 ü ß O i-t 1-4

1 i
•H
V ts a
n
•H
4»
C0

£§.3
Cd «H
4» +>
G} C0
'} N
—( N
(H O
+> fl
3) • •>
1
O
ß
N
N

• 9> T» o
4» O
© o ä
■PII
o
a *o •> © tj e-
O ß . M O ß .
e «II O MO © « r-( © «H © i-l
X U |l
S<ß M

Is
°5^
5O r^S4 . r-( "»fr © 3
O
©ß

^»43
•»
O 43 co
© M
r-{ ß ©
© O .ß
C

43
-.
• 4» ß
ß ••> E ß ß © ß C
rrf H ,-1 «TV -4 © CM +3 cd © d bo d ce
a> © «H ••» © Ji 4J fl .C 43 ja +3
|H ß ••> © w H © ü co r-t • ü CG
<H -H © <w 4S H ß © 4> ß © -fr ß
S iH
© r-t
N i-» § B o «H a o
'J 71 C O
ü •
O
*H
ß O
•* O O
60 © ü N « 4J iH>k 43 0 «H H CT. +3 43 -H II S
C o w n j-i ß ß +3 £
«H dC "V, 4J '( COCV • © © KfV
+1 O y T- T3 CH
•3 4» 4^
c © *o
© B ß
ß ü »(13
5 II H •
^>V

B O- y
O TJ 43 tf»v
^Sl'o.
n TJ O Cd
■
• <*\ o o • O O
•H © *-!>©« CV ü • ■<-» fV O ü »
i-t tto n -P 0 a ••> IT-. T3
s,
■H
•
rsj
w ß
• © cd
"^■S 4>
• • O i-i ••> w
CV ST © t3 ■P Ü
«* «»,
b0 K ß «H
Q crj 43
ß
t-t ••»
43 O
•
•♦»CM
43
ß •
fl
TH
4-»
••<
O n C0 C © fr\ er' ci
(3 4s •» 3 a *•* 4» • ß .►3 03 © © »« 43 © ©
o C ß c» © "H -H 4-» 1^, ■n ••><r» 43 , •
£ igt) o
+> cd
«0 4> ß
© cd T ß M
4» 4» •* ß o 2
ü +3
43 0! LJ
~H a
t, o H
7>
c: « 43 ej LJ
0 w-i ^. 0 ,'
fl co o > ß ßL II Ü © o w II 43 © 0 u
O C -H 0 cd H% «-» 0 « H % ►
O O 4» f «4 •" JS tTl
# HOB
OJ C0.G
© B^^- M ß "^
> e-i o. T) g |r( ü mTl \i) H+) co 13 vO|
©
. .8
H(A ü xD
• •
H C*i »45 O I
• pC ! « JO
1
H M ,■4
ensuos «4
-<
i 1
- X2 -

- -".—r^' i—
 ■ --

CO
u ■p
o a.
05

Q>
to
M w
j3< ,o
§
O <D
O J3
g
9
•O O U
ß -P a>
S XI £
>
« o
o E •*
fi 3 M
a u C
-p « <H
v.1 «J iH
S tö •
(1) w
P, ß tn r-<
§ •»
P -P
+J ©
US
-P o
0> »-» «9 0)
r-1 .Q e a
ß O O 01
•H a) ©
r-i ÜJ ß
« (ti
;>> © er •
p. 4S-g
a. a SO O
2 JZ O •>-;
W -p T-t .a
•o e
ß TJ ß
•ri C «H rH
öS cd «0
O 0)
ß r4 H ß
C ©
ß T3 cc t;
VH 0 -p ß
C3
a C3
C
01 C .E rH
O Ä ■P <D
r-l +>
ß o
ri JC •n E
<D +>
> o tS <P
c x>
0.
>., ß S
5U.H ►7 O
c tl -P
•H W o
-P O ^c
rt «H « B O
rH O H *o
3 R «J ß
'J <D Ü c »
I—1 «ri "f-l •H
« 0+! ■P >o
O -H ß CtJ ß
1
«M <D *J
c a)
C <^i *G
M © «H AX
1

V
-p
O
Si

S^JOTUSy

- 13 -

--
 ™^iffflffffMlffffll^^ -■_ — -, ■*.--■. -■. -..-...

rS <B 01 © -P
r-4 s W c a
© M Q) © ©
c « T< r-l CUT»
o
«H
P
<f *
d
C
•H
4* P
£g
9
► X
•rl +>
.p <H O
m « «H
r-l a P
© © jo

-H-
1
1.

1
!!

: v '
i -
 I

<H -p
c^ 2 ü
3 o
O (-1
CD >

P c>
u
(U +3
CO
a i
o 3
■p
01
6?
a>
JZ O <H
■P
t-t o
U H

!0
3 >
o
a
o
z ko
•r»
-P
to
O
fc
3
u 3 03
,a o)
-p S a>
O U
CJ a,

a«
U

o h
ß
r-t <U

as
3
*>
O -P

o
•P H
« 0)
"3 >

CM >
•H «
G
fl) o
-t
I

** <H
1
«-1 -P
t> c P C C O
3 flj « TO 3 C
0) C H e BOH
. 4> H «
«: -o i-i jQ >

15
 :

•-

o
CM «X o
a »<l

Vi r-<
D
K &
a
o
3
| 9 i
a
(X,

- 16 -

1 ;
by using a centrifugal nozzje in the combustion chamber, as the mean

diameter of droplets injected by the nozzle in a model or in a real

object must be directly proportional to the square root of the combus-

tion chamber design dimension. Data for Scheme II are obtained from

Scheme I revised by supplementation. Because of the effect induced by

the coefficient of nozzle flow on droplet lize, which i- takeü in+-o

account by the theory suggested in this paper, and of the variation in

relationship between pressure and combustion velocity at an index of •£,

the fuel-to-air ratios f thus obtained are not identical, and the

dimensionless combustion times are also nonidentical. To sum up, the

results obtained do not differ greatly from those of Scheme I; but if

the pressure has no effects-:^n combustion velocity as assumed in

reference [7J, then similarity in combustion can be closer than that

shown in Scheme I.

Scheme III •& equivalent to the scheme described by Way'-2-', who

assumes reaction in a combustion chamber on the basis of homogeneous

gas reaction. According to this assumption, the relationships of the

working pressure in a combustion chamber with airflow velocity and

design dimensions for soaliug can be obtained by Damkohler'3 five

similarity criteria. Scheme Il^b in the table is obtained on the basis

if Way's assumption by talcing n = 1.7 to calculate various relative data.

Scheme Ilia is obtained on the basis of the nonhomogenecus drop control

theory, using Way's original data in calculation. When the theory

suggested in this article is applied, the relative dimensionless combus-

tion time will not be equal to A. In other words, the same combustion

- 17 -
 fasse«

time cannot be maintained when the process is controlled by a non-

homogeneous reaction, according to the Way scaling rule. It must be

pointed out that the Vtay scaling method uses the injection velocity

as the design velocity for calculating retardation time, and the writer

believes that the airflow velocity should be used as the design velocity.

Calculation results in Scheme III show that similarity in atomization

varies greatly from those in Schemes I and II j not only is it impossible

to keep dimonsionlese parameters for the process identical, especially

velocity ratio JL, but final results for atomization also differ greatly.

The model has a lower combustion efficiency, as the dimensionless droplet

diameter is much larger and the dimensionless combustion time shorter

than the values in a real object. Thus, according to theoretical analysis,

Scheme III is inferior to Schemes I and II.

The principal scaling rule for Scheme IV is suggested by Lebedev{8j.

In fact, this rule is merely Damkohler's first similarity criterion

D = -£ (where t< is the reaction time) and the Re number is assumed for

a "self-scaling" [Translator's note: May mean dynamically similar or

self-similar! condition (it is doubtful whether or not the experimental

condition reported by that paper actually attained a self-scaling condi-

tion). ThJs scheme is thus based entirely on homogeneous gas reaction

without taking mixing factors into sufficient account. Referenced

does not describe the experimental seeling conditions for a ->.czzle fully,

but only briefly mentions invariable supplied oi] pressure and reduction in

dimensionless nozzle orifice diameter with the geometric characteristics

- 18 -
 J
of a nozzle in the model remaining unchanged. The writer made calcula-

tions for such a condition on the basis of his own theory. From the

calculations it ia revealed that if initial conditions are maintained,

the nozzle orifice diameter in the model is reduced (L^/LH)" times

as compared to that in a real object. Although the mean droplet diameter

is reduced I^/LH times from its actual value through geometric scaling,

the dimensionless droplet diameter still varies greatly with the reduced

nozzle orifice diameter; thus making it impossible for atomization

processes to reach similarity. As the retardation time of droplets in

the real object and that of droplets with reduced diameter in the model

are the same, the relative dimensionless combustion time, not to mention

similar?.ty in various mixing parameters, also differs greatly. The writer

believes that the Lebedev gas scaling rule at p = constant and »,«L

is groundless, because fineness of atomization was not checked in the

original article for satisfactory explanation of rapid droplet valoriza-

tion into gas and rapid mixing of the gas with air in a model or under

actual conditions; thus the role of the similarity criterion controlling

atomization is rendered unimportant.

Brisid.nl.9-' applied a scaling rule in tests conducted under the

following requirements: Pj,v_ = constant, 1ÜÜ = uL • Ji, and assumption

e K
-3H si PII ^H
of Re in the self-scaling region of a combustion chamber. He failed,

however, to check the droplet diameter before combustion tests were

conducted. Nozzle scaling conditions were not clearly described and the

supplied fuel pressure at 40 kg/cm*- was only vaguely mentioned. Scaling

conditions were suggested, but not theoretically explained in detail.

- 19 -
Finally, he c. ^aidered that working similarity in combustion chambers

can be obtained with conditions verified by tests. Another noteworthy

phenomenon was the comparatively narrow working range obtained in the

tests. The coefficient of excess airo^ varied only between 1.3 and

1.7. The reason for such a narrow working simulation range is not

explained. According to the Briskin scaling conditions (assuming

v» = constant and dj^/djir = L^/LJJ) » ^he calculation results shown in

Scheme Va of Table 1 are obtained. The scaling method is characterized

by having nearly identical dimensionless combustion times. Atomisation

and mixing results as well aa various similarity parameters are un-

satisfactory. The blowoff characteristic in the model is far inferior

to the actual condition. This may be the main reason for the narrow test

range; inability to keep the processes in both systems similar during

changes in operation may also limit the working range in tests. Scheme Vb
1 a
is Va revised by assuming that the injection velocity is v,ocL • . The

pressure ratios governed by the law of change thus obtained may be closer

to Briskin1s test data.

IV. Search for possible Scaling Schemes for

• Combustion Chambers on the Basis of
Theories Governing Nonhomogeneous Droplet
Combustion

It can be seen from the theoretical analysis in the preceding

sections that the optimum scheme among the existing scaling schemes is

the one using pL = constant as a scaling rale. For further discussion

of scaling schemes and 'dependence relationships between parameters under

a given necessary scaling condition, investigation of the following

- 20 -
 -.T.-- .^ar«^--«^?"T w

aspects is made on the basis of the theory suggested in Section II.

According to the theoretical data for controlling droplet combus-

tion in jet flow, the writer takes 1) fuel inlet temperature T0 = constant!

2) identical fuel characteristicsj 3) fuel-to-air ratio f = constant;

4.) geometric similarities in combustion chambers and nozzles, but dj/L

not necessarily identical; 5) Reg = constant; and 6) dimensionless

combustion time tD/tr = constant and other conditions to be considered

as necessary scaling conditions for the problems under discussion. The

first three conditions simplify theoretical analysis greatly, because

many effects induced by physical parameters can be largely eliminated when

the same fuel (including f = constant) and identical reaction temperature

conditions are used in a model and a real object. Geometric similarity

of nozzles is especially necessary for scaling under condition Ui other-

wise difficulties are involved in obtaining identical atomization charac-

teristic parameters such as injection angle Q and distribution of various

dropl-yt sizes. To make requirements less rigid in construction of nozzles

for a model, dj/L need not necessarily be a constant. In scheme selection,

suppose the injection angles are identical, the Reg numbers of the

principal similarity criterion controlling nixing are equivalent, and the

mean dimensionless droplet diameters and penetration depths in both systems

can be made as nearly identical as possible. Then similar4.ty in concentra-

tion distribution will not be seriously affected, even if the geometric

dimensions of the nozzle and combustion do not strictly concur. The most

important similarity criterion is tb/tr = constant, because it determines

similarity in steady vaporization and corbustion of droplets in jet flow.

-a-
 HJ32 -^KSXß^'i^BI&l^s&mtz^ttiTlse^--~ ■ .--^r-

In searching for possible schemes in addition to these five necessary

scaling conditions, the atomization effects must be kept as closely
identical as possible, first by making ds/dj = constant and then by
giving more attention to the velocity ratio XL, because the latter plays
an important role in atomization concerning the mean droplet diameter.
tfonateady vaporisation is not a deciding factor, as its duration is very
short as compared with the steady state. The blowoff characteristic is
a comparatively important simulation index under certain conditions, but
can be disregarded when certain equipment is used. Nevertheless, the
blowoff limit may often become a parameter determining the selection of
a scheue, because many schemes have various equally satisfactory Indices,
but a low diraonaionless blowoff velocity, These schemes cannot be
applied unless a special flame stabilizer is installed.
A system of fundamental equations for solving various problems can
be developed on the basis of the above conditions and the theory in Section
II.
1) Overall fuel-to-air equation {from ecaation (6)j :

d
jH PH yl H V
gK L
H

2) Atomization equatj.cn l_fr~- ■wuaiion (3)1:

d d V
sH jH /H PS

3) Equation for combustion of droplets in jet flow [obtained from

equations (10) and (11)1 :

- 22 -
 u L v
sH fl gM Pfc

where x is an Index determining the effect of combustion pressure on combus-

tion velocity or vaporization rate. If we lot Kffp1"'^ from expression (13),

then x = 1/8 in the above equation. If prsssare has no effects on vaporiza-

tion rate K, then x = 0. If we 1st K(Xp~l/3 a3 in expression (14.) for drop-

let combustion under high pressure, then x = -I/o. la the writer's opinion,

pressure mu3t have effects OK vaporization velocity, particularly when the

working pressure in a real object differs greatly from that in a model.

When the working pressure exceeds the critical pressure during droplet

vaporisation, the vaporization mechanism may undergo a pronounced change.

Twc separate results are shown in the following calculations by taking

x as 1/8 and as 0.

4-) Mixing equation (Re„ ~ constant): ■i

SHaä.Jl.
v
(23)
gH PM %

Six parameters, i.e., £üf, li«, M, % M and % are included in

d
jH ¥JH vgH PH dsH «H
1S
the four fundamental equations. If (IMAH) taken as a known parameter,
one of the five unknowns must be taken as an independent variable, leaving
the variation relationships Of the other four dependent variables to be
determined.

Thus, a set of function relations can be do rived from equations (20)

and (23) at x = 1/8 as follows;.

- Zi -
 L v
(£5)
jH R iH
and

\26,

(27)

■ ■

................. , etc.
Siailarly, at x = 0 when the eoabustion time is not affected by pressure,
s sure,
i
another set of function relations is obtained as follows:
1:
3i- Ä-0.925/Ü«»-1.4. (24a)
PH " V W ' ■
■

(25a)
'gH ^B

daÄs (^0.52^-0.7^ (26a)

■

d
sK h. ;H i
and

d v
(27a)
jH % ?M(sicj

...<....«......*.......,..,.,,..., etc.
■

Injection velocity is used as a vaiiatlij in calculation oy r.r.in 3et of

equations, and the conpletu re&ults are sh^wn in Tables £A and 2B, where
the injection velocity varies fron *tccL no «-,cc.:'' , Table 2A snows what
dixensionloss droplet diameter and penetration depth ratios are close to 1

- ?J

\
¥.:

I• c
TS

0
n •H
ß p $
m'S 9) jj
O a)
St OJtn
ß a>
0
X!
w .. S) as«
i o >»
•H rH
JJ <-<
f>3 H
<n
5n>ß
tö n S
•r-4 0)
P
ß
m «ö ü
00 O O cn to
in H ü %-,
g O ß
t> ß ü u
Oj

ti r-4 ♦3 I
4i
■»■» ß
a
> ß gV 'J o
G »
4J
$
3 it
€ ß
01
ß
<n
ß 4. rn ^ !
Ü3 1-4° ■»->
a.
R C0
3 N
!7i DJ o T4 ß
n UJ H c O
* g 8f •<H
OJ
M »H
ß l-<
p ß
•» «H
cö P
ß. e ß •H i-t
ai C « ■ö
ha n +■'
a
v
•H "H
•P-P-O
W '-> c
c> ß n
J es
1ß t^
a
01
0)
r-l
o
o
i S ß
5)
O
c •u a> rt to
•H •ö -r-, ß f)
r-< •H ß -
o ü< ß es
C! ■H +> *H o p
O OB 0
tr. p
(0 © r-< ff* 0 -P ß p * ß
d <D ß o o
ß 4>
3,? «
4-
o u
e •T-»

0 C5 *3 fl ß
0) g^n S i- «
ü
ß ä n ß
c © c O i-4 o
o ^
a * " in a t>t
? -n ß 4". e ■5. cn
ß jq
p
4» h 43
ti
ß •P X> ja
c a c
p j^
u H O o <3
c •H P 4) X
«rl C P
ty c x: P
■*- V.
c C a) -P TH
(C •• O P tr. (Q X> O
x: «'■HD •.» ä ß ß
ü n b4> C £ 5 ■P 4-> O 3
o sö n o +3 t. •r-f
UOi-i UJ o t: -H>
ß -P es .ß ß -p
V •»■■ »3 R II 1
o <u
TJ O O O * ü
;c ^-, D.
c o •HOW) SD, > ■* & +>
o r-t 4J i
c o ß O C
O. V T- "> « 0 ß c. o
w > TJ +3 O ß
CJ :-H -a
Ö ß — 4-i Kl
u C O « >>
-■>
a' ß ä O C
u o
a ■H
O -P r;
in ci p r-t °i
o +»
s
u> ß >
ß O -rl
•rt O -P c c
r.
ä i GI T3l

* P
< 1-» F~i ü •H Ci T4
o. c ■o o r2 re c
•H
03 II
O
O
P. o
<£ : « H 09 ß -*J
S) u «ß. m ß j^ a- -C ß
e> ß Ri U-j o ß
r-t r-i
1
a> •-» •>. J4*| « Öl
XJ OJ r- ••> ß c -P m US oj
ß 9 <C r-i •rt ■>-!
H ■■-. C aJ - O <v
•H O ß r-t O
O TH SS © r-4 ill 5 O .;l
3
ß 4-
wH C
—"i
«rj is* >>SE n, <5; es
A.
ii« £
Pk TJ il
—-«H U
aoatiüS rs M *r> •< w\>c r^ O iO C -- -\ r^ 3 O

- 25

«- -i-v.n uix . i r.a M,

•:
^«*«iA^«ffWKas<^^
 ^^VB^^^^mmswx-m^-^ri - :<j g*re*B^g£7--V^ - ■

R O
P
fi
r-( T<
<B 0)

o CD .3 § 1 "8O 5CO
1-1 «
.? «
00 H
09
"•1
TJ •
00
r-4
•H
03
•n a
ffl
0 2S
P CO
Ct> fi
bl
2$
^ x:
a; o
•H rH

CO o
«)
p. '"■>
o? C E *-< P
rH J o a o •» CD
Ü o to P
m TJ a o
c: 0 o CD CD
rH o o P 03 e .C
•H >H CD 43
P P r-" f-i f»

* a ö
gf
O C
t/3 O

r^i
et
o o
T3 O
O O CD
r~i
o a . c 0)
35 fi
Oi
i—1
z>
p •.
6
no
d> j3

§ TJ
to to co X! «
en m 0 fi
#H •H a> co 02 TH
rH P H 3 d
■B
CD a» C O
O
^ Xi
3 P
o
•H
ca
«H
S
0 0
i-l to 01 1)
fi a CD
a> <D CD CO cd U
0s a -ri CP <d
CU>H
.. -P TJ u t> <D
CO |p CO c a
e c fi p cd oi
c- 5 O P x:
S»H P
o m
a)
tXVH
■P o
M 01
CD
vO * 3 V. CD
X> fn cd
VN -? O
@ > P 'ö x:
CD K H-^
rä ° O P
■sf cd P
■p
33 CD V.

(*v
V
r-t
tn a. ffl
• 2 fi. ~P
cd

SV
•H
£ fc 3
O ri
fe °
■P TH
fi <d
0 tj
P «ö to 01 !
h C 4)
u a!
O i-H
•r1 f> ä
rH OOP rH fi
H 3
cs 0 I
v. a © o
in > rH a w
C P 4- 0) 0 <d •
v^ O o o
■H a»
P tu -P
o CD
H
TJ
JO t-i fi
C
-P M
c-<
O •
a
CCS
o <D a
• 3 TJ Oi TJ
O O fi C^ CD £t X. TO
e p
o r-t fi O cd ö H
cd P. CD Pi a» P x:
«r> p cd (JO* H->
-P CD U> C-. CD
o -P Sss 2c c >4 fi
H X! «H
^;
H h
CD ox; c,
fr, P «-
01 +' o
fi +> G r.
--* ..o 1
O o s% • • •
rH
^•tf i-t C
O H ii
Hg > *
ÄBei^cg t-t oi c<^-«* <r.\f. £^ «0 ^OHNC\«4
r-i r H i-H rH m «
i

26 -
 when Vi varies from Ir'8 to L ' . The velocity ratios of the model and

the real object do not exceed 0,8$ at I#Ad - $. The nozstle orifice

diameter relative to the injection velocity changes between (WJ[»H)

and (I^AH)0'^« Therefore, it is possible to obtain similarity in combus-

tion conditions, if this r- «jion ie used for scaling. Scheme 7, in which

v, = constant, happens to fall in this region. Scheme 6 (Stewart scheme)

is also inserted in Table 2A especially for comparison. Prom a theoretical

viewpoint, the Stewart scheme is not a unique possible scaling scheme.

The next scheme* however, uses the same fuel injection and air inlet

velocities to eliminato the effect of velocity ratio on atomization results,

and its blowofi characteristic is superior to that of Schemes 5—10 in

Table 2A,/ As noted from the blowoff characteristic, it is a disadvantage

to use an injection velocity in the model higher than that in the real

object; otherwise, the applicability of all the above scaling schemes can-
not be ensured.

Curves showing the variation of relative dimension!ess similarity

parameters and of principal scaling indices with injection velocity at a

geometric scale of m, = pi - £ are plotted in Fig. 1 and 2.

Fig. 1 shows that all dimension!ess similarity parameters obtained

at the ^ a 0. U—1.09 variation range (equivalent to the VjCCir1'8

region) are ratter satisfactory. The most seriously impaired is the

blowoff characteristic. Blowoff velocity in the model ux-^er the above

scaling conditions is always lower than that in a real object; thus it

ia difficult to make (V^o/V») a constant unless the air inlet velocity

is greatly reduced. It is possible only by aaking Y£M equal to 0,8, or

- 27 -
equivalent to about VjocL*. Ae noted, the optimum scaling region should

be in the v,otL*—V, = v» range. Various important scaling indices

relative to this variation region may be obtained from Fig. 2 or Table 2A.

The dotted lines in Fig. 1 and 2 indicate v, in the variation region from

VjOtL» to VjOCL"*. Parameters deviate greatly outside this region, with

similarity in processes seriously Impaired and, especially, reduction of

blowoff velocity in the model. As a result, similarity in combustion can-

not be realised unless a flame stabilizer is installed in the model. Since

It Is still difficult to evaluate blowoff characteristics in a combustion

chamber quantitatively, the above analysis can be used only as a qualitative

reference.

.Suppose pressure has no effects on vaporization rate (see Table 2B);

conditions are then slightly different. First, with other important

scaling conditions identical, a higher working pressure is to be applied • j

to the model. Secondly, the optimum scaling scheme deviates slightly, but

the region in which similarity of processes may be maintained remains at * |

VjOtL1'8—v.diir1/8, and is optimum at variations between v^ccL1^ and

VjCCL . |

Therefore, regardless of whether or not pressure has effects (not

counting negative index effects) on combustion velocity, the optimum ,

soeling scheme occurs in a region where injection velocity is a constant

i
or la slightly lower than that of a real object. If scaling is carried
I
out in or near this range, there is a possibility of obtaining satisfactory

results. If the flame stabilization capability and simulation performance

of a model with variable working conditions are to be Improved, it is <

-28-" !

• i
 f

'''ig. 1, Variation relationship betveo? relative

dimensionless parameters and fuel injection
velocities at a geometric scale of 1 : 2

1 - Dimensionless droplet diameter ratio (da/dj);

2 - dimension penetration depth ratio (s/L)» 3 -
dimensionless blowoff velocity ratio (v^^/v );
U - dimensionless droplet Reynolds ratio (Re^/dj);
5 - dimensionless nozzle geometry ratio (d*/L);
and 6 - dimensionless velocity ratio (vi/v ).
■ ©

, (Ordinates on the left are used for curves 1, 2, 3,

and 6; ordinates on the right are used for curves
k and 5)

Dotted lines indicate the variation region at injec-

tion velocities from V.OCL*" to v,<ciJ~T.

_ 2o -
 \

^f;i2;ndC^aSl°2 ri ClPal SCalinß

? VersU3
bJeCt P»~te» 1» a
MML f? x injection velocity
ratios at a geometric scale of 1x2 -"«W

1 - Nozzle orifice diameter (dJM/djH); 2 - blowoff

velocity (vboM/vboH). 3 - airflow velocity {v^/v ).
A - combustion pressure (j^), 5 . a«^ JJJ

(Ordinates on the left are used for curves 1 2 «nrt

5; ordinates on the right ar.usec fServes 3'a£\)
Dotted lines indicate,the variation region at injection
velocities from Vj«i> to y,3°

- 30 -
 necessary to select a lower injection velocity, a correspondingly-

reduced airflow velocity to match, and an appropriately higher working

pressure. The latter, however, is disadvantageous to power loss fron

air supply. Under the above scaling conditions, the blowoff ratio

does not vaiy with the changing injection velocity, but power loss

increases with increasing combustion, pressure. Therefore, regardless

s
of whether scaling is carried out according to the rule with pL

constant 'or to the rule suggested in this section, the pressure in a

model is to be higher than that in a real object. This is applicable

only to models having a low initial working pressure, as in the case

of scaling an aircraft combustion chambers; otherwise, elaborate air

supply equipment is needed.

V. Investigation of Approximate Scaling

Schemes under Isobaric Condition

The application of a high-pressure scheme for scaling a large

complicated gas turbine engine with high-compression ratio has the

following two disadvantages:

1) The combustion press-re in the original engine is already

high. If the working pressure in the model is further increased, then

it is necessary to use a compressor having a higher compression ratio

than the actual equipment, and this is disadvantageous to test condi-

tions.

2) The effects of combustion pressure on vaporization and combus-

tion are as yet unclarified at present, and test data on combustion under

high pressure are still scarce. The vaporization and combustion

-.31 -
mechanisms ma r undergo a pronounced change when the working pressure

in a model exceeds that in a real object.

Oil the basis of the above principles, it is best to use an
isobaric scaling scheme (the same combustion pressure for model and
real object). It is difficult, however, to obtain the same Reynolds
number for airflow by applying the isobaric scaling method whon
geometric dimensions of the model are reduced. As analyzed above,
the airflow velocity in the model may exceed the blowoff velocity
permitted by the flame stabilizer. When scaling under the isobaric
condition is carried out, leniency in. mixing requirements is inevitable.
In other words, Reg cannot be maintained strictly identical, and the
atomization criterion also loses its effectiveness.
Comparison of calculations shows that a rather satisfactory
approximate scaling scheme under the isobaric condition is: 1) T0 =
constant; 2) fuel properties identical, f = constant; 3) vg = constant;
and 4) tt/tj. = constant. The corresponding important scaling indices
are:
UM- Ä0.21. diM ,%0.96 M % = ,^0.5. Gv _ q, 2

'!(?)= (=)
'g V V Li-'

(U) = &)-°'°U: (Re.) = Ä,^^

v :
"H 41 boH "H
Suppose airflow velocity in a model is lowe. than that in a real
object. Then the atomizavLon, vaporization, and mixing conditions will

- 32 -
be less favorable. Further increase of the airflow velocity in a

model will affeot the blowoff condition even more seriously.

As noted, an isobaric scaling scheme is applicable only to high-

flow engines having a high compression ratio. Re_, even reduced in

the model, can still remain in the self-scaling region, because the

Reynolds number of the airflow in the ret..? object is very high.

VI. Conclusions

1. The approximate scaling rule for a combustion chamber is not

unique, under conditions analyzed in Section 17, satisfactory approxi-

mation may be obtained with fuel injection velocities in the region

from V,OCL* to v.ocL"1/16.

2. Variation of fuel Injection pressure and nozzle outlet diameter

has a great effect on approximation in scaling. Thus, a scaling rule

obtained by coincidence in a single test without clearly designated

working conditions for the nozzle is undefendable. Even the accuracy

of an established scaling rule can be affected by a slight error in

nozzle construction.

3. Among the existing scaling rules, a rule with pL = constant is

the most suitable, particularly for scaling of jet engine combustion

chambers.

4.. Nothing prevents testing high-flow gas turbine engines having

a high compression ratio with the approximate scaling scheme. Similarity

in combustion conditions may be obtained by the scaling rule using

p « constant, Vg ■ constant, VjOtL1'^, and dj<*L when the Reynolds number

FTD-HT-23-1U83-67 - 33 -
f\>r airflow in a real object exceeds 4 x 10->. Low-pressure simulation

teats conducted under low Reynolds numbers can only be used as a rough

check on performance sensitivity,

5. The theoretical analytical method in this paper may be used to

evaluate scaling materials qualitatively for reference in future studies.

Bibliography

(1) Stewart, D. G., Agard Selected Combustion Problems Vol. II,

p. 384(1956).

(2) Way, S., Agard Selected Combustion Problems Vol. II, p. 296(1956).

(3) Way, S., Jet Propulsion Vol. 27, p. 162(1957).

(4) Bragg,S. L., and Holliday, J. B., Agard Selected Combustion

Problems Vol. n, p. 270(1956).

(5) Greenhough, V. W., and Lefebvre, A. H., 6 th. Sym on Combustion

p. 858(1956).

(6) Lefebvre, A. H., and Halls, G. A., Advanced Aero-Engine Testing

p. 177(1959).

(7) Herbert, M. V., and Bamford, J. A., (Combustion and Flame) no. 3,

p. 360(1957): no. 1, p. 35(1961). •

(8) Lebedev, B. P., Izv. vyssh. ucheb. sav. (Mashinostroyeniye),

no. 2, p. 177(1960).

(9) Briskin, S. L., Sudostroyeniye, no. 11, p. 22(1957).

(10) Giffen, E., and Muraszew, A., The Ax-oiization of Liquid Fuels(1953).

(11) Basic Considerations in The Combustion of Hydrocarbon Fuels With

Air, 1957, Edited by H. C. Barnett and R. R. Hibbard.

FTD-KT-23-1U83-67 - A -
 (12) Ml esse, C, Indust. and Engng, Chea. Vol. 47, no. 9, p. 1690(1955).

(13) Novikov, I, I., Engng. Dig. Vol. 10, p. 72(1949).

(14) Garner, P. fU, and Henny, V. E., Fuel (London) Vol. 32, p. 151(1953).

(15) Radcliffe, A., Proc. Inst. Mech. Engng. Vol. 16?, p. 93(1955).

(16) Spaldlng, D. B., ABS Journal, Vol. 29, no. 11, p. 828(1959).

(17) Godsave, G. A., 4th Sym. on Combustion p. 819(1953).

(18) Spaldlng, D. B., Some Fundamentale of Combustion, 1955.

(19) Rumagai, S., and Iaoda H., 5th. Sym. on Combustion p, 129 (1955).

(20) Hall, A. R., and Diedrichaen, J., 4th. Sym. on Combustion p. 837
(1953).

(21) Bolt, D., and Boyle, T., Arbor, A», Transc of ASME Vol. 78, no. >,
p. 610(1956).

(22) Rex, J. F,, Funs, A. E., Penner, S. S., Jet Propulaion Vol. 26,
p. 179(1956).

F?D-:-!T-23-li1e3-67 - 35 -