(Reprrnkdfrorn ARS J O C R V ~ L , 4u;lust.
1962)
Copyright, 1962, by the Arnerican I k d e t Society, Inv.. and reprinted 1~ p e r i r k i o n of the copyright owner.
Thermodynamic Extrapolation of
Rocket Performance Parameters
First and second partial derivatives of some thermodynamically defined rocket performance
parameters are presented. Theae derivatives are useful for the extrapolation and interpolation of
thermodynamic performance calculations for changes in combustion pressure, combustion en-
thalpy, pressure ratio, and area ratio. The accuracy of the extrapolation or interpolation is indi-
cated for a typical propellant combination.
FOR SOME propellant systems, performance data are
needed over a wide range of conditions such as chamber
pressure, pressure ratio or area ratio, oxidant to fuel ratio,
and, occasionally, initial enthalpy. The thermodynamic cal-
culation of rocket performance is usually sufficiently difficult
to require the use of automatic computers. Because of the
cost and time involved in the computations, it is often not
feasible to explore the complete range of independent vari-
ables. For this reason, the problem of extrapolating or inter-
polating a limited number of performance calculations occurs
frequently.
The use of partial derivatives in extrapolating performance
data with considerable accuracy for moderate changes in the
independent variables was discussed previously by Gordon
and Huff (l).a In Ref. 1, expressions were derived for first
partial derivatives of the logarithm of specific impulse, charac-
teristic velocity, and area ratio with respect to the logarithm
of chamber pressure. In addition, first- and second-order
corrections to impulse for a change in combustion enthalpy
were also given. A similar first-order correction for the effect
of combustion enthalpy (or heat of formation) on impulse was
described by Gordon (2).
I n this report, the list of partial derivatives is extended to
include the derivatives of specific impulse I , specific impulse
in vacuum I,.,, area per unit mass flow rate A / w , area ratio e,
and characteristic velocity c* with respect to combustion
pressure P,, combustion enthalpy h,, pressure ratio P J P , and
area ratio e. Examples are given to illustrate the use of the
derivatives.
RRceived March 7, 1962.
1 Head, Thermodynanics Section. Member ARS.
* Aeronautical Research Scientist.
Numbers in parentheses indicate References at end of paper.
AUGCIST
1962
SANFORD GORDON'
FRANK J. ZELEZNIK2
NASA Lewis Research Center
Cleveland, Ohio
Analytical expressions for the derivatives of parameters
with respect to oxidant to fuel ratio O/F can be derived; how-
ever, the method of derivation of these expressions is some-
what different from that used to obtain derivatives for a fixed
O/F, and, therefore will not be presented in this paper. All
derivatives in this paper are for a fixed O/F.
Effect of Change in Initial Conditions
on End Point of a Process
All the thermodynamic properties of a system of known
composition can be specified uniquely in terms of any two
thermodynamic functions, say a! and 8. Thus the system can
be represented by a point in a tnwdimensional space with
coordinates (a,@. At any point in (a$) space, not only ape
all the thermodynamic properties of the system known, but
it is also possible to determine the rate of change of these
properties along some curve in (a$) space. Thus if I) is a
third thermodynamic function, the derivative (bj3/da)+ ex-
presses the rate of change of j3 with respect to a change in a
along a curve of constant I). This partial derivative is the
usual thermodynamic first partial derivative discussed in
thermodynamic textbooks.
Thermodynamic Processes
A process in thermodynamics means that a system orig-
) been moved to a new point
inally a t some point ( ~ r ~ , j 3 ~has
(a!,@. An infinitesimal process can be completely charac-
terized by a derivative of the form (bj3/ba)+. A finite process
can be specified by giving a starting point (o~g,&), a path,
say a curve of constant I), and one of the coordinates of the
end point a. For a given path, the only independent vari-
ables of the process are the coordinates of the initial point
(a!o,j30) and a coordinate, say a,of the final point. The other
1195
coordinate of the final point, 0, and all other thermodynamic
functions are dependent variables in the process.
Let cp be any dependent thermodynamic variable associated
with the end point of a finite process. The rates of change of
cp with respect to the independent thermodynamic variables
of the process are of two different types. For a change in a,
the usual type of derivative (3cp/ba)+is needed. However,
for a change in one of the coordinates of the initial points,
derivatives of the type ( b p / a a o ) p oand (bp/b/3,,),, are needed.
Expressions for the latter type of derivative are obtained in
the following section.
Derivatives Associated With a Finite Process
As indicated in the previous section, the process is defined
as taking place along a curve of constant J.. Therefore
+o(ao, Po) = $(a,P) = +(a,cp)
Whatever change in +o results from a change in the initial
point (aO,Po)must be equal to the change in J.. This may be
expressed in differential form as
Imposing in turn the conditions of constant a. and constant Po
gives the following desired expressions for the partial deriva-
tives of a function a t the end point of a process with respect to
the initial coordinates:
In Eqs. [3 and 41 all the derivatives except (da/b/30)~,
and
(3a/bao)po are the standard thermodynamic first derivatives
and can be immediately evaluated. The two exceptions can
be evaluated for a specified form of the relation
a = d a o , Po)
Two forms of Eq. [5] are considered in this paper: a = kl
and a = kZaO,where kl 2nd k2 are constants. For a = kl, Eqs.
(3and 41 reduce to
For the particular choice QI = klao,Eqs. [3 a d 41 give
It may be seen that the right-hand sides of Eqs. [6 and S]
are identical.
Application to Rocket Performance
First Derivatives of Thermodynamic Functions With
Respect to P,, h,, and P J P for an Isentropic Process
In t,his paper, Eqs. [6-91 are applied to an isentropic proc-
ess. For an isentropic expansion from P , and h, to an exit
pressure P I the follon-ing correspondences to functions in
Eqs. [6-91 apply: s, = Go, P, = (yo, P = a, h, = PO, and
s = $. From Eqs. [Sand 81
1196
FromEq. [7]
From Eq. [9]
For any cp selected, the right-hand sides of Eqs. [lo-121
[I I are in terms of the usual first partial derivatives. Bridgman
(3) presents a convenient scheme (given in many thermody-
namic textbooks) for expressing all first partial derivatives in
terms of three first partial derivatives, namely, ( a h / b T ) ~=
e,, (bv/bT)p,and ( a ~ / b P ) ~
An. equation of state is required
to evaluate these derivatives. I n rocket performance calcu-
lations it is convenient to assume that the reaction products
obey an ideal equation of state with a variable molecular
weight
PV = R T / M [I31
I t has been found useful in performance calculations to
calculate molecular weight derivatives rather than volume
derivatives to indicate changes in composition. From Eq.
[I 31 there follows
A further discussion of the molecular weight derivatives and a
[51 method for their numerical evaluation is given by Gordon et
al. (4). With the aid of Bridgman's tables (3) and Eqs.
(13-151, all other first partial derivatives ran he expressed
in terms of c, ( b In A f / b In T ) p and ( b In M/a In P ) T . As
an example, assuming cp = T , Eq. [12] becomes
[71 -v, T
- -
T, c, P. c, ( r T ) p
3 In hl
-
In a manner similar to the previous example, derivatives
were obtained from Eqs. [10-12] for cp = T I h, and p . Ex-
pressions for the logarithmic form of these derivatives are
given in Table 1. The logarithmic form was selected t o pre-
sent the results in dimensionless form. Expressions for
molecular weight derivatives are also given in Table 1. These
are obtained by considering molecular weight to be a function
of temperature and pressure and expanding by the chain rule
in differentiation. Derivatives of the form [b In cp/b In
(P,/P)1p,,he = [b In p / ( b In ( P J P )I, = - ( b In cp/b In P)*are
also given in Table 1. Henceforth, derivatives a t constant
P, and h, will be shown as derivatives at constant s. The use
of these derivatives will be discussed in a later section.
A word of caution is required about the derivatives with
respect to In h,. If a reference level for enthalpy is chosen so
that h, is negative, the logarithmic derivative must be re-
placed by a derivative with respect to h,.
ARS JOURNAL
 W
cr
'y-

0 * I2

A
n
a

+
3
-1 E 4

>-

T
4
4
h

h
10 im 4
w >
4
I W

-
a -
n
h h
u

h\
c d
c
4
b
4
n
P/,.
d

I I

+
4 *I%
*u s
3 i13 u 4

AUGUST1962 1197
Derivatives of Rocket Performance Parameters With
- -
Respect to P,, h,, and P , / P for Isentropic Expansion bxby
Parameters that are usually of interest in performance cal- Detailed expressions for the second partial derivatives of I2
culations are defined in the following equations, where it is obtained by means of [ a bJ are given in Table 2.
assumed that a consistent set of units is employed: The expressions for the f i s t derivatives of the remainder of
I = [2(hc - h)]"2 the performance parameters in Table 1, unlike specificim-
~ 7 1
pulse, include first derivatives of thermodynamic functions.
(S/w) = l / p u = l/pI 1181 The second derivatives of these performance parameters are
therefore not included in this paper, inasmuch as they involve
€ = (A/w)/(A/w)t= A / A t ~ 9 1 second derivatives of thermodynamic functions which are
E* = P,(A/w), [201 generally not available.
I",, = I + (A/w)P 1211 First Derivatives of Rocket Performance Parameters With
c* = I / c * (221 Respect to P,, h,, and E

A specific set of units and associated dimensional constants for
Eqs. [17-221 is given in Ref. 4.
Derivatives of the logarithm of I can be obtained directly
from the derivatives of the logarithm of h in Table 1. Thus,
for example
Eqs. [18-221 can be used to obtain derivatives of the re-
maining parameters in terms of the derivatives of the loga-
rithm I and the thermodynamic functions. The results are
included in Table 1.
Second Derivatives of Specific Impulse With Respect to
Pc, h,,and P c / P
An examination of Table 1 shows that the expressions for
the derivatives of the logarithm of I include only the thermc-
dynamic functions h, T, and M . Therefore, the second
derivative of the logarithm of I can be obtained by the
methods previously discussed. However, for purposes of
extrapolation, Izis a better form than In I . (This will be dis-
cussed further in a later section.) The first and second
derivatives of I* can be expressed in terms of logarithmic
derivatives as follows:
1198
The previous discussion of thermodynamic processes
showed that a process could be specified by giving the value
of two thermodynamic functions at the initial point, a path
connecting the final point and initial point, and one of the
coordinates of the final point. Thus for a given process, any
parameter of the process X can be considered a function of the
coordinates of the initial point and the coordinate of the
final point. For the isentropic expansion of gases in a rocket
nozzle, i t is permissible to write
X = U P , , h,, P d P ) [=1
and for area ratio in particular
€ = €(P., h., P J P ) t26 1
Eliminating P J P between the previous two equations gives
A = w,,h., E) [271
where the functional forms of Eqs. 125 and 271 are not the
same.
If the functional relation in Eq. [25] were known, then this
would provide an alternate method of obtaining the first,
second, and fourth columns of derivatives in Table 1, Le.,
( b in A/b In Pc)PJP,h., ( b In Xjb In ~ J P J P , P ~and , [bIn X/b
In ( P c / P ) . ] . Similarly, if the functional relation in Eq. [27J
were known, the partial derivatives (b In X/b In Pc)t,ho(b In
X/b In hC)*,pc, and ( b In X/b In E ) , could be obtained. How-
ever, expressions for these last three derivatives need not be
AF& JOGRNAL
 C O O m h * 3 f r n Q L n t o m - o --Ob m-m O N f O O O
r o e s .C

- ..
- C - N O N om-a -morn E)=* ONIIO-o
x .ommo 0 0 0 - N f *m-- D m D O

.... .... . . .
L O - 0 0 D 0 0 0
YOO-om
0 .
00
0
e.
N O
1
. O O Q N .
m o
.oo-
.rn
0 1
O N t -
0000
0 0 0 0
- f N D
-=NO
0000
NE0
0-h
0 0 0
......
O O D O O 3
0 0 0 0 0 0
0 0 0 0 0 0

.
I f
t-OCODf N Q C l C f D CDCNN- e m - D CfNe- C O W rnma-mc
-ON- .C ~ f ~ m e 0o. .
0 N N m t 00I-m D O C o m ~ o o m
oo--mc -emm
f. - m e e -
X O N Q f O NO DO^ e c m o N m 0 - 0
tu mo-om
0.
00 N O
.OOC.-.
C O
.oo-
* . .f F .NmQ
N N t f
.... .... . . .
0-n-I-
0 0 0 0
-I---
-"-0
m m o
0-m
......
O O ~ O O O
O O D O O O
0 1 0 1 0 0 0 0 0000 0 0 0 000000
m t I I
C O O 0 - a
r o m m
h N Q l c Q 0 N--0 O O O C cco m - s ~ - e
.C o-m-am m N D * W Q * * O Q D m m ~ 0 - m
x o o 0 . C o
?222)c"? 0mm- mtmrn
e o . . .f
.... .... ... ......
C W 0 O D N O f O
Ly . O - f D

-
0.
00
0
N .
Nm
1
- 0
.oo-
1
OON-
0000
0000
I t
0 0 - 0
-N-0

0 0 0 0
fN-
0-m
000
I
O OD
O mOOO OO

0 0 0 0 0 0
o

CO-mQf mobfm- CmNO t O C C 0-m L n C Q f t t

- 0 m m .C ~ c a m o o e*-* CCo.rn -NC m-Q-Ot
xomc-mo m ~ m ~ m - om-- *&Om omm
UI
0.
00
eo-czm
C

~m
.

-.000.-.
C O . .
.00-
.f .... .... ...
OON-
0 0 0 0
OI-OD
-N-0
mN-
o-m
O N C O m f

...
0 - C O O 0
y9eyoo
M I 0 1 0000 000 0 0 0 0 0 0
0 0
I 0
1 0

tx8o S $?E f m O C - 0
mm~r-am
mmwoo DmmD COOC - m e ~ 0 m t r - 0
O N 0 m a m a fmmm o m m m Q m ~ * ? -
0 -orno m m o m t o
.... ....
o m 0 0 9 0
......
Nb * f O C N t - -#*U7m

- ..
-3-0
m - ~ m m .o-o-.
0 Y
0 . C O . * .* OON- om(D0 m-N
3-m O-COOO

-
c
4 C . 0000 - N O 0 0 0 D 0 0 0
0 0 0 N O .o-r
v) 0 0 0 1 0 1 0 0 0 0 0 0 0 0 000 0 0 0 0 0 0

--
I 1 I

0
II Co--hs
-0ovI .h
LOC-mo
W O O O N
om-
~m
. 0-Nf-
D m r m
mem
oeo
0--t00
a - m f f e
--o.omm
.... .... ...
e o - - 0 c m
c
I
w
VI .-NO9
0 .
00
f
0.
N O
I
OON-
0000
0 0 0 0
o m e o
- N O 0
0000
2 2%
o-m
000
I
......
O N m O 0 0
0 0 0 0 0 0
0 0 0 0 0 0
U
z
Ly
C O W W N t N O Q f D I C mr-mom O m f D mrnmv- O f N m-emmm
-0o- .h o m m ~ - m O f O . . mme: 00CU7 NNN O f O Q O f
X O N N ~ O CQCQOC N m .Or) -*e- O f m m

......
m t o NmOOO-

- .... .... ...

4
-1 UJ . N N Q m e o - N - e .ma- 00-- 0-mD QON ONmO--
> 0. 0 . am -mart o-m ooeooo
" 0 0 -m
C O S
.o-- 0 0 0 0 - N O 0

3
0 - 1 0 1 0000 0000 0 0 0 0 0 0 0 0 0
I I

-
W
C O m D O f
- 0 o o .e
x o m * 0 0
mfmernb
N'VCYC-O
fN-rUNN
mU7--0
oo*.
. f N
. - Q Q D
0 f r - f
mf"-
Q-OC
- 0 m m
9 a m m
QQm
* D O
m-I-
N O ~ N O Q
m m c e = a
mmm-0m
3
-
0 u .ONCID
0 .
00
-
0 .
-m
1
-
.-me-.
-0.. .O
.o--
0 1
c s f - m
--a*
....
0-c-
0 0 0 0
0 0 0 0

I
OQNO
9 799
0 0 0 0
...
m o o
o o m
0 0 0
I
......
O f O O - N
0 0 0 0 0 0
000000

C Q O O N f otomm,r- mm0r-N O N O D ODmC 0 - O m ~ m e f m

I
A -*om .e -mmoom o z m . . N-Of m-fO c m m -C-mCm
m t o m - o -me cmmm
......
X 0 0 m . 0 0 NO ehN- moo. C O N N ~ C O
3
U

c
W w
Q
NO^
m- m .
-m
1
-.-Oh-.
C O .

0
am
.0--
1
c . 0 0 Q
--am
....
o---
0 0 0 0
0 0 0 0
I
0rn-D
9799
0000
...
o m 0
00-

0 0 0
I
O O D O - N
O O ~ O O O
0 0 0 0 0 0

z
W +cc.f-t mlnmom w o o e --Ne Nl-W NODQYIO
U -No+ .C ttOQIDU7 0 - 0 . h m 0 f mmmm C D N

......
C O ~ Q Q O
a x m o o w o -mmo-m

.... ....
NU7000 bob- QN-m 0omm-U3
Ly
n
m
Ly
0-
.N-m
f
s.
-I)
I
-.-to.-.
e o . . .N
0
.o--
1
"-'<25 ONO-
0 0 0 0
0 0 0 0
I
et-*
o-oo
0 0 0 0
0 9 m
- 0 -
. I .
0 0 0
I
0 0 0 O W O
o o m o o o
0 0 0 0 0 0
0
In
Q COU3LnOt
- 0 0 - .e
QmmaCO
m m - o e o
m m o o n
O C f . Q t h D
0 - 0 2
-m-c
DQfm
QOO
De-
IDfN-ON
fmcr-mf

.... .... ...

m x o m o w o **-e-- "=?-e me-m mmnDcm0
......
Ohm- O f N

-
3

-
-1 w . .mfm
0 .NQN-. e. . f a -NO- 0-00 o m 0 - W 0 3 N l n
L 0 0 0 . 00 .N -NOW 0 0 0 0 o-oo -0- oor-000
a c I " *ON-
2 I(
b-
0 1 0000
I
0 0 0 0 000
1
000000
U

-
U
2 0
\ uc00c-t fO--mo mmoon DC 0 e m c -m C0V)fF-N
O N f O el- e o o m c o o m o . . fm m
.-
f OQ
Q
0Qc mm~mu-im
U a c . ~ Y ) oD m O f N V N Q O Q Q N Q c t m m

.... ...
Ln

......
Q Q

I:
I
W
u
I . e m f m
b-c)
.*mm-.
N .
1 - -
mo..
.ON-
0 1
e . .t-*
0-N-
eo
0-
0 0 0 0
0
0
mm
QO
0 0 0
m-o-mo
O - c O O O
0 0 0 0 0 0

"
d o m o o f
t u o o m .C
Nmm--
rnomo-
- I

. 20..
0
m o .QOO C o t e m D
smmm m .mmm-. m
IO u-Q .O 0
I m .ow-
U 0 1 0
c I
z
4 an

-
d o Y U
z
3y -X
L
- I \\uu-1
uunau
mm-
--a
0
I

.. na--u
V O U -nu c
a
In
c XU-
U
u
- tw
mln
--uur
uuxx
S X O O Z
uua
aa-1
-1dn
on-
U
a - -
- 1 1 OD\\-
\- \c=4 V 0 0

----.---
a -\am

-
c .ID
4 U U A -cuuu Ly 00-vv-
e -u -1-1-1-1X
noao-
-1-1-1 -1 --o--z
Y)VW> ana 0 - N W - W -
uu4-r I rxxooo

qotained explicitly inasmuch as they can be related to the first

set of three derivatives as shown in the next paragraph. dlnX = ()bbInIn X p c qhc
dlnP,+ (-)b In1n X
b
hc f,Pc
x
The total differentials of Eqs. [25-271 are
dlnX=(-) b In X dlnP.+(-) blnX x
b In Pc Pc/P,hc b In hc PJP,P,
Eliminating d In E in Eq. [30] with Eq. [29] and then compar-
b In X ing the coefficients of the differentials with those of Eq. [28]
p,lp [281
with some rearrangement, the following relations give :

dlne = (-)bbInInP,
e
P,/P,he
dInP,+ (-)b InIn hc
b
e
PJP.P,
X

h u w s 1962
~ 1199 '
 Table 5 Comparison of accurately calculated characteristic
H z ( g ) + 1/303 ( g )
( h c = 6 2 9I C A L I G ,
velocity data for stoichiometric H,-0, with data extrapolated from
500 P c = I O O P/S
.A
.I-)
reference point

h, = 0 cal/g h, = -190.6cal/g h, = 629.1 cal/g

P,, Accu- Extrapo- Accu- Extrapo- Accu- Extrapo-
psia rate lated rate lated rate lated
1000 7208" 7208 7075 7077 7625 7641
_I
600 7165 7166 7035 7034 7573 7599
< 400- 100 7008 7017 6887 6886 7386 7451
Pc=lOOO P S I A )
a Reference point for extrapolation.

"'H~(2)+1120~(1)
(h,--=- 190.6 C A L / G I
P c = 100 P S I A )

!2 (5(xi) "- x?)

- zi0)(zi [35]
2 i,j dXi3Xj

100
P R E S S U R E RATIO, Pc/P
Fig. 1 Comparison of accurately calculated specific impulse data
for stoichiometric Hz-02or Hz-0, with data extrapolated from the
reference point
(OInX) = ( O-InX) -( blnX
b In h, ..pC b In hc P J P , P ~ 3 In ( P c / P )
Second Derivative of Impulse With Respect to Area Ratio
An approximate second derivative of impulse with respect
to area ratio for constant P. and h, can be obtained by dif-
ferentiation of [33], assuming that [ b In e/b In (PC/P)]#is
constant. This is a very good approximation for pressure
ratios greater than 10. With this assumption, there follows
Partial Derivatives for Extrapolation
The first and second partial derivatives given in Tables 1
and 2 can be used for extrapolation purposes by assuming
that the change in any one of the parameters X or functions (o
can be obtained from a truncated Taylor series. Thus, for
any function f, neglecting derivatives higher than the second
When only first derivatives are available, the last term in Eq.
[35] is omitted.
The accuracy obtained with Eq. [35] depends considerably
on the form off and 2,. Those forms whose first derivatives
are the most nearly constant over the range of interest can be
expected to give the most accurately extrapolated values.
) An indication of the desirable forms for f and Z, can be ob-
tained by an examination of some performance data and
I
derivatives. In Table 3, data are given for stoichiometric
H,(g)-O,(g) for P, = 1000 psia assuming equilibrium com-
position during isentropic expansion. Table 3 is the direct
output from an IBM 704 program.
The symbols used in Table 3 for some parameters are some-
what different from those used in the rest of this paper be-
cause the IBM printer does not contain characters such as
lower case letters, Greek letters, subscripts, or superscripts.
The following examples illustrate the differences:
(DLI/DLPC)PC/P = ( b In I / b In Pc)pC/p,hc
(DLCS/DHC)PC/P = (b In c * / b h c ) p c / p , p ,
(DLAR/DLPCP)S = [ b In ~ / dIn (PJP)Is
It can be seen from Table 3 that the derivatives [ bIn e/b In
( P J P ) are very nearly constant over a considerable range
of pressure ratio. I t is to be expected, therefore, that a good
form of Eq. [35] for extrapolating area ratios to other pressure
ratios is
In contrast, the large differences in the values for [b In I /
b In ( P J P ) ] ,indicate that using In I forfand ln(P,/P) for x
in Eq. [35] would not give particularly good results. A more

Table 4 Comparison of accurately calculated area ratio data for stoichiometric H2-02with data extrapolated from reference point

h, = 0.0 cal/g h, = -190.6 cal/g h, = 629.1 cal/g

P,/P P, = 1OOOpsia P , = 100 psia P , = 100 psia
Accurate ExtraDolated Accurate Extrauolated Accurate Extrapolated
10.0 2.440 2.228 2.468 2.289 2.466 2.337
40.83 7.009 6.941 7.151 7.130 7.169 7.279
68.05 10.50 10.49 10.75 10.77 10.81 11 .oo
100.0 14.31" 14.31 14.69 14.70 14.81 15.01
300.0 35.05 34.75 36.28 35.70 37.02 36.45
400.0 44.39 43.84 46.04 45.04 47.22 45.98
600.0 61.92 60.83 64.44 62.49 66.66 63.80
800.0 78.38 76.75 81.77 78.83 85.21 80.49
1000.0 94.08 91.90 98.31 94.40 103.2 96.38

a Reference point for extrapolation.

1200 ARS JOURNAL

nearIy constant set of derivatives can be obtained in this case
by multiplying the logarithmic derivatives by Z, i.e.

A still more nearly constant set of derivatives are

For specific impulse, therefore, a good form of Eq. [35] for

extrapolating to other pressure ratios is

A different situation occurs for temperature. In Table 3

the derivative [b In T / b In ( P J P ) ] ,is fairly constant over
part of the range of interest (from P , / P = 1 to about loo),
whereas another derivative form [bT/b In (PC/P)l8is more
nearly constant over the remainder of the pressure ratio in-
terval. A further complication arises because the derivative
of temperature or logarithm of temperature with respect to
pressure ratio has a maximum. Extrapolation may, there-
fore, give very poor results no matter which derivative form
is used. For most chemical systems, the derivatives [bIn T/b
In (PJP)]. are more nearly constant as a function of pressure
ratio than are the derivatives [bT/bIn (PC/P)l8.

Numerical Examples of Extrapolation

Specific Impulse
The use of the first and second derivatives permits the ex-
trapolation of specific impulse data with considerable ac-
curacy. This is illustrated by two numerical examples.
Both examples start with the d a h in Table 3 for a pressure
ratio of 100.

Example I
Extrapolate Z from P c / P = 100 to P,/P = lo00 (A In
( P e / P ) = 2.30259).
From Table 3, Z = 382.0, [bIn Z/b In (PJP)]. = 0.08396,
and y = 1.1215. From Eq. [ 241, [bZ2/bIn ( P J P ) ] .= 24,504,
whereas from the equation in Table 2, [b2(P)/b(ln P,/P)Z], =
-2654.6. From Eq. [35]
1' = (382.0)' 4-24,504 (2.30259) + %(-2654.6)(2.30259)'
Z2 = 195309
z = 441.9
This is an excellent agreement a i t h the accurately cdcu-
lated value of 442.2 given in Table 3.

Example 2
Extrapolate Z from P,/P = 100 to lo00 (A In (PJP) =
2.30259), from P , = 1000 to 100 psia (A In P , = -2.30259),
from h, = 0 to -190.6 cal/g (from gaseous to liquid pro-
pellant).
Using the data in Table 3 for combustion chamber condi-
tions and for a pressure ratio of 100, the three first derivatives
of Z* from Eq. [24] are
(bT2/bIn P s ) P J P , h , 2600.4
@Z2/dhJ P,/P.P = 28,648
[bZ2/bIn (PJP)], = 24,504
AUGUST1962 1201
 Numerical values for all the quantities that appear in the equations derived from the data of Table 3 at pressure ratios
expressions for the second derivatives in Table 2 are given in of 100 and 1000. The functional forms for I and c which gave
Table 3 with the following three exc,eptions: the best results as functions of the logarithm of P,/P are Z 2

rs) hc
= 0.015198 (see footnote, Table 2 )
and In e. Both the linear and logarithmic forms gave es-
sentially the same results for T and M .
The interpolated results are compared in Table 6 with the
accurately calculated values of Table 3. Included in Table 6
= 0.041197 (see footnote, Table 2) are values of CP and I,,,. Values of CF were calculated from
(=)he
c * and the interpolated values of I by means of Eq. [22], using
consistent units. Values of Ivacwere calculated from c*
= 0.007821 (see Table I) and the interpolated values of I and E by means of an alternate
form of Eq. [19]:
The six second derivatives of I 2 are (with subscripts omitted
for brevity) I,,, = I i-C*€/(P,/P) 1411

b2(12)/b(In P,) = - 182.69 again using consistent units. As may be seen from Table 6,
the interpolated results are in excellent agreement with the
b2(Z2)/b
In P,bh, = 1.5610 accurate results.
In the case of specific impulse, a quintic equation can be ob-
In P,b In ( P J P ) = 162.67
bZ(Z*)/b tained from data a t two pressure ratios, inasmuch as both first
by1 2) /bh,2 = -0.0036205 and second derivatives are available. Specific impulse
values interpolated over the entire range of pressure ratios
In (PJP)
b2(12)/bh,b = 5.3690 from 10 to 10,000 from a quintic formed from data at these
two points were within 0.1 Ib-sec/lb of accurately calculated
b 2 ( I Z ) / d[In (Pe/P)I’ = -2654.6 values.
Using the first and second derivatives in Eq. [35] gives Data are often desired at assigned area ratios. By use of
derivatives obtained from Eq. [33], accurately interpolated
1’ = (382.0)’ +
2600.4(-2.30259) +
28,648(-0.1906) + parameters corresponding to assigned area ratios can be ob-
24,504(2.30259) f $( -182.69)( -2.30259)’ + tained in a manner similar to that just discussed for assigned
~(-0.0036205)(-190.6)2 +
$(-2654.6)(2.30259)2 pressure ratio interpolation.
+ 1.5610(-2.30259)(-190.6) 4-162.67 X
Nomenclature
(-2.30259)(2.30259) +
5.3690(- 190.6)(2.30259)
A , B, C , 1) = coefficients,Eq. [401
Iz = 180,778 A /w = area per unit mase flow rate
C* = characteristic velocity
Z = 425.2 CP = heat capacity at constant pressure per unit mass
CP = thrust coefficient
This is in very good agreement with an accurately calculated h = enthalpy per unit niasa
value of 426.5. I = specific impulse, Eq. [I71
Some additional numerical comparisons of accurate data I\%L = specific impulse in vacuum, Eq. [21 I
with data extrapolated for various changes In P,, h,, and ki, k2 = constants
P,/P are given in Fig. 1 for specific impulse, in Table 4 for ’If = molecular weight
area ratio, and in Table 5 for characteristic velocity. The O/F = oxidant to fuel weight ratio
extrapolated data are all obtained from one accurately cal- P = pressure
culated reference point corresponding t o P, = 1000 psia, h, = PC/P = pressure ratio
0, and P c / P = 100. The derivative forms used were first and R = universal gas constant, 1.98726 cal/mole-”K
S = entropy per unit mass
second derivatives of 1 2 , first derivatives of In e, and first 1‘ = absolute temperature
derivatives of c*. It is apparent that the extrapolated values U = velocity
are in excellent agreement with the accurately calculated 0 = specific volume
values for a considerable range of extrapolation. X,Y? = any variable
a$, e,$ = any thermodynamic variable
e = area ratio
Partial Derivatives for Interpolation x = process parameter
P = density
Derivatives can be used to increase the accuracy of inter- Y = isentropic exponent ( b In P / b In p ) $
polation in a specified range. This is because each derivative
is approximately equivalent t o having an additional point Subscripts
in the specified interval. For example, if only functions are
known at two points, only linear interpolation is possible. C = combustion chamber
0 = initial coordinate of a finite process
However, if the first derivatives of these functions are also t = throat
known a t the two points, cubic interpolation is possible.
With second derivatives also known, quintic interpolation is Superscript
possible. Thus, for example, the following equation may be 0 = reference condition
obtained:
References
1 Gordon, S. and Huff, V. N.. “Theoretical performance of liquid hy-
drazine and liquid fluorine a s a rocket propellant,” NACA R M E53E12
(1953).
where the coefficients A , B, C, and D are determined by the 2 Gordon, L. J., “Uncertainty in a calculated specific impulse due to a n
solution of four simultaneous equations involving the values uncertainty in the heat of formation of the propellant,” ARS J. 30, 115-116
(1960).
of Z2and first derivatives of P, each at two pressure ratios. 3 Bridginan. P. W . , Condensed Colleclaon of ThernodVnanLic Formula
To illustrate the accuracy of interpolation which can be ob- (Harvard University Press, Cambridge. Mass.. 1925).
tained by this technique, values were interpolated for T, IN,c, 4 Gordon, S., Zeleznik, F. J., and Huff, V. N., “ A general method for
automatic computatians of equilibrium compositions and theoretical rocket
and I a t pressure ratios of 300, 400, 600, and 800 using cubic performance of propellanta,” NASA T N D-132 (October 1959).

I202 ARS JOURNAL