www.rasaero.

com

G,3 DEPARTURES FROM IDEAL PERFORMANCE

mentioned. This is logical since flight performance depends directly on
I,D, and only indirectly on 8 .
Consideration of the thermal efficiency leads to the idea of a heat
balance for the rocket motor. The chemical energy introduced with the
injected propellants is distributed in four directions: (1) e small part
(up to 16 per o a t ) remaina unconverted due to incomplete reaction; (2)
about 1 per cent, more or less, of the heat reaction is traneferred to the
motor walls where it may be loat unless the motor is cooled by the liquid
propellant (regenerative cooling); (3) the kinetic energy of the jet com-
prises a large part, from one third to two thirds, of the heat of reaction;
and (4) &ally, the remsjnder of the reaction heat is carried away in the
hot jet as thermal energy.
Typical values of the performance parameters of rocket motors in
the moderate and high performance olassea, respectively, are indicated
in Table Q,2. It is notable that the flame temperatures range from 2000
Table U,#. Typioal perfomnoe ohmaoteriatice.

Performmoe Ordinary High

pmmeter range range

to 6000eK, and that dissociation is prominent in most of the range. The

effects of dissociation constitute the greateat source of inaccuracy in the
rtpplication of the idea analysis just presented to actual rocket proowat.
The modification of the ideal a n a l y l to handle dissociation will be
discussed in Art. 4. Modi5crttions due to other departuresl from the ideal
conditions of this analysis are discuaaed in the following article.

6,3. Departures from Ideal Performanme. The preceding ideal

performance analyais requires correction to take c@reof the following
actual conditions:
a. Conical divergence of the exhaust jet. .
b. Surface friction and flow disturbances in the exhaunt norale.
c. Constriction of the exit area due to boundary layer build-up.
d. Jet detachment.
e. Heat loss from the hot gas to the cold motor walls,
f. Suspended liquid or solid particles in the exhauet jet.
g. Pressure drop in the combustion chamber due to heat releaee.

www.rasaero.com
 C - THE LIQUID PROPELLANT ROCKET ENGINE
Consideration of the effects of temperature dependence of the apeoifio
heats and of chemical reaction during the expansion in the noaale will be
postponed until Art. 4.
The effect of ezhuust noaala divergenee. A loss in thrust occurs as a
result of the divergence of the exhaust jet a# it leaves the noaale, in oom-
parison with the ideal cam of parallel flow. The loss may be viewed as the
deorease in the axial component of momentum due to the outward
inclination of the streamlines. I n m u c h as the loerr is usually only
a few per cent, a fairly approximate theory is adequate to determine its
magnitude.
section

If the exit-to-throat area ratio r is sulfidently large, say at lemt 3,

and if the half angle a of the exit cone is sufficiently mall, say not more
than 20•‹, it is reasonable to assume a radial flow pattern in the exha&
oone with concentric spherical caps for isobaric surfaces. The exit cone
defines the location of fhe effe~tiveoenter for this radiaJ flow pattern. The
axis of the noezle corresponds to 0 = 0, and the surface of the exit oone
corresponds to 0 = a (Fig. C,3a).
To calculate the thrust, divide the exit flow area into ,differential
+
aanular areas between 0 and 9 dB. The axial component of the thrust
of thia element of flow is
dF - p,V.(* sin 8 * rd9)(Va008 0 ) + (p, - p,)(% .
sin g rag) g,
Integrating with respect to 8 from 0 to a, F becomes

In this formula, Vet p, A: come out in the analyais to be the

velocity, premure, and area of the 8pherioal d8obaric cap d the exit,
and not the planar area at the exit. The quantity in braoketa is the
t h t of a aero divergence motor whose exit area equals A:. However, for
angle a, the planar exit area A. is so close to A: that the t h w t
of the sera divergence motor would be almost unaffected if A: were
rephoed by A , and Veand P, were.interpreted to apply to A. instead of
 G,3 . DEPARTURES FROM IDEAL PERRORMANCE
A:. Therefore, if Po is the thrust of a zero divergence motor having an
exit area ratio s equal to the planar exit area ratio of the a divergence
motor,
F, = -0, +
where X = j ( l cos a) (3-2)
For u = 10•‹,h = 0.992, and for u = 20•‹, h = 0.970. Consequently,
the correction is a small one, and the preceding bplified model for the
flow field is justiied. To measure the loss accurately in order to check
the theory requires experiments of high precision. Some experimental
reaults have been reported, which show that the formula (Eq. 3-2) holds
reasonably well from u = loQto u = 30'. For smaller angles, the nozales
are so long that frictional losses dominate the results, and for angles in
excess of 40' there is clear evidence of flow separation.
The most commonly employed divergence angle in practice is about
15". This produces e loss of only 1.7 per cent, yet the nozzle is not so long
as to be excessive in either weight or length. It is possible to design an
exit section with a curved contour eo &e to produce a pardel jet and
thereby recover thia divergence loss, but the pedtiea of complexity
and cost must then be accepted. For a design to produce the utmwt in
performance, e.g., a satellite launcher, thia is worthwhile.
The e&$ects of &id frict6on amd fiow loaeea. It is not possible to calcu-
late reliably the thrust loss due to surface friction, because so little is
known of the nature of the boundary layer under the conditions of prea-
sure gradient and curvature that in the nozule, Abo, the flow field
in a practical nozzle is likely to exhibit strong pressure waves originating
in the sharply curved region near the throat. The rooket designer ordi-
narily cannot eliminate such aharp aurvature, a8 can the supersonic wind
tunnel designer, beoause rocket nozzles must be made cheaply [hence
simply) and must be compact. Consequently, the flow disturbances are
accepted as long as the loss in thrust is small. Fortunately, it Beerne that
drastically poor de&n is necessary to cause a thrust low greater than,
say 10 per cant, due to bad flow, snd if the designer merely uses hie French
CUNeB with good technique, the loss can be cut to about 1or 2 per oent.
For example, if the nozzle contour consists of a converging oone of
aOOhalf mgle, a diverging oone of 15' half &@el and a throat section
with a longitudinal radius of cwature equal to twice the radius of c w a -
ture of the throat section, the loss will ordinarily not exceed 3 per cent
and will probably be nearer to 1per oent if there are no sharp cornera.
In test results, thia type of 166s is reported in the form of a di~charge
ooefficient Cd that normally lie between 0.87 and 0.99.
(Cn),.t = C&[(CP)-OIIW ($8)
Conetriction of th8 6 x l area by the bowday, layer. The effect of the
boundary layer in constricting the flow passage of the nozzle can be con-
( 466 )
 O - T H E LIQUID PROPELLANT ROCKET ENGINE
sidered in two places, at the throat and at the exit. At the throat, a
boundary layer would reduce the mass flow ch; at the exit, it would reduce
the effective expansion by reducing e. It turns out that the first effect
is negligible and the aecond one is small.
Observations of boundary layer growth in supersonic wind tunnel#
indicate that, in the Reynolds number range from 6 X los to 6 X loo,
and in the Mach number range from 1.6 to 2.6, the boundary layer is
turbulent and the displacement thickness grow0 in proportion to the
distance from the throat. This Reynolds number is baaed on the velocity
and kinematic viscosity at the particular station and the distance from
the throat to this station. Although the strong cooling of the boundary
layer and the much larger pressure gradient in the rocket noazle would
suggest caution in applying wind tunnel results to rocket norslea, there
are some observations in cold flow through rocket-type nozalea whioh
indicate similar behavior. (This conclusion will be utilized in Art. 6
in the discuwion of convective heat transfer in rocket nozzlee.)
6* (displacement thicknew) E 0.WL (distance from throat) (3-4)
With thie relation it turns out that the exit area congtriction for a
16' half angle nozzle is only 3 per cent. Inspection of Fig. GI2e and G,2f
showa that the effect of a 3 per cent area reduction can be neglected
if the nozzle area ratio is near the optimum. For narrower exit cones
and for off-design norales, it may be necessary to allow for this effect.
This is another instance of the remarkable insensitivity of rocket
nozzle performance to flow disturbancart that are of great importance in
wind tunnel nozzle design.
Jet detuchment or fow aeparatim. It is not always poasible to provide
a nossle with the o p t i i area ratio for each condition of ambient
pressure. In particular, the exhaust nos& of a rocket airplane that must
fly at altitudes from sea level to perhapn 60,000 feet, or of a guided missile
that darts at sea level and reaches 100,000 at burnout, generally operates
in a very much overexpanded condition at the lower altitudes. The ques-
tion arises whether the flow in the divergent section remains isentropic.
The podbility of a stationary shock wave perpendicular to the flow
axis haa been recognized for a long time, .and the early literature on
rocket performance indicates that this ie to be expected. However,
although such normal shocks have been observed in supersonic flow
channels with small angles of divergence, normal ahocks have not been
found at all in rocket nozsles. Instead,% is found that when the back
pressure (pressure of the surrounding atmosphere) exceeds the isentropic
exit preasure by a large enough amount, the flow separatee symmetriody
from the nozzle wall, and a complioated pattern of oblique shock waves
appears in the stream beyond the point of separation (Plate U,LI).
The fundamental fluid mechanics of this type of separation phenome-
( 466 1
 .
0,3 DEPARTURES FROM IDEAL PERFORMANCE
non i s discussed in IVIB. In brief, the region of importmoe is the place
near the wall where the shock wave penetrates the boundary layer. Ilere
the sharp adverse pressure gradient connected with the shock thickens
the boundary layer and leads to deflection of the flow, while the pressure
rise across the shock is supported by the momentum transfer from the

angle
Obllque
Axis of~ &- ~ c -k w o v e
nozzle . / u atm;spheric pressure
-
maspheric pressure
Region of
undisturbed gas'
section ' area I section
Area of
detachment

A I Ath
Fig. QJb. Typioal ~Cahiopresaure distribution in &ast nozzle.

supersonic &ream to the "stagnant1' fluid in the flow corner. Conse-

quently the strength of the shack and the separation location depend on
the state of the boundary layer (laminar or turbulent) and particularly
on its development dong the nomle wall prior to separation. For the
purpose of the reader interested in rockets, the subject oan be presented
moet usefully in terms of the experimental results and the empirical
correlations.
The static presaure distribution in the nos& is shown in ESg. 0,3b.
The pressure follows the isentropic expansion ourve until the separation
 G . THE LIQUID PROPELLANT ROCKET ENGINE
point is reaahed. At thie point, the pressure rises steeply to almost the
level of the ambient pressure at the exit, indicating a rather clean break-
away of the flow. From the standpoint of thrust, the noznle may well be
cut off at the separation station, since the internal and external pressures
are very neerly in balance beyond this point. Experimentally, it makes
little difference if a longer nozzle is used, once separation has occurred.
The extra length of nozzle does not affect the location of separation and
ha8 no appreciable effect on the thrust. (There is a small reduction
in thrust due to the extra length of cone, because the internal pressure
is slightly less than the external pressure, but t h i ~is usually neglected
in performance estimates.)
It is obvious that the complicated flow S ' & N C ~ Udownstream
~~ of
separation is of no consequence in the computation of thrust. The thrust
of a rocket motor with separation in the nozzle is calculated on the
assumption that the noszle area ratio is not that corresponding to the
actual exit but rather that of the separation station. Therefore it is
necessary to be able to predict the location of separation.
In tests with nitric aoid-mihe rocket motors employing noszles with
half anglae u of lo0, 16', 20•‹,and 30•‹,with actual area ratioa a of 10 and
20, and with pressure ratios p,/p. from 11 to 26, it was found that

Almost identical results were obtained at all mixture ratios, and in faot,
the resultn obtained with the nitrogen flow channel previously mentioned
were also nearly the same. Consequently, as a general rule for estimating
the performance of overexpanded nozzles in the absence of speoifio test
data, the relation (Eq. 3-5) may be used.
Heat lose porn the ooduelim gas. The transfer of heat from the
combustion gas to the relatively oold motor walls, by convection and by
radiation, a h t s the performance of the motor. Two cases of heat trans-
fer are of practical interest: (a) the heat traderred is completely removed
from the thermodynamic process, ss in a water-cooled motor; and (b)
the heat is reintroduced into the process aa in a motor cooled by the
propellants (regenerative cooling). These two processes, together with
the adiabatic one, are ahown in the A, s diagrams of Fig. G,30.
In diagram 1,the caw of no heat loss, the working fluid (product gas)
is injected into the chamber oold at A, is heated adiabatically at constant
pressure by combustion to point B, and ia expanded kentropically to
ambieat pressure at C. The enthalpy drop from B to C is transformed to
the kinetio energy of the jet (see Eq. 2-10). In diagram 2, the case of
heat lost to the cooling water, it is aasumed in order to eimplify the calcu-
latiom that the heat removal is concentrated at two atationa, the hot
end of the chamber and the throat of the exhaust nozzle. (Thia is reason-
( 45s )
 .
G,3 DEPARTURES FROM IDEAL PERFORMANCE
able on the basis of experimental heat t r a d e r distributions.) The procesa
is then as follows: adiabahio combustion from A to B, oonatant preesvre
heat loss from B to D, iaentropio expansion from D to 1,constant pres-
sure heat loss from E to F, and isentropio expansion from F to (3. The
regenerative process, diagram 8,Woes the following path: conetent pres-
sure heating from A to J, adiabatic combustion from J to K, heat loss
from K to L, expanaion from L to M Iheat loss from M to N,and exp~n-
&on from N to 0.

The exhauat velocities for the three diagrams are:

The heat of wmbustion: Aha 5 h~ - Q = hx - h~ (3-7)

In owes 2 and 8, the heat tran~ferredper unit mass is given by:

For a perfect-gaa working fluid with oonetant apeoifio heat, the ratio
of O e exhaust velocity of the aooled mofor to thaa of the ideal motor
can be given explicit form, in eaoh oase, in terms of the heat of combustion
and the preasure ratio. In general, to oaloulate the oorreotion with pre-
&on. it ia neoesaary to refer to aotual h,8 computations on diagrams that
incordorate the te&peraturedependent speoifio heata and the effeots of
diaaooiation. It ia immediately evident from the diagrams that the water-
oooled motor produoepl & lower speoifio impulee than the ided one. In prao-
-
tioe, the approximate formula: V,to LY V , ( i ) [ ( ~ hAqa)/~ha]iis applied
( 469 )
 G . THE LIQUID PROPELLANT ROCKET ENGINE
to relate the two results. For mall test motors, this can amount to w
much as 5 per cent. It is also evident that the specific impulse of a regen-
eratively cooled motor is never less than that of the ideal one, and can
-
ezosed the ideal performance if the heat transfer h~ h~ is large in
comparison with hg - h ~ In . practice, this affect ia too small to be
measured.
8uepmded liquid or solid particle@in the exhaust jet. Condensed phases
can appear in the exhaust jet if elements are present in the reactants that
form refractory products. For example, fud compounds based on the
light metals Li, Be, All and Mg produce oxides LipO, BeO, Alsos and
MgO when reacted with oxygen. These oxides have normal vaporiration
temperatures of 2250, 3800, > 1700, and >280OoC, reapeotively, and
therefore these vapora would condense to droplete or solid pmticles
whenever the temperature drops below these levels, either in the com-
bustion or in the expansion process. dmilarly, carbon (sublimation point
>4200•‹C) which ~m form in large amounts in the decomposition of
acetylenio monopropellants, and KC1 (sublimation point, 1500•‹C)which
forms in solid propellant combustion where potassium perchlorate is
used as the oxidirer, preaent the problem of flow with oondensed phases.
The occurrence of condensation leads to a reduction in specific
impulee for two reaeons. The portion of the working fluid that condenses
cannot perform any expansion work and therefore cannot contribute
to the acceleration of the jet; also, the heat in the condensed phase is
partly ejected with the jet and not transformed to kinetic energy,
because of the low rate of heat transfer from the hot particles to the
surrounding gas. The rate of heat transfer dependa on the particle sire,
which in turn depends on whether condensation occurs in the chamber
or at aome position in the nossle. Average .partiole sir% have been
observed in the range from 1 0 to ~ 50~.
To analym the problem in the moat exact way, it ia necessary to
know at what place in the expansion prooess nucleation occurs, the rate of
growth of the condensed nuclei, the rate of heat transfer from the hot
particles to the cooler gas stream, and the velocity lag between the eus-
pended particles and the gas stream. For practical purposes, it is usually
adequate to simplify the analysis by assuming that the velooity lag is
sero. Concerning heat trander, two extreme assumptions are possible: (1)
the temperature of the condensed phase remains the same throughout
the expansion prooesa as that of the gas where oondensation took place,
that is, rero h e d transfer; and (2) the oondenaed phase temperature
follows exactly the temperature hiatory of the gas, that is, perfect heat
exchange. The first assumption would apply to very large partioles; the
second assumption would apply to very mall particles.
The equation of state of the mixture of gas and suspended particlea,
and the density of the mixture p, are debable in terms of the mean
 0,3 . DEPARTURES FROM IDEAL PERFORMANCE
molecular weight % of the gas, the molecular weight of the solid a,
the mean molecular weight of the mixture I ,
and the respective numbers
of moles of gas and solid, n. and n,,in a standard molar volume UO.

For adiabatic, frictionless flow, dh - ( l / p ) d p . For the mixture,

Introducing mole fractions XI and X.,this equation becomes

For adiabatic flow, V I V

velocity lag,
+ dh - 0 , For the mixture, with zero

Eq.3-11 and 3-12 are applicable upstream of the condensation loca-

tion, if X. is set equal to unity, and X. equal to zero. Downstream of this
location, these equations aan be applied if it is mumed that the male
fractions X,and X,are fixed. This is a reasonable assumption since most
of the condensation takes place in a very short region.
A satisfactory solution of the problem is to obtain V. in t e r n of the
pressure ratio p./po. This can be obtained by integrating Eq. 3-11 and
8-12 if the ratio dT,/dT, can be speoified. For oase 1 mentioned above,
the ratio is zero; for case 2, the ratio i a unity; and other cases can be
treated. The most extreme cage is to assume that dT,/dT. is sero and that
oondensation oocura in the oombustion chamber, so that the particles
are at temperature T . at the exit, This leads to:

For a typical propellant gaa compoeition, bmed on hydrocarbon

combustion. the condensation of 20 per cent of the mass in the oombuetion
chamber l e k s to about 10 per cent loss in I,, by formula (Eq.3-13).
The effect can therefore be of eoneiderableimportance for some propellant
ayatems.