Вы находитесь на странице: 1из 71

Proo. Aerospace Sci. Vol. 28, pp.

1-71, 1991

0376-O421/9150.00 + 0.50
1991 Pergamon Press plc

Printed in Great Britain. All fights reserved.

PLUME FLOW AND PLUME IMPINGEMENT


IN SPACE T E C H N O L O G Y
GEORG DETTLEFF

DLR lnstitut.f~r Experimentelle Str6mungsmechanik, Bunsenatrasse 10, D-3400 G6ttingen. F.R.G.


(Received 18 January 1991)

Almtraet--Rocket thrusters of various types are the common means of propulsion in space technology. In
the space vacuum the exhaust gases form a large free jet, called a plume, which can impinge on neighbouring
surfaces. The impingement effects, often unavoidable due to the unrestricted spreading of the gas, are
disturbing forces and unwanted heat load. They can reduce the lifetime of the spacecraft or lead to damages.
The solution of the problem includes the determination of the non-uniform plume flow field (hypersonic
continuum and free molecule flow together with the non-equilibrium transition), the treatment of the nozzle
flow with the characteristic viscosity influence, the assessment of the impinging plume flow and finally the
determination of forces and heat transfer. Various theoretical and experimental investigations in the past
and the present state are presented.

CONTENTS
NOMENCLATURE
INTRODUCTION
THRUSTERS AND PLUME IMPINGEMENT SITUATIONS
PLUME FLOW
NOZZLE FLOW AND PLUME NEAR FIELD
PLUME IMPINGEMENT
5.1. Impinging plume flow characteristics
5.2. Plume impingement forces
5.3. Plume impingement heat transfer
6. INTERACTION OF TWO PLUMES
7. TEST FACILITIES AND EXPERIMENTAL TECHNIQUES
8. SUMMARY AND CONCLUSIONS
ACKNOWLEDGEMENT
REFERENCES
1.
2.
3.
4.
5.

1
3
5
9
25
33
33
37
45
52
56
66
67
67

NOMENCLATURE
A
fm']
A~
Ap.ezp, Ap.med

A~
I'm2]

a
B

[m/see]
[kg/m]

e
ca
cp
c'p

Cm/sec]

c,

[- ]

28:1-A

cv

[Ws/kg.K]

[Ws/kg.K]

c,
d...
d

ffi A,. P*/Po

A,

JPAS

area
plume constant
plume constant determined by experiment
and model calculation,
respectively

dE
e
F
F*

rm]
I'm]
[m]
[m]
[Ws]
[W/m 2]
L-N]
L-N]

f
H
h

[J]
I'W/m2.K'l

l,p

rm/sec]

dp

projected
area
of
sphere
speed of sound
factor
of proportionality defined in
Eq.(2)
thermal velocity
discharge coefficient
pressure coefficient
specific heat at constant pressure
constant used in Eq.
(22)

d~

dx, d2

Kn
1

specific heat at constant volume


shear stress coefficient
elemental. . . . for example dA: elemental
area
diameter
probe diameter
sphere diameter
diameter of molecules
energy
energy flux
(1) force; (2) thrust
thrust produced by
sonic orifice
function
enthaipy
heat transfer coefficient
specific impulse
Knudsen number

G. DETTLEFF
gnp

k
k'
l
IE
M
Ma
Mab
Maw
m

fa
rhid

Nu
Nuds
n

P:
Pr
P
Ph
PPAT

P,
Pt2

P~
q

penetration Knudsen
number
[J/K]
Boltzmann constant
[W/m. K]
coefficient of thermal
conductivity
[m]
length
[m]
distance nozzle throat
to exit
[kg/k-mole] relative molecule mass
Mach number
Mach number on
border streamline
Mach number at wall
[kg]
mass of molecule
[kg/sec]
mass flow
[kg/sec]
mass flow, Eq. (32)
[kg/sec]
total mass flow
Nusselt number
Nusselt number based
on diameter d,
[1/m 3]
number density
breakdown parameter
(Bird), Eq. (25)
onset of breakdown
Prandtl number
[N/m 2]
pressure
[N/m 2]
background pressure
[N/m 2]
pressure indicated by
Patterson probe
(free molecular pressure probe)
[N/m 2]
pressure on impinged
surface
[N/m 2]
Pitot pressure
[N/m 2]
vapour pressure
[W]
heat flow
[N/m 2]
dynamic
pressure
1/ 2 pu 2

0
R
Re

[W/m 2]
[Ws/kg K]

Rep

rf

[m]
[In]

rt

[m]

r'

S
St
S

7",
t
u, o
Ultra, Ulim

[K]
[K]
[see]
[m/see]
[m/see]

v[

[m a]

X
XM
Xs

[m3/sec]
[m/see]
[m]
[ml
[m]

Xt
Xo

[m]
[m]

Y
Yw

[m]
[m]

heat flux
gas constant
Reynolds number
probe Reynolds number
radius
location of freezing
surface
radius of curvature at
nozzle throat
recovery factor
speed ratio
Stanton number
Reynolds
analogy
factor
temperature
recovery temperature
time
velocity
maximum attainable
gas velocity after
expansion
volume of vacuum
chamber
pumping speed
velocity
coordinate
location of Mach disk
location of shock intersection
distance along plate
location of virtual
source point
coordinate
coordinate of nozzle
wall

z
z,h

[m]
[m]

ZN

[m]

coordinate
shock stand-off distance
(1) distance
platenozzle,
(2) distance interaction
plane-nozzle

Greek Symbols

~t

[]

ct'
fl

[rad]

7
6

[m]

e
O
OE
O1i=

[o]
[o]
[o]

Oo

[]

O~

[]

angle of attack (angle


between streamline
and surface)
angle of attack
constant defined in
Eq. (If)
ratio of specific heats
boundary layer thickness
nozzle area ratio
polar angle
nozzle angle at exit
angle
of
border
streamline
angle of streamline
separating isentropic core
and boundary layer expansion region
deflection
angle
veM(Mab)
-- vpu(Ma~)

.9

[]

r
2
2p

[m]
[In]

#
v
v~u
n
p
a

[Ns/m 2]
[I/s]

[kg/m 3]

ae
ap
at
T
~0
f2

[N/m 2]
[]
[m 2]

cant angle of thruster


axis
temporary parameter
mean free path
mean free path at interaction plane
viscosity
collision frequency
Prandtl-Meyer function
3.14
density
accommodation coefficient
thermal accommodation coefficient
accommodation coefficient for normal
momentum transfer
accommodation coefficient for shear stress
shear stress
angle
collision cross section

Indices

If not otherwise defined above, indices have the


following general meaning:
c
continuum flow
CL, cl
on center line
E
at nozzle exit
i
incoming molecules
FM
free molecule flow
mod N
modified Newton theory
n
normal to streamline
R
reflected molecules
ref
reference state/quantity
tr
transition flow
w
wall conditions
x
parallel to streamline
o
stagnation conditions

Plume flow and plume impingement in space technology


o, eft
2

*
c~

effective
stagnation
conditions
conditions behind a norma! shock
at nozzle throat
free stream conditions

Abbreviations
DSMC

Direct
Simulation
Monte Carlo Method
Method of Characteristics

MOC

1. INTRODUCTION
Rocket thrusters are the common means of propulsion in space technology, with different
sizes and propellants. To launch a rocket big engines with several million N thrust may be
necessary, while for an attitude control manoeuvre of a satellite small engines with about
1 N thrust are sufficient.
Firing of a thruster leads inevitably to the formation of a jet flow downstream Of the
nozzle. This flow, called a plume, is relatively slender when a thruster is fired in the lower
atmosphere, but spreads more and more with decrease of the surrounding pressure as in the
upper atmosphere..If the surroundings can be considered a vacuum (pressure < 10- ~ mbar)
a maximum spreading of the plume results.
In many situations in space technology the plume flow causes severe problems: In Fig. 1
a plume is formed during the firing of a satellite attitude control thruster. Due to the
position of the thruster, parts of the satellite body and of the solar arrays are impinged by

plume

attitude control
thruster axis

J
solar arrays

!
-lm

1.34 tn
FIG. 1. Plume formation and impingement during attitude control thruster firing (Orbital Test Satellite).

G. DETTLEFF

the exhaust gases. The impingement effects are forces, heat load, and contamination: Due to
the large size of the solar arrays the forces can lead to considerable undesired torques. Since
the torque vector, which is additional to the thrust torque, is generally unknown, uncontrolled movements occur which can only be compensated by additional propellant
consuming manoeuvres. The heat load, alone or in addition to the solar radiation, can
damage surfaces. Finally the quality of optically sensitive surfaces can be reduced by gas
deposition or by droplets of unburned propellant. It is therefore clear that the impingement
effects lead to a reduction of function and lifetime of the spacecraft.
The direct way to solve the problem would be to put the spacecraft into a vacuum
chamber, fire the thrusters and then correct their positions to avoid or to minimize the
impingement effects. This, however, is very difficult since enormous pumping capabilities
would be necessary to maintain vacuum quality (,~ 10 - 6 mbar) during thruster firing
(a I N-hydrazine thruster has a mass flow of about 0.5" 10- a kg/sec). Another way would be
to solve the proper flow equations and the equations describing the interaction between
flow and surfaces. This, however, turns out to be likewise difficult. In Fig. 2 the basic
features of a plume expanding into vacuum are sketched schematically. The flow starts in
the combustion chamber (stagnation chamber). In the nozzle a restricted expansion takes
place with the establishment of a boundary layer enclosing the is'cntropic core in the near
axis region. The nozzle area ratio ~ = ~

is typically in the order of 50 so that the exit

Mach number is typically about 5. The flow downstream of the exit in the free expansion is
therefore hypersonic. Since the streamlines are divergent, continuum flow is only maintained in the vicinity of the nozzle passing into a free molecular flow further downstream.
According to the different flow types, different types of interaction between flow and
spacecraft surfaces occur. In Fig. 3 continuum flow impingement with the establishment of
a compression shock and the free molecule flow interaction are sketched. The general,
complete investigation of such impingement problems therefore comprises the treatment of
nozzle flow with boundary layer, hypersonic continuum and rarefied flow, expansion and
compression phenomena and gas-surface interaction. In addition chemical reactions and
condensation can take place in the flow and complications arise from difficult geometrical
situations. Some fundamental literature on these subjects is cited in the References.t1-4) By
means of experiments and theoretical work, however, a considerable approach to the
solution of the basic problems and of single projects was possible. It is the intention of this
article to expose these efforts and to work out the ruling parameters and simple formulae to
estimate plume impingement effects.
An early survey on plume flow problems was given in 1964 by Vick et al., ~s} who discussed
among other things the problems of plume impingement. A literature survey concentrating
on simple plume flow and impingement forces descriptions was published in 1980 by
Boettcher and Legge. ~6)Recently Lengrand {7}gave a concise survey on plume impingement
problems at the Rarefied Gas Dynamics Symposium.

FIo. 2. Flow types in a plume expanding into vacuum. Adapted from Ref. (121).

Plume flowand plumeimpingementin space technology

Continuum ftow impin~lement


//!.,cypersonicf
r~~._~nllnuwll
olfw
/

undisturbed"L:tr~,nrnlln)~
/ disturbed__./~..~._J
.........
y

layer

r / i / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /

Surface

Free motecuteflow imlaingement


~ n s i t i o n

)--Wo,ecu,e
\

,,ffu.e

surface
FIG. 3. Continuumand free moleculeflowimpingement.
The determination of mass flux, momentum flux, and energy flux effectively impinging on
a given surface and the local distribution of these fluxes on the surface is necessary to solve
a plume impingement problem. We will therefore mainly treat these aspects of fluid
mechanics and we will consider the methods and results to determine especially the density
and velocity, which are the dominating quantities determining the fluxes in the plume
(Section 3). We will also look in some detail at the main features of nozzle flow (Section 4),
since the nozzle exit flow represents the initial conditions for the free expansion. Then in
Section 5 we deal with the plume impingement forces and heat transfer. Plume impingement
contamination as the third important effect (besides forces and heat load) will not be treated
here. This would not only include the consideration of the fluid mechanical aspects of the
plume but also of the chemical reactions under various firing conditions. Moreover the
impinged surface itself must be considered in detail, since contamination depends on the
specific material, its cleanness or pro-layers of exhaust products and on other factors.
Experimental facilities will be presented in Section 6 together with some measurement
techniques. Plume impingement problems arising during multi-thruster firing are of
increasing importance. The treatment here will deal with the interference characteristics of
two plumes (Section 7). A survey on thrusters, on their positions and other requirements,
and on the resulting impingement situations will be given in the next section, Section 2. The
considerations will be limited to steady plume flow in vacuum and will emphasize thruster
sizes typical for the application in such surroundings, i.e. in general the lower thrust level
range.
2. THRUSTERS AND PLUME IMPINGEMENT SITUATIONS
Before going into the details of plume flow and plume impingement it is worthwhile to
look at the variety of thrusters and propellants and on the different impingement situations.
The requirements for the thrusters depend in the first place on the specification of the
mission. As an example we consider a satellite in the so-called geostationary or 24-hours

G. DETI'LEFF

orbit. (a-l) The mission is characterized by its purpose, payload, dimensions and lifetime.
The requirements for the attitude and orbit control thrusters are moreover determined by
the desired accuracy of position and by the possible disturbances: lateral gravity acceleration leading to east-west drift, which must be compensated for, and varying gravity
influence of the moon and sun, leading to north-south station keeping man0~uvrcs. Further
perturbations are possible due to solar radiation pressure, gas leakage or micrometeorite
impact. Knowing or having estimated all these factors and having defined the thruster
positions, we can determine the different impulse increments, F" At, and finally the total
burn time t and impulse St0F. dr.
The thruster positions for a satellite in a geostationary orbit are sketched in Fig. 4. For
a thruster to produce a rotation of a satellite the position is determined by the requirement
to obtain a maximum torque.
To perform man0~uvres in space, engines with different thrust, specific impulse, system
weight, size of impulse bits, power requirements, reliability and cost are available. An
important selection criterion is the weight of the propulsion system, since it is one of the
factors determining the launch cost. (1~l For this reason the specific impulse, defined as the
attainable thrust per unit mass flux of propdlant

Isp

= - -

F
vh

is an essential characterizing quantity of a thruster.


In Table 1 are listed some common thruster types, propellants and exhaust products. In
the case of bipropeUant thrusters the fuel and oxidant chemically react in the combustion
chamber. The often used combination of hydrazine and dinitrogentetraoxid (N20,)
provides thrust not less than 1ON. (~l The combustion temperature is in the range of
TABLE 1. SOMETHRUSTERTYPES USED IN SPACETECHNOLOGY

Thruster

Main exhaust
gas components

Propellant

Liquid bipropcllant

Monomethyl-hydrazine
MMH (fuel), N20 (oxidant)
N2H 4 (hydrazine)
H202
N 2, CF,, H 2
N 2, CF,, H 2, NH a, He and CO 2, H20
(biowast products)

Liquid monoprol~llant
Cold gas
Rcsistojet

~
I ~
station -~_ ~ _ . . ~ - ' 7
kee#ng ~
Jj~
- - e ~ - - - W - - - ' e~
N-S

attitude

I~

~ ~,

control--",., ~ ' e

.........

~,,.:, .u,,u<,.u,.~

control Wn~l

n~odi~

u ........

Ill

t_

~
( ,
~

rail/yaw
control

'o

c:~

_1

----]-o- - ----Ti- - 7 - -

_ , / _ /
(AI representative 3-axis

satellite

E-W station
Keepln.g, .

repos,t,onlng

offset

H2, N2, NH a
H20, O2

Same as propellant

yaw control,
spin/despin,

~....

.,/, "

H20, N2, CO, H2, CO2

# ~
IP~( ,
~,1X~,, /

E-Wstailo- n~r~
keepina
/
repositloning J,-,
axio is : . I . ~ --~
N-S station

keeping,
precession

<o,,o<

. .

LsPm{

i ~ i ~ _ oe~'n

radials,--._ i

cp0
/

c~
/
,~J

/
~

thrusters
(B) representative spinning
satellite thrusters

FIG. 4. Repr~cntative

thruster locations and functions for three-axis and spin-stabilized satellites. N-S:
north-south, E-W: east-west. Adapted from Ref. (11).

Plume flow and plume impingement in space technology

cold/warm gas IGN2.KcCF/.,Ar,H2.etc)

N ~"

.c"6

hydrogen

~ f ~ " ~

cototytic hydrozine.N2HL,
I

ISp:250t310s~

hipeht/hydrozine
Low thrust N202/MMt-~:~
llsp=2r-JO0s . . ~

~c,
='E
I
1960

(,~

el-

I
1970

mercury ion propulsion


1980
and beyond

FIG. 5. Generations of spacecraft on-orbit control propulsion system applications. Adapted from Ref. (11).

3000 K. t4) Monopropellants like hydrazine (N2H4) are decomposed by a catalyst and
provide thrust down to the milli-Newton range. The combustion temperature is about
1000 K. The thruster efficiency (specific impulse) can be increased if the exhaust products,
before entering the nozzle, are heated electrothermally,tl 2) In cold gas thrusters no chemical
reactions take place. The pressurized gas is let into the stagnation chamber and expands
through the nozzle. If electrothermal energy is available, resistojets can be employed. Such
thrusters are relatively easy to handle. On manned space stations biowaste gases (CO 2,
H20) may be used as a propellant. ~13) For one single manoeuvre with defined thrust and
burning time solid rockets can be used. An example is the orbit injection manoeuvre. The
thrust is typically several hundred to several thousand Newton.
Sackheim e t al. ~ ) have subdivided the on-orbit control propulsion systems into five
generations (Fig. 5). Except for hydrogen peroxide (H202) all generations are in use now
(ion propulsion is out of the scope of this article). Note that the data beyond 1980 in Fig. 5
are an estimate.
An early review of microrocket technology is given by Sutherland and Maes,t~4) with
additional information for example on thrust and total impulse.
Thruster nozzles for space applications have an expansion ratio e > 50, since the ambient
pressure is very low. Therefore undesired flow separation will not occur. The nozzle shape is
in general conical or contoured and in some cases trumpet shaped (Fig. 6).
Plume impingement effects play a role in thruster positioning and propellant selection
besides the criteria mentioned before. In Fig. 7 the orientation of a thruster giving a
maximum torque is sketched (~ = 0). Canting of the thruster axis reduces plume impingement effects on the solar array but also reduces the thrust torque. The proper orientation is
therefore balanced between a minimum of plume impingement effects and a maximum of
attainable torque. This is demonstrated in the example in the lower part of Fig. 7. For 9 = 0,
maximum thrust torque, but also a maximum opposite disturbing torque and maximum
heat load due to plume impingement are obtained. Variation of ~ leads to an effective
torque as shown by the dashed line. Also indicated is the heat load on the solar panel which,
in this example, exceeds a certain critical value for small angles 9. Thus plume impingement
effects prevent the utilization of the maximum torque orientation (9 = 0) of the thruster.
The consequence is an undesired increased fuel consumption.
With respect to plume flow calculations the knowledge of the exhaust gas composition
is necessary, since it determines essentially the gas dynamic behaviour. For cold gas and
resistojet thrusters the composition is identical to the propellant gas. In the case of
monopropellant and bipropellant thrusters, however, chemical reactions take place in the

G. DETTLEFF
a)

7~a~.7

b)

9.8

~
~ 6 ~ 8 . 5 3
c)

7.85

10./,8"

FIG. 6. Typical nozzle shapes used in space technology. (a) Conical (ERNO 0.5 N thruster); (b) contoured (ERNO
2 N thruster); (c) trumpet (adapted from Ref. (60)).

,_ ~

impinged surface:
solar ,pnnet

'-thruster

~'ume axis
,~
canted

heat toad

torque

....... "'.."*...

thrustertorque~

critical vatue

~heattoad at
certain position

effective/
torque
(sumof thruster
torqueand
plumeimp.
torque)

FIG. 7. Diminution of plume impingement effects by canting of the thruster (schematic).

Plume flowand plume impingement in space technology

combustion chamber and also in the nozzle flow. This makes it necessary to determine the
exhaust gas composition either by experimental investigation or by theoretical treatment of
the chemical reactions. The knowledge of the exhaust gas composition is furthermore
necessary to assess contamination effects.
In Table 1 we have also listed the main exhaust gas components. Their composition in
a liquid propellant thruster cannot be considered as fixed, since it depends on various
parameters, for example the firing mode (pulse or continuous), thrust level, and mixture
ratio. The thrust level is either constant or decreases in the course of the mission ('blow
down mode'). The decreasing combustion pressure is accompanied by decreasing density
and also temperature, 15) which may lead to different compositions.
Continuous and pulse mode firing influences the thermal interaction between the reacting
gases and their solid surroundings in different ways and may therefore also influence the
reaction process. Different temperatures in the center of the combustion chamber and near
the chamber wall can lead to locally different gas compositions in the plume. ~xtJ
Another factor determining the details of the exhaust gas composition is the time
available for the chemical reactions. Due to rarefaction and temperature decrease in the
flow the chemical reactions may remain incomplete causing a chemically frozen flow.
A large variety of other manoeuvres in space operation exists, in addition to the attitude
and orbit control manoeuvres of a geostationary satellite sketched above as an example to
demonstrate the relation between plume impingement on one hand and thruster position
and propellant selection on the other hand. Some are depicted in Fig. 8. Orbit injection and
stage separation may be performed with a solid rocket thruster. For each such manoeuvre
an analysis must be made for the propulsion system. Instructive examples can be found in
Refs (17) and (18).
The example discussed in Fig. 7 demonstrated that plume impingement can be diminished by canting of the thruster axis, i.e. the impingement situation is shifted into the outer,
more rarefied region of the plume with less mass flux. During rendezvous and docking,
however, impingement also takes place in the core region of the plume (Fig. 8), where the
mass flux is high and the possible impingement effects can become especially severe.
Deceleration of the chaser is best if the thruster fires along the trajectory. This may be
accepted if the distance to the target is large. During further approach this can no longer be
accepted so that the thruster axis should be canted to prevent too much heat load,
contamination and momentum on the target. During the last phase it is necessary to use
cold gas thrusters to prevent an unacceptable heat transfer.
A particular rendezvous situation was the moon landing during the Apollo mission. The
plume of the decelerating thruster was directed towards the moon surface, where dust could
be raised endangering the safe landing. x9' 2o)
The proper choice of propellant is also important during scientific missions with
(unmanned) landing operations on other planets or on comets similar to the moon landing.
If material samples are to be taken they must not be contaminated by the exhaust gas or
altered by the heat load from the retro-rocket plume.
In this section we have shown that the plume impingement problem can affect each
spacecraft and mission and that in some cases it cannot be avoided but only minimized. The
possible appearance of this problem which can affect reliability and lifetime of a mission
necessitates in almost all projects an analysis of thruster plume flow and resulting
impingement.
Considerable efforts have been made first to give a proper description of the non-uniform
plume flow, on which we will report in the next section. We will do this close to the
chronological order to demonstrate the requirements for increasing refinements.

3. P L U M E F L O W
Early plume flow determinations for impingement calculations date back about thirty
years in the context of the preparation for the moon landing operation. Prior to this specific

10

G. DETTLEFF

a)

.. \\ U

b)

II/

c,..D ~i.

t hr uster
.Axial I"

" ~ 6 rt~

/. 629.6

d)

plume 1
<

199.5

Overlapping region
of plume1 and plume 2

~T

%
~.

(~) (~

0.6 Newton Hydrazin


thrusters (ERNO)
(length dimension : ram)

1/.10

plume

FIG. 8. a-d.

11

Plume flow and plume impingement in space technology

~8m

~(Apple)
(Cat Module)

~ e t r o

rockets

f)
upper stage

lower

stege
shock

-i

....

TARGET I
e

plume boundary

haust nozzles
otot of 12)

b)

3
:\

~ . x

x x ' ~

~,

1.ROCKET EXHAUSTFLOW
2.SURFACESHEAR DISTRIBUTION
3.DUST LAYER EROSION
4.DUST CLOUD,VISIBILITY

exhuust
plume

FIG.8. Examplesof plumeimpingementsituations.(a) Transferorbit; (b) spin-up;(c) attitudecontrol;(d) orbit


control (north-south station keeping), (adapted from Ref. (100)); (e) stage separation ARIANE; (f) stage separation
(adapted from Ref. (89)); (g) rendezvous and docking; (h) moon landing (adapted from Ref. (19)); (i) backpack
manoeuvring (adapted from Ref. (122)).

12

G. DETTLEFF

interest of space technology, the flow field was already the subject of investigation when
interacting with the external flow during rocket flight and affecting the stability/5' 21-2a~
First attempts to determine the plume expansion behaviour in vacuum surroundings
profited from this field of activity and its results.
The complete flow field with the different flow types as sketched in Fig. 2 was not of
interest at that time. The specific impingement problem during the moon landing operation
resulted from the interaction of the more dense exhaust flow with the dust layer of the moon
surface (where rarefied flow in the off-axis region played no role). The boundary layer in the
thruster nozzle was relatively thin (adjustable thrust 5,000 N-30,000 N), so that the nozzle
and plume flow could be determined by the well known Method of Characteristics (MOC)
for inviscid flow. At that time, however, this exact method seemed to be applied only in
exceptional cases for two reasons: cost of time and/or money and severe computational
difficulties when reaching the high hypersonic flow.(24' 22)
Examples of the method of characteristics calculations of plumes are shown in Fig. 9. The
flow properties at the mesh points are known from the calculation. For a qualitative
assessment it is to be noted that increasing mesh size in the flow direction indicates
expansion and decreasing mesh size indicates compression.
In Fig. 9a the characteristics net of a plume expanding into vacuum is shown. The
computational difficulties mentioned above arise from the too large mesh sizes with
resulting uncertainties, here demonstrated in the region with large flow angles to the jet axis.
With the initial conditions selected the region near the border streamline could not be
calculated. In Fig. 9b the streamline pattern corresponding to Fig. 9a is drawn, and in
Fig. 9c the characteristics net of the same plume expanding in low pressure surroundings is
shown with the typical lateral compression shock (an explanation will be given in Section 6).
These calculations have been performed with a code t26~ adopted from Vick et al. (27~
Fortunately it is possible to describe the complicated plume flow into the vacuum to
a sufficiently exact degree by analytical methods. In many cases of the early impingement
problems only the continuum far field was of interest. It is characterized by high Math
numbers and gas velocity and therefore almost straight streamlines, which seem to originate
in the nozzle exit (Fig. 10). The identification of such a source point is possible, when the
dimensions of the flow field are much larger than the dimensions of the nozzle exit. The
outer shape of the axisymmetric plume, produced by the border streamlines, is that of a
cone (also see Fig. 1). In such a flow the density p decreases according to
1

p ~ r~

(1)

where r is the distance along the straight streamline. For fixed r the flow has a maximum
density on the plume axis. The lateral density distribution is therefore described by a
function f(), which has a maximum for = 0 and is zero or has a minimum value for
= am (border streamline), and the plume flow density distribution p(r, 0 ) is approximately determined by

p(r, 0 ) = Brl-~f().

(2)

This basic form of the density function for a plume has been fitted with various expressions
for B and f(). The different expressions arise from the search for the essential flow
parameters, which determine these quantities. These parameters stem either from theoretical reasoning, from experimental observation, or from fitting proeexlures with results of the
method of characteristics calculation. For an inviscid flow of a pure perfect gas they can be
subdivided into three groups:
The stagnation conditions: pressure po, density Po, temperature To;
The gas properties: ratio of specific heats 7, the relative mass M of a molecule;
The nozzle geometry: expansion ratio 8 = -~,~, nozzle length IE or exit angle E.

Plume flow and plume impingement in space technology

a)

flow

13

direction
jet axis

border
streamli,

b)

flow direction

jet axis

c)

flow direction

FIG. 9. (a) Characteristics net of a plume expanding into vacuum, p~ = 0. Radial source flow in the nozzle
assumed, E = 10.48; 7 = 1.24. (b) Streamline pattern derived from the characteristics net in Fig. 9a. (c)
Characteristics net of the plume from Fig. 9a, here expanding into low pressure surroundings, PE/P~ = 20.

G. DETTLEFF

14

virtuntsourcepaint

nozzle

I;m

OXi$

--

~"b~

streomtine

voctJum

~,.l-'~.

FIG. 10. Plume far field modelling.

Under these conditions the density p(r, O) at a fixed position (r, O) is directly proportional
to Po, and the border streamline angle ,m is approximated by the sum of the nozzle angle
n and the deflection angle ~o = VpM(Mab)--vpu(MaE). vpM is the Prandfl-Meyer function
and Mab, Man are the M a t h numbers on the border streamline and at the nozzle exit,
respectively. The determination of ]in, is approximate since the Prandtl-Meyer function is
derived for two-dimensional flow.
Sibulkin and Gallaher (24) obtained the expression
B = (0.4rt/~koo)

(3)

where ~b~ois a cone solid angle related to ~o by


,= = 2n(1 - c o s = ) .
In this simple form B is only a function of 0~o, which in turn is given by Y and 5.
Roberts and South (2s) deny that the angle oo is a significant parameter (since at the
border streamline the density is zero) and introduce a hypersonic parameter
x,, = ~(y -- 1)" M a 2

(4)

so that their density function becomes


/

_p -

(cos)"'.

PE

2 ~rE J

The parameter x4 is interpreted as


x4 _ ~(~ - 1)Ma~ u~/2
( exhaust gas kinetic energy
2
2
= ~
= \exhaust gas internal energy JE

and x4 is also, as 0 = , a function of T and 5.


Albini (29) specifies an angular M a t h number distribution
FMa( O - - O )
lim,_,= L Ma(O)

lr
cos,/2(=
'~
\ 2 0 l i m)

which can be transformed into an angular density function f():


- -

PcL

cos,

- 1

(5)

"

Hill and Draper (25) derive from an exact method of characteristics solution an approximation
f(O) = exp[ - ~:6(1 -- COS)2 ]
(6)

Plume flow and plume impingement in space technology

15

where ~c6 is a quantity which depends on the vacuum thrust coefficient and on the Crocco
u
number - -

Ulim

Boynton (3) finds near the leading edge of the expansion

P--- ~ r - 2 ( O -- Olim) " - 1


Po

(7)

which is reproduced by the approximate formula


2
p

r_2[

-po ~

cos ~ ~

17 -1 "

_J

(8)

Boynton's (3) and Albini's (zg) forms differ by the factor 2 in the exponent of the cosine
function. The plot of these two forms in Fig. 11 shows that considerable differences of the
density distribution occur for larger O. In this off-axis region the resulting differences of the
density can easily reach a factor of 10. The impingement effects are roughly proportional to
the mass flux p u and their determination would therefore have the same uncertainty.
The validity of the expression
O
f(O) = cos ~21 (~- O---l~mim)

(9)

has been checked experimentally. The realization ofisentropic flow is best achieved in a free

2 (~ O~im)

jet expansion from a sonic orifice. In this case the function f(O)= cos~-I

describes exactly the experimental profiles at various distances r (Refs (31) and (32)): This
confirms the concept of straight streamlines resulting in similar profiles f(O) independent of
the distance r at least in this case with a sonic orifice (in none of the expressions for f(O)
presented above does r appear as a parameter). The coincidence of experimental data and
the approximate form off(Q) was even suitable to confirm the ratio of specific heats 7 from a
set of experimental profile data. (31) For the isentropic core expansion flow with nitrogen as
test gas from an original thruster nozzle the agreement for f(O) was also good.
In the considerations presented so far the nozzle boundary layer and its expansion into
vacuum (Fig. 2) plays no role. The appearance of plume impingement problems in the
boundary layer expansion region (example: Fig. 7) was the occasion to extend the simple
isentropic treatment. The introduction of viscosity changes first of all the quality of the
nozzle flow. It must now fulfil the boundary condition M a = 0 (with p ~ 0, p ~ 0,
T = T , ( x ) at the nozzle wall). As a consequence of the reduced Mach number at the nozzle
lip the deflection angle of the border streamline becomes much larger than in a plume with a
pure isentropic nozzle flow, and the complete density distribution in the boundary layer

10-1

lO-Z

10-3
.1 .2 .3 .~ .5 .6 .?

e/e tim

FIG. 11. Angular dependence of density in undisturbed flow far from the nozzle exit. Adapted from Ref. (30).

16

G. DETrLEFF
10

lO-t
q/qcc
lO-Z
e/eeL
10-3

10-t
lO-S

e/eCL

.2 .t, .6 B 101.211.161829

8 { radians )

P/PcL,

q/qcL

FIG. 12. Limiting (self-similar) angular distribution of normalized density


dynamic pressure
and kinetic energy flux
for a plume. The dotted line shows distribution without boundary layer. Adapted
from Rd. (33).

e/eeL

expansion region must deviate from the functions f() presented above. The gas velocity in
the pure isentropic flow reaches its maximum value
U l i m ~-"

~ 1

R TO

(10)

everywhere in the far field; now in the boundary layer expansion region it is expected to be
lower.
Boynton t33J presented numerical plume flow field calculations from viscid nozzle flow. He
finds self-similar angular distributions of the density p, dynamic pressure q = pu2 and
kinetic energy flux e = pu3 in the far field (Fig. 12). The comparison with the results of the
calculation from inviscid nozzle flow reveals the enormous influence of the gas viscosity on
the expansion. The method itself does not allow treatment of the subsonic boundary layer
which can have a considerable thickness. The Mach number at the nozzle wall is therefore
assumed to be Maw ,~ 1. The boundary layer profile is joined to the computed inviscid
profile at the exit.
Simons t3*~ has generalized Boynton's numerical results. In addition to the parameters
already ruling the inviscid case as mentioned above, he introduced the nozzle boundary
layer thickness 6 and described the plume properties analytically. The starting point for his
generalization is the exponential decay of the density in the boundary layer expansion
region (Fig. 12), quantitatively
f(@) = f(@o)exp[ -/~(@ - @o)];

> @o

(11)

(where @o denotes the edge streamline of the boundary layer expansion region (Fig. 2)),
whereas in the isentropic core
f(@) = cosy-1 -2- -@,m ; < o

(9)

is used as proposed by Boynton. ~3) From mass flow considerations Simons derives
approximate expressions for @o and/~ and relates the streamline angle to the coordinate y
in the boundary at the nozzle exit:
7-___.L1

(Y-- l~l/2(26~?+lln(3_~
= 0 .~ k~' + l /

\rE/
2Ap

\y/
?-1

(12)

Plume flow and plume impingement in space technology

17

The quantity ~ in Eq. (11) is

~,-1
h //' q- 1 ~1/2 (2 t~nm ~//rE ~ '+1

(14)

As average limiting velocity tllim in the boundary layer expansion region Simons suggests
1

~Ulim ~ I~lim <~ Ulim.

AFis the

(15)

so-called plume constant similar to B in Eq. (2)

A~r*~ 2

, [ - ~ - ] f((~)

(16)

which is calculated by mass flow considerations


Ap

U*/(2 Ulim)

(17)

rejo,,= frO) sinO dO


In the equations the boundary layer thickness appears now as the ruling quantity. No
statement is made on the boundary layer details, i.e. the determination of the boundary
layer profile and thickness.
Simons' model has found a broad application in plume flow calculations either in its
original or in modified and extended form. To apply the model the determination of the
boundary layer thickness is necessary. This was done by Lengrand ~3s~whose model is also
based on Boynton's results and is similar to Simons' determination of the flow field
quantities. Lengrand considers boundary layers with 6 ,~ re, following the original concept
of boundary layer theory. Later Lengrand, Allrgre and Raffin ~36) improved Lengrand's
model by introducing a calculation for thick boundary layers (up to 6 ~ 0.2rn) which is an
essential feature of small thruster nozzle flow. This introduction also leads to modified
equations for the constant Ap, the angle Oo and the relation between y and O. In their
experiments to test the validity of the model they used small supersonic nozzles with
different area ratios e (leading to different Ma~) and nozzle half angles Oe. The density was
measured directly. Along the axis the authors found excellent agreement between experiment and theory, confirming the validity of their determination of the constant Ap. In the
transverse profiles the isentropic core function frO) (Eq. (9)) was found to fit the experimental results at large distances from the nozzle (x/rE = 40-60).
Experimental access to the boundary layer expansion region was possible in the
experiments of Calla and BrookJ 37) Mass flux measurements and estimates of the velocity
were performed in the region O = 30-90 of plumes emanating from a 1/10 scale model of
a LMRCS engine (Lunar Module Reaction Control System). Nitrogen and simulated
propulsion products were used as test gases. The evaluation of the velocity estimates
revealed an exponential decrease with increasing angle O. The main measurements of the
mass flux showed dearly that neglecting viscous effects in the nozzle flow leads to a
dramatic underestimation in the outer regions of the plume (by a factor 0.1 and less,
compare also Fig. 12). For O > Olim, where the method of characteristics calculation
delivers no mass flux, values of IO-3(pU)cL were still measured (r fixed).
In the comparison of the experimental data with Boynton's calculations and results from
Simons' model 'qualitative agreement' and 'reasonable agreement' were found, respectively.
In both cases the calculations in the region O ~ 90 gave normalized dynamic pressures
(Boynton) and densities (Simons) by a factor 2 to 3 times smaller than the experiments. This
is a considerable discrepancy which in part may be a result of the omission of the subsonic
boundary layer in the calculation.
Genovese ~3s)found the Simons model equations to be also suitable for a rapid estimation
of plume interaction. For the limiting velocity in the boundary layer expansion region tillm
= 0.75U.m is proposed and the boundary layer thickness is determined using the equation
JPAS 28:1-~

18

G. DE'VrLEFF

for a flat plate. ~39)


1

Now hydrazine monopropellant thrusters are considered and a further problem in the
theoretical treatment of thruster plumes is faced, namely the determination of the exhaust
gas properties viscosity # and ratio of specific heats V. Monopropellant hydrazine is
decomposed according to the chemical reaction t4~
3N2H 4 ~ 4NH a + N 2 + 80.15 kcal
4NH 3 -. 2N 2 + 6H 2 - 44 kcal.
Depending on the dissociation degree of NH3, the stagnation temperature, the gas
composition and thereby the viscosity p(To) and the ratio of specific heats ~,(To) are fixed.
Genovese t3a~ assumes V = 1.26 which corresponds to a dissociation degree of 50% and a
mean relative molecule mass M - 13.7.
Since y is the essential gas property determining the expansion behaviour this quantity
should be known as exactly as possible. Dettleff et alY~1~have determined V = 1.37 for the
plume gas, a value, which has also been calculated by KewleyYTM The difference of about
8% between this result and Genovese's value, ~ = 1.26, leads to unacceptably different
results. The resulting plume flows are completely different (compare the influence of ~ later
in Fig. 15). The values ofy reported here are assumed to be constant throughout the flow, i.e.
the gas is considered to be perfect.
As a last example of adoption and extension of Simons' model we present the work of
Boettcher and Legge.t43-45~ They use the Eqs (9-17) of Simons and approximate the
boundary layer thickness by
6E = 6.25~-~er

(19)

A quantitative measure of the viscosity effects in the nozzle flow is the Reynolds number
Reg
PruElr
ReE = - (20)
/xo
where Ie is the nozzle length, PE, u~ the density and velocity respectively, at the nozzle exit
(calculated for pure isentropic flow) and/~o is the viscosity of the gas at the stagnation
temperature To. Boundary layer thickness c~and Ree are related to each other according to
boundary layer theory (39) by
1
6~ - (21)
The usefulness of Eqs (9)-(17) for engineering purposes has in part been shown by
experiments. Evaluating Pitot pressure measurements Dettleff tal~ and Legge et al.t46~ have
determined indirectly the density along the plume axis and thereby the plume constant as
proposed by Simons. Using two nozzles from original monopropellant hydrazine thrusters
(ERNO 0.5 N and 2 N, Fig. 6) with different geometrical forms, nitrogen was selected as the
test gas. In the experiments the Reynolds number Re~ was varied as the parameter. The
Pitot pressure profiles from which the plume constants were derived are depicted in Fig. 13.
In hypersonic flow, Pt2 P, and therefore the density along the plume axis exhibits
qualitatively the same profile as the Pitot pressure. The completely different shapes for the
two nozzles in the region x < 30 mm will be discussed in Section 4.
Further, three test gases were selected to study the influence of the ratio of specific heats
on the plume constant (CF 4, 7 = 1.17; N 2 , )~ -- 1.4; Ar, ~ = 1.67). The parabola shaped
curves Ap(~) (Fig. 14) demonstrate the importance of y when determining the plume
expansion behaviour. This can also be found in a theoretical parameter study performed by
Boettcher and Legge.t43~ A result of this work is shown in Fig. 15.
'~

Plume flow and plume impingement in space technology


a)

'

I0"i

10-2

..'~. "~

10-3

19

7s

:1

I~--I7

Pt=

.............

)0q3560
F%

po ,or,

--'--

borll~2~01

--'-'-- '~~~1

"%
%
I

10!

30

X
b)

,o_,l

mm

300

i
~8

10./.,8

8"53

10";

10-:
Pt2
Po
10-t

.........
--x--I "-

PQ ReEN
2 bar 5600[
8 bar 12~0 I
,6 bor

lO-S

i,,8001

30

-q
mm

300

FIG. 13. (a) Axial Pitot pressure profiles. Nozzle ERNO 0.5 N (Fig. 6a), test gas N2, To = 300 K, parameter po.
Adapted from Ref. (123). (b) Axial Pitot pressure profiles. Nozzle ERNO 2 N (Fig. 6b), test gas Nz, To = 300 K,
parameter po. Adapted from Ref. (123).

The extensions of Simons' model, introduced by Boettcher and Legge, concern in the first
place the complete description of the thermal state of the gas in the boundary layer
expansion region. For this purpose the concept of effective stagnation conditions is
introduced. Although the flow in the nozzle boundary layer is strongly influenced by
viscosity, it is assumed that the free expansion beyond the nozzle exit is isentropic.
Nevertheless, as the experiments of Calia and Brook (aT) indicate, the gas does not reach the
limiting velocity u~im =/2~__~ RTo (Eq. (10)). Instead,
y?-i

a smaller velocity ulim((9)is reached:

U,m((9) ~ exp( -- O)
O1"

Ulim(O ) = Ulim,CL exp( -- Cu( -- (90)).

(22)

20

G. DETTLEFF

13
ERNO 0.5 N

1o

j"

Ap 5

1.0 1.14

1.4

1.67

-~

FIG. 14. Plume constant Ap as a function of 7. Three different test gases CF4 (7 = 1.17); N2, (7 = 1.4); Ar, (y = 1.67);
T o =- 800 K, Re r = 3600. Adapted from Ref. (41).

y=167
.....

10"
P
p"~

_ . _

,~
~

10":

'~
~,
~,
~,

"'.....

...,.
\
"".,..
\
.......
"~
"...

\~,

10--~

"~ \

,'oo

;i;= 6
,=1,7

', "< . . . . . . . . . . . . . .

~"""

8'0 o

"""'"-..

1~o o

160

FIG. 15. Angular far field density distribution. Conical nozzle, e = 25, OE = 10. Parameter: ratio of specific heats
~. Adapted from Ref. (43).

According to the form of Eq. (10) an effective stagnation temperature To, e:: () is to be
defined

u..,(O) = ~ / 2 _ ~ 1 R To.e:: (O)


which can be found by combining Eqs (10), (23) and (22)

(Idli m =

(23)
Ulim, CL)."

Unra(O) "]2 = T'eH = e x p ( - - 2cu( O -- O o) ) .

Unm,cL ]

(24)

To

In a similar way effective stagnation pressure and density are introduced, allowing the
calculation of not only the density in the boundary layer expansion region but also the static
pressure, temperature and M a c h number. This can be of advantage in certain plume
impingement situations, w h e n the k n o w l e d g e of density and velocity alone is not sufficient.
As an example of the application of this m o d e l i s o - M a c h lines in a plume flow have been
determined (Fig. 16).

Plume flow and plume impingement in space technology

21

.2(:

.15
"E .10
.05

35
.0 _.z,,--

nozzle
(not scaled)

.10

.20

30
x(rn)

.40

.50

FIG. 16. Contours of Mach number and freezing surface P = 2, calculated with the model of Ref. (44).
ERNO 0.5 N hydrazine thruster.

The second extension of Simons' model introduced by Boettcher and Legge concerns
rarefaction effects. The plume flow can only be considered as isentropic as long as the
perfect gas is assumed to be in thermal equilibrium despite the rapid changes which it
undergoes. This is achieved by a sufficient number of intermolecular collisions. The flow in
1
a plume has diverging streamlines with p ~ ~ . Since the collision frequency v ~ p, v also
1
decreases according to v ~ ~ (assuming constant, collision cross-section). At a certain
distance r, a density decay Ap along a streamline will therefore no longer be accompanied
by a decay A T of the temperature corresponding to the adiabatic equation of state, since the
number of intermolecular collisions is not sufficient.
The onset of this rarefaction effect in the expanding plume flow is commonly described by
Bird'sc47) breakdown parameter
1 d(lnp)
P = v ~

(25)

which for one-dimensional steady flow (as in simple plume flow models) takes the form
=u
P

Sincep~

r3,V

pv

dp
dr

"

(26)

~andu~const,
P~r.

This relation shows that increasing rarefaction is indicated by increasing P. As onset of nonequilibrium effects Bird (47'4s) found in his example calculations with the Direct Simulation
Monte Carlo method (DSMC)
PI "~ 0.05.
(27)
A result of the calculations is shown in Fig. 17. The deviation from isentropic flow occurs
gradually. The curve for Tx seems to approach a horizontal line (constant value for Tx)
while Tn continues to decrease even faster than in continuum (in this case). As a result of the
approach to a constant value the temperature T~ is said to become frozen.
An earlier proposal has been a sudden freezing instead of a gradual process. It means that
the transition from continuum to free molecule flow occurs discontinuously at a certain
location on the streamlines. This concept of a freezing surface is nowadays used in plume
modelling to avoid the difficult description of the details of the transition. The location is
placed within the transition region with P > 0.05. Boettcher and Legge(*' have defined
P = 2 as the freezing surface in their model. The contour with this value is also shown in
Fig. 16.

22

G. DETTLEFF

0.2

0,1
0

0o

0.05

T
%

0.02

0,01
Tn/T

~~~inuum

0.005
%\

0.002 2
I

, I

10

r/r"

20

. ~ 7

SO

FIG. 17. Breakdown of translational equilibrium in the spherical expansion of a hard sphere gas. Kno = 2Jr*
-- 0.002. T~ and 7",: parallel and normal components of the kinetic temperature, resp. Adapted from Ref. (47).

So far we have emphasized the temperature T as a quantity which is affected by


rarefaction effects and deviates from equilibrium behaviour. It is obvious that quantities
u

,/y

which depend on T as for example the Mach number Ma . . . .


leading to Max

~,

are affected too

which becomes frozen and Ma, ~ ~/T, which according to Fig. 17

will continue to increase. Another quantity affected by rarefaction is the ratio of specific
heats ? of the gas. In liquid propellant thrusters the combustion temperature is above
1000 K and this means that the vibrational degrees of freedom of the molecules are also
excited. In the expansion flow the temperature decreases and the internal energy is
transferred to kinetic energy of the flow by intermolecular collisions with a corresponding
increase of ?. If the number of collisions is not sufficient to allow y to adjust to the
temperature T, thermal equilibrium is not maintained. This lag in energy exchange is called
vibrational relaxation. The rate of energy exchange depends strongly on the gas species: for
example, the characteristic collision number to regain equilibrium between translational
and rotational energy modes is about 60 times larger for hydrogen than for nitrogen (Tang
and Fenn(49)).
There are two consequences for the gas flow: Firstly it is possible that the maximum
velocity Ulimis not reached, since the kinetic energy potential remains stored in the internal
energy modes (and is not transferred into bulk motion of the gas). Secondly the gas behaves
as if the ratio of specific heats y is effectively larger than if the gas would exhibit no
relaxation.
The consideration of rarefaction effects and its description by the concept of the freezing
surface offers the possibility for a refined treatment of mass flux distribution. In the model
descriptions presented before all flow came along apparently straight streamlines from the
source at the nozzle. This is still true for the main mass flow in the free molecule flow region,
but now additional contributions can come from the whole freezing surface (see Fig. 18).
The subdivision of the flow into a continuum and a free molecule region includes also the
assumption that no inter-molecular collisions at all occur in the latter. Last collisions take
place at the freezing surface and then the molecules are released downstream according to
the Maxwellian velocity distribution at r s (Fig. 18). Therefore the principle of cone of sight is

Plume flowand plumeimpingementin space technology


Oo

23

streamlines

cone of sight

\
freezing

surface

FIG. 18. Freezingsurfaceand cones of sight at two differentlocationsO1, 02; 02 in the backflowregion.

applied. It means that at a certain location O in the flow molecules arrive from all points of
the freezing surface, which can be seen from this point O.
A further consequence of plume rarefaction is mass separation, which can be of particular
importance with respect to contamination effects. Disregarding for the time being
the simplifying freezing surface concept with a collisionless free molecule regime, the
increasingly rarefied expansion flow is characterized by increasing mean free paths, 2. The
main flow direction is still radial, but deflections occur due to inter-molecular collisions.
Larger deflections are possible for light molecules, while heavier ones tend to maintain their
flow direction. Since there is a density gradient perpendicular to the streamlines with a
density maximum on the axis heavy molecules will tend to remain in the near axis region,
while lighter species with its larger mean free path will be shifted into the outer region. This
species separation effect has been investigated recently by Allen et al. (5) in a free jet
expansion with various gas mixtures.
Rarefaction of the plume does not only occur downstream of the isentropic core far away
from the nozzle where the streamlines are practically straight. In the off-axis region these
thermal non-equilibrium phenomena can already appear in the plume near field relatively
close to the nozzle, where the application of continuum methods (MOC for example) would
still predict curved streamlines. Naumann (s 1) has investigated experimentally the streamline
pattern in this region and has in fact found a 'freezing' ( = ceasing) of flow deflection. Its
onset could be shown to coincide with Bird's parameter P ~ 0.06 (compare Eqs (25)-(27)).
The exact method of characteristics can be used in continuum flow with thermal
equilibrium, but its validity fails when rarefaction effects appear. The same holds for the
model equations, which are based on continuum principles. Efforts to take into account
rarefaction effects would complicate the desired simplicity. A certain compromise is the
freezing surface concept. The only numerical method by which the flow downstream of the
continuum region can be calculated is the DSMC method.
Boyd(Sz) has performed a detailed comparison of MOC, DSMC, and Simons' model with
experimental data from Pitot pressure measurements in the plume of the ERNO 0.5 N
thruster. As expected, the MOC results were throughout closer to the experimental data
than the results of the model calculation. The comparison of MOC- and DSMC-results
in the transition flow regime then dearly revealed the error inherent in the application of
MOC in this regime, where the assumption of thermal equilibrium is not valid. While the
flow velocity differences are negligible, density and temperature differ considerably at the
free molecule flow locus in the region investigated in this study (near axis region, < 24).
MOC underestimates the density about 15-20% in this case.
Following 1960, the isentropic continuum flow was initially of interest. Then consideration was extended to the nozzle boundary layer and its continuum expansion and finally
to rarefaction effects. In recent years the back flow region of the plume has attracted
considerable attention. In this highly rarefied flow regime with flow angles > 90 plume
impingement can hardly be avoided, since canting of the plume axis as explained in Fig. 7 is
not appropriate. Moreover forces or heat load are not the most serious impingement effects
in this region with highly rarefied flow but rather gas deposition. Scientific satellites like the

24

G. DETTLEFF

Infrared Spectroscopy Observatorium (ISO) are equipped with cryogenically cooled instruments exposed to space. Gas deposition in the order of some molecule layers can readily
impair the quality of the experimental equipment.
The source of the back flow is the boundary layer flow at the nozzle exit. Contributions
can also be expected from the freezing surface in the plume (see Fig. 18). An important step
in understanding the onset of the boundary layer expansion at the nozzle lip has been
contributed by Bird. (4s'53) Using the numerical DSMC method it was shown that the
contour M a = 1 in the nozzle flow ends at the nozzle lip, if the plume is underexpanded.
Thus the flow conditions in the exit plane are completely supersonic. According to
boundary layer theory Maw = 0 would be expected along the whole nozzle wall; Maw > 0
in the vicinity of the exit is an indication of rarefied flow features (slip flow~3)).Since the back
flow is rarefied and since its source at the nozzle exit with its very rapid transition from
continuum to free molecule flow can only be treated correctly by the DSMC method, this
numerical technique has been used almost exclusively in recent years.
Hueser et al. ~54) have treated the case of an upper stage solid rocket motor at a height of
about 280 km to investigate the flow angle contours and effects of species separation in the
back flow. The exhaust gas components are light like H2, medium like N z, and heavy like
COz. Light molecules are found in larger abundance at large angles, while heavier
molecules are represented less than in the nozzle flow. This effect has also been reported by
Bird.(4s)
Another investigation with the DSMC method has been performed by Campbell. ~55) In
this study a tube instead of a nozzle is considered. The geometry (square lip edge and knife
edge lip) and the gas density and temperature are used as parameters and their influence
on the mass flux in the back flow region is investigated systematically. The two different lip
shapes result in considerably different mass fluxes by a factor of about 2. Increased
stagnation temperature also leads to a substantial increase of mass flux, while increased
density in the boundary layer results in decreased mass flux into the back flow region for
thin lips. For thick lips the mass flux is almost unaffected by density variations in the
boundary layer according to this study.
Bayleyts6) has investigated the back flow experimentally and reports on measurements of
the flow angle. The nozzle lip peculiarities with a local flow angle in the exit plane larger
than E (Bird~4s' 53)) thus find the first experimental support. In addition it is found that
the background gas density influences the expansion flow behaviour in this region very
strongly.
Recently Boyd~s2) has performed a further numerical study (DSMC) to assess the
influence of several basic assumptions. The influence of boundary layer profile variation on
the degree of back flow is investigated systematically for the ERNO 0.5 N thruster. As a
strong parameter the gas temperature at the nozzle wall is traced: the higher the temperature, the greater the mass flow around the lip. Again a large degree of species separation for
this thruster is found (main exhaust constituents: Hz, N2, NH3). With reference to
Bailey'sIs6) experimental study of flow angles, Boyd finds that it is insensitive to his basic
assumptions.
Finally we wish to mention condensation, an effect associated with the expansion (but not
rarefaction) in the plume. The continuous decrease of the temperature in the expanding flow
can lead the gas to the phase change border, appropriately represented in a vapour-pressure
diagram (Fig. 19). Condensation is possible since the isentropes of the main exhaust gas
constituents have a smaller slope than their vapour-pressure curves. Note that while
relaxation is related to the rate of change in the flow, condensation is associated with the
thermal state of the gas achieved in the flow. The extent of mass fraction transferred to the
liquid or solid phase depends of course on the degree of rarefaction, since droplet growth is
a function of the collision rate.
The formation of the liquid phase in the flow is associated with an energy transfer to the
gas phase (latent heat is released), which in turn increases the gas temperature and
influences the flow Mach number. Thus the formation of dusters or liquid droplets (or solid

Plume flow and plume impingement in space technology

o:tripte point

,0I

25

T--300K 1000K

/, \

10 ~

Pv 10 e

10"I2

100

'

10~
T

10 2

103

FIG. 19. Vapour pressure of thruster exhaust gas components and isentropes for ? = 1.4 (perfect gas).

particles) basically changes the type of flow. Instead of a single gas phase a two-phase flow
(gas/droplets or gas/particles) needs to be considered.
In Fig. 19 an isentrope is drawn through po = 10 bar, T O = 1000 K for 7 = 1.4, which is
a typical stagnation condition for hydrazine monopropellant thrusters with its main
constituents H 2 and N 2. The intersections with the vapour-pressure curves are at T ~ 30 K
for N 2 and T ~ 5 K for H 2. For N 2 a temperature ratio T / T o = 30 K/1000 K - 0.03
corresponds to a Mach number M a ~ 12, which for the ERNO 0.5 N thruster plume is
achieved at r = 0.034 m on the axis. This estimation means that beyond this distance to the
nozzle exit, condensation of N 2 is possible as far as the thermal state is concerned (if the gas
is already too rarefied, droplets will hardly be formed).
Liquid droplets in plume flow are not only a result of a condensation process in the
expanding gas but are sometimes unburned propellant. Their existence and distribution in
the plume must be treated in a different way than the droplets considered above. Here we
have reached the limit of the scope of this paper, because this problem includes not only
the consideration of two-phase flow (fluid mechanics), but also the combustion process
(chemical reactions). Nevertheless we note that extensive work has been performed in this
area, in part connected with a computer code called CONTAM. The computation includes
the treatment of combustion chamber processes with a prediction of the engine performance
and the time-dependent formation of gaseous and liquid combustion products. It includes
the consideration of two-phase flow and contaminant deposition on surfaces,t57~ Extensive
laboratory work has been performed to obtain experimental validation of the predicted
results (Refs (57) and (58) with additional references).
The complete flow field to be considered in plume impingement problems consists not
only of the free expanding plume, but comprises also the nozzle flow. This already became
clear when introducing the boundary layer expansion region and particularly the back flow.
Some more details of this part of the flow will be considered in the next section.
4. N O Z Z L E F L O W AND P L U M E NEAR FIELD
Considering the linear dimensions of a plume relevant to impingement, the nozzle flow
typically occupies only less than 1%. As an example consider the plume in Fig. 1, where the
length of the thruster nozzle is about 1 cm. Only along this comparatively small length is the
expansion of the gas guided and influenced by viscosity.

26

G. DETTLEFF

I.

Free jet from sonic orifice, d =0.6ram


I
I
Nz,Po=2 bar, To= 300 K

m
8ooo
4~oo

0= 60"

,/0

20o

.,~

I0O0

J /streamlines

,~

#
0

Iooo

8OOO

4ooo

x/d ~

P(ume:from
(=62

iozzle

eE=15 d ~ = 0 6 m m
f
' I
]

IN2.po=2bar, To=300K

y/d*

l~l##
Io##

str'eamtines

1o

I
O

.20"

o
x/d N

FIG. 20. Contours of constant density in free jet (above) and plume (below). Identical stagnation conditions,
identical diameters of sonic orifice and nozzle throat, ~, = 1.4.

The considerable effect of the guided expansion in the nozzle prior to the free expansion
into vaccum can easily be demonstrated by comparing such a plume with a free jet from a
sonic orifice (Fig. 20). Lines of constant density have been determined for both flow fields.
The contours for the plume lie close to the axis, while those for the free jet resemble circles,
indicating a more lateral spreading. This is also demonstrated by the comparison of the
mass flow distributions in Fig. 21. The plume near axis region with _< 30 contains more
than 90% of the total mass flow rotor,but the corresponding region in the free jet contains
just 35%.

O" 10"

30"

,~(o)

FIG. 21. Mass flow distribution - -

TfJtot

60"

90"

as a function of O for free jet and plume jn Fig. 20.

Plume flow and plume impingement in space technology

27

The effect of viscosity was clearly demonstrated in Fig. 12 by comparing plumes


emanating from nozzles with inviscid and with viscid flow. Regarding the relevance of
nozzle flow with respect to the plume flow demonstrated in Figs 20 and 12 it is worthwhile
looking at the details of the source of the plume which, with increasing requirements for a
refined flow and impingement description, will receive increasing importance.
The main task of the nozzle is to produce thrust, F. A minimum thrust F* is obtained by
F-F*

the flow through a simple orifice (sonic conditions), and the ratio ~

is defined as thrust

gain. It is the intention of the nozzle designer to make the thrust gain as large as possible,
which can be achieved by fulfilling three requirements: (a) the flow at the nozzle exit is
uniform, (b) the velocity at the exit is close to U~im,and (c) the boundary layer is thin. In
principle a nozzle with a uniform high velocity flow at the exit is achievable. However, the
length IE of such a nozzle would be considerable, and since the nozzle weight is proportional
to l~, and ~ ~ x/~E, in practice a good gain must be achieved with a limited length lE.
Under this restriction the selection of the area ratio ~ and of the nozzle shape becomes
important. The conical nozzle has the simplest shape (Fig. 6). A few percent additional
thrust gain can be achieved with a contoured nozzle (Conway and Celmins (59)) and for a
highly viscous flow with a trumpet shaped nozzle (Kallis et al.(6)). The different shapes
result in different flows, demonstrated for the conical and the contoured nozzle in Fig. 22. In
both cases compression waves are observed, though at a first glance the nozzle shape would
lead one to expect only gas expansion (Fig. 22a and 22b). The contoured nozzle shape
reduces the flow divergence angle near the nozzle wall (Fig. 22d) accompanied by a
recompression of the gas (Fig. 22f).
The profile obtained with the method of characteristics shows qualitative agreement with
the experimental Pitot pressure profile for the contoured nozzle (Fig. 22f) except in the
vicinity of the wall (boundary layer). In the case of the conical nozzle there is not even
qualitative agreement. The reason is that viscosity plays a dominant role in this nozzle flow
as is demonstrated by the experimental profiles in Fig. 22c where the Reynolds number is
varied. The range of R e E corresponds to the real thruster. In Fig. 23 it is demonstrated how
the initial spreading of the plume is affected by such a variation. Two streamlines passing
a)

.nozzle

shock

axis

FIG. 22. (Contd.)

'

' ~ " " ~ " - ' ~ ~-~'~ ~ ~

'~

28

G. DETTLEFF

through the points y/rE = 0.42 and 0.63 of the nozzle exit have been obtained for two
different values of Ree. The initial spreading is much more pronounced for the larger value
of R e E.
The strong viscosity influence is an inherent feature of nozzle flow in small thrusters as
used in space technology. Different expressions exist for the Reynolds number:

peur, lE
ReE = - -

(18)

#o

(nozzle exit conditions are taken into account);

Re*

p*u* d*

(28)

#o

(expresses Re in terms of throat conditions);


c)

16
14
12
10'
8"

6"
4.
e

2"
O-

de 9

in

-2-4-

-8-10 -12 -

rE

-14-16

y (mm)
d)

14
12
10
8

6
4
0
in

deg
o
-2
-4
-6

-8
-10,

rE

-12
-14

-4

-3

-2

-1
y (ram)
FIG. 2 2

(Contd.)

'

Plume flow and plume impingement in space technology

29

oo5 t
o.ot.
,"

0.03

) .

~+ ~

'? Yy

Yv,

PI2

Po
0.02

:
-

.......

:J'

I-'-

o I E
0.5 bo~

890

I 1.0bor11780

0.01
I

-2.37

-2

-1

I
mm

2 2,37

f) 0.75

0.50

IIV.T"

/I

Pt 2

Po

0.25

-j

J
-4

-2

/, rnm

FIG. 22. Nozzle flow study with conical and contoured nozzle (s~ Fig. 6a and 6b); test gas N2. (a) and (b)
Characteristics net; (c) and (d) Flow angle in nozzle exit plane (coordinate y) derived from characteristics net; (e)
and (f) Pitot prt~surc profilos in nozzle exit plane. Experimental results from Ref. (123). MOC calculations from
Ref. (124). Navier-Stokes calculations by Bcrnd Miiller and Klaus Hannemann, DLR, Institute for Theoretical
Fluid Mechanics, G6ttingen.

Reo =

P o Ul|m r ~

- #o

(29)

(expresses Re in terms of stagnation condition). T h e b o u n d a r y layer is l a m i n a r for F < 40 N


and turbulent for F > 400 N (Boynton(3a)).
Values for the viscosity coefficient # must be taken from the literature (for example (61' 62)).
T h e t e m p o r a t u r e variation is expressed by

#(T)

#(T,,:)

(30)

30

G. DETTLEFF

10

Po=O.Sbar
(ReE= 8 9 0 )

y(mm)

YE
rE

0,63"
. . . . .

0.42

0.42
Po= 8.0 bar
(ReE=14240)

I
-0
-I

-5

"~
" ~ 0.63

0 1

10

15
X (mr'n)

-10

FIG. 23. Streamline detection in a plume near field. Nozzle E R N O 0.5 N (Fig. 6a), test gas N2,
nozzle exit coordinate of the streamlines. Adapted from Ref. (125).

yE/rEdenotes

the

where the suffix refdenotes known reference values of T and #, and co = 0.75. (44) Boettcher
and Legge(44) use the formula
5~
m
# = ~zrRT~
(31)
where m is the molecular weight and f~ is the collision cross-section. For mixtures of such
gases a somewhat more complicated equation is available (the non-constant composition of
the exhaust gases, depending on the firing mode, requires such formulae).
The application of refined models makes it necessary to determine the boundary layer
thickness 6. This can be done either by calculations based on the Navier-Stokes equations
or by evaluating measurements like those shown in Fig. 22c and 22f. Down to Reynolds
numbers Rer of about 1000 the subdivision of the nozzle flow into the isentropic core and
1

boundary layer is justified and 5r ,-~. R/_k_~p(Tw = To), but below Rer = 1000 deviations
occur from this simple picture.
The boundary layer profile is very difficult to measure. With a very thin probe at least
the Pitot pressure can be measured in small original thruster nozzles (Fig. 22c and 22f).
The determination of wall (static) pressure, density or velocity makes it necessary to increase
the scale of the nozzle. Wall pressure combined with PitOt pressure measurements in the
boundary layer at the nozzle exit have been performed by Naumann (51) to determine the
Mach number profile. A linear relation Ma(y) is found (To = Tw), but as is pointed out, such
linearity is not the general behaviour, and in fact in the investigation of Whitfield(63) the
profile is far from linear for a reduced wall temperature Tw/To = 0.33 (To = 300 K).
Whitfield has thoroughly investigated the influence of the temperature T~ on the boundary
layer profile. It acts very sensitively on the displacement thickness and likewise on the static
temperature, enthalpy and velocity.
The wall temperature Tw is not explicitly mentioned in the models presented in Section 3.
The interaction between nozzle flow and wall also means an energy transfer, which among
other things, determines Tw. The heat flux ~ is a maximum at the throat and decreases to
considerably lower values ( < 10% with respect to this maximum) a few throat diameters
further downstream (as an example, see Ref. (64)).
If the thruster is not fired the wall is cold since the nozzle is exposed to space (though
often protected with a radiation shield). In the continuous firing mode the nozzle will be
heated up until a steady temperature distribution is reached. In the pulse mode firing the
wall temperature will remain at a generally lower temperature level depending on the pulse
characteristics. The variation of Tw during the different firing modes combined with the
strong dependence of the boundary layer profile and thickness on Tw leads one to expect a

Plume flow and plume imPingement in space technology

31

broad variation of flow field properties in the off-axis region including back flow.
A study of viscous supersonic nozzle flow with variation of the wall temperature has
recently been performed by Chang et al., t6s) using the thin layer Navier-Stokes equations.
A very detailed experimental study of nozzle flow has been performed by Rothe. t66~Using
electron beam techniques he determined density and rotational temperature in the nozzle.
The Reynolds number range was Reo = 100-1500. He found that the axial temperature
distribution is monotonically decreasing as expected for Re > 500. However for Re < 300
the temperature has a minimum downstream of the throat, i.e. there is an increase in
temperature towards the exit indicating a thermalization of supersonic flow energy. For
Reo ~ 100 a viscous shock-free transition to subsonic flow towards the exit takes place. This
means that a supersonic bubble exists in the nozzle flow. The transition is shock free, since
the density could be shown to decrease all along the nozzle axis. Another feature of the
viscous flow with Reo < 300 is the almost similar normalized radial density profiles P(Y/Yw)
for different axis positions x, while they are dissimilar for Re > 500.
Viscosity plays not only a role in the supersonic part of the nozzle but also in the throat.
The mass flow through the nozzle is expressed by
r*2~

(32)

rhid = P o ~
x/ RTo

with
~+1

r(y) =

3'

(33)

if a uniform inviscid flow with velocity u* at the throat can be assumed. At low Reynolds
numbers a significant boundary layer thickness 6 compared to r* exists at the throat. Then
r* - 6 = r*sl must be taken instead of r* when describing the mass flow which, reduced by
viscosity, is expressed by the discharge coefficient
co = mi--d"

(34)

This is important for quantities which are related to or normalized by po, since the relation
mid ~ Po (Eq. (32)) no longer holds.
A systematic determination of the discharge coefficient has been performed by Tang and
Fenn, 49)validating a theoretical expression given by the first of the authors. For R e < 700
for example, cD < 0.9 for nitrogen, where
1

Re =

#.

\ rt /

A thruster type operating at such very low Reynolds numbers is the biowaste resistojet,
which is a potential candidate for attitude control on manned space stations using biowaste
products like H20, CO2, CH4. A comprehensive study of such thrusters with F < 0.1 N has
been performed by Kallis et al. ~6) The specific impulse I,p, in Section 2 introduced as one of
the essential thruster data, is about 10% less than in inviscid flow. In the experiments a
trumpet shaped nozzle (see Fig. 6) was used, since for highly viscous flow the thrust gain due
to increased area ratio over-compensates the thrust loss due to increased divergence of the
flow. The values of Re* were below 4000.
The existence of shocks in the nozzle flow has been found in MOC calculations (Migdal
and Landis, t67) Darwell and Badhamt6S)), and has then been shown experimentally (Migdal
and Kosgon, (69) Back and Cuffel~7)). Examples of such shocks are shown in Fig. 22a and
22b. They are a result of the over-expansion behind the throat followed by a recompression
at the conical or contoured part of the nozzle wall (note that this phenomenon is not
observed in two-dimensional flow).
Shocks are not considered in the simple models, but as is demonstrated in Fig. 22 and
Fig. 13 these shocks are not only dominant in the flow pattern of the nozzle but also of the

G. DETTLEFF

32

plume. Pitot pressure profiles, measured perpendicular to the plume axis in the near field,
are shown in Fig. 24. A shock pattern in the x-y plane is depicted in Fig. 25. The intersection
of the axisymmetric shock system with the plume axis (location xs) corresponds to the
pressure peak in Fig. 13b. In Fig. 26 it is demonstrated how viscosity (parameter ReE)
influences the shock pattern for ReE < 4800 (in Fig. 13b it is Re~ > 3200 and the shock
position on the plume axis is hardly influenced by viscosity).
Instructive Schlieren photographs of the shocks which continue beyond the nozzle exit
into the plume near field are shown by Stilt. (2)
The thrusters considered here are fired in hard vacuum surroundings. Effects caused by
external gas or flow around the nozzle like separation are not present. However, the vacuum
leads to an effect not observed in thrusters fired in the atmosphere and known under the
term nozzle lip problem. Bird (53) showed by means of calculations with the DSMC method
that the contours with Ma = 1 end at the nozzle lip, so that the boundary layer at the nozzle
exit is completely supersonic. This feature of the nozzle flow is a result of the boundary
condition Pk = 0. The resulting pressure gradient at the nozzle lip leads to an acceleration of
the gas associated with a streamline angle (with respect to the nozzle axis) larger than the
nozzle exit angle.
Though the expansion in the nozzle is guided (prior to the free expansion into vacuum) it
can be rapid enough to cause relaxation effects which influence the ratio of specific heats of

10.L,8"

0.075 . ~ S ~ S . 5 3

~,

0.05

* 0 (nozzle exit)
n3
x6

+
Pt2
Po

**

0.025

-6

-5

-4

-3

-2

-1

1
mm

Y
FIG. 24. Transverse Pitot pressure profiles Pt2/Po -- f(Y). Nozzle ERNO 2 N (Fig. 6b), test gas N2, To = 290 K,
po = 1 bar. Adapted from Ref. (123).

mm

10

/0

~ 0 - -

~ o ~

o ~
"

FIG. 25. Shock structure in the plume of the ERNO 2 N thruster. Adapted from Ref. (126).

Plume flow and plume impingement in space technology


I

x
nozzle

exit
)
Po[bor] ReE

Pt._j_Z
PO
x

0.08

n5

0.2
1.0

560
2800
8~00
I
I
3
5

10z.8,"

10 .2

I0 "3

33

3.0

I
2

\.
",.

+++ [

.?

"%j
.J"
~, ~

I
10

%1

20

~1~
%-I

I
mm

50

~---

F]o. 26. Axial Pitot pressure profiles Pt2/Po = f(x) in the plume near field. Nozzle ERNO 2 N (Fig. 6b), test gas N2,
To = 290 K, parameter po. Adapted from Ref. (123).

the gas expanding downstream into vacuum. Also condensation as a result of expansion is
possible in the nozzle flow (these two effects and their influence on the flow have been
discussed in Section 3).
Summarizing, we can therefore state that the thrust gain-producing nozzle flow on close
inspection exhibits a variety of phenomena which can be relevant for the plume flow
description in the context of plume impingement. We will see in the following chapter that
in continuum and transition flow impingement effects can only be determined with a
relatively high degree of uncertainty even if the undisturbed plume flow is known exactly.
On the other hand, in the free molecule flow the methods to determine the impingement
effects are refined and exact but major uncertainties can arise from the lack of exact flow
field determination. In such cases efforts to improve the flow prediction necessarily include
a refined treatment of nozzle flow.
5. P L U M E I M P I N G E M E N T
5.1. IMPINGING PLUME FLOW CHARACTERISTICS

In aerodynamics the flow is usually uniform and the body is totally immersed in this flow.
In contrast, with plume impingement the flow is not uniform and very often the total surface
is not impinged.
In Section 3 we have seen that there are steep gradients especially of the density along and
perpendicular to the streamlines. The flow types range from continuum to highly rarefied
flow resulting in different types of interaction with solid surfaces (Fig. 3). In continuum flow
impingement a shock is established in front of the impinged surface (surface shock) and a
boundary layer is formed, i.e. compression and viscosity come into play.
In free molecule flow impingement, the incoming molecules reach the surface undisturbed. Their interaction with the wall consists of an exchange of momentum and energy
and reflection or sticking, commonly denoted as gas-surface interaction.
The distinction between the impingement flow types is characterised by the Knudsen
number
Kn =
(36)
/,ef
JPAS

28:1-C

34

G. DETTLEFF

where 2 is the mean free path of the molecules and l,ef is a reference length. This can be
a typical length of the impinged body or the distance to the nozzle. For Kn ~ 1 the flow
impingement is free molecular, for Kn ~ 1 it is continuum. For Kn ~ 1 we have transition
flow impingement. It is, however, not the same as the rarefaction phenomena, termed flow
transition and described in Section 3.
For Kn ~ 1 ( > 10) there is practically no interaction between the incoming and reflected
molecules. If Kn is decreased below 10, collisions occur and thus the incoming flow
experiences small disturbances (Fig. 27). With further decreased Knudsen number the
disturbance becomes stronger, the features of compression of the gas in front of the body
appear and finally, for Kn ~ 1, a surface shock and boundary layer exist. Some illustrative
examples of flow-surface interaction with varying Knudsen number are given in Ref. (71).
In aerodynamics similarity rules are often applied. The forces for example are expressed
in dimensionless form as drag or lift coefficients dependent on dimensionless flow parameters, mainly Mach number and Reynolds number. This similarity procedure is hardly
applicable to plume impingement problems. The reason is the non-uniformity of the flow,
which by its nature, cannot be characterized by one Mach number a~d one Reynolds
number. These can only be specified locally.
Another problem is the determination of the flow at the surface, which is necessary to
understand the details of the interaction with the body in the continuum and transition
regimes. In this case the flow has a history with shock compression and influence of
viscosity and is difficult to describe theoretically.
The shock structure and the quality of the disturbed flow depend on the shape of the
surface, which can vary widely. From the plume impingement situations sketch (Fig. 8),

\, \ .
free mo[ecute flow
impingement,

Kn>>l

J
incoming t
reflected

motecuLes

transition flow
impingement,
Kn=l

streamtines

co uom

,tow

,iiiiiiiiil

Jrnpingement,

Kn <<

J!iiii

FXG.27. Impingement flow types.

Plume flow and plume impingement in space technology

35

o)

shock

'
7 7 / / 7 7 7 / / / / 7
v

IN

supersonmc

plate

7 7 7 7 7 7 7 / / / / / / / / 7

I'.,

.....

subsonic

supersonic

flow behind the surface shock

y~

:x

surface
shock

"~
bockftow

./N._~.

subsonic supersonic
flow behind the surface shock

F]o. 28. Basic geometrical arrangements for plume impingement investigations. (a) Plate perpendicular to the
plume axis; (b) plate parallel to the plume axis.

however, and from experience we can deduce two particular cases as basic geometrical
arrangemeAts, namely the plate perpendicular and the plate parallel (adjacent) to the plume
axis (Fig. 28). Both cases have been investigated extensively and will be used to explain the
main features of the disturbed flow.
Two types of shocks are observed simultaneously: a strong one with subsonic flow
downstream and a weak (oblique) shock with supersonic flow downstream. The strong
shock is observed near the point where a contour of constant Mach number (see Fig. 16)
is tangent to the impinged surface. (72) If the plume axis is perpendicular to the surface,
a stagnation point exists at the intersection of the axis with the surface coinciding with a
pressure maximum. For the impingement on an adjacent surface the pressure distribution is
also characterized by a pronounced peak associated with strong gradients (Fig. 29). The
plume centerline
nozzle

P$
/ Z ~/
/ /
//
/ /
J
,~ /f

/ Z
7
-//.
~7/

,J6

10 ~ \ f l a t

projection
of plume
center|ine
on plQte

plate

FIG. 29. Pressure distribution schematic. Adapted from Ref. (72).

36

G. DETTLEFF

stagnation point lies at the peak or slightly upstream. The sonic line in Fig. 30 indicates that
the subsonic flow along the plate is accelerated to supersonic Mach numbers, i.e. the shock
on the far right hand side in Fig. 30 is oblique.
In the oblique shock region (the main part of the shock in Fig. 31) the distance between
the shock and surface increases along the plate. Also seen is that the boundary layer in the
flow direction thickens considerably with 6 ~ X~(K > 1). The interaction of the gas with the

hock

/ / I/ / / / / / / / / / / / / / / / /

-----Lines of constant pressure,ps/po ,103

streamlines

sonic (ine

~,.~
rE

shock

/ / / / / / / / / /
0 .

I IIII

)'II =/f/l)lll/l

CL

- - calculated

o experiment

.001-~

!0[

....

Po

o
I

axis of

o-----___.o....__..__
i

symmetry

'

1'2

'

I'~

'

2'0

x/rE

Fla. 30. Flow field of a conical nozzle exhausting cold air normal to a flat plate in a near vacuum ( z d r E = 38,
M a ~ = 2.94, e ~ = 15). Adapted from Ref. (74).

a)

plume

streamline

,o[E~--F'--T_~

--

~L:,/ / / l l / / / / / / / / /l/ / / / / ) ,/ / / / / /~/ / / / / / / / / / / / / / / / /./ / /.J / /. .


plate
50
ZN= 26.6mm

sho;k -/- --- -boundary layer edge i--

axis

100

~---

b)

T
150

mm

boundary layer e d g e s
(almost identical for

plume a~'~
I

O= 0-5=5

~'1'~4-~:-~
111
eL

~ -- o" : ) \

surface shock for

o_-~.

-~,

200

,o" \

50

F.~m

',,/,~H;~Z,H,~;HHHH~;HH)HHY
plate
50
100
150
200
mm
250

FIG. 31. Surface shock and boundary layer edge in impinging plume flow. Conical nozzle e = 25, OE = 10, test
gas N2, To = 290 K, po = 16 bar. (a) zN/r E = 10, 9 = 0;, (b) zN/r E = 5, variation of 9. Adapted from Ref. (127).

Plume flow and plume impingement in space technology

37

surface in this region is characterized by viscosity, while in the strong shock region with
subsonic flow near the stagnation point it is characterized by compression.
When the impinging flow is free molecular, the particles reach the surface undisturbed.
The gas has a certain bulk velocity u and the thermal motion for an observer moving with
u is Maxwellian. The Maxwellian distribution function is (drifting MaxweUian velocity
distribution)
/l

f = (2nkT/m)3/2 exp[ - m(v - u)2/2kT]

(37)

where v = u 4- c; e is the thermal velocity. When the molecules hit the surface there is an
exchange of energy and momentum with the body and then the molecules leave the surface
with a half MaxweUian distribution function corresponding to the wall temperature Tw. In
this case the molecules are said to have completely accommodated to the wall temperature
and a maximum possible energy dE: - dE w is transferred to the body ( d E , dEw: energy of
incoming and reflected molecules, respectively). If the accommodation is not complete the
degree is expressed by the thermal accommodation coefficient
dEi -- dER
ae = dEi - dEw

(38)

where dE R is the energy of the reflected molecules, ae = 1 denotes complete accommodation


dE R = dE w, and ae = 0 means no energy exchange (dE~ = dER). In a similar way accommodation coefficients for shear stress and normal momentum transfer are defined by
1~i -- ~R
Ti

(39)

P~ -- PR
P~ -- Pw

(40)

as - - (Zw = 0 for the reflected Maxwellian gas)


av =

For free molecule flow impingement exact formulae are available. For the continuum case
at least the approximate theory of Newtonian flow is available.
5.2. PLUME IMPINGEMENT FORCES

To determine forces in continuum flow impingement the Newtonian theory has often
been applied. Together with a local determination of the impingement effects it offers a way
out of the problems associated with the non-uniformity of the flow. Newton assumed fluid
to consist of particles which on impinging on a surface lose their momentum component
normal to the surface but keep the tangential one. Applying Newton's laws of mechanics,
the force exerted on the body can be determined. This theory has some relevance to
hypersonic flow, characterized by very high Mach numbers. The compression shock lies
close to the impinged body and the undisturbed plume flow reaches close to the surface.
Therefore the flow properties at the surface can be assumed to be the ones of the
undisturbed plume at that location. Thus, the problem of treating the disturbed flow is
avoided. The surface is then subdivided into sufficiently small plane elements so that the
flow in the stream tube reaching a certain element can be considered as uniform. In this
local treatment the problem is reduced to the interaction between each stream tube flow
(characterized by Ma and Re for example) with a small plane element under a given angle of
inclination. In the original form Newton's theory neglected the existence of a tangential flow
coming from neighbouring surface elements and interfering with the incoming stream tube.
The total impingement effect on the surface is obtained by adding up the contributions from
all surface elements.
The force F is equal to the rate of momentum change of the fluid; according to Newton's
theory only the flow momentum perpendicular to the surface need be considered.

38

G. DETTLEFF

Therefore the impingement pressure is


Ps = P u2 sin2~t

(41)

with = the angle between the streamline and surface (angle of attack). For the pressure
coefficient, defined as
P, - P~o
%=
pu2
(42)
one obtains
Cp = 2 sin2g

(43)

(Poo = 0 in Newton's theory). A so-called modified Newtonian pressure coefficient is


~+3

. 2
sin ~.

Cp. modN = ~

(44)

Early investigations of pressure forces on perpendicularly impinged surfaces (simulation


of moon landing, Fig. 8) gave 'excellent agreement' between experiment and application of
Newton's theory (Eastman and Radtke, (73) experimental results from Stitt(2)). Improvement for the theoretical prediction was obtained by taking the undisturbed flow (calculated
with the method of characteristics) at the surface shock rather than at the surface. For the
shock shape a suitable formula is for example proposed by Eastman and Bonnema. 74) In a
recent study ~75) the usefulness of Newton's theory for this case of impingement could again
be demonstrated, but for angles of attack smaller than about 45 considerable deviations
occur (Fig. 32, ~ < 45 corresponding to x / z N > 1).
The shock location and the influence of nozzle-surface distance and stagnation pressure
have been investigated by Vick and Andrews. 76) For fixed location they found a direct
proportionality Ps ~ Po.
Early investigations of pressure forces on adjacent surfaces were reported by Piesik
et al., ~72) Rochelle and Kooker, (77) and Vick and Andrews37s) Piesik et al. compare

/////"N/~////
ZN

o__,
Po
0.1

=l
OOlJ

-' I

-2

-1

X/Zs

~--

FIG. 32. Free jet impingement on a fiat plate (jet axis perpendicular to plate). Distance orifice-plate ZN/r* =
orifice radius r* = 1 mm; test gas N2, p o = 4 bar, T O = 293 K. Adapted from Ref. (75).

40,

Plume flowand plumeimpingementin space technology

39

experimental data with Newton's theory and with calculated results from the twodimensional shock theory. However, no conclusion is reached concerning a generalization
of the accuracy of this comparison. Vick and Andrews also showed in their experiments
with an adjacent plate that all curves P~ = f(x) with parameter Po and fixed Zs are identical,
Po
i.e. Ps ~ Po at a fixed position on the surface. Rochelle and Kooker find a considerable
underestimation by Newtonian theory compared to experimental data. This simple theory,
approved in the case of perpendicular impingement, obviously fails for conditions where the
surface shock is oblique and tends to withdraw from the plate and where a thick boundary
layer has developed.
A further attack on the impingement force problem was made about 15 yrs ago. In the
meantime rarefaction phenomena in the plume impingement problem had been recognized
as important and consequently Lengrand et al.(79~ examined if the prediction methods for
impingement forces based on continuum flow assumption can be applied to the rarefied
conditions. The prediction methods are again Newton's theory and a method based on the
work of Rochelle and Kooker~77~ (local oblique shock). A comparison with the earlier
results of Vick and AndrewstTa~ lead to 'reasonable agreement'. However, a comparison
with their own results obtained with a similar arrangement except for a lower stagnation
pressure revealed a considerable discrepancy. These experimental conditions simulated
situations typically found when satellite attitude control thrusters are fired. In an attempt to
take into account rarefaction effects, a closer approach to the experimental data was
obtained, indicating that the physical interpretation for the observed discrepancy to the
prediction based on Newton's theory led in the right direction (in a later numerical study~8~
with the DSMC method satisfactory agreement with the experimental results was found). A
part of the study dealt with the investigation of the shock structure discussed in Section 5.1,
especially with the position of the transition point between strong and weak shock regions
and the location of the pressure maximum. The transition point lies in all cases upstream
from the peak (see Fig. 29), i.e. the maximum itself belongs to the weak shock region. The
height of the peak could be shown to be a strong function of the nozzle exit Mach number:
an increase from M a r = 3.23 to M a r = 6.53 (other conditions unchanged) caused the
pressure peak to decrease by a factor of 30.
In Section 3 we have demonstrated that much effort has been devoted to obtain simple
plume flow descriptions containing the most important flow parameters. If the simple
formulae describing the forces (for example Eq. (43) or (44)) are combined with the plume
model equations, an insight into the relevance of the parameters determining the impingement effects should be possible. Work on these lines has been made by Lengrand et al. tsl~
Starting with the adoption of Simons' model and applying Newton's theory the surface
pressure becomes
ZN 4 f (O)~,
1
--Ps -(~--)

(45)

PS,ref

with
l = 0.613 I-(~' - 1)/2] 0.835 I-Olim/(rc/2)]1.92 ,

(46)

fr* ~2
P~,,el = P~-~N) 0.978[(y -- 1)/2] -aav ,

(47)

and
f() = cosz--21 ~-

(9)

It was found earlier that the agreement of Newton's theory with experimental data is not
satisfactory. But the shape of the corresponding two curves Ps/P,ef = f ( x / z s ) is very similar.
Therefore Eq. (45) is suitable for an assessment of the relevant parameters. We note that the

40

G. DETTLEFF

proportionality Ps ~ Po, already experimentally found by Vick and Andrews, (78) is reproduced. For fixed flow conditions (Olim, 7, and r* given) the impingement pressure along the
plate is solely a function of : if the plume axis is parallel to the plate, sin a = ZN and = .
r
Therefore sin4 f() determines the pressure distribution.
Another extensive study of plume impingement forces has been performed recently by
Boetteher and Legge(82) and Legge. (75' 83) Using the plume flow model (44) based on Simons'
and Boynton's work (Chapter 3), with Eq. (42)
Ps = cpl p u2,

and with Eq. (15) for p and Eq. (10) for u,


sin a = z N / r

(48)

= a + ~

one obtains

~o\T~/

(49)

+
= CpSln ~ApCOS~,-1 / . . . . .

\ O,/m 2 )

(50)

Newton's pressure coefficient cp (Eq. (43)) underestimates the experimental data. But
these data can be used to find empirically a modified coefficient
cp = - 7+1

0.6 +

sin2e

(51)

with 6' in radians, by a fitting procedure where the angular dependence cp(a) is retained as in
Eqs (43) and (44). The form of formula (51) is a modification of an expression for ep proposed
by Chiu e t al. (s4)
7+ 3~
6.88'~
cp - 7 + 1 \0.814.+ ~ . 8 ] sin2a
(52)
Results obtained with Eq. (51) are presented in Figs 32-34, where the cant angle ~ is the

,-.,..
/ / / / / / / / / / / / / /

50

-theot: cp=cp.modN(0.6+0.63/O.)
O
O O

N2, nozzle, r*=0.3rnm, E=15"


rE= 2.3?Smm,z~i~0mm , po=4 bor
To: 300 K
theor
3

10
.......

0.

Ps/z~ 2
1

\
O

0,1

,/

-1

,iifi1'~'

ta

,o ,

\
2

;...

,\,
3

x/z
FIG. 33. Nozzle plume impingement pressure x-profiles. Comparison of simple local approximation with experimental data at different cant angles ~. Adapted from Ref. (75).

Plume flow and plume impingement in space technology

41

~=___0--

ZN

plume:
Ap:l.77, O=;m=66./,/,"
re:2.66 mm r%-O.6mm, Oe=t0" , u= =0*
T0=285 K, Nz

impingement:

PoZN

Px
Po

t ~ r ] re
10-1

16

P= [ ZN'2

~'~,"~rxlllh

5133I0"~ \ ~

back flow
on plate

~.EM.

0 ~ I Tw/To=I

5 8.33.10 "t

o//

0.1

I :'-],-

\\

..~

L,, -1
X
ZN

OdN
"ewO
tn

....

~,lll \
10

FIG. 34. Experimental and theoretical plume impingement pressure, x-profiles. Adapted from Ref. (83).

curve parameter. We note that the single curve for a given configuration is again solely a
function of the angle of attack =, as is expressed in Eq. (50).
The comparison in Fig. 33 shows that the experimental values at the pressure peak are
greater than the theoretical values, while along the plate overestimation as well as
underestimation is observed. This is of course also due to the desired simplicity of the
expression (51) for the pressure coefficient, which does not distinguish between the weak and
strong shock region, for example, and which contains no consideration of the cant angle.
In Fig. 34 the quality of the proposed theoretical prediction is demonstrated with a
variation of the stagnation pressure, Po, and nozzle plate distance, zN. Also to be seen is the
improvement as compared with the modified Newtonian pressure coefficient
7+3
cp = - - sin2~ .
7+1

(44)

In the Newtonian theory the shear stress "c is zero, which is not true in reality. For the
shear stress coefficient c~ Leggeits) proposes for all angles of attack 0c and cant angles 0
2.6

c,

pu2 c"rU~/Re2 + 6.7

(53)

Further, some improvement at small cant angles O gave the equation


c, = c,,rM /

2.6

Re2

(54)

~ cos2=[ + 6.7
where c,, r u is the value for free molecule flow and

Re2 = p2u2r

/4(T2)

(55)

(Index 2 denotes conditions behind a normal shock at the surface location of consideration;
upstream shock conditions are given by the undisturbed plume flow at that location r, .)

42

G. DETTLEFF

Re 2 expresses the degree of rarefaction in this problem. The Knudsen number K n can be

written in the form


F-

K n = l"26x/;MaRe"

(56)

Thus the Reynolds number is some kind of a reciprocal Knudsen number. In fact, for large
Kn, respectively small Re 2, c, approaches C,,FM, (2.6 ~ X / ~ ) . On the other side, in the
continuum case, there is no restriction on Re 2 as long as the flow is laminar.
Combining Eq. (53) with Eq. (15) for the density p and Eq. (10) for the velocity u one
obtains
1

~(ZN~ 2

0tApf(O)~

CzSln

(57)

po\r* ]

The validity of this equation is demonstrated in Fig. 35, where the degree of flow rarefaction
2o
expressed by Kno = ~ has been varied; with increasing Knudsen number tlie curves tend to
approach the free molecule limit. The shape of the single curves very much resembles those
for the impingement pressure (compare with Figs 33, 34). This means that the dependence
2.7
z(0t) is still dominant (decreasing along the plate) while
increases along the
~/Re2 + 6.7
plate (mainly because the density decreases).
In Fig. 36 the cant angle has been turned to 45 (in Fig. 35: ~ = 0 ) and the x-scale has
been extended to x < 0. The single curve now seems t0 consist of two parts. The indicated
minimum of shear stress (exact: z = 0) defines the stagnation point of the flow on the plate.
The values of z on the left side of this point must be read with a negative sign (as indicated in
the figure) indicating that the shear stress vector has changed its direction. For continuum

~0~
zN/r ~

1
C =C~FM
2.6
,
~ReffcosZct~ 6.7'

/
10-1

impingement

EM

10-~

10 -3

0.16

40

~.19.10-s

377.10-, - - ~ . r - - ~ \

o ~ . t o ' :

~Mo-s-----~ -

1 , s ~ . 1 0 - ' ~

0,1

\\\
10

ZN

FIG. 35. Freejet impingementshear stress z on plate x-axis;jet axis parallel to plate. Kno = 2o/r*. Adaptedfrom
Ref.(83).

Plume flow and plume impingement in space technology

43

N1.Ae. ns63, e,....13(1~


zNIr%/.0 , t'=lmm
~ "/.5". Te=298K
theory pi[bar]

EM

0.1

= / ~ / ~
po ~

r'/

0.01

X/Z

m-

FiG. 36. Free jet impingement shear stress x-profiles. Influence of rarefaction by varying po. Adapted from
Ref. (75).

flow conditions the location of the stagnation point is between x = 0 (plane of nozzle exit)
and the location of the corresponding pressure peak (compare Fig. 28b). With increasing
rarefaction the stagnation point is shifted upstream on the plate until x = 0 in free molecule
flow impingement.
The results demonstrated in Fig. 35 have been obtained with a free jet from a sonic orifice.
The same experiment with a plume from a conical nozzle gave results which did not
compare satisfactorily with Eq. (57).
It has been shown by the experimental results that in continuum flow impingement the
shear stress is by a factor of about 10 smaller than the corresponding pressure. So far, to a
certain extent, Newton's theory approaches reality quite well. It must, however, be
emphasized that the fact that < p in plume impingement problems does not imply that the
shear stress can be neglected compared to the pressure, since not the forces as such but the
resulting moments are often posing the problems on spacecraft vehicles.
The relation (53) has been deduced using Reynolds' analogy. In its original form for
incompressible fluids this analogy leads to
St = c,

(58)

where St is the Stanton number (see later Section 5.3). The physical meaning is that heat and
tangential momentum are transferred by similar processes. In its modified form Reynolds'
analogy is
St = s \ 2 - ]

(59)

with s as the Reynolds' analogy factor. This factor should take into account turbulence,
compressibility or high speed flow effects not considered in the original form (58) and the
shape of the body. Relation (53) had been derived from heat transfer data (aS) (see later
Eq. (88)) and had been applied to the shear stress determination by making use of Reynolds'
analogy (we will return to this in Section 5.3).
Further, the relation (53) is a particular case of the general bridging procedure in
transition flow impingement, where the coefficients like cp and c, become functions of the
degree of rarefaction, expressed by K n or R e (the coefficients are fixed in continuum and free
molecule flow with different values which depend on various geometrical and flow

44

G. DETTLEFF

parameters). The bridging relation in a general form reads

c,(Re) = (1 -- f(Re))c,,c + f(Re)c~.rM

(60)

%(Kn) = (1 - f(Kn))%,c + f(Kn)%,r M.

(61)

or

Bridging therefore means a 'weighting' of the continuum value (cp,c for example) and the
corresponding free molecule value (%, FM) by the function f(Kn), which is in general deduced
from experimental data and can depend strongly on flow characteristics and geometry. A
certain subtle feeling and experience is therefore required to deal with transition flow
impingement problems. Examples of bridging functions can be found in Refs (86) and (87).
More details of bridging will be given in the next section.
In free molecule flow impingement exact formulae can be derived for the pressure and
shear stress exerted on non-concave surfaces (non-concave, since otherwise a molecule
could hit the surface several times). The force dF on a surface element dA is (3)

dF = dF i + dFa

(62)

where index i denotes incoming and index R reflected molecules. Specification yields
(complete accommodation)
dF
dFi dFR
d--A-= d A + ~
= p~ + z, + PR

(63)

Pi + Pa = P

(64a)

~i = z .

(64b)

For the quantities in these two latter equations exact formulae can be derived. As an
example we write (Schaaf and Chambr6 ~3~) for complete accommodation
1 2 ~1 L
[ -~/ 'S s i1n c t
p =~pu

+ 1 //~-~'~e_lS.in~)2
2X/ T ]

+{[(Ssina)2+]+~XIrt~(Ssina)}[l+erf(Ssina)] 1 .

(65)

Here we have introduced the molecular speed ratio

S=~

Ma

(66)

which relates the gas velocity u to the most probable molecule speed ~
(free molecule
gas is not 'elastic;' sound waves will not occur and therefore the Mach number u/a, which
can formally still be calculated, is physically meaningless).
For S sin a ~> 1, i.e. if the angle of attack is not too small, an approximation of Eq. (65)
yields
cp = cp.~ +

x/nTw sin
T S

(67)

with c~,~ = 2 sin%t ( = %,c, Eq. (43)). Therefore Cp,FM> %,c, and the FM curve in Figs 32
and 34 is higher than the continuum curve, and the contribution of the reemitted molecules
(expressed by PR) is responsible for the difference. The region between the two curves
represents the transition flow impingement with
%,c < %,t, < %, rM

(68)

The theoretical approach to the determination of impingement forces has mainly been
based on the local methods presented so far, and the local measurements were in most cases
performed along the projection line of the plume axis on the plate, in some cases also

Plume flow and plume impingement in space technology

45

perpendicular to this line. The final goal of all these efforts is of course to predict the overall
effect of the impingement forces, i.e. the total moment they produce. Instead of making local
measurements Allrgre eta/. (Ss) have determined the total force (pressure and shear stress) on
a plate. Parameters were the inclination angle (cant angle) and the distance between nozzle
and plate. Using again the local Newtonian theory for the prediction of normal forces, they
found also that the total normal force component was much underestimated. Therefore the
results discussed before considering only one line on the plate are in fact representative for
the overall discrepancy. But satisfactory estimates were obtained for the tangential forces.
Allrgre et aL (88) also investigated an impingement configuration where the plate has been
shifted along its plane downstream so that the flow around the leading edge with its
peculiarities came into play. It turned out that the bluntness of the leading edge affected the
tangential force but hardly the pressure force.
The plate with the leading edge effect is a simple example of a configuration for which the
available theoretical prediction methods were not intended. In such cases experimental
investigations are necessary either to make exploratory measurements which help to
approach the problem theoretically or to confirm that a theoretical approach is reasonable.
One such situation is the stage separation sketched in Fig. 8. The geometrical configurations and the problems encountered are quite different. In the first case (Fig. 8f) the
impinged lower stage lies in the near-axis plume flow. The particular problem is the
establishment of a cavity flow which by escaping may affect the upper stage. Su and
Mullen (89) present a simple theoretical treatment of this configuration and they find good
agreement with experimental data. In the second case (All~gre e t a / . (9)) the impingement
effects caused by retro-rocket firing are treated (Fig. 8c). The configuration has been scaled
down 1/36 and measurements of density in the flow field and of wall pressure and heat
transfer have been performed. The unknown real flight data were estimated in the following
way. With a model calculation the wall pressure is determined first for the real and then for
the scaled configuration, and the ratio is formed. The experimental value is then multiplied
with this ratio to obtain the expected real flight wall pressure. The ratio was 40.9 which is
different from the geometric scaling factor.
Particular problems also arise when the impinged surface is close to the nozzle. Such
situations occur during rendezvous manoeuvres and during the reverse separation
manoeuvres, but also on adjacent surfaces close to the thruster. One problem is the shock
system established in a contoured nozzle and extending into the plume (Fig. 25). If such a
shock intersects a surface a peak of pressure, and also of heat load, must be expected
(Stitt, {2) Leng et a/.(91)). These effects are very difficult to predict quantitatively.
Another problem especially on perpendicularly impinged configurations is the flow
escaping after impingement from the gap between the chaser and target (Fig. 8). This flow
may get in contact with the thruster vehicle and also influence its motion. Schlieren
photographs of the gap flow and some experimental pressure data measured on the chaser
are presented in Ref. (20). This example is one case of pre-impingement where the gas after
impinging on a first surface hits a second one. Such situations can especially occur during
attitude control manoeuvres on satellites (Fig. 8). However, there are not many investigations in the literature (an example is given by Rochelle and Kooker(77)).
Impingement forces cause undesired motions of spacecraft and therefore additional fuel
consumption for compensation. Impingement heat load can cause damage or destruction of
surfaces if a critical value is exceeded. Reynolds analogy already gives a hint where severe
problems of heat transfer may be expected. This will be examined in the next section.
5.3. PLUME IMPINGEMENT HEAT TRANSFER

The basic equation describing the heat transfer from a flow to a body is
t~ = h(T, - Tw)

(69)

which relates the heat flux ~ to the driving temperature difference ( T , - Tw). T~ is the
recovery temperature of the surface which it takes in thermal equilibrium (q -- 0). It is

46

G. DETTLEFF

expressed in non-dimensional form by the recovery factor


r,-T~
r' = - -

To-T

(70)

and depends on the Prandtl number


Pr=

k'

(71)

with c~ the specific heat at constant pressure, k' the coefficient of thermal conductivity. For
example, for a plate in continuum uniform gas flow with a laminar boundary layer it is
r'= x/~.

(72)

For nitrogen we have Pr = 0.72 and r' = 0.85 and for hydrogen Pr = 0.67 and r' = 0.81.
Therefore T, < To in these cases, i.e. the equilibrium temperature T, is smaller than the
stagnation temperature To.
The heat transfer coefficient h in dimensionless form can be expressed by
St -

puc'~

(73)

(St is the Stanton number), where p, u, c~, are the density, velocity and specific heat

respectively at the edge of the boundary layer, and


St = f(Nu, Re, Pr)

(74)

h l,es
where N u = - - ~ - is the Nusselt number. The wall temperature Tw together with T,

determines the heat flux q; but Eq. (69) does not say how Tw changes during the
impingement and how it adjusts in interaction with the radiation absorbing and emitting
surroundings. This, together with the heat conduction in the body, depends on the material
properties and is not a subject of this article. Nevertheless Twis the quantity of interest, since
a critical value must not be reached or exceeded. This value for a given surface depends on
the local heat flux due to impingement, which will be treated here.
Reynolds' analogy (Eq. 58) relates c, and St to each other, and therefore similar
distributions of ~(x) and ~(x) as in Figs 35 and 36 can be expected. However, it must not be
inferred that ~ = 0 if T = 0. From the considerations in the foregoing Sections, 5.1 and 5.2,
we know that there are two types of impingement flow: the strong shock region with
compression and minor shear stress and the weak shock region with dominant shear stress.
Along a plate as in Fig. 28 we have to treat both types when dealing with heat transfer
problems: the first one where thermal energy is primarily transferred from the compressed
gas and the second one where Reynolds' analogy is expected to hold.
Exact or approximate heat transfer formulae exist for certain configurations, for example
the flat plate or axisymmetric blunt or slender bodies in uniform flow (see for example
Ref. (92)). The non-uniformity of the flow is again the problem which is encountered during
plume impingement. Therefore a local treatment of the problem must be tried again. Then it
is to be decided if for a certain location on the surface a blunt or slender body solution shall
be applied (note that for a given configuration, for example the plate in Fig. 28 the angle of
attack can vary in a wide range between 90 and ,~ 0). If both are to be applied matching is
necessary and can be problematic.
An early systematic study on plume impingement heat transfer was performed by
Lehrer. t93~ A first order approach was to determine locally the stagnation point heat
transfer on a small blunt body thought to be immersed in the plume at given points of
interest. If the stream tube reaching the immersed body has a sufficiently small cross section
its flow can be considered as uniform. For a symmetric blunt body in such a flow the
stagnation point heat transfer is a maximum. Therefore this approach represents the worst
case. For the laminar stagnation point heat transfer on a sphere as a blunt body Lehrer{93}

Plume flow and plume impingement in space technology

47

used the equation


Nuds

1.32 ~

(75)

Pr '4

or

hE ~ k---Z]

\T-]

Assuming in the first order approach that k', u, #, Cp are constant, d, is fixed, one obtains

h~

(76)

and with ~ ~ constant throughout the (continuum) flow, and Tw fixed, it follows from
Eq. (69)
4 '~ x/~.

(77)

For the heat transfer on adjacent surfaces Lehrer uses the equation for laminar flow
conditions along flat plates
Nu = 0.332 Pr 1/3 R e a/z
(78)
The comparison of some experimental with predicted data is denoted as good.
An experimental investigation with a small sphere immersed in the flow as described
above has been performed recently by Legge and Detfleff~94)in a hydrazine thruster plume.
The temperature, Tw, of the probe was measured as a function of time (Fig. 37a) and the
resulting heat transfer as a function of T~ is shown in Fig. 37b (Fig. 37a is taken from a pulse
mode firing, Fig. 37b from steady state firing). The linear relation 4 ~ T~ from Eq. (69) is
clearly reproduced, and the linear extrapolation of the experimental data 4(T~) to 4 = 0
gives by definition the recovery temperature T,. T, depends especially on the stagnation
pressure po, because the decomposition of hydrazine is influenced by po.
For the subsonic flow region along the adjacent plate Piesik et al. ~Tz) used the formula of
Kemp and Riddell~95) for stagnation point heating. The form of the equation, originaUy
obtained for reentering satellites, considered as a blunt spherical body, has been retained,
but a term was changed to account for the particular configuration,
4=

15200

( p ~1/2( U ~3"25(1

e + 0.75z~kP'ef/
re

\u,ef/

Hw~"
1"1/

(79)

The nozzle area ratio e and the nozzle plate distance zN/rE have been introduced. Note that
relation (77)
4~,/p
is found again in Eq. (79). Besides the geometrical quantities, 4 depends on the flow
properties p, velocity u and temperature T(H = CpT). The wall temperature is expressed by
1

Hw. The heat transfer equation for the oblique shock region contains the term - - ,

where

x, is the distance on the plate from a reference point (location between normal and oblique
shock flow). In the vicinity of this point (x, ~ 0) according to

4~

(80)

we would obtain 4 ~ ~ . Therefore a continuous matching of the two theoretical solutions


is not possible. The results of the heat flux 4 obtained from calculation and experiment agree
quite wall, also with variation of the cant angle. A strong influence is shown for the nozzle
area ratio ~ (and therefore for the nozzle exit Mach number, similar to the influence on the
pressure distribution demonstrated by Lengrand et al. t79)) and for the distance zs/re, as can
already be expected from Eq. (79). With increasing zn/r~ the peak is shifted downstream.
Roughly, the peak location on the plate varies with zs/rE.

48

G. DETrLEFF

a)
K V

~'~"~Jtpulse: 71

58O-T

!on.=

/ ~

I/

to(~

~.

48CJ
O.5N t h r u s t e r

47C

pul~= mode: 1oe=03s, tofr=27s, X/re=t6

Tw
460
T

l = l l t l l n l a l = l *

At

b)

1.5.1~N ~
0.5N thruster, steady stQte
x/r E
p0[ bar]
o 32
15.6
/.0
15.8
32
7.95
L56
32

1.0.

x,%
o.5.1o-

"~%
+ "~.~+.,~

0
250

I
500

I
750
Tw

~r
1000

1250 K 1500

----

FIG. 37. (a) Exampleof probe temperaturerecordingfor pulsemodefiringof a 0.5 N hydrazinethruster. Adapted
from Ref.(15). (b) Aerodynamicheat transfer on a sphere probe and recoverytemperature determinationfor
differentstagnationconditionsand probe distances.AdaptedfromRef.(15).

Empirical factors are used in the theory (as for example in Eq. (79)) and the matching of
the two different heat transfer equations is problematic. This has been found to be a
deficiency which is the starting point of Maddox's theory.(96) Maddox assumed ideally that
in continuum flow the stagnation point on the impinged surface coincides with the location
of the pressure peak (see for example Fig. 29). Then the pressure distribution on the plate
looks qualitatively like the one on an asymmetrical blunt body in uniform flow. Under this
assumption the plume impingement effects were calculated by means of equations for blunt
body heat transfer in uniform flow with local similarity. Maddox used the equation
1

h = Kc'pPr- '6(p,ef#,f u/x)~

(81)

with K = 0.57 for plane stagnation flow and K = 0.763 for rotational symmetric stagnation flow. Equation (81) was originally derived for laminar flow with constant gas
properties, but can be applied to varying properties p and #, when these are taken at a
certain reference temperature T , f (Eckert(92)). Comparison with experimental data of
Piesik gave the best agreement, while differences were observed in the comparison with
Lehrer's results. But here the assumptions in the theory were not completely met, for
example Tw <~ To.

Plume flowand plume impingement in space technology

49

In preparing a more general validation of Maddox's theory by experiments, Lengrand


et al. t97) have put the heat transfer Eq. (81) into a non-dimensional form 4~4,el with
(~,ef =

poUnmRe-l~2

\ZN/

(82)

C'p(To -- Tw)C -'5

with Re = Re* r* and C is the Chapman-Rubesin constant. The reference heat flux thus
ZN
depends mainly on the stagnation conditions Po, To(uum~ V/-~o) and the geometrical
arrangement, characterized by r*/zN. In the non-dimensional form q/4,,f the number of
parameters is considerably reduced and the heat flux depends on 7, Olimand a and is nearly
independent of Pr and the temperature level (a similar procedure for the impingement
pressure led to Eqs (45) and (50)). The experimental conditions were selected typical for
ZN
microthrusters, and the pa/ameters were Po, To, OE, Maz, 7 " Results for various zn are
shown in Fig. 38. Experimental and theoretical (Maddox) values exhibit a small dependence
on zn. The maximum heat flux is predicted satisfactorily, and on the whole a qualitative
agreement for the distribution is found. The comparison with the theoretical curve obtained
with the method of Piesik et al. (72) is not as good.
The experimental conditions correspond to the transition flow regime along the plate, but
no indication of rarefaction is found. In free molecule flow impingement the calculated heat
flux should be a factor 3-10 times greater than the continuum values shown in Fig. 38. The
factor depends o n ZN/r*, since continuum heat transfer decreases with (zN/r*)- t.s, while the
corresponding free molecule heat flux decreases with (zn/r*) -2.
In their experiments I.~ngrand et al. (97) have investigated another important case of
impingement. If the plate is shifted such that the kading edge is downstream of the nozzle, a
tremendous increase of the maximum heat flux is observed. For the configuration shown in
Fig. 39 the peak value for th e blunt plate is a factor of about 5 greater than for the infinite
plate, and the heat flux distribution further downstream is also affected.
The considerations so far were for the continuum flow impingement. In free molecule
flow again, as for the forces, exact formulae can be derived. (3) The heat flux 4 is given by

RTV ~R~T([
7
- ~ ' $2 + ~'----I

4 = (TeP

7 + 1 i)
2(~=

~] {e-(Ssin=)2

+ x/~-(S sin a) [1 + erf(S sin a)] } - e-(S'in=)2~.


/

Exp.

~
qref

~,

Piesik e

(83)

ZNlr"
1o

2o

4o
60
80

I00

" t ~ a ~

Maddox120~
lao /

-~

Io"
0

&

;o

x/z N

FIG. 38. Reducedconvectiveheat flux vs. reduceddistance along the plate axis of symmetryfor plate parallel to
the plume axis. Conical nozzle OE = 10 , Ma~ = 4.63 for test gas N 2, po = 4 bar, To = 1100 K; parameter zs/r*.
Adapted from Ref.(97).
JPAS 28:1-D

50

G. DETTLEFF
1-

]unt plate

~il~ate)
10
x/z N

FIG. 39. Reduced convective heat flux vs. reduced distance along the plate axis of symmetry for plate parallel to
the plume axis. Leading edge inside and outside of the jet. Conical nozzle OE = 10, M a e - 4.63 for test gas N2,
po = 4 bar, To = 1100 K. Adapted from Ref. (97).

A considerable simplification of this equation is possible if S sin ~ >> 1 (i.e. if the angles of
attack are not too small)
~ ae P-2u a/'l
~

72V
+lTw)Too sin~.

(84)

The Stanton number in free molecule flow Stem for a flat plate is (both plate sides thermally
insulated)
Strn = ae y + 1

~-

{e-(s sin~)2

4x/~ S

N//~(S sin ~t)[ 1


+

erf(S sin at)] }

(85)

and the recovery factor in this case is


rru' = _ _+y
_ _1 2S21
' {2S2 + 1 - {e -(s'i"=)2 + x/~(Ssin~t)[1 + eff(Ssin00]}}

(86)

Again, if S sin ~ >> 1, one obtains


r~ ~ 2

7
7+1

(87)

since 7 > 1
i

r~M > i.

The recovery factor is greater than 1, i.e. the equilibrium temperature of the impinged
body is greater than the stagnation temperature. This is a remarkable feature of free
molecule impingement, where no energy is required to recompress the gas in front of the
body as in continuum flow.
In the transition regime between continuum and free molecule flow some bridging as for
the forces (Section 5.2) is necessary. Boettcher (sS) has used measurements of Legge and
Dankert (9s) on cones with various angles (Fig. 40) to derive empirically
2.6

St = S t ~ x / R e 2 + 6.7

(88)

Plume flow and plume impingement in space technology

51

a)
St FM

St

0.5

bridging
' function

0.4
0.31

Nz ,Mo-=9 + 25
@:15"+90

range of J
experimental
data

0.2

100

10t
Re 2

b)

1,3
r'.FM

1.2

Re,~2
-'~w~

range of
experimental

/ data

1.1
t

bridging
function

1.0

r' 0.9
r'

0.8

Iref

0.7

Q~O-2

10-1

N2, Mo= 10
~=15.90

10o
Rez

101

102

---

FIG. 40. (a) Stanton number as function of the Reynolds number;,experimental results for varying cone angles;
p u I,f
Rez =
. Adapted from Ref. (85). (b) Recovery factor as function of the Reynolds number; experimental results
P
p U Iref
for varying cone angles; R e 2 =
. Adapted from Ref. (85).
P

and
r'

'

r'ru

--

r'c

= rFM -- lgl00 -- lg2 (lg Re2

-- lg2)

(89)

where rim and r'c are the recovery factors for free molecule and continuum flow, respectively.
In a recent experimental study Dgring (99) showed that the bridging relation (88) also holds
for transition flow impingement on an adjacent plate. For various conditions the estimated
error does not exceed 50%. An example is shown in Fig. 41 together with the recovery
temperature determination (note that 7", is a local quantity).
Equation (88) has been derived as a bridging relation from experimental data. With
c~ ,~ S t

according to the Reynolds analogy, Eq. (53), Section 5.2 had been established.
In this section about plume impingement effects we find that simple theoretical ideas like
the Newtonian approach together with the evaluation of experimental results lead to a
reasonable prediction method. The description of the impingement effects was preferably
deduced with the assumption of an undisturbed plume. The difficult task of determining the
properties of the disturbed flow, however, may appear again as soon as a more accurate
prediction is required. Moreover it seems that it must be attacked in the context of
simultaneous firing of two thrusters, which will be treated in the next section.

52

G. DETTLEFF

a)

0.1

Nl

0.09

S: =S~
tem -

t26

0.08

S:60"
Pk

pe= t,~ r
~:
6 '10"zmbar

FN~

z.I

0.07

r==asoK
Tw:30OK

" \

0.06

ptate

St

0.05

0.04

0 03
0.02

0.01
f

-0.2

-0.5

0.1

0.4

0.7

1.3

1.6

1.9

2.2

2~5

ZN

1.2

b)

1.16

' PO

1.12

N2
S :60"

d~

ZN: LO m m
Po =/" bar

i .09
*
Tr
m
To

Z,,[

r'~N.~

pK= 6.lO'Zmbar
To=SSOK

9..04
C

ptete

0.96

0.92
0.99
0.84

O.fl
-O. 5

FIG. 41. F r e e j e t i m p i n g e m e n t

-0.2

0.1

0.4

0.7

1.3

~.6

1,9

2.2

on

x
ZN

2.5

a fiat plate. (a) S t a n t o n n u m b e r x-profile; (b) C o r r e s p o n d i n g


t e m p e r a t u r e x-profile. A d a p t e d f r o m Ref. (99).

recovery

6. INTERACTION OF TWO PLUMES


During some manoeuvres in space it is necessary to fire two thrusters simultaneously.
Two such situations are sketched in Fig. 8 (rendezvous and docking and north-south
station keeping). The particular feature is that it involves an interaction between the plumes
before impingement and the task is to determine the resulting plume flow.
The problem of the interaction is known and has been treated extensively in the context
of rocket plumes in atmospheric surroundings. In vacuum surroundings rarefaction effects
have to be considered in addition. Moreover the whole plume looks different, since no
barrel shock system exists (this will be explained in the next section). The possible plume

Plume flowand plume impingement in space technology

53

o -

--

__

'

thick Shock

tnteroct~n

plane

plane
mixin 9 due tO
incipient penetration

--

--

--.,gh

;,urn, 2

Incipient plume penetration regime (Knp,cl)

Free plume penetrotion regime (Knp>>l)

plume 1

1
2

Shock wave

tnterac tio~
ptane

--

plane

Plume 1 molecules
collide with
' Pturnl 2 molecules
plume 2

Disturbecl plume I:~netroti~ regime (Knp=l)

Shock interoction regime (Knp<<l)

Fzo. 42. Definition of interaction regimes of two plumes. Adapted from Ref. (103).

interference flow regimes have been defined by Koppenwallner (1) and are sketched in
Fig. 42. A configuration of two equal thrusters with parallel plume axes is considered. In
between is a plane of symmetry, called the interaction plane. In the first case (free
penetration regime) the flow is so rarefied everywhere that an undisturbed mutual penetration is possible. If the stagnation pressure is increased we obtain first collisions between
molecules of the two plumes (disturbed penetration regime). Further increase of the density
leads to the appearance of two thick shock waves which are symmetric to the interaction
plane. In this incipient penetration regime the flow upstream from the shocks is undisturbed. Finally we have the shock interaction regime with continuum flow features.
A stagnation point exists and consequently back flow in the interaction region will appear
(the back flow heating is a main problem during rocket multi-thruster firing). In this case no
gas of a plume can (ideally) pass the interaction plane.
If the interaction plane is replaced by a flat plate we have a similar disturbed flow field
except for the existence of the boundary layer (compare with Fig. 3). As in this case of the
impinging flow a theoretical description or at least a simple modelling of the flow in the
interaction region is missing. A few experimental investigations gave a first insight into
the features of the flow produced by two interacting plumes.
Due to the similarity the distribution of the flow quantities along the interaction plane
resembles very much the one along the plate since the governing parameters are the
same,t1 o 1) namely the Math number Mar, ratio of specific heats ~, stagnation conditions po
and To and linear dimensions Zs and d*. Ermolov et al. " ~ have investigated the density
distribution along the interaction plane. Expressing the density in the non-dimensional
form
~\~-~] = f ~

(90)

we find that its distribution very much resembles that of the pressure p, in Fig. 34. With
increasing Reynolds number Re* (Eq. (28)) the non-dimensional density increases and the
maximum of the curve is shifted upstream. Thus viscosity has a considerable influence on
the flow pattern in the interaction region (in the similar case of the impinging flow in front of
a plate the role of viscosity is evident by the establishment of a boundary layer).

54

G. DETTLEFF

S.10"3t

I'Y'-----O--

~,.lff
~"'~"k
3.10":
P,=._JL
Po
2.10":

110": ~ Po=300 mbar Knp=0,02


~--- po= 75rnbar Knp=0.00
o...... p0 = 30 mb(:lr Knp=0.18
0 ~ -10

-5

10

FIG. 43. Interaction of two free jets. Transverse profiles measured with a free molecular pressure probe.
Orientation of probe orifice,see insert. Test gas N2, x/d* = 12, d* = 2 ram,ZN = 20 mm. Adaptedfrom Ref.(103).

The structure of the interaction region has also been investigated experimentally by Soga
et al.(12) They found that the shape of the two shocks in the vicinity of the plane formed by
the two plume axes is wedge-like, but that the flow in between exhibits no two-dimensional
features. (These authors call the interaction region 'secondary jet'. It has, however, no source
point, but a source area since it is fed by the gas crossing the shocks.)
Another investigation of the interaction region is reported by Dankert and
Koppenwallner. (13) The pressure distributions shown in Fig. 43 reflect the different flow
regimes according to Fig. 42. In continuum the shocks are deafly recognizable by the two
peaks. They broaden more and more when in this experiment the stagnation pressure is
reduced, until finally a mutual penetration of molecules from one jet into the other is
possible.
As noted before a theoretical description of the interaction flow is not available. For an
assessment of plume impingement effects in this region it is already valuable to know which
of the interaction regimes will be met in a particular problem. The different situations
sketched in Fig. 42 resemble very much those in Fig. 27, which are distinguished by different
values of the Knudsen number Kn = 2/l,e:. This parameter has been introduced to the
problem of the interaction of two plumes by Koppenwallner/l) According to the sketch in
Fig. 44 the mean free path 2p considered is the one of the molecules on a certain streamline
at the intersection with the interaction plane. Then the so-called penetration Knudsen
number Knp is defined by
K%

2~
lref

(91)

where l,e: is the distance along the streamline to the axis of the other plume. The reference
length l,ey is
ZN

l,e: -- sin
and 2p ,,~ 1, where p is the density on the streamline at the interaction plane. It can be
P
expressed by the equations of the Simons model (Eq. (16)), so that finally the formula
1 Kno ZN
1
1
Knp(O) = 2 Ap r* sin2Of(O)

(92)

Plume flow and plume impingement in space technology

55

2r m

,o,..o.,,oo

Penetration Knudsen number


~kp

kp
Iref

Knp =

Mean free path of plume 1 particle moving


through plume 2 flow field particles
{ Collision - free penetration distance)

I,.f

Distance

from interaction

2 centerlines measured
h.f -- zs
sin O

plane to plume

along particle path

FIG. 44. Definition of the penetration Knudsen number Knp in rarefied plume interaction. Adapted from
Reg. (103).

is formed, with

~o
Kno r*
p*

,4; = A,~.
Knp for a given configuration thus depends on the penetration function

1
P(O) = 2 s i n 2 f ( O ) .

K.p(e) A;
=

Kno

.~

zN/r*)

(93)

which is shown for a certain configuration in Fig. 45. It is reasonable to define the minimum
1000

--

y
Omax Ag
- - 1.67
78.2" 0.6
1.40
95.2" 0.3/*5
..... 1.286 108.17" 0.237

,.

I
I

/ .-

/
I

P l e ) ~ c
O"

values for interact~


50"

100"

Fla. 45. Penetration function P(O) for the streamlines of intersecting jets. Adapted from Ref. (100).

56

G. DETTLEFF

value of the function P(O)min as characteristic for the interaction; it indicates the most dense
region along the interaction plane (a most rarefied region with P() ---, ~ always exists for
O = 0 and = ~im). From experiments it was found c13) that Knp.mt, < 0.02 indicates
interaction with continuum flow features (shocks) while for Knp.mt, > 2 a mutual free
molecule penetration of two plumes is possible. We find that the governing parameters in
this problem mentioned above (~, Mar, po, To, zN, d*) are directly or indirectly represented
in Eq. (92).
These considerations on the interaction of two plumes in vacuum surroundings close this
treatment of plume flow and impingement problems.

7. TEST FACILITIES AND EXPERIMENTAL T E C H N I Q U E S


In the preceding sections results from experimental investigations have been presented.
Without going into too many details some measurement techniques for the determination
of mass flux (density and velocity), forces and heat transfer will now be presented together
with some useful visualization techniques.
Moreover a mass spectrometer analysis(1~' lo4~ has been used to analyze the exhaust gas
composition either to determine the gas properties 7, #, R or to estimate possible
contamination effects. This also included the effects of mass separation or transverse
gradients of the composition cl6>due to non-uniform conditions in the combustion chamber
and in the nozzle flow. The goal of the experiments is the validation of the theoretical
models or to supply a data base or exploratory results for a theoretical approach.
For many purposes it is sufficient to perform experiments with inert test gases depending
on the aim. To check the validity of Newton's theory for example (impingement pressure) it
is not necessary to fire an original thruster. Here it is essential that the density and velocity
of the impinging flow are known. Nitrogen is very often used as the test gas. It is easy to
handle, represents a main constituent of real thruster exhaust gases and its ratio of specific
heats ~ is close to that for original thrusters. As long as an elevated stagnation temperature
near real conditions is irrelevant for the particular investigation, the test gas can be used at
room temperature. This no longer holds if, for example, a real thruster nozzle flow is
essential for the problem to be considered, where the nozzle wall temperature sensitively
influences the boundary layer development (Section 4).
In those cases where an original thruster is not used, the rules of similarity in fluid
mechanics must be observed. These rules allow direct transformation of the result of one
configuration to another geometrically similar one if the similarity parameters are the same
in both cases. Important flow similarity parameters are the Reynolds number, Mach
number and Knudsen number, and important parameters for impingement effects are the
Knudsen number and the Reynolds number.
Complete similarity cannot be achieved in some cases. This is demonstrated by
considering the Reynolds number similarity in nozzle flow

pEuElr
Rer = - -

(18)

#o
Expressing Pr and uE by the equations for isentropic flow one obtains 5:J
y+l

ReE = lrPo1

Mar( 1 +

Ma 2

2t~ x,.

(94)

If the ratio of specific heats ~ is allowed to vary, the Mach number Ma r can only be
maintained if the expansion ratio e is changed, i.e. if the geometrical similarity is given up.
This example demonstrates that simultaneous similarity of Rer and Mar cannot be
obtained, if all other parameters ~, #, R etc. are allowed to be selected freely. Similarity in
Rer can only be achieved, if the gas is the same (identical ~). Otherwise only a simulation
of one parameter (Rer) is possible, while the other (Mar) is not identical. Detailed

Plume flow and plume impingement in space technology

57

considerations on similarity and scaling laws in plume flow and impingement problems can
be found in Refs (93), (106) and (107).
To allow the expansion of the plume gas the vacuum surroundings of space must be
simulated in a ground test facility. It is no technical problem to achieve a vacuum quality of
10 -6 mbar as long as there is only a very small mass flow into the chamber. A typical
attitude control thruster, however, produces about 1 g/see gaseous exhaust products. The
resulting background pressure p~ in the chamber can be determined by the formula
.RT
p~ = m /~

(95)

where V is the pumping speed. For the plume flow this background gas represents an
obstacle, and since the flow is supersonic a compression shock system, consisting of a barrel
shock and Mach disk, is formed and terminates the free expansion. The resulting plume
shape which is similar to the one of the rockets in atmospheric surroundings, is shown in
Fig. 46.
The flow upstream of the barrel shock is practically unaffected by the existence of the
background gas, i.e. the flow is identical to the one in space within the region bounded by
the dotted line in Fig. 46a (a slight influence of the chamber background pressure on the exit
plane center line Math number M a E has been reported by Bailey et al.(~s)).
The size of the shock system and hence the plume or free jet size used in experiments has
been investigated in numerous studies. (l9-1 z3) A first approximation also for plumes is the

a)
border
st reom

[in-"~-"

t
prope[Iont
~ u

n;:;otrisCi~~

b)
iet" .':: vacuum~
"oxls " : pu'mp

propettnnt~ ,

: ':: '. :.' :. ' : ' : '

C)
Px=10"Imba~'~

propeltont.-~,

:'I) ~ ";ID'tmbor"
.....

LHe- cryopump

xl \
.....

"~

P~

FIG. 46. Plume expansion. (a) In space; (b) in vacuum chamber with mechanical pump; (c) in vacuum chamber
with cryogenic pump.

58

G. DETTLEFF

formula of Ashkenas and Sherman "12)


= 0.67

(96)

which was found for free jets emanating from sonic orifices with a diameter of d* and which
expresses the M a t h disk location xM as a function of Po and Pk (easy to measure). It
represents the general result Moran "18) derived:
x u ~ = f(~, Mr, OE . . . . )
dE "~ Po

(97)

For a first guess of the maximum plume diameter a value x u / 2 can be assumed.
The existence of a bounding compression shock system means that the expansion of the
plume flow is terminated in the continuum regime. To achieve rarefied flow conditions, the
background gas pressure Pk must be reduced by increasing the pumping speed I? according
to Eq. (95). Then the shock system is shifted downstream (Eq. (96)). In the transition regime
it will thicken more and more and a penetration of background gas into the expanding
plume flow becomes possible, (114-116) i.e. there is a gradual transition from undisturbed to
disturbed expansion flow. The grade of penetration must be known if measurements are
performed in this flow region. The drag on a flat plate, for example, is proportional to the
background pressure. "17) Such results underline the necessity to examine very carefully the
background gas for experiments in the rarefied regime.
Since large pumping capabilities require a large investment, attempts have been made to
gain access to the rarefied flow regime by means of a pulsed plume source technique. In this
case a steady plume flow is allowed to expand into the vacuum chamber only for a few msec
after the chamber has been evacuated to a very low pressure, Pk = 1 0 - 6 mbar for example.
Then the background pressure increases roughly according to
dpk
1
= v~kmRT,
dt

(98)

As an illustrative example we assume a chamber volume VR = 2 m 3, R = 700 Ws/kgK


(hydrazine exhaust), th = 10-3 kg/sec, T = 300 K. Within 5 msec we would achieve Pk ~" 5
X 10- a mbar (to maintain this pressure during continuous firing, a relatively large pumping
speed ~" = 420 ma/sec would be necessary). To produce a short duration plume flow Long
et al. "18) have used the gas behind a reflected shock in a shock tube as the plume source.
The same apparatus has been used by Calia and Brook (37) for their investigations in
the boundary layer expansion region (Section 3). In addition they have subdivided the
vacuum facility into two sections, one taking the main near axis mass flow (not to be
investigated) and a second into which the outer plume gas expands (to be investigated).
According to Eq. (98) the increase of Pk is thus reduced.
Another way to gain access to the more rarefied regime (with a given pumping capability)
is a reduction of the stagnation pressure po and hence of the mass flow. In this case the rules
of similarity must be checked, depending on the particular problem.
An undisturbed plume expansion into the free molecule flow regime (compare Fig. 2) is
only possible if the gas is pumped instantaneously after the expansion and is not allowed to
penetrate into the plume. This can be achieved by cryogenic pumps (Fig. 46). They consist of
cold walls where the exhaust gas condenses in the solid state, i.e. the walls must have a
temperature at least below the triple point of the gas component (see Fig. 19). But in
addition the temperature must be so low that the vapour pressure of the condensate does
not destroy the high vacuum, i.e. Pv -< 10-5 mbar is necessary. As can be seen in Fig. 19
h y d r o g e n H 2 (which appears in hydrazine mono- and bipropellant thruster exhaust gases) is
the most critical component to pump cryogenically. The wall temperature must be about
4 K, which can only be achieved with liquid helium (LHe) as the cooling agent. Both
mechanical and cryogenic pumps or a combination of the two types are used in practice
depending on the type of investigation.

Plume flowand plumeimpingementin space technology

59

For measurements of forces and heat transfer on extended surfaces, for example on the
plate configuration in Fig. 28, the location and possible influence of the barrel shock system
must be considered. Glow discharge flow visualization~I19) is an effective means to make the
shocks in low density gases visible. As an example we present the shock system established
during the impingement of a free jet on a flat plate (Fig. 47). Such pictures help to identify
the regions on the plate which are not affected by the presence of the barrel shock. In Fig. 48
the shock system due to the interaction of two plumes (Section 6) is shown.
Oblique shocks like these interaction shocks, nozzle shocks, surface shocks and barrel
shocks can be detected with a Pitot probe. Examples of the typical steep pressure increase
can be found in Figs 43 and 13b.
From Pitot pressure data the density in the flow field can be derived. For hypersonic flow
the approximation
pt2~T+ 3 1
? +----f~ pu 2

(99)

holds. In the isentropic core the velocity is almost constant (u ~ Utim).Assuming perfect gas
behaviour (y = const) one obtains
Pt2 "~ P
Under these conditions the Pitot pressure profiles in Fig. 13 served to determine the plume
constant Ap.clp (Fig. 14). Other investigators have used an electron beam probe to determine
the local density in the plume (Lengrand et al. ~36)) or in the nozzle (Rothe66)).
With an electron beam probe measurements of the gas velocity can also be performed
(Fig. 49). Molecules of the plume gas are ionized by a pulsed electron beam. The ions are
assumed to travel with the same velocity as the neutral gas and are detected further
downstream. From the distance Ar along a streamline and the lapsed time At the velocity is
calculated. This experimental method is also suitable for streamline detection as shown in
Fig. 23.
pudp .
Pitot probes can only be used as long as the probe Reynolds number Rep = ~
Is larger
than about 2. For the range below that, i.e. for increasing rarefied flow conditions, a free
molecular pressure probe is used for the determination of particle flux (Fig. 50). To apply
A
the probe free molecule flow conditions must exist, i.e. Kn = - > 10, with de: probe
dp
diameter. Its operation principles do not only allow determination of the particle flux along
the streamline. By rotating the probe around its axis mass flux contributions reaching the
probe from other directions can also be detected (compare Fig. 18 and the principle of cone
of sight). If the probe is combined with a mass spectrometer, species detection and
measurement of their concentration is possible. 12J Under continuum flow conditions it is
a Pitot probe.
While the glow discharge method also helps to identify shocks in the rarefied gas flow in
front of an impinged surface, there are two methods to help identify the flow on the surface.
Oil flow pictures show the streamline pattern (Fig. 51) and the colour response of a liquid
crystal layer gives a survey of the heat transfer distribution (Fig. 52). Moreover these
pictures allow an assessment of the penetration of background gas into the impingement
region.
Local forces on a plate have been measured with an experimental set-up as shown in
Fig. 53. A floating circular element with a diameter of several millimeters is mounted on the
lever arm of a sensitive balance, Its position is in a hole in the impingement plate. The
balance in this arrangement responds to forces perpendicular to the paper plane, i.e. shear
stress r. The working principle of the balance is to keep the lever arm in the same position
under each permitted shift by means of an electromagnetic adjustment mechanism. Thus the
disturbance of the flow along the plane wall is kept at a minimum. With a modified
arrangement the normal force component can be measured. The results presented in Figs
32-36 have been obtained with this balance. The maximum load was 10 mN.

60

G. DETTLEFF

detail

detai._._./

barrel shock

PK

//P/~//"
slip

stI.~

not suitable

1
.

surface shock

suitable for impingement investigation


corresponding to space conditions

11o. 47. Glow discharge visualization of a free jet impinging perpendicularly on a flat plate. Test gas N z,
Po ffi 2 bar, T o = 300 K. Detail; 1: undisturbed plume; 2: subsonic flow downstream from surface shock;
3: supersonic flow downstream from surface shock; 4: supersonic flow, new expansion behind subsonic flow
region 2. Compare with Fig. 30. Adapted from Ref. (128).

Plume flow and plume impingement in space technology

free

61

jet

sources

interaction
shocks

=3
re[

shock

FIG. 48. Glow discharge visualization of two interacting free jets. Penetration Knudsen number
(continuum), Po/PK= 4 x 104.Adapted from Ref.(129).

Kn, ffi 0.01

Y
e-gun
L - ~

pulsedelectron beam
barrel shock

~ r o b e
~J
g&s

-"~ ,
~

de T pressure
I prObe

pressur~_p

gaugel/ [
data recording and
processing
FIG. 49. Experimental set-up for Pitot pressure and 8as velocity measurements in a plume.

G. DETTLEFF

62

cross section at orifice


rotation
Yl
51

slot
orifice

probe tube
~vPressure~.~
o uge

olume

d I -,1 0 ['-

t -%--

V ~

circular
orifice

--

FIG. 50. The free molecular probe with slot and circular orifice. Adapted from Ref. (130).

,position of

sonic orifice

OreQ
wetted by
free jet

prate
~7

fine of vision

/ / / / / / / / / / / / / / / / / / / / / / / / /

plate
F]G. $1. Oil flow pattern on a fiat plate impinged by a free jet. Free jet, test gas N2, po = 8 bar, To = 295 K
zM/d* = 20, d* = 2 mm, ~ = 0 . Adapted from Ref. (131).

a)

h)

FIG. 52. Colour patterns of liquid crystals (surface temperature pattern) on a flat plate impinge d by a free jet.
Experimental conditions, see Fig. 51 except for To = 417 K (instead of 295 K). Time t = 6.4 sec after removal of
a protecting shield in front of the plate (similar to Fig. 54). Parameter:. background pressure Px- (a) pz/po = 10-s;
(b) P~c/Po= 5 x 10-5; (c) Px/Po= 8.5 x 10 -5. Adapted from Ref. (131); experimental arrangement see Fig. 51.

65

Plume flow and plume impingement in space technology

floating
element, F~

uo

y
- /////,.~

impingement plate

"~/11///11~117/
?S mm

leaf spring

"~

tarsion
wire

E
suspensionn _ ~
beam

quoftz beams
( lever -arm)

Ion

leaf Ipring

FIG. 53. Shear stress balance arrangement, mounted on a flat plate.

density
of Nickel

a
i

heat
flux

thickness
of Nickel wall

I (dTw.
co. s. td-T- ]
!
I

specific heat
of Nickel

/ t e m p e r a t u r e change
per time to be
measured

~ pO'T'N2

water cooled
Iwith-drawn)

~\\'-\

\}"

'\.

~',

\ \ \ ' - "~l

201~~ ~ N i

Tw

Tw measured by

FIG.54. Determination

L NiCr-wire A Sl~=0"2mm
NiCr-Ni Ihermocouples

of aerodynamic heat transfer by thin wa]] technique.

To measure the heat transfer to a surface with given wall temperature Tw the so-called
thin wall technique is applied (Fig. 54). Thermocouples are mounted on a thin surface which
is initially protected from the continuously flowing plume by a shield. The shield is removed
very quickly and the initial linear temperature history T(t) is used for the determination of
the heat transfer. Variation of the adjusted wall temperature Tw before removal of the shield
allows the determination of the recovery temperature Tr as is shown in Fig. 37b. The results
in Fig. 37b have been obtained with the calorimetric method. A copper sphere with good
thermal conduction and a diameter of a few millimeters is heated up by the plume. A
thermocouple in the sphere indicates the almost uniform temperature of this body.
Information is again deduced from the temperature history T(t).

66

G. DETTLEFF
8. SUMMARY AND CONCLUSIONS

Plume impingement is a problem encountered during practically every space mission,


since firing of thrusters with the inevitable formation of large plumes is necessary to perform
manoeuvres. Fuel consumption is to be kept at a minimum and therefore disturbing plume
impingement forces must be avoided as much as possible. Fortunately efforts to minimize
undesired impingement effects in the off-axis region require the same provisions as desired
optimization of the thrust, for example to gain an almost parallel nozzle exit flow with high
Mach numbers. To a certain extent reduction of impingement effects can be achieved by
canting of the thruster axis, but then an undesired reduction of the thrust torque must be
accepted. The same holds for the diminution of heat load below a critical value.
Impingement situations can arise in all parts of the plume, in the near-axis, off-axis, and
backflow region. The fluid mechanical problems in the different regions look quite different.
Continuum flow, non-equilibrium effects due to rarefaction, and free molecule flow with
different histories (boundary layer and isentropic core) and with different gas compositions
due to species separation must be considered to determine first of all the impinging plume
quality. Secondly the surfaces determine the nature of the plume impingement by their
location, size and shape. Continuum, transition and free molecule impingement are possible
even simultaneously on an extended surface. The superposition of such problems and the
variety of different thrusters, propellants, firing modes and impingement situations make it
necessary to identify the most important parameters and to put them into simplifying
formulae to obtain an engineering means for the prediction of impingement effects with
sufficient accuracy. The steady far field plume flow with constant gas properties and
constant gas velocity along the streamlines is such a suitable simplification. From the large
variety of different impingement situations the fiat plate oriented parallel and perpendicularly to the thruster axis has been treated as characteristic.
Proceeding from these assumptions and arrangements, the density distribution in the
flow field can be described by the general formula

p/po = B~ f(O).
Parameters entering the quantity B and function f() are the ratio of specific heats y of the
gas and the nozzle exit Mach number Man. In small thrusters the boundary layer occupies
a large part of the nozzle flow and must be considered. Thus as another important gas
property the viscosity must be known. Now the nozzle exit conditions can no longer be
characterized by one quantity Man but by profiles over the exit plane. This makes it
necessary to treat in more detail the nozzle flow with its boundary conditions. The wall
temperature is then an essential parameter. Compression shocks appear and render more
difficult a simple description. Moreover the boundary layer is a source of backflow which
has attracted increasing attention in recent years. Here a full model description is still
missing. Some results from basic work and parameter studies are, however, available.
The application of the method of characteristics has helped us to determine the ruling
parameters and density distributions in continuum flow under various conditions. A similar
role is now played by the Direct Simulation Monte Carlo method for the physical
properties of rarefied gas flow and for non-equilibrium effects. Effects of gradual transition
from continuum to free molecule flow are modelled by the introduction of a freezing surface.
Complications in the description of impingement force and heat transfer arise in part
from the non-uniformity of the plume flow. Application of Newton's theory provides in
cases of large angles of attack a suitable means to predict forces. When a pronounced
boundary layer is formed along the impinged surface (smaller angles of attack) the
assumptions of this theory obviously no longer hold and consequently its application
is questionable. Nevertheless a qualitative agreement of the pressure distribution with
experimental results can be found. This indicates that the simple formulae contain the
governing parameters. These are the distance to the nozzle, the angle of attack and y and

Plume flow and plume impingement in space technology

67

MaE. Viscosity (/~, Re) is an additional essential quantity especially in transition flow
impingement, and the specific heat c~, of the gas comes into play if heat load is considered.
Many of the equations are written in normalized form containing stagnation quantities as
reference quantities.
The formulae and criteria presented here allow the solution or at least the assessment of
certain plume flow and impingement problems. However, it must be emphasized that the
force and heat transfer equations have been developed and checked experimentally for
simple geometric shapes, for steady state firing and in the absence of pre-impingement, to
mention some limitations. The example of heat transfer on a plate with leading edge showed
that the application of the formulae derived for the infinite flat plate lead to a dramatic and
therefore eventually dangerous underestimation of destructive impingement effects. The
discussion of the bridging procedure showed that a certain experimental experience is
required to obtain reasonable predictions.
In the simple methods a treatment of disturbed plume flow is avoided. Disturbed means
that the flow in front of the impinged surface deviates from the radial source flow
assumption and is characterized by a three-dimensional shock in continuum flow and the
establishment of a boundary layer. This avoidance can be accepted as long as some
inaccuracy is tolerated in the predicted results. The interaction of two plumes and the
resulting impingement effects seem to be a problem where reasonable results can only be
obtained if the disturbed flow is included. It may therefore be the starting point for progress
in the treatment of disturbed flow. Another field of activity is the backflow where even a
modelling attempt is missing. By means of DSMC some insight has been gained and the
physical effects have been revealed. Experimental validation is still missing due to the
difficulties in obtaining proper vacuum surroundings in ground facilities. The requirements
for experimental work have therefore been extended to simulate such conditions by the use
of novel forms of pumps. Only in such facilities can the rarefaction effects in real thruster
plumes be investigated.

ACKNOWLEDGEMENT
The author should like to thank SHM for very helpful support.

REFERENCES
1. HAYES,W. D. and PROBSTEIN,R. F. (1959) Hypersonic Flow Theory, Academic Press, New York, London.
2. DORRANCE,H. (1962) Viscous Hypersonic Flow, McGraw-Hill, New York.
3. SCHAAF,S. A. and CHAMBRl~,P. L. (1961) Flow of Rarefied Gases. In: High Speed Aerodynamics and Jet
Propulsion, Princeton University Press, Princeton, N.J.
4. DADIEU,A., DAMM, R. and SCHMXDT,E. W. (1968) Raketentreibstoffe, Springer, Wien, New York.
5. VICK, A. R., CUBBAGE,J. M. and ANDREWS,E. H. (1966) Rocket exhaust plume problems and some recent
related research. In: The Fluid Dynamic Aspects of Space Flight, AGARDooraph 87, Vol. II, Gordon and
Breach Science Publishers.
6. BOETrCHER, R. D. and LEGGE, H. (1980) A study of rocket exhaust plumes and impingement effects on
spacecraft surfaces--I. Literature Survey. DFVLR Internal Report IB 251-80 A 27, G6ttingen.
7. LENGRAND,J. C. (1984) Plume impingement upon spacecraft surfaces. In: Proceedinos of the 14th International Symposium on Rarefied Gas Dynamics, pp. 217-228 (ed. H. Oguchi), University of Tokyo Press,
Tokyo.
8. OLDS, R. H. (1963) Attitude control and station keeping of a communication satellite in a 24-hour orbit,
AIAA J., 1(4), 852-858.
9. MoLrroR, J. H. (1964) Ion propulsion system for stationary satellite control, J. Spacecraft Rockets, 1(2),
170-175.
10. BOUCHER,R. A. (1964) Electrical propulsion for control of stationary satellites, J. Spacecraft Rockets, 1(2),
164-169.
11. SACKHEIM,R. L., FRITZ,D. E. and MACKLIS,H. (1980) Performance trends in spacecraft auxiliary propulsion
systems, J. Spacecraft Rockets, 17(5), 390-395.
12. SCI-IMITZ,H. D. and S'rEENBORO,M. (1983) Augmented electrothermal hydrazine thruster development,
J. Spacecraft Rockets, 20(2), 178-181.
,IP~S 28:1-E

68

G. DETTLEFF

13. PUGMIRE,T. K., SHAW, R. and ENOS, G. R. (1971) Applied resistojet technology, J. Spacecraft Rockets, 8(1),
63-68.
14. SUTHERLAND, G. S. and MAES, M. E. (1966) A review of microrocket technology: 10 -6 to 1 lbf thrust,
J. Spacecraft Rockets, 3(8), 1153-1165.
15. LEGGE, H, and DETTLEFF, G. (1986) Pitot pressure and heat transfer measurements in hydrazine thruster
plumes, J. Spacecraft Rockets, 23(4), 357-362.
16. GREENWOOD,T., SEYMOUR, D., PROZAN, R. and RATLIFF, A. (1971) Analysis of liquid rocket engine exhaust
plumes, J. Spacecraft Rockets, 8(2), 123-128.
17. JANSEN,D. P. L. F. and KLETZKINE, P. (1988) Preliminary design for a 3AN hybrid propellant engine, ESA J.,
12, 421-439.
18. BURLAGE, H., JR, GIN, W. and RIEBLING, R. W. (1972) Unmanned planetary spacecraft chemical rocket
propulsion, J. Spacecraft Rockets, 9(10), 729-737.
19. ROBERTS,L. (1966) The interaction of a rocket exhaust with the lunar surface. In: The Fluid Dynamic Aspects
of Space Flight, AGARDograph 87, Vol. II, Gordon and Breach Science Publishers.
20. STITT,L. E. (1961) Interaction of highly underexpanded jets with simulated lunar surfaces. NASA TN D-1095.
21. ALPINIERI, L. J. and ADAMS, R. H. (1966) Flow separation due to jet pluming, AIAA J., 4(I0), 1865-1866.
22. HINSON, W. F. and HOFEMAN, S. (t963) Analysis of jet-pluming interference by computer simulation of
measured flight motions of two ram at fourth stages. NASA TN D-2018.
23. HINSON, W. F. and FALANGA,R. A. (1902) Effect of jet pluming on the static stability of cone-cylinder-flare
configurations at a Mach number of 9.65. NASA TN D-1352.
24. SIBULKIN,M. and GALLAHER,W. H. (1963) Far field approximation for a nozzle exhausting into a vacuum,
AIAA J., 1(6), 1452-1453.
25. HILL, J. A. F. and DRAPER, J. S. (1966) Analytical approximation for a flow from a nozzle into vacuum,
J. Spacecraft Rockets, 3(10), 1552-1554.
26. DETTLEEE, G. (1982) A study of rocket exhaust plumes and impingement effects on spacecraft surfaces. II.
Plume Profile Analysis, Method of Characteristics Computer Program for Axisymmetric Free Jets. Part 3:
Extended version "Free Jet". DFVLR Internal Report IB 222-82 A 07, G6ttingen.
27. VICK, A. R., ANDREWS, E. H., DENNARD, J. S. and CRAIDON, C. B. (1964) Comparisons of experimental free jet
boundaries with theoretical results obtained with the method of characteristics. NASA TN D-2327.
28. ROBERTS,L. and SOUTH, J. C., JR (1964) Comments on exhaust flow field and surface impingement, AIAA J.,
2(5), 971-973.
29. ALBINI, F. A. (1965) Approximate computation of underexpanded jet structure, AIAA J., 3(8), 1535-1537.
30. BOYNTON, F. P. (1967) Highly underexpanded jet structure: exact and approximate calculations, AIAA J.,
5(9), 1703-1704.
31. DETTLEFE, G. (1984) Experimental verification of rocket exhaust plumes and impingement effects on
spacecraft surfaces. Work package 1: Plume Model Testing. Part l: Pitot Pressure Measurements. DFVLR
Internal Report IB 222-84 A 42, G6ttingen.
32. GOFFE, D, (1984) Moddlisation num6rique par la m6thode des charact6ristiques et 6tude exp6rimentale de jets
supersonique libres, Th6se de docteur ing6nieur, Ecole Centrale des Arts et Manufacture, France.
33. BOYNTON,F. P. (1968) Exhaust plumes from nozzles with wall boundary layers, J. Spacecraft Rockets, 5(10),
1143 1147.
34. SIMONS,G. A. (1972) Effect of nozzle boundary layers on rocket exhaust plumes, AIAA J., 10(ll), 1534-1535.
35. LENGRAND,J. C. (1974) Approximate calculation of rocket plumes with nozzle boundary layers and external
pressure. In: Proc. 9th Int. Syrup. Rarefied Gas Dynamics, DFVLR Press, Porz-Wahn.
36. LENGRAND,J. C., ALLEGRE,J. and RAFFIN, M. (1976) Experimental investigation of underexpanded exhaust
plumes, AIAA J., 14(5), 692-694.
37. CALLA,V. S. and BROOK, J. W. (1975) Measurements of a simulated rocket exhaust plume near the PrandtlMeyer limiting angle, J. Spacecraft Rockets, 12(4), 205-208.
38. GENOVESE, J. E. (1978) Rapid estimation of hydrazine exhaust plume interaction, AIAA/SAE 14th Joint
Propulsion Conference, AIAA paper 78-1091.
39. SCHLICHTING, H. (1965) Grenzschicht-Theorie, G. Braun, Karlsruhe.
40. LEwis, B., PEASE, R. N. and TAYLOR, H. S. (eds) (1956) High Speed Aerodynamics and Jet Propulsion, Vol. II:
Combustion Processes, Princeton University Press, Princeton, N. J.
41. DETTLEEF,G., BOETTCHER, R. D., DANKERT, C., KOPPENWALLNER,G. and LEGGE, H. (1986) Attitude control
thruster plume flow modelling and experiments, J. Spacecraft Rockets, 23(5).
42. KEWLEY, D. J. (1985) Predictions of the exit conditions, including species concentrations and the ratio of
specific heat. DFVLR Internal Report IB 222-85 A 05, G6ttingen.
43. BOETTCHER, R. D. and LEGGE, H. (1980) A study of rocket exhaust plumes and impingement effects on
spacecraft surfaces, lI: Plume Profile Analysis. Part l: Continuum Plume Modelling. DFVLR Internal
Report IB 251-80 A 29, G6ttingen.
44. LEGGE, H. and BOETrCHER, R. D. (1985) Modelling control thruster plume flow and impingement. In: Proc.
13th Int. Syrup. Rarefied Gas Dynamics, pp. 983-992, Plenum, New York.
45. BOETTCHER, R. D. and LEGGE, H. (1981) A study of rocket exhaust plumes and impingement effects on
spacecraft surfaces. II: Plume Profile Analysis. Part 3: Rarefaction Effects. DFVLR Internal Report IB 222-81
A 19, G6ttingen.
46. LEGGE,H., DANKERT, C. and DETTLEEE, G. (1984) Experimental analysis of plume flow from small thrusters.
In: Proc. 14th Int. Syrup. Rarefied Gas Dynamics, pp. 279-286, Tokyo Press, Tokyo.
47. BIRD, G. A. (1970) Breakdown of translational and rotational equilibrium in gaseous expansions, AIAA J.,
8(11), 1998-2003.
48. BIRD, G. A. (1981) Breakdown of continuum flow in free jets and rocket plumes. In: Proc. 12th Int. Syrup.
Rarefied Gas Dynamics, Prog. Astronaut. Aeronaut., 74, 681-694, AIAA, New York.
49. TANG, S. P. and FENN, J. B. (1978) Experimental determination ofthe discharge coefficients for critical flow
through an axisymmetric nozzle, AIAA J., 16(1), 41-46.

Plume flow and plume impingement in space technology

69

50. ALLEN, G. A., KOPPENWALLNER,G. and LENERS, K. H. (1986) A study of species separation in free jet
expansions. In: Proc. 15th Int. Syrup. Rarefied Gas Dynamics, pp. 66-75, B. G. Teubner, Stuttgart.
51. NAUMANN, W. (1985) Die Nachexpansion von Dfisenstr6mungen mit Grenzschicht im hypersonischen
Freistrahl. DFVLR Report DFVLR-FB 85-10.
52. BOYD,I. D. (1988) Modelling of satellite control thruster plumes, Ph.D. "Ilaesis, University of Southampton.
53. BIRD, G. A. (1974) The nozzle lip problem. In: Proc. 9th Int. Syrup. Rarefied Gas Dynamics, DFVLR Press,
Porz-Wahn.
54. HUESER, J. E., MELFI, L. T., JR, BIRD, G. A. and BROCK, F. J. (1986) Rocket nozzle lip flow by Direct
Simulation Monte Carlo method, J. Spacecraft Rockets, 23(4), 363-367.
55. CAMPBELL,D. H. (1989) Nozzle lip effects on argon expansions into the plume backflow, J. Spacecraft
Rockets, 26(4), 285-292.
56. BAILEY,A. B. (1987) Flow angle measurements in a rarefied nozzle plume, AIAA J., 25(10), 1301-1304.
57. HOFFMAN,R. J., KAWASAKI,A., TRINKS, H., BINDEMANN,I. and EWERING, W. (1985) The CONTAM 3.2
plume flow field analysis and contamination prediction computer program: analysis model and experimental
verification, AIAA 20th Thermophysics Conference, Williamsburg, AIAA Paper 85-0928.
58. TRINKS, H. and HOFFMAN, R. J. (1985) Standard experimental exhaust plume analysis procedure in
connection with CONTAM computer model predictions, AIAA 21st Joint Propulsion Conference, Montery.
59. CONWAY,W. E. and CELMINS,A. (1972) Maximum thrust nozzles, AIAA J., 10(9), 1241-1242.
60. KALLIS,J. M., GOODMAN, M. and HALBACH,C. R. (1972) Viscous effects on biowaste resistojet nozzle
performance, 2. Spacecraft Rockets, 9(12), 869-875.
61. LANDOLT-BORNSTEIN(1969) Eigenschaften der Materie in,ihren Aggregatzustllnden, 5. Teil, Transportphiinomene, Springer.
62. REID, R. C., PRAUSNITZ,J. M. and SHERWOOD,T. K. (1977) The Properties of Gases and Liquids, 3rd edn,
McGraw-Hill.
63. WHITEIELD, D. A. (1968)-Theoretical and experimental investigation of boundary layers in low density
hypersonic axisymmetric nozzles, AEDC-TR-68-193.
64. CARDEN,W. H. (1965) Local heat transfer coefficients in a nozzle with high speed laminar flow, AIAA J., 3(12),
2183-2188.
65. CHANG,C. L., KRONZON,Y, and MERKLE,C. L. (1988) Time-iterative solutions of viscous supersonic nozzle
flows, AIAA J., 26(10), 1208-1215.
66. ROTHE,D. E. (1971) Electron beam studies of viscous flow in supersonic nozzles, A1AA J., 9(5), 804-811.
67. MIGDAL,D. and LANDIS,F. (1962) Characteristics of conical supersonic nozzles, ARS J., 32, 1898-1901.
68. DARWELL,H. M. and BADHAM,H. (1963) Shock formation in conical nozzles, AIAA J., 1(8), 1932-1934.
69. MIGDAL,D. and KOSSON,R. (1965) Shock predictions in conical nozzles, A1AA J., 3(8), 1554-1556.
70. BACK, L. H. and COFFEL, R. F. (1966) Detection of oblique shocks in a conical nozzle with a circular arc
throat, AIAA J., 4(12), 2219-2221.
71. VOGENITZ,F. W., BIRD, G. A., BROADWELL,J. E. and RUNGALDIER,H. (1968) Theoretical and experimental
study of rarefied supersonic flows about several simple shapes, AIAA J., 6(12), 2388-2394.
72. PIESIK,E. T., KOPPANG,R. R. and SIMKIN,D. J. (1966) Rocket exhaust impingement on a flat plate at high
vacuum, J. Spacecraft Rockets, 3(11), 1650-1657.
73. EASTMAN,D. W. and RADTKE,L. P. (1963) Flow field of an exhaust plume impinging on a simulated lunar
surface, AIAA J., 1(6), 1430-1431.
74. EASTMAN,D. W. and BONNEMA,J. P. (1966) Flow field of a highly underexpanded jet impinging on a surface,
AIAA J., 4(7), 1302-1303.
75. LEGGE,H. (1990) Plume impingement forces on inclined flat plates. In: Proc. 17th Int. Syrup. Rarefied Gas
Dynamics, VCH Veriag, Weinheim.
76. VICK, A. R. and ANDREWS, E. H., JR (1966) An investigation of highly underexpanded exhaust plumes
impinging upon a perpendicular flat surface, NASA TN D-3269.
77. ROCHELLE, W. C. and KOOKER, D. E. (1969) Heat transfer and pressure analysis of rocket exhaust
impingement on flat plates and curved panels, J. Spacecraft Rockets, 6(3), 248-256.
78. VICK,A. R. and ANDREWS,E. H., JR (1964) An experimental investigation of highly underexpanded free jets
impinging upon a parallel flat surface, NASA TN D-2326.
79. LENGRAND,J. C., ALLI~GRE,J. and RAFFIN, M. (1977) Interaction of underexpanded jets with adjacent flat
plates. In: Proc. lOth Int. Syrup. Rarefied Gas Dynamics, Prog. Astronaut. and Aeronaut., 51, 447-458, AIAA,
New York.
80. LENGRAND,J. C., RAFFIN, M. and ALLI~GRE,J. (1981) Monte Carlo simulation method applied to jet wall
interactions under continuum flow conditions. In: Proc. 12th Int. Syrup. Rarefied Gas Dynamics, Prog.
Astronaut. Aeronaut., 74, 994-1006, AIAA, New York.
81. LENGRAND,J. C., ALLI~GRE,J. and RAFFIN, M. (1982) Underexpanded free jets and their interaction with
adjacent surfaces, AIAA J., 20(1), 27-28.
82. BOETTCHER,R. D. and LEGGE, H. (1981) A study of rocket exhaust plumes and impingement effects on
spacecraft surfaces, DFVLR Internal Report IB 222-81 A 28, Grttingen.
83. LEGGE,H. (1986) Shear stress and pressure in plume impingement flow. In: Proc. 15th Int. Syrup. Rarefied Gas
Dynamics, pp. 523-538, B. G. Teubner, Stuttgart.
84. CHII), P, B., PEARSON,D. J., MUHM, P. M., SCHOONMAKER,P. B. and RADAR,R. J. (1977) SVDS plume
impingement modelling development. Sensitivity analysis supporting level B requirements, NASA-TM74768.
85. BOETTCHER,R. D. (1982) Computational analysis of plume impingement effects from the reaction control
thrusters of the GIOTTO spacecraft, DFVLR Internal Report IB 222-82 A 32, Grttingen.
86. BOETTCHER,R. D. (1989) DLR technical assistance to CNES on HERMES, Contract Report H-NT-O-2010,
DLR Grttingen.
87. MATTING,F. W. (1971) Approximate bridging relations in the transitional regime between continuum and
free molecule flows, J. Spacecraft Rockets, 8(1), 35-40.

70

G. DETTLEFF

88. ALLI~GRE,J., RAFFIN,M. and LENGRAND,J. C. (1984) Forces induced by a simulated rocket exhaust plume
impinging upon a flat plate. In: Proc. 14th Int. Syrup. Rarefied Gas Dynamics, pp. 287-294, University of
Tokyo Press, Tokyo.
89. Su, M. W. and MULLEN,C. H., JR (1972) Plume impingement force during tandem stage separation at high
altitudes. J. Spacecraft Rockets, 9(9), 715-717.
90. ALL~GRE,J., RAFFIN,M. and LENGRAND,J. C. (1986) Experimental study of the plume impingement problem
associated with rocket stage separation. J. Spacecraft Rockets, 23(4), 368-372.
91. LENt, J., OSONXTSCH,C. W. and LACINSKI,T. M. (1969) Effects of oblique shock waves in the near field of
rocket plumes, J. Spacecraft Rockets, 6(11), 1316-1318.
92. ECKERT,E. R. G. (1960) Survey of boundary layer heat transfer at high velocities and high temperatures,
WADC Technical Report 59-624.
93. LEHRER, S. (1966) Microthrust engines for the investigation of spacecraft surface heating. J. Spacecraft
Rockets, 3(7), 983-988.
94. LEGOE,H. and DETTLEFF,G. (1984) Experimental verification of rocket exhaust plumes and impingement
effects on spacecraft surfaces. WPS: Hydrazine thruster near field plume profile measurements. Part 1: Heat
transfer measurements in combination with Pitot pressure and thruster data, DFVLR Internal Report
IB 222-84 A 48, G6ttingen.
95. KEMP, N. H. and RIDDELL, F. R. (1957) Heat transfer to satellite vehicles reentering the atmosphere, Jet
Propul., 27, 132-137.
96. MADDOX,A. R. (1968) Impingement of underexpanded plumes on adjacent surfaces. J. Spacecraft Rockets,
5(6), 718-724.
97. LENGRAND,J. C., ALLI~GRE,J. and RAFFIN,M. (1981) Heat transfer to a surface impinged upon b~' a simulated
underexpanded rocket exhaust plume. In: Proc. 12th Int. Syrup. Rarefied Gases, pp. 980-993, AIAA,
New York.
98. LEGGE,H. and DANKERT,C. (1981) Influence of incomplete accommodation in wind tunnel experiments on
heat transfer and drag of sharp cones in the transition regime. In: Proc. 12th Int. Syrup. Rarefied Gas
Dynamics, pp. 964-979, AIAA, New York.
99. DORING, S. (1990) Experimental plume impingement heat transfer on inclined flat plates, DLR Internal
Report IB 222-90 A 36, G6ttingen.
100. KOPPENWALLNER,G. (1983) Rarefied plume interference and scaling laws, DFVLR Internal Report IB
222-83 A 27, G6ttingen.
101. ERMOLOV,V. I., LENGRAND,J. C., REBROV,A. K. and KHRAMOV,G. A. (1983) Experimental study of the
interaction of a pair of hypersonic jets. J. M~c. Appl. Phys. Tech, (3), 379-383.
102. SOGA,T., TAK.,NISHI,M. and YASUHARA,M. (1984) Experimental study of interaction of unexpanded free jets.
In: Proc. 14th Int. Syrup. Rarefied Gas Dynamics, pp. 485-492, University of Tokyo Press, Tokyo.
103. DANKERT,C. and KOPPENWALLNER,G. (1984) Experimental study of the interaction between two rarefied free
jets. In: Proc. 14th Int. Syrup. Rarefied Gas Dynamics, pp. 477484, University of Tokyo Press, Tokyo.
104. MCCAY, T. O., WEAVER, D. P., WILLIAMS,W. D., POWELL, H. M. and LEWIS, J. W. L. Exhaust plume
contaminants from an aged hydrazine monopropellant thruster. In: Proc. USAF/NASA Int. Spacecraft
Contamination Conf. (ed. J. Jemiola). NASA-CP-2139, 456-517.
105. McCAY, T. D., POWELL,H. M. and BusBY, M. R. (1978) Direct mass spectrometric measurements in a highly
expanded rocket exhaust plume, J. Spacecraft Rockets, 15(3), 133-138.
106. LEGGE,H. and KOPPENWALLNER,G. (1986) Similarity in plume impingement with application to satellites.
In: Proc. 15th Int. Syrup. Space Technology Science, Tokyo.
107. LEGGE, H. (1982) A study of rocket exhaust plumes and impingement effects on spacecraft surfaces. VI.
experiment proposal, DFVLR Internal Report IB 222-82 A 14, G6ttingen.
108. BAILEY,A. B., PRICE, L. L. and PIPES, J. G. (1985) Effect of ambient pressure on nozzle center line flow
properties, AIAA J., 23(6), 953-954.
109. LEWIS,C. H., JR and CARLSON,D. J. (1964) Normal shock location in underexpanded gas and gas particle jets,
AIAA J., 2(4), 776-777.
110. D'ATTORRE,L. and HARSHBARGER,F. C. (1965) Parameters affecting the normal shock location in underexpanded gas jets, AIAA J., 3(3), 530-531.
111. CRIST,S., SHERMAn,P. M. and GLASS,D. R. (1966) Study of highly underexpanded sonic jets, AIAA J., 4(1),
68-71.
112. ASHKENAS,H. and SHERMAN,F. S. (1967) The structure and utilization of supersonic free jets in low density
wind tunnels. In: Proc. 5th Int. Syrup. Rarefied Gas Dynamics, pp. 84-105, Academic Press.
113. MORAN,J. P. (1967) Similarity in high altitude jets, AIAA J., 5(7), 1343-1345.
114. MUNTZ,E. P., HAMEL,B. B. and MAGUIRE,B. L. (1970) Some characteristics of exhaust plume rarefaction,
AIAA J., g(9), 1651-1658.
115. Murerz, E. P. and MAGUIRE,B. L. (1971) An experimental study of the rarefaction of underexpargied jets.
In: Proc. 7th Int. Syrup. Rarefied Gas Dynamics, pp. 619-626, Editrice Tecnico Scientifica, Pisa.
116. MEYER,J.-T. (1990) Free molecular pressure probe measurements in the continuum, transitional and free
molecular regimes of a free jet. In: Proc. 17th Int. Syrup. Rarefied Gas Dynamics, VCH Verlag, Weinheim.
117. LEGGE, H. and RAGHURAMAN,P. (1974) A study of background molecule scattering in free jet expansions
through flat plate drag measurements. In: Proc. 9th Int. Syrup. Rarefied Gas Dynamics, DFVLR Press,
Porz-Wahn.
118. LEIqG,J., OMAtq,R. A. and HOPKINS,H. B. (1968) A detonation tube technique for simulating rocket plumes
in a space environment, J. Spacecraft Rockets, 5(10), 1148-1154.
119. FISHER,S. S. and BHARATAN,D. (1973) Glow discharge flow visualization in low density free jets, J. Spacecraft
Rockets, 10(10), 658-662.
120. KOPPENWALLNER,G. (1986) The free molecular orifice pressure probe for mass spectroscopic studies. In: Proc.
15th Int. Symp. Rarefied Gas Dynamics, pp. 24-33, B. G. Teubner, Stuttgart.
121. BOETTCHER,R. D., DETTLEFE,G., KOPPENWALLNER,G. and LEGGE, H. (1982) A study of rocket exhaust

Plume flow and plume impingement in space technology

71

plumes and impingement effects on spacecraft surfaces. Executive Summary, DFVLR Internal Report IB 22282A 11.
122. WALKER, S. C., PRATT, C. L. and GOODNIGHT, F. H. (1963) Integrated back-pack manceuvering unit
propulsion study and exhaust plume heating analysis, ASD-TDR-63-729, Chance Vought Corp., Astronautics Division.
123. DETTLEFr,G. and DANKERT,C. (1988) Studies on rocket exhaust plumes and impingement effects related to
the Columbus space station program. Work package 3: Experimental study of the boundary layer expansion
region in the plume, DFVLR Internal Report IB 222-88 A 14, G6ttingen.
124. VOEGT,S. (1990) Berechnung isentroper, rotationssymmetrischer Diisenstr6mungen mit dem Charakteristikenverfahren, DLR Internal Report IB 222-90 A 05, G6ttingen.
125. DANKERT,C. and DETTLEFF,G. (1990) Near field expansion in thruster plumes. In: Proc. 17th Int. Syrup.
Rarefied Gas Dynamics, VCH Verlag, Weinheim.
126. DETTLEFF,G. and LEGGE, H. (1985) Experimental verification of rocket exhaust plumes and impingement
effects on spacecraft surfaces. Work package 5: Hydrazine thruster near field plume profile measurements,
DFVLR Internal Report IB 222-85 A 15, G6ttingen.
127. DETTLEFF,G. and LEGGE, H. (1985) Experimental verification of rocket exhaust plumes and impingement
effects on spacecraft surfaces. Work package 2: Continuum and transition flow impingement, Part 2, DFVLR
Internal Report IB 222-85 A 42, G6ttingen.
128. LEC,GE, H. (1988) Plume impingement on inclined flat plates in continuum and rarefied flow. Part 1: Flow
survey by glow discharge and liquid crystal surface temperature visualization, DFVLR Internal Report IB
222-88 A 02, G6ttingen.
129. DANKERT,C. and KOPPENWALLNER,G. (1984) Influence of penetration Knudsen number on interaction of
two rarefied free jets, DFVLR Internal Report IB 222-84 A 23, G6ttingen.
130. KOPPENWALLNER,G. (1984) The free molecular pressure probe with finite length slot orifice. In: Proc. 14th
Int. Syrup. Rarefied Gas Dynamics, pp. 415-422, University of Tokyo Press, Tokyo.
131. PRASAD,J. K. and LEGGE,H. (1983) Studies of plume impingement on a flat plate, DFVLR Internal Report IB
222-83 A 10, G6ttingen.
132. LEC,GE, H. and DETTLEFF,G. (1985) Experimental verification of rocket exhaust plumes and impingement
effects on spacecraft surfaces. Work package 2: Continuum and transition flow impingement. Part 1, DFVLR
Internal Report IB 222-85 A 41, G6ttingen.

Вам также может понравиться