Professional Documents
Culture Documents
r u
COMPUTATIONAL COMBUSTION
FLUID ANALYSIS
DYNAMICS EVALUATION
Report D.O.116
NAS8-36955
Number
CCFD-92-02
By
Y.M. University
Kim,
H.M.
Shang,
C.P.
Chen,
and
of Alabama
in Huntsville,
Huntsville,
April,
(NASA-CR-IG4326) FLUIO OYi_AHICS EVALUATION Univ.) 94 Fin,_1 p
1992
N93-12688 --TH_U-N93-12689 Unclas
G3/_4
0120220
For
C. Marshall
Center,
\ \
f v
ABSTRACT This spray ciency physical dense solved study involves We the development discuss particle several tracking of numerical issues modelling the in dilute and dense effiof and are the
combustion. in the
concerning
stochastic
method
swirling
equations correction
in Eulerian procedure
equations technique.
coordiis used
discrete length
scales
turbulence
each computarepre-
The
testings
the capability
of accuratly turbulent
dispersion efficiency
in.homogeneous separated
stochastic employed
effects,
we have analogy
coalescence
a Taylor
breakup(TAB) all-speed
dense
development
tracking
method E.
wall-boundary
documented
in Appendix
/? .:
TABLES
OF
CONTENTS
ABSTRACT NOMENCLATURE 1. INTRODUCTION 1.1 General 1.2 Turbulent 1.3 Dilute 1.4 Dense Particle Models Models MODELLING Equations Equations Models Models Equations Mass & Heat Interaction and Collision Transfer Model Model Model CONDITIONS Equations OF THE MULTI-PHASE FLOWS Dispersion
Spray Spray
Turbulence Combustion
Lagrangian
Droplet/Wall
Impingement AND
MODEL
BOUNDRAY
Dispersion Flows
Spray
Flames Atomizer
a Rotating-Cup
4.3 Dense
ii
...d
Spray Spray
Sprays
5. CONCLUSIONS 6. RECOMMENDATIONS
APPENDIX A. Two-Phase B. Numerics Interaction of Beta Source Terms Density Function Fuels
Probability
C. Stoichiometric D. Droplet
Relations
For Hydrocarbon
Distribution
Wall-Boundary Report
To The FDNS
Code
from SECA,
III
"x
J i o
NOMENCLATURE
Bm Bt :
"
Cp : Cp,_ : CD :
dp : D: E: Ep :
specific internal
f:
Fpi g: h: hI : K: k: L:
fuel vapor
Nt:
Np :
of droplets
particle
of computational
P: p:
Pr
pressure Prandtl Reynold's radius velocity velocity for gases for droplets number number
I_ep : rp : U i :
instantaneous instantaneos
S: _..j
SC t :
source turbulent
T:
gas temperature
iv
time Td:
2:
droplet
Y: Y:
coordinate direction
normal
to the streamwise
density
dissipation fraction
mixture time
of pdf
fuel gas phase time eddy liquid particle step index phase or parcel square index in an eddy
9: i: k: l:
.p: TTn8
root
mean
t: Supercripts -: '
turbulent
density-averaged fluctuating
f ....
NASA
I. Report No.
5. Report Duo April 10, 1992 6. Performing Organization Code University of Alabama in Huntsville
7. Aurthor(s) Y. M. Kim H. M. Shang C. P. Chen J. P. Ziebarth }. Performing Organization Name and Address
11. Contract or Grant No. University of Alabama in Huntsville Huntsville, Alabama 35899 1":2. Sponsoring Agency Name and Addre_ss"L National Aeronautics and Space Administration Washington, D.C. 20546-001 Marshall Space Flight Center, AL 35812 r5. Supplementary Notes NAS8-36955, D.O. 116
16.Abstract
This study involves the development of numerical modelling in spray combustion. These modelling efforts are mainly motivated to improve the computational efficiency in the stochastic particle tracking method as well as to incorporate the physical subrnodels of turbulence, combustion, vaporization, and dense spray effects. The present mathematical formulation and numerical methodologies can be casted in any time-marching pressure correction methodologies (PCM) such as FDNS code and MAST code. A sequence of validation cases involving steady burning sprays and transient evaporating sprays will be included.
r!
Ig. Security Class. (of trois report) Unclassified NASA FORM 1626 OCT 86
22. Price
_s
1.
INTRODUCTION
1.1 General
have and
been physical
a number models
efforts[I-6]
the
of
combustion.
fuel and
evaporation
analysis studies
are
better demand
turbine,
engines. of the local flow and continuous interphase properties accounts of spray for complex flows. flames requires the between approaches The as the continuum soluthe have
The tions
prediction
of multi-phase droplets
interactions Various
phenomena. classified
combustion statistical
two-fluid well
approaches model
gained
to its comconvenient Eulerianflow(SSF) to acstochastic and gas two comphase phases terms coorbased
putaional interphase Lagrangian approach[5] count separated bustion and the for
in handling
poly-disperse
spray, diffusion.
and
elimination
s_ochastic
model transport.
is usually In the
effects
interphase
of the
two-phase
in terms The
dinate on the
a time-marching technique.
operator-splitting
droplet-phase
Lagrangian
_-VJ
discrete length
droplet scales
technique.
The
k - e model for
of the
turbulence vaporization
and
droplet/turbulence of variable
interactions.
includes
the effects
properties,
non-unitary of internal
Lewis
the $tefan
circulation
heating. motivated capabilitiy In the to improve over following the physical range submodels of fuel spray as well as conditions issues method va-
the
a wider
combustor
subsections,
we adress particle
several
concerning
tracking
of turbulence,
combustion,
swirling
Turbulent In the
Particle stochastic
flow(SSF) droplets
each same
computat]onal droplet
parcel
a collection sampling
characteristics
a random
is entailed and
for instantaneous
on a specified
turbulence
model
fluctuations tracking.
Lagrangian a large
particles
to produce
distributions
for rather
width(group)
of physical model
distribution. dispersion
width Each
for the
due to the
dispersion
in the Lagrangian or
position tracking,
Lagrangian density
probability
function(PDF). dispersion
mean-squared
on prior
advantages represent
is to reduce and
of computational
par-
which
dynamics
grid-independent
solutions
Eulerian of "group"
gas-flow evap-
or "group" report
dispersion
is basically procedure
similar
is somewhat
Jeng[T],
in which
time less
Furthermore, memory.
transport interactions,
model
to calibrate
the stochastic
of particledispersion
turbulence
round
width
has successfully
capability
representing improved
flows with
1.3
Dilute
Spray
Models
spray
combustion
models, droplet
separated turbulence
is
for the turbulent and rate. the combustion The turbulent density effects
processes fluctuations
an irreversible
properties centrifugal
are force
function(pdf) are
associated equations.
the swirl
in the
we evaluate combustion,
procedure swirling
distributions.
present
numerical
for the
multi-phase
tested
by applying swirling
it to predict
confined,
the liquid-fuel
a hollow
cone
mixing.
Two
considered
influence
of swirl
evaporation
and
trajectories, The
and
of of
predictive
with
experimenfeatures
tal data.
experimental locations, at
differences
at near-burner centerline.
distributions chemistry
the deficiencies
of the k - e turbulence
curvature.
1.4
Dense
Spray
Models aspects in spray process, occurs grid on combustion drop time sizes breakup, and and modeling droplet scale is the dense collision too short spray co-
and
length time
to be
steps.
Thus,
process.
To account
collision
model,
ity distributions drops method. ing the able droplet reasonable processes are sampled The
the number
particle and
tracking distorton
TAB
and
system.
is based
within
a dense drops
spray which
the nozzle
exit.
to Reitz's to input
has several
an easy
introduction
informations of mass,
of distortion and
effects
exchange
momentum, cases, 4
and burning
dense
spray
the predictions
agreement
with
available
experimental
results
in terms
of spray
drop spray.
characteristics comparison
the detailed
available formulation
of a volume-of effects.
fluid (VOF)
2.
MATHEMATICAL
MODELLING
OF
THE
MULTI-PHASE
FLOWS
and
liquid-phase
by a system equation
of unis writ-
ten in an Eulerian coordinates. interaction transfer. 2.1 Basic The source These
is presented phases
between the
is described mass
represent below.
of momentum,
equations
are given
Eulerian
Equations
2.1.1
Mean The
Flow
Equations conservation coordinate can equation be written of mass, momentum, and scalar
density-weighted in an Eulerian
variables
as follows: (2.1)
(pu ) =
(puiuj)
azi
azj
(2.2)
(pu,) density
= -_-_-x [purz'] + S_, + S_,,t mixture, part fluctuating u, and u_ are the i component of the instantaneous part velocity,
(2.3) of
of the
and scalar
Detailed
expressions
... r
terms to model
can
be found
To close
the
system
of equations,
we
uiu_"':77-' and
ui..'727.
effective
diffusivity
model the
is used
to represent fluxes,
diffusivity
models, through
turbulent
of an isotropic
and
a constant
Prandtl
or Schmidt
pu_,; = -#,( _
au,
Oui)
au,
(2.4) (2.5)
eddy
viscosity(pt) turbulence
appearing length
in (2.4)
and
(2.5)
is defined scale
in terms so that
of a #t
scale(k3/2/e)
and
a velocity
(kl/2),
The the
turbulent turbulent
kinetic transport
energy,
k, and
its dissipation
equations:
#, Op Op
Opk _j 0 #t Ok ..-TST.j Oui p20Xj OX j
pe
(2.7)
0
Here, density tion terms
"
0
involving These
'
,,)a,a,;
0"{ '
_ 8z,i
in (2.7) terms
and
effects[12].
originally
come
in the Reynolds
stress
equation. effect
For reacting
flows,
partially tion.
from combus-
2.1.3
Combustion
Model
sources
combustion
process
flames.
Experimental approach
13. An idealized
is to invoke intermediate
assumption play
do not
a significant
diffusion account
flame model,
of turbulence fractions.
of mass but
oxidizer way
at a different
convenient
to include
the effect
of turbulent
properxi). This
ties is via the introduction function mixture equations. convoluting contains fraction. The information These
variance
variables
density-weighted functions
the property
a probability
fuction,
= f Numerous study detailed delta pdf probability the/3 density pdf which pdf functions
adopts
is known
as the widely
numerics procedures
of the/3
in Appendix
B. The
in the program. is optionally rate incorporated is determined in the present as follows: reaction In an is
eddy Using
sigle-step
reaction,
mixing-controlled
rate[14]
= Ap kmin(Yi,
(2.10) Y! and
s is the
fraction chemical
ignition
to be considered. Arrhenius 7
chemically
controlled
rate,
formula[15].
pYf a PYo)be(-,fi,)
Rch, The reaction rate, Rfu = A( Wf ) ( from either slower. of the mixing rates is determined rate,
(2.11)
of the reactants
or the chemical
reaction
whichever
R D, = rnin(R,,,iz, For simple one-step reaction of the hydrocarbon-air composition. the remaining Once species
be easily
in Appendix
is described is essentially
by a discrete a statistical
particle approach
method
formulated track-
number
particle
a number
of droplets equations
velocity,
temperature,
The governing
- vi
t_i ! -- Vi
(2.13)
dpdvi
ui
dt
ri
m
+ Fbi
(2.14)
drp dt and
dTd
m,u
4_rrp2 pd
(2.15)
QL
(2.16) force
the centrifugal
The particle
r '-1 ' =
relaxation
+ tti'
ri can expresses
3P--_-CDtlli 8 rp
-- Vi I
(2.17)
Cv
is the drag
coefficient
given 24
by 1 + _Repl); 0.424; for for Rep Rep > 1000 < 1000 (2.18)
Re v = lu, + u/ # In equation [16,17] m, In equation the droplet, (2.16), is found = 27rdp(pD)(1 + 0.3Re_ (2.15), the droplet evaporation
(2.19) correlation
rate
Scd _ )/n(1
+ B,,,) to be constant
(2.20) within
which QL
is assumed :
energy
QL - 47rrp2Qc
j
(2.21)
rate to
where
heat
the droplet
per unit
by the _,,
_ z'rd
Ranz-MarshaU l,/n(l+B,,)
) B-m
correlation
Qc =
The Schmidt as number,
2K(T-Td)(
dp
,.,,,,
u.o.,t.ep
(2.22)
Prandtl
number
and
mass
transfer
number
are defined
re-
spectively
la pD'
Prd
= pCp K
1_
dependent
mass
fraction
a reference
to the
is used. mass
fraction,
Y, is obtained
P W_ -l Y, = [1 + ("x- - 1),--;7-,1 1% 9
Here
the mass
fraction
pressure
surface,
equations
are described
A. The C.
droplet
sprays
of the
droplet
passage
through
the
plane
of symmetry, but
the
droplet image
instantaneous is injected
properties
and physical
dimensions,
the mirror
velocity
vector,
On impingement momentum[10].
on a wall,
the droplets
to bounce
the reduced
Interactions is described by a discrete particle method we follow formulated the concept computais calcufor each
To account
dispersion, distribution
a normal
(Gaussian)
tional lated
particle.
location tracking
by a stochastic particle
computational
uk
Vk
dt dxk -dt with 3 I.k-1 =8pp p The location particle's calculated by the above
rk
_-Pk
+ F6k
(2.25)
(2.26)
CDIuk-
dv
Vk[
equations The
corresponding and
of each statistical
turbulent turbulent
effects. dispersion,
the variance
velocity
can be splitted
from equations
dvtk
(2.25)
u'k -- Vtk
-dt
rk 10
(2.2S)
dz'_ dt With chosen the isotropic a Gaussian turbulence distribution assumption, with each component deviation within (2.28)
from
standard
as the time
the eddy
lifetime,
fluctuating
locations
(2.30)
Vtki = Utkrrns + (Vtk(i_l) _ Utkrms)e'_(i-t)
We then
sum
up the m steps
for which
Irtl
the particle
fully interact
E
i=!
Atti = Ark
(2.32)
The
change
of variance
particle dispersion
rn 2
within
the
k th eddy
is
represented
by a characteristic
in the form:
ak 2 = ak-i
2 + (_
i=1
z'ki)
(2.33)
(2.33),ak-1 within
variance
of
eddy
of time
step/_tki
2.1 well describes is easy particle, implemented from particle In their errors the pdf recent
eddy and
with
particles. memory.
present each
less x'ki,
computer u'k,._s,v'ki,
need
current
time-marching
numerical
is summed study,
truncation
of unnecessary
time
history efflcency
terms was
was discussed
and additional
computational
obtained[35]. 11
particles, number
the
variances The
of particles.
can be written
(2.34)
represents factor
particle
position, number
K is of exist
is the total
symmetry
and
condition
in coordinate may be
boundary,
= location
distribution = 0.5[err(
y - yp ) + err(
(2.36)
where
erf( ) =
In accordance sitions with the approach particles
of Litchford
of computational
is calculated this
gas velocity),
using
of gas-phase gas
turbulent
in Eqs.(2.25)-(2.27)
velocity),
as a stochasticdispersion
model. function by SSF model, Carlo of u'k the method turbulence effects that on droplet dis-
are u'k,
simulated where
a Monte
a fluctuating disThus
is randomly
with
standard
, is added 12
to the mean
the turbulence
dispersion cally
is assumed
This
type
of simulation
has been
used Main
[20,21,22]
stationary
dispersed to specify
eddy
eddy. within
procedures
aspects
the interaction
can be found
2.2.3
Drop The
Breakup study
and
Collision the TAB This and (Taylor model Analog" Breakup) model between force proan of
present
employs
posed
by O'Rourke
is based
on an analogy The
system. The
restoring
to the suface
external forces
force on the mass due to liquid the TAB an easy viscosmodel introand
model[34], angle,
advantages
of liquid effects
viscosity
effects,
of distortion
momentum,
limitation
only
tracked. spite
However,
many between
analogy.
agreement times
breakup require
has been
two normalized
particle
arrays(deformation
oscillation number.
drop
Weber
is then
chosen breakup,
distribution temperature
$MR. drop,
Following
with
and oscillating
are set to zero. drop collision model among suggested by O'Rourke[8] liquid phase. is employed The collision to calculate routine is
and coalescence
the dispersed 13
operatinging forthe pairof particles if, and only if,they are in the same computationalcell.For the collision calculation, the drops associated with each computation parcel axe considered to be uniformly distributedthroughout the computational cell where they are located. For allparcelsin each computational cell, a collision frequency between drops between the parcel(parcell ) of largerdrop radius(rl ) and the parcel(parcel2)of smallerdrop radius(r2)isobtained from the relationship in terms of the number ofdrops in parcel2,the relative velocitybetween parcell and parcel2,
the area based on rl + r2,and the volume of computational cell.The probability with n collisions is assumed to follow a Poisson distributionbased on a collision frequency and the computational time step. Using the probabilityinformations, the collision impact parameters are stochastically calculated.Ifthe collision impact parameter is lessthan a critical impact parameter, the outcome of every collision
is coalescence. In opposite case, each collision is a grazing collision. The critical impact parameter depends on the drop radii,the relativevelocitybetween drops, and the liquidsurfacetension coefficient.
3.
NUMERICAL
MODEL
AND
BOUNDARY
CONDITIONS
The present method isbased on the operator splitting technique[23]attempting to reach accurate transientsolutionafter prescribed predictor- correctorsteps for each time-marching step. The previous multiple pressure-correction method[t1.27] isextended to to handle the strong nonlinear couplings arisingin the multi-phase, fast-transient, and reacting flows. This method is non-iterativeand applicableto
all-speedflows. The additionalscalarconservation equations such as species, and enery are incorporated intothe same predictor-corrector sequence. Discretization of the gas phase governing equation uses the finite volume approach. To enhance the numerical stability, the implicit Euler scheme isemployed in differencing the temporaldomain. Allthe dependent and independent variablesare stored at thesame grid location and the variablesat the finite controlvolume boundaries are interpolated between adjacent grid points.The discretizations have been performed on a general non-orthogonal curvilinearcoordinatesystem with a second order upwind scheme for convection resulting terms and the central differencing scheme for diffusion gadient terms. solver. The In
discretized
equations
were solved 14
by a conjugate
(CGS)
into particle
a one-predictor/two-corrector and gas are evaluated are used solution PSIC(particle The method to treat by the mo-
mentum described
exchanges above
iterations
and time-accurate. the convergence and the numerical equations stability for the fast transient are
except
the continuity
equation
in the advective
pn+l
- p,_
. u,),_+l
(3.1)
pn
-SiC
By using corrector operator-splitting method, the transport as follows: equations with procedure step can be discretized
(3.3)
Scalar('):
pA (-_-Bo)C*=J'_(*)+Sq, Here, the operators scheme terms. ,_ . +S_,,t+ p. ,, A--"_ from the third-order scheme reacting (3.4) upwind
Ao, Bo, H, and J are constructed terms and the central stability resulting
differencing
treatment
turbulence S_,,
source
interaction
The 15
quantities
S_,.'_ t are
determined from the existing flow fields. The general scalar dependent variables, may represent the enery, the mass fraction, and the turbulent transport quantities. In this stage, the velocity field(u_) does not satisfy the continuity equation. The temperature T is calculated
predictor step. step ): , p, p) are sought to satisfy the continuity equation: from the flowfield(energy, species,momentum) at the
flowfleld(u
(3.5)
momentum
equations
- Aip as:
+ S"_' + S.,,t ,,
equation
- p")u;']
= -,...ki(p"u;) correction
+ S_,t
(3.7)
the velocity
equation.
(3.s)
correction Eq.(3.7)
equation. yields
taking
the divergence
and substituting
1
[ AtRT -A,(p"u:) Equation culated (3.9) from + S_,t + Ai(_)
47
- A,(p"Du"A,)](p - p") = (3.9) 1 1 pn 1 + ( RT------Z - RT-----:).-_ + A,[( RT" for the corrected 1 )p"u'] RT
can be solved
the equation
of state.
pressure, p'. The density(p*) is cal. The _elocltms(u, ) are computed from Eq.(3.8).
I10
16
Scalar("):
These update new the flowfield(u_*,#') B coefficient. satisfying the continuity equation (3.5) are used to
([
The temperature T*"
(3.10) flowfield(eneKy',
axe sought
to satisfy
the continuity
equation:
_(p
Subtracting Eq.(3.5) from
(3.11)
Eq.(3.11)
,....z
(3.12)
momentum
equations
are: (3.13)
- AO)u,'" from
= g*(u;*) Eq.(3.13)
-/kip" yields
Eq.(3.6)
Ui
- u_" = Du'[(A**
- A'_)u_"
+ H'(u_')-
H"(uT)
] - Du'[A,(p'"
-p')]
(3.14)
Here,
pn 1
=(_--_
Eq.(3.12),
o - A'_)u_" 17
1 +( RT"
Using of state. the corrected The
1 )p'u'] RT'"
is obtained from the
(3.15)
equation
pressure(p"), ')
velocities(u_
are computed
Eq.(3.15).
Scalar('"):
These update new flowfield(u_",p") the B coefficient. p'= . p_" satisb'ing the continuity equation (3.11) are used to
(_
The temperature
- ,;')""
T"" The is updated
= j"(")
calculated
(3.16)
the field
values
at the
completes
in the solution
of the equation
over the time-step. the left hand involve nature side of the corrected a convection term flows, PISO treatment preswhich and en-
In the present sure can ables does equations properly capturing not include
method, and
Eq.(3.15)
of supersonic
shock this
developed
non-iterative
pressure-based of continuity
and steady
is more state
fast-transient procedure
calculations, Bo).
the present
can be simplified
the coefficient
inlet
boundary, components
and
the
total
pressure
The
velocity components
by the the
extrapolation vorticity
velocity
by enforcing
boundary. velocity
the normal
grdients
variables
component from
variables
are extrapolated
4.
RESULTS
AND 18
DISCUSSIONS
To evaluate
the stochastic performed particle dense 4.1
the
present
dispersion
width
transport interactions,
model
and
simulation
of particle-turbulence dispersion
for the
spray Turbulent
sprays.
Particle
4.1.1
Turbulent experimental
dispersion
, the transit
controls
(87.0
times
undergo
transit
In this
experiment, The
turbulence calculations
particle
of x/m
equal
of 6.55 m/sec.
computational with
the resulting
dispersion
to time.
computations,
a single
to evaluate by using
representing
parameters
medium,
particles. These
experimental a single
results
with
computational
following and
deterministic prediction
the efficiency,
the overall
nearly-homogeneous Inhomogeneous
turbulent Turbulent
Dispersion 19
The next example problem is a particle laden round jet[29] in which the turbulence is inherently inhomogeneous. The turbulent gas-phasetransport properties are provided by using the k-e
tration profiles of the delta parcels, 10,000 the 200 computational axial tions. locations. Using model. function at various ate Figures SSF 4.2 and 4.3 show the particle and the SDWT model and SSF concen-
model
of the
factor, function
particles
in the
the distribution
smooth 50 parcel
model
regions
the correction
the uncertainty
and smooth
considerably. function
In Figure
case for 200 particle The 200 parcel factor In Figure with results case since
is very model
shows
sensitive
of increased shows
favourable These
agreement numerical
the SDWT
representing
dispersion function
turbulent
reacting
flows has
been
tested
by applying
it to predict
flow properties
in two axisymmetric,
confined, 4.2.1
swirling
in Fig.
4.5.
Experimenobtained droplet
tal from
data
for temperature,
components
were
measurement
of Khalil
a1.[36].
conditions
are given
in Table
1.
Liquid
kerosene
was
used
was fixed at 20.17. study, influence show vectors, two swirling of swirl numbers(S=0.72 and evaporation such 1.98) were considered characdroplet In the recircu-
In the present to investigate teristics. trajectories, lower lation swirl swirl zone Fig. the
& burning
4.6-4.8 velocity
temperature
contours
large
portion
of droplets
survive
in the
central
to evaporate droplets
region.
case(S=l.98), there,
evaporate With
intensive
burning
in this
increasing
spreading
increases
particle
In addition, swirl
corresponding
to the higher
is contributed
to recirculate at near
and to increase
the temperature
temperature
profiles
in Fig. result
siginificant
discrepancies
uncertainties locations
impingement with
However,
deviations combustion
of turbulence
temperature
profiles Radial
of axial present
model
velocities.
numerical velocity
size of central
zone based
the reverse
in Fig. by the
distribution areobtained
velocities.
estimated to the
axial solid
angular
flux. to the
velocity errors in
body
rotation
with the
a Rotatlng-Cup liquid-fueled
Atomizer combustor with a rotating spray. This cup small atomnearinitial process.
which
is capable
of producing
spray
of the
droplet
impingement intial
gas-phase
boundary
distributions) Liquid
from as fuel
kerosene
fuel/air out
at 0.0228. with
In present
study,
carried diameter(rk
condiSince and
the swirl
droplet
X,.J
the
limited some
conditions,
distributions
velocity.
case with
small distance
in droplet and
velocity
some error
droplet
Based
the droplet
70 % fuel mass flow rate). vectors, that zone, two and and
the smaller
predicted in Figure
velocity
temperature
are exist
4.16-4.18. around
layer
around results of
recirculation these
high-temperature
trapping and
droplets,
chemprofiles at four
reaction.
measured and
radial
of carbon results
oxygen(O2) agreement
qualitative
with
the
--,_j
data.
However,
quantitative chemical
differences
exist The
in the sigificant
profiles
of tem-
corresponding
to the
attributed turbulent
of swirl
as soot model.
monoxide(CO) regions,
associated
chemistry
mass
explained
existence to the
of the relatively
fractions(" rate
release
calculated
much
higher
than flame,
measured
values. between
and measured
temperature
decreases
4.3
Dense
Spray
4.3.1
Spray of Hiroyasu spray above. model Liquid and which Kadato[30] includes were coUison, through Spray tip used to
numerical models
described pressure,
into constant
spray.
cross-section
nozzle
physical
The mesh
spacing The
close this
Since
calculation
is sensitive solution.
resolution,
used
number 1500,
thisparcel number
10-4rn2/s3). The numerical resultswere insensitive to these initial values. The spray parcel distributionfor three sprays is shown in Figure 4.22. This
plot indicatesthat the spray tip penetration and the core length decrease with the increase of the gas density.Figure 4.23 shows the predicted and measured spray tip penetration versus time. It can be seen that there is reasonably good agreement between the predictionand the measurement . In the present computations, the
spray tip was defined to be the location of the leading spray drop parcel. It is necessary to note that a far-field spray penetration is not a sensitiveindicatorof model performance. Previous studies[26,31] indicated that a far-field spray penetration is mostly influencedby the turbulence difr-usivity. However, a near-field spray penetration could be more sensitive to the physicalsubmodels such as breakup and collision.Figure 4.24 shows the variationof SMD injector. The three soliddata at 65 mm computed with axial distance from the The
drop sizeisa time average over the spray cross-section at each axiallo-
cation. At the nozzle exit,the drop diameter is equal to the nozzle diameter, 0.3 ram. Generally these curves can be broken into two sections.Close to the injector, the drop sizedecreasesrapidlydue to drop breakup. Further downstream, the drop size increasesgradually due to drop coalescence. In the low gas pressure case(1. I MPa), the drop sizeremains relatively uniform afterinitial breakup region and then
increasesslightly in the far-downstream region.For the high pressure cases(3.0and 5.0 MPa), the drop sizeincreaseslargelyin far-downstream region,because higher gas densitiespromote collisions and coalescence.This trend is alsoobserved in the measuments. The predicted drop sizesat 65 mm axe qualitatively agreed with the
experimental data for all three cases. The discrepancy could be associated to the fact that the experimental sprays were pulsed while the computations assumed a constant pressure injectionforthe entirecomputational time period.
Hollow-Cone
Spray
The hollow-cone spray tippenetration data of Shearer and Groff[32] have been used forthe model validation. In the experiment, the liquidisinjectedintoquiescent 24
at P = parcels
550kPa.
The
numerical
timestep
is used
as
are used in the computation. the ms. flow The rate 0.0165 computational measured condition
is 60 degrees, about
and 0.58
mL/injection injection
of duration
set to the experimental in the early stage Figure the predicted indicate spray indicate tip toward
pictures
is listed
2.25 shows
distribution
vectors,
and
and measured
The numerical
results
has a relatively
in a hollow-cone vectors
spreading
of a vortex of spray.
the head
A substantial
of strong These
inward
of the cone near the injector compared the predictions quite favorably
flow patterns
observations[32]. spray
reasonably
the experimental
tip penetration.
4.3.3
and
Burning
Solid-Cone spray
Spray measurement dense nozzle into of Yokoda spray model. et. al. [33] Liquid highsprays domain and 44 and conditurbulent as
have
been
used
to
numerical hole
is injected nitrogen
through
conditions
for evaporating
in Table in radius
4. The and
diameter in length
A computational by a 21 radial
The mesh
spacing The
was nonuniform
number
quantities a solid
The upstream
boundary
is treated
as open boundaries. and the contours the spray however of the fuel
sprays. to time
penetrathe penethough
at early
become
nearly
constant
after
t = 0.2ms 25
due to evaporation.
Even
the
liquid
drop with
does respect
more,
the evaporated
penetrate spray
Comparisons
penetration agrees
in Figure
results[33]. of the
parcel mass
distribution, fraction
fuel mass
fraction,
at different of a burning
configuration ones.
agreement
a considerable ratio
the spray
is low and
Therefore,
properties
available
have From
been
developed
of dilute
the present
numerical
the following
1. Present fully
implementation
width
transport
model dispersion
demonstrated
the capability
improved
ettlciency
function model
prediction
of the
stationary
spray-
by comparison procedure
available
correctly the
and
yields
qualitative exist
agreement especially
at near-burner centerline.
at near-wall
combustion
chamber mainly
observed
in the results
are attributed
spray
ment model,
interaction,
of the k-e
dealing
streamline numerical
curvature. procedure,
the works
capabilities the
consistent
evaporating, such
and burning
sprays stress
by utilizing
the non-isotropic
model
as the algebraic
For non-evaporating, tions terms spray merical dispersion ing dense corporation models. show a reasonably of spray flame.
evaporating, good
burning with
spray
cases,
the predicresults in
experimental
penetration, To improve
configuration
and efficiencies
models,
works
must
include
transport
the incorporation
vaporization of atomization
of soot model
and further
and breakup
6.
RECOMMENDATIONS
The and
following
recommendations spray
are intended combustion for the following efficient combusting coupling Volume
extensions
physical
modelling
are need
approach
for multi-
Development pressure
multiple
correction
procedure of equilibrium
of Fluid(VOF)
non-equilibrium
chemitry
for effi-
reacting
flow calculations. of breakup conjunction and with coalescence multistep procedure. pressure correction
of turbulence of wall/droplet
modulation impingement 27
effects
by droplets.
process:
Incorporation
of supercritical
vaporization
model.
The
validations
and
the applications
spray
combustion
models
will
studied solution
non-evaporating,
evaporating,
and
burning
dense
flame
propagation
in a two-dimensional
spray
with
transient
droplet
analysis
of SSME
injector spray
atomization
and combustion
process.
to bipropellant simulation
of combusting
Acknowledgements
The
x.._./
authors
would
like
to express
their
appreciation comments.
to Dr.
T.S.
Wang
of is
participation Litchford
and critical
Acknowledgement PDF
for valuable
discussions
on the parcel
modby the
The authors
time supplied
Supercomputer
28
REFERENCES
1.
Asheim,
J.P.
and
Peters,
J.E.(1989), 391-398.
"Alternative
Fuel
Spray
Behavior",
J.
5, pp. "Ignition
Behavior 26th
of a Dilute Aerospace
Vaporizing Sciences
MulticomReno,
AIAA-88-0635,
Meeting,
Jan. W.A.(1983), Comb. "Fuel Sci., Droplet Vaporization and Spray Combustion" ,
Energy
291-322. in Droplet 171-201. and Combustion in Spray", Prog. EnVaporization and Combustion",
5.
Faeth, ergy
Transport
293-345. "The Formulation ASME of Combustion of Heat Models: Traa_,fer, 108, resolution pp. 633-
6.
W.A.(1986), to droplet
spacing,"
Journal
7.
Litchford, Turbulent
R.J.
"Efficient AIAA
Model
for
in Sprays", Drop
8.
O'Rourke, Alamos
Effects
on Vaporing
report
LA-9069-T. for Numerical 872089. "Calculation AIAA J., 18, of the Flow ProperCalculation of Spray
9.
O'Rourke, Droplet
P.J.(1987), Breakup",
Technical
10.
E1 Banhawy, ties
11.
Kim, Effects
C.P.(1991), Paper
on Spray
Combustion",
91-2i96,
Sacramento,
CA, June
W.P.(1980),
"Models
Combustion", Hemisphere
Prediction 379-422.
Method
for Turbulent
Flows,
13. Tuttle,
Fueled pp. 14.
J.H.,
Shishr,
R.A.,
and
Mellor,
A.M.(1976),
of Liquid 14,
Turbulent
Diffusion
Flames",
Comb_fion
Technology,
229-241. B.F. and Hjertager, With B.H.(1977), Emphasis "On Mathematical Formation Modeling of
Combustion
Special
on Soot
arid Com-
p. 719. "Chemical Kinetic Vol.10, Modelling pp. 1-57. Gerlands BeIof Hy-
15.
Westbrook, drocarbon
Prog. the
Comb,
Sci.,
16.
of Falling
Droplets",
52, pp.
17.
vaporization 26th
model
for spray
com-
Aerospace
Science
Meeting,
Nevada. G.L. et al.(1975), InL "Droplet J. Heat and Vaporization Mass Transfer, : Effects 18, pp. of Transient 1003Table", 1008. 2nd ediand
18.
Hubbard, Vaxiable
19.
Stull tion,
D.tL N.W.
H.(1974),
20.
Gosman, Liquid
E.(1981), AIAA
"Aspects
Simulation
of
Fueled
21.
22.
Shuen,
et al(1985). AIAA
Jets
: Measurements
and
"Solutions J.
Fluid
Flow
Equations
"Statistical Dispersion
M.M
in a Turbulent
Gas",
J. Comp.
3O
...........
25.
Crowe,
C.T.,
Sharma,
M.P.
and
Stock,
D.E.(1977), Eng.
Particle 325-332.
Source
in
Cell Method 26. Chen, Method Group Numer. 27. Chen, Phase NASA 28. Snyder,
Flows",
J. Fluid Y.(1990),
and Jiang,
Numerical Working J.
Combustion",
Eigth April
Liquid
17-19,
1990, to be appeared
in the Int.
Jiang,
Y.M.,
and
Shang, Code
H.M.(1991)
"MAST
- A Multi-
in Generalized
Coordinates",
J.L.(1971),
"Some
Functions
in a Turbulent
J. Fluid
N.,
Hirosawa, Laden
Yasuo, Round
and Jet",
Jotaki, AIChE
T.(1978), Journal,
in a Dust
509-519. H. and Kadato, T.(1974), SAE Paper "Fuel 740715. Numerical Model for Liquid Sprays", Droplet Size Distribution in Diesel
Hiroyasu, Combustion
31.
Du.kowicz, J. Comp.
32.
Shearer,
E.G.(1984), Proceedings
Effects Diesel
Conference,
New York,
H.(1988), Bulletin
"A Study
of Diesel
Spray
Technique, "Structure
of JSME, Pressure
of High
870598, R.J.
and Jeng,
for Turbulent
Dispersion,"
to be appeared
in AIAA
"x...i
36. Khalil,
K.H.,
F.M.,
and
Moneib,
H.A.,
"Effect
of Combustion Sixteenth
Air Swirl
on the
in a Cylindrical 31
Oil Fired
Furnace",
The
Combustion
Institute,
Pittsburgh,
PA,
pp.
J.H.,
"Experimental Combustion
Stud), Air,"
a Fuel
253-275, Chen,
38.
Kim,
H.M.,
Ziebarth, AIAA
J.P., Paper
and
Wang,
T.S.,
"Numer-
Sprays,"
92-0223,
30th Aerospace
Meeting, et.
Reno,
a/., "Solution
of the
Implicitly Phi.
Reacting 388-410,
Flow
Equa-
by Operator-Splitting,"
,7. Comp.
1991.
32
Table
1.
Gas-phase
B.C.
and
droplet-phase
I.C.
Inlet Air Temperature Droplet Sauter Droplet Number Dlatrlbution Mean Diameter Size Range of Size Range Velocity
x_../
Droplet
Velocity
Temperature
33
Table 2.
Nozzle diameter:
Case
Pgas (MPa)
Pgas (kg/m3)
Vinj (m/s)
Mini (kg/s)
SMD (_m)
1 2 3
Table 3.
Pgas (kPa)
Pgas (kg/m3)
Vinj (m/s)
VOUnj (ml/inj)
550
6.36
60
0.0165
60
Table 4.
Case
Pinj (MPa)
Pgas (UPa)
"lamb (K)
Minj (kg/s)
Atmosphere
30
3.0
900
0.00326
N2
30
3.0
900
0.00326
Air
34
J oil
Jl
J J
owmn
_J c_
Omll
r_ q_
om
lml
S.
qwml
35
_-
_ _ .._
I. @@
_m
17
[_-._ID]NOISU3dSICI
86
_ ,_lll
'_I_
tirol
L..
oj
M'O 180 I1_ _ 0OO
o_
O .'-
,.u
1_
01
ZTO
!
IOtO MO WC _lC 18 0BeL
,_m
Tm
o!
;.":
__ff
.. .'S_
!
ooc
37
// Lo
i
cl im In
o7
itgo ml_
r_ ulu
ow.i _
_'li
/
0
m olm
!
_a _0 NO _ WQ mO in
_wmml
c_
_a
"i
L.
_Dm
38
0
I.
ellUl
i MI m Iml _ m oral
03,3
9m
r.
Q m
A qP
o111 m m
I.
elm
*m-I
OC;;;Z: I_C ;|: 01: _" 00_,
_%..jJ
39
40
__v._
I'll
,f.
mu
,too io atom
m m
*m
_M
e...,
41
aC_ _111t111 111ttttlttt 11ttttllf _1111111 _1111111 ltt_tTtlftl 1111tttlttl !11_ttllttl _1111111 Itflftltttl fl
t111tt tttftt!
II
tt_ltt fttltt! _
Itlltt
tlflttt
ttlltt
tlttt!
11111!
IIttt_
Tttltt
Ittttt
ttlttt
tttttt
tttttt
tttttt
ttlttt
tt
ttt
ttlltf
l!
ttt
tllfft
tt
ttt
II
tl$1tt
It
tit
_t
Iltttt
||
ttl
_t
|||if|
|I
ill
_l
||tilt
II
If!
tt
tttftl
IF
III
_t
_m
Fe_tffet_tele_e.,
I. ......... ..........
_J >
0
_1_ff_ttl_tr_,,,,, .................
_flfl_ll_*,
.........................
,.... ,
....;,,'"
_m
,,
"'[
Trrrz,,,
42
ao
c_ C
r,1
_L
kl
eml
/
/
/
/
kl
q_
q_ fu
_u
pm
emm
43
X " O.II
m
"1
'( "
'3.24
'_
X " 0.52
o i
/0 _ -_
/O
el"
o
f_
w
L.J _J Z
0
(Ec_
0 0
O O / 0
,0
150O 2000
SOil
I000
s0o _ooo
(_ _s0o
20oo
Figure
4.9
Radial
profiles
of
mean
temperature(S=0.72)
o ..2-
X -
O.1l
o
.L-
X - 0.24
o
.L-
X "
0.52
f
o o
uO
c;-
o r_ r_
0 0 =;-
o
0
coo Z CE
crl o
3
O"
o 0
..J
0
e-t r,.- _.
0 o
o-
=;-
0 0 0 ,
20OO
50O
I{_
1500
2OO0
_OO
I0_1
150O
20OO
500
!000
50O
'EMPERATURE { K }
Figure
4.10
Radial
profiles
of
mean
temperature(S=1.98)
44
X-Oil
X "0.24
(_
X - 0.52 1
\ 0
0 0
O n."
_-
CDO Z
0
tn N o J C m O 0"
0 0 0 0
O
O" O
O
O 0
0 -I0.0
0.0
:5
C_ [o.o
O7
1
20.0
Figure
4.11 Radial
profiles
of mean
axial
velocity(S=0.72)
X " 0. Ii m
X - 0.24 m
X " 0.52 m
,.20
q@
0
tf_ m
r._q"
J (I:
0 0
O"
0 0
t
O O O
_ 0
I
-lO.O
I0.0
30.0
-10.0
AXIAL
:I
0 I0.0 30.0
Figure
4.12 Radial
profiles
of mean
axial
velocity(S--1.98)
45
j"
X - 0.ii
X " 0.24
- 0.52
_,
x-,,t O O
r_
w
c, _
C._o Z rr FU_
i
(_ r'_ n_ .
_
_-
//
/
i
/
10.0 211.11 0.0 10.0 20.0
/: _t/o
0.0 I0.0 20.0
0.0
TBNGENTIRL
VELOCITY(m/s)
Figure
4.13
Radial
profiles
of
mean
tangential
velocity(S--0.?2)
X - 0.II
X - 0.24
X " 0.52 \ 0
0
qD
_-
_0
lID
0 c5"
0 t I
0.0
15.0
30.0
0.0
15.0
30.0
0.0
15.0
30.0
TRNGENTIRL
VELOClT?(m/s)
Figure
4.14
Radial
profiles
of
mean
tangential
velocity(S=l.98)
46
C_ q_
47
Figure
4.16
Droplet
trajectories
in kerosene
spray
flame
fields
Figure
4.17
Velocity
vectors
in kerosene
spray
flame
fields
X--fl.23m
Figure
4.18
Temperature
contours
in kerosene
spray
flame
fields
48
II
X
O" I
8"'0
9"0
i,'O
C'O
I 0"0
r_
m m
qo
v
L_
X
O'I
8"0
9"0
_'0
_'0
0"0 c_
n_ cJ Q_
_..
_@
u
@
0"! U_
CN
8"0
9"0
_"0
C"O
0"0
,
!
OI
!
cxl
O'I
(o_I_]T_ONWL_iO
7_IOWU
49
'!
8"0
g'O
_'0
_'0
O"
0
! I t
o
I
o _'0
( 0"0
ql I
0"!
C_ u')
8"0
9"0
_,'0
f._
P41
6..
-c_
c_
-c_
X
m_
e_
c_pJ
O'T 8'0 9"0 _'0 _'0 0"0
6
I
o
_m
n,,,
X 0
o
0'! 8'_0 9"0 _,'0 _'0 0'0
( _/_
5O
r_
'_
ii
C_
X
( 0"{ @
llm
IJ
!
C_
@@ @@
x
I I
o'I
CD
9"0
0 9"0
i
aQ _'o
_'0
o'0
,r
CD
6=,
>u
@
9@
r_
=0 o'I
Ln
8"0 0
9"0
"'0
0"0
ql"
r_
X
O'I
8"0
9'0
_'0
_'0
0'0
51
.._.:.v,.,;-." .,.-..;+:._,_, :+
Q
?
5.0
I I I I I
MPa
""
'
.+Q 'L
+. "lep
"
.:.."
-.,; ,-:-_o.'-
.
.. .
+
I
3.0
I I I I
MPa
+..
,,
.. '
.- .
--_
. _ _
%
;._
:+.'_ f.+_m.
.'."._'_li".i_
D::: ?:f.i ;-:._; "'"_ " ::;_" " '_'""#"'" -" :-'_._'_,_.'_l,.,t_'._"._,_._. ""_"_ " :":_ .'_"'"_',-. _'" ,'."_: _._".,:P"'_-.". _..-. _._._._._._._._._._w_.'t_
,-, +. +;* q,,+o'o+,'" , o, ,
1.1 MPa
I I I I T T" I .... ' ....
000
0.0_
004
008
0.08
0.10
012
Figure
4.22
Spray
parcel
distribution
in a solid-cone
spray(_
= 3.0ms)
52
ii |1
\0
8 .a
<i-.
'
/o
!i
,_,..
,=.
[. "o\ _"-... ko
Oo\
o\
I.,
o_
f,.
"_.\.o
,,.} ,I
11 , I , _, ,, ., I , I , _" -_'_"
bO 0
OOt
Og
09
0'I,
O_
53
om
OOg
Og_
00_
0_I
00I
Og
(_ 9-3)
(]:MS
64
#"
/X
B, e..
::
::: :1 111111"1"'1
....
""
ms
,
........ o
,lee_eeeg_99e_leegv_
,," :: ::
....
:::::::::::::::
, -o
"::::::::::::::
/ X
0.88 ms
i i i I , ", ,
:::::::::::::::::::::::
::::::::
_IlllUI! II_lll
::::::",,
l_plllll Illltlll IlllIIIU lUll. Ill II II
::,,1:,,
::
.....
Ii
.........
X,
........ :,.',_-III _ _-_,., ......... ......... _J_tl_.! "i_.k_ .........
........ :_-_,_J_ :: _',_ 1: .........
....... _".:.
.......
:.!!!.,.
_,, .........
__ .:. .....
"_9/I .................
...........=..._
s.
loll
I.Ss
s.,_...,
ms
::::::::'
'::::1
!
......................
i ! i ! _ i ! !
Figure
4.25
Spray
parcel
distribution
and
velocity
vectors
in
a hollow-cone
spray
i!
0
,,,
O_
O_
_I
OI
(_)
NOIVU_.__ad
56
i
--......#
57
0
O_
0 0 0 0 0 0 0 0
1,i
Om
Ion o
N_
em
o
sm
o_
L.
4
0
I I , I 0
om
tw
Og
O_
OC
O_
OI
(_) _oIv_za_aa
58
_o
"-'
eL 0 k r,_
0
t
_'0 ' IO'O ' ' (30'0 ' ' IO'O--_0"080"0 'IO_O
i
00:0 IO0- _00-
_o _o!o ooo_00-_00-
..
8o
59
.i
!!I :
[00 00'0 I0'0-80"080'0 I0'0 00'0 I0"0-80'0-800 [0:0 000 I0"_-800--
6O
A. interaction
Interaction in the
Source governing
Sm,t
= _
p=l
Npmev,p/dV
(A.1)
NP
s.,., =
p-'l NP p=l
7;
(A.2)
s,,, = F_,{x,,_,_,,(h,L+ _) Z
_37rpdrp3 where Np[( Cp,d dTp
dvi . + (-_-l,,iu_)_)l}/m/
(A.4)
(-_-)p
Here, enthaipy To action dV denotes and improve source the the latent the term, volume heat
dvi
= u'_ +1 + u_ -
v_ t+l
+ Fbi
(A.4)
cell
and
hp and
L are
the
droplet
be treated
implicitly.
S"+_ ni,l
Here, S,, and R. are obtained
-S.u_ +1 + R,
(A.4) into (A.2):
(A.5)
by substituting
NP
S. = -_ _
p NP
N,m,/(
At + rp)
(A.6)
R_
1 = d'--V _
p
Npmp/(At
+ rp)(v'_
-- u i + Fbirp)
(A.7)
and
mp=
47rr3ppd volume
is the
particle
mass. on
The
parameters particle
S_
and
Ru
are
momentum at previous
control timestep.
quantities
depending
available
information
61
APPENDIX
B.
Numerics
of
Beta
Probability
Density
Function
The
mean
mixture
properties with
(era)
at any density
location function,
by
convoluting
functions
a probability
Cm(xi) where
_0 1
Cm(_)P(_,xi)d_
(B1)
--
fl Beta
calculated
function,
of the
following
relationship:
r(a)r(b)
B(a,b) - r(a + b)
Substituting Eq.(B2) into Eq.(B1) yields
(B3)
Cm(Xi)--" The numerator of However, chosen of the when the to be Beta mean pdf Eq.(B5) the quite when
Cm(_')(a-'(1--()b-]d( integrated errors These variance can by a trapzoidal unless axe due rule the to or
mixture
fraction(f) of the (0
is close conserved
to 0 or 1. To scalax(_). into
problem, function 0 to _s as
is expressed the
monotonic, and _
< _ <
1) is split The
sections:
to 1, where
stoichiometric
value.
property
,,_ is expressed
(B5) (B6)
dm.C
62
n = 1,...,N
(N = degree
of
the
polynomial)
Substituting
Eqs.(B5)
and
(B6)
into
Eq.(B4)
results
in
1 fo"
+
where equations a and b varying
f_[ E
n
,-R._ , ,4 _rt_a-I
(1-- _)b-'d_]
from the solutions of the
(BT)
transport
with
of Eq.(BT)
is simplified
as follows:
T = _'_(cm,,
n
- dR,,)
1 --
d_
+ X: dR. fo
n
I _'"+"-I(1
_)b-'d_
as the convenient expression:
(BS)
For using
IMSL
routines,
(BS)
can be transformed
T=
_[dm,,
,,
+(c_.--dR.)
f_ ,_"+_-'(I--_)b-_d_]
f_,_"+-'(1 - _)b-ld_
f01
_'_+"-_(1--_)b-'d_
= _'_[d,_.+(c,nn-d,nn)BETAI(s,a+n,b)]BETA(a+n,b)
(Bg)
In present mixture
study, property,
of the polynomial as
is used
as N = 6. Finally,
the mean
can be calculated
era(x,)
T BETA(a,b)
(B10)
63
APPENDIX
C.
Stoichlometrie
Relations
For
Hydrocarbon
Fuels
For
the
hydrocarbon-air
mixtures,
the
irreversible
single-step
reaction
is expressed
as follows:
C, tt, + (x +
Here, H20) and the n is 3.76. are oxidizer mass In the
(Cl)
and of fuel
determined remaining
transport the
fraction relations.
be obtained
from
ichimetric
(C2)
(c3)
+ Yco, + Yo= + Yf.)
(c4)
K2
"--
K3m
64
D.
Droplet
assumes spray
as a fiilly distribution
atomized with
of spherical by a finite
droplet-size ranges.
is represented and
a Rosin-Rammler have
implemented
distributions
the following
forms:
Nukiyama-Tanasawa
Distribution
Ot
D = A( SMO)
e_B(D/SMD)a
of computational of computational
to D +dD Sauter
respectively;
SMD
is the
mean
a,/3, A, and
B are experimental/determined
constants.
Rosin-Rammler
Distribution
dQ qDq-l -(D/X)' d-'D = X---'-'7 e X SMD where D, and Q is the fraction X and of the total
1
(O.2)
- r(1 - -)
q contained in drops of diameter
(D.3)
less than
volume
q axe constants.
65
SECA-TR-g2-06
'
ADAPTION OF MULTIDIMENSIONAL AND PARTICLE WALL-BOUNDARY CONDITION MODEL TO THE FDNS CODE GROUP PARTICLE TRACKING /
Prepared
by:
Y. S. Chen R. C. Farmer
Prepared
March,
1992
3313
Hunt._vi_le,
SECA-TR-92-06
Table
of
Contents
1 1 3 4 7 10 10 15 16 17 19 20
Chemistry
Equations Particle
Solution
.........................
..................................................... Tomahawk Tomahawk Nozzle Plume Flowfield Flowfield .......................... ........................... .............................
Liquid
Injector
Flowfields
Closure
Conclusions References
SECA-TR-92-06
Introduction
flow
CFD
model is
was
based
which
a pressure flow
plus
multi-corrector with
Navier-Stokes
models
and A
the
function
finite-rate flow
(Refs.
reacting a
simulation.
particulate method
efficient
developed particle
and not
size in
change
agglomeration
fragmentation
considered
this
investigation.
At
V
the which
onset had
of been
the
present
version of
of
FDNS
(FDNS-2DEL) The
too in
use,
because
it had on the
full
utilized.
objective single
reorder
the
automatic
and to
to
implement the
produce
a very study to
increase necessary,
simulated
particle
tracking
methodology
Governinq
_quation
The
gas-phase
governing
equations
of
the
FDNS
module
are
the
k_/
1
SECA-TR-92-06
equations fluxes in
the
of
equation, differences
density
particles
gas, to
the the
drag
considered
primary gas-phase
momentum as:
exchange.
equations
j.1(apq/at)
a[-pUig
_effGij(aq/a_j)
]/a_ i +
Sq
where
1,
u,
v,
w,
h,
k,
and model
ei
for
the
energy,
and
chemical
respectively. viscosity,
the given
transformation as:
j ut
Gij /Ueff
= a(_,n,_)/a(x,y,z) =
=
(u/J)
(a_i/ax
j)
(a_i/ax k) (a_/axk)/J
+ pt)/Oq
=: (p
The
source
terms
in
the
governing
equations,
Sq,
are
given
as:
0
--Px + V[/_eff(Uj)x] --
x + +
+
Dx Dy DZ Wp Dz
-py
--Pz
+
+
V[_eff(uj)y
V[g'.ff(Uj)z]
--
Sq
j-1
Dp/Dt
hv +
Hp
P(Prp(e/k)
_n
')
[(CI+C3Pr/_) Pr Cze]
where
Dx,
Dy
and
Dz
represent
the
drag
forces
and
takes
on
SECA-TR-92-06
values
between
N.
up, rate
wp
are
the
transfer heat
for
viscous
flux energy
Pr stands is written
turbulence
kinetic
production
as:
P: =
(_t/p) [ (Suj/Sx!
+ aui/Sxj)2/2
2(SUk/SXk)Z/3]
An
equation of
of
state,
p = p/(RT/_), Turbulent
is
used and
to
close
the
above aq,
equations.
Schmidt other 7.
Prandtl
numbers,
governing taken
and 4,
turbulence
model
constants
given
6 and
Finite
Rate
Chemistry
Model
For of
gas-phase reactions
chemical is
reaction in
modeling, terms of
a general the
system
chemical
written the
vij' ) and as
i-th
chemical
The i due to
net
rate
of j
change
in
the
molar as:
concentration
of
species
reactions
, Xij, is
written
Xi j == (vij,_vij)
[Kfjii(pai/Mwi
) _,ij _ Kbjii(Pai/Mui)=,ij']
and is
the
species by
production summing
_i,
(in
terms
of
mass
fraction)
calculated
reactions.
wi
= 1_!
ZXij J
SECA-TR-92-06
a! - mass p fluid
species
K_| = _j
forward
of of
reaction reaction
j j = Kfj/K,j
= backward
K,j = equilibrium -
constant fivij)} i
f! = Gibbs Kf =
A T a exp{-E/RT}
Finally,
the
species
continuity
equations
are
written
as:
p D,=! - V[(#ef_/o,)V_
i] = wi
where
o,
(assumed
to
be
0.9)
the is
number to of
for
ensure every
end for
marching
the This
numerical is
stability by
a CFD
model. in species
accomplished
allowable
concentrations, equations, fractions limited species turbulence for are changes source
species
continuity
species limits.
within so
are
similar reported
chemistry previously
approach (Ref.
detailed
submodels
PartiGulate-Ph_se
Equations
A in FDNS
particle of the
tracking and
was
employed
momentum particle
exchanges particle
between
phase
phase.
SECA-TR-92-06
calculated for
using
an
several fluxes
groups are
and
heat
equations. and
The
equations
constitute are
trajectory
temperature
history
written
DVI/Dt Dh_Dt
= C_
where
U! Vi td
= = =
Gas
Particle Particle
= Gas =
Recovery
Particle
Thermal-Equilibrium
=
G
(pp Up)/[12
Nu _/(Pr
dp)]
Constant
= = =
Pp
Nu
for are
drag
coefficient of
and
number and
for
functions
Reynolds given 9)
relative I0.
correlations correlation
are (Ref.
9 and as:
Carlson
Hoglund's
is written
SECA-TR-92-06
Cd "
(24/Re)
(I [I
+ +
0.15 M
Re '_7) +
(1 1.28
e")/
(3.82
e'1"_"'m)/Re]
Nu
(I
0.2295 [I +
where
0.427/_
"_
3.0/Re for
0"M. the
more
accurate is
but
more by
complicated Henderson
drag for
coefficient Mach a 1,
provided
That
is,
Ca -
24
[Re
{4.33 /(I
+ +
Re/S)
(3.65
1.53
TJT)
exp(-0.5*M/Re /(i
+ [i
(4.5 -
0.38a)
0.6
exp(-M/Re)]
where 0.48
M(7/2) For
1"2 is Mach _
the
molecular
speed
ratio.
=0.03
Re+
Re I/2.
1.75,
Cd =
[0.9
+ +
0.34/M 1.058
2 +
1.86(M/Re) I/2/S -
1/2
{2
+ /
2/S [i
2 + 1.86 (M/Re)'/2]
(TJT)
I/S 4}]
And,
for
<
Mach
<
1.75,
Cd
"
Cd
N*1 +
(4/3)
(M
I)
(Cd.=1.75
Cd .--1 )
which
assumes
linear
variation
between
and
1.75.
It motor
has
been
shown
that
the
law
performance and is
predictions possible
data.
applicability correlation
Nusselt
currently
researched
SECA-TR-92-06
Details
of
the
Particle
Solution
Method
In was flux
the
flow of flow
an
employed
to
and
to
evaluate
governing of the
the
(i.e.
energy
gives better as
similar
liquid made
layers,
shown of
by which
the
simulates made.
rocket
plumes
Particle specified wall. velocity Therefore, with turned be wall must due the
are particles in
treated which
by
using to
a the
stick particle
stick wall
a decreased size
more only
parallel to the
exchange This
assumed of
particles be in
model
particle data
interaction become
can
improved, order
experimental
test
available
In
the
2-D
version was
of
the
FDNS to test
flow
solver, the
Runge-Kutta trajectories. \ i
method After
employed
integrate of the
a thorough
integration
SECA-TR-92-06
it
was
found
that
the when
explicit the
scheme
can
give
diverged
solutions an model.
source
terms scheme
implicit For
integration
convenience, of motion.
X-component
equation
dX_dt dUp/dt
= Up = A (U Up)
where
A -
U = gas
Up = particle Xp = particle
In
finite
difference
form
the
above
equations
can
be
written
as:
Xp(_I)
Up(n_l)
Xp(") Up(n) I
(_t/2)
_tA [U
[Up (_1)
Up
+ Up (")]
(n+l)]
or Xp(ml) = Xp (") + _t/2 Up (_I) [Up (n) + atA [Up(_I) + Up (") ] U]/(I+_tA)
two of
equations source
the
the
step
size
4 time
across
that the
an
improved
scheme was
was
for in
particle The
scheme
acceptable
solutions, showed
detailed
comparisons pressure
FDNS-2DEL
analyses
unacceptable
losses
SECA-TR-92-06
predicted.
causing the form this of
Several
solution the
other
factors
initially the
suspected
of
energy
the
particle were of
suspected, were
before
effects the
found
poor good
FDNS-2DEL to
results
plume
and
method
poorly of
the
Resolving would
consumed to run
much
resources of
which test
otherwise
a wider
variety
cases.
k.j 9
SECA-TR-92-06
Test
C_s_$
major
test nozzle
case
which
was
studied
was of was
the
Tomahawk
solid
motor an are
analysis.
Consideration injector
flowfield These
coaxial following
paragraphs.
The
Tomahawk
Nozzle
Flowfield
The and is
Tomahawk in
nozzle Figs.
was test
with chosen
shown
RAMP
codes were
comparison
velocity,
number,
chamber,
made.
predicted (2250
somewhat the
(2470
K)
solutions
same the to
temperatures. the centerllne An in This flowfield includes centerline case) shape. and
temperature be due to
the
appears
oblique at is the
shock. centerline
apparent the is
non-zero
temperature of the
subsonic due to
portion
a very
inlet
boundary the
simulation flow
burning the
uniform
probably number,
sharp
temperature of the
contours particle
limiting
streamline
10
.......
.........
,-4
.,.-_
11
I-i t_ .,-t
12
t_ -,-4
13
"
,.
_J
.--
..
14
SECA-TR-92-06
to
gas
only
fills energy
the
nozzle.
the
static
enthalpy
equation which
the the
same stick
nozzle or
solutlons.
reflect there of is
give
The the
place at the
is
the hit
analysis spectrally
particles in order
which to
account
the very
plume
particle in
drag the
parallel nozzle,
streamlines reflection
transonic occur
of case
the being
does
not
considered.
The
Tomahawk
Plume
Flowfield
near
plume free
appears shear
to
be is
well
predicted defined
with and
layer
sharply
from of the
afterburning energy
reactions. were
Both considered.
the
energy the on
the
such exhibit
number
correction
turbulence
this
simulation.
the of
plume plane is
is
for
a long
distance rapid to and the when remedy the k-_ using in near be
with
crossing much
creating model
turbulent A plume
kinetic
energy
used.
similar code
exists The
SPF/II SPF/II
standard code is
JANNAF to switch
turbulence
models
between
15
SECA-TR-92-06
and run
far with
An
number if the
of Mach
test
cases
have
been
number
will
this when
behaved
used.
This must
plume future
FDNS-3DEL model It at
change of
other
computed the
afterburning so much of
existence
of
afterburning
apparent.
Liquid
Injector
Flowfields
version the
of
does the
not
treat (or
mass
particle particle
gas
the
treated is
with
particle instant
temperature of time
any
during
These for
restrictions spray
should
describing of
combustion. fuel if or
without mixture
one i.e.
not the of
one
lump
remains
being
analyzed. could be
energy treated
not
have
considered. such
models
arbitrarily not
supplying
vaporization spray
do The
supercritical
J
phenomena.
only
16
SECA-TR-92-06
work
at
all
is of
that the
the oxygen
heat lump
of
vaporization crudely at
evaluated the
at
temperature capacity oxygen described constant nature strong was were not of spray
high
the
llquid-like from by
lump a
single
calculation would that equation would the engine more heat be the of
considered, of
temperature real-gas
used,
stated
spray is
simulation
developing
because not
FDNS-3DEL
completed
Closure
calculation Several
of
flows to was
at make
best this
is
computed,
particle calculated
entire
these
conditions of the
results rocket
overall and
that in within in
flowfield to use
be
analysis, range
number
region.
option and
incorporated be analyzed
example,
should
computed
17
SECA-TR-92-06
be then be near
used be
to
calculate The
the
break
near
chosen field
between
balanced development
predict is also
large
plume
other
recommended.
".
18
SECA-TR-92-06
i.
CFD
code
was
developed the
and CFD
Tomahawk
nozzle
indicates
simulates
this
flow.
Particle the
mass
effects inclusion
are of
not
currently effects
included would be
in
current
these
relatively
simple.
More
test
cases of the
should
be
run
to
establish The be in
the
range
of of the
validity
require should be
to not
verify now
these The to
Compare
type
with
any
methods
carefully
parabolized computational
should
employed.
19
SECA-TR-92-06
Re ferences:
1. 2. 3. 4. 5.
Sanford, Launder
91-1789,
and Vol.
6. 7.
Chen Chen
and and
R.C. R.C.
Farmer, Farmer,
June
1991.
Vol.l,
8.
Wang, T.S., Y.S. Chen, and R.C. Investigation of Transient SSME with a Pressure Based Reactive Working Group Carlson Henderson, & Meeting, R.F. AIAA and P. Rates MSFC, Hoglund, J., Vol.
Farmer, "Numerical Fuel Preburner Flowfield CFD Method," 7th SSME CFD 1989. J., p. Vol. 707, 2, June of Nov. 1976. Gas/Particle to be presented 1964.
9,
D.J. C.B.
i0. II.
Meeting,
12.
Smith, S.D., Y.S. Chen, and B.L. Myruski:, "Model Development for Exhaust Plume Effects on Launch Stand Design," March 1991 Progress Report on NAS8-38472, SECA, Inc., Huntsville, AL, April 8, 1991. Dash, S.M., Version II 4, Science Shuen, J.S., et al, "The JANNAF (SPF-II), Volume I" Applications, Inc., and V. Yang, AIAA Standard Technical Princeton, Plume Flowfield Code Report CR-RD-SS-90NJ, July, 1990. January 1991.
13.
14. 15.
Paper-91-0078,
Freeman, J.A., R.C. Farmer in Rocket Engine Combustion Cooled Nozzles," November, Report on NAS8-38961, SECA, 1992.
and P.G. Anderson, "Heat Transfer Chambers and Regeneratively 1991 - February, 1992 Progress Inc., Huntsville, AL, February,
2O