coupled nozzle flow and spray simulations Alan Kastengren Argonne National Laboratory Qingluan Xue Argonne National Laboratory Julien Manin Sandia National Laboratory Chawki Habchi ID: 911097
Download Presentation The PPT/PDF document "Topic 1.2: Near field spray developmen..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Topic 1.2: Near field spray development and coupled nozzle flow and spray simulations
Alan Kastengren*:
Argonne National Laboratory
Qingluan Xue:
Argonne National Laboratory
Julien
Manin
:
Sandia National Laboratory
Chawki
Habchi
:
IFPEN
Slide2Modeling Part
Facilitate
dynamic coupling of in-nozzle flow and external spray approaches
Compare the different coupling approaches of in-nozzle flow and external spray
Study
the capability of the different modeling approaches (
Lagrangian-Eulerian
,
Eulerian-Eulerian
) and CFD frameworks (RANS, LES, DNS) for the simulation of the primary atomization and the cavitation
Encourage high-fidelity simulations near nozzle sprays and liquid jet atomization
Slide3InstitutionModel approachCFD Code Compressibility
Turbulence model
Argonne
Single-mixture
EulerianCONVERGEIncompressible liquidCompressible gas (EOS)RANSStandard k-epsilonCMTSingle-mixture EulerianOpenFOAM (ESA)Compressible liquid (function of pressure & temperature) and gas (EOS)RANSSST k-omegaIFPENTwo-Fluid multi-species EulerianIFP-C3DCompressible liquid and gasLiquid : Stiffened gas EOS Gas : Perfect gas EOSLES SmagorinskySandiaMultiphase/multicomponent with transcritical thermodynamics Raptor (CRF-Sandia)Compressible liquid and gasLESdynamic SmagorinskyUMassSingle-mixture EulerianOpenFOAM (HRMFoam)Compressible liquid and gasLESone-eq. eddy
Model description
Slide4Spray A
Slide5InstitutionMesh resolutionNeedle motionInitial conditions inside injector
2-D or 3-D
Argonne
10
μm inside nozzle and first 6mm; fixed embeddingTransient needle motion – with wobbleFuel filled sac and orifice; 150 MPa and 343 K in Sac and orifice3-DHeight=200mm; D=50mmCMTAbout 30 μm uniformFixed needle lift at 100 μmFuel filled sac and orifice: 150 MPa and 343 K in sac and orifice2-D axisymmetricHeight = 12 mmWidth = 6 mmIFPENNon-uniform hexahedra : Minimum cell size=(5µm in the hole,2µm in the seat, 15µm near-nozzle region)With needle motion - Without wobble(LVF,P,T)=Chamber-needle-seat : (0.99, 150 Mpa, 343K)Sac and hole : (0.0001,2MPa,, 343K)Chamber : (0.0001,2MPa,303K)3D configuration including:The needle (15 cm) control volumeChamber (bore=6 cm, height =6cm)Sandia4 μm close to injector; stretched fixed messPlug-flow BC used a few nozzle diameter upstream of nozzle exit; needle not simulatedGas filled sac and orifice; 6MPa and 900K3-Dheight=90mmDiameter=45mmUMassAbout 12 μm cross nozzle and
50 μm for sprayWithout needle motion; fixed needle lift at 100 μmGas filled with sac and orifice at 2MP and 346K3-DHeight = 9 mmDiameter = 2 mm
Simulation set up (Spray A)
Slide6InstitutionGeometry (210675)Total number of computational cellsComputational time for first
100
microseconds (CPU hours)
ArgonneLow resolution4.50 millions 12,288CMTLow resolution0.09 millions100IFPENHigh-resolution (CNRS scanned)1.10 millions6,144SandiaLow resolution50.00 millions 140,000UMassLow resolution 2.80 millions50,000Spray A: Computational time (first 100µs)
Note: the computational domain size, initial conditions, and needle motions are different for the institutions
Slide7Computational domain (Spray A)
CMT
IFPEN
Sandia
The inflow boundary conditions are specified at upstream for all simulations
Argonne
UMASS
Slide8Spray A (675) modeling results
Mass flow rate
at the nozzle exit
Fuel spray penetration vs. time
(0.1% liquid mass fraction)Contour plots of projected density in the 0°plane at 0.1 and 0.5 ms after actual SOIProjected fuel mass/density (ug/mm2) profiles across nozzle axis at 0.5 ms after SOIx = 0.1, 0.6, 2, 6, and 10 mm downstream to nozzle exit@ x= 0.6 and 6 mm, @ 0.75 ms and 1 ms after SOI2D contours of liquid volume fraction at x = 0.1, 0.6, 2, 6, 10 mmTransverse integrated mass profiles at 0.5 ms after SOI: x = 0.1 mm, 0.6 mm, 2 mm, 6 mm, and 10 mm Mean droplet size (SMD) at x = 1, 4, 8 mm at 0.5 ms after SOIMean SMD at the above axial positions vs. axial positionDistributions of SMD vs. radial position at the above axial positions
Dynamics: peak projected density and Full Width Half Maximum (FWHM) of distribution at x = 0.1, 2, 6 mm from nozzle for entire duration of the injection event (in intervals of 20 μs)
Slide9Spray A:
Predicted mass
flow rate at nozzle exit
All simulations predict similar mass flow rates as the ROI from CMT’s Virtual generator
The transient process at the start of injection is different from the modeling groups since the minimum or fixed needle lift used are differentStarting from empty sac, initial LES penetration for IFPEN (up to 10 μs) compares well with the ROI from CMT’s generator, but it is oscillating; LES for UMass under-predictsEnlarged @ first 0.1ms
The start of injection is defined as first evidence of liquid outside of the nozzle exit
Slide10Spray A: spray
penetration
Note: Sandia simulated conditions (Ambient=N2@900K
and
60bar); Standard conditions for topic 1.2 are (Ambient=N2@303K and 20bar) The data measured by Argonne and SandiaArgonne, CMT and IFPEN: The liquid penetration defined by 0.1% mass fractionSandia: Mixture-fraction of 0.79 cut-off for the dense flow modelInitially Sandia simulations over-predict penetration but the liquid length is lowerCMT predict higher mass flow rate but lower penetration compared to Sandia Argonne over-predict penetration due to high initial minimum needle lift
Slide11Mass/area(μg/mm2)
RANS
LES
Spray A: projected mass
density – 0.1ms ASOI
X-ray
data
Argonne
CMT
IFPEN
All simulations predict quite different liquid jets
A dense liquid core near nozzle region is captured by all simulations
All simulations
over-predict liquid penetration
RANS models tend to be overly diffusive compared to LES
LES over-predict the core
length and some averaging to be comparable to x-ray results
Sandia
Slide12Spray A: projected mass
density – 0.5ms ASOI
X-ray
Argonne
CMT
IFPEN
Mass/area
(
μ
g/mm
2)
At steady-state, the observations are consistent with earlier injection time (0.1ms)
LES simulation captures the fact that spray is off-axis, because IFPEN consider a more detailed geometry with finer mesh
G
eometry
RANS
LES
UMASS
Slide13Spray
A: IFPEN calculation @ 0.5ms
Velocity vector plots in 3-D (left) and 2-D show the asymmetric distributions
Geometry effect on downstream spray predicted
Slide14Spray A: projected mass
density – 0.5ms ASOI
Argonne
CMT
IFPEN
RANS simulations tend to over-predict radial diffusion;
LES simulations tend to over-predict liquid core length
IFPEN consider a more detailed geometry with finer mesh and predict off-axis spray
G
eometry
RANS
LES
Sandia
@0.25ms
UMASS
Slide15Spray A:
2-D (YZ) Liquid Volume Fraction (LVF)
Argonne
Very high LVF (close to 1) are seen from Tomographic reconstruction x-ray data
RANS results are symmetric due to symmetric geometry usedAsymmetric liquid jet is shown from LES due to asymmetric geometry used Low LVF (max=0.65) obtained LES results due to the presence of gaseous cavitation in the [0.1,2] mm region.
X=0.1mmX=0.6mmX=2mm
RANS
0.5ms ASOI
LES
IFPEN
Sandia
Tomographic
Reconstruction
from x-ray
G
eometry
IFPEN
Others
Slide16Spray A:
Radial profiles of projected density
density 0.5ms ASOI
0 degree view from x-ray is asymmetric at all locations
Argonne and CMT use the same geometry, and fair comparison between the EE model is obtained Sandia and UMass LES predict quite symmetric with the geometry usedAll simulations capture the shape of the radial profile, however, significant differences in the peak values and tails (IFPEN GERM model predicts some gaseous cavitation in the [0.1,2] mm region)Note: The profiles from simulations are shifted to have the same phase as the x-ray data
0.5ms ASOI
Slide17Spray A: Radial profiles of projected density
Multiple realizations are necessary for LES to get smoother profiles
Significant difference between Argonne and CMT, although the similar EE model used (some difference in details)
Suggest
to use the new geometry for the RANS cases in future work0.5ms ASOI
Slide18Spray A: Projected
fuel
density
0.5, 0.75,
1.0 ms Argonne RANS predicts liquid jet reaches steady-state after 0.5 ms ASOI similar than x-ray experiments.IFPEN Single-realization LES shows liquid jet reaches steady state after 0.5 ms, and may be at 0.75 ms
LESRANS
Slide19Spray A:
Transverse Integrated Mass (TIM)
TIM increases with axial distance, all the simulations capture this trend
All simulations over-predicted the TIM which may be due to over dispersion in the radial direction
TIM at nozzle exit corresponds with the mass flow rate trends from topic 1.1CMT profile is not monotonic perhaps due to initial conditions0.5ms ASOI
Slide20Spray A:
peak density & FWHM vs. time
Similar peak values predicted by ANL and CMT at 0.1mm and 2.0mm, IFPEN under-predicted
We see variations from RANS results from CMT due to pressure wave and compressibility effect
Fluctuations are expected from LES @ SGS which may enhance primary atomization
PeakprojecteddensityFWHM
Slide21Spray A
:
SMD
at first 50
μs for IFPENSub-microns droplets observed for IFPEN simulation which is consistent with x-ray measurement
Slide22Spray A:
EE vs. LE at Argonne
Projected mass density along spray axis
Mass-averaged velocity along axis
Eulerian model is better than traditional Lagrangian approach in the near nozzle region Lagrangian simulations: 62.5μ
m minimum resolution, blob injection model, 300,000 parcelsX-ray dataEulerianLagrangian
Slide23Spray B
Slide24Spray B: Internal flow to external sprayHole 1
Hole 2
Hole 3
UMass contribution
Mass fraction of gasHole # 3 has wider distribution Phoenix geometry
Slide25Spray
B: spray velocity at 100
μ
s ASOI
Hole #1Hole #2Hole #3Argonne contributionPreliminary results for 3-D transient simulation at 10 microns resolutionZoomed-in view shows holes #3 has wider spray compared to other holesThe effect of geometry, needle wobble, grid alignment and
resolution to be investigated in the future workCONVERGE reduced High-resolutiongeometry
Slide26Conclusions
Eulerian
model is suitable for the coupling nozzle and near-field dense spray
For the first time, quantitative data on spray morphology in the near nozzle region was extracted from simulation
Significant insight from different simulation approaches were obtained when compared against experimental dataThe Eulerian simulations seem to work better than Lagrangian simulations in the near nozzle regionA few key challenging areas for modeling:Liquid compressibilityTemperature variationGas in the sac
Grid resolution2-D results are good and computationally cheap, needle off-axis motion and geometric asymmetries need 3-D simulations
Slide27All Contributors:Argonne National Laboratory: Qingluan Xue, Michele Battistoni, Sibendu SomCMT - Motores Térmicos: Pedro Martí, Raúl Payri
IFPEN:
Chawki
Habchi, Rajesh KumarSandia National Laboratories: Guilhem Lacaze, Joseph C. OefeleinUniversity of Massachusetts Amherst: Maryam Moulai, David P. SchmidtAcknowledgements