ECN 4 Topic 8: Internal & Near Nozzle Flow Modeling Spray G Organizer: Chris Powell Argonne National Laboratory Ronald Grover, Presenter General Motors R&D General introduction Talk dovetails to Dan Duke’s presentation

Presentation Contents Modelling Approaches Simulation Techniques Boundary Conditions Meshing ECN 4 Simulation Results Injector coefficients Representative contour plots Next Steps for Spray G Encouraging more contributors!

Contributors Acknowledgement Three institutions contributed simulation results University of Massachusetts-Amherst and General Motors Maryam Moulai, David Schmidt (UMass)/Ronald Grover (GM) * Results published in SAE 2015-01-0944 Delphi Bizhan Befrui, Andreas Aye, Adrien Bossi * Representative modelling approach outlined in SAE 2015-01-0943 Argonne National Laboratory Kaushik Saha, Sibendu Som, Michele Battistoni *Results will be available in ASME paper ICEF2015-1112 in Nov, 2015 Politecnico di Milano Ehsanallah Tahmasebi , Tommaso Lucchini, Gianluca D’Errico *Details about the solver and its validity presented by (Salvador et al 2010) Acknowledgement GridPro Thank the contributors and acknowledge GridPro

Spray G Nominal Baseline Data/Conditions Fuel Isooctane Injection Pressure 20 MPa Fuel Temperature 90° C (363.15 K) Ambient Temperature 300° C (573.15 K) Ambient Density 3.5 kg/m3 - Ambient Pressure 5.97 bar (N2); 5.77 bar (Air) Injected Quantity 10 mg Injection Duration 780 µs (“actual”) Number of Nozzle Holes 8 Hole Diameter 165 µm Fully Included Angle 80° NOTES: Delphi simulation was conducted at injection pressure = 10MPa, atmospheric conditions (Spray G condition forthcoming) Results for the Spray G2 condition are also available but will be presented at a later time (teaser shown at the end of the presentation)

Modelling Approaches

Internal Modeling Codes Institution UMass/GM ANL Delphi PoliMi Code HRMFoam CONVERGE OpenFOAM Origin UMass Convergent Science OpenCFD OpenCFD (?) External Coupling NO Made best guess on PoliMi contribution UMass pulls base source from the extend branch of OpenFOAM External coupling = used these data to initialize an external spray solver

Approaches Institution/Code UMass/GM ANL Delphi PoliMi Liquid fuel Iso-Octane N-Heptane Equation of State Compressible Incompressible Compressible? Cavitation Enabled? Yes No Cavitation Model Homogenous Relaxation Homogeneous Relaxation - cavitatingFoam multiphase solver Inclusion of turbulent viscous energy generation? No? Turbulence RANS k-epsilon K-epsilon VOF-LES, SGS Smagorinsky-Lilly EVM, k-Eq. k-Omega SST Spatial Discretization 2nd order Fuel Properties REFPROP (input table) CONVERGE, Dymond et al. 1985 Reference (from REFPROP) Ambient Properties Ideal Gas Ideal gas Submerged ? Liquid/Gas interface Eulerian, diffuse-interface (i.e., pseudo-fluid) Eulerian, Mixture Model VOF Transport + interface compression n/a (?) Heat Transfer Enabled? No; fuel is isenthalpic No; isothermal Liquid/gas interface  make point that only computing one velocity vector per cell

Computational Domain Institution/Code UMass HRMFoam ANL Converge Delphi PoliMi Dimensionality 3 Cell Type Polyhedral with extrusion layer Cut-cell Cartesian Cubic Types Hexahedra Cell count (total interior and exterior) 414k 2.8 & 4.5 million 5.2 million 1.2 million (GridPro) Adaptive or Static Refinement? Static Needle motion? NO Geometry “Ideal” geometry with 3mm plenum “Ideal” geometry with mainly 9mm plenum 1/8 CAD seat geometry + 4mm plenum “Ideal” geometry with 9mm plenum Different code resolutions Engineering (UMass) Mid (PoliMi) High (ANL) Very High (Delphi) Mention that all contributions are ideal geometry Needle motion will be measured by ANL later this year … noted in future work

Geometry & Meshes Available on ECN Website Working with GridPro to finalize OpenFOAM grids for “measured” CAD geometry GridPro Mesh (9mm Ideal) Used by PoliMi Team

Comparison of “Ideal” & “Measured” Geometry Plenum # Feature Ideal Measured 1 Inner hole inlet diameter after inlet radius 3 0.165 mm 0.170 mm, all holes 2 Inner hole exit diameter before exit radius 4 0.173 mm, all holes 3 Inner hole inlet radius of curvature 0 mm 0.005 mm, all holes 4 Inner hole exit radius of curvature 9 Outer hole inlet reverse radius of curvature 0.040 mm estimate 0.080 mm 9 4 3 2 1 needle

Computational Meshes UMass/GM ANL 15 µm min. grid size

Computational Mesh domain

Time Accurate ROI Profile? Boundary Conditions Institution/Code UMass ANL Delphi PoliMi Time Accurate ROI Profile? Yes No Inlet Time varying velocity Constant Pressure Time varying Pressure Constant Pressure (?) Wall BCs L.O.W. Log. Law L.O.W. (?) Needle motion? Time accurate =used the ROI measurements to compute velocity inlet condition

Simulation Results

Hole Orientation & Numbering Spray G Convention CAD Geometry 2 3 4 5 6 7 8 1 Mention 5 needle guides and 8 holes  injectors are not fed symmetrically Axis of symmetry between holes 1 & 5

Summary of Submissions Terminology Fuel  liquid + vapor Ambient  non-condensable gas Injector Coefficients CD, Cv, Ca (individual hole & injector averaged) Contour Plots Goal: use these data to initialize external spray solvers 2-D Contour plots Plot X Axis Min, Max Y Axis Min, Max Hole exit cut planes -2 to +2 mm See following slide for example. Counter -bore cut planes See following slide for example Spray cut plane -4, +4 mm at z=2mm Variable Color Map Min, Max Density (kg/m3) & Velocity (m/s) 0, round max value to nearest multiple of 100 Temperature (K) Round min & max to nearest multiples of 100 Turbulent Kinetic energy (m2/s2) Void Fraction 0 to 1

Injector coefficients Hole # UMass/GM (ECN 3) ANL Delphi Individual hole 1 0.47 0.53 0.60 Single hole analysis 2 0.52 0.61 3 0.58 0.54 4 0.55 5 0.50 6 0.51 7 0.48 0.57 8 Overall Injector CD 0.59 Cv 0.73 0.78 - Ca 0.69 0.68 CD Highest flowing holes Lowest flowing holes Little consistency on hole-to-hole flow variability, possibly caused by: Mesh resolution and asymmetry Accuracy of flow separation Inlet radius resolution GM Measurement CD ~ 0.52

Mean Velocity (m/s) UMass/GM ANL PoliMi 1 2 3 4 5 6 7 8 Counter bore exit Both approaches are showing a bias in the velocity field towards the injector axis that initiates at the hole exit and propagates to the exit of the counter bore Some hole-to-hole variability is predicted

Internal Nozzle Flow Field DELPHI Inner bias of the velocity field is initiated at the inlet corner of the holes ANL PoliMi UMass/GM

Entrainment of ambient gas into the counter bore ANL UMass/GM DELPHI Ambient gases mix with injected fuel RANS results shows strong bias of fuel towards inner edge LES result shows fluctuations in fuel that fill counter bore

“Effective” Plume Trajectory Spray G Injector Geometry Specification ANL PoliMi UMass/GM DELPHI Velocity magnitude used to ‘eye-ball’ trajectory of spray plume ~32o ~35o ~33o ~33o 37o 37o 37o 37o

Temperature (K) UMass/GM ANL 1 2 3 4 5 6 7 8 Mention that no plots from PoliMi & Delphi here ANL results shows hotter ambient gasses (nominally 573 K) entrained into the outer edge of the counter bore

Turbulent Kinetic Energy (m2/s2) UMass/GM ANL 1 2 3 4 5 6 7 8 Turbulence is generated within the holes in the ‘separation’ region. The magnitude of TKE to the counter bore exit differs  possibly due to heat transfer (the team will test this)

Mixture Density (kg/m3) UMass/GM ANL 1 2 3 4 5 6 7 8 Mention that no plots from PoliMi & Delphi here PoliMi results show density with fuel and fuel vapor only so not directly comparabe Mixture density contours look qualitatively similar. Both codes show some hole-to-hole variation.

Void Fraction of Vapor UMass/GM ANL 1 2 3 4 5 6 7 8 Both codes show some vapor at the hole exit, but differ in the persistence of the vapor at the counter bore exit.

Additional representation of phase change Vapor Mass Fraction PoliMi Vapor initiated at the hole inlet and eventually condenses at the counter bore exit

Down steam (2mm) Liquid Distribution DELPHI ANL “Probability of finding liquid” Z=2mm Both simulation approaches show qualitatively similar structures

Teaser … Spray G2 Condition Gas Mass Frac. UMass/GM Liquid Vol. Frac. z=2mm ANL Flash boiling leads to increased plume angle  plume-to-plume interaction

Summary General findings Future Work All modeling approaches show a inward bias in the velocity field internal to the nozzle Modeling approaches were consistent in predicting entrainment of ambient gases within the counter bore Modeling approaches differ in the prediction of vapor persistence due to flow separation at the hole entrance & turbulence levels at the exit of the counter bore Very little consistency amongst models on hole-to-hole flow variability Future Work Exercise meshes from GridPro Improve CAD geometry with recent experiments Simulation of needle motion (ANL plans to measure) Full presentation of Spray G2 results Comparison to increasing experimental database Invite more contributors!