1. 1 ECN Topic 1.1 Modeling Results Compiled by David P. Schmidt UMass Amherst *Hanabusa Itchō (1652–1724)

2. 2 Organization Techniques Spray A Early Injection Spray A Main Injection Spray B

3. 3 Internal Modeling Codes InstitutionANLUMassCMTIFPENSandia CodeConvergeHRMFoamEulerian Spray Atomization IFP-C3DCLSVOF OriginConvergent Science In-house External Coupling No for A, Yes for B Yes * Also included a few results from M. Bode, Aachen RWTH for a simplified geometry and lower Reynolds number

4. 4 Approaches Institution/CodeANL Converge UMass HRMFoam CMT ESA IFP C3D Sandia Liquid fueln-dodecane Equation of StateIncomp.Const. compressibility (input error!), perfect gas Non-linear function of p,T, perfect gas Stiffened gas EOS, ideal gas Non-linear function of p,T, perfect gas Cavitation Enabled? YesNo for A, Yes for B NoYesNo Cavitation ModelHomogenous Relaxation --GERM- Inclusion of turbulent viscous energy generation? NNYYN TurbulenceRANS k-epsilon LES one-eq. eddy RANS SST k-ω LES Smagorinksy None Spatial Discretization 2 nd order2 nd / 1 st order1 st order2 nd orderCell-integrated semi-Lagrangian

5. 5 Computational Domain Institution/Cod e ANL Converge UMass HRMFoam CMT ESA IFP C3D Sandia CLSVOF Dimensionality33233 Cell TypeHex cut cellHex, polyhedral Quad, polyhedral HexEmbedded boundary Cell count (total interior and exterior) 2.5M early, 0.8M main 2.8M64K1M74M Needle motion?YesNo Yes (lift, no wobble) Yes GeometryCNRS yearly, Same as last year for main Same as last year Last year(2D) CNRS scanBased on description

6. 6 Boundary Conditions Institution/CodeANL Converge UMass HRMFoam CMT ESA IFP C3D Sandia CLSVOF Time Accurate ROI Profile? Yes InletTime varying total pressure Time varying velocity Time varying static pressure Fixed static pressure Wall BCsL.O.W.SlipNo-slipSlipNo-slip Needle motion?YesNo Yes (lift, no wobble) Yes

7. 7 Spray A Early Injection Submissions by – Sandia, Marco Arienti – ANL, Michele Battistoni, Qingluan Xue, Sibendu Som – IFP, Chawki Habchi, Rajesh Kumar

8. 8 Early Injection A subset of the contributors submitted results that focused on the details of: – early needle lift – air ingestion – the start of fuel flow – trapped air

9. 9 Needle-Filling Movies by IFP Noted how fuel follows the needle surface into the sac

10. 10 Simulation starts at 200 µs (4 µm lift) Ends at 400 µs 210675 LIFT CNRS stl file for sac and nozzle ANL Results

11. 11 Notes: 1.Based on End-of-Injection studies, the sac is likely to be filled with ambient gas, resulting from a previous injection event. 2.Spray A penetration starts at ~310 µs after command 3.Needle rises slowly until ~310 µs after command; afterwards steep rise 11 ANL Results

12. 12 ANL Results

13. 13 ANL Results: Mostly dissolved gas, very little vapor

14. 14 ANL Results:

15. 15 t = 2.6  st = 76.5  st = 173.3  st = 187.6  st = 235.5  s t = 315.5  st = 325.0  st = 330.0  st = 320.0  s Axial velocity – early opening transient 40 m/s -40 m/s 0m/s t = 317.5  s t = 158.8  s Axial velocity Sandia Results

16. 16 520 m/s 0m/s t = 390.0  s t = 336.2  st = 334.8  st = 337.2  st = 332.7  s t = 338.6  s under-expanded jet Axial velocity – opening transient trapped gas t = 390.7  s Axial velocity Sandia Results

17. 17 time [ms]mass flow – AIR [g/s] mass flow – FUEL [g/s] 0.002557532.40E-060 0.1587736-1.23E-030 0.1732709-3.82E-030 0.1876011.08E-030 0.198372.05E-030 0.2355461.42E-030 0.280205-2.83E-030 0.3155186879.29E-070 0.317500687-6.54E-060 0.319629-8.46E-040 0.3249891.12E-030 0.3296547.11E-030 0.33265562.71E-020 0.3347924.90E-020 0.33614971.41E-010 0.337172.00E-010 0.338061.69E-010 0.3390211.34E-011.37E-01 0.33974943.73E-031.442248 Mass flows – opening transient time [ms] The flows are calculated by integration of the axial flux over a cross-section of the orifice located just before the exit Sandia Results

18. 18 Trapped gas at t = 390.7  s Estimated gas volume ~ 3 10 -7 cm 3 = 0.0015 V sac The average density of the gas inside the bubble is ~ 0.2 g/cm 3 The estimated residual gas mass is therefore 6 10 -8 g Z Y Sandia Results

19. 19 Observations Computationally, we can predict air ingestion and delay due to sac filling Unresolved: Is fuel filling the sac as a quasi one-dimensional process or does the fuel follow the needle? Prediction of under-expanded jet Unresolved: Prediction of trapped air

20. 20 Spray A Main Injection Submissions by – ANL, Qingluan Xue, Michele Battistoni, Sibendu Som – UMass, Maryam Moulai, David Schmidt – CMT, Pedro Marti Gomez-Aldaravi, Raul Payri – IFP, Chawki Habchi

21. 21 Mass Flow Rate

22. 22 Coefficient of Discharge

23. 23 Coefficient of Velocity

24. 24 Coefficient of Area

25. 25 Spray A Snapshots All taken at t = 0.75 ms

26. 26 Velocity, Transverse (x-y) View ANL UMass CMT IFPEN

27. 27 Velocity at the Exit ANL IFPEN CMT UMass Aachen

28. 28 Density, Transverse (x-y) View UMassCMTANLIFPEN * Due to input error, the UMass density was too low.

29. 29 Idea from Pedro Marti Perhaps viscous dissipation in the turbulent boundary layer raises the temperature and decreases the density of the liquid near the wall

30. 30 Density at the Exit ANL UMass IFPEN CMT Density at nominal exit t,p : 741 kg/m 3 * Due to input error, the UMass density was too low.

31. 31 Temperature, Transverse (x-y) View ANL UMass IFPEN CMT

32. 32 Temperature at the Exit UMass IFPEN ANL CMT

33. 33 Turbulence, Transverse (x-y) View ANLUMass IFPEN CMT

34. 34 Turbulence at the Exit ANL UMass IFPEN Aachen CMT

35. 35 Spray B ESRF geometry is incomplete, but more accurate Initial simulation with the Phoenix tomography ESRF Phoenix

36. 36 Contributions InstitutionsUMassANL PeopleMaryam Moulai, David Schmidt Michele Battistoni, Qingluan Xue, Sibendu Som ApproachEulerian cavitation, spray development Adaptive mesh Eulerian-Eulerian CodeIn-houseConverge Needle motionnoneyes GeometryReduces ESRFPhoenix

37. 37 ANL Contribution

38. 38 Hole 3 Hole 2Hole 1 Spray B: Hole slices of liquid volume fraction and velocity ANL Contribution

39. 39 A closer examination of hole 3 ANL Contribution t = 100  s ASOI

40. 40 Spray B Rough geometry from Phoenix STL is smoothed, slightly Roughness causes large spatial pressure fluctuations in nozzle, especially near exit Hole 3 produces a wider spray than the other holes UMass contribution Hole 1Hole 2 Hole 3

41. 41 Connecting internal flow to external spray Hole 1 Hole 2Hole 3 UMass contribution Mass fraction of gas

42. 42 Cavitation? Volume fraction of fuel vapor As the holes diverge near the exit, trace amounts of vapor appear No cavitation predicted at nozzle inlets UMass contribution Hole 1 Hole 2 Hole 3

43. 43 Main Injection Conclusions Some differences between results with new and old 675 geometries With no slip walls, temperature variation due to viscous energy generation Stiffened gas EOS is overly compressible, UMass results included incorrect density input Mostly, matched experimental Cv,Ca, Cd Roughness effects in Phoneix spray B geometry Matched mass flow trends with CSI spray B geometry Wide spray from hole 3, with perhaps, a little cavitation, just at the exit