1. ECN4 – Diesel Combustion REACTIVE FLOW ANALYSIS
Coordinator: José M. García-Oliver, CMT
September 6th, 2015

2. MOTIVATION AND OBJECTIVES EXPERIMENTAL ANALYSIS
CONTENTS
MOTIVATION AND OBJECTIVES
EXPERIMENTAL ANALYSIS
EXPERIMENTS VS CALCULATION
Inert flow
Reacting flow – Nominal spray A
Reacting flow – Low reactivity cases
CONCLUSIONS

3. MOTIVATION
Existence of a database of PIV measurements by IFPEN available for CFD validation
900K, 22.8kg/m3, 0%O2
900K, 22.8kg/m3, 15%O2
Ref: Eagle et al ILASS 2014

4. Flow analysis @ECN2 MOTIVATION Large scattering in modelling results
Boundary conditions?
No analysis of velocity/reactive penetration data

5. MOTIVATION Reacting spray tip penetration
Detailed knowledge on the transient dynamics of a reacting jet from high-speed schlieren imaging
Mixture shifts from inert to reacting within a transient flow, so there are deviations compared to the well-known inert spray evolution
How do tip penetration and radial dispersion evolve with time?

6. MOTIVATION Reacting spray tip penetration
Acceleration of the reacting vs inert spray in terms of the Sr/Si ratio defines some stages in spray tip evolution
Autoignition small ‘bump’ in the curve followed by a stabilization period
Acceleration compared to the inert case
Quasi-steady: Penetration speed is amplified in a constant factor compared to the inert case
SOC
window limit
window limit
Ref: Payri et al Applied Thermal Engineering 2015

7. MOTIVATION Radial dilation vs flame structure
Radial expansion as detected by schlieren starts at around the OH* LOL
Significant increase in spray width is observed for Spray A conditions R
Inert
Quasi-steady
Transient
Ref: Payri et al Applied Thermal Engineering 2015

8. OBJECTIVES
Further investigation on the flow characteristics under reacting conditions
PIV field analyis
Comparison vs inert evolution
Assessment of the capability of CFD models to reproduced the flow characteristics
Reacting spray penetration
Velocity field

9. MOTIVATION AND OBJECTIVES EXPERIMENTAL ANALYSIS
CONTENTS
MOTIVATION AND OBJECTIVES
EXPERIMENTAL ANALYSIS
EXPERIMENTS VS CALCULATION
Inert flow
Reacting flow – Nominal spray A
Reacting flow – Low reactivity cases
CONCLUSIONS

10. EXPERIMENTAL ANALYSIS
Conditions
Experimental information from IFPEN databases
PIV
LIF (ECN3)
Schlieren reacting tip penetration from ECN3
CONDITION Ta
[K]
rhoa
[kg/m3]
XO2
[%]
Pinj
[bar]
InjDuration [ms] SA
900
22.8
15/0
1500
1.5 T2
800
7.3 EX
780
14.8
5.0

11. DESCRIPTION OF REACTING FLOW
NOMINAL Spray A
PIV derived data

12. DESCRIPTION OF REACTING FLOW
NOMINAL Spray A
Evidences of:
Increased spray tip penetration
Flow acceleration
Radial dilation

13. DESCRIPTION OF REACTING FLOW
NOMINAL Spray A
Evidences of:
Radial dilation
Flow acceleration
Radial dilation

14. DESCRIPTION OF REACTING FLOW
T2/EX
Longer ID/LOL
Flow before/after LOL
Increased radial expansion

15. DESCRIPTION OF REACTING FLOW
Radial dilation
Radial dilation in the velocity field is found downstream of LOL, i.e. high temperature zone SA
OH/355 LIF
Ru,reac
Ru,inert
LOL
OH*

16. DESCRIPTION OF REACTING FLOW
Radial dilation
Radial dilation in the velocity field is found downstream of LOL, i.e. high temperature zone
Dilation increases with lower temperatura and density EX
OH/355 LIF
Ru,reac
Ru,inert
LOL
OH*

17. MOTIVATION AND OBJECTIVES EXPERIMENTAL ANALYSIS
CONTENTS
MOTIVATION AND OBJECTIVES
EXPERIMENTAL ANALYSIS
EXPERIMENTS VS CALCULATION
Inert flow
Reacting flow – Nominal spray A
Reacting flow – Low reactivity cases
CONCLUSIONS

18. MODELLING CONTRIBUTIONS
ANL
USYD
CMT
TUE
POLIMI
ETH Zurich
UNSW
Code name
CONVERGE
OpenFOAM (mmcFoam)
OpenFoam
OpenFOAM
OpenFOAM with LibICE
STAR-CD 4.20
FLUENT 14.5
TURBULENCE
Turbulence model
LES
RANS Standard k-ε
RANS k-ε
RANS Realizable k-ε
Sub-grid or turbulent scalar transport
Dynamic structure
Smagorinsky/Sparse-Lagrangian
gradient transport
SPRAY MODEL
Used Lagrangian discrete phase model (Y/N)? Y
Equivalent gas jet
Y,N
Injection
Blob,
Gas-jet
Blob
Atomization & Breakup
KH-RT
None
KH-RT (with break-up length)
KH-RT (w/wo break-up length), Huh, KH, Reitz-Diwakar, ...
KH-RT (without break-up length)
Reitz-Diwakar
No breakup model
Collision
NTC collision No
O'Rourke
No collision
Drag
Dynamic
Standard model
standardDragModel (OpenFOAM)
Dynamic,…
Strokes-Cunningham
Evaporation
Frossling
standardEvaporationModel (OpenFOAM)
Spalding
Heat Transfer
Ranz-Marshall
Dispersion
Stochastic
Stoachastic DRW

19. MODELLING CONTRIBUTIONS
ANL
USYD
CMT
TUE
POLIMI
ETH Zurich
UNSW
Code name
CONVERGE
OpenFOAM (mmcFoam)
OpenFoam
OpenFOAM
OpenFOAM with LibICE
STAR-CD 4.20
FLUENT 14.5
GRID
Dimensionality
3D domain, 80x80x80 mm 3D
2D axisymmetric
Type
Structured AMR
unstructured
Block structured
Block structured Cartesian
Cartesian 2D
Grid size range (mm)
mm to 1mm
0.01 mm to 1 mm
0.5x0.25
min 0.2x0.2 mm max 1x1 mm
0.1 mm mm
0.15 x 0.42mm to 4x11mm
Total grid number (#cells)
2.00E+07
6.36E+05
4.67E+04
2.33E+04
2.30E+04
1.60E+04
9.54E+03
TIME ADVANCEMENT
Time discretisation scheme
PISO
Implicit
PIMPLE
SIMPLE
Time-step (sec)
Variable with max Courant number equal to 0.75
variable with max Courant No equal to 0.5
1.E-07
1.E-06
4.E-06
Identical injection rate/boundary conditions for all institutions
Details on combustion model/chemical mechanisms in the following sections

20. NOMENCLATURE FOR PIV ANALYSIS
Width
Same methodology for both PIV and CFD
Normalization to accomodate for nozzle differences (675 vs 678)
ucl
Area
R5%
0.05ucl
Area~Vdot= 0 𝑅 𝑢∗2𝜋𝑟 𝑑𝑟
Peak value

21. ETH > TUE-UNSW > CMT-Polimi
INERT CASES
Spray tip penetration
Overall good agreement
UNSW deviations in the early penetration zone
ETH deviations during the whole time period
ETH > TUE-UNSW > CMT-Polimi

22. INERT CASES Flow variables Overall good agreement
ETH/UNSW highest differences in volumetric flow
time = 1500 us

23. MOTIVATION AND OBJECTIVES EXPERIMENTAL ANALYSIS
CONTENTS
MOTIVATION AND OBJECTIVES
EXPERIMENTAL ANALYSIS
EXPERIMENTS VS CALCULATION
Inert flow
Reacting flow – Nominal spray A
Reacting flow – Low reactivity cases
CONCLUSIONS

24. REACTING SPRAY A Sr/Si as derived from CFD
General features of the reacting flow are reproduced
CMT/Polimi/TUE combustion models produce essentially a quantitatively similar effect on the inert flow
UNSW, ETH slower reaction to onse of heat release
CMT Polimi
TUE
UNSW
ETH

25. The inert field does have an influence on reacting flow evolution!
REACTING SPRAY A
Rective penetration
Results are stratified according to inert spray penetration
TUe above Polimi-CMT
ETH comes together to UNSW
The inert field does have an influence on reacting flow evolution!
For all cases, si = experiment
TUE
Polimi
CMT
ETH-UNSW

26. REACTING SPRAY A TUE Overpredicted velocity on axis and integral
time = 1500 us
EXP-REACT
EXP-INERT

27. REACTING SPRAY A CMT/Polimi Good overall agreement time = 1500 us
EXP-REACT
EXP-INERT

28. REACTING SPRAY A
ETH
Good description of the flow under reacting conditions
Very high expansion in the Inert-to-reacting transition
time = 1500 us
EXP-REACT
EXP-INERT

29. REACTING SPRAY A UNSW ETH/UNSW very different flow, same penetration
High radial expansion
time = 1500 us
ETH/UNSW very different flow, same penetration
EXP-REACT
EXP-INERT

30. MOTIVATION AND OBJECTIVES EXPERIMENTAL ANALYSIS
CONTENTS
MOTIVATION AND OBJECTIVES
EXPERIMENTAL ANALYSIS
EXPERIMENTS VS CALCULATION
Inert flow
Reacting flow – Nominal spray A
Reacting flow – Low reactivity cases
CONCLUSIONS

31. LOW REACTIVITY CASES Consistency of experimental data T2 ID(ms)
LOL(mm) T2
IFPEN (schlieren)
ECN3 (chemilum)
0.77 ± 0.06
0.95
24.6
27.5 EX
1.19 ± 0.18
39.5

32. LOW REACTIVITY CASES Reacting penetration
Reasonable agreement of modelling results T2 EX

33. LOW REACTIVITY CASES T2 case Overall good agreement
Polimi lower reactivity, resulting in longer LOL/ID
TUE: At a late time instant, no indication of radial dilation
time = 4500 us
EXP-REACT
EXP-INERT

34. MOTIVATION AND OBJECTIVES EXPERIMENTAL ANALYSIS
CONTENTS
MOTIVATION AND OBJECTIVES
EXPERIMENTAL ANALYSIS
EXPERIMENTS VS CALCULATION
Inert flow
Reacting flow – Nominal spray A
Reacting flow – Low reactivity cases
CONCLUSIONS

35. EXPERIMENTS CFD CONCLUSIONS
Quantification of combustion-induced flow changes
Low reactivity cases:
Inert to reacting transition in velocity can be quantified
Extended flame
CFD
Accuracy highly improved compared to ECN3
Inert spray evolution has an influence on the reacting case
Not a single model predicts penetration/flow characteristics under both inert and reacting conditions

36. Backup slides

37. DESCRIPTION OF REACTING FLOWEX
Longest ID/LOL
Flow before/after LOL
Radial expansion ~T2

38. Dirty laundry…. Data consistency

39. INERT CASES Spray tip penetration Overall good agreement
UNSW deviations in the early penetration zone
ETH deviations during the whole time period
Momentum flux plot to clarify penetration trend TUE-UNSW > CMT-Polimi
time = 1500 us

40. REACTING SPRAY A
ETH
Good description of the flow under reacting conditions
Inert-to-reacting transition?
Much higher radial expansion compared to experiments/other CFD
time = 1500 us
Polimi

41. Sensitivity to TCI / Chemistry
REACTING SPRAY A
Sensitivity to TCI / Chemistry
TCI affects ignition timing/LOL, but not subsequent reacting flow dynamics
For WM models, Chemistry does have an effect on flow evolution

42. Sensitivity to TCI / Chemistry
REACTING SPRAY A
Sensitivity to TCI / Chemistry
TCI affects ignition timing/LOL, but not subsequent reacting flow dynamics
Chemical mechanism for WM models, strong effect on flow evolution

43. REACTING SPRAY A Sensitivity to TCI / Chemistry
Later ID/LOL higher flow expansion due to autoignition progressing different spatially
Later autoignition  higher radial dilation, which is observed experimentally!!

44. LOW REACTIVITY CASES T2 case Similar overall flow
Lower reactivity of Polimi vs TUE  longer LOL/ID
EXP-REACT
EXP-INERT

45. LOW REACTIVITY CASES EX case Overall good agreement
Polimi too long transition to reacting conditions
time = 4300 us
EXP-REACT
EXP-INERT

46. Correlation between Mdot stablishment and end of acceleration??

49. SPRAY A ANALYSIS LOW REACTIVITY CASES CONCLUSIONS CMT/Polimi/TUE
Good flow description of flow
Reacting tip overprediction
TUE overprediction was already happening for the inert case
ETH/UNSW good sr prediction
Very high radial dilation compared to the inert case
LOW REACTIVITY CASES
Polimi calculations predict a too long LOL/transition to reaction
TUE does better, but radial dilation is not adequately quantified