Embed Size (px)
Citation preview
Topic 3 – Vaporizing Diesel Spray Spray A and Spray B
Presenter: Caroline Genzale, Georgia Tech September 5th, 2015
2
ECN1: Detection of liquid boundary near the LL is sensitive to optical method and setup
Extinction offers referenced intensity measurement.
Some history: We’ve made several attempts to standardize liquid length measurements within the ECN.
Side-illumination, high-res., f/4
Side-illumination, fast, f/4
Side-illumination, saturated, f/1.2
Sheet-illumination, saturated, f/1.2
Head-illumination, f/2
Diffuser back-illumination
I/I0
532 nm CW
532 nm CW
532 nm CW
532 nm CW
Halogen lamp
Halogen lamp
0 2 4 6 8 10 12 14 16 18 200
0.20.40.60.8
11.21.41.61.8
2
Axial distance [mm]
Cen
terli
ne o
ptic
al th
ickn
ess
τ [-]
Beam steering
3
• “Extinction profiles measured on the centerline are somewhat different between institutions.”
• “The variations observed are believed to come from differences in optical arrangement (mainly illumination and collection angles).”
• “A methodology must be setup to consistently measure liquid length from extinction without the influence of beam steering.”
Some history: We’ve made several attempts to standardize liquid length measurements within the ECN.
ECN2: Adoption of DBI imaging to measure I/Io
Beam steering
ECN4: Fredrick Westlye will provide us an update on this topic next!
4
More history: We’ve seen consistent predictions of liquid/vapor penetration between CFD codes, but not in the details of the spray.
ECN2: Significant differences between CFD codes in the predicted spray structure.
5
More history: We’ve seen consistent predictions of liquid/vapor penetration between CFD codes, but not in the details of the spray.
ECN3: Even when downstream profiles are consistent, upstream spray structure can vary greatly. These regions can be coincident with LOL region.
6
Following up on prior themes: •Further assess institutional variations in DBI measurements and develop a standardized (and consistent) measurement approach for liquid penetration. •Pursue further insight into phase transition behavior at Spray A conditions. Vaporization data at new target conditions: •Spray A multiple injection conditions •Spray B •Spray C and D injectors
Experimental Objectives for Vaporizing Diesel Sprays
7
Following up on prior themes: • Deep dive into near-nozzle spray structure variations between
codes and sensitivities to ambient conditions. • Understand impact of these differences on combustion
predictions. Simulations at new target conditions: • Spray A multiple injection conditions.
Modeling Objectives for Vaporizing Diesel Sprays
Spray A Parametric Modeling Study
CMT Adrian Blanco, Raul Payri
PoliMi Tommaso Lucchini, Gianluca D’Errico
GA Tech Gina Magnotti, Caroline Genzale
Bosch Edward Knudsen
Sandia Guilhem Lacaze, Joe Oefelein
9
• Injector 675
• 89.4 µm, Ca = 0.98, Aeff = 6.15x10-
9 m2, ρliq = 713 kg/m3
• Standard injection rate profile provided to all institutions
Target conditions for Spray A parametric modeling study
Parametric Variable O2 [%] Tamb [K] ρamb [kg/m3] Pinj [bar]
Standard Spray A 0 900 22.8 1500
Temperature 0 700 22.8 1500
0 1200 22.8 1500
Density 0 900 7.6 1500
0 900 15.2 1500
Injection Pressure 0 900 22.8 500
0 900 22.8 1000
00.5
11.5
22.5
3
-0.5 1.5 3.5 5.5
Mas
s Flo
w R
ate [
g/s]
Time [ms]
10
Comparison of spray and CFD models
PoliMi GA-Tech CMT Bosch Sandia
Code OpenFOAM CONVERGE OpenFOAM Cascade Technologies Raptor
Spray model Blob, KH-RT Blob, KH-RT Eulerian Σ-Y model Eulerian dense fluid
Eulerian dense fluid
Evaporation model
Frossling drop evaporation
Frossling drop evaporation
State relationships from locally
homogeneous flow assumption
Peng-Robinson EOS
Real fluid model
Turbulence model RANS, k-ε RANS, k-ε RANS, k-ε LES LES Dynamic
Model Mesh
Dimensionality Grid size
2D 0.5 mm
3D 0.0625 –
2 mm
2D 0.009 mm –
0.7 mm
3D w/ internal nozzle
3.2M cells
3D 0.002 mm –
0.1 mm
Standard Spray A Condition
12
Time ASI [ms]0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
LP [m
m]
0
5
10
15
BoschPolimiCMTGA TechSandiaSandia - 675
Simulations of quasi-steady liquid penetration cluster around the experimental data, similar to previous workshops.
Liquid boundary definition: Axial position where path-averaged liquid volume fraction is 0.15%
0.5 mm x 0.5 mm LVF computation grid
Spray A
13
Differences in fuel vapor fraction upstream arise from combined effects of spray model and momentum coupling.
Spray A
Axial Distance [mm]0 2 4 6 8 10 12
Ua
vg
[m/s
]
0
100
200
300
400
500
600CMTPolimiSandia
Axial Distance [mm]0 5 10 15 20 25
Z [-]
0
0.1
0.2
0.3
0.4
0.5
0.6
CMTPolimiSandia
Axial Distance [mm]0 2 4 6 8 10 12
LVF
[-]
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
CMTPOLIMIGATechSandia
14 T Di t [ ]-2 -1 0 1 2
LVF
[-]
0
0.005
0.01
0.015
0.02
CMTPOLIMIGA Tech
Jet dispersion well matched, but predicted spray structures vary significantly between institutions as we approach the LL.
Transverse Distance [mm]-2 -1 0 1 2
Z [-]
0
0.1
0.2
0.3
0.4
CMTPOLIMI
Spray A time averaged spray structure
1 - 3 ms
7.5 mm
15
Predicted spray structures vary significantly between institutions as we approach the LL.
Spray A time averaged spray structure
1 - 3 ms
9.0 mm
16
SMD profiles can take on different distribution behaviors as breakup and evaporation proceed.
Spray A time averaged spray structure
1 - 3 ms
Which solution is the “accurate” one?
3 mm
7.5 mm
9 mm
Spray A Parametric Study: Exploration on what information we’d really like to have
to validate evaporating sprays
18
4
6
8
10
12
14
16
18
20
650 850 1050 1250
LL [m
m]
Temperature [K]
CMTPolimi - MassGA TechSandia Data
4
6
8
10
12
14
16
18
20
650 850 1050 1250
LL [m
m]
Temperature [K]
CMTPolimi - MassPolimi - LVFGa Tech - MassGA Tech - LVFSandia Data
Revisit of global model response to parametric variations in ambient and operating conditions.
19
0
5
10
15
20
25
5 10 15 20 25
LL [m
m]
Density [kg/m3]
CMTGA Tech - MassGA Tech - LVFPolimi - MassPolimi - LVFSandia Data - 675
Revisit of global model response to parametric variations in ambient and operating conditions.
20
4
6
8
10
12
14
16
18
20
40 80 120 160
LL [m
m]
Injection Pressure [MPa]
CMTPolimi - MassPolimi - LVFSandia Data - 677Sandia Data - 675
Revisit of global model response to parametric variations in ambient and operating conditions.
21
-1 -0.5 0 0.5 1
1 0 5 0 0 5 1
How do the spray and evaporation models respond to changes in ambient and operating conditions?
CMT
PoliMi
22
-1 -0.5 0 0.5 1
Comparison of LVF behavior and accumulated mass… they both show *some* sensitivity to ambient conditions.
PoliMi
-1 -0.5 0 0.5 1
Acc
umul
ated
Mas
s [%
]
96.5
97
97.5
98
98.5
99
99.5
100
= 22.8 kg/m 3
= 15.2 kg/m 3
= 7.6 kg/m 3
23
Axial Distance from LPL [mm]-1 -0.5 0 0.5 1
Ta m b
= 1200 K
Ta m b
= 900 K
Ta m b
= 700 K
1 0 5 0 0 5 1
LVF
av
g
-0.03
-0.02
-0.01
0
0.01
0.02
1
Axial Distance from LPL [mm]-1 -0.5 0 0.5 1
Pin j
= 50 MPa
Pin j
= 100 MPa
Pin j
= 150 MPa
Gradients near LL show that LVF shows more sensitivity to correctly capture the liquid boundary
PoliMi
24
Is there a potential pathway forward to validated modeling of the local liquid structure and evaporation process?
Magnotti & Genzale, Atomization and Sprays, 2015.
10-2 10-1 100 101 10210-25
10-20
10-15
10-10
10-5
Droplet Size [µm]
Cex
t [m2 ]
Cext data
Rayleighα d 6
Mie α d 2
Theoretical dependence of light-scattering cross-section on drop size
Modeling of light extinction measurements could provide some validation at vaporizing conditions.
Quality extinction measurements needed (next talk).
Validation of near-spherical droplets in these regions (previous talk).
25
• Old news revisited: Validation of evaporating sprays solely against liquid and vapor penetration measurements does not adequately validate the evaporation process.
• Somewhat old news revisited: It is difficult to isolate effects of spray model choices on near-nozzle mixture discrepancies. – Lagrangian models lead to errors in gas momentum exchange
• Error cascades to affect breakup and vaporization models – Eulerian models don’t necessarily predict evaporation well either
• To accurately validate evaporating sprays, we need a validation methodology that responds to changes in the mixing and evaporation rate appropriately.
• Liquid volume fraction is physically consistent with extinction and scattering measurements AND drops rapidly near the liquid length.
• High quality extinction measurements may help.
Conclusions from Spray A parametric modeling study:
Spray A Multiple Injections
Modeling CMT: Adrian Blanco, Raul Payri
Polimi: Tommaso Lucchini, Gianluca D’Errico Experiments
Sandia: Julien Manin, Scott Skeen, Lyle Pickett
27
Spray A multiple injection target conditions
O2 [%] Tamb [K] ρamb [kg/m3] Pinj [bar]
Standard Spray A 0 900 22.8 1500
Split Injection 0.5 ms – 0.5 ms dwell – 0.5 ms Pilot-Main 0.3 ms – 0.5 ms dwell – 1.2 ms
• Same model setup as parametric study. Nozzle 675.
28
Multiple injection simulations compare well between institutions and with Sandia vapor penetration data.
CMT PoliMi
Split Injection 0.5 ms – 0.5 ms dwell – 0.5 ms Pilot-Main 0.3 ms – 0.5 ms dwell – 1.2 ms
Sandia vapor data
29
Transient mixing process is well matched between CMT and PoliMi downstream of liquid regions.
Spray B
Experiments CMT: Alberto Viera, Raul Payri
Nozzle 211200 Sandia: Julien Manin, Scott Skeen, Lyle Pickett
KAIST: Yongjin Jung Nozzle 211201
31
Simultaneous Mie scatter and DBI at CMT
Continuous light source
Blue LED
HS Camera Mie-scatter 32 kfps HS Camera with long-range
microscopic lens DBI 82.1 kfps
32
Two different liquid penetration measurement techniques and two spray orientations: •Horizontal Configuration
– Mie scattering: Phantom v710, 105 mm, f/2.8, 80 kfps
•Vertical Configuration – Diffused back illumination (DBI),
Photron, 100 kfps
Mie scatter and DBI imaging at Sandia
33
Sandia has shown that liquid penetration profile is correlated with transient spreading angle behavior.
Jung et al. SAE 2015-01-0946
34
CMT also measures a transient liquid penetration rate, but with a shorter penetration length.
35
• First look at modeling multiple injections: predictions of mixture profiles and transient behavior is not any worse that steady-state Spray A. – Near-nozzle mixture discrepancies, but these are rapidly diffused by
vaporization and transient mixing process. Spray B measurements: • Transients in measured liquid penetration is well correlated with
transient spreading angle behavior. • Measurements at CMT display similar transient behavior, but injection
conditions are different so it’s hard to make further conclusions. • Seems to be a consistent trend that CMT measures shorter LL’s than
Sandia. • Modeling of the transient behavior will require more than the usual
Lagrangian spray modeling approach.
Moving forward with Spray A multiple injections and Spray B:
36
• Old news revisited: Validation of evaporating sprays solely against liquid and vapor penetration measurements does not adequately validate the evaporation process.
• Somewhat old news revisited: It is difficult to isolate effects of spray model choices on near-nozzle mixture discrepancies. – Lagrangian models lead to errors in gas momentum exchange
• Error cascades to affect breakup and vaporization models – Eulerian models don’t necessarily predict evaporation well either
• To accurately validate evaporating sprays, we need a validation methodology that responds to changes in the mixing and evaporation rate appropriately.
• Liquid volume fraction is physically consistent with extinction and scattering measurements AND drops rapidly near the liquid length.
• High quality extinction measurements may help.
Conclusions from Spray A parametric modeling study:
37
EXTRA SLIDES
38
ECN4: Fredrick Westlye will provide us an update on this topic next!
Some history: We’ve made several attempts to standardize liquid length measurements within the ECN.
ECN3: Still differences in measured LL between institutions. No significant advancement of “standardized”
LL measurement method. Spray A Measured Liquid Length
Sandia - 675 CMT - 675 IFPen - 678 TU/e - 679
11.7 9.7 10.5 10.3
Parametric Spray A Conditions
Sandia - 675 CMT - 675 ∆
900 K, 7.6 kg/m3 16.0 22.8 43%
700 K, 22.8 kg/m3 14.5 17.6 21%
