Why LES for Atomization

3 downloads 0 Views 6MB Size Report
internal flow of automotive injectors: URANS & LES improving the understanding of the atomization. Dr. J. Hélie (Continental Automotive) invited presentation of ...
The role of unsteadiness and coherent structures in the internal flow of automotive injectors: URANS & LES improving the understanding of the atomization Dr. J. Hélie (Continental Automotive) A

invited presentation of LES4ICE conference with contributions of: J. Cousin, A. Berlemont, T. Menard, C. Dumouchel, CORIA, G. Wigley LBoro Univ, A. Heather, OpenCFD, M Gavaises, CUL Univ. O. Soriano, R. Rotondi, J. Shi, H. Nuglisch, B. Imoehl, Continental Automotive

Introduction: Objectives Overview of the atomization process Efficiency Losses

Available energy (∆Pi)

Flowing energy

Cavitation

CASE

Eatom

2 ⋅ σ ⋅ mL  3 2   = ⋅  − ρL  D32 d nozzle 

© Continental AG

Theoritical Efficiency (%)

MPI 5bar, SMD=50um

0.5%

HPDI 200bar, SMD=11um

0.0657%

DIESEL 2000bar versus 30bar, SMD= 5um

0.0015 %

Still a lot to do ! JH / Oct. 2010

Secondary flow

Turbulence

Introduction: Objectives Injection Trends for 2015-2020 (for Diesel & Gasoline)

Increase the Pressure

Increase the Temperature

Increase the fuel diversity

+ tendency of Diesel & last year

+ overheated conditions

+ less dependency/ petroleum

- request more and more pressure

- difficulty of heated tip

- system quality

- energy spent Increase the Nozzle complexity

Increase the Holes number

Increase the needle dynamic

+ better atomisation

(Decrease the hole diameter)

+ multi injection

- tolerances

better(Diesel) or worst (Gasoline) atomisation

- actuator (Piezo faster)

- coking - jet-to-jet interaction (Group-hole)

JH / Oct. 2010

© Continental AG

Great challenges ! R. Rotonti, O. Soriano, J. Hélie, SIA 2010

- constraint on cost

Introduction: Objectives Overview of the Design Process for industrial application performances

Design variation CFD investigate (RANS)

Design rules Turbulence intensity Cavitation intensity

Prototypes

rules

Referent Design CFD investigate (detailed URANS, LES)

Disavantage: long CPU time JH / Oct. 2010

© Continental AG

Atomization factors

methodology

Introduction: Context Overview of current injectors designs in the litterature Valve Covered ( Diesel): lift (levée)

needle (aiguille)

seat (siege, fermeture)

sac volume Micro Sac, (Diesel & Gasoline type) : Simplified : academic world (Hiroyasu et al. ..)

divergent

Flat Sac, (Gasoline type) :

step

JH / Oct. 2010

© Continental AG

Introduction: Why LES for Atomization ? How it proceeds in automotive atomizer ? Let's consider a liquid sheet that is expanding (through tangential velocity, swirl, turbulence...)

Regime of inertial Force

mass balance :

r ∫V S1

r2 kinetic energy balance ( V ∫

) – Surface tension force

S1

cannot be conserved both without dissipation !

JH / Oct. 2010

© Continental AG

Introduction : Why LES for Atomization ? 2D numerical Experiment 1/3 Geometrical parameters

MODEL VORTEX X

l d

Y

GAS

LIQUID

l d  L  X = 60l  Y = 30l

Physical parameters

ρ G , µG  ρ L , µ L  σ u θ max

Temporal evolution of the Interface and Vorticity field

JH / Oct. 2010

© Continental AG

J. Cousin, T. Menard, A. Berlemont, J. Hélie, ILASS 2010

Non Dimensional parameters (PI theorem)

d /l

ρ G/ ρ L µ G/ µL Re = ρ L uθ max l / µ L We = ρ L u 2θ max l / σ

Introduction : Why LES for Atomization ? 2D numerical Experiment 2/3 Re Re (-) (-) Re (-) (-)

10000 10000 10000 10000

Re varies We varies R = 1.126 10-3 M = 1.566 10-2 d = 2l L=15l l=0.001 m

Detachment Detachment Detachment Detachment

1000 1000 1000 1000

No detachment No detachment No detachment 100 100 100 11 100 1 1

10 10 10 10 (-) We

A minimum energy is required to eject a drop Simple model from Energy & Mass conservation + Rayleigh Breakup

JH / Oct. 2010

© Continental AG

ρ LU 2l We = = 7.56 σ

We (-) (-) We (-)

100 100 100 100

Existence of a single curve that separates the two behaviors

J. Cousin, T. Menard, A. Berlemont, J. Hélie, ILASS 2010

Introduction : Why LES for Atomization ? 2D numerical Experiment 2/3 Re (-) 10000

Re varies We varies R = 1.126 10-3 M = 1.566 10-2 d = 2l L=15l l=0.001 m

1000

100 1.E-04

1.E-03

1.E-02

1.E-01

Oh (-)

Ohnesorge-Reynolds diagram

JH / Oct. 2010

© Continental AG

We Oh = = Re

J. Cousin, T. Menard, A. Berlemont, J. Hélie, ILASS 2010

µL ρ Lσl

Low Pressure Atomizer (MPI, SCR, DDS) Low Weber & Reynolds Number •Low Pressure Atomizer •Long Opening time

MPI (Deka VII)

•Holes direction for MPI targetting •Dosing & atomization for DDS & SCR

DDS injector

•Various Fluids used •Cost limited

SCR injector

JH / Oct. 2010

© Continental AG

Low Pressure Atomizer (MPI, SCR, DDS) Triple Disk Experiment The atomization energy balance was established Swirl velocity at the nozzle exit Turbulence intensity at the nozzle exit

x z

vs Surface creation σ/D32 (kg/ms ) 2

700

1000

σ/D32 (kg/ms2) Water Linear regression CSL2 (model k-ε) CSL2 (model RSM)

600 800

500 400

600

300 400

200 200

100 0

σ = 9.63 x10 − 3 (Ek + ρL Tke ) + 170 D32

0

0

10000

20000

30000

Ek + ρLTke (kg/ms ) 2

40000

50000

0

10000

20000

30000

40000

50000

60000

70000

Ek + ρLTke (kg/ms ) 2

Formalism of the maximum entropy principle and the interface density balance were used to establish a theoretical modeling of the break up process JH / Oct. 2010

C. Dumouchel & J. Cousin CORIA group published in Experiments in Fluids 2005 Int. J. of Multiphase Flow 2007, J. of flow visualization & image processing 2008...

© Continental AG

y

Low Pressure Atomizer (MPI, SCR) Triple Disk Experiment : RANS Approach Variant090

Variant000

m²/s²

Variant200

Variant500

Variant700

Variant900

Ek

0.006

0.006

0.006

0.006

0.004

0.004 0.002

0.000 0

100

200 300 D (µm)

400

0.000 0 500

0.004 0.002

100

200 300 D (µm)

400

0.000 500 0

0.008 075076700

0.004 0.002

100

200 300 D (µm)

400

0.000 500 0

0.004 0.002

100

Volume based drop size distribution

200 300 D (µm)

C. Dumouchel & J. Cousin CORIA group published in Experiments in Fluids 2005 Int. J. of Multiphase Flow 2007, J. of flow visualization & image processing 2008...

© Continental AG

075076900

0.006 fv (µm-1)

075076500 0.008

fv (µm-1)

075076200 0.008

fv (µm-1)

0750760900.008

0.002

S1I1 WATER ∆Pi = 3 bar

JH / Oct. 2010

0.008

fv (µm-1)

fv (µm-1)

Tke

400

0.000 5000

100

200 300 D (µm)

400

500

Low Pressure Atomizer (MPI, SCR) Triple Disk Experiment : RANS Approach coupled with Interface Solver: ARCHER (CORIA) Level Set+VOF+Ghost Fluid + Input statistical at intlet

Solver: Fluent6.2 RNG k-eps simulation

removed turbulence (a)

non axial velocity

rebuilt turbulence

(b)

turbulent kinetic energy

Acceptable morphology of the liquid system was found BUT Overestimated jet angle at the vicinity of the nozzle exit ⇒more accurate internal flow simulation is necessary to provide better quantitative results. JH / Oct. 2010

© Continental AG

A. Berlemont, J. Cousin, S. Grout, T. Menard, ASME 2010

Low Pressure Atomizer (MPI, SCR) Triple Disk Experiment : LES Approach of Internal Flow

Solver: Fluent6.2 LES, cst smago

Computational mesh

Fluid trajectories

Experimental validation based on hot wire measurements applied on a large gas scale model 2 -2

Tke(m s ) 16

Turbulent kinetic energy profile along the x and Tke(mys )directions 2 -2

25

(b)

(a)

14

20

12 10 8

15

LES EXP

LES

10

6

EXP

4 5

2 0 0.E+00

2.E-05

4.E-05

x (m)

6.E-05

8.E-05

0 0.0E+00

4.0E-05

8.0E-05

y (m)

A. Berlemont, J. Cousin, S. Grout, T. Menard, ASME 2010 JH / Oct. 2010

© Continental AG

1.2E-04

1.6E-04

Low Pressure Atomizer (MPI, SCR) Triple Disk Experiment : LES Approach Coupled to Interface

Solver: ARCHER (CORIA) Level Set+VOF+Ghost Fluid + LES results at intlet

The calculated jet angle (26°) is fairly close to the measured one (24°). LES simulations provide much better initial conditions for the jet simulation compared to what was previously with RANS

Jet angle

JH / Oct. 2010

© Continental AG

One of the main visible difference on these two images is the number of droplets in the visualized domain. When the simulation starts, the velocity of the discharged jet is high and a large number of droplets are created in the jet leading edge. This is not the case for the real application; when the injection starts, the jet velocity continuously increases and the production of droplets is limited ⇒ Spurious difference which can be easily corrected

A. Berlemont, J. Cousin, S. Grout, T. Menard, ASME 2010

Low Pressure Atomizer (MPI, SCR) Triple Disk Experiment : LES Approach coupled to Interface

Solver: ARCHER (CORIA) Level Set+VOF+Ghost Fluid + LES results at intlet

During steady state operation, the simulation reports well the particular breakup mechanism. 1. The liquid jet flattening in the vicinity of the discharge orifice is well reproduced. Rims formation (left: visualization, right: simulation)

2. The reorganization into a ligament network is also observed. 3. This leads to the formation of two external rims. Perforations occur in the jet center.

Perforation process left: visualization, right: simulation

The numerical simulation predicts perforations closer to the orifice It is due to the jet flattening, the liquid sheet becomes too thin to be predicted with a sufficient accuracy with respect to the computational cell size set to 5 µm. ⇒ AMR requested for fully predictive VOF

JH / Oct. 2010

© Continental AG

A. Berlemont, J. Cousin, S. Grout, T. Menard, ASME 2010

High Weber & Reynolds Number – example of Diesel Injectors

switch leakage reduction 2000 bar capability Smart Electronic control algorithms ... Nozzle improved

PCR Servo Valve Euro 6

(…)

Development targets: -

Injector Closed Loop Control

-

Hydraulic-free: Full rate shaping possible

(...) PCR NG – direct drive

JH / Oct. 2010

© Continental AG

L. Passilly, F. Atzler, K. Wenzlawski, SIA 2010

High Weber & Reynolds Number – Diesel Injectors Cavitation for industrial application: Shear cavitation, RANS & LES Academic investigation, including both real-size and largescale, sac-type and VCO-type Diesel nozzles, with either cylindrical or tapered holes, inlet sharp or rounded, as here from City University London

Solver: Homecode City University london Model: URANS + lagrangian cavitation model*

Mean LES

RANS Instantaneous LES Similar result for the structure of RANS with the LES mean pressure field. However the instantaneous solution provides a more detailed representation of the turbulent structures © Continental AG A Theodorakakos, E Giannadakis, D. Papoulias, M Gavaises, SAE 2009-01-0833 A Papoutsakis,

JH / Oct. 2010

High Weber & Reynolds Number – Diesel Injectors Cavitation for industrial application: cavitation vortex, also named cavitation string Cavitation strings are forming at areas where large-scale vortical structures develop Such large-scale vortical structures although they are evident in CFD simulations, are rather transient and strongly affected by the nozzle geometry, operating conditions, needle motion, etc.

0

Vortex core between adjacent holes

JH / Oct. 2010

Swirl (-)

0.5

Sharp flow turn inside the sac volume & separarions points

Vortex core inside the injection hole

© Continental AG

A Papoutsakis, A Theodorakakos, E Giannadakis, D. Papoulias, M Gavaises, SAE 2009-01-0833

Solver: Homecode City University london Model: URANS + lagrangian cavitation model*

High Weber & Reynolds Number – Diesel Injectors Cavitation for industrial application: modeling the effect on atomization Cavitation affects the atomization, flow direction & the flow profile Two Zones model.

Solver: CFX R-P cavitation model CONTI Improved Baumgarten Model*

(Baumgarten,C. – U. Hannover 2003)

Vapour Zone (Zone 2)

Liquid Zone (Zone 1)

New droplets 2 Phase-cavitating turbulent nozzle flow

Injected ligament

Collapsing ligament

Φ 2/2 Φ1

Rayleigh´s tcoll

Acting break-up mechanisms: Cavitation Turbulence Nozzle flow

Primary break-up Diesel – Euler approach Diesel Vapour – Euler approach

JH / Oct. 2010

© Continental AG

O. Soriano et al., ILASS Europe 2008

Acting break-up mechanisms: Fuel - ambient gas interaction Secondary break-up

Diesel – Lagrange approach Gas – Euler approach

Medium Weber & Reynolds Number – Pintle Gasoline Injectors Technical Data Working Flow Range: Static flow:

Thermal Compensator

> 50

and inlet tubes

> 35 g/s

Minimum dynamic flow. < 2 mg/str Spray angle:

60 deg to 120°

SMD size:

~ 10 - 15 µm

Opening/Closing time:

≥ 150 µs

Temperature range:

-30...+140°C

System Pressure:

5...20 MPa

Piezo Electrical Activator

-Outward opening -fine liquid corrona cartridge -geometrical angle

seat needle

JH / Oct. 2010

© Continental AG

Medium Weber & Reynolds Number – Pintle Gasoline Injectors Specific Atomization Process

Piezo produces a new (non linear) mode of atomization: Hollow cone with steaks (micro-jets)

Delta P = 5 bar

JH / Oct. 2010

Delta P = 200 bar

G Delay, B. Prosperi, R. Bazile, H Nuglisch, J. Hélie , Experiments in Fluids, 2007 & Work of J. Cousin, CORIA & G. Wigley, Louborough Univ

© Continental AG

Medium Weber & Reynolds Number – Pintle Gasoline Injectors Specific Atomization Process

Recirculations in the needle sac

Solver: Homecode City University london Model: URANS + lagrangian cavitation model*

Injector

Main flow jet Counter-rotating recirculation zones formed inside the dead volume below the lower needle guide

E Giannadakis, D Papoulias, A Theodorakakos, M Gavaises, Proc. IMechE Vol. 222 Part D: J. Automobile Engineering © Continental AG Gavaises, M., Tonini, S., Marchi, A., Theodorakakos,A., Bouris, D., and Matteucci, L., Int. J.Engine Res., 2006, 7, 381–397.

JH / Oct. 2010

Medium Weber & Reynolds Number – Pintle Gasoline Specific Flow - Atomization Process : results of large scale nozzle Injector

Solver: Homecode City University london Injectors Model: URANS + lagrangian cavitation model*

High Speed Imaging: Flow images reveal cavitation formation in

a rather unstable pattern consisting of vapour ‘pockets’

Cavitation

Simulation: Sample vapour bubbles and vapour volume fraction: A similar pattern is predicted by the simulation model

0% vapour 40% Visualisation . window E Giannadakis, D Papoulias, A Theodorakakos, M Gavaises, Proc. IMechE Vol. 222 Part D: J. Automobile Engineering © Continental AG JH / Oct. 2010

Gavaises, M., Tonini, S., Marchi, A., Theodorakakos,A., Bouris, D., and Matteucci, L., Int. J.Engine Res., 2006, 7, 381–397.

Medium Weber & Reynolds Number – Pintle Gasoline Injectors LES + VOF: Internal Flow Results: real injector geometry Solver: OpenFOAM Model: LES (smago) + VOF*

injection of turbulence at intlet

10 angular degrees domain

JH / Oct. 2010

© Continental AG

Results Continental Automotive/OpenCFD, PREDIT IDE2 PROJECT

Medium Weber & Reynolds Number – Pintle Gasoline Injectors LES + VOF: External Flow Result : real injector geometry nozzle exit

Liquid/Gas interface colored by the air velocity

Streaky structures

JH / Oct. 2010

Results Continental Automotive/OpenCFD, PREDIT IDE2 PROJECT Picture CORIA, PREDIT IDE2 PROJECT

© Continental AG

Solver: OpenFOAM Model: LES (smago) + VOF*

Medium Weber & Reynolds Number – Low L/D MultiHole Gasoline Injectors Present generation of Multihole gasoline injectors DI gasoline homogeneous combustion Oriented downsized engine 200bar sample operational fuel pressure fast magnetic actuation Studied injectors, special prototype: 3 holes 90deg wide spray angle XL2 5.5g/s N-Heptane @ 100bar No Step-Hole

seat sac

6 holes = L/D = 1 to 1.2 hole

JH / Oct. 2010

© Continental AG

Solver: CFX Model:U-RANS SST LES SAS

Medium Weber & Reynolds Number – Low L/D Numerical Study, validation at low pressure without cavitation Mesh:

SST_basic

SAS_fine

Q8.1E+11

1.66 M Cells up to 3.2M

Q2.5E+12

"URANS" solution

0.01 0.009 0.008 0.007 0.006 0.005 0.004

Mass flow at outlet

0.003 0.002 0.001 0 0

Power Spectrum

Experimental evidence: -non unique frequency, -one important frequency identified around 25kHz

0.5

x

1

1.5

2

2.5

3

3.5

4

4.5

5 4

x

10

x

10

x

10

x

10

x

10

x

10

7

10

16

14

12

10

8

6

4

Pressure

2

0 0

0.5

x

1

1.5

2

2.5

3

3.5

4

4.5

5 4

11

10

7

6

5

4

3

Total pressure

2

1

0 0

0.5

x

1

1.5

2

2.5

3

3.5

4

4.5

5 4

15

10

2.5

2

1.5

1

0.5

0 0

12

0.5

x

1

1.5

2

2.5

3

3.5

4

4.5

Velocity U

5 4

14

10

10

8

6

4

2

0 0

2.2

0.5

x

1

1.5

2

2.5

3

3.5

4

4.5

Velocity V

5 4

14

10

2 1.8 1.6 1.4 1.2 1 0.8 0.6

Velocity W

0.4 0.2 0 0

0.5

1

x

1.5

2

2.5

20

10

0

3

3.5

4

4.5

5 4

40

30

50

"LES" solution

−3

10

4

3.5

3

2.5

2

1.5

1

0.5

0 0

x

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5 4

x

10

x

10

x

10

x

10

9

10

6

Mass flow at outlet

Power Spectrum

5

4

3

2

1

0 0

6

x

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5 4

9

10

5

Pressure

4

3

2

1

0 0

x

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5 4

13

10

3.5

3

Total pressure

2.5

2

1.5

1

0.5

0 0

x

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5 4

13

10

Velocity U

10

9

8

7

6

5

4

3

2

1

0 0

x

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5 4

x

10

x

10

13

10

2.5

Velocity V

2

1.5

1

0.5

0 0

JH / Oct. 2010

© Continental AG

0

0.5

1

1.5

2

2.5

3

3.5

Frequency [k Hz]

J Shi, K. Wenzlawski, J. Helie, H. Nuglisch, J. Cousin, ILASS 2010

4

4.5

5 4

Velocity W

50

Medium Weber & Reynolds Number – Low L/D MultiHole Gasoline Injectors Internal Flow: Fluid Properties, cavitation Simplified 2D Nozzle flow AVL testcase exhibits the main features.

Solver: CFX Model:RANS + R-P cavitation

"shear cavitation"

Cavitation dominated by the shear :

Experimental data available for diesel

low effect of sat. vapour pressure

100bar versus 33 bar, Geometry J

large effect of viscosity

Diesel, 20 °C

JH / Oct. 2010

© Continental AG

n-Heptane, 20 °C

J.Shi & M. Arafin, ILASS 2010

n-Heptane, 70 °C

Medium Weber & Reynolds Number – Low L/D Unsteadiness of the flow Coherent structures at various pressure, developing from the nozzle :

5 bar

10 bar

50 bar

Unsteadiness during injection (various times shown, @50bar):

JH / Oct. 2010

© Continental AG

Results CORIA, FUI "MAGIE" Project

120 bar

CONCLUSIONS Great Challenges in CFD to support design of next generation of automotive injectors Most of the time, URANS returns correct trends for a fast answer LES is requested as input of external VOF Cavitation model is a limiting factor for accuracy–no consensus in the community.

Cases of Low Pressure, High Pressure Diesel, Medium Pressure nozzle with Pintle needle and Short L/D have been presented Internal Flow & Atomization are highly coupled Coherence of the internal flow has to be captured for the simulation of the breakup Validation: Qualitative results satisfying, quantitative results request continuous efforts &industry-academy collaborations

JH / Oct. 2010

© Continental AG

Thank you for your Attention

Questions ? JH / Oct. 2010

© Continental AG

Suggest Documents