workshop hpc enabling of openfoam for cfd …€¦ · in-cylinder flow and combustion modeling...

In-cylinder flow and combustion modeling using the OpenFOAM® technology

Department of Energy, Politecnico di Milano

http://www.engines.polimi.it

https://twitter.com/ICEPoliMi/

T. Lucchini, G. Montenegro, G. D’Errico, A. Onorati, L. Cornolti

Workshop "HPC enabling of OpenFOAM for CFD applications"


Background

• Target for next IC engines generation: contemporary reduction of fuel consumption and pollutant emissions.

• Development and design focused on fuel-air mixing and combustion:

GDI engines (premixed or stratified)

Diesel engines (complex injection strategies)

New combustion modes (HCCI, PCCI)

• CFD widely used for IC engines design:

Development of a platform that can be used both for model implementation and engine design.


Topics

• Development of new combustion models for IC engines using the OpenFOAM® technology.

• Lib-ICE, model development for in-cylinder flow simulations:

Mesh management

Fuel-air mixing

Combustion

• Examples of application on real engine cases


Model implementation: Lib-ICE

• Mesh motion for engine geometries

• Combustion (spark-ignition, diesel)

• Lagrangian sprays + liquid film

• Unsteady flows in intake and exhaust systems:

Plenums, silencers, 1D-3D coupling

• Reacting flows in after-treatment devices

DPF, SCR, catalyst

Libraries and solvers for IC engine simulations based on OpenFOAM® technology


OpenFOAM-2.0.x

Lib-ICE

src

• Spray • Thermo-physical models • Combustion models • Liquid film • Engine mesh management • Boundary conditions • 1D-3D coupling

solvers

• Cold-flow • Diesel • Spark-ignition • Combustion • Spray

utilities

• Pre-processing • Parallel-processing • Post-processing • Mesh-management

Lib-ICE – code structure


MESH MANAGEMENT


MESH 1

XXX CA YYY CA

MESH 2

YYY CA ZZZ CA

1) Multiple meshes cover the entire cycle simulation.

2) Each mesh is valid in a user-specified interval.

3) During each time-step: Grid points are moved using

automatic mesh motion and/or pre-defined points motion.

Mesh topology can be eventually changed

4) Mesh-to-mesh interpolation.

Full-cycle simulation


Automatic mesh motion

• Grid points motion computed by an automatic mesh motion solver, based on the Laplace equation.

02 u

toldnew uxx

Layout for deforming mesh simulations



Topological changes


• Reduction of the overall number of meshes.

• Mesh is modified according to user-defined parameters, no need to specify exactly when a specific event should take place.

Dynamic mesh layering Sliding interface


Topological changes: Adaptive Local Mesh Refinement


• Grid resolution increased to properly describe the most important physical and chemical processes and reduce grid dependency:

Fuel-air mixing

Combustion


Mesh management

engine library in Lib-ICE

engine

engineTopoChangerMesh

fvMotionEngineMesh

pistonLayerAREngineMesh

twoStrokeEngineMesh

pistonRefineEngineMesh

Deforming grid for full-cycle simulations in four-stroke engines

Compression-expansion with dynamic mesh layering

Full-cycle simulation in four-stroke engines

Compression-expansion with adaptive local mesh refinement

multiCycleEngineTime



1) Generation of multiple meshes 2) Definition of a template case

3) engineCaseSetUp utility case

constant

init

system

case0300

case0320

……

case0720

case1440

……

dataOutput

4) runMultiCycleCase solverName simpleColdSpeciesEngineDyMFoam;

startTimes

(/* list of startTime for each mesh */);

endTimes

(/* list of endTimes for each mesh */);

writeInterval

(/* list of writeInterval for each mesh */);

parallelRun on;

5) Post-processing

Specific utilities for case set-up


Generation of engine meshes

Grid imported from different tools

(Gambit®, ICEM-CFD®, Pointwise®,

STAR-CD®)

Fully automatic hexahedral mesh

generation from stl file



Mesh generation: new approach based on Blender®

Blender + snappyHexMesh: engine mesh generation exclusively based on open-source tools




Hydrogeninjector

Cylinderhead

Exhaust

Intake

Tumbleplate

Laser532/266 nm

Piston

Engine block

Opticalliner

Displaced volume 560 cm3

Compression ratio 11

Intake pressure/temperature 1 bar / 36 °C

Engine speed 1500 rpm

SANDIA Hydrogen Engine (data from ECN)

Fixed zones

Deforming zones

Motored conditions. Experimental pressure and temperature profiles imposed at intake and exhaust ports.

• Mesh validity: 15 deg each. Average cell size ranging from 300 to 600 thousand.

• Grid size: Cylinder: 2.5 mm Valve region: 0.75 mm Ducts: 2.5 – 5 mm

• Turbulence model: standard k-e


Exp. Calc.

The main flow features at BDC are almost preserved at this time, when the intake valve is closing.

-140 CA (IVC: -140)

SANDIA Engine: velocity field comparison




Gas pressure now higher than ambient pressure.

Still rather good agreement between computed and experimental data.

Exp. Calc.

-70 CA (IVC: -140)

SANDIA Engine: velocity field comparison



Post-processing

0

1

2

3

4

5

6

7

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

435 465 495 525 555 585 615 645 675 705 735

Crank Angle [deg]

Tumble ratio

Turb. Kin. energy

SOI EOI

IVC

Cylinder pressure In-cylinder charge motions

Specific utilities to estimate relevant quantities: charge motions (tumble, swirl ratio, turbulence intensity); air-fuel mixing: homogeneity index.


SPRAY AND WALL FILM


Comprehensive spray model for Diesel and GDI sprays

Turbulence induced breakup

Wave + Catastrophic

breakup

• Breakup transition: Weber number.

Spray model


• Extending the capabilities of the standard OpenFOAM spray library, by introducing new models for:

High pressure injection

Liquid fuel atomization

Droplet-wall interaction

Liquid evaporation

• Reducing the grid-dependency and CPU time of spray simulations:

Easy development of new models.

Fast tuning of existing models.

Spray model

Comprehensive spray model for Diesel and GDI sprays


Mesh management

dieselSpray library in Lib-ICE

dieselSpray

spray

parcel

spraySubModels

injectorModel

atomizationModel

wallModel

breakupModel

Fully re-implemented

Partially re-implemented

Huh model with and without cavitation

Huh-Gosman model

Bai-Gosman model

Bug fixes


Liquid film model

Liquid film

• Approach proposed by Bai and Gosman (SAE-960626).

• Mass, momentum and energy equations solved for a thin film using the Finite Area Method accounting for both interaction with liquid (impact) and gas phase (evaporation).


Spray model

Non-evaporating diesel spray

Experimental data courtesy of Dr. Montanaro and Dr. Allocca (CNR-Istituto Motori)


Spray model

Evaporating diesel spray (ECN data)

• Fuel: n-dodecane (“Spray-A”)

• Effects of ambient conditions (temperature, density) on liquid and vapor penetration.

Exp data from ECN: www.ca.sandia.gov/ecn


• Air/fuel mixing and velocity comparison

• Comprehensive spray model validation absolutely necessary for a correct prediction of the combustion process.

Spray model

Evaporating diesel spray (ECN data)

Exp. Calc.

20

30

40

50

Dis

tan

ce [m

m]

(a) (b) (c)

• Fuel: n-heptane (Spray-H)

Exp data from ECN: www.ca.sandia.gov/ecn


Spray model

Impinging gasoline spray

(a) t = 400 μs (b) t = 2300 μs

Liquid film formation on the plate

Experimental data courtesy of Dr. Montanaro and Dr. Allocca (CNR-Istituto Motori)


COMBUSTION MODELS


Background

• Combustion models for IC engines. Requirements:

Flexibility multiple-injections stratified charge combustion different combustion regimes

Detailed chemistry pollutant formation (soot) auto-ignition multi-component fuels

Interacting processes Energy transfer during spark-ignition Turbulence-chemistry interaction


Background

• Implementation of combustion models in Lib-ICE

• Design-oriented tools:

Fast, robust and suitable for industrial simulations

Reduced chemistry

• Diagnostic-oriented tools:

Allowing a detailed analysis of the flame structure and study of new combustion modes

Detailed kinetics


COMBUSTION MODELS Spark-ignition


Combustion models for SI Engines

• Combustion phases: spark-ignition and turbulent combustion.

• State of the art: different combustion models available (G-Equation, Weller, ECFM) and extensively validated.

• However, initial flame development plays a big role and modeling efforts are still required to describe such phase. Influence of local flow, turbulence conditions and energy transfer from the electrical circuit to be accounted for.

• Proposed approach: Lagrangian ignition model + turbulent combustion model (Eulerian)



• Lagrangian ignition model

1) The spark channel is represented by a set of Lagrangian particles, convected by the mean flow.



KaKa

2) Flame kernels are launched at the particle locations satisfying the ignition criterion (Karlovitz). For each particle mass and energy equations are solved.


plasma

u

p

tup

pssr

dt

dm

4

ppp

spk

bp

p

pp

cm

QTT

m

m

dt

dT

,



3) Flame surface density distribution reconstructed from particle distribution and their size.

cell

N

i

ii

spkV

Sf

1





4) Initial thermal transient of the plasma channel correctly described by solving the heat conduction equation on it.

ri

Ti (>10000 K) Tu (300-600 K)

Qel

Temperature [K]

a [

m2/s

]

Temperature [K]

c p [

kJ/j

kgK

]

plp

pl

plp

elpl

pl

Vc

titVT

Vc

QT

t

T

a

a

22




5) Simplified model for the secondary circuit to estimate the amount of energy transferred to the gas phase

titVQ

lVVtV

L

Eti

titVtiRdt

dE

spk

spkgcacs

s

ss

sssss

2

2

Rp Rs

Lp Ls Spark

gap




6) Ignition source term included in the flame surface density transport equation:

k

it

t

ii

i PDPxScScxx

u

t

Possibility to choose different expressions for P and D depending on the selected FSD approach.



sparkIgnition library in Lib-ICE

flameKernelIgnition

flameKernel

electricalCircuit

plasmaChannelModel regionModel

particle

Cloud<flameKernel>

sparkIgnitionSubModels ignition, restrike, wall, dispersion, efficiency

Solver: lagrangianPlasmaFlameKernelEngineFoam

flameSurfaceDensityModels CFM, ECFM, CFM-2, Chang-Huh, ..



• Pre-chamber with optical access. • Fuel: propane; laminar flame speed:

Gulder

• Tested operating conditions:

1) Effect of engine speed 2) Effect of air/fuel ratio 3) Effect of spark-plug position

• Grid generation: snappyHexMesh

Engine geometry details, operating conditions and complete set of experimental data available in SAE Paper 922243.

Experimental validation: optical pre-chamber engine




Non-combusting conditions

Correct prediction of turbulence intensity and velocity very important for realistic simulation of the combustion process

0

4

8

12

16

310 320 330 340 350 360

no

n-d

ime

nsi

on

al ra

tio

Crank Angle [deg]

Calc.

Exp.

u'/up Umean/up

velocity Turbulence intensity

Ignition locations




Computed vs experimental burned volume

Effect of engine speed Effect of air/fuel ratio at 1000 rpm


COMBUSTION MODELS Diesel


Diesel combustion

• Complex process involving:

Detailed fuel chemistry (auto-ignition, pollutant formation)

Spray evolution

Multi-component fuels

Turbulence-chemistry interaction

• Implemented approaches:

CTC (reduced chemistry) Well-mixed model (detailed chemistry) Flamelet-combustion models (detailed chemistry)


Diesel combustion: CTC model

• 11 chemical species (fuel, O2, N2, CO, CO2, H2O, O, OH, NO, H, H2)

• Auto-ignition computed by the Shell auto-ignition model; turbulent combustion simulated accounting for both laminar and turbulent time scales:

• Temperature threshold to switch between auto-ignition (a=1)

and turbulent combustion (a=0).

CTCiShellii YYY ,, 1 aa

Characteristic time-scale combustion model


Validation: passenger-car, common-rail engine

Operating conditions:

• Full load with post-injection to reduce soot

• Medium high load with:

Variable EGR rate

Three injection events (pre, main, post)

Bore 85 mm

Stroke 88 mm

Compression ratio ~15

Models:

• Modified Huh-Gosman for atomization. Diesel fuel approximated as C12H26

• CTC+Shell for combustion. C12H26 Shell-model constants proposed by Reitz.



Cylinder pressure validation at full load

0%

100%

EGR Qpil Qmain Qpost

Full load, single

injection, No EGR



Soot emissions, full load

0.5

0.6

0.7

0.8

0.9

1

0 1 3 5 7 9

No

n d

ime

nsi

on

al s

oo

t

Mass of fuel in the post-injection [mg/stroke]

Effect of post injection on soot-emissions Measured

Computed

Post-injection effect on soot emissions



Medium-high load, 0% EGR

3 injections

Cylinder pressure validation at mid-load

0%

100%


2300 rpm imep = 10.5 bar



Medium-high load, 34% EGR

3 injections

Cylinder pressure validation at mid-load

0%

100%


2300 rpm imep = 10.5 bar



Soot emissions, medium-high load

Injection strategy and external EGR effects on soot and NOx emissions

0

0.2

0.4

0.6

0.8

1

3.5 19.5 23.5 33.5 (std.)

33.5 (adv.)

No

n-d

ime

nsi

on

al s

oo

t

Measured

Computed

0

100

200

300

400

500

600

700

3.5 19.5 23.5 33.5 (std.)

33.5 (adv.)

NO

x[p

pm

] Measured

Computed

External EGR [%] External EGR [%]



Diesel combustion: well-mixed model

• Direct integration of complex chemistry in each computational cell. Each cell is an homogeneous reactor and species reaction rates are computed by means of an ODE stiff solver:

tt

t

iiii dtW

tYttY '*

t

tYttYY ii

i

*

• Detailed approach but very time-consuming, mainly when a realistic mechanism (>50 species) is employed.


Diesel combustion: TDAC

• TDAC: Tabulation of Dynamic Adaptive Chemistry

• Combination of mechanism reduction and tabulation techniques

• Work developed in collaboration with Dr. F. Contino (Vrije Universiteit Brussel) and Prof. H. Jeanmart (Univ. of Louvain)

CPU time speed-up compared to direct-integration is of the order of n x m (n: number of species, m: number of reactions) .

Very general approach, flexible with respect to the mechanism and fuel composition.

Up to 300 species and 1000 reactions can be used in a reasonable amount of time.



chemistryModelPolimi library in Lib-ICE

chemistryModelPolimi

chemistryModel

chemistrySolversTDAC

mechanismReduction

tabulation

Solvers: dieselFoamAutoRefine, dieselEngineDyMFoam

ISAT

DAC, DRG, DRGEP, ..

Developed in collaboration with Dr. F. Contino and Prof. H. Jeanmart.



Auto-ignition and flame lift-off in SANDIA ECN Spray-A Diesel flame

103 species, 370 reactions mechanism for C12H26 (Sarathi et al.)

Effects of oxygen concentration and ambient temperature: IGN. DELAY



Auto-ignition and flame lift-off in SANDIA ECN Spray-A Diesel flame

103 species, 370 reactions mechanism for C12H26 (Sarathi et al.)

Effects of oxygen concentration and ambient temperature: LIFT-OFF



Flame structure in the SANDIA ECN Spray-A Diesel flame

300 species, 8900 reactions mechanism for C12H26 (Ranzi et al.)

Temperature OH Acetylene BIN1-A



Air-fuel mixing model validation

7 mm, exp 12 mm, exp 18 mm, exp

7 mm, calc 12 mm, calc 18 mm, calc

Combustion model validation

PRF30 fueled engine:

ALMR employed to better predict both the fuel-air mixing process and flame propagation.

Detailed mechanism for PRF fuels oxidation

PCCI combustion in an optical engine


Diesel combustion: unsteady flamelets

• Multiple unsteady flamelets: a set of unsteady diffusion flames represents Diesel combustion.

• First model towards the definition of a generic model for partially-premixed combustion

xH~

pst ,~

xT~ xYhxH ii

~~

dtYtxftxY izi ,~

,;~

,~

1

0

CFD Code

xZxZ 2'',

~

FlameletCode

tZYi ,~

Detailed chemistry and turbulence-chemistry interaction

Flamelet Equations

Y

Z

YZ

t

Y ii

2

2

2

chem

ss qZ

hZ

t

h

2

2

2



Detailed chemistry and turbulence-chemistry interaction

• CFD domain: transport equations are solved for mixture fraction Z

and variance Z’’2 , momentum, mass, enthalpy h.

• Flamelet domain takes the average scalar dissipation rate conditioned on the stoichiometric mixture fraction (st) and use it in the flamelet equations for mass and energy in the mixture fraction space.

• Possibility to use multiple flamelets, with two possible criteria for division: mass or scalar dissipation rate.

• Chemical composition in CFD cells:

N

f

fifi dZZpZYIY1

1

0,xx



Implementation of a flamelet combustion model

• Use of any pre-implemented capability to exploit the entire potentialities that are offered by the OpenFOAM technology:

Chemistry model + in-house developments (ISAT, DAC, …)

Thermodynamics

Parallelization using existing domain decomposition techniques.

Pre-implemented functions (average, integration, matrix algebra…) to solve flamelet equations and to exchange data between the phase space and the physical domain.


• regionModel approach for flamelet combustion models:

Flamelet mesh, representing the Z space

Flamelet equations (species and enthalpy) solved on the flamelet mesh, relying on native OpenFOAM matrix algebra and ODE solvers.

PDF averaging:

Standard OpenFOAM functions (integration, averaging) applied on the flamelet mesh, that requires accurate grading at the Z = 0 and Z = 1 boundaries.


Implementation of a flamelet combustion model



flameletCombustionModels

RIF

flameletCombustionModel

scalarDissipationRate

Peters

scalarDissipationRate

pdfFunctions

Abstract class

Implementation of Representative Interactive Flamelet Model

Operations to compute scalar dissipation rate (st, (Z))

Solver: RIFdieselFoam

flameletCombustionModels library in Lib-ICE



Simulation: SANDIA ECN Spray H flame (n-heptane)

Flamelet mesh

0 1 z

The grid is refined at the boundaries (Z=0, Z=1), to correctly integrate the PDF function in presence of high mixture fraction variances, as suggested by Liu et al.


Initial conditions in the flamelet domain: • Temperature in Z-domain • Species in the Z-domain


Flamelet mesh

Temperature in Z-domain

Species




Flamelet mesh


Species

Scalar dissipation

rate

Cool flame



Diesel combustion: unsteady flamelets Simulation: SANDIA ECN Spray H flame (n-heptane)

Flamelet mesh


Species

Scalar dissipation

rate

Main ignition


t = 0.9 ms

Diesel combustion: unsteady flamelets Simulation: SANDIA ECN Spray H flame (n-heptane)

Mixture fraction Mixture fraction variance

Stoichiometric scalar dissipation rate Temperature

Flame structure in CFD domain computed at steady state conditions




0

0.5

1

1.5

2

8 12 16 20 24

Ign

itio

n d

ela

y [m

s]

Oxygen concentration [%]

Calc.

Exp.

amb= 14.8 kg/m3

amb= 30 kg/m3

0

0.3

0.6

0.9

1.2

800 1000 1200 1400

Ign

itio

n d

ela

y [m

s]

Ambient temperature T [K]

Calc.

Exp.


Acknowledgments

• Ing. Rita Di Gioia, Ing. Giovanni Bonandrini, Ing. Mario Picerno, Ing. Luca Venturoli, Magneti Marelli, Bologna.

• Dr. Francesco Contino, Vrije Universiteit Brussel.

• Ing. Marco Fiocco, Politecnico di Milano.

• Paolo Colombi, MSc. Student, Politecnico di Milano

• Dr. Luigi Allocca, Dr. Alessandro Montanaro, CNR-Istituto Motori


Thanks for your attention!


https://twitter.com/ICEPoliMi/

workshop hpc enabling of openfoam for cfd …€¦ · in-cylinder flow and combustion modeling...

Documents