APPLICATION OF ZONAL HYBRID URANS/LES MODELING TO …projet.ifpen.fr/Projet/upload/docs/application/pdf/2019... · 2019. 1. 3. · APPLICATION OF ZONAL HYBRID URANS/LES MODELING TO

APPLICATION OF ZONAL HYBRID URANS/LES MODELING TO INTERNAL COMBUSTION ENGINE FLOWS

V. K. Krastev1, A. D’Adamo2, S. Breda3, S. Fontanesi2

1 Department of Economics, Engineering, Society and Business Organization, University of Tuscia,

01100 Viterbo, Italy

2 DIEF - "Enzo Ferrari" Engineering Department, University of Modena and Reggio Emilia, 41125

Modena, Italy

3 R&D CFD SRL - Via Pietro Vivarelli 2, 41125 Modena, Italy

LES4ICE 2018

Outline

V. K. Krastev et al. 2

• Background o URANS/LES hybrids in ICE modeling

o Motivations

• Zonal modeling formulation & calibration

• Applications o Static valve intake flow

o TCC-III Engine multi-cycle analysis

• Conclusions

LES4ICE 2018

Outline



o Motivations




• Conclusions

LES4ICE 2018

Outline



o Motivations




• Conclusions

LES4ICE 2018

Outline



o Motivations




• Conclusions

LES4ICE 2018

Background


URANS/LES hybrids in the ICE modeling community

Cumulative graph of the published journal papers with relevant hybrid

URANS/LES applications in the ICE field (source+: www.scopus.com)

+The collected data might not be fully exhaustive

o LES methods for ICE modeling are being developed

for 25 years, but …

o …in recent years the interest for URANS/LES hybrids

is apparently growing

o Potential gains in computational robustness and

efficiency, keeping scale-resolving capabilites

where needed

o Several seamless (SAS, DES, DLRM) and zonal

(ZDES) alternatives proposed

http://www.scopus.com/

LES4ICE 2018

Background












where needed




LES4ICE 2018

Background












where needed




LES4ICE 2018

Background


Meanwhile, in the aerospace/turbomachinery engineering fields…

Complete aircraft at scale 1:1 in true flight conditions (Ma=0.8, Re ~ 5∙107, ~ 2∙108 grid points)+

+Picture taken from: S.Deck et al, Phil. Trans. R. Soc. A 372: 20130325 (2014).

o DES and Zonal-DES modeling have proven to be highly efficient in complex aerospace

applications

o The flexibility of hybrid modeling allows the accurate resolution of different types of internal

and external, flows within the same computational domain

o Care is needed for the optimal domain decomposition and numerics choice

LES4ICE 2018

Background


o Single, URANS-based turbulence modeling framework

o Activation of LES only where actually needed

o Less troublesome BCs

o Reported to be competitive with LES on sub-optimal grids,

especially for wall-impinging flows+

+See e. g. K. Keskinen et al. International Journal of Heat and Fluid Flow 65 (2017)

141–158

Why (Z)DES for ICE?

LES4ICE 2018

Background


o Assess if a (relatively) basic ZDES model is capable of

handling complex multi-cycle engine simulations with a

reasonable level of accuracy

Comparable with experimental data sets

Comparable with modern LES approaches on engineering-

grade meshes

More specifically:

LES4ICE 2018

Modeling


From two-equation URANS to (Z)DES

Basis:

o General URANS two-equation form:

o Strelets et al. (2001) showed that URANS can be

turned into DES by modifying Sk only

TKE sink term modification (1)

,

k k k k

t

kt

t

f k

C P S D

C P S D

LES4ICE 2018

Modeling



Basis:





3/2

,; ,

RANSk RANSRANS

kl f k

l S

3/2

,; ,

DES RANS DESk DESDES

kl min l C

l S

(gridf )

,

k k k k

t

kt

t

f k

C P S D

C P S D

DESC O(1)

LES4ICE 2018

Modeling



Basis:





3/2

,; ,

RANSk RANSRANS

kl f k

l S

3/2

,; ,

DES RANS DESk DESDES

kl min l C

l S

,

k k k k

t

kt

t

f k

C P S D

C P S D

, ,DESk DES k RANSFS S

max / ,1DES RANS DESlF C

Final

form

LES4ICE 2018

Modeling



TKE sink term modification (2) Seamless DES:

o URANS-to-LES managed entirely by the model (user

decisional load kept to a minimum), but…

o …the triggering mechanism is not always efficient

Zonal DES:

o URANS and LES (or DES) regions explicitly marked by

the user

o Better control of the solution behavior (at the expense

of nontrivial a-priori decisions)

o Good candidate for complete ICE simulation

* *, ,DESk DES k RANSFS S

*1 11DES z DES z ZDESC F C FF

2 21 RANSZDES z z

DES

lC C

CF

LES4ICE 2018

Modeling



TKE sink term modification (2) Seamless DES:

o URANS-to-LES managed entirely by the model (user

decisional load kept to a minimum), but…

o …the triggering mechanism is not always efficient

Zonal DES:

o URANS and LES (or DES) regions explicitly marked by

the user

o Better control of the solution behavior (at the expense

of nontrivial a-priori decisions)

o Good candidate for complete ICE simulation

* *, ,DESk DES k RANSFS S

*1 11DES z DES z ZDESC F C FF

2 21 RANSZDES z z

DES

lC C

CF

Cz1 Cz2 Simulation

type Mode

0 1 URANS I

1 1/0 DES II

0 0 LES III

LES4ICE 2018

Modeling


Implementation overview

Which turbulence model?

o RNG k-e model

• Improved, shear-dependent e-equation

• Widely adopted in the ICE community

Methodology implementation/calibration:

o The CDES value must ensure a consistent turbulent

energy decay in mode III (pure LES)

o The optimal value depends on:

• the specific underlying turbulence model (k-e,

k-w…)

• the accuracy/behavior of the discretization

schemes

o Calibration tests needed

Which CFD code?

o STAR-CD v4.22c (Siemens PLM)

• Unstructured FVM code

• Second-order accurate in space and time

• ZDES formulation implemented through user-

supplied Fortran subroutines

LES4ICE 2018

Modeling


CDES calibration: turbulence box

o Standard test for DNS and SGS models

o Cubic domain with cyclic BCs in each direction, N3

perfectly cubic cells (N=64)

o Flow field initialized with an incompressible

divergence-free turbulent spectrum

o Turbulence is left to spontaneously decay driven

by mode III of ZDES

o CDES is varied in the 0.4-0.61+ range

o MARS scheme for momentum convection (BF = 0.5

and BF = 1)

o Comparisons with the Comte-Bellot and Corrsin§

database (energy spectra)

+CDES = 0.61 is the first “standard” value reported in the literature for k-e based DES models

§G. Comte-Bellot G and S. Corrsin Journal of Fluid Mechanics 1971; 48(2): 273–337.

LES4ICE 2018

Modeling


CDES calibration: turbulence box

o The BF effect is dominant

o BF = 1 assumed as default for the LES mode (mode III)

o CDES = 0.61 assumed as standard

t = t1 t = t2

LES4ICE 2018

Applications – Static valve


Case overview

Preliminary remarks: o Intake port geometry with an axis-centered fixed poppet valve, Reb ≈ 3 x 104

o LDA (velocity) and axial pressure measurements available

Valve Stem Diameter (Ds) 16 mm

Intake Duct Diameter (Di) 34 mm

Cylinder Diameter (Dc) 120 mm

Valve Head Diameter (Dv) 40 mm

Intake Duct Length (Li) 140 mm

Cylinder Length (Lc) 300 mm

Valve Lift (LL) 10 mm

Inlet Bulk Velocity (Ub) 60 m/s

Fluid air at NIST standard conditions

Inlet Bulk Reynolds Number (Reb) ~ 3 x 104

LES4ICE 2018



Case overview

Preliminary remarks: o Intake port geometry with an axis-centered fixed poppet valve, Reb ≈ 3 x 104

o LDA (velocity) and axial pressure measurements available

o Mode I numerics: MARS 0.5 for momentum and scalars

o Mode II/III numerics: MARS 1 for momentum, MARS 0.5 for scalars

o Standard inflow/outflow incompressible BCs, WFs at the walls

o Comparisons with experiments and OpenFOAM results (same turbulence model and grid, similar numerics)

Mode II = Z1

Mode III = Z2

LES4ICE 2018



Results (x = 20 mm)

Mean axial velocity RMS axial velocity fluctuation

o Significant improvements compared to steady RANS profiles (code-by-code differences due to numerics)

o Z1 and Z2 time-averaged profiles very close (Mode II is likely to trigger LES in most of the domain)

LES4ICE 2018



Results (axial pressure development)

LES4ICE 2018



Results (axial pressure development)

o Even more pronounced improvements compared to RANS (same code-by-code differences)

o Z1 and Z2 time–averaged profiles very close (Mode II is likely to trigger LES in most of the domain)

LES4ICE 2018

Applications – TCC III


Case overview

Preliminary remarks (1): o Selected for the ease of reproduction and large experimental PIV data set

o 1300 RPM and 40 kPa intake manifold pressure

o Availability of LES data sets from UniMoRe

Engine Data

Bore 92 mm

Stroke 86 mm

Connecting rod length 231 mm

Geometrical compression ratio 10:1

Operation 4-stroke

N. of valves 2

Combustion chamber type Pancake

RPM 1300

Intake manifold pressure 40 kPa

LES4ICE 2018



Case overview

Preliminary remarks (2): o Full domain splitted in 5 zones (cylinder, intake/exhaust ports, plenums)

o Only the cylinder zone treated in scale-resolving mode (Mode II or Mode III)

o Max 2M cells at BDC, ~ 0.75 mm base cell dimension in the cylinder

Mode II = TCC-Z1

Mode III = TCC-Z2

LES4ICE 2018



Case overview

,1

1 n

i i jj

u un

2

, ,1

1

1

n

i rms i j ij

u u un

PIV grid point

cycle number

50

i

j

n

Preliminary remarks (3): o Postprocessing initially focused on the vertical planes

o S_2013_10_24_01 and S_2014_02_05_02 reference data sets (same windows/resolution in the simulations)

o Previous LES DSM results (on the same CFD grid) added for comparison

LES4ICE 2018



Results (Y = 0): 475 CA E

ns

. A

vg

. R

MS

LES4ICE 2018



Results (Y = 0): 475 CA

o Good average jet shape prediction by TCC-Z1 (added modeled visosisty?)

o Underestimation of RMS fluctuations close to the piston head

En

s. A

vg

. R

MS

LES4ICE 2018




ns

. A

vg

. R

MS

LES4ICE 2018




ns

. A

vg

. R

MS

o DSM closer to experiments

o TCC-Z1 slighlty more consistent with DSM and experiments

LES4ICE 2018




ns

. A

vg

. R

MS

LES4ICE 2018




ns

. A

vg

. R

MS

o DSM still in good agreement with the experiments

o Anomalous shift of the RMS peak values

LES4ICE 2018



Results (X = 0): 475 CA E

ns

. A

vg

. R

MS

o Good overall consistency for all numerical predictions

o Small differences persist between TCC-Z1 and TCC-Z2

LES4ICE 2018




ns

. A

vg

. R

MS

LES4ICE 2018




ns

. A

vg

. R

MS

o Much better main vortex average shape description for TCC-Z1 and TCC-Z2

o DSM seems to overestimate RMS fluctuations

LES4ICE 2018




ns

. A

vg

. R

MS

LES4ICE 2018




ns

. A

vg

. R

MS

o Vortex compression better described by DSM

o TCC-Z1 (still) slightly more consistent than TCC-Z2 (average flow)

LES4ICE 2018

Conclusions


Final remarks

o Is the evaluated ZDES formulation reasonably adequate for

engine multi-cycle simulation?

LES4ICE 2018

Conclusions


Final remarks

o Is the evaluated ZDES formulation reasonably adequate for

engine multi-cycle simulation?

Yes, but…

LES4ICE 2018

Conclusions


What’s next?

o More TCC-III results (in progress):

• complete the analyses on the vertical planes (LES

quality indicators, LES or DES for the cylinder?);

• more planes;

• different zonal setups.

o Modeling aspects (LES mode, wall modeling)

o Different engines (pentroof, 4 valves, GDI?)

o Combustion?

LES4ICE 2018

Conclusions


Acknowledgments

GM (through the GM University of Michigan

Automotive Cooperative Research Laboratory,

Engine Systems Division).

Siemens PLM Software Inc.

TCC-III database

STAR-CD software licensing (through

UniMoRe HPC resources)

LES4ICE 2018

Conclusions


Documents

APPLICATION OF ZONAL HYBRID URANS/LES MODELING TO …projet.ifpen.fr/Projet/upload/docs/application/pdf/2019... · 2019. 1. 3. · APPLICATION OF ZONAL HYBRID URANS/LES MODELING TO