28
Compressible Flow LES Using OpenFOAM I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands March 10, 2011 I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B Compressible Flow LES March 10, 2011 1 / 27

Compressible Flow LES Using OpenFOAM - Dutch OpenFOAM… · Compressible Flow LES Using OpenFOAM I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl PhD student, TU Delft, Delft,

Embed Size (px)

Citation preview

Compressible Flow LES Using OpenFOAM

I. B. Popovsupervisor: S. J. Hulshoff

promoter: H. Bijl

PhD student, TU Delft, Delft, The Netherlandsresearcher, NEQLab Research B.V., The Hague, The Netherlands

March 10, 2011

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 1 / 27

Introduction: Nanosecond plasma actuation forseparation control

Actuation designTwo electrodes, dielectriclayerPulse: U = 12 kV,τ = 12ns

Typical frequency∼ 100Hz . . . 1 kHz

-10

-5

0

5

10

-50 0 50 100 150 200 250V

olta

ge

, kV

Time, ns

Single: Ein = 11.6 mJDouble: Ein = 10.0 mJ

Triple: Ein = 9.5 mJ

Incident energy: 15 mJ

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 2 / 27

Experimental resultsGiuseppe Correale, TUDelft, 2010

Integral effectsCL ↑, CD ↓Stall delay of severaldegreesRe up to 3× 106 (60 cmchord, V = 80m/s)

Observed mechanismsAir jet up to 0.2 m/sShock waveHeated gas

α [deg]35

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 3 / 27

Schlieren imaging results

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 4 / 27

Proposed mechanism

1 Discharge ∼ 10ns

2 Fast heating of gas ∼ 1µs3 Formation of the shock wave4 Shock-flow interaction5 Creation of some coherent flow structures6 Separation elimination

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 5 / 27

Nanosecond actuator modelInstantaneous (1µs� Tflow)Constant-volume∆T ∼ 100K for 0.5× 0.5mm

Flow-wise distribution: filaments length ∼ 3 . . . 5mm,most energy released in ∼ 0.5mm at the electrodeSpan-wise distribution — two models:

1 Uniform2 Filaments with thickness ∼ 0.5mm and pitch of several

mm

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 6 / 27

Hybrid simulation

LES model

Nanosecond actuator

Separation zoneBoundary layer

DES or RANS model

1 3D LES simulation of near-actuator region2 RANS simulation of the rest of the flow3 Interaction between them

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 7 / 27

Periodic channel case

Flow

Wall

Wall y

xz

2

46

PropertiesDomain size 6× 4× 2Channel flow case, Reτ = 180Constant pressure gradient ∇p = 1x, z — periodic, top and bottom — no slipDNS data by Jimenez used as a reference

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 8 / 27

LES of incompressible channel flow

Verification pointsTemporal convergenceMesh refiningNumerical dissipationSGS modelsComputational costs

DifficultiesAveraging in OpenFoam’sdynamic Smagorinskymodel⇒ implementedtruly dynamic model

13

13.5

14

14.5

15

15.5

16

16.5

0 20 40 60 80 100 120

Ubulk

t

Reference value (15.68)Smag 64

dynSmag 32dynSmag 64

dynSmag 128

0.1

100

RM

S e

rror

n

1/nUmean

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 9 / 27

locDynSmagorinsky turbulent model

dynSmagorinsky

Averaging over whole domain⇒ used with vanDriestdumping

locDynSmagorinsky

Local computation of turbulent viscosityClipping of negative values

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 10 / 27

Temporal convergence

13

13.5

14

14.5

15

15.5

16

16.5

0 20 40 60 80 100 120

Ubulk

t

Reference value (15.68)Smag 64

dynSmag 32dynSmag 64

dynSmag 128

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 11 / 27

Comparison of turbulent modelsIntroduction

linear spatial schemebackward time scheme

Smagorinsky

dynSmagorinsky

locDynSmagorinsky

laminar (= no SGS model)

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 12 / 27

Comparison of turbulent modelsMean velocity profiles

0

2

4

6

8

10

12

14

16

18

20

0 0.2 0.4 0.6 0.8 1

Um

ea

n

y

DNSno model 64

3

Smag 643

dynSmag 643

locDynSmag 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 13 / 27

Comparison of turbulent modelsu variations profiles

0

0.5

1

1.5

2

2.5

3

0 0.2 0.4 0.6 0.8 1

<u

’2>

y

DNSno model 64

3

Smag 643

dynSmag 643

locDynSmag 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 14 / 27

Comparison of turbulent modelsv variations profiles

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 0.2 0.4 0.6 0.8 1

<v’2

>

y

DNSno model 64

3

Smag 643

dynSmag 643

locDynSmag 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 15 / 27

Comparison of turbulent modelsw variations profiles

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1

<w

’2>

y

DNSno model 64

3

Smag 643

dynSmag 643

locDynSmag 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 16 / 27

Comparison of turbulent modelsReynolds stress profiles

-0.8

-0.7

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

0 0.2 0.4 0.6 0.8 1

<u

’v’>

y

DNSno model 64

3

Smag 643

dynSmag 643

locDynSmag 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 17 / 27

Comparison of turbulent modelsConclusions

Laminar equations have surprisingly the bestperformanceStatic Smagorinsky with default coefficient has worstperformanceDynamic variants of Smagorinsky are somewhereinbetween.

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 18 / 27

Performance on stretched meshesMean velocity profiles

0

2

4

6

8

10

12

14

16

18

20

0 0.2 0.4 0.6 0.8 1

Um

ea

n

y

DNSuniform 64

3

grading 25 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 19 / 27

Performance on stretched meshesu variations profiles

0

0.5

1

1.5

2

2.5

3

0 0.2 0.4 0.6 0.8 1

<u

’2>

y

DNSuniform 64

3

grading 25 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 20 / 27

Performance on stretched meshesv variations profiles

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 0.2 0.4 0.6 0.8 1

<v’2

>

y

DNSuniform 64

3

grading 25 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 21 / 27

Performance on stretched meshesw variations profiles

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1

<w

’2>

y

DNSuniform 64

3

grading 25 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 22 / 27

Performance on stretched meshesReynolds stress profiles

-0.8

-0.7

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0

0 0.2 0.4 0.6 0.8 1

<u

’v’>

y

DNSuniform 64

3

grading 25 643

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 23 / 27

Comparison of turbulent modelsConclusions

Stretching does not improve performanceFilter delta for SGS model? (used cubeRootVol)

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 24 / 27

LES of compressible channel flow

RequirementsTransient solverConservative variablesTurbulence modelling

Self-made solverρ, ρU, ρETemporal scheme: implicit iterated or Runge-Kutta 4/3

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 25 / 27

Periodic hill case

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 26 / 27

Periodic hill case

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 27 / 27

Appendix: Discharge simulation

Poisson equation for electric field ε∆ϕ = qSeveral species: neutrals, positive and negative ions,electronsSeparate convection velocitiesKinetics (coupled solution is desired)Boltzman equation

I. B. Popov supervisor: S. J. Hulshoff promoter: H. Bijl ( PhD student, TU Delft, Delft, The Netherlands researcher, NEQLab Research B.V., The Hague, The Netherlands )Compressible Flow LES March 10, 2011 28 / 27