arsm -asfm reduction

28
ARSM -ASFM reduction RANS LES DNS 2-eqn. RANS Averaging Invariance Application DNS 7-eqn. RANS Body force effects Linear Theories: RDT Realizabi lity, Consisten cy Spectral and non-linear theories 2-eqn. PANS Near-wall treatment, limiters, realizability correction Numerical methods and grid issues Navier-Stokes Equations Dr. Girimaji Research Road map

Upload: cassia

Post on 22-Feb-2016

47 views

Category:

Documents


0 download

DESCRIPTION

Navier-Stokes Equations. DNS. Body force effects. Linear Theories: RDT. 7-eqn. RANS. Realizability, Consistency. Spectral and non-linear theories. ARSM -ASFM reduction. 2-eqn. RANS. Averaging Invariance. 2-eqn. PANS. Near-wall treatment, limiters, realizability correction. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: ARSM -ASFM  reduction

ARSM -ASFM reduction

RANSLESDNS

2-eqn. RANS

Averaging Invariance

Application

DNS

7-eqn. RANS

Body force effects

Linear Theories: RDT

Realizability, Consistency

Spectral and non-linear theories

2-eqn. PANS

Near-wall treatment, limiters, realizability correction

Numerical methods and grid issues

Navier-Stokes Equations

Dr. Girimaji Research Road map

Page 2: ARSM -ASFM  reduction

• Need for a new approach to modeling the scalar flux considering compressibility effects Mg effect • Application: Turbulent combustion/mixing in hypersonic aircrafts

Objective

Physical sequence of mixing:

(1) (( , i

Mixing Process steps

Large - Scale 2)Molecular - Scale scalar due to turbulent, u ) (fuel & oxidizer)

6444444444447444444444448

Turbulent Stirring Molecular Mixing Chemical Reaction

Page 3: ARSM -ASFM  reduction

Velocity Field(ARSM),

Scalar Flux Field(ASFM),

[Mona]

[Carlos] [Gaurav]

Scalar Dissipation Rate,

iu

i ju u

Turbulent Stirring Molecular Mixing Chemical Reaction

Turbulent mixing

Page 4: ARSM -ASFM  reduction

iu

Differential Transport eq.

Reduced Differential algebraic

Modeling

Weak Equilibriumassumption

Representation theory

,( , , , , , , , )i i i j i j ,i gu F u u U Θ k k M

Scalar Flux molding approaches

2

2

6 .

3 .

1 . / 2

1 . / 2

11 .

i j

i

i

eqn s u u

eqn su

eqn k u

eqn k

Total eqn s

Constitutive Relations

Page 5: ARSM -ASFM  reduction

ii

unormalized scalar flux

kk

:

: mm

k knormalized gradeint mean scalark x

ktime scale ratio r

k

/:

/

i i

2

TKE k ( u u )/2

half scalar variance : k ( )/2

i i m g tTrans(ξ )= f ( Θ r,Ma ,Ma )mn mnS ,W , ,

( , , )mn mn mn mnb b S W rARSM:

ij( , , )i mn mn g t jS W r,Ma ,Ma D

Weak equilibrium assumption

3/2

( . )gS kMa Pressureeffect vs Inertia effecta

:tMa Turbulent Mach number

ASFM with variable Pr_t effect

ξi i m g tTrans(ξ )= f ( Θ r,Ma ,Ma )mn mn mnb ,S ,W , ,

Algebraic Scalar Flux modeling approach:

Page 6: ARSM -ASFM  reduction

Algebraic Scalar Flux modeling approach (step-by-step)

Step (1) the evolution of passive scalar flux

ii i i

Du DiDt P i i j j j j iu u u U P

ii

ukk

( ) 1-2

i i iKD i

i iDt k k kk

D P PP/ 2i ik u u 2 / 2k

Step (2) Assumptions: • the isotropy of small scales • weak equilibrium condition, advection and diffusion terms 0

0i ( ) 0

D iiDt

D

12

i iKi k k kk

P PP

Step (3) Pressure –scalar gradient correlation i

Page 7: ARSM -ASFM  reduction

Algebraic Scalar Flux modeling approach (step-by-step)

Step (3) Modeling Pressure-scalar gradient correlation i

1. High Mg- pressure effect is negligible.

2. Intermediate Mg - pressure nullifies inertial effects.

3. Low Mg – Incompressible limit

[Craft & Launder, 1996]

( )ri i P 0iDu

Dt

( ) ( )r s θi θi θiΠ Π Π

( ) :rθiModifying Π

(s)θiΠ unmodified .

( )1 5 2 3 4( ) jr i

i k i j j i jk j i j

UUkc c u u c u c u c u uk x k x x x

( ) 0ri

0iu

Step (4) Applying ARSM by Girimaji’s group

Page 8: ARSM -ASFM  reduction

Algebraic Scalar Flux modeling approach (step-by-step)

Step (4) using ARSM developed by Girimaji group, ( , )ij ij ij ijb b S W

,1 1, (1 )2θ i ij j i, j j ij i j i i

k kN ξ = b Θ +U ξ N b Uε r

2 3

i j ijij

u ub

k

[Wikström et al, 2000])3

T

ijθ4i ij , j

θ

ij

1- cξ = ( )(b ΘN

1444444442444444443

3 21 1 1 22 2 ( ) 0N GN Q R N GQ R

/,/

θ1

21 θ5 θ4

22 θ5 θ4 S W

1 1G = 2c - tr bS -1-2 r

1R = 4( - c ) 1- c tr bΘ2

k1R = 4( - c ) 1- c tr (c S +c W)bΘ r2 k

T

θ4i j

θ

ij

1- c( )u u

N

: = Tensorial eddy diffusivity

Page 9: ARSM -ASFM  reduction

,( , , , , , )ij TT ij ij ij j gS W k k M

1. Standard k-ε model

1-a) with constant- Cμ =0.09

1-b) variable- Cμ with Mg effects which uses the linear ARSM [Gomez & Girimaji ]

Assume Pr_t = 0.85

2. Variable tensorial diffusivity

,Prp t

i eff eff lami t

cu k k

x

Ti ijj

ux

Preliminary Validation of the Model

2 /Pr

pi lam i

t

C C ku k

Page 10: ARSM -ASFM  reduction

Isentropic relations (compressible flows)

y

x

Fast stream Tt1 = 295 K, M=2.01 Pressure inlet

0.025

- 0.025X=0 X=0.5 X=0.1 X=0.15 X=0.2 X=0.25 X=0.3

slow stream Tt2 = 295 K, M=1.38 Pressure inlet

Geometry of planar mixing layer

for both free-stream inletsthe turbulent intensity =0.01 %, turbulent viscosity ratio = 0.1

Page 11: ARSM -ASFM  reduction

Fast stream

Slowstream

Pressure-inletPtot,1

Pstat,1

Pressure-inletPtot,1

Pstat,1

U1M1T1

U2M2T1

Pressure-outletTout

NRBC: avg bd. press.

Case 2 Case 3r Case 4 Case 5

R= U2/ U1 0.57 0.25 0.16 0.16

s =ρ2/ ρ1 1.55 0.58 0.60 1.14

Mr 0.91 1.44 1.73 1.97

M1, M2 1.91, 1.36 2.22, 0.43 2.35, 0.3 2.27, 0.38

Tt1, Tt2, (K) 578, 295 315, 285 360, 290 675,300

U1, U2, (m/s) 700, 399 561, 142 616, 1000 830,131

P(kPa) Inlet pressure [G&D, 1991] 49 53 36.05 32

Ps(kPa) inlet pressure [simulation] 56,49.5 57 40 37

Schematic of planar mixing layer

Page 12: ARSM -ASFM  reduction

•Normalized mean total temperature The mean total temperature is normalized by initial mean temperature difference of two streams and cold stream temperature. Due to the Boundedness of the total temperature, the normalized value, in theory, should remain between zero and unity.

2norm

T TTT

•Eddy diffusivity (eddy diffusion coefficient)For the approach (a), in which the turbulence model is the standard k-ε, the scalar diffusion on coefficient or eddy diffusivity is obtained by modeling the turbulent scalar transport using the concept of “Reynolds’ analogy” to turbulent momentum transfer. Thus, the modeled energy equation is given by

t i i i j eff j i ij heffkinetic energy work pressure work

E u E u p k T u S

144444424444443

23ij eff j i i j k k ijeff

u u u

12v i iE c T u u

Post-processing

Page 13: ARSM -ASFM  reduction

•Flux components1. Constant-/variable-Cμ

2. Tensorial eddy diffusivity Streamwise scalar flux:

Transversal scalar flux:.

Post-processing

•Thickness growth rate [ongoing]

1 1 2 ,2 1 1 ,1θ4

θ

1- cu ( ) u u u uN

2 2 2 ,2 2 1 ,1θ4

θ

1- cu ( ) u u u u

N

,i eff iu k

Page 14: ARSM -ASFM  reduction

1-a) Standard k-ε model with constant-Cμ

Page 15: ARSM -ASFM  reduction

Case -5Mr = 1.97

Case -3rMr = 1.44

Case -4Mr = 1.73

Normalized Temp Contours

Case -2Mr = 0.91

2norm

T TTT

1-a) Standard k-ε model with constant-Cμ

Page 16: ARSM -ASFM  reduction

Case -5Mr = 1.97

Case -3rMr = 1.44

Case -4Mr = 1.73

Bounded Normalized Temp Contours

Case -2Mr = 0.91

0 1normT

Page 17: ARSM -ASFM  reduction

Normalized Temp Profile 1-a) Standard k-ε model with constant-Cμ

0( )y yb

Fast stream

Slow stream

2norm

T TTT

Page 18: ARSM -ASFM  reduction

Eddy diffusivity profile 1-a) Standard k-ε model with constant-Cμ

Fast stream

Slow stream

0( )y yb

effk

Page 19: ARSM -ASFM  reduction

Scalar flux components 1-a) Standard k-ε model with constant-Cμ

Streamwise scalar flux @ x=0.2

Fast streamSlow stream

0( )y yb

1u

Page 20: ARSM -ASFM  reduction

Scalar flux components 1-a) Standard k-ε model with constant-Cμ

Transversal scalar flux @ x=0.2

Fast streamSlow stream

0( )y yb

2u

Page 21: ARSM -ASFM  reduction

Eddy diffusivity profile for case 5 (Mr=1.97), @ different stations

Fast stream

Slow stream

0( )y yb

effk

Toward outlet

Page 22: ARSM -ASFM  reduction

Comparing Scalar flux components, Axial vs. Transversal for Mr-1.8 (case5) and Mr 0.97 (case2)

1-a) Standard k-ε model with constant-Cμ

Page 23: ARSM -ASFM  reduction

1-b) Standard k-ε model with variable Cμ (Mg effect)

Page 24: ARSM -ASFM  reduction

Normalized Total Temp Profile @ x=0.02

2norm

T TTT

1-a) Standard k-ε model with constant-Cμ

1-b) Standard k-ε model with variable Cμ (Mg effect)

Fast stream

Slow stream

Page 25: ARSM -ASFM  reduction

1-a) Standard k-ε model with constant-Cμ

1-b) Standard k-ε model with variable Cμ (Mg effect)

Eddy Diffusivity Profile @ x=0.02

Page 26: ARSM -ASFM  reduction

1-a) Standard k-ε model with constant-Cμ

1-b) Standard k-ε model with variable Cμ (Mg effect)

Streamwise scalar flux @ x=0.02 1u

Page 27: ARSM -ASFM  reduction

1-a) Standard k-ε model with constant-Cμ

1-b) Standard k-ε model with variable Cμ (Mg effect)

Transversal scalar flux @ x=0.02 2u

Page 28: ARSM -ASFM  reduction

• All simulations were continued until a self-similar profiles (for mean velocity and temperature) are achieved in different Mach cases.

• Main Criterion to check convergence : imbalance of Flux (Mass flow rate ) across the boundaries (inlet & outlet) goes to zero. < 0.2%

• Error-function profile self-similarity state1.Normalized mean stream-wise velocity2.Normalized mean temperature

Convergence issues