71
Turbulent bluff-body separation: recent advances of a self-consistent flow description for large Reynolds numbers Bernhard Scheichl [email protected] Institute of Fluid Mechanics and Heat Transfer Seminar lecture Department of Mathematics, UCL, 10 Nov 2009 B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 1 / 34

Turbulent bluff-body separation: recent advances of a self ... · Turbulent bluff-body separation: recent advances of a self-consistent flow description for large Reynolds numbers

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Turbulent bluff-body separation: recent advances of aself-consistent flow description for large Reynolds numbers

Bernhard [email protected]

Institute of Fluid Mechanics and Heat Transfer

Seminar lecture

Department of Mathematics, UCL, 10 Nov 2009

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 1 / 34

The research presented has been carried out in collaboration with

Prof Alfred Kluwick

Institute of Fluid Mechanics and Heat Transfer

Prof Frank T. Smith, FRS

Department of Mathematics

Their contributions are greatly acknowledged.

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 2 / 34

Sandborn & Liu (JFM, 1968)

“Turbulent boundary-layer separation is normally listed as one ofthe most important unsolved problems in fluid mechanics...”

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 3 / 34

Overview

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 4 / 34

Wall-bounded high-Reynolds-number flows : Re := UL/ν ≫ 1

x ,u

y

y ∼ δ

ue(x)

uBL

inviscid, irrotational

Navier–Stokes eqs (ρ = const)

∇ · u = 0 ,∂u∂t

+ u · ∇u = −∇p +1

Re∇2u , u|y=0 = 0

Reynolds-averaging, ergod hypothesis (nominally steady 2D flow)

Q(x , t ; Re) = 〈Q〉(x , y ; Re) + Q′(x , t ; Re)

〈Q〉 := limtav→∞

1tav

∫ tav /2

−tav /2Q(x , t + t ′; Re) dt ′

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 5 / 34

Wall-bounded high-Reynolds-number flows : Re := UL/ν ≫ 1

x ,u

y

y ∼ δ

ue(x)

uBL

inviscid, irrotational

Navier–Stokes eqs (ρ = const)

∇ · u = 0 ,∂u∂t

+ u · ∇u = −∇p +1

Re∇2u , u|y=0 = 0

Reynolds-averaging, ergod hypothesis (nominally steady 2D flow)

Q(x , t ; Re) = 〈Q〉(x , y ; Re) + Q′(x , t ; Re)

〈Q〉 := limtav→∞

1tav

∫ tav /2

−tav /2Q(x , t + t ′; Re) dt ′

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 5 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 6 / 34

Numerically-based methods

• DNS : resolution of all scales ⇒ no modelling needed but at presentrestricted to moderately large values of Re

• LES : modelling of short scales ⇒ reduction of computational efforts,which, however, are still too massive to be useful for the solutionof engineering problems

• RANS : modelling of all scales ⇒ engineering problems can be solvedwith an acceptable amount of computational efforts, which,however, increase with increasing values of Re

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 7 / 34

Numerically-based methods

• DNS : resolution of all scales ⇒ no modelling needed but at presentrestricted to moderately large values of Re

• LES : modelling of short scales ⇒ reduction of computational efforts,which, however, are still too massive to be useful for the solutionof engineering problems

• RANS : modelling of all scales ⇒ engineering problems can be solvedwith an acceptable amount of computational efforts, which,however, increase with increasing values of Re

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 7 / 34

Numerically-based methods

• DNS : resolution of all scales ⇒ no modelling needed but at presentrestricted to moderately large values of Re

• LES : modelling of short scales ⇒ reduction of computational efforts,which, however, are still too massive to be useful for the solutionof engineering problems

• RANS : modelling of all scales ⇒ engineering problems can be solvedwith an acceptable amount of computational efforts, which,however, increase with increasing values of Re

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 7 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 8 / 34

Analytically-based methods

• full NS eqs : starting efforts, but still no complete theory existsDeriat & Guiraud (J Mec Theor Appl, 1986), Walker (AIAA J, 1989, CISM, 1998),

Scheichl & Kluwick (JFS, 2008)

• non-dimensional Reynolds-averaged NS eqs(nominally 2D, curvature effects on BL flow of higher order):

∂u∂x

+∂v∂y

= 0−τ

u∂u∂x

+ v∂u∂y

= −∂p∂x

− ∂〈u′2〉∂x

− ∂︷ ︸︸ ︷

〈u′v ′〉∂y

+1

Re∇2u

∇2 =∂2

∂x2 +∂2

∂y2

u∂v∂x

+ v∂v∂y

= −∂p∂y

− ∂〈u′v ′〉∂x

− ∂〈v ′2〉∂y

+1

Re∇2v

y = 0 : u = v = u′ = v ′ = 0, y ∼ δ(x ; Re) : u ∼ ue(x), τ ∼ 0

⇒ asymptotic theory faces closure problem for τ

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 9 / 34

Analytically-based methods

• full NS eqs : starting efforts, but still no complete theory existsDeriat & Guiraud (J Mec Theor Appl, 1986), Walker (AIAA J, 1989, CISM, 1998),

Scheichl & Kluwick (JFS, 2008)

• non-dimensional Reynolds-averaged NS eqs(nominally 2D, curvature effects on BL flow of higher order):

∂u∂x

+∂v∂y

= 0−τ

u∂u∂x

+ v∂u∂y

= −∂p∂x

− ∂〈u′2〉∂x

− ∂︷ ︸︸ ︷

〈u′v ′〉∂y

+1

Re∇2u

∇2 =∂2

∂x2 +∂2

∂y2

u∂v∂x

+ v∂v∂y

= −∂p∂y

− ∂〈u′v ′〉∂x

− ∂〈v ′2〉∂y

+1

Re∇2v

y = 0 : u = v = u′ = v ′ = 0, y ∼ δ(x ; Re) : u ∼ ue(x), τ ∼ 0

⇒ asymptotic theory faces closure problem for τ

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 9 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 10 / 34

Classical theory of turbulent small-defect BLs : uτ:=

√τw → 0

Yajnik (JFM, 1970), Bush & Fendell (JFM, 1972), Fendell (J Astronaut Sci, 1972),

Mellor (Int J Eng Sci, 1972), Gersten & Herwig (1992), Walker (CISM, 1998)

uue

O(uτ ) uuτ

∼ 1κ

ln y+ + Ci

{

outer predominately

inviscid region overlap region : y+ := yuτ Re → ∞

viscous wall layer

τw = Re−1(∂u/∂y)y=0

assumptions Walker (CISM, 1998), Scheichl, Kluwick & Alletto (Acta Mech, 2008)

(a) locally isotropic turbulence ⇒ 〈u′2〉, 〈u′v ′〉, 〈v ′2〉 of same magnitude

(b) wall layer : total shear stress essentially unaffected by pressure gradient

(c) direct match with outer fully turbulent region ⇒ two-tiered BL

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 11 / 34

Classical theory of turbulent small-defect BLs : uτ:=

√τw → 0

Yajnik (JFM, 1970), Bush & Fendell (JFM, 1972), Fendell (J Astronaut Sci, 1972),

Mellor (Int J Eng Sci, 1972), Gersten & Herwig (1992), Walker (CISM, 1998)

uue

O(uτ ) uuτ

∼ 1κ

ln y+ + Ci

{

outer predominately

inviscid region overlap region : y+ := yuτ Re → ∞

viscous wall layer

τw = Re−1(∂u/∂y)y=0

assumptions Walker (CISM, 1998), Scheichl, Kluwick & Alletto (Acta Mech, 2008)

(a) locally isotropic turbulence ⇒ 〈u′2〉, 〈u′v ′〉, 〈v ′2〉 of same magnitude

(b) wall layer : total shear stress essentially unaffected by pressure gradient

(c) direct match with outer fully turbulent region ⇒ two-tiered BL

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 11 / 34

Classical theory of turbulent small-defect BLs : uτ:=

√τw → 0

Yajnik (JFM, 1970), Bush & Fendell (JFM, 1972), Fendell (J Astronaut Sci, 1972),

Mellor (Int J Eng Sci, 1972), Gersten & Herwig (1992), Walker (CISM, 1998)

uue

O(uτ ) uuτ

∼ 1κ

ln y+ + Ci

{

outer predominately

inviscid region overlap region : y+ := yuτ Re → ∞

viscous wall layer

τw = Re−1(∂u/∂y)y=0

assumptions Walker (CISM, 1998), Scheichl, Kluwick & Alletto (Acta Mech, 2008)

(a) locally isotropic turbulence ⇒ 〈u′2〉, 〈u′v ′〉, 〈v ′2〉 of same magnitude

(b) wall layer : total shear stress essentially unaffected by pressure gradient

(c) direct match with outer fully turbulent region ⇒ two-tiered BL

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 11 / 34

Classical theory of turbulent small-defect BLs : uτ:=

√τw → 0

Yajnik (JFM, 1970), Bush & Fendell (JFM, 1972), Fendell (J Astronaut Sci, 1972),

Mellor (Int J Eng Sci, 1972), Gersten & Herwig (1992), Walker (CISM, 1998)

uue

O(uτ ) uuτ

∼ 1κ

ln y+ + Ci

{

outer predominately

inviscid region overlap region : y+ := yuτ Re → ∞

viscous wall layer

τw = Re−1(∂u/∂y)y=0

assumptions Walker (CISM, 1998), Scheichl, Kluwick & Alletto (Acta Mech, 2008)

(a) locally isotropic turbulence ⇒ 〈u′2〉, 〈u′v ′〉, 〈v ′2〉 of same magnitude

(b) wall layer : total shear stress essentially unaffected by pressure gradient

(c) direct match with outer fully turbulent region ⇒ two-tiered BL

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 11 / 34

Distinguished limit γ := uτ/ue → 0 , Re → ∞

viscous wall layer : y+ = yuτ Re outer defect layer : η = y/δ

uue

∼ γ(x ,Re) u+(x , y+) + · · ·τ

u2e∼ γ2(x ,Re) t+(x , y+) + · · ·

p ∼ p0(x) + · · ·

uue

∼ 1 − γ∂F1(x , η)

∂η+ O(γ2)

τ

u2e∼ γ2T1(x , η) + O(γ3)

p ∼ pe(x) + O(γ2)

experimental observation / result from first principles (?)Österlund et al. (Phys Fluids, 2000) Walker (AIAA J, 1989, CISM, 1998)

u+(y+) ∼ κ−1 ln y+ + Ci , y+ → ∞ , κ ≈ 0.384 , Ci ≈ 4.1

provides expansion in outer layer and matching with wall layer

η → 0 : ∂F1/∂η ∼ −κ−1 ln η + Co(x) , T1 → 1 ; p0(x) = pe(x)

skin-friction law κ/γ ∼ ln(Re γδue) + κ(Ci + Co) ∼ ln Re

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 12 / 34

Distinguished limit γ := uτ/ue → 0 , Re → ∞

viscous wall layer : y+ = yuτ Re outer defect layer : η = y/δ

uue

∼ γ(x ,Re) u+(x , y+) + · · ·τ

u2e∼ γ2(x ,Re) t+(x , y+) + · · ·

p ∼ p0(x) + · · ·

uue

∼ 1 − γ∂F1(x , η)

∂η+ O(γ2)

τ

u2e∼ γ2T1(x , η) + O(γ3)

p ∼ pe(x) + O(γ2)

experimental observation / result from first principles (?)Österlund et al. (Phys Fluids, 2000) Walker (AIAA J, 1989, CISM, 1998)

u+(y+) ∼ κ−1 ln y+ + Ci , y+ → ∞ , κ ≈ 0.384 , Ci ≈ 4.1

provides expansion in outer layer and matching with wall layer

η → 0 : ∂F1/∂η ∼ −κ−1 ln η + Co(x) , T1 → 1 ; p0(x) = pe(x)

skin-friction law κ/γ ∼ ln(Re γδue) + κ(Ci + Co) ∼ ln Re

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 12 / 34

Substitution into Reynolds-averaged NS eqs

viscous wall layer : du+/dy+ + t+ = 1 . . . constant total shear stress

outer defect layer : δ ∼ γ∆1(x) + O(γ2)

leading-order equation

(E + 2β0)ηF ′

1 − E F1 − ∆1F1,e ∂F1/∂x = F1,e (T1 − 1) , F1,e = F1(x ,1)

E := 1 − ∆1 dF1,e/dx , β0 := −(∆1F1,e due/dx)/ue

layer thickness ratio (Kármán number) δ+ exponentially small

δ+ =(uτ Re)−1

δ∼ 1

Re γ2∆1ue∼ 1

∆1ue

exp(−κ/γ)γ2

consequences

classical theory not capable of describing BL separation Sykes (JFM, 1980)

“manipulate” velocity defect 1 − u/ue = O(ǫ), BL thickness δ

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 13 / 34

Substitution into Reynolds-averaged NS eqs

viscous wall layer : du+/dy+ + t+ = 1 . . . constant total shear stress

outer defect layer : δ ∼ γ∆1(x) + O(γ2)

leading-order equation

(E + 2β0)ηF ′

1 − E F1 − ∆1F1,e ∂F1/∂x = F1,e (T1 − 1) , F1,e = F1(x ,1)

E := 1 − ∆1 dF1,e/dx , β0 := −(∆1F1,e due/dx)/ue

layer thickness ratio (Kármán number) δ+ exponentially small

δ+ =(uτ Re)−1

δ∼ 1

Re γ2∆1ue∼ 1

∆1ue

exp(−κ/γ)γ2

consequences

classical theory not capable of describing BL separation Sykes (JFM, 1980)

“manipulate” velocity defect 1 − u/ue = O(ǫ), BL thickness δ

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 13 / 34

Substitution into Reynolds-averaged NS eqs

viscous wall layer : du+/dy+ + t+ = 1 . . . constant total shear stress

outer defect layer : δ ∼ γ∆1(x) + O(γ2)

leading-order equation

(E + 2β0)ηF ′

1 − E F1 − ∆1F1,e ∂F1/∂x = F1,e (T1 − 1) , F1,e = F1(x ,1)

E := 1 − ∆1 dF1,e/dx , β0 := −(∆1F1,e due/dx)/ue

layer thickness ratio (Kármán number) δ+ exponentially small

δ+ =(uτ Re)−1

δ∼ 1

Re γ2∆1ue∼ 1

∆1ue

exp(−κ/γ)γ2

consequences

classical theory not capable of describing BL separation Sykes (JFM, 1980)

“manipulate” velocity defect 1 − u/ue = O(ǫ), BL thickness δ

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 13 / 34

Substitution into Reynolds-averaged NS eqs

viscous wall layer : du+/dy+ + t+ = 1 . . . constant total shear stress

outer defect layer : δ ∼ γ∆1(x) + O(γ2)

leading-order equation

(E + 2β0)ηF ′

1 − E F1 − ∆1F1,e ∂F1/∂x = F1,e (T1 − 1) , F1,e = F1(x ,1)

E := 1 − ∆1 dF1,e/dx , β0 := −(∆1F1,e due/dx)/ue

layer thickness ratio (Kármán number) δ+ exponentially small

δ+ =(uτ Re)−1

δ∼ 1

Re γ2∆1ue∼ 1

∆1ue

exp(−κ/γ)γ2

consequences

classical theory not capable of describing BL separation Sykes (JFM, 1980)

“manipulate” velocity defect 1 − u/ue = O(ǫ), BL thickness δ

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 13 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 14 / 34

Routes to separation

η

1

uue

O(ǫ)

δ+ =1

Re δ2

log law

defectlayer

viscouswall layer

consider BLs having . . .

• ǫ≫ γ ⇒ ∂p/∂x = O(1) ⇒marginal separation Scheichl & Kluwick (AIAA J, 2007)

• δ+ =(δRe)−1

δ= O(δr ) ⇒ locally ∂p/∂x ≫ 1 ⇒

massive separation Scheichl, Kluwick & Smith

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 15 / 34

Routes to separation

η

1

uue

O(ǫ)

δ+ =1

Re δ2

log law

defectlayer

viscouswall layer

consider BLs having . . .

• ǫ≫ γ ⇒ ∂p/∂x = O(1) ⇒marginal separation Scheichl & Kluwick (AIAA J, 2007)

• δ+ =(δRe)−1

δ= O(δr ) ⇒ locally ∂p/∂x ≫ 1 ⇒

massive separation Scheichl, Kluwick & Smith

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 15 / 34

Routes to separation

η

1

uue

O(ǫ)

δ+ =1

Re δ2

log law

defectlayer

viscouswall layer

consider BLs having . . .

• ǫ≫ γ ⇒ ∂p/∂x = O(1) ⇒marginal separation Scheichl & Kluwick (AIAA J, 2007)

• δ+ =(δRe)−1

δ= O(δr ) ⇒ locally ∂p/∂x ≫ 1 ⇒

massive separation Scheichl, Kluwick & Smith

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 15 / 34

Routes to separation

η

1

uue

O(ǫ)

δ+ =1

Re δ2

log law

defectlayer

viscouswall layer

consider BLs having . . .

• ǫ≫ γ ⇒ ∂p/∂x = O(1) ⇒marginal separation Scheichl & Kluwick (AIAA J, 2007)

• δ+ =(δRe)−1

δ= O(δr ) ⇒ locally ∂p/∂x ≫ 1 ⇒

massive separation Scheichl, Kluwick & Smith

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 15 / 34

Problem formulation – motivationLocal description of turbulent break-away separation . . .

. . . requires answers to three basic questions:

(1) Global topology of flow for ‘vanishing viscosity’?

(2) Characteristics of incident boundary layer flow?

(3) How do these issues interdepend?

basic assumptions

• flow incompressible,nominally steady, 2D

• free-stream turbulencedisregarded

• Re := UL/ν → ∞

canonical example: circular-cylinder flow

How turbulent are the BL and the separated SL ?Scheichl, Kluwick & Alletto (Acta Mech, 2008, AIAA J)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 16 / 34

Problem formulation – motivationLocal description of turbulent break-away separation . . .

. . . requires answers to three basic questions:

(1) Global topology of flow for ‘vanishing viscosity’?

(2) Characteristics of incident boundary layer flow?

(3) How do these issues interdepend?

basic assumptions

• flow incompressible,nominally steady, 2D

• free-stream turbulencedisregarded

• Re := UL/ν → ∞

canonical example: circular-cylinder flow

How turbulent are the BL and the separated SL ?Scheichl, Kluwick & Alletto (Acta Mech, 2008, AIAA J)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 16 / 34

Problem formulation – motivationLocal description of turbulent break-away separation . . .

. . . requires answers to three basic questions:

(1) Global topology of flow for ‘vanishing viscosity’?

(2) Characteristics of incident boundary layer flow?

(3) How do these issues interdepend?

basic assumptions

• flow incompressible,nominally steady, 2D

• free-stream turbulencedisregarded

• Re := UL/ν → ∞

canonical example: circular-cylinder flow

How turbulent are the BL and the separated SL ?Scheichl, Kluwick & Alletto (Acta Mech, 2008, AIAA J)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 16 / 34

Problem formulation – motivationLocal description of turbulent break-away separation . . .

. . . requires answers to three basic questions:

(1) Global topology of flow for ‘vanishing viscosity’?

(2) Characteristics of incident boundary layer flow?

(3) How do these issues interdepend?

basic assumptions

• flow incompressible,nominally steady, 2D

• free-stream turbulencedisregarded

• Re := UL/ν → ∞

canonical example: circular-cylinder flow

How turbulent are the BL and the separated SL ?Scheichl, Kluwick & Alletto (Acta Mech, 2008, AIAA J)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 16 / 34

Problem formulation – motivationLocal description of turbulent break-away separation . . .

. . . requires answers to three basic questions:

(1) Global topology of flow for ‘vanishing viscosity’?

(2) Characteristics of incident boundary layer flow?

(3) How do these issues interdepend?

basic assumptions

• flow incompressible,nominally steady, 2D

• free-stream turbulencedisregarded

• Re := UL/ν → ∞

canonical example: circular-cylinder flow

u∞ = 1L = 1

F R

S

nn ∼ δ

TBL TSL

How turbulent are the BL and the separated SL ?Scheichl, Kluwick & Alletto (Acta Mech, 2008, AIAA J)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 16 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 17 / 34

Global flow picture

Re ' 3 × 106 : postcritical regime, Re → ∞ : transition point approximates F

Re−1 = 0 : ultimate or T-state of flow, transition in F

u∞ = 1

F R ∼ S ?

SθS ≈ 115◦ uS . u∞

pS . p∞

transition

TBL TSL

Neish & Smith (JFM, 1992)

unlikely Scheichl & Kluwick (JFS, 2008)

corroborated experimentally up to Re ≈ 4 × 107

Roshko (JFM, 1961), Achenbach (JFM, 1968, Int J Heat Mass Trans, 1975), Zdravkovich (1997),

Schewe (J Wind Eng Ind Aerod, 2001), Gölling (2006)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 18 / 34

Global flow picture

Re ' 3 × 106 : postcritical regime, Re → ∞ : transition point approximates F

Re−1 = 0 : ultimate or T-state of flow, transition in F

u∞ = 1

F R ∼ S ?

SθS ≈ 115◦ uS . u∞

pS . p∞

transition

TBL TSL

Neish & Smith (JFM, 1992)

unlikely Scheichl & Kluwick (JFS, 2008)

corroborated experimentally up to Re ≈ 4 × 107

Roshko (JFM, 1961), Achenbach (JFM, 1968, Int J Heat Mass Trans, 1975), Zdravkovich (1997),

Schewe (J Wind Eng Ind Aerod, 2001), Gölling (2006)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 18 / 34

Global flow picture

Re ' 3 × 106 : postcritical regime, Re → ∞ : transition point approximates F

Re−1 = 0 : ultimate or T-state of flow, transition in F

u∞ = 1

F R ∼ S ?

SθS ≈ 115◦ uS . u∞

pS . p∞

transition

TBL TSL

Neish & Smith (JFM, 1992)

unlikely Scheichl & Kluwick (JFS, 2008)

corroborated experimentally up to Re ≈ 4 × 107

Roshko (JFM, 1961), Achenbach (JFM, 1968, Int J Heat Mass Trans, 1975), Zdravkovich (1997),

Schewe (J Wind Eng Ind Aerod, 2001), Gölling (2006)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 18 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 19 / 34

SeparationIs there really a fully developed TBL as Re → ∞ ?

oil-flow measurements : ReC := UC/ν , C = airfoil chord length

sub- trans- super- post-

by courtesy of G. Schewe (Göttingen, 2001)

-critical

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 20 / 34

SeparationIs there really a fully developed TBL as Re → ∞ ?

oil-flow measurements : ReC := UC/ν , C = airfoil chord length

C

W ≈ 6 × C

ReC = 7.4 × 105 :

ReC = 7.7 × 106 :

cD ≈ 0.11,

cD ≈ 0.14,

cL ≈ 1.1

cL ≈ 0.65

3.5 2.5 7.5

cD

cL

ReC

sub- trans- super- post-

by courtesy of G. Schewe (Göttingen, 2001)

-critical

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 20 / 34

U

S

T

T

A

A

Re

laminar

turbulent

Prandtl (Göttingen, 1914)

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 21 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 22 / 34

‘Ideal-fluid limit’Hierarchy of external flow – shear layer: potential flow approached as Re → ∞Oseen (1927), Lamb (1932), Batchelor (1956), Prandtl (1961), Sychev et al. (1998)

Re−1 = 0 : external potential flows

Imai (J Phys Soc Japan, 1953),

Birkhoff & Zarantonello (1957), Gurevich (1966)

• free streamlines confine (open) dead-water region

• class of flows / θS controlled by

Brillouin–Villat parameter k ≥ 0 , free-stream velocity uS ≤ u∞

ue

∼ b(k)θ + O(θ2) , θ → 0+

∼ uS[1 + 2k(−s)1/2 + O(−s)

], s → 0−

≡ uS , s ≥ 0

κ

{

= κB(s) , s < 0

∼ −k s−1/2 + κB(0) + O(s1/2) , s → 0+

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 23 / 34

‘Ideal-fluid limit’Hierarchy of external flow – shear layer: potential flow approached as Re → ∞Oseen (1927), Lamb (1932), Batchelor (1956), Prandtl (1961), Sychev et al. (1998)

Re−1 = 0 : external potential flows

Imai (J Phys Soc Japan, 1953),

Birkhoff & Zarantonello (1957), Gurevich (1966) 1u∞ = 1 F R

S n

s

ue, κB

uS , κ

θ θS

• free streamlines confine (open) dead-water region

• class of flows / θS controlled by

Brillouin–Villat parameter k ≥ 0 , free-stream velocity uS ≤ u∞

ue

∼ b(k)θ + O(θ2) , θ → 0+

∼ uS[1 + 2k(−s)1/2 + O(−s)

], s → 0−

≡ uS , s ≥ 0

κ

{

= κB(s) , s < 0

∼ −k s−1/2 + κB(0) + O(s1/2) , s → 0+

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 23 / 34

‘Ideal-fluid limit’Hierarchy of external flow – shear layer: potential flow approached as Re → ∞Oseen (1927), Lamb (1932), Batchelor (1956), Prandtl (1961), Sychev et al. (1998)

Re−1 = 0 : external potential flows

Imai (J Phys Soc Japan, 1953),

Birkhoff & Zarantonello (1957), Gurevich (1966) 1u∞ = 1 F R

S n

s

ue, κB

uS , κ

θ θS

• free streamlines confine (open) dead-water region

• class of flows / θS controlled by

Brillouin–Villat parameter k ≥ 0 , free-stream velocity uS ≤ u∞

ue

∼ b(k)θ + O(θ2) , θ → 0+

∼ uS[1 + 2k(−s)1/2 + O(−s)

], s → 0−

≡ uS , s ≥ 0

κ

{

= κB(s) , s < 0

∼ −k s−1/2 + κB(0) + O(s1/2) , s → 0+

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 23 / 34

Kirchhoff-type flows – numerical solutionsOpen cavity: uS ≡ u∞ = 1

surface velocity ue(θ, k)

k = 0, 0.05, 0.1, . . . ,0.5

55◦ 2′ 30′′ ≤ θS / 126◦

k = 0 :

Brillouin–Villat condition

0

2

1.5

1

0.5

030 6055 12012690≈ θ [◦]

k

k = 0

k = 0.5

free streamlines

126◦ / θS ≤ 180◦

0.5 / k < ∞?

1 ≥ uS ≥ 0

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 24 / 34

Kirchhoff-type flows – numerical solutionsOpen cavity: uS ≡ u∞ = 1

surface velocity ue(θ, k)

k = 0, 0.05, 0.1, . . . ,0.5

55◦ 2′ 30′′ ≤ θS / 126◦

k = 0 :

Brillouin–Villat condition

0

2

1.5

1

0.5

030 6055 12012690≈ θ [◦]

k

k = 0

k = 0.5

free streamlines

126◦ / θS ≤ 180◦

0.5 / k < ∞?

1 ≥ uS ≥ 0

2.5 2 1.5 1 0.5 0

1

2.5 2

1.5

0.5 0−1 2 3 1 4 5 6 7 0

kk = 0

k = 0.5

S

R?

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 24 / 34

Question: overall flow structure as Re → ∞

125

105

55

95

85

75

65

0 0.45 .05 .1 .15 .2 .25 .3 .35 .4 .5

115

θS

k ?

(1) Global topology of flow for ‘vanishing viscosity’?

(2) Characteristics of incident boundary layer flow?

(3) How do these issues interdepend?

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 25 / 34

Question: overall flow structure as Re → ∞

125

105

55

95

85

75

65

0 0.45 .05 .1 .15 .2 .25 .3 .35 .4 .5

115

θS

k ?

(1) Global topology of flow for ‘vanishing viscosity’?

(2) Characteristics of incident boundary layer flow?

(3) How do these issues interdepend?

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 25 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 26 / 34

Asymptotic framework – turbulence intensity levelNeish & Smith (1992), Scheichl, Kluwick & Alletto (Acta Mech, 2008, AIAA J)

Prandtl-type BL: k , T , Re ≫ 1

N = Re1/2n{

Re1/2[ψ, −〈u′v ′〉

], p

}

∼{[

Ψ,T r](θ,N; k ,T ), pF −u2

e(θ; k)/2}

+ O(Re−1/2

)

∂NΨ ∂NθΨ − ∂θΨ ∂NNΨ = ue ∂θue + ∂N [T r + ∂NNΨ], 0 ≤ T <∞N → 0 : Ψ → 0, ΨN → 0, r = O(N3), N → ∞ : ΨN → ue, r → 0

solution as s = θ − θS → 0−

T ≫ 1 :

Goldstein singularity at

s = sG = O[k6T−8(ln T )16

]

⇒ interaction length scale δTD

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 27 / 34

Asymptotic framework – turbulence intensity levelNeish & Smith (1992), Scheichl, Kluwick & Alletto (Acta Mech, 2008, AIAA J)

Prandtl-type BL: k , T , Re ≫ 1

N = Re1/2n{

Re1/2[ψ, −〈u′v ′〉

], p

}

∼{[

Ψ,T r](θ,N; k ,T ), pF −u2

e(θ; k)/2}

+ O(Re−1/2

)

∂NΨ ∂NθΨ − ∂θΨ ∂NNΨ = ue ∂θue + ∂N [T r + ∂NNΨ], 0 ≤ T <∞N → 0 : Ψ → 0, ΨN → 0, r = O(N3), N → ∞ : ΨN → ue, r → 0

solution as s = θ − θS → 0−

T ≫ 1 :

Goldstein singularity at

s = sG = O[k6T−8(ln T )16

]

⇒ interaction length scale δTD

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 27 / 34

Asymptotic framework – turbulence intensity levelNeish & Smith (1992), Scheichl, Kluwick & Alletto (Acta Mech, 2008, AIAA J)

Prandtl-type BL: k , T , Re ≫ 1

N = Re1/2n{

Re1/2[ψ, −〈u′v ′〉

], p

}

∼{[

Ψ,T r](θ,N; k ,T ), pF −u2

e(θ; k)/2}

+ O(Re−1/2

)

∂NΨ ∂NθΨ − ∂θΨ ∂NNΨ = ue ∂θue + ∂N [T r + ∂NNΨ], 0 ≤ T <∞N → 0 : Ψ → 0, ΨN → 0, r = O(N3), N → ∞ : ΨN → ue, r → 0

solution as s = θ − θS → 0−

T ≫ 1 :

Goldstein singularity at

s = sG = O[k6T−8(ln T )16

]

⇒ interaction length scale δTD

−s

τw = ∂NNΨ(θ,0; k ,T )

−sG

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 27 / 34

Concept of ‘underdeveloped’ TBLBL originating in F parametrised by T ≫ 1

small defect: ue − u = O(ǫ), r = O(ǫ2)

BL eq: −〈u′v ′〉 = Re−1/2T r = O(σǫ), δ = O(σ)

}

⇒ T = Re1/2σ/ǫ

fully developed TBL: −〈u′v ′〉 = O(ǫ2), σ ∝ ǫ ∼ κ/ ln Re

small-defect layer: σ ≪ 1, ǫ = κ/(σ2 ln Re) ≪ 1{[uen − ψ

ue δǫ,−〈u′v ′〉

u2e δǫ

]

σ

}

∼{

[F ,Σ](θ, η; k), ∆(θ; k)}

+ O(ǫ), η =nδ

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 28 / 34

Concept of ‘underdeveloped’ TBLBL originating in F parametrised by T ≫ 1

small defect: ue − u = O(ǫ), r = O(ǫ2)

BL eq: −〈u′v ′〉 = Re−1/2T r = O(σǫ), δ = O(σ)

}

⇒ T = Re1/2σ/ǫ

fully developed TBL: −〈u′v ′〉 = O(ǫ2), σ ∝ ǫ ∼ κ/ ln Re

small-defect layer: σ ≪ 1, ǫ = κ/(σ2 ln Re) ≪ 1{[uen − ψ

ue δǫ,−〈u′v ′〉

u2e δǫ

]

σ

}

∼{

[F ,Σ](θ, η; k), ∆(θ; k)}

+ O(ǫ), η =nδ

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 28 / 34

Concept of ‘underdeveloped’ TBLBL originating in F parametrised by T ≫ 1

small defect: ue − u = O(ǫ), r = O(ǫ2)

BL eq: −〈u′v ′〉 = Re−1/2T r = O(σǫ), δ = O(σ)

}

⇒ T = Re1/2σ/ǫ

fully developed TBL: −〈u′v ′〉 = O(ǫ2), σ ∝ ǫ ∼ κ/ ln Re

small-defect layer: σ ≪ 1, ǫ = κ/(σ2 ln Re) ≪ 1{[uen − ψ

ue δǫ,−〈u′v ′〉

u2e δǫ

]

σ

}

∼{

[F ,Σ](θ, η; k), ∆(θ; k)}

+ O(ǫ), η =nδ

η

1

uue

O(ǫ)1

Re δ2

log law

defectlayer

viscouswall layer

η

sS

δ ≪ ǫδTD

δTD

TD

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 28 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 29 / 34

Solutions of outer-defect eqs for 0 ≤ θ < θS.= 113.5◦ k = 0.45

u2e ∂θ(ue∆)ηF ′−∂θ(u3

e∆F ) = u3e(Σ−1), Σ = l2F ′′|F ′′|, F (θ, η; k), ∆(θ; k)

η → 0 : F ′ ∼ −κ−1 ln η + B(θ) ⇔ Σ ∼ 1, η = 1 : F ′ = F ′′ = Σ = 0

θ → 0 : F ∼ F0(η), ∆ ∼ ∆0 θ, ue ∼ b(k) θ stagnant-flow

• algebraic closure for ℓ = δl , Klebanoff’s intermittency factorMichel, Quémard & Durant (1969), Klebanoff (1955)

• Keller–Box scheme / method of lines, adaptive grid remeshing

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 30 / 34

Solutions of outer-defect eqs for 0 ≤ θ < θS.= 113.5◦ k = 0.45

u2e ∂θ(ue∆)ηF ′−∂θ(u3

e∆F ) = u3e(Σ−1), Σ = l2F ′′|F ′′|, F (θ, η; k), ∆(θ; k)

η → 0 : F ′ ∼ −κ−1 ln η + B(θ) ⇔ Σ ∼ 1, η = 1 : F ′ = F ′′ = Σ = 0

θ → 0 : F ∼ F0(η), ∆ ∼ ∆0 θ, ue ∼ b(k) θ stagnant-flow

• algebraic closure for ℓ = δl , Klebanoff’s intermittency factorMichel, Quémard & Durant (1969), Klebanoff (1955)

• Keller–Box scheme / method of lines, adaptive grid remeshing

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 30 / 34

Solutions of outer-defect eqs for 0 ≤ θ < θS.= 113.5◦ k = 0.45

u2e ∂θ(ue∆)ηF ′−∂θ(u3

e∆F ) = u3e(Σ−1), Σ = l2F ′′|F ′′|, F (θ, η; k), ∆(θ; k)

η → 0 : F ′ ∼ −κ−1 ln η + B(θ) ⇔ Σ ∼ 1, η = 1 : F ′ = F ′′ = Σ = 0

θ → 0 : F ∼ F0(η), ∆ ∼ ∆0 θ, ue ∼ b(k) θ stagnant-flow

• algebraic closure for ℓ = δl , Klebanoff’s intermittency factorMichel, Quémard & Durant (1969), Klebanoff (1955)

• Keller–Box scheme / method of lines, adaptive grid remeshing

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 30 / 34

Solutions of outer-defect eqs for 0 ≤ θ < θS.= 113.5◦ k = 0.45

u2e ∂θ(ue∆)ηF ′−∂θ(u3

e∆F ) = u3e(Σ−1), Σ = l2F ′′|F ′′|, F (θ, η; k), ∆(θ; k)

η → 0 : F ′ ∼ −κ−1 ln η + B(θ) ⇔ Σ ∼ 1, η = 1 : F ′ = F ′′ = Σ = 0

θ → 0 : F ∼ F0(η), ∆ ∼ ∆0 θ, ue ∼ b(k) θ stagnant-flow

• algebraic closure for ℓ = δl , Klebanoff’s intermittency factorMichel, Quémard & Durant (1969), Klebanoff (1955)

• Keller–Box scheme / method of lines, adaptive grid remeshing

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

0 0 5 10 15 20 25 30

η

F ′

i

θ = θi = i θS/10i = 0, 1, . . . , 10

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

0 0 0.5 1 1.5 2 2.5 3 3.5

η

Σ

i

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 30 / 34

Solutions of outer-defect eqs, cont’d k = 0.45Comparison with leading-order theory as s = θ − θS → 0−

1

0

1.5

0.5

−0.5

−1

−1.5

0 90 80 50 40 30 20 10 100 113

0.5

0.4

0.3

0.2

0.1

0

−0.1

−0.2 70

ue

cp = 1−u2e

∆ ∼ ∆0 θ

θ

[F , ∆

]∼

[FS(η; k), ∆S(k)

]− ∑

i=1

[fi(s, k)Fi(η; k), gi(s, k)∆i(k)

]

[F , ∆

]∼

[FS(1 − 2g1), ∆S(1 − g1)

]+ O(s), g1 = 2k(−s)1/2

F ′

i = O[ln(η)], η → 0, i = 1,2, . . . ⇒ sublayer for η = O(−s)

indicates onset of viscous/inviscid interaction

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 31 / 34

Solutions of outer-defect eqs, cont’d k = 0.45Comparison with leading-order theory as s = θ − θS → 0−

1

0

1.5

0.5

−0.5

−1

−1.5

0 90 80 50 40 30 20 10 100 113

0.5

0.4

0.3

0.2

0.1

0

−0.1

−0.2 70

ue

cp = 1−u2e

∆ ∼ ∆0 θ

θ

0.1

0.01

0.001 0.001 0.1 0.01 0.25

∆−∆S

− lg ∆s

∆S.= 0.4464

g1

[F , ∆

]∼

[FS(η; k), ∆S(k)

]−

∑∞

i=1

[fi(s, k)Fi(η; k), gi(s, k)∆i(k)

]

[F , ∆

]∼

[FS(1 − 2g1), ∆S(1 − g1)

]+ O(s), g1 = 2k(−s)1/2

F ′

i = O[ln(η)], η → 0, i = 1,2, . . . ⇒ sublayer for η = O(−s)

indicates onset of viscous/inviscid interaction

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 31 / 34

Solutions of outer-defect eqs, cont’d k = 0.45Comparison with leading-order theory as s = θ − θS → 0−

1

0

1.5

0.5

−0.5

−1

−1.5

0 90 80 50 40 30 20 10 100 113

0.5

0.4

0.3

0.2

0.1

0

−0.1

−0.2 70

ue

cp = 1−u2e

∆ ∼ ∆0 θ

θ

0.1

0.01

0.001 0.001 0.1 0.01 0.25

∆−∆S

− lg ∆s

∆S.= 0.4464

g1

[F , ∆

]∼

[FS(η; k), ∆S(k)

]−

∑∞

i=1

[fi(s, k)Fi(η; k), gi(s, k)∆i(k)

]

[F , ∆

]∼

[FS(1 − 2g1), ∆S(1 − g1)

]+ O(s), g1 = 2k(−s)1/2

F ′

i = O[ln(η)], η → 0, i = 1,2, . . . ⇒ sublayer for η = O(−s)

indicates onset of viscous/inviscid interaction

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 31 / 34

Outline

1 Turbulent wall-bounded flowsNumerically-based methodsAnalytically-based methods

2 Classical theory of turbulent small-defect BLsGeneral frameworkRoutes to separation

3 Problem formulation – motivationGlobal flow pictureSeparation – experimental findings

4 Asymptotic structure of the flow‘Ideal-fluid limit’ – Kirchhoff-type potential flowBoundary layer

5 Incident boundary layer – numerical results

6 Preliminary conclusions – outlook

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 32 / 34

status quo of research

• increasing turbulence intensity shifts separation downstream, increases k

• BL never attains fully developed turbulent state

interactive BL ⇒ scaling in terms of k , Re ≫ 1

ǫ ∼ κ/ ln Re, δTD = k8/9Re−4/9, δ = O(δTD/ǫ), −〈u′

i u′

j 〉 = O(ǫδ)

ongoing research

• TD problem (akin to laminar counterpart)

• self-induced separation ⇔ (maximum) values of k , θS(k) ?

• structure of (large-scale) separated flow

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 33 / 34

status quo of research

• increasing turbulence intensity shifts separation downstream, increases k

• BL never attains fully developed turbulent state

interactive BL ⇒ scaling in terms of k , Re ≫ 1

ǫ ∼ κ/ ln Re, δTD = k8/9Re−4/9, δ = O(δTD/ǫ), −〈u′

i u′

j 〉 = O(ǫδ)

ongoing research

• TD problem (akin to laminar counterpart)

• self-induced separation ⇔ (maximum) values of k , θS(k) ?

• structure of (large-scale) separated flow

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 33 / 34

status quo of research

• increasing turbulence intensity shifts separation downstream, increases k

• BL never attains fully developed turbulent state

interactive BL ⇒ scaling in terms of k , Re ≫ 1

ǫ ∼ κ/ ln Re, δTD = k8/9Re−4/9, δ = O(δTD/ǫ), −〈u′

i u′

j 〉 = O(ǫδ)

ongoing research

• TD problem (akin to laminar counterpart)

• self-induced separation ⇔ (maximum) values of k , θS(k) ?

• structure of (large-scale) separated flow

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 33 / 34

status quo of research

• increasing turbulence intensity shifts separation downstream, increases k

• BL never attains fully developed turbulent state

interactive BL ⇒ scaling in terms of k , Re ≫ 1

ǫ ∼ κ/ ln Re, δTD = k8/9Re−4/9, δ = O(δTD/ǫ), −〈u′

i u′

j 〉 = O(ǫδ)

ongoing research

• TD problem (akin to laminar counterpart)

• self-induced separation ⇔ (maximum) values of k , θS(k) ?

• structure of (large-scale) separated flow

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 33 / 34

Thank you for your attention !

B. Scheichl (TU Vienna) Turbulent Bluff-Body Separation 34 / 34