![Page 1: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/1.jpg)
Wall-Bounded Shear Flow TransitionBeyond Classical Stability Theory
Bassam Bamieh
MECHANICAL ENGINEERINGUNIVERSITY OF CALIFORNIA AT SANTA BARBARA
UCSB, Nov 2013 1 / 18
![Page 2: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/2.jpg)
Transition & Turbulence as Natural Phenomena
All fluid flows transition (as 0 R−→∞) from laminar to turbulent flows
Bluff bodies dominant phenomenon: separation
Re = 0.16Re = 13.1 Re = 2,000
Re = 10,000
Streamlined bodies dominant phenomenon: friction with walls
wall-bounded shear flows
UCSB, Nov 2013 2 / 18
![Page 3: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/3.jpg)
Boundary Layer Turbulence
Flow direction
boundary layer turbulence side view top view
Laminar boundary layer Turbulent boundary layer
MeanFlow Pro!le
skin-friction drag: laminar vs. turbulent
Transition & Turbulence in Boundary Layer and Channel Flows
Technologically Important: Skin-Friction Drag
UCSB, Nov 2013 3 / 18
![Page 4: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/4.jpg)
Boundary Layer Turbulence
Flow direction
boundary layer turbulence side view top view
Laminar boundary layer Turbulent boundary layer
MeanFlow Pro!le
skin-friction drag: laminar vs. turbulent
Transition & Turbulence in Boundary Layer and Channel FlowsTechnologically Important: Skin-Friction Drag
UCSB, Nov 2013 3 / 18
![Page 5: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/5.jpg)
Control of Boundary Layer Turbulence
in nature: passive control active control withsensor/actuator arrays
︷ ︸︸ ︷
flow direction
rigid base
flexible membraneflow
corrugated skin compliant skin
Intuition: must have ability to actuate at spatial scale comparable to flow structures
spatial-bandwidth of controller ≥ plant’s bandwidth
Caveat: Plant’s dynamics are not well understood :(obstacles
{not only device technologyalso: dynamical modeling and control design
UCSB, Nov 2013 4 / 18
![Page 6: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/6.jpg)
Control of Boundary Layer Turbulence
in nature: passive control active control withsensor/actuator arrays
︷ ︸︸ ︷
flow direction
rigid base
flexible membraneflow
corrugated skin compliant skin
Intuition: must have ability to actuate at spatial scale comparable to flow structures
spatial-bandwidth of controller ≥ plant’s bandwidth
Caveat: Plant’s dynamics are not well understood :(obstacles
{not only device technologyalso: dynamical modeling and control design
UCSB, Nov 2013 4 / 18
![Page 7: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/7.jpg)
Control of Boundary Layer Turbulence
in nature: passive control active control withsensor/actuator arrays
︷ ︸︸ ︷
flow direction
rigid base
flexible membraneflow
corrugated skin compliant skin
Intuition: must have ability to actuate at spatial scale comparable to flow structures
spatial-bandwidth of controller ≥ plant’s bandwidth
Caveat: Plant’s dynamics are not well understood :(obstacles
{not only device technologyalso: dynamical modeling and control design
UCSB, Nov 2013 4 / 18
![Page 8: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/8.jpg)
Mathematical Modeling of Transition: Hydrodynamic Stability
The Navier-Stokes (NS) equations:
∂tu = −∇uu− grad p + 1R∆u
0 = div u
x
y
z
uv
w
Hydrodynamic Stability: view NS as a dynamical system
laminar flow uR := a stationary solution of the NS equations (an equilibrium)
UCSB, Nov 2013 5 / 18
![Page 9: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/9.jpg)
Mathematical Modeling of Transition: Hydrodynamic Stability
The Navier-Stokes (NS) equations:
∂tu = −∇uu− grad p + 1R∆u
0 = div u
x
y
z
uv
w
Hydrodynamic Stability: view NS as a dynamical system
laminar flow uR := a stationary solution of the NS equations (an equilibrium)
laminar flow uR stable ←→ i.c. u(0) 6= uR,
u(t) t→∞−→ uR
I typically done with dynamics linearized about uR
I various methods to track further “non-linear behavior”
u(0)u(t)
uR
UCSB, Nov 2013 5 / 18
![Page 10: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/10.jpg)
Mathematical Modeling of Transition: Hydrodynamic Stability
The Navier-Stokes (NS) equations:
∂tu = −∇uu− grad p + 1R∆u
0 = div u
x
y
z
uv
w
Hydrodynamic Stability: view NS as a dynamical system
A very successful (phenomenologically predictive) approach for many decades
UCSB, Nov 2013 5 / 18
![Page 11: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/11.jpg)
Mathematical Modeling of Transition: Hydrodynamic Stability
The Navier-Stokes (NS) equations:
∂tu = −∇uu− grad p + 1R∆u
0 = div u
x
y
z
uv
w
Hydrodynamic Stability: view NS as a dynamical systemhowever: problematic for wall-bounded shear flows
Flow type Classical linear theory Rc Experimental Rc
Channel Flow 5772 ≈ 1,000-2,000Plane Couette ∞ ≈ 350Pipe Flow ∞ ≈ 2,200-100,000
UCSB, Nov 2013 5 / 18
![Page 12: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/12.jpg)
Mathematical Modeling of Transition: Hydrodynamic Stability
The Navier-Stokes (NS) equations:
∂tu = −∇uu− grad p + 1R∆u
0 = div u
x
y
z
uv
w
Hydrodynamic Stability: view NS as a dynamical systemhowever: problematic for wall-bounded shear flows
Flow type Classical linear theory Rc Experimental Rc
Channel Flow 5772 ≈ 1,000-2,000Plane Couette ∞ ≈ 350Pipe Flow ∞ ≈ 2,200-100,000
I was widely believed: this theory fails because it is linearand “nonlinear effects” are important even for infinitesimal i.c.
I however, since 90’s: story is actually more interesting than thatNonmodal Stability Theory, Schmid, ARFM ’07
UCSB, Nov 2013 5 / 18
![Page 13: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/13.jpg)
Mathematical Modeling of Transition: Linearized Stability
Decompose the fields as u = uR + u↑ ↑
laminar fluctuationsFluctuation dynamics: In linear hydrodynamic stability, − ∇uu is ignored
∂tu = −∇uR u −∇uuR − grad p + 1R∆u − ∇uu
0 = div u
Linearization in “Evolution Form”
∂t
[∆vω
]=
[U′′∂x − U∆∂x + 1
R ∆2 0−U′∂z −U∂x + 1
R ∆
] [vω
]v := wall-normal velocity
ω := wall-normal vorticity
=:
[L 0C S
] [vω
]=: A
[vω
]
Classical Linear Hydrodynamic Stability:
Transition ←→ Instability ⇔ A has spectrum in right half plane
The existence of “exponentially growing normal modes” (of etA)
UCSB, Nov 2013 6 / 18
![Page 14: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/14.jpg)
Mathematical Modeling of Transition: Linearized Stability
Decompose the fields as u = uR + u↑ ↑
laminar fluctuationsFluctuation dynamics: In linear hydrodynamic stability, − ∇uu is ignored
∂tu = −∇uR u −∇uuR − grad p + 1R∆u − ∇uu
0 = div u
Linearization in “Evolution Form”
∂t
[∆vω
]=
[U′′∂x − U∆∂x + 1
R ∆2 0−U′∂z −U∂x + 1
R ∆
] [vω
]v := wall-normal velocity
ω := wall-normal vorticity
=:
[L 0C S
] [vω
]=: A
[vω
]
Classical Linear Hydrodynamic Stability:
Transition ←→ Instability ⇔ A has spectrum in right half plane
The existence of “exponentially growing normal modes” (of etA)
UCSB, Nov 2013 6 / 18
![Page 15: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/15.jpg)
Mathematical Modeling of Transition: Linearized Stability
Decompose the fields as u = uR + u↑ ↑
laminar fluctuationsFluctuation dynamics: In linear hydrodynamic stability, − ∇uu is ignored
∂tu = −∇uR u −∇uuR − grad p + 1R∆u − ∇uu
0 = div u
Linearization in “Evolution Form”
∂t
[∆vω
]=
[U′′∂x − U∆∂x + 1
R ∆2 0−U′∂z −U∂x + 1
R ∆
] [vω
]v := wall-normal velocity
ω := wall-normal vorticity
=:
[L 0C S
] [vω
]=: A
[vω
]
Classical Linear Hydrodynamic Stability:
Transition ←→ Instability ⇔ A has spectrum in right half plane
The existence of “exponentially growing normal modes” (of etA)
UCSB, Nov 2013 6 / 18
![Page 16: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/16.jpg)
The Eigenvalue Problem
For parallel channel flowsA is translation invariant in x, z, ⇒ Fourier transform in x and z:
∂
∂t
[vω
]=
[ikx∆−1U′′ − ikx∆−1U∆ + 1
R ∆−1∆2 0−ikzU′ −ikxU + 1
R ∆
] [vω
]
kx, kz: spatial frequencies in x, z directions (wave-numbers).
∂
∂t
[v(t, kx, ., kz)ω(t, kx, ., kz)
]= A(kx, kz)
[v(t, kx, ., kz)ω(t, kx, ., kz)
]
Essentially: spectrum (A) =⋃kx,kz
spectrum(A(kx, kz)
)
UCSB, Nov 2013 7 / 18
![Page 17: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/17.jpg)
Tollmien-Schlichting Instability
Poiseuille flow at R = 6000, kx = 1, kz = 0Typical stability regions in K, R space: (for
Poiseuille and Blasius boundary layer flows)
Unstable eigenvalue corresponds to a slowly growing traveling wave:the Tollmien-Schlichting wave
First seen in experiments bySkramstad & Schubauer, 1940
UCSB, Nov 2013 8 / 18
![Page 18: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/18.jpg)
Tollmien-Schlichting Instability
Poiseuille flow at R = 6000, kx = 1, kz = 0Typical stability regions in K, R space: (for
Poiseuille and Blasius boundary layer flows)
Unstable eigenvalue corresponds to a slowly growing traveling wave:the Tollmien-Schlichting wave
First seen in experiments bySkramstad & Schubauer, 1940
UCSB, Nov 2013 8 / 18
![Page 19: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/19.jpg)
A Toy Example
[ψ1
ψ2
]=
[−1 0k −2
] [ψ1ψ2
]D
raft
M. JOVANOVIC 8
A toy example 1
2
�=
�1 0
k �2
� 1
2
�
1(t)
,
2(t)
1
2
v1v2
1/k
UCSB, Nov 2013 9 / 18
![Page 20: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/20.jpg)
A Toy Example
[ψ1
ψ2
]=
[−1 0k −2
] [ψ1ψ2
]D
raft
M. JOVANOVIC 8
A toy example 1
2
�=
�1 0
k �2
� 1
2
�
1(t)
,
2(t)
1
2
e�2t
e�t
UCSB, Nov 2013 9 / 18
![Page 21: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/21.jpg)
Mathematical Modeling of Transition: Adding Signal Uncertainty
Decompose the fields as u = uR + u↑ ↑
laminar fluctuations
Fluctuation dynamics: In linear hydrodynamic stability, − ∇uu is ignored
∂tu = −∇uR u −∇uuR − grad p + 1R∆u − ∇uu + d
0 = div u
I a time-varying exogenous disturbance field d
NSR
(spatio-temporal system)
d u
Input-Output view of the Linearized NS EquationsJovanovic, BB, ’05 JFM
UCSB, Nov 2013 10 / 18
![Page 22: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/22.jpg)
Input-Output Analysis of the Linearized NS Equations
∂t
[∆vω
]=
[U′′∂x − U∆∂x + 1
R ∆2 0−U′∂z −U∂x + 1
R ∆
] [vω
]+
[−∂xy ∂2
x + ∂2z −∂zy
∂z 0 −∂x
][ dxdydz
] u
vw
=(∂2
x + ∂2z
)−1[
∂xy −∂z∂2
x + ∂2z 0
∂zy ∂x
][vω
]
x
y
z
uv
wNSR
(spatio-temporal system)
d u
∂tΨ = A Ψ + B du = C Ψ
UCSB, Nov 2013 11 / 18
![Page 23: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/23.jpg)
Input-Output Analysis of the Linearized NS Equations
∂t
[∆vω
]=
[U′′∂x − U∆∂x + 1
R ∆2 0−U′∂z −U∂x + 1
R ∆
] [vω
]+
[−∂xy ∂2
x + ∂2z −∂zy
∂z 0 −∂x
][ dxdydz
] u
vw
=(∂2
x + ∂2z
)−1[
∂xy −∂z∂2
x + ∂2z 0
∂zy ∂x
][vω
]
x
y
z
uv
wNSR
(spatio-temporal system)
d u
∂tΨ = A Ψ + B du = C Ψ
eigs (A): determine stability
System norm d −→ u: determines response to disturbances
UCSB, Nov 2013 11 / 18
![Page 24: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/24.jpg)
Input-Output Analysis of the Linearized NS Equations
∂t
[∆vω
]=
[U′′∂x − U∆∂x + 1
R ∆2 0−U′∂z −U∂x + 1
R ∆
] [vω
]+
[−∂xy ∂2
x + ∂2z −∂zy
∂z 0 −∂x
][ dxdydz
] u
vw
=(∂2
x + ∂2z
)−1[
∂xy −∂z∂2
x + ∂2z 0
∂zy ∂x
][vω
]
x
y
z
uv
wNSR
(spatio-temporal system)
d u
∂tΨ = A Ψ + B du = C Ψ
Surprises:Even when A is stable the gain d −→ u can be very large
Input-output resonances very different from least-damped modes of A
UCSB, Nov 2013 11 / 18
![Page 25: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/25.jpg)
Spatio-temporal Impulse and Frequency Responses
Translation invariance in x & z implies
x
y
z
uv
w
Impulse Response
u(t, x, y, z) =
∫G(t − τ, x− ξ, y, y′, z− ζ) d(τ, ξ, y′, ζ) dτdξdy′dζ
u(t, x, ., z) =
∫G(t − τ, x− ξ, z− ζ) d(τ, ξ, ., ζ) dτdξdζ
G(t, x, z) : Operator-valued impulse response function
Frequency Response
u(ω, kx, kz) = G(ω, kx, kz) d(ω, kx, kz)
G(ω, kx, kz) : Operator-valued frequency response. Packs lots of information!
UCSB, Nov 2013 12 / 18
![Page 26: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/26.jpg)
Modal vs. Input-Output Analysis
x
y
z
uv
wNSR
(spatio-temporal system)
d u∂tΨ = A Ψ + B d
u = C Ψ
IR: G(t, x, z)
FR: G(ω, kx, kz)
UCSB, Nov 2013 13 / 18
![Page 27: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/27.jpg)
Modal vs. Input-Output Analysis
x
y
z
uv
wNSR
(spatio-temporal system)
d u∂tΨ = A Ψ + B d
u = C Ψ
IR: G(t, x, z)
FR: G(ω, kx, kz)
Modal Analysis: Look for unstable eigs of A
Flow type Classical linear theory Rc Experimental Rc
Channel Flow 5772 ≈ 1,000-2,000Plane Couette ∞ ≈ 350Pipe Flow ∞ ≈ 2,200-100,000
UCSB, Nov 2013 13 / 18
![Page 28: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/28.jpg)
Modal vs. Input-Output Analysis
x
y
z
uv
wNSR
(spatio-temporal system)
d u∂tΨ = A Ψ + B d
u = C Ψ
IR: G(t, x, z)
FR: G(ω, kx, kz)
Modal Analysis: Look for unstable eigs of A
Channel Flow @ R = 6000, kx = 1, kz = 0:
Flow structure of corresponding eigenfunction:Tollmein-Schlichting (TS) waves
UCSB, Nov 2013 13 / 18
![Page 29: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/29.jpg)
Modal vs. Input-Output Analysis
x
y
z
uv
wNSR
(spatio-temporal system)
d u∂tΨ = A Ψ + B d
u = C Ψ
IR: G(t, x, y,−1, z)
FR: G(ω, kx, kz)
Impulse Response Analysis: Channel Flow @ R = 2000
similar to “turbulent spots”
UCSB, Nov 2013 13 / 18
![Page 30: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/30.jpg)
Spatio-temporal Frequency Response
G(ω, kx, kz) is a large object! one aggregation method: supω σmax
(G(ω, kx, kz)
)
What do the corresponding flow structures look like?
UCSB, Nov 2013 14 / 18
![Page 31: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/31.jpg)
Spatio-temporal Frequency Response
G(ω, kx, kz) is a large object! one aggregation method: supω σmax
(G(ω, kx, kz)
)
What do the corresponding flow structures look like?
UCSB, Nov 2013 14 / 18
![Page 32: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/32.jpg)
Spatio-temporal Frequency Response
G(ω, kx, kz) is a large object! one aggregation method: supω σmax
(G(ω, kx, kz)
)
What do the corresponding flow structures look like?
UCSB, Nov 2013 14 / 18
![Page 33: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/33.jpg)
Spatio-temporal Frequency Response
G(ω, kx, kz) is a large object! one aggregation method: supω σmax
(G(ω, kx, kz)
)
What do the corresponding flow structures look like?closer (than TS waves) to structures seen in turbulent boundary layers
UCSB, Nov 2013 14 / 18
![Page 34: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/34.jpg)
Modal vs. Input-Output Response
Correspondence: poles of a transfer function←→ frequency response
Typically: underdamped poles←→ frequency response peaks
cf. The “rubber sheet analogy”:
UCSB, Nov 2013 15 / 18
![Page 35: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/35.jpg)
Modal vs. Input-Output Response
Correspondence: poles of a transfer function←→ frequency response
However: no connection necessary between
pole locations and FR peaks
Theorem: Let z1, . . . , zn be any locations in the left halfof the complex plane.Any stable frequency response function in H2 can bearbitrarily closely approximated by a transfer function ofthe following form:
H(s) =
N1∑i=1
α1,i
(s− z1)i + · · · +
Nn∑i=1
αn,i
(s− zn)i
by choosing any of the Nk ’s large enough
Re(s)
Im(s)
|H(s)|
X X
X
UCSB, Nov 2013 15 / 18
![Page 36: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/36.jpg)
Modal vs. Input-Output Response
Correspondence: poles of a transfer function←→ frequency response
However: no connection necessary between
pole locations and FR peaks
Theorem: Let z1, . . . , zn be any locations in the left halfof the complex plane.Any stable frequency response function in H2 can bearbitrarily closely approximated by a transfer function ofthe following form:
H(s) =
N1∑i=1
α1,i
(s− z1)i + · · · +
Nn∑i=1
αn,i
(s− zn)i
by choosing any of the Nk ’s large enough
Re(s)
Im(s)
|H(s)|
X X
X
For large-scale systems: IO behavior not predictable from modal behavior
UCSB, Nov 2013 15 / 18
![Page 37: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/37.jpg)
Implications for Turbulence
For large-scale systems: IO behavior not predictable from modal behavior
x
y
z
uv
wNSR
(spatio-temporal system)
d u∂tΨ = A Ψ + B d
u = C Ψ
IR: G(t, x, z)
FR: G(ω, kx, kz)
“modal behavior”: Stability due to i.c. condition uncertainty“IO behavior”: behavior in the presence of ambient uncertainty
I forcing terms from wall roughness and/or vibrationsI Free-stream disturbances in boundary layersI Thermal (Langevin) forcesI uncertain dynamics
UCSB, Nov 2013 16 / 18
![Page 38: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/38.jpg)
Implications for Turbulence
For large-scale systems: IO behavior not predictable from modal behavior
x
y
z
uv
wNSR
(spatio-temporal system)
d u∂tΨ = A Ψ + B d
u = C Ψ
IR: G(t, x, z)
FR: G(ω, kx, kz)
“modal behavior”: Stability due to i.c. condition uncertainty“IO behavior”: behavior in the presence of ambient uncertainty
I forcing terms from wall roughness and/or vibrationsI Free-stream disturbances in boundary layersI Thermal (Langevin) forcesI uncertain dynamics
UCSB, Nov 2013 16 / 18
![Page 39: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/39.jpg)
The Nature of Turbulence
Fluid flows are described by deterministic equationsOLD QUESTION: why do fluid flows “look random” at high R?
UCSB, Nov 2013 17 / 18
![Page 40: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/40.jpg)
The Nature of Turbulence
A common view of turbulencest
ate
spac
e
R
unstable
stable
u0
UCSB, Nov 2013 17 / 18
![Page 41: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/41.jpg)
The Nature of Turbulence
A common view of turbulencest
ate
spac
e
R
unstable
stable
u0
chaoticdynamics
UCSB, Nov 2013 17 / 18
![Page 42: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/42.jpg)
The Nature of Turbulence
A common view of turbulencest
ate
spac
e
R
unstable
stable
u0
chaoticdynamics
Intuitive reasoning:Complex, “statistical looking” behavior ←→ chaotic dynamics
UCSB, Nov 2013 17 / 18
![Page 43: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/43.jpg)
The Nature of Turbulence
A common view of turbulencest
ate
spac
e
R
unstable
stable
u0
chaoticdynamics
Intuitive reasoning:Complex, “statistical looking” behavior ←→ chaotic dynamicsAssumes NS eqs. with perfect BC, no disturbances or uncertainty
(i.e. a a closed system)
UCSB, Nov 2013 17 / 18
![Page 44: Wall-Bounded Shear Flow Transitionbamieh/talks/UCSB... · 2013-11-13 · Transition & Turbulence as Natural Phenomena All fluid flows transition (as 0 ! 1R) from laminar to turbulent](https://reader034.vdocuments.us/reader034/viewer/2022042113/5e8f414993e5c04a943f4825/html5/thumbnails/44.jpg)
The Nature of TurbulenceAn Alternate Possibility
A driven (open) system
Dynamics of the NS Equations
NoiseSurface RoughnessThermal ForcesFree Stream disturbances
FluctuatingFlow Field
(looks statistical)
The NS equations act as an amplifier of ambient uncertainty at high R
Qualitatively similar to
Dynamics of the Linearized
NS Equations
NoiseSurface RoughnessThermal ForcesFree Stream disturbances
FluctuatingFlow Field
(looks statistical)
UCSB, Nov 2013 18 / 18