SlideShare a Scribd company logo
1 of 87
Download to read offline
Introduction Global stability theory Roughness-induced transition Conclusion
Dynamics and global stability analysis of
three-dimensional flows
Jean-Christophe Loiseau1,2
supervisor: Jean-Christophe Robinet1
co-supervisor: Emmanuel Leriche2
(1): DynFluid Laboratory - Arts & M´etiers-ParisTech - 75013 Paris, France
(2): LML - University of Lille 1 - 59655 Villeneuve d’Ascq, France
PhD Defence, May 26th 2014
1/79
Introduction Global stability theory Roughness-induced transition Conclusion
What are hydrodynamic instabilities?
• Let us consider the flow of water (ν = 15.10−6 m2.s−1) past a
two-dimensional cylinder of diameter D = 1.5 cm.
• If water flows from left to right at U = 4.5 cm.s−1 (Re = 45),
nothing really fancy takes place: the flow is steady and stable.
2/79
Introduction Global stability theory Roughness-induced transition Conclusion
What are hydrodynamic instabilities?
• If you increase the velocity to U = 5 cm.s−1 (Re = 50), the flow
looks very different.
• The steady flow became (globally) unstable and has experienced a
(supercritical) bifurcation.
3/79
Introduction Global stability theory Roughness-induced transition Conclusion
How do we study these instabilities?
• Let us consider a non-linear dynamical system
B
∂Q
∂t
= F(Q) (1)
1. Compute a fixed point (or base flow): F(Qb) = 0
2. Linearise the dynamics of an infinitesimal perturbation q in the vicinity
of this solution:
B
∂q
∂t
= Jq with J =
∂F
∂q
(2)
3. Investigate the stability properties of this linear dynamical system.
4/79
Introduction Global stability theory Roughness-induced transition Conclusion
How do we study these instabilities?
• In the context of fluid dynamics, this includes several different
approaches depending on the nature of the base flow:
• Local stability analysis for parallel flows:
֒→ Temporal stability, Spatial stability, Absolute/Convective stability,
Response to harmoning forcing, Transient growth
• Global stability analysis for two-dimensional and three-dimensional
flows:
֒→ Temporal stability, Response to harmoning forcing (Resolvent),
Transient growth
5/79
Introduction Global stability theory Roughness-induced transition Conclusion
Local stability analysis
• The base flow depends on a single space coordinate:
Ub = (Ub(y), 0, 0)T
• Linear dynamical system (2) is now autonomous in time and in the x
and z coordinates of space.
֒→ The perturbation q can be decomposed into normal modes:
q(x, y, z, t) = ˆq(y) exp(iαx + iβz + λt) + c.c with λ = σ + iω
6/79
Introduction Global stability theory Roughness-induced transition Conclusion
Local stability analysis
• Introducing such decomposition into the system (2) yields to a
generalised eigenvalue problem:
λBˆq = J(y, α, β)ˆq (3)
• The stability of the base flow Ub is governed by the growth rate σ:
֒→ If σ < 0, the base flow is said to be locally stable.
֒→ If σ > 0, the base flow is said to be locally unstable.
7/79
Introduction Global stability theory Roughness-induced transition Conclusion
Local stability analysis
Theoretical point of view
• Relies on the parallel flow assumption.
• Provides insights into the local stability properties of the flow.
֒→ Requires a good theoretical and mathematical background.
Practical point of view
• The generalised eigenproblem involves small matrices (∼ 100 × 100)
• Can be solved using direct eigenvalue solvers in a matter of seconds
even on a 10 years old laptop.
8/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability theory
• The base flow has two components both depending on the x and y
space coordinates:
Ub = (Ub(x, y), Vb(x, y), 0)T
• Linear dynamical system (2) is now only autonomous in time and in
z.
֒→ The perturbation q can be decomposed into normal modes:
q(x, y, z, t) = ˆq(x, y) exp(iβz + λt) + c.c with λ = σ + iω
Base flow of the 2D separated boundary layer at Re = 600 as in Ehrenstein & Gallaire (2008).
9/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability theory
• Introducing such decomposition into the system (2) yields to a
generalised eigenvalue problem once again:
λBˆq = J(x, y, β)ˆq (4)
• The stability of the base flow Ub is governed by the growth rate σ:
֒→ If σ < 0, the base flow is said to be globally stable.
֒→ If σ > 0, the base flow is said to be globally unstable.
Streamwise velocity component of the leading unstable global mode for the 2D separated boundary layer.
10/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability analysis
Theoretical point of view
• Got rid of the parallel flow assumption.
• Allows to investigate more realistic configurations as separated flows
very common in Nature and industries.
Practical point of view
• The generalised eigenproblem involves relatively large matrices
(∼ 105 × 105)
• Mostly solved using iterative eigenvalue solvers on large workstations.
11/79
Introduction Global stability theory Roughness-induced transition Conclusion
Objectives
• Bagheri et al. (2008) and Ilak et al. (2012) performed the first global
stability analysis ever on a 3D flow (jet in crossflow).
• Extension of the global stability tools to a fully three-dimensional
framework.
֒→ Mostly a numerical problem due to the (extremely) large matrices
involved.
12/79
Introduction Global stability theory Roughness-induced transition Conclusion
Objectives
λ2 visualisation of the hairpin vortices shed behind a hemispherical roughness element. Courtesy of P. Fischer.
13/79
Introduction Global stability theory Roughness-induced transition Conclusion
Context
• PhD thesis part of a larger project: Simulation and Control of
Geometrically Induced Flows (SICOGIF)
֒→ Funded by the French National Agency for Research (ANR)
֒→ Involves several different parties (IRPHE, EPFL, Arts et M´etiers
ParisTech and Universit´e Lille-1)
֒→ Aims at improving our understanding of instability and transition in
complex 2D and 3D separated flows both from an experimental and
numerical point of view.
14/79
Introduction Global stability theory Roughness-induced transition Conclusion
Context
• Three flow configurations have been investigated:
֒→ The lid-driven cavity flow
֒→ The asymmetric stenotic pipe flow
֒→ The roughness-induced boundary layer flow
Vertical velocity component of the leading global mode for a LDC having a spanwise extent Λ = 6 at Re = 900.
15/79
Introduction Global stability theory Roughness-induced transition Conclusion
Context
• Three flow configurations have been investigated:
֒→ The lid-driven cavity flow
֒→ The asymmetric stenotic pipe flow
֒→ The roughness-induced boundary layer flow
Streamwise velocity component for the two existing steady states of an asymmetric stenotic pipe flow at Re = 400.
16/79
Introduction Global stability theory Roughness-induced transition Conclusion
Context
• Three flow configurations have been investigated:
֒→ The lid-driven cavity flow
֒→ The asymmetric stenotic pipe flow
֒→ The roughness-induced boundary layer flow
Identification of the vortical structures by the λ2 criterion (Jeong & Hussain 1995).
17/79
Introduction Global stability theory Roughness-induced transition Conclusion
Introduction
Global stability theory and algorithm
Base flows
Global stability theory
How to solve the eigenvalue problem?
Roughness-induced transition
Motivations
Fransson 2005 experiment
Parametric investigation
Physical analysis
Non-linear evolution
Conclusions & Perspectives
Conclusions
LDC & Stenosis
Perspectives
18/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability analysis of
three-dimensional flows
18/79
Introduction Global stability theory Roughness-induced transition Conclusion
How to compute base flows?
• Base flow are given by:
F(Qb) = 0 (5)
• Various techniques can be employed to compute these peculiar
solutions:
֒→ Analytical solutions, impose appropriate symmetries, Newton and
quasi-Newton methods, ...
• In the present work, we use the Selective frequency damping
approach (see Akervik et al. 2006).
19/79
Introduction Global stability theory Roughness-induced transition Conclusion
Selective frequency damping
• Enables the stabilisation of the solution by applying a low-pass filter
to the Navier-Stokes equations.
֒→ A forcing term is added to the r.h.s of the equations.
֒→ The system is extended with an equation for the filtered state.



∂Q
∂t
= F(Q) + χ(Q − ¯Q)
∂ ¯Q
∂t
= ωc(Q − ¯Q)
(6)
• The cutoff frequency ωc is connected to the frequency of the most
dominant instabilities and should be smaller than this frequency
(ωc < ω).
• The gain χ needs to be large enough to stabilise the system (χ > σ).
20/79
Introduction Global stability theory Roughness-induced transition Conclusion
Selective frequency damping
Pros
→ Really easy to implement within
an existing DNS code.
→ Memory footprint similar to that
of a simple direct numerical
simulation.
→ Easy to use/tune the low-pass
filter.
Cons
→ As time-consuming as a direct
numerical simulation.
→ Requires a priori information
regarding the instability of the
flow.
→ Unable to stabilise the system if
the instability is non-oscillating.
21/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability theory
• Dynamics of a three-dimensional infinitesimal perturbation
q = (u, p)T evolving onto the base flow Qb = (Ub, Pb)T are
governed by:



∂u
∂t
= −(u · ∇)Ub − (Ub · ∇)u − ∇p +
1
Re
∆u
∇ · u = 0
(7)
• If projected onto a divergence-free vector space, this set of equations
can be recast into:
∂u
∂t
= Au (8)
with A the (projected) Jacobian matrix of the Navier-Stokes
equations.
22/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability theory
• Using a normal mode decomposition
u(x, y, z, t) = ˆu(x, y, z)e(σ+iω)t
+ c.c
• System (8) can be formulated as an eigenvalue problem
(σ + iω)ˆu = Aˆu (9)
• The sign of σ determines the stability of the base flow Ub:
֒→ If σ < 0, the base flow is said to be asymptoticaly linearly stable.
֒→ If σ > 0, the base flow is said to be asymptoticaly linearly unstable.
• ω determines whether the instability is oscillatory (ω = 0) or not
(ω = 0).
23/79
Introduction Global stability theory Roughness-induced transition Conclusion
How to solve the eigenvalue problem?
• Depends on the dimension of the discretised problem.
Base Flow Inhomogeneous Dimension Storage
direction(s) of ˆu of A
Poiseuille U(y) 1D 102 ∼ 1 Mb
2D bump U(x, y) 2D 105 ∼ 1-50 Gb
3D bump U(x, y, z) 3D 107 ∼ 1-100 Tb
• For 3D global stability problem, A is so large that it cannot be
explicitely constructed.
Matrix-free approach is mandatory!
24/79
Introduction Global stability theory Roughness-induced transition Conclusion
Time-stepping approach
• Time-stepping approach (Edwards et al. 1994, Bagheri et al. 2008) is
based on the formal solution to system (8):
u(∆t) = eA∆t
u0
• The operator M(∆t) = eA∆t is nothing but a matrix. Its application
on u0 can be computed by time-marching the linearised Navier-Stokes
equations.
֒→ Its stability properties can be investigated by eigenvalue analysis.
MˆU = ˆUΣ (10)
with ˆU the matrix of eigenvectors and Σ the eigenvalue matrix of
M = eA∆t.
25/79
Introduction Global stability theory Roughness-induced transition Conclusion
Arnoldi algorithm
• The eigenvalue problem (10) is solved using an Arnoldi algorithm.
1. Given M and u0, construct a small Krylov subspace (compared to the
size of the initial problem),
Km(M, u0) = span u0, Mu0, M2
u0, · · · , M(m−1)
u0
2. Orthonormalize: U = [U1, · · · , Um]
3. Project operator M ≈ UHUT
−→ MUk = Uk Hk + rk eT
k
with Hk : upper Hessenberg matrix.
4. Solve small eigenvalue problem (ΣH , X): HX = XΣH, (m × m),
m < 1000
5. Link with the initial eigenproblem (ΛA, ˆu):
ΛA =
log(ΣH)
∆t
, ˆu = UX
26/79
Introduction Global stability theory Roughness-induced transition Conclusion
Arnoldi algorithm
27/79
Introduction Global stability theory Roughness-induced transition Conclusion
Summary
• All calculations have been performed with the code Nek 5000
֒→ Legendre spectral elements code developed by P. Fischer at Argonne
National Laboratory.
֒→ Semi-implicit temporal scheme.
֒→ Massively parallel code based on an MPI strategy.
• Base flow computation
֒→ Selective frequency damping approach : application of a low-pass filter
to the fully non-linear Navier-Stokes equations (Akervik et al. 2006).
• Global stability analysis
֒→ Arnoldi algorithm similar to the one published by Barkley et al. (2008).
28/79
Introduction Global stability theory Roughness-induced transition Conclusion
Roughness-induced transition
28/79
Introduction Global stability theory Roughness-induced transition Conclusion
Motivations
• Roughness elements have numerous applications in aerospace
engineering:
֒→ Stabilisation of the Tollmien-Schlichting waves,
֒→ Shift and/or control of the transition location, ...
• Their influence on the flow has been extensively investigated since the
early 1950’s.
Experimental visualisation of the flow induced by a roughness element. Gregory & Walker, 1956.
29/79
Introduction Global stability theory Roughness-induced transition Conclusion
Delay of the natural transition
• Cossu & Brandt (2004): Theoretical prediction of the stabilisation of
TS waves by streamwise streaks.
• Fransson et al. (2004-2006): Experimental demonstration using a
periodic array of roughness elements.
Schematic setup
Experimental observations
Figures from Fransson et al. (2006).
30/79
Introduction Global stability theory Roughness-induced transition Conclusion
Roughness-induced transition
• Problem: If the Reynolds number is too high, transition occurs right
downstream the roughness elements!
Illustration of the early roughness-induced transition. λ2 visualisation of the vortical structures.
31/79
Introduction Global stability theory Roughness-induced transition Conclusion
Roughness-induced transition
• Since the early 1950’s: Numerous experimental investigations.
֒→ Transition diagram by von Doenhoff & Braslow (1961).
• Despite the large body of literature, the underlying mechanisms are
not yet fully understood.
Transition diagram from von Doenhoff & Braslow (1961).
32/79
Introduction Global stability theory Roughness-induced transition Conclusion
Motivations
• Methods used until now rely on a parallel flow assumption:
֒→ Local stability theory (Brandt 2006, Denissen & White 2013, ...),
֒→ Local transient growth theory (Vermeersch 2010, ...)
• Objective:
֒→ Might a 3D global instability of the flow explain the roughness-induced
transition?
֒→ If so, what are the underlying physical mechanisms?
• Methods:
֒→ Fully three-dimensional global stability analyses,
֒→ Direct numerical simulations,
֒→ Comparison with available experimental data.
33/79
Introduction Global stability theory Roughness-induced transition Conclusion
Problem formulation
z X
y
d h
Lz
l
Lx
Ly
δ
0
Sketch of the computational arrangement and various scales used for DNS and stability analysis.
- (Lx , Ly , Lz ) = (105, 50, 8η)
- η = d/h = 1, 2, 3
- Re = Ueh/ν
- Reδ∗
= Ueδ∗/ν
- Inflow: Blasius profile,
- Outflow: ∇U · x = 0,
- Top: U = 1, ∂y V = ∂y W = 0,
- Wall: no-slip B.C.
- Lateral: periodic B.C.
34/79
Introduction Global stability theory Roughness-induced transition Conclusion
Numerical informations
η Number of SEM Gridpoints (N = 6-12) Number of cores used
1 10 000 2-17.106 256
2 17 500 3.5-30.106 512
3 20 000 4.5-35.106 512
Typical size of the numerical problem investigated. N is the order of the Legendre polynomials used in the three directions
within each element.
Typical SEM distribution in a given horizontal plane. Full mesh with Legendre polynomials of order N = 8.
35/79
Introduction Global stability theory Roughness-induced transition Conclusion
Roughness-induced transition
The Fransson 2005 experiment
35/79
Introduction Global stability theory Roughness-induced transition Conclusion
Experimental setup
• Experimental demonstration of the ability for finite amplitude streaks
to stabilise TS waves.
• Unfortunately, transition takes place right downstream the array of
roughness elements if the Reynolds number is too high.
h D η Lz/h xk/h Rec
δ∗
1.4mm 4.2mm 3 10 57.14 ≃ 290
36/79
Introduction Global stability theory Roughness-induced transition Conclusion
Base flow
(a)
(b)
• Upstream and downstream
reversed flow regions:
֒→ Induces a central low-speed
region.
• Vortical system stemming:
֒→ Investigated by Baker (1978)
֒→ Horseshoe vortices whose legs
are streamwise oriented
counter-rotating vortices.
֒→ Creation of streamwise
velocity streaks (lift-up effect)
37/79
Introduction Global stability theory Roughness-induced transition Conclusion
Base flow
(a) X=20 (b) X=40 (c) X=60 (d) X=80
Visualisation of the base flow deviation from the Blasius boundary layer flow in various streamwise planes for Re = 466. High
speed streaks are in red while low-speed ones are in blue.
• Low-speed region generated by the roughness element’s blockage.
֒→ Fades away quite rapidly in the streamwise direction.
• High- and low-speed streaks on each side of the roughness element
due to the horshoe vortex.
֒→ Sustains over quite a long streamwise distance.
38/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability
Eigenspectrum of the linearised Navier-Stokes operator.
• Hopf bifurcation taking place in-between 550 < Rec < 575.
֒→ Linear interpolation: Rec = 564, i.e. Rec
δ∗
= 309.
39/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability
(a) Top view of u = ±10% iso-surfaces
(b) X = 23 (c) X = 40
Visualisation of streamwise velocity component of the leading unstable mode for Re = 575.
40/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability
• Leading unstable mode exhibits a varicose symmetry:
֒→ Streamwise alternated patches of positive and negative velocity mostly
localised along the central low-speed region.
֒→ Non-linear DNS have revealed that it gives birth to hairpin vortices.
• Rec predicted by global stability analysis only 6% larger than the
experimental one from Fransson et al. (2005):
֒→ Global instability of the flow appears as one of the possible
explanations to roughness-induced transition.
41/79
Introduction Global stability theory Roughness-induced transition Conclusion
Roughness-induced transition
Parametric investigation
41/79
Introduction Global stability theory Roughness-induced transition Conclusion
Parametric investigation
• Aims of the parametric investigation:
֒→ How do the Reynolds number and the aspect ratio of the roughness
elements impact the base flow and its stability properties?
֒→ Does the leading unstable mode always exhibit a varicose symmetry?
• To do so:
֒→ The spanwise extent of the domain is taken large enough so that the
roughness element behaves as being isolated.
֒→ δ99/h is set to 2 to isolate the influence of the Reynolds number only.
֒→ The roughness element’s aspect ratio varies from η = 1 up to η = 3.
42/79
Introduction Global stability theory Roughness-induced transition Conclusion
Base flows
(a) (Re, η) = (600, 1) (b) (Re, η) = (1250, 1)
Influence of the Reynolds number on the flow within the z = 0 plane. Streamwise velocity iso-contours ranging from U = 0.1
up to U = 0.99.
• Influence of the Reynolds number:
֒→ Does not qualitatively change the shape of the downstream reversed
flow region.
֒→ Strengthen the gradients and reduces the thickness of the shear layer.
43/79
Introduction Global stability theory Roughness-induced transition Conclusion
Base flows
(a) (Re, η) = (600, 1)
(b) (Re, η) = (1250, 1)
Top view of the low-speed (white) and high-speed (black) streaks induced by the roughness element. Streak have been
identified using the deviation of the base flow streamwise velocity from the Blasius boundary layer flow.
• Influence of the Reynolds number:
֒→ Strongly increases the amplitude and the streamwise extent of the
central low-speed region.
֒→ Slightly increases the amplitude of the outer velocity streaks.
44/79
Introduction Global stability theory Roughness-induced transition Conclusion
Base flows
(a) (Re, η) = (600, 2) (b) (Re, η) = (600, 3)
Influence of the aspect ratio on the flow within the z = 0 plane. Streamwise velocity iso-contours ranging from U = 0.1 up to
U = 0.99.
• Influence of the aspect ratio:
֒→ Strengthen the gradients and reduces the thickness of the shear layer.
֒→ Strongly increases the amplitude and the streamwise extent of the
central low-speed region.
֒→ Strongly increases the amplitude of the outer velocity streaks.
45/79
Introduction Global stability theory Roughness-induced transition Conclusion
Base flows
(a) (Re, η) = (600, 2)
(b) (Re, η) = (600, 3)
Top view of the low-speed (white) and high-speed (black) streaks induced by the roughness element. Streak have been
identified using the deviation of the base flow streamwise velocity from the Blasius boundary layer flow.
46/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability
ω
σ
0 0.5 1 1.5 2
-0.04
-0.02
0
0.02
0.04
(a) (Re, η) = (1200, 1)
ω
σ
0 0.5 1 1.5
-0.04
-0.02
0
0.02
0.04
(b) (Re, η) = (900, 2)
ω
σ
0 0.5 1 1.5
-0.04
-0.02
0
0.02
0.04
(c) (Re, η) = (700, 3)
Eigenspectra of the linearised Navier-Stokes operator for different roughness elements.
47/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability
(a) (Re, η) = (1200, 1)
(b) (Re, η) = (900, 2)
Evolution of the leading unstable mode when the roughness element’s aspect ratio is changed from η = 1 to η = 2. Isosurfaces
u = ± 10% of the modes streamwise velocity component.
48/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability
(a) (Re, η) = (1200, 1) (b) (Re, η) = (900, 2)
Evolution of the leading unstable mode when the roughness element’s aspect ratio is changed from η = 1 to η = 2 in the
X = 25 plane.
49/79
Introduction Global stability theory Roughness-induced transition Conclusion
Global stability
• Increasing the roughness element’s aspect ratio decreases the critical
Reynolds number.
η 1 2 3 Fransson (η = 3)
Rec 1040 850 656 564
Rec
h 813 630 513 519
Symmetry S V V V
Summary of the global stability analyses. V: varicose, S: sinuous. Reh is the roughness Reynolds number.
• Exchange of symmetry in qualitative agreements with Sakamoto &
Arie (1983) and Beaudoin (2004).
50/79
Introduction Global stability theory Roughness-induced transition Conclusion
Roughness-induced transition
Physical analysis
50/79
Introduction Global stability theory Roughness-induced transition Conclusion
Physical analysis
• Aims of the analysis:
֒→ Unravel the underlying physical mechanisms for each mode.
֒→ How and where do they extract their energy?
֒→ Where do they originate?
• Type of analysis:
֒→ Kinetic energy transfer between the base flow and the perturbation
(Brandt 2006).
֒→ Computation of the wavemaker region (Giannetti & Luchini 2007).
51/79
Introduction Global stability theory Roughness-induced transition Conclusion
Kinetic energy budget
• The evolution of the perturbation’s kinetic energy is governed by the
Reynolds-Orr equation:
∂E
∂t
= −D +
9
i=1 V
Ii dV (11)
• with the total kinetic energy E and dissipation D given by:
E =
1
2 V
u · u dV , and D =
1
Re V
∇u : ∇u dV (12)
52/79
Introduction Global stability theory Roughness-induced transition Conclusion
Kinetic energy budget
• The integrands Ii representing the different production terms are
given by:
I1 = −u2 ∂Ub
∂x
, I2 = −uv
∂Ub
∂y
, I3 = −uw
∂Ub
∂z
I4 = −uv
∂Vb
∂x
, I5 = −v2 ∂Vb
∂y
, I6 = −vw
∂Vb
∂z
I7 = −wu
∂Wb
∂x
, I8 = −wv
∂Wb
∂y
, I9 = −w2 ∂Wb
∂z
(13)
• Their sign indicates whether the associated local transfer of kinetic
energy acts as stabilising (negative) or destabilising (positive).
53/79
Introduction Global stability theory Roughness-induced transition Conclusion
Kinetic energy budget: Sinuous mode
0
0.5
1
I1 I2 I3 I4 I5 I6 I7 I8 I9 D
(a) (Re, η) = (1125, 1)
0
0.5
1
I1 I2 I3 I4 I5 I6 I7 I8 I9 D
(b) (Re, η) = (1250, 1)
X0 30 60 90
2.0x10
-03
4.0x10
-03
6.0x10
-03
∫I2dydz
∫I3dydz
(c) (Re, η) = (1125, 1)
X0 30 60 90
2.0x10
-03
4.0x10
-03
6.0x10
-03
∫I2dydz
∫I3dydz
(d) (Re, η) = (1250, 1)
Top: Sinuous unstable mode’s kinetic energy budget integrated over the whole domain. Bottom: Streamwise evolution of the
production terms y,z I2 dydz (red dashed line) and y,z I3 dydz (blue solid line).
54/79
Introduction Global stability theory Roughness-induced transition Conclusion
Kinetic energy budget: Sinuous mode
(a) I2 = −uv∂U/∂y (b) I3 = −uw∂U/∂z
Spatial distribution of the I2 = −uv∂Ub/∂y (a) and I3 = −uw∂Ub/∂z (b) production terms in the plane x = 25 for
(Re, η) = (1125, 1). Solid lines depict the base flows streamwise velocity isocontours from Ub = 0.1 to 0.99, whereas the red
dashed lines stand for the location of the shear layer.
55/79
Introduction Global stability theory Roughness-induced transition Conclusion
Kinetic energy budget: Sinuous mode
(a) I2 = −uv∂U/∂y
(b) I3 = −uw∂U/∂z
Spatial distribution of I2 (c) and I3 (d) in the y = 0.75 horizontal plan.
56/79
Introduction Global stability theory Roughness-induced transition Conclusion
Kinetic energy budget: Varicose mode
0
1
2
I1 I2 I3 I4 I5 I6 I7 I8 I9 D
(a) (Re, η) = (850, 2)
0
1
2
I1 I2 I3 I4 I5 I6 I7 I8 I9 D
(b) (Re, η) = (1000, 2)
X0 30 60 90
.0x10
+00
4.0x10
-03
8.0x10
-03
∫I2dydz
∫I3dydz
(c) (Re, η) = (850, 2)
X0 30 60 90
.0x10
+00
4.0x10
-03
8.0x10
-03
∫I2dydz
∫I3dydz
(d) (Re, η) = (1000, 2)
Top: Varicose unstable mode’s kinetic energy budget integrated over the whole domain. Bottom: Streamwise evolution of the
production terms y,z I2 dydz (red dashed line) and y,z I3 dydz (blue solid line).
57/79
Introduction Global stability theory Roughness-induced transition Conclusion
Kinetic energy budget: Varicose mode
(a) I2 = −uv∂U/∂y (b) I3 = −uw∂U/∂z
Spatial distribution of the I2 = −uv∂Ub/∂y (a) and I3 = −uw∂Ub/∂z (b) production terms in the plane x = 25 for
(Re, η) = (850, 2). Solid lines depict the base flows streamwise velocity isocontours from Ub = 0.1 to 0.99, whereas the red
dashed lines stand for the location of the shear layer.
58/79
Introduction Global stability theory Roughness-induced transition Conclusion
Kinetic energy budget: Varicose mode
Spatial distribution of the I3 = −uw∂Ub/∂z production term in the plane y = 0.5 for (Re, η) = (850, 2).
59/79
Introduction Global stability theory Roughness-induced transition Conclusion
Wavemaker
• Kinetic energy budgets provide valuable insights into the mode’s
dynamics but very limited about its core region, i.e. the wavemaker.
• Defined by Giannetti & Luchini (2007) as the overlap of the direct
global mode u and its adjoint u†:
ζ(x, y, z) =
u† u
u†, u
(14)
• Allows the identification of the most likely region for the inception of
the global instability under consideration.
60/79
Introduction Global stability theory Roughness-induced transition Conclusion
Wavemaker
Figure: Sinuous wavemaker in the y = 0.75 plane.
Figure: Varicose wavemaker in the z = 0 plane.
61/79
Introduction Global stability theory Roughness-induced transition Conclusion
Wavemaker
Figure: Sinuous wavemaker in the y = 0.75 plane.
Figure: Varicose wavemaker in the z = 0 plane.
61/79
Introduction Global stability theory Roughness-induced transition Conclusion
Wavemaker
• Sinuous wavemaker:
֒→ Exclusively localised within the spatial extent of the downstream
reversed flow region.
֒→ Shares close connections with the von K´arm´an global instability in the
2D cylinder flow (Giannetti & Luchini 2007, Marquet et al. 2008).
• Varicose wavemaker:
֒→ Localised on the top of the central low-speed region shear layer.
֒→ Quite extended in the streamwise direction.
֒→ Yet, its amplitude in the reversed flow region is almost ten times larger
than its amplitude in the wake.
62/79
Introduction Global stability theory Roughness-induced transition Conclusion
Sinuous instability mechanism
What we know from local stability
approaches?
• Central low-speed region can
sustain local convective
instabilities (Brandt 2006).
• Related to the work of the
Reynolds stresses against the
wall-normal and spanwise
gradients of Ub.
• Not the dominant local
instability though.
What global stability analyses
revealed?
• Existence of a global sinuous
instability.
• Related to the downstream
reversed flow region.
• Similar to the von K´arm´an
instability in the 2D cylinder
flow.
63/79
Introduction Global stability theory Roughness-induced transition Conclusion
Varicose instability mechanism
What we know from local stability
approaches?
• Central low-speed region can
sustain local convective
instabilities (Brandt 2006,
Denissen & White 2013).
• Related to the work of the
Reynolds stresses against the
wall-normal gradient of Ub.
• Dominant local instability and
possible large transient growth
(Vermeersch 2010)
What global stability analyses
revealed?
• Existence of a global varicose
instability.
• Find its roots in the reversed
flow region.
• Mechanism might be similar to
the one proposed by Acarlar &
Smith (1987).
64/79
Introduction Global stability theory Roughness-induced transition Conclusion
Roughness-induced transition
Non-linear evolution
64/79
Introduction Global stability theory Roughness-induced transition Conclusion
Non-linear evolution
• Varicose instability
֒→ Induces a varicose modulation of the central low-speed region and
surrounding streaks.
֒→ Numerous hairpin vortices are shed right downstream the roughness
element and trigger very rapid transition to turbulence.
֒→ Dominant frequency and wavelength of this vortex shedding is well
captured by global stability analyses.
Streamwise velocity distribution in the y = 0.5 plane for (Re, η) = (575, 3).
65/79
Introduction Global stability theory Roughness-induced transition Conclusion
Non-linear evolution
• Varicose instability
֒→ Induces a varicose modulation of the central low-speed region and
surrounding streaks.
֒→ Numerous hairpin vortices are shed right downstream the roughness
element and trigger very rapid transition to turbulence.
֒→ Dominant frequency and wavelength of this vortex shedding is well
captured by global stability analyses.
Identification of the vortical structures by the λ2 criterion (Jeong & Hussain 1995).
66/79
Introduction Global stability theory Roughness-induced transition Conclusion
Non-linear evolution
• Sinuous instability
֒→ Induces a sinuous wiggling of the central low-speed region (Beaudoin
2004, Duriez et al. 2009).
֒→ Frequency of this sinuous wiggling well captured by global stability
analysis.
֒→ Hairpin vortices are nonetheless observed to be shed downstream the
roughness element..
Streamwise velocity distribution in the y = 0.5 plane for (Re, η) = (1125, 1).
67/79
Introduction Global stability theory Roughness-induced transition Conclusion
Non-linear evolution
• Sinuous instability
֒→ Induces a sinuous wiggling of the central low-speed region (Beaudoin
2004, Duriez et al. 2009).
֒→ Frequency of this sinuous wiggling well captured by global stability
analysis
֒→ Hairpin vortices are nonetheless observed to be shed downstream the
roughness element..
Identification of the vortical structures by the λ2 criterion (Jeong & Hussain 1995).
68/79
Introduction Global stability theory Roughness-induced transition Conclusion
Non-linear evolution
• Sinuous instability
֒→ Monitoring the amplitude of the spanwise velocity in the central
mid-plane revealed the bifurcation is supercritical.
−20 0 20 40 60 80 100
−0.1
−0.05
0
0.05
0.1
ε=Re−Rec
Amplitude
69/79
Introduction Global stability theory Roughness-induced transition Conclusion
Conclusion & Perspectives
69/79
Introduction Global stability theory Roughness-induced transition Conclusion
Conclusion
• Sinuous instability
֒→ Dominant instability for low aspect ratio roughness elements.
֒→ von K´arm´an-like global instability of the reversed flow region.
֒→ Vortices shed from this region then experiences weak spatial transient
growth.
֒→ The creation of hairpin vortices by sinuous global instability is not yet
understood.
70/79
Introduction Global stability theory Roughness-induced transition Conclusion
Conclusion
• Varicose instability
֒→ Dominant instability for large aspect ratio roughness elements.
֒→ Mechanism similar to the one proposed by Acarlar & Smith (1987).
֒→ Triggers rapid transition to a turbulent-like state by promoting the
creation of hairpin vortices.
71/79
Introduction Global stability theory Roughness-induced transition Conclusion
Conclusion
• Critical roughness Reynolds numbers and observations from DNS in
qualitatively good agreements with the transition diagram by von
Doenhoff & Braslow (1961).
֒→ Three-dimensional global instability of the flow appears as one of
the possible explanations to roughness-induced transition.
72/79
Introduction Global stability theory Roughness-induced transition Conclusion
Lid-driven cavity flow
• Same instability mechanism as
before:
֒→ Centrifugal instability of the
primary vortex core.
• For large LDC, Rec in good
agreements with predictions
from 2.5D stability analysis.
73/79
Introduction Global stability theory Roughness-induced transition Conclusion
Lid-driven cavity flow
• DNS revealed bursts of kinetic energy related to intermittent chaotic
dynamics.
֒→ Koopman modes decomposition suggests it would type-2 intermittent
chaos (Pomeau & Manneville 1980).
74/79
Introduction Global stability theory Roughness-induced transition Conclusion
Stenotic pipe flow
• Asymmetry of the stenosis triggers the wall-reattachment at lower Re
compared to the axisymmetric case.
• Existence of a hysteresis cycle related to a subcritical pitchfork
bifurcation.
75/79
Introduction Global stability theory Roughness-induced transition Conclusion
Stenotic pipe flow
• Nonetheless, predictions from global stability analyses are
uncorelatted to the experimental observations (Passaggia et al.)
֒→ Transition is dominated by transient growth.
• Preliminary optimal perturbation analysis appears to be more
conclusive.
76/79
Introduction Global stability theory Roughness-induced transition Conclusion
Perspectives
• Several questions are still unanswered and require further in-depth
investigations:
֒→ What is the mechanism responsible for the creation of hairpin vortices
in the sinuous case?
֒→ Is the varicose bifurcation super- or subcritical?
֒→ How does global optimal perturbation influence these transition
scenarii? Can they trigger subcritical transition to turbulence?
֒→ How does the shape of the roughness element impact the stability
properties of the flow?
77/79
Introduction Global stability theory Roughness-induced transition Conclusion
Perspectives
• How to answer these questions?
֒→ More direct numerical simulations!
֒→ Non-linear analyses of these DNS (Koopman modes decomposition,
POD, statistical analysis, ...).
֒→ Linear and non-linear transient growth analysis.
֒→ Conduct similar investigations for smooth bumps and hemispherical
roughness elements to assess the robustness of these results.
78/79
Dynamics and global stability of three-dimensional flows

More Related Content

What's hot

Fluid dynamics
Fluid dynamicsFluid dynamics
Fluid dynamicsCik Minn
 
Grlweap frank rausche
Grlweap  frank rauscheGrlweap  frank rausche
Grlweap frank rauschecfpbolivia
 
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...Lake Como School of Advanced Studies
 
Tachyon inflation in DBI and RSII context
Tachyon inflation in DBI and RSII contextTachyon inflation in DBI and RSII context
Tachyon inflation in DBI and RSII contextMilan Milošević
 
Fundamentals of Seismic Refraction
Fundamentals of Seismic RefractionFundamentals of Seismic Refraction
Fundamentals of Seismic RefractionAli Osman Öncel
 
Tachyonic and Localy Equivalent Canonical Lagrangians - The Polynomial Case -
Tachyonic and Localy Equivalent Canonical Lagrangians - The Polynomial Case -Tachyonic and Localy Equivalent Canonical Lagrangians - The Polynomial Case -
Tachyonic and Localy Equivalent Canonical Lagrangians - The Polynomial Case -Milan Milošević
 
Physics formula list
Physics formula listPhysics formula list
Physics formula listJSlinkyNY
 
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...Lake Como School of Advanced Studies
 
SPM Physics Formula List Form4
SPM Physics Formula List Form4SPM Physics Formula List Form4
SPM Physics Formula List Form4Zhang Ewe
 
13.1.1 Shm Part 2 Circular To Shm
13.1.1 Shm Part 2 Circular To Shm13.1.1 Shm Part 2 Circular To Shm
13.1.1 Shm Part 2 Circular To ShmChris Staines
 
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...Lake Como School of Advanced Studies
 
Simple harmonic motion
Simple harmonic motion Simple harmonic motion
Simple harmonic motion Sachin Jangid
 

What's hot (20)

Wave Motion Theory Part1
Wave Motion Theory Part1Wave Motion Theory Part1
Wave Motion Theory Part1
 
Lecture 1 sapienza 2017
Lecture 1 sapienza 2017Lecture 1 sapienza 2017
Lecture 1 sapienza 2017
 
Fluid dynamics
Fluid dynamicsFluid dynamics
Fluid dynamics
 
Wave force
Wave forceWave force
Wave force
 
Grlweap frank rausche
Grlweap  frank rauscheGrlweap  frank rausche
Grlweap frank rausche
 
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
 
Tachyon inflation in DBI and RSII context
Tachyon inflation in DBI and RSII contextTachyon inflation in DBI and RSII context
Tachyon inflation in DBI and RSII context
 
Note chapter9 0708-edit-1
Note chapter9 0708-edit-1Note chapter9 0708-edit-1
Note chapter9 0708-edit-1
 
Fundamentals of Seismic Refraction
Fundamentals of Seismic RefractionFundamentals of Seismic Refraction
Fundamentals of Seismic Refraction
 
Tachyonic and Localy Equivalent Canonical Lagrangians - The Polynomial Case -
Tachyonic and Localy Equivalent Canonical Lagrangians - The Polynomial Case -Tachyonic and Localy Equivalent Canonical Lagrangians - The Polynomial Case -
Tachyonic and Localy Equivalent Canonical Lagrangians - The Polynomial Case -
 
Physics formula list
Physics formula listPhysics formula list
Physics formula list
 
Soil Dynamics
Soil DynamicsSoil Dynamics
Soil Dynamics
 
1 relativity
1 relativity1 relativity
1 relativity
 
ENG1040 Lec06
ENG1040 Lec06ENG1040 Lec06
ENG1040 Lec06
 
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
 
SPM Physics Formula List Form4
SPM Physics Formula List Form4SPM Physics Formula List Form4
SPM Physics Formula List Form4
 
13.1.1 Shm Part 2 Circular To Shm
13.1.1 Shm Part 2 Circular To Shm13.1.1 Shm Part 2 Circular To Shm
13.1.1 Shm Part 2 Circular To Shm
 
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
The Analytical/Numerical Relativity Interface behind Gravitational Waves: Lec...
 
Simple harmonic motion
Simple harmonic motion Simple harmonic motion
Simple harmonic motion
 
on Thomesn's strange anisotropy parameter
on Thomesn's strange anisotropy parameteron Thomesn's strange anisotropy parameter
on Thomesn's strange anisotropy parameter
 

Similar to Dynamics and global stability of three-dimensional flows

ETC14: Dynamics of a boundary layer flow over a cylindrical rugosity
ETC14: Dynamics of a boundary layer flow over a cylindrical rugosityETC14: Dynamics of a boundary layer flow over a cylindrical rugosity
ETC14: Dynamics of a boundary layer flow over a cylindrical rugosityJean-Christophe Loiseau
 
ICOMASEF 2013: Influence of the shape on the roughness-induced transition
ICOMASEF 2013: Influence of the shape on the roughness-induced transitionICOMASEF 2013: Influence of the shape on the roughness-induced transition
ICOMASEF 2013: Influence of the shape on the roughness-induced transitionJean-Christophe Loiseau
 
W. t. koiter’s elastic stability of
W. t. koiter’s elastic stability ofW. t. koiter’s elastic stability of
W. t. koiter’s elastic stability ofAbdollah Ghavami
 
DSD-INT - SWAN Advanced Course - 04 - Numerics in SWAN
DSD-INT - SWAN Advanced Course - 04 - Numerics in SWANDSD-INT - SWAN Advanced Course - 04 - Numerics in SWAN
DSD-INT - SWAN Advanced Course - 04 - Numerics in SWANDeltares
 
Part 1 Last weeks summary (1).pdf
Part 1 Last weeks summary (1).pdfPart 1 Last weeks summary (1).pdf
Part 1 Last weeks summary (1).pdfSajawalNawaz5
 
Cfx12 09 turbulence_printing
Cfx12 09 turbulence_printingCfx12 09 turbulence_printing
Cfx12 09 turbulence_printingMarcushuynh66
 
CE 72.32 (January 2016 Semester) Lecture 3 - Design Criteria
CE 72.32 (January 2016 Semester) Lecture 3 - Design Criteria CE 72.32 (January 2016 Semester) Lecture 3 - Design Criteria
CE 72.32 (January 2016 Semester) Lecture 3 - Design Criteria Fawad Najam
 
Cfd notes 1
Cfd notes 1Cfd notes 1
Cfd notes 1sach437
 
Part 2 Boundary Conditions.pdf
Part 2 Boundary Conditions.pdfPart 2 Boundary Conditions.pdf
Part 2 Boundary Conditions.pdfSajawalNawaz5
 
Vol. 1 (1), 2014, 7–11
Vol. 1 (1), 2014, 7–11Vol. 1 (1), 2014, 7–11
Vol. 1 (1), 2014, 7–11Said Benramache
 
DYNAMIC RESPONSE OF SIMPLE SUPPORTED BEAM VIBRATED UNDER MOVING LOAD
DYNAMIC RESPONSE OF SIMPLE SUPPORTED BEAM VIBRATED UNDER MOVING LOAD DYNAMIC RESPONSE OF SIMPLE SUPPORTED BEAM VIBRATED UNDER MOVING LOAD
DYNAMIC RESPONSE OF SIMPLE SUPPORTED BEAM VIBRATED UNDER MOVING LOAD sadiq emad
 
Abaqus CFD-Sample Problems
Abaqus CFD-Sample ProblemsAbaqus CFD-Sample Problems
Abaqus CFD-Sample ProblemsAndrea Porter
 

Similar to Dynamics and global stability of three-dimensional flows (20)

ETC14: Dynamics of a boundary layer flow over a cylindrical rugosity
ETC14: Dynamics of a boundary layer flow over a cylindrical rugosityETC14: Dynamics of a boundary layer flow over a cylindrical rugosity
ETC14: Dynamics of a boundary layer flow over a cylindrical rugosity
 
CFD Course
CFD CourseCFD Course
CFD Course
 
Cfx12 09 turbulence
Cfx12 09 turbulenceCfx12 09 turbulence
Cfx12 09 turbulence
 
ICOMASEF 2013: Influence of the shape on the roughness-induced transition
ICOMASEF 2013: Influence of the shape on the roughness-induced transitionICOMASEF 2013: Influence of the shape on the roughness-induced transition
ICOMASEF 2013: Influence of the shape on the roughness-induced transition
 
W. t. koiter’s elastic stability of
W. t. koiter’s elastic stability ofW. t. koiter’s elastic stability of
W. t. koiter’s elastic stability of
 
Presnt3
Presnt3Presnt3
Presnt3
 
DSD-INT - SWAN Advanced Course - 04 - Numerics in SWAN
DSD-INT - SWAN Advanced Course - 04 - Numerics in SWANDSD-INT - SWAN Advanced Course - 04 - Numerics in SWAN
DSD-INT - SWAN Advanced Course - 04 - Numerics in SWAN
 
GDRR Opening Workshop - Variance Reduction for Reliability Assessment with St...
GDRR Opening Workshop - Variance Reduction for Reliability Assessment with St...GDRR Opening Workshop - Variance Reduction for Reliability Assessment with St...
GDRR Opening Workshop - Variance Reduction for Reliability Assessment with St...
 
Part 1 Last weeks summary (1).pdf
Part 1 Last weeks summary (1).pdfPart 1 Last weeks summary (1).pdf
Part 1 Last weeks summary (1).pdf
 
Cfx12 09 turbulence_printing
Cfx12 09 turbulence_printingCfx12 09 turbulence_printing
Cfx12 09 turbulence_printing
 
CE 72.32 (January 2016 Semester) Lecture 3 - Design Criteria
CE 72.32 (January 2016 Semester) Lecture 3 - Design Criteria CE 72.32 (January 2016 Semester) Lecture 3 - Design Criteria
CE 72.32 (January 2016 Semester) Lecture 3 - Design Criteria
 
Hydrological Modelling of Slope Stability
Hydrological Modelling of Slope StabilityHydrological Modelling of Slope Stability
Hydrological Modelling of Slope Stability
 
10 rans
10 rans10 rans
10 rans
 
Cfd notes 1
Cfd notes 1Cfd notes 1
Cfd notes 1
 
Fish Propulsion
Fish PropulsionFish Propulsion
Fish Propulsion
 
Part 2 Boundary Conditions.pdf
Part 2 Boundary Conditions.pdfPart 2 Boundary Conditions.pdf
Part 2 Boundary Conditions.pdf
 
Vol. 1 (1), 2014, 7–11
Vol. 1 (1), 2014, 7–11Vol. 1 (1), 2014, 7–11
Vol. 1 (1), 2014, 7–11
 
Momentum theory
Momentum theoryMomentum theory
Momentum theory
 
DYNAMIC RESPONSE OF SIMPLE SUPPORTED BEAM VIBRATED UNDER MOVING LOAD
DYNAMIC RESPONSE OF SIMPLE SUPPORTED BEAM VIBRATED UNDER MOVING LOAD DYNAMIC RESPONSE OF SIMPLE SUPPORTED BEAM VIBRATED UNDER MOVING LOAD
DYNAMIC RESPONSE OF SIMPLE SUPPORTED BEAM VIBRATED UNDER MOVING LOAD
 
Abaqus CFD-Sample Problems
Abaqus CFD-Sample ProblemsAbaqus CFD-Sample Problems
Abaqus CFD-Sample Problems
 

Recently uploaded

(9818099198) Call Girls In Noida Sector 14 (NOIDA ESCORTS)
(9818099198) Call Girls In Noida Sector 14 (NOIDA ESCORTS)(9818099198) Call Girls In Noida Sector 14 (NOIDA ESCORTS)
(9818099198) Call Girls In Noida Sector 14 (NOIDA ESCORTS)riyaescorts54
 
Neurodevelopmental disorders according to the dsm 5 tr
Neurodevelopmental disorders according to the dsm 5 trNeurodevelopmental disorders according to the dsm 5 tr
Neurodevelopmental disorders according to the dsm 5 trssuser06f238
 
Dubai Calls Girl Lisa O525547819 Lexi Call Girls In Dubai
Dubai Calls Girl Lisa O525547819 Lexi Call Girls In DubaiDubai Calls Girl Lisa O525547819 Lexi Call Girls In Dubai
Dubai Calls Girl Lisa O525547819 Lexi Call Girls In Dubaikojalkojal131
 
BUMI DAN ANTARIKSA PROJEK IPAS SMK KELAS X.pdf
BUMI DAN ANTARIKSA PROJEK IPAS SMK KELAS X.pdfBUMI DAN ANTARIKSA PROJEK IPAS SMK KELAS X.pdf
BUMI DAN ANTARIKSA PROJEK IPAS SMK KELAS X.pdfWildaNurAmalia2
 
STOPPED FLOW METHOD & APPLICATION MURUGAVENI B.pptx
STOPPED FLOW METHOD & APPLICATION MURUGAVENI B.pptxSTOPPED FLOW METHOD & APPLICATION MURUGAVENI B.pptx
STOPPED FLOW METHOD & APPLICATION MURUGAVENI B.pptxMurugaveni B
 
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝soniya singh
 
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptxRESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptxFarihaAbdulRasheed
 
Vision and reflection on Mining Software Repositories research in 2024
Vision and reflection on Mining Software Repositories research in 2024Vision and reflection on Mining Software Repositories research in 2024
Vision and reflection on Mining Software Repositories research in 2024AyushiRastogi48
 
Functional group interconversions(oxidation reduction)
Functional group interconversions(oxidation reduction)Functional group interconversions(oxidation reduction)
Functional group interconversions(oxidation reduction)itwameryclare
 
《Queensland毕业文凭-昆士兰大学毕业证成绩单》
《Queensland毕业文凭-昆士兰大学毕业证成绩单》《Queensland毕业文凭-昆士兰大学毕业证成绩单》
《Queensland毕业文凭-昆士兰大学毕业证成绩单》rnrncn29
 
Call Girls in Majnu Ka Tilla Delhi 🔝9711014705🔝 Genuine
Call Girls in Majnu Ka Tilla Delhi 🔝9711014705🔝 GenuineCall Girls in Majnu Ka Tilla Delhi 🔝9711014705🔝 Genuine
Call Girls in Majnu Ka Tilla Delhi 🔝9711014705🔝 Genuinethapagita
 
Environmental Biotechnology Topic:- Microbial Biosensor
Environmental Biotechnology Topic:- Microbial BiosensorEnvironmental Biotechnology Topic:- Microbial Biosensor
Environmental Biotechnology Topic:- Microbial Biosensorsonawaneprad
 
Fertilization: Sperm and the egg—collectively called the gametes—fuse togethe...
Fertilization: Sperm and the egg—collectively called the gametes—fuse togethe...Fertilization: Sperm and the egg—collectively called the gametes—fuse togethe...
Fertilization: Sperm and the egg—collectively called the gametes—fuse togethe...D. B. S. College Kanpur
 
Pests of Bengal gram_Identification_Dr.UPR.pdf
Pests of Bengal gram_Identification_Dr.UPR.pdfPests of Bengal gram_Identification_Dr.UPR.pdf
Pests of Bengal gram_Identification_Dr.UPR.pdfPirithiRaju
 
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptxLIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptxmalonesandreagweneth
 
Four Spheres of the Earth Presentation.ppt
Four Spheres of the Earth Presentation.pptFour Spheres of the Earth Presentation.ppt
Four Spheres of the Earth Presentation.pptJoemSTuliba
 
The dark energy paradox leads to a new structure of spacetime.pptx
The dark energy paradox leads to a new structure of spacetime.pptxThe dark energy paradox leads to a new structure of spacetime.pptx
The dark energy paradox leads to a new structure of spacetime.pptxEran Akiva Sinbar
 
User Guide: Orion™ Weather Station (Columbia Weather Systems)
User Guide: Orion™ Weather Station (Columbia Weather Systems)User Guide: Orion™ Weather Station (Columbia Weather Systems)
User Guide: Orion™ Weather Station (Columbia Weather Systems)Columbia Weather Systems
 
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfBehavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfSELF-EXPLANATORY
 

Recently uploaded (20)

(9818099198) Call Girls In Noida Sector 14 (NOIDA ESCORTS)
(9818099198) Call Girls In Noida Sector 14 (NOIDA ESCORTS)(9818099198) Call Girls In Noida Sector 14 (NOIDA ESCORTS)
(9818099198) Call Girls In Noida Sector 14 (NOIDA ESCORTS)
 
Neurodevelopmental disorders according to the dsm 5 tr
Neurodevelopmental disorders according to the dsm 5 trNeurodevelopmental disorders according to the dsm 5 tr
Neurodevelopmental disorders according to the dsm 5 tr
 
Dubai Calls Girl Lisa O525547819 Lexi Call Girls In Dubai
Dubai Calls Girl Lisa O525547819 Lexi Call Girls In DubaiDubai Calls Girl Lisa O525547819 Lexi Call Girls In Dubai
Dubai Calls Girl Lisa O525547819 Lexi Call Girls In Dubai
 
BUMI DAN ANTARIKSA PROJEK IPAS SMK KELAS X.pdf
BUMI DAN ANTARIKSA PROJEK IPAS SMK KELAS X.pdfBUMI DAN ANTARIKSA PROJEK IPAS SMK KELAS X.pdf
BUMI DAN ANTARIKSA PROJEK IPAS SMK KELAS X.pdf
 
STOPPED FLOW METHOD & APPLICATION MURUGAVENI B.pptx
STOPPED FLOW METHOD & APPLICATION MURUGAVENI B.pptxSTOPPED FLOW METHOD & APPLICATION MURUGAVENI B.pptx
STOPPED FLOW METHOD & APPLICATION MURUGAVENI B.pptx
 
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
 
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptxRESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
RESPIRATORY ADAPTATIONS TO HYPOXIA IN HUMNAS.pptx
 
Vision and reflection on Mining Software Repositories research in 2024
Vision and reflection on Mining Software Repositories research in 2024Vision and reflection on Mining Software Repositories research in 2024
Vision and reflection on Mining Software Repositories research in 2024
 
Functional group interconversions(oxidation reduction)
Functional group interconversions(oxidation reduction)Functional group interconversions(oxidation reduction)
Functional group interconversions(oxidation reduction)
 
《Queensland毕业文凭-昆士兰大学毕业证成绩单》
《Queensland毕业文凭-昆士兰大学毕业证成绩单》《Queensland毕业文凭-昆士兰大学毕业证成绩单》
《Queensland毕业文凭-昆士兰大学毕业证成绩单》
 
Call Girls in Majnu Ka Tilla Delhi 🔝9711014705🔝 Genuine
Call Girls in Majnu Ka Tilla Delhi 🔝9711014705🔝 GenuineCall Girls in Majnu Ka Tilla Delhi 🔝9711014705🔝 Genuine
Call Girls in Majnu Ka Tilla Delhi 🔝9711014705🔝 Genuine
 
Environmental Biotechnology Topic:- Microbial Biosensor
Environmental Biotechnology Topic:- Microbial BiosensorEnvironmental Biotechnology Topic:- Microbial Biosensor
Environmental Biotechnology Topic:- Microbial Biosensor
 
Fertilization: Sperm and the egg—collectively called the gametes—fuse togethe...
Fertilization: Sperm and the egg—collectively called the gametes—fuse togethe...Fertilization: Sperm and the egg—collectively called the gametes—fuse togethe...
Fertilization: Sperm and the egg—collectively called the gametes—fuse togethe...
 
Hot Sexy call girls in Moti Nagar,🔝 9953056974 🔝 escort Service
Hot Sexy call girls in  Moti Nagar,🔝 9953056974 🔝 escort ServiceHot Sexy call girls in  Moti Nagar,🔝 9953056974 🔝 escort Service
Hot Sexy call girls in Moti Nagar,🔝 9953056974 🔝 escort Service
 
Pests of Bengal gram_Identification_Dr.UPR.pdf
Pests of Bengal gram_Identification_Dr.UPR.pdfPests of Bengal gram_Identification_Dr.UPR.pdf
Pests of Bengal gram_Identification_Dr.UPR.pdf
 
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptxLIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
LIGHT-PHENOMENA-BY-CABUALDIONALDOPANOGANCADIENTE-CONDEZA (1).pptx
 
Four Spheres of the Earth Presentation.ppt
Four Spheres of the Earth Presentation.pptFour Spheres of the Earth Presentation.ppt
Four Spheres of the Earth Presentation.ppt
 
The dark energy paradox leads to a new structure of spacetime.pptx
The dark energy paradox leads to a new structure of spacetime.pptxThe dark energy paradox leads to a new structure of spacetime.pptx
The dark energy paradox leads to a new structure of spacetime.pptx
 
User Guide: Orion™ Weather Station (Columbia Weather Systems)
User Guide: Orion™ Weather Station (Columbia Weather Systems)User Guide: Orion™ Weather Station (Columbia Weather Systems)
User Guide: Orion™ Weather Station (Columbia Weather Systems)
 
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdfBehavioral Disorder: Schizophrenia & it's Case Study.pdf
Behavioral Disorder: Schizophrenia & it's Case Study.pdf
 

Dynamics and global stability of three-dimensional flows

  • 1. Introduction Global stability theory Roughness-induced transition Conclusion Dynamics and global stability analysis of three-dimensional flows Jean-Christophe Loiseau1,2 supervisor: Jean-Christophe Robinet1 co-supervisor: Emmanuel Leriche2 (1): DynFluid Laboratory - Arts & M´etiers-ParisTech - 75013 Paris, France (2): LML - University of Lille 1 - 59655 Villeneuve d’Ascq, France PhD Defence, May 26th 2014 1/79
  • 2. Introduction Global stability theory Roughness-induced transition Conclusion What are hydrodynamic instabilities? • Let us consider the flow of water (ν = 15.10−6 m2.s−1) past a two-dimensional cylinder of diameter D = 1.5 cm. • If water flows from left to right at U = 4.5 cm.s−1 (Re = 45), nothing really fancy takes place: the flow is steady and stable. 2/79
  • 3. Introduction Global stability theory Roughness-induced transition Conclusion What are hydrodynamic instabilities? • If you increase the velocity to U = 5 cm.s−1 (Re = 50), the flow looks very different. • The steady flow became (globally) unstable and has experienced a (supercritical) bifurcation. 3/79
  • 4. Introduction Global stability theory Roughness-induced transition Conclusion How do we study these instabilities? • Let us consider a non-linear dynamical system B ∂Q ∂t = F(Q) (1) 1. Compute a fixed point (or base flow): F(Qb) = 0 2. Linearise the dynamics of an infinitesimal perturbation q in the vicinity of this solution: B ∂q ∂t = Jq with J = ∂F ∂q (2) 3. Investigate the stability properties of this linear dynamical system. 4/79
  • 5. Introduction Global stability theory Roughness-induced transition Conclusion How do we study these instabilities? • In the context of fluid dynamics, this includes several different approaches depending on the nature of the base flow: • Local stability analysis for parallel flows: ֒→ Temporal stability, Spatial stability, Absolute/Convective stability, Response to harmoning forcing, Transient growth • Global stability analysis for two-dimensional and three-dimensional flows: ֒→ Temporal stability, Response to harmoning forcing (Resolvent), Transient growth 5/79
  • 6. Introduction Global stability theory Roughness-induced transition Conclusion Local stability analysis • The base flow depends on a single space coordinate: Ub = (Ub(y), 0, 0)T • Linear dynamical system (2) is now autonomous in time and in the x and z coordinates of space. ֒→ The perturbation q can be decomposed into normal modes: q(x, y, z, t) = ˆq(y) exp(iαx + iβz + λt) + c.c with λ = σ + iω 6/79
  • 7. Introduction Global stability theory Roughness-induced transition Conclusion Local stability analysis • Introducing such decomposition into the system (2) yields to a generalised eigenvalue problem: λBˆq = J(y, α, β)ˆq (3) • The stability of the base flow Ub is governed by the growth rate σ: ֒→ If σ < 0, the base flow is said to be locally stable. ֒→ If σ > 0, the base flow is said to be locally unstable. 7/79
  • 8. Introduction Global stability theory Roughness-induced transition Conclusion Local stability analysis Theoretical point of view • Relies on the parallel flow assumption. • Provides insights into the local stability properties of the flow. ֒→ Requires a good theoretical and mathematical background. Practical point of view • The generalised eigenproblem involves small matrices (∼ 100 × 100) • Can be solved using direct eigenvalue solvers in a matter of seconds even on a 10 years old laptop. 8/79
  • 9. Introduction Global stability theory Roughness-induced transition Conclusion Global stability theory • The base flow has two components both depending on the x and y space coordinates: Ub = (Ub(x, y), Vb(x, y), 0)T • Linear dynamical system (2) is now only autonomous in time and in z. ֒→ The perturbation q can be decomposed into normal modes: q(x, y, z, t) = ˆq(x, y) exp(iβz + λt) + c.c with λ = σ + iω Base flow of the 2D separated boundary layer at Re = 600 as in Ehrenstein & Gallaire (2008). 9/79
  • 10. Introduction Global stability theory Roughness-induced transition Conclusion Global stability theory • Introducing such decomposition into the system (2) yields to a generalised eigenvalue problem once again: λBˆq = J(x, y, β)ˆq (4) • The stability of the base flow Ub is governed by the growth rate σ: ֒→ If σ < 0, the base flow is said to be globally stable. ֒→ If σ > 0, the base flow is said to be globally unstable. Streamwise velocity component of the leading unstable global mode for the 2D separated boundary layer. 10/79
  • 11. Introduction Global stability theory Roughness-induced transition Conclusion Global stability analysis Theoretical point of view • Got rid of the parallel flow assumption. • Allows to investigate more realistic configurations as separated flows very common in Nature and industries. Practical point of view • The generalised eigenproblem involves relatively large matrices (∼ 105 × 105) • Mostly solved using iterative eigenvalue solvers on large workstations. 11/79
  • 12. Introduction Global stability theory Roughness-induced transition Conclusion Objectives • Bagheri et al. (2008) and Ilak et al. (2012) performed the first global stability analysis ever on a 3D flow (jet in crossflow). • Extension of the global stability tools to a fully three-dimensional framework. ֒→ Mostly a numerical problem due to the (extremely) large matrices involved. 12/79
  • 13. Introduction Global stability theory Roughness-induced transition Conclusion Objectives λ2 visualisation of the hairpin vortices shed behind a hemispherical roughness element. Courtesy of P. Fischer. 13/79
  • 14. Introduction Global stability theory Roughness-induced transition Conclusion Context • PhD thesis part of a larger project: Simulation and Control of Geometrically Induced Flows (SICOGIF) ֒→ Funded by the French National Agency for Research (ANR) ֒→ Involves several different parties (IRPHE, EPFL, Arts et M´etiers ParisTech and Universit´e Lille-1) ֒→ Aims at improving our understanding of instability and transition in complex 2D and 3D separated flows both from an experimental and numerical point of view. 14/79
  • 15. Introduction Global stability theory Roughness-induced transition Conclusion Context • Three flow configurations have been investigated: ֒→ The lid-driven cavity flow ֒→ The asymmetric stenotic pipe flow ֒→ The roughness-induced boundary layer flow Vertical velocity component of the leading global mode for a LDC having a spanwise extent Λ = 6 at Re = 900. 15/79
  • 16. Introduction Global stability theory Roughness-induced transition Conclusion Context • Three flow configurations have been investigated: ֒→ The lid-driven cavity flow ֒→ The asymmetric stenotic pipe flow ֒→ The roughness-induced boundary layer flow Streamwise velocity component for the two existing steady states of an asymmetric stenotic pipe flow at Re = 400. 16/79
  • 17. Introduction Global stability theory Roughness-induced transition Conclusion Context • Three flow configurations have been investigated: ֒→ The lid-driven cavity flow ֒→ The asymmetric stenotic pipe flow ֒→ The roughness-induced boundary layer flow Identification of the vortical structures by the λ2 criterion (Jeong & Hussain 1995). 17/79
  • 18. Introduction Global stability theory Roughness-induced transition Conclusion Introduction Global stability theory and algorithm Base flows Global stability theory How to solve the eigenvalue problem? Roughness-induced transition Motivations Fransson 2005 experiment Parametric investigation Physical analysis Non-linear evolution Conclusions & Perspectives Conclusions LDC & Stenosis Perspectives 18/79
  • 19. Introduction Global stability theory Roughness-induced transition Conclusion Global stability analysis of three-dimensional flows 18/79
  • 20. Introduction Global stability theory Roughness-induced transition Conclusion How to compute base flows? • Base flow are given by: F(Qb) = 0 (5) • Various techniques can be employed to compute these peculiar solutions: ֒→ Analytical solutions, impose appropriate symmetries, Newton and quasi-Newton methods, ... • In the present work, we use the Selective frequency damping approach (see Akervik et al. 2006). 19/79
  • 21. Introduction Global stability theory Roughness-induced transition Conclusion Selective frequency damping • Enables the stabilisation of the solution by applying a low-pass filter to the Navier-Stokes equations. ֒→ A forcing term is added to the r.h.s of the equations. ֒→ The system is extended with an equation for the filtered state.    ∂Q ∂t = F(Q) + χ(Q − ¯Q) ∂ ¯Q ∂t = ωc(Q − ¯Q) (6) • The cutoff frequency ωc is connected to the frequency of the most dominant instabilities and should be smaller than this frequency (ωc < ω). • The gain χ needs to be large enough to stabilise the system (χ > σ). 20/79
  • 22. Introduction Global stability theory Roughness-induced transition Conclusion Selective frequency damping Pros → Really easy to implement within an existing DNS code. → Memory footprint similar to that of a simple direct numerical simulation. → Easy to use/tune the low-pass filter. Cons → As time-consuming as a direct numerical simulation. → Requires a priori information regarding the instability of the flow. → Unable to stabilise the system if the instability is non-oscillating. 21/79
  • 23. Introduction Global stability theory Roughness-induced transition Conclusion Global stability theory • Dynamics of a three-dimensional infinitesimal perturbation q = (u, p)T evolving onto the base flow Qb = (Ub, Pb)T are governed by:    ∂u ∂t = −(u · ∇)Ub − (Ub · ∇)u − ∇p + 1 Re ∆u ∇ · u = 0 (7) • If projected onto a divergence-free vector space, this set of equations can be recast into: ∂u ∂t = Au (8) with A the (projected) Jacobian matrix of the Navier-Stokes equations. 22/79
  • 24. Introduction Global stability theory Roughness-induced transition Conclusion Global stability theory • Using a normal mode decomposition u(x, y, z, t) = ˆu(x, y, z)e(σ+iω)t + c.c • System (8) can be formulated as an eigenvalue problem (σ + iω)ˆu = Aˆu (9) • The sign of σ determines the stability of the base flow Ub: ֒→ If σ < 0, the base flow is said to be asymptoticaly linearly stable. ֒→ If σ > 0, the base flow is said to be asymptoticaly linearly unstable. • ω determines whether the instability is oscillatory (ω = 0) or not (ω = 0). 23/79
  • 25. Introduction Global stability theory Roughness-induced transition Conclusion How to solve the eigenvalue problem? • Depends on the dimension of the discretised problem. Base Flow Inhomogeneous Dimension Storage direction(s) of ˆu of A Poiseuille U(y) 1D 102 ∼ 1 Mb 2D bump U(x, y) 2D 105 ∼ 1-50 Gb 3D bump U(x, y, z) 3D 107 ∼ 1-100 Tb • For 3D global stability problem, A is so large that it cannot be explicitely constructed. Matrix-free approach is mandatory! 24/79
  • 26. Introduction Global stability theory Roughness-induced transition Conclusion Time-stepping approach • Time-stepping approach (Edwards et al. 1994, Bagheri et al. 2008) is based on the formal solution to system (8): u(∆t) = eA∆t u0 • The operator M(∆t) = eA∆t is nothing but a matrix. Its application on u0 can be computed by time-marching the linearised Navier-Stokes equations. ֒→ Its stability properties can be investigated by eigenvalue analysis. MˆU = ˆUΣ (10) with ˆU the matrix of eigenvectors and Σ the eigenvalue matrix of M = eA∆t. 25/79
  • 27. Introduction Global stability theory Roughness-induced transition Conclusion Arnoldi algorithm • The eigenvalue problem (10) is solved using an Arnoldi algorithm. 1. Given M and u0, construct a small Krylov subspace (compared to the size of the initial problem), Km(M, u0) = span u0, Mu0, M2 u0, · · · , M(m−1) u0 2. Orthonormalize: U = [U1, · · · , Um] 3. Project operator M ≈ UHUT −→ MUk = Uk Hk + rk eT k with Hk : upper Hessenberg matrix. 4. Solve small eigenvalue problem (ΣH , X): HX = XΣH, (m × m), m < 1000 5. Link with the initial eigenproblem (ΛA, ˆu): ΛA = log(ΣH) ∆t , ˆu = UX 26/79
  • 28. Introduction Global stability theory Roughness-induced transition Conclusion Arnoldi algorithm 27/79
  • 29. Introduction Global stability theory Roughness-induced transition Conclusion Summary • All calculations have been performed with the code Nek 5000 ֒→ Legendre spectral elements code developed by P. Fischer at Argonne National Laboratory. ֒→ Semi-implicit temporal scheme. ֒→ Massively parallel code based on an MPI strategy. • Base flow computation ֒→ Selective frequency damping approach : application of a low-pass filter to the fully non-linear Navier-Stokes equations (Akervik et al. 2006). • Global stability analysis ֒→ Arnoldi algorithm similar to the one published by Barkley et al. (2008). 28/79
  • 30. Introduction Global stability theory Roughness-induced transition Conclusion Roughness-induced transition 28/79
  • 31. Introduction Global stability theory Roughness-induced transition Conclusion Motivations • Roughness elements have numerous applications in aerospace engineering: ֒→ Stabilisation of the Tollmien-Schlichting waves, ֒→ Shift and/or control of the transition location, ... • Their influence on the flow has been extensively investigated since the early 1950’s. Experimental visualisation of the flow induced by a roughness element. Gregory & Walker, 1956. 29/79
  • 32. Introduction Global stability theory Roughness-induced transition Conclusion Delay of the natural transition • Cossu & Brandt (2004): Theoretical prediction of the stabilisation of TS waves by streamwise streaks. • Fransson et al. (2004-2006): Experimental demonstration using a periodic array of roughness elements. Schematic setup Experimental observations Figures from Fransson et al. (2006). 30/79
  • 33. Introduction Global stability theory Roughness-induced transition Conclusion Roughness-induced transition • Problem: If the Reynolds number is too high, transition occurs right downstream the roughness elements! Illustration of the early roughness-induced transition. λ2 visualisation of the vortical structures. 31/79
  • 34. Introduction Global stability theory Roughness-induced transition Conclusion Roughness-induced transition • Since the early 1950’s: Numerous experimental investigations. ֒→ Transition diagram by von Doenhoff & Braslow (1961). • Despite the large body of literature, the underlying mechanisms are not yet fully understood. Transition diagram from von Doenhoff & Braslow (1961). 32/79
  • 35. Introduction Global stability theory Roughness-induced transition Conclusion Motivations • Methods used until now rely on a parallel flow assumption: ֒→ Local stability theory (Brandt 2006, Denissen & White 2013, ...), ֒→ Local transient growth theory (Vermeersch 2010, ...) • Objective: ֒→ Might a 3D global instability of the flow explain the roughness-induced transition? ֒→ If so, what are the underlying physical mechanisms? • Methods: ֒→ Fully three-dimensional global stability analyses, ֒→ Direct numerical simulations, ֒→ Comparison with available experimental data. 33/79
  • 36. Introduction Global stability theory Roughness-induced transition Conclusion Problem formulation z X y d h Lz l Lx Ly δ 0 Sketch of the computational arrangement and various scales used for DNS and stability analysis. - (Lx , Ly , Lz ) = (105, 50, 8η) - η = d/h = 1, 2, 3 - Re = Ueh/ν - Reδ∗ = Ueδ∗/ν - Inflow: Blasius profile, - Outflow: ∇U · x = 0, - Top: U = 1, ∂y V = ∂y W = 0, - Wall: no-slip B.C. - Lateral: periodic B.C. 34/79
  • 37. Introduction Global stability theory Roughness-induced transition Conclusion Numerical informations η Number of SEM Gridpoints (N = 6-12) Number of cores used 1 10 000 2-17.106 256 2 17 500 3.5-30.106 512 3 20 000 4.5-35.106 512 Typical size of the numerical problem investigated. N is the order of the Legendre polynomials used in the three directions within each element. Typical SEM distribution in a given horizontal plane. Full mesh with Legendre polynomials of order N = 8. 35/79
  • 38. Introduction Global stability theory Roughness-induced transition Conclusion Roughness-induced transition The Fransson 2005 experiment 35/79
  • 39. Introduction Global stability theory Roughness-induced transition Conclusion Experimental setup • Experimental demonstration of the ability for finite amplitude streaks to stabilise TS waves. • Unfortunately, transition takes place right downstream the array of roughness elements if the Reynolds number is too high. h D η Lz/h xk/h Rec δ∗ 1.4mm 4.2mm 3 10 57.14 ≃ 290 36/79
  • 40. Introduction Global stability theory Roughness-induced transition Conclusion Base flow (a) (b) • Upstream and downstream reversed flow regions: ֒→ Induces a central low-speed region. • Vortical system stemming: ֒→ Investigated by Baker (1978) ֒→ Horseshoe vortices whose legs are streamwise oriented counter-rotating vortices. ֒→ Creation of streamwise velocity streaks (lift-up effect) 37/79
  • 41. Introduction Global stability theory Roughness-induced transition Conclusion Base flow (a) X=20 (b) X=40 (c) X=60 (d) X=80 Visualisation of the base flow deviation from the Blasius boundary layer flow in various streamwise planes for Re = 466. High speed streaks are in red while low-speed ones are in blue. • Low-speed region generated by the roughness element’s blockage. ֒→ Fades away quite rapidly in the streamwise direction. • High- and low-speed streaks on each side of the roughness element due to the horshoe vortex. ֒→ Sustains over quite a long streamwise distance. 38/79
  • 42. Introduction Global stability theory Roughness-induced transition Conclusion Global stability Eigenspectrum of the linearised Navier-Stokes operator. • Hopf bifurcation taking place in-between 550 < Rec < 575. ֒→ Linear interpolation: Rec = 564, i.e. Rec δ∗ = 309. 39/79
  • 43. Introduction Global stability theory Roughness-induced transition Conclusion Global stability (a) Top view of u = ±10% iso-surfaces (b) X = 23 (c) X = 40 Visualisation of streamwise velocity component of the leading unstable mode for Re = 575. 40/79
  • 44. Introduction Global stability theory Roughness-induced transition Conclusion Global stability • Leading unstable mode exhibits a varicose symmetry: ֒→ Streamwise alternated patches of positive and negative velocity mostly localised along the central low-speed region. ֒→ Non-linear DNS have revealed that it gives birth to hairpin vortices. • Rec predicted by global stability analysis only 6% larger than the experimental one from Fransson et al. (2005): ֒→ Global instability of the flow appears as one of the possible explanations to roughness-induced transition. 41/79
  • 45. Introduction Global stability theory Roughness-induced transition Conclusion Roughness-induced transition Parametric investigation 41/79
  • 46. Introduction Global stability theory Roughness-induced transition Conclusion Parametric investigation • Aims of the parametric investigation: ֒→ How do the Reynolds number and the aspect ratio of the roughness elements impact the base flow and its stability properties? ֒→ Does the leading unstable mode always exhibit a varicose symmetry? • To do so: ֒→ The spanwise extent of the domain is taken large enough so that the roughness element behaves as being isolated. ֒→ δ99/h is set to 2 to isolate the influence of the Reynolds number only. ֒→ The roughness element’s aspect ratio varies from η = 1 up to η = 3. 42/79
  • 47. Introduction Global stability theory Roughness-induced transition Conclusion Base flows (a) (Re, η) = (600, 1) (b) (Re, η) = (1250, 1) Influence of the Reynolds number on the flow within the z = 0 plane. Streamwise velocity iso-contours ranging from U = 0.1 up to U = 0.99. • Influence of the Reynolds number: ֒→ Does not qualitatively change the shape of the downstream reversed flow region. ֒→ Strengthen the gradients and reduces the thickness of the shear layer. 43/79
  • 48. Introduction Global stability theory Roughness-induced transition Conclusion Base flows (a) (Re, η) = (600, 1) (b) (Re, η) = (1250, 1) Top view of the low-speed (white) and high-speed (black) streaks induced by the roughness element. Streak have been identified using the deviation of the base flow streamwise velocity from the Blasius boundary layer flow. • Influence of the Reynolds number: ֒→ Strongly increases the amplitude and the streamwise extent of the central low-speed region. ֒→ Slightly increases the amplitude of the outer velocity streaks. 44/79
  • 49. Introduction Global stability theory Roughness-induced transition Conclusion Base flows (a) (Re, η) = (600, 2) (b) (Re, η) = (600, 3) Influence of the aspect ratio on the flow within the z = 0 plane. Streamwise velocity iso-contours ranging from U = 0.1 up to U = 0.99. • Influence of the aspect ratio: ֒→ Strengthen the gradients and reduces the thickness of the shear layer. ֒→ Strongly increases the amplitude and the streamwise extent of the central low-speed region. ֒→ Strongly increases the amplitude of the outer velocity streaks. 45/79
  • 50. Introduction Global stability theory Roughness-induced transition Conclusion Base flows (a) (Re, η) = (600, 2) (b) (Re, η) = (600, 3) Top view of the low-speed (white) and high-speed (black) streaks induced by the roughness element. Streak have been identified using the deviation of the base flow streamwise velocity from the Blasius boundary layer flow. 46/79
  • 51. Introduction Global stability theory Roughness-induced transition Conclusion Global stability ω σ 0 0.5 1 1.5 2 -0.04 -0.02 0 0.02 0.04 (a) (Re, η) = (1200, 1) ω σ 0 0.5 1 1.5 -0.04 -0.02 0 0.02 0.04 (b) (Re, η) = (900, 2) ω σ 0 0.5 1 1.5 -0.04 -0.02 0 0.02 0.04 (c) (Re, η) = (700, 3) Eigenspectra of the linearised Navier-Stokes operator for different roughness elements. 47/79
  • 52. Introduction Global stability theory Roughness-induced transition Conclusion Global stability (a) (Re, η) = (1200, 1) (b) (Re, η) = (900, 2) Evolution of the leading unstable mode when the roughness element’s aspect ratio is changed from η = 1 to η = 2. Isosurfaces u = ± 10% of the modes streamwise velocity component. 48/79
  • 53. Introduction Global stability theory Roughness-induced transition Conclusion Global stability (a) (Re, η) = (1200, 1) (b) (Re, η) = (900, 2) Evolution of the leading unstable mode when the roughness element’s aspect ratio is changed from η = 1 to η = 2 in the X = 25 plane. 49/79
  • 54. Introduction Global stability theory Roughness-induced transition Conclusion Global stability • Increasing the roughness element’s aspect ratio decreases the critical Reynolds number. η 1 2 3 Fransson (η = 3) Rec 1040 850 656 564 Rec h 813 630 513 519 Symmetry S V V V Summary of the global stability analyses. V: varicose, S: sinuous. Reh is the roughness Reynolds number. • Exchange of symmetry in qualitative agreements with Sakamoto & Arie (1983) and Beaudoin (2004). 50/79
  • 55. Introduction Global stability theory Roughness-induced transition Conclusion Roughness-induced transition Physical analysis 50/79
  • 56. Introduction Global stability theory Roughness-induced transition Conclusion Physical analysis • Aims of the analysis: ֒→ Unravel the underlying physical mechanisms for each mode. ֒→ How and where do they extract their energy? ֒→ Where do they originate? • Type of analysis: ֒→ Kinetic energy transfer between the base flow and the perturbation (Brandt 2006). ֒→ Computation of the wavemaker region (Giannetti & Luchini 2007). 51/79
  • 57. Introduction Global stability theory Roughness-induced transition Conclusion Kinetic energy budget • The evolution of the perturbation’s kinetic energy is governed by the Reynolds-Orr equation: ∂E ∂t = −D + 9 i=1 V Ii dV (11) • with the total kinetic energy E and dissipation D given by: E = 1 2 V u · u dV , and D = 1 Re V ∇u : ∇u dV (12) 52/79
  • 58. Introduction Global stability theory Roughness-induced transition Conclusion Kinetic energy budget • The integrands Ii representing the different production terms are given by: I1 = −u2 ∂Ub ∂x , I2 = −uv ∂Ub ∂y , I3 = −uw ∂Ub ∂z I4 = −uv ∂Vb ∂x , I5 = −v2 ∂Vb ∂y , I6 = −vw ∂Vb ∂z I7 = −wu ∂Wb ∂x , I8 = −wv ∂Wb ∂y , I9 = −w2 ∂Wb ∂z (13) • Their sign indicates whether the associated local transfer of kinetic energy acts as stabilising (negative) or destabilising (positive). 53/79
  • 59. Introduction Global stability theory Roughness-induced transition Conclusion Kinetic energy budget: Sinuous mode 0 0.5 1 I1 I2 I3 I4 I5 I6 I7 I8 I9 D (a) (Re, η) = (1125, 1) 0 0.5 1 I1 I2 I3 I4 I5 I6 I7 I8 I9 D (b) (Re, η) = (1250, 1) X0 30 60 90 2.0x10 -03 4.0x10 -03 6.0x10 -03 ∫I2dydz ∫I3dydz (c) (Re, η) = (1125, 1) X0 30 60 90 2.0x10 -03 4.0x10 -03 6.0x10 -03 ∫I2dydz ∫I3dydz (d) (Re, η) = (1250, 1) Top: Sinuous unstable mode’s kinetic energy budget integrated over the whole domain. Bottom: Streamwise evolution of the production terms y,z I2 dydz (red dashed line) and y,z I3 dydz (blue solid line). 54/79
  • 60. Introduction Global stability theory Roughness-induced transition Conclusion Kinetic energy budget: Sinuous mode (a) I2 = −uv∂U/∂y (b) I3 = −uw∂U/∂z Spatial distribution of the I2 = −uv∂Ub/∂y (a) and I3 = −uw∂Ub/∂z (b) production terms in the plane x = 25 for (Re, η) = (1125, 1). Solid lines depict the base flows streamwise velocity isocontours from Ub = 0.1 to 0.99, whereas the red dashed lines stand for the location of the shear layer. 55/79
  • 61. Introduction Global stability theory Roughness-induced transition Conclusion Kinetic energy budget: Sinuous mode (a) I2 = −uv∂U/∂y (b) I3 = −uw∂U/∂z Spatial distribution of I2 (c) and I3 (d) in the y = 0.75 horizontal plan. 56/79
  • 62. Introduction Global stability theory Roughness-induced transition Conclusion Kinetic energy budget: Varicose mode 0 1 2 I1 I2 I3 I4 I5 I6 I7 I8 I9 D (a) (Re, η) = (850, 2) 0 1 2 I1 I2 I3 I4 I5 I6 I7 I8 I9 D (b) (Re, η) = (1000, 2) X0 30 60 90 .0x10 +00 4.0x10 -03 8.0x10 -03 ∫I2dydz ∫I3dydz (c) (Re, η) = (850, 2) X0 30 60 90 .0x10 +00 4.0x10 -03 8.0x10 -03 ∫I2dydz ∫I3dydz (d) (Re, η) = (1000, 2) Top: Varicose unstable mode’s kinetic energy budget integrated over the whole domain. Bottom: Streamwise evolution of the production terms y,z I2 dydz (red dashed line) and y,z I3 dydz (blue solid line). 57/79
  • 63. Introduction Global stability theory Roughness-induced transition Conclusion Kinetic energy budget: Varicose mode (a) I2 = −uv∂U/∂y (b) I3 = −uw∂U/∂z Spatial distribution of the I2 = −uv∂Ub/∂y (a) and I3 = −uw∂Ub/∂z (b) production terms in the plane x = 25 for (Re, η) = (850, 2). Solid lines depict the base flows streamwise velocity isocontours from Ub = 0.1 to 0.99, whereas the red dashed lines stand for the location of the shear layer. 58/79
  • 64. Introduction Global stability theory Roughness-induced transition Conclusion Kinetic energy budget: Varicose mode Spatial distribution of the I3 = −uw∂Ub/∂z production term in the plane y = 0.5 for (Re, η) = (850, 2). 59/79
  • 65. Introduction Global stability theory Roughness-induced transition Conclusion Wavemaker • Kinetic energy budgets provide valuable insights into the mode’s dynamics but very limited about its core region, i.e. the wavemaker. • Defined by Giannetti & Luchini (2007) as the overlap of the direct global mode u and its adjoint u†: ζ(x, y, z) = u† u u†, u (14) • Allows the identification of the most likely region for the inception of the global instability under consideration. 60/79
  • 66. Introduction Global stability theory Roughness-induced transition Conclusion Wavemaker Figure: Sinuous wavemaker in the y = 0.75 plane. Figure: Varicose wavemaker in the z = 0 plane. 61/79
  • 67. Introduction Global stability theory Roughness-induced transition Conclusion Wavemaker Figure: Sinuous wavemaker in the y = 0.75 plane. Figure: Varicose wavemaker in the z = 0 plane. 61/79
  • 68. Introduction Global stability theory Roughness-induced transition Conclusion Wavemaker • Sinuous wavemaker: ֒→ Exclusively localised within the spatial extent of the downstream reversed flow region. ֒→ Shares close connections with the von K´arm´an global instability in the 2D cylinder flow (Giannetti & Luchini 2007, Marquet et al. 2008). • Varicose wavemaker: ֒→ Localised on the top of the central low-speed region shear layer. ֒→ Quite extended in the streamwise direction. ֒→ Yet, its amplitude in the reversed flow region is almost ten times larger than its amplitude in the wake. 62/79
  • 69. Introduction Global stability theory Roughness-induced transition Conclusion Sinuous instability mechanism What we know from local stability approaches? • Central low-speed region can sustain local convective instabilities (Brandt 2006). • Related to the work of the Reynolds stresses against the wall-normal and spanwise gradients of Ub. • Not the dominant local instability though. What global stability analyses revealed? • Existence of a global sinuous instability. • Related to the downstream reversed flow region. • Similar to the von K´arm´an instability in the 2D cylinder flow. 63/79
  • 70. Introduction Global stability theory Roughness-induced transition Conclusion Varicose instability mechanism What we know from local stability approaches? • Central low-speed region can sustain local convective instabilities (Brandt 2006, Denissen & White 2013). • Related to the work of the Reynolds stresses against the wall-normal gradient of Ub. • Dominant local instability and possible large transient growth (Vermeersch 2010) What global stability analyses revealed? • Existence of a global varicose instability. • Find its roots in the reversed flow region. • Mechanism might be similar to the one proposed by Acarlar & Smith (1987). 64/79
  • 71. Introduction Global stability theory Roughness-induced transition Conclusion Roughness-induced transition Non-linear evolution 64/79
  • 72. Introduction Global stability theory Roughness-induced transition Conclusion Non-linear evolution • Varicose instability ֒→ Induces a varicose modulation of the central low-speed region and surrounding streaks. ֒→ Numerous hairpin vortices are shed right downstream the roughness element and trigger very rapid transition to turbulence. ֒→ Dominant frequency and wavelength of this vortex shedding is well captured by global stability analyses. Streamwise velocity distribution in the y = 0.5 plane for (Re, η) = (575, 3). 65/79
  • 73. Introduction Global stability theory Roughness-induced transition Conclusion Non-linear evolution • Varicose instability ֒→ Induces a varicose modulation of the central low-speed region and surrounding streaks. ֒→ Numerous hairpin vortices are shed right downstream the roughness element and trigger very rapid transition to turbulence. ֒→ Dominant frequency and wavelength of this vortex shedding is well captured by global stability analyses. Identification of the vortical structures by the λ2 criterion (Jeong & Hussain 1995). 66/79
  • 74. Introduction Global stability theory Roughness-induced transition Conclusion Non-linear evolution • Sinuous instability ֒→ Induces a sinuous wiggling of the central low-speed region (Beaudoin 2004, Duriez et al. 2009). ֒→ Frequency of this sinuous wiggling well captured by global stability analysis. ֒→ Hairpin vortices are nonetheless observed to be shed downstream the roughness element.. Streamwise velocity distribution in the y = 0.5 plane for (Re, η) = (1125, 1). 67/79
  • 75. Introduction Global stability theory Roughness-induced transition Conclusion Non-linear evolution • Sinuous instability ֒→ Induces a sinuous wiggling of the central low-speed region (Beaudoin 2004, Duriez et al. 2009). ֒→ Frequency of this sinuous wiggling well captured by global stability analysis ֒→ Hairpin vortices are nonetheless observed to be shed downstream the roughness element.. Identification of the vortical structures by the λ2 criterion (Jeong & Hussain 1995). 68/79
  • 76. Introduction Global stability theory Roughness-induced transition Conclusion Non-linear evolution • Sinuous instability ֒→ Monitoring the amplitude of the spanwise velocity in the central mid-plane revealed the bifurcation is supercritical. −20 0 20 40 60 80 100 −0.1 −0.05 0 0.05 0.1 ε=Re−Rec Amplitude 69/79
  • 77. Introduction Global stability theory Roughness-induced transition Conclusion Conclusion & Perspectives 69/79
  • 78. Introduction Global stability theory Roughness-induced transition Conclusion Conclusion • Sinuous instability ֒→ Dominant instability for low aspect ratio roughness elements. ֒→ von K´arm´an-like global instability of the reversed flow region. ֒→ Vortices shed from this region then experiences weak spatial transient growth. ֒→ The creation of hairpin vortices by sinuous global instability is not yet understood. 70/79
  • 79. Introduction Global stability theory Roughness-induced transition Conclusion Conclusion • Varicose instability ֒→ Dominant instability for large aspect ratio roughness elements. ֒→ Mechanism similar to the one proposed by Acarlar & Smith (1987). ֒→ Triggers rapid transition to a turbulent-like state by promoting the creation of hairpin vortices. 71/79
  • 80. Introduction Global stability theory Roughness-induced transition Conclusion Conclusion • Critical roughness Reynolds numbers and observations from DNS in qualitatively good agreements with the transition diagram by von Doenhoff & Braslow (1961). ֒→ Three-dimensional global instability of the flow appears as one of the possible explanations to roughness-induced transition. 72/79
  • 81. Introduction Global stability theory Roughness-induced transition Conclusion Lid-driven cavity flow • Same instability mechanism as before: ֒→ Centrifugal instability of the primary vortex core. • For large LDC, Rec in good agreements with predictions from 2.5D stability analysis. 73/79
  • 82. Introduction Global stability theory Roughness-induced transition Conclusion Lid-driven cavity flow • DNS revealed bursts of kinetic energy related to intermittent chaotic dynamics. ֒→ Koopman modes decomposition suggests it would type-2 intermittent chaos (Pomeau & Manneville 1980). 74/79
  • 83. Introduction Global stability theory Roughness-induced transition Conclusion Stenotic pipe flow • Asymmetry of the stenosis triggers the wall-reattachment at lower Re compared to the axisymmetric case. • Existence of a hysteresis cycle related to a subcritical pitchfork bifurcation. 75/79
  • 84. Introduction Global stability theory Roughness-induced transition Conclusion Stenotic pipe flow • Nonetheless, predictions from global stability analyses are uncorelatted to the experimental observations (Passaggia et al.) ֒→ Transition is dominated by transient growth. • Preliminary optimal perturbation analysis appears to be more conclusive. 76/79
  • 85. Introduction Global stability theory Roughness-induced transition Conclusion Perspectives • Several questions are still unanswered and require further in-depth investigations: ֒→ What is the mechanism responsible for the creation of hairpin vortices in the sinuous case? ֒→ Is the varicose bifurcation super- or subcritical? ֒→ How does global optimal perturbation influence these transition scenarii? Can they trigger subcritical transition to turbulence? ֒→ How does the shape of the roughness element impact the stability properties of the flow? 77/79
  • 86. Introduction Global stability theory Roughness-induced transition Conclusion Perspectives • How to answer these questions? ֒→ More direct numerical simulations! ֒→ Non-linear analyses of these DNS (Koopman modes decomposition, POD, statistical analysis, ...). ֒→ Linear and non-linear transient growth analysis. ֒→ Conduct similar investigations for smooth bumps and hemispherical roughness elements to assess the robustness of these results. 78/79