**Page Web du cours** > **Introduction aux instabilités hydrodynamiques** **ATTENTION : ceci est la page du cours édution 2019-2020.** **L'adresse du cours de cette année est la suivante :** [http://basilisk.fr/sandbox/easystab/M2DET/Instabilities.md]() #Presentation of the course ##Objectives of the course The concept of *Instability* generally means the spontaneous transition of a system from a "regular state" (steady, symmetric, ...) to a "less regular state" (unsteady, asymetric, ...). The identification and description of instabilities is a general situation encountered in many situation in fluid dynamics... and more generally in other fields of physics. The mathematical study of instabilities originates from the second half of the 19th century and is associated with the name of classical scientists (Rayleigh, Kelvin, Helmholtz, ...). However this discipline is still in flourishing development nowadays. This course is an introduction to: - The various kinds of instabilities encountered in fluid dynamics, with emphasis on their *physical* origin, - The *mathematical* formalism allowing to describe an instability, - The specific *numerical* methods used in the study of instability phenomena. ##Material provided for this course. This webpage gathers the online material for the course. The material consists of various documents, including lecture notes, complementary documents, exercices, and **commented programs**. The commented programs aim at providing simple, self-contained programs incorporating their own documentation, written in Matlab/Octave language. The proposed programs allow to study the main classes of instabilities considered in this course, and to produce most of the numerical results illustrating the course. The programs are largely built using the support and the philosophy of the [**easystab**](/sandbox/easystab/README) project launched by J. Hoeffner and his colleagues from Université Paris VI. The website of the present course is actually hosted by the website of the easystab project, which is itself hosted by the website of the basilisk numerical simulation code. ## How to use the programs provided in these pages : - Get the basic bricks of the project (functions to build matricial operators, etc...) by downloading and unpacking [easypack.zip]() - To download a program, open its web page, click "raw page source" in the left column, cut-paste into the matlab/octave editor window, and save it with the corresponding name and ".m" extension. - Once it is downloaded, just play with the programs and have fun ! - If you have improved a program and wish to share it, you are welcome ! the collection of programs is handled under the form of a wiki, so you can easily edit, discuss, and create new programs. ## Personal work The interactive teaching part of this course consists of only 10 lectures of 1h30. To complement this limited number of teaching hours, a signicant personal work is required from you, to prepare and digest in the best **Before the lecture :** - Check the suggested preparation work, - Read the material for the lecture (possibly print it so that you can write on it during the lecture), - Download the commented programs and check that they correctly run on your computer. The programs will be used and modified during the lectures, so it is advised that you bring your personal laptops with you. **After the lecture :** - Reconstruct the results produced during the course with the programs. - It is recommended that you bring your personal laptops with you. - Do the suggested exercices. ## Bibliography for the course - F. Charru, *Instabilités hydrodynamiques*, CNRS editions (main source for this course) - Drazin & Reid, *Hydrodynamic instabilities* - Schmid & Henningson, *Instabilities and transition in shear flows* - Glendinning, *Instabilities, chaos & Dynamical systems* - Godrèche & Manneville, *Hydrodynamics and Nonlinear Instabilities* - Bender & Orzag, *Advanced mathematical methods for scientists and engineers.* # Organisation of the course ## **Lecture 1** Dynamical systems I : Introduction, classification of fixed points ### Preparation work : - Review your knowledge about linear systems of differential equations, linear algebra, eigenvalues of real and complex matrices. - Advised lectures : F. Charru, sections 1.1 and 1.2; Glendining. ### Material for the lecture : - Lecture notes for lectures 1 and 2 : [DynamicalSystems.md]() - Complements on linear systems [LinearSystems.md]() - Commented programs [/sandbox/easystab/PhasePortrait_Linear.m](PhasePortrait_Linear.m) and [/sandbox/easystab/PhasePortrait_NonLinear.m](PhasePortrait_NonLinear.m). ### Personal work : 1. Consider the damped harmonic oscillator, with the form : $$m \ddot x = - k x - \mu \dot x $$ a. Write this problem under the form of a linear dynamical system of order 2, with the form $X = [x,\dot x]$. b. Give the nature of the fixed-point $[0,0]$ as function of the parameters (node, focus, etc...) c. Explain the correspondance between the identified cases and the classical analysis of this equation (pseudo-pediodic, periodic, critical, aperiodic and unstable régimes) c. Using the program [PhasePortrait_Linear.m](), draw phase portraits for this problem for the five regimes of solution. 2. Brusselator [correction](http://basilisk.fr/sandbox/easystab/Correction_Exercices.md#exercice-1.2) Consider the "Brusselator" 2D dynamical system: $$\frac{dx}{dt} = A + x^2 y - (B+1) x$$ $$\frac{dy}{dt} = B x-x^2 y $$ This dynamical system is a classical model initially inttroduced to model an oscillating chemical reaction. Here $x$(t) and $y(t)$ correspond to the concentrations of two chemical products, and $A$ and $B$ are constants. a. Show that this system admits a unique equilibrium point $[x,y] = [x_0,y_0]$. b. Study the linear stability of this equilibrium point. Specify the nature of this point (node, focus, ...) as function of the parameters. [Additional Exercices for lecture 1](LinearSystems.md#exercices) ## **Lecture 2** Dynamical systems II : Bifurcations ### Preparation work: - Charru, section 1.3 (see in particular section 11.6) ### Material for the lecture : - Lecture notes for lectures 1 and 2 : [DynamicalSystems.md]() - Commented programs [/sandbox/easystab/PhasePortrait_Linear.m]() and [/sandbox/easystab/PhasePortrait_NonLinear.m]() ### Personal work : 1. (Charru, exercice 11.7.12) Build the bifurcation diagram of the following amplitude equation, where $\mu$ is the bifurcation parameter: $$ \frac{d x}{d t} = 4 x ( (x-1)^2-\mu-1 ) $$ Verify that the bifurcation diagram is the one given in figure 11.20 of Charru. (warning : there is a little error in the book ! will you find it ???) 2. Bifurcation analysis of the rotating pendulum equation. Consider a pendulum with mass $m$ length $L$. The pivot is characterized by a friction coefficient $\mu_f$. The pendulum is in a unuformly rotating frame with rotation rate $\Omega$ with vertical axis. a. Show that the motion of the pendulum is driven by the following equation : $$ m L^2 \ddot \theta = - \mu_f \dot \theta - m g L \sin \theta + m L^2 \Omega^2 \sin \theta \cos \theta $$ b. Using convenient nondimensionalization (including redefinition of the time-scale), and under the hypothesis of small angles, show that the equation can be reduced to the "two-dimensional pitchfork equation": $$ \ddot x + \dot x = r x - x^3 $$ where $$ r = \frac{m ( \Omega^2 - g L) }{\mu_f} $$ c. Justify that for $r \ll 1$ The inertial term (second time-derivative) is negligible with respect to the friction term (first time-derivative) and that the classical pitchfork equation is recovered. Deduce the equilibrium solutions. d. Study the linear stability of the equilibrium points and give their nature. Show that a bifurcation occurs for $r=0$ and that a qualitative change of the nature of the equilibrium points occurs for $r=-1/4$ and for $r=1/8$. (this question may be treated either by calculus of by using the PhasePortrait_NonLinear.m program) [Additional exercices on lecture 2](http://basilisk.fr/sandbox/easystab/david/DynamicalSystems.md#exercices) ### Going further: - Charru, chapter 11 - Glendinning ## **Lecture 3** Numerical methods for eigenvalue problems ### Preparation work: - Review your knowledge about the finite difference method for the numerical resolution of differential equations. - Study the numerical implementation of this method provided in easystab project: see programs [/sandbox/easystab/diffmat.m](), [/sandbox/easystab/diffmat_dif1D.m]() and [sandbox/easystab/differential_equation.m]() ### Material for the lecture: Lecture notes [NumericalMethodsForEigenvalueProblems.md](http://basilisk.fr/sandbox/easystab/NumericalMethodsForEigenvalueProblems.md) Example program [diffusion_eigenmodes.m]() Commented program (2018) [/sandbox/easystab/david/GinsburgLandau_Linear.m]() Commented program (2019) [/sandbox/easystab/david/GinsburgLandau_NonLinear_OneMode.m]() ### Personal work: [Exercices for lecture 3](http://basilisk.fr/sandbox/easystab/NumericalMethodsForEigenvalueProblems.md#exercices-for-lecture-3) ## **Lecture 4** Gravity-capillary waves and Rayleigh-Taylor instabilities ### Personnal work This chapter is left as personal work. The situation investigated corresponds to two fluids separated by a horizontal interface. The region $y>0$ is occupied by a fluid of density $\rho_1$ and the region $y<0$ is occupied by a fluid of density $\rho_2$. We note $\gamma$ the surface tension and we neglect viscous effects. 1. Show that modal perturbations proportional to $e^{i k x - i \omega t}$ are governed by the following relation between $k$ and $\omega$ (*dispersion relation*) : $$ \omega^2 = \frac{(\rho_2-\rho_1)}{\rho_2+\rho_1} g k + \frac{\gamma}{\rho_2+\rho_1} k^3. $$ 2. In the case where the lower fluid is heavier ($\rho_2> \rho_1$), show that the dispersion relation predicts *surface waves* with real frequencies. Recall the definition and physical interpretation of the *group velocity* $c_g$. Justify that the group velocity has a minimum value $c_m$ corresponding to a particular wavelength $\Lambda_m = 2 \pi / k_m$. 3. In the case where the lower fluid is lighter ($\rho_1> \rho_2$), show that the dispersion relation predicts either *surface waves* with real frequencies, or *unstable modes* with purely imaginary frequencies $\omega$ (hence purely real eigenvalues $\lambda$). Show that instabilities occur for wavelength $\Lambda = 2 \pi / k > 2 \pi l_c$ where $l_c$ is the capillary length defined by $$ l_c = \sqrt{\frac{\gamma}{|\rho_1-\rho_2| g}} $$ ### Suggested lectures : - Charru, sections 2.3 and 2.4 - Wikipedia : [https://en.wikipedia.org/wiki/Dispersion_(water_waves)]() [https://en.wikipedia.org/wiki/Capillary_wave]() [https://en.wikipedia.org/wiki/Rayleigh%E2%80%93Taylor_instability]() ### Exercices - Rayleigh-Taylor instability with confinement effects (Charru, exercice 2.8.1) - Rayleigh-Taylor instability of a suspended film (Charru, exercice 2.8.2) - Derive the dispersion relation and instability criteria for a double layer of fluid, defined as follows: $$ \rho = \left\{ \begin{array}{ll} \rho_1 \qquad & (y<-L/2); \\ \rho_2 & (-L/2L/2) \end{array}\right. $$ ## **Lecture 5** Thermal instabilities (Rayleigh-Benard) ### Preparation work: - Review your knowledge about thermal convection (Bousinesq approximation) - Charru, section 2.5 ### Material for the course: - Commented program [/sandbox/easystab/RayleighBenard.m]() performing the resolution of the linear eigenvalue problem for convection in a horizontal cell. - Commented program [/sandbox/easystab/lorenz_convection.m]() displaying the solution of the Lorenz system, with reconstruction of the convection pattern. - Commented program [/sandbox/easystab/lorenz.m]() displaying the divergence of two initially close trajectories (characteristic of a chaotic behaviour). ### Personal work: **Exercice 1** Mathematical resolution of the Rayleigh-Bénard problem for "free-free" boundaries (sources: Drazin & Reid, sections 2.8 and 2.10 ; Rieutort ; Chandrasekar) 1. Starting from the stability equations in primitive form for $[\hat u, \hat v, \hat, p, \hat T]$, show that the problem can be reduced to a single differential equation of order 6 with the following form (Drazin & Reid, Eq. 8.39, p. 43) : $$ (\partial_y^2 - k^2) (\partial_y^2 - k^2 - \lambda) (\partial_y^2 - k^2 - \lambda/Pr)\hat v = 0 $$ 2. precise the set of boundary conditions to be used with this equation, in the respective cases of (a) rigid boundaries, and (b) "free-slip" boundaries. 3. In the case of "free-slip" boundaries, show that the least-stable eigenvalue is (D&R, eq. 10.5): $$ \lambda = -\frac{1}{2}(1+Pr)(\pi^2 + k^2) \pm \frac{1}{2} \sqrt{ (Pr-1)^2(\pi^2+k^2)^2+k^2 \, Ra \, Pr/(\pi^2+k^2)} $$ 4. Show that for a given $k$, the flow is unstable for $Ra >Ra_j(k)$, with $$ Ra_j(k) = (\pi^2+k^2)^3/k^2 $$ (D&R, Eq. 10.6) Draw this curve in the $(Ra,k)$ plane. 5. Show that the instability threshold is $Ra_c = min [Ra_j(k)] = 27 \pi^4/4 \approx 657.5$ and give the corresponding value of $k$. **Exercice 2** Study of the Lorenz system. 1. Find all fixed-point solutions of the Lorenz system. Show that the system undergoes a supercritical Pitchfork bifurcation for $r>1$. 2. Show that if if $P>b+1$, the stable solution existing for $r>1$ becomes unstable through a Hopf bifurcation for $r>r_c$ with $$ r_c = \frac{ P ^2 + b P + 3P}{P - b - 1}. $$ Give the numerical value for $b=1$ (Charru) and for $b=8/3$ (initial choice of Lorenz). Tip : Study the stability of the fixed-point solution. Write the characteristic polynomial governing its stability under the form $P(\lambda) = \lambda^3 + A \lambda^2 +B \lambda + C$. Remarking that if a Hopf bifurcation occurs, this polynomial has two purely imaginary roots and one real root, write a condition on the coefficients $A$,$B$,$C$ of the polynomial. 3. Using the program, check that the bifurcation is subcritical. 4. Using the program, verify the existence of a "strange attractor" for $r>r_s$ with $r_s\approx 24$ (try to find the best approximation of $r_s$ !) 5. Verify that for $r_c Commented program [/sandbox/easystab/KH_temporal_viscous.m]() * Viscous stability analysis of the plane Poiseuille flow > Commented program [/sandbox/easystab/poiseuille_uvp.m]() for explaining the numerical method and validation. > Program [/sandbox/easystab/david/TS_PlanePoiseuille.m]() for parametric study. * Viscous stability analysis of the Blasius boundary layer (Program [/sandbox/easystab/blasius.m]() to compute the base-flow solution for a boundary layer (Blasius equation)) (Program [/sandbox/easystab/blasiusStability.m]() to compute the stability properties of the Blasius boundary layer (maybe next year...) ### Personal work: [Exercices for lecture 7](/sandbox/easystab/LectureNotes_Viscous.md#exercices) ## **Lecture 8** Shear-flow instabilities III (three-dimensional perturbations and transient growth mechanisms) ### Preparation work: - Review the theory of dynamical systems (lectures 1 and 2) - Study the "green" sections in the document [LinearSystems.md]() ### Material for the lecture: * Lecture notes [LectureNotes_NonModal.md](/sandbox/easystab/LectureNotes_NonModal.md) ### Personal work: [Exercices for chapter 8](/sandbox/easystab/LectureNotes_NonModal.md#exercices-for-chapter-8) ## **Lecture 9** Shear-flow instabilities IV (spatial an spatio-temporal theory) ### Preparation work: Charru, Chapter 3. ### Material for the lecture: * Lecture notes [LectureNotes_SpatioTemporal.md](/sandbox/easystab/LectureNotes_SpatioTemporal.md) * Commented program [KH_spatial_invinviscid.m](/sandbox/easystab/KH_spatial_inviscid.m) * Commented program [KH_spatial_viscous.m](/sandbox/easystab/KH_spatial_viscous.m) * Commented program [SpatioTemporal_ModelEquations.m](/sandbox/easystab/SpatioTemporal_ModelEquations.m) ### Personal work: * [Exercices for chapter 9](/sandbox/easystab/LectureNotes_SpatioTemporal.md#exercices-for-lecture-9) ## **Lecture 10** Bluff-body Wake instabilities ### Preparation work: ### Material for the lecture: This final lecture is based on a paper available [here:](https://gitlab.com/stabfem/StabFem/tree/develop/99_Documentation/ARTICLE_STABFEM) The programs used for this lecture are part of the StabFem project, available [here](https://gitlab.com/stabfem/StabFem). To get this project, you will have to do the following steps : - Install the FreeFem++ software (free, available [here](http://freefem.org) ) - open a terminal (linux/unix/mac systems) or a console (windows) and type the following command : ~~~ { .haskell } git clone https://gitlab.com/stabfem/StabFem ~~~ ### Personal work # Tips for the exam - A few exam subjects from the past years are available here: [http://basilisk.fr/sandbox/easystab/M2DET/]() - Corrections for a number of exercices are regrouped here: [http://basilisk.fr/sandbox/easystab/Correction_Exercices.md]()