Complexity

Volume 2018, Article ID 4602485, 15 pages

https://doi.org/10.1155/2018/4602485

## Symmetries, Dynamics, and Control for the 2D Kolmogorov Flow

Department of Mathematics, Kuwait University, P.O. Box 5969, 13060 Safat, Kuwait

Correspondence should be addressed to Nejib Smaoui; moc.oohay@46iuoamsn

Received 13 July 2017; Accepted 19 March 2018; Published 13 May 2018

Academic Editor: Sigurdur F. Hafstein

Copyright © 2018 Nejib Smaoui. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

#### Abstract

The symmetries, dynamics, and control problem of the two-dimensional (2D) Kolmogorov flow are addressed. The 2D Kolmogorov flow is known as the 2D Navier-Stokes (N-S) equations with periodic boundary conditions and with a sinusoidal external force along the -direction. First, using the Fourier Galerkin method on the original 2D Navier-Stokes equations, we obtain a seventh-order system of nonlinear ordinary differential equations (ODEs) which approximates the behavior of the Kolmogorov flow. The dynamics and symmetries of the reduced seventh-order ODE system are analyzed through computer simulations for the Reynolds number range . Extensive numerical simulations show that the obtained system is able to display the different behaviors of the Kolmogorov flow. Then, we design Lyapunov based controllers to control the dynamics of the system of ODEs to different attractors (e.g., a fixed point, a periodic orbit, or a chaotic attractor). Finally, numerical simulations are undertaken to validate the theoretical developments.

#### 1. Introduction

In recent years, a lot of efforts have been devoted to construct dynamical systems that arise from solving the 2D Navier-Stokes equations. In the literature, the dynamics of the Navier-Stokes equations were approximated by using several reduced order models [1–20]. The concept of approximate inertial manifold (AIM) (see [1–11] and references therein) and Fourier Galerkin methods [12–20] is among the methods used to obtain such reduced order systems with the task of approximating the long-time behavior of the 2D Navier-Stokes equations. The behavior of the 2D Navier-Stokes equations depends on the nature of the forcing. When the force is of a single mode, the 2D N-S equations with periodic boundary conditions are known as the 2D Kolmogorov flow.

During the last three decades, numerous numerical studies of the 2D Kolmogorov flow with different forcing terms have appeared [11–29]. In 1981, Franceschini et al. [14–17] constructed a system of ODEs approximating the dynamics of the Kolmogorov flow when the force acts on the mode . In [15], a number of steady states and Hopf bifurcation have been observed up to Reynolds number 100. In 1996, Armbruster et al. [25] analyzed the dynamics of the 2D Navier-Stokes equation when the force acts on the mode by exploiting as much as possible the symmetries in the problem and by using the Karhunen-Loeve decomposition (K-L). Later on and in 1997, Smaoui and Armbruster [27] described a computationally effective way to obtain a reduced order equivariant system of the 2D N-S equations using K-L decomposition with symmetries; the reduced order system obtained consists of 12 nonlinear ODEs. They showed that when Re = 16.6, the ODE system exhibits the same dynamics as the original simulation of the 2D N-S equations. However, the obtained Galerkin system does not show signs of a homoclinic behavior as observed in the original PDE simulation [25].

On the other hand, the control problem of the Navier-Stokes equations and especially to the 2D Navier-Stokes equations has not been completely investigated (see [30–38] and the references therein). In 2003, using a global pinning coupling strategy, Guan et al. [34] designed an adaptive controller to control flow turbulence governed by the 2D Navier-Stokes equations. In 2009, Gambino et al. [35] designed an adaptive controller to drive the state of the system to the stationary solution. Recently, Smaoui and Zribi [36–38] constructed reduced order ODE models using the truncated Fourier expansion method for approximating the dynamics of the 2D N-S equations when the force acts on the mode . They showed that, for , the dynamics of the reduced order models exhibit periodic doubling bifurcation leading to chaotic attractors. In addition, they designed Lyapunov based control laws to drive the states of the reduced order model to the basic state solution and to synchronize two reduced order ODE models having different Reynolds and starting from two different initial conditions.

In this paper, we construct a system of seven ODEs that approximates the dynamics of Kolmogorov flow when the force acts on the mode . It should be noted that the approach used in this paper to construct the system of ODEs is different than the one used in [36–38]. In addition, unlike the system obtained earlier by Smaoui and Armbruster [27], this system shows a homoclinic gluing bifurcation similar to the one observed in the original PDE simulations [25]. Furthermore, Lyapunov based control laws are designed to control the dynamics of the system for a given Re.

The paper is organized as follows. In Section 2, the 2D Navier-Stokes equations are presented and a seventh-order nonlinear ODE system is obtained to approximate the behavior of these equations. The reduced order ODE system is also analyzed in Section 2. Section 3 presents the design of a control law which is used to regulate the states of the reduced order ODE system to a desired fixed point. Section 4 deals with the design of a control scheme to synchronize two reduced order systems obtained from the 2D N-S equations having the same or different Reynolds number but they start from different initial conditions. Finally, some concluding remarks are given in Section 5.

#### 2. The 2D Kolmogorov Flow

##### 2.1. The Seven-Mode Reduced Order System of the 2D Kolmogorov Flow

The “basic 2D Kolmogorov flow” was introduced by Kolmogorov in 1958 as an example on which to study transition to turbulence [39]. This basic flow is the solution of the 2D Navier-Stokes equations with periodic boundary conditions in two directions given bywith and force , which is assumed to be stationary and spatially biperiodic. The kinematic viscosity is , where is the Reynolds number and the pressure is .

The perturbed nondimensional vorticity formulation of the Kolmogorov flow is where the scaled time , , , and .

In Smaoui and Zribi [36–38], the derivation of the reduced order system of ODEs was based on expanding the stream function in (2) as follows:

In this paper, we use a totally different approach than the one used in [36–38]. We derive a system of seven ODEs from the Navier-Stokes equations by expanding in the following form:where is a wave vector with integer components, , and the reality condition must hold.

The equation for iswhere and is the component of with respect to

Next, if the basic flow , then one can consider the set of vectors, , , , , , , and , and their negatives in (5) to obtain the following system of ODEs:

Let , , , , , , and ; then system (6) becomes

After changing the length scale for , the time scale and , and considering the forcing term to act on mode , system (7) becomes

The system given by the equations in (8) can be written in the following form: where the the vector is such that and the diagonal matrix is such that:and the nonlinear vector is such thatwith

*Remark 1. *It is noted that system (8) is invariant under the following symmetries:where , , and are reflection symmetries across the -axis, the -axis, and the origin, respectively. Hence, , , with the identity transformation form an Abelian group: .

##### 2.2. The Dynamics of the Seven-Mode Truncation System

In this subsection, we analyze the dynamics of the seven-mode truncation ODE system presented in (8) for different Reynolds numbers. The DsTool software [40] is used in all numerical simulations presented in this section using the 4th-order Runge-Kutta method as the numerical integrator and with the time step .(i)For , the basic fixed point is the only stable solution and, numerically, it is globally attractive for all . This is a special case of the general results on the theory of Navier-Stokes equations [41].(ii)For there are three fixed points: the old one , which becomes unstable as a result of the crossing of the imaginary axes by one of the eigenvalues of the Lyapunov matrix, and two more, and , that bifurcated from are stable and attracting. Numerical evidence shows that any randomly chosen initial condition is either attracted by or .(iii)For , there are seven fixed points: four asymptotically stable and three unstable. Figure 1 presents the phase portrait of the four asymptotically stable fixed points at . The first asymptotically state fixed point (Figure 1, top left) was obtained using the following initial conditions: The second asymptotically stable fixed point (Figure 1, bottom right) was obtained by applying the reflection symmetry on the initial conditions used for the first asymptotically stable fixed point. That is, The third asymptotically stable fixed point (Figure 1, top right) was generated by applying the reflection symmetry on the initial conditions used for the second asymptotically stable fixed point. That is, The fourth asymptotically stable fixed point (Figure 1, bottom left) was obtained by applying the reflection symmetry on the initial conditions used for the second asymptotically stable fixed point. That is,(iv)For the four asymptotically stable fixed points become unstable because a pair of complex eigenvalues crosses the imaginary axes, and bifurcate via a Hopf bifurcation into a four stable periodic orbits around these four fixed points. The four stable periodic orbits remain stable up to . The other three remain unstable (see Figure 2).(v)For . At each of the four periodic orbits loses stability and bifurcates into a new periodic orbit with double the period.(vi)For , a homoclinic gluing bifurcation similar to the one observed by Armbruster et al. [25] occurs (see Figure 3). It should be noted that the homoclinic gluing bifurcation observed leads to two chaotic or strange attractors each connecting a pair of two stable periodic orbits. Figure 3(b) shows that the dynamics of one of the chaotic attractors follow a quasiperiodic regime; then either bursts to the same attractor or to its symmetric counterpart; then other bursts follow. Numerical simulations indicate that intervals between bursts become shorter as the Reynolds number, , is increased up to .