Global Exponential Stabilization for a Simplified Fluid-Particle Interaction System

Marius Tucsnak, Zhuo Xu · 2025

Read Global Exponential Stabilization for a Simplified Fluid-Particle Interaction System by Marius Tucsnak, Zhuo Xu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-11446.txt.

Introduction

Marius Tucsnak and Zhuo Xu prove that a point mass drifting in a simple fluid can be held steady forever. Their paper appeared in two thousand twenty-five. The fluid follows a viscous Burgers equation, a one-dimensional stand-in for flow. A point mass moves inside it, pushed by the current. Feedback control can act on the mass. The authors ask whether calm is reachable from any start.

Two results answer yes, with exponential speed. First, the free velocity field settles globally on its own. Second, with feedback on, the mass also glides to a chosen target spot. A special test function keeps the mass away from the walls. A perturbed Lyapunov function then seals the decay. Control theory gains a clean benchmark for fluid-particle systems.

Catalog record

CREATORMarius Tucsnak, Zhuo Xu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-11446

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Global Exponential Stabilization for a Simplified Fluid-Particle Interaction System Marius Tucsnak, Zhuo Xu Abstract This work considers a system coupling a viscous Burgers equation (aimed to describe a simplified model of 1 ​ D 1D fluid flow) with the ODE describing the motion of a point mass moving inside the fluid. The point mass is possibly under the action of a feedback control. Our main contributions are that we prove two global exponential stability results. More precisely, we first show that the velocity field corresponding to the free dynamics case is globally exponentially stable. We next show that, in the presence of the feedback control both the velocity field and the distance from the mass point to a prescribed target position decay exponentially. The proofs of these results heavily rely on the use of a special test function allowing both to prove that the mass point stays away from the boundary and to construct a perturbed Lyapunov function. Key words: Viscous Burgers equation; fluid-solid interaction system; finite energy solution; well-posedness; global exponential stability. AMS subject classification: 35R35, 35Q35, 74F10, 93B52, 93D15 1 Introduction In recent years, the equations that describe fluid-structure interactions have attracted significant attention in mathematics and physics. Early references such as Conca, San Martín and Tucsnak [ 2 ] , Desjardins and Esteban [ 3 ] , Tucsnak [ 11 ] , Takahashi [ 12 ] , Vázquez and Zuazua [ 14 ] , [ 15 ] provided well-posedness results for such interaction systems. There exist also results on the stabilization and the controllability of such type of systems. These results typically require smooth and close to equilibrium initial data. One persistent difficulty in obtaining global results is to ensure the fact that there are no collisions of the solid with the boundary. This can be granted by the choice of the data of the problem (see, for instance, Hesla [ 6 ] , Hillairet [ 7 ] , Hillairet and Gérard Varet [ 5 ] or Hillairet and Takahashi [ 8 ] ) and/or by the design of the control mechanism, Takahashi, Tucsnak, and Weiss [ 13 ] . In this paper, we consider a one-dimensional model first introduced by [ 14 ] , [ 15 ] , where the existence, uniqueness and large-time behavior of the solutions have been first investigated, in the case of a fluid filling the entire space. Subsequently, Doubova and Fernandez-Cara [ 4 ] addressed the boundary controllability problem for this system, proving local null controllability for the nonlinear system when control acts on both boundaries. Then, specifically regarding the establishment of null controllability when control is applied at only one end of the domain, Liu, Takahashi, and Tucsnak [ 10 ] solved this problem by employing spectral methods and introducing a the so called source point method . In 2015, Cîndea, Micu, Roventa, and Tucsnak [ 1 ] considered a model in which control is applied directly to the particle. Their analysis focused on a closed-loop system with initial data close to equilibrium, and they examined the long-time behavior of the solutions. Using a combination of spectral and fixed-point techniques, they established local null controllability for the nonlinear interaction system. There are two questions left open in [ 1 ] . The first one is the global well-posedness for the closed loop interaction system. The second consists in obtaining decay rates in the presence of the considered feedback control. In this paper we consider the system: \displaystyle\dot{v}(t,y)+v(t,y)v_{y}(t,y)-v_{yy}(t,y)=0 \displaystyle t \displaystyle\in(0,T),\quad y\in(-1,1),\quad y\neq h(t), (1.1) \displaystyle v(t,-1)=v(t,1)=0 \displaystyle t \displaystyle\in(0,T), (1.2) \displaystyle\dot{h}(t)=v(t,h(t)) \displaystyle t \displaystyle\in(0,T), (1.3) \displaystyle\ddot{h}(t)=[v_{y}](t,h(t))+u(t) \displaystyle t \displaystyle\in(0,T), (1.4) \displaystyle v(0,y)=v_{0}(y) \displaystyle y \displaystyle\in(-1,1), (1.5) \displaystyle h(0)=h_{0},\qquad\qquad\dot{h}(0)=g_{0}. (1.6) where \left[v_{y}\right](t,h(t))=v_{y}\left(t,h(t)^{+}\right)-v_{y}\left(t,h(t)^{-}\right)\qquad\qquad(t\geqslant 0). (1.7) Here, we consider a fluid velocity field denoted by v=v(t,y) , which is governed by the viscous Burgers equation. Additionally, the motion of a particle within this field obeys Newton’s second law (as referenced in ( 1.4 )). The position of the particle at any given time is denoted by h(t) , and its velocity is represented by \dot{h}(t) , both of which are functions of time only. The input force acting on the particle, denoted by u , also depends solely on time t . We observe that the system defined by ( 1.1 )- ( 1.6 ) constitutes a free boundary value problem since the position h(t) of the particle, thus the domain filled by the fluid, is one of the unknowns of the problem. This is a property which ( 1.1 )- ( 1.6 ) shares with Stefan problems, see Koga and Krstic [ 9 ] for the control theoretic properties of the corresponding systems. However, in our case, the presence of a second derivative of h in ( 1.4 ), which is the main difference with respect to Stefan type systems (where a single time derivative appears) yields more regularity of the state trajectories. The presence of a free boundary necessitates a precise definition of the concept of a finite energy solution. Such a definition has been meticulously formulated in [ 1 , Definition 1.1] . Definition 1.1 . Given T>0 , v_{0}\in L^{2}(-1,1) , g_{0}\in\mathbb{R} , h_{0}\in(-1,1) and u\in L^{2}(0,T) , we say that \begin{bmatrix}v\\ g\\ h\end{bmatrix}\in\left[C\left([0,T];\;L^{2}(-1,1)\right)\cap L^{2}\left((0,T);\;H^{1}_{0}(-1,1)\right)\right]\times L^{2}(0,T)\times H^{1}(0,T), (1.8) is a finite energy solution of system ( 1.1 )-( 1.6 ) on [0,T] if h(0)=h_{0} , \dot{h}(t)=g(t)=v\left(t,h(t)\right) , h(t)\in(-1,1) for almost every t\in[0,T] and \displaystyle\int_{-1}^{1}v(t,y)\psi(t,y)\,{\rm d}y-\int_{-1}^{1}v_{0}(y)\psi(0,y)\,{\rm d}y+g(t)l(t)-g_{0}l(0)-\int_{0}^{t}g(\sigma)\dot{l}(\sigma)\,{\rm d}\sigma \displaystyle-\int_{0}^{t}\int_{-1}^{1}v(\sigma,y)\dot{\psi}(\sigma,y)\,{\rm d}y\,{\rm d}\sigma+\int_{0}^{t}\int_{-1}^{1}v_{y}(\sigma,y)\psi_{y}(\sigma,y)\,{\rm d}y\,{\rm d}\sigma \displaystyle-\frac{1}{2}\int_{0}^{t}\int_{-1}^{1}v^{2}(\sigma,y)\psi_{y}(\sigma,y)\,{\rm d}y\,{\rm d}\sigma=\int_{0}^{t}u(\sigma)l(\sigma)\,{\rm d}\sigma, (1.9) for every t\in[0,T] and for every \displaystyle\left[\begin{gathered}\psi\\ l\end{gathered}\right]\in\left[H^{1}\left((0,T);\;L^{2}(-1,1)\right)\cap L^{2}\left((0,T);\;H^{1}_{0}(-1,1)\right)\right]\times H^{1}(0,T), \displaystyle l(t)=\psi(t,h(t))\qquad\qquad\qquad\left(t\in[0,T]\right). (1.12) A triple \begin{bmatrix}v\\ h\\ g\end{bmatrix} which is a finite energy solution on [0,T] for every T>0 is called a global finite energy solution. Note that the test functions used above depend on the solution and more precisely on its component h . Here and in the rest of this paper, for each m\in\mathbb{N} , H^{m} and H^{m}_{0} denote the classical Sobolev spaces and H^{-m} stands for the dual of H^{m}_{0} with respect to the pivot space L^{2} . One of the contributions of our work is that we prove the existence and uniqueness of global finite energy solution of ( 1.1 )-( 1.6 ) when u is given by a feedback law of the form u(t)=\mathcal{K}(h_{1}-h(t))\qquad\qquad(\mathcal{K}\geqslant 0,\qquad t>0), (1.13) with a given h_{1}\in(-1,1) and \mathcal{K}\geqslant 0 . This feedback may be regarded as a proportional derivative (PD) controller, as is often used in control engineering. Another interpretation is that the particle is actuated by a force which is generated by a spring (with elastic constant \mathcal{K} and fixed at h_{1} ). Unlike in [ 1 ] , in Theorem 1.2 below we do not assume that the initial velocity field v_{0} and the initial data g_{0} and h_{1}-h_{0} are small, in a suitable sense. The first main result of this paper is: Theorem 1.2 . Let v_{0}\in L^{2}(-1,1) , g_{0}\in\mathbb{R} , h_{0},\ h_{1}\in(-1,1) and \mathcal{K}\geqslant 0 . Then system ( 1.1 )-( 1.6 ), with u given by the feedback law ( 1.13 ), admits a unique global finite energy solution. Moreover, there exists \alpha>0 depending on \|v_{0}\|_{L^{2}(-1,1)} , g_{0} h_{0} , h_{1} and \mathcal{K} , such that the trajectory h of the particle satisfies \displaystyle-1+\alpha\leqslant h(t)\leqslant 1-\alpha\qquad\qquad\qquad(t\geqslant 0).

Read the complete text