Strong solutions and stability for a thin-film equation of shear-thinning fluids with contact line in partial wetting

Manuel V. Gnann, Christina Lienstromberg, Katerina Nik · 2026

Read Strong solutions and stability for a thin-film equation of shear-thinning fluids with contact line in partial wetting by Manuel V. Gnann, Christina Lienstromberg, Katerina Nik (2026) 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-2604-22590.txt.

Introduction

Gnann, Lienstromberg, and Nik study strong solutions of a thin-film equation for fluids that thin under shear, where the film meets a surface at a nonzero contact angle. Earlier work had built only weak solutions, leaving behavior near the vanishing film height unclear. By perturbing a linear profile and passing a time-discretized variational problem to its limit, they prove existence and stability of such solutions.

Catalog record

CREATORManuel V. Gnann, Christina Lienstromberg, Katerina Nik
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE199346
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.22590

Browse the full library · About the Tower of Babel

Full text

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

Strong solutions and stability for a thin-film equation of shear-thinning fluids with contact line in partial wetting Thanks: The authors acknowledge the Lorentz Center in Leiden for its hospitality during the workshop Nonlinear diffusion models: analytical & numerical challenges in January 2026, where fruitful discussions regarding this work took place. This publication is part of the project Codimension two free boundary problems (with project number VI.Vidi.223.019 of the research program ENW Vidi ) which is financed by the Dutch Research Council ( NWO ). Both MVG and KN have received funding from this project during the preparation of this paper. Manuel V. Gnann, Christina Lienstromberg, Katerina Nik Abstract . We consider a power-law thin-film equation for strongly shear-thinning fluids. Weak solutions to this equation have been constructed more than twenty years ago by Ansini and Giacomelli. Here, we pass over to analyzing strong solutions with nonzero contact angle (partial-wetting regime), and place emphasis on studying the behavior of solutions near points where the film height vanishes (the contact-line region) by considering perturbations of a linear profile. The leading-order equation in von-Mises coordinates shows similarities with the evolution equation for the p p -Laplace, though being of fourth order. Using a time discretization, we reduce the leading-order problem to finding a variational solution, and pass to the limit in the discretization scheme on suitably estimating higher-order nonlinear terms in conjunction with compactness arguments. This proves existence and asymptotic stability of strong solutions that are perturbations of the linear profile, and yields control on the contact-line velocity on carefully tracking singular terms in our estimates. While we believe that the transformed equation shows mathematical features the analysis of which stands on its own merit, it also physically corroborates shear thinning behavior as an alternative in resolving the no-slip paradox, as opposed to more standard approaches like introducing slip at the liquid-solid interface. 2020 Mathematics Subject Classification 35K65, 35K35, 76D08, 76A05 1. Introduction 1.1. Setting Consider the non-Newtonian thin-film equation (1.1a) h_{t}+\big(h^{\alpha+2}|h_{yyy}|^{\alpha-1}h_{yyy}\big)_{y}=0\quad\text{in }\{h>0\}, where the dependent variable h=h(t,y) describes the height of an incompressible viscous thin film as a function of time t\geq 0 and lateral position (base point) y\in\mathbb{R} . The parameter \alpha>0 is the so-called flow-behavior exponent which features the strain-dependent rheology of the fluid. More precisely, for 0<\alpha<1 , the corresponding fluid is shear-thickening, while it is Newtonian for \alpha=1 and shear-thinning for \alpha>1 . In this paper, we consider the strongly shear-thinning regime in which \alpha>2 . Equation ( 1.1a ) can be derived through a lubrication approximation (small film heights h at moderate Reynolds number) from the underlying (Navier-) Stokes system for a non-Newtonian power-law fluid with a no-slip condition at the substrate. For a formal derivation in case of the related Ellis law, see for instance [ 28 ] . We solve ( 1.1a ) with the contact-angle condition (1.1b) \lim_{\{h>0\}\owns\tilde{y}\to y}|\partial_{\tilde{y}}h(t,\tilde{y})|=1\quad\text{for }y\in\partial\{h>0\}, and the condition for the contact-line velocity (1.1c) \lim_{\{h>0\}\owns\tilde{y}\to Y_{0}}h^{\alpha+1}(t,\tilde{y})|\partial^{3}_{\tilde{y}}h(t,\tilde{y})|^{\alpha-1}\partial^{3}_{\tilde{y}}h(t,\tilde{y})=\frac{{\rm d}Y_{0}}{{\rm d}t}\quad\text{for the contact lines }Y_{0}\in\partial\{h>0\}, which implies no flux at the triple junction (the contact line), where liquid, air, and solid meet. A special solution to ( 1.1 ) is the linear profile (1.2) h=y_{+}\coloneqq\begin{cases}y&\text{ for }y>0,\\ 0&\text{ for }y\leq 0.\end{cases} The present paper deals with perturbations of this linear profile ( 1.2 ). Specifically, this enables us to precisely study the properties of ( 1.1 ) close to the contact line \partial\{h>0\} , separating wetted from dry regions. Here, we prove that for \alpha>2 strong solutions to ( 1.1 ) with moving free boundary exist, thus rigorously justifying shear-thinning behavior as an alternative to assuming slip at the liquid-solid interface in resolving the no-slip paradox. Indeed, for the Newtonian thin-film equation with no-slip condition at the substrate h_{t}+(h^{3}h_{yyy})_{y}=0\quad\text{in }\{h>0\}, the contact line cannot move. See the first discussion of the no-slip paradox in [ 16 ] and the reviews [ 8 , 25 , 3 ] for formal results from the physics literature. Its regularization on introducing (nonlinear) slip at the liquid-solid interface is given by (1.3) h_{t}+(h^{n}h_{yyy})_{y}=0\quad\text{in }\{h>0\}, where n\in[1,3) . Then it is somewhat natural also from the mathematical perspective to ask for regularizations thereby incorporating the term h_{yyy} in the flux. In this regard, from the physical perspective ( 1.1a ) turns out to be a natural regularization. See [ 28 , 1 ] for a discussion of the no-slip paradox in the case of Ellis’ law. 1.2. Transformation onto a fixed domain: von-Mises transform Next, we transform the free-boundary problem ( 1.1 ) onto a fixed domain. Therefore, we suppose \{h>0\}=(Y(t,0),\infty) and use the von-Mises transform (cf. Figure 1 ) (1.4) h(t,Y(t,x))=x. Y(t,x) x h(t,\color[rgb]{0,0,1}Y(t,x)\color[rgb]{0,0,0})=\color[rgb]{1,0,0}x\color[rgb]{0,0,0} gas liquid triple junction film height h base point x or y Figure 1. Schematic plot of a liquid thin film and the von-Minses transform ( 1.4 ). The transformation ( 1.4 ) is well-defined provided [0,\infty)\owns x\mapsto Y(t,x) is, for t\geq 0 fixed, a strictly increasing profile. We differentiate ( 1.4 ) with respect to t and x and obtain (1.5) h_{y}(t,Y(t,x))\,Y_{x}(t,x)=1\quad\text{and}\quad\partial_{t}h(t,Y(t,x))+h_{y}(t,Y(t,x))\,Y_{t}(t,x)=0. Defining (1.6) F\coloneq Y_{x}^{-1} and using (1.7) \partial_{y}=F\partial_{x}, we obtain with ( 1.4 ) and ( 1.5 ) that (1.8) h_{y}(t,Y(t,x))=F(t,x). Additionally using ( 1.1a ) in combination with ( 1.4 ), ( 1.5 ), ( 1.7 ), and ( 1.8 ), we find (1.9) F\partial_{t}Y=F\partial_{x}\big(x^{\alpha+2}\,|F\partial_{x}(F\partial_{x}F)|^{\alpha-1}\,F\partial_{x}(F\partial_{x}F)\big)\quad\text{for }t,x>0, so that with F_{t}\stackrel{{\scriptstyle\eqref{f-mises}}}{{=}}-F^{2}Y_{xt} and noting that \partial_{x}(F\partial_{x}F)=\frac{1}{2}\partial_{x}^{2}F^{2} , we obtain (1.10a) F_{t}+2^{-\alpha}\,F^{2}\,\partial_{x}^{2}\big(x^{\alpha+2}\,F^{\alpha}g_{\alpha}(\partial_{x}^{2}F^{2})\big)=0\quad\text{for }t,x>0, where (1.10b) \displaystyle g_{\alpha}(v) \displaystyle\coloneqq|v|^{\alpha-1}v. With ( 1.1b ) and ( 1.8 ), the contact-angle condition translates into (1.10c) F=1\quad\text{at }x=0. With ( 1.1c ) the condition for the contact-line velocity takes the form (1.10d) \lim_{x\downarrow 0}\,x^{\alpha+1}g_{\alpha}(F\partial_{x}^{2}F^{2})=2^{\alpha}\partial_{t}Y(t,0). 1.3. Leading-order equation Multiplying the evolution equation ( 1.10a ) by F , we have 2^{\alpha-1}(F^{2})_{t}+(F^{2})^{\frac{3}{2}}\,\partial_{x}^{2}\big(x^{\alpha+2}\,(F^{2})^{\frac{\alpha}{2}}g_{\alpha}(\partial_{x}^{2}F^{2})\big)=0\quad\text{for }t,x>0, so that writing (1.11) F^{2}\eqqcolon 1+u, and rescaling time according to (1.12) t\longmapsto 2^{\alpha-1}t, we have \displaystyle u_{t}+(1+u)^{\frac{3}{2}}\,\partial_{x}^{2}\big(x^{\alpha+2}\,(1+u)^{\frac{\alpha}{2}}g_{\alpha}(\partial_{x}^{2}u)\big)=0\quad\text{for }t,x>0. Therefore, equation ( 1.10a ) attains the leading-order form (1.13a) u_{t}+\partial_{x}^{2}\big(x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u)\big)=f\quad\text{for }t,x>0, where f=N(u) contains all terms of order higher than |u|^{\alpha} : (1.13b) \displaystyle N(u) \displaystyle\coloneqq(1-(1+u)^{\frac{3}{2}})\,\partial_{x}^{2}(x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u))+(1+u)^{\frac{3}{2}}\,\partial_{x}^{2}\big((1-(1+u)^{\frac{\alpha}{2}})x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u)\big), From ( 1.10c ) and ( 1.11 ) we infer that the dependent variable u additionally satisfies the boundary condition (1.13c) u=0\quad\text{at }x=0, and from ( 1.10d ) we obtain that (1.13d) \lim_{x\downarrow 0}x^{\alpha+1}g_{\alpha}(\partial_{x}^{2}u)=2\partial_{t}Y(t,0). 1.4. Heuristics: leading-order estimates On assuming that a sufficiently regular solution u\colon(0,\infty)^{2}\to\mathbb{R} to ( 1.13 ) already exists, in this subsection we derive a-priori estimates for u . To this end, we test ( 1.13a ) with u in space and, integrating by parts twice, we obtain (1.14a) \frac{1}{2}\frac{{\rm d}}{{\rm d}t}\int_{0}^{\infty}u^{2}\,{\rm d}x+\int_{0}^{\infty}x^{\alpha+2}\,|\partial_{x}^{2}u|^{\alpha+1}\,{\rm d}x=\int_{0}^{\infty}f\,u\,{\rm d}x. Next, we test ( 1.13a ) with \partial_{x}^{2}(x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u)) and get with \displaystyle\int_{0}^{\infty}u_{t}\,\partial_{x}^{2}(x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u))\,{\rm d}x=\int_{0}^{\infty}x^{\alpha+2}\,(\partial_{x}^{2}u_{t})\,g_{\alpha}(\partial_{x}^{2}u)\,{\rm d}x=\frac{1}{\alpha+1}\frac{{\rm d}}{{\rm d}t}\int_{0}^{\infty}x^{\alpha+2}\,|\partial_{x}^{2}u|^{\alpha+1}\,{\rm d}x the identity \frac{1}{\alpha+1}\frac{{\rm d}}{{\rm d}t}\int_{0}^{\infty}x^{\alpha+2}|\partial_{x}^{2}u|^{\alpha+1}\,{\rm d}x+\int_{0}^{\infty}\big|\partial_{x}^{2}(x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u))\big|^{2}\,{\rm d}x=\int_{0}^{\infty}f\,\partial_{x}^{2}(x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u))\,{\rm d}x, which by Young’s inequality entails the estimate (1.14b) \frac{2}{\alpha+1}\frac{{\rm d}}{{\rm d}t}\int_{0}^{\infty}x^{\alpha+2}|\partial_{x}^{2}u|^{\alpha+1}\,{\rm d}x+\int_{0}^{\infty}\big|\partial_{x}^{2}(x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u))\big|^{2}\,{\rm d}x\leq\int_{0}^{\infty}f^{2}\,{\rm d}x. In time-integrated form, the identity ( 1.14a ) entails (1.15a) \displaystyle\sup_{t\geq 0}\int_{0}^{\infty}u^{2}\,{\rm d}x+2\int_{0}^{\infty}\int_{0}^{\infty}x^{\alpha+2}\,|\partial_{x}^{2}u|^{\alpha+1}\,{\rm d}x\,{\rm d}t\leq 2\int_{0}^{\infty}u_{0}^{2}\,{\rm d}x+4\int_{0}^{\infty}\int_{0}^{\infty}|f|\,|u|\,{\rm d}x\,{\rm d}t, where u_{0}\colon(0,\infty)\to\mathbb{R} is the initial data. Time integration of estimate ( 1.14b ) leads to \displaystyle\frac{2}{\alpha+1}\sup_{t\geq 0}\int_{0}^{\infty}x^{\alpha+2}|\partial_{x}^{2}u|^{\alpha+1}\,{\rm d}x+\int_{0}^{\infty}\int_{0}^{\infty}\big|\partial_{x}^{2}(x^{\alpha+2}\,g_{\alpha}(\partial_{x}^{2}u))\big|^{2}\,{\rm d}x\,{\rm d}t (1.15b) \displaystyle\quad\leq\frac{4}{\alpha+1}\int_{0}^{\infty}x^{\alpha+2}|\partial_{x}^{2}u_{0}|^{\alpha+1}\,{\rm d}x+2\int_{0}^{\infty}\int_{0}^{\infty}f^{2}\,{\rm d}x\,{\rm d}t. On noting that \alpha>1 , estimates ( 1.15 ) suggest that \sup_{t,x\geq 0}|u| is controlled by these inequalities in conjunction with an interpolation inequality. Indeed, for p>1 we have, using the fundamental theorem of calculus and Hölder’s inequality,

Read the complete text