Luigi C. Berselli,Alex Kaltenbach
Abstract:In this paper, we examine a finite element approximation of the steady $p(\cdot)$-Navier-Stokes equations ($p(\cdot)$ is variable dependent) and prove orders of convergence by assuming natural fractional regularity assumptions on the velocity vector field and the kinematic pressure. Compared to previous results, we treat the convective term and employ a more practicable discretization of the power-law index $p(\cdot)$. Numerical experiments confirm the quasi-optimality of the a priori error estimates (for the velocity) with respect to fractional regularity assumptions on the velocity vector field and the kinematic pressure.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about error analysis in the finite - element approximation of the steady - state p(·)-Navier - Stokes equations. Specifically, the authors aim to:
1. **Study the finite - element approximation**: Conduct a finite - element approximation of the steady - state Navier - Stokes equations with a variable exponent \(p(·)\), and prove the convergence order under the assumption that the velocity vector field and the dynamic pressure have natural fractional - order regularity.
2. **Handle the convection term**: Compared with previous works, this paper specifically deals with the convection term \([∇v]v\) and adopts a more practical discretization method for the power - law exponent \(p(·)\).
3. **Verify the quasi - optimality of a priori error estimates**: Confirm the quasi - optimality of a priori error estimates (especially for velocity) with respect to the fractional - order regularity assumptions of the velocity vector field and the dynamic pressure through numerical experiments.
### Specific problem description
The steady - state p(·)-Navier - Stokes equations can be expressed as:
\[
-\text{div} S(·, Dv) + [\nabla v]v + \nabla q = f \quad \text{in } \Omega,
\]
\[
\text{div} v = 0 \quad \text{in } \Omega,
\]
\[
v = 0 \quad \text{on } \partial \Omega.
\]
where:
- \( \Omega \subseteq \mathbb{R}^d \) is a bounded polyhedral Lipschitz domain.
- \( f: \Omega \to \mathbb{R}^d \) describes the external force.
- \( v: \Omega \to \mathbb{R}^d \) is the velocity vector field.
- \( q: \Omega \to \mathbb{R} \) is the dynamic pressure.
- \( S(x, A) := \mu_0 (\delta + |A_{\text{sym}}|)^{p(x)-2} A_{\text{sym}} \), where \( A_{\text{sym}} \) is the symmetric part of the strain - rate tensor \( Dv \).
### Main contributions
1. **Four - point quadrature rule**: Propose a general one - point quadrature rule to define the cell - constant approximation of the power - law \(p(·)\), instead of finding the minimum power - law exponent \(p(·)\) on each cell, thereby avoiding potentially slow and high - cost computations.
2. **Fractional - order regularity and error decay rate**: Consider a non - Lipschitz but only H¨older continuous power - law exponent \(p \in C^{0,\alpha}(\Omega)\), and expect to obtain partial regularity \(F(·,Dv) \in N^{\beta,2}(\Omega; \mathbb{R}^{d \times d})\) and \(q \in H^{\gamma,p'(·)}(\Omega)\).
3. **Handling of the convection term**: Handle the convection term under a non - constant power - law exponent \(p \in C^{0,\alpha}(\Omega)\) and give solutions only under the fractional - order regularity assumption.
4. **Quasi - optimality**: Establish the quasi - optimality of the derived error decay rate (for velocity) with respect to the fractional - order regularity assumptions of velocity and pressure through numerical experiments.
In summary, this paper is dedicated to improving and expanding the understanding of the finite - element approximation of the steady - state p(·)-Navier - Stokes equations, especially in handling the convection term and the fractional - order regularity assumption.