Highly efficient linear energy stable methods for preserving the original energy dissipation law of the incompressible Navier-Stokes equation

Zihan Weng, Qi Hong, Yuezheng Gong \comma \corrauth · 2025

Read Highly efficient linear energy stable methods for preserving the original energy dissipation law of the incompressible Navier-Stokes equation by Zihan Weng, Qi Hong, Yuezheng Gong \comma \corrauth (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-2506-07141.txt.

Introduction

Zihan Weng and co-authors build computer methods for simulating fluid flow that respect the flow's own energy budget exactly. The target is the incompressible Navier-Stokes equations, the standard model of slow, dense fluids. Good simulations should let energy drain away at exactly the rate physics dictates; many methods distort that law.

The authors first recast the equations into a form designed for structure-preserving computation. On that foundation they build four time-stepping schemes, mixing first- and second-order accuracy, that keep the original dissipation law at every step while solving only a few linear systems. Space is handled with differences on a staggered grid, and the team proves each fully discrete method preserves the law and has a unique solution. Numerical experiments confirm the accuracy and speed of the schemes.

Catalog record

CREATORZihan Weng, Qi Hong, Yuezheng Gong \comma \corrauth
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-07141

Browse the full library · About the Tower of Babel

Full text

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

Highly efficient linear energy stable methods for preserving the original energy dissipation law of the incompressible Navier-Stokes equation Zihan Weng, Qi Hong, Yuezheng Gong \comma \corrauth Abstract In this paper, we introduce a comprehensive computational framework to construct highly efficient linear energy stable methods for the incompressible Navier-Stokes equation, which preserve the original energy dissipation law. By multiplying the convection term by an identity-one term and incorporating a zero stabilization term, we recast the original model as a strongly equivalent system, while ensuring the retention of the original energy dissipation law. Such nonlinear system is then discretized in time based on the Crank-Nicolson schemes and the backward differentiation formulas, resulting in highly efficient time-discrete schemes. The proposed schemes are designed to preserve the original energy dissipation law while requiring only the solutions of three linear Stokes systems and a 2 × 2 2\times 2 linear system at each time step. The finite difference approximation on a staggered grid is employed for the time-discrete systems to derive fully discrete energy stable schemes, which are proven to preserve the original energy dissipation law and be uniquely solvable. We present the efficient implementation of these methods. Various numerical experiments are carried out to verify the accuracy, efficacy, and advantageous performance of our newly developed methods. keywords Incompressible Navier-Stokes equations; Linear energy stable scheme; Original energy dissipation law; Staggered gird method 1 Introduction The incompressible Navier-Stokes(NS) equations play an important role in today’s fluid dynamics simulations. In this paper, we consider the dimensionless incompressible Navier-Stokes equations \displaystyle\left\{\begin{array}[]{lr}\bm{u}_{t}-\nu\Delta\bm{u}+\bm{u}\cdot\nabla\bm{u}+\nabla p=\bm{f},&\ \mathrm{in}\ \Omega\times(0,T],\\[4.26773pt] \nabla\cdot\bm{u}=0,&\ \mathrm{in}\ \Omega\times(0,T],\end{array}\right. where the velocity \bm{u}(\bm{x},t) subject to the initial condition \bm{u}(\bm{x},0)=\bm{u}_{0}(\bm{x}) and the Dirichlet boundary conditions or the periodic boundary conditions. \bm{x} and t are the spatial coordinates and time, respectively and in ( 1 ), \Omega is an open bounded domain in R^{d}(d=2,3) . The notation p(\bm{x},t) is the pressure and \nu is the kinematic viscosity which means the inverse of the Reynolds number Re . \bm{u}\cdot\nabla\bm{u} is the nonlinear convection and \bm{f} is an external force. There are various numerical methods for the incompressible NS equations, and one class of the most popular schemes is the projection method, which was originally proposed by Chorin [ 4 ] and Temam [ 27 ] . The projection methods consist of a prediction step for the viscous effects and a projection step for the divergence-free conditions, which successfully decouple the momentum equations and arrive at efficient calculation. Whereas, artificial boundary conditions are introduced for the pressure which can lead to boundary layers and reduce the convergence rates [ 13 , 7 , 8 ] . To tackle this issue, the gauge formulation was developed by using the Helmholtz-Hodge decomposition to reformulate ( 1 ) in terms of two auxiliary variables [ 9 , 15 ] . The main advantages of this formulation are that one can choose the boundary conditions flexibly and arrive at a decoupled system by treating the convection term explicitly. However, it still remains challenging for the schemes of gauge system to achieve energy stability. It is known that the incompressible NS equations satisfy the energy dissipation law with the periodic or the homogeneous Dirichlet boundary conditions (3) \displaystyle\bm{u}|_{\partial\Omega}=0. Given the kinetic energy E[\bm{u}]=\frac{1}{2}\int_{\Omega}|\bm{u}|^{2}\mathrm{d}\bm{x} in a smooth domain \Omega\in\mathbb{R}^{d} and ignoring the external force, system ( 1 ) has the energy dissipation law (4) \displaystyle\frac{\mathrm{d}}{\mathrm{d}t}E[\bm{u}]=-\nu\int_{\Omega}|\nabla\bm{u}|^{2}\mathrm{d}\bm{x}. By rewriting the convection term into an antisymmetry form and applying the Crank-Nicolson method, one can arrive a second-order scheme which maintains the energy dissipation law [ 10 ] unconditionally. However, such scheme is not efficient for it needs to solve a coupled nonlinear system with iteration technique at each time step. Recently, the scalar auxiliary variable (SAV) schemes were proposed for the gradient flow [ 24 , 25 ] which can be extended to the incompressible NS equations [ 20 , 18 , 19 ] and the rotating dynamics [ 29 ] . The SAV scheme reformulates system ( 1 ) by introducing a time-dependent auxiliary variables to transform the energy functional into a quadratic form. The schemes derived from new system allow for an explicit treatment of the convection terms and maintain a modified energy dissipation law. Additionally, the implementation of the SAV scheme is efficient for that it only includes solving two generalized Stokes equations and a nonlinear algebraic equation. Simultaneously, inspired by the application of Lagrange multiplier technique on the gradient flow [ 3 ] , a Lagrange multiplier was introduced into a dynamic equation for the original kinetic energy in [ 31 ] . The proposed schemes are combined with the projection method and satisfy the energy dissipation law with respect to the introduced Lagrange multiplier and some pressure gradient terms. Moreover, by treating the convection terms explicitly, the implementation also consists of two generalized Stokes equations and a nonlinear algebraic equation. In addition to that, Doan et al. modified the systems in [ 6 ] to ensure the uniqueness solution of the Lagrange multiplier and avoid iteration for solving the multiplier. For the incompressible NS equations with penalty terms, Yang et al. proposed an auxiliary variable viscosity splitting(AVVS) idea [ 26 , 30 ] , and constructed energy stable method for the simulation of fluid–solid interactions on a uniformed mesh. Besides the schemes above, notice that the convection and the stress term satisfy a so-called ”zero-energy-contribution” feature. To be specific, when deducing the energy dissipation law, after applying the inner products of the momentum equation with velocity, the results of the convection and velocity, together with the results of the gradient of pressure and velocity will completely cancel out. Therefore, by applying this property and combining the projection method, Yang et al. developed fully decoupled linear schemes for the phase field model (the DIEQ scheme in [ 32 ] and the DSAV scheme in [ 35 , 33 , 36 , 34 ] ). These schemes maintain a modified energy dissipation law containing the pressure gradient terms, the auxiliary terms and the stabilized terms which allow for calculation with large time steps. Although most of the energy stable schemes mentioned above are efficient to implement, which require solving several generalized Stokes equations (for the coupled approach) or a sequence of Poisson-type equations (for the decoupled approach) at each time step, it still remains unknown that how to construct a class of schemes for the incompressible NS equations which could maintain all the following properties: (i)An unconditionally original energy dissipation law can be satisfied; (ii)The schemes are linear and can be implemented efficiently; (iii)The schemes can be proved uniquely solvable. In this paper, we aim to develop a class of linear schemes for the incompressible Navier-Stokes equations to overcome the challenges mentioned above. At first, by introducing a zero stabilization term, we reformulate the original model ( 1 ) into an equivalent system. First- and second-order schemes based on the Crank-Nicolson method and backward differentiation formula are obtained with the explicit treatment of the nonlinear convection term. It holds that our schemes are linear and unconditionally maintain the original energy dissipation law. The unique solvability and fully discrete energy dissipation law are proved with the spatial discretization on the staggered grid [ 11 , 14 , 17 ] . Additionally,

Read the complete text