A Unified Norm Error Analysis of SAV-BDF Schemes for the Incompressible Navier–Stokes Equations

Bingquan Ji,Hong-lin Liao
DOI: https://doi.org/10.1007/s10915-024-02555-9
2024-05-23
Journal of Scientific Computing
Abstract:This article is concerned with the norm error analysis of high-order BDF methods for the incompressible Navier–Stokes equation subjected to no-slip boundary conditions. To mitigate the complexity of high-order time-stepping algorithms caused by decoupling strategies, we combine the prediction–correction technique with a generalized scalar auxiliary variable approach to devise a family of linear, energy stable BDF schemes up to fifth-order accuracy in time. All the proposed schemes can be decoupled into a sequence of Poisson-type equations for the pressure and velocity fields, which significantly reduces the size of the linear systems at each time step. To deal with the essential difficulty raised by the non-A-stability of high-order BDF schemes, we introduce a class of discrete orthogonal convolution kernels to develop a unified framework for the norm convergence analysis of the proposed high-order BDF schemes. The velocity attains optimal rate of convergence in the norm under some mild mesh restrictions. Furthermore, building on the recent results concerning the positive definiteness of high-order BDF convolution kernels, we also provide an error estimate of the pressure in a weak norm. Benchmark examples including the Taylor–Green vortex problem and the regularized lid-driven cavity flow with Reynolds numbers up to 5000 are included to verify the stability and high accuracy of our numerical schemes.
mathematics, applied
What problem does this paper attempt to address?