Unconditional Stability and Error Estimates of Modified Characteristics FEMs for the Navier–Stokes Equations

Zhiyong Si,Jilu Wang,Weiwei Sun

DOI: https://doi.org/10.1007/s00211-015-0767-9

IF: 2.5

2016-01-01

Numerische Mathematik

Abstract:The paper is concerned with the unconditional stability and convergence of characteristics type methods for the time-dependent Navier–Stokes equations. We present optimal error estimates in \(L^2\) and \(H^1\) norms for a typical modified characteristics finite element method unconditionally, while all previous works require certain time-step restrictions. The analysis is based on an iterated characteristic time-discrete system, with which the error function is split into a temporal error and a spatial error. With a rigorous analysis to the characteristic time-discrete system, we prove that the difference between the numerical solution and the solution of the time-discrete system is \(\tau \)-independent, where \(\tau \) denotes the time stepsize. Thus numerical solution in \(W^{1,\infty }\) is bounded and optimal error estimates can be obtained in a traditional way. Numerical results confirm our analysis and show clearly the unconditional stability and convergence of the modified characteristics finite element method for the time-dependent Navier–Stokes equations. The approach used in this paper can be easily extended to many other characteristics-based methods.