A posteriori error estimates for the finite element approximation of the convection–diffusion–reaction equation based on the variational multiscale concept

Ramon Codina, Hauke Gravenkamp, Sheraz Ahmed Khan · 2025

Read A posteriori error estimates for the finite element approximation of the convection–diffusion–reaction equation based on the variational multiscale concept by Ramon Codina, Hauke Gravenkamp, Sheraz Ahmed Khan (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-2505-02531.txt.

Introduction

Ramon Codina, Hauke Gravenkamp, and Sheraz Khan turn a stabilization trick into an error gauge for flow simulations. The convection-diffusion-reaction equation models transport with spread and source terms. Finite element answers need honest error bars, above all when flow dominates. Their multiscale method splits the problem into solved coarse scales and modeled fine scales.

The fine-scale model, tied to the leftover residual, doubles as the error estimate. Contributions count both inside elements and along their edges. A scaled norm of those fine scales prices the error in a stabilized measure. That measure keeps grip on the flow term itself. Tests show the gauge tracks true error reliably against rival estimators.

Catalog record

CREATORRamon Codina, Hauke Gravenkamp, Sheraz Ahmed Khan
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE74 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-02531

Browse the full library · About the Tower of Babel

Full text

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

A posteriori error estimates for the finite element approximation of the convection–diffusion–reaction equation based on the variational multiscale concept Ramon Codina, Hauke Gravenkamp, Sheraz Ahmed Khan Abstract In this study, we employ the variational multiscale (VMS) concept to develop a posteriori error estimates for the stationary convection-diffusion-reaction equation. The variational multiscale method is based on splitting the continuous part of the problem into a resolved scale (coarse scale) and an unresolved scale (fine scale). The unresolved scale (also known as the sub-grid scale) is modeled by choosing it proportional to the component of the residual orthogonal to the finite element space, leading to the orthogonal sub-grid scale (OSGS) method. The idea is then to use the modeled sub-grid scale as an error estimator, considering its contribution in the element interiors and on the edges. We present the results of the a priori analysis and two different strategies for the a posteriori error analysis for the OSGS method. Our proposal is to use a scaled norm of the sub-grid scales as an a posteriori error estimate in the so-called stabilized norm of the problem. This norm has control over the convective term, which is necessary for convection-dominated problems. Numerical examples show the reliable performance of the proposed error estimator compared to other error estimators belonging to the variational multiscale family. 1 Introduction The finite element (FE) method is a numerical technique to approximate solutions of partial differential equations computationally and is widely used to solve engineering problems. In fluid mechanics, applying the standard Galerkin FE method proves to be challenging in two main ways. Firstly, the numerical approximation of convection-diffusion type problems lacks stability when convection is dominant and exhibits numerical oscillations. Secondly, it is frequently difficult to satisfy the inf-sup stability condition for mixed interpolations. Alternatively, stabilized FE methods can be used, which contain additional stabilization terms and provide proper stability without the need to satisfy the inf-sup conditions. Moreover, the singularly perturbed nature of the continuous problem causes the stability bound to explode when diffusion (or viscosity) approaches zero. This drawback is mitigated when stabilized methods are employed. The addition of new stabilization terms enables control over the norm of the convective term and enhances accuracy. A comprehensive comparison of stabilization techniques applied to the convection-diffusion-reaction equation (CDRE) can be found in [ 1 ] . The variational multiscale (VMS) method introduced by Hughes et al. [ 2 , 3 ] serves as the foundation to derive the stabilized method. The point of departure is the notion that the FE approximation cannot capture the solution precisely, hence, the VMS method is based on splitting the continuous solution into a resolved scale (coarse scale) and an unresolved scale (fine scale), where the unresolved scale, also known as the sub-grid scale (SGS), is represented or estimated analytically. The SGS model we have used is known as the orthogonal SGS (OSGS) method, introduced in [ 4 ] . In the OSGS approach, the SGS problem is modeled in a specific fashion by choosing the SGSs orthogonal to the FE space. The stability and convergence analysis of this method for the stationary and transient CDRE are presented in [ 5 ] and [ 6 ] , respectively. The complete analysis for the Oseen problem, including non-uniform FE meshes, can be found in [ 7 ] . A detailed discussion of the VMS method and its application to computational fluid dynamics problems can be found in [ 8 ] . While substantial expertise has been developed in solving physical problems, it has remained a computational challenge to predict and control the error of the computed solutions. In recent years, a posteriori error estimation has made significant progress, becoming an essential tool for FE practitioners. An overview of the topic is given by Ainsworth and Oden in [ 9 ] . See also [ 10 ] for an overview of a posteriori error estimation using the Galerkin method, including problems related to fluid mechanics. A recent survey on the development of a posteriori error estimation in fluid mechanics based on the VMS theory has been conducted by Hauke and Irisarri in [ 11 ] . In general, a posteriori error estimation is a broad discipline that originated with Babuška and Rheinbolt [ 12 ] for elliptic problems and was later successively expanded by Zienkiewicz and Zhu [ 13 , 14 ] , Eriksson and Johnson [ 15 ] , and Ainsworth and Oden [ 16 ] , among others. In addition, Verfürth investigated a posteriori error estimates for the Stokes problem [ 17 ] , for the convection-diffusion problem [ 18 ] , and for elliptic problems [ 19 ] . Moreover, we refer to the recent work using different numerical methods for the Navier-Stokes problem in [ 20 ] , for semi-linear elliptic problems in [ 21 ] , and for linear and non-linear singularly perturbed problems in [ 22 ] . Previously, several studies have been conducted to estimate the a posteriori error for convection-diffusion-type problems. In [ 23 ] , an a posteriori error analysis was performed for the stationary and transient CDRE using stabilized FE schemes. In this reference, error estimates are measured in an energy-like norm for the symmetric part of the differential operator, while a dual norm is employed to bound the convective term. Likewise, Verfürth used the same strategy to measure the error for stationary CDREs in [ 24 ] , for non-stationary CDREs in [ 25 ] , and for non-linear non-stationary CDREs in [ 26 ] . A similar approach has been by followed by Sharma [ 27 ] to explore a posteriori error estimates for the weak Galerkin method applied to the CDRE. Moreover, a similar choice of norm has been made by Ainsworth et al. in [ 28 ] , where the authors derived a posteriori error estimators using the streamline-upwind Petrov-Galerkin (SUPG) method to approximate solutions of the stationary CDRE in three dimensions. Du et al. [ 29 ] also investigated recovery-type a posteriori error estimates using the SUPG method for singularly perturbed problems. Already in the initial development of the VMS method, it was suggested by Hughes et al. [ 3 ] to use the SGSs as an a posteriori error estimator. One of the first attempts to estimate a posteriori the error using the VMS technology for convection-dominated problems was presented in [ 30 ] , verifying that VMS-based stabilized methods are equipped with an inherent error estimator. In recent years, several studies have been conducted to derive a posteriori error estimates using the VMS method [ 31 , 32 , 33 , 34 , 35 ] . The same strategy is applied to higher-order elements in [ 36 , 32 ] and later extended to 2D domains in [ 34 ] , where the error contribution on the element boundaries is also discussed. In addition, error estimators have been derived for the Navier-Stokes equations in [ 37 , 35 , 38 ] , for the Stokes equations in [ 39 ] , for linear elasticity in [ 33 ] , and for higher-order ordinary differential equations in [ 40 ] , all based on the same concepts. In this study,

Read the complete text