EPSRC logo
 Home | GoW Home | Back | Programme | Scheme | Topic | Sector | Theme | Region | Organisation     
 
Details of Grant
 
EPSRC Reference: GR/M00077/01
Title: ANISOTROPIC MESH ADAPTION AND ERROR ESTIMATION FOR THE F.E. SOLUTION OF 3-D CONVECTION DOMINATED P.D.E.S.
Principal Investigator: Professor PK Jimack
Other Investigators:
Professor M Berzins
Researcher Co-investigator:
Project Partner:
Shell Global Solutions UK
Department: Sch of Computing
Organisation: University of Leeds
Scheme: Standard Research
Starts: 01 December 1998 Ends: 30 November 2000 Value (£): 87,655
EPSRC Research Topic Classifications:
Numerical Analysis
EPSRC Industrial Sector Classifications:
Related Grants:
Panel History:  
Summary
The main purpose of this research is to develop and apply directional error estimates and associated anisotropic mesh refinement strategies for the finite element solution of steady and time-dependent convection dominated problems in three dimensions. The starting point for this work will be our own research experience using anisotropic refinement in 2-d and isotropic refinement in 3-d, along with recently published work on a posteriori error estimates for hyperbolic problems.

The main new contribution will be to modify existing a posteriori error estimates so as to provide adaptive software with reliable information on how to refine in given directions throughout the solution domain. The feasibility of this work is suggested by the successful experiences of others for elliptic problems and, importantly, by the results of a priori analysis which clearly demonstrate that anisotropic refinement can yield much better representations of typical solutions than isotropic refinement.

To demonstrate the advantages of using our new error estimates and our anisotropic refinement strategies, these will be applied to the solution of two non-trivial test problems, including a challenging nonlinear gas jet problem for which we and our collaborators have significant 2-d (axisymetric) experience. Advantages over using conventional isotropic refinement will be quantified.

Final Report Summary
The main purpose of this research is to develop and apply directional error estimates and associated anisotropic mesh refinement strategies for the finite element solution of steady and time-dependent convection dominated problems in three dimensions. The starting point for this work will be our own research experience using anisotropic refinement in 2-d and isotropic refinement in 3-d, along with recently published work on a posteriori error estimates for hyperbolic problems.

The main new contribution will be to modify existing a posteriori error estimates so as to provide adaptive software with reliable information on how to refine in given directions throughout the solution domain. The feasibility of this work is suggested by the successful experiences of others for elliptic problems and, importantly, by the results of a priori analysis which clearly demonstrate that anisotropic refinement can yield much better representations of typical solutions than isotropic refinement.

To demonstrate the advantages of using our new error estimates and our anisotropic refinement strategies, these will be applied to the solution of a number of test equations. Advantages over using conventional isotropic refinement will be quantified.

Further Information:  
Organisation Website: http://www.leeds.ac.uk
Terms and conditions