Previous Talks

Wednesday, April 8th, 2009

Adaptive Meshing in Geomechanics

Jack Chieslar, Independent Contractor

Abstract

Adaptive meshing is reviewed with the goal of balancing the competing objectives of a high quality calculation and a cost-effective one. Perspectives of both human time and effort as well as computational effort are considered. Attention is restricted to regular (structured) meshes assuring compatibility with reservoir simulation grids, which are typically of high refinement but of limited geographical extent, compared to their mechanical counterparts. The refinement is achieved with incompatible meshes. Examples in mechanical-thermal-porous flow are provided to establish the validity of these constrained, incompatible meshes. One advantage that this technique has over compatible mesh gradation techniques is that rapid rates of change of mesh resolution are supported allowing localized capturing of detail.

As a mesh generation technique refinement may be placed locally around a fluid mass source or sink, for example. The location of a high resolution mesh to capture such detail is known a priori and generally does not migrate. Other phenomena, such as plasticity, may become active in areas of the mesh not anticipated at the mesh generation phase. For cases such as these, adaptive meshing as an on-the-fly solution technique is demonstrated.

When compared to a mechanical-porous flow calculation, performed on a mesh with high resolution to capture flow detail and with this resolution propagated to the boundaries, substantially less computational effort is realized with adapted meshes. Yet solution detail is preserved. Simply, the computational effort is avoided where it is not required.  A well-shaped, fully-integrated, linear finite element (4-node quad, 8-node brick) will capture linear gradients in nodal variables.

Other conveniences accrue as a result of adaptive meshing. For post-processing,graphical and otherwise,adaptive meshing provides a natural basis for culling the output, leaving the regions with more refined meshing.

Speaker Biography

Jack Chieslar is an independent contractor working in the area of continuum mechanics software development. Jack has worked since 1986 developing nonlinear geomechanics software. Until 1996 he worked for Re/Spec, Inc., a contractor to Sandia Laboratories, where he developed two applications for creep of salt, supported by the Waste Isolation Pilot Plant (WIPP) in New Mexico. Since 1996 he has taken contract work and otherwise has been developing an application for coupled thermal-mechanical-porous flow. Jack has a PhD in Civil Engineering from the University of Calgary (1986).