Two multiscale discrete-continuum models for predicting fluid-saturated geomaterials from grain- to field-scales
Prof. Steve Waiching Sun
Department of Civil Engineering and Engineering Mechanics
Columbia University, New York
Monday, August 10, 2015
2:00 PM – 3:00 PM
Building 823 Room 2279
The mechanical behavior of a fluid-infiltrating porous solid is significantly influenced by the presence and diffusion of the pore fluid in the void. This hydro-mechanical coupling effect can be observed in a wide range of materials, including rocks, soils, concretes, bones and soft tissues. Due to the high computational demand, explicitly simulating the pore scale solid-fluid interactions remains impractical for engineering problems commonly encountered in the field and basin scales. The objective of this talk is to present two classes of multiscale technologies that couple across different spatial and temporal scales.
The first class of model is a concurrent coupling model in which deformation-diffusion problems are casted as the two-fold saddle point that optimizes the constrained partitioned incremental work of a multi-field energy functional. By enforcing compatibility across length scales, pore-scale simulations in confined domain can be coupled with large-scale field problems while maintaining numerical stability and accuracy. The combined usage of temporal operator split and Arlequin model to resolve highly refined details of space-time porous continuum will be discussed.
The second class of multiscale model is a nonlocal hierarchical multiscale framework that couples grain-scale discrete element simulations with a hydromechanical mixed finite element model. This hierarchical nonlocal DEM-mixed-FEM coupling retains the simplicity and efficiency of the continuum-based finite element model, while possessing the original length scale of the granular system. Techniques for two-scale material identification with Dakota will be discussed. The pros and cons of these two different coupling strategies will be demonstrated in numerical examples.
News about Computational Poromechanics lab at Columbia University.