Speaker
Description
Planar or thin-walled elements of structural concrete are ubiquitous within the civil infrastructure. Common examples include shear walls and floor slabs in building construction. The analyses of such structural elements for out-of-plane loading is challenging, due to the concomitant needs for capturing the distribution of damage in the plane and through the thin wall dimension. Three-dimensional models are apt for this purpose, but are computationally expensive, to the extent that analyses are limited to individual or small assemblages of structural members. Extension to larger systems, or more effective parametric studies of moderately sized systems, requires efficient modeling techniques. Layered finite element models help address concerns with computational expense, but such formulations rely on continuum representations of damage, which are problematic when considering cracking and other forms of displacement discontinuity.
This research utilizes a layered Voronoi-cell lattice model (L-VCLM), which is based on a two-dimensional discretization of the planar or thin-walled structure. Similar to layered finite element approaches, kinematic constraints are used to model elastic behavior and nonlinear material response (e.g., plastic yielding or crack formation) and its propagation in the wall thickness direction. The discrete nature of the L-VCLM has advantages in simulating distributed cracking and the process of fracture localization. Whereas the L-VCLM is elastically uniform under in-plane (membrane-type) loading, it exhibits spurious heterogeneity under flexural loading. However, such errors disappear with grid refinement. This suggests adaptive grid refinement as a means for modeling nonlinear behavior in large systems, which is the main topic of this research. Grid refinement is guided by cluster analyses based on maps of damage events produced from coarse-grid model(s). The process is first developed for one-way structural elements and then generalized for planar systems. Accuracy of the L-VCLM is established through benchmark comparisons with theoretical and physical test results.
