Evaluation of Concrete Constitutive Models for Finite Element Simulation of Dam Shear Keys

Abstract

Three-dimensional performance-based structural assessment of concrete dams subjected to floods and earthquakes requires an accurate estimation of inter-monolith-relative displacements. These displacements could be limited by the presence of shear keys, used in many concrete gravity and arch dams’ contraction joints as interlocking mechanisms for transferring shear forces between adjacent monoliths. The key shear strength that could be mobilized is a function of several parameters that control the load–displacement response and failure mechanisms: friction, cohesion, key geometry, confinement pressure, etc. The influence of these parameters is not well understood. This paper first presents an evaluation of five concrete constitutive models implemented in three commercially available general-purpose finite element software packages for numerically simulating the behavior of plain concrete shear keys (ABAQUS, LS-Dyna, ATENA). First, a brief review of each model is presented, including their theoretical formulations, required inputs, advantages, and limitations. Experimental data on shearing of trapezoidal shear keys, available from the literature, are used to evaluate the performance of each constitutive model. The study is then extended to different shear key geometries. It is shown that the load–displacement responses, ultimate shear capacities, failure mechanisms, and post-peak softening behavior can be accurately predicted by a well-calibrated LS-Dyna Concrete Surface Cap Model. It is also shown that the geometry and the boundary conditions, with either unrestrained or restrained lateral shear dilatancy, can greatly influence the failure mechanisms and consequently the shear key load-bearing capacity and residual strength.