Performance of Simplified Damage-Based Concrete Models in Seismic Applications

Abstract

Significant progress in the finite-element (FE) modeling at the member-level of reinforced-concrete (RC) structures under seismic excitation has been achieved in the past decades; however, reliable and accurate analysis models for full-scale 3D system-level RC structures validated with experimental data are scarce. As the development of a complete nonlinear model is expensive and time consuming, simpler models, typically elastic or lumped-plastic in nature, are employed in practice with additional provisions prescribed to account for nonlinear behavior. Depending on the assumptions made by the analyst, there may be substantial uncertainties related to the response of a complete structure. Capturing global failure modes is challenging, and the assessment of strength and ductility capacities may be inaccurate, resulting in potentially unsafe designs. The objective of this research is to assess the performance of simplified damage-based concrete biaxial models in analyzing and capturing the structural behavior of full-scale RC systems. Damage-based models for concrete require minor input from the analyst, facilitating their use in a design setting, while their explicit, non-conditional convergence formulations allow for non-iterative solutions. This results in damage-based models featuring efficient computational analysis while accounting for complex phenomena such as the capacity to account for stiffness recovery in reversal loading (crack closing) and permanent strains in the concrete. A comparison between analytical and experimental data at both the element- and system-levels is conducted, and a viable damage-based model is proposed for a full-scale structure analysis. The results of the study show that damage-based models are a viable alternative to developing efficient analysis models for elements and whole structures.