Predicting Time-Dependent Behavior of Post-Tensioned Concrete Beams: Discrete Multiscale Multiphysics Formulation

Abstract

Time-dependent deformations, including creep and shrinkage, are essential factors that govern multiple design aspects of prestressed/post-tensioned concrete structures. These include (but are not limited to) time to initial post-tensioning, prestressing losses, time to shoring removal, and serviceability in general. Excessive creep and shrinkage deformations can render a structure unusable aesthetically or even lead to eventual collapse. This is becoming more and more important because many of the recently developed advanced cementitious materials are characterized by larger and more evident long-term deformations (e.g., prolonged self-desiccation in high-strength concrete). This paper presents the prediction of long-term deformations of post-tensioned concrete beams due to creep, shrinkage, and steel relaxation under sustained loading and varying environmental conditions. This is achieved by using the lattice discrete particle model (LDPM) framework, in which time-dependent deformations are imposed at the coarse aggregate level following an explicit solidification-microprestress formulation and a code-based model for steel relaxation. Time-dependent deformations are formulated as functions of spatial and temporal evolutions of temperature, humidity, and cementitious materials’ hydration within the concrete mesostructure, which are modeled by using a semidiscrete multiphysics hygro-thermo-chemical (HTC) model. The coupling between the different models allows for capturing the time-dependent deformations relevant to the different design stages of post-tensioned concrete beams. To show the predictive capabilities of the proposed multiscale physics-based framework, all model parameters are calibrated by simulating the response of companion specimens (lab scale) only, then used to predict blindly the behavior of full-scale post-tensioned beams. The predictions show very good agreement with experimental data.