Prediction of time-dependent displacements from concrete to structure under creep and shrinkage: A data-poromechanics-driven study

Kefu Yao, Matthieu Vandamme, Jean-Michel Torrenti, Huaizhi Su

Abstract:

The time-dependent displacement is a crucial indicator of the long-term safety of concrete structures, which have decades to centuries of service life, as it reflects both material properties and structural behavior. Little attention is systematically devoted to bridging material-level behavior and structural response in time-dependent deformation and displacement. For this reason, our study provides a solution, from understanding the physical phenomena to engineering applications, to characterize the long-term behavior of concrete structures submitted to creep and shrinkage. The main works include:

• A poromechanical constitutive model based on the work of Aili et al. (Aili et al., Cem Concr Res, 2020) is derived for coupled creep and shrinkage, incorporating mass and energy balances in an unsaturated poro-viscoelastic framework. We implemented the poromechanical model computationally by developing toolsets, which include preprocessing modules, mechanical subroutines, and posttreatment analysis. The developed tool employs a stepwise recursive increment algorithm, allowing for calculating viscoelastic hydric stress and retrieving delayed strains from creep and shrinkage under unified parameter settings, thereby reducing the computational overhead of storing stress and strain histories.
• Comparisons with long-term laboratory and field data, which encompasses moisture transport and deformation under concrete drying, creep, and shrinkage experiments, and actual concrete dam displacement, indicate that i) The proposed model reasonably captures the delayed strains arising from concrete creep and shrinkage. ii) At the concrete dam crest, the time-dependent dam displacement over 7 years constitutes a non-negligible portion of the total dam displacement.
• Furthermore, to enhance the generalization capability of the proposed poromechanical model in dam engineering, an intelligent multi-objective back analysis algorithm is incorporated, enabling the real-time prediction of time-dependent dam displacement based on field data. Additionally, in the spirit of the linear poro-viscoelasticity used in the poromechanical model, an interpretable data-driven method is proposed for extracting time-dependent dam displacement from field data.
  

Short bio:

Hello everyone! I began my PhD in September 2021 at Hohai University. Since February 2023, I have been working under the supervision of Prof. Matthieu Vandamme and Prof. Jean-Michel Torrenti. My PhD topic focuses on the delayed deformations and time-dependent displacements of structural concrete submitted to creep and shrinkage.