Prof Sonia Antoranz Contera

Abstract:

This study presents an integrated finite-element–machine-learning framework for predicting early-age stress evolution in concrete materials/structures by combining an enhanced thermo-chemo-mechanical (TCM) model, deep sequential learning (DSL), and active learning (AL). The proposed TCM model incorporates experimentally informed viscoelasticity, a stable exponential creep–relaxation conversion, and an efficient exponential algorithm for the Maxwell-chain formulation in finite element analysis, which is further validated by a temperature stress testing machine. This model generates high-fidelity stress–time data across diverse mixtures, temperatures, and structural configurations. These simulations are used to train a Gated Recurrent Unit with Monte Carlo Dropout (GRU-MCD) model, whose predictive performance surpasses conventional point-wise approaches such as Light Gradient Boosting Machine and Gaussian Process Regression, yielding higher accuracy with reduced overfitting. The AL strategy further enhances efficiency by enabling the GRU-MCD model to achieve the accuracy of ∼900 Latin Hypercube samples using only ∼200 samples selected by active learning. Although demonstrated on a wall–base structure, the proposed framework is general and applicable to other cementitious material or structural systems, providing an effective tool for cracking-risk evaluation, reliability analysis, and the design of low-carbon concrete structures.

Abstract:

This study presents an experimentally informed coupled creep–damage modelling framework for the time-dependent micromechanical behaviour of hardened cement paste. Using realistic microstructures, a viscoelastic formulation solved by an exponential algorithm, and a continuum damage model, the framework consistently captures microscale creep, creep recovery, and strain-rate-dependent response. The results show that microstructural discretization strongly affects predictions: coarse discretizations underestimate porosity and overestimate hydration, leading to overpredicted stiffness and strength. The model reproduces measured flexural strength, elastic modulus, and the observed brittle failure. Low-stress creep and recovery are predicted accurately, with high recovery ratios indicating predominantly linear creep. The calibrated HD-CSH/ LD-CSH creep modulus ratio is consistent with experimental insights, supporting that calcium hydroxide increases the creep modulus of CSH. Strain-rate effects are also captured: slower loading allows more creep and earlier damage in weaker phases, while faster loading drives damage into stronger phases, yielding higher apparent stiffness and strength with more pronounced damage patterns. Overall, the framework provides a physically consistent basis for studying microscale creep–damage interactions. Future work will incorporate temperature and moisture effects and extend the approach to multiscale simulations of long-term concrete behavior under realistic environments.

Abstract:

A novel methodology, enhanced by artificial intelligence (AI), to achieve zero emission building (ZEB) designs is being developed within the European ZEBAI project. The methodology will integrate all interdependent analyses and partial decision-making processes into a holistic approach that simultaneously assesses the building’s energy performance, environmental impact, indoor environmental quality and economic costs. The results of a building energy simulation are significantly influenced by the accuracy of the heat and moisture transfer calculations of the individual components of the building envelope. Consequently, a comprehensive and up-to-date database of building material properties is required to obtain reliable simulation results. However, most databases on the market do not include parameters essential for accurate energy calculations, such as optical properties (solar and visible absorption, reflectance and transmittance), colour coordinates and thermal emittance of envelope materials. Furthermore, the behaviour of these properties with changes in temperature and humidity, as well as the coupled mechanical and thermal properties of building materials, are rarely considered. As a result, engineers/technicians are forced to adopt standardised values that may differ significantly from the true values, increasing the discrepancy between simulated and measured energy performance of buildings. This work presents an overview of the existing databases at European level containing information on the mechanical, optical and thermal properties of building materials. From their analysis, the requirements for the development of a new catalogue of materials suitable for AI-enhanced design based on machine learning and heuristic optimisation of ZEBs are established, i.e. a characterisation plan for materials and systems, the appropriate conditions for processing the characterisation data and collecting the results, and the proposed format for the final catalogue.

Abstract:

The mechanical properties of the polymeric substrate or matrix where a cell grows affect cell behavior. Most studies have focused on relating elastic properties of polymeric substrates, which are time-independent, to cell behaviors. However, polymeric substrates and biological systems exhibit a time-dependent, often viscoelastic, mechanical response. While less is known about how time responses dictate cell behavior, cells are likely sensitive to substrate time responses rather than elasticity alone. However, testing this hypothesis is complex due to the lack of nanoscale tools. To overcome this limitation, photothermal actuation-atomic force microscopy nanoscale dynamic mechanical analysis (PT-AFM nDMA), a novel AFM technique that measures sample viscoelasticity over a broad and continuous frequency range, was applied to measure the viscoelastic properties of cell culture substrates made of poly(2-hydroxyethyl methacrylate) (poly(HEMA)) and collagen I (pureCol) in liquid at frequencies ranging continuously from 0.1 Hz–5000 Hz. PT-AFM nDMA to date has not been performed in liquid, but successfully and accurately characterized substrate viscoelasticity and synergized with measurements obtained using a more established AFM technique, bimodal imaging. The results of this study demonstrate that PT-AFM nDMA can be performed in liquid environments, especially relevant to biological samples. Additionally, comparing PT-AFM nDMA measurements of the poly(HEMA) and pureCol substrates in this study to cell behaviors described in the scientific literature on similar substrates suggests that longer substrate time responses at low measurement frequencies promote cell attachment, proliferation, and migration, while shorter substrate time responses promote ECM remodeling and differentiation.