A Deep Learning Approach to Design and Discover Sustainable Cementitious Binders: Strategies to Learn From Small Databases and Develop Closed-form Analytical Models

Abstract

To reduce the energy-intensity and carbon footprint of Portland cement (PC), the prevailing practice embraced by concrete technologists is to partially replace the PC in concrete with supplementary cementitious materials [SCMs: geological materials (e.g., limestone); industrial by-products (e.g., fly ash); and processed materials (e.g., calcined clay)]. Chemistry and content of the SCM profoundly affect PC hydration kinetics; which, in turn, dictates the evolutions of microstructure and properties of the [PC + SCM] binder. Owing to the substantial diversity in SCMs’ compositions–plus the massive combinatorial spaces, and the highly nonlinear and mutually-interacting processes that arise from SCM-PC interactions–state-of-the-art computational models are unable to produce a priori predictions of hydration kinetics or properties of [PC + SCM] binders. In the past 2 decades, the combination of Big data and machine learning (ML)—commonly referred to as the fourth paradigm of science–has emerged as a promising approach to learn composition-property correlations in materials (e.g., concrete), and capitalize on such learnings to produce a priori predictions of properties of materials with new compositions. Notwithstanding these merits, widespread use of ML models is hindered because they: 1) Require Big data to learn composition-property correlations, and, in general, large databases for concrete are not publicly available; and 2) Function as black-boxes, thus providing little-to-no insights into the materials laws like theory-based analytical models do. This study presents a deep learning (DL) model capable of producing a priori, high-fidelity predictions of composition- and time-dependent hydration kinetics and phase assemblage development in [PC + SCM] pastes. The DL is coupled with: 1) A fast Fourier transformation algorithm that reduces the dimensionality of training datasets (e.g., kinetic datasets), thus allowing the model to learn intrinsic composition-property correlations from a small database; and 2) A thermodynamic model that constrains the model, thus ensuring that predictions do not violate fundamental materials laws. The training and outcomes of the DL are ultimately leveraged to develop a simple, easy-to-use, closed-form analytical model capable of predicting hydration kinetics and phase assemblage development in [PC + SCM] pastes, using their initial composition and mixture design as inputs.

Predicting mechanical properties of ultrahigh temperature ceramics using machine learning

“…Moreover, unique features designed for UHTCs can be affixed to ML models to reduce the complexity of the database. Our studies 73,74 have demonstrated that the customized features can significantly improve the prediction accuracy of ML models for complex materials.…”

On the Prediction of the Mechanical Properties of Limestone Calcined Clay Cement: A Random Forest Approach Tailored to Cement Chemistry

“…Meanwhile, the variable importance provides critical knowledge to develop analytical models. Our previous studies [53,65,67,74,[81][82][83] successfully harnessed this tool to craft user-friendly, closed-form analytical models for different materials.…”

“…This combination of bagging and two-step randomization effectively reduces both variance and bias errors, enhancing the model's reliability [49,50]. To avoid overfitting and underfitting, the hyperparameters of RF are optimized via the 10-fold cross-validation (CV) [51,52] and grid-search methods [51,53].…”

Predicting Dissolution Kinetics of Tricalcium Silicate Using Deep Learning and Analytical Models

“…The rigorously trained model produces high-fidelity, a priori predictions of the C3S dissolution rate. It is notable that ML models can predict the hydration kinetics of PC at any given age, which has been shown in our previous studies [51][52][53]. This study only focuses on the dissolution kinetics at the initial period (i.e., undersaturated solution) because the hydration products precipitate and cause the solution to reach saturation after a short time of the dissolution of C3S.…”