arXiv subjects

Gaurav Sant

Publications and source records attributed to Gaurav Sant.

7 recordsLinked to original sources

arXiv2025

A Roadmap for Applying Graph Neural Networks to Numerical Data: Insights from Cementitious Materials

Machine learning (ML) has been increasingly applied in concrete research to optimize performance and mixture design. However, one major challenge in applying ML to cementitious materials is the limited size and diversity of available databases. A promising solution is the development of multi-modal databases that integrate both numerical and graphical data. Conventional ML frameworks in cement research are typically restricted to a single data modality. Graph neural network (GNN) represents a new generation of neural architectures capable of learning from data structured as graphs, capturing relationships through irregular or topology-dependent connections rather than fixed spatial coordinates. While GNN is inherently designed for graphical data, they can be adapted to extract correlations from numerical datasets and potentially embed physical laws directly into their architecture, enabling explainable and physics-informed predictions. This work is among the first few studies to implement GNNs to design concrete, with a particular emphasis on establishing a clear and reproducible pathway for converting tabular data into graph representations using the k-nearest neighbor (K-NN) approach. Model hyperparameters and feature selection are systematically optimized to enhance prediction performance. The GNN shows performance comparable to the benchmark random forest, which has been demonstrated by many studies to yield reliable predictions for cementitious materials. Overall, this study provides a foundational roadmap for transitioning from traditional ML to advanced AI architectures. The proposed framework establishes a strong foundation for future multi-modal and physics-informed GNN models capable of capturing complex material behaviors and accelerating the design and optimization of cementitious materials.

cs.CE↗

arXiv2020

Learning from Sparse Datasets: Predicting Concrete's Strength by Machine Learning

Despite enormous efforts over the last decades to establish the relationship between concrete proportioning and strength, a robust knowledge-based model for accurate concrete strength predictions is still lacking. As an alternative to physical or chemical-based models, data-driven machine learning (ML) methods offer a new solution to this problem. Although this approach is promising for handling the complex, non-linear, non-additive relationship between concrete mixture proportions and strength, a major limitation of ML lies in the fact that large datasets are needed for model training. This is a concern as reliable, consistent strength data is rather limited, especially for realistic industrial concretes. Here, based on the analysis of a large dataset (>10,000 observations) of measured compressive strengths from industrially-produced concretes, we compare the ability of select ML algorithms to "learn" how to reliably predict concrete strength as a function of the size of the dataset. Based on these results, we discuss the competition between how accurate a given model can eventually be (when trained on a large dataset) and how much data is actually required to train this model.

cs.LG↗

arXiv2017

Cooling-Rate Effects in Sodium Silicate Glasses: Bridging the Gap between Molecular Dynamics Simulations and Experiments

Although molecular dynamics (MD) simulations are commonly used to predict the structure and properties of glasses, they are intrinsically limited to short time scales, necessitating the use of fast cooling rates. It is therefore challenging to compare results from MD simulations to experimental results for glasses cooled on typical laboratory time scales. Based on MD simulations of a sodium silicate glass with varying cooling rate (from 0.01 to 100 K/ps), here we show that thermal history primarily affects the medium-range order structure, while the short-range order is largely unaffected over the range of cooling rates simulated. This results in a decoupling between the enthalpy and volume relaxation functions, where the enthalpy quickly plateaus as the cooling rate decreases, whereas density exhibits a slower relaxation. Finally, we demonstrate that the outcomes of MD simulations can be meaningfully compared to experimental values if properly extrapolated to slower cooling rates.

cond-mat.mtrl-sci↗

arXiv2015

Direct Experimental Evidence for Differing Reactivity Alterations of Minerals following Irradiation: The Case of Calcite and Quartz

Concrete, a mixture formed by mixing cement, water, and fine and coarse mineral aggregates is used in the construction of nuclear power plants (NPPs), e.g., to construct the reactor cavity concrete that encases the reactor pressure vessel, etc. In such environments, concrete may be exposed to radiation (e.g., neutrons) emanating from the reactor core. Until recently, concrete has been assumed relatively immune to radiation exposure. Direct evidence acquired on Ar$^+$-ion irradiated calcite and quartz indicates, on the contrary, that, such minerals, which constitute aggregates in concrete, may be significantly altered by irradiation. Specifically, while quartz undergoes disordering of its atomic structure resulting in a near complete lack of periodicity, i.e., similar to glassy silica, calcite only experiences random rotations, and distortions of its carbonate groups. As a result, irradiated quartz shows a reduction in density of around 15%, and an increase in chemical reactivity, described by its dissolution rate, similar to a glassy silica; i.e., an increase of around 3 orders of magnitude. Calcite however, shows little change in dissolution rates - although its density noted to reduce by around 9%. These differences are correlated with the nature of bonds in these minerals, i.e., being dominantly ionic or covalent, and the rigidity of the mineral's atomic network that is characterized by the number of topological constraints (n$_c$) that are imposed on the atoms in the network. The outcomes are discussed within the context of the durability of concrete structural elements formed with calcitic/quartzitic aggregates in nuclear power plants.

cond-mat.mtrl-sci↗

arXiv2015

Topological Control on Silicates Dissolution Kinetics

Like many others, silicate solids dissolve when placed in contact with water. In a given aqueous environment, the dissolution rate depends highly on the composition and the structure of the solid, and can span several orders of magnitude. Although the kinetics of dissolution depends on the complexities of both the dissolving solid and the solvent, a clear understanding of which critical structural descriptors of the solid control its dissolution rate is lacking. Through pioneering dissolution experiments and atomistic simulations, we correlate the dissolution rates - ranging over four orders of magnitude - of a selection of silicate glasses and crystals, to the number of chemical topological constraints acting between the atoms of the dissolving solid. The number of such constraints serves as an indicator of the effective activation energy, which arises from steric effects, and prevents the network from reorganizing locally to accommodate intermediate units forming over the course of the dissolution.

cond-mat.mtrl-sci↗

arXiv2015

Does a dissolution-precipitation mechanism explain concrete creep in moist environments?

Long-term creep (i.e., deformation under sustained load) is a significant material response that needs to be accounted for in concrete structural design. However, the nature and origin of creep remains poorly understood, and controversial. Here, we propose that concrete creep at RH (relative humidity) > 50%, but fixed moisture-contents (i.e., basic creep), arises from a dissolution-precipitation mechanism, active at nanoscale grain contacts, as is often observed in a geological context, e.g., when rocks are exposed to sustained loads, in moist environments. Based on micro-indentation and vertical scanning interferometry experiments, and molecular dynamics simulations carried out on calcium-silicate-hydrates (C-S-H's), the major binding phase in concrete, of different compositions, we show that creep rates are well correlated to dissolution rates - an observation which supports the dissolution-precipitation mechanism as the origin of concrete creep. C-S-H compositions featuring high resistance to dissolution, and hence creep are identified - analysis of which, using topological constraint theory, indicates that these compositions present limited relaxation modes on account of their optimally connected (i.e., constrained) atomic networks.

cond-mat.mtrl-sci↗

arXiv2015

Stretched Exponential Relaxation of Glasses at Low Temperature

The question of whether glass continues to relax at low temperature is of fundamental and practical interest. Here, we report a novel atomistic simulation method allowing us to directly access the long-term dynamics of glass relaxation at room temperature. We find that the potential energy relaxation follows a stretched exponential decay, with a stretching exponent $β= 3/5$, as predicted by Phillips' diffusion-trap model. Interestingly, volume relaxation is also found. However, it is not correlated to the energy relaxation, but is rather a manifestation of the mixed alkali effect.

cond-mat.dis-nn↗