Date of Original Version
Civil and Environmental Engineering
With the advent of 3D printing, auxetic cellular cementitious composites (ACCCs) have recently garnered significant attention owing to their unique mechanical performance. To enable seamless performance prediction of the ACCCs, interpretable machine learning (ML)-based approaches can provide efficient means. However, the prediction of Poisson’s ratio using such ML approaches requires large and consistent datasets which is not readily available for ACCCs. To address this challenge, this paper synergistically integrates a finite element analysis (FEA)-based framework with ML to predict the Poisson’s ratios. In particular, the FEA-based approach is used to generate a dataset containing 850 combinations of different mesoscale architectural void features. The dataset is leveraged to develop an ML-based prediction tool using a feed-forward multilayer perceptron-based neural network (NN) approach which shows excellent prediction efficacy. To shed light on the relative influence of the design parameters on the auxetic behavior of the ACCCs, Shapley additive explanations (SHAP) is employed, which establishes the volume fraction of voids as the most influential parameter in inducing auxetic behavior. Overall, this paper develops an efficient approach to evaluate geometry-dependent auxetic behaviors for cementitious materials which can be used as a starting point toward the design and development of auxetic behavior in cementitious composites.
Publication Title, e.g., Journal
Materials & Design
Lyngdoh G., Zaki M., Krishnan N.M.A., Das S., (2022) “Prediction of Concrete Strengths Enabled by Missing Data Imputation and Interpretable Machine Learning” Cement and Concrete Composites, Volume 128, April 2022, 104414.
Available at: https://doi.org/10.1016/j.matdes.2021.110341
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.