Skip to main content
Elsevier BV

Predicting the Mechanical Response Profile of Porous Materials Via Microstructure-Informed Neural Networks

Preprints
Lindqwister, W; Peloquin, J; Dalton, L; Gall, K; Veveakis, M
2023

Porous media represent a wide range of materials that have yet to be fully understood and present a challenge in being harnessed to their full potential. From rocks to energy storage materials and bones, porous materials are ubiquitous across engineering disciplines. However, due to the non-uniformity and near infinite variations of their structure it is inherently challenging to accurately and quickly assess meso-scale properties such as mechanical strength. The non-uniformity of structure, created by the inherent heterogeneity and stochasticity of a porous material’s microstructure, creates an added complexity to determining an exact relationship between the structure and properties of a porous material. Although destructive testing and finite element analysis (FEA) are proven methods for calculating properties of a porous material, the former is costly and impractical at larger scales, and the latter is computationally intensive. Machine learning has proven itself over the past few decades as a method of expediting the discovery of correlations within data and creating powerful predictive models. To expedite the discovery of structure-property relationships for porous materials and overcome the challenges presented when training a machine learning model, we have developed a systematic methodology for defining a porous material by scalar geometric descriptors, performing physical and computational experiments to obtain structure performance data, and ultimately training a machine learning model to predict effective compressive strength directly from geometry. Through this work we aim to accelerate the analysis of porous material structure-property relationships via a machine learning model trained on microstructural features.

Duke Scholars

Altmetric Attention Stats
Dimensions Citation Stats

DOI

Publication Date

2023
 

Citation

APA
Chicago
ICMJE
MLA
NLM
Lindqwister, W., Peloquin, J., Dalton, L., Gall, K., & Veveakis, M. (2023). Predicting the Mechanical Response Profile of Porous Materials Via Microstructure-Informed Neural Networks. Elsevier BV. https://doi.org/10.2139/ssrn.4605136
Lindqwister, Winston, Jacob Peloquin, Laura Dalton, Ken Gall, and Manolis Veveakis. “Predicting the Mechanical Response Profile of Porous Materials Via Microstructure-Informed Neural Networks.” Elsevier BV, 2023. https://doi.org/10.2139/ssrn.4605136.
Lindqwister W, Peloquin J, Dalton L, Gall K, Veveakis M. Predicting the Mechanical Response Profile of Porous Materials Via Microstructure-Informed Neural Networks. Elsevier BV. 2023.
Lindqwister, Winston, et al. “Predicting the Mechanical Response Profile of Porous Materials Via Microstructure-Informed Neural Networks.” Elsevier BV, 2023. Crossref, doi:10.2139/ssrn.4605136.
Lindqwister W, Peloquin J, Dalton L, Gall K, Veveakis M. Predicting the Mechanical Response Profile of Porous Materials Via Microstructure-Informed Neural Networks. Elsevier BV. 2023.

DOI

Publication Date

2023