← Back to VOLUME 13, ISSUE 10, OCTOBER 2026
This work is licensed under a Creative Commons Attribution 4.0 International License.
Artificial Intelligence and Machine Learning for Predicting Concrete Performance: A Critical Review of Modified Mixes, Durability, Structural Response and Sustainable Material Systems
Srushti Vijay Doke, Anand B. Kudoli*
👁 4 views📥 1 download
Abstract: The current research field of machine learning (ML) uses prediction models for concrete property assessment but these models show different levels of accuracy in their results. The review analyzes ML performance for concrete through four progressive levels of difficulty which include compressive-strength prediction and modified mixture prediction for fibres and polymers and waste materials and bacterial agents and freeze–thaw prediction and durability assessment and blast-structural response surrogate modeling. The review evaluates each task by assessing available data sets together with needed input characteristics and chosen algorithms and methods which assess model performance. The field of strength prediction stands as the most developed area because artificial neural networks and tree ensembles produce excellent results on benchmark datasets through their ability to perform interpolation on known mixtures. The available datasets do not include material descriptors which should contain information about fibre geometry and polymer chemistry and waste fineness and composition for creating sustainable modified mixtures. The prediction of durability faces obstacles because exposure data remains limited while showing different patterns and structural surrogate models draw their assumptions from the simulations which trained them. The research shows three main weaknesses which include limited dataset size and tests which split related specimens randomly and scientists who depend only on determination coefficient values. The study presents a ten-step process together with a reporting checklist which focuses on validation grouping and uncertainty reporting and verification through physical experiments. ML operates best as a screening system which supports decision-making processes through its ability to enhance engineering judgment and testing methods without replacing them.
Keywords: Machine learning; Artificial neural networks; Concrete strength prediction; Durability prediction; Explainable AI; Surrogate modelling; Sustainable concrete
Keywords: Machine learning; Artificial neural networks; Concrete strength prediction; Durability prediction; Explainable AI; Surrogate modelling; Sustainable concrete
How to Cite:
[1] Srushti Vijay Doke, Anand B. Kudoli*, “Artificial Intelligence and Machine Learning for Predicting Concrete Performance: A Critical Review of Modified Mixes, Durability, Structural Response and Sustainable Material Systems,” International Advanced Research Journal in Science, Engineering and Technology (IARJSET), DOI: 10.17148/IARJSET.2026.131043
