Data driven tensile strength prediction for fiber-reinforced rubberized recycled aggregate concrete using machine learning

The structural integrity and long-term durability of concrete depend on its tensile strength, which endows the material with the capacity to resist crack initiation and propagation. The tensile strength of concrete is largely influenced by the mixing proportions, the type of aggregates, and the pres...

Full description

Saved in:
Bibliographic Details
Main Authors: Avijit Pal, Khondaker Sakil Ahmed, Nur Yazdani
Format: Article
Language:English
Published: Elsevier 2025-09-01
Series:Cleaner Materials
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2772397625000322
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849330665599795200
author Avijit Pal
Khondaker Sakil Ahmed
Nur Yazdani
author_facet Avijit Pal
Khondaker Sakil Ahmed
Nur Yazdani
author_sort Avijit Pal
collection DOAJ
description The structural integrity and long-term durability of concrete depend on its tensile strength, which endows the material with the capacity to resist crack initiation and propagation. The tensile strength of concrete is largely influenced by the mixing proportions, the type of aggregates, and the presence of fibers or additives. The incorporation of different ingredients and mixing proportions makes this property nearly unpredictable. To tackle this, this research examined the tensile strength behavior of fiber-reinforced rubberized recycled aggregate concrete (FR3C) using nine machine learning (ML) models. In this study, nine machine learning models—Random Forest, K-Nearest Neighbors, Support Vector Regression, Decision Tree, Artificial Neural Network, AdaBoost, Gradient Boost, CatBoost, and Extreme Gradient Boost—were trained and tested using a dataset of 346 samples representing various mix proportions. The models were applied to predict the tensile strengths of the concrete and to determine the optimal proportions of ingredients. Key input characteristics include water-to-cement ratio (W/C), nominal aggregate size, rubber content, amount of recycled coarse aggregate (RCA), type of fiber and usage, plasticizer use, fly ash (%), and compressive strength. The findings showed that K-Nearest Neighbors performed best in predicting FR3C tensile strength, achieving the lowest mean absolute error MAE (0.001) and root mean squared error (RMSE 0.001) and highest coefficient of determination (R2 = 0.999) in test scores. The Shapley Additive Explanations (SHAP) analysis indicated that compressive strength, W/C ratio, and fiber (%) are the most influential parameters affecting the tensile strength of FR3C. Moreover, increased W/C ratios and higher plasticizer content were associated with a 60–72 % reduction in tensile strength. This research may contribute to practical concrete mix design in the construction industry and also in the design process of structural elements particularly for crack width control and mitigation. Therefore, it is feasible to increase the usage of FR3C concrete by precisely forecasting its tensile strength, transforming wastes into resources, and minimizing the adverse environmental effects of construction materials.
format Article
id doaj-art-7b4bab4c4b274223a7eca4be4701068a
institution Kabale University
issn 2772-3976
language English
publishDate 2025-09-01
publisher Elsevier
record_format Article
series Cleaner Materials
spelling doaj-art-7b4bab4c4b274223a7eca4be4701068a2025-08-20T03:46:50ZengElsevierCleaner Materials2772-39762025-09-011710032310.1016/j.clema.2025.100323Data driven tensile strength prediction for fiber-reinforced rubberized recycled aggregate concrete using machine learningAvijit Pal0Khondaker Sakil Ahmed1Nur Yazdani2Department of Civil Engineering, The University of Texas at Arlington, USADepartment of Civil Engineering, Military Institute of Science and Technology (MIST), Dhaka, Bangladesh; Corresponding author.Department of Civil Engineering, The University of Texas at Arlington, USAThe structural integrity and long-term durability of concrete depend on its tensile strength, which endows the material with the capacity to resist crack initiation and propagation. The tensile strength of concrete is largely influenced by the mixing proportions, the type of aggregates, and the presence of fibers or additives. The incorporation of different ingredients and mixing proportions makes this property nearly unpredictable. To tackle this, this research examined the tensile strength behavior of fiber-reinforced rubberized recycled aggregate concrete (FR3C) using nine machine learning (ML) models. In this study, nine machine learning models—Random Forest, K-Nearest Neighbors, Support Vector Regression, Decision Tree, Artificial Neural Network, AdaBoost, Gradient Boost, CatBoost, and Extreme Gradient Boost—were trained and tested using a dataset of 346 samples representing various mix proportions. The models were applied to predict the tensile strengths of the concrete and to determine the optimal proportions of ingredients. Key input characteristics include water-to-cement ratio (W/C), nominal aggregate size, rubber content, amount of recycled coarse aggregate (RCA), type of fiber and usage, plasticizer use, fly ash (%), and compressive strength. The findings showed that K-Nearest Neighbors performed best in predicting FR3C tensile strength, achieving the lowest mean absolute error MAE (0.001) and root mean squared error (RMSE 0.001) and highest coefficient of determination (R2 = 0.999) in test scores. The Shapley Additive Explanations (SHAP) analysis indicated that compressive strength, W/C ratio, and fiber (%) are the most influential parameters affecting the tensile strength of FR3C. Moreover, increased W/C ratios and higher plasticizer content were associated with a 60–72 % reduction in tensile strength. This research may contribute to practical concrete mix design in the construction industry and also in the design process of structural elements particularly for crack width control and mitigation. Therefore, it is feasible to increase the usage of FR3C concrete by precisely forecasting its tensile strength, transforming wastes into resources, and minimizing the adverse environmental effects of construction materials.http://www.sciencedirect.com/science/article/pii/S2772397625000322FiberCrumb rubberRecycle concrete aggregateTensile strengthMachine Learning
spellingShingle Avijit Pal
Khondaker Sakil Ahmed
Nur Yazdani
Data driven tensile strength prediction for fiber-reinforced rubberized recycled aggregate concrete using machine learning
Cleaner Materials
Fiber
Crumb rubber
Recycle concrete aggregate
Tensile strength
Machine Learning
title Data driven tensile strength prediction for fiber-reinforced rubberized recycled aggregate concrete using machine learning
title_full Data driven tensile strength prediction for fiber-reinforced rubberized recycled aggregate concrete using machine learning
title_fullStr Data driven tensile strength prediction for fiber-reinforced rubberized recycled aggregate concrete using machine learning
title_full_unstemmed Data driven tensile strength prediction for fiber-reinforced rubberized recycled aggregate concrete using machine learning
title_short Data driven tensile strength prediction for fiber-reinforced rubberized recycled aggregate concrete using machine learning
title_sort data driven tensile strength prediction for fiber reinforced rubberized recycled aggregate concrete using machine learning
topic Fiber
Crumb rubber
Recycle concrete aggregate
Tensile strength
Machine Learning
url http://www.sciencedirect.com/science/article/pii/S2772397625000322
work_keys_str_mv AT avijitpal datadriventensilestrengthpredictionforfiberreinforcedrubberizedrecycledaggregateconcreteusingmachinelearning
AT khondakersakilahmed datadriventensilestrengthpredictionforfiberreinforcedrubberizedrecycledaggregateconcreteusingmachinelearning
AT nuryazdani datadriventensilestrengthpredictionforfiberreinforcedrubberizedrecycledaggregateconcreteusingmachinelearning