Predicting flexural strength of sustainable basalt fiber-reinforced concrete using machine learning algorithms incorporating a graphical user interface
Year: 2026
Type: article
Abstract: The development of basalt fiber-reinforced polymers (BFRP) is attributed to the revolutionary nature of basalt fiber (BF) technology. Composite material made from basalt rock, which is widely available in nature, BFRP produces around 74% fewer carbon emissions than traditional steel, aligning with international sustainability targets. Notably, the adoption of sustainable basalt fiber-reinforced concrete (BFRC) aligns with the Kingdom of Saudi Arabia’s (KSA) Vision 2030, which emphasizes sustainable infrastructure development, reduction of carbon emissions, and integration of advanced technologies such as AI in engineering practices. This study uses BF to improve the mechanical properties of BFRC. A total of 245 data points were used to predict the flexural strength (FS) of the BFRC, which were collected from several previous articles. These data points were randomly divided into two parts: a training phase (70%) and a testing phase (30%). In the present study, the prediction of FS of BFRC employed five machine learning (ML) models, such as the Decision Tree algorithm (DT), K nearest neighbors (KNN), Adaptive Boosting Regression (ABR), random forest regression (RFR), and Support Vector Regression (SVR) with grid search hyper-tuning. The results found that the RFR and SVR models’ performance is the most accurate prediction of FS compared to other models, achieving an R 2 of 0.980 for the training stage and R 2 of 0.919 for the testing stage. The root mean squared error (RMSE) values for the SVR and RFR models are 0.5853 and 0.5743, respectively, at the testing stages of the predicted BFRC concrete. Additionally, the model generated Shapley Additive Explanations (SHAP) interaction plots to display the influence of each input variable on an individual prediction. However, SHAP revealed that the cement and silica fume had the strongest positive influence on the BFRC of FS. Finally, the Graphical User Interface was developed to allow designers to efficiently and economically predict the FS of BFRC, providing a convenient alternative to expensive computational simulations and labor-intensive experimental procedures.
Authors: Hafiz Farooq Ahmad, MD Arifuzzaman, Sourov Paul, Pobithra Das
Institutions: King Faisal University, Leading University
Language: English
Subfield: Civil and Structural Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $1,600
Funder: King Faisal University
OpenAlex ID: W7147325871