Novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approaches
The performance and durability of conventional concrete (CC) are significantly influenced by its weak tensile strength and tensile strain capacity (TSC). Thus, the intrusion of fibers in the cementitious matrix forms ductile engineered cementitious composites (ECCs) that can cater to this weak area...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
De Gruyter
2025-03-01
|
| Series: | Reviews on Advanced Materials Science |
| Subjects: | |
| Online Access: | https://doi.org/10.1515/rams-2025-0088 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850032376115101696 |
|---|---|
| author | Alahmari Turki S. Farooq Furqan |
| author_facet | Alahmari Turki S. Farooq Furqan |
| author_sort | Alahmari Turki S. |
| collection | DOAJ |
| description | The performance and durability of conventional concrete (CC) are significantly influenced by its weak tensile strength and tensile strain capacity (TSC). Thus, the intrusion of fibers in the cementitious matrix forms ductile engineered cementitious composites (ECCs) that can cater to this weak area of CC. Moreover, ECCs have become a reasonable substitute for brittle plain concrete due to their increased flexibility, ductility, and greater TSC. Thus, its prediction of ECC concrete is crucial without the need for laborious experimental procedures. Thus, to achieve this, machine learning approaches (MLAs), namely light gradient boosting (LGB) approach, extreme gradient boosting (XGB) approach, artificial neural network (ANN), and k-nearest neighbor (KNN), were developed. The data gathered from the literature comprise input parameters in which the fiber content, fiber length, cement, fiber diameter, water-to-binder ratio, fly ash (FA), age, sand, superplasticizer, and TSC as output parameters are utilized. The assessment of the models is gauged with coefficient of determination (R
2), statistical measures, and uncertainty analysis. In addition, an analysis of feature importance is carried out for further refinement of the model. The result demonstrates that ANN and XGB perform well for train and test sets with R
2 > 0.96. Statistical measures show that all models give fewer errors with higher R
2, in which XGB and ANN depict robust performance. Validation via K-fold confirms that models perform by showing fewer errors and a greater correlation of determination. In addition, the analysis of parameters reveals that the fiber diameter, cement, and FA have a major contribution in the prediction of TSC of ECC. Moreover, the graphical user interface is also developed to help users/researchers that will facilitate them to estimate the strength of ECC in practical applications. |
| format | Article |
| id | doaj-art-58a1eb54064e485dbe5fcc9d1463559f |
| institution | DOAJ |
| issn | 1605-8127 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | De Gruyter |
| record_format | Article |
| series | Reviews on Advanced Materials Science |
| spelling | doaj-art-58a1eb54064e485dbe5fcc9d1463559f2025-08-20T02:58:40ZengDe GruyterReviews on Advanced Materials Science1605-81272025-03-01641pp. 313810.1515/rams-2025-0088Novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approachesAlahmari Turki S.0Farooq Furqan1Department of Civil Engineering, Faculty of Engineering, University of Tabuk, P.O. Box 741, Tabuk, 71491, Saudi ArabiaNUST Institute of Civil Engineering (NICE), School of Civil and Environmental Engineering (SCEE), National University of Sciences and Technology (NUST), Sector H-12, Islamabad, 44000, PakistanThe performance and durability of conventional concrete (CC) are significantly influenced by its weak tensile strength and tensile strain capacity (TSC). Thus, the intrusion of fibers in the cementitious matrix forms ductile engineered cementitious composites (ECCs) that can cater to this weak area of CC. Moreover, ECCs have become a reasonable substitute for brittle plain concrete due to their increased flexibility, ductility, and greater TSC. Thus, its prediction of ECC concrete is crucial without the need for laborious experimental procedures. Thus, to achieve this, machine learning approaches (MLAs), namely light gradient boosting (LGB) approach, extreme gradient boosting (XGB) approach, artificial neural network (ANN), and k-nearest neighbor (KNN), were developed. The data gathered from the literature comprise input parameters in which the fiber content, fiber length, cement, fiber diameter, water-to-binder ratio, fly ash (FA), age, sand, superplasticizer, and TSC as output parameters are utilized. The assessment of the models is gauged with coefficient of determination (R 2), statistical measures, and uncertainty analysis. In addition, an analysis of feature importance is carried out for further refinement of the model. The result demonstrates that ANN and XGB perform well for train and test sets with R 2 > 0.96. Statistical measures show that all models give fewer errors with higher R 2, in which XGB and ANN depict robust performance. Validation via K-fold confirms that models perform by showing fewer errors and a greater correlation of determination. In addition, the analysis of parameters reveals that the fiber diameter, cement, and FA have a major contribution in the prediction of TSC of ECC. Moreover, the graphical user interface is also developed to help users/researchers that will facilitate them to estimate the strength of ECC in practical applications.https://doi.org/10.1515/rams-2025-0088machine learningengineered cementitious compositesfeature analysisgraphical user interfacestatistical analysisuncertainty analysis |
| spellingShingle | Alahmari Turki S. Farooq Furqan Novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approaches Reviews on Advanced Materials Science machine learning engineered cementitious composites feature analysis graphical user interface statistical analysis uncertainty analysis |
| title | Novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approaches |
| title_full | Novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approaches |
| title_fullStr | Novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approaches |
| title_full_unstemmed | Novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approaches |
| title_short | Novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approaches |
| title_sort | novel approaches in prediction of tensile strain capacity of engineered cementitious composites using interpretable approaches |
| topic | machine learning engineered cementitious composites feature analysis graphical user interface statistical analysis uncertainty analysis |
| url | https://doi.org/10.1515/rams-2025-0088 |
| work_keys_str_mv | AT alahmariturkis novelapproachesinpredictionoftensilestraincapacityofengineeredcementitiouscompositesusinginterpretableapproaches AT farooqfurqan novelapproachesinpredictionoftensilestraincapacityofengineeredcementitiouscompositesusinginterpretableapproaches |