Compressive strength prediction models for concrete containing nano materials and exposed to elevated temperatures
The addition of nanomaterials to concrete is widely employed in modern construction to improve its durability and mechanical properties. In the present study, two machine learning algorithms, random forest (RF) and M5P decision tree, and linear regression were used for developing prediction models f...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025000635 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841545949633052672 |
|---|---|
| author | Hany A. Dahish Ahmed D. Almutairi |
| author_facet | Hany A. Dahish Ahmed D. Almutairi |
| author_sort | Hany A. Dahish |
| collection | DOAJ |
| description | The addition of nanomaterials to concrete is widely employed in modern construction to improve its durability and mechanical properties. In the present study, two machine learning algorithms, random forest (RF) and M5P decision tree, and linear regression were used for developing prediction models for the compressive strength (CS) of concrete containing nano alumina (NA) and carbon nanotubes (CNT) and being subjected to elevated temperatures. Datasets of 169 tested concrete specimens of 100 × 100 × 100 mm were gathered from the literature. Four variables were considered in the development of the prediction models, including temperature, exposure duration, and NA and CNT ratios. K-fold cross-validation was used to confirm the predicted output. The performance of the created models was compared to experimental data and earlier developed models: fuzzy logic models, artificial neural networks, genetic algorithms, and water cycle algorithms, using several evaluation metrics. The response surface methodology (RSM) was employed for optimizing the input parameters in order to achieve the maximum compressive strength of concrete containing NA and CNT exposed to elevated temperatures. The results demonstrated that the RF model for predicting the compressive strength of concrete outperformed the other developed models in terms of performance and accuracy. The RF model has the highest R2 value of 0.9921 and the lowest error values: RMSE of 0.7721 MPa, MAE of 0.5757 MPa, and MAPE of 1.72% compared to other developed models in this study. The SHAP and sensitivity analyses indicated that the temperature has the greatest impact on predicting the CS. The optimal compressive strengths of concrete were achieved by replacing cement with NA and CNT at replacement levels of (2.14, 0.08), (1.4, 0.11), (1.5, 0.08), (1.0, 0.1), (1.2, 0.1), and (2.1, 0.1) for concrete exposed to temperatures of 200 °C, 400 °C, 500 °C, 600 °C, 700 °C, and 800 °C, respectively. In comparison with the previously developed models in literature, the RF model demonstrated superiority for predicting the CS of concrete containing NA and CNT and exposed to elevated temperatures. |
| format | Article |
| id | doaj-art-f586e8b4854b4823ba2acd594416c7a9 |
| institution | Kabale University |
| issn | 2590-1230 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Results in Engineering |
| spelling | doaj-art-f586e8b4854b4823ba2acd594416c7a92025-01-11T06:41:54ZengElsevierResults in Engineering2590-12302025-03-0125103975Compressive strength prediction models for concrete containing nano materials and exposed to elevated temperaturesHany A. Dahish0Ahmed D. Almutairi1Department of Civil Engineering, College of Engineering, Qassim University, Buraydah 51452, Saudi ArabiaCorresponding author.; Department of Civil Engineering, College of Engineering, Qassim University, Buraydah 51452, Saudi ArabiaThe addition of nanomaterials to concrete is widely employed in modern construction to improve its durability and mechanical properties. In the present study, two machine learning algorithms, random forest (RF) and M5P decision tree, and linear regression were used for developing prediction models for the compressive strength (CS) of concrete containing nano alumina (NA) and carbon nanotubes (CNT) and being subjected to elevated temperatures. Datasets of 169 tested concrete specimens of 100 × 100 × 100 mm were gathered from the literature. Four variables were considered in the development of the prediction models, including temperature, exposure duration, and NA and CNT ratios. K-fold cross-validation was used to confirm the predicted output. The performance of the created models was compared to experimental data and earlier developed models: fuzzy logic models, artificial neural networks, genetic algorithms, and water cycle algorithms, using several evaluation metrics. The response surface methodology (RSM) was employed for optimizing the input parameters in order to achieve the maximum compressive strength of concrete containing NA and CNT exposed to elevated temperatures. The results demonstrated that the RF model for predicting the compressive strength of concrete outperformed the other developed models in terms of performance and accuracy. The RF model has the highest R2 value of 0.9921 and the lowest error values: RMSE of 0.7721 MPa, MAE of 0.5757 MPa, and MAPE of 1.72% compared to other developed models in this study. The SHAP and sensitivity analyses indicated that the temperature has the greatest impact on predicting the CS. The optimal compressive strengths of concrete were achieved by replacing cement with NA and CNT at replacement levels of (2.14, 0.08), (1.4, 0.11), (1.5, 0.08), (1.0, 0.1), (1.2, 0.1), and (2.1, 0.1) for concrete exposed to temperatures of 200 °C, 400 °C, 500 °C, 600 °C, 700 °C, and 800 °C, respectively. In comparison with the previously developed models in literature, the RF model demonstrated superiority for predicting the CS of concrete containing NA and CNT and exposed to elevated temperatures.http://www.sciencedirect.com/science/article/pii/S2590123025000635Machine learning algorithmsNanoparticleCompressive strengthM5PRandom forestResponse surface methodology |
| spellingShingle | Hany A. Dahish Ahmed D. Almutairi Compressive strength prediction models for concrete containing nano materials and exposed to elevated temperatures Results in Engineering Machine learning algorithms Nanoparticle Compressive strength M5P Random forest Response surface methodology |
| title | Compressive strength prediction models for concrete containing nano materials and exposed to elevated temperatures |
| title_full | Compressive strength prediction models for concrete containing nano materials and exposed to elevated temperatures |
| title_fullStr | Compressive strength prediction models for concrete containing nano materials and exposed to elevated temperatures |
| title_full_unstemmed | Compressive strength prediction models for concrete containing nano materials and exposed to elevated temperatures |
| title_short | Compressive strength prediction models for concrete containing nano materials and exposed to elevated temperatures |
| title_sort | compressive strength prediction models for concrete containing nano materials and exposed to elevated temperatures |
| topic | Machine learning algorithms Nanoparticle Compressive strength M5P Random forest Response surface methodology |
| url | http://www.sciencedirect.com/science/article/pii/S2590123025000635 |
| work_keys_str_mv | AT hanyadahish compressivestrengthpredictionmodelsforconcretecontainingnanomaterialsandexposedtoelevatedtemperatures AT ahmeddalmutairi compressivestrengthpredictionmodelsforconcretecontainingnanomaterialsandexposedtoelevatedtemperatures |