PSI - Issue 70

Abdul Musavvir et al. / Procedia Structural Integrity 70 (2025) 432–439

438

Fig. 5. Bar Chart Representation between Prediction Values & Experimental Results.

6. Conclusion The predictive model has been developed using ML to predict the 28 th day compressive strength of HSC with the inclusion of supplementary cementitious material from industrial wastes like Fly Ash, Ground Granulated Blast Furnace Slag, Silica Fumes and Nano Silica. For this model four different algorithms were used to compare the best suitable model for the collected dataset. The XGBoost regression model showed the highest R2 Score of 0.85 less Mean Squared Error of 23.28, Mean Absolute Error of 2.66 and Root Mean Squared Error of 4.82, thus makes the predicted value 95% accurate. The developed model was also validated by using real-time experimentation results. The model performs well when focused on 50-80 MPa and also for Fly Ash and Ground Granulated Blast Furnace Slag due to larger data availability in the specified range and materials. The ML model could be further developed, when more data (100-200) is trained in the weaker areas i.e., 70-100 MPa and combinational mixes with Silica Fumes and Nano Silica. By using this model, the optimization of HSC mixes has become easier, when the cement is replaced by the supplementary cementitious material mentioned by reducing the time and waste caused due to traditional trial and error method.

References

Abobakr Khalil Al-Shamiri, Joong Hoon Kim, Tian-Feng Yuan, Young Soo Yoon, 2019. Modeling the compressive strength of high-strength concrete: An extreme learning approach. Construction and Building Materials. Adebanjo, A. U., Shafiq, N., Razak, S. N. A., Kumar, V., Farhan, S. A., Singh, P., & Abu (Singh, 2024)bakar, A. S. 2024. Design and modeling the compressive strength of high-performance concrete with silica fume: a soft computing approach. Soft Computing, 28(7), 6059-6083. Alexey N. Beskopylny , Sergey A. Stel’makh , Evgenii M. Shcherban, Levon R. Mailyan, Besarion Meskhi, Irina Razveeva , Andrei Chernil’nik A (Adebanjo, 2024)nd Nikita Beskopylny 2022. Concrete Strength Prediction Using Machine Learning Methods Catboost, K-Nearest Neighbors, Support Vector Regression .Applied Sciences.