Microstructural behavior and explainable machine learning aided mechanical strength prediction and optimization of recycled glass-based solid waste concrete

One effective way to accomplish sustainability goals is to use solid waste as sustainable building materials to help reduce environmental footprint and enhance cementitious compounds. Compressive strength (CS), one of the most crucial characteristics of concrete, must be measured through expensive and time-consuming studies. Furthermore, numerous research has shown the effect of using solid waste (waste glass) as both cement and aggregate replacement on the mechanical properties of concrete. Therefore, this study determines the CS for solid waste concrete (SWC) using an extensive database of 451 data points gathered from experimental and earlier research using waste glass (used as solid waste) as aggregate and cement replacement. For experimental evaluation, SWC was prepared by replacing cement with 10 % SF and 5 %, 10 %, 15 %, and 25 % solid waste (SW). A compressive strength test was conducted to assess the mechanical strength, and microstructural analysis was performed to highlight the interplay between SW and the concrete matrix. The CS was predicted using Random Forest (RF), XGBoost (XGB), Gradient Boosting (GB), and AdaBoost (ADB) models. The ML model’s performance was validated using a variety of performance metrics, including R 2 , RMSE, MAE, and MAPE. To further illustrate how each feature affected the CS, SHAP plots and partial dependence plots were created. According to the results, the XGB model outperformed the others, with R 2 values between 0.948 and 0.994. SHAP analysis revealed that curing age, SCM, cement, and SW were the most impactful features for predicting strength. Microstructure analysis revealed that the incorporation of SW showed denser microstructure and greater C-S-H bonds, which enhanced strength. These results are useful for the sustainable manufacturing of waste-based concrete and help designers comprehend the importance of every element in SW.