Predictive modeling of compressive strength in glass powder blended pervious concrete

Pervious concrete, known for its high porosity, is crucial in sustainable construction and effective stormwater management. This study explores the predicting compressive strength of glass powder blended pervious concrete. A comprehensive methodology was employed, utilizing machine learning algorithms - specifically Support Vector Regression (SVR), Artificial Neural Networks (ANN), K-Nearest Neighbors (KNN), and Extreme Gradient Boosting (XGB) - to predict the compressive strength. The models were trained on a diverse dataset that included parameters such as water-to-cement ratio, binder content, and curing conditions. Results indicated that the SVR model outperformed others, achieving high predictive accuracy with minimal error margins. Additionally, sensitivity analysis underscored the significant impact of the curing period and admixture content on compressive strength. This research underscores the potential of using crushed glass powder in pervious concrete, promoting both enhanced performance and sustainability in modern construction practices.