Predicting the Compressive Strength of PVC-Confined Concrete via Machine Learning

Polyvinyl Chloride (PVC) is a promising sustainable alternative to traditional materials for confining concrete in structural applications due to its corrosion resistance, durability, and cost-effectiveness. The present research is focused on the axial compressive strength of PVC-confined concrete short columns with machine learning models for superior predictive accuracy. A database gathered from FEA simulations was utilized to train the Artificial Neural Network (ANN) and Support Vector Machine (SVM) models, in which the performance of each model was compared with an available empirical formula. The ANN and SVM models could achieve a high predictive accuracy with R² values close to 1.0 and smaller RMSE values than those by traditional empirical approaches. Results have shown that machine-learning models succeed in capturing complex interactions among the parameters, including PVC thickness, column diameter, and concrete compressive strength, providing a versatile and powerful method for strength prediction. These models offer construction engineers a rapid, cost-effective tool for predicting PVC-confined concrete column strengths without extensive physical testing, potentially accelerating the adoption of sustainable materials in structural design. By reducing experimental costs and design time, the approach demonstrates significant practical value for innovative construction technologies.