mix design

recycled concrete

predicting compressive

Construction and Building Materials

• Machine learning techniques predict compressive strength of concrete made with RAC. • GB_PSO is the best Machine Learning model for prediction of compressive strength. • Mix design of concrete made with RAC can be estimated with aided of PDP 1D. To reduce the environmental impact of construction and demolition waste of concrete, recycled concrete aggregate (RCA) has been widely utilized in concrete. The compressive strength of recycled concrete is one of the most important parameters governing the quality of concrete. The compressive strength is determined from the compression test, which requires a huge amount of materials as well as consumes cost and time. Thus, to solve those limitations, this study focused on evaluating the compressive strength of concrete made from RCA using different single and hybrid models of machine learning. Six machine learning models including Gradient Boosting (GB), Extreme Gradient Boosting (XGB), Support Vector Regression (SVR), and three hybrid models of those single models with Particle Swarm Optimization (PSO) namely GB_PSO, XGB_PSO, and SVR_PSO were used to estimate the compressive strength of recycled concrete. The input variables for modeling consisted of cement content, water content, aggregate content, natural aggregate content, recycle concrete aggregate content, sand content, water absorption rates of natural aggregate and RCA. The results of this study show that hybrid models performed better than single models in terms of prediction accuracy. The results indicated that the GB_PSO has the highest prediction accuracy with R = 0.9356, RMSE = 5.5604 MPa, and MAE = 4.2882 MPa. The results of feature importance analysis and partial dependence plots (PDP) analysis revealed that the most important variable effect on compressive strength of concrete made with RAC is cement content, whatever performance strategies of concrete made with RAC. From the results of PDP, the quantity of each material can be computed easily for the designed compressive strength. In the end, this study...

Evaluating compressive strength of concrete made with recycled concrete aggregates using machine learning approach