Ensemble boosting-based model for predicting compressive strength of recycled aggregate concrete
Mosbeh R. Kaloop,Mohamed Rezaik,Ash Ahmed,Jong Wan Hu,Mohamed Eldessouki,Emad Elbeltagi
Abstract
Using recycled aggregate concrete (RAC) has recently been growing rapidly as an alternative to conventional concrete for sustainable development in construction. Nevertheless, there are limitation in computational guidance for compressive strength (CS) of RAC. Thus, this study aims to develop ensemble data-driven models for estimating the CS of RAC. Five ensemble models, namely category-boosting (CatBoost), gradient-boosting (GBoost), extreme gradient-boosting (XGBoost), K-nearest neighbor boosting (KNN), and random forest (RF), were developed, examined and compared for estimating the 28-day CS. A total of 578 datasets of different RAC mixtures were used to develop and test the proposed models. SHapley Additive exPlanations (SHAP) was used to analyze the impact of the used input variables on CS. The multivariate adaptive regression splines (MARS) algorithm was used to formulate the relationship between significant input variables and CS. The results show that CatBoost model outperformed the other proposed models with correlation coefficient (r) and mean absolute error (MAE) scores of 0.92 and 4.36 MPa, respectively. SHAP results of the CatBoost model show the impact of water/cement ratio is highly significant in modelling CS of RAC followed by nominal maximum recycled concrete aggregate (RCA) size, RCA replacement ratio, and bulk density of natural aggregate. Although parent concrete strength and the Los Angeles abrasion index of RCA have limited influence on the CS of RAC, they can be used to enhance the CS when both are greater than 40 MPa and 35 MPa, respectively. The proposed MARS equations can be used as a guideline in the design stage of RAC mixtures.
Mosbeh R. Kaloop — 1) Department of Civil and Environmental Engineering, Incheon National University, Korea, 2) Incheon Disaster Prevention Research Center, Incheon National University, Korea, 3) Public Works and Civil Engineering Department, Mansoura University, Egypt, 4) Digital InnoCent Ltd., London, UK
Mohamed Rezaik — Appout ITs, Tanta, Egypt
Ash Ahmed — Civil Engineering Department, Leeds Beckett University, Leeds, UK
Jong Wan Hu — 1) Department of Civil and Environmental Engineering, Incheon National University, Korea, 2) Incheon Disaster Prevention Research Center, Incheon National University, Korea
Mohamed Eldessouki — Digital InnoCent Ltd., London, UK
Emad Elbeltagi — Department of Civil Engineering, College of Engineering, Qassim University, Buraydah 51452, Saudi Arabia
PDF Viewer
Preview is limited to the first 3 pages. Sign in to access the full PDF.
