Performance and optimisation study of waste plastic aggregate based sustainable concrete – A machine learning approach

The use of waste plastic made aggregates in cement concrete not only can solve the problem of the disposal of waste plastics in a sustainable way but also it reduces the necessity of extracting natural aggregates, which causes erosion and environmental degradation. Therefore, feasibility of the usage of single used plastic (PET and PE) made aggregates in M20 and M25 grade cement concrete as partial replacement of natural coarse and fine aggregate were studied in this research work with 8 volumetric substitution levels (5 %, 10 %, 15 %, 20 %, 25 %, 30 %, 35 %, and 40 %) The influence of substitution on fresh and hardened properties of concrete was assessed by analyzing workability, wet and dry density, compressive strength at different ages and split tensile strength. The experimental results indicate that both the densities decrease about 10 % and workability increases about 60 % with increase in the ratio of plastic aggregate. Compressive strength and split tensile strength decrease with increase in plastic replacement upto 14 % and 34 % respectively. Based on the experimental results models were developed using four different Machine learning (ML) techniques namely ANN, FL, ANFIS and GA to predict the workability, density and compressive strength of the waste plastic aggregate based concrete. The result clearly signifies that the ANFIS model yields the best result (R2 = 0.98) for predicting the properties of concrete accurately. Furthermore, GA based multi-objective optimization models were developed using different aggregates for obtaining concrete with maximum workability and compressive strength along with minimum density and cost.