Machine learning based prediction of mechanical strength in self-compacting concrete incorporating electronic waste aggregates
摘要
This study focuses on the impact of using electronic waste (E-waste) as partial replacement for gravels in the Self-Compacting Concrete (SCC) with the mixing of 10% Alccofinite, and 20% fly ash as supplementary cementitious materials. To assess performance in both fresh and hardened states, the study examined thirteen distinct SCC formulations, incorporating E-waste in 2.5% increments ranging from 0% up to 30%. The experimental data show that the values of the slump flow vary between 652.5 mm and 809.2 mm, the values of V-funnel times range from 6.3 s to 10.3 s, and the value of L-box ratio is between 1.023 and 0.787. Maximum compressive strength, split tensile strength, flexural strength, modulus of elasticity, density, and bond strength of the composite are 38.9 MPa, 3.6 MPa, 5.4 MPa, 29,833.3 MPa, 2406.7 kg/m3, and 11.33 MPa, respectively. The results show that the mechanical performance reaches its optimum with a 15% E-waste replacement; however, higher contents of E-waste reduce the workability and strength. Mechanical strength prediction using four machine learning algorithms (Linear Regression, Support Vector Regression, Random Forest, and XGBoost) was carried out by using fresh concrete properties and mix variables. It is interesting to note that the XGBoost model provided the best prediction, as it delivered an R2 value of more than 0.93 and an RMS prediction error of less than 0.8 MPa. Six 1.5 m long beams of 100 mm × 200 mm dimensions is subjected to two-point loading to determine strength, deflection and failure modes. The results show the 15% E-waste mix reached a maximum flexural strength of 40.50 MPa, which is a + 12% increase in load capacity over the 10% mix. The study thus confirms a moderate addition of E-waste (15%) improves the sustainability of SCC while maintaining good structural integrity.