Machine Learning Models for Mechanical and Micro Structural Properties of Recycled Fine Aggregate Concrete Using Different Mixing Approaches

The construction industry is primarily responsible for the depletion of natural resources and the disruption of environmental equilibrium due to unregulated mining activities. In this particular context, the utilization of recycled fine aggregate (RFA) derived from construction and demolition (C&D) waste presents itself as a viable solution. The conventional method of mix proportioning for RFA in concrete is not applicable in this case. The main innovation of our research lies in the fulfilment of one of the principles of circular economy, namely the reduction of carbon emissions, through the recycling of locally collected concrete waste. To tackle this issue, a novel triple mix-proportioning approach has been developed using the concepts of maximum packing density and minimum paste theory. The fresh and hardened properties were evaluated and microstructural characterization was carried out for the newly formulated mixes incorporating RFA with optimized combined aggregates. The compressive strength of concrete with recycled fine aggregate increases by ۵.۰۴% for ۲۵% and, ۲۱.۶۹% for ۵۰% replacement, and decreases by ۳۵.۴۴% for ۱۰۰% replacement as compared to controlled concrete at the age of ۲۸ days using the triple mixing approach. The findings indicate that replacing approximately ۵۰% of sand with RFA is the optimal amount, as further replacement leads to a decrease in compressive strength, particularly at ۱۰۰% replacement due to the presence of adhered mortar in RFA. In this study, the performance evaluation of RFA concrete has been conducted by comparing six established ML regression models and sensitivity analysis was performed to assess the variable's performance.

کلیدواژه ها:

Recycled aggregate concrete ، Recycled Fine Aggregate ، particle packing ، Compressive strength ، Micro structure ، Machine Learning