Optimization of Environmentally Friendly Asphalt Mix Formulation Using Artificial Intelligence Based on Genetic Algorithm

This study develops an intelligent system to optimize asphalt mix formulations with the goal of reducing carbon dioxide equivalent (CO₂e) emissions using a data-driven approach. A Random Forest regression model predicts emissions based on mix design parameters, while a Genetic Algorithm (GA) identifies the optimal combination. The dataset includes 50 asphalt mix records with variables such as binder content, mixing temperature, RAP proportion, aggregate type, and production method. The Random Forest model achieved a coefficient of determination (R²) of 0.772 and a Root Mean Square Error (RMSE) of 5.27 kg CO₂e/ton, indicating strong predictive capability, identifying mixing temperature, binder content, and RAP proportion as the most influential factors. GA optimization resulted in a mix with 4.65% binder, 125 degrees Celsius mixing temperature, and 48% RAP, producing 67.5 kg CO2e/ton—around 20% lower than conventional Hot Mix Asphalt. The findings demonstrate that combining artificial intelligence with optimization techniques provides an effective approach for sustainable asphalt design and supports low-carbon infrastructure planning.