A Random Forest-Based Model for Optimizing PAO Concentration in Biodiesel-Petroleum Diesel Blends

Biodiesel, while offering significant environmental and renewable energy benefits, faces challenges in cold environments due to the crystallization of fatty acid methyl esters, which impedes its flow properties and limits its utilization. To address this, blending biodiesel with petroleum diesel or adding pour point depressants (PPDs), such as Poly-alpha-olefin (PAO), has been used to improve its cold flow performance. However, the effectiveness of these methods varies, and previous studies have shown inconsistent results in determining the optimal concentration of PAO. In this paper, we enhanced the dataset using spline interpolation and developed a predictive model using a random forest model to optimize PAO concentration for biodiesel-petroleum diesel blends. The results of this approach provide a more reliable prediction of the ideal PAO concentration for improving the cold flow properties of biodiesel, thereby enhancing its practical application in low-temperature environments.