Abstract
Ester biofuels have emerged as promising renewable alternatives to fossil fuels due to their environmental compatibility and favorable combustion characteristics. Accurate knowledge of their liquid thermal conductivity (LTC) is essential for optimizing energy systems, engine performance, and thermal modeling applications. However, existing literature lacks generalizable models capable of estimating LTC across diverse ester biofuels and operating conditions. This study addresses this gap by developing robust machine learning models using a comprehensive dataset comprising 1,641 experimental LTC measurements for 22 different ester biofuels under varied pressures and temperatures. Three advanced computational approaches, including Support Vector Machine (SVM), Decision Tree (DT), and Genetic Programming (GP), were employed to predict LTC based on temperature, pressure, critical thermodynamic properties, and molar weight of the biofuels. Among the developed models, the SVM technique exhibited superior predictive performance with a determination coefficient (R(2)) of 99.53%, and a mean absolute percentage error (MAPE) of 0.60% for the unseen dataset. The GP model, in turn, produced an explicit mathematical correlation with MAPE of 1.16% and R² of 97.94%, offering a reliable and interpretable alternative. The models demonstrated strong agreement with experimental data across a wide operating range and successfully captured the influence of key parameters on LTC. Sensitivity analysis revealed that temperature plays the most fundamental role in controlling LTC. Compared to existing empirical correlations, the proposed intelligent models provide broader applicability and enhanced accuracy for LTC prediction of ester biofuels. These findings contribute to advancing predictive capabilities in biofuel thermal property modeling and support the efficient integration of ester biofuels into energy systems.