Reliable predictive frameworks for thermal conductivity of ester biofuels using artificial intelligence approaches

Area of Science:

• Renewable Energy
• Chemical Engineering
• Computational Science

Background:

• Ester biofuels are key renewable energy sources.
• Accurate liquid thermal conductivity (LTC) data is crucial for optimizing biofuel energy systems.
• Existing models for LTC prediction lack generalizability across diverse ester biofuels and conditions.

Purpose of the Study:

• To develop robust machine learning models for predicting the LTC of ester biofuels.
• To address the limitations of existing empirical correlations.
• To provide accurate and generalizable tools for biofuel thermal property estimation.

Main Methods:

• Utilized a dataset of 1,641 experimental LTC measurements for 22 ester biofuels.
• Employed machine learning algorithms: Support Vector Machine (SVM), Decision Tree (DT), and Genetic Programming (GP).
• Predicted LTC based on temperature, pressure, critical thermodynamic properties, and molar weight.

Main Results:

• The SVM model achieved high accuracy (R²=99.53%, MAPE=0.60%) on unseen data.
• The GP model provided an explicit, interpretable correlation (R²=97.94%, MAPE=1.16%).
• Both models showed strong agreement with experimental data and identified temperature as the primary factor influencing LTC.

Conclusions:

• The developed intelligent models offer superior accuracy and broader applicability than existing empirical correlations.
• These models enhance predictive capabilities for biofuel thermal properties.
• Findings support the efficient integration of ester biofuels into energy systems.