Session: 09-14-01: Fundamentals and Applications of Thermodynamics

Paper Number: 166904

Estimation of the Heating Value of Fuels by a Regression Analysis Model 

The Lower Heating Value (LHV) is a fundamental property in fuel characterization, playing a critical role in energy efficiency assessments, optimization of combustion systems, and fuel performance evaluation. Accurately determining the LHV is essential for various applications, including power generation, waste-to-energy processes, and biomass utilization, amongst others. Traditionally, LHV is obtained through experimental methods such as bomb calorimetry, which, while precise, can be time-consuming, costly, and impractical for large-scale or continuous assessments. To overcome these limitations, this study aims to develop a predictive model for LHV estimation based on elemental composition, providing a more efficient and cost-effective alternative to experimental determination.

A comprehensive database of 520 fuel samples was compiled, ensuring a diverse representation of different fuel types, including 38% solids, 31% gases, and 31% liquids. Each sample in the database was characterized by its elemental composition, including carbon (C), hydrogen (H), oxygen (O), nitrogen (N), sulfur (S), and ash content, alongside its corresponding LHV. To establish the relationship between elemental composition and LHV, a correlation analysis was performed, revealing that carbon, hydrogen, and oxygen had the most significant influence on LHV, while nitrogen, sulfur, and ash exhibited lower correlation levels. Based on these findings, a second-degree polynomial equation was formulated using C, H, and O as input variables.

To optimize the polynomial coefficients and enhance the predictive performance of the model, a Python-based linear regression algorithm was developed. The model was trained using the compiled dataset, with the objective of minimizing the error between the predicted and experimentally measured LHV values. The final equation demonstrated high accuracy, achieving a coefficient of determination (R²) of 0.9871 and a mean squared error (MSE) of 2.2. Additionally, the model exhibited a maximum deviation of 9.1 MJ/kg between predicted and actual LHV values, indicating strong predictive reliability.

Furthermore, the proposed equation was compared with existing empirical correlations from the literature, demonstrating superior performance in terms of predictive accuracy and error minimization. This improvement underscores the advantage of using a tailored polynomial approach based on elemental analysis rather than relying on generalized empirical formulas.

These results confirm the reliability of the developed model as an efficient and accurate alternative for LHV estimation. The equation provides a practical tool for researchers and engineers working in fuel analysis, energy system modeling, and combustion engineering applications, reducing the need for extensive laboratory testing while maintaining high precision. This model can facilitate advancements in fuel characterization, sustainability assessments, and energy efficiency optimization.

Presenting Author: Jose Teixeira Universidade do Minho

Presenting Author Biography: Jose Teixeira is professor of Fluid Mechanics and Heat Transfer at the Universiy of Minho, Portugal

Authors:

Carlos Castro Unversidade do Minho
Didier Sanchez FCT-NOVA University of Lisbon
Rui Ribeiro 3HyLab – Green Hydrogen Collaborative Laboratory
Jose Teixeira Universidade do Minho
Margarida Gonçalves FCT-NOVA University of Lisbon