📄 chemistry

Semi-empirical thermodynamic correlation of vapor pressure behavior in biodiesel–ethanol mixtures using group-structured molecular descriptors

This study develops and validates a semi-empirical thermodynamic framework incorporating group-structured molecular descriptors to accurately model the non-ideal vapor–liquid equilibrium behavior of biodiesel–ethanol mixtures, achieving significantly lower prediction errors than conventional Raoult's law.

Original authors: Esmaeil Bahrani, Mohammad Saleh Shafeeyan, Morteza Banihashemi

Published 2026-08-03

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Esmaeil Bahrani, Mohammad Saleh Shafeeyan, Morteza Banihashemi

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to bake the perfect cake, but instead of flour and sugar, your ingredients are liquid fuels. In the world of engines, getting the fuel to turn into a gas at just the right moment is crucial. If it turns to gas too early, the engine sputters; too late, and it runs rough. This "turning to gas" tendency is called vapor pressure. Scientists have long known that mixing different fuels can change this pressure, much like adding lemon juice changes the taste of a soup. However, when you mix a heavy, oily fuel (biodiesel) with a light, watery one (ethanol), they don't play nice together. They are like oil and water, or a shy introvert and an extrovert at a party; they interact in messy, unpredictable ways that simple math formulas can't easily explain. This is where thermodynamics comes in—the branch of science that studies how heat and energy move and how substances behave when mixed. Understanding these mixtures is vital because as the world looks for cleaner energy, we need to know exactly how these new fuel blends will behave inside an engine to make them safe and efficient.

This paper tackles the tricky problem of predicting the vapor pressure of a specific fuel cocktail: biodiesel (made from soybean oil) mixed with ethanol. The researchers found that the old, standard way of calculating this pressure—using a rule called Raoult's Law—was like trying to predict the weather using only a calendar; it was too simple and often got the answer wrong, especially when there was very little ethanol in the mix. The error was significant, missing the mark by an average of about 11.63 kPa.

To fix this, the authors built a new, smarter tool. Instead of treating the fuel molecules as single, boring blocks, they broke them down into their "Lego pieces"—functional groups like carbon chains and oxygen atoms. They then used a computer program called a genetic algorithm (which works a bit like natural selection, where the best guesses survive and improve over time) to find the perfect way to arrange these pieces in their math. By feeding the computer data from experiments where they heated up nine different mixtures (ranging from 2% ethanol to 80% ethanol) and measured the pressure, the new model learned the secret handshake between the molecules.

The result? The new "group-structured" model was a massive improvement. It predicted the pressure with an average error of only 4.07 kPa, which is about 65% better than the old method. The authors suggest that this approach works because it accounts for the specific shapes and interactions of the molecules, rather than just guessing based on the total amount of liquid. They tested this on mixtures with ethanol concentrations of 2%, 4%, 6%, 8%, 10%, 20%, 30%, 60%, and 80%, and found that while the old method struggled the most with low ethanol amounts (missing by as much as 27.70 kPa in the 2% mix), their new method kept the error down to just 2.6 kPa for that same difficult mix.

It is important to note that this isn't a magic wand that solves every fuel problem in the universe. The authors explicitly state that their model is a "semi-empirical" correlation, meaning it is a mathematical bridge built specifically for this type of biodiesel and ethanol mix. It is not a universal law that can be instantly applied to any random chemical mixture without further testing. They also ruled out the idea that the old, simple formulas were sufficient for these complex, non-ideal mixtures. While the model is highly accurate within the tested range (from atmospheric pressure up to 1 bar gauge), the authors caution that it is designed for interpolation—filling in the gaps between the data they collected—rather than for wild guesses far outside those conditions. Ultimately, they have shown that by looking at the molecular "Lego bricks" and using a smart computer search, we can understand these fuel blends much better than before, paving the way for safer, more efficient renewable energy.

Technical Summary: Semi-empirical thermodynamic correlation of vapor pressure behavior in biodiesel–ethanol mixtures using group-structured molecular descriptors

Problem Statement
The study addresses the challenge of accurately modeling the vapor–liquid equilibrium (VLE) behavior of biodiesel–ethanol mixtures. While biodiesel is a viable renewable fuel alternative, its blending with ethanol introduces significant non-ideal thermodynamic behavior, particularly at low ethanol concentrations. These deviations arise from strong molecular interactions and polarity differences between ethanol and the long-chain methyl esters found in biodiesel. Classical thermodynamic models, such as the standard Raoult's law, fail to capture these non-idealities, leading to substantial prediction errors. Furthermore, established activity coefficient models (e.g., Wilson, NRTL, UNIQUAC) often rely on fitted binary interaction parameters derived from experimental data, limiting their predictive capability when such data is scarce. The authors identify a need for a simplified, structurally informed thermodynamic correlation framework that can represent the equilibrium pressure trends of these renewable fuel blends without requiring extensive experimental parameter fitting for every new system.

Methodology
The research employed a combined experimental and semi-empirical modeling approach:

Material Synthesis and Characterization: Biodiesel was synthesized via alkali-catalyzed transesterification of soybean oil using methanol and potassium hydroxide. The resulting fatty acid methyl esters (FAMEs) were characterized using Gas Chromatography (GC) to determine their specific compositional fractions (e.g., C16:0, C18:1, C18:2). Pure ethanol and biodiesel–ethanol mixtures with varying ethanol weight percentages (2% to 80%, plus pure components) were prepared for testing.
Experimental VLE Data Acquisition: Equilibrium vapor pressure–temperature data were collected for nine specific mixtures and pure components over a pressure range from atmospheric up to 1 bar gauge. The experiments utilized a custom-fabricated stainless steel chamber equipped with Pt1000 RTD temperature sensors and Hogller pressure sensors. Equilibrium was defined by the stabilization of temperature and pressure readings. The uncertainty was estimated at ±0.1 °C for temperature and ±0.01 bar for pressure.
Thermodynamic Modeling Framework: The authors developed a semi-empirical framework based on a modified Raoult's law. Instead of using rigorous excess Gibbs energy formulations, the model introduced effective activity-correction terms ( $\gamma$ $γ$ ) derived from group-structured molecular descriptors.
- Group Decomposition: Ethanol and the identified methyl esters were decomposed into functional groups (e.g., $-CH_3$ , $-CH_2$ , $-CH=CH-$, $-OH$, $-COOCH_3$ ).
- Correction Terms: The correction factors were formulated as functions of component mole fractions and the weighted sums of functional group contributions, containing adjustable coefficients ( $\alpha, \beta, \gamma, \delta$ ).
Optimization: A genetic algorithm was employed to optimize the group-associated fitting parameters and adjustable coefficients. The algorithm utilized 60% of the experimental dataset over 100 generations to minimize the absolute error between calculated and experimental vapor pressures. The remaining 40% served as a validation set (implied by the error analysis structure).

Key Contributions

Experimental Dataset: The study provides meticulously acquired equilibrium vapor pressure–temperature data for a wide range of biodiesel–ethanol compositions (0% to 100% ethanol), filling a gap in the literature regarding the phase behavior of these specific blends.
Group-Structured Correlation: The authors propose a novel semi-empirical correlation that integrates molecular structural information (functional group distributions) directly into the activity correction terms of a modified Raoult's law.
Optimization Strategy: The application of a genetic algorithm to optimize the structural descriptors and fitting parameters allows for a robust correlation of non-ideal behavior without relying on pre-existing binary interaction parameters from other systems.

Results

Experimental Trends: The data confirmed that vapor pressure increases with temperature and ethanol concentration. However, significant non-ideal deviations were observed, particularly in mixtures with low ethanol content.
Model Performance: The proposed semi-empirical framework significantly outperformed the conventional Raoult's law.
- Conventional Raoult's Law: Yielded an average absolute error of approximately 11.63 kPa. The highest error occurred at 2% ethanol concentration (27.70 kPa).
- Proposed Group-Structured Model: Reduced the average absolute error to approximately 4.07 kPa, representing a roughly 65% reduction in error magnitude.
- Specific Improvements: The modified model showed the most dramatic improvement at low ethanol concentrations (e.g., reducing error from 27.70 kPa to 2.6 kPa at 2% ethanol).
Antoine Correlations: The study also derived mixture-specific Antoine equation coefficients for each composition to facilitate interpolation within the experimental domain, noting that these are empirical representations specific to the tested conditions.

Significance and Claims
The paper claims that its primary contribution is the development of a "simplified yet structurally informed thermodynamic representation" for renewable fuel mixtures. The authors assert that incorporating molecular group descriptors into a modified Raoult's law framework significantly improves the correlation of experimentally observed non-ideal vapor–liquid equilibrium behavior compared to classical approaches.

The study explicitly positions its framework as a system-specific correlation approach rather than a universal predictive group contribution model transferable to arbitrary chemical systems. The authors emphasize that the model is intended to represent the experimentally measured equilibrium pressure data for the specific biodiesel–ethanol system investigated. They acknowledge that independent transferability validation against external binary systems was beyond the scope of this work. The significance lies in demonstrating that a semi-empirical approach, grounded in molecular structure and optimized via genetic algorithms, can effectively capture the complex thermodynamics of biodiesel blends, offering a practical tool for predicting vapor pressure behavior where traditional ideal or fitted-parameter models fail.

More like this