A hybrid molecular dynamics - machine learning framework for boiling point estimation in aromatic fluids

Aug 2, 2026

Precise estimation of boiling points in organic fluids is essential for designing thermal systems that are both efficient and safe. This study introduces a hybrid molecular dynamics–machine learning framework for aromatic fluids, using biphenyl and diphenyl ether as representative cases. Two force fields (OPLS-AA and COMPASS) are benchmarked in equilibrium MD, then density-based and inflection-point methods extract boiling onset and completion. MD-derived data train NNR, NN, and SVR, enabling reproducible and accurate boiling prediction with model-driven evaluation.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages23

Reading Time58 min

AccessFree to download

Introduction

Importance of thermophysical property prediction

Prior work on boiling point modelling

Role of molecular dynamics simulations

Case Studies in Thermal Engineering 73 (2025) 106684 Contents lists available at ScienceDirect Case Studies in Thermal Engineering journal [homepage: www.elsevier.com/locate/csite](homepage: www.elsevier.com/locate/csite) A hybrid molecular dynamics–machine learning framework for boiling point estimation in aromatic fluids Amirali Shateri , Zhiyin Yang , Nasser Sherkat , Jianfei Xie * School of Engineering, University of Derby, DE22 3AW, UK | G R A P H I C A L A B S T R A C T | | --- | | | A R T I C L E I N F O Keywords: Machine learning Molecular dynamics simulation Boiling point Thermal fluids Phase change A B S T R A C T Precise estimation of boiling points in organic fluids is critical for designing efficient and safe thermal systems. This study presents a hybrid molecular dynamic (MD)–machine learning (ML) framework for boiling point estimation in two representative aromatic fluids: biphenyl (C 12H10) and diphenyl ether (C 12H10O). Two force fields, OPLS-AA and COMPASS, were tested in equilibrium MD simulations. OPLS-AA produced density predictions with a relative error below 2 % compared to experimental values, while COMPASS showed reduced accuracy at elevated temperatures. Boiling point was estimated using a density threshold method (yielding 525.66 K) and * Corresponding author. E-mail address: [j.xie@derby.ac.uk](j.xie@derby.ac.uk) (J. Xie). [https://doi.org/10.1016/j.csite.2025.106684](https://doi.org/10.1016/j.csite.2025.106684) Received 4 April 2025; Received in revised form 4 July 2025; Accepted 10 July 2025 Available online 10 July 2025 2214-157X/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license ([http://creativecommons.org/licenses/by/4.0/](http://creativecommons.org/licenses/by/4.0/)). A. Shateri et al. Case Studies in Thermal Engineering 73 (2025) 106684 a thermodynamically rigorous inflection-point method (508.18 K), revealing ~3.3 % deviation between boiling onset and completion. MD data were used to train and evaluate three regression models—Nearest Neighbours Regression (NNR), Neural Network (NN), and Support Vector Regression (SVR). The NNR model achieved the best match with MD data, predicting a boiling point of 524.97 K and density of 0.064 g/cm3. The NN model accurately estimated boiling temperature (525.3 K) but overestimated density, while SVR underestimated both. This work contributes a novel, interpretable MD–ML framework to integrate the inflection-point detection with data-driven model selection, offering a reproducible and accurate method for boiling point estimation that can be extended to other organic thermal systems. 1. Introduction Accurate predictions of thermophysical properties of fluids are essential to design efficient and sustainable thermal fluid systems for many industrial applications, e.g., chemical processing and thermal management. The chemical makeup of heat transfer fluids (HTFs) along with their operating conditions determines their thermal stability and heat transfer efficiency. Previous studies show that thermal fluids degrade because of thermal and chemical stresses during normal operations, underlining the importance of fluid chemical decomposition and environmental conditions for long-term performance [1,2]. Rigorous system design practices are essential to maximize the service life of HTFs [3]. The design of heat exchangers and heaters demands that engineers meticulously evaluate the heat flux control, temperature gradients, and thermal boundary conditions, with specific attention to energy recovery units [4]. In scenarios where flame impingement is present localized overheating often happens and raises the risk of premature thermal degradation [5]. Proper fuel-to-air mixing becomes essential because incorrect mixing leads to changes in flame dynamics which amplify thermal stresses [6,7]. The stability of HTF fluids is seriously compromised by chemical contamination beca

FAQs

What aromatic fluids are used to validate the hybrid MD–ML framework?

The framework is tested on biphenyl and diphenyl ether as representative aromatic fluids.

How are boiling points estimated from MD simulations in this study?

Boiling points are obtained using a density-threshold method and a thermodynamically rigorous inflection-point method.

Which regression model performs best and what is its key outcome?

Nearest Neighbours Regression (NNR) gives the best match with MD data, predicting a boiling point of 524.97 K and a density of 0.064 g/cm3.

Aug 2, 2026

Upload by River Wang