Abstract:Dimethyl ether (DME) is arousing attention as a novel water-soluble additive in oil industry. The accurate prediction of phase behavior of DME and water system is crucial important for DME-based research. An equation of state has been a powerful tool for solving phase equilibrium problem, which is based on an ideal molecular force model with binary interaction coefficients reflecting the deviation of a non-ideal system. However, the current knowledge of interaction coefficient is limited by obtaining from backward by using phase equilibrium data. This leads to the unclear meaning of the interaction coefficients, insufficient response to the molecular force interaction relationship at the molecular level and no ability to predict phase equilibrium properties without primary use of experimental data. To overcome this, in this study a correction by combining an activity coefficient and a linear solvation energy relationship (LSER) has been proposed to address the solubility deviation from an idealized model. This deviation at the micro level indicates the deviation of nonideal molecular forces from the ideal Raoult's law. Based on this novel correction, binary interaction coefficients have been obtained with a more practical molecular interaction relationship, which leads to more consistent results with experimental data from the literature. Moreover, in asymmetric phase systems, concentration-dependent binary interaction coefficients have been proposed based on previous discussion to represent intermolecular forces in non-ideal phase systems. Furthermore, partitioning coefficients, which lead to a potential criterion for solvent selection for heavy oil recovery, have been discussed from the perspective of intermolecular forces. This study provides new insights into non-ideal phase behavior. By combining a LSER and activity coefficients, a quantitative analysis on non-ideal molecular behavior and related complex phase problems becomes possible.

Bridging Machine Learning and Redlich–Kister Theory for Solid–Liquid Equilibria Prediction of Binary Eutectic Solvent Systems

Evaluation of COSMO-RS for solid–liquid equilibria prediction of binary eutectic solvent systems

Rational Eutectic Solvent Design by Linking Regular Solution Theory with QSAR Modelling

Extending UNIFAC Models for Solid-Liquid Equilibria Prediction and Design of Eutectic Solvent Systems

Study on the Factors Affecting Solid Solubility in Binary Alloys: an Exploration by Machine Learning

Machine-Learning-Assisted Design of Deep Eutectic Solvents Based on Uncovered Hydrogen Bond Patterns

Machine Learning for Predicting Thermodynamic Properties of Pure Fluids and Their Mixtures

Designing solvent systems using self-evolving solubility databases and graph neural networks

Prediction of Hydration energies of Adsorbates at Pt(111) and Liquid Water Interfaces using Machine Learning

Liquid–Liquid Equilibrium with Ab Initio Molecular Dynamics Simulation and NRTL Parameter Estimation

Machine learning force fields for molecular liquids: Ethylene Carbonate/Ethyl Methyl Carbonate binary solvent

Machine learning with physicochemical relationships: solubility prediction in organic solvents and water

Random forest models to predict the densities and surface tensions of deep eutectic solvents

Prediction of the viscosity of green deep eutectic solvents by constructing ensemble model based on machine learning

Modeling Chemical Processes in Explicit Solvents with Machine Learning Potentials

Solvent Selection Scheme Using Machine Learning Based on Physicochemical Description of Solvent Molecules: Application to Cyclic Organometallic Reaction

Machine-Learning Prediction of Vegard's Law Factor and Volume Size Factor for Binary Substitutional Metallic Solid Solutions

Interpretable Machine Learning Model for Predicting Interaction Energies Between Dimethyl Sulfide and Potential Absorbing Solvents

Vapor-liquid Equilibrium (VLE) Prediction for Dimethyl Ether (DME) and Water System in DME Injection Process with Peng-Robinson Equation of State and Composition Dependent Binary Interaction Coefficient

Prediction of Liquid-Liquid Equilibrium Using the Group Solubility Parameter Model

Rational Electrolytes Design for Li-Metal Batteries Operated Under Extreme Conditions: A Combined DFT, COSMO-RS, and Machine Learning Study