Graph neural network improves liquid mixture predictions for process design
Learning atomic interactions for the thermodynamics of liquid mixtures
Machine LearningComputational Engineering, Finance, and Science
Summary
Predicting how liquids mix is important for designing chemical processes, but current methods either lack accuracy or are costly to run. The authors created COGNAC, a machine learning model that learns how individual atoms interact in liquid mixtures and uses this to make better predictions. It works faster than other modern methods and is more accurate than traditional ones. This helps engineers simulate and optimize chemical processes more effectively.
What this means in practice
- •For chemical process engineers: Run accurate activity coefficient predictions within existing simulators to optimize chemical mixture processes more efficiently.
- •For pharmaceutical formulation teams: Predict liquid mixture behavior to better design multi-component drug formulations with reliable thermodynamic properties.
Authors
Marco Hoffmann, Sebastian Schmitt, Hans Hasse, Fabian Jirasek
Abstract
Predictive models for activity coefficients, which describe the non-ideality of liquid mixtures, are foundational for process design and optimization. Group-contribution methods, such as UNIFAC, are computationally cheap and well established in process simulators. Modern machine learning models, such as HANNA, offer higher accuracy and broader applicability, but their integration into process simulation is expensive or even infeasible. In this work, we introduce COGNAC, a graph neural network that learns atom-level geometry and interaction parameters that depend on the molecular neighborhood and evaluates them in the UNIFAC framework. Trained on about 820,000 experimental data points, COGNAC achieves significantly higher accuracy than UNIFAC and HANNA for binary and multi-component mixtures, reduces error relative to HANNA by 13-63%, reliably predicts multi-phase equilibria and identifies azeotropes, yields interpretable parameters, and is strictly thermodynamically consistent. COGNAC evaluations run at a cost comparable to UNIFAC, thereby unlocking the potential of machine-learning models for process simulations.