HANNA: Hard-constraint Neural Network for Consistent Activity Coefficient Prediction

TL;DR

HANNA is a novel neural network model that enforces thermodynamic consistency in predicting activity coefficients, outperforming existing models and applicable to any binary mixture using only component SMILES.

Contribution

This work introduces the first hard-constraint neural network for activity coefficient prediction that ensures thermodynamic consistency and symmetry, advancing the accuracy and applicability of such models.

Findings

HANNA achieves higher prediction accuracy than UNIFAC.

It maintains thermodynamic consistency through physical constraints.

The model is applicable to any binary mixture using only SMILES.

Abstract

We present the first hard-constraint neural network for predicting activity coefficients (HANNA), a thermodynamic mixture property that is the basis for many applications in science and engineering. Unlike traditional neural networks, which ignore physical laws and result in inconsistent predictions, our model is designed to strictly adhere to all thermodynamic consistency criteria. By leveraging deep-set neural networks, HANNA maintains symmetry under the permutation of the components. Furthermore, by hard-coding physical constraints in the network architecture, we ensure consistency with the Gibbs-Duhem equation and in modeling the pure components. The model was trained and evaluated on 317,421 data points for activity coefficients in binary mixtures from the Dortmund Data Bank, achieving significantly higher prediction accuracies than the current state-of-the-art model UNIFAC. Moreover, HANNA only requires the SMILES of the components as input, making it applicable to any binary mixture of interest. HANNA is fully open-source and available for free use.