Predicting Temperature‐Dependent Activity Coefficients at Infinite Dilution Using Tensor Completion

Damay, Julie; Ryzhakov, Gleb; Jirasek, Fabian; Hasse, Hans; Oseledets, Ivan; Bortz, Michael

doi:10.1002/cite.202200230

Cited by 4 publications

(6 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This phenomenon aligns with expectations, as extrapolating to new solvents within polymer–solvent mixtures presents a more challenging task compared to interpolating among known chemical species. This can be confirmed, by the fact that interpolation can even be approached (with remarkable accuracy) using matrix or tensor completion techniques that do not need explicit chemical structure insights. − …”

Section: Results and Discussionmentioning

confidence: 91%

See 1 more Smart Citation

Gibbs–Helmholtz Graph Neural Network for the Prediction of Activity Coefficients of Polymer Solutions at Infinite Dilution

Sanchez Medina,

Kunchapu,

Sundmacher

2023

J. Phys. Chem. A

View full text Add to dashboard Cite

Machine learning models have gained prominence for predicting pure-component properties, yet their application to mixture property prediction remains relatively limited. However, the significance of mixtures in our daily lives is undeniable, particularly in industries such as polymer processing. This study presents a modification of the Gibbs− Helmholtz graph neural network (GH-GNN) model for predicting weightbased activity coefficients at infinite dilution (Ω ij ∞ ) in polymer solutions. We evaluate various polymer representations ranging from monomer, repeating unit, periodic unit, and oligomer and observe that, in data-scarce scenarios of polymer−solvent mixtures, polymer representation specifics have a reduced impact compared to data-rich environments. Leveraging transfer learning, we harness richer activity coefficient data from small-size systems, enhancing model accuracy and reducing prediction variability. The modified GH-GNN model achieves remarkable prediction results in mixture interpolation and solvent extrapolation tasks having an overall mean absolute error of 0.15, showcasing the potential of graph-neural-network-based models for property prediction of polymer solutions. Comparative analysis with the established models UNIFAC-ZM and Entropic-FV suggests a promising avenue for future research on the use of data-driven models for the prediction of the thermodynamic properties of polymer solutions.

show abstract

Section: Results and Discussionmentioning

confidence: 91%

“…This can be confirmed, by the fact that interpolation can even be approached (with remarkable accuracy) using matrix or tensor completion techniques that do not need explicit chemical structure insights. 43 − 45 …”

Section: Results and Discussionmentioning

confidence: 99%

Gibbs–Helmholtz Graph Neural Network for the Prediction of Activity Coefficients of Polymer Solutions at Infinite Dilution

Sanchez Medina,

Kunchapu,

Sundmacher

2023

J. Phys. Chem. A

View full text Add to dashboard Cite

show abstract

“…Within KEEN, these methods have been complemented with physical knowledge [18,19] and were applied to different physical properties [20,21]. In this special issue, matrix-completion is generalized to tensor completion techniques [43]. It is shown that this allows to systematically extract information on the dependence of mixture properties on a physical variable, e.g., temperature.…”

Section: Modeling and Simulationmentioning

confidence: 99%

AI in Process Industries – Current Status and Future Prospects

Bortz

Dadhe

Engell

et al. 2023

Chemie Ingenieur Technik

Self Cite

View full text Add to dashboard Cite

The chemical industry is one of the key industrial sectors in Germany and at the same time one of the largest consumers of energy and raw materials. A successful energy transition and the development of a circular economy can only succeed if they are actively supported and shaped by the chemical industry – through the redesign of existing production processes and the exploration and implementation of new process routes. The challenge is to realize this transformation within a very short time and for many production processes, whereby a much larger number of process routes must be explored. Digital technologies are key to master this transformation towards more sustainability, climate, and environmental protection. The KEEN project aims to explore and leverage artificial intelligence (AI) opportunities in process industry. The newly developed AI methods are tested wherever possible in real working environments and production plants to prove the economic benefit, applicability, and reliability of the methods and technologies.

show abstract

“…The crux of MCMs lies in their proficiency in inferring missing entries by identifying patterns within the matrix and leveraging similarities across rows and columns. Recent explorations, such as those by Jirasek et al, 16 Damay et al, 17 and Chen et al, 18 have applied MCMs to predict γ values in both traditional solvent systems and ionic liquid (IL) systems, demonstrating the method's versatility. For instance, Jirasek et al 16 applied a MCM for predicting infinite dilution γ values (γ ∞ ) at ambient temperature, showing favorable performance compared to UNIFAC.…”

Section: Introductionmentioning

confidence: 99%

Neural Network-Based Tensor Completion: Advancing Predictions of Activity Coefficients and Beyond

Averbeck,

Sadowski,

Held

et al. 2024

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

Although existing tensor completion methods have progressed in predicting two-and three-dimensional data, they still struggle to capture nonlinearities and temporal dependencies in relational data effectively. We introduce an innovative solution to this research gap: our novel 3D-DMF-H method for tensor completion. Developed as a neural network-based matrix completion approach, our method extends the Deep Matrix Factorization (DMF) method, handling nonlinear data structures and effortlessly incorporating additional data points. Our method applies to a wide range of three-dimensional tensor completion problems, and exceptional accuracy was achieved in predicting activity coefficients to model phase equilibria. Notably, the 3D-DMF-H method outperforms the benchmark thermodynamic g E model UNIFAC (standard) and significantly enhances the accuracy of azeotrope predictions when integrated with the equation of state PC-SAFT. Our findings demonstrate the promising potential of machine learning for advancing applications in the chemical industry and highlight the necessity for further algorithmic refinement and exploration.

show abstract

Predicting Temperature‐Dependent Activity Coefficients at Infinite Dilution Using Tensor Completion

Cited by 4 publications

References 39 publications

Gibbs–Helmholtz Graph Neural Network for the Prediction of Activity Coefficients of Polymer Solutions at Infinite Dilution

Gibbs–Helmholtz Graph Neural Network for the Prediction of Activity Coefficients of Polymer Solutions at Infinite Dilution

AI in Process Industries – Current Status and Future Prospects

Neural Network-Based Tensor Completion: Advancing Predictions of Activity Coefficients and Beyond

Contact Info

Product

Resources

About