20 December 2024 in Chemical Engineering Journal
6 June 2022 in AIChE Journal
Methods for predicting Henry's law constants Hij are important as experimental data are scarce. We introduce a new machine learning approach for such predictions: matrix completion methods (MCMs) and demonstrate its applicability using a data base that contains experimental Hij values for 101 solutes i and 247 solvents j at 298 K. Data on Hij are only available for 2661 systems i + j. These Hij are stored in a 101 × 247 matrix; the task of the MCM is to predict the missing entries. First, an entirely data‐driven MCM is presented. Its predictive performance, evaluated using leave‐one‐out analysis, is similar to that of the Predictive Soave‐Redlich‐Kwong equation‐of‐state (PSRK‐EoS), which, however, cannot be applied to all studied systems. Furthermore, a hybrid of MCM and PSRK‐EoS is developed in a Bayesian framework, which yields an unprecedented performance for the prediction of Hij of the studied data set.
11 October 2022 in Digital Discovery
Experimental data on diffusion in binary liquid mixtures at 298 ± 1 K from the literature were systematically consolidated and used to determine diffusion coefficients D∞ ij of solutes i at infinite dilution in solvents j in a consistent manner. The resulting database comprises basically all data on D∞ ij at 298 K that are available and includes 353 points, covering 208 solutes and 51 solvents. In a first step, the new database was used to evaluate semiempirical methods for predicting D∞ ij from the literature, namely the methods of Wilke and Chang, Reddy and Doraiswamy, Tyn and Calus, and SEGWE, of which SEGWE yielded the best results. Furthermore, a new method for the prediction of D∞ ij based on the concept of matrix completion from machine learning was developed, which exploits the fact that experimental data for D∞ ij can be represented as elements of a sparse matrix with rows and columns corresponding to the solutes i and solvents j; it is demonstrated that matrix completion methods (MCMs) can be used for closing the gaps in this matrix. Three variants of this approach were studied here, a purely data-driven MCM and two hybrid MCMs, which use information from SEGWE together with the experimental data. The methods were evaluated using the new database. The hybrid MCMs outperform both the data-driven MCM and all established semiempirical models in terms of predictive accuracy.
9 May 2025 in Industrial & Engineering Chemistry Research
8 January 2020 in Journal of the Electrochemical Society
Redox flow batteries (RFBs) are considered an outstanding candidate for the integration of renewable energy sources into the existing power grids. A key property of RFBs is the open circuit voltage (OCV) corresponding to the currentless equilibrium state. In the literature, the Nernst equation describing this property is often simplified by neglecting the activity coefficients. In this work, using a thermodynamically rigorous approach, we show that activity coefficients have a significant influence on the OCV of the Iron-Cadmium and All-Vanadium RFBs. Moreover, this influence varies with the state of charge. Therefore, activity coefficients should not be neglected in the Nernst equation. We show that when doing so, the resulting offset in OCV is actually comparable to typical voltage losses occurring during operation. Hence, fitting kinetic parameters to measurement data of voltage losses can lead to ambiguous results if only the idealized OCV, obtained by neglecting the activity coefficients, is used in that evaluation. Therefore, the implementation of a thermodynamically rigorous model has the potential to significantly improve state-of-the-art models for RFBs.
30 December 2024 in The Journal of Physical Chemistry B