An investigation of automated physicochemical property modelling of aqueous strong electrolyte solutions and their mixtures

Darren Rowland

An investigation is described into the thermodynamic modelling of aqueous electrolyte solutions. Solutions relevant to environment and industry are emphasised. Computational facilities are described that automate the processing of physicochemical property data. Various computer models for predicting the thermodynamic quantities of aqueous solutions are developed. Used together these facilities allow data to be assessed and harmonised in a more effective manner than existing methods. Considerable attention is paid to the popular Pitzer theoretical framework but other frameworks are also studied, especially the application of so-called `mixing rules' to describe multicomponent solutions from the properties of their individual solute components. The thermodynamic properties of strong electrolyte solutions under ambient conditions are well described by Pitzer equations provided a sufficient number of accurate data are available. However, extending the pressure and temperature ranges reveals fundamental limitations. Gaps or large errors (which are inevitable over large dimensional spaces) cause unrealistic extrapolations and other problems requiring human intervention (e.g., highly specifc and subjective judgements) to redress. Objective and robust automated modelling based on the Pitzer equations is thus precluded. Accordingly, a selection of possible alternative theoretical frameworks are investigated. One, based on the Huckel equation for activity coefficients, has fewer empirical parameters than the corresponding Pitzer equation and its extrapolations of the apparent molar volumes and heat capacities to infinite dilution, performed for the first time in this work, are more robust. However, its correlation of activity coefficients is correspondingly less accurate. Different mixing rules for density, heat capacity and water activity are compared with one another. These comparisons show that various rules give essentially equivalent property predictions for a range of ternary systems. Contrasted with experimental osmotic coefficients, predictions from Zdanovskii's rule are often within assessed experimental uncertainty, which can be up to an order-of-magnitude worse than that claimed by investigators. Simple mixing rules, like those of Zdanovskii, therefore appear to offer advantages for future model development of multicomponent aqueous strong electrolyte solutions.

An investigation of automated physicochemical property modelling of aqueous strong electrolyte solutions and their mixtures

Files and links (2)

Abstract

Details

Metrics