Thermodynamic modeling and parameter estimability analysis of a wet phosphoric acid process with impurities

Abstract

In this work, a thermodynamic-based model of an industrial phosphoric acid process is developed. It consists of mass and charge balance equations along with equilibria and Pitzer’s model equations. The latter involves several unknown parameters to be identified from experiments. Experimental measurements consisting mainly of speciation data of sulfur and phosphate-based systems, solubility data of ten minerals, and water activity of eight binary systems, are used. Part of the measurements are collected from the literature and the rest is carried out in the laboratory. A global estimability analysis method recently developed allowed us to determine which parameters can be identified from the available measurements, and those to be fixed from the literature or from previous works. The Pearson product moment coefficient is computed and found to be higher than 97% for all the identification cases, demonstrating the accuracy of the model. The resulting model is then validated by means of additional measurements which are not used for its identification, using the t-Student, Fisher–Snedecor and Kolmogorov–Smirnov tests. Normalized root mean square errors are computed for all the validation cases and found to be less than 7%, showing good model performance. The validated model can now be used to predict several properties needed for the simulation and optimization of the industrial process under study.