An efficient model for estimation of gypsum (calcium sulfate di-hydrate) solubility in aqueous electrolyte solutions over wide temperature ranges

Abstract

Thermodynamic solubility of calcium sulfate di-hydrate (gypsum) plays a major role in many operations involved in oilfields and process industries. In this study, several solubility models were developed using the approaches including Multilayer Perceptron (MLP) networks, Radial Basis Function (RBF) networks, Least Squares Support Vector Machine (LSSVM), Adaptive Network-based Fuzzy Inference system (ANFIS), and Committee Machine Intelligent System (CMIS) for estimating the phase behavior of calcium sulfate dihydrate (or gypsum) in strong electrolyte solutions at high temperatures. The developed CMIS model was found to outperform all other modeling approaches indicating an excellent accordance with experimental data and yielding an overall correlation coefficient of (R2) 0.9968. The developed CMIS model was also found to outperform the ion interaction model of Pitzer providing superior predictability for gypsum solubility in aqueous electrolyte solutions. The results of this study indicate that gypsum solubility could be estimated by employing the CMIS approach with high accuracy in Na-Ca-Mg-Fe-Al-H-Cl-H2O system over temperatures ranging from 5 to 98 °C.