Modeling Solubility of Anhydrite and Gypsum in Aqueous Solutions: Implications for Swelling of Clay-Sulfate Rocks

Abstract The swelling of clay-sulfate rocks is a well-known phenomenon often causing threats to the success various geotechnical projects, including tunneling, road and bridge construction, geothermal drilling. origin usually explained by physical due clay expansion combined with chemical associated transformation anhydrite (CaSO 4 ) into gypsum ∙2H 2 O). latter occurs through dissolution subsequent precipitation. Numerical models that simulate rock must consider hydraulic, mechanical, processes. simulation processes performed solving thermodynamic equations, which contribute significant portion overall computation time. This paper employs feed-forward neural network (FFNN) cascade-forward (CFNN) trained Bayesian regularization (BR) algorithm as an alternative approach determine solubility in aqueous phase. are developed using calcium sulfate experimental data collected from literature. Our results indicate FFNN-BR most accurate model for regression task. comparison analysis Pitzer ion interaction well previously published data-driven shows highly determining minerals acid salt-containing solutions. We conclude our can be used needed address typical subsurface engineering problems such rocks.