A model for determining the cyclic swell–shrink behavior of argillaceous rock

The swelling potential of rocks including clay minerals in arid and semi-arid regions, where the rocks experience completely dry to fully saturated conditions, is a complex problem. A method is lacking for determining the swell–shrink behavior of such rocks under variable water contents and boundary conditions prevailing in underground excavations. This paper suggests a method, which finally leads to a model, for determining the swell–shrink behavior of mudrock at the Masjed–Soleiman underground hydroelectric power plant (UHEPP). The recorded nonlinear behavior of the mudrock is modeled using an artificial neural network (ANN). The ANN model utilizes the information from the previous swell–shrink cycles to learn and model the dynamics of swell–shrink behavior. Using this method facilitates prediction of swelling pressure of argillaceous rocks with different swelling potentials and site conditions.