Input data for indirect estimation of swelling pressure of expansive soils

Soil swelling is one of the most complex geotechnical phenomena that an engineer has to deal with in a wide range of civil engineering structures. Various methods can be used to determine the swelling soil potential. Some, such as the mineralogical identification and direct swelling measurements, are more or less time consuming and require specific equipment. However, there are other relatively fast and less expensive methods that are based on the use of multiple nonlinear regression and artificial neural networks. These procedures evaluate the behavior of swelling soils based on easily determined basic geotechnical parameters. This paper focuses on the reliability of the input data used to determine the swelling pressure. The most frequently occurring input parameters in indirect predictions were selected (Atterberg limits, percentage of grains, index of colloidal activity). These parameters also use the methodology "Identification and Solution of the Problem of Swelling Prone Soils" certified by the Ministry of the Environment of the Czech Republic.