Estimation of the Swelling Index and Swelling Pressure of Expansive Soils Using Multiple Artificial Intelligence Techniques

The performance and safety of geoinfrastructures, such as foundations, pipelines, pavements, retaining walls, and slopes, constructed with or within expansive soils offer challenges due to their swelling behavior characteristics. Swelling index and swelling pressure are two key properties required for characterizing the expansive soils behavior. These properties that are determined from conventional oedometer tests are time-consuming, require trained professionals, and hence are expensive. A simple approach for the reliable estimation of the swelling index and swelling pressure that can be used for expansive soils of all regions in the world will be valuable for practicing engineers. For achieving this objective, three artificial intelligence (AI) algorithms, multilayer perceptron, support vector machine, and extreme learning machine, have been employed in this study. By utilizing these advanced AI models along with the aid of regression analyses tools, two equations are proposed for the estimation of the swelling index based on the plasticity index information for 12 expansive soils. In addition, multiple AI models and equations have also been developed to estimate the swelling pressure of diverse expansive soils. The swelling index and swelling pressure can be estimated using spreadsheets or hand calculators for both the AI models and the proposed equations in this study. The good agreement between the estimated values and measured data using the simple approaches is encouraging for geotechnical engineers to implement them in conventional practice.