Graduation Semester and Year




Document Type


Degree Name

Doctor of Philosophy in Civil Engineering


Civil Engineering

First Advisor

Anand Puppala


Expansive soils are spread over different climatic zones around the world with varying mineralogical content and microstructure, making the soil to absorb moisture between their mineral layers and therefore undergo volume changes. This behavior generates swell and shrinkage surface movements which affect the integrity of infrastructure built on them. A practice currently employed correlates swell-shrink behavior to soil index parameters, however, variable soil mineralogy limits this approach, which generates poor soil swell characterization. Thus, characterizing efforts must be emphasized on identification of real parameters governing the swell/shrink behavior, such as mineralogy, variation of suction with moisture content and pore distribution. The present work intends to validate previously formulated swell behavior models for clays based on these parameters and to utilize the data obtained from the validation process to generate a unified formulation that assesses the effect of the mentioned key parameters in the prediction and characterization of soil swell behavior. The unified model was defined using multiple linear regression (MLR) and artificial neural network (ANN) techniques. Both approaches exhibited acceptable prediction capacity, however, ANN models showed higher prediction capability than MRL models. ANN proved its usefulness for complementing or replacing MLR in soil swell behavior characterization


Civil and Environmental Engineering | Civil Engineering | Engineering


Degree granted by The University of Texas at Arlington