Modeling the Crack Network of Swelling Clay Soils

Abstract

The dynamic crack network in swelling clay soils is a major factor determining their hydraulic properties and particularly preferential flow, with important consequences for ecologically adverse processes. We developed a physically based probabilistic model for the prediction of crack network geometry that takes into account the dependence of the crack concentration on soil depth. Starting with the similarity of crack networks in rocks and soils, we used the multiple cracking and fragmentation model available for rocks. In the case of cracked soils, two fundamental parameters of the model, the average spacing between cracks and the crack connectedness, become specific functions of depth. We constructed these functions for “steady‐state” cracking after long drying times using a number of reasonable physical assumptions and introducing two additional parameters: the maximum crack depth, z m , and the thickness of the intensively cracked layer, z o . With continued drying, the ratio z m / z o ≅ 10 and the limiting values of these parameters are determined. Applications of the model include estimates of crack width, cross‐sectional area, and volume in clay soils, given a knowledge of the water content profile and the shrinkage curve of the clay soil matrix. Published data of four field and lysimeter experiments from different geographical areas were analyzed for validation of the model. A satisfactory correspondence was obtained between predicted and measured crack parameters.