This study examines the behavior of subsurface fractures in fine-grained soil such as clays in response to changing environmental conditions. Fractures serve as conduits for moisture transfer, which can lead to substantial shrinking and swelling of the surrounding fracture boundary soils. These volume changes, in turn, affect fracture geometry and moisture transmission rates. A new predictive model, termed the 'Fracture Volume Change Model' (FVC Model), has been developed to relate moisture transfer, soil volume change and associated changes in fracture aperture. The model assumes a discrete horizontal fracture in a laterally-infinite, saturated, expansive clay with rigid, outer no-flow boundaries and an inner flexible yielding boundary along the fracture. The FVC Model is based on the one-dimensional diffusion equation, which is solved analytically for both constant moisture and constant flux fracture boundary conditions. Changes in fracture aperture are predicted assuming normal shrinkage and either isotropic or anisotropic volume change. The model is expandable to bulk scale analysis of geologic formations with multiple stacked fractures.

The model was validated and calibrated in the laboratory using a custom fabricated horizontal infiltrometer device. Tests were conducted on a problematic clay soil from Fairfax County, Virginia, belonging to the southern montmorillonite facies of the Potomac Formation. Moisture content was varied from 17% to 33% by forcing air through an artificially created discrete fracture. Moisture changes in the fracture boundary soils caused the effective fracture aperture to fluctuate from near closure to 0.031 in. (0.79 mm). Upon application of excess moisture, it was riot possible to effect full closure of the fracture. Moisture values predicted with the FVC Model demonstrated good agreement with the laboratory data, deviating 6% on average. Predictions of fracture aperture were generally overestimated. The model confirmed the dominance of internal hydraulic properties of the soil matrix over evaporation or infiltration mechanisms. The model was also used to predict soil desiccation rates for an environmental remediation project in expansive clay in Santa Clara, California. Model application to agriculture, geotechnical engineering, and resource geology is also described.