Spatial uncertainty quantification of desiccation cracks in clays with limit state-adjusted linear elasticity

Periodic cycles of flood and drought are aggravated by global warming, thereby inducing critical desiccation cracks in soils. However, the estimation and prevention of crack propagation is imperfect due to the imprecise field measurements and the underlying soil uncertainties. To address these issues, this study presents an alternative framework to assess the probability of crack propagation. Thus, a series of Monte Carlo sampling (MCS) and Gaussian random fields are employed to quantify the associated uncertainties. The results suggest that matric suction has a governing effect in the course of crack propagation and induces cracking particularly at shallow depths. Crack propagation tends to decrease with the increase of tensile strength and reduction in matric suction. Normal probability distribution is established as the distribution carrying the least uncertainty, followed by the Weibull distribution. Furthermore, the probability of crack propagation is directly proportional to the reduction of the soil compaction density and the variations of matric suction. Also, random field sampling is superior to MCS due to the application of several correlation length scales between the simulated samples. Consequently, the application of wider length scales leads to lesser variance and similar estimations to the field observation of crack. These new findings suggest that measuring crack propagation, considering the spatial uncertainty, yields dependable estimations with only 8% deviation from the field observation. The developed probabilistic framework also provides a promising alternative to the deterministic risk quantification analyses of crack propagation without recourse demanding experiments, and complex simulations, and hence it can be a feasible option for a reliable design.