Geological origin as an input variable in reliability-based designs: for an accurate exploration in geotechnical engineering

Soil is one of the most difficult materials to characterize realistically, which partly explains the uncertainty between the designs and the geostructures real behavior. Different recommendations have arisen with respect to carrying out field investigations in order to reduce the uncertainties inherent to the soil. However, the field exploration and the implementation of sophisticated geotechnical models have proven to be insufficient to mitigate the geotechnical uncertainty. Therefore, Reliability-Based Designs (RBD) emerge as a decision-making tool through the definition of the probability of failure in conjunction with the typical Factors of Safety. RBD requires a previous understanding of the most appropriate soil probabilistic models, such as the Shear Strength Varying with Depth (SSVD) analysis, traditional Monte Carlo simulations or random fields. Soil shear strength uncertainty is related to soil geological characteristics, however, geology has been commonly used in geotechnical engineering as a definition of the layers’ distribution on the soil mass, where the definition of the accurate RBD models according to the geological origin has been missing. Therefore, two geological formations were analyzed: residual soils (stationary origin) and mudflows (dynamic origin). The results show that random fields are more related to the mudflows due to the random nature of these soils, thus the exploration resources should be focused on the determination of the Probability Density Functions (PDF) and the spatial variability of the shear strength (SS) properties (laboratory tests have priority over the in situ tests). Residual soils present a higher SS space uniformity because these soils have not been previously mobilized, thus the exploration resources should be focused on the determination of the SSVD (field tests have priority over the laboratory tests). Therefore, defining the geological origin as an “input variable” will allow recognizing the most important variables and the definition of the best soil exploration for an accurate and cost-effective RBD in geotechnical engineering.