Uncertainty Propagation Assessment in CPTu-Based Lithological Modeling Using Stochastic Co-Simulation

Reliability-based designing in geotechnical engineering represents a modern approach to estimating the probability of failure due to spatial variability of soil and rock properties as well as the errors and uncertainties related to measured parameters and calculated designing variables. As spatial data are often used as inputs in models of spatially distributed variables, such as CPTu-based litho-mechanical models, the propagation of spatial uncertainty in model predictions becomes a critical issue even in engineering design. This is commonly known as model uncertainty and several different methods to handling it has been proposed in the literature. This paper proposes an efficient geostatistical simulation approach, based on the use of the Sequential Gaussian Co-Simulation, for uncertainty propagation assessment in the calculation of a Soil Behavior Index from raw CPTu profiles (i.e., q_c, f_s, u₂). These measurements have been collected in a portion of the Bologna district (Italy) by the Emilia Romagna Region.

A Linear Model of Coregionalization was fitted to the matrix of experimental direct and cross-variograms of the input data, and one thousand realizations that honor the experimental data were provided. These realizations were used to calculate the corresponding values of the litho-mechanical subsoil index. As a result, the uncertainty estimates of subsoil lithology can be efficiently used within geotechnical designing of several structures so avoiding assumptions on probability distribution of natural materials that may in turn show site-specific hydro-mechanical behavior, which needs to be investigated in detail.

Keywords: Uncertainty propagation, 3D litho-mechanical model, CPTu profiles, Sequential Gaussian Co-Simulation