Estimating the spatial distribution of hydromechanical properties in the investigated subsoil by defining an Engineering Geological Model (EGM) is crucial in urban planning, geotechnical designing and mining activities. The EGM is always affected by (i) the spatial variability of the measured properties of soils and rocks, (ii) the uncertainties related to measurement and spatial estimation, as well as (iii) the propagated uncertainty related to the analytical formulation of the transformation equation. The latter is highly impactful on the overall uncertainty when design/target variables cannot be measured directly (e.g., in the case of piezocone Cone Penetration Test–CPTu measurements). This paper focuses on assessing the Propagated Uncertainty (PU) when defining 3D EGMs of three CPTu-derived design/target variables: the undrained shear resistance (s_u), the friction angle (φ^′), and the hydraulic conductivity (k). We applied the Sequential Gaussian Co-Simulation method (SGCS) to the measured profiles of tip (q_c) and shaft resistance (f_s), and the pore pressure (u₂), measured through CPTus in a portion of Bologna district (Italy). First, we calculated 1000 realizations of the measured variables using SGCS; then, we used the available transformation equations to obtain the same number of realizations of s_u, φ^′, and k. The results showed that PU is larger when the transformation equation used to obtain the design/target variable is very complex and dependent on more than one input variable, such as in the case of k. Instead, linear (i.e., for s_u) or logarithmic (i.e., for φ^′) transformation functions do not contribute to the overall uncertainty of results considerably.

In the past, soil-layer delineation methods can usually only take a single type of input data, e.g., soil-type data at boreholes. However, this does not fit in the geotechnical engineering practice where multiple types of data are usually available during site investigation (e.g., borehole data and cone penetration test data are both available). This paper adopts a novel data-driven method for soil-layer delineation that accommodates multiple types of site investigation data. The basic idea is to include liquid limit (LL), plasticity index (PI), and fines content (FC) into the soil parameters of analysis. According to the Unified Soil Classification System (USCS), the information of (LL, PI, FC) can be used to determine whether the soil is sand, silt, or clay. As a result, the conditional random field simulation results for (LL, PI, FC) can be used to delineate sand, silt, and clay layers. If extra soil parameters (such as cone penetration test results) are incorporated, the novel method can accommodate multiple types of site investigation data. A real example of the Fucino Basin in Italy is adopted to demonstrate the application of the novel data-driven soil-delineation method.

Using CPTu profiles for subsoil characterisation, transformation equations must be used to obtain the hydro-mechanical properties for structures and infrastructure designing. Additionally, the uncertainty and the spatial variability of measured parameters must be taken into account for a reliable geotechnical design. In this work, we used a Stochastic Simulation approach to define reliable 3D models of two geotechnical designing variables for granular soils (friction angle–ϕ’ and the Darcy permeability coefficient–k) from tip resistance (qc), sleeve friction (fs), and pore pressure (u2) profiles. The selected method – the Sequential Gaussian Co-Simulation (SGCS) – provided reliable optimized 3D models of the spatial distribution of the variables of interest and allowed quantifying the propagation of the estimation uncertainty associated with the raw measurement models through the transformation equations. Overestimation (OE) and Underestimation (UE) percentages for a confidence interval of 68% were calculated throughout the 3D model: granular soils showed a larger uncertainty than fine soils concerning the measured variables (qc, fs, and u2). In granular soils, the measured variable uncertainty varies up to 100% but the derived variables show different behavior: ϕ’ shows UE and OE less than 25% while k reaches 100%. These differences in the propagated uncertainties depend on the transformation equations and the measured variable dependence.

To perform a geotechnical reliable design, the spatial variability and the uncertainties related to the adopted engineering geological model (EGM) must be taken into account. However, any conceived EGM is characterized by uncertainties covering (1) the bias in the mathematical expression that transforms the measured parameters into design ones; and (2) the uncertainty associated with the variability of the soil and rock parameters in the prediction equations. Hereinafter, the sequential Gaussian co-simulation method (SGCS) has been applied to propagate the uncertainty in the calculation of the undrained shear resistance su from measured CPTu profiles (i.e., qc, fs, u2) through a linear model of co-regionalization. The studied area is located in the Po River alluvial plain (Bologna Province, Italy), where the mixture of silts, sands, and clays gets thicknesses of hundreds of meters. These heterogeneous deposits have been mechanically characterized through a 3D EGM to be used in reliability-based designing.