Uncertainty assessment of effective friction angle of non-cohesive materials combining data from cone penetration and shear tests

The effective shear strength is a critical parameter for evaluating the ultimate and serviceability limit state of geotechnical structures. To conduct a fully probabilistic assessment or to determine characteristic values according to the second generation of Eurocodes, it is essential to quantify the uncertainty of ground properties that arises due to inherent variability, measurement error, transformation and statistical uncertainty. However, unlike other ground properties, even in the laboratory, the shear strength parameters are not directly measured but are derived from the relationship between shear and normal stresses, making the uncertainty analysis non-trivial. This study applies two different regression approaches and the extended multivariate approach (EMA) to estimate the effective friction angle for non-cohesive soils. Firstly, an Ordinary Least Squares (OLS) and a Bayesian Linear Regression (BLR) approach are utilized to quantify the uncertainties inherent to data from direct shear and triaxial tests from an offshore wind project. Secondly, the EMA is utilized to integrate CPT and shear test data via Bayesian inference. The results are discussed based on characteristic values according to Eurocode 7 (EN 1997-1:2024) highlighting the importance of an accurate and precise mean and uncertainty estimation.