Full text loading...
This abstract presents a method for automated, stress-normalized stratigraphic classification of cone penetration test (CPT) data. The approach addresses key limitations in current practice, including manual interpretation, inconsistent layering, and lack of integration with geotechnical design models. The method uses friction ratio and stress-normalized cone resistance (referenced at 100 kPa effective stress) as primary classification parameters, defined through an engineer-controlled configuration file. Stress normalization is performed using a formulation inspired by the Hardening Soil model, allowing direct linkage between CPT-derived materials and stiffness-moduli used in numerical design. The method was applied to six CPTs from the Wilhelmshaven site in northern Germany, representing varied subsurface conditions. Each CPT was processed through a two-pass normalization and classification loop, with a minimum layer thickness constraint of 0.3m to reduce interpretive noise. Outputs include model materials with geotechnical properties and layer boundaries, suitable for integration into 2D/3D modelling workflows. Results demonstrated consistency across CPTs, preserved thin but interpretable layers, and highlighted site variability via 3D visualization. The method enables transparent, reproducible ground modelling from raw CPT data and improves the connection between site characterization and design-stage geotechnical modelling. Future extensions will focus on uncertainty quantification and full 3D model generation.