Triangulated density models for detecting brittle materials with CPTu

Highly sensitive marine clays, referred to as quick clays, are characterized by extreme strain-softening behavior that can result in near total strength loss from small disturbances. This makes them a potential hazard for large retrogressive or progressive landslides. Identifying such sensitive soils is essential for assessing geotechnical risks in marine clay environments. In current Norwegian practice, identification of sensitive clays relies on laboratory tests where sample extraction and testing are both time-consuming and costly. A field investigation method is desired to efficiently screen and delineate zones for sensitive materials. For this the cone penetration test with pore pressure measurements (CPTu) is a strong candidate among available methods. In this work, data from 38 road projects across Norway is used, where laboratory test results have been linked to 277 CPTu tests. The data are divided into Brittle and Non-brittle materials based on their remolded undrained shear strength. Kernel density estimation (KDE) method is used to create density distribution models for these groups in three dimensions. Although the KDE models can be used directly to screen for sensitive materials, boundaries for selected density ratios between both models are triangulated to generate a visual representation of the resulting volumes. A method to classify CPTu data using triangulated models is presented, and examples of classification result are given.