Full text loading...
-
Application of Machine Learning to Shallow Site Investigations for Jack-Up Installation
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, NSG 2025: 31st Meeting of Environmental and Engineering Geophysics, Sep 2025, Volume 2025, p.1 - 5
Abstract
Summary
For the safe installation of a Jack-Up rig alongside a wellhead platform on the Ebro continental margin, the acquisition of UUHR (Ultra-Ultra-High Resolution) seismic data along with five boreholes and CPT (Cone Penetration Test) data has enabled the detailed mapping and prediction of the expected soils profile needed to characterise the foundation conditions, where punch-through failures pose significant risks. The integration of geophysical and geotechnical data has been further leveraged with quantitative rock property predictions using seismic inversion and machine learning techniques which has allowed to derive the expected cone resistance (Qc) at a location not directly sampled by the geotechnical campaign, but equally important for foundation design. This work highlights the need for robust geological modelling based on high-quality seismic data as accurate machine-learning-derived Qc predictions mostly rely on well-defined stratigraphic interpretations; these enable the building of low-frequency models which are consistent with the lateral and vertical geological and geotechnical variations observed in the data, while machine learning complements these with the high-frequency components extracted from the inverted seismic trace.
© EAGE Publications BV