Ground model development – Geostatistical and Machine Learning prediction of soil properties
Installation of new offshore wind park requires a good knowledge of the shallow subsurface over a large area. Integration of sparse geotechnical data and high-resolution geophysical data is a necessity to build a consistent ground model that can be used for the design of the wind turbine foundations.
Even if the oil and gas industry has been working with data integration and geological model building for years, integration of high-resolution shallow subsurface data remains challenging.
In this project, we will build a ground model for the Holland Kust Zuid offshore wind farm using the data (geotechnical and geophysical) provided by the Dutch authorities. The data sets are in 1D and 2D and new tools to populate these findings into a 3D need to be developed.
To do this, geo-statistics and machine learning will be used to populate the ground model with soil properties while tracking the uncertainties. This will improve the input into engineering and conceptual design of the wind turbine foundation and reduce the production cost at the same by optimizing data usage.
The goal of the project will be to re-do the already existing geophysical and geotechnical interpretation to ensure an integrated interpretation. Secondly, to test different models to populate the ground model with geotechnical data and finally evaluate the accuracy of the models performing cross-validation.