Quantitative analysis approach to assess variability in seabed conditions across a large offshore windfarm site

Offshore windfarms are growing in size and are being constructed across areas with significant variability of seabed sediments. Traditional in-situ site investigation protocols such as the cone penetrometer test (CPT) are time consuming, and thus costly, and provide data at a single discrete location with interpolation necessary to estimate soil properties between testing points. Emerging techniques utilising Artificial Neural Networks (ANN) allow for the generation of a 'synthetic' CPT profile at any location based on geophysical survey data, but ANN-based approaches are only as good as their inputs. It is necessary to ensure the limited number of 'real' CPT profiles are representative and provide maximum coverage of the problem space. This paper demonstrates how time series analysis techniques can be applied to CPT 'depth' series data to quantitatively determine similarity of CPT profiles across a site. The method could inform future geotechnical site investigation planning, enabling efficient and effective layout of test locations to calibrate an ANN for derivation of synthetic CPTs anywhere across the site. The method also has potential downstream applications in clustering design solutions for areas with similar geotechnical properties. To demonstrate this methodology, the Burbo Bank Extension windfarm site investigation dataset (provided by the Crown Estate) has been used.