Advancing offshore site characterization practices through intelligent data utilisation

1. Home
2. Publications
3. Online Library
4. Advancing offshore site characterization practices through intelligent data utilisation

Site investigation (SI) campaigns for offshore infrastructure require substantial vessel time and resource, invoking £10s of millions and large carbon footprints. Investigations typically cover large areas, utilise a wide range of testing methods, and involve multiple phases. Despite these complexities, the planning, interpretation, and integration of datasets is still relatively manual. Furthermore, characterisation methods (e.g., parameter correlations) can vary significantly, and there is a general lack of consensus in the choice and suitability of methods for a given site. The resources required to develop rigorous characterisation can be reduced through intelligent SI campaigns and interpretation methods which: leverage existing datasets and prior knowledge, maximise value from low-cost forms of acquisition (e.g., in situ testing), and handle uncertainty rigorously and correctly. This paper presents a new framework for the development and propagation of characterisation methods for both research and industry purposes. This follows the curation of a large database of site characterisation data at over 20 offshore sites, facilitating an initial base of prior knowledge. In particular, automated methods for cleaning, pre-processing, interpretation, and spatial pairing of data (where typically, manual processes are required) are detailed, enabling seamless preparation of reliable datasets for model training. Such procedures allow site characterisation models to be continually refined in a straightforward, standardised manner, while maintaining robustness, transparency, and objectivity. From these prepared datasets, following Bayesian principles, initial probabilistic (prior) models can be built from existing data or updated using site-specific data resulting in tailored (posterior) models.

5th International Symposium on Frontiers in Offshore Geotechnics (ISFOG2025)

5 - Data Analytics and Machine Learning