AI-Powered Pipeline Freespan Detection
Daniel Mercado and Pedro Santos of NEO NEXT +ENERGY take a highly practical look at how AI can transform subsea pipeline assessment, moving freespan detection from time-consuming manual review to automated risk flagging. Their presentation connects engineering integrity challenges with advanced data science, showing how deep learning can help identify pipeline freespans at scale using 3D bathymetry data.
The session begins with the engineering fundamentals behind pipeline freespans, where sections of subsea pipeline lose contact with the seabed and become exposed to hydrodynamic loading. Daniel explains how vortex-induced vibration, cyclic loading and mechanical fatigue can create serious integrity risks, making accurate detection, assessment and mitigation essential for safe pipeline operations.
Pedro then explores the AI modelling approach, using ungridded 3D point cloud data from multi-beam sonar surveys. The presentation outlines how a blended architecture combines PointNet++ deep learning with physics-informed manual rules, improving detection performance while keeping outputs practical for production use.
A key value of the approach is its ability to automate year-on-year survey comparison, track span movement, support lateral displacement detection and reduce dependence on subjective categorisation. What once required days of engineering review can be processed far more quickly, with consistent model outputs and clearer visibility of pipeline risk evolution.
The full presentation shows how NEO NEXT +ENERGY is applying AI, high-performance computing and interactive visualisation to make subsea pipeline integrity assessment faster, more consistent and better aligned with real operational decision-making.
