Mooring Line Failure Detection Using Machine Learning

Abstract

Offshore floating vessel mooring failure and subsequent loss of station can have catastrophic consequences for the vessel and the associated subsea infrastructure. Therefore, integrity management and timely detection of mooring failure is critical. Traditional methods of failure detection rely on line tension measurements and watch circle approaches. Both these approaches have limitations and are not reliable. Alternate methods of detecting line failure are therefore required. This paper discusses a novel approach of using measured vessel positions and 6-degrees-of-freedom accelerations along with a deep machine learning algorithm to detect mooring line failure in near real time. Results from a numerical case study for a turret moored FPSO with over 4000 test cases demonstrate that this approach can accurately identify failed mooring line cases over 99% of the time.