In recent years, there has been several cases of mooring line failures on various floating offshore units. In several of those cases, the failures were identified months or years after they initially occurred. Most assets being designed for single line failure only, this means that the risk of a catastrophic failure of the whole mooring system is quite high if the failure of a single line cannot be detected reliably. To mitigate that risk, class societies such as DNV GL have introduced requirements for the use of lines tensions monitoring systems as part of the mooring class notations (POSMOOR). However, most of the line tension monitoring systems available on the market today have proven unable to remain functional after more than 2 years in operation, due to the harsh conditions and loads they are exposed to. For that reason, an alternative system for line failure detection is needed.
In this paper, a system is developed to detect reliably a single line failure based solely on GPS and motions sensors data installed on the asset. The GPS and motion time series are used to train a neural network which can then reproduce any motion signal as a function of the others, capturing all the complex nonlinear correlations between the wave frequency and drift motions of the asset along its 6 degrees of freedom. Any change in the mooring system properties such as a line break has an impact on those correlations, this change is captured by the neural network, therefore enabling it to detect a line failure. The accuracy of the system is demonstrated using numerical simulations for an FPSO in various sea states, where a line break occurs at one time instant during the simulation.
A moored floater is a 6-DOFs system where the loads are stochastic environmental loads, wind waves and current. The behavior of the system in a given sea state is governed by its inertia, damping and stiffness properties. The failure of one or several mooring lines has an obvious effect on the stiffness, but also on the damping of the whole system since the drag on the lines is one of the damping contributions. The floater displacement in the same sea state will therefore be different for an intact or damaged system. Those differences might not be easily detectable since depending on the mooring system configuration, and the sea state, a line failure might not change significantly the basic statistics of the motions such as mean and standard deviation. In those cases, changes in the higher order statistics, as well as in the correlations between different degrees of freedom motions need to be detected to identify a line failure.
