A hybrid FEM-Kriging approach for fatigue assessment of a steel pipe riser from field-measured motions

Resumo

The establishment of monitoring systems for Floating Production Units motions provides an important input needed for real-time assessment of riser fatigue through Finite Element Method analysis. Standard practice relies on these simulations to obtain the loads required to calculate fatigue life along the riser, however, it is a CPU-intensive approach which makes it difficult to achieve results in real-time. Since interest often lies on few critical points on the riser, such as 1st weld and TDP, the number of outputs of is vastly reduced, making surrogate models an interesting and cost-effective alternative. Previous work such as DAMASCENO (2020) has shown Kriging to be a good choice for time series prediction, yielding adherent results for outputs such as forces and moments by having vessel motions as inputs. This method requires a short FEM simulation (~600 s) to train the models that will expand the output time-series to full 3,600 s duration, which will be shown to be far less CPU intensive. A case study for a SCR connected to a semi-submersible platform is presented, evaluating fatigue life from 5715 hours worth of data from the year of 2018 through both methods. Analysis of the obtained results yields additional tools which correlate damage to vessel motions, which are then used to prioritize sea states and evaluate the most cost-effective FEM and Kriging combination.