Feasibility of Using Hindcast Data for Fatigue Assessment of Permanently Moored Offshore Units in West-Africa

To register the wave loads acting on offshore floating structures, wave buoys have been the generally accepted as the most accurate method. However, over the last couple of years, hindcast data has become increasingly accurate. The availability of this data allows assessment of the response of offshore floating structure without the need of local wave buoys.
The accuracy when using hindcast data to assess fatigue accumulation on FPSO hulls is assessed in this paper. For this analysis, calculations have been executed for multiple spread-moored production units in the West-Africa region. The purposely deployed wave buoys provide a full wave spectrum including multiple wave components and directional variation of the wave energy. This data is used to derive statistical data description of the sea state.
In this paper, we consider the use of WaveWatch III, ERA-5 and Copernicus hindcast models. These are all models which provide sea state descriptions on a global scale. The statistical data from the hindcast models is compared against the reference data provided by the wave buoys. Overall, the three models provide decent results in this mild environment. However, it is found that Copernicus provides more accurate estimates for the larger wave heights.
Fatigue analyses have been executed. The full spectral data provided by the wave buoys is used to conduct a spectral fatigue assessment. The hindcast models generally provide more limited details of the sea state, information such as spectral shapes and directional spreading may be missing. Understanding the influence of these missing characteristics is vital for reliable long term fatigue assessment. Multiple fatigue analyses have been executed to examine the sensitivity to these missing characteristics.
It has been shown that the multi-modality of the sea state and spectral shape are the most important parameters driving the deviation between the fatigue assessment based on hindcast data and wave buoy data. These differences can accumulate to a factor of 2 on lifetime consumption. The influence of the accuracy of the statistical parameters provided by the hindcast models and wave spreading is considerably less with a typical contribution of around 30% on lifetime consumption.