An Innovative Regional Approach to the Analysis of Hurricane Impact on Offshore Platforms

Abstract

Hurricanes are one of the biggest threats to offshore platform integrity and personnel security in the Gulf of Mexico. In hurricane-dominated regions, history of extreme events at a determined location, like an offshore platform, are often too scarce for a direct Extreme Value Analysis (EVA). Extreme Value Analysis being essential for structural design, it is necessary to explore new methodologies. The spatial method presented in this article is based on the Regional Frequency Analysis (RFA), adapted to hurricane-generated sea states. The goal is to increase the size of the sample of extreme events considered for a given location, to ensure a more reliable view of the distribution of very extreme events. For a concrete application, a hindcast dataset of significant wave height over the Gulf of Mexico is decomposed into homogeneous regions where distinct locations show similar hurricane impacts. The identification of these homogeneous regions is performed by applying the Ward hierarchical clustering method on peaks of significant wave heights generated by hurricanes. Five homogeneous regions are identified and described. It is then possible to trade space for time: we thus build samples of extreme events for each of the homogeneous regions, hence gathering a larger number of events than if we had only considered hurricanes which traveled above one given location. A parametric distribution is then used to describe the distribution of extreme events inside each homogeneous region. Extreme value analysis at a given location is then computed by transposing the corresponding regional distribution of extreme events. This methodology allows to reduce significantly the confidence interval width around the significant wave height extreme values estimate, especially for the 100 and 1000-year return values. The methodology presented is quite an innovation as it had never been applied before to a dataset of sea states resulting from hurricane activity. The results are very valuable to obtain better estimate of potential extreme wave height due to hurricanes, hence increasing the confidence in offshore platforms structural design and the overall security. Moreover, this methodology can be applied to other regions of the world impacted by very extreme events, and to other kind of spatial data, such as winds, currents, or storm surge.