m: 07741 003180

The Challenge in Deriving Extreme Values and Final Design Criteria

When you see a table of say 50 or 100 year Metocean design criteria have you given much thought to how they were derived and perhaps more importantly whether they are reliable.

Before finalising design criteria it is necessary to carry out what is known as extreme value analysis (EVA), a technique that enables you to take a dataset, plot it in terms of probability of exceedance and extrapolate it to various return periods from say 1 to 100 years by fitting a line through the data.

Example of EVA Graphical & Tabular Output

The plot above (courtesy of MetoceanWorks) is an example of this and shows a 100 year value for Hs of 7.49m. But if you change the amount of data points used, and/or focus on different parts of the data the best fit line will change and a different set of return period values will be produced which could be higher or lower than 7.49m.

In fact there are 3 main factors influencing the variability of extreme values: what input data is used; how the EVA is carried out; and what contractor is doing the work. This means that if you require Metocean design criteria for a location it is important to understand what data is to be used, how it will be analysed and maybe most importantly who is doing the work.

All of this means that it is vital to carry out a range of EVA runs – reliance on 1 run could be dangerous as the resultant value may be an over or under-estimate..

The 50 or 100 year extreme values generated from the different EVA runs pose a key challenge: how do you go from a number of different extreme values to the final design value? I will discuss this in my next blog.

Share this post