Get Started with ESOX

ESOX uses the ERA5 reanalysis data which is produced by the European Center for Medium-Range Weather Forecasts (ECMWF) on behalf of the European Unions Copernicus Climate Change Service (C3S). ERA5 is ECMWF/C3S’s latest reanalysis product, providing a state-of-the-art, global data set on atmosphere, land surface and ocean waves covering the time period from 1979 until a few days from the current date, with a one-hourly time resolution. 

Each location in the ESOX map is the centroid of an area with a width of 0.25 x 0.25 degrees and the data at each point is representative of this grid box area, resulting in a spatial resolution of approximately 10 to 30 km, depending on latitude.

ESOX includes data out to 200+ km from shore, from 70 degrees North to 60 degrees South – in total more than 150.000 data points. Each data point covers the period from 01-01-1990 to 31-12-2019 (30 years) and includes 4 metocean parameters:

• Mean wind speed at 10 metre height calculated with this formula using the parameters
  10 metre U wind component and
  10 metre V wind component
• Mean wind speed at 10 metre height calculated with this formula using the parameters
  100 metre U wind component and
  100 metre V wind component
• Significant wave height
• Peak wave period

The full ERA5 documentation is available here.

We have benchmarked the ERA5 meteorological and oceanic data against both measurements and site-specific reanalysis products developed by industry leading companies for multiple of the offshore wind project developers at several locations in the North Sea, the US Atlantic, the US Pacific, Japan, South Korea, Taiwan and Australia. Overall, ERA5 has shown excellent results, outperforming many reanalysis data sets by combining an advanced data assimilation system with an increased spatial and temporal resolution. However, we see limitations in specific areas, such as areas with tropical climate (typhoons) and locations with sudden bathymetry changes which the spatial resolution of ERA5 cannot capture.

The land-sea mask is the proportion of land, as opposed to ocean or inland waters (lakes, reservoirs, rivers and coastal waters), in a grid box. This parameter has values ranging between zero and one and is dimensionless.

Each location in the ESOX map is the centroid of an area with a width and height of 0.25 x 0.25 degrees. The land-sea mask value represents  the proportion of land over that area.

See full land-sea mask description here and a practical land-sea mask example here.

Some onshore locations hold wave values that are obviously non-representative of the local conditions.

The wave values for a location are generally representative if the land-sea mask parameter for the location is below 0.5 – and the reliability of the wave data increases as the land-sea parameter approaches zero. A grid box (with the centre of the box marked by a dot on the map) is considered to be land if more than 50% of it is land, otherwise it’s considered to be water (ocean or inland water, e.g. rivers, lakes, etc.). 

The inland points with high land-sea mask parameters are included in the ESOX map in order to provide meterological data for analysis of activities in port. 

Periods with negative wave values (wave height and/or wave period) represent that sea-ice has formed and that no wave values are available.

ESOX includes 4 examples that can be used as a starting point:

• Monopile/Transition piece installation using a jackup vessel
• Monopile/Transition piece installation using a floating vessel
• Wind Turbine Generator erection using a jackup vessel
• Single weather window analysis

The VBA macro reads the campaign assumptions defined by the user and generates the full list of activities to be executed during the full installation campaign. The user can modify the full list of activities before running the analysis – or alternatively define the full list of activities directly.

Starting from the selected start date, ESOX will identify the first point of time where the time series weather data meets the requirements for the first activity to be executed (e.g. 3 hour weather window of significant wave height below 2 meter and 10 minute average wind speed below 15 m/s). The period of time elapsed while waiting for the required weather window is recorded as weather downtime (WDT).

If the weather window required is shorter than the base duration of the activity, the activity will be executed during one or more weather windows while pausing between periods of WDT. At the point of time where the first activity is completed, ESOX starts looking for the weather window for the next activity. This continues until all activities have been executed.

ESOX will analyze the recorded milestones and WDT during the simulation and present the results with a confidence percentile defined by the user (e.g. P50, P75, P90).

Monthly statistics can be deceiving when looking at a campaign that takes more than one month to execute. Unless you are looking at average or P50 values, the monthly weather downtime percentiles may not add up: the P80 downtime of January + the P80 downtime of February + the P80 downtime of March is normally very different from the P80 downtime of the period of January to March.

As an example let’s consider a precipitation data time series covering the last 10 years. When looking at each calendar month, you have 10 sample years from which you can identify the 3rd worst January (similar to a P80 January). And you can do this for all 12 calendar months. But if you look at the combined period from January to March (Q1) then most likely the 3rd worst January did not happen the same year as the 3rd worst February. And most likely the 3rd worst March happened in yet another year. In reality, we don’t see long periods of consistently very good weather or periods of consistently very bad weather. Generally, fluctuations evens out when looking at longer periods. So if you want to know the 3rd worst Q1 period from the 10 year dataset, then you need to compare the ten Q1 periods and select the 3rd worst year.

Monthly weather calendars constructed for non-average scenarios with a duration of more than one calendar month are calculated by first calculating the PX duration of the total campaign with the selected start date – and then using this total duration to scale the average monthly downtime percentages up or down to fit the total PX campaign duration. This means that the monthly weather calendars for such cases are only valid for the selected start date – hence the start date and other assumptions cannot be moved around in the scheduling software without re-calculating the corresponding weather calendar.