Recommended workflow for coupled Black Sea wind–wave–current simulations in OpenFAST and WEIS

[Petrov.Mihov]

Hello everyone,

I am developing an OpenFAST workflow for site-specific offshore wind simulations in the Black Sea and would appreciate guidance from the community on the correct treatment of coupled wind, wave and current environmental data.

Current project status

I am using OpenFAST v4.2.1 with ROSCO 2.10.1 on macOS ARM64.

My reference-turbine validation work currently includes:

• IEA 15 MW fixed-bottom reference turbine;
• IEA 15 MW floating reference turbine;
• IEA 22 MW fixed-bottom reference turbine;
• IEA 22 MW floating reference turbine, for which the staged validation is currently being completed.

The validation workflow progresses from smoke tests and isolated-physics cases to still-water, aerodynamic, controller, wave and eventually current-loaded simulations.

My longer-term objective is to use WEIS to investigate possible improvements or adaptations of the reference turbine designs and then evaluate their behaviour under representative Black Sea environmental conditions.

Available Black Sea environmental data

I have collected approximately five years of reanalysis data for three preselected Black Sea areas of interest.

The available datasets contain:

• wind time series at several heights;
• significant wave height;
• wave period;
• wave direction;
• current velocity and direction at several depths;
• separate spatial points within each selected area.

The source data have different temporal resolutions. The wind data are approximately three-hourly, the wave data are approximately hourly, and the current data are approximately daily.

For every area, I intend to retain the individual point time series while also calculating an area-representative mean using only the points located inside the corresponding area boundary.

My main challenge is now to convert these datasets into physically consistent and technically defensible environmental inputs for OpenFAST.

Main objective

I do not want to select wind, wave and current conditions independently and then combine unrelated extremes.

The objective is to preserve, as far as the source data permit, the observed relationships between:

• wind speed and direction;
• significant wave height and peak period;
• wave direction;
• current velocity and direction;
• seasonal conditions;
• wind–wave misalignment;
• wind–current and wave–current misalignment.

The final workflow should support both fixed-bottom and floating wind-turbine simulations.

For floating turbines, it should also support investigation of platform motion, mooring loads, controller response and coupled aerodynamic–hydrodynamic behaviour.

Questions for the OpenFAST and WEIS community

I would be grateful for recommendations on the following points.

1. Construction of coupled environmental states

What is the recommended method for constructing representative wind–wave–current states from multi-year time-series data?

Would it be preferable to:

• align all datasets to a common timeline;
• create joint probability distributions;
• classify the observations into environmental clusters;
• select representative medoids or nearest real observations;
• retain complete historical events rather than constructing synthetic combinations?

I would particularly like to avoid generating environmental combinations that are statistically possible but were never physically observed.

2. Different temporal resolutions

What is the most defensible method for combining three-hourly wind data, hourly wave data and daily current data?

For example, should the slower variables be held constant over their native interval, or should all variables be resampled to a common interval?

I am concerned that interpolation, particularly of direction and daily current data, could create artificial short-term variability.

3. Spatial aggregation

For each area of interest, I have several data points located inside a defined geographical boundary.

Is an area mean suitable for generating representative OpenFAST conditions, provided that the original point-level time series are retained?

Would it be better to:

• simulate the area mean;
• simulate a representative point;
• simulate several spatial points;
• use the area mean for general cases and individual points for sensitivity or extreme cases?

For directional variables, I am using circular averaging rather than ordinary arithmetic averaging.

4. Direction conventions and environmental misalignment

What is the recommended convention for transferring wind, wave and current directions into OpenFAST?

In particular, I would like to ensure consistent treatment of:

• direction “from” versus direction “toward”;
• meteorological and oceanographic conventions;
• global headings;
• nacelle yaw;
• wave propagation direction;
• current direction;
• relative wind–wave–current misalignment.

Would it be better to rotate every environmental state into a turbine-relative coordinate system before creating the OpenFAST cases?

5. Current profiles

The current data are available at several depths.

What is the recommended approach for creating an OpenFAST current profile from discrete reanalysis depths?

Should I use:

• direct interpolation between measured or reanalysis depths;
• a power-law profile;
• a logarithmic profile;
• a depth-uniform current;
• a user-defined profile based directly on the available levels?

How should below-seabed levels, missing depth levels and differences between model water depth and site water depth be treated?

6. Fixed-bottom versus floating workflows

Should the same environmental-state matrix be used for both fixed-bottom and floating reference turbines?

My current assumption is that the environmental states should remain the same, while the model configuration and relevant acceptance checks differ.

For example:

• fixed-bottom cases would focus on structural loads, tower response and foundation-related outputs;
• floating cases would additionally require platform six-degree-of-freedom motion, mooring tension, offset, stability and controller–platform interaction checks.

Is this the recommended approach, or should the floating environmental matrix contain additional cases specifically selected around platform natural periods and low-frequency excitation?

7. Representative operating and extreme cases

How should five years of environmental observations be reduced into a practical simulation matrix?

I am considering separating the cases into:

• representative operational states;
• below-rated, near-rated and above-rated wind conditions;
• common seasonal states;
• high-wave states;
• high-current states;
• strong wind–wave misalignment states;
• strong wave–current misalignment states;
• combined high-load events;
• shutdown or parked conditions;
• selected historical storms.

Would this be an appropriate engineering screening structure before developing formal IEC design-load cases?

8. Duration and transient removal

For environmental states based on real data, what simulation duration would be recommended for initial screening and later production simulations?

My current staged approach is:

• very short smoke tests;
• 10–30 s numerical and configuration checks;
• 120 s intermediate validation cases;
• longer production simulations after the configuration is accepted.

For irregular waves, turbulent wind and current-loaded floating cases, what minimum duration and transient-removal period would normally be considered sufficient for meaningful statistical comparison?

9. Relationship between OpenFAST and WEIS

Would it be better to complete the full Black Sea environmental case-definition workflow in OpenFAST first and only then introduce WEIS optimisation?

My preliminary plan is:

1. validate the reference OpenFAST models;
2. prepare and quality-control the Black Sea environmental datasets;
3. define the coupled environmental-state matrix;
4. execute baseline Black Sea simulations using the unmodified reference turbines;
5. identify the dominant load and response limitations;
6. use WEIS for targeted design adaptation or optimisation;
7. rerun the same environmental matrix with the modified design;
8. compare the baseline and optimised configurations.

Is this a reasonable sequence?

Intended outcome

The intended result is an auditable workflow that connects:

raw five-year environmental data → quality-controlled area time series → coupled environmental states → OpenFAST input cases → fixed-bottom and floating simulations → WEIS design investigations.

I would greatly appreciate references to existing OpenFAST, WEIS, IEA Wind or NREL examples that demonstrate a similar site-specific wind–wave–current workflow.

Thank you in advance for any recommendations, particularly regarding environmental-state construction, temporal alignment, current-profile definition and the separation between fixed-bottom and floating simulation matrices.

[Jason.Jonkman]

Dear @Petrov.Mihov,

I would generally recommend you review and follow the requirements and guidance for performing loads analysis of wind turbines based on IEC design standards, e.g., IEC 61400-3-1 for fixed-bottom offshore wind turbines and IEC 61400-3-2 for floating offshore wind turbines. While these are not publicly available (they require purchasing through IEC), they answer many of your questions. Here is my quick answer to your specific questions:

1. IEC requires that a joint probability of wind and wave conditions be created, based on a 1-hour reference period, with the data binned in terms of hub-height wind speed, significant wave height, peak-spectral wave period (and directionality, when available). Different load-case simulations will then sample from this distribution.
2. It is often assumed that the wind and wave data are independent of averaging period for normal conditions; for N-year return periods, correction to averaging period is needed, with guidance provided in IEC.
3. The IEC requires site-specific data, but clustering of turbines with similar wind/wave conditions, depth, etc. may be justified.
4. OpenFAST does not use compass directions, but rather assumes the global X axis is aligned with the zero-degree wind and wave directions. I would say it is common to define the joint probably distribution of wind and wave conditions in terms of compass directions, than adopt this data to the OpenFAST convention when setting up design load case simulations.
5. I’m not expert on site assessments, but my understanding is that it is common to fit to the current profiles defined in IEC.
6. The wind and wave site assessment should be nearly identical, but the load case simulation results differ. See the IEC standards for details.
7. I would generally follow IEC, unless there are project-specific conditions that justify a change.
8. The IEC states simulation requirements for load case simulations. You should remove simulation start-up transients before post-processing simulation results. The length of the simulation start-up transients can be reduced by setting appropriate initial conditions–as has been discussed in other forum topics.
9. Sounds reasonable.

Best regards,