OC3 Phase I
1 Test description
In Phase I, the NREL 5-MW wind turbine is mounted atop a monopile with a fixed support in 20 m water depth. Therefore we do not study the geotechnical models in this benchmark.
Several load cases of increased complexity are compared throughout the benchmark, starting from a stiff model with only wind or only waves to a fully flexible model with wind and waves.
This tests uses the Offshore template as the starting point.
The CSV file to run this test can be downloaded from here.
The model used for this test is shown in the figure below:
This model is the NREL 5-MW reference wind turbine as described in Jonkman et al. (2009k).
The results from the different participants as well as a detailed definition of the load cases and sensors can be found on this link.
A brief summary of the load cases simulated in Ashes is gievn below:
- LC 2.1a: fully stiff model, steady wind, no hydrodynamics, constant RPM
- LC 2.1b: fully stiff model, steady wind, no hydrodynamic, NREL 5-MW controller
- LC 2.4: fully stiff model, no aerodynamics, regular linear waves, parked rotor
- LC 2.6: fully stiff model, no aerodynamics, regular nonlinear waves, parked rotor
- LC 3.1: flexble blades and tower, stiff monopile, steady wind, no hydrodynamics, NREL 5-MW controller
- LC 4.1: flexble tower and monopile, stiff blades, no aerodynamics, regular linear waves, parked blades
- LC 4.3: flexble tower and monopile, stiff blades, no aerodynamics, regular nonlinear waves, parked blades
- LC 5.1: fully flexible model, steady wind, regular linear waves, NREL 5-MW controller
The sensors used for this benchmark are
- the Generator sensor
- the Rotor sensor
- a Beam element sensor at the mudline, half way from the mudline to the MSL, at the MSL and at the tower top, measuring the response loads in the elements
- a Beam element sensor at half the span of Blade 1
- the Blade [Time] sensor
- a Node sensor at the transition piece and the tower top, measuring the displacements of the nodes
- the Sea sensor, measuring the wave elevation
Since this is a code to code comparison, there is by definition no correct result, the objective is to match as closely as possible the other participants of the project. However, the spread among participant is very wide and there are several outliers, which makes it hard to define the values that Ashes should be aiming for. Furthermore, it possible that although two sets of results are very similar, a phase difference in the tower shadow make them take very different values at a given instant.
For those reasons, there is no automatic fail/pass criteria on these benchmarks, only a visual inspection is performed. This report can be downloaded in the following link:
Note that since the Stream function wave is not currently available in batch simulations, the load cases with stream function wave kinematics (i.e. LC 2.6 and LC 4.3) are run separately. Their reports can be found in the following links:
Note that only a visual check is performed here. Once we have a set of results that are in agreement with the other participants, we use that set as a base case to compare the results produced by Ashes daily. If these results differ from more than 0.01% with the base, a failed test notification is raise. The report can be found on the following link:
5 Additional notes on the OC3 benchmark
- The transients from the results of the OC3 benchmarks have been removed, but we have not removed them in Ashes. It is thus expected that only the last 30 sec or so of the simulation should match the results from the other participants
- The results from the participants Vestas_FLEX5 were removed from the benchmark as they were too different than the rest of the participants to fit in the same graph
- In several load cases, some degrees of freedom should be deactivated. In Ashes, this is done by having an arbitrarily high stiffness. This implies that although displacements and deflections will be very small, they can be non-zero.
- It appears that no buoyancy loads were applied during this benchmark (because the axial force in the monopile is constant), we have therefore deactivated buoyancy loads in Ashes.
- There is a noticeable difference between Ashes and some other participants in the blade root torque for the cases where the rotor is not rotating. This seems to be due to differences in the position of the center of mass of the airfoils between the original definition of the NREL 5-MW blade (defined in Jonkman et al. (2009k) and which was presumably used in the OC3 benchmark) and the more recent BeamDyn definitions, which we have used to create our blade in Ashes (see NREL 5-MW). Future work includes implementing the old blade to see whether we obtain a better match. However, it should be noted that the spread among participants is so wide that it is not trivial to decide what the Ashes results should look like to improve our match.