Preprocessing Hydrodynamic Properties for BModesJJ / OpenFAST of a floating substructure

[Daniel.Walia]

Dear Jason,

thank you. It took me a while to come back to your suggestions and answers, as I had some problems to acess the results from our decay tests in the wave tank, in order to answer your questions. And also to dig deeper and understand the relations between the different matrices and their meanings for HydroDyn/BModes/AQWA/WAMIT. Because it got me confused again. Even though you have it explained quite often in this forum.

@Jason.Jonkman

• Are the frequencies output by BModes what you’d expect them to be? If not, this could point to a possible bug in the input file settings.

When having our 1:50 physical model tested, we have carried out also dedicated decay tests to find out the natural frequencies of the substructure. However, from my understanding the test setup was not suitable to account for the natural frequencies of the tower. We have used an optical motion tracking system and attached reflective bulbs to the substructure. Therefore, only the motions of the substructure have been tracked. Unfortunately we did not monitor the tower or RNA motions themselves. I think this setup does not provide sufficient information to process the towers natural frequency, as the resolution of the optical tracking system is not high enough to track the low amplitude, higher frequency mode shapes of the tower which might be also hidden in the motions of the substructure.
However, the natural frequency of the substructure was found to be at about ~0.021 to 0.025 Hz for Surge as well as for Sway (for the full scale floater as simulated in BModes/FAST).

@Jason.Jonkman

• I see that you have the (4,4) and (5,5) elements of the hydrostatic stiffness matrix (hydro_K) as negative-valued. This makes sense for the FAST / HydroDyn input, because gravitational restoring is intrinsically accounted for in the ElastoDyn module. But gravitational restoring is not intrinsically accounted for in BModes, so, I would guess hydro_K (4,4) and (5,5) should be positive-valued in BModes (perhaps this is a bug in your BModes input?).

Here is where I got confused a little. In the second post of this thread you wrote, that I should augment the (4,4) and (5,5) values of the hydrostatic stiffness matrix with the body weight-related restoring. You have repeated that in your last post and also in other threads. However, in the second post of the thread BModes : Input parameters about tower support subsystem
you wrote that of the OC3Hywind.bmi you have augmented the mooring_K matrix with the contributions from the body weight-related restoring instead of the hydro_K matrix. So in my matrices above I have added body weight to the (4,4) and (5,5) of the mooring_K matrix instead of the hydro_K matrix. Is the implementation for the HywindOC3 something special? Generally spoken, the correct approach is to add the augmentation to the hydro_K matrix, not to the mooring_K.
Is this correct?

The augmentation I for the body weight-related restoring is quite small for our TLP. You have explained it before that the importance is high for deep drafted floaters such as a spar. In our case, the augmentation for our TLP is quite small and equals -mgz = -25.05e+06 because our Center of Gravity is located near water level. Therefore, the possible errors in the matrices above will have minor imfluence on our results. Nevertheless, of course it is important to understand where the augmentation should be included correctly.

@Jason.Jonkman

• BModes does not distinguish between tower modes and platform modes. For a floating turbine, often the first six modes are rigid-body modes of the platform at low frequency, perhaps with a bit of coupling to tower deflections. The heave mode is often difficult to spot from the BModes output file because the tower axial displacement is not output in this file. Regardless, in this case, the tower modes would not start until mode 7.

• Often the first tower fore-aft and side-to-side modes have quite similar frequency. Likewise for the second tower fore-aft and side-to-side modes, etc.

• Not including the six rigid-body modes of the platform, I would expect the first tower fore-aft and side-to-side modes to have a slope that doesn’t change sign, and the second tower fore-aft and side-to-side modes to have a slope that changes sign once, etc.

• Assuming your floating wind turbine is symmetric about the XZ plane, I would expect the tower fore-aft modes to be uncoupled from tower torsion (twist), whereas the side-to-side modes would be coupled to tower torsion because of the offset of the rotor mass from the tower centerline.

• When calculating the mode shape for use in FAST / ElastoDyn, I suggest setting the slope at bottom (dy/dx) to equal the slope at span_loc = 0.0.

.

I will consider these, when moving on after making sure that I have the correct inputs for BModes. For the moment I will try to meet with the expected natural frequencies for the substructure, to be sure that the implementation of the latter is correct in BModes. When considering the frequencies from the tank tests I think there is at least one more mistake of major influence in the preprocessing of the hydro_K, the hydro_M, or the mooring_K matrix other then the augmentation of the body weight-related restoring.

Kind Regards,
Daniel