Dear Jason
Sorry to bother you again!
I’m confused to some problems about hydrodynamic model.
As it is referred to hydrodyn user guide. the Vertical center of gravity(VCG) is used to determine the pitch and roll restoring associated with platform weight. In order to neglect the effect of these terms in WAMIT, VCG can be set to zero when solving the first-order problem. There are my questiones:
The cooridinate of COG of semisubmersible FOWT is (-32,0,-17)m. When I calculate the hydrodynamic coeficients by using WAFDUT or AQWA. Should the coordinate of COG be set to (0,0,0) or (-32,0,0)m？ And should the center of rotation be set to (0,0,0) or (-32,0,0)?
2)IF the COG is set to (-32,0,0), the restoring coeficient, the value of added mass and radiation damping et al are refered to the COG in AQWA. Should I convert these datas to the original coordinate systems by using the transmat martix?
HydroDyn expects the hydrodynamic coefficients for the potential-flow solution to be defined about the WAMIT reference point, which is (0,0,0). And the pitch and roll restoring associated with the center of mass of the floating wind turbine are intrinsically accounted for in the ElastoDyn module (and so, should not be accounted for in HydroDyn). Thus, the COG and the hydrodynamic reference point should both be set to (0,0,0) in AQWA (or WAMIT) for the first-order problem.
I am working on a similar 3 column V-shaped platform. Unlike OC3 spar and OC4 semi whose system LCG align with the centreline of the turbine system. The LCG of the platform sits at the middle of the V shape pontoon while the turbine is installed on top of one of the column. When preparing the hydrodynamic model (e.g. WAMIT), it is convenient to set the WRP at one of the column.
I can understand that by setting the VCG of the platform to zero, the restoring associated with the centre of mass of the floating wind turbine is ignored in the hydrodynamic calculation, so that this part will not be double calculated by ElasDy. I assume the hydrodynamic reference point you mentioned above is the rotation point, I can not understand why the rotation point should be set to (0,0,0) as well. The hydrodynamic coefficients, for example: added inertia, will be different depends on where the rotation axis is defined. Would be indeed apricated if you can spare some time to explain the reason why the hydrodynamic reference point should be set to (0,0,0).
The HydroDyn module currently assumes that the WAMIT reference point–i.e., the location where the WAMIT-derived potential-flow solution is specified in HydroDyn–is at (0,0,0). This was a natural choice when HydroDyn was first developed, but is problematic if you have WAMIT data defined about a difference reference point.
That said, we are working on an update to HydroDyn that will allow the user to specify multiple WAMIT bodies (the so-called “N-Body” option in WAMIT), and for each WAMIT body, will allow the user to specify the origin and orientation of that WAMIT body with respect to coordinate system at (0,0,0). While this capability is not yet released, it will provide some powerful functionality in the future.
Can I conclude that at the current stage, a conventional way of simulating this kind of V shaped platform is not accurate due to the difference in the hydrodynamic coefficients introduced by different reference point? That is the required hydrodynamic coefficients for HydroDyn should be specified at (0,0,0) while the actual correct hydrodynamic coefficients should be specified at the COG? The resulting hydrodynamic coefficients due to rotation motions about different reference points are different.
Since the whole purpose of setting the COG to (0,0,0) is to avoid double counting of the restoring due to mass in the ElsaDyn, do you think this can be achieved by only changing the hydrostatic output from hydrodynamic calculation, say wamit, and keep the hydrodynamic coefficients obtained at the cog? This would require run the WAMIT twice, once about (0,0,0) and the other one about cog. Use the hst file from the (0,0,0) and 1,3 files from the cog simulation as the input to FAST.
Alternatively, I guess I can set the WRP to the LCG of the V platform, however, this will require movement of the tower centreline to one of the platform column. Is there a way to achieve this?
I’m not understanding how you are arriving at your conclusions.
There should be no problem to simulate the V-shaped semi in OpenFAST / HydroDyn. But without changing the the source code, the hydrodynamic reference point for the hydrodynamic coefficients and hydrostatic restoring, must be (0,0,0), which is equivalent with the undisplaced / undeflected tower centerline at the mean sea level (which I assume would be the vertex of the V). And the hydrostatic restoring should be defined without consideration of center of mass offsets, i.e., with the center of mass also at (0,0,0), so as to not double count the body-mass-based restoring. The center of buoyancy and center of mass can be specified in OpenFAST’s HydroDyn and ElastoDyn modules, respectively, with offsets relative to (0,0,0).
First of all, as usual, thank you for your prompt reply. Indeed appricated.
Maybe I misunderstood something here. Attached figure is the roll added inertia of the platform, where the baseline denotes the simulation result of the case rotates about the COG, and the RotateZero refers to the case rotates about (0,0,0). The simulations were done in Open Source package Nemoh. There is quite a bit difference between the two as indicated. I originally thought that Elasdyn only corrects the restoring part by setting the cog and cob offsets as you mentioned above, this will add the restoring contribution due to the mass into the hydrostatic restoring matrix, and will not have any impact on the hydrodynamics, for example the added inertia. Does Elasdyn take care of the hydrodynamics coefficients due to different rotation point as well?
The translation and rotation DOFs in ElastoDyn are about (0,0,0) (assuming that you’ve set PtfmRefzt = 0 in ElastoDyn), just like they are in HydroDyn. So, the rotational added mass should be defined about (0,0,0) as well.
Followed with your answer regarding to the COG problem, I would like to know when I calculate PtfmRIner and PtfmPIner in ElastoDyn, should I calculate about (0,0,0) as used in WAMIT or the real centre of mass of the platform as specified by PtfmCMzt.
The platform mass moments of inertia in the ElastoDyn structural-dynamics module of FAST / OpenFAST (PtfmRIner, PtfmPIner, PtfmYIner) are all specified bout the platform center of mass location (PtfmCMxt, PtfmCMyt, PtfmCMzt).
Dear Jason,
I have several questions about the hydrodynamic coefficient files(.hst .1 .3), hope you can help me.
(1) I found that the order of magnitudes is different between .hst file and the documentation “Definition of the Semisubmersible Floating System for Phase II of OC4” . for example, the value of hydrostatic stiffness C33=3.8e6N/m, C44=-3.8e8Nm/rad in your documentation,while C33=3.8e2, C44=-3.8e4 in the .hst files. Is it because of different units? And this difference is also found in .1 and .3 files.
(2) in .3 file, with respect to each period and direction,why there are 6×4 wave excitation forces? are the 4 column wave excitation forces represent 4 floating column of the OC4 ?
The WAMIT output data is nondimensional. See the following forum post for information on how to dimensionalize the WAMIT output: WAMIT - #4 by Jason.Jonkman. That forum topic links to the WAMIT documentation that clarifies the formatting of the WAMIT output files. The four columns in the *.3 file are the amplitude and phase and real and imaginary components of the nondimensional wave-excitation loads.
First of all, thanks for your reply~~Now, I am confused about the definition of attitude. I noticed that you described the roll, pitch, yaw as the rotations of the platform about the axes of the inertial reference frame(X,Y,Z), which means that roll rotates about OX, pitch rotates about OY, yaw rotates about OZ. Can I see that as a way of extorsion ? and how do you define the sequence of rotation, z-y-x or x-y-z ? And I want to know how the FAST deal with the hydrostatic stiffness. For example, assuming that the platform’s yaw(or pitch) is 10 degrees, the roll stiffness about OX is different from that when platform’s yaw(or pitch) is 0. Especially when yaw is 90 degree, the roll stiffness about OX should be equal to the “pitch” stiffness.
The WAMIT output assumes small rotations such that rotation sequence doesn’t matter (angles less than 10 deg).
The ElastoDyn module of FAST / OpenFAST also employs small angle assumptions (again, without a specific sequence), but with a correction on orthogonality of the rotation transformation matrices to ensure that they are orthonormal. This allows the platform rotations to become about 15 degrees before the errors start to get large. See the Equations (1) and (2) in my 2009 Wind Energy Paper for more information: onlinelibrary.wiley.com/doi/abs/10.1002/we.347.
I am a bit confused about reference frames, between WAMIT and OpenFAST.
In BEM (WAMIT/HAMS), my floater was modeled with the direction of positive x exactly opposite (180 deg) to the main wind direction to be simulated
(please, see attached diagram).
I have rotated hydro_k, hydro_m matrices calculated by the BEM for BModes, and calculated eigenmodes and eigenfrequencies, where the 1st pitch eigenfrequency turns up to be at ~0.078Hz, which is what we expected.
I then proceeded to run two decay tests in still water and no wind in OpenFAST where I offset the initial platform pitch by 5 deg, and setting in HydroDyn, PtfmRefztRot = 0 deg and 180 deg, respectively.
When I run the first test (0 deg), from a quick FFT I get the same frequency of 0.078Hz, but when I run the PtfmRefztRot =180deg, I get a completely different spectrum that does not show that frequency at all.
So I am a bit puzzled. I was under the impression that to simulate my dominant wind and wave direction, I would have to use PtfmRefztRot =180deg to account for the fact that BEM had the 180deg offset, yet I was not expecting to get the response so much altered in terms of pitch frequency. And now I am not sure on how to proceed with PtfmRefztRot.
PtfmRefztRot in HydroDyn allows the user to use WAMIT data whose orientation does not align with the OpenFAST global coordinate system. Specifying a nonzero PtfmRefztRot will rotate the WAMIT data (Aij, Bij, Cij, Xi) from the local WAMIT coordinate to OpenFAST global coordinates. This is summarized in the following NREL technical report: nrel.gov/docs/fy20osti/76822.pdf (PtfmRefztRot = theta in section 3.2). As far as I can tell, your use of PtfmRefztRot = 180deg makes sense in your case.
If you are not getting the results you expect, perhaps there is a bug in the implementation of PtfmRefztRot? I’ll need to ask others at NREL what testing was done when PtfmRefztRot was implemented.
