Inertia issue

Wind & Water Computer-Aided Engineering Software Tools
1

Dear Jason，
I am using wamit to do some hydrodynamic analysis of the OC3 spar wind turbine model you’ve published .
I’d like to first do some alteration on the model parameters, thus make some OC3 Spar modified models ,then using Wamit to see how the hydrodynamic performance changes along with the physical parameters of the model.
I just start learning to deal with these kind of engineering problems,so the question is I don’t know how to calculate the mass matrix or the inertia of the modified model.Could you please teach me something on that ?
Best regards,
Qian Yi

2

Dear Qian,

I’m assuming you mean that you want to derive the 6x6 rigid-body mass matrix of the full (rotor + nacelle + tower + platform) system using FAST? You’ll find several examples on the forum where this has been done using either (1) the linearization of FAST v7 with only the six platform DOFs enabled or (2) the FAST-to-ADAMS preprocessor of FAST v7 to generate an ADAMS model, from which the aggregate mass can be calculated within MSC.ADAMS. However, without modification of the source code, it is not possible to directly output the 6x6 mass matrix from FAST v8.

Best regards,

3

Dear Jason,

Thank you so much.

Best regards,

4

Dear Jason，

Since I am now working on a frequency analysis using Wamit. Is there any way that I could output the frequency response transfer matrix via Wamit( not the RAO which is included in the output files by default)?

Best regards.

5

Dear Qian,

I’m sorry, but I don’t know what you mean when you say:

What frequency response transfer matrix are you referring to?

Best regards,

6

Dear Jason,

I’m sorry for my unclear expression, the frequency response transfer matrix I am referring to is actually a transfer matrix which relates the input force with output response, it is often obtained from the inverse of the impedance matrix.

Best regards,

7

Dear Qian,

I don’t believe that WAMIT will calculate any transfer function other than the RAO; however, you can output from WAMIT the matrices you’ll need to compute the frequency response transfer matrix manually outside of WAMIT (e.g. hydrostatic stiffness matrix and frequency-dependent hydrodynamic added mass and damping matrices).

Best regards,

8

Dear Jason,

Thank you so much, this really helps.

Best regards,

9

Dear Jason,

In the published article “Definition of the Floating System for Phase IV of OC3”, there is an explanation about the inertia of the platform.

"The roll and pitch inertias of the floating platform about its CM are 4,229,230,000 kg•m 2 and the yaw inertia of the floating platform about its centerline is 164,230,000 kg•m 2 . These inertias were calculated using a mass distribution appropriate to the floating platform. "

I tried using a simplified calculation “I=m*r^2” to get the yaw inertia of the platform ,and turns out it somehow matches the result given in the article ablove.(Not matched perfectly ,but quite close)

So I wonder what approximation method you are using when calculating the inertias of the platform, both pitch and yaw inertias. Could you please give me a lead on this ? Thanks so much.

Best regards,

10

Dear Qian,

The design details of the Hywind spar where not provided to NREL by Statoil for the development of the OC3-Hywind spar. Instead, the OC3-Hywind spar properties were set to support the NREL 5-MW turbine and to match the design information that was provided by Statoil as much as possible. I’m afraid I can’t provide further details than what is already provided in the OC3-Hywind documentation. You’ll have to make your own assumptions for your own purposes if you need further details on the spar design.

Best regards,

11

Dear Jason,

i try to reproduce the OC3 Phase IV Hywind-Spar Model in Bladed v4.7 to compare these. Do you have any general advice or additional source for variables fitted to Bladed needs?

To model the floating support platform in Bladed no inertia needs to be defined, but the material, wall thickness, density and other basic parameters. As far as I understood there are no specific information about this public available. In order to estimate these values to get the same results as with FAST 8.16 I would like to know weather the Platform inertias about the platforms C.M. are calculated for the whole turbine (with Blades, Hub,…) or only for the support structure till the tower base at 10m above MSL.

I hope you can give me a hint for my problem.

Best regard,
Simon

12

Dear Simon,

The platform mass, center of mass, and inertias specified in the ElastoDyn module of FAST v8 are for the platform itself, not including the tower, nacelle, drivetrain, rotor, or moorings.

It sounds like for Bladed you’ll need to derive a wall thickness/density, and ballast so as to mimic the global platform mass, center of mass, and inertias, but I’ve not done this myself so I don’t have suitable properties to give you.

Best regards,

13

Dear Jason,
I just read this paper nrel.gov/docs/fy13osti/58098.pdf, and tried a fast linearization (steady state) with a OC3 spar model as input,meanwhile I diabled all the DOFs except 6 platform DOFs, and set the mdlorder=2 to get the linearized mass ,damping ,stiffness matrix.
And then I get the result calculated by this linearized process as below.

M - Mass Matrix 1.583E+07 0.000E+00 0.000E+00 0.000E+00 -1.110E+09 -3.228E+06 0.000E+00 1.583E+07 0.000E+00 1.110E+09 0.000E+00 -5.116E+07 0.000E+00 0.000E+00 8.307E+06 3.228E+06 5.116E+07 0.000E+00 0.000E+00 1.110E+09 3.228E+06 1.055E+11 2.824E+07 -5.481E+09 -1.110E+09 0.000E+00 5.116E+07 2.824E+07 1.059E+11 3.468E+08 -3.228E+06 -5.116E+07 0.000E+00 -5.481E+09 3.468E+08 6.367E+08

  1. I am confused with this result ,shouldn’t the(1,1),(2,2),(3,3)elements of the mass matrix equal to the total mass of the whole system ,which should be around 8066048.000?
    2.If I want to consider the aerodynamics effect on the wind turbine and tower (above the SWL) in different wind speed cases ,could I just change the windfile content to make it work ?
    3.When I am doing a linearization process with only 6 platform DOFs enabled. ,what is the difference between an “True” CompAero and a “Fasle” CompAero? Does it mean FAST will calculate the aero impact on the structure above SWL if I set the “CompAero” to True?
    I am a newbie in FAST ,so the questions maybe simple ,hope you could help me on this .
    Best Regards,
14

Dear Qian,

Here are my answers to your questions:

  1. If you are linearizing your FAST v7 model with hydrodynamics enabled, the mass matrix computed by FAST will include the influence of hydrodynamic added mass in addition to the physical mass.

  2. Yes; you’ll also likely need to change the rotor speed and pitch angle appropriate for the given wind speed.

  3. Yes.

Best regards,

15

Dear Jason,

I tried to calculate suitable properties for ballast density, steal density and wall thickness. I got two equations: One for the separate inertias around the z-axis summed up (to the one specified for OC3) and one equation for vertical balance of mass in CM_Ptfm. I assumed reasonable densities for ballast (around 2000 kg/m³) and steal (around 8000 kg/m³). Further I assumed to have constant wall thickness for the whole structure and no structures inside the cylinder.
Then I combined both equations to have only the wall thickness as unknown.
As it is hard to solve the equation I wrote a little MATLAB iteration script.
The result were 0.0951 m for the wall thickness. But, and that’s my problem, the overall platform+ballast mass sums up to 15% less than stated in OC3 Phase IV.
So I changed the densities and for roh_steal=14 000 kg/m³ and roh_ballast=10 000 kg/m³ the difference went down to less than 1% (with a wall thickness of 0.066 m).
do you have an idea where I done a mistake?
I attach the MATLAB-File but its commented in german.

thank you very much and best regards,
Simon
Wallthickness(.m).txt (4.59 KB)

16

Dear Simon,

I haven’t looked at your MATLAB script, but your approach sounds OK. There are likely further design details (e.g. plates and stiffeners) in the original spar design that would lead to differences relative the simple hollow cylinder assumption. I’m not sure the details are really that critical as long as the total mass, center of mass, and inertias are correct and the stiffness is large enough to avoid low-frequency flexural modes, which would be unphysical.

Best regards,

17

Dear Jason,
Thanks for your help.
I tried using the zeros wamit file ( *.1 and *.2 ),and the linearized M matrix seems reasonable which proves you are right about the hydrodynamic effect.
Then when I ran a sanity check on the damping matrix ,it turns out that the linearized damping matrix I got isn’t a zero matrix. C - Damp Matrix 0.000E+00 0.000E+00 0.000E+00 -1.311E-01 -1.802E+01 1.967E-01 0.000E+00 0.000E+00 0.000E+00 8.997E+00 -1.302E-01 4.665E+04 0.000E+00 0.000E+00 0.000E+00 2.531E-05 -9.328E+04 6.126E-06 0.000E+00 0.000E+00 0.000E+00 -1.050E+03 -1.046E+01 5.029E+06 0.000E+00 0.000E+00 0.000E+00 1.418E+01 9.773E+00 -2.127E+01 0.000E+00 0.000E+00 0.000E+00 1.119E+03 4.538E-05 1.053E+03
But I don’t know what causes this problem, I try to describe my configuration in this sanity check below.

  1. I am using the original FASTv7 exe (without recompile) from nwtc.nrel.gov/FAST7
  2. In the primary input file ,I set AnalMode=2,PCMode =0,RotSpeed =0 , Disabled all the DOFs and CompAero as well .
    In the linearization control file ,I set CalStdy =True MdlOrder=2.
    In the platform file ,I set the
    PtfmLdMod =FltngPtfmLd ,
    WAMITFile to zeroed *.1 and *.2 file and a correct *.hst file,
    PtfmCD=0
    wavemod =0
    CurrentMode=0

Is there anything wrong with this configuration? If there is please tell me how to get it fixed ,I really appreciate your help .
I attached the major input file below in case I didn’t make it clear to you.
Best regards.
OC3SanityCheck-Linear.dat.txt (2.09 KB)
OC3SanityCheck.fst.txt (22.5 KB)
OC3SanityCheck-Platform.dat.txt (7.91 KB)

18

Dear Qian,

Your files approach look reasonable to me.

My guess is that there is small residual motion of the platform that is leading to a small amount of damping from linearization of the nonlinear structural model. You’ve eliminated all aerodynamic and hydrodynamic damping from your model, which may make it difficult to find a static-equilibrium solution through a time-domain-based steady state calculation. If you initialize the platform displacement so that it is at (or very close) to the natural steady state solution, my guess is the damping terms will decrease.

Best regards,

19

Dear Jason,
First of all ,thank you for your help as always.
I noticed there is a ptfmCD option in the platform input file to define the viscous drag coefficient .
So I tried two tests .
one with the ptfmCD=0 like the sanity test I mentioned in the former post,one with the ptfmCD=0.6.
And it turns out the linearized C,K matrix are different,I wonder what causes this difference since I didn’t define any morison element in the input file(excatly the same with the attachment in my post above).
Hope you could help me figure this out ,many thanks.
Best regards.

20

Dear Qian,

I’m not sure why you say that you “didn’t define any Morison element”. In your OC3SanityCheck-Platform.dat.txt file, you have PtfmNodes, PtfmDraft, and PtfmDiam all set nonzero. This will lead to Morison-style viscous drag when PtfmCD is nonzero.

Best regards,