Question regarding underestimation of Surge response for IEA-15-240-RWT-UMaineSemi

Wind & Water Structural Analysis
1

Thank you for all your help.

I have used the published model “IEA-15-240-RWT-UMaineSemi” and input the wave (Hs=13.4m,Ts=13.0) corresponding to DLC6.1 for my analysis. The wave direction is 0 degrees. The wind input is a variable wind of 54 m/s in the direction of 0 degrees.

Q1
The results show that the surge response becomes smaller after about 400s. My expectation was that the surge would be large at a natural period of about 140s. Please tell me the reason why it does not oscillate significantly at the period of 140s. The results do not change even if I extend the computation time.

note.
I am concerned that the response near the natural period of the surge that coincides with the direction of wave action is small, while the Sway response that is orthogonal to the wave direction clearly shows the response near the natural period.

Q2
Is it OK to set WvLowCOff and WvHiCOff of HydroDyn lower than the natural period of Surge? I also tried the calculation with WvLowCOff=0, but the result was almost the same.

image
image1899×1657 429 KB

2

Dear @Yosuke.Matsumoto,

Regarding (1), the large surge response at the natural frequency before 400 s appears to be related to a simulation start-up transient that slowly dies out. The response after 400 s appears to be directly tied to the wave excitation in the surge direction. Because your waves are only exciting the structure in the surge direction, the excitation in the sway direction must be driven by wind or coupling to other modes of motion. (Because you are simulating with a very high wind speed, presumably your rotor is idling with the blades feathered to around 90 degrees, which means the aerodynamic loads on the rotor will be larger in the sway direction than the surge direction.) To identify frequencies in the motion response, I would suggest computing power spectral densities (PSDs) of the motion variables.

Regarding (2), we typically recommend setting the cut-off frequencies of the first-order wave components at the lower and upper limits of the wave energy range to minimize computation expense. Setting WvLowCOff = 0 will increase the computational expense, but likely not change the answer, relative to setting it to the lower limit of the wave energy range because there is no energy there anyway. Regardless, I would guess most of the wave excitation of the platform-surge mode to be driven by second-order difference-frequency effects rather than first-order effects.

Best regards,

3

Thanks for the reply.

I used the FFT and checked the results in the frequency domain.

The above question was to resolve the following issue with the results of IEA-15-240-RWT-UMaineSemi. Below is the actual problem we are facing now. This question relates to a semi-sub floating structure with a 15 megawatt wind turbine, which I asked about in another post. The floating structure is larger than what is publicly available.
The FATS analysis uses SudDyn to account for the elastic response of the floating body. The Orcaflex calculations for comparison also model the floating body with elastic elements.
In both cases, the WAMIT results are

Q1
The surge calculated with FAST has a very small response above 100 seconds, but the Orcaflex results show a large response above 100 seconds. I would like to know what causes this difference; I am having trouble with the main response of the serge not showing up when calculated with FAST.
If we look at the period domain, the FAST response appears to be flat below 2s and above 20s. Is this due to a difference in the dynamic analysis method?

Q2
By second-order difference-frequency, do you mean the WvDiffQTF setting in HydroDyn? Am I correct in understanding that additional calculations are required in WAMIT to obtain this?
Can this effect be taken into account when calculating with the Morrison formula?

Q3
Is it possible that the difference between Orca and FAST results is also due to second-order difference-frequency?

Q4
I have also included a comparison of other tower base moments. I am also troubled by the fact that the FAST results are smaller than the Orcaflex results. In particular, the response at about 2.5 seconds, which is the natural period of the tower.

image
image1613×1302 267 KB

4

Dear @Yosuke.Matsumoto,

Here are my responses:

  1. We’ve compared OpenFAST and OrcaFlex many times and the two models can typically calculate quite similar output when the input settings are consistent. There is little first-order wave excitation below 2 s and above 20 s; instead, second-order effects dominate for lower (sum frequency) and higher (difference frequency) periods.
  2. HydroDyn input WvDiffQTF will enable difference-frequency wave kinematics for the strip-theory solution. But to enable the difference-frequency potential-flow solution, you should set DiffQTF nonzero. Strip theory cannot capture the full effect of potential flow for the difference-frequency solution, but viscous effects from the strip-theory solution may also be important at difference frequencies.
  3. Yes, I would guess that is what is causing the differences at the surge natural period.
  4. From your plots, it looks like both OpenFAST and OrcaFlex predict the tower natural period, but the damping of OpenFAST looks larger. I would guess this is related to some differences in the structural damping of the tower or substructure and/or hydrodynamic damping or viscous drag.

Best regards,

5

Thank you for your answer …

I would like to confirm the details regarding #2.

Q1
Can I use WvDiffQTF and WvSumQTF only when PotMod=1? I want to consider 2ND-ORDER WAVES in strip theory only calculations (PotMod=0).

The reason and situation is as follows.

I am trying to find the sectional force of a floating body using SubDyn. If I set “PotMod=1”, I get very large moments at the reference point, as discussed in the forum below.

I am trying to find the hydrodynamic loads acting on the floating body using only strip theory with “PotMod=0” for HydroDyn.

If WvDiffQTF or WvSumQTF is set to True under these conditions, the following error occurs

FAST_InitializeAll:HydroDyn_Init:HydroDynInput_ProcessInitData:DiffQTF can only be used with PotMod==1. Turning off
HydroDynInput_ProcessInitData:SumQTF can only be used with PotMod==1. Turning off

Q2
I would like to know how to calculate the cross-sectional force of a floating body while taking into account the secondary wave forces.
Is the only way to do this the following method you suggested in the same forum?
Create multiple WAMIT files and assign them to the multiple elements that make up the float.

6

Dear @Yosuke.Matsumoto,

Regarding (1), HydroDyn inputs WvDiffQTF and WvSumQTF only effect the strip-theory solution; they are independent from, and do not effect, the potential-flow solution. HydroDyn inputs DiffQTF and SumQTF enable the second-order potential-flow solution, but these can only be enabled when potential-flow is used.

Regarding (2), for a structure composed of thin members where strip-theory applies, you can enable second-order waves and only use the strip-theory solution. For a large-volume structure where potential-flow applies, if you want internal member-level loads, you must split the potential-flow model up using multiple potential-flow bodies.

Best regards,

7

---------------------- WAVES ---------------------------------------------------
2 WaveMod - Incident wave kinematics model {0: none=still water, 1: regular (periodic), 1P#: regular with user-specified phase, 2: JONSWAP/Pierson-Moskowitz spectrum (irregular), 3: White noise spectrum (irregular), 4: user-defined spectrum from routine UserWaveSpctrm (irregular), 5: Externally generated wave-elevation time series, 6: Externally generated full wave-kinematics time series [option 6 is invalid for PotMod/=0]} (switch)
0 WaveStMod - Model for stretching incident wave kinematics to instantaneous free surface {0: none=no stretching, 1: vertical stretching, 2: extrapolation stretching, 3: Wheeler stretching} (switch) [unused when WaveMod=0 or when PotMod/=0]
4050 WaveTMax - Analysis time for incident wave calculations (sec) [unused when WaveMod=0; determines WaveDOmega=2Pi/WaveTMax in the IFFT]
0.25 WaveDT - Time step for incident wave calculations (sec) [unused when WaveMod=0; 0.1<=WaveDT<=1.0 recommended; determines WaveOmegaMax=Pi/WaveDT in the IFFT]
13.4 WaveHs - Significant wave height of incident waves (meters) [used only when WaveMod=1 or 2]
13.0 WaveTp - Peak spectral period of incident waves (sec) [used only when WaveMod=1 or 2]
1.0 WavePkShp - Peak-shape parameter of incident wave spectrum (-) or DEFAULT (string) [used only when WaveMod=2; use 1.0 for Pierson-Moskowitz]
0.00 WvLowCOff - Low cut-off frequency or lower frequency limit of the wave spectrum beyond which the wave spectrum is zeroed (rad/s) [unused when WaveMod=0, 1, or 6]
100 WvHiCOff - High cut-off frequency or upper frequency limit of the wave spectrum beyond which the wave spectrum is zeroed (rad/s) [unused when WaveMod=0, 1, or 6]

---------------------- 2ND-ORDER WAVES ----------------------------------------- [unused with WaveMod=0 or 6]
True WvDiffQTF - Full difference-frequency 2nd-order wave kinematics (flag)
True WvSumQTF - Full summation-frequency 2nd-order wave kinematics (flag)
0 WvLowCOffD - Low frequency cutoff used in the difference-frequencies (rad/s) [Only used with a difference-frequency method]
100 WvHiCOffD - High frequency cutoff used in the difference-frequencies (rad/s) [Only used with a difference-frequency method]
0 WvLowCOffS - Low frequency cutoff used in the summation-frequencies (rad/s) [Only used with a summation-frequency method]
100 WvHiCOffS - High frequency cutoff used in the summation-frequencies (rad/s) [Only used with a summation-frequency method]

---------------------- 2ND-ORDER FLOATING PLATFORM FORCES ---------------------- [unused with WaveMod=0 or 6, or PotMod=0 or 2]
0 MnDrift - Mean-drift 2nd-order forces computed {0: None; [7, 8, 9, 10, 11, or 12]: WAMIT file to use} [Only one of MnDrift, NewmanApp, or DiffQTF can be non-zero]
0 NewmanApp - Mean- and slow-drift 2nd-order forces computed with Newman’s approximation {0: None; [7, 8, 9, 10, 11, or 12]: WAMIT file to use} [Only one of MnDrift, NewmanApp, or DiffQTF can be non-zero. Used only when WaveDirMod=0]
0 DiffQTF - Full difference-frequency 2nd-order forces computed with full QTF {0: None; [10, 11, or 12]: WAMIT file to use} [Only one of MnDrift, NewmanApp, or DiffQTF can be non-zero]
0 SumQTF - Full summation -frequency 2nd-order forces computed with full QTF {0: None; [10, 11, or 12]: WAMIT file to use}

Thank you for your reply.

Q1
Is the following setting correct when using strip theory? Note that the range of angular frequencies to be considered is quite wide.

---------------------- WAVES ---------------------------------------------------
2 WaveMod - Incident wave kinematics model {0: none=still water, 1: regular (periodic), 1P#: regular with user-specified phase, 2: JONSWAP/Pierson-Moskowitz spectrum (irregular), 3: White noise spectrum (irregular), 4: user-defined spectrum from routine UserWaveSpctrm (irregular), 5: Externally generated wave-elevation time series, 6: Externally generated full wave-kinematics time series [option 6 is invalid for PotMod/=0]} (switch)
0 WaveStMod - Model for stretching incident wave kinematics to instantaneous free surface {0: none=no stretching, 1: vertical stretching, 2: extrapolation stretching, 3: Wheeler stretching} (switch) [unused when WaveMod=0 or when PotMod/=0]
4050 WaveTMax - Analysis time for incident wave calculations (sec) [unused when WaveMod=0; determines WaveDOmega=2Pi/WaveTMax in the IFFT]
0.25 WaveDT - Time step for incident wave calculations (sec) [unused when WaveMod=0; 0.1<=WaveDT<=1.0 recommended; determines WaveOmegaMax=Pi/WaveDT in the IFFT]
13.4 WaveHs - Significant wave height of incident waves (meters) [used only when WaveMod=1 or 2]
13.0 WaveTp - Peak spectral period of incident waves (sec) [used only when WaveMod=1 or 2]
1.0 WavePkShp - Peak-shape parameter of incident wave spectrum (-) or DEFAULT (string) [used only when WaveMod=2; use 1.0 for Pierson-Moskowitz]
0.00 WvLowCOff - Low cut-off frequency or lower frequency limit of the wave spectrum beyond which the wave spectrum is zeroed (rad/s) [unused when WaveMod=0, 1, or 6]
100 WvHiCOff - High cut-off frequency or upper frequency limit of the wave spectrum beyond which the wave spectrum is zeroed (rad/s) [unused when WaveMod=0, 1, or 6]

---------------------- 2ND-ORDER WAVES ----------------------------------------- [unused with WaveMod=0 or 6]
True WvDiffQTF - Full difference-frequency 2nd-order wave kinematics (flag)
True WvSumQTF - Full summation-frequency 2nd-order wave kinematics (flag)
0 WvLowCOffD - Low frequency cutoff used in the difference-frequencies (rad/s) [Only used with a difference-frequency method]
100 WvHiCOffD - High frequency cutoff used in the difference-frequencies (rad/s) [Only used with a difference-frequency method]
0 WvLowCOffS - Low frequency cutoff used in the summation-frequencies (rad/s) [Only used with a summation-frequency method]
100 WvHiCOffS - High frequency cutoff used in the summation-frequencies (rad/s) [Only used with a summation-frequency method]

---------------------- 2ND-ORDER FLOATING PLATFORM FORCES ---------------------- [unused with WaveMod=0 or 6, or PotMod=0 or 2]
0 MnDrift - Mean-drift 2nd-order forces computed {0: None; [7, 8, 9, 10, 11, or 12]: WAMIT file to use} [Only one of MnDrift, NewmanApp, or DiffQTF can be non-zero]
0 NewmanApp - Mean- and slow-drift 2nd-order forces computed with Newman’s approximation {0: None; [7, 8, 9, 10, 11, or 12]: WAMIT file to use} [Only one of MnDrift, NewmanApp, or DiffQTF can be non-zero. Used only when WaveDirMod=0]
0 DiffQTF - Full difference-frequency 2nd-order forces computed with full QTF {0: None; [10, 11, or 12]: WAMIT file to use} [Only one of MnDrift, NewmanApp, or DiffQTF can be non-zero]
0 SumQTF - Full summation -frequency 2nd-order forces computed with full QTF {0: None; [10, 11, or 12]: WAMIT file to use}

Q2
What is the recommended range of angular frequencies?

Q3
I have set WvDiffQTF and WvSumQTF to True. Because we did not have time, the results are for the range of 0 to 800s.
Blue indicates that WvDiffQTF and WvSumQTF are False, and orange indicates True.
When set to True, the long-period component appears to increase at the spectral level. On the other hand, the waveform does not show the large amplitude of about 100s as in the Orcaflex waveform already shown.

Are there any other possible causes?

image
image2425×2179 424 KB

8

Dear @Yosuke.Matsumoto,’

Regarding Q1 and Q2, please find guidance for setting the first- and second-order wave cut-off frequencies in the online HydroDyn documentation on readthedocs: 4.2.8.4. Modeling Considerations — OpenFAST v3.5.1 documentation.

Regarding Q3, again, I would say that strip-theory cannot capture the full effect of potential flow at difference frequencies, so, I’m not surprised an OpenFAST model with second-order strip-theory is not matching an OrcaFlex model with second-order potential flow.

Best regards,

9

Thank you very much.
I understand.