Error during linearization with VolturnUS-S

Andre.White · June 11, 2021, 1:55pm

Hi Jason:

I have come up on a hitch when trying to carry out linearization using the VolturnUS-S model is OpenFAST. I amended the necessary VolturnUS-S input files using the DeepCwind linerization model. However, I got the error shown below. I should also point out that I carried out a simulation using the original VolturnUS-S model and the model ran satisfactorily. You will see that the first error that comes up when I attempted linearization is about ElastoDyn nodal output section, which has left me bemused as that error did not come up during the normal simulation. See error below.
Running ElastoDyn.
Nodal outputs section of ElastoDyn input file not found or improperly formatted.
Running ServoDyn.
Running HydroDyn.
FAST_InitializeAll:HydroDyn_Init:HydroDynInput_ProcessInitData:Cannot use full
difference-frequency 2nd-order wave kinematics for linearization. Set WvDiffQTF=FALSE.
HydroDynInput_ProcessInitData:Full difference-frequency 2nd-order forces computed with full QTF
cannot be used for linearization. Set DiffQTF=0. FAST encountered an error during module initialization.

Now this error occurs for ExctnMod = 0 and RdtnMod = 0. However, setting them to 2 as per the DeepCwind model resulted in this error:
Running ElastoDyn.
Nodal outputs section of ElastoDyn input file not found or improperly formatted.
Running ServoDyn.
Running HydroDyn.
FAST_InitializeAll:HydroDyn_Init:HydroDynInput_ProcessInitData:Cannot use full
difference-frequency 2nd-order wave kinematics with state-space wave excitations. Set
WvDiffQTF=FALSE.
HydroDynInput_ProcessInitData:Full difference-frequency 2nd-order forces computed with full QTF
cannot be used with state-space wave excitations. Set DiffQTF=0.
HydroDynInput_ProcessInitData:Cannot use full difference-frequency 2nd-order wave kinematics for
linearization. Set WvDiffQTF=FALSE.
HydroDynInput_ProcessInitData:Full difference-frequency 2nd-order forces computed with full QTF
cannot be used for linearization. Set DiffQTF=0.
FAST encountered an error during module initialization.

I have checked a few other forums but I have not seen any solution to this. Would be very grateful for your input on how to fix it.

Thank you.

Regards,
AOAW

Jason.Jonkman · June 11, 2021, 3:04pm

Dear Andre,

The “nodal output” warning will not effect the simulation results; you can ignore the warning.

Otherwise, the error messages are pointing to the solution:

Set WvDiffQTF = False – You cannot linearize HydroDyn with 2nd-order wave kinematics enabled
Set DiffQTF = 0 – You cannot linearize HydroDyn with 2nd-order potential flow enabled

Best regards,

Andre.White · June 14, 2021, 2:45pm

Jason:

Many thanks. However, I ran into another glitch. I seem to be missing two WAMIT files from the HydroData folder. See below for the message about the .ssexctn file. When I checked said HyrodData folder I realized that the .ss file is also missing. Do you know where I could locate these files?

FAST_InitializeAll:HydroDyn_Init:WAMIT_Init:SS_Exc_Init:OpenFInpFile:The input file,
“.\HydroData/IEA-15-240-RWT-UMaineSemi.ssexctn”, was not found.

Thank you.

Regards,
AOAW

Daniel.Zalkind · June 14, 2021, 9:48pm

Hi Andre,

I just added the state space hydrodynamic files to my fork of the IEA-15MW repository here: [url]https://github.com/dzalkind/IEA-15-240-RWT/tree/hydro_linearization/OpenFAST/IEA-15-240-RWT-UMaineSemi/HydroData[/url]

I submitted a PR to get them into the main branch soon. I hope this helps!

Best, Dan

Andre.White · June 15, 2021, 2:55pm

Many thanks Dan.
That solves the problem.

Regards,
AOAW

Andre.White · June 15, 2021, 5:34pm

Hi Jason/Dan
I have attached the natural frequencies that I obtained for the VolturnUS-S. Are the values obtained satisfactory (values in second column)?
I only managed to get exact values for the yaw and heave DOFs.

Thank you.
Regards,
AOAW

Jason.Jonkman · June 15, 2021, 5:39pm

Dear Andre,

The values you have obtained through the OpenFAST linearization are quite close to those reported. I’m not really familiar enough with the Volturn-US model or report to know how the reported values were obtained. Of course, you could check that the OpenFAST linearization results are converged be ensuring the solution is in steady state before linearizing and by outputting with higher precision (via OpenFAST input parameter OutFmt).

Best regards,

Andre.White · June 16, 2021, 1:22pm

Jason:
Thank you. Just seeking some further clarification based on your last response. You said:

“by ensuring the solution is in steady state before linearizing” - How is this done? Is there something to check for to ensure steady state before linearizing?
“by outputting with higher precision (via OpenFAST input parameter OutFmt)”- I currently have this set to “ES10.3E2”. Is there is way to output with higher precision than this?

Regards,
AOAW

Jason.Jonkman · June 16, 2021, 3:20pm

Dear Andre,

Here are my answers to your questions:

Yes, you can plot the time series (e.g., of platform motions) to ensure that the solution is in equilibrium before linearizing. Enabling CalcSteady = True provides an automated way of ensuring the solution is in steady state before linearizing.
Yes, this is standard Fortran format statement. OutFmt = “ES.10.3E2” is a scientific notation with three digits behind the period and two digits in the exponent, so, only 4 digits of precision. A higher precision is, e.g., OutFmt = “ES17.9E3” which has 10 digits of precision with three digits in the exponent. We generally recommend outputting the linearization with high precision because of the sensitivity of eigensolution to precision.

Best regards,

Andre.White · June 21, 2021, 9:46am

Thanks Jason.

Regards,
AOAW

Andre.White · October 8, 2021, 3:25pm

Hi Jason:
I am getting the following error in MATLAB where I try to process the .lin file. See below.
[mbc_data,matData,FAST_linData]=fx_mbc3(‘5MW_OC4Semi_Linear_original.1.lin’);
[b]Index exceeds the number of array elements (1).

Error in ReadFASTLinear (line 37)
if strfind( C{1}{2}, ‘Yes’ )

Error in fx_getMats (line 46)
data(matData.NAzimStep) = ReadFASTLinear(FileNames{matData.NAzimStep}); %we’ll read this twice so we can
allocate space first; putting it at matData.NAzimStep saves some reallocation later

Error in fx_mbc3 (line 41)
[matData, FAST_linData] = fx_getMats(FileNames);[/b]

I am not sure why this is happening since everything worked fine before. Do you have any idea what could be the source of this error?

Thank you.

Regards,
AOAW

Jason.Jonkman · October 8, 2021, 5:44pm

Dear Andre,

From the error message, I see that the line of code generating the error has changed in the newest code base. I’m guessing upgrading to the newest version of MBC3 and the MATLAB Toolbox will solve the problem. Can you try the version currently on the main branch: github.com/OpenFAST/matlab-toolbox?

Best regards,

Andre.White · October 11, 2021, 10:33am

Thanks Jason. That solves the problem. Including the problem I was having with the ReadFastBinary.m function.

Regards,
AOAW

Andre.White · April 4, 2022, 10:56am

Hi Daniel:
I am not sure if you are familiar with carrying out free decay test in OpenFAST. I have carried out free decay test but I had to use different values for the initial displacements of the platform DOFs (in ElastoDyn) to get similar natural frequencies to the ones in the paper by Allen et al (2020). Can you please say if the natural frequencies are affected by the initial displacements in the six DOFs?

Thank you.

Regards,
AOAW

Andre.White · April 4, 2022, 11:00am

Hi Daniel:
Just to clarify that the previous query is for the VolturnUS-S floater.

Regards,
AOAW

Jason.Jonkman · April 4, 2022, 12:39pm

Dear @Andre.White,

I’m not too familiar with this specific model, but is this related to geometric nonlinearities in the mooring system? Can you clarify what initial conditions you are changing the different results you are obtaining?

Best regards,

Andre.White · April 4, 2022, 2:14pm

Hi @Jason.Jonkman ,
The initial conditions I have used are shown below:
1.5 m PtfmSurge -
1.5 m PtfmSway
1.5 m PtfmHeave
0.25 m PtfmRoll
0.25 m PtfmPitch
1.5 m PtfmYaw

The initial conditions that were used in paper by Allen et al (2020) are:
20.0 m PtfmSurge -
2.0m m PtfmSway
4.0 m PtfmHeave
10.0 m PtfmRoll
10.0 m PtfmPitch
10.0 m PtfmYaw

The initial conditions by Allen gives the system natural frequencies as per the paper. However, those values do not yield system natural frequencies (in my model) so I had to tweak the values to the ones shown first above to get similar natural frequencies.
AOAW

Jason.Jonkman · April 4, 2022, 2:37pm

Hi @Andre.White,

How different are the natural frequencies when you use the larger initial conditions? Regardless, this sounds like it is cause by geometric nonlinearities in the mooring system. That said, I’m not sure why your results are not matching the paper.

I would suggest posting your question on the issues page of the IEA Wind 15-MW reference wind turbine repository: Issues · IEAWindTask37/IEA-15-240-RWT · GitHub.

Best regards,

Andre.White · April 4, 2022, 2:56pm

Thanks @Jason.Jonkman . I will raise the query on the IEA Wind 15-MW reference wind turbine repository.

Regards,
AOAW

Andre.White · April 6, 2022, 4:29pm

Hi @Jason.Jonkman I have one more query.
When doing wind turbine design/analysis (in this case with OpenFAST), do you have to run simulations for random seeds and use the highest value from the different seeds as the design value?

Regards,
AOAW