## The prebent blade and aerodyn15

Provide feedback, request enhancements, and get help with wind-turbine computer-aided engineering tools.

Moderators: Bonnie.Jonkman, Jason.Jonkman

Ruiliang.Wang
Posts: 87
Joined: Sun Nov 24, 2013 6:21 pm
Organization: YANSHAN UNIVERSITY
Location: China

### The prebent blade and aerodyn15

hello,Jason
I have a problem about the prebent blade.I want to calculate the long prebent blade in my FAST model,I want to use the BeamDyn originally,but the convergence problem occured when i use the beamdyn,I have to give up the beamdyn option,I choose the ElastDyn to calcualte the blade dynamic. But i want to use the Aerodyn15. In aerodyn15,We must input the parameter like BlCrvAC , BlCrvAng and so on .As I know,the ElastDyn can calculate the blade without prebending. But aerodyn15 calculate the aerodynamic load by the prebent blade we define.
Is it reasonable to use the ElastDyn and prebent Aerodyn15?
How did the two moudle work about the prebent blade ? Could the ElastDyn calculate the blade by prebending in aerodyn15?
Convergence error happended frequently when I use BeamDyn.Do you have some good advices about how to build own prebent blade BeamDyn model in order to avoid the convergence problem. To be honest ,the converagence problem is boring.

Best regard
Ruiliang.Wang

Jason.Jonkman
Posts: 5733
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

### Re: The prebent blade and aerodyn15

Dear Ruiliang,

The mesh-mapping functionality of FAST permits each module to have its own statial discretization and spatial extent. This functionality allows one to transfer motions and loads between ElastoDyn, which assumes an initially straight blade, and AeroDyn v15, which permits the modeling of precurved and preswept blades. See the following papers for more information on the mesh-mapping functionality of FAST: http://www.nrel.gov/docs/fy14osti/60742.pdf and http://www.nrel.gov/docs/fy16osti/63203.pdf.

There are several topics on this forum that have discussed how to set up BeamDyn to model precurved or preswept blades e.g. see: viewtopic.php?f=4&t=1451&p=6609. Have you already followed this advice? Also, which version of FAST are you using? There have a been a number of improvements in BeamDyn and AeroDyn in the upgrade from FAST v8.16 and OpenFAST v1.0.0 and newer.

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Ruiliang.Wang
Posts: 87
Joined: Sun Nov 24, 2013 6:21 pm
Organization: YANSHAN UNIVERSITY
Location: China

### Re: The prebent blade and aerodyn15

Hello,Jason
It help me a lot .I am using OpenFAST-1.0.0.Recently I am trying to recompiling OpenFAST by VS2013 and IVF2013.I do nothing changed,But error happened like below

Code: Select all

The "Running Registry for InflowWind" error happened in the FASTlib,and others in FAST.
but I can compile FAST8.16 normally.
Can you give me some suggestions?

Best Regard
Ruiliang.Wang

Jason.Jonkman
Posts: 5733
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

### Re: The prebent blade and aerodyn15

Dear Ruiliang,

It looks like the FAST Registry didn't run. Are you using the Visual Studio project provided with OpenFAST v1.0.0?

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Ruiliang.Wang
Posts: 87
Joined: Sun Nov 24, 2013 6:21 pm
Organization: YANSHAN UNIVERSITY
Location: China

### Re: The prebent blade and aerodyn15

hello,Jason
Yes, I'm using the the Visual Studio project named FAST.vfproj in vs-build folder provided with OpenFAST v1.0.0 what i download from github relases.

Best Regard

Ruiliang.Wang

Jason.Jonkman
Posts: 5733
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

### Re: The prebent blade and aerodyn15

Dear Ruiliang,

You are showing errors 27-47; what are the errors 1-26?

Did you try closing the Visual Studio project, reopening it, and try compiling again?

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Ruiliang.Wang
Posts: 87
Joined: Sun Nov 24, 2013 6:21 pm
Organization: YANSHAN UNIVERSITY
Location: China

### Re: The prebent blade and aerodyn15

Dear Jason.

The whole errors and warnning are displayed below

Code: Select all

I try for many times ,I think there are something wrong in VS setting? Do I need special setting in VS tool?
The "引用的局部变量" is unreferenced local variables.

Best Regard
Ruiliang.Wang

Jason.Jonkman
Posts: 5733
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

### Re: The prebent blade and aerodyn15

Dear Ruiliang,

What is reported in the Output window in Visual Studio when you try to compile?

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Abhinay.Goga
Posts: 40
Joined: Tue Oct 25, 2016 8:12 am
Organization: SETEC elektronische Antriebsregelung GmbH
Location: Germany

### Re: The prebent blade and aerodyn15

Dear Jason and all,

Currently I am working on an onshore turbine with prebent rotor using ElastoDyn+BeamDyn. For the convergence issue, I kept the glue code time step as 0.0001 and its taking a humongous amount of time. Is there any possible ways to reduce the simulation time while coupled with BeamDyn?

When I tried to simulate an entire power production DLC, some of the simulations are aborting in the middle due to convergence. Could it be solved by changing NRMax input from default(10) to a higher value?

Thanks and regards
Abhinay Goga

Jason.Jonkman
Posts: 5733
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

### Re: The prebent blade and aerodyn15

Dear Abhinay,

The small time step requirement is driven by frequencies transferred between BeamDyn and ElastoDyn. The only way to increase the time step is reduce that very high frequency in some way, e.g. by reducing the stiffness (are all stiffness values specified realistic or are some made up and artificially high?) or eliminating the structural DOFs that result in the very high frequency, e.g., dropping order_elem.

I would suggest identifying the troublesome high frequency through an OpenFAST linearization analysis to understand the source of the problem. This should provide insight on how best to solve it.

I generally don't touch the BeamDyn inputs in the SIMULATION CONTROL section of the primary input file. It may be that the time step is indeed higher than it should be and reducing the time step further would solve the problem.

Please note that there is a known issue in BeamDyn associated with blades with prebend. Until this bug is fixed, I would use caution interpreting the results of a BeamDyn model with a prebent blades.

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Abhinay.Goga
Posts: 40
Joined: Tue Oct 25, 2016 8:12 am
Organization: SETEC elektronische Antriebsregelung GmbH
Location: Germany

### Re: The prebent blade and aerodyn15

Dear Jason,

Thanks for the insights. I will use caution while working with BeamDyn. Could you please explain or guide me to a thread where I can learn more about this bug that you mentioned. My present aim is to work with a prebent blade. The Stiffness matrix of the blade sections are well defined as well as the axis inputs for the prebent curve are adjusted for a smooth curve.

I will follow your suggestion for switching of the DOFs in ElastoDyn. I am relatively new to linearization analysis. After the initial linearization, there are certain .lin files containing Linearized state matrices. Using matlab I calculated eigen frequencies for the resulting matrices. Is there any material where can I understand how to analyze the results for identifying the troublesome frequencies for my setup? Your guidance is always well appreciated.

Kind regards
Abhinay Goga

Jason.Jonkman
Posts: 5733
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

### Re: The prebent blade and aerodyn15

Dear Abhinay,

The known issues in BeamDyn are summarized well in the following issue on the OpenFAST github site: https://github.com/OpenFAST/openfast/issues/366.

There have been many discussions about how to interpret the results of linearization analysis, both here on this forum as well as on the OpenFAST issues page: https://github.com/OpenFAST/openfast/issues. I would suggest searching for "linearization", "MBC3", "Eigenanalysis", and/or "Campbell".

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Abhinay.Goga
Posts: 40
Joined: Tue Oct 25, 2016 8:12 am
Organization: SETEC elektronische Antriebsregelung GmbH
Location: Germany

### Re: The prebent blade and aerodyn15

Dear Jason,

I am facing exactly the same issue of convergence for some simulations, with OpenFASTv2.3.0 and a time step of 0.0001s

the maximum number of iterations
FAST encountered an error at simulation time 324.34 of 650 seconds.
Simulation error level: FATAL ERROR
Aborting OpenFAST.''

Correct me, if I am wrong. Through linearization I obtained state matrices. Considering: Mod(A-lamda*I)*X=0 , I have obtained eigen vector. Using FFT, I can obtain natural frequencies, right? In your previous comment you mentioned 'dropping order_elem'. Did you mean to reduce the value in BeamDyn input? Currently, I am using a value of 4.

I am not sure, if I understand you correct or not. In the thread 'https://github.com/OpenFAST/openfast/issues/366' you mentioned that for linearization analysis parked/idling is sufficient. But to calculate the operational frequencies the turbine needs to be in operational condition, right?

Thanks and kind regards
Abhinay Goga

Jason.Jonkman
Posts: 5733
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

### Re: The prebent blade and aerodyn15

Dear Abhinay,

Yes, you can obtain the natural frequencies both through an eigenanalysis and through FFT of the time series, but it is typically far easier to obtain all of the natural frequencies and damping and to interpret the eigenmodes through an eigenanalysis. (With an FFT analysis, you'll need to ensure that the modes get excited and it is often difficult to interpret and to quantify the damping level.)

A value of OrderElem = 4 is already quite low, so, I'm a bit surprised such a small time step is required. Are all of the mass and stiffness values in your BeamDyn model set to physically correct values?

Calculating the natural frequencies in a parked/idling condition is more straightforward than in the operational condition and should provide insight into which modes are introducing high frequencies and the associated time-step requirements. Certainly the natural frequencies during operation will differ a bit, with some frequencies increasing and some decreasing with rotational speed due to, e.g., centrifugal stiffening. But I would analyze the parked/idling condition before moving to an eigenanalysis in an operational condition.

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Abhinay.Goga
Posts: 40
Joined: Tue Oct 25, 2016 8:12 am
Organization: SETEC elektronische Antriebsregelung GmbH
Location: Germany

### Re: The prebent blade and aerodyn15

Dear Jason,

With the insights from previous threads on Natural frequencies, using MBC3 script I have calculated the natural frequencies. Parked DLC with exclusion of Aero and Inflow(Only Elasto, Beam and Servo activated), and the linearization performed for debug (all inputs, outputs & inclusion of Jacobians).
But I am not completely sure what I am looking at. Could you please explain anything from the results that I attached.

The damping coefficients are of the order of -03 and blade distributed properties for the assigned sections are relatively correct. I understood that because of a certain bug in BeamDyn, when the blades experience edgewise loads the resulting forces near the tip section are inaccurate. But I would very much like to increase the time step from 0.0001 to at least 0.001. Any guidance in this direction is very much appreciated.

Thanks and kinds regards
Abhinay Goga
Attachments
Natural frequencies.txt