Defining more points along span wise to Blade ED file

[PaulKSmith]

Hello,

Is it possible to discretize the blade ED file more (0.1 m apart) to study the effect of elasticity? For this purpose, does the linear interpolation work among properties already available in the file? What are the settings I need to be careful to avoid any imbalance? It would be helpful if anyone could guide me.

Regards,

[PaulKSmith]

I tried discretizing the Elastodyn file from 51 to 1001. I wanted to check the modal coefficients with BModes but it cannot allocate the array (See the following error). Instead of discretizing uniformly, can I discretize the a smaller portion of the beam where I want to change the stiffness? Will it create any instability ?

Running BModes (v3.00.00, 20-Mar-2008, compiled using double precision).
Linked with the NWTC Subroutine Library (v1.01.08, 26-Sep-2007).
Unable to allocate array, gm, in main program.

Aborting BModes.

[jjonkman]

Dear @PaulKSmith,

You can increase the number of structural analysis nodes in ElastoDyn by increasing BldNodes; the table of distributed blade mass and stiffness properties in the ElastoDyn blade file will be linearly interpolated to these nodes. That said, given the limited number of blade degrees of freedom in ElastoDyn (two flapwise bending and one edgewise bending), there is little benefit of increasing the number of structural analysis nodes to 1000.

Likewise, I don't see any need to increase the resolution of finite elements in BeamDyn so high, but BeamDyn certainly allows you to specify a nonlinear distribution of finite elements if you want to discretize finer in specific areas of the blade.

Best regards,

[PaulKSmith]

Dear @jjonkman,

Thank you for your kind reply and confirmation.
You are right that there will be a little benefit to increase the nodes up to 1000. I just need to adjust the local stiffness of the blade. As the discretized length is more than 1.5m currently, I cannot adjust the stiffness locally. So, the goal is to discretize finer so that I can adjust stiffness over a length of my choice (for example 0.5m). So, I will keep original discretization fixed except for the region in interest. Probably ElastoDyn allow it? I see you mentioned BeamDyn allows the non-linear distribution.

Is there any restriction on no of nodes on BModes to calculate modal coefficients? for 1000 elements it throws an allocation error. Please see the first comment for error.

In my current setup, I only have ElastoDyn file. Do you think I can convert it to use in BeamDyn as well? I am working with DTU10MW.

Regards,

[jjonkman]

Dear @PaulKSmith,

Internally within ElastoDyn, the blade structural analysis nodes can be discretized nonuniformly; however, there is no input file setting for that, so taking advantage of this capability will require a small source code change and recompile.

Regarding the BModes error, sounds like an allocation error (likely due to the large memory requirement of 1000 finite elements), but I'm not familiar with the BModes source code to to comment. To get around this error, I would suggest lowering the number of blade elements (nselt) to something more reasonable and use a nonuniform element distribution.

BeamDyn requires additional information beyond what ElastoDyn does, including torsional and axial stiffness, mass/elastic offsets, and coupling terms. While not in ElastoDyn, this data should exist for the DTU 10-MW RWT. If you are using the IEA Wind version of the DTU 10-MW RWT, you can find an OpenFAST model with BeamDyn here enabled here: https://github.com/IEAWindTask37/IEA-10.0-198-RWT/tree/master/openfast.

Best regards,

[PaulKSmith]

Dear @jjonkman

Thank you for clarification. For my existing setup, the Blade ElastoDyn file looks like it has non-uniform distribution. Therefore, I suppose it linearly interpolates this data again inside code before proceeding further? My model is the old DTU10MW, not sure I am naming it correctly (Got it from another lab member who does not have BeamDyn file as well). For IEATask37 the blade length and other properties are different.

See the Different column, which is BlFract(i) -BlFract(i+1).

I am quite beginner to this field, therefore changing the source code and recompile would take some time. Is there any guide on how to do that?

For my problem I just need to change the flap and edge stiffness at some locations for a element size of [0.1, 0.5, 1.0]m . Would you suggest any possible way out of this situation ? Appreciate your feedback.

Regards,

[jjonkman]

Dear @PaulKSmith,

Yes, ElastoDyn will linearly interpolate the distributed blade property inputs from the blade file to the structural analysis nodes, which are then used in the calculation.

Guidance on compiling OpenFAST can be found here: https://openfast.readthedocs.io/en/main/source/install/index.html#compile-from-source. If you are using Windows, I would install the Intel Fortran compiler and Visual Studio, which are both free.

The uniform distribution of blade analysis nodes is set within SUBROUTINE SetBladeParameters() of ElastoDyn.f90 (p%DRNodes and p%RNodes); that calculation is the only thing you'll need to change.

Best regards.

[PaulKSmith]

Dear @jjonkman,

Thank you for your guidance.
Before attempting to compile the source code, I was checking the BeamDyn input files for IEATask37. From what I noticed, it needs stiffness and mass matrices for individual segments, which can be derived using VABS, BECAS, or NuMAD/BPE. To run NuMAD/BPE I need to have yaml file for the blade. I am not sure for BECAS, probably it also needs 3D FEM blade model. Please correct me if I am wrong. Without having access to yaml/ 3D FEM model it is difficult to discretize finer (more h-refinement) in specific area and alter any parameters. In that case, even if I had BeamDyn file for DTU10MWRWT it would not help unless I have yaml file. Please correct me if I am wrong.

Regards,

[jjonkman]

Dear @PaulKSmith,

I agree that you'll need more information to run VABS, BECAS, and/or NuMAD/BPE.

But as far as I can tell, you have the beam properties needed to populate the 6x6 mass and stiffness matrices, at least the two sets of 3x3 matrices, assuming the bending-axial terms are uncoupled from the shear-torsion terms. See the following document for more information on how to derive the BeamDyn mass and stiffness matrices from the beam properties you already have: https://openfast.readthedocs.io/en/main/_downloads/fd7a8bc10f2371a50828391f75170032/beamdyn_inputs_sectional_props.pdf.

Best regards,