search

OpenFAST / openfast

Main repository for the NREL-supported OpenFAST whole-turbine and FAST.Farm wind farm simulation codes.
http://openfast.readthedocs.io
Apache License 2.0
693 stars 458 forks source link

Question regarding the ElastoDyn #2434

Open PaulKSmith opened 1 month ago

PaulKSmith commented 1 month ago

Hello everyone,

Currently I am trying to study the effect of blade flexibility on system response. For that I am reducing the stiffness. I am wondering if that makes sense as I am keeping other parameters in Elastodyn file same. I think the mode shape coefficients, natural frequency, and damping might change as well. That means I have to generate new ElastoDyn file ? Could anyone share any thought on how to do that? I read about Bmodes but would appreciate any suggestions.

Regards,

jjonkman commented 1 month ago

Dear @PaulKSmith,

For ElastoDyn's blade model, the user must specify the mass and distributions, as well as the mode shape and structural damping ratio for each mode. The mode shapes are intrinsically tied to the mass and stiffness distributions, so, I agree that if you change the latter, you should recalculate the mode shapes (e.g., using BModes).

The structural damping ratios do not necessarily scale with a stiffness change, but it is difficult to assess damping levels without more information, and in the absence of that, it may be preferred to assume a conservatively low level of structural damping.

Best regards,

PaulKSmith commented 1 month ago

Dear @jjonkman Thank you for the reply! Appreciate it very much. Now, I am trying to run BModes using the download link mentioned in this previous post: (https://forums.nrel.gov/t/bmodes-download/2306/14)

I could easily run the test cases. Howver, I my intention is to reproduce the modal coefficients first mentioned in the ElastoDyn file for the DTU10MW. Here are the parameters I have in ElastoDyn Blade input file:

BlFract PitchAxis StrcTwst BMassDen FlpStff EdgStff GJStff EAStff Alpha FlpIner EdgIner PrecrvRef PreswpRef FlpcgOf EdgcgOf FlpEAOf EdgEAOf

However, I see that cg_offst sc_offst tc_offst missing from here. So, I am wondering how to calculate these parameters to run BModes?

I checked ElastoDyn Blade file of the (https://github.com/IEAWindTask37/IEA-10.0-198-RWT) and noticed it has only six columns [ BlFract PitchAxis StrcTwst BMassDen FlpStff EdgStff]. Does that mean it does not need other parameters from ElastoDyn which I am using above? I am using OpenFAST-v3.5.2-dirty to run my simulations and it is perfectly working with the above mentioned ElastoDyn input file.

Regards,

jjonkman commented 1 month ago

Dear @PaulKSmith,

The blade input file of ElastoDyn within OpenFAST only needs 6 columns of distributed blade data (BlFract PitchAxis StrcTwst BMassDen FlpStff EdgStff); older versions within FAST v7 supported additional columns for use in the FAST-to-ADAMS preprocessor that is no longer supported.

Regarding how to set up the BModes input file for consistency with the ElastoDyn model, see my response dated March 18, 2015 in the following topic on our forum: https://forums.nrel.gov/t/nrel-5mw-blade-mode-shape/1092.

Best regards,

PaulKSmith commented 1 month ago

Dear @jjonkman ,

Thank you for the reply. Always appreciate it. It took some time to read all the forums associated with the one you mentioned. I noticed that there are differences between IEAWindTAsk37 and old FAST v7 ElastoDyn input of the DTU 10MW model, therefore the modal coefficients changed between these two models.

To build my confidence, I wanted to validate the IEAWindTAsk37 ElastoDyn input file modal co-efficients with BModes. However, I see some differences. Here is the summary:

<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:x="urn:schemas-microsoft-com:office:excel" xmlns="http://www.w3.org/TR/REC-html40">

-0.33667 | BldFl1Sh(2) | Flap mode 1, | coeff | of | x^2 | ==> | -0.35311 -- | -- | -- | -- | -- | -- | -- | -- 3.86045 | BldFl1Sh(3) | , | coeff | of | x^3 | ==> | 3.883281 -9.50716 | BldFl1Sh(4) | , | coeff | of | x^4 | ==> | -9.51792 12.17967 | BldFl1Sh(5) | , | coeff | of | x^5 | ==> | 12.07877 -5.19628 | BldFl1Sh(6) | , | coeff | of | x^6 | ==> | -5.09102   |   |   |   |   |   |   |     |   |   |   |   |   |   |   -2.16796 | BldFl2Sh(2) | Flap mode 2, | coeff | of | x^2 | ==> | -2.02892 16.57105 | BldFl2Sh(3) | , | coeff | of | x^3 | ==> | 15.71578 -52.2934 | BldFl2Sh(4) | , | coeff | of | x^4 | ==> | -49.9678 62.58792 | BldFl2Sh(5) | , | coeff | of | x^5 | ==> | 59.74453 -23.6977 | BldFl2Sh(6) | , | coeff | of | x^6 | ==> | -22.4636   |   |   |   |   |   |   |     |   |   |   |   |   |   |   0.268236 | BldEdgSh(2) | Edge mode 1, | coeff | of | x^2 | ==> | 0.495838 1.718565 | BldEdgSh(3) | , | coeff | of | x^3 | ==> | 0.463422 -3.79766 | BldEdgSh(4) | , | coeff | of | x^4 | ==> | -1.21502 5.201944 | BldEdgSh(5) | , | coeff | of | x^5 | ==> | 2.931682 -2.39109 | BldEdgSh(6) | , | coeff | of | x^6 | ==> | -1.67592

Right side of the table is the newly calculated coefficients while those are on the left side are copied from IEAWindTask37. Although there are some small differences in flap modes, edge mode coefficients are deviated the most. I am wondering if the coefficients in IEAWindTask37 are also derived the same way (uncoupled)? I understand from the previous forum discussion, the flap, and edge modes are needed to be calculated uncoupled from BModes and later it will be coupled during runtime in OpenFAST. Without knowing the insights of ElastoDyn, I am assuming any mode derived other than the settings you recommended will be incorrect? Please correct me if I am wrong.

You also commented that With these features, my experience has been that while the prescribed mode shapes input to FASTshould be reasonable, they do not need to be exact for accurate response predictions. in https://forums.nrel.gov/t/the-consideration-of-interactive-effect-between-blade-and-tower-rotor-when-the-bmodes-mode-shapes-are-entered-to-fast/1507/2

Saying that, how do I confirm if the small change in modal coefficients due to change in elasticity (which I am studying currently) is different from the changes of two models (e.g., IEAWindTask37 and freshly calculated BModes) ?

I also wanted to know the insight on how the modes are coupled during runtime ? Also if torsion and axial can be modelled as well through ElastoDyn? I would like to request some reference about these.

Thank you always!

Regards,