search

IEAWindTask37 / IEA-15-240-RWT

15MW reference wind turbine repository developed in conjunction with IEA Wind
Apache License 2.0
203 stars 125 forks source link

Questions on subdyn TP #183

Open Ran-Tu opened 4 months ago

Ran-Tu commented 4 months ago

Hi developer,

Could you please check if below configurations are correct? The subdyn TP is originally set at Joint 10 which is -10 m below mean sea level. Should it be Joint 19 which is 15 m above mean sea level which is the tower bottom and tower reference point?

https://github.com/IEAWindTask37/IEA-15-240-RWT/blob/f09558d500cfefd0c6e18b26b6bf359c04ea7757/OpenFAST/IEA-15-240-RWT-Monopile/IEA-15-240-RWT-Monopile_SubDyn.dat#L24C1-L56C77

And could you please explain the reason why each joint has a redundant joint very close to each other, e.g. -30.0 and -29.999, -25.0 and -24.999. I checked the MPropSetID, and it seems that they adopted the same MPropSetID for two joints which are close to each other. I am confused by the reason setting such a redundant joint.

Kind regards, Ran

gbarter commented 4 months ago

Great catch. Definitely a bug in the interface joint label.

The reason for the joints very close to one another is to achieve a piecewise-constant thickness distribution, instead of linearly varying thickness, as it better captures monopile manufacturing realities from rolling steel plates. https://github.com/IEAWindTask37/IEA-15-240-RWT/blob/master/OpenFAST/IEA-15-240-RWT-Monopile/IEA-15-240-RWT-Monopile_SubDyn.dat#L79-L91

Ran-Tu commented 4 months ago

Hi @gbarter

In the monopile subdyn file, the joints very close to one another adopted the same cross-section property. For example, both joint1 and joint2 of member1 adopted PropSet1. And both joint2 and joint3 of member2 adopted PropSet1. In my understanding, that is equivalent to a member staring from joint1 and ending at joint3 which adopts a constat PropSet1. Therefore, joint2 is not necessary.

Similarly, member 3 and member4 have the same cross-section property PropSet2, meaning joint4, which is in between member 3 and member4, is unnecessary.

I am confused why those joints, e.g. joint2, joint4, joint6 are remained. They will lead to members with a short length of 0.001m. I think it may do harm to the convergency performance. However, if removing those unnecessary joints, e.g. joint2, joint4, joint6 and so on, the monopile is still distributed as a piecewise-constant thickness. But we can avoid short-length members.

The reason for the joints very close to one another is to achieve a piecewise-constant thickness distribution, instead of linearly varying thickness, as it better captures monopile manufacturing realities from rolling steel plates.

Kind regards, Ran

gbarter commented 4 months ago

Yup, I should have been more thorough in my review. You are correct that the current element-property assignment did not achieve the desired distribution. This has now been fixed. I also caught that the transition piece point mass needed to be moved to the correct node as well. See #185

Ran-Tu commented 4 months ago

Thanks for the fix.