Flange-bolt joint, Abaqus, wind turbine tower

[Kanwarosama]

Hi all,
I am trying to model a wind turbine tower. I was successful in modelling the geometry and thickness but I am stuck when it comes to connections which are flange-bolt type. I can simplify my problem by totally eliminating the need to model the flange and bolt and use the tie constraint to connect the cylindrical sections together. But this approach is very simple. On the other hand, the approach to model the flange and the bolts will make the model very complicated because there are few hundred bolts in my whole tower.
I am thinking to simplify the problem in such a way that either I can pull out the bolt forces or provide some stiffness as in case of a springA element so that i can parametrize it easily. The aim being to understand the joint effect on whole tower. I have tried using spring by creating attachment points on the ends of each sections and then connecting them in a straight line. But this model isnt stable. Ofcourse, i have to provide a gap between sections which is also not realistic but to provide springs i had to do that. Also, I have tried using cartesian connector but that also fails when i do a frequency analysis as the first three modes are rigid body modes. The same result in observed for a springA elememt. So, I am looking for any help regarding modelling these joints without contact. Is it possible? Or any suggestions that might help? One more thing that the bottom section is fixed and the top is free. I would like to verify my model against the first six modes of frequency, but in my case, as i described i just get O frequency meaning rigid body modes. I have tried to add a spring in the circular direction but that doesnt help much as well. The result is the same rigid body modes. I have provided springs at 200 mm intervals along the circumference.

 

[FEA way]

You could model the bolts using 1D (e.g. beam or connector) elements attached to the joined parts via coupling constraints. There are even plug-ins for Abaqus that make it easier to define this.

 

[Kanwarosama]

Thank you for your quick response. I have tried using cartesian connectors along the circumference of the sections. First I created a set of attachment points then I used tie to fix them to the tower sections on each ends. Then i used a very high stiffness of the cartesian in three directions. But still it didnt work.
Also, I have tried to find the definition of beam connector. It says failure or integration. But what does this mean?

 

[SWComposites]

The aim being to understand the joint effect on whole tower. Why? Properly designed joints will have no overall effect on the tower.

If the tower is fixed at the base it won't have any rigid body modes. You would need to remove the bottom fixity and run a frequency analysis. And what are you going to "verify" against? some sort of hand calculations?

Please show some pictures of which tower joints you are referring to.

 

[Kanwarosama]

Hi,
The tower model will be verified by the modes of frequency provided by the manufacturer. When I used a tie constraint between the tower section, i got very close to the actual modes.
So, my assumption is that if i use a connector element which behaves like a bolt (as this is the actual case), I should be able to reach the same frequency modes.
The idea is to prevent the relative motion of the tower sections and make them work like one single body. I am not modelling the falnges or the bolts, as this will make my model too complex. I will attach a picture of the actual bolt connection.

 

[Kanwarosama]

Here I will attach a spring model assembly between two sections.

 

[Kanwarosama]

This is what the whole model looks like. I think the fix boundary condition cannot be actually removed as this is considered that the foundation provides a complete restraint to all degrees of freedom.

 

[Kanwarosama]

The modes of frequency that I get from using cartesian and springs are almost same. First three as rigid body and afterwards some values which dont seem right.

 

[FEA way]

The simplified approach with tie constraints seems to be reasonable here (especially if the results are good) but if you want to model the bolts then you could use mesh-independent fasteners (easy to define) or such 1D bolt models connecting the holes as described above. Check the article "Modeling Bolted Connections with Abaqus FEA" on the Technia Simulation blog to learn how to do it.

 

[SWComposites]

Ok, re the joints shown in the first picture - those have a lot of bolts and should be torqued to a high level providing a high clamping force. So there should be very little (if any) flexibility in those joints; a spring representation that does not include the initial joint clamp-up stiffness will be way too flexible. I would not waste any time trying to model the fasteners.

Re the frequency results, you should (must) also look at the eigenvector (mode shape displacements). Its too hard to sort out just looking at the eigenfrequency values.

If you are getting very close to the manufacturer's modes, and it would help to see the numerical comparisons, then there could be other subtle differences, such as material properties, as built thickness variations, etc. Does the manufacturer provide the mode shapes?