Convergence of multiple span cable structure

[t.sautierr]

Here is two cable model. One does converge and the other does not.

Sole difference : cable have been divided into several part between supports and pdelta analysis has been set for the second as it needs to be when cable are made of "chains" or part of cables.

I does not understand why the second one does not converge.

Assembling parameters : 

first model : each cable is a little in compression (negative little relative strain)

second model : each cable is a little in compression (the middle part is affected with little negative strain, the other part are free of assembling parameters)

Thanks for your help.

[t.sautierr]

Second model.

[t.sautierr]

Could anyone help?

Thx a lot.

[t.sautierr]

zzzz 😞

[StefanoPasquini6790]

Hi t.sautier,

I toke a look to your model and I found something strange.

First of all, why do you divided all the horizontal bars in al lot of very little bars? I tried to merging it uaing the tool mergingmultiplebars, but after two hours the api was still running and I've stopped.

Why do you divide also the cable into small continous cables? This give you a lot of nodes that create an instability....

My suggestion, for your facade design, is to start with a simple model studying a 3 span portion of the structure, use only two or tree load condition and make your own tests to set up all the settings to let work your analysis. After you ended this phase, you can make a complete model of the facade using all the settings that you have tested in the test model.

Cable structure is very uneasy to manage, take your time to evaluate exactly its behaviour.

I'm sorry for haven't fixed your model, but this will committ all my afternoon, and I'm

[StefanoPasquini6790]

...I'm in a difficult time managing phase...a lot of project in the same time....


I hope to help you more in next posts...

Good luck and "forza robot"

[t.sautierr]

thanks for your points stefano.

First of all, why do you divided all the horizontal bars in al lot of very little bars? I tried to merging it uaing the tool mergingmultiplebars, but after two hours the api was still running and I've stopped.

Why do you divide also the cable into small continous cables? This give you a lot of nodes that create an instability....

I did that because in the same time I'm making my analysis, a wind numerical analysis is performed. Once i will recieve the results, I will need to load the cables with loads that can be different along it. So we seen with the design office in charge of the wind study, that we will not go below bars 1m long (6 parts for 6 m).

Actually if no wind study would have been performed I would have let all cable and horizontal bars in one piece and the model would compute ... 😞

My suggestion, for your facade design, is to start with a simple model studying a 3 span portion of the structure, use only two or tree load condition and make your own tests to set up all the settings to let work your analysis. After you ended this phase, you can make a complete model of the facade using all the settings that you have tested in the test model.

I did, the model converges when the cable are in one piece between columns ... and Pawel tought me how make a cable in several part converge but it seems that with such a number of cable, it doesn't work. However, I have not tried arc-lenght method to make it converge.

[t.sautierr]

On pp2008 hints, I removed the asssemling case with no OW and I increased some section to avoid to much compression and then buckling. The model still contains some type 3 instabilitie, I will make a sum up of what I learned after removing the last instabilities.

[t.sautierr · Accepted solution]

Summary of pp2008 conclusions :

- When using cables, do not suppress OW of cables of the assembling case (eg: at least self-weight of cable shall be included in case n°1), otherwise the software will not be able to calculate the assembling case has the cable is not loaded except the starting force/elongation condition.

- When using cable divide in several part : 2nd and 3rd order analysis (1st order and pdelta) shall be performed in order to obtain results. When using cable in one part, simple non linear analysis can be performed.

- When setting force or elongation in the assembling case combined with the activation of DSC Algoritmn (for modal analysis or for pdelta analysis), instabilities of type 3 can occur if the set value is very small, it disappears when greater value is set (it increases the strain-stress stiffening effect). If a very small value is kept, this can be ignored by desactivating the dsc algorithmn if the results with or without are exactly the same. These instabilities come from the spare element at the end of the cables added by the algorithmn which have a very small stiffness compared to the rest of the structure.

- When splitting calble into several part for one cable, it seems that there is no need to distiguish several conditions for asssembling cases for different parts of one cable (for example setting traction only in the middle part and not on the side parts is useless.

Thanks to french and polish team for theeir help.