bandwidth warning

[sancat]

I would like to know if there is a method to chase the cause of a "max/min bandwidth too high warning".
I have found that sometimes it is due to elements that are very close to each other, but not in contact from the program point of view. I have also found that some lines used to construct the model remain and form host elements that also cause the problem.
I turn off part of the model or erase them, and solve over and over. While this method works, it takes forever to find the lost sheep.
Is there an alternate method? Or the model has just to be built right from the beginning?
By the way, the program I had used is Algor.

 

[vonlueke]

sancat: For nodes that are too close together, in most programs this is resolved by performing what is called a "Coincident Node Check." When you execute this feature, it finds and highlights all nodes in your model that are too close together, then asks you whether or not you want it to go ahead and merge these nearly-coincident nodes and delete the resultant disconnected nodes. Look for this feature in the meshing application you are using, or its Help index. Nearly-coincident nodes create very bad elements and need to be eradicated (merged).

However, this function permanently changes your model, so execute it on a copy of your model unless you are certain it cannot damage your FEM. Good luck.

 

[EdR]

sancat:

Also you may be able to get around this warning by redefining some nodes and elements....Generally what this type of warning refers to is the difference in node numbers that are connected to an element(which translates directly to degrees of freedom). Thus an element with node #'s of 10 15 & 20 has a nodal bandwidth of 10 (20-10) while an element with node numbers of 20 70 & 30 has a BW of 50 (70 -20). The only result of such a warning is that the solution may take considerably longer with a higher bandwidth (assuming that it will solve). If there is an option to use a bandwidth minimizer that may also reduce the bandwidth..

Ed.R.

 

[sancat]

Thanks for your answers.
I think I forgot to write the main issue: max/min stiffness ratio is too high: I understand that some parts of my model are too stiff compared to others, and the warning is triggered with differences above 1.0Exp08, that is the resolution of the mathematical solver.
What I am looking is for a way to know easily where are the weaker parts of the structure; these parts are often due to loose elements, and cause my headaches. I have tried to block each degree of liberty at the time, and solve again the problem, but a graphical aproach will be the best.
Thanks again
sancat

 

[pja]

Do a modal analysis..you can look at the mode shape to see
where the part isn't constrained properly.

 

[sancat]

pja,
thanks, I'll try that
sancat