Releases to the edges of panels

[PaulGrimes2282]

I am following the ideas in this post:

http://forums.autodesk.com/t5/Autodesk-Robot-Structural/Rigid-or-semi-rigid-diaphragms/td-p/3198174 (post 5)

and i don't know quite how to release the edges of the slabs correctly to get diaphragm action with no composite action. I have tried to release moment perpendicular tot he edge and shear parallel to the edge (i.e. Ux and Rx linear releases), but it is giving me instability errors in the nodes connecting the panel to the bar at the edge.

As a seperate issue (i think) I also get type 3 instability errors if i try to make my panels assymettric (i.e. if i try to reduce the material stiffness coeficient n2 below 0.1)

I hae attached the model - would someone be able to have a look and see what i am doing wrong?

thanks!

[Pawel.Pulak]

The linear shear releases you have defined on the edges of panel are not necessary and efficient because the diaphragm in X and Y defined for this panel will prevent from any in-plane shear in finite elements of the panel (shear will be transferred by the diaphragm). It only results in additional instabilities, It seems that generally FE mesh on this panel is not necessary because even the load is transferred from it via tributary area (one-way distribution) directly to beams 1 8 698.

That is why I would recommend changing the panel calculation model as below:

After this change the type 3 instability in node 1655 in RY is reported. When checking the neighborhood of this node it can be noticed tha all bars are connected to beam 708 via xxxxff releases (local RY and RZ rotations free) - it means that this beam can rotate about its longitudinal axis.

After fixing it by modifying release at one end of the adjacent beam 899 and running analysis again the type 3 instability in node 1835 in RY is reported suggesting some problems with beam 1017. In this case more detailed check has confirmed that it is only numerical instability resulting from small stiffness and it can be ignored.

Such "more detailed check" may contain applying temporary MX moment in node 1835, running analysis and checking displacements and rotations (sorting and/or "Global extremes" tab in the table of displacements).

Attached the model modified in above way.

---------------------------------------------
If this post answers your question please click the "Accept as Solution" button. It will help everyone to find answer more quickly! 

Regards,

[PaulGrimes2282]

Hi,

thank you for your response.

I have a question about investigating type 3 instabilities and ignoring them if you can explain them - in some models taht i have created in the past the type 3 instability seems to mask other instabilities that may occur - i.e. once i have found an corrected the error creating the type 3 instability then other instabilities are reported on the next run. Therefore is it not dangerous to leave Type 3 instabilities even if you are happy with the cause?

[Artur.Kosakowski]

Starting from Robot 2014 all detected instabilities are reported in the dialog that is displayed after calculations so that you are aware of them without checking the log file.

[PaulGrimes2282]

but by checking the log file in 2012 it should still report them all?

[Artur.Kosakowski]

Correct provided you allow the calculation process to finish.

[PaulGrimes2282]

excellent - thank you!

[Pawel.Pulak]

Additional info to provided by Artur:

All instabilities are reported in log (and in 2014 release in the dialog) in case of type 1 and type 2 instabilities, which are related to zeros on the diagonal of stiffness matrix.

It is not a case for type 3 instabilities, which are related to big ratio between max and min stiffness on the diagonal. In such case it is possible that after correcting the reason of one instability and repeating analysis another instability of type 3 will be reported. That is why it is important that if one source of instability is identified and corrected it may be useful to reflect before repeating analysis whether similar situation is not observed in many locations in the model.

---------------------------------------------
If this post answers your question please click the "Accept as Solution" button. It will help everyone to find answer more quickly!

Regards,

[PaulGrimes2282]

So if there is a type 3 stability that you investigate and accept (as referred to earlier in this post), this may be masking other type 3 instabilities? These may be caused by something completely different, and so picking up one may not lead you to pick up another.

[Pawel.Pulak]

Yes, it is possible:(

Fortunately type 3 instabilities related to modeling errors are reported first than type 3 instabilities related to model properties (because these first ones are related to bigger max/min stiffness ratio).

So if correcting the model you reach type 3 instability related to model properties there is very little chance of modeling errors related to instabilities still existing in the model:)

---------------------------------------------
If this post answers your question please click the "Accept as Solution" button. It will help everyone to find answer more quickly!

Regards,