(API) macro for highlighting nodes with instabilities

[t.sautierr]

Hi everybody,

please find attached the beginning of a xls macro.

purpose :

to be able to check quickly where are located the portion of model which contains instabilities.

process:

after a calculation,

search for the nodes with instability in log file,

transform the list in an appropriate layout,

make a list with type of instability and highlight it in the model (copy paste the list).

 At this stage, it is for its major part this si hand manipulations, if someone could implement a little API (same spirit than round up / round down) that makes this automatic.

In the attached file, a little guidelines is given (in French, sorry ...but with snapshots!)

What I usuallly to after having run the macro and which is not implemented is :

- copy/paste point list

- add the connected mesh/bars to the selection

-  new window with the selction only.

Thanks a lot Rafal (I guess .. 🙂 )

[Rafal.Gaweda]

 

I have added a pannel in order to see if it is possible (for FEM model) to take a node with instability and to display all related pannels but doens't give me anything.

Yes.

I don't know if it possible as it seems not to be possible to select pannels bar nodes with the classic selection box.

 ????????

Any idea?

Yes it is possible but you have to write the function to find this panel(s) by your own or....

(there is getconnedbars function in API so it can be used directly)

In the example : pannel is linked to node 14 but it doesn't display the pannel with the macro at this stage.

So you send me wrong model :

[t.sautierr]

I see the procedure.

i'll to implement this and come back a new version.

Thx

[Rafal.Gaweda]

I see the procedure.

i'll to implement this and come back a new version.

HINT: loop of all panels , then find Nodes (Nodes method returns string) for each, make selection from it, compare selection with selection made by list of nodes from excel and build selection of panels, then display this selection, turn on display of panels numbers.

20-30 additional lines of code

When can we expect new version 😉  ?

[Rafal.Gaweda]

Update for version 3 - instabilities in calculation nodes also reported.

[t.sautierr]

With a "short" delay ... update 6 -> it highlights also pannels now.

Rafal thx for your help I didn't open the file or let's say it was ajar 🙂

an example of file is attached.

2 questions: as I was testing the macro on the model attached. My preferences for the solver is by default set to multiprocessor (bad thing I know, i'll change this), with this, it always calculate the solution even if the model contains instabilities so the questions :

- what is the criteria that stops the calculation process with this solver? does the results come from a stop in the process or are they a the "true" solution even if inconsistent?

- is the multiprocessor solver able to detect instabilities or will it calculate whatever is the structure?

Thx

[Rafal.Gaweda]

Multithreded solver is not our solver. It is Intel Solver implemented in Robot. So in previous versions it did not detect any instability or reported as instability type 34 or 225 or something, but its main goal is\ was to solve matrices even if they are not well conditioned.

For 2014 they improved solver, we improved its implementation and :

Instabilities in every solver "pause" calculations waiting to confirm what is displayed in warning message window.

So you need IgnoreWarnings = true to run calculations to the end neglecting all instabilities which may appear.

[t.sautierr]

OK Rafal,

this means till RSA2014, for multithreaded solver "a" solution is calculated without any limit on the matrix properties, reactions, displacements, etc .....

Good to know if it is the case.

By cusiosity, if the model is strongly divergent, there shall be some kind of limitation as the algorithm cannot goes to the infinite? 

[Rafal.Gaweda]

I have not looked at your model or macro yet 😞

I meant that (every version) Multithreded solver is designed to calculate any matrices even not well conditioned ;  in Robot understanding with all types of instabilities.

IN earlier multithreded solver versions no instabilities were reported at all, later - strange instability type number (strange from Robot \ Robot user numberation point of view), then in 2014 it seems the instabilities will be \ are reported in "robot way".

[Rafal.Gaweda]

By cusiosity, if the model is strongly divergent, there shall be some kind of limitation as the algorithm cannot goes to the infinite? 

---

Of course there are.

In short

1. converging but slow or diverging but slow  : tolerance parameter is not reached in given maximum number of iterations - then program reduced last increment step by half and tries again - as far as I remember 3-4 times then warning message about divergence is displayed

2. fast divergence - when the different between last and before last ratio is above certain level  then program reduced last increment step by half and tries again - as far as I remember 3-4 times then warning message about divergence is displayed

[tony.ridley]

Rafal this is very good information. Can you seperate out the parts of this post relating to new multi threaded capability and make it a new post?

Thanks, Tony

[rsousa_]

Rafal,
In your excel sheet, you have the following line in vba
RVP.Set I_VDA_STRUCTURE_ONLY_FOR_SELECTED_OBJECTS, True

With this code you select the option "display attributes only for selected objects". When running this code Robot display all node numbers and not only the selected ones, making hard to find the desired node. What can I do to correct this issue?
Note: i'm using RSA 2009

Regards

[Rafal.Gaweda]

I do not have such old robot version.
Display (I_VDA_STRUCTURE_ONLY_FOR_SELECTED_OBJECTS) by API corrected starting from 2010 version.

[rsousa_]

Ok. Maybe that's it. Thanks!