Continue to Site

Welcome to MCAD Central

Join our MCAD Central community forums, the largest resource for MCAD (Mechanical Computer-Aided Design) professionals, including files, forums, jobs, articles, calendar, and more.

pro/mechanica optimization problem

gasser3000

New member
hi,


When i perform design study (optimization), i want proM to generate NEW- elements with each shape update ... INSTEAD of trying to Fix Invalid Elemenrs ... which cause warning message ....... i checkes the box in options that ask to remesh with each shape update ... but PROM not doing that ...?? anyone has an idea what is going on ??!! Thanks
 
Don't forget that Mechanica is purely based on p-adaptivity (higher order elements) and you need to do something else instead of creating new elements. Also it works only in elastic range.


Israr
 
i know, but sometimes the p-adaptivity does not work well, if geometry shape change caused some elements to be invalid. I guess i have to increase number of elements. Did you try to generate new elements with each iteration before??
 
I think you are running independent Mechanica - if this is true and you can run integrated mode, try this mode and check the option "Create Elements during run". I know it sounds the same as the one in independent mode, but this one does work. I remember an issue with the independent mode option and it's communication link back to Pro/E not allowing clean mesh updates. There is, or used to be, an environment variable you can set to force Mechanica to remesh all elements before solving. Try "AV_REMESH_ALL" set to "YES" and see if this will do the trick for independent mode.


Kaz Z06
 
Hi,


Well, i work with integrated mode. and i chose the option you mention but did not work ???? the command u mentioned is not avilable anymore PROE 3. I do not understand why they put an option that does not work ???!!!!!!!
 
Are you running the optimization from BMX or from inside Mechanica? I know the re-mesh option works when using BMX to drive the optimization running WF3. I have not used the optimization facility inside Mechanica forthe lastfew versions. If you are running an optimization from BMX and cannot get it to work as you stated, you need to call PTC's technical help desk and get the issue resolved by fowarding your model to them. Good luck.


Kaz Z06
 
hi,


Well, i well try once you tell me what is "BMX" stands for ??!!
smiley1.gif
. thanks for your time.


Gasser
 
BMX is "Behavioral Modeling Extension". It is a module thatallows you to run sensitivity, optimization/feasibilty and DOE studies from inside Pro/E on anything that can be a measure (length, area, radius, mass, etc.). It works with Mechanisms and Mechanica as well.


Kaz Z06
 
Behavioral Modeling eXtension


BMX lets you optimize based on things other than stress / deflection results from Mechanica. Optimize things like size / distance,total mass, center of gravity, volume, etc.


Does your optimization cause drastic changes in the shape of your model? I have had experiences where drastic changes that require a complete re-mesh will fail to optimize. In these cases, I break the study into parts that can regenerate using the same mesh. So I end up with more than one study to obtain my complete answer.
 
Well, my optimization problem depends on stress and deflections, so if BMX can not include that, it is of no use to me at the moment.


I can remesh during optimiztion using ANSYS Workbench, but ANSYS W uses h-elements and not p-elements, so Pro/M gives better results. So, my geometry has no problem to remesh during geometry update.


So, i need to use Pro/M instead for optimization and the p-element results which is more accurate ..... so guys i am still do not know how to Re-mesh in Pro/M during design study .... Even with the "remesh with geometry update ..." option checked !!!
smiley5.gif
....... i even tried a simple problem ... and yet it does not re-mesh !!
 
gasser, there is debate as to whether p-element meshes provide more accurate results than h-element ones -- or vice versa. If you alreadyhave results from ANSYS, are you concerned that they are not accurate enough?


As I understand it, the "Create Elements during run" option is strictly for determining whether Mechanica uses an existing mesh file for the analysis or creates a newmesh when you start it. Either way, it still usesa single set ofmeshelements for the entire optimization.It is probablythatyour variable dims are changing the design in a discontinuous manner (causing some p-elements to become invalid), and autoGEM is having trouble with it. Do you have large ranges for your variable dims? Do the dim ranges span gaps in the geometry? If so, you may have to reduce the range of your variable dims.


Do you have a significant number of variable dims? Do all possible combinations of your variables yield a feasible model inside proe?Iassume you havedone a shape-animate to ensure that the model is not failing to regen.
 
I did this simulation on ANSYS Workbench for 25 design points and no problem as it re-mesh each time. but i am having problem with no able to re-mesh on Pro/M.


I did the shap animation and no problem. and i tested on simple geometry and still not able to re-mesh !!!!!!!
 
I don't think that you can do what you are trying to do. As I understand it, Pro/Mechanica does not create an entirely new mesh for each step inan optimization. It simply adjusts the mesh to fit the changed geometry. If your variables causefeatures move in such a way as to render elements invalid, the study will fail. You must define your study around this reality. In other words, the variable dims cannot impose drastic changes to the model -- and the changes must not cause existing features to disappear or new ones to emerge.
 
So,what is the option "remesh with each geometry update .." is for ??? !!! i guess it is a backup option in case many invalid elements. but how i do not knoe
smiley5.gif
 
I have run a very simple cantilever model in WF3 M040 that demonstrates remeshing capabilities during a sensitivity study driven by both BMX and from within Mechanica (integrated mode). The study simply changes the number of holes in the part through a Pro/E parameter - obviously requiring a re-mesh before solving properly. I've included the model file as well assome result files to demonstrate its successful run. Each configuration (1 and 2 holes) was solved individually to validate sensitivity study results - all values match. Please pay close attention to the setting dialog box image I included as this must be checked for the volume to be remeshed between sensitivity steps. The two stress images are from independent runs, and thetwo ".txt" files are plot data from the sensitivity analyses by BMX and Mechanica. Report files are also included. If this model does not produce the same resultsfor you, there is an issue with your install, or you may not have BMX as part of your license.

BMX will allow you to run all studies using stress results, or any other measure from Mechanica as well as Pro/E's mechanism capabilities (MDX and MDO).

I hope this helps!

2007-01-31_082558_remesh.zip
 
kaz, I still use wf1 (although I have access to wf2 and occasionally use it). Are you saying that you define the optimization within Mechanica and then select it from within BMX? Or are you saying that you define the Analysis within Wildfire and then define the optimization from within BMX? This is great news if simply running it from BMXallows total re-meshing for every shape change. What are the steps you took to run it?


I have not seen an option: "re-mesh with each geometry update." I have only seen "Create Elements during run" in the analysis run settings dialogue. Perhaps this is new in wf3?
 
rmckinley,


With BMX, you define the Mechanica analysis, as typical, but you then go back to the Pro/E "Standard" mode side to define the study you wish to perform (sensitivity, optimization, DOE)using any measure you created on the Mechanica side. There is much more flexibility using BMX than there is from inside Mechanica.


Cheers
 
well thanks KAZ Z06,


it worked for a simple problem fine. now i guess i well go now for my complected assembly to try it ..:)


i have another question about "total_cost" .... how to relate it to Mass of part??? i can not figure it out ...??
 
Maybe this old setting helps:


You could try the environment variableAV_TOTAL_REMESH and set it to YES or ON (I forgot which one was applicatble in this case). This setting was used to force MECHANICA to remesh between two steps of a sensitivity or optimization study. Point is that this is defined as a system variable so be sure to disable it again after the problem analysis has run or all yoursensitivity studies will be remeshing......
I'm not sure if this setting is stillactive in WF3
 

Sponsor

Back
Top