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.

COSMOSXpress Accuracy?

abm23

New member
Hi All,

I have been running an analysis using COSMOSXpress on a simple sheetmetal part and am finding the results to be confusing. The first configuration of the part I run is just a piece of bent sheetmetal with no strengthening ribs. The second configuration is the same piece of sheetmetal, but with several ribs to add strength. If I'm reading COSMOS' results correctly, it's telling me that the configuration without ribs is by far stronger (less deflection) than the configuration with ribs. The lowest Factor of Safety for the no-rib config is just over 1,000, and the lowest Factor of Safety for the rib config is around 15. This seems absurd. So, either COSMOS doens't work well, or there is some heavy user error going on here.

I have applied SS 304 to each part, fixed the bottom to ground, applied a load of 150lb to the top, and accepted all the default values to run the analysis. The part is about 50 in in diameter, and I simply suppressed the rib features when running the no-rib-configuration analysis. Anyone have ideas why I'm getting such strange results?
 
The higher the safety factor the better. Safety factor is defined as:


permissible stress / actual stress.


However, be very carefull. Cosmosxpress does not allow you to configure the analysis correctly, even for simple cases. The results should not be trusted.


A valid analysis requires the full version of Cosmos, a person the is experienced in using it, and that can interpret the results properly.
Edited by: bjulien99
 
So basically you're saying COSMOSXpress is worthless.

In the case I described above, the problem wasn't so much that the FOS was too high, it's that it was higher (much higher) on the part without ribs than it was on the same part with ribs - exactly the opposite of what I expected and what, as far as I can tell, it should actually be. Is there any conceivable way ribs could weaken a part? And weaken it significantly?
 
mesh quality is important for all analysis.


if you make smaller meshes,are the resultsclose to the previousones withbigger meshes?


if no,the meshesarestill not small enough. Try smaller ones until the results converge tosome certain values.


Use Higher order meshes (more nodes per mesh) can reach convergencemore easily, especially for bending problems.


itis no meaning to draw any conclusion if convergence is not reached.


bigmesh, bigerror.
 
Without seeing the model, my guess is your mesh size is too large in areas of high stress concentration, and therefore not accurate. You need to apply mesh control which I believe is not possible with CosmosXpress, or a very tight mesh over the whole model which will take a long time to solve.

For more information on mesh control, refer to this link: [url]http://gabijack.com/2009/11/troubles-corner-suggestions-to-i mprove-the-study/[/url] Example images (click images to enlarge) are shown of how the mesh should appear for a solid. For sheet metal, a shell mesh should be used instead of a solid. Does CosmosXpress permit the use of shell meshing?

My friend was writing a blog about using simulation (static FEA) in SolidWorks, and I provided her with some suggestions for improving the accuracy of the FEA study. As she is a better writer than me, she improved upon the write-up I had previously sent to her.

Regards,

Chris Thompson
www.appianwaytech.com


Edited by: c_thompson_68
 
For sheetmetal "THIN" models, i suggest you create a surface model (mid-surface of the solid), then create 2d meshes on this surface model. By this way, you can use bigger meshes than that used for a solid model.


Meshes should not compressed/distorted too much before any analysis.
 

Sponsor

Articles From 3DCAD World

Back
Top