Quantcast
Channel: PTC Community: Message List - Analysis and Simulation
Viewing all 4134 articles
Browse latest View live
↧

Mesh Question

May 7, 2015, 12:50 am
$
0
0

Hi,

Looking for some advice or clarification that I understand how Mechanica works with respect to meshing.I have created a simple test part shown below and gone through many of the mesh settings to understand the influence they have. Note I am using Wildfire 4.0.

 

My question is, is the default mesh shown below on the top sufficient?

I understand Mechanica uses the p-element method which means a mesh can be coarse and results are obtained by increasing the order of equations until convergence is achieved, so I think this mesh is OK?

 

I am working with some people who used to different programs and they desire that I create a more uniform mesh - kind of like I have created in the bottom picture by limiting the max element size. But is this approach good practice with mechanica? I guess it would significantly increase the analysis time?

 

If anyone can give some advice and let me know if I am understanding this correctly it would be much appreciated.

Capture1.PNGCapture2.PNG

Thanks

↧
Search

Re: Mesh Question

May 7, 2015, 7:38 am
$
0
0

Hi s,

 

The best way to answer your question "is it sufficient" is to run an analysis using Multi-Pass Adaptive, and look at the convergence graph.  If it levels off smoothly before reaching 9th order, the mesh is sufficient; if it's still climbing (or oscillating) then a finer mesh is needed.

 

Using a maximum element size as you've shown is rather a blunt tool.  I much prefer to use Edge Length by Curvature (which can be applied to the whole part, or to selected geometry) which varies the size in proportion to the geometry radius.  The default value of 2 is pretty good; 1.5 or 1.0 is usually overkill.  This produces a more efficient mesh, with smaller elements only where they're needed.

 

In general, Mechanica is pretty fast to run so I tend to use SPA but make the mesh better than I think is necessary - so an ELC of 1.5 or lower in the areas I'm concerned about, or globally if it's a simple model.

 

If your H-element colleagues complain about the rapid mesh transitions Mechanica can use, just tell them that you'll use small elements everywhere if they'll run their models using 9th-order elements!

 

HTH!

↧
↧

Re: Mesh Question

May 7, 2015, 7:47 am
$
0
0

It looks like your model is not that big, so run an analysis for each mesh, and compare the results. Then show the results to your colleagues. And post a copy here

↧

Apply Moment, TLAP Menu not available

May 7, 2015, 8:28 am
$
0
0


I´m trying to apply moment to a part but the menu to select "total load at point" is not available. I checked it with my colleagues where it works but on my workstation/installation/config it is not shown.

Any idea?

 

Problem solved

FEM Mode was selected

↧

Re: Mesh Question

May 7, 2015, 10:52 pm
$
0
0

Thanks for the response guys. I'll pay close attention to the convergence graph in my analysis.

 

Appreciate the 'Edge Length by Curvature' advice also.

 

I'm going to be doing this on a much much more complex assembly so will put it into practice then. With this simple model computation time is so quick that its hard to see effects of different approaches yet.

 

For an unconstrained modal analysis I got the following results. They have very similar mode frequencies (as expected) and the convergence seems better?

 

delta.PNG

↧
↧

Re: Mesh Question

May 8, 2015, 4:51 am
$
0
0

As Jonathand mentioned, run a MPA-analysis, create measures of the quantities you are interested in, and see how those converges as the P-level is increased. If a mesh is "sufficiently dense" is difficult to say; every model is unique. Beware of singularities; you have some sharp reentrant corners in your model that may not converge.

 

The default mesh settings are, so I've heard, is a compromise between solution acccuracy, meshing time, and solution time. My experience is that for geometries that are very simple, like a square, or a square block, the mesh becomes too coarse. If the geometry is more complex, then the mesh becomes finer automatically. 

 

/Mats L/

↧

Re: Mesh Question

May 8, 2015, 5:19 am
$
0
0

In that case also demand of your colleagues that they do the same, so decrease mesh size and compare results for each mesh size. If the results no longer change it shows that the mesh size is okay. (This is what all the text books say one should do, but in practice I've rarely seen this done).

↧

COMPARING RESULTS OF DEFLECTION IN CREO VS HYPERMESH

May 10, 2015, 1:11 am
$
0
0

I have tried to do a chassis analysis in Creo Simulate using beam idealization and same in hypermesh but when I am comparing the results both show different value for the same loading scenario.can any one suggest why this of problem exist.

↧
Search

Re: Representation of curves in Simulate

May 10, 2015, 11:04 am
$
0
0

Hello,

It's justa documentof apractical example forthe representation ofdeformationcurvescompared toSimulate.

Thisis achieved witha free,RDM6

Sincerely.

Denis.

↧
↧

Re: Mesh Question

May 10, 2015, 11:46 pm
$
0
0

For modal analysis the mesh size is much less of an issue than for stress analysis. Basically you only need to get the stiffness and mass distribution right, so for most parts (in H-elements) any 2nd order mesh will do.

 

I suspect you will see larger differences in stress analysis (if you will be doing any?)

↧

Re: COMPARING RESULTS OF DEFLECTION IN CREO VS HYPERMESH

May 10, 2015, 11:51 pm
$
0
0

This is impossible to answer without more information.

 

Tell us more about the way you modelled the chassis and the loading you applied?

↧

Atlas Stress Strain Curves

May 11, 2015, 5:36 am
$
0
0

Hello,

In responseto a requeston mattersalreadywithan integratedstresscurvein the definition, which does not seem to exist, I found a bookwith somealreadypredefined curves.

Iencloseda copyof the bookif it willhelp.

Sincerely.

Denis.

↧

Re: COMPARING RESULTS OF DEFLECTION IN CREO VS HYPERMESH

May 11, 2015, 9:57 am
$
0
0

CREO SETUP

I modelled the chassis wireframe and used beam idealization for assigning material and geometry of pipe to the wireframe in Creo and constrained all the four rear suspension pickup points in all 6DOF's.Then I applied vertically upwards force on Front right suspension points and downward force on front left suspension points and then I ran the analysis.

Same I did in hypermesh using same material properties .

If you have hyperworks 12.0 and creo 3.0 i can send you the files and you can look into it.

↧
↧

Re: COMPARING RESULTS OF DEFLECTION IN CREO VS HYPERMESH

May 11, 2015, 11:37 pm
$
0
0

Yes if the files are not confidential please share them.

I'm on Creo2 and Hypermesh 13. Not sure if Creo2 will read v3 files, but i'll give it a try.

↧

Re: Atlas Stress Strain Curves

May 12, 2015, 2:14 am
↧
Search

Re: COMPARING RESULTS OF DEFLECTION IN CREO VS HYPERMESH

May 12, 2015, 12:12 pm
↧

Re: COMPARING RESULTS OF DEFLECTION IN CREO VS HYPERMESH

May 12, 2015, 10:32 pm
$
0
0

Sorry, I can't get to Dropbox on my workplace

↧
↧

How to fix degenerate boundary faces in Simulate

May 13, 2015, 4:52 am
$
0
0

Hallo,

 

anyone knows what is this kind of problem and what is it due to? (here below the screenshot)

 

6.png

 

how could I fix it?

 

thanks

bye

↧

Re: Mesh Question

May 13, 2015, 4:58 am
$
0
0

sjohnstone,


Creo Simulate is based on the p-version of the finite element method, in which the fidelity of the finite element model is increased by raising the polynomial orders of the finite element basis functions. This approach is in contrast to traditional h-version finite element programs, in which the fidelity of the model is increased by adding additional elements.  Also, Creo Simulate has convergence algorithms that automatically increase the polynomial orders so that accurate solutions are efficiently found without requiring manual intervention from the user. The efficacy of this approach is the reason why the same result was obtained for the two PTC problems attached.  I've attached two Word documents to clarify these methods and to compare results with an h-element solver.  Please look at the document where three different FEA programs were used, ANSYS, StressCheck and Mechanica (now Creo Simulate).

↧

Re: Mesh Question

May 13, 2015, 5:47 am
$
0
0

sjohnstone,

 

I ran your test model with my own boundary conditions and mat props, and as you can see in the below images, there is very little change in the solution and convergence.  I made three separate 2D runs, 1) with default automesh, 2) autogem set to 1 for element size, 3) autogem set to .25 for element size.

 

 

Study 1:

study_default_vm_stress.JPG

study_default_vm_conv.JPG

 

Study 2:

 

study_auto_1_vm_stress.JPG

study_auto_1_vm_conv.JPG

 

Study 3:

study_auto_25_vm_stress.JPG

study_auto_25_vm_conv.JPG

 


↧
Viewing all 4134 articles
Browse latest View live
Search
<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>