How to get rid of "Spikes" - inaccurate VM stress calculations in Simulate?

[JS_9824412]

See above screenshot, the spikes of stress calculations among shell elements. Does anyone see this before?! The stresses should be less than 20ksi, however, these spikes are around 2.0e+7 ksi. I am using Simulate 6.

[SweetPeasHub]

Please give us more information. If you can attach a zip file that would help greatly. My only comment on the images is that the elements look strangely shaped.

[JS_9824412]

The image on the left side shows the shell elements of a metal bracket of thickness 0.1". The image on the right side shows the von Mises stress distribution in the latest simulation just now. There is no more "spikes" shown in yesterday's post. 

It never happened before according to my limited memory. 

On yesterday afternoon, when the "spikes" showed up, Simulate popped up a message "Fatal error encountered", then quitted.

Today, this is no error message.

[JS_9824412]

The traceback.log about yesterday's incidence is attached here.

[JS_9824412]

Simulate allowed the Total Effective Mass to be greater than 100%!??

[JS_9824412]

NEVER HAPPENED BEFORE! Numerical errors in 1-sigma von Mises (ksi) calculations are absurd!

Total Effective Mass should be approaching 100%, not 351,854%! Creo Simulate should not allow this happen.

[JXBWk]

Can the mesh (nodes /element ) and the data in the result (file) be for a different mesh? I have seen such issue in some tools as long as there is a node with say the same id then the tool will try to plot something. Is the plot for disp & rms stress ? looks like simulate is plotting rms disp hence the "spikes". You pointed to total MEFFMASS > 100% in another post so clearly something's not right

[JS_9824412]

The plot is for rms stress. 

This is caused by some software bug. Because it can work sometimes, i.e., without changing anything, sometimes Simulate can produce normal results, at some other time it can't.

[pausob · Accepted solution]

I don't have an answer but point out that it is very possible that there are bugs in any software...

However, this thread and your examples do make me remember an old discussion from few years ago.  In it, (in no small part due to how PTC handled it), I developed severe mistrust in Creo Simulate FEA capabilities:

Non-symmetric results for symmetric model and load? 

Disclaimer: I don't have maintenance, so I can't track if the SPR 2258467 has been closed.  I have to say that it seems to me that PTC has gotten better lately, so I'd be curious if there was a resolution to a very basic issue that somehow got past PTC QA and made it to a released product costing $$$

[JXBWk]

Fully agree with you. And possibly compounded by the click-the-button and black-box approach

[TomU]

---

@pausob wrote:

I don't have maintenance, so I can't track if the SPR 2258467 has been closed.  I have to say that it seems to me that PTC has gotten better lately, so I'd be curious if there was a resolution to a very basic issue that somehow got past PTC QA and made it to a released product costing $$$

---

No.

[JS_9824412]

Now they are trying to sell ANSYS to Simulate users. On 9/16/2021, when my Simulate simulation encountered a serious weird issue, ANSYS sales agent called me out of a sky. What a coincidence. Most of my time were spent on creating manufacturing drawings in CREO. How could they know I was using Simulate? Might be just a coincidence. 

If they are focusing on ANSYS sales, there would be less funding and manpower on improving Simulate. Just a guess.

[pausob]

Well, if you want accurate results, then ANSYS might be the way to go.  By the outcome of the SPR 2258467, it seems that PTC feels that they would spend too much effort on fixing Creo Simulate and since it gives "almost right" answers most of the time, that is good enough.

I would not be surprised if there is no one left at PTC (be it product manager, software developer, mathematician) that knows about this closed issue.  And since it was filed away under Creo Elements Direct - Drafting (???), most FEA users won't come to know that they are using sub-standard solution.

But then again, I kind of doubt that Creo Simulate is used in any serious, mission critical type of projects.

[JXBWk]

Very good points.

On the last statement. “… mission critical type of projects”

With the move or incorporation of ANSYS capabilities (I’ll admit I have not read any flyers/marketing glossy brochures on it) how does one know how far one can push the tool? Short of doing huge amount of testing and comparison?

[JS_9824412]

Hopefully NASA can explicitly ban some FEA software tools with sub-standard solutions. Soon they will ban Simulate as an analysis tool used by their contractors.

[pausob]

I'd think any NASA mission-critical designs would have to be validated in a real life test.

Personally, I don't fully trust FEA results because... well, I'm just ignorant about the limitations of computer modeling and always have doubts about high stresses being an artefact of improper mesh element size, eg.

FEA is being used with great success.  I'm just under the impression that FEA experts use software that is just better.

Anyway, I thought ANSYS was already integrated into Creo somehow - for the PTC simulate "live" technology? So maybe the inherent flaw I pointed out in the other thread is simply gone and this discussion which basically involves me ranting about quality of PTC software is moot.

But I'd be curious to know what FEA experts think of latest Creo Simulate?

[skunks]

for amusement device: https://www.youtube.com/watch?v=Fllzv_mkRwQ

Mechanism Dynamics Option, Simulate and Fatigue (Creo 2.0 M080)

I don't have a subscription

[pausob]

Thanks for the example and an amusing break to my day @skunks ! 

So what do you think of that whole SPR 2258467 affair chronicled in the thread Non-symmetric results for symmetric model and load? ?

I guess I still can't get over how PTC handled this, despite clear demonstration of the issue and how other software (Solidworks, Abaqus) didn't require any "mesh refinement" mentioned as a work-around.  Also, it seemed that some knowledgeable users pointed out the root cause.  Yet PTC didn't seem to fix it.  My impression: they don't care enough, or they can't.  Either way, kills my confidence.  It's like using a spreadsheet application and accepting that, with value in cell A1 = 2.000 and one in A2 = 2.000, then formula "=A1+A2" in cell B1 will, in some cases, display 4.010

[skunks]

I think: PTC will direction toward Ansys

[skunks]

"...So what do you think of that whole SPR 2258467 affair chronicled in the thread Non-symmetric results for symmetric model and load? ?..."

example att.

[skunks]

creo simulate for a another new coaster kart

[skunks]

creo simulate for a new amusement device: coaster kart

https://www.youtube.com/watch?v=gW3mF5hDND8

[rrabe]

A late feedback...

Well, the topic is really drifting from the original question but still connected. If the issue is a software bug the question is if PTC will fix it.

The answer is most probably no. PTC is not investing anymore in Creo Simulate and we should not wait for new features or quick corrections. Big clients with important simulation bugs may have it fixed but the PTC simulation focus now is on selling the integrated Ansys solution.

If this is so, it should be time to have a clear path and we should start to accept more and more the fate. I like very much the integration between Simulate and Creo, to be able to make CAD modifications and simulate it straightforward. I use a lot of optimization and nonlinear analysis (currently with nonlinear materials). Can I have this with the current Creo-Ansys package?

If you move to the current Creo-Ansys solution, will you not get the full Simulate (advanced) package! You would need to go to the full Ansys (nonintegrated) package – and pay extra for it!

The following articles can give you an overview about what is currently included.

CS331359 (Frequently Asked Questions about Creo Ansys Simulation)

CS349429 (What is the difference of the Creo CAE product modules Creo Simulation, Creo Simulation Live and Creo ANSYS Simulation?)

[skunks]

"...PTC is not investing anymore in Creo Simulate and we should not wait for new features or quick corrections.
Big clients with important simulation bugs may have it fixed but the PTC simulation focus now is on selling the integrated Ansys solution."

yes, the Ansys community is large and young
but simulate (rasna) is old, without a future

[rrabe]

Well, just now I am still shaping the future with Creo Simulate – it may take some time until extinction 😊

As a R&D engineer I use FEM to develop and validate my designs. I am not a FEM specialist or living doing FEM for others.

I would not mind to learn another FEM software, but based on the available material about the Creo-Ansys integration I still feel it as an intermediate solution. It is ok for some daily FEM work but still lacks the advanced features found in the Simulate full package. At the end it is not part of the “Ansys community” and also not being co-developed by PTC as quick as it should be. In the future will be hard to think why I would pay extra (and spend time learning) for an OEM Ansys version if what I need is only available in the original Ansys only…

[JS_9824412]

I have shifted to using SolidWorks Simulation by managerial order two months before. I have been wondering why the simulation time to finish a run is so greatly shorter than using CREO Simulate. Does SolidWorks cut corner? However, I have not yet figure out how to run batch file in SolidWorks Simulation. I miss the batch runs in Simulate.

[JXBWk]

are you comparing like-for-like? 1st things that springs to mind: h- vs. p-element

[JS_9824412]

Managers want to see results from me ASAP. I did what ever I could to "cut corner" such as coarse mesh, quick check run. In general, I feel the results can be delivered sooner in using SW simulation. I am also doubting about it.

[rrabe]

Not easy answer. Some Simulate features will not take full profit of quick machines for example due to a non-updated algorithm. In his case other FEMs can be indeed quicker.

But we can make Simulate quicker – if you cut some corners. There are some quick options to try depending the required accuracy.

Is the “simplified” solution the default by SolidWorks? You see, if we do not know what to tune in each software it hard to say. On the other hand you can simply benchmark your simulation and verify the similarities of both results x simulation time.

[JS_9824412]

The reality is managers are all satisfied with the beauty of result plots. They have not yet been trained to understand what has been missed in the models. It takes time and training to finally know what they don't know, like accuracy.