Result comparison

I have calculated the same construction in PPM and creo simulate
I have noticed that displacement are different
PPM ca 240 mm and creo simulate ca 85 mm (3 times bigger more or less)
bonduary conditions material and loads are identical in both calculations.
I have no idea why such big difference , stress level more less ok (the same )
best regards

You have to compare each and every aspect of the analysis in both programs - mesh density, element type, what you’ve already listed (maybe double-check to make sure that there are no mistakes in BCs, loads or material definitions), analysis procedure and solver settings (e.g. Nlgeom), connections (constraints, contact) and so on.

If only displacements are not OK, check the Youngs modulus. If this is the same then the boundary condition should be checked. Check both animations.

2 Likes

You can perform a fast free-free modal analysis to compare. No loads, no BC. Just contacts in case you have set some. Mostly free to vibrate. If natural frequencies are not the same without any BC and loads, the difference could be the element type / mesh density / material properties or contacts set up. If they are the same, then you can proceed to check the BC and loads. Seems too much difference to sleep well.

2 Likes

Good point !!!..

after checking model and recalculation (error was in one force -





wrong direction (+/-) displacement are more or less ok but next problem apperaed difference in sigma XX or in PPM sigma 11. Could anybody explain me why such difference , I have made reference model (significantly smaller size ) and result in this model was ok

It is hard to say without more detail, but it might be the boundary condition. Are you using a rotating boundary condition and how did you model it?

I used rigid body connection between point and named node set and I used dispacement rotation BC ,
for one BC I have used *transform by calculix keywords to obtain inclined support

HI Mishal,

I’m happy you keep going and sorting main differences.

Numbers commonly refers to Local coordinate systems and letters to global. That maybe would mean your S11 is stress in the local direction and SXX is Stress in Global X direction. They may not coincide if you have set up a local coordinate system.

Check principal Stresses S1, S2 and S3 first.

What do you mean by smaller size ? What were the exact differences between that reference model and your actual model under consideration ?

As you can see on attached photos difference is between supports in compression /axial stresses

Best regards

image001.jpg

Neighter Sxx or S11 are invariants.

¿Are your sure Direction 1 is aligned with the Global coordinate system X?

You comment about a *transform card.

¿How could we visualize in Prepomax where the local direction 1 is pointing?

Did you check the other components of the stress tensor as well ? Maybe there are some important relations between them (like swapped directions and so on).

Someone need to remove transform feature at first model to review, be sure using exact the same geometry, material and boundary conditions. Also, mesh density and similarity in element type, quadratic is recommended. Then review the results in resultant deflection and Mises stress.

If all are okay, so probably the transform is not properly assigned.

I solved many variants with option transform and without -reaction as Rax fully horizontal from that time and difference is too big Results from Creo Simulate are much closer to reality-hand calculations

I did not found any sketch of problems (dimension, section shape, load and boundary condition) so make it hard to trace

if there’s any report from analytical and Creo Simulate results also CAD files provided, maybe someone can try to reproduce the problem and clarify.

But i’m still in doubt about your conclusion regarding CalculiX is so far from açcurate, probably there’s something wrong in the model

probably with model is something wrong but it is hard to find it
best regards

Creo Simulate could be wrong.

¿What if you request your results in global coordinates. It should be independent of any previous Transform card isn’t it. If I’m not wrong that’s done with:

*EL FILE,GLOBAL=YES
S

By other way I would first compare principal stresses. If they are not the same the stress tensor components will not neither.

For *EL FILE, the global coordinate system is used by default unless *ORIENTATION or shell elements are used.

For *NODE FILE, the global coordinate system is used by default unless *TRANSFORM is used.

problem is that only on one segment of lenght of structure stresses are different the rest is ok . so I think that problem is with model but I’m unable to solve it (as you can see on previous photos/screens