Non-zero Mises stress in ABAQUS simulation!!

[zhang1026]

Hi everyone.

I have a question about ABAQUS simulaiton results.

I am running a trial simulation to test ABAQUS codes and get results that puzzle me a lot.

The FE model consists of 4 parts- 2 cyliners with some distance and 2 shells sandwiching the 2 cylindrs (Pls see attachement for model setup).

The boundary condition is as follows: 2 cylinders are tied respectively with each of the 2 shells.

The loading condition is like this: one shell is fixed while the other shell is moving to stretch the cylinder.

Numerial analysis was done via ABAQUS/Standard.

The result is that, the cylinder tied with fixed shell show zero Mises stress, while the cylinder tied with moving shell dispaly non-zero stress.

What puzzles me is that, both the two cylinders should show zero mises stress (because the two cylinders are separated such that no deformation was undergone by the cylinders) but actually not.

Could anyone help to explain why and how non-zero stress come out?

Thanks in advance.


Regards,
Bao

 

[DanStro]

That looks like an artifact of the numerical approach. I am not sure why it would not be in both cylinders but 10^-16 is small enough to consider zero in any case that I've come across.

Han primo incensus

 

[zhang1026]

Hi DanStro.

Thank you for your reply.

In that model, the cylinder properties are arbitrarily defined and have no physical meaning, so the stress values are not 'true' but reveal that the stress deviates from the theoretical value - zero.

I have confirmed that stress connected with fixed shell display zero stress everywhere. But if this shell is allowed to move, then the stress in the cylinder connected is no longer zero. It seems that the stress state in the cylinder is affected by the kinetic energy of the shell with which it is connected. However, this does not accord with physics.

Have you realized this 'strange thing' before? Is it because the ABAQUS codes are not reliable?

I have used ABAQUS for some time and this is my first time to meet this, but I can not explain how and why it happened.

Hope to hear more and more views on this.

Have a good weekend, everyone.


Regards,
Bao

 

[ESPcomposites]

I tend to agree with DanStro. If the value were 1.0E-30 would still be worry about it? At what point do you say the result is numerically equivalent to zero? 1E-16 is zero for all practical purposes. FEM solutions are numerically "perfect" for only the simplest of problems.

Brian

http://www.espcomposites.com

 

[zhang1026]

Hi ESOcomposites.

I totally agree with you and DanStro that the stess value in order of 1E-16 is negligible practically.

The problem I have is that I can not figure out why and how the non-zero stress come out from theoretical viewpoint.

My limited knowledge on ABAQUS prevents me from giving a reasonable answer.

Could you shed a light on this? Thanks in advance.


Regards,
Bao

 

[corus]

With numerical methods and matrix manipulation, you won't get zero as there will be rounding errors. If you don't believe the results then put in real material properties.

 

[zhang1026]

Hi corus.

I put in real properties and the conclusion is unchanged. The cylinder connected with fixed shell display zero stress, while cylinder conected with moving shell show non-zero stress.

how come the strss state relate to kinetic energy?

Is it because the stress wave was genenated at the interface between shell and cylider and then propogating across the cylindr in the direction normal to shell, i.e. the direction of shell and cylinder moving??

 

[zhang1026]

Hi everyone. I found that the stress in the moving cylinder increases with its stiffness and the moving velocity .

It appears that the interia effect should be the reason for non-zero stress state in the cylinder. This means that the cylinder is not in equilibirum internally and zero-stress only occur when stress is distributed uniformly.


Anyone have other thoughts on this?

I realize i am weak in theroy and therefore can not explain the non-zero stress in the moving cylinder.

Look forward to more ideas.


Regards,
Bao