Analysis of a Cylindrical Tube with Explicit

[Biglaugh]

Hi All,
Recently I just started to work on a really simple analysis regarding a cylindrical tube with Explicit. What I have is a Tube with the inner diameter of 14mm (as shown in Fig. 1),

http://files.engineering.com/getfile.aspx?folder=81f9504a-c04e-44fa-92bd-84c81b0cc98e&file=Help1.png

and would like to expand it to 26mm in about 25 seconds. Now the analysis can be completed, but the deformed results is unacceptable. As shown in Fig. 2

http://files.engineering.com/getfile.aspx?folder=476371a8-7296-4d25-866a-de24b26c606d&file=Help2.png

I really don't know what went wrong here, and I believe I have encountered similar problems before. So I wonder if any people here have experienced this before, or is this a quite common problem that there is even a term to call it.
Thanks a lot.

Also the input file is attached if anyone doesn't mind taking a look.

http://files.engineering.com/getfil...3-8664-450d9d141e8c&file=Elastic_Explicit.zip

 

[corus]

Try it without the reduced integration element, ie. C3D8

The manager

 

[Biglaugh]

Are you suggesting this might be caused by hourglassing? I thought as long as I have 4 or more elements along thickness direction, hourglassing won't be a problem. Am I right about this?
Thanks,

 

[rstupplebeen]

Even with C3D8 elements this model eventually fails. LE gets to 0.6 which is way outside what Young's modulus is accurate to. You need to model this with a hyperelastic material model. I hope this helps.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[Biglaugh]

@rstupplebeen: Actually, Hyperelastic material is the material I would like to try after I have the pure elasticity work. Now I am a little confused now. If the accuracy of the material definition is a problem, I would think it may only bring inaccuracy to the result, but it should at least finish the computation without craziness.

Thanks for you reply!

PS: Any people know how to reply to one specific reply in this forum? I cannot find a proper "reply" button......

 

[iamlinkedin]

Biglaugh, how did you apply the loading and boundary conditions to your cylinder? I am trying to model a similar problem but cannot get past the first increment in my analysis, "Too many attempts for this increment". I am however using an incompressible hyperelastic material for my cylinder. I have tried fixing one end of the cylinder in the axial direction and applying an internal pressure, but it does not follow through. Is this the correct way to do it?

Mr Stupplebeen, if you could share some light on this as well, it will be appreciated.

 

[Biglaugh]

iamlinkedin, apparently you are using Standard instead of Explicit for your analysis, so I don't think my BCs will be of help to your problem. Actually, I also tried to solve this problem with Standard before, and it ran OK for pure elastic material, but encountered the same problem as you when I changed it to Hyperelastic. So my suggest would be try to run it in pure elasticity and see if it works, and then you can determine if something is wrong with your hyperelastic material definition.
I would greatly appreciate your further comment on this problem if you ever figure it out.
Thanks,

 

[corus]

Your results look like 'noise' coupled with hour-glassing. Have you tried using double precision in case of a build up of rounding errors?

The manager

 

[MechIrl]

Like corus says...looks like you may be having kinetic energy problems. I notice from your input file you aren't using any mass scaling. Since you seem to want a quasi-static response make sure your kinetic energy is maintained sufficiently small relative to total internal energy (5% is a general rule of thumb).

 

[corus]

Incidentally, why isn't this run as a 2D axisymmetric cylinder? You could put in a lot more elements and/or reduce the run time considerably.

The manager

 

[Biglaugh]

Thank you all very much for the helpful replying!

I double checked my result, and I believe that what I have is a severe hourglassing problem. I didn't realize it since I didn't check results during the simulation where hourglass began to form. However, even after I switched my element from reduced to full integration, the simulation still aborted. Still working on it.

Corus, thanks a lot for your suggestion.The reason I didn't use 2d Axisymmetric is that this cylinder is only the first part of a contact problem, which is not symmetric when other parts are considered.

MechIrl, I have never thought of the mass scale would be a problem, I will definitively take a look. Thanks,