Viscoelastic material - Linear or Non-linear?

[xhbg]

Hi, I am using CosmosWorks 2007 SP1 without non-linear study. I would like to know if my model has certain parts whose material is viscoelastic, can I analyse it with linear study? Thank you.
Regards.

 

[cbrn]

Hi,
no, the term "linear" refers to ALL the possible sources of non-linearities, thus including material ones. The only materials you can use are purely-linear elastic ones (Hooke's law).

Regards

 

[xhbg]

Ok, so you meant that I should run the model with non-linear study? And whatever results that I got with linear study are completely wrong???

 

[gfbotha]

Xhbg, as Cbrn pointed out, a viscoelastic material strictly spoken requires a non-linear solver.

However, under certain circumstances one could perform a linear analysis to test drive your FE model or to get some indication of stress & strain. In fact you could get quite useful results IF your material input values were obtained at the relevant temperature and strain rate, and the strain levels are sufficiently low. Some viscoelastic materials exhibit near linear behaviour at low strain levels (say < 3%). Even if you only have viscoelastic properties (relaxation modulus) available, you might read off an effective modulus and perform a first order linear analysis.

Beware that if your material has a Poisson's ratio very near to 0.5 you could potentially run into serious accuracy/solving problems when using a standard linear material/element formulation. You also need to be aware that a linear study also excludes non-linear geometrical effects (like considering thinning or thickening of cross sections). Normally deformation results will be more useful than stress results.

One advantage of a linear analysis is that it typically runs much quicker than a non-linear study. Sometimes a non-linear study might not even solve. All depends on what you actually want to achieve, experience, interpretation and engineering judgement.

Gert

 

[xhbg]

Thank you cbrn and gfbotha,

I have tried different elastic modulus values (within the range of viscoelastic material) and it shows the difference which is smaller URES (displacement) result. However, the result seems to be still far from reality. So I really don't know if I should proceed with linear, how to get a reliable result or if I should use non-linear analysis, which I hope not because of its time-consuming and also, as what you mentioned, sometimes it can't even solve....

Please suggest me a way...Thank you so much indeed....

Regards,

xhbg

 

[cbrn]

Hi,
at a first sight it could seem that your material is complex enough so a simple variation of E is about meaningless. But as Gfbotha said, it could also be a matter of Poisson ratio.
But the question is: are you absolutely sure of the "extension" of the field in which linearity / elasticity can be assumed? Do you have knowledge of what your material "mean" by being viscoelastic?
Bear in mind that, if there is some condition (APART from material characteristics) in the real experiment which violates linearity, your linear analysis will not be correct, but a non-linear analysis uniquely focused on the material won't be correct as well!
Don't be scared by non-linear analysis: if you find you need it, do it!!! It will be a little more difficult, and slower to solve, that's all.

Regards

 

[gfbotha]

Xhbg, one might be able to better assist if you could (or are indeed allowed) to briefly describe what material and what exactly you want to simulate.

 

[rtmpxr]

You can have small deformations (which obbey the Hook's law) but large rotations. This is a non-linear analysis. See "Foundations of non-linear elasticity" by VV Novozhilov