Tensile damage in concrete

[compe_ad]

Hi all! I am seeing some unusual damage pattern in my model. Can anyone help me understand why I have this horizontal crack in my model and how can I eliminate this? It is occurring at the region between fine and coarse mesh region. My fine meshes are of 1 in. and coarse mesh are of 3 in.

 

[FEA way]

Is there some constraint connecting those refions with different meshes or is it a continuous mesh ?

 

[centondollar]

If I remember correctly, the results obtained by damage models can be mesh-dependent, similar to how fracture mechanics problems can be mesh-dependent. The odd results at the boundary of varying mesh density would seem to support this faint memory of mine.

You have been posting about this problem for quite some time now. If it is a thesis project or something similar, I would assume that your supervisor could provide more detailed and elaborate answers to these questions.

 

[compe_ad]

Thank you for the response. No there is no constraint in between. Yes, I am working on my thesis project, and I found "eng-tips" to be more informative than any other options, so I like to ask questions here to know more about things.

 

[FEA way]

Change the mesh (move the refinement, remove it, use uniform and non-uniform meshes, try different densities and so on) and check how this influences the results. Crack analyses can indeed be quite mesh sensitive but it also depends on the way you model them.

 

[swimfar]

These questions should be posted to the Student's Area:

https://www.eng-tips.com/threadcategory.cfm?lev2=105

1. Do you understand the material model that you are using? Where did the parameters come from? Have the parameters been validated already on other similar problems? Concrete damage models are complicated. And as centondollar pointed out, they are often mesh-dependent. If this is the case for your material model, it may include a mesh size/characteristic length parameter. Find out which constitutive model your software is using and look for papers that describe it.

2. Instead of just randomly making changes, look at the stress state in those elements. You should have the entire stress tensor history and you can easily hand-calculate the damage in a few of the elements. Does the stress state make sense? Does the calculated damage make sense? Start there.

 

[driftLimiter]

I am expecting to see a transition region from the fine mesh to the coarse mesh. It looks like an abrupt change, doesn't that lead to many nodes not being connected across the surface? All elements should be connected continuously.