Mesh Convergence Issue

[ebdoep22]

I have a 2D model made using CPE8 and CPE6 elements in a non-linear analysis. The model was running fine, but before I moved any further I wanted to do a mesh conv study. What I found I can't explain. Attached is a plot with the X axis being mesh seed size (finer mesh to the left courser to the right) and the resulting deflection of a point in the model. I am used to these plots having monotonic convergence, but clearly this model does not.

I thought maybe this had to do with the nonlinear analysis, so I ran a linear analysis and got the same pattern and grouping (just shifted). I tried a couple of the other plane strain elements and even for kicks switched it to plane stress. All the same groupings just slightly different displacements, but overall the same pattern. Ran it by a couple of people and no one had any idea.

Brady

 

[IceBreakerSours]

Interesting. What kinds of non-linearities do you have in the model?

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[ebdoep22]

Using several orthotropic materials (CFRP and honeycomb) and have large deflections. Also the materials have extremely different modulus. The honeycomb has a 0.0 Poisson's in two directions. The model is of a morphing trailing edge for an aircraft. Looking at the cross section, the flexiable portion is 0.5m in the chord direction, but will have ~20cm tip deflection.

From the graph it almost looks like oscillatory convergence instead of monotonic, but even if that were the case (which I have not seen previously in mesh conv studies) it does not explain why it holds a value to a certain point then jumps.

I am will to try anything to test out possible fixes, I just really don't want to have to start the whole model over again.

 

[ebdoep22]

Anyone have any ideas?