Dear Simulia community,

I am working on an Abaqus simulation representing carbon fibers in a PETG matrix. Having finished coding the script to generate the model, I am now conducting a mesh convergence analysis to determine the optimal mesh density for the model. While doing this, I encountered some odd behaviour in my fibre values. I noticed that the stress values within the fibre would decrease as the mesh became finer, which seems quite odd to me.

To perform the convergence analysis, I wrote a code to extract the node ID of a list of nodes at the same coordinate (in order to measure the stress at the same points through different models/meshes) at different points on the fibre (middle node and sides), as shown in the numbered pictures (node 7 and 8 could not fit in the picture).
As shown in the picture on the right, I extracted the values using the Probe tool.

Here are the 3 model meshes I tested :

 

The table below shows the results I obtained for Mises/22 stress at different points of the fiber mesh (numbered as in the first picture) for the three different models. As you can immediately notice, the values consistently decrease as you go from model 1 to model 3.

These results are confusing me, as from my understanding of mesh convergence, the finer the mesh, the more accurate and higher the stress results will be, as more factors are considered.

For more information and context, the model I am using consists of a rectangular matrix (C3D10) body that surrounds the fibers (C3D6). The two parts are connected by a tie constraint and are under a uniform load in the y-direction, with the base constrained, as can be seen in the pictures below.

I find these results very odd and am wondering if any other forum members have encountered the same problem and could offer some advice.