Best way to connect QUAD4 elements to a neighbour of half the size?
A simple model is defined: left side fixed, right side fixed displacement.
We should get a uniform stress field, as shown in below Figure with uniform mesh:
The problem is that red nodes on the T-junction are not connected as continuous solid, shown below (magnified displacement X):
This causes a Von Mises stress deviation of approx. +10%/−18%:
Potential solution 1:
RBE2 elements, they reduce this deviation to approx. +3.7%/−10%.
Potential solution 2:
Thanks to Greg Harte's answer to an earlier forum question, RSpline elements were tested as well. They give the best result so far at +2%/−10% percent.
The question:
A 10% deviation still feels large to me. Is this really the best way to do it? Or is there an even better solution (aside from using a triangle/trapezoid) transition mesh? Model is attached.
Answers
-
is this a solid or shell?
If this is solid, you might think on using contacts (FREEZE or TIE) to couple these.
RSPLINE and RBE3 connections are alternatives.
But usually these transition zones are expected to present some deviation. Usually we try to keep them away from the interest area.
0 -
Adriano A. Koga_21884 said:
is this a solid or shell?
If this is solid, you might think on using contacts (FREEZE or TIE) to couple these.
RSPLINE and RBE3 connections are alternatives.
But usually these transition zones are expected to present some deviation. Usually we try to keep them away from the interest area.
It is a shell.
Following your advice, I tried to couple them with FREEZE contact. However, they don't seem to work because it is inside the volume? The out file warns (among others) for: "... nodes are included by both the Slave and the Master of TIE(CONTACT)".
Attached is the file, can you help me understand how to properly create the FREEZE or TIE contact?
0
Altair Employee