assafwei
Materials
- Jul 20, 2008
- 89
Hi,
In order to try and validate a Hyperelastic material model I tried to simulate the following simple experiment and then compare it to real life results. Since I am using this material model to simulate sealing, this looked to me as a simple and quite similar to the strains the seal will undergo duting its life time.
The experiment is simple: I modeled a rod made of steel, the end of the rod is ball shaped 10mm in diameter (I have 10mm balls so I can use it in the real testing later), I then modeled a small piece of the elastomer (20x20x2.3 mm) and beneath a thick rigid plate of steel. I cut it to quareter (1/4 symmetry) and fixed the bottom steel plate, created a frictional contact pair between the sttel plate and rubber plate, and a frictionless contact pari between the rod and the rubber plate. I also added a downward displacement on the rod. Additionally I added a nonlinear adaptive mesh zone, based on skewness criteria in order for it to converge (large compression, that will have problems converging otherwise). As mentioned in the subject - I used linear tetras all around (mainly because this is the limitation of the nonlinear adaptive meshing).
I started with a medium mesh (used default settings, only changed global size to medium) and ran the analysis, the results look good, as I expected the edges of the rubber plate raised becuse of the compression by a few milimeters.
I then changed the mesh size to fine and re-ran the analysis, I got the same force reaction on the rod, and the same compression, and generraly the same results with one big difference - the edges of the rubber plate didnt raise at all, except of the compression, and some expansion of the plate to the sides - there was no displacement at all in the direction opposite to the displacement of the rod.
Can anyone explain how can the change in mesh size made such a difference in the results?
Thanks.
In order to try and validate a Hyperelastic material model I tried to simulate the following simple experiment and then compare it to real life results. Since I am using this material model to simulate sealing, this looked to me as a simple and quite similar to the strains the seal will undergo duting its life time.
The experiment is simple: I modeled a rod made of steel, the end of the rod is ball shaped 10mm in diameter (I have 10mm balls so I can use it in the real testing later), I then modeled a small piece of the elastomer (20x20x2.3 mm) and beneath a thick rigid plate of steel. I cut it to quareter (1/4 symmetry) and fixed the bottom steel plate, created a frictional contact pair between the sttel plate and rubber plate, and a frictionless contact pari between the rod and the rubber plate. I also added a downward displacement on the rod. Additionally I added a nonlinear adaptive mesh zone, based on skewness criteria in order for it to converge (large compression, that will have problems converging otherwise). As mentioned in the subject - I used linear tetras all around (mainly because this is the limitation of the nonlinear adaptive meshing).
I started with a medium mesh (used default settings, only changed global size to medium) and ran the analysis, the results look good, as I expected the edges of the rubber plate raised becuse of the compression by a few milimeters.
I then changed the mesh size to fine and re-ran the analysis, I got the same force reaction on the rod, and the same compression, and generraly the same results with one big difference - the edges of the rubber plate didnt raise at all, except of the compression, and some expansion of the plate to the sides - there was no displacement at all in the direction opposite to the displacement of the rod.
Can anyone explain how can the change in mesh size made such a difference in the results?
Thanks.