Learning Forum

[peteroznewman]

Fabricio wondered why his fastener model that consisted of a Head, Shaft and Nut had such a different value of peak stress depending on whether it was a 3-body multibody part or it was Boolean United into a single body. Below is the difference he was seeing.

---

The first image is the 3-body result. Note the peak is 821 MPa.

---

The second image is the 1-body part. Note the peak is 627 MPa.

---

Most of the difference is due to the setting Average Across Bodies set to No, which is the default. 

If the Average Across Bodies is set to Yes, then there is little difference between the mulitbody part and the single body part. Note the peak goes down to 615 MPa.

It is understood that this model includes a stress singularity due to the sharp interior corner at the shaft, so comparing peak stress is somewhat pointless since any reduction in element size around the corner is going to result in a higher stress. The point of this post is to show that stress results on multibody parts should have Average Across Bodies set to Yes if they are all the same material.

---

Fabricio also asked how to get the full element display on the section view. It is by clicking the diamond shaped icon on the Section window.

Now here are the three section view:

It seems the elements are actually a little better quality in the 1-body mesh than the 3-body mesh above. I expect with a little work, the 3-body mesh could be made to look like the 1-body mesh.

Attached is an 18.2 archive.

[Bhargava Sista]

That's a nice post with a good example! To further your point, I would like to add a few more notes:

• In FEA, forces and displacements are calculated at nodes and the stresses and strains are calculated at the integration points (internal to elements). The stresses are then either interpolated or copied to the nodal locations.


• Therefore, each node receives a stress value from each element it is connected to. An average of all these values is reported as the stress at that location. This is valid only when all these elements have the same material assignment.


• In a multi-body part, the bodies share nodes at the interface and they may have different material assignments. This is the reason the option to average across bodies is turned off by default. As stated in the above post, if all the bodies in the multi-body part have the same material assignment, this option can be manually turned ON.

Regarding the mesh, it might help to use multi-zone mesh method on the 3-body mesh to get a structured hex mesh as in the 1-body case. This is a common recommendation for modeling bolts.

P.S: You may want to mark the sentence "The point of this post..." in bold.