I am working on a simplified problem before moving on the actual one. I am trying to simulate pull out test of a simple nail from a drilled hole. I have considered the materials to be isotropic and deformable.

The assembly is constrained co-axially and the top of the nail in translated to coincide with the face of the block. The analysis used is Dynamic, explicit with surface-to-surface penalty friction co-efficient (=0.75, chosen arbitrarily). For load, I have fixed the bottom of the block and provided a displacement of 10 mm to the top of the nail.

Now, when meshed using hex elements there is no mesh convergence. The results are scattered. So, I tried looking up about it and I found hertz contact problem (https://abaqus-docs.mit.edu/2017/English/SIMACAEBMKRefMap/simabmk-c-hertzcontact.htm#simabmk-c-hertzcontact-t-DynamicAnalysisInASLB-sma-topic5). Although, the problem is for cylinder to cylinder contact, but I tried applying the mentioned C3D10M mesh for 3D problem. The results still don't seem to converge.

I also had a doubt about hex meshes. There is a particular value in hex elements for block and nail when the meshes align perfect and at those values the von mises and the reaction forces are the least. Does it have anything to do with contact area and overclosure?

And finally what is the suitable way to perform the simulation where the result is the reaction force?

I am attaching a picture of the C3D10M meshing and the one of the hex mesh where the values drop suddenly.