Trying to analyze a press fit gear ( with SOL101). Right now it struggles to solve. Using a linear contact. I restrained the end of the shaft and tried to prevent the gear from moving axially and in theta. Gave me some wierd results (probably did not specify the cylindrical coordinate system correctly.
I would like to run an analysis with forward and reverse tooth loading to calculate fatigue. Since I can't seem to figure out the right way to constrain it in theta (lazy or cluelesss, thought I was doing it right but...), I thought about simply using a glued contact, but not sure if that works with the press fit.
Solved! Go to Solution.
1) Split problem into two parts. Calculate pressfit without tooth loads, use symmetry BC and calculate only 1/4 part of model. Use contact instead of glue. Two orthogonal symmetry planes should give additional constraint and prevent gear from rigid body move.
2) Use glue and calculate tooth loads, glue should constrain model without any problems.
3) Post example file with the problem. Problem description is unclear and one screenshot isn`t enough to find what is wrong.
Seems to be working now, more or less (after running it about 10 times). Glued connection did not work with the press fit, although I think I was doing something funny when merging nodes and mixing coordinate systems where the nodes would not merge.
I ran with the press fit (interference) with contact (with friction) and a tooth load. Interestingly enough, I only restrained the gear in axial. I would have been more comfortable using a theta restraint, but it worked. Go figure.
If you are curious, I attached the file.
See your model. Here is some tips.
To get uniform pressfit pressure you can set overlap at contact region options and set Initial Penetration option to 3..Zero gap/penetration.
Also all geometry is swappable or revolvable. It is possible to build fine hexmesh and dramatically reduce element count and increase element quality, hexmesh rules)
Check your contact defintion and region offset. What Karachun had said regarding INIPENE is spot on.
Using INIPENE should produce no pressfit loads, when region offset not changed
Pressfit stress can be calculated by including region offset ,INIPENE (this method is also useful for checking different interfaces without changing the geometry)
Using INIPEN will produce the pressfit loads based off the nodal positions (all models the exact same just changed PEN type)
Another way you can aid the contact when using TET(10) &HEX(20) is to use midside nodes on surface, this will help maintain the suraface curvature:
I'd also split the problem into two parts:
1. gear/shaft interface
2. tooth loads
you can do this by adding a subcase to the SOL101 and setting the contact status to START FROM PREV SUBCASE.
I've worked with a similar model. As there is an interference press fit between the shaft and the gear hub, I assume you are looking for contact pressures and stresses at the interface region. For this problem to converge accurately using a contact setup, define an offset magnitude (the amount of radial interference) within the region of the cylinder OR the gear hub (not both) under connection region > region options > offset distance.
Also, in the contact region properties dialog box, select the INIPENE parameter = 3 (zero gap/no penetration), this will allow for the offset to take effect under the solver. The friction values defined by the physics of your modeled materials.
As for the boundary conditions, you can constraint the shaft cylindrical surfaces from moving in the axial direction, and apply a pinned constraint to one end of the shaft. The gear wheel can have a translation constraint from axial displacement in one face.
Reminder that the stresses at the interface region must have a smooth profile (no hot spots) to arrive at precise results.
You can handle with building hexmesh on sharp edge by using tool Mesh->Revolve->Element.
First, create 2D mesh on edge.
Then revolve 3D mesh from 2D about axis of rotation.
Now you have neat hexmesh. Set mesh sizing on outer gear solid body to get coincident nodes and then merge them.