XANSYS Message: 176 [Go back to message list] [bookmark on del.icio.us]
No rating yet Subject: Re: Stress concentrations Author: Paris Altidis Date: 1999-01-11 13:56:00Couple of comments based on other responses that I've seen:
1. I don't know what kind of contact elements you've been using, but switching to CONTAC48s you have to change the PLANE82s to PLANE42s (linear). The mid-side nodes in PLANE82 is nothing but trouble when used along with CONTAC48s, let alone problems in getting the problem to converge. An option is to remove the mid-side nodes for the elements whose nodes are used to generate the CONTAC48s.
2. As others suggested, a finer mesh with PLANE42 and CONTAC48 will be ok. Along with the AUTOTS, ON option don't forget the LNSRCH, ON option which has helped me in all the interference fit problems (recommended in ANSYS doc'n - search under "interference"). Include the NLGEOM,ON option as well and use the sparse solver for this class of problems. I've heard some very positive feedback about the new PCG solver in v. 5.5 when used in contact problems but haven't tried it yet.
3. As far as the yield modulus (a.k.a. tangent modulus) is concerned, for steel start with 10% of the material's Young's modulus and as Rick Fischer mentioned, you should not have any convergence problems. Unless, you have the actual stress - strain curve for the material.
Paris Altidis Borg Warner Automotive
> -----Original Message----- > From: Rick Fischer [SMTP:Rick.Fischer@c...] > Sent: Monday, January 11, 1999 12:00 PM > To: PTibbits > Subject: [xansys] Re: Stress concentrations
> From: "Rick Fischer"
> Plasticity, yes, I forgot that one. Good point, Patrick! I've > done this, it can be done, but can be tricky. Some pointers:
> 1. Use plastic material only where yielding is likely. Set up > a one material for the elastic part of the shaft and one for > the plastic part. Use the elastic material everywhere except > near the contact. The solution will run faster with with fewer > nonlinear elements.
> 2. Let AUTOTS do the work by using NSUBST to set a wide range > of min and max timesteps.
> 3. If convergence is difficult, set the gap stiffnesses real > low, run to convergence, then restart with higher gap > stiffness. Warning: If you use a CONTACT48 with a fine mesh > and a low gap stiffness, it is possible to push the contact > node so far thru the target that it will exit the back of the > pinball and release.
> 4. Use a realistic yield modulus. The flatter the curve, the > harder the convergence.