
Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Erik Kostson Oct 12, 2018 4:02 AM (in response to Babis Lazaridis)Negative stiffness is not really good, and of course things fail (in stable).
A possible common cause is that the load is just too large for the size and material used. Try solving a linear static small deflection analysis and that will give you an idea roughly of the order of magnitude for the displacements. If it is hundreds/thousands of mm (many times bigger than the model size), then it will not solve with nonlinear. In that case go through the model and apply realistic materials, loads, and check dimensions (perhaps part is tiny).

Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Babis Lazaridis Oct 12, 2018 3:20 AM (in response to Erik Kostson)the part is the left half of a wristband, the distance from one edge to another is ~50mm so it is not very small. i applied a load of 40N and in another case a prescribed displacement of 4 mm. i defined the material as an elastomer with a stress strain curve going up to ε=400% and σ=2,8 MPa so i think it should be running fine..

Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Janko Stellaard Oct 12, 2018 4:03 AM (in response to Babis Lazaridis)I have looked at your model. By changing the convergence parameters the simulation will run but the autostepper will use very (very) small (pseudo)timesteps.
The stiffness of the model is really low which gives troubles. I would suggest changing the method from force control to displacement control.

Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Babis Lazaridis Oct 12, 2018 4:20 AM (in response to Janko Stellaard)ok thanks i'll look into it! do you think the displacement control method would be also suitable for a study wich includes this part, a polypropylene part and a metal sheet? (the other two are the "frame" of the assembly, i put them as bonded inside the initial part)

Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Janko Stellaard Oct 12, 2018 4:29 AM (in response to Babis Lazaridis)What Erik said: the displacement with this load and Emodulus is way too high. So please review your material properties/loads first, something might be wrong there.



Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Erik Kostson Oct 12, 2018 4:35 AM (in response to Babis Lazaridis)Run it with linear static you will see that even with the 1 N and 4 N loads applied it deflects about 1300 mm (for a rubber stiffer than the one defined, initial linear E modulus of 3 MPa). Nonlinear FEA is not a magic black box where everything can be solved, and it would be quite impressive if this can be solved for this loads and nonlinear elastic model with large deflections (geometric nonlinear), that wants to deflect many times it's original size (so apply smaller loads and see what happens, in nonlinear FEA loads are applied gradually, especially for rubber parts, in order to help the NR iterations to reduce the residuals to acceptable values below the convergence criteria).
Another point is that the mesh is not really good and that itself can lead to problems when trying to obtain a converged solution. Trying to improve/simplify the geometry and to obtain a better mesh is important (not sure how one can look on mesh quality parameters, like Jacobian determinants, tetrahedral volumes, aspect ratios, mixed products,..).
As suggested, if displacement can be used (then look on reactions to get total load applied), that might help also, especially if your part is to soft and we get a negative stiffness (load scaling cannot deal with that very well).

Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Babis Lazaridis Oct 12, 2018 5:58 AM (in response to Erik Kostson)thanks for the replies guys! eric i run a static study and i noticed the large deformation even for small loads, that you mentioned.. so i should probably use prescribed displacement (and change the control method to displacement control) and calculate the loads afterwards right? unfortunately i can't change the geometry, so i use a fine curvature based mesh with a high aspect ratio to make it work. i will also check the materials again.

Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Erik Kostson Oct 12, 2018 6:30 AM (in response to Babis Lazaridis)No worries.
That could be OK. Yes, you can get the applied load by looking on the reaction forces (resultant) on the fixed face or perhaps on the edge with the enforced displacements. (The results tool is called List Result Force)
Try that and see how it goes (by how much are you going to pull that edge with?).

Re: Negative diagona on Assembled Stiffness Matrix error on nonlinear static study
Babis Lazaridis Oct 12, 2018 9:11 AM (in response to Erik Kostson)i just need to move it about 1cm (in the final assembly with another two parts which are bonded inside the part i posted). i started the study but i guess it's going to take a while due to the high mesh and the weird geometry of the parts (i also ran the study on the part on its own and it finished with no problems and expected results!)




