I am struggling with the right parameters for the exponential settings.
How do I determine the correct parameters?
You can check the “Troubleshooting Finite Element Modeling with Abaqus” book for detailed guidelines, but the exact values aren’t so important - the way it changes (exponential increase) is the key here. You can do some initial estimation based on the standard rules taking into account the stiffness of the components in contact and then vary it. Most importantly, look for excessive penetration.
I flicked through the comments and here is my 2p, mainly from my Abaqus times, but in case is similar in CCX:
1.- Your master surf (static part) seems much finer vs slave (moving) one? Normally it is the other way around. Have you tried swapping slave/master? When slave side is too coarse vs master + curvature, slave nodes can ‘miss’ and show large penetrations.
2.- The static part can be coarsened a lot to speed up tests. It’s a chunky part so you can use growth of tets internally (gradient thingy in meshing). For sliding contacts the master doesn’t need to be so fine along a cyl, straight ‘road’; but around the circumference and/or fillets contacting slave at the start.
3.- If you are using high friction values, say > 0.20, you will need to solve the full matrix (UNSYM parameter ON). This is to store and solve all the off-diagonal terms required in [K] matrix to model high friction properly. Not sure how it work in CCX, but just in case.
I hope that master/solave assignment rules are clear at this point, but it might be good to remind them just in case: Selection master and slave in contact
Or made rigid with RB constraints, as I suggested above.
CalculiX doesn’t have such a switch, but in the log you can see messages like this:
Solving the system of equations using the symmetric pardiso solver.
or:
Solving the system of equations using the unsymmetric pardiso solver.
In fact, it’s recommended to use unsymm=yes in Abaqus also for lower friction coefficients and sometimes even when there’s no friction.
1 Like
I have started with coarse models to get the simulation running to an end. Improved the model step by step. The problem was to find the right parameters for acceptable penetration and realistic values. Have put the overview of the K values with results previous in this topic.
In previous models the mesh of the slave was smaller then of the master, but still had some error.
I have now modeled the simulation with surface interaction with the exponential setting. Played with the parameters in smaller model to find a good value and coppied in this new model.
The penetration is much less, and the insertion force graph is also realistic to what to expect. Both mateials of the copper plated hole and the contact are with plastic settings
Next step is run it with friction.
2 Likes