How can I tie 2 surfaces together, but only w.r.t. 1 specific DoF? Moreover, this 1 DoF is w.r.t. a non-standard coordinate system, i.e. as a result of *TRANSFORM.
Example: frictionless “contact” of a bolt in a borehole. The pin is allowed to freely move in the direction of its main axis and to rotate about it, but it must follow the borehole for all other possible movements.
will connect the nodes in the set dep to element-faces described by nodes in-
cluded in the set indep. The set dep must contain all nodes which should be
”glued” and the set indep should contain all nodes of the elements surfaces to
which the dep nodes should be glued. The numbers ”123” are the degrees of
freedom which will be connected (”t” will create a thermal connection, “p” a
press-fit connection). The ”c” triggers the correction of the position of the de-
pendent nodes to a position on the surface of the independent elements (highly
recommended and default), ”u” would prevent the correction, “f” forces the
dependent node away from the independent face.
Tie constraint and MPC generated are sensitive to surface definition of master and slave, it may lead to unexpected or wrong results. However, this not occurs for tied contact type.
Another advantages using tied contact seems to be possible working for any element type (beam, shell and solid) at arbitrary plane e.g circular, flat or curves.
However, setting parameter values for stick slope and/or friction still need more test to validated since it’s not officially documented. I’m only doing an opposite way from implementor reference document, setting to lowest values make it released at normal and/or tangential movement.
This is probably the way that Calc_em suggested, “No separation contact”. From the example in the linked answer, people seem to have verified that it works. However contact simulation is computationally costly. I would prefer an MPC-variant because coupling DoF’s should not require the solution of anything.
Is contact with TIED and low stick slope really nonlinear? It seems to just be linear springs, and any nonlinearity might just be for large deflections and fairly low-cost compared to full-blown contact.
Interesting to conduct testing on computational times between MPC constraint and tied contact type, even it negligible for tie constraint. Hard point are seeking a percentage of sharing force or displacement based on master node projection to the slave. Only CGX as precessor can do automatically, another is not yet capable.
Tied contact types with parameter setting can be use as alternative way of working, movement at normal and/or tangential also can be set.