I've used a constraint equation to stabilze the four bottom corners of a model that is being lifted with straps. It seems to have worked (results are reasonable). However, I'm not sure what it's doing, because the bottom four corners aren't remaining planar. Is this configuation ok? How can I tell how much load this coupling is taking/transferring? Is it ok to use such a coupling between distant nodes?
I would use an RBE2, but then I couldn't turn it on and off between load cases.
Solved! Go to Solution.
One of the key advantages of using the RBE2 is that Nastran internally accounts for the geometry of the nodes. Writing your own relationships can be pretty easy for coincident nodes, but if they are not coincident, you must account for the geometry to avoid issues like grounding.
The displacement equation you have created is the following:
0 = [(1.0)*tz(node202)]+[(-.333333)*tz(node203)]+[(-.333333)*tz(node204)]+[(-.333333)*tz(node205)]
You can look at the .dat file at the resulting MPC equation and then the QRG to understand the syntax.
Without knowing your structure, it is impossible to say if this correct, but it certainly is not the same relationship that would be created by using an RBE2. The RBE2 accounts for the geometry of the nodes and the rigid body motion of the included nodes.
It is possible to print out the coefficients created by the RBE2 and then include that via MPC equations where you can turn them on/off per subcase.
To understand the load transfer that results from this relationship, you just need to look at the grid point forces and pay attention to the force of mpc. You should also look at the forces of constraint, if your nodes are not coincident then I expect that relationship may result in unintended grounding also.
Is there an MPC that ensures that the DOF of several nodes are coupled such that they are the same? So, that they always move together? In Strand7 we had a link element (Master-Slave Link) that was like that, which could be used in such situations (i.e. to enforce coplanar motion).
When creating MPC or constraint equations, it is totally up to the user to come up with proper relationships to fully define the desired behavior.
The RBE2 and RBE3 are the only "automatic" coupling choices.
Your feedback was very helpful, and led to a configuration which fits my needs perfectly. In my case, the structure is point-reflection-symmetric. I've attached one corner to the opposite corner (catty-corner from it) with an MPC which couples Z (dz1 + dz2 = 0). I did that twice, so that each corner is coupled only to it's opposite corner. That way, the structure is allowed to twist, but not tip. Allowing twist is important, because the 2-point lifting load induces twist in the structure.
The induced vertical forces from the MPC are very small (< 0.1 kN vs. a 52 kN lift, or 0.2%).