Minimal constraints with unbalanced loads

[marcrichard]

Hi all,

I would like to use the topology optimization module in NX 8 with loads imported from motion analysis in NX Motion.

At first sight, topology optimization in NX can not be used with inertia relief. So i am trying to see how i should place constraints on the mesh to obtain results similar to the analysis with inertia relief.

I found some interesting publications on how to apply minimal constraints on a model to do analysis with balanced loads (for example, this article). But the loads imported from NX Motion analysis are unbalanced, so I face a new problem.

For example, let's say i work on a crankshaft mechanism and want to run a topology optimization on a connecting rod. First, i run a motion simulation of the mechanism in NX motion and find the timesteps for which i have the peak forces. Then i transfer the loads to NX Advanced Simulation. I can import the reaction forces and accelerations for a given motion object into a time-dependent boundary condition field as a force/moment. The loads are imported at the joint origins (You must manually connect the load transfer nodes to your mesh). Two gravity loads are also added: The gravity defined in the Motion Simulation solution and the link acceleration, imported at the link's center of mass, to counter-balance the reaction loads (i suppose in the case you use inertia relief).

Have you got any idea of a good methodology to "convert" an inertia relief analysis into an analysis with constraints?

I would appreciate any reference on how to work with motion analysis to determine loads for stress/deformation/fatigue analysis.

Sorry for the beginner questions...

Kind regards,
Marc Richard

 

[marcrichard]

I don't know if it's possible to edit a thread so I post this correction.

Corrections to the previous post:

Then I transfer the loads to NX Advanced Simulation. I can import the reaction forces and accelerations for a given motion object into a time-dependent boundary condition field as a force/moment. The loads are imported at the joint origins (You must manually connect the load transfer nodes to your mesh). Two gravity loads are also added: The gravity defined in the Motion Simulation solution and the link acceleration, imported at the link's center of mass, to counter-balance the reaction loads (I suppose in the case you use inertia relief).

 

[rb1957]

if you have balanced loads and reactions then you only need to restrain the 6 rigid body motions.

you can fully contrain (6dof) one node ... but i wouldn't advise it.

i prefer to restrain 3 noes using (XYZ), (YZ), and (Z) constraints. you need to be careful in selecting nodes to make sure you react the required 6 dof. selected nodes should also be on reasonably stiff structures, if possible away from areas of interest.

the 6 constraints should (always check) have pretty much zero load in them (with balanced applied loads).

 

[marcrichard]

Thank you very much rb1957 for your help and sorry for the delay of my reply.

Practical advices are valuable, I have tons of stuff to read to improve my FEA skills and it's not always easy to find practical examples without getting too deep into the mathematics behind.

As I am working on topology optmization, the BC's matter seems of primary importance to obtain accurate stress/displacements distributions.

I found page 21 of this doc on "Cessna 441 Life Extension" a good illustration to an approach similar with what you described.

Are the results similar with inertia relief analysis? Is there any disadvantage with using inertia relief?

If you have any more suggestion/reference, it's always much appreciated!

Thanks.
Kind Regards,
Marc