MemsGeek
Mechanical
- Oct 29, 2007
- 3
(Initially posted in Mechanical Acoustics/Vibration. Resubmitted per a recommendation from another member.)
I am trying to use the ansys mode shapes for reduced order modeling outside of ansys. The reason for exporting the model is that the structure is a fairly simple arrangement of beams/solids/and or shells, but the loading is nonlinear and time dependent. The response of the structure always takes approximately the same shape.
Obtaining the displacements at each node is trivial, but the same is not true for estimating the amount of mass apportioned to each node. It seems like this should be trivial, but I am having trouble getting it out of ansys.
I can export the mass matrix, but these are element effective masses, not physical masses, and therefore depend on the specific shape functions of the elements. I tried using the lumped mass option in a modal analysis and wrote out the mass matrix using the hbmat command, but recognized that these masses were not simply the mass at each node.
I am hoping to make this algorithm generic enough to support Beam44, Beam188, Mass21, solid45, and shell43 for the analysis. I could brute force the approximation by looping through each element type, computing volumes, and multiplying by densities, but this wouldn't necessarily work for beam 188's where the cross sections can contain multiple materials.
So... If you wanted to get the true physical mass attributed to each node in a lumped mass formulation, how would you do it?
I am trying to use the ansys mode shapes for reduced order modeling outside of ansys. The reason for exporting the model is that the structure is a fairly simple arrangement of beams/solids/and or shells, but the loading is nonlinear and time dependent. The response of the structure always takes approximately the same shape.
Obtaining the displacements at each node is trivial, but the same is not true for estimating the amount of mass apportioned to each node. It seems like this should be trivial, but I am having trouble getting it out of ansys.
I can export the mass matrix, but these are element effective masses, not physical masses, and therefore depend on the specific shape functions of the elements. I tried using the lumped mass option in a modal analysis and wrote out the mass matrix using the hbmat command, but recognized that these masses were not simply the mass at each node.
I am hoping to make this algorithm generic enough to support Beam44, Beam188, Mass21, solid45, and shell43 for the analysis. I could brute force the approximation by looping through each element type, computing volumes, and multiplying by densities, but this wouldn't necessarily work for beam 188's where the cross sections can contain multiple materials.
So... If you wanted to get the true physical mass attributed to each node in a lumped mass formulation, how would you do it?