I've been attempting to solve a problem regarding a stiffness of an assembly. The analysis that I'm performing has two requirements: (1) determine the modal response of the global system
(2) determine the stiffness of each component and provide the contribution of each component to the modal response
My question is regarding the second requirement. I've extracted the stiffness of the using
param, extse, dmigpch into a punch file but the results are provided in an element by element basis.
Stiffness is dependant on the BC's and the shape along with the material properties. The element size is dependant on its stiffness output. There must be a way for Nastran to export this for the entire system.
How can I extract the global stiffness of a component? Is there a method by which I can format the output of the punch file to a 6DOF stiffness matrix?
Some comments that may help:
Standard practice for setting up an EXTSE creation is to define an ASET that represents the boundary between your component and the rest of the system. Nastran then performs a Guyan reduction of the stiffness and mass down to this boundary and this gets output as your DMIG in whatever format you choose(op2,pch, etc). This is then added to the system run as an EXTSE.
So if you perform normal modes on this boundary mass/stiffness that is the behavior passed along to the system. If want to increase the accuracy of the behavior of that component in the system run, you can choose to add residual vectors and/or craig-bampton technique (fixed boundary modes) to the reduction so those modes are also added to the system response.
Not sure what you mean by the output being element by element? If you do not include an ASET, then you do get the global mass and stiffness, but that is for each GRID/DOF not by elements.
Thanks for your prompt responce.
In regards to your answer, I am not attempting to perform CMS at the moment. My question is regarding the stiffnesses at grid points in the punch file.
My issue is that the stiffnesses is given at a grid point on on a single element. If the component mesh is refined - the comparison of stiffness at a given point shows the values to be different (I checked this for a model with beam elements). The stiffness I desire is for the entire part/component. The stiffness at a node is not the same as a [6x6] stiffness matrix for the entire model. There must be a way to output this.
I've extracted the stiffness of the mode by node into a punch file via a static run using the following techniques which yielded the same answer.
(1) 'PARAM, EXTOUT,DMIGPCH'
or via a SE creation
Also, specifying an ASET and not did not change stiffness values.