A data table is a easy way for a user to specify a region of a model from which to output desired quantities from a postprocessor to a file. I’m going to demonstrate how this is done for FEMAP, doing it with other postprocessors should be a similar process. For the purposes of breakout modeling the quantities needed to be output are node ids and displacements.
The motivation for doing this is that for large models, from which you want to analyze a relatively small local region, even the translation step can become time consuming. You may have a million element model and want more accurate results in a region of 10000 elements. It is inefficient to have an analysis tool like stressRefine wade through the results of a million elements to extract those needed in the local region.
A data table allows you to simply “rubber-band” select a region of interest. I’m going to show doing this with box-selecting.
So here’s how it goes in Femap:
Now we need to tell the Femap selector to box select nodes:
Note there are lots of other choices besides box. Now select a box a bit larger than the local region of interest. It doesn’t have to be precise, but it helps to include the whole part for “breakout by part”:
The data for all the nodes in the box will go into the data table. Now we specify what quantities to output (node ids and displacements).
Table is now ready to be saved:
This will all be in the stressRefine user’s guide that will be updated shortly. This procedure is a lot simpler to do than it is to explain, it takes just a few seconds once you have tried it a couple of times.
The example of breakout by part in an assembly from my previous post used a data table. Extraction of the breakout model took 2.4 secs using the data table, while it took over 30 additional seconds to translate the results and extract the model directly from the Nastran binary output results. This is for a model with less than 300,000 elements. The time savings will be much more for larger models.
Automatically Determining The Point of Maximum Stress In The Region Selected
In stressRefine, the user can specify the center of the local region using a node id. But this is not necessary by default:
The point of maximum stress in the region enclosed by the data table is used by default. Determining this point quickly without having to translate the results for the full model requires stressRefine to recreate the stresses in the elements that own the nodes in the data table. This is straightforward because displacements are known, and it is known that Nastran used quadratic elements (e.g. 10 node tets) in the solution. Calculating the stresses in the local region in the example above took much less than a second (for 55000+ nodes in the data table, owned by 33000+ nodes).