First, some good news…
In Workbench R14.5, ANSYS introduced nodal Named Selections, and in R15.0, they have added the ability to create Named Selections of elements. So now you can make groups of nodes or elements just like you can in MAPDL. You can use these name selections for result plots to show just specific portion of the results. 
In R15.0, you can right-click on a Name Selection in the tree and hit, “Create Nodal Name Selection”. This creates a Name Selection of all the nodes associated with the particular piece of geometry in the original Named Selection, whether that is a body, surface, edge, or vertex. Highlighting the nodal named selection in the tree will then take you to the Worksheet where you can add rows for limiting the selection of nodes to a location value or some other criteria.
This is also where you can add a row to “Convert” the “Mesh Node” entity type to “Mesh Element”. The Mesh Element entity type has a criterion choice for how the elements are selected from the nodes.
“Any Node” will select all the elements that have any of their nodes in the list of nodes that make up the current named selection. “All Nodes” will select only those elements that have all of their nodes in the current set. Many of you may already know this, and it is a great new feature, but there is a catch, and that brings us to the telling of the “Bad News”.
The Bad News…
After noticing the generation time of the name selection drastically increase when using the “All Nodes” criteria, I ran a small test case. With just a cube meshed to two different refinement levels, I tracked the generation time for the element name selection using the two different criterion. Here is what I found.
I am not even going to speculate what is different with the “All Nodes” node-checking algorithm, but an increase in element count by a factor of eight caused more than a 13300% increase in generation time. But look at the generation time for the “Any Node” criteria. It stayed right on par for the different mesh sizes.
So, back to the Good News, and the Really Good News…
The Good News is that you can avoid the long generation times, in R15.0, by not using the “All Nodes” criteria. The Really Good news is that when I ran the same test in R16.0, I got 6.0 Sec for the “Any Node” criteria, and 6.3 Seconds for the “All Nodes” criteria. So ANSYS has already fixed the problem in R16.0, which just gives you another reason to upgrade. If you are going to continue using R15.0, then just stay away from the “All Nodes” criteria for the element named Selections. It is much better to use the location based filtering to cut down your nodal selection so that you can use the “Any Node” criteria.
