When data is internalized into Femap, each output set gets a header, containing information about the analysis program, analysis type, time step value or frequency value, adding a little to the total size of the data. In addition, FEMAP will calculate some data that’s not output by the solver, i.e. Major Stress, vonMises Stress. The end result is that the storage requirement for the data actually expands a little bit when it’s internalized by Femap.
With v11’s “Attach”, the header is still created, but that’s all that is stored with FEMAP. This keeps the model file itself small and light. In addition, all the results data has not been run through the FEMAP database and now does not bloat FEMAP’s database cache.
In a case where one points to a large number of results files, it’s very obvious how this can keep the demands on storing the Femap model to a minimum, speeding up File – Open, and keeping Femap’s memory footprint lower when running.
No change how everything is done in FEMAP, all post-processing commands are the same and don’t differentiate between attached and internalized data. One can have any number of attached files and internalized output sets.Nastran .op2 support in v11, looking to add the other formats as fast as we can. The code has been architected such that once the first one has been done (.op2), it’s now much easier to plug in other output file formats.Attached results can be selectively internalized. Once output data has been attached, the user can then specifiy that some, or, all of it, can be internalized as before to reside in the FEMAP .modfem file.
Boeing International Space Station Laboratory model as a baseline. Without any output data, it’s roughly 69MB on Disk, and 303 MB of RAM used when running. Attaching to the file results file as opposed to internalizing is better across the board, it’s faster, smaller, and doesn’t clog Femap’s database cache with output data.
Vertex Buffer Objects use a lot of RAM, but they’re a lot faster than not using them.
Vertex Buffer Objects can make a 4.5 Million Node/Element model 6x faster as long as you have a 2 GB Graphics card. After v11 (v11.1), OpenGL based graphics will be faster still with a lot less RAM requirements.
FE models can be updated or remeshed more easily. Geometry can be created based on the original mesh and is automatically associated tothat mesh. With the new geometry in place, it can be modified using the existing geometry modeling tools and remeshed. Further mesh changes can be made using the Meshing Toolbox. In addition, you can simply delete the mesh once you have the geometry, add cutouts and other modifications and then remesh. You can work on individual patches of shell elements and turn them into geometry. Alternatively, you can select the entire mesh and generate a surface, but some of the cusps may not be created. For most effective results, discreet regions should be picked manually.
In shipping Femap, one can already assign Superelement IDs to nodes and create a Superelement run. With v11, support for External Superelements will be available. Femap makes it easy to create a store external superelements, and then once these are available (or supplied to you by a 3rd party), Femap add new functionality to create the full system model using any number of superelements and a residual structure.
A complete do-over. No more limits on number of curves and a lot more options. User input at this time would be greatly appreciated – is there something users would like added to current XY-plotting?
Currently one can plot plate and solid data in the same post-processing windows, we’re adding line data as well.