Strange TIE and EBOU behaviour

[Jfrisch]

Dear FEAP users,

I am trying to model a crack-tip problem and get some strange behaviour according to my Ofile. I am using several bloc-statements (combined with regi) to define my mesh blockwise and want to tie the matching nodes an almost every edge except the crack-edges by tying the regions. Unfortunately there are two strange behaviours/outputs in the Ofile:

• the last region/block I am defining is tied completely (every single node)
• my first block gets an edge boundary condition defined accorind to the Ofile that isn't defined in my Ifile

The first issue dissappears if I remove the "TIE,REGIon,16,17" line from the attached Ifile so I suggest it has something to do with this. I have no clue why the second issue occurs and no startpoint of debugging.

Has anyone had some similar problems and can give me a hint why I get this strange results of meshing?

[Prof. R.L. Taylor]

It looks like block 20 has bad geometry.

To look at the mesh I used interactive mode after all the ties.

Give the command PLOT

Then plot the mesh with MESH followed by PICK to use the mouse to plot the region around the crack.
Then give ELEM and you will see that element 20 (block 20 probably) is not there.

What version are you using?

[Jfrisch]

I am using FEAP8.5 released this October and another strange issue occurs. If I try to start a simulation with this inputfile but more nodes per edge FEAP starts, merges nodes and freezes there without any error-message (started it yesterday and this morning after nearly 1000 minutes it was at the same state right after the tying.

EDIT: I have checked both, the first block that seems corrupted at my plot and the last block defined for the element 20 that you mentioned. Both seem correct in my opinion and are defined in the same manner than the other equivalent blocks that work:

• "REGIon 1" and "REGIon 2" differ only in the sign of the second coordinate of the defined nodes (the first two defined nodes are the same)
• "REGIon 17" and "REGIon 13" differ only in the sign of the second coordinate of the defined nodes and the ordering

In both listed comparisons the latter defined region is meshed nicely in contradiction to the first mentioned one and I don't get why this is the case.

EDIT2: Even if I delete all the blocks except for the inner four ones the mesh isn't created correct. The block located at north-east is degenerated and problematic but I can't see any failures in my inputfile. I have attached a screenshot of the BLOCk-declarations opened in emacs and the plot FEAP generates.

[Prof. R.L. Taylor]

OK, let's debug first the four blocks.  Post the mesh for that.

[Jfrisch]

Dear Prof. Taylor,

please find attached the inputfile "Icrack_debug", produced outputfile "Ocrack_debug" and a screenshot of the plotted mesh after starting FEAP with "Icrack_debug".

[Prof. R.L. Taylor]

Your main problem is the ordering of the nodes on the blocks.  They must be numbered anti-clockwise (Right Hand Rule) around the blocks.
The TIE COOR does bad things and probably is not needed.

My suggestion is to create the mesh with a standard feap element and then use OUTM to get the mesh you use for analysis.  You can optimize the profile before doing the OUTM if necessary.

I attach my revised file

[Jfrisch]

Thanks for your advice. I haven't thought about the node ordering being consistent. I'll give it a try.

Why does the TIE COOR bad things? Because I added this TIE-command to tie the node at the crack tip itself so that on the edge between region 3 and 4 all nodes except for the cracktip-node are split and all other edges between regions are tied together.

The usage of OUTM sounds quite nice to overcome issues caused by the block or any tie commands after I got a working mesh - thats a great advice, thanks for that!