My last model in RSA was a concrete strucure with a lot of walls and floors. So, the most annoying stage of process was meshing: quite slow (however Coons was used), quite unreliable and so on. Because of necessary to implement a lot of correction in model, I had to remesh it many times. And each time it was about 10 minutes in quite modern hardware (Intel Core i7-2600 processor (HT was switching off), 16 GB of RAM, 7200 rpm hard drive) for unfreezed mesh (with freezed of course not so long, but not quick too). The FE's count about 65000...70000.
So, I think that it’s quite necessary to improve meshing process, for example making it multithreading (like you, guys, made multithreading solver, which one is wonderful). So, this enhancement can save a lot of time for us!.
And related problem. I have worked in link Revit Structure -> RSA to transfer building structure geometry. I have created special calculation model, not structural, with great accuracy and avoiding any sources of problems (for example, walls were aligned by axis, etc.). But for my surprise, I have obtained model with a little inaccuracies in coordinates (I have used genius Rafal’s Excel-based macros from message 5 of this topic http://forums.autodesk.com/t5/Autodesk-Robot-Struc
It’s quite little but… In the result in have always obtain “Incoherent mash problem…”. Have you any advices how I can avoid this situation in future?
Thank you in advance!
I have had quite a lot of problems similar to yours, when I bring a model in from Revit. The panels may be out of alignment by 0.000001m, which does not even show up unless you change units and add many decimal points.
It then can take a considerable amount of time to get all of the panels to be on the exact co-ordinates required. I think that Robot should be rounding all co-ordinates (nodes, panels etc) to whatever accuracy is defined in user settings. This would greatly reduce these errors.
Are you using iterative meshing or kinematic constraints? Obviously the iterative method runs slower.
My advice is to say select all walls and mesh. Then you should get nice even mesh for your walls. Then select all slabs and mesh. The slab ondes should then mesh to the previously defined wall nodes. Sometimes, after meshing the slabs you need to go back and re-mesh the walls. I've found this works quite well, and is much faster than trying to get Robot to mesh all panels simultaneously.
>Are you using iterative meshing or kinematic constraints? Obviously the iterative method runs slower.
I have to use only interative meshing, because right now I use RSA only for meshing and loads defining. Then using my own little utility (http://dwg.ru/dnl/7096) I transfer FE model into old but reliable Ukrainian FEA software LIRA (http://www.liraland.ru/). Reason, why I act in this way are:
- Some little problems in RC module in RSA 2013 for Russian code. After issuing SP2 things are much better, but unfortunately aren't problemless.
- Very long time to calculate and obtain dynamic's results in RSA. For example, for Modal Analisys of this model (69000 FE), it takes about an 3/4 of hour only to calculate ("Model consolidation...", "Applying of loads..." and other) on computer which one I've described in message #1 of this topic. On the other hand, it takes only 30 minutes to calculate Modal on LIRA (on the 4 years old 32-bit laptop). Nothing to say, it's quite strange and significant difference and I hope ADSK guys must pay attention for RSA performance on large models.
>My advice is to say select all walls and mesh. Then you should get nice even mesh for your walls. Then select all slabs and mesh. The slab ondes should then mesh to the previously defined wall nodes. Sometimes, after meshing the slabs you need to go back and re-mesh the walls. I've found this works quite well, and is much faster than trying to get Robot to mesh all panels simultaneously.
Thanks a lot, Tony! I will try your method for meshing as soon as possible!
Please, find here: http://files.ramboll.ru/download.php?id=506066041&
You can find password for 7z archive in private message.
But this model is corrected in most problem places. Unfortunately, I haven't saved uncorrected model...
But some walls and floors still has inaccurancies!
Thank you in advance!
Take a look at geometry of panels 553 569
Meshed model : https://www.filemail.com/dl.aspx?id=HTEADHGVSGJYEE
Password in PM.
Dear Rafal, thanks a lot! Unfortunately, I can't receive meshed files (please, find reasons in Private)
Yes, this element with incorrect geomenry (it was created like wall in RST, than opening with width = wall's width, so invalid contour occured).
But you know, the most problematic place in model here:
Also, I have an issue with linear load. I have redefine it RSA, but it still exist.