running analysis

[Anonymous]

Im trying to run an analysis on an inventor assemby file and it has been running for about 6hours and the window for meshing prgress is still going. Right now it is on part 14 of 14 and says it is closing remaining voids: 618231 elements, 14884 faces. I have a AMD quad core processor computer running windows 7 64 bit which is only a few months old so I know this long analysis time is not due to a slow computer. Im wondering if it normally takes this long to run an analysis as this is the first one I have run. One of the parts when meshed would come up with errors due to not being watertight so I had to turn the meshing up to fine. My initial guess was that this is what is taking it so long, but I am unsure. Any help would be greatly appreciated.

[xli]

Hi,

From the message you provided: " ...closing remaining voids: 618231 elements" we know it is still in meshing phase. Nothing comes into analysis yet.

It looked not a too small models in terms of mesh size. Some times the structure is not easy to mesh as whole, especially some locations have many small and detailed geometry features that brings down overall mesh size and raises difficulty to mesher. Usually, showing failure message like "not being watertight" indicates there are such problems. 

Maybe you need to consider use shell or beams to model thin and narrowed parts in your model. this is not normal. Maybe a picture would share with some of our expert to make suggestions on meshing. Thanks.

-xli 

[S.LI]

how many parts are in your assemby file?

My guess is that most of meshing time are taken by part-part connection.

[Anonymous]

There are 26 parts all together.  I also had used shell elements for all thin parts and used mid plane meshing for them as well.  After about 12 hours it was still meshing and decided to stop it.  The project is a tunnel/bridge and the part it seems to have been stuck on was the top layer of soil.  Unfortunatly I had to shut the program down so Im pretty sure a lot of the settings were not saved.  I' tried attaching the file with no success, the size of the fem file is over 500 MB.  Is this a typical size fem file or could this be the reason the analysis is running so slow.  I will attach the inventor assembly file if anybody might have a suggestion as far as how to go about analyzing this structure.  Thankyou again for any help you can provide. 

[S.LI]

The attached .iam file could be opened by Inventor, since all parts file are missed.

26 parts are not too many to me.

There is one thing I'm not sure if you know: the mesh size could be set part by part. In another word, you can use fine mesh for some parts (such as the bridge) and coarse one for other parts (such as soil).

[xli]

Here is one of reasons fem file is larger than Inventor simulation model files: it include surface and solid meshes because Autodesk (Algor) Simulation's modeling is based on a specific mesh. So when you mesh is very fine fem file size could be very large. We do see some big model with 500M~1GB.

Specifying indivule part meshing size (use coarser mesh for some parts) might help in certain cases but may not in general resolve the problem if you do have some parts requiring fine mesh. Because mesher needs to maintain the consistence between parts for part matching as well.

One question: can you mesh and analysis this model in Inventor simulation? And if you can just share a snapshot of your model it can let us have a quess what difficulty could be, because we could not open your attached iam file. Thanks.

Another idea, if you change all you part to plate/shell with relative coarser mesh, we will  see if surface/shell mesher can make it or not. If that can reduce your fem model size, and share with us thanks.

-xli

[Anonymous]

This is not unusual for me. It seems time for the solid-meshing part of the analysis increases exponentially with the number of DoF. I've had reasonable large, but not huge, models (about 1 million DoF) meshing for over 24 hours. There's really nothing to do about, if you want to retain the mesh as it is.

Still, a general tip: Small, thin geometry usually causes insane solid mesh times. Try to avoid e.g. rebars in big concrete structures just as an example. Those thin parts will be first of all very hard to get a good surface mesh, but they will also require a [i]very[/i] long time to solid mesh since they are so fine, and the elements created will most likely be very ugly.

[xli]

I agree with Bjorn that solid meshing will get into difficulty for many models with some fine features or geometry complex connection between parts. What we can do is modifying models for improved (with some approximations) conditions besides playing mesh size to let mesher working in a model based matter. Go simpler model first then gradually back to as realistic as we can is a general strategy. 

-xli

[Anonymous]

After messing with it for quite awhile I was able to get all the parts surface and solid meshed pretty quickly except for a part simulating soil.  The soil seems to just get stuck on filling voids.  When I mesh this part by itself with all other parts deactivated it meshes completely in less then a minute.  It seems I was having a similiar problem with other parts as well.  When a part would mesh by itself it would take only a couple of seconds but then adding other parts to it the program would say that the part is not watertight.  I've also found that if I only increase the fineness of the mesh for that one part it is still not watertight, but when I increase the mesh for all other parts it seems to work. 

To better explain the project I am trying to analyze I attached a snapshot this time instead of the assembly file.  If anyone is familiar with the bridge in a backpack system, that is essentially what I am trying to analyze.  All of the tubes in the picture are concrete encased in a frp tube.  All of the tubes are very thin and I have been using plate element types for linear and shell for nonlinear for them and have had no issues.  This is the first time I have used this program and am now wondering if I'm going about this all wrong.  Thankyou again for all your input.