Dangerous build in triangle neighbor list

Submitted by alexander.polson on Tue, 06/15/2010 - 13:30

Hi Everyone

I am running a model of a drum with spheres in it. The drum is imported from an STL file. When I run this, I get the following warning repeatedly:

"WARNING: Dangerous build in triangle neighbor list."

Do I need to be concerned with this? Is there perhaps a problem with the STL file? I did notice that my spheres might be escaping the drum, but I need to double-check for a reason.

I could only find the following in the "fix_tri_neighlist.cpp" file:

//if a fix move/mesh/gran is registered, advise the user to increase skin safety (and eventually shorten time-step)
//if a fix pour is registered, advise the user to increase distance between insertion region and the mesh/gran wall

I think none of the above advice is applicable since I am not using the mentioned fixes in the model.

Thank you

Alexander

ckloss's picture

ckloss | Tue, 06/15/2010 - 13:47

Hi Alex,

that is most probably a bug - I recently discovered a bug in that routine that will be fixed in the next release - probably it is related to that.

If you mail me the input script and the STL file I will have a look.

Best,
Christoph

ckloss's picture

ckloss | Tue, 06/15/2010 - 16:39

Hi Alex,

some of your particles that you read into the simulation from the data file have an inital overlap with the wall - which is bad!
You should generate them in a way that they have a safe distance from the wall when they are read.
Please see if the error persists then (it hopefully shouldn't)

Christoph

raguelmoon's picture

raguelmoon | Wed, 06/16/2010 - 04:12

Hi Christoph,
What do you mean by safe distance from the wall? Should it be large or small gap?

Best..
Ram

Ram

ckloss's picture

ckloss | Wed, 06/16/2010 - 14:32

Hi ram,

there should be no initial overlap with the wall. In earlier versions you had to obey a certain distance from the wall to the insertion volume, but in 1.0.4, this is no longer necessary.

You only have to be careful when using fix pour/legacy because it generates the particles in a way that only the particle centers are in the insertion region, whereas the new fix pour/dev generates the particles in a way that they are completely contained within the insertion region

Christoph