You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am attempting to run a DEM of a large imported mesh and am having some difficulty with getting the solver to detect collisions between the particles and the mesh. I am using a modified version of the SimpleChute example and an STL file that has dimensions of approximately 5m on a side but is made up of mostly a thin shell (dump body on a truck).
The first thing I noticed is that the algorithm for determining the grid size goes to 1 cell for this particular STL as it finds a very large radius (>100) and this drives the cell counts to 1 for this domain. Manually setting the grid size to a small number has some minimal success as it detects a very small fraction of the wall collisions but decreasing the grid size introduces issues with memory consumption. I haven't dug into the code but I am assuming that it uses some form of sdf on a grid to detect collisions and that getting a fine enough grid to detect the shape essentially just runs out of memory due to the dense grid?
I am wondering if it is possible to select an alternative form of collision detection for the walls?
I have included an example STL as a text file below.
I am attempting to run a DEM of a large imported mesh and am having some difficulty with getting the solver to detect collisions between the particles and the mesh. I am using a modified version of the SimpleChute example and an STL file that has dimensions of approximately 5m on a side but is made up of mostly a thin shell (dump body on a truck).
The first thing I noticed is that the algorithm for determining the grid size goes to 1 cell for this particular STL as it finds a very large radius (>100) and this drives the cell counts to 1 for this domain. Manually setting the grid size to a small number has some minimal success as it detects a very small fraction of the wall collisions but decreasing the grid size introduces issues with memory consumption. I haven't dug into the code but I am assuming that it uses some form of sdf on a grid to detect collisions and that getting a fine enough grid to detect the shape essentially just runs out of memory due to the dense grid?
I am wondering if it is possible to select an alternative form of collision detection for the walls?
I have included an example STL as a text file below.
underbody.txt
The text was updated successfully, but these errors were encountered: