The particle clump is made up by some overlapping spheres ,I have confirmed that the volume and mass can be caculated correctly in liggghts. Now I wonder IF THE INERTIA TENSOR CAN BE CACULATED CORRECTLY ? since it is concerned with shape and density distribution. It says the caculated inertia tensor is based on the mass center of the clump, does that means the coordinate origin is moved to the masscenter automaticly ,or should I give the spheres locations based on mass center ?
PS: does anyone knows how to find the masscenter of a irregular shape clump using MATLAB?
Dedicated to open source high performance scientific computing in fluid mechanics and particle science
A project by DCS Computing and CFDEMresearch
- About
-
Aspherix® solver and GUI
-
Aspherix® Features
- Particle shapes: convex, concave, fibers, boxes, cylinders, elipsoids...
- Advanced Multi-sphere: Resolved non-spherical particles
- Rigid body dynamics - 6DOF & MDB coupling
- Bonded Particles
- Powder compaction
- Deforming meshes & Resolved wear
- Post-processing, spatial and temporal averaging
- Particle attrition, simplified fluid forces, area evaluations
- Mass transfer and chemical reactions
-
Aspherix® Features
-
LIGGGHTS®
-
For everyone: LIGGGHTS®-PUBLIC
- Mesh import & moving mesh
- Conveyor model
- Stress analysis & Wear prediction
- Multi-sphere: Resolved non-spherical particles
- Non-resolved non-spherical particles
- Cohesion & Liquid Bridges
- Particle insertion & Packing generation
- Stress-controlled wall ("Servo wall")
- Heat transfer
- Particle growth & shrinkage
- SPH
- Electrostatics
- More Examples
-
For everyone: LIGGGHTS®-PUBLIC
-
CFDEM®coupling
- CFD
- Forums
- Contact
- DCS Computing
cstoltz | Tue, 11/24/2015 - 12:30
I haven't looked lately, but
I haven't looked lately, but I think it is in fact calculating correctly using a Monte Carlo algorithm. For a multisphere particle, it's pretty straightforward to generate a bunch of trial points within a rectilinear bounding box and calculate the fraction that fall within the bounds of the multisphere particle, thus providing the fraction of the rectilinear box occupied by the particle. Using the same set of points, you can then calculation the inertia tensor.
For an irregular shape in Matlab, if it is multispheres, you can use the technique I just outlined. You can also do something similar if you have a faceted shape, though it gets more complicated to determine whether or not a trial point is actually inside the geometric body. Check the Matlab file exchange to see if anyone has posted such a routine.
Regards,
Chris
zhou_yu | Wed, 11/25/2015 - 08:09
Thank you very much.
Thank you very much.
I have caculate the properties of a multisphere clump earlier in matlab, the algorithm is just as you said. The problem is the inertia tensor I get is based on a uncertain coordinate system, I.E.the origin is not on the mass center of the clump. But liggghts needs the inertia tensor based on mass center. So the most important problem is how to get the mass center in matlab with Monte Carlo algorithm.I confirm that is practically viable cause liggghts does it, but I just dont know how, so I give up this way.
A day or two ago, I used CT data of several true sand particles to calulate volume ,mass,inertiatensor, and then caculate these in liggghts, and found they have very little gap. This confirmed that liggghts can give the mass,volume and inertia tensor correctly,which is just what I need.
Thanks again.
ckloss | Thu, 12/10/2015 - 22:43
Hi all,
Hi all,
>>IF THE INERTIA TENSOR CAN BE CACULATED CORRECTLY
"Correctly" is of course a difficult word - but yes, it's using a Monte Carlo algorithm to get close to "correct" :-)
>>The problem is the inertia tensor I get is based on a uncertain coordinate system, I.E.the origin is not on the mass center of the clump. But liggghts needs the inertia tensor based on mass center
Get yourself a mechanics textbook! This is doable :-)
Best wishes
Christoph