Particles passing through physical geometry in simulation
Dear Officer,
I conducted a granulator simulation in EDEM, but found that several particles would pass through the closed geometry and get into the outer layer of the granulator. I have checked the geometry and am sure that the geometry is fully closed without any interspaces. Here is an animation for your reference, you may find that some particles will leave the inner layer and accumulate at the bottom of the reactor.
Here are the parameters that might concern, the particle diameter is 1mm, the mesh size is 2.5R (0.25mm), and the time step size is set as auto. Are there any solutions for dissolving such problems?
Thank you for your great help in advance.
Best Regards,
Yu
Answers
-
Hi Yu,
It could be a few things and it's not immediately clear how you have meshed your model, but the mostly likely problem is the geometric constraints of your model are not being adhered to. You will notice that the particles only go through the geometry at the kneading element because in this zone particles can easily get caught between the kneading element and barrel wall if the particles are bigger than the clear gap. In a real TSG, when this kind of jamming happens, the machine will often stop as there is limited torque but in a simulation they just get pushed through the walls because of the unlimited force. So usually, the solution is to ensure your particles are small enough to fit through the gap.
Another possibility is that the mesh refinement is not small enough to create a very smooth barrel (your barrel may be more hexagonal than cylindrical) which cause the particles to stick in the first place so an improved mesh can often solve this.
JP0 -
J.P. Morrissey said:
Hi Yu,
It could be a few things and it's not immediately clear how you have meshed your model, but the mostly likely problem is the geometric constraints of your model are not being adhered to. You will notice that the particles only go through the geometry at the kneading element because in this zone particles can easily get caught between the kneading element and barrel wall if the particles are bigger than the clear gap. In a real TSG, when this kind of jamming happens, the machine will often stop as there is limited torque but in a simulation they just get pushed through the walls because of the unlimited force. So usually, the solution is to ensure your particles are small enough to fit through the gap.
Another possibility is that the mesh refinement is not small enough to create a very smooth barrel (your barrel may be more hexagonal than cylindrical) which cause the particles to stick in the first place so an improved mesh can often solve this.
JPDear JP,
Great thanks for your reply and help.
For the 1st solution, may I interpret that the particle size I set is so large that they would easily get stuck between kneading elements and inner walls, so they are unable to move with the kneading elements? May I suppose that the center of the particles is already outside the barrel wall, and the software recognizes that the particles shall not contact with kneading elements and move with them? I will change the particle size to a smaller one and attempt another time.
For the 2nd solution, I just import the geometry from SolidWorks to EDEM, and set the geometry import parameters as auto mesh size (min mesh scaling factor: 0.33, max mesh scaling factor: 4, max deviation scaling factor: 1, max angle: 15deg). Here is the screenshot for viewing from X-axis, and I suppose the cross-section is a circle rather than a polygon. From your perspective, do you think I need to re-import the geometry with different parameters or set the mesh size as a smaller one (like 2R or 1.5R)?
Looking forward to your reply.
Best Regards,
Yu
0 -
Hi Yu,
I don't know what speed your are running at, but in some cases these screws can be at speeds of up to 2000 RPM. It only takes a slight imperfection in the barrel for it to momentarily stick against the wall, meaning that when it gets hit by the kneading element there is a huge force. Because these are not real particles that can break, like those that would in the real system, they only thing that can happen is the particle is forced through the wall. From experience I've found that this usally happens along the intermeshing line where the two barrels meet. You can see in the attached screen shot that the PSD being used here to minimise this effect is small enough that the particles can fit between the screw and the wall.
In this image, you can also see what I mean by the mesh resolution:
The mesh is refined to use small triangular elements that are refined close to the shapr edge - there are no lon elements running along the barrel as can happen with the default EDEM settings.
This is an example of a poor mesh that will cause problems:
The defualt meshing options of EDEM cause these long elements which are really bad for contact detection for a TSG - this type of mesh should be avoided at all costs as it will increase the number of particles that pass through the walls.
Hope this helps.
JP
0 -
J.P. Morrissey said:
Hi Yu,
I don't know what speed your are running at, but in some cases these screws can be at speeds of up to 2000 RPM. It only takes a slight imperfection in the barrel for it to momentarily stick against the wall, meaning that when it gets hit by the kneading element there is a huge force. Because these are not real particles that can break, like those that would in the real system, they only thing that can happen is the particle is forced through the wall. From experience I've found that this usally happens along the intermeshing line where the two barrels meet. You can see in the attached screen shot that the PSD being used here to minimise this effect is small enough that the particles can fit between the screw and the wall.
In this image, you can also see what I mean by the mesh resolution:
The mesh is refined to use small triangular elements that are refined close to the shapr edge - there are no lon elements running along the barrel as can happen with the default EDEM settings.
This is an example of a poor mesh that will cause problems:
The defualt meshing options of EDEM cause these long elements which are really bad for contact detection for a TSG - this type of mesh should be avoided at all costs as it will increase the number of particles that pass through the walls.
Hope this helps.
JP
Dear JP,
Thank you for your kind clarifications.
The screw velocity I set is 450rpm, which is relatively slower than the value you mentioned.
May I know the complete spelling of PSD? As I assume it should be some technologies that can help avoid the crossing phenomenon in EDEM, but I am not so familiar with this function. If possible, could you please introduce it to me for some onboarding skills?
Here is the mesh in the geometry, I think it is quite dense, not as loose as the poor mesh you shared. If I would like to avoid bad contact detection, should I set the manual mesh size at a smaller value (default min size: 10mm, max size: 100mm, max deviation: 1e+03mm, max angle: 15deg) when importing into the geometry? The total length of the TSG is 500mm, may I wonder if you have any suggestions on the determination of these values?
Best Regards,
Yu
0 -
By PSD I simply meant particle size distribution and selecting something representative of your real powder and not too big that it affects the flow and granualtion behaviour in the TSG. Large particles that are similar in size to the free space between kneading elements will not give you realsitic results.
Yes, itsnot a bad mesh, a bit coarse but better than the oblong mesh that was previosuly the default in EDEM. As for good input parameters, the max size and max angle are the most likely to be the controlling factors. Set them as small as possible to give you a reasonable mesh without being excessively small - extra elements can slow down the simualtion, particluarly if it's for the screw.
I would suggest a max angle of 1 or 2 degrees and possibly a max size of 5-10mm. I can't remeber what was used in the above images. You will just need to explore.
JP
0 -
J.P. Morrissey said:
By PSD I simply meant particle size distribution and selecting something representative of your real powder and not too big that it affects the flow and granualtion behaviour in the TSG. Large particles that are similar in size to the free space between kneading elements will not give you realsitic results.
Yes, itsnot a bad mesh, a bit coarse but better than the oblong mesh that was previosuly the default in EDEM. As for good input parameters, the max size and max angle are the most likely to be the controlling factors. Set them as small as possible to give you a reasonable mesh without being excessively small - extra elements can slow down the simualtion, particluarly if it's for the screw.
I would suggest a max angle of 1 or 2 degrees and possibly a max size of 5-10mm. I can't remeber what was used in the above images. You will just need to explore.
JP
Thank you for your great guidance!!!!
Let me have a try and investigate whether there will be more promising results!!
Best Regards,
Yu
0