Hi all
I am a beginner of the CFDEM. I have no idea on how to select the unresolved and resolved CFD-DEM approach for a project. As is stated in the paper, the resolved CFD-DEM approach is applicable to those cases where particle sizes are bigger than the computational grid, thus the particles are assumed to cover multiple (e.g., at least ten) cells. And the unresolved CFD-DEM approach is applicable to those cases where particle sizes are smaller than the computational grid, thus the particles are assumed to not completely fill a cell. It seems that the CFD grid size is quite essential for the selection of the unresolved and resolved approach. Does it mean that I must determine the CFD grid size before I select these two CFD-DEM approaches? But how can I determine the CFD grid size for a special project? Is there any relationship between the CFD grid size and the DEM particle size?
Specifically, I want to build a model to simulation the fluid-particle flow in a wedge fracture. The fracture model has a length of 100mm, a height of 25.4mm, a width of 1mm and 0.33mm at the inlet and outlet respectively. The particles size will be 0.4mm and 0.15mm. I'm now using the ANSYS ICEM CFD to mesh the fluid computational domain. So, how can I determine the CFD grid size? Which approach should I select? Unresolved or resolved?
Is there a good book/paper that explains how to select these two CFD-DEM approaches?
Any ideas/comments would be appreciated. Thanks.
Chong
alice | Thu, 12/06/2018 - 07:49
Hello Chong,
Hello Chong,
resolved CFD-DEM is a method that is suitable when you want to depict processes that involve only few particles (some hundreds, perhaps thousands) and want to resolve the flow behaviour between the particles. This method is computationally quite expensive (you already mentioned the approx. 10 cells in particle diameter, which easily leads to 100 thousands or some millions of cells for a small number of particles).
unresolved CFD-DEM can handle large amounts of particles, the flow field between the particles is not resolved, but the impact of the particles in one cell is averaged/summed up. This method is widely applied for industrial applications due to the number of particles. However, it is important that the particles do not cover the cells entirely, as the equations are not designed for such cases. Even if numerical barriers ensure that the calculation still yields results, this does not mean that the results reflect the true behaviour.
The width of the geometry that you state above is fairly small in comparison to the particle size (partially even smaller than the larger particles?) so you might still have to go for the resolved approach.If you have only few particles you can try to use local mesh refinement to represent them accordingly, but if there are many particles (i.e. the domain is packed with particles) you will end up with 100 millions of cells or more...
The alternative would be that you find a way to change the geometry such that unresolved CFD-DEM is applicable.
I hope this helps a bit at least!
Cheers,
Alice
lcraul | Thu, 01/24/2019 - 02:07
Thank you for you reply.
Thank you for you reply. Sorry to reply so late. There is something wrong with the geometry. My geometry is a wedge shaped fracture with an inlet width of 1mm and an outlet width of 0.34mm. The length and height of the fracture is 100mm and 25mm respectively. And the particle has two sizes: 0.33mm and 0.1386mm. In this suiation, the large particle size is just a little smaller than the outlet width. And in a finall state of my simulation, the geometry will be fully packed by particles, which means there may be tens of thousands to hundreds of thousands of particles. I creat this geometry based on my experimental equipment. So that I can compare the experimental results and the simulation results. So my questions are:
1) In the abovementioned case, is the unresloved approach still suitable?
2) Accoring to your answer, selection of resloved and unresloved approach should be based on the smallest size of geometry and the biggest particle size, right? In other words, if the biggest particle size is larger than the smallest geometry size, the resloved approach should be used. Otherwise, the unresloved approach.
3) In a general situation, how many times the smallest geometry larger than the biggest particle size should the unresloved approach be used? Such as five, ten or hundred. And how many times the biggest particle size larger than the smallest geometry should the resloved approach be used?
4) In the case that the smallest geometry is closed to the biggest particle size, how to decide?
5) If I use the resolved approach for my geometry, as you stated there may be 100 millions of cells or more, it is must be extremly time-consuming, right?
6) If the the resolved approach is used, the mesh size should be determined based on the biggest or the smallest particle? i.e. the mesh size should be 1/10 of the biggest particle or the smallest particle?
I'm new in this software, so there are too many questions. But I can't find the answer in the online document or the forum or the literature or your doctoral dissertation, so I have to ask you for help. Thank you for your patient. I am looking forward your reply.
Best wisehes,
Chong