model and simulation attached
I'm old school (2009 limited) so I can't open your model, but based on description: can you utilize symmetry? if it is simply 2 concentric cylinders you could set symmetry boundary conditions (i.e. top and right plane) so you only analyze 1/4 of the model. Then you can use the same number of total cells, but they will be smaller cells and better able to resolve the narrow gap.
Have you tried manually defining the narrow gap options to help resolve that area?
Thank you Chris for your replay.
1. At first this problem is symetric, but soon after ill have success in this attempt i will make the cylinders unconcentric and solve it as a bearing. so it might work, but it won't help my next simulation.
2. This is my attempt with Narrow Channel, no use...
3. While starting with 150X150X5 elements and refinement for narrow channels i can find fluid only cell but my computer is unable to continue more than 2 iterations and i'm out of memory. I think if i could change the elements shapes to un-square ratios it would run better.
what are my options now ?
Flow will only work with square elements.
Did you just use the small channel? or did you use the "resolve partial cells" slider on the Refining Cells tab?
What are you looking for as far as flow? Axial movement of oil in the gap? do you need the 5 cells along the axial direction? once it refines the mesh 8 times (max) to find the fluid all of those cells will get chopped in the axial direction also. If you can either reduce the length simulated or decrease the number of cells in the axial it will let you use more of the memory to resolve the fluid space.
How many cells total is it generating currently that you have memory issues?
1. I have used narrow channel (as in the picture) and also "Refine Partial cells" both become too large.
2. I'm looking for pressure. In this simulation i should result a constant pressure for the fluid. But, when ill change the cylinders to be not co-centric the analytical Ocvirk solution of the Reynolds equation for a short submerged journal bearing.
So right now i can use less than 5 elements along the axial direction but eventually no.
3. I can't tell exactly how many cells are too much, but this computer is fairly new and has 3 GB of RAM total.
I might check it later if you think its a problem with this specific station.
4. I guess running this on few computers at the same time should solve the memory problem ? I might be able to do so if you think its the only way.
5. I saw in the help section about initial mesh that there can be unstraight corners, isn't it ? i didn't have success with that.
3) if you look at the current mesh and only show the fluid cells (or partial cells) - how many are there axially? does the refinement give you enough cells along the axis to be realistic? (i.e. if you go from 5 to 4 along the axis you could go 165 on each edge with the same total cells, and when it refines to resolve the fluid you would have 32 along the axis because it split each cell 8 times during the refinement)
4) I haven't used the network solver, but my understanding is that only works for multiple models (batch solver).
5) That may be newer than 2009 as I've never seen anywhere that allows non-square corners.
on 2009 I typically limit myself to 2.2M cells on an i7e with 12GB of RAM, with only 3GB you're pretty limited on how big of a model you can run.