There are no standard fluid cells in the computation

MFiX How to
#1

Hello, everyone!
I am encountering the following error when simulating pure fluid flow into a parallel plate microchannel:
NUMBER OF CELLS = 648000
NUMBER OF STANDARD CELLS = 0 ( 0.00 % of Total)
NUMBER OF CUT CELLS = 0 ( 0.00 % of Total)
NUMBER OF FLUID CELLS = 0 ( 0.00 % of Total)
NUMBER OF BLOCKED CELLS = 648000 (100.00 % of Total)
Error from cartesian_grid/vtk_out.f:2563
FATAL Error: There are no standard fluid cells in the computation.

I have set the grid size to 50 microns with parallel plates that fit snugly on the top and bottom of the full domain. What puzzles me is that when I leave a gap between the parallel plate and the top and bottom of the region, I find that the above error will no longer exist. Is it a problem with the grid settings? Since I want to explore some parameters under this mesh size, so it is not convenient to change the mesh size, I would like to know if it is possible to solve this problem by changing some meshing parameters while the parallel plates are tightly fit to the top and bottom of the full region? Here is my case, I hope you can make some suggestions, thanks!

1
2
fluid_channel.zip (386.8 KB)

#2

If your geometry is rectangular, then you should not use STL files, just use regular regions and assign a NSW BC to them.

#3

Thank you very much for your reply, Jeff. Do you mean that you don’t need to check the “select facets(stl)” box in the “regions” column? And in the “Boundary conditions” column, do you only need to set no-slip boundary conditions on the top, bottom, front and back of the whole region?If so, don’t I need to set the boundary conditions between the fluid and my parallel plate (box)?

1722903396(1)
1722903396(1)743×740 35.7 KB

1722903455(1)
1722903455(1)736×608 28.1 KB

#4

I have not looked at your file, but based on the picture, my suggestion is to define 4 separate regions (one for each box) by adjusting xmin, xmax, ymin,ymax, zmin,zmax. Then you can define a boundary condition for each region and set it to no-slip wall. You don’t need to define any box in the geometry pool.

#5

Thank you very much for your reply jeff. Your advice was very helpful. Another problem that really bothered me was that I needed to perforate the board, the diameter of the hole was 300 microns, and let the fluid flow in through the hole in the board. The plate is fitted to the periphery of the area, both of which are non-slip boundaries. The model works with a 100 micron grid, but a 50 micron grid diverges. Why?Is the overlap of my slab wall and zone wall the cause of this problem

1
1876×446 62.4 KB

#6

Not sure. How does your mesh look like? You may have to play with the mesh tolerances at such tiny scales.

#7

Hi jeff, thank you very much for your reply. My mesh is a uniform orthogonal mesh of 50μm50μm50μm. I have tried to increase the mesh tolerances to 0.1, but they are still not convergent, and the following errors appear:
Outflow BC defined: IJK_P_g remaining undefined.
t= 0.000000 Wrote SPx: 1,2,3,4,5,6,7,8,9; .RES;
t= 0.0000 Dt= 0.1000E-05 NIT= 12 Sm= 0.0000 CPU= 21.s
Mass imbalance%(0,MMAX): 0.7670E-07
Timestep walltime, fluid solver: 21.016 s
Recovered: Dt= 0.90000E-06 :slight_smile:
Recovered: Dt= 0.81000E-06 :slight_smile:
Recovered: Dt= 0.72900E-06 :slight_smile:
Recovered: Dt= 0.65610E-06 :slight_smile:
Recovered: Dt= 0.59049E-06 :slight_smile:
Recovered: Dt= 0.53144E-06 :slight_smile:
Recovered: Dt= 0.47830E-06 :slight_smile:
Recovered: Dt= 0.43047E-06 :slight_smile:
Recovered: Dt= 0.38742E-06 :slight_smile:
Recovered: Dt= 0.34868E-06 :slight_smile:
Recovered: Dt= 0.31381E-06 :slight_smile:
Recovered: Dt= 0.28243E-06 :slight_smile:
Recovered: Dt= 0.25419E-06 :slight_smile:
Recovered: Dt= 0.22877E-06 :slight_smile:
Recovered: Dt= 0.20589E-06 :slight_smile:
Recovered: Dt= 0.18530E-06 :slight_smile:
Recovered: Dt= 0.16677E-06 :slight_smile:
Recovered: Dt= 0.15009E-06 :slight_smile:
Recovered: Dt= 0.13509E-06 :slight_smile:
Recovered: Dt= 0.12158E-06 :slight_smile:
Recovered: Dt= 0.10942E-06 :slight_smile:
Recovered: Dt= 0.98477E-07 :slight_smile:
Recovered: Dt= 0.88629E-07 :slight_smile:
Recovered: Dt= 0.79766E-07 :slight_smile:
Recovered: Dt= 0.71790E-07 :slight_smile:
Recovered: Dt= 0.64611E-07 :slight_smile:
Recovered: Dt= 0.58150E-07 :slight_smile:
Recovered: Dt= 0.52335E-07 :slight_smile:
Recovered: Dt= 0.47101E-07 :slight_smile:
Recovered: Dt= 0.42391E-07 :slight_smile:
Recovered: Dt= 0.38152E-07 :slight_smile:
Recovered: Dt= 0.34337E-07 :slight_smile:
Recovered: Dt= 0.30903E-07 :slight_smile:
Recovered: Dt= 0.27813E-07 :slight_smile:
Recovered: Dt= 0.25032E-07 :slight_smile:
Recovered: Dt= 0.22528E-07 :slight_smile:
Recovered: Dt= 0.20276E-07 :slight_smile:
Recovered: Dt= 0.18248E-07 :slight_smile:
Recovered: Dt= 0.16423E-07 :slight_smile:
Recovered: Dt= 0.14781E-07 :slight_smile:
Recovered: Dt= 0.13303E-07 :slight_smile:
Recovered: Dt= 0.11973E-07 :slight_smile:
Recovered: Dt= 0.10775E-07 :slight_smile:
Recovered: Dt= 0.96977E-08 :slight_smile:
Recovered: Dt= 0.87280E-08 :slight_smile:
Recovered: Dt= 0.78552E-08 :slight_smile:
Recovered: Dt= 0.70697E-08 :slight_smile:
Recovered: Dt= 0.63627E-08 :slight_smile:
Recovered: Dt= 0.57264E-08 :slight_smile:
Recovered: Dt= 0.51538E-08 :slight_smile:
Recovered: Dt= 0.46384E-08 :slight_smile:
Recovered: Dt= 0.41746E-08 :slight_smile:
Recovered: Dt= 0.37571E-08 :slight_smile:
Recovered: Dt= 0.33814E-08 :slight_smile:
Recovered: Dt= 0.30433E-08 :slight_smile:
Recovered: Dt= 0.27389E-08 :slight_smile:
Recovered: Dt= 0.24650E-08 :slight_smile:
Recovered: Dt= 0.22185E-08 :slight_smile:
Recovered: Dt= 0.19967E-08 :slight_smile:
Recovered: Dt= 0.17970E-08 :slight_smile:
Recovered: Dt= 0.16173E-08 :slight_smile:
Recovered: Dt= 0.14556E-08 :slight_smile:
Recovered: Dt= 0.13100E-08 :slight_smile:
Recovered: Dt= 0.11790E-08 :slight_smile:
Recovered: Dt= 0.10611E-08 :slight_smile:
DT < DT_MIN. Recovery not possible!
This model can run in a 62.5μm grid, but I want to compare it with the data obtained in a 50μm grid, so I want to know how to make the model run successfully in a 50μm grid. The following is my case, I hope you can give me some advice, thank you
pl4pores0.5.zip (607.6 KB)

#8

It runs better when I increase the normal distance tolerance:

image
image824×618 67.9 KB

#9

Thank you very much for your reply, jeff. this advice is very useful for me. I have another problem that needs to be solved, can you please take a look at it, thanks.https://mfix.netl.doe.gov/forum/t/program-received-signal-sigfpe-floating-point-exception/5774?u=zxc

#10

Hello jeff! I set up two regions as no-slip walls as you suggested. The other confusing thing is that the size of these two regions (boxes) and the distance to the wall changes with the size of the mesh as shown below. It seems that the distance of these two regions to the wall and their widths can only be integer multiples of the grid size and will not cut the mesh, why is that?Does this have anything to do with the fact that the tops of my two box areas overlap with the top of the full domain

1
1580×551 104 KB

2
2939×711 266 KB

The real size of my model is shown below:
3
3261×501 5.52 KB

#11

When you use a rectangular region, the mesh is not cut, but the region snaps to the closest grid line. You need to adjust the grid spacing so it aligns with the regions.

The region does not coincide with the boundary mesh
#12

Ok, thank you very much for your answer jeff. this suggestion was very useful for me and I have solved the problem!