Learning Forum

[Ivosp]

I'd like to simulate the inflow/flow around a box. The fluid (air) itself is separated into 2 fluid bodies due to simulations planned for later. The inner body is a sphere, the outer is a cuboid.
 

My setup:

Meshing

• Box as dead
• Inverse Quality: Max = 0.78 (only a few), Average = 0.02
• round about 2000000 cells
• I've already connected the congruent spherical surfaces of the 2 fluid bodies in the geometry (in SpaceClaim) and it's therefore treated as one (in simulation as interface). Therefore I used a conformal mesh.
• There is one velocity inlet (left side of the outer fluid body) and one pressure outlet (right side of the outer fluid body). All the other zones are walls.
• I answered the following questions as:
  • Geometry Type -> The geometry consists of both fluid and solid regions and/or voids
  • Change all fluid-fluid boundary types from 'wall' to 'internal'? -> Yes
  • Do you need to apply Share Topology? -> No (Since the 2 spherical surfaces are already treated as one due to the settings in SpaceClaim)
  • Enable Multizone Meshing? -> No
  • Will you cap openings and extract fluid regions? -> No
• I used a poly-hexcore mesh

Solution

• velocity inlet = 13 m/s
• I left the rest of the settings as default 
  • The congruent spherical surfaces, let's say I name them zone1 (from outer fluid body body1) and zone2 (from inner fluid body body2), are listed as Internal under Boundary Conditions as follows:
    • zone1-zone2-body1-body2
    • interior--body1
    • interior--body2
• The hybrid initialization was done after 10 iterations (with 7.3e-09)
• For my first attempt, I set the number of iterations to 1000.
• I clicked on Check Case... and it said No recommendations to make at this time.

But after 1000 iterations the scaled residuals are still too high . The whole problem seems not that complex to me, but I am still a beginner and I have absolutely no idea what could be the reason for this. Can anyone help me?

The scaled residuals stayed like this even after 1000 iterations...

 

[Ivosp]

 

 

For a better understanding, this is the model in Fluent Solver:

 

 

 

[Rob]

Plot contours of velocity around every 5 iterations for 50 iterations and look carefully at the output. Or put a couple of point monitors for velocity in the wake region. I suspect you've found a transient in the flow field from the way the residuals look, the monitors and contours will provide evidence of that.