Error: divergence detected after mesh refinement

  • 23 Views
  • Last Post 27 August 2019
  • Topic Is Solved
trippleD posted this 14 August 2019

Hello everybody,

I'm doing a steady state simulation of the flow in an annular gap. The inner cylinder is modelled as a rotating wall and the outer one is a stationary wall. The left and right side are stationary walls as well. All in all, it is a closed system. I'm using the realizable k-epsilon model with enhanced wall treatment. I'm using the pressure-based coupled solver.

I want to do the mesh refinement.  The first simulation with the original mesh works without any problems. For the refinement, i decrease the size of the elements and starts the simulation again with the refined mesh. But then, i get the following error:

Divergence detected in amg solver pressure coupled

Does anyone have an idea why the error occured and how to solve it? 

Thanks a lot for your help,

 

trippled

Order By: Standard | Newest | Votes
abenhadj posted this 14 August 2019

Try to interpolate from the obtained results of first coarse mesh.

Best regards,

Amine

trippleD posted this 15 August 2019

Hey Amine,

I have tried it, but I still got the same error and I don't know why

abenhadj posted this 15 August 2019

Check if the mesh is valid at first.

Best regards,

Amine

trippleD posted this 16 August 2019

I checked all the statistic values in Meshing. In my opionion, the statistic looks good. I'm just wondering why the statistic in Workbench Meshing (Aspect Ratio) is different to the check in fluent.

  • Orthogonal Quality:
    • Min: 0,99982
    • Max: 1
    • Average: 0,99999   
  • Aspect Ratio:
    • Min: 1,3871
    • Max: 2,146
    • Average: 1,5421
  • Element Quality
    • Min: 0,76243
    • Max: 0,94858
    • Average: 0,90365
  • Jacobian Ratio
    • Min: 1,0205
    • Max: 1,0463
    • Average: 1,0299
  • Warping Factor
    • Min: 6,4811e-015
    • Max: 8,9758e-003
    • Average: 8,3415e-006
  • Parallel Devision
    • Min: 1,193
    • Max: 1,238
    • Average: 1,2 
  • Maximum Corner Angle
    • Min: 90,597
    • Max: 91,441
    • Average: 90,601
  • Skewness
    • Min: 6,284e-003
    • Max: 1,6159e-002
    • Average: 6,6797e-003

I also check the mesh in fluent and report the quality. Nothing out of the ordinary. There is no error:

Mesh Quality:

  • Minimum Orthogonal Quality = 9.99824e-01

         (Orthogonal Quality ranges from 0 to 1, where values close to 0 correspond to low quality.)

 

  • Maximum Ortho Skew = 1.76141e-04

         (Ortho Skew ranges from 0 to 1, where values close to 1 correspond to low quality.)

 

 

  • Maximum Aspect Ratio = 2.83212e+00

 

Domain Extents:

x-coordinate: min (m) = -5.700000e-02, max (m) = 5.700000e-02

y-coordinate: min (m) = -5.700000e-02, max (m) = 5.700000e-02

z-coordinate: min (m) = 0.000000e+00, max (m) = 5.000000e-01

Volume statistics:

minimum volume (m3): 5.012315e-10

maximum volume (m3): 1.120107e-09

total volume (m3): 4.121465e-03

Face area statistics:

minimum face area (m2): 4.288502e-07

maximum face area (m2): 1.359810e-06

Checking mesh.........................

 

Done.

abenhadj posted this 16 August 2019

Slightly Different algorithm and allocation of neighborhood(but the theoretical formulation of the metrics are almost the same). Stick to the ones from Fluent.

Quality seems to be okay. You are using coupled solver with pseudo-transient? Can you please add more details?

Best regards,

Amine

trippleD posted this 16 August 2019

It is a steady state simulation with pressure-based, coupled solver with pseudo transient. 

At Soution Methods: I'm using First Order Upwind for all except for pressure. There i"m using PRESTO!

The outher settings are default. If you need more information, please tell me.

Thank you very much for the good help

abenhadj posted this 16 August 2019

Settings are okay. Can you set the time scale to 0.01/Angular Speed and try again

Best regards,

Amine

  • Liked by
  • trippleD
trippleD posted this 26 August 2019

Do you have an explanation in your words, what the option pseudo transient does? 

abenhadj posted this 27 August 2019

Adding an implicit under relaxation to the discretized equations which might enhance diagonal dominance.

Best regards,

Amine

Close