Haven't reach convergence even after 2000 iterations

Project Support
1

Hey everyone,

I’m having issue with convergence. I’m using Academic License so if you’d like to see the simulation, let me know.

My simulation details:

  • Steady-state analysis
  • Incompressible External Flow
  • Turbomachine
  • Boundary Conditions: Velocity inlet, Pressure Outlet and Symmetry at all sides.
  • MRF Rotating Zone
  • Standard Mesher

I think, I already have a mesh with fairly nice metrics as you may see in the pictures below. About 8M cells. But, when I run the simulation, the values are not stabilizing. I’ve attached some plots down here from my latest simulation as well. The important parameter for me is torque.

image
image
image

MOMENTS PLOT

image
image1056×500 57 KB

FORCES PLOT

image
image1057×497 45.1 KB

RESIDUALS

image
image1075×486 35.5 KB

I’m currently running a 3000 s simulation. Do you think it will converge? Does running above 2000 iterations make sense? Previously, did 1000 and 2000 s. Is there a way I can improve convergence? I did notice that my y+ has been decreasing when I’m using 2 BL instead of 1. Do you think perhaps this has something to do with convergence as well? Honestly, I want to optimize my simulation runtime. Running it for 2000 s did consume almost 100 cores. Can someone guide me on this?

Thanks in advance.

2

Latest Update:

My moments plot look like this after 3000 iterations. It’s converged right? How can I reduce the simulation run?

image
image1077×470 21.5 KB

3

Hi @edwintjy thanks for posting on the forum! Sorry for the late reply here

The last plot you’ve shared looks converged, however maybe a better tracker for that would be some sort of area average on the inlets/outlets.

Also, check out these resources on simulation convergence:

Since your mesh’s quality looks alright, maybe there’s an issue with your boundary conditions or some other parameters! If possible, please share your project’s URL for us to take a closer look :slight_smile:

Best,
Igor