[O'Neill1976]

Dear CFX-Experts,


I am simulating a flow through a pipe which has three inlets and one outlet. The simulation is steady-state, incompressibel and isothermal.

I first ran the simulation on a coarse grid (~100 000 cells). The solution converged after approx. 50 minutes.

After refining the grid (now ~ 900 000 cells) convergence is not achieved (see attached picture). I know that the computational effort scales significantly with the number of cells but my simulation is now running for more than 24 hours and still has not converged (convergence criterion 1E-6). Is this reasonable?

I started the simulation with a physical timestep of 1E-5 seconds and switched to a local timestep (local timestep factor = 5) after 50 iterations.

The simulation has the same setup as the simulation on the coarse mesh.

I used an automatic domain initialization.
What can be done to make the solution converge?



I am very grateful for your answers.
O'Neill1976

[ghorrocks]

FAQ number 1 covers exactly this issue: https://www.cfd-online.com/Wiki/Ansy...gence_criteria