Is this solution converged?
2 Attachment(s)
Hi
can you please help me to assess if this solution is converged? This is simulation of wind tunnel nozzle. Boundary conditions as in given attachemtn? Can i consider this values of residuals as low enough since they are straight line? This is simulation with standard k-epsilon turbulance model. Any ideas how to improve resiuadals? Do you think no slip wall outside of the nozzle is a good idea? |
Cześć Michał,
There are several ways to assess convergence: - the level of residuals (for example 1e-3 for continuity, 1e-6 for energy) - mass and energy balance - tracking some monitor plot, for example area-weighted average of pressure |
I would say they are.
1. Your contours look nice, meaning the initial condition seems to be gone. 2. Your residuals are flat/horizontal. Regardless of how high/low they are, the solution probably is not changing at all. Even if it is wrong or poor quality result, you probably cannot improve it anymore. I don't really care what is the magnitude unless it is 1000. But if it is 1e-03, 1e-05, does not make much difference. Anyway you know it will not go any lower than what it is. If you are not convinced. It is a good idea to verify the flow hasn't changed. You can make the same contour screenshot at different iterations and visually inspect. More quantitatively you make a point monitor or line monitor or surface monitor and check the any or all of the pressure, temperature, velocity, and so on and verify that nothing changes. |
1 Attachment(s)
hi Guys,
thanks you so much for help - actually i have changed all oustide nozzle boundaries to pressoure outlet and it has resulted in pretty nice convergance (as attached). Right now i am sure my solution is quiet ok with Y+ at wall in bound 13-22. thanks |
Looks nice.
# of cells with reversed flow not changing is also a reassuring indicator that the solution is converged (i.e. always 358 faces). |
All times are GMT -4. The time now is 21:39. |