Find answers, ask questions, and connect with our <br>community around the world.

Home Forums ANSYS Fluent Beginners to Advanced level Lesson 16 : Internal Flow | Solution not converging

• # Lesson 16 : Internal Flow | Solution not converging

Posted by on October 5, 2023 at 12:17 pm

Hi everyone,

I am trying to solve the globe valve simulation for a delta pressure of 1 bar. However, I am not getting a converged solution even after 5000 iterations.

these are my case setup parameters,

Pressure based solver (SIMPLE)

First order upwind

Used 0.5 under relaxation factor for momentum

used PRESTO! scheme for pressure

If anyone could share some suggestions on how to solve the non-convergence issue, it would be quite helpful

Best regards,

replied 3 months, 3 weeks ago 3 Members · 4 Replies
• 4 Replies
• ### Rahul

Member
October 5, 2023 at 12:39 pm

Convergence need not mean that the residuals always reduce below 1e-5 or so. Solution convergence depends on various factors including the mesh size, turbulence models, solution methods etc. If the residuals are sufficiently low or are periodic with a low value, like in your case, you should also evaluate other important flow parameters such as mass imbalance between the inlet and outlet, the mass flow rate/velocity, pressure etc. If any of these parameters are still not stable and the values are still increasing or decreasing with iterations, the solution cannot be considered converged and you should run more iterations. If these values are varying cyclically, you can take the average of these values as the results.
Regards
Flowthermolab Support

Member
October 5, 2023 at 6:02 pm

Thank you, Rahul

In my case, I gave two report definitions for the lift force on the disc and the outlet mass flow rate and they were really stable at the end. So I will take these values into account and consider the simulation to be successful. The convergence criteria were bugging me.

Thanks again, Rahul.

Best regards,

• ### Rahul

Member
October 6, 2023 at 7:55 am

One more point to note, if there is reverse flow at the outlet/inlet faces, residuals may not converge well. Or even though residuals may be low, correct results may be obtained only if the inlet or outlet domains are extended so that the reverse flow is properly resolved.
Best regards
Rahul

• ### Burak

Member
December 23, 2023 at 12:02 pm

I’d like to highlight another issue that may be preventing the achievement of a converged solution, based on my experience. I believe someone else might encounter a similar challenge and find a solution within this context.

Upon completing the simulations, I initially struggled to obtain a converged solution and explored various solver methods, including simple, simplec, piso, and coupled methods, adjusting under-relaxation factors, pseudo time step, and experimenting with different alternatives of first and second-order schemes. Despite these efforts, I couldn’t achieve convergence.

Eventually, I realized that a seemingly simple mistake related to ‘initialization’ was the main problem. As emphasized in the CFD Foundation course, initialization plays a crucial role in preventing divergence. I discovered that hybrid initialization was not sufficient for obtaining a converged solution. Instead, I preferred for standard initialization from the inlet, using the initial gauge pressure and related velocity. This adjustment led to a converged solution ðŸ™‚