Please kindly check you equations & Codes carefully. In iterative methods, the convergence mostly varies due to initial guess values as well as number of iterations.

thanks a lot i'll do it

Mr. Ben Tra, the convergence of solution depends on many factor which includes element size, mesh density, mesh dimensions and convergnece criteria. During calculation, software used to verify the unbalancing forces, the unbalancing forces are carried forwarded to next interations, so the one way is to increase the number of interations. then, to get better accuracy, you need to define the convergnece criteria i.e. either in terms of displaccement 0.005 mm or force 0.01 kN...in your case, better increase the number of iterations..ok all the best in your research..

Mr. Karthigeyan i thank you very much for your detailed response but I don't understand (convergence displacement criteria i.e. either in terms of displaccement 0.005 mm ) bécause I'm simulating an incompressible turbulent flow in a pipe through an obstacle

Hi,

Without knowing more details of your simulation it is difficult to comment, anyhow assuming you are doing a steady-state simulation and your mesh quality is high:

1. Monitor quantities that are important for your simulation e.g. via probes etc. If they do not change then you can consider your solution is converged.

2. Refining mesh may lead to this behavior sometimes. In short, coarser mesh tends to blend the unsteady region of the domain.

3. You can start the simulation with first-order scheme. check if you can get convergence, then switch to second-order.

4. Changing the Solver and its setting are also might help. e.g. For Simple type solvers you have relaxation factors, for Coupled solver you have pseudo time step value.

Again, without knowing your case it is difficult to provide more remedies for your problem.

Mr. Shervin Shahriari thank you very much for your interest in my problem.

the case is as follows: I'm comparing results of turbulent flow in a pipe through several flowmeters. i reached an orthogonal quality mesh of 0.80 and the solution converged without a problem, but when i changed the flowmeters i reached a quality mesh of 0.84 the solution didn't converge even i took the same physical conditions. about the simulation i'm using k-omega sst second-order simple shame as fluid I'mspecification using water inlet velocity and pressure outlet with intensity and hydraulic diameter as a pecification method .

Dear Ben

could you let me know about the chosen turbulence model and value of Y+ in your simulation?

Regards

Mr. Hamed, the turbulence model chosen for a Reynolds number od 24880 is k-w sst (it gave me better accuracy than realizable k-e ) the specification method is for boundary conditions is intensity and hydraulic diameter. velocity inlet and pressure outlet

thanks very much

Ben,

Achieving convergence with SST model is difficult (based on my personal experience) compared to other models like k-e or SA. Again, back to my points, monitor values that are important to you (velocity, pressure etc.). If they do not change you can consider your simulation converged. Another point is to use first-order and then switch to second order. This way you will have a better initial condition for the second simulation.