CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)

 andrea March 1, 2002 09:35

Convergence

I have some problems because the residuals oscillate around an average value . I have decreased relaxation factors of velocities and pressure. What else can i do?

Andrea

 Jing March 6, 2002 11:33

Re: Convergence

Please check if it is a unsteady flowfield. If it is, it is normal the residuals oscillate around an average value when you calculte it as a steady flow field.

 Andrea March 6, 2002 12:21

Re: Convergence

Andrea

 john March 6, 2002 23:17

Re: Convergence

Jing's response is quite possible. Several things may also be considered: 1. relax. factor for p is not small enough, have you tried <0.1 2. mesh is too fine, or bad quality, 3. wide open pressure boundary 4. you are too picky, the solution is good enough

 andrea March 7, 2002 11:50

Re: Convergence

Thank you for your answer,first. I can't understand what' s the meaning of "wide open pressure boundary": in our model we define a pressure boundary to simulate the flow exit in the open air (environment pressure); it takes quite a big area: do you think this could create problems? Why?

Andrea

 john March 9, 2002 01:33

Re: Convergence

Depend on how big and if inflow is induced at pressure boundary, and how stable this inflow is. Basically, when solving Navier Stoke eq. pressure boundary is not required, but velocity boundary is necessary. For some reasons, they allow you to define pressure boundary but internally still velocity to be applied at the pressure boundary, and mostlikely it is velocity gradient. When pressure boundary is used, and if you did not provide velocity boundary at other boundary as a stronger reference, how flow knows where it should go. Then, it get lost or get to endless solutions. Check if your flow pattern is stable, and if inflow occured, check it closely.

 All times are GMT -4. The time now is 01:55.