CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Standard initialization with an expected pressure (http://www.cfd-online.com/Forums/fluent/115239-standard-initialization-expected-pressure.html)

APUT March 26, 2013 17:08

Standard initialization with an expected pressure
 
I'm using the SIMPLE solver to find the area-weighted average of static pressure at the inlet of a laminar, steady-state, incompressible flow within a porous material. My boundary conditions are a velocity inlet and a pressure outlet.

From theory, I know what the pressure inlet should roughly be. However, if I try to initialize the problem with this theoretical pressure inlet, the solver gives me worse convergence and the pressure starts diverging at a high rate (though I did not keep running the simulation for long due to this initial observation).

If I standard initialize using the default values at the velocity inlet, the convergence trend is much more re-assuring and the values do not tend toward divergence. Also, hybrid initialization with default settings also gives me worse convergence.

Is there a better way to specify an expected solution at the initialization stage to speed up the convergence process? What are the exact steps?


Thank you!

oj.bulmer March 27, 2013 11:30

Often it takes a bit of experimentation to deduce the right approach for initialization strategy. As long as you get a converged solution, it doesn't exactly matter how you initialized it, but right initialization yields faster convergence.

In your case, in addition to standard initialization, you can also "patch" the inlet pressure to your inlet boundary.

Also, have a look at FMG initialization, wherein your geometry is meshed with virtually with coarse mesh and the Euler inviscid equations are solved. Be aware that FMG doesn't initialize turbulence quantities (k, \epsilon etc). So it is important to specify appropriate quantities at boundaries. The solution is then applied to successive meshes until the final fine mesh. Process takes few minutes but provides a good initialization.

Another way is to start having a coarse mesh solution as initial solution. This is possible in CFX, but not sure about FLUENT.

OJ

jalay September 4, 2013 02:18

i get smooth curve while going with hybrid initialization when i go for air in with standard initialization it continuously converge diverse ..
why this would be happening?


All times are GMT -4. The time now is 12:43.