[checasino]

Hi guys,
in my new job they asked me to perform a CFD analysis with OpenFoam for a customized duct (almost rectangular) for HVAC application on a coach of a train.The duct has 1 inlet, 1 outlet (at the end of the duct itself) and mutliple outlets at the bottom (ventilation grids).The flow of air is turbulent and , as an initial approximation, I am performing steady state simulations. As said until now I performed a steady simulation using algorithm SIMPLE, imposing the boundary and initial conditions, other than the simplified physical and momentum models. Notice that the duct is symmetric in the longitudinal direction, so the model was halved.
After imposing the number of iterations, the simulation start and SIMPLE gives the following message:
SIMPLE: Convergence criteria found
p: tolerance 0.01
U: tolerance 0.001
"(k|epsilon|omega|f|v2)": tolerance 0.001
But when it converges (at iteration 77) the results are unfeasible and I cannot understand the reason.
I am gonna upload the 0,constant and system files.

[tomf]

Hi,

My guess is that it is because your convergence criteria are not strict enough:

SIMPLE: Convergence criteria found
p: tolerance 0.01
U: tolerance 0.001
"(k|epsilon|omega|f|v2)": tolerance 0.001

I would go at least one order of magnitude stricter, but more importantly I would monitor some flow rates through openings or forces to see if those converge. Just using residuals alone is typically not enough.

Cheers,
Tom

[checasino]

Originally Posted by tomf

Hi,

My guess is that it is because your convergence criteria are not strict enough:



I would go at least one order of magnitude stricter, but more importantly I would monitor some flow rates through openings or forces to see if those converge. Just using residuals alone is typically not enough.

Cheers,
Tom

Hi,
thanks for your suggestion.
Since I am very new in this kind of analysis, are there any documentations you advice on how to set those parameters for a good resolution of the simulation? I know those "parameters" (as tolerances, relaxationfactors ....) are in fvSolution, their meaning, but still I don't have any sensibility on their set up. So if you could suggest me some readings/articles/journals/guides I would appreciate.

cheers;
Silvia

[tomf]

Hi Silvia,

There is not a one solution fits all setup I am afraid. But as far as reading is concerned:

Computational Methods for Fluid Dynamics by Joel H. Ferziger , Milovan Perić , Robert L. Street

An Introduction to Computational Fluid Dynamics: The Finite Volume Method by H. Versteeg and W. Malalasekera

Those may cover more than you need now, but do provide a very good overview.

For convergence you would typically look at the evaluation of your parameter of interest (pressure drop/flow rate/force/moment/temperature) and see if that convergences within a certain tolerance (that you decide). The residuals do give you in indication about the numerical error, but that also depends on how accurate you model the flow: vortex shedding cannot be captured in steady state, so do not expect residuals for simpleFoam to drop to low values for a case with significant vortex shedding. You may have similar transient phenomena in you setup that hinder your residuals dropping below a certain threshold.

Best of luck,
Tom