[medaouarwalid]

hello, i'am new in simulation . after100 itertations, this is the residuals results, what do you think about te convergence and, why always the energy residual is very low compared to the other residuals ?

[lcarasik]

Originally Posted by medaouarwalid

hello, i'am new in simulation . after100 itertations, this is the residuals results, what do you think about te convergence and, why always the energy residual is very low compared to the other residuals ?

You should run it longer. The residuals are still decreasing and likely the quantities of interest are changing. Usually, I run it until the residuals are oscillating around low residual value.

Also, please do not post FLUENT questions in the main forum. Please keep these questions to the other forum.

[medaouarwalid]

Originally Posted by lcarasik

You should run it longer. The residuals are still decreasing and likely the quantities of interest are changing. Usually, I run it until the residuals are oscillating around low residual value.

Also, please do not post FLUENT questions in the main forum. Please keep these questions to the other forum.

Thank you and i appologize

[lcarasik]

You likely want to monitor a few quantities such as velocity, pressure, temperature at a few different points in the domain as well. Those can be used in addition to monitoring residuals for numerical convergence.

[BlnPhoenix]

Energy residual is lower because solution is approached faster than other quantities, perhaps because intial guess is closer to what the simulation calculates as solution. I would run also longer and carefully inspect other flow characteristics like change in mean velocity on a plane etc. to judge convergence. Also inspect whats going on with reversed flow at outlet. This can also go away with further iterations.

[medaouarwalid]

Originally Posted by BlnPhoenix

Energy residual is lower because solution is approached faster than other quantities, perhaps because intial guess is closer to what the simulation calculates as solution. I would rund also longer and carefully inspect other flow characteristics like change in mean velocity on a plane etc. to judge convergence. Also inspect whats going on with reversed flow at outlet. This can also go away with further iterations.

I posted a thread about that warning byt i dident recieve any good anwser, it is a nightmare for me, i tried to modify boundry conditions but the warnig still displaying, one of the members suggest that i extand the domain to give the flow mor Space to fully developpd but i guesse that i can not extend it more than this because the dimensions are the same as the real domain.

[BlnPhoenix]

Let it run further and see what happens to the reversed flow. Then if you think you achieved convergence, try to find out where in your domain reversed flow is present. Reversed flow often occurs if e.g. a vortex extends over your boundary, so that the flow re-enters the domain from outside because of this vortex. If this is the case, extension of Boundary is usually the way to get rid of the problem. I saw that you try to model a cylinder. Is this cylinder a technical device with fixed diameter and length or are you modelling release into atmosphere? When the later, what speaks agains increasing the cylinders diameter..

Good luck

[medaouarwalid]

Originally Posted by BlnPhoenix

Let it run further and see what happens to the reversed flow. Then if you think you achieved convergence, try to find out where in your domain reversed flow is present. Reversed flow often occurs if e.g. a vortex extends over your boundary, so that the flow re-enters the domain from outside because of this vortex. If this is the case, extension of Boundary is usually the way to get rid of the problem. I saw that you try to model acylinder. Is this cylinder a technical device with fixed diameter and length or are you modelling release into atmosphere? When the later, what speaks agains increasing the cylinders diameter..

Good luck

The cylindre is a room, i chose the cylindre shape and not a parallelepipede to gain little time of calcul in the corners. And yes i am modeling an air jet release into atmosphère . It is a problem of cooling and heating

[BlnPhoenix]

I see. But your room has walls, right? Or i'm i missing something here... If your room has walls, then Pressure Outlet does not look correct as BC. You should use wall BC instead and have the top as pressure outlet.

Edit: What i try to say is, model it as a parllelepipede (what a word ) that it is in real life. Use wall BC and Pressure Outlet at top. I think it is more realistic and probably gets also rid of the reversed flow if its occuring on this boundary.

[medaouarwalid]

Originally Posted by BlnPhoenix

I see. But your room has walls, right? Or i'm i missing something here... If your room has walls, then Pressure Outlet does not look correct as BC. You should use wall BC instead and have the top as pressure outlet.

Edit: What i try to say is, model it as a parllelepipede (what a word ) that it is in real life. Use wall BC and Pressure Outlet at top. I think it is more realistic and probably gets also rid of the reversed flow if its occuring on this boundary.

The reason why i chose pressure outlet rather than wall is because the jet i am modelling is a free jet, lets say that there is two type of jets, free jets where the effects of walls Is negligible and confined jets where we take the effefts of the walls in concideration.

[KaLium]

Originally Posted by lcarasik

You likely want to monitor a few quantities such as velocity, pressure, temperature at a few different points in the domain as well. Those can be used in addition to monitoring residuals for numerical convergence.

I agree. Monitoring the results is the best method for judging the convergence. It gives more information that the residuals.

[venturi]

This is a very good post about convergence:
Convergence