turbulent viscosity limited to viscosity ratio of 1.000000e+05

[Sylnet]

Thanks I imagined that he was not "very good" to get this load loss...in fact the accounts don't come back much! I didn't think, or rather, I forgot that the cfd doesn't do much with friction!

I didn't see the point in front of 60... after 6 hours of study the head goes a little....:biggrin:

Yes, the values are definitely better, I have reported much larger values, but this is obvious since the air has a lower viscosity the speed profiles extend more. I would extend to the sides a little more, as well as behind the bulb... then it would be to see the turbulence, maybe after 4 m it will be sufficiently low to avoid recirculations.

cmq agree, with an image it would be much easier to understand the situation

 

[Random86]

I attach image of the domain.

 

[Sylnet]

the mesh so it seems ok, even the distance in front and behind the bulb should be ok. Do you still have problems? or now works better? in case of problems still try to put a much lower inlet speed, like 5 m/s, so you could maybe check if the problem of a possible non-convergence is due to mesh or the outlet still too close

 

[Random86]

I think I solved the problem in case with the domain presented above. Now I'm doing tests in the case of split domain with transition set to roll to turbulent. However, while in the case of single domain and I count all turbulent I can get to a solution that is what I expected, in the case of separate domains return to have the problem. I can't figure out whether it's a condition problem set to the interface (which I defined internal) or what else. The mesh is basically the same in the two simulations.

 

[Spes]

Good evening,
I'm a student of bari tech. the case I'm dealing with is quite similar to what you discussed. I have, at the moment, two bulbs, of equal size and profile, one in the tail to the other. I isolated a generator and built a 2d grid, in gambit, which I would like to solve in fluent, with solver axisymmetric. while using all the shortcomings on beds, however already suggested by the engineers who support me, the limit of the turbulent viscosity ratio is punctually reached, with the progress of iterations, by an increasing number of cells.
to instabilize the calculation, probably, the high size of the domain (every bulb is about 8 meters long and the entrance, of the domain, is from the exit 64 meters) in front of a need to totally solve the flow (from the laminar substrate up to the totally developed turbulent flow).
I believe that the solution is not immediate and I hope in some advice, which I am searching from every source.

 

[Random86]

check the boundary conditions you have imposed and then check that the limit layer mesh is correct. especially for the elements of the first cell.

 

[Spes]

you are saying well, in the sense that they are exactly the aspects you mention those that usually cause the above problem. Unfortunately for the case in question I have already coined more volts the aforementioned aspects, focusing on grow factor, height of the first cell to wall and so on.. .
However, it would seem, somehow, that if he could leave; Tonight the pc is traveling with the right turbulence length.
a curiosity: but what solutor did you use for viscosity? Did you lower the sub-relaxing factors? why relying on the coupled quano is not high-speed compress fluid we are dealing with but water?
But I hope you will give me your interpretation. I'm not hiding that I'm finding not a few differences between teachers, engineers, fluent help and so on.. .

 

[Random86]

the coupled helps solve the problem on the turbulent viscosity, but for me it is a little risk. Consider that fluent is a program that seeks in every way to force convergence. with the coupled solutor the holes even more and risk to find solutions not very consistent. It also takes much more time than simple, which is what I used. Try to check that gambit doesn't eat
some cell element of layer limit to wall. you can also see it from fluent: view y+ and see that they are all under 4 or 5. the problem however arises from the mesh and the conditions imposed. What model do you use then?? have you activated enhanced wall treatment? ?
then, even if it definitely works, try to make a mesh only tetraedrica for potential analysis.
In this way you will definitely take away the doubts that the pure cad model has no problem.

 

[Spes]

I used the simple too. with the coupled jumped into the air all :s
I am working with a y+ almost equal to 0.7 (value desired as I want to solve the laminar substrate).
I tried to use the enhanced wall treatment but without positive outcome. in particular the solutor k-epsilon fails to find solution.
the k-omega sst at the moment turned out to be the only stable. I am looking, with what it acquired, to find confirmation, to the data obtained, through other solutors (k-omega standard, the same k-epsilon) but at the moment I have not had any positive outcome.
I think, in any case, that I'm missing something, not very immediate!

 

[daniele.arlotti]

Good day to all,
I am a student of the Milan Polytechnic (energetic) and write in this forum because I also had problems regarding the turbulent vicosity ratio:
I am simulating a wind blade trunk with a profile du-97-w-300 characterized by "blunt trailing edge" and I created the mesh in icem by adding vortic generators on extradox (vortex generator vgs) in order to increase the stall angle of the shovel; the model of turbulence that I use is the spalart-allmaras considering the compress length.
as density imposed ideal gas and as viscosity use sutherland model and as equations all to the second order (use density-based approach).
performing a fmg (full multigrid initialization ) and launching simulation by "solution" I get the message in the image I attach (I also did the display of the contour of the viscosity ratio in the cells where this problem arises )Can anyone help me?
nb: the problem is found after making the adapting on fluent in the region around the vgs
Thank you very much
Daniele claws