CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Turbulent viscosity in a riser (https://www.cfd-online.com/Forums/fluent/31291-turbulent-viscosity-riser.html)

 ap April 15, 2003 08:31

Turbulent viscosity in a riser

I'm modeling a gas-solid flow in a riser using FLUENT 6.0 Eulerian model together with kinetic theory for granular flows. I use RNG k-e with differential viscosity model activated for turbulence, and wall-functions for wall treatment. I tried to solve turbulence equations for both phases and for the mixture.

FLUENT always warns me that the turbulent viscosity has been limited to viscosity ratio. With a contour plot of the viscosity ratio, I determined that this happens in the center of the riser, where the solid concentration is low, so also solid molecular viscosity is low.

I tried to increase the viscosity ratio in the Limits window. This seems to solve the problem (FLUENT doesn't warn me anymore), but gas and solid velocities are underextimated and solid mass flux profile is not correct.

I also tried to change boundary conditions and to use second order discretization, but this just made the warn appear later.

I don't have convergence problem neither with nor without the limitation.

What do you suggest?

Thanks for any help.

 Alex Munoz April 15, 2003 11:17

Re: Turbulent viscosity in a riser

Hi

Refine your mesh in those region that you identify the turbulent viscosity ratio higher than 100000

Regards

Alex Munoz

 ap April 15, 2003 16:57

Re: Turbulent viscosity in a riser

I tried to refine the mesh, but the place where the limitation tooks place changes, so I'd have to refine the whole domain. This is not possible because I already have a high number of nodes.

I think the problem is in the boundary conditions of turbulence, but changing them just delays the problem and reduces the number of cells where the limitation took place.

P.S. In some simulation using the Mixture approach for turbulence I don't have the limitation, but I think it would be better to use the "Per phase" approach in my case. What do you think?

Thanks again

 Jongdae Kim April 15, 2003 17:30

Re: Turbulent viscosity in a riser

Why don't you use default solution schemes (i.e. usually 1st order scheme) at first. And then you may increase the order.

What kind of boundary condition do you use? If you use high turbulence intensity, you may have this warning also. Why don't you decrease turb. intensity?

 ap April 15, 2003 17:52

Re: Turbulent viscosity in a riser

I use the first order discretization scheme. I also tried to use the second order, but this didn't solved the problem.

I use two velocity inlet and a pressure outlet boundary condition. Turbulence conditions are given with Turbulence intensity (%) and Hydraulic diameter. Turbulence intensity is set to 4% for gas and to 1% for solid, but these data have been extimated through empirical correlations, so I don't know if their values are correct.

Thanks again

 J. Kim April 16, 2003 20:28

Re: Turbulent viscosity in a riser

Although you used lower-order(i.e. 1st order) schemes, if you still have the same warnings, I recommend you to remesh your domain.

Lower-order schemes may give you looks-good results. But don't trust too much.

Kim.

 ap April 17, 2003 11:49

Re: Turbulent viscosity in a riser

Hi

 Alex Munoz April 17, 2003 14:09

Re: Turbulent viscosity in a riser

Hi

I apologize to you becuase i did not read carefully the phase that you try to model!

Of course you will get high turbulence ratio if you try to mode a SOLID.

I will explain to you, FLUENT is telling you that in two adjanced cells the differences in velocity is too high. In other words, you fluid has caracteristics of concrete!! of coarse, if you are modeling water as a fluid, it should not be posible and I should refine the mesh.

However, in your case you try to model a SOLID with equation and limits for LIQUID phase!!!

Think twice if you can use the turbulence model design FLUID to model a SOLIDS, Do not forget that the turbulence models were built under the assumption of newtonian rheology and a solid hardly meet this condition!!

Another point RSM models is a very dificult model to with!!

try first k-e model or one of its relatives

Best regards

Alex

 ap April 19, 2003 08:00

Re: Turbulent viscosity in a riser

I'll explain better. I'm modeling a GAS-SOLID flow: solid is made of GRANULAR particle of 76*10^-5 m of diameter.

I'm not using RSM model, because FLUENT doesn't allow to use it with the Eulerian multiphase model, I'm trying to use the RNG-k-e model.

The molecular viscosity of solid phase is evaluated using kinetic theory for granular flows, but it's value is too low. This may be due to a mistake in the calculation of the granular temperature, because the equation implemented by FLUENT is not complete.

I posted another question regarding the possibility to add terms to the granular temperature equation.

Do you think it's possible? I can't find a macro in UDF manual to access the granular temperature value in a cell.