[Ritah]

Hello,

I study liquid Zn with some Al (0.2 wt%) flowing in a 2D channel and I apply an Al mass flux at the lower wall of the channel.The mass transport is computed in the User Defined Scalar frame. I want to use a course mesh (y+>30) , but the results from the course mesh differ a lot from a y+1 mesh (see Fig.1 in Attachments), so the modelling in UDS is not mesh independent. Probably the solution of the implemented UDS mass transport equations doesn't take into account the near wall/ wall function modelling.

In the case of heat Transport, the solution to Energy equation is mesh independent (see Fig.2 in Attachments), where a heat flux is applied at the lower wall.

Does anyone have an idea how I could implement a mesh independent mass Transport solution in UDS Frame?

Thank you,
Andreea

[vinerm]

You are partly right in saying that UDS equation doesn't take into account the wall model or wall function model. The fact is that by default the eddy diffusivity is not considered for the diffusion of UDS. User needs to add it. So, you have to write a UDF that returns the sum of the molecular diffusivity and the ratio of eddy diffusivity and turbulent Schmidt number.

[Ritah]

Thank you for your answer, Vinerm.

The turbulent diffusivity is already included (D_eff= rho*D_mol+mu_t/Sc , in Kg/m*s as Fluent uses). So there should be something else that is not working or should be taken into account.

But again, thank you

[vinerm]

Do you mean you are already using UDF for applying diffusivity for the UDS?

[Ritah]

yes, we use a UDF for diffusivity of Al /Zn.

[vinerm]

Turbulent viscosity inherently takes into account everything it needs for the wall modeling. Furthermore, you have density multiplying the molecular diffusion. Though it won't have much effect in the core if eddy viscosity is significant but it would apply about two orders of magnitude higher diffusion coefficient in the near wall region. Higher diffusion coefficient implies smaller gradient and that's what the simulation seems to predict.

[Ritah]

Hello,

We have multiplied the molecular Diffusion coeff by Density because Fluent works this way: the Units of mass Diffusion coefficient in fluent are Kg/ms (normally it should have been m^2/s).

The problem is that the Al concentration profile across the channel looks different for different meshes (resp. y+ values), which shouldn't be the case (as seen in Fig2 with temperature profiles).

Thanks

[vinerm]

Yes, you are right. Fluent expects kg/m-s because that is the unit for dynamic coefficient; the one used in conservation equations. is kinematic coefficient and valid only for constant density systems. But I suppose the value you have available for D_mol is kinematic value and not dynamic one. In that case, you are doing it correctly.

I looked at the attached images and yes, the gradient is very low for the scalar. But any particular reason for using UDS instead of species transport? That can certainly be used for validation purpose. You used energy equation, however, for the energy equation, Reynolds analogy is used and a similar law is implemented for species. No such law is applied for UDS, however. So, you have to either use species transport or use the law used for species.

[Ritah]

Thank you for answer.

Yes, we use UDS since this will be part of a bigger Simulation and UDS Frame is more flexible than Species Transport (like defining mass fluxes on the walls,etc)
But at least for the 2D channel it makes sense to compare the results from UDS with those from Species Transport.
So, thank you for suggestion.

[vinerm]

UDS will show differences with respect to species as well because species also has modeling similar to energy. So, I suppose you have to include the effect as you initially thought of. The equation used in the first cell is given in the manual.