CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Free Surface Modeling (

willy July 12, 2001 23:04

Free Surface Modeling
Hello, I am modeling a square watertank with free surface. It has an influnet on the front and effluent at the back Goes like this.


I (free surface) I


I I ->>(in)= water (out)->>

------------------------- One suggested that I should see the suface as a wall and ignore the roughness factor 0. So far I modeled several tanks and it seeems to me.. the assumption makes sense. Is it correct to see the free surface as a wall? Otherwise, is there other to model free surfaces? Thanks in advance.

P.S : I got terrible result when I defined the free surface as "Pressure inlet."

willy July 12, 2001 23:07

Re: Free Surface Modeling
sorry. The graphical description messed up.

John C. Chien July 12, 2001 23:46

Re: Free Surface Modeling
(1). If your interest is the free surface behavior, then the answer is obvious. (2). Otherwise, simplified model is good enough for you, if the solution serves the purpose.

rajeev July 13, 2001 00:04

Re: Free Surface Modeling
Hi !

I too have a similar problem. I am modeling flow of molten steel in tundish. The molten metal is let into a trapezoidal (boat like tank made of refractory and steel shell) tundish through a ladle placed on top. The metal from ladle flows to tundish through a submerged pipe called shroud. Metal level in tundish is maintained around 900mm with 150 mm of shroud under the metal. Outlet is a hole in the bottom of the tank. If I am not looking for surface waves can I not use symmetry as the boundary condition for the top surface of metal rather than using wall as boundary condition. Because on surface I will have velocity in horizontal direction.

Will someone advise me on what should be the correct boundary condition.

Thanks in advance Rajeev

Livo July 13, 2001 02:01

Re: Free Surface Modeling
You can set the boundary type as wall and then specify the shear stress as zero for all the components in the boundary condition panel. I hope I understand you correctly?

wilyban July 13, 2001 02:02

Re: Free Surface Modeling
Thanks very much, So it seems that if I am interested in the behavior of the flow in the tank, I can simply regard free surface as a wall (cause no vertical flow thus no pressure generated..) If I got the message wrong, please corect me. Thanks million.

rajeev July 13, 2001 06:06

Re: Free Surface Modeling

The process taking place in tundish is similar to the what I am describing below. Consider a bath tub which is being filled with hot water through a vertical pipe. Initially water is allowed to collect in the tub by closing the drain. Once a satisfactory level ia achieved the drain is opened and the flow of hot water from tap is adjusted to maintain a constant level in the tub. What type of problem this will be. Can I consider it a steady state.

If we use wall as boundary condition for the free surface then what should be wall roughness parameter.

Regards Rajeev

Livo July 13, 2001 06:59

Re: Free Surface Modeling

I do not know enough about your problem, but it depends on whether you are interested in modeling the problem from the beginning until the required level has been reached (unsteady) or whether you are only interested in a model when a certain level has already been reached (steady). The wall roughness will only be applicable if you have a fluid flowing along a wall or then fluid/solid interaction. If you have a free surface then you can specify the boundary type as a wall with zero shear stress. Hope it helps?

willy July 14, 2001 01:19

Thanks million guys.One other thing.
Thanks million guys. I am modeling steady flow so I can set the free surface as a wall. I still have problem with roughness factor though. It seems to have little difference whether I set the factor 0 or 0.5(For aluminium). and I found out that by defalut, Fluent set the XY shear force term 0 each.

And one other thing, how can I vertfy the model is correct? I believe I have to do some experiment to support the model, (like measuring velicity in certain point..- hard one isn't it??). Or one suggest that I can use dye test for velicicy vector verification. What about U guys? What kind of experiment do U use to back up the model generated by Fluent. Thanks in advance.

John C. Chien July 14, 2001 03:14

Re: Thanks million guys.One other thing.
(1). Talk to the guys in the laboratory.

rajeev July 15, 2001 05:06

Re: Thanks million guys.One other thing.
We have a particle image velocimeter. Using a physical model and water we simulate the same in lab and using the PIV take photographs. The software then calculates the velocity vector in 2d planes. This is used to validate the results from the Fluent model. Although we haven't done this yet. But I came across some literature where people have done the same and compared the two results.


Tomas Börjesson July 17, 2001 07:07

Re: Free Surface Modeling
According to Rodi, Turbulence Models and their Application in Hydroddynamics, A state of the art review 1993, the free surface can be considered to be a symmetry plane as a first approximation when wind induced shear stresses are absent.

Note that symmetry plane must not be confused with a frictionless wall (very different turbulence characteristics). Of course when the Froude number becomes significant one has to use a free surface model and the symmetry plane is not valid.

The free surface models of Fluent v5.3 and v5.4 (geo-recunstruct and the implicit) gave me crap results compared with experimental data on a lab scale mixer tank, however the v5.5 gave reasonable good agreement (compared with velocity profiles in the tank).

All times are GMT -4. The time now is 12:43.