CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Marangoni Stress At the Wall (

Peter Nagy January 23, 2004 17:16

Marangoni Stress At the Wall
Dear Community

I have had a very difficult time in modeling the competition between Marangoni and buoyancy effect(inside a "pinned" droplet of R=1.5mm) using VOF. Because of the "parasitic error" (i think) I won't even get the right flow.

Right now, I am compromising with the non-deformable interface with some prescribed convective heat transfer coefficient to model the above flow (no boussinesque yet).

I switched off the multi-phase model and defined a solution domain that is only consist of the liquid phase. The boundary condition is modified for the Marangoni stress. How does this model the boundary incorporate Marangoni stress conditions???

I thought that the one phase model does not incorporate the effect of surface tension. (no panel for the surface tension...)

is it like this del(sigma)=d(sigma)/dT del(T)

where the gradient of surface tension is added to the N-S as some source term as I have seen in some literatures (like VOF, etc)?

Thanks for your help and precious time!

mateus January 26, 2004 04:47

Re: Marangoni Stress At the Wall

This post probabbilly won't be of any help to you but what Fluent version are you using? I've allways had problems with VOF and Fluent 6.0 or older. In 6.1 it works ok.



ozgur January 26, 2004 10:37

Re: Marangoni Stress At the Wall

What I've experienced with VOF is, the selection of time step is quite important. If you use a bit larger time step than you have to, the flow field may become distorted, even if the solution seems to be converged!

Does your drop move in the domain (i.e. falling)? By the way, what does "pinned drop" mean?

Peter Nagy January 26, 2004 11:50

Re: Marangoni Stress At the Wall
To mateus: I am using ver. 6.1 for computation. I really don't know about the known error associated with the VOF in any version of the FLUENT, however this error introduced at the interface is inherent to CSF model that FLUENT uses. If there's a bulk flow of associated with the free surface flow, the error introduced by this "parasitic error" is somewhat negligible, and some people live with it. However when the flow of interest is expected to be same order of magnitude (of Marangoni flow inside a small drop) then the error may be of great deal...

To ozgur: The time step seems to be important as well. I have checked my time scales for the flow. Capillary time scale is O(1e-2 s) for my system. The viscous time scale is O(1 s) and thermal diffusion time scale is O(10s). The time step was chosen to be 1e-7 s for my VOF computation but the end result was not-so-good convergence for the continuity. After each time step, the residual stops falling after certain iterations(~1e-2) and flattens out. The end results of my solution is that there's this fictitious velocity created at the interface level which is O(1mm/s), a comparable velocity magnitude that I am to expect out of the Marangoni flow of my interest.

A pinned drop is a deposited droplet on a solid surface. Pinning is achieved experimentally by having a sharp edge at the edge of the solid so that the liquid-solid-air interface maintains its location. Therefore, the drop is not moving with respect to the surrounding gas. The flow is induced by the thermocapillarity, the surface tension force variation wrt temperature.

ozgur January 26, 2004 16:06

Re: Marangoni Stress At the Wall

Are u using double precision solver? Maybe it helps. If you are using PISO algortihm with under-relax. factors of all 1, try 0.7 for momentum, 0.5 for pressure and volume fraction. Moreover, sometimes the residuals may not fall down 3 orders of magnitude, even the solution converges. For instance, when your initial conditions are not so far away from your final solution. I think you better to monitor convergence history by volume & surface integrals as well.

All times are GMT -4. The time now is 12:57.