I used a define_init function to initialize the force field. Its a simple 2d geometry, the field value is zero at the bottom and raises linear in direction of y, flow direction is x. The init function also sets the face values for the field boundaries equal to the adjacent cell value to ensure there is no gradient in direction of the wall. I saw from you pictures that you are using fluent 6.1 so i cant send you the case and dat files (i use 6.2). I will send you my mesh, the udf source and some text so you can set it all up.

RoM

Hi

It seems from the above long discussion that u have very good experience in species transport. In regard to this, I would like to post my problem which was already done earlier but was not given any reply.

Problem defn:

I tried to simulate a simple vertical plate free convective mass transfer which is similar to that of free convective heat transfer. i gave pressure outlet at all the three sides (top, bottom and one side) and the other side is a plate. The plate is specified with constant mass fraction and the complete domain i.e., ambient is specified with a different mass fraction sothat the concentration difference will induce bouyancy and the flow takes place. In doing so, I found that the velocity vectors are not proper. I get velocity vectors everywhere in thedomain and not just near the wall. Is the problem with the bounday condition or is it sth else?

Could you please help me in correcting it,, coz.. I need to do this base case and then extend it to my problem.

Thanks in advance

Hi,

"the concentration difference will induce bouyancy and the flow takes place" is not entirely correct. Concentration difference will induce diffusion from the wall into the domain (that is if wall concentration>domain contration). The diffusion will change the mixture composition wich will result in density change and then bouancy will occur. You wont see any effects if density in material panel is set to constant for the mixture or all species in your mixture have almost the same density (like co and n2).

RoM

Yes, I agree to what u have written that the density change in the mixture after diffusion, results in buoyancy.

I tried with the fluid 'air' comprising o2, h2o and n2. On the wall, the concentration of h2o is specified more and that of o2 is less when compared to that in the ambient.. i.e., domain. I found the concnetration boundary layer profiles to be growing upwards as expected and are similar to that of temperature boundary layer. But I found the velocity bounary layer is not as expected. I suppose, it should be similar to that in free convection heat transfer i.e., there should be a growing boundary layer near wall and the velocities should be zero at all other regions. Am I correct in this, I feel there wont be any net flow in the domain (away from the wall) under equilibrium and steady state conditions.

I hope I am clear in defining my problem. If required, I will send that case and data file.

Please correct me if I am wrong.

From you problem definition and setup i dont see any obvious problem. It may help to expand the wall in direction of the bottom (add a small inert zone without species gradients). What turbulence model do you use? Have you tried to solve it transient even if the solution will be steady state? Try some other mixture like CO2,H2,N2 for evaluation. Initialize the domain to CO2 and let H2 diffuse in, this should give some strong density changes. Although i would expect some backflow from the pressure outlet at the bottom.

RoM

This is a laminar flow problem, cause the height of the wall is 5cm and the concnetration diff is not significant to make it turbulent. I also tried with extended domain initialised to the same species concentrations as that in the ambient. Shall I send you the case and data file to ur mail id. Kindly go through it in ur free time,