I have two incompressible fluids (water and ethanol) that flow in rectangular canals in a turbulent counter-flow. The fluids are separated by a steel plate (5 mm). I want to study the heat transfer between the two fluids through the plate. My model is made in GAMBIT 2D.The questions are:

1. How I must define the boundary condition for the two interfaces (fluid1-plate and plate-fluid2): interface, interior, internal or wall? 2. For fluid entrance is good if I defined velocity_inlet and for outlet: outflow? 3. For the symetry axe (the model is simetric by OX) should I define symetry or axis?

I think that my problem is not so complicated but I didn't find a good explanation for my problem in Gambit and Fluent Help! So, thank you for your help!

> 1. How I must define the boundary condition for the two interfaces (fluid1-plate and plate-fluid2): interface, interior, internal or wall?

If you need only to define the two fluid zones and the solid zone in gambit (no walls an interfaces). If you read your grid in fluent it will create walls which separate the different zones. Those walls are already set in a way that heat flux can occur across these walls.

> 2. For fluid entrance is good if I defined velocity_inlet and for outlet: outflow?

Yes.

>3. For the symetry axe (the model is simetric by OX) should I define symetry or axis?

If your geometry is actually a small pipe within a larger pipe you should use axis. To use this axis in fluent you will have to switch the slover to axis symmetric.

RoM

Thank you very much, RoM! I made the model following your indication. All is good, but continuity residual is stabilised at 2.5 value!!! The residuals for X and Y velocity is also high (they passed up the 1e-3 threshold). Only the energy residual is good: 6e-8. Is need more than 1e4 iterations or I need to change something else?

You should not judge convergeance only by residuals. Look here for mor information

http://www.cfd-online.com/Forum/fluent.cgi?read=31920

RoM

All is good, but continuity residual is stabilised at 2.5 value!!! Typically in mixing problems you need to go for smaller timesteps to represent good mixing especially in turbulent mixing - in the mass balance 'dm' term which could be in the form dm=const*fn(t/C)*dt, fn could be anything between linear to exponential andso 'dt' has to be defined properly to obtain the appropriate dm, I think this is a modification you may have to try.

The residuals for X and Y velocity is also high (they passed up the 1e-3 threshold). Again in navier stokes the solver would be using the dm and so I would doubt the same problem.

Only the energy residual is good: 6e-8. Is need more than 1e4 iterations or I need to change something else? Probably you are assuming that the mixing is taking place at a constant (room) temperature and to energy conservatiuon may not have much to do? you assume a temperature rise during the process then this would also jump up

Arunvel.

But there is no fluid mixing! The fluids are separated by a steel plate.