Hello,

We are planning to model a packed column, with countercurrent gas liquid flow. Experiments have shown that due to radial distribution of porosity in the bed, liquid tends to accumulate towards the walls, due to radial distribution of the flow resistance (as a result of porosity distribution). This phenomenon is seen even with a single phase liquid downflow. My concern is: As far as I have heard (I have still not tried) , FLUENT treats a packed bed as a momentum sink. Is it possible to predict this behavior of the flow by defining the radial distribution of the porosity of this "momentum sink"?

Hi Rakib,

Indeed, FLUENT allows you to define the Viscous and Inertial Resistance coefficients in both x, y and z directions. You can use the Ergun equation, for example to determine these coefficients.

HOWEVER... what is a big problem in FLUENT, is that the porosity term that can be defined, is only used to calculate an effective heat transfer coefficient k eff, a linear mix of the fluid k and the solid k. For the fluid mechanics side, the fluid zone is still considered 100 % open, id est no obstacles to the fluid which, in nature, will accelerate. This acceleration, which is not simulated, will certainly have an influence to the flow distribution you will simulate, and therefore, I suggest you to take this into account.

I hope my input will benfit your project, which is very interesting, even if it seems simple, its interesting.

Good luck

Greetings

Damien

Thanks for the reply. But is it possible to define any user defined function to define the radial variation of the porosity of the bed, so that the flow resistance will be evaluated accordingly?

I don't think it is possible to define a radii dependant porosity via UDF.

There is no list to choose between "constant" or other value on the right side of the porosity definition window in the Define->BC->Fluid panel. Unlike for Source Terms.

But maybe, could you look further about the conical porous medium in the manual (I have never used it).

Is it necessary for you to use the porosity?

If you want to have a radii dependant heat transfer coefficient, you should be able to handle the thermal conductivity values via UDF directly. Use the equation 6.19.9 in the FLUENT manual to calculate the k_eff. Use your own internal variable to define a radii dependent porosity. I hope its possible.

Good luck

Actually, as I mentioned earlier, even in a single phase liquid downflow through a packed column, liquid tends to accumulate towards the walls. This is because of the porosity variation radially (hence flow resistance variation). This is the reason why I want the porosity to be varying radially in my model. In this case what do you suggest? Thanks for the reply.

If its fluid dynamics parameter like Viscous Resistance and Inertial resistance you want to control via UDF, I think it's not directly accessible. However, you will be able to assign values (radii dependant in your case) of momentum sources (negative for sinks) There is an excelent, simple example in the UDF manual (FLUENT6.0) chapter 4-3 page 4-43

Thanks Damien.

There's a couple of things you CAN do to overcome Fluents limitations.

First as Damien mentions the porous media model is simply a momentum sink term added to the momentum equations. In the standard model constants are used for the porosity and viscous and inertial resiatnces. Thus the spatial variation in these quantities is not readily handled by the standard Fluent model.

To overcome this you can write a source term udf to add to the momentum equations which enables you to model spatial variations in say the porosity. This is very simple to do.

As Damien also points out, the equations used by Fluent aren't rigorously correct under all conditions. If you spatially average the N-S eqns you actually get a (1/porosity) term in the convection parts which is ignored in Fluent. For the most part, though the convection terms, ie d/dx(rho*u*v) etc are small due to the squared velocity terms. So this is probbably not an issue in your problem.

If you want to look at gas-liquid flow rather than a single phase flow, you will probably have to move to the granular type model which enables multiple phases.

But with some mods the porous media model is probably a good place to start.

By the way, are you talking about the variation in porosity due to the presence of the wall - where the porosity right at the wall must head to 1 due to the assumed spherical nature of particles close to the wall?? Or does it vary macroscopically throughout the reactor?

Greg

Yes, I am talking about the variation in porosity due to the presence of the wall - where the porosity right at the wall must head to 1 due to the assumed spherical nature of particles close to the wall. Thanks for ypur advice.