[simospanu]

Hi to everybody.

I have to simulate a porous sparger used to insufflate Nitrogen in a Olive Oil flow. A simplified geometry of the system is attached.

Boundary conditions on Inlets, Outlet and Walls are the following:
In_Oil: Mass Flow Rate = 1.28 kg/s
In_N2: Mass Flow Rate = 3.9e-04 kg/s
Out: Opening Condition with Relative Pressure = 71000 Pa
Walls: Fluid Dependent; for Nitrogen Free Slip Wall and for Olive Oil No Slip Wall.

The sparger has been modeled using a porous domain. For Nitrogen has been set an Isotropic Loss Model (Loss Velocity Tipe: Superficial) based on Permeability and Loss Coefficient. These two parameters have calculated using the following equations:

Permeability = [(Dp^2)*(epsilon^3)/[150*(1-epsilon)^2]
Loss Coefficient = 2*[1.75*(1-epsilon)/(Dp*(epsilon^3)]

Where epsilon is the Volume Porosity (set to 0.7) and Dp is the diameter of the Nitrogen particle (set to 40 μm, which is the diameter of the holes located on the sparger).
For the Olive Oil has not been set a Loss Model.

The adopted turbulence model is k-e and there is no heat exchange (Isothermal Model, Temperature fixed at 25°C).
In the multiphase settings has not been set the homogeneous model.

The analysis is stationary and a physical timescale of 0.5e-03 is adopted.

The problem is that, after a certain number of iterations, the volume inside the porous media is partially filled by Olive Oil, thus obstructing Nitrogen flow.
If I try to insert a Loss Model also for Oil (setting a very low permeability in order to avoid that Oil goes inside the sparger) the simulation crashes after a few iteration (Overflow).

Any suggestions?

Thanks in advance.

[ghorrocks]

I don't think the Eularian non-homogenous model you have selected is appropriate for this model because it is based around small particles (presumably you had N2 as the disperse phase). The N2 exists as a pure gas in the N2 inlet to the porous membrane, so you cannot model this as N2 bubbles in a continuous fluid.

Am I correct in saying that the N2 flow has essentially no pressure loss in the pipe, but a large pressure loss across the porous membrane into the oil, and when in the oil it forms small bubbles?

If this is correct then why model the N2 inlet and porous material at all? Replace the oil surface of the porous material with an inlet of N2 bubbles of the appropriate size and flow rate. Then you don't have to worry about modelling pure N2 gas or porous materials.