Hello
I hope some of you use Comsol.
So, I’m doing simulation of airflow through porous media. According to the characteristics of the flow and the geometry, the flow is turbulent, so after doing some simulations, the physics chosen was k-epsilon Low-Re, because of the wall treatment (Low Re). Geometry is a 2D rectangle representing a long tube with the porous bed in the middle of it.
I included a volume force in order to simulate porous media. At flow conditions, the equation that represent it is Ergun Equation (that could be expressed as Forchheimer eq). My implementation was as simple as :
Fx= -(spf.mu*u)/(ka )- b*spf.rho*u*sqrt((u^2)+(v^2))
Fy=-(spf.mu*v)/(ka)-b*spf.rho*v*sqrt((u^2)+(v^2))
So, when I defined a horizontal cut line in the middle of the bed in order to see velocity profile, I found a problem because a velocity peak is shown at each side, near to the wall (fig 2 and fig 3). I think profile should be just horizontal. Since properties are isotropic, there no reason for this, so, there is something wrong.
I think it is not a boundary layer problem because I simulated it when having a fully developed flow and with 12 boundary layers (as default), shown at fig 1. Minimum mesh quality is 0.45 (in fluent it would be 0.55 as maximum skewness), which I think is not so bad. I tried also using laminar flow conditions and Fluid and Matrix Properties as default which uses Darcy's Law, but the profile I found was similar.
You can see the images as attachments below.
Has anyone had this kind of problem? Am I missing something? How could I overcome this problem??
Any help is really appreciated, since I've been stuck there several days.
Thanks in advance,
Jhoan G.
Hello Jhoan,
I have also noticed the same when I apply brinkman equation for laminar flow. I am not sure why this happens. It seems like at the interface between porous and free domain there is always a spike in fluid flow velocity. I also think your implementation of Ergun equation is correct.
- Badges
- Science topic