High Reynolds number simulation

I have simulated incompressible flow in porous media using low body force (low Reynolds number) and the results were satisfactory. However, when I increased the body force beyond a particular value (at the same viscosity as the low body force), I get nan as simulation results for velocity.

I would be appreciate if there is an idea on how to simulate flow at high Reynolds number/body force as that is my main interest in the flow simulation


what is the critical value for the body force density? Please be aware that increasing the Reynolds number usually also requires increasing the resolution. It is possible that your Mach number becomes too large. Please post your simulation parameters.



I understand that increasing the Reynolds number requires increasing the resolution. However, my resolution is fixed as I work with CT images whose resolution I cannot alter. Thus, I am left with working with the default resolution that comes with the image data file.

My parameters are
lattice viscosity = 0.181884
dx = 5.80E-06 (voxel dimension)
N = 250;
The maximum body force I could simulate with was 8.60E-03 (i.e 1.33E+06N/m^2) and I had a velocity of 2.60E-02m/s. This was equivalent to a Reynold’s number of about 2.4.

However, I want to simulate beyond this Reynolds number to see the onset of inertia effects but when I try with higher body force, I get nan for velocity.

I used OpenLB for my simulations

I see. Try to reduce the lattice viscosity. Is it possible in your case? A body force of order 0.01 is very high. What is your velocity in lattice units?

The velocity in lattice unit at that maximum body force is 2.74337E-02. I think I can decrease lattice viscosity. However, I have read that for good simulation results in flow through porous media, the relaxation time should be around 1.0. Which again fixes a limit on the range of values my lattice viscosity can take.

The lattice velocity seems to be okay. You are right that the relaxation time should be kept constant in porous media simulations if you use BGK since the bounceback boundary condition depends on tau. This can cause large deviations from the expected values if the pores are very small. What is the typical width of the pores? What kind of boundary conditions do you have at inlet/outlet? Periodicity, I assume.

Well, I do not have a defined pore size. It is a collection of spherical particle in my simulation domain. However, the diameter of the particles is 123micrometre. I have implemented periodic boundary conditions on all sides of the simulation domain, including the inlet and outlet faces.

I am still hoping I obtain answers to how to simulate large body force (high Reynolds number) flows using OpenLB.


Did you get it solved?

If not: Couldn’t you just use a higher resolution and then downsample the results?

Thanks, I have been able to find a solution, atleast to move ahead. I have confirmed that resolution is my main issue, so I decided to work with a computer generated domain that I can vary the resolution at will and I was able to simulate high Reynolds number flow. Are you into porous media simulation?

I currently don’t do porous media simulations; only blood.

I just saw that no other answer was there and grew curious what happened. So thanks for the update! It’s good to know you got it to work.