search

OPM / LBPM

Pore scale modelling
https://lbpm-sim.org/
GNU General Public License v3.0
65 stars 31 forks source link

Viscosity ratio in lbpm_color_simulator #38

Open maldiH opened 3 years ago

maldiH commented 3 years ago

Hi LBPM Community,

I have a question regarding the limitations with viscosity ratio. How high can we set M(mu_nw/mu_w), but still get reasonable results (relative permeability at steady-state)? I see that in the tutorial, recommendations are M around 5, but wondering, if it is just the maximum value at which the code was tested. Any comments on this is greatly appreciated!

JamesEMcClure commented 3 years ago

Note that mu_w and mu_n are the dynamic viscosity, which includes the contribution of both the density and the kinematic viscosity. The relaxation times control the kinematic viscosity, and you will max out around M=5 if you only change the kinematic viscosity for the color LBM scheme implemented in LBPM. However, you can get a larger viscosity ratio by also controlling the density ratio, e.g.

    rho_n = 0.05
    rho_w = 1.0
    tau_n = 0.7
    rho_w = 1.5

This will give you kinematic viscosities nu_n = (tau_n-0.5)/3 and nu_w = (tau_w-0.5)/3, with M = (rho_w*nu_w)/(rho_n*nu_n). In many cases (e.g. liquid-gas systems) you do also have a large density contrast in addition to the viscosity contrast.

Using this strategy you should be able to achieve M=100 or possibly up to M=350. You can run into instabilities as M increases, and these effects are geometry dependent. Lowering the IFT will tend to help in these cases, since it instabilities are often due to the IFT moving mass too quickly somewhere in the simulation domain.

ahmedsrizk95 commented 3 years ago

Thanks James for clarification, if I have density of rho_1=1000 kg/m3 and viscosity mu_1=1 mPa.s, so what will be the tau_1 and rho_1 in lattice units?

JamesEMcClure commented 3 years ago

You don't need to match the physical units. You only need to match the dimensionless ratios (e.g. viscosity and density ratio). This is due to the Buckingham Pi theorem.

Generally you want to choose the LBM units so that the timestep is as big as possible.

ahmedsrizk95 commented 3 years ago

Thank you very much. I have found this paper also to covert from physical units to lattice units: https://doi.org/10.1016/j.compfluid.2021.104925