Here are some questions about multi-component simulations in a square box without top cover. Like this:
| hhhhhhhhhhhhhhh|
hhhhhhh light fluid
hhhhhh heavy fluid
hhhhhhhhhhhhhhhh
Upper is light fluid, and the bottom is heavier one with higher visicosity. A upward force is set at the bottom of the heavy fluid, then some phenomenon is supposed to see.
Now here are the questions:
the viscosity ratio of light and heavy fluid is 1:100. According to the equation (1/omega = 3*viscosity+0.5 ), I set the omega of upper one is 1.98, while that of the heavier one is 1. But the simulation broke down only after several steps. How to solve this problem, or palabos can only solve low viscosity ratio problems?
As to the uncover boundary condition, should I set neumann pressure boundary ? Or are there any other methods? I don’t thinks it’s called free surface. Do you know the key words of such boundary situation?
I had similar problems with two fluids with high omega difference.
You could try to reduce the upper ones omega to find the highest difference palabos can simulate.
Another test you could do is increasing the resolution of the simulation (N). With 6GiB Memory my simulation ran at a resolution of 600x600x600 (though pushing the computer to its limit).
I’ll try to give an answer to your questions. I guess you are using th MultiComponent ShanChen model.right?..
the viscosity ratio of light and heavy fluid is 1:100. According to the equation (1/omega = 3*viscosity+0.5 ), I set the omega of upper one is 1.98, while that of the heavier one is 1. But the simulation broke down only after several steps. How to solve this problem, or palabos can only solve low viscosity ratio problems?
well … the model you are using only resolve for low viscosity ratios. In the literature you will have trouble in finding
calculation of immiscible fluid lower than 1:5. Palabos won’t do better then.
As to the uncover boundary condition, should I set neumann pressure boundary ? Or are there any other methods? I don’t thinks it’s called free surface. Do you know the key words of such boundary situation?
BC with the ShanChen method for multicomponent fluid (would you like to have a sort of free surface at the top boundary right??)… well… I don’t really now how to help you here. As far as I know there is not BC in the SC model for the problem you want to resolve.
Thanks heaps. Andrea. Your advice really helped a lot.
Still have a question. How to lower the viscosities of fluids.
I tried to modify density, omega, G, the external force, but none of them worked. All the fluids seem to be too creamy and thick, not like water but a kind of honey…
I’m using shan-chen model? or should I change the model to another one? could someone give me some hints?
Thanks a lot!
Here I talk about lattice units only. You can decrease the kinematic viscosities of the two phases by playing with omega (1/relaxation time). So, smaller the relaxation time lower the viscosities.
However, from my experience (it is plenty of literature however that is much more clear then this post), the Shan-Chen method for multi-component fluid with BGK collision term can be prone to numerical instabilities when either the relaxation times of the two fluids are small (not such a big news with LB) or if you decide to play with a viscosity ratio different from 1. More important the viscosity ratio you choose, higher the chance of having numerical instabilities.
In the literature I have seen that some people report working calculation for viscosity ratios of 10. Lucky them !!!
About your second question …
I’m using shan-chen model? or should I change the model to another one? could someone give me some hints?
Well, I don’t know… Personally, I think you should give a try to the model. For sure the Shan-Chen model is one of the “classics” of the LB. It is somehow easy to implement and understand. For sure you can only learn cool stuff by using it. It works “well” if you have to deal with complex geometries (porous media flows) at low Reynolds numbers.
