Geometry/ initial conditions/ convergence

[gesx123]

Hello, Dr. Karnakov. I have a few questions I hope you can help with:

1. I’m creating a geometry made up of rectangular blocks, but there is a significant difference between the maximum and minimum dimensions. Is there a way to locally refine the mesh to reduce computational load?

2. In the initial conditions, the default volume fraction in the computational domain is set to 0. Does this represent the continuous phase? I’ve initialized the dispersed phase near the inlet and also set the inlet to the dispersed phase, but there seems to be an interface between these. Why is that?

3. Can increasing Max_iter improve convergence? Is it the number of iterations within a single time step, or does it cover multiple steps (e.g., 10)?Also, is the convergence condition related to the residual?
   

4. Is the main function responsible for solving located in the Run function in hydro.ipp?

Thank you in advance for your help!

[pkarnakov]

I’m creating a geometry made up of rectangular blocks, but there is a significant difference between the maximum and minimum dimensions. Is there a way to locally refine the mesh to reduce computational load?

unfortunately not, the solver only supports a uniform Cartesian grid

In the initial conditions, the default volume fraction in the computational domain is set to 0. Does this represent the continuous phase? I’ve initialized the dispersed phase near the inlet and also set the inlet to the dispersed phase, but there seems to be an interface between these. Why is that?

the volume fraction represents the dispersed phase; if the case you are running uses MultiVOF then this is expected, those distinct volumes correspond to different bubbles with an interface between them

Can increasing Max_iter improve convergence? Is it the number of iterations within a single time step, or does it cover multiple steps (e.g., 10)?Also, is the convergence condition related to the residual?

max_iter is within one time step, increasing it beyond 10 is unlikely to improve convergence, I'd start with reducing the density ratio

Is the main function responsible for solving located in the Run function in hydro.ipp?

yes, that function puts all the other components together

[gesx123]

Thank you, I understand now! However, I still have two questions that have been bothering me:

1. How are the units converted from experiments to simulations? I understand that length is scaled based on a reference dimension, but how do the units of time change between simulations and real-world experiments? Why does the density change from 1000 to 1, and the velocity from 2.08 to 1?

(https://github.com/user-attachments/assets/3bd5d5ad-b906-4aef-84de-0aa55245e8ef)

2. If I want to simulate droplet flow, how should I modify the definition of the Reynolds number?![image]

[gesx123]

@pkarnakov Hello, Dr. Karnakov.

1. How are the units converted from experiments to simulations? I understand that length is scaled based on a reference dimension, but how do the units of time change between simulations and real-world experiments? Why does the density change from 1000 to 1, and the velocity from 2.08 to 1?
   The velocity is normalized using the inlet velocity as the reference, and the density is normalized using the density of the continuous phase as the reference. Is this understanding correct? However, when I set Re=0.5 and Ca=0.1, the droplet seems to shrink.
   
   But when I set Re=0.8 and Ca=0.07, it appears normal again.
   

[pkarnakov]

How are the units converted from experiments to simulations? I understand that length is scaled based on a reference dimension, but how do the units of time change between simulations and real-world experiments? Why does the density change from 1000 to 1, and the velocity from 2.08 to 1?
The velocity is normalized using the inlet velocity as the reference, and the density is normalized using the density of the continuous phase as the reference. Is this understanding correct? However, when I set Re=0.5 and Ca=0.1, the droplet seems to shrink.

Yes, the velocity is normalized so that the mean inlet velocity is 1, and the density of the continuous phase is 1.
The droplets may shrink at Re=0.5 due to numerical errors as the viscosity is too high, perhaps changing some parameters of the linear solver could help it.

If I want to simulate droplet flow, how should I modify the definition of the Reynolds number?![image]

The Reynolds number is defined here based on the continuous phase, so if you do droplets (with a density ratio near 1) then there is not need to change the definition.