[kinneas]

Dear forum,

We are currently testing a simulation that involves a multiphase flow (Water and air), basically is a rectangular channel that ultimately leads to a dissipation well:






The current setup, which uses InterDyMFoam as solver is limited to a CFL number of 1, is able to somewhat represent the expected behavior of the flow with a peak velocity of ~50 m/s in the main channel, unfortunately in the waterfall zone that has an elevation delta of ~200m (where the fluid separates from the channel and falls to the dissipation well), the dynamic mesh chases small particles of water that are quite spread from the main flow, increasing cells count and hence computation time significantly because the time step to model this behavior, unnecessary for our goals, and maintain the desired CFL number is sitting at 7e-4 s (which is a no-go).
We have already tried limiting the maximum cell count for the dynamic mesh with no success.

I was wondering, is InterDyMFoam a good choice for this type of simulation or is there another way (solver, procedure…) to tackle this model? We would greatly appreciate any suggestion.

Kind regards,

José D.

[BlnPhoenix]

Originally Posted by kinneas

Dear forum,

We are currently testing a simulation that involves a multiphase flow (Water and air), basically is a rectangular channel that ultimately leads to a dissipation well:






The current setup, which uses InterDyMFoam as solver is limited to a CFL number of 1, is able to somewhat represent the expected behavior of the flow with a peak velocity of ~50 m/s in the main channel, unfortunately in the waterfall zone that has an elevation delta of ~200m (where the fluid separates from the channel and falls to the dissipation well), the dynamic mesh chases small particles of water that are quite spread from the main flow, increasing cells count and hence computation time significantly because the time step to model this behavior, unnecessary for our goals, and maintain the desired CFL number is sitting at 7e-4 s (which is a no-go).
We have already tried limiting the maximum cell count for the dynamic mesh with no success.

I was wondering, is InterDyMFoam a good choice for this type of simulation or is there another way (solver, procedure…) to tackle this model? We would greatly appreciate any suggestion.

Kind regards,

José D.

In the DynamicMeshDict you can also limit the number of refinement steps.
Alternatively you can mesh your model uniformly but in my mind it usually is better to use Dynamic mesh sizing like in InterDyMFoam.

[Tobi]

CFL number is sitting at 7e-4 s

is in my opinion acceptable and it depend on how much you are going to refine the mesh. I would use the suggestion of BinPhoenix and limit the maximum amount of cells or limit your refinement level e.g. to 2.
Hope that Medellin is nice to be. I am going to that place soon

[Santiago]

Originally Posted by Tobi

is in my opinion acceptable and it depend on how much you are going to refine the mesh. I would use the suggestion of BinPhoenix and limit the maximum amount of cells or limit your refinement level e.g. to 2.
Hope that Medellin is nice to be. I am going to that place soon

Nice! If youre open to see different things from Europe, I think you might like the city. Some places to visit: Arvi park (is a peripheral natural reserve enclosing the city), "El castillo" museum (a castle, not like Neuschwanstein, but one of the few in S. America), plaza Botero and Antioquia Museum (have you heard of Fernando Botero, the painter? Hes from medellin, youll find there a lot of his sculptures and paintings), Lleras Park (to do apperitiv), and if you like "adventure" you can do the graffiTour. Basically they take you around the city's Ghettos, many of which are painted with Graffitis telling stories about the particular "barrio".

Then, if you have the time, you should go to Rionegro and Guatape. These are two villages close to Medellin. You will not regret.

Foods to eat: pandeyuca and pandebono (like a Brioche, but not quite). Bunuelos (very similar to the german quarkballchen or italian frittelle, but again not quite). If you are up to eating A LOT, you may try the "bandeja paisa" (similar to the Irish Breakfast).

Thats it. Have fun.

[Tobi]

Thanks for replying to the Medellin stuff. Sounds great - I should invite you on a beer. However, to kinneas, there could be another option too. You could use my extended library in order to refine the interface higher while keeping the internal droplet coarse. That just came into my mind. My extended library is hosted on my server https://holzmann-cfd.de

[kinneas]

Originally Posted by Tobi

Thanks for replying to the Medellin stuff. Sounds great - I should invite you on a beer. However, to kinneas, there could be another option too. You could use my extended library in order to refine the interface higher while keeping the internal droplet coarse. That just came into my mind. My extended library is hosted on my server https://holzmann-cfd.de

Dear Tobi and Santiago,

Thank you for responses. Currently we switched to a standard mesh with a different refinement taking into account our previous results, things look a little bit more promising than using a dynamic mesh, fortunately we still have some resources left to tune and play a little bit more with this case.

We'll take a deep look at the resources hosted in your web page!

Kind regards,

José Díaz.