[salumi]

Hey everyone,

I am trying to model the flow inside a certain gravity separator.
The first case shall be conducted isothermally (later heat-exchange should be added to the model to see the difference, but first i need to get isothermal correctly).

I have created a fully structured mesh. The quality measurements displayed by FLUENT seem to be good (max. skewness around 0.6 and min. ortho quality around 0.4), cell count is around 1.2 million for a geometry of 40 mm diameter, 300mm length of main body and additional inlet and outlet pipes.
There are some high aspect ratio cells of ~450 but they are at the beginning of the inlet wher exact resolution of flow pattern is not of interest. In the main domain there shouldnt be any cells with aspect ratio above 100. There are necessarily some high aspect ratio cells because of the structured mesh (blocking created in ICEM).
When checking the mesh, FLUENT gives me the warning that there are some high aspect ratio cells near walls and suggests the use of an alternative algorithm, which i enable.
Except for that, fluent doesnt display any bad quality cells etc.

Firstly i want to only build up the flow field using only the fluid (supercritical water), in this case only specified with constant density and viscosity (isothermal).
However, i cannot get the solution to converge....
The geometry has one inlet and two outlets.
The operating pressure is 28 MPa, which i set in the BC setup.
The Reynolds number is expected to be high in the inlet pipe and the orifice and low flow velocity and Re are expected in the body of the separator. Therefore i tried using the Transitional SST and SST k omega models.

The inflow is given at 0.0003 kg/s, therefore i specify it as a massflow inlet.

I also know that the outflow at one outlet should be approx. 0.0000533 kg/s (around 18 percent of inflow), the rest should exit through the other outlet.
I tried two options of specifiying this:
1) i specify both outlets as outflows and specify their rating as 0.18 and 0.82.
When i try to calculate the solution i usually end up with a continuity resiudal around 5e-3 and local mass imbalances up to around 5e-10 in some cells.
Looking at the fluxes i see that the outflows seem to devide up correctly.
However there is also a flux for the int_fluid part which is my interior?
I don't think i can trust this solution since the mass imbalance seems to be too high? What else could i look at to decide if this solution is okay?
I tried Transitional SST and K-omega SST models, SIMPLE scheme , first order , second order, etc., coupled scheme + pseudo transient..... Nothing seems to change

What is also really strange to me: When i lower the UR factors after the resiuals dont chagne much anymore, they suddenly drop to a new level (within a couple iterations) and then stay there. When i re-increase them, the residuals also suddenly increase again (shown in attached picture)

2) I set both outlets to outlet-vent or pressure-outlet and set a target mass flow rate for both of them. The rest of the settings of the outlets are left as standard.
The continuity residual also doesnt converge far enough and looking at the fluxes and velocitys in the domain show a completely wrong solutions, since the outflows have completely different values than specified and the flow field is physically completely unrealistic.
I tried different UR factors and solution methods (SIMPLE and Coupled + pseudo transient).
I have no idea why this is the case or what might be wrong about this specification?


Does anyone have an idea about what is the problem?
I will attach an exemplaric case and data of 3000 iterations with standard UR factors where the Outflow condition was used.



Now if the solution could be finally be properly converged i would like to inject a disperse phase and examine the particle trajectories. Therefore i tried this starting from my (semi-trustworthy) solution i obtained using the outflows as outlets.
I inject some inert particles at a certain mass flow rate. I am using a steady solution (is this correct?), i want to know where the particles end up, in the effluent or brine outlet, depending on their size, and if the separation would be efficient.
Whenever i try to do this, Fluent displays this warning after usually 9 iterations in the console: "Warning: DPM_Locate_Point: This might not behave the same way in parallel Fluent".
It then keeps saying "Calculating..." but nothing happens anymore and the program doesn't react. I have to close the program over the task manager and that's it.


Here are the exemplaric case and data of only the flow field (called 3000standardvalues) using the outflows and the case of the injection:
https://drive.google.com/drive/folde...mc?usp=sharing

Any ideas?

I don't know how i could my solution for the flow field, and the DPM doesn't work at all.. I am desperate since i feel i've tried everything that i thought could help.

Let me know if any more info is needed

THANKS!

[salumi]

I managed to get the DPM running, the reason it always crashed was that i gave my injections too many different diameters (100) in the rosin rammler distribution which apparently was too much for the PC.

The uncoupled DPM ends up with a continuity residual of around 6e-3...

The coupled DPM doesn't seem to go towards convergence at all....

Any ideas? Is it the mesh? Or the case setup? IF it is the mesh, what could i further improve?


I updated the case files in Google Drive:
https://drive.google.com/drive/folde...mc?usp=sharing

Please let me know if you have an idea!!

Thanks