Help on model

[Sidemountyucatan]

Hi,

could anybody take a look at the attched model and give me advice why it doesn't work. It's the first version of a cleaning tray.
On top there is a outlet aswell on the bottom. The inlet is via the 4 nozzles (maybe later on there can be more).

Thx a lot

[thschrader]

For me your case works.
I changed mesh refinement a little bit, laminar case, single cpu run.
My file:

tank_ts.FCStd

(487.26 KiB) Downloaded 44 times

OF-version is bluecfd 2017-2 with
OS: Windows 10 (10.0)
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.19.23546 (Git)
Build type: Release
Branch: master
Hash: 6b017f9a16b15b0e628c8d874c4058442dee5548
Python version: 3.6.8
Qt version: 5.12.1
Coin version: 4.0.0a
OCC version: 7.3.0
Locale: German/Germany (de_DE)

nozzle.JPG (60.79 KiB) Viewed 1850 times

[thschrader]

Aaah..I forgot:
I dont know if the volume-flow is spread across the 4 inlets or
if it is adressed to each face...

inlet.JPG (45.26 KiB) Viewed 1845 times

[Sidemountyucatan]

- Thx, looks fine. It is meant to be volume flow each face. Must it be corrected ?
- You processed it as a laminar case but isn't the flow through the nozzles turbulent ? Does the laminar model describe the flow near the nozzles correctly ?
- So, I thought maybe it is necessary to define each nozzle as a separate inlet in the model. But your answer tells me it's not, right ?
Could it be done ?

[thschrader]

- Thx, looks fine. It is meant to be volume flow each face. Must it be corrected ?
- You processed it as a laminar case but isn't the flow through the nozzles turbulent ? Does the laminar model describe the flow near the nozzles correctly ?
- So, I thought maybe it is necessary to define each nozzle as a separate inlet in the model. But your answer tells me it's not, right ?
Could it be done ?

Laminar case: start with the most simple arrangement, even when the flow is turbulent.
Volume-flow: after case writing for me it is not clear if the volume flow of 2600 mm^3/s is adressed
to each inlet (==> 2600 mm^3/s at each face) or is it 2600/4 for each face.
So in my opinion it would be better to use 4 separat inlets with a clear definition.
Hope that helps. Weeekend....

[Sidemountyucatan]

One last question to the "Laminar"-calculation mode:
You wrote "Laminar case: start with the most simple arrangement, even when the flow is turbulent." What does it mean specifically ?
I mean, which calculation scheme hides behind this mode ? Are RANS-equation being solved ? How is turbulent viscosity handled ?


I try to understand what the software is doing.


Thx for answer.

[thschrader]

One last question to the "Laminar"-calculation mode:
You wrote "Laminar case: start with the most simple arrangement, even when the flow is turbulent." What does it mean specifically ?
I mean, which calculation scheme hides behind this mode ? Are RANS-equation being solved ? How is turbulent viscosity handled ?
I try to understand what the software is doing.
Thx for answer.

Starting with the most simple arrangement means that you start with the most simple arrangement .
In your case I would do this:
Setup a laminar case ==> test if meshing works, no boundary elements necessary. Mesh ok? (use "checkMesh"). Does the solver run?
If so, refine your modell ==> setup turbulent case, define boundary-layers. Mesh still ok? Does the solver run?
When the basic setup runs, you can do your parameter study by testing different meshes and/or boundary conditions.

You can compare this with the FEM-wb. Before running a complicated case from scratch, for instance using nonlinearities, it
is better to start with a simple linear problem. If this doesnt work, the rest wont either.

The RANS-equations are only solved when using a turbulence model. For understanding the turbulence model and the
solvers you should have a look at this:
https://www.youtube.com/channel/UCcqQi9 ... oUu8eYaEkg

[Sidemountyucatan]

Hi,

I appreciate the discussion a lot. But maybe we missunderstood each other. What I was asking for is the set of formulas being solved in the "laminar case".

I suppose Navier Stokes equations are solved aswell, right ? Is it kind of DNS ?

But now another point: which further approach would you suggest. Currently the "laminar case" is running and it looks fine. Afterwards I will try a run with k-w-SST. But I guess the error message will show up again. What approach would you use ?

Thx

[thschrader]

I appreciate the discussion a lot. But maybe we missunderstood each other. What I was asking for is the set of formulas being solved in the "laminar case".
I suppose Navier Stokes equations are solved aswell, right ? Is it kind of DNS ?
But now another point: which further approach would you suggest. Currently the "laminar case" is running and it looks fine. Afterwards I will try a run with k-w-SST. But I guess the error message will show up again. What approach would you use ?

You can see the used solvers/gradient schemes in the fvSchemes/fvSolution files in the system folder after case writing.
In my opinion a problem with your model is the very small inlet compared to the whole fluid domain. I dont know what you
want to calculate, but I would start to check which mesh-refinement is necessary to simulate the correct inflow at the inlet.
If the inflow is wrong, you cant expect a correct simulation in the rest of the domain. Imagine you want to simulate a pipe-flow
with your inlet diameter. You need a much finer mesh. I would do 2 steps.
1. Create a separate model with 1 inlet to clarify which mesh/boundary condition/turbulence setting must be used to get a correct inflow.
2. Transfer the result to your complete model.

Openfoam/the Navier-Stokes-equation can solve the turbulence without using a turbulence model ==> DNS-simulation.
But you need a superfine mesh to resolve the micro-turbulence at sub-millimeter-scales. For a practical use (time!)
with a normal PC that wont work. But...give it a try

inlet.JPG (26.86 KiB) Viewed 1524 times

inlet_mesh.JPG (49.88 KiB) Viewed 1524 times