[yahmed]

Hi, I am new in Ansys world.
I am starting my phD research on the cfd simulations of UV reactor used in waste water disinfection.
The reactor consists of rectangular open channels, each channel (2.5 m x 1.2 m x 9 m) has two banks of UV lamp (each bank has 8x24 lamps).
I will be using Ansys Fluent for these simulations on a 12 cores and 32 GB RAM workstation.
For simplicity I am starting now with modelling only one channel with one bank of lamps. but I am facing some problems.
1- Geometry : See image
Geometry.jpg
2- Meshing: Used Sweep Meshing for the lamps and Proximity and Curvature meshing method (coarse) for the whole domain. – The generated number of elements is 30 Million Cells!!! See images for the mesh and the named selections.
mesh1.jpg
mesh2.jpg

3- Fluent:
Models:
- Transient - Pressure Based - Gravity Enabled = -9.8 y direction
-Viscous >> Turbulent >> realizable k-epsilon model >> scalable wall functions
-Multiphase >> VOF model >> Open Channel >> Explicit
Primary Phase: Air & Secondary Phase: Water
Boundaries:
-MassflowInlet ( one inlet for both water and air)
-Pressure Outlet
-Symmetry >> Top face of the channel (not sure about this one?!!)
-Walls >> No slip walls

Solution Methods:
Gradient: Least Square Cell Based
Pressure: PRESTO
Momentum: Second Order Upwind
VOF: Gro-Reconstruct
Turbulent Kinetic Energy: Second Order Upwind
Turbulent Dissipation: First Order Upwind
Transient: First Order Implicit
Under relaxation Factors: Default values
Initialization: Standard Initi./ Compute from Inlet / Relative to cell zone
Calculations: Time step 0.075 s

َQuestions:
When I open the model in fluent, it takes long time to open, to initialize the solution, and when I started to run the calculations I waited for about 8 hours and it wasn’t even giving the residuals for first time step. It showed the global courant number =11, I assume this is a lot and not accepted!!!
So can you please tell me what is wrong my model and how can I modify it to give more smooth and accurate results? Is it limited by the RAM that is why the computer can’t handle such big model, do I need larger memory size than 32GB ?!!



I have watched a lot of tutorials and read many posts from this forums that really helped me to learn how to work with Ansys as a beginner but I still need to learn more, so I appreciate your advice and help to get my model work correctly..Thanks for you all and sorry for the long post..

[flotus1]

There clearly is something wrong if the mesh for the first simplified model contains 30 million cells. You wont be able to run the "full" setup on any workstation.
So before buying more ram you should take a step back from the details of the setup and ask why you need 30 million cells.
And if a simplified model, whose purpose is to get familiar with the setup options, could not be represented on a mesh with a lower cell count.
It it often a good idea to use 2D models in this first stage, especially since your simulation seems to include lots of physical effects that make the computation expensive anyway.

[yahmed]

Originally Posted by flotus1

There clearly is something wrong if the mesh for the first simplified model contains 30 million cells. You wont be able to run the "full" setup on any workstation.
So before buying more ram you should take a step back from the details of the setup and ask why you need 30 million cells.
And if a simplified model, whose purpose is to get familiar with the setup options, could not be represented on a mesh with a lower cell count.
It it often a good idea to use 2D models in this first stage, especially since your simulation seems to include lots of physical effects that make the computation expensive anyway.

Thank you flotus for your reply, so do you think there is a better meshing method that can reduce number of elements while maintaining good quality ??

[flotus1]

There has to be some way to reduce the computational cost of this first model, either through a more suitable meshing method, a lower overall cell count or a simplified 2D model.

[yahmed]

I did a 2D model instead of 3D for one channel- one bank of lamps.
This is the velocity contour in the domain please...
velocity.jpg

Is is relevant to used periodic boundary condition for my case?, to replicate the the first bank of lamps to have a second bank downstream?
And also can I use symmetry BC to model only half of the channel with only 12 arrays of lamps?? Please look at this figure to understand where do I want to apply symmetric and periodic BCs. Boundaries.png

[flotus1]

We would need to see the actual geometries and physical effects you are trying to model to make recommendations for suitable simplifications.
But to be honest, this is rather something you have to make up your mind about for your own scientific work.