[anuprdk]

Hi, I am simulating a turbulent flow behind a 2d cylinder at Re 15000 using the k-w SST model without any wall functions. The distance of the first node from the wall comes to around 0.15 mm to get a y+ value of 1. A region around the cylinder with this thickness has been made and divided into 15 sections using the edge sizing bias option. How do I know if the boundary layer is being captured or not? Am I to mesh each division of the boundary layer such as the buffer layer, inertial layer etc separately? I am getting a Cd of around 0.97, Strouhal number of 0.24 and the mass conservation is also being obeyed.

Also, I have used a time step which needs more than 50 iterations in the first two to three time steps and later converges in 20 steps. Is this all right?

Please advise

Anup

[anuprdk]

Hi again,

I did try to crack up on getting the plot for y+. But I am not sure if this is correct. Why such a low value, I do not understand. And should I increase my mesh element size to get a value close to y+ = 1.

The cylinder is placed at 0,0 and has a radius of 0.075 m.

y+.jpg

Anup

[lehoanganh07]

Hi,
How do you use SST k-o without wall modeling. i think SST k-o use enhanced wall treatment as default and it can hand all range of y+

[anuprdk]

Originally Posted by lehoanganh07

Hi,
How do you use SST k-o without wall modeling. i think SST k-o use enhanced wall treatment as default and it can hand all range of y+

I found the distance to the first node from the wall for y+ = 1 using the relation y = (y+) x (kinematic viscosity of water) / (frictional velocity)
and it comes to around 0.15 mm.

Why is it that then Fluent shows y+ value to be in the range 0.001 to 0.007 ?

thanks
Anup

[virendra_p]

Just to help you recheck your calculations of y+ -> http://www.computationalfluiddynamic...t-cell-height/
If your calculations are correct and 0.15mm gives y+ ~ 1 then put the first mesh layer around cylinder at that height. Then use growth factor of 1.2 and put 10 layers above it so at y+ ~ 30 you will have around 10 layers and this is sufficient to capture the boundary layer (viscous effects) and beyond it you enter the turbulent region where the model switches to k-e method and thus you are able to resolve both Low-Re (viscous region) and High-Re (turbulent region) effectively.
To further improve accuracy you can involve more sophisticated turbulent models which i guess will require a bit a survey. Good luck and let me know how it works out.

[anuprdk]

Originally Posted by virendra_p

Just to help you recheck your calculations of y+ -> http://www.computationalfluiddynamic...t-cell-height/
If your calculations are correct and 0.15mm gives y+ ~ 1 then put the first mesh layer around cylinder at that height. Then use growth factor of 1.2 and put 10 layers above it so at y+ ~ 30 you will have around 10 layers and this is sufficient to capture the boundary layer (viscous effects) and beyond it you enter the turbulent region where the model switches to k-e method and thus you are able to resolve both Low-Re (viscous region) and High-Re (turbulent region) effectively.
To further improve accuracy you can involve more sophisticated turbulent models which i guess will require a bit a survey. Good luck and let me know how it works out.

Hi Virendra,

Thank you so much for the reply. I did create a region around the cylinder at 0.15mm but I did the inflation meshing inside this region and that could be the reason why I am getting a y+ value in the range 0.003 - 0.007. Is it all right if it is this low? And should be I using the 'Low Re Correction' option with SST k-w ??

Thanks

Anup

[virendra_p]

if y+ ~1 is around 0.15mm then thats where your first point should be, creating an inflation layer within is unnecessary (and the cause for low y+ you get). Viscous sub-layer will last till y+ ~ 10 to 30 for such high Re so if you put first point at 0.15 mm and then use a growth factor of 1.2 for subsequent 10 layers then both viscous sub-layer and boundary layer will be resolved and all the details will be captured. You can just play with the position of subsequent layers like say use growth factor of 1.1 and put 15 layers and so on and then compare your results to get gird in-dependency.
As for Low-Re corrections, i think it won't be necessary but for further help, read this --> http://www.cfd-online.com/Forums/flu...ega-model.html

[anuprdk]

Originally Posted by virendra_p

if y+ ~1 is around 0.15mm then thats where your first point should be, creating an inflation layer within is unnecessary (and the cause for low y+ you get). Viscous sub-layer will last till y+ ~ 10 to 30 for such high Re so if you put first point at 0.15 mm and then use a growth factor of 1.2 for subsequent 10 layers then both viscous sub-layer and boundary layer will be resolved and all the details will be captured. You can just play with the position of subsequent layers like say use growth factor of 1.1 and put 15 layers and so on and then compare your results to get gird in-dependency.
As for Low-Re corrections, i think it won't be necessary but for further help, read this --> http://www.cfd-online.com/Forums/flu...ega-model.html

Hi Virendra,

I did follow your advise and now the value of y+ is very close to 1. Thank you for all the advice.

Anup