# [ANSYS Meshing] Infilation - Number of layers

 Register Blogs Members List Search Today's Posts Mark Forums Read

 September 4, 2017, 13:28 Infilation - Number of layers #1 Member   numan Join Date: Sep 2014 Posts: 30 Rep Power: 8 Hi guys, For my CFD simulation, during meshing, I calculate first layer height of the inflation according to the desired y+. I wanted to ask how we decide the number of layers? I see some studies use 5, some studies use 20 layers. Thanks .

 September 4, 2017, 18:15 #2 Member   Join Date: Jan 2015 Posts: 62 Rep Power: 8 From what a CFD consultant with a PhD told me, "use as many inflation layers as you can afford". Also you don't want a huge size difference between your last inflation layer and the next cell. So for a real small first layer height (SST on turbomachinery perhaps) you may need 40 layers or more to get the last layer to be a similar size to the rest of the elements. You'll need less for higher y+ turbulence models. Or less if you have a finer mesh outside the inflation layer. Sent from my iPhone using CFD Online Forum mobile app

 September 5, 2017, 14:52 #3 Member   numan Join Date: Sep 2014 Posts: 30 Rep Power: 8 Thank you, I will consider that.

September 6, 2017, 03:19
#4
Super Moderator

Sijal
Join Date: Mar 2009
Posts: 4,542
Blog Entries: 6
Rep Power: 51
Quote:
 Originally Posted by mrswordf1sh Hi guys, For my CFD simulation, during meshing, I calculate first layer height of the inflation according to the desired y+. I wanted to ask how we decide the number of layers? I see some studies use 5, some studies use 20 layers. Thanks .
If you using hybrid wall fucntions (Y+ can be 10 not necessarily 1, which currently fluent and cfx uses) 20 Layers are enough. USe expansion ratio 1.2

For Y+=1 when you want to fully resove viscous sub layer, use 40 layers. Use expansion ratio 1.1 to 1.15.

Any one who is uing 5 layers is just trying to show that he is resolving boundary layer. Just useless.

But if you are studying laminar to turbulne transition, I tend to use 100-120 Layers in boundary layer with expansion ratio of 1.05 or lower.

September 13, 2017, 10:07
#5
Member

numan
Join Date: Sep 2014
Posts: 30
Rep Power: 8
Quote:
 Originally Posted by Far If you using hybrid wall fucntions (Y+ can be 10 not necessarily 1, which currently fluent and cfx uses) 20 Layers are enough. USe expansion ratio 1.2 For Y+=1 when you want to fully resove viscous sub layer, use 40 layers. Use expansion ratio 1.1 to 1.15. Any one who is uing 5 layers is just trying to show that he is resolving boundary layer. Just useless. But if you are studying laminar to turbulne transition, I tend to use 100-120 Layers in boundary layer with expansion ratio of 1.05 or lower.
I am using k-e Realizable. I think it is best for me to use 20 inflation layers. I am using Enhanced-Wall Treatment and Menther-Lechner and my y+ is below 5. Do you think it is ok?

February 22, 2021, 05:10
#6
New Member

lucas s
Join Date: Jul 2013
Location: Grenoble, France
Posts: 12
Rep Power: 9
Quote:
 Originally Posted by Far Any one who is uing 5 layers is just trying to show that he is resolving boundary layer. Just useless.

Even if I agree that 5 layers is not enough to solve a boundary layer, you cannot say that it is useless. When you cannot even afford more than 5 layers, you can still get interesting enough results.

However you have to be aware that your results are not very precise. Particularly, you can expect to have significant discrepancies in the computed skin friction.

But it is always a compromise between the precision you need and the precision you can afford.

February 22, 2021, 08:13
#7
Super Moderator

Sijal
Join Date: Mar 2009
Posts: 4,542
Blog Entries: 6
Rep Power: 51
Quote:
 Originally Posted by LoUcAsss Even if I agree that 5 layers is not enough to solve a boundary layer, you cannot say that it is useless. When you cannot even afford more than 5 layers, you can still get interesting enough results. However you have to be aware that your results are not very precise. Particularly, you can expect to have significant discrepancies in the computed skin friction. But it is always a compromise between the precision you need and the precision you can afford.
Well, I tend to agree with you.