Prism Layer (Boundary Layer) Thickness

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

 March 27, 2013, 08:55 Prism Layer (Boundary Layer) Thickness #1 New Member   Peter Join Date: Mar 2013 Posts: 1 Rep Power: 0 Hi, i'm using the Star-CCM+ software to model a vehicle and a little confused about the prism layer thickness. I know that the prism layer is to model the boundary layer, but the equation for boundary layer contains the Reynolds number. The Reynolds number changes at each point along the length of the car which will mean that the boundary layer will change. Therefore, what should my prism layer thickness be as prism layer thickness can't change along the length of the car .. Thanks in advance, Peter.

March 27, 2013, 14:32
#2
Senior Member

Ryne Whitehill
Join Date: Aug 2009
Posts: 312
Rep Power: 16
Quote:
 Originally Posted by Peter-27 Hi, i'm using the Star-CCM+ software to model a vehicle and a little confused about the prism layer thickness. I know that the prism layer is to model the boundary layer, but the equation for boundary layer contains the Reynolds number. The Reynolds number changes at each point along the length of the car which will mean that the boundary layer will change. Therefore, what should my prism layer thickness be as prism layer thickness can't change along the length of the car .. Thanks in advance, Peter.

Calculate boundary layer for the end of the vehicle, set prism layet to capture that.

i.e. if you were doing a flat plate 1 m long, you would use "1 m" in your boundary layer calculation. Overkill upfront, but it ensures you capture it all.

 April 6, 2013, 13:37 #3 Senior Member   Gajendra Gulgulia Join Date: Apr 2013 Location: Munich Posts: 144 Rep Power: 11 Hey Peter I am also very new to Star CCM+ and this question has perplexed me too.. Even I am trying to figure out the logic behind the first prism layer thickness value, however the default values are 33 % of the base value. If your flow is laminar or highly viscous in the region of interest, you can reduce the value to 20% or even lower and use the all y+ wall treatment model with fine mesh around the region of interest. This model blends the viscous and turbulent profiles smoothly. The finer mesh in the region of interest can be obtained by accessing the following in Star CCM+ Region of interest(Wall Type Boundary)> Mesh Conditions>Custom Boundary Growth Rate> Slow, Medium, Fast options are available you can choose among them depending upon your flow. I hope this was useful

 June 18, 2013, 06:54 #4 New Member   Alamaas Join Date: Aug 2009 Posts: 5 Rep Power: 14 Hi everyone! If my prism layer has 10 cells (for example) and the overall thickness of the prim layer is equal to the boundary layer thickness, do i still need to worry about y+ wall treatment or is it automatically taken care of? Thank you!

January 5, 2015, 11:49
#5
Member

Yuanchuan Liu
Join Date: Oct 2012
Posts: 31
Rep Power: 11
Hi Alamaas, I know it has been a while since you posted. But I recently have an exact problem as yours. Have you finally figured it out?
As far as I am concerned, there are two parameters defining the boundary: the overall boundary layer thickness and the distance between the first cell and the wall. I read from elsewhere that there should be at least 10 cells inside the boundary layer while in order to use wall functions, you also need to make sure y+ should lie between 30 and 300. Besides, the growth rate should be best set as 1.2-1.25. There are just so many variables to consider. And I doubt all of them can be satisfied at the same time. BTW how do you determine the boundary layer thickness when you generate mesh? Do you use the Blasius solution for flat plates? Cheers,

Best wishes,

Yuanchuan
Quote:
 Originally Posted by Alamaas Hi everyone! If my prism layer has 10 cells (for example) and the overall thickness of the prim layer is equal to the boundary layer thickness, do i still need to worry about y+ wall treatment or is it automatically taken care of? Thank you!

May 6, 2015, 03:55
#6
Senior Member

Join Date: Mar 2015
Posts: 250
Rep Power: 10
Quote:
 Originally Posted by SailorLiu Hi Alamaas, I know it has been a while since you posted. But I recently have an exact problem as yours. Have you finally figured it out? As far as I am concerned, there are two parameters defining the boundary: the overall boundary layer thickness and the distance between the first cell and the wall. I read from elsewhere that there should be at least 10 cells inside the boundary layer while in order to use wall functions, you also need to make sure y+ should lie between 30 and 300. Besides, the growth rate should be best set as 1.2-1.25. There are just so many variables to consider. And I doubt all of them can be satisfied at the same time. BTW how do you determine the boundary layer thickness when you generate mesh? Do you use the Blasius solution for flat plates? Cheers, Best wishes, Yuanchuan
Hi Yuanchuan,

I am just answering to your last question now. Generally, I would use the Blasius solution for laminar flow and the Prandtl solution for a turbulent boundary layer.
Regarding your other question, I have to do some research first. You should at least take into account your choice of turbulence model and whether you are using wall functions or not.

Have a nice day,

Kate

Last edited by KateEisenhower; May 6, 2015 at 04:03. Reason: addition

 May 19, 2015, 13:01 #7 Member   Nils Hennig Join Date: Apr 2015 Posts: 44 Rep Power: 9 Hi, iīm also working at my boundary layer. My scenario is a propeller at open-water conditions. I try to set a boundary-layer which is scientific correct. I used the Prandtl-formula and calculated the over-all thickness of the layer. After that i calculated the closest cell to the wall. The results gave me an over-all thickness of 2.1mm and a cloesest cell of 0.00456mm for a y+=1. If the closest cell is larger than 0.01mm the solution converge, but if itīs smaler it doesnīt. Does anyone have a hint for my problem? Thank you in advance Nils

 May 21, 2015, 09:59 #8 Member   Nils Hennig Join Date: Apr 2015 Posts: 44 Rep Power: 9 I solved my problem. I used the Distribution Mode: "Wall Thickness" and the "Hyperbolic Tangent" Stretching Funktion. If the mesh is still bad, i can use "Cell Quality Remediation"...it works fine

April 7, 2020, 13:23
#9
Senior Member

MA
Join Date: Mar 2020
Posts: 163
Rep Power: 4
Quote:
 Originally Posted by Fiedde1887 I solved my problem. I used the Distribution Mode: "Wall Thickness" and the "Hyperbolic Tangent" Stretching Funktion. If the mesh is still bad, i can use "Cell Quality Remediation"...it works fine

Have you used the empirical formula provided by ITTC: Y = y+*L/0.712*Re^0.9 to get thickness of first prism layer? And how did you get the total thickness of the prism layer to be fed into the mesher? In my case, I am using a blade geometry.

 March 27, 2022, 03:46 Prandtl formula #10 New Member   Saurav Joshi Join Date: Mar 2022 Posts: 2 Rep Power: 0 Hi, Everyone seems to be referring to the prandtl formula to determine the prism layer thickness. Can you please let me know what is the exact prandtl formula?

 March 27, 2022, 05:41 #11 Senior Member   Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 4,784 Rep Power: 56 It's referring to a correlation for the boundary layer thickness over a flat plate. The one for the laminar case is the Blasius solution, the turbulent one is sometimes called the Prandt solution. You would typicall find this in your fluids textbook or from a handbook. Note that these are for the canonical flat plate. There's many other variants that may be more suitable for your specific problem, such as a plate with a trip strip. There's also different correlations for internal flows in pipes.

 March 27, 2022, 08:22 #12 New Member   Saurav Joshi Join Date: Mar 2022 Posts: 2 Rep Power: 0 Can the flat plate assumption be taken for a cylinder to straight up overkill and assume the prism layer thickness of the cylinder based on that?

 March 27, 2022, 12:25 #13 Senior Member   Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 4,784 Rep Power: 56 Doing a lot of work is overkill because it'll still won't be that accurate. You pretty much just want to be in the right order of magnitude. And then run the case and verify your prism layer was sufficient. You also need to check the wall y+ after the fact anyway. Don't spend more than 10 minutes on it.