# k-w SST and Wall functions: How many layers? (Automotive aerodynamics)

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

 April 17, 2020, 10:21 k-w SST and Wall functions: How many layers? (Automotive aerodynamics) #1 New Member   Andrea Stedile Join Date: Feb 2020 Posts: 11 Rep Power: 6 Greetings, I am using the OpenFOAM CFD suite to study the aerodynamic forces that act on a vehicle (more precisely, an Ahmed body). I chose the k-w SST model, with wall functions boundary conditions. I am still a beginner, and unfortunately I have no background in CFD. Right now, I am able to create some simple but fine meshes (with some millions of cells). I taught myself how to add layers on the vehicle's surface mesh and check the y+ values after the solver has run. I usually reach ~ 98 / 99 % of layer thickness. However, I do not understand: How many layers should the vehicle's surface have? I noticed that y+ somewhat depends on the size of the first layer. On the Internet, I saw that, when using Wall functions, y+ should be between 30 and 200 (or 30 and 300); whereas, when using a "full resolution" approach, y+ should be <= 1. I also saw that, with the full resolution approach, the required number of layers should be 10, 20, or even more. In my case, I am using wall functions, but I did not find a resource that points out the required number of layers. Do you have any idea? If this can help, the motorbike tutorial in OpenFOAM, which uses the same model and wall functions, uses just 1 layer.

 April 17, 2020, 17:01 #2 Senior Member   Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,665 Rep Power: 65 It's like asking how many cells you should have in your mesh. The answer is, however many you need to get the answer that you need. That is, we can't tell you unless you have an objective in mind. Even still, there aren't hard rules. Some people need to capture all the boundary layer development in their models and so they pack the boundary layer full of cells. You can do this with layers or you can even do it with regular cells. Some people don't care about the boundary layers. They might use very few layers just to have a nice mesh near surfaces so that you don't end up with velocity vectors pointing into walls. Some people don't even about this and will use no layers. Some people are just doing CFD for funsies, just for the colors and don't do any postprocessing, none of these settings matter for them. The scientific method applies. If you think it might change something, try it and find out. mazhar16823 likes this.

April 22, 2020, 13:42
#3
Senior Member

duri
Join Date: May 2010
Posts: 245
Rep Power: 16
Quote:
 Originally Posted by steddy Greetings, I am using the OpenFOAM CFD suite to study the aerodynamic forces that act on a vehicle (more precisely, an Ahmed body). I chose the k-w SST model, with wall functions boundary conditions. I am still a beginner, and unfortunately I have no background in CFD. Right now, I am able to create some simple but fine meshes (with some millions of cells). I taught myself how to add layers on the vehicle's surface mesh and check the y+ values after the solver has run. I usually reach ~ 98 / 99 % of layer thickness. However, I do not understand: How many layers should the vehicle's surface have? I noticed that y+ somewhat depends on the size of the first layer. On the Internet, I saw that, when using Wall functions, y+ should be between 30 and 200 (or 30 and 300); whereas, when using a "full resolution" approach, y+ should be <= 1. I also saw that, with the full resolution approach, the required number of layers should be 10, 20, or even more. In my case, I am using wall functions, but I did not find a resource that points out the required number of layers. Do you have any idea? If this can help, the motorbike tutorial in OpenFOAM, which uses the same model and wall functions, uses just 1 layer.

In wall integration number of layers implies number of layers in viscous sub laminar usually y+ < 4. But in case of wall function no. of layers is not defined. The first cell should be in valid log law region typically y+ 30 to 200. It doesn't matter how many cells you have in log law region because shear stress is calculated from the log law and not from velocity gradient.

 Tags aerodynamics, automotive, layers, mesh, wallfunction