[Fco.Herbert]

Greetings CFDers,
I hope everything is well.

I am trying to predict the 2D aerodynamics (i.e., lift, drag and moment coefficients) of two parallel airfoils using ANSYS 18.1 (Design Modeler + ANSYS Meshing + Fluent). The problem is that I have been unable to properly discretize (mesh) the domain, so as to obtain y+ values below 1, particularly at the interface region between the airfoils. Thus, I was wondering if someone with experience on complex surface meshing could provide guidance/insights. Below is a description of the case study and the points of issue:

1. 2D Geometry:
For simplicity, two identical NACA 4412 airfoils with chord length of 1 m are considered. A centerline separation distance of 0.15 m is imposed between the parallel airfoils. Each airfoil has a sharp trailing edge. A graphical representation of the geometry and the airfoils’ coordinates are attached (see link at the end of the post).

2. 2D Boundary Topology
The case study is formulated as a traditional external flow problem. Modified “C-type” and “H-type” topologies are considered. The “C type” topology encompass both airfoils and the “H type” topology is used at the interface region between the airfoils. The complete region was (initially) divided into 7 faces; 6 surrounded the airfoils and 1 was located between the airfoils. The boundaries were extended up to 10 times the chord length (10 m) in all directions from the trailing edges of the airfoils. A graphical representation of the 2D boundary topology is attached (see link at the end of the post).

3. 2D Inflow and Boundary Conditions
The Reynolds numbers to be considered are 3 x 10^6 and below. The angle of attack (AoA) will vary from -20 deg to + 20 deg. The inlet turbulence intensity and length scale will be arbitrarily defined as 1% and 0.01m, respectively. The airfoils’ surfaces will be smooth with zero roughness (non-slip condition).

4. 2D Spatial Discretization (Meshing)
Ideally, I would like to work with a structured quadrilateral discretization, such as in traditional aerodynamic studies of isolated airfoils (e.g., https://confluence.cornell.edu/displ...ver+an+Airfoil). This is mainly because the boundary layer physics near the airfoils’ surfaces, as well in the far wake region, will be accurately solved. However, it is understandable that a multizone quad/tri discretization (or hybrid mesh) might be required to accurately model the geometry and to spare computational resources.

I have estimated that the first set of nodes should be placed at a distance of about 5e-6 m from the airfoils’ surfaces in order to obtain maximum y+ values below 1 (e.g., this requirement handles the case of Re = 3x10^6 and AoA +-20 deg). I have tried 2 different approaches to create a mesh that is compliant with such requirement:

Case 1: I have considered different “methods” in ANSYS Meshing (e.g., multizone quad/tri, only quad and only tri), as well as different combinations of “sizings” and bias factors at the edges that define the 7 faces of the boundary topology. For some reason, ANSYS Meshing always struggles to mesh the inner region between the airfoils (see link at the end of the post).

Case 2: After several attempts, I decided to merge the 7 faces into a unique face. Then, I considered an “edge sizing” function with 800 elements around each airfoil (soft and no bias). I also specified an “inflation” function at the airfoils’ surfaces with the following properties: “first layer thickness” of 9e-4 m, “maximum layers” of 10 and “growth rate” of 1.05, considering the “pre” inflation algorithm (see link at the end of the post). I got a decent preliminary mesh, however, I could not specify a lower “first layer thickness” without generating an error in ANSYS Meshing: “The mesher tried to modify the mesh for the following edges that have hard sizing”. No other combination of parameters allowed me to define a lower first layer of nodes. The maximum y+ value for this mesh is about 204, which is extremely high.

My sincere thanks in advance for any advice on how to deal with this meshing problem.
Best regards,

PD. The ANSYS projects are located in the following OneDrive link, in case they are of use: https://1drv.ms/f/s!AvQajjZrGwh6h0iqpI_g6aTM9LUY

[Fco.Herbert]

Hi CFDers,
The solution to this problem essentially consists in adopting the approach described in "case 2" and then play with the "inflation" free parameters until a sufficiently refined and quality mesh is obtained. Focus on the inflation growth rate and the maximum layers. Try to avoid defining large growth rates, as this will create an overlapped region of cells between the inflations defined in each airfoil, which in turn will result in the error described in the first post.

I have updated the OneDrive folder (see first post) with a successful mesh for future reference.
Best regards,