CFD simulation of a NACA airfoil – Geometry making, meshing, processing and post processing
The geometry consists of multiple joined parts that I will use later to refine the mesh in certain areas. You can download the geometry here – Download.
The total length of the fluid domain is approximately 14.1 m with a maximum width of 5.3 m. Moreover, the chord length of the airfoil is 0.76 m and the maximum thickness is 0.09m. I uploaded a picture of the whole 2D model:
I dissected the fluid domain into 3 major parts. The first part which is the outer domain has the biggest size cells. The second part, the middle zone, has relatively smaller size cell than the first part. Finally, the inner domain with the smallest mesh cell size.
In total, I meshed my geometry with 362,060 elements, with 95 % hexahedral cells. Additionally, I used a generic maximum and minimum face size of 6.10-2 m and 3.10 -4 m respectively. Furthermore, I created 2 face sizing sets for both the middle and inner parts. I chose an element size of 8.10-3 m for the middle part and 4.10-3m for the inner part.
I added an inflation layer at the boundary of the airfoil wall to capture the viscous layer accurately. I set 10 layers with smooth transitions as shown in the figure below:
Processing – ANSYS Fluent
I decided to carry out a transient simulation so I chose the transient option from the general menu and added the gravitational acceleration (-9.81 m/s2 in the Y direction). I also used the density based solver because I am dealing with a compressible flow at relatively high velocities.
I selected the K-Omega SST model which is suitable for this type of simulations (adverse pressure gradients, transonic flows, and airfoils).
I activated the Energy model and then I changed the fluid material from air with constant density to air as an ideal gas.
During the meshing process, I created 2 boundaries, a Pressure Farfield boundary and a wall boundary that represents the airfoil itself. I entered 0.45 for the Mach Number.
In the last part of running the simulation, I initialised using the hybrid initialisation method and finally, I chose a time step of 0.00005 S and adjusted (increasing /decreasing) accordingly.