Hello,

I'm currently developing a code to solve the 2-D Navier-Stokes equations for external flow over a NACA 0012 airfoil. The airfoil is contained within a circular boundary. Unstructured meshes are being used.

May I know the boundary conditions to be imposed in order to minimize the size of the circular boundary (and hence the computation time)? Please let me know of suitable references on this matter.

Thanks very much.

(1). The size of the computational domain depends on the Mach number and the angle of attack. So, there is no simple rule to set the outer boundary. This is the reason why CFD simulation is more flexible than the wind tunnel testing,where it outer boundary is constrained by the wind tunnel wall. (2). You can set the outer boundary far away from the airfoil first, and run cases with smaller ones to determine the proper size. The computer is supposed to do the number crunching. So, don't worry about the computing time.(3). The computing time depends on the machine used, the mesh distribution and the method used. It is easier to use a faster computer to reduce the computing time. A proper mesh distribution is also important to reduce the mesh size. (4). In other words, it is not a good idea to reduce the computational domain to reduce the computing time. It defeats the purpose of using CFD simulation.

Hello John,

Thanks very much for your advice.

Is there any radius for the circular boundary that should be used to make the first calculation? The literature seems to suggest an order of 50 chord lengths.

Suppose the AOA were zero, how many chord lengths away from the trailing edge would you suggest be used for generating stretched cells for the wake region?

Regards, Chan K I

(1). I don't have any good number for you, so you can pick a number you like. (2). I have noticed that some methods and boundary condtions can have stronger effect on the computed results, especially for the density-based methods.

Hello John,

May I know the types of boundary conditions which would least affect accuracy? Would they be characteristic-based or simple application of upstream conditions?

Regards

(1). It depends on the numerical formulation. (2). For compressible, transient, density based formulation, it is a good idea to use some kind of characteristics condition. (3). For steady, incompressible formulation, down-stream parallel condition should be good enough.