MacCormack's Solution
 

For my undergrad thesis, I recently implemented the MacCormack's space marching method in Matlab for compressible, inviscid flow over two plate angles to output two shocks in an attempt to verify my fluent solutions. The plate angles are: A1 = 10 deg, and A2 = 11.4 deg. I used the method found in Anderson's book CFD - The Basics with Applications (1995) around pg. 400. I use Courant number of .001 to avoid instabilities and CFL# = .5.

Shock angle and flow properties between the first and second shock match fluent and 1D solutions. Unfortunately, for the second shock, any A2 > 10.2 deg. yields imaginary flow properties behind the shock. Interestingly, the real part of all the imaginary results match fluent and 1D solutions. The 2nd shock is even slightly curved due to the nonuniform flow field, just like FLUENT!

What's the underlying reason for these imaginary results and can I simply ignore the imaginary parts? Any help is appreciated. Thank you.

http://postimg.org/image/gs1mb0ov3/http://postimg.org/image/txh4h4iqn/