The artificial perturbation imposed at the inlet is the sum of three sinusoidal functions with the same frequency (1.8 randian/s and f=0.3Hz)and different phases. And it is concentrated in the shear layer, the numerical scheme I use is extended MacCormack--- a explicit time integration method. while I take a nondimensinal timestep=0.01, the result oscilates intensily at the inlet, mass density and temperature decrease to a unphysical extent. How to solve this problem, please help!

Without seeing your problem firsthand, I'll offer this... It seems like some sort of entry shock. Could be numeric, or physical.

Perhaps you could offer a few more insights on your model?

diaw...