[maksay]

I am trying to solve a crossflow problem where air is fed to the domain at certain freestream Mach number condition. For simplicity let's assume, air is also injected via the injector at a different condition than the inlet air.
In the figure above there is a problem with imposing boundary condition at the injector.

When the flow in the domain is initialised with the inlet conditions, the results are as following:
If alpha = 90 degrees (angle of injection) with respect to the horizontal plane, injection boundary conditions are well applied (i.e. the properties at the injector surface remain constant with time as expected).
However, if alpha is between 0 to 90 degrees with respect to the horizontal plane, the properties at the injector surface (i.e. including injector pressure or temperature or Mach number) change with every iteration and does not remain what was prescribed (i.e. sonic injection with a constant static pressure and temperature). I tried all the allowable inlet boundary conditions; however it does not help.

However, when the flow is initialised with the injection boundary conditions regardless of alpha, value the injection boundary condition values are always correct and preserved.

It is not reasonable to initialise the problem with the injection boundary conditions as it would take longer to converge. My question is, how to impose the right injector boundary conditions such that the injected flow properties are imposed at the injector and remain constant with time (like they are supposed to). Which boundary condition should I use for the injection. Thank you for your answers.

Solver:
Ansys 19.1
Fluid: Air
Density: ideal gas
Viscosity: Inviscid
Solver: Density based

Boundary conditions are as following:

Inlet:
Velocity inlet boundary conditions
M1=2.5
P1=50 kPa
T1=600 K

Injection:
M2=1
P2=1.7MPa
T2=250K
Alpha = 30 deg

Outlet : Pressure outlet
Bottom boundary: isothermal wall at T=300K

Others : Symmetry