I am doing transient simulation of compressible flow over forward facing step. The pressure ratio maintained is 5.5 bar. Whether it is k-epsilon or k-omega it is diverging only when the low mach number flow is reaching the pressure outlet. I was running the simulation at a CFL of 0.5 in a density based solver. I have also tried changing the mesh but nothing seems to work
I have attached the contour just before the result diverges.
Please help.
First, I do not understand if you want to run a time dependent case or you are running a transient case to reach a statistically steady state.
Then, I see that you start from very low Mach number n inflow, expanding then the flow up to supersonic conditions (M=3) conditions. I suspect that your BC.s are not correctly prescribed according to the characteristic directions.
Try first a case at lower supersonic conditions, check the BC.s and the numerical stability constraints
Filippo Maria Denaro The bubble in front of the step would be traveling back and forth. So thought of doing a transient flow so that i can track the movement at regular intervals.
It is a pressure inlet, so you just have to give the inlet pressure, it would evolve in time. Its a CD nozzle, so a flow starts from subsonic and reaches supersonic at the exit.
Earlier I ran the simulation with just the nozzle(no step) and it ran without any problem-no divergence. But now when i do it with the step with the same conditions, it diverges when the low mack number flow reaches the exit.
In my opinion, that makes no sense. The supersonic flow cannot transfer upstream the presence of the step, therefore if your simple C-D case works, the problem can be likely due to a numerical instability.
However, to catch a real physical time-dependent behavior, the URANS is not the correct choice.
@ Filippo Maria Denaro I tried steady but same thing is happening again
Have you tried to simulate only the straight duct with the frontal step setting a supersonic inflow?
@ Anant Diwakar i have given non reflecting pressure outlet at 0 gauge pressure with 5% backflow Turbulent intensity and 10% backflow turbulent viscosity ratio
the outflow seems to develop both subsonic and supersonic conditions, that should be managed with care
Sourabh Bhardwaj I tried supersonic 3D turbulent flow in a square duct using my density base code (AUSM+up with SA model), which has supersonic and subsonic conditions (in the boundary layer) at the outflow, using zero gradient conditions. It worked for me. You can check that testcase on NASA turbulence modelling webpage.
For only subsonic conditions at the outflow, I use characteristic based BCs.
Can you try with zero gradient bc and let me know if it works.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running