CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   Flow Phenomena in Nozzle (https://www.cfd-online.com/Forums/main/224136-flow-phenomena-nozzle.html)

jcownbey February 8, 2020 12:58

Flow Phenomena in Nozzle
 
1 Attachment(s)
I simulated a nozzle I created using method of characteristics at mach 6.13. The nozzle is simulated with SU2. I am getting a strange flow phenomena I don't think should be there. The boundary layer appears to be relatively large (x and y are in meters). I have attached a photo. Would this be a result of numerical error?

FMDenaro February 8, 2020 14:36

Quote:

Originally Posted by jcownbey (Post 757235)
I simulated a nozzle I created using method of characteristics at mach 6.13. The nozzle is simulated with SU2. I am getting a strange flow phenomena I don't think should be there. The boundary layer appears to be relatively large (x and y are in meters). I have attached a photo. Would this be a result of numerical error?




I don't know what SU2 solves in your problem. You should better clarify what you are doing. Is that a characteristic method applied for Euler equation ? In the case of omo-entropic expansion you can use the Riemann invariant to develop the exact solution and there is no physical boundary layer.

jcownbey February 8, 2020 15:26

I am doing a RANS calculation using SST to solve my boundary layer. The inlet is defined by the freestream values (calculated with isentropic relations) with a supersonic outlet. The inlet is slightly after the sonic point, so Mach no. is around 1.03 to start.

FMDenaro February 8, 2020 16:15

Quote:

Originally Posted by jcownbey (Post 757241)
I am doing a RANS calculation using SST to solve my boundary layer. The inlet is defined by the freestream values (calculated with isentropic relations) with a supersonic outlet. The inlet is slightly after the sonic point, so Mach no. is around 1.03 to start.




Ok, now it is clear to me, you have first developed the geometry using the Euler equation and the characteristic method, then you are using such geometry to solve a real viscous turbulent flow.
I suppose you are working at a very high Re number, are you using a grid so fine to resolve the BL until the laminar sublayer? Clearly the geometry designed for inviscid isoentropic flow does not ensure that the BL is fully attached to the wall, the case you are simulating is quite challenging.

I have no direct experience on this case at high Mach number, I suggest to check the grid resolution near the wall, I suggest also to try some different turbulence model to see the differences.
What is more, I suggest to try replicate a case where data are available in literature.
You should post also the 1D velocity profiles along some stations in streamwise direction

jcownbey February 10, 2020 10:44

I have done a RANS simulation successfully of existing nozzles, but not with hypersonics (Mach no. was around 2 and 3.5). The spacing at the wall is not the issue according to my y+ value. I checked the vector field of the velocity, and it looks like a boundary layer. I am hoping that changing the boundary conditions to Reimann BC's will mitigate this issue.


All times are GMT -4. The time now is 06:47.