[rvl565]

Hi i wanted to use rotor 67 as a test case for fan simulations.
To speed up the process i am using the mesh provided by stanford university ( http://aero-comlab.stanford.edu/vdweide/CGNSFiles/ )

I removed the tip clearance from this mesh, other than that the mesh is the same.

I am using the MRF model. The boundary conditions are:
Pressure inlet: total pressure = 101325 ( 0 reference pressure)
Initial gauge pressure =99825
% intensity =1
length scale =0.01

Pressure outlet: gauge pressure 114500
backflow length scale and intensity the same as inlet

Rotor RPM: -16043

I extended the inlet and outlet by 0.15m to create two stationary zones.

I used the komegaSST model with compressibility effects and curvature correction.
The energy equation has been turned on and used the ideal gas equation for the density.

I get reversed flow at the inlet and outlet.

Please help me correct this problem.

Thanks!

[CFD-fellow]

Hi
Have you checked "Radial equilibrium pressure distribution"in pressure outlet B.C?
On how many faces you get reversed flow?No. of iteration?

[rvl565]

Hi,

Number of iterations > 7000, i check the inlet mass flow rate and inlet pressure for convergence as well and they are stable.
Approximately 1000 faces in the pressure outlet and 700 face on the pressure inlet.
I checked the z-velocity contours at the interface at the rotor inlet and mostly the forward flow is concentrated at the tip of the rotor, rest of the flow is in the opposite direction.
I havent tried the radial equilibrium pressure distribution. I use the average pressure specification at the outlet.

[CFD-fellow]

Try outlet vent for outlet and let me know the answer. I will be back tomorrow

[rvl565]

Hi,

I used radail equilibrium and outlet vent at the outlet and i still get reversed flow at the inlet (>1000faces).
If you have time, i could share the case and data files with you on dropbox or google drive.

[CFD-fellow]

Try initialization with a proper velocity value so that the air enter the domain and exit.

[rvl565]

During initialization, the air is entering the domain and exiting the fan. Thus for the first few 100 iterations there is no reversed flow. I tried lowering the exit pressure but that didnt work either.

[sehs15]

i have the same problem .. the back flow is large .. seems like the flow is driven by the pressure difference which is simply from outlet to inlet as the pressure outlet is larger by the pressure ratio

[sehs15]

im using Single reference frame with the fluid,blade,hub and shroud rotating with the reference frame in the compressor direcion , i elimenated the tip clearance to avoid extra errors and instability issues

[rvl565]

I think i did the same, just to be on the same page, you put RPM as -16043 with the flow direction being -z ?
I have extended the inlet and outlet, so that i have 2 zones ( inlet and outlet ) which are stationary and one zone ( rotor ) which is rotating.
If so, what do you think the problem is?
They seem to have used this grid with no problems, take a look at this:
http://aero-comlab.stanford.edu/Pape...2002-30313.pdf

[sehs15]

with your grid the rpm direction is +x

[rvl565]

I think i rotated my grid so that the flow direction was -z ... anyway i can share my grid with you using this forum? if not i will post a dropbox link or something.

[sehs15]

i downloaded the grid from the link you provided in your post ... the flow direction is +x (rotational axis is x)

http://aero-comlab.stanford.edu/vdweide/CGNSFiles/

[rvl565]

here's the link to the cgns file. I have just rotated the grid and removed the tip clearance

If you want i can upload a pointwise file as well (if thats what you prefer)

Thanks a lot!

https://drive.google.com/file/d/0BxN...it?usp=sharing

[sehs15]

i see your grid is rotating about -z .. i dunno if the extended boundaries could help or not .. i think we could add a big room instead to get rid of pressure waves

[rvl565]

my previous experience suggests that adding a plenum to the outlet is not a good idea. I had a lot of convergence issues.... so much so that i finally ended up using the profiles at the outlet (of velocity,turbulence,pressure,etc. ) as the inlet condition to the plenum.
You think it could be a problem with the mesh?
The skewness seems ok and so does the first cell height.

[sehs15]

the aspect ratio is very large about 1000 and the skewness is large also .. but the people in stanford ran the case successfully .. and this is weird

[rvl565]

since this is a low Re mesh the aspect ratio will be large right? unless you want a lot more cells.... i also tried running with a low outlet static pressure but that doesnt work either... any ideas? do you by any chance have access to a better mesh?

[sehs15]

try to solve it laminar with inlet and outlet pressure 1 bar .. if the mass flow rate was large increase the out pressure slightly to 1.5 bar .. i did that with nasa rotor 37 it converged ... now i will solve it with appropriate turbulence model with the correct boundary conditions
.. im using density base solver , i tried the outlet non reflecting boundary but it was unstable

[rvl565]

I will try that and let you know how it goes, but i have tried putting 1atm total pressure at the inlet and 1atm static pressure at the outlet... but that didnt work... Could you share your rotor 37 mesh?

What do you mean by "outlet non reflecting boundary"

with the actual boundary conditions it was unstable you mean?

I dont want to mesh these rotors ... i have a rotor at hand and meshing it is not what i enjoy doing :P