Application of SST model by CEL

 Register Blogs Members List Search Today's Posts Mark Forums Read

 October 26, 2017, 15:59 Application of SST model by CEL #1 New Member   linlin.geng Join Date: Oct 2017 Posts: 22 Rep Power: 8 S11=Velocity u.Gradient X S12=0.5*(Velocity u.Gradient Y+ Velocity v.Gradient X) S13=0.5*(Velocity u.Gradient Z+ Velocity w.Gradient X) S22=Velocity v.Gradient Y S23=0.5*(Velocity v.Gradient Z+Velocity w.Gradient Y) S33=Velocity w.Gradient Z S21=S12 S23=S32 S31=S13 S=sqrt(2*S11^2+2*S12^2+2*S13^2+2*S21^2+2*S22^2+2*S 23^2+2*S31^2+2*S32^2+2*S33^2) a1=5/9 UtSST=Density*a1*Turbulence Kinetic Energy /max(a1 *Turbulence Eddy Frequency, S *Second Blending Function for SST model) The above CEL languages are applied to the CFX code to be the eddy viscosity of SST turbulent model. But the predicted results using the above CEL is not the same as the results predicted by the SST model. I do not know why, is it the CEL problem. If anyone knows why, please inform me in this forum or contact me with my e-mail(mosu2010@outlook.com). thanks

 October 26, 2017, 18:05 #2 Super Moderator   Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,746 Rep Power: 143 I have not checked your implementation in detail, and I have to assume that the simulation you are using to make this assessment is correct as well. In that case a possible explanation of the difference is the different way the function is coupled into the other equations. In the solver version of the SST model the turbulence equations are coupled into the solver tightly, but in CEL the coupling is looser. This means achieving convergence will be harder. linlin.geng likes this.

 October 27, 2017, 04:22 #3 Senior Member     Mr CFD Join Date: Jun 2012 Location: Britain Posts: 361 Rep Power: 14 Have you linearised your source terms? If you haven't read up a book by Patankar called Numerical Heat and Fluid Flow. linlin.geng likes this.

October 27, 2017, 05:18
#4
New Member

linlin.geng
Join Date: Oct 2017
Posts: 22
Rep Power: 8
Quote:
 Originally Posted by ghorrocks I have not checked your implementation in detail, and I have to assume that the simulation you are using to make this assessment is correct as well. In that case a possible explanation of the difference is the different way the function is coupled into the other equations. In the solver version of the SST model the turbulence equations are coupled into the solver tightly, but in CEL the coupling is looser. This means achieving convergence will be harder.
i also consider the situation as you think, but the fact that the big difference between these two method when i simulation the cavitating flows makes me confused, and can you give me some suggestions. thank you very much.

October 27, 2017, 05:21
#5
New Member

linlin.geng
Join Date: Oct 2017
Posts: 22
Rep Power: 8
Quote:
 Originally Posted by JuPa Have you linearised your source terms? If you haven't read up a book by Patankar called Numerical Heat and Fluid Flow.
I just replicate the eddy viscosity of SST model by the CEL languages, so what does it have to do with the source terms. I do not understand. can you give me a clear explaniation. thank you very much.

 October 27, 2017, 05:52 #6 Super Moderator   Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,746 Rep Power: 143 Please attach the CCL of your simulation. To understand your results we will need to know how you have setup the model.

October 27, 2017, 06:31
#7
New Member

linlin.geng
Join Date: Oct 2017
Posts: 22
Rep Power: 8
Quote:
 Originally Posted by ghorrocks Please attach the CCL of your simulation. To understand your results we will need to know how you have setup the model.
# State file created: 2017/10/27 18:22:19
# Build 17.1 2016.04.12-14.50-136032

LIBRARY:
CEL:
EXPRESSIONS:
S = 2^0.5*(S11 ^2+S12 ^2+S13 ^2+S21 ^2+S22 ^2+S23 ^2+S31 ^2+S32 ^2+S33 \
^2)^0.5
S21 = S12
S31 = S13
S32 = S23
a1 = 5/9
utSST = Density *a1 *water.Turbulence Kinetic Energy /max(a1 \
*water.Turbulence Eddy Frequency, S *Second Blending Function for SST \
model)
END
END
END
COMMAND FILE:
Version = 17.1
END

Please check. Thank you very much.

 October 27, 2017, 06:55 #8 Super Moderator   Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,746 Rep Power: 143 Please attach ALL the CCL. Not just the first bit.

October 27, 2017, 07:02
#9
New Member

linlin.geng
Join Date: Oct 2017
Posts: 22
Rep Power: 8
Quote:
 Originally Posted by ghorrocks Please attach ALL the CCL. Not just the first bit.
Attached Files
 utsst.txt (18.0 KB, 7 views)

 October 28, 2017, 06:15 #10 Super Moderator   Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,746 Rep Power: 143 I note that this is a multiphase cavitation model. This adds a whole new dimension of complexity to the simulation. How about you see if your function is the same as the built in SST model on a simpler single phase simulation?

October 29, 2017, 15:30
#11
New Member

linlin.geng
Join Date: Oct 2017
Posts: 22
Rep Power: 8
Quote:
 Originally Posted by ghorrocks I note that this is a multiphase cavitation model. This adds a whole new dimension of complexity to the simulation. How about you see if your function is the same as the built in SST model on a simpler single phase simulation?
yes, when i simulate the single phase using the SST model and the CEL of SST, the difference is very small. but in the case of cavitation prediction, the difference will be not negligible. May be the question of my CEL. i will check it in detail. thank you again.

 Tags sst cel