CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > General Forums > Main CFD Forum

Choosing Turbulence Model for transitionl flows

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

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   May 27, 2015, 05:48
Post Choosing Turbulence Model for transitionl flows
  #1
New Member
 
Hannes
Join Date: Apr 2015
Posts: 3
Rep Power: 9
hk04 is on a distinguished road
Hello guys,

I am currently working on a 3D flow simulation of an orifice for high pressure homogenization for my thesis. The orifice is part of a 2D2C µPIV (Particle Image Velocity) flow measurement system. Since the system cannot handle velocities above 100-150 m/s the Reynolds numbers only range from about 300-3000. This is clearly not fully turbulent. I understand that most turbulence models base on fully turbulent flows to be fully applicable. I am looking for the most accurate model for my case. So if you have any experience which model provides the best solutions – please share it with me


Here are some more information:

First of all – I am using Fluent – but since this is a question regarding turbulence models I hope it is okay to post it in the Main CFD Forum
The goal of my investigation is to compare the limitations of both techniques – CFD and PIV. Most CFD Users use the experimental results of PIV measurements to validate their simulations but I would like to do it both ways. The 2D2C µPIV measurement produces errors due to particles moving out of the 2D measurement plane. The movement in z-direction produces errors since PIV measurement compares two images and calculates the speed of the particles due to the change of position. Since my simulation should give an idea of this error it must be as accurate as it can be. There are 3D3C measurement techniques such as stereo PIV – but it is not applicable (and affordable) to our problem.
At the moment I am thinking about a 4 equ Transition SST or 7 equ Reynolds Stress model.

Thank you for your input!!!!

Hannes


Last but not least some images of my orifice (results of a laminar calculation – Re=230)

First picture: View of the orifice from above (diameter of the canal = 0,2mm)

Second picture: Side view – the orifice is build like this to be able to visualize the flow in the canal. The orifice is made from steel with acrylic glas on top to be accessible for the laser and camera
hk04 is offline   Reply With Quote

Old   May 27, 2015, 14:45
Default
  #2
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,045
Rep Power: 64
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
My opinion is that for your goal the only correct approach would be DNS...any turbulence modelling is by itself a source of errors therefore the comparison will be not really meaningful
FMDenaro is offline   Reply With Quote

Old   May 29, 2015, 02:20
Default
  #3
New Member
 
Hannes
Join Date: Apr 2015
Posts: 3
Rep Power: 9
hk04 is on a distinguished road
Unfortunally fluent does not support DNS.

The question is:
Does anyone have the experience to predict which turbulence model would suit best for my case with transitional flow (Re 300-3000), a rather simple geometry (see pictures) and pressure inlet and outlet.

I know that there is no such thing as an exact model and I will have to be very careful with my results and their interpretation.
hk04 is offline   Reply With Quote

Old   May 29, 2015, 02:45
Default
  #4
New Member
 
Join Date: Sep 2012
Posts: 11
Rep Power: 11
JulianP is on a distinguished road
Quote:
Originally Posted by hk04 View Post
Unfortunally fluent does not support DNS.
This is not correct. DNS are nothing a CFD code needs to "support". Simply choose to do a laminar flow without a turbulence model and resolve your grid down to the Kolmogorov scale. What indeed is a problem is that you probably won't have enough computational power to finish your run in a reasonable amount of time.
JulianP is offline   Reply With Quote

Old   May 29, 2015, 02:45
Default
  #5
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,045
Rep Power: 64
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
Quote:
Originally Posted by hk04 View Post
Unfortunally fluent does not support DNS.

The question is:
Does anyone have the experience to predict which turbulence model would suit best for my case with transitional flow (Re 300-3000), a rather simple geometry (see pictures) and pressure inlet and outlet.

I know that there is no such thing as an exact model and I will have to be very careful with my results and their interpretation.

Fluent does support, DNS, it is simply the laminar model! You must use secondo order scheme and an a fine grid, nothing else...

Or you can try LES
FMDenaro is offline   Reply With Quote

Old   May 29, 2015, 04:02
Default
  #6
New Member
 
Hui Zhu
Join Date: May 2015
Posts: 4
Rep Power: 8
zhuhui is on a distinguished road
Hi,
I believe the only way for FLUENT to simulate a transitional jet flow is the transitional SST model. DNS may be conceptually correct, but unfortunately the numerical dissipation in FLUENT code will fail you.
Remember typical LES (such as Smagorinsky model) can never simulate transition from laminar to turbulence, since the sub-grid stress model is designed for a fully turbulent flow.
zhuhui is offline   Reply With Quote

Old   May 29, 2015, 04:11
Default
  #7
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,045
Rep Power: 64
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
Quote:
Originally Posted by zhuhui View Post
Hi,
I believe the only way for FLUENT to simulate a transitional jet flow is the transitional SST model. DNS may be conceptually correct, but unfortunately the numerical dissipation in FLUENT code will fail you.
Remember typical LES (such as Smagorinsky model) can never simulate transition from laminar to turbulence, since the sub-grid stress model is designed for a fully turbulent flow.

FLuent can be used in DNS switching to second order central discretization instead of upwinding. However, the grid should be very fine.

Furthermore, LES with dynamic Smagorinsky model could be used to simulate transitional flows, however some past experience with LES implemented in Fluent showed to take some care in the analysis of the results
FMDenaro is offline   Reply With Quote

Old   May 29, 2015, 04:16
Default
  #8
New Member
 
Hui Zhu
Join Date: May 2015
Posts: 4
Rep Power: 8
zhuhui is on a distinguished road
Quote:
Originally Posted by FMDenaro View Post
FLuent can be used in DNS switching to second order central discretization instead of upwinding. However, the grid should be very fine.

Furthermore, LES with dynamic Smagorinsky model could be used to simulate transitional flows, however some past experience with LES implemented in Fluent showed to take some care in the analysis of the results
You are correct. I have no idea whether FLUENT supports Dynamic Smagorinsky.
zhuhui is offline   Reply With Quote

Old   May 29, 2015, 04:26
Default
  #9
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,045
Rep Power: 64
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
Quote:
Originally Posted by zhuhui View Post
You are correct. I have no idea whether FLUENT supports Dynamic Smagorinsky.
yes, it has the option
FMDenaro is offline   Reply With Quote

Old   May 29, 2015, 14:09
Default
  #10
Senior Member
 
sbaffini's Avatar
 
Paolo Lampitella
Join Date: Mar 2009
Location: Italy
Posts: 1,790
Blog Entries: 29
Rep Power: 35
sbaffini will become famous soon enoughsbaffini will become famous soon enough
Send a message via Skype™ to sbaffini
Be aware that when the 'transitional' label is attached to a RANS like model it usually means that it has a sort of 'smart' switcher to predict the trsndition location for, say, attached flows over wings. They clearly cannot work on overall transitional flows. Still, the results with some low Re RANS model will not be that bad if the dimulation converges. I confirm everything said by FMDenaro on Fluent and DNS/LES. You can do both, the schemes snd the models are there. My experience with a similar flow (Re=22000 at the orifice of a lab scale burner) says that LES with Fluent is achievable and trustable for such case. Still, you need to set it up properly.
sbaffini is offline   Reply With Quote

Old   May 29, 2015, 14:24
Default
  #11
New Member
 
Hannes
Join Date: Apr 2015
Posts: 3
Rep Power: 9
hk04 is on a distinguished road
Thank you all for your input!

I will measure and simulate 5-8 different flows. I will keep you updated on my results.

If anyone has more experince or ideas to share: you are more than welcome!!!
hk04 is offline   Reply With Quote

Reply

Tags
orifice, piv, tranisitional flow, turbunce model

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Modified k-e turbulence model UDF Travis Fluent UDF and Scheme Programming 7 November 11, 2018 20:21
An error has occurred in cfx5solve: volo87 CFX 5 June 14, 2013 17:44
K - epsilon VS SST turbulence model Maicol Main CFD Forum 0 November 30, 2012 16:25
What model of turbulence choose to study an external aerodynamics case raffale OpenFOAM 0 August 23, 2012 05:45
Turbulence model for vortical flows Dan Chambers STAR-CCM+ 0 December 30, 2010 09:07


All times are GMT -4. The time now is 16:16.