Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Turbulence in pipe flow

Hi,

I am studying the Navier Stokes equation in the context of pipe flow. As of now, I am using Water, modeling it as incompressible, with outlet at 0 Pa and a parabolic inlet profile. (currently at 2D, but moving to 3D as things advance).

I am currently using Fluid Flow>Single-Phase Flow>Laminar Flow, as my inlet is a stable velocity profile and the turbulence, if any, will arise due to obstacles inside the pipe.

I have been playing with it, doing the Von Karman vortexes simulation (www.comsol.com/showroom/gallery/361/), looking into a moving mesh (www.comsol.com/showroom/gallery/361/), as my object might be able to move subjected to the flow.

My question is: how far can I push the Laminar Flow module? I am looking for turbulence, so I don't think I should be using a Turbulence specific model (it forces turbulence, and requires me to input parameters specific for turbulence, that clearly doesn't exist at the start). On the other hand, I don't know if the Laminar Flow module will be able to propely simulate turbulence.

What should I be using then?

Thank you for input.

11 Replies Last Post May 14, 2014, 9:45 PM EDT
Posted: 5 years ago May 11, 2012, 9:58 AM EDT
Hi

I'm no CFD expert, but still, I do not beleive the laminar flow should be used to approach turbulence, however you can plot the Reynolds number and use that to get a feeling where you are. Check the litterature where you should put the Reynolds limit for laminar to turbulent transition. then use the average laminar flow as input intial conditions to a turbulent model

--
Good luck
Ivar
Hi I'm no CFD expert, but still, I do not beleive the laminar flow should be used to approach turbulence, however you can plot the Reynolds number and use that to get a feeling where you are. Check the litterature where you should put the Reynolds limit for laminar to turbulent transition. then use the average laminar flow as input intial conditions to a turbulent model -- Good luck Ivar

Posted: 5 years ago May 11, 2012, 10:31 AM EDT
I understand.

I am computing the Reynold number as the average of the velocity magnitude is the pipe main direction, per pipe cross section. I have experimented with various Re.

The literature is a bit vague as to what Reynold number is the critical one, and the experiments and simulations I found led me to believe the critical Reynold number varies with the perturbation used (cylinder, perpendicular plane, whatever).

Your suggestion is, after simulating for a bit with the Laminar Flow model, I should take the end values and input that to the turbulence model?

Thank you
I understand. I am computing the Reynold number as the average of the velocity magnitude is the pipe main direction, per pipe cross section. I have experimented with various Re. The literature is a bit vague as to what Reynold number is the critical one, and the experiments and simulations I found led me to believe the critical Reynold number varies with the perturbation used (cylinder, perpendicular plane, whatever). Your suggestion is, after simulating for a bit with the Laminar Flow model, I should take the end values and input that to the turbulence model? Thank you

Posted: 5 years ago May 11, 2012, 10:41 AM EDT
Hi

exactly that is "my way" but i'm not regularly using CFD, (more structural, thermal, ACDC, RF ...)

COMSOL has a predefined variable for the "cell" reynolds number related to the local mesh size, by the way most "CFD numbers" are predefined in COMSOL now, check your Postprocessing variable list

--
Good luck
Ivar
Hi exactly that is "my way" but i'm not regularly using CFD, (more structural, thermal, ACDC, RF ...) COMSOL has a predefined variable for the "cell" reynolds number related to the local mesh size, by the way most "CFD numbers" are predefined in COMSOL now, check your Postprocessing variable list -- Good luck Ivar

Posted: 5 years ago May 11, 2012, 11:03 AM EDT
Thank you, I will look into that.
Thank you, I will look into that.

Posted: 4 years ago Sep 20, 2013, 5:18 AM EDT
Hi,

i have almost the same issues. i'm working electromotor, which (should) oil cooled. oil flows through air gap. we are suspecting that the oil causing viscous heating, since it has high viscosity (dynamic).

firstly i assume that this should be a laminarer case. so i calculate with conjugate heat transfer (NITF). At 1000 rpm for 700 second the temp rising fron 20 to 30 degC. The maximum "cellRe" 14 dan Prandt 1500. it's fine, laminar case has Re <10^5 and 0,6<Prandt<2000.

then i tried to use turbulence model type: RANS and turbulence model: k-epsilon.

Doc says "The standard k-ε model is the most widely used turbulence model since it is often a good compromise between accuracy and computational cost (memory and CPU-time)".

But the calculated Re is different. "cellRe" = 1e-10 and Pr is same. It is still laminar case actualy. I'm still wondering, why "cellRe" has a huge difference?

"Low Reynolds number k-ε model" instead of "k-epsilon" wont run.

i calculate it manualy Re =9 and Pr =ca. 1000, also laminar.

I'm using 4.3b

regards
akmal
Hi, i have almost the same issues. i'm working electromotor, which (should) oil cooled. oil flows through air gap. we are suspecting that the oil causing viscous heating, since it has high viscosity (dynamic). firstly i assume that this should be a laminarer case. so i calculate with conjugate heat transfer (NITF). At 1000 rpm for 700 second the temp rising fron 20 to 30 degC. The maximum "cellRe" 14 dan Prandt 1500. it's fine, laminar case has Re

Martijn Weterings
Posted: 4 years ago Sep 20, 2013, 9:33 AM EDT
---- "Low Reynolds number k-ε model" instead of "k-epsilon" wont run. ----

This might be because the wall distance is not calculated in the study node.

This happens if you first add a study that does not require the calculation of wall distance (as standard k-epsilon) and then change the physics to something that does require the wall distance to be calculated.

The study node does not change when you adjust the physics. The solution is to add an extra study node (then you automatically get all the standard stuff for the newly selected physics).
---- "Low Reynolds number k-ε model" instead of "k-epsilon" wont run. ---- This might be because the wall distance is not calculated in the study node. This happens if you first add a study that does not require the calculation of wall distance (as standard k-epsilon) and then change the physics to something that does require the wall distance to be calculated. The study node does not change when you adjust the physics. The solution is to add an extra study node (then you automatically get all the standard stuff for the newly selected physics).

Dr. Ammar Alsabery
Posted: 3 years ago Apr 29, 2014, 10:22 AM EDT
i want do turbulence in porous media,,which model should i choose from add physics?



thank you
i want do turbulence in porous media,,which model should i choose from add physics? thank you

Martijn Weterings
Posted: 3 years ago Apr 30, 2014, 11:59 AM EDT
If you select the brinkman equation (instead of the Darcy flow) then you can include an inertial term (but this is not turbulence).

What kind of problem requires you to consider turbulence in porous media?
If you select the brinkman equation (instead of the Darcy flow) then you can include an inertial term (but this is not turbulence). What kind of problem requires you to consider turbulence in porous media?

Dr. Ammar Alsabery
Posted: 3 years ago Apr 30, 2014, 8:57 PM EDT
actually, i want do the model in the paper i attached, but i don't know which one i should from add physics list
actually, i want do the model in the paper i attached, but i don't know which one i should from add physics list


Martijn Weterings
Posted: 3 years ago May 14, 2014, 8:07 PM EDT
In the paper they consider macroscopic flow.

You might be able to do the same with the k-epsilon equations (not with porosity) in comsol but you do need to be carefull for the difference between modified quanties (with subscript capital D) versus microscopic quantities. Further than that I only notice a difference in the the fifth term on the right of equation 4 which, I believe, you can manually add to the comsol equations by adding a force term. This term stems from darcy's law and is the friction between the porous medium and the fluid.
In the paper they consider macroscopic flow. You might be able to do the same with the k-epsilon equations (not with porosity) in comsol but you do need to be carefull for the difference between modified quanties (with subscript capital D) versus microscopic quantities. Further than that I only notice a difference in the the fifth term on the right of equation 4 which, I believe, you can manually add to the comsol equations by adding a force term. This term stems from darcy's law and is the friction between the porous medium and the fluid.

Dr. Ammar Alsabery
Posted: 3 years ago May 14, 2014, 9:45 PM EDT
thank you very much..
thank you very much..

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.