About Turbulence Intensity (Pipe flow assimilated)

gRomK13 · July 8, 2009, 06:05

Hello everybody,

I am currently running simulations meant to assess the pressure drop across different designs of gagging located at the entrance of a pipe. The 3D meshed domain comprises a first cylinder (A = plenum) with no shear stress on the walls, then the gagging (B = let's say a simple hole), and after that another cynlinder with no-slip condition this time (C = pipe).

I'm carrying out theses cases for different mass flow rates, the Reynolds number ranging from 8000 to 800000. The turbulence model I use is the Realizable K-Epsilon Two Layers in StarCCM+. The fluid is helium under 70 bars/400°C (673K).

(GENERAL QUESTIONS)
1) What range should my Y+ lie in? Between 30 in 100? Actually I'm not sure I really understood this "All-Re" stuff: does it mean the model behaves like a low-Re if the mesh is fine enough, and as a high-Re if it is too coarse??

2) Some cells will locally show up with Y+ values of 300, 500, or even 800. It is very local (especially on the gagging = B), due to the unstructured mesh. Can this be a problem for the exactness of the solution?

(MY PROBLEM)
I decided to specify turbulence boundary conditions (for inlet and outlet) as a length scale and a turbulence intensity ratio.

3) According to your experience of CFD, is there is significant influence of turbulence intensity upon the computation of pressure drop?

4) I know turbulence intensity won't exceed 5% anywhere for the fastest flowrate, but how could I evaluate it for slower conditions?

- Before the gagging (B), the plenum (A) is actually much larger than the pipe (C), even if in my simulation I chose to reduce it to the same order of magnitude (to respect the splitting of the plenum flow into the many pipes existing in reality). I don't know how the turbulence intensity can be assessed for my different operating conditions...

- After the gagging (B), I extruded the pipe (C) just long enough to recover from strong instabilities: the pressure profile turns near linear, even if the velocity profile is not established still (not constant). Anyway, the formula giving turbulence intensity for fully developped pipe flow should remain a good approximation no? (By the way, I don't see why in this relation turbulence intensity decreases with flow velocity......)

I hope my case and my questions are explicited enough (first post out there), don't hesitate to ask for more information if need be.

Thanks for your help

Regards.

gRomK13 · July 10, 2009, 04:11

As noboby will answer, I will just write down what I think or found out, so that this thread will not be completely useless for those who would encounter the same issues:

1) Excerpt from StarCCM+ help :

"The two-layer approach, first suggested by Rodi, is an alternative to the low-Reynolds number approach that allows the K-Epsilon model to be applied in the viscous sublayer. In this approach, the computation is divided into two layers. In the layer adjacent to the wall, the turbulent dissipation rate $\epsilon$ and the turbulent viscosity $\mu_{t}$ are specified as functions of wall distance. The values of [math]\epsilon[/epsilon] specified in the near-wall layer are blended smoothly with the values computed from solving the transport equation far from the wall. The equation for the turbulent kinetic energy is solved in the entire flow. This explicit specification of $\epsilon$ and $\mu_{t}$ is arguably no less empirical than the damping function approach, and the results are often as good or better. In STAR-CCM+, the two-layer formulations will work with either low-Reynolds number type meshes $y^{+}\approx$ 1 or wall-function type meshes $y^{+}$ >30"

Still don't see the difference with high Re approach then....

2) As far as I'm concerned, the simulations converge with not so much oscillations and with residues below $10^{-6}$ , despite these few cells with high $y^{+}$ .

3) For my case, decreasing turbulence intensity from 5% to 1% yields in a variation by less than 1% of the pressure drop. I don't know if such a low influence of turbulence intensity upon pressure drop is a generality.

4) I use the same value ( $I=0.16 \cdot Re^{-0.125}$ , valid for developped pipe flows) for both the inlet and the outlet. This remains a source of uncertainty to me...

Hope this can help

July 8, 2009, 06:05	About Turbulence Intensity (Pipe flow assimilated)	#1
gRomK13 New Member Join Date: Jul 2009 Location: near Marseille, France Posts: 7 Rep Power: 17	Hello everybody, I am currently running simulations meant to assess the pressure drop across different designs of gagging located at the entrance of a pipe. The 3D meshed domain comprises a first cylinder (A = plenum) with no shear stress on the walls, then the gagging (B = let's say a simple hole), and after that another cynlinder with no-slip condition this time (C = pipe). I'm carrying out theses cases for different mass flow rates, the Reynolds number ranging from 8000 to 800000. The turbulence model I use is the Realizable K-Epsilon Two Layers in StarCCM+. The fluid is helium under 70 bars/400°C (673K). (GENERAL QUESTIONS) 1) What range should my Y+ lie in? Between 30 in 100? Actually I'm not sure I really understood this "All-Re" stuff: does it mean the model behaves like a low-Re if the mesh is fine enough, and as a high-Re if it is too coarse?? 2) Some cells will locally show up with Y+ values of 300, 500, or even 800. It is very local (especially on the gagging = B), due to the unstructured mesh. Can this be a problem for the exactness of the solution? (MY PROBLEM) I decided to specify turbulence boundary conditions (for inlet and outlet) as a length scale and a turbulence intensity ratio. 3) According to your experience of CFD, is there is significant influence of turbulence intensity upon the computation of pressure drop? 4) I know turbulence intensity won't exceed 5% anywhere for the fastest flowrate, but how could I evaluate it for slower conditions? - Before the gagging (B), the plenum (A) is actually much larger than the pipe (C), even if in my simulation I chose to reduce it to the same order of magnitude (to respect the splitting of the plenum flow into the many pipes existing in reality). I don't know how the turbulence intensity can be assessed for my different operating conditions... - After the gagging (B), I extruded the pipe (C) just long enough to recover from strong instabilities: the pressure profile turns near linear, even if the velocity profile is not established still (not constant). Anyway, the formula giving turbulence intensity for fully developped pipe flow should remain a good approximation no? (By the way, I don't see why in this relation turbulence intensity decreases with flow velocity......) I hope my case and my questions are explicited enough (first post out there), don't hesitate to ask for more information if need be. Thanks for your help Regards.

July 10, 2009, 04:11		#2
gRomK13 New Member Join Date: Jul 2009 Location: near Marseille, France Posts: 7 Rep Power: 17	As noboby will answer, I will just write down what I think or found out, so that this thread will not be completely useless for those who would encounter the same issues: 1) Excerpt from StarCCM+ help : "The two-layer approach, first suggested by Rodi, is an alternative to the low-Reynolds number approach that allows the K-Epsilon model to be applied in the viscous sublayer. In this approach, the computation is divided into two layers. In the layer adjacent to the wall, the turbulent dissipation rate $\epsilon$ and the turbulent viscosity $\mu_{t}$ are specified as functions of wall distance. The values of [math]\epsilon[/epsilon] specified in the near-wall layer are blended smoothly with the values computed from solving the transport equation far from the wall. The equation for the turbulent kinetic energy is solved in the entire flow. This explicit specification of $\epsilon$ and $\mu_{t}$ is arguably no less empirical than the damping function approach, and the results are often as good or better. In STAR-CCM+, the two-layer formulations will work with either low-Reynolds number type meshes $y^{+}\approx$ 1 or wall-function type meshes $y^{+}$ >30" Still don't see the difference with high Re approach then.... 2) As far as I'm concerned, the simulations converge with not so much oscillations and with residues below $10^{-6}$ , despite these few cells with high $y^{+}$ . 3) For my case, decreasing turbulence intensity from 5% to 1% yields in a variation by less than 1% of the pressure drop. I don't know if such a low influence of turbulence intensity upon pressure drop is a generality. 4) I use the same value ( $I=0.16 \cdot Re^{-0.125}$ , valid for developped pipe flows) for both the inlet and the outlet. This remains a source of uncertainty to me... Hope this can help GV77 likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
High speed compressible flow through pipe	Munni	Main CFD Forum	6	December 7, 2015 12:33
decay of turbulence intensity	littlelz	CFX	3	May 16, 2014 17:44
Pipe Flow and Turbulence	cfd101	Main CFD Forum	0	May 13, 2007 13:19
Domain for Pipe Flow	Swarup	FLUENT	0	September 29, 2005 00:25
Hydrostatic pressure in 2-phase flow modeling (CFX4.2)	HB &DS	CFX	0	January 9, 2000 14:19