Implicit LES

selig5576 · December 15, 2017, 14:31

I am currently having a problem with my implicit LES code. I am currently modelling a round jet using a profile similar to that described in Sagaut. I am using a non-dynamic Smagorinsky model as its the simplest to implement. I am using $\nu_{laminar} = 15,000$ and I see no turbulence. The jet acts like its laminar. If I remove the SGS model and do DNS I see turbulence. My domain is 8 x 8 x 8 and am using 90^3 grid points. I am using simple second order discretization for both advection and viscous terms. As far as I know for implicit LES this should be sufficient to get decent results.

Given I am in the incompressible regime the way I handle my viscous terms is as follows

$\frac{D\bar{u}_{i}}{D t} = - \frac{\partial \bar{p}}{\partial x_{i}} + \frac{\partial}{\partial x_{j}} \left[ (\nu_{laminar} + \nu_{SGS}) \frac{\partial \bar{u}_{i}}{\partial x_{j} }\right]$

It seems like the SGS viscosity is damping out all the turbulence. With that said, the way am discretizing the stress tensor magnitude in the Smagorinsky model is using central differencing as well, which I presume is appropriate. Thoughts?

FMDenaro · December 15, 2017, 15:01

Quote:

Originally Posted by selig5576

I am currently having a problem with my implicit LES code. I am currently modelling a round jet using a profile similar to that described in Sagaut. I am using a non-dynamic Smagorinsky model as its the simplest to implement. I am using $\nu_{laminar} = 15,000$ and I see no turbulence. The jet acts like its laminar. If I remove the SGS model and do DNS I see turbulence. My domain is 8 x 8 x 8 and am using 90^3 grid points. I am using simple second order discretization for both advection and viscous terms. As far as I know for implicit LES this should be sufficient to get decent results.

Given I am in the incompressible regime the way I handle my viscous terms is as follows

$\frac{D\bar{u}_{i}}{D t} = - \frac{\partial \bar{p}}{\partial x_{i}} + \frac{\partial}{\partial x_{j}} \left[ (\nu_{laminar} + \nu_{SGS}) \frac{\partial \bar{u}_{i}}{\partial x_{j} }\right]$

It seems like the SGS viscosity is damping out all the turbulence. With that said, the way am discretizing the stress tensor magnitude in the Smagorinsky model is using central differencing as well, which I presume is appropriate. Thoughts?

what do you mean for this viscosity value??

Then, the grid seems quite coarse and the static Smagorinsky model can give too dissipation. What about the value of the constant?

selig5576 · December 15, 2017, 15:09

So in Saugaut's book they are using 10 x 10 x 10 with domain with 100 grid points in each direction. What I meant by viscosity value I mean the laminar viscosity. My C_s value is 0.15.

FMDenaro · December 15, 2017, 15:19

Quote:

Originally Posted by selig5576

So in Saugaut's book they are using 10 x 10 x 10 with domain with 100 grid points in each direction. What I meant by viscosity value I mean the laminar viscosity. My C_s value is 0.15.

viscosity or Reynolds number??
This value of Cs can be too large and the lenght of the streamwise direction can depend on the inflow conditions (base + superimposed perturbation).

You should post your results for a better discussion.

selig5576 · December 15, 2017, 16:56

"viscosity or Reynolds number?? "

I am talking about viscosity. I am running my simulation with Cs=0.01 to see if I get a change. When it is done I will post my results.

LuckyTran · December 15, 2017, 19:12

[QUOTE=selig5576;675267
It seems like the SGS viscosity is damping out all the turbulence. With that said, the way am discretizing the stress tensor magnitude in the Smagorinsky model is using central differencing as well, which I presume is appropriate. Thoughts?

[/QUOTE]

It should be central yes. I guess you have a problem that needs debugging.

FMDenaro · December 16, 2017, 05:05

Quote:

Originally Posted by selig5576

"viscosity or Reynolds number?? "

I am talking about viscosity. I am running my simulation with Cs=0.01 to see if I get a change. When it is done I will post my results.

Still not clear the value 15,000 for the laminar viscosity ...

December 15, 2017, 14:31	Implicit LES	#1
selig5576 Senior Member Selig Join Date: Jul 2016 Posts: 213 Rep Power: 11	I am currently having a problem with my implicit LES code. I am currently modelling a round jet using a profile similar to that described in Sagaut. I am using a non-dynamic Smagorinsky model as its the simplest to implement. I am using $\nu_{laminar} = 15,000$ and I see no turbulence. The jet acts like its laminar. If I remove the SGS model and do DNS I see turbulence. My domain is 8 x 8 x 8 and am using 90^3 grid points. I am using simple second order discretization for both advection and viscous terms. As far as I know for implicit LES this should be sufficient to get decent results. Given I am in the incompressible regime the way I handle my viscous terms is as follows $\frac{D\bar{u}_{i}}{D t} = - \frac{\partial \bar{p}}{\partial x_{i}} + \frac{\partial}{\partial x_{j}} \left[ (\nu_{laminar} + \nu_{SGS}) \frac{\partial \bar{u}_{i}}{\partial x_{j} }\right]$ It seems like the SGS viscosity is damping out all the turbulence. With that said, the way am discretizing the stress tensor magnitude in the Smagorinsky model is using central differencing as well, which I presume is appropriate. Thoughts?

December 15, 2017, 15:09	Implicit LES	#3
selig5576 Senior Member Selig Join Date: Jul 2016 Posts: 213 Rep Power: 11	So in Saugaut's book they are using 10 x 10 x 10 with domain with 100 grid points in each direction. What I meant by viscosity value I mean the laminar viscosity. My C_s value is 0.15.

December 15, 2017, 16:56	Problem in LES	#5
selig5576 Senior Member Selig Join Date: Jul 2016 Posts: 213 Rep Power: 11	"viscosity or Reynolds number?? " I am talking about viscosity. I am running my simulation with Cs=0.01 to see if I get a change. When it is done I will post my results.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
implicit and explicit filtering in LES	cfdmms	Main CFD Forum	9	April 13, 2018 11:45
Setup for implicit LES	newbie384	STAR-CCM+	2	November 28, 2016 14:06
implicit and explicit filtering in LES	kkpal	OpenFOAM	0	February 19, 2014 03:27
Is it feasible to apply implicit scheme in LES	EagleGao	Main CFD Forum	2	October 24, 2011 15:04
Komogrov time scale and implicit scheme for LES	Li Yang	Main CFD Forum	1	January 17, 2004 02:13

December 15, 2017, 19:12		#6
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,747 Rep Power: 66	[QUOTE=selig5576;675267 It seems like the SGS viscosity is damping out all the turbulence. With that said, the way am discretizing the stress tensor magnitude in the Smagorinsky model is using central differencing as well, which I presume is appropriate. Thoughts? [/QUOTE] It should be central yes. I guess you have a problem that needs debugging.