[ashvinc9]

Hi Xianbei,

Sorry for my late reply. I just noticed it now.

Sure, we have made some comparison between the two solvers: rk4+projection and pisoFoam. I would encourage you to look the following two papers in order to see/learn the differences between RK-4 method and the built-in second order time integration.

1) http://www.sciencedirect.com/science...45793014000334
2) http://www.sciencedirect.com/science...65997814001513

- Ashvin

[Adlak]

Originally Posted by ashvinc9

Dear all,

I have been running the Open channel flow simulation using wall function approach. I have done done several LES for open channel flow using different wall-functions, Sgs models, numerics, etc, and I would like to share my results for further suggestions and results corrections.

These all simulations are for relatively higher Re_tau, e.g. 7500 to 60000, and in most of the cases, the grid is really coarse. Earlier (in OF 2.0 and 2.1), I was using "nuSgsUSpaldingWallFunction" wall-function but now in the newer version of OF, it also possible to employ almost all the wall-function which are made for RANS. In my case, I have used nuSgsUSpaldingWallFunction and nutUSpaldingWallFunction, and nutURoughWallFunction for smooth, and rough wall conditions, respectively.

The LES results for open channel flow using wall-functions are reasonably good as well as the velocity solutions agree reasonably well with the Log-law for both smooth and rough wall surfaces. Actually, in all the cases, LES velocity profile underestimate the Log-law in a near-wall region and produce a kink only for few cell centred values after the first one from the wall (please see the results in the attached file: Static_SGSModel.pdf ). I have been trying a lot in order to rid out of that kink near wall boundary but didn't find any reason for such a kink so far. I also discussed with other researcher those who use LES-wall-function approach, and they say that it is common to have a such a behaviour for few cell centred values where the velocity solution go away from the Log-law. Also, that velocity defect occurs only for few cells (or nodes) so it shouldn't be too serious problem in my opinion. Please correct me if I am wrong? Please share your ideas if you have had experienced such a behaviour of velocity using wall function in LES.

SGS model validation:
Since, the coarse grid has been used in almost all LES for open channel flow, the SGS models play important roles in the simulations, and thus, the results are very sensitive to the SGS model used. From the open channel flow testing, it seems that one equation eddy viscosity (oneEqEddy) model gives good results for the coarse LES. I also tried standard Smagorinsky model but no improvement was observed in the near-wall agreement as well as poor agreement with Log-law compared to that of using oneEqEddy model. When using any Dynamic or Lagrangian SGS models on the same case (same grid, Re_tau, numerical schemes, averaging time, solver), the LES results are really bad, the velocity profile seems to be non-stable as well as the velocity profile is not even comparable to the Log-law.I do not know why any Dynamic or Lagrangian SGS models are not working as good as static SGS model, although they are suppose to produce better results because of the dynamic procedure of finding the model constant. I have tried, e.g. dynOneEqEddy, homogeneousDynOneEqEddy, homogeneousDynSmagorinsky and dynLagrangian SGS models on the same case. But the results are not acceptable. . The results using dynamic SGS models are attached in the file: Dynamic_SGSModel.pdf. Please find the attachments.

Does any one have idea behind this problem? Did I miss something to take into account when employing the dynamic SGS model? I appreciate your comments and suggestions.

Regards,
Ashvin

Hi ashvin,

Can u help me to run dynamicLagrangian SGS model ? I run my case but facing following error :

--> FOAM FATAL ERROR:
incompatible dimensions for operation
[flm[0 4 -5 0 0 0 0] ] + [flm[1 1 -5 0 0 0 0] ]

From function checkMethod(const fvMatrix<Type>&, const fvMatrix<Type>&)
in file /mnt/home/SW/CFDSupportFOAM3.0/OpenFOAM-3.0.x/src/finiteVolume/lnInclude/fvMatrix.C at line 1295.

FOAM aborting

#0 Foam::error:rintStack(Foam::Ostream&) at ??:?
#1 Foam::error::abort() at ??:?
#2 void Foam::checkMethod<double>(Foam::fvMatrix<double> const&, Foam::fvMatrix<double> const&, char const*) at ??:?
#3 Foam::tmp<Foam::fvMatrix<double> > Foam:perator+<double>(Foam::tmp<Foam::fvMatrix<d ouble> > const&, Foam::tmp<Foam::fvMatrix<double> > const&) at ??:?
#4 Foam::LESModels::dynamicLagrangian<Foam::EddyDiffu sivity<Foam::ThermalDiffusivity<Foam::Compressible TurbulenceModel<Foam::fluidThermo> > > >::correct() at ??:?
#5 ? at ??:?
#6 __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
#7 ? at ??:?
Aborted (core dumped)

If u can help, it will be great for my work.

[Santiago]

Hi guys,

I just saw your discussion and I wanted to put my 2 cents on the issue. For the last month I have developed a projection method similar (see Guermond&Shen) to that of Chaudhari for the incompressible N-S equations, and as validation I run the classic channel flows with different Re. The difference of my case with oodlesChannel (or channel365) is that I drive my flow using a certain pressure gradient, not by setting some flow.

Additionally, with the risk of sparking a salty debate, its my opinion that whenever you run openFOAM without any turbulence model you should not call your results 'DNS' but more 'NO-MODEL LES'. This given the 2nd-order FV nature of the library itself, and the boundary conditions that you can impose. In fact, calling 'cyclic' boundary conditions 'periodic' carries a promise that the developers where well aware of and thus decided to call them as the former.

In any case, you'll find attached the velocity profiles and RMS of channel flows run using Re_tau=171 and Re_tau=410. In both the dynamic lagrangian model of Meneveau was used. In general results seem decent.

[vonboett]

Tahnk you, that really helps!

[godfatherBond]

Originally Posted by Santiago

Hi guys,

I just saw your discussion and I wanted to put my 2 cents on the issue. For the last month I have developed a projection method similar (see Guermond&Shen) to that of Chaudhari for the incompressible N-S equations, and as validation I run the classic channel flows with different Re. The difference of my case with oodlesChannel (or channel365) is that I drive my flow using a certain pressure gradient, not by setting some flow.

Additionally, with the risk of sparking a salty debate, its my opinion that whenever you run openFOAM without any turbulence model you should not call your results 'DNS' but more 'NO-MODEL LES'. This given the 2nd-order FV nature of the library itself, and the boundary conditions that you can impose. In fact, calling 'cyclic' boundary conditions 'periodic' carries a promise that the developers where well aware of and thus decided to call them as the former.

In any case, you'll find attached the velocity profiles and RMS of channel flows run using Re_tau=171 and Re_tau=410. In both the dynamic lagrangian model of Meneveau was used. In general results seem decent.

Hi Santiago,
I am trying to use the dynamic Lagrangian model of Meneveau(1996) for reacting flows, but somehow for me the flm and fmm values become unbounded at some later stage during the simulation and the Co shoots above 1 and time steps sturuggles to maintain the maxCo and crashes finally. This however is not to be the case with Smagorinksy model.
I am using the IC of flm and fmm form Meneveau(1996) and BC as zero gradient. The flow a is free shear jet flow. I am using `backward` scheme time differencing and `linear` spatial discretisation schems with maxCo 0.35. The mesh is fully structured with max aspect ratio of 15 and max non-orthogonality of 18.

It will be beneficial for the CFD community if you could kindly share the setup values i.e. IC, BC of flm and fmm and the fvSchmes for the dynLag case.


Thanks

[Santiago]

Originally Posted by godfatherBond

Hi Santiago,
I am trying to use the dynamic Lagrangian model of Meneveau(1996) for reacting flows, but somehow for me the flm and fmm values become unbounded at some later stage during the simulation and the Co shoots above 1 and time steps sturuggles to maintain the maxCo and crashes finally. This however is not to be the case with Smagorinksy model.
I am using the IC of flm and fmm form Meneveau(1996) and BC as zero gradient. The flow a is free shear jet flow. I am using `backward` scheme time differencing and `linear` spatial discretisation schems with maxCo 0.35. The mesh is fully structured with max aspect ratio of 15 and max non-orthogonality of 18.

It will be beneficial for the CFD community if you could kindly share the setup values i.e. IC, BC of flm and fmm and the fvSchmes for the dynLag case.


Thanks

The boundary condition you set (zeroGradient) for flm is not correct, if there's a wall. One question: What is flm? is an integral of Lij*Mij. How Lij behaves at a wall? Must be zero, by definition. Hence, flm is *zero* at a wall. In general, there is no "rule" on how to define flm/fmm, it's case dependent. I suggest you to start from the definition of the variables themselves.

Another thing: The dynLagrangian Smagorinsky model implemented in OF (whichever version, fork, etc) is not consistent at the test filter level, thus the results obtained are not going to be "correct". I recently published an article discussing this issue:

https://www.mdpi.com/2311-5521/4/3/171

Cheers,

S

PS.: Initial conditions for a channel flow? A TS wave (doesn't matter the dimensionality). Set-up? Gauss linear all the way. Time? Not Euler.

[godfatherBond]

Originally Posted by Santiago

The boundary condition you set (zeroGradient) for flm is not correct, if there's a wall. One question: What is flm? is an integral of Lij*Mij. How Lij behaves at a wall? Must be zero, by definition. Hence, flm is *zero* at a wall. In general, there is no "rule" on how to define flm/fmm, it's case dependent. I suggest you to start from the definition of the variables themselves.

Another thing: The dynLagrangian Smagorinsky model implemented in OF (whichever version, fork, etc) is not consistent at the test filter level, thus the results obtained are not going to be "correct". I recently published an article discussing this issue:

https://www.mdpi.com/2311-5521/4/3/171

Cheers,

S

PS.: Initial conditions for a channel flow? A TS wave (doesn't matter the dimensionality). Set-up? Gauss linear all the way. Time? Not Euler.

Hi Santiago,
Thanks for the reply.
1.For my case, it is a jet flow with no walls. Hence I set the values to ZeroGradient. I hope it is correct?
2. The inlet BC for velocity is derived from a precursor pipe flow simulation.
The IC for flm and fmm are based on Meneveau(1996) so that Cs=0.16 initially.
3. Time Scheme: backward Convection: U->linear, flm,fmm->limitedLinear 1.0.
4. fvSolution(flm,fmm)->PBiCG(DILU)

Somehow, I feel that the setup is very unstable as it caused the fmm value to explode somewhere in between the simulation. Is there something incorrect?
Thanks in advance!

[Santiago]

Originally Posted by godfatherBond

Hi Santiago,
Thanks for the reply.
1.For my case, it is a jet flow with no walls. Hence I set the values to ZeroGradient. I hope it is correct?
2. The inlet BC for velocity is derived from a precursor pipe flow simulation.
The IC for flm and fmm are based on Meneveau(1996) so that Cs=0.16 initially.
3. Time Scheme: backward Convection: U->linear, flm,fmm->limitedLinear 1.0.
4. fvSolution(flm,fmm)->PBiCG(DILU)

Somehow, I feel that the setup is very unstable as it caused the fmm value to explode somewhere in between the simulation. Is there something incorrect?
Thanks in advance!

From the precursor flow you should be able to determine flm and fmm. For the slip conditions flm/fmm may be zeroGradient. for the solution of the equations put upwind/euler, as you want everthing to be bounded. Notice that meneveau also does upwinding in time, given that flm/fmm are integrated along flow trajectories. Notice that the clipping done by openfoam is "wrong", since the clipping should be made to the source term of the transport equations, via a heaveside function.

[godfatherBond]

Originally Posted by Santiago

The boundary condition you set (zeroGradient) for flm is not correct, if there's a wall. One question: What is flm? is an integral of Lij*Mij. How Lij behaves at a wall? Must be zero, by definition. Hence, flm is *zero* at a wall. In general, there is no "rule" on how to define flm/fmm, it's case dependent. I suggest you to start from the definition of the variables themselves.

Another thing: The dynLagrangian Smagorinsky model implemented in OF (whichever version, fork, etc) is not consistent at the test filter level, thus the results obtained are not going to be "correct". I recently published an article discussing this issue:

https://www.mdpi.com/2311-5521/4/3/171

Cheers,

S

PS.: Initial conditions for a channel flow? A TS wave (doesn't matter the dimensionality). Set-up? Gauss linear all the way. Time? Not Euler.

Hi Santiago,
About the inconsistency you mentioned for the test filter, can't we use the Laplace test filter( or anisotropic filter) implemented in OF? I understand that the grid level filtering is implicit but for test filter may be specified explicitly as Simple/anisotropic/Laplace. Will it still not give "correct" the results as you have achieve in your article using a Laplace test filter?

[Santiago]

Originally Posted by godfatherBond

Hi Santiago,
About the inconsistency you mentioned for the test filter, can't we use the Laplace test filter( or anisotropic filter) implemented in OF? I understand that the grid level filtering is implicit but for test filter may be specified explicitly as Simple/anisotropic/Laplace. Will it still not give "correct" the results as you have achieve in your article using a Laplace test filter?

Well, the anisotropic filter is the only valid (strictly speaking) explicit filter for hexahedral, non uniform grids. And no, using the anisotropic filter does not resolve the inconsistency I refer to when using the dynamic smagorinsky models.