CFD Online Discussion Forums
Page 2 of 3 1 2 3
Show 40 post(s) from this thread on one page

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM (https://www.cfd-online.com/Forums/openfoam/)
-   -   Copy of kwSST paper (https://www.cfd-online.com/Forums/openfoam/60745-copy-kwsst-paper.html)

MisterX February 13, 2013 06:55
I am looking for the paper
"Heat Transfer Predictions based on Two-Equation Turbulence Models with Advanced Wall Treatment" and
"Elements of Industrial Heat Transfer Prediction" 16th Brazilian Congress of Mechanical Engineering (COBEM), Nov. 2001 respectively, too.
I am studying the implementation of the kOmegaSST model.

It would be very nice if someone could send me the paper.

Markus.Trompa@t-online.de

Thank you very much.

mb.pejvak April 26, 2013 02:15
Quote:
Originally Posted by mb.pejvak (Post 352043)
I found it. if anyone wants it please send an email to me I can send it.
the paper is:
"Heat Transfer Predictions based on Two-Equation Turbulence Models with Advanced Wall Treatment"
T.Esch and F.R. Menter - Turbulence, Heat and Mass Transfer 4.

mb.pejvak@gmail.com
Dear friends;
I am so sorry for the mistake I made. the paper I have is
"Heat Transfer Predictions using Advanced Two Equation Turbulence Models" by Wolfgang Vieser, Thomas Esch and Florian Menter.
if anyone needs this paper, please send an Email to me.I will be happy to send it to him/her.

Bojan August 15, 2013 05:38
Elements of Industrial Heat Transfer Prediction
 
Thank you in advance for sending to me this paper! :)

bojans86@gmail.com

david89 August 19, 2013 09:45
Hi, could anyone send to me the papers dealt in this thread, please?
I mean:
"Heat Transfer Predictions using Advanced Two Equation Turbulence Models" by Wolfgang Vieser, Thomas Esch and Florian Menter
and
"Heat Transfer Predictions based on Two-Equation Turbulence Models with Advanced Wall Treatment"
T.Esch and F.R. Menter - Turbulence, Heat and Mass Transfer 4

davide.peloso@gmail.com

Thanks in advance.

ehsoon February 25, 2014 12:08
request for 2 papers
 
Hi friends. I am looking for these 2 papers. this will be my pleasure if you can email them to me.

"Heat Transfer Predictions using Advanced Two Equation Turbulence Models" by Wolfgang Vieser, Thomas Esch and Florian Menter
and
"Heat Transfer Predictions based on Two-Equation Turbulence Models with Advanced Wall Treatment"
T.Esch and F.R. Menter - Turbulence, Heat and Mass Transfer 4

my email address is: e.a.tahmasebi[at]gmail.com

Thanks in advance.

littlescorpion March 25, 2014 04:20
The paper exists.
 
Hi, everybody.
According to the bugs report below, the paer, Elements of Industrial Heat Transfer Prediction, does exist. But we can not read it.
http://openfoam.org/mantisbt/print_b...php?bug_id=232

Are there anyone find it helpful to understand the komegasst model in OF after reading that paper, Heat Transfer Predictions based on Two-Equation Turbulence Models with Advanced Wall Treatment?

Besides, can anybody send me a copy of Heat Transfer Predictions based on Two-Equation Turbulence Models with Advanced Wall Treatment. Thank you a lot. My email is shawnqin2014 at gmail.com

shawn

msuaeronautics March 25, 2014 21:22
These two are available online:

http://www.doc88.com/p-387479694178.html

http://nippon.zaidan.info/seikabutsu...okyo_ts059.pdf

dancfd April 22, 2014 21:09
Hello all,

I contacted a number of authors who made reference to the COBEM paper, and found someone who pointed me to what looks to be the complete proceedings from COBEM 2001, available in the public domain no less:

http://www.abcm.org.br/pt/wp-content...f/Lectures.pdf

Therein is the difficult to find Menter 2001 paper. Hopefully this will make the paper available to one and all, and end the aura of mystery surrounding it!

Regards,

Daniel

dancfd April 22, 2014 23:28
I have a question about how the equation in the 2001 paper was corrected, as indicated in kOmegaSST.H:

Code:
Also note that the error in the last term of equation (2) relating to
    sigma has been corrected.
Equation 2 is the omega transport equation, and the last term is given by

+(1-F_1)2\rho\sigma_{\omega 2}\frac{1}{\omega}\frac{\partial k}{\partial x_j}\frac{\partial\omega}{\partial x_j}

kOmegaSST.C lines 409-434 give the same equation as follows:

Code:
    tmp<fvScalarMatrix> omegaEqn
    (
        fvm::ddt(rho_, omega_)
      + fvm::div(phi_, omega_)
      - fvm::Sp(fvc::ddt(rho_) + fvc::div(phi_), omega_)
      - fvm::laplacian(DomegaEff(F1), omega_)
    ==
        rhoGammaF1*GbyMu
      - fvm::SuSp((2.0/3.0)*rhoGammaF1*divU, omega_)
      - fvm::Sp(rho_*beta(F1)*omega_, omega_)
      - fvm::SuSp
        (
            rho_*(F1 - scalar(1))*CDkOmega/omega_,
            omega_
        )
    );

    omegaEqn().relax();

    omegaEqn().boundaryManipulate(omega_.boundaryField());

    solve(omegaEqn);
    bound(omega_, omegaMin_);
And lines 401-404:

Code:
    volScalarField CDkOmega
    (
        (2*alphaOmega2_)*(fvc::grad(k_) & fvc::grad(omega_))/omega_
    );
And kOmegaSST.H Lines 38-41:

Code:
Note that this implementation is written in terms of alpha diffusion
    coefficients rather than the more traditional sigma (alpha = 1/sigma) so
    that the blending can be applied to all coefficients in a consistent
    manner.
Not sure what "SuSp" means in the transport equation, but otherwise the only change I see is that the \sigma_{\omega 2} is replaced by alphaOmega2, which is \frac{1}{\sigma_{\omega 2}}. Is that all, or am I missing something?

Thanks for any advice,

Daniel

msuaeronautics April 23, 2014 20:29
Quote:
Originally Posted by dancfd (Post 487598)
I have a question about how the equation in the 2001 paper was corrected, as indicated in kOmegaSST.H:

Code:
Also note that the error in the last term of equation (2) relating to
    sigma has been corrected.
Equation 2 is the omega transport equation, and the last term is given by

+(1-F_1)2\rho\sigma_{\omega 2}\frac{1}{\omega}\frac{\partial k}{\partial x_j}\frac{\partial\omega}{\partial x_j}

kOmegaSST.C lines 409-434 give the same equation as follows:

Code:
    tmp<fvScalarMatrix> omegaEqn
    (
        fvm::ddt(rho_, omega_)
      + fvm::div(phi_, omega_)
      - fvm::Sp(fvc::ddt(rho_) + fvc::div(phi_), omega_)
      - fvm::laplacian(DomegaEff(F1), omega_)
    ==
        rhoGammaF1*GbyMu
      - fvm::SuSp((2.0/3.0)*rhoGammaF1*divU, omega_)
      - fvm::Sp(rho_*beta(F1)*omega_, omega_)
      - fvm::SuSp
        (
            rho_*(F1 - scalar(1))*CDkOmega/omega_,
            omega_
        )
    );

    omegaEqn().relax();

    omegaEqn().boundaryManipulate(omega_.boundaryField());

    solve(omegaEqn);
    bound(omega_, omegaMin_);
And lines 401-404:

Code:
    volScalarField CDkOmega
    (
        (2*alphaOmega2_)*(fvc::grad(k_) & fvc::grad(omega_))/omega_
    );
And kOmegaSST.H Lines 38-41:

Code:
Note that this implementation is written in terms of alpha diffusion
    coefficients rather than the more traditional sigma (alpha = 1/sigma) so
    that the blending can be applied to all coefficients in a consistent
    manner.
Not sure what "SuSp" means in the transport equation, but otherwise the only change I see is that the \sigma_{\omega 2} is replaced by alphaOmega2, which is \frac{1}{\sigma_{\omega 2}}. Is that all, or am I missing something?

Thanks for any advice,

Daniel
"SuSp" is a numeric function in OpenFOAM that handles the term in question either implicitly or explicitly based on its sign.

Menter's 2003 paper is more in line with the OpenFOAM implementation than the 2001 paper that's referenced in the header file (if, for nothing else, the fact that the coefficients are written as 1/sigma, which gives values written in the 2003 paper).

http://cfd.mace.manchester.ac.uk/flo...-SST-paper.pdf

dancfd April 28, 2014 10:39
Hello Eric,

Thanks for your response. I am having a hard time finding the alpha = 1/sigma difference in the papers, though; looking at the equations in the two papers, they both have the sigma terms (sigma_omega1, etc) and they are used the same way.

I think I found the error in sigma, but it is not in the last term as indicated in the kwSST.H code; rather, it is in the second-to-last term where Menter 2001 gives:

\frac {\partial}{\partial x_j} \left[ \left( \mu + \frac{\mu_t}{\sigma_\omega} \right) \frac{\partial \omega}{\partial \ x_j}\right]

But Menter 2003 gives:

\frac {\partial}{\partial x_j} \left[ \left( \mu + \mu_t\sigma_\omega \right) \frac{\partial \omega}{\partial \ x_j}\right]

Which is consistent with OpenFOAM:

Code:
alphaOmega(F1)*mut_ + mu()
Except for the alphaOmega term:

Code:
tmp<volScalarField> alphaOmega(const volScalarField& F1) const
        {
            return blend(F1, alphaOmega1_, alphaOmega2_);
        }
I do not see how the implementation of alpha as 1/sigma is consistent with either Menter paper. What am I missing, or is there a mistake in the OpenFOAM implementation?

Thanks,

Daniel

msuaeronautics April 28, 2014 11:23
Dan,

I think the OpenFOAM implementation is very accurate. The coefficients are definitely written as alpha = 1/sigma, but the nomenclature differs from the literature.

For instance, if you look at Menter 2001, \sigma_{k1} = 1.176. The reciprocal of this is 0.85034, which is the exact value given to \alpha_{k1} in the code.

Now, OpenFOAM writes these as \alpha etc, even though Menter keeps the \sigma nomenclature in 2003 (even though the 2003 and 2001 coefficients are certainly not the same in face value).

Long story short, the \sigma coefficients that are in 2003 are in fact what OpenFOAM calls "alpha" in the code, and \alpha = 1/\sigma (sigma being the value of any coefficient from 2001).

Naturally, this also means that:

Code:
alphaOmega(F1)*mut_ + mu()
is correct and consistent with Menter 2001 and 2003. ;)

Hopefully I have answered your questions properly. I have combed through the incompressible portion of the code thoroughly and have read several of Menter's papers. Looking at the compressible code, there doesn't seem to be any fundamental differences.

dancfd April 28, 2014 12:07
Hello Eric,

Thanks for your response. I understand that alpha and sigma are reciprocals, therefore it follows that to remain consistent each instance of sigma in the Menter papers would have to appear as 1/alpha in the OpenFOAM code, would it not?

If I look at the code for the turbulent kinetic energy equation, I understand the laplacian term to be implemented as follows:

\frac{\partial^2 k}{\partial x_j^2} \left[ \alpha_k \mu_t + \mu \right]
with alpha_k1 = 0.85.

The [Menter 2001] paper shows the same term as follows:

\frac{\partial}{\partial x_j} \left[ \frac{\partial k}{\partial x_j}\left( \frac{\mu_t}{\sigma_k} + \mu \right) \right]
with sigma_k1 = 1.176.

The [Menter 2003] paper shows the same term as follows:

\frac{\partial}{\partial x_j} \left[ \frac{\partial k}{\partial x_j} \left( \sigma_k \mu_t + \mu \right) \right]
with sigma_k1 = 0.85.

I am satisfied that the above expressions are equivalent. So if sigma from Menter 2003 is equivalent to alpha in OpenFOAM, which is equivalent to 1/sigma in Menter 2001, then how can it be explained that the equations for CD_kw, F1, as well as the last term in the omega equations, are the same in both papers, even though each contains a sigma term to the same power, but they have reciprocal values? The difference in the omega equation can be explained as an error in the Menter 2001 paper as explained by the OpenFOAM code, but if the equations for CD_kw and F1 are the same in both papers yet the constants are different, I do not see how they can be equivalent.

Perhaps there is an inconsistency between the two Menter papers and not with the OF code.

Daniel

msuaeronautics April 28, 2014 12:22
Quote:
Originally Posted by dancfd (Post 488702)
So if sigma from Menter 2003 is equivalent to alpha in OpenFOAM, which is equivalent to 1/sigma in Menter 2001, then how can it be explained that the equations for CD_kw, F1, as well as the last term in the omega equations, are the same in both papers, even though each contains a sigma term to the same power, but they have reciprocal values? The difference in the omega equation can be explained as an error in the Menter 2001 paper as explained by the OpenFOAM code, but if the equations for CD_kw and F1 are the same in both papers yet the constants are different, I do not see how they can be equivalent.

Perhaps there is an inconsistency between the two Menter papers and not with the OF code.

Daniel
Daniel,

I have pondered the same on these two terms before but dropped the issue some time back. It may just be an inconsistency between Menter's papers, as you say, but may be worth a try with reciprocated coefficients.

dancfd April 28, 2014 19:05
Hello Eric,

Thanks for your feedback. I looked through the code with a fine-toothed comb to try to convince myself that it matched one or the other of the references, and found some more differences:

F1:
- arg1 (OpenFOAM) = min(10, arg1(Menter)); why was this done?
- Within arg1, OpenFOAM removed the rho from the numerator and from the CD_kw expression, which is fine if CD_kw < 1e-10; if not, CD_kw = 1e-10 and then the rho does not cancel, making the OF expression = Menter * 1/rho (and therefore a units mismatch).

F2:
- arg2 (OpenFOAM) = min(100, arg2(Menter)); why was this done?

kEqn, omegaEqn:
- Not clear what the influence is of the Su term on the LHS
- Pk appears to be related to the SuSp and rhoGammaF1*GbyMu or min(G...) terms, but I am not certain.

Any ideas why these changes were made to the model?

Thanks,

Daniel

msuaeronautics April 28, 2014 23:07
Quote:
Originally Posted by dancfd (Post 488774)
Hello Eric,

Thanks for your feedback. I looked through the code with a fine-toothed comb to try to convince myself that it matched one or the other of the references, and found some more differences:

F1:
- arg1 (OpenFOAM) = min(10, arg1(Menter)); why was this done?
- Within arg1, OpenFOAM removed the rho from the numerator and from the CD_kw expression, which is fine if CD_kw < 1e-10; if not, CD_kw = 1e-10 and then the rho does not cancel, making the OF expression = Menter * 1/rho (and therefore a units mismatch).

F2:
- arg2 (OpenFOAM) = min(100, arg2(Menter)); why was this done?

kEqn, omegaEqn:
- Not clear what the influence is of the Su term on the LHS
- Pk appears to be related to the SuSp and rhoGammaF1*GbyMu or min(G...) terms, but I am not certain.

Any ideas why these changes were made to the model?

Thanks,

Daniel
The limiting implementations for both F1 and F2 are likely just numerical tools.

I.E., values for F2 very near the wall can get very large, depending on your first cell height (especially thanks to the y^2 value in the denominator). So, adding a ceiling of scalar(100) to this term prevents this (and, I'm assuming, any funny stuff that might happen near the wall when calculating the eddy viscosity).

If you can imagine, since nut is calculated as:

Code:
nut_ = a1_*k_/max(a1_*omega_, F2()*sqrt(S2));
The term

Code:
max(a1_*omega_, F2()*sqrt(S2))
could be dominated by F2, which, without the limiter, might be on the order of 10^20 or more (again, depending on first cell height). That would give very insignificant values of eddy viscosity everywhere in your domain, which would render the model as a whole useless to the solver. The ceiling of 100 may either have been arbitrarily chosen by the programmers, or purely a result of trial-and-error.

Quote:
- Within arg1, OpenFOAM removed the rho from the numerator and from the CD_kw expression, which is fine if CD_kw < 1e-10; if not, CD_kw = 1e-10 and then the rho does not cancel, making the OF expression = Menter * 1/rho (and therefore a units mismatch).
I am not sure about this one- will take another look and comment, if I can. I will also try to comment on your other observation.

msuaeronautics April 29, 2014 09:27
An edit to that post. The tanh makes the large F2 irrelevant, but there is no point in storing so many digits I suppose. Hence the limit.

fabiocfd December 15, 2014 04:05
Menter's papers
 
Good morning,
I'm a PhD student working on heat transfer problems, and it would be very useful for me having a copy of Menter's papers. In particular i'm looking for:

-Menter, F., Esch, T.: "Elements of Industrial Heat Transfer Prediction". 16th Brazilian Congress of Mechanical Engineering (COBEM),Nov. 2001

-Grotjans, H. and Menter, F. "Wall functions for general CFD codes". In Proc. of 4th European Comp. Fluid Dynamics conference, Athens, Greece, 1998, pp. 1112-1117

-"Heat Transfer Predictions based on Two-Equation Turbulence Models with Advanced Wall Treatment" T.Esch and F.R. Menter - Turbulence, Heat and Mass Transfer 4

This is my e-mail address: fabio.berni@unimore.it
Thank you in advance for the help

s_shimz January 21, 2015 23:48
Could someone provide me with the following paper?
"Elements of Industrial Heat Transfer Prediction" 16th Brazilian Congress of Mechanical Engineering (COBEM), Nov. 2001

e-mail is: shimizu.shunsuke@gmail.com

Thanks in advance.

romant January 22, 2015 03:48
Final version Online
 
Dear all,

the paper together with many other papers can be found in the following link. The paper belongs to the invited lectures

http://www.abcm.org.br/pt/wp-content.../2001/home.htm

Please post further comments when the link becomes dead.


All times are GMT -4. The time now is 01:18.
Page 2 of 3 1 2 3
Show 40 post(s) from this thread on one page