CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Community Contributions (https://www.cfd-online.com/Forums/openfoam-community-contributions/)
-   -   [swak4Foam] Difference between snGrad(U) using swakExpression and wallgradU utility ? (https://www.cfd-online.com/Forums/openfoam-community-contributions/146015-difference-between-sngrad-u-using-swakexpression-wallgradu-utility.html)

Dan1788 December 16, 2014 22:52
Difference between snGrad(U) using swakExpression and wallgradU utility ?
 
Hello,

I am doing an LES of pipe flow. The streamwise direction is the z-direction. In my controlDict I enter the following "swakExpression" to keep track of my wall shear stress:

Code:
wallShear
    {
        type    swakExpression;
        valueType  patch;
        patchName  Side;
        accumulations  (
            average
        );
        expression  "nu*1.16*snGrad(U)";
        verbose true;
    }
The output of the following in my "postProcessing" folder is as follows:

Code:
      Time             
      1.15            (0.00158395 -0.00195003 -0.309623)
      1.15001            (0.0015569 -0.0018457 -0.31036)
      1.15002            (0.00154682 -0.00179581 -0.310094)
      1.15002            (0.0015546 -0.00176741 -0.30988)
      1.15003            (0.0015752 -0.00177392 -0.309709)
      1.15003            (0.00154428 -0.00172741 -0.309567)
      1.15004            (0.00152869 -0.00170179 -0.309448)
      1.15004            (0.00154396 -0.00171494 -0.309346)
      1.15005            (0.00153824 -0.00171052 -0.309258)
      1.15005            (0.00153934 -0.00171405 -0.309179)
      1.15006            (0.00153327 -0.00170792 -0.30911)
      1.15006            (0.00152485 -0.00169658 -0.309047)
      1.15007            (0.00152928 -0.00169544 -0.308991)
      1.15007            (0.00153558 -0.00170495 -0.30894)
      1.15008            (0.00153235 -0.00170306 -0.308893)
      1.15008            (0.00153008 -0.00169775 -0.308849)
      1.15009            (0.00154298 -0.00171269 -0.30881)
      1.15009            (0.00154351 -0.00171235 -0.308773)
      1.1501            (0.00153823 -0.00170273 -0.308739)
      1.1501            (0.00153198 -0.00169867 -0.308707)
      1.15011            (0.00152493 -0.00169242 -0.308677)
      1.15011            (0.00153547 -0.00170078 -0.30865)
      1.15012            (0.00153531 -0.00170344 -0.308624)
      1.15012            (0.00152695 -0.00169185 -0.3086)
      1.15013            (0.00152394 -0.00168702 -0.308578)
      1.15013            (0.00152902 -0.00170066 -0.308557)
      1.15014            (0.00152878 -0.00170005 -0.308537)
You see that the tau_w(z-direction) is always in the range of ~ 0.3.

However when I compute the wall shear stress using the wallShearStressLES utility, the output is as follows:

Code:
(0.00922472 -0.00948885 -0.436616)
(0.0167652 -0.0139423 -0.497059)
(0.027834 -0.0217006 -0.573286)
(0.0405144 -0.0274527 -0.648546)
(0.0479752 -0.029261 -0.717811)
(0.0462766 -0.0244107 -0.7791)
(0.0379119 -0.0177215 -0.833495)
(0.0251241 -0.010028 -0.885577)
(0.0158647 -0.00529415 -0.91607)
(-0.00224175 0.000942265 -0.846752)
(-0.0139847 0.0031963 -0.695487)
(-0.0133073 0.00202354 -0.506113)
(0.00138907 -0.000239348 -0.352859)
(-0.00568237 0.000631173 -0.211232)
(-0.0198709 0.000730177 -0.169419)
(-0.00914825 -0.00054803 -0.258058)
(0.0156156 0.00114809 -0.448939)
(0.00519621 0.0010823 -0.597595)
(0.00899102 0.00132361 -0.671042)
(0.0455127 0.0108841 -0.716581)
(0.0519395 0.0163323 -0.705576)
(0.0428748 0.0163925 -0.644185)
(0.0271054 0.0124011 -0.564419)
(0.0127216 0.00706786 -0.551247)
(0.00512223 0.00361148 -0.617746)
(0.00279277 0.00222053 -0.690317)
(0.00107463 0.0039771 -0.730843)
(0.00670205 0.00761867 -0.778388)
(0.0236784 0.0209093 -0.886947)
(0.0405041 0.0391589 -1.10094)
(0.011279 -0.0109334 -0.438007)
(0.0185427 -0.0161606 -0.500274)
(0.0297334 -0.0231098 -0.57867)
(0.0411742 -0.0284098 -0.654419)
(0.0476158 -0.029062 -0.72111)
(0.0457186 -0.0245445 -0.77872)
(0.0385896 -0.018128 -0.830116)
(0.0269545 -0.0107781 -0.883763)
(0.0166884 -0.00548417 -0.914038)
(-0.00277345 0.00124665 -0.838704)
(-0.0161723 0.00388169 -0.685918)
(-0.0180005 0.00293102 -0.501037)
(3.30381e-05 -1.50492e-05 -0.35243)
(-0.00945636 0.00110753 -0.208022)
(-0.0277763 0.000992707 -0.172966)
(-0.0160046 -0.000732155 -0.269808)
(0.00941981 0.000497363 -0.453516)
(0.00323716 0.000629968 -0.600543)
(0.0046372 0.000533136 -0.683728)
Clearly the values of the z-component here are really far apart.

Which one is correct and what is the difference between both using snGrad(U) in the first approach and then using the wallShearStressLES utility ??

Please help :confused:

Dan1788 December 17, 2014 21:26
Can anyone help please?

gschaider December 18, 2014 04:40
Quote:
Originally Posted by Dan1788 (Post 524291)
Hello,

I am doing an LES of pipe flow. The streamwise direction is the z-direction. In my controlDict I enter the following "swakExpression" to keep track of my wall shear stress:

Code:
wallShear
    {
        type    swakExpression;
        valueType  patch;
        patchName  Side;
        accumulations  (
            average
        );
        expression  "nu*1.16*snGrad(U)";
        verbose true;
    }
The output of the following in my "postProcessing" folder is as follows:

Code:
      Time             
      1.15            (0.00158395 -0.00195003 -0.309623)
      1.15001            (0.0015569 -0.0018457 -0.31036)
      1.15002            (0.00154682 -0.00179581 -0.310094)
      1.15002            (0.0015546 -0.00176741 -0.30988)
      1.15003            (0.0015752 -0.00177392 -0.309709)
      1.15003            (0.00154428 -0.00172741 -0.309567)
      1.15004            (0.00152869 -0.00170179 -0.309448)
      1.15004            (0.00154396 -0.00171494 -0.309346)
      1.15005            (0.00153824 -0.00171052 -0.309258)
      1.15005            (0.00153934 -0.00171405 -0.309179)
      1.15006            (0.00153327 -0.00170792 -0.30911)
      1.15006            (0.00152485 -0.00169658 -0.309047)
      1.15007            (0.00152928 -0.00169544 -0.308991)
      1.15007            (0.00153558 -0.00170495 -0.30894)
      1.15008            (0.00153235 -0.00170306 -0.308893)
      1.15008            (0.00153008 -0.00169775 -0.308849)
      1.15009            (0.00154298 -0.00171269 -0.30881)
      1.15009            (0.00154351 -0.00171235 -0.308773)
      1.1501            (0.00153823 -0.00170273 -0.308739)
      1.1501            (0.00153198 -0.00169867 -0.308707)
      1.15011            (0.00152493 -0.00169242 -0.308677)
      1.15011            (0.00153547 -0.00170078 -0.30865)
      1.15012            (0.00153531 -0.00170344 -0.308624)
      1.15012            (0.00152695 -0.00169185 -0.3086)
      1.15013            (0.00152394 -0.00168702 -0.308578)
      1.15013            (0.00152902 -0.00170066 -0.308557)
      1.15014            (0.00152878 -0.00170005 -0.308537)
You see that the tau_w(z-direction) is always in the range of ~ 0.3.

However when I compute the wall shear stress using the wallShearStressLES utility, the output is as follows:

Code:
(0.00922472 -0.00948885 -0.436616)
(0.0167652 -0.0139423 -0.497059)
(0.027834 -0.0217006 -0.573286)
(0.0405144 -0.0274527 -0.648546)
(0.0479752 -0.029261 -0.717811)
(0.0462766 -0.0244107 -0.7791)
(0.0379119 -0.0177215 -0.833495)
(0.0251241 -0.010028 -0.885577)
(0.0158647 -0.00529415 -0.91607)
(-0.00224175 0.000942265 -0.846752)
(-0.0139847 0.0031963 -0.695487)
(-0.0133073 0.00202354 -0.506113)
(0.00138907 -0.000239348 -0.352859)
(-0.00568237 0.000631173 -0.211232)
(-0.0198709 0.000730177 -0.169419)
(-0.00914825 -0.00054803 -0.258058)
(0.0156156 0.00114809 -0.448939)
(0.00519621 0.0010823 -0.597595)
(0.00899102 0.00132361 -0.671042)
(0.0455127 0.0108841 -0.716581)
(0.0519395 0.0163323 -0.705576)
(0.0428748 0.0163925 -0.644185)
(0.0271054 0.0124011 -0.564419)
(0.0127216 0.00706786 -0.551247)
(0.00512223 0.00361148 -0.617746)
(0.00279277 0.00222053 -0.690317)
(0.00107463 0.0039771 -0.730843)
(0.00670205 0.00761867 -0.778388)
(0.0236784 0.0209093 -0.886947)
(0.0405041 0.0391589 -1.10094)
(0.011279 -0.0109334 -0.438007)
(0.0185427 -0.0161606 -0.500274)
(0.0297334 -0.0231098 -0.57867)
(0.0411742 -0.0284098 -0.654419)
(0.0476158 -0.029062 -0.72111)
(0.0457186 -0.0245445 -0.77872)
(0.0385896 -0.018128 -0.830116)
(0.0269545 -0.0107781 -0.883763)
(0.0166884 -0.00548417 -0.914038)
(-0.00277345 0.00124665 -0.838704)
(-0.0161723 0.00388169 -0.685918)
(-0.0180005 0.00293102 -0.501037)
(3.30381e-05 -1.50492e-05 -0.35243)
(-0.00945636 0.00110753 -0.208022)
(-0.0277763 0.000992707 -0.172966)
(-0.0160046 -0.000732155 -0.269808)
(0.00941981 0.000497363 -0.453516)
(0.00323716 0.000629968 -0.600543)
(0.0046372 0.000533136 -0.683728)
Clearly the values of the z-component here are really far apart.

Which one is correct and what is the difference between both using snGrad(U) in the first approach and then using the wallShearStressLES utility ??

Please help :confused:
Depends on your definition of "correct". I think the names are pretty self-explanatory: snGrad(U) in swak ONLY looks at the velocity field and calculates the gradient. wallShearStressLES (as the name says) adds the effect of the turbulence model. I'd only see a problem if wallGradU (which you mention in the title) differs significantly from swak

Dan1788 December 21, 2014 17:21
Test between nu*rho*snGrad(U) and nu*rho*wallGrad(U)
 
Hi Bernhard,

Thanks for your reply.

So I did a small test to compare results of the wall Shear stress using snGrad(U) and using wallGrad(U) utility.

Following are a couple of lines as output from wallGrad(U) -latestTime

Code:
(158.301 -63.0679 9752.01)
(-18.023 -183.104 9801.31)
(70.9486 -45.88 9654.49)
(119.605 -84.467 9432.72)
(138.58 -85.0173 9607.43)
(138.097 -74.9961 10680.3)
(240.245 -116.391 12213)
(439.328 -183.63 12451.1)
(545.342 -192.911 11860.8)
(507.157 -148.288 10896.3)
(177.236 -40.1328 9856.27)
(-232.057 45.4857 10258.2)
If I look at the 3rd collumn and multiply that by nu*rho (1.53e-05*1.16), the values I get are really off from the output of nu*rho*snGrad(U). The nu*rho*snGrad(U) from the swakExpresssion looks like this:

Code:
Time             
      1.15            (0.00158395 -0.00195003 -0.309623)
      1.15001            (0.0015569 -0.0018457 -0.31036)
      1.15002            (0.00154682 -0.00179581 -0.310094)
      1.15002            (0.0015546 -0.00176741 -0.30988)
      1.15003            (0.0015752 -0.00177392 -0.309709)
      1.15003            (0.00154428 -0.00172741 -0.309567)
      1.15004            (0.00152869 -0.00170179 -0.309448)
      1.15004            (0.00154396 -0.00171494 -0.309346)
      1.15005            (0.00153824 -0.00171052 -0.309258)
Clearly there is a huge difference between the output of wallGrad(U) and snGrad(U) and I dont know why :confused: Please help me understand this. Thanks :)

gschaider January 7, 2015 15:50
Quote:
Originally Posted by Dan1788 (Post 524874)
Hi Bernhard,

Thanks for your reply.

So I did a small test to compare results of the wall Shear stress using snGrad(U) and using wallGrad(U) utility.

Following are a couple of lines as output from wallGrad(U) -latestTime

Code:
(158.301 -63.0679 9752.01)
(-18.023 -183.104 9801.31)
(70.9486 -45.88 9654.49)
(119.605 -84.467 9432.72)
(138.58 -85.0173 9607.43)
(138.097 -74.9961 10680.3)
(240.245 -116.391 12213)
(439.328 -183.63 12451.1)
(545.342 -192.911 11860.8)
(507.157 -148.288 10896.3)
(177.236 -40.1328 9856.27)
(-232.057 45.4857 10258.2)
If I look at the 3rd collumn and multiply that by nu*rho (1.53e-05*1.16), the values I get are really off from the output of nu*rho*snGrad(U). The nu*rho*snGrad(U) from the swakExpresssion looks like this:

Code:
Time             
      1.15            (0.00158395 -0.00195003 -0.309623)
      1.15001            (0.0015569 -0.0018457 -0.31036)
      1.15002            (0.00154682 -0.00179581 -0.310094)
      1.15002            (0.0015546 -0.00176741 -0.30988)
      1.15003            (0.0015752 -0.00177392 -0.309709)
      1.15003            (0.00154428 -0.00172741 -0.309567)
      1.15004            (0.00152869 -0.00170179 -0.309448)
      1.15004            (0.00154396 -0.00171494 -0.309346)
      1.15005            (0.00153824 -0.00171052 -0.309258)
Clearly there is a huge difference between the output of wallGrad(U) and snGrad(U) and I dont know why :confused: Please help me understand this. Thanks :)
And it doesn't confuse you that wallGrad(U) produces widly varying results? ;)

You are comparing an averaged value at different times (swak) with the raw data at one timestep (wallGrad(U), snGrad(U)) Of course they differ. But if you average the wallShearStressLES-output you should get similar values to swak

BTW: if your patch has faces of different sizhes you'll probably want to use weightedAverage as the accumulation


All times are GMT -4. The time now is 23:36.