[mich_K]

Is there a way to obtain the skin friction coefficient over a surface, or am I going to have to calculate it manually?

Thanks in advance

[tomf]

If you use the wallShearStress utility you have the wall shear stress (tauw). You can than use ParaView to calculate the skin friction coefficient (2*tauw/(rho*V^2))

Regards,
Tom

[mich_K]

Thanks Tom.

I have only used openFOAM for a few days now and cannot find any documentation on the actual implementation of the Standard Utilities.

Can you perhaps explain its use to me?

In what file and how do I implement the wallShearStress utility?

Also, how do I do such types of calculations in paraView?

Thanks in advance.

[tomf]

Hi, just type "wallShearStress" in the terminal and it will calculate this after your normal run. It is a standard postProcessing OpenFOAM utility. In ParaView you can load the wall patches (not the internalField) and use the calculator filter on it.

I bet you can find a lot of information on both of these by searching the forum.
Regards,
Tom

[mich_K]

Thank you very much Tom.

I did indeed try and search for something, but didn't find anything.

I think in the beginning there are a few things that are so basic that no one really talks about it, such as this.

Thanks again, cheers.

[huangxianbei]

Originally Posted by tomf

If you use the wallShearStress utility you have the wall shear stress (tauw). You can than use ParaView to calculate the skin friction coefficient (2*tauw/(rho*V^2))

Regards,
Tom

Hi,Tom:
Can we calculate the wallShearStress using tau_wall=rho*nu*gradU? Two different values are obtained in the two ways, which one would be correct? Should we just use wallShearStress utility?


Xianbei

[tomf]

Hi Xianbei,

I think it depends a lot on the type of system that you are analyzing. The wallShearStress utility also takes turbulence into account. I guess that the results should be the same if you have a laminar case. Also if you have an incompressible case I think the wallShearStress is actually the wallShearStress divided by rho!

In version 2.3.x you can actually also use the utility as a function object (at least the source code indicates this)

Regards,
Tom

[huangxianbei]

Originally Posted by tomf

Hi Xianbei,

I think it depends a lot on the type of system that you are analyzing. The wallShearStress utility also takes turbulence into account. I guess that the results should be the same if you have a laminar case. Also if you have an incompressible case I think the wallShearStress is actually the wallShearStress divided by rho!

In version 2.3.x you can actually also use the utility as a function object (at least the source code indicates this)

Regards,
Tom

Hi,Tom:
Thank you for your reply. In fact, my case is not laminar, but I don't think wallShearStress is just written for laminar only. I agree with you that the wallShearStress is actually the wallShearStress divided by rho, however, the results are not the same (My case is incompressible)

Does it related to the Beff in LES, in my opinion, i_normal&Beff seems not equal to wallShearStress, what's your opinion?

Xianbei

[tomf]

Hi Xianbei,

Maybe I needed to clarify my statement, wallShearStress is not only for laminar indeed, but also for turbulent flow (using nuEff instead of nu), but if you calculate wall shear stress by nu*rho*gradU than there is a difference in the values between both approaches. You would have to use nuEff in the second approach as well.

I am not sure how things would change for LES, but still I would expect a difference since even with LES you will have a shear stress from the unresolved eddies, depending on LES model, grid refinement, wall model etcetera. I have not studied LES simulations enough to give you any guidance. Best way would probably be to take a simple flat plate flow and look at the results from different methods.

Good luck,
Tom

[huangxianbei]

Originally Posted by tomf

Hi Xianbei,

Maybe I needed to clarify my statement, wallShearStress is not only for laminar indeed, but also for turbulent flow (using nuEff instead of nu), but if you calculate wall shear stress by nu*rho*gradU than there is a difference in the values between both approaches. You would have to use nuEff in the second approach as well.

I am not sure how things would change for LES, but still I would expect a difference since even with LES you will have a shear stress from the unresolved eddies, depending on LES model, grid refinement, wall model etcetera. I have not studied LES simulations enough to give you any guidance. Best way would probably be to take a simple flat plate flow and look at the results from different methods.

Good luck,
Tom

Hi,Tom:
Thank you for your advice. In fact, in LES, nuEff=nu+nuSgs, however, nuSgs on the wall is 0, so when we calculate wallShearStress, we can just ignore nuSgs. Additionally, the grid near the wall is refined enough with average y+=3, so it should be enough. I performed another case, a straight duct, however, the results still differs between 2 methods

Xianbei

[tomf]

Hi Xianbei,

I guess the next step would be to compare with the pressure loss for your duct flow, which should be related to the shear stress, at least you would know if one of both methods is correct.

Regards,
Tom

[huangxianbei]

Originally Posted by tomf

Hi Xianbei,

I guess the next step would be to compare with the pressure loss for your duct flow, which should be related to the shear stress, at least you would know if one of both methods is correct.

Regards,
Tom

Hi,Tom:
Sorry for forgotting metioning that I compare both results with the experimental data, the rho*nu*gradU is at the same order with the experiment's, however, the wallShearStress returns a much more larger value, about 70 times. That's why I doubt about the wallShearStressLES.

Xianbei

[tomf]

Hi Xianbei,

Ok I did not realize the difference was that big! In the source code of the utility (not the functionObject) it is explicitly stated that a RAS type turbulence model is expected however, maybe something goes wrong because of that?

Other than that I would not know what might be causing this error, I guess the best thing may be to file a bug report if you can make a simple (small enough) test case.

Regards,
Tom

[huangxianbei]

Originally Posted by tomf

Hi Xianbei,

Ok I did not realize the difference was that big! In the source code of the utility (not the functionObject) it is explicitly stated that a RAS type turbulence model is expected however, maybe something goes wrong because of that?

Other than that I would not know what might be causing this error, I guess the best thing may be to file a bug report if you can make a simple (small enough) test case.

Regards,
Tom

Hi, Tom:
Thank you for your patient explaination. Although it's not solved yet, I learn more from the communication. Thank you again

Xianbei

[beatlejuice]

Can we calculate the wallShearStress using tau_wall=rho*nu*gradU? Two different values are obtained in the two ways, which one would be correct? Should we just use wallShearStress utility?

Xianbei

Hello,

seems like the wallShearStress utility uses tau_wall = nu * gradU since the dimension is m²/s² (see dimesions [0 2 -2 0 0 0 0]; )

So the kinematic visc. is used instead of dynamic visc....