[chripasch]

Hej everybody

I am wondering about the following thing: In my simulation of a desilting basin, I used a constant dynamic viscosity as input in the "fluids" tab (see attached picture "input"). The value of 0.00152 Pa-s correlates to water of a temperature of 5°C. All units in SI.

Once starting the simulation and checking the results, I obtain values orders of magnitude higher, in the range of 100 Pa-s (see attached picture "output").

What is wrong ovr there? My dynamic viscosity should be constant and only dependent on the water temperature (which is not changed during the simulation). Or is the output "dynamic viscosity" in FLOW-3D kind of a numerical, "fictitious" value with a non-distinct denotation?

Help would be great, because my results of the following sediement camculation are dependet also on the dynamic viscosity.

Thanks a lot

[JBurnham]

The output viscosity = molecular + turbulent. Turbulent viscosity is fictitious: it's how the 2-equation RANS turbulence models (like k-e RNG) represent mixing eddies on scales smaller than a cell. The viscosity is the extra drag that is not resolved by the mesh.

In general, the output viscosity will be 1E3 to 1E4 higher than the molecular viscosity. If it is greater than that, you are either seeing extreme turbulence (like in a Type III dentated stilling basin) or you should specify a constant global 'maximum turbulent mixing length' TLEN instead of letting FLOW-3D pick it dynamically.

[chripasch]

Thanks for the explanation, Jeff.
I now know which term is meant by the expression "dynamic viscosity", it is the turbulent eddy viscosity.

From my point of view, the denotation "dynamic viscosity" within the software is a bit irritating, because (we) hydraulic engineeers call dynamic viscosity the product of the kinetic viscosity (of water) and the density (of water).

For example, which viscosity is meant in the manual of v11, page. 279, eq. 7.237: The dynamic viscosity as you described it, or as I understand/described it? : )

You see, where the problem is!?

Thanks and best regards
Chris

[JBurnham]

7.237 is the equation for the drag of a single grain of sediment. The viscosity in that equation, MU_f, is the pure liquid dynamic viscosity, aka molecular viscosity, defined on the Fluids tab as Viscosity MU1. The -output- viscosity on the Analyze tab is molecular + turbulent eddy. The outputs are not the inputs.

In general, when you see the subscript f on MU or RHO in the User Manual it means the pure fluid density or viscosity.

[chripasch]

Yes, I agree on that and I got it : )
But do you also agree with me, that the name "dynamic viscosity" is not correct in the analyze tab, but should be called "eddy viscosity"?
Because the molecular viscosity is also called dynamic viscosity in the fluid tab.

Nevertheless, thanks for answering the question. It helped me a lot!

[JBurnham]

I understand your point, but I disagree. The output variable is molecular + eddy, not just eddy. Furthermore, if you're mixing two fluids, then the output is the mixture viscosity, and only includes eddy viscosity if turbulence models are on. Still further, if you're using a non-Newtonian viscosity model, then the output is the viscosity due to strain rate and temperature, and only includes eddy viscosity if a turbulence model is active. "Dynamic viscosity" includes all of these possibilities, and differentiates the output from kinematic viscosity, which includes density. Just calling it turbulent eddy viscosity would ignore the molecular, non-Newtonian, and mixture effects.

[chripasch]

Yes, that is true.
Maybe I was too much into my problem and discarded all the other cases.
I now totally agree with you.
The most important thing is to be now sure which output is given by the software.

Thanks a lot for that gainful and informative discussion.
Chris

[chripasch]

Dear Jeff and other experts,
we discussed about this topic quite along time ago. Now, a linked question arose: In version 11.1, the particle settling is calculated according to the approach of Soulsby (1997) as shown in Eq. (10.241) in the manual.
To calculate the settling velocity, the dimensionless diamter d* is used, determined according to Eq. (10.235). In that equation, the symbol for the molecular viscosity has an "f" as index, indicating that it is "the pure fluid [...] viscosity", explained by Jeff in one of the last replies.

I am now wondering if it really is the "original" viscosity, or if it rather is the modified viscosity according to Eq. (10.287), taking the effect of turbulence by means of an additional eddy viscosity into account.

If the latter is not the case, I am wondering how the effect of turbulence onto the particle settling velocity is considered in FLOW-3D.

Could you give me a hint or a short explanation on this?
Thanks a lot in advance and best regards

[bfox48]

The settling the equation in FLOW3D uses the true kinematic viscosity of the fluid and does not account for eddy viscosity. Otherwise, the effect of settling can be affected by the turbulent diffusion multiplier by increasing it from zero to 1. Other than that, turbulence will effect velocities, so it can effect settling if there are vertical velocities.

Keep in mind the richardson-zaki correlation also effects settling velocities. It reduces the settling velocity for increases sediment concentrations.

[chripasch]

Okay, thanks for the answer. Could a reason for not using the modified viscosity (based on the influence of turbulence/ eddies) be that it is not clear in research how exactly turbulence changes the terminal settling velocity of particles?
In this case, I could "misuse" the Richardson-Zaki model to change the settling velocity of particles according to my desires.

Best regards

[bfox48]

You are correct. The soulsby equation we use for settling is derived for a the pure kinematic viscosity. Using the modified viscosity is an interesting idea to account for turbulence effects, but I am not sure what the accuracy of that would be.

Keep in mind with the R-Z, the settling velocity is only affected by changes in concentration. So while you can "misuse" this for your purposes, it might very hard to get it to do what you want.

[chripasch]

Yes, I totally agree. Changing the R-Z number might become a very iterative and time consuming process. Thanks a lot again for your feedback on that topic.

Best regards