Horizontally homogeneous/constant in empty computational domain

Using SimScale Fluid Flow / CFD
1

According to Blocken (https://pure.tue.nl/ws/files/7993534/687442152079717.pdf, section 5.7) one needs to check if one’s empty computational domain (but using the correct roughnes length [inlet) and roughness height [for ground wall]) is horizontally homogeneous/constant.
I don’t know if it has to be homogeneous/constant in speed and turbulant kinetic energy, but I assume minimally for the speed. Correct?

So I did some experiments in SIMSCALE using an empty EFV,; using an inlet logarithmic ABL of z0=0.03m with:

  1. a ground wall with wall roughness of ks=0.33m (=10.9*z0) and Cs=0.9:
    https://www.simscale.com/workbench/?pid=5399799295123716855&mi=run%3A220%2Csimulation%3A134&mt=SIMULATION_RUN
    or
  2. a ground wall without wall roughness:
    https://www.simscale.com/workbench/?pid=5399799295123716855&mi=run%3A222%2Csimulation%3A134&mt=SIMULATION_RUN

One would assume that the wall with explicit roughness height (equivalent wih z0) would have the most constant speed over the 165m, but infact the wall without roughness is slightly more constant (see the related Solution Fields and the deviation of the z=9m line).

In both cases the plot of the turbulent kinetic energy is certainly not constant.

So my questions are:

  • do both speed and turbulent kinetic energy need to be constant over the whole length?
  • or did I make somewhere an error in defining my EFV and/or its parameters?
  • or do I define ‘constant’ too strict?
  • why is turbulent kinetic energy so depending on distance (in both a wall with roughness height and/or without it)?

I hope someone can help me.

Thanks.

All the best,

Victor

2

Is there someone who has some idea on this? I understand it is one of the important test to verify that the CFD simulation can be ok, so I hope someone can help me.
Thanks.

All the best,

Victor

3

Hi Victor,

I am Renan, from the SimScale support team. Thank you for reaching out to us!

Those are great questions! I will take a look at your model and get back to you as soon as possible.

Best,
Renan

4

Hi @nl_victor_reij ,

here some comments concerning your questions:

  1. They do not need to be constant per say. Nonetheless, it depends how the "constancy"is defined. Since the turbulent kinetic energy depends on the divergent of the velocity vector at a certain coodinate point, in order for the tubulent kinectic energy to be constant wrt to the velocity, the velocity vector needs to vary linearly. Therefore, the relation constant velocity component = constant turbulent kinect energy is not direct, since it would also depend on the other velocity components;
  2. Your EFV seems to be set-up correctly;
  3. The "constant"might be misleading, because it is not a direct relation. Meaning, just because the “horizontal” component of the velocity is constant, does not mean that the corresponding kinetic energy will constant. If anything, the part of the kinetic energy that depends on this constant horizontal component of velocity will be zero, since the derivative of a constant is zero;
    4)The K-Epsilon method (realized) is not advised for investigations close to the wall. Nonetheless, the change of with distance comes from the dependency of the divergent of the velocity field. Therefore, if the velocity changes with the distance, most likelly the turbulent kinetic energy will change.

Here is a piece do documentation that might help you with further simulations.

I hope these comments help you further. If you have anyfurther questions, just let us know.

Best,
Renan

5

Thanks Renan, your feedback is much appreciated. I will digest your answer. I am glad to hear that my EFV does not seem to be incorrect;-)

6

I found out that the number if iterations is quite important for horizontally homegenous investigations of k. U achieves faster the final version. Of ourse URF en initial condition settings are also important for the convergance of k.