[H--]

If I understand it correctly, The smallest (microscale) eddies in a turbulent flow field are supposed to get their energy from the medium and or large scale eddies through the energy cascade notion of turbulent flow.

If no large or medium scale eddies exist in the flow (as evident from simulation results, assuming the mesh cell sizes are small enough to capture both the large and medium scales) - does that not indicate the lack of existence of microscale eddies in the flow?

otherwise, where do they get their energy (to even exist) from?

Any light shed on this topic is greatly appreciated.

Thanks.

[FMDenaro]

Originally Posted by H--

If I understand it correctly, The smallest (microscale) eddies in a turbulent flow field are supposed to get their energy from the medium and or large scale eddies through the energy cascade notion of turbulent flow.

If no large or medium scale eddies exist in the flow (as evident from simulation results, assuming the mesh cell sizes are small enough to capture both the large and medium scales) - does that not indicate the lack of existence of microscale eddies in the flow?

otherwise, where do they get their energy (to even exist) from?

Any light shed on this topic is greatly appreciated.

Thanks.

If there are no large scales that creates the intertial energy transfer, the flow is simply laminar, the viscous term strongly acting at low wavenumbers components destroyng the energy cascade.

[H--]

Originally Posted by FMDenaro

If there are no large scales that creates the intertial energy transfer, the flow is simply laminar, the viscous term strongly acting at low wavenumbers components destroyng the energy cascade.

so, just to verify..., lack of large scale eddies in the flow field indicates a completely laminar flow?

Thank you for your reply!

[FMDenaro]

Originally Posted by H--

so, just to verify..., lack of large scale eddies in the flow field indicates a completely laminar flow?

Thank you for your reply!

you can have large structures in laminar flow, immagine for example the backward facing step flow (Re<400) or the flow around a cylinder at low Re. The key is that they cannot transfer energy to smallest structures in an intertial way as the dissipation due to the viscosity acts at large length.

[H--]

Originally Posted by FMDenaro

you can have large structures in laminar flow, immagine for example the backward facing step flow (Re<400) or the flow around a cylinder at low Re. The key is that they cannot transfer energy to smallest structures in an intertial way as the dissipation due to the viscosity acts at large length.

I see. In that case I want to further ask -

If a simulation conducted with a space and time resolution sufficient to capture large and medium scale eddies yields results that don't indicate any turbulent or transitional behavior (time-resolved curves of velocity and WSS at various points in the domain, for example, do not contain any high frequency fluctuations and resemble the inlet waveform of the problem) -
I could conclude that further refinement of the simulation's resolution (down to the Kolmogorov scale) definitely won't yield results indicating turbulent phenomena?

Thanks again!

[FMDenaro]

Actually, before doing a CFD simulation you should be aware of the main physical features of the flow problem...Generally, the Reynolds number is the first indicator. Setting a grid size able to produce everywhere cell Re number O(1) is a guarantee to solve all the scales of the motion (DNS).

However, assuming in your case all the setup is correct, if you reach a steady solution it is for sure not turbulence. If you get an unsteady solution having one or very few and low wavenumber components it is laminar, too.

[H--]

Originally Posted by FMDenaro

Actually, before doing a CFD simulation you should be aware of the main physical features of the flow problem...Generally, the Reynolds number is the first indicator. Setting a grid size able to produce everywhere cell Re number O(1) is a guarantee to solve all the scales of the motion (DNS).

However, assuming in your case all the setup is correct, if you reach a steady solution it is for sure not turbulence. If you get an unsteady solution having one or very few and low wavenumber components it is laminar, too.

I'm actually trying to justify not utilizing DNS by showing that the results that I have obtained using a coarse (relative to Kolmog. scale) mesh and time steps are sufficient to indicate the lack of turbulence without having to scale down to the Kolmogorov scale.

My problem has a pulsatile inlet, so by definition the solution isn't steady. Did you mean something else by 'steady solution'?

Plus, can one conclude the lack of turbulence without evaluating the wavenumber, but only on the basis of temporal curves of the physical parameters (e.g. WSS and Velocity)?

Thanks for your reply and your patience

[FMDenaro]

Originally Posted by H--

I'm actually trying to justify not utilizing DNS by showing that the results that I have obtained using a coarse (relative to Kolmog. scale) mesh and time steps are sufficient to indicate the lack of turbulence without having to scale down to the Kolmogorov scale.

My problem has a pulsatile inlet, so by definition the solution isn't steady. Did you mean something else by 'steady solution'?

Plus, can one conclude the lack of turbulence without evaluating the wavenumber, but only on the basis of temporal curves of the physical parameters (e.g. WSS and Velocity)?

Thanks for your reply and your patience

I do not understand your sentence "I'm actually trying to justify not utilizing DNS by showing that the results that I have obtained using a coarse (relative to Kolmog. scale) mesh and time steps".

If you already know that your grid size cut away the viscous part (and maybe also some part of the inertial?), this is already an indication you have a turbulent spectrum and you are performing a sort of LES without model. Refining the grid you will capture the part of the spectrum that can be relevant in your solution.

You need to do a frequency analysis to see the extension of your energy spectra.

[H--]

Originally Posted by FMDenaro

I do not understand your sentence "I'm actually trying to justify not utilizing DNS by showing that the results that I have obtained using a coarse (relative to Kolmog. scale) mesh and time steps".

If you already know that your grid size cut away the viscous part (and maybe also some part of the inertial?), this is already an indication you have a turbulent spectrum and you are performing a sort of LES without model. Refining the grid you will capture the part of the spectrum that can be relevant in your solution.

You need to do a frequency analysis to see the extension of your energy spectra.

I'll try to phrase it better:

1. My model involves a pulsatile flow (transient) in a pipe that has a complex geometry (it is slightly curved, has a few fenestrations, etc.). The Reynolds number is ~1400.

2. I want to show that if I use a certain mesh cell size larger than the Kolmogorov length scale (in my case: cell_size =~ 5*Kolmog_length_scale)
but smaller than the Taylor scale (Taylor_length_Scale =~ 4*cell_size),
I can conclude that the flow field is laminar
as long as the temporal curves of wall shear stress vs. time and velocity vs. time
are 'smooth' and don't have high frequency fluctuations ('noise') riding on the low frequency waveform
and are thus behaving like the velocity inlet waveform to my domain, which is 'smooth' and has a low frequency.

3. If this is indeed true, then there is no need to use DNS to find the smallest scale eddies in the flow field, because they would not exist.

Are 2. and 3. indeed true?



Thank you again for your help, I really appreciate it

[FMDenaro]

Again, I do not understand... if you have an estimation of the Taylor micro-scale and Kolmogorov scale, that means you have estimation of turbulence characteristic scales therefore you contradicts that fact that the flow can be laminar...
in a laminar flow you have only the large integral scale, you can have some structures but no clear distinction between inertial and viscous ranges such to distinguish the viscous scales. In practice, the effect of the viscous term will dump any inertial range so that defining the Taylor micro-scale (the largest of the viscous one) makes no sense.
For your Re number, being the geometry complex, I am quite sure you cannot have a laminar flow everywhere.

Conversely, I can immagine more correct the question if the further resolving of the Kolmogorov scale would add something in the physics of the numerical solution.

[H--]

Originally Posted by FMDenaro

Again, I do not understand... if you have an estimation of the Taylor micro-scale and Kolmogorov scale, that means you have estimation of turbulence characteristic scales therefore you contradicts that fact that the flow can be laminar...
in a laminar flow you have only the large integral scale, you can have some structures but no clear distinction between inertial and viscous ranges such to distinguish the viscous scales. In practice, the effect of the viscous term will dump any inertial range so that defining the Taylor micro-scale (the largest of the viscous one) makes no sense.
For your Re number, being the geometry complex, I am quite sure you cannot have a laminar flow everywhere.

Conversely, I can immagine more correct the question if the further resolving of the Kolmogorov scale would add something in the physics of the numerical solution.

Thank you for all of your responses, they were really helpful. I am trying to put everything together to get a better understanding of the matter.