open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello everyone,
I'm currently trying to simulate a flow through a Tidal turbine in a Basin.
As I'm using a simple Dragforce Model, the Mean speed is correctly simulated and fits to the Mesure-datas I already have.
But when I consider the turbulence Intensity, I have huge differences:



As you can see the turbulence intensity directly below the turbine is very low.

Is there a way to simulate a higher intensity below the turbine ?

I added the .cas_file and the Dragforce routine

Best regards,

Maxime Jeannin

Hello,

I do not fully understand how you compute the turbulence intensity because you have a turbulence model with constant 1.D-6. Anyway we are also working on this topic, but using the k-epsilon model, which should be the first thing to try in your case. We find that the velocities fairly (10%) fit the measurements, except within a few diameters in the wake of the turbines. This is due to the fact that we do not add in the model the turbulence due to the action of the turbine. What we plan to do now is to get an idea of the turbulence added by the turbines (literature ?) and then add this knowledge in the k-epsilon model. It will be done by changing the production terms in k and epsilon equations (but maybe not in the same proportions, this is all a theory to find).

With best regards,

Jean-Michel Hervouet

DearJean-Michel,

actually, that was the last case of a series of simulation I ran to see the effect of viscosity on turbulence intensity.

I already tried with the k-epsilon model but the turbulence intensity seems to be even lower than with Constant Viscosity model.

I calculate the turbulence intensity in post-processing by calculating mean-speed and instantaneous speed and using the formula of turbulence intensity.

I imagine it would be better to compute it in the processing but I don't really know how to do it (maybe using the K variable and defining Turbulence Intensity relation in User variables?)

best regards,

Maxime

Hello,

In your case the turbulence obtained with the k-epsilon model certainly depends on the boundary conditions at the entrance, by default 0 turbulence, and it probably has no time to set up. Anyway what happens in the wake of the turbine will highly depend on what we'll add in the turbine. This is our state of the art, and we think now to play on the turbulence production in the k-epsilon model. So far it is computed as a function of the velocity gradients, now it should be given in the turbine, as a data (provided by experiments on real turbines ?).

With best regards,

Jean-Michel Hervouet

Ok I'll try to find how to change the boundary conditions at the entrance. That should help me, thank you very much!
Good luck to you for your further work on the subject.

best regards,

Maxime

Last quick question Jean michel,

When you say you work on the topic, do you work in 3D or in 2D ?
And what model do you use for the turbine also drag-force ?

Thank you and best regards

Maxime

Hello,

We did both 2D and 3D. In 3D you need to add local head losses where the turbine is, instead of a 2D drag force. I did not do it myself but basically the head losses must represent the energy taken away by the turbine.

If you send me your E-mail I can redirect you to my colleagues who did the tests.

Regards,

JMH

Hello,
there have been new evolutions in my work.
I have been running simulations with k-epsilon model and new boundary conditions in entrance, as Jean-michel told me to try.

And it turns out that I get quite satisfying results (Sorry the colorbar don't match):






Problem: It seems that I have used a wrong formula to compute the Turbulence Intensity.

In fact I read that K=1/2*(u'²+v'²) so I thought that I could use: TI=sqrt(K/(U²+V²)) with K->Turbulent energy from the telemac variables

But 1rst problem:
If I use the turbulent energy from telemac variables I get the results showed by the image. But if I compute the turbulence intensity afterwards, using only the speed variations(u' and v') and Mean speeds (U and V):
(TI=sqrt(u'²+v'²/2(U²+V²)) , I get results that are far below that.

2nd problem: I checked in the Telemac manual and saw that Turbulence energy is given in J/kg this means that I should multiply Turbulence intensity by density before calculating the Turbulence intensity (TI=sqrt(k*density/(U²+V²))). But then my results with the two formulas would be even more different where the should be the same.

Does anyone know where the problem could be? Have I been using wrong formulas?

Best regards,

Maxime JEANNIN

Hi,

I am also trying to figure out how to compute the turbulence intensity from the turbulent energy k.

First, as the SI units of the Joule are [kg.m**2/s**2], the SI units of k given from telemac in J/kg are in fact [m**2/s**2]. You shouldn't multiply k by the density.

But I found two different forms for the turbulence intensity equation:

r0 = sqrt(k)/U
r0 = sqrt(2/3.k)/U

The first one seems to be rather classic and the latter is used by CFD software (ANSYS...).

Which form should we use for free surface flows? Is there a difference between 2D and 3D?

Thank you in advance for your help!

Best regards
PL

Hello,

since I struggled and still struggle with the different approaches found in the literature... I would say r0 = sqrt(2/3.k)/U is more correct since you can deduce it if you compute the turbulence intensity from the fluctuations in the three spatial directions.
However be aware that the turbulence intensity can be somewhat an ambiguous quantity and you have to be really careful if you compare measured values from e.g. papers with computed values. Different authors employ different concepts for describing the turbulence intensity. So you first equation has also a meaning in that sense. Turbulence intensity is only a measure to compare different turbulence levels in different flows.
The mean velocity U can be the local velocity magnitude in a point or the depth-averaged bulk velocity or only the streamwise velocity or the turbulence intensity is calculated in all three spatial directions separately which is also the original definition of the turbulence intensities if we don't assume turbulence isotropy. And often instead of a velocity the turbulence intensity is scaled with the friction velocity which has a good physical meaning.
2D and 3D... again I think it depends on how the values coming from the measurements have been computed.

Best regards,
Clemens