[price_s]

I have an analysis of a transverse axis tidal turbine in a channel. The turbine is rotating within the channel. Therefore a rotating mesh is used. I ran an initial analysis that gave reasonable results. However I created a new mesh, as there were some problems with the old mesh, i.e gaps at mesh interface and also to extend the mesh higher. The new mesh is near identical to the original. I also added gravity effects. Apart from these two changes I believe the set-up is the same. However the new analysis does not seem to be accurately modelling the interface, as can be seen in the velocity plot, there is a clear line at the interface. I've attached images of the velocity and pressure distributions for both cases. Any ideas as to why the new analysis does not seem to work as well?

[vinerm]

Line depicting the interface is always there; it's only a matter of user's choice, whether user wants it to be displayed or not. It appears to be 2D simulation. For 2D cases, do no select any boundary while displaying Contours. Then, you won't see the line.

[price_s]

Hi Vinerm,

I'm talking about in picture 4, the velocity plot for the new analysis, the turbine does not seem to affect the flow above the interface. Compared to picture 1, the original analysis, where there is a clear effect, which is what I would expect. Not about a line indicating where the interface is.

[vinerm]

I apologize for misunderstanding the situation. At this scale, the difference is not very clear.

I would suggest doing a simple check. Increase the viscosity to a very high value or use a very high rotation velocity to check if there is an effect across the interface or not. If it is not there, then check the type of interface. Sometimes, recreating the interface helps, particularly when you go from Serial to Parallel or vice-versa.

[price_s]

Hi Vinerm, no worries about the misunderstanding. Thanks for the advice, the fluid is water at the moment, so viscosity fairly high already and a tip speed ratio of 3.5. Would you suggest increasing these further?

[price_s]

I may have changed the solver settings inadvertently as the cases are being solved as a batch job on nodes on a shared computing system. So I will check the settings for this and for the fluent session that the case was created in.

[vinerm]

For debugging, you need to increase the viscosity by 1000 times. That would be similar to assuming a homogeneous turbulence.

[price_s]

Hi Vinerm, thanks for all the help. I've been back and remade the case from scratch. ensuring it was made with the same solver settings as the batch job solver and recreating the interface. Currently running a case for normal viscosity and x1000. I'll see what the results are in a few hours. Thank you

[vavnoon]

Hello everybody,

Many thanks for starting this thread and valuable comments.

I also face the same problem. I have a wavy cross-section. I made a structured mesh in ICEM. Because of the special cross-section, an undesirable dense mesh was created in the middle. So after trying several options in ICEM setting for resolving this issue, I decided to apply the non-conformal mesh method. Now the mesh looks somewhat better.

I simulated the fluid flow using FLUENT and ran two times, first with viscosity= 0.0006965 and then with viscosity=1 (units in SI).

The contours have been attached. Labels one and two are related to cases one and two, respectively.
In both cases interfaces are evident. Is this matter trivial?

Thanks for your time