Learning Forum

[enricomarco.dimarzo]

Hello, 

I'm trying to mesh a ducted wind turbine with Fluent meshing using the watertight workflow. However, viewing the CFD results I got some pressure spots on the duct surface at the leading edge, even if I use the polyhedral surface mesh.

Is there a way to improve the local mesh on the leading edge ? 

So far, I used the curvature sizing for the blade and the duct, and the body of influence for some local refinements.

I also tried with Ansys meshing imposing the mapped surface mesh but the mesh is not always generated (blue circle) and no warning messages appear. 

 

Thanks in advance. 

Regards. 

[Nikhil Narale]

Hello, 

 

Can you post some more images, perhaps from different angles for easy understanding? 

Why do you think the mesh is not sufficiently refined? Are those pressure spots unphysical? 

 

'Is there a way to improve the local mesh on the leading edge ?'  -  By this do you mean refining the mesh or improving the mesh quality? 

 

Nikhil

[enricomarco.dimarzo]

Helllo, thank you for your reply. 

As you can see from the picture below (frontal view of the duct), the spots are on the leading edge of the duct. 

I'm not really concerned about the mesh refinement but more about the quality around the leading edge, which is a region characterised by high gradients. However, the pressure spots should not be there. 

I was wondering if there is the possibility of generating a mapped mesh in the watertight geometry workflow or if there is a way to have a better mesh in that area. As an example, here there is a picture of a mapped mesh obtained by ansys meshing. 

Unluckly, I'm not so experienced with this tool so that I'm not always able to get a mapped mesh and I lose control over the local refinements. 

[Swathi V. V.]

Hi, 

Is the solution converged? What about the flux imbalance? 

 

[enricomarco.dimarzo]

Hello, 

thank you for your answer. 

The solution is converging even if it's not fully converged. 

I monitored these spots during the iterations but they seem to be fixed in space and do not change during the iterations. 

Additional info: I'm using the pressure-based solver with the coupled algorithm and the second order accuracy schemes for all the equations. The CFL is 10. The turbulence model is the k-w.

Thank you for your time.

[Swathi V. V.]

What about the flux imbalance and mesh quality. 

[enricomarco.dimarzo]

Hello, I'm sorry I replied to you but not using the reply button. 

The answer to your question is in the post below. 

Thanks. 

[enricomarco.dimarzo]

Hello, thanks again for the reply.

The only flux imbalance that I can access is the mass one, but I do not see anything wrong with it. 

Here there are the contours of the mass imbalance in a plane which cuts the duct profile from leading edge to trailing edge (orthogonal to the previous pictures).

Additionally, here there is also the mesh orthogonal quality on the duct surface. 

Thanks in advance. 

Marco

 

[Rob]

Are you modelling a blade in your sector? If so, where is it?

[enricomarco.dimarzo]

Hello, thank you for your repply.

Yes, I'm modelling also the blade but I did not activate the surface plot in the previous pictures.

 

[Rob]

Where is the blade relative to the pressure rise on the duct? 

[enricomarco.dimarzo]

 

Hello, thank you for you reply. 

Here you have a 2D sketch of the problem:

To better understand in 3D, the duct goes from an angular position of -60° degree to 60° while the stacking line of the blade profiles (red line in previous figure) is placed at an angular position equal to 0° (It is the NREL Phase VI blade).

If I did not make it clear let me know.

 

[Rob]

So the blade lines up with the higher pressure region? 

[enricomarco.dimarzo]

Yes, it does.

[Rob]

Have you had a look to see what the rest of the flow field is doing?  Ie how would I think to ask where the blade was? 

[enricomarco.dimarzo]

 

Hello, thank you for yor reply. 

Yes, of course I had a look to the rest of the flow field: the stagnation point on the duct is where it should be, the blade is correctly rotating, the wake is shedding downstream the rotor and its diverging, as it should. The simulation is all fine, the integral parameters are within the typical range of the application. 

Therefore, my only problem is that, if I zoom in at the duct stagnation point, the pressure contours have this local irregolarities (first figure from the top) instead of a nice levels distribution (see for instance the third picture from the top).

 

[Rob]

Replot the pressure on the duct with node values off. Also plot a slice through the blade & duct showing velocity and separately pressure. Post these. 

[enricomarco.dimarzo]

Hello, 

here you have the pressure contours with the node values off:

and also the slice through the blade and the duct of pressure and velocity magnitude

Thanks in advance.

[Rob]

 

I suspect you’re picking up some tip effects and possibly the turbine speed isn’t quite right for the flow speed. Increase the rotation speed a bit and see what happens. 

 

[enricomarco.dimarzo]

Hello, thank you for your reply.

The problem that I highlighted is not case-dependent, since the variation of the turbine speed will change the aerodynamic load and duct stagnation point too.

Therefore, I will have the same local peaks at the stagnation point (more inwords or outwards depending on the rotational speed), which, I think, is more related to the local mesh.

My question still is:

In Fluent meshing, is there a way to improve the local mesh ? is it not possible to apply a mapped surface mesh as in Ansys meshing ?

Thanks in advance,

Marco

[Rob]

In Fluent Meshing you've got size functions, and local sizing, the latter allows bodies of influence. In Ansys Meshing (Workbench) you can set surface mesh types. 

[enricomarco.dimarzo]

Thank you for your reply. 

I applied the curvature functions on the duct surface setting the min/max element size and the reference angle.

I also tried with the Meshing tool (workbench), third figure from the top, but not always the mapped mesh is generated on the duct whereas a blue circle appears near the function, without showing any warning messages. Moreover, it takes almost the whole day to generate the mesh. 

Thanks in advance,

Marco

 

[Rob]

Mapped mesh can also be a regular tri-pave according to Ansys Meshing, you may be seeing that.