[ricod-stareng]

Hi All,

I am new to the forum, I have been using the site for information and found it very helpful so far!

I have been taking my first steps in CFD lately, running a numerical analysis for a real-world problem at work, as an experimental path parallel to the usual code-based and theoretical calculations we ordinarily do.

The problem we are facing is working out a solution to upgrade an 18 meter steel chimney (diameter 0.4 meter) that is vibrating in the wind.

To the point:

I have been getting results that look seem to make some sense but I know there is still a way to go.

Flow speeds are 2-27 m/s giving Re of 2.75e3-7.4e5, so I expect organized vortex shedding at most of the range.

I have run 2d CFD on Midas NFX, for the different speeds. My first model consists of triangular element meshes with element size graded properly towards the cylinder (circular), with element size of 0.01 m at the interface.

I ran a 2 equation K-e analysis with what I believe to be properly defined field parameters (eddie kinematic energy and eddie length scale calculated from turbulence intensity level of 0.03) and time steps (1/100 of expected frequency of shedding). Got nice results showing shedding with the expected frequency. For the wall boundary condition I chose no-slip. The weird thing about the results is that I get a biased oscillation of the lift force, oscillating around a non zero force, despite the entire model being perfectly symmetrical. I have no clue why this is happening.

Following some reading I realized that I should give particular attention to the size of the layers immediately adjacent to the wall, to capture the different flow layers in the area, and should use an SST model and a properly defined wall distance. Here is where things went wrong. No matter what size mesh i use and which wall distance, I get very nasty results. I tried different mesh sizes, layering the mesh in different ways, calculating the wall distance with the online calculator on this site. nothing.. Basically, what I get is a flow that seems to emerge from the back of the cylinder, without any effect on the front and sides (relative to flow direction) and without any oscillation. Also, the norms of velocity and look very erratic during the run (compared to smooth oscillation in k-e).

A few specific questions:

I have tried resolving the viscous sub layer with a mesh point inside the layer, should I use a no-slip wall in this case?

I can't work out how to see the y+ values on Midas NFX, does anyone know how to?

Can someone please explain the difference between wall distance and dimensionless wall distance?

I would love some input, suggestions or discussion on the matter!

I will be happy to share the parameters I used, and show the analysis results. I tried adding pics to my post but it seems the only way is to link to images hosted somewhere online, and I don't have where at the moment.

Thanks, Rico.

[FMDenaro]

Despite that fact that I think this flow problem is not suitable, I assume you are performing a 2D URANS simulation.
First of all, could you show the plot of the force vs time (and drag vs. time)?
Then, is your grid exactly specular (I mean obtained by reflecting the half domain)?
What about the reference data you are comparing?

[ricod-stareng]

Originally Posted by FMDenaro

Despite that fact that I think this flow problem is not suitable, I assume you are performing a 2D URANS simulation.
First of all, could you show the plot of the force vs time (and drag vs. time)?
Then, is your grid exactly specular (I mean obtained by reflecting the half domain)?
What about the reference data you are comparing?

Thanks for the quick reply!

Yes, as far as I understand 2-equation k-e and 2-equation k-w(SST) turbulence models are URANS. Why do you ask this?

The grid is not exactly specular, all the dimensions of the domain and all of the boundary conditions are symmetrical in relation to the flow direction.

I will try to work out how to attach an image and then I'll show the results.

What do you mean by reference data? I am not comparing to any test results at the moment, only trying to get some seemingly reasonable results as a starting point. Also, as a structural engineer I am familiar with the expected wind loads on such a structure and I can compare the results of the drag force for reference.

[FMDenaro]

Originally Posted by ricod-stareng

Thanks for the quick reply!

Yes, as far as I understand 2-equation k-e and 2-equation k-w(SST) turbulence models are URANS. Why do you ask this?

The grid is not exactly specular, all the dimensions of the domain and all of the boundary conditions are symmetrical in relation to the flow direction.

I will try to work out how to attach an image and then I'll show the results.

What do you mean by reference data? I am not comparing to any test results at the moment, only trying to get some seemingly reasonable results as a starting point. Also, as a structural engineer I am familiar with the expected wind loads on such a structure and I can compare the results of the drag force for reference.

Use an exactly specular grid, then post the results, you can attach images