Learning Forum

[Royale Ohoka]

Hi guys,

I'm struggling to improve the Minimum Orthogonal Quality of my C-Mesh to analyse a 2D Supercritical Aerofoil (RAE 2822) in ANSYS Fluent for speeds of Mach 0.6-0.8. I made the mesh in the mesh tool of the 'Fluid Flow (Fluent)' toolbox of the software. 

I am certain the mesh I created is fine enough (especially around the boundary layer of the Aerofoil) with 450,000 cells. Image attached below. I have also included the Details tab in case there is something I have missed.

I would really appreciate the help on this one. I'm a final year Aero Eng student from Ireland.

Thanks in advance!

 

[Rob]

Have a look in the Courses on here (Learning, book icon upper left) and https://ansyshelp.ansys.com/account/secured?returnurl=/Views/Secured/prod_page.html?pn=Meshing&pid=Meshing&lang=en  (you'll need to open Help from the software and then paste the link into the same browser). 

[Royale Ohoka]

Hi Rob, none of the courses seem to have a solution to this or tutorials on Youtube because most seem to use ICEM CFD for Meshing which is no longer offered on ANSYS 2022 R1. 

I have tried researching different methods for mesh generation but it seems that the way I have done it up until now (C-Mesh) is the most popular method. 

I have tried refining the mesh in mesh tool but still get warning in Setup tool that my minimum orthogonal quality is less than 0.01 which is far less than acceptable, however, I am still getting very reasonable results from my simulations.

Is it possible that this warning message is a glitch or something?

Thanks 

 

[Rob]

If you have a 2d hex mesh you're better off using skew and aspect ratio. Ortho quality works but is better suited to poly cells. 

Re the meshing, you can also break up the domain manually to replicate blocking. Pave is also a good approach. 

[Royale Ohoka]

I have a 2-dimensional Mesh with quadrilateral cells. It says my Maximum Skewness is 0.54 which is a good value from any literature I have seen.

Attached is a snippet of my Skewness values and Aspect Ratio. Are these all acceptable?

What is a poly cell?

In relation to the domain, I do have my mesh broken into 6 sections or so following a particular tutorial.

Thanks 

[Rob]

Aspect ratio is too high, 2-300:1 may be OK in the boundary layer, over 67k is a touch excessive. That'll explain the ortho quality as it's angle (ratio) based and the ratio will be very large or near zero. 

[Royale Ohoka]

Ok that would make more sense for sure. Are they any good solutions for changing my Aspect ratio? Thanks 

 

[Royale Ohoka]

I have 200 divisions input in the domain around the aerofoil but I have still failed to bring down the aspect ratio by increasing this number. 

[Rob]

Look to see where the bad cells are using the graph functions. 

[Royale Ohoka]

For this simulation in the transonic mach regime of a 2D Airfoil, what Bias Factor would be reasonable for me to use when creating my mesh? Some tutorials said 50000 for Reynolds numbers of a fe million whereas my analysis is approx. 15,000,000 Reynolds. 

Also, should I keep the selected 'Bias Type' or choose another option?

I am doing this in the hopes of cleaning up my aspect ratio and hence ortho quality.

Thanks 

[Rob]

Bias will cluster cells based on the Bias Type. Have a look at the tutorials. 

[Royale Ohoka]

Thanks Rob, what value of Y+ would be suitable for this transonic aerofoil simulation? I have been told a Y+ of less than 1 is good but I am getting between 5 and 10 usually. I can reduce this Y+ by increasing the mesh bias but then my orthogonal quality value reduces to below the desired level. 

Its difficult to balance these quality checks. 

Thanks for the time.

 

[Rob]

You really want y+ of around 1 to capture the near wall effect, but also then need 15+ cells in the boundary layer region. Plus a sensible aspect ratio, good cell ortho quality and good streamwise resolution. That all adds up to a high cell count. There's a reason people modelling this sort of thing properly use a lot of parallel compute and cell counts in the 10's to 100's of millions. 

[Royale Ohoka]

Yeah this is understandable, using the student license, I am only limited to approximately 512000 cells or so.

I guess what I'm really asking is if a Y+ of more than 5 or so is still reasonable in order to produce relatively accurate results in your opinion. My Ortho Quality and Aspect Ratios are now better with 0.2 and 2000 respectively.

If this Y+ is too high, would you sacrifice Ortho Quality in order to get a good Y+?

Thanks 

 

 

[Rob]

I'd sacrifice y+ but then include why, and what I'd do to improve the result in any reports. Wall functions exist for a reason, and I'd always trust those over results from a poor quality mesh. 

Student exists to allow you to learn what simulation can do, and to acquire some of the skills to use it. The results you'll get are "good enough" to compare with any literature or undergrad Uni lab projects; for postgrad work you really want to be looking at Research level licences. To give you some comparison, my PhD models had 95k cells and took around a month to run with 128Mb RAM on one core: a resonable UNIX box back in the 90's. 

[Royale Ohoka]

Hi Rob, Thanks for that. I will probably mention that these results could be improved with an increased limit in Cell count. 

With this slightly 'sacrificed' value of Y+ in favour of Ortho Quality, Skewness and Aspect Ratio, I seem to be getting less variance in lift and drag coefficient values as I increase my Mach number from 0.6 to 0.8 in increments of 0.01.

I wonder if this is an expected effect of a lower Y+?

My lift and drag graphs converge perfectly every time but my residuals not so much probably due to the Y+ value.

 

[Rob]

Pass, I don't know what you're modelling to say if there should be much of a change: I'd also not know if you showed me it as I don't really do wings & air flow.

Re the residuals, have a look at the wake region too. Too many new users fall into the trap of only looking at y+ and one or two numerical outputs: you need to study the whole solution to see what's going on. 

[Royale Ohoka]

Hi Rob,

I had a look at the poor cells that are causing my ortho quality to go down and here they are. They are only on the leading edge of this aerofoil. How would I edit this portion alone to improve the cell value??

I have tried messing about with the bias and sizing etc in that portion of my meshing but to no success.

Thanks again for the help.

 

 

 

[Rob]

Zoom in and have a look. For quad cells use skew. Ortho is great for poly cells but for hex & tet I tend to use skew and aspect ratio.