Learning Forum

[pvenkatesh]

Is it acceptable to create meshes with high skewness around areas of inflation? If so, would it affect whether the solution with converge or not?

I'm currently simulating an FSAE car, and I've added 10 layer of inflation to the car walls and the ground, which results in a good chuck of cells being highly skewed (around 3 million cells). 

Here is the geometry for the simulation:

And here are my inflation parameters, quality, and mesh metrics. The first layer thickness was determined using a y+ calculator with a reference length of 2.5 meters.

   

For the inflation, the scope is the entire domain, and the boundaries are the car walls. The same parameters apply for the ground walls. The growth rate is set to 1.5 to reduce skewness for higher levels of inflation layers.

1) The ANSYS manual mentions that the max skewness should not exceed 0.99, but a lot of my elements have a skewness close to 0.99. (And the maxiumum is 1, but I'll have to take a look at the geometry for that.) Will the skewness in inflation significantly affect the time it takes for the solution to converge? Are there any ways to improve mesh quality other than reducing face sizing? 

2) Would the solution be accurate if I only included inflation in some parts of the ground? (maybe a small patch under the car)

 

 

[Rob]

If you click on the worst bar in the graph you'll be able to see where the poor cells are. You can then see why they're bad and revise the mesh (or geometry) to resolve this. This could mean adjusting surface sizing or altering geometry (removing sliver faces for example).

In terms of skew, the worse it is the more likely you'll have problems converging the solution and in getting the right answer.  Y+ is a measure of the near wall cell height and is used for checking/setting up wall functions. You also need a good volume mesh to capture wake & wake interaction regions. I see far too many "good y+ meshes" which are rubbish as the volume mesh is poorly resolved for the flow. 

[blackbaddl]

What does your tire-street contact look like? Sharp contacts in this area can cause a lot of trouble. Try it like this (if this not already the case)

 

[Rob]

Pretty much, you're aiming to smooth corners and help the mesh fit in/around the geometry. 

In the above look at how the wheel touches the ground: will a simple block give a better mesh than the approach you've used? 

[vganore]

Please see this tutorial as well showing fluent meshing for FSAE cars. We have made a complete series on external aerodynamics.  

https://youtu.be/YSDH0YNfP9o?t=14