[n0t_sm0L]

Hello CFD Community,

I found this online tool, called AirShaper, that does simple and detailed CFD analyses of 3D models that can be uploaded to its server. It is a cloud computing service, and since it is paid, I wanted to know if anyone has used it, and if it is any good, with regard to accuracy of flow physics. I know how to use ANSYS Fluent, but I wanted to know if there's an alternative which can give me quick results for rudimentary analyses.

Regards,
n0t_sm0L

[arjun]

Originally Posted by n0t_sm0L

Hello CFD Community,

I found this online tool, called AirShaper, that does simple and detailed CFD analyses of 3D models that can be uploaded to its server. It is a cloud computing service, and since it is paid, I wanted to know if anyone has used it, and if it is any good, with regard to accuracy of flow physics. I know how to use ANSYS Fluent, but I wanted to know if there's an alternative which can give me quick results for rudimentary analyses.

Regards,
n0t_sm0L

As far as I understand it is openFOAM solver in the back ground so it is like any other CFD solver.


As far as alternatives go, since I develope Wildkatze I can suggest that. If you have mesh with you then till 5million cells you can use it for free from https://fvus.github.io/wildkatze/


As far as accuracy goes I do not think it will do worse than AirShaper because in last Automotive workshop for RANS Wildkatze had most accurate results using the same mesh. So expect it to be at least as good as openfoam or Fluent etc.




Here all red dots are RANS calculations from various solvers (Fluent, CFX, OpenFOAM and others).

Grey dots are LES calculations

[n0t_sm0L]

This was helpful, thanks. Didn't know they use openFOAM. Will give wildkatze a look.

[arjun]

Originally Posted by n0t_sm0L

This was helpful, thanks. Didn't know they use openFOAM. Will give wildkatze a look.

While I could not find the page that said that they used openfoam. This is what I could locate https://airshaper.com/courses

Anyway by the looks of it, what they have done is that they do the meshing probably based on oct tree based method and then perform the simulation. Since it is an wind tunnel type situation a templated input file could be created.

I was working on similar thing last year and never finished it. Basically there exists a rudimentry oct tree based mesher that might be shaped to do automated meshes for given closed surface.

In wildkatze though if you have mesh then also you can do automated solution in a single command.

For example

solver.sh -lc flow meshFile -vel 25 -wd -kw -forces -lc pf o,txt -lc iterate 500 -lc export-ensight results

would also set up flow, turbulence, then iterate and then export results. So that could be painless too.

I should note that

1. For the Full detailed car in automative workshop the CD should be 0.254 and pretty much no solver got it. Most got it around 0.27 or worse. Wildkatze got it around 0.27.

2. This was with 128 and 244 million cell meshes.

3. With 17 million (We created) coarse mesh Wildkatze got 0.33 and this part does not make me happy. (So this is reason workshop provided 128 million and 244 million cells meshes)

4. In order to get more accurate results with coarser meshes, I intend to just refine first layer of cells from wall. (Automatically) So that results could be improved even on coarse meshes. So this is something you could look out for. I intended to work on it in a month or two but if someone is using this part and not happy with the results I will put it on higher up of the list.