Transonic 2D flow around Onera OAT15a airfoil at Mach 0.73

Transonic flow around the supercritical Onera OAT15a airfoil at Mach 0.73 and 7 degrees angle of attack. The free stream state is that of air at sea level, the airfoil has a chord length of 23cm. The chord Reynolds number is about 3 million.
Invariant-domain preserving compressible Navier-Stokes solver (github.com/con...) based on the finite element toolkit deal.II (www.dealii.org).
References: arxiv.org/abs/... arxiv.org/abs/...

[@petterlarsson7257]

i love how theres a periodic wave of shockwaves at the front every 8 seconds

[@professionalprocrastinator8103]

Thank you so much Matthias. This simulation is my muse.

[@daikiwatanabe9697]

It seems to use a very fine mesh. What is the resolution?

[@jaromirandel543]

Nice, Now, please show the same with supercritical airfoil.

[@johnthebun]

why do the waves behind the shockwave not outpace the shockwave itself even though they seem to be traveling faster? just noticed that waves at the top seem to be moving ahead of the shockwave

[@southfox2012]

great work

[@spacejunk2186]

What is the meaning of this straight line on the top of the airfoil?

[@matthiasmaier8956]

This is a schockwave that forms due to the flow above the airfoil reaching supersonic speeds. A supercritical airfoil such as the OAT15a is specifically designed so that this shockwave is as far back and controlled as possible.

[@petterlarsson7257]

@@matthiasmaier8956 if the airfoil is traveling at mach 0.73 then why is the shockwave stationary relative to it

[@matthiasmaier8956]

@petterlarsson7257 because the pressure/density distribution is "stationary" and not the air molecules themselves. An individual air molecule will travel at Mach 0.73 and will run into this shock (with a sudden increase in density and pressure). There is actually a nice wikipedia article about this: en.m.wikipedia.org/wiki/Supercritical_airfoil

[@southfox2012]

how about internal flow ? such as blade ?

[@uwuowo4856]

Why is there so much flow separation? Are aeorofoils this inefficient at this sppeed?

[@tensorific]

Terrific video. What flow variable are you visualising here? Is this solver validated?

[@matthiasmaier8956]

We're visualizing the magnitude of the gradient of the density on a log scale.

[@fahadabdullah3046]

I got multiple questions 1. Can you share the mesh density and CFD turbulence model, CFD software and parameters used in simulation for this case? 2. What system did you run it on? Desktop or Cluster? What were the no of CPUs, memory etc. used? Or was it GPU based? 3. It looks like a transient simulation where the shockwave boundary interaction seperates the flow downstream but also results in an upstream periodic separation of flow. Am I right? Also, your work is fascinating and mesmerizing

[@matthiasmaier8956]

(1.) This is a direct numerical simulation without a turbulence model. So in order to resolve the boundary layer we had to use a heavily graded mesh which resulted in a significant runtime of around a day (?) on roundabout 10 compute nodes (320 cores). I do not recall precisely (2.) The code runs on CPUs, you can find it here: github.com/conservation-laws/ryujin I will update and check our example parameter files over the coming and ping you when I have updated the OAT15a parameter file to our development version. (3.) Yes, this is the buffeting effect that can be observed, see doi.org/10.2514/1.9885

[@aravindm8512]

Bro your results are extremely insane. What software are you using to postprocess?

[@matthiasmaier8956]

Thanks. I created temporal snapshots (vtk/vtu files) from the running simulation and used Paraview to postprocess and writing out PNGs. Finally the images were converted to a movie with ffmpeg.

[@jacobjivanov]

What is actually being shown here? log(grad p)?

[@matthiasmaier8956]

@@jacobjivanovsomething close to log(|grad(rho)|)

[@PedroGarcia-ib6tt]

Nice video. Can you provide the color map that you used in it?

[@juanpabloruscio6045]

@Matthias Maier congratulations for your work, that looks impressive. I need to perform a simulation of an OAT15A at the same conditiones except for the angle of attack, 4.5º (in my case). The objective of obtaining this simulation is to analyse it afterwards by the exergy methodology. Do you know if there is any open source data of the OAT15A, like meshes or coordinates? Thanks very much!

[@matthiasmaier8956]

Sorry, I just saw your comment. Data for the OAT15a airfoil is annoyingly hard to get. For the OAT15a we have used the following description of the OAT15a airfoil: github.com/conservation-laws/ryujin/blob/development/source/geometry_airfoil.h#L407-L570 The values are extracted from here lume.ufrgs.br/handle/10183/28925 (see also the discussion here: www.cfd-online.com/Forums/main/114894-oat15a-geometry.html)

[@antony2857]

Could you please share the mesh file?

[@matthiasmaier8956]

The mesh is generated in the code directly [1]. So you could just run the mesh generation routine and the output the mesh. What format would you need? [1] github.com/conservation-laws/ryujin/blob/development/source/geometry_airfoil.h

[@DeanHolder]

@@matthiasmaier8956 I need this mesh in a .cgns do you think that is possible with that?

[@priyanshusingh7365]

@@matthiasmaier8956 Could you please share the mesh file in either .msh or .cas format or provide a method to read the mesh file?

[@badlamplampgod1024]

dude holy cow

[@DGSEM]

Nice simulation Matthias! I will try this example with my DG framework. 😉 What BCs do you use? Is the airfoil curved?

[@matthiasmaier8956]

Yes, the airfoild boundary (and the mesh) is curved. You can find the airfoil description here tinyurl.com/2fnyu43v (github), this parameter file contains the free stream state tinyurl.com/twchum7j and other information. On the outer boundary a subsonic inflow and outflow condition is used (see § 4.3 in www.math.tamu.edu/~maier/arxiv.pdf - on the arxiv early next week). Also § 6.4 for some 3D for the airfoil. And if you're up for the challenge there is also § 6.3.

[@DGSEM]

@@matthiasmaier8956 Recently found some time to setup the simulation: kzbin.info/www/bejne/hqC1oWaQiNRqbcmqq=hd1440

[@matthiasmaier8956]

@@DGSEM looks good 🙂