Видео

You can't see shedding in a geometry like this even in significantly high Re numbers.

@@erencolak7387 is that due to the narrow walls?

It did the piu~ psh boingg

I guess you mean the solver, and I seriously don't know. Probably pimpleFoam or interFoam... I would bet on interFoam.

Deneysel olarak da yapması çok zor bir şey değil aslında

Actually i want to do 3D version it WITH LES. Lucky for me I have access to super computers right now... But really lack time... So I want to do it but I don't know when will I do it. Maybe in next week's...

I can't because I deleted the solution files. But you may see the 3D version in higher FPS in next week's 👌🏻

@@erencolak7387it’s been 3 weeks

Thnx! Its OpenFOAM, ESI version. But IDK which version is. Probably 1912 or newer.

Yes it does. Whole point is the lowering the number of mesh while capturing physics accurately. It doesn't matter if it's steady or transient.

Hi, I don't really remember the solver. I need to check the case files, which will take about a week because I am away from my PC, and you can send me message from linkedIn, I don't want to write my e-mail in here.

@@erencolak7387 ok, i'll keep in touch. thanks a lot. you did a great work actually. keep it up!

Hi Michael! Sorry for late reply, now I checked and yes it is chtMultiRegionTwoPhaseEulerFoam.

Thank you for the insight, well, In comments I also explained need of corrector steps, and I am well aware of the initial mesh, which is (I think) the main reason of bad unorthogonality in this case. Basically, I am not saying OpenFOAM is the reason of bad elements, but the initial state I provide, and refinement condition I provide is the problem. But as you said it's not really bad. You also have some very good simulations on your channel! I am also using immersed boundary on my PhD thesis, it is brilliant. 100 lines of code, makes our life much easier.

I believe if you have good quality background mesh(low in number) you can avoid problems like this. Because it just splits the cells, does not morph them to have high quality.

@@erencolak7387 generally if you define pressure based differentials, the cells are divided in two in the direction of largest pressure difference. The mesh coarsens as the flow stabilizes in your case (the video). So even the narrow cells close to the circular geometry, generally become more rectangular which actually improves element quality.

Idk what the heck you both are talking about but that sounds very very interesting to physics nerd like me

@@arnavtete7793 You can think the mesh as dx in any numerical solution, mesh is actually 3D version of it. Adaptive methods are being used in ODE solutions too. He basically tells that if I decrease/increase dx with respect to pressure gradient the problem(this is hard to explain with simple terms) will be solved. Generally he is right but not for this background mesh (initial dx).

@@arnavtete7793 Hahah are you in college? CFD is always fun

This can be done on CFDEM, but it would take "some" effort. I did this with palabos , LBM-IBM

I would say so since bottom mesh is not moving

I did this simulation on Ryzen 5 3600 PC

As he stated, they ended up with a lot of non orthogonal elements.

Look at the mesh near the obstacles

@@pointinpolyhedron kinda obvious since the angle of each line is not orthonormal near bend/curved topological sufaces.

@@user-hk8yp7cw1v why is that bad? i don't know anything about this

@@NutchapolSal it's not bad perse, fact is that the lines are just not orthogonally placed next to each other since the angle between both the axis are less than 90 degrees...

Yes it can. Well, my thesis subject will also be blood flow. I hope they open pos doc position for in 2-3 years!

Indeed, CFD always amazes me!

OpenFOAM is great environment to learn CFD, good luck on your progress! It took approx. 24 hrs.

@@erencolak7387 Could you share your turbulenceProperties file for this case, I have try a lot , but cant get a reasonable result for any "2D LES" case.