SPH fluid simulation: cylinder moving review

This is a revision of another simulation to show the vortex formation process of an incompressible fluid in a pipe and flowing through a cylinder (Re~150).

I used color bands to show the vortex formation and evolution process.

The code is a modification of SPH (named XSPH) with multifluid support, using ~30.000 particles (at 7 fps in a PIV 2.6GHz, RAM 1.5GB, Debian Etch), and was reprogrammed using C (Ansi C) and allegro libraries to increase the performance. The animation was built in slow camera mode (1/20) joining 950 image captures using avidemux.

Category:

Science & Technology

Tags:

• SPH
• fluid
• simulation
• cylinder
• moving
• CFD
• Von
• Karman
• Street
• Vortex

[snailmr2turbo]

Sweet simulation! I've been trying to do this but having a tough time making vortexes form... my fluid must not be too compressible... Any suggestions for papers to read on this approach?

[jahkr]

Hi, snailmr2turbo, thanks for comment. I don't understand the question, I saw your video "second fluid vortex video" (good simulation!) and appear the vortex formation process, please, reformulate your question.

[snailmr2turbo]

@jahkr no problem. I was just wondering if there is a research paper or tutorial that you read to help you implement your SPH simulation? in my simulation I actually found I got good vortexes when I have over 150,000 particles but not with only ~30,000 :(...Thanks for the response :)

[jahkr]

Hi again, I have no papers about vortex formation using SPH, I'm sorry. When I did this simulation I saw that if use less particles it's like to increase the viscosity (the vortex formation is related with Re number, and is inverse proportional to the viscosity), but you can try increase the speed or density (particle mass) to increase the Re number too.

[jahkr]

I have some question:

-Is your simulation in real time?

-What kind of neighbor search algorithm you use?

By the other hand, my simulator is for incompressible fluids, but I think that your results is for a compressible fluid, I'll try to change my simulator and reproduce your results using 30k particles.

[ZnahNah]

Some time ago I simulated vortex street formation with Lattice-Boltzmann method. It is fun to see how completely methods produce similar results.

[jahkr]

Hi ZnahNah, I saw your videos about LBM, and they are great, my first work in fluid dnamics was modify the original LBM to solve free surface problems.

Test the code or the method with well known problems is fundamental part of the development.

[akg26giga]

hi, it's very good simulation. tube flow is a hard job to simulate. what kind of the external force on the particle you used?