Simulation of Filling of a Cylinder by a Non-Newtonian Fluids (Power Law Fluid)

Hello

I simulated the filling of a cylinder by a Power-Law fluid.

But as it was shown in the attached video some particles come out of the container. That phenomenon does not happen in reality.

Is there a rationale behind the phenomenon according to SPH?

Thanks a lot


Comments

  • Hi,

    I am not sure of the details of the simulation. Some pointers, make sure you have more than one layer of wall boundaries. If the problem persists, use shifting only on the fluid (ignore boundaries) instead of "Full"

    if you have more problems, let me know.

    Thanks,

    George

  • I appreciate your guidance.

  • Dear Dr George

    Following your comments, the issue is solved.

    Having observed the results it seems that there is an instability in velocity results however there is also a bit of instability in pressure results.

    Do you have any recommendations to mitigate instability in velocity results?

    Best regards

  • I carried out the following measures to address instability in results:

    As you know simulation of Non-Newtonian fluids is associated with highly non-linear behavior in viscosity which increases the risk of instability in SPH simulation. According to EOS of WCSPH the pressure is proportional to the seventh power of density. Any fluctuation in the density may cause pressure oscillation and noise. Pressure term is also directly implemented in the momentum equation. So instability in pressure has an effect on velocity results.

    Density diffusion term (DDT): An available option in DualSPHysics (DSPH) is to implement density diffusion term (DDT). The default value of DDT in DSPH is 0.1. I increased the value step by step to 0.6 and could see the higher pressure at the gate and a steady decrease in the cylinder as expected. There are, however, still some noises.

    Time stepping: It is recommended to use Symplectic time integration scheme if numerical stability is an issue, but I surprisingly observed the fluid was collapsed using this time integration scheme and even the injection process cannot be accomplished.  Then I used Verlet scheme. Verlet scheme has a second part to stop divergence of integrated values after N time-steps. The default value of the N in DSPH is 40. I decreased N step by step to 8. It is recommended N should be higher than 10. I could not see any significant changes in the results by changing the N parameter of Verlet scheme.

    Shifting algorithm: Anisotropic particle spacing results in the introduction of noise in the velocity and pressure field. There is an option in DSPH to reduce instability in velocity and pressure. Shifting algorithm in DSPH is regulated by shifting coefficient and shifting TFS.

    By increasing the sifting coefficient the instability still persists but I could observe that range of velocity magnitude decreased significantly.  

    Decreasing IDP: Actually this measure is the main measure to improve stability in the SPH simulation results. I have investigated this solution too but it does not seem to mitigate instability. I decreased IDP from 1mm to 0.6 mm step by step. There are no significant changes in results signaling improvement of stability.

    Please let me know if you have any comments.

    Thanks

  • Hello to DualSPHysics team

    I wanted to let you know I have realized that the code works very well for some Power Law fluids.

    If want to give you the specifications of those fluids for n higher than 0.75 and k lower than 0.002 I haven't seen any noises or problems in the results.

    I just wanted to ask you what is the relaxation parameter to ease the effects of increasing viscosity at a low flow behavior index (I mean 0.7 for example). How can I check the viscosity calculations?

    Please take a look at the figure (apparent viscosity variation in the cross-section of Poiseuille flow ) as you noted before we can see quite large viscosity at the low flow behavior index.

    Thank you for your kind response.

  • Thank you @salehmeiabadi good to know all works okay!

Sign In or Register to comment.