Hello all,

I've recently played with the multpiphase code and encountered a strange behavior. I have a box half_filled with sediment with water on top. During the simulation the sediment particules slowly but surely are leaking through the bottom of the box (see attached image).

Has anyone encountered this problem before ? What can I do to avoid this ?


  • It looks like you shoould adjust the shifting or augment the wall thickness (number of particles across the wall).

    Shifting moves particles to fill voids of fluid. It is a tool that must be used carefully because it can produce particles cross walls. To ways to avoid it is reduce shifting or increase wall thickness.

    Shifting is discussed in the DualSPHysics 4.2 manual "DualSPHysics_v4.2_GUIDE.pdf" within the chapter 3.6 and in the XML guide "XML_v4.0_GUIDE.pdf"at the page 84.
  • To add to what Agavino said you also have to be careful about which shifting mode you use (1,2,3), if you use a boundary layer which is only one element thick. You can read more about this in the "DualSPHysics_v4.0_LiquidGas_GUIDE"
  • Thanks !

    I did not use shifting.
    I already have increased the thickness of my boundaries and particules have stoped leaking.
    However, I have encountered other problems with my results which I am discussing directly with Dr Fourtakas.

    Thanks anyway !
  • Hi All,

    Although I have never seen this on my simulations in the past , I think TPouzol has highlighted an issue I was not aware of, I will be looking at his with the user closely and I will update you with my findings.

    Thank you!
  • Hi, may I know, how DualSPHYSICS deals with MULTIPLE layers of wall/boundary particles, as mentioned by TPouzol? The densities on these multiple boundary particle layers are solved using the continuity equation (The dynamic BC approach)?

    If I am not mistaken, DualSPHYsics applies only ONE (1) layer of boundary particle (instead of multiple layers of particles for fulfilling the full compact support of kernel as mainly reported in academic SPH journals).

    Can we actually enforce DUalSPHYSICS to use more than 1 layer of particles (where the number of layers depends on the SPH kernel used) to better represent the SPH summation near the boundary?
  • You can define the number of layers you want, that is only an issue of creating the initial condition...

  • Dear Alex, may I know what you meant by issue of creating the initial condition? Is this problem occurs when the number of layers > 1?

    Can you please elaborate?

    Thanks ya Alex.
Sign In or Register to comment.