shifting and artificial viscosity

Hi all,

I know that the artificial viscosity can be tuned by some amount, and the "recomended value" is 0.01, which I tried to stick with. In the case moving square the art. viscosity value is set to 0.00001 in combination with the shifting alg.

I did some settling / drop tests with floating particles and I get better values using art. visco = 0.00001 then art visco 0.01 in combination with the shifting.

So is it possible that the shifting "slows down" the falling ability of a floating object, and it is better to use 0.00001 instead of 0.01, when one uses shifting ?????

Thank you so much for any help !!!


  • Hello,

    the artificial viscosity parameter is, as its name implies, an empirical parameter. Its optimal value depends on each case: the domain, the particle movement, the fluids used etc.It is different for each case, as you mentioned so there is no need to keep to it. It is given as an initial suggestion and to show people a typical value used in WCSPH simulations. If you have identified a better value for your case, use that instead.

    As for the shifting, the change in particle position due to it is relatively small compared to the change by the momentum equation (which is why DualSPHysics does not correct particle properties after shifting) so I do not see it have any effect on the movement of the floating object. The improved particle distribution should also give you a more evenly distributed force.

  • Dear Athanasios,

    thank you so much for you answer this clarified a lot … I feel way more confident to have a "official" statement that it is good practice to tune the artificial viscosity value, as I get a better agreement with experimental data I intended to compare with.

    For the shifting your answer is also extremely helpful, since a more evenly distributed drag also prevents the floating from rotation in a certain direction, they fall way more straight.
    However, may be I was not clear enough in my first thread. I observed lower values of the settling velocity only in the cases with the laminar – sps viscosity enabled. In the cases when I disabled the shifting I found better results that matched with the literature values basically the floating body was dropping as expected …

    Anyway, for now I decided to stick with the artificial viscosity model and shifting, since with a slightly tuned viscosity value I get values that seem to match the literature values …

    Again thank you very much for your help !!!!
    BEST H
  • Note that for wave tanks, the value of visco(alpha)=0.01 was shown to provide very good results as you can read in Altomare et al., 2015 and Altomare et al., 2017.
    However for dry-wet problems (dam breaks, sloshing tanks, etc) the value of alpha would change also with resolution.

    On the other hand, the use of shifting is recomended for internal flows and for free-surface problems only when holes can be a problem.

  • Dear Alex, thank you so much for your grat answer.
    I know these papers very well ... However due to the great validation therein I was unsure if the articivicial viscosity value can change a bit... So I am very happy that I can tune that parameter a bit so the experimental dataset from the literature I am comparing with, fits somewhat better for a siligthy lower art value of 0.007 than 0.01. However the general pattern is in the similar range also with 0.01 ...

    As you mentioned it I am struggeling with a internal flow and in particular under conditions as high flow speeds around paricles holes can become a problem... That is why I am very happy taht the shifting is implemented !!! ...

    Thanks again !!!
Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!