Relating artificial viscosity to DP

Is there a way to relate artificial viscosity to particle spacing? Presently I am using the default value of 0.01 and attempted the problem with different dp (ranging from 0.005 to 0.01).

I am attempting a dam break problem on a slope and all the literature I referred to says that I might have to tune the values of alpha with respect to the dp. Can someone please shed some light?

Thank you.


  • There is no linear relation between dp and alpha and, in addition, it will be very case dependent...

    Your only option is to have reference data (experimental data) to validate with and you will have to increase "alpha" when you decrease "dp" in order to obtain the same numerical results (in terms of velocity of the tip of the dam for example)


  • Your other alternative would be to try doing it using the other turbulence model, SPS-LES which is based on a more conventional turbulence/viscosity modelling theory.

    It can be hard to get stable, but atleast you will not have to tune alpha.

    Kind regards

  • @Asalih3d , Thank you for the comment. The model shown here is roughly 14.0X1.0X3.0 (m) with a minimum attempted particle spacing of 0.005 m. I did see a couple of comments on some papers pointing to the turbulence scheme. But when I attempted it for a small volume (1.0X0.3X0.4 (m) ), the SPS-LES-based approach gave me some really weird results. (clumping of particles). I will explore more on this

    @Alex , Thank you. I do have access to the experimental data. I like your generic approach and I did read that that it is very case-dependent. Most of the data I referred to came from wave-related problems using an alpha of 0.01. But specifically mentioning, that users should calibrate the alpha for violent interaction problems. The problem attempted here is a base case of swash zone simulation. Thank you for the inputs, I will post an update once I come through with the simulation.

  • In the case of wave-related problems where most of the tank is already "wet" the value of 0.01 has been shown to give good results during different validations cases, however with dam-break evolution where the tip of the dam reaches "dry" zones then alpha is strongly dependent on the resolution.... at least using DualSHysics and with DBC boundary conditions... This means that different formulation or different SPH options or different SPH code may not present this dependency. For example formulation based on Rieman Solver solutions and some advanced delta-SPH formulations do not need to use extra viscosity dissipative treatments


  • @Alex , This condition makes more sense in my test case. I have these dry wet zones and the effect is more pronounced at the interfaces. I have had not much of a change in deep areas where the field is always wet. I will update the thread with more findings once I compile it. Thank you.

Sign In or Register to comment.