Should I always use delta-SPH with Laminar-SPS Turbulence scheme?
Hey again
I've been reading the DualSPHysics guide and other papers through, in particle about the implementation of the laminar-SPS turbulence modelling scheme.
In Dalrymple 2006 I read that it is nessecary to apply a Shepherd filter since in this method density fluctuations will appear at the free surface. But since Shepherd filtering was abandoned in v4 of DualSPHysics there currently is no way of correcting the density fluctuations at the free surface unless I use delta-SPH?
So I guess my question is, if I use the laminar-SPS scheme I have to use delta-SPH to artificially diffuse the density, so:
- How do I know which artificial value to use? I know default is 0,1
- When I use the delta-SPH scheme will I be sure that the density fluctuations which would happen due to the turbulence scheme will be corrected and the weakly compressible fluid state is preserved?
Hope my questions makes sense, kind regards Ahmed
I've been reading the DualSPHysics guide and other papers through, in particle about the implementation of the laminar-SPS turbulence modelling scheme.
In Dalrymple 2006 I read that it is nessecary to apply a Shepherd filter since in this method density fluctuations will appear at the free surface. But since Shepherd filtering was abandoned in v4 of DualSPHysics there currently is no way of correcting the density fluctuations at the free surface unless I use delta-SPH?
So I guess my question is, if I use the laminar-SPS scheme I have to use delta-SPH to artificially diffuse the density, so:
- How do I know which artificial value to use? I know default is 0,1
- When I use the delta-SPH scheme will I be sure that the density fluctuations which would happen due to the turbulence scheme will be corrected and the weakly compressible fluid state is preserved?
Hope my questions makes sense, kind regards Ahmed
Comments
I would argue that yes, you should. Lam-SPS is useful for very energetic flows, with high shear-rates and large gradients in time and space. That is also precisely the type of situations were delta-sph, the implemented version at least, is useful, since you don't need to worry about free-surface kernel truncation creating an upwelling force (the order of magnitude of that force is many times smaller that any other component in such a flow). For close to hydrostatic situations you don't want to use delta-sph (that force will corrupt your simulation) neither lam-sps (it's a waste of resources since the SPS terms will be virtually null).
- The default value of 0.1 is a surprisingly good bet. I can't remember any papers with a profound analysis of the impact of this coefficient.
- the delta-sph terms will only filter very high wave-number fluctuations. These should be orders apart from ordinary turbulence fluctuations, especially considering the spatial resolutions typically used with sph.
I'll try to find some references to back this up.
Cheers
I am using Lam-SPS primarily since I want to be consistent in my turbulence model and have a better understanding of this one, than the artificial viscosity approach.
I am still looking into understanding delta-SPH approach, but will be using it in the future for the purpose of being consistent in my simulations. I would be happy to read any references you could bring, especially those about turbulent fluctuations and spatial resolutions in SPH.
Kind regards, thanks for your time - Ahmed