CRKSPH: A Novel Approach to SPH Simulation

Nicholas Frontiere, University of Chicago

We demonstrate a version of smoothed particle hydrodynamics (SPH) that employs a first-order consistent smoothing function that exactly reproduces linear fields with particle interpolants. Moreover, we present an improved “limited” form of the artificial viscosity, which reduces excess diffusion by capitalizing on the increased accuracy of the kernel function. This scheme confers all of the benefits of traditional particle methods, such as Galilean invariance and natural conservation of momentum, while also eliminating some of their shortcomings, such as overly aggressive artificial viscosity and their inability to reproduce linear fields. Previous efforts in this realm have run into difficulties maintaining conservation of momentum when the kernel functions are no longer symmetric. Here, we utilize a reformulation of the fluid equations that maintains conservation without any assumption on kernel symmetries. Application of the outlined scheme exhibits improved solutions to problems featuring fluid shearing and contact mixing such as the Kelvin-Helmholtz instability, which hitherto proved difficult for SPH codes.

Abstract Author(s): Nicholas Frontiere, J. Michael Owen, Cody Raskin