Refinement on non-hydrostatic shallow granular flow model in a global Cartesian coordinate system

TL;DR

This paper refines a non-hydrostatic shallow granular flow model in a Cartesian coordinate system, enabling better simulation of flows over arbitrary topography with improved analytical and numerical methods.

Contribution

The paper reformulates the vertical normal stress in a quadratic polynomial form, leading to a refined non-hydrostatic model suitable for numerical implementation.

Findings

The refined model accurately simulates shallow granular flows over complex topography.

Numerical instability issues are addressed with an approximate enhanced gravity formula.

Comparison shows the Cartesian model performs well against bed-fitted models.

Abstract

Current shallow granular flow models suited to arbitrary topography can be divided into two types, those formulated in bed-fitted curvilinear coordinates, and those formulated in global Cartesian coordinates. The shallow granular flow model of Denlinger and Iverson \cite{Denlinger2004} and the Boussinesq-type shallow granular flow theory of Castro-Orgaz \emph{et al}. \cite{Castro2014} are formulated in a Cartesian coordinate system (with $z$ vertical), and both account for the effect of nonzero vertical acceleration on depth-averaged momentum fluxes and stress states. In this paper, we first reformulate the vertical normal stress of Castro-Orgaz \emph{et al}. \cite{Castro2014} in a quadratic polynomial in the relative elevation $\eta$. This form allows for analytical depth integration of the vertical normal stress. We then calculate the basal normal stress based on the basal friction law and scaling analysis. These calculations, plus certain constitutive relations, lead to a refined full non-hydrostatic shallow granular flow model, which is further rewritten in a form of Boussinesq-type water wave equations for future numerical studies. In the present numerical study, we apply the open-source code TITAN2D to numerical solution of a low-order version of the full model involving only a mean vertical acceleration correction term. To cure the numerical instability associated with discretization of the enhanced gravity, we propose an approximate formula for the enhanced gravity by utilizing the hydrostatic pressure assumption in the bed normal direction. Numerical calculations are conducted for several test cases involving steep slopes. Comparison with a bed-fitted model shows that even the simplified non-hydrostatic Cartesian model can be used to simulate shallow granular flows over arbitrary topography.