Hostname: page-component-7b7567c8c-fvjsw Total loading time: 0 Render date: 2026-11-16T06:20:59.785Z Has data issue: false hasContentIssue false

Two-dimensional continuum modelling granular column collapse by non-local peridynamics in a mesh-free method with $\mu(I)$ rheology

Published online by Cambridge University Press:  30 April 2021

Tibing Xu*
Affiliation:
School of Civil and Environmental Engineering, Ningbo University, 315211 Ningbo, PR China
Yee-Chung Jin
Affiliation:
Faculty of Applied Science and Engineering, University of Regina, Regina S4S 0A2, Canada
*
†Email address for correspondence: xu320@uregina.ca

Abstract

Granular column collapse involves complicated granular flow bridging initial and final solid-like states, with intermediate multi-regime spatial and temporal flow patterns and transitions, which include large free-surface variation. In this study, a non-local mesh-free numerical method is proposed to model these flows and capture the entire process from flowing to arresting. Free surface evolution is tracked by the mesh-free method while the non-local theory of peridynamics is used to capture the arresting for the flow. The constitutive relation to calculate the effective viscosity and stress is based on the μ(I) rheology. The non-local mesh-free method is validated to simulate a granular column collapse in which effects from the peridynamic horizon and particle distance are both examined. Non-local modelling is then used to simulate more types of granular column collapses, by comparison with other numerical results and experimental observations in terms of the free surface and velocity variations in both the fluid-like and solid-like states. The collision between two adjacent collapsing granular columns is also simulated, and the interface variations between material from each collapsing column are compared with experimental observations. The non-local modelling is shown to reflect the internal flow characteristics in the granular flow. Throughout these simulations, the non-local mesh-free method is able to calculate the free surface, velocity and interface variation in the granular flows.

Information

Type
JFM Papers
Copyright
© The Author(s), 2021. Published by Cambridge University Press
Figure 0

Figure 1.Flow chart to calculate the granular flow with peridynamics in the mesh-free method. The predictor-and-corrector time splitting scheme is used to solve the governing equations. In the predictor, the external force of gravity is solved to obtain intermediate information. With the intermediate particle number density, the pressure field is calculated by the equation of state. In the corrector, the force state is calculated, and the intermediate velocity is corrected by solving the stress tensor term in the governing equations in the framework of peridynamics.

Figure 1

Figure 2.Numerical set-up for the granular column collapse: (a) diagram for the granular column collapse and (b) bottom position of the two plates when lifting them by using a weight according to the experiments (Xu et al.2017).

Figure 2

Figure 3.Free surface profiles simulated by using different peridynamic horizons for the granular column collapse with the aspect ratio a = 1.25: (a) at t = 0.12 s, (b) t = 0.22 s and (c) t = 0.42 s. The experimental free surface profiles are from Xu et al. (2017).

Figure 3

Figure 4.Free surface variation in the solid-like state for the granular column collapse a = 1.25 by using different peridynamic horizons: (a) Ωi = 4.0Δl and (b) Ωi = 2.2Δl.

Figure 4

Figure 5.Velocity profiles at different time steps and sections calculated by the peridynamic horizon Ωi = 4.0Δl for the granular column collapse a = 1.25: (a) t = 0.129 s across the section y/L = 0.125, (b) t = 0.129 s across the section y/L = 0.375, (c) t = 0.189 s across the section y/L = 0.125 and (d) t = 0.189 s across the section y/L = 0.375. The experimental results are from Xu et al. (2017).

Figure 5

Figure 6.Free surface simulated by using different particle distances of L/Δl = 128, 64, 32 and 16 at different time steps for the granular column collapse a = 1.25: (a) t = 0.12 s, (b) t = 0.22 s and (c) t = 0.42 s.

Figure 6

Figure 7.Velocity distribution across the section y/L = 0.125 simulated by different particle distances for the granular column collapse a = 1.25 compared with the experimental measurements by Xu et al. (2017): (a) at t = 0.129 s and (b) t = 0.189 s.

Figure 7

Figure 8.Comparison of the free surface profiles among non-local modelling, local modelling and experimental measurements by Xu et al. (2017) for the granular column collapse a = 1.25: (a) at t = 0.12 s, (b) t = 0.22 s and (c) t = 0.42 s.

Figure 8

Figure 9.Comparison of the free surface profiles among non-local modelling, local modelling and experimental measurements by Xu et al. (2017) for the granular column collapse a = 5.0: (a) at t = 0.20 s, (b) t = 0.36 s and (c) t = 0.67 s.

Figure 9

Figure 10.Horizontal velocity comparison between local and non-local modelling at t = 1.0 s across the section y/L = 0.25 in the granular column collapse a = 5.0.

Figure 10

Figure 11.Snapshots of the final deposit for the granular column collapses (the dashed line represents the original shape for the column): (a) the non-local modelling for the granular column collapse a = 1.25, (b) the local modelling for the granular column collapse a = 1.25, (c) the non-local modelling for the granular column collapse a = 5.0 and (d) the local modelling for the granular column collapse a = 5.0.

Figure 11

Figure 12.Diagram for the granular column collapse occurring on one side.

Figure 12

Figure 13.Free surface comparison for the granular column collapse on one side among non-local modelling, FEM and the experimental measurements by Mangeney et al. (2010): (a) at t = 0.18 s, (b) t = 0.30 s, (c) t = 0.42 s and (d) t = 1.02 s.

Figure 13

Figure 14.Numerical set-up for the collision of two adjacent collapsing columns: (a) diagram for the granular flow and (b) bottom position of the four plates when lifting the plates according to the experimental measurements by Xu et al. (2019).

Figure 14

Figure 15.Snapshots at different time steps for the collision of two adjacent granular columns simulated by the non-local mesh-free method for Case I (The dashed line represents the original shape for the two granular columns): (a) t = 0.15 s, (b) t = 0.25 s, (c) t = 0.5 s, (d) t = 0.8 s, (e) t = 1.25 s and (f) t = 1.61 s.

Figure 15

Figure 16.Free surface profiles at different time steps from the non-local modelling and local modelling compared with the experimental measurements by Xu et al. (2019) for Case I (The dashed line represents the original shape for the two granular columns): (a) t = 0.1 s, (b) t = 0.2 s, (c) t = 0.3 s and (d) t = 0.4 s.

Figure 16

Figure 17.Final deposit for the collision of two adjacent granular columns of Case I (the dashed line represents the original shape for the two granular columns): (a) final deposit by the non-local modelling and (b) final deposit by the local modelling.

Figure 17

Figure 18.Interface variation during the collision of two adjacent granular columns simulated by the non-local mesh-free method for Case II (The dashed line represents the original shape for the two granular columns): (a) t = 0.1 s, (b) t = 0.3 s, (c) t = 0.5 s and (d) t = 1.5 s.

Figure 18

Figure 19.Interface variation at different time steps from the non-local modelling and local modelling for the collision of two adjacent granular columns (Case II) compared with the experimental results by Xu et al. (2019): (a) t = 0.15 s, (b) t = 0.20 s, (c) t = 0.40 s and (d) t = 0.60 s.

Figure 19

Figure 20.The interface in the collision of two adjacent collapsing columns for Case II at t = 0.60 s (The dashed line represents the original shape for the two granular columns): (a) the experimental measurement, (b) the non-local modelling and (c) the local modeling.

Figure 20

Figure 21.Simulation of the plane Couette granular flow in the presence of gravity by the non-local mesh-free method: (a) diagram of the granular shearing flow, (b) the averaged velocity profile for the flow by the dragging velocity Uwall = 0.0003 m s−1 and (c) the averaged velocity profile for the flow by the dragging velocity Uwall = 0.0013 m s−1. The imposed pressure for the upper plate is Pwall = 102.3 Pa and the bottom plate is fixed.