Physical Process-Based Entrainment Behaviour Modelling of Diluted Debris Flow Using SPH Incorporated with HBP-DP Approach

Yangfan Ma, M. Asai, Guihai Chen, Zheng Han2023 · DOI

TL;DR

This paper introduces a novel 3D SPH model that accurately simulates debris flow entrainment by coupling a Herschel-Bulkley-Papanastasiou (HBP) constitutive model for the flow with a Drucker-Prager (DP) softening model for the erodible bed. This approach captures the crucial role of pore water pressure in reducing soil strength and driving progressive erosion, showing good agreement with large-scale flume experiments.

What Problem Does This Paper Address?

Debris flows are a significant natural hazard, capable of immense destruction due to their ability to entrain vast amounts of sediment, exponentially increasing their volume and destructive potential. A key factor in this entrainment process is the development of excess pore water pressure within the bed sediment, which significantly reduces its shear strength. While previous research has explored debris flow dynamics, a comprehensive numerical model that accurately captures the progressive entrainment behavior, particularly the interaction between a viscoplastic debris flow and an erodible bed influenced by pore water pressure, has been lacking. Existing models often simplify the problem by assuming shallow water flow or neglecting the complex interplay of fluid-sediment mechanics and soil strength reduction. This paper aims to bridge this gap by developing a robust 3D numerical framework.

Methods

The core of this study is a three-dimensional Smoothed Particle Hydrodynamics (SPH) model. SPH is a mesh-free Lagrangian method well-suited for simulating large deformations and free-surface flows, making it ideal for complex phenomena like debris flows.

Key methodological innovations include:

  • Two-Phase Modeling: The model treats the debris flow and the underlying bed sediment as distinct phases, allowing for detailed interaction simulation.
  • Herschel-Bulkley-Papanastasiou (HBP) Constitutive Model: Both the debris flow and the bed sediment are modeled using the HBP model, which describes non-Newtonian fluid behavior, specifically Bingham-type fluids. This is crucial for capturing the complex rheology of debris flows.
  • Drucker-Prager (DP) Softening Model for Erosion: To simulate mass erosion from the bed, the study incorporates a Drucker-Prager (DP) yield criterion. Crucially, this DP model is adapted into a softening model. This softening is driven by the pore water pressure ratio, directly linking the pore water pressure development (due to undrained loading from the debris flow) to a reduction in the bed sediment's shear strength.
  • Phase-Change Approach: A straightforward phase-change approach is implemented based on the mutation of effective viscosity. This helps to prevent unrealistic minute displacements of viscoplastic materials on steep slopes.
  • Dynamic Boundary Conditions (DBC): For the rigid boundaries of the flume, dynamic boundary conditions are employed, which offer straightforward implementation and compatibility with complex geometries.
  • Explicit Second-Order Verlet Scheme: This scheme is used for time integration, with the time step governed by the Courant-Friedrichs-Lewy (CFL) condition and an additional restriction from viscous forces.

Datasets

The study uses data from a series of large-scale flume experiments conducted by the United States Geological Survey (USGS). These experiments, totaling 163 runs, investigated debris flow behavior under three configurations: rigid bed, erodible dry bed, and erodible saturated bed. The paper compares its simulation results against these established experimental datasets. No new datasets were generated or introduced by this study.

Key Insights and Ideas

  • Pore Water Pressure is King for Entrainment: The study strongly reinforces the idea that excess pore water pressure in saturated bed sediments is the primary driver for significant debris flow entrainment and momentum gain. As pore pressure rises, it directly reduces the bed's shear resistance, leading to a positive feedback loop of increased entrainment and flow velocity.
  • Viscoplastic Flow and Soil Softening are Essential: Accurately modeling the non-Newtonian, viscoplastic nature of debris flows (using HBP) and the pressure-dependent strength reduction of the soil (using a softened DP criterion) is critical for capturing progressive entrainment.
  • A Unified Approach for Erosion: Integrating the HBP rheology with a DP softening model provides a more physically consistent way to model mass erosion compared to simpler approaches. The pore water pressure ratio directly influences the internal friction angle, a key parameter in the DP criterion.
  • Phase Change for Stability: The implemented phase-change approach based on effective viscosity mutation is a clever way to handle the transition between yielding and non-yielding states in viscoplastic materials, preventing spurious particle movements on slopes.

Key Results

The numerical model was validated against the USGS flume experiments for three scenarios:

  • Rigid Bed: The simulation accurately reproduced the rapid collapse, acceleration, and subsequent deceleration of the debris flow as it encountered the curved bed section and exited the flume.
  • Wet Erodible Bed: The model successfully captured the "explosive" behavior of debris flows on saturated beds. It showed an initial slight velocity decrease, followed by a significant increase in momentum and entrainment as pore water pressure built up, leading to a larger run-out extent than the rigid bed case.
  • Dry Erodible Bed: The simulation demonstrated a stark contrast, with the debris flow velocity progressively decreasing upon encountering the dry bed. Minimal erosion was observed, and the run-out was significantly shorter compared to the other scenarios.

In all three cases, the 3D numerical results showed good agreement with the experimental data, encompassing flow momentum feedback and erosion patterns. The paper highlights that the simulation results were consistent with experimental observations regarding the front position propagation of the debris flow.

Limitations and Future Work

  • Parameter Sensitivity: The authors acknowledge that the initial particle distance influences the simulation outcomes. However, a detailed sensitivity analysis of all parameters was beyond the scope of this paper.
  • Computational Cost: While SPH is powerful, 3D simulations of complex multi-phase flows can be computationally intensive. The paper mentions using a large number of particles (over 1 million boundary particles in total for the simulation).
  • Simplifications: The study uses a simplified approach for phase change and boundary conditions.

Future work could involve:

  • Conducting comprehensive parameter sensitivity studies to better understand the influence of various input parameters.
  • Exploring more advanced phase-change mechanisms or coupling with other erosion models.
  • Investigating the impact of different debris flow compositions (e.g., varying boulder content) and more complex topography.
  • Optimizing the SPH implementation for improved computational efficiency.

Summary updated

Abstract

Debris flows overriding steep valleys can cause a significant decrease in bed friction resistance due to undrained excess pore water pressure, leading to an exponential increase in both destructiveness and volume. This study develops a two-phase numerical model based on Smoothed Particle Hydrodynamics to simulate the progressive entrainment behavior of debris flow accurately. The fluid and bed-sediment materials are modeled using the non-Newtonian Bingham-type Herschel-Bulkley-Papanastasiou (HBP) constitutive model. The mass erosion behavior of debris flow is achieved and augmented by incorporating the Drucker-Prager (DP) softening model, which accounts for variations in the pore water pressure ratio across different saturation states. A straightforward phase-change approach is implemented according to the mutation of effective viscosity to prevent any minute displacements of viscoplastic materials when subjected to steep inclinations. The multi-phase model has been compared with the large scale flume experiments conducted by the United States Geological Survey. The 3-D numerical results obtained from the rigid bed, dry and wet erodible bed exhibit a good agreement with the experimental data, encompassing flow momentum feedback and erosion patterns. This paper initially attempts to simulate the entrainment of multiple phases in a steep valley by incorporating viscoplastic flow.
BibTeX
@misc{ma2023physical,
  author = {Yangfan Ma and M. Asai and Guihai Chen and Zheng Han},
  title = {{Physical Process-Based Entrainment Behaviour Modelling of Diluted Debris Flow Using SPH Incorporated with HBP-DP Approach}},
  year = {2023},
  doi = {10.23967/c.particles.2023.029},
}

Loading related content...

Figure