Research Paper
Reseeding of particles in the material point method for soil–structure interactions

https://doi.org/10.1016/j.compgeo.2020.103716Get rights and content

Abstract

Lagrangian particles in the material point method (MPM) are free to flow through the background Eulerian mesh to represent material deformation. Excessive compression or tension in the kinematic field tends to entangle the particles from the initial uniform configuration, causing artificial voids or aggregations that lead to loss of the continuity of the stress field and the singular deformation gradient of the particles, undermining computational robustness and accuracy. In this paper, the reseeding operation is performed in elements depending on the numbers of the accommodated particles and their accumulated deformations. The state variables of the reseeded particles are recovered from their surrounding old counterparts. The stresses of the reseeded particles around the structure are adjusted to mitigate the fluctuations of contact force from errors in the state recovery. Benchmark problems of deep penetration of a wished-in-place T-bar and the dynamic impact of a submarine landslide on a partially buried mudmat are analysed. The results of the MPM analyses are validated by comparison with the exact solutions by theoretical analyses and the numerical predictions using computational fluid dynamics (CFD) simulations.

Introduction

Large deformation problems are frequently encountered in geotechnical practices, which include the impact of submarine landslides on pipelines (Dong et al. 2017), free-falling torpedo anchors penetrating into seabed (Kim et al. 2018), and the dynamic installation of suction caissons (Cox et al. 2014). To avoid excessive mesh distortion in the simulation of large deformation problems, Lagrangian methods seek to remesh periodically and interpolate field variables from the old meshes to the new ones (Hu and Randolph, 1998, Wang et al., 2013, Zhang et al., 2013, Tian et al., 2014); Eulerian methods allow the material to flow through a fixed spatial mesh by decoupling the mesh and material movement (Qiu and Grabe, 2012, Zheng et al., 2014); and arbitrary Lagrangian Eulerian (ALE) methods combine the merits of the Lagrangian and Eulerian approaches (Di et al., 2007, Nazem et al., 2008).

The material point method (MPM), originated from the particle-in-cell method in computational fluid dynamics (CFD) (Harlow 1964) and then extended to solid mechanics (Sulsky et al. 1995), falls into the category of the ALE methods. The material in the MPM is represented by a cloud of discrete particles that inherit history-dependent information such as mass, volume, density, velocities, deformation gradients, and stresses. A fixed Eulerian mesh is used to calculate the governing equations in each incremental step without carrying permanent information. The large deformation of the material can be derived by tracking the particles moving through the background mesh. The MPM has been applied to mimic the flow of granular materials (Bardenhagen et al. 2000), the evolution of subaerial and submarine landslides (Andersen and Andersen, 2010, Soga et al., 2016), the dynamic process of pile driving (Hamad 2016), and the large-amplitude displacement of structural elements through soil (Phuong et al. 2016).

However, the large deformation of the material entangles the particles from the initial uniform arrangement, often causing artificial voids among stretched particles or virtual particle aggregations in compressive domains (Ando et al., 2012, Sołowski and Sloan, 2015, Yue et al., 2015), somehow similar to mesh distortion in the conventional finite element method (Hu and Randolph, 1998, Wang et al., 2013). Also, virtual interpenetration or departure between soil and structure particles may be induced by the ‘penalty’ contact algorithm adopted, although other contact algorithms are also available, such as moving-mesh contact (Ceccato et al. 2016) and generalised frictional contact (Nairn et al. 2018). The varied particle arrangements may introduce numerical noises given the loss of the continuity of the stress field or the singular deformation gradient of the particles. As a result, computational robustness and accuracy tend to be undermined (Steffen et al. 2010). For instance, convergence is not necessarily enhanced with mesh fineness (Liang et al. 2019). Techniques splitting particles in sparse domains and merging particles in dense areas were then developed, Tan and Nairn, 2002, Ma et al., 2009 in the simulation of high-energy explosions as well as Ando et al., 2012, Yue et al., 2015, and Gao et al. (2017) for computer animations. Most of the previous explorations of the rearrangement of particles were based on a qualitative description and limited to visual treatment, which needs to be enhanced for the quantitative analysis of soil–structure interactions in terms of mitigating the fluctuations of contact force.

In this paper, a technique for reseeding particles specialised for soil–structure interactions based on explicit integration is presented. The reseeding operation is performed in elements depending on the number of the accommodated particles and their accumulated deformations. The state variables of the reseeded particles are recovered from their surrounding old counterparts. The stresses of the reseeded particles are adjusted to mitigate the fluctuations of contact force from errors in the state recovery. Then benchmark problems of deep penetration of a wished-in-place T-bar and the dynamic impact of a submarine landslide on a partially buried mudmat are analysed.

Section snippets

Standard material point method

The standard MPM analyses were undertaken using an in-house program, MPM-GeoFluidFlow, which stems from an open-source package, Uintah (Guilkey et al. 2012), and features a novel contact algorithm ‘Geo-contact’ (Ma et al. 2014), as well as a GPU parallel computing strategy (Dong et al., 2015b, Dong and Grabe, 2018). Meshes with identical sizes of square elements were used. The explicit updated Lagrangian calculation in each incremental step was based on the GIMP method presented by Bardenhagen

Tracking of soil boundary

The soil domain needs to be outlined before the reseeding of the particles inside. Previous studies include those by Remmerswaal (2017) based on the support domain of particles and Bing et al. (2019) using the B-spline. A new scheme is used here based on the engagement of elements, which are categorised into inner, boundary and outer individuals. An empty element with no particle inside is seen as an outer element. An engaged element is distinguished as inner if all the surrounding elements

Deep penetration of T-bar

For in-situ geotechnical investigation under deep waters, the T-bar (Fig. 4(a)), a penetrometer with no requirement of extra correlation of the penetration resistance by mobilising the full-flow failure mechanism in the surrounding soil (Fig. 4(b)), has been proven to be more accurate than the conventional vane shear test and cone penetration test (Randolph and Houlsby, 1984, Einav and Randolph, 2005). The shallow penetration of structures (less than half the diameter or width of the structure)

Conclusions

A technique for reseeding particles in an entangled arrangement was presented, specialised for soil–structure interactions. The reseeding operation was performed in inner lattices, i.e. quarter elements, depending on the number of accommodated particles and their deformed lengths. The state variables of the reseeded particles were recovered from their original neighbours. The stresses of the reseeded particles around the structure were adjusted to reduce the errors in the new stress field and

CRediT authorship contribution statement

Y. Dong: Conceptualization, Methodology, Validation, Formal analysis, Writing - original draft, Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This paper was supported by the National Natural Science Foundations of China (Grant No. 51909248) and the Open Research Fund of State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences (Grant No. Z018011).

This work was also supported by the NVIDIA Corporation with the donation of the GPUs Geforce Titan Xp and GeForce Titan V.

The author would also like to acknowledge the valuable input of Profs. Mark Randolph and Yuxia

References (63)

  • Y. Dong et al.

    Large scale parallelisation of the material point method with multiple GPUs

    Comput. Geotech.

    (2018)
  • F. Hamad

    Formulation of the axisymmetric CPDI with application to pile driving in sand

    Comput. Geotech.

    (2016)
  • Y. Liang et al.

    An efficient staggered grid material point method

    Comput. Methods Appl. Mech. Eng.

    (2019)
  • J. Liu et al.

    Impact forces of submarine landslides on offshore pipelines

    Ocean Eng.

    (2015)
  • J.A. Nairn

    Numerical simulation of orthogonal cutting using the material point method

    Eng. Fract. Mech.

    (2015)
  • N.T.V. Phuong et al.

    Numerical investigation of pile installation effects in sand using material point method

    Comput. Geotech.

    (2016)
  • D. Sulsky et al.

    Implicit dynamics in the material-point method

    Comput. Methods Appl. Mech. Eng.

    (2004)
  • D. Sulsky et al.

    Application of a particle-in-cell method to solid mechanics

    Comput. Phys. Commun.

    (1995)
  • Y. Tian et al.

    A simple implementation of RITSS and its application in large deformation analysis

    Comput. Geotech.

    (2014)
  • P.C. Wallstedt et al.

    An evaluation of explicit time integration schemes for use with the generalized interpolation material point method

    J. Comput. Phys.

    (2008)
  • D. Wang et al.

    A dynamic large deformation finite element method based on mesh regeneration

    Comput. Geotech.

    (2013)
  • B. Wang et al.

    Investigation of retrogressive and progressive slope failure mechanisms using the material point method

    Comput. Geotech.

    (2016)
  • A. Zakeri

    Submarine debris flow impact on suspended (free-span) pipelines: Normal and longitudinal drag forces

    Ocean Eng.

    (2009)
  • X. Zhang et al.

    Particle finite element analysis of large deformation and granular flow problems

    Comput. Geotech.

    (2013)
  • S. Andersen et al.

    Modelling of landslides with the material-point method

    Comput. Geosci.

    (2010)
  • R. Ando et al.

    Preserving fluid sheets with adaptively sampled anisotropic particles

    IEEE Trans. Visualization Comput. Graphics

    (2012)
  • ANSYS. (2011). ANSYS FLUENT 14.0 theory guide, ANSYS, Inc....
  • S.G. Bardenhagen et al.

    An improved contact algorithm for the material point method and application to stress propagation in granular material

    Comput. Modeling Eng. Sci.

    (2001)
  • S.G. Bardenhagen et al.

    The generalized interpolation material point method

    Comput. Modeling Eng. Sci.

    (2004)
  • N. Boukpeti et al.

    Analytical modelling of a submarine slide under steady flow in relation to the impact load on a pipeline

    Géotechnique

    (2012)
  • F. Ceccato et al.

    Analysis of piezocone penetration under different drainage conditions with the two-phase material point method

    J. Geotech. Geoenviron. Eng.

    (2016)
  • Cited by (15)

    • A methodology to predict the run-out distance of submarine landslides

      2023, Computers and Geotechnics
      Citation Excerpt :

      Experimental studies include conventional gravity flume experiments with the small-scale (Mohrig et al., 1998; Ilstad et al., 2004; Haza et al., 2013; Yin and Rui, 2018; Yu et al., 2022) and centrifugal experiments (Bowman et al., 2006; Boylan et al., 2009; Gue et al., 2010; Yin et al., 2019); qualitative and quantitative assessment and analysis were carried out, and important physical phenomena were obtained, such as hydroplaning, morphological characteristics, soil–water mixing observation, evolution of debris flow into turbidity current, etc. In terms of numerical simulation, commonly used numerical methods are based on the continuum mechanics framework, such as the large-deformation finite element method (FEM) (Dey et al., 2016; Zhang et al., 2019b), particle finite element method (PFEM) (Zhang et al., 2019a and 2022), material point method (MPM) (Dong et al., 2015, 2017 and 2022; Dong, 2020), finite difference method (FDM) (Imran et al., 2001), depth average method (Dong et al., 2020; Qian et al., 2020), smoothed-particle hydrodynamics (SPH) (Cheng et al., 2014; Tao, 2015), or computational fluid dynamics (CFD) (Gauer et al., 2005; Xiu et al., 2015; Debnath et al., 2018; Liu et al., 2019); in parallel, based on the discontinuous mechanics framework, the discrete element method (DEM) (Utili et al., 2015) and CFD-DEM coupling method (Zhao et al., 2014; Jiang et al., 2015; Cheng et al., 2018; Nian et al., 2021) also attracted attention. Numerical simulations enable efficient quantification of governing processes and their visualization and can analyze complex and/or large-scale conditions.

    • An improved quadrature scheme in B-spline material point method for large-deformation problem analysis

      2022, Engineering Analysis with Boundary Elements
      Citation Excerpt :

      The combination of the merits of both Lagrangian and Eulerian methods makes the MPM quite suitable for solving solid mechanics problems with large deformations [3,4], material discontinuity [5,6], and multiphase interaction [7,8] that are challenging for the traditional mesh-based methods such as finite element method (FEM) and finite difference method (FDM), and finite volume method (FVM) [9]. Since the introduction in the 1990s [1,2], the MPM has gained significant progress and been used in many scientific and engineering problems, e.g., penetration and contact [10–12], explosion [13,14], crack propagation [15,16], landslide [17–19], fluid-structure interaction [20–22], thermomechanics [23,24], large deformation elasto-plasticity [25,26], geological engineering [27–29], etc. It has been demonstrated that there exists numerical noise in the original MPM when the particles move across the background cell boundary [30].

    View all citing articles on Scopus
    View full text