Research Paper
Shear band static evolution by spatially mobilized plane criterion based Drucker-Prager model and numerical manifold method

https://doi.org/10.1016/j.compgeo.2020.103962Get rights and content

Abstract

The shear band due to strain localization is deemed a strong discontinuous plane in this study, and simulated by the numerical manifold method (NMM). The SMP (Spatially Mobilized Plane) criterion is incorporated into the Drucker-Prager (DP) model by the transformed stress (TS) method. The constitutive integration of plasticity is carried out on the new yielding surface with the tensile part being cut off, leading to a mixed complementarity problem (MiCP). The Gauss-Seidel based projection contraction (GSPC) algorithm is invoked to solve the MiCP. In NMM, the cohesion is a constant when the contact is in slipping state. This study modifies the cohesion so that the decaying of cohesion could be reflected with the growth of the sliding. To accelerate the convergence, the contact stiffness matrix in open state is proposed to be constructed in a new way in this paper. The NMM formulation for solving the strain localization problem is derived in this study. The displacement controlled method is extended to the material nonlinearity accompanied with contact problem. Two examples, including the soil block and soil slope under compression, have been studied to verify the efficiency of the proposed method in simulating the process of shear band evolution.

Introduction

The development of shear band is accompanied with highly localized deformation which initially emerges from a homogeneous deforming displacement field. Associated with the loss of material strength, the shear band induces tangential relative sliding which causes the failure of geo-materials (Armero and Linder, 2008, Gravouil et al., 2002, Rudnicki and Rice, 1975, Hobbs and Ord, 1989, Pijaudier-Cabot and Bažant, 1987). Actually, the shear band is a strain localization band with a finite width. Therefore, the classic finite element method which uses the classic rate-independent plastic model fails to accurately predict the localization phenomenon.

To tackle the problem above, some efforts have been made including using an elastic-viscoplastic material model (Loret and Prevost, 1990, Loret et al., 1995, Prevost and Loret, 1990), incorporating a length scale into the plastic evolution equations (Pietruszczak and Óz, 1981), and also other theories including nonlocal integral method (Chen et al., 2000, Di Prisco and Imposimato, 2003), high gradient formulation (Swaddiwudhipong and Poh, 2019), and the Cosserat theorem (Ehlers and Volk, 1999, Khoei and Karimi, 2008). Nevertheless, the introduction of a length scale into the governing equation does not guarantee to accurately capture the jump in the displacement gradient without the refinement of mesh in the area of shear band. Deb and Prevost (Deb et al., 1996) used adaptive meshing to accurately capture the shear band behavior by refining the mesh in the area of the weak discontinuity, which, however, causes high computation cost. Considering that the shear band width is very small compared with other length scales and that the calibration of the length scale requires special experiment approaches (Wolf et al., 2003), some researchers preferred to treat the width of shear band as zero (Belytschko et al., 1988, Borja, 2000, Oliver et al., 1999, Regueiro and Borja, 2001, Simo et al., 1993) so that the kinematics of the weak discontinuity in strain field were reduced to those of strong discontinuity in displacement field. In this way, numerical computation is simplified by avoiding both the required mesh refinement near the discontinuity as well as the need for a more complex plastic model (Sanborn and Prévost, 2011). Hence, in this study the shear band is treated as strong discontinuities with zero width.

By assuming the distribution of the plastic multiplier takes on the form of Dirac-delta function, researchers (Simo et al., 1993, Sanborn and Prévost, 2011, Oliver, 1996, Borja, 2002, Armero and Garikipati, 1996, Armero, 2009, Borja and Regueiro, 2001) derived that the onset of the strong discontinuity strain localization occurred when the acoustic tensor became singular. Leroy (Leroy and Ortiz, 1989) and Oliver (Oliver, 1996) regarded the searching algorithm for the localization of associative plasticity as the constrained optimization problem. And this algorithm was extended, by Sanborn (Sanborn and Prévost, 2011) and Wells (Wells and Sluys, 2001), to the non-associative plasticity. Compared with the numerical algorithm, scholars (Rudnicki and Rice, 1975, Oliver et al., 1999, Armero and Garikipati, 1996, Borja and Regueiro, 2001, Runesson et al., 1991, Ottosen and Runesson, 1991) also derived the analytical solution of the initiation condition. This study revises the Drucker-Prager model by the SMP criterion (Sun et al., 1999, Matsuoka et al., 1999, Matsuoka and Nakai, 1974, Matsuoka and Nakai, 1985) of which the influence on the evolution of shear band is studied, which is not covered by existing literatures. Since it is still in progress to derive the closed-form solution of the localization condition for the SMP-based Drucker-Prager model, the numerical algorithm is used in this study.

To advance the solution after the strong discontinuity is inserted into the mesh, it is necessary to formulate a contact constitutive model to account for the behavior of the contact interfaces. Cohesive laws (Samaniego and Belytschko, 2005, Rabczuk et al., 2007, Rabczuk and Samaniego, 2008, Zi and Belytschko, 2003) and frictional laws (Yu, 2010, Khoei and Nikbakht, 2007, Kim et al., 2007, Liu and Borja, 2008) are the two primary approaches to govern the post-localization behavior along the shear band. The contact in NMM is governed by the Mohr-Coulomb frictional law. Inherited from DDA (discontinuous deformation analysis), the contact in NMM is treated as a collection of contact pairs. Each contact pair could be in one of the three contact state: fixed, slipping, or open. And the selection and the shifting of the contact state is determined by the open-close iteration (Shi, 1992). This study revises the cohesion when the contact is in slipping state so that the decaying of the cohesion is realized with the growth of the sliding. Besides, when the contact state is shifted to open state from fixed state or slipping state, the contact stiffness suddenly becomes zero. The big change in contact stiffness deters the convergence of a nonlinear problem (Sanborn and Prévost, 2011). Therefore, this paper proposes to construct the contact matrix in open state in three different cases to alleviate the change in stiffness.

In (Liu and Borja, 2008), Borja compared the performance of the Newton method and the LATIN (large time increment) method in solving the nonlinear constitutive integration. In this study, the constitutive integration of continuum could be reduced to the MiCP (mixed complementarity problem) which is solved by the GSPC (Gauss-Seidel based projection contraction) algorithm (Zheng et al., 2020). Compared with Newton method, GSPC possesses high accuracy and it could be easily implemented because the inverse of Jacobian matrix is not needed. GSPC is obtained by improving the projection contraction algorithm (PCA) (He, 1997) by Gauss-Seidel iterative method. PCA has achieved much success in dealing with complementarity related problems including the simulation of static growth of multiple cracks (Zheng et al., 2015), the force method DDA (Zheng et al., 2016), and the constitutive integration for multiple yield surfaces problem (Zheng et al., 2020).

The evolution of shear band has been simulated by various numerical methods such as the assumed enhanced strain method (AES) (Simo et al., 1993, Borja and Regueiro, 2001), the extended finite element method (XFEM) (Sanborn and Prévost, 2011, Mikaeili and Liu, 2018, Liu, 2015), the numerical manifold method (Wu and Wong, 2013); etc. In AES, the enhancement equation is introduced to account for the localized elements and the elastic stiffness matrix is weakened with the extra unknowns condensed out (Borja and Regueiro, 2001). In XFEM, jump function is introduced for the elements containing discontinuities and the level set method is used to locate the shear band. The shortcoming of the XFEM is that the explicit representation of the discontinuity surface is needed by means of the level set, which adds complexity to this method. To circumvent this deficiency, cracking particles method (CPM) is proposed by Rabczuk in (Rabczuk and Belytschko, 2007, Rabczuk and Belytschko, 2010). CPM, a simplified meshfree method, simulates the growth of the discontinuity by the activation of a series of discontinuous surfaces which are defined at the corresponding particles. A major advantage of CPM is that this method could trace complex evolution of discontinuities because no continuity of the evolution path is necessary. In this paper by NMM, the shear band is treated as physical boundary, by cutting the mesh. Hence, similar to CPM, no explicit representation of the discontinuity surface is needed in NMM. Further, the critical advantage of NMM is that the displacement interpolation possesses high accuracy because the interpolation is provided by the regularly deployed the star nodes associated with each physical patch. While in XFEM or AES, the interpolation accuracy would be weakened because the element nodes used to interpolate would be distorted and severely irregular with the big growth of the shear band.

This paper is organized as follows. Section 2 briefly introduces the numerical manifold method (NMM) used in the simulation of shear bands. Section 3 revises the Drucker-Prager model by the SMP criterion with recourse to the transformed stress method, of which the constitutive integration can be reduced to the mixed complementarity problem. And how the shear band initiates and evolves is briefly introduced in this section. Section 4 modifies the contact treatment in NMM. Section 5 establishes the NMM formulation for the shear band of strain localization. Also, the displacement loading method is extended to the material nonlinearity accompanied with contact problem. Section 6 studies two typical examples, the soil block and the soil slope under displacement loading, by the proposed approach. And some conclusions are reached in Section 7.

Section snippets

Brief introduction to NMM

Fig. 1, representing a problem domain with a shear band inside, is used to introduce the basic concepts of NMM involved in this study. More aspects of NMM could be found in Yang et al., 2020, Yang et al., 2018, Yang et al., 2016, Wu et al., 2019a, Wu et al., 2019b, Wu et al., 2020, Zheng and Xu, 2014.The problem domain is denoted by Ω and the shear band is Γ. Mathematical cover can be generated by any type of meshes as long as these meshes cover Ω. In this case, the triangular finite element

The constitutive model and the initiation and evolution of shear band

The original Drucker-Prager (D-P) constitutive model is isotropic in terms of strength on the deviatoric plane, causing the strength of soil in general states not to be properly described. To overcome the deficiency, the transformed stress method (Sun et al., 1999) is applied to incorporate the SMP criterion into the D-P model. Subsequently, how the strain localization is initiated and evolved is briefly introduced.

Analysis for the contact interfaces

Inherited from DDA (discontinuous deformation analysis), the contact between the interfaces in NMM is treated as a collection of contact pairs. The contact is governed by one dimensional Mohr-Coulomb frictional law. The contact state for each contact pair could be in fixed state, slipping state or open state. For clarity, a brief introduction of contact state is given here. A contact pair comprises a vertex and edge. For an incremental load step, contact of fixed state means the vertex is fixed

Governing equations

For the convenience of presentation, it is specified here that all symbols used in Section 5 are in matrix or vector forms, even though some letters has already been used as tensors in Section 3.

The weak form of the boundary value problem with displacement discontinuities readsΩ(δε)TσdΩ=Ω(δu)Tb¯dΩ+Sp(δu)Tp¯dSp+Suβ(δu)T(u¯-u)dSu+Sc(δς)TτsdSc

Here δ is the variation operator. σis the stress vector. εis the strain vector. b¯, p¯ and u¯are respectively the prescribed body force vector in the

Numerical examples

In this section, two numerical examples are presented to demonstrate the capability of the new approach to capture the static evolution of shear band. Comparisons have been made between the influence of SMP-based DP model and the original DP model on the formation of shear band. Note that all the material parameters except hsc and μs used in the following two examples are based on those in (Borja, 2000, Regueiro and Borja, 2001, Mikaeili and Liu, 2018).

Conclusions and future work

This paper studies the effect of the SMP criterion on the formation of shear band and the load–displacement curves by using the transformed stress method to revise the D-P model by the SMP criterion. As expected, the results show that the DP model overestimates the yield strength in plane strain state than that by the SMP based criterion. Due to the direction of the major compressive stress, the shear band paths are very close for both criterions in the soil block example. But in the soil slope

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

The support of the National Natural Science Foundation of China (Grant Nos. 51538001 and 52079002) is gratefully acknowledged.

References (65)

  • P. Liu

    Extended finite element method for strong discontinuity analysis of strain localization of non-associative plasticity materials

    Int J Solids Struct

    (2015)
  • B. Loret et al.

    Finite element simulation of dynamic strain-localization: A multi-scale problem

    Comput Methods Appl Mech Eng

    (1995)
  • H. Matsuoka et al.

    Relationship among Tresca, Mises, Mohr-Coulomb and Matsuoka-Nakai failure criteria

    SOILS Found

    (1985)
  • H. Matsuoka et al.

    The Cam-Clay models revised by the SMP criterion

    J Japanese Geotech Soc Soils Found

    (1999)
  • E. Mikaeili et al.

    Numerical modeling of shear band propagation in porous plastic dilatant materials by XFEM

    Theor Appl Fract Mech

    (2018)
  • J. Oliver et al.

    Strong discontinuities and continuum plasticity models: The strong discontinuity approach

    Int J Plast

    (1999)
  • M. Ortiz et al.

    A finite element method for localized failure analysis

    Comput Methods Appl Mech Eng

    (1987)
  • N.S. Ottosen et al.

    Properties of discontinuous bifurcation solutions in elasto-plasticity

    Int J Solids Struct

    (1991)
  • T. Rabczuk et al.

    A three-dimensional large deformation meshfree method for arbitrary evolving cracks

    Comput Methods Appl Mech Eng

    (2007)
  • T. Rabczuk et al.

    Discontinuous modelling of shear bands using adaptive meshfree methods

    Comput Methods Appl Mech Eng

    (2008)
  • R.A. Regueiro et al.

    Plane strain finite element analysis of pressure sensitive plasticity with strong discontinuity

    Int J Solids Struct

    (2001)
  • J.W. Rudnicki et al.

    Conditions for the localization of deformation in pressure-sensitive dilatant materials

    J Mech Phys Solids

    (1975)
  • K. Runesson et al.

    Discontinuous bifurcations of elastic-plastic solutions at plane stress and plane strain

    Int J Plast

    (1991)
  • G.N. Wells et al.

    Analysis of slip planes in three-dimensional solids

    Comput Methods Appl Mech Eng

    (2001)
  • H. Wolf et al.

    Experimental investigation of shear band patterns in granular material

    J Struct Geol

    (2003)
  • W. Wu et al.

    Enriched three-field numerical manifold formulation for dynamics of fractured saturated porous media

    Comput Methods Appl Mech Eng

    (2019)
  • W. Wu et al.

    Enriched mixed numerical manifold formulation with continuous nodal gradients for dynamics of fractured poroelasticity [J]

    Appl Math Modell

    (2020)
  • Y.T. Yang et al.

    Hydraulic fracturing modeling using the enriched numerical manifold method

    Appl Math Modell

    (2018)
  • Y.T. Yang et al.

    Three-dimensional fracture propagation with numerical manifold method

    Eng Anal Bound Elem

    (2016)
  • Y.T. Yang et al.

    Modeling the entire progressive failure process of rock slopes using a strength-based criterion

    Comput Geotech

    (2020)
  • H. Zheng et al.

    Complementarity problem arising from static growth of multiple cracks and MLS-based numerical manifold method

    Comput Methods Appl Mech Eng

    (2015)
  • H. Zheng et al.

    Dual form of discontinuous deformation analysis

    Comput Methods Appl Mech Eng

    (2016)
  • Cited by (10)

    • Investigation of the shear band evolution in soil-rock mixture using the assumed enhanced strain method with the meshes of improved numerical manifold method

      2022, Engineering Analysis with Boundary Elements
      Citation Excerpt :

      Furthermore, the contact technique, used for the contact interface, is mature in NMM. Therefore, the NMM has been preferably used to solve many geotechnical problems [42–49]. Notice that the SRM has extremely heterogeneous characters because the distinct properties between soil and rock.

    • Influence of soil non-homogeneity on shear band evolution investigated by the numerical manifold method combined with the assumed enhanced strain method

      2021, Engineering Analysis with Boundary Elements
      Citation Excerpt :

      Besides, this assumption is the basis of such well-known theories as Coulomb's [15] and Rankine's [16] methods for earth pressure. Nowadays, the evolution of strong discontinuity shear band or crack has been simulated by many kinds of numerical techniques, including the numerical manifold method (NMM) [17,18], the extended finite element method (XFEM) [19,20], the slipped particle method [21], the assumed enhanced strain (AES) method [13,22], and the enriched element free Galerkin (XEFG) [23–25]. The XFEM accounts for the shear band by introducing Heaviside function into the approximating function and enriched degrees of freedom are generated.

    View all citing articles on Scopus
    View full text