Scribd Logo
Upload

Welcome to Scribd!

  • 25 views

    Implementation of Linear Viscoelasticity Model in Pdlammps

    Uploaded by

    Saeed Abd
    This document summarizes the implementation of a linear viscoelasticity model in PDLAMMPS. It describes how the total extension state is decomposed into volumetric and deviatoric components to account for viscoelastic effects. It details the algorithm, including how the peridynamic force state is modified to include a back stress that depends on material properties like lambda and tau_b. The implementation includes adding a new pair style "peri/ves" and modifying existing codes to include arrays that store deviatoric extension and back extension at each timestep to incorporate viscoelasticity into the peridynamic formulation.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Implementation of Linear Viscoelasticity Model in Pdlammps

    Uploaded by

    Saeed Abd
    25 views7 pages
    This document summarizes the implementation of a linear viscoelasticity model in PDLAMMPS. It describes how the total extension state is decomposed into volumetric and deviatoric components to account for viscoelastic effects. It details the algorithm, including how the peridynamic force state is modified to include a back stress that depends on material properties like lambda and tau_b. The implementation includes adding a new pair style "peri/ves" and modifying existing codes to include arrays that store deviatoric extension and back extension at each timestep to incorporate viscoelasticity into the peridynamic formulation.

    Original Description:

    PDMS

    Original Title

    pdbs

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document summarizes the implementation of a linear viscoelasticity model in PDLAMMPS. It describes how the total extension state is decomposed into volumetric and deviatoric components to account for viscoelastic effects. It details the algorithm, including how the peridynamic force state is modified to include a back stress that depends on material properties like lambda and tau_b. The implementation includes adding a new pair style "peri/ves" and modifying existing codes to include arrays that store deviatoric extension and back extension at each timestep to incorporate viscoelasticity into the peridynamic formulation.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    25 views7 pages

    Implementation of Linear Viscoelasticity Model in Pdlammps

    Uploaded by

    Saeed Abd
    This document summarizes the implementation of a linear viscoelasticity model in PDLAMMPS. It describes how the total extension state is decomposed into volumetric and deviatoric components to account for viscoelastic effects. It details the algorithm, including how the peridynamic force state is modified to include a back stress that depends on material properties like lambda and tau_b. The implementation includes adding a new pair style "peri/ves" and modifying existing codes to include arrays that store deviatoric extension and back extension at each timestep to incorporate viscoelasticity into the peridynamic formulation.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 7

    Implementation of linear viscoelasticity model

    in PDLAMMPS
    R. Rahman and J.T. Foster
    Center for Simulation, Visualization and
    Real-time prediction (SiViRt)
    The University of Texas at San Antonio
    San Antonio, TX 78249
    Email: [email protected],
    [email protected]
    September 12, 2014

    1 Introduction
    In this documentation the discussion is focused on implementation of linear
    viscoelasticity model in PDLAMMPS [PLPS08],[Pli95]. The peridynamic
    viscoelasticity formulation (and its time integration) used in this work was
    developed by John Mitchell at Sandia National Lab [Mit11]. In the PD-
    LAMMPS a new pair-peri-style is added in order to incorporate the viscoelas-
    ticity. The new source codes pair-peri-ves.cpp and pair-peri-ves.h are introduced. Besi
    peri-neigh.h are modified for introducing viscoelasticity.

    2 Algorithm and implementation


    In order to include the viscoelasticity the total extension state can be de-
    composed into two parts [Mit11] 1 :
    1
    For detail please see the document on implementation of viscoelasticity in state based
    peridynamics model [Mit11].

    1
    Total extension: e (Y ) = ei (Y ) + ed (Y ) (1)
    θ (Y ) |X|
    Volumetric extension: ei (Y ) = (2)
    3
    θ (Y ) |X|
    Deviatoric extension: ed (Y ) = |Y | − |X| − (3)
    3
    Here, |Y |, |X| and θ are the reference state, defromation state and dilation
    state, respectively. The deviatoric extension ca be written as:

    ed (Y ) = ede (Y ) + edb(i) (Y ) (4)


    Here, ede (Y ) and edb(i) are the elastic and back parts of the deviatoric
    extension. Considering viscoelastic effect the peridynamic force state can be
    written as:
    3p
    t = ti + td = − ωx + (α∞ + αi ) ed − αi ωedb (5)
    m
    p, k, αi , ti and td are hydrostatic pressure, bulk modulus, elastic properties
    volumetric and deviatoric scalar force states, respectively. α∞ + αi = 15µ m
    ;
    where, µ, m are the shear modulus and weighted volume, respectively. The
    1
    influence function: ω hξi = kξk , kξk is the scalar reference state. At the
    current or tn+1 timestep the scalar force state based on LPS is:
     
    i θ (i) θ (i)
    tn+1 = 3k ω+ + ω− eν (xj − xi ) Vj (6)
    m (i) m (i)
    Based on viscoelasticity the deviatoric scale force state can be written as:

     
    ω+ ω−
    tdn+1 = 15 (1 − λ) µ + eν (xj − xi ) Vj
    m (i) m (i)
     
    ω+ ω−
    e − edb
    
    + 15µλ + n+1 ν (xj − xi ) Vj (7)
    m (i) m (i)

    Here, Vj , edb
    n+1 , ν (xj − xi ) are volume fraction, back extension at current
    timestep and particle volume scaling factor, respectively. λ varies within zero
    to one. The back extension at current state can be calculated as 2 [Mit11]:
    2
    https://software.sandia.gov/trac/peridigm/

    2
    edb d db
    n+1 = (1 − decay) en + decay · en + β∆e
    d
    (8)
    ∆ed = edn+1 − edn (9)
     
    dt
    decay = exp − (10)
    τb
    τb
    β = 1 − (1 − decay) (11)
    dt
    At any current timestep (i.e. n + 1), for a bond between ith and j th PD
    nodes, the deviatoric extension is stored on the fly. The value is stored in
    deviatorextension[i][j]. Similarly, the back extension is also stored in
    deviatorBackextension[i][j]. Both of the arrays initialized with zero.
    In the fix_peri_neigh.cpp an integer flag based switch is used in order
    to trigger the LPS, PMB or VES (Viscoelastic solid) models. The class
    FixPeriNeigh looks like:

    FixPeriNeigh::FixPeriNeigh(LAMMPS *lmp,int narg, char **arg) :


    Fix(lmp, narg, arg)
    {

    //Get the pair information


    Pair *anypair01 = force->pair_match("peri/pmb",1);
    Pair *anypair02 = force->pair_match("peri/lps",1);
    Pair *anypair03 = force->pair_match("peri/ves",1);

    isPMB = 0; //Check if PMB


    isLPS = 0; //Check if LPS
    isVES = 0; //Check if VES

    //Select the flag based on which model is being used


    if (anypair01 != NULL)
    {
    isPMB=1;
    }
    else if (anypair02 != NULL)
    {
    isLPS=1;

    3
    }
    else
    {
    isVES=1;
    }

    restart_global = 1;
    restart_peratom = 1;
    first = 1;

    // perform initial allocation of atom-based arrays


    // register with atom class
    // set maxpartner = 1 as placeholder

    maxpartner = 1;
    npartner = NULL;
    partner = NULL;
    if (isVES == 1){
    deviatorextention = NULL;
    deviatorBackextention = NULL;
    }
    r0 = NULL;
    vinter = NULL;
    wvolume = NULL;

    grow_arrays(atom->nmax);
    atom->add_callback(0);
    atom->add_callback(1);

    // initialize npartner to 0 so atom migration is OK the 1st time

    int nlocal = atom->nlocal;


    for (int i = 0; i < nlocal; i++) npartner[i] = 0;

    // set comm sizes needed by this fix

    comm_forward = 1;
    }

    4
    There are three integer flags isPMB , isLPS, isVES .PMB, LPS and VES
    stand for bond based PD model, linear peridynamic solid and viscoelastic
    peridynamic solid, respectively. Based on the input file any one of these flags
    are set to one based on:

    Pair *anypair01 = force->pair_match("peri/pmb",1);


    Pair *anypair02 = force->pair_match("peri/lps",1);
    Pair *anypair03 = force->pair_match("peri/ves",1);

    The stored values are used in:

    PairPeriVES::compute(int eflag, int vflag)

    double PairPeriVES::single(int i, int j, int itype, int jtype,


    double rsq, double factor_coul,
    double factor_lj,double &fforce)

    At any current step the total scalar force state t is calculated and the
    current deviatoric extension and back extension are stored in order to use
    in next step. λ and time constant τb are material dependent parameters.
    These are defined in the LAMMPS input script. This will be discussed in
    the next section. The new LAMMPS pair style {pair_peri_ves}is defined
    with addition arguments in the input file (λ and τb ):

    void PairPeriVES::settings(int narg, char **arg)


    {
    if (narg) error->all(FLERR,"Illegal pair_style command");
    }

    /* ----------------------------------------------------------------------
    set coeffs for one or more type pairs
    ------------------------------------------------------------------------- */

    void PairPeriVES::coeff(int narg, char **arg)


    {
    if (narg != 9) error->all(FLERR,"Incorrect args for pair coefficients");
    if (!allocated) allocate();

    5
    int ilo,ihi,jlo,jhi;
    force->bounds(arg[0],atom->ntypes,ilo,ihi);
    force->bounds(arg[1],atom->ntypes,jlo,jhi);

    double bulkmodulus_one = atof(arg[2]);


    double shearmodulus_one = atof(arg[3]);
    double cut_one = atof(arg[4]);
    double s00_one = atof(arg[5]);
    double alpha_one = atof(arg[6]);
    double mlambdai_one = atof(arg[7]); // New
    double mtaui_one = atof(arg[8]); //New

    int count = 0;
    for (int i = ilo; i <= ihi; i++) {
    for (int j = MAX(jlo,i); j <= jhi; j++) {
    bulkmodulus[i][j] = bulkmodulus_one;
    shearmodulus[i][j] = shearmodulus_one;
    cut[i][j] = cut_one;
    s00[i][j] = s00_one;
    alpha[i][j] = alpha_one;
    m_lambdai[i][j] = mlambdai_one; // New
    m_taubi[i][j] = mtaui_one; // New
    setflag[i][j] = 1;
    count++;
    }
    }

    if (count == 0) error->all(FLERR,"Incorrect args for pair coefficients");


    }

    3 LAMMPS command for PD VES


    There is no significant change in the LAMMPS input script for viscoelastic
    model. For PD VES the LAMMPS commands are:

    pair_style peri/ves
    pair_coeff i j Bulk_modulus Shear_modulus s00 alpha lambda tau

    6
    The LAMMPS compilation with the VES is same. The updated Install.sh
    should be used.

    4 Conclusion
    The LAMMPS implementation of peridynamic viscoelastic model is still
    in beta phase. Any bug or issue can be informed to the authors through
    [email protected] .

    5 Acknowledgment
    The authors are thankful to Steve Plimpton and S. M. Raiyan Kabir for
    their useful suggestion regarding the code development for PDLAMMPS im-
    plementation of viscoelasticity.

    References
    [Mit11] John A. Mitchell. A nonlocal, ordinary, state-based viscoelas-
    ticity model for peridynamics. Technical Report SAND2011-
    8064, Sandia National Laboratories, May 2011. Available at
    https://cfwebprod.sandia.gov/cfdocs/CompResearch/docs/SAND2011-Viscoelast

    [Pli95] S. Plimpton. Fast parallel algorithms for short-range


    molecular dynamics. J Comp Phys, 117:1–19, 1995.
    http://lammps.sandia.gov.

    [PLPS08] Michael L. Parks, Richard B. Lehoucq, Steven J. Plimpton, and


    Stewart A. Silling. Implementing peridynamics in molecular dy-
    namics code. Computer physics communication, 179:777–783,
    2008.

    You might also like