C. T. Wu ctwu@lstc.

com
Wei Hu* whu@lstc.com
Yong Guo yguo@lstc.com
Bo Ren boren@lstc.com
Youcai Wu ycwu@lstc.com
Xiaofei Pan xfpan@lstc.com

Manufacturing and Material Failure Analyses

Using Advanced FEM and Meshfree Methods
LS-DYNA® Training Class, 07/18/2017, Shanghai
▪ Download links of class material
▪ http://ftp.lstc.com/anonymous/outgoing/whu/Class/AdvFem_Meshfree_2017Workshop_Hengstar.zip

▪ Related links
▪ http://ftp.lstc.com/anonymous/outgoing/ycwu/SPGdocs/IntroSPG.ppsx
▪ http://ftp.lstc.com/outgoing/boren/perioutgoing/102016/peridynamicpackage.zip

▪ Journal articles and conference papers are available upon request

Resources 2 / 138
▪ Non-destructive manufacturing (page 6-74)
▪ Forging, extrusion, spring back, compression molding
▪ Materials: metal, composite
▪ Large deformation, coupled thermal-mechanical, implicit and explicit

▪ Numerical approaches Courtesy of JSOL

▪ Adaptive FEM/EFG (element free Galerkin)

▪ Adaptive procedure
Remeshing: mesh quality & refinement, surface representation
Remapping: transfer internal variables through remapping functions
▪ FEM: fast and robust, low order method
▪ EFG: highly accurate, enhanced capabilities, slower and memory demanding

▪ SPG (smooth particle Galerkin)

▪ Particle based truly meshfree method
Computationally stable and efficient to handle extreme large deformation
Only available in LS-DYNA

Manufacturing and material failure analysis (1) 3 / 138

▪ Destructive manufacturing (page 75-113)
▪ Grinding, riveting, cutting, flow drill screw,
Self-pierce riveting, self-tapping screw, drilling
▪ Materials: metal, composite
▪ Workpiece-tool interaction and stress analysis

▪ Numerical approaches
▪ SPG (smooth particle Galerkin)
▪ Multiple kernel options for different types of material failure behavior
▪ Ability to help prevent “numerical failure”
▪ Flexibility to incorporate different failure indicators including those in material laws

▪ Peridynamics
▪ Bond-based version: brittle and composite laminate material Inner-layer failure
▪ Discontinuous Galerkin framework: use FEM model as input
▪ Laminate Peridynamics model: inner-layer and inter-layer bonds

Delamination

Manufacturing and material failure analysis (2) 4 / 138

▪ Material failure and fragmentation analysis (page 114-135)
▪ Impact and penetration, 3D crack propagation in solid
Delamination and folding, fragmentation
▪ Material: metal, concrete, rock and soil, composite, glass type brittle material
▪ Material fragmentation due to dynamic wave propagation

▪ Numerical approaches
▪ SPG (smooth particle Galerkin)
▪ Ability to handle flying particles
▪ Particle-based BODY-TO-BODY and SELF contacts
▪ Support EOS materials

▪ Peridynamics
▪ Bond-based version: brittle material
▪ Ability to effectively capture dynamic propagation of multiple cracks

Manufacturing and material failure analysis (3) 5 / 138

▪ Meshfree Galerkin method
▪ Meshfree approximation: accuracy and smoothness
▪ Numerical integration: mesh-based and node-based
▪ EFG and its keyword *SECTION_SOLID_EFG Non-destructive
▪ SPG and its keyword *SECTION_SOLID_SPG manufacturing
▪ 3D solid adaptivity
▪ General introduction Forging, extrusion, spring back,
compression molding
▪ Main features
▪ *CONTROL_ADAPTIVE and *CONTROL_REMESHING
▪ Mesh size: local refinement, *DEFINE_BOX_ADAPTIVE
▪ Orbital remeshing
▪ Adaptive EFG enhanced features

▪ Constraints and contacts

Fiber-reinforced solid *CONSTRAINED_BEAM_IN_SOLID
▪ Implicit analysis, thermal coupling, spring back
▪ Run-time control files for adaptive procedure
Manual remeshing, controls on multiple adaptive parts
▪ Construction of approximation (shape) functions based on discrete nodes
▪ Discrete nodes and associated supports covering the computational domain
▪ Bell-shape functions: locality and smoothness
▪ Continuous function and its derivative: smooth results in displacement and stain/stress fields

Meshfree shape function

Meshfree approximation (1) 7 / 138

▪ Meshfree Galerkin approach
▪ Each nodal point represents a
chunk of material
▪ 1st order shape function
▪ Galerkin weak formulation
▪ Solid material (continuum)
Continuum
(deformation, stress)
Meshfree approximation (2)
▪ SPH (smooth particle hydrodyn)
▪ Each nodal point represents a
chunk of material
▪ Low order approximation
▪ Collocation method
▪ Fluid / EOS (Diffusive)
▪ DEM (discrete element method)
▪ Deformable particles with
various sizes interact through
bonds and contacts
▪ Particle dynamics
▪ Grangular material (Discrete)
Discrete
(momentum, EOS)
8 / 138
▪ Lagrangian kernel (EFG)
▪ Support is defined in the initial configuration
▪ Support covers the same set of material
points throughout time
▪ Neighbor searching only once (initialization)

▪ Semi-pseudo Lagrangian kernel (SPG)

▪ Support is defined in the initial configuration
▪ Remove neighboring points from support due
to “bond” failure

▪ Eulerian kernel (SPG) ▪ Updated Lagrangian kernel (SPG)

▪ Support is defined in the current configuration
▪ Support covers different material points
throughout time
▪ Neighbor searching regularly
▪ Support size is modified according to
deformation to help prevent numerical failure
▪ Support is defined in the current configuration
▪ Support covers ABOUT the same number of
material points throughout time
▪ Neighbor searching regularly
▪ Adaptive EFG
(1) remesh: redefine material points
(2) new mesh: new reference (“initial”)
configuration for defining neighbor

Meshfree kernel 9 / 138

▪ EFG accepts FEM mesh as background integration mesh
▪ ELFORM=41: TET, HEX, PENT (3D orbital adaptivity)
▪ ELFORM=42: TET (optimized in efficiency, 3D general TET adaptivity)
▪ Pros: define physical domain, use various contact types, compute nodal mass,
Tetrahedron (TET)
impose boundary conditions, create stress points
▪ Cons: mesh distortion issue, difficult in failure analysis

▪ SPG accepts FEM mesh as input

▪ Nodal mass and support based on TET, HEX, PENT
▪ Pros: truly meshfree, failure analysis Hexahedron (HEX)

SPG nodal
integration points
e
EFG stress
(integration) points
Pentahedron (PENT)

Spatial domain integration in Galerkin weak formulation 10 / 138

▪ EFG background mesh distortion in severe deformation
▪ Negative volume error Ω0 Particle

▪ Beyond Lagrangian Description

▪ More tolerant on mesh distortion than conventional FEM Stress point

▪ Severe deformation: SPG and adaptive EFG

▪ Material failure and separation: SPG

x   ( X , t) •Is continuously differentiable

•Is one-to-one
dx  F  dX •J >0
J  det( F ) •F is invertible
Lagrangian

Ωt
J>0
NO material overlapping
NO gap
NO negative volume

Negative volume issue of EFG 11 / 138

▪ EFG approximation function for displacement
NP
u (x)   wa[n] (x  xI ) u I
h

I 1
T
wa[n] (x  xI )  H [n] (0) M [n] ( x) H [n] (x - xI )wa (x  xI )
-1

  
n th order completeness kernel function

▪ Basis function: 1st order accuracy

Conservation of linear and angular momentum
Numerical convergence
▪ Kernel function: locality and smoothness
▪ NOT interpolation function: u ( xI )  u I , need special boundary treatment
h

▪ Discrete equation of motion under Galerkin framework

A-T MA-1Dd + A-T KA-1Dd = -A-T R

▪ Stable and NO low energy mode (NO need for hour-glass control)
▪ More CPU time and memory demanding than FEM

Overview of EFG 12 / 138

▪ Meshfree nodal integration
▪ Direct nodal integration leads to numerical instability
▪ Stabilized nodal integration suitable for industry application is challenging

▪ SPG introduces smoothed displacement field

▪ Perform kernel update regularly to avoid negative volume issue
▪ Consistent numerical stabilization
NO problem-dependent stabilization parameter
▪ Provide numerical regularization
▪ Extremely large deformation and failure analysis
Redefine the neighbor and re-compute the approximation
Maintain time step size in explicit analysis
▪ Generally needs larger support size than EFG
More CPU time and memory demanding than EFG

Overview of SPG 13 / 138

▪ *SECTION_SOLID_EFG
▪ Card 1 ELFORM EQ.41: EFG solid (TET, HEX, PENT mesh)
Variable SECID ELFORM
EQ.42: EFG solid (TET mesh)
Type I I SMP & MPP
Default Implicit and explicit analysis
Thermal-mechanical coupling
Available from R5

▪ Card 2
Variable DX DY DZ ISPLINE IDILA IEBT IDIM TOLDEF
Type F F F I I I I F
Default 1.01 1.01 1.01 0 0 -1 2 0.01

▪ Card 3
Variable IPS STIME IKEN SF MID IBR DS ECUT
Type I F For fracture
I F analysis
I & other
I features
F F
Default 0 1.0E20 0 0.0 0 1 1.01 0.1

EFG solid keyword (1) 14 / 138

▪ *SECTION_SOLID_EFG
▪ Card 2
Variable DX DY DZ ISPLINE IDILA IEBT IDIM TOLDEF
Type F F F I I I I F
Default 1.01 1.01 1.01 0 0 -1 2 0.01

▪ Normalized dilation parameters (normalized support size) in X, Y and Z directions

1.0 < DX, DY, DZ < 1.6 is recommended
CPU time and memory usage increase with support size

Regular mesh Irregular mesh

Foam 1.0 ~ 1.2 1.0 ~ 1.2
Metal 1.2 ~ 1.4 1.0 ~ 1.2
Fluid or EOS 1.4 ~ 1.6 1.2 ~ 1.4

EFG solid keyword (2) 15 / 138

▪ *SECTION_SOLID_EFG
▪ Card 2
Variable DX DY DZ ISPLINE IDILA IEBT IDIM TOLDEF
Type F F F I I I I F
Default 1.01 1.01 1.01 0 0 -1 2 0.01

▪ Kernel function
ISPLINE EQ.0: Cubic spline function with base function (1, x, y, z )
EQ.1: Quadratic spline function
EQ.2: Cubic spline function (circular support, DX defines radius coefficient) Cubic B-spline

EQ.10: Cubic spline function with bilinear base function (1, x, y, z, xy, yz, xz )

EFG solid keyword (3) 16 / 138

▪ *SECTION_SOLID_EFG
▪ Card 2
Variable DX DY DZ ISPLINE IDILA IEBT IDIM TOLDEF
Type F F F I I I I F
Default 1.01 1.01 1.01 0 0 -1 2 0.01

▪ Essential boundary condition treatment

IEBT EQ. 1: Full transformation
EQ.–1: (w/o transformation)
EQ. 2: Mixed transformation
EQ. 3: Coupled FEM/EFG (most efficient) with transformation
EQ. 4: Fast transformation
EQ. 7: Maximum Entropy approximation
Stand shape function

maximum entropy

EFG solid keyword (4) 17 / 138

▪ *SECTION_SOLID_EFG
▪ Card 2
Variable DX DY DZ ISPLINE IDILA IEBT IDIM TOLDEF
Type F F F I I I I F
Default 1.01 1.01 1.01 0 0 -1 2 0.01

▪ Spatial domain integration

IDIM EQ. 1: Local boundary condition method
EQ. 2: Gauss integration
EQ.-1: Stabilized EFG method (apply to PENT and HEX background mesh)
One-point integration scheme + gradient type hourglass control
Designed especially for foam and rubber materials
Computational cost is between reduced integration FEM (#1) and full integration FEM (#2)
EQ.-2: Fractured EFG method (apply to TET, SMP only)

EFG solid keyword (5) 18 / 138

▪ *SECTION_SOLID_EFG
▪ Card 3
Variable IPS STIME IKEN SF MID IBR DS ECUT
Type I F I F I I F F
Default 0 1.0E20 0 0.0 0 1 1.01 0.1

▪ Pressure smoothing
IPS EQ. 0: NO pressure smoothing
Before
EQ. 1: Moving-least squared pressure recovery
▪ ONLY works for EFG ELFORM=42 (TET mesh)
▪ Material behaves slightly more compressible
▪ NOT recommended if After

(1) There is numerical instability in contact (“hard contact”)

(2) Implicit analysis has convergence problem

EFG solid keyword (6) 19 / 138

▪ Crashworthiness
▪ Large deformation of bumper
▪ Explicit analysis
▪ EFG has ability to perform full analysis
▪ EFG captures the peak impact force Courtesy of GM

Impact Force

FEM + high
viscosity

FEM EFG

EFG solid applications (1) 20 / 138

▪ Dummy model
▪ EFG for hyper-elastic jacket and foam ribs

Internal Energy

Courtesy of GM

Total 84583 nodes Ribs Jacket

15966 EFG nodes

EFG solid applications (2) 21 / 138

▪ Material design
▪ Particle-reinforced (nano-scale) rubber material
▪ Unit cell model with periodic boundary conditions
▪ Up to 100% ~ 200% effective strain
▪ Implicit analysis

ME-FEM
EFG-PS
EFG

Courtesy of YRC

EFG solid applications (3) 22 / 138

▪ *SECTION_SOLID_SPG
▪ Card 1 ELFORM EQ.47: SPG solid (TET, HEX, PENT mesh)
Variable SECID ELFORM
SMP & MPP
Type I I Explicit analysis
Default Thermal-mechanical coupling* (SMP)
Available from R9, recommend R10 & Beta

▪ Card 2
Variable DX DY DZ ISPLINE KERNEL LSCALE SMSTEP SUKTIME
Type F F F I I F I F
Default 1.5 1.5 1.5 0 0 15

▪ Card 3
Variable IDAM FS STRETCH ITB
Type I For
F failure
F analysis
I & other features
Default 0 0

SPG solid keyword (1) 23 / 138

▪ *SECTION_SOLID_SPG
▪ Card 2 Variable DX DY DZ ISPLINE KERNEL LSCALE SMSTEP SUKTIME
Type F F F I I F I F
Default 1.5 1.5 1.5 0 0 15

▪ Normalized dilation parameters (normalized support size) in X, Y and Z directions

1.4 < DX, DY, DZ < 2.0 is recommended
▪ Ductile material with severe deformation: larger support size
▪ Semi-brittle material with local deformation: smaller support size
▪ Generally larger support size is numerically more stable
▪ CPU time and memory usage increase with support size

SPG solid keyword (2) 24 / 138

▪ *SECTION_SOLID_SPG
▪ Card 2 Variable DX DY DZ ISPLINE KERNEL LSCALE SMSTEP SUKTIME
Type F F F I I F I F
Default 1.5 1.5 1.5 0 0 15

Initial
▪ Kernel types support

KERNEL EQ.0: Updated Lagrangian kernel (NO failure, less shear deformation)
EQ.1: Eulerian kernel (failure analysis, global extreme deformation)
EQ.2: Semi-pseudo Lagrangian kernel (failure analysis, local extreme)
current
▪ Neighbors are redefined and approximation functions are re-computed support

Updated Lag kernel: the new support (circular/rectangular) in the current

configuration at least covers all the neighboring points
found in the initial configuration
Rubber material, foam
Eulerian kernel: the new support in the current configuration is adjusted
according to the current deformation
Ductile and semi-brittle material

SPG solid keyword (3) 25 / 138

▪ *SECTION_SOLID_SPG
▪ Card 2 Variable DX DY DZ ISPLINE KERNEL LSCALE SMSTEP SUKTIME
Type F F F I I F I F
Default 1.5 1.5 1.5 0 0 15

▪ Interval of time step to perform kernel update and displacement smoothing

▪ (Time step size) * SMSTEP = CONSTANT
TSSFAC (*CONTROL_TIMESTEP): recommended value (0.1 ~ 0.3) if controlled by SPG parts
Increase TSSFAC, decrease SMSTEP, and vice versa.
▪ Smaller SMSTEP: numerically more stable, less accurate
Larger SMSTEP: numerically less stable
▪ Default value works for most of extreme large deformation and failure analyses

SPG solid keyword (4) 26 / 138

▪ Punch test
▪ Extreme large deformation
▪ Explicit analysis
▪ NO material failure
▪ Time step size does NOT drop
Updated Lagrangian
SPG solid: numerical examples (1-1)
d
Particle plotting in LS-PrePost for EFG & SPG
• Select Model and Part
• In Appearance, check Sphere and Shrn for
target parts
• Select Settings
• In General settings, check SPH/Particle and
set appropriate values of Radius and Divs,
choose Smooth as Style, check Fixed Radius
Eulerian kernel
27 / 138
▪ Punch test
▪ NO material failure
▪ NO Zero-energy mode and numerical oscillation
▪ Time step size does NOT drop

EPS Updated Lagrangian

EPS Eulerian (Tensile instability)

SPG solid: numerical examples (1-2) 28 / 138

▪ Taylor bar impact
▪ SPG: updated Lagrangian kernel
EPS
▪ Explicit analysis (FEM)
▪ SPG: time step size does NOT drop
FEM: time step size drops by one order

Ø7.62x23.5mm
11520 hexes
Al: 6061-T6
12985 nodes
Johnson-Cook
EPS
(SPG)

-373m/s

SPG solid: numerical examples (2) 29 / 138

▪ Rigid ball impacting plate
▪ SPG: updated Lagrangian kernel
▪ NO material failure t=0.02
∆t=7.47×10-6
▪ Explicit analysis

t=0.04
∆t=7.44×10-6
Ball: Radius=5.0
Plate: Radius=20.0, Thickness=5.0
25721 nodes

Elastic perfectly-plastic material

ρ0=7.85×10-3, E=6.9x104, v=0.3
σy=200.0, Vz= -600.0
t=0.08
Bottom view ∆t=7.39×10-6

SPG solid: numerical examples (3) 30 / 138

 main routine Adaptive control
▪ 3D solid adaptivity ADPFREQ
Interactive control
▪ Achieves high accuracy
Use least number of adaptive steps
▪ Represents complex surface (1) to avoid distortion related issues Remesh surface
(2) to improve convergence in triangulation
▪ Captures high gradient field
implicit analysis
▪ Hot forging and extrusion, cold forming, (3) to improve contact penetration Generate 3D mesh
and surface representation based on new surface
Molding, riveting, rolling, tapping triangulation
▪ Current implementation (available in R6, recommend Beta)
▪ Adaptive FEM/EFG I/O output all the
Restart information of old
▪ R-adaptivity: remesh the entire adaptive part model and new mesh
Expensive
▪ Two remeshers: general 3D TET based, orbital PENT/HEX based
▪ Multiple adaptive parts I/O input and re-
▪ Explicit and implicit analysis initialization everything

▪ Thermal-mechanical coupling
Remap
▪ Spring back analysis
from old to new mesh
▪ SMP and MPP FEM: low order
▪ NO material failure, limited capability for material separation EFG: high order
Numerical error
Switch to adaptive SPG for material failure (future development)

Adaptivity introduction 31 / 138

▪ *CONTROL_ADAPTIVE across all adaptive parts
▪ Card 1 Variable ADPFREQ ADPTOL ADPOPT MAXLVL TBIRTH TDEATH LCADP IOFLAG
Type F F I I F F I I
Default 1.0E20 1 3 0.0 1.0E20 0 0.01

▪ 3D solid adaptivity
ADPOPT EQ.7: 3D solid adaptivity
ADPOPT (card 2 in *PART): mark the adaptive parts
EQ.2: 3D general TET remesher, FEM(ELFORM=1,13), EFG(ELFORM=41,42)
EQ.3: 3D orbital PENT/HEX remesher, FEM(ELFORM=1), EFG(ELFORM=41)
▪ General control on adaptive steps
ADPFREQ : time interval between adaptive steps
TBIRTH : birth time at which the adaptive remeshing begins
TDEATH : death time at which the adaptive remeshing ends
LCADP : curve ID to define the adaptive intervals as a function of time

Adaptivity keyword (1) 32 / 138

▪ *CONTROL_ADAPTIVE across all adaptive parts
▪ Card 2
Variable ADPSIZE ADPASS IREFLG ADPENE ADPTH MEMORY ORIENT MAXEL
Type F I I F F I I I
Default 0 0 0.0 0

▪ Card 4
Variable CNLA MMM2D ADPERRR D3TRACE IADPFC IFSAND
Type F I I I I I
Default 0 0 0 0 0 0

Adaptivity keyword (2) 33 / 138

▪ *CONTROL_REMESHING(_EFG) across all adaptive parts
▪ Card 1 Variable RMIN RMAX VF_LOSS MFRAC DT_MIN ICURV CID SEGANG
Type F F F F F I I F
Default 1.0 0.0 0.0 4 0 0.0

▪ Card 2 (_EFG)

Variable IVT IAT IAAT IER MM

Type I I I I I
Default 1 0 0 0 0

▪ Card 3 (_EFG)

Variable IAT1 IAT2 IAT3

Type F F F
Default 1.0E20 1.0E20 1.0E20

Adaptivity keyword (3) 34 / 138

▪ *CONTROL_REMESHING(_EFG)
▪ Card 1 Variable RMIN RMAX VF_LOSS MFRAC DT_MIN ICURV CID SEGANG
Type F F F F F I I F
Default 1.0 0.0 0.0 4 0 0.0

▪ RMIN defines the globally minimum mesh size for all adaptive parts
▪ RMAX defines the globally maximum mesh size for all adaptive parts
▪ Global remeshing (local refinement is turned off)
RMAX/RMIN ~ 3.0: better mesh quality and robustness in remeshing Global remeshing Local refinement
▪ Local refinement (driven by contact surface curvature)
RMIN: the mesh size to well represent fine features of contact surface
RMAX: the maximum mesh size for least deformed area to speed up
the analysis

Adaptive remesh: mesh size (1) 35 / 138

▪ *CONTROL_REMESHING(_EFG)
▪ Card 1 Variable RMIN RMAX VF_LOSS MFRAC DT_MIN ICURV CID SEGANG
Type F F F F F I I F
Default 1.0 0.0 0.0 4 0 0.0

▪ ICURV defines the number of elements along the radius

▪ ONLY supporting 3D general TET remesher

• Smoothing process for

soft edges
• NO smoothing for hard
edges, feature lines

Adaptive remesh: mesh size (2) 36 / 138

▪ *CONTROL_ADAPTIVE (major update in R9)
▪ Card 2
Variable ADPSIZE ADPASS IREFLG ADPENE ADPTH MEMORY ORIENT MAXEL
Type F I I F F I I I
Default 0 0 0.0 0

▪ ADPENE EQ.0: global remeshing

GT.0: local refinement
▪ The mesh size of local refinement is based on contact surface curvature
▪ Bigger ADPENE, larger and smoother transition from fine to coarse mesh
▪ Works for contacts defined between adaptive part and ADPENE=3.0
(1) rigid / deformable / adaptive solid part
(2) rigid / deformable shell part
▪ Supporting contact types: SURFACE_TO_SURFACE (SOFT=0,1)
Recommend FORMING_, AUTOMATIC_, _MORTAR
▪ ONLY supporting 3D general TET remesher ADPENE=4.0

Adaptive remesh: mesh size (3-1) 37 / 138

▪ Local refinement
▪ ADPENE=1
▪ Self-riveting application Deformable
▪ Two adaptive EFG parts tool
▪ *CONTACT_FORMING_SURFACE_TO_SURFACE_MORTAR_THERMAL
▪ MPP Implicit analysis
▪ Thermal-mechanical coupling

▪ Avoid unnecessary refinement (if possible) Adaptive parts

▪ Extend the tool edges beyond
adaptive part boundaries
and symmetric planes
▪ Use multiple tool parts
to construct corners

Adaptive remesh: mesh size (3-2) 38 / 138

▪ *DEFINE_BOX_ADAPTIVE (available in Beta)
▪ Card1
Variable BOXID XMN XMX YMN YMX ZMN ZMX
Type I F F F F F F
Default 0.0 0.0 0.0 0.0 0.0 0.0

▪ Card2
Variable PID LEVEL LIDX/NDID LIDY LIDZ BRMIN BRMAX
Type I I I I I F F
Default 0 1 0 0 0 0.0 0.0

PID : 3D adaptive part ID

BRMIN : RMIN for remeshing the mesh contained in this box
BRMAX : RMAX for remeshing the mesh contained in this box
▪ Multiple keyword cards can be defined for different adaptive parts or different areas of adaptive parts

Adaptive remesh: mesh size (4-1) 39 / 138

▪ *DEFINE_BOX_ADAPTIVE
▪ Taylor bar impact
NO contact definition to trigger local refinement
▪ Define multiple boxes to optimize the mesh size distribution
▪ ONLY support 3D general TET remesher

w/ boxes

NO boxes

Adaptive remesh: mesh size (4-2) 40 / 138

▪ *CONTROL_REMESHING(_EFG)
▪ Card 1 Variable RMIN RMAX VF_LOSS MFRAC DT_MIN ICURV CID SEGANG
Type F F F F F I I F
Default 1.0 0.0 0.0 4 0 0.0

▪ CID defines the orbital axis orbital axis // global z

The orbital axis has to be in parallel to the global z-axis in current practice
▪ SEGANG defines the angular mesh size
The absolute mesh size is adjusted between RMIN and RMAX
▪ ADPOPT=3 (card 2 in *PART): mark the orbital adaptive parts
▪ PENT/HEX remesher, FEM(ELFORM=1), EFG(ELFORM=41)
▪ IDIM=-1 (card 2 in*SECTION_SOLID_EFG)
Stabilized EFG method (very efficient for PENT/HEX mesh)
▪ NO local refinement
▪ NOT support *DEFINE_BOX_ADAPTIVE

Orbital remeshing (1) 41 / 138

▪ Orbital remeshing
▪ Adaptive EFG captures and maintains the high gradient field
▪ *CONTACT_TIED_SURFACE_TO_SURFACE
▪ Explicit analysis

EPS

Orbital remeshing (2) 42 / 138

▪ Remapping
▪ Internal variables (stress / strain / …) transfer from old to new mesh
▪ High order remapping function
Minimizing numerical error
Maintaining high gradient field
Improving the convergence in implicit analysis right after adaptive step
▪ Multiple options to balance between accuracy and efficiency

▪ Interactive adaptive control

▪ Automatically detects mesh distortion and triggers adaptivity
▪ Optimizes adaptive intervals to minimize total number of adaptive steps for both accuracy and efficiency
▪ Suitable for explicit analysis

▪ Mesh size control

▪ Monotonic resizing

Adaptive EFG enhanced features 43 / 138

▪ *CONTROL_REMESHING_EFG
▪ Card 2
Variable IVT IAT IAAT IER MM
Type I I I I I
Default 1 0 0 0 0

IVT EQ. 1: 1st order meshfree remapping function (interpolation)

EQ.-1: 1st order meshfree remapping function (NOT interpolation)
EQ. 2: 0th order meshfree remapping function (interpolation)
EQ.-2: 0th order meshfree remapping function (NOT interpolation)
EQ.-3: FEM remapping scheme

FEM remapping EFG remapping

Adaptive EFG remapping (1) 44 / 138

▪ Taylor bar impact
▪ Adaptive EFG

EPS

Smeared gradient

IVT=1 IVT=2

Adaptive EFG remapping (2) 45 / 138

▪ Self-piercing riveting
▪ Adaptive EFG
EPS
▪ Multiple adaptive parts
▪ Explicit / implicit analysis
▪ MPP

IVT=2 IVT=1

Adaptive EFG remapping (3) 46 / 138

▪ Pre-defined adaptive control in *CONTROL_ADAPTIVE
▪ ADPFREQ: constant time intervals
▪ LCADP: curve data defining time intervals
▪ NOT suitable to maintain mesh quality for complex material deformation
NOT able to avoid unnecessary adaptive steps (introducing error, more expensive and less robust)

▪ *CONTROL_REMESHING_EFG
▪ Card 2
Variable IVT IAT IAAT IER MM
Type I I I I I
Default 1 0 0 0 0

▪ Card 3
Variable IAT1 IAT2 IAT3
Type F F F
Default 1.0E20 1.0E20 1.0E20

Adaptive EFG: interactive adaptive control (1) 47 / 138

▪ *CONTROL_REMESHING_EFG
▪ Card 2
Variable IVT IAT IAAT IER MM
Type I I I I I
Default 1 0 0 0 0

IAT EQ.0: No interactive adaptivity

A constant period of adaptivity is defined by ADPFREQ
ADPFREQ t

EQ.1: Interactive adaptivity combined with predefined adaptivity

Extra adaptivity is triggered interactively within every period defined by ADPFREQ
ADPFREQ t

EQ.2: Purely interactive adaptivity

The time interval between two successive adaptive steps is bounded by ADPFREQ
DT DT<ADPFREQ t

EQ.3: Purely interactive adaptivity

Interactive adaptivity t

Adaptive EFG: interactive adaptive control (2) 48 / 138

▪ *CONTROL_REMESHING_EFG
▪ Card 2
Variable IVT IAT IAAT IER MM
Type I I I I I
Default 1 0 0 0 0

▪ IAAT=1: automatically adjusts the tolerance of interactive control parameters based on ADPFREQ and
the history of adaptive time intervals to avoid over-activation of interactive adaptivity

▪ Card 3 IAT1: shear deformation

Variable IAT1 IAT2 IAT3
Type F F F Fij  Fji , i j

Default 1.0E20 1.0E20 1.0E20

IAT2: volumetric change

1  det  F 
▪ IAT1, IAT2, IAT3: user-define tolerance of three
interactive control parameters
(distortion indicators) IAT3: unbalanced nodal distribution
Max  edge length 
▪ The rate of indicator change over ONE time step
Interactive adaptivity when the rate is over 50% Min  edge length 

Adaptive EFG: interactive adaptive control (3) 49 / 138

▪ Wheel forging Resultant force

▪ Adaptive EFG with interactive control

IAT 0 3

Normalized CPU time 1.0 0.72

# of adaptive steps 50 22

Adaptive EFG: interactive adaptive control (4-1) 50 / 138

▪ Wheel forging
▪ Adaptive EFG with interactive control
▪ IAT=3
▪ IAAT=1
Interactive adaptivity triggered by rate of indicator change

Tolerance
Shear deformation Indicator
IAT1
t (sec)

Unbalanced nodal
Distribution
IAT3

Volumetric change
IAT2

Adaptive EFG: interactive adaptive control (4-2) 51 / 138

▪ Gear forging
▪ Adaptive EFG with interactive control
▪ IAT=3
▪ IAAT=1

Courtesy of JSOL

Adaptive EFG: interactive adaptive control (5) 52 / 138

▪ *CONTROL_REMESHING_EFG
▪ Card 2
Variable IVT IAT IAAT IER MM
Type I I I I I
Default 1 0 0 0 0

▪ MM=1: monotonic resizing

▪ Captures high gradient
▪ Maintains mesh density
MM=0
▪ Useful if local refinement is not applicable

MM=1
MM=0 MM=1

Adaptive EFG: monotonic resizing 53 / 138

▪ New nodes and elements after adaptive remeshing
▪ *SET_ based on old node and element IDs is UNDEFINED
▪ Problematic keywords to impose constraints on 3D adaptive parts
*BOUNDARY_SPC_SET
*BOUNDARY_PRESCRIBED_MOTION_SET

▪ Constraints on flat surfaces

▪ Symmetry planes, fixed planes
▪ Recommended keywords
*CONSTRAINED_GLOBAL
*CONSTRAINED_LOCAL
*BOUNDARY_SPC_SYMMETRY_PLANE

▪ Constraints on complex surfaces

▪ Fixed cylinder surface in orbital forming
▪ Define *CONTACT_TIED_ to rigid body

Imposing constraints in 3D r-adaptivity 54 / 138

▪ Remeshing does not take contact surface into account
▪ Change of geometry in remeshing may result in contact penetration right after adaptivity
▪ *CONTROL_CONTACT
▪ Card 1
Variable SLSFAC RWPNAL ISLCHK SHLTHK PENOPT THKCHG ORIEN ENMASS
Type F F I I I I I I
Default 0.1 0 1 1 0 1 0
▪ SLSFAC: scale factor for sliding interface penalties (1.0~6.0 recommended in adaptivity)
▪ SHLTHK: shell thickness in contact (also in *CONTACT_)
EQ.1: thickness is considered excluding rigid bodies (AUTOMATIC_SURFACE_TO_SURFACE)
EQ.2: thickness is considered including rigid bodies (SURFACE_TO_SURFACE) offset thickness/2
▪ PENOPT: penalty stiffness value (4 or 5 are recommended for high gradients when using adaptivity)
EQ.4: use slave node value, area or mass weighted
EQ.5: proportional to shell thickness
▪ Recommended contact types in adaptivity
*CONTACT_(AUTOMATIC)_SURFACE_TO_SURFACE
*CONTACT_(FORMING/AUTOMATIC)_SURFACE_TO_SURFACE_MORTAR
Contacts in 3D r-adaptivity (1) 55 / 138
▪ *CONTACT_
▪ Optional card A
Variable SOFT SOFSCL LCIDAB MAXPAR SBOPT DEPTH BSORT FRCFRQ
Type I F I F F I I I
Default 0 0.1 0 1.025 0.0 2 10-100 1

▪ SOFT: soft constraint option

EQ.0: penalty formulation
local refinement
EQ.1: soft constraint formulation
EQ.2: segment-based contact global remeshing
EQ.4: constraint approach for FORMING contact option
▪ BSORT: number of cycles between bucket sorts
(smaller values are recommended in adaptivity)
▪ Singular conditions in contact result in numerical instability
Extend the tool edges beyond adaptive part boundaries and symmetric planes
Use multiple tool parts to construct corners

Contacts in 3D r-adaptivity (2) 56 / 138

▪ Contact force oscillation in adaptivity
▪ Gaps between contact surfaces are changed after remeshing
The influence of changing contact gaps is amplified by penalties
▪ Current implementation of contact algorithm resets contact
parameters and builds up contact force from zero after adaptivity

▪ BODY_TO_BODY contact
▪ Much less sensitive to the change in contact surface gaps
▪ Under development for SPG (see page 116-117 for detail)

Contact
Force

Contacts in 3D r-adaptivity (3) 57 / 138

▪ *CONSTRAINED_BEAM_IN_SOLID (coupling with adaptivity, available in Beta)
▪ Card 2
Variable SLAVE MASTER SSTYP MSTYP NCOUP CDIR
Type I I I I I I
Default 0 0 0 0

▪ NCOUP: coupling points per beam element

▪ CDIR: allowing slip or not

▪ Card 3 Master Node Slave Node

Variable START END AXFOR
Type F F I
Default 0 1.0E10 0

▪ AXFOR: de-bonding law

EQ.-n: function ID n
Courtesy of JSOL
EQ.1001: user subroutine

Fiber-reinforced compression molding using adaptivity (1) 58 / 138

▪ *CONSTRAINED_BEAM_IN_SOLID
▪ Supports TET, PENT and HEX elements
▪ Coupling multiple beams in one solid element

▪ Additional coupling points by defining NCOUP

Master Node Slave Node Coupling Point

Fiber-reinforced compression molding using adaptivity (2) 59 / 138

▪ Compression molding analysis
▪ Long glass fiber reinforced ThermoPlastics (320k beam and 228k~358k TET solid elements)
▪ Explicit analysis (termination time 0.05s)
▪ Thermal-mechanical coupling
▪ MPP (64 cores, 6h clock time)

Courtesy of JSOL

Fiber-reinforced compression molding using adaptivity (3-1) 60 / 138

▪ Compression molding analysis
▪ Long glass fiber reinforced ThermoPlastics (320k beam and 228k~358k TET solid elements)
▪ Explicit analysis (termination time 0.05s)
▪ Thermal-mechanical coupling
▪ MPP (64 cores, 6h clock time)

Courtesy of JSOL

Fiber-reinforced compression molding using adaptivity (3-2) 61 / 138

▪ Compression molding analysis
▪ Long glass fiber reinforced ThermoPlastics (320k beam and 228k~358k TET solid elements)
▪ Explicit analysis (termination time 0.05s)
▪ Thermal-mechanical coupling
▪ MPP (64 cores, 6h clock time)

Courtesy of JSOL

Fiber-reinforced compression molding using adaptivity (3-3) 62 / 138

▪ *CONTROL_IMPLICIT_AUTO
▪ Card 1
Variable IAUTO ITEOPT ITEWIN DTMIN DTMAX DTEXP KFAIL KCYCLE
Type I I I F F F
Default 0 11 5 DT/1000 DT*10
▪ IAUTO=1: automatically adjusts time step (recommended to improve convergence right after adaptivity)
▪ DTMAX: combined with ADPFREQ to maintain multiple implicit time steps in each adaptive time interval

▪ *CONTROL_IMPLICIT_GENERAL
▪ Card 1
Variable IMFLAG DT0 IMFORM NSBS IGS CNSTN FORM ZERO_V
Type I F I I I I I I
Default 0 2 1 2 0 0 0

▪ DT0: initial time step size (the first time step size right after every adaptive step)
smaller DT0 improves convergence in adaptivity

Implicit analysis in adaptivity 63 / 138

▪ *INTERFACE_SPRINGBACK_LSDYNA
▪ Card 1
Variable PSID NSHV
Type I I
Default
▪ Stage 1: creates dynain OUTPUT file containing history variables of the final deformation
▪ Stage 2: includes dynain file to perform ONE-STEP implicit spring back analysis

Spring back analysis using adaptivity 64 / 138

▪ 3D adaptivity (FEM/EFG) takes full advantage of thermal-mechanical coupling solver
▪ Thermal induced material laws
▪ Thermal contacts
▪ Thermal boundary and initial conditions (NOT based on *SET_ of nodes and elements of adaptive parts)

Temperature von Mises Stress von Mises Stress

(spring back)

Thermal-mechanical coupling in adaptivity 65 / 138

▪ *CONTROL_ADAPTIVE
▪ Card 4
Variable CNLA MMM2D ADPERRR D3TRACE IADPFC IFSAND
Type F I I I I I
Default 0 0 0 0 0 0

▪ IADPFC=1: turns on run-time control for 3D TET r-adaptivity (available in R9)

▪ Main features
Manually trigger additional adaptive remeshing
Manually remesh and edit mesh
Define adaptive parameters for individual adaptive part
▪ Two control files in the job folder
▪ adapt.fc1
Trigger additional adaptive steps either immediately or any time later
Manually remesh/edit the mesh instead of using LS-DYNA remesher
▪ adapt.fc2
Set specific RMIN, RMAX, TBIRTH and TDEATH for individual adaptive part

Adaptivity user control files (1) 66 / 138

▪ adapt.fc1 defines three parameters (C1, C2, C3)
▪ (1,0.0,0): trigger additional normal adaptivity immediately
▪ (1, 0.0025,0): trigger additional normal adaptivity at t=0.0025
▪ (1, 0.0025,3): trigger special adaptivity with manual remeshing on
adaptive part 3 at t=0.0025

▪ Manual remeshing and mesh editing procedure:

(1) LS-DYNA hangs up waiting for new mesh from user
(2) "user.mesh" file in the job folder contains the mesh for users to
remesh / edit
(3) Change C1 to -1 to continue LS-DYNA with new "user.mesh"

Adaptivity user control files (2) 67 / 138

▪ adapt.fc2 defines parameter list for adaptive parts
▪ Example:

Line 1: 2 (number of adaptive parts)

Line 2: 2, 0.0,0.1,1,4 (for adaptive part 2, TBIRTH is 0.0,
TDEATH is 0.1, RMIN=1, RMAX=4)
Line 3: 3, 0.01,0.2,2,4 (for adaptive part 3, TBIRTH is 0.01,
TDEATH is 0.2, RMIN=2, RMAX=4)

▪ If new mesh quality of some adaptive parts using manual remeshing

is NOT good enough for the remesher in LS-DYNA (TET remesher in
LS-PrePost is a good tool to test that), it is recommended to stop
adaptivity by setting corresponding TDEATH to be just slightly larger
than C2 in adapt.fc1 to avoid error termination due to failure of
automatic remeshing

Adaptivity user control files (3) 68 / 138

▪ Self-riveting
▪ Adaptive EFG
Manually erode a thin layer of elements Manually cut the mesh using Hypermesh
▪ Thermal coupling Stop adaptivity of top sheet after erosion Adaptivity of top sheet continues on
▪ Implicit analysis
▪ MPP

Adaptivity user control files (4) 69 / 138

▪ Complex forming
▪ Adaptive EFG
▪ MPP implicit analysis

Adaptivity: more applications (1) 70 / 138

▪ Squeezing
▪ Adaptive EFG
▪ MPP implicit analysis
▪ Deformable tools

Work piece

Top tool

Adaptivity: more applications (2) 71 / 138

 ▪ Friction stir welding (FSW)
▪ Adaptive EFG
▪ MPP implicit analysis

Temperature

Adaptivity: more applications (3) 72 / 138

▪ Cutting
▪ Adaptive EFG
▪ MPP implicit analysis

Adaptivity: more applications (4) 73 / 138

▪ Piercing EPS
▪ Adaptive EFG
▪ MPP implicit analysis

Rigid

Piercing force

Adaptivity: more applications (5) 74 / 138

▪ Material failure analysis
▪ Weak and strong discontinuity
▪ Failure in ductile and brittle materials
▪ Numerical failure and physics-based material failure Destructive
▪ SPG for ductile and semi-brittle failure manufacturing
▪ Peridynamics for brittle failure

▪ SPG Grinding, riveting, cutting, flow drill

screw, self-tapping screw, drilling
▪ Eulerian and semi-pseudo Lagrangian kernels
▪ Two different failure analysis
Continuum material damage and particle erosion
“Bond”-break based on EPS and “bond” stretching

▪ Peridynamics
▪ Nonlocal theory
▪ Micro modulus and critical bond stretching
▪ Discontinuous Galerkin framework
▪ Hybrid bonds for composite laminate
▪ Weak discontinuity: discontinuous deformation gradient
▪ Continuum damage constitutive law + nonlocal strain smoothing
Material erosion (loss of mass and volume)
▪ Implicit cracks with assumed width (mesh size dependent)
weakening
▪ Small time step size in explicit analysis: fine mesh and distortion
material
▪ Strong discontinuity: discontinuous displacement
▪ Cohesive model + Interface element / eXtended FEM (X-FEM)
▪ Phenomenological failure + nonlocal strain (SPG, peridynamics)
Fine-scale modeling to improve failure criteria in continuum scale
▪ Explicit cracks: NO mesh (size/orientation) sensitivity issue
▪ Expensive and difficult to construct free surfaces in 3D

▪ Idea: weak + strong discontinuities

▪ Phenomenological material laws provide material failure indicator
Erode damaged material
▪ “Bond”-break in SPG (invisibility) and Peridynamics for post failure
Create free surface
▪ Avoid the difficulty to construct free surfaces in 3D

Material failure analysis (1) 76 / 138

▪ Ductile failure ▪ Brittle failure
▪ Nonlocal failure zone (material length scale) ▪ Crack tip (stress singularity)
Fine-scale modeling + experiments Regularization (e.g. cohesive model)
▪ SPG smoothing ▪ Peridynamics
▪ Numerical stabilization ▪ Bond stretching
▪ Strain regularization to bring in nonlocal effect Sc related to fracture energy release rate
Regularization support ~ material length scale
▪ Failure zone consists of broken bonds
▪ SPG “bond”-break (invisibility) ▪ Extended to failure in composite laminate
▪ Failure indicators: EPS, … Semi-brittle
▪ “bond” stretching
SPG

Crack tip

Material failure analysis (2) 77 / 138

▪ Physics-based failure before numerical failure (Eulerian kernel)
▪ Adjust support to secure the “bond” until its physical breaking based on failure criteria

Tensile instability in SPH Eulerian kernel (numerical failure)

Updated Lag (NO material failure) Eulerian kernel (material failure)

Material failure analysis (3) 78 / 138

▪ *SECTION_SOLID_SPG (available in R9, recommend Beta version)
▪ Card 2
Variable DX DY DZ ISPLINE KERNEL LSCALE SMSTEP SUKTIME
Type F F F I I F I F
Default 1.5 1.5 1.5 0 0 15

KERNEL EQ.0: Updated Lagrangian kernel (NO failure, less shear deformation)
Initial
EQ.1: Eulerian kernel (failure analysis, global extreme deformation) support
Support defined in the current configuration is adjusted
according to the current deformation
EQ.2: Semi-pseudo Lag kernel (failure analysis, local extreme deformation)
Support is defined in the initial configuration current
support
Remove neighboring points from support due to “bond” failure

KERNEL=1 KERNEL=2

Failure analysis SPG keyword (1) 79 / 138

▪ *SECTION_SOLID_SPG
▪ Card 3
Variable IDAM FS STRETCH ITB Re-compute the shape
Type I F F I functions at node I

Default 0 0

▪ IDAM EQ. 0: Continuum damage mechanics

J
Nodes fail based on the criteria defined in material law
or *MAT_ADD_EROSION
Failed nodes become free particles and interact with
others through contact I
EQ. 1: Phenomenological EPS strain-based failure criteria
“Bond” is failed (broken) by satisfying two conditions
(1) The average EPS of two nodes of this “bond” reaches
user-define failure EPS value FS
(𝐸𝑃𝑆𝐼 + 𝐸𝑃𝑆𝐽 )/2 ≥ FS
(2) The “bond” is stretched over the user-define critical
|𝑥𝐼 − 𝑥𝐽 |/|𝑋𝐼 − 𝑋𝐽 | ≥STRETCH
value STRETCH
Practical guideline: STRETCH < (1+FS) < DX,DY,DZ
Failure analysis SPG keyword (2) 80 / 138
▪ *CONTACT_
▪ NODES_TO_SURFACE contact between SPG (slave nodes) and FEM
▪ SPG to SPG contact
▪ NODES_TO_SURFACE
▪ Particle based BODY_TO_BODY and SELF contacts under development (trial version in Beta SMP)
▪ SOFT=1: soft constraint contact

▪ Thermal coupling
▪ Explicit and implicit SPG thermal solver (Beta SMP)
▪ Convert contact frictional energy to heat
FRCENG=1 (card 4, *CONTROL_CONTACT)
▪ Convert internal plastic work to heat
▪ Thermal expansion is currently NOT available

▪ Explicit time step size

▪ TSSFAC (*CONTROL_TIMESTEP)
Recommended value (0.1 ~ 0.3) if controlled by SPG parts

Failure analysis SPG related keyword 81 / 138

▪ Metal cutting
▪ Material
ρ0=2.7×10-6 kg/mm3 EPS
E=78.2GPa
v=0.3
σy=0.29(1+125ep)0.1
▪ Cutting Speed = 10m/s
▪ Fixed ∆t=3.0×10-5
▪ SPG
▪ DX=DY=DZ=1.6
▪ Eulerian kernel
▪ SMSTEP=15
▪ IDAM=1, FS=0.5, STRETCH=1.2

▪ *CONTACT_NODES_TO_SURFACE

Failure analysis using SPG: manufacturing applications (1-1) 82 / 138

▪ Metal cutting
▪ Material
ρ0=2.7×10-6 kg/mm3
E=78.2GPa EPS
v=0.3
σy=0.29(1+125ep)0.1
▪ Cutting Speed = 10m/s
▪ Fixed ∆t=3.0×10-5
▪ SPG
▪ DX=DY=DZ=1.6
▪ Eulerian kernel
▪ SMSTEP=15
▪ IDAM=1, FS=0.5, STRETCH=1.2

▪ *CONTACT_NODES_TO_SURFACE
▪ Different cutting angles

Failure analysis using SPG: manufacturing applications (1-2) 83 / 138

▪ Metal cutting
▪ Material
ρ0=2.7×10-6 kg/mm3
E=78.2GPa
EPS
v=0.3
σy=0.29(1+125ep)0.1
▪ Cutting Speed = 10m/s
▪ Fixed ∆t=3.0×10-5
▪ SPG
▪ DX=DY=DZ=1.6
▪ Eulerian kernel
▪ SMSTEP=15
▪ IDAM=1, FS=0.5, STRETCH=1.2

▪ *CONTACT_NODES_TO_SURFACE
▪ Different cutting depth

Failure analysis using SPG: manufacturing applications (1-3) 84 / 138

▪ Metal cutting

EPS

Failure analysis using SPG: manufacturing applications (1-4) 85 / 138

▪ Metal cutting
Rigid blade
(FEM shell)
Cutting depth: 2.5mm

20x5x1mm
Aluminum

* Espinosa C, Lacome JL, Limido J, Salaun M, Mabru C, Chieragatti R.,

Modeling high speed machining with the SPH method, 10 international
LS-DYNA users conference, Dearborn, MI, June 8-10, 2008

Failure analysis using SPG: manufacturing applications (1-5) 86 / 138

▪ Metal shearing
▪ Material: Aluminum
ρ0=2.7×10-6 kg/mm3
E=78.2GPa
v=0.3
σy=0.29(1+125ep)0.1
▪ Fixed ∆t=1.0×10-5
▪ SPG EPS
▪ DX=DY=DZ=1.6
▪ Eulerian kernel
▪ SMSTEP=15
▪ IDAM=1, FS=0.5, STRETCH=1.2 Part
▪ EPS is monotonically increased
Burr Geometry
NO numerical diffusion
▪ Major applications
Blanking, bolt/rivet shearing, Scrap
AHSS trimming …

Failure analysis using SPG: manufacturing applications (2-1) 87 / 138

▪ Metal shearing
▪ Rigid tool (FEM)
▪ Metal: isotropic hardening
▪ Metal thickness: 1mm
▪ Shearing clearance: 0.1mm

R0.05mm

FEM

SPG (mesh size 0.02 mm)

Failure analysis using SPG: manufacturing applications (2-2) 88 / 138

▪ Metal shearing
▪ Rigid tool (FEM)
▪ Metal: isotropic hardening
▪ Metal thickness: 1mm
▪ Shearing clearance: 0.1mm

Failure analysis using SPG: manufacturing applications (2-3) 89 / 138

▪ Punching hole in metal
▪ SPG Time-Punch Force
▪ DX=DY=DZ=1.6
▪ Eulerian kernel
▪ SMSTEP=15
▪ IDAM=1, FS=0.5, STRETCH=1.2

Pressure Contour
(Cross-section view)

Effective Stress Contour (full model)

Failure analysis using SPG: manufacturing applications (3-1) 90 / 138

▪ Punching hole in metal
▪ Time step size ∆t=7.69~8.16 ×10-8

Effective Stress Contour (Cross-section view)

Failure analysis using SPG: manufacturing applications (3-2) 91 / 138

▪ Self-piece riveting
▪ Additional failure criteria (in-house code): NO material failure under compression
Current development: new failure criteria considering stress triaxiality

Constant v

Rigid Deformable tool

(FEM)

Solid plates
(SPG) EPS

Failure analysis using SPG: manufacturing applications (4-1) 92 / 138

▪ Self-piece riveting
▪ MPP explicit analysis

EPS

Von Mises stress

Failure analysis using SPG: manufacturing applications (4-2) 93 / 138

▪ Flow drill screw (FDS)
stage 1
Head with external
drive system Thread forming zone

Flow drilling zone stage 2

sharing nodes SPG FEM

Courtesy of Ford Motor Co.

Stage 1

EPS

Failure analysis using SPG: manufacturing applications (5-1) 94 / 138

▪ Flow drill screw (FDS)

Stage 1 Stage 2

von Mises stress

EPS EPS

Failure analysis using SPG: manufacturing applications (5-2) 95 / 138

▪ Flow drill screw (FDS): pull-out test
Constant v Stage I Stage II

Rigid

Solid plates (SPG)

39941 nodes

Constant v

v decreases by x10
Rotational speed increases by x4

Rotation creates material vertical transportation Start unscrewing

in the direction opposite to the prescribed v

Failure analysis using SPG: manufacturing applications (5-3) 96 / 138

▪ Flow drill screw (FDS): thermal coupling

Temperature

EPS

Failure analysis using SPG: manufacturing applications (5-4) 97 / 138

▪ Self-drilling screw

Tighten the screw

Contact resultant force

T3.5
screw
steel

Fix

Fix

Failure analysis using SPG: manufacturing applications (6) 98 / 138

 von Mises stress
▪ Metal tearing
▪ Multiple cracks
▪ SPG only handles failure zone
Sharing nodes between SPG and FEM

EPS

Failure analysis using SPG: manufacturing applications (7) 99 / 138

▪ Metal grinding

Ø10mm blade 8000rpm

Cutting depth: 10mm
Cutting thickness: 1mm

50

10

100

Failure analysis using SPG: manufacturing applications (8-1) 100 / 138

▪ Metal grinding

Failure analysis using SPG: manufacturing applications (8-2) 101 / 138

▪ Metal grinding

SPG: KERNEL=2, IDAM=1 FEM with element erosion

Failure analysis using SPG: manufacturing applications (8-3) 102 / 138

 Courtesy of Dr. Steward Silling at Sandia Nat. Lab.

Peridynamics: nonlocal theory 103 / 138

▪ PDE of explicit dynamic analysis

Bond
▪ Bond force fi
▪ Micro-elastic brittle material
Horizon
     
f  cs , s
  

▪ Micro modulus c
▪ Limitation: fixed Poisson’s ratio 0.25

▪ Conservation of linear and angular momentums

Peridynamics: bond force 104 / 138

▪ Classic elastic energy density (small deformation)

▪ Micro elastic energy in one stretched bond

▪ Classic energy and collective micro elastic energy are equivalent

▪ varies from bond to bond

▪ boundary correction

Peridynamics: from Young’s modulus to micro modulus 105 / 138

 ▪ varies from bond to bond
Madenci, et. al., 2014
Wu, et. al., 2014

Peridynamics: from fracture energy release rate to critical bond stretching 106 / 138
▪ Weak form of the governing equation

Node detach
Element tool in LS-PrePost

“Crack” propagates along element edges

Peridynamics: discontinuous Galerkin weak form 107 / 138

▪ *SECTION_SOLID_PERI (Beta version)
▪ Card 1
Variable SECID ELFORM
Type I I
Default

ELFORM EQ.48: Peridynamic formulation for TET, PENT, HEX solid elements

▪ Card 2
Variable DR PTYPE
Type F I
Default 1.01 1

DR: normalized horizon size, 𝟎.𝟔~𝟏.𝟐 is recommended

PTYPE EQ.1: bond based formulation
EQ.2: state based formulation

Peridynamics keyword (1) 108 / 138

▪ *MAT_ELASTIC_PERI (Beta version)
▪ Card 1
Variable MID RO E G
Type I I F F
Default 1.0E20

RO: density
E: Young’s modulus
G: fracture energy release rate

Peridynamics keyword (2) 109 / 138

 600 elements 38400 elements

4800 elements

Mode I crack: convergence study 110 / 138

▪ Composite laminate failure analysis
▪ Decompose laminate model into two parts
▪ Inter-layer and inner-layer bonds
▪ Compute micro-modulus c based on spherical
harmonic functions of angles from fiber
direction [Ghajari, et. al., 2014]
▪ Use structure mesh to model arbitrary cracks

Composite laminate Peridynamics model (under development) 111 / 138

▪ Compact-tension test of a cortical bone plate
▪ 𝐺𝐼𝑓 = 2190𝐽/𝑚2 and 𝐺𝐼𝑚 = 1030𝐽/𝑚2
▪ Different fiber orientations: 00 , 300 , and 600

Composite laminate Peridynamics model: examples (1-1) 112 / 138

▪ Compact-tension test of a cortical bone plate
▪ 𝐺𝐼𝑓 = 2190𝐽/𝑚2 and 𝐺𝐼𝑚 = 1030𝐽/𝑚2
▪ Different fiber orientations: 00 , 300 , and 600

Composite laminate Peridynamics model: examples (1-2) 113 / 138

▪ Fiber reinforced composite laminate
▪ T700 carbon/epoxy material Numerical results Experiment results
▪ Eight laminate [45/0/-45/90]s layers
▪ 229504 nodes

Longitudinal

transverse
Elastic deformation

Composite laminate Peridynamics model: examples (2-1) 114 / 138

▪ Fiber reinforced composite laminate

Inner-layer failure Delamination

Composite laminate Peridynamics model: examples (2-2) 115 / 138

▪ SPG
▪ Fluid particle method to handle flying particles
▪ Particle based BODY_TO_BODY and SELF contacts Material failure and
▪ Applications on impact, penetration and fragmentation fragmentation
analysis
▪ Peridynamics
▪ Applications on 3D crack propagation in brittle solid Impact and penetration, 3D crack
propagation in solid, delamination
and folding, fragmentation
▪ *SECTION_SOLID_SPG (Beta version)
▪ Card 3
Variable IDAM FS STRETCH ITB
Type I F F I
Default 0 0

T-bond
ITB: stabilization
EQ.0: standard meshfree approximation + T-bond failure algorithm
T-bond: symmetric and asymmetric
EQ.1: fluid particle approximation (accurate but expensive)
EQ.2: simplified fluid particle approximation (fast and robust)
▪ Recommended setting combination
KERNEL=1 and ITB=1 for metal cutting and riveting
KERNEL=2 and ITB=2 for impact and penetration

Ghost particles

SPG flying particles 117 / 138

▪ *CONTACT_SPG (Beta SMP version)
▪ Card 1
Variable IPART1 IPART2 IPART3 IPART4 IPART5 IPART6 IPART7 IPART8
Type I I I I I I I I
Default

IPARTn: SPG part IDs for multi-body contact

▪ Card 2
Variable ISELF ISELF ISELF ISELF ISELF ISELF ISELF ISELF
Type I I I I I I I I
Default 0 0 0 0 0 0 0 0

ISELF EQ.0: The corresponding SPG part has NO self-contact

EQ.1: The corresponding SPG part has self-contact

SPG BODY_TO_BODY and SELF contact (1) 118 / 138

▪ *CONTACT_SPG (Beta SMP version)
▪ Card 3
Variable PEN1 PEN2 PEN3 PEN4 PEN5 PEN6 PEN7 PEN8
Type F F F F F F F F
Default 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

PENn: penalty number for SPG parts in contact (recommend Young’s modulus)
▪ Card 4
Variable FS FD DC NFEQ
Type F F F I
Default 10

FS: static coefficient of friction contact

FD: dynamic coefficient of friction contact 𝜇𝑐 = 𝐹𝐷 + 𝐹𝑆 − 𝐹𝐷 𝑒 −𝐷𝐶∙|𝑣𝑟𝑒𝑓 |
DC: exponential decay coefficient
NFEQ: frequency of contact sorting

SPG BODY_TO_BODY and SELF contact (2) 119 / 138

▪ Aluminum projectile impacts into aluminum plate
▪ Four different models with uniform HEX mesh
3249 SPG nodes (1.6mm mesh), 128480 FEM
12493 SPG nodes (1.0mm mesh), 469200 FEM
30258 SPG nodes (0.75mm mesh), 944160 FEM
93025 SPG nodes (0.5mm mesh), 1230528 FEM

Slab: 8’x8’x0.5’
Projectile: Ø16.7x23mm

m=13.6g
FEM 970m/s

SPG

* Schwer L.E. “Aluminum plater perforation: a comparative case study using Lagrange with erosion, multi-material ALE, and smooth
particle hydrodynamics,” 7th European LS-DYNA Conference, Salzburg, Austria, May 14th-15th, 2009

SPG failure analysis: impact and penetration (1-1) 120 / 138

▪ Aluminum projectile impacts into aluminum plate

EPS

Effective
stress

0.5 mm 0.75 mm 1.0 mm

SPG failure analysis: impact and penetration (1-2) 121 / 138

▪ Aluminum projectile impacts into aluminum plate

0.5 mm 0.75 mm 1.0 mm

SPG failure analysis: impact and penetration (1-3) 122 / 138

▪ Aluminum projectile impacts into aluminum plate

Test

0.5 mm 0.75 mm 1.0 mm

SPG failure analysis: impact and penetration (1-4) 123 / 138

▪ Aluminum projectile impacts into aluminum plate
▪ Convergence study

Slab thickness
Stress is recorded at the
impact center

Trial test: both parts are elastic

Test residual

SPG failure analysis: impact and penetration (1-5) 124 / 138

▪ Tungsten rod impacts into steel

Projectile:
42752 FE elements
Tungsten alloy
Ø7.87x78.74mm 56953 SPG nodes
65.4g mesh size 0.52mm
1500m/s

Target: 282240 FE elements

RHA steel
Ø100x127mm

* Rajendran A.M. “Penetration of Tungsten alloy rods into shallow-cavity steel target,” Int. J. Impact
Engng, 21(6), p451-460, 1998

SPG failure analysis: impact and penetration (2-1) 125 / 138

▪ Tungsten rod impacts into steel

EPS

von Mises
stress

FS=0.45 FS=0.90 FS=1.50

SPG failure analysis: impact and penetration (2-2) 126 / 138

▪ Tungsten rod impacts into steel

DOP in test

SPG failure analysis: impact and penetration (2-3) 127 / 138

▪ Perforation on high strength concrete
▪ SPG IDAM=1: damage indicator (critical value FS) is the one in concrete material law
Projectile
Ø75 x L225mm
6.3kg @ 621m/sec
Elastic FEM Damage distribution
Frictional coef.: 0.15

391040
FE elements

44649
SPG nodes

Concrete * Unosson M, Nilsson L. Projectile penetration and perforation of high

Ø1400 x H400mm cylinder performance concrete: experimental results and macroscopic modeling.
Concrete strength: 153MPa International Journal of Impact Engineering 2006; 32: 1068-1085.

SPG failure analysis: impact and penetration (3-1) 128 / 138

▪ Perforation on high strength concrete
▪ SPG IDAM=1: damage indicator (critical value FS) is the one in concrete material law

SPG failure analysis: impact and penetration (3-2) 129 / 138

▪ Perforation on high strength concrete
▪ SPG IDAM=1: damage indicator (critical value FS) is the one in concrete material law

Projectile
Ø75 x L225mm Damage distribution
6.3kg @ 623m/sec
Elastic FEM
Frictional coef.: 0.15

782080 FE
elements

88209
SPG nodes Concrete
Ø1400 x H800mm cylinder * Unosson M, Nilsson L. Projectile penetration and perforation of high
performance concrete: experimental results and macroscopic modeling.
Concrete strength: 153MPa
International Journal of Impact Engineering 2006; 32: 1068-1085.

SPG failure analysis: impact and penetration (3-3) 130 / 138

▪ Perforation on high strength concrete
▪ SPG IDAM=1: damage indicator (critical value FS) is the one in concrete material law

DOP
Unreliable data in test
due to ejecta and fragmentation

SPG failure analysis: impact and penetration (3-4) 131 / 138

▪ Kalthoff-Winkler problem

30800 elements 97608 elements

52272 elements

Peridynamics 3D crack propagation (1) 132 / 138

▪ Three point bending of concrete beam Experimental results

γ=0, Θ=00 γ=0.5, Θ=210 γ=0.7, Θ=290

Peridynamics 3D crack propagation (2) 133 / 138

▪ 3D Glass-PC-Glass system

Rigid body
E=211Gpa
ρ=2g/cm2
V=30 m/s

Polycarbonate (FEM) Soda-Lime glass Top view Back view

Elastic material Peridynamic model
E=2Gpa E=72Gpa
Υ=0.25 G=8J/m2

Contact force
Ball velocity

Peridynamics 3D crack propagation (3) 134 / 138

▪ Windshield impact

Glass layers: Peridynamic solid

*SECTION_SOLID_PERI, *MAT_ELASTIC_PERI

Vinyl layer: FEM solid

*MAT_PIECEWISE_LINEAR_PLASTICITY

Interface of vinyl and glasses:

*CONTACT_TIED_SURFACE_TO_SURFACE_OFFSET

Peridynamics 3D crack propagation (4-1) 135 / 138

▪ Windshield impact

EPS

Damage pattern (top) Damage pattern (bottom)

Vinyl layer

Peridynamics 3D crack propagation (4-2) 136 / 138

▪ Windshield impact

von Mises stress on vinyl interlayer

Top view Bottom view

Peridynamics 3D crack propagation (4-3) 137 / 138

Thank you!

Questions? 138 / 138