Title: DYNAMICAL EVOLUTION OF TEXTURE IN POLYCRYSTALLINE METALS UNDER HIGHSTRAIN RATE CONDITIONS
1DYNAMICAL EVOLUTION OF TEXTURE IN POLYCRYSTALLINE
METALS UNDERHIGH-STRAIN RATE CONDITIONS
- Zisu Zhao
- Raul Radovitzky
- Department of Aeronautics and Astronautics
- Massachusetts Institute of Technology
- Cambridge, MA
Stephen Kuchnicki Alberto M. Cuitiño Department
of Mechanical and Aerospace Engineering Rutgers
University Piscataway, NJ
International Symposium on Plasticity and its
Current Applications Quebec City, Canada July
7-11, 2003
2Main Motivation
- Physically-based models that properly capture the
physics (e.g. multiscale modeling) - Algorithms and modeling techniques that properly
exercise the models (e.g. DNS or Taylor) - Implementation and computational capabilities
that provide proper spatial and temporal
resolution (e.g. massive parallel environments)
- To develop high-fidelity models and simulation
capabilities for describing multiphysics
scenarios subjected to high strain rates, for
example fluid-solid systems under explosive
conditions
Pressure
Detonation
Only the solid region is shown
Plastic Deformation
ASCI Center for the Dynamic Response of Materials
3Models Multiscale Paradigm
time
hours minutes seconds microsec nanosec picos
ec femtosec
Multiphysics Multicomponent Systems
Grains
- Resolve (as opposed to model) mesoscale behavior
exploiting the power of high-performance
computing - Enable full-scale simulation of engineering
systems incorporating micromechanical effects.
Single Crystal
Stainier, Cuitino and Ortiz (JMPS) 2002 Cuitino
et al (JCAMD) 2002
Energetics
distance
Å nm micron
mm cm meters
4Objectives
- Quantitative assessment of microstructural
effects in macroscopic material response through
Direct Numerical Simulations (DNS) computation of
full-field solutions of polycrystals - Inhomogeneous plastic deformation fields
- Grain-boundary effects
- Stress concentration
- Localization of slip and dislocation pile-up
- Constraint-induced multislip
- Size dependence (inverse) Hall-Petch effect
- Texture evolution
- Assess the degree of approximation of main field
approaches, such as Taylor, by consistently
comparing Taylor to DNS
5Inhomogeneities
Taylor Anvil Test
Direct Numerical Simulation (DNS)
Taylor Average
Radovitzky and Cuitino, 2003
6InhomogeneitiesTaylor anvil test 100 grains
1M elements
Plastic Activity
7Inhomogeneities (grain to grain)Taylor anvil
test 100 grains 1M elements
8Inhomogeneities (within grains)
9Textures
- Selection of a well-characterized case
experimental test (rolling) - Selection of a well-characterized grain shape and
topology (Space filling polyhedron) - Selection of a well-know material model (Pierce,
Asaro, Needleman)
Advantages1. Retains basic properties of grain
shape and topology (usually violated by
grain-growth models due to meshing
difficulties)2. Boundaries are flat, sharp
interfaces3. Easy to mesh4. Highly scalable
(with the aid of scalable meshing)5. Useful tool
for polycrystal simulations DisadvantagesRestric
tive grain geometry and topology
10Single Crystal accumulated slip
11Single Crystal 111 Pole Evolution
12Single Crystal Misorientation
Accumulated misorientation between crystals
initial orientation and current orientation
13Polycrystals
- Sample size 1e-3 x 1e-3 x 1e-3 (m3)
- Strain speed -50 (m/s)
- 3 grains along each side
- 91 grains were used
- parallel computing on 91 processor
- crystalline orientations were assigned randomly
to each grain - 10368 tetrahedral elements
- 42 reduction was reached.
14Polycrystals 111 Pole Evolution
15Polycrystals Texture (42 reduction)
Levels 1, 2, 3, 4
16Orientation Distribution Function
Texture components in rolled FCC metals.
17Polycrystals Texture (49 reduction)
Levels 1, 2, 3, 4, 5
TAYLOR
Same mesh to keep the same spatial resolution
(91 orientations per sampling point)
18Taylor and DNS
RD
RD
DNS
TAYLOR
TD
TD
TAYLOR (Blue) DNS (Red)
19Taylor - DNS
Levels 1, 2, 3
20Taylor and DNS
1
C
S
B
G
21Taylor and DNS
F
F
j 2
j 2
j 1
j 1
Sharper Texture
DNS
TAYLOR
22Taylor and DNS
Macroscopic Response
TAYLOR
DNS
Stiffer Response
- Number of orientations
- Spatial Resolution
Two issues
23Taylor Orientations
91 orientations per sampling point
or 384 orientations per sampling point
Negligible differences are observed
24Taylor Resolution
O subdivision 10,368 tetrahedral elements 1
subdivision 82,944 tetrahedral elements
More high-frequency dumping for coarser meshes
25DNS Orientations
384 grains 50 reduction
Increase number of orientations by reducing
spatial resolution within each grain. Extreme
case one element per grain,
26Taylor and DNS
Spatially resolved DNS
Lowest Resolution DNS
Taylor
27Role of Spatial Resolution on DNS
192 cells/grains
12,288 cells/grains
1,536 cells/grains
Required to properly capture the physics at
mesoscale
28Feasibility with current platform
103 elements
- 1Gel at our disposal (1Mel/proc x 1000
processors), 1000 elements - per direction
- 25K elements/2 grains to resolve inhomogeneous
plastic fields - (many more for subgrain structures, PSBs,
etc.) - 80K grains (40 per side)
- 2 mm specimen size for 50 mm grain size
103 e
109
103 e
2 grains (25K at subdivision 2) 1 full grain
(12K at subdivision 2) 8 1/8 of grain (12K at
subdivision 2)
29Summary and future work
- Two complementary full-field studies are
considered to investigate the texture evolution
Direct Numerical Simulation (DNS) of each crystal
and a mean field approach, Taylor-averaging
(TA). - DNS allows for explicit account of
microstructural features including grains shape,
size, orientation and distribution in macroscopic
response - Size Dependency
- Other metals (BCC)
- Grain Boundary effects
- Multiscale models
With the increasing of grains number, grains
sizes are reduced
Stainier, Cuitino and Ortiz (JMPS) 2002
30ABAQUS 1 element - 1 grain, 512 grains, 50
reduction