Vous êtes sur la page 1sur 11


• Traditional way of material optimization would be
• Come up with a structure and composition based on known theories
• Prepare lot of specimens
• Physical testing
• Advances in computational mechanics have paved way for doing all these in a virtual
environment – A virtual Lab
• behavior of materials at the macroscopic scale, is determined by the microscopic
interactions between atoms, mesoscopic phenomenon of dislocation dynamics etc

• Multiscale modeling aims to solve problems at the engineering (macro) scale while considering
the complexity of the microstructure
• Generally, two scales are considered in multiscale modelling:
• small scale, which is designed to capture the mechanical phenomena at the atomistic, molecular or
molecular cluster level
• large scale which is connected to continuous description
• well-established numerical methods have been developed over the years to handle the relevant
• In addition a coupling is required between these scales
Length Scales
• The nanoscopic (or atomic) scale (∼10−9 m),where
phenomena related to the behavior of electrons become
significant. At this scale, the interaction between particles
(electrons, atoms, etc.) is directly dictated by their
quantum-mechanical state.
• The microscopic scale (∼10−6 m), where phenomena
related to the behavior of atoms are considered. The
interaction between atoms is governed by empirical inter-
atomic potentials, which are generally derived from
Quantum Mechanical computations. The classic Newtonian
mechanics is used to compute the atoms displacements
and rotations.
• The mesoscopic scale (∼10−4 m), where phenomena
related to lattice defects are considered. At this scale,the
atomic degrees of freedom are not explicitly treated, and
only larger-scale entities (clusters of atoms, clusters of
molecules, etc.) are considered. The interaction between
particles are also described by the classic Newtonian
• The macroscopic scale (∼10−2 m), where macroscopic
phenomena which can be described by Continuum
Mechanics are considered. At this scale, the studied
physical systems are regarded as continua, whose the
behavior is described by constitutive laws.

Quantum Atomistic Mesoscopic

Methods Methods Methods

Quantum Molecular
Lattice Models
Monte Carlo Mechanics

Quantum Molecular
Particle Models
Chemistry Dynamics

Functional Monte Carlo Hybrid Models

Local Density Contact

Approximation Dynamics
• QM
• electrons are the players
• solving the electronic Schrödinger equation for atoms
and molecules
• Quantum Monte Carlo (QMC), Quantum Chemistry (QC)
• Density-Functional Theory (DFT)
• Local Density Approximation (LDA)
• Very accurate, very expensive
• but limited to few tens of atoms
• Timescale of pico seconds
• AM
• atoms are the players
• interaction laws between atoms can be described by
empirical interatomic potentials that encapsulate the effects
of bonding
• obtained from
• fitting material properties from experimental data
• quantum mechanical calculations
• distance between the atoms, angles between bonds, angles
between planes
• Molecular Mechan- ics (Statics) (MM), MD, Monte Carlo (MC)
• Relatively inexpensive
• Can process large systems involving 10^9 atoms
• limited to timescales of a few nanoseconds
• Mesoscopic Discrete Methods
• To overcome the timescale limitations
• lattice defects such as dislocations, crack propagation, and other
microstructural elements are the play- ers
• interaction laws are used usually derived by calibration or from
phenomenological theories that encompass the effects of interactions between
the atoms
• the atomic degrees of freedom are not explicitly treated and only larger-scale
particles are modelled
• Lattice Models- a solid is modeled by a set of nodes connected with truss or
beam elements
• Particle Models -These elements often have disc shape (in 2D) or spherical
shape (in 3D)
• Hybrid Models- The complementary advantages of these models has given rise
to a family of “hybrid” methods combining particle and lattice models
• Contact Dynamics - alternative approach based on a “non-smooth” formula-
tion of mutual exclusion and dry friction between elements
Discrete –Continuum Coupling (D-
• none of the DMs and CMs methods alone is able to
describe the entire system
• behavior of materials at the macroscopic scale, is
determined by the microscopic interactions
between atoms
• combination of discrete and continuum methods
• less expensive and more suitable to study realistic
engineering applications
• hierarchical approaches, concurrent approaches,
hybrid hierarchical-concurrent approaches
Hierarchical D-C Coupling
• hierarchy of computational approaches, large scale models use
information obtained from the fine scale models sequentially which are
weakly coupled
• sequential, serial, implicit, or message passing methods
• homogenization methods for multi-phase elastic heterogeneous media
are typical examples of hierarchical coupling methods
• the response of a representative volume element at the fine scale is first
computed over a range of expected inputs, and from this, a stress-strain
law is extracted which can be used at a coarser scale to describe the
mechanical behaviour of the homogenized material
Concurrent D-C Coupling
• linking numerical approaches having dif- ferent scales
together in a single combined model
• fine scale model communicates directly and instantly
with the coarse scale model though some coupling
• parallel or explicit cou- pling methods
• problem domain is often par- titioned into several
subdomains characterized by different scales and

Decomposition of the global domainΩG into two subdo- mainsΩD andΩC,modeled respectively by the
discrete model and the continuum model, with a bridging region ΩB, ug and tg are respectively the
given (prescribed) displacement and traction
• General notion is that MMCs have lesser ductility than
parent material
• Also MMCs have higher strength than parent material
• Both these properties have a very strong relationship to
particle type, size, volume fraction etc.
• Toughness or energy absorbing characteristics of a
material is a function of both strength and ductility
• Using the tools of Molecular dynamics and multiscale