Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userAtomistic-to-Meso Multi-Scale Data-Driven Graph Surrogate Modeling of Dislocation Glide
69
0
0.0
(
0
)
Added by
Eduardo Augusto Barros De Moraes
Publication date
2020
fields
Physics
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
From their birth in the manufacturing process, materials inherently contain defects that affect the mechanical behavior across multiple length and time-scales, including vacancies, dislocations, voids and cracks. Understanding, modeling, and real-time simulation of the underlying stochastic micro-structure defect evolution is therefore vital towards multi-scale coupling and propagating numerous sources of uncertainty from atomistic to eventually aging continuum mechanics. We develop a graph-based surrogate model of dislocation glide for computation of dislocation mobility. We model an edge dislocation as a random walker, jumping between neighboring nodes of a graph following a Poisson stochastic process. The network representation functions as a coarse-graining of a molecular dynamics simulation that provides dislocation trajectories for an empirical computation of jump rates. With this construction, we recover the original atomistic mobility estimates, with remarkable computational speed-up and accuracy. Furthermore, the underlying stochastic process provides the statistics of dislocation mobility associated to the original molecular dynamics simulation, allowing an efficient propagation of material parameters and uncertainties across the scales.
rate research
Read More
The stress-driven motion of dislocations in crystalline solids, and thus the ensuing plastic deformation process, is greatly influenced by the presence or absence of various point-like defects such as precipitates or solute atoms. These defects act as obstacles for dislocation motion and hence affect the mechanical properties of the material. Here we combine molecular dynamics studies with three-dimensional discrete dislocation dynamics simulations in order to model the interaction between different kinds of precipitates and a $frac{1}{2}langle 1 1 1rangle$ ${1 1 0}$ edge dislocation in BCC iron. We have implemented immobile spherical precipitates into the ParaDis discrete dislocation dynamics code, with the dislocations interacting with the precipitates via a Gaussian potential, generating a normal force acting on the dislocation segments. The parameters used in the discrete dislocation dynamics simulations for the precipitate potential, the dislocation mobility, shear modulus and dislocation core energy are obtained from molecular dynamics simulations. We compare the critical stresses needed to unpin the dislocation from the precipitate in molecular dynamics and discrete dislocation dynamics simulations in order to fit the two methods together, and discuss the variety of the relevant pinning/depinning mechanisms.
The validity of the structure-property relationships governing the deformation behavior of bcc metals was brought into question with recent {it ab initio} density functional studies of isolated screw dislocations in Mo and Ta. These existing relationships were semiclassical in nature, having grown from atomistic investigations of the deformation properties of the groups V and VI transition metals. We find that the correct form for these structure-property relationships is fully quantum mechanical, involving the coupling of electronic states with the strain field at the core of long $a/2<111>$ screw dislocations.
We develop a model for the gliding of dislocations and plasticity in solid He-4. This model takes into account the Peierls barrier, multiplication and interaction of dislocations, as well as classical thermally and mechanically activated processes leading to dislocation glide. We specifically examine the dc stress-strain curve and how it is affected by temperature, strain rate, and dislocation density. As a function of temperature and shear strain, we observe plastic deformation and discuss how this may be related to the experimental observation of elastic anomalies in solid hcp He-4 that have been discussed in connection with the possibility of supersolidity or giant plasticity. Our theory gives several predictions for the dc stress strain curves, for example, the yield point and the change in the work-hardening rate and plastic dissipation peak, that can be compared directly to constant strain rate experiments and thus provide bounds on model parameters.
Dislocation velocities and mobilities are studied by Molecular Dynamics simulations for edge and screw dislocations in pure aluminum and nickel, and edge dislocations in Al-2.5%Mg and Al-5.0%Mg random substitutional alloys using EAM potentials. In the pure materials, the velocities of all dislocations are close to linear with the ratio of (applied stress)/(temperature) at low velocities, consistent with phonon drag models and quantitative agreement with experiment is obtained for the mobility in Al. At higher velocities, different behavior is observed. The edge dislocation velocity remains dependent solely on (applied stress)/(temperature) up to approximately 1.0 MPa/K, and approaches a plateau velocity that is lower than the smallest forbidden speed predicted by continuum models. In contrast, above a velocity around half of the smallest continuum wave speed, the screw dislocation damping has a contribution dependent solely on stress with a functional form close to that predicted by a radiation damping model of Eshelby. At the highest applied stresses, there are several regimes of nearly constant (transonic or supersonic) velocity separated by velocity gaps in the vicinity of forbidden velocities; various modes of dislocation disintegration and destabilization were also encountered in this regime. In the alloy systems, there is a temperature- and concentration-dependent pinning regime where the velocity drops sharply below the pure metal velocity. Above the pinning regime but at moderate stresses, the velocity is again linear in (applied stress)/(temperature) but with a lower mobility than in the pure metal.
Atomistic computations of the Peierls stress in fcc metals are relatively scarce. By way of contrast, there are many more atomistic computations for bcc metals, as well as mixed discrete-continuum computations of the Peierls-Nabarro type for fcc metals. One of the reasons for this is the low Peierls stresses in fcc metals. Because atomistic computations of the Peierls stress take place in finite simulation cells, image forces caused by boundaries must either be relaxed or corrected for if system size independent results are to be obtained. One of the approaches that has been developed for treating such boundary forces is by computing them directly and subsequently subtracting their effects, as developed by V. B. Shenoy and R. Phillips [Phil. Mag. A, 76 (1997) 367]. That work was primarily analytic, and limited to screw dislocations and special symmetric geometries. We extend that work to edge and mixed dislocations, and to arbitrary two-dimensional geometries, through a numerical finite element computation. We also describe a method for estimating the boundary forces directly on the basis of atomistic calculations. We apply these methods to the numerical measurement of the Peierls stress and lattice resistance curves for a model aluminum (fcc) system using an embedded-atom potential.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
