[SY-J3] Fundamentals of Generalized Particle (GP) Multiscale Methods with Applications to Analyses of Alternatively-Arranged Soft and Hard Layers
Invited
Alternatively-placed soft and hard layers are investigated using multiscale generalized particle methods, namely the GP and/or XGP methods (Fan 2009, Xu, Fan et al. 2016, Fan et al. 2017). In the methods, material models are divided as several domains with generalized particle of different scales. The distinguished characteristics of the GP methods are that each higher scale particle is formed by lumping the lower scale particle or atoms but their structures are all the same as the atomistic one, such as BCC, FCC and HCP. The fundamentals of this type of multiscale analysis are based on the local bottom-up multiscale analysis. Specifically, the position, mass and physical property of the upper-level particle is determined by all lower-level particles (e.g. atoms) within the individual particle’s neighbor link cell (NCL), thus the material behavior naturally follows the displacement and motion of these atoms consisted of the material and the numerical algorithm follows the typical molecular dynamics.
Extrinsic and intrinsic size effects has been successfully investigated to confirm the accuracy and efficiency of the proposed multiscale methods. Among several interesting observations of the atomistic/multiscale simulations, results show that defects in the soft, hard layers and their interface due to loading and manufacturing determine, to a large extent, their deformation and failure behavior. It has been found that some observations are different from previous predictions in the literature for a long period. For instance, simulations show that Langford’s famous prediction (1977) of dislocations appearance in cementite layer is not fully accurate. In addition, a direct check of the dislocation pile-up theory proposed by Eshelby, Frank and Nabarro (1951) indicate that the simulation-determined interface stress is much lower than their theory-predicted one.
Extrinsic and intrinsic size effects has been successfully investigated to confirm the accuracy and efficiency of the proposed multiscale methods. Among several interesting observations of the atomistic/multiscale simulations, results show that defects in the soft, hard layers and their interface due to loading and manufacturing determine, to a large extent, their deformation and failure behavior. It has been found that some observations are different from previous predictions in the literature for a long period. For instance, simulations show that Langford’s famous prediction (1977) of dislocations appearance in cementite layer is not fully accurate. In addition, a direct check of the dislocation pile-up theory proposed by Eshelby, Frank and Nabarro (1951) indicate that the simulation-determined interface stress is much lower than their theory-predicted one.