• 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9

Multi-scale Modelling of Polycrystalline Fracture

Project Details

Materials Science
Other
ARCHER RAP
EPSRC
Co-investigator
Abstract

Interfacing HPC software for cellular automata with ParaFEM to build a world class open source platform for multi-scale modelling of materials.

A team from the University of Bristol, University of Manchester and the Science and Technology Facilities Council have been awarded 11 million allocation units by the UK National HPC Service ARCHER for a pilot project to develop the world's first massively parallel hybrid finite element/cellular automata (CAFE) simulation platform. This will involve interfacing two massively parallel software platforms, CGCAPACK and ParaFEM.

In a CAFE simulation, the finite element (FE) layer is used to compute the structural response, i.e. the nodal forces and displacements. The constitutive response(s) are calculated by the cellular automata layer(s). In each FE time increment the FE layer passes the strain rate, the deformation gradient and other fields, such as temperature, to the cellula automata (CA). The CA layer calculates the material response, updates the state of the microstructure, including crack propagation and void evolution, and returns back to the FE part a set of damage variables, which are typically used to reduce the element stiffness. Of critical importance is the simple fact that the size and time scales of the FE and the CA layers are completely independent. Thus the CAFE model can be used to simulate fracture processes where fast cleavage propagation is combined with relatively slow plastic flow, and very fast formation of adiabatic shear bands.

This work is novel because rather than creating a particular constitutive model for a particular physical process, it will deliver a generic, flexible, expandable, standard compliant and highly portable framework for creating multi-scale fracture models, expressly designed for massively parallel computing platforms. The proposed framework can be used for a wide range of structural integrity predictions for structures made of nano- and micro-crystalline materials.

To find out about our progress in this project, please consult the following two papers:

For further information and/or to discuss sponsoring this initiative through PhD studentships, industrial collaboration or consultancy, please contact:

Share This Page