Development and identification of the cellular automata phase transformation model

Danuta Szeliga,Lukasz Madej,Maciej Pietrzyk,Mateusz Sitko,Roman Kuziak,Wolfgang Bleck,Yuling Chang

doi:10.25518/esaform21.2640

Abstract

The development and identification of the complex microscale austenite to ferrite transformation model during continuous cooling based on the Cellular Automata method and DigiCore library is the main goal of the work. The model is designed to predict phase transformation from a fully austenitic range and involves nucleation of ferrite grains with their further growth. The major driving force for the CA grain growth is based on the carbon concentration differences across the microstructure. The model parameters are identified with the inverse analysis method with the goal function defined on the basis of the dilatometric investigation. The basic assumption of the developed model, experimental procedure, as well as subsequent identification stages, are presented within the work.

Highlights

The development of reliable material models for metal forming simulations has been in scientists' interest for a number of years [1,2,3]
This work is a step towards the development of the full-field cellular automata phase transformation model in steels focused on transformation from austenite to ferrite phase under continuous colling
A set of dilatometric experiments was conducted to reveal the behavior of investigated material during cooling in a small cooling range regime. These experiments supported by the ThermoCalc calculations provided the required set of Cellular Automata (CA) model boundary conditions and data for the definition of the goal function

Summary

Introduction

The development of reliable material models for metal forming simulations has been in scientists' interest for a number of years [1,2,3]. There are three different types of boundary conditions in the classical CA method: absorbing, where the state of cells located at the edges of the computational domain are adequately fixed with a specific state to absorb moving quantities; reflective, where the state of cells located at the edges are adequately fixed to reflect moving quantities; and most commonly used in material science application the periodic boundary conditions In the latter case, the CA neighborhood takes into account cells located at the opposite edges of the computational domain and assume their interaction. It should be mentioned that all the available CA models for the material science applications are based on the same common major elements, e.g., CA space as a computational domain, type of neighborhood, transition rule definitions, and additional components like absorbing or periodic boundary conditions [5,6,7]. The DigiCore library is based on the unified data-structures, allowing all the developed microstructure evolution models to be coupled for simulation of complex thermo-mechanical operations

DigiCore library

CA phase transformation model

Determination of equilibrium diagram

Inverse analysis

Conclusions