Bainitic steels have been studied for nearly a century. In fact, they are highly valued for their unique combination of toughness and mechanical strength, making them ideal for a wide variety of applications, particularly in the nuclear industry. Despite the extensive knowledge gained about bainitic steel, many unanswered questions still remain. In particular, multiple observations and measurements suggest that strains and stresses are likely to play an important role. Although they can be introduced into the usual mean-field models used to predict kinetics, the calibration of the terms that depend on them is very delicate, as is the interpretation of their values. The reason for this is that the proper evaluation of these terms must take into account the complexity of the mechanical fields in a polycrystalline, multiphase material.

Motivated by the previous statement, the present study focuses on the modeling of bainite formation in steels at the scale of the initial austenite polycrystal using a cellular automaton coupled with an FFT-based elasto-visco-plastic solver. The model is designed to efficiently handle 3D polycrystals with significant numbers of austenite grains. The morphology of the bainite blocks is simplified by considering ellipsoids or plates. The blocks are assumed to elongate in the main growth directions given by the phenomenological theory of martensite crystallography at a constant rate under isothermal conditions given by adapting either autocatalytic nucleation models or diffusion-controlled models. Independent of the growth law, carbon partitioning is considered using mass balance and non-equilibrium thermodynamic conditions at interfaces. Furthermore, in a first step, the mechanical fields are considered in the selection of block orientations. In a second step, they are introduced ad hoc into the growth laws.