Multiscale microstructure-based modelling of cleavage in high strength steels

The need for more accurate cleavage modelling is particularly acute for a new generation of high-strength steels because they obtain their favourable properties through complex, multi-phase microstructures. One of the challenges in cleavage modelling is the strong sensitivity to material characteristics at the microlevel. This thesis proposes a novel framework that can quantitatively capture the interaction of complex microstructures in cleavage, allowing to calculate the probability of cleavage failure in high strength steels of complex multiphase microstructures. This method is validated with detailed experiments on different types of steels and on steels that have been subjected to heat treatments.
The framework is developed from a multi-barrier theory with the particular intention to include the effect of plastic strain and deactivation of hard inclusions. In order to quantitively determine the inclusion stress from far-field stress on a matrix, analytical equations are first derived. The proposed framework is first validated with examples of specimens taken from a S690 QT steel plate fractured at -100°C. Centreline segregation bands (CLs) appear in the middle-section specimens, containing smaller grains and elongated inclusion clusters. Two modelling approaches are compared to discuss the effect of CLs in cleavage modelling. A sensitivity study is performed to explore the influence of volume fractions, yield strength, and spacing of CLs. Then, the modelling approach is applied to determine the cleavage parameters across different types of steels. Cleavage parameters are compared among three tempered bainitic (S690) steels, an as-quenched martensitic steel, and a ferritic steel. The variation of cleavage parameters is discussed considering the influence of the matrix types and the hard particle types. Finally, cleavage simulations of the high strength steel after rapid cyclic heating and microstructures representing heat affect zones are performed. The simulations are compared with experiments that feature parametric variations of grain size, second particle size, and second particle density. The effect of different types of microstructures generated by heat treatments is quantitatively established.