Development of a Multi-Scale creep modeling approach

The high-temperature and low-stress creep deformation response of Incoloy 800H is characterized by two consecutive creep strain rate minima. The first minimum creep rate peak is attributed to a solid-solution hardening phenomenon, whereas the second one is associated to the steady-state creep rate of the material.
In this thesis, the accurate prediction of the complex microstructure-dependent creep behavior of 800H alloy is achieved through a multi-scale creep modeling approach involving a micromechanical mean-field model (MFC) and a macromechanical finite element constitutive law (VMVP).
A thorough experimental campaign addressing creep tests and microstructural characterization techniques was first organized to understand the high-temperature creep deformation response of 800H alloy. Microstructural characterization tests of specimens in as-received, furnace-aged and creep-tested conditions revealed a complex microstructure evolution involving aging, oxidation and nitridation. Additionally, curves showing non-classical characteristics were observed during low-stress creep tests.
Given the limited number of experimental creep curves, these were complemented with virtual creep curves generated using the MFC model developed during this PhD.
MFC computes the creep strain rate by solving a non-linear equation system where dependency on the mean dislocation density and precipitate kinetics is established. The latter was introduced from ex-situ thermokinetic simulations conducted in Thermo-Calc PRISMA. The 1st creep strain rate peak was assimilated as an internal stress contribution slowing down dislocation glide, whereas the 2nd peak was captured by an improved stress sensitivity formulation of the mechanical energy threshold in the mean dislocation glide velocity equation. The virtual tests generated with MFC demonstrated remarkable physical consistency and accuracy within the stress-temperature loadings of interest for this work, improving our understanding of the non-classical creep deformation response at creep loadings found between our creep experiments (interpolation) and at very-low stresses (extrapolation, where conventional tensile creep tests would take years to complete).
The database comprised of experimental and virtual creep curves was then used for the identification and validation of VMVP in Lagamine. To predict the non-classical creep behavior of 800H, VMVP incorporates a new viscosity function: a classical Norton law recovering the steady-state creep strain rate of the material modified by an activation function factor which reduces the initial creep strain rate to capture the 1st solid-solution hardening creep strain peak. The activation function Norton (afN) equation demonstrated good numerical stability while capturing the 2-peak creep behavior of Incoloy 800H.
The creep deformation response predicted by VMVP and identified through the MSCreep approach (VMVP-MSCreep) was found in good agreement with ULiège experiments and MFC virtual creep curves associated to high-temperature and low-stress creep tests. Post-assessment tests recommended by the European Creep Collaborative Committee (ECCC) show that the log.-log. prediction error for the steady-state creep rate and the time-to-1% creep deformation are found within the $\pm\ln(2)$. According to the ECCC standard, the predictability of the VMVP-MSCreep model is considered admissible. Additional experimental validation for complex loadings was ascertained by modeling the creep deflection of 6 cantilever beams. Numerical results achieved with VMVP-MSCreep were found in reasonable agreement with experiments. Additionally, the cantilever bending configuration allowed to establish a mesh size dependency on the stress gradient, thereby ensuring the consistency of the creep prediction of VMVP-MSCreep for realistic industrial applications.