Reduced order modeling and multiscale investigations of high performance alloys under monotonic and cyclic loading

Abstract

This dissertation is devoted to the development of multiscale modeling strategies for polycrystalline materials at both micro- and structural scales. At the microscale, a dislocation glide based crystal plasticity finite element (CPFE) model considering large deformation and cyclic loading conditions has been formulated. This model has been used to investigate the hysteresis response of nickel-based alloy IN 617 subjected to fatigue and creep-fatigue cycles at 950 oC. A new slip resistance evolution equation is proposed to account for the transient softening induced by solute-drag creep phenomenon. The CPFE model is further extended to capture the creep deformation and rupture of IN 617 at 950 oC by incorporating the dislocation climb mechanism that characterizes the creep strain evolution at relatively low stress levels, and a cohesive zone model that captures the intergranular creep damage describing the tertiary creep. The model is fully calibrated using experimental data, and numerical simulations are performed to illustrate the capability of the proposed model in capturing damage initiation and growth under creep loads. To upscale the behavior of polycrystalline materials from the scale of the microstructure to the structure or structural component, an eigenstrain based reduced order modeling (EHM) approach has been developed. EHM pre-computes certain microstructure information and approximates the microscale problem using a much smaller basis by prescribing spatial variation of inelastic response fields over the microstructure. To further improve the efficiency of the EHM and its scalability with respect to microstructure size, accelerated, sparse and scalable EHMs have been developed. The acceleration is achieved by introducing sparsity into the linearized reduced order system generated by the EHM through selectively considering the interactions between grains. The proposed approach results in a hierarchy of reduced models that recover EHM when full range of interactions are considered, and degrade to the Taylor model when all inter-grain interactions are neglected. Performance of the proposed approach is evaluated by comparing the results against the full EHM and CPFE simulations.