Integrated Experimental-Numerical Approach to Analyse the Evolution of Damage at Different Loading Conditions

Abstract

Understanding and predicting the evolution of damage in metals and rocks under various loading conditions is crucial to avoid unexpected failure in diverse industrial applications from aerospace and automotive to civil engineering. Furthermore, the progressive growth of damage in rocks can facilitate rock breakage in mining operations if the loading conditions are well understood and tuned through modelling strategies. This dissertation demonstrates how a combination of experimental activities, inverse methods, and numerical modelling can be used to enhance the accuracy of damage evolution simulations in metals and rocks. This work also aims to improve the understanding of how damage accumulates in granite when exposed to alternating current excitations.

A calibration strategy with objective regularization is proposed to identify material model parameters using scattered experimental data. The calibration strategy identifies the parameters of a high-cycle fatigue model for metals using experimental data from uniaxial and multiaxial cyclic loading tests. The results show an excellent match between the experiments and simulations of fatigue tests of two aluminium alloys. In addition, this study presents an experimental-numerical approach to analyse the evolution of fatigue damage in Kuru granite when exposed to high-voltage alternating current excitations. Mesostructural experimental data is used to inform a Finite Element model to achieve a numerical representation that closely matches the actual mesostructure of the rock. The model includes explicit representations of grain boundaries using cohesive elements. A traction separation relationship is proposed to simulate the damage accumulation of rock during the preconditioning process. The calibration strategy, initially developed to identify damage evolution in metals, was adapted to identify the parameters controlling the mechanical behaviour of the cohesive elements under monotonic and cyclic loading. The calibration strategy uses experimental data of the rock loaded under monotonic quasistatic compression, dynamic tension, and cyclic loads. The model reproduces the experimental data used for calibration very well, and the simulated fragmentation patterns of dynamic Brazilian disc specimens compare very well with the in-situ X-ray phase contrast images obtained during testing. The experiments evidence reductions in the dynamic tensile strength of granite Brazilian disc specimens exposed to alternating voltage and the model reproduces such strength reductions well. The results of the presented framework suggest that strength is reduced by the introduction of weakened spots in the rock structure during the AC excitations, which facilitate crack nucleation and coalescence during loading. Overall, the methodology presented in this dissertation enables accurate simulations of the damage evolution in both rock and metals during cyclic loading. Furthermore, it provides a framework to combine diverse experimental data into numerical models via inverse methods with enough flexibility to provide representative calibrations of constitutive models for rocks and metals under various loading conditions.