A Multiscale Parametric Framework for Thermo-Mechanical Fracture Prediction

Thesis Abstract

The study addresses the critical challenge of predicting thermo-mechanical fracture in heterogeneous materials under complex loading by integrating multiscale phenomena into a cohesive parametric framework. Recognizing that fracture behavior emerges from interactions across microscale defect structures, mesoscale grain boundaries, and macroscale structural constraints, the research aims to develop a predictive model that preserves essential physics while enabling efficient design optimization. The objectives are to (i) formulate a multiscale parametric framework that couples thermo-poro-mechanical fields with fracture indicators, (ii) incorporate two complementary theoretical lenses—continuum damage mechanics and cohesive zone modeling—within a unified hierarchical structure, (iii) calibrate and validate the framework against experimental data from metallic and composite specimens under varied thermal and mechanical loading, and (iv) demonstrate the framework’s utility in predicting crack initiation, growth paths, and critical load for design safety margins. The methodology adopts a sequential mixed-methods approach anchored in computational mechanics and empirical validation. The population comprises metallic alloys (titanium alloy Ti-6Al-4V and nickel-based superalloy Inconel 718) and carbon/epoxy composites subjected to controlled thermal gradients and mechanical loads in a servo-hydraulic testing rig. A sample of 60 specimens (20 per material class) is used for experiments, with each specimen instrumented for digital image correlation (DIC) to capture full-field strain distributions and infrared thermography to record surface temperature evolution. The data collection instruments include high-resolution cameras for DIC, infrared thermography systems, thermocouple arrays for internal temperature monitoring, and load-displacement sensors to determine fracture initiation and crack growth. Complementary microstructural data are obtained from electron backscatter diffraction (EBSD) and scanning electron microscopy (SEM) to characterize grain orientations and defect populations. Analytical methods combine computational simulation with statistical calibration. The framework integrates a multiscale finite element model wherein microscale defect statistics inform mesoscale cohesive zone properties, which in turn drive macroscale fracture response. Thermo-mechanical coupling is implemented via coupled heat diffusion and solid mechanics equations, with material constitutive behavior represented by continuum damage mechanics (CDM) at the macroscale and cohesive zone models (CZM) at interfaces. Model calibration employs Bayesian updating to fuse prior material knowledge with experimental observations, enabling probabilistic prediction of fracture metrics. Regression analysis and analysis of variance (ANOVA) are used to assess the sensitivity of fracture outcomes to temperature gradients, loading rates, and microstructural features. Model validation hinges on quantitative agreement with experimentally observed crack initiation times, crack lengths, and load-displacement curves, assessed via root mean square error (RMSE) and coefficient of determination (R^2). Nonlinear optimization techniques are employed to identify parametric sets that minimize mismatch between predictions and observations. Key expected findings include robust correlations between microscale defect statistics and macroscale fracture thresholds, demonstrating that incorporating temperature-dependent CZM parameters significantly enhances prediction accuracy under thermo-mechanical loading. It is anticipated that the multiscale framework will capture crack initiation at lower nominal stresses in thermally stressed composites due to differential thermal strains, while metallic alloys will exhibit distinct crack paths governed by grain boundary orientation distributions. The study also expects to reveal material-specific calibration patterns showing that CDM parameters are more influential for metals, whereas CZM properties dominate for composites. The study contributes to knowledge by delivering a rigorously validated multiscale parametric framework that bridges microscale defect populations, mesoscale interface behavior, and macroscale fracture response under thermo-mechanical loads. It advances theory by integrating CDM and CZM within a coherent hierarchical model and demonstrates a practical pathway for material-specific calibration using a combination of experimental and computational data. Practical implications include improved design safety margins for aerospace and automotive components subjected to thermal cycling. Recommendations emphasize extending the framework to include anisotropic diffusion effects, environmental corrosion interactions, and real-time predictive updates using online monitoring data to support proactive maintenance decisions.