<p>  </p> <p>There has been a growing interest within the aerospace industry for shifting towards a digital twin approach, for reliable assessment of individual components during the product lifecycle - across design, manufacturing, and in-service maintenance, repair & overhaul (MRO) stages. The transition towards digital twins relies on continuous updating of the product lifecycle datasets and interoperable exchange of data applicable to components, thereby permitting engineers to utilize current state information to make more-informed downstream decisions. In this thesis, we primarily develop a framework to store, track, update, and retrieve product lifecycle data applicable to a serialized component, its features, and individual locations. </p> <p>From a structural integrity standpoint, the fatigue performance of a component is inherently tied to the component geometry, its material state, and applied loading conditions. The manufacturing process controls the underlying material microstructure, which in turn governs the mechanical properties and ultimately the performance. The processing also controls the residual stress distributions within the component volume, which influences the durability and damage tolerance of the component. Hence, we have demonstrated multiple use cases for fatigue life assessment of critical aerospace components, by using the developed framework for efficiently tracking and retrieving (i) the current geometric state, (ii) the material microstructure state, and (iii) residual stress distributions.</p> <p>Model-based definitions (MBDs) present opportunities to capture both geometric and non-geometric data using 3D computer-aided design (CAD) models, with the overarching aim to disseminate product information across different stages of the lifecycle. MBDs can potentially eliminate error-prone information exchange associated with traditional paper-based drawings and improve the fidelity of component details, captured using 3D CAD models. However, current CAD capabilities limit associating the material information with the component’s shape definition. Furthermore, the material attributes of interest, viz., material microstructures and residual stress distributions, can vary across the component volume. To this end, in the first part of the thesis, we implement a CAD-based tool to store and retrieve metadata using point objects within a CAD model, thereby creating associations to spatial locations within the component. The tool is illustrated for storage and retrieval of bulk residual stresses developed during the manufacturing of a turbine disk component, acquired from process modeling and characterization. Further, variations in residual stress distribution owing to process model uncertainties have been captured as separate instances of the disk’s CAD models to represent part-to-part variability as an analogy to track individual serialized components for digital twins. The propagation of varying residual stresses from these CAD models within the damage tolerance analysis performed at critical locations in the disk has been demonstrated. The combination of geometric and non-geometric data inside the MBD, via storage of spatial and feature varying information, presents opportunities to create digital replica or digital twin(s) of actual component(s) with location-specific material state information.</p> <p>To fully realize a digital twin description of components, it is crucial to dynamically update information tied to a component as it evolves across the lifecycle, and subsequently track and retrieve current state information. Hence, in the second part of the thesis, we propose a dynamic data linking approach to include material information within the MBDs. As opposed to storing material datasets directly within the CAD model in the previous approach, we externally store and update the material datasets and create data linkages between material datasets and features within the CAD models. To this end, we develop a model-based feature information network (MFIN), a software agnostic framework for linking, updating, searching, and retrieving of relevant information across a product’s lifecycle. The use case of a damage tolerance analysis for a compressor bladed-disk (blisk) is demonstrated, wherein Ti-6Al-4V blade(s) are linear friction welded to the Ti-6Al-4V disk, comprising well-defined regions exhibiting grain refinement and high residuals stresses. By capturing the location-specific microstructural information and residual stress fields at the weld regions, this information was accessed within the MFIN and used for downstream damage tolerant analysis. The introduction of the MFIN framework facilitates access to dynamically evolving as well as location-specific data for use within physics-based models.</p> <p>In the third part of thesis, we extend the MFIN framework to enable a physics-based, microstructure sensitive and location-specific fatigue life analysis of a component. Traditionally, aerospace components are treated as monolithic structures during lifing, wherein microstructural information at individual locations are not necessarily considered. The resulting fatigue life estimates are conservative and associated with large uncertainty bounds, especially in components with gradient microstructures or distinct location-specific microstructures, thereby leading to under usage of the component’s capabilities. To improve precision in the fatigue estimates, a location-specific lifing framework is enabled via MFIN, for tracking and retrieval of microstructural information at distinct locations for subsequent use within a crystal plasticity-based fatigue life prediction model. A use case for lifing dual-microstructure heat treated LSHR turbine disk component is demonstrated at two locations, near the bore (fine grains) and near the rim (coarse grains) regions. We employ the framework to access (a) the grain size statistics and (b) the macroscopic strain fields to inform precise boundary conditions for the crystal plasticity finite-element analysis. The illustrated approach to conduct a location-specific predictive analysis of components presents opportunities for tailoring the manufacturing process and resulting microstructures to meet the component’s targeted requirements.</p> <p>For reliably conducting structural integrity analysis of a component, it is crucial to utilize their precise geometric description. The component geometries encounter variations from nominal design geometries, post manufacturing or after service. However, traditionally, stress analyses are based on nominal part geometries during assessment of these components. In the last part of the thesis, we expand the MFIN framework to dynamically capture deviations in the part geometry via physical measurements, to create a new instance of the CAD model and the associated structural analysis. This automated workflow enables engineers for improved decision-making by assessing (i) as-manufactured part geometries that fall outside of specification requirements during the materials review board or (ii) in-service damages in parts during the MRO stages of the lifecycle. We demonstrate a use case to assess the structural integrity of a turbofan blade that had experienced foreign object damage (FOD) during service. The as-designed geometry was updated based on coordinate measurements of the damaged blade surfaces, by applying a NURBS surface fit, and subsequently utilized for downstream finite-element stress analysis. The ramifications of the FOD on the local stresses within the part are illustrated, providing critical information to the engineers for their MRO decisions. The automated flow of information from geometric inspection within structural analysis, enabled by MFIN, presents opportunities for effectively assessing products by utilizing their current geometries and improving decision-making during the product lifecycle.</p>

Partial discharge (PD) measurements have been widely accepted as an efficient online insulation condition assessment method in high voltage equipment. Two sets of experimental PD measuring setups were established with the aim to study the variations in the partial discharge characteristics over the insulation degradation in terms of the physical phenomena taking place in PD sources, up to the point of failure. Probabilistic lifetime modeling techniques based on classification, regression and multivariate time series analysis were performed for a system of PD response variables, i.e. average charge, pulse repetition rate, average charge current, and largest repetitive discharge magnitude over the data acquisition period. Experimental lifelong PD data obtained from samples subjected to accelerated degradation was used to study the dynamic trends and relationships among those aforementioned response variables. Distinguishable data clusters detected by the T-Stochastics Neighborhood Embedding (tSNE) algorithm allows for the examination of the state-of-the-art modeling techniques over PD data. The response behavior of trained models allows for distinguishing the different stages of the insulation degradation. An alternative approach utilizing a multivariate time series analysis was performed in parallel with Classification and Regression models for the purpose of forecasting PD activity (PD response variables corresponding to insulation degradation). True observed data and forecasted data mean values lie within the 95th percentile confidence interval responses for a definite horizon period, which demonstrates the soundness and accuracy of models. A life-predicting model based on the cointegrated relations between the multiple response variables, trained model responses correlated with experimentally evaluated time-to-breakdown values and well-known physical discharge mechanisms, can be used to set an emergent alarming trigger and as a step towards establishing long-term continuous monitoring of partial discharge activity. Furthermore, this dissertation also proposes an effective PD monitoring system based on wavelet and deflation compression techniques required for an optimal data acquisition as well as an algorithm for high-scale, big data reduction to minimize PD data size and account only for the useful PD information. This historically recorded useful information can thus be used for, not only postault diagnostics, but also for the purpose of improving the performance of modelling algorithms as well as for an accurate threshold detection.