In the combustion zone of industrial- and aero- gas turbines, thermomechanical fatigue (TMF) is the dominant damage mechanism. Thermomechanical fatigue is a coupling of independent creep, fatigue, and oxidation damage mechanisms that interact and accelerate microstructural degradation. A mixture of intergranular cracking due to creep, transgranular cracking due to fatigue, and surface embrittlement due to oxidation is often observed in gas turbine components removed from service. The current maintenance scheme for gas turbines is to remove components from service when any criteria (elongation, stress-rupture, crack length, etc.) exceed the designed maximum allowable. Experimental, theoretical, and numerical analyses are performed to determine the state of the component as it relates to each criterion (a time consuming process). While calculating these metrics individually has been successful in the past, a better approach would be to develop a unified mechanical modeling that incorporates the constitutive response, microstructural degradation, and rupture of the subject material via a damage variable used to predict the cumulative “damage state” within a component. This would allow for a priori predictions of microstructural degradation, crack propagation/arrest, and component-level lifing. In this study, a unified mechanical model for creep-fatigue (deformation, cracking, and rupture) is proposed. It is hypothesized that damage quantification techniques can be used to develop accurate creep, fatigue, and plastic/ductile cumulative- nonlinear- damage laws within the continuum damage mechanics principle. These damage laws when coupled with appropriate constitutive equations and a degrading stiffness tensor can be used to predict the mechanical state of a component. A series of monotonic, creep, fatigue, and tensile-hold creepfatigue tests are obtained from literature for 304 stainless steel at 600°C (1112°F) in an air. iv Cumulative- nonlinear- creep, fatigue, and a coupled creep-fatigue damage laws are developed. The individual damage variables are incorporated as an internal state variable within a novel unified viscoplasticity constitutive model (zero yield surface) and degrading stiffness tensor. These equations are implemented as a custom material model within a custom FORTRAN onedimensional finite element code. The radial return mapping technique is used with the updated stress vector solved by Newton-Raphson iteration. A consistent tangent stiffness matrix is derived based on the inelastic strain increment. All available experimental data is compared to finite element results to determine the ability of the unified mechanical model to predict deformation, damage evolution, crack growth, and rupture under a creep-fatigue environment.

Remaining useful life (RUL) prediction is a problem that researchers in the prognostics and health management (PHM) community have been studying for decades. Both physics-based and data-driven methods have been investigated, and in recent years, deep learning has gained significant attention. When sufficiently large and diverse datasets are available, deep neural networks can achieve state-of-the-art performance in RUL prediction for a variety of systems. However, for end users to trust the results of these models, especially as they are integrated into safety-critical systems, RUL prediction uncertainty must be captured. This work explores an approach for estimating both epistemic and heteroscedastic aleatoric uncertainties that emerge in RUL prediction deep neural networks and demonstrates that quantifying the overall impact of these uncertainties on predictions reveal valuable insight into model performance. Additionally, a study is carried out to observe the effects of RUL truth data augmentation on perceived uncertainties in the model.

Many real-world systems experience deterioration with usage and age, which often leads to low product quality, high production cost, and low system availability. Most previous maintenance and reliability models in the literature do not incorporate condition monitoring information for decision making, which often results in poor failure prediction for partially observable deteriorating systems. For that reason, the development of fault prediction and control scheme using condition-based maintenance techniques has received considerable attention in recent years. This research presents a new framework for predicting failures of a partially observable deteriorating system using Bayesian control techniques. A time series model is fitted to a vector observation process representing partial information about the system state. Residuals are then calculated using the fitted model, which are indicative of system deterioration. The deterioration process is modeled as a 3-state continuous-time homogeneous Markov process. States 0 and 1 are not observable, representing healthy (good) and unhealthy (warning) system operational conditions, respectively. Only the failure state 2 is assumed to be observable. Preventive maintenance can be carried out at any sampling epoch, and corrective maintenance is carried out upon system failure. The form of the optimal control policy that maximizes the long-run expected average availability per unit time has been investigated. It has been proved that a control limit policy is optimal for decision making. The model parameters have been estimated using the Expectation Maximization (EM) algorithm. The optimal Bayesian fault prediction and control scheme, considering long-run average availability maximization along with a practical statistical constraint, has been proposed and compared with the age-based replacement policy. The optimal control limit and sampling interval are calculated in the semi-Markov decision process (SMDP) framework. Another Bayesian fault prediction and control scheme has been developed based on the average run length (ARL) criterion. Comparisons with traditional control charts are provided. Formulae for the mean residual life and the distribution function of system residual life have been derived in explicit forms as functions of a posterior probability statistic. The advantage of the Bayesian model over the well-known 2-parameter Weibull model in system residual life prediction is shown. The methodologies are illustrated using simulated data, real data obtained from the spectrometric analysis of oil samples collected from transmission units of heavy hauler trucks in the mining industry, and vibration data from a planetary gearbox machinery application.

Many real-world systems experience deterioration with usage and age, which often leads to low product quality, high production cost, and low system availability. Most previous maintenance and reliability models in the literature do not incorporate condition monitoring information for decision making, which often results in poor failure prediction for partially observable deteriorating systems. For that reason, the development of fault prediction and control scheme using condition-based maintenance techniques has received considerable attention in recent years. This research presents a new framework for predicting failures of a partially observable deteriorating system using Bayesian control techniques. A time series model is fitted to a vector observation process representing partial information about the system state. Residuals are then calculated using the fitted model, which are indicative of system deterioration. The deterioration process is modeled as a 3-state continuous-time homogeneous Markov process. States 0 and 1 are not observable, representing healthy (good) and unhealthy (warning) system operational conditions, respectively. Only the failure state 2 is assumed to be observable. Preventive maintenance can be carried out at any sampling epoch, and corrective maintenance is carried out upon system failure. The form of the optimal control policy that maximizes the long-run expected average availability per unit time has been investigated. It has been proved that a control limit policy is optimal for decision making. The model parameters have been estimated using the Expectation Maximization (EM) algorithm. The optimal Bayesian fault prediction and control scheme, considering long-run average availability maximization along with a practical statistical constraint, has been proposed and compared with the age-based replacement policy. The optimal control limit and sampling interval are calculated in the semi-Markov decision process (SMDP) framework. Another Bayesian fault prediction and control scheme has been developed based on the average run length (ARL) criterion. Comparisons with traditional control charts are provided. Formulae for the mean residual life and the distribution function of system residual life have been derived in explicit forms as functions of a posterior probability statistic. The advantage of the Bayesian model over the well-known 2-parameter Weibull model in system residual life prediction is shown. The methodologies are illustrated using simulated data, real data obtained from the spectrometric analysis of oil samples collected from transmission units of heavy hauler trucks in the mining industry, and vibration data from a planetary gearbox machinery application.

Tato diplomová práce se zabývá posouzením tepelně-mechanické únavy výfukového potrubí. Nejprve byla provedena rešeršní studie, ve které je rozebrán fenomén tepelně-mechanické únavy. Byly prezentovány hlavní mechanismy poškození a přístupy k jejich modelování. Diskutována byla i specifická chování materiálu vystavenému tepelně-mechanickému zatěžování. Byl vypracován přehled vhodných modelů materiálu a modelů únavové životnosti společně s algoritmem predikce tepelně-mechanické únavy komponenty. Poté byl tento teoretický základ aplikován na praktický případ výfukového potrubí podléhajícího tepelně-mechanickému zatěžování. Dva tepelně závislé elasto-plastické modely materiálu byly nakalibrovány a validovány na základě experimentálních dat. Byl vytvořen diskretizovaný konečnoprvkový model sestavy výfukového potrubí. Model tepelných okrajových podmínek byl předepsán na základě výpočtů ustáleného sdruženého přestupu tepla. Slabě sdružená tepelně-deformační úloha byla vyřešena metodou konečných prvků pro oba modely materiálů. Bylo použito paradigma nesvázaného modelu únavy, které je vhodné pro nízkocyklovou únavu. Životnost byla tedy vyhodnocena jako součást post-procesoru. Použity byly dva modely únavové životnosti – energeticky založený model a deformačně založený model. Získané hodnoty životnosti byly porovnány vzhledem k použitým modelům materiálu a modelům únavové životnosti. Nakonec jsou diskutovány závěry této práce, oblasti dalšího výzkumu a navrženy možnosti na zlepšení použitých přístupů.

The thesis deals with the modeling and prediction of corrosion of radiation tube snake in the heating furnace. Specifically it is focused on vertical cylindrical furnace which is included in the catalytic hydrocracking unit and serves for heating aggressive circulation gas which is the cause of high temperature corrosion. An important basis for the creation of computational models are available records about the operation of the furnace and about the corrosion and degradation mechanisms during the lifetime of the tube system in furnace. Such information enables the creation of a computational model which is based on the prediction of high-temperature corrosive damage of radiation tube snake. The computational model involving all relevant factors may serve as the basis for a predictive life management system of radiation snakes in the heating furnace.

One of prominent issues related to failures in nuclear power components is attributed to material degradation due the aggressive environment conditions, and mechanical stresses. For instance, reactor core support components, such as fuel claddings, are under prolonged exposure to an intense neutron field from the fission of fuel and operate at elevated temperature under fatigue loadings caused by start up, shut down, and unscheduled emergency shut down. Additionally, exposure to highluence neutron radiation can lead to microscopic defects that result in material hardening and embrittlement, which significantly affects the physical and mechanical properties of the materials, resulting in further reduction in fatigue life of reactor structural components. The effects of fatigue damage on material deterioration can be further exacerbated by the presence of thermal loading, hold-time, and high-temperature water coolant environments. In this study, uniaxial fatigue models were used to predict fatigue behavior based only on simple monotonic properties including ultimate tensile strength and Brinell hardness. Two existing models, the Bäumel Seeger uniform material law and the Roessle Fatemi hardness method, were employed and extended to include the effects of test temperature, neutron irradiation fluence, irradiation induced helium and irradiation induced swellings on fatigue life of austenitic stainless steels. Furthermore, a methodology to estimate fatigue crack length using a strip-yield based model is presented. This methodology is also extended to address the effect of creep deformation in a presence of hold- times, and expanded to include the effects of irradiation and water environment. Reasonable fatigue life predictions and crack growth estimations are obtained for irradiated austenitic stainless steels types 304, 304L, and 316, when compared to the experimental data available in the literature. Lastly, a failure analysis methodology of a mixer unit shaft made of AISI 304 stainless steel is also presented using a conventional 14-step failure analysis approach. The primary mode of failure is identified to be intergranular stress cracking at the heat affected zones. A means of circumventing this type of failure in the future is presented.

No description available.

[pt] Critérios de fadiga multiaxial para vida infinita tinham por objetivo apenas avaliar a ocorrência de fratura em um componente mecânico quando submetido a carregamentos multiaxiais totalmente reversíveis. Carpinteri e Spagnoli propuseram uma modificação em seu próprio modelo, substituindo por outros parâmetros os limites de resistência à fadiga em flexão f−1 e torção t−1 para ensaios totalmente reversíveis, introduzindo na equação uma variável nf que permitiu realizar uma previsão de vida em fadiga finita. O objetivo do presente estudo é verificar experimentalmente a consistência dessa modificação. A metodologia consistiu em obter experimentalmente curvas de Wohler para tração e torção referentes ao aço DIN 42CrMo4 a fim de obter os parâmetros m e m(asterisco), que são os coeficientes angulares das mesmas em escala log-log, produzindo meios para a aplicação do critério. Como o equacionamento do modelo não apresenta solução analítica, foi desenvolvido uma solução numérica para obter junto ao critério uma previsão teórica de vida em fadiga. Adicionalmente, o estudo busca discutir acerca de uma possível relação direta entre amplitude de tensão normal, amplitude de tensão cisalhante e número de ciclos para falha. O modelo em si apresentou consistência parcial com os experimentos, tendo sido assertivo nos ensaios de torção pura, mas discrepante em ensaios de tração pura. Para carregamentos combinados, houve razoável precisão em dois casos e grande dispersão em outra, mas a avaliação final depende de mais pontos experimentais. / [en] Infinite-life multiaxial fatigue criteria had only the ability to evaluate whether or not fatigue failure is to occur to a mechanical componente once subjected to multiaxial fatigue loadings. Carpinteri e Spagnoli proposed a modification to their own model, substituting both fully reversed bending and torsion fatigue endurance limits, f1 and t−1 respectively, introducing into the equation a new variable nf, allowing the model to predict the fatigue-life of the mechanical component. The main goal of the presente study is to assess the accuracy of the modified model via experiments. The research methodology consisted in determining m and m (asterisk), which are the slopes of the S-N curves for fully reversed bending and torsion experiments on regards to DIN 42CrMo4 steel when plotted into a log-log scale, providing means to apply the model. Since there is no analytic solution to the model, the criterion s equation has to be solved numerically. Furthermore, the present study discusses the possibility of a direct relation between amplitude of normal stress, amplitude of shear stress and number of cycles to failure. The modified Carpinteri & Spagnoli s criterion proved itself to be partially consistent, presenting both accurate predictions of torsional fatigue-life and discrepant results for axial loadings. For combined loadings, the model provided two consistent results while another experimental point was proved far off. The final assessment on regards to the model s accuracy depends on more experimental points.

Reducing unscheduled maintenance is important for aircraft operators. There are significant costs if flights must be delayed or cancelled, for example, if spares are not available and have to be shipped across the world. This thesis describes three methods of aircraft health condition monitoring and prediction; one for system monitoring, one for forecasting and one combining the two other methods for a complete monitoring and prediction process. Together, the three methods allow organizations to forecast possible failures. The first two use decision trees for decision-making and genetic optimization to improve the performance of the decision trees and to reduce the need for human interaction. Decision trees have several advantages: the generated code is quickly and easily processed, it can be altered by human experts without much work, it is readable by humans, and it requires few resources for learning and evaluation. The readability and the ability to modify the results are especially important; special knowledge can be gained and errors produced by the automated code generation can be removed. A large number of data sets is needed for meaningful predictions. This thesis uses two data sources: first, data from existing aircraft sensors, and second, sound and vibration data from additionally installed sensors. It draws on methods from the field of big data and machine learning to analyse and prepare the data sets for the prediction process.