Search

Object detection is an important research problem in computer vision. Convolutional Neural Networks (CNN) based deep learning models could be used for this problem, but it would require a large number of manual annotated objects for training or fine-tuning. Unfortunately, fine-grained manually annotated objects are not available in many cases. Usually, it is possible to obtain imperfect initialized detections by some weak object detectors using some weak supervisions like the prior knowledge of shape, size or motion. In some real-world applications, objects have little inter-occlusions and split/merge difficulties, so the spatio-temporal consistency in object tracking are well preserved in the image sequences/videos. Starting from the imperfect initialization, this paper proposes a new easy-to-hard learning method to incrementally improve the object detection in image sequences/videos by an unsupervised spatio-temporal analysis which involves more complex examples that are hard for object detection for next-iteration training. The proposed method does not require manual annotations, but uses weak supervisions and spatio-temporal consistency in tracking to simulate the supervisions in the CNN training. Experimental results on three different tasks show significant improvements over the initialized detections by the weak object detectors.

A method of quantifying fiber chirality, the twist of continuous fibers through a volume, is defined and applied to both phantom data and real data. Specifically, a field quantity termed the fiber chirality based on the anti-symmetric part of the gradient of the fiber orientation is introduced. For this method of estimation, the input is the set of fiber positions gathered from the stack of images which represent the sample volumes. The phantom sample is generated and several different real continuous fiber reinforced matrix composites are experimentally characterized. Each phantom dataset contains a bundle of fibers that rotate about a center axis with a user-defined angle at each step in the z-direction. The chirality of the fibers is calculated based on the pre-characterized positions using a machine learning algorithm. To validate the method of quantification, our chirality estimation method results in a colormap with an angle of rotation that becomes increasingly more similar to the user-defined angle with decreasing inter-slice distance. Fiber positions from real data are then input into the estimation method and the results are compared.

Successful deployment of the highly heterogeneous, laminated, polymer matrix composites (PMCs) in high-performance structural applications is currently hindered by the lack of reliable experimental protocols for evaluation of the local mechanical responses at the salient meso-length/structure scales present in these material systems. Our main interest in this paper lies in establishing and demonstrating protocols for high-throughput evaluation of the local mechanical responses in PMCs at a length scale larger than the fiber diameter but smaller than the individual laminate (i.e., ply) thickness. This goal was accomplished in this work through a successful extension of the spherical indentation stress–strain protocols demonstrated recently for metallic samples. Specifically, plies with fibers at 0°, 30°, 60°, and 90° to the indentation direction were tested, and the means and standard deviations of their indentation moduli and the indentation yield strengths were measured and reported in this paper. The measured values of the indentation moduli were validated with finite element (FE) simulations performed using estimated values of the effective single laminate stiffness parameters. Furthermore, the measured variation in the indentation moduli was shown to correlate extremely well with the corresponding FE predictions that accounted for the measured variation in the local fiber volume fractions in the primary indentation deformed zones in the sample. These comparisons provided strong support for the validity of the extended spherical indentation protocols developed in this work for PMC samples.

With ceramic matrix composites (CMCs) being targeted for use as hot section components in gas turbine engines , it is important to understand the thermal-mechanical behavior of these materials within extreme application environments. This work focused on characterization of the in-situ response of a melt-infiltrated (MI) composite (i.e., a laminated Hi-Nicalon™, BN-interphase, melt-infiltrated (MI) SiC matrix composite) subjected to combined thermal-mechanical loads representative of intended operational environments. Laboratory scale test specimens were loaded in uniaxial tension at room temperature and at elevated temperature under both isothermal and non-isothermal conditions . During laser heating experiments, a number of in-situ techniques enabled monitoring of damage formation, strain evolution, and temperature up to failure at 1150 °C. Results indicate that applying mechanical loads axially in the presence of thermal gradients may induce interply delamination cracks not observed after CMC test specimens are loaded axially under isothermal conditions .

A computational framework is developed to model the transverse failure of fiber-reinforced polymer-matrix composites, with an emphasis on capturing fiber debonding with a cohesive failure model along the fiber/matrix interfaces. We introduce a nonlinear material sensitivity formulation to quantify how variations in the interfacial cohesive zone properties affect the transverse failure response. The analytic sensitivity formulation is implemented in an interface-enriched generalized finite element method (IGFEM) framework that allows for the simulation of transverse failure in a composite layer consisting of hundreds of closely packed fibers discretized with finite element meshes that do not need to conform to the composite microstructure.

The concept of Integrated Computational Materials Engineering (ICME) is aimed at accelerating the development and insertion of new materials in engineering applications. ICME approach relies on the development and use in design of relationships between processing and structure, and its corresponding property/performance. This poses a constraint on computational speed, which is difficult to achieve without losing the physics. The concept of building data-driven, material agnostic models to describe process-structure-property linkages has the potential to satisfy this need. In recent works, this has been introduced on a wide variety of materials at multiple length scales of interest. We review these developments. More specifically, a review of the fusion of material science and data science is presented. The framework addresses curation of materials’ knowledge from the available datasets in computationally efficient manner to extract and use the processing-structure-property relationships.

This manuscript presents a computationally efficient method based on a geometric model to simulate the transverse cracking of a 90∘ cross-ply in a composite laminate. The model expands on existing homogenized solutions of transverse cracking by accounting for the random microstructure of the transverse ply extracted from optical micrographs of a hybrid [0∕90∕0]T glass/carbon/epoxy composite laminate. The chapter summarizes the three steps of the method, which allows to model the creation of multiple transverse cracks in realistic transverse plies composed of tens of thousands of fibers. The model is calibrated against experimental measurements of the critical values of the applied transverse strain corresponding to the appearance of transverse cracks and then used in a statistical analysis of the impact of the interface strength distribution on the evolution of the transverse cracking process.

The statistically equivalent RVE or P-SERVE have been introduced in Swaminathan et al. (J Compos Mater 40(7):583–604, 2006) and Ghosh (Micromechanical analysis and multi-scale modeling using the voronoi cell finite element method. CRC Press/Taylor & Francis, Boca Raton, 2011) as the smallest microstructural volume element in non-uniform microstructures that has effective material properties equivalent to those of the entire microstructure. An important consideration is the application of appropriate boundary conditions for optimal SERVE domains. The exterior statistics-based boundary conditions or ESBCs have been developed in Ghosh and Kubair (J Mech Phys Solids 96:1–24, 2016), Kubair and Ghosh (Int J Solids Struct 112:106–121, 2017), Kubair et al. (J Comput Mech 52(21):2919–2928, 2018), accounting for the statistics of fiber distributions and interactions in the domain exterior to the SERVE. The ESBC-based SERVEs have been validated for effective convergence in evaluating homogenized stiffnesses and optimal domains for micromechanical analysis. Validation is also conducted with an experimentally studied carbon-fiber epoxy-matrix polymer matrix composite (PMC). The performance of the SERVE with ESBCs is compared with other boundary conditions, as well as with the statistical volume elements (SVE). The tests clearly show the significant advantages of the ESBCs in terms of accuracy of the homogenized stiffness and efficiency.

Raman spectroscopy was utilized to characterize the chemical composition and residual stresses formed in melt infiltrated SiC/SiC CMCs during processing. Stresses in SiC fibers, in SiC chemical vapor (CVI) infiltrated matrix, in SiC melt infiltrated matrix, and in free silicon were measured for two different plates of CMCs. Stresses in the free silicon averaged around 2 GPa in compression, while stresses in the matrix SiC were 1.45 GPa in tension. The SiC CVI phase had stresses ranging between 0.9 GPa and 1.2 GPa in tension and the SiC fibers experienced stresses of .05–0.7 GPa in tension. These results were validated with the proposed model of the system. While the mismatch in the coefficients of thermal expansion between the constituents contributes to the overall residual stress state, the silicon expansion upon solidification was found to be the major contributor to residual stresses within the composite.

The property-based statistically equivalent RVE or P-SERVE has been introduced in the literature as the smallest microstructural volume element in non-uniform microstructures that has effective material properties equivalent to those of the entire microstructure. An important consideration is the application of appropriate boundary conditions for optimal property-based statistically equivalent representative volume element domains. The exterior statistics-based boundary conditions have been developed, accounting for the statistics of fiber distributions and interactions in the domain exterior to the property-based statistically equivalent representative volume element. This paper is intended to validate the efficacy of the exterior statistics-based boundary condition-based property-based statistically equivalent representative volume elements for evaluating homogenized stiffnesses of a unidirectional polymer matrix composite with a polydispersed microstructure characterized by nonuniform dispersion of carbon fibers of varying sizes in an epoxy matrix. Experimental tests and microstructural characterization of the polymer matrix composite are conducted for calibration and validation of the model. Statistically equivalent microstructural volume elements are constructed from experimental micrographs for direct numerical simulations. The performance of the property-based statistically equivalent representative volume element with exterior statistics-based boundary conditions is compared with other boundary conditions, as well as with the statistical volume elements. The tests clearly show the significant advantages of the exterior statistics-based boundary conditions in terms of accuracy of the homogenized stiffness and efficiency.

The degradation of SiC-based ceramic matrix composites (CMCs) in conditions typical of gas turbine engine operation proceeds via the stress-rupture of fiber bundles. The degradation is accelerated when oxygen and water invade the composite through matrix microcracks and react with fiber coatings and the fibers themselves. We review micromechanical models of the main rate-determining phenomena involved, including the the diffusion of gases and reaction products through matrix microcracks, oxidation of SiC (in both matrix and fibers) leading to the loss of stiffness and strength in exposed fibers, the formation of oxide scale on SiC fiber and along matrix crack surfaces that cause the partial closure of microcracks, and the concomitant and synergistic loss of BN fiber coatings. The micromechanical models could be formulated as time-dependent coupled differential equations in time, which must be solved dynamically, e.g., as an iterated user-defined material element, within a finite element simulation. A paradigm is thus established for incorporating the time-dependent evolution of local material properties according to the local environmental and stress conditions that exist within a material, in a simulation of the damage evolution of a composite component. We exemplify the calibration of typical micromechanical degradation models using thermodynamic data for the oxidation and/or volatilization of BN and SiC by oxygen and water, mechanical test data for the rate of stress-rupture of SiC fibers, and kinetic data for the processes involved in gas permeation through microcracks. We discuss approaches for validating computational simulations that include the micromechanical models of environmental degradation. A special challenge is achieving validated predictions of trends with temperature, which are expected to vary in a complex manner during use.. This article is protected by copyright. All rights reserved.

High-temperature mechanical properties and tension-tension fatigue behavior of three advanced SiC/SiC composites are discussed. The effects of steam on high-temperature fatigue performance of the ceramic-matrix composites are evaluated. The three composites consist of a SiC matrix reinforced with laminated, woven SiC (Hi-Nicalon™) fibers. Composite 1 was processed by chemical vapor infiltration (CVI) of SiC into the Hi-Nicalon™ fiber preforms coated with boron nitride (BN) fiber coating. Composite 2 had an oxidation inhibited matrix consisting of alternating layers of silicon carbide and boron carbide and was also processed by CVI. Fiber preforms had pyrolytic carbon fiber coating with boron carbon overlay applied. Composite 3 had a melt-infiltrated (MI) matrix consolidated by combining CVI-SiC with SiC particulate slurry and molten silicon infiltration. Fiber preforms had a CVI BN fiber coating applied. Tensile stress-strain behavior of the three composites was investigated and the tensile properties measured at 1200°C. Tension-tension fatigue behavior was studied for fatigue stresses ranging from 80 to 160 MPa in air and from 60 to 140 MPa in steam. Fatigue run-out was defined as 2 × 105 cycles. Presence of steam significantly degraded the fatigue performance of the CVI SiC/SiC composite 1 and of the MI SiC/SiC composite 3, but had little influence on the fatigue performance of the SiC/SiC composite 2 with the oxidation inhibited matrix. The retained tensile properties of all specimens that achieved fatigue run-out were characterized. Composite microstructure, as well as damage and failure mechanisms were investigated.

An environmental barrier-coated Sylramic-iBN/pyrolytic carbon/chemical vapor infiltration silicon carbide ceramic matrix composite was characterized in a harsh combustion environment under two conditions; (i) with a coating surface temperature of 1480℃, and (ii) with a combustion flame distance of 120 mm (constant flux). Three coating systems were evaluated; Coating “A” (ytterbium silicate using hafnium–silicate bond), Coating “B” (ytterbium silicate using alumina bond), and Coating “C” (ytterbium–silica compounds using alumina bond). All three coatings survived with essentially no mass loss up to 25 h when tested at 1480℃. At a flame distance of 120 mm coatings “A” and “B” experienced extensive degradation and strength loss while coating “C” had practically no degradation in retained strength. Coating “C” outperformed the other two coatings exposed to the harsh combustion environment. It performed best in terms of material loss, thermal barrier capability, and residual tensile strength.

Carbon fiber reinforced plastics (CFRPs) have become the material of choice for many low weight, high strength applications. One problem associated with these materials is a scatter of mechanical properties. Efforts have been made to correlate these results with random fiber packing using 2-D representative volume elements (RVEs). While these efforts have shown that fiber distribution is important, scans have shown fiber position is a function of 3-D morphology. For example, fiber meandering and entanglement can only be seen when variation along the fiber direction is studied. While the optimal scale for 2-D RVEs is well researched, little has been done to find the effective length in which the 3-D behavior of real samples can be observed. As such, the construction and analysis of 3-D RVEs with a requisite length in the fiber direction is necessary. The scope of this work is to study the fibers within a CFRP sample in order to recreate the microstructure. In this work, long micro-CT scans from an aerospace-grade are used to obtain the fiber positions. In addition, scans from an automotive grade specimen with a high fiber count manufactured using VARTM were analyzed due to the high potential for entanglement. Descriptive metrics were then generated to fiber behavior within both samples. The variance of these metrics was captured using the change in Pearson’s Correlation Coefficient (R) with respect to position along the fiber length. One

A novel methodology is introduced for quantifying the severity of damage created during testing in composite components. The method uses digital image correlation combined with image processing techniques to monitor the rate at which the strain field changes during mechanical tests. The methodology is demonstrated using two distinct experimental datasets, a ceramic matrix composite specimen loaded in tension at high temperature and nine polymer matrix composite specimens containing fibre-waviness defects loaded in bending. The changes in the strain field owing to damage creation are shown to be a more effective indicator that the specimen has reached its proportional limit than using load-extension diagrams. The technique also introduces a new approach to using experimental data for creating maps indicating the spatio-temporal distribution of damage in a component. These maps indicate where damage occurs in a component, and provide information about its morphology and its time of occurrence. This presentation format is both easier and faster to interpret than the raw data which, for some tests, can consist of tens of thousands of images. This methodology has the potential to reduce the time taken to interpret large material test datasets while increasing the amount of knowledge that can be extracted from each test.

Continuous ceramic fiber reinforced ceramic matrix composites (CMCs) offer an innovative damage tolerant structural material for temperature regimes inaccessible to high temperature metals. However, when subjected to application environments, these materials exhibit damage and degrade over time depending on the severity of those conditions. The ultimate strength of CMCs is dictated by the properties of the load bearing fibers, but matrix cracking weakens the composite and can expose the load bearing fibers to harsh environments. The objective of the work performed was to develop a methodology for linking in situ detection of localized damage to final failure in continuous fiber reinforced CMCs. Initiation and growth of matrix cracking are measured and located linearly along the gage length via acoustic emission (AE) detection. High amplitude events at relatively low static loads can be associated with initiation of large matrix cracks. When there is a localization of high amplitude events in a given area, a measurable effect on the strain field can be observed. Full field surface strain measurements were obtained using digital image correlation (DIC). An estimation of the matrix cracking stress as well as localized areas of initiation were measured in real time.

Experimental data on grain growth and oxidation kinetics of SiC-based fibers, as well as the accompanying strength degradation, in argon, air, and moist air are interpreted using a mechanistic model. The grain growth from thermal history is modeled using conventional models, and its influence on strength is modeled assuming that the flaw size scales with grain size. The model for fiber oxidation uses available relevant thermodynamic and kinetic data for reactions, vapor pressures, oxygen permeation, and boundary layer effects to capture scale thickness data reported by several prior works, in static or flowing air, moist air, and steam. The effect of the oxide scale on strength was modeled assuming that the flaw size scaled with scale thickness. The resulting model is compared with experimental data and is shown to capture most of the data in the literature on degradation of HiNicalon™ and HiNicalon™ type S fibers.

Toughness in continuous ceramic fiber reinforced ceramic matrix composites (CMCs) with dense matrices depends on the properties of the fiber coating or interphase. Multiple criteria have been proposed to describe the mechanism of crack propagation/deflection at the filament scale in brittle matrix continuous fiber reinforced composites; however, most of these criteria fail to account for the presence of an interphase of finite thickness and/or employ unrealistic boundary conditions. Recent simulations employing the extended finite element method (XFEM) have shown that variations in interphase thickness and strength relative to the fibers and/or matrix can have a significant influence on the crack propagation/deflection mechanism. It is shown that primary crack deflection most often occurs when conditions favor secondary cracking in the interphase in front of an approaching matrix crack. Although this mechanism is similar to that argued by Cook and Gordon (1964), the simulations here indicate that the conditions for secondary crack initiation and deflection of the primary crack can be much different than that which was originally presented in their analytical model. Variations in the properties of the interphase are simulated to produce large deviations in the local crack growth behavior as a matrix crack grows into interphase. Results are discussed relative to what has been observed experimentally.

This study employs Peridynamic (PD) theory to predict crack deflection in axisymmetric ceramic matrix composites (CMC). It specifically employs the weak form of PD equations of motion which enables the direct imposition of natural and essential boundary conditions. The results indicate that the critical stress ratio between the coating and matrix have a strong influence on crack deflection in a CMC. Also, it was observed that the smaller this ratio is, the earlier the crack deflection in the fiber coating occurs.

A reduced-order, data-driven, probabilistic predictive model to quantify damage initiation in continuous SiC ceramic fiber SiC ceramic matrix composites (CMCs) at pertinent lengths scales using machine learning tools is proposed and explored. A novel framework is developed to characterize the influence of key stochastic microstructure attributes on matrix crack initiation. The approach is illustrated for the case of transverse crack initiation in the matrix surrounding fibers oriented perpendicular to the loading direction. A variety of stochastic microstructure attributes were considered including fiber spacing, fiber diameter, and coating thickness. Statistics of a commercial CMC microstructure were digitally represented and used to instantiate microstructures. In addition, discrete digital instantiations generated over a range of the distributed microstructural attributes were considered. The statistics of the distributed microstructure attributes were quantified using n-point statistics and reduced using principal component analysis. The elastic responses of the instantiated microstructures were characterized using finite element analysis (FEA). Results from the FEA were used as the ground truth to calibrate and validate a data-driven machine learning (ML) model. The quantified stochastic microstructure attributes were correlated with the statistics of the simulated damage response. The predictive capabilities of the model for a new microstructure class were demonstrated.

A computationally efficient method based on digital image correlation (DIC) has been introduced in this paper for fast and accurate characterisation of fibre orientation fields from a series of optical mosaics of a continuous fibre-reinforced composite (CFRC) specimen, which were obtained using a well-established automated serial sectioning technique. The newly developed method has the capability of determining fibre paths from the sequence of mosaics in scenarios where the serial section spacing between the adjacent mosaics is atypically large, resulting in shifts in the fibre locations on the order of 90 pixels or twelve fibre diameters. The performance of the proposed method was quantitatively compared with a state-of-the-art fibre-tracking algorithm based on a Kalman filter, by applying them to the mosaic sequences from two CFRC material specimens. It was demonstrated from the fibre-tracking results that the proposed DIC-based method outperformed the Kalman filter algorithm in terms of both reliability and speed.

Advance ceramic reinforced ceramic matrix composites (CMCs) materials exhibit hierarchical internal structure with rich details at multiple length scales of interest. These details of the materials’ internal structure are referred to as the microstructure which dictates the overall performance of the material. In this work, we show that a rigorous statistical quantification of the material internal structure (i.e. microstructure) is possible for CMCs. In particular, a 2-point spatial correlation function is shown to capture the spatial heterogeneity for the two different class of microstructures generated using Uniform and Gaussian distribution. Additionally, it is demonstrated how these stochastic representations of the distributed filaments can be imported into a finite element framework to simulated transverse cracking strength. The damage in these simulations in modeled using a novel Regularized eXtended Finite Element Method (Rx-FEM) that predicts damage initiation and propagation without a priori defining where that damage occurs.

When considering structural materials used in aerospace applications and time-dependent behavior, primary concern are material/microstructural changes and damage initiation and growth as a result of complex loading (creep and/or fatigue) scenarios and/or environmental attack. The degradation and damage in the material can result in a decrease in load-carrying capability. It is the decrease of capability as a function of time/usage/exposure that must be understood and predicted to optimize the design and life management strategies of aerospace components that comprise aircraft structures and turbine engines. Historically predictive models in these domains were empirically based; relying on accelerated test methods, extensive amounts of test data, and mathematical fits to that data. More recent research in time-dependent material properties has shifted the focus to understanding the underlying mechanisms of material degradation and developing predictive capabilities incorporating that understanding. Specifically, to realize more accurate and robust performance prognosis for structural materials, a shift from empirical descriptions of time-dependent material behavior to more mechanistic-based models that capture the physics of failure is needed.

This paper provides some background and historical review of how microstructure-sensitive finite element simulations can play a role in understanding the effects of stress amplitude, R -ratio, and microstructure on fatigue crack formation and early growth at notches, including pores and non-metallic inclusions for Ti alloys and Ni-base superalloys. The simulations employ fatigue indicator parameters (FIPs) computed over finite volumes that relate to processes of fatigue crack formation and early growth at the scale of individual grains. It is argued that both coarse scale (uncracked, mesoscale) and fine scale FIPs (computed in the vicinity of cracks in single grains or crystals) serve as a driving force for crystallographic fatigue crack growth, and correlate directly with the cyclic crack tip displacement (CTD). Furthermore, variability in high cycle fatigue (HCF) and very high cycle fatigue (VHCF) responses is computationally assessed using multiple statistical volume elements and the distribution of FIPs of extreme value character. The concepts of marked correlation functions and weighted probability density functions are reviewed as a means to quantify the role of multiple microstructure attributes that couple to enhance the extreme value FIPs in the HCF regime. An algorithm for estimation of the cumulative probability distribution of cycles for crack formation and growth from notches in HCF and VHCF is also described.

Automatic tracking of large-scale crowded targets are of particular importance in many applications, such as crowded people/vehicle tracking in video surveillance, fiber tracking in materials science, and cell tracking in biomedical imaging. This problem becomes very challenging when the targets show similar appearance and the interslice/ inter-frame continuity is low due to sparse sampling, camera motion and target occlusion. The main challenge comes from the step of association which aims at matching the predictions and the observations of the multiple targets. In this paper we propose a new groupwise method to explore the target group information and employ the within-group correlations for association and tracking. In particular, the within-group association is modeled by a nonrigid 2D Thin-Plate transform and a sequence of group shrinking, group growing and group merging operations are then developed to refine the composition of each group. We apply the proposed method to track large-scale fibers from microscopy material images and compare its performance against several other multi-target tracking methods. We also apply the proposed method to track crowded people from videos with poor inter-frame continuity.

The objective of the work performed here was to develop a methodology for linking in-situ detection of localized matrix cracking to the final failure location in continuous fiber reinforced CMCs. First, the initiation and growth of matrix cracking are measured and triangulated via acoustic emission (AE) detection. High amplitude events at relatively low static loads can be associated with initiation of large matrix cracks. When there is a localization of high amplitude events, a measurable effect on the strain field can be observed. Full field surface strain measurements were obtained using digital image correlation (DIC). An analysis using the combination of the AE and DIC data was able to predict the final failure location.

The distributions of the extreme-value driving force(s) for surface vs. subsurface fatigue crack formation (nucleation and early growth) in high-cycle fatigue are evaluated for a microstructure variant of duplex Ti–6Al–4V. In polycrystalline metals, previous work has explored estimation of the driving force(s) for fatigue crack formation at the scale of the grains by computing non-local fatigue indicator parameters (FIPs) based on the cyclic plastic strain averaged over domains on the length scale of the grains. Instantiated statistical volume elements (SVEs), which sample the distributed microstructure attributes of interest for a given material system, can be simulated via the finite element method with embedded polycrystalline plasticity models to estimate the distributed plasticity and resulting FIPs. This strategy of simulating multiple SVEs is in contrast to the simulation of a single representative volume element which is typically untenably large for extreme-value distributions of microstructure attributes or response variables. In this work, multiple SVEs are instantiated with both traction-free (i.e. surface) boundary conditions and fully periodic (i.e. subsurface) boundary conditions. In addition to estimating the extreme-value distributions of the FIPs, newly introduced extreme-value marked correlation functions are applied to characterize the coupled crystallographic microstructure attributes (e.g. grain size, grain orientation, grain misorientation) that most influence the extreme-value distributions of the FIPs. It is shown that there is overlap in the distributions of the driving forces for surface vs. subsurface crack formation in the low to moderate range of failure probability based on FIPs; however, at higher failure probability levels, the driving forces are highest for surface crack formation. The overlap in the distributions of the driving forces for fatigue crack formation in the low to moderate probability range may assist in describing the competing surface vs. subsurface failure modes that are observed experimentally.

Strategies are described for microstructure-sensitive computational methods for estimating variability of high cycle fatigue (HCF) crack formation and early growth in metallic polycrystals to support design of fatigue resistant alloys. We outline a philosophy of employing computational simulation to establish relations between remote loading conditions and microstructure-scale slip behavior in terms of Fatigue Indicator Parameters (FIPs) as a function of stress amplitude, stress state and microstructure, featuring calibration of mean experimental responses for known microstructures. Effects of process history (carburization and shot peening) and resulting residual stresses are considered in the case of subsurface crack formation at primary inclusions in martensitic gear steel. The need to characterize extreme value correlations of microstructure attributes coupled to the local driving force (i.e., features) for HCF crack formation is outlined, along with a strategy involving a set of FIPs relevant to different mechanisms of crack formation. Surface to subsurface transitions are considered in terms of competing mechanisms in the transition from HCF to very high cycle fatigue (VHCF) regimes.

Ceramic matrix composites (CMC) with continuous fiber reinforcements have the potential to enable the next generation of high speed hypersonic vehicles and/or significant improvements in gas turbine engine performance due to their exhibited toughness when subjected to high mechanical loads at extreme temperatures (2200F+). Reinforced fiber composites (RFC) provide increased fracture toughness, crack growth resistance, and strength, though little is known about how stochastic variation and imperfections in the materi al effect material properties. In this work, tools are developed for quantifying anomalies within the microstructure at several scales. The detection and characterization of anomalous microstructure is a critical step in linking producti on techniques to properties, as well as in accurate material simulation and property prediction for the integrated computation materials engineering (ICME) of RFC based components. It is desired to find statistical outliers for any number of material character istics such as fibers, fiber coatings, and pores. Here, fiber orientation, or ‘velocity’, and ‘velocity’ gradient are developed and examined for anomalous behavior. Categorizing anomalous behavior in the CMC is approached by multivariate Gaussian mixture modeling. A Gaussian mixture is employed to estimate the probability density function (PDF) of the features in question, and anomalies are classified by their likelihood of belonging to the statistical normal behavior for that feature. Keywords: Fiber reinforced composites, Ceramic matrix composites, Anomaly detection, Gaussian mixture modeling, Color visualization, Texture anomalies, Velocity gradient tensor

The Markov Random Field (MRF) has been used extensively in Image Processing as a means of smoothing interfaces between differing regions in an image. The MRF applies a total boundary length ‘energy’ penalty that is subsequently minimized by an inversion algorithm. The minimization of energy implies a force associated with boundaries, the sum of which must equal zero at every point at equilibrium. This requirement leads to long range interactions, resulting from the short-range interactions of the MRF, which biases segmentation results. This work uses a simple Bayesian MRF regularized segmentation method to show that classical results from Surface Science are reproduced when segmenting regions of low contrast. This has implications, both in the Materials Science and Image Processing fields.

Oxidation of SiC-containing turbine and other structural components during high temperature operation generates residual stresses that can potentially lead to life-limiting environmentally assisted crack growth and failure. Currently, residual stress generation and/or relaxation during thermal oxidation of SiC is poorly understood. Here a combined experimental-modeling framework is introduced that provides new insight into the mechanisms of oxide formation, and, the magnitudes of the residual stresses induced when SiC is subjected to an oxidizing environment. SiC plates are oxidized at several temperatures over varying time periods, where after laser interferometry is employed to characterize the deformation induced in the plate due to the buildup of residual stress from the formation of the oxide. In this manner, values of the mismatch strain generated upon oxidation are extracted. In parallel, an integrated modeling approach for SiC oxidation is introduced that incorporates the essential physics associated with the diffusion of oxidant and effluent species, reaction at/near the interface, stress generation and viscoelastic/plastic relaxation mechanisms that accompany oxidation, respectively, and their interplay. Results are discussed regarding the generation and evolution of stress as the passive oxide layer grows at different temperatures in the range 500-1000ºC range. Insights obtained on the role of viscoelastic/plastic accommodation in oxidation and residual stress relaxation, and, the mechanisms underlying stress generation, are discussed. The impact of oxidation on environmentally assisted failure of SiC structures is also addressed.

One of research thrusts to push the limits of advanced fiber reinforced composites is to determine the link between manufacturing, resulting microstructure, and final structural properties. By bridging the gap between these topics, not only can we better understand how and why composites structurally work as they do, but we can potentially tailor the manufacturing processes for a desired resultant set of properties. To better illuminate the effects of manufacturing on microstructure and microstructure on properties, computational models are often employed. Using these models, we can gain insight on relationships that may otherwise remain unexplored. This research thrust is often labelled as ICME, integrated computational materials engineering.