Repository logo

Browsing by Author "Simske, Steven, committee member"

Now showing 1 - 8 of 8
  • Thumbnail Image
    ItemOpen Access
    A combined classification and queuing system optimization approach for enhanced battery system maintainability
    (Colorado State University. Libraries, 2022) Pirani, Badruddin, author; Cale, James, advisor; Simske, Steven, committee member; Miller, Erika, committee member; Keller, Josh, committee member
    Battery systems are used as critical power sources in a wide variety of advanced platforms (e.g., ships, submersibles, aircraft). These platforms undergo unique and extreme mission profiles that necessitate high reliability and maintainability. Battery system failures and non-optimal maintenance strategies have a significant impact on total fleet lifecycle costs and operational capability. Previous research has applied various approaches to improve battery system reliability and maintainability. Machine learning methodologies have applied data-driven and physics-based approaches to model battery decay and predict battery state-of-health, estimation of battery state-of-charge, and prediction of future performance. Queuing theory has been used to optimize battery charging resources ensure service and minimize cost. However, these approaches do not focus on pre-acceptance reliability improvements or platform operational requirements. This research introduces a two-faceted approach for enhancing the overall maintainability of platforms with battery systems as critical components. The first facet is the implementation of an advanced inspection and classification methodology for automating the acceptance/rejection decision for batteries prior to entering service. The purpose of this "pre-screening" step is to increase the reliability of batteries in service prior to deployment. The second facet of the proposed approach is the optimization of several critical maintenance plan design attributes for battery systems. Together, the approach seeks to simultaneously enhance both aspects of maintainability (inherent reliability and cost-effectiveness) for battery systems, with the goal of decreasing total lifecycle cost and increasing operational availability.
  • Thumbnail Image
    ItemOpen Access
    Bio-inspired design for engineering applications: empirical and finite element studies of biomechanically adapted porous bone architectures
    (Colorado State University. Libraries, 2020) Aguirre, Trevor Gabriel, author; Donahue, Seth W., advisor; Ma, Kaka, committee member; Heyliger, Paul, committee member; Simske, Steven, committee member
    Trabecular bone is a porous, lightweight material structure found in the bones of mammals, birds, and reptiles. Trabecular bone continually remodels itself to maintain lightweight, mechanical competence, and to repair accumulated damage. The remodeling process can adjust trabecular bone architecture to meet the changing mechanical demands of a bone due to changes in physical activity such as running, walking, etc. It has previously been suggested that bone adapted to extreme mechanical environments, with unique trabecular architectures, could have implications for various bioinspired engineering applications. The present study investigated porous bone architecture for two examples of extreme mechanical loading. Dinosaurs were exceptionally large animals whose body mass placed massive gravitational loads on their skeleton. Previous studies investigated dinosaurian bone strength and biomechanics, but the relationships between dinosaurian trabecular bone architecture and mechanical behavior has not been studied. In this study, trabecular bone samples from the distal femur and proximal tibia of dinosaurs ranging in body mass from 23-8,000 kg were investigated. The trabecular architecture was quantified from micro-computed tomography scans and allometric scaling relationships were used to determine how the trabecular bone architectural indices changed with body mass. Trabecular bone mechanical behavior was investigated by finite element modeling. It was found that dinosaurian trabecular bone volume fraction is positively correlated with body mass like what is observed for extant mammalian species, while trabecular spacing, number, and connectivity density in dinosaurs is negatively correlated with body mass, exhibiting opposite behavior from extant mammals. Furthermore, it was found that trabecular bone apparent modulus is positively correlated with body mass in dinosaurian species, while no correlation was observed for mammalian species. Additionally, trabecular bone tensile and compressive principal strains were not correlated with body mass in mammalian or dinosaurian species. Trabecular bone apparent modulus was positively correlated with trabecular spacing in mammals and positively correlated with connectivity density in dinosaurs, but these differential architectural effects on trabecular bone apparent modulus limit average trabecular bone tissue strains to below 3,000 microstrain for estimated high levels of physiological loading in both mammals and dinosaurs. Rocky Mountain bighorn sheep rams (Ovis canadensis canadensis) routinely conduct intraspecific combat where high energy cranial impacts are experienced. Previous studies have estimated cranial impact forces up to 3,400 N and yet the rams observationally experience no long-term damage. Prior finite element studies of bighorn sheep ramming have shown that the horn reduces brain cavity translational accelerations and the bony horncore stores 3x more strain energy than the horn during impact. These previous findings have yet to be applied to applications where impact force reduction is needed, such as helmets and athletic footwear. In this study, the velar architecture was mimicked and tested to determine suitability as novel material architecture for running shoe midsoles. It was found that velar bone mimics reduce impact force (p < 0.001) and higher energy storage during impact (p < 0.001) and compression (p < 0.001) as compared to traditional midsole architectures. Furthermore, a quadratic relationship (p < 0.001) was discovered between impact force and stiffness in the velar bone mimics. These findings have implications for the design of novel material architectures with optimal stiffness for minimizing impact force.
  • Thumbnail Image
    ItemOpen Access
    Integrating geometric deep learning with a set-based design approach for the exploration of graph-based engineering systems
    (Colorado State University. Libraries, 2024) Sirico, Anthony, Jr., author; Herber, Daniel R., advisor; Chen, Haonan, committee member; Simske, Steven, committee member; Conrad, Steven, committee member
    Many complex engineering systems can be represented in a topological form, such as graphs. This dissertation introduces a framework of Graph-Set-Based Design (GSBD) that integrates graph-based techniques with Geometric Deep Learning (GDL) within a Set-Based Design (SBD) approach to address graph-centric design problems. We also introduce Iterative Classification (IC), a method for narrowing down large datasets to a subset of more promising and feasible solutions. When we combine the two, we have IC-GSBD, a methodological framework where the primary goal is to effectively and efficiently seek the best-performing solutions with lower computational costs. IC-GSBD is a method that employs an iterative approach to efficiently narrow down a graph-based dataset containing diverse design solutions to identify the most useful options. This approach is particularly valuable as the dataset would be computationally expensive to process using other conventional methods. The implementation involves analyzing a small subset of the dataset to train a machine-learning model. This model is then utilized to predict the remaining dataset iteratively, progressively refining the top solutions with each iteration. In this work, we present two case studies demonstrating this method. In the first case study utilizing IC-GSBD, the goal is the analysis of analog electrical circuits, aiming to match a specific frequency response within a particular range. Previous studies generated 43,249 unique undirected graphs representing valid potential circuits through enumeration techniques. However, determining the sizing and performance of these circuits proved computationally expensive. By using a fraction of the circuit graphs and their performance as input data for a classification-focused GDL model, we can predict the performance of the remaining graphs with favorable accuracy. The results show that incorporating additional graph-based features enhances model performance, achieving a classification accuracy of 80% using only 10% of the graphs and further subdividing the graphs into targeted groups with medians significantly closer to the best and containing 88.2 of the top 100 best-performing graphs on average using 25% of the graphs.
  • Thumbnail Image
    ItemOpen Access
    Modeling and simulation to investigate the electrification potential of medium- and heavy-duty vehicle fleets
    (Colorado State University. Libraries, 2023) Trinko, David A., author; Bradley, Thomas H., advisor; Quinn, Jason C., committee member; Simske, Steven, committee member; Hurrell, James, committee member
    This project involves developing and integrating new modeling tools to simulate the dynamics of electric medium- and heavy-duty fleet vehicle adoption. A technical and economic modeling tool, combining a data-driven hardware cost model with a cost-optimal charging strategy microsimulation, enables tailored analysis of the costs and benefits of electrifying individual fleets. Next, a novel text synthesis process, applied to a curated corpus of literature, quantifies trade-offs between technical, economic, and other factors in the fleet vehicle procurement decision. The outcomes of these tasks combine with knowledge from recent literature on fleet decision processes to specify the vehicle procurement model used by fleets in an agent-based model of the medium- and heavy-duty electric vehicle market. This model embodies an especially disaggregated approach to adoption modeling, internalizing factors and dynamics that conventional adoption models externalize. In particular, explicitly modeling the formation and diffusion of opinions among agents enables experiments that conventional models cannot support. Demonstrations show, for example, that increasing the extent of interactions between populations with different proclivities to electric vehicles has an asymmetrical outcome. High-proclivity electric vehicle adoption is generally unaffected as interactions increase, but low-proclivity adoption is accelerated. By representing individual fleets' requirements and costs at a high level of detail, incorporating an adoption decision model informed by a wide body of empirical research, and broadening the array of variables and dynamics available for experimentation, this integrated model offers a new way to understand the urgent challenge of eliminating emissions from the most emissions-intensive transportation sectors.
  • Thumbnail Image
    ItemOpen Access
    Proactive extraction of IoT device capabilities for security applications
    (Colorado State University. Libraries, 2020) Dolan, Andrew, author; Ray, Indrakshi, advisor; Majumdar, Suryadipta, advisor; Simske, Steven, committee member; Ghosh, Sudipto, committee member
    Internet of Things (IoT) device adoption is on the rise. Such devices are mostly self-operated and require minimum user interventions. This is achieved by abstracting away their design complexities and functionalities from users. However, this abstraction significantly limits a user's insights on evaluating the true capabilities (i.e., what actions a device can perform) of a device and hence, its potential security and privacy threats. Most existing works evaluate the security of those devices by analyzing the environment data (e.g., network traffic, sensor data, etc.). However, such approaches entail collecting data from encrypted traffic, relying on the quality of the collected data for their accuracy, and facing difficulties in preserving both utility and privacy of the data. We overcome the above-mentioned challenges and propose a proactive approach to extract IoT device capabilities from their informational specifications to verify their potential threats, even before a device is installed. More specifically, we first introduce a model for device capabilities in the context of IoT. Second, we devise a technique to parse the vendor-provided materials of IoT devices and enumerate device capabilities from them. Finally, we apply the obtained capability model and extraction technique in a proactive access control model to demonstrate the applicability of our proposed solution. We evaluate our capability extraction approach in terms of its efficiency and enumeration accuracy on devices from three different vendors.
  • Thumbnail Image
    ItemOpen Access
    Redundant complexity in deep learning: an efficacy analysis of NeXtVLAD in NLP
    (Colorado State University. Libraries, 2022) Mahdipour Saravani, Sina, author; Ray, Indrakshi, advisor; Banerjee, Ritwik, advisor; Simske, Steven, committee member
    While deep learning is prevalent and successful, partly due to its extensive expressive power with less human intervention, it may inherently promote a naive and negatively simplistic employment, giving rise to problems in sustainability, reproducibility, and design. Larger, more compute-intensive models entail costs in these areas. In this thesis, we probe the effect of a neural component -- specifically, an architecture called NeXtVLAD -- on predictive accuracy for two downstream natural language processing tasks -- context-dependent sarcasm detection and deepfake text detection, and find it ineffective and redundant. We specifically investigate the extent to which this novel architecture contributes to the results, and find that it does not provide statistically significant benefits. This is only one of the several directions in efficiency-aware research in deep learning, but is especially important due to introducing an aspect of interpretability that targets design and efficiency, ergo, promotes studying architectures and topologies in deep learning to both ablate the redundant components for enhancement in sustainability, and to earn further insights into the information flow in deep neural architectures, and into the role of each and every component. We hope our insights highlighting the lack of benefits from introducing a resource-intensive component will aid future research to distill the effective elements from long and complex pipelines, thereby providing a boost to the wider research community.
  • Thumbnail Image
    ItemOpen Access
    Structural health monitoring in adhesively bonded composite joints
    (Colorado State University. Libraries, 2024) Caldwell, Steven, author; Radford, Donald W., advisor; Simske, Steven, committee member; Cale, James, committee member; Adams, Henry, committee member
    Composite bonded aircraft structure is a prevalent portion of today's aircraft structural composition. Adequate bond integrity is a critical aspect of the fabrication and operational service life of aircraft structure. Many of these structural bonds are critical for flight safety. Thus, a major concern is related to the assurance of quality in the structural bond. Over the last decade, non-destructive bond evaluation techniques have improved but still cannot detect a structurally weak bond that exhibits full adherend/adhesive contact. Currently, expensive, and time-consuming structural proof testing is required to verify bond integrity. The objective of this work is to investigate the feasibility of bondline integrity monitoring via piezoelectric sensors embedded in the composite joint. Initially, a complex composite joint, the Pi preform, was analytically evaluated for health monitoring viability, with the results showing promising capability. Subsequently, due to experimental complexities, a simple, state-of-the-art composite single lap shear joint is selected for experimentation and analysis to measure and quantify the effect of incorporating a sensor within the bondline to evaluate and expand on the ability of the embedded sensor to monitor and assess the joint integrity. Simple flatwise tension joints are also studied to investigate an orthogonal loading direction through the sensor. The experimental results indicate that the embedded piezoelectric sensors can measure a change in the joint before the integrity degrades and fails on subsequent loadings, resulting in a novel approach for prognostic performance evaluation without detrimentally affecting the performance of the structural joint.
  • Thumbnail Image
    ItemOpen Access
    Techniques in reactive to proactive obsolescence management for C5ISR systems
    (Colorado State University. Libraries, 2023) Chellin, Matthew D., author; Miller, Erika, advisor; Daily, Jeremy, committee member; Herber, Daniel, committee member; Simske, Steven, committee member; Prawel, David, committee member
    Obsolescence is a significant challenge for the Command, Control, Communications, Computers, Cyber, Intelligence, Surveillance and Reconnaissance (C5ISR) community. Obsolescence can negatively affect a C5ISR system's cost, schedule, performance, and readiness. This research examines the challenge of obsolescence for C5ISR systems by focusing on the U.S. Army at Aberdeen Proving Ground, Maryland and their industry partners. The objective of this research is to synthesize insights from the experiences of government and industry practitioners that mitigate diminishing manufacturing sources and material shortages (DMSMS) challenges into mitigation techniques. The obsolescence mitigation areas described in this research include proactive and reactive obsolescence mitigation, obsolescence mitigation methods, and the importance of DMSMS contracting language. This research also offers approaches grounded in practitioner experiences to mitigate obsolescence through a proactive obsolescence management model, risk mitigation framework, metrics, modeling & simulation, and systems thinking methods. The combination of the models, methods, and approaches discussed from this research have the potential to achieve greater system readiness, more availability, better maintainability, and lower costs for C5ISR systems.