{"id":{"repo_id":"mtu","oai_identifier":"oai:digitalcommons.mtu.edu:etdr-3183"},"canonical_url":"https://search.ndltd.org/etd/mtu/oai:digitalcommons.mtu.edu:etdr-3183","repository":{"repo_id":"mtu","name":"Michigan Technological University","base_url":"https://digitalcommons.mtu.edu/do/oai/"},"display":{"title":"NONDESTRUCTIVE EVALUATION OF ADDITIVELY MANUFACTURED PARTS USING RESONANT INSPECTION AND FREQUENCY DOMAIN-BASED CORRELATION CRITERIA","abstract":"Additive manufacturing (AM) enables the production of highly customized and geometrically complex components; however, these parts remain susceptible to a wide range of defect types and severities. The combination of complex geometries and distributed defects presents a significant challenge for post-process nondestructive evaluation (NDE). Conventional inspection techniques, such as computed tomography and ultrasonic testing, are often limited by geometry, material, accessibility, and cost, while in situ monitoring is not yet sufficiently mature to replace post-process inspection. This dissertation establishes the Frequency Domain Assurance Criterion (FDAC) as a quantitative, vibration-based framework for defect detection in AM components. FDAC captures spatial–spectral correlations between operating deflection shapes, providing a physically interpretable representation of structural health. Scalar damage indicators derived from FDAC matrices reliably distinguish defective components and reflect defect severity. A sensitivity analysis identified spatial resolution as the dominant factor influencing detection performance, while frequency resolution had minimal impact. These findings enabled optimization of test and analysis parameters, reducing test duration by 92.9% and memory requirements by approximately 98% without compromising detection accuracy. To address data limitations, a synthetic data generation framework was developed using physics-constrained mode shape synthesis and feature-level alignment. The results demonstrate that synthetic datasets must be validated at multiple levels, including physics, representation, and decision level, to ensure reliable synthetic-to-experimental transfer. Machine learning (ML) and deep learning (DL) models trained using FDAC-based features achieved strong detection performance, outperforming statistics-based methods and achieving 100% classification accuracy. A hybrid framework combining a convolutional autoencoder with a support vector machine further improved performance by leveraging structure-aware feature extraction from FDAC matrices. Overall, this work demonstrates that FDAC provides a robust, efficient, and cost-effective defect detection framework that overcomes the limitations of conventional inspection methods and is well suited for high throughput AM applications.","abstract_html":"Additive manufacturing (AM) enables the production of highly customized and geometrically complex components; however, these parts remain susceptible to a wide range of defect types and severities. The combination of complex geometries and distributed defects presents a significant challenge for post-process nondestructive evaluation (NDE). Conventional inspection techniques, such as computed tomography and ultrasonic testing, are often limited by geometry, material, accessibility, and cost, while in situ monitoring is not yet sufficiently mature to replace post-process inspection. This dissertation establishes the Frequency Domain Assurance Criterion (FDAC) as a quantitative, vibration-based framework for defect detection in AM components. FDAC captures spatial–spectral correlations between operating deflection shapes, providing a physically interpretable representation of structural health. Scalar damage indicators derived from FDAC matrices reliably distinguish defective components and reflect defect severity. A sensitivity analysis identified spatial resolution as the dominant factor influencing detection performance, while frequency resolution had minimal impact. These findings enabled optimization of test and analysis parameters, reducing test duration by 92.9% and memory requirements by approximately 98% without compromising detection accuracy. To address data limitations, a synthetic data generation framework was developed using physics-constrained mode shape synthesis and feature-level alignment. The results demonstrate that synthetic datasets must be validated at multiple levels, including physics, representation, and decision level, to ensure reliable synthetic-to-experimental transfer. Machine learning (ML) and deep learning (DL) models trained using FDAC-based features achieved strong detection performance, outperforming statistics-based methods and achieving 100% classification accuracy. A hybrid framework combining a convolutional autoencoder with a support vector machine further improved performance by leveraging structure-aware feature extraction from FDAC matrices. Overall, this work demonstrates that FDAC provides a robust, efficient, and cost-effective defect detection framework that overcomes the limitations of conventional inspection methods and is well suited for high throughput AM applications.","abstract_has_math":false,"creators":["Deonarain, Gita"],"institution":null,"degree_name":"Doctor of Philosophy in Mechanical Engineering-Engineering Mechanics (PhD)","degree_level":"Open Access Dissertation","degree_discipline":"Department of Mechanical and Aerospace Engineering","degree_department":null,"school":null,"contributors":["Jason R. Blough"],"advisors":[],"committee_chairs":[],"committee_members":["Gordon G. Parker","David M. Labyak","Chad M. Walber","Vijaya V.N. Sriram Malladi"],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-11-19T08:19:26Z","subjects":["additive manufacturing","nondestructive evaluation","frequency domain assurance criterion","structural health monitoring","vibration-based damage detection","Acoustics, Dynamics, and Controls"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["https://digitalcommons.mtu.edu/etdr/2085"],"render_values":[{"text":"https://digitalcommons.mtu.edu/etdr/2085","href":"https://digitalcommons.mtu.edu/etdr/2085","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/10.37099/mtu.dc.etdr/2085","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/10.37099/mtu.dc.etdr/2085","record_label":"Repository record","record_kind":"doi","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://digitalcommons.mtu.edu/do/oai/?verb=GetRecord&metadataPrefix=oai_etdms&identifier=oai%3Adigitalcommons.mtu.edu%3Aetdr-3183","prefix":"oai_etdms"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Jason R. Blough"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Gordon G. Parker","David M. Labyak","Chad M. Walber","Vijaya V.N. Sriram Malladi"]},{"key":"dc:creator","label":"Author","values":["Deonarain, Gita"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.created","label":"Dc Date Created","values":["2026-01-01T08:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Department of Mechanical and Aerospace Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Open Access Dissertation"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy in Mechanical Engineering-Engineering Mechanics (PhD)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["additive manufacturing","nondestructive evaluation","frequency domain assurance criterion","structural health monitoring","vibration-based damage detection","Acoustics, Dynamics, and Controls"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["10.37099/mtu.dc.etdr/2085","https://digitalcommons.mtu.edu/etdr/2085"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Additive manufacturing (AM) enables the production of highly customized and geometrically complex components; however, these parts remain susceptible to a wide range of defect types and severities. The combination of complex geometries and distributed defects presents a significant challenge for post-process nondestructive evaluation (NDE). Conventional inspection techniques, such as computed tomography and ultrasonic testing, are often limited by geometry, material, accessibility, and cost, while in situ monitoring is not yet sufficiently mature to replace post-process inspection. This dissertation establishes the Frequency Domain Assurance Criterion (FDAC) as a quantitative, vibration-based framework for defect detection in AM components. FDAC captures spatial–spectral correlations between operating deflection shapes, providing a physically interpretable representation of structural health. Scalar damage indicators derived from FDAC matrices reliably distinguish defective components and reflect defect severity. A sensitivity analysis identified spatial resolution as the dominant factor influencing detection performance, while frequency resolution had minimal impact. These findings enabled optimization of test and analysis parameters, reducing test duration by 92.9% and memory requirements by approximately 98% without compromising detection accuracy. To address data limitations, a synthetic data generation framework was developed using physics-constrained mode shape synthesis and feature-level alignment. The results demonstrate that synthetic datasets must be validated at multiple levels, including physics, representation, and decision level, to ensure reliable synthetic-to-experimental transfer. Machine learning (ML) and deep learning (DL) models trained using FDAC-based features achieved strong detection performance, outperforming statistics-based methods and achieving 100% classification accuracy. A hybrid framework combining a convolutional autoencoder with a support vector machine further improved performance by leveraging structure-aware feature extraction from FDAC matrices. Overall, this work demonstrates that FDAC provides a robust, efficient, and cost-effective defect detection framework that overcomes the limitations of conventional inspection methods and is well suited for high throughput AM applications."]},{"key":"dc:title","label":"Title","values":["NONDESTRUCTIVE EVALUATION OF ADDITIVELY MANUFACTURED PARTS USING RESONANT INSPECTION AND FREQUENCY DOMAIN-BASED CORRELATION CRITERIA"]}]}],"canonical_facts":{"dc:contributor":["Jason R. Blough"],"dc:contributor.committeemember":["Gordon G. Parker","David M. Labyak","Chad M. Walber","Vijaya V.N. Sriram Malladi"],"dc:creator":["Deonarain, Gita"],"dc:date.created":["2026-01-01T08:00:00Z"],"dc:description.abstract":["Additive manufacturing (AM) enables the production of highly customized and geometrically complex components; however, these parts remain susceptible to a wide range of defect types and severities. The combination of complex geometries and distributed defects presents a significant challenge for post-process nondestructive evaluation (NDE). Conventional inspection techniques, such as computed tomography and ultrasonic testing, are often limited by geometry, material, accessibility, and cost, while in situ monitoring is not yet sufficiently mature to replace post-process inspection. This dissertation establishes the Frequency Domain Assurance Criterion (FDAC) as a quantitative, vibration-based framework for defect detection in AM components. FDAC captures spatial–spectral correlations between operating deflection shapes, providing a physically interpretable representation of structural health. Scalar damage indicators derived from FDAC matrices reliably distinguish defective components and reflect defect severity. A sensitivity analysis identified spatial resolution as the dominant factor influencing detection performance, while frequency resolution had minimal impact. These findings enabled optimization of test and analysis parameters, reducing test duration by 92.9% and memory requirements by approximately 98% without compromising detection accuracy. To address data limitations, a synthetic data generation framework was developed using physics-constrained mode shape synthesis and feature-level alignment. The results demonstrate that synthetic datasets must be validated at multiple levels, including physics, representation, and decision level, to ensure reliable synthetic-to-experimental transfer. Machine learning (ML) and deep learning (DL) models trained using FDAC-based features achieved strong detection performance, outperforming statistics-based methods and achieving 100% classification accuracy. A hybrid framework combining a convolutional autoencoder with a support vector machine further improved performance by leveraging structure-aware feature extraction from FDAC matrices. Overall, this work demonstrates that FDAC provides a robust, efficient, and cost-effective defect detection framework that overcomes the limitations of conventional inspection methods and is well suited for high throughput AM applications."],"dc:identifier":["10.37099/mtu.dc.etdr/2085","https://digitalcommons.mtu.edu/etdr/2085"],"dc:subject":["additive manufacturing","nondestructive evaluation","frequency domain assurance criterion","structural health monitoring","vibration-based damage detection","Acoustics, Dynamics, and Controls"],"dc:title":["NONDESTRUCTIVE EVALUATION OF ADDITIVELY MANUFACTURED PARTS USING RESONANT INSPECTION AND FREQUENCY DOMAIN-BASED CORRELATION CRITERIA"],"thesis:degree_discipline":["Department of Mechanical and Aerospace Engineering"],"thesis:degree_level":["Open Access Dissertation"],"thesis:degree_name":["Doctor of Philosophy in Mechanical Engineering-Engineering Mechanics (PhD)"]},"updated_at":"2026-11-19T08:19:26Z"}