Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Mechanical Engineering

Committee Chair/Advisor

Laine Mears

Committee Member

Adam Hoover

Committee Member

Cameron Turner

Committee Member

John Wagner


Manufacturing competitiveness is related to making a quality product while incurring the lowest costs. Unexpected downtime caused by equipment failure negatively impacts manufacturing competitiveness due to the ensuing defects and delays caused by the downtime. Manufacturers have adopted condition monitoring (CM) techniques to reduce unexpected downtime to augment maintenance strategies. The CM adoption has transitioned maintenance from Breakdown Maintenance (BM) to Condition-Based Maintenance (CbM) to anticipate impending failures and provide maintenance actions before equipment failure. CbM is the umbrella term for maintenance strategies that use condition monitoring techniques such as Preventive Maintenance (PM) and Predictive Maintenance (PdM). Preventive Maintenance involves providing periodic checks based on either time or sensory input. Predictive Maintenance utilizes continuous or periodic sensory inputs to determine the machine health state to predict the equipment failure. The overall goal of the work is to improve bearing diagnostic and prognostic predictions for equipment health by utilizing surrogate systems to generate failure data that represents production equipment failure, thereby providing training data for condition monitoring solutions without waiting for real world failure data. This research seeks to address the challenges of obtaining failure data for CM systems by incorporating a third system into monitoring strategies to create a Digital Triplet (DTr) for condition monitoring to increase the amount of possible data for condition monitoring. Bearings are a critical component in rotational manufacturing systems with wide application to other industries outside of manufacturing, such as energy and defense. The reinvented DTr system considers three components: the physical, surrogate, and digital systems. The physical system represents the real-world application in production that cannot fail. The surrogate system represents a physical component in a test system in an offline environment where data is generated to fill in gaps from data unavailable in the real-world system. The digital system is the CM system, which

provides maintenance recommendations based on the ingested data from the real world and surrogate systems. In pursuing the research goal, a comprehensive bearing dataset detailing these four failure modes over different collection operating parameters was created. Subsequently, the collections occurred under different operating conditions, such as speed-varying, load-varying, and steadystate. Different frequency and time measures were used to analyze and identify differentiating criteria between the different failure classes over the differing operating conditions. These empirical observations were recreated using simulations to filter out potential outliers. The outputs of the physical model were combined with knowledge from the empirical observations to create ”spectral deltas” to augment existing bearing data and create new failure data that resemble similar frequency criteria to the original data. The primary verification occurred on a laboratory-bearing test stand. A conjecture is provided on how to scale to a larger system by analyzing a larger system from a local manufacturer. From the subsequent analysis of machine learning diagnosis and prognosis models, the original and augmented bearing data can complement each other during model training. The subsequent data substitution verifies that bearing data collected under different operating conditions and sizes can be substituted between different systems. Ostensibly, the full formulation of the digital triplet system is that bearing data generated at a smaller size can be scaled to train predictive failure models for larger bearing sizes. Future work should consider implementing this method for other systems outside of bearings, such as gears, non-rotational equipment, such as pumps, or even larger complex systems, such as computer numerically controlled machine tools or car engines. In addition, the method and process should not be restricted to only mechanical systems and could be applied to electrical systems, such as batteries. Furthermore, an investigation should consider further data-driven approximations to specific bearing characteristics related to the stiffness and damping parameters needed in modeling. A final consideration is for further investigation into the scalability quantities within the data and how to track these changes through different system levels.

Author ORCID Identifier




To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.