Alexander Caputo
Advisor: Rick Neu
will defend a doctoral thesis entitled,
Towards Born Qualification of AM components: High Temperature Fatigue Testing and Microstructural Characterization of AM IN718
On
MRDC Room 3510
and
Virtually via Microsoft Teams
Meeting ID: 295 166 181 165
Passcode: AP5o8L
Committee
Prof. Richard W. Neu – George W. Woodruff School of Mechanical Engineering (advisor)
Prof. Aaron Stebner – School of Materials Science and Engineering
Prof. Chris Saldaña– George W. Woodruff School of Mechanical Engineering
Prof. Josh Kacher – School of Materials Science and Engineering
Dr. Xuan Zhang – Principal Materials Scientist, Argonne National Lab
Abstract
Additive manufacturing (AM) of metallic components offers significant advantages over traditional manufacturing. AM provides a rapid and highly customizable means to create components of complicated geometries in a fraction of the lead time and cost of conventional means. Additionally, AM has been proven valuable for producing replacement parts and repairing components originally produced by now-defunct manufacturers. While AM can solve many manufacturing problems that have been prohibitively expensive by conventional means, it also introduces an additional uncertainty in material properties. The AM process is complicated and introduces many uncontrolled cofactors that induce variability in observed mechanical properties.
Traditional metallurgical processes like casting and forging have been studied and optimized for thousands of years, whereas AM is still an emerging technology. The inherent variability of the AM process, and corresponding variation in properties, is holding it back from broader industrial implementation. Post build testing and characterization of witness specimens has been employed to certify AM components for use in critical applications, but this process is expensive and time consuming. For AM to become a reliable means of manufacturing metallic components, a rapid and effective certification method to enable the immediate determination of material properties as soon as the components are finished being built must be established. Achieving this born qualification of AM components requires an intimate knowledge of the process-structure-property relationships that dictate the critical performance requirements of the intended application.
Turbine components used in high temperature applications in aerospace and energy production stand to uniquely benefit from the lightweighting and enhanced cooling channel geometries unlocked by AM. IN718 components, specifically, are expensive and difficult to manufacture and machine using conventional means, and AM is uniquely suited to produce IN718 turbine components more effectively. The AM process, however, introduces defects and large variability in the fatigue properties which are critical to turbine applications. To unlock the benefits of AM for turbine applications, a robust procedure for ensuring the born qualification of fatigue properties must be established.
This work advances the goal of achieving reliable AM fatigue performance by investigating the intricate process-structure-property relationships determining the high temperature fatigue performance of AM IN718. In this work, the porous and crystallographic microstructure of AM IN718 specimens was characterized by X-ray computed tomography and scanning electron microscopy and correlated to the high temperature fatigue properties through data analysis and machine learning. The AM IN718 specimens investigated were produced using various combinations of process parameters, or process pedigrees, throughout AM process window to capture the full range of possible AM defects and microstructures.
The results of this investigation revealed that in a majority of the AM IN718 process window, the fatigue properties are determined by the competitive influence of microstructure and porosity. In a select few cases, pores of high criticality, being some combination of large in volume, having sharp geometries, or being close to other pores or the surface, proved to severely decrement the observed fatigue properties. Analysis of the compiled characterization and fatigue data through the training of multiple ML algorithms reinforced the competing influence of the porous and crystallographic microstructural features in the accurate determination of fatigue properties.