To develop and disseminate metrology methods, tools, data, and standards applied to in-situ monitoring of AM processes, such that manufacturers and their customers can accurately and precisely determine the quality of the fabrication process and resulting parts.
What is the Technical Idea?
The modus operandi for qualification of metal AM parts requires extensive machine testing and tweaking for optimal performance and repeatability, in addition to numerous and expensive post-process destructive and non-destructive part performance tests. Real-time process monitoring is an idealized solution in which by applying in-situ sensors that monitor the fabrication process, one might determine the performance of the AM machine and the predict the quality of the produced part as it is fabricated. Over the last several years, there has been an enormous research effort into AM process monitoring that spans hundreds of different institutions, countries, companies, and monitoring methods . Several monitoring technologies have exhibited widespread interest and are recently implemented on commercial AM systems. These include melt pool monitoring (MPM) systems that observe melt pool temperature, geometry, or dynamics, or layer-wise monitoring (LM) systems, which capture intermittent observations of the solid or powder layers. Although these methods are on the verge of becoming standard tools in industrial AM settings, there is still much to be improved such as sensor system design, best practices for image analysis and signal processing, instrument calibration and characterization, and general understanding of these system’s capabilities and limitations. Ultimately, these factors need to be addressed to enable in-situ monitoring to accelerate the qualification and certification processes. Numerous forward-looking (low technology readiness level, or TRL) non-destructive technologies are continuously being invented or developed, such as in-situ acoustic monitoring methods, methods for measuring surface structure, measurement of multispectral optical emission, and many more. Despite demonstrated or promising results from these various monitoring tools, significant research is still required to determine robust correlations between the sensor signatures and part qualities. In addition, these sensor systems differ in their design, application, and results interpretation despite a shared goal of rapid part qualification. Another key barrier in AM monitoring research is the heterogeneity of instruments, and lack of standard protocols for their application, calibration, and characterization.
What is the Research Plan?
A principle objective of the Real-time Monitoring for AM (RTM) project will be to explore the sensor signature – part quality relationships by generating numerous intercomparable and well-controlled process monitoring reference datasets utilizing industrially relevant monitoring systems, with potential combination with new or experimental sensor systems. Compilation of large number of comparable datasets are the core requirement for traditional statistical process control (SPC), empirical model-based control (MBC) methodologies, or analytics approaches such as artificial intelligence (AI) or machine learning. In conjunction with the dataset generation, the monitoring instrumentation will be thoroughly characterized. That is, the qualities of the instrumentation itself will be measured via calibration, or measurement of resolution, dynamic range, frequency response, etc. These characterizations may then be disseminated as standards, so that proper description of process monitoring tools can be better established and better communicated. Furthermore, the RTM project will explore the feasibility of new sensor technologies by making use of open platform AM systems, such as the AMMT, or utilizing existing and established measurement systems to correlate or characterize the newer techniques.
Generation of carefully designed and well-controlled process monitoring datasets will primarily be generated on the Additive Manufacturing Metrology Testbed (AMMT), and make use of its flexibility, available calibration instruments, and well-characterized system performance. The RTM project will collaborate with NIST and external experts in various part quality metrology methods such as X-ray computed tomography (XCT) for defect detection, surface structure and morphology characterization, and microstructure characterization. These datasets will be shared publicly, with focused collaborations with experts in signal and data processing, statistics, and data analytics. These collaborations, which will be informed by domain expertise of the RTM project team, will enable broad investigation and eventual discovery of robust relationships between AM monitoring data and ex-situ part quality data. By creating many open, public, well-controlled, and intercomparable AM process monitoring datasets, this project aims to utilize broad collaborative efforts towards the analysis and use of process monitoring data.
In addition to working with custom instrumentation at NIST, the RTM project will utilize its broad collaborations to help understand, characterize, and for commercial AM process monitoring systems. These include laser powder bed fusion (LPBF) and direct energy deposition (DED) systems. In order to fulfill the process monitoring standards gaps (D22 and PC16) outlined in the AMSC roadmap, this project will accelerate more standard approaches to AM monitoring by 1) disseminating best practices through reports and standards development 2) developing new calibration and characterization tools and artifacts with intended applicability and deployment to commercial AM systems and 3) investigating the relationship between the true physical phenomena (e.g., temperature and emittance) in AM processes, and the resultant monitoring system observations to determine sources or error or misinterpretation. Furthermore, the RTM project will utilize the flexible AM research platforms and instrumentation expertise at NIST to investigate new, high-risk, cutting-edge AM monitoring methods.