Investigation of melt pool monitoring-based deposition quality prediction in directed energy deposition

Abstract

Directed energy deposition (DED) is one of the representative metal additive manufacturing. In the DED process, a high intensity energy beam such as a laser or electron beam is irradiated onto a substrate to form a melt pool and then supply filler materials such as metal powder or wire to the melting region for bead formation. The DED process, which uses a laser as an energy source and metal powder as a filler material, is currently the most widely used because of its relatively simple equipment configuration and precise fabrication. The laser powder DED process parameters, such as laser power, travel speed, powder feed rate, and inert gas supply amount, are usually optimized considering the light absorption rate, melting point, and heat transfer rate of the substrate and feed material. However, the optimal parameter values often vary during the process dependent on the process conditions such as scan path, dwell time, the amount of heat accumulation, and heat transfer condition determined by the substrate. If the proper parameter values deviate from the optimal range because of the process condition change, the melt pool become unstable, leading to process failure. Accordingly, a technology for maintaining a stable melt pool state through real-time quality prediction and feedback control using process monitoring is essential. The state-of-the-art DED process monitoring system uses a method that measures the temperature and shape of the melt pool via a pyrometer and vision camera. In most process control studies, process control only focuses on keeping the melt pool state stable by simply controlling one process parameter, the laser power, without judging process quality. However, it is difficult to maintain the stable process by the simple control strategy because the process state is all affected by the various variables mentioned above. To compensate the effect of heat accumulation and changed heat transfer condition during a long process time, it is necessary to comprehensively consider and control multiple variables simultaneously. Therefore, a deep understanding of the relationship between process conditions and parameters and what the process monitoring data means must be preceded to establish the advanced control strategy. In this study, a melt pool monitoring system using a 2-color pyrometer and a CMOS-vision camera optically installed coaxially with the laser beam is developed, and the melt pool monitoring data in the DED process is acquired. In addition, the correlation between process variables-melt pool state-deposited shape is identified based on the data analysis, leading to obtaining a regression model that can predict the deposited quality. Through this, a stable process maintenance strategy to suppress defects is considered.