NSF Award
2015693
This award advances the understanding of defect formation in wire arc additive manufacturing, leading to improved processing and greater process control. This research addresses surface waviness and nonuniform wall thickness challenges in wire arc additive manufacturing, making it acceptable for many U.S. industries, including aerospace, defense, and automotive, and thereby benefiting the nation?s economy and well-being. Wire arc additive manufacturing uses a welding arc as the energy source in making three-dimensional objects with high throughput. This process can be easily integrated with existing computer-numerical-control routers, or robot arms. Thus, the knowledge and methodology developed through this project can be directly transferred to small- and medium-sized enterprises interested in making or repairing metallic parts. Additionally, this project develops the professional skills of K-12, undergraduate, and graduate students, including women and underrepresented minorities, by training them through a unique set of integrated education and multidisciplinary research opportunities in the areas of wire arc additive manufacturing and data analytics. Accordingly, students are familiarized with emerging technology-intensive manufacturing and data science disciplines, thereby preparing a future workforce equipped with these new skills and knowledge. <br/><br/>The two overarching research goals of this project are to (1) gain fundamental knowledge about surface waviness and effective wall thickness formation mechanisms in wire arc additive manufacturing and (2) investigate the design rules for the monitoring and control of the manufacturing process. An integrated experimental-characterization and theoretical-modeling framework are considered to achieve these goals. The surface quality of wire arc additively manufactured structures is controlled by fine-tuning and balancing the surface tension, arc force, droplet impact, gravity, buoyancy, and friction for different manufacturing conditions. The project framework consists of four components: (1) measurement and analysis of real-time process signatures (e.g., voltage and current) and visualization data of the arc, droplet, and weld pool features; (2) data measurements and analysis of defect (e.g., balling effect and voids) generation and propagation as a function of process parameters, wall thickness and inclination, and multi-bead/multilayer deposition; (3) characterization of the surface tension-based computational fluid dynamics model and its verification and validation through experiments; and (4) establishment of design rules using data-driven, low-fidelity surrogate models. Machine learning algorithms (e.g., convolutional neural network, AnoGAN, transfer learning, and reinforcement learning) are developed in detail for defect detections and classifications as well as process control. This integrated framework provides unique, transformative, and efficient opportunities to synergistically understand the arc, droplet, and weld pool characteristics and defect formation in wire arc additive manufacturing.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
