Metallic materials are widely adopted in the construction industry. While traditional manufacturing techniques such as hot-rolling, cold-forming and extrusion often leads to prismatic structural components, adopting metal additive manufacturing (AM) technologies, such as Directed Energy Deposition (DED), will allow greater flexibility in the geometry of construction components. However, geometry inconsistency in DED is one of the main problems that hinders its application in various fields, including construction industry. Lack of timely geometry assessment during DED process results in dimensional errors and defects of final products. Therefore, this research proposes to develop and integrate advanced process monitoring techniques into a DED system with a goal of online geometry qualit...[ Read more ]

Metallic materials are widely adopted in the construction industry. While traditional manufacturing techniques such as hot-rolling, cold-forming and extrusion often leads to prismatic structural components, adopting metal additive manufacturing (AM) technologies, such as Directed Energy Deposition (DED), will allow greater flexibility in the geometry of construction components. However, geometry inconsistency in DED is one of the main problems that hinders its application in various fields, including construction industry. Lack of timely geometry assessment during DED process results in dimensional errors and defects of final products. Therefore, this research proposes to develop and integrate advanced process monitoring techniques into a DED system with a goal of online geometry quality assessment in support of quality assurance and process certification.

To achieve geometry quality control, geometry quality assessment including geometry monitoring and geometry estimation are needed. According to existing literature review about geometry quality assessment during DED process, the research gaps are identified: (1) there is a lack of online real-time geometry monitoring technique to inspect each deposited track profile, which requires the integration of 3D geometry monitoring system with DED system; (2) real-time geometry estimation that can establish explicit relationship between process parameters and geometry attributes is not well studied, which impedes the development of geometry quality control; (3) there is a lack of real-time geometry monitoring and geometry estimation for depositions with sharp features, such as corners and intersections. Therefore, this research aims to tackle these research gaps to better adopt DED in construction industry. The research objectives are achieved with three studies. The first and second studies aim to tackle research gap (1) and research gap (3), which is to achieve online real-time geometry monitoring with consideration of sharp feature. The third study serves for research gap (2) and research gap (3), which is to achieve geometry estimation considering sharp feature.

For geometry monitoring, in the first study, an online geometry monitoring methodology for continuous monitoring during the DED process was developed using a laser line scanner. The proposed methodology achieved track-wise inspection during multi-layer single-track deposition process, including (1) real-time scanning of each deposited track’s profile, (2) online extraction of the track’s geometry, and (3) online plotting and comparison of the as-designed and as-built models. The effectiveness of the developed methodology is examined by comparing the profiles of the multi-layer single-track deposition estimated during the DED process with the reference profiles obtained via laser line scanning and microscopy after the completion of the DED process.

Furthermore, in the second study, an improved online real-time geometry monitoring methodology is developed to achieve real-time inspection during multi-layer deposition with sharp feature. A geometry monitoring system was developed using two laser line scanners, one for real-time inspection of the melt pool area or newly solidified area of each track, and another for layer-wise inspection of each deposited layer. For real-time inspection, an image processing method with an encoder-decoder based profile completion network was developed to obtain accurate track profiles on images. Then, a point cloud generation method was proposed to convert the obtained track profiles to 3D point cloud that represents the deposited object by calibrations and fusing the position information from the printer. Experiments have been successfully conducted to validate the proposed methodology by depositing multi-layer X-shape objects.

After geometry monitoring, geometry estimation to identify the relationship between the process parameters and geometry attributes is vital toward geometry control. In the third study, a real-time layer height estimation technique is developed using a laser line scanner, vision camera, and domain adaptive neural networks (DaNN) for multi-layer deposition with sharp feature. The proposed technique can achieve real-time layer height estimation for both straight-line deposition and corner deposition, and an emphasis was placed on layer height estimation at sharp corners.

This research provides approaches to achieve online geometry quality assessment during the DED process using laser line scanner, including real-time geometry monitoring and real-time geometry estimation. In addition, this research put emphasis on geometry with sharp features which are often problematic and not well studied in previous literatures. The proposed online geometry assessment techniques could effectively facilitate DED to achieve zero-defect manufacturing and assist its adoption in various fields including the construction industry.