Automatic reconstruction of nominal freeform surfaces in aero-components for remanufacturing

• Main Author: Dung, Van Than
• Other Authors: Tegoeh Tjahjowidodo
• Format: Theses and Dissertations
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Manufacturing::Flexible manufacturing systems; DRNTU::Engineering::Manufacturing::CAD/CAM systems
• Online Access: https://hdl.handle.net/10356/88841; http://hdl.handle.net/10220/45979
• Institution: Nanyang Technological University

Remanufacturing is a new branch of manufacturing industry in which used mechanical components are repaired and reconditioned to be used as new ones. In remanufacturing processes, used components were cleaned, filled in at the worn locations and reprofiled before being used again. Most of cases, the design geometries of the workpieces are not used as a reference because of many reasons, e.g. they might not be provided by suppliers and/or distortions of the workpieces during service life create discrepancies to the design geometries. Therefore, remanufacturing processes rely heavily on skilled workers. In this project, we propose a framework to automate reprofiling processes in remanufacturing industries for freeform surface workpieces in the absence of design geometries. Many components, e.g. in aerospace industries, have complex geometries that are usually generated in freeform surfaces, such as turbine valves, blades, blisks, dies, molds, etc. This brings a significant challenge in automating the reprofiling processes. In the proposed framework, a workpiece (component) will be mounted on a reference table and initially digitized to acquire its geometries in the form of cloud data. Subsequently, the reference table and the Workpiece Coordinate System (WCS) are identified to transform the measured data to the WCS. A depth map image is then generated by resampling the measured data. Based on the B-spline fitting technique, an algorithm is proposed to automatically estimate the nominal profile surfaces of the components from the measured data. Some defects (e.g. from weld beads) are then directly obtained by comparing the estimated nominal surface with the digitized geometries. A tool path program then adaptively generates G-code for each identified defect and it is sent to a milling CNC machine to remove the defect. The performance of the proposed method is finally evaluated using numerical simulation and real measurement data. The results confirm that the method is capable of exactly estimating the nominal profile surface and identify weld beads on freeform surface components with minimal human intervention.