| description abstract | This paper presents a semi-automated grinding system for the postprocessing of metalcastings. Grinding is an important procedure in the “cleaning room” of a foundry, where the removal of gate contacts, parting line flash, surface defects, and weld-repaired areas is performed, and almost always manually. While the grinding of repetitive locations on medium to high production castings can be automated using robotics or otherwise, it is not as practical for larger castings (e.g., > 200 kg) that are typically produced in smaller production volumes. Furthermore, automation is even more challenging in that the locations of the required grinding are not a constant depending on the unique conditions and anomalies of each pouring of a component. The proposed approach is intended for a simple x−y−z positioner (gantry) device with a feedback controlled grinding head that enables automated path planning. The process begins with touch probing of the surfaces that contain the anomaly requiring grinding, and then the system automatically handles the path planning and force control to remove the anomaly. A layer-based algorithm for path planning employs a search-and-destroy technique where the surrounding geometry is interpolated across the grind-requiring surface patch. In this manner, each unique condition of the casting surface after initial torch or saw cutting can be handled cost effectively without the need for human shaping and the egregious ergonomic problems associated. Implementation of the proposed grinding control is prototyped at a lab scale to demonstrate the feasibility and versatility of this strategy. The average error for the prototype was on the order of 0.007 in (0.2 mm) with a flatness of the ground surface within 0.035 in (0.9 mm), which is within the cleaning room grinding requirements, as per ISO and ASTM dimensional and surface tolerance requirements. A significant contribution of the work is the layer-based algorithm that allows an effective automation of the process planning for grinding, avoiding robot programming or numerical control code generation altogether. This is a key to addressing the largely unknown and unpredictable conditions of, for example, the riser contact surface removal area on a metalcasting. | |
