Point-Based Shape Monitoring of Plate Bending for Large-Scale Storage Tanks

Storage tanks are constructed using thousands of large curved steel plates, which are manufactured from flat plates. In conventional manufacturing of curved plates, operators fit wooden templates to specific positions on steel plates and measure differences between the current shape and the template. However, it is costly to create many wooden templates for a variety of plates. In addition, it is time-consuming and requires skills to precisely place wooden templates on specific positions to measure differences. In this paper, we discuss methods to automatically calculate differences of shapes during bending processes without wooden templates. We capture dense points on steel plates using a terrestrial laser scanner, and analyze shapes of curved plates using point-clouds. In our method, the system extracts only the points on curved plates, and tracks the amount of deformation on reference lines defined on the plates. Corresponding positions between intermediate curved plates and the original flat shape are calculated using mesh flattening techniques. In our experiments, our method could calculate the amount of differences of steel plates in reasonable performance and precision.