Incremental sheet forming (ISF) is a manufacturing process, which has a high flexibility and is suitable for producing small lot sizes. However, without solid support beneath the blank, for example through a die or a counter tool, complex geometries with a good geometrical accuracy cannot be formed.
The first funding period of the research project on incremental sheet forming with active medium (IFAM) met this challenge by providing a soft support through a liquid or gaseous medium on the blank’s underside. The combination of tool path and pressure level enabled both the manufacturing of convex and concave elements as well as combined convex-concave parts. One aspect of the first funding period was analysing the convex bulging behaviour by including
analytical, experimental and numerical methods. To achieve a target geometry accurately, the intricate correlations between the process parameters and the part shape had to be considered and led to an interdisciplinary collaboration with the field of machine learning. A neural network was trained to predict the required pressure in an online closed-loop control of the geometrical properties. Compared to the manufacturing process with constant pressure, the geometrical
accuracy was significantly improved by using a dynamically adjusted pressure level.
Beside the great potential of this controlling concept, it has the disadvantage of being restricted to one specific material and one predetermined geometry. The motivation behind the second funding period is to solve this problem and extend the applicability of the controlling concept to any arbitrary material and geometry. The extended controlling concept is based on the research hypothesis that a strong relationship between target shape, forming energy and tangential forming forces exists and can be expressed by a mathematical equation. The analytical model, which includes the aforementioned mathematical equation, was proven by numerical results as preliminary work to this project proposal. In this approach, the pressure level of the medium as a pivotal correcting variable was replaced by the tangential forces and thus requires further investigation of the interrelation between the process parameters. To ensure that the extended
controlling concept includes all dependencies, the multidisciplinary cooperation with the field of machine learning is necessary. The underlying goal of this cooperation is to develop an online closed-loop control for a reliable and a reproducible manufacturing process. The controlled forming process will be completed by design guidelines, which include the processing of CAD-
files and the compensation of geometric deviations through adapting manufacturing sequences of combined convex-concave parts and tool paths