Series production of sheet metal components in press shops is permanently subjected to an increasing pressure in time, cost and quality. Varying sheet metal properties due to batch fluctuations and changing process conditions lead to an increase of try out time phase and scrap rate, resulting in a demand for adaptive control strategies for deep drawing. Today, these chal-lenges are met by integrating sensors and actuators into tool structure which measure and control part quality, mainly through blank draw-in. However, such tool-based control systems require a complex and cost-intensive modification of the existing die or press technology. Against this background, the active adjustment of blank position prior the deep drawing process realized with an intelligent transfer and positioning system as a promising approach towards economic and technical aspects. This paper deals with blank positioning and its sensitivity to quality-related failures of deep drawn sheet metal components like splits and wrinkles. Therefore, a numerical study was conducted on a deep drawing process of an exemplary structural part geometry, wherein a typical fluctuation of material parameters and variation of local friction conditions was simulated. Subsequently, correlations between process disturbance and blank draw-in were elaborated, thus using local draw-in values as controlled variables for a closed loop system. Sim-ulation results showed that the manipulating parameter, i.e. the blank position prior deep draw-ing, reveals a significant influence on the draw-in and therefore on the component s quality. An essential finding of this study is the numerical proof of concept for this new deep drawing control strategy, demonstrated for different process conditions. Finally, disturbances in the deep drawing process considered could be successfully controlled by adapting the blank position.

The subjective perception of the quality of sheet metal components mainly depends on geometric characteristics and surface structures. In this context, particular attention must be paid to avoiding surface defects such as skid lines during the sheet metal forming process for components with high surface quality requirements (e.g. outer skin passenger car panels). In principal, FEM simulation can provide an effective tool for predicting such surface defects. However, the numerical modelling approaches available today do not yet allow an adequate basis for assessing their quality relevance in production, which especially applies to skid lines. In a previously published study, new skid line criteria were developed in this regard, considering unbending strain, thinning and major strain. Using a simple steel sheet part (DC06) as reference, the study showed that these criteria allow the quality relevance of skid lines to be predicted relatively accurately. This paper focuses on the validation of the proposed skid line criteria and their applicability to the materials DC06 and AA6016. Furthermore, the numerical studies presented show that one of the novel skid line criteria, which considers unbending strain and thinning, is able to accurately locate and predict the quality relevance of skid lines even for complex shaped parts.

Bei der Serienproduktion von Blechbauteilen können schwankende Blecheigenschaften zu zeitintensiven Anlaufphasen und erhöhten Ausschussraten führen. Diesen Herausforderungen wird heute mit werkzeugbasierten Regelungssystemen begegnet, welche aber eine komplexe und kostenintensive Modifikation der bestehenden Werkzeuge beziehungsweise Pressen erfordern. Vor diesem Hintergrund befasst sich der Beitrag mit einem relativ einfachen und kostengünstigen Regelungsansatz auf Basis einer adaptiven Platinenpositionierung.   In series production of sheet metal components, varying sheet metal properties may lead to an increase of try-out time phase and scrap rate. Today, these challenges are met by tool-based control systems, which, however, require a complex and cost-intensive modification of the existing die or press technology. Against this background, this paper deals with a relatively simple but cost-effective adaptive control concept based on adjustable blank positioning.

The determination and validation of material parameters required for the finite element simulation of sheet metal forming processes can be realized by a full-field optical measurement of the deformation of a test specimen in combination with a simulation-based inverse approach. Development of such an inverse approach can be quite time consuming and requires programming skills and also expertise in FEM analysis and optimization. Emerging machine learning techniques offer a practical alternative to optimization and inverse approaches provided that the ground truth is completely known and generalized by the machine learning model. To be more precise, a machine learning model can directly compute the material parameters from the experimental measurements if the hypothetical mapping function between material parameters and deformation behavior is learned as ground truth. This paper presents such a machine learning based approach for the determination of validated yield locus parameters.

Die für die FEM (Finite-Elemente-Methode)-Simulation von Blechumformprozessen benötigten validierten Materialparameter können heutzutage durch eine vollflächige optische Messung der Verformung eines Prüfkörpers in Kombination mit einem simulationsbasierten inversen Ansatz ermittelt werden. Dieser inverse Ansatz erfordert jedoch Fachwissen in der FEM-Analyse, Optimierung sowie Programmierung und kann zudem recht zeitaufwendig sein. Vor diesem Hintergrund wird in diesem Beitrag eine auf maschinellem Lernen basierende Methode zur Bestimmung von validierten Materialparametern vorgestellt.   Today, validated material parameters required for FE simulation of sheet metal forming processes can be identified via full-field optical measurement of test specimen‘s deformation combined with a simulation-based inverse approach. This inverse approach normally requires deep expertise in FE analysis, optimization, and programming and can be very time-consuming. This paper proposes a novel machine-learning approach for determining such validated material parameters.