Composites Part A: Applied Science and Manufacturing, Volume 51, 2013, Pages 89-98.
E. Kappel, D. Stefaniak, D. Holzhüter, C. Hühne, M. Sinapius.
DLR, Institute of Composite Structures and Adaptive Systems, Lilienthalplatz 7, 38108 Braunschweig, Germany and
Technical University of Braunschweig, Institute of Adaptronics and Functional Integration, Langer Kamp 6, 38106 Braunschweig, Germany.
Counteracting undesired manufacturing distortions of carbon-composite structures is still a challenge for manufacturers. This paper presents a semi-numerical methodology to predict manufacturing-induced distortions of complex composite parts. The novel approach uses a combination of experimental, analytical and numerical procedures whereas standard shell elements are used for the FE calculations. The main innovation is the significant parameter reduction. Only one simulation parameter is necessary to predict the manufactured shape of a complex CFRP structure. For validation purposes, two CFRP box structures with dimensions of 798 mm × 206 mm × 55 mm (l × w × h) were manufactured and process-induced distortions are evaluated in detail using a high-precision 3D full-field measurement system. Measurements show that manufacturing distortions are mainly driven by the spring-in effect. However, a global curvature of the part is obtained which is likely due to forced-interaction induced by the aluminum male tool. The predicted process-induced deformations match precisely with the measurement results. Evaluation of the predicted distortions allow a clear distinction between spring-in induced and forced-interaction induced deformations. In consequence, obtained results show that effects due to forced-interaction need to be considered for suchlike structures fabricated on metallic tools when CTEtool is larger than CTECFRP.
Process-induced distortions (PID) are an inherent issue in composite manufacturing. The composite’s through-thickness anisotropy as well as manufacturing boundary conditions are sources of those undesired deviations between the nominal and the manufactured part shape. This lack of dimensional fidelity is a significant cost driver in aerospace and automotive applications nowadays, as it complicates assembly processes and necessitates cost-intensive tool rework/redesign. Consequently, there is a strong desire for prediction capabilities in industry which can be beneficially utilized and integrated in an existing part-design process.
The challenge, which is addressed in the featured paper, is to provide a prediction approach that possesses a well-balanced ratio between benefit and effort. With other words, the aim is to provide a PID prediction while at the same time modeling and simulation parameter costs are significantly reduced compared to existing process-simulation tools.
The featured paper represents an excerpt of ongoing studies at the DLR that focus on PID. Therein, a new semi-numerical prediction methodology is presented and applied to an integral CFRP box structure. Measured distortions on L-profiles of laboratory scale are analytically transferred into a corresponding simulation parameter. This parameter is subsequently used in combination with a shell-element based FE model of the box. Predicted PIDs are compared to full-field measurement results obtained from two manufactured boxes. As shown in the paper, the achieved accuracy of the predicted distortions is convincing what underlines the potential of the comparably simple approach. Moreover, due to the use of conventional shell-elements, modeling and computational efforts have been reduced to a minimum.
In recent studies at the DLR the focus is on PID inducing mechanisms as well as further simplification of the proposed prediction methodology. In addition, case studies of realistic industrial complexity are conducted in order to increase confidence.