1 Introduction In the cardboard package production process, the cardboard bottom and the paper roll are joined together (flanged and glued) in a flanging process. In Figure 1a, the flanging process is schematically illustrated and it can be observed that in the first step, the paper roll is pushed onto a rigid aluminium holder and the cardboard bottom is inserted into the paper roll. Hot glue is applied radially onto the outer side of the cardboard bottom and onto the inner side of the paper roll. In the second step, by means of a pneumatic press, the forming tool rotates and presses the paper roll onto the cardboard bottom. Thereby, flanging arises, which in combination with hot glue fixes the paper roll onto the paper tray. In Figure 1b, the bottom of a cardboard food can assembled with the described flanging process is shown. Depending on the desired end product, the paper composite system can be uncoated or coated with a transparent varnish. In this process, the cardboard is plastically deformed and large strains may occur in the final product, which under unfavourable process conditions lead to damage of the cardboard surface. Paper is an anisotropic material with a pronounced hygroscopicity [1]. Typically, it has a high strength in the machining direction (MD) and high elasticity in the cross direction (CD). In addition, paper is characterised by a certain inhomogeneity related to fibre orientation, filler content, etc. [2]. In the case of the flanging process considered herein, the material stays in the area of elastic deformation. Aus Wissenschaft und Forschung 19 Tribologie + Schmierungstechnik · 66. Jahrgang · 2/ 2019 DOI 10.30419/ TuS-2019-0009 Modelling and simulation of a paper forming tool - Optimization of the flanging process Florian Ausserer, Igor Velkavrh, Stefan Klien, Joel Voyer, Georg Vorlaufer, Alexander Abbrederis* Für den Herstellprozess von Papier-Lebensmitteldosen werden Verpackungsboden und eine auf Maß geschnittene Papierrolle in einem Bördelprozess mit einander verbunden (gebördelt und verklebt). Bei diesem Prozess wird das Papier plastisch verformt und große Dehnungen können im Produkt auftreten, welche unter ungünstigen Prozessbedingungen zu einer Beschädigung der Papieroberfläche führen. Durch die Kombination eines Design of Experiments (DoE) Ansatzes mit realitätsnaher Systemsimulation (digitaler Zwilling, FEM-Simulation) und experimenteller Validierung in Laborversuchen konnte der Zeit- und Kostenaufwand für eine optimierte Werkzeuggeometrie (Bördelprozess) deutlich reduziert werden. Schlüsselwörter Bördelprozess, Papier, FEM, Design of Experiments (DoE), Reibung For the production process of cardboard cans, the bottom of the packaging and the paper roll are joined together (flanged and glued) in a flanging process. During this process, the paperboard is plastically deformed and high elongations can occur in the product, which, under unfavourable process conditions, lead to damage of the paperboard surface. By combining a Design of Experiments (DoE) approach with a realistic system simulation (digital twin, FEM simulation) and experimental validation in laboratory tests, the time and cost required to obtain an optimized tool geometry (flanging process) could be significantly reduced. Keywords Flanging Process, Paper, FEM, Design of Experiments (DoE), Friction Kurzfassung Abstract * DI (FH) Florian Ausserer MSc. Orcid-ID: https: / / orcid.org/ 0000-0001-8541-3063 DI Dr. Igor Velkavrh Orcid-ID: https: / / orcid.org/ 0000-0002-4293-9978 DI (FH) Stefan Klien Orcid-ID: https: / / orcid.org/ 0000-0001-8727-4909 Dr. Joel Voyer Orcid-ID: http: / / orcid.org/ 0000-0002-0649-8140 V-Research GmbH, Stadtstrasse 33, 6850 Dornbirn, Austria DI Dr. Georg Vorlaufer AC2T research GmbH, 2700 Wiener Neustadt, Austria Mag. Alexander Abbrederis pratopac GmbH, 6833 Klaus, Austria T+S_2_2019.qxp_T+S_2018 16.04.19 13: 48 Seite 19 2.2 Experimental Design By applying the methods of statistical planning, the experimental parameters and the amount of required variables can be designed so that objective conclusions about the system under investigation can be made as accurately and time-efficiently as possible [3]. The application of the so-called Design of Experiments (DoE) methods to a real problem presupposes that the factors influencing the system under investigation are known and definable. In the general case, the input variables (X) and the disturbance variables (Z) act on a system and as the system response, an output variable or a quality characteristic (Y) is received. In the system under investigation, the experimental model was based on the FEM model where the input variables (X) were defined as the geometrical parameters of the forming tool: radii R1, R2 and the inlet taper W1 (Figure 3); the disturbances (Z) were neglected; while the system response was defined as the reaction force acting Aus Wissenschaft und Forschung 20 Tribologie + Schmierungstechnik · 66. Jahrgang · 2/ 2019 DOI 10.30419/ TuS-2019-0009 In addition to the mechanical material properties, the tribological properties of the paper surface and the forming tool are of crucial importance for the production of geometrically accurate flanges. During the flanging process, considerable frictional and forming forces may occur between the forming tool and the paper roll, which can lead both to tool wear as well as wear on the surface of the deformed paper (flange) in the form of cracks or tears. By reducing the frictional and the forming forces while maintaining the wear resistance, the flanging process can be optimized so that surface damages may be greatly minimized or completely eliminated. In addition, under lower frictional and forming forces, smaller and visually appealing flanging radii can be realized in the flanging process. 2 Modelling and Experimental Setup 2.1 FE Simulation For the optimization of the tool geometry, the flanging process was modelled using a 3D digital model, which described the movement dynamics (turning, closing and opening), the material behaviour (elasto-plastic material response) and the contact conditions (frictional forces, surface roughness) as accurately as possible. With the help of these system simulations, the so-called digital twins, geometrical variations in “virtual experiments” were subsequently studied. In Figure 2, the 3D digital model is presented. The model is based on the finite element method (FEM) and describes the structural mechanical deformations as well as the mechanical stresses occurring in the paper and in the forming tool by utilizing a radial symmetry, i.e. a two-dimensional cross-section is considered, which is converted by rotation around the vertical axis into a three-dimensional representation. The calculation is quasi-static, i.e. the simulated closing and opening processes are time-dependent and the inertial forces are neglected. For the paper or the cardboard, an elastically ideal plastic material model was chosen, while for the forming tool a linear elastic material behaviour was applied. The contact between the cardboard and the forming tool was described using the Coulomb friction coefficient variable friction model. Figure 1: (a) Schematic of the flanging process and (b) bottom of a cardboard food can produced in the flanging process. Figure 2: Modelling of the flanging process by means of the so-called digital twins. (a) (b) T+S_2_2019.qxp_T+S_2018 16.04.19 13: 48 Seite 20 on the paper roll. A fully-fractional experimental plan with a factor variation on two levels was prepared, comprising 2 3 trials (Table 1). 2.3 Tribological Tests The tribological tests were performed with a modular friction and wear tribometer RVM (Dr. Tillwich GmbH Werner Stehr, Germany) shown in Figure 4a. Figure 4b represents schematically the experimental setup used for the tribological tests. In Table 2, the experimental parameters applied in the tribological tests are summarised. For the experimental investigations, a surface contact with an area of 130 mm 2 was applied. In order to better understand the influence of the surface properties of different paper-composite systems on their frictional behaviour, their surface energies were previously measured. It was found that the dispersive fraction of the surface energy is dominant in all paper-composite systems studied and that on the paper-composite systems with a high surface energy, the test liquids have a smaller contact angle and spread more than for paper systems having low surface energy. Therefore, as candidates for the coatings for the forming tool, coatings with low surface adhesion were selected. Aus Wissenschaft und Forschung 21 Tribologie + Schmierungstechnik · 66. Jahrgang · 2/ 2019 DOI 10.30419/ TuS-2019-0009 Figure 3: Input variables (X) for the DoE model: R1, R2 and W1. Figure 4: (a) Modular friction and wear tribometer RVM and (b) schematic representation of the experimental setup. R1 R2 W1 1 -1 1 -1 1 -1 -1 -1 1 -1 1 1 1 1 -1 1 -1 -1 -1 -1 -1 1 1 1 Table 1: Non-dimensionalized fully-fractional experimental design (-1 corresponds to the minimum value, 1 corresponds to the maximum value). Test parameter (SI-Units) Value Normal force F N (N) 65, 130, 195, 260 (continuous load level increase) Type of movement linear oscillating Oscillation frequency n (s -1 ) 0.07 Stroke (m) 0.024 Test duration (s) 60 per load level Table 2: Experimental parameters applied in the tribological tests. (b) (a) T+S_2_2019.qxp_T+S_2018 16.04.19 13: 48 Seite 21 Figure 7 shows the calculated contact force curve for exemplary flange geometry. It can be clearly seen that the occurring stress peaks are accompanied by significant structural changes of the paper tube. Figure 8 shows the dependence of the reaction force (qualitative parameter) on the flange tool radii R1 and R2, calculated by means of DoE. In addition, geometry parameters of the original Aus Wissenschaft und Forschung 22 Tribologie + Schmierungstechnik · 66. Jahrgang · 2/ 2019 DOI 10.30419/ TuS-2019-0009 In Table 3, the coatings for the forming tool which were tested in the tribological tests are listed. 3 Results 3.1 Tribological Tests Figure 5 shows the coefficients of friction measured in tribological tests under contact pressures of 0.5, 1, 1.5 and 2 N/ mm 2 . From Figure 5, it is clear that for 42CrMoS4 steel when uncoated, treated with plasma nitriding or coated with DLC coating, coefficients of friction were between 0.24 and 0.28 and thus, higher compared to the uncoated cast iron EN-GJS-400-15, which provided a coefficient of friction of around 0.18. However, when 42CrMoS4 steel was coated with a PAEK coating (variants V1 and V2), coefficients of friction of 0.05 to 0.06 were measured; i.e. the reduction of the coefficient of friction compared to the benchmark was around 70 %. Based on the results from the tribological tests, a constant coefficient of friction of 0.06 was subsequently used for the FEM model. 3.2 Numerical Simulations and DoE One of the central issues of the present study is the determination of the changes of the contact force during the flanging process. Since these data cannot be determined directly from the flanging machine, the flanging process was modelled in a tension-compression machine. Figure 6 shows the force-displacement diagrams measured in the tension-compression test (strain press) and calculated in the FEM simulation. By applying suitable variations of the yield strength and the modulus of elasticity, the elastic-ideal-plastic material model of the paper roll could be adjusted so that it corresponded to the measured force-displacement curve with sufficient accuracy. The FEM simulations were performed by using the material model with the force-displacement curve identified with an arrow in Figure 6b. Substrate material Surface treatment / coating S a (µm) EN-GJS-400-15* - 1.274 42CrMoS4* (1.7225) - 0.712 Plasma nitriding* 0.253 DLC 0.346 PAEK-V1 8.247 PAEK-V2 4.960 Table 3: Concepts for the forming tool coatings and their surface roughness values S a . Sample A corresponds to the benchmark material. Samples denoted with * were grinded. Figure 5: Measured coefficients of friction of the investigated coating concepts at contact pressures of 0.5, 1, 1.5 and 2 N/ mm 2 . Figure 6: Force-displacement diagrams (a) measured in the tension-compression test and (b) calculated in the FEM model. Due to simulation constraints, the starting values on the X-axis are different. (a) (b) T+S_2_2019.qxp_T+S_2018 16.04.19 13: 48 Seite 22 flange contour Y1 are shown in comparison to the optimized flange contour Y2. It can be clearly seen that the change of the radii results in a reduction of the reaction force. In real component tests, it was confirmed that solely by applying the described geometry optimization, i.e. using the benchmark EN-GJS- 400-15 flange tool without additional material or surface optimization, the reaction force is reduced by up to 20 %. 4 Conclusions The developed material model (elastic-ideal-plastic) of the FEM simulations has shown to be well suited for the applied comparative parameter investigations. Paper and paper composite systems are very difficult to model due to their multi-layered structure. Therefore, a compromise between the model accuracy and the computational effort needs to be found. In this study, it was found that the yield point has a greater influence on the simulation results than the modulus of elasticity. By using a DoE approach, linear model dependence between the reaction forces and the radii of the flange tool was calculated, which saved additional computation time. Through the combination of DoE, numerical simulations (digital twins), and experimental validation in laboratory experiments, the time and cost required for the optimization of tool geometry could be significantly reduced. In addition, the flange tool geometries which cannot ensure a proper flanging action (due to e.g., buckling or blocking of the bending process) could be predicted. Based on the results of this work, a new flanging tool was manufactured, which is currently undergoing testing in the real production process and is providing an improved performance as compared to the benchmark. The reduction of the tool force is comparable to the force predicted in the numerical simulations. Acknowledgements This work was funded by the Austrian COMET Programme (Project XTribology, no. 849109) and carried out at the “Excellence Centre of Tribology” (AC2T research GmbH) in cooperation with V-Research GmbH and pratopac GmbH. References [1] E. Dörsam, Papier: Eigenschaften und Verwendung, TU Darmstadt, WS 2010/ 11 [2] R. Heiss: Verpackungen von Lebensmittel, Springer Verlag, 1980, ISBN-13: 978-3-540 [3] J. Antony, Design of experiments for engineers and scientists. Oxford ; Burlington, MA: Butterworth-Heinemann, 2003. Aus Wissenschaft und Forschung 23 Tribologie + Schmierungstechnik · 66. Jahrgang · 2/ 2019 DOI 10.30419/ TuS-2019-0009 Figure 7: Simulated contact force and stress curve during the flanging process for the exemplary flange geometry. Figure 8: (a) Dependence of the reaction force (qualitative parameter) on flange tool radii R1 and R2 and (b) example of optimized flange tool geometry Y2 compared to the benchmark flange tool geometry. (a) (b) T+S_2_2019.qxp_T+S_2018 16.04.19 13: 48 Seite 23