Within the FormPlanet project, AP&T has investigated in which ways the tool try-out effort may be reduced in the sheet forming industry in order to limit the cost and lead time. The study has focused on a hot forming process of high-strength aluminium for two different aluminium grades (AA7075 and AA6010).
One of the ways that has been identified is to implement a digital twin that allows virtual forming using real tool surfaces and forming parameters, such as press and thermal data from a real forming cycle.
By applying this additional information in a forming simulation model, along with well-characterised material data, more accurate forming results can be obtained compared to simulation models with nominal tool geometries and assumed constant cycle parameters.
In addition, virtual forming can run in parallel with the tool try-out process and can allow for virtual tool compensation based on the simulation results prior to actual re-work of tooling is performed.
Parameters identified for simulation model setup
The most crucial component in this process is the setup of the simulation model itself. In AP&T’s first blogpost published in May 2021 (Creating a digital twin for hot forming of aluminium to reduce the effort during tool try-out), it was described which input parameters from the forming cycle that are required to set up the simulation model of a door ring in high strength aluminium. These were defined to be:
- 3D-scan of the real tool surfaces – to be used as the tool surfaces in the simulation model.
- Friction coefficient between tool surfaces and blank – a lubricant was applied to the tool surfaces prior to forming to reduce potential galling.
- Input parameters from the servo press such as slide speed and cushion force – to represent the kinematics of the forming cycle.
- Blank temperature at the start of the forming cycle – due to the transportation time between the furnace and the press the blanks cool down a bit, specifically from a solutionising temperature in the furnace of 482°C to 432°C at the start of forming for AA7075, and from 560°C to 530°C for AA6010.
- Quenching time when the tool is closed – held for 10 s at 6000 kN.
- Heat transfer coefficients for aluminium depending on gap and pressure.
- Thickness measurements of formed parts – for thinning and thickness comparison with simulated parts
In addition to these parameters, it was essential to construct a viable material card to be used in the simulation model. Together with research institutes Fraunhofer IWU and COMTES FHT, AP&T has defined the requirements of the material characterisation scheme for the two aluminium grades in order to finalise material cards for the alloys to be used in the simulations.
The goal for the research institutes was to determine the true stress-strain curves and anisotropy values for the different aluminium grades by performing hot tensile tests at different temperatures and strain rates. By implementing the true stress-strain curves into the Hockett-Sherby equation it was possible to extrapolate flow curves to higher strain values. Flow curves for AA7075 are depicted in Figure 1.
Figure 1: Flow curves for AA7075 at different temperatures and strain rates
In addition to these tests, also hot tensile tests in three different load states to describe the forming limit characteristics (FLC) of the alloys were performed. This was done for uniaxial strain, plane strain and biaxial strain. Within the investigation, the temperature 280°C, 360°C and 400°C have been analysed for AA7075 and 300°C, 350°C, 400°C and 450°C for AA6010. Unfortunately, due to equipment restrictions the strain rate used during these laboratory tests for AA7075 was very low – in the 0.005 – 0.01 s-1 range. As will be shown in the results later, this strain rate does not reflect the real strain rate in the material behaviour for this hot stamping application. Hence why the experimental FLC data for AA7075 are not applicable. The forming limit curves for AA7075 and AA6010 are described in Figure 2.
Figure 2: Forming limit diagram for AA7075 and AA6010 at 2 mm and different temperatures.
Simulation results and validation process
After the practical forming, the parts were 3D-scanned in order to have an overview of the thickness and thinning results. There are several critical areas of interest in the part that can be used to compare the thickness of the formed part to the simulation results. The critical areas are marked in red in Figure 3 and were used to validate the simulations. These regions exhibit different load states which enable a meaningful validation of the material input data used in the simulation model. In particular one region can be used to validate the result with the help of forming limit curves, as this region shows a crack for AA7075 and strong necking for AA6010 for all formed parts respectively.
Figure 3. Critical regions marked in red in the 3D-scanned formed part with highlights of the zone that exhibits a crack for AA7075 and strong necking for AA6010
Comparing the results of the simulation with the 3D-scanned part for AA7075 shows an accurate outcome, illustrated in Figure 4. A deviation of ± 0.2 mm occurs which is acceptable.
Figure 4. Thickness of a 3D-scanned AA7075 part (top) and thickness result from the simulated part (bottom).
The large deviations of the sheet thickness in the critical region where the crack is formed during physical testing, defined in Figure 3, can be explained by not using failure models to determine crack behaviour in sheet metal forming simulations. This leads to incorrect results when the crack starts to develop. Classic sheet metal forming simulations only calculate the evolution of the sheet thickness, temperature, stress and strain of the formed part. To predict necking, which is defined as material failure, the forming limit curve can be used.
In the region where the crack occurs, in the simulation there is a strain rate of 0.001 ms-1. Since the time unit system is milliseconds used in PamStamp this equals a strain rate of 1 s-1. Applying the FLC data to the result can be seen in Figure 5, showing that this zone is safe compared with the same zone in Figure 3 for AA7075. The reason for this is because the strain rate used during the experimental tests was 0.005 – 0.01 s-1, which is lower than the strain rates during forming.
Figure 5. FLC data from Fraunhofer IWU for AA7075 indicating that the critical zone has no crack.
The simulation results for both aluminium alloys show promise in terms of thickness behaviour compared to the formed parts. However, the material was described as isotropic in the sheet metal forming simulation model which might lead to less accuracy of the calculated results. To enhance the simulation performance, an anisotropic characterisation of the material is suggested to be implemented in the material model in future projects. Furthermore, the heat up rate to solutionising temperature and the cooling rate after solutionising to the specified test temperatures differ in the industrialised process. The cooling rate was set constant during the tensile tests, but during part production the cooling rate of the blank varies and is inhomogeneous as certain areas of the blank are cooled faster than others in the tool. The cooling rate during part forming depends on the contact conditions in the tool, such as pressure and distance between blank and tool surfaces. Considering this, the flow curves received by the laboratory tensile tests may differ from the actual flow behaviour of the material in the forming process and can be the reason that a deviation in the results between formed and simulated part are seen.
Apart from the material cards and flow curves, the most influencing parameters during the simulation trials that affect the result heavily are the forming speed of the slide and the friction between tool and blank. The forming speed cannot be changed as this is an input from the press but the friction coefficient between blank and tool is influenced by the temperature, speed, pressure and amount of lubricant. Those parameters are changing during the forming process which leads to a variable coefficient of friction. However, in the current simulation model, the friction coefficient was simplified and set to be constant. This assumption affects the draw-in of material and as such the thinning and thickening results of the simulated part.
To conclude, the simulation results are satisfactory in terms of feasibility analysis of the hot forming process of aluminium to predict thinning and thickening behaviour of the part. The material input data from Fraunhofer IWU and COMTES FHT can be used for future work and forming simulations of other parts geometries. The obtained material data can serve as a base when formulating the material card in the FEM simulation setup when studying springback and virtual tool compensation.
Furthermore, the general approach to validate the simulation results is reliable. The practical trials showed repeatable forming results which meant the formed parts could be 3D-scanned. As such, the press output data, along with the 3D-scans of the tool surfaces, could be implemented in the forming simulation to represent precise kinematics once the FEM model was built. Combining this together with the material cards that were made using the material data from Fraunhofer IWU and COMTES FHT indicate that any future work regarding FEM simulation for hot formed aluminium sheet metal alloys can use the same validation approach.
Vedran Kovacevic
Master’s degree in mechanical engineering at Lund University.
Part of the R&D team at AP&T’s department Forming Technologies as a Research and Test Engineer.
Main research topics and development areas are within aluminium and composite materials, forming simulations and forming techniques.
