Automated Optimization for High-pressure Die Casting Dies
High-pressure die casting (HPDC) is a common industrial mass production process. In HPDC, molten metal is injected into a reusable steel die at high pressure. Due to repetitive filling however, exists an intensive thermal flow from the molten metal to the die. In order to maintain a stable temperature level of the die, it is therefore mandatory
to cool the die, so a temperature control system exists built up by cooling channels inside the die. The main advantages of HPDC are its high productivity, the good surface quality of the castings and the capability to produce thin-walled structures. On the other hand, the manufacturing effort of the dies is relatively high, so that HPDC is particularly efficient for large lot sizes. During the last couple of years however, the demand for individualized products has increased, so that the aim of the demonstrator for the automated optimisation of HPDC-dies is now to make the mass production process of HPDC more suitable for smaller lot sizes.
The gating and the temperature control systems are crucial for the design of an HPDC-die. So far there are no standardized procedures that guide the die-designer towards an ideal die. Nowadays, the toolmaker designs the die based on empirical formulas and his experience. For complex parts, casting-simulations are used to reach a good design for the gating system. Usually a lot of simulations need to be done in order to achieve a satisfactory filling behavior of the die. This iterative and therefore time-consuming approach makes the design process expensive and causes the main part of the total amount of costs for the die design. As long as the HPDC-process is used for mass production this is not a major issue. The high costs of the dies are compensated for by the large amount of parts that are produced with one die. For individualized production however, lot sizes tend to be much smaller, and the die costs are distributed over fewer parts resulting in the newly cast part not being cost competitive.
The motivation of the automated optimization approach for HPDC is to find an optimal die design with minimal effort by the die-designer, and the quality of the resulting casting should be improved. With this improvement of efficiency in the die-design process, a mass production process such as HPDC can be operated economically even for lot sizes of individualized production. The automated optimization approach has already been tested successfully for the plastics profile extrusion process. This has also been part of the Cluster of Excellence. The starting point for the optimization is the numerical simulation of the HPDC-process. The simulation data is used to evaluate the filling and solidification behaviour in the die. Also, the temperature distribution is examined. All of this can be achieved without running expensive experiments. These simulations take into account the whole die including the temperature control system. In addition to that, a quality measure is needed that allows the evaluation of the overall quality, a so-called objective function. The aim of the optimization is the minimization of this objective function, which is developed in close collaboration with casting experts. The last component that is needed for the optimisation is the representation and deformation of the geometry. In this regard it is important to determine the constraints for the geometry deformation (in this case: for the gating- and temperature control system). Again the experience of casting experts must be taken into account. The automated optimization process is designed as follows: Using a start-geometry a numerical simulation of the filling process is performed. Based on the results of this simulation, the objective function is evaluated. An optimization algorithm now uses the objective function and generates a new geometry by deforming the old one, eg., for the gating system a free-form deformation as well as a translation is possible. The modified geometry is then used as the basis for the next optimization cycle. This process is reiterated until the results show the desired quality. A die which is manufactured based on such an optimization can reduce the amount of manual simulations/trial-runs and therefore the overall manufacturing costs. In this context, the application of generative manufacturing processes for die manufacturing will be investigated as well. Using an exemplary geometry, the simulation results will be validated and the numerical models will be improved with the help of the gained data.
The numerical modelling of the filling process is a big challenge for the optimization.This has two reasons. On the one hand, the filling process lasts only fractions of a second, and on the other hand the metal flow is very complex. Due to the high flow-velocities, the melt is exposed to high pressures in the range of 300-1200 bar, and the melt front underlies turbulent movements. The complexity of the modelling is even further increased as the solidification of the molten metal also needs to be taken into account. Large variations in the material properties like density and viscosity inevitably lead to higher computational costs. In order to perform the numerical investigations in a reasonable amount of time, big computing clusters have to be used, thus implying that the simulation tools must be able to support them. In the context of automated optimization, the definition of a suitable objective function is another challenge. When developing an objective function several parameters need to be taken into account. Possible ingredients for objective functions used for the HPDC-process are a complete filling of the cavity or the compliance to certain temperature-boundaries for the die. In addition, an approach for a simple and robust way to deform the geometry during the optimization process needs to be developed.