Automated Optimization of Profile Extrusion Dies

  Flow simulation of a die Flow simulation of a die

Profile extrusion is an established process for the mass production of plastics profiles. However, the development
and manufacturing of the profile die is combined with high efforts and costs. These are mainly caused by elaborate running-in trials before the start of production. Because of this, a cost-effective use of profile extrusion in small series is not possible yet. Nevertheless, the customer demands are increasingly individualized, and product life cycles continue getting shorter resulting in smaller lot sizes. Numerical flow simulations have the potential to provide substantial time and cost savings in the design of profile dies. However, the currently available simulation tools lack the ability to automatically improve the flow channel geometry based on calculation results. Therefore, an optimization framework for the design of profile extrusion dies is being developed in the demonstrator for the automated optimization of profile extrusion dies. The framework combines the flow simulation with an optimization algorithm. This combination enables an automated optimization of the flow channel geometry in profile extrusion dies.


Practical Issues

Design features in the flow channel

Due to the almost unlimited complexity of possible profile cross-sections, an analytically exact design of the profile extrusion die is usually not possible. Instead, the design and manufacturing often take place in form of a “trial and error” process. Starting from a first draft, the die designer changes the flow channel geometry iteratively based on empirical knowledge and insights from the running-in trials. This happens via deposition or removal of material until the produced profile satisfies the previously defined quality criteria. Depending on the complexity of the geometries this can take up to 15 iterations. This type of production is very time-consuming and costly, the extruder is not available during the running-in trials for the regular production and considerable costs occur for the experimental materials. A shortening of the design Figure 4: Approach in the design and manufacturing of profile extrusion dies process is offered by flow simulations with the stabilised FEM (Finite-Element-Method) which transfers the iterative design process from the machine to the computer. Through the numerical analysis of the flow processes, the running-in process is simplified as vulnerabilities can be detected even before the manufacturing of the die. However, in this case the experience and the knowledge of the die maker are still necessary to decide how the
geometry has to be changed in order to achieve the design goals. Therefore, the design process is still dominated
by its iterative, time-intensive nature.




The fundamental approach to finding a solution to the problem is to imitate the real process of continuous improvement of the die. The realisation of this process progresses through the usage of optimization algorithms which minimize a so-called objective function. The objective function mainly reflects the quality of the die. In order to simulate the real process, it is necessary to model the reality within certain limits. Here we use several methods that replicate all the relevant properties and the behavior of the real process, but are still calculable by high performance computers or ideally ordinary computers within acceptable times. It is crucial for this optimization application to get to the target in as few steps as possible. The background is that one tries to simulate as few geometries as possible, because each simulation is time consuming and computationally intensive and, in the end, again costly. In this respect, this demonstrator goes beyond the numerical and experimental components and integrates ”best practices” of the industry and the manufacturing process. Another future step is the use of Laser Powder Bed Fusion (L-PBF) produced components and modules within the dies, so that not only previously unmanufacturable but also rapidly deployable modules can be used.


Technical Challenges

Due to the non-intuitive behaviour of the plastics, the representation of reality is one of the biggest technical challenges. The core property which complicates the behavior of the plastics to such a degree is called “viscoelasticity”. Viscoelasticity is defined as the superposition of elastic and viscous flow behavior; its mathematical description is very complicated. This additional complexity also flows into the simulation. Furthermore, the definition of the objective function has a decisive influence on the optimization process. Both the choice of the measured quantity and the actual mathematical formulation play an important role. Ultimately, the goal is to achieve the maximum efficiency in the implementation of the parametric design and optimization algorithms and hereby realize a minimum number of numerical iterations. It is necessary and also especially challenging to involve institutions into a close work process to validate theoretical simulations in practical experiments and vice versa to influence the models.