Digital Product Engineer
Complex Systems Engineer
Digital Physics AI
Hybrid Modeling - the combination of (Digital) Physics and AI is accelerating the efficiency of engineering of today.
Using an AI tool for engineers makes it easy to apply AI without being a specialist or data scientist. Experience the power of AI demonstrated by the application of AI to Fluid-Assisted Injection Molding by Plastic Innovation.
In this webinar we will cover
Hybrid Modeling: Merging engineering with AI
Discrete manufacturing and also process manufacturing have been shaped in the past by research into the mathematical and physical description (causality-based) of production technologies. With the advent of digitization and machine learning in manufacturing comes the shift and focus on data-driven approaches. These typically do not build on these causalities and only consider data. Combining these approaches, called hybrid modeling, helps overcome the limitations of each approach. The idea is to integrate the available knowledge and causalities into a data-driven model to achieve higher accuracy with less training effort. Therefore, hybrid modeling can provide more accurate predictions at a better cost level and has the potential to save resources, shorten schedules and improve manufacturing quality.
To illustrate the value and efficiency of combining engineering and AI, in this workshop we will explain hybrid modeling using the example of “Accelerating Parameter Studies with AI in Fluid-Assisted Injection Molding to Form Hollow Structural Plastic Parts”.
About the showcase: Fluid Assisted Injection Moulding
Fluid Assisted Injection Moulding (FAIM) allows the creation of complex hollow polymeric parts, such as bicycle frames and other hollow structural parts with high torsional stiffness.
After the filling process and an adequate delay time a fluid is injected into the part and the remaining melt gets pushed out, either into an overflow cavity or back into the plasticising unit of the injection moulding machine. Thus, it is possible to produce more complex hollow structures than with an ordinary core-puller. Additionally, the fluid enables weight reduction, lower warpage, better surface finish and, when water is used as fluid, additional cooling from the inside out.
The residual wall-thickness of the part can be estimated, but for structurally loaded parts, knowledge of the thickness distribution of the hollow cross-section is crucial prior to committing to series-production. The wall-thickness depends on various process parameters affecting the viscosity of the polymer. A further complication is that these parameters are set at two processes (i.e., the injection moulding and the fluid injection process) and have a strong interdependency. A machine learning algorithm is applied here to predict the wall-thickness depending on a range of process parameters and it enables a better understanding of the interplay of the parameters. The objective is to determine the parameter window required to realise a specified residual wall-thickness needed to fulfil the structural performance of a product.