Injection molding AI Tool Calculates injection-Molded Components up to 100 Times Faster

Related Vendors

Universität Augsburg

dataTec AG

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

SIMTOS (Seoul International Machine Tool Show)

Whether it's cordless screwdriver housings or toothbrush handles—many everyday products are created through injection molding. Before a new tool is manufactured, simulations are required that often take hours. Researchers have developed software that combines classical physics with AI—delivering calculations up to 100 times faster than conventional solutions.

The Most Important Points at a Glance

• Researchers at the University of Augsburg (Germany) combine physical models with AI-supported methods
• Simulations for component design in injection molding run in seconds instead of hours
• The model requires only a few hundred training datasets thanks to incorporated prior knowledge
• A specially developed AI-compatible finite element software is set to continue learning from production data in the future
• Long-term goal: an AI tool that generates complete components suitable for injection molding

Accuracy vs. Speed: The Trade-Off in Practice

If a company wants to produce a new component via injection molding, feasibility and mold design must first be examined. Typically, software solutions based on physical flow simulations are used for this purpose. While these deliver precise results, they are computationally intensive—a single simulation can take several hours. Especially in the early phase of product development, where many variants need to be compared quickly, this quickly becomes a bottleneck. 

This is where the team led by Prof. Dr.-Ing. Nils Meyer from the Centre for Future Production at the University of Augsburg comes in. Meyer heads the Chair for Data-driven Product Engineering and Design. "A tool that can check multiple possibilities in seconds—such as the most suitable injection point—can save valuable resources for companies: time, computing power, money, and energy," explains Meyer. "We have managed to accelerate the process by up to a factor of 100." The highly accurate physical methods are not rendered obsolete in the process—according to Meyer, they are used as a second step for the final calculation of the component.

Few Data, Arbitrary Geometries

A key challenge: The AI-based tool must be able to make predictions for any component geometry—and with as few training data as possible. Instead of relying on vast amounts of data, the Augsburg team has deliberately integrated physical prior knowledge into the AI model."For example, we already know that an area far from the injection point will be filled later than one right next to it. We don't need to learn that from data first," says Meyer. This prior knowledge significantly reduces the training effort: A few hundred examples are sufficient for the model to make predictions for entirely new components.

AI-compatible finite element software as a key element

A special feature of the development is an AI-compatible finite element software developed by Meyer's research group. The finite element method breaks down complex components into many small, easily calculable parts – for example, to predict the warping of a component during cooling. The crucial advantage: "Through this software, we can also continue training our AI model. This allows the model to keep learning from measured deformations in the production process and become increasingly better," explains Meyer.

Outlook: AI as a Design Partner

Currently, the software primarily supports the optimization of already designed components. However, Meyer's team is already thinking ahead: In the future, the AI is expected to accompany the entire design process."We envision a program where you specify your component requirements, and it is then generated injection-molding-ready with AI support. This helps designers quickly find the best solution from the vast range of possible components. It relieves them of repetitive tasks and gives them more time for creative processes and product requirements," Meyer summarizes the vision.

We envision a program where you communicate your component requirements, and it is then AI-generated to be suitable for injection molding.

Prof. Dr.-Ing. Nils Meyer