July 8, 2024
Following two years of collaborative research, the SKZ Plastics Center and Fraunhofer Institute for Manufacturing Engineering and Automation IPA have recently announced the successful completion of the ProBayes development project.
The newly developed software can monitor and control injection molding systems based on Bayesian networks or structured statistical models that learn from data to make accurate predictions. It detects variations in product quality, pinpoints likely causes, and provides guidance on corrective actions.
Injection molding is widely used to process thermoplastic polymers — plastics that can be melted and reshaped repeatedly. It involves melting plastic pellets in a heated barrel and injecting the molten plastic into a mold cavity for cooling and hardening into the desired shape.
Topically appearing efficient, the process outcome actually depends largely on several factors, which may not be easy to control. Unplanned variations in these factors often lead to product defects, reflected in part weight, dimensions, warpage, shrinkage, and surface defects.
These are dependent on machine settings. They include:
They depend on machine settings, the material used, and the mold design. These variables include melt temperature and pressure in the cavity.
There is also a shortage of skilled workers despite increasing political and social demand for more use of recycled materials.
Identifying the root cause of these defects and implementing corrective measures is essential for improving product quality and reducing waste. As such, this was the primary focus of the project.
The software collects data from measurement systems in the injection molding machine and utilizes Bayesian networks to analyze it. In the Bayesian networks, as explained by the research partners, process parameters — machine variables, process variables, and quality characteristics — are represented and visualized as nodes.
The relationships between them are represented by corresponding arrow connections. These distributions enable the system to interpret dependencies among various process variables and thus make precise predictions.
The software then continuously monitors these process parameters while comparing them with the expected values. Deviations are flagged as defects, and suggestions for corrective actions are made.
The project, funded by the German Federal Ministry of Economics and Technology, will help reduce production costs and improve product quality, which are integral to customer satisfaction. It will also set the stage for digitalizing injection molding production processes, paving the way for Industry 4.0.
The software’s efficiency was tested and validated in a fully integrated injection molding cell, the setup of which comprised the injection molding machine, sophisticated measurement systems, and other essential peripherals. The integration was via OPC UA and MQTT, industrial connectivity for data exchange.
This networking allows for real-time data collection and analysis and provides immediate feedback on the quality of each manufactured part. The system can also record and store data for every production cycle.
In addition to precision in quality control, this technology benefited the partnering companies through knowledge of interfaces and networking.
The developed Bayesian network maintains high-quality production standards and minimizes waste by performing root cause analysis when the part weight deviates from the desired value.
It’s able to determine if such deviations occur and their causes by mapping the part weight against various settings and process variables of the injection molding machine.
This operation, which is trained on real process data, was validated on the injection molding cell in live operation and showcased to vested companies. The Bayesian network was capable of detecting deviations in part weight beyond the adjustable tolerance and most likely cause of said deviations, as well as recommended corrective actions.
Interested parties are invited to witness the capabilities of the new process monitoring software at the SKZ Plastics Center. This opportunity will also allow them to have the suitability of the process for their own production processes.
Jonathan Lambers, a senior scientist and project manager at SKZ, took to the press to highlight the significance of the accomplishment and his satisfaction with the results. Christoph Kugler, Group Manager Digitalization SKZ, also made some positive remarks about the project.
Kugler noted the economic significance of the new process monitoring software, saying that it has set the stage for the profitable integration of the process into real production processes and the digitalization of injection molding production.
Lambers also highlighted the contributions of the team and collaborators, which were essential for the project’s success. Some helped with data collection, others with data science, and others with the injection molding machine. This level of collaboration is integral in advancing technology and improving processes in the injection molding industry.
The Plastics Industry Association (PLASTICS) is committed to supporting technological advancements in the plastics supply chain, such as the new process monitoring software from the ProBayes development project. As a leading advocate for the plastics industry, PLASTICS promotes tech-driven innovations that enhance productivity, improve quality, and promote sustainability in the sector.
Companies interested in new technology can benefit from PLASTICS’ extensive network of indu
PLASTICS and the Future Leaders in Plastics (FLiP) Committee are devoted to supporting and encouraging the next generation of plastics leaders who will play a crucial role in the innovation, technology and future of the plastics industry. FLiP’s mission is to provide young professionals under the age of 40 the exposure, education and resources they need to build lifelong careers in plastics. Want to join? Want to get your employees involved? Email: [email protected]
