ComoNeoPREDICT and STASA QC Optimize reduce production costs and enhance quality
Like many other sectors, the plastics industry is currently experiencing significant cost pressures. Many production facilities are struggling to cope with rising raw material, energy and labor costs. And these financial pressures are exacerbated by new resource conservation and carbon pricing regulations. Especially at times when production machinery is not operating at full capacity, it is worth taking time to examine the basic process for optimization and cost-saving opportunities. Digital solutions can provide invaluable assistance, as demonstrated by ComoNeoPREDICT and STASA QC Optimize, two software solutions specially developed for this purpose by the experts at STASA Steinbeis Angewandte Systemanalyse GmbH.
Technological applications face two general challenges in real-world environments. The first is that production processes involve a combination of multiple complex interactions. The second is that, for technical and financial reasons, big data is rarely available for process setup and control. Moreover, the algorithms based on big data often only produce very general conclusions that lack the necessary degree of precision. Injection molding is one of the production processes confronted with both of these challenges. This complex process is influenced by the large number of possible machine settings, the technical specifications of the injection molding machine and the internal regulation processes, leading to chaotic variability in melt quality. Furthermore, the geometry of the injection mold is itself extremely complex, and additional processes are required to control and regulate the temperature profiles in the mold over the course of the production cycle. Each and every molded part produced is thus unique, since its quality is determined in the mold.
In addition to this, strong national and international competition creates severe cost pressures in the plastics industry. Two areas where cost-saving opportunities exist are process setup and optimization and the reduction and identification of rejects. Accordingly, it is important to minimize the number of setup adjustments needed to get the process up and running. At the same time, the requirements in terms of molded part quality, emission reduction and rejection of molded parts that do not meet the required quality standard are getting ever higher.
AI can help
The STASA GmbH team has developed two AI products that can overcome these challenges without using big data. STASA QC Optimize helps to optimize production processes, while ComoNeoPREDICT provides a process monitoring solution.
STASA QC Optimize creates a digital twin of the production system in order to find the optimal settings. These can be evaluated on the basis of quality and dimensional accuracy, but also other criteria such as energy efficiency improvements or reduced cycle time. The process can also be used at no additional cost in other industries or for any cyclical process. Building on STASA QC, the Steinbeis experts have also developed the ComoNeoPREDICT software solution in partnership with Swiss company Kistler Instrumente AG. It can be integrated into Kistler’s existing ComoNeo system and enables cavity pressure-based monitoring, control, regulation and documentation of the injection molding process. ComoNeo can also access process data from the current cycle and statistical production data from injection molding operations all over the world from any location using a Web browser.
STASA QC Optimize: digital twin improves quality
Production processes are often optimized by trial and error. The setup technician uses his or her experience to keep tweaking a machine’s parameters until they find the right settings to allow production to begin. This approach makes it almost impossible to find the truly optimal settings. And once a production setting has been found, there is often no way of knowing how stable the process will be. Even tiny variations in the external factors can mean that the machine suddenly starts producing rejects instead of in-spec parts. It is also difficult to reduce energy consumption or cycle times without deviating from production tolerances.
STASA QC Optimize makes it easier to find the right settings through a systematic design of experiments (DoE) that minimizes the number of adjustments needed to cover all possible settings. The AI process enables a significant reduction of production tolerances by automatically generating an AI quality model for each individual quality attribute and, if need be, each individual cavity. This also allows subtle differences between different mold shapes to be modeled. In addition, process stability is predicted for every possible setting. This makes it possible to see how stably the process will behave within the tolerances. In other words, a digital twin models and displays the entire process. Correlations between input and output variables are also displayed, providing detailed insights into the process. This means that STASA QC Optimize also serves as a valuable teaching tool.
Zero-defect production with ComoNeoPREDICT
Defective parts can be detected by manual spot checks or using special sensors. In injection molding, these are typically temperature and pressure sensors in the mold. The cavity pressure profile is like a fingerprint for the forming of molded parts and is the most significant variable for the assessment of part quality. Deviations from the optimal profile are a clear sign of process variability that results in defective molded parts. STASA has developed a patented AI process to achieve the desired correlation between part quality, machine settings and the time-dependent sensor data. The network topology is automatically generated and individually adapted to the necessary complexity of the underlying correlations, without the user having to intervene in the process. A continuous analysis of the production process predicts the final molded part dimensions. Faults, process variability revealed by the sensor data, and long-term process changes that impact quality are detected immediately by the self-generating neural network in the ComoNeoPREDICT process monitoring system. Changes in the process curves that have no impact on quality, for example due to a replaced sensor, are
identified and compensated for by an additional AI process. This means that the part can still be evaluated despite the change in the curves. ComoNeoPREDICT makes it possible to decide during each individual cycle whether a molded part will meet the required quality standard while it is still being produced. The measurement and AI-based analysis of the sensor signals thus enables zero-defect production. When used together, STASA QC Optimize and Kistler’s ComoNeoPREDICT can significantly reduce production tolerances, enabling the production of very large volumes to the highest quality standards.
Optimizing the entire process
The tolerance limits for process optimization and process monitoring can be taken directly from the part specifications and set as monitoring criteria. The relevant correlations between process curve shape and quality criteria are identified during mold validation and subsequently used for process monitoring. This constitutes a significant step forward in in-process quality control.
In practice, production costs can be reduced by an average of ten to 15% compared to conventional processes. STASA QC Optimize’s AI processes reduced optimization time to between a third and a fifth of the time needed for conventional optimization methods (Kunststoffe 10/2022). Material efficiency is increased by ten to 20%, while reduced cycle times improve energy efficiency and thus reduce the process’s carbon footprint (BINE Projektinfo 07/05; fact sheets KUZ, Leipzig). For high-quality molded parts, the prevention of rejects enabled by ComoNeoPREDICT translates into an energy saving of approximately 15% (BINE Information Service 07/05).
Contact
Niklas Janssen (author)
Managing Director
STASA Steinbeis Angewandte Systemanalyse GmbH (Stuttgart)
www.stasa.de