Cross phase process concepts for injection moulding using modern control strategies

• Applicants: Professor Dr.-Ing. Dirk Abel; Professor Dr.-Ing. Christian Hopmann
• Subject Area: Plastics Engineering
• Term: since 2017
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 378417139

In the first funding period, a cross-phase process control was investigated. This consists of trajectory planning and process control. The trajectory planning is based on a quality model and specifies the cavity pressure trajectory to be realized by the process control. The process control is an adaptive model-based predictive control, which can be parameterized with low effort. The cross-phase control of cavity pressure allows to reduce the number of process control variables. In addition, the process control is more robust since no switching occurs in the controlled variable. An essential part of the process control are the pre-defined material properties. Processing post-consumer recyclate (PCR) comes with strong differences in the processing conditions as the composition and history of the batches vary. In terms of the recycling economy, the central goals of the research project are for one, the realization of a constant molding quality irrespective of batch property variations and for another, the reduction of the required process set-up effort. In this context, changing material properties of the granulate require an adjustment of both the trajectory planning and the process control. The adaptation is to be automated by an extension of the existing cross-phase process control strategy. For this purpose, a PCR-oriented quality model is combined with a learning-based model-predictive cavity pressure control, so that overall high process stability is achieved. First, differences in PCR batches are quantified by testing laboratory analysis and injection molding experiments. Then, the identified relationship between material, process variables and part quality are mapped in a quality model. With the derived quality model, the influence of the recyclate can be explicitly considered in an inline quality control. To achieve a robust control performance with a minimum of adjustment effort, the controller model of the process controller is extended by a data-based error model. This error model is based on a Gaussian Process Regression, in which batch-dependent model deviations are learned online via process data. The effect of the learned model component on process stability is quantitatively assessed. Finally, the developed process control is to be integrated at the technical machine and iteratively improved. The evaluation and validation of the recyclate control are carried out on a complex component geometry with at least two PCR types. By the end of the project, it should be possible to reduce batch-related deviations in part quality by at least 50%. In the long term, the project should thus help to reduce rejects as well as material and energy consumption. In addition, the attractiveness of the use of PCR is to be increased in order to enable material recycling in the sense of the recycling economy.

• DFG Programme: Research Grants
• Co-Investigator: Professorin Dr.-Ing. Heike Vallery