The fourth and final full scale demo trial of the PREVIEW system was performed recently at Smithers Rapra Polymer Processing and Development Centre (PPDC), at their site in Shawbury, UK. Two of the partners in this European project (Horizon 2020), Plastia (Spain) and Humboldt-Universität zu Berlin (Germany) travelled to Smithers Rapra’s facilities to continue the implementation and testing of the PREVIEW technology. The partners worked collaboratively to evaluate the overall performance of the system in a real case and industrially-driven scenario (Figure 1).
PREVIEW’s four components:
• Data Acquisition System (DAS)
• Wireless Communication Nodes (WCN)
• Advanced Prediction System (APS)
• Location Based Content Delivery (LBCD)
worked together to build the Cyber Physical System (CPS).
The fundamental goal of PREVIEW is to reduce waste, and improve productivity through the smart monitoring, control and optimisation of the injection moulding process. The DAS records cavity and machine signals (temperature, pressure), and they are processed and transferred via wireless communication to a central wireless node that collects the data. This node acts as a centralised server that works in conjunction with the advanced predictive system (APS) which analyses the data, determines process inconsistencies and recommends optimum production parameters that are delivered to the user through a mobile application named the location based content delivery (LBCD).
In this demo trial, all of the PREVIEW subsystems were set up on the first day. The DAS was connected to both: the ARBURG 420C injection moulding machine, and the sensorised mould to read the signals. The DAS interacted properly with the machine and sensor interfaces, timely receiving cavity and machine signals (temperature, pressure). The rate of the data transfer process from the DAS to the WCN was seen to function efficiently, and the main information of the signals was seen to arrive at the node within the first few seconds.
The reach of the wireless network was evaluated by installing an additional node positioned at least 200 meters away from wireless server in the opposite extreme of the production floor. Simulated data incoming from this node was adequately received and processed by the central server node. It was possible to confirm that both WCNs and the server created a robust wireless network (Figure 2) that benefitted from the customised communication protocol to prioritise information and communicate effectively with the APS.
The optimum parameters of the injection moulding process were established and a Design of Experiment (DoE) was designed based on the following parameters: injection speed (cm3/s), injection pressure (bar) and holding pressure (bar). This was done to identify any deviations that could pose a risk in part quality to feedback into the APS system.
On the second day, the connection between the DAS-WCNs-CMS-APS was checked again to ensure the whole PREVIEW system loop was working smoothly. The DoE was conducted, and throughout these runs the DAS was able to acquire and display digital data coming from the machine and cavity signals that projected the corresponding plot deviations resulting from the changes of the injection speed, injection pressure, and holding pressure.
In Figures 3 and 4 two examples are presented of the observed changes in injection speed and holding pressure, respectively, as obtained from the data captured by the DAS. These results prove the adaptability and capability of the DAS to accurately monitor the changes in injection moulding parameters. The quality of the produced parts, in addition to any production defect (e.g. flash, flow mark, shrinkage etc.) was monitored and visually inspected by the operator. Part dimensionality and weight were also measured for the purpose of quality control.
On the third day, after completion of the training process for establishing the production parameters in the APS, this subsystem was evaluated and it was corroborated it could effectively monitor the injection moulding process, providing part quality classification and recommendations to secure optimum set-up. All this information was easily retrieved remotely by the user through the location based content delivery mobile application. In addition to classifying the quality of parts (e.g. good/bad), the APS was capable of assisting the user in identifying the type of defect, which was seen to greatly improve the quality control process by guiding the operator into what to specifically look for.
The whole third day was focused on running the system autonomously, e.g. as in full scale production trial, and monitoring the stability, smoothness and performance of both: the individual subsystems, and the PREVIEW system as a whole. The overall observation from the partners was that PREVIEW can work smoothly in an industrial production scale environment. Yet another successful trial for the PREVIEW Project team after spending more than two and a half years on the development of an inclusive, innovative and smart technological system.