LAUSR

Abstract The mass production of metal molds for glass bottles is a highly automated and technologically demanding sector of activity. Mold manufacturing involves several operations and one of the most critical is the one related to mold surface welding. The main goal here is to insert a more durable metal into specific areas of the mold to create regions with enhanced characteristics (e.g. improved wear resistance) in order to improve the quality and durability of the mold. Such an operation typically involves three steps, namely: i) mold preheating, ii) manipulation and iii) welding. Overall, this is a rather complex process, demanding tight control and monitoring of its variables. Our industrial partner, a medium-size manufacturer of molds for glass bottles, already had an extensively automated manufacturing process. However, there were frequent reports of processing flaws during the welding stage (e.g. quality of metal deposition, fissures, welding pores) with the mold being lost in some cases. The reasons behind most of the reported flaws were difficult to discern from the available sensors and methods. From our early assessment it became clear that the range of sensors available during the welding operation was insufficient and that the communication between the machines inside the production cell was either too simple or disconnected from the rest of the factory. This resulted in an absence of an integrated record of the production process. In view of this, it was necessary to develop and implement a digitalization plan, which we describe in this paper, of the production cell responsible for mold surface welding. As a result of this planning, additional sensors were introduced to the production cell, related to pressure, temperature, humidity, flow, and voltage. The intent here was to have monitoring variables that could be used to diagnose and/or prognosticate problems in the process. Also, since the sensors lacked integration, a network was established using IoT technologies supported by the PlugThings Framework together with Universal Plug and Play (UPnP) technology. The PlugThings, an IoT platform developed by our partner iTime, acted as a mediator between the sensor network and a Web server. The IoT platform also allowed to perform the vertical integration of the process, from sensors on the production cell to the Enterprise Resource Planning (ERP) system. Finally, this initiative enabled access to charting tools and to data in the form of XML files. Currently, the system in undergoing exhaustive tests before expanding to other production cells in the shop floor. Future steps will involve the use of analytics and artificial intelligence to diagnose and predict failures of the manufacturing process and to better streamline production.