At the beginning of the project the IREAS solution architecture was defined, and a roadmap was structured for development throughout the project along with an IP strategy and a dissemination plan. Partnerships with DTE’s customers and partners were signed for assistance in developing the platform, including requirement analysis, feedback on features and user testing. The platform’s first version was built on Pheonix LiveView and incorporated white-labeled Dundas BI for Dashboards and visualizations, with the central DTE tenant hosted in Amazon cloud services and the platform being cloud agnostic. The platform is based on a BI schema and an ETL from the replicated DTE analyzer’s database that is done every 1 minute. The data delivery is done with pg-logical, a Postgres data replication tool that is highly effective for real-time data flow and handles lossy connectivity well. To further support connectivity and data delivery to the platform, edge devices are hosted at customer sites to aggregate and forward both device and other plant data to the cloud. The edge device further enables setup of the platform on customer premises for those who request that. After the platform infrastructure had been set up and basic process visualization for DTE device results implemented and validated with customers the development of the metal production optimization apps was started. To construct the models for the electrolysis pot maintenance system an AI&ML contractor was hired to create models that predict the likelihood of failure of the electrolysis cells within different timeframes. An Aluminum alloying optimization system was next developed with a customer along with the implementation of four fully automatic analyzers from DTE. The system assists the customer in directing crucibles to furnaces and optimize the furnace batch. The last metal app produced was the electrolysis pot tapping optimization system, to automate the process to arrange pots to crucibles in the pot tapping process to meet the casthouse production schedule requirements. The metal apps have been implemented and tested in production environment and their value generated validated by customers. At last, in the development of the platform an integration between the analyzers and platform was established to calibrate and manage the devices out in the field. This is to ensure precision and accuracy of the analyzers that support usage of multiple calibrations for different alloys and use cases. During the project period DTE applied for three patents for the molten metal analysis technology to protect the proprietary technology that the IREAS solution is built on. As a part of the dissemination plan DTE attended over 15 conferences during the project time, published six technical papers and published content on social media channels such as LinkedIn. DTE has experienced a strong interest and demand for the solution and the project has enabled DTE to enhance the IREAS solution along with growing the presence of DTE in the industry with the mission to provide a solution to drive improvements, optimize process control, improve safety and increase sustainability in the metals industry.
