At the PAAT Annual Meeting, in Frankfurt/Main, Germany on November 20, 2023, INOSIM’s Head of Innovation, Christian Sonntag, will give a talk in the Digitalization II session. Below you find the abstract for the lecture with the title Digital Twins in Practice – Predictive Decision Support For A Batch Plant At Bayer, Wuppertal.
Additionally, INOSIM will be represented in the poster program of the event with a poster on A Multi-Fidelity Approach for Optimal Production Planning of an Industrial Formulation Plant with Intralogistics.
Digital Twins in Practice – Predictive Decision Support For A Batch Plant At Bayer, Wuppertal
Christian Sonntag (INOSIM), Andreas Schluck(BAYER), Dominik Wolff (INOSIM)
Industrial batch plants are complex technical systems in which – often simultaneously – dozens of different products are manufactured using complicated, sometimes dynamic production recipes. Hundreds of individual steps, a high degree of automation and dynamically changing framework conditions make it almost impossible for the operating personnel to make reliable predictions about the long-term behavior of the plant. As a result, plant operators lose a lot of potential in production, resource and staff planning.
In this lecture, we present a decision support system for industrial batch plants that enables highly accurate predictions of plant operation in real time by using digital twins. This allows production, resource utilization and operations personnel activities within the plant to be planned far more accurately and less conservatively than before, which can lead to large savings and efficiency gains. In addition, the effects of unplanned or unexpected events can be reliably predicted, so that negative consequences can be recognized and prevented before they occur.
The system was implemented at a Bayer batch plant in Wuppertal, among others, and is in productive operation there. The system automatically initializes the digital twin to the state of the real plant in real time using data from the IT/OT systems (MES and PCS systems) and generates a prediction of the plant behavior through dynamic simulation. The information relevant to various stakeholders is then generated from the simulation results and presented in graphical dashboards that can be displayed on monitors in the plant and on mobile devices. Depending on the objective, different visualizations are provided, e.g. a “Departure Board” that gives the operating staff an accurate prediction of the manual work steps required in the near future, predictions of the time windows for the (decentralized) maintenance of plant components, visualizations of the effects of malfunctions or predictive KPI cockpits.