Project description
In order to remain competitive, food producers are responding to rapidly changing market requirements with a growing variety of products and short product life cycles. This necessitates the ability to develop and commission new production systems quickly. However, the increasing complexity of production plants, the growing use of automation and information technologies, and the shortage of skilled workers are making plant development more challenging. This applies in particular to the development of automation solutions. Therefore, new approaches and technologies are required for control system development as well as suitable testing and analysis tools for their validation.
Against this background, the virtASI research project is rethinking the engineering of food process engineering production plants and placing innovative methods and technologies at the center of plant development. The central object of the project is to generate both modular simulation models and the corresponding control algorithms simultaneously and automatically for the first time. Based on the standardized, modular planning data stored in the administration shell, the system automates itself with the help of AI methods. The generated automation solution can then be connected to the automatically generated simulation model in the sense of software and hardware-in-the-loop, allowing the system to be put into virtual operation quickly and easily.
Consortium
In addition to the working group Smart Production Systems at the Chair of Brewing and Beverage Technology, which acts as project manager, innovative technology companies are involved in the virtASI project.
The project partner SimPlan is a leading service provider for simulation and virtual commissioning (VIBN, emulation) and is active in the fields of production and logistics. Various third-party tools are used, e.g. Plant Simulation from Siemens, AnyLogic from the AnyLogic Company or Emulate3D from Rockwell Automation, as well as proprietary software packages such as SimVSM, jasima and PacSi. SimPlan has extensive know-how in the application and development of simulators/emulators and simulations/emulations. SimPlan also has expertise in linking digital planning tools such as simulators with other operational IT systems and integrating these tools into IT platforms. The foundations and preliminary developments for some of the software solutions (SimVSM, SimAssist) sold by SimPlan on the market today have been laid in joint research projects over the last ten years.
As a highly specialized technology company, Gimbio develops virtual expert systems that optimize production processes and measuring systems with ultrasound technology. These systems are used successfully in breweries and companies in the liquid food industry worldwide. In addition, Gimbio develops basic automation components and development tools for use in small and microbreweries.
Ziemann Holvrieka is a global leader in brewery and process solutions. These include tanks and process technology for the brewing, beverage, food, chemical and pharmaceutical industries, which are developed and produced at the German sites in Ludwigsburg (Baden-Württemberg) and Bürgstadt (Bavaria), among others. With the introduction of the Modular Type Package and a digital, cloud-based engineering platform, Ziemann Holvrieka is a pioneer of digitalization in the food industry.
Project schedule
The research project will run for three years. It begins with an analysis of existing solutions and standards, a definition of suitable use cases, and the development of a solution approach. Based on this, a modular, standardized food processing plant will be modeled - from a mechanical, electrical, automation, and process engineering perspective. The increasingly established Modular Type Package (MTP), which is stored together with other engineering data in the administration shell of Industry 4.0, the so-called Asset Administration Shell, will be the central object of consideration. This data is then used for the (semi-) automated generation of a modular simulation model and a control solution. AI methods, particularly large language models, are evaluated regarding their suitability for analyzing the available data and generating models and code. Based on this, a concept for merging the module solutions into a system network will be developed and the solution approach will be systematically validated on industry-related systems. The results will serve as a basis for further development work and will be transferred to science and practice.
