Today, the planning of industrial production is primarily driven by the goal of optimal utilization of production facilities, taking into account the time constraints for individual orders. Considerations for optimized energy usage play only a minor role, if at all, e.g. in avoiding peak loads.
However, the ambitious goal of changing Austria’s energy supply to 100% renewables (#mission 2030) changes the structure of energy production considerably. With this new direction, electricity as energy source gains emphasis. Many areas that currently still operate on fossil energy are gradually changed to electricity. This ranges from heating systems to process heat generation to mobility. A higher electricity demand together with the inevitable volatility of renewable energy means that production planning must take into account the availability of energy in order to achieve the primary goals of #mission 2030. With the expansion of renewable energy sources, the baseload share of electrical power plants decreases, and energy production becomes increasingly volatile. Under these circumstances, a fixed load profile is often no longer realistic; instead, the load should be adapted to the available energy.
The primary objective and innovation of the project is to enable production plants to plan their manufacturing steps according to a predetermined yet dynamic energy availability in such a way that a given energy profile is maintained by more than 95%. This means that the given energy it is not only not exceeded but also used as fully as possible while at the same time ensuring efficient utilization of manufacturing resources. Production, therefore, becomes a plannable (electric) load for the power grid. This constitutes an essential innovation for both production planning and energy grids, allowing to guarantee reliability and safety of the processes and to prevent negative effects up to possible instability of the energy system. The big challenge is in the handling of the inherent prediction uncertainty of the energy profiles, which increases with increasing planning horizon and production complexity.
The goal is achieved by an iterative planning process, in which production planning is executed for energy profiles with different probabilities. The results are assessed and evaluated with respect to their compliance with the set of production constraints, and the best plan is successively adapted to the prediction uncertainty that is changing over time. Adaptive energy consumption models that are optimized using machine learning techniques increase the planning accuracy, and the inclusion of production-dependent energy storage and recuperation potentials improves the overall efficiency. The concept presented in the project will enable energy-aware production planning for individual production lines, however it will be scalable to allow multiple independent processes to be jointly adapted and optimized for an overarching energy profile.
Within the scope of the project, a simulation-based analysis of the system will be carried out in a first step, complemented by an industrial laboratory test in the planned battery production facility of AVL. The results have therefore highly validity for the entire manufacturing industry. A further generalization to other sectors (e.g., food production, logistics) will also be continuously examined.
The project is part of the CELTIC-NEXT project IEoT (Intelligent Edge of Things) where it constitutes a central use case.
