The SteamDry project has reached a key milestone: the pilot plant is now fully operational. Since early May, initial trials have been underway, marking the transition from theoretical development to real-world experimentation. At the center of this phase, VDEh-Betriebsforschungsinstitut (BFI) is leading the effort to understand, model, and optimise the drying process under realistic industrial conditions.
With the pilot plant running, BFI has begun a series of controlled experiments designed to explore how the system behaves. By applying step changes in parameters such as temperature, researchers observe how the plant responds dynamically. Throughout these trials, a wide range of data is recorded, including temperature, air velocity, moisture levels, and web speed. This data is essential for building accurate representations of the process.
Using these measurements, BFI has developed initial dynamic models of the plant through advanced system identification techniques. Because the drying process is inherently non-linear, multiple models are created across different operating points and combined into a nominal model with defined uncertainty ranges. These models form the backbone for the next step: designing an intelligent and efficient control system.
A key objective is to control moisture in a way that is both energy- and time-efficient. One of the main challenges lies in optimally distributing the drying load between the plant’s two dryers. To address this, BFI applies optimisation strategies that determine how best to split the drying effort, ensuring efficient use of resources.
The development process is highly iterative. Experiments inform the models, models guide controller design, and the resulting performance is validated through further testing. When uncertainties or gaps are identified, new experiments are carried out to refine the system. This rapid cycle allows the team to continuously improve results while gaining deeper insight into the process.
At the same time, BFI is building a multi-layer control architecture that combines performance with safety. At the base level, conventional PID controllers regulate core variables such as temperature, airflow, and web speed. Above this sits a robust process controller designed to handle disturbances and model uncertainties. On top of these safety layers, an advanced AI-based learning system will be introduced, enabling the plant to adapt and improve over time while operating within safe boundaries.
Looking ahead, the solutions developed at the pilot plant are intended for real industrial application. The control strategies can be transferred and adapted to full-scale facilities, with future developments aiming to incorporate learning capabilities that continuously optimise performance.
With the pilot plant now in operation, SteamDry has entered one of its most dynamic phases, where experimentation, modelling, and intelligent control come together to shape the future of industrial drying.