ARCO-CHAIN: Autonomous Robust Predictive Control of a Continuous Crystal Process Chain

Crystallization is a widely used technology to produce particulate materials with a defined particle size distribution (PSD), crystal form and purity. In the pharmaceutical and fine chemical industries, batch operation remains the standard for industrial crystallization, especially at small production scales; more than 90% of active pharmaceutical ingredients and many fine chemicals are crystalline organic compounds produced at rates of only 250–1000 kg/year. Conventional batch processes suffer from high variability and inconsistent product quality, making them inefficient and costly. Continuous crystallization processes such as slug-flow crystallization promise lower investment and operating costs, better controllability and higher flexibility, but they raise major challenges in modelling, sensing, actuation and adaptive control.

In the first phase of the project (ARCO-CRYSTAL), a slug flow crystallizer (SFC) was equipped with sensors and actuators enabling online control of a defined PSD; static and dynamic models were developed and robust model predictive control approaches were formulated to achieve autonomous operation of the SFC. In its second phase, ARCO-CHAIN extends autonomous operation to the full continuous crystal process chain (CPC) by connecting the SFC to a continuous vacuum screw filter (CVSF), which combines filtration, washing and drying in a single apparatus.

The central goal of ARCO-CHAIN is to achieve autonomous process control of a fully continuous CPC, delivering high-quality products with minimum resource and energy consumption and maximum yield. The project pursues four objectives:

1. Integration of sensors and actuators that enable control of the CVSF;
2. Development of detailed models of the CVSF together with linking models that transfer outputs from one unit to the next (including the SFC), as well as surrogate models that respect basic physical balances;
3. Development of a model predictive control algorithm that explicitly accounts for uncertainties and inter-unit couplings; and
4. Demonstration of autonomous operation of the continuous CPC.

The main outcome will be an adaptive and robust process control strategy for a continuous CPC at laboratory scale, capable of coping with uncertainties and disturbances during autonomous operation.

ARCO-CHAIN is the second phase of the ARCO-CRYSTAL project and is funded by the DFG within Priority Programme SPP 2364 "Autonomous Processes in Particle Technology" (project number 504676854). It is carried out jointly with PD Dr.-Ing. Kerstin Wohlgemuth (TU Dortmund).