Project Details
Projekt
Model-based control of spray synthesis of structured granules with specified properties, using transfer functions derived by multivariate stochastic models and machine learning
Subject Area
Mechanical Process Engineering
Term
since 2022
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 504580586
This project focuses on a spray drying process chain, where a carrier liquid, containing dispersed particles, is atomized into droplets, which dry during the residence time in the dryer, leading to solidification. The inner morphology of these granules determines their eventual application in various fields. Here, porosity and nano-pore structure play a central role, e.g., for adsorptive or catalytic behavior. Thus, product design works on tailoring the inner morphology of granules. The latter depends on both, the properties of the feed, e.g. the primary particles, their interactions and the concentration, and the process conditions, e.g., the temperature field, droplet size, drying conditions. To control the entire process, all these parameters have to be considered, leading to a multi-parametric problem. In-line measurements are a valuable tool for process monitoring. However, in the present case, such measurements cannot directly assess the nano-pore structure of the granules and, consequently, the final product specifications. Off-line measurements like Hg-porosimetry and 3D imaging methods like high-resolution computer tomography (CT) are required to obtain full information on the pore structure. This is time consuming and undergoes several steps from granule sampling, sample preparation, and imaging, to image segmentation and data analysis. To be able to track pore descriptor distributions during a running process (e.g. for using them in an autonomous control scheme), the data that can be collected in-line must be linked to these distributions, making them so-called proxies. In the 2nd funding period (FP), data-driven proxies (e.g. compressed image data of granule systems obtained by automated encoding) will be used. The forward and backward regression of process parameters and feed properties with interpretable and data-driven proxies is achieved by methods of machine learning, e.g., feed-forward networks. In the 1st FP, the relationship between droplet formation and final dried product morphologies was addressed, which will be extended in the 2nd FP by the pre-conditioning of the feed composition, i.e., in-line modifications of the feed concentration and viscosity. In this way, the project generates scientific insights on the performance and applicability of a data-driven control scheme, using in-line and off-line information. The control scheme is experimentally validated to demonstrate that granules with desired nano-pore structure can be produced even under fluctuating process conditions. Additionally, it accounts for costs associated with process parameters, making it particularly suitable in times of fluctuating energy prices.
DFG Programme
Priority Programmes
