Scientific Deep Machine Learning Concepts for the Prediction of Concentration Profiles and Chemical Reaction Kinetics: Consideration of Reaction Conditions.

Emerging concepts from scientific deep machine learning such as physics-informed neural networks (PINNs) enable a data-driven approach for the study of complex kinetic problems. We present an extended framework that combines the advantages of PINNs with the detailed consideration of experimental parameter variations for the simulation and prediction of chemical reaction kinetics. The approach is based on truncated Taylor series expansions for the underlying fundamental equations, whereby the external variations can be interpreted as perturbations of the kinetic parameters. Accordingly, our method allows for an efficient consideration of experimental parameter settings and their influence on the concentration profiles and reaction kinetics. A particular advantage of our approach, in addition to the consideration of univariate and multivariate parameter variations, is the robust model-based exploration of the parameter space to determine optimal reaction conditions in combination with advanced reaction insights. The benefits of this concept are demonstrated for higher-order chemical reactions including catalytic and oscillatory systems in combination with small amounts of training data. All predicted values show a high level of accuracy, demonstrating the broad applicability and flexibility of our approach.

A linear scale-up approach to fluid bed granulation.

Fluid bed granulation (FBG) is used extensively in the pharmaceutical industry and it is known to be a complex process, because the final product quality of the FBG process is determined by a complex interplay between the process parameters, fluid dynamics, and material properties. Due to this complexity, the FBG process is inherently nonlinear and as such difficult to scale-up. The field of chemical engineering has shown that complex nonlinear processes can be assumed to be linear under limiting conditions. We leverage this idea and present a linear scale-up approach (LiSA) to the FBG process. We derive the key LiSA equation from first principles, and then use it in combination with the similarity principle for scale-up purposes. Furthermore, we present a novel regression-based LiSA. The regression-based LiSA is founded on the hypothesis that there is a linear relationship between the moisture content and a scaling parameter called the Maus factor. This hypothesis is based on our experience and it is shown to be plausible due to high R2 values ranging from 0.86 to 0.98. Moreover, we successfully demonstrate that LiSA is effective under typical industrial process settings by applying it to two different formulations during pharmaceutical drug product development.

A systematic reactor design approach for the synthesis of active pharmaceutical ingredients.

Today's highly competitive pharmaceutical industry is in dire need of an accelerated transition from the drug development phase to the drug production phase. At the heart of this transition are chemical reactors that facilitate the synthesis of active pharmaceutical ingredients (APIs) and whose design can affect subsequent processing steps. Inspired by this challenge, we present a model-based approach for systematic reactor design. The proposed concept is based on the elementary process functions (EPF) methodology to select an optimal reactor configuration from existing state-of-the-art reactor types or can possibly lead to the design of novel reactors. As a conceptual study, this work summarizes the essential steps in adapting the EPF approach to optimal reactor design problems in the field of API syntheses. Practically, the nucleophilic aromatic substitution of 2,4-difluoronitrobenzene was analyzed as a case study of pharmaceutical relevance. Here, a small-scale tubular coil reactor with controlled heating was identified as the optimal set-up reducing the residence time by 33% in comparison to literature values.