Agunloye, Emmanuel; Petsagkourakis, Panagiotis; Yusuf, Muhammad; Labes, Ricardo; Chamberlain, Thomas; Muller, Frans; Bourne, Richard; Agunloye, Emmanuel; Petsagkourakis, Panagiotis; Yusuf, Muhammad; Labes, Ricardo; Chamberlain, Thomas; Muller, Frans; Bourne, Richard; Galvanin, Federico; - view fewer (2023) Application of MBDoE techniques to a cloud-based platform for automated chemical manufacturing in flow reactor systems. Presented at: 2023 Workshop on Model Based Design of Experiments, London, UK.

Preview

Text Agunloye et al CPSE MBDoE Workshop 2023.pdf - Published Version Download (831kB) | Preview

Abstract

Industry 4.0 has birthed a new era for the chemical manufacturing sector, transforming reactor design and automating process control. Towards autonomous chemistry development, on-demand manufacturing, and real-time optimization, we developed a cloud-based platform driven by model-based design of experiment (MBDoE) to coordinate remotely smart flow reactors, also known as “LabBots”, sited in different locations. A cloud-based iterative MBDoE framework was proposed characterised by five elemental stages: i) formulation of candidate models, ii) preliminary design of experiments (DoE), iii) parameter estimation using the experimental data acquired from designed experiments, iv) MBDoE for parameter precision, and v) model validation. The framework has been applied to two pharmaceutically relevant case studies: 1) homogeneous amide formation from amine and 2) nucleophilic aromatic substitution of 2,4-difluoronitrobenzene with morpholine. The first case study involves a single-forward reaction step and yields products that can also react in the reverse direction, presenting two alternative mechanisms. After employing rate expressions to model the two mechanisms and then estimating their Arrhenius parameters through a preliminary DoE-driven experimental investigation, the MBDoE framework was used to analyse the relative model performance. While rejecting the single-forward model, a χ^2 lack-of-fit test accepted the single-reversible model as the best model for the amide formation. With one single additional experiment designed using MBDoE for parameter precision, the numerical values of the reversible model parameters were estimated with statistically satisfactory confidence intervals. In the model validation stage, the reversible model performed well in describing new data from experiments designed by a full factorial design of experiments. The second case study (nucleophilic aromatic substitution) is characterised by a unique structure but involves a more complex reaction mechanism. Analogous to the amide formation, the MBDoE framework employed rate equations to describe the parallel and consecutive steps and identified a mechanistic kinetic model with the minimum number of runs. Following MBDoE for parameter precision, the resulting model predictions improved in terms of uncertainty, particularly in unexplored regions of the experimental design space.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item