Development of Digital Twin Models for the Mixing of Complex Rheology Fluids in Stirred Tanks

Johnson Matthey (JM) has a number of fluid mixing processes that are critical in the manufacture of a wide range of products which possess a complex microstructure and rheology. Whilst JM has a well-established expertise in experimental methods and Computational Fluid Dynamics (CFD) modelling, it is difficult to exploit these optimally in the scale up of new products and processes. This is due to the significant computational expense and run time for complex simulations. The objective of this project, therefore, is to employ an Industry 4.0 approach to derive simplified mixing models. These will possess a sufficiently low computational demand that they can be run in real time as “digital twins” for enhanced design and control. As JM is a UK-based FTSE 100 listed company leading in sustainability and sustainable technologies, these models are a vital component towards net zero aspirations for our own operations, and an important step towards digitisation of manufacturing processes. The simplified mixing models will employ reduced order, or zonal continuity models, that are simple and computationally small enough to be used more readily than CFD for model based control of manufacture. These will be used alongside advanced data analytical approaches, including Machine Learning (ML), to extract data and information from more complex CFD simulation results as well as from advanced experimental flow visualisation techniques available in JM and at the University of Birmingham. These analyses and ML models will be combined with key physical and scaling relationships to provide a predictive but largely data driven tool for the scale up and optimisation of mixing processes.