Lead Engineer Altair Engineering Edinburgh, Scotland, United Kingdom
The rapid and cost-effective design of reliable manufacturing processes is key to maximizing the profitability of pharmaceutical products, a large proportion of which are in the Oral Solid Dose (OSD) form. OSD manufacturing processes pose significant challenges because of the complex nature of the particulate materials they manipulate. The mechanics of these processes is not fully understood leading to overreliance on expensive and time-consuming physical trials in the process development stage where delays can be costly due to time limits on patents and competitive market conditions.
Simulation-based virtual process design and optimization can reduce development timelines and costs and has become an important component of the digital transformation strategy in the pharmaceutical industry. High-fidelity Discrete Element Method (DEM) simulations that can resolve the mechanics of particulate materials at the particle scale have already been used to generate new and deeper understanding of process mechanics, improve process reliability, and reduce time to market for new formulations. Machine learning can increase the befits of DEM simulations by reducing computational expense and maximizing learning.
In this work we present an efficient virtual optimization methodology for OSD manufacturing processes which combines DEM modelling, design of experiments, machine learning and optimization methods. A key feature of the methodology is the generation of a machine learning based process digital twin using synthetic data from a statistically efficient set of DEM simulations. This computationally efficient digital twin is then used in conjunction with a gradient descend optimization algorithm to identify the globally optimum operational parameter set in seconds rather than the weeks required by the equivalent purely simulation-based approach.
The validity of the methodology is evaluated through the optimization of two common mixing systems in OSD manufacturing - a continuous mixer and a bin blender. The focus of the optimization is on improving the mixing efficiency by optimizing operational parameters. The accuracy of the predicted optimal operational parameter set for both systems is evaluated in silico and the advantages and limitations of the methodology are discussed
Learning Objectives:
Understand how to perform high-fidelity physics-based simulations of OSD manufacturing processes using the Discrete Element Method (DEM) and how to generate data via virtual trials.
Understand how to develop accurate machine learning based Digital Twins of OSD manufacturing processes from DEM simulation-derived synthetic data.
Understand how to rapidly optimize process operational parameters using a machine learning based Digital Twin in conjunction with genetic and gradient descend optimization algorithms.
