In terms of digitalization, the biotech industries are quite significantly behind others, like the automotive or aeronautical industries, possibly because the digital tools available are not very well suited to their needs.
Roger-Marc explains how numerical tools can support process development by minimizing the time required and the experimental effort required.
To this end, he presents and distinguishes two main approaches for process modeling: the statistical approach and the mechanistic approach.
The statistical approach relies on representing as well as possible the experimental datasets available with mathematical functions.
This type of model can predict the behavior of the process within the experimental boundaries investigated.
In contrast, the mechanistic approach, based on first principles, relies on understanding the physical and chemical phenomena ruling the system and on building a model that represents these phenomena. A few experiments allow determining a limited number of parameters. Such a model can predict the process behavior even outside the “experimental space”.
The selection of one approach or the other (or both or none!) will depend on: -
- the actual need: scaling-up, optimization, …
- the knowledge available: simple reaction stoichiometry, separation principles…
- the data available: production data only, well-thought experiments…
Roger-Marc illustrates the advantages and limitations of each approach with a few simple examples, emphasizing the experimental burden associated with each.
Do not hesitate to contact us for any inquiry : contact@ypso-facto.com