Transition to digital process development and biomanufacturing
Imagine if you could simulate entire experiments at the computer instead of performing time-consuming laboratory work. Simply set the process conditions at which the process should be performed, start the simulation and watch how the process responses you want to investigate (e.g., biomass and product titer) evolve over the process time. A powerful modeling tool, called digital process twin, enables this operation.
How does a digital process twin work?
Based on a suitable hybrid model structure, the application of a digital process twin is possible. Therefore, the prerequisite is that the utilized hybrid model is developed using controllable process parameters. Once this model is built, it already can be used as a digital twin. Herein, the value range of the controllable model inputs, and the step size in between, can be set, e.g., a cultivation temperature from 30°C to 37°C, in steps of 0.5°C, and a pH value from 5 to 9, in steps of 1. Based on the developed hybrid model, all possible process condition combinations are simulated, and the process responses are evaluated for each time point in each process. This results in a high number of time-resolved process simulations that are rapidly available at hand.
What can be done with this powerful modeling tool?
The digital twin tool already is implemented in our toolbox and enables
simple evaluation at any time of the process
examine ideal process conditions not only at the process endpoint but within
As a result of such a process simulation, a heatmap of the chosen process conditions and the response variables is derived. In the digital twin example below, derived from our hybrid modeling upstream case study, a heatmap of the biomass for all investigated cultivation temperature and growth rate settings, using one particular induction strength, is displayed. This feature enables easy detection of process optima.
Figure 1. Digital Twin of the Novasign hybrid modeling toolbox
The slider for the induction strength can be put to another value in the simulated range, which would modify the heatmap accordingly, allowing a systematic investigation of the entire simulation for the process responses of interest.
Moreover, each simulated process can be investigated separately by displaying your wanted process variables in a time-resolved manner for one particular process, i.e., from the start of the process until the end. Again, the process conditions can be set and the simulated process is plotted over the entire process time. In our toolbox example below, the trend of the biomass (turquoise) and the product titer (violet) are displayed for the entire process.
Figure 2. Bioprocess simulation in the Digital Twin
This visualization of the investigated process responses, via the digital process twin, as a heatmap for all simulated process conditions and for one particular process in a time-resolved manner, accelerates and simplifies optimization tasks and highly contributes to demanded process understanding.
The transition to digital process development and optimization is favorable in various aspects, e.g., in saving
and raw materials
A, by now, not covered topic is the long-lasting procedure to get a reliable and robust hybrid model in the first place for this powerful modeling application. We answer this subject by performing intensified DoE (iDoE) to reduce the experimental effort in the laboratory by more than 66%, once again accelerating time-critical process steps for our customers.