Skip to main content

Project - Data science

Cell Culture: Biological Modelling and Link with Monitoring Flows

The challenge

The scale-up of processes, i.e., their scaling up to a larger size, always requires some thought, especially when they are bioprocesses: living beings have very non-linear behaviour. Understanding the influence of environmental variables is necessary to assess the impact of a change within biology.

In this case, mammalian cells are cultured to produce antibodies for use in a specific treatment. Scale-up to a larger bioreactor requires some physical tests for oxygen transfer, but it is difficult to assess the impact on cell performance.

Another difficulty is the data available:

  • When establishing a culture recipe in the pharmaceutical field, the operating conditions are fixed, which implies that the batches are mostly the same. The scale-up will surely change this operating point (different bioreactor, control devices, etc.) so it is tricky to assess any impact on process performance;
  • The sampling rate is high for environmental variables such as oxygen levels, but it is especially low for biology: 1 measurement per day.

The solution

The current operating point has been modelled with differential equations representing changes in variables (biomass, antibodies, oxygen supply, kLa). Consequently, the evolution of biology is linked to measures with high sampling rates and, as such, can be described with the same kind of dynamics as these measures. The model becomes an online software sensor describing what happens in the bioreactor.

To evaluate the impact of scale-up on process performance, the next step may be to scale-down the process into small (inexpensive) batches exploring different operating conditions. The impact of the change in operating point on biology can then be assessed by interpolating the known and modelled operating points. Therefore, the result is a decision support tool that can help in batch management (guaranteed performance or not? Start a new batch or not in case of a problem?).

The potential gain

The software sensor becomes a monitoring tool that can track deviations from defined operating points and send alarms to the culture operators to save the batch. The decision support tool saves production time: if the batch has deviated for a few hours, is it worth continuing with the batch, or is it better to start a new one and save production time?

Again, operational gain can easily translate into financial gain.

Project Details

Client Activity

  • Pharmaceutical industry
  • Antibody Production

Solution Brand

  • Wonderware
  • Historian
  • Intouch
  • Microsoft
  • Excel
  • Matlab

Nature of the work carried out

  • Golden batch



  • Data Science Offline et Optimisation
  • Software sensors

Other references More to be discovered

Let's stay in touch

Subscribe to our newsletter