Real–Time Data on Manufacturing Bio–Processes with NIR spectroscopy

Related Vendors

Frewitt fabrique de machines SA

Sartorius AG

Batch Trajectories – Processes at a Glance

Alongside individual parameters, NIR spectroscopy also displays the status of a fermentation in terms of a positive/negative assessment. This multivariate statistical process control (MSPC) is gaining ground in the biopharmaceutical industry. This approach does not require any laboratory analysis for calibration; instead, it is solely based on the variances within the spectra which react sensitively to changes in the overall process. This allows deviations from the desired (optimal) process behavior to be recognized early on and to be displayed in graphs which are easy to interpret.

A common method is to display process behavior as trajectories through the Batch Evaluation Model (BEM). The main variances that occur within the spectra, which are expressed through the main components of a principal component analysis (PCA), are displayed as trajectories over time. As a result, those overseeing the process have a constant overview of proceedings and can react to undesired changes instantaneously.

Process Industry Trends

10 Trends for the Process Industry You Should Know

Real–Time Data on Manufacturing Processes

Figure 2, shows the temporal progression of the first main components. These trajectories are taken from three batches which showed a particularly good performance. The mean from these optimal processes is called a ‘golden batch’ (green dotted line in Figure 2) and future processes will be compared to this batch. Alarm criteria would then be, for example, three times the standard deviation of the processes which have been used to define the model (red dotted line). These lines show the permitted range of fluctuation (design space) of processes without taking critical quality attributes into consideration. As long as a new batch remains within these boundaries, it will be considered statistically identical to the golden batch and it is presumed that the process has run optimally until this point. Should readings occur outside of these boundaries, a real-time alarm will be sounded so that those in charge of the process can investigate the cause of this deviation from the optimal behavior.

Figure 3 displays the model processes of two CHO cell cultivations, the behavior of which started to deviate from the expected pattern after a certain period of time due to contamination (blue process). In addition, a modified aeration rate led to a slowdown of cell growth (black curve).

* First published in PROCESS India. The authors work for Sartorius Stedim Biotech

(ID:42525331)