By Dan Hebert, Senior Technical Editor
The pharmaceutical industry has often lagged other process industries in terms of advanced process control, but in terms of first-principle models the industry has made many recent advances. Because the underlying first-principle modeling technology is applicable to batch and continuous processes in other process industries, reviewing recent pharmaceutical industry advances can help automation professionals in other industries improve their process control systems.
With the help of Process Automation Hall- of-Famer Terry Blevins, Ive outlined how.
One reason pharmaceutical companies have lagged in adopting new control technology is the regulatory regime. Realizing the impact this delay has on manufacturing innovation and efficiency, the U.S. Food and Drug Administration published Guidance for Industry, PAT (Process Analytical Technology)A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance.
The PAT framework addresses the use of multivariate data analysis, process endpoint monitoring and knowledge management tools for process modeling; thus sparking renewed interest in these tools and their application to batch processes.
The most commonly applied multivariate analysis tools are principal component analysis (PCA) for fault detection and projection of latent structures (PLS) for the prediction of key quality parameters at end of batch. However, recent technical advancements in PCA and PLS permit the on-line use of these techniques.
In the development of the underlying data-driven PCA and PLS models, dynamic time warping can be applied to compensate for variations in the batch time. Dynamic time warping is a technique for alignment of batch data to compensate for batch holdups and delays in batch processing.
A new technique for processing batch data called hybrid unfolding gives better PCA models for on-line fault detection and improves on-line fault detection and end-of-batch prediction of quality parameters.
These analytic models provide some process insight, but their use is limited for on-line process control because most batch processes are non-linear.
First-principle models may be the answer, using experimental data instead of statistical methods to estimate model parameters. They are not as quick and easy to build, but they have many advantages. In terms of simulation, first-principle models provide extrapolation in addition to the interpolation provided by data-driven models. But they also can be used for monitoring, control and optimization.
Cost and the time and skill required to develop an application-specific model have been barriers to using first-principle modeling tools. Integrating the resulting model into the plant control system to evaluate the impact of a change in control strategy is also difficult.
Embedding application-specific models into the control system addresses many of these issues. For example, pharmaceutical companies are using fungal, bacterial and mammalian cell bioreactor models to support process and control studies.
Design of experiments can minimize the time and tests required to identify model parameters. Once a process model is developed, then this model may be used to explore the best approach to stepping process inputs or to changing operating conditions to improve batch cycle time and product quality.
Details on these techniques are highlighted in the book, New Directions in Bioprocess Modeling and Control: Maximizing Process Analytical Technology Benefits by Michael A. Boudreau and Gregory K. McMillan, ISA, 2007.
If you are exploring ways to improve your process plant operations, these pharmaceutical industry advances are worth considering.