(117f) Dynamic Flowsheet Simulation and Application of Soft Sensors on an Intensified and Integrated Purification Step for Pharmaceutical Upstream Manufacturing

Recent disruptions in the global supply chain of critical medicines, active pharmaceutical ingredients (APIs), and key synthesis precursors to APIs have emphasized the need for process modernization and intensification in the pharmaceutical industry. Moreover, manufacturing processes must adapt to newfound global demand for critical medicines by considering higher throughput alternatives such as pivoting from batch to continuous manufacturing. This paradigm shift has raised the need for the integration of process knowledge, quality control strategy, and real-time process management. In this regard, model-based digital design has gained an important role as a platform to simulate the dynamics of actual plant operation, while simultaneously enabling the design of robust processes and identification of process operating parameters under conflicting design objectives.¹

Considering that more than 90% of the currently produced APIs are in crystalline form², the drug purification and isolation is a critical bottleneck, while achieving end-to-end continuous manufacturing (CM). Drug purification and isolation is carried out with a sequence of crystallization, filtration, washing, and drying operations. These operations not only serve as critical steps to pass stringent regulatory screening on the critical quality attributes (CQAs) of the solid API, but also on the purity requirements related to the residual material after the drug crystal form is recovered. Since impurities remaining in the solid at the end of the drying step will be directed to the downstream process, achieving high purification while conserving crystal quality during filtration, washing, and drying will yield fewer off-spec products in the final stage of a drug manufacturing. While crystallization has been already established in terms of both experimental and modeling perspectives³, the overall integration of the drug purification section has been explored only marginally and several challenges need to be addressed further.⁴

Understanding the interactions between process parameters from different unit steps is critical. However, streamlining crystal slurry purification makes it difficult to operate due to its propagation of error arising from the interconnectivity of consecutive operations. To systematically analyze optimal process design for the desired CQAs of a product on CM, model-based process design has gained importance as a way to simulate process dynamics without incurring the usage of expensive API and lengthy experimental efforts. As a tool to simulate and identify the critical process parameters (CPPs) for developing continuous manufacturing system, creating a digital twin of a given physical platform has been identified as strong alternative to repeating various open-loop experiments.⁵

In this work, we develop a digital twin of a novel continuous filtration carousel (CFC) unit, where a drying component has been recently added, as an intensified platform to perform solid-liquid-separation, integrated with a precursory crystallization step. The digital twin is constructed as a set of first-principle mechanistic models of each step of the drug purification process in a continuous manner, using the Python-based, object-oriented platform, PharmaPy.⁶ The series of unit operations composing the carousel platform are arranged as a sub-flowsheet inside PharmaPy, which allows the creation of a simulation executive that coordinates the execution of the carousel models, the timing of the integrated processing steps, and the material transfer between them. The models are calibrated by parameter estimation on each processing step in a standalone fashion, using data from an API processed on the physical pilot-scale continuous crystallizer-CFC platform. The resulting model has been utilized as a soft-sensor to identify the CPPs that impact the impurity content throughout the process. The setpoint values for the selected CPPs will be determined by predicting the impurity concentration of the cake and subsequently tuned to produce the final product with admissible range of impurity concentration, while meeting the sufficient product throughput. The suggested soft sensor is validated on a physical continuous crystallizer-CFC platform running in parallel with the digital twin on the compound of interest. The communication interface that allows measurement from the process analytical tools (PATs), and allows setpoint changes from the digital twin has been created utilizing the Python OPC-UA package.⁷ The validation experiments not only confirm the functionality of the digital twin as a guide for mapping the CPPs to attain the desirable CQAs of the final product over multiple operation steps, but also demonstrate its application in a real-time process monitoring and fault detection tool, that can identify critical process disturbances, i.e. filtration mesh fouling.

References

[1] Boukouvala, Fani, et al. "An integrated approach for dynamic flowsheet modeling and sensitivity analysis of a continuous tablet manufacturing process." Computers & Chemical Engineering 42 (2012): 30-47.

[2] Schaber, S. D.; Gerogiorgis, D. I.; Ramachandran, R.; Evans, J. M. B.; Barton, P. I.; Trout, B. L. Economic Analysis of Integrated Continuous and Batch Pharmaceutical Manufacturing: A Case Study. Ind. Eng. Chem. Res. 2011, 50 (17), 10083–10092.

[3] Braatz, R. D. Advanced Control of Crystallization Processes. Annu. Rev. Control 2002, 26 I, 87–99.

[4] D. Acevedo, R. Peña, Y. Yang, A. Barton, P. Firth, Z.K. Nagy, Evaluation of mixed suspension mixed product removal crystallization processes coupled with a continuous filtration system, Chemical Engineering and Processing: Process Intensification, 108 (2016) 212-219.

[5] Nagy, B., Szilágyi, B., Domokos, A., Tacsi, K., Pataki, H., Marosi, G., ... & Nagy, Z. K. (2020). Modeling of pharmaceutical filtration and continuous integrated crystallization-filtration processes. Chemical Engineering Journal, 127566.

[6] Casas-Orozco, D., Laky, D.J., Wang, V., Abdi, M., Feng, X., Wood, E., Laird, C.D., Reklaitis, G. V., Nagy, Z.K., 2021. PharmaPy: an object-oriented tool for the development of hybrid pharmaceutical flowsheets. Comput. Chem. Eng.

[7] Mai, S., & Yi, M. J. (2011, August). An OPC UA client development for monitoring and control applications. In Proceedings of 2011 6th International Forum on Strategic Technology (Vol. 2, pp. 700-705). IEEE.