(470g) Sensitivity Analysis and Identification of Feasible Region of a Wet Granulation Continuous Pharmaceutical Manufacturing Process

Driven by the advances in continuous pharmaceutical manufacturing when compared to batch processes, extensive research is focused on developing predictive models for continuous unit operations. While this is leading to detailed and more accurate models, it is accompanied by an increase in model complexity and computational expense. This is further amplified when several such models are integrated to develop a flowsheet model that simulates the dynamic behavior of a plant scale manufacturing process. The limitations are particularly exacerbated for integrated models simulating the wet granulation solid oral dosage manufacturing route as different approaches are used to model unit operations in this route. For example, feeders and blenders are modeled using semi-empirical approaches where as units involving particle size changes such as granulation and milling are simulated using population balance models¹. Discrete element method models are developed to understand and capture particle scale interactions that would render more accuracy to the model. In order to fully realize the potential of the developed models through applications such as sensitivity analysis and feasibility analysis, there is a need to develop and implement strategies to overcome the computational limitations.

Sensitivity analysis is the study of how uncertainty in a model output can be assigned to different sources of uncertainty in the model input². Sources of uncertainty exist through potential variability in raw material quality, operating conditions and production rate demands. As flowsheet models have several input and output variables, it is important to identify the variables that have a significant effect on the properties of interest. This allows prioritization of the input factors and simplification of the high dimensional problem that aids in further applications such as feasibility analysis. Feasibility analysis is used to identify the feasible region of a process that is characterized by the range of input variables within which the process meets equipment, quality and production constraints³. Traditional methods to identify feasible regions incur large sampling costs which is prohibitive for a wet granulation flowsheet model. In addition, some constraints are not available in closed form expression. Alternatively, they may be available but expensive to evaluate⁴.

Current work utilizes surrogate based feasibility analysis method⁵ that is based on building surrogate models used to approximate the function of interest. In this work, a reduced order model is built for feasibility function, which characterizes the maximum constraint violation. The surrogate model accuracy is improved through selection of new samples that are identified using an adaptive sampling strategy. The strategy uses a modified Expected improvement function to direct search towards feasible region boundaries and unexplored regions⁶. The developed approach is utilized to identify feasible regions for an integrated wet granulation line that simulates a plant scale operation.

References

1. Boukouvala F, Chaudhury A, Sen M, et al. Computer-Aided Flowsheet Simulation of a Pharmaceutical Tablet Manufacturing Process Incorporating Wet Granulation. J Pharm Innov. 2013;8(1):11-27.
2. Saltelli A, Annoni P, Azzini I, Campolongo F, Ratto M, Tarantola S. Variance based sensitivity analysis of model output. Design and estimator for the total sensitivity index. Computer Physics Communications. 2010;181(2):259-270.
3. Wang Z, Escotet-Espinoza MS, Ierapetritou M. Process analysis and optimization of continuous pharmaceutical manufacturing using flowsheet models. Computers & Chemical Engineering.
4. Rogers A, Ierapetritou M. Feasibility and flexibility analysis of black-box processes Part 1: Surrogate-based feasibility analysis. Chem Eng Sci. 2015;137:986-1004.
5. Wang Z, Ierapetritou M. A novel feasibility analysis method for black‐box processes using a radial basis function adaptive sampling approach. Aiche J. 2017;63(2):532-550.
6. Boukouvala F, Ierapetritou MG. Derivative‐free optimization for expensive constrained problems using a novel expected improvement objective function. Aiche J. 2014;60(7):2462-2474.