Characterization and propagation of RTD uncertainty for continuous powder blending processes

Area of Science:

• Chemical Engineering
• Process Systems Engineering
• Pharmaceutical Manufacturing

Background:

• Residence time distribution (RTD) is crucial for characterizing mixing and traceability in pharmaceutical processes.
• RTD measurements inherently contain uncertainties from process fluctuations, measurement errors, and experimental variations.
• Accurate characterization of RTD uncertainty is essential for reliable predictions and quality control in drug manufacturing.

Purpose of the Study:

• To develop and evaluate methods for quantifying and propagating residence time distribution (RTD) uncertainty in pharmaceutical manufacturing.
• To assess the impact of RTD uncertainty on downstream process predictions and quality control applications.
• To provide a framework for robust decision-making under uncertainty in pharmaceutical production.

Main Methods:

• A model-based approach using RTD model parameters and Monte Carlo sampling for uncertainty propagation.
• A data-based approach employing raw experimental data and interval arithmetic for uncertainty characterization.
• A constrained optimization technique to mitigate limitations of interval arithmetic in the data-based approach.

Main Results:

• The developed approaches successfully characterized RTD uncertainty.
• Probability intervals were generated for upstream disturbance tracking and funnel plots.
• The results facilitate improved decision-making for real-time quality control under uncertainty.

Conclusions:

• Both model-based and data-based approaches effectively address RTD uncertainty in pharmaceutical manufacturing.
• The proposed methods enhance the reliability of RTD-based predictions and quality control strategies.
• This work contributes to more robust and accurate process monitoring and control in the pharmaceutical industry.