Text

theses

In the context of the pharmaceutical manufacturing, a recent momentum has developed in using continuous manufacturing lines as opposed to conventional batch manufacturing systems. Processes that are continuous in nature traditionally have been adapting process systems engineering (PSE) tools to assist in their quality management process. The pharmaceutical industry, which has been newly initiated into the domain of continuous manufacturing, presents new and challenging problems within the PSE domain. The primary reason for these challenges is the particulate nature of the raw materials. The design, development and implementation of control systems in such an environment lacks a comprehensive literature base. This work attempts to fill in this void through an exploration of control schemes that can be implemented into a Direct Compaction (DC) continuous manufacturing line. Focus was given to model predictive control (MPC) systems due to their expected augmented performance in comparison to the classical Proportional Integral Derivative (PID) controller. Multiple control strategies were developed in the domain of tablet compaction. A key result was the development and implementation of a multi input multi output (MIMO) MPC that was capable of controlling tablet weight and compression force simultaneously under the assumption that real time tablet weight data was available.
Building upon this Model Predictive Control scheme, an optimization algorithm that was adapted from a previous simulation based study was modified for implementation into the DC manufacturing line. The methodology for its implementation along with some key experimental results is presented here. Here, the demand was a user input to the optimization. The output of this calculation was the production rate set point which was relayed to the MPC. The actual value of the production rate is treated as a disturbance variable. Main compression force was monitored and controlled during various demand scenarios to give an indication of tablet quality.
Finally, a Residence Time Distribution (RTD) based control system was implemented insilico for proof of concept. The RTD of a system can be used to predict outlet parameters if input parameters are known. This was used to predict the concentration of the active pharmaceutical ingredient (API) in tablets at the outlet of the compaction process. This information was used to develop a rejection system that would divert tablets that violate specified tolerance limits.

ETD_9176

doi:10.7282/T3QR51RC

M.S.

Includes bibliographical references

by Aparajith Bhaskar

ETD graduate

The author owns the copyright to this work.