(652c) Systematic Retrofitting Methodology for Pharmaceutical Drug Purification Processes
- Conference: AIChE Annual Meeting
- Year: 2015
- Proceeding: 2015 AIChE Annual Meeting Proceedings
- Group: Pharmaceutical Discovery, Development and Manufacturing Forum
- Time: Thursday, November 12, 2015 - 9:10am-9:30am
Pharmaceutical production processes are gaining research attentions in the field of CAPE/PSE. In this work, we present a systematic method for retrofitting pharmaceutical manufacturing plants, together with an industrial case study.
The method consists of five tasks: (I) collection of process understanding/data, (II) creation of process model, (III) adaptation of the model for optimization, (IV) optimization and (V) task outcome interpretation. In Task I, an overview of the process is obtained by performing Path Flow Decomposition (PFD), which determines the sources and sinks of the all substance in a batch, and thus facilitates the creation of a model. In Task II descriptive mathematical models are created for every unit, which will then be evaluated for feasibility in Task V. The aim of Task III is to transform the descriptive model into an analytical model, which relates the optimization variables to the objective function. In parallel to identifying variables to be optimized, analysis is performed on constraints, especially on the pharma-specific, GMP and product quality aspects. All these information are considered in Task IV, i.e., optimization, where MINLP plays a central role. Task V runs simultaneously to all the other tasks, evaluating the feasibility of the outcomes and the necessity to iterate corresponding Task(s), which resembles iterative nature of the design.
The methodology was applied to a manufacturing plant of a solid drug, where crude crystal is purified through the process of Dissolution-Filtration-Crystallization (DFC). In Task I, a trade-off was identified: the higher the process temperature, TH, the more degradation of the crude crystal; the lower the TH, the more amount of undissolved crystal. The process flow rate, F, affects also the overall yield. In Task II, process model was created with considering measured data, e.g. particle size distribution (PSD) during the dissolution for the heat exchanger. The descriptive model was adapted for optimization by incorporating accurate information, e.g., liquid-wall-liquid heat transfer. We were able to model reactive-dissolution of crystals with PSD inside the heat exchanger, which is the secondary novelty of the work. As Task IV, three alternative process layouts were considered. The insertion of coolers before and after the filtration unit enabled a substantial improvement, lowering the degradation reaction and thus increasing the yield and the net present value. The feasibility of the model was supported by the result of the performance of lab scale experiments on a plant with the same new layout.
The methodology is a practical decision-supporting tool for the retrofitting of pharmaceutical purification processes, and its application is not only limited to the DFC processes. The novelty lays in its integrative nature, indeed the combination of different technics, e.g., information gathering in GMP environment, mathematical modelling and multiobjective optimization, and decision making, in the field of pharmaceutics. The methodology can show its weakness in case the data measurement is difficult, however, elucidation of trends and behaviours inside the plant can still serve as a basis for process retrofitting.