Dynamic modelling, transient behaviour analysis and scheduling of volatile organic compound (VOC) abatement systems for the pharma industry

Abstract

Active Pharmaceutical Ingredient (API) manufacturing is heavily reliant on solvents for reactions and separations. Among them, Volatile Organic Compounds (VOCs) reign supreme, although their emissions pose a risk both to the environment and human health. With the effects of climate change becoming more prominent, stringent regulations are in place to ensure the enforcement of environment-conscious industrial. Adsorption on activated carbon beds is an established VOC control technology, often preferred on an industrial level due to the low cost of installation and maintenance, as well as its effectiveness in treating large volumetric streams containing trace amounts of pollutants. However, the simultaneous feeding of varying composition and load waste streams from plant-wide process vents under batch operation causes quick and irregular bed saturation, leading to sub-optimal process efficiency, with higher operational costs due to frequent adsorbent material regeneration outsourcing. Despite the multitude of studies on adsorption over the years, there is a profound mismatch towards the proportion of research focused on multicomponent VOC adsorption under realistic industrial conditions. Mathematical modelling and simulations are valuable tools towards understanding the complex interactions of competing solvent emission streams and allow process optimisation without the need for costly and resource-intensive pilot plant experiments.

This PhD thesis aims to elucidate the intrinsic relationships between complex pharmaceutical manufacturing solvent vapour mixtures and propose operational optimisation scenarios to minimise environmental leak risks as well as maximise process efficiency, at an even lower operational cost. In order to address the dire need for reliable thermodynamic parameter values, a Langmuir Isotherm parameter database is established for organic solvents based on published experimental data. Next, a dynamic, multicomponent, nonisothermal adsorption model is constructed and used to highlight the interactions of key VOC binary mixtures under a range of steady-state and transient feed conditions, temperatures, activated carbon bed structures, column lengths and stream velocities. Furthermore, Hodograph Theory breakthrough metrics are compared with validated simulation results to test the extrapolation potential of single component mixture predictions to multicomponent mixtures as a quick, operational plant decision-making tool. The dynamic model simulation results for the different mixtures are then employed to inform Mixed Integer Linear Programming (MILP) scheduling models for process optimisation and comparative economic evaluation. This PhD thesis highlights the immense value of systematic and rigorous model-based simulation and optimisation campaigns for VOC emission abatement systems in the pharmaceutical industry.