(200p) Virtual Screening of Process Parameters for Pharmaceutical Drying Operation: A Combined DoE-CFD Approach

In recent years, the mechanistic modeling of pharmaceutical unit operations has made significant progress. Modern process modeling has been recognized as an important tool for developing fundamental understanding of processes, optimization, scale-up, sensitivity analysis, transient dynamics and development of predictive control strategies. There has been growing interest in quality-by-design (QbD) based thinking for various purposes including cost reduction and ensuring consistent product quality. Process modeling gives a perfect platform to achieve these objectives by reduction in batch cycle time, improvement in energy efficiency and miniaturization of equipment sizes.

For example, spray and fluidized bed drying operations are carried out with short residence times due to excellent gas-particle contact, very high heat and mass transfer rates. These help in uniform drying of all particles and also significantly reduce thermal degradation of product. Although the operation is simple, the design engineer has to make choices related to large number of operating parameters including inlet gas humidity, temperature and flow rate, incoming slurry temperature, slurry concentration, position and type of atomizer, size of the equipment and permissible exhaust air temperature. The effect of these variables on product quality and throughput needs systematic investigation. However, the number of experiments/simulations can be extremely high. Design of experiments (DoE) plays a key role in addressing this issue.

In this work, operating conditions of existing spray and fluid bed drying were optimized to improve drying of drug products and drug substances with known drying kinetics. The optimization was carried out using virtual screening of operating parameters using a combined design of experiments (DoE) and computational fluid dynamics (CFD) approach. The inlet gas humidity, temperature and flow rate are considered as manipulated variables. The set of CFD simulations were shortlisted using central composite design (CCD). For each parameter, three levels were considered and their impact on the final moisture content has been assessed. The CFD simulations were carried out in a Eulerian-Lagrangian framework. In the constant and falling rate periods, the drying kinetics were modelled by modified vapor pressure approach. The simulation results were validated with the actual data in terms of product moisture and outgoing air temperature. In addition, the residence time distribution (RTD) of gas and solid phase has been studied using CFD. A simplified process model for spray and fluid bed drying supplemented with RTD information has been proposed and incorporated in Dynochem. The Dynochem predictions were found to correlate well with the rigorous CFD approach.

The methodology has been demonstrated using following case studies:

The first case study is the use of spray drying to obtain crystal morphologies that could not be obtained using conventional solution phase crystallization. We show how we can predict, from the droplet size and residence time distribution in the spray dryer, the distribution of the crystal morphologies finally obtained.

The second case study involves drying in a fluidized bed. We have used the aforementioned approach to predict characteristics of the dried material as a function of temperature, flow-rate and moisture content of the drying air and the drying time. Also, this approach has been combined with the models of nonnewtonian fluid processing for extending these calculations to wet granulation.