Fluid bed granulation is a complex, multi-phase, multi-component unit operation often used in the pharmaceutical industry. Various attributes such as particle morphology, size, porosity, wettability, and binder viscosity play key roles in determining the overall size and composition of the granules that are produced. Many researchers have sought to investigate, from both an experimental and theoretical perspective, the key parameters that impact granulation. The research presented within this dissertation follows suit.;In this work, the concept of a multi-component population balance equation was utilized as a mathematical framework for modeling fluidized bed granulation. Ultimately, it was desired to develop a computational model that can simultaneously consider the evolution of granule size, moisture content, and compositional distribution. Specifically, the end goal was to develop a model that can evaluate the dynamic evolution of a four-component system containing two powders, binder polymer content, and binder moisture content. To accomplish this task, a very systematic approach was taken.;First, three different numerical methodologies for the population balance equation were examined: a "rigorous" discrete method, the direct quadrature method of moments (DQMOM) technique, and constant-number Monte Carlo. The system under examination contained two morphologically distinct powders that were assumed to already have binder present on the surface. The rigorous discrete method was found to be very computationally efficient. DQMOM was found to be very fast, but was unreliable when analyzing kernels with significant compositional dependence. Constant-number Monte Carlo was determined to be the best choice when analyzing multi-dimensional, composition-dependent population balance equations.;Any model that is developed, though complex, is of no practical good if it cannot faithfully recreate experimental results. As such, simulation results were compared with experimental results at nearly every step along the way in this research. It was found that the constant-number Monte Carlo model continually matches the experimental data for three granulation cases of increasing difficulty: a two-component case that considers the continuous addition of binder to a granulator, a three-component case that evaluates the impact of evaporation and moisture content on the granule growth profile, and ultimately the four-component case mentioned above where granule growth, moisture content, binder distribution, and composition distribution are evaluated.;An effective pharmaceutical wet granulation model should include at least three components: (i) a population balance methodology that tracks the distribution and composition of all species of interest, (ii) a physically based description of agglomeration and breakage, and (iii) hydrodynamic modeling. While the third piece of the puzzle was beyond the current scope of work, significant strides were made in regards to the other two. It was shown that the constant-number Monte Carlo methodology can utilize a complex morphologically-based compositional-dependent coagulation kernel (criteria ii) and track the size evolution of two different powders while also considering the impact of moisture loss on the system (criteria i). Thus, the development of the constant-number Monte Carlo model contained within this dissertation helps to fill a critical need within the granulation community for a model that can provide a thorough analysis of multi-component granulation problems.
展开▼
