Granulation is an industrial process where fine particles are bound together into larger granules. The process has numerous applications including the manufacture of pharmaceuticals and the production of cosmetics, chemicals, detergents and fertilisers. This thesis studies aspects of wet granulation which involves the application of a viscous binder, usually in the form of a spray, to an agitated bed of powder particles. Individual powder particles may adhere together, joined by small quantities of binder fluid called liquid bridges. By a process of collision and adherence additional particles may join the newly formed agglomerates. Agglomerates may also coalesce together which is a process that leads to granule formation. On the completion of this process, granules are typically dried.This thesis studies wet granulation on three different levels. First, micro-level investigations of liquid bridges between two and three particles are performed. For the two-particle case, the fluid profile of static (stationary) and dynamic (moving) liquid bridges is investigated. For the static case, a numerical solution to the Young-Laplace equation is obtained; this relates the volume of binder fluid to liquid bridge properties such as the inter-particle force. An analytic solution is also obtained, providing the liquid bridge profile in terms of known mathematical functions. For both solutions, the radii of the (spherical) primary particles may be different. The dynamic case is then studied using the Navier-Stokes equations with the low Reynolds number approximation. The motion of the approaching particles is shown to be damped by the viscosity of the liquid bridge. Static liquid bridges between three equally sized primary particles are then studied. Symmetry of the problem is used to obtain a numerical solution to the Young-Laplace equation. Liquid bridge properties are calculated in terms of the binder fluid volume. Experimental agreement is provided.Secondly, a model to estimate the stickiness (fractional wet surface area) of agglomerates is proposed. Primary particles are approximated as spheres and are added one at a time in a closely packed arrangement. The model includes parameters to control the inter-particle separation distance and the fluid saturation state. Computational geometry is used to obtain results which relate the number of particles and the volume of binder fluid to the stickiness of the agglomerates.Finally, a population balance model for wet granulation is developed by extending an earlier model to incorporate the effects of binder fluid. Functions for the inter-particle collision rate and drying rate are proposed, including functions which are derived from the geometric model, described above, for the case of maximum particle consolidation. The model is solved numerically for a range of coalescence kernels and results are presented which show the effect of binder volume and the drying rate.

Crystallization is a key unit operation used for obtaining purified products by many process industries. The key properties of the crystalline products, such as size and shape distribution, purity and polymorphic form are controlled by the crystallization process. All these properties impact significantly the downstream operations such as drying or filtration. Therefore, monitoring and controlling this process is fundamental to ensure the quality of the final product. Process analytical technology (PAT) brings numerous new methods and opportunities in the process analytics and real time process monitoring systems, which can be integrated into the control algorithm and provide high level optimal control strategies as well as deeper understanding of the process. Process monitoring helps develop mathematical models which can, in one hand, help in better understanding the processes and consecvently the development and application of advanced control methods in order to achieve better product quality. In this work, image processing and image analysis based direct nucleation control (IA-DNC) is developed in order to investigate the evolution of the crystal properties, such as crystal size, and crystal shape distribution. The IA-DNC approach is also compared to alternative DNC techniques, in which particle number were measured by Focused Beam Reflectance Measurement (FBRM) in order to control crystal size. A control approach is introduced that control the nucleation and disappearance of crystals during cooling and heating segments related to the changes of the number of counts (measured by Particle Vision Measurment, so called PVM or combination of FBRM and PVM). The approach was applied to investigate crystallization of compounds with different behavior: potassium dihydrogen phosphate (KDP) water, contaminated KDP -water and Ascorbic acid water systems. The results demonstrate the application of imaging technique for model-free feedback control for tailoring crystal product properties. The second main aim of the thesis is to investigate and control crystallization processes in impure media in the presence of multiple impurities, with an impact on the crystal shape via growth kinetics. The broad impact of the crystal growth modifiers (impurities) on the growth kinetics is observed in real time by using in situ video imaging probe and real-time image analysis. A morphological population balance model is developed, which incorporates a multi-site, competitive adsorption mechanism of the impurities on the crystal faces. The kinetic parameters of primary nucleation, growth and impurity adsorption for a model system of potassium dihydrogen phosphate crystallization in water in the presence of two impurities, were estimated and validated with experimental results. It was demonstrated that the model can be used to describe the dynamic evolution of crystal properties, such as size and aspect ratio during crystallization for different impurity profiles in the system. Manual, feedback and hybrid feedback-feedforward control techniques are developed and investigated numerically for continuous processes, while model-based and model-free control approach for crystal shape are developed for batch processes. The developed morphological population balance model is implemented and applied in the model-based control approaches, which are suitable to describe multicomponent adsorption processes and their influence on the crystal shape. Case studies show the effectiveness of crystal growth modifiers based shape control techniques. Comparison of different control approaches shows the effectiveness of the techniques. The third part of the thesis deals with purification of crystals when adsorption of impurities on crystal surfaces and its incorporation into crystals are considered. A purification method, called competitive purity control (CPC) is proposed and investigated. A morphological population balance model, including nucleation, growth and competitive impurity adsorption kinetics is developed to describe the case when multiple impurities can adsorb competitively on the crystal surface. The model is also combined with liquid phase chemical reaction model, in order to investigate the purity control case when an additive is introduced in the system that reacts with the impurity forming a non-adsorbing reaction product. Both competitive purity control approaches proposed: the adsorption based competitive purity control (A-CPC) and the reaction based competitive purity control (R-CPC); are investigated using detailed numerical simulations then compared with the alternative widely used purification method, called recrystallization. In the last contribution chapter, an integrated process optimization of a continuous chemical reactor and crystallizer is performed and studied numerically. The purpose of this study is to show the way in which the byproduct produced in the chemical reactor may affect the crystallization process and how its negative effect can be reduced by applying integrated process optimization. Sensitivity analysis of the system was performed by considering the flow rate and the concentration of substances in the input stream of the chemical reactor as manipulated process variables. Model based integrated process optimization and the sensitivity analysis in order to obtain improved quality product in terms of crystal size, shape and purity.

La simulation de réacteurs à bulles en régime industriel est un grand défi. L'objectif principal de ce travail est la prédiction de la taille des bulles à l’aide d’un modèle numérique de bilan de population, basé sur la modélisation des phénomènes de brisure et de coalescence, et pouvant être couplé aux conditions hydrodynamiques présentes dans les réacteurs. Différentes données expérimentales sont obtenues pour valider le modèle. La taille des bulles est mesurée à l'aide d'une technique innovante de corrélation croisée. Les essais, réalisés en eau du réseau (partiellement contaminée) et en eau déminéralisée avec ajout éventuel d'éthanol, montrent que les additifs réduisent la coalescence et diminuent la taille moyenne des bulles. Deux distributeurs du gaz différents sont utilisés pour découpler l'étude de la brisure et de la coalescence. Les données expérimentales sont utilisées initialement pour valider des simulations CFD 3D transitoires Eulériennes-Eulériennes. La loi de traînée est corrigée par un facteur de swarm pour intégrer l’effet d’une fraction de gaz élevée. Différents modèles de turbulence sont testés. La contribution de la turbulence induite par les sillages de bulles au mélange de scalaires est évaluée. Enfin, pour prédire la taille des bulles, un bilan de population est couplé au modèle hydrodynamique préalablement validé et est résolu par la méthode de quadrature des moments (QMOM). Un set original de kernels de brisure et coalescence est proposé, capable de prédire la taille des bulles pour différentes conditions opératoires. Le comportement du modèle lors de l’extrapolation des réacteurs est également examiné / The simulation of bubble column reactors under industrial operating conditions is an exciting challenge. The main objective of this work is to predict the bubble size, in turn interconnected to the reactor hydrodynamic conditions, with computational models, by modelling bubble breakage and coalescence. Experimental data is collected for model validation, including bubble size measurements with an innovative cross-correlation technique. Experiments are carried out with tap water and demineralized water, with or without the addition of ethanol, and gathered results show that additives reduce coalescence and lower the mean bubble size. Two different spargers are used, in order to decouple the investigation of breakage and coalescence. The experimental data set is used to validate out unsteady three-dimensional Eulerian-Eulerian CFD simulations. A drag law for oblate bubbles is considered, together with a swarm factor, that accounts for the swarm effect. Several turbulence models are tested. The contribution of bubble induced turbulence (BIT) to scalar mixing is assessed. To predict bubble size, a population balance model is coupled to the hydrodynamic model and is solved with the quadrature method of moments. A set of breakage and coalescence kernels is proposed, capable of predicting the bubble size for different operating conditions. Scale-up effects are also investigated

Milling is a common unit operation in industry for the purpose of intentional size reduction. Considerable amount of energy is consumed during a grinding process and much of the energy is dissipated as heat and sound, which often makes grinding into an energy-intensive and highly inefficient operation. Despite many attempts to interpret particle breakage during a milling process, the grindability of a material in a milling operation remains aloof and the mechanisms of particle breakage are still poorly understood. Hence the optimisation and refinement in the design and operation of milling are in great need of an improved scientific understanding of the complex failure mechanisms. This thesis aims to provide an in-depth understanding of particle breakage associated with stressing events that occur during milling. A hybrid of experimental, theoretical and numerical methods has been adopted to elucidate the particle breakage mechanics. This study covers from single particle damage at micro-scale to bulk comminution during the whole milling process. The mechanical properties of two selected materials, i.e. alumina and zeolite were measured by indentation techniques. The breakage test of zeolite granules subjected to impact loading was carried out and it was found that tangential component velocity plays an increasingly important role in particle breakage with increasing impact velocity. Besides, single particle breakage via in-situ loading was conducted under X-ray microcomputed tomography (μCT) to study the microstructure of selected particles, visualize the progressive failure process and evaluate the progressive failure using the technique of digital image correlation (DIC). A new particle breakage model was proposed deploying a mechanical approach assuming that the subsurface lateral crack accounts for chipping mechanism. Considering the limitation of existing models in predicting breakage under oblique impact and the significance of tangential component velocity identified from experiment, the effect of impact angle is considered in the developed breakage model, which enables the contribution of the normal and tangential velocity component to be rationalized. The assessment of breakage models including chipping and fragmentation under oblique impact suggests that the equivalent normal velocity proposed in the new model is able to give close prediction with experimental results sourced from the public literature. Milling experiments were performed using the UPZ100 impact pin mill (courtesy by Hosokawa Micron Ltd. UK) to measure the comminution characteristics of the test solids. Several parameters were used to evaluate the milling performance including product size distribution, relative size span, grinding energy and size reduction ratio etc. The collective data from impact pin mill provides the basis for the validation of numerical simulation results. The Discrete Element Method (DEM) is first used to model single particle breakage subject to normal impact loading using a bonded contact model. A validation of the bonded contact model was conducted where the disparity with the experimental results is discussed. A parametric study of the most significant parameters e.g. bond Young’s modulus, the mean tensile bond strength, the coefficient of variation of the strength and particle & particle restitution coefficient in the DEM contact model was carried out to gain a further understanding of the effect of input parameters on the single particle breakage behavior. The upscaling from laboratory scale (single particle impact test) to industrial process scale (impact pin mill) is achieved using Population Balance Modelling (PBM). Two important functions in PBM, the selection function and breakage function are discussed based on the single particle impact from both experimental and numerical methods. An example of predicting product size reduction via PBM was given and compared to the milling results from impact pin mill. Finally, the DEM simulation of particle dynamics with emphasis on the impact energy distribution was presented and discussed, which sheds further insights into the coupling of PBM and DEM.

In some species, including humans, red blood cells (RBCs) under low shear stress tend to clump together and form into regular stacks called rouleaux. These stacks are not static, and constantly move and break apart. This phenomenon is referred to as red blood cell aggregation and disaggregation. When modelled as a single liquid, blood behaves as a non-Newtonian fluid. Its viscosity varies, mainly due to the aggregation of RBCs. The aim of this research is to develop a mesoscale computational model for the simulation of RBCs in plasma. This model considers RBC interaction and aggregation to predict blood-flow characteristics such as viscosity, rouleaux size and velocity distribution. In this work, the population-balance modelling (PBM) approach is utilized to model the RBC aggregation process. The PBM approach is a known method that is used for modelling agglomeration and breakage in two-phase flow fluid mechanics to find aggregate size. The PBM model is coupled to the incompressible Navier-Stokes equations for the plasma. Both models are numerically solved simultaneously. The population-balance equation has been used previously in a more restricted form, the Smoluchowski equation, to model blood viscosity, but it has never been fully coupled with the Navier-Stokes equation directly for the numerical modelling of blood flow. This approach results in a comprehensive model which aims to predict RBC aggregate size and their velocities for different flow configurations, as well as their effects on the apparent macro-scale viscosity. The PBM approach does not treat the microscopic physics of aggregation directly but rather uses experimental correlations for aggregation and disaggregation rates to account for the effects of aggregation on the bulk. To find the aggregation rate, a series of experiments on RBC sedimentation due to gravity is designed. In these tests, aggregated RBCs (rouleaux) tend to settle faster than single RBCs and, due to low shear stresses, disaggregation is very low and can be neglected. A high-speed camera is used to acquire video-microscopic pictures of the process. The size of the aggregates and their velocities are extracted using image processing techniques. For image processing, a general Matlab program is developed which can analyze all the images and report the velocity and size distribution of rouleaux. An experimental correlation for disaggregation rate is found using results from a previous steady-state Couette flow experiment. Aggregation and disaggregation rates from these experiments are used to complete the PBM model. Pressure-driven channel flow experiments are then used for the final validation of the model. Comparisons of the apparent viscosity of whole blood in previous experiments show reasonable agreement with the developed model. This model fills a gap between micro-scale and macro-scale treatments and should be more accurate than traditional macro-scale models while being cheaper than direct treatment of RBCs at the micro-scale.

Den pågående elektrifieringen av transport och samhälle kräver utveckling av nya metoder för återvinning av batterier. Hydrometallurgi som fokuserar på selektiv kristallisation av metaller är ett intressant alternativ för dessa ändamål. Dessa system kan studeras genom modellering och simulering. Många matematiska modeller finns tillgängliga för att beskriva de olika involverade processerna i kristallisationen av metaller. Dessa processer inkluderar övermättnad, nukleation, kristalltillväxt och aggregation. Denna rapport sammanställer ett antal av de tillgängliga matematiska modellerna och presenterar ett numeriskt tillvägagångssätt för modellering av den tidsberoende nummerdensiteten av partiklar genom en populationsbalansekvation. Populationsbalansen kan lösas med olika metoder såsom momentmetoden och metoden av viktade residualer. Här löses ekvationen genom diskretisering. Diskretisering av den inre koordinaten i ett flertal längdintervall möjliggör simulering av partikel-storleksfördelningen som en funktion av tid. Det numeriska tillvägagångssättet applicerades på bariumsulfatutfällning i en perfekt blandad satsreaktor och två- och tre-dimensionella T-mixer-system, såväl som en perfekt blandad satsreaktor för förträngningskristallisation av nickelsulfat med groddning. Den simulerade storleksfördelningens placering visade sig ha bra överenstämmelse med experimentell data vid låga Reynolds-tal. Här undersöktes även påverkan av en mängd parametrar såsom diskretisering, aggregation och magnituden av diffusion. Aggregation hade en märkbar inverkan på välblandade system. Inverkan av aggregation i diffusions-kontrollerade system med kort retentionstid var låg. Diffusionsmagnituden hade liten påverkan på den normaliserade distributionen men större på det totala antalet partiklar. / The currently ongoing electrification of society and transport necessitates the development of novel methods for battery recycling. Hydrometallurgy with a focus on selective metal crystallisation is an interesting prospect to these ends. The resource recovery systems of interest can be studied through simulation where many mathematical models are available to describe the varying processes involved. These processes include supersaturation generation, nucleation, growth and aggregation. This work compiles some of these mathematical models and presents a numerical approach for the modelling of the time-dependent particle number density with a population balance equation. The population balance equation can be solved using a variety of different methods such as method of moments and method of weighted residuals. Here, the balance equation was solved by discretisation. Discretising the inner coordinate (crystal length) into a number of length intervals allows for the particle size distribution to be modelled as a function of time for various crystallisation systems. The framework was successfully applied to barium sulphate precipitation in a perfectly mixed batch reactor and two- and three-dimensional T-mixer systems, as well as a seeded perfectly mixed nickel sulphate anti-solvent crystallisation system. The simulated size distribution showed promising similarity to experimental data at low Reynolds number. The influence of a variety of parameters such as aggregation and magnitude of diffusion was investigated. Aggregation had a significant impact on well-mixed systems increasing with retention time. The impact of aggregation on diffusion-controlled systems with low retention time was low. The magnitude of diffusion had little impact on the particle size distribution of the crystal population but a large impact on the total number of crystals.

<p dir="ltr">The dissolution profile is one of the most important critical quality attributes (CQAs) for pharmaceutical solid oral dosage forms, as failure to meet the dissolution specification can impact bioavailability. Dissolution tests are essential to assess lot-to-lot product quality and guide the development of new formulations. Therefore, predictive dissolution reduced-order models (ROM) are crucial for the successful implementation of any real-time release testing (RTRT) strategy. Mechanistic and semi-mechanistic ROMs of tablet dissolution for realizing quality by control (QbC) and RTRT frameworks in continuous manufacturing are still scarce or nonexistent. Moreover, realizing the underlying coupled mechanics of wetting, swelling, disintegration, and dissolution is still an open question. This dissertation contributes to developing a mechanistic ROM of pharmaceutical tablet dissolution for the design, optimization, and control of manufacturing processes. We follow several steps towards the progression of the mechanistic model development. First, we develop a semi-mechanistic ROM to capture the relationship between critical process parameters (CPPs), critical material attributes (CMAs), and dissolution profiles. We demonstrate the versatility and the capability of the semi-mechanistic ROM to estimate changes in dissolution due to process disturbances in tablet porosity, lubrication conditions, and moisture content in the powder blend. Next, to understand the underlying coupled mechanism of wetting, swelling, disintegration, and dissolution, we use dynamic micro-computed tomography (micro-CT) with a high temporal resolution to visualize water penetration through the porous network of immediate-release tablets. We couple liquid penetration due to capillary pressure described by the Lucas-Washburn theory with the first-order swelling kinetics of the excipients to provide a physical interpretation of the experimental observations. From the mechanistic understanding of the water penetration kinetics using the micro-CT tests, we propose a two-stage mechanistic ROM, which is comprised of (i) a mechanistic dissolution model of the active pharmaceutical ingredient (API) that solves a population balance model (PBM) for a given API crystal size distribution and dissolution rate coefficient, and (ii) a tablet wetting function that estimates the rate at which the API is exposed to the buffer solution. These two sub-models are coupled by means of convolution in time to capture the start time of the API dissolution process as water uptake, swelling, and disintegration take place. Finally, we demonstrate the versatility and the capability of the mechanistic API dissolution model and the two-stage tablet dissolution ROM to represent the dissolution profile of different pharmaceutical formulations and its connection with CMAs, CPPs, and other CQAs, namely initial API crystal size distribution, porosity, composition, and dimensions of the tablet. In all of the cases considered in this work, the estimations of the model are in good agreement with experimental data. </p>