Many industrial processes involve powders in some form when making products, and the behavior of the powders processed is impacted by the adhesion of the individual particles which comprise it. This adhesion behavior, in turn, is critically influenced by the complementarity between the topography of a surface and the shape and roughness of the particles that adhere to that surface. Problems such as poor flowability, dust hazards, and equipment wear arise due to uncontrolled particle adhesion and can lead to production challenges. Computational models have been developed to predict the behavior of highly idealized powders (i.e., powders comprised of simple geometries such as spheres) under various processes but are limited in their ability to model and optimize the manufacturing and handling of powders comprised of many complex particles. This work focuses on further developing an experimental and modeling framework, called the Enhanced Centrifuge Method (ECM), that maps particle-scale and surface properties onto experimentally-validated ‘effective’ adhesion distributions that describe the adhesion between particles in powders. These distributions represent an engineering approach that allows powders comprised of particles of complex shape and roughness, which are challenging to model, to be described as if they were perfect, smooth spheres, which are comparatively simple to model. The complexity associated with the shape and size distributions of the individual particles is captured by the ‘effective’ adhesion parameters. These ‘effective’ adhesion parameter distributions provide a quantitative guide as to how the specific particle properties are interacting with the surface topography which directly impacts the overall powder adhesion. The initial framework of the ECM is constructed around characterizing the van der Waals adhesion of silica and polystyrene powders. The impact of the surface topography and the particle properties of each of the powders is captured in ‘effective’ Hamaker constant distributions. These distributions provide a quantitative guide for specifically how the particles interact with the surface topography based on the respective scales of the particle and surface features. The ECM framework is further adapted here to investigate the effects of topographical changes of stainless steel due to polishing on the adhesion properties of three different pharmaceutical powders to the stainless steel. In this adaptation of the ECM framework, the force of adhesion was described by modifying the Johnson, Kendall, and Roberts (JKR) model describing elastic-like particle contact to a flat plate. Within the modified JKR adhesion description, the work of adhesion is tuned to be an ‘effective’ work of adhesion parameter. These size-dependent ‘effective’ work of adhesion distributions provide a quantifiable measure of the change in the powder and surface adhesion that reflects the size, shape, and topographical features on the powder and surface with which the powder interacts. To investigate environmental effects on the adhesion properties, the ECM framework is also extended to characterize the effect humidity has on altering surface and particle interactions of the three pharmaceutical powders to stainless steel. In addition to the work with the pharmaceutical powders, the investigation of the effect of humidity on the powder’s adhesion includes a study of the influence of water on the interactions between silica particles and a silica substrate. In all cases, the ‘effective’ adhesion force distributions developed through the ECM provide the ability to quickly determine quantitatively how environmental and process conditions alter particle and surface properties, and overall powder behavior.

Confined particulate systems, particularly powder compacts, are widely used in various applications in industries such as pharmaceutical, automotive, agriculture, and energy production. Due to their extensive applications, characterization of these materials is of great importance for optimizing their performance and manufacturing processes. Modeling approaches capable of capturing the heterogeneity and complex behavior are effective at predicting the macroscopic behavior of granular systems. These modeling approaches utilize information about the microstructure evolution of these materials during compaction processes at the mesoscale (particle-scale). Using these types of modeling depend on accurate contact formulation between inter-particle contacts. The challenge comes in formulating these contact models that accurately predict force-area-deformation relationships. In this work, contact laws are presented for elastic-ideally plastic particles and plastic particles with power-law hardening under unconfined (simple compression) and confined (die and hydrostatic compaction) compression. First, material properties for a set of finite element simulations are obtained using space-filling design. The finite element simulations are used for verification and building an analytical framework of the contact radius and contact pressure which allows for efficient determination of the contact force. Semi-mechanistic contact laws are built for elastic-ideally plastic spherical particles that depend on material properties and loading configuration. Then, rigid-plastic assumption is used to modify the contact laws to consider power-law hardening effects while keeping loading configuration dependency. Finally, after building and verifying the contact laws, they are used to estimate hardening properties, contact radius evolution, and stress response of micro-crystalline cellulose particles under different loading configurations using experimental data from simple compression.

<p>Controllable continuous feeding of biomass feedstock in a biorefinery is critical to upscaling current ethanol conversion techniques to a commercial scale. Mechanical pretreatment of biomass feedstock performed using a compression feed screw (CFS) improves the ethanol yield but is subject to flowability issues, especially the plugging of biomass. The mechanical behavior, and hence, the flowability of biomass feedstock, is strongly affected by several factors, including preparation method, moisture content, physical composition, and particle size distribution. In addition, the current design of CFS is guided by limited experimentation and even fewer theoretical correlations. This thesis aims at developing computational methods to model the flow of densified feedstock in a CFS and experimental techniques to characterize the mechanical properties required for the model. We adopted a modified Drucker-Prager Cap constitutive (mDPC) law for milled corn stover (a widely used feedstock for bioethanol production) to model the material’s rate-independent bulk behavior in a CFS. The mDPC elastoplastic law captures the frictional shear and permanent volumetric changes in corn stover using a continuous porosity-dependent yield surface. The parameters of the mDPC model are calibrated using a unified set of single-ended die compaction and multiple shear failure tests. In addition, we quantified the changes in the mDPC parameters with moisture content up to the water-holding capacity of corn stover particles. A Coupled Eulerian-Lagrangian Finite Element Method model developed for the CFS geometry predicts the deformation of the material using the calibrated mDPC parameters. We model the interaction between the material and the CFS surface using a Coulomb wall friction coefficient calibrated using the Janssen-Walker method for a punch and die system. A laboratory-scale compression feed screw is designed and fabricated to characterize the flow of dense granular materials in collaboration with undergraduate students in the School of Mechanical Engineering. FEM model predictions of feeding torque and mass flow rate are validated against the laboratory-scale feeder for microcrystalline cellulose Avicel PH-200 and milled corn stover. The model predictions agree with the experiments for Avicel PH-200 but have a higher error in the case of corn stover. Some physical effects, such as shear hardening and particle erosion observed in milled corn stover, are not captured using the current implementation of the mDPC model, which explains the different model accuracies for both materials. The continuum model is used to uncover material density distribution, torque, and pressure inside the CFS, otherwise challenging through experiments. The FEM model showed a significantly higher sensitivity of the feeder performance to two material properties, namely the hydrostatic yield stress and the wall friction coefficient. The characterized variation of material properties with moisture content and the effect of each material property on the feeder performance provide strategies to engineer the feedstock for better flowability. Further, the continuum model offers a method to study design changes before manufacturing the equipment. Finally, we propose the possibility of a reduced-order analytical model based on the critical material properties and the material deformation mechanism demonstrated by the FEM model.</p>

<p>  </p> <p>Crystallization is a purity and particle control unit operation commonly used in industries such as pharmaceuticals, agrochemicals, and energetics. Often, the active ingredient’s crystal mean size, polymorphic form, morphology, and distribution can impact the critical quality attributes of the final product. The active ingredient typically goes through a series of process development iterations to optimize and scale-up production to reach production scale. Guided by the FDA, the paradigm shift towards continuous processing and crystallization has shown benefits in introducing cheaper and greener technologies and relieving drawbacks of batch processing. To achieve successful batch scale-up or robust continuous crystallization design, process intensification of unit operations, crystallization techniques, and utilizing data driven approaches are effective in designing optimal process parameters and achieving target quality attributes. </p> <p>In this thesis, a collection or toolbox of various process intensification techniques was developed to aid in control, optimization, and digitalization of crystallization processes. The first technique involves developing a novel control algorithm to control agrochemical crystals of high aspect ratio to improve the efficiency of downstream processes (filtration, washing, and drying). The second technique involves the further improvement of the first technique through digitalization of the crystallization process to perform simulated optimization and obtain a more nominal operating profile while reducing material consumption and experimentation time. The third method involves developing a calibration procedure and framework for in-line video microscopy. After a quick calibration, the in-line video microscopy can provide accurate real-time measurements to allow for future control capabilities and improve data scarcity in crystallization processes. The last technique addresses the need for polymorphic control and process longevity for continuous tubular crystallizers. Through a sequential stirred tank and tubular crystallizer experimental setup, the control of polymorphism, particle mean size, and size distribution were characterized. Each part of this thesis highlights the importance and benefits of process intensification by creating a wholistic process intensification framework coupled with novel equipment, array of PAT tools, feedback control, and model-based digital design.</p>

<p>Crystallization is a key solid-liquid separation and purification technique used in pharmaceutical industry. Some of the critical quality attributes (CQAs) of a product from crystallization process include crystal size distribution (CSD), purity, polymorphic form, morphology, etc. Different size and polymorphs of a drug substance may have different dissolution profiles and different bioavailability, which can have adverse effect on human health. Therefore, it is important to design and control crystallization process to meet product CQAs. In recent years, drug substances are becoming more complex, often being heat sensitive, which may limit the temperature that can be used in the crystallization step. Consequently, the traditional cooling only crystallization may not be well suited to recover the high value drug substances. For these systems, antisolvent crystallization is typically employed to improve the yield. On the other hand, the solvent composition can significantly impact the polymorphic outcome. Therefore, designing combined cooling and antisolvent crystallization (CCAC) processes to solve the challenges of active pharmaceutical ingredient (API) crystallization in a highly regulated environment is a complex engineering problem. </p> <p>With rising energy costs and intense price competition from generic pharmaceutical companies, the pharmaceutical industry is looking for ways to reduce the cost of manufacturing via process intensification (PI). This thesis focuses on different PI techniques for CCAC of drug substances. Continuous or smart manufacturing is gaining popularity due to its potential to lower the cost of manufacturing while maintaining consistent quality. Continuous crystallization is an important link in the continuous manufacturing process. The first part of the thesis shows PI of a commercial drug substance, Atorvastatin calcium (ASC) for target polymorph development via continuous CCAC using an oscillatory baffled crystallizer (OBC). An existing batch CCAC process for ASC was compared with the continuous CCAC in OBC and it was found the continuous process 30-fold more productive compared to the batch process. An array of process analytical technology (PAT) tools was used in this work to assess key process parameters that affect the polymorphic outcome and CSD. The desired narrower CSD product was obtained in the OBC compared to that from a batch crystallizer.</p> <p>The next part of the thesis focused on model-based PI technique for efficient determination of crystallization kinetics of a polymorphic system in CCAC. A novel experimental design was proposed which significantly reduced the number of experiments required to determine crystallization kinetics in a CCAC process. The kinetic parameters were validated, and a validated polymorphic model was used to perform an in-silico design of experiment (DoE) to develop a design space that can be used to identify operating conditions to achieve a desired crystal size and polymorphic form. </p> <p>The final part of the thesis combines the experimental and model-based approach for designing a continuous CCAC process for ASC in a cascade of Coflore agitated cell reactor (ACR) and three-stage mixed suspension mixed product removal (MSMPR). A combined artificial neural network (ANN) and principal component analysis (PCA) method was used to calibrate an ultraviolet (UV) probe which was used to monitor ASC solute concentration in the cascade process. The crystallization kinetic parameters were estimated in ACR and MSMPR which was used to build a digital model of the cascade process. The digital model was then used to obtain a design space with different temperature profile in the three-stage MSMPR that yielded narrow CSD of ASC form I. Overall, this thesis demonstrates the benefits of applying PI in the CCAC of drug substances using a holistic approach including novel equipment, application of an array of PAT tools, and model-based digital design to achieve desired CQAs of the product.</p>

The research presented in this work addresses open questions regarding (i) the fundamental understanding of powder compaction, and (ii) the complex mechanical response of particle-binder composites under large deformations. This work thus benefits a broad range of industries, from the pharmaceutical industry and its recent efforts on continuous manufacturing of solid tablets, to the defense and energy industries and the recurrent need to predict the performance of energetic materials. Powder compacts and particle-binder composites are essentially confined particulate systems with significant heterogeneity at the meso (particle) scale. While particle mechanics strategies for modeling evolution of mesoscale microstructure during powder compaction depend on the employed contact formulation to accurately predict macroscopic quantities like punch and die wall pressures, modeling of highly nonlinear, strain-path dependent macroscopic response without a distinctive yield surface, typical of particle-binder composites, requires proper constitutive modeling of these complex deformation mechanisms. Moreover, continued loading of particle-binder composites over their operational life may introduce significant undesirable changes to their microstructure and mechanical properties. These challenges are addressed with a combined effort on theoretical, modeling and experimental fronts, namely, (a) novel contact formulations for elasto-plastic particles under high levels of confinement, (b) a multi-scale experimental procedure for assessing changes in microstructure and mechanical behavior of particle-binder composites due to cyclic loading and time-recovery, and (c) a finite strain nonlinear elastic, endochronic plastic constitutive formulation for particle-binder composites.

<p dir="ltr">Metal speciation and structure can be influenced by the deposition method used during synthesis, interactions with the support, and by post-deposition treatments and reaction conditions experienced during its lifetime of carrying out a catalytic reaction. Supported metal particles of different size contain different surface structures and coordination environments, which may not only influence reaction rates but also the interconversion between agglomerated metallic domains and dispersed metal atom or ion sites. Here, we address the influence of post-deposition treatments and support properties on the structural interconversion of Pd and Cu on aluminosilicate chabazite (CHA) zeolites, Pt on gamma-alumina (γ-Al2O3), and Pd on amorphous oxides (γ-Al2O3, La-doped Al2O3, ΘΔ-Al2O3). The fundamental insights from these studies can be used to design catalysts used widely in automotive exhaust aftertreatment systems, including Pd-exchanged zeolites for passive NOx (x = 1,2) adsorbers (PNA), Cu-exchanged zeolites for NOx (x = 1,2) selective catalytic reduction (SCR), Pt/Al2O3 for NH3 oxidation, and Pd/oxides for three-way catalysts (TWC). Incipient wetness impregnation (IWI) and colloidal methods were used to prepare Pd nanoparticles deposited on CHA zeolites with distinct Pd nanoparticle sizes and distributions. These Pd-CHA samples were used to investigate the effects of Pd particle size distribution on structural interconversion between ion-exchanged Pd and agglomerated Pd domains under realistic operating conditions. Smaller Pd nanoparticles had larger fractions of agglomerated Pd that converted to ion-exchanged Pd2+ sites at fixed air treatment temperatures (598–973 K) and H2O pressures (2–6 kPa H2O), consistent with thermodynamic predictions from DFT calculations. Furthermore, the addition of H2O during air treatment of different Pd nanoparticles (2–14 nm) inhibited the formation of ion-exchanged Pd2+ (thermodynamics), but not the rate of redispersion (kinetics). This demonstrates that, regardless of Pd nanoparticle size, water vapor in automotive exhaust streams facilitate metal sintering in PNA applications. Aqueous-phase exchange of Cu on CHA zeolites with varying support properties (i.e., number of paired Al sites in the 6 membered ring) were used to prepare materials with distinct types and numbers of extraframework Cu species (Cu2+, CuOH+). These Cu-CHA materials were used to analyze Cu structural changes before and after exposure to hydrothermal aging conditions. In the absence of H2O, some Cu2+ sites condense to form binuclear Ox-bridged Cu species that can be reduced with H2 to form Cu-hydride sites and reject H2O, leading to a sub-stoichiometric H2 consumption (H2/Cu < 0.5). In the presence of H2O, all nominally isolated Cu2+ species convert to [CuOH]+ structures, which can subsequently be reduced by H2 to form a Cu-hydride and reject H2O, leading to stoichiometric H2 consumption (H2/Cu ~ 0.5). Furthermore, the presence of H2O led to reduction features in H2 temperature programmed reduction (TPR) profiles that were similar among Cu-CHA materials, regardless of the initial Cu2+ speciation, further supporting the proposal that all nominally isolated Cu2+ sites convert to a similar [CuOH]+ motif. This demonstrates how water influences Cu speciation on CHA materials of varying origin or treatment history, aiding in quantifying SCR-active isolated Cu ions and SCR-inactive Cu species (e.g., CuO, CuAl2O4). Pt supported on γ-Al2O3 were prepared with different average Pt particle sizes (2–13 nm) by increasing the temperature of post-deposition air treatment (523–873 K). This suite of materials was interrogated to isolate the effects of Pt particle size on NH3 oxidation rates and selectivities during conditions relevant to NH3 slip applications in diesel exhaust aftertreatment. For all Pt particle sizes, NH3 oxidation rates displayed a hysteresis with temperature, with high rates measured during temperature decreases than during temperature increases. Smaller Pt particles (2 nm) had lower rates (per surface Pt, quantified by CO chemisorption) than larger Pt particles (13 nm), signifying that NH3 oxidation is a structure-sensitive reaction. Furthermore, surfaces of Pt particles restructure under NH3 oxidation reaction conditions, influencing effective Pt oxidation states, surface structures (numbers and types of exposed Pt sites), and surface coverages of intermediates leading to the observed hysteresis in rate. These findings demonstrate that Pt particles undergo dynamic structural changes during reaction, influencing their ability to convert NH3 to environmentally benign products in NH3 slip applications. The influence of treatment conditions, support properties, and initial Pd particle size and distribution on the kinetics of nanoparticle sintering were investigated to identify which material properties allow maintaining high dispersion to maximize metal utilization for three way catalysts (TWC) during the conversion of regulated pollutants (CO, hydrocarbons, NOx). Pd was deposited by IWI methods to generate polydiserse particle size distributions, and using colloidal Pd nanoparticle solutions to generate monodisperse size distributions, onto various supports (γ-Al2O3, La-doped Al2O3, ΘΔ-Al2O3) and subjected to aging under oxidative and reductive conditions relevant for TWC operation. The average Pd particle size for all materials increased with treatment time under both reductive and oxidative environments. For samples prepared with IWI (i.e., log normal distribution of Pd particle sizes), reductive aging treatments led to higher sintering rates than oxidative treatments. In contrast, for samples prepared using colloidal Pd solutions (i.e., normal distribution of Pd particle sizes), oxidative aging treatments led to higher sintering rates than reduction treatments. Furthermore, after the same treatment condition and time, samples prepared with IWI resulted in higher average Pd particle sizes. These results indicate that more monodisperse initial Pd particle size distributions lead to lower sintering rates, providing guidance to design of supported metal TWCs with improved metal utilization during their lifetimes. Here, the combination of synthesis approaches to prepare a suite of model (e.g., powder) supported metal catalysts of varying structure and composition, interrogated using site and structural characterizations and steady-state and transient kinetic measurements, along with predictions from theoretical calculations, enabled unraveling the influence of material properties and gas environments that affect metal speciation, structure, and oxidation state in real-world aftertreatment systems that use more complex catalytic architectures (e.g., layered washcoats) and reactor designs (e.g., monoliths). This approach provides insights into the fundamental thermodynamic and kinetic factors influencing metal restructuring and interconversion under realistic conditions encountered in automotive exhaust aftertreatment applications, and the kinetic and mechanistic factors that underlie complex phenomena (e.g., reaction rate hysteresis) from data measured in the absence of hydrodynamic artifacts. The overall approach used in this work enabled development of synthesis-structure-function relationships on various metal supported catalysts for automotive exhaust aftertreatment applications, which can provide guidance for material design and treatment strategies to form and retain desired metal structures throughout the material lifetime, including synthesis, reaction, and regeneration treatments.</p>

<p>To apply Industry 4.0 technologies and accelerate the modernization of continuous pharmaceutical manufacturing, digital twin (DT) and advanced process control (APC) strategies are indispensable. The DT serves as a virtual representation that mirrors the behavior of the physical process system, enabling real-time monitoring and predictive capabilities. Consequently, this facilitates the feasibility of real-time release testing (RTRT) and enhances drug product development and manufacturing efficiency by reducing the need for extensive sampling and testing. Moreover, APC strategies are required to address variations in raw material properties and process uncertainties while ensuring that desired critical quality attributes (CQAs) of in-process materials and final products are maintained. When deviations from quality targets are detected, APC must provide optimal real-time corrective actions, offering better control performance than the traditional open loop-control method. The progress in DT and APC is beneficial in shifting from the paradigm of Quality-by-Test (QbT) to that of Quality-by-Design (QbD) and Quality-by-Control (QbC), which emphasize the importance of process knowledge and real-time information to ensure product quality.</p> <p><br></p> <p>This study focuses on four key elements and their applications in a continuous dry granulation tableting process, including feeding, blending, roll compaction, ribbon milling and tableting unit operations. Firstly, the necessity of a digital infrastructure for data collection and integration is emphasized. An ISA-95-based hierarchical automation framework is implemented for continuous pharmaceutical manufacturing, with each level serving specific purposes related to production, sensing, process control, manufacturing operations, and business planning. Secondly, investigation of process analytical technology (PAT) tools for real-time measurements is highlighted as a prerequisite for effective real-time process management. For instance, the measurement of mass flow rate, a critical process parameter (CPP) in continuous manufacturing, was previously limited to loss-in-weight (LIW) feeders. To overcome this limitation, a novel capacitance-based mass flow sensor, the ECVT sensor, has been integrated into the continuous direct compaction process to capture real-time powder flow rates downstream of the LIW feeders. Additionally, the use of near-infrared (NIR)-based sensor for real-time measurement of ribbon solid fraction in dry granulation processes is explored. Proper spectra selection and pre-processing techniques are employed to transform the spectra into useful real-time information. Thirdly, the development of quantitative models that establish a link between CPPs and CQAs is addressed, enabling effective product design and process control. Mechanistic models and hybrid models are employed to describe the continuous direct compaction (DC) and dry granulation (DG) processes. Finally, applying APC strategies becomes feasible with the aid of real-time measurements and model predictions. Real-time optimization techniques are used to combine measurements and model predictions to infer unmeasured states or mitigate the impact of measurement noise. In this work, the moving horizon estimation-based nonlinear model predictive control (MHE-NMPC) framework is utilized. It leverages the capabilities of MHE for parameter updates and state estimation to enable adaptive models using data from the past time window. Simultaneously, NMPC ensures satisfactory setpoint tracking and disturbance rejection by minimizing the error between the model predictions and setpoint in the future time window. The MHE-NMPC framework has been implemented in the tableting process and demonstrated satisfactory control performance even when plant model mismatch exists. In addition, the application of MHE enables the sensor fusion framework, where at-line measurements and online measurements can be integrated if the past time window length is sufficient. The sensor fusion framework proves to be beneficial in extending the at-line measurement application from just validation to real-time decision-making.</p>

<p>  </p> <p>In the pharmaceutical industry, special care must be taken by companies to guarantee high quality medications that are free from byproducts and impurities. The development process involves various considerations including solvent selection, solubility screening, unit operation selection, environmental, and health impact evaluations. Traditionally, pharmaceutical manufacturing consisted of large, centralized facilities to meet pharmaceutical demands; however, there has been a recent shift toward distributed manufacturing. With distributed manufacturing platforms, rapidly changing supply chain needs can be met regionally in addition to supplying small-volume medications and personalized medicines to hospitals and pharmacies. To produce quality pharmaceuticals, distributed manufacturing platforms should integrate digital design, novel unit operations, and process analytical technology (PAT) tools for quality monitoring and control. In this dissertation, a process design and development framework is proposed and implemented for a small-scale pharmaceutical manufacturing platform: MiniPharm.</p> <p>Various approaches to process design are detailed in this dissertation, which include heuristic-based and digital or simulation-based design. For heuristic-based design, the knowledge of the researchers was utilized to provide unit operation evaluation and screening of process alternatives. In cases when unit operations were highly complex, digital or simulation-based design was utilized to conduct sensitivity analyses and simulation-based design of experiments. With the implementation of simulation-based design, material and time needs were reduced while gaining knowledge about the system. The integration of various unit operations comes with increased understanding of start-up dynamics and operational constraints. What was found to be the most successful approach was the combination of heuristics and digital design to combine researcher knowledge and experience with the information gained from process modeling and simulation to create process alternatives that utilized system dynamics to reach desired process outcomes. </p> <p>Additionally, MiniPharm was used for process model development at the small-scale. Various software packages have been made commercially available that focus on production scale; however, models for small-scale operations are not typically implemented in these packages. Models for unit operations were fit with collected experimental data to estimate model parameters for small-scale synthesis, liquid-liquid extraction, and crystallization unit operations. The models were implemented to better capture the heat and mass transfer of the milli-fluidic scale platform, which consist of unit operations housed within microchannels. MATLAB was utilized for estimation of parameters such as kinetic rate constants and overall mass transfer coefficients. These parameters were used for design space determination and process disturbance simulation. The exploration of the impact of various process parameters on quality attributes helps researchers gain a deeper understanding about the manufacturing process and helps to demonstrate how to control the process. </p> <p>An important aspect of MiniPharm is the process development progress that has been demonstrated. With the construction of a modular and reconfigurable platform, various process alternatives can now be experimentally validated. The integration of unit operations operated at a decreased scale makes MiniPharm an example of process intensification. The implementation of integrated unit operations decreases handling time of intermediates and reduces the overall footprint for manufacturing. Designed to allow for increased flexibility of operation, perfluoroalkoxy alkane (PFA) tubing was used for synthesis and purification. With PFA tubing clean in place procedures can be implemented using continuous solvent flow or the low cost, PFA tubing can be replaced. The modular nature of the platform also allows for the investigation of individual unit operations for performance evaluation. </p> <p>Finally, a novel continuous solvent switch distillation unit operation was designed and constructed along with customized reactor and crystallizers for process alternative screening for the synthesis and purification of two compounds: Diphenhydramine hydrochloride and Lomustine. Diphenhydramine hydrochloride is a low-value, high volume allergy medication commonly found in Benadryl and Lomustine is a high-value, low volume cancer medication used to treat glioblastoma and Hodgkin Lymphoma. The production of the compounds demonstrated the flexibility of the manufacturing platform to produce both a generic and a specialty medication. A versatile platform is needed for distributed manufacturing because of quickly changing supply chain needs. Overall, this dissertation successfully demonstrates the process design, development, and simulation for small-scale manufacturing.</p>

<p dir="ltr">Condition-based maintenance is a proactive maintenance strategy that prevents failures or diminished functionality in process systems through proper monitoring and management of process conditions. Despite being considered a mature maintenance management strategy in various industries, condition-based maintenance remains underutilized in pharmaceutical manufacturing. This situation needs to change, especially as the pharmaceutical industry continues to shift from batch to continuous manufacturing, where the implementation of CBM as a maintenance strategy assumes a greater importance.</p><p dir="ltr">This dissertation focused on addressing the challenges of implementing CBM in a continuous drug product manufacturing system. These challenges stem from the unique aspects of pharmaceutical drug product manufacturing, which includes the peculiar behavior of particulate materials and the evolutionary nature of pharmaceutical process development. The proposed solutions to address these challenges revolve around an innovative framework for the practical development of condition monitoring systems. Overall, this framework enables the incorporation of limited process knowledge in creating condition monitoring systems, which has the desired effect of empowering data-driven machine learning models.</p><p dir="ltr">A key feature of this framework is a formalized method to represent the process condition, which is usually vaguely defined in literature. This representation allows the proper mapping of preexisting condition monitoring systems, and the segmentation of the entire process condition model into smaller modules that have more manageable condition monitoring problems. Because this representation methodology is based on probabilistic graphical modelling, the smaller modules can then be holistically integrated via their probabilistic relationships, allowing the robust operation of the resulting condition monitoring system and the process it monitors.</p><p dir="ltr">Breaking down the process condition model into smaller segments is crucial for introducing novel fault detection capabilities, which enhances model prediction transparency and ensures prediction acceptance by a human operator. In this work, a methodology based on prediction probabilities was introduced for developing condition monitoring systems with novel fault detection capabilities. This approach relies on high-performing machine learning models capable of consistently classifying all the initially known conditions in the fault library with a high degree of certainty. Simplifying the condition monitoring problem through modularization facilitates this, as machine learning models tend to perform better on simpler systems. Performance indices were proposed to evaluate the novel fault detection capabilities of machine learning models, and a formal approach to managing novel faults was introduced.</p><p dir="ltr">Another benefit of modularization is the identification of condition monitoring blind spots. Applying it to the RC led to sensor development projects such as the virtual sensor for measuring granule flowability. This sensor concept was demonstrated successfully by using a data-driven model to predict granule flowability based on size and shape distribution measurements. With proper model selection and feature extraction guided by domain expertise, the resulting sensor achieved the best prediction performance reported in literature for granule flowability.</p><p dir="ltr">As a demonstration exercise in examining newly discovered faults, this work investigated a roll compaction phenomenon that is usually concealed from observation due to equipment design. This phenomenon results in the ribbon splitting along its thickness as it comes out of the rolls. In this work, important aspects of ribbon splitting were elucidated, particularly its predictability based on RC parameters and the composition of the powder blend used to form the ribbon. These findings have positive ramifications for the condition monitoring of the RC, as correspondence with industrial practitioners suggests that a split ribbon is desirable in some cases, despite being generally regarded as undesirable in the limited literature available on the subject.</p><p dir="ltr">Finally, this framework was primarily developed for the pharmaceutical dry granulation line, which consists of particle-based systems with a moderate level of complexity. However, it was also demonstrated to be feasible for the Tennessee Eastman Process (TEP), a more complex liquid-gas process system with a greater number of process faults, variables, and unit operations. Applying the framework resulted in machine learning models that yielded one of the best fault detection performances reported in literature for the TEP, while also introducing additional capabilities not yet normally reported in literature, such as fault diagnosis and novel fault detection.</p>