Search

Crystallisation is an important unit operation for separation and purification in the pharmaceutical and food industrials. Crystallisation affects the product properties of an active pharmaceutical ingredient (API), such as purity, crystallisation size distribution (CSD) and crystal shape. Therefore, the crystallisation process, especially process kinetics should be understood and well controlled to produce the required size, shape and purity crystals. The purpose of this research is to design the trajectory through the phase diagram to control the relative rates of various kinetics mechanisms using (a) digital tools and (b) structured experimental investigations. Paracetamol (PCM) in propanol/water mixture solvent and Compound X offered by Takeda Pharmaceutical Company in methanol/water were used as the model system in this work. Experiments on each process were real-time monitored by Process Analysis Technology (PAT) tools, such as focused beam reflectance measurement (FBRM) for nucleation detection, particle vision and measurement (PVM) for crystal morphology detection (crystal growth and agglomeration) and attenuated total reflectance (ATR)UV/vis spectroscopy for measuring solute concentration (supersaturation which is the driving force of a crystallisation process). In addition, the availability of a mathematical model is necessary to improve a crystallisation process by implementing advanced modelbased optimisation and control of the desired product properties, which can also minimise the use of API and save cost; developing a model can be challenging due to the complexity of the process kinetics and corresponding parameter estimation (kinetics expression form and parameter number).This thesis addresses these issues for the first model system. A sequential design of experiment was applied to isolate the different kinetics in different batches by controlling supersaturation levels and seeds loading; then different kinetic mechanisms can be understood from each experiment and the corresponding mathematical expressions can be determined. PCM in propanol/water system was used as case system to demonstrate this method. Agglomeration and breakage were studied first in saturated solution to separate the growth; the experiments demonstrated agglomeration and breakage simultaneously occur and the sum kernel and symmetric daughter distribution were applied to simulate the agglomeration and breakage. In series of growth experiments, seeds were added to a low supersaturation solution to avoid secondary nucleation. A size-independent power law expression was used to describe the growth. The estimated parameters were fixed for the nucleation study. A higher seeds loading and a higher supersaturation level was applied to study secondary nucleation. Secondary nucleation rates were found to be dependent on the second moment. Unseeded cooling experiments were designed to study the primary nucleation; a power-law expression involving the supersaturation was applied to describe the kinetics. In addition, a model-based design of experiments method was applied to improve the precision of kinetic parameters for unseeded process and seeded process which include four parameters and six parameters, respectively. A D-optimal criterion was used to optimise the temperature cooling profile and sampling points to obtain the most informative process data, resulting in a reduction in the confidence intervals for the parameter estimates. Using the kinetics from the previously described batch experiments, the steady state operation of a cascade mixed-suspension, mixed-product removal (MSMPR) continuous crystallisation process was optimised using the process model to control the nucleation and growth in different stages and hence control the crystal size. To produce crystals with larger mean size, a relative smaller nucleator with higher temperature was optimised to restrict the primary nucleation and secondary nucleation, so that the remaining longer mean residence time was used for crystal growth. The start-up time was also optimised to reduce the time to reach the steady state, for stages starting full of feed solution or empty. An industrial API, Compound X which crystallised from a sequential antisolvent and cooling batch process was studied in the final section. The purpose was to design the sequence of crystallisation stepsto reduce agglomeration in the final product. Crystals with narrower product CSD and reduced agglomeration, were achieved by applying feedback-controlled temperature cycling after the wet milling (by dissolving undesired fine crystals) with an improved sequence of antisolvent and cooling. The approaches provide a systematic framework to study and control process kinetics for CSD design. The methodologies developed include applying sequential design of experiments and model-based design of experiments for estimation of kinetic parameters, optimising the continuous process by separating nucleation and growth and applying wet-milling and temperature cycle to reduce the degree of crystal agglomeration. The trajectory through the phase diagram determines the generation of supersaturation which affects the process kinetics, and in turn determines the consumption of the supersaturation and properties of the crystals such as the mean size.

Automated guided vehicles (AGVs), a type of unmanned moving robots that move along fixed routes or are directed by laser navigation systems, are increasingly used in modern society to improve efficiency and lower the cost of production. A fleet of AGVs operate together to form a fully automatic transport system, which is known as an AGV system. To date, their added value in efficiency improvement and cost reduction has been sufficiently explored via conducting in-depth research on route optimisation, system layout configuration, and traffic control. However, their safe application has not received sufficient attention although the failure of AGVs may significantly impact the operation and efficiency of the entire system. This issue becomes more markable today particularly in the light of the fact that the size of AGV systems is becoming much larger and their operating environment is becoming more complex than ever before. This motivates the research into AGV reliability, availability and maintenance issues in this thesis, which aims to answer the following four fundamental questions: (1) How could AGVs fail? (2) How is the reliability of individual AGVs in the system assessed? (3) How does a failed AGV affect the operation of the other AGVs and the performance of the whole system? (4) How can an optimal maintenance strategy for AGV systems be achieved? In order to answer these questions, the method for identifying the critical subsystems and actions of AGVs is studied first in this thesis. Then based on the research results, mathematical models are developed in Python to simulate AGV systems and assess their performance in different scenarios. In the research of this thesis, Failure Mode, Effects and Criticality Analysis (FMECA) was adopted first to analyse the failure modes and effects of individual AGV subsystems. The interactions of these subsystems were studied via performing Fault Tree Analysis (FTA). Then, a mathematical model was developed to simulate the operation of a single AGV with the aid of Petri Nets (PNs). Since most existing AGV systems in modern industries and warehouses consist of multiple AGVs that operate synchronously to perform specific tasks, it is necessary to investigate the interactions between different AGVs in the same system. To facilitate the research of multi-AGV systems, the model of a three-AGV system with unidirectional paths was considered. In the model, an advanced concept PN, namely Coloured Petri Net (CPN), was creatively used to describe the movements of the AGVs. Attributing to the application of CPN, not only the movements of the AGVs but also the various operation and maintenance activities of the AGV systems (for example, item delivery, corrective maintenance, periodic maintenance, etc.) can be readily simulated. Such a unique technique provides us with an effective tool to investigate larger-scale AGV systems. To investigate the reliability, efficiency and maintenance of dynamic AGV systems which consist of multiple single-load and multi-load AGVs traveling along different bidirectional routes in different missions, an AGV system consisting of 9 stations was simulated using the CPN methods. Moreover, the automatic recycling of failed AGVs is studied as well in order to further reduce human participation in the operation of AGV systems. Finally, the simulation results were used to optimise the design, operation and maintenance of multi-AGV systems with the consideration of the throughputs and corresponding costs of them. The research reported in this thesis contributes to the design, reliability, operation, and maintenance of large-scale AGV systems in the modern and rapidly changing world.

This work approaches the little studied topic of compiler optimisations directed to network applications. It starts by investigating if there exist any fundamental differences between application domains that justify the development and tuning of domain-specific compiler optimisations. It shows an automated approach that is capable of identifying domain-specific workload characterisations and presenting them in a readily interpretable format based on decision trees. The generated workload profiles summarise key resource utilisation issues and enable compiler engineers to address the highlighted bottlenecks. By applying this methodology to data intensive network infrastructure application it shows that data organisation is the key obstacle to overcome in order to achieve high performance. It therefore proposes and evaluates three specialised data transformations (structure splitting, array regrouping, and software caching) against the industrial EEMBC networking benchmarks and real-world data sets. It also demonstrates on one hand that speedups of up to 2.62 can be achieved, but on the other that no single solution performs equally well across different network traffic scenarios. Hence, to address this issue, an adaptive software caching scheme for high frequency route lookup operations is introduced and its effectiveness evaluated one more time against EEMBC networking benchmarks and real-world data sets achieving speedups of up to 3.30 and 2.27. The results clearly demonstrate that adaptive data organisation schemes are necessary to ensure optimal performance under varying network loads. Finally this research addresses another issue introduced by data transformations such as array regrouping and software caching, i.e. the need for static analysis to allow efficient resource allocation. This thesis proposes a static code analyser that allows the automatic resource analysis of source code containing lists and tree structures. The tool applies a combination of amortised analysis and separation logic methodology to real code and is able to evaluate type and resource usage of existing data structures, which can be used to compute global resource consumption values for full data intensive network applications.