The world population is increasing at a very rapid rate while the natural water resources remain constant. During the past decades industrial desalination (reverse osmosis (RO) and multistage flash desalination (MSF)) became a viable, economical, and sustainable source of fresh water throughout the world. In the MSF units, the flashing of seawater involves formation of pure vapour, which flows through a wire mesh demister to remove the entrained brine droplets and then condenses into product water. The study presented in this thesis is motivated by the absence of detailed modelling and analysis of the dynamics of the MSF process and the demister. A detailed dynamic model can be used in design, control, startup/shutdown and troubleshooting. Most of the previous studies on MSF plant focused on model development and presented limited amount of performance data without any validation against plant data. Literature models of the MSF demister are either empirical or semi-empirical. This motivated use of a computational fluid dynamics (CFD) software to design a new demister that will reduce the pressure/temperature drop in the vapour stream without affecting the separation efficiency of brine droplets and allows the optimal design of complete MSF units. Lumped parameter dynamic models were developed for the once through (MSF-OT) and the brine circulation (MSF-BC) processes. The models were coded using the gPROMS modelling program. The model predictions for both MSF-OT and MSF-BC in steady state and dynamic conditions showed good agreement against data from existing MSF plants with an error less than 1.5%. Dynamic analysis was made to study plant performance upon making step variations in system manipulated variables and identify stable operating regimes. New stable operating regimes were reached upon changing the cooling water flow rate by + 15% and increasing the recycle brine flow rate by 15% and decreasing it by 7%. This was not the case for the steam temperature where its variation was limited to + 2-3 %. This behavior is consistent with the actual plant data. The FLUENT software was used to model the MSF demister using different combinations of Eulerian and Lagrangian approaches to model the vapour and the brine droplets. This provided the open literature with novel and new methodologies for design and simulation of the MSF demister using CFD. A new demister design was made upon varying the wire diameter. This led to an efficient design with low pressure drop and high separation efficiency. This design was used in the MSF/gPROMS model to predict its effect on the heat transfer area. The new design provided reductions of 3-39% in the condenser heat transfer area without affecting dynamic performance. Since the tubing system accounts for almost 70% of the capital cost, then this would reduce the plant capital cost and product unit cost. The modelling approach presented in this thesis enables design of thermal desalination units to determine optimal heat transfer area and optimized operating conditions.

This thesis investigates some important problems in the supply chain management (SCM) for the process industry to fill the gap in the literature work, covering production planning and scheduling, production, distribution planning under uncertainty, multiobjective supply chain optimisation and water resources management in the water supply chain planning. To solve these problems, models and solution approaches are developed using mathematical programming, especially mixed-integer linear programming (MILP), techniques. First, the medium-term planning of continuous multiproduct plants with sequence-dependent changeovers is addressed. An MILP model is developed using Travelling Salesman Problem (TSP) classic formulation. A rolling horizon approach is also proposed for large instances. Compared with several literature models, the proposed models and approaches show significant computational advantage. Then, the short-term scheduling of batch multiproduct plants is considered. TSP-based formulation is adapted to model the sequence-dependent changeovers between product groups. An edible-oil deodoriser case study is investigated. Later, the proposed TSP-based formulation is incorporated into the supply chain planning with sequence-dependent changeovers and demand elasticity of price. Model predictive control (MPC) is applied to the production, distribution and inventory planning of supply chains under demand uncertainty. A multiobjective optimisation problem for the production, distribution and capacity planning of a global supply chain of agrochemicals is also addressed, considering cost, responsiveness and customer service level as objectives simultaneously. Both ε- constraint method and lexicographic minimax method are used to find the Pareto-optimal solutions Finally, the integrated water resources management in the water supply chain management is addressed, considering desalinated water, wastewater and reclaimed water, simultaneously. The optimal production, distribution and storage systems are determined by the proposed MILP model. Real cases of two Greek islands are studied.

The objective of this work is to study experimentally and theoretically novel multiphase microreactors and characterize them in relation to hydrodynamics and mass transfer, in order to evaluate, understand and improve their performance. In order to achieve this CO2 absorption in sodium hydroxide and amine solutions an example of a fast gas-liquid reaction has been investigated in a single microstructured metallic mesh reactor, CRL reactor, PTFE single channel membrane reactor and the silicon nitride mesh reactor. CO2 absorption in sodium hydroxide solution was initially studied experimentally and theoretically in a metal microstructured mesh reactor. The differential mass balances to describe the concentration profiles of components in the three domains (gas/membrane/liquid), were solved with Comsol Multiphysics (modeling software for finite element analysis of partial differential equations). The model indicated that the carbon dioxide is consumed within few microns from the gas – liquid interface, and the dominant resistance for mass transfer is located in the mesh because it is wetted by the liquid reactant. In order to overcome the limitation of the extra resistance to the mass transfer in the metallic mesh, PTFE membranes were used in the single channel reactor, which are considered as hydrophobic to aqueous solutions of NaOH and amines. Monoethanolamine solution (MEA) absorbed more CO2 than diethanolamine (DEA) since the reaction rate constant for MEA is higher than DEA. 8 channel (PTFE) microreactor showed much higher CO2 removal efficiency than the metallic mesh microreactor. Furthermore the model indicated partial-wetting of the PTFE membrane when NaOH solution was used as an absorbent. In order to enhance mass transfer staggered herringbones were used on the floor of the liquid side of the single channel PTFE microreactor. No enhancement of mass transfer was observed with the use of staggered herringbones. A possible reason for that is that a limit for the fast second-order reaction is reached for enhancement and that the apparent reaction rate is independent from mass transfer for our case, or that the herringbones are far away from the reaction zone and cannot create the appropriate stirring for enhancement. In order to increase throughput, carbon dioxide absorption in sodium hydroxide solution was performed in the metallic mesh ‘scale-out’ reactor (with 4 meshes). CO2 removal efficiency for the ‘scale-out’ reactor was significantly lower than the single mesh reactor, which is probably due to breakthrough of liquid in the gas phase (stagnant liquid) or uneven flow distribution in each plate of the ‘scale-out’ reactor. Finally a silicon nitride mesh reactor developed by Bayer Technology Services and FluXXion was used for CO2 absorption in aqueous solutions of NaOH and DEA. The silicon nitride mesh reactor showed better performance than the PTFE single channel reactor, the metallic 8 channel reactor and the CRL mesh reactor when NaOH was used, due to the very thin membrane of 1 μm thickness, which makes the resistance to mass transfer very small.

Many of the companies involved in powder processing are seeking to improve their economic performance via improved control of the physical properties of their products. Pharmaceutical manufacturers, for example, are considering the potential benefits of moving from batch to continuous operations to improve consistency. Furthermore, in recent years interest in the application of ultrasound to crystallization has received a significant impetus with the increased requirement to prepare complex chemical entities to very exacting standards. Literature reports about sonocrystallization indicate several interesting benefits, including superior crystal habit, purer and easily handled product, narrowed crystal size distribution, and prolific nucleation. The sonocrystallization literature, however, mainly focuses on batch crystallization operations, with less information relating to continuous crystallization processes. The present thesis concerns an evaluation of the use of ultrasonic technology during batch and continuous crystallization of adipic acid crystals and compares it with common industrial operations for product particle engineering. Firstly, the effect of a continuous sonication on a continuous crystallization process has been investigated. Cooling crystallization of adipic acid from aqueous solution is the selected case study. Analogous experiments have been carried out both under silent and continuous insonation regimes in order to investigate the effects of sonication on the time required to reach the steady state, particle size distribution (PSD), solids yield, and crystal habit. The results reveal that under continuous ultrasonic irradiation the steady state particle size distribution is achieved after shorter times than in silent continuous crystallization experiments. Continuous crystallization with ultrasonic irradiation results in significantly smaller crystal sizes, reduced agglomeration and an improved habit of crystals with highly reproducible product characteristics. Furthermore, the product yield is increased. The crystallization kinetics, focused on both the nucleation and growth rates, has been determined using the continuous Mixed-Suspension Mixed-Product-Removal (MSMPR) crystallizer model. Kinetics have been extracted sequentially from experimental data relating the particle size distribution, using the Population Balance Equation (PBE) in terms of moments, and evaluating the effect of mean residence time, supersaturation at steady state and ultrasonic power amplitude on the growth and nucleation rates. Application of ultrasound resulted in the most pronounced effects on the nucleation process, with rates increased by one order of magnitude with the results without sonication. The Mydlarz and Jones three parameters (MJ3) model for size dependent growth fits the non-linearity in the insonated experimental population density data for crystal sizes up to 10 μm. For larger sizes, whereas the population density plot is linear, the growth rate is deemed to be independent of crystal size. Further analysis of the kinetics of nucleation and growth at steady state, in continuous crystallization under continuous insonation, has been developed using the commercially available software package PARSIVAL based on the fully adaptive Galerkin h-p method. The population balance has been modeled with secondary nucleation and a growth rate depending on both supersaturation and particle size, according to the MJ3 model. Numerically derived results from the population balance modelled with PARSIVAL are in reasonable agreement with experimental observations, in terms of population density values. The use of ultrasound in the particle engineering of micron scale adipic acid crystals has been implemented by evaluating its size reducing power compared with the product of industrially established milling processes. Specifically, the steady state particles characteristics of a continuous operation under ultrasonic irradiation and the final product characteristics of batch cooling crystallization under continuous ultrasonic have been compared with hammer milling, micronization, and High Shear Wet Milling (HSWM). Ultrasound applied to batch and continuous crystallization produces particle sizes comparable with those from micronization. Continuous insonation during batch crystallization provides spherical particles, with regular surface roughness and highly reproducible results. The use of ultrasound in the crystal product engineering has been addressed to the achievement of large particles by generating seeds crystals in-situ by means of controlled primary nucleation; the results were compared with the product of conventionally seeded crystallization. Seeded batch cooling crystallization of adipic acid from aqueous solution has been investigated to determine the effects of the method used to produce seeds and optimize seeding load, cooling rate, initial concentration, and supersaturation at seeding to achieve large particle sizes and mono-modal crystal size distribution. Finally, the analysis of final particle size distributions and particle surface characteristics has demonstrated that seed crystals generated in-situ by ultrasound offer advantages comparable with conventionally inoculated seeds, eliminating the need of previous preparation and selection of seeds and the drawbacks associated with seed handling and selection of a suitable inoculation time.

Technological advancements over the last century have lead large and continuous growth in the output of plastic materials. This exponential growth has created public concern over the environmental impact caused by the polymeric waste produced. These have acted as driving forces for a lot of current research aimed at the development of plastic recycle processes. As a result, the conversion of plastic waste to useful products is gaining increasing attention. The aim of this work was to study aspects of polymer catalytic degradation using zeolite based catalysts. More specifically the study focused on identifying the role of the external catalytic surface on overall polymer decomposition reactions, the reusability of the catalysts as well as temperature and acidity effects. The first stage of this investigation aimed to explore the premise behind the assumption that polymer catalytic degradation takes place initially on the external catalytic surface by selectively poisoning the external sites of a zeolite catalyst (ZSM-5). Degradation results in a semi-batch reactor as well as thermogravimetric analysis demonstrated that the activity of poisoned catalyst samples was indeed lower than that of fresh catalyst. The next stage of the study involved an investigation of the extent of catalytic reusability of four zeolite catalysts - HZSM-5, USY and two commercial cracking catalysts containing 20 % and 40 % USY respectively. While the performance of US-Y showed deterioration with each cycle, ZSM-5 and both commercial cracking catalysts retained consistent levels of activity that enabled full polymer conversion in each cycle. Finally, the temperature effect on catalytic reactions was studied as well as the effect of catalyst acidity. While temperature effects were not conclusive regarding selectivity towards gas or liquid products prompting the suggestion of further work using a continuous flow reactor system, the formation of liquid products showed a maximum with the acidity content.

Adsorption in zeolites is widely and increasingly used in many industrial processes. For the design of new processes and improvement of the performance of existing ones, basic adsorption data is needed but due to the difficulties of experimentation it is lacking. In this thesis, different methods have been used to calculate the adsorption isotherms of benzene, methane, ethane and CO, mixtures as well as Propane, i-butane and n-butane and their binary and ternary mixtures in zeolites. Firstly, Lattice Model has been used to calculate benzene adsorption isotherms in silicalite zeolite whose experimental adsorption isotherms exhibit unusual features. The zeolite is modelled as two types of quasi one- dimensional pores. The lattice model has dimer states to represent molecules lying in extended states with the aromatic ring on average parallel to the pore wall, the monomer state to represent molecules standing perpendicular to the principle axis of the pores. Vacant sites or holes allow for incomplete filling of the lattice sites. For a wide range of interaction parameters the model gives steps in the adsorption isotherms similar to those observed experimentally for benzene adsorption in silicalite. The model attributes the experimentally observed steps in the level of adsorption with rising pressure, to re-orientational transitions amongst molecules in the adsorbed phase. Secondly Conventional Grand Canonical Monte Carlo techniques have been used to calculate methane, ethane and co, binary mixture adsorption isotherms in silicalite as well as propane, i-butane and n-butane equimolar binary and ternary mixture adsorption in silicalite. In the last part of the thesis, studies have made of lattice models of CH, I C,H, and CO, adsorption in silicalite and the results compared with the Monte Carlo data. One dimensional lattice models have also been used to calculate the binary and ternary mixture adsorption of propane, i-butane and n-butane. The isotherms were compared with the Monte Carlo results.

The main reasons behind the success of the petrochemicals industry are not only the vast array of products that it provides - considered vital to our daily functions - but also the added value that it brings to the crude oil barrel price, making it a reliable venture for any concerned party. However, the industry is now faced with a fluctuating market and an unstable economy, which makes it imperative to find a more abundant and sustainable feedstock. Of all petrochemical derivatives, polymers (and their related industries) occupy the major share, and this makes the plastics industry a growing sector in terms of processing and conversion. Both virgin and waste plastics represent an attractive source of energy and product recovery. The main objective of this work was to investigate the thermo-chemical treatment (TCT) of polymers at different scales, and the reactors studied ranged from micro laboratory scale to industrial units suitable for covering large market demands. Within this framework, the degressive behaviour of polyolefin polymers (three virgin grades and two recyclate ones) was investigated alongside the products yielded (gases (C1-C4), liquids (non-aromatic C5-C10), aromatics (single ring structures) and waxes (> C11). This was achieved in a micro scale isothermal pyrolysis process, using 15 mg in a laboratory thermogravimetric analyser covering the temperature range of 500-600°C. The analysis led to the development of an nth order novel model on the basis of lumped products yielded by pyrolysis. The degradation mechanism was used to develop the mathematical breakdown of the primary, secondary and tertiary reactions. The model developed predicts the yield of the four different products and the polymer residual fraction at any operating condition proving to be a useful tool for reactor design and simulation, where the production of a specific chemical at a certain operating condition is paramount. In addition, laboratory scale isothermal pyrolysis experiments on end of life tyres (ELTs) were also conducted. This was achieved as a means to demonstrate the application of the concept previously applied to the polyolefins. A thermal cracking (degradation) scheme was proposed based on the global yielded products, which were lumped into four categories, namely gases (C1-C4), liquids (non-aromatic C5-C10), single ring aromatics (C5-C10), and char. The depolymerization kinetics (from primary, secondary and tertiary reactions) evaluation showed a high match with the experimental results obtained in this work. Finally, a life cycle assessment (LCA) was conducted for three integrated scenarios that reflect the current (2012) treatment of waste plastics in the Greater London area. The scenarios studied utilised a fraction of the polymers treated as a feedstock for two industrial scale TCT technologies; namely a low-temperature pyrolysis reactor that works using BP® technology and a hydrocracking unit that utilises the Veba-Combi Cracking (VCC®) concept. The scenarios studied also include transfer stations, a dry materials recovery facility (MRF) and a combined heat and power (CHP) incineration unit. The energy recovered via the different processes studied, as well as the chemicals and petrochemicals recovered, were all considered as credits in the LCA conducted. Chemicals obtained by the TCT units are very valuable and can replace refinery cuts and petrochemicals (e.g. syncrude (crude oil), naphtha, heavy (waxes) fraction (comparable to atmospheric residue), gases (C3 and C4) refinery cuts, etc.). This led to a technoeconomic analysis of the three integrated scenarios in order to assess the overall profitability. The analysis included capital, operating and maintenance costs, gate fees, transportation costs and corporation tax. The eligibility for governmental incentives (i.e. renewable obligation certificates (ROCs), levy exemption certificates (LECs) and packaging recovery notes (PRNs)) was also considered. The results obtained from the work carried out and reported in this thesis point towards ideal strategies for the treatment of polymers within the urban environment. It also provides a detailed understanding of potential products from polymers introduced to TCT units. This also aids the optimum recovery of petrochemicals, chemicals and energy from different TCT processes, and could help the UK Government in meeting its energy policy targets. It can also contribute to the energy security through diversification of supply. Finally, it provides a perspective on the integration between the crude oil upstream industry and different petrochemical complexes and oil refineries, through the use of different TCT units to increase the production of petrochemicals in existing plants.

The thesis presents investigations on two-phase gas-liquid microstructured reactors operating in Taylor flow and the dependence of reactor performance on design parameters. Literature review revealed that flow patterns in microchannels are affected not only by channel dimension, fluids flowrates and surface tension, but also by wall wettability and gas inlet size. A universal flow regime map does not seem to exist. The hydrodynamic parameters of Taylor flow were investigated both by Computational Fluid Dynamics simulations and experiments in microstructures with sizes 0.3mm – 1mm and various inlet configurations such as T- and Y- junctions fabricated in-house. The same parameters that influence flow patterns and their transitions were also found to affect Taylor bubble sizes. To account for the effect of inlet conditions, correlations were developed for predicting bubble/slug size in the T- and Y- inlet geometries that were used subsequently. Mass transfer with and without chemical reaction was investigated numerically in Taylor flow microreactors using CO2 physical absorption into water or chemical absorption into NaOH aqueous solution. Chemical absorption was enhanced by a factor of 3-18 over physical absorption. With reaction present, the reactor performance depended mainly on the gas-liquid interfacial area, while mixing within the phases was only important in physical absorption. This agreed with the experimental results of a similar reaction system, which showed that bifurcating main channels, where new interfaces are generated, significantly improved reaction conversion while meandering channels that enhance liquid mixing had little impact. Finally, the performance of a Taylor flow microreactor was evaluated for an industrial fast gas-liquid reaction of CO2 absorption from fuel gas into amine solutions. The Taylor flow microreactor offered the largest specific area and the smallest reactor volume compared to other microreactor types. However, in order to meet absorption specifications for the case considered multistage absorption would have been necessary.

In our work, we employed molecular dynamics and Monte Carlo (MC) simulations to investigate the supercritical phase of carbon dioxide near its critical point. Three systems have been studied. The pure carbon dioxide, mixture methane + carbon dioxide at infinite dilution of supercritical carbon dioxide and water + carbon dioxide at infinite dilution of supercritical carbon dioxide. The usage of molecular simulation methods in supercritical region gave us a distinct advantage of knowing the microstructure of the systems in a qualitative and quantitative way. The Kirkwood-Buff theory, which predicts the influence of the solvent on the solute, enabled us to predict thermodynamic properties of supercritical phase and compare them with experimental values. We have examined the density effect on structure of the pure carbon dioxide and its solutions along its critical isotherm 4 K above its critical point. We focused our research and we present results for two basic sections, A. Equilibrium and transport properties, namely Volumetric properties; Average configurational energy; Isothermal compressibility; Diffusivity; and the Isochoric heat capacity B. Solution structures at infinite solutions, namely Radial distribution function; and Coordination number We discuss the outcomes based on the density inhomogeneities of the solvent and critical fluctuations, which are maximised at the critical point. We found that the addition of methane to supercritical carbon dioxide increases the volume of the solution and a cavitation is formed around it. On the hand, the addition of water gives a cluster around it in local structure and decrease the volume of solution. We report results also of the diffusion coefficients for the pure carbon dioxide and the mixtures in this study, which it shows an anomalous decrease close to the critical point of the pure carbon dioxide. It is a general conclusion for all the properties we have studied that the density dependence along the isotherm is maximised at densities close to the critical one. Further, the usage of both molecular dynamics and Monte Carlo in supercritical regions validates the extension of the techniques in the supercritical region and reveals their limitations.

No description available.