With Industrial Internet, businesses can pool their resources to acquire large amounts of data that can then be used in machine learning tasks. Despite the potential to speed up training and deployment and improve decision-making through data-sharing, rising privacy concerns are slowing the spread of such technologies. As businesses are naturally protective of their data, this poses a barrier to interoperability. While previous research has focused on privacy-preserving methods, existing works typically consider data that is averaged or randomly sampled by all contributors rather than selecting data that are best suited for a specific downstream learning task. In response to the dearth of efficient data-sharing methods for diverse machine learning tasks in the Industrial Internet, this work presents an end-to end working demonstration of a search engine prototype built on PriED, a task-driven data-sharing approach that enhances the performance of supervised learning by judiciously fusing shared and local participant data. / Master of Science / My work focuses on PriED - a data sharing framework that enhances machine learning performance while also preserving user data privacy. In particular, I have built a working demonstration of a search engine that leverages the PriED framework and allows users to collaborate with their data without compromising their data privacy.

The widespread popularity of Federated Learning (FL) has led researchers to delve into its various facets, primarily focusing on personalization, fair resource allocation, privacy, and global optimization, with less attention puts towards the crucial aspect of ensuring efficient and cost-optimized communication between the FL server and its agents. A major challenge in achieving successful model training and inference on distributed edge devices lies in optimizing communication costs amid resource constraints, such as limited bandwidth, and selecting efficient agents. In resource-limited FL scenarios, where agents often rely on unstable networks, the transmission of large model weights can substantially degrade model accuracy and increase communication latency between the FL server and agents. Addressing this challenge, we propose a novel strategy that integrates a knowledge distillation technique with a Particle Swarm Optimization (PSO)-based FL method. This approach focuses on transmitting model scores instead of weights, significantly reducing communication overhead and enhancing model accuracy in unstable environments. Our method, with potential applications in smart city services and industrial IoT, marks a significant step forward in reducing network communication costs and mitigating accuracy loss, thereby optimizing the communication efficiency between the FL server and its agents.

In the kingdom of Saudi Arabia (KSA), the current seawater desalination technologies are completely relying on burning unsustainable crude oil as their main energy driver. Saudi authorities have realized that the KSA is not going to be protected from the future global energy crisis and have started to set up a plan to diversify its energy resources. Membrane Distillation (MD) has emerged as an attractive alternative desalination process. It combines advantages from both thermal and membrane-based technologies and holds the potential of being a cost-effective separation process that can utilize low-grade waste heat or renewable energy. MD has four different configurations; among them is Air Gap Membrane Distillation (AGMD) which is the second most commonly tested and the most commercially available pilot-plant design. AGMD has a stagnant thin layer of air between the membrane and the condensation surface. This layer introduces a mass transfer resistance that makes the process require a large membrane surface area if a large quantity of fresh water is desired. This dissertation reports on experimental and theoretical work conducted to enhance the AGMD flux by removing non-condensable gases from the module and replacing it with either vacuum, liquid water or porous materials. At first, a mathematical model for AGMD was developed and validated experimentally to create a baseline for improvements that could be achieved after the removal of non-condensable gases. The mathematical model was then modified to simulate the process under vacuum where it showed a flux enhancement that reached 286%. The Water Gap Membrane Distillation (WGMD) configuration improved the flux by almost the same percentage. Since enhancing the flux is expected to increase temperature polarization effects, a theoretical study was conducted on the effect of temperature polarization in a Vacuum Membrane Distillation (VMD) configuration. The study showed that the effect of temperature polarization at small temperature difference (3-7) degree Celsius between the bulk feed and coolant temperatures is significantly high. This may indicate the importance of mitigating the effect of temperature polarization in large scale modules operating at small temperature difference across the membrane. The dissertation concluded with some recommendations for future work.

Accurate and reliable vapour-liquid equilibrium (VLE) and liquid-liquid equilibrium (LLE) data are the key to a successful design and simulation of most important industrial separation processes (traditional distillation, extractive and azeotropic distillation). This work focuses on measurement of new phase equilibrium data for systems comprising of propan-1-ol, water and diisopropyl ether which are of important use in the petrochemical industry. In addition, an investigation of phase equilibrium behavior for systems of interest constituted by solvents and high added-value oxygenated compounds deriving from lignocelluloses biomasses (bio-fuels) was conducted at the Ecole des Mines de Paris CEP/TEP laboratories (France).Various data bases such as Science Direct, ACS publications and Dortmund Data Bank (DDB, 2009) were used to confirm that no literature data is available for these systems. The VLE data measurements for the system of propan-1ol + water and propan-1ol + diisopropyl ether (DIPE) ( 333.15, 353.15 and 373.15 K ) were carried out using a dynamic still of Lilwanth (2011), with a test system (ethanol + cyclohexane at 40 kPa) undertaken prior measurements to confirm the accuracy of the method and apparatus.The phase equilibrium (VLE and LLE) behaviours for furan + n-hexane and furan + Methylbenzene, furfural + n-hexane and furan + water were determined at 101.3 kPa. The atmospheric dynamic ebulliometry was used to measure VLE systems at 101.3 kPa. A set of LLE data for furfural + n-hexane and furan + water systems were obtained using a static analytical method, with a newly commissioned LLE apparatus. Furfural + n-hexane system was compared used as test system, to verify the reliability of the new equipment. The NRTL model was used to correlate the LLE data, with Cox- Herington model used to predict the entire LLE curve for furfural+ n-hexane system. The experimental VLE data were correlated using the combined y − y method. The vapour phase non idealities were described using the methods from Nothnagel et al. (1973), Hayden and O’Connell (1975) and the Peng-Robinson (1976) model. The activity coefficients were correlated using the NRTL model of Renon and Prausnitz (1968) and the modified UNIQUAC model of Abrams and Prausnitz (1976). A propan-1-ol dehydration process was simulated using Aspen to illustrate the use and importance of thermodynamic models in industrial process design and simulation. The model used in the simulation was validated with measured VLE and literature LLE data. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2012.

Student Number : 9510423G PhD Thesis School of Chemical and Metallurgical Engineering Faculty of Engineering and the Built Environment / Techniques for the design and analysis of simple column separations are well established. Shortcut design techniques have been employed in the initial design of these “traditional” distillation systems for a number of years and these columns are well understood. However, few currently available techniques are useful in the design of novel or complex configurations. The techniques that are available tend to be configuration specific. An all inclusive or universal, design and analysis tool, that can be applied to any and all configurations, is required. Tapp et al (2004) introduced Column Profile Maps (CPMS) as a means of addressing this issue. These are maps of composition profiles for column sections with defined net-molar-flow and reflux ratio. It is suggested that by producing CPMs for a configuration a designer can essentially superimpose these, determine feasible operating profiles and hence column operating parameters. In this thesis we show that this technique can be used to, not only produce quick and easy complex column designs but gain a comprehensive understanding of the steady-state operation of these arrangements. We demonstrate this analytical potential first by application of the CPM technique to the two-product feed distribution problem. It is shown that feed distribution can lower the minimum required reflux ratio for non-sharp separations and in some cases produce feasible separations from previously infeasible product specifications. A composition region of operation for all distributed feed policies is also found. The potential for detailed analysis, design and optimisation of complex configurations is demonstrated via application of the CPM procedure to the fully thermally coupled (Petlyuk) distillation column at both sharp and non-sharp split conditions. A detailed design methodology for any configuration results from this. It is found that the Petlyuk column can operate under five possible bulk/net flow conditions and that very interesting and counter-intuitive net-molar-flows are possible. A feasible column parameter region equivalent to the optimality region (Halvorsen and Skogestad, 2001) is found for zeotropic systems. Importantly a minimum reflux condition for the Petlyuk column is found. This condition can be applied to all zeotropic systems for all product specifications. It is also demonstrated that the CPM technique can be used for design optimisation of separation systems.

La distillation dite « charentaise » est une méthode de distillation discontinue permettant la production d’eaux-de-vie de cognac à partir de vin. La distillation est effectuée dans des alambics en cuivre avec un chauffage à feu nu qui peut être favorable à la génération de composés volatils. Or, les précurseurs et les mécanismes de réactions formant ces composés, pendant la distillation, restent mal connus. La première partie de l’étude consiste à identifier les composés volatils formés au cours de la distillation charentaise. Le bilan matière effectué sur de nombreux composés lors du procédé de distillation a révélé que 2 esters, 3 aldéhydes, 3 terpènes et 12 norisoprénoïdes étaient générés. Par la suite, deux distillations utilisant un mode de chauffe différent (feu nu et vapeur) ont été menées sur un alambic pilote. Le but était d’évaluer l’impact du mode de chauffe sur la génération en composés volatils. L’étude a montré que le mode de chauffe a peu d’effet sur la génération en composés volatils. De plus, les composés formés sont similaires à ceux formés lors de la distillation en alambic traditionnel. Enfin, le troisième axe de l’étude porte sur la caractérisation de l’hydrolyse acide de l’α- terpenyl-O-β-glucopyranoside, précurseur impliqué dans la formation de l’α-terpineol, identifié comme étant généré lors de la distillation charentaise ainsi qu’en alambic pilote. Le suivi de la dégradation dans un réacteur hermétiquement fermé de l’α-terpenyl-O-β-glucopyranoside et dans des conditions représentatives du vin durant la distillation montre l’hydrolyse pour former l’α- terpineol, le trans-terpin et son isomère. Cette étude a révélé que l’hydrolyse du précurseur est favorisée en milieu aqueux et suit une cinétique d’ordre 1 / The « charentaise » distillation is a batch distillation method that allows the production of cognac spirits from wine. The distillation is performed in copper alembics through a direct open flame heating that could favor the formation of volatile compounds. The first part of this study consists in identifying the volatile compounds formed during the distillation of cognac spirits. The mass balance performed on volatile compounds revealed that 2 esters, 3 aldehydes, 3 terpenes and 12 norisoprenoids were generated. Thereafter, two distillations using a different heating mode (direct open flame and steam) were conducted on a small-scale alembic. The goal was to assess the impact of the heating mode on the formation of volatile compounds. The study showed that the mode of heating has little effect on volatile compounds’ generation. Moreover, the compounds formed are similar to the ones during the traditional distillation of cognac spirit. Finally, the third part of the study focuses on the characterization of the acid hydrolysis of α-terpenyl-O-β-glucopyranoside which is the suspected precursor to be involved in the formation of α-terpineol, identified as generated during the charentaise distillation and in small-scale distillations. α-terpenyl-O-β- glucopyranoside was placed in two representative model solutions corresponding to the initial wine and the stillage. Both solutions were exposed to 100 °C in a closed reactor system. Results showed that the hydrolysis of the precurseur formed α- terpineol, trans-terpin and its isomere that seems to be 4-(2-hydroxypropan-2-yl)-1- methylcyclohexan-2-ol). Data also revealed that the hydrolysis of the precursor follows a first order reaction model ant that an aqueous media promotes the formation of trans-terpin.

L’eau est une ressource qui va manquer dans les prochaines années, pourtant l’eau couvre ¾ de la planète et environ 97% de l’eau sur Terre est de l’eau salée. Du fait de sa nette abondance, dessaler l’eau de mer, ou les eaux saumâtres, est une solution présentant un immense attrait. C’est dans ce contexte global que s’inscrivent ces travaux. Le but de cette thèse est de modéliser des installations de dessalement d’eau de mer, basées sur la distillation multiple-effets (MED), soumises à des régimes transitoires. Ce procédé permet d’assurer à la fois une grande flexibilité aux variations de charge et des performances énergétiques élevées. Dans le cadre de cette étude, un couplage d’une usine MED avec une source d’énergie renouvelable, donc intermittente, est envisagé. Pour étudier le potentiel et la faisabilité d’un tel couplage, un modèle dynamique représentatif des installations MED est présenté. Ce modèle permet le suivi et l’étude de toutes les variables agissant sur les performances globales de l’installation ainsi que l’identification des variables critiques pouvant nuire au fonctionnement de ces installations en régime transitoire. Ce modèle a ensuite été couplé à un modèle d’une centrale solaire à concentration afin de conclure sur le potentiel d’une telle association. Pour pouvoir assurer un fonctionnement optimal de l’installation MED en régime transitoire, une stratégie de régulation est présentée et intégrée / Demand for water will continue to increase over the next decades, leading to more stress on limited resources and ecosystems. However, ¾ of the Earth is water and 97% of it is seawater. Therefore, it is highly interesting to desalinate seawater. It is within this framework that this work was carried out. The goal of this PhD was to model desalination plants, based on the Multiple-Effect Distillation (MED) process, when they are operated under dynamic conditions. The MED plants are mature desalination processes with high energetic performances and are very flexible to load variations. For this study, a coupling between a MED plant and a renewable energy source, therefore transient, was assumed. To study the potential and the feasibility of such an association, a dynamic model representative of the MED plants was developed. With this, it is now possible to study the evolution of all the variables that have a consequence on the plant’s performances as well as to identify which variables are the most critical to the dynamic operation of the MED plant. This model was then combined with a concentrated solar power plant to conclude about the potential of this coupling. To assure an optimal operation of the MED plant under dynamic conditions, a regulation strategy was developed and implemented

Distillation systems consist of one or more distillation columns, in which a mixture is separated into higher-value products, and a heat exchanger network (HEN) that recovers and reuses heat within the system. For example, crude oil distillation systems comprise crude oil distillation units (CDU), in which crude oil is distilled into products for downstream processing, a HEN and a furnace. Heat-integrated distillation systems present complex interactions between the distillation columns and HEN. These interactions, together with the many degrees of freedom and process constraints, make it challenging to retrofit or modify the operating conditions of existing distillation processes to accommodate changes in process operating conditions. Retrofit designs aim to re-use existing equipment when process objectives change, for example to increase throughput, improve product quality, or reduce energy consumption or environmental impact. To achieve these retrofit objectives, operational, structural and/or flowsheet modifications to the overall system (distillation columns and HEN) may be considered, subject to specifications and system constraints. This work proposes an optimisation-based approach to retrofit design for the capacity expansion of heat-integrated distillation systems, with a particular focus on crude oil distillation systems. Existing retrofit approaches found in the open research literature consider operational optimisation, replacing column internals, adding preflash or prefractionation units and HEN retrofit to increase the capacity of existing systems. Constraints considered usually relate to the distillation column hydraulic limits, product quality specifications and heat exchanger performance (e.g. minimum temperature approach and, pressure drop). However, no existing methodologies consider these possible modifications simultaneously; thus, beneficial interactions between flowsheet modifications, operational changes, heat integration and equipment modifications may be missed. In this work, retrofit design solutions for crude oil distillation are developed using a stochastic optimisation framework implemented in MATLAB to optimise the system operating parameters and to propose flowsheet, column and HEN modifications. Within the framework, the optimiser can propose addition of a preflash unit, modifications to the CDU internals and changes to its operating conditions; the separation system is then simulated using Aspen HYSYS (via the MATLAB interface) and the hydraulic performance of the column is analysed using published hydraulic correlations. The optimiser also proposes modifications to the HEN (i.e. installed heat transfer area, HEN structure and operating conditions), which is then simulated to evaluate heating and cooling utility demand. Either simulated annealing and global search optimisation algorithms are applied to identify the optimal design and operating conditions that meet the production requirements and product specifications. Industrially relevant case studies demonstrate the effectiveness and benefits of using the proposed retrofit approach. The case studies illustrate that combined structural and operational modifications can be effectively and systematically identified to debottleneck an existing crude oil distillation system with a relatively short payback time, while simultaneously reducing energy consumption per barrel of crude oil processed.

Este trabalho aborda a predição da eficiência de pratos valvulados de dois passes, de escoamento cruzado com downcomers de uma coluna de destilação industrial, através do uso de diferentes modelos com e sem o uso do método Stagnant Regions Model proposto por Lockett em 1986. Contrariando o esperado, poucos artigos foram publicados sobre as principais características de pratos valvulados e bem menos sobre eficiência de pratos valvulados de dois passes. Os resultados encontrados sugerem que a correlação de O'Connell, o modelo de Chan e Fair aplicado aos pratos valvulados de dois passes convertidos para um passe e o modelo de Chan e Fair com uso do método SRM são os que melhor predizem a eficiência de pratos valvulados de dois passes, de escoamento cruzado com downcomers, no caso estudado.Em consonância com os estudos prévios, a correlação de O'Connell subestima o valor da eficiência, mostrando-se levemente conservativa, enquanto as predições pelo uso do modelo de Chan e Fair e do modelo modificado de Chan e Fair com uso do comprimento equivalente proposto por Klemola e Ilme superestimam o valor da eficiência, mostrando-se levemente não-conservativas. / This work approaches the prediction of the efficiency of two passes, crossflow valve trays with downcomers of an industrial distillation column, through the use of different models with and without the use the Stagnant Regions Model, method proposed by Lockett in 1986.Contradicting what it could be expected, few papers were published about the main characteristics of valve trays and fewer about efficiency of two passes valve trays. The results suggest that the O'Connell's correlation, the model of Chan and Fair applied to the two passes valve trays converted for one pass valve tray and the model of Chan and Fair with SRM method are those that best predict the efficiency of two passes, crossflow valve trays with downcomers. In consonance with the previous published studies, the O'Connell'correlation underestimates the value of the efficiency, showing a slightly conservative value, while the predictions using Chan and Fair model and the modified Chan and Fair model with equivalent length proposed by Klemola and Ilme overestimates the value of the efficiency, showing a slightly non-conservative results.

Le présent travail de thèse met l'accent sur divers aspects de la distillation membranaire dans l'objectif de concevoir des procédés de dessalement proches du " zéro effluent liquide ". De manière générale, deux sujets sont discutés en détail: (i) la corrélation entre les caractéristiques de la membrane et les performances du procédé de distillation membranaire (ii) la compréhension et le contrôle de la polarisation thermique en DM. L'analyse de l'état de l'art en distillation membranaire porte notamment sur les progrès dans le développement des membranes, dans la compréhension des phénomènes de transport, les récents développements dans la conception des modules et le colmatage. Des phénomènes annexes et les applications innovantes sont également discutés dans la partie introductive de la thèse. L'effet des conditions de fabrication et de la composition des collodions sur les caractéristiques des membranes et la corrélation entre ces dernières et leurs performances a été discuté dans la section suivante. Il est établi que la morphologie de la membrane joue un rôle crucial dans ses performances pour des applications sur des fluides réels. En outre, on met en évidence que l'impact de la morphologie de la membrane est différente selon que la procédé fonctionne avec une phase liquide froide du côté distillat (Direct Contact Membrane Distillation - DCMD)) ou avec un courant d'air sec ou le vide (Air Gap ou Vacuum Membrane Distillation). Dans une deuxième partie, les aspects théoriques et expérimentaux de la polarisation thermique en distillation membranaire (DCMD) ont également été étudiés. Les phénomènes de polarisation thermique sur une membrane plane ont été étudiés en utilisant une cellule spécialement conçue. L'effet des conditions de fonctionnement et de la concentration de la solution sur la polarisation thermique a été étudié expérimentalement. Nous avons observé que l'augmentation de concentration de la solution favorise la polarisation thermique à cause d'une détérioration de l'hydrodynamique résultante à la surface de la membrane. Certaines techniques actives et passives pour réduire la polarisation thermique et le colmatage en distillation membranaire ont également été examinées dans l'étude cette étude. Nous avons montré que la polarisation thermique peut être considérablement réduite en générant des écoulements secondaires dans le fluide circulant à l'intérieur du canal d'alimentation, donc à l'intérieur de la fibre creuse si c'est cette configuration qui est retenue. Dans la présente étude, l'induction d'un écoulement secondaire a été réalisée en utilisant les fibres torsadées en hélice et une configuration ondulée. En raison de l'amélioration du niveau de polarisation thermique du côté de l'alimentation et du distillat, les géométries de fibres ondulées fournissent des flux et des taux de rendement supérieurs à ceux des autres configurations. La mise en œuvre d'un écoulement pulsé et intermittent pour contrôler la polarisation en distillation membranaire a également été examinée. Notre étude permet de conclure que ces types d'écoulements ont un impact positif sur les taux de rendement et le facteur d'amélioration volumique (gain en flux ramené par rapport à l'augmentation du volume de l'équipement) sans compromis sur le taux de remplissage des carters de fibres creuses. / The current PhD work emphasizes on various aspects of membrane distillation for approaching zero liquid discharge in seawater desalination. In broader sense, two themes have been discussed in detail: (i) correlation between membrane features and their performance in MD (ii) understanding and control of thermal polarization in MD. Introduction and state-of-the-art studies of MD including progress in membrane development, understanding the transport phenomenon, recent developments in module fabrication, fouling and related phenomenon and innovative applications have been discussed in introductory part of the thesis. The effect of operating conditions and dope compositions on membrane characteristics and correlation between membrane features and their performance has been discussed in subsequent section. It has been established that membrane morphology plays a crucial role in performance of the membrane for real applications. Furthermore, it has been demonstrated that the effect of membrane morphology is different for direct contact and vacuum configurations. Theoretical and experimental aspects of thermal polarization in direct contact membrane distillation have also been investigated. Thermal polarization phenomenon in a flat sheet membrane has been studied by using a specifically designed cell. The effect of operating conditions and solution concentration on thermal polarization has been explored experimentally. It has been observed that increased solution concentration favors the thermal polarization due to resulting poor hydrodynamic at the membrane surface and increase in diffusion resistance to the water vapors migrating from bulk feed phase to the membrane surface. Some active and passive techniques to decrease thermal polarization and possible fouling in membrane distillation have also been discussed in the current study. Thermal polarization can be greatly reduced by inducing secondary flows in the fluid flowing inside the fiber. The induction of secondary flows in the current study has been realized by using the fibers twisted in helical and wavy configurations. Due to improvement of thermal polarization coefficient on up and downstream, the undulating fiber geometries provide high flux and superior performance ratio. Application of intermittent and pulsatile flow to control thermal polarization in MD has also been discussed. It has been inferred that these flows have positive impact on performance ratio and volume based enhancement factors without compromising on packing density of the system. The application of MD for treatment of produced water has also been studied.