In the past decade a large number of studies concerning the application of modern state estimation techniques to chemical processes have been reported. However, the vast majority of these studies have been in simulation, and do not adequately demonstrate the utility of an estimator. For this reason, a practical study of Kalman-type estimation algorithms has been made. A real chemical process was interfaced to a small digital computing system and the process modelled analytically. Extensive simulation studies, which verified previous work, were performed before applying the Kalman estimation algorithms to the real process. It was found that the extended Kalman filter could usefully be applied to the problem of chemical process state estimation. Techniques for improving estimation when the process' was erroniously modelled in the filter, were successfully demonstrated. Incorporation of the filter in a feedback control loop enabled the use of modern control algorithms which gave good control. The Kalman estimation algorithms were then applied to the process which was under either open-loop or feedback control. The results, which were presented both graphically and in tabular form, clearly showed the practical advantages of using this kind of state estimator.

The work consists of an optimization study of the design of a complete chemical plant. The process chosen is the manufacture of acetic anhydride by thermal cracking of acetone. There are involved fourteen design variables, two major recycles and six iterative, computational loops. The process includes the most important unit operations of chemical engineering. Emphasis is placed in two areas: developing computer procedures which perform the design of individual items of plant in considerable detail and in producing an optimization program for the integrated plant. An improved version of the Pattern Search method is presented, known as MOSP, and it is shown to be competitive with the best Direct Search techniques available. A new approach is offered for achieving global rather than local optima. The results show clearly the feasibility of optimization in process design and give quantitative informations, for the chosen example, of the optimum conditions.

A continuous stirred tank reactor operating under digital computer control has been designed and built, and a comparison made between digital control methods that employ recent advances in estimation theory, and an equivalent classical analog control system. The digital control method required process identification, optimal control, and Kalman filtering for its implementation. The unknown parameters in the nonlinear mathematical model describing the process were identified by the method of quasi-linearization. This technique not only linearizes the non-linear system equations but also provides a sequence of functions that in general converge rapidly to the solution of the true non-linear equations. The optimal controller is believed to be original and comprises an adaptation of the quasi-linear algorithm. The controller calculates, at preset time intervals, those changes in the desired values of supervisory feedback controllers such that a weighted sum of squared errors at the end of the time interval is minimised. The predicted value of the state vector obtained in this calculation was then used in a fixed gain formulation of the Kalman filter. Simulation studies verified the viability of both the identification and optimal control methods, which were then applied to the process. The classical controller was designed using either the method of Ziegler and Nicholl or that of Bode, which utilised the frequency response. Experimental work showed good agreement with the simulation studies. However, it was found that the Kalman filter was unnecessary because of the essentially noise-free nature of the measurements. The identification method performed satisfactorily, and gave the values of the parameters in the mathematical model which described the process. The digital control system was shown to give better control than the classical scheme, especially in the steady state, where the predictive nature of the optimal controller effectively prevented overshoot, whilst assuring a fast stable response to disturbances.

No description available.

Reactor designs to collect data for determining the nature and rates of chemical reactions (particularly those in the gas phase at elevated temperatures) are reviewed. The limitations and weaknesses of the equipment presently employed are discussed and inadequate mixing in a so called 'ideally stirred' reactor is illustrated using a technique in which smoke is generated by chemical reaction within the vessel itself. The selection of a 'perfectly mixed' flow reactor from a wide range of alternative schemes is then reported (use being made where appropriate, of computer programs written for the purpose). This basic design concept is developed on prototypes and finally, the design, fabrication and testing of a stirred gas phase quartz reactor and its oven, suitable for operation up to approximately 500°C are described in detail. In addition, with only slight modification to the reactor body, it is suggested that the equipment would be suitable for the study of heterogeneously catalysed reactions.

The aim of this work was to design novel reactors to increase the reaction conversion and enantioselectivity for the asymmetric transfer hydrogenation (ATH) of acetophenone. The reactor designs should also be amenable to scale-out for increase of productivity. Asymmetric transfer hydrogenation of acetophenone with isopropanol is a reversible reaction and backwards reaction limits the reaction conversion and enantioslectivity. Novel reactor design aims to efficient acetone removal, which is a byproduct in the reaction system. By gas stripping, reaction equilibrium can be shifted and hence conversion and enantioselectivity improved. Reaction conditions optimization was initially conducted in a laboratory batch reactor and a simple kinetic model was built. The catalyst deactivation, the effect of the temperature, substrate concentration, substrate/catalyst concentration ratio and acetone concentration on the ATH were investigated. The metal-ligand bifunctional mechanism was considered for the kinetic model. The gPROMS/gEST commercial software was used for kinetic parameters estimation. Three continuous reactors (tubular reactor, rotating disc reactor and micromesh reactor) were designed and fabricated, each one encompassing a different gas/liquid contacting method. In the tubular reactor, gas/liquid contact is achieved through slugs. In the rotating disc reactor, gas/liquid contact is achieved through a thin film formed on the disc which rotates. In the micromesh reactor, a micromesh forms and stabilizes the gas/liquid interface. Acetone removal and asymmetric transfer hydrogenation studies were carried out under different conditions in these reactors and the performance of the different reactors were compared. The tubular reactor showed similar performance as the batch reactor. The rotating disc reactor enhanced acetone removal, thus improved the conversion and enantioselectivity. Acetone was removed most efficiently in the micromesh reactor. Therefore, the highest conversion and enantioslectivity were also obtained in this reactor. By simplified calculations, it was established that in order to increase the acetone removal efficiency, the ratio of gas to liquid flowrate and the gas-liquid interfacial area has to be increased. The scale out/up concept was also demonstrated with the micromesh reactor. Scale out/up was achieved by increasing the number of meshes in parallel and enlarging the single mesh reactor. The conversion and enantioselectivity was slightly lower in the scaled out version reactor which is probably due to the fact that the flow distribution inside the reactor was not uniform.

The main focus of this thesis is to characterise and model a novel multiphase mesh microreactor (continuous-phase microsystem) in relation to hydrodynamics, mass transfer/chemical reaction and scale-out operation. The physical understanding gained by the study of interface stability within the mesh pores and the analysis of single-phase flow distribution is then applied to dispersed-phase microsystems, in particular to the bubble formation in a Taylor flow type microreactor and its influence in the design of two-phase scale-out manifolds. In the mesh microcontactor, two phases flow in different channels separated by a thin mesh with openings (pores) through which the two phases can come in direct contact allowing mass transfer by diffusion to take place. A resistance network model was developed to optimise the shape of the reaction plate for even flow distribution and minimum sample dispersion. The meniscus shape, position and stability within the mesh pores was modelled and supported by experimental results in single pores and meshes. Different factors that affect meniscus stability in real meshes were identified and modelled. Parametric maps that define the boundaries of the kinetic control regime were developed (i.e. negligible mass transfer resistance) which allow the reaction kinetics to be directly retrieved from the experimental data of conversion vs. time. Flow distribution between parallel channels, required in the scale up operation of manifolds for high throughput, was analysed by means of a resistance network model validated via CFD simulations and experimentation. The model was applied to understand the effect of manufacturing tolerances, channel blockages and additional pressure losses on flow distribution. The knowledge gained in meniscus stability in single pores and single-phase flow distribution in manifolds was applied to analyse the effect of bubble formation on flow distribution in manifolds during Taylor/bubble flow. Pressure drop during Taylor flow (a function of bubble/liquid slug lengths) and pressure fluctuations during bubble formation were investigated and implemented in the resistance network model. A two-channel manifold structure for water/air and octane/air systems was successfully demonstrated.

The definition, history and applications of Microfiltration (MP) are briefly reviewed in Chapter 1. The physical mechanisms and mathematical models of the filtration process including concentration polarization (CP), gel polarization (GP) and pore blocking are given in Chapter 2. Crossflow microfiltration membrane fouling and the deposition of solids onto the filter surface have been investigated using a process fluid (seawater), latex and a ground mineral. The performance of various membrane materials has also been studied, including: acrylonitrile, polypropylene, PTFE, ceramic and stainless steel. The seawater filtration work showed in Chapter 3 that good filtrate flux rates can be maintained if material fouling or depositing on the membrane can be prevented from entering the membrane structure. A surface deposit may be removed by mechanical means such as backflushing with permeate or compressed air. This aspect of the work indicated that a more comprehensive study of fouling was required. Existing crossflow filtration membrane models did not adequately represent even the simplest filtration when penetration of the membrane structure applied. Such conditions occurred during latex filtration in Chapter 4. Latex of varying sizes and density were manufactured and filtrations using acrylonitrile membranes were performed. Considerable deposition of latex inside the membrane pores occurred despite the nominal rating of the membrane being less than the latex particle diameter. Thus the membranes relied on a depth filtration mechanism rather than a surface straining mechanism for filtration effectiveness. A standard filtration blocking model was modified for use in crossflow microfiltration, coupled with a mass balance on the amount of material filtered. This mathematical model was then used to predict and correlate the rate of filtration flux decay with respect to filtration time during crossflow filtration. The model provided acceptable accuracy and is an improvement on existing empirical models for the flux decay period. Under the circumstances of membrane penetration it is advisable to minimise the amount of material entering the membrane structure. Mechanical means to achieve this were investigated and a novel anti-fouling method using a centrifugal field force and enhanced shear stress at the membrane surface was developed. The filtration of limestone slurries with three different tubular filters are presented in Chapter 5, in which one filter was conventional, the other two novel ones were specially designed for the separation of particles with a density different from that of the liquid, one used a helical channel around the filter, and the other had tangential inlet and outlet endcaps. The centrifugal force produced by the spinning flow around these two filters retarded the approach of particles towards the membrane surface so that the particle deposition was reduced. The results showed such a system was energy efficient, saving 20 % of the energy required to effect a separation of mineral material compared with using the membrane in a more conventional way.

This work focuses on the characterisation of a commonly encountered flow pattern, Taylor flow, for the purpose of using it in high throughput experimentation (HTE). Taylor flow consists of elongated gas bubbles of equivalent diameter larger than the tube diameter separated by liquid slugs. The bubbles adopt a characteristic capsular shape almost entirely filling the channel cross section. This configuration enhances the mixing within the liquid slugs and significantly decreases axial mixing along the liquid compared to single phase liquid flow. In this work, the characterisation of a Taylor flow reactor was divided into three parts: The first part deals with the hydrodynamics of the flow, identifying the dimensionless parameter governing Taylor flow and the size of forming bubbles in a coaxial inlet arrangement. One of the findings was that a minimum channel size 100 mum exists below which the Taylor flow may becomes non-periodic or unstable. The experimentally observed mechanisms of Taylor bubble formation are reported and a simple model is provided for predicting the most common of the mechanisms. In the second part, two models were developed which enable the prediction of axial mixing and the residence time distribution curves of the Taylor flow reactor. The first model is applicable when Peclet numbers Pe 100. It was used to evaluate existing literature models and then for determining the rate of mixture injection in HTE. The second model is a special extension of the first applicable when Pe 1000 and accounts for forward as well as back mixing. In the third part a model was developed for identifying the flow rates and channel and slug lengths in the Taylor reactor which allow its use for determining kinetics of chemical reactions. An important finding from both axial and the kinetics models is that the slug lengths should be kept to the minimum possible size for best reactor performance.

This thesis carries out a detailed study of a nonlinear spectral theory that is useful for modeling and controlling chemical reactors. The motivation for this work originates from a few reports which have demonstrated in the past that the nonlinear spectral method offers a useful mathematical framework for classifying and quantifying nonlinear complexities of large degrees of freedom, as well as for qualifying a general nonlinear dynamic behavior. We present and discuss this new theory and show that it extends the familiar linear systems notion of characteristic modes (eigenmodes), as well as the notions of mathematical quantities known as the eigenvectors, and eigenvalues, into a multi-dimensional nonlinear domain, i. e., applies to model dimensions one, two, three and higher. This approach offers a new insight into nonlinear phenomena, and as such has a significant theoretical and practical value. In the theory of nonlinear systems the spectral framework provides some useful answers regarding the issues of multivariate process complexity, stability and control. Similarly, in applications it often leads to a simple relation between a desired process behavior and control parameters. We demonstrate this by showing how a process operating point, its behavior, and its domain of attraction are determined by nonlinear structures which characterize both a process and its control realization. In addition, we show that by a correctly modeling and regulating process nonlinearities one can obtain a nonlinear control solution that often outperforms the conventional first-order realizations. That is, there exist important nonlinear structural and dynamic process relations which determine a feasibility of a control realization. This is demonstrated by studying control behaviors of several highly exothermic continuously stirred tank reactor processes.