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.

Salts in solution should be removed by desalination techniques to prevent equipment fouling and corrosion. Common desalination technologies are energy intensive such as Multi Stage Flash (MSF) distillation which requires 14.5 J/m3 (Ribeiro. J, 1996) of energy. Desalination technologies produce purified water and a concentrated salt solution, where the salt concentration is dependent on the desalination technology used. This work investigates gas hydrate technology as a possible desalination technology. Hydrates are composed of guest molecules and host molecules. Guest molecules may be in the form of a liquid or gas. During hydrate formation, host molecules, water, form a cage enclosing the guest molecule. Common hydrate formers or guest molecules such as; methane, ethane, propane and carbon dioxide are currently being investigated in literature, for use in gas hydrate desalination technology. Common hydrate formers form hydrates at low temperatures; below 288 K and high pressures; above 2 MPa. To increase the temperature and reduce the pressure at which gas hydrates form, commercially available hydrofluorocarbon hydrate formers such as R14, R32, R116, R134a, R152a, R218, R404a, R407c, R410a and R507 are preliminarily investigated in this work. The criteria for choosing the most suitable fluorine-based formers require the former to be: environmentally acceptable where it is approved by the Montreal Protocol; non-toxic where it has a low acute toxicity; non-flammable; chemically stable; a structure II hydrate to simplify the washing process; available in commercial quantities; low cost in comparison to other hydrate formers; compatible with standard materials and contain a high critical point for a large heat of vaporisation (McCormack and Andersen, 1995). Taking all these criteria into account, R134a was chosen for further investigation as a possible hydrate former. In this work, hydrate-liquid-vapour phase equilibrium measurements are conducted using the isochoric method with a static high pressure stainless steel equilibrium cell. The Combined Standard Uncertainty for the 0-1 MPa pressure transducer, 0-10 MPa pressure transducer and the Pt100 temperature probes are ±0.64 MPa, ±5.00 MPa and ±0.09 K respectively. Vapour pressure measurements for Hydrofluoropropyleneoxide, CO2, R22 and R134a were measured to verify the pressure and temperature calibrations. Hydrate test systems for R22 (1) + water (2) and R134a (1) + water (2) were measured to verify calibrations, equipment and procedures. New systems measured included R134a (1) + water (2) + {5wt%, 10wt% or 15wt%} NaCl (3). For the system R134 (1) + water (2) at 281 K the dissociation pressure is 0.269 MPa. However, addition of NaCl to the system resulted in a shift of the HVL equilibrium phase boundary to lower temperatures or higher pressures. The average shift in temperature between the system R134a (1) + water (2) containing no salt and the systems containing {5, 10 and 15} wt% NaCl are -1.9K, -4.8K and -8.1K respectively. In this work, the measured systems were modelled using two methods of approach. The first method is where hydrofluorocarbon hydrate former solubility is included, (Parrish et al., 1972) and the second is where hydrofluorocarbon hydrate former solubility is ignored, (Eslamimanesh et al., 2011). From these models, it is found that hydrofluorocarbon solubility could not be neglected. In this work, the hydrate phase was modelled using modifications of the van der Waals and Platteeuw model, (Parrish et al., 1972). The liquid and vapour phases are modelled using the Peng- Robinson equation of state with classical mixing rules (Peng, 1976). The electrolyte component is modelled using the Aasberg-Peterson model (Aasberg-Petersen et al., 1991) modified by Tohidi (Tohidi et al., 1995). The percent absolute average deviation (%AAD) for the systems, which includes solubility, is 0.41 for R22 (1) + water (2) and 0.33 for R134a (1) + water (2). For the system R134a (1) + water (2) + {5 wt%, 10 wt% or 15 wt%} NaCl (3) the % AAD is 5.14. Using the hydrate former, R134a, is insufficient to ensure gas hydrate technology is competitive with other desalination technologies. Hydrate dissociation temperature should be increased and pressure decreased further to ambient conditions. As evident in literature, promoters, such as cyclopentane, are recommended to be added to the system to shift the HLV equilibrium phase boundary as close to ambient conditions as possible. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2011.

In this work VLE data was measured on three different pieces of equipment. Measurements were undertaken in the laboratory of Professor Gmehling in Oldenburg, Germany using two different static cells and in the Thermodynamics Research Unit (TRU), University of Natal, South Africa using a specially designed dynamic still. The three pieces of equipment used are as follows: i.) Static apparatus of Rarey and Gmehling (1993), ii.) Static apparatus of Kolbe and Gmehling (1985) as modified by Fischer and Wilken (2001), and, iii.) Dynamic apparatus ofHarris et al. (2003b). In total 370 data points were measured; fourteen sets of VLE data and eight vapour pressure data sets were measured. The work undertaken in Germany measured the systems hexane (1) + N-methylformarnide (2), benzene (1) + N-methylformamide (2), cWorobenzene (1) + N-methylformarnide (2) and acetonitrile (1) + N-methylformamide (2), at 363.15 K using the equipment of Rarey and Gmehling (1993). The systems CO2 (1) + Napthalene (2) at T = 372.45 K, 403.85 K and 430.65 K and CO2 (1) + Benzoic acid (2) at T= 403.28 K, 432.62 K and 458.37 K were measured on the equipment of Kolbe and GmeWing (1985) (as modified by Fischer and Wilken (2001)). Apart from the CO2 (1) + Napthalene (2) system at T = 372.45 K, all the above-mentioned data are new data. The equipment designed in the TRU was designed to operate between 300 and 700 K and between 1 kPa and 30 MPa. The equipment is of the dynamic recirculating VLE still type (DRVS) and is based on the principles of low-pressure stills. The still is constructed from uniquely machined Stainless-steel components and standard commercial Stainless-steel tubing and valves and is computer controlled to operate either isobarically or isothermally. Vapour pressures were measured on the new equipment for n-heptane, n-decane, n-dodecane, n-hexadecane, l-octadecene, 1-hexadecanol and d,l-menthol at low pressures and for acetone at high pressures. These vapour pressure measurements were used as test systems and ranged from 1.00 kPa to 1 000 kPa and from 308.33 K to 583.90 K. Cyclohexane (1) + ethanol (2) at 40 kPa and n-dodecane (1) + l-octadecene (2) at 26.66 kPa were measured as two isobaric VLE test systems. The VLE data measured for d,l-menthol (1) + l-isomenthol (2) at T= 448.15 K and n-dodecane (1) + l-octadecene (2) at P = 3.0 kPa represent new data measured on the equipment. All the VLE systems were modeled. Two data reduction methods were investigated: i.) the combined (r-rf) method, and, ii.) the direct method (H) method. Several different Gibbs excess models (Wilson, NRTL and UNIQUAC), equations of state (PengRobinson and virial) and mixing rules (Huron-Vidal, Wong-Sandler and Twu-Coon) were used in different combinations to find the best fit for the data. The Maher and Smith (1979) method was used to determine infinite dilution activity coefficients from the very smooth data of the N-methylformamide systems. Excess properties were determined for the CO2 (1) + Napthalene (2) and the CO2 (1) + Benzoic acid (2) systems. Although the equipment of Hams et al. (2003b) was able to measure data at high temperatures and elevated pressures, the precission of the data was not as good as was expected. Measuring the system temperature at elevated temperatures was especially problematic. The problem is attributed to the large mass of Stainless-steel used in the construction of the apparatus. To rectify this problem it is suggested that the equipment be modified to be lighter in weight and only capable of measuring VLE at moderate pressures (less than 3 MPa). / Thesis (Ph.D.)-University of Natal, Durban, 2004.

Computational fluid dynamic (CFO) modelling has been applied to examine the operation of the prc-ozonation system at Wiggins Waterworks, operated by Umgeni Water in Durban, South Africa. Ozonation is employed in water treatment process primarily to achieve the oxidation of iron and manganese, the destruction of micro-organisms and the removal of taste and odour causing compounds. It also aids in the reduction of the colour of the final water, enhancement of algae removal and possible reduction of coagulant demand. A hydrodynamic model has been satisfactorily verified by experimental tracer tests. The effect of the gas injection was modelled by increasing the level of turbulence intensity at the ozone contactor inlet. The model prediction of the overall tracer response corresponded closely to the experimental results. The framework of ozone reaction modelling was subsequently established using values of rate constants from the literature. An accurate prediction of the ozone concentration profile requires the application of the correct ozone kinetics involved. In raw waters, the depletion of ozone is influenced by the presence of natural organic matters (NOM). The observed ozone decay was found in good agreement using the pseudo first-order rate law. By measuring the total organic carbon (TOC) as a surrogate for NOM, the experimentally determined rate constants can be calculated to account for the effects of the ozone doses and the water quality. The characterisation study also aimed to provide sufficient information on ozone depletion and to be operated easily, without the lengthy and costly analyses ofa detailed kinetics study. The predicted profile of residual ozone concentration suggests the current operating strategy can be improved to optimise the ozone utilisation. The proposed monitoring point was suggested to be at the end of second companment where most ozone reactions have been completed. By coupling the transport equations of the target compounds with their chemical reaction rates, the concentration profile of these compounds such as ozone can be predicted in order to assist the understanding of an operation and to attain better interpretation of experimental results. / Thesis (Ph.D.)-University of KwaZulu-Natal, 2005.

The deteriorating water quality in South Africa and changing legislation requiring the industrial implementation of waste minimisation and pollution prevention technologies has highlighted the need for the investigation of new effluent treatment technologies such as advanced oxidation processes. This investigation details the evaluation of ultrasound, an emerging advanced oxidation process, to degrade organic compounds during water treatment. The objectives of the investigation included the design of a suitable ultrasonic laboratory reactor to investigate ultrasound chemistry and the sub-processes occurring during sonication. Atrazine was used as a model compound to compare the performance of ultrasound with that of ozone and hydrogen peroxide, already established advanced oxidation processes. Recommendations have also been made for the scale-up of ultrasonic processes. A 500 mL ultrasonic cell containing an ultrasonic horn as an energy source was designed and constructed. The measurement of hydrogen peroxide concentration was used as a tool to indicate the process conditions under which the formation of free radical reactions during sonication are enhanced. These include the application of oxygen and air sparging or the addition of a commercial source of hydrogen peroxide. It was found that oxygen sparging and a high acoustic power input should be used in ultrasonic processes with a short retention time, and conversely, that air sparging and a lower acoustic energy source should be used in processes with a long retention time. A flow loop system should be considered to maximise oxidation both within and beyond the sonicated zone, gas sparging should only occur within the sonication zone else the degradation of hydrogen peroxide is encouraged. Ultrasound is most effectively applied in water treatment as a pretreatment stage in combination with other technologies and not as a stand-alone process. Atrazine was used. as a model compound to compare the performance of ultrasound with ozone because of its persistence in the environment and resistance to degradation. Atrazine was degraded during sonication and ozonation. degradation increased wim the addition of hydrogen peroxide. Ozone decomposition (and hence free radical reactions) was enhanced when ozone was combined with ultrasound or hydrogen peroxide. Enhanced ozone decomposition during ozonation combined with sonication is due to the conditions (high temperatures and pressures) as well as the free radical reactions occurring within the collapsing cavitation bubbles and at the gas-liquid interface. The enhancing effect of combining ultrasound with ozone was greatest at the low ozone concentrations typically applied during water treatment. Atrazine degradation during sonication and ozonation is predominantly due to the reaction with hydroxyl radicals. Atrazine degradation products identified using gas chromatography and mass spectrometry were deethylatrazine. hydroxyatrazine and deethyldeisopropylatrazine (tentatively identified). / Thesis (Ph.D.)-University of Natal, Durban, 2001.

Vapour-Liquid Equilibrium (VLE), Liquid-Liquid Equilibrium (LLE) and Vapour-Liquid-Liquid Equilibrium (VLLE) are of special interest in chemical engineering as these types of data form the basis for the design and optimization of separation processes such as distillation and extraction, which involve phase contacting. Of recent, chemical companies/industries have required thermodynamic data (especially phase equilibrium data) for chemicals that are expensive or costly to synthesize. Phase equilibrium data for such chemicals are scarce in the open literature since most apparatus used for phase equilibrium measurements require large volumes (on average 120 cm3) of chemicals. Therefore, new techniques and equipment have to be developed to measure phase equilibrium for small volumes across reasonable temperature and pressure ranges. This study covers the design of a new apparatus that enables reliable vapour pressure and equilibria measurements for multiple liquid and vapour phases of small volumes (a maximum of 18 cm3). These phase equilibria measurements include: VLE, LLE and VLLE. The operating temperature of the apparatus ranges from 253 to 473 K and the operating pressure ranges from absolute vacuum to 1600 kPa. The sampling of the phases are accomplished using a single Rapid-OnLine-Sampler- Injector (ROLSITM) that is capable of withdrawing as little as 1μl of sample from each phase. This ensures that the equilibrium condition is not disturbed during the sampling and analysis process. As an added advantage, a short equilibrium time is generally associated with a small volume apparatus. This enables rapid measurement of multiple phase equilibria. A novel technique is used to achieve sampling for each phase. The technique made use of a metallic rod (similar in dimension to the capillary of the ROLSITM) in an arrangement to compensate for volume changes during sampling. As part of this study, vapour pressure and phase equilibrium data were measured to test the operation of the newly developed apparatus that include the following systems: • VLE for 2-methoxy-2-methylpropane + ethyl acetate at 373.17 K • LLE for methanol + heptane at 350 kPa • LLE for hexane + acetonitrile at 350 kPa • VLLE for hexane + acetonitrile at 348.20 K New experimental vapour pressure and VLE data were also measured for systems of interest to petrochemical companies. These measurements include: • VLE for methanol + butan-2-one at 383.25, 398.14 and 413.20 K ABSTRACT • VLE for ethanol + butan-2-one at 383.26, 398.23 and 413.21 K • VLE for ethanol + 2-methoxy-2-methylbutane at 398.25 and 413.19 K • VLE for ethanol + 2-methylpent-2-ene at 383.20 K These measurements were undertaken to understand the thermodynamic interactions of light alcohols and carbonyls as part of a number of distillation systems in synthetic fuel refining processes which are currently not well described. Two of these above mentioned systems include expensive chemicals: 2-methoxy-2-methylbutane and 2-methylpent-2-ene. The experimental vapour pressure data obtained were regressed using the extended Antoine and Wagner equations. The experimental VLE data measured were regressed with thermodynamic models using the direct and combined methods. For the direct method the Soave-Redlich-Kwong and Peng-Robinson equations of state were used with the temperature dependent function (α) of Mathias and Copeman (1983). For the combined method, the virial equation of state with the second virial coefficient correlation of Tsonopoulos (1974) was used together with one of the following liquid-phase activity coefficient model: TK-Wilson, NRTL and modified UNIQUAC. Thermodynamic consistency testing was also performed for all the VLE experimental data measured where almost all the systems measured showed good thermodynamic consistency for the point test of Van Ness et al. (1973) and direct test of Van Ness (1995). / Thesis (Ph.D.)-University of KwaZulu-Natal, Durban, 2011.

A 20 litre experimental batch and continuous test rig and 5 litre batch and 50 litre continuous test rigs for stirred ball milling were built at the University of Natal and Mintek respectively. All the mills featured a grinding vessel with a central shaft equipped with pins and a torque measurement system. A washed chrome sand from the Bushveld Igneous Complex was used for the grinding experiments. Particle size analysis of products was performed using standard sieves and a Malvern Particle Sizer. Batch tests were run in the 20 litre stirred ball mill to achieve efficient grinding conditions. The effects of grinding conditions such as pulp density, media size, media density and shaft rotation speed and mill design parameters such as ball load, pin spacing and pin diameter on product size, power consumption and media wear were studied. It has been shown that the median size of the product can be calculated by the Charles' Energy-Size Equation. The stirred ball mill has been found to be more energy efficient than the tumbling ball mill. An energy reduction of 50% was possible for a product size of 6 microns when the stirred ball mill was employed instead of the tumbling ball mill. The energy input per ton of grinding media in the stirred ball mill could be 10 times higher than for the tumbling ball mill. Although during coarse grinds the media wear was higher in the stirred ball mill than in the tumbling mill, it became less so as grinding proceeded and for a product median size of 4.8 microns it was the same. Using a 5 litre batch mill, an experimental programme was designed to study the comminution characteristics of the stirred mill. A factorial design was prepared with the following parameters, which influence grinding in the stirred ball mill: pulp density, pin tip velocity and ball density and size. The energy required for grinding the chromite sand in the stirred ball mill was determined by the use of Charles' Equation. The findings were in agreement with the results predicted by this equation. It was shown that the Rosin-Rammler size distribution equation was a suitable procedure for presenting and comparing grinding data obtained from the stirred ball mill. The factors that had the greatest effect on grindability were, in order of importance: ball size, pin tip velocity and ball density. Interactions between grinding parameters were negligible. results implied that accurate predictions can be made to determine the grinding conditions required to achieve a desired product specification. An attempt was made to study the grinding kinetics the chromite are using the mass population·balance model. Grinding tests were performed with two mono size fractions ·53+38 and -38+25 microns and natural feed ·100 microns using various pin tip velocities, ball densities and within the normal stirred ball milling operating range. relationship between the ball diameter and the particle was explained by the "angle of nip" theory which applied for roller crushers. It was shown that the particle giving the maximum breakage rate was directly proportional to the ball diameter. Estimated grinding kinetic parameters from monosize provided a good basis for predictions of natural feed. However, the breakage rate obtained from monosize tests appeared to be lower than those from the natural feed It was found that if the selection and breakage functions were determined by monosize tests, it was possible to modify selection function parameters by back-calculation which gave the best fit to the natural feed size. A good correlation was obtained between the experimental and product distributions using a population-balance model. The links between the empirical model combining Charles' and Rosin-Rammler equations and the first-order batch grinding equation were also shown. The stirred ball mills were operated in batch and continuous mode. The median size of the products from the batch stirred ball mill experiments closely matched those of the continuous grinding experiments under similar grinding conditions. Using a salt solution as a tracer material, an attempt was made to estimate the residence time distribution based on a simplified analysis of the motion of the water in tile mill. The current scale-up methods for the stirred ball mill are discussed. A torque model was developed for given shaft geometry and ball relating the power rements of the stirred ball mill to the following prime design and operating parameters : mill diameter, mill height, pin tip velocity and effective density of the mill load. The basic assumptions underlying the model were that the mill content behaved as a fluidised bed, consequently a P effg h type model for the pressure was applied throughout the grinding media bed the effective charge velocity was proportional to the pin tip velocity. It was found that pin spacing, pin diameter and ball diameter significantly affected the mill torque. A semi-empirical torque model was derived to include these parameters. The relationships formulated from these models were shown to be in excellent agreement with experimental results. / Thesis (Ph.D.)-University of Natal, Durban, 1993.

In the absence of any significant legislation, paper recycling in South Africa has grown to a respectable recovery rate of 43% in 2008, driven mainly by the major paper manufacturers. Recently introduced legislation will further boost the recovery rate of recycled paper. Domestic household waste represents the major remaining source of recycled paper. This source will introduce greater variability into the paper streams entering the recycling mills, which will result in greater process variability and operating difficulties. This process variability manifests itself as lower average brightness or increased bleaching costs. Deinking plants will require new techniques to adapt to the increasingly uncertain composition of incoming recycled paper streams. As a developing country, South Africa is still showing growth in the publication paper and hygiene paper markets, for which recycled fibre is an important source of raw material. General deinking conditions pertaining to the South African tissue and newsprint deinking industry were obtained through field surveys of the local industry and assessment of the current and future requirements for deinking of differing quality materials. A large number of operating parameters ranging from waste mixes, process variables and process chemical additions, typically affect the recycled paper deinking process. In this study, typical newsprint and fine paper deinking processes were investigated using the techniques of experimental design to determine the relative effects of process chemical additions, pH, pulping and flotation times, pulping and flotation consistencies and pulping and flotation temperatures on the final deinked pulp properties. Samples of recycled newsprint, magazines and fine papers were pulped and deinked by flotation in the laboratory. Handsheets were formed and the brightness, residual ink concentration and the yield were measured. It was determined that the type of recycled paper had the greatest influence on final brightness, followed by bleaching conditions, flotation cell residence time and flotation consistency. The residual ink concentration and yield were largely determined by residence time and consistency in the flotation cell. The laboratory data generated was used to train artificial neural networks which described the laboratory data as a multi-dimensional mathematical model. It was found that regressions of approximately 0.95, 0.84 and 0.72 were obtained for brightness, residual ink concentration and yield respectively. Actual process data from three different deinking plants manufacturing seven different grades of recycled pulp was gathered. The data was aligned to the laboratory conditions to take into account the different process layouts and efficiencies and to compensate for the differences between laboratory and plant performance. This data was used to validate the neural networks and select the models which best described the overall deinking performances across all of the plants. It was found that the brightness and residual ink concentration could be predicted in a commercial operation with correlations in excess of 0.9. Lower correlations of ca. 0.5 were obtained for yield. It is intended to use the data and models to develop a predictive model to facilitate the management and optimization of a commercial flotation deinking processes with respect to waste input and process conditions. / Thesis (Ph.D.)-University of KwaZulu-Natal, Durban, 2011.

A process was developed for the production of hexafluoropropylene oxide via the gas-phase oxidation of hexafluoropropylene with molecular oxygen. The non-catalytic oxidation reaction was investigated in an isothermal, laminar flow reactor at a total pressure of 4.5 bar and over the temperature range of 453 to 503 K. Feed mixtures comprising 20 to 67% HFP in oxygen were used at total flow-rates between 150 and 550 cm3 min-1. The reactor was fabricated from 1/8 inch nominal sized, copper refrigeration tubing and was 114.3 m long. The reactor was used in the form of a helical coil. Gas-chromatographic analysis was used for reactant and stable product quantification. The reaction gave hexafluoropropylene oxide, trifluoroacetyl fluoride and carbonyl fluoride as major products. Minor products included tetrafluoroethylene and hexafluorocyclopropane. The oxidation reaction also produced high molecular weight oligomers that were retained on the inner surface of the reactor tube. The operating conditions for the non-catalytic oxidation were optimized independently for HFPO selectivity and yield using quadratic response surface methodology. A maximum HFPO selectivity of 55.81% was identified at 478.2 K, a HFP/O2 molar feed ratio of 1.34 mol mol-1 and a space time of 113 seconds. An optimum HFPO yield of 40.10% was identified at 483.2 K, a HFP/O2 molar feed ratio of 1.16 mol mol-1 and a space time of 121 seconds. Using the weighted-sum-of-squared-objective-functions (WSSOF) multi-response optimization technique, a combined optimum HFPO selectivity and yield of 56% and 40%, respectively, was obtained at 480 K, with a HFP/O2 molar feed ratio of 1.21 and a space time of 118 seconds. This represented the best trade-off between these two performance criteria. A kinetic reaction scheme involving 8 species and 7 reactions was developed, based on the results of the experimental study, and was used to model the non-catalytic oxidation of HFP. The initial steps in this scheme encompassed the addition of oxygen to the double bond of the fluoro-olefin and transformation of the resultant dioxetane intermediate to form HFPO and the haloacetyl fluorides. Subsequent steps included the thermal decomposition of HFPO to yield CF3COF, C2F4 and c-C3F6, as well as elimination of C2F4, and to a lesser extent CF3COF, through oxidation. Rate parameters for the oxidation reactions were determined through a least-squares minimization procedure. The investigation was extended by considering the catalyzed synthesis of HFPO. Four different catalysts were studied viz., 1wt% Au/A12O3, 1wt% Au/ZnO, 10wt% CuO/SiO2 as well as 10wt% CuO/SiO2 doped with caesium. The gold-based catalysts were found to be completely inactive for the oxidation reaction. The caesium promoted, copper-based catalyst appeared to be the most stable and active, with no observable decomposition to copper fluoride. At 453 K, a HFP/O2 molar feed ratio of 0.86 and a weight-hourly-space-velocity of 0.337 h-1, a HFPO selectivity of 85.88% was obtained. This was significantly greater than what was achieved in the non-catalytic system. / Thesis (Ph.D.)-University of KwaZulu-Natal, Durban, 2012.

Artificial neural networks (ANN) can detect complex non-linear relationships between independent and dependent variables. Properly trained ANNs have repeatedly demonstrated superior predictive accuracy to other predictive technologies when applied to non-linear systems. Currently there are no studies that have been carried out on predicting survival of leukaemia patients at all. The neural network prediction method adopted in this study aims to provide a robust and accurate method for predicting survival of leukaemia patients for both censored and uncensored patient data. The aim of this research was also to find out the effectiveness of neural networks in modelling leukaemia prognosis and to determine the factors that have the most influence. There is ongoing research into finding ways and means of extending the life span of diseased patients. There is great interest in identifying factors that will yield better predictions of survival for terminally ill leukaemia patients. Prognostic factors generally differ with the treatment of leukaemia. Clinicians face the problem of how to choose the appropriate treatment regime, therefore an analysis of prognostic factors that predict success or failure may identify patients who require an alternative approach of specialist or targeted treatment. Being able to predict an individual patient’s prognosis will enable clinicians to categorise them into the relevant high and low risk treatment groups for conventional treatment or allow for the patients to be incorporated into specialised treatment schedules and clinical trials if available. In this study there is believed to be relationship that exists between the results gained on diagnosis and the period of survival. A patient’s health status is dependent on various symptoms and the complexity of the medical condition is dependent on an individual’s biological system. This complexity allows for the application of artificial neural networks (ANN) in predicting outcomes in medical application, especially in prognosis prediction and survival rate. This thesis contains contributions to the development of neural network models for survival analysis of leukaemia patients. The feed forward back propagation algorithm (BPA) modified to the gradient descent BPA was identified for the training and building of the neural network for predicting survival of leukaemia patients. The prognostic factors that affect survival have also been determined by the neural networks. The comparisons of models were based on using combined groups of leukaemia patients and comparing them with individual groups of the sub-types of leukaemia, i.e. acute lymphoid leukaemia (ALL), acute myeloid leukaemia (AML), chronic myeloid leukaemia (CLL) and chronic myeloid leukaemia (CML). A combination of 38 variables was used in the development of the neural networks. The variables were age, race, sex, gender, and results of full blood counts, differential tests and flow cytometry. The survival period of patients was based on the diagnosis date and the date of treatment. Those patients who status of mortality was known as of October 2008 were considered to be uncensored and were used for the 2-year and 3-year case studies. The patients with unknown mortality were considered as censored patients and used for the censored case study. The patient data was processed into a coded system and used to build the neural networks for each data set. The choice of patient groups used for the model building was prompted by the availability of uncensored data for analysis. For the group of combined leukaemia patients and the sub-group CML-CLL, it is recommended that the 2-year neural network model be used. The main prognostic factors affecting leukaemia survival were found to be the patient’s age, the mean haemoglobin concentration, % neutrophils and the markers CD13, CD20 and CD56. The race group, platelet count, % monocytes and the markers CD3, CD4, CD34 and LC lambda were found to significantly affect the CML-CLL group of patients. For the ALL and AML groups the 3-year neural network models were favoured. Prognostic factors for the survival of ALL patients were their age, the mean corpuscular haemoglobin concentration, % blasts and the markers CD8 and CD22. For the AML group the important prognostic factors were the patient’s age, the mean corpuscular haemoglobin concentration, the % neutrophils, % lymphocytes, and the markers CD7 and CD34. / Thesis (Ph.D.)-University of KwaZulu-Natal, Durban, 2009.