Thesis (MScEng)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: In this research work, auto-associative neural networks have been used for changepoint detection. This is a nonlinear technique that employs the use of artificial neural networks as inspired among other by Frank Rosenblatt’s linear perceptron algorithm for classification. An auto-associative neural network was used successfully to detect change-points for various types of time series data. Its performance was compared to that of singular spectrum analysis developed by Moskvina and Zhigljavsky. Fraction of Explained Variance (FEV) was also used to compare the performance of the two methods. FEV indicators are similar to the eigenvalues of the covariance matrix in principal component analysis. Two types of time series data were used for change-point detection: Gaussian data series and nonlinear reaction data series. The Gaussian data had four series with different types of change-points, namely a change in the mean value of the time series (T1), a change in the variance of the time series (T2), a change in the autocorrelation of the time series (T3), and a change in the crosscorrelation of two time series (T4). Both linear and nonlinear methods were able to detect the changes for T1, T2 and T4. None of them could detect the changes in T3. With the Gaussian data series, linear singular spectrum analysis (LSSA) performed as well as the NLSSA for the change point detection. This is because the time series was linear and the nonlinearity of the NLSSA was therefore not important. LSSA did even better than NLSSA when comparing FEV values, since it is not subject to suboptimal solutions as could sometimes be the case with autoassociative neural networks. The nonlinear data consisted of the Belousov-Zhabotinsky (BZ) reactions, autocatalytic reaction time series data and data representing a predator-prey system. With the NLSSA methods, change points could be detected accurately in all three systems, while LSSA only managed to detect the change-point on the BZ reactions and the predator-prey system. The NLSSA method also fared better than the LSSA method when comparing FEV values for the BZ reactions. The LSSA method was able to model the autocatalytic reactions fairly accurately, being able to explain 99% of the variance in the data with one component only. NLSSA with two nodes on the bottleneck attained an FEV of 87%. The performance of both NLSSA and LSSA were comparable for the predator-prey system, both systems, where both could attain FEV values of 92% with a single component. An auto-associative neural network is a good technique for change point detection in nonlinear time series data. However, it offers no advantage over linear techniques when the time series data are linear. / AFRIKAANSE OPSOMMING: In hierdie navorsing is outoassosiatiewe neurale netwerk gebruik vir veranderingspuntwaarneming. Dis is ‘n nielineêre tegniek wat neurale netwerke gebruik soos onder andere geïnspireer deur Frank Rosnblatt se lineêre perseptronalgoritme vir klassifikasie. ‘n Outoassosiatiewe neurale netwerk is suksesvol gebruik om veranderingspunte op te spoor in verskeie tipes tydreeksdata. Die prestasie van die outoassosiatiewe neurale netwerk is vergelyk met singuliere spektrale oontleding soos ontwikkel deur Moskvina en Zhigljavsky. Die fraksie van die verklaarde variansie (FEV) is ook gebruik om die prestasie van die twee metodes te vergelyk. FEV indikatore is soortgelyk aan die eiewaardes van die kovariansiematriks in hoofkomponentontleding. Twee tipes tydreeksdata is gebruik vir veranderingspuntopsporing: Gaussiaanse tydreekse en nielineêre reaksiedatareekse. Die Gaussiaanse data het vier reekse gehad met verskillende veranderingspunte, naamlik ‘n verandering in die gemiddelde van die tydreeksdata (T1), ‘n verandering in die variansie van die tydreeksdata (T2), ‘n verandering in die outokorrelasie van die tydreeksdata (T3), en ‘n verandering in die kruiskorrelasie van twee tydreekse (T4). Beide lineêre en nielineêre metodes kon die veranderinge in T1, T2 en T4 opspoor. Nie een het egter daarin geslaag om die verandering in T3 op te spoor nie. Met die Gaussiaanse tydreeks het lineêre singuliere spektrumanalise (LSSA) net so goed gevaar soos die outoassosiatiewe neurale netwerk of nielineêre singuliere spektrumanalise (NLSSA), aangesien die tydreekse lineêr was en die vermoë van die NLSSA metode om nielineêre gedrag te identifiseer dus nie belangrik was nie. Inteendeel, die LSSA metode het ‘n groter FEV waarde getoon as die NLSSA metode, omdat LSSA ook nie blootgestel is aan suboptimale oplossings, soos wat soms die geval kan wees met die afrigting van die outoassosiatiewe neural netwerk nie. Die nielineêre data het bestaan uit die Belousov-Zhabotinsky (BZ) reaksiedata, ‘n outokatalitiese reaksietydreeksdata en data wat ‘n roofdier-prooistelsel verteenwoordig het. Met die NLSSA metode kon veranderingspunte betroubaar opgespoor word in al drie tydreekse, terwyl die LSSA metode net die veranderingspuntin die BZ reaksie en die roofdier-prooistelsel kon opspoor. Die NLSSA metode het ook beter gevaaar as die LSSA metode wanneer die FEV waardes vir die BZ reaksies vergelyk word. Die LSSA metode kon die outokatalitiese reaksies redelik akkuraat modelleer, en kon met slegs een komponent 99% van die variansie in die data verklaar. Die NLSSA metode, met twee nodes in sy bottelneklaag, kon ‘n FEV waarde van slegs 87% behaal. Die prestasie van beide metodes was vergelykbaar vir die roofdier-prooidata, met beide wat FEV waardes van 92% kon behaal met hulle een komponent. ‘n Outoassosiatiewe neural netwerk is ‘n goeie metode vir die opspoor van veranderingspunte in nielineêre tydreeksdata. Dit hou egter geen voordeel in wanneer die data lineêr is nie.

Thesis (MScEng)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: Distillation is the most widely used separation technique in the chemical process industry and typically accounts for approximately one-third of the total capital cost and more than half of the total energy consumption of a typical petrochemical-chemical plant. Therefore, the design and optimization of the distillation sequence are of critical importance to the economics of the entire process. Azeotropic mixtures cannot be separated into their pure components via normal distillation. Enhanced distillation techniques such as heterogeneous azeotropic distillation should be considered for these mixtures. Isobaric vapour-liquid-liquid equilibrium (VLLE) data are highly important for the design and analysis of heterogeneous distillation columns. However, limited VLLE data are available in literature due to the difficulties involved with measuring such data. The objective of this work was to systematically evaluate and compare the performance of selected entrainers (including benzene, DIPE and cyclohexane) for the dehydration of C2 and C3 alcohols. To meet this objective, phase equilibrium data had to be measured. Isobaric VLLE at standard atmospheric conditions were measured with a dynamic Guillespie unit equipped with an ultrasonic homogenizer, which prevented liquid-liquid separation. Vapour-liquid equilibrium (VLE) and VLLE data were measured for ethanol/water/di-isopropyl ether (DIPE), n-propanol/water/DIPE and n-propanol/water/isooctane. The VLE data were found to be thermodynamically consistent according to the L-W (Wisniak 1993) and McDermott-Ellis consistency tests. No thermodynamic consistency test, specifically for VLLE data, could be found in literature, but the LLE part of the data followed a regular profile according to the Othmer-Tobias correlation. The binary DIPE/water and isooctane/water azeotropes, as well as ternary ethanol/DIPE/water and n-propanol/isooctane/water azeotropes, as measured in this work, agree well with those found in literature. Regressed parameters for the NRTL, UNIQUAC, and UNIFAC models, generally improved the model predictions compared with built-in Aspen parameters. This confirmed the importance of having actual measured VLLE data available for evaluation and improvement of estimations by thermodynamic models. NRTL predicted the ethanol/DIPE/water and n-propanol/DIPE/water VLLE most accurately. Despite the improved regressed parameters for n-propanol/isooctane/water predictions, the models are still considered unsuitable for accurate prediction of the VLLE behaviour of this system. Separation sequences were simulated in Aspen with built-in and regressed parameters, respectively, to illustrate the significant effect such inaccurate parameters have on these simulations. Phase diagram (VLLE data) evaluation of ethanol and isopropanol (IPA) with various entrainers, as found in literature, indicated that DIPE might be a good entrainer for the dehydration of these alcohols. Benzene and cyclohexane are generally used as entrainers in industry for these processes. Benzene is however carcinogenic and therefore an alternative has to be found (United States Department of Labour - Occupational Safety & Health Administration 2011). Separation sequences were simulated for ethanol dehydration with benzene and DIPE as entrainers, respectively. Taking cost and safety into account, DIPE can be considered an acceptable replacement for benzene as entrainer for ethanol dehydration. A separation sequence was also simulated for the dehydration of IPA with DIPE as entrainer and compared to a simulation with cyclohexane (Arifin, Chien 2007) as entrainer. DIPE was found to be a reasonable alternative to cyclohexane as entrainer for IPA dehydration. Two other separation sequences were simulated as practical examples where DIPE could be used as entrainer for the recovery of ethanol or n-propanol from aqueous Fischer Tropsch waste streams. DIPE is therefore found to be a feasible alternative entrainer to benzene and cyclohexane for the dehydration of ethanol and IPA via heterogeneous azeotropic distillation, based on pre-liminary cost considerations, separation ability and safety. Better entrainers than DIPE may exist, but from the data available in literature and the measurements made in this work DIPE appears to be superior to benzene, cyclohexane and isooctane. / AFRIKAANSE OPSOMMING: Distillasie is die mees algemeen-gebruikte skeidingstegniek in die chemiese proses-industrie. Dit is tipies verantwoordelik vir ʼn derde van die totale kapitaalkoste en meer as die helfte van die totale energie verbruik op ʼn tipiese petrochemiese chemiese aanleg. Daarom is die ontwerp en optimering van ʼn distillasie trein van kardinale belang vir die winsgewendheid van die proses. Azeotropiese mengsels kan nie slegs deur normale distillasie in suiwer komponente geskei word nie. Gevorderde distillasie tegnieke soos heterogene azeotropiese distillasie moet dus oorweeg word vir sulke mengsels. Isobariese damp-vloeistof-vloeistof ewewigsdata is een van die belangrikste fisiese eienskappe vir die ontwerp van heterogene distillasie kolomme. Die hoeveelheid damp-vloeistof-vloeistof ewewigsdata wat beskikbaar is in die literatuur is egter baie beperk omdat dit moeilik is om die data te meet. In hierdie werk is isobariese damp-vloeistof-vloeistof ewewigsdata met ʼn dinamiese Guillespie eenheid, by standaard atmosferiese druk gemeet. Die eenheid is toegerus met ʼn ultrasoniese homogeniseerder om vloeistof-vloeistof skeiding te voorkom. Temperatuur is gemeet met ʼn akkuraatheid van 0.03oC by 0oC. Die sisteem se druk is konstant gehou op 101.3 kPa met ʼn akkuraatheid van 0.35 % VSU (Vol Skaal Uitset). Die ewewigsamestellings is met ʼn relatiewe akkuraatheid van 2 % gemeet. Daar is damp-vloeistof en damp-vloeistof-vloeistof ewewigsdata van etanol/water/di-isopropiel eter (DIPE), npropanol/ water/ DIPE en n-propanol/water/iso-oktaan gemeet. Die damp-vloeistof ewewigsdata is deur die LW (Wisniak 1993) en McDermott-Ellis termodinamiese konsistensie toetse getoets en konsistent bevind. Geen termodinamiese konsistensie toets spesifiek vir damp-vloeistof-vloeistof ewewigsdata kon gevind word nie. Die Othmer-Tobias korrelasie dui egter aan dat die vloeistof-vloeistof ewewig gedeelte van die data ʼn reëlmatige gang volg. Die binêre DIPE/water en iso-oktaan/water azeotrope en ternêre etanol/DIPE/water en n-propanol/iso-oktaan/water fase-ewewigte wat in hierdie werk gemeet is, stem ooreen met die wat in die literatuur te vind is. Die parameters vir die modelle (NRTL, UNIQUAC en UNIFAC) wat in hierdie werk bestudeer is, is in die algemeen verbeter deur regressie van die eksperimentele data. Dit dui daarop dat dit belangrik is om eksperimentele damp-vloeistof-vloeistof ewewigsdata te hê om die voorspellings van termodinamiese modelle mee te evalueer en te verbeter. Die etanol/water/ DIPE en n-propanol/water/DIPE damp-vloeistof-vloeistof ewewigsdata is die akkuraatste deur NRTL voorspel. Ten spyte van die verbeteringe wat deur regressie behaal is met die NRTL en UNIQUAC parameters vir n-propanol/water/isooktaan, word hierdie modelle steeds nie as gepas vir die voorspelling van hierdie datastel beskou nie. Skeidingsreekse is gesimuleer met die ingeboude Aspen parameters en regressie parameters, onderskeidelik, om te illustreer dat onakkurate parameters ʼn beduidende effek op sulke simulasies het. Die evaluasie van fase diagramme van etanol en IPA met verskeie skeidingsagente, wat in die literatuur te vind is, dui aan dat DIPE ʼn goeie skeidingsagent kan wees vir die dehidrasie van hierdie alkohole. Skeidingsreekse vir die dehidrasie van etanol met benseen en DIPE, onderskeidelik, is gesimuleer. Met koste en veiligheid in ag geneem, is daar gevind dat DIPE ʼn aanvaarbare plaasvervanger vir benseen as skeidingsagent vir etanol dehidrasie kan wees. Daar is ook ʼn skeidingsreeks vir die dehidrasie van IPA met DIPE as skeidingsagent gesimuleer en met ʼn simulasie (Arifin, Chien 2007) wat sikloheksaan as skeidingsagent gebruik, vergelyk. Daar is bevind dat DIPE ʼn redelike alternatief vir sikloheksaan kan wees as skeidingsagent vir IPA dehidrasie. Nog twee skeidingsreekse is gesimuleer om as praktiese voorbeelde te dien van die gebruik van DIPE as skeidingsagent om etanol of n-propanol vanaf waterige Fischer-Tropsch afvalstrome te herwin. Daarom is daar bevind dat DIPE ʼn geldige alternatiewe skeidingsagent vir benseen en sikloheksaan is, gebaseer op koste, skeidingsvermoë en veiligheid. Daar kan beter skeidingsagente as DIPE bestaan, maar vanuit die data beskikbaar in literatuur en die metings geneem in hierdie werk, is DIPE die beste.

Thesis (MScEng)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: See item for full text / AFRIKAANSE OPSOMMING: Sien item vir volteks

Thesis (MScEng (Process Engineering))--Stellenbosch University, 2008. / Developments in the field of liquid detergents and cosmetics have increased the demand for surfactants, processing aids and emollients. Alcohols are often used in liquid products where they serve as solvents for the detergent ingredients, adjust the viscosity and prevent product separation. Industrial scale oxygenation of the alkane to the alcohol is often incomplete, resulting in a significant amount of residual alkane. Application of these alcohols often requires a low residual alkane content and a post-production separation process is thus required. Traditional separation methods such as distillation and crystallisation are not technically viable as crossover melting and boiling points prevent the successful implementation of such processes and it is envisaged to use supercritical extraction to separate a mixture of n-alkanes and 1- alcohols. The project scope revolves around a product stream consisting of detergent range alcohols and corresponding n-alkanes that need to be separated. To model such a system, a typical detergent range alkane – alcohol feed with an average of 13 carbon atoms was selected, although a large number of the molecules have between 12 and 14 carbon atoms each. Generally the presence of the hydroxyl group as well as an increase in the number of carbon atoms decreases the solubility in supercritical solvents [17]-[19]. The most difficult separation will therefore be that of the alcohol with the least number of carbon atoms, that is 1-dodecanol (alcohol with 12 carbon atoms, CH3-(CH2)9-CH2-OH ) and the alkane with the most number of carbon atoms, that is n-tetradecane (alkane with 14 carbon atoms, CH3-(CH2)12-CH3 ). To model the system, it is assumed that the hydrocarbon backbone is linear and the alcohol is primary. Therefore 1-dodecanol and n-tetradecane are used. If it is possible to separate 1-dodecanol and ntetradecane with the use of supercritical fluids, it should be possible to separate an industrial mixture. The deliverables of this study are: a comparison of the high pressure solubility of n-tetradecane and 1-dodecanol with a selection of possible solvents; a selection of potential solvents to be tested on a pilot plant to confirm practical separation. From the literature and measured phase equilibria, all three solvents (carbon dioxide, ethane and propane) have the ability to distinguish (based on a difference in the pressure required for solubility) between 1-dodecanol and n-tetradecane. Both carbon dioxide and ethane have favourable temperature considerations. Propane has superior solubility of n-tetradecane and 1- dodecanol. Carbon dioxide demonstrates the highest selectivity. Pilot plant experiments have shown that both carbon dioxide and ethane have the ability to separate a 50-50% (mass percentage) mixture of 1-dodecanol and n-tetradecane. Both solvents perform at their best at low temperatures. Carbon dioxide shows the best results at low temperature and conditions of reflux. The highlight of pilot plant experiments with supercritical carbon dioxide is achieving a selectivity of 16.4 with the mass percentage of n-tetradecane at 5% and 82% for the bottoms and overheads product respectively. Very good separation is achieved using supercritical ethane as solvent even without reflux. Attention is drawn to pilot plant experiments where the selectivity is as high as 82 with the mass percentage of n-tetradecane in the bottoms and overheads product at 1% and 82% respectively. It is recommended to measure ternary phase equilibria for the system n-tetradecane, 1-dodecanol and carbon dioxide/ethane to establish the interaction between n-tetradecane and 1-dodecanol. The measured binary phase equilibrium data need to be expanded to include the vapour mass fraction composition in the isothermal solubility data.

Thesis (MScEng (Process Engineering))--University of Stellenbosch, 2007. / The catalytic selective oxidation of hydrocarbon molecules is the process where a selectively oxidized intermediate molecule is formed instead of the thermodynamically favoured total oxidation products, in the presence of a suitable catalyst. Examples are the selective oxidation of methane to synthesis gas at moderate temperatures, for which a catalyst is still needed and the selective oxidation of propene to acrolein over α-Bi2Mo3O12. The selective oxidation of propene over α-Bi2Mo3O12 occurs via a Mars-van Krevelen mechanism where the bulk oxygen in the catalyst is inserted into the propene molecule and leaves as part of the product, while being replaced with gaseous oxygen. From an economic perspective there is a need to produce synthesis gas from methane at low temperatures. It was seen in the literature that α-Bi2Mo3O12 is a mixed metal oxide that might be capable of achieving this. The feasibility of the selective oxidation of methane to synthesis gas with α-Bi2Mo3O12 was therefore investigated. However, it was found that the selective oxidation of methane over α-Bi2Mo3O12 is not feasible at moderate temperatures. To circumvent the problem of producing synthesis gas at low temperatures a membrane reactor was suggested that might be able to produce synthesis gas at moderate temperatures with conventional selective methane oxidation catalysts that thermodynamically favours carbon dioxide formation at low temperatures. No time on-stream experiments had been done previously for the selective oxidation of ...

Thesis (MScEng (Process Engineering))--University of Stellenbosch, 2006 / Fouling is a serious problem in membrane filtration, caused by pore plugging and adsorption of rejected macromolecules or other solutes in the membrane system. This requires periodic cleaning of membranes, which can add considerably to the overall cost of plant operation owing to lost productivity related to down-time, the cost of the chemicals used in cleaning, higher pressures and associated pumping costs to maintain membrane productivity, as well as reduced lifetime of the membranes. Ultrasound has recently been suggested as a promising approach to combating fouling in membranes. In principle it can be used on-line and may even eliminate the use of chemical cleaning or alternative measures completely, which could lead to major advances in the development and implementation of membrane technology. The objective of this investigation was therefore to assess the feasibility of using ultrasound to mitigate fouling in capillary ultrafiltration systems applied to water containing natural organic matter. Experimental work was conducted with a small laboratory-scale capillary membrane module. Ultrasound was introduced into the system by means of an ultrasonic probe operating at a fixed frequency of approximately 30 kHz, generating a maximum acoustic power density of 130 W/cm2 with a nominal power output of 50 W (IKA Labortechnik Staufen, United Kingdom, U50). Five systems were investigated, viz. aqueous solution of Congo Red dye, ultrapure water, coloured ground water from the George region, water from the Steenbras dam, as well as an aqueous solution of dextran. In most cases, ultrasonication resulted in an increase in the permeate flux. This increase could partly be attributed to an increase in the temperature and thus a decrease in the viscosity of the fluid and partly to enhanced mass and energy transfer due to sonication. Based on experiments done with the Congo Red dye and ultrapure water, no damage as a result of ultrasonication could be discerned in the membrane filter, except when there was direct contact between the ultrasonic probe and the membrane materials. Permeate quality analyses confirmed that sonication does not damage the membrane material – no degradation of permeate quality was found specifically during sonication intervals. In conclusion, ultrasound indeed appeared to be an effective approach to remove foulants associated with natural organic matter from membranes. However, an issue not addressed by this study, but apparent from the literature, is that the effect of ultrasound is strictly local and this has major implications for the scaleup of such ultrasound systems.

Thesis (MScEng (Process Engineering))--University of Stellenbosch, 2009. / ENGLISH ABSTRACT: The aim of this study was to modify the surface of polyethersulfone (PES) ultrafiltration (UF) membranes to produce a more hydrophilic membrane by cross-linking poly(vinyl alcohol) (PVA) with sodium tetraborate (Na2B4O7.10H2O) (SB) on the surface. Key preparation factors were identified as PVA molecular weight, concentrations of the PVA and SB, cross-linking reaction time, number of coatings and the mode of coating. The effect of these factors on the membrane performance (salt retention and permeate flux) is discussed. These PVA-SB membranes typically had 11.46% retention and 413.30 L/m2.h flux for a feed containing 2000 ppm NaCl (0.45 MPa, 20°C, 45 – 50 L/h). The coating was shown to be uniform and stable by Fourier transform infrared spectroscopy (FT-IR) analyses. Coating significantly increased hydrophilicity and a maximum flux increase of 500 L/m2.h was reached. Measurements showed a reduced water contact angle and this confirmed the obvious enhancement of surface hydrophilicity. As a control, the role of the PVA base layer without cross-linking and the effects of its drying and heating on the water permeability of the PES-UF membrane were also studied, in order to ascertain maximum treatment conditions. Retention and permeate flux were determined (feed solution: 2000 ppm NaCl, applied pressure 0.45 MPa, 25°C, 45 – 50 L/h). It was found that the heating had the largest effect on the reduction of water permeability and therefore 50°C was the limit for treatment of this specific PES-UF membrane. Thin-film composite (TFC) membranes were prepared by an interfacial polymerization (IP) reaction between a polyfunctional amine and tri- or di-functional carboxylic chloride and then evaluated for their reverse osmosis (RO) performance. The salt retention of the PVA-SB membranes was improved when covering the cross-linked PVA gel sub-layer with a polyamide (PA) layer. However, the permeate flux decreased to below 30 L/m2.h (2000 ppm NaCl, 1 – 2 MPa, 20°C, 45 – 50 L/h). Two TFC membranes made from trimesoyl chloride (TMC) with m-phenylenediamine (MPD) or 2,6-diaminopyridine (DAP) exhibited retentions of 96.71% to 89.65% and fluxes of 10.93 to 27.91 L/m2.h, depending on the type of diamine used, when tested with a 2000 ppm NaCl solution (2 MPa, 25°C, 45 – 50 L/h). Two TFC membranes made from a n ew 2,5-furanoyl chloride (FC) with MPD or DAP exhibited retentions of 34.22% to 58.54% and fluxes of 49.21 to 25.80 L/m2.h, depending on the type of diamine used, when tested with a 2000 ppm NaCl solution (1 MPa, 25°C, 45 – 50 L/h). Novel PVA-SB-DAP-FC membranes made from the DAP with FC had the highest hydrophilicity value and exhibited >58.54% NaCl retention and 25.80 L/m2.h flux, and 75.08% MgSO4 retention and 34.75 L/m2.h flux, when tested with (2000 ppm feed, 1 MPa, 25°C, 45 – 50 L/h). The effect of the chemical structures of the different amines and carboxylic chlorides used on the RO performances of the TFC membranes prepared by two amines reacting with TMC or FC, on the surfaces of the modified asymmetric PES-UF membranes, was investigated. FT-IR and water contact angle determination were used to characterize the chemical structure, morphology and hydrophilicity of the PA layers of the composite membranes. The response surface methodology (RSM) was used to optimize the preparation conditions that had the largest effects on the RO performance of the PVA-SB-DAP-FC membranes. Good membrane performance could be realized particularly by manipulating three variables: DAP concentration, FC concentration and polymerization time (PT). The regression equation between the preparation variables and the performance of the composite membranes was established. Main effects, quadratic effects and interactions of these variables on the composite membrane performance were investigated. The membranes were characterized in terms of pure water permeation (PWP) rate, molecular weight cut off (MWCO), solute separation and flux. Mean pore size (μp) and standard deviation (σp) of the membranes were determined using solute transport data. The results revealed that PVA-SB membranes have almost the same pure water permeation that PES-UF membranes have. The MWCO of the PES-UF membranes decreased from 19,000 to 13,000 Daltons when the membrane was coated with PVA. / AFRIKAANSE OPSOMMING: Die doel van hierdie studie is die modifikasie van die oppervlakte van poliëtersulfoon ultrafiltrasie (PES-UF) membrane om meer hidrofiliese membrane te berei deur die kruisbinding van polivinielalkohol (PVA) met natriumtetraboraat ((Na2B4O7.10H2O) (NaB) op die membraanoppervlakte. Sleutelfaktore in die bereidingsproses is geïdentifiseer, naamlik: PVA molekulêre massa, PVA en NaB konsentrasies, kruisbindingsreaksietyd, die aantal bestrykingslae, en die manier waarop die bestrykingslae aangewend is. Die invloed van hierdie faktore op die membraanontsouting en vloed is ondersoek, en word hier bespreek. Hierdie PVA-NaB membrane het die volgende tipiese resultate getoon: 11.46% ontsouting en 413.30 L/m2.h vloed (Kondisies: 2000 dpm NaCl oplossing, 0.45 MPa toegepaste druk, 20 °C, vloeitempo 45–50 L/h). Die deklaag was uniform en stabiel, soos bepaal d.m.v. FTIR. Die aanwesigheid van die deklaag het die hidrofilisiteit verhoog en 'n maksimum vloed van 500 L/m2.h is behaal. Die waterkontakhoek is ook gemeet; 'n laer waarde het 'n verbetering in die hidrofilisiteit van die oppervlakte bevestig. Die rol van die PVA basislaag, sonder kruisbinding (kontrole), en die effek van uitdroging en verhitting hiervan, is ook bestudeer, om sodoende optimale behandelingskondisies te bepaal. Membraanontsouting en vloed is bepaal (Kondisies: 2000 dpm NaCl oplossing, 0.45 MPa toegepaste druk, 25 °C, vloeitempo 45–50 L/h). Verhitting het die grootste effek gehad op die afname in vloed. Daar is bevind dat 'n maksimum temperatuur van 50°C geskik is vir die behandeling van hierdie spesifieke PES-UF membraan. Dunfilmsaamgestelde (DFS) membrane is berei d.m.v. 'n tussenvlakpolimerisasiereaksie tussen 'n polifunksionele amien en 'n di- of tri-funksionele karbonielchloried, en daarna is die tru-osmose (TO) gedrag bepaal. Die ontsouting van die PVA-NaB membrane was hoër nadat die kruisgebinde PVA jel sub-laag met 'n poliamied (PA) laag bedek is. Die vloed het egter afgeneem, tot onder 30 L/m2.h (Kondisies: 2000 dpm NaCl oplossing, 1–2 MPa toegepaste druk, 20 °C, vloeitempo 45–50 L/h). Twee DFS membrane is berei met trimesoïelchloried (TMC), naamlik met m-fenieldiamien (MFD) of 2,6-diaminopiridien (DAP). Afhangend van die diamien wat gebruik is, is die volgende ontsoutingsresultate en vloede verkry: 96.71% tot 89.65% en 10.93 to 27.91 L/m2.h (Kondisies: 2 000 dpm NaCl oplossing, 2 MPa toegepaste druk, 25 °C, v loeitempo 45–50 L/h). Twee DFS membrane is ook berei met 'n nuwe verbinding, 2,5-furanoïelchloride (FC), en MFD of DAP. Afhangend van die diamien wat gebruik is is die volgende ontsoutingsresultate en vloede behaal: 34.22% tot 58.54% en 49.21 tot 25.80 L/m2.h (Kondisies: 2000 dpm NaCl oplossing, 1 MPa toegepaste druk, 25 °C, vloeitempo 45–50 L/h). Die PVA-NaB-DAP-FC membrane het die hoogste hidrofilisiteit getoon: 58.54% NaCl ontsouting en 25.80 L/m2.h vloed, en 75.08% MgSO4 ontsouting en 34.75 L/m2.h vloed (2000 ppm NaCl oplossing, 1 MPa toegepaste druk, 25 °C, vloeitempo 5–50 L/h). Die invloed van die chemiese struktuur van die verskillende diamiene en karboksielsuurchloriedes wat gebruik is in die bereiding van die DFC membrane op die oppervlakte van die gewysigde PES-UF membrane is in terme van TO ondersoek. FTIR en kontakhoekbepalings is gebruik om die chemiese struktuur, morfologie en hidrofilisiteit van die PA lae van die saamgestelde membrane te bepaal. Die eksperimentele oppervlakte ontwerp metode is gebruik om die bereidingskondisies vir die TO aanwending van die PVA-NaB-DAP-FC membrane te optimiseer. Goeie resultate is verkry deur die volgende veranderlikes te manipuleer: DAP en FC konsentrasies en die tydsduur van die polimerisasie. 'n Regressie-vergelyking tussen die bereidingsverandelikes en die funksionering van die saamgestelde membrane is bepaal. Die volgende is ook ondersoek vir hul effek op die funksionering van die saamgestelde membrane: hoof-effekte, vierkantseffekte, en interaksie tussen veranderlikes. Die eienskappe van die membrane wat bepaal is, is: deurlatingstempo van suiwer water (DSW), molekulêre massa-afsnypunt (MMAP), skeiding van opgeloste sout en vloed. Deurlating van opgeloste sout data is gebruik om gemiddelde poriegrootte (μp) en standaard afwyking (σp) van die membrane te bepaal. Resultate het getoon dat die PVA-NaB membrane amper dieselfde DSW gehad het as die PES-UF membrane. Die MMAP van die PES-UF membrane het afgeneem van 19,000 tot 13,000 Daltons na behandeling met PVA.

Thesis (MScEng (Process Engineering))--University of Stellenbosch, 2009. / Surfactants are compounds that reduce interfacial surface tension, resulting in detergency, emulsifying, foaming and dispersing properties. Surfactants produced via biochemical processes (biosurfactants) form a niche market with their low toxicity, biodegradability and high specificity attributes. Biosurfactants have recently received considerable attention owing to their potential as biomedical molecules. In this study a knowledge base was established for the development of a process which produces biosurfactants for use as antimicrobial agents. Specifically, rhamnolipid biosurfactants were produced from Pseudomonas aeruginosa and tested for antimicrobial activity against target organisms. Accurate and reproducible analyses for the quantification of rhamnolipids and antimicrobial activity were developed. The amount of rhamnolipid was determined indirectly by measuring the rhamnose concentration. A novel HPLC method as well as an orcinol colorimetric method were developed for rhamnose measurement. In order to obtain accuracy with the orcinol method it was found that samples must be extracted at least three times prior to the analysis. An examination of literature on rhamnolipid production showed that many studies used colorimetric methods without extraction. Antibacterial activity was quantified by zone clearing around wells of supernatant in soft agar containing the target organism Mycobacterium aurum. This target organism is especially important in a South African context, since it is used to indicate possible susceptibility of tuberculosis to antibiotics. This method was developed for antibacterial testing, after a standard disk diffusion method proved to be ineffective. Antifungal activity of rhamnolipids was evaluated against the fungus Botrytis cinerea, by growing a lawn of fungus on a plate and adding rhamnolipid. The factors influencing rhamnolipid production were studied by growing different Pseudomonas aeruginosa strains from the ATCC culture collection, namely ATCC 9027 and ATCC 27853 as well as a locally isolated strain under different media conditions. The initial focus was on production of biosurfactants in media containing glucose as substrate. Alkanes were subsequently investigated as an alternative substrate, since they are readily available in South Africa as byproducts from the petrochemical industry. The rhamnolipids produced from the culture collection strains were evaluated for their antibacterial activity against Mycobacterium aurum. A number of key factors were identified which were important for the development of a rhamnolipid production process. Of critical importance were the media conditions. Good production was achieved on glucose media containing a phosphate limitation, pH buffering around neutral pH and a high carbon concentration (2 % carbon). When Pseudomonas aeruginosa ATCC 9027 was cultured on this medium (a minimal salts phosphate limited medium with a Tris buffer), it produced 1.31 g/l rhamnose, equivalent to 4.0 g/l rhamnolipid. This rhamnolipid concentration is 2.7-fold higher that of 1.47 g/l reported in the literature with the same strain (cultured on a different phosphate limited medium The particular strain also proved to be a factor which influenced the yield of rhamnolipids. A rhamnose concentration of 0.43 g/l was obtained with Pseudomonas aeruginosa ATCC 27853 grown on MSM+Tris medium, compared to 1.31 g/l produced by Pseudomonas aeruginosa ATCC 9027 on the same medium. The most promising strain and medium, Pseudomonas aeruginosa ATCC 9027 and MSM+Tris medium, were evaluated under controlled conditions in an instrumented bioreactor. Nearly double the rate of growth and production were obtained in the bioreactor, indicating that production time can be shortened considerably under controlled conditions. However, when compared to shake flask studies, only a 4 % increase in growth and a 5 % increase in rhamnolipid production were achieved in the bioreactor, indicating that the yield was limited by the media components or process conditions. With media containing hexadecane as sole carbon source, negligible rhamnolipid production was achieved. Slow growth was observed and the stationary phase had not been reached even after 2 weeks of growth. It was shown that in glucose media rhamnolipid production only commenced in the stationary phase. Since the stationary phase was not reached during growth on hexadecane, rhamnolipids, which are known to increase the availability of alkanes through emulsification and solubilisation, could not be produced. A strategy was devised to accelerate growth on alkane media. A dual substrate medium containing both glucose and hexadecane was investigated. It was hypothesised that growth would be promoted by glucose leading to rhamnolipid production, which would then increase the uptake of hexadecane. Rhamnolipid was produced in the dual substrate experiments, but the hexadecane uptake was still poor. This was suggested to be due to the exposure of the cells to glucose in the inoculum or test flask, which hampered the ability of the cells to utilise hexadecane. It was reasoned that the ability to utilise hexadecane was determined by the cell hydrophobicity, which was influenced by the exposure to hydrophilic or hydrophobic substrates. Rhamnolipids from Pseudomonas aeruginosa ATCC 9027 and ATCC 27853 were shown to have antibacterial activity against Mycobacterium aurum. The largest zone of clearing of 45 mm was obtained with 4 g/l rhamnolipid from Pseudomonas aeruginosa ATCC 9027. The activity was shown to be directly related to the rhamnolipid concentration, highlighting the importance of maximising the biosurfactant yield when developing a process for the production of rhamnolipids as antimicrobial agents. Antifungal activity tests against Botrytis cinerea were inconclusive. Future studies should expand the antimicrobial application of rhamnolipids by testing their activity against a larger range of target organisms. In order to maximise the rhamnolipid yield in future studies, a fed batch process is proposed which would increase the cell density thereby increasing rhamnolipid production and prolonging the stationary phase, which was found to be the phase associated with rhamnolipid production. Different feeding strategies should be investigated, depending on the kinetics of substrate consumption. It is desirable to feed the smallest volume of substrate that is necessary with a high concentration in order to keep the dilution rate low and maximise the product concentration. A factorial design is recommended for this purpose. Further studies with alkanes as carbon source should be conducted using strains that have been maintained and cultured on media containing alkanes as sole carbon source. Alternative biosurfactant producing strains should also be investigated, which have higher natural cell hydrophobicities.

Thesis (MScEng (Process Engineering))--University of Stellenbosch, 2006. / Earthquakes and other seismic events are known to have catastrophic effects on people and property. These large-scale events are almost always preceded by smallerscale seismic events called precursors, such as tremors or other vibrations. The use of precursor data to predict the realization of seismic hazards has been a long-standing technical problem in different disciplines. For example, blasting or other mining activities have the potential to induce the collapse of rock surfaces, or the occurrence of other dangerous seismic events in large volumes of rock. In this study, seismic data (T4) obtained from a mining concern in South Africa were considered using a nonlinear time series approach. In particular, the method of surrogate analysis was used to characterize the deterministic structure in the data, prior to fitting a predictive model. The seismic data set (T4) is a set of seismic events for a small volume of rock in a mine observed over a period of 12 days. The surrogate data were generated to have structure similar to that of T4 according to some basic seismic laws. In particular, the surrogate data sets were generated to have the same autocorrelation structure and amplitude distributions of the underlying data set T4. The surrogate data derived from T4 allow for the assessment of some basic hypotheses regarding both types of data sets. The structure in both types of data (i.e. the relationship between the past behavior and the future realization of components) was investigated by means of three test statistics, each of which provided partial information on the structure in the data. The first is the average mutual information between the reconstructed past and futures states of T4. The second is a correlation dimension estimate, Dc which gives an indication of the deterministic structure (predictability) of the reconstructed states of T4. The final statistic is the correlation coefficients which gives an indication of the predictability of the future behavior of T4 based on the past states of T4. The past states of T4 was reconstructed by reducing the dimension of a delay coordinate embedding of the components of T4. The map from past states to future realization of T4 values was estimated using Long Short-Term Recurrent Memory (LSTM) neural networks. The application of LSTM Recurrent Neural Networks on point processes has not been reported before in literature. Comparison of the stochastic surrogate data with the measured structure in the T4 data set showed that the structure in T4 differed significantly from that of the surrogate data sets. However, the relationship between the past states and the future realization of components for both T4 and surrogate data did not appear to be deterministic. The application of LSTM in the modeling of T4 shows that the approach could model point processes at least as well or even better than previously reported applications on time series data.

Thesis (MScEng (Process Engineering))--University of Stellenbosch, 2009. / ENGLISH ABSTRACT: The study of water-in-oil emulsion stability and separation was carried out for this thesis. The main objectives were as follows: to rank crude oil samples in terms of creating stable emulsions; to assess the effect of the brine pH on emulsion stability; to investigate the influence of different organic acids on emulsion stability; and to determine the efficiency of an electric separator in removing water droplets from a flowing organic liquid. Seven crude oil samples from different sources such as A, C, H, M, P, U, and V were used to investigate the water-in-crude-oil emulsion. Two crude oil blends were also used. Brine solution comprising 4 wt% NaCl and 1 wt% CaCl2 was used. In this study the gravity settling, critical electric field (CEF) and centrifuge test methods were used to estimate the emulsion stability created by the crude oil and crude oil blend samples. The experiments were carried out at 60°C. In the gravity test method, the brine pH, stirring speed, stirring time and water-cut (the fraction of water in the emulsion) were changed in 2IV-1 factorial design. The parameters for the centrifuge and CEF test methods were selected on the basis of the gravity test method. The crude oil samples were ranked in terms of creating stable emulsion in the following order V, U, P, H, A, M and C. The crude oil blends created more stable emulsions than their respective constituents. The ranking order of the crude oil samples did not correlate to asphaltenes, resins, wax or total acid number (TAN). There was a good correlation between the test methods used. There was an increase and decrease in the brine pH when different crude oil samples were in contact with the brine. It is believed that the structure of the surfactants present in crude oil may explain the emulsion-forming characteristics of different crude oil deposits around the world. To account for the effect of organic acids on emulsion stability, different organic acids were used. In this case, a mixture of equal volumes of heptane and toluene (here referred to as heptol) was used as the model for crude oil. The brine solution composition was the same as the one used in the crude oil experiments. Equal volumes of heptol and brine were mixed for a period of time and then separated. The brine pH was changed from acidic to basic. In this regard, gas chromatography and liquid chromatography were used to analyse the concentration of the acids in the brine and heptol samples. It was found that the partitioning coefficient for acids containing a straight-chain hydrocarbon moiety decreased with an increase in molecular weight. However, the partitioning coefficient depended on the structure of the acid. The presence of a benzene ring in the organic acid increased the partitioning coefficient. Organic acids with rings created an interface layer when the heptol sample was mixed with basic brine solution. This confirmed that the emulsion of water and crude oil starts with the formation of a film, and it also provides insight into the formation of naphthenate soap. It is believed that the naphthenic acids that cause stable emulsions have rings. More organic acids should be tested. It is recommended that the interaction of asphaltenes, resins and naphthenic acids should be investigated at different pH levels, temperatures and pressures. The separation of water droplets from a flowing organic liquid was carried out using a direct current (d.c.) electric separator. The separator used centrifugal forces and a d.c. electric field to enhance the removal of water drops from a flowing organic liquid. For this, vegetable oil, crude oil blend and heptane were used as the continuous phase. The experiments were carried out at room temperature (for heptane and vegetable oil) and at 70°C (for vegetable oil and crude oil blend). The flow rate to the separator was kept constant. The separator removed water droplets from flowing organic liquids. A maximum of 97% (at 100 V)of water droplets was removed from the heptane liquid; a maximum of 28% (at 100 V) of water droplets was removed from the vegetable oil at 70°C and 5% (at 100 V) of water droplets was removed from the crude oil blend. The d.c. electric field enhanced the efficiency of the separator in removing water droplets. The break-up of the droplets is suspected to decrease the efficiency of the separator. This separator can easily be installed into existing process lines and does not require much space. However, further improvements are needed in the design of this separator. Emulsions created in the petroleum industries are quite complex to deal with. The identification of the structure of the components in crude oil is a matter that still has to be investigated. An improvement in the techniques may lead to a better understanding of the cause of the ultra-stable emulsion encountered in the petroleum and related industries. / AFRIKAANSE OPSOMMING: Die studie van die stabiliteit en skeiding van water-in-olie-emulsies is vir hierdie tesis uitgevoer. Die hoofdoelstellings was as volg: om ruolie-monsters in terme van die skepping van stabiele emulsies te klassifiseer; om die effek van die pekel-pH op emulsie-stabiliteit te assesseer; om die invloed van verskillende organiese sure op emulsie-stabiliteit te ondersoek; en om die doeltreffendheid van ’n elektriese skeier in die verwydering van waterdruppels uit ’n vloeiende organiese vloeistof te bepaal. Sewe ruolie-monsters uit verskillende bronne soos was A, C, H, M, P, U en V gebruik om die water-in-ruolie-emulsie te ondersoek. Twee ruolie-mengels is ook gebruik. ’n Pekeloplossing wat 4 wt% NaCl en 1 wt% CaCl2 bevat, is gebruik. In hierdie studie is die gravitasie-afsakkings-, kritieke elektriese veld- (KEV-) en sentrifuge-toetsmetodes gebruik om die emulsie-stabiliteit te beraam wat deur die ruolie- en ruolie-mengsel-monsters geskep is. Die eksperimente is teen 60°C uitgevoer. In die gravitasietoetsmetode is die pekel-pH, roertempo en watersnyding (die fraksie van water in die emulsie) is in ‘n 2IV-1-faktoriaalontwerp ondersoek. Die parameters vir die sentrifuge- en KEV-toetsmetodes is op grond van die gravitasietoetsmetode resultate gekies. Die ruolie-monsters is in terme van die skepping van ’n emulsie stabiliteit geklassifiseer in die volyende orde V, U, P, H, A, M, en C. Die rudie-menysels het meer stabiele emulsies gerorm as die respektiewe samestellende dele. Die rangorde van emulsie stabiliteit van die ruolie-monsters het nie met asfaltene, hars, waks of totale suurgetal gekorreleer nie. Daar was ’n goeie korrelasie tussen die toetsmetodes wat gebruik is. Daar was ’n toename of afname in die pekel-pH wanneer verskillende ruolie-monsters in kontak met die pekel was. Die aanname is dat die struktuur van die surfaktante wat in die ruolie teenwoordig is, die emulsievormende karaktereienskappe van verskillende ruolie-neerslae regoor die wêreld kan verklaar. Om die effek van organiese sure op emulsie-stabiliteit te verklaar, is verskillende organiese sure gebruik. In hierdie geval is ’n mengsel van gelyke hoeveelhede heptaan en tolueen (voortaan verwys na as heptol) as die model vir ruolie gebruik. Die pekeloplossing-samestelling was dieselfde as die een wat in die ruolie-eksperimente gebruik is. Gelyke hoeveelhede heptol en pekel is vir ’n tydperk gemeng en toe geskei. Die pekel-pH is van suurvormend tot basies verander. Gaschromatografie en vloeistofchromatografie is gebruik om die konsentrasie van die sure in die pekel- en heptoloplossings te analiseer. Daar is gevind dat die verdelingskoëffisiënt vir sure wat ’n reguitketting-koolwaterstofhelfte bevat met ’n toename in molekulêre gewig afneem. Die verdelingskoëffisiënt het egter van die struktuur van die suur afgehang. Die teenwoordigheid van ’n benseenring in die organiese suur het die verdelingskoëffisiënt verhoog. Organiese sure met ringe het ’n tussenvlaklaag geskep toe die heptolmonster met die basiese pekeloplossing gemeng is. Dit het bevestig dat die emulsie van water en ruolie met die vorming van ’n vlies begin, en gee ook insig in die vorming van naftenaatseep. Dit blyk dat die naftenaatsure wat stabiele emulsies veroorsaak, ringe het. Meer organiese sure moet getoets word. Daar word aanbeveel dat die interaksie van asfaltene, hars en naftenaatsure teen verskillende pH-vlakke, temperature en drukke getoets word. Die skeiding van waterdruppels uit ’n vloeiende organiese vloeistof is uitgevoer met behulp van ’n gelykstroom- elektriese skeier. Die skeier het sentrifugiese kragte en ’n wisselstroomelektriese veld gebruik om die verwydering van waterdruppels uit ’n vloeiende organiese vloeistof te verhoog. Hiervoor is plantolie, ’n ruoliemengsel en heptaan gebruik as die deurlopende fase. Die eksperimente is teen kamertemperatuur (vir heptaan en plantolie) en teen 70°C (vir plantolie en ruolie-mengsel) uitgevoer. Die vloeitempo na die skeier is konstant gehou. Die skeier het waterdruppels uit die vloeiende organiese vloeistowwe verwyder. N’ maksimum van 97% (by 100 V) van die water drupples is verweider van die heptaan vloeistof; a maksimum van 28% (by 100 V) van die water druppels was verweider van die plantolie by 70°C en 5% (by 100 V) van die water druppels was verweider van die rudie mengsel. Die gelykstroom- elektriese veld het die doeltreffendheid van die skeier om waterdruppels te verwyder, verhoog. Daar word vermoed dat die afbreek van die waterdruppels die doeltreffendheid van die skeier verlaag. Die skeier kan met gemak in bestaande proseslyne geïnstalleer word en benodig nie veel spasie nie. Verdere verbeterings is egter nodig ten opsigte van die ontwerp van hierdie skeier. Emulsies wat in die petroleumbedrywe geskep word, is kompleks om te hanteer. Die identifikasie van die struktuur van die komponente in ruolie verg verdere ondersoek. ’n Verbetering in hierdie tegnieke kan tot beter begrip lei van die oorsaak van die ultrastabiele emulsie wat in die petroleum- en verwante bedrywe aangetref word.