Thesis (MScEng)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: Desalination, by means of reverse osmosis (RO), in combination with other processes, can produce potable water at high recoveries. Antiscalants are generally used to reduce scaling on equipment surfaces and to improve water recovery during RO by slowing down the precipitation kinetics of sparingly soluble salts in the RO feed, thereby allowing concentration levels in the RO brine at several times the solubility limit of these salts. In addition, a fraction of the concentrate may be recycled back to the feed of the RO-membrane to improve the overall recovery, but only after the super saturated salts in the concentrate have been precipitated. The inhibitory character of the antiscalants (which are rejected into the concentrate stream) complicates the removal of salt from the concentrate and therefore prohibits such recycling. The focus of this study is aimed at properly understanding some of the parameters that influence the functionality or effectiveness of antiscalants used in high sulphate waters, with the purpose to override the effect of the antiscalant in the concentrate stream and force precipitation of the super saturated salts in solution. A batch crystallization technique, which considers the precipitation of calcium sulphate dehydrate (gypsum) from a solution of changing super saturation, was used to perform precipitation tests 1) on synthetically prepared solutions, super saturated with gypsum and 2) industrial concentrate, rich in sulphate (produced by concentrating acid mine drainage (AMD) by means of a lab scale RO unit). During batch crystallization, the precipitation process was observed by means of monitoring the depletion of calcium, using a calcium selective electrode (ISE). Deductions concerning the kinetics of precipitation were made from observing two kinetic variables (response variables) e.g. the induction time and the growth rate (tC80 – inferential variable). Two antiscalants have been evaluated in this study: a phosphonate based antiscalant (HYDREX) and a polyacrylate antiscalant (BULAB), at concentrations of 4 mg/l and 12 mg/l. The objective was to chemically and physically manipulate the antiscalant effectiveness, override its effect and force precipitation of gypsum by means of changing parameters in the system, such as the temperature (15°C- 25°C), pH (4-10), ferric chloride concentration (2-10 mg/l) or seeding the solution with gypsum seed at a concentration of 0-2000 mg/l. In addition, lime and a combination of gypsum and lime were also used for seeding at concentrations of 2000 mg/l. The induction time, prior to precipitation, was found to be most strongly affected by the change in seed concentration and pH at a given antiscalant concentration. Seed at a concentration of 2000 mg/l was sufficient in most cases to immediately override the effect of HYDREX and BULAB (at 4-12 mg/l) and produce ~ 0 minutes induction time. A pH of 10 increased the adsorption capacity of HYDREX and BULAB, leading to longer induction times (exceeding 24 hours in some cases). At a pH of 4 the adsorption capacity was very low for both HYDREX and BULAB (lower) leading to shorter induction times (zero to 100 minutes). It was especially in the ‘no-seed’ cases that the effect of pH on the induction time was prominent. The rate of precipitation (crystal growth rate) was increased at a temperature of 25°C, compared to 15°C (the rate increased two fold for an increase in 10°C). The addition of lime-seed, instead of gypsum, (at 2000 mg/l) produced growth rates, two times higher compared to when gypsum was used at the same conditions. In Addition, seeding with lime produced induction times (150 minutes for HYDREX and 50 minutes for BULAB) prior to precipitation, compared to zero induction time when gypsum was used at the same conditions. It was proven that an induction time could be eliminated by adding a combination of gypsum and lime both at a concentration of 2000 mg/l. with the added benefit of the higher growth rate. An increase in the calcium concentration increased the crystal growth rate in the presence of HYDREX. The presence of a high pH, however caused the effect of calcium on the growth (in the presence of BULAB) to be overshadowed. At a higher pH the growth rate of gypsum slowed down as a result of the increase in adsorption capacity of the polymer onto the crystal surface. The interaction of the antiscalant with FeCl3 seemed to be important with regard to crystal growth. Higher ferric concentrations (10 mg/l) were sufficient to limit the inhibitory effect of 12 mg/l antiscalant (HYDREX and BULAB) on the crystal growth rate. Conversely, low ferric concentration resulted in slower growth rates in the presence of an antiscalant. The best conditions (within the scope of the current study), sufficient 1) to override the inhibitory effect of antiscalants (HYDREX and BULAB) and 2) to produce rapid precipitation of gypsum, lie in the use of seeding with gypsum and lime (2000 mg/l), adding ferric chloride (10 mg/l), lowering the pH to 4 or lower (which can only be obtained when lime is not added) and setting the solution temperature to a moderate value of 25°C or higher. These ‘best’ conditions were subsequently applied to a concentrate, produced from concentrating AMD in a RO unit, and proved to be even more successful in overriding the effect of HYDREX and BULAB than in synthetic aqueous solutions. The induction times of precipitation of AMD in all cases were ~ 0 minutes, whereas the growth rate increased threefold compared to the synthetic tests. The presence of additional foreign precipitates of aluminum, calcium and magnesium as well as an increased [SO4ª-] x [Caª+] product of 3.73 (AMD concentrate) vs. 3.46 (synthetic solutions) is thought to be responsible for the increase in precipitation kinetics when only gypsum seed was used. The addition of lime caused an increase in the precipitation potential of the brine by increasing the calcium concentration. Although the addition of lime caused an increase in the pH to 12.3 (at which point the antiscalant was most effective), the increase in pH is likely to cause an increase in the natural carbonate in the water, which would stimulate CaCO3 precipitation. The CaCO3 precipitate would be responsible for the adsorption of antiscalants, reducing their efficiency. / AFRIKAANSE OPSOMMING: Ontsouting by wyse van tru-osmose (TO), in samewerking met ander prosesse, kan help om drinkwater te lewer teen verhoogte herwinning. Tipies word antiskaalmiddels gebruik om bevuiling op die oppervlak van toerusting te verminder en terselfdetyd herwinning te verhoog deurdat dit die presipitasiekinetika van superversadigde soute in die TO voerwater vertraag. Dit lei daartoe dat water (superversadig met soute) deur die membraansisteem kan beweeg, sonder om bevuiling te veroorsaak. ‘n Breukdeel van die konsentraat kan herwin word na die TO voer om sodoende die algehele waterherwinning te verhoog. Dit kan egter eers gebeur nadat die soute in die konsentraat neergeslaan en verwyder is. Die inhirente ‘vertragingskarakter’ van antiskaalmiddels (wat ook in die konsentraat stroom beland) kompliseer die verwydering van sout vanuit die konsentraat en verhoed so herwinning. Die fokus van hierdie studie is daarop gemik om die parameters wat die funksionaliteit of effektiwiteit van antiskaalmiddels (wat in sulfaatryke waters gebruik word), beter te verstaan. Die doel is daarop gemik om die betrokke antiskaalmiddel se effek te kanselleer asook presipitasie van die superversadigde soute in oplossing aan te help. ‘n Lot (‘batch’) kristallisasietegniek wat die presipitasie van kalsiumsulfaatdehidraat (gips) beskou vanuit ‘n oplossing waar die konsentrasie verander soos presipitasie plaasvind, is gebruik om presipitasietoetse uit te voer 1) op oplossings wat sinteties versadig is met gips en 2) op sulfaatryke AMD (gekonsentreer met behulp van ‘n laboratoriumskaal TO eenheid). Die presipitasie proses is in elke geval waargeneem, deur die vermindering van die kalsium konsentrasie in die oplossing dop te hou, met die gebruik van ‘n kalsiumselektiewe elektrode. Afleidings rakende die kinetika van presipitasie is gemaak deur twee responsveranderlikes dop te hou: die induksietyd en die kristal groeitempo (tC80). Twee antiskaalmiddels by ‘n konsentrasies van 4 dpm (deetjies per miljoen) en 12 dpm is evalueer: ‘n fosfonaat (HYDREX) and poliakrilaat (BULAB). Die doel was om die antiskaalmiddel se werking chemies en fisies te manipuleer, hul werking teen te werk en presipitasie van gips te forseer. Die manipulasie het geskied deur die volgende parameters te verander: temperatuur (15°C-25°C), pH (4-10), FeCl3 (2-10 mg/l) of saad byvoeging (gips: 2000 mg/l). Kalsiumhidroksied (gebuste kalk) en ‘n kombinasie van gips en gebluste kalk is ook gebruik by konsentrasies van 2000 mg/l. Die induksietyd (by ‘n spesifieke antiskaalmiddel konsentrasie) is die sterkste beïnvloed deur ‘n verandering in saad konsentrasie en pH verandering. In die meeste gevalle was ‘n saad konsentrasie van 2000 mg/l voldoende om die induksie effek van beide HYDREX en BULAB te vernietig en nulminute induksietyd is verkry. ‘n pH van 10 het gelei tot die verhoging van die adsorpsiekapasiteit van HYDREX en BULAB wat gelei het tot langer induksietye (in sommige gevalle het dit 24 uur oorskry). By ‘n pH van 4 was die adsorpsie kapasiteit van beide antiskaalmiddels baie laag (laer vir BULAB) en induksie-tye is beperk tot 100 minute. Dit is veral wanneer geen saad toegevoeg is nie wat die effek van pH prominent was. Die tempo van presipitasie was verhoog by ‘n temperatuur van 25°C (2 keer hoër as by 15°C). Die byvoeging van gebluste kalk teen 2000 mg/l het ‘n kristal groeitempo, 2 keer hoër as in die teenwoordigheid van gips gelewer. Gebluste kalk saad byvoeging het egter gelei tot ‘n indukisetyd (150 minute vir HYDREX en 50 minute vir BULAB). Hierdie probleem is oorkom deur ‘n kombinasie van gips en gebluste kalk te gebuik teen ‘n konsentrasie van 2000 mg/l. Geen induksie tyd is waargeneem met die voordeel van ‘n hoër presipitasietempo (kristal groei). ‘n Verhoging van kalsium konsentrasie verhoog die kristal groei tempo in die teenwoordigheid van HYDREX. Nietemin, die invloed van pH oorskadu die invloed van kalsium op die groei tempo (in die teenwoordigheid van BULAB). By ‘n hoë pH word die kristal groei tempo vertraag as gevolg van die verhoging van die adsorpsiekapasiteit van die antiskaalmiddel. Die interaksie van FeCl3 met die antiskaalmiddel blyk van belang te wees. By hoë FeCl3 konsentrasies (10 dpm), is die werking van beide HYDREX en BULAB (12 dpm) beperk. Die ‘beste’ kondisies (verkry binne die konteks van hierdie studie), 1) om die vertragingseffek van HYDREX en BULAB teen te werk en 2) spoedige presipitasie van gips te bewerk, lê in die gebruik van saad (gips en gebluste kalk teen 2000 mg/l), die byvoeging van FeCl3 (10 mg/l), ‘n lae pH (4 of laer, wat natuurlik net tersprake is wanneer slegs gips as saad gebruik word aangesien geluste kalk die pH sal lig) asook ‘n relatiewe hoë temperatuur (25°C). Hierdie ‘beste’ kondisies is toegepas in AMD konsentraat om die effek van HYDREX en BULAB te vernietg en gips te presipiteer en die gevolg was dat dit selfs meer suksesvol was as in sintetiese oplossings. In elke geval is die induksietyd na nul minute toe verminder, terwyl die kristal groei tempo 3 maal verhoog het in vergelyking met die sintetiese toetse. Die teenwoordigheid van onsuiwerhede insluitende aluminium, kalsium, magnesium sowel as ‘n verhoging in die [SO4ª-]x[Caª+] produk (3.73 teenoor 3.46 vir sintetiese toetse), blyk verantwoordelik te wees vir die versnelling van die kinetika. Met die byvoeging van gebluste kalk is dit waarskynlik dat die verhoging van die pH (12.3) lei tot die verhoging van natuurlike karbonate in die water wat weer CaCO3 stimlueer. Die teenwoordigheid van CaCO3 kan verantwoordelik gehou word vir bykomende nukleasie en groei, sowel as die deaktivering van antiskaal effektiwiteit.

Thesis (MScEng)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: Increasing oil prices along with the climate change threat have forced governments, society and the energy sector to consider alternative fuels. Biofuel presents itself as a suitable replacement and has received much attention over recent years. Thermochemical conversion processes such as pyrolysis is a topic of interest for conversion of cheap agricultural wastes into clean energy and valuable products. Fast pyrolysis of biomass is one of the promising technologies for converting biomass into liquid fuels and regarded as a promising feedstock to replace petroleum fuels. Corn residues, corn cob and corn stover, are some of the largest agricultural waste types in South Africa amounting to 8 900 thousand metric tonnes annually (1.7% of world corn production) (Nation Master, 2005). This study looked at the pyrolysis kinetics, the characterisation and quality of by-products from fast pyrolysis of the corn residues and the upgrading of bio-oil. The first objective was to characterise the physical and chemical properties of corn residues in order to determine the suitability of these feedstocks for pyrolytic purposes. Secondly, a study was carried out to obtain the reaction kinetic information and to characterise the behaviour of corn residues during thermal decomposition. The knowledge of biomass pyrolysis kinetics is of importance in the design and optimisation of pyrolytic reactors. Fast pyrolysis experiments were carried out in 2 different reactors: a Lurgi twin screw reactor and a bubbling fluidised bed reactor. The product yields and quality were compared for different types of reactors and biomasses. Finally, a preliminary study on the upgrading of bio-oil to remove the excess water and organics inorder to improve the quality of this liquid fuel was performed. Corn residues biomass are potential thermochemical feedstocks, with the following properties (carbon 50.2 wt. %, hydrogen 5.9 wt. % and Higher heating value 19.14 MJ/kg) for corn cob and (carbon 48.9 wt. %, hydrogen 6.01 wt. % and Higher heating value 18.06 MJ/kg) for corn stover. Corn cobs and corn stover contained very low amounts of nitrogen (0.41-0.57 wt. %) and sulphur (0.03-0.05 wt. %) compared with coal (nitrogen 0.8-1.9 wt. % and sulphur 0.7-1.2 wt. %), making them emit less sulphur oxides than when burning fossil fuels. The corn residues showed three distinct stages in the thermal decomposition process, with peak temperature of pyrolysis shifting to a higher value as the heating rate increased. The activation energies (E) for corn residues, obtained by the application of an iso-conversional method from thermogravimetric tests were in the range of 220 to 270 kJ/mol. The products obtained from fast pyrolysis of corn residues were bio-oil, biochar, water and gas. Higher bio-oil yields were produced from fast pyrolysis of corn residues in a bubbling fluidised bed reactor (47.8 to 51.2 wt. %, dry ash-free) than in a Lurgi twin screw reactor (35.5 to 37 wt. %, dry ash-free). Corn cobs produced higher bio-oil yields than corn stover in both types of reactors. At the optimised operating temperature of 500-530 °C, higher biochar yields were obtained from corn stover than corn cobs in both types of reactors. There were no major differences in the chemical and physical properties of bio-oil produced from the two types of reactors. The biochar properties showed some variation in heating values, carbon content and ash content for the different biomasses. The fast pyrolysis of corn residues produced energy products, bio-oil (Higher heating value = 18.7-25.3 MJ/kg) and biochar (Higher heating value = 19.8-29.3 MJ/kg) comparable with coal (Higher heating value = 16.2-25.9 MJ/kg). The bio-oils produced had some undesirable properties for its application such as acidic (pH 3.8 to 4.3) and high water content (21.3 to 30.5 wt. %). The bio-oil upgrading method (evaporation) increased the heating value and viscosity by removal of light hydrocarbons and water. The corn residues biochar produced had a BET Brynauer-Emmet-Teller (BET) surface area of 96.7 to 158.8 m2/g making it suitable for upgrading for the manufacture of adsorbents. The gas products from fast pyrolysis were analysed by gas chromatography (GC) as CO2, CO, H2, CH4, C2H4, C2H6, C3H8 and C5+ hydrocarbons. The gases had CO2 and CO of more than 80% (v/V) and low heating values (8.82-8.86 MJ/kg). / AFRIKAANSE OPSOMMING: Die styging in olie pryse asook dreigende klimaatsveranderinge het daartoe gelei dat regerings, die samelewing asook die energie sektor alternatiewe energiebronne oorweeg. Biobrandstof as alternatiewe energiebron het in die afgope paar jaar redelik aftrek gekry. Termochemiese omskakelingsprosesse soos pirolise word oorweeg vir die omskakeling van goedkoop landbou afval na groen energie en waardevolle produkte. Snel piroliese van biomassa is een van die mees belowende tegnologië vir die omskakeling van biomassa na vloeibare brandstof en word tans gereken as ’n belowende kandidaat om petroleum brandstof te vervang. Mielieafval, stronke en strooi vorm ’n reuse deel van die Suid Afrikaanse landbou afval. Ongeveer 8900 duisend metrieke ton afval word jaarliks geproduseer wat optel na ongeveer 1.7% van die wêreld se mielie produksie uitmaak (Nation Master, 2005). Hierdie studie het gekk na die kinetika van piroliese, die karakterisering en kwaliteit van by-produkte van snel piroliese afkomstig van mielie-afval asook die opgradering van biobrandstof. Die eerste mikpunt was om die fisiese en chemiese karakteristieke van mielie-afval te bepaal om sodoende die geskiktheid van hierdie afval vir die gebruik tydens piroliese te bepaal. Tweendens is ’n kinetiese studie onderneem om reaksie parameters te bepaal asook die gedrag tydens termiese ontbinding waar te neem. Kennis van die piroliese kinetika van biomassa is van belang juis tydens die ontwerp en optimering van piroliese reaktore. Snel piroliese ekspermente is uitgevoer met behulp van twee verskillende reaktore: ’n Lurgi twee skroef reaktor en ’n borrelende gefluidiseerde-bed reaktor. Die produk opbrengs en kwaliteit is vergelyk. Eindelik is ’n voorlopige studie oor die opgradering van bio-olie uitgevoer deur te kyk na die verwydering van oortollige water en organiese materiaal om die kwaliteit van hierdie vloeibare brandstof te verbeter. Biomassa afkomstig van mielie-afval is ’n potensiële termochemiese voerbron met die volgende kenmerke: mielie stronke- (C - 50.21 massa %, H – 5.9 massa %, HHV – 19.14 MJ/kg); mielie strooi – (C – 48.9 massa %, H – 6.01 massa %, HHV – 18.06 MJ/kg). Beide van hierdie materiale bevat lae hoeveelhede N (0.41-0.57 massa %) and S (0.03-0.05 massa %) in vergelyking met steenkool N (0.8-1.9 massa %) and S (0.7-1.2 massa %). Dit beteken dat hieride bronne van biomassa laer konsentrasies van swael oksiedes vrystel in vergelyking met fossielbrandstowwe. Drie kenmerkende stadia is waargeneem tydens die termiese afbraak van mielie-afval, met die temperatuur piek van piroliese wat skuif na ’n hoër temperatuur soos die verhittingswaarde toeneem. Die waargenome aktiveringsenergie (E) van mielie-afval bereken met behulp van die iso-omskakelings metode van TGA toetse was in die bestek: 220 tot 270 kJ/mol. Die produkte verkry deur Snel Piroliese van mielie-afval was bio-olie, bio-kool en gas. ’n Hoër opbrengs van bio-olie is behaal tydens Snel Piroliese van mielie-afval in die borrelende gefluidiseerde-bed reakctor (47.8 na 51.2 massa %, droog as-vry) in vergelyking met die Lurgi twee skroef reakctor (35.5 na 37 massa %, droog as-vry). Mielie stronke sorg vir ’n hoër opbrengs van bio-olie as mielie strooi in beide reaktore. By die optimum bedryfskondisies is daar in beide reaktor ’n hoër bio-kool opbrengs verkry van mielie stingels teenoor mielie stronke. Geen aansienlike verskille is gevind in die chemise en fisiese kenmerke van van die bio-olie wat geproduseer is in die twee reaktore nie. Daar is wel variasie getoon in die bio-kool kenmerkte van die verskillende Snel Piroliese prosesse. Snel piroliese van mielie-afval lewer energie produkte, bio-olie (HVW = 18.7-25.3MJ/kg) en bio-kool (HVW = 19.8-29.3 MJ/kg) vergelykbaar met steenkool (HVW = 16.2-25.9 MJ/kg). Die bio-olies geproduseer het sommige ongewenste kenmerke getoon byvoorbeeld suurheid (pH 3.8-4.3) asook hoë water inhoud (21.3 – 30.5 massa %). Die metode (indamping) wat gebruik is vir die opgradering van bio-olie het gelei tot die verbetering van die verhittingswaarde asook die toename in viskositeit deur die verwydering van ligte koolwaterstowwe en water. Die mielie-afval bio-kool toon ’n BET (Brunauer-Emmet-Teller) oppervlakte area van 96.7-158.8 m2/g wat dit toepaslik maak as grondstof vir absorbante. The gas geproduseer tydens Snel Piroliese is geanaliseer met behulp van gas chromotografie (GC) as CO2, CO, H2, CH4, C2H4, C2H6, C3H8 and C5+ koolwaterstowwe. Die vlak van CO2 en CO het 80% (v/V) oorskry en met lae verhittingswaardes (8.82-8.86 MJ/kg).

Thesis (MScEng)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: The difficulties of chemical process control using plain Proportional-Integral- Derivative (PID) methods include interaction of process manipulated- and control variables as well as difficulty in tuning. One way of eliminating these problems is to use a centralized non-linear control solution such as a feed-forward neural network. While many ways exist to train such neurocontrollers, one of the promising active research areas is reinforcement learning. The biggest drawing card of the neurocontrol using reinforcement learning paradigm is that no expert knowledge of the system is neccesary - all control knowledge is gained by interaction with the plant model. This work uses episodic reinforcement learning to train controllers using two types of process model - non-linear dynamic models and non-linear autoregressive models. The first was termed model-based training and the second data-based learning. By testing the controllers obtained during data-based learning on the original model, the effect of plant model mismatch and therefore real-world applicability could be seen. In addition, two reinforcement learning algorithms, Policy Gradients with Parameter-based Exploration (PGPE) and the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) were compared to one-another. Set point tracking was facilitated by the use of integral error feedback. Two control case studies were conducted to test the effectiveness of each type of controller and algorithm, and allowed comparison to multi-loop feedback control. The first is a ball mill grinding circuit pilot plant model with 5 degrees of freedom, and the second a 41-stage binary distillation column with 7 degrees of freedom. The ball mill case study showed that centralized non-linear feedback control using neural networks can improve on even highly optimized PI control methods, with the proposed integral error-feedback neural network architecture working very well at tracking the set point. CMA-ES produced better results than PGPE, being able to find up to 20% better solutions. When compared to PI control, the ball mill neurocontrol solution had a 6% higher productivity and showed more than 10% improvement of the product size set point tracking. In the case of some plant-model mismatch (88% fit), the data-based ball mill neurocontroller still achieved better set point tracking and disturbance handling than PI control, but productivity did not improve. The distillation case study showed less positive results. While reinforcement learning was able to learn successful controllers in the case of no plant-model mismatch and outperform LV - and (L/D)(V/B)-based PI control, the best-performing neurocontroller still performed up to 20% worse than DB-based PI control. Once again, CMA-ES showed better performance than PGPE, with latter even failing to find feasible control solutions. While on-line learning in the ball mill study was made impossible due to stability issues, on-line adaptation in the distillation case study succeeded with the use of a partial neurocontroller. The learner was able to achieve, with a success rate of just over 50%, greater than 95% purity in both distillate and bottoms within 2,000 minutes of interacting with the plant. Overall, reinforcement learning showed that, when there is sufficient room for improvement over existing control implementations, it can make for a very good replacement control solution even when no model is available. Future work should focus on evaluating these techniques in lab-scale control studies. / AFRIKAANSE OPSOMMING: Die probleme van prosesbeheer met behulp van gewone Proporsioneel-Integraal- Afgeleide (PID) metodes sluit interaksie van proses gemanipuleerde- en beheerveranderlikes, sowel as probleme met in-stemming in. Een manier om hierdie probleme te elimineer, is deur ’n gesentraliseerde nie-lineêre oplossing te gebruik, soos ’n vorentoe-gevoerde neurale netwerk. Daar is baie maniere is om sulke neurobeheerders op te lei, waarvan die meer innoverende maniere versterkingsleer is. Die grootste trekpleister van versterkingsleer is dat geen deskundige kennis van die stelsel nodig is nie - alle beheerkennis word opgedoen deur interaksie met die aanleg model. Hierdie werk gebruik episodiese versterkingsleer om beheerders met behulp van twee tipes van prosesmodel op te lei - nie-lineêre dinamiese modelle en nie-lineêre outoregressiewe modelle. Die eerste was model-gebaseerde opleiding en die tweede data-gebaseerde opleiding genoem. Deur die beheerders wat verkry is tydens datagebaseerde opleiding op die oorspronklike model te toets, kon die effek van die verskil tussen aanleg en model gesien word, en ’n aanduiding van werklike wêreld toepaslikheid gee. Twee versterkingsleer algoritmes was met mekaar vergelyk - Policy Gradients with Parameter-based Exploration (PGPE), en die Covariance Matrix Adaptation Evolution Strategy. Stelpunt volging was deur integraalfout-terugvoer gefasiliteer. Twee gevallestudies is uitgevoer om die doeltreffendheid van elke tipe beheerder en algoritme te toets, deur vergelyking met PI terugvoerbeheer. Die eerste is ’n balmeul toetsaanleg met ’n vryheidsgraad van 5 en die tweede ’n binêre distillasie kolom met ’n vryheidsgraad van 7. Die balmeul gevallestudie het getoon dat gesentraliseerde nie-lineêre terugvoerbeheer met behulp van neurale netwerke selfs op hoogs-geoptimeerde PI beheer metodes kan verbeter. In vergelyking met PI beheer, kon die balmeul neurobeheer oplossing ’n 6% hoër produktiwiteit handhaaf en het meer as 10% verbetering in die handhawing van die produkgrootte stel punt getoon. In die geval van ’n 12% aanleg-model verskil, het die data-gebaseerde balmeul neurobeheerder steeds beter stel punt handhawing en versteuring hantering as PI beheer gewys, alhoewel produktiwiteit nie verbeter het nie. In beide gevalle het die integraalfout oplossing sukses getoon, en CMA-ES het tot 20% beter as PGPE gevaar. Die distillasie gevallestudie het getoon dat die sukses van die balmeul gevallestudie nie noodwendig na ander aanlegte uitbrei nie. Alhoewel versterkingsleer in staat was om suksesvolle beheerders in die geval van geen aanleg-model verskil te leer, het die beste presterende neurobeheerder steeds tot 20% swakker as DB-gebaseerde PI beheer gevaar. Weereens het CMA-ES beter as PGPE gevaar, met die laasgenoemde wat selfs nie daarin kon slaag om werkende oplossings te vind nie. Alhoewel onstabiliteit aan-lyn aanpassing in die balmeul gevallestudie onmoontlik gemaak het, is an-lyn aanpassing in die distillasie gevallestudie moontlik gemaak deur die gebruik van ’n gedeeltelike neurobeheerder. Die leerder was in staat om, met ’n slaagsyfer van net meer as 50 %, meer as 95 % suiwerheid in beide uitlaatstrome te bereik in 2,000 minute van die interaksie met die aanleg. Op die ou end het versterkingsleer getoon dat, wanneer daar voldoende ruimte is vir verbetering oor bestaande beheer implementasies, kan dit ’n baie goeie vervanging wees selfs wanneer daar geen model beskikbaar is nie. Toekomstige werk moet fokus op laboratoriumskaal toepassings van hierdie tegnieke.

Thesis (MScEng)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: Crude NiSO4 solutions are often produced as a product of Sherrit based matte leach processes leading to iron and cobalt contaminated solutions of NiSO4. To upgrade the quality of these solutions for either, the production of NiSO4 crystals or cathode/precipitated nickel, the iron and cobalt must be removed. Conventional processes use either pure or saponified Cyanex 272 in solvent extraction to extract iron and cobalt from pregnant nickel leach solutions. These processes require the addition of an alkali like NaOH to neutralise the protons being exchanged for the different metal species since extraction is a strong function of pH. Hence, while removing iron and cobalt from the solution, sodium is added instead. This limitation can be dealt with by pre-loading of Cyanex 272 with a portion of the purified nickel product prior to impurity extraction. During the extraction stage nickel is then exchanged for the impurities instead of hydrogen and no NaOH addition is necessary, resulting in a pure nickel product. The extracted metals are recovered from the organic phase by stripping it with H2SO4 and the organic phase is recycled. A solvent extraction process that purifies an 80 g/l nickel sulfate solution from 1 g/l cobalt and 3 g/l iron was developed. The key variables include pH, organic/aqueous (O/A) mixing ratio, nickel loading, H2SO4 concentration, Cyanex 272 concentration, various organic diluents and temperature. These variables were investigated via preliminary and three 24 factorial design batch experiments which reduced the variables to pH/NaOH addition and O/A ratio for the pre-loading section, nickel loading and O/A ratio for the extraction section and H2SO4 concentration and A/O ratio for the stripping section. The reduced variables were considered in further batch experiments from which the data was used to develop models and design a simulation sheet in Microsoft Excel of the extraction circuit. These final batch tests revealed the conditions that resulted in the purest aqueous nickel product after extraction, where nickel and cobalt can selectively be stripped from iron and what H2SO4 concentration and A/O ratio is necessary to finally strip iron and regenerate the organic phase. Ultimately four tests were conducted on a mixer-settler setup to test the validity of the simulation sheet. The proposed process conditions for the solvent extraction circuit are to use a 20V% Cyanex 272 solution in Shellsol D70 as diluent that are pre-loaded with an 80 g/l purified nickel product solution at an O/A ratio of 25 and a pH of 7 generating a nickel loading of 3.2 g/l for the extraction section. This loading at an O/A ratio of 1.5 produces a high pure nickel aqueous product (0.05 g/l Co, 0.01 g/l Fe). From the generated organic phase co-extracted nickel and cobalt can selectively be stripped with 0.1 M H2SO4 at an A/O ratio of 2.25 followed by an 1 M H2SO4 solution at an A/O ratio of 0.75 to strip the remaining iron. The temperature at all the stages should be controlled between 40-50°C to ensure phase separation. / AFRIKAANSE OPSOMMING: Ongesuiwerde NiSO4 oplossings word gedurig vervaardig as ʼn produk van die Sherrit gebaseerde mat logingsproses wat lei tot yster en kobalt gekontamineerder oplossings van NiSO4. Om die graad van hierdie oplossings op te gradeer vir die produksie van of NiSO4 kristalle of katode/gepresipiteerde nikkel, moet die yster en kobalt eers verwyder word. Meeste prosesse gebruik of suiwer Cyanex 272 of die natrium sout van Cyanex 272 in ʼn vloeistof-vloeistof ekstraksie siklus om yster en kobalt vanuit versadigde nikkel oplossing te ekstraheer. Hierdie prosesse verg die gebruik van ʼn alkali soos NaOH om die protone te neutraliseer wat geruil word vir die verskillende metaal spesies, aangesien ekstraksie ʼn baie sterk funksie van pH is. Dit het tot gevolg dat natrium tot die oplossing bygevoeg word soos wat yster en kobalt verwyder word. Hierdie tekortkoming kan oorkom word deur Cyanex 272 eers te laai met nikkel deur ʼn porsie van die gesuiwerde nikkel produk te gebruik voor die onsuiwerhede geëkstraheer word. Gedurende die ekstraksie stadium word nikkel dan geruil vir die onsuiwerhede, in plaas van waterstof, en geen NaOH toevoeging is nodig nie. Die gevolg is ʼn suiwer nikkel oplossing. Die geëkstraheerde metaal word herwin vanaf die organies fase deur dit te stroop met ʼn H2SO4 oplossing en die organiese fase kan weer gehersirkuleer word. ʼn Vloeistof-vloeistof ekstraksie proses wat ʼn 80 g/l nikkelsulfaat oplossing van 1 g/l kobalt en 3 g/l yster suiwer, is ontwikkel. Die belangrikste veranderlikes sluit in pH, organies/waterige (O/A) mengingsverhouding, nikkel lading, H2SO4 konsentrasie, Cyanex 272 konsentrasie, verskillende organiese oplosmiddels en temperatuur. Hierdie veranderlikes was ondersoek deur verskeie voorlopige en drie 24 faktoriaal ontwerp enkel-lading eksperimente wat gevolglik die veranderlikes verminder het na pH/NaOH toevoeging en O/A verhouding vir die voorbelading seksie, nikkel lading en O/A verhouding vir die ekstraksie seksie en H2SO4 konsentrasie en A/O verhouding vir die stroping seksie. Die verminderde veranderlikes was verder ondersoek in nog enkel-lading eksperimente, waarvan die data gebruik was om modelle op te stel en ʼn simulasie sigblad in Microsoft Excel te ontwerp van die ekstraksie siklus. Die finale enkel-lading toetse het die kondisies weergegee wat die mees suiwer waterige nikkel produk sal oplewer na ekstraksie, waar kobalt en nikkel selektief van yster gestroop kan word en watter H2SO4 konsentrasie en A/O verhouding nodig is om yster finaal te stroop en die organiese fase te regenereer. Laastens is vier toetse gedoen op ʼn menger-afskeidingstenk opstelling om die betroubaarheid van die simulasie sigblad te toets. Die bedryfstoestande wat voorgestel word vir die vloeistof-vloeistof ekstraksie siklus is om ʼn 20V% Cyanex 272 oplossing in Shellsol D70 as oplosmiddel vooraf te laai met ʼn 80 g/l gesuiwerde nikkel produk oplossing by ʼn O/A verhouding van 25 en ʼn pH van 7 om ʼn nikkel lading van 3.2 g/l vir die ekstraksie seksie te genereer. Hierdie lading by ʼn O/A verhouding van 1.5 produseer ʼn suiwer nikkel waterige oplossing produk (0.05 g/l Co, 0.01 g/l Fe). Vanaf die gegenereerde organiese fase kan die geëkstraheerde nikkel en kobalt selektief gestroop word met 0.1 M H2SO4 by ʼn A/O verhouding van 2.25 gevolg deur ʼn 1 M H2SO4 oplossing by `n A/O verhouding van 0.75 om die oorblywende yster te stroop. Die temperatuur waarby al die stadiums beheer moet word is tussen 40-50°C om fase skeiding te verseker.

Thesis (MScEng)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: At Skorpion Zinc mine, in south-west Namibia, zinc oxide ore is refined through sulphuric acid leaching, solvent extraction, electrowinning and casting of the final 99.995 % Zn metal. Over the last four years, the rare earth element concentrations, with particular reference to Y, Yb, Er and Sc, have significantly increased in the circulating electrolyte and the zinc-stripped organic phase streams in the electrowinning and solvent extraction processes. This project had two main objectives: firstly, the effect(s) of rare earths on the zinc solvent extraction and electrowinning processes were to be determined; based on these results, the second objective was to find a suitable method for removing rare earth elements from the organic phase during zinc solvent extraction. The investigation into the effect of the rare earths on zinc electrowinning showed that an increase of 100 mg/l in the electrolyte Y concentration caused a decrease of 6 % in current efficiency. The elemental order of decreasing current efficiency was found to be: Y > Yb > Er > Sc. In the zinc solvent extraction process, it was found that an increase in the total organic rare earth elements and iron concentration from 3100 to 6250 mg/l resulted in doubled viscosity and an increase in phase disengagement time from 100 to 700 seconds. The organic zinc loading capacity after two extraction stages was reduced by 1 – 3 g/l depending on the pregnant leach solution used. The detrimental effect of rare earth elements on solvent extraction and electroplating of zinc therefore justified the development of a rare earth element removal process. Stripping of low concentrations of rare earth elements from 40% D2EHPA diluted in kerosene to produce a clean organic for zinc extraction was investigated using bench-scale experiments in a glass jacketed mixing cylinder. For the rare earths, the best stripping agent was found to be H2SO4, followed by HCl and then HNO3. Hydrochloric acid achieved better Fe stripping than sulphuric acid. Acid concentration was tested in the range of 1 to 7 M, organic-to-aqueous ratio for the range of 0.25 to 6.0 and temperatures between 30 and 55 °C. More than 80% stripping of yttrium and erbium could be achieved at an optimum hydrochloric acid concentration of 5 M and more than 90% rare earth element (specifically Y, Er, Yb) stripping from the organic phase could be achieved with 5 M sulphuric acid. Stripping was improved by reducing the organic-to-aqueous ratio to as low as 0.5 and increasing the temperature. Stripping increased with increasing temperature in an S-shaped curve, flattening off at 50°C. The effect of O:A ratio was more significant than the effect of temperature on rare earth stripping. The results showed good repeatability, and were not limited by the rare earth concentration, agitation rate or equilibrium time in the range of set points used in the experiments. Statistical models were compiled to fit the experimental data obtained for Y, Yb, Er and Fe when stripped with sulphuric and hydrochloric acid respectively. All models showed dependence on the acid concentration and squared-concentration and interaction effects between the O:A ratio and temperature and stripping agent concentration were significant. The models were compiled for the experimental data obtained from stripping synthetically prepared organic and then tested on results obtained when stripping the plant organic phase. The following three process solutions were discussed for implementation on a plant scale for the removal of rare earths from the organic phase during zinc solvent extraction: Sulphuric acid stripping mixer settler or stripping column, improvement of available HCl stripping section and replacement of the organic inventory. The possibility of an oxalic acid precipitation process to obtain value from the rare earths as by-product was also discussed. It was concluded that the current process that uses HCl to strip off iron and rare earths would be the best practically and financially feasible process. Value can be gained from the rare earths if a rare earth element - oxalic acid precipitation section that is financially feasible can be established. / AFRIKAANSE OPSOMMING: By Skorpion Zinc myn, in suidwes Namibië, word sinkoksied erts gesuiwer deur middel van swaelsuur-loging, oplosmiddel ekstraksie, elektroplatering en gieting van die finale 99.995 % Zn metaal. Oor die afgelope vier jaar het die seldsame aardmetale konsentrasie, spesifiek Y, Yb, Er en Sc, noemenswaardig in die sirkulerende elektroliet en sink-gestroopte organiese fase toegeneem. Hierdie projek het twee hoofdoelstellings gehad: eerstens moes die effek van seldsame aardmetale op die sink oplosmiddel ekstraksie en elektroplatering prosesse bepaal word; gebaseer op hierdie resultate, was die tweede doelstelling om ‘n geskikte metode vir die verwydering van seldsame aardmetale vanaf die organiese fase gedurende sink oplosmiddel ekstraksie te vind. Die ondersoek na die effek van seldsame aardmetale op sink elektroplatering het gewys dat ‘n verhoging van 100 mg/l in die elektroliet Y konsentrasie ‘n verlaging van 6 % in kragdoeltreffendheid veroorsaak het. Die element-orde van verminderende kragdoeltreffendheid was Y > Yb > Er > Sc. Vir die sink oplosmiddel ekstraksie proses, is gevind dat ‘n verhoging in die totale organiese seldsame aardmetaal- en yster konsentrasie van 3100 tot 6250 mg/l ‘n verdubbelde viskositeit en ‘n verlenging in faseskeidingstyd van 100 tot 700 sekondes tot gevolg gehad het. Die organiese sink ladingskapasiteit na twee ekstraksie stappe is met 1 – 3 g/l verminder afhangende van die logings oplossing wat gebruik is. Die nadelige effek van seldsame aardmetale op oplosmiddel ekstraksie en sink elektroplatering het die ontwikkeling van ‘n seldsame aardmetale verwyderingsproses regverdig. Die verwydering van lae konsentrasies seldsame aardmetale vanaf die D2EHPA-keroseen organiese fase om ‘n skoon organiese fase vir sink-ekstraksie te verkry is ondersoek deur banktoetsskaal eksperimente. Vir die seldsame aardmetale is bevind dat H2SO4 die beste stropingsagent is, gevolg deur HCl en dan HNO3. Soutsuur het beter yster verwydering as swaelsuur bewerkstellig. Suurkonsentrasies van 1 tot 7 M, O:A verhoudings van 0.25 tot 6 en temperature tussen 30 en 55°C is getoets. Meer as 80% verwydering van yttrium en erbium kon bereik word met ‘n optimum HCl konsentrasie van 5 M en meer as 90% seldsame aardmetale (spesifiek Y, Er en Yb) verwydering vanaf die organiese fase met 5 M swaelsuur. Seldsame aardmetale en yster verwydering kon verbeter word deur die organies-tot-waterfase (O:A) verhouding te verminder tot so laag as 0.5 en deur die temperatuur te verhoog. Stroping het in ‘n S-kurwe verhoog soos die temperatuur verhoog het, en het afgeplat teen 50°C. Die effek van die O:A verhouding was belangriker as die effek van temperatuur op stroping. Die resultate het goeie herhaalbaarheid gewys, en is nie deur massaoordrag beperk nie. Statistiese modelle is saamgestel om die eksperimentele data wat vir Y, Yb, Er en Fe verkry is vir stroping met swaelsuur en soutsuur te pas. Al die modelle het afhanklikheid van die suurkonsentrasie en kwadratiese suurkonsentrasie gewys en interaksie effekte tussen die O:A verhouding, temperatuur en suurkonsentrasie was belangrik. Die modelle is saamgestel vir die eksperimentele data wat verkry is vanaf stroping van ‘n sintetiese organiese fase en is toe getoets op resultate wat verkry is vanaf stroping van die aanleg se organiese fase. Die volgende drie proses-oplossings is oorweeg vir implementering op ‘n aanlegskaal vir die verwydering van seldsame aardmetale vanaf die organiese fase gedurende sink oplosmiddel ekstraksie: Swaelsuur stroping menger-afskeidingstenk of stropingskolom, verbetering van die bestaande HCl aanleg en vervanging van die organiese inventaris. ‘n Moontlike oksaalsuur presipitasie proses om waarde vanaf die seldsame aardmetale as by-produk te verkry is ook bespreek. Daar is tot die gevolgtrekking gekom dat die huidige proses wat HCl gebruik om yster en seldsame aardmetale van die organiese fase af te stroop, die beste praktiese en finansieel vatbare proses huidiglik is. Waarde kan vanaf die seldsame aardmetale verkry word as ‘n seldsame aardmetale – oksaalsuur presipitasie afdeling wat finansieel haalbaar is daargestel kan word.

Thesis (MScEng)--Stellenbosch University, 2011. / ENGLISH ABSTRACT: According to recent studies, biochar has the potential to improve soil fertility, mitigate climate change, reduce off-site pollution and assist in managing wastes. The application of biochar to soil is not a new concept; Amazonian dark earths are carbon-rich soils with high soil fertility that were created before 1541. Vacuum pyrolysis is a thermo-chemical conversion technique in which biomass is transformed into bio-oil, biochar and non-condensable gas. The objective of this work was to investigate the chemical and physical properties of biochar produced from vacuum pyrolysis of black wattle, vineyard annual prunings and sugar cane bagasse for their potential as soil amendment and adsorbent. The vacuum pyrolysis of black wattle, vineyard prunings and sugar cane bagasse (pyrolysis temperature: 460°C, pressure: 8kPaabs, heating rate: 17°C/min) resulted in biochar yields of 23.5%, 31.0% and 19.7% on a weight basis, respectively. The nature of the biomass had a substantial effect on yields of the products. High ash content combined with high lignin composition led to higher biochar yields for vineyard prunings. The highest surface acidity was observed for sugar cane bagasse (2.3 mmol/g), whereas the lowest surface acidity was observed for vineyard biochar (1.67 mmol/g). Consequently, the pH of the biochars was in the order: vineyard (10.43)> black wattle (9.74)> sugar cane bagasse (6.56). The cation exchange capacities (CEC) of biochars were 122 cmol/kg, 101 cmol/kg and 65 cmol/kg for sugar cane bagasse, black wattle and vineyard, respectively. The electrical conductivities (EC) were highly correlated with feedstock nature. The Ca and K rich vineyard biochar resulted in the highest EC (0.83 dS/m), whilst EC values of black wattle and sugar cane bagasse were 0.67 dS/m and 0.17 dS/m, respectively. Biochars contained substantial amounts of plant-available nutrients, while being low in toxic inorganic content (Pb, As, Cd). The BET surface areas of sugar cane bagasse, black wattle and vineyard were 259 mª/g, 241 mª/g and 91 mª/g, respectively. The adsorption capacity was found to increase with increased contact time and initial solution concentration. The experimental equilibrium time were found to be 3505 min, 1350 min and 150 min for adsorption of 20 mg/L methylene blue solution for vineyard, black wattle and sugar cane bagasse, respectively. Equilibrium data were well fitted to Langmuir and Freundlich isotherms. The maximum adsorption capacities were found to be 15.15 mg/g, 14.49 mg/g and 19.23 mg/g for vineyard, black wattle and sugar cane bagasse when modelled with Langmuir isotherms. The adsorption kinetics was found to follow the pseudo-second order kinetic model. In summary, biochar from sugar cane bagasse is a promising adsorbent for the removal of basic dyes due to its high surface area and microporous structure. This biochar can be applied to slightly acidic soils for nutrient retention and the exchange of nutrients. On the other hand, possessing high amounts of nutrients, biochars from black wattle and vineyard are potential soil amendentment agents. Biochar from black wattle is more beneficial compared to biochar from vineyard due to its higher surface area, microporosity and cation exchange capacity. / AFRIKAANSE OPSOMMING: Volgens onlangse studies, het houtskool die potensiaal om grond vrugbaarheid te verbeter, klimaat verandering te versag, besoedeling te verlaag en ondersteuning te verleen in die bestuur van afval. Die toevoeging van houtskool in grond is nie ‘n nuwe konsep nie; Amazone donker gronde is koolstof ryk gronde met hoë vrugbaarheid wat voor 1541 geskep is. Vakuum pirolise is ‘n termo-chemiese omskakelings tegniek waarin biomassa afgebreek word na bio-olie, houtskool en nie-kondenseerbare gasse. Die doelwit van hierdie werk was om die chemiese en fisiese eienskappe van houtskool, wat geproduseer is deur die vakuum pirolise van swart wattel, jaarlikse wingerd snoeisels, en suikerriet bagasse, vir hulle potensiaal vir grond verbetering en adsorpsie toepassings te ondersoek. Die vakuum pirolise van swart wattel, jaarlikse wingerd snoeisels, en suikerriet bagasse (pirolise temperatuur: 460°C, druk: 8kPaabs, verhittingstempo: 17°C/min) het houtskool opbrengste van 23.5%, 31.0% en 19.7% op massa basis, respektiewelik tot gevolg. Die tipe biomassa het ‘n beduidende effek op die opbrengs van die produkte. Hoë as-inhoud, gekombineer met hoë lignien inhoud, lei tot hoër houtskool opbrengste vir wingerd snoeisels. Die hoogste oppervlak suurheid is gevind vir suikerriet bagasse (2.3 mmol/g), terwyl die laagste waarde gevind is vir die wingerd snoeisels (1.67 mmol/g). Gevolglik, is die pH van die houtskole in die volgorde van: wingerd (10.43) > swart wattle (9.74) > suikerriet bagasse (6.56). Die katioon uitruiling vermoë (CEC) van die houtskole was 122 cmol/kg, 101 cmol/kg and 65 cmol/kg vir suikerriet bagasse, swart wattel en wingerd snoeisels respektiewelik. Die elektriese konduktiwiteite (EC) is gekorreleer met die eienskappe van die biomassas. Die Ca en K ryke wingerd snoeisel houtskool het die hoogste EC waarde (0.83 dS/m) tot gevolg, terwyl die EC waardes vir swart wattel en suikerriet bagasse bepaal is as 0.67 dS/ 0.16 dS/m respektiewelik. Die houtskole het groot hoeveelhede plant-beskikbare voedingstowwe bevat, terwyl dit laag was in toksiese anorganiese stowwe (Pb, As, Cd). Die BET oppervlak areas van suikerriet bagasse, swart wattel en wingerd snoeisels was 259 mª/g, 241 mª/g en 91 mª/g respektiewelik. Daar is gevind dat die adsorpsie kapasiteit toeneem met toenemende kontak tyd met die aanvanklike oplossing. Die eksperimentele ewewigs tye is gevind as 350 min, 1350 min en 150 min vir die adsorpsie van ‘n 20 mg/L metileen blou oplossing vir wingerd snoeisels, swart wattel en suikerriet bagasse, respektiewelik. Die ewewigs data het die Langmuir en Freundlich isoterme goed gepas. Die maksimum adsorpsie kapasiteite is gevind as 15.15 mg/g, 14.9 mg/g en 19.23 mg/g vir wingerd snoeisels, swart wattel en suikerriet bagasse wanneer dit gemodeleer is met Langmuir isoterme. Daar is bevind dat die adsorpsie kinetika ‘n pseudo-tweede orde kintika model volg. In opsomming, houtskool van suikerriet bagasse is ‘n veelbelowende adsorpsie middel vir die verwydering van basiese kleurstowwe, as gevolg van die hoë oppervlak area en mikroporie-struktuur van hierdie houtskool. Dié houtskool kan gebruik word op effense suur gronde vir voedingstof behoud en uitruiling. Aan die ander kant, houtskole van swart wattel en wingerd snoeisels wat hoë hoeveelhede voedingsstowwe bevat, is potensiële grond verbeterings middels. Houtskool afkomstig van swart wattel is meer voordelig as die van wingerd snoeisels, as gevolg van die hoër oppervlak area, mikroporositeit en katioon uitruilings vermoë van die swart wattel houtskool.

Thesis (MScEng (Process Engineering))--Stellenbosch University, 2008. / A substance called CRUD (Chalk River Unidentified Deposit) was deposited together with gold tailings to the East Paydam tailings dam. Previous research conducted on the material has shown that the crud leaches Mn and Ni at concentrations that are above their acceptable risks limits as well as Zn which leaches at concentration slightly below its acceptable limits thereby posing an environmental risk. The main objective of the research was to test the hypothesis stating that soil washing in series with bioremediation can be used to remove the heavy metals from the material from the East Paydam tailings dam. Various laboratory and pilot scale tests were conducted to investigate critical soil washing and bioremediation parameters and their respective influence on the treatment process. Laboratory work involved column tests and batch tests. These tests were crucial in determining the critical parameters for the pilot scale tests such as the selection of the suitable lixiviant from the four that were investigated. The optimal concentration of the lixiviant was also determined together with the optimum soil: liquid ratio. These parameters were employed in the pilot scale tests. Pilot scale tests involved soil washing in series with bioremediation. The bacterial growth over the bioremediation period was also determined. Precipitation of the heavy metals from leachate was investigated by varying the pH and temperature. Results showed that the soil from the East Paydam can effectively be treated by soil washing in series with bioremediation. Oxalic acid was selected for soil washing of the payable slimes at a concentration of 0.001M. However, material that contains high amount of CRUD (deeper down the slime dam) required the relatively concentrated 0.1M oxalic acid and mechanical agitation. Bioremediation was determined to increase the amount of heavy metals that was leached from the material from the East Paydam slimes dam. Precipitation of the heavy metals at a pH of 12 achieved up to 98% removal of heavy metals from leachate.

Thesis (MScEng (Process Engineering))--University of Stellenbosch, 2009. / The abundance of alkane by-products formed in South Africa presents a feedstock opportunity for the production of a wide range of commercially important products, such as long-chain dioic acids and alcohols. These compounds are formed as intermediates through the biological conversion of alkanes, a route which is particularly attractive when compared with chemical conversion due to its operation under milder process conditions. Furthermore, advances in genetic manipulation, which enable the accumulation of a range of metabolic intermediates, make the biological route remarkably flexible. From the literature review Yarrowia lipolytica was identified as a promising organism for use in studying alkane bioconversion because of its ability to produce large quantities of fatty acids when grown on n-paraffins as a sole carbon source. The bioconversion of alkanes will not only depend on the genetic modification but also on the process conditions to maximise growth and bioconversion. The overall objective of this project was therefore to investigate the potential of Y. lipolytica for alkane bioconversion by defining the conditions that maximise both cell growth and bioconversion. The Y. lipolytica strains supplied (TVN348, TVN493 and WT), however, were not yet modified to the extent that accumulation of metabolic intermediates was possible. Use was therefore made of a model system in which the alkane substrate was substituted with an even chain alkylbenzene. Since Y. lipolytica is unable to metabolise the benzene ring, the alkylbenzene is converted to the metabolic intermediate, phenyl acetic acid (PAA), and the potential for bioconversion assessed through measuring the accumulation of PAA. The specific objectives of the project were therefore 1) to define and quantify the parameters for the establishment of an effective model system in shake flasks with respect to trace elements, buffering, added nitrogen, oxygen supply, glucose concentration, alkylbenzene substrate and inducer requirements 2) to use the defined model system to identify the most promising strain of Y. lipolytica TVN348, TVN493 and WT 3) to use the defined model system and selected strain for evaluation of the influence of time of substrate addition and glucose concentration on cell growth and bioconversion of Y. lipolytica under controlled conditions in an instrumented bioreactor Furthermore, since poor reproducibility in cell growth and bioconversion had been prevalent in previous studies, it was also aimed to identify and statistically quantify the reproducibility between duplicate or triplicate samples in each experiment and between sets of different experiments with respect to PAA formation and cell concentrations. Studies were conducted in shake flask cultures to define and quantify the parameters for the model system. The parameters assessed included trace elements, buffering, nitrogen concentration, oxygen supply, glucose concentration, alkylbenzene substrate type and possible inducer requirements. Trace elements, phosphate buffering and added nitrogen did not significantly affect the cell growth of Y. lipolytica TVN348. The cell concentration of Y. lipolytica TVN348 and TVN493 was increased by 65% and 43% respectively for an increase in oxygen supply by decreasing the working volume from 150ml to 50ml, while the cell concentration of Y. lipolytica WT was increased by 41% when oxygen supply was increased by switching from non-baffled to baffled flasks in 50ml cultures. Bioconversion was also increased for an increase in oxygen supply: 2.4mM to 29.0mM PAA (Y. lipolytica TVN348) and 1.2mM to 21.7mM PAA (Y. lipolytica TVN493) for a decrease in working volume; 10.5mM to 46.6mM PAA (Y. lipolytica WT) when switching from non-baffled to baffled flasks. These results indicated that adequate oxygen supply is crucial to both growth and bioconversion, and that further study should be conducted in 50ml working volumes. Cell concentrations obtained in 1.6% (wt/v) and 3.2% (wt/v) glucose cultures (3.95x108cells/ml and 4.03x108cells/ml respectively) indicated that cell growth was neither enhanced nor inhibited by 3.2% (wt/v) glucose. Of the range of substrates examined (propylbenzene, butylbenzene, sec-butylbenzene, hexylbenzene, ethyltoluene and tert-butyltoluene for Y. lipolytica TVN348 and TVN493; octylbenzene and decylbenzene for Y. lipolytica WT), hexylbenzene was regarded as the best substrate for bioconversion (14.7mM and 14.1mM PAA for TVN348 and TVN493 respectively; 42.6mM PAA for WT). Lastly, the absence of a requirement for an additional inducer such as ethanol or oleic acid was confirmed when PAA was formed from hexylbenzene in the culture containing additional glucose (25.0mM). This suggested that when using hexylbenzene as substrate, bioconversion was induced provided sufficient glucose was available for cell maintenance. Results from duplicate or triplicate flasks in each individual shake flask experiment were reproducible and conclusions were based solely on results which showed 95% confidence intervals. However, reproducibility problems were experienced with results between different sets of experiments carried out under the same conditions. The model system was therefore defined by: 1) no addition of trace elements, additional buffering or added nitrogen, 2) cultures grown in 50ml volumes to supply an adequate amount of oxygen crucial for growth and bioconversion, 3) 3.2% (wt/v) glucose and 4) addition of 1% (v/v) hexylbenzene at 24h with no inducer requirements. Use of the model system in shake flask cultures to identify the most promising of the three strains of Y. lipolytica supplied demonstrated that there was no significant difference in cell growth or bioconversion between these strains. Y. lipolytica WT (which has no genetic modifications) was therefore used for further investigation until an appropriate strain could be substituted when it became available. The growth and bioconversion of Y. lipolytica WT was further investigated under controlled conditions in a bioreactor. The influence of time of substrate addition (11h, 24h, 48h) and glucose concentration (3.2% and 6.4% (wt/v)) on growth and bioconversion was examined. When hexylbenzene was added at 48h, cell growth was increased (8.90x108cells/ml) when compared to two of the triplicate cultures with hexylbenzene addition at 24h (4.74x108cells/ml and 3.92x108cells/ml) and the culture with hexylbenzene addition at 11h (2.82x108cells/ml). The third of the triplicate cultures with hexylbenzene addition at 24h, on the other hand, exhibited the strongest growth (2.23x109cells/ml). The poor reproducibility between the triplicate cultures with hexylbenzene addition as 24h made it difficult to determine whether hexylbenzene addition at 24h or 48h maximised cell growth. Furthermore, the cell growth was not significantly improved when the glucose concentration was increased from 3.2% (wt/v) to 6.4% (wt/v) (7.47x108cells/ml for 6.4% glucose culture), however it was also not inhibited. The highest amount of specific PAA formed by Y. lipolytica WT was found when hexylbenzene was added at 11h (7.4x10-11mmol PAA/cell), however the highest accumulated PAA was produced in the culture that exhibited the strongest growth with hexylbenzene addition at 24h (41.4mM). This suggested that the bioconversion of hexylbenzene was maximised when it was added during the active growth phase. It is therefore recommended to conduct fed-batch experiments in future to maintain the active growth phase. Accumulated PAA was increased in 6.4% (wt/v) glucose culture (15.2mM PAA) when compared with two of the 3.2% (wt/v) glucose cultures (5.4mM and 4.3mM PAA). These results indicated that the increased glucose concentration did not inhibit the bioconversion. Furthermore, PAA was formed when 5% (wt/v) residual glucose was observed in the culture, suggesting that the bioconversion of hexylbenzene was not inhibited at glucose concentrations as high as 5.0% (wt/v). If future work were to be conducted in bioreactor culture where glucose is added in fed-batch operation, glucose concentrations in cultures of up to 5% (wt/v) could be considered for initial studies. During bioconversion by Y. lipolytica, the PAA measured after hexylbenzene exhaustion did not, however, correspond to 100% conversion. Further, poor reproducibility was found in the bioreactor cultures. The disappearance of hexylbenzene without a corresponding accumulation of PAA and poor reproducibility was investigated by determining whether PAA was further degraded or alternatively, whether other metabolic intermediates were being formed and accumulated from the hexylbenzene. However, substitution of the hexylbenzene with PAA as substrate confirmed that PAA could not be metabolised. Further, NMR analyses of both the aqueous and organic phases of the culture did not identify any additional metabolic intermediates. It is recommended that additional analyses be conducted on the aqueous and organic phases to further assess the possible accumulation of intermediates. The development of the model system in shake flask cultures demonstrated the importance of adequate oxygen supply for both cell growth and bioconversion. It was also shown that no inducer was needed because hexylbenzene acted as its own substrate inducer. Furthermore, comparison of Y. lipolytica strains TVN348, TVN493 and WT under the defined conditions of the model system revealed that the genetically modified strains (TVN348, TVN493) did not exhibit enhanced bioconversion. Bioreactor cultures using the model system under controlled conditions further showed that bioconversion was not inhibited at a 5% (wt/v) residual glucose concentration and suggested that bioconversion was maximised when hexylbenzene was added during active growth phase. This informs on future work, suggesting fed-batch operation in order to extend the active growth phase, where glucose concentrations in the bioreactor of up to 5% (wt/v) can be considered.

Thesis (MScEng (Process Engineering))--University of Stellenbosch, 2006. / As easy to mine high grade ore bodies are being depleted, many mining industries are experiencing an increasing need to process lower grade ores, and thus the high costs involved in the mineral recovery from these ores (of which comminution energy costs are a large component) are of major concern. It has been estimated that up to 70% of the total energy consumption in mineral processing is used up by comminution processes, which characteristically may have efficiencies of less than 0.1% in terms of the transfer of electrical energy into particle breakage. In many cases, very fine grinding is required to liberate the valuable inclusions in such low grade ores, which also leads to slimes losses of valuable minerals due to the inefficiencies of recovery methods in the ultra-fine size ranges. For many years the use of thermal pretreatment has been suggested as a way to decrease the costs of size reduction, and improve the liberation of valuable minerals in ores to aid later beneficiation technologies, but it was not until exploration into the use of microwaves to selectively heat only some of the minerals in ores, that this form of treatment became economically viable. A low grade copper ore from Palabora was subjected to microwave treatment and then tested for ore strength in a laboratory rod mill, using the developing cumulative size distributions of the rod mill products with time to quantitatively determine the effects of microwave treatment on ore strength. It was seen that after microwave treatment the ore responded more readily to milling, producing a finer grind than for untreated ore at every measured time interval of milling. From this data, comminution models were created to describe the grinding of this ore in various flowsheet simulations. An investigation was also performed to determine the effect of the application of microwave treatment on the liberation of minerals, due to the preferential breakage induced along grain boundaries during the selective thermal expansion of certain mineral inclusions in ores during microwave treatment. To ensure consistency between results for microwave treated and untreated material, it was decided to use the same grinding time for both when preparing ore for the next stage of testing. A grinding time was chosen which would produce an 80% passing size of 800 μm for the microwave treated ore. This time was determined from the previous grinding tests and was found to be approximately 16 minutes. After particle size classification of the mill products through sieving, a size range suitable for gravity separation processes was chosen for sink-float testing, with the aim of investigating whether microwave treatment had liberated enough gangue material at large particle sizes to offer the possibility of removing this hard gangue material early on in the process, before costly fine grinding is required. XRF analysis of the products showed little difference in recoveries of gangue material to the floats between treated and untreated material, and that while most of the copper reported to the sinks, that some of the copper was always entrained in the floats. These losses of valuable minerals to the gravity tailings will lead to overall losses in copper mineral recovery from the plant. QEMSCAN® analysis showed that there was a significant increase in mineral liberation in the size ranges associated with flotation as a result of the microwave treatment. An increase in liberation of the copper minerals which are easily recovered by flotation (i.e. chalcopyrite, cubanite, bornite, chalcocite and digenite) of 8.4% over that of the untreated ore was seen. This indicates that significant increases in copper recovery are possible after microwave treatment, and also that less fine grinding is then required to extract the valuable minerals from the ore, which leads to a reduction in loss of these valuable minerals to slimes. Palabora Mining Company supplied enough data on their plant operations from 1989 to enable models to be built to describe the operation of the mills and classifiers used in their comminution circuit. This data, together with the work performed to compare the performance of microwave treated and untreated Palabora ore in both milling and liberation (which allowed for basic recovery models to be built), allowed flowsheet simulations of the plant operations. Simulations of the plant after the addition of microwave pretreatment of the ore showed that the total energy used in comminuting the ore (including that of the microwave treatment) to the correct size distribution for mineral recovery by flotation were reduced by 19% from that required for untreated ore, and was mainly due to reductions in the circulating loads over the mills. By exploiting the greater milling capacity allowed for by these lower circulating loads, it was shown that it was theoretically possible to obtain increases of up to 46% in maximum throughput after microwave treatment, while retaining the same final grind size in the feed sent to flotation as is required for untreated ore. The addition of gravity separation processes to remove liberated gangue material from the comminution circuit early on, led to further savings in energy and also grinding media, and also decreased the requirements for flotation reagents and smelter fuel later on in the flowsheet. Unfortunately, the losses of entrained copper to the gravity separation tailings were such that overall economic losses were incurred by the operation. It was concluded that when dealing with low grade ores, only the implementation of very efficient and mineral specific separation technologies could make the removal of gangue material at large particle sizes (i.e. > 1 mm) viable. Economic analyses based on the simulations of the plant under various operating conditions showed potential increases in plant profitability after the addition of microwave pretreatment of the ore before milling, and were reported using net present value (NPV) calculations for the plant over a 10 year period with monetary values discounted at 20%. When operating under the same conditions and throughput as in the 1989 data provided by Palabora Mining Company, an increase in the NPV of the plant of 23% over that for the reported operation was seen after the addition of microwave pretreatment, and an increase of 72% in NPV given a 10% increase in throughput which is made possible by microwave pretreatment of the ore. In real money terms, after 10 years of operation the increase in NPV of the plant with the addition of microwave pretreatment of the ore was seen to be around R259 million (under the conditions reported for the plant operation in 1989), and around R795 million if the 10% increase in throughput which is only made possible by microwave pretreatment is realized. Current conditions at Palabora are very different from those supplied by the plant for the operation in 1989, however, as the mining operation has since been moved underground resulting in the throughput of the plant being greatly reduced, with the consequence that the plant is currently operating at a loss. Palabora mining company posted a net loss of R158 million over the 6 months leading up to June 2004, while an economic analysis of the proposed addition of microwave pretreatment of the ore at an increased throughput of 10% made possible by this treatment, indicated that a loss of only R138 million would have been incurred over the same 6 month period had this been implemented. Thus, while benefits from the introduction of microwave pretreatment of the ore before milling can still be seen under the operating conditions of the plant during the time period investigated, these alone would not have be able to bring the plant to profitable operation.

Thesis (MScEng (Process Engineering))--Stellenbosch University, 2008. / The general objective for this thesis is to assess the ability of cellular automata to model relatively complex processes or phenomena, in particular thermodynamic scenarios. The mass transfer in packing materials of distillation columns was selected as an example due to the sufficient level of complexity in the distillation process, and its importance in a wide range of applications. A literature survey on cellular automata that summarizes the information currently available in formal publications and the internet is included to provide a general overview on the basic theoretical principles and the application of cellular automata models in the process engineering industry. The literature study was also used to identify potential requirements for the new research project. The study objective includes the construction of a cellular automata model that is able to represent transition of solutes from the fluid on the micro-surfaces of packing materials to the by-passing vapour stream, as well as the steady-state equilibrium between evaporation and condensation. Iterated model parameters sufficient for the realistic modelling of mass transfer as a result of thermodynamic driving forces, are required to meet this objective. The model behaviour was compared and the parameters subsequently adjusted according to the behaviour that is theoretically expected from the system being simulated. Qualitative (although sometimes in a quantitative format) rather than quantitative observations and comparisons were made seeing that the model has not yet been calibrated. The model that has been developed to date is not able to simulate the individual effects of chemical and thermodynamic properties although a realistic simulation of the cumulative effect exerted by these factors, or change thereof, on a system has been achieved. The accuracy of the results that have been obtained by using iterated parameters cannot be guaranteed for scenarios that deviate too much from the systems that have already been modelled successfully. The trade-off between the ability of the model to incorporate the effect of polarization, its ability to represent separation, in particular the condensation of hydrophilic substances, for strong hydrophilic packing materials and its ability to incorporate a large number of species limits the range of scenarios that can be successfully modelled. The model is able to represent the effect of a declining driving force (difference between the component vapour pressure of the gas phase and that of the liquid phase) that is typical of a system which is allowed to reach equilibrium after an initial disturbance. The model is also able to represent an additional driving force for separation caused by the effect of intermolecular forces. The model also displays the potential ability to represent the effect of different surface structures of the packing material on the extent of separation achieved at steady state as well as the rate at which such steady state conditions have been achieved. The model must be correctly scaled to minimize inaccurate results. Although several adjustments are needed to eliminate some limitations, the model has proven itself worthy of further development due to its capability to represent the basic characteristics of mass transfer in packing materials.