Thesis (MScEng)--University of Stellenbosch, 2001. / ENGLISH ABSTRACT: Liquid-liquid extraction is a branch of solvent extraction that employs addition of an immiscible solvent, as a separating agent, to a liquid feed. Various types of equipment can be used, however if the process requires more than three stages, typically an extraction column, operated in a countercurrent manner would be employed. In order to scale-up and design a commercial extraction column, it is necessary to quantify the extraction system hydrodynamics and mass transfer characteristics. The principal objectives of countercurrently operated extraction columns concern the mass transfer rate and permissible throughput. The performance of a countercurrent extraction column can be adversely affected by axial mixing, which disturbs countercurrent plug flow. Various methods have been devised whereby it is possible to evaluate the performance of a column extraction and scale it up to ensure that the commercial operation achieves the same separation achieved on a pilot scale. Classical axial dispersion models allow quantification of axial mixing and mass transfer rates. Two Backflow models were derived to describe the performance of a Vibrating Plate Extraction (VPE) Column, one for steady state and one for unsteady state operation. The steady state model consisted of a series of simultaneous equations, which were solved using the Excel solver function. The unsteady state model consisted of 54 ordinary differential equations, which were solved stagewise using a fourth order Runge Kutta procedure. The steady state model was based on a dissociation extraction process, whereby meta-cresol (mcresol) was separated from para-hydroxy-benzaldehyde (PHB). The process used the differing de-protonation constants of the two components and the fact that the solubility of the ionic species of each was low in the organic solvent. The extraction system was quantified using a combination of acid-base and extraction theory. Experimentally determined concentration profiles, measured along the length of the column, were force fitted to the model, thereby allowing determination of the model parameters. The mass transfer coefficients ranged between 0.0098 and 0.189 Imin, and it was found that backmixing of the dispersed phase was negligible, while that of the continuous phase was low (varying between 0 and 0.3). The unsteady state model, used to describe the dynamic response of a VPE, was based on a system whereby tert-butyl hydroquinone (TBHQ) was recovered from a purge stream. Conductivity measurements of the raffinate were used to determine the residence time distribution in the column, and hence allowed determination of the extent of axial mixing. It was preferable that the column be operated with minimum settler volumes, otherwise buffering in the settlers occurred, thereby masking axial mixing effects. This method did not facilitate accurate determination of backmixing, at least two other conductivity measurements in the column needed to be measured. / AFRIKAANSE OPSOMMING: Vloeistof-vloeistofekstraksie is 'n vertakking van oplosmiddelekstraksie wat gebruik maak van die toevoeging van 'n onmengbare oplosmiddel as 'n skeidingagent tot die vloeistofvoer. Verskeie tipes apparaat kan gebruik word, maar as die proses meer as drie stadia vereis, sal 'n ekstraksiekolom, in teenstroom bedryf, tipies gebruik word. Om dit moontlik te maak om 'n kommersiele ekstraksiekolom te skalleer en te bedryf, moet die ekstraksiesisteem se hidrodinamika en massa-oordragkarakteristieke gekwantifiseer word. Die hoof doelwitte van ekstraksiekolomme wat teenstroom bedryf word, gaan om die massa-oordrag en toelaatbare deurset. Die skeidingsdoeltreffendheid van 'n teenstroom ekstraksiekolom kan nadelig beinvloed word deur aksiale vermenging, wat teenstroom propvloei versteur. Verskeie metodes is voorgestel wat dit moontlik maak om die doeltreffendheid van 'n ekstraksiekolom te evalueer en te verseker dat dieselfde skeiding verkry word vir 'n kornmersiele aanleg as vir 'n loodsaanleg. Klassieke aksiale dispersiemodelle laat kwantitatiewe berekening van aksiale vermenging en massa-oordragtempos toe. Twee terugvloeimodelle is afgelei om die werksverrigting van 'n Vibrerende Plaat Ekstraksiekolom (VPE) te beskryf. Die gestadidge toestand model bestaan uit 'n stelsel gelyktydige vergelykings wat opgelos is d.m.v. Excel. Die ongestadige toestand model bestaan uit 54 gewone differensiaalvergelykings, wat stapsgewys opgelos is d.m.v. die vierde orde Runge-Kutta metode. Die gestadigde teostand model is gebaseer op 'n dissosiasie ekstraksieproses, waardeur m-kresol geskei is van p-hidroksiebensaldehied (PHB). Die proses maak gebruik van die verskillende protoneringskonstantes van die twee verbindings en die feit dat die oplosbaarheid van beide die ioniese spesies laag is in die organiese oplosmiddel. Die ekstraksiestelsel is gekwantifiseer deur gebruik te maak van 'n kombinasie van suur-basis- en ekstraksieteorie. Die model is gepas op eksperimenteel bepaalde konsentrasieprofiele, gemeet langs die lengte van die kolom. Die massa-oordragkoeffisiente het waardes aangeneem tussen 0.0098 en 0.189 Imin en daar is gevind dat die terugvermenging van die verspreide fase weglaatbaar was, terwyl die van die kontinue fase laag was (tussen 0 en 0.3). Die ongestadige toestand model wat gebruik is om die dinamiese respons van die VPE te beskryf, is gebaseer op 'n stelsel waar tert-butielhidrokinoon (TBHQ) herwin is vanuit 'n bloeistroom. Geleidingsmetings van die raffinaat is gebruik om die residensietydverspreiding in die kolom te bepaaI en het derhalwe toegelaat dat die mate van aksiale vermenging bepaaI kon word. Die kolom moet by voorkeur met minimale skeiervolumes bedryf word, anders is daar 'n buffereffek in die skeiers, wat die aksiale vermenging verskuiI. Hierdie metode Iaat nie die akkurate bepaling van terugvermenging toe nie en minstens twee ander geleidingsmetings in die kolom was benodig.

The Fluidised Bed Combustion (FBC) technology has been widely used internationally for power generation. This technology has good fuel flexibility and reduced S02 emissions with dry sorbent (Limestone or Dolomite) addition. South Africa has large reserves of coals that are difficult to combust in conventional pulverised fuel fired boilers. These reserves could be potential feedstocks for new build FBC boilers. The chemical composition of these coals is site specific and could have an impact on the combustion performance of the fuel. This necessitates the need for FBC coal tests in the presence of a sorbent. The objectives of this study were to investigate the changes in the production NO" SO" and the combustion efficiency of the three test coals under conditions of fluidised bed combustion, with the same sorbent. Tests with no sorbent were performed to evaluate the coals inherent calcium capabilities of capturing sulphur. Tests with varying ratios of sorbent were performed to evaluate the sorbent's capabilities for further levels of in-bed desulphurisation. The experimental equipment used in this investigation was the Eskom Fluidised Bed Test Facility (FBTF). This facility is a bubbling fluidised bed combustorlgasifier. The investigated bed temperature range was between 800 to 900°C, in intervals of 20°C. The operating pressure was 50kPa (gauge). The three coals were compared at CalS molar ratio of 1. Carbon in ash has shown to decrease with an increase in bed temperature for Coal A, Band C. The best performing coal in terms of least quantity of remaining carbon in ash was Coal A. The NO emissions increased for an increase in bed temperature for Coal A, Band C. The greatest NO emissions were recorded during Coal B tests. The N 20 emissions decreased with an increase in bed temperature for Coal A and B tests. Higher N 20 emissions were observed for Coal B than Coal A tests. In terms of S02 retention Coal C performed the best. The optimal operating bed temperature for S02 retention observed for the three coals was in the region of 800-860°C. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2007.

This study focuses on the effect that particle properties (size, density and shape) have on fluidized bed hydrodynamics and entrainment rate. The experimental work was carried out using two Plexiglas columns having internal diameters of 0.05 m and 0.14 m respectively and a total height of 6 m from the gas distributor. The particle density was in the range 1300 to 4600 kg/m3, the Sauter mean diameter was in the range 23 to 60 um and the fines content (% < 22 um) was in the range 1 to 29 %. Particle shapes, which ranged from angular to spherical, were characterized by image analysis of SEM photographs. Air was used, as the fluidizing gas and the superficial velocity was kept constant at 0.38 m/s in the 0.05 m column. In the 0.14 m column it was varied in the range 0 to 0.8 m/s. The dense-phase voidage, bubble fraction and entrainment rate of the powders were measured at ambient conditions. In general it was found that the bubble fraction and entrainment rate increased with an increase in the superficial gas velocity. The dense phase voidage was found to increase with an increase in the fines content of the powder and it was only a weak function of the superficial gas velocity. Most importantly, it was found that angular-shaped particles had a higher dense phase voidage, a lower bubble fraction and a lower entrainment rate when compared to spherical-shaped particles having similar particle density and size. Possible reasons for the lower entrainment rate for the angular-shaped particles are given. The measured dense phase voidage, bubble fraction and entrainment flux was compared with predictions from published correlations and it was found that none of the correlations provided a good fit to the data obtained in this work and that different correlations predicted widely different entrainment rates for the same system. It is therefore recommended that literature correlations should be used with caution in the absence of experimental data. Empirical correlations for the dense phase voidage and bubble fraction are developed. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2007.

Sulphur dioxide (S02) is an atmospheric pollutant that has the ability to negatively impact on local vegetation, farming activities and human health. South Africa's coal fired power stations release this pollutant into the atmosphere during the combustion of coal. Current coal fired power stations operating in South Africa are not required to install any form of S02 removal equipment however, the new Air Quality Act to be implemented in South Africa could change this situation. The use of Fluidised Bed Technology with the addition of limestone or dolomite (sorbent) has the ability to absorb and convert S02 from a gaseous phase into a solid phase for easy disposal. The objective of this study was to evaluate potential commercial sorbent sources in South Africa that could potentially be used for the reduction of S02 released into the atmosphere during fluidised bed combustion of coal. Eight commercially mined sorbents within a two hundred kilometre radius of large economically mineable coalfields were selected. The study was divided into two parts in order to identify any possible links between the physical and chemical composition of the sorbents and their performance under fluidised bed combustion conditions. In Part 1, the chemical composition of the sorbents was determined by X-Ray Fluorescence (XRF) and X-Ray Diffraction (XRD) analysis. The sorbents hardness property was determined by Hardgrove Grindability Index (HGI) testing. The physical structure of the sorbent was analysed by both Petrographical and Scanning Electron Microscope (SEM) analysis of the original/parent sorbents. In Part 2, S02 absorption capability by the sorbents was determined through batch tests conducted in a 1.6m high stainless steel, 10kW electrically heated Atmospheric Fluidised Bed Reactor (AFBR). Three different bed temperatures (800, 850 and 900°C) and three different particle size ranges (425-500, 600-710 and 850-lOOOllm) were tested for each of the eight sorbents. The highest Maximum Sulphur Retention for all of the sorbents was found to occur at a temperature of 850°C and at the smallest particle size tested, 425-500llm. The best desulphurisation sorbent of the eight sorbents tested was found to be Sorb1 with a S02 Maximum Sulphur Retention of 92.30% and a Removal Efficiency of 84.54%. Additional tests were also performed on the sorbents to get a better understanding of their desulphurisation ability. For the area calculation on the performance test graphs, it was found that the sorbent that produced the best S02 removal efficiency was not necessarily the sorbent that had the highest maximum sulphur retention. For varying quantities of sorbent added to the AFBR, it was found that each sorbent had an optimum quantity that produced the best removal efficiency. However, for desulphurisation beyond certain limits any further increase in the amount of sorbent added to the AFBR resulted only in a marginal increase in the sorbent's S02 removal. The calcium and magnesium composition of the sorbents was found to have no noticeable influence on the sorbents ability to reduce S02. The silica and inherent moisture content of the sorbent showed signs whereby an increase in their compositions produced an increase in desulphurisation. The Hardgrove Grindability Index of the sorbents indicated that the softer the sorbent, the better the S02 reduction. The petrographical analysis performed on the eight sorbents showed no obvious reason for the difference between the sorbents ability to remove S02. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, 2006.

The South African based petrochemical company, SASOL, operates three large plants for the recovery and purification of the chemical 1-hexene. The thermodynamic models available in commercial simulation packages fail to predict or correlate the plant data presently observed by SASOL. The focus of this project is the accruement of comprehensive and accurate modelling parameters that would assist SASOL in optimizing the operation of the three plants and meet their purity specifications. The experimental requirements of the project are the measurement of isothermal vapour-liquid equilibrium (VLE) data for selected binary systems, using a dynamic still. The binary systems investigated were 1-hexene + n-methylpyrrolidone (NMP) at 40, 62 and 90 °C, water + NMP at 70, 90 and 107 °C and 1-hexene + 3methylcyclopentene (3MCP) at 40, 50 and 60 °C. With respect to the modelling of the VLE data, the combined (gamma-phi) and direct (phi-phi) regression procedures were utilized. The results of the analysis show the combined method as the more flexible of the two, when used for low pressure systems. The excess Gibbs energy correlations investigated were the Margules, Van Laar, Wilson, NRTL and UNIQUAC. The NRTL and Van Laar models dominated the modelling results across the range of temperatures for each binary system and for both the direct and combined methods of data regression. The experimental data for the systems of water (1) + NMP (2) at 107 °C and at 70 °C were compared to literature data. The first system showed excellent correlation with the literature results while the second plot at 70 °C showed a positive bias of the experimental data between xj of 0.3 and 0.8. Thermodynamic consistency tests for the VLE data are also required to verify the accuracy of the data. For this project, the point and direct (Van Ness) consistency tests were used as the area test was considered as too mild. All systems passed the point and direct tests for the combined method and therefore verify the thermodynamic consistency of the experimental data. The systems failed in most cases for the direct method as the combined method is the more flexible of the two modelling methods. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2007.

Ion exchange for the recovery of metals from solutions is a well established process. It features significantly in terms of being able to recover valuable substances from what would otherwise be waste streams as well as recovering substances that could be harmful to the environment if left in the waste stream. The more popular options for ion exchange processes could be batch, fixed bed, fluidized, moving bed, and chromatographic columns. Although most ion exchange processes tend to be batch processes making use of the fixed bed columns, technological developments enable the use of fluidized beds to be explored. The main purpose of this research was to compare the performance of a fixed bed ion exchange system with a fluidized ion exchange system for the recovery of copper from a cupric sulphate solution. By experimentation the bed depth required for each type of equipment (in order to achieve a specified percentage recovery of copper from a specified feed) was determined. The comparative advantage of one type of equipment over the other ensures the correct type of system to be used for a sulphate solution of a particular concentration. This study provides a basis for comparative studies of contacting equipment for the removal of other substances from dilute solutions. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2005.

Sappi's Ngodwana integrated Kraft pulp and paper mill was used as case study for the application and evaluation of the water pinch technique. The technique of water pinch originates from energy pinch, but uses mass flow and contaminant concentration to identify water and effluent reduction opportunities. The classical meaning of pinch, as defmed by energy pinch has however been changed to a more modem meaning. Historically the terms water or energy pinch was used to refer to the points where two composite curves touched on energy or water graphs. This graphical meaning of pinch is gradually being replaced to refer to the optimal po int proposed by a numerical solver beyond which improvement of the water network is no longer possible for the given inputs. The water pinch technique was applied by means of a numerical solver that used mixed integer non-linear programming to optimise to the minimum cost for running the water network under investigation. The problem defmition was defined in terms of costs associated with the use of utilities, raw material, treatment facilities and process units. It was also possible to define factors such as environmental impact, corrosion, fouling, scaling, cooling tower treatment cost, legal risk etc in terms of a penalty cost. The water pinch technique has been refined in software packages that are user friendly, ca~ble of handling multi-contaminants and suitable for varying flows. The software package WaterPinch by Linnhoff March was used. The case study was applied on Ngodwana mill that has an already highly closed water system with effluent generation rates as low 20 kL per ton of pulp and paper. The pinch study included sodium, chloride, calcium, suspended solids and COD as contaminants. The study investigated different applications of the pinch technique. The following was concluded: • The mill's understanding of its current restrictions, ()l pinch points, of its water network was confirmed. No new pinch points have been identified of which the mill was not aware. This indicates that the mill was already highly knowledgeable about its water system. This was expected of a mill that has a very low specific-effluent-generation rate. Water pinch was unable to significantly improve on the effluent generation rate of the mill. • The pinch analysis has identified opportunities of mixing small quantities of waste streams into process water streams to replace fresh water. These changes can introduce minor water savings and new risks to the process that have to be understood better before implementation. • The mill has progressed far with the design and costing of a proposed effluent treatment plant (ERPl). The integration of this treatment plant into the water network was investigated using the pinch technique. The pinch solver suggested a totally different approach to the integration of the ERPI plant compared to design of the mill. The mill's design revolves around the treatment of low chloride streams in the ERPl plant and using of the treated water as make-up to the cooling towers. Sodium was recovered as raw material from the cooling towers' blow-down. Pinch proposed treatment of the high cWoride containing streams and returning the streams to users suitable of using high chloride water. The network proposed by mill's design generates 8.2 MLlday effluent less than the pinch proposal, and recovers sodium as raw material. The proposal presented by pinch is not recommended and points to the difficulty in simulating factors, such as raw material recovery, in a pinch analysis. • Users for the excess storm water were identified using water pinch and will be suitable for implementation. The mill has mwever decided on alternative sinks for the storm water based on considerations such as process inter-dependency, risks associated with contamination and general management philosophy for the different systems in the mill. These considerations could have been included into the pinch solver, but were not because it was of interest what the second best option would be. • The pinch investigation proved useful to confirm certain understandings of the mill. The investigation confirmed the difficulty of improving the water systems of the mill due to the fact that Ngodwana is already a highly closed and integrated mill. Numerous smaller process changes have been identified by the pinch solver and could be investigated further for smaller process improvements. • It i; recommended that pinch technology be applied again when the mill plans to make major process changes or expansions. It is also recommended to use water pinch on a more frequent basis in smaller sections of the mill or for other evaluations in the mill. As a group Sappi could benefit from the use of water pinch, especially in situations where the water network of the mill is not already water efficient. • The recommendations and conclusions in this report have not been subjected to technical and economical feasibility studies. Extensive further studies must be conducted before implementation of any of the results. Further studies must include impacts from process dynamics, long term effects, impacts from other contaminants that have not been simulated, etc. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2009.

VLE data is essential for the design and optimisation of industrial separation processes. Carboxylic acids are of significant interest because of their importance in both industrial and biological processes. Carboxylic acids are used as raw materials for a wide range of products, which include soaps , detergents, nylon , biodegradable plastics, medical drugs and food additives, They are also used both as solvents and as additives and co solvents under a wide range of conditions. Carboxylic acids exhibit strong self and cross -association through hydrogen bonds in both liquid and vapour phases. A thorough understanding of how these molecules interact both with themselves and with other solvents becomes necessary if existing processes may be optimised and new processes developed. Vapour-liquid equilibrium (VLE) data were measured for carboxylic acid systems ranging from C3 to C6• New vapour-liquid equilibrium data were measured for the following binary carboxylic acid systems: • Propionic acid + Hexanoic acid at 20 kPa, 403.15 K, 408.15 K and 413.15 K. • Isobutyric acid + Hexanoic acids at 20 kPa, 413.15 K and 423.15 K. • Valerie acid + Hexanoic acid at 15 kPa, 423 .15 K and 433.15 K. • Hexanoic acid + Heptanoic acid at 10 kPa and 443.15 K. A highly refined dynamic VLE Still by Raal (Raal and Muhlbauer [1998]) was used to undertake the VLE measurements. The still was operated either isothermally or isobarically using a computer control scheme. The isobaric and isothermal control was measured to be ±0.03 kPa and ±0.02 K respectively. The experimental procedure was verified with the highly volatile cyclohexane (l) + ethanol (2) system. The cyclohexane (l) + ethanol (2) measured VLE data was found to be in good agreement with that of Joseph (2001) and passed both the direct test and point test for thermodynamic consistency. A high degree of confidence was then placed on the equipment set-up and experimental procedure, as well as the new carboxylic acids VLE data . The VLE data for all the systems measured were modelled. Two data reduction methods were used: I. The combined ( r- ¢ ) method u. The direct method (¢ - ¢ ) method. In the combined method, the vapour phase non-idea lity was corrected using the Pitzer-Curl (1957) correlation and the Hayden and O' Connell (1975) chemical theory approach. Three liquid phase activity coefficient models were used namely the Wilson, NRTL, and UNIQUAC equations. The Peng-Robinson equation of state (Peng and Robinson [1976]) in combination with the Twu and Coon mixin g rule was used in the direct method. Thermodynamic consistency tests were done on all the systems measured. The point test (Van Ness et a!. [1973]) and the direct test Van Ness ([ 1995]) were used for consistency tests . The direct test could not be carried out on the carboxylic acids data because of the model's inability to adequately characterise the experimental activity coefficients. Generally the models fitted the data well but failed to accurately predict the "S" shape of the carboxylic acids phase diagrams. Considerable work still exists for further investigation into carboxylic acids. Currently, many experimentalists are working in this area . Penget a!. (2004) present ed their progress on developing an equation of state incorporating chemical theory to specifically handle carboxylic acids at the ICCT conference in Beijing, 2004. Raal and Clifford (University of Kwa-Zulu Natal, Thermodynamics Research Unit) are currently developing activity coefficient models incorporating chemical theory for a binary mixture of carboxylic acids. This work is part of the continuing research to under stand the phase behaviour of carboxylic acids. / Thesis (M.Sc.)-University of KwaZulu-Natal, Durban, 2005.

The use of black liquor gasification as an alternative to conventional chemical and energy recovery systems for spent liquors is an area of particular interest to the pulp and paper industry. The motivation to explore this technology is to improve the thermal efficiency of the recovery process by utilizing the energy content of the spent black liquor more effectively and provide chemical recovery for sodium and sulphur containing liquors for a local pulp and paper mill. A study of the available gasification technologies showed that the steam reforming process marketed by ThermoChem Recovery International is particularly suited to the mill in that it can handle a change to a sulphite pulping chemistry and also handle silica removal which is an inherent problem with the bagasse raw material that the mill uses. However the technology required further development and confirmation of process suitability before implementation at the mill. This aim of this project was to build and operate a gasifier based on the TRI concept to determine if this process is suitable for recovery of SASAQ black liquor from bagasse pulping. This included gaining an understanding of the process variables like the black liquor solids composition and the non-process element levels and required carrying out a mass balance on inorganic components across the reactor as well. The focus of this investigation was primarily on the front end of the project and entailed basic and detailed design of a pilot gasification unit. The pilot unit was subsequently constructed, commissioned and operated to prove the unit met the design intent. Preliminary results showing the conceptual proof of the technology are presented as well as performance tests showing the unit capability of gasifying a 3.1 1Ihr 60% solid content black liquor feed. Problematic areas that could influence the design of a scale-up unit were identified and highlighted for further development, with proposed solutions. / Thesis (M.Sc.)-University of KwaZulu-Natal, Durban, 2005.

The dynamic still originally designed by Raal (Raal and Muhlbauer [1998]) has been transformed into a valuable still that can now be used for measuring both low pressure vapour-liquid equilibria (VLE) for systems that are completely miscible and vapour-liquid-liquid equilibria (VLLE) for systems that are partially miscible. The resultant equilibrium data are important in the design and analysis of distillation and allied separation process equipment, with VLLE data, in particular, being useful in the design of heterogeneous azeotropic distillation columns. The original Raal still was based on the designs of Heertjies [1960] and Yerazunis [1964], who successfully used a packed equilibrium chamber where the liquid and vapour phases are forced downward co-currently to achieve rapid and dynamic equilibrium (Joseph et al [2001]). Direct analysis of the vapour composition prior to condensation through a new heated valving system with superheated sample conveyance to a gas chromatograph, a modification incorporated into the Raal still, ensured that accurate and reproducible equilibrium data were obtained. This new arrangement dispenses with the impossible task of getting the actual vapour composition that would result were the vapours allowed to condense and form two liquid phases. The initial testing of the still which established the operating procedures was conducted on two previously measured systems - the first which was homogeneous and the second heterogeneous. For the homogeneous system the new vapour sampling system was tested by comparing the measured composition to that of a condensed sample sent manually to the GC using a gas-tight syringe. In order to completely describe the VLLE for the systems studied, the liquid-liquid equilibrium (LLE) data for these systems were also measured. The LLE measurements were conducted in a newly-developed small jacketed glass cell with temperatures maintained constant by circulating water from a bath maintained at the desired temperatures. The main focus of this project was thus the development of an apparatus and procedures for measuring low pressure vapour-liquid-liquid equilibria. The project also went on to measuring and modeling VLE, VLLE and LLE data for selected binary and ternary systems. Both the Gamma-Phi and the Phi-Phi methods of VLE analysis were carried out on the measured data. The NRTL, Wilson, TK-Wi1son and UNIQUAC activity coefficient models were used in the Gamma- Phi method together with the Virial equation of state for vapour phase non-idealities. In the Phi-Phi method, The Peng and Robinson Equation of State (EOS), the Soave Redlich-Kwong EOS and the Stryjek and Vera modified Peng and Robinson EOS were all used, first with the classical mixing rules and then with the theoretically correct Wong and Sandler [1992] mixing rules. Ternary LLE binodal curves were correlated to the Hlavaty correlation, the beta function and the log gamma function while the corresponding tie-lines were fitted to the NRTL model. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2005.