Browsing by Author "Fourie, Frederick Christiaan van Niekerk"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    The high pressure phase behaviour of detergent range alcohols and alkanes
    (Stellenbosch : Stellenbosch University, 2018-03) Fourie, Frederick Christiaan van Niekerk; Knoetze, J. H.; Schwarz, C. E.; Stellenbosch University. Faculty of Engineering. Dept. of Process Engineering.
    ENGLISH SUMMARY: Supercritical fluids fulfil specialised roles in the food and beverage, petroleum, natural gas, polymers, pharmaceuticals and novel materials development industries. Supercritical fluid extraction (SFE) represents a separations niche and utilises the tuneable properties of a high pressure solvent to execute challenging extractions. One such example is the purification of detergent alcohol product streams. These streams often exhibit a range of carbon backbone lengths and contain significant residual alkanes, linear and branched alcohol isomers. The boiling and crystallisation points of these species are often narrowly distributed or overlap and SFE is considered as an alternative means of product purification. Robust thermodynamic models are key to designing SFE processes but model development is partially reliant on access to accurate and comprehensive high pressure phase equilibria data. Vapour-liquid equilibria (VLE) data of ternary solvent + solute A + solute B systems are particularly useful because the solute A-solute B interaction, a determining factor in mixture phase behaviour, is inherently incorporated. Unfortunately these data remain scarce because the requisite analytical equipment is complex and costly. A high pressure multi-component static analytic phase equilibria setup capable of operation at 150 °C and 300 bar was developed. It features a visual, variable volume equilibrium cell incorporating two ROLSITM samplers. Simultaneous sampling from two phases coupled with concurrent parallel online analysis via gas chromatography (GC) is possible. Additional features include vertical adjustment of both samplers whilst maintaining system pressure, reduced cell dead volume and a high definition camera enabling the observation of internal phenomena not visible with the naked eye. Heating is achieved using jacketed fluid circulation coupled with a forced convection oven. The GC contains two inlets, four columns, two switch valves and three detectors arranged in parallel pathways and is equipped to handle mixture constituents from volatile gases to mid-length hydrocarbons. Adapted manual injection techniques enabled quantitative GC calibration methods that are decoupled from the equilibrium cell. The setup was validated by sampling from a one-phase ternary of known composition and comparison with binary and ternary literature data. Observation of the cell contents has proven essential for accurate vapor phase sampling. Visuals were used to indicate that sampling can disrupt a high-pressure equilibrium system even if pressure and temperature remain constant to within 0.1 bar and 0.01 °C. Such disruptions may manifest in one of three phenomena: global mist formation, localized mist formation, or no-warning droplet formation. In a similar fashion, the impact of temperature gradients and particularly the small scale of the resultant disruptions, are conveyed. The new apparatus was used to study nine ternary mixtures comprising the solvent CO2 and different combinations of the solutes n-dodecane (nC12), 3,7-dimethyl-1-octanol (37DM1O) and 1-decanol (C10OH). Three ternary mixtures – [CO2 + nC12 + 37DM1O], [CO2 + 37DM1O + C10OH] and [CO2 + nC12 + C10OH] – were each investigated at bulk CO2-free solute A:solute B mass ratios of 75:25, 50:50 and 25:75. Phase compositions were measured at 35, 55 and 75 °C and pressures between 68 and 237 bar. The experimental outline and target reduced pressures in particular enable a comparative study using ethane as solvent. Phase behaviour in the two alkane-containing mixtures – nC12 + 37DM1O and nC12 + C10OH – displayed similarities and deviated from behaviour in the alcohol + alcohol mixture, 37DM1O + C10OH. The presence of nC12 led to enhanced solubility in the ternary mixture and, in the high-nC12 region, co-solvency. On Gibbs diagrams this presents as s-shaped convex-to-concave liquid curvature, concave vapour curvature and, at intermediate pressures, a pinched two-phase band. In the CO2 + 37DM1O + C10OH system enhanced solubility or pinched two-phase bands were not detected and liquid curves remained convex throughout. Relative solubility (ij) and separation potential (SPij) were used to evaluate solvent efficacy. The former is an indicator of attainable fractionation sharpness whilst the latter incorporates ij coupled with vapour phase loading of the more soluble solute. The suitability for fractionation using high pressure CO2 decreased in the order [nC12 + C10OH] > [nC12 + 37DM1O] > [37DM1O + C10OH]. In both nC12-containing systems, ij’s and to a large extent also SPij’s were positively correlated with bulk fraction of the less soluble solute species. In CO2 + 37DM1O + C10OH, ij’s were insensitive to solute ratio and, as a result, SPij’s were positively correlated with bulk fraction of the more soluble species. The alkane + alcohol ij compositional dependency may enable optimisation of a SFE process with a variable or manipulable feed composition. Four thermodynamic models were evaluated for their ability to correlate the experimental vapour and liquid phase equilibrium pressures and compositions. The models, presented in order of decreasing performance, were RK-ASPEN, SR-POLAR, PR-BM and PC-SAFT. In the two C10OH-containing systems at 35 °C, all four models essentially failed. Regardless the mixture, RK-ASPEN and PC-SAFT respectively produced the best and worst pressure correlations with percentage average absolute deviations of 3.1, 4.0 and 3.1 % (RK-ASPEN) and 8.5, 18.3 and 12.9 % (PC-SAFT). RK-ASPEN and SR-POLAR were superior at capturing the s-shaped liquid phase complexity and co-solvency pinch in the high-nC12 regions. In general, however, RK-ASPEN produced the best composition estimates, particularly for the vapour phase. Model-predicted relative solubilities and separation potentials were also evaluated. RK-ASPEN and PC-SAFT correctly ranked the systems in terms of suitability for SFE with CO2 and, on average, produced the smallest absolute deviations in ij. Regardless the mixture, SPij estimates via PC-SAFT, surprisingly, were superior. Density inversions or barotropy was detected in the high-C10OH region of both C10OH-containg systems at 35 °C and pressures in the approximate range 180 to 190 bar. Interesting images illustrating surface tension effects at the inverted conditions and a pressure reduction sequence passing through the inversion point are presented. A number of valuable academic contributions have emanated from this work. Content from the equipment design review was published as a chapter in an edited volume and serves as a valuable point of departure for those who intend to construct similar equipment. In: M.R. Belinsky (Ed.), Supercritical Fluids, Nova Publishers, New York, 2010, Ch.6. The apparatus, design considerations and unique aspects of the equilibrium cell were discussed in a journal publication. This expands the body of knowledge related to high-pressure research. Chemical Engineering & Technology 38 (2015) 1165-1172. A second methodology paper covered GC design criteria and quantitative calibration methods, and presented novel visuals illustrating the challenges associated with high-pressure sampling. Chemical Engineering & Technology 39 (2016) 1475-1482. New phase composition data and modelling results for CO2 + nC12 + 37DM1O were presented in a journal publication. Given the scarcity of and challenges associated with producing high-pressure ternary VLE data, this represents a valuable contribution. The Journal of Supercritical Fluids 130 (2017) 105-117. New phase composition data and modelling results for CO2 + 37DM1O + C10OH, and visuals illustrating density inversion behaviour, were in September 2017 submitted for publication. New phase composition data were measured for CO2 + nC12 + C10OH and the data were used to evaluate four thermodynamic models. A manuscript presenting these results and a comparative discussion on phase behaviour of the three ternaries is currently in preparation.