The hydrodynamics of supercritical packed countercurrent columns

Thesis (PhD)--Stellenbosch University, 2020.ENGLISH ABSTRACT: Supercritical fluids are enjoying ever-increasing popularity as a solvent medium for extraction, stripping, absorption and fractionation processes. Although the potential of supercritical fluids as solvents have been known for more than a century, there are still several areas of uncertainty, one being the hydrodynamics of fractionation columns operating under supercritical conditions. Supercritical fractionation columns are readily tuneable and able to achieve sharp, highly efficient separations, presenting an attractive alternative to traditional solvents in specific niche applications. Robust hydrodynamic models are key to the design of supercritical fractionation processes, but no such models are available in the open literature. To create models, investigations into fundamental hydrodynamics are required. Two aspects are of particular concern when investigating hydrodynamics. Firstly, the fluid properties of the respective phases involved must be known. Secondly, mass transfer should be minimised or quantified. No study found in the literature presents hydrodynamics under supercritical conditions with measured, not estimated, saturated fluid properties taken into consideration. This study had the overarching aim to investigate hydrodynamics in countercurrent columns operating under supercritical conditions while laying groundwork for the eventual development of accurate predictive models and design methods. This aim was broken down into five objectives to remove obstacles and generate the needed data to achieve the aim. To address the lack of fluid property data, the first objective was to develop new equipment capable of concurrently measuring the required fluid properties of density and dynamic viscosity. The equipment setup (Pmax 35 MPa; Tmax 393 K) included a variable volume view cell to determine bubble/dew points and density, with a quartz-crystal resonator to measure dynamic viscosity. The equipment was validated, firstly using pure component fluid property measurements with n-dodecane and benzene at 0.1 - 30 MPa and 313 - 353 K and, secondly, using CO2 + ethyl tetradecanoate (ET) for binary phase equilibrium measurement. New data were presented on the dynamic viscosity and density of both saturated CO2 and ET phases. Using the developed equipment, it was possible to evaluate binary systems implicitly for use in supercritical hydrodynamics, addressing the second objective of the study. Two Poly [dimethylsiloxane] (PDMS) fluids graded at 100 cSt and 200 cSt were selected. Saturation pressure was determined for 1 – 70 wt% PDMS in CO2 at 313 – 353 K. The density (~900 – 800 kg.m-3) and dynamic viscosity (~0.7 – 7 mPa.s) were measured at saturation. The selected systems exhibited low mutual solubility and fluid properties in the ranges desired for further work. The third objective, to conduct hydrodynamic pilot plant studies, could be planned and executed using the measured fluid property and phase data. The equipment was operated at 14 MPa and 333 - 323 K to investigate the hydrodynamic operability for a 38 mm diameter column packed with ¼” Dixon rings. The temperature/pressure combination was selected to investigate a wide array of fluid properties while presenting the opportunity to differentiate between the influence of viscosity and density. The pressure drop, liquid hold-up, mass flow and massfractions were measured. Importantly, it was found that liquid hold-up and pressure drop are not reliable indicators of operability in supercritical systems. Three distinct types of inoperability were identified, namely liquid layer flooding, bubble column flooding and entrainment. The influence of the density and dynamic viscosity on hydrodynamics was found to be complex, yet significant. Further, no observable loading operating regime was observed, with the column only operating in the pre-loading or inoperable (flooded/entrained) regimes. The fourth objective was to evaluate three hydrodynamic models for their ability to predict experimental and literature data. The Stichlmair et al. model was able to predict the liquid hold-up after a modification was made to take supercritical fluid properties into account, and the empirical constants were recalculated. The model could, however, not predict the experimental pressure drop or operability limits regardless of any modification. Maćkowiak’s SBD model failed to predict any hydrodynamic properties. The model by Woerlee provided an order of magnitude estimation of liquid hold-up and pressure drop for specific conditions using very little empiricism. The model could not predict flooding at all. None of the models could present an accurate view of the hydrodynamics of the system regardless of attempted adjustments and modifications, with the models displaying different qualitative trends than the gathered experimental data. The final objective tested the literature hypothesis that supercritical hydrodynamics are fundamentally similar to ‘classical’ hydrodynamics. Three conclusions made during this study cast doubt on the fundamental similarity. Firstly, the lack of a detectable loading zone was in contradiction with classical hydrodynamics where the loading zone plays a significant role. Secondly, pressure drop and liquid hold-up were found to be unreliable predictors of operability in the supercritical systems investigated, in contrast to classical systems. Thirdly, the investigated hydrodynamic models cannot predict pressure drop, flooding, or hydrodynamic capacity for the supercritical systems investigated. The thesis presents the following novel contributions: a)Equipment, presenting a novel combination of measurement techniques for concurrent determination of phase equilibria, density and viscosity at fluid saturation,as published in J. Supercrit. Fluids, 133 (2018) 444-454. The publication also contains novel data on benzene viscosity, phase equilibria, and saturated phase density and dynamic viscosity for CO2 + ethyl tetradecanoate. It further presents the first measurement of saturated fluid properties of the supercritical phase found by the author in the literature. b) New data on two PDMS + CO2 systems, 100 cSt and 200 cSt, including phase equilibria,and saturated phase density and viscosity for both phases, as published in J. Supercritical Fluids, 139 (2018) 1-7. Given the lack of saturated fluid properties under supercritical conditions and the complexity of measurement, this represents a valuable contribution, especially being the only measurement of saturated supercritical fluidphase properties found in the literature. c) New data on supercritical hydrodynamics allows various observations and conclusions. Observations include the identification of three different inoperability modes, the lack of an observable loading zone, as well as describing the influence of fluid properties on the hydrodynamic behaviour. Published in part in Chem. Eng. Trans., 69 (2018), with a second publication planned. d) The full evaluation of the Stichlmair et al. model, Maćkowiak’s SBD model and Woerlee’s model against the gathered data showcases the inability of the investigated classical hydrodynamic models to predict supercritical hydrodynamics. This finding, along with other findings mentioned above, highlight the possibility of a fundamental difference between classical and supercritical systems. However, there is insufficient information to make a definite statement.AFRIKAANSE OPSOMMING: Superkritiese vloeistowwe geniet toenemende populariteit as ’n oplosmiddel in ekstraksie-, stroping-, absorpsie- en fraksioneringsprosesse. Alhoewel die potensiaal van superkritiese vloeistowwe as oplosmiddels al vir meer as ’n eeu bekend is, is daar steeds verskeie areas van onsekerheid in die literatuur. Een so ‘n area is die hidrodinamika van fraksioneringskolomme wat onder superkritiese toestande werk. Superkritiese fraksioneringskolomme is maklik verstelbaar en in staat daartoe om skerp, doeltreffende skeidings te bewerkstellig wat ’n aantreklike alternatief tot tradisionele oplosmiddels in spesifieke nistoepassings bied. Robuuste hidrodinamiese modelle is belangrik vir die ontwerp van superkritiese fraksioneringsprosesse, maar geen sulke modelle is in die oop literatuur beskikbaar nie. Om modelle te skep, is ondersoek tot die fundamentele hidrodinamika benodig. Twee aspekte is van besondere belang wanneer hidrodinamika ondersoek word. Eerstens, moet die vloeistofeienskappe van die betrokke fases bekend wees. Tweedens, moet massa-oordrag tot ‘n minimum beperk word, of gekwantifiseer word. Geen studie in die literatuur bevat hidrodinamika onder superkritiese kondisies met die gemete, nie geskatte, versadigde vloeistofeienskappe wat in ag geneem is nie. Hierdie studie het die oorkoepelende doelwit gehad om hidrodinamika in superkritiese teenstroomkolomme te ondersoek en ‘n grondslag lê vir die uiteindelike ontwikkeling van akkurate modelle en ontwerpsmetodes. Hierdie doelwit is opgedeel in vyf doelstellings om struikelblokke te verwyder en die nodige data te genereer. Om die gebrek aan vloeistofeienskapdata aan te spreek, was die eerste doelstelling om nuwe toerusting te ontwikkel wat die vermoë het om die nodige vloeistofeienskappe, digtheid en dinamiese viskositeit, gelyktydig te meet. Die toerusting (Pmax 35 MPa; Tmax 393 K) bestaan uit ’n veranderlikevolume-sel met ‘n sigglas om borrel-/doupunte en digtheid te bepaal, asook ’n kwarts-kristalresonator om dinamiese viskositeit te meet. Die toerusting is gevalideer, eerstens, met behulp van suiwer-komponent vloeistofeienskapsmetings met n-dodekaan en benseen by 0.1 – 30 MPa en 313 – 353 K en, tweedens, met behulp van CO2 + etieltetradekanoaat (ET) vir die meting van binêre fase-ewewig. Nuwe data is voorgelê vir die dinamiese viskositeit en digtheid van albei die versadigde CO2- en ET-fases. Met behulp van die ontwikkelde toerusting was dit moontlik om binêre sisteme te evalueer vir die implisiete doel om superkritiese hidrodinamika te meet, wat die tweede doelstelling van die studie aanspreek. Twee poli[dimetielsiloksaan] (PDMS) vloeistowwe, gegradeer as 100 cSt en 200 cSt, is gekies. Versadigingsdruk is bepaal vir 1-70 wt% PDMS in CO2 by 313–353 K. Die digtheid (~900 - 800 kg.m-3) en dinamiese viskositeit (~0.7 - 7 mPa.s) is by versadiging gemeet. Die gekose stelsels het lae wedersydse oplosbaarheid, asook vloeistofeienskappe in die gewenste bereik, getoon. Die derde doelstelling, om hidrodinamiese loodsaanlegstudies uit te voer, kon beplan en uitgevoer word met behulp van die gemete vloeistofeienskap- en fase-data. Die toerusting is bedryf by 14 MPa en 333 - 323 K om die hidrodinamiese bedryfsgrense van ’n 38 mm-deursnee kolom gepak met ¼’’-Dixon ringe, te ondersoek. Die temperatuur/druk kombinasie is gekies om ’n wye verskeidenheid vloeistofeienskappe te ondersoek, asook om ’n geleentheid skep om die tussen invloed van viskositeit en digtheid te kan onderskei. Die drukval, vloeistof-oponthoud, massavloei en massafraksie is gemeet. Van belang is die bevinding dat vloeistof-oponthoud en drukval nie betroubare aanduiders van beryfbaarheid in superkritiese sisteme is nie. Drie afsonderlike bedryfsgrens-oorgange is geïdentifiseer, naamlik vloeistoflaagvloeding, borrelkolomvloeding en meesleuring. Die invloed van die digtheid en dinamiese viskositeit op hidrodinamika is gevind om kompleks te wees, dog beduidend. Verder is daar geen waarneembare ladingsbedryfregime gevind nie, met die kolom wat slegs in die voorladings- of onbedryfbare-(gevloede/meegesleurde) regimes werk. Die vierde doelstelling was om drie hidrodinamiese modelle te evalueer vir hul vermoë om eksperimentele- en literatuurdata te voorspel. Die Stichlmair-et-al.-model kon die vloeistof-oponthoud voorspel nadat ’n modifikasie aangebring is om superkritiese vloeistofeienskappe in ag te neem, en die empiriese konstantes herbereken is. Die model kon egter nie die eksperimentele drukval of bedryfsgrense voorspel nie, ongeag enige modifikasies. Maćkowiak se SBD model kon geen hidrodinamiese eienskappe voorspel nie. Die model deur Woerlee was in staat tot ‘n ordegrootte skatting van die gemete vloeistof-oponthoud en drukval vir van die stelsels met baie min empiriese invloed. Hierdie model kon glad nie die bedryfsgrense voorspel nie. Nie een van die modelle kon ’n akkurate oorsig van die hidrodinamika van die stelsel bied nie, ongeag pogings tot aanpassings en modifikasies, met die modelle wat anderse kwalitatiewe tendense as die eksperimentele data toon. Die finale doelstelling het die literatuurhipotese getoets dat superkritiese hidrodinamika, in wese, dieselfde is as ‘klassieke’ hidrodinamika. Drie gevolgtrekkings is tydens die studie gemaak wat die fundamentele ooreenkoms in twyfel trek. Eerstens, is die gebrek aan ’n waarneembare ladingsbedryfregime in teenstelling met klassieke hidrodinamika waar die ladingsbedryfregime ’n belangrike rol speel. Tweedens, is daar gevind dat drukval en vloeistof-oponthoud onbetroubare voorspellers van die bedryfsgrense in die superkritiese sisteme is, in teenstelling met klassieke hidrodinamiese stelsels. Derdens, kan die ondersoekte hidrodinamiese modelle nie die drukval, vloeding, of hidrodinamiese kapasiteit voorspel vir die superkritiese sisteme wat ondersoek was nie. Die proefskrif bied die volgende nuwe bydraes aan: a) Toerusting wat ’n nuwe kombinasie van metingstegnieke vir die gelyktydige bepaling van fase-ewewig, digtheid en viskositeit by vloeistof versadiging, voorstel, soos gepubliseer in J. Supercrit. Fluids, 133 (2018) 444–454. Die publikasie bevat ook nuwe benseen-viskositeitsdata, asook fase-ewewig, en versadigde fase digtheids- en dinamiese viskositeitsdata vir CO2 + etieltetradekanoaat. Dit stel verder ten toon die eerste meting van versadigde vloeistofeienskappe van die superkritiese fase wat deur die outeur in die literatuur gevind is. b) Nuwe data op twee PDMS + CO2-stelsels, 100 cSt en 200 cSt, wat insluit fase-ewewig, en versadigde fase digtheids- en viskositeitsdata vir beide fases, soos gepubliseer in J. Supercritical Fluids, 139 (2018) 1–7. Gegewe die gebrek aan versadigde vloeistofeienskappe onder superkritiese kondisies en die kompleksiteit van meting, verteenwoordig hierdie ’n waardevolle bydrae, veral omdat dit die enigste mate van versadigde superkritiese vloeistof fase-eienskappe is wat in die literatuur voorkom. c) Nuwe data op superkritiese hidrodinamika laat verskeie waarnemings en gevolgtrekkings toe. Waarnemings sluit in die identifisering van drie verskillende bedryfsgrens-oorgange, die gebrek aan ’n waarneembare ladingsbedryfregime, asook die beskrywing van die invloed van vloeistofeienskappe op die hidrodinamiese gedrag. Die bevindings is gedeeltelik gepubliseer in Chem. Eng. Trans. 69 (2018), met ’n tweede publikasie wat beplan word. d) Die volle evaluering van die Stichlmair-et-al.-model, Maćkowiak se SBD-model, en Woerlee se model teen die versamelde data wys die onvermoë van hierdie klassieke hidrodinamiese modelle om superkritiese hidrodinamika te voorspel. Hierdie bevinding, tesame met ander bevindings hierbo genoem, beklemtoon die moontlikheid van ’n fundamentele verskil tussen klassieke en superkritiese stelsels. Daar is egter onvoldoende inligting om ’n definitiewe standpunt te maak.The financial assistance of the National Research Foundation (NRF) towards this research is hereby acknowledged. Opinions expressed and conclusions arrived at, are those of the author and are not necessarily to be attributed to the NRF.Doctora