Phase Equilibria & thermodynamic modelling of the ternary system CO2 + 1-decanol + N-Tetradecane

Ferreira, Machelle (2018-12)

Thesis (PhD)--Stellenbosch University, 2018.

Thesis

ENGLISH ABSTRACT: Experimental data and predictive process models, tested at various operating conditions, have shown that supercritical fluid fractionation is a feasible process when aimed at the separation of detergent range 1-alcohols and n-alkanes with similar boiling points. Although this process shows good separation performance, it was previously found that distinct solute + solute interactions occur that influence the predictive capability of thermodynamic models. The aim of this study was to obtain a fundamental understanding of the solute + solute interactions in the CO2 + 1-decanol + n-tetradecane ternary system; firstly, through the generation of phase equilibria data and secondly, through the evaluation of thermodynamic models, with solute + solute binary interaction parameters (BIPs) incorporated into their algorithm, to correlate the new VLE data. The aim was achieved through the following objectives: (1) Studying the high-pressure phase equilibria of the CO2 + 1-decanol + n-tetradecane ternary system; (2) Studying the low-pressure phase equilibria of the 1-decanol + n-tetradecane binary system; (3) Selecting 4 suitable thermodynamic models available within a commercial process simulator and studying the modelling of the ternary and binary phase equilibria data with new solute + solute BIPs obtained from the experimental data. The first objective was met in two parts namely, the measurement of new high-pressure bubble- and dew-point data (HPBDP) and the measurement of new high-pressure vapour-liquid equilibria data (HPVLE). The HPBDP experiments were conducted between T = 308 K and T = 358 K using a visual static synthetic method. CO2 free n-tetradecane mass fractions (wcred) of 0.2405, 0.5000, 0.6399, 0.7698, 0.8162 and 0.9200 g/g were investigated, and the total solute mass fractions were varied between 0.015 g/g and 0.65 g/g. An increase in the solute + solute interactions were observed when increasing the n-tetradecane composition and decreasing the temperature. The distinct solute + solute interactions lead to the formation of a liquid-gas hole in the three-phase surface, cosolvency effects and miscibility windows. For the HPVLE data, a state of the art high-pressure analytical view cell was used to study four ternary mixtures at T = 308 K, 328 K and 348 K and pressures between P = 8.0 and 16.4 MPa. The equipment allowed for equilibrium to be achieved after which samples of the co-existing phases were taken simultaneously. Phase composition data for four tie lines were obtained and ternary phase diagrams constructed. A similar outcome to the HPBDP experimental results were observed. In general, for wcred ≥ 0.9004 g/g, 1-decanol will be the more soluble compound and for wcred ≤ 0.2403 g/g, n-tetradecane will be the more soluble compound. Furthermore, within the complex phase behaviour region (wcred = ± 0.6245 g/g), separation of residual n-tetradecane from 1-decanol in the mixtures are postulated to be impossible. However, separation experiments are required on a pilot plant setup to verify this assumption. To achieve the second objective, the phase behaviour complexities brought on by the 1-decanol + n-tetradecane interactions were further evaluated through the measurement of new low-pressure vapour-liquid equilibria data (LPVLE). The experiments were conducted at sub-atmospheric pressure (P = 40 kPa) using an all glass dynamic recirculating still. The binary system displayed positive azeotropy, inferring Type I-A fluid phase behaviour. The presence of the azeotrope and the non-unity activity coefficients confirmed that the binary system exhibits non-ideal phase behaviour. Four thermodynamic models, available within Aspen Plus®, were evaluated for their ability to correlate (RK-Aspen, SR-Polar and PC-SAFT) and to predict (PSRK) all three sets of experimental data. PSRK made use of previously determined low-pressure activity coefficient group-group parameters and thus served as a purely predictive model. For the remaining three models, the HPVLE and LPVLE data were used with the built-in data regression function in Aspen Plus® to regress solute + solute BIPs. For the HPBDP BIPs a plug-and-play method was applied to manually regress representative values instead of exact values. Objective 3 was achieved by evaluating the performance of the models with varying solute + solute BIPs in their specific model algorithm, i.e. BIPs regressed using low-pressure data were used to represent high-pressure data and vice versa as summarised in Table i. RK-Aspen was the only model to produce an accurate representation of each set of experimental data, which included the complex phase behaviour regions. SR-Polar was a close second, lagging in the representation of the LPVLE data. On a purely predictive front, PSRK can be used to represent accurate HPBDP and LPVLE data but should not be used to predict HPVLE data. Lastly, PC-SAFT can produce reasonable LPVLE and HPVLE data but failed to correlate the HPBDP data accurately. The models, presented in order of decreasing performance, were RK-Aspen > SR-Polar > PSRK > PC-SAFT. Lastly, the RK-Aspen model with HPBDP solute + solute BIPs provided the most accurate model fit (quantitatively and qualitatively) within the complex phase behaviour region of the HPBDP, HPVLE and LPVLE data. Overall, this thesis provided valuable insight into the role that solute + solute interactions play in generating complex phase behaviour and fractionation difficulties. In future studies, solute + solute interactions should not be ignored as they will help improve not only the design of pilot plant experiments, but also process models.

AFRIKAANSE OPSOMMING: Eksperimentele data en voorspellende prosesmodelle, wat by verskeie bedryfstoestande getoets is, het getoon dat superkritiese vloeier fraksionering ’n lewensvatbare proses is wanneer dit gemik is op die skeiding van 1-alkohole en n-alkane met soortgelyke kookpunte. Alhoewel hierdie proses goeie skeiding toon, is daar voorheen bevind dat opgeloste stof interaksies die voorspellingsvermoë van termodinamiese modelle beïnvloed. Die doel van hierdie ondersoek was om ’n fundamentele begrip van die opgeloste stof interaksies in die CO2 + 1-dekanol + n-tetradekaan ternêre sisteem te verkry: eerstens, deur fase-ewewigsdata te genereer, en tweedens, deur die evaluering van termodinamiese modelle, wat van binêre interaksie parameters (BIPs) gebruik maak, om die nuwe fase-ewewigsdata te korreleer. Hierdie doel is deur die volgende doelwitte bereik: (1) Bestudeer die hoë druk fase ewewig van die CO2 + 1-dekanol + n-tetradekaan ternêre sisteem; (2) Bestudeer die lae druk fase ewewig van die 1-dekanol + n-tetradekaan binêre sisteem; (3) Kies 4 gepaste termodinamiese modelle wat in ’n kommersiële proses simulator beskikbaar is en bestudeer die termodinamiese modellering van die ternêre en binêre fase-ewewigsdata met nuwe opgeloste stof-BIPs bepaal deur die eksperimentele data. Die eerste doelwit is in twee dele bereik: nuwe hoë druk borrel- en doupunt data (HPBDP) en nuwe hoë druk damp-vloeistof ewewigsdata (HPVLE) is gemeet. Die HPBDP eksperimente is tussen T = 308 K en T = 358 K met ’n visuele staties-sintetiese metode uitgevoer. CO2-vry n-tetradekaan massafraksies (wcred) van 0.2405, 0.5000, 0.6399, 0.7698, 0.8162 en 0.9200 g/g is bestudeer en die totale opgeloste stof massafraksies is tussen 0.015 g/g en 0.65 g/g varieer. ’n Toename in opgeloste stof interaksies is met ’n toename in n-tetradekaan samestelling en ’n afname in temperatuur waargeneem. Die opgeloste stof interaksies het gelei tot die vorming van ’n vloeistof-gas gaping in die drie-fase gebied, saam-oplosbaarheid, en mengbaarheidsgebiede. Om die HPVLE data te meet is ’n moderne hoë druk visuele analitiese sel gebruik. Vier mengsels is by T = 308 K, 328 K en 348 K, en by drukke tussen P = 8.0 en 16.4 MPa bestudeer. In hierdie toerusting kan monsters van die ekwilibrium fases gelyktydig geneem word. Die fase-samestellingsdata van vier bindlyne is verkry en ternêre fasediagramme is gekonstrueer. Die HPVLE data het soortgelyke gevolgtrekkings as die HPBDP data na vore gebring. Oor die algemeen was 1-dekanol meer oplosbaar vir die wcred ≥ 0.9004 g/g mengsel en n-tetradekaan meer oplosbaar vir die wcred ≤ 0.2403 g/g mengsel. In die komplekse fasegedragsgebied (wcred = ± 0.6245 g/g) word daar postuleer dat skeiding van residuele n-tetradekaan van 1-dekanol onmoontlik is. Skeidingseksperimente op ’n loodsaanleg word egter benodig om hierdie aanname te bevestig. Om die tweede doelwit te bereik, is die komplekse fasegedrag wat deur 1-dekanol + n-tetradekaan interaksies veroorsaak word, verder evalueer. Nuwe lae druk damp-vloeistof ewewigsdata (LPVLE) is vir hierdie binêre stelsel by sub-atmosferiese druk (P = 40 kPa) in ’n dinamiese hersirkulerende distilleerder gemeet. Hierdie binêre sisteem toon ’n positiewe aseotroop, wat moontlik Tipe I-A vloeier fasegedrag aandui. Die teenwoordigheid van ’n aseotroop en die aktiwiteitskoëffisiënt waardes het bevestig dat die sisteem nie-ideale fasegedrag uitoefen. Vier termodinamiese modelle, almal beskikbaar in Aspen Plus®, is evalueer op grond van hul vermoë om die drie stelle eksperimentele data te korreleer (RK-Aspen, SR-Polar, en PC-SAFT) en te voorspel (PSRK). PSRK gebruik voorheen bepaalde lae druk aktiwiteitskoëffisiënt groep-groep parameters en dien dus as ’n suiwer voorspellende model. Vir die ander drie modelle is die HPVLE en LPVLE data saam met die ingeboude regressiefunksie in Aspen Plus® gebruik om opgeloste stof BIPs te bepaal. Verteenwoordigende waardes van die HPBDP BIPs is met die hand bepaal. Die derde doelwit is bereik deur die werksverrigting van die modelle met wisselende opgeloste stof BIPs te evalueer, met ander woorde om die LPVLE BIPs te gebruik om hoë druk data te voorspel, en anders om, soos opgesom in Tabel i. RK-Aspen was die enigste model wat ’n akkurate voorstelling van elke datastel kon lewer, ook in die komplekse fasegedragsgebiede. SR-Polar se werksverrigting was amper so goed soos dié van RK-Aspen, maar kon nie die LPVLE data so goed voorstel nie. PSRK kan gebruik word om HPBDP en LPVLE data akkuraat voor te stel, maar word nie vir HPVLE data aangeraai nie. PC-SAFT kan LPVLE en HPVLE data redelik voorstel, maar kan nie HPBDP data akkuraat korreleer nie. Die modelle word in volgorde van afname in werksverrigting voorgestel as: RK-Aspen > SR-Polar > PSRK > PC-SAFT. RK-Aspen, in kombinasie met die HPBDP opgeloste stof BIPs, bied die mees akkurate model passing (kwantitatief en kwalitatief) in die komplekse fasegedragsgebied van die HPBDP, HPVLE, en LPVLE data. Oor die algemeen het hierdie proefskrif waardevolle insig gegee tot die rol wat opgeloste stof BIPs speel. Die opgeloste stof interaksies veroorsaak komplekse fase-gedrag en fraksionele probleme. In toekomstige studies moet opgeloste stof interaksies nie geïgnoreer word nie, aangesien dit nie net die ontwerp van proefnemingsaanlegte sal verbeter nie, maar ook proses modelle sal verbeter.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/105046
This item appears in the following collections: