Vapour-liquid-liquid equilibria measurements for the dehydration of low molecular weight alcohols via heterogeneous azeotropic distillation

Brits, Leanne (2015-03)

Thesis (MEng)--Stellenbosch University, 2015.

Thesis

ENGLISH ABSTRACT: The operation and optimisation of a distillation train directly effects the total energy consumption of a typical processing plant. With this in mind, the efficient separation of low molecular weight alcohol azeotropes, using heterogeneous azeotropic distillation, is of great economic and environmental importance. Heterogeneous azeotropic distillation involves the addition of an extraneous component, known as an entrainer, to the mixture to facilitate separation. Benzene has long been replaced as the entrainer of choice, due to its carcinogenic nature, and research into finding a more suitable entrainer has commenced. To determine if an entrainer is suitable for a particular separation, detailed phase behaviour information of the ternary alcohol/entrainer/water system is required; vapour-liquid (VLE), vapour-liquid-liquid (VLLE) equilibria data and the composition of all azeotropes present. This is complicated by the fact that thermodynamic models (like the nonrandom two-liquid (NRTL), universal functional (UNIFAC) and universal quasichemical (UNIQUAC) activity coefficient models) often fail to predict the phase equilibria of ternary systems. The lack of available experimental phase equilibria data, and the inability of thermodynamic models to predict phase equilibria data, has fueled the need for the experimental determination of accurate, repeatable isobaric VLE, VLLE and azeotropic data. With this in mind, this research is focused on the experimental determination of VLE, VLLE and azeotropic data for three low molecular weight alcohol/entrainer/water systems at 101.3 kPa. Following an extensive literature study on azeotropes, applicable separation techniques and available VLE and VLLE data in literature, the ethanol/2-butanone/water, n-propanol/2-butanone/water and iso-propanol/2-butanone/water systems were chosen for experimental investigation. The experimental determination was carried out in a Gillespie type still, equipped with an ultrasonic homogenizer. The temperature and pressure accuracies of the equipment were found to be 0.03°C and 2mbar respectively. The chosen experimental methodology was verified, and its repeatability tested, through the measurement of isobaric VLE and VLLE data of ethanol/isooctane, ethanol/n-butanol/water and n-propanol/isooctane/water systems at 101.3 kPa and subsequent comparison of the measured data with literature data. The compositional error reported, taking into account experimental and analysis effects, is ±0.014 mole fraction. All experimentally determined data sets, verification and new data, were tested for thermodynamic consistency by using the Wisniak modification of the Herrington test, the L/W consistency test, as well as the McDermott-Ellis consistency test, and found to be consistent. The Othmer-Tobias correlation was used to ensure the measured LLE data followed a steady trend, with all R-values larger than 0.910. For all three of the new systems chosen, the absence of ternary heterogeneous azeotropes was noted. The presence of a ternary homogeneous azeotrope was found for both the ethanol/2-butanone/water and iso-propanol/2-butanone/water systems. No ternary azeotropes are present for the n-propanol/2-butanone/water system. Suitable entrainers were compared to 2-butanone (MEK) by plotting measured data and literature information of five similar alcohol/entrainer/water systems on a ternary phase diagram. It was found that MEK could not be considered as a suitable entrainer for heterogeneous azeotropic distillation of ethanol, n-propanol and IPA. This is due to the absence of a ternary heterogeneous azeotrope for the aforementioned alcohol/MEK/water systems. Finally, the ability of thermodynamic models (NRTL, UNIFAC and UNIQUAC) to predict experimental data was determined both visually and through descriptive statistics. This entailed the inspection of ternary phase diagrams and the calculation and evaluation of average absolute deviation (AAD) and and average absolute relative deviation (AARD%) values. The measured data were modelled in Aspen Plus®. It was found that none of the models could predict the ternary systems with acceptable accuracy and the data were regressed. In general, the regressed parameters for the NRTL, UNIFAC and UNIQAC models improved the model predictions when compared to the built-in Aspen parameters. The UNIFAC model predicted the ethanol/MEK/water and n-propanol/MEK/water systems most accurately while none of the models could predict the IPA/MEK/water systems with acceptable accuracy.

AFRIKAANSE OPSOMMING: Die ontwerp en optimering van 'n distillasietrein het ‘n duidelike effek op die totale energieverbruik van ‘n tipiese prosesaanleg. Met dit in gedagte, is ‘n meer doeltreffende skeiding van lae molekulêre massa alkohol aseotrope, met behulp van heterogene aseotropiese distillasie, voordelig vir die ekonomie en die omgewing. Heterogene aseotropiese distillasie behels die toevoeging van 'n eksterne komponent, wat bekend staan as 'n skeidingsagent, om uiteindelik die skeiding te fasiliteer deur die komponente se dampdrukke te verander. Benseen was in die verlede ‘n gewilde skeidingsagent, maar dit is a.g.v. sy karsenogeniese eienskappe nie meer aanvaarbaar om te gebruik nie. Nuwe navorsing in hierdie veld fokus dus onder andere op die identifisering van meer geskikte skeidingsagente. Om te bepaal of 'n skeidingsagent geskik is, word indiepte fasegedrag inligting benodig, i.e. damp-vloeistof en damp-vloeistof-vloeistof ewewigsdata en die samestelling van alle aseotrope teenwoordig. Ongelukkig kan termodinamiese modelle dikwels nie die fasegedrag van ternêre stelsels voorspel nie. Dit, sowel as die beperkte beskikbaarheid van eksperimentele ewewigsdata in die literatuur, het dus hierdie navorsing aangevuur. Die projek het gefokus op die experimentele bepaling van damp-vloeistof en damp-vloeistof-vloeistof ewewigsdata en aseotropiese data vir drie alkohol/skeidingsagent/water-stelsels by 101.3 kPa. Na ‘n indiepte literatuurstudie van aseotrope, gepaste skeidingstegnieke en beskikbare damp-vloeistof en damp-vloeistof-vloeistof ewewigsdata, is 2-butanone (MEK) gekies as ‘n moontlike skeidingsagent en die etanol/MEK/water-, n-propanol/MEK/water- en iso-propanol/MEK/water-stelsels gekies vir eksperimentele ondersoek. Die data is met ‘n dinamiese Gillespie eenheid gemeet, toegerus met ‘n ultrasoniese homogeniseerder om vloeistof-vloeistof skeiding te voorkom. Die akkuraatheidsbande van temperatuur- en druk meetinstrumente was 0,03°C en 2 mbar, onderskeidelik. Die eksperimentele metode en die herhaalbaarheid van metings is bevesting, deur die isobariese damp-vloeistof en damp-vloeistof-vloeistof ewewigsdata van etanol/iso-oktaan, etanol/n-butanol/water en n-propanol/iso-oktaan/water te vergelyk met onafhanklike stelle ooreenstemmende data uit die literatuur. Die gesamentlike eksperimentele en analitiese fout wat gemaak kon word tydens bepaling van molfraksie samestellings was ±0.014 molfraksie. Alle gemete eksperimentele data is getoets vir termodinamiese samehang deur middel van beide die L/W en McDermott-Ellis konsekwentheidstoetse. Die Othmer-Tobias korrelasie is gebruik om seker te maak dat die gemete LLE data ‘n konstante tendens volg, met alle R-waardes groter as 0.910. Vir al drie van die nuwe stelsels wat gekies is, was ‘n drieledige heterogene aseotroop afwesig. Die teenwoordigheid van drieledige homogene aseotrope is egter waargeneem vir die etanol/MEK/water- en IPA/MEK/water-stelsels. Geen drieledige aseotrope is vir die n-propanol/MEK/water-sisteem gevind nie. Alle gemete data, asook literatuur inligting van vyf soortgelyke alkohol/skeidingsagent/water sisteme, is op ‘n drieledige fase diagram voorgestel om die skeidingsagente met mekaar te vergelyk. Hiervolgens word dit getoon dat MEK nie as ‘n gepaste skeidingsagent vir heterogene aseotropiese distillase beskou kan word nie a.g.v. die afwesigheid van ‘n drieledige heterogene aseotroop in die voorgenoemde alkohol/MEK/waterstelsels. Die vermoë van die termodinamiese modelle (NRTL, UNIFAC en UNIQUAC) om die eksperimentele data te voorspel is visueel (per grafiek) sowel as deur beskrywende statistiek bepaal. Dit behels die inspeksie van drieledige fasediagrame en die berekening en evaluasie van die gemiddelde absolute afwyking en gemiddelde absolute relatiewe afwykingswaardes. Hierdie teoretiese data is met Aspen Plus® bepaal. Nie een van die modelle kon die drieledige stelsels se fasegedrag met aanvaarbare akkuraatheid voorspel nie. Die parameters vir die NRTL-,UNIFAC- en UNIQUAC-modelle kan verbeter word deur middel van regressie, in vergelyking met die ingeboude Aspen parameters. Dit is bevind dat die UNIFAC model die etanol/MEK/water- en n-propanol/MEK/water-stelsel die beste kan voorspel. Nie een van die bogenoemde modelle kon egter die fasegedrag van die IPA/MEK/water-stelsel voorspel nie.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/96850
This item appears in the following collections: