Behaviour of oxygen transfer in a simulated multiphase hydrocarbon-based bioprocess in a bubble column reactor

Abufalgha, Ayman A. (2018-03)

Thesis (MEng)--Stellenbosch University, 2018.

Thesis

ENGLISH SUMMARY: Linear alkane hydrocarbons can be converted into higher value products and product-precursors through the addition of an oxygen moiety. One method of upgrading these hydrocarbons is through biologically-mediated oxidation, whereby an oxygen group is added to the alkane backbone by a microorganism to produce a more reactive, and more valuable molecule. To achieve this, operating conditions conducive to this biological process need to be reached; one key is liquid-phase oxygen concentration, which is governed by the oxygen transfer rate into the reaction fluid. This study focuses on the behaviour of the overall volumetric oxygen transfer coefficient (KLa) and, the interfacial area (a), in a model four-phase (air-water-hydrocarbon-deactivated yeast) hydrocarbon-based bioprocess in a bubble column reactor (BCR). It sought to give an understanding on the impact of the alkane concentration, superficial gas velocity and yeast loading on the KLa and the interfacial area of the system. The experiments were conducted at different alkane concentrations of (2.5 to 20% v/v n-C14-20), superficial gas velocities of (1 to 3 cm/sec), and yeast loading of (0.5 to 6 g/l Saccharomyces cerevisiae). The yeast cells were deactivated by heating, to suppress oxygen utilisation, while maintaining cell integrity. The KLa was measured using the gassing-out method (GOP) using a dissolved oxygen (DO) probe which recorded the rate of DO in the system every 5 seconds until oxygen saturation was achieved. The probe constant (Kp) was evaluated at all experimental conditions to ensure the accuracy of the KLa determination. The interfacial area was calculated from the gas hold-up (Ɛ𝐺), and Sauter mean diameter (𝐷32) data. 𝐷32 was evaluated by high-speed photography and image analysis on the acquired images, performed using MATLAB software, while the Ɛ𝐺 was calculated as the difference in the liquid level before and after sparging the reactor. For three phase systems (air-water-hydrocarbons), it was found that at low superficial gas velocity the system was non-homogeneous, with an increased hydrocarbon concentration at the top of the reactor. The spatial homogeneity of the system has not been previously investigated in the literature, particularly in the the bubble column reactor when the process contains hydrocarbons. The KLa measurements were thus affected by the spatial variations in concentrations and therefore could not be reliable. To overcome this, the system was analysed for spatial liquid-phase homogeneity using a physical sampling methodology, which established the optimal superficial gas velocity (2 cm/sec) to achieve homogenous flow in the three phase system. For the four-phase system, the addition of yeast to the BCR resulted in a change in the flow regime and therefore increased the homogeneity in the system at low superficial gas velocity (1 cm/sec). This result suggets that the addition of yeast cells to the reactor likely changed the fluid properties such as viscosity and the surface tension. Investigations on the impact of yeast loading, alkane concentration and superfical gas velocity on KLa and interfacial area in a multiphase hydrocarbon-based bioprocess in a BCR were undertaken for this project. It was found that an increase in the yeast loading resulted in a decrease in KLa at constant superficial gas velocity in the three phase (air-water-deactivated yeast) system. This decrease might be attributed to the existence of diffusion blocking effects as well as the increasing fluid viscosity in the system. A similar effect was found in a four phase (air-water-hydrocarbon-deactivated yeast) system, when the values of KLa were depressed with increasing yeast concentration at high superficial gas velocities and constant alkane concentration (2.5% v/v). However, increasing the constant alkane concentration at the same conditions caused a decrease in KLa for all superficial gas velocities. At low yeast loading (0.5 g/l), the interfacial area increased to an average of 190 m2/m3 when the superficial gas velocity increased from 1 to 3 cm/sec. However, increasing yeast loading from 0.5 g/l to 6 g/l saw a decrease in the interfacial area to approximately 70 m2/m3 at the maximum levels of superficial gas velocities. The observed decrease in the interfacial area at high yeast loadings was a result of the increasing 𝐷32, likely due to the effect the yeast has on fluid properties. Increasing the hydrocarbon concentration from 2.5 to 10% v/v decreased KLa, possibly due to the increase in fluid viscosity and the surface tension, dampening turbulence in the system. However, further increases in alkane concentration above 10% v/v caused a significant increase in KLa for the lowest superficial gas velocities and constant yeast loading. An increase in the superficial gas velocity also decreased the values of KLa with an average of 0.3 s-1 at highest alkane concentration (20% v/v) for all constant yeast loading. The interfacial area had marginal increases of about 20 m2/m3 when the alkane concentration was increased from 2.5 to 20% v/v. This increase in interfacial area with increasing superficial gas velocity was due to the increase in the gas hold up, since 𝐷32 was not influenced by the alkane concentration. It can be concluded that these trends of KLa values were not entirely as expected, since the strongest effect came from alkane concentration, rather than superficial gas velocity, which is usually identified as the strongest factor in oxygen transfer studies. Further, the interfacial area was significantly influenced by the variations in superficial gas velocity (1 to 3 cm/sec). The oppositional effects between the interfacial area and KLa values suggest that the interaction between the yeast cells and the hydrocarbons may have changed the system hydrodynamics such that the oxygen transfer coefficient KL was effected, since interfacial area is seen to be fairly constant over the variable ranges. This work significantly furthers our understanding of the fluid mixing properties and KLa in a bubble column reactor, as well as the behaviour of the overall volumetric oxygen transfer coefficient in simulated hydrocarbon-based processes in this reactor configuration, which has the potential to impact and inform the operation of an industrially relevant hydrocarbon based bioprocess.

AFRIKAANS OPSOMMING: Dit is moontlik om lineêre alkaan koolwaterstowwe te omskep in hoër waarde produkte asook voorloperprodukte deur die toevoeging van ‘n suurstofgedeelte. Sulke koolwaterstowwe kan opgegradeer word deur biologies-bemiddelde oksidasie: ‘n suurstofgroep kan by die alkaan basis bygevoeg word deur ‘n mikro-organisme om ‘n meer reaktiewe en meer waardevolle molekule te produseer. Om dit te bewerkstellig is dit belangrik om operasionele toestande bevorderlik vir hierdie biologiese proses te behaal. Een noodsaaklike item is vloeibare fase suurstofkonsentrasie, wat geaffekteer word deur die suurstof oordragstempo in die reaksievloeistof in. Dié studie fokus op die gedrag van die algehele volumetriese suurstofoordragskoëffisiënt (KLa) en die gas-vloeistof oppervlakarea (a) in ‘n vierfase model (water, koolwaterstof, lug en mikrobiese vastestowwe) koolwaterstofbioproses wanneer dit deur ‘n borrel-kolom reaktor (BCR) gestuur word. Dit mik om die impak van die alkaan konsentrasie, oppervlakkige lugsnelheid en gisbelading op die KLa en die gas-vloeistof oppervlakarea van die stelsel te verduidelik. Die eksperimente is uitgevoer teen verskillende alkaan konsentrasies (van 2.5 – 20 vol% n-C14-20), diskrete oppervlakkige gas snelhede (van 1 – 3 cm/sek), en gisbelading (van 0.5–6 g/l Saccharomyces cerevisiae). Om suurstofbenutting te onderdruk sonder om handhawing van sel integriteit te verloor was die gisselle gedeaktiveer deur dit vir 60 minute teen 70°C te verhit. Die KLa is gemeet met ‘n opgeloste-suurstof (DO) meter en die gas uitlatings metode (GOP). Die stelsel se DO was elke 5 sekondes gemeet tot suurstof-versadiging bereik is. Die instrument se reaksie sloering (KP) was tydens alle eksperimentele toestande geëvalueer om die akkuraatheid van die KLa bepaling te verseker. Die gas-vloeistof oppervlakarea is gedefinieer as ‘n funksie van die gas ophou (Ɛ𝐺), en gemiddelde Sauter deursnit (D32). D32 is geëvalueer deur ‘n foto-analise algoritme. Die beeldontleding is uitgevoer met behulp van MATLAB sagteware, terwyl die Ɛ𝐺bereken is deur veranderinge in die werkende vloeibare vlakke in die reaktor te meet. Vir driefase-stelsels (gas-water-koolwaterstowwe) is dit gevind dat die stelsel nie-homogeen was by lae oppervlakkige lug snelheid met ‘n toeneming in koolwaterstof konsentrasie by die bopunt van die reaktor. Dus is die KLa en gas-vloeistof oppervlakarea metings deur die ruimtelike verskille in konsentrasies geaffekteer en kan dit nie as betroubaar beskou word nie. Om ‘n betroubare meting te vind is die stelsel geanaliseer vir ruimtelike vloeibare fase-homogeniteit deur ‘n fisiese steekproefneming metodologie te gebruik. Dit is dus bevestig dat (𝑢𝐺) teen 1.8 cm/sek optimale homogene vloei in ‘n driefase stelsel bereik word. In die vierfase-stelsel het die byvoeging van gis tot die BCR gelei tot ‘n verandering in die vloeisisteem en dus ‘n verhoging in die homogeniteit van die stelsel teen lae vloei (1 cm/s). Hierdie resultaat bevestig dat die toevoeging van gisselle in die reaktor die vloeistof eienskappe soos viskositeit, borrel samesmelting, sowel as die oppervlakspanning verander. Die impak van gisbelading op KLa en gas-vloeistof oppervlakarea tydens meervoudige fase koolwaterstof-gebasseerde bioprosesse in ‘n BCR is eerste ondersoek. Dit is gevind dat ‘n toename in die gisbelading gelei het tot ‘n afname in KLa in die driefase (lug-water-gis) stelsel teen ‘n konstante. Hierdie afname kan toegeskryf word aan die bestaan van diffusie blokkerende effekte asook ‘n toenemende vloeistofsviskositeit. ‘n Soortgelyke effek is in ‘n vierfase (lug-water-koolwaterstof-gis) stelsel aangetref; verlaagde KLa waardes en toenemende gislading is opgemerk by hoë oppervlakkige gas snelheid met ‘n konstante alkaan konsentrasie (onder 10% v/v). Daar is egter ‘n afname in KLa waardes opgemerk by alle oppervlakkige gas snelheid wanneer die alkaan konsentrasie op dieselfde toestande verhoog is. By lae gislading (0,5 g/l) het die gas-vloeistof oppervlakarea toegeneem tot ‘n gemiddelde van 190 m2/m3 wanneer die oppervlakkige gas snelheid verhoog is (van 1 – 3 cm/sek). Wanneer gislading van 0,5 g/l tot 6 g/l verhoog is is ‘n afname in die gas-vloeistof oppervlakarea na ongeveer 70 m2/m3 bevestig teen maksimum vlakke van oppervlakkige gas snelhede. Die afname in die gas-vloeistof oppervlakarea by hoë gisbeladings was die gevolg van hierdie toename, waarskynlik as gevolg van die effek van die gis op vloeibare eienskappe. ‘n Verhoging van 2,5 tot 10% v/v koolwaterstof konsentrasie het gelei tot ‘n afname in KLa met enige verhogings in vloeistof viskositeit en oppervlakspanning sowel as ‘n demping van die onstuimigheid in die stelsel. Verdere toename in alkaan konsentrasie (bo 10% v/v) het wel tot ‘n beduidende toename in KLa gelei vir die laagste oppervlakkige gas snelhede en konstante gislading. ‘n Verhoging in die oppervlakkige gassnelheid het verder die KLa waardes met ‘n gemiddeld van 0.3 s-1 verlaag by hoogste alkaan konsentrasie (20% v/v) vir alle konstante gisladings. Die gas-vloeistof oppervlakarea het minimale verhogings van ongeveer 20 m2/m3 getoon wanneer die alkaan konsentrasie van 2,5 tot 20% v/v verhoog is. Hierdie toename in die gas-vloeistof oppervlakarea met toenemende oppervlakkige gassnelheid was te danke aan die klein toename in 𝐷32 aangesien dit nie deur die alkaan konsentrasie beïnvloed is nie. Daar kan afgelei word dat die KLa waardes se neigings nie heeltemal na verwagting was nie aangesien die sterkste verandering van alkaan konsentrasie af gekom het eerder as oppervlakkige gas snelheid, wat gewoonlik die sterkste faktor in suurstofoordrag is. Verder was die gas-vloeistof oppervlakarea aansienlik beïnvloed deur die variasies in oppervlakkige gassnelheid (1 tot 3 cm/sek). Die teenoorgestelde effekte tussen die gas-vloeistof oppervlakarea en KLa dui dat die interaksie tussen die gisselle en die koolwaterstowwe die stelsel hidrodinamika verander het wat gelei het tot ‘n veranderde suurstof oorplasing koëffisiënt KL. Hierdie werk bevorder ons begrip van vloeistofmengsel eienskappe en KLa in ‘n borrel-kolom reaktor wat die potensiaal het om die ‘n industrieel relevante koolwaterstof gebaseerde bioproses beter te verstaan en moontlik te bevorder.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/103483
This item appears in the following collections: