Measurement and behavior of the overall volumetric oxygen transfer coefficient in aerated agitated alkane based multiphase systems

[摘要] ENGLISH ABSTRACT: Hydrocarbons provide excellent feed stocks for bioconversion processes to producevalue added products using various micro-organisms. However, hydrocarbon-basedaerobic bioprocesses may exhibit transport problems where the bioconversion islimited by oxygen supply rather than reaction kinetics. Consequently, the overallvolumetric oxygen transfer coefficient (KLa) becomes critical in designing, operatingand scaling up of these processes. In view of KLa importance in hydrocarbon-basedprocesses, this work evaluated KLa measurement methodologies as well asquantification of KLa behavior in aerated agitated alkane-solid-aqueous dispersions.A widely used KLa measurement methodology, the gassing out procedure (GOP) wasimproved. This improvement was done to account for the dissolved oxygen (DO)transfer resistances associated with probe. These resistances result in a lag in DOresponse during KLa measurement. The DO probe response lag time, wasincorporated into the GOP resulting in the GOP (lag) methodology. The GOP (lag)compared well with the pressure step procedure (PSP), as documented in literature,which also incorporated the probe response lag time.Using the GOP (lag), KLa was quantified in alkane-solid-aqueous dispersions, usingeither inert compounds (corn flour and CaCO3) or inactive yeast cells as solids torepresent the micro-organisms in a hydrocarbon bioprocess. Influences of agitation,alkane concentration, solids loading and solids particle sizes and their interactions onKLa behavior in these systems were quantified.In the application of an accurate KLa measurement methodology, the DO proberesponse lag time was investigated. Factors affecting the lag, which included processconditions such as agitation (600-1200rpm), alkane concentration (2.5-20% (v/v),alkane chain length (n-C10-13 and n-C14-20), inert solids loading (1-10g/L) and solidsparticle sizes (3-14μm) as well as probe characteristics such as membrane age andelectrolyte age (5 day usage), were investigated. Kp, the oxygen transfer coefficient ofthe probe, was determined experimentally as the inverse of the time taken for the DOto reach 63.2% of saturation after a step change in DO concentration. Kp dependenceon these factors was defined using 22 factorial design experiments. Kp decreased onincreased membrane age with an effect double that of Kp decrease due to electrolyteage. Additionally, increased alkane concentration decreased Kp with an effect 7 times higher compared to that of Kp decrease due to increased alkane chain length. Thiswas in accordance to Pareto charts quantification.KLa was then calculated, using the GOP (lag), according to equation [1] whichincorporates the influence of Kp. Equation 1 is derived from the simultaneous solutionof the models which describe the response of the system and of the probe to a stepchange in DO.11*L ppp K at K tLp p LaCK e K aeC K K= -  - - -  -  [1]The KLa values documented in literature from the PSP and KLa calculated by theGOP (lag) showed only a 1.6% difference. However KLa values calculated by theGOP (lag) were more accurate than KLa calculated by the GOP, with up to >40% errorobserved in the latter according to t-tests analyses. These results demonstrated thatincorporating Kp markedly improved KLa accuracy. Consequently, the GOP (lag) waschosen as the preferred KLa measurement methodology.KLa was determined in n-C14-20-inert solid-aqueous dispersions. Experiments wereconducted in a stirred tank reactor with a 5L working volume at constant aeration of0.8vvm, 22ºC and 101.3kPa. KLa behavior across a range of agitations (600-1200rpm), alkane concentrations (2.5-20% (v/v)), inert solids loadings (1-10g/L) andsolids particle sizes (3-14μm) was defined using a 24 factorial design experiment. Inthese dispersions, KLa increased significantly on increased agitation with an effect 5times higher compared to that of KLa increase due to interaction of increased alkaneconcentration and inert solids loading. Additionally, KLa decreased significantly onincreased alkane concentration with an effect 4 times higher compared to both that ofincreased solids particle sizes and the interaction of increased agitation and solidsparticle size.In n-C14-20-yeast-aqueous dispersions, KLa was determined under narrowed processconditions better representing typical bioprocess conditions. KLa behavior across arange of agitations (600-900rpm), alkane concentrations (2.5-11.25% (v/v)) and yeastloadings (1-5.5g/L) using a 5μm-yeast cell was defined using a 23 factorial designexperiment. In these dispersions, KLa increased significantly on increased agitation.Additionally, KLa decreased significantly on increased yeast loading with an effect 1.2times higher compared to that of KLa decrease due to interaction of increased alkaneconcentration and yeast loading. In this study, the importance of Kp for accurate KLa measurement in alkane basedsystems has been quantified and an accurate and less complex methodology for itsmeasurement applied. Further, KLa behavior in aerated alkane-solid-aqueousdispersions was quantified, demonstrating KLa enhancement on increased agitationand KLa depression on increased alkane concentration, solids loading and solidsparticle sizes.

[发布机构] Stellenbosch University