A fluid dynamic study of the retrofitting of large agitated bioreactors: Turbulent flow

Studies were conducted in three 19-m(3) fermentors (14 m(3) working volume, aspect ratio = 3:1), one fitted with four Rushton turbines (D/T = 0.35), one with three Lightnin' A315 hydrofoil impellers (D/T = 0.46). The power drawn under the same aerated conditions relative to the unaerated ones was always greater with the hydrofoils, which gives them the potential for enhanced mass transfer rates under practical operating conditions. However, the power draw was also sensitive to the magnitude of the unaerated power. Indeed, at low unaerated specific power ( approximately 0.6 W-kg) and high air flow rates ( approximately 1vvm), the relative power draw with the hydrofoils could be even greater than 1. The hold-up with each of the impellers was broadly similar at the same aeration rate and power input, though the later had a much smaller impact in these large vessels than has been reported in the literature based on smaller scale work. As usual, repressed coalescence caused increased hold-up, and, with the hydrofoils, this increase was associated with a lower power draw. Because of the greater mechanical vibration of the reactors with the hydrofoils, vibration characteristics of the vessels were measured and they were very similar. The results showed that provided care is taken in the mechanical design of the system, such impellers can operate reliably in large-scale fermentations with the potential for enhanced biological performance. (c) 1994 John Wiley & Sons, Inc.     

Power consumption of three impeller combinations in mixing Xanthan fermentation broths

Xanthan gum fermentation represents a good model for the study of the mixing of rheologically complex culture broths. Most of the previous work on power consumption dealt with ‘standard’, single impellers and used model fluids to simulate xanthan broths. This work describes the characterization of three dual-impeller combinations (D/T = 0·53) for the mixing of dehydrated—reconstituted fermentation broths of Xanthomonas campestris that had matched rheology to the actual broths. The bottom impeller was a Rushton turbine (RT) and the top impeller was another RT, a 45° pitched blade turbine (PT) or an A-310 Lightnin mixer (A310). The experiments were carried out in a tank of 0·0094 m³ working volume equipped with an air bearing dynamometer. The power was measured in a wide range of xanthan concentrations (5–40 kg m⁻³) in aerated (0·25, 0·5 and 1·0 vvm) and unaerated conditions. Unaerated power number (Po) vs. Reynolds number (Re) curves showed similar trends for the three combinations. Exponents close to −1 were obtained in the laminar region. A minimum in Po (Po_min) occurred at Re = 30–40, then increasing to a plateau value which was evident at Re> 200. In the transition region Po_min values were 4·3 (RT and RT), 3·6 (RT and PT) and 2·4 (RT and A310). The aerated power data for (RT and PT) and (RT and A-310) showed higher torque instabilities than the dual RT combinations at higher xanthan concentrations. The higher the xanthan concentrations, the higher the drop in power and the less important the effect of the aeration rate. Among the combinations tested, when using Rushton turbines, the well-mixed ‘cavern’ reached the tank wall (i.e., fluid motion was observed) at the lowest volumetric power input. High     

Mass transfer and mixing rates in fermentation vessels

The effects on mass-transfer and overall mixing rates of varying impeller geometry and operating speed have been studied for flat-bladed turbines in laboratory fermentors, in aerated aqueous solutions, and in unaerated and aerated suspensions (1.6% w/v) of paper pulp. In the absence of suspended solid, oxygen absorption rates could be correlated directly with power input. In the pulp suspension, oxygen absorption at a given power input was influenced by impeller geometry and operating speed. The data for the three-phase system can be correlated by a dimensionless equation relating oxygen-transfer rates and mixing times to the geometrical and operating parameters of the impellers.     

Gas-liquid dispersion with dual Rushton turbine impellers

Aerated and unaerated power consumption and flow patterns in a 0.56 m diameter agitated vessel containing water with dual Rushton turbines have been studied. Under unaerated conditions with a liquid height-to-diameter ratio of 2, an impeller spacing of 2 to 3 times the impeller is required for each to draw an amount of power equal to a single impeller. For aerated conditions, if a similar spacing is used, equations for the flooding-loading transition and for power consumption for a single Rushton impeller can be extended relatively easily to dual systems. All results for this spacing are explained by reference to bulk flow patterns and gassed-filled cavity structures and the proportion of sparged gas flowing through the upper impeller is also estimated. Such a spacing is generally recommended since it maximizes the power draw and hence the potential for oxygen mass transfer. Data are presented for other spacings but the results do not fit in easily with single agitator studies because strong impeller-impeller flow pattern interactions occur.     

Impact of nozzle operation on mass transfer in jet aerated loop reactors. Characterization and comparison to an aerated stirred tank reactor

The impact of mass transfer on productivity can become a crucial aspect in the fermentative production of bulk chemicals. For highly aerobic bioprocesses the oxygen transfer rate (OTR) and productivity are coupled. The achievable space time yields can often be correlated to the mass transfer performance of the respective bioreactor. The oxygen mass transfer capability of a jet aerated loop reactor is discussed in terms of the volumetric oxygen mass transfer coefficient k_La [h^?1] and the energetic oxygen transfer efficiency E [kgO₂ kW^?1 h^?1]. The jet aerated loop reactor (JLR) is compared to the frequently deployed aerated stirred tank reactor. In jet aerated reactors high local power densities in the mixing zone allow higher mass transfer rates, compared to aerated stirred tank reactors. When both reactors are operated at identical volumetric power input and aeration rates, local k_La values up to 1.5 times higher are possible with the JLR. High dispersion efficiencies in the JLR can be maintained even if the nozzle is supplied with pressurized gas. For increased oxygen demands (above 120 mmol L^?1 h^?1) improved energetic oxygen transfer efficiencies of up to 100 % were found for a JLR compared to an aerated stirred tank reactor operating with Rushton turbines.     

Mixing design for enzymatic hydrolysis of sugarcane bagasse: methodology for selection of impeller configuration

One of the major process bottlenecks for viable industrial production of second generation ethanol is related with technical–economic difficulties in the hydrolysis step. The development of a methodology to choose the best configuration of impellers towards improving mass transfer and hydrolysis yield together with a low power consumption is important to make the process cost-effective. In this work, four dual impeller configurations (DICs) were evaluated during hydrolysis of sugarcane bagasse (SCB) experiments in a stirred tank reactor (3 L). The systems tested were dual Rushton turbine impellers (DIC1), Rushton and elephant ear (down-pumping) turbines (DIC2), Rushton and elephant ear (up-pumping) turbines (DIC3), and down-pumping and up-pumping elephant ear turbines (DIC4). The experiments were conducted during 96 h, using 10 % (m/v) SCB, pH 4.8, 50 °C, 10 FPU/g_biomass, 470 rpm. The mixing time was successfully used as the characteristic parameter to select the best impeller configuration. Rheological parameters were determined using a rotational rheometer, and the power consumptions of the four DICs were on-line measured with a dynamometer. The values obtained for the energetic efficiency (the ratio between the cellulose to glucose conversion and the total energy) showed that the proposed methodology was successful in choosing a suitable configuration of impellers, wherein the DIC4 obtained approximately three times higher energetic efficiency than DIC1. Furthermore a scale-up protocol (factor scale-up 1000) for the enzymatic hydrolysis reactor was proposed.     

Improved performance in viscous mycelial fermentations by agitator retrofitting

For viscous mycelial fermentations it was demonstrated at the pilot-plant scale that the replacement of standard radial flow Rushton turbines with larger diameter axial-flow Prochem hydrofoil impellers significantly improved oxygen transfer efficiency. It was also determined that the Streptomyces broth under evaluation is highly shear thinning. Separate experiments using a Norcardia broth with similar Theological properties demonstrated that the oxygen transfer coefficient, K(L)a, can be greatly increased by use of water additions to reduce broth viscosity. These observations are consistent with the hypothesis that the improvement in oxygen transfer by changing agitator types is primarily due to an improvement in bulk mixing. A model is presented, based on the concepts of Bajpai and Reuss, which explains this improvement in performance in terms of enlargement of the well mixed micromixer region for viscous mycelial broths.     

Optimization of oxygen mass transfer in a multiphase bioreactor with perfluorodecalin as a second liquid phase

Oxygenation is an important parameter involved in the design and operation of mixing-sparging bioreactors and it can be analyzed by means of the oxygen mass transfer coefficient (k(L)a). The operational conditions of a stirred, submerged aerated 2-L bioreactor have been optimized by studying the influence of a second liquid phase with higher oxygen affinity (perfluorodecalin or olive oil) in the k(L)a. Using k(L)a measurements, the influence of the following parameters on the oxygen transfer rate was evaluated: the volume of working medium, the type of impellers and their position, the organic phase concentration, the aqueous phase composition, and the concentration of inactive biomass. This study shows that the best experimental conditions were achieved with a perfluorodecalin volume fraction of 0.20, mixing using two Rushton turbines with six vertical blades and in the presence of YPD medium as the aqueous phase, with a k(L)a value of 64.6 h(-1). The addition of 20% of perfluorodecalin in these conditions provided a k(L)a enhancement of 25% when pure water was the aqueous phase and a 230% enhancement when YPD medium was used in comparison to their respective controls (no perfluorodecalin). Furthermore it is shown that the presence of olive oil as a second liquid phase is not beneficial to the oxygen transfer rate enhancement, leading to a decrease in the k(L)a values for all the concentrations studied. It was also observed that the magnitude of the enhancement of the k(L)a values by perfluorodecalin depends on the biomass concentration present.     

Studies on transfer processes in mixing vessels: effect of particles on gas-liquid hydrodynamics using modified Rushton turbine agitators

The work presents the effect of solid particles having a mean diameter between 15–1000?μm, on the gas dispersion in a mechanically agitated vessel with standard and modified Rushton turbine agitators positioned singly or doubly on same shaft. For the dispersing and uniform distribution of the three phase (gas-liquid-solid) through the entire vessel section, the modified blade turbines, with the surface fraction of the perforations equal to 0.353, were found to be more efficient, the power consumption being reduced by approximately 50%in comparison with the standard Rushton turbines. The power number in the turbulent mixing of the three phase system is dependent on the aeration rate, the surface fraction of the perforations, the turbine number and the physical and rheological properties of the suspensions.     

Interaction of cultural conditions and end-product distribution in Bacillus subtilis grown in shake flasks

Summary A culture of Bacillus subtilis, in which the relative production of acetoin (Ac) and butanediol (Bu) is highly sensitive to oxygen tension as well as to mixing conditions, was used to evaluate several culture conditions in 500-ml shake flasks. The concentration ratio of these metabolites (Ac/Bu) produced in a defined period of culture time was used as a parameter for comparative purposes. The influence of working volume, shaking speed, broth viscosity and the presence of baffles were evaluated. Using unbaffled flasks it was found that working volume had the most influence on oxygenation in shake flasks, especially below 10%, where differences in Ac/Bu ratios up to ten times could be measured. Shaking speed played an important role only at values higher than 400 rpm or when small working volumes were used. The addition of xanthan gum decreased the Ac/Bu ratio nearly four times under equivalent working conditions and also diminished the influence of shaking speed. In general, Ac/Bu was higher when sulphite oxygen transfer rate (OTR) values were higher. However, the test culture was able to detect differences which were not evident using the OTR method. Comparing Ac/Bu ratios in stirred fermentors from the literature, it seems that similar oxygenation conditions can be reached in non-baffled shake flasks only at very high shaking speeds using small working volumes. With baffled flasks, our data suggest that better oxygenation and mixing can be achieved in shake flasks if compared with those obtained in stirred fermentors at conventional power inputs.     

Flow Pattern and Mixing in the Multi‐Turbine Agitated Bioreactor Filled with Poly(γ‐Glutamic Acid) Solution

We investigated the flow pattern and mixing behavior of a poly(γ‐glutamic acid) (γ‐PGA) solution in a bioreactor equipped with two Rushton turbines by simulation and experiment. Computational fluid dynamics (CFD) is used to solve the three‐dimensional hydrodynamics in the bioreactor and to obtain the flow patterns and tracer concentration at every point. The flow circulation patterns by inter‐impeller clearance and viscosity and their effects on overall mixing time were studied. Based on the results we can conclude that the impeller clearance should not be larger than 0.2 D for the efficient mixing under non‐aerated condition when the liquid viscosity is above 20 cp, which corresponds to concentrations of 20 g/L or above for γ‐PGA.     

Scale-up of stirring as foam disruption (SAFD) to industrial scale

Foam disruption by agitation—the stirring as foam disruption (SAFD) technique—was scaled up to pilot and production scale using Rushton turbines and an up-pumping hydrofoil impeller, the Scaba 3SHP1. The dominating mechanism behind SAFD—foam entrainment—was also demonstrated at production scale. The mechanistic model for SAFD defines a fictitious liquid velocity generated by the (upper) impeller near the dispersion surface, which is correlated with complete foam disruption. This model proved to be scalable, thus enabling the model to be used for the design of SAFD applications. Axial upward pumping impellers appeared to be more effective with respect to SAFD than Rushton turbines, as demonstrated by retrofitting a 12,000 l bioreactor, i.e. the triple Rushton configuration was compared with a mixed impeller configuration from Scaba with a 20% lower ungassed power draw. The retrofitted impeller configuration allowed 10% more broth without risking excessive foaming. In this way a substantial increase in the volumetric productivity of the bioreactor was achieved. Design recommendations for the application of SAFD are given in this paper. Using these recommendations for the design of a 30,000 l scale bioreactor, almost foamless Escherichia coli fermentations were realised. Electronic Publication     

States of aggregation and phase transformations in mixtures of phosphatidylcholine and octyl glucoside

The result of mixing varying concentrations of the nonionic detergent octyl glucoside (OG) with small unilamellar vesicles (SUV) of egg phosphatidylcholine (PC) made by sonication depends on the ratio between OG and PC in the mixed aggregates. When this molar ratio (Re) is lower than 1.4, the detergent partitions between the PC vesicles and the aqueous medium with a partition coefficient of K = 0.033 mM-1. As a consequence of introduction of OG into the bilayers, the vesicles grow in size. The resultant vesicles have a mean diameter that is an increasing function of Re and is independent of the total PC concentration. Experiments in which the vesicles were loaded with high molecular weight dextran prior to being exposed to OG suggest that the mechanism responsible for the size growth involves lipid transfer rather than fusion. Mixtures with Re values within the range of 1.4-3.2 separate into two macroscopic phases: The lower phase is clear but very viscous. It contains constant OG and PC concentrations and is characterized by an Re value of 3.2, independent of the composition of the whole dispersion. The upper phase contains vesicles of varying concentrations of OG and PC, but a constant Re of 1.4. When the saturating level of 1.4 OG molecules per PC molecule is approached, the concentration of OG monomers in the aqueous medium reaches the value of 16.6 +/- 0.3 mM, which is the apparent cmc of OG in the lipid-containing medium. OG-PC mixed micelles contain at least 3.2 OG molecules per PC molecule. The mixed micelles present at Re = 3.2 apparently have the shape of oblate ellipsoids with a minor axis of about 2 nm and two major axes of about 25 nm. The surface area of the mixed micelles at this point is just sufficient for them to undergo conversion into the smallest possible spherical vesicles of a radius of 12 nm. At Re values above 3.2, the major axis of the mixed micelles becomes smaller as Re increases, while at values of Re below 3.2 the micelles would have been expected to grow very rapidly with decreasing Re. This may explain the partial vesicle closure occurring below Re = 3.2.     

Temperature limitation of methanogenesis in aquatic sediments.

Microbial methanogenesis was examined in sediments collected from Lake Mendota, Wisconsin, at water depths of 5, 10, and 18 m. The rate of sediment methanogenesis was shown to vary with respect to sediment site and depth, sampling date, in situ temperature, and number of methanogens. Increased numbers of methanogenic bacteria and rates of methanogenesis correlated with increased sediment temperature during seasonal change. The greatest methanogenic activity was observed for 18-m sediments throughout the sampling year. As compared with shallower sediments, 18-m sediment was removed from oxygenation effects and contained higher amounts of ammonia, carbonate, and methanogenic bacteria, and the population density of methanogens fluctuated less during seasonal change. Rates of methanogenesis in 18-m sediment cores decreased with increasing sediment depth. The optimum temperature, 35 to 42 C, for sediment methanogenesis was considerably higher than the maximum observed in situ temperature of 23 C. The conversion of H2 and [14C]carbonate to [14C]methane displayed the same temperature optimum when these substrates were added to sediments. The predominant methanogenic population had simple nutritional requirements and were metabolically active at 4 to 45 C. Hydrogen oxidizers were the major nutritional type of sediment methanogens; formate and methanol fermentors were present, but acetate fermentors were not observed. Methanobacterium species were most abundant in sediments although Methanosarcina, Methanococcus, and Methanospirillum species were observed in enrichment cultures. A chemolithotropic species of Methanosarcina and Methanobacterium was isolated in pure culture that displayed temperature optima above 30 C and had simple nutritional requirements.     

Development of a method for reliable power input measurements in conventional and single‐use stirred bioreactors at laboratory scale

Power input is an important engineering and scale‐up/down criterion in stirred bioreactors. However, reliably measuring power input in laboratory‐scale systems is still challenging. Even though torque measurements have proven to be suitable in pilot scale systems, sensor accuracy, resolution, and errors from relatively high levels of friction inside bearings can become limiting factors at smaller scales. An experimental setup for power input measurements was developed in this study by focusing on stainless steel and single‐use bioreactors in the single‐digit volume range. The friction losses inside the air bearings were effectively reduced to less than 0.5% of the measurement range of the torque meter. A comparison of dimensionless power numbers determined for a reference Rushton turbine stirrer (N_P = 4.17 ± 0.14 for fully turbulent conditions) revealed good agreement with literature data. Hence, the power numbers of several reusable and single‐use bioreactors could be determined over a wide range of Reynolds numbers between 100 and >10⁴. Power numbers of between 0.3 and 4.5 (for Re = 10⁴) were determined for the different systems. The rigid plastic vessels showed similar power characteristics to their reusable counterparts. Thus, it was demonstrated that the torque‐based technique can be used to reliably measure power input in stirred reusable and single‐use bioreactors at the laboratory scale.     

Mixing and power characteristics of drag board device in shallow pool

Mixing and stirring of a 20-m(2) shallow pool by means of a drag board device has been investigated. The board closes the pool cross section except for a slit of a few centimeters above the bottom, and it is slowly moved back and forth, forcing the water to run through the slit and thereby creating a turbulent backwhirl. Power drawn and the drag on the board has been measured together with the velocities of the water at different locations in the wake of the board. Power number N(p) has been correlated with the Reynolds N(Re) and the bottom clearance numbers N(c) by the expressions \documentclass{article}\pagestyle{empty}\begin{document}$$ N_p = 13,465N_{\rm Re};{-0.774} N_c;{0.1016} N_{\rm Re} < 80.000\\N_p = 5.4N_{\rm Re};{-0.0863} N_c;{0.104} N_{\rm Re} > 80.000 $$\end{document} Power and Reynolds numbers are defined as usual with the square root of the board-immersed-area as the characteristic length. The bottom clearance number is defined as the ratio of the water depth in the pool to the difference between water depth and the board width immersed in water. Flow pattern behind the board consists of large vortex loops causing the fluid to circulate from bottom to top and producing a thorough mixing effect. The drag board seems to have several advantages over conventional paddle wheels for the mixing and stirring of algal cultures in shallow ponds.     

Polysaccharide production: Experimental comparison of the performance of four mixing devices

This investigation was undertaken with the objective to compare experimentally the performance of four different mixing devices for the production of the polysaccharide pullulan with Aureobasidium pullulans (2552). Fermentations were performed using identical bioreactors with respectively an assembly of three Rushton turbines (RTB), a helical ribbon impeller (HR) and two different reciprocating plates (RPB1, RPB2). Each mixing vessel had identical geometry and working volume (18 L). These four fermentations were performed with an equal level of power input per unit volume (1000?W/m³) and gas flow rate (0.5 vvm, 9 L/min). For each system, the evolution of biomass, polysaccharide concentration, dissolved oxygen and agitation speed or frequency were recorded as a function of time along with the rheological properties of the culture broths. The type of mixing device used had a significant impact on the rate of biomass production and on polysaccharide physical properties. However, the rate of polysaccharide production appears to be less sensitive to the bioreactor design. The overall productivity, which represents the ability of micro-organisms to convert rapidly substrate into biomass and polysaccharide, was maximised using the RPB2 system. The quality of the synthesised polysaccharide, in terms of viscosity producing power, was highest using the HR system but the overall productivity was very low. Thus, the best compromise between productivity and product quality was achieved with the RPB2.     

Study of drop and bubble sizes in a simulated mycelial fermentation broth of up to four phases

The mean sizes and size distributions of air bubbles and viscous castor oil drops were studied in a salt-rich aqueous solution (medium), first separately, and then simultaneously as a three-phase system. The dispersion was created in a 150-mm-diameter stirred tank equipped with a Rushton turbine, and the sizes were measured using an advanced video technique. Trichoderma harzianum biomass was added in some experiments to study the effect of a solid phase under unaerated and aerated conditions to give either three-or four-phase systems. In all cases, the different dispersed phases could be clearly seen. Such photoimages have never been obtained previously. For the three phases, air-oil-medium, aeration caused a drastic increase in Sauter mean drop diameter, which was greater than could be accounted for by the reduction in energy dissipation on aeration. Also, as in the unaerated case, larger drops were observed as the oil content increased. On the other hand, mean bubble sizes were significantly reduced with increasing oil phase up to 15% with bubbles inside many of the viscous drops. With the introduction of fungal biomass of increasing concentration (0.5 to 5 g L(-1)) under unaerated conditions, the Sauter mean drop diameter decreased. Finally, in the four-phase system (oil [10%]-medium-air-biomass) as found in many fermentations, all the phases (plus bubbles in drops) could clearly be seen and, as the biomass increased, a decrease in both the bubble and the drop mean diameters was found. The reduction in size of bubbles (and therefore increase in interfacial area) as the oil and bio- mass concentration increased provides a possible explanation as to why the addition of an oil phase has been reported to enhance oxygen transfer during many fermentations.     

First- and second-approximation calculations in the relaxation function of high-sugar/polysaccharide systems

The distribution function of Maxwellian relaxation times (phi) was derived from the small- deformation dynamic properties of high-sugar/agarose, /deacylated gellan and /high-methoxy pectin mixtures. First-approximation calculations of phi employed the time derivative of the experimentally measured storage and viscous moduli, with the two traces converging at the theoretically predicted slope of - 0.5. Second-approximation calculations were based on phi, as derived by the first approximation, being a simple power function of relaxation times (tau(-m)). The slope m was measured at various points and used to derive correction factors for shifting the relaxation function to the second approximation. Thus, values of phi calculated from G' and G" were brought into satisfactory agreement, particularly in the recorded portion of the glass transition region. Once the function phi is accurately determined, it can be readily used for calculations of other viscoelastic properties.     

Substrate gradient formation in the large-scale bioreactor lowers cell yield and increases by-product formation

A heterogeneous micro-environment was identified in a 12 m³ bioreactor with a height-to-diameter ratio of 2.5. The reactor was aerated by a ring sparger and stirred by three Rushton turbines. E. coli cells were cultivated in minimal medium to a cell density in the order of 30?g/l. Samples of glucose, the growth limiting component fed to the process, were taken at three levels in the bioreactor (top/middle/bottom). These showed that glucose concentration declined away from the feedpoint. The gradients depended on the mixing characteristics of the feedpoint, and concentrations of up to 400 times the mean value were found when feed was added to a relatively stagnant mixing zone. This resulted in up to 20% lower biomass yield compared to the bench scale. Gradients also affected the by-product formation, resulting in acetate formation in the large-scale bioreactor. IPTG induction of a recombinant protein was shown to influence important cell parameters and considerably increased the yield of carbon dioxide per glucose added, indicating an increased maintenance. The product formation rate was, however, not notably affected by the scale-up.