Experimental visualization of mixing pathologies in laminar stirred tank bioreactors

Experimental visualization of mixing pathologies in laminar stirred tank bioreactors

Chemical Engineering Science 60 (2005) 2449 – 2457 www.elsevier.com/locate/ces Experimental visualization of mixing pathologies in laminar stirred ta...

621KB Sizes 0 Downloads 13 Views

Chemical Engineering Science 60 (2005) 2449 – 2457 www.elsevier.com/locate/ces

Experimental visualization of mixing pathologies in laminar stirred tank bioreactors M.M. Alvareza,∗ , A. Guzmána , M. Elíasb a Centro de Biotecnología. Instituto Tecnológico y de Estudios Superiores de Monterrey, Avenida Eugenio Garza Sada 2501, Sucursal de Correos “J”,

Monterrey, N.L. 64849, México b Planta Piloto de Fermentaciones, Universidad Autónoma de Nuevo León, Monterrey, N.L., México

Received in revised form 15 October 2004; accepted 5 November 2004

Abstract Most of the bio-reactive processes now in development by Pharmaceutical and Biotechnological firms will be based on the culture of shear-sensitive mammalian, insect or plant cells. To properly address the need of designing, optimizing, and scaling up such bioprocesses, a precise characterization of the flow and mixing behavior of typical and novel bioreactor geometries is required. Using experimental visualization techniques (3-D UV light pattern visualization, 2-D laser-induced fluorescence techniques, and acid base reactions in the presence of pH indicators) we experimentally describe and analyze common mixing pathologies occurring in frequently used bioreactor geometries: heterogeneity, coexistence of regular and chaotic regions, flow segregation, compartmentalization, flow bypass, cell focusing, etc. We also analyze the role of baffles on mixing in bioreactors operated at low and moderate Re. and discuss the risk and validity of assuming perfectly mixed conditions in a bio-reaction system. 䉷 2005 Elsevier Ltd. All rights reserved.

1. Introduction Mixing has been recognized as crucial in reactive process operations. However, reactive mixing in bio-systems has received poor research attention, despite its increasing industrial relevance. Transport limitations have been regarded as one of the major problems leading to process yield reduction in largescale bioprocesses (Vrábel et al., 2001). At the macroscale, mixing issues arising from geometrical and operational factors have been documented conclusively. The surprisingly wide distribution of shear stresses within a vessel has also been demonstrated in laboratory and industrial-scale equipment (see for example Alvarez et al., 2002; Alvarez, 2000; Zalc et al., 2001, and the very complete critical review on multiple impeller geometries by Gogate et al., 2000). Also,

∗ Corresponding author.

E-mail address: [email protected] (M.M. Alvarez). 0009-2509/$ - see front matter 䉷 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2004.11.049

various problems related to mixing have been reported on the operation of large-scale bioreactors. Recently, Jian Li et al. (2002) reported the first large-scale study (80 m3 ) involving quantitation of recombinant enzyme expression at different impeller power supplies in a fungus fermentation. Further evidence on the relationship between adequate mixing and appropriate bioreactor performance has been published. Ariff et al. (1997) discussed the effect of mixing on an enzymatic reaction; Vrábel et al. (2001) documented top–bottom gradients in glucose concentrations (limiting substrate) in large-scale fermentors; and Enfors et al. (2001) present some experimental evidence of complex physiological responses to mixing in large-scale bioreactors. Although most of the current mixing operations in Biotechnology applications are turbulent, laminar and transitional scenarios are becoming industrially more relevant. Consequently they are beginning to capture more research attention. As examples, Peña et al. (2002) referred to mixing effects in high viscosity fermentations; Unger et al. (2000) studied particle trajectories in roller bottles;

2450

M.M. Alvarez et al. / Chemical Engineering Science 60 (2005) 2449 – 2457

Berson et al. (2002) documented a more than two-fold improvement in reactor performance in roller bottles with enhanced mixing; and Weiss et al. (2002) experimentally studied mixing in 96-well microplates using fluorescence indicators. Pharmaceutical and Biotechnological companies are increasingly focusing efforts on the development of high-value drugs based on biotechnology (bio-therapeutics). This includes the use of shear-sensitive mammalian and insect cells, or highly concentrated and viscous cell broths. Although today there are only a few of these products already in the market, in the years to come more of the now “in development” drugs will need to hit pilot plant and industry-scale reactors. A decade ago, when Genentech Inc. (the second largest Biotech company worldwide) started operations, 90% of their products were obtained from turbulent bacterial bioreactors. Today, Genentech already produces more than 75% of their products using recombinant mammalian cells. We are not technically ready to resolve the mixing issues that can potentially arise in such scenarios. Many particularities and constraints of these bio-systems preclude the successful application of mixing technology (and mixing criteria) normally used in inert turbulent reactive systems. Indeed, we are far from reaching goals such as an optimum mixing design of a bioreactor for highly shearsensitive cells, and even further away from understanding how to scale up such unit operations. To mix in situations where mammalian, insect or plant cells are handled is an extremely challenging proposition (Elias et al., 1995; Nikolai and Hu, 1992; Gregoriades et al., 2000; Aloi and Cherry, 1994). We intent to assure homogeneous conditions for cell growth, while preserving the metabolic functionality of shear-sensitive cells. This can be better done by operating in laminar conditions. While in turbulent regime we rely on the injection of energy through a cascade of eddies, in the laminar regime the only mechanism for mixing is chaos. However, as described by numerous authors, chaos in a stirred tank is not necessarily widespread homogeneously throughout the tank space (Lamberto et al., 1996; Alvarez et al., 1998; Lamberto et al., 1999, 2001). Chaotic and regular regions do coexist in dynamical systems, and stirred tanks are not the exception. In terms of flow behavior, laminar stirred tanks are a very rich system. Alvarez et al. (2002) documented the complex structural patterns occurring in un-baffled laminar stirred tanks and the mixing mechanisms responsible for chaotic mixing on them. More recently, Arratia et al. (2004) explored mixing and flow patterns in continuous stirred tanks operated in laminar regimes. In this communication, we expand those observations to the context of typical bioreactor geometries, using simple visualization techniques to demonstrate the most common mixing pathologies (deviations from the assumption of a perfectly mixed environment) observable in laminar and transitional stirred tanks. Although other impeller configurations have been documented as more efficient for laminar flow regimes in stirred tanks, here we mainly use radial impellers in order to adhere to the most

widely used lab-scale fermentor configuration: a multipleRushton impeller stirred tank system. In the absence of a specific design for a particular application, this would be the default reactor configuration to use. The pathological scenarios described here are still observable when axial impellers are used, since it has been established (Alvarez et al., 2002) that even axial impeller geometries behave as radial when operated at low Re numbers.

2. Materials and methods 2.1. Experimental systems For all experiments reported here, glycerin or glycerin– water solutions have been used as a working fluid. The range of viscosities achieved with these mixtures goes from 500 to 950 cp. Different tank geometries have been used. Most experiments were run in two or three Rushton impeller systems, typical bioreactor geometries. Both Acrylic stirred tank models and actual fermentors (5 and 14 L fermentors from New Brunswick Scientific, Edison N.J.) were used. 2.2. Visualization techniques Throughout the study, different experimental techniques were use to reveal mixing pathologies. Acid–base visualization experiments were performed on the described stirred tank systems. For this type of visualization, a pH indicator (Bromothymol blue) was added to the working fluid (glycerin–water mixtures). The indicator changes from yellow (below pH=6.5) to blue (above 8.2). Injection of NaOH 1 M or HCl 1 M allows for the observation of the evolution of reactive mixing patterns in 3D, and the presence of segregated areas in the flow domain (see Lamberto, 1997). In addition, 3-D UV visualizations were conducted. In an otherwise dark room, the use of a UV lamp allows for the observation of 3-D dispersion patterns of a fluorescent tracer (fluorescein-glycerin solution) injected into the stirred tank systems. In a third kind of visualization experiment, mixing structure information in a particular plane was gathered using 2-D LIF (Laser Induced Fluorescent) techniques. In a dark room, a laser sheet was directed to a particular plane of a transparent stirred tank. When a fluorescent tracer (rhodamine or fluorescein) is injected, the mixing patterns in the illuminated plane can be captured by a photographic camera located perpendicularly with respect to the laser plane. The same laser illumination principle was applied to gather information of particle concentration fields in a particular plane in tanks seeded with PIV (Particle Image Velocimetry) particles (10 m silver-coated particles). Illuminated particles reflect laser light. Thus, areas with higher particle density appear brighter in photographic images taken at high speed.

M.M. Alvarez et al. / Chemical Engineering Science 60 (2005) 2449 – 2457

2451

Fig. 1. Heterogeneity in typical stirred tank bioreactor geometries: (a) coexistence of chaotic and regular regions in a 3-impeller system operated at Re = 50 as revealed by an acid–base experiment, (b) regular toroidal region in a two-impeller fermentor.

3. Results and discussion 3.1. Segregation in laminar bioreactors Fig. 1a demonstrates the heterogeneous flow and mixing environment in a typical multiple impeller bioreactor geometry. In this experiment, a pH indicator has been added to the tank content, such that acidic regions will display a yellow color, and basic regions will appear in blue. Several acid injections have been applied in the originally basic fluid agitated by a three-Rushton impeller system. Therefore, the yellow areas denote zones in which, even after more than 10 min of agitation, the neutralization reaction has still not taken place. Such areas are located above and below the impeller planes. Fig. 1b shows an experiment that demonstrates the prevalence and stability of the segregated regular toroidal regions in baffled stirred tank bioreactors. These flow structures were first reported by Lamberto et al. (1996) for unbaffled systems. Fluorescent dye has been carefully injected in one of the toroidal segregated regions of this system, the one located above the lower impeller in a two-Rushton stirred tank bioreactor. The experiment takes place under UV-light in an actual 5 L baffled commercial fermenter operated at Re < 100. Fifteen minutes after the injection, the well-defined segregated region still prevails. Inside it, regular behavior exists, and complete segregation (other than by diffusion) prohibits mixing with material from the chaotic region in practical time scales. In the situation depicted here, any type of mixing time estimation lacks any realistic value, since the tank will only achieve homogeneity eventually, by the much slower action of diffusion. From the previously presented analysis, it becomes clear that a laminar stirred tank bioreactor is not a well-mixed environment, as we have insisted in the modeling and

design approaches that we have used for decades. In the next few pages, we address the structural complexity of such heterogeneity.

3.2. Compartmentalization Fig. 2a reveals another common mixing pathology of laminar systems: compartmentalization. A common laboratory and industrial practice is to control pH in bioreactors by acid or base injection at the top of the tank, or even by dripping them above the liquid surface. In this experiment, base is added by injection at the top section of a three-impeller stirred tank system. Two minutes after injection (Re = 100), only the upper portion of the tank has become basic, while the rest of the flow domain has not yet felt the basic pulse. A well-established separation plane demarks the sharp transition of pH zones. In practical situations, this observation questions the convenience of pH control by the currently conventional strategy of injection in a single location at the top tank compartment. Fig. 3 shows the evolution of pH vs time (as reported by an electrode located 2.5 cm below the second impeller mid-plane) in an experiment in which base or acid is dripped at the surface of a three-impeller system. Images of the status of the tank at different moments along the pH progression are included (see Fig. 3a–d). Bromothymol blue has been added into the tank, such that acidic regions will appear yellow, and basic regions above 8.2 will appear blue. During the first 16 min of the experiment, base is added to reach the pH set-point value of 8.2. The reader will observe that the lecture reported by the electrode has no correlation with the actual pH of upper regions of the tank. Twelve minutes after injection, the electrode still reports an acidic lecture (Fig. 3b), while the upper region of the tank has clearly reached the set-point. The vertical dotted line indicates the moment at which acid addition starts. From this point in time, there is still a delay of 2 min before the

2452

M.M. Alvarez et al. / Chemical Engineering Science 60 (2005) 2449 – 2457

Fig. 2. Compartmentalization in multiple impeller bioreactors (Re = 50): (a) acid–base experiment showing compartmentalization after a basic injection at the top of a three-impeller fermentor, (b) 2-D laser visualization of mixing patterns revealing separation planes at and inbetween impellers.

9.5 acid injection

9

(c)

8.5

(a)

8

Basic injection

(b)

pH treshold for indicator

7.5

(d)

7 0

5

10

15

20

25

Fig. 3. Progression of pH vs time for an acid–base experiment performed in a compartmentalized laminar stirred tank system. Images of the tank at different times are presented. The dotted line indicates time for acid injection.

electrode begins to respond to the acidic injection. In point (d), the electrode lecture is 8.2 again (threshold value) when the upper part of the tank became acidic several minutes before. In an actual scenario, in which acid or base is added continuously to a bioreactor to maintain a pH set-point, we can clearly see that the position of the electrode will

determine the pH reading. This measure is, as observed here, not necessarily representative of an average realistic cell environment. Fig. 2b presents results from an experiment in which a fluorescent dye is injected in a three-impeller system with no baffles. A laser sheet is projected toward the transparent tank, revealing the mixing patterns in the 2-D illuminated plane. A remarkable heterogeneity in mixing conditions through the tank is revealed by the fluorescent tracer. Compartmentalization is clearly observed. Six flow cells are delimited by impeller planes or curves created by impeller interactions. In each flow cell, regular areas are also observed. This bioreactor, which we will normally conceive and model as a homogeneous well-mixed environment, behaves as a series of non-ideal stirred tanks connected in a non-intuitive fashion. Fig. 4 elaborates on the dynamics of the process of distribution of a tracer throughout a three-impeller 14 L stirred bioreactor. In this particular experiment, a fluorescent tracer was injected at the liquid surface, near the impeller shaft. The dye distributes descending near the shaft in a spiral fashion to first get distributed in the top compartment (a process that takes in the order of 2 min for Re < 200). Later, the dye reaches and invades the chaotic region of the second compartment. The progressive invasion of successive compartments is severely limited at low Re, even in bladed systems such as the one used in this experiment. The reader will note that neither the dye coverage complete is nor is the concentration field homogeneous within each cell flow region.

M.M. Alvarez et al. / Chemical Engineering Science 60 (2005) 2449 – 2457

2453

Fig. 4. Dispersion patterns of fluorescent dye injected at the liquid surface in a 14 L compartmentalized 3-impeller stirred tank bioreactor.

3.3. Local effects of geometrical features: shaft, impellers and baffles Figs. 5 and 6 illustrate different aspects of the complicated nature of the process of dispersion of a tracer pulse within a single flow compartment. In Fig. 5, the time progression of a fluorescent dye injection experiment is presented. A stream of dye (initial radius of 4 mm) is injected at 4.0 cm from the shaft, above the top impeller of a dual Rushton fermentor system. The injected dye forms a filament that wraps around the shaft to reach the impeller plane (see Fig. 5a; see also Fig. 8). The impeller (in this case a radial impeller) stretches the filament by the periodic action of the passing of its blades, and projects it to the tank walls in a spiral fashion (see Fig. 6a and b). Alvarez et al. (2002) and Alvarez-Hernández et al. (2002) demonstrated that indeed this periodic perturbation of the otherwise regular manifold of the flow is the mechanism that triggers chaos in a laminar stirred tank. At Re < 100 (the situation depicted in the series of snapshots in Fig. 5), the stretched filament “totally respects” and aligns to the separation plane originated by the impeller rotation, and the stream of fluorescent dye is redirected exclusively upwards, suffering not significant disturbance by its encounter with baffles. Each impeller plane (and in fact each one of the separation planes observed in Fig. 2b) acts as a barrier to mass transfer (substrate, oxygen, etc.). These transport barriers are more or less “permeable” depending on the Re number (see Alvarez et al., 2002). At higher Re values (Re > 250), the effect of the baffles is more evident (see Fig. 6c and d). The filament suffers both upwards and downwards redirection and the separation planes are therefore less of a rigid barrier for transport (becoming unsteady, but still present in turbulent flows).

Later in the dispersion process (see Fig. 5c, and d) the stretched filament will evolve into a convoluted sheet that wraps around the regular regions depicted in Fig. 1b, without intruding them.

3.4. Flow short-cuts In a chaotic system, particle trajectories strongly depend on initial conditions. This fact, translated to the context of a stirred tank bioreactor, implies that an adequate selection of the point for substrate, acid, or base addition might be an important consideration for mixing efficiency. Chaotic systems, and laminar stirred tanks are not the exception, have a defined skeleton (unsteady manifold) that directs the particle paths (Muzzio et al., 2000). As observed before (see Figs. 2, 5 and 6) this underlying structure is defined by geometry (presence of baffles and impellers, distance between impellers, etc.) and operational conditions (agitation speed). Fig. 7 illustrates the presence of a “fluid funnel” or flow shortcut that directly connects regions from the tank surface with the upper impeller in a 3-Rushton system operated at Re = 250. In this experiment, acid was continuously dispensed by dripping at the liquid surface (in this case a basic water–glycerin solution saturated with NaCl). The path of the acid stream is marked by the precipitation of sodium chloride. The geometry of this fluid shortcut to the impeller is highly stable. It is also not intuitive, indeed different from the most commonly observed ribbon-like structures portrayed in Fig. 8 (see also Figs. 5 and 6), since it does not wrap around the impeller shaft. Bypassing in stirred tanks is not a new observation. For example, Arratia et al. (2004) report similar flow behavior in stirred tanks operated continuously. The implications of the existence of these shortcuts in

2454

M.M. Alvarez et al. / Chemical Engineering Science 60 (2005) 2449 – 2457

Fig. 5. Dispersion patterns within a single compartment of a fluorescent dye injected in a compartmentalized 5 L bioreactor.

Fig. 6. Mechanisms of mixing in a bioreactor: (a) the periodic passing of the impeller blades stretches fluid pockets into filaments that advance towards the tank walls, (b) close-up of spiral filament patterns, (c) effect of baffles deflecting and redirecting streams of fluid, (d) at Re = 250; the separation planes are “more permeable” to mass transfer in the vicinity of baffles.

M.M. Alvarez et al. / Chemical Engineering Science 60 (2005) 2449 – 2457

2455

Fig. 7. Fluid shortcut in a stirred tank bioreactor as observed in a salt precipitation experiment in NaCl-saturated water–glycerin solution.

bioreactors cannot be dismissed. A bad selection of an injection location in a laminar stirred tank can clearly favor the presence of zones highly acidic, basic, or highly concentrated in terms of a particular substrate. 3.5. Cell focusing Fig. 9 illustrates another source of non-ideality in bioreactor performance, particle focusing. Cell bioreactors are multiple phase systems with at least one solid phase, precisely the cell biomass. Here we present results of an experiment in which a single impeller tank, filled with glycerin and equipped with an axial impeller, is seeded with solid silver-coated particles (under 10 m). Due to their density (approximately 1.3 g/ml) and size (10 m in diameter), the particles are expected to closely follow the flow streamlines and experience velocities similarly matching those of fluid particles. Indeed, the particles used here are recommended for PIV studies. When the seeded tank is illuminated by a laser plane, the particle density field is revealed. However, the particles disperse in a highly non-homogeneous way. The reader will notice highly illuminated areas (richer in particle density) and areas of much lower particle content. Particles seeded within regular regions do not leave them even after several hours. Those seeded in the chaotic regions apparently exhibit a higher frequency of visitation to the high shear areas, as revealed by the higher intensity of laser light reflection in areas previously associated with high shear stresses (see for example Zalc et al., 2001). Cells will certainly not be exempt of such dynamics. The implications of particle

Fig. 8. Injection of fluorescent dye at the liquid surface in laminar stirred tank bioreactors. Ribbon-like structures develop around the impeller shaft: (a) frontal view, (b) top view.

focusing are relevant to animal cell cultures, since according to the observations presented here, cells will frequently visit flow areas where we do not want them to go: those where a higher mechanical stress is felt. Particle focusing has been widely reported in different contexts. For example, Leighton and Acrivos (1987) and Phillips et al. (1992) refer to the migration of particles flowing in channels, from areas of high shear to areas of low shear. In such systems, the migration is perpendicular to the stream lines of the flow. Here, however, we observe the opposite effect, higher particle density in high shear regions. The particle dynamics observed in stirred tanks is more complex than in channel flows, due to the nature of the flow field. Shinbrot et al. (2001) demonstrated numerically and experimentally that even very small particles can be forced to concentrate in certain regions in model-stirred tank systems, when they are subject to periodic forcing or even to a single perturbation event. 4. Conclusions In this communication, we studied mixing in bio-reactor geometries with a focus on an increasingly relevant industrial

2456

M.M. Alvarez et al. / Chemical Engineering Science 60 (2005) 2449 – 2457

the injection location for nutrients, oxygen, substrates, and acid or base solutions for pH control. Cell focusing, and the fact that this phenomena concentrates cell in high shear areas, has severe implications for the design and operation of cell culture systems for shear-sensitive cells. The mixing pathologies studied here are controlled by the geometry of the system. In Alvarez et al. (2002) a study of mixing behaviors in eccentric laminar stirred tanks is presented. We are currently carrying out a more comprehensive characterization of eccentric bioreactors as an alternative to alleviate some of the pathologies discussed here.

Acknowledgements Some of the figures presented correspond to experiments conducted at Rutgers University at Dr. Fernando Muzzio’s lab. We thank him for his support. We thank Dr. Troy Shinbrot, and Dr. Paulo Arratia at Rutgers University for their feedback on this manuscript. References

Fig. 9. Focusing of solid particles (emulating cells) in a laminar stirred tank. Areas of higher illumination denote higher particle concentration.

scenario: laminar and transitional stirred tanks. We had presented experimental evidence of mixing pathologies observable in typical bio-reactor geometries in the laminar regime. Using visualization techniques, we illustrate the presence of segregated regular regions below and above impellers, the existence of flow compartmentalization in multiple impeller systems, the effect of the presence of deflectors in bioreactors operated at different Re values, and the focalization of solid particles to high shear regions. All of these pathologies and flow features have profound implications on the mixing performance of stirred tank bioreactors. The presence of segregated regions itself implies a deviation from the commonly used assumption of a perfectly mixed environment that is typically used to design and model bioreactors. Compartmentalization in multiple impeller systems imposes severe mass transfer limitations, particularly when accurate control of pH, dissolved oxygen, or other parameters is attempted by a single measurement point. The existence of fluid shortcuts has implications for the selection of

Alvarez-Hernández, M.M., Shinbrot, T., Muzzio, F.J., 2002. Practical chaotic mixing. Chemical Engineering Science 57, 3749–3753. Alvarez, M.M., Muzzio, F.J., Cerbelli, S., Adrover, A., Giona, M., 1998. Self-similar spatiotemporal structure of intermaterial boundaries in chaotic flows. Physical Review Letters 81 (16), 3395–3398. Alvarez, M.M., 2000. Using spatio-temporal asymmetry to enhance mixing in chaotic flows: from maps to stirred tanks. Ph. D. Dissertation, The State University of New Jersey, Rutgers. Alvarez, M.M., Zalc, J.M., Shinbrot, T., Arratia, P.E., Muzzio, F.J., 2002. Mechanisms of mixing and creation of structure in laminar stirred tanks. A.I.Ch.E Journal 48 (10), 2135:2148. Arratia, P.E., Lacombe, J., Shinbrot, T., Muzzio, F., 2004. Segregated regions in continuous laminar stirred tank reactors. Chemical Engineering Science 59 (7), 1481–1490. Aloi, L.E., Cherry, R.S., 1994. Intracellular calcium response of Sf-9 insect cells exposed to intense fluid forces. Journal of Biotechnology 33 (1), 21–31. Ariff, A.B., Sabih, B.A., Azudin, M.N., Kennedy, J.F., 1997. Effect of mixing on enzymatic liquefaction of sago starch. Carbohydrate Polymers (33), 101–108. Berson, R.E., Pieczynski, W.J., Svihla, C.K., Hanley, T.R., 2002. Enhanced mixing and mass transfer in a recirculation loop results in high cell densities in a roller bottle reactor. Biotechnology Progress 18 (1), 72–77. Enfors, S.-O., Jahic, M., Rozkov, A., Xu, B., Hecker, M., Jurgen, B., Kruger, E., Schweder, T., Hamer, G., O’Beirne, D., NoisommitRizzi, N., Reuss, M., Boone, L., Hewitt, C., McFarlane, C., Nienow, A., Kovacs, T., Tragardh, C., Fuchs, L., Revstedt, J., Friberg, P.C., Hjertager, B., Blomsten, G., Skogman, H., Hjort, S., Hoeks, F., Lin, H.Y., 2001. Physiological responses to mixing in large scale bioreactors. Journal of Biotechnology 85, 175–185. Elias, C.B., Rajiv, B.D., Milind, S.P., Jyeshtharaj, B.J., Raghunath, A.M., 1995. Turbulent shear stress-effect on mammalian cell culture and measurement using laser Doppler anemometer. Chemical Engineering Science 50 (15), 2431–2440. Gogate, P.R., Beenackers, A.A.C.M., Prandit, A.B., 2000. Multiple impeller systems with a special emphasis on bioreactors: a critical review. Biochemical Engineering Journal 6, 109–144.

M.M. Alvarez et al. / Chemical Engineering Science 60 (2005) 2449 – 2457 Gregoriades, N., Clay, J., Ma, N., Koelling, K., Chalmers, J.J., 2000. Cell damage of microcarrier cultures as a function of local energy dissipation created by a rapid extensional flow. Biotechnology and Bioengineering 69 (2), 171–182. Jian Li, Z., Shukla, V., Wenger, K.S., Fordyce, A.P., Petersen, A.G., Marten, M.R., 2002. Effects of increased impeller power in a productionscale aspergillus oryzae fermentation. Biotechnology Progress 18, 437–444. Lamberto, D.J., 1997. Enhancing laminar mixing in stirred tank reactors using dynamic flow perturbations. Ph. D. Thesis Dissertation, The State University of New Jersey, Rutgers. Lamberto, D.L., Alvarez, M.M., Muzzio, F.J., 1999. Experimental and computational investigation of the laminar flow structure in a stirred tank. Chemical Engineering Science 54 (7), 919–942. Lamberto, D.J., Muzzio, F.J., Swanson, P.D., Tonkovich, A.L., 1996. Using time-dependent RPM to enhance mixing in stirred vessels. Chemical Engineering Science 51 (5), 733–741. Lamberto, D.J., Alvarez, M.M., Muzzio, F.J., 2001. Computational analysis of regular and chaotic mixing in a stirred tank reactor. Chemical Engineering Science 56 (16), 4887–4899. Leighton, D., Acrivos, A., 1987. The shear-induced migration of particles in concentrated suspensions. Journal of Fluid Mechanics 181, 415. Muzzio, F.J., Alvarez, M.M., Cerbelli, S., Giona, M., Adrover, A., 2000. The intermaterial area density and striation thickness distribution generated by time- and spatially-periodic 2D chaotic flows. Chemical Engineering Science 55 (8), 1497–1508.

2457

Nikolai, T.J., Hu, W.S., 1992. Cultivation of mammalian cells on macroporous microcarriers. Enzyme and Microbial Technology 14 (3), 203–208. Peña, C., Galindo, E., Díaz, M., 2002. Effectiveness factor in biological external convection: study in high viscosity systems. Journal of Biotechnology 95 (1), 1–12. Phillips, R.J., Armstrong, R.C., Brown, R.A., 1992. A constitutive equation for concentrated suspensions in two-dimensional channel flow, part I, monodisperse systems. Journal of Fluid Mechanics 363, 25. Shinbrot, T., Alvarez, M.M., Zalc, J.M., Muzzio, F.J., 2001. Attraction of minute particles to invariant regions of volume preserving flows by transients. Physical Review Letters 86 (7), 1207–1210. Unger, D.R., Muzzio, F.J., Aunins, J.G., Singhvi, R., 2000. Computational and experimental investigation of flow and fluid mixing in the roller bottle bioreactor. Biotechnology and Bioengineering 70 (2), 117–130. Vrábel, P., van der Lans, G.J.R., van der Schot, F.N., Luyven, K.Ch.A.M., Xu, B., Enfors, S., 2001. CMA: integration of fluid dynamics and microbial kinetics in modeling of large-scale fermentations. Chemical Engineering Journal 84, 463–474. Weiss, S., John, G.T., Klimant, I., Heinzle, E., 2002. Modeling of mixing in 96-well micro plates observed with fluorescence indicators. Biotechnology Progress 18 (4), 821–830. Zalc, J.M., Alvarez, M.M., Shinbrot, T., Arik, B.E., Muzzio, F.J., 2001. Extensive validation of computed laminar flow fields and mixing dynamics in a stirred tank equipped with three rushton turbines. A.I.Ch.E Journal 47 (10), 2144–2154.