Experimental studies of hydrodynamics and regime transition in bubble columns.

Thet, May Khin ; Wang, Chi-Hwa ; Tan, Reginald B.H. 等

The detailed flow structures in bubble columns with and without internal draught tube have been investigated using the PIV technique. The onsets of transition due to vortex formation and different flow patterns with and without draught tube have been studied using the drift-flux model and the experimentally measured Reynolds stresses. The role of solid particles and liquid viscosity, as well as bubbling orifice configuration on the flow patterns and regime transition has also been studied and discussed.

On a etudie par la technique PIV les structures d'ecoulement detaillees dans des colonnes a bulles avec et sans tube d'aspiration interne. L'apparition de la transition due a la formation de vortex et a differents profils d'ecoulement avec et sans tube d'aspiration a ete etudiee a l'aide du modele de flux d'aspiration et des contraintes de Reynolds mesurees experimentalement. On a egalement examine le role des particules solides et de la viscosite du liquide ainsi que de la configuration des orifices de bullage sur les profils d'ecoulement et la transition de regime.

Keywords: transition regime, PIV, Reynolds stresses, liquid phase viscosity, solid particle

Bubble columns are widely used for their simple construction and economically favourable operation. Gas is usually sparged upwards through a perforated plate or a series of nozzles into a continuous liquid phase. The liquid phase may contain solids such as catalyst, reagent or biomass, etc., which are distributed in the column. Modified bubble columns including external loop airlift reactors, internal loop airlift reactors are sometimes used to increase liquid circulation and gas residence times. The hydrodynamics of the bubble column generally depends on the gas-liquid or gas-liquid-solid physical properties as well as column geometry and internal configuration.

This paper presents experimental studies of hydrodynamics and regime transition in bubble columns. The liquid velocity distribution in terms of flow pattern and Reynolds stresses in bubble column with and without draught tube using multiple orifices are addressed and liquid phase viscosity and solid particle effects on transition regime described. Finally, the measurement of liquid flow patterns of single to quadruple bubbling orifices using the particle image velocimetry (PIV) are presented and discussed.

HYDRODYNAMICS IN BUBBLE COLUMNS AND AIRLIFT REACTORS

Knowledge of hydrodynamic behaviour in a bubble column is very important for prediction of the design parameters such as heat and mass transfer coefficients, critical gas flow rates, etc. The hydrodynamic behaviour of bubble columns consists of the macroscopic or large-scale phenomena and the microscopic or local phenomena (Deckwer, 1992). The macroscopic flow phenomena include flow regimes, gas hold-up and the gross liquid circulation. From the overall gas hold-up and superficial gas velocity, three different flow regimes namely homogeneous, transition and heterogeneous regimes are observed in bubble column.

Liquid phase behaviour caused by gas distribution and regime transition has been studied in gas-liquid systems using Laser-Doppler Anemometry (LDA) (Mudde et al., 1997a and b; Vial et al., 2001b; Olmos et al., 2003) and in three phase systems by applying PIV (Chen and Fan, 1992; Chen et al., 1994; Lin et al., 1996; Tzeng et al., 1993). Vial et al. (2001b) and Olmos et al. (2003) reported that transition was characterized by local liquid recirculation near the wall where the mean liquid velocity is negative (downwards). High positive values were observed at the centre of the column. In addition, the magnitudes of horizontal normal stresses were found to be twice that of vertical normal stresses. Another non-intrusive method that works independently from the void fraction is the Computer-Aided Radioactive Particle Tracking (CARPT) (Chen et al., 1999). It is a single point method that tracks a single radioactive particle over a long period of time and can provide the radial profiles of the Reynolds shear stress, and the axial and radial normal stresses. Yang et al. (1993) found that it was difficult to measure the Reynolds shear stress near the column centre and close to the column wall. Interestingly, they reported that the axial and normal stresses were considerably greater than the Reynolds shear stress. The present study proposes to examine the full field measurement of Reynolds shear stress and normal stresses at the wall region in 2D bubble column with and without draught tube through the use of a PIV system.

The geometry of the gas distributor is known to have a significant effect on the flow regime. Furthermore, the flow structure in the absence or presence of draught tube system is extended in this present work where the liquid velocities are generally higher than in bubble columns (Wild et al., 2003).

Van Benthum et al. (1999) identified three flow regimes in internal draught tube bubble columns, namely, (regime I) no air bubbles (when liquid velocity in the downcomer is lower than slip velocity of bubbles in the liquid) (regime II) bubbles remain stationary (when liquid velocity in the downcomer is equal to the slip velocity of bubbles in the liquid) and (regime III) bubbles flow downwards and into the riser (when liquid velocity in the downcomer is higher than slip velocity of bubbles in the liquid). Their study shows that transition from regime II to III occurred at a gas hold-up of 10% to 12%. Our study aims to provide further information on the Reynolds stresses at these regimes. Wild et al. (2003) reported that in airlift reactors, no maximum of the gas hold-up is observed, even with efficient gas distribution systems; this is due to the relation between the gas and the liquid velocity, which have opposite effects on the gas hold-up. An increasing gas velocity leads to an increase of the overall liquid velocity, which reduces in turn the increase of the gas hold-up. Deckwer (1992) suggests that microscopic flow phenomena are more likely to be associated with gas phase phenomena such as bubble wake interaction with the continuous phase, bubble coalescence, and bubble break up.

Thus, in any reactor design or modelling of bubble columns, both the macroscopic and microscopic flow regimes have to be taken into account. There are many influencing factors that affect the flow phenomena such as column dimensions, pressure effect, particle concentrations, liquid phase properties, orifice plate configurations, as reported in several studies on homogeneous to heterogeneous regime transition (Ruzicka et al., 2001a; Vial et al., 2001a). Firstly, the effect of solid loading in the bubble column on gas hold-up has been studied both experimentally and by CFD simulation (Jamialahmadi and Muller-Steinhagen, 1993 and ref: there in; Saxena and Chen, 1994; Garcia-Ochoa et al., 1997; Gentile et al., 2003). Increasing solid concentration, particle size and solid-liquid density difference results in a decrease in gas hold-up. This trend was reported to be reversed if the particle size is below 10 [micro]m. However, there have been very limited studies on the effect of solids on transition gas velocity in bubble columns. It should be mentioned that although addition of catalyst particles enhanced the regime transition as reported by Krishna et al. (1999), no significant effect was found for dilute particle concentration of 1 to 10% using 0.5 mm glass beads and 1.5 mm acetate beads (Chen et al., 1994).

In addition, the particle effect on macroscopic flow structure was studied by Tzeng et al. (1993) and Lin et al. (1996). The vortex size, wavelength, frequency, and vortex descending velocity were strongly influenced by increasing bubble coalescence rate. Due to the limitations of experimental measurements in three phase systems, there is still a lack of detailed understanding of the effect of particles within the transition regime, such as the particle size and density effect on the flow structure and the liquid phase properties in different operating conditions.

Another influencing parameter on microscopic flow structures is the liquid phase viscosity and its effect on the gas hold-up, as studied by Krishna et al. (1997) and (2000). An increase in liquid viscosity lowered the bubble break up rates as reported by Saxena and Chen (1994). They found that [[epsilon].sub.g] increases with viscosity in the range 1 to 3 cp and then decreases sharply till about 11 cp and then decreases slowly up to about 39 cp. The changes were relatively sharp at higher gas velocities. This was explained on the basis of the bubble coalescence phenomena on liquid-phase viscosity. But the viscosity effect on the transition regime in bubble column is not yet well understood. Zahradnik et al. (1997) explained that the effect of viscosity of saccharose solution on the transition regime is found to be significant because an increase of drag forces promotes bubble coalescence in the distributor region. Reduction in the surface tension of the liquid led to a decrease in bubble stability and thus to smaller bubbles.

Several authors have studied the influence of different aspect ratios and distributor configuration on the flow structure (Tzeng et al., 1993; Becker et al., 1994; Borchers et al., 1999; Becker et al., 1999; Deen et al., 2000). In a shallow column when the air is injected from the central part of the column, the "gulf stream" phenomena was observed, in which a pair of symmetrical liquid circulation cells were formed. Becker et al. (1994) measured the flow pattern with LDA in the case of a decentralized gas inlet for a number of different aspect ratios and gas flow rates. The gas inlet was placed at the left of column. They found that the lower part of the bubble plume was stationary at low gas flow rates and directed to the left wall, under the influence of a large liquid vortex on the right-hand side. The upper part of the bubble plume was meandering in a quasi-periodic way. For high gas rates the meandering behaviour was not observed. Experiments on the flow of a bubble plume in a 3D bubble column were performed by Deen et al. (2000).

Therefore, we aim to study the effect of different number of orifices and their relative positions on the column hydrodynamics using PIV technique as part of this work. In addition, more work on the investigation of whole field velocity data with high spatial and temporal resolution is necessary to obtain new experimental information about the flow fields of interest.

EXPERIMENTAL SET-UP

PIV (TSI Company, Shoreview, Minnesota, U.S.) was used to observe the liquid phase flow structure. The liquid phase images were captured by PowerView[TM] 4M 2K ??2K camera that is connected to LaserPulse[TM] Computer Controlled Synchronizer.

A colour filter was used to increase the image contrast between gas and liquid phases. Thus the scattered light from the 20-40 [micro]m Rhodamine-B dye particles in the liquid phase was captured, but direct laser light was filtered out. The measurement was done under atmospheric condition. 30 image pairs were recorded near the wall when the gas was introduced at critical value of superficial gas velocity for transition in each bubble column with and without draught tube. Instantaneous velocity vectors were obtained after calibration and image processing.

Figure 1 represents the schematic diagram of the experimental apparatus. The bubble column had a height of 1.2 m, 0.15 m by 0.15 m square area and was open to atmosphere at the top. Different perforated orifice plates of 3 mm thick Plexiglas[R] could be installed (Figure 2). The internal draught tube had a height of 0.455 m and a clearance of 0.1 m from the bottom of the column. The annular space was 0.02 m wide. Water is filled in the column up to 0.55 m height so that the liquid can circulate from the top to the annular region until bottom of the column. The combinations of operating conditions are listed in Table 1. For the partial aeration system Figure 2 three different orifice conditions were applied, namely one orifice, (1) two orifices (1, 2) and four orifices (1, 2, 3, 4). Three test sections near the bottom, y = 0.03-0.11 m, middle, y = 0.11-0.21 m, and top of the column, y = 0.21-0.26 m were investigated. PIV measurements were taken at three X-Y planes at Z = 0.065 m.

[FIGURES 1-2 OMITTED]

Gas hold-up can be calculated as [[epsilon].sub.g] H' - H / H' where H' is aerated liquid height and H is static liquid height. Filtered tap water ([[rho].sub.L] = 1000kg/[m.sup.3], [sigma] = 0.07N/m) was used and the experiments were done at room temperature and at a gas supply (compressed air) pressure of 60 psi to obtain gas velocities up to 0.07 m/s. For the three phase systems, low density polycarbonate particles and high density glass beads were used (Table 2, Hartman et al., 1994). Glucose, the viscosity modifier, was well mixed with deionized water before adding to the column. The column was then operated at a high gas flow rate for approximately 20 min to ensure that the system has reached a steady state.

RESULTS AND DISCUSSION

Characterization of Flow Regime With and Without Draught Tube Figure 3a represents the characterization of different flow regime transition points with gas voidage vs. superficial gas velocity in the column with and without draught tube (DT). Also the flow regime transition points can be determined by using Wallis' drift-flux model (Wallis, 1969) as shown in Figure 3b. From Figure 3a, it is clear that the gas hold-up in the case without DT has a local maximum value occurring at a certain point of superficial gas velocity. The value of q corresponding to this point is denoted as [q.sub.max] the maximum gas superficial velocity corresponding to maximum gas voidage in the transition regime.

[FIGURE 3 OMITTED]

This physical quantity will be further discussed in this paper. In the column with DT, no local maximum voidage was found throughout the trend (Figure 3a). Changes of slope in Figure 3b suggest that the regime changes from regime I to regime III (regime descriptions according to Van Benthum et al., 1999) as noted earlier. The gas hold-up at the transition is found to be approximately 13% by visual observation of the superficial gas velocity corresponding to the onset of bubbles in the downcomer.

Time Averaged Flow Structures

In Figure 4a, the profiles of the time averaged flow field of the liquid without DT were determined by averaging the information of 30 dual frames PIV vector fields per profile with a height of 0.03-0.11m at an inlet superficial gas velocity of q = 0.04 m/s. It is observed that the existence of a small pair of liquid vortices in the time averaged flow field is caused by the liquid circulation structure that consists of bubbly upward flow in the column core region and bubble-free liquid downward flow region near the side wall (Figure 4a). At low gas velocities, gross liquid circulation is not observed (homogeneous regime). Deckwer (1992) reported that a radial cross-exchange of .uid elements leads to axial circulation and gives rise to high radial intermixing. In the heterogeneous regime, liquid down flow was very pronounced. At gas velocities intermediate between those causing homogeneous and heterogeneous regimes, the trend of bubble streams appears to oscillate in the column core and to show a tendency towards recirculation. Simultaneously, vortices appear near the sidewall caused by the momentum transfer to central region from wall region. These vortices have been termed descending vortices (Tzeng et al., 1993).

[FIGURE 4 OMITTED]

Figure 4b shows the time averaged flow field at q = 0.04 m/s in the column with DT. The results show a predominant downflow near the wall region, showing the effectiveness of the DT in facilitating liquid circulation. The area enclosed by the dashed line is the region where the liquid changes direction prior to entering the riser. A similar flow pattern was observed for all values of gas flow rate. No wall-region vortex was observed in the case of columns with DT.

Interpretation of Wall Region Flow

The downward liquid flow region near the sidewall represents the descending flow region (Tzeng et al., 1993). There appears to be no bubbles occupying this region at very low gas velocities. As the gas velocities increase, small bubbles were observed in the descending liquid as evidenced by small scale liquid recirculation.

In the case of the fluid flow field without DT, a vortical flow region is observed close to the descending flow region (Figure 4a). Vortices seem to be formed at the confluence of descending liquid flow and the upward bubble plume region. The sizes of vortices increased in size and instability with increasing gas velocities. The development of vortical structure near the wall is disappeared with the presence of DT and pure descending region is only observed at wall region (Figure 4b).

A third region may be discerned in Figure 4a that is the central descending region located between the two vortical flow regions. This central region is characterized by complex flow associated with bubble wakes or liquid slip flow between bubble streams in the core region. The size of this region depends on the geometry of the column. In this region, the liquid velocity is lower than that in the sidewall region.

Reynolds Stresses

Figure 5a shows the typical Reynolds stresses profiles at various radial positions in a bubble column with no DT. The longitudinal normal stress profile is flat in the middle with two distinct maxima close to each sidewall. The lateral normal stresses on the other hand, display a flat maximum in the centre and minima at the walls. Our earlier observation of vortices migrating towards the core region could explain the result that vertical normal stresses are higher at the walls. The Reynolds shear stresses are generally lower than the normal stresses. Similar profiles for Reynolds stresses have been reported by Olmos et al. (2003) in a two-dimensional bubble column. Mudde et al. (1997a) reported that "the shear stress had its greatest absolute value in the region where the axial velocity gradient was highest and had a value of zero where the averaged axial velocity was maximum, that is, the column centre, thereby suggesting a Boussinesq type of correlation between the averaged axial velocity and the shear stress."

[FIGURE 5 OMITTED]

In Figure 5b, the typical Reynolds stresses profiles are shown for a bubble column with DT since no vortices are formed, normal stresses are lower as compared with Figure 5a, and the shear stresses are virtually zero. The DT allows a regular circulation pattern of liquid upwards in the central DT and downwards through the annulus.

EFFECT OF SOLID AND LIQUID PHASE PROPERTIES ON FLOW REGIME

Effect of Solid Particles on Flow Regime Figure 6 shows the drift-flux plot as a function of gas hold-up at different concentrations of 0.5 mm diameter glass beads (C = 0 to 14% wt.). At low particle concentrations (C<1%), the data are virtually indistinguishable from those for C = 0%. The effect of solids loading on [q.sub.max], the gas superficial velocity corresponding to local maximum gas voidage in the transition regime, can be clearly seen from these results. The corresponding results with 3 mm diameter polycarbonate beads are shown in Figure 7. It was found that, at corresponding gas flow rates, the gas hold-up was slightly higher in the system with glass spheres of small particles as compared to the larger and less dense polycarbonate beads. A possible explanation is that the smaller and denser glass beads had a less pronounced effect on the effective suspension viscosity as compared with larger particles with density close to the liquid density.

[FIGURES 6-7 OMITTED]

At high solids loadings around 14% wt., the curves in Figures 6 and 7 do not show a distinct kink as is the case for lower values of particle concentration. This is due to the high interaction between bubbles and solid particles, leading to instabilities in the flow regime, and no clear value of [q.sub.max] can be obtained. Tzeng et al. (1993) reasoned that the development of vortices can be retarded at high solid hold-ups due to the disappearance of gross circulation pattern, and the interaction between solid and fluid phases become significant and results in changes in the macroscopic flow behaviour.

Saxena and Chen (1994) reported that the presence of particles changes the rheology of two- and three-phase dispersions in bubble columns. Increasing solid concentration in a liquid tended to increase the effective viscosity of the slurry. Krishna et al. (1999) added fine silica catalyst particles of 40 [micro]m to the liquid phase and found that the presence of particles promotes regime transition due to bubble coalescence.

Our results for the effect of solids on the gas superficial velocity corresponding to local maximum gas voidage in the transition regime (as characterized by [q.sub.max]) can be summarized in Figure 8, where results for [q.sub.max] are plotted for three particle types. The results show that the larger 3 mm particles of glass and polycarbonate tend to promote earlier transition, as shown by the decreasing trend of [q.sub.max], whereas increasing concentration of small particles (0.5 mm glass spheres, up to 8% by weight) tended to extend the homogeneous regime.

[FIGURE 8 OMITTED]

For these experiments, the small glass beads were of the same size as the orifice diameter and poorly suspended particles circulated near the bottom of the column could cause misdistribution of the sparged gas and possibly lower bubble rise velocity and liquid circulation velocity. This may be a reason why presence of the small glass beads tended to extend the transition regime by inhibiting the transition from homogeneous to heterogeneous flow.

On the other hand, low density polycarbonate beads were observed to be suspended throughout the liquid. The presence of these particles increased particle-bubble interactions and promoted bubble coalescence/breakage, which would result in earlier regime transition to heterogeneous flow.

It was also observed that in the experiments involving the larger glass spheres, the solids tended to settle towards the bottom of the bubble column. For solid concentration greater than about 3% wt. of 3 mm glass spheres, complete suspension could not be achieved, and the clustering of particles above the distributor resulted in a purely heterogeneous flow regime.

Effect of Liquid Viscosity on Transition Regime

Figure 9 shows the drift-flux profiles for solutions with different glucose concentrations. Glucose concentrations between 3.2 and 41 wt. % with the range of viscosities from 1.2 to 6.7 mPa.s were used (Table 3). It is found that gas hold-up generally decreases with increasing viscosity as expected.

[FIGURE 9 OMITTED]

The value of [[epsilon].sub.g] corresponding to [q.sub.max], the gas superficial velocity corresponding to local maximum gas voidage in the transition regime, decreased with increasing liquid viscosity, showing a narrowing of the homogeneous regime at high viscosities. Beyond about 4 mPa.s, the bubbling was in the heterogeneous regime for all superficial gas velocities, and no value of [q.sub.max] was measurable. Zahradnik et al. (1997) also found that the homogeneous bubbling column was suppressed at high liquid viscosities.

Figure 10 shows the values of [q.sub.max] due to the effect of solution viscosity for columns with different aspect ratios. It is clear that earlier transition was observed with increasing viscosity up to 3.5 mPa.s. At a high viscosity, shear stresses between the liquid and bubble phase are higher. Bubbles remain in the liquid for a longer time, and this in turn enhances the possibility of bubble coalescence and fluctuation of fluid flow patterns. Thus, higher viscosity promotes the transition from homogeneous towards heterogeneous flow by increasing the instability of the flow regime. The results also show that increasing the height to diameter aspect ratio caused earlier transition for the range of viscosities studied. Jamialahmadi and Muller-Steinhagen (1993) and Wilkinson et al. (1992) observed that a higher column height to diameter ratio decreases the gas hold-up due to bubble coalescence. Ruzicka et al. (2001b) and Sarra. et al. (1999) also found that the gas velocity for transition decreased with column size.

[FIGURE 10 OMITTED]

PARTIAL AERATION EFFECTS

Experiments were carried out to study the liquid flow patterns for different combinations of a small number of orifices by employing plate type 2 (Figure 2) without DT.

Single Aeration Effect

Figure 11 shows the 3D surface plot of time averaged liquid flow pattern from 30 frames for a single non-centrally placed orifice. Liquid flow induced by a single bubble stream closer to the left wall can be seen. The intensity of shading corresponds to the turbulence intensity. To the right of the rising bubble stream, the liquid velocity is stable and negative (downward flow), showing that an effective pattern of liquid circulation has been formed.

[FIGURE 11 OMITTED]

Some gas bubbles were observed to be entrained into the downward flowing liquid, and when these bubbles encounter the primary upward flowing bubble stream, small vortices are formed and these can be discerned in the contour plot.

The magnitudes of Reynolds stresses near the bottom of the column are shown in Figure 12. Clearly, the stresses only exist close to the position of the single orifice. Interestingly, the lateral normal stresses are significant, and this is reflected in the fact that the bubbles do not rise vertically but in an unsteady swirling helical path.

[FIGURE 12 OMITTED]

Double Aeration Effect

Figure 13 shows the effect of symmetrical double orifice bubbling in the column. The average velocities show two distinct rising bubble streams. The bubble plume is not stable and the two streams subsequently merge to form a single bubble stream. Two distinct peaks can also be seen in both the horizontal and vertical normal Reynolds stresses (Figure 14).

[FIGURES 13-14 OMITTED]

Quadruple Aeration Effect

Figures 15 and 16 show the data of time-averaged liquid velocities and Reynolds stresses respectively for the case of symmetrical bubbling through four orifices. Four peaks are just discernible in Figure 15, which also shows the existence of several small vortices. Figure 16 shows that the measured Reynolds stress values are rather low. These data may not be reliable because of limitations of the PIV measurements in a region of very high gas hold-up close to the multiple orifices.

[FIGURES 15-16 OMITTED]

CONCLUSIONS

In this work, the time-averaged liquid flow structures in fully aerated bubble columns with and without DT have been measured using PIV up to gas hold-up values of 19%. The liquid circulation structure at the wall region has been observed in the transition regime. The Reynolds stresses in columns without DT were larger than those in DT, reflecting the influence of vortical structure. When a DT was introduced the normal stresses were almost uniform across the column and a pure descending region was observed at wall region.

It was observed that increasing the viscosity of liquid caused an earlier transition from the homogeneous to heterogeneous regime, as measured by a decrease in [q.sub.max]. Moreover, the transition gas velocity was found to decrease with increasing aspect ratio in a glucose aqueous solution. When solid particles were added to the bubble column, the overall gas hold-up generally decreased. However, our results suggest that solid particles either promote or inhibit the transition from homogeneous to heterogeneous flow depending on the size and density of particles. Earlier transition was observed with large and high density particles. The configuration and position of the gas sparger is another factor that may affect the regime transition and the behaviour of the flow pattern.

The strong influence of the sparger configuration on the liquid flow structure and the resolution of PIV have been highlighted by the evolution of time averaged surface plots and Reynolds stresses. By varying the number and placement of bubbling orifices, it was observed that the number of vortices is low for asymmetrical aeration, and symmetrical aeration generates symmetrical vortices. The PIV technique is found to be a useful tool to characterize the number of aeration modes through the time averaged surface plot of 30 dual frames in one second, and PIV spatial resolution is applicable for up to four orifices with a frequency of 60 Hz.

ACKNOWLEDGEMENTS

This work was funded by National University of Singapore under the grant R279-000-095-112. The authors are grateful to Rensheng Deng for consultation on PIV set-up.

REFERENCES

Becker, S., A. Sokolichin and G. Eigenberger, "Gas-Liquid Flow in Bubble Columns and Loop Reactors: Part II. Comparison of Detailed Experiments and Flow Simulations," Chem. Eng. Sci. 49, 5747-5762 (1994).

Becker, S., H. De Bie and J. Sweeney, "Dynamic Flow Behaviour in Bubble Columns," Chem. Eng. Sci. 54, 4929-4935 (1999).

Borchers, O., C. Busch, A. Sokolichin and G. Eigenberger, "Applicability of the Standard k-??Turbulence Model of the Dynamic Simulation of Bubble Columns. Part II: Comparison of Detailed Experiments and Flow Simulations," Chem. Eng. Sci. 54, 5927-5935 (1999).

Chen, J. A. K., M. H. Al-Dahhan, M. P. Dudukovic, D. J. Lee and L.-S. Fan, "Comparative Hydrodynamics Study in a Bubble Column using Computer-Automated Radioactive Particle Tracking (CARPT) / Computed Tomography (CT) and Particle Image Velocimetry (PIV)," Chem. Eng. Sci 54, 2199-2207 (1999).

Chen, R. C. and L.-S. Fan, "Particle Image Velocimetry for Characterizing the Flow Structure in Three-Dimensional Gas-Liquid-Solid Fluidized Beds," Chem. Eng. Sci. 47, 3615-3622 (1992).

Chen, R. C., J. Reese and L.-S. Fan, "Flow Structure in a Three-Dimensional Bubble Column and Three-Phase Fluidized Bed," AIChE J. 40, 1093-1103 (1994).

Deckwer, W. D., "Bubble Column Reactors," John Wiley and Sons, New York, NY (1992).

Deen, N. G., B. H. Hjertager and T. Solberg, "Comparison of PIV and LDA Measurement Methods Applied to the Gas-Liquid Flow in a Bubble Column," 10th Int. Symp. on Appl. of Laser Techniques to Fluid Mech., Lisbon, Portugal (10-13 July 2000).

Garcia-Ochoa, J., R. Khalfet, S. Poncin and G. Wild, "Hydrodynamics and Mass Transfer in a Suspended Solid Bubble Column with Polydispersed High Density Particles," Chem. Eng. Sci. 52, 3827-3834 (1997).

Gentile, F., H. Oleschko, P. Veverka, V. Machon, A. Paglianti, W. Bujalski, A. W. Etchells and A. W. Nienow, "Some Effects of Particle Wettability in Agitated Solid-Gas-Liquid Systems: Gas-Liquid Mass Transfer and the Dispersion of Floating Solids," J. Chem. Eng. Japan 78, 581-587 (2003).

Hartman, M., O. Trnka and K. Svoboda, "Free Settling of Non-Spherical Particles," Ind. Eng. Chem. Res. 33, 1979-1983 (1994).

Jamialahmadi, M. and H. Muller-Steinhagen, "Gas Hold-Up in Bubble Column Reactors," in "Encyclopedia of Fluid Mechanics," N. P. Cheremisinoff, Ed., Gulf Publishing Company, Houston, 387-407 (1993).

Krishna, R., J. W. A. de Swart, J. Ellenberger, G. B. Martina and C. Maretto, "Gas Hold-Up in Slurry Bubble Columns: Effect of Column Diameter and Slurry Concentrations," AIChE J. 43, 311-316 (1997).

Krishna, R., J. Ellenberger and C. Maretto, "Flow Regime Transition in Bubble Columns," Int. Comm. Heat Mass Transfer 26, 467-475 (1999).

Krishna, R., M. I. Urseanu, J. W. A de Swart and J. Ellenberger, "Gas Hold-Up in Bubble Columns: Operation with Concentrated Slurries Versus High Viscosity Liquid," Can. J. Chem. Eng. 78, 442-447 (2000).

Lin, T.-J., J. Reese, T. Hong and L.-S. Fan, "Quantitative Analysis and Computation of Two-Dimensional Bubble Columns," AIChE J. 42, 301-318 (1996).

Mudde, R. F., D. J. Lee, J. Reese and L.-S. Fan, "Role of Coherent Structures on Reynolds Stresses in a 2-D Bubble Column," AIChE J. 43, 913-926 (1997a).

Mudde, R. F., J. S. Groen and H. E. A. Van Den Akker, "Liquid Velocity Field in a Bubble Column: LDA Experiments," Chem. Eng. Sci. 52, 4217-4224 (1997b).

Olmos, E., C. Gentric and N. Midoux, "Identification of Flow Regimes in a Flat Gas-Liquid Bubble Column Via Wavelet Transform" Can. J. Chem. Eng. 81, 382-388 (2003).

Ruzicka, M. C., J. Zahradnik, J. Drahos, and N. H. Thomas, "Homogeneous-Heterogeneous Regime Transition in Bubble Columns," Chem. Eng. Sci. 56, 4609-4626 (2001a).

Ruzicka, M. C., J. Drahos, M. Fialova, N. H. Thomas, "Effect of Bubble Column Dimensions on Flow Regime Transition," Chem. Eng. Sci. 56, 6117-6124 (2001b).

Sarrafi, A., M. Jamialahmadi, H. M. Steinhagen and J. M. Smith, "Gas Hold-up in Homogeneous and Heterogeneous Gas-Liquid Bubble Column Reactors," Can. J. Chem. Eng. 77, 11-21 (1999).

Saxena, S. C. and Z. D. Chen, "Hydrodynamics and Heat Transfer of Baf.ed and Unbaf.ed Slurry Bubble Columns," in "Rev. Chem. Eng.," Freund Pub. House 10, 236-300, London (1994).

Tzeng, J.-W., R. C. Chen and L.-S. Fan, "Visualization of Flow Characteristics in a 2-D Bubble Column and Three-Phase Fluidized Bed," AIChE J. 39, 733-743 (1993).

Van Benthum, W. A. J., R. G. J. M. Van der Lans, M. C. M. Van Loosdrecht and J. J. Heijnen, "Bubble Recirculation Regimes in an Internal-Loop Airlift Reactor," Chem. Eng. Sci. 54, 3995-4006 (1999).

Vial, C., S. Poncin, G. Wild and N. Midoux, "A Simple Method for Regime Identi.cation and Flow Characterisation in Bubble Columns and Airlift Reactors," Chem. Eng. and Proc. 40, 135-151 (2001a).

Vial, C., R. Laine, S. Poncin, N. Midoux and G. Wild, "In.uence of Gas Distribution and Regime Transitions on Liquid Velocity and Turbulence in a 3-D Bubble Column," Chem. Eng. Sci. 56, 1085-1093 (2001b).

Wallis, G. B., "One-Dimensional Two-Phase Flow," McGraw-Hill, New York, NY (1969).

Wild, G., S. Poncin, H.-Z. Li and E. Olmos, "Some Aspects of the Hydrodynamics of Bubble Columns," Int. J. Chem. Rec. Eng. 1, R7, 1-36 (2003).

Wilkinson, P. M., A. P. Spek and L L. Van Dierendonck, "Design Parameters Estimation for Scale-Up of High-Pressure Bubble Columns," AIChE J. 38, 544-553 (1992).

Yang, Y. B., N. Devanathan and M. P. Dudukovic, "Liquid Back-Mixing in Bubble Column via Computer-Automated Radioactive Particle Tracking (CARPT)," Exp. Fluids 16, 1 (1993).

Zahradnik, J., M. Fialova, M. Ruzicka, J. Drahos, F. Kastanek, and N. H. Thomas, "Duality of the Gas-Liquid Flow Regimes in Bubble Column Reactors," Chem. Eng. Sci. 52, 3811-3826 (1997).

Manuscript received January 7, 2005; revised manuscript received October 27, 2005; accepted for publication November 7, 2005.

Department of Chemical and Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, Singapore, 117576

May Khin Thet, Chi-Hwa Wang and Reginald B. H. Tan *

* Author to whom correspondence may be addressed.

E-mail address: chetanbh@nus.edu.sg

Table 1. Different operating conditions for all experiments

System Orifice dia, No. of
 mm orifice

Bubble column 0.5 49
Column with DT Plate 1(centre)

Bubble column 0.5 225
(Two- and three-phase) (Plate 1)

Bubble column 1.6 1-4
(Two-phase) (Plate 2)

System H/D, Test section
 aspect ratio

Bubble column 3.7 y = 0.03-0.11 m
Column with DT at z = 0.011 m

Bubble column 1.7, 2.3, 3.7, 5
(Two- and three-phase)

Bubble column 1.7 y = 0.03-0.11 m
(Two-phase) y = 0.11-0.21 m
 y = 0.21-0.26 m
 at z = 0.065 m

Table 2. Physical properties of the particles.

 Mean diameter Sphericity
 ([micro]m) [psi] = [A.sub.sphere]/
 [A.sub.[particle]
 (Hartman
 et al., 1994)

Type of material

Polycarbonate 3000 0.75
Glass sphere 500 1
Glass sphere 3000 1

 Density
 (kg/[m.sup.3])

Type of material

Polycarbonate 1200
Glass sphere 2500
Glass sphere 2500

Table 3. Apparent viscosity data for glucose-deionized water

Mass fraction Viscosity
(% wt.) (mPa s)

0 1.24

0.032 1.32

0.061 1.4

0.163 1.93

0.279 2.9

0.29 3.1

0.31 3.5

0.34 4

0.41 6.7