database expansion for bubble column flows
by Peter Spicka
1Gas-liquid Bubble columns
A chronological summary of the experimental work reported for DOE in period from 1995 to 2001 is given. This summary contains an overview of the experimental data obtained by Computed Tomography (CT) and Computer-Automated Radioactive Particle Tracking (CARPT). The data was collected in different bubble columns and various gas-liquid and gas-liquid-solid systems for relatively broad range of gas superficial velocities, encompassing bubble and churn-turbulent regimes. Furthermore, effect of gas distributor internals, solid loading and pressure was studied. This report provides the summary of findings, major conclusions and samples of results for each reported study.
1.1 Study on Liquid properties and Internals
Among the first reported data from gas-liquid bubble column were CARPT and CT experiments from 18” diameter column using air-water and air-drakeoil systems (quarterly reports No.6 and 7). A perforated plate distributor was used with pore size of 0.7 mm and porosity of 0.076%. Experiments were conducted at three : 2, 5 and 10 cm/s. It was noticed that the measured gas holdup distributions are similar to those observed in smaller diameter columns and qualitative similarities were found both in the air-water and air-drakeoil systems. However, the overall gas holdup was found lower for air-drakeoil system when compared to air-water system due to the large bubbles formed in the drakeoil. The higher gas holdup (and also higher radial gradient of gas holdup) in air-water leads to higher axial liquid velocity as shown in Figure 1a at gas superficial velocities of 2.0. The maximum up flow liquid velocities are about 23 cm/s for air-drakeoil system and 28 cm/s for air-water at superficial gas velocity of 2 cm/s. At the same conditions, the maximum down flow velocities for the systems are 10 cm/s for air-drakeoil and 25 cm/s for air-water.
Similar experiments were performed in 18” column containing internals consisting of a bunch of 16 aluminum rods extending vertically over whole column height. Comparison between the systems with/without internals was reported in quarterly reports No. 9, 10 and 11. The reported data includes gas holdup, time-averaged liquid velocities, shear and Reynolds stresses and turbulent kinetic energy. It was concluded that the internals do not affect significantly main recirculation velocity (see Figure 1b). It was further observed that in the column with internals, the velocity inversion point moves radially inward as the superficial gas velocity increase. That is not case of the column without internals, where the velocity inversion occurs almost at the same radial position, independently on . The magnitudes and the trends of the radial profiles of the turbulent stresses in the column with internals are quite similar to those obtained in the column without internals. However, the shear stress in the column with internals is lower than that in the column without internals indicating that the presence of the internals indeed reduces the radial length scales of turbulence.
a) / b)Figure 1Effect of: a) liquid properties and; b) internals on axial liquid velocity
1.2eddy diffusivities and recirculation Velocities
Effect of column diameter and gas superficial velocity on eddy diffusivities in air-water column was reported in Quarterly report No.13 and further extended in 2nd Topical report 1998 for air-drakeoil systems with/without internals. The outcome of this study was a methodology that allows the estimation of the mean liquid recirculating velocity and turbulent eddy diffusivities in churn-turbulent regime. CARPT data in an air-water system and three different column sizes, 14 cm, 19 cm, and 44 cm, were considered. Figure 2 shows axial and radial eddy diffusivities, and , as a function of gas superficial velocity, . The following dependencies have been developed for and (based on CARPT data), which are applicable to large diameter columns (> 10 cm) in the churn-turbulent flow regime (Ug > 5):
(1)
(2)
Furthermore, CARPT results for the average radial and axial eddy diffusivities in the churn-turbulent flow regime(Figure 3), indicate that the radial profiles of the turbulent diffusivities can be approximately expressed as follows:
(3)
(4)
Parameters and are fourth order and second order polynomials which are independent of gas velocity and column diameter. This is illustrated in Figure 3, which show the profiles evaluated using Equations 3 and 4. The reasonably good comparisons suggest that the above Equations 3 and 4 in combination with Equations 1and 2 can be used to estimate the profiles for the axial and radial eddy diffusivities as a function of column diameter, , and superficial gas velocity, , in air-water bubble columns operating in the churn-turbulent flow regime. Ax expected, the magnitude of both the axial and radial eddy diffusivities increases with column diameter
Finally, the axial eddy diffusivities in air-water were found to be comparable with air-drakeoil system. Small differences in magnitude of eddy diffusivities were attributed to larger turbulence time scale in air-drakeoil compared to air-water and inability of CARPT to capture eddies with frequencies less than 30 Hz.
a) / b)Figure 2Effect of superficial gas velocity and column diameter on the average a) axial and; b) radial eddy diffusivity.
a) / b)Figure 3Radial profiles of axial (a) and radial (b) eddy diffusivity compared with Equation 3 and 4.
1.3Scale-Up issues
Effect of column diameter and gas superficial velocity on gas holdup and liquid recirculations was reported in Quarterly report No.12. Using experimental data, both from CREL and from the literature, equations have been developed for the prediction of the mean liquid recirculating velocity in air-water atmospheric systems as a function the both gas superficial velocity and column diameter: :
(5)
Predictions using the above equation compare reasonably well with data of Menzel et al. (1990) and Nottenkamper et al. (1983) at = 32.4 cm/s, as shown in Figure 4b. A similar dependency of ()0.33 has been reported by Joshi and Sharma (1979) and Zehner (1982) for the liquid circulation velocity that was derived based on the assumption of the existence of multiple circulation cells. Equation 3 can be used to estimate the mean recirculation liquid velocity in an air-water bubble column (> 10 cm), at atmospheric pressure operating in the churn turbulent flow regime.
a) / b)Figure 4Effect of superficial gas velocity and column diameter on: a) overall gas holdup and; b) mean liquid recirculating velocity.
1.4Effect of Gas Distributor
In Quarterly report No.14, the experimental measurements obtained in CREL using different distributor designs were reported. The data from different bubble columns of 14 cm (6"), 19 cm (8") and 44 cm (18") in diameter were compared over the range of between 2 cm/s and 12 cm/s which spans both the bubble flow and churn turbulent flow regime. These studies were performed at atmospheric pressure and CARPT was used to obtain liquid velocities and mixing parameters.. Table 1 summarizes their designs as well as the experimental conditions used.
Overall gas holdup measurements were performed in each column prior to the CARPT experiments, by measuring the bed expansion. The results are presented in Figure 5. No specific trend of the gas holdup with respect to the column diameter was observed. It was found that the global gas holdup is sensitive to the static height of the liquid in the column, for a given column diameter. For larger column aspect ratios (e.g. length to diameter ratios greater than 8) this dependence becomes small and eventually disappears, since the contribution of the end zones to the overall gas holdup decreases with an increase in aspect ratio.
An example of CARPT results from this study is shown in Figure 6. It displays the time averaged axial liquid velocity profile, as a function of radial position for various distributors in the 19 cm diameter column at a high superficial gas velocity of 12 cm/s. In the region next to the wall, it was observed that in the column with the bubble cap distributor, 8B, and cone distributor, 8C there are larger bubble sizes than observed for the perforated plate distributor 8A. In addition, the flow appeared more violent for the case of the cone and the bubble cap, with large structures frequently moving up the system in a spiraling motion. On the other hand, with the perforated plate the flow appeared less violent and the large structures were less distinct and less frequent.
As shown in Figure 7a, the turbulent kinetic energy in the fully developed flow region obtained with the bubble cup and cone distributors is much larger than that obtained with the perforated plate distributor. The difference between center-line kinetic energies produced by the bubble cup or cone and perforated plate can be as much as 40%. This indicates that the bubble cup and cone distributor generate larger scale turbulence than perforated plate. These large eddies are more effectively recorded by the trace particle in CARPT which can follow frequencies of up to 25 Hz. Figure 7b indicates that there is no significant difference between the turbulent shear stresses for the cone and bubble cup distributor used. The perforated plate distributor gives slightly larger Reynolds shear stress values in the region between r/R ~ 0.4 to r/R ~ 0.8 compared to those obtained by the other two distributors. The maximum difference is 10%.
Table 1Operating Conditions and Experimental Details for the Gas Distributor Effects Study
[cm] / Distr. /
[cm/s] /
[cm] /
[cm] / / .
[hr] /
[cm] /
[cm]
14.0 /
6A
/ 2.4 / 120.2 / 133.2 / 0.098 / 18.0 / 70.0 / 115.0/ 4.8 / 98.0 / 120.4 / 0.186 / 18.0 / 50.0 / 95.0
/ 9.6 / 98.0 / 126.0 / 0.222 / 18.0 / 50.0 / 95.0
/ 12.0(1) / 98.0 / 129.1 / 0.241 / 18.0 / 50.0 / 95.0
/ 12.0(2) / 98.0 / 129.5 / 0.243 / 18.0 / 50.0 / 95.0
14.0 /
6B
/ 2.4 / 98.0 / 105.0 / 0.067 / 18.0 / 45.0 / 80.0/ 4.8 / 98.0 / 112.0 / 0.125 / 18.0 / 50.0 / 80.0
/ 9.6 / 98.0 / 123.0 / 0.203 / 18.0 / 50.0 / 90.0
/ 12.0 / 98.0 / 126.0 / 0.222 / 18.0 / 40.0 / 90.0
19.0 /
8A
/ 2.0 / 104.0 / 114.7 / 0.093 / 18.0 / 80.0 / 100.0/ 5.0 / 103.5 / 128.0 / 0.191 / 18.0 / 55.0 / 95.0
/ 12.0 / 95.5 / 124.0 / 0.230 / 18.0 / 40.0 / 90.0
8B
/ 12.0 / 95.5 / 118.0 / 0.191 / 18.0 / 50.0 / 90.08C
/ 12.0 / 95.5 / 117.0 / 0.184 / 18.0 / 50.0 / 90.044.0 /
18A
/ 2.0 / 179.0 / 193.1 / 0.073 / 36.0 / 115.0 / 155.0/ 5.0 / 179.0 / 209.8 / 0.147 / 36.0 / 90.0 / 160.0
/ 10.0 / 176.1 / 217.6 / 0.191 / 36.0 / 85.0 / 180.0
Distributors used
6Aperforated plate with 0.4 mm diameter of holes and 0.05 % open area
6Bperforated plate with 1.0 mm diameter of holes and 0.62 % open area
8Aperforated plate of 0.33 mm holes diameter of holes and 0.05% open area
8Bbubble cap distributor
8Cone point inlet (1.27 cm in diameter) cone distributor
18Aperforated plate distributor with 0.7 mm diameter of holes and 0.077% open area
Figure 5Global gas holdup as a function of superficial gas velocity in the different columns.
The general observations from CARPT experiments were that in the middle column region, the axial liquid velocities dominate (15 cm/s to 60 cm/s), and the radial and azimuthal velocities are negligible ( 1 cm/s), and can be considered to be zero. For the columns with large aspect ratio ( 7), symmetry exists with respect to the column axis in this region. However, near the distributor zone, a symmetric flow pattern about the column axis is absent. The extent of asymmetry seems to depend on the distributor used. The asymmetry in the time-averaged flow in this region is therefore attributed to the influence of the distributor. For most cases in the middle well-developed region of the column, the axial variation of the time-averaged velocities is not significant. This study also tried to suggest suitable spargers for further experiments. The proposed spargers were: one point inlet sparger, perforated plate distributor with holes uniformly distributed and with half the holes blocked.
.
Figure 6Effect of distributor on the time averaged liquid velocity in a column of diameter 19 cm, = 12.0 cm/s. Distributors are: cone (8C), bubble cap (8B), and perforated plate (8A).
a) / b)Figure 7Effect of distributor on the turbulent kinetic energy (a) and Reynolds stresses (b) in a column of diameter 19 cm, = 12.0 cm/s. Distributors are; cone (8C), bubble cap (8B), and perforated plate (8A).
Another study concerning the gas distributor effect on gas holdup was reported in Quarterly report No. 18. Distributor effects have been studied in a 6.4” diameter column at atmospheric conditions at two superficial gas velocities of 14 cm/s and 30 cm/s. Five different gas distributors were used in this study as shown in Figure 8 (D1-D5)
Gas holdup profiles at a high superficial gas velocity of 30 cm/s (Figure 9b) under atmospheric condition confirmed the findings of Shollenberger et al. (1999) that in churn-turbulent flow the gas distributor does not affect the gas holdup or radial gas holdup profile. Moreover, the entry length is confined to z/D 2.
Figure 8A schematic diagram of the gas distributors.
However, the findings from this study indicate that there is a significant variation in gas holdup from perforated plate type distributors to nozzle type distributors at at =14 cm/s. Such differences were not observed at =30 cm/s where the turbulent intensities are much higher.At a superficial gas velocity of 14 cm/s (Figure 9a), the perforated plate spargers exhibit a similar behavior of axially decreasing gas holdup, whereas the cross and single nozzle spargers show insignificant variation of gas holdup with height at heights above two column diameter. In addition, it is interesting to note that Kumar (1994) and Degaleesan (1997) also observed similar trends when using perforated plate distributors compared to nozzle type distributors in a different column geometry.
a) / b)Figure 9Effect of spargers on gas holdup radial profiles for distributor at: a) = 14 cm/s and; b) =30 cm/s at two different axial positions.
Finally, recent study (Quarterly report No.27) focused on an effect of nozzle orientation of the cross sparger D2 (Figure 8). Liquid velocity profiles at two nozzle orientations (nozzles facing upwards and downward) were reported and compared at two different pressures, =1 and 4 bars. As can be seen from Figure 10a, in the bottom of the column, the nozzles facing upward exhibit slightly steeper liquid velocity profiles than nozzles facing downward. This trend is consistent at both pressures, =1 and 4 bars. Furthermore, the nozzle orientation effect diminishes with the increased elevation and basically, it is irrelevant at 5.5. An interesting fact is revealed by the comparison of the liquid velocity profiles at two different pressures where steeper liquid velocity profile were observed at higher pressure. This was explained by increased gas density, resulting from higher pressure, which increases liquid re-circulation. Similar conclusions, concerning the nozzle orientation effect, can be drawn from turbulent kinetic energy, , profiles, shown in Figure 10b. Unlike liquid velocity profiles, no significant change in profiles was observed when the system pressure was increased.
a) / b)Figure 10Radial profiles of: a) axial liquid velocity and; b) kinetic turbulent energy. Legend: ND - nozzles facing downward; NU – nozzles facing upward.
1.5Effect of Pressure
Pressure effect on gas holdup in bubble columns via computed tomography (CT) in 6.4” diameter column were reported in Quarterly report No.22. The experiments were performed at = 1, 4 and 10 bars and room temperature of approximately 20 C. The superficial gas velocities were =2, 8, 14 and 30 cm/s. Moreover, the pressure effect was examined for five different spargers D1, D2, D3, D4 and D6, depicted in Figure 8.
Some interesting results from this study are presented in Figure 11. It shows the cross-sectional time-averaged gas holdup distribution in the 6.4” diameter bubble column at selected pressures and at different superficial gas velocities. The plots confirm that gas holdup always increases with superficial gas velocity and increases with pressure but way in churn turbulent flow at higher superficial gas velocities. It can also be observed from Figure 11 that gas holdup distribution is almost uniform at lower superficial gas velocities of 2 and 8 cm/s. Other conclusions from this study can be drawn as following:
- The pressure effect is insignificant at lower superficial gas velocity when the column is in bubbly flow.
- The flow regime transition is shifted to higher gas superficial velocity at higher pressure. In addition, the steepness of the radial gas holdup profile increases marginally with pressure;
- At Ug = 30 cm/s and P = 1 atm, the distributor effect on gas holdup is insignificant.. On the other hand, at P = 4 atm and Ug = 30 cm/s, some variation of the gas holdup near the wall of the column was observed;
- Overall, we concluded that radial gas holdup profiles at a pressure of 4 bars and superficial gas velocity of 30 cm/ are very close to each other if one excludes the radial gas holdup profile obtained with the single nozzle distributor.
Figure 11Effect of pressure on gas holdup profiles at scan level z/D = 5.5 using distributor D4.
2Slurry bubble columns
In the first project budget year 1996, experimental data for the solid motion in a slurry bubble column of 6” diameter has been reported. The solid loading used was 7% and the experiment were conducted at three different superficial gas velocities , 2, 5 and 10 cm/.. The study of slurry bubble columns later continued to investigate the effect of solid loading and gas distributor on major hydrodynamic parameters, as described in the following sections.
2.1Solid loading and Sparger EFFECT in Slurry bubble columns. Comparison with G-L Systems
The effect of column diameter, gas superficial velocity, sparger design and solids loading on liquid/slurry velocity have been studied in Quarterly report 12. Two different columns of diameter 4” and 6” were used in this study. The operating conditions used for G-L and G-L-S slurry systems are listed in Table 2 and Table 3, respectively. Sparger design specifications used in the experiments are listed in Table 4.
The 4” column data showed that sparger design has a much smaller effect on time averaged axial velocity profiles in slurry systems than in gas-liquid systems (15% variation in G-L-S compared to up to 100% in G-L). Similarly, variation in has a smaller influence on the time averaged axial velocity profiles in slurries than in G-L systems. In line with this finding the effect of is more noticeable in slurries with lower (7 wt.%) than higher (20 wt.%) solids loading. As shown in Figure 12, a change in solids loading while keeping superficial gas velocity constant changes the axial velocity profiles up to only 20% (6 inch column).
All but one experiment in slurry systems exhibited a typical one-cell recirculation pattern with column core up-flow and wall down-flow regions. Only in the 6 inch column slurry experiment with 7 wt.% of solids loading and at 2.0 cm/s superficial gas velocity a two-cell recirculation pattern was seen, with cell at the bottom of the column with core down-flow and wall up-flow regions.