Three-phase slurry bubble columns are widely employed in
several applications in chemical and petrochemical industries, for example, in
Fischer-Tropsch synthesis, catalytic hydrogenation of edible oils, coal
liquefaction etc. Three-phase flows in slurry reactors operating under dense
conditions offer challenges not only to their experimental characterization,
but also to the development of computational flow model capable of predicting
the dynamic and time-averaged flow characteristics. While there exists a large
volume of literature on development and experimental verification of two-fluid and multi-fluid models coupled with population balance models to simulate
the dynamics of dispersed gas-liquid
flows [1-3], relatively limited information exists on numerical simulation of
dense gas-liquid-solid flows in slurry bubble columns.

Most of the earlier simulations of gas-liquid-solid flows
in slurry bubble columns were carried out using two-/three-fluid Eulerian
simulations [4, 5]. In these
investigations, the inter-phase momentum exchange was accounted through the
drag force based on averaged (constant) bubble size. The experimental
validation of such two-/three-fluid models was often limited only to low gas
velocities. While the predictive capability of multi-fluid models was found to
be better since the inter-phase drag force was evaluated using bubble diameter
of individual size groups/gas phases. However, experimental verification of the
multi-fluid models was still performed using time-averaged profiles of gas
hold-up that comprised of all bubble groups/sizes. It is not clear if such
models could predict the spatial distribution of hold-up of individual bubble
size groups/phases and dynamic characteristics correctly. The predictive
abilities of the computational models to simulate the effects of solid loading
and superficial gas velocity on instantaneous flow characteristics for dense
three-phase flow conditions are not yet fully understood.

In the present work, we have performed experimental
and numerical investigations of dispersed gas-liquid-solid flows in a
rectangular slurry bubble column with the objectives of (a) quantitative
measurements on the effect of solid loadings (5 - 40%) on spatial
distribution of gas hold-up and bubble size distribution for superficial
gas velocities up to 30 cm/s and (b) verification of the predictive
capabilities of two?fluid and multi?fluid Eulerian models in
terms of time-evolution of instantaneous gas hold-up fluctuations and spatial distribution
of hold-up of individual bubble classes/ size groups.

Experiments were performed in a rectangular slurry
bubble column of 141 cm (height) x20 cm (width) x 5 cm (depth) as shown in
Figure 1. The demineralized water was used as the liquid phase and air was
injected using a uniform sparger at the bottom cross-section (10 cm x 5 cm).
Glass particles of ~ 250 µm size were used as the solid phase. The dispersion
height was maintained at 90 cm in all experiments. ?In-house? developed voidage
probes were used to measure the instantaneous and time-averaged gas hold-up at
various locations in the column for various solid loading 0, 20 and 40 %. The
detailed experimental work will be discussed in full length manuscript.

Figure
1. Experimental set-up

Two-/three-fluid
Eulerian simulations of transient gas-liquid/gas-liquid-solid flows in a
three-dimensional rectangular slurry bubble column, with the dimensions same as
that was used in experiments, were performed using commercial flow solver Fluent v14 (Ansys Inc.). Air was introduced from
the bottom cross-section that corresponds to the uniform sparger used in
experiments. The model
based on kinetic theory of granular flow (KTGF) was used for the solid phase
and a set of standard k-ε equations was used to simulate turbulence in the
liquid phase. The effect of gas-liquid momentum exchange was accounted through
drag force model proposed by Tsuchiya et al. [6] and liquid-solid momentum
exchange through drag model proposed by Wen and Yu [7]. Further details of the computational model, boundary
conditions and numerics will be described in the full manuscript.

The dynamic
and time-averaged flow behavior of gas-liquid/gas-liquid-solid flows in slurry
bubble column was investigated at superficial gas velocity in the range 5 - 30
cm/s and for solid loading 0 - 40 vol. %. The effect of solid loading on
measured and simulated instantaneous gas hold-up and solid hold-up distribution
in slurry bubble column is shown in Figure 2 (at superficial gas velocity of 20
cm/s and for solid loading (a) 0 %, (b) 10 %, (c) 20 %, (d) 30 % and (e) 40 %).
The
simulated effects of superficial gas velocity and solid loading on overall gas
hold-up were found in a qualitative agreement with the measurements.

(a) G-L	(b) G-L-S (10 %)	(c) G-L-S (20 %)	(d) G-L-S (30 %)	(e) G-L-S (40 %)

Figure 2. Snapshots
of gas-liquid/gas-liquid-solid flows in slurry bubble column at superficial gas
velocity of 20 cm/s for various solid loadings showing the (i) experiments (ii)
simulated instantaneous gas hold-up and (iii) simulated instantaneous solid
hold-up [20 uniform color contours form 0 (blue) to 0.5 (red)]

It
can be seen that the addition of solids (in different volume %) leads to
formation of gas pockets/slugs of different sizes (Figures 2(i) (a)-(e)). At
solid loading of 40 vol. %, the gas slugs were seen to span almost entire width
of the column (Figure 2(i) (e)). The corresponding simulated instantaneous gas
and solid hold-up distributions are shown in Figure 2(ii) (a)-(e) and 2(iii)
(a)-(e), respectively. Though the non-uniformity in simulated gas hold-up distribution
was seen to increase with increased solid loading, three-fluid model (with
certain interphase closure with appropriate corrections) was not able to
predict the formation of large gas slugs as observed in the experiments. A
comparison of simulated time evolution of local gas hold-up for different solid
loadings (0, 10 and 40 vol. %) at two different spatial locations (marked by
arrows in Figure 2) is shown in Figure 3(a) and (b). In case of gas-liquid flow (0%
solid), the local gas hold-up was found to fluctuate from 0.15-0.55 (see Figure
3(b)). However, with addition of solids, the gas hold-up fluctuations exhibit
formation of large gas slugs (also indicated by peaks where gas hold-up
fluctuates between ~ 0 to ~ 0.8).

Figure 3. Effect
of solid loading on gas hold-up fluctuation time-series recorded at superficial
gas velocity of 20 cm/s and solid loading of 0, 10 and 40 vol. %

Experiments will be performed to
measure such instantaneous gas hold-up fluctuations and time-averaged gas
hold-up profiles for different solid loadings. 
The measured instantaneous gas hold-up fluctuations will help to
understand the bubbling/slugging characteristics as a function of solid
loading. These measurements will be used to verify the simulated results shown
in Figure 3. Further simulations will be performed using the three-fluid model
to understand the effect of different interphase drag force formulations and
multi-fluid models (to account for bubble size distribution). The ability of
computational models with interphase drag formulations and bubble size to
simulate instantaneous gas hold-up fluctuations in different regimes
(bubbling/slugging) and time-average gas hold-up distribution will be verified
using the measurements.

References

[1]     
Bhole,
M. R., J. B. Joshi and D. Ramkrishna, Chem. Eng. Sci. 63, 2267-2282,
2008.

[2]     
Chen,
P., J. Sanyal, M. P. Dudukovic, Chem. Eng. Sci. 59, 5201-5207, 2004.

[3]     
Díaz,
M. E., A. Iranzo, D. Cuadra, R. Barbero, F. J. Montes and M. A. Galan, Chem.
Eng. J., 139, 363-379, 2008.

[4]     
Panneerselvam,
R., S. Savithri, G. D. Surender, Chem. Eng. Sci. 64, 1119-1135, 2009.

[5]     
Rampure,
M. R., V. V. Buwa and V. V. Ranade, Can.  J.  Chem. Eng. 81,
692-706, 2003.

[6]     
Tsuchiya,
K., A. Furumoto, L. S. Fan and J. Zhang, Chem. Eng. Sci. 52, 3053-3066,
1997.

[7]     
Wen,
C. ?Y. and Y. H. Yu, Chem. Eng. Prog. Symp. Series, 62, 100-111, 1966.