Improving Flooding Performance for Countercurrent Monolith Reactors


Improving Flooding Performance for Countercurrent Monolith Reactorshttps://pubs.acs.org/doi/pdf/10.1021/ie034289fSimilar...

6 downloads 121 Views 181KB Size

4848

Ind. Eng. Chem. Res. 2004, 43, 4848-4855

Improving Flooding Performance for Countercurrent Monolith Reactors Achim K. Heibel,*,† Joshua A. Jamison,‡ Pierre Woehl,§ Freek Kapteijn,| and Jacob A. Moulijn| Corning Environmental Technologies and Corning Science & Technology, Corning Inc., Corning, New York 14831, Corning Science & Technology, Fontainebleau Research Center, Corning S.A.S., 77210 Avon, France, and Reactor and Catalysis Engineering, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands

The limiting factors on the flooding performance of monoliths have been investigated in detail. In addition to the channel size and the void fraction of the monolith itself, the flooding limits are strongly impacted by inlet and outlet effects of the reactor bed as well as the stacking of monolith segments. Solutions to overcome these additional hydraulic restrictions, based on stacking slices of high-void-fraction monoliths with strategic step changes in cell density, are identified and evaluated. Implementation of the improvements allowed for the countercurrent operation of monoliths with a channel size as small as 1.25 mm even when a high-surfacetension liquid (water) is used. With the enhanced configurations, flooding experiments over a wide range of channel geometries of single and multiple monoliths were performed. Results from small-scale testing are in fair agreement with those obtained from the large-diameter tests, indicating a good scale-up behavior of the technology. The flooding limits are conveniently described with a flooding correlation based on an adjusted capacity plot taking the geometry properties of the monolith into account: CG/(xgdh) ) -0.025 + 0.12FLG-0.475 for 0.1 e FLG e 10. The measured flooding performance is in good agreement with recently reported data for flooding investigations of a falling film in an annular-flow configuration (Stockfleth, R.; Brunner, G. Ind. Eng. Chem. Res. 2001, 40, 6014-6020) that were performed with a significantly larger hydraulic diameter (12.5 mm). Monoliths enable stable countercurrent film flow operation with capillary-sized channels (1.25-4.0 mm), which so far was not feasible for any other packing structure. The findings are summarized in a design guide for the configuration of a monolith reactor for countercurrent operation. Introduction Countercurrent film flow in single tubes is a wellstudied phenomenon18 and is widely applied in industry for falling film absorbers, evaporators, heat pipes, reflux condensers, and emergency cooling of nuclear reactors.17 In the film flow regime, the gravity-driven liquid flows down the channel wall, and the gas occupying the center of the tube flows in the opposite direction. The clear separation and low interaction of the flowing phases in the single tube allows for countercurrent operation over a wide operating window.1 The significant impact of the channel inlet and outlet on the flooding performance is a well-known issue for larger tube sizes7 and is even magnified for capillary-sized channels.10,12 Recently, the hydrodynamics of monoliths in countercurrent operation was the subject of several fundamental research programs.5,6,9,11 In principle, the monolith can be considered as an assembly of many parallel channels, which should also allow countercurrent operation upon application of the appropriate operating * To whom correspondence should be addressed. Address: A. K. Heibel, Corning, Inc., MP-HQ-W2-21, Corning, NY 14831. Tel.: +1-607-974-8430. Fax: +1-607-974-4617. Email: [email protected]. † Corning Environmental Technologies, Corning Inc. ‡ Corning Science & Technology, Corning Inc. § Corning Science & Technology, Corning S.A.S. | Delft University of Technology.

conditions. The main difficulties for establishing stable countercurrent flow are, in general, the smaller size of the channels (dh < 5.0 mm) and the complications at the inlet and outlet of the monolith block. A dedicated study6 on square-channel monoliths confirmed especially the importance of the liquid drainage at the outlet. Special outlet devices were required to effectively drain the liquid flowing out of the monolith. Because of these strong drainage effects, the countercurrent operation of monoliths has been mostly limited to monolith channels with hydraulic diameters larger than 2.5 mm and quite often has been feasible only for organic liquids having low surface tension.6,11 In agreement with these limitations, a flooding correlation was developed, showing a different dependency on the Bond number, the ratio between buoyancy and surface tension forces, compared to flooding correlations for more common packings. It was concluded that this effect was mostly due to drainage phenomena at the monolith outlet and the applied outlet device. Furthermore, initial studies of multiple monolith segments revealed that the hydraulic capacity for a 25 cpsi monolith decreased by roughly 15% as a result of stacking layers of monoliths on top of each other such that the cells were not aligned vertically with each other. Based on these studies, this work summarizes the results of an extensive research program to better understand the impact of limiting phenomena at the

10.1021/ie034289f CCC: $27.50 © 2004 American Chemical Society Published on Web 07/03/2004

Ind. Eng. Chem. Res., Vol. 43, No. 16, 2004 4849

Figure 1. Monolith for small-scale testing (14.5 mm diameter): (A) cross section, (B) machined outlet section for improved liquid drainage. Table 1. Geometric Properties of the Monoliths cell density/ wall thickness (cpsi/mil)

lw (mm)

dh (mm)

 (%)

S/V (m2/m3)

dMono (mm)

16/40 25/34 50/19 50/24 100/14 100/23 110/19 200/12 200/21

1.02 0.86 0.48 0.61 0.36 0.58 0.48 0.30 0.53

5.33 4.22 3.11 2.98 2.18 1.96 1.94 1.49 1.26

71 69 75 69 74 59 64 69 49

529 654 964 925 1354 1213 1323 1849 1566

95 95 and 102 95 102 95 95 14.5 95 95

Table 2. Fluid Properties at 25 °C and 1.015 bara19

decane water air a

surface tension (mN/m)

dynamic viscosity (cP)

density (kg/m3)

capillary lengtha (mm)

23.4 73.6 -

0.86 0.91 0.0185

728 997 1.185

1.81 2.74 -

Based on eq 1.

monolith outlet and to explore methods to overcome these limitations. The investigations are extended to flooding phenomena initiated at the monolith inlet. Additional focus is given to the flooding performance of stacked assemblies of several monolith segments and configurations to enable enhancements to the hydraulic capacity. As a baseline, the results for a smallerdimension monolith are used and compared to those for a full-size monoliths. Experimental Section Materials. All experiments were performed using monoliths of ceramic material (Cordierite) with a material porosity of roughly 30%. A summary of the geometric properties of the monoliths is given in Table 1. All monoliths had channels with square cross sections. Monoliths with 95- and 101-mm diameters had lengths of 38, 75, 190, 305, or 380 mm as indicated in the text. The monolith used for the small-scale investigations had a diameter of 14.5 mm and a length of 300 mm. The small-scale monolith had a special outlet device machined into the monolith, removing every second row of monolith channels over a length of 20 mm (Figure 1). Table 2 summarizes the fluid properties at 25 °C and 1.015 bara. Pressurized air was used in all experiments as the primary gas phase. Experimental Setup and Procedures. The details of the experimental setup and the installation/packaging of the monoliths are described elsewhere.6 Very similar equipment and approaches were used in the current investigations. It is important to notice that different

spray nozzles were used and the height was adjusted to guarantee a uniform distribution of the liquid5 over a wide range of flow conditions. For evaluation of the small-scale monolith, the uniform introduction of liquid is challenging. Therefore, liquid was sprayed over a larger area, with the monolith cross section covering only part of this area. The liquid flow rate was determined by the flow exiting the monolith. Before each experiment, the porous monolith structures were sufficiently wetted with a high liquid flow for a period of 30 min. This procedure ensured that local wetting differences were minimized. Experiments were conducted for a fixed liquid flow rate, with the gas flow rate increased in steps. After each step, a stabilization period of 15 s was followed by a measurement period of 15 s with a data acquisition frequency of 2 Hz. The pressure drop over the measurement time was averaged, and the standard deviation was calculated. These measurements were performed over a wide range of liquid flow rates. Data Analysis. Flooding was determined by evaluating changes in the pressure drop and the pressure drop fluctuations as a function of the gas flow rate for a given liquid flow rate.6 The experimental setup did not allow for all experiments to be performed at constant temperature. Therefore, the temperatures at the inlet and outlet of the monolith test section were continuously measured, and the actual fluid properties were determined according to the method of Reid et al.13 The necessary reference data were obtained from Yaws.19 Furthermore, it was assumed that the gas phase was completely saturated by the corresponding vapor of the liquid phase. Therefore, the gas flow rate and the gas-mixture properties were adjusted accordingly. The change in liquid flow rate due to evaporation was negligible in all cases. Results and Discussion Exit Effects. The liquid outlet configuration has a significant effect on the flooding performance of monoliths with capillary-sized channels. It is essential that the liquid drains efficiently from the monolith channels; otherwise, droplets will build up at the channel outlet and block the monolith channels, triggering flooding. To enhance the hydraulic capacity, special outlet devices are used to help drain the liquid. The outlet device is built up from several plates with drip points on one side. To understand the impact of the alignment of the plates in this device, two test configurations were investigated. In one case, the plates were aligned with the monolith walls, and in the other case, the channel walls were rotated by 45° with respect to the outlet device (Figure 2A). The latter configuration is a bounding case for maximum interference between the monolith and outlet device. The flooding results (Figure 2B) indicate the importance of proper alignment. In the case of the nonaligned outlet device, additional hydraulic restrictions are created at the interface between the monolith and the outlet device, resulting locally in a smaller equivalent hydraulic diameter. This decreases the flooding limits by about 60% compared to those obtained with an aligned outlet device. In accordance with these findings, all subsequent experiments were performed with an aligned outlet device to determine the upper limits of the hydraulic capacity.

4850 Ind. Eng. Chem. Res., Vol. 43, No. 16, 2004

Figure 2. (A) Alignment of outlet device: aligned (top) and nonaligned (bottom). (B) Flooding performance with aligned (b) and nonaligned (0) outlet device for 95-mm-diameter 25/35 monolith, water/air.

Previous work6 has confirmed that the capillary length

lc )

x

σ g(FL - FG)

Figure 3. (A) Example of outlet stacking configuration: 200/21, 100/23, 50/19, 25/35, 16/40, and outlet device. (B) Flooding performance of a 200/12 monolith with different outlet configurations: (4) with a 25/35 monolith slice and (b) with 100/14, 50/19, and 25/35 monolith slices. Length of monolith slices ) 75 mm, random orientation of slices to each other, outlet device used for each configuration, length of 200 cpsi monolith ) 305 mm, diameter of monolith ) 95 mm, water/air.

(1)

gives a good indication of the droplet size relevant for drainage at the outlet of a given monolith packing. As seen from eq 1, the capillary length scale is a function of the fluid properties, dependent on both its surface tension and its density. With liquid water, the high surface tension increases the capillary length significantly. To prevent flooding, the characteristic length dimension of the monolith channel at the outlet should be sufficiently larger (>125-150%) than the capillary length scale. Therefore, for smaller monolith channels, countercurrent operation can be quickly limited by the liquid drainage at the outlet. Furthermore, it is quite challenging to align the outlet device accurately with the monolith channel walls. The thinner channel walls for the higher-cell-density monoliths and the larger number of cells per unit area make it practically impossible to properly align the outlet grid. Previous work has shown that countercurrent operation using water and air is not feasible for 50 and 100 cpsi monoliths.6 This was mostly attributed to the low ratio between the channel dimension and the capillary length preventing effective liquid drainage. To improve the drainage behavior, monolith configurations with larger channel sizes at the outlet are investigated here. Therefore, monolith segments of different cell densities are stacked together. The segments are arranged in decreasing cell density in the flow direction (Figure 3A). Two different outlet configurations are investigated: (1) 200/12 monolith stacked on a 25/35 monolith with an outlet device and (2) 200/12 monolith stacked on a 100/ 14 monolith, followed by a 50/19 monolith, followed by a 25/35 monolith with an outlet device. The cell walls of the monoliths stacked on each other were not aligned; the orientation was rather random. The 200/12 monoliths had a length of 305 mm. All other monolith segments were 75 mm long. In general, Figure 3B indicates the improvements in flooding performance due to the larger channel dimensions at the outlet enabling better liquid drainage. The hydraulic capacity is limited not only by the channel diameter of the final monolith segment, but also by the transitions between the monolith segments of different cell densities. The more gradual transition from 200 to 100 to 50

Figure 4. (A) Water/air flooding performance of a 50/19 monolith (305 mm long) for different outlet configurations: (4) no outlet stack, (0) with a 75-mm-long 25/35 monolith outlet stack and (b) with 25/35, 16/40 monoliths (each 75 mm long) outlet stack. Outlet device used for each configuration, monolith diameter ) 95 mm. (B) n-Decane/air flooding performance of a 50/24 monolith for different outlet configurations and monolith lengths: (4) 50/24 monolith (380 mm) without outlet stack, (0) 50/24 monolith (380 mm) with a 25/35 monolith (380 mm), and (b) 50/24 monolith (190 mm) with a 25/35 monolith (190 mm). Outlet device used for each configuration, monolith diameter ) 102 mm.

to 25 cpsi shows a better overall flooding performance. If the steps in cell density between monolith segments are too large, some channels might not drain into subsequent channels. However, if the steps are too small, the hydraulic restriction at the interface will lead to a low hydraulic capacity. The optimum is found between the two bounding regimes, with changes in cell density in steps around a factor of 2. Therefore, the channel sizes will change with the square root of two from segment to segment. Similar investigations were performed using 50 cpsi monoliths (Figure 4). Again, a significant improvement in flooding performance was observed when a stack of monoliths at the exit was used. Using water and air, it was found that the opening to a 16 cpsi monolith provides additional benefits in flooding performance compared to the 25 cpsi outlet segment (Figure 4A). In addition to the better performance, the 16 cpsi outlet device requires fewer metal plates and is easier to align with the monolith. In another set of experiments, decane/air was used as the fluid system (Figure 4B). Especially at lower liquid velocities, improvements were found by opening the channels at the outlet to a 25 cpsi monolith. As the liquid velocity increased, the flooding curves appeared to converge, indicating that the outlet

Ind. Eng. Chem. Res., Vol. 43, No. 16, 2004 4851

Figure 5. Liquid bridging at the inlet of a single tube.7

configuration is of less importance at higher liquid velocities. Measurements on the stacked configuration were performed with two different monolith lengths 190 and 360 mm. In both cases, the same hydraulic capacity was found, thus allowing the conclusion that the segment length of the monoliths has a minor impact. Inlet Effects. As indicated above (Figure 4B), at higher liquid velocities, a different mechanism for the onset of flooding seems to become important. Earlier work by Jeong and No7 focused on inlet and outlet flooding in single tubes. Their findings showed that, at lower liquid velocities, exit phenomena dominated. At higher liquid velocities, inlet effects were more relevant. The introduction of liquid into the channel resulted in a thicker liquid film at the entrance, thus leading to bridging between opposite walls and, hence, flooding (Figure 5). For monoliths, similar effects can be envisioned. However, the bridging of a single liquid channel at the inlet, by itself, might not necessarily induce flooding of the entire monolith. It is more likely that the local hydraulic restriction of the blocked channel will actually redirect and increase the gas flow in adjacent channels, which then have a higher potential to flood. This process will then propagate over the monolith cross section. This difference from single channels, with a somewhat selfregulating behavior and a rather broad transition region, has also been encountered in previous work with tube bundles.12 The flooding performance of single channels under high liquid fluxes can be improved by the use of less acute inlet geometries. In the case of monoliths, it was found that a more gradual transition to smaller channel diameters is beneficial. This approach will significantly decrease channel blocking due to bridging. In Figure 6A, an example of an inlet stack using monolith slices is illustrated. From preliminary experiments, it was found that a cell density of 25 cpsi provides sufficiently large channels to prevent inlet-induced flooding. These results were used for an experimental investigation to properly identify the effects of increasing the monolith channels with stacking slices of different cell density monoliths. Figure 6B shows the results of flooding experiments for 100 and 200 cpsi monoliths with and without inlet stacks. Both configurations were tested with the optimum outlet stack configuration. For lower liquid velocities, no effect of the inlet stack could be determined. However, as expected, in the case of the inlet stack, a wider nonflooded operating window was feasible at higher liquid velocities. Stacking Effects. From a manufacturing and handling perspective, monoliths with a length of more than 500 mm are not very attractive. Therefore, in a reactor, multiple monolith segments need to be stacked together to assemble the required length of the packed bed. Previously, it was determined that random stacking of the same monoliths will reduce the hydraulic capacity

Figure 6. (A) Example of inlet stacking configuration: 16/40, 25/ 35, 50/19, 100/23, 200/21. (B) Impact of inlet stack configuration on flooding performance for (4/2) 100/14 and (O/b) 200/12 monoliths: closed symbols, without inlet stack; open symbols, with inlet stack. Inlet stack configuration: 25/35 and 50/19 monolith slices for the 100 cpsi monolith; 25/35, 50/19 and 100/12 monolith slices for the 200 cpsi monolith. Length of monolith slices ) 75 mm, length of 100 and 200 cpsi monoliths ) 305 mm, optimum outlet stack configuration used in each case, monolith diameter ) 95 mm, water/air.

Figure 7. Schematic illustrations of (A) the flow field of the liquid film at the interface between two monolith segments, (B) singleand (C) multiple-spacer monolith stacking configurations.

compared to a single reference piece.6 This effect can be attributed to the local hydraulic restrictions introduced at the interfaces of the monolith segments (Figure 7A) and locally induced flooding. A similar phenomenon was previously determined for sheet metal and wire gauze structured packings by γ ray investigations measuring the local hold-up.16 The authors found that, close to the load/flooding point, the local hold-up increases at the interfaces of two packing units and eventually induces flooding. To overcome these limitations, a special wire gauze packing structure was introduced with a smoother transition at the interfaces between the packing units.2 The advanced packing resulted in higher flooding limits as well as lower pressure drops. In line with these findings and from the understanding developed for the exit configurations, it seemed plausible to investigate the effect of stacking monolith segments with spacers in between, enabling a smoother transition at the interfaces and therefore reducing the introduction of additional hydraulic resistances. In this context, spacers refer to monolith slices (75%) are essential for maintaining the flooding performance and can be of the same cell density as the connecting monolith segments. In another set of experiments, stacking effects for 50 cpsi monoliths were studied using n-decane as an organic liquid (Figure 9A). The experiments were performed using an outlet stacking configuration to guarantee good liquid drainage. A single monolith segment, a stack of two segments, and a stack of three monolith segments with 25/34 spacers were investigated. At higher liquid velocities, the three different configurations exhibited very similar flooding limits. In these cases, flooding was most likely initiated at the inlet, and

Figure 9. (A) Impact of stacking on flooding for 50/24 monoliths: (0) single segment of 190 mm, (4) two 190-mm segments stacked, and (b) three 190-mm segments stacked with a 38-mm-thick 25/ 35 monolith spacer between. All experiments with 190-mm-long 25/35 outlet configuration followed by an outlet device. (B) Impact of gaps between monoliths on flooding performance: Pressure gradient as a function of gas velocity for stacked monolith assemblies (three 190-mm 50/24 monolith segments stacked with 38-mm 25/35 monolith spacers followed by one 190-mm 25/35 outlet configuration with outlet device) for two different liquid velocities, uL0 ) 0.005 m/s (closed symbols) and uL0 ) 0.01 m/s (open symbols). (4/2) Stacking configuration with gaps between monolith segments and (0/9) stacking configuration without gaps between monolith segments. Monolith diameter ) 102 mm, n-decane/air.

therefore, the interfaces between the monoliths were of less importance. This observation is also consistent with the findings reported above for inlet stacking, which was not applied in these experiments. For lower liquid velocities, the results show no differences between the single segment and the multiple segments with spacers, indicating that the stacking with spacers functions well. On the other hand, the configuration with the two monolith segments stacked on top of each other without any spacers exhibited significantly lower flooding limits, illustrating the significant effects of stacking monolith segments without spacers. Even though the appropriate stacking configurations have been identified above, the implementation has a significant impact on the flooding limits. Caution has to be taken that the interfaces between monoliths are even and the end faces of the monoliths are perpendicular to the long axis. Furthermore, gaps between monoliths should be prevented. The impact of a gap is shown in Figure 9B, depicting the pressure drop gradient as a function of the gas velocity for a low and a high liquid velocity. In the case of the stacked monoliths without a gap, the pressure drop increases gradually, nearly linearly, in the nonflooded region.6 Around the flooding point, the pressure drop gradient shows a stronger dependence on the gas velocity, before a clear jump is observed clearly to indicate the onset of flooding. In the case of the monoliths stacked with a gap, flooding occurs at lower gas velocities for both liquid velocities. Additional experimental investigations with water showed that the impact of gaps is even more pronounced. Very often, countercurrent operation was not feasible at all. This might be attributed to the pronounced capillary effects (high surface tension of water), which lead to liquid bridging at the outlet of certain channels at the gap. These channels are then blocked for countercurrent gas flow, resulting in higher gas flows in neighboring channels, which will initiate flooding at lower gas velocities compared to a similar configuration without gaps. Effect of Monolith Channel Dimensions and Flooding Correlation. With the advanced inlet and

Ind. Eng. Chem. Res., Vol. 43, No. 16, 2004 4853

Figure 10. (A) Flooding performance for various single monolith segments: ([) 50/19 monolith, (9) 100/14 monolith, and (2) 200/ 12 monolith. Lines represent best fit, optimum inlet and outlet stack configurations used in each case, monolith diameter ) 95 mm, segment length ) 305 mm, water/air. (B) Dimensionless description of flooding performance for square-channel monoliths: ([) single 50/19 segment, (9) single 100/14 segment, (0) five 100/14 segments, (b) single 100/23 segment, (O) two 100/23 segments, (2) single 200/12 segment, (]) single 200/21 segment. Optimum inlet and outlet stack configurations as well as spacer configuration used in each case, monolith diameter ) 95 mm, segment length ) 305 mm. (×) Small-scale test with single 110/ 19 segment, special machined outlet, monolith diameter ) 14.5 mm, segment length ) 300 mm. The solid line represents the flooding correlation (eq 4) with (30% bounds; the dashed line represents the flooding correlation determined by Stockfleth et al.15 (water/air).

outlet stacking configurations, the impact of channel geometry on flooding needed to be revisited. On the basis of our recent improvements, monoliths with hydraulic channel diameters as low as 1.25 mm were successfully operated in countercurrent flow with highsurface-tension liquids such as water. In Figure 10A, the results for single-segment-monolith experiments with inlet and outlet stacking configurations for three different channel geometries are summarized. In all cases, the void fractions of the monoliths are similar (68-75%), so that changes in cell density mostly reflect differences in hydraulic diameter (1.50-3.18 mm). As expected, the flooding limits decrease with smaller hydraulic diameters. In the case of the 200 cpsi monolith, the maximum liquid velocity for countercurrent flow is limited to about 0.011 m/s compared to 0.021 m/s for the 100 cpsi monolith. Typically, the flooding performance of a packing is summarized in a so-called capacity plot,3 which correlates the capacity parameter

CG ) uG0

x

FG FL - FG

(ms)

(2)

as a function of the flow parameter

FLG )

uL0 ‚ uG0

x

FL FG

(3)

The capacity and flow parameters do not contain any geometrical information, and therefore, each capacity plot is applicable to only a certain packing geometry. In this work, the capacity parameter is extended to include information on the monolith geometry. The underlying assumption is that the flooding of monoliths is dependent on the actual velocity in the channel and the hydraulic diameter of the channel. This description is very similar to the one used in the early flooding investigations of Sherwood et al.14 and applied by

Stockfleth et al.15 To include geometrical effects, the capacity parameter is divided by the void fraction to relate it to the open cross section, the square root of the hydraulic diameter, and the gravity constant. This representation makes the adjusted capacity parameter dimensionless and can be interpreted as the ratio between inertial and buoyancy forces in the gas or vapor phase. The results of all experiments using the optimized inlet and outlet stacking configuration are summarized in the modified capacity plot (Figure 10B). Water and air were used as fluids for the experimental investigations underlying the flooding correlation. Therefore, the impact of fluid properties cannot be derived from this study. From previous work,6 it is known that, qualitatively, both lower surface tension and lower viscosity will lead to higher flooding limits. Single- as well as multiple-segment experiments are included in the graph. Overall, the experimental results follow similar trends and can be well described ((30%) by the following flooding correlation (95% confidence interval range in brackets) over the range investigated

CG  ‚xgdh

)

-0.025 ((0.025) + 0.12 ((0.028)FLG-0.475((0.093) for 0.1 e FLG e 10 (4)

Equation 4 can be used for design purposes for countercurrent monolith reactors over a wide range of channel geometries down to hydraulic diameters of 1.25 mm. For comparison reasons, some small-scale (14.5-mmdiameter) experiments were performed. In the case of the small-diameter monoliths, the exit face of the monoliths were machined (Figure 1) to allow good drainage of the liquid. This configuration was identified from a large screening study and gave excellent flooding results. The approach could not be successfully implemented on larger-diameter monoliths because of the complex machining and potentially negative impact of the larger distances the liquid has to travel before it reaches the circumference of the monolith. These smallscale experiments can be considered as the upper flooding limits that can be for monoliths and similar to a single-channel baseline. The results of the tests with the larger-diameter monoliths are somewhat lower than those obtained in the small-scale experiments, indicating that the scale-up leads to some loss in flooding performance (Figure 10B). Overall, both sets of results are still in reasonable proximity, indicating that the current large-scale configurations leverage the majority of the hydraulic capacity potential of monoliths. Recently, Stockfleth et al.15 reported the flooding performance of a falling film in an annular flow arrangement with a hydraulic diameter of 12.5 mm using corn oil and carbon dioxide under high pressure. The correlation derived by Stockfleth et al. is also represented in Figure 10B and shows good agreement with the results of the current study, even though the experimental conditions and the setup were quite different in the two studies. Stockfleth et al.15 also found that their data showed a spread of (30% in the adjusted capacity plot. For smaller flow parameters, the monoliths exhibited a better flooding performance. This encouraging agreement between the data set of Stockfleth et al.15 and the current work might be fundamen-

4854 Ind. Eng. Chem. Res., Vol. 43, No. 16, 2004

Figure 11. Schematic of recommended bed assembly for countercurrent monolith reactor:(1) monolith inlet stack, (2) main monolith segment for reaction zone, (3) spacer between main monolith segments, (4) outlet stack with outlet device.

tally explained by the similarity of the underlying physical phenomenon of a falling film in both cases. Stockfleth et al.15 confirmed the fundamental importance of the falling film flow for flooding by the good correlation of their data with other structured and nonstructured packings in the same range of hydraulic diameters (∼10 mm). However, one of the major enhancements in the case of the monoliths is the extension of the stable film flow regime into the range of capillary-sized channel dimensions (1.25-4.0 mm). So far, this has not been feasible for any other packing structure. Smaller hydraulic diameters can be very useful for applications where high surface-to-volume ratios (S/V) are required or beneficial. For example, for heterogeneously catalyzed reactions involving a liquid phase, pore diffusion effects very often have a detrimental impact on the overall reaction rate.8 With a given catalyst load per unit volume, structures with a higher surface-to-volume ratio will suffer less from pore diffusion limitations and therefore will improve the overall reactor performance.4 Conclusions In confirmation of previous work, it was found that the drainage of liquid has a significant impact on the overall flooding performance of a monolith reactor operated in the film flow regime. This dominant effect can be addressed by applying an appropriate outlet configuration to the exit of the monolith bed. Stacking of monolith segments with increasing cell density has been established as a viable option. At higher liquid velocities, another phenomenon at the liquid inlet determines the flooding limits of a monolith operated in countercurrent flow mode. To widen the operating window in this regime, a stack of monolith slices with decreasing cell density toward the liquid inlet of the reactor can be used. Additional attention needs to be given to the stacking of monolith segments. Thin slices of high-void-fraction

monoliths acting as spacers between main bodies of monolith segments enable flooding performance comparable to that of a single segment. Figure 11 summarizes the findings of the current work and shows a schematic of a preferred monolith reactor configuration for countercurrent operation. Basically, four different zones can be differentiated: 1. Inlet Stacking Configuration. The top monolith should have a cell density of around 25 cpsi. Monolith layers below should change by a maximum factor of 2 in cell density from slice to slice up to the cell density of the main monolith segments. For handling purposes, the slice thickness of the monoliths used for the inlet configuration should be around 25-50 mm, and for a reduced hydraulic restriction, a void fraction larger than 70% is recommended. 2. Main Monolith Segments. The cell density and wall thickness of the main monolith segments are selected on the basis of the reactive/separation requirements (reaction kinetics and transport phenomena) and the hydraulic capacity (eq 4). For handling and manufacturing purposes, a segment length of around 300 mm is reasonable. 3. Spacers To Stack Main Monolith Segments. Spacers should be of the same or at least more than half the cell density of the connecting monolith segments. For handling purposes, the slice thickness of the monoliths used for spacers should be around 25-50 mm, and for a reduced hydraulic restriction, a void fraction larger than 70% is recommended. 4. Outlet Stacking Configuration. The bottom monolith should have a cell density of around 10-20 cpsi and be attached to an outlet device holding the weight of the bed. The cell density of the subsequent monolith slices should change by a maximum factor of 2 from slice to slice up to the cell density of the main monolith segments. For handling purposes, the slice thickness of the monoliths used for the outlet configuration should be around 25-50 mm, and for a reduced hydraulic restriction, a void fraction larger than 70% is recommended. In general, with the addition of spacers and inlet/ outlet configurations, drainage rivulets are created, allowing liquid to flow along the walls. All of the interfaces should be even and perpendicular to the long axis of the monolith, and no gaps should be apparent between monolith segments or slices. The arrangement of the cell structures of the spacers was not controlled in the current work, but rather was random. This seems to be most representative for the implementation in a real reactor. As with other packings used in countercurrent applications, the flooding limits of monoliths are sufficiently and conveniently described by an adjusted capacity plot representation (eq 4), taking into account the geometric properties of the monolith. Currently, the flooding correlation is based on water/air as the fluid systems. Additional investigations to determine the effects of fluid properties are necessary. Good agreement with recently reported flooding results for a falling film configuration was found. The major benefit of monoliths is the extension of the stable countercurrent film flow operation to capillary-sized channels with high S/V ratios. Acknowledgment Cooperation and discussions with Dr. G. Y. Adusei are highly appreciated. The authors thank Mr. J. Leloup

Ind. Eng. Chem. Res., Vol. 43, No. 16, 2004 4855

for performing the small-scale flooding experiments. We also thank Dr.-Ing. Thorsten Boger (Corning GmbH) and Dr. Charles Sorensen (Corning Inc.) for the revision of the manuscript. Notation d ) diameter, m g ) gravitational constant, m/s2 l ) length, m lc ) capillary length (eq 1), m S/V ) surface-to-volume ratio, m2/m3 u ) velocity, m/s Dimensionless Groups CG ) capacity parameter (eq 2) FLG ) flow parameter (eq 3) Greek Letters ∆p ) pressure drop, Pa  ) void fraction F ) density, kg/m3 η ) dynamic viscosity, Pa‚s σ ) surface tension, N/m Subscripts 0 ) reactor-based c ) capillary G ) gas h ) hydraulic L ) liquid Mono ) monolith w ) wall

Literature Cited (1) Bankoff, S. G.; Lee, S. C. A Critical Review of the Flooding Literature; Hemisphere Publishing Corporation: New York, 1986. (2) Bennett, K.; Pilling, M. Efficiency benefits of high performance structured packings. 2003. (3) Fair, J. R.; Bravo, J. L. Distillation columns containing structured packings. Chem. Eng. Prog. 1990, 86 (1), 19-29. (4) Gosselink, J. W.; Stork, W. H. J. A Simple Multicomponent Description of the Inluence of the Particle and Pore Size of Commercial Catalysts on the Hydrodesulfurization of Heavy Gas

Oil in (Small) Trickle-Flow Reactors. Chem. Eng. Process. 1987, 22, 157-162. (5) Heibel, A. K.; Heiszwolf, J. J.; Kapteijn, F.; Moulijn, J. A. Influence of channel geometry on hydrodynamics and mass transfer in the monolith film flow reactor. Catal. Today 2001, 69, 153-163. (6) Heibel, A. K.; Kapteijn, F.; Moulijn, J. A. Flooding performance of square channel monolith structures. Ind. Eng. Chem. Res. 2002, (7) Jeong, J. H.; No, H. C. Experimental study of the effect of pipe length and pipe-end geometry. Int. J. Multiphase Flow 1996, 22 (3), 499-514. (8) Krishna, R.; Sie, S. T. Strategies for multiphase reactor selection. Chem. Eng. Sci. 1994, 49 (24A), 4029-4065. (9) Lebens, P. J. M. Development and design of a monolith reactor for gas-liquid countercurrent operation. Ph.D. Thesis, TU Delft, Delft, The Netherlands, 1999. (10) Lebens, P. J. M.; Edvinsson, R. K.; Sie, S. T.; Moulijn, J. A. Effect of entrance and exit geometry on pressure drop and flooding limits in a single channel of an internally finned monolith. Ind. Eng. Chem. Res. 1998, 37, 3722-3730. (11) Lebens, P. J. M.; Stork, M. M.; Kapteijn, F.; Sie, S. T.; Moulijn, J. A. Hydrodynamics and mass transfer issues in a countercurrent gas-liquid internally finned monolith reactor. Chem. Eng. Sci. 1999, 54, 2381-2389. (12) Lebens, P. J. M.; vander Meijden, R.; Edvinsson, R. K.; Kapteijn, F.; Sie, S. T.; Moulijn, J. A. Hydrodynamics of gas-liquid countercurrent flow in internally finned monolith structures. Chem. Eng. Sci. 1997, 52 (21/22), 3893-3899. (13) Reid, R. R.; Prausnitz, J. M.; Poling, B. E. The Properties of Gas & Liquids, 4th ed.; McGraw-Hill: New York, 1987. (14) Sherwood, T. K.; Shipley, G. H.; Holloway, F. A. L. Flooding velocities in packed columns. Ind. Eng. Chem. 1938, 30, 765-769. (15) Stockfleth, R.; Brunner, G. Film Thickness, Flow Regimes, and Flooding in Countercurrent Annular Flow of a Falling Film at High Pressures. Ind. Eng. Chem. Res. 2001, 40, 6014-6020. (16) Suess, P.; Spiegel, L. Hold-up of Mellapak structured packings. Chem. Eng. Process. 1991, 31, 119-124. (17) Vijayan, M.; Jayanti, S.; Balakrishnan, A. R. Effect of tube diameter on flooding. Int. J. Multiphase Flow 2001, 27, 797-816. (18) Wallis, G. B. One-Dimensional Two-Phase Flow; McGrawHill: New York, 1969. (19) Yaws, C. L. Chemical Properties Handbook; McGrawHill: New York, 1999.

Received for review December 7, 2003 Revised manuscript received April 18, 2004 Accepted May 12, 2004 IE034289F