Using ‘The Apprentice’ to Teach a Managerial Economics Course

Studies in Surface Science and Catalysis, Vol. 139 J.J. Spivey, G.W. Roberts and B.H. Davis (Editors) 9 2001 Elsevier Science B.V. All rights reserved.

407

A Comparison of Fischer-Tropsch Synthesis in a Slurry Bubble Column Reactor and a Continuous Stirred Tank Reactor James K. Neathery, Robert L. Spicer, Dennis E. Sparks, and Burtron H. Davis University of Kentucky, Center for Applied Energy Research 2540 Research Park Drive, Lexington, Ky 40511-8410

A Slurry Bubble Column Reactor (SBCR) is a gas-liquid-solid reactor in which the finely divided solid catalyst is suspended in the liquid by the rising gas bubbles. SBCR offers many advantages over fixed-bed type reactors such as: 1) improved heat transfer and mass transfer; 2) isothermal temperature profile is maintained; and 3) relatively low capital and operating cost. Fischer-Tropsch Synthesis (FTS) takes place in a SBCR where the synthesis gas is converted on catalysts suspended as fine particles in a liquid. The synthesis gas flows in a bubble phase through the catalyst/wax suspension. The volatile products are removed with unconverted gases, and the liquid products are separated from the suspension. A gas distributor located in the bottom of the reactor produces the bubbles in the reactor. A considerable interest has been expressed in using the SBCR to carry out FTS particularly for the conversion of stranded natural gas into liquids. Currently, the Center for Applied Energy Research (CAER) is utilizing a Prototype Integrated Process Unit (PIPU) system for scale-up research of the FTS. The purpose of this study was to compare the performance and activity decline of a precipitated Fe/K Fischer Tropsch Synthesis (FTS) catalyst in a revamped slurry bubble column reactor (SBCR) to that of previous CSTR and SBCR runs using the same catalyst and operating conditions. The activity decline measured in the revamped SBCR system was shown to be similar to that of the CSTR experiments. The apparent activity decline in a previous SBCR run was due a transient startup effect from the slurry filtration system.

1. INTRODUCTION The PIPU is a pilot plant system built in the early 1980s for studying a multitude of synthetic fuel/chemical processes. In the mid 1990s, a direct coal liquefaction reactor within the PIPU plant was reconfigured as a SBCR for FTS studies (see Figure 1.). The reactor was originally designed to operate with coarse catalyst pellets (>500 ~tm). Consequently, the reactor system did not contain a wax separation system sufficient for smaller catalyst particles that are typically used in FTS. Therefore, a slurry accumulator and a batch wax filtration system were installed. During the period from 1995-96, attempts to operate the direct liquefaction reactor in a F-T mode were successful in that a clear wax product could be obtained. However, the initial activity observed in the bubble column was about 10-15% less than that of comparable CSTR

408 runs. Also, the rate of conversion decline (and apparent catalyst deactivation) in the SBCR was much greater than that observed in the CSTR. It was hypothesized that the apparent increased deactivation rate in the SBCR was caused by the depletion of catalyst inventory due to the nature of the wax/catalyst separation system. The CAER SBCR plant was overhauled and redesigned to incorporate automatic slurry level control and wax filtration systems. These design changes will allow a more constant inventory of the catalyst to be maintained in the reactor while reducing slurry hold-up in the catalyst/wax separation system. In addition, the wax filtration system was rearranged to accept a variety of filter elements. These additions were meant to enhance the stability of the reactor operation so that long-term tests can be conducted to study catalyst deactivation and attrition under real-world conditions. In the following discussion, we will detail the results and operational experiences the enhanced SBCR system. Objectives of the run were to: 1) test the new slurry level control system; 2) compare the performance of a precipitated Fe/K Fischer Tropsch Synthesis (FTS) catalyst in the enhanced SBCR and a continuous stirred tank reactor (CSTR); and 3) determine the effectiveness of the catalyst/wax filtration system.

2. EXPERIMENTAL All FTS runs were conducted in either CSTR or SBCR systems. Two types of SBCR configurations were used in this study: 1) SBCR (old)- a bubble column with a large volume filtration/settling tank arrangement; and 2) SBCR (new)- a bubble column with a flow through filter arrangement with a small slurry hold-up volume. Activation and synthesis conditions for each reactor configurations are listed in Table 1. A precipitated iron catalyst having atomic composition of 100 Fe/4.4 Si/1K was used for each reactor experiment.

2.1. CSTR Apparatus The one-liter CSTR used in this study has been described in detail in the literature [ 1-2]. The following is a brief description of the reactor system. Catalysts were suspended in Ethyflo 164 hydrocarbon (Ethyl Corp.), which is reported to be a C30 1-decene homopolymer. The initial loading of the catalyst in the slurry was 20 wt%. Hydrogen and carbon monoxide were metered by mass flow controllers to attain a H2/CO ratio of 0.7. The synthesis gas was delivered to the catalyst slurry via a sparger tube located below an impeller blade turning at 750 rpm. The reactor effluent exited the reactor and passed sequentially through two traps maintained at 333 and 273 K. Accumulated reactor wax was removed daily through a tube fitted with a porous metal filter. A dry flow meter was used to measure the exit gas flow rate. The catalysts were activated with syngas with a H2/CO ratio of 0.7. In general, the activation gas flow was started at ambient conditions and the reactor temperature was ramped to the desired set point at a 2 K min -~ rate. After the activation temperature was reached, the conditions were maintained for 24 h. Following the activation treatment, the reactor was brought to FT synthesis conditions: 1.21 MPa, 543 K, 5.0 normal L h-' (g ofFe)-:.

409

Table 1. Operating Conditions: SBCR and CSTR Comparison Experiments SBCR(old)

SBCR (new)

CSTR

CO+H2

CO

CO+H 2

H J C O Ratio

0.7

--

0.7

Gas soace velocity (SL/hr-~ Fe)

5.3

1.0

5.15

Temoerature (~

270

270

270

Pressure ( atm. )

1

12

1

H,/CO Ratio

0.7

0.7

0.7

Gas soace velocity (NL/hr-~ Fe)

5.3

5.2

5.15

Temoerature (~

270

270

270

Pressure (MPa)

1.21

1.21

1.21

3

2.7

750 RPM

Catalyst. Activation: Gases

Synthesis Conditions:

U~ (cm/sec)

2.1. SBCR Apparatus The SBCR apparatus, shown in Figure 1, was originally designed as a direct coal liquefaction reactor. In the current configuration, the bubble column has a 5.08 cm diameter and a 2 m height with an effective reactor volume of 3.7 liters. The synthesis gas was passed continuously through the reactor and distributed by a sparger near the bottom of the reactor vessel. The product gas and slurry exit the top of the reactor and pass through an overhead receiver vessel where the slurry was disengaged from the gas-phase. Vapor products and unreacted syngas exit the overhead vessel, enter a warm trap (333 K) followed by a cold trap (273 K). A flow meter downstream of the cold trap was used to measure the exit gas flow rate. A dip tube was added to the reactor vessel so that the F-T catalyst slurry could be recycled internally via a natural convection loop. The unreacted syngas, F-T products, and slurry exited into a side port near the top of the reactor vessel and entered a riser tube. The driving force for the recirculation flow was essentially the difference in density between the fluid column in the riser (slurry and gas) and that of the dip-tube (slurry only). The dip tube provided a downward flow path for the slurry without interfering with the upward flow of the turbulent syngas slurry mixture. Thus, to some degree, back mixing of the slurry phase and wall effects in the narrow reactor [3] tube were minimized. Based upon the analysis of the previous SBCR runs (in 1995-96), several more design changes were carried out to the SBCR system to increase the conversion stability. An automatic level controller was added to the overhead slurry/gas separation tank. This insured a constant inventory of catalyst particles was being maintained in the reactor vessel if the superficial gas velocity within the column was constant.

410

Overhead Receiver ,t

• - C-h i l l•e d W a t e r

Q-

Ill J

n

Gas Samples to vent and/or

_~ Argon Purge ~

I

I

Filter ~

=~ Slurry and \ Gas Exit

Wax

-I r Control Valve m ~ u ry Downcomer to Dip-Tube

Hot Trap/ Clear Wax Storage

Warm

Trap

Cool Trap

Slurry Bubble Column Reactor

Syngas Inlet

Figure 1. Schematic of the SBCR pilot plant system.

Originally, the overhead separator vessel was designed to enhance settling of the catalyst particles. Thus, slurry to be filtered was extracted near the top of the vessel where the catalyst concentration would be lower than that near the bottom. Unfortunately, this approach required a large hold-up volume of slurry outside the reactor (greater than the reactor volume itself). Decreasing the volume of the overhead vessel from 18 to 4 liters lowered slurry hold-up outside the reactor. A sintered metal filter tube was moved to the liquid down comer below the overhead separation vessel. Currently, the filter is a flow-through device having a sintered metal tube in a shell. Filtered wax was extracted radially through the tube while slurry flows downward in the axial direction. The shear force of the axial slurry flow prevented excessive caking of the catalyst around the filter media. Filtered wax was metered into a storage tank through a letdown valve operated by the overhead liquid level controller. Pressure drop across the filter media can be varied manually by varying the wax storage tank pressure. The filter assembly was configured such that the filter media could be replaced on-line, without aborting or interrupting the reactor run. The level or volume of the slurry within the receiver was continuously monitored by measuring the differential pressure across the height of the vessel. Argon was purged through each of the pressure legs to keep the lines free of slurry. Slurry volume within the receiver was controlled to be no more than 1.3 liters by removing wax from the reactor system via the level

411

control valve. The unfiltered slurry flowed back to the reactor via a natural convection loop through a dip-tube exiting near the bottom of a reactor. In preparation for catalyst activation, the SBCR was filled with 2.8 liters (N75% of the reactor volume) of slurry consisting of 20-wt% iron catalysts and C30oil. An additional 1.3 liters of the C30 oil was isolated in the overhead separation vessel. The reactor was pressurized with flowing CO gas at 175 psig (12 atm) while the slurry temperature was increased to 543 K at a 50 K/hour rate. Once the reactor temperature stabilized, the exit gas was periodically monitored for CO2 to observe the progress of activation. During the activation period, the down-comer leg from the overhead vessel to the reactor was valved-off so that the catalyst remained isolated inside the reactor. Likewise, the C30 oil in the overhead vessel did not mix with reactor catalyst during activation. After the catalyst had been activated (~-24 hours), hydrogen gas flow was phased in with the CO feed gas. Once the desired gas space velocity had been attained, the down-comer valve used to isolate the C30 oil in the overhead vessel was opened to allow circulation between the reactor, riser and down-comer legs. Once the C30 oil became mixed with the activated catalyst slurry and the reactor temperatures were stabilized, CO, H2, and syngas conversions were calculated at least once a day to monitor the reactor performance.

2.2. Gas/Liquid Analysis The composition of the exit gases for CSTR and SBCR runs were determined by GC techniques. The condensed liquid phases were sampled on a 24-hour basis. The aqueous phase was analyzed for water and oxygenates using a GC fitted with Porpack Q column. The oil and wax phase samples were combined according to their mass fraction, O-xylene was added as an internal standard, then this sample was analyzed for hydrocarbons by GC with a DB-5 column.

3. DISCUSSION OF RESULTS

3.1. SBCR Shakedown/Conversion Comparisons between CSTR and SBCR runs One of the objectives of the shakedown run was to compare the performance of the enhanced SBCR performance with the previous SBCR configuration. It was anticipated that the modified SBCR system performance, in terms of catalyst deactivation, would be comparable to that of the CSTR experiments. The shakedown run/activation conditions for the enhanced SBCR system along with the comparison SBCR and CSTR conditions are listed in Table 1. The CO gas conversions versus time-on-stream for the SBCR and CSTR systems are displayed in Figure 2. The CO conversion for the enhanced SBCR with level control reached a maximum of 78% after 72 hours time-on-stream (TOS). After this catalyst initiation period, the gas conversion started to steadily decline to about 72% after 192 hours TOS. Carbon dioxide selectivity stabilized to 45% while the methane selectivity averaged 4%.

412

~

SBCR-new 9t'=0 .t'=0

r

.o r 9 > to O

o

O

CSTR

60 t'=0

50 40

9

CR-old

3o 2o

Time where -rF-T

0

50

is at maximum for baseline activity.

100

150

200

250

T i m e on Stream, hrs

Figure 2. CO conversion vs. TOS for the CSTR and SBCR configurations.

Significant differences in conversion between the two SBCR configurations are apparent. The enhanced SBCR (run # SBCR-new) conversion continued to increase after +70 hours. The older SBCR (run # SBCR-old) conversion continued to drop at a significant rate after activation and was consistently lower than that of the CSTR and enhanced SBCR. Slurry back-mixing in the enhanced SBCR is significantly reduced by the addition of the down-comer/dip-tube flow path; consequently, the gas and liquid phases likely exhibited more plug-flow behavior. Thus, for a given space velocity, the enhanced SBCR should yield a higher conversion than that of a CSTR [4]. Differences in conversion between the enhanced SBCR and CSTR reactor types may also be caused by the dissimilarity of heat and mass transfer phenomena. In addition, the relatively large L/D ratio of the SBCR may also contribute to its plug-flow characteristics. 3.2. Catalyst Deactivation Rate Comparisons Catalyst deactivation rates were compared between the different reactor configurations using the activity function defined as: a(t) =

r FT ( t ' ) r FT ( t ' = O)

~

X co ( t ' )

(1)

X co ( t ' = O)

Where t' is the time after attaining the maximum total reaction rate or conversion. The maximum reaction rate was identified for each conversion curve, as shown in Figure 2. The relative activity functions were calculated from the maximum conversion and plotted in Figure 3 versus the relative t' time-scale. In this fashion, each of the deactivation rates could be compared on an equal basis, independent of the conversion levels. The deactivation rates calculated for the SBCR-new and CSTR cases followed a linear zero order fashion with decay constants of 0.0130 and 0.0142 day -1, respectively. The apparent catalyst activity decline of the SBCR-old appeared to have two distinct rate deactivation periods:

413

SBCR-new, >, ._ >

0 95

"o-

0 90

o ~,

085

d a y -1

k d = 0 0142

d a y "1

"..0

CSTR, ""..... "O.

.

SBCR-old

Initial act=voty d e c h n e

""..period, k d = 0 . 0 6 2 4 0 d a y -1 ..

0 80

"0 9...

~.

k d = 0.013

07s

Secondary

"'~.....~ . ~

m

period,

0.01560 d a y "1 "o

0

20

t', t i m e

o []

0

SBCR CSTR SBCR

40

60

after

maximum

80

(new) Enhanced

100

120

reaction

140

rate

160

180

200

(hours)

desogn wath slurry level control.

(old) O l d e r d e s i g n with large filter h o l d u p v o l u m e

Figure 3. Activity decline vs. relative TOS for the CSTR and SBCR configurations.

a relative rapid decay of 0.0624 day-1 followed by another linear decay period with a slope of 0.0156 day -1. The first decay period was apparently due to a transient effect from the accumulation of catalyst within the large overhead vessel and filtering system. Thus, the decrease in reaction rate and conversion during this initial decay period was caused by a steady increase in the space velocity as catalyst was removed from the reactor. Once the SBCR-old system reached steady state, the activity decline rate was comparable to the other reactor configurations.

4. CONCLUSIONS Tight control of catalyst inventory within SBCRs must be maintained in order to quantify activity decline, especially for small pilot plant systems. Transient problems with previous SBCR experiments were caused by a maldistribution of catalyst between the reactor and slurry filtration system. The level indication/control system installed in an enhanced SBCR was robust and effective in maintaining a steady inventory of catalyst slurry in contact with the gas-phase. Measured deactivation rates in the enhanced SBCR system were comparable to that of CSTR experiments under similar conditions.

414 ACKNOWLEDGEMENT This work was supported by the U.S. Department of Energy and the Commonwealth of Kentucky.

NOMENCLATURE a(t) kd rF_T SV t' TOS Ug Xco

Catalyst activity function First order deactivation rate constant, day -I Rate of Fischer-Tropsch synthesis, mole s-1 Fe-g-1 Gas space velocity, N L h-1 (Fe.g)-1 Time after maximum CO conversion, hours Time-on-stream, hours Superficial gas velocity based on inlet reactor conditions, cm s-~ CO conversion

REFERENCES

1. O'Brien, R., L.Xu, D. Milburn, Y. Li, K. Klabunde, and B. Davis, Top. Catal. 1995,2, 1. 2. R. J. O'Brien, L. Xu, R. L. Spicer, and B. H. Davis, Energy & Fuels, 10 (1996) 921. 3. Marretto, C. and R. Krishna, Modeling of Bubble Column Slurry Reactor for FischerTropsch Synthesis , Catalysis Today 52 (1999) 279-289. 4. Fogler, H., Elements of Chemical Reaction Engineering, 1stEdition, pgs. 273-278, PrenticeHall, Englewood Cliffs, New Jersey 07632, 1986, ISBN 0-13-263476-7.