You are on page 1of 11

Pergamon

Chemical Engineenng Science, Vol, 49, No. 24A, pp. 4615...4625, 1994
Copyright 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0009-2.509/94 $7.00 + 0.00

0009-2509(94) 00265-7

FISCHER-TROPSCH SYNTHESIS IN A STIRRED TANK SLURRY R E A C T O R


D R A G O M I R B. BUKUR,* LECH N O W l C K I t and X I A O S U L A N G Department of Chemical Engineering, Kinetics, Catalysis and Chemical Reaction Engineering Laboratory, Texas A&M University, College Station, TX 77843-3122, U.S.A.

(Received 25 May 1994; accepted for publication I0 August 1994)


Abstract--Three precipitated iron catalysts were evaluated in a stirred tank slurry reactor to determine their performance during Fischer-Tropsch synthesis under steady-state conditions_ High syngas conversions (>80%) were achieved in all three tests, and the catalyst deactivation rates were moderate (1.3-2.1% per day). Selectivities of all three catalysts were fairly stable with time (up to 560h of continuous testing). Methane selectivity (2.1-3.7 wt% of total hydrocarbons) and gaseous hydrocarbon (C2~4) selectivity (9--15 wt%) were very low, whereas the selectivity of C~ hydrocarbons was high. Methane and gaseous hydrocarbon selectivities obtained in these tests are comparable to those obtained in the two most successful slurry bubble column reactor tests. The activity of one of these three catalysts (100 Fe/5 Cu/6 K/24 SIO2, in parts per weight) was higher than that of any other known iron Fischer-Tropsch catalyst developed for slurry phase operation. to coking is decreased. This, in turn, allows much higher conversions per pass, minimizing synthesis gas recycle, and offers the potential to operate with CO-rich synthesis gas feeds without the need for prior water-gas shift. Due to the simpler reactor design, capital investment in a slurry phase FT reactor is expected to be substantially smaller than in conventional fixed or circulating fluidized bed systems (Fox et al., 1990; Gray et al., 1991). The Rheinpreussen demonstration plant performance (K61bel et al., 1955; K61bel and Ralek, 1980) is regarded as the most successful bubble column slurry reactor (BCSR) operation to date. All subsequent attempts to reproduce or exceed the Rheinpreussen demonstration plant performance have been unsuccessful (Mitra and Roy, 1963; Farley and Ray, 1964; Kuo, 1983, 1985). In Rheinpreussen's test, a high single pass synthesis gas conversion of 89% was achieved, and the bulk catalyst activity (space-time yield) was high. The yield of Cs--CI1 hydrocarbons (gasoline fraction) was 53.6 wt%, and was accompanied by low yield of methane + ethane (3.2 wt% only). This product distribution is very unusual and could not be reproduced in any subsequent studies. In all other studies (regardless of the reactor type), low methane yields were accompanied by high yields of Ci2* products. The maximum yield of transportation fuels, Cs-Cls hydrocarbons, based on classical AndersonSchulz-Flory (ASF) distribution is about 66% (corresponding to chain growth probability factor of 0.816). This yield was nearly achieved in the Rheinpreussen BCSR. An alternative way to increase total transportation fuel yield is through maximization of the reactor wax yield, while simultaneously minimizing the light gas yield. Reactor

INTRODUCTION

Several technologies are currently available or are under development for conversion of coal-derived synthesis gas to liquid transportation fuels or fuel precursors. Technologies that have been commercially proven, or that are close to commercialization, include Fischer-Tropsch synthesis (FTS), methanol synthesis, and Mobil's methanol to gasoline (MTG) process. Of these technologies, Fischer-Tropsch (FF) hydrocarbon synthesis produces the widest slate of products and has been in operation for the longest period. F F hydrocarbon synthesis was first developed and practised in Germany during the 1930s and 1940s using cobalt catalysts_ Subsequently, the process was commercialized on a large scale by SASOL in South A f r i c a . The SASOL process includes both tubular fixed bed and circulating fluidized bed operation and uses promoted iron catalysts (Dry, 1981)_ Recently, a conventional fluidized bed reactor, without circulation of the catalyst, and a bubble column slurry reactor were constructed and placed on stream at SASOL in 1989 and 1993, respectively. In the late 1940s, slurry phase Fischer-Tropsch technology was developed in Germany (Krlbel and Ralek, 1980). Slurry processing provides the ability to more readily remove the heat of reaction, minimizing temperature rise across the reactor and eliminating localized hot spots. As a result of the improved temperature control, yield losses to methane are reduced and catalyst deactivation due

*To whom correspondence should be addressed. tPresent address: Institute of Chemical and Process Engineering, E6d~ Technical University, 90-924 E6dz', Poland.

4615

4616

DRAGOMIRB. BUKUR el al.

wax can be easily upgraded, via conventional organic phases. Oxygenates (alcohols, ketones and organic acids) in the aqueous phase were analyzed processes, into high quality diesel fuel (Kuo, 1985; Shah and Fullerton, 1990). using a 1.8 m Porapak Q column and flame ionizaAt Texas A & M University (TAMU), the tion detector (FID). The amount of water present authors' group have been working on development in the aqueous phase was determined by a K a r l of improved iron FT catalysts since 1987. In this - Fisher titration method. Hydrocarbons and oxypaper, results are presented from process evaluagenated compounds in the organic phase were tion studies of three iron FT catalysts in a stirred quantified using a 30 m fused silica capillary column tank slurry reactor (STSR). The catalysts tested and an FID. The high molecular weight hydrocarbons (wax) were developed with the objective of minimizing production of methane and gaseous hydrocarbons, accumulate in the reactor and have to be removed while maximizing the yield of high molecular periodically to prevent the reactor from filling with weight hydrocarbons (wax). The tests were conwax, and to obtain complete product distributions and accurate mass balance closures. The wax was ducted over long periods of time to evaluate the removed from the reactor through a porous cylincatalysts under steady-state conditions, and to obtain information on catalyst deactivation rate and drical sintered metal filter with a nominal pore size of 0.5/~m, and analyzed using a wide-bore fused changes of product selectivity (hydrocarbon distribution, total olefin content, 2-olefin content) silica capillary column (10m x 0 . 5 3 m m ) and an with time-on-stream (TOS). A comparison of the FID. performance of catalysts synthesized in the authors' The reactants and noncondensable products leavlaboratory with that of the state-of-the-art iron ing the ice traps were analyzed on an on-line GC catalysts is also presented. (Carle A G C 400) equipped with multiple columns, and both flame ionization and thermal conductivity detectors. Further details on the reactor and product analysis systems can be found elsewhere EXPERIMENTAL (Bukur et al., 1989, 1990a,c). Reactor system The reactor used in this study was a 1 dm 3 stirred Catalysts tank reactor (Autoclave Engineers). The system Catalyst preparation involved three steps: prepaprocess diagram is shown in Fig. 1_ The feed gases ration of the iron-copper precursor, incorporation (H2 >99_5% purity, and CO >99.3% purity) or a of binder/support (silicon oxide), and finally potaspremixed gas passed through a series of oxygensium impregnation. The preparation procedure has removal, alumina, and activated-charcoal traps to been described in detail previously (Bukur et al., remove 02, iron carbonyls, water, and hydrocar1989, 1990b). In brief, the catalyst precursor was bon impurities. The feed gas flow rate was controlcontinuously precipitated from a flowing aqueous led using calibrated mass flow controllers, and the solution containing iron and copper nitrates at the feed was introduced into the reactor below a flat desired Fe/Cu ratio, using aqueous ammonia_ Imblade impeller, used to agitate the slurry. After pregnation with SiO2 binder/support was accomleaving the reactor, the exit gas passed through a plished by addition of an appropriate amount of high pressure trap, which was maintained at about dilute (26 wt%) K2SiO3 solution to undried, reslur150"(2, to condense high molecular weight proried Fe/Cu co-precipitate. ducts. After releasing pressure through a back After a vacuum drying step, the potasssium pressure regulator, the gas passed through a low promoter was added as aqueous KHCO3 solution pressure ice trap to collect any remaining condensvia an incipient wetness pore-filling technique. The ables. The flow rate of the tail gas exiting the dried catalyst was calcined in air at 300C for 5 h, system was measured frequently with a soap film and then crushed and sieved to a diameter between flow meter and a wet test meter. 45 and 53/.Lm (270/325 mesh)_ A purified nDuring mass balance periods, typically 6--8h, octacosane (c. 250 g) was used as a slurry liquid liquid products were allowed to accumulate in high medium in all tests, and a catalyst was added in and low pressure mass balance (steady-state) traps. amounts to give 3-7 wt% slurry. The catalysts were At the conclusion of the mass balance period, flow activated in situ prior to FTS with hydrogen at was directed to waste traps placed in parallel with 220-250C for 2-4 h. the mass balance traps, and liquid products from Three catalysts with nominal compositions: the mass balance traps were collected and weighed. 100Fe/3 Cu/4 K/8 SiO2 (Run SB-0931); 100Fe/3 After startup, and following any change in process Cu/4 K/16 SiO2 (Run SB-0261) and 100 Fe.5 Cu/6 conditions, the reactor was allowed to operate K/24 SiO2 (Run SB-1931) were employed in this undisturbed for at least 40h in order to achieve study. The catalyst compositions are given on a steady-state conditions before the next mass balmass basis. The test duration was 560-720 h. The ance was performed. All products collected in the reaction pressure and/or temperature were varied steady-state traps were analyzed by gas chromain all three tests; however, results from these tography after physical separation into aqueous and periods are not reported in this paper.

Fischer-Tropsch synthesis in a stirred tank slurry reactor

4617

'rr
SAMPLE PORT. '
STIRRER[

PURGE GAS

'~,,,
'1"

IW
13.

VENT

rr w
I-

ONLINE
GC

w O
_.J I1

-g
ICE AT

_..I

< x'~ <rr

-r I.---

HEATED BYPASS

MASS BALANCE WASTE TRAP TRAP

HIGH R E TRAP

I
S

WASTE S U R

BACK PRESSURE REGULATOR


Fig. 1. Schematic of stirred tank slurry reactor system.
RESULTS AND D I S C U S S I O N

Catalyst activity
Variations in (He + CO) conversion with timeon-stream (TOS) for the three catalysts are shown in Fig, 2. Process conditions in all tests were: 260C, 1.48 MPa, Ha/CO = 0_66--0.70, whereas gas space velocity varied between 2.2 and 3.4 NI/g-Fe/h (where Nl/h denotes volumetric gas flow rate at standard temperature and pressure, i.e. 0C and 1 bar). Three types of catalyst behavior were

observed in these tests. Catalyst with nominal composition 100 Fe/3 Cu/4 K/8 SiO2 (SB-0931) had initially high conversion ( - 8 8 % ) , but its activity decreased gradually with TOS. During the test of catalyst 100Fe/3 Cu/4 K./16 SiO2 (SB-0261) the ( H 2 + C O ) conversion was initially about 81%, decreasing to 76% at 150h on stream_ Between 160 h and 240 h the catalyst was tested at 263C (results not shown in Fig. 2), and it continued to deactivate. Upon returning to the baseline conditions, the activity became stable as evidenced by a

4618

DRAGOMIR B. B U K U R et al.

100

'

'

'

'

'

'

'

'

'

'

'

'

'

'

[] o 90
v

Catalyst: 100 Fe/5 Cu/6 K/24 SiC 2, SV=2.3-3.4 NI/g-Fe/h Catalyst: 100 Fe/3 Cu/4 K/16 SiC 2, SV=2.4 NI/g-Fe/h Catalyst: 100 Fe/3 Cu/4 K/8 SiC 2, SV=2.5 NI/g-Fe/h

SV=3.0
tO
(D > tO

80

70
0 U + "1"
N

0 0

0(-~

60

T = 260C P = 1.48 MPa

H /CO = 2/3
Z

50 0

l O0

200

300

400

500

600

Time-on-Stream (h)
Fig. 2. Variations in (H2 + CO) conversion with time-on-stream in STSR tests of three precipitated iron catalysts.

500

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

400
13..

-"- Catalyst: 100 Fe/3 Cu/4 K/8 SIC2, (SB-0931) o Catalyst: 100 Fe/3 Cu/4 K/16 SiC 2, (SB-0261) [] Catalyst: 100 Fe/5 Cu/6 K/24 SiC 2, (SB-1931)

~r
(D
U-

300

&
O

E E
v v

200

n
1 O0
T = 260C P = 1.48 MPa H /CO = 2 / 3
2 , L , , i i i , , I , , = t l , , , , I , , , , I I

] O0

200

300

400

500

600

Time-on-Stream (h)
Fig. 3. Apparent first-order reaction rate constant as a function of time.

Fischer-Tropsch synthesis in a stirred tank slurry reactor nearly constant value of (H2 + CO) conversion (66--68%) between 240 h and 530 h on stream. In the test of catalyst 100 Fe/5 Cu/6 K/24 SiO2 (SB1931), gas space velocity was decreased twice to obtain higher conversions_ At a constant gas space velocity (3.4 and 3.0Nl/g-Fe/h) the activity increased slightly with time, up to about 390h. During the last portion of the test (400-500h) at gas space velocity of 2.3NI/g-Fe/h, the catalyst exhibited some deactivation. Activities of the three catalysts are compared in Fig. 3, in terms of values of apparent first-order reaction rate constant. The latter was calculated from experimental data, assuming perfect mixing and negligible interphase and intraparticle transport resistances:

4619

k = SV

x XH2+C O (mol H 2 + CO/g-Fe/h/MPa).

22.4 x PH2

(1)
The catalyst used in Run SB-1931 was the most active, whereas the catalyst used in Run SB-0261 was the least active. All three catalysts deactivated with time, but the rate of deactivation was the most rapid during the first 150-200 h on stream. The average catalyst deactivation rate was estimated using the following expression:

provides a conservative estimate, since the deactivation rate is based on the initial catalyst activity. In some cases the catalyst activity goes through a maximum (induction period) before it starts decreasing or leveling off. Values of the (H2/CO) usage ratio, UR, and partial pressure quotient, Kp = Pco2PH2/ PcoPH2o, provide relative measures of the watergas shift (WGS) activity (lower values of the usage ratio or higher values of Kp correspond to higher WGS activity) and are summarized in Table 1. The catalyst used in Run SB-0931 had the highest WGS activity (UR = 0.54--0.61, Kp = 40--60), whereas the catalyst used in Run SB-1931 had the lowest WGS activity (UR = 0.58-0.62, Kp = 10-24). This is consistent with previous studies (Bukur et al., 1990b) where it was shown that the WGS activity decreases with increasing amounts of silicon oxide. However, in all three runs the WGS activity was rather high and this reaction was close to the equilibrium (the equilibrium value of Kp at 260C is about 71, cf_ Newsome, 1980).

Catalyst selectivity
Selectivities of methane and (C1 + C2) hydrocarbons obtained in tests of these three catalysts are shown in Fig. 4. In tests SB-0931 and SB-1931, selectivities of methane and (C~ + C2) hydrocarbons increased gradually with TOS, whereas in test SB-0261, these two selectivities passed through a maximum at about 150 h. Methane selectivity of all three catalysts was less than 3% (mol% carbon basis), whereas (Ci + C2) selectivity was less than 7% throughout the entire test. Changes in hydrocarbon product distributions as a function of TOS for Run SB-0261 are shown in Fig. 5. In this figure, cumulative values are shown for the lumped product distribution of selected

DR = [1 - k(t)/k(to)] x 100/t

(2)

where DR = deactivation rate (%/day); k(t) and k(to) = apparent reaction rate constants at time t and to, respectively; t = time in days; to = time at which the first mass balance was conducted (usually, after about 40 h on stream). Deactivation rates during 500 h of testing were: 1.3, 1.6 and 2.1% per day, for Runs SB-1931, SB-0261 and SB-0931, respectively. This procedure

Table 1_ Results from slurry reactor tests of precipitated iron catalysts Run ID: Catalyst composition (Fe/Cu/K/SiO2): Activity parameters: (H2 + CO) (%) CO (%) UR k (mmol/g-Fe/h/MPa) Nm a [(H 2 + CO)/kg-Fe/h] SB-0931 100/3/4/8 70-88 75-93 0.54-0.61 170-320 1.7-2_1 40-60 2.1 200 188 5.3-7.1 2.5-3.4 11-13 16-22 65-70 SB-0261 100/3/4/16 67-83 71-90 0_53-0.60 160-255 1.5-2.0 17-40 1.6 205 194 5.0-6.4 2.1-2.8 9-13 12-16 69-77 SB-1931 100/5/6/24 74-81 78-86 0.58-0_62 260-350 1.7-2.6 10-24 1.3 210 199 5.4-8.2 2.5-3.7 10-15 13-20 62-74

Kp = Pco2P~J Pco PH2o DR (%)


Selectivity/yield: g HC/Nm3 (H 2 + C O ) reacted g C3 + Nm3 (H2 + CO) reacted (C, + C2) (wt%) CH4 C2--C4 C~CI, C,2,

Process conditions: T = 260C, P = 1.48 MPa, H2/CO = 0.66--0.70, SV = 2.3-3.4 Nl/g-Fe/h.

4620
.... I .... I ....

DRAGOMIRB. BUKURet
I .... I .... I .... I ....

al.
I .... I .... I .... I ....

A |

z~ Catalyst: 100 Fel3 Cu/4 K/8 SiO2 n Catalyst: 100 Fe/5 Cu/6 K/24 SiO2 o Catalyst: 100 Fe/3 Cu/4 K/16 SiO2

E 3
,i,-,

to

O r) 2 ',d" "I" (3

0 )
i .... A I I .... I .... I .... I .... I .... I .... I .... I .... I .... I ....

a
O

[]

-$
03

G)

C',J 4 (.3

,Z3
, , . I , , F , I . . . . 1 . . . . I . . . . L . . . . I . . . . I . . . . I . . . . [ . . . . I . . . .

50

100

150

200

250

300

350

400

450

500

550

Time on Stream (h) Fig. 4. Selectivities of methane and (C~ + C_~)hydrocarbons as a function of time. product groups: methane, CI-C4, and C~-C~. Differences in numerical values of different groups provide information on mass percentages of methane, C2---C4 (light gases) and Cs--Cll (gasoline fraction), whereas the mass percentage of C~2 (diesel and wax) hydrocarbon products is given by 1 0 0 - % (Ci-Ci 1). The selectivities of low molecular weight hydrocarbons (CH4, C2-C4 and Cs--Cll) passed through a maximum, whereas the C~2+ selectivity had a minimum value at about 150 h on stream. The initial increase in selectivities of lower molecular weight hydrocarbons coincides with catalyst deactivation and high conversions during the first 150h of testing (Fig. 2). After the (H2 + CO) conversion had stabilized at about 67% (300-550 h), the hydrocarbon selectivities reached their steady-state values_ It appears that both the catalyst activity and selectivity were changing during the first 250 h on stream as a result of changes in the state of the catalyst (bulk and surface composition), and then became stable when the catalyst had stabilized. In the other two tests, hydrocarbon distribution shifted slowly towards lower molecular weight products. Ranges of values for different groups of products for all tests are summarized in Table 1. A typical carbon number product distribution is shown in Fig. 6, in the form of an AndersonSchulz-Flory (ASF) plot. Figure 6(a) shows results based on the analysis of all the products collected (gas, liquid phase and wax withdrawn from the reactor) in Run SB-0261 at about 214 h on stream, whereas Fig. 6(b) shows results from the analysis of the reactor wax only. The data in Fig. 6(a) were fitted by a three-parameter model of Huff and Saterfield (1984), and numerical values of model parameters (two chain growth probability factors, a~ and t~2, on type 1 and 2 sites, respectively, and a fraction of type 1 sites----/3) are shown in the figure. Positive deviations from the modified ASF distribution were observed in the C~-C22 carbon number range, possibly due to hydrocarbon cracking. This type of behavior was also observed with the other two catalysts. Also, the experimental values of C~ and C2 products were lower than the model predictions_ The chain growth probability factor for products withdrawn from the reactor was found to he 0.94 [Fig. 6(b)], which is in good agreement with the value obtained from the three-parameter model (t~2 = 0.92). Primary products of Fischer-Tropsch synthesis are 1-alkenes, which are partly converted to

Fiscber-Tropsch synthesis in a stirred tank slurry reactor

4621
I

50

'

'

'

'

'

'

'

'

"

'

'

'

'

'

Catalyst: 100 FoI3 Cul4 W16 Si@2 {SB-0261)


40
4i~

t.O m -I

30

~"4~"~ ~~"*'~,~,.........,...~

C 1 - C l

t~
tO
t~ U

20

t~

01-0 4
I0
A3

o
"o -r
, ~. , , , I , " , . . I ~ . , . , JI , . .m, , . I ~. , , . I j_ .

1 .

100

200

300

400

500

;00

Time-on-Stream

(h)

Fig. 5. Change in cumulative hydrocarbon product distribution with time-on-stream (Run SB-0261).

paraffins, and 2-alkenes via secondary reactions. The olefin content (selectivity), defined as 100 x total olefin/(total olefin + paraffin), is a measure of hydrogenation activity. Variations of olefin content with carbon number (up to C15) and TOS are shown in Fig. 7 (Run SB-0931). The olefin content (particularly C2 and C8+ hydrocarbons) increases with TOS_ This may be attributed to changes in the catalyst composition with time, as well as the decrease in conversion (lower conversion favors primary reactions). At a given time, all curves have the same shape, i.e. the ethylene content is lower than that of other gaseous hydrocarbons, and the olefin content decreases with carbon number. This is generally ascribed to higher reactivity of ethylene and its readsorption to the catalyst surface resulting in secondary hydrogenation and/or incorporation into growing chains. High molecular weight olefins have higher adsorptivity and higher solubility, which increases their residence time in the reactor and results in a higher probability for secondary hydrogenation (Dictor and Bell, 1986; Herzog and Gaube, 1989). The olefin content did not vary much with time in the other two tests (Runs SB-0261 and SB-1931). Figure 8 illustrates effects of TOS and gas space velocity on 2-alkene selectivity, defined as 100 x (2alkene)/(1-alkene + 2-alkene), as a function of carbon number (Run SB-1931). In all cases the fraction of 2-alkenes increases with carbon number. Longer residence time of high molecular weight 1-aikenes (see above) favors secondary isomerization reactions. At constant gas space velocity

(1.6NI/g-cat/h) there is no effect of time on 2alkene selectivity (208 vs 351h). However, the secondary isomerization reactions increase with decrease in gas space velocity, i.e. with increase in conversion.

Comparison with other iron FTS catalysts


Comparison of performance of the catalyst with nominal composition 100 Fe/5 Cu/6 K/24 SiO 2 (Run SB-1931) with other catalysts tested in the authors' laboratory, and elsewhere, is presented in Table 2. Performance of this catalyst is very similar to that of the best Mobil catalyst in the wax mode of operation (Kuo, 1985). The latter catalyst was tested in a bubble column slurry reactor (BCSR), the behavior of which approaches that of a plug flow reactor. An apparent first-order reaction rate constant evaluated at a common temperature of 260C was chosen as a measure of catalyst activity. Data obtained at other reaction temperatures were converted to 260"C, assuming the activation energy of 90 kJ/mol (Zimmerman and Bukur, 1990). STSR was modeled as a perfectly mixed flow reactor, whereas the rate constant from a BCSR was estimated using a model which assumes that the gas phase is in plug flow and the liquid is unmixed (Bukur, 1983), The range of values is given for data from BCSR tests, due to uncertainties in the magnitude of the gas-liquid mass transfer resistance. The minimum and maximum values were obtained by assuming that the fractional mass transfer resistance is 0.15 and 0.40 of the total

4622

DRAGOMIRB. BUKURet

al.

(a)

lOo
--1 ' l' ' ' ' C I ' ' ' 1 a O 0 ~ ' F ' e t 1 13 ' t / '

Cu/4 I</16 Si02


1 a' ' ' ' Il ' [ ' 'y ' ' [

I s'

'

'

'

t'

'

'

':

'

'

'

'

'

'

'

'

'

'

'

'

10"

lO'
0)

10"

H2/CO = 0.70
S V . =. , 1 . 3 9 . N . l / g - c a t . : h ....... 24 Carbon (b) 10" : ....... , ....... , ....... , ....... 32 number , ....... , ....... , ........ , ....... 40 , ...... 48

:1 .I
58

10"'
0 8 16

, .......

Catalyst: 100 Fe/3 Cu/4 K/16 Si02


~ (:x= 0.94

10"
o

~ 10.,
0
10.s T = 2 6 3 " C

o~

P = 1.48MPa H2/CO SV
10 -e ......

= 0.70
NI/g-cat*h , ....... , ....... , ....... , ....... , .......

= 1.39
, .......

16

24 Carbon

32 number

40

48

56

Fig. 6. Carbon number product distribution in Run SB-0261 at 214 h on stream of (a) all products; (b) reactor wax (open circles: data points excluded from parameter estimation).

Fischer-Tropsch synthesis in a stirred tank slurry reactor

4623

Table 2. Catalyst performance in slurry bed reactors Catalyst designation: Run ID: Process conditions: Temperaure (*C) Pressure (MPa) S V (Nl/g-Fe/h) Feed (H2/CO) TOS (h) % (H= + CO) conversion Usage ratio Rate constant at 260"C (rel) Hydrocarbon selectivities (wt%): CH,t C,-C4 C5---C] i
C j2+

TAMU SB-1931 260 1.48 2.2-3.4 0.66--0_69 40--520 74-82 0.58--0.62 100 3.0 10.5 16.0 70.5 6.1 1.7-2.6 203--218 199 0.35-0.54

Ruhrchemie SA-0888 250 1.48 3.8 0.67 0-343 40-43 0.74-0.84 88 4.7 20.6 23.2 51.5 10.8 1.7 186-217 179 0.34

UCI SA-3391 265 2.10 2.4 0.7 227-322 72-74 0.58.--0.62 40 4.4 16.5 23_6 55.5 9.2 1.8 204 185 0.37

UOP (1991) 265 2. I0 2_4 0.7 15-370 69 0.57 33 4.5*

Mobil t CT-256-13 257 1.48 2.3 0.73 475 82 0.59 49-84 2.7 11.1 18. I 68.1 5.6 1.9 206 195 0.39

K61bel et al. t (1955) 268 1.20 3_1 0.67

89 0_63 52-74 3.2 I 31.3 53.6 11.9 6.8 2.8 178 166 0_49

C, + C2 Yields: Nm3/kg-F'e/h g HC/Nm 3 (Hz + CO) g C3+/Nm 3 (H2 + CO) g HC/g-Fe/h *Slurry bubble column reactor test. *(mol%). ~CH4 + C2H6. I(m01%) of CH4 + C2H6.

5.84

I O0

'

'

Catalyst: l O0 Fe/3 Cu/4 K/8 SiO 90


A

109 h 244 h 554 h

.m

80

[] []
A

[]

t-

70 SB-0931 T : 260C 60
P = 1 . 4 8 MPa SV = 1.5 N I / g - c a t / h H /CO = 0 . 6 6
2

[]
A

50

6
Carbon

10 Number

12

14

Fig. 7_ Change of olefin selectivity with carbon number and time-on-stream in Run SB-0931.

4624

DR.AGOMIR B, B U K U R

et al.
I I I

50

'

SB-1931 T = 260"C

Catalyst: 100 Fe/ 5 Cu/6 K/24 SiO


MPa

2
V

40
4=1 ,m

P = 1.48 H/CO
Z

= 0.69

30

tJ o3 e"
Nk=

20

98 h, 162 h, zx 351 h, 496 h,


[]

1.8 1.6 1.6 1.2

NI/g-cat/h NI/g-cat/h NI/g-cat/h NI/g-cat/h


zx [] []

[]

O N

10

[]

6 Carbon

10 Number

12

14

16

Fig. 8. Variation of 2-olefin selectivity with carbon number and time-on-stream in Run SB-1931.

resistance, respectively. The assumed range of fractional mass transfer resistances is fairly conservative and is consistent with the analysis presented by Deckwer et al. (1981), as well as with more recent measurements of hydrodynamic parameters (Gas holdups and bubble sizes) conducted in the authors' laboratory (Patel et al_, 1990). The apparent rate constant obtained in Run SB-1931 at 350 h on stream, 317 mmol/g-Fe/h/MPa, was chosen as a reference value (relative value of 100). The results summarized in Table 2 show that catalyst 100Fe/5 Cu/6 K/24 SiO2 is more active than Mobil's catalyst used in Run CT-256-13, and the catalyst used in the Rheinpreussen demonstration plant (Krlbel et al., 1955)_ Selectivities of methane and C2 hydrocarbons on this catalyst are low and comparable to those obtained in these two studies conducted in the bubble column slurry reactors. Catalyst 100 Fe/5 Cu/6 K/24 SiO2 is also more active than the Ruhrchemie (Bukur et al., 1990c), United Catalysts Inc. (Bukur et al., 1993) and UOP (Abrevaya et al., 1991) catalysts, and it produces less methane and gaseous hydrocarbons than these catalysts. The productivity of the catalyst, expressed as g hydrocarbons produced/g-Fe/h, is equal to or greater than that of the other catalysts in Table 2. This quantity varies with process conditions. In test SB-1931, the highest productivity (0.54 g HC/gFe/h) was obtained at (H2 + CO) conversion of 75% (SV = 3.4 Nl/g-Fe/h), and the lowest productivity (0.35 g HC/g-Fe/h) at (H2 + CO) conversion of 82% (SV = 2.25 Nl/g-Fe/h). The highest syngas conversion, 89%, was obtained in the Rheinpreussen's demonstration

plant BCSR. Almost the same conversion was achieved here in Run SB-0931 during the early portion of the test. However, the conversion decreased with time due to catalyst deactivation. It is difficult to achieve syngas conversion of 89% in an STSR, the behavior of which approaches that of a perfectly mixed reactor.

SUMMARY

Three precipitated iron catalysts synthesized in the authors' laboratory were tested in the stirred tank slurry reactor to determine their performance (activity, selectivity and stability) as a function of time-on-stream (up to 560h). Syngas conversions higher than 80% were achieved in all three tests, and the catalyst deactivation rate was moderate (1.3-2.1% per day). Selectivities of all three catalysts were fairly stable with time, and production of methane and gaseous hydrocarbons was very low, whereas the selectivity of liquid hydrocarbons and wax was high. The activity of one of these three catalysts (100Fe/5 Cu/6 K/24 SIO2) was higher than that of any other state-of-the-art catalyst developed for the slurry phase operation. It would be desirable to evaluate the three catalysts in a BCSR, at a high syngas conversion, to determine the catalyst deactivation and product distribution as a function of time-on-stream.
Acknowledgements--This work was supported by the U.S. Department of Energy (Pittsburgh Energy Technology Center) under contract DE-AC22-89PC89868 and the Texas Advanced Technology Program under Grant No. 999903-222_

Fischer-Tropsch synthesis in a stirred tank slurry reactor


NOTATION

4625

HC k Kp P Pi SV t UR X

hydrocarbon compounds apparent first-order reaction rate constant, tool (H2 + C O ) / g - F e / h / M P a partial pressure quotient for the water-gas shift reaction, Kp = Pco2PH/PcoPn2o total reaction pressure, MPa partial pressure of c o m p o n e n t i (i = CO, CO2, H2 or H 2 0 ) , MPa gas hourly space velocity per mass of iron at standard conditions, NI/g-Fe/h reaction time, h or days usage ratio (moles of hydrogen consumed per mole of carbon monoxide) conversion (XH2+co---synthesis gas conversion)

REFERENCES

Abrevaya, H., Frame, R. R. and Targos, W. M., 1991, Technology development for iron Fischer-Tropsch catalysts, in DOE Liquefaction Contractors" Review Meeting Proceedings (Edited by G. J. Stiegel and R. D. Srivastava), pp. 219-235_ Bukur, D.B., 1983, Some comments on models for Fischer-Tropsch reaction in slurry bubble column reactors. Chem. Engng Sci. 38, 441-446. Bukur, D. B., Lang, X., Rossin, J. A., Zimmerman, W. H., Rosynek, M. P., Yeh, E. B. and Li, C., 1989, Activation studies with a promoted precipitated iron Fischer-Tropsch catalyst. Ind. Engng Chem. Res. 28, 1130-1140. Bukur, D_ B., Mukesh, D. and Patel, S. A., 1990a, Promoter effects on precipitated iron catalysts for Fischer-Tropsch synthesis. Ind. Engng Chem_ Res. 29, 194-204. Bukur, D. B., Lang, X., Mukesh, D., Zimmerman, W. H., Rosynek, M. P. and Li, C., 1990b, Binder/support effects on the activity and selectivity of iron catalyst in the Fischer-Tropsch synthesis. Ind. Engng Chem. Res. 29, 1588-1599. Bukur, D. B., Patel, S. A. and Lang, X., 1990c, Fixed bed and slurry reactor studies of Fischer-Tropsch synthesis on precipitated iron catalyst. Appl. Catal. 61, 329-349. Bukur, D. B., Lang, X., Koranne, M. and Nowicki, L_, 1993, Pretreatment effect and process evaluation studies of precipitated iron Fischer-Tropsch catalyst, in DOE Coal Liquefaction and Gas Conversion Contractors Review Conference Proceedings Vol. 2 (Edited by S. Rogers, P. Zhou, K. Lockhart and N_ Maceil), pp. 943-970. Deckwer, W.-D., Serpemen, Y., Relek, M. and Schmit, B., 1981, On the relevance of mass transfer limitations in the Fischer-Tropsch slurry process. Chem. Engng Sci. 36, 765-771.

Dictor, R. and Bell, A. T., 1986, Fischer-Tropsch synthesis over reduced and unreduced iron oxide catalysts. J. Catal. 97, 121-136. Dry, M. E., 1981, The Fischer-Tropsch synthesis, in Catalysis--Science and Technology, Vol_ 1, pp. 160255. Springer-Verlag, New York. Farley, R. and Ray, D., 1964, The design and operation of a pilot-scale plant for hydrocarbon synthesis in the slurry phase. J. Inst_ Petr_ 50, 27-46. Fox, J. M., Degen, B. D., Cady, G., Deslate, F. D. and Summers, R. L., 1990, Slurry reactor design studies: slurry vs. fixed-bed reactors for Fischer-Tropsch and methanol synthesis. Final Report prepared for DOE Contract No. DOE/PC/89867-T2, Bechtel, San Francisco, CA. Gray, D., EI-Sawy, A. and Tomlinson, G., 1991, Quantification of progress in indirect coal liquefaction, in DOE Liquefaction Contractors' Review Meeting Proceedings (Edited by G. J. Stiegel and R. D. Srivastava), pp. 344-356. Herzog, K. and Gaube, B., 1989, Kinetics studies for elucidation of the promoter effect of alkali in FischerTropsch synthesis_ J. Catal. 115, 337-346. Huff, G. A., Jr and Satterfield, C. N., 1984, Evidence for two chain growth probabilities on iron catalyst in the Fischer-Tropsch synthesis. J. Catal. 85, 370-379. KOlbel, H. and Ralek, M., 1980, The Fischer-Tropsch synthesis in the liquid phase. Catal. Rev--Sci. Engng 21,225-274. K61bel, H., Ackerman, P_ and Engelhardt, F., 1955, New developments in hydrocarbon synthesis, in Proceedings of Fourth World Petroleum Congress, Section IV/C, pp. 227-247. Carlo Colombo Publishers, Rome. Kuo, J. C. W., 1983, Slurry Fischer-Tropsch/Mobil two stage process of converting syngas to high octane gasoline. Final Report prepared for DOE Contract No. DE-AC22-80PC30022, Mobil Research and Development Corp., Paulsboro, NJ. Kuo, J. C. W., 1985, Two stage process for conversion of synthesis gas to high quality transportation fuels. Final Report prepared for DOE Contract No. DE-AC2283PC600019, Mobil Research and Development Corp, Paulsboro, NJ. Mitra, A. K. and Roy, A. N., 1963, Performance of slurry reactor for Fischer-Tropsch and related synthesis. Indian Chem. Engng July, 127-142_ Newsome, D. S., 1980, The water-gas shift reaction. Catal. Rev.--Sci. Engng 21,275-318. Patel, S. A., Daly, J. G. and Bukur, D. B., 1990, Bubble size distribution in Fischer-Tropsch derived waves in a bubble column. A.I.Ch.E.J. 36, 93--105. Shah, P_ P. and Fullerton, H. W., 1990, Economics of upgrading Fischer-Tropsch products, in DOE Indirect Liquefaction Contractors" Review Meeting Proceedings (Edited by G. J. Stiegel and R. D. Srivastava), pp. 383--.409_ Zimmerman, W. H. and Bukur, D. B., 1990, Reaction kinetics over iron catalysts used for the FischerTropsch synthesis. Can. J. Chem. Engng 68, 292-301.

You might also like