Conclusions
Literature Cited
T h e regenerative COZ-removal system designed for the three-man AAP command module was tested in a closed chamber for 47 days. N o electrical regeneration of the molecular sieve beds was required during the test. At no time during the 47 days did the average daily level of COZ partial pressure in the test chamber exceed the design limit of 7.6 mm. of Hg. The analytical techniques developed provide excellent description of system performance and accurate prediction of performance degradation because of water poisoning. These predictions were compared with experimental results from both normal and accelerated operation.
AiResearch Manufacturing Co., Rept. 67-2779 (December 1967) Dell'Osso, L., Jr., Winnick, J., IND.ENG.CHEM.PROCESS DESIGNDEVELOP. 8,468 (1969). Fukunaga, Paul, Hwang, K. C., Davis, S. H., Jr., Winnick, J., IND.ENG.CHEM.PROCESS DESIGNDEVELOP. 7 , 269 (1968). Schumacher, W. J., York, R., IND.ENG. CHEM.PROCESS DESIGNDEVELOP. 6,321 (1967).
RECEIVED for review November 1, 1968 ACCEPTED May 23, 1969
METHANATION OF LOW-CONCENTRATION CARBON MONOXIDE FEEDS OVER RUTHENIUM 5 . E.
S . R A N D H A V A , A M l R A L l H . C A M A R A
Institute of Gas Technology, Chicago, Ill.
R E H M A T ,
A N D
60616
The methanation of carbon monoxide a t parts per million levels was studied over 0.5% ruthenium metal catalyst, dispersed on alumina catalyst in a fixedbed reactor. Gas mixtures of 3450, 1090, and 505 p.p.m. carbon monoxide in hydrogen were used. The rate of reaction of carbon monoxide follows simple pseudo-first-order kinetics. The rate constant follows the Arrhenius temperature dependence at low temperatures. Evidence of diffusion control of the reaction rate was found in the higher temperature regions investigated.
METHANE synthesis by the reaction of CO and Hz over metal catalyst was first reported a t the beginning of the century. This reaction was soon recognized to have commercial significance in gas manufacture because it offered a possible alternative to oil carburetion. Akers and White (1948) studied the kinetics of methane synthesis over reduced nickel catalyst a t atmospheric pressures. A detailed study of the synthesis of methane with empty stainless steel tubes and with steel balls was reported by Gilkeson et al. (1953). Nicolai et al. (1946) determined the kinetics of the methanation reaction over ruthenium catalyst a t elevated pressures. Extensive work has also been reported by the U.S. Bureau of Mines (Karn et al., 1965). A detailed survey of the studies in methanation appears in a research bulletin (Dirksen and Linden, 1963). The best catalysts for hydrocarbon synthesis and the hydrogenation reactions belong to the eighth group of the periodic system of the elements that comprises iron, cobalt, nickel, ruthenium, rhodium, palladium, osmium, iridium, and platinum. Although the metals of the platinum group are all generally active, they differ considerably in their ability to catalyze various reactions between CO and HZ. Ruthenium has been known for many years to be a highly effective catalyst in the Fischer-Tropsch reaction and methane synthesis. Recently, McKee (1967) studied the interaction of H2 and CO with platinumgroup metals and suggested that the highly specific behavior of ruthenium was probably due to its lower affinity for CO than the other noble metals. Many thermodynamic analyses of the methanation reaction system have been conducted to evaluate the influence 482
I & E C PROCESS D E S I G N A N D DEVELOPMENT
of the major operating variables. Under conditions of the normal chemical equilibrium with a fair degree of CO conversion, the direct methanation reaction is accompanied by the shift reaction. However, in the presence of a highly active methanation catalyst, CO is consumed so rapidly that the CO water-gas shift equilibrium is not approached, and CO concentrations are well below computed equilibrium values. Consequently, to negate any effects of the CO-shift reaction in determining the exact kinetics of the CO methanation, it is advisable to work with feed gases having very low concentrations of CO. At the energy conversion laboratory of the Institute of Gas Technology investigations have been under way to remove traces of CO from the H 2 anode gas, which would otherwise act as poisons in the normal operations of low-temperature fuel cells. In the present study of methane synthesis, parts-permillion mixtures of CO and H z were passed through a tubular reactor containing ruthenium catalyst. The effects of temperature, feed rate, and feed composition were investigated. Experimental
Experimental Apparatus. The diagram of the flow system is shown in Figure 1. T o facilitate the description, the apparatus is considered in operation. The mixture of carbon monoxide and hydrogen passes through a pressure regulator and control valve to the rotameter, which has been calibrated by using a wettest meter, enters the top of the reactor, and flows downward through the catalyst bed.
I
CO LIRA
7 SPAN G A S
I
t b
3050 ppm CO1
o CHI L I R A ZERO GAS
-Cn.
LIRA
SWN GAS [ 1000 Gum1
Figure 1. Equipment and apparatus
WATER
L1RA B Y P P S S TC VEN'
+ CONDENSATE
OUT
The reactor is built from %-inch i.d., 1-inch o.d., and 22-inch-long stainless steel 304 tubing. An 11 gage 4.inch o.d., 2 'IJ-inch-long stainless steel 304 tube is inserted in both ends and welded to provide connections to the reactor. A multipoint thermocouple enters from the bottom. A perforated screen supports the catalyst and the catalyst support within the reactor. The screen was carefully welded to a h-inch 0.d. thermocouple well, l Y 4 inches from the tip, and sits 11 inches above the bottom of the reactor. This arrangement makes it easy to discharge the catalyst. A multipoint thermocouple consists of three Chromel-Alumel thermocouples placed a t %-inch intervals starting from the tip of the thermocouple well. This measures the top, middle, and bottom temperatures of the catalyst bed. These thermocouples are connected to a Brown temperature recorder. After leaving the reactor, the moisture in the exit gas was condensed. The exit stream was continuously analyzed until steady state was achieved. An infrared analyzer similar to that of Cohen et al. (1966) (MSA Lira Model I R 300) was used t o determine the concentration of carbon monoxide and MSA Lira Model I R 300 was used to determine the concentration of methane. The CO infrared analyzer was calibrated to read in the three ranges 0 to 5000 p.p.m., 0 to 1000 p.p.m., and 0 to 500 p.p.m. and had an accuracy of 1 2 5 a t full scale. The CH, analyzer was calibrated to read from 0 to 5000 p.p.m. Both units were calibrated periodically, using a special span gas, and were checked for gas drift prior to each run. All data were reproducible to within & 3 5 conversion. The catalyst used in this investigation, furnished by Englehard Industries, Inc., consisted of 0.5% ruthenium impregnated on alumina pellets. T h e catalyst pellets were % x yb inch cylinders. The average weight of each pellet was 0.044 gram. The catalyst support consisted of small alumina pellets. The catalyst bed was changed prior to a set of runs with different concentrations of gas. The bed was activated by heating for 6 hours a t 400°C. by passing helium through the bed. The CO-H, assayed gas mixtures and the zero and span gases for the two Liras were furnished by The Matheson Co.
initial concentration of carbon monoxide was varied from 505 to 3450 p.p.m. There was no decrease in the catalytic activity of ruthenium over the range of temperatures investigated. Specific runs repeated before and after a set of experimental data showed excellent reproducibility within the margin of experimental error. The time factor, V u0, in the present case was varied by changing the feed rate while maintaining the same volume of catalyst. The data obtained by varying the time factor a t 175", 200°, 225", 250°, and 275°C. are shown in Figure 6 for the feed gas containing 3450 p.p.m. Space velocities were held constant in several other runs, while the quantity of catalyst and the feed flow rate were changed by simultaneously doubling the catalyst volume and the feed flow rate. No effect of this change in catalyst volume or flow rate was noted as long as the ratio of these two was held constant. The methanation of CO may proceed by either or both of the following reactions:
CO + 3H2 + CH, + H?O 2CO + 2H2- CH, + CO?
(1) (2)
Results Experimental data for the conversion of carbon monoxide as a function of temperature for different space times and feed gas concentrations are shown in Figures 2 to 4 and for the corresponding production of methane for feed gas containing 3450 p.p.m. of CO in Figure 5. The
Figure 2. Conversion of CO as a function of temperature with 505 p.p.m. feed gas VOL. 8 NO. 4 OCTOBER 1 9 6 9
483
T,*C
Figure 5. Production of methane as a function of temperature with 3450 p.p.m. feed gas
Figure 3. Conversion of C O as a function of temperature with 1090 p.p.m. feed gas
0
10 0
01
03
02
04
05
06
V c c . catalyst v, c c . / s c c . gas
- 2
Figure 4. Conversion of CO as a function of temperature with 3450 p.p.m. feed gas
Dirksen and Linden (1963) have shown that the HzOforming methanation reaction is favored over the COPforming reaction by high H,/CO ratios and that the COPforming reaction starts predominating with increasing reactor pressures. Since the Hz/CO ratios are extremely high in this investigation, the effects of the COP-formingreac484
I&EC PROCESS DESIGN A N D DEVELOPMENT
Figure 6. CO concentration as a function of space time with 3450 p.p.m. feed gas
tion should be minimal compared t o the HrO-forming reaction. We observed this by examining methane formation as a function of CO depletion. There is a fair degree of correspondence a t all temperatures and space times; however, some difference (maximum 155) is noticeable, more so for feed gases containing the higher concentrations
of carbon monoxide. For gas containing 3450 p.p.m. of
CO there is a slight difference over the entire range of temperatures investigated, the degree of deviation increasing with the increases in temperature. For the feed gas containing 505 p.p.m. of CO, the difference first becomes apparent a t temperatures above 200" C. The rate of CO conversion increases rapidly with a decrease in the partial pressure of CO a t the same temperature. Kicolai et al. (1946) indicated that, in the presence of a large excess of H r , the H20-forming methanation reaction proceeds a t a constant speed for the majority of the time and then clearly accelerates before the end. The feed concentration levels in this investigation are roughly those mentioned by Nicolai et al. just before their reactions were terminated. There is consequently substantial agreement between our results and the hypothesis proposed by Sicolai et al. For the atmospheric pressures a t which the investigation was conducted, the fractional conversion of carbon monoxide increased continuously with an increase in temperature. This increase is expected from the Arrhenius equation, which indicates the effect of temperatures on the reaction. However, in the higher temperature ranges, the final apparent effect is probably due to the combined effects of both specific rate constants (Arrhenius equation) and the adsorption constants (diffusion effects). Discussion
The complete set of data was used to determine the order of reaction and the reaction rate constant in the kinetic equation:
-reo = kC;o
The rate of methane synthesis reaction on the surface of the catalyst can be given by (Akers and White, 1948)
For very low CO, CH,, and CO? concentrations, Equation 7 reduces to a pseudo-first-order equation:
where the reaction rate constant incorporates the effect of the constant concentration of hydrogen. Consequently, even before analyzing the data, one would suspect that the rate of disappearance of CO was first-order for CO. The fact that the experimental data fit the assumption of the first-order reaction so well indicates the accuracy of the data. The rate constants for the three feed gases as a function of temperature were derived from the slopes of the lines similar to those shown in Figure 6. Figure 7 shows the derived rate constants as a function of reciprocal absolute temperature. As the results of all three CO concentrations are presented in the figure, the scatter does not seem excessive. In lower temperature regions. the data follow the linear relationship, but start deviating in the higher temperature ranges. These results indicate that the reaction is kinetically controlled in the lower temperature ranges, while the effects of diffusion control on the kinetics become progressively apparent in the higher temperature ranges.
(3)
by utilizing the design equation for the catalytic flow reaction:
L',
Jc co
(4)
-r
10
The simple empirical rate equation such as Equation 3 has been suggested by Levenspiel (1962) for engineering design. This equation does not incorporate a term for the concentration of hydrogen, because that gas is present in such large excess as to be considered essentially constant. Since the product methane concentration is very small, the total number of moles is unchanged throughout the reaction. Consequently, one can safely assume that the total number of moles remains constant during the passage of the gas through the reactor. With the substitution of Equation 3 into Equation 4 we can write:
: -
t
V/u,was plotted as a function of C , , ' ~ " for various '
i
I I
I
I
I
1
1
19
20 I/T
=+=I
I I
--
h
--
I
c-! 001
1
;
,I
I
(6)
values of n and also as a function of log Cco. The value of n yielding the best set of straight lines through the data was used as a reaction order. Figure 6 shows the result of assuming n = 1 for the CO feed concentration of 3450 p.p.m. used in the investigation.
1
\
01
n = l
e.
I
\ \
1
I
,
I
21
2 2
2 3
24
x 103.K1
Figure 7. Arrhenius plot for methanation of CO VOL. 8 N O . 4 OCTOBER 1 9 6 9
485
As noted by Caretto and Nobe (1966), this behavior is typical of pore diffusion. Further, radial temperature gradients that become progressively larger as the temperature is increased could also affect the shape of the Arrhenius plot. In our investigation, because of the shape and the diameter of the reactor, the highest radial gradients were not in excess of 1” to 2’C. and, as such, would have a minimal effect on the Arrhenius plot. I t seems reasonable to attribute the curvature in the Arrhenius plot to the pore diffusion only. The values of constants k , and E in the equation:
k = hoe-”
(9) were obtained using the lower temperatures only. The values were:
Apparent activation energy, E = 37.2 x IO3 cal. per g. mole Frequency factor, h, = 1.25 x 10” cc. gas/cc. catalyst-second Conclusions
The methanation of carbon monoxide a t parts per million levels was studied over a 0.5% ruthenium catalyst in a fixed-bed reactor. The rate of reaction of carbon monoxide follows simple pseudo-first-order kinetics:
where h follows the Arrhenius temperature dependence a t low temperatures. Evidence of diffusion control of the reaction rate was found in higher regions of the temperatures investigated.
k = pseudo-reaction rate constant, cc. gas/cc. cat.sec. h, = Arrhenius frequency factor, cc. gas/cc. cat.-sec. n = reaction order rA = reaction rate of species A , (p.p.m. CO/sec.) act. gas/cc. cat. R = gas constant, cal./g. rnole-OK. T = temperature, K. v, = volumetric flow rate of feed gas, cc./sec. V = volume of catalyst, cc. XCH4= moles of CH1produced per mole of CO in feed X,, = moles of CO converted per mole of CO in feed All volumes measured a t 60’ F. and atmospheric pressure. literature Cited
Akers, W. W. White, R. R., Chem. Eng. Progr. 44, 554 (1948). Caretto, L. S., Nobe, K., IND. ENG. CHEM. PROCESS DESIGNDEVELOP. 5, 217 (1966). Cohen, A. E., Nobe, K., IND.ENG.CHEM.PROCESS DESIGN 5, 214 (1966). DEVELOP. Dirksen, H. A., Linden, H. R., “Pipeline Gas from Coal by Methanation of Synthesis Gas,” Institute of Gas Technology, Res. Bull. 31 (1963). Gilkeson, M. M., White, R. R., Sliepcevich, C. M., Znd. Eng. Chem. 45, 460 (1953). Karn, F. S., Shultz, J. F., Anderson, R. G., Znd. Eng. Chem. Prod. Res. Develop. 4, 265 (1965). Levenspiel, O., “Chemical Reaction Engineering,” Chap. 14, Wiley, New York, 1962. McKee, D. W. J., Catalysis 8, 240 (1967). Nicolai, J., d’Hont, M., Jungers, J. C., Bull. SOC.Chim. (Belges) 55, 160 (1946).
Nomenclature
B, D C,, C,, E
RECEIVED for review August 8, 1968 ACCEPTED June 2, 1969
= integration constants = concentration of CO in feed, p.p.m.
= concentration of CO in effluent, p.p.m.
Institute of Gas Technology sponsored this work through its basic research program.
= activation energy, cal./g. mole
M I N I M U M CRITICAL VELOCITY FOR ONE-PHASE FLOW OF LIQUIDS ALFONSO
GUTIERREZ’
AND
S C O T T
LYNN
Department o f Chemical Engineering, University of California, Berkeley, Calif. 94720
IT IS frequently necessary when
designing chemical processing plants t o provide for transporting a relatively hot liquid from a reactor, heat exchanger, or holding tank through a pipe to a receiver where the pressure is relatively low. If flashing occurs in the pipe or in a valve, control of the liquid flow may become difficult. If the flashing is accompanied by the precipitation of dissolved solids, the designer may be faced with a decidedly vexing problem.
’ Present address, Carrera 20, No. 54-45, Bogota D.E.2, Colombia 486
1 8 E C PROCESS D E S I G N A N D DEVELOPMENT
Many authors have investigated the flow of flashing liquids in pipes (Allen, 1951; Benjamin and Miller, 1942; Bottomley, 1936; Starkman et al., 1964), and virtually all of the possible two-phase flow regimes have been studied. Several have noted that critical flow is frequently obtained (Cruver and Moulton, 1967; Isbin et al., 1957; Levy, 1965; Moody, 1965; Zivi, 1964)-i.e., that the flow rate in the pipe is unaffected, within limits, by variation of the pressure in the receiver. The velocities of these critical flows are always far below sonic velocity in either phase of the flowing mixture.