should be negative or close to zero. Thermodynamic calculations show this to be the case with four other metals [Mo, Ge, V(VO2 + VO), and Sn] as well as iron. I t is interesting to note that, of all the metals tried as catalysts, four (Fe, V, Mo, and Sn) of the five mentioned above were the most effective of all the metals that were investigated. In addition, Reaction 12 was found to take place in molten carbonate. I t was found that 4.8% iron oxide dissolved in the carbonate eutectic was, reduced by 31% in 6.75 hours. Reaction 13 has also been shown to take place (Vinogradov and Belyi, 1967). This indirect information suggests that iron is directly involved in the reduction. literature Cited
Ahlgren, P., Teder, A., Acta Chem. Scand. 21, 1119 (1967). Bagbanly, I. L., Mirbabaeva, F. Yu, Azerbaidzahn. Khim. Zhur. (Russ.) 111 (1959); CA 55, 10817c (1961). Birk, J. R., Larsen, C. M., Wilbourn, R . G., Anal. Chem. 42, 273 (1970). Fotivev. A. A., Izuest. Sibir. Otdel. Akad. Nauk S.S.S.R. (Russ.) No. 9, 107 (1960). Heredy, L. A., McKenzie, D. E., Yosim, S. J. (to North American Rockwell), U. S. Patent 3,438,722 (April 15, 1969).
Kunin, V. T., Kirillov, I. P., Izu. Vyssh. Ucheb. Zaued., Khim. Tekhnol (Russ.) 11, 569 (1968); C A 69, 64258x (1968). Meyer, R. J., “Gmelins Handbook of Inorganic Chemistry,” 8th ed. (Ger.), Vol. 21, pp. 184-205, Verlag Chemie, Weinheim, Germany, 1965. Nikitin, V. A., Kunin, T. I., Zhur. Vsesyuz. Khim. Obshchestua im D. I . Mendoleeua (Russ.) 5 , 350 (1960). Nyman, C. J., O’Brien, T. D., Ind. Eng. Chem. 39, 1021 (1947). Polyvyannyi, T. R., Demchenko, R. S., Izuest. Akad. Nauh,
Kazakh., SSR, Ser. Met., Obogashchen. i Ogneuporou (Russ.) 34, (1960); CA 55, 198e (1961). Puttagunta, V. R., Ph.D. thesis, Univ. of Saskatchewan, Saskatoon, Saskatchewan, 1967. Vinogradov, V. I., Belyi, V. M., Izotopy Sery Vop Rudoobrazou. (Russ.) 118 (1967); CA 70, 5899e (1969). White, F. M., White, A. H., Ind. Eng. Chem. 28, 244 (1936). RECEIVED for review November 3, 1969 ACCEPTED August 3, 1970 The basic process is a proprietary development of North American Rockwell Corporation. Subsequent work was performed pursuant to contract P H 86-67-128 with the U. S. Public Health Service, Department of Health, Education and Welfare.
Adsorption of light Hydrocarbons from Nitrogen with Activated Carbon William C. McCarthy Phillips Research Center, Phillips Petroleum Co., Bartlesuille, Okla.
74003
Dynamic adsorption data were obtained on activated carbon for single, binary, and multicomponent hydrocarbons in nitrogen at 100 and 300 psig. Component concentration, gas velocity, presaturation, and pressure were investigated and correlated with the length of the mass transfer zone. Presaturation of the carbon with the next lighter hydrocarbon component had the greatest effect on zone length. Bed length, component concentration, gas velocity, and pressure also affected the zone length substantially. Since the zone length varied considerably, depending upon conditions, it was not considered a good design tool for multicomponent systems. An exposure-recovery plot facilitates the design of an adsorption unit but a separate, experimentally-determined plot i s required for each feed composition, pressure, and velocity encountered.
T h e recovery of propane from natural gas using a shortcycle carbon adsorption unit has been demonstrated by Bray et al. (1965) to be an attractive processing method. To accurately design short-cycle adsorption units dynamic loading data for the adsorbed hydrocarbons are required. Dale et al. (1961) and Campbell et al. (1963) have experimentally determined the dynamic loading of several hydrocarbons on silica gel and have correlated the data through the use of the mass transfer zone concept. Dynamic loading data for carbon, however, has been seriously lacking in the literature. The purpose of these studies
was to obtain dynamic loading data for single, binary, and multicomponent hydrocarbon systems on carbon and to determine the effects of concentration, gas velocity, and pressure on the length of the mass transfer zone for the various hydrocarbons. Experimental
The adsorbent used was Columbia Grade NXC 4-6 mesh activated carbon. Nitrogen was used for the carrier gas. All hydrocarbons were Phillips Petroleum Co. pure grade. Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
13
SURGE VESSEL
BATH TO HEAT NITROGEN DURING REGENERATION
ADSORPTION BED
Figure 1. Gas adsorption unit
CAPILLARY BACKPRESSURE
HYDROCARBON
MOTO? VALVE ON CONTROL I R HAND VALVE PHILLIPS MODEL 12 ANALYZER
1.1
w
0.t
w
+! FEED SAMPLES
LAST EFFLUENT /SAMPLE 142 MIN
2 0
I
x
z
0 I< a
0.6
I
t-
z w
I
’
I
I
h:
V
8
0.4
w
z
I
INLET
I! !
; I z
BED LENGTH-
OUTLET
Figure 3. Gas composition gradient through bed
w
n
I
MASS TRANSFER ZONE
0.2
EFFLUE
0 TIME INCREASES ANALYSIS EVERY TWO MINUTES Figure 2. Typical one component in nitrogen fractogram Pentane in N?at 100 psig
Figure 1 is a flowsheet of the experimental unit. The adsorbent bed was a section of 2-in. schedule 40 stainless steel pipe 67 in. (5.58 f t ) long containing 3.68 Ib of carbon. The nitrogen was controlled by an integral d P cell-orifice meter connected to a motor valve. Hydrocarbon was then added to the nitrogen in a small glass beadpacked evaporator. A surge vessel was placed in the gas line to help smooth any concentration fluctuations. Flow was downward through the bed, and column pressure was maintained by a pressure controlled motor valve in the outlet. After completion of the adsorption step, the charcoal was regenerated with nitrogen heated to 600750”F in a salt bath. 14
Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
A Model 12 Phillips gas-liquid, on-stream chromatograph was used for analyses. The output was a bar-peak for each hydrocarbon component on chart paper and is illustrated in Figure 2 . The chromatograph was connected to both the inlet and outlet of the bed. There was no temperature control on the adsorption column as a11 adsorptions were run a t room temperature. The bed was cooled prior to a run by purging with nitrogen. Discussion
One method for using dynamic adsorption data for design purposes is through the use of the “mass transfer zone” concept. The mass transfer zone as described by Michaels (1952) is that section of the adsorbent bed in which a change in gas composition is occurring. This is illustrated in Figure 3. At the inlet of the bed there is Zone A saturated to essentially equilibrium with the adsorbable component in the fluid. At the effluent end of the bed is Zone C where the fluid is essentially in equilibrium with regenerated adsorbent. Between these two is Zone B in which the fluid concentration of the
Figure 4. Separation of components by selective adsorption
W
I a
A = section of bed in equilibrium with denuded carrier gas
2 %
G d I-
5=
Y a
1.0
s INLET
BED LENGTH INCREASES
-
OUTLET
l!d
e E
(0,
- 0,)
(1)
Fo)(e€- eB)
- (1 -
where: 2 = length of mass transfer zone, f t ; L = length of adsorbent bed, ft; O B = time of breakthrough of adsorbed component a t the end of the bed, min; eE = time when the bed is exhausted--i.e., effluent; gas composition matches feed gas, min; F , = fractional residual adsorbent capacity in mass transfer zone. The fractional residual capacity could be determined from the experimental fractogram given in Figure 2. The area enclosed by the dotted lines below the height of the feed peaks, but above the breakthrough curve formed by the effluent gas bar peaks (Zone A ) , represents the amount of feed adsorbed as the mass transfer zone moved out of the bed. This area divided by the area that would represent the amount of feed for the same increment of time is the factor F,.Zone length determinations were arbitrarily based on times corresponding to the 5 and
L c;
3*0I 2.01
W'
z
s
1
1.0
% z
0.8 0.6
a U t-
v)
2 =
0.5 0.4
0.3
g
0.21
0
I
=
replaces
mass transfer zone where pentane replaces some butane
95% concentrations of the breakthrough curve. If the mass transfer zone is symmetrical, F equals 0.5. Experimental values were usually below this. The foregoing discussion applies to one adsorbable component in a nonadsorbable gas. If there are two or more adsorbable components, the breakthrough curves are more like that shown in Figure 4 than in Figure 2 . A detailed discussion of the breakthrough curves for multicomponents is presented later. When multicomponents are considered. the definition of the term 81:in the zone length equation must be modified to accommodate adsorbable component concentrations in the effluent gas greater than in the feed. @ E then becomes the time when the component concentration in the effluent gas first approaches the component concentration in the feed gas. Single Component Data
Effect of Velocity. Single hydrocarbons in nitrogen were first studied. The nitrogen is but slightly adsorbed a t the temperature and pressures investigated, hence the hydrocarbon can be considered almost the only adsorbable component. Severe heat of adsorption effects were encountered if the hydrocarbon feed concentration approached one mole per cent.
t: wz 0 N
a w L v,
z
U a tv,
v,
a
1
LL
F
mass transfer zone where propane i s being adsorbed
L
l
a W
=
and propane G = section of bed in equilibrium with feed gas Assumed feed: carrier gas plus propane, butane, and pentane
adsorbable component is decreasing. This is the mass transfer zone. The length of this transfer zone was calculated by the method developed by Michaels (1952). The equation is:
z=
=
section of bed loaded with propane = mass transfer zone where butane partially propane E = section of bed loaded with propone ond butane
L L m
d
B C D
I 0
0.3
1
z W -I
I
0.1.
I
I
I
I 1 1 1 1
I.
Figure 5. Effect of velocity on mass transfer zone length propane 0 n-butane
v
isobutane
A
n-pentane
0 isopentane One component in N?, 100 psig; adsorption temp range, 95' to 120" F; hydrocarbon concn, 0.36 to 0.51 mole %
0.1
I
I
I
I l l l l
I
I
Figure 6. Effect of feed concentration on mass transfer zone length 0 isopentane propane 0 n-butane v isobutane (0.80 to 0.83 ft/sec) A n-pentane One component in Nr, 100 psig; adsorption temp, 92" to 140" F; superficial gas velocity, 0.41 to 0.45 ft/sec
Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
15
WITH n-BUTANE
/
/A
IW
5
/BED
AI
UPPER LIMIT LENGTH= 5.58 F
The variable that most affected the length of the mass transfer zone was the superficial gas velocity as shown in Figure 5 . Also, the isoparaffins have a longer transfer zone than the normal paraffins. This, no doubt, is a steric effect since there is less than 105 difference in the equilibrium loading between iso- and normal-paraffins of the same molecular weight. The transfer zone length is independent of molecular weight when considering the isobutane and isopentane and the normal butane and normal pentane. That is, the net adsorptive rate for both molecular weights are equal for each isomeric type. Effect of Hydrocarbon Concentration. The mass transfer zone length increased as the hydrocarbon concentration increased as shown in Figure 6. At the higher concentrations, however, this effect may be exaggerated because of the higher temperature of the zone. Again the normal paraffins had a shorter zone than the isoparaffins. Binary and Multicomponent Data
-
1/
ISOBUTANE PRESATURATED WITH PROPANE ISOPENTANE PRESATURATED WITH PROPANE I 1
I
I
I
I
I
2
3
4
5
6
CONCENTRATION OF PRESATURATION COMPONENT, MOLE X Figure 7. Effect of presaturation component on mass transfer zone length Heovy component concn, nominol 0.5 mole 0.4 ft/sec; pressure, 100 psig (1 14.4 psio)
Oh;
superficial velocity, nominal
For binary and multicomponent mixtures the mass transfer zones in the bed would tend to separate and move along a t different velocities. This is because the higher molecular weight hydrocarbons were preferentially adsorbed while the lower molecular weight hydrocarbons would pass through the bed and load the bed t o a smaller extent. This zone separation effect is illustrated in Figure 4. The first zone, A . is denuded, regenerated carbon. Zone B is the mass transfer zone for the lightest component-eg , propane. Zone C represents bed saturated with propane. Propane loading in this zone may be substantially greater than the final feed equilibrated loading of zone G. D is the mass transfer zone for the next heavier (butane). In this zone butane is displacing a very substantial amount of propane. Zone E is equilibrated with the propane and butane in the feed. Zones F and G would follow in a similar manner for pentane. Frequently, in actual practice. the mass transfer zones ( B , D , and F ) overlap and the equilibrated zones ( C and E ) do not exist. Effect of Lighter Component. The effect of the lighter component (presaturation) on the adsorption of a heavier component is very pronounced. This is illustrated in Figure 7 for isobutane and isopentane. Propane presaturation
! r
I-
2 W
-I
1 0.1
0.2
0.3 0.4 0.6 0.8 1.0
2.0
3.0 4.0
ISOBUTANE CONCENTRATION, MOLE %
Figure 8. Effect of pressure on mass transfer zone length
Figure 9. Effect of concentration on mass transfer zone length for isobutane
Presoturoted with propane at 2.2 psi0 partial pressure; superficial velocity, 0.4 ftjsec
Adsorption temp range, 80" to 100" F; superficial velocity, presoturoted with 2 . 2 psi0 partial pressure of propane
16
Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
0.4 ft/sec;
was accomplished by loading with propane and then discontinuing the propane flow when the heavier hydrocarbon was started. n-Butane saturation was accomplished as with propane except butane flow was continued while the heavy hydrocarbon was fed. Calculated mass transfer zone lengths greater than the bed length are not real and would be shorter than the real zone length. Calculated values greater than the bed length were included in Figure 7 to illustrate the trend of the data. Presaturation with the next lighter hydrocarbon could, depending on the concentration of the presaturator, increase the zone length an order of magnitude. Presaturation with the third lighter hydrocarbon (isopentane-propane) had a much smaller effect. Effect of Pressure. Increasing system pressure increased the mass transfer zone length with constant partial pressure of adsorbable component as shown in Figure 8. Figure 8 also shows the effect of presaturation on isobutane zone length. Effect of Concentration. The effect of isobutane concentration on propane presaturated beds is shown in Figure 9 a t two pressure levels. Again zone length increases with concentration and per cent-wise increases a t about the same rate as for pure components. Compare slopes of curves in Figures 6 and 9.
Effect of Multicomponents. Adsorption data for systems having three or more components are in Table I. The zone length for the lightest component. either propane or n-butane, was about what would be expected for the pure component. That is. the presence of hea\''ier components had no effect on the lightest component. The zone lengths for the heavier components, however, were somewhat greater than for the pure component. Also, contrary to expectations, the zone lengths increased with boiling point. This is illustrated especially well for runs 68-70. Apparently, the overlapping of zones in the short bed used (5.6-ft long) caused the intermediate hydrocarbons (isobutane through n-pentane) to be displaced shortly after being adsorbed. This suggests that the zone length is very much a function of bed length. I t has already been shown that the next lighter component has a very significant effect on the zone length of a given component. The effect of the third lighter component, however, had very little effect (runs 61. 62, and 64). Propane had very little influence on isopentane; in fact, the zone length for isopentane was about the same as for the pure component. Thus, for multicomponent systems the mass transfer zone length varies substantially with velocity. component concentration, lighter component concentration, pressure,
Table 1. Multicomponent Adsorption Data 59
Run
Feed, mole C
cc
61
62
64
68
69
70
in N 0.689
0.647
0.791
0.555 0.558
0.464 0.467
0.592 0.595
IC
C IC C C Time to 5 5 breakthrough, min
0.922 0.432 0.381
C IC4 CI IC
C C Zone length. min'
55.36 102.19 134.94 158.76
C
96.86
iCi Ci Zone length, ft C
100.07 103.24
189.54 193.96
111.81 115.30
16.54
21.08
17.10
16.11 122
21.25 226.69
16.01 118.79
1.50
1.12
1.43
IC4
C, IC
C Ch Loading a t end of cycle, g g charcoal
0.59 2.09 3.7
C
0.84 5.1
0.60 4.83
0.75 4.46
0.00777
0.0101
0.0040
IC4
C4
IC
C C, Gal C i adsorbed cu ft carbon Superficial velocity, ft /sec Adsorption t e m p , Fd Adsorption pressure, psia
0.0444 0.0904 0.1380 1.57 0.285 90 114.4
0.1000 0.1680 0.418 95 114.4
0.1129 0.1882 0.284 90 114.4
2.15 0.529 0.541 0.401 0.403 0.416
3.00 0.359 0.587 0.439 0.437 0.440
60.17
IC * 11.26 60.69 135
1.075' 0.194 0.307 0.234 0.228 0.237
0.1129 0.1904 1.77 0.410 100 114.4
71.60 80.48 110.71 126.93 266.28
34.23 39.88 53.55 62.55 118.32
50.66 59.06 80.94 94.70 189.12
20.70 31.70 53.36 104.52 213.87
15.92 18.87 34.47 51.82 146.88
18.94 25.18 40.25 76.33 151.87
1.42 1.87 2.23 3.45 3.49
2.11 2.16 2.81 3.40 4.69
1.74 1.99 2.24 3.38 3.38
-0.001 1 0.0073 0.0214 0.0364 0.2059 1.49 0.448 90 114.5
-0.0020 0.0055 0.022; 0.0295 0.2021 1.43 0.282 85 114.3
0.0086 0.0098 0.0191 0.0301 0.2074 0.397 80 114.3
'Propane not measured by analysis, but calculated from hydrocarbon addition rates. 'From 5'; t o 9 5 ' ~ of feed concentration Temperature of last zone through.
of that component. ' At adsorption temperature and pressure.
Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
17
Recovery is that fraction of the feed adsorbed for each component. An E-R plot for run 69 is shown in Figure 10. Data points for run 70 determined a t 0.62 the velocity of run 69 are also plotted to show the effect of velocity. The E-R plot facilitates the rapid design for amulticomponent adsorber. But, unfortunately, it also suffers from some of the same disadvantages as the mass transfer zone length concept. Therefore, a separate E-R plot is usually required for each feed composition, velocity, and pressure likely to be encountered. In the absence of a suitable method for correlating dynamic loadings, data for each E-R plot must be determined experimentally. Such data can be reliably obtained in a small diameter, long tube adsorption unit as suggested by Barrere (1968). Conclusions
The mass transfer zone length for pure hydrocarbons in nitrogen a t 100 psig varied from 0.5 ft for n-pentane to 2.0 f t for propane. The zone lengths increased almost proportionally with gas velocity and somewhat less than proportionally with concentration. For binary and multicomponent gases the zone length was greatly affected by the next lighter component. Depending on the concentration of the next lighter. the zone length could increase more than an order of magnitude. For multicomponent gases the zone length increased as the boiling point increased. This is contrary t o previous experience and suggests severe overlapping of the mass transfer zones in the 5.6-ft long pilot plant bed. The mass transfer zone then is a critical function of bed length. Since the mass transfer zone length is strongly affected by bed length, concentration of the next lighter component, gas velocity, and, t o a lesser extent, by pressure and concentration, the zone length concept alone is a poor tool in the design of multicomponent adsorption beds. An exposure-recovery plot facilitates the design of an adsorption unit. A separate plot, however, is required for each feed composition, pressure, and velocity that may be encountered and must be determined experimentally. literature Cited
30t
2ob
rb
2;
3b
46
sb
66
7b
1 810
do
I60
% RECOVERY Figure 10. Exposure vs. recovery 0 superficial velocity, 0.45 ft/sec 0 superficial velocity, 0.28 ft/sec
Run 69 Run 70
and bed length. Zone length alone, then, is not a very suitable tool for multicomponent design purposes. The effects of zone length and bed length can be combined in an exposure-recovery (E-R) plot. Exposure is the pounds of propane-plus fed per 100 pounds carbon.
18
Ind. Eng. Chem. Process Des. Develop., Vol. 10, No. 1, 1971
Barrere, C. A., Hydrocarbon Process 47 ( l o ) , 141 (1968). Bray, R . K., Moffat, E. C., Greathouse, W. P., Cheek, J. H., “Operation of Fruita Gas Adsorption Plant,”
Proceedings of Forty-Fourth Annual Convention of the NGPA, Dallas, Texas (March 24-26, 1965). Campbell, J . M., Ashford, F . E., Needham, R. B., Reid, L. S.,Hydrocarbon Process Petrol. Refiner 42 (12), 89 (1963). Dale, G. H., Haskell, D. M., Keeling, H. E., Warzel, L. A . , Chem. Eng. Progr. Symp. Ser. 57 (34), 42 (1961). Michaels, A . S., Ind. Eng. Chem. 44, 1922 (1952).
RECEIVED for review June 4, 1969 ACCEPTED August 13, 1970