Professional Documents
Culture Documents
1, 1979
maintains that fact throughout the calculation. Actually, yielding a solution to each column in a few iterations. The
after each column solution, the distribution of all com- method has been used on systems in which the equilibrium
ponents can be checked by infinite section equations constants were independent of liquid composition, but it
utilizing the conditions a t the top and bottom of the could be readily extended to include such effects. One
calculated column, and the distribution of each component interesting result of the test calculations, which were done
can be easily found. This serves either to verify that a on hydrocarbon systems, was the finding that Underwood's
component is not distributing or to calculate the amount method was usually reasonably correct. If the minimum
in each product if it does distribute. sectional flows given by Underwood are assumed to be
Also, it is apparent that other specifications of the light flows at the respective pinch points, and energy balances
and heavy keys could be used, for example, the concen- are used to obtain flows and loads a t the top and bottom
tration of a key in a product, or even a total product of the column, the error was normally only a few percent.
amount or a product boiling point. However, setting the However, the calculation by Newton-Raphson is suffi-
amounts of the two key components in the products is the ciently rapid that there is no reason not to use the correct
most common way of specifying the separation. calculation. A typical close-boiling four-component system
The number of stages to be inserted between the feed which required solution of a 14-stage column, a 24-stage
stage and the corresponding infinite section in each ex- column, and a 34-stage column was solved in 12.5 s on a
pansion of the column is arbitrary. If the nondistributing CDC 6400 computer. A six-component system which
diluents are considerably different from the corresponding required a 14-stage calculation and a 24-stage calculation
key components, only a few plates are necessary. However, required 16.9 s. The authors have written a Fortran
since many distillation systems are fairly close-boiling, program for single-feed two-product columns for systems
there are often components which differ in volatility from up to ten components. The program is available a t no
a key component by only a small amount, perhaps l o % , charge on request.
and which still do not distribute. To reduce the amount
of these components at the ends of the calculated column
Literature Cited
Bachelor, J. B., Pet. Refiner, 36 (6), 161 (1957).
to amounts which do not affect the reboiler and condenser Erbar, R. C., Maddox, R. N., Can. J . Chem. Eng., 40, 25 (1962).
loads usually requires 10 to 15 stages. If the number of Naphtali, L. M., Sandholrn, D. P., AIChE J . , 17, 148 (1971).
stages added in each column expansion is small, many Newman, J., Ind. Eng. Chem. Fundam., 7, 514 (1968).
Ricker, N. L., Grens, E. A. 11, AIChE J . , 20, 238 (1974).
columns are calculated unnecessarily. The authors have Shiras, R. N., Hanson, D. N., Gibson, C. H., Ind. Eng. Chem., 42, 871 (1950).
found it reasonable to add five stages at a time. Underwood, A . J. V., J . Inst. Pet., 32, 614 (1946).
Underwood, A . J. V., Chem. Eng. Prog., 44, 603 (1948).
Results
The calculational procedure has been tested on many Received for reuiew February 9, 1978
examples and has been found to be extremely stable, Accepted August 7 : 1978
The design and operation of a PDU-scale tube wall methanation system are discussed. The tube wall reactor was
constructed from a stainless steel pipe with a surrounding jacket. The inside surface of the tube was flame sprayed
with Raney nickel catalyst for 14 ft of length. The catalyst coating was activated by leaching with a caustic solution.
A liquid coolant in the outer shell of the reactor removed the heat of methanation and maintained the system at
near isothermal conditions. Exposure velocity and recycle ratio were varied during the run. Concentration and
temperature profiles were periodically measured along the length of the catalyst bed. Catalyst performance, methane
production, deactivation, and other results are presented for the test which was terminated after 1179 h on stream.
condenser
T C wells
- Reactant pas
I
Dowlhcrm 4 s C h 40 plpe
reservoir dowlherm jacket
Reactor t
Used t u ~ s t i c
tokeoff
cootea internally
with Raney nickel
L
- 7
Product par
Table I. Catalyst Bed Data for TWR-Run 16 within 20 "C. The product gas exits the reactor and splits;
catalyst type flame sprayed Raney nickel one stream leaves the system and is flared and the other
weight percent 42a is recycled. The recycle stream is cooled and dried by a
percent activated 70.gb double pipe heat exchanger and then compressed. Flow
estimated catalyst 0.025 streams are metered and checked by orifices.
thickness, in, During the test volumetric gas samples which were
reactor diameter x 2.067 x 168
length, in. representative of a 24-h period were analyzed by both mass
void reactor volume, 0.311 spectrometry and gas chromatography. An on-line gas
ft3 chromatograph checked sample results. The gas analyses
weight of unactivated 1.63 along with condensed product water weights and metered
catalyst, lb flows were used to calculate the mass balances for each
geometric area of 7.39 24-h period.
catalyst, ft2
effective diameter, ft 0.168 A sample stream from the mixed gas before the reactor
was sent to an iron deposit unit consisting of a transparent
a Before leaching. Theoretical. Vycor tube heated to 300-350 "C in a furnace through
GOI which the sample stream flows. As shown in the past, any
prehealer
Fresh 90s iron or nickel carbonyl in the sample gas will decompose
n and form a metallic coating on the tube.
w
Results and Discussion
Reoclor
TWR-Run 16 was terminated after 1179 h on stream.
The following conditions were investigated during the
experiment: a fresh feed exposure velocity of 10, 20, 30,
comprersar Condenser
and 40 (1 unit exposure velocity = 1 SCFH of fresh
synthesis gas/ft2 of geometric catalyst surface area) and
Woler Product a cold-recycle-to-fresh-gasratio ranging from 3:l to 0. Inlet
gas temperature (av 353 " C ) , Dowtherm coolant tem-
90s
perature (av 375 "C), and system pressure (300 psig) were
Figure 3. Flow scheme of the tube wall reactor system. held reasonably constant throughout the test. Three brief
unscheduled power outages occurred but did not affect
reacted. Temperature of the leaching reaction never catalyst life. No shutdowns occurred during the run.
exceeded 50 "C. Previous experience indicated that the Operating parameters and product gas characteristics are
unleached Raney nickel substrate acted as an adhering shown in Figure 4.
agent between the leached catalyst and the bond coat. The following select representative periods are tabulated
When leaching was finished, the reactor was drained in Table I1 to provide more detailed information in run
under a helium atmosphere, and a continuous stream of 16.
deionized water was flowed over the catalyst. Many times Period 3 represents performance of the fresh catalyst
during the rinsing procedure the catalyst was batch flushed at a 10-exposure velocity and a 3.06:l recycle ratio.
with this water and always under a helium atmosphere. Period 8 indicates the effect of decreasing the recycle
No traces of spalled catalyst were found in the effluent ratio to 1.50:l at a constant exposure velocity of 10.
water. Washing was stopped after 93.5 h when the pH of Periods 11 and 18 show the results of decreasing the
the rinse water (6.7) was approximately the same as the recycle ratio from 1.54:l to 0.50:l at an exposure velocity
demineralized water (6.2). The leaching apparatus was of 20. The effect of increasing fresh feed can be compared
disassembled and the reactor was incorporated into the between period 8 and period 11.
system, all steps performed under a helium flow. The Periods 21 and 35 provide a comparison of identical flow
reactor system was then placed under hydrogen until the conditions a t a 30-exposure velocity and a 0.50:l recycle
temperature and pressure were brought to methanation ratio but at different times.
conditions, at which time the synthesis feed gas was Period 26 represents performance a t an exposure ve-
gradually fed into the system to start the run. Pertinent locity of 30 with no recycle.
catalyst bed data are shown in Table I. Periods 41 and 44 represent performance at the highest
Reactor System exposure velocity attained (40) and the change of recycle
The synthesis gas in the experiment was made by steam from 0.25:l to 0 a t the end of the run.
reforming natural gas in a Girdler plant. The hydrogen Since carbon monoxide was the limiting reactant, a good
to carbon monoxide ratio ranged from 3.1:l to 3.3:1, always indication of the Raney nickel catalyst performance was
with a slight excess of hydrogen to prevent possible carbon the percent carbon monoxide in the product gas. In
deposition. Two activated carbon traps were located in general, an increase in exposure velocity a t a constant
the fresh gas inlet line and one in the product recycle line. recycle ratio brought about an increase in product CO
This precaution was taken to eliminate the possibility of concentration. At 263 h on stream, the fresh gas was
sulfur contamination of the nickel catalyst. Sulfur con- increased from an exposure velocity of 10 to 20 a t 1.50:l
centrations entering the system were maintained around recycle ratio and the product gas CO concentration in-
66 ppb and were detected by the methylene blue method. creased from 0.13 to 0.52%. At 503 h the exposure velocity
Figure 3 illustrates a simplified flow diagram of the tube was boosted from 20 to 30 a t 0.50:l recycle ratio with an
wall methanation system. Fresh synthesis gas enters the increase in product gas CO concentration. At 959 h the
system and mixes with a dry product gas recycle stream 30-exposure velocity condition was raised to 40 and again
or "cold recycle" stream. In TWR-run 16 this mixed gas the greater the fresh feed the lower the CO conversion.
stream was preheated through several heat exchangers and Although the fresh feed was decreased at 1127 h from a
entered the reactor a t 353 "C. The boiling Dowtherm 40-exposure velocity to 30, the percent CO in the product
coolant system was maintained a t 375 "C and controlled did not decrease as would be expected. The catalyst was
the maximum temperature variance of the catalyst to appreciably deactivated a t this point.
Ind. Eng. Chem. Process Des. Dev., Vol. 18, No. 1, 1979 159
2 51 I I I I 1 nrl
Exp TWR-16
1
400 .-
Maximum catalvst 1
350
c
300 t 05
1000
W Product qos heating value
U
1
vi 800
a
W
l/"t
the CO percent in the product increased. Later the CO
concentration decreased when the recycle was brought Figure 5. Least-squares fit through plotted representative data.
back to 1.5:l. A t a 30-exposure velocity the recycle ratio by dimensionlesss group analogies between heat and mass
was changed from 0.51 to 0 back to 0.51. Here the best transfer. Specifically, the following correlation of Sieder
reactor performance was at the no recycle level. A decrease and Tate for laminar flow was used (Bennett and Myers,
in recycle at an exposure velocity of 20 or 40 also brought 1962).
(;y3(
about a lowered product CO percent and thus a better CO
conversion.
From the previous tlndings a simple plug flow model can Nu = 1.86Re1/3Sc1/3 $14 (2)
be developed. For a catalyst sprayed tubular reactor, mass
transfer of reactants and products to and from the tube Finding the mass transfer coefficient and substituting
wall becomes an important factor. In the reactor the bulk
gas diffusion resistance is assumed much greater than that
of the surface reaction, and thus the global rate of reaction
is limited by the rate of flow of reactant to the surface.
SD0.667
2*37D2.333L0.333
(:)"'I
into the design equation yields
- 1-
u0.667
-
0 - 5
2 4 6 8 IO 12 e1;4
,TIPI
REACTOR LENGTH, f e e t
Figure 6. Actual gas concentration profiles for two different times Figure 7. Actual temperature profiles for various times on stream.
on stream but same exposure velocity and recycle ratio.
The design equation can further be simplified in terms
of the fresh synthesis gas rate and the recycle flow. It can
be seen that the reactor CO conversion decreases with an
increase in fresh feed or an increase in cold recycle. These
results have been verified as shown before in the data.
Besides the predominant methanation reaction a sec-
ondary reaction is noticeable due to the CO, concentration
in the product gas. The "water gas shift" reaction com-
bines CO and HzO to form COPand Hz 0 240 480 720 960 I200
CO + HzO + COZ + Hz (4) 7 ME ON STWEAM hour5
Table IV. BET Surface Areas, Pore Volumes, and Pore advantage in future scale-up work. Even without the use
Radii of Spent Raney Nickel Catalyst of a recycle flow, catalyst temperatures can be controlled
reactor reactor with relative ease. The reactor is capable of converting
top bottom fresh feeds with a CO concentration as high as 25%. With
2 ft middle 2 ft different flows there is negligible pressure drop across the
(inlet) 2 ft (outlet) reactor. Recent improvements in the catalyst flame
surface area. m2/e 43.17 39.40 28.80 spraying techniques and equipment have decreased
pore volume, cA'/g 0.118 0.121 0.132 spraying laydown time by 20-fold. Although this particular
average pore radius, 54.88 61.22 91.87 run lasted only 1179 h with a wide range of fresh feed and
A recycle flows, the catalytic reactor did show relative
percent pore volume stability during the test with little difficulty in operation.
with the following
radii: Since run 16, the process development system has been
<30 A 12.69 6.04 0 modified. Vessels and piping of carbon steel have been
30-40 A 14.14 13.74 0 replaced with stainless steel to retard or eliminate the
40-50 A 11.45 12.14 4.09 formation of the iron carbonyl poison. A future test will
50-60 A 9.91 10.69 11.69 utilize this new system.
>60 A 51.81 57.39 84.22
percent total surface Nomenclature
area with the
following radii: D, inside diameter of reactor, ft
<30 A 26.12 13.46 0 D, diffusivity, ft2/s
30-40 A 22.46 24.20 0 e, contraction factor
40-50 A 14.04 16.59 8.25 h , mass transfer coefficient, ft/s
50-60 A 9.93 11.95 19.55 L , catalyst length, ft
>60 A 27.45 33.80 72.20 Nu, Nusselt number
Sc, Schmidt number
lowest A1 content (28.5%). Also, X-ray diffraction results Re, Reynold number
indicate A1-Ni compounds, which are found in unleached Q, reactor inlet volumetric flow, ft3/s
Raney nickel, in the top section of the catalyst. S, geometric catalyst surface area, ft2
The other possibility is replacement of the active nickel wo, catalyst surface gas viscosity, lb/ft-s
W , bulk gas viscosity, lb/ft-s
a t the reactor inlet by migration of the aluminum from u , reactor inlet gas velocity, ft/s
pockets of unleached catalyst or from aluminum complexes Xco, CO conversion
formed by leaching. Most of the reaction took place along
the catalyst top length during the experiment and the local L i t e r a t u r e Cited
heat of reaction could provide a driving force for the Bennett, C. O., Myers, J. E., "Momentum, Heat, and Mass Transfer", Mc(3awHIl1,
New York, N.Y.. 1962.
migration. Haynes, W. P., Elliott, J. J., Forney, A. J., Am. Cbem. Soc. Div. FuelChem.
Although metal surface area measurements were not Prepr., 16 (2),47 (1972).
Haynes, W. P., Forney, A . J. Elliott, J. J.. Pennline, H. W., Am. Chern. Soc.
performed on the spent catalyst samples, BET surface Div. Fuel Chern. Prepr.. 19(3), 10 (1974).
measurements were obtained and shown in Table IV. The Haynes, W. P., Schehl, R. R., Weber, J. K.,Forney, A. J., Ind. Eng. Chem.
surface areas of the samples are not as great as freshly Process Des. Dev., 16, 113 (1977).
Savinell, R. F., Youngblood, A. J.. Haynes, W. P., ERDA PERU RI-7614,(1976).
leached Raney nickel (64 m2/g). However, there is a Schehl, R. R., Pennllne, H. W., Strakey, J. P., Haynes, W. P., Am. Chern. Soc.
decreasing trend from reactor inlet to outlet. The high Div. Fuel Cbem. Prepr., 21 (4),2 (1976).
Schehl, R. R., Pennline, H. W., Youngblood, A. J., Baird, M. J., Strakey, J. P.,
amount of carbon at the inlet would account for the large Haynes, W. P., ERDA PERC/RI-77/10, (1977).
surface area. Amorphous carbon would also indicate the
smaller pore radii associated with carbon a t the reactor Received f o r review April 7, 1978
inlet as compared to the outlet. Accepted July 20, 1978
Summary Presented at the Division of Colloid and Surface Chemistry, 2nd
The tube wall methanator provides an efficient method Joint Conference Chemical Institute of Canada/American
of heat removal and a relative simple design which is an Chemical Society, Montreal, Canada, May 1977.