I n d . Eng. Chem. Res. 1989,28, 1567-1570
1567
n = integer exponent in eq 2 t = deviation, %
N = number of experimental data points Subscripts exp = experimental data cal = calculated value from eq 2
min = minimum pressure of measurement Registry No. LiBr, 7550-35-8; LiC1, 7447-41-8;ZnBr2,769945-8.
Literature Cited
55
60
65
70
75
Absorbent concentration (wt Yo)
Bach, R. 0.;Boardman, W. W. Vapor Pressure of Aqueous Lithium Iodide Solutions. ASHRAE J . 1967,11,33-36. Boryta, D. A.; Maas, A. J.; Grant, C. B. Vapor Pressure-Temperature-Concentration Relationship for System Lithium Bromide and Water (40-7070 Lithium Bromide). J . Chem. Eng. Data 1975,20,316-319. Ivoki. S.: Hanafusa. Y.: Koshivama. H.: Uemura. T. Studies on the Water-Lithium Bromide-lithi& Thiocyanate Absorption Refrigerating Machine. Reito 1981,56,661-671. Matsuda, A,; Munakata, T.; Yoshimaru, T.; Kubara, T.; Fuchi, H. Measurement of Vapor Pressures of Lithium Bromide-Water Solutions. Kagaku Kogaku Ronbunshu 1980,6,119-122. Pennington, W. How to Find Accurate Vapor Pressures of LiBr Water Solutions. Refrig. Eng. 1955,63,57-61. Steam Tables;The Japan Society of Mechanical Engineers: Tokyo, 1980. Takagi, S. Teiryo Bunseki no Jikken to Keisan; Kyoritau Shuppan: Tokyo, 1976. Takigawa, T. Master's Thesis, Kansai University, Osaka, Japan, 1988. Timmermans, J. Physico-Chemical Constants of Pure Organic Compounds; Elsevier: Amsterdam, 1965. Uemura, T.; Hasaba, S. Studies on the Lithium Bromide-Water Absorption Refrigerating Machine. Technol. Rep. Kansai Univ. 1964,6,31-55. Ueno, K. Kireito Tekiteihou; Nankodo: Tokyo, 1976. I
Figure 3. Comparison of vapor pressures for HzO-LiBr-ZnBrz-LiC1 system: (-) calculated values from eq 2,(- - -) extrapolated values from the literature (Takigawa, 1988).
absorbent were measured at low temperatures. Experimental data were correlated by means of an Antoine-type equation. The calculated values from this equation were in good agreement with the experimental data. Maximum and average absolute deviations between the experimental data and the calculated values from this equation were 2.45% and 0.95%,respectively. The vapor pressure data for this four-component system at low temperatures are very useful for the design of the absorber of absorption refrigerating machines, absorption heat pumps, and absorption heat transformers. Nomenclature V = volume of the apparatus, cm3 X = absorbent concentration, w t % T = absolute temperature, K p = vapor pressure, Pa A,, B, = constants in eq 2 u = volumetric air flow rate leaked on the vacuum side in the apparatus, cm3-Pa/s 0 = time of measurement, s b = allowable pressure elevation, %
.
I
Shigeki Iyoki,* Shozo Iwasaki, Tadashi Uemura Department of Chemical Engineering Faculty of Engineering Kansai University Yamate-cho, Suita, Osaka 564,Japan Received for review December 20, 1988 Revised manuscript received June 28, 1989 Accepted July 18,1989
Laboratory Evaluation of Clays in the Treatment of Benzene-Toluene-Xylene Feedstocks Clays are used to treat BTX (benzene-bluene-xylene) feedstocks in refineries. During clay treatment, the olefin content of BTX as measured by bromine index is reduced. A BTX aging test is described which can differentiate the olefin removal performance of clays under accelerated aging conditions. The purpose of this paper is to describe a laboratory test for the comparative evaluation of clays in BTX clay treatment under accelerated aging conditions. Test equipment, procedures, and measures of clay performance will be described. Data will be presented to demonstrate test reproducibility, the effect of process variables, and the ability of the test to differentiate olefin removal performance of clays in the treatment of BTX feedstocks. BTX is an acronym for benzene-toluene-xylene. Petroleum-sourced BTX is produced by catalytic reforming and pyrolysis. The BTX produced is used as an octane booster in gasoline and as a feedstock for the production of plastics and fibers (Ransley, 1978). Clay treatment of BTX is practiced when it is desired
to clean up a BTX stream. During clean-up, the olefin content of BTX is reduced. Previous Work Olefin Removal Mechanisms. Reidel (1954)stated that BTX impurities polymerize during clay treatment and are subsequently removed by downstream distillation. Kawakami et al. (1972)concluded that primarily heptene dimers were formed by polymerization when heptene-2-spiked toluene was passed in the liquid phase over a clay at 200 "C. These investigators published a gas chromatograph (GC) trace with four unknown "heavy" peaks. Their unpublished mass spectroscopy data led them to conclude that some of these peaks corresponded
0888-5885/89/2628-1567$01.50/0 0 1989 American Chemical Society
1568 Ind. Eng. Chem. Res., Vol. 28, No. 10, 1989
to heptene dimers. In the absence of published mass spectroscopy data, one can speculate about another interpretation. Since toluene and heptene are both C7's and toluene is in excess, some of the heavy peaks observed by Kawakami et al. (1972) may have corresponded to toluene-heptene alkylation products. Interpretation of the olefin removal mechanisms in the work by Kawakami et al. (1972) was confounded by the number of potential reaction products in the tolueneheptene-2 system. Previous work (unpublished) in our laboratory was conducted with a system with fewer potential reaction products. A batch autoclave reaction was run a t 175 "C using a benzene-cyclohexene (98%-2%) mixture with a granular clay contained in an internal basket. Product samples were analyzed as a function of time by GC. As the reaction proceeded, the GC peak corresponding to cyclohexene decreased and a single peak corresponding to cyclohexyl benzene grew. These findings suggest that alkylation was the only apparent major olefin reduction reaction occurring under these conditions. In view of the Bronsted and Lewis acidities of BTX clays (Ishiguro et al., 1982) and the high ratio of BTX to olefins, it is not surprising that BTX-olefin alkylation can occur. Clay Aging Mechanisms. Potential clay deactivation mechanisms include deactivation of active sites by poisoning or carbon deposits and pore plugging by carbon or polymer buildup. Ishiguro et al. (1982) studied the effect of heavy aromatics on clay life by a model system of heptene-2-spiked benzene with various components at the 2 wt % level. Fixed bed experiments were conducted a t 180 "C,210 psig, and LHSV = 2. Anthracene was found to significantly deactivate the clay. Ishiguro et al. suggested that deactivation with anthracene occurred by adsorption on Lewis acid sites with a concurrent reduction in the activity of neighboring Br~nstedsites due to steric hindrance.
BTX Clay Aging Test The key performance criteria for BTX clays is clay life because of the impact on BTX treatment cost. Activity and selectivity of the clay catalyst is generally not an issue in BTX treatment. As a supplier of BTX clays, it was desirable to develop a laboratory test to evaluate the comparative performance of clays in the treatment of BTX feedstocks. The test development objectives were the following: (1)differentiate the olefin removal performance of clays in the treatment of BTX feedstocks under accelerated aging conditions; (2) minimize the time and cost per test; (3) minimize the BTX feedstock consumed per test. The above objectives indicated that a microreactor rapid aging test was needed. During test development, a microreactor test system was built, and the liquid hourly space velocity (LHSV) was increased well beyond typical commercial LHSV's to meet the test objectives. BTX Test Equipment and Procedures. A simplified schematic diagram of the BTX test equipment is illustrated in Figure 1. The equipment consists of four major components: feed system, reactor system, product recovery system, and monitoring and control system. The monitoring and control system allows unattended operation by providing safety shut-down and data logging capabilities. If preset temperature or pressure limits are exceeded, the power to the equipment is shut off, the BTX feed valve is closed, and the system vent valve is opened. Glass burets are used for feed reservoirs and flow rate monitoring. Stainless steel is used for the balance of the wetted equipment. The fixed bed microreactor has a clay capacity of 5 cm3 and operates in a down-flow mode. A
(w)
FEED SYSTEM
E FEEDBURET I
\ CONYROL /
n
PRODUCT RECOVERY SYSTEM A
+-
POSITIVE DISPLACEMENT
I
F.0
Y
L
REACTORTUBE
1
SAMPLING LINE
REACTOR SYSTEM
Figure 1. BTX test equipment.
100-mesh stainless steel screen is used as a bed support. Thermocouples located at the reactor inlet and in the hot oil bath are automatically monitored and logged. Execution of a clay aging test consists of a start-up, run, and shut-down procedure. During start-up, BTX is fed to the hot reactor at atmospheric pressure in order to flush noncondensable gases out of the system. The reactor is then temporarily dead-ended to allow pressure to build to the normal 200 psig operating pressure. This pressure was chosen to ensure liquid-phase operation. At 200 psig, a valve is opened, which allows the reactor product stream to flow into the product receiver which is maintained a t 200 psig with Nz. As BTX product fills the receiver, displaced N2 is vented through a back-pressure regulator to maintain 200 psig. A dozen BTX product samples are typically collected from the receiver over the course of a test and analyzed for the bromine index. At the end of the test, the system is depressurized and the reactor is removed from the hot oil bath. The bromine index data on the BTX product as a function of time, or BTX throughput, form the basis for evaluating clay performance in this test. The bromine index is a broad indication of olefin content. The reaction of olefins with bromine is an electrophilic addition to the carbon-carbon double bond. An automated version of ASTM D 1491-78 is used to determine the bromine index of product samples. The bromine index is defined as the number of milligrams of bromine consumed per 100 g of sample under given conditions. In this procedure, the bromine index is determined by potentiometric titration. During titration, bromine is generated. As the rate of bromine consumption by the BTX product sample slows down, the potential of the titration mixture begins to change. The potentiometric end point is determined by the inflection point or the parameters set in the automatic titrator. General Test Conditions. Microreactor, 1/2-in.0.d. X 3/8-in. i.d. x 3 in.; amount of clay, up to 5 cm3 (load by weight based on packed apparent bulk density); feedstock, variable BTX sources; pressure, 200 psig; temperature; 150-200 OC; LHSV (h-l), adjustable to get a Br index of 5 or more. The actual feed rate or LHSV is dependent on the feedstock, clay, amount of clay, and temperature. Data Analysis and Measures of Clay Performance. Raw data consist of the BTX product bromine index as a function of time or BTX throughput (cm3 of BTX fed per cm3 of clay bed). One approach to comparing clays would be to make comparative plots of the product bromine index versus BTX throughput. This approach was abandoned in favor of a more fundamental approach. An apparent kinetic rate constant was chosen as a measure of activity and plotted against the BTX
1
Ind. Eng. Chem. Res., Vol. 28, No. 10, 1989 1569
--I
TEST CONDITIONS: TEMPERATURE = 1854: LHSV 12hr" CLAY = A = B FEEDSlOCK
2
-
0
4
+
200
++:
+
6
5 +.
+ +
+
I 18Ag4:
FEEDmCKacwY TEMPERATURE
-
+
LHSV = 100 hr , 4 ccCLAY
c
LHSV
200 hr.', 2 CCCLAY
(Linear Velocity is held constant)
++=
+++ + ++
+
+=
0
-
+t
+ +
--
0
c 0
B 0 0
500 BTXTHROUGHPUT (cdcc)
1
1
,
l
1
1
1
1
1000
Figure 2. Performance test reproducibility.
throughput. The advantage of this approach is that intrinsic particle kinetic rate constants can be determined by subtracting inter-particle mass-transfer intrusions from the apparent bed kinetics (Levenspiel, 1979). This has the advantage of making the test more discriminating. It also offers the potential of correlating the intrinsic particle kinetic rate constant with catalyst models that incorporate the fundamental properties of the clay. First-order and second-order kinetic models were considered. Kinetics that are first order in bromine index should correspond to the alkylation of BTX by small amounts of a specific olefin. Since BTX is in excess, the BTX concentration remains essentially unchanged during the alkylation reaction and the reaction should appear to be first order in olefin concentration. A second-order model would correspond to the dimerization of a specific olefin. The rate constants for first- and second-order models are given below for an isothermal plug flow reactor: first order LHSV[ln (Co/C)] = K (1) second order LHSV[Co/C - 11 = K'Co
(2)
where Co is the bromine index of the feed BTX, C is the bromine index of the product BTX, LHSV is the liquid hourly space velocity (cm3of BTX/(h.cm3 of clay bed)), K is the apparent first-order rate constant (h-l), and K' is the apparent second-order rate constant ((hebromine index)-'). The appropriate model and the effect of inter-particle mass transfer will be discussed later in this paper. Another advantage of this fundamental approach is the inherent ability of the model to correct for possible bromine index instrument drift in the determination of rate constants. Instrument drift between experiments causes random variation in the measured bromine index. The division of Co by C was found to cancel out the effect of instrument drift on the determination of rate constants. Figure 2 illustrates BTX aging test data as a plot of rate constant versus BTX throughput for a first-order reaction. The rate constant used in all graphs in this paper is the apparent rate constant defined in eq 1. In this and subsequent graphs, the proprietary feedstocks and clays are identified as A, B, C, etc. The information in Figure 2 allows one to obtain two measures of clay performance in the BTX aging test: (1) initial activity or rate constant and (2) activity decay rate. The initial activity is obtained by extrapolating the rate constant back to zero BTX throughput. The activity decay rate is the slope of the decay line.
The constant activity decay rate demonstrated in Figure 2 can be shown to correspond to a sharp deactivating front moving down the clay bed at constant velocity. Wheeler and Robell (1969) have developed a more complex catalyst deactivation model which includes four mechanistic parameters. Their model predicts an initial constant activity decay rate followed by a continual reduction in the activity decay rate. BTX Aging Test Reproducibility. Figure 2 also demonstrates the reproducibility of the BTX aging test. Extrapolation of the least-squares lines to the y and x intercepts provides an indication of reproducibility in terms of initial activity and extrapolated life at zero activity. They intercept in Figure 2 ranges between 58 and 62 h-' and the x intercept ranges between 650 and 810 cm3/cm3. It should be emphasized that these laboratory numerical results cannot be directly translated to commercial numerical values because the laboratory velocities and bed lengths are much smaller than commercial values. Effect of Bed Length. One method of evaluating the kinetic models represented by eq 1and 2 is to vary the bed length and hence LHSV at fixed velocity. Operation at fixed velocity eliminates the possible variation of the apparent rate constant with velocity due to the effect of inter-particle mass transfer. Under constant velocity conditions, the apparent rate constant should be independent of bed length as LHSV is varied if the model is correct. Figure 3 illustrates the effect of bed length at fixed velocity with the first-order model (eq 1). These data are for clay A and feedstock A. The longer bed (4 cm3) data appear to deviate negatively from the shorter bed (2 cm3) data. These results indicate some apparent deviation from the first-order model (eq 1). It can be shown in a test of the second-order model (eq 2) that positive deviations of even greater magnitude occur. These findings suggest that the apparent kinetics are closer to first order. Calculations using an axially dispersed first-order model (Montagna and Shah, 1975) demonstrate that a plug flow model is valid under our conditions. Therefore, confounding end effects should not be causing deviations from first-order kinetics. Deviations from a first-order model may be due to the use of a "real" feedstock containing a mixture of olefins of varying reactivities. Under these conditions, the observed kinetics of an olefin mixture will appear to be greater than first order. Schuit and Gates (1973) demonstrated how first-order reactions of species with varying reactivities can give the appearance of higher order kinetics for the mixture.
1570 Ind. Eng. Chem. Res., Vol. 28, No. 10, 1989 TEST CONDITIONS TEMPERATURE = 185% = 12 hr
FEEDSOCK a CLAY = A TEMPERATURE = 185% +
+
200 -
-
L n s - 100 hr LHW
.tisI
4 ccCLA'Y
50 hr
4 ccCLAY
25 hr
4 ccCLAY
FEEDSTOCK
8
2;
40
1 +
5
=
B
CLWD
iu
h
moo
1000
0
500
Figure 6. Comparative performance.
Figure 4. Effect of velocity. I
I
moval performance of clays under accelerated aging conditions. Commercial information has supported the ability of this accelerated clay aging test to differentiate and rank olefin removal performance of clays during BTX clay treatment. The Wheeler-Robe11 (1969) model can be used to curve fit the activity-aging curves in Figures 2-6 by using a least-squares-based parameter estimation technique. Unfortunately, one cannot guarantee that the four mechanistic Wheeler-Robe11 parameters are unique.
TESTCONDITIONS
TEMPERATURE =
150 a
=
12 h r '
=
A
=
B
0
4
1000
BTXTHROUGHPUT (cclcc)
BTX THROUGHPUT (cclcc)
FEEDSTOCK
t 1854:
5 -+
8
500
0
1000
BTX THROUGHPUT (cclcc)
Figure 5. Effect of temperature.
Effect of Velocity. Figure 4 demonstrates a significant effect of the velocity on the extrapolated initial activity (rate constant) and on the activity decay rate (slope). The shape of the activity decay curves in Figures 3 and 4 is consistent with the previously discussed Wheeler-Robe11 theory-constant activity decay rate followed by declining activity decay rate. It is evident that the extrapolated initial rate constants increase significantly with velocity. This behavior demonstrates the classical indication of inter-particle masstransfer intrusions, which can decrease the sensitivity of an activity test. The activity decay rate can also be seen to change with velocity. A t higher velocities, curvature of the activity decay curves becomes noticeable. This behavior is consistent with the Wheeler-Robe11 theory for first-order catalytic reactions, with adsorptive poisoning occurring in fixed bed reactors. As velocity increases, the shape of the poison profile in the bed broadens and results in the curvature of the activity decay curve. Effect of Temperature. Figure 5 demonstrates the effect of temperature on the extrapolated initial activity and the activity decay 'curves. The initial activity appears to be relatively insensitive to temperature. However, aging appears to be significantly affected by temperature. This behavior suggests that initial activity is essentially interphase mass-transfer controlled, and fewer active sites are effective for olefin removal at lower temperatures. This might be due to enhanced poison adsorption and/or sites that are inherently less active at lower temperatures due to stronger adsorption of reactants or products. Comparative Performance of Clays. Figure 6 demonstrates the ability of the accelerated aging test to differentiate the olefin removal performance of clays during BTX clay treatment. The region labeled "clay A" encloses all of the data from the five nonsequential runs illustrated in Figure 2 . The test clearly differentiates the olefin re-
Conclusions A test procedure was developed that can differentiate the olefin removal performance of clays during BTX clay treatment under accelerated aging conditions. The small equipment size and accelerated aging minimize the time and BTX consumed per test. Although alkylation and polymerization are potential olefin removal mechanisms, alkylation appears to be the most likely major olefin removal mechanism. Acknowledgment The authors express their appreciation to Engelhard Corporation for granting permission to publish this work.
Literature Cited Ishiguro, T.; Kato, T.; Kataoka, K. A Study of Clay Deactivation. Aromatikkusu 1982, 34(1/2), 12-15. Kawakami, H.; Teramato, K.; Soda, M.; Hanano, K.; Abe, K. The Function of Activated Clay in the BTX Refining Process: 1. The Mechanism of Removal of Olefins. Sekiyu Gakkaishi 1972,15(9), 763-766. Levenspiel, 0. The Chemical Reactor Omnibook; OSU Bookstores: Corvallis, OR, 1979. Montagna, A. A.; Shah, Y. T. The Role of Liquid Holkup, Effective Catalyst Wetting, and Backmixing on the Performance of a Trickle Bed Reactor for Residue Hydrodesulfurization. Znd. Eng. Chem. Process Des. Dev. 1975, 14(4), 479-483. Ransley, D. L. BTX Processing. In Kirk-Othmer Encyclopedia of Chemical Technology, 3rd ed.; John Wiley & Sons: New York, 1978; Vol. 4. Reidel, J. C. 53 Million Gallons of BTX Annually: Capacity of Sun's Marcus Hook Refinery. Oil Gas J. 1954,52(47), 94-96. Schuit, G. C. A.; Gates, B. C. Chemistry and Engineering of Catalytic Hydrodesulfurization. AZChE J. 1973, 19(3), 417-438. Wheeler, A,; Robell, A. J. Performance of Fixed-Bed Catalytic Reactors with Poison in the Feed. J. Catal. 1969, 13, 299-305.
Lawrence T. Novak,* Kasey F. Petraitis Engelhard Corporation Engelhard Research & Development Center 23800 Mercantile Road Beachwood, Ohio 44122 Received for review November 21, 1988 Accepted June 21, 1989
