Ind. Eng. Chem. Res. 1988,27, 1095-1103
Alfano, 0. M.; Romero, R. L.; Cassano, A. E. Chem. Eng. Sci. 1985, 40,2119. Alfano, 0. M.;Romero, R. L.; Cassano, A. E. Chem. Eng. Sci. 1986a, 41, 1155. Alfano, 0. M.; Romero, R. L.; Negro, A. C.; Cassano, A. E. Chem. Eng. Sci. 1986b,41,1163. Astarita, G. Znd. Eng. Chem. 1966,58,18. Astarita, G . Mass Transfer with Chemical Reaction; Elsevier: Amsterdam, 1967. Astarita, G.; Savage, D. W.; Bisio, A. Gas Treating with Chemical Soluents; Wiley: New York, 1983. Bhagwat, W. V.; Dhar, N. R. J. Indian Chem. SOC.1932,335. Calderbank, P. H.; Moo-Young, M. B. Chem. Eng. Sci. 1961,16,39. Chiltz, G.; Goldfinger, P.; Huybrechts, G.; Martens, G.; Verbeke, G. Chem. Reu. 1963,63,355. Claril, M. A. Ph.D. Dissertation, Universidad Nacional del Litoral, Santa Fe, Argentina, 1984. Clegg, G. T.; Tehrani, M. A. J. Chem. Eng. Data 1973,18,59. De Bernardez, E. Ph.D. Dissertation, Universidad Nacional del Litoral, Sante Fe, Argentina, 1984. Felder, R. M.; Hill, F. B. Znd. Eng. Chem. Fundam. 1970,9, 360. Gavalas, G.R. Chem. Eng. Sci. 1966,21,133. Harada, J.; Akehata, T.; Shirai, T. Kagaku Kogaku 1971,35,233. Harris, P. R.;Dranoff, J. S. AZChE J. 1965,11, 497. Hill, F. B.; Felder, R. M. AZChE J. 1965,11, 873. Huybrechts, G.;Meyers, L.; Verbeke, G. Trans. Faraday SOC.1962, 58. 1128.
1095
Irazoqui, H. A.; Cerdl, J.; Cassano, A. E. AZChE J. 1973,19, 460. Jacob, S.M.; Dranoff, J. S. Chem. Eng. Prog. Symp. Ser. 1968,64, 54. Matsuura, T.; Smith, J. M. AZChE J. 1970,16, 321. Otake, T.; Tone, S.; Higuchi, K.; Nakao, K. Kagaku Kogaku Ronbunshu 1981,7,57. Otake, T.; Tone, S.; Higuchi, K.; Nakao, K. Int. Chem. Eng. 1983, 2.-7,.288. - -.
Ramage, M. P.; Eckert, R. E. Znd. Eng. Chem. Process Des. Deu. 1973,12, 248. Romero, R. L.; Alfano, 0. M.; Marchetti, J. L.; Cassano, A. E. Chem. Eng. Sci. 1983,38, 1593. Schaftlein, R. W.; Rusell, F. T. W. Znd. Eng. Chem. 1968,60, 12. Spadoni, G.; Bandini, E.; Santarelli, F. Chem. Eng. Sci. 1978,33,517. Stramigioli, C.; Santarelli, F.; Foraboschi, F. P. Zng. Chim. It. 1975, 11, 143. Yokota, T.; Iwano, T.; Deguchi, H.; Tadaki, T. Kagaku Kogaku Ronbunshu 1981, 7,157. Yokota, T.; Iwano, T.; Tadaki, T. Kagaku Kogaku Ronbunshu 1976, 2,298. Yokota, T.; Iwano, T.; Tadaki, T. Kagaku Kogaku Ronbunshu 1977, 3,248. Zolner, W. J., 111; Williams, J. A. AZChE J. 1971,17, 502. Received for review May 8, 1986 Revised manuscript received October 12, 1987 Accepted November 2, 1987
Modeling of a Gas-Liquid Tank Photoreactor Irradiated from the Bottom. 2. Experiments Orlando M. Alfanot and Albert0 E. Cassano*t INTEC,%Casilla de Correo No. 91, 3000 Santa Fe, Argentina
The experimental verification of the models developed in part 1through the photochlorination of trichloroethylene and pentachloroethane in carbon tetrachloride solution was performed. The dominant subregime for each reaction was studied, finding that the chlorination of trichloroethylene was under a diffusional subregime and the chlorination of pentachloroethane was mainly under a kinetic subregime. Afterward, a comparison between the experimental values and the predictions obtained with the theoretical models was performed. The agreement found in the addition step is only attributable to the knowledge of the mass-transfer parameters. Instead, the agreement observed for the substitution step is mainly due to the accurate evaluation of the reaction rate and, therefore, through the initiation rate, to the proper prediction of the volumetric rate of radiant energy absorption. Photochlorination reactions of saturated or nonsaturated hydrocarbons share the common feature of presenting reactants and products with highly corrosive or very agressive solvent properties. The presence of chlorine, hydrochloric acid, and chlorinated derivatives creates difficulties for the selection of the materials for the equipment that has to be in contact with these chemicals. Furthermore, in the case of gas-liquid heterogeneous systems, the need for vigorous stirring of the reacting mixture requires the utilization of a hermetically sealed device to avoid loss of corrosive gases at the stuffing box of the stirrer shaft. Different alternatives have been used to solve those problems, among which we could mention the following: liquid-phase chlorination of n-dodecane in a stirred tank reactor irradiated from the bottom by means of a UV 'Research Assistant from CONICET and U.N.L. Member of CONICET's Research Staff and Professor at U.N.L. Instituto de Desarrollo Tecnoldgico para la Industria Qdmica. Universidad Nacional del-Litoral (U.N.L.) and Consejo Nacional de Investigaciones Cientificas y TBcnicas (CONICET). 0888-5885/88/2627-1095$01.50/0
radiation source (Ramage and Eckert, 1975, 1979), the photochlorination of trichloroethylene in a stirred tank reactor irradiated from the outside by eight UV sources parallel to the reactor axis and-equally spaced one from the other (Gebhard, 1978),and the chlorination of toluene in an annular reactor of the bubbling column type with the lamp located at the axis (Yokota et al., 1981b, 1983). In the present work, a semibatch stirred tank reactor irradiated from the bottom by a UV source located at the focal axis of a cylindrical reflector of parabolic cross section has been used. This means that we have tried not only to avoid the introduction of the lamp inside the reactor but to use a reflecting mirror as well, with the object of improving the device irradiation efficiency. The main purpose of this study is to investigate, from an experimental viewpoint, the behavior of the gas-liquid tank photoreador which was mathematically modeled and described in part 1. The idea is to verify which one of the two limiting representations for the mixing states can better describe the performance of a well-stirred twephase photochemical reactor and, at the same time, the quality of the radiation model to predict the effect of the UV light in a heterogeneous system. In order to achieve these ob0 1988 American Chemical Society
1096 Ind. Eng. Chem. Res., Vol. 27, No. 7, 1988
Figure 1. Flow sheet of the experimental device.
jectives, the verification of the models previously proposed in part 1 was carried out through the photochlorination reactions of trichloroethylene and pentachloroethane in the reactor described above. It is possible that real situations could fall between both limiting cases of mixing, but if at least one of the reactions is kinetically controlled, the results should be very useful for a rigorous test of the quality of the radiation model to predict the effect of radiation in a system where Lambert's equation does not strictly apply. Hence, the dominant subregime (mass transport or chemical reaction) was studied for each reaction through the modification of key operating variables of the system (agitation rate, composition of the feed gas, change of the radiation field, etc.). Afterward, the experimental data that were obtained were compared with the theoretical predictions for two limiting cases of mixing: perfect mixing and no mixing of atomic and free-radical intermediates. Both models require the description of the radiation field in the heterogenous system. With this purpose, the rate of the initiation reaction was obtained by applying the Extense Source Model with Volumetric Emission (ESVE model) developed by Irazoqui et al. (1973). Thus, predictions of the behavior of the two-phase photoreactor were made by starting from fundamental principles, using the mechanistic sequence of the reaction and their respective reaction rate constants and avoiding the introduction of any experimentally adjustable parameters. Consequently, the proposed methodology opens a route for reliable design procedures in the area of gasliquid photochemical reactions.
Experimental Setup The experimental device is schematically illustrated in Figure 1. The gas feed in the more general case consists of a fixed mixture of chlorine and nitrogen. The nitrogen obtained from the cylinder (1) was of 99.99% purity and is fed at a pressure of 27 kPa by means of the pressure regulator (R). The nitrogen stream afterward goes through the trap (6) in order to prevent the back-flow of the liquid and through a purification system made up of four absorbers arranged in series, two of them to retain liquid drops, while absorber 7 contains a solution of sodium hydrosulfite to eliminate oxygen impurities and absorber 8 has concentrated sulfuric acid to remove the water content. Valve 9 is used to equalize the pressures between both ends of the purification train. The nitrogen stream is adjusted by the needle valve (11)and measured in 12 through a mass flow meter. The three-way stopcock (14) makes it possible to feed the reactor with nitrogen or flush the chlorine tubes once every experiment is finished. Liquid chlorine of 99.5% purity is contained in cylinder 2 and is fed as a gas at pressures of 27-40 kPa by means of the pressure regulator (R). The chlorine stream goes
through the trap (6) and then bubbles in concentrated sulfuric acid contained in 8. The following absorber retains part of the liquid drops carried by the gaseous stream, while 9 again fulfills the functions of equalizing pressures whenever necessary. The microporous glass fiber filter (10) is capable of retaining any small drops that could have remained in the gas. The flow rate of chlorine is adjusted by means of the needle valve (11)and measured in 12 with a mass flow meter. Those parts of the flow meter in contact with chlorine are built of monel metal. The chlorine flow rate during the flow stabilization stages and during the reaction period is plotted by means of a potentiometric recorder (13). The chlorine and nitrogen streams, once purified and of known flow rate, are mixed in 15 before entering the reactor. The three-way valves (16 and 23) fulfill a double function. They are used for the flow stabilization stage, deriving the gaseous stream through auxiliary pipelines and finally makiig it bubble in concentrated sodium hydroxide contained in the chlorine absorber (25). In the reaction stage, both valves facilitate the sampling from the inlet or outlet gaseous streams of the reactor and the subsequent evaluation of their chlorine molar fraction. The chlorination reactor is built of a Pyrex glass tube, 154" o.d., 1 4 4 " i.d., and 228 mm high, which renders a working reactor volume of 2000 cm3. Both ends of the tube have glass fins which are used to attach the bottom and top plates of the reactor to the cylindrical tube, by means of a system of metal flanges. The bottom of the reactor is a circular plate made of quartz or Pyrex glass and has a system of hermetic sealing made up of aluminium flanges, rubber joints covered with Teflon, and a Teflon ring especially made to achieve a perfect adjustment with the reactor walls. The inlet window for the radiant energy has a diameter of 130 mm. The reactor cover is made up of a stainless steel disk protected by a thin Teflon disk. This cover has a central hole where the stirrer is located together with eight openings circularly arranged for the gas inlet and outlet, sampling, temperature measurement, and cooling of the reacting system. The stirring system has a sealing device made of Teflon and graphite especially constructed to avoid leaks of corrosive gases. The gas feed is performed through a hollow stirring rod built with a Pyrex tube and with a Teflon turbine agitator located at the bottom end of the shaft. There is a minimum stirrer speed (approximately 11 rps) under which there is no suction of the turbine agitator and, therefore, there is no gas feed. Above this minimum speed, an increase of the revolutions per second is translated into an increase of the holdup inside the reactor. The sampling is performed through a Teflon valve (19) connected to a Pyrex glass tube which is introduced into the liquid up to a few millimeters from the bottom plate. Externally, this valve has a Teflon adapter especially built for the connection with the glass syringe employed for the sampling. The temperature measurement in the liquid is performed by means of a digital thermometer with a PTR probe (21). It also has an analog output which allows the continuous recording of the temperature in 13 during the experiment. The cooling of the reactor is performed through the use of a couple of cooling coils made of Pyrex glass (18). Through them, water circulates at constant temperature, coming from a thermostatic bath. The gas stream coming from the photoreactor goes through a Pyrex glass condenser (22) to condense the hydrocarbons carried by the stream. Water at room temperature is used as a cooling fluid. The total pressure in the reador is measured through a U-tube manometer (24),
Ind. Eng. Chem. Res., Vol. 27, No. 7, 1988 1097 filled with concentrated sulfuric acid as manometric fluid. Finally, the effluent gases go through a three-way valve (23), bubble through a concentrated solution of sodium hydroxide (25), and are vented to the atmosphere. The emitting system of radiant energy consists of a UV tubular source with the axis located on the focal axis of a parabolic reflector. The radiation source employed is a General Electric Uviarc UA-3, medium-pressure, mercury-arc lamp. Its characteristics are 360-W input power, 152-mm arc length, and 19-mm diameter. The reflector is built with an aluminium sheet, specularly finished with Alzac treatment, which is 158 mm long and with a focal distance of 21 mm. The system is completed with a stream of controlled room-temperature circulating air to exhaust the ozone generated and cool the source; the continuous measurement of the operating voltage and the measurement of the radiant energy emission through the use of an Eppley thermopile Bi-Ag type are performed to control the stabilization process of the source. Further details on the characteristics of the experimental system may be found in Alfano (1984).
Experimental Procedure The experimental procedure begins by turning the lamp on together with the fan. The stabilization of the source demands about 4 h and can be monitored through the readings of the operating voltage and the thermopile output. During the stabilization stage of the emission, and about 2 h before the initiation of the reaction, nitrogen is turned on and the whole system is flushed with the object of eliminating the oxygen and the humidity that may be present. At that moment, it is d S 0 possible to check for leaks by turning off the nitrogen flow and observing possible variations of pressure in the manometer (24). The pressure was increased up to about 1.5-2 times the working pressure during the reaction. An hour before the initiation of the reaction, the nitrogen flow is cut off and the reactor is charged with about 2000 cm3 of a solution of trichloroethylene (or pentachloroethane) dissolved in carbon tetrachloride. The nitrogen flow is again connected to strip off any possible impurities in the liquid (oxygen and moisture, mainly) by nitrogen bubbling in the solution. Due to the gas feed system employed, at this moment it is necessary to initiate the agitation and fix the desired stirrer speed.1 I t is accurately controlled during all the experience and whenever necessary it is adjusted to the desired value. Then the cooling system is started and the temperature of the water circulating in the coils (18)is fixed between 15 and 18 OC, according to the desired reaction temperature. Afterward, the valve (4) is closed and the chlorine cylinder is opened, regulating the outlet pressure. Valves 16 and 23 have to be positioned in such a way as to permit the chlorine and nitrogen stream, after being mixed in 15, to circulate through the auxiliary pipeline and bubble in the scrubber (25), being vented to the atmosphere. Under these conditions, the fine adjustment of the flow rate of both gases is performed until an adequate stabilization of the flow rates of the gaseous streams is achieved. When the reactor is fed with pure chlorine, the nitrogen stream is cut off with 11 and a similar procedure is followed. Once the stabilization of the flow rates in the gaseous stream is achieved, valves 16 and 23 are positioned in such a way so as to allow the gas feed to enter the reactor. Simultaneously, the cover of the emitting system is removed to let the radiant energy enter the reactor, thus marking the initial time of the reaction. The temperature inside the reactor and the inlet flow rate of the gaseous stream are continuously recorded. At the beginning, a
temperature rise is produced and then a stabilization is reached, both of which coincide with the addition step of chlorine to trichloroethylene. This temperature rise is never higher than 5 OC. When almost all the trichloroethylene has been consumed, the substitution step takes place together with a temperature drop until reaching another stabilization value. This temperature drop is seldom higher than 2 "C. The presence of carbon tetrachloride used as solvent decreases the observed rate of reaction, thus contributing to control the thermal effects it may cause. The sampling of the liquid is effected at predetermined intervals with a syringe of 10 cm3. It is performed through valve 19 taking advantage of the fact that the reactor operates at positive pressure. The sample is immediately introduced in a flask containing a potassium iodide solution, thus generating a mixture with two phases: an aqueous phase and an organic phase. Chlorine is determined by titrating the liberated iodine with sodium thiosulfate. Afterward, both phases are separated and the organic phase (free from chlorine) is used to analyze the concentrations of trichloroethylene, pentachloroethane, and hexachloroethane by gas chromatography. For these determinations, an internal standard method was used. In those experiments in which the reactor is fed with a given relationship of chlorine and nitrogen, the sampling of the inlet and outlet gaseous streams of the reactor is also performed to know the molar fraction of chlorine in the gas. The duration of each experimental run is never longer than 2 h, while the total duration of each experiment demanded about 6 h including the previous steps of source stabilization, flushing with nitrogen, and obtention of the operating conditions. At the end of each run, the lamp is turned off and the chlorine feed is closed. The flushing with nitrogen is then started, in which case the three-way valve (14) and valve (4) are used. All the auxiliary systems of the equipment are disconnected, and afterward the positive pressure in the reactor is used to discharge the remaining liquid solution through 19. Some of the experimental runs that had been performed were repeated for the same operating conditions, finding in general a good reproducibility of the results. For two runs performed under the same conditions, differences in conversions were always smaller than 4%. This reproducibility was verified for the whole range of experimental concentrations used along the work. Table I summarizes the operating conditions used along the experimental work.
Results and Discussion In part 1of this work, it was concluded that the reaction takes place through a slow reaction regime; hence, in this section the dominant subregime for the addition and substitution reactions will be studied. Afterward, the theoretical and experimental results will be compared, discussing the validity of the results obtained with the ESVE model for the limiting mixing cases under consideration. Experimental Study for the Reaction Subregimes. In the first part of the study, the predominant subregime for the addition reaction of chlorine to trichloroethylene dissolved in carbon tetrachloride is studied under different conditions of stirring and of gas feed composition. The temporal evolution of the experimental concentration of chlorine and of pentachloroethane will be discussed, as well. Figure 2a shows the behavior of the photoreactor for increasing stirring velocities: N = 13.3 and 16.6 rps. It
1098 Ind. Eng. Chem. Res., Vol. 27, No. 7, 1988 Table I. Experimental Operating Conditions liquid reactant trichloroethylene
pentachloroethane
run T-1 T-2 T-3 T-4 T-5 T-6
C, (kmol/m3) 0.1
yz,in 0.1
0.1
0.1 0.15 0.40 1.0
P- 1 P-2 P-3 P-4
0.1
1.0
0.1
1.0 1.0
, ,32
0.1 0.1
0.2 0.2
1.0
0.1 0.1
104Fi, (mol/s) 5.8 5.8 5.8 3.9 4.5 4.5
N (rps) 13.3 16.6 13.3 13.3 13.3 16.6
bottom plate quartz quartz quartz quartz quartz quartz
10-V' (Pa) 1.06 1.06 1.05 1.06 1.06 1.06
T (K)
4.5 4.5 4.5 4.5
13.3 16.6 20.0 13.3
quartz quartz quartz Pyrex
1.06 1.06 1.07 1.07
293 292 293 293
1.0
10,
Table 11. Values of Model Parameters Dimensions of the Photoreactor rL = 9.5 mm LL = 152 mm a = 21 mm 1=84mm Lw = 158 mm d, = 130 mm dR = 144 mm LR = 228 mm kloA, = 1.5 X tg = 0.01-0.03
io)
294 295 294 295 293 294
fbi
Figure 2. Experimental results of trichloroethylene chlorination. Effect of mechanical agitation: (a) YZ,~,, = 0.1; (b) y2,in =: 1.0.
can be observed that the conversion of the reaction increases with the stirring, since for a given value of the dimensionless time the dimensionless concentration of pentachloroethane (G5) becomes greater. With respect to it can be the dimensionless concentration of chlorine (GZ), observed that the concentration level of this species with respect to pentachloroethane is notably small (the corresponding scale has been enlarged 5 times). The concentration of chlorine for which a quasi steady state is reached in the reactor is approximately 5 X lo4 M,a value which is quite far from the saturation composition for the working conditions that have been used. This low concentration of chlorine produces a low radiant energy absorption by the reacting medium, with which the effect provoked by the increase of the stirring velocity is totally attributed to an increase of the gas to liquid absorption rate (see the second condition for the existence of mixing effecb in part 1). On the other hand, it should also be remarked that the concentration of pentachloroethane as a function of time is practically a straight line for both stirring velocities, which suggests the existence of an apparently constant reaction rate. Besides, for the experiment performed with N = 16.6 rps, a noticeable increase of the chlorine concentration is observed immediately after the total conversion is reached. This makes evident a possible change in the reaction subregime at the end of the addition of chlorine to trichloroethylene and the beginning of the hydrogen substitution. This will be analyzed in detail below. Figure 2b studies the same reaction under similar operating conditions but with a gaseous feed constituted by pure chlorine to eliminate the possibility of a mass-transfer control in the gas film altogether. Again, it is observed that an increase in the stirring velocity produces an increase of conversion even though not so significant as in the former case. Besides, the concentration of chlorine in the bulk of the liquid phase is still small, the quasi-steady-state M. As in Figure 2a, concentration being about 6 X the concentration-time curves are practically straight, again indicating the existence of a constant rate of reaction. In this case, the reaction is quite fast and the total conversion is reached in about 15 min; for this reason it is
Mass-Transfer Parameters l/s H = 5.3 X lo7 (Pa.cm3)/mol db = 1.5 mm
Kinetic Constants for the Chlorination Reactions k , = 6.31 X lo'* cm3/(mol.s) k i = 0 k3 = 8.07 X lo7 cm3/(mol.s) k i 0 k , = 2.13 X 1O1O cm3/(mol-s) k4/ = 5.18 X lo2 cm3/(mol.s) k , = 1.81 X lo7 cm3/(mol-s) kg) = 0 ke = 2.51 X l O I 5 cm3/(mol.s) k7 = 2.00 X lo'* cm3/(mol-s) k, = 5.01 X IO" cm3/(mobs) k9 = 3.16 x 1014cm3/(mol.s) k , , = 1.00 x 1014cm3/(mol.s) k , , = 1.00 x 10l2 cm3/(mol.s)
=
impossible to obtain a greater number of data for these experiments. Taking into account the phenomena described above (increasing conversion with the stirring velocity, very low concentration of chlorine, etc.), it is possible to conclude that, for the operating conditions included in Figure 2, the reaction is under a net "diffusional subregime" (Astarita, 1967; Astarita et al., 1983). According to Astarita et al. (1983), the diffusional subregime is that condition where the chemical reaction is slow enough not to result in any enhancement ( t
