Sei sulla pagina 1di 5
hel Erin Se, 9p. 36 3,19 Saxe Peamon row ple RADIAL DISPERSION MODEL FOR BUBBLE PHENOMENA IN THREE-PHASE FLUIDIZED BEDS S. L. P. LEE and H. I. DE LASA Faculty of Engineering Science, The University of Western Ontario, London, Ontario, Canada N6A 5B9 (Receiced 23 January 1987; ecceped for publication 8 February 1988) Abstract “The local gas hold-up and average gas bubble velocity in three-phase uiized beds (TPFBs) were measured by optical fiber probes. radial gas dispersion model was applied to analyse the gas hold-up distribution in the TPFB. The gas dispersion was found high near the grid and the dispersion deceased sharply toa much lower ievel and stabilized at higher axial positions. intropucTiON ‘Thee-phase fluidization received much attention in the past few decades because of various applications in many chemical processes. The characteristics of three-phase fluidization beds (TPFBs) have been re- viewed by Baker (1981), Darton (1985), de Lasa and Lee (1986), Epstein (1981), Muroyama and Fan (1985), (Ostergaard (1971) and Wild (1983). Bubble dynamics strongly influence the performance of a TPFB reactor. For example, the gas hold-up and the bubble velocity strongly affect the gas-liquid interfacial area and the ‘a5 residence time. Consequently, the mass transfer and conversion in TPFB reactors are highly influ- enced by the bubble dynamics. There are two different experimental approaches in the study of gas hold-up in a TPFB, The first method (eg. Kim et al. (1972)] gives overall gas hold-up through the use of pressure profile along the uidized ‘bed, the total amount of solid particles and the conti- ‘nuity of the three phases. In this case, the phase hold- ‘ups are assumed to be uniform throughout the fluidized bed. The sevond method of gas hold-up measurement [eg Rigby and Capes (1970) and Begovich and Watson (1978)] uses in situ measuring probes. The advantage of in situ measuring probes is that the detailed information of local phase hold-ups can be obtained. The probe measurements show that there isa distribution, instead of constant-phase hold ups in TPFB. ‘The gas bubble rising velocity in previous studies ‘was obtained from either the in situ measuring probes ‘or the analysis of bubble images on a flat two-dimen- sional bed. Various types of measuring probes were used by Righy et al. (1970), de Lasa et al. (1984) and Lee et al, (1984). Examples of two-dimensional bed studies were investigations by Henriksen and Ostergaard (1974) and Kim etal (1977). ‘The bubbles in a freely bubbling fluidized bed were found to rise faster than the single isolated bubbles in a liquid fuidized bed (de Lasa et al, 1984; Kim et al, 1977; Lee, 1986; Lee et al, 1984; Rigby et al, 1970). Various relationships between bubble velocity and bubble size have been observed in freely bubbling fluidized beds. The dependency of bubble velocity on ‘bubble size was found to be higher than the Davies and Taylor relationship (Kim etal, 1977: Righy and Capes, 1970), Observations by Lee (1986) and Lee and de Lasa (1986), however, showed that the overall velocity-size relationship was similar to the Davies and Taylor equation. In addition, it was found that dependency was lower fr small bubbles than for lage bubbles. The bubble size distributions in TPFBs were found to be log-normal and the mean bubble size ‘depended strongly on the flow regime (Matsuura and Fan, 1984; Lee, 1986; Lee et al. 1984) Previous studies on the bubble phenomena in ‘TPFBs were focused individually on either the gas hold-up or the bubble rising velocity. In the present work, a radial gas dispersion model which considered both parameters was proposed to describe the gas hold-up distribution throughout the bed. The infor- ‘mation required for the analysis were the local gas hhold-up and the local bubble velocity. These par- meters were obtained through several optical ‘measuring probes, EXPERIMENTAL SET-UP ‘The experiments were performed in a cylindrical uidized bed of 02m internal diameter. Air. water and glass beads of 250 um were used as the gas, li and solid phases. The local gas hold-up and bubble velocity were measured by an optical fiber probe ‘measuring system. The design of the probe was de- scribed in de Lasa et al. (1984) and Lee (1986). The details of the apparatus and the experimental pro- cedures were presented by Lee and de Lasa (1987). DATA ANALYSIS (Once the stationarity was established (Lee and de Lasa, 1987), the signals recorded were treated stat- isticaly to obtain the information desired. In order to ‘measure the bubble velocity. two optical fiber probes -were aligned vertically with a separation of 0.94 em. The normalized cross-correlation function (NCCF) was used to measure the time delay between the mass 26 S.LP. Letand H. I. og Lasa, signals from the two probes: iapl®) Ray w and ifr Y(t) +f End bxe+9-n,)4 2 were x(t) and y(t) are the signals from the lower and the upper probe, respectively AA cross correlogram was constructed by plotting the NCCF against the time displacement t. An example ofa cross correlogram is shown in Fig. 1. The average time required for gas bubbles to pass through the two probes was obtained from the time delay fous which was established by noting the time displace- ‘ment at the peak of the cross correlogram. Since the distance between the two probes was known and fixed, the average bubble velocity was calculated by ing the distance by the time delay, tyax. The ‘ross-spectral density function (CSDF) was also used to obtain the time delay as a function of frequency in ‘ase the cross correlogram did not show a distinct peak and for the purpose of confirming the cross- correlation analysis. ‘The CSDF of the two signals was obtained from the Fourier transform of the NCCF as Gal) @ and. Gal N= CN IQ) 4 where cotsvo2 f Rylt)cosQnfede (5) 2nisr=2 agi Qxfad. (6) The phase angle of the cross spectrum gave the phase shift between the two signals at frequency f: Of Sd tan™ [ej SUC AS a os; Ray os os 02 ° 0.01 0.02 003 004 Fig. 1. Sample of cross cortelogram for :=0.075 m, r,=0, Vou tl ems, P,=039 em’. The time delay was obtained for '=0,)2«fand, consequently, the average bubble velocity at fre- quency f was calculated. Note that the vse of the CSDF gives time delay information as a function of frequency, which is not available directly from the NCCF. Obviously, the case where time delay is in dependent of frequency, a constant time delay should be observed throughout the frequency range. Hence the cross-specral density funtion can aso be used to complement the cross-coreation analysis ‘THEORETICAL DEVELOPMENT ‘The radial gas hold-up profile in a TPFB was found to be mostly parabolic (Hu er al, 1985, 1986; Lee and de Lasa, 1986; Morooka et al, 1982). Therefore, itis assumed that the gas hold-up in a TPFB can be represented in dimensionless form: BLee)] © with boundary conditions 1 Faby) Vem Volt) ° 20. The boundary condition at r, =O.considers that the radial gas hold-up profiles are symmetric. This con- dition generally applies if the gas distributors (in- jectors) are arranged symmetrically. Equation (8) can be solved by separation of variables to give: ant susie where the coefficients 4, are f Patel aalaarsddry ay rebitarexo( 2 Ver. where J, is the Bessel function of the first kind of order zero and 3, are the positive zeros of J,(2 ‘The gas hold-up. and the average bubble velocity, were measured at various axial and radial pos- itions in the TPFB. The gas hold-up was defined as. the fraction of total measuring time when the detec- ting probe was in contact with gus bubbles, The average bubble velocity was assessed through the application of the cross-correlation technique to the signals from the measuring probes. Once these exper- imental data were obtained, the dispersion coeficent, D, was assessed through the Marquardt method of optimization (Kuestler and Mize, 1973). It should be noted that both the gas hold-up. ¢, and the average bubble velocity, V. were obtained through analyses of signals from the same time frame. Different operating conditions, listed in Table 1, were used in the experiments. The data analysis was performed ee Racial dispersion model for bubble phenomena eT Table |. Standard deviations of model esti- mations of gas hold-up Ve % Standard (emis) (ems) _ deviation 128 079 0.0095 8 ur aos 18. 156 0.0086 20 079 013s 20 ut 00136 20 136 00132 20 195 0108 ‘Table 2 Average bubble velocity in ems ‘measurements are listed in Table 2. Both the cross correlation and the phase angle were used for analyses and the results obtained through the latter method are listed within brackets. These two methods show ‘good consistency. In addition, it was observed from the phase angle analysis that the average bubble velocity was independent of bubble frequency. The average bubble velocities were mostly in the range of 50 to 80cmys. Occasionally, velocities as low as 30 cm/s were also detected. At the grid level, 7=0075,, the average bubble velocities were found to peak near the wall of the column, r,=0.5 and 0.75 (Table 2(a) and (b)]. However, the radial positions of the peak bubble velocity moved toward the center of the ‘column as the gas bubbles rose inthe fluidized bed. In Radial po addition, the magnitude of the peak velocity de- : ‘creased with height. Furthermore, the jet effect atthe i) 035 075 (@ You=20 emis, ¥z=079 cays ey © 41079 « oss 588 32 470 “ ae (8946) on ao gem geet 7 eee 7 0 156 omy 8] 2) S573) Gs eee sR eeeari cease nas ey AD | bag cee SAS sea ses eases 69 89 G29 ” meets senses eg (65 (653) 89) | © Yex20em/, Yea .t7 ems oss msi 4 (9 Ga 8O ons ok az 88 3 Gas) oss 88880 ons (ee Sas) aS) ig. 2, Radial dspesioncoeficients for Yo=1.28 om. (1) 4TH ea ete ry) 4 z (9 Yon20ems, Yy=1.36 cm/s ate geen Se 2 MELA ems Cr a © YL=156 anys 3) SN 8) ors Se SGD + MEr185 emt @H 06) 6S HD) oss Lata (663) (646) 646) (48-2) 0375 83078} SB «80 5 m5 gy 0225 or 6h ‘388 (5m 6S) BB) 553) ors orl 83 STL (656) 072) 661) (65) individually for each operating condition to obtain the dispersion coefficient at various axial heights. [RESULTS AND DISCUSSION ‘The bubble rising velocity in TPFBs were measured at 0.075, 0.225, 0375, 0.525, 0.675 and 0:825 m above the distributor. At each axial height, the measure- ‘ments were performed at four radial positions, ,=0, (025, 0.5 and 0.75. Typical average bubble velocity Fig. 3. Radial dispersion coefficients for Y¢=20 ems 28, S.LP. Leeand H. 1. 0€ Lasa arid level was diminished by a higher liquid velocity; flatter radial bubble velocity profiles were observed. In general,-once above the grid region, the average bubble velocities near the wall are a little lower than those in the rest of the bed. This is presumably due to the wal effect. uch a phenomenon is more noticeable when a high liquid velocity is used. ‘The relationship between the gas dispersion coef- ficient and the dimensionless bed height are shown in Figs 2 and 3. The dispersion coefficient was found to be quite large near the grid. The coefficient decreased rapidly toa much lower level ata slightly higher Position. From there on, the dispersion coefficient reduced further but only slightly as the height in- ‘creased. The sharp change ofthe dispersion coefficient took place within a dimensionless height of 0.15, approximately. ‘An example of the gas hold-up distribution in the fluidized bed is shown in Fig. 4 where the gas hold-up predicted by the dispersion model is also illustrated. . © Exp i © Medet cy or} o Fig. 4. Gas holdsups for Vg=20.m/s, ¥,=195 emis. tp. enn is Fig. 5, Comparison between the experimental gas hold-ups and the predictions by the di ‘model, Vo=20em/5, "yclssems Since gas bubble injectors were used in the exper- iments, a high degree of radial dispersion can be expected in the grid region. The result in Fig. 4(a) indicates thatthe radial dispersion coefficient inthis region is also a function of the gas hold-up which varies over a wide range radially, Therefore, the assumption of radial independence ofthe dispersion coefficient is not perfect for the low part of the bed. Nevertheless, the correct tendency of gas hold-up Profile was predicted by the present analysis. If the radial dependency were considered, eg. (8) would have to be solved numerically instead of analytically. However, numerical solution is not justified at this stage of the model development. ‘The fit ofthe model to the experimental gas hold- ups corresponding to the results in Fig. 4 is shown in Fig. 5. As expected, the error of predictions at the grid level appears to be higher than the rest of the est- mation. The standard deviations of gas hold-ups est- mated by the dispersion model ar listed in Table ! for experiments under various operating conditions. Tn conclusion, the present dispersion model de- scribes the gas hold-up distribution throughout the TPFB, There is a shallow grid region where intensive 1s dispersion occurs. The radial gas hold-up distr- butions stabilize very quickly at higher axial pos- itions where the gas dispersion coeficients are relatively smal. NOTATION coefficient in eq, (10) coincident spectral density function dispersion coefficient, cm/s é HD ‘modified dispersion coefficient, bubble frequency, 1/8 cross-spectral density function fluidized bed height, cm Bessel function of the first kind of order Seo Bop Radial dispersion model for bubble phenomena number of data points Quadrature spectral density function radial position, cm radius of the fluidized bed, em time, s total measuring time, s dimensionless radial position 1/R superficial gas velocity, cm/s superficial liquid velocity, em/s average bubble velocity, em/s average bubble velocity at z, time history of probe signals z aial positon, cm initial axial positon, om z dimensionless axial position, z/H, ce iitaldimensioniss axial a Positive zeros of Ja) ty covariance function Ho By mean value of x() and y() o, phase angle & local gas hold-up, cm? gas/em? bed be local gas hold-up at z=z,, em? gas/em? bed we ye e [ WE, Commit eerie ] 4x6, standard deviation of x) and y(e) r time displacement isplacement at the peak of R,y(*) [REFERENCES Baker,C.G. 5, 1981, In Three-phase Flidzation, Multiphase Chemical Reactors (Edited by S.A. Rodrigues. J. Calo and N. Sweed), pp. 343-387, Stioff and Noordhoft The therlands. Begovich, J. M. and Watson, J. S, 1978, An electroconduc- tivity technique for the measurement of axial variation of holdups in three-phase Muidized beds, A,L.CKE. J. 24, 351-354, Datton, R. C. 1985, In The Physical Behaviour of Three- phase Fluidized Beds, Fluldzation (Edited. by J. F. Davidson, R. Clift. and D. Harrison), 2nd edn, p. 495-526, Academic Pres, New York e Lasa. H Lee, S. LP. and Bergougnou, M. A. 1984, ‘Bubble measurement in three-phase fluidized beds ising a Usshaped optical fiber. Cam J. chem. Engng 62, 165-169 de Lazo, H. and Lee, LP, 1986, In Thre-phase Fluiized Bed Reaciors (Edited by H. de Last, pp. 349-39 Martinus Nijhof, The Hague, aug Epstein. N 1981, Three phase uiization: some knowledge ‘ps. Cat. J chem. Eng $9. 689-659, Henriksen, H. K. and Ostergaard, K.. 1973 Characteristics of large two-dimensional air bubbles in liquids and in three-phase fuisized beds. Chem. Enyny J, 7. 14-196, Hu, T.T. Yu, BT. and Wang. ¥.P, 1986. in Holdaps and Models of Three Phase Fluidized Bed, Flidization. V (Edited by K. Ostergaard, and A. Sorensen. pp. 353-460, Engineering Foundation. 3. Hu, T. T, Zang, Y. Chao, H. Yu, B.T, and Wang. YP. 1985, A Study of Hol-up and Ligud Mixing in Three ‘phase Fluidized Bed. Flidcation "8S Science and Teche nology (Edited by M, Kwauk, D. Kuni, J. S- Zheng ‘and M. Hasatani, pp. 262-273. Kim, S. D, Baker, C. G. J. and Bergougnou, M. A. 19; Hold up and axial mining characteristics of two and three hase uidied beds. Can. J. chem. Engng 32, 695-701. Kim, 8. D. Baker, C.G. J. and Bergougnou, M. A. 1975, Phase hold-up characteristics of three phase Auidized beds. Can J. chem, Engng 53, 134-139, Kim. . D, Baker, . G.J. and Bergougnou, M. A. 1977, ‘Bubble characterstis in thee phase fluidized beds. Chem Engng Sel 32, 1299-1306, Kuestler J, Land Mize. J.H, 1973, Optinication Techniques with FORTRAN. McGraw-Hill New York. Lee. S. LP. 1986, Bubble dynamics in thre phase Nuiized beds. Ph.D. dissertation, University of Western Ontario. London, Canada. Lee, S. LP. and de Lasa, H.1, 1986, In Characteriation ‘of Bubbles in Three-phase Fuidized Beds Using U-shaped Optical Fiber Probes, Flidisation V (Edited. by K. (Ostergaard and A. Sorensen), pp. 361-368. Engineering Foundation Le, SL. P.and de Lasa,H.L, 1987, Phatehold-upsin three hase fuidized beds. A1.CKE. J. 33, 1359-1370, Les, S. L. P. de Lasa, H. and Bergougnou. M. A. 1984, Bubble phenomena in’ three-phase Muidized beds us viewed by a U-shaped optic probe. AJ.ChE. Symp. Ser 80, 110-116. Matsuura, A. and Fan, LS, 1984, Distribution of bubble ‘ropertis in a gas-lquid-soid fuidiaed bed. 4J.ChE. J. 30, 894-503 Morooka, S, Uchida, K. and Kato, Y., 1982, Recirculating turbulent flow of liquid in gas-lquid-solid Auidized bed J. chem. Engng Japan 18, 29-34, Muroyama, K. and Fan, LS, 1985, Fundamentals of ‘ga-liquid-soid Nuidization, A1.ChE. J. 31 1-34 Ostergaard, K. 1971. In Three-phase Fluidization. Fluid- ‘ation (Edited by J. F. Davidson and D. Harrison, pp. 751-180, Academic Press, New York Rigby, G. R. and Capes, C. E, 1970, Bed expansion and ‘bubble wakes in three-phase Muidiztion. Can. J. chem, Engng 48, 343-348, Wile. G. "1985, Les reacteurs alts fuidise gas liquide-solide. Etat de Tart et perspectives industrielles. Enaropie, 3-37.

Potrebbero piacerti anche