Você está na página 1de 11

Chemical Engineering Journal 175 (2011) 494504

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal


journal homepage: www.elsevier.com/locate/cej

Hydrodynamics and mass transfer characteristics in an internal loop airlift reactor with different spargers
Lijia Luo, Fengna Liu, Yuanyuan Xu, Jingqi Yuan
Department of Automation, Shanghai Jiao Tong University, and Key Laboratory of System Control and Information Processing, Ministry of Education of China, Shanghai 200240, China

a r t i c l e

i n f o

a b s t r a c t
The effect of the sparger structure on the hydrodynamics and mass transfer characteristics of an internal loop airlift reactor was investigated. Three spargers with different diameters and numbers of orices were tested. Two different ow regimes, i.e., homogenous ow and heterogeneous ow, were identied within the experimental range of supercial gas velocities. It is found that the sparger structure has more signicant effect on hydrodynamic parameters in the heterogeneous ow regime than in the homogenous ow regime. The largest gas holdup and downcomer liquid velocity were obtained with the 4-orice nozzle, followed by the O-ring distributor and the 2-orice nozzle. The sparger structure has less effect on the oxygen transfer coefcient kL , while it strongly affects the specic interfacial area a. The 4-orice nozzle improved the volumetric mass transfer efciency for that it generated a smaller mean bubble diameter and therefore a larger specic interfacial area than other spargers. Based on the material conservation and pressure balance principles, a hydrodynamic model was established to predict the liquid velocity in the downcomer, as well as the liquid velocity and the cross sectional area of the up-ow in the riser. Empirical correlations were proposed for different spargers, which well predicted the gas holdup and the volumetric mass transfer coefcient. 2011 Elsevier B.V. All rights reserved.

Article history: Received 13 May 2011 Received in revised form 13 September 2011 Accepted 15 September 2011 Keywords: Airlift reactor Sparger structure Gas holdup Liquid velocity Mass transfer coefcient

1. Introduction Airlift reactors have been widely applied in chemical, biotechnological as well as water treatment industries due to their simple construction, low power consumption, good mixing and relatively high mass transfer efciency [13]. Generally, the airlift reactors are classied into two categories: external loop airlift reactor (ELAR) and internal loop airlift reactor (ILAR). The former is composed of two conduits connected at the top and the bottom, while the latter consists of two concentric cylinders [4]. A typical internal loop airlift reactor includes four important sections named as riser, downcomer as well as top and bottom sections, respectively. The density difference induced by the gas holdup difference between riser and downcomer is the main driving force to form the liquid circulation in airlift reactors [5]. Gas holdup, liquid circulation velocity and mass transfer coefcient are the most important hydrodynamic parameters for airlift

Corresponding author at: Department of Automation, Shanghai Jiao Tong University, Shanghai 200240, China. Tel.: +86 21 34204055; fax: +86 21 34204055. E-mail address: jqyuan@sjtu.edu.cn (J. Yuan). 1385-8947/$ see front matter 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.cej.2011.09.078

reactors. The knowledge of their dependences on supercial gas velocity, reactor geometry and sparger structure is essential for the design of airlift reactors. Extensive studies on the hydrodynamics in airlift reactors have been reported in the last decades. Kilonzo et al. [6] investigated the hydrodynamics and mass transfer characteristics in an inverse internal loop airlift-driven brous-bed z bioreactor. Bla ej et al. [7] studied the scale inuence on the hydrodynamics of an internal loop airlift reactor. Lu et al. [8] compared the hydrodynamics and mass transfer characteristics between a modied square airlift reactor and traditional airlift reactors. Some researchers investigated the local hydrodynamic characteristics of airlift reactors [2,9,10]. There are also some studies regarding to the inuence of the gas sparger structure in airlift reactors [1114]. Meanwhile, many empirical correlations have been proposed to predict the hydrodynamic parameters in airlift reactors [1,3,15]. However, as pointed out by van Baten et al. [4], the applicability of these correlations is generally restricted by the reactor geometry. This work focuses on the inuence of the sparger structure on gas holdup, liquid velocity and volumetric mass transfer coefcient in an annulus sparged internal loop airlift reactor. Three spargers with different structure parameters, such as the diameter and numbers of orices, are tested. A hydrodynamic model based on the

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504

495

Nomenclature a A CL C* CL0 dB do DD DL DR f HD HL Hd He g KL Kg kf kL kL a R2 Re t Ug specic interfacial area (m1 ) cross sectional area (m2 ) instantaneous dissolved oxygen concentration (mg L1 ) saturation dissolved oxygen concentration (mg L1 ) initial dissolved oxygen concentration (mg L1 ) mean bubble diameter (m) diameter of sparger hole (mm) hydraulic diameter of the down-ow region in the reactor (m) diffusivity of oxygen in liquid (m2 s1 ) hydraulic diameter of the up-ow region in the reactor (m) frictional factor aerated liquid height (m) unaerated liquid height (m) height of the downcomer (m) Henry constant of oxygen gravitational acceleration (m s2 ) liquid-side mass transfer coefcient (m s1 ) gas-side mass transfer coefcient (m s1 ) frictional loss coefcient overall mass transfer coefcient (m s1 ) volumetric mass transfer coefcient (s1 ) correlation coefcient Reynolds number time (s) supercial gas velocity (m s1 ) linear velocity (m s1 )

the volumetric mass transfer coefcient in the reactor for different spargers. The effectiveness of these correlations is validated by experimental data. 2. Experimental apparatus and methods 2.1. Apparatus The schematic diagram of the experimental apparatus used in this study is shown in Fig. 1. The airlift reactor was constructed of a Plexiglas column with a height of 1300 mm and an inside diameter of 284 mm. A draft tube with 820 mm in height and 70 mm in inside diameter was mounted in the center of the column, where the clearance between the draft tube and the bottom of the reactor was xed at 40 mm. There are two groups of holes at the trisection positions of the height of the draft tube (see Fig. 1). Each group consists of three 33 mm-diameter holes equally distributed along the circumference of the draft tube. These holes served as entrances for the draft tube when the liquid surface was below the top of the draft tube. The air was injected into the reactor through a sparger located at the bottom of the annulus section between the draft tube and the external column. The annulus channel and the draft tube served as the riser and the downcomer, respectively. As shown in Fig. 2, three types of spargers were tested: 2-orice nozzle (each orice has a diameter of 2.6 mm and is down-curved at a 45 angle), 4orice nozzle (4 orices are arranged in a symmetry conguration and each orice has a diameter of 1.84 mm. Moreover, each orice is also rotated in a clockwise direction at a 45 angle, except for being down-curved at a 45 angle as the orice of the 2-orice nozzle) and O-ring distributor (made of a 6 mm-diameter copper ring pipe, with 63 pores of 1 mm-diameter spaced equidistantly along the circumference). All the orices or pores were mounted facing down, where the clearance between the orice of the 2-orice nozzle or the 4-orice nozzle and the bottom of the reactor was 10 mm, while the O-ring distributor was located 70 mm above the bottom of the reactor. The airow rate was varied from 4.44 105 to 1.78 104 m3 s1 , giving a range of supercial gas velocity from 0.7 103 to 2.81 103 m s1 (based on the cross sectional area of the reactor). Tap water was used as the continuous phase. The unaerated liquid level was 1032 mm, holding the top clearance of 172 mm between the liquid surface and the upper end of the draft tube. In addition, the reactor was equipped with nine dissolved oxygen (DO) electrodes. As shown in Fig. 1, three electrodes (pO2 13) were inserted into the downcomer at different horizontal positions along the same axis, while the other six electrodes (pO2 49) were divided into two sets and mounted in the riser with the same way as that in the downcomer. All experiments were carried out at room temperature and atmospheric pressure. 2.2. Measurements 2.2.1. Gas holdup The overall gas holdup g,o was calculated from the unaerated liquid height HL and the aerated liquid height HD by: g,o = 1 HL HD (1)

Greek letters distance between two measuring points (m) h P pressure difference between two measuring ports (Pa) t time difference (s) gas holdup density (kg m3 ) density difference (kg m3 ) time constant of the DO electrode (s) p time constant of the reactor (s) R viscosity (Pa s) Subscripts b bottom downcomer d D down-ow dn down friction f g gas homogeneous-phase h L liquid overall o r riser R up-ow t top total T up up

The riser gas holdup g,r and the downcomer gas holdup g,d were estimated from the pressure difference P between two pressure taps with a vertical space of h: g,r = 1 g,d = 1
Lg

material and pressure balance principles is established to predict the liquid velocity in the downcomer, as well as the liquid velocity and the cross sectional area of the up-ow in the riser. Some empirical correlations are proposed to predict the overall gas holdup and

Pr hr Pd hd

(2) (3)

Lg

496

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504

Fig. 1. Schematic diagram of the annulus sparged internal loop airlift reactor.

The pressure differences were measured with the U-type manometer connected to pressure taps. 2.2.2. Liquid velocity The liquid velocity in the downcomer was measured using the trace technique [4]. The saturated KCl solution was used as the tracer. After the ow reached a steady state, 15 ml of saturated KCl solution was injected into the downcomer through an injection port (see Fig. 1). Two conductivity probes with a distance of 290 mm were used to detect changes in conductivity. The response signals of two conductivity probes were sampled at a frequency of 100 Hz by a data acquisition system and then low-pass ltered at 5 Hz. The liquid velocity was determined from the time difference, t, in responses of two probes:

vLd =

0.29 t

(4)

Due to the low ratio between cross sectional areas of downcomer and rise as well as the low airow rate performed in all experiments, the up-ow and the down-ow were simultaneously observed in the riser, which is known as the internal liquid circulation [16]. This phenomenon makes the measurement of the liquid velocity in the riser difcult, and thus the liquid velocities in the riser were calculated from the material balance. The riser was assumed to consist of two distinct regions, i.e. up-ow and down-ow regions, according to the ow direction of the liquid phase. Experimental observations indicated that hydrodynamic characteristics in the down-ow region of the riser were similar to those in the downcomer. Thus, the gas holdup, the liquid velocity and the frictional coefcient in the down-ow region of the riser were assumed to be approximately equal to those in the downcomer. At the top and bottom sections of the reactor, most of gas bubbles moved upward except that few bubbles were entrained downward by the liquid recirculation. Neglecting the recirculation of small bubbles in view of the

Fig. 2. Sparger types: (a) 2-orice nozzle; (b) 4-orice nozzle; (c) O-ring distributor.

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504

497

low gas holdup in the down-ow, it was assumed that gas holdups in the top and bottom sections were similar to that in the up-ow region of the riser. The real cross sectional areas of the down-ow (AD ) and the up-ow (AR ) in the reactor are obtained from the gas volume balance: AD = Ad + Ar,dn = AR = AT AD The liquid velocity in the up-ow region of the riser is: HD AT (g,r g,o ) Hd (g,r g,d ) (5) (6)

vLr,up =

vLd AD (1 g,d )
AR (1 g,r )

(7)

2.2.3. Mass transfer coefcient The mass transfer rate in an airlift reactor is expressed as [1]: dCL = kL a(C CL ) dt (8)
Fig. 3. Overall gas holdups in the reactor for different spargers.

where C* and CL are the saturation and instantaneous dissolved oxygen concentrations, respectively. The overall mass transfer coefcient kL is related to the gas-side and the liquid-side mass transfer coefcients Kg and KL through the Henry constant He by [1,18]: 1 1 1 = + KL HeKG kL (9)

For oxygen, because He is much larger than unity and Kg is considerably larger than KL , the second term on the right side of Eq. (9) is negligible and Eq. (9) reduces to kL KL [1]. Therefore, the main resistance to mass transfer lies basically on the liquid-side. According to Eq. (8), the knowledge of the volumetric mass transfer coefcient kL a is necessary for the evaluation of the oxygen transfer performance in the reactor. The kL a was determined by the gassing out dynamic method [17]. The reactor was rstly fed with the nitrogen gas to remove the dissolved oxygen from the system, and then the nitrogen ow was stopped and the fresh air was fed into the reactor. The time-variant dissolved oxygen concentration was measured by the nine DO electrodes mentioned above. The signals were recorded at an interval of 1 s by a data acquisition system until the dissolved oxygen concentration was saturated. Under the assumption of idea mixing in the liquid phase as well as a constant oxygen concentration in the gas phase, and taking the inuence of the DO electrode into account, the volumetric oxygen transfer coefcient kL a is calculated by [1]: (e(kL at) kL a C CL = C CL0 1 kL a
p p

Fig. 4. Gas holdups in the riser for different spargers.

et/ p )

(10)

where CL0 is the initial dissolved oxygen concentration and p is the time constant of the DO electrode. If the time constant of the reactor, R = 1/kL a, is much greater than p , according to the suggestion in [18], Eq. (10) is simplied by neglecting p : C CL = e(kL at) C CL0 (11)

Thus, the kL a can be calculated from the slope of the linear part of ln[(C* CL )/(C* CL0 )] against t. 3. Results and discussion 3.1. Gas holdup Figs. 35 show the evolutions of gas holdups as a function of the supercial gas velocity for the three spargers. It is clear that the gas holdup increases with the supercial gas velocity. Generally, there exist two typical ow regimes in an airlift reactor

Fig. 5. Gas holdups in the downcomer for different spargers.

498

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504

Fig. 6. Comparison of gas holdups in the downcomer and riser for different spargers.

depending on the airow rate [19,20]: (a) homogeneous ow, which may be characterized by a narrow distribution of bubble size, a radial uniform gas holdup distribution and the weak liquid phase turbulence in the riser, as well as no bubble entrainment or the stagnation of bubbles in the downcomer; (b) heterogeneous ow, which may correspond to a wide distribution of bubble size formed by bubble coalescence and breakup, a radial nonuniform gas holdup distribution and the intensive liquid phase turbulence in the riser, as well as the bubble recirculation in the downcomer. These two ow regimes can be identied from the slope changes of gas holdup and liquid velocity curves. As shown in Figs. 35, the slope of the gas holdup curve decreases a little after crossed Ug = 1.75 103 m s1 . In Fig. 7, the liquid velocity increases rapidly before Ug = 1.75 103 m s1 , while increases slightly at larger supercial gas velocities. Therefore, the ow regime transition point may be at around Ug = 1.75 103 m s1 . In fact, bubbles rise through the riser in a straight line at low supercial gas velocities, and thus the bubble coalescence is not expected. Meanwhile, only small bubbles are entrained into the downcomer and tend to remain stationary in the downcomer. A higher supercial gas velocity implies that more bubbles are formed in the riser and entrained into the downcomer. As a result, the gas holdups both in the riser and the downcomer increase almost linearly with the supercial gas velocity. After the supercial gas velocity reached a critical value, which may be at around 1.75 103 m s1 , vortices of bubble swarms and the bubble coalescence make ow structures more complicated, resulting in that the ow regime breaks away from the uniform bubbly ow. This critical supercial gas velocity may correspond to the ow regime transition point. As the supercial gas velocity increases further, bubbles become closely packed and coalesce to form lager bubbles, and the vortices of bubble swarms and the liquid eddies in the riser become more intensive, meanwhile, more large bubbles recirculate in the downcomer, which slow down the increases of gas holdups at high supercial gas velocities.Fig. 6 shows the comparison of gas holdups in the downcomer and in the riser. It indicates that the gas holdup in the downcomer is approximately proportional to that in the riser. Similar results also reported by Chisti [1], Lu et al. [21] and van Benthum et al. [22]. They found that the ratio between gas holdups in the downcomer and in the riser was a constant between 0.8 and 1 for the two-phase internal loop airlift system without gas disengagement device. Kilonzo et al. [6] reported that this ratio was 0.672 in an unpacked inverse internal loop airlift reactor. However, in this study, this ratio is only about 0.55 in the homogeneous ow regime, and 0.6 in the heterogeneous ow regime. This may be caused by

the specic conguration of our reactor, where the small diameter of the draft tube induces high ow resistances and thus a low liquid circulation velocity, which further makes less bubbles enter into the downcomer, resulting in a small value of g,d /g,r . As shown in Fig. 3, the largest overall gas holdup is obtained with the 4-orice nozzle, followed by the O-ring distributor, while the 2-orice nozzle produces the smallest overall gas holdup. It is worth to mention that the gas outlet velocity of each orice of the 4-orice nozzle is guaranteed equal to that of the 2-orice nozzle at the same airow rate. Therefore, the increase of the overall gas holdup for the 4-orice nozzle may be attributed to two reasons: the smaller orice diameter and the more number of orices. The former is benet to the formation of small bubbles at a given gas outlet velocity, resulting in a high gas holdup in the riser. Moreover, the smaller bubble size implies a lower bubble rise velocity, hence the larger recirculation of bubbles, leading to a higher gas holdup in the downcomer. The latter may provide more paths for the bubble formation, which is equivalent to increase the number of bubbles, and thus improves the gas holdup. However, despite that the Oring distributor has the most number of pores and the smallest pore diameter, the overall gas holdup of the O-ring distributor is smaller than that of the 4-orice nozzle. This may be explained by the fact that a large number of pores results in a decrease of gas outlet velocity of each pore, which weakens the impact between the gas phase and the liquid phase, leading to a larger bubble size in the sparger region, and thus the lower overall gas holdup. 3.2. Liquid velocity Fig. 7 shows the evolutions of the liquid velocity in the downcomer with the supercial gas velocity. For all the three different spargers, the downcomer liquid velocity increases rapidly in the homogenous ow regime, while increases slightly in the heterogeneous ow regime. The ow regime transition has a signicant effect on the downcomer liquid velocity regardless of the sparger type. This inuence is indirectly reected on changes of gas holdup differences between the riser and the downcomer. As shown in Fig. 8, the gas holdup difference follows the trend of the liquid velocity because it is the mainly driving force to form the liquid circulation in the reactor. In the homogeneous ow regime, both the gas holdups in the riser and the downcomer increase with the supercial gas velocity (Figs. 4 and 5), while the gas holdup in the downcomer increases relatively slowly due to the low liquid circulation velocity, leading to the increase of the gas holdup difference in this ow regime (Fig. 8). In the heterogeneous ow regime, the

Fig. 7. Comparison of downcomer liquid velocities for different spargers.

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504

499

Fig. 8. Gas holdup differences between the riser and the downcomer for different spargers.

Fig. 9. Evolutions of the volumetric oxygen transfer coefcient with the supercial gas velocity for different spargers.

intensive bubble coalescence slows down the increase of the gas holdup in the riser. However, the high liquid circulation velocity allows more and more bubbles to be entrained into the downcomer, and thus the gas holdup in the downcomer increase rapidly. The interaction between these two effects results in a slow increase of the gas holdup difference in the heterogeneous ow regime. In addition, the frictional losses also have a great effect on the liquid velocity, where the higher liquid velocity induces the higher frictional loss, which in turn reduces the liquid velocity. This may be another reason for that the increase of liquid velocity slows down at high supercial gas velocities. On the other hand, the liquid velocity in the downcomer is also affected by the sparger structure. Fig. 7 indicates that the liquid velocity of the 4-orice nozzle is higher than those of other spargers, especially in the heterogeneous ow regime. This may result from the inuence of the sparger structure on the gas holdup difference. As shown in Fig. 4, when the 4-orice nozzle is used, the gas holdup in the riser is higher than that of the 2-orice nozzle due to the smaller bubble size generated by the smaller diameter of the orice. It is also higher than that of the O-ring distributor because of the more effective impact between the gas phase and the liquid phase, which results in smaller bubbles in the sparger region. However, the differences of the downcomer gas holdup among three spargers are small. As a result, Fig. 8 shows that the largest gas holdup difference corresponds to the 4-orice nozzle, which induces the highest liquid velocity in the downcomer. There is little difference in the liquid velocity between the 2-orice nozzle and the O-ring distributor because they have the similar gas holdup difference. 3.3. Mass transfer Experimental data of the volumetric oxygen transfer coefcient kL a for different spargers are plotted in Fig. 9. For each supercial gas velocity, the value of kL a is the average of nine dissolved oxygen electrodes pO2 19. Although the experimental data are not provided here, it should be mentioned that, for any given supercial gas velocity and arbitrary sparger structure, the kL a at different positions in the reactor are similar to each other. This result indicates that the mixing is fairly well in the reactor, which conrms the assumption regarding the ideal mixing of the liquid phase in the reactor, and thus the calculation of kL a using Eq. (10) is valid in this study. Moreover, since p 2 s but R > 277 s ( p R ), it is also

reasonable to neglect the time constant of the electrode p in Eq. (10). Fig. 9 shows that the volumetric oxygen transfer coefcient increases almost linearly with the supercial gas velocity for all spargers. The ow regime transition appears to have less inuence on the volumetric oxygen transfer coefcient. It indicates that the largest kL a is obtained with the 4-orice nozzle and the middle one corresponds to the O-ring distributor, while the kL a of the 2-orice nozzle is the lowest. Since the volumetric oxygen transfer coefcient kL a is a product of the oxygen transfer coefcient kL and the specic interfacial area a, the differences in kL a for different spargers may arise from differences both in kL and a. According to Eqs. (12) and (13) proposed by Cerri et al. [23], as well as Eq. (14) proposed by Calderbank [24], the mean bubble diameter dB , the specic interfacial area a and the oxygen transfer coefcient kL corresponding to different spargers are calculated and shown in Figs. 1012, respectively. = kL a(1 ) 6 4DL 0.694 (1 ) 2
2/5

(12) g
L 1/5

dB =

(13)

Fig. 10. Evolutions of the mean bubble diameter with the supercial gas velocity for different spargers.

500

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504

area while not in the oxygen transfer coefcient kL . Since the specic interfacial area is strongly related to the gas holdup, the kL a is expected to depend indirectly on the gas holdup. These results agree well with the conclusions reported by Cerri et al. [23] and Wongsuchoto et al. [25]. 4. Correlation of experimental data 4.1. Gas holdup It is widely recognized that the gas holdup is a function of the supercial gas velocity. The following equation has been frequently used to establish the correlation between the gas holdup and the supercial gas velocity [1]: = Ug

(15)

Fig. 11. Evolutions of the specic interfacial area with the supercial gas velocity for different spargers.

a=

6 dB (1 )

(14)

Fig. 10 shows that the dB of the 4-orice nozzle is in the range from 6.3 to 7.1 mm. It is a little smaller than those of other two spargers, where dB is in the range from 7.2 to 8.2 mm for the 2orice nozzle and from 6.7 to 7.3 mm for the O-ring distributor. This result conrms the statement that the 4-orice nozzle generates the smallest bubble size. According to Eq. (14), a higher gas holdup and a smaller mean bubble diameter lead to a larger specic interfacial area, and thus the largest value of a is obtained with the 4-orice nozzle at a given supercial gas velocity as shown in Fig. 11. Furthermore, Fig. 12 indicates that the 4-orice nozzle also induces a slightly larger oxygen transfer coefcient kL than other spargers. Hence, the 4-orice nozzle generates the largest kL a at a given supercial gas velocity. Fig. 12 also shows that the kL maintains almost a constant in the experimental range of supercial gas velocity for the given sparger type, which indicates that the supercial gas velocity less affects the oxygen transfer coefcient. However, Fig. 11 shows that the specic interfacial area increases linearly with the supercial gas velocity. Therefore, the increase of kL a arises mainly from the increase in the specic interfacial

where depends on liquid physical properties as well as the reactor geometry, and is generally found to be a value ranged from 0.4 to 1 [3]. Fitting Eq. (15) to the experimental data of different spargers, the parameters determined by the regression are listed in Table 1. Fig. 13 shows that the calculated gas holdups agree well with experimental data. For comparison, some available empirical correlations in the literature are listed in Table 2. It can be found that the correlation between the gas holdup and the supercial gas velocity is strongly affected by ow regimes and sparger structures. The in correlations of the 2-orice nozzle and the O-ring distributor are similar to those reported by Chist [1], Kilonzo et al. [6] and Bla ej et al. [7], while the for the 4-orice nozzle is close to z those obtained by Lu et al. [8] and Abashar et al. [20]. However, the parameter in these correlations are quite different because it also depends on other geometric parameters of the reactor, such as area ratio between riser and downcomer, aspect ratio of the reactor, top and bottom clearances and so on. 4.2. Liquid velocity In recent years, many hydrodynamic models based on the energy conservation or the force balance principles have been proposed to predict the liquid circulation velocity in airlift reactors. However, most of these models are not applicable to the reactor in this study due to the existence of the internal liquid circulation in the riser. A hydrodynamic model is thus proposed to predict the downcomer liquid velocity in this reactor. Based on

Fig. 12. Evolutions of the oxygen transfer coefcient with the supercial gas velocity for different spargers.

Fig. 13. Comparison between predicted gas holdups and experimental data.

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504 Table 1 Parameters and correlation coefcients in different ow regimes for the three spargers. Sparger type Homogeneous ow regime Parameters 2-Orice nozzle 4-Orice nozzle O-ring distributor 2.821 0.7071 3.574 0.9548 0.7264 0.9806 Coefcient R2 0.9974 0.989 0.997 Heterogeneous ow regime Parameters 0.3218 0.9837 0.3388 0.6118 0.7761 0.6078

501

Coefcient R2 0.9911 0.9828 0.987

the assumptions mentioned in Section 2.2.2, the pressure balance over the loop in the reactor is: Ph = Pf (16)

less than 10%, frictional factors fR and fD can be estimated by the Blasius equation [30]:
0.25 fi = 0.0791ReLi ,

i = R, D

(23)

Ph represents the driving force that induced by the pressure difference between the up-ow and the down-ow: Ph = [ +
L (1 g,d ) + g g,d ]gHd

where the Reynolds numbers (ReLR and ReLD ) of the liquid phase are [27]: ReLR =
L vLr,up DR L L vLd DD L

L (1 g,r )

(24) (25)

g g,r ]gHd

L (g,r

g,d )gHd

(17)

ReLD =

where Hd is the height of the draft tube. The dynamic pressure drop ( Pf ) is the summation of the pressure drops in the up-ow region of the riser ( Pr,up ), down-ow region of the riser ( Pr,dn ), downcomer ( Pd ), as well as top ( Pt ) and bottom ( Pb ) sections of the reactor [26]: Pf = + 1 2 Pr,up +
2 L kfD vLd

Generally kft kfb in internal loop airlift reactors [31], thus kft is negligible. kfb has a value of 2 obtained by tting the experimental data. Substituting Eqs. (17) and (18) into Eq. (16), the pressure balance in the reactor is:
L (g,r

Pr,dn + 1 2
2 L kfD vLd

Pd + + 1 2

Pt +

Pb = + 1 2

1 2

2 L kfR vLr,up

g,d )gHd =

kfD +

kfb 2

2 L vLd

1 k 2 fR

2 L vLr,up

(26)

2 L kft vLr,up

2 L kfb vLd

(18)

where kfR , kfD , kft and kfb are the frictional loss coefcients. The frictional loss coefcients in the up-ow region (kfR ) and the down-ow region (kfD ) of the reactor are [27]: kfR = 4fR kfD = 4fD Hd DR Hd DD (19) (20)

Combining Eq. (26) with Eq. (7) leads to the nal expression of the liquid velocity in the downcomer:

vLd =

2gHd (g,r g,d ) 2kfD + kfb + (kfR (1 g,d )2 )/(1 g,r )2 (AD /AR )
2

1/2

(27)

where hydraulic diameters of the up-ow region (DR ) and the down-ow region (DD ) in the reactor are: DR = 4AR (21)

DD =

4AD

(22)

Wallis [28] and Kemblowski et al. [29] suggested that at a low gas holdup (<1015%) the two-phase frictional factor could be approximated by the frictional factor of the liquid phase. Since the gas holdups in the riser and downcomer of our system are far
Table 2 Empirical gas holdup correlations for airlift reactor. Reference Chist [1] Correlation
0.8920.075 Bubble ow: = 1.488Ug Coalesced bubble ow: 0.4300.015 = 0.371Ug 1 = 0.499Ug 0.815 Bubble ow: = 1.488Ug 0.449 Churn ow: = 0.371Ug 0.647 Ug 0.74 = 0.29Ug (0 Ug 0.02)

It should be noted that the parameters AR and AD in Eq. (27) are not the cross sectional areas of the riser and downcomer, but the cross sectional areas of the up-ow region and the down-ow region in the reactor due to the existence of the internal liquid circulation in the riser. AD and AR can be calculated by Eqs. (5) and (6), respectively. Based on the equations established above, the liquid velocities in the downcomer are predicted under different operation conditions. Comparisons between the experimental data and the predicted results for different spargers are shown in Fig. 14. The predicted results are generally in well agreement with the measured data for all the spargers. Figs. 15 and 16 show the calculated cross sectional area and liquid velocity of the up-ow in the reactor, respectively. In all cases, the calculated cross sectional area of the up-ow is smaller than that of the riser (0.059 m2 ), which conrms the existence of the down-ow in the riser. For all the spargers, the cross sectional area of the up-ow increases almost linearly with supercial gas velocity in the homogeneous ow regime, while reaches a

Reactor ILAR

Liquid Water or salt solution

Sparger Perforated plate, 40 holes, do = 1 mm O-ring distributor, do = 0.8 mm Perforated plate, 25 holes, do = 0.5 mm Single nozzle Perforated plate, 193 holes, do = 1 mm

Kilonzo et al. [6] Bla ej et al. [7] z Lu et al. [8] Abashar et al. [20]

ILAR ILAR ILAR ELAR

Water Water Water Water

502

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504

Fig. 14. Comparison between experimental data and predicted liquid velocities in the downcomer.

Fig. 16. Calculated liquid velocities of the up-ow in the riser. Table 3 Parameters of correlations between the volumetric oxygen transfer coefcient and the supercial gas velocity for different spargers. Sparger type Parameters 2-Orice nozzle 4-Orice nozzle O-ring distributor 0.2557 0.4661 0.384 0.8496 0.8957 0.8845 Coefcient R2 0.996 0.945 0.999

to the highest gas holdup difference, while the cross sectional area of the up-ow is smallest as a result of the smallest mean bubble diameter and the largest liquid rise velocity. 4.3. Mass transfer coefcient kL a The volumetric mass transfer coefcient kL a is commonly correlated with the supercial gas velocity or the gas power uptake per unit volume [32]. Actually, the kL a can be correlated only with the gas supercial velocity because the gas power uptake is not independent from the supercial gas velocity [33]. Similar to the case of overall gas holdup, the following equation is applied to correlate the volumetric mass transfer coefcient with the supercial gas velocity: kL a = Ug

Fig. 15. Calculated cross sectional areas of the up-ow in the riser.

constant value in the heterogeneous ow regime. The similar tendency is also observed in the evolution of the liquid velocity of the up-ow. A possible explanation may be that, in the homogeneous ow regime, because of the low gas holdup and the small liquid circulation velocity, the behavior of the reactor is close to bubble columns, where the wall effect plays an important role in retarding the movement of liquid phase as a loop within the reactor, but the driving force is larger than the ow resistances. However, in the heterogeneous ow regime, the driving force and the ow resistances may reach a balance state, resulting in the relatively stable cross sectional area and liquid velocity of the up-ow. When the 4orice nozzle is used, the liquid velocity of the up-ow is largest due
Table 4 Empirical mass transfer correlations for airlift reactor. Reference Chist [1] Bello et al. [15] Correlation kL a = 0.349(1 + Ad /Ar ) kL a = 0.76(1 + Ad /Ar )
1 0.837 Ugr

(28)

Table 3 shows the parameters obtained for all spargers by tting Eq. (28) to experimental data. The calculated results and the experimental data are compared in Fig. 17. All correlations yield a satisfactory agreement with the experimental results with less than 10% error. Some available empirical correlations in the literature are presented in Table 4. There is little difference between parameters

Reactor ELAR ELAR/ILAR

Parameter range Water, 0.026 Ugr 0.21, Ad /Ar = 0.25 and 0.44 Water, 0.0137 Ugr 0.086, ELAR: Ad /Ar = 0.11 to 0.69, ILAR: Ad /Ar = 0.13, 0.35 and 0.56 Water, 0.002 Ugr 0.0125 Water or dilute alcohol solution, 0.004 < Ug < 0.04

Sparger Perforated plate, 52 holes, do = 1 mm

0.8 Ugr

Nikakhtari and Hill [33] Albijanic et al. [34]

0.762 kL a = 0.531Ugr 0.77 kL a 0.28Ug

ELAR ILAR

6 orices, do = 1.6 mm Single orice, do = 4 mm

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504

503

a, while the oxygen transfer coefcient kL was less affected by the supercial gas velocity. Based on the material conservation and pressure balance principles, a hydrodynamic model was established to predict the liquid velocity in the downcomer. This model could be also used to estimate the cross sectional area and the liquid velocity of the up-ow in the riser. Empirical correlations were proposed to predict the overall gas holdup and the volumetric oxygen transfer coefcient for different spargers. The hydrodynamic model and empirical correlations yielded a satisfactory agreement with the experimental results. It is indicated that the relationship between the supercial gas velocity and the volumetric oxygen transfer coefcient may be more complicated than a simple power law correlation. A more reasonable correlation of the volumetric oxygen transfer coefcient should consider inuences of the reactor geometry, two-phase mixture properties, sparger structures, the gas holdup as well as the liquid circulation velocity simultaneously.
Fig. 17. Comparison between the predicted kL a and experimental data.

Acknowledgements This study is supported by the National Science Foundation of China (No. 60974068) and the National High Technology Research and Development Program of China (No. 2009AA04Z162). J.Q. Yuan acknowledges the nancial support of Alexander von HumboldtStiftung/Germany during the early stage of the study.

( and ) obtained in this study and those reported in the literature due to the differences in reactor geometry, operating variables and sparger structures. Actually, the direct effect of the supercial gas velocity is mainly reected on the gas holdup and the liquid circulation velocity. The gas holdup and the liquid circulation velocity further affect the specic interfacial area a and the mass transfer coefcient kL , respectively. Therefore, the relationship between Ug and kL a is not straightforward. Moreover, the inuences of reactor geometry, sparger structures, two-phase mixture properties and ow regimes make this relationship more complicated. Thus, a simple power law correlation may be not enough to describe the relationship between kL a and Ug . Similar conclusion was also reported by Ayazi Shamlou et al. [35]. 5. Conclusions The inuences of the sparger structure on hydrodynamics and mass transfer characteristics in an annulus sparged internal loop airlift reactor have been investigated. It is found that gas holdup, liquid velocity and volumetric oxygen transfer coefcient are strongly related to the sparger structure. The comparison between experimental data of the 2-orice nozzle and the 4-orice nozzle indicated that the sparger with the smaller orice diameter and the lager number of orices was more efcient for the mass transfer, because it produced a smaller mean bubble diameter but a higher liquid circulation velocity, and therefore a higher volumetric mass transfer coefcient. Although the O-ring distributor has the smallest pore diameter and the largest number of pores, it generated a lower mass transfer coefcient than the 4-orice nozzle. The reason may be that the low gas outlet velocity of each pore weakened the impact between gas and liquid phases, resulting in an increase of the mean bubble diameter and a decrease of the liquid circulation velocity, and thus the decrease of volumetric mass transfer coefcient. The internal liquid circulation in the riser was observed due to the large working volume, low airow rates, as well as the low ratio between cross sectional areas of the downcomer and the riser performed in all experiments. For each sparger, two ow regimes, i.e. homogeneous ow and heterogeneous ow, were identied from the evolutions of the gas holdup and the downcomer liquid velocity. In the heterogeneous ow regime, the inuence of the sparger structure on hydrodynamic parameters of the reactor was more signicant than that in the homogeneous ow regime. Moreover, for a given sparger structure, the effect of the supercial gas velocity on the kL a was mainly reected on the specic interfacial area

References
[1] M.Y. Chisti, Airlift Bioreactors, Elsevier, London and New York, 1989. [2] H.P. Luo, M.H. Al-Dahhan, Local characteristics of hydrodynamics in draft tube airlift bioreactor, Chem. Eng. Sci. 63 (2008) 30573068. [3] B. Jin, P. Yin, P. Lant, Hydrodynamics and mass transfer coefcient in threephase air-lift reactors containing activated sludge, Chem. Eng. Proc. 45 (2006) 608617. [4] J.M. van Baten, J. Ellenberger, R. Krishna, Hydrodynamics of internal air-lift reactors: experiments versus CFD simulations, Chem. Eng. Proc. 42 (2003) 733742. [5] Y.Y. Xu, L.J. Luo, J.Q. Yuan, CFD simulations to portray the bubble distribution and the hydrodynamics in an annulus sparged air-lift bioreactor, Can. J. Chem. Eng. 89 (2011) 360368. [6] P.M. Kilonzo, A. Margaritis, M.A. Bergougnou, Hydrodynamics and mass transfer characteristics in an inverse internal loop airlift-driven brous-bed bioreactor, Chem. Eng. J. 157 (2010) 146160. [7] M. Bla ej, M. Ki a, J. Marko , Scale inuence on the hydrodynamics of an internal z s s loop airlift reactor, Chem. Eng. Proc. 43 (2004) 15191527. [8] X.P. Lu, J. Ding, Y.R. Wang, J. Shi, Comparison of the hydrodynamics and mass transfer characteristics of a modied square airlift reactor with common airlift reactors, Chem. Eng. Sci. 55 (2000) 22572263. [9] G. Olivieri, A. Marzocchella, J.R. van Ommen, P. Salatino, Local and global hydrodynamics in a two-phase internal loop airlift, Chem. Eng. Sci. 62 (2007) 70687077. [10] M.A. Young, R.G. Carbonell, D.F. Ollis, Airlift bioreactors: analysis of local twophase hydrodynamics, AIChE J. 37 (1991) 403428. [11] J.C. Merchuk, A. Contreras, F. Garcia, E. Molina, Studies of mixing in a concentric tube airlift bioreactor with different spargers, Chem. Eng. Sci. 53 (1998) 709719. [12] W.J. McManamey, D.A.J. Wase, S. Raymahasay, K. Thayanithy, The inuence of gas inlet design on gas holdup values for water and various solutions in loop-type airlift fermenter, J. Chem. Technol. Biotechnol. 34 (1984) 151164. [13] J.B. Snape, J. Zahradnik, M. Fialova, N.H. Thomas, Liquid-phase properties and sparger design effects in an external-loop airlift reactor, Chem. Eng. Sci. 50 (1995) 31753188. [14] A. Contreras, F. Garcia, E. Molina, J.C. Merchuk, Inuence of sparger on energy dissipation, shear rate, and mass transfer to sea water in a concentric-tube airlift bioreactor, Enzyme Microb. Technol. 25 (1999) 820830. [15] R.A. Bello, C.W. Robinson, M. Moo-Young, Prediction of the volumetric mass transfer coefcient in pneumatic contactors, Chem. Eng. Sci. 40 (1985) 5358. [16] P. Wongsuchoto, P. Pavasant, Internal liquid circulation in annulus sparged internal loop airlift contactors, Chem. Eng. J. 100 (2004) 19. z s [17] M. Bla ej, J. Annus, J. Marko , Comparison of gassing-out and pressure-step dynamic methods for kL a measurement in an airlift reactor with internal loop, Chem. Eng. Res. Des. 82 (2004) 13751382. [18] E.S. Gaddis, Mass transfer in gasliquid contactors, Chem. Eng. Proc. 38 (1999) 503510. [19] C. Vial, S. Poncin, G. Wild, N. Midoux, A simple method for regime identication in bubble column and airlift reactors, Chem. Eng. Process. 40 (2001) 135151.

504

L. Luo et al. / Chemical Engineering Journal 175 (2011) 494504 [28] G.B. Wallis, One-Dimensional Two-Phase Flow, McGraw-Hill, New York, 1969. [29] Z. Kemblowski, J. Przywarski, A. Diab, An average gas holdup and liquid circulation velocity in airlift reactors with external loop, Chem. Eng. Sci. 48 (1993) 40234035. [30] R.H. Perry, D.W. Green, Perrys Chemical Engineers Handbook, 6th ed., McGraw Hill, New York, USA, 1984, p. 564. [31] M.Y. Chisti, B. Halard, M. Moo-Young, Liquid circulation in airlift reactors, Chem. Eng. Sci. 43 (1988) 451457. [32] M.S. Puthli, V.K. Rathod, A.B. Pandit, Gasliquid mass transfer studies with triple impler system on a laboratory scale bioreactor, Biochem. Eng. J. 23 (2004) 2530. [33] H. Nikakhtari, G.A. Hill, Hydrodynamic and oxygen mass transfer in an external loop airlift bioreactor with a packed bed, Biochem. Eng. J. 27 (2005) 138145. [34] B. Albijanic, V. Havran, D.L. Petrovic, M. Duric, M.N. Tekic, Hydrodynamics and mass transfer in a draft tube airlift reactor with dilute alcohol solutions, AIChE J. 53 (2007) 28972904. [35] P. Ayazi Shamlou, D.J. Pollard, A.P. Ison, Volumetric mass transfer coefcient in concentric-tube airlift bioreactor, Chem. Eng. Sci. 50 (1995) 15791590.

[20] M.E. Abashar, U. Narsingh, A.E. Rouillard, R. Judd, Hydrodynamic ow regimes, gas holdup, and liquid circulation in airlift reactors, Ind. Eng. Chem. Res. 37 (1998) 12511259. [21] W.J. Lu, S.J. Hwang, C.M. Chang, Liquid velocity and gas holdup in three-phase internal loop airlift reactors with low-density particles, Chem. Eng. Sci. 50 (1995) 13011310. [22] W.A.J. van Benthum, R.G.J.M. van der Lans, M.C.M. van Loosdrecht, J.J. Heijnen, Bubble recirculation regimes in an internal-loop airlift reactor, Chem. Eng. Sci. 54 (1999) 39954006. [23] M.O. Cerri, J.C. Baldacin, A.J.G. Cruz, C.O. Hokka, A.C. Badino, Prediction of mean bubble size in pneumatic reactors, Biochem. Eng. J. 53 (2010) 1217. [24] P.H. Calderbank, Physical rate processes in industrial fermentation. Part I. The interfacial area in gasliquid contacting with mechanical agitation, Chem. Eng. Res. Des. 36 (1958) 443463. [25] P. Wongsuchoto, T. Charinpanitkul, P. Pavasant, Bubble size distribution and gasliquid mass transfer in airlift contactors, Chem. Eng. J. 92 (2003) 8190. [26] J.J. Heijnen, J. Hols, R.G.J.M. van der Lans, H.L.J.M. van Leeuwen, A. Mulder, R. Weltevrede, A simple hydrodynamic model for the liquid circulation velocity in a full scale three phase airlift reactor, Chem. Eng. Sci. 52 (1997) 25272539. [27] C. Freital, M. Fialova, J. Zahradnik, J.A. Teixeira, Hydrodynamic model for threephase internal- and external-loop airlift reactors, Chem. Eng. Sci. 54 (1999) 52535258.

Você também pode gostar